summaryrefslogtreecommitdiff
path: root/ThirdParty
diff options
context:
space:
mode:
authorchai <chaifix@163.com>2021-10-25 23:29:21 +0800
committerchai <chaifix@163.com>2021-10-25 23:29:21 +0800
commit7ecf913256fb396e3027aac3318d996a716a52ef (patch)
tree4540835c881a63b665e2a692bf30115fd29e8bb0 /ThirdParty
parent0816cd70ca1a213b6ed872bcf3c0bf0912473722 (diff)
+ job system
Diffstat (limited to 'ThirdParty')
-rw-r--r--ThirdParty/stb/stb_image.h1461
-rw-r--r--ThirdParty/stb/stb_image_resize.h2634
-rw-r--r--ThirdParty/stb/stb_image_write.h1724
-rw-r--r--ThirdParty/stb/stb_truetype.h7372
-rw-r--r--ThirdParty/stb/stb_vorbis.c8503
5 files changed, 13498 insertions, 8196 deletions
diff --git a/ThirdParty/stb/stb_image.h b/ThirdParty/stb/stb_image.h
index 4df317a..d60371b 100644
--- a/ThirdParty/stb/stb_image.h
+++ b/ThirdParty/stb/stb_image.h
@@ -1,5 +1,5 @@
-/* stb_image - v2.15 - public domain image loader - http://nothings.org/stb_image.h
- no warranty implied; use at your own risk
+/* stb_image - v2.27 - public domain image loader - http://nothings.org/stb
+ no warranty implied; use at your own risk
Do this:
#define STB_IMAGE_IMPLEMENTATION
@@ -22,7 +22,7 @@
JPEG baseline & progressive (12 bpc/arithmetic not supported, same as stock IJG lib)
PNG 1/2/4/8/16-bit-per-channel
-
+
TGA (not sure what subset, if a subset)
BMP non-1bpp, non-RLE
PSD (composited view only, no extra channels, 8/16 bit-per-channel)
@@ -48,6 +48,18 @@ LICENSE
RECENT REVISION HISTORY:
+ 2.27 (2021-07-11) document stbi_info better, 16-bit PNM support, bug fixes
+ 2.26 (2020-07-13) many minor fixes
+ 2.25 (2020-02-02) fix warnings
+ 2.24 (2020-02-02) fix warnings; thread-local failure_reason and flip_vertically
+ 2.23 (2019-08-11) fix clang static analysis warning
+ 2.22 (2019-03-04) gif fixes, fix warnings
+ 2.21 (2019-02-25) fix typo in comment
+ 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs
+ 2.19 (2018-02-11) fix warning
+ 2.18 (2018-01-30) fix warnings
+ 2.17 (2018-01-29) bugfix, 1-bit BMP, 16-bitness query, fix warnings
+ 2.16 (2017-07-23) all functions have 16-bit variants; optimizations; bugfixes
2.15 (2017-03-18) fix png-1,2,4; all Imagenet JPGs; no runtime SSE detection on GCC
2.14 (2017-03-03) remove deprecated STBI_JPEG_OLD; fixes for Imagenet JPGs
2.13 (2016-12-04) experimental 16-bit API, only for PNG so far; fixes
@@ -58,10 +70,6 @@ RECENT REVISION HISTORY:
correct channel count for PNG & BMP
2.10 (2016-01-22) avoid warning introduced in 2.09
2.09 (2016-01-16) 16-bit TGA; comments in PNM files; STBI_REALLOC_SIZED
- 2.08 (2015-09-13) fix to 2.07 cleanup, reading RGB PSD as RGBA
- 2.07 (2015-09-13) partial animated GIF support
- limited 16-bit PSD support
- minor bugs, code cleanup, and compiler warnings
See end of file for full revision history.
@@ -77,39 +85,52 @@ RECENT REVISION HISTORY:
Thatcher Ulrich (psd) Nicolas Guillemot (vertical flip)
Ken Miller (pgm, ppm) Richard Mitton (16-bit PSD)
github:urraka (animated gif) Junggon Kim (PNM comments)
- Daniel Gibson (16-bit TGA)
+ Christopher Forseth (animated gif) Daniel Gibson (16-bit TGA)
socks-the-fox (16-bit PNG)
- Jeremy Sawicki (handle all ImageNetManager JPGs)
- Optimizations & bugfixes
- Fabian "ryg" Giesen
- Arseny Kapoulkine
+ Jeremy Sawicki (handle all ImageNet JPGs)
+ Optimizations & bugfixes Mikhail Morozov (1-bit BMP)
+ Fabian "ryg" Giesen Anael Seghezzi (is-16-bit query)
+ Arseny Kapoulkine Simon Breuss (16-bit PNM)
+ John-Mark Allen
+ Carmelo J Fdez-Aguera
Bug & warning fixes
- Marc LeBlanc David Woo Guillaume George Martins Mozeiko
- Christpher Lloyd Jerry Jansson Joseph Thomson Phil Jordan
- Dave Moore Roy Eltham Hayaki Saito Nathan Reed
- Won Chun Luke Graham Johan Duparc Nick Verigakis
- the Horde3D community Thomas Ruf Ronny Chevalier Baldur Karlsson
- Janez Zemva John Bartholomew Michal Cichon github:rlyeh
- Jonathan Blow Ken Hamada Tero Hanninen github:romigrou
- Laurent Gomila Cort Stratton Sergio Gonzalez github:svdijk
- Aruelien Pocheville Thibault Reuille Cass Everitt github:snagar
- Ryamond Barbiero Paul Du Bois Engin Manap github:Zelex
- Michaelangel007@github Philipp Wiesemann Dale Weiler github:grim210
- Oriol Ferrer Mesia Josh Tobin Matthew Gregan github:sammyhw
- Blazej Dariusz Roszkowski Gregory Mullen github:phprus
-
+ Marc LeBlanc David Woo Guillaume George Martins Mozeiko
+ Christpher Lloyd Jerry Jansson Joseph Thomson Blazej Dariusz Roszkowski
+ Phil Jordan Dave Moore Roy Eltham
+ Hayaki Saito Nathan Reed Won Chun
+ Luke Graham Johan Duparc Nick Verigakis the Horde3D community
+ Thomas Ruf Ronny Chevalier github:rlyeh
+ Janez Zemva John Bartholomew Michal Cichon github:romigrou
+ Jonathan Blow Ken Hamada Tero Hanninen github:svdijk
+ Eugene Golushkov Laurent Gomila Cort Stratton github:snagar
+ Aruelien Pocheville Sergio Gonzalez Thibault Reuille github:Zelex
+ Cass Everitt Ryamond Barbiero github:grim210
+ Paul Du Bois Engin Manap Aldo Culquicondor github:sammyhw
+ Philipp Wiesemann Dale Weiler Oriol Ferrer Mesia github:phprus
+ Josh Tobin Matthew Gregan github:poppolopoppo
+ Julian Raschke Gregory Mullen Christian Floisand github:darealshinji
+ Baldur Karlsson Kevin Schmidt JR Smith github:Michaelangel007
+ Brad Weinberger Matvey Cherevko github:mosra
+ Luca Sas Alexander Veselov Zack Middleton [reserved]
+ Ryan C. Gordon [reserved] [reserved]
+ DO NOT ADD YOUR NAME HERE
+
+ Jacko Dirks
+
+ To add your name to the credits, pick a random blank space in the middle and fill it.
+ 80% of merge conflicts on stb PRs are due to people adding their name at the end
+ of the credits.
*/
+
#ifndef STBI_INCLUDE_STB_IMAGE_H
#define STBI_INCLUDE_STB_IMAGE_H
// DOCUMENTATION
//
// Limitations:
-// - no 16-bit-per-channel PNG
// - no 12-bit-per-channel JPEG
// - no JPEGs with arithmetic coding
-// - no 1-bit BMP
// - GIF always returns *comp=4
//
// Basic usage (see HDR discussion below for HDR usage):
@@ -133,11 +154,12 @@ RECENT REVISION HISTORY:
// with each pixel consisting of N interleaved 8-bit components; the first
// pixel pointed to is top-left-most in the image. There is no padding between
// image scanlines or between pixels, regardless of format. The number of
-// components N is 'req_comp' if req_comp is non-zero, or *comp otherwise.
-// If req_comp is non-zero, *comp has the number of components that _would_
-// have been output otherwise. E.g. if you set req_comp to 4, you will always
-// get RGBA output, but you can check *comp to see if it's trivially opaque
-// because e.g. there were only 3 channels in the source image.
+// components N is 'desired_channels' if desired_channels is non-zero, or
+// *channels_in_file otherwise. If desired_channels is non-zero,
+// *channels_in_file has the number of components that _would_ have been
+// output otherwise. E.g. if you set desired_channels to 4, you will always
+// get RGBA output, but you can check *channels_in_file to see if it's trivially
+// opaque because e.g. there were only 3 channels in the source image.
//
// An output image with N components has the following components interleaved
// in this order in each pixel:
@@ -149,14 +171,50 @@ RECENT REVISION HISTORY:
// 4 red, green, blue, alpha
//
// If image loading fails for any reason, the return value will be NULL,
-// and *x, *y, *comp will be unchanged. The function stbi_failure_reason()
-// can be queried for an extremely brief, end-user unfriendly explanation
-// of why the load failed. Define STBI_NO_FAILURE_STRINGS to avoid
-// compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly
+// and *x, *y, *channels_in_file will be unchanged. The function
+// stbi_failure_reason() can be queried for an extremely brief, end-user
+// unfriendly explanation of why the load failed. Define STBI_NO_FAILURE_STRINGS
+// to avoid compiling these strings at all, and STBI_FAILURE_USERMSG to get slightly
// more user-friendly ones.
//
// Paletted PNG, BMP, GIF, and PIC images are automatically depalettized.
//
+// To query the width, height and component count of an image without having to
+// decode the full file, you can use the stbi_info family of functions:
+//
+// int x,y,n,ok;
+// ok = stbi_info(filename, &x, &y, &n);
+// // returns ok=1 and sets x, y, n if image is a supported format,
+// // 0 otherwise.
+//
+// Note that stb_image pervasively uses ints in its public API for sizes,
+// including sizes of memory buffers. This is now part of the API and thus
+// hard to change without causing breakage. As a result, the various image
+// loaders all have certain limits on image size; these differ somewhat
+// by format but generally boil down to either just under 2GB or just under
+// 1GB. When the decoded image would be larger than this, stb_image decoding
+// will fail.
+//
+// Additionally, stb_image will reject image files that have any of their
+// dimensions set to a larger value than the configurable STBI_MAX_DIMENSIONS,
+// which defaults to 2**24 = 16777216 pixels. Due to the above memory limit,
+// the only way to have an image with such dimensions load correctly
+// is for it to have a rather extreme aspect ratio. Either way, the
+// assumption here is that such larger images are likely to be malformed
+// or malicious. If you do need to load an image with individual dimensions
+// larger than that, and it still fits in the overall size limit, you can
+// #define STBI_MAX_DIMENSIONS on your own to be something larger.
+//
+// ===========================================================================
+//
+// UNICODE:
+//
+// If compiling for Windows and you wish to use Unicode filenames, compile
+// with
+// #define STBI_WINDOWS_UTF8
+// and pass utf8-encoded filenames. Call stbi_convert_wchar_to_utf8 to convert
+// Windows wchar_t filenames to utf8.
+//
// ===========================================================================
//
// Philosophy
@@ -169,12 +227,12 @@ RECENT REVISION HISTORY:
//
// Sometimes I let "good performance" creep up in priority over "easy to maintain",
// and for best performance I may provide less-easy-to-use APIs that give higher
-// performance, in addition to the easy to use ones. Nevertheless, it's important
+// performance, in addition to the easy-to-use ones. Nevertheless, it's important
// to keep in mind that from the standpoint of you, a client of this library,
// all you care about is #1 and #3, and stb libraries DO NOT emphasize #3 above all.
//
// Some secondary priorities arise directly from the first two, some of which
-// make more explicit reasons why performance can't be emphasized.
+// provide more explicit reasons why performance can't be emphasized.
//
// - Portable ("ease of use")
// - Small source code footprint ("easy to maintain")
@@ -217,11 +275,10 @@ RECENT REVISION HISTORY:
//
// HDR image support (disable by defining STBI_NO_HDR)
//
-// stb_image now supports loading HDR images in general, and currently
-// the Radiance .HDR file format, although the support is provided
-// generically. You can still load any file through the existing interface;
-// if you attempt to load an HDR file, it will be automatically remapped to
-// LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1;
+// stb_image supports loading HDR images in general, and currently the Radiance
+// .HDR file format specifically. You can still load any file through the existing
+// interface; if you attempt to load an HDR file, it will be automatically remapped
+// to LDR, assuming gamma 2.2 and an arbitrary scale factor defaulting to 1;
// both of these constants can be reconfigured through this interface:
//
// stbi_hdr_to_ldr_gamma(2.2f);
@@ -253,11 +310,10 @@ RECENT REVISION HISTORY:
//
// iPhone PNG support:
//
-// By default we convert iphone-formatted PNGs back to RGB, even though
-// they are internally encoded differently. You can disable this conversion
-// by by calling stbi_convert_iphone_png_to_rgb(0), in which case
-// you will always just get the native iphone "format" through (which
-// is BGR stored in RGB).
+// We optionally support converting iPhone-formatted PNGs (which store
+// premultiplied BGRA) back to RGB, even though they're internally encoded
+// differently. To enable this conversion, call
+// stbi_convert_iphone_png_to_rgb(1).
//
// Call stbi_set_unpremultiply_on_load(1) as well to force a divide per
// pixel to remove any premultiplied alpha *only* if the image file explicitly
@@ -299,7 +355,14 @@ RECENT REVISION HISTORY:
// - If you use STBI_NO_PNG (or _ONLY_ without PNG), and you still
// want the zlib decoder to be available, #define STBI_SUPPORT_ZLIB
//
-
+// - If you define STBI_MAX_DIMENSIONS, stb_image will reject images greater
+// than that size (in either width or height) without further processing.
+// This is to let programs in the wild set an upper bound to prevent
+// denial-of-service attacks on untrusted data, as one could generate a
+// valid image of gigantic dimensions and force stb_image to allocate a
+// huge block of memory and spend disproportionate time decoding it. By
+// default this is set to (1 << 24), which is 16777216, but that's still
+// very big.
#ifndef STBI_NO_STDIO
#include <stdio.h>
@@ -309,7 +372,7 @@ RECENT REVISION HISTORY:
enum
{
- STBI_default = 0, // only used for req_comp
+ STBI_default = 0, // only used for desired_channels
STBI_grey = 1,
STBI_grey_alpha = 2,
@@ -317,6 +380,7 @@ enum
STBI_rgb_alpha = 4
};
+#include <stdlib.h>
typedef unsigned char stbi_uc;
typedef unsigned short stbi_us;
@@ -324,11 +388,13 @@ typedef unsigned short stbi_us;
extern "C" {
#endif
+#ifndef STBIDEF
#ifdef STB_IMAGE_STATIC
#define STBIDEF static
#else
#define STBIDEF extern
#endif
+#endif
//////////////////////////////////////////////////////////////////////////////
//
@@ -351,36 +417,46 @@ typedef struct
// 8-bits-per-channel interface
//
-STBIDEF stbi_uc *stbi_load (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
STBIDEF stbi_uc *stbi_load_from_memory (stbi_uc const *buffer, int len , int *x, int *y, int *channels_in_file, int desired_channels);
STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk , void *user, int *x, int *y, int *channels_in_file, int desired_channels);
#ifndef STBI_NO_STDIO
-STBIDEF stbi_uc *stbi_load_from_file (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_uc *stbi_load (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_uc *stbi_load_from_file (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
// for stbi_load_from_file, file pointer is left pointing immediately after image
#endif
+#ifndef STBI_NO_GIF
+STBIDEF stbi_uc *stbi_load_gif_from_memory(stbi_uc const *buffer, int len, int **delays, int *x, int *y, int *z, int *comp, int req_comp);
+#endif
+
+#ifdef STBI_WINDOWS_UTF8
+STBIDEF int stbi_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input);
+#endif
+
////////////////////////////////////
//
// 16-bits-per-channel interface
//
-STBIDEF stbi_us *stbi_load_16(char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_us *stbi_load_16_from_memory (stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels);
+STBIDEF stbi_us *stbi_load_16_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels);
+
#ifndef STBI_NO_STDIO
+STBIDEF stbi_us *stbi_load_16 (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
STBIDEF stbi_us *stbi_load_from_file_16(FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
#endif
-// @TODO the other variants
////////////////////////////////////
//
// float-per-channel interface
//
#ifndef STBI_NO_LINEAR
- STBIDEF float *stbi_loadf (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
STBIDEF float *stbi_loadf_from_memory (stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels);
STBIDEF float *stbi_loadf_from_callbacks (stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels);
#ifndef STBI_NO_STDIO
+ STBIDEF float *stbi_loadf (char const *filename, int *x, int *y, int *channels_in_file, int desired_channels);
STBIDEF float *stbi_loadf_from_file (FILE *f, int *x, int *y, int *channels_in_file, int desired_channels);
#endif
#endif
@@ -405,7 +481,7 @@ STBIDEF int stbi_is_hdr_from_file(FILE *f);
// get a VERY brief reason for failure
-// NOT THREADSAFE
+// on most compilers (and ALL modern mainstream compilers) this is threadsafe
STBIDEF const char *stbi_failure_reason (void);
// free the loaded image -- this is just free()
@@ -414,11 +490,14 @@ STBIDEF void stbi_image_free (void *retval_from_stbi_load);
// get image dimensions & components without fully decoding
STBIDEF int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp);
STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp);
+STBIDEF int stbi_is_16_bit_from_memory(stbi_uc const *buffer, int len);
+STBIDEF int stbi_is_16_bit_from_callbacks(stbi_io_callbacks const *clbk, void *user);
#ifndef STBI_NO_STDIO
-STBIDEF int stbi_info (char const *filename, int *x, int *y, int *comp);
-STBIDEF int stbi_info_from_file (FILE *f, int *x, int *y, int *comp);
-
+STBIDEF int stbi_info (char const *filename, int *x, int *y, int *comp);
+STBIDEF int stbi_info_from_file (FILE *f, int *x, int *y, int *comp);
+STBIDEF int stbi_is_16_bit (char const *filename);
+STBIDEF int stbi_is_16_bit_from_file(FILE *f);
#endif
@@ -435,6 +514,13 @@ STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert);
// flip the image vertically, so the first pixel in the output array is the bottom left
STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip);
+// as above, but only applies to images loaded on the thread that calls the function
+// this function is only available if your compiler supports thread-local variables;
+// calling it will fail to link if your compiler doesn't
+STBIDEF void stbi_set_unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply);
+STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert);
+STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip);
+
// ZLIB client - used by PNG, available for other purposes
STBIDEF char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen);
@@ -502,7 +588,7 @@ STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const ch
#include <limits.h>
#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR)
-#include <math.h> // ldexp
+#include <math.h> // ldexp, pow
#endif
#ifndef STBI_NO_STDIO
@@ -514,6 +600,12 @@ STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const ch
#define STBI_ASSERT(x) assert(x)
#endif
+#ifdef __cplusplus
+#define STBI_EXTERN extern "C"
+#else
+#define STBI_EXTERN extern
+#endif
+
#ifndef _MSC_VER
#ifdef __cplusplus
@@ -525,6 +617,23 @@ STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const ch
#define stbi_inline __forceinline
#endif
+#ifndef STBI_NO_THREAD_LOCALS
+ #if defined(__cplusplus) && __cplusplus >= 201103L
+ #define STBI_THREAD_LOCAL thread_local
+ #elif defined(__GNUC__) && __GNUC__ < 5
+ #define STBI_THREAD_LOCAL __thread
+ #elif defined(_MSC_VER)
+ #define STBI_THREAD_LOCAL __declspec(thread)
+ #elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 201112L && !defined(__STDC_NO_THREADS__)
+ #define STBI_THREAD_LOCAL _Thread_local
+ #endif
+
+ #ifndef STBI_THREAD_LOCAL
+ #if defined(__GNUC__)
+ #define STBI_THREAD_LOCAL __thread
+ #endif
+ #endif
+#endif
#ifdef _MSC_VER
typedef unsigned short stbi__uint16;
@@ -555,7 +664,7 @@ typedef unsigned char validate_uint32[sizeof(stbi__uint32)==4 ? 1 : -1];
#ifdef STBI_HAS_LROTL
#define stbi_lrot(x,y) _lrotl(x,y)
#else
- #define stbi_lrot(x,y) (((x) << (y)) | ((x) >> (32 - (y))))
+ #define stbi_lrot(x,y) (((x) << (y)) | ((x) >> (-(y) & 31)))
#endif
#if defined(STBI_MALLOC) && defined(STBI_FREE) && (defined(STBI_REALLOC) || defined(STBI_REALLOC_SIZED))
@@ -638,15 +747,19 @@ static int stbi__cpuid3(void)
#define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name
-static int stbi__sse2_available()
+#if !defined(STBI_NO_JPEG) && defined(STBI_SSE2)
+static int stbi__sse2_available(void)
{
int info3 = stbi__cpuid3();
return ((info3 >> 26) & 1) != 0;
}
+#endif
+
#else // assume GCC-style if not VC++
#define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16)))
-static int stbi__sse2_available()
+#if !defined(STBI_NO_JPEG) && defined(STBI_SSE2)
+static int stbi__sse2_available(void)
{
// If we're even attempting to compile this on GCC/Clang, that means
// -msse2 is on, which means the compiler is allowed to use SSE2
@@ -654,6 +767,8 @@ static int stbi__sse2_available()
return 1;
}
#endif
+
+#endif
#endif
// ARM NEON
@@ -663,14 +778,21 @@ static int stbi__sse2_available()
#ifdef STBI_NEON
#include <arm_neon.h>
-// assume GCC or Clang on ARM targets
+#ifdef _MSC_VER
+#define STBI_SIMD_ALIGN(type, name) __declspec(align(16)) type name
+#else
#define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16)))
#endif
+#endif
#ifndef STBI_SIMD_ALIGN
#define STBI_SIMD_ALIGN(type, name) type name
#endif
+#ifndef STBI_MAX_DIMENSIONS
+#define STBI_MAX_DIMENSIONS (1 << 24)
+#endif
+
///////////////////////////////////////////////
//
// stbi__context struct and start_xxx functions
@@ -688,6 +810,7 @@ typedef struct
int read_from_callbacks;
int buflen;
stbi_uc buffer_start[128];
+ int callback_already_read;
stbi_uc *img_buffer, *img_buffer_end;
stbi_uc *img_buffer_original, *img_buffer_original_end;
@@ -701,6 +824,7 @@ static void stbi__start_mem(stbi__context *s, stbi_uc const *buffer, int len)
{
s->io.read = NULL;
s->read_from_callbacks = 0;
+ s->callback_already_read = 0;
s->img_buffer = s->img_buffer_original = (stbi_uc *) buffer;
s->img_buffer_end = s->img_buffer_original_end = (stbi_uc *) buffer+len;
}
@@ -712,7 +836,8 @@ static void stbi__start_callbacks(stbi__context *s, stbi_io_callbacks *c, void *
s->io_user_data = user;
s->buflen = sizeof(s->buffer_start);
s->read_from_callbacks = 1;
- s->img_buffer_original = s->buffer_start;
+ s->callback_already_read = 0;
+ s->img_buffer = s->img_buffer_original = s->buffer_start;
stbi__refill_buffer(s);
s->img_buffer_original_end = s->img_buffer_end;
}
@@ -726,12 +851,17 @@ static int stbi__stdio_read(void *user, char *data, int size)
static void stbi__stdio_skip(void *user, int n)
{
+ int ch;
fseek((FILE*) user, n, SEEK_CUR);
+ ch = fgetc((FILE*) user); /* have to read a byte to reset feof()'s flag */
+ if (ch != EOF) {
+ ungetc(ch, (FILE *) user); /* push byte back onto stream if valid. */
+ }
}
static int stbi__stdio_eof(void *user)
{
- return feof((FILE*) user);
+ return feof((FILE*) user) || ferror((FILE *) user);
}
static stbi_io_callbacks stbi__stdio_callbacks =
@@ -782,6 +912,7 @@ static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp);
static int stbi__png_test(stbi__context *s);
static void *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp);
+static int stbi__png_is16(stbi__context *s);
#endif
#ifndef STBI_NO_BMP
@@ -800,6 +931,7 @@ static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp);
static int stbi__psd_test(stbi__context *s);
static void *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc);
static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp);
+static int stbi__psd_is16(stbi__context *s);
#endif
#ifndef STBI_NO_HDR
@@ -817,6 +949,7 @@ static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp);
#ifndef STBI_NO_GIF
static int stbi__gif_test(stbi__context *s);
static void *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
+static void *stbi__load_gif_main(stbi__context *s, int **delays, int *x, int *y, int *z, int *comp, int req_comp);
static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp);
#endif
@@ -824,21 +957,27 @@ static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp);
static int stbi__pnm_test(stbi__context *s);
static void *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri);
static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp);
+static int stbi__pnm_is16(stbi__context *s);
#endif
-// this is not threadsafe
-static const char *stbi__g_failure_reason;
+static
+#ifdef STBI_THREAD_LOCAL
+STBI_THREAD_LOCAL
+#endif
+const char *stbi__g_failure_reason;
STBIDEF const char *stbi_failure_reason(void)
{
return stbi__g_failure_reason;
}
+#ifndef STBI_NO_FAILURE_STRINGS
static int stbi__err(const char *str)
{
stbi__g_failure_reason = str;
return 0;
}
+#endif
static void *stbi__malloc(size_t size)
{
@@ -877,11 +1016,13 @@ static int stbi__mul2sizes_valid(int a, int b)
return a <= INT_MAX/b;
}
+#if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR)
// returns 1 if "a*b + add" has no negative terms/factors and doesn't overflow
static int stbi__mad2sizes_valid(int a, int b, int add)
{
return stbi__mul2sizes_valid(a, b) && stbi__addsizes_valid(a*b, add);
}
+#endif
// returns 1 if "a*b*c + add" has no negative terms/factors and doesn't overflow
static int stbi__mad3sizes_valid(int a, int b, int c, int add)
@@ -891,18 +1032,22 @@ static int stbi__mad3sizes_valid(int a, int b, int c, int add)
}
// returns 1 if "a*b*c*d + add" has no negative terms/factors and doesn't overflow
+#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM)
static int stbi__mad4sizes_valid(int a, int b, int c, int d, int add)
{
return stbi__mul2sizes_valid(a, b) && stbi__mul2sizes_valid(a*b, c) &&
stbi__mul2sizes_valid(a*b*c, d) && stbi__addsizes_valid(a*b*c*d, add);
}
+#endif
+#if !defined(STBI_NO_JPEG) || !defined(STBI_NO_PNG) || !defined(STBI_NO_TGA) || !defined(STBI_NO_HDR)
// mallocs with size overflow checking
static void *stbi__malloc_mad2(int a, int b, int add)
{
if (!stbi__mad2sizes_valid(a, b, add)) return NULL;
return stbi__malloc(a*b + add);
}
+#endif
static void *stbi__malloc_mad3(int a, int b, int c, int add)
{
@@ -910,11 +1055,13 @@ static void *stbi__malloc_mad3(int a, int b, int c, int add)
return stbi__malloc(a*b*c + add);
}
+#if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) || !defined(STBI_NO_PNM)
static void *stbi__malloc_mad4(int a, int b, int c, int d, int add)
{
if (!stbi__mad4sizes_valid(a, b, c, d, add)) return NULL;
return stbi__malloc(a*b*c*d + add);
}
+#endif
// stbi__err - error
// stbi__errpf - error returning pointer to float
@@ -944,13 +1091,29 @@ static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp);
static stbi_uc *stbi__hdr_to_ldr(float *data, int x, int y, int comp);
#endif
-static int stbi__vertically_flip_on_load = 0;
+static int stbi__vertically_flip_on_load_global = 0;
STBIDEF void stbi_set_flip_vertically_on_load(int flag_true_if_should_flip)
{
- stbi__vertically_flip_on_load = flag_true_if_should_flip;
+ stbi__vertically_flip_on_load_global = flag_true_if_should_flip;
+}
+
+#ifndef STBI_THREAD_LOCAL
+#define stbi__vertically_flip_on_load stbi__vertically_flip_on_load_global
+#else
+static STBI_THREAD_LOCAL int stbi__vertically_flip_on_load_local, stbi__vertically_flip_on_load_set;
+
+STBIDEF void stbi_set_flip_vertically_on_load_thread(int flag_true_if_should_flip)
+{
+ stbi__vertically_flip_on_load_local = flag_true_if_should_flip;
+ stbi__vertically_flip_on_load_set = 1;
}
+#define stbi__vertically_flip_on_load (stbi__vertically_flip_on_load_set \
+ ? stbi__vertically_flip_on_load_local \
+ : stbi__vertically_flip_on_load_global)
+#endif // STBI_THREAD_LOCAL
+
static void *stbi__load_main(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri, int bpc)
{
memset(ri, 0, sizeof(*ri)); // make sure it's initialized if we add new fields
@@ -958,9 +1121,8 @@ static void *stbi__load_main(stbi__context *s, int *x, int *y, int *comp, int re
ri->channel_order = STBI_ORDER_RGB; // all current input & output are this, but this is here so we can add BGR order
ri->num_channels = 0;
- #ifndef STBI_NO_JPEG
- if (stbi__jpeg_test(s)) return stbi__jpeg_load(s,x,y,comp,req_comp, ri);
- #endif
+ // test the formats with a very explicit header first (at least a FOURCC
+ // or distinctive magic number first)
#ifndef STBI_NO_PNG
if (stbi__png_test(s)) return stbi__png_load(s,x,y,comp,req_comp, ri);
#endif
@@ -972,10 +1134,19 @@ static void *stbi__load_main(stbi__context *s, int *x, int *y, int *comp, int re
#endif
#ifndef STBI_NO_PSD
if (stbi__psd_test(s)) return stbi__psd_load(s,x,y,comp,req_comp, ri, bpc);
+ #else
+ STBI_NOTUSED(bpc);
#endif
#ifndef STBI_NO_PIC
if (stbi__pic_test(s)) return stbi__pic_load(s,x,y,comp,req_comp, ri);
#endif
+
+ // then the formats that can end up attempting to load with just 1 or 2
+ // bytes matching expectations; these are prone to false positives, so
+ // try them later
+ #ifndef STBI_NO_JPEG
+ if (stbi__jpeg_test(s)) return stbi__jpeg_load(s,x,y,comp,req_comp, ri);
+ #endif
#ifndef STBI_NO_PNM
if (stbi__pnm_test(s)) return stbi__pnm_load(s,x,y,comp,req_comp, ri);
#endif
@@ -1028,6 +1199,44 @@ static stbi__uint16 *stbi__convert_8_to_16(stbi_uc *orig, int w, int h, int chan
return enlarged;
}
+static void stbi__vertical_flip(void *image, int w, int h, int bytes_per_pixel)
+{
+ int row;
+ size_t bytes_per_row = (size_t)w * bytes_per_pixel;
+ stbi_uc temp[2048];
+ stbi_uc *bytes = (stbi_uc *)image;
+
+ for (row = 0; row < (h>>1); row++) {
+ stbi_uc *row0 = bytes + row*bytes_per_row;
+ stbi_uc *row1 = bytes + (h - row - 1)*bytes_per_row;
+ // swap row0 with row1
+ size_t bytes_left = bytes_per_row;
+ while (bytes_left) {
+ size_t bytes_copy = (bytes_left < sizeof(temp)) ? bytes_left : sizeof(temp);
+ memcpy(temp, row0, bytes_copy);
+ memcpy(row0, row1, bytes_copy);
+ memcpy(row1, temp, bytes_copy);
+ row0 += bytes_copy;
+ row1 += bytes_copy;
+ bytes_left -= bytes_copy;
+ }
+ }
+}
+
+#ifndef STBI_NO_GIF
+static void stbi__vertical_flip_slices(void *image, int w, int h, int z, int bytes_per_pixel)
+{
+ int slice;
+ int slice_size = w * h * bytes_per_pixel;
+
+ stbi_uc *bytes = (stbi_uc *)image;
+ for (slice = 0; slice < z; ++slice) {
+ stbi__vertical_flip(bytes, w, h, bytes_per_pixel);
+ bytes += slice_size;
+ }
+}
+#endif
+
static unsigned char *stbi__load_and_postprocess_8bit(stbi__context *s, int *x, int *y, int *comp, int req_comp)
{
stbi__result_info ri;
@@ -1036,8 +1245,10 @@ static unsigned char *stbi__load_and_postprocess_8bit(stbi__context *s, int *x,
if (result == NULL)
return NULL;
+ // it is the responsibility of the loaders to make sure we get either 8 or 16 bit.
+ STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16);
+
if (ri.bits_per_channel != 8) {
- STBI_ASSERT(ri.bits_per_channel == 16);
result = stbi__convert_16_to_8((stbi__uint16 *) result, *x, *y, req_comp == 0 ? *comp : req_comp);
ri.bits_per_channel = 8;
}
@@ -1045,21 +1256,8 @@ static unsigned char *stbi__load_and_postprocess_8bit(stbi__context *s, int *x,
// @TODO: move stbi__convert_format to here
if (stbi__vertically_flip_on_load) {
- int w = *x, h = *y;
int channels = req_comp ? req_comp : *comp;
- int row,col,z;
- stbi_uc *image = (stbi_uc *) result;
-
- // @OPTIMIZE: use a bigger temp buffer and memcpy multiple pixels at once
- for (row = 0; row < (h>>1); row++) {
- for (col = 0; col < w; col++) {
- for (z = 0; z < channels; z++) {
- stbi_uc temp = image[(row * w + col) * channels + z];
- image[(row * w + col) * channels + z] = image[((h - row - 1) * w + col) * channels + z];
- image[((h - row - 1) * w + col) * channels + z] = temp;
- }
- }
- }
+ stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi_uc));
}
return (unsigned char *) result;
@@ -1073,8 +1271,10 @@ static stbi__uint16 *stbi__load_and_postprocess_16bit(stbi__context *s, int *x,
if (result == NULL)
return NULL;
+ // it is the responsibility of the loaders to make sure we get either 8 or 16 bit.
+ STBI_ASSERT(ri.bits_per_channel == 8 || ri.bits_per_channel == 16);
+
if (ri.bits_per_channel != 16) {
- STBI_ASSERT(ri.bits_per_channel == 8);
result = stbi__convert_8_to_16((stbi_uc *) result, *x, *y, req_comp == 0 ? *comp : req_comp);
ri.bits_per_channel = 16;
}
@@ -1083,55 +1283,57 @@ static stbi__uint16 *stbi__load_and_postprocess_16bit(stbi__context *s, int *x,
// @TODO: special case RGB-to-Y (and RGBA-to-YA) for 8-bit-to-16-bit case to keep more precision
if (stbi__vertically_flip_on_load) {
- int w = *x, h = *y;
int channels = req_comp ? req_comp : *comp;
- int row,col,z;
- stbi__uint16 *image = (stbi__uint16 *) result;
-
- // @OPTIMIZE: use a bigger temp buffer and memcpy multiple pixels at once
- for (row = 0; row < (h>>1); row++) {
- for (col = 0; col < w; col++) {
- for (z = 0; z < channels; z++) {
- stbi__uint16 temp = image[(row * w + col) * channels + z];
- image[(row * w + col) * channels + z] = image[((h - row - 1) * w + col) * channels + z];
- image[((h - row - 1) * w + col) * channels + z] = temp;
- }
- }
- }
+ stbi__vertical_flip(result, *x, *y, channels * sizeof(stbi__uint16));
}
return (stbi__uint16 *) result;
}
-#ifndef STBI_NO_HDR
+#if !defined(STBI_NO_HDR) && !defined(STBI_NO_LINEAR)
static void stbi__float_postprocess(float *result, int *x, int *y, int *comp, int req_comp)
{
if (stbi__vertically_flip_on_load && result != NULL) {
- int w = *x, h = *y;
- int depth = req_comp ? req_comp : *comp;
- int row,col,z;
- float temp;
-
- // @OPTIMIZE: use a bigger temp buffer and memcpy multiple pixels at once
- for (row = 0; row < (h>>1); row++) {
- for (col = 0; col < w; col++) {
- for (z = 0; z < depth; z++) {
- temp = result[(row * w + col) * depth + z];
- result[(row * w + col) * depth + z] = result[((h - row - 1) * w + col) * depth + z];
- result[((h - row - 1) * w + col) * depth + z] = temp;
- }
- }
- }
+ int channels = req_comp ? req_comp : *comp;
+ stbi__vertical_flip(result, *x, *y, channels * sizeof(float));
}
}
#endif
#ifndef STBI_NO_STDIO
+#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8)
+STBI_EXTERN __declspec(dllimport) int __stdcall MultiByteToWideChar(unsigned int cp, unsigned long flags, const char *str, int cbmb, wchar_t *widestr, int cchwide);
+STBI_EXTERN __declspec(dllimport) int __stdcall WideCharToMultiByte(unsigned int cp, unsigned long flags, const wchar_t *widestr, int cchwide, char *str, int cbmb, const char *defchar, int *used_default);
+#endif
+
+#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8)
+STBIDEF int stbi_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input)
+{
+ return WideCharToMultiByte(65001 /* UTF8 */, 0, input, -1, buffer, (int) bufferlen, NULL, NULL);
+}
+#endif
+
static FILE *stbi__fopen(char const *filename, char const *mode)
{
FILE *f;
+#if defined(_WIN32) && defined(STBI_WINDOWS_UTF8)
+ wchar_t wMode[64];
+ wchar_t wFilename[1024];
+ if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, filename, -1, wFilename, sizeof(wFilename)/sizeof(*wFilename)))
+ return 0;
+
+ if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, mode, -1, wMode, sizeof(wMode)/sizeof(*wMode)))
+ return 0;
+
#if defined(_MSC_VER) && _MSC_VER >= 1400
+ if (0 != _wfopen_s(&f, wFilename, wMode))
+ f = 0;
+#else
+ f = _wfopen(wFilename, wMode);
+#endif
+
+#elif defined(_MSC_VER) && _MSC_VER >= 1400
if (0 != fopen_s(&f, filename, mode))
f=0;
#else
@@ -1190,6 +1392,20 @@ STBIDEF stbi_us *stbi_load_16(char const *filename, int *x, int *y, int *comp, i
#endif //!STBI_NO_STDIO
+STBIDEF stbi_us *stbi_load_16_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *channels_in_file, int desired_channels)
+{
+ stbi__context s;
+ stbi__start_mem(&s,buffer,len);
+ return stbi__load_and_postprocess_16bit(&s,x,y,channels_in_file,desired_channels);
+}
+
+STBIDEF stbi_us *stbi_load_16_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *channels_in_file, int desired_channels)
+{
+ stbi__context s;
+ stbi__start_callbacks(&s, (stbi_io_callbacks *)clbk, user);
+ return stbi__load_and_postprocess_16bit(&s,x,y,channels_in_file,desired_channels);
+}
+
STBIDEF stbi_uc *stbi_load_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp)
{
stbi__context s;
@@ -1204,6 +1420,22 @@ STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk, void *u
return stbi__load_and_postprocess_8bit(&s,x,y,comp,req_comp);
}
+#ifndef STBI_NO_GIF
+STBIDEF stbi_uc *stbi_load_gif_from_memory(stbi_uc const *buffer, int len, int **delays, int *x, int *y, int *z, int *comp, int req_comp)
+{
+ unsigned char *result;
+ stbi__context s;
+ stbi__start_mem(&s,buffer,len);
+
+ result = (unsigned char*) stbi__load_gif_main(&s, delays, x, y, z, comp, req_comp);
+ if (stbi__vertically_flip_on_load) {
+ stbi__vertical_flip_slices( result, *x, *y, *z, *comp );
+ }
+
+ return result;
+}
+#endif
+
#ifndef STBI_NO_LINEAR
static float *stbi__loadf_main(stbi__context *s, int *x, int *y, int *comp, int req_comp)
{
@@ -1287,12 +1519,16 @@ STBIDEF int stbi_is_hdr (char const *filename)
return result;
}
-STBIDEF int stbi_is_hdr_from_file(FILE *f)
+STBIDEF int stbi_is_hdr_from_file(FILE *f)
{
#ifndef STBI_NO_HDR
+ long pos = ftell(f);
+ int res;
stbi__context s;
stbi__start_file(&s,f);
- return stbi__hdr_test(&s);
+ res = stbi__hdr_test(&s);
+ fseek(f, pos, SEEK_SET);
+ return res;
#else
STBI_NOTUSED(f);
return 0;
@@ -1341,6 +1577,7 @@ enum
static void stbi__refill_buffer(stbi__context *s)
{
int n = (s->io.read)(s->io_user_data,(char*)s->buffer_start,s->buflen);
+ s->callback_already_read += (int) (s->img_buffer - s->img_buffer_original);
if (n == 0) {
// at end of file, treat same as if from memory, but need to handle case
// where s->img_buffer isn't pointing to safe memory, e.g. 0-byte file
@@ -1365,6 +1602,9 @@ stbi_inline static stbi_uc stbi__get8(stbi__context *s)
return 0;
}
+#if defined(STBI_NO_JPEG) && defined(STBI_NO_HDR) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM)
+// nothing
+#else
stbi_inline static int stbi__at_eof(stbi__context *s)
{
if (s->io.read) {
@@ -1376,9 +1616,14 @@ stbi_inline static int stbi__at_eof(stbi__context *s)
return s->img_buffer >= s->img_buffer_end;
}
+#endif
+#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC)
+// nothing
+#else
static void stbi__skip(stbi__context *s, int n)
{
+ if (n == 0) return; // already there!
if (n < 0) {
s->img_buffer = s->img_buffer_end;
return;
@@ -1393,7 +1638,11 @@ static void stbi__skip(stbi__context *s, int n)
}
s->img_buffer += n;
}
+#endif
+#if defined(STBI_NO_PNG) && defined(STBI_NO_TGA) && defined(STBI_NO_HDR) && defined(STBI_NO_PNM)
+// nothing
+#else
static int stbi__getn(stbi__context *s, stbi_uc *buffer, int n)
{
if (s->io.read) {
@@ -1417,18 +1666,27 @@ static int stbi__getn(stbi__context *s, stbi_uc *buffer, int n)
} else
return 0;
}
+#endif
+#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC)
+// nothing
+#else
static int stbi__get16be(stbi__context *s)
{
int z = stbi__get8(s);
return (z << 8) + stbi__get8(s);
}
+#endif
+#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD) && defined(STBI_NO_PIC)
+// nothing
+#else
static stbi__uint32 stbi__get32be(stbi__context *s)
{
stbi__uint32 z = stbi__get16be(s);
return (z << 16) + stbi__get16be(s);
}
+#endif
#if defined(STBI_NO_BMP) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF)
// nothing
@@ -1444,13 +1702,16 @@ static int stbi__get16le(stbi__context *s)
static stbi__uint32 stbi__get32le(stbi__context *s)
{
stbi__uint32 z = stbi__get16le(s);
- return z + (stbi__get16le(s) << 16);
+ z += (stbi__uint32)stbi__get16le(s) << 16;
+ return z;
}
#endif
#define STBI__BYTECAST(x) ((stbi_uc) ((x) & 255)) // truncate int to byte without warnings
-
+#if defined(STBI_NO_JPEG) && defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM)
+// nothing
+#else
//////////////////////////////////////////////////////////////////////////////
//
// generic converter from built-in img_n to req_comp
@@ -1466,7 +1727,11 @@ static stbi_uc stbi__compute_y(int r, int g, int b)
{
return (stbi_uc) (((r*77) + (g*150) + (29*b)) >> 8);
}
+#endif
+#if defined(STBI_NO_PNG) && defined(STBI_NO_BMP) && defined(STBI_NO_PSD) && defined(STBI_NO_TGA) && defined(STBI_NO_GIF) && defined(STBI_NO_PIC) && defined(STBI_NO_PNM)
+// nothing
+#else
static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int req_comp, unsigned int x, unsigned int y)
{
int i,j;
@@ -1490,19 +1755,19 @@ static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int r
// convert source image with img_n components to one with req_comp components;
// avoid switch per pixel, so use switch per scanline and massive macros
switch (STBI__COMBO(img_n, req_comp)) {
- STBI__CASE(1,2) { dest[0]=src[0], dest[1]=255; } break;
+ STBI__CASE(1,2) { dest[0]=src[0]; dest[1]=255; } break;
STBI__CASE(1,3) { dest[0]=dest[1]=dest[2]=src[0]; } break;
- STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0], dest[3]=255; } break;
+ STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=255; } break;
STBI__CASE(2,1) { dest[0]=src[0]; } break;
STBI__CASE(2,3) { dest[0]=dest[1]=dest[2]=src[0]; } break;
- STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0], dest[3]=src[1]; } break;
- STBI__CASE(3,4) { dest[0]=src[0],dest[1]=src[1],dest[2]=src[2],dest[3]=255; } break;
+ STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=src[1]; } break;
+ STBI__CASE(3,4) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2];dest[3]=255; } break;
STBI__CASE(3,1) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); } break;
- STBI__CASE(3,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]), dest[1] = 255; } break;
+ STBI__CASE(3,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); dest[1] = 255; } break;
STBI__CASE(4,1) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); } break;
- STBI__CASE(4,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]), dest[1] = src[3]; } break;
- STBI__CASE(4,3) { dest[0]=src[0],dest[1]=src[1],dest[2]=src[2]; } break;
- default: STBI_ASSERT(0);
+ STBI__CASE(4,2) { dest[0]=stbi__compute_y(src[0],src[1],src[2]); dest[1] = src[3]; } break;
+ STBI__CASE(4,3) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2]; } break;
+ default: STBI_ASSERT(0); STBI_FREE(data); STBI_FREE(good); return stbi__errpuc("unsupported", "Unsupported format conversion");
}
#undef STBI__CASE
}
@@ -1510,12 +1775,20 @@ static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int r
STBI_FREE(data);
return good;
}
+#endif
+#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD)
+// nothing
+#else
static stbi__uint16 stbi__compute_y_16(int r, int g, int b)
{
return (stbi__uint16) (((r*77) + (g*150) + (29*b)) >> 8);
}
+#endif
+#if defined(STBI_NO_PNG) && defined(STBI_NO_PSD)
+// nothing
+#else
static stbi__uint16 *stbi__convert_format16(stbi__uint16 *data, int img_n, int req_comp, unsigned int x, unsigned int y)
{
int i,j;
@@ -1539,19 +1812,19 @@ static stbi__uint16 *stbi__convert_format16(stbi__uint16 *data, int img_n, int r
// convert source image with img_n components to one with req_comp components;
// avoid switch per pixel, so use switch per scanline and massive macros
switch (STBI__COMBO(img_n, req_comp)) {
- STBI__CASE(1,2) { dest[0]=src[0], dest[1]=0xffff; } break;
+ STBI__CASE(1,2) { dest[0]=src[0]; dest[1]=0xffff; } break;
STBI__CASE(1,3) { dest[0]=dest[1]=dest[2]=src[0]; } break;
- STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0], dest[3]=0xffff; } break;
+ STBI__CASE(1,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=0xffff; } break;
STBI__CASE(2,1) { dest[0]=src[0]; } break;
STBI__CASE(2,3) { dest[0]=dest[1]=dest[2]=src[0]; } break;
- STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0], dest[3]=src[1]; } break;
- STBI__CASE(3,4) { dest[0]=src[0],dest[1]=src[1],dest[2]=src[2],dest[3]=0xffff; } break;
+ STBI__CASE(2,4) { dest[0]=dest[1]=dest[2]=src[0]; dest[3]=src[1]; } break;
+ STBI__CASE(3,4) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2];dest[3]=0xffff; } break;
STBI__CASE(3,1) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); } break;
- STBI__CASE(3,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]), dest[1] = 0xffff; } break;
+ STBI__CASE(3,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); dest[1] = 0xffff; } break;
STBI__CASE(4,1) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); } break;
- STBI__CASE(4,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]), dest[1] = src[3]; } break;
- STBI__CASE(4,3) { dest[0]=src[0],dest[1]=src[1],dest[2]=src[2]; } break;
- default: STBI_ASSERT(0);
+ STBI__CASE(4,2) { dest[0]=stbi__compute_y_16(src[0],src[1],src[2]); dest[1] = src[3]; } break;
+ STBI__CASE(4,3) { dest[0]=src[0];dest[1]=src[1];dest[2]=src[2]; } break;
+ default: STBI_ASSERT(0); STBI_FREE(data); STBI_FREE(good); return (stbi__uint16*) stbi__errpuc("unsupported", "Unsupported format conversion");
}
#undef STBI__CASE
}
@@ -1559,6 +1832,7 @@ static stbi__uint16 *stbi__convert_format16(stbi__uint16 *data, int img_n, int r
STBI_FREE(data);
return good;
}
+#endif
#ifndef STBI_NO_LINEAR
static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp)
@@ -1574,7 +1848,11 @@ static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp)
for (k=0; k < n; ++k) {
output[i*comp + k] = (float) (pow(data[i*comp+k]/255.0f, stbi__l2h_gamma) * stbi__l2h_scale);
}
- if (k < comp) output[i*comp + k] = data[i*comp+k]/255.0f;
+ }
+ if (n < comp) {
+ for (i=0; i < x*y; ++i) {
+ output[i*comp + n] = data[i*comp + n]/255.0f;
+ }
}
STBI_FREE(data);
return output;
@@ -1704,7 +1982,8 @@ typedef struct
static int stbi__build_huffman(stbi__huffman *h, int *count)
{
- int i,j,k=0,code;
+ int i,j,k=0;
+ unsigned int code;
// build size list for each symbol (from JPEG spec)
for (i=0; i < 16; ++i)
for (j=0; j < count[i]; ++j)
@@ -1720,7 +1999,7 @@ static int stbi__build_huffman(stbi__huffman *h, int *count)
if (h->size[k] == j) {
while (h->size[k] == j)
h->code[k++] = (stbi__uint16) (code++);
- if (code-1 >= (1 << j)) return stbi__err("bad code lengths","Corrupt JPEG");
+ if (code-1 >= (1u << j)) return stbi__err("bad code lengths","Corrupt JPEG");
}
// compute largest code + 1 for this size, preshifted as needed later
h->maxcode[j] = code << (16-j);
@@ -1764,7 +2043,7 @@ static void stbi__build_fast_ac(stbi__int16 *fast_ac, stbi__huffman *h)
if (k < m) k += (~0U << magbits) + 1;
// if the result is small enough, we can fit it in fast_ac table
if (k >= -128 && k <= 127)
- fast_ac[i] = (stbi__int16) ((k << 8) + (run << 4) + (len + magbits));
+ fast_ac[i] = (stbi__int16) ((k * 256) + (run * 16) + (len + magbits));
}
}
}
@@ -1773,7 +2052,7 @@ static void stbi__build_fast_ac(stbi__int16 *fast_ac, stbi__huffman *h)
static void stbi__grow_buffer_unsafe(stbi__jpeg *j)
{
do {
- int b = j->nomore ? 0 : stbi__get8(j->s);
+ unsigned int b = j->nomore ? 0 : stbi__get8(j->s);
if (b == 0xff) {
int c = stbi__get8(j->s);
while (c == 0xff) c = stbi__get8(j->s); // consume fill bytes
@@ -1789,7 +2068,7 @@ static void stbi__grow_buffer_unsafe(stbi__jpeg *j)
}
// (1 << n) - 1
-static stbi__uint32 stbi__bmask[17]={0,1,3,7,15,31,63,127,255,511,1023,2047,4095,8191,16383,32767,65535};
+static const stbi__uint32 stbi__bmask[17]={0,1,3,7,15,31,63,127,255,511,1023,2047,4095,8191,16383,32767,65535};
// decode a jpeg huffman value from the bitstream
stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h)
@@ -1842,7 +2121,7 @@ stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h)
}
// bias[n] = (-1<<n) + 1
-static int const stbi__jbias[16] = {0,-1,-3,-7,-15,-31,-63,-127,-255,-511,-1023,-2047,-4095,-8191,-16383,-32767};
+static const int stbi__jbias[16] = {0,-1,-3,-7,-15,-31,-63,-127,-255,-511,-1023,-2047,-4095,-8191,-16383,-32767};
// combined JPEG 'receive' and JPEG 'extend', since baseline
// always extends everything it receives.
@@ -1852,13 +2131,12 @@ stbi_inline static int stbi__extend_receive(stbi__jpeg *j, int n)
int sgn;
if (j->code_bits < n) stbi__grow_buffer_unsafe(j);
- sgn = (stbi__int32)j->code_buffer >> 31; // sign bit is always in MSB
+ sgn = j->code_buffer >> 31; // sign bit always in MSB; 0 if MSB clear (positive), 1 if MSB set (negative)
k = stbi_lrot(j->code_buffer, n);
- STBI_ASSERT(n >= 0 && n < (int) (sizeof(stbi__bmask)/sizeof(*stbi__bmask)));
j->code_buffer = k & ~stbi__bmask[n];
k &= stbi__bmask[n];
j->code_bits -= n;
- return k + (stbi__jbias[n] & ~sgn);
+ return k + (stbi__jbias[n] & (sgn - 1));
}
// get some unsigned bits
@@ -1885,7 +2163,7 @@ stbi_inline static int stbi__jpeg_get_bit(stbi__jpeg *j)
// given a value that's at position X in the zigzag stream,
// where does it appear in the 8x8 matrix coded as row-major?
-static stbi_uc stbi__jpeg_dezigzag[64+15] =
+static const stbi_uc stbi__jpeg_dezigzag[64+15] =
{
0, 1, 8, 16, 9, 2, 3, 10,
17, 24, 32, 25, 18, 11, 4, 5,
@@ -1908,7 +2186,7 @@ static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman
if (j->code_bits < 16) stbi__grow_buffer_unsafe(j);
t = stbi__jpeg_huff_decode(j, hdc);
- if (t < 0) return stbi__err("bad huffman code","Corrupt JPEG");
+ if (t < 0 || t > 15) return stbi__err("bad huffman code","Corrupt JPEG");
// 0 all the ac values now so we can do it 32-bits at a time
memset(data,0,64*sizeof(data[0]));
@@ -1965,11 +2243,12 @@ static int stbi__jpeg_decode_block_prog_dc(stbi__jpeg *j, short data[64], stbi__
// first scan for DC coefficient, must be first
memset(data,0,64*sizeof(data[0])); // 0 all the ac values now
t = stbi__jpeg_huff_decode(j, hdc);
+ if (t < 0 || t > 15) return stbi__err("can't merge dc and ac", "Corrupt JPEG");
diff = t ? stbi__extend_receive(j, t) : 0;
dc = j->img_comp[b].dc_pred + diff;
j->img_comp[b].dc_pred = dc;
- data[0] = (short) (dc << j->succ_low);
+ data[0] = (short) (dc * (1 << j->succ_low));
} else {
// refinement scan for DC coefficient
if (stbi__jpeg_get_bit(j))
@@ -2006,7 +2285,7 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__
j->code_buffer <<= s;
j->code_bits -= s;
zig = stbi__jpeg_dezigzag[k++];
- data[zig] = (short) ((r >> 8) << shift);
+ data[zig] = (short) ((r >> 8) * (1 << shift));
} else {
int rs = stbi__jpeg_huff_decode(j, hac);
if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG");
@@ -2024,7 +2303,7 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__
} else {
k += r;
zig = stbi__jpeg_dezigzag[k++];
- data[zig] = (short) (stbi__extend_receive(j,s) << shift);
+ data[zig] = (short) (stbi__extend_receive(j,s) * (1 << shift));
}
}
} while (k <= j->spec_end);
@@ -2111,7 +2390,7 @@ stbi_inline static stbi_uc stbi__clamp(int x)
}
#define stbi__f2f(x) ((int) (((x) * 4096 + 0.5)))
-#define stbi__fsh(x) ((x) << 12)
+#define stbi__fsh(x) ((x) * 4096)
// derived from jidctint -- DCT_ISLOW
#define STBI__IDCT_1D(s0,s1,s2,s3,s4,s5,s6,s7) \
@@ -2166,7 +2445,7 @@ static void stbi__idct_block(stbi_uc *out, int out_stride, short data[64])
// (1|2|3|4|5|6|7)==0 0 seconds
// all separate -0.047 seconds
// 1 && 2|3 && 4|5 && 6|7: -0.047 seconds
- int dcterm = d[0] << 2;
+ int dcterm = d[0]*4;
v[0] = v[8] = v[16] = v[24] = v[32] = v[40] = v[48] = v[56] = dcterm;
} else {
STBI__IDCT_1D(d[ 0],d[ 8],d[16],d[24],d[32],d[40],d[48],d[56])
@@ -2805,7 +3084,7 @@ static int stbi__process_marker(stbi__jpeg *z, int m)
if (t > 3) return stbi__err("bad DQT table","Corrupt JPEG");
for (i=0; i < 64; ++i)
- z->dequant[t][stbi__jpeg_dezigzag[i]] = sixteen ? stbi__get16be(z->s) : stbi__get8(z->s);
+ z->dequant[t][stbi__jpeg_dezigzag[i]] = (stbi__uint16)(sixteen ? stbi__get16be(z->s) : stbi__get8(z->s));
L -= (sixteen ? 129 : 65);
}
return L==0;
@@ -2955,6 +3234,8 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan)
p = stbi__get8(s); if (p != 8) return stbi__err("only 8-bit","JPEG format not supported: 8-bit only"); // JPEG baseline
s->img_y = stbi__get16be(s); if (s->img_y == 0) return stbi__err("no header height", "JPEG format not supported: delayed height"); // Legal, but we don't handle it--but neither does IJG
s->img_x = stbi__get16be(s); if (s->img_x == 0) return stbi__err("0 width","Corrupt JPEG"); // JPEG requires
+ if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+ if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
c = stbi__get8(s);
if (c != 3 && c != 1 && c != 4) return stbi__err("bad component count","Corrupt JPEG");
s->img_n = c;
@@ -2967,7 +3248,7 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan)
z->rgb = 0;
for (i=0; i < s->img_n; ++i) {
- static unsigned char rgb[3] = { 'R', 'G', 'B' };
+ static const unsigned char rgb[3] = { 'R', 'G', 'B' };
z->img_comp[i].id = stbi__get8(s);
if (s->img_n == 3 && z->img_comp[i].id == rgb[i])
++z->rgb;
@@ -2986,6 +3267,13 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan)
if (z->img_comp[i].v > v_max) v_max = z->img_comp[i].v;
}
+ // check that plane subsampling factors are integer ratios; our resamplers can't deal with fractional ratios
+ // and I've never seen a non-corrupted JPEG file actually use them
+ for (i=0; i < s->img_n; ++i) {
+ if (h_max % z->img_comp[i].h != 0) return stbi__err("bad H","Corrupt JPEG");
+ if (v_max % z->img_comp[i].v != 0) return stbi__err("bad V","Corrupt JPEG");
+ }
+
// compute interleaved mcu info
z->img_h_max = h_max;
z->img_v_max = v_max;
@@ -3092,8 +3380,8 @@ static int stbi__decode_jpeg_image(stbi__jpeg *j)
} else if (stbi__DNL(m)) {
int Ld = stbi__get16be(j->s);
stbi__uint32 NL = stbi__get16be(j->s);
- if (Ld != 4) stbi__err("bad DNL len", "Corrupt JPEG");
- if (NL != j->s->img_y) stbi__err("bad DNL height", "Corrupt JPEG");
+ if (Ld != 4) return stbi__err("bad DNL len", "Corrupt JPEG");
+ if (NL != j->s->img_y) return stbi__err("bad DNL height", "Corrupt JPEG");
} else {
if (!stbi__process_marker(j, m)) return 0;
}
@@ -3541,12 +3829,16 @@ static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp
else
decode_n = z->s->img_n;
+ // nothing to do if no components requested; check this now to avoid
+ // accessing uninitialized coutput[0] later
+ if (decode_n <= 0) { stbi__cleanup_jpeg(z); return NULL; }
+
// resample and color-convert
{
int k;
unsigned int i,j;
stbi_uc *output;
- stbi_uc *coutput[4];
+ stbi_uc *coutput[4] = { NULL, NULL, NULL, NULL };
stbi__resample res_comp[4];
@@ -3610,20 +3902,20 @@ static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp
} else if (z->s->img_n == 4) {
if (z->app14_color_transform == 0) { // CMYK
for (i=0; i < z->s->img_x; ++i) {
- stbi_uc k = coutput[3][i];
- out[0] = stbi__blinn_8x8(coutput[0][i], k);
- out[1] = stbi__blinn_8x8(coutput[1][i], k);
- out[2] = stbi__blinn_8x8(coutput[2][i], k);
+ stbi_uc m = coutput[3][i];
+ out[0] = stbi__blinn_8x8(coutput[0][i], m);
+ out[1] = stbi__blinn_8x8(coutput[1][i], m);
+ out[2] = stbi__blinn_8x8(coutput[2][i], m);
out[3] = 255;
out += n;
}
} else if (z->app14_color_transform == 2) { // YCCK
z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n);
for (i=0; i < z->s->img_x; ++i) {
- stbi_uc k = coutput[3][i];
- out[0] = stbi__blinn_8x8(255 - out[0], k);
- out[1] = stbi__blinn_8x8(255 - out[1], k);
- out[2] = stbi__blinn_8x8(255 - out[2], k);
+ stbi_uc m = coutput[3][i];
+ out[0] = stbi__blinn_8x8(255 - out[0], m);
+ out[1] = stbi__blinn_8x8(255 - out[1], m);
+ out[2] = stbi__blinn_8x8(255 - out[2], m);
out += n;
}
} else { // YCbCr + alpha? Ignore the fourth channel for now
@@ -3648,10 +3940,10 @@ static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp
}
} else if (z->s->img_n == 4 && z->app14_color_transform == 0) {
for (i=0; i < z->s->img_x; ++i) {
- stbi_uc k = coutput[3][i];
- stbi_uc r = stbi__blinn_8x8(coutput[0][i], k);
- stbi_uc g = stbi__blinn_8x8(coutput[1][i], k);
- stbi_uc b = stbi__blinn_8x8(coutput[2][i], k);
+ stbi_uc m = coutput[3][i];
+ stbi_uc r = stbi__blinn_8x8(coutput[0][i], m);
+ stbi_uc g = stbi__blinn_8x8(coutput[1][i], m);
+ stbi_uc b = stbi__blinn_8x8(coutput[2][i], m);
out[0] = stbi__compute_y(r, g, b);
out[1] = 255;
out += n;
@@ -3667,7 +3959,7 @@ static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp
if (n == 1)
for (i=0; i < z->s->img_x; ++i) out[i] = y[i];
else
- for (i=0; i < z->s->img_x; ++i) *out++ = y[i], *out++ = 255;
+ for (i=0; i < z->s->img_x; ++i) { *out++ = y[i]; *out++ = 255; }
}
}
}
@@ -3683,6 +3975,7 @@ static void *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int re
{
unsigned char* result;
stbi__jpeg* j = (stbi__jpeg*) stbi__malloc(sizeof(stbi__jpeg));
+ if (!j) return stbi__errpuc("outofmem", "Out of memory");
STBI_NOTUSED(ri);
j->s = s;
stbi__setup_jpeg(j);
@@ -3695,6 +3988,7 @@ static int stbi__jpeg_test(stbi__context *s)
{
int r;
stbi__jpeg* j = (stbi__jpeg*)stbi__malloc(sizeof(stbi__jpeg));
+ if (!j) return stbi__err("outofmem", "Out of memory");
j->s = s;
stbi__setup_jpeg(j);
r = stbi__decode_jpeg_header(j, STBI__SCAN_type);
@@ -3719,6 +4013,7 @@ static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp)
{
int result;
stbi__jpeg* j = (stbi__jpeg*) (stbi__malloc(sizeof(stbi__jpeg)));
+ if (!j) return stbi__err("outofmem", "Out of memory");
j->s = s;
result = stbi__jpeg_info_raw(j, x, y, comp);
STBI_FREE(j);
@@ -3738,6 +4033,7 @@ static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp)
// fast-way is faster to check than jpeg huffman, but slow way is slower
#define STBI__ZFAST_BITS 9 // accelerate all cases in default tables
#define STBI__ZFAST_MASK ((1 << STBI__ZFAST_BITS) - 1)
+#define STBI__ZNSYMS 288 // number of symbols in literal/length alphabet
// zlib-style huffman encoding
// (jpegs packs from left, zlib from right, so can't share code)
@@ -3747,8 +4043,8 @@ typedef struct
stbi__uint16 firstcode[16];
int maxcode[17];
stbi__uint16 firstsymbol[16];
- stbi_uc size[288];
- stbi__uint16 value[288];
+ stbi_uc size[STBI__ZNSYMS];
+ stbi__uint16 value[STBI__ZNSYMS];
} stbi__zhuffman;
stbi_inline static int stbi__bitreverse16(int n)
@@ -3835,16 +4131,23 @@ typedef struct
stbi__zhuffman z_length, z_distance;
} stbi__zbuf;
+stbi_inline static int stbi__zeof(stbi__zbuf *z)
+{
+ return (z->zbuffer >= z->zbuffer_end);
+}
+
stbi_inline static stbi_uc stbi__zget8(stbi__zbuf *z)
{
- if (z->zbuffer >= z->zbuffer_end) return 0;
- return *z->zbuffer++;
+ return stbi__zeof(z) ? 0 : *z->zbuffer++;
}
static void stbi__fill_bits(stbi__zbuf *z)
{
do {
- STBI_ASSERT(z->code_buffer < (1U << z->num_bits));
+ if (z->code_buffer >= (1U << z->num_bits)) {
+ z->zbuffer = z->zbuffer_end; /* treat this as EOF so we fail. */
+ return;
+ }
z->code_buffer |= (unsigned int) stbi__zget8(z) << z->num_bits;
z->num_bits += 8;
} while (z->num_bits <= 24);
@@ -3869,10 +4172,11 @@ static int stbi__zhuffman_decode_slowpath(stbi__zbuf *a, stbi__zhuffman *z)
for (s=STBI__ZFAST_BITS+1; ; ++s)
if (k < z->maxcode[s])
break;
- if (s == 16) return -1; // invalid code!
+ if (s >= 16) return -1; // invalid code!
// code size is s, so:
b = (k >> (16-s)) - z->firstcode[s] + z->firstsymbol[s];
- STBI_ASSERT(z->size[b] == s);
+ if (b >= STBI__ZNSYMS) return -1; // some data was corrupt somewhere!
+ if (z->size[b] != s) return -1; // was originally an assert, but report failure instead.
a->code_buffer >>= s;
a->num_bits -= s;
return z->value[b];
@@ -3881,7 +4185,12 @@ static int stbi__zhuffman_decode_slowpath(stbi__zbuf *a, stbi__zhuffman *z)
stbi_inline static int stbi__zhuffman_decode(stbi__zbuf *a, stbi__zhuffman *z)
{
int b,s;
- if (a->num_bits < 16) stbi__fill_bits(a);
+ if (a->num_bits < 16) {
+ if (stbi__zeof(a)) {
+ return -1; /* report error for unexpected end of data. */
+ }
+ stbi__fill_bits(a);
+ }
b = z->fast[a->code_buffer & STBI__ZFAST_MASK];
if (b) {
s = b >> 9;
@@ -3895,13 +4204,16 @@ stbi_inline static int stbi__zhuffman_decode(stbi__zbuf *a, stbi__zhuffman *z)
static int stbi__zexpand(stbi__zbuf *z, char *zout, int n) // need to make room for n bytes
{
char *q;
- int cur, limit, old_limit;
+ unsigned int cur, limit, old_limit;
z->zout = zout;
if (!z->z_expandable) return stbi__err("output buffer limit","Corrupt PNG");
- cur = (int) (z->zout - z->zout_start);
- limit = old_limit = (int) (z->zout_end - z->zout_start);
- while (cur + n > limit)
+ cur = (unsigned int) (z->zout - z->zout_start);
+ limit = old_limit = (unsigned) (z->zout_end - z->zout_start);
+ if (UINT_MAX - cur < (unsigned) n) return stbi__err("outofmem", "Out of memory");
+ while (cur + n > limit) {
+ if(limit > UINT_MAX / 2) return stbi__err("outofmem", "Out of memory");
limit *= 2;
+ }
q = (char *) STBI_REALLOC_SIZED(z->zout_start, old_limit, limit);
STBI_NOTUSED(old_limit);
if (q == NULL) return stbi__err("outofmem", "Out of memory");
@@ -3911,18 +4223,18 @@ static int stbi__zexpand(stbi__zbuf *z, char *zout, int n) // need to make room
return 1;
}
-static int stbi__zlength_base[31] = {
+static const int stbi__zlength_base[31] = {
3,4,5,6,7,8,9,10,11,13,
15,17,19,23,27,31,35,43,51,59,
67,83,99,115,131,163,195,227,258,0,0 };
-static int stbi__zlength_extra[31]=
+static const int stbi__zlength_extra[31]=
{ 0,0,0,0,0,0,0,0,1,1,1,1,2,2,2,2,3,3,3,3,4,4,4,4,5,5,5,5,0,0,0 };
-static int stbi__zdist_base[32] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,
+static const int stbi__zdist_base[32] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,
257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577,0,0};
-static int stbi__zdist_extra[32] =
+static const int stbi__zdist_extra[32] =
{ 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13};
static int stbi__parse_huffman_block(stbi__zbuf *a)
@@ -3969,7 +4281,7 @@ static int stbi__parse_huffman_block(stbi__zbuf *a)
static int stbi__compute_huffman_codes(stbi__zbuf *a)
{
- static stbi_uc length_dezigzag[19] = { 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 };
+ static const stbi_uc length_dezigzag[19] = { 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 };
stbi__zhuffman z_codelength;
stbi_uc lencodes[286+32+137];//padding for maximum single op
stbi_uc codelength_sizes[19];
@@ -3999,11 +4311,12 @@ static int stbi__compute_huffman_codes(stbi__zbuf *a)
c = stbi__zreceive(a,2)+3;
if (n == 0) return stbi__err("bad codelengths", "Corrupt PNG");
fill = lencodes[n-1];
- } else if (c == 17)
+ } else if (c == 17) {
c = stbi__zreceive(a,3)+3;
- else {
- STBI_ASSERT(c == 18);
+ } else if (c == 18) {
c = stbi__zreceive(a,7)+11;
+ } else {
+ return stbi__err("bad codelengths", "Corrupt PNG");
}
if (ntot - n < c) return stbi__err("bad codelengths", "Corrupt PNG");
memset(lencodes+n, fill, c);
@@ -4029,7 +4342,7 @@ static int stbi__parse_uncompressed_block(stbi__zbuf *a)
a->code_buffer >>= 8;
a->num_bits -= 8;
}
- STBI_ASSERT(a->num_bits == 0);
+ if (a->num_bits < 0) return stbi__err("zlib corrupt","Corrupt PNG");
// now fill header the normal way
while (k < 4)
header[k++] = stbi__zget8(a);
@@ -4051,6 +4364,7 @@ static int stbi__parse_zlib_header(stbi__zbuf *a)
int cm = cmf & 15;
/* int cinfo = cmf >> 4; */
int flg = stbi__zget8(a);
+ if (stbi__zeof(a)) return stbi__err("bad zlib header","Corrupt PNG"); // zlib spec
if ((cmf*256+flg) % 31 != 0) return stbi__err("bad zlib header","Corrupt PNG"); // zlib spec
if (flg & 32) return stbi__err("no preset dict","Corrupt PNG"); // preset dictionary not allowed in png
if (cm != 8) return stbi__err("bad compression","Corrupt PNG"); // DEFLATE required for png
@@ -4058,7 +4372,7 @@ static int stbi__parse_zlib_header(stbi__zbuf *a)
return 1;
}
-static const stbi_uc stbi__zdefault_length[288] =
+static const stbi_uc stbi__zdefault_length[STBI__ZNSYMS] =
{
8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8, 8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,8,
@@ -4104,7 +4418,7 @@ static int stbi__parse_zlib(stbi__zbuf *a, int parse_header)
} else {
if (type == 1) {
// use fixed code lengths
- if (!stbi__zbuild_huffman(&a->z_length , stbi__zdefault_length , 288)) return 0;
+ if (!stbi__zbuild_huffman(&a->z_length , stbi__zdefault_length , STBI__ZNSYMS)) return 0;
if (!stbi__zbuild_huffman(&a->z_distance, stbi__zdefault_distance, 32)) return 0;
} else {
if (!stbi__compute_huffman_codes(a)) return 0;
@@ -4228,7 +4542,7 @@ static stbi__pngchunk stbi__get_chunk_header(stbi__context *s)
static int stbi__check_png_header(stbi__context *s)
{
- static stbi_uc png_sig[8] = { 137,80,78,71,13,10,26,10 };
+ static const stbi_uc png_sig[8] = { 137,80,78,71,13,10,26,10 };
int i;
for (i=0; i < 8; ++i)
if (stbi__get8(s) != png_sig[i]) return stbi__err("bad png sig","Not a PNG");
@@ -4274,7 +4588,7 @@ static int stbi__paeth(int a, int b, int c)
return c;
}
-static stbi_uc stbi__depth_scale_table[9] = { 0, 0xff, 0x55, 0, 0x11, 0,0,0, 0x01 };
+static const stbi_uc stbi__depth_scale_table[9] = { 0, 0xff, 0x55, 0, 0x11, 0,0,0, 0x01 };
// create the png data from post-deflated data
static int stbi__create_png_image_raw(stbi__png *a, stbi_uc *raw, stbi__uint32 raw_len, int out_n, stbi__uint32 x, stbi__uint32 y, int depth, int color)
@@ -4294,13 +4608,14 @@ static int stbi__create_png_image_raw(stbi__png *a, stbi_uc *raw, stbi__uint32 r
a->out = (stbi_uc *) stbi__malloc_mad3(x, y, output_bytes, 0); // extra bytes to write off the end into
if (!a->out) return stbi__err("outofmem", "Out of memory");
+ if (!stbi__mad3sizes_valid(img_n, x, depth, 7)) return stbi__err("too large", "Corrupt PNG");
img_width_bytes = (((img_n * x * depth) + 7) >> 3);
img_len = (img_width_bytes + 1) * y;
- if (s->img_x == x && s->img_y == y) {
- if (raw_len != img_len) return stbi__err("not enough pixels","Corrupt PNG");
- } else { // interlaced:
- if (raw_len < img_len) return stbi__err("not enough pixels","Corrupt PNG");
- }
+
+ // we used to check for exact match between raw_len and img_len on non-interlaced PNGs,
+ // but issue #276 reported a PNG in the wild that had extra data at the end (all zeros),
+ // so just check for raw_len < img_len always.
+ if (raw_len < img_len) return stbi__err("not enough pixels","Corrupt PNG");
for (j=0; j < y; ++j) {
stbi_uc *cur = a->out + stride*j;
@@ -4311,7 +4626,7 @@ static int stbi__create_png_image_raw(stbi__png *a, stbi_uc *raw, stbi__uint32 r
return stbi__err("invalid filter","Corrupt PNG");
if (depth < 8) {
- STBI_ASSERT(img_width_bytes <= x);
+ if (img_width_bytes > x) return stbi__err("invalid width","Corrupt PNG");
cur += x*out_n - img_width_bytes; // store output to the rightmost img_len bytes, so we can decode in place
filter_bytes = 1;
width = img_width_bytes;
@@ -4499,6 +4814,7 @@ static int stbi__create_png_image(stbi__png *a, stbi_uc *image_data, stbi__uint3
// de-interlacing
final = (stbi_uc *) stbi__malloc_mad3(a->s->img_x, a->s->img_y, out_bytes, 0);
+ if (!final) return stbi__err("outofmem", "Out of memory");
for (p=0; p < 7; ++p) {
int xorig[] = { 0,4,0,2,0,1,0 };
int yorig[] = { 0,0,4,0,2,0,1 };
@@ -4619,19 +4935,46 @@ static int stbi__expand_png_palette(stbi__png *a, stbi_uc *palette, int len, int
return 1;
}
-static int stbi__unpremultiply_on_load = 0;
-static int stbi__de_iphone_flag = 0;
+static int stbi__unpremultiply_on_load_global = 0;
+static int stbi__de_iphone_flag_global = 0;
STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply)
{
- stbi__unpremultiply_on_load = flag_true_if_should_unpremultiply;
+ stbi__unpremultiply_on_load_global = flag_true_if_should_unpremultiply;
}
STBIDEF void stbi_convert_iphone_png_to_rgb(int flag_true_if_should_convert)
{
- stbi__de_iphone_flag = flag_true_if_should_convert;
+ stbi__de_iphone_flag_global = flag_true_if_should_convert;
+}
+
+#ifndef STBI_THREAD_LOCAL
+#define stbi__unpremultiply_on_load stbi__unpremultiply_on_load_global
+#define stbi__de_iphone_flag stbi__de_iphone_flag_global
+#else
+static STBI_THREAD_LOCAL int stbi__unpremultiply_on_load_local, stbi__unpremultiply_on_load_set;
+static STBI_THREAD_LOCAL int stbi__de_iphone_flag_local, stbi__de_iphone_flag_set;
+
+STBIDEF void stbi__unpremultiply_on_load_thread(int flag_true_if_should_unpremultiply)
+{
+ stbi__unpremultiply_on_load_local = flag_true_if_should_unpremultiply;
+ stbi__unpremultiply_on_load_set = 1;
}
+STBIDEF void stbi_convert_iphone_png_to_rgb_thread(int flag_true_if_should_convert)
+{
+ stbi__de_iphone_flag_local = flag_true_if_should_convert;
+ stbi__de_iphone_flag_set = 1;
+}
+
+#define stbi__unpremultiply_on_load (stbi__unpremultiply_on_load_set \
+ ? stbi__unpremultiply_on_load_local \
+ : stbi__unpremultiply_on_load_global)
+#define stbi__de_iphone_flag (stbi__de_iphone_flag_set \
+ ? stbi__de_iphone_flag_local \
+ : stbi__de_iphone_flag_global)
+#endif // STBI_THREAD_LOCAL
+
static void stbi__de_iphone(stbi__png *z)
{
stbi__context *s = z->s;
@@ -4653,9 +4996,10 @@ static void stbi__de_iphone(stbi__png *z)
stbi_uc a = p[3];
stbi_uc t = p[0];
if (a) {
- p[0] = p[2] * 255 / a;
- p[1] = p[1] * 255 / a;
- p[2] = t * 255 / a;
+ stbi_uc half = a / 2;
+ p[0] = (p[2] * 255 + half) / a;
+ p[1] = (p[1] * 255 + half) / a;
+ p[2] = ( t * 255 + half) / a;
} else {
p[0] = p[2];
p[2] = t;
@@ -4674,12 +5018,12 @@ static void stbi__de_iphone(stbi__png *z)
}
}
-#define STBI__PNG_TYPE(a,b,c,d) (((a) << 24) + ((b) << 16) + ((c) << 8) + (d))
+#define STBI__PNG_TYPE(a,b,c,d) (((unsigned) (a) << 24) + ((unsigned) (b) << 16) + ((unsigned) (c) << 8) + (unsigned) (d))
static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
{
stbi_uc palette[1024], pal_img_n=0;
- stbi_uc has_trans=0, tc[3];
+ stbi_uc has_trans=0, tc[3]={0};
stbi__uint16 tc16[3];
stbi__uint32 ioff=0, idata_limit=0, i, pal_len=0;
int first=1,k,interlace=0, color=0, is_iphone=0;
@@ -4705,8 +5049,10 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
if (!first) return stbi__err("multiple IHDR","Corrupt PNG");
first = 0;
if (c.length != 13) return stbi__err("bad IHDR len","Corrupt PNG");
- s->img_x = stbi__get32be(s); if (s->img_x > (1 << 24)) return stbi__err("too large","Very large image (corrupt?)");
- s->img_y = stbi__get32be(s); if (s->img_y > (1 << 24)) return stbi__err("too large","Very large image (corrupt?)");
+ s->img_x = stbi__get32be(s);
+ s->img_y = stbi__get32be(s);
+ if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+ if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
z->depth = stbi__get8(s); if (z->depth != 1 && z->depth != 2 && z->depth != 4 && z->depth != 8 && z->depth != 16) return stbi__err("1/2/4/8/16-bit only","PNG not supported: 1/2/4/8/16-bit only");
color = stbi__get8(s); if (color > 6) return stbi__err("bad ctype","Corrupt PNG");
if (color == 3 && z->depth == 16) return stbi__err("bad ctype","Corrupt PNG");
@@ -4818,8 +5164,13 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp)
if (req_comp >= 3) s->img_out_n = req_comp;
if (!stbi__expand_png_palette(z, palette, pal_len, s->img_out_n))
return 0;
+ } else if (has_trans) {
+ // non-paletted image with tRNS -> source image has (constant) alpha
+ ++s->img_n;
}
STBI_FREE(z->expanded); z->expanded = NULL;
+ // end of PNG chunk, read and skip CRC
+ stbi__get32be(s);
return 1;
}
@@ -4850,10 +5201,12 @@ static void *stbi__do_png(stbi__png *p, int *x, int *y, int *n, int req_comp, st
void *result=NULL;
if (req_comp < 0 || req_comp > 4) return stbi__errpuc("bad req_comp", "Internal error");
if (stbi__parse_png_file(p, STBI__SCAN_load, req_comp)) {
- if (p->depth < 8)
+ if (p->depth <= 8)
ri->bits_per_channel = 8;
+ else if (p->depth == 16)
+ ri->bits_per_channel = 16;
else
- ri->bits_per_channel = p->depth;
+ return stbi__errpuc("bad bits_per_channel", "PNG not supported: unsupported color depth");
result = p->out;
p->out = NULL;
if (req_comp && req_comp != p->s->img_out_n) {
@@ -4908,6 +5261,19 @@ static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp)
p.s = s;
return stbi__png_info_raw(&p, x, y, comp);
}
+
+static int stbi__png_is16(stbi__context *s)
+{
+ stbi__png p;
+ p.s = s;
+ if (!stbi__png_info_raw(&p, NULL, NULL, NULL))
+ return 0;
+ if (p.depth != 16) {
+ stbi__rewind(p.s);
+ return 0;
+ }
+ return 1;
+}
#endif
// Microsoft/Windows BMP image
@@ -4941,11 +5307,11 @@ static int stbi__high_bit(unsigned int z)
{
int n=0;
if (z == 0) return -1;
- if (z >= 0x10000) n += 16, z >>= 16;
- if (z >= 0x00100) n += 8, z >>= 8;
- if (z >= 0x00010) n += 4, z >>= 4;
- if (z >= 0x00004) n += 2, z >>= 2;
- if (z >= 0x00002) n += 1, z >>= 1;
+ if (z >= 0x10000) { n += 16; z >>= 16; }
+ if (z >= 0x00100) { n += 8; z >>= 8; }
+ if (z >= 0x00010) { n += 4; z >>= 4; }
+ if (z >= 0x00004) { n += 2; z >>= 2; }
+ if (z >= 0x00002) { n += 1;/* >>= 1;*/ }
return n;
}
@@ -4959,29 +5325,62 @@ static int stbi__bitcount(unsigned int a)
return a & 0xff;
}
-static int stbi__shiftsigned(int v, int shift, int bits)
-{
- int result;
- int z=0;
-
- if (shift < 0) v <<= -shift;
- else v >>= shift;
- result = v;
-
- z = bits;
- while (z < 8) {
- result += v >> z;
- z += bits;
- }
- return result;
+// extract an arbitrarily-aligned N-bit value (N=bits)
+// from v, and then make it 8-bits long and fractionally
+// extend it to full full range.
+static int stbi__shiftsigned(unsigned int v, int shift, int bits)
+{
+ static unsigned int mul_table[9] = {
+ 0,
+ 0xff/*0b11111111*/, 0x55/*0b01010101*/, 0x49/*0b01001001*/, 0x11/*0b00010001*/,
+ 0x21/*0b00100001*/, 0x41/*0b01000001*/, 0x81/*0b10000001*/, 0x01/*0b00000001*/,
+ };
+ static unsigned int shift_table[9] = {
+ 0, 0,0,1,0,2,4,6,0,
+ };
+ if (shift < 0)
+ v <<= -shift;
+ else
+ v >>= shift;
+ STBI_ASSERT(v < 256);
+ v >>= (8-bits);
+ STBI_ASSERT(bits >= 0 && bits <= 8);
+ return (int) ((unsigned) v * mul_table[bits]) >> shift_table[bits];
}
typedef struct
{
int bpp, offset, hsz;
unsigned int mr,mg,mb,ma, all_a;
+ int extra_read;
} stbi__bmp_data;
+static int stbi__bmp_set_mask_defaults(stbi__bmp_data *info, int compress)
+{
+ // BI_BITFIELDS specifies masks explicitly, don't override
+ if (compress == 3)
+ return 1;
+
+ if (compress == 0) {
+ if (info->bpp == 16) {
+ info->mr = 31u << 10;
+ info->mg = 31u << 5;
+ info->mb = 31u << 0;
+ } else if (info->bpp == 32) {
+ info->mr = 0xffu << 16;
+ info->mg = 0xffu << 8;
+ info->mb = 0xffu << 0;
+ info->ma = 0xffu << 24;
+ info->all_a = 0; // if all_a is 0 at end, then we loaded alpha channel but it was all 0
+ } else {
+ // otherwise, use defaults, which is all-0
+ info->mr = info->mg = info->mb = info->ma = 0;
+ }
+ return 1;
+ }
+ return 0; // error
+}
+
static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info)
{
int hsz;
@@ -4992,6 +5391,9 @@ static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info)
info->offset = stbi__get32le(s);
info->hsz = hsz = stbi__get32le(s);
info->mr = info->mg = info->mb = info->ma = 0;
+ info->extra_read = 14;
+
+ if (info->offset < 0) return stbi__errpuc("bad BMP", "bad BMP");
if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) return stbi__errpuc("unknown BMP", "BMP type not supported: unknown");
if (hsz == 12) {
@@ -5003,10 +5405,11 @@ static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info)
}
if (stbi__get16le(s) != 1) return stbi__errpuc("bad BMP", "bad BMP");
info->bpp = stbi__get16le(s);
- if (info->bpp == 1) return stbi__errpuc("monochrome", "BMP type not supported: 1-bit");
if (hsz != 12) {
int compress = stbi__get32le(s);
if (compress == 1 || compress == 2) return stbi__errpuc("BMP RLE", "BMP type not supported: RLE");
+ if (compress >= 4) return stbi__errpuc("BMP JPEG/PNG", "BMP type not supported: unsupported compression"); // this includes PNG/JPEG modes
+ if (compress == 3 && info->bpp != 16 && info->bpp != 32) return stbi__errpuc("bad BMP", "bad BMP"); // bitfields requires 16 or 32 bits/pixel
stbi__get32le(s); // discard sizeof
stbi__get32le(s); // discard hres
stbi__get32le(s); // discard vres
@@ -5021,21 +5424,12 @@ static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info)
}
if (info->bpp == 16 || info->bpp == 32) {
if (compress == 0) {
- if (info->bpp == 32) {
- info->mr = 0xffu << 16;
- info->mg = 0xffu << 8;
- info->mb = 0xffu << 0;
- info->ma = 0xffu << 24;
- info->all_a = 0; // if all_a is 0 at end, then we loaded alpha channel but it was all 0
- } else {
- info->mr = 31u << 10;
- info->mg = 31u << 5;
- info->mb = 31u << 0;
- }
+ stbi__bmp_set_mask_defaults(info, compress);
} else if (compress == 3) {
info->mr = stbi__get32le(s);
info->mg = stbi__get32le(s);
info->mb = stbi__get32le(s);
+ info->extra_read += 12;
// not documented, but generated by photoshop and handled by mspaint
if (info->mr == info->mg && info->mg == info->mb) {
// ?!?!?
@@ -5045,6 +5439,7 @@ static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info)
return stbi__errpuc("bad BMP", "bad BMP");
}
} else {
+ // V4/V5 header
int i;
if (hsz != 108 && hsz != 124)
return stbi__errpuc("bad BMP", "bad BMP");
@@ -5052,6 +5447,8 @@ static void *stbi__bmp_parse_header(stbi__context *s, stbi__bmp_data *info)
info->mg = stbi__get32le(s);
info->mb = stbi__get32le(s);
info->ma = stbi__get32le(s);
+ if (compress != 3) // override mr/mg/mb unless in BI_BITFIELDS mode, as per docs
+ stbi__bmp_set_mask_defaults(info, compress);
stbi__get32le(s); // discard color space
for (i=0; i < 12; ++i)
stbi__get32le(s); // discard color space parameters
@@ -5084,6 +5481,9 @@ static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req
flip_vertically = ((int) s->img_y) > 0;
s->img_y = abs((int) s->img_y);
+ if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+
mr = info.mr;
mg = info.mg;
mb = info.mb;
@@ -5092,13 +5492,21 @@ static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req
if (info.hsz == 12) {
if (info.bpp < 24)
- psize = (info.offset - 14 - 24) / 3;
+ psize = (info.offset - info.extra_read - 24) / 3;
} else {
if (info.bpp < 16)
- psize = (info.offset - 14 - info.hsz) >> 2;
+ psize = (info.offset - info.extra_read - info.hsz) >> 2;
+ }
+ if (psize == 0) {
+ if (info.offset != s->callback_already_read + (s->img_buffer - s->img_buffer_original)) {
+ return stbi__errpuc("bad offset", "Corrupt BMP");
+ }
}
- s->img_n = ma ? 4 : 3;
+ if (info.bpp == 24 && ma == 0xff000000)
+ s->img_n = 3;
+ else
+ s->img_n = ma ? 4 : 3;
if (req_comp && req_comp >= 3) // we can directly decode 3 or 4
target = req_comp;
else
@@ -5120,36 +5528,56 @@ static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req
if (info.hsz != 12) stbi__get8(s);
pal[i][3] = 255;
}
- stbi__skip(s, info.offset - 14 - info.hsz - psize * (info.hsz == 12 ? 3 : 4));
- if (info.bpp == 4) width = (s->img_x + 1) >> 1;
+ stbi__skip(s, info.offset - info.extra_read - info.hsz - psize * (info.hsz == 12 ? 3 : 4));
+ if (info.bpp == 1) width = (s->img_x + 7) >> 3;
+ else if (info.bpp == 4) width = (s->img_x + 1) >> 1;
else if (info.bpp == 8) width = s->img_x;
else { STBI_FREE(out); return stbi__errpuc("bad bpp", "Corrupt BMP"); }
pad = (-width)&3;
- for (j=0; j < (int) s->img_y; ++j) {
- for (i=0; i < (int) s->img_x; i += 2) {
- int v=stbi__get8(s),v2=0;
- if (info.bpp == 4) {
- v2 = v & 15;
- v >>= 4;
+ if (info.bpp == 1) {
+ for (j=0; j < (int) s->img_y; ++j) {
+ int bit_offset = 7, v = stbi__get8(s);
+ for (i=0; i < (int) s->img_x; ++i) {
+ int color = (v>>bit_offset)&0x1;
+ out[z++] = pal[color][0];
+ out[z++] = pal[color][1];
+ out[z++] = pal[color][2];
+ if (target == 4) out[z++] = 255;
+ if (i+1 == (int) s->img_x) break;
+ if((--bit_offset) < 0) {
+ bit_offset = 7;
+ v = stbi__get8(s);
+ }
}
- out[z++] = pal[v][0];
- out[z++] = pal[v][1];
- out[z++] = pal[v][2];
- if (target == 4) out[z++] = 255;
- if (i+1 == (int) s->img_x) break;
- v = (info.bpp == 8) ? stbi__get8(s) : v2;
- out[z++] = pal[v][0];
- out[z++] = pal[v][1];
- out[z++] = pal[v][2];
- if (target == 4) out[z++] = 255;
+ stbi__skip(s, pad);
+ }
+ } else {
+ for (j=0; j < (int) s->img_y; ++j) {
+ for (i=0; i < (int) s->img_x; i += 2) {
+ int v=stbi__get8(s),v2=0;
+ if (info.bpp == 4) {
+ v2 = v & 15;
+ v >>= 4;
+ }
+ out[z++] = pal[v][0];
+ out[z++] = pal[v][1];
+ out[z++] = pal[v][2];
+ if (target == 4) out[z++] = 255;
+ if (i+1 == (int) s->img_x) break;
+ v = (info.bpp == 8) ? stbi__get8(s) : v2;
+ out[z++] = pal[v][0];
+ out[z++] = pal[v][1];
+ out[z++] = pal[v][2];
+ if (target == 4) out[z++] = 255;
+ }
+ stbi__skip(s, pad);
}
- stbi__skip(s, pad);
}
} else {
int rshift=0,gshift=0,bshift=0,ashift=0,rcount=0,gcount=0,bcount=0,acount=0;
int z = 0;
int easy=0;
- stbi__skip(s, info.offset - 14 - info.hsz);
+ stbi__skip(s, info.offset - info.extra_read - info.hsz);
if (info.bpp == 24) width = 3 * s->img_x;
else if (info.bpp == 16) width = 2*s->img_x;
else /* bpp = 32 and pad = 0 */ width=0;
@@ -5167,6 +5595,7 @@ static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req
gshift = stbi__high_bit(mg)-7; gcount = stbi__bitcount(mg);
bshift = stbi__high_bit(mb)-7; bcount = stbi__bitcount(mb);
ashift = stbi__high_bit(ma)-7; acount = stbi__bitcount(ma);
+ if (rcount > 8 || gcount > 8 || bcount > 8 || acount > 8) { STBI_FREE(out); return stbi__errpuc("bad masks", "Corrupt BMP"); }
}
for (j=0; j < (int) s->img_y; ++j) {
if (easy) {
@@ -5184,7 +5613,7 @@ static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req
int bpp = info.bpp;
for (i=0; i < (int) s->img_x; ++i) {
stbi__uint32 v = (bpp == 16 ? (stbi__uint32) stbi__get16le(s) : stbi__get32le(s));
- int a;
+ unsigned int a;
out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mr, rshift, rcount));
out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mg, gshift, gcount));
out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mb, bshift, bcount));
@@ -5208,7 +5637,7 @@ static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req
stbi_uc *p1 = out + j *s->img_x*target;
stbi_uc *p2 = out + (s->img_y-1-j)*s->img_x*target;
for (i=0; i < (int) s->img_x*target; ++i) {
- t = p1[i], p1[i] = p2[i], p2[i] = t;
+ t = p1[i]; p1[i] = p2[i]; p2[i] = t;
}
}
}
@@ -5232,14 +5661,14 @@ static void *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req
static int stbi__tga_get_comp(int bits_per_pixel, int is_grey, int* is_rgb16)
{
// only RGB or RGBA (incl. 16bit) or grey allowed
- if(is_rgb16) *is_rgb16 = 0;
+ if (is_rgb16) *is_rgb16 = 0;
switch(bits_per_pixel) {
case 8: return STBI_grey;
case 16: if(is_grey) return STBI_grey_alpha;
- // else: fall-through
+ // fallthrough
case 15: if(is_rgb16) *is_rgb16 = 1;
- return STBI_rgb;
- case 24: // fall-through
+ return STBI_rgb;
+ case 24: // fallthrough
case 32: return bits_per_pixel/8;
default: return 0;
}
@@ -5388,6 +5817,11 @@ static void *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req
int RLE_repeating = 0;
int read_next_pixel = 1;
STBI_NOTUSED(ri);
+ STBI_NOTUSED(tga_x_origin); // @TODO
+ STBI_NOTUSED(tga_y_origin); // @TODO
+
+ if (tga_height > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (tga_width > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
// do a tiny bit of precessing
if ( tga_image_type >= 8 )
@@ -5428,6 +5862,11 @@ static void *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req
// do I need to load a palette?
if ( tga_indexed)
{
+ if (tga_palette_len == 0) { /* you have to have at least one entry! */
+ STBI_FREE(tga_data);
+ return stbi__errpuc("bad palette", "Corrupt TGA");
+ }
+
// any data to skip? (offset usually = 0)
stbi__skip(s, tga_palette_start );
// load the palette
@@ -5551,6 +5990,7 @@ static void *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req
// Microsoft's C compilers happy... [8^(
tga_palette_start = tga_palette_len = tga_palette_bits =
tga_x_origin = tga_y_origin = 0;
+ STBI_NOTUSED(tga_palette_start);
// OK, done
return tga_data;
}
@@ -5635,6 +6075,9 @@ static void *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req
h = stbi__get32be(s);
w = stbi__get32be(s);
+ if (h > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (w > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+
// Make sure the depth is 8 bits.
bitdepth = stbi__get16be(s);
if (bitdepth != 8 && bitdepth != 16)
@@ -5698,7 +6141,7 @@ static void *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req
// Else if n is 128, noop.
// Endloop
- // The RLE-compressed data is preceeded by a 2-byte data count for each row in the data,
+ // The RLE-compressed data is preceded by a 2-byte data count for each row in the data,
// which we're going to just skip.
stbi__skip(s, h * channelCount * 2 );
@@ -5989,6 +6432,10 @@ static void *stbi__pic_load(stbi__context *s,int *px,int *py,int *comp,int req_c
x = stbi__get16be(s);
y = stbi__get16be(s);
+
+ if (y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+
if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (pic header)");
if (!stbi__mad3sizes_valid(x, y, 4, 0)) return stbi__errpuc("too large", "PIC image too large to decode");
@@ -5998,6 +6445,7 @@ static void *stbi__pic_load(stbi__context *s,int *px,int *py,int *comp,int req_c
// intermediate buffer is RGBA
result = (stbi_uc *) stbi__malloc_mad3(x, y, 4, 0);
+ if (!result) return stbi__errpuc("outofmem", "Out of memory");
memset(result, 0xff, x*y*4);
if (!stbi__pic_load_core(s,x,y,comp, result)) {
@@ -6034,11 +6482,13 @@ typedef struct
typedef struct
{
int w,h;
- stbi_uc *out, *old_out; // output buffer (always 4 components)
- int flags, bgindex, ratio, transparent, eflags, delay;
+ stbi_uc *out; // output buffer (always 4 components)
+ stbi_uc *background; // The current "background" as far as a gif is concerned
+ stbi_uc *history;
+ int flags, bgindex, ratio, transparent, eflags;
stbi_uc pal[256][4];
stbi_uc lpal[256][4];
- stbi__gif_lzw codes[4096];
+ stbi__gif_lzw codes[8192];
stbi_uc *color_table;
int parse, step;
int lflags;
@@ -6046,6 +6496,7 @@ typedef struct
int max_x, max_y;
int cur_x, cur_y;
int line_size;
+ int delay;
} stbi__gif;
static int stbi__gif_test_raw(stbi__context *s)
@@ -6094,6 +6545,9 @@ static int stbi__gif_header(stbi__context *s, stbi__gif *g, int *comp, int is_in
g->ratio = stbi__get8(s);
g->transparent = -1;
+ if (g->w > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+ if (g->h > STBI_MAX_DIMENSIONS) return stbi__err("too large","Very large image (corrupt?)");
+
if (comp != 0) *comp = 4; // can't actually tell whether it's 3 or 4 until we parse the comments
if (is_info) return 1;
@@ -6107,6 +6561,7 @@ static int stbi__gif_header(stbi__context *s, stbi__gif *g, int *comp, int is_in
static int stbi__gif_info_raw(stbi__context *s, int *x, int *y, int *comp)
{
stbi__gif* g = (stbi__gif*) stbi__malloc(sizeof(stbi__gif));
+ if (!g) return stbi__err("outofmem", "Out of memory");
if (!stbi__gif_header(s, g, comp, 1)) {
STBI_FREE(g);
stbi__rewind( s );
@@ -6121,6 +6576,7 @@ static int stbi__gif_info_raw(stbi__context *s, int *x, int *y, int *comp)
static void stbi__out_gif_code(stbi__gif *g, stbi__uint16 code)
{
stbi_uc *p, *c;
+ int idx;
// recurse to decode the prefixes, since the linked-list is backwards,
// and working backwards through an interleaved image would be nasty
@@ -6129,10 +6585,12 @@ static void stbi__out_gif_code(stbi__gif *g, stbi__uint16 code)
if (g->cur_y >= g->max_y) return;
- p = &g->out[g->cur_x + g->cur_y];
- c = &g->color_table[g->codes[code].suffix * 4];
+ idx = g->cur_x + g->cur_y;
+ p = &g->out[idx];
+ g->history[idx / 4] = 1;
- if (c[3] >= 128) {
+ c = &g->color_table[g->codes[code].suffix * 4];
+ if (c[3] > 128) { // don't render transparent pixels;
p[0] = c[2];
p[1] = c[1];
p[2] = c[0];
@@ -6206,11 +6664,16 @@ static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g)
stbi__skip(s,len);
return g->out;
} else if (code <= avail) {
- if (first) return stbi__errpuc("no clear code", "Corrupt GIF");
+ if (first) {
+ return stbi__errpuc("no clear code", "Corrupt GIF");
+ }
if (oldcode >= 0) {
p = &g->codes[avail++];
- if (avail > 4096) return stbi__errpuc("too many codes", "Corrupt GIF");
+ if (avail > 8192) {
+ return stbi__errpuc("too many codes", "Corrupt GIF");
+ }
+
p->prefix = (stbi__int16) oldcode;
p->first = g->codes[oldcode].first;
p->suffix = (code == avail) ? p->first : g->codes[code].first;
@@ -6232,62 +6695,77 @@ static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g)
}
}
-static void stbi__fill_gif_background(stbi__gif *g, int x0, int y0, int x1, int y1)
-{
- int x, y;
- stbi_uc *c = g->pal[g->bgindex];
- for (y = y0; y < y1; y += 4 * g->w) {
- for (x = x0; x < x1; x += 4) {
- stbi_uc *p = &g->out[y + x];
- p[0] = c[2];
- p[1] = c[1];
- p[2] = c[0];
- p[3] = 0;
- }
- }
-}
-
// this function is designed to support animated gifs, although stb_image doesn't support it
-static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, int req_comp)
+// two back is the image from two frames ago, used for a very specific disposal format
+static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, int req_comp, stbi_uc *two_back)
{
- int i;
- stbi_uc *prev_out = 0;
-
- if (g->out == 0 && !stbi__gif_header(s, g, comp,0))
- return 0; // stbi__g_failure_reason set by stbi__gif_header
+ int dispose;
+ int first_frame;
+ int pi;
+ int pcount;
+ STBI_NOTUSED(req_comp);
- if (!stbi__mad3sizes_valid(g->w, g->h, 4, 0))
- return stbi__errpuc("too large", "GIF too large");
+ // on first frame, any non-written pixels get the background colour (non-transparent)
+ first_frame = 0;
+ if (g->out == 0) {
+ if (!stbi__gif_header(s, g, comp,0)) return 0; // stbi__g_failure_reason set by stbi__gif_header
+ if (!stbi__mad3sizes_valid(4, g->w, g->h, 0))
+ return stbi__errpuc("too large", "GIF image is too large");
+ pcount = g->w * g->h;
+ g->out = (stbi_uc *) stbi__malloc(4 * pcount);
+ g->background = (stbi_uc *) stbi__malloc(4 * pcount);
+ g->history = (stbi_uc *) stbi__malloc(pcount);
+ if (!g->out || !g->background || !g->history)
+ return stbi__errpuc("outofmem", "Out of memory");
+
+ // image is treated as "transparent" at the start - ie, nothing overwrites the current background;
+ // background colour is only used for pixels that are not rendered first frame, after that "background"
+ // color refers to the color that was there the previous frame.
+ memset(g->out, 0x00, 4 * pcount);
+ memset(g->background, 0x00, 4 * pcount); // state of the background (starts transparent)
+ memset(g->history, 0x00, pcount); // pixels that were affected previous frame
+ first_frame = 1;
+ } else {
+ // second frame - how do we dispose of the previous one?
+ dispose = (g->eflags & 0x1C) >> 2;
+ pcount = g->w * g->h;
- prev_out = g->out;
- g->out = (stbi_uc *) stbi__malloc_mad3(4, g->w, g->h, 0);
- if (g->out == 0) return stbi__errpuc("outofmem", "Out of memory");
+ if ((dispose == 3) && (two_back == 0)) {
+ dispose = 2; // if I don't have an image to revert back to, default to the old background
+ }
- switch ((g->eflags & 0x1C) >> 2) {
- case 0: // unspecified (also always used on 1st frame)
- stbi__fill_gif_background(g, 0, 0, 4 * g->w, 4 * g->w * g->h);
- break;
- case 1: // do not dispose
- if (prev_out) memcpy(g->out, prev_out, 4 * g->w * g->h);
- g->old_out = prev_out;
- break;
- case 2: // dispose to background
- if (prev_out) memcpy(g->out, prev_out, 4 * g->w * g->h);
- stbi__fill_gif_background(g, g->start_x, g->start_y, g->max_x, g->max_y);
- break;
- case 3: // dispose to previous
- if (g->old_out) {
- for (i = g->start_y; i < g->max_y; i += 4 * g->w)
- memcpy(&g->out[i + g->start_x], &g->old_out[i + g->start_x], g->max_x - g->start_x);
+ if (dispose == 3) { // use previous graphic
+ for (pi = 0; pi < pcount; ++pi) {
+ if (g->history[pi]) {
+ memcpy( &g->out[pi * 4], &two_back[pi * 4], 4 );
+ }
}
- break;
+ } else if (dispose == 2) {
+ // restore what was changed last frame to background before that frame;
+ for (pi = 0; pi < pcount; ++pi) {
+ if (g->history[pi]) {
+ memcpy( &g->out[pi * 4], &g->background[pi * 4], 4 );
+ }
+ }
+ } else {
+ // This is a non-disposal case eithe way, so just
+ // leave the pixels as is, and they will become the new background
+ // 1: do not dispose
+ // 0: not specified.
+ }
+
+ // background is what out is after the undoing of the previou frame;
+ memcpy( g->background, g->out, 4 * g->w * g->h );
}
+ // clear my history;
+ memset( g->history, 0x00, g->w * g->h ); // pixels that were affected previous frame
+
for (;;) {
- switch (stbi__get8(s)) {
+ int tag = stbi__get8(s);
+ switch (tag) {
case 0x2C: /* Image Descriptor */
{
- int prev_trans = -1;
stbi__int32 x, y, w, h;
stbi_uc *o;
@@ -6306,6 +6784,13 @@ static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, i
g->cur_x = g->start_x;
g->cur_y = g->start_y;
+ // if the width of the specified rectangle is 0, that means
+ // we may not see *any* pixels or the image is malformed;
+ // to make sure this is caught, move the current y down to
+ // max_y (which is what out_gif_code checks).
+ if (w == 0)
+ g->cur_y = g->max_y;
+
g->lflags = stbi__get8(s);
if (g->lflags & 0x40) {
@@ -6320,19 +6805,24 @@ static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, i
stbi__gif_parse_colortable(s,g->lpal, 2 << (g->lflags & 7), g->eflags & 0x01 ? g->transparent : -1);
g->color_table = (stbi_uc *) g->lpal;
} else if (g->flags & 0x80) {
- if (g->transparent >= 0 && (g->eflags & 0x01)) {
- prev_trans = g->pal[g->transparent][3];
- g->pal[g->transparent][3] = 0;
- }
g->color_table = (stbi_uc *) g->pal;
} else
return stbi__errpuc("missing color table", "Corrupt GIF");
o = stbi__process_gif_raster(s, g);
- if (o == NULL) return NULL;
-
- if (prev_trans != -1)
- g->pal[g->transparent][3] = (stbi_uc) prev_trans;
+ if (!o) return NULL;
+
+ // if this was the first frame,
+ pcount = g->w * g->h;
+ if (first_frame && (g->bgindex > 0)) {
+ // if first frame, any pixel not drawn to gets the background color
+ for (pi = 0; pi < pcount; ++pi) {
+ if (g->history[pi] == 0) {
+ g->pal[g->bgindex][3] = 255; // just in case it was made transparent, undo that; It will be reset next frame if need be;
+ memcpy( &g->out[pi * 4], &g->pal[g->bgindex], 4 );
+ }
+ }
+ }
return o;
}
@@ -6340,19 +6830,35 @@ static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, i
case 0x21: // Comment Extension.
{
int len;
- if (stbi__get8(s) == 0xF9) { // Graphic Control Extension.
+ int ext = stbi__get8(s);
+ if (ext == 0xF9) { // Graphic Control Extension.
len = stbi__get8(s);
if (len == 4) {
g->eflags = stbi__get8(s);
- g->delay = stbi__get16le(s);
- g->transparent = stbi__get8(s);
+ g->delay = 10 * stbi__get16le(s); // delay - 1/100th of a second, saving as 1/1000ths.
+
+ // unset old transparent
+ if (g->transparent >= 0) {
+ g->pal[g->transparent][3] = 255;
+ }
+ if (g->eflags & 0x01) {
+ g->transparent = stbi__get8(s);
+ if (g->transparent >= 0) {
+ g->pal[g->transparent][3] = 0;
+ }
+ } else {
+ // don't need transparent
+ stbi__skip(s, 1);
+ g->transparent = -1;
+ }
} else {
stbi__skip(s, len);
break;
}
}
- while ((len = stbi__get8(s)) != 0)
+ while ((len = stbi__get8(s)) != 0) {
stbi__skip(s, len);
+ }
break;
}
@@ -6363,28 +6869,130 @@ static stbi_uc *stbi__gif_load_next(stbi__context *s, stbi__gif *g, int *comp, i
return stbi__errpuc("unknown code", "Corrupt GIF");
}
}
+}
- STBI_NOTUSED(req_comp);
+static void *stbi__load_gif_main_outofmem(stbi__gif *g, stbi_uc *out, int **delays)
+{
+ STBI_FREE(g->out);
+ STBI_FREE(g->history);
+ STBI_FREE(g->background);
+
+ if (out) STBI_FREE(out);
+ if (delays && *delays) STBI_FREE(*delays);
+ return stbi__errpuc("outofmem", "Out of memory");
+}
+
+static void *stbi__load_gif_main(stbi__context *s, int **delays, int *x, int *y, int *z, int *comp, int req_comp)
+{
+ if (stbi__gif_test(s)) {
+ int layers = 0;
+ stbi_uc *u = 0;
+ stbi_uc *out = 0;
+ stbi_uc *two_back = 0;
+ stbi__gif g;
+ int stride;
+ int out_size = 0;
+ int delays_size = 0;
+
+ STBI_NOTUSED(out_size);
+ STBI_NOTUSED(delays_size);
+
+ memset(&g, 0, sizeof(g));
+ if (delays) {
+ *delays = 0;
+ }
+
+ do {
+ u = stbi__gif_load_next(s, &g, comp, req_comp, two_back);
+ if (u == (stbi_uc *) s) u = 0; // end of animated gif marker
+
+ if (u) {
+ *x = g.w;
+ *y = g.h;
+ ++layers;
+ stride = g.w * g.h * 4;
+
+ if (out) {
+ void *tmp = (stbi_uc*) STBI_REALLOC_SIZED( out, out_size, layers * stride );
+ if (!tmp)
+ return stbi__load_gif_main_outofmem(&g, out, delays);
+ else {
+ out = (stbi_uc*) tmp;
+ out_size = layers * stride;
+ }
+
+ if (delays) {
+ int *new_delays = (int*) STBI_REALLOC_SIZED( *delays, delays_size, sizeof(int) * layers );
+ if (!new_delays)
+ return stbi__load_gif_main_outofmem(&g, out, delays);
+ *delays = new_delays;
+ delays_size = layers * sizeof(int);
+ }
+ } else {
+ out = (stbi_uc*)stbi__malloc( layers * stride );
+ if (!out)
+ return stbi__load_gif_main_outofmem(&g, out, delays);
+ out_size = layers * stride;
+ if (delays) {
+ *delays = (int*) stbi__malloc( layers * sizeof(int) );
+ if (!*delays)
+ return stbi__load_gif_main_outofmem(&g, out, delays);
+ delays_size = layers * sizeof(int);
+ }
+ }
+ memcpy( out + ((layers - 1) * stride), u, stride );
+ if (layers >= 2) {
+ two_back = out - 2 * stride;
+ }
+
+ if (delays) {
+ (*delays)[layers - 1U] = g.delay;
+ }
+ }
+ } while (u != 0);
+
+ // free temp buffer;
+ STBI_FREE(g.out);
+ STBI_FREE(g.history);
+ STBI_FREE(g.background);
+
+ // do the final conversion after loading everything;
+ if (req_comp && req_comp != 4)
+ out = stbi__convert_format(out, 4, req_comp, layers * g.w, g.h);
+
+ *z = layers;
+ return out;
+ } else {
+ return stbi__errpuc("not GIF", "Image was not as a gif type.");
+ }
}
static void *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp, stbi__result_info *ri)
{
stbi_uc *u = 0;
- stbi__gif* g = (stbi__gif*) stbi__malloc(sizeof(stbi__gif));
- memset(g, 0, sizeof(*g));
+ stbi__gif g;
+ memset(&g, 0, sizeof(g));
STBI_NOTUSED(ri);
- u = stbi__gif_load_next(s, g, comp, req_comp);
+ u = stbi__gif_load_next(s, &g, comp, req_comp, 0);
if (u == (stbi_uc *) s) u = 0; // end of animated gif marker
if (u) {
- *x = g->w;
- *y = g->h;
+ *x = g.w;
+ *y = g.h;
+
+ // moved conversion to after successful load so that the same
+ // can be done for multiple frames.
if (req_comp && req_comp != 4)
- u = stbi__convert_format(u, 4, req_comp, g->w, g->h);
+ u = stbi__convert_format(u, 4, req_comp, g.w, g.h);
+ } else if (g.out) {
+ // if there was an error and we allocated an image buffer, free it!
+ STBI_FREE(g.out);
}
- else if (g->out)
- STBI_FREE(g->out);
- STBI_FREE(g);
+
+ // free buffers needed for multiple frame loading;
+ STBI_FREE(g.history);
+ STBI_FREE(g.background);
+
return u;
}
@@ -6508,6 +7116,9 @@ static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int re
token += 3;
width = (int) strtol(token, NULL, 10);
+ if (height > STBI_MAX_DIMENSIONS) return stbi__errpf("too large","Very large image (corrupt?)");
+ if (width > STBI_MAX_DIMENSIONS) return stbi__errpf("too large","Very large image (corrupt?)");
+
*x = width;
*y = height;
@@ -6650,12 +7261,18 @@ static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp)
info.all_a = 255;
p = stbi__bmp_parse_header(s, &info);
- stbi__rewind( s );
- if (p == NULL)
+ if (p == NULL) {
+ stbi__rewind( s );
return 0;
+ }
if (x) *x = s->img_x;
if (y) *y = s->img_y;
- if (comp) *comp = info.ma ? 4 : 3;
+ if (comp) {
+ if (info.bpp == 24 && info.ma == 0xff000000)
+ *comp = 3;
+ else
+ *comp = info.ma ? 4 : 3;
+ }
return 1;
}
#endif
@@ -6663,7 +7280,7 @@ static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp)
#ifndef STBI_NO_PSD
static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp)
{
- int channelCount, dummy;
+ int channelCount, dummy, depth;
if (!x) x = &dummy;
if (!y) y = &dummy;
if (!comp) comp = &dummy;
@@ -6683,7 +7300,8 @@ static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp)
}
*y = stbi__get32be(s);
*x = stbi__get32be(s);
- if (stbi__get16be(s) != 8) {
+ depth = stbi__get16be(s);
+ if (depth != 8 && depth != 16) {
stbi__rewind( s );
return 0;
}
@@ -6694,6 +7312,33 @@ static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp)
*comp = 4;
return 1;
}
+
+static int stbi__psd_is16(stbi__context *s)
+{
+ int channelCount, depth;
+ if (stbi__get32be(s) != 0x38425053) {
+ stbi__rewind( s );
+ return 0;
+ }
+ if (stbi__get16be(s) != 1) {
+ stbi__rewind( s );
+ return 0;
+ }
+ stbi__skip(s, 6);
+ channelCount = stbi__get16be(s);
+ if (channelCount < 0 || channelCount > 16) {
+ stbi__rewind( s );
+ return 0;
+ }
+ STBI_NOTUSED(stbi__get32be(s));
+ STBI_NOTUSED(stbi__get32be(s));
+ depth = stbi__get16be(s);
+ if (depth != 16) {
+ stbi__rewind( s );
+ return 0;
+ }
+ return 1;
+}
#endif
#ifndef STBI_NO_PIC
@@ -6765,7 +7410,6 @@ static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp)
// Known limitations:
// Does not support comments in the header section
// Does not support ASCII image data (formats P2 and P3)
-// Does not support 16-bit-per-channel
#ifndef STBI_NO_PNM
@@ -6786,19 +7430,23 @@ static void *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req
stbi_uc *out;
STBI_NOTUSED(ri);
- if (!stbi__pnm_info(s, (int *)&s->img_x, (int *)&s->img_y, (int *)&s->img_n))
+ ri->bits_per_channel = stbi__pnm_info(s, (int *)&s->img_x, (int *)&s->img_y, (int *)&s->img_n);
+ if (ri->bits_per_channel == 0)
return 0;
+ if (s->img_y > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+ if (s->img_x > STBI_MAX_DIMENSIONS) return stbi__errpuc("too large","Very large image (corrupt?)");
+
*x = s->img_x;
*y = s->img_y;
if (comp) *comp = s->img_n;
- if (!stbi__mad3sizes_valid(s->img_n, s->img_x, s->img_y, 0))
+ if (!stbi__mad4sizes_valid(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0))
return stbi__errpuc("too large", "PNM too large");
- out = (stbi_uc *) stbi__malloc_mad3(s->img_n, s->img_x, s->img_y, 0);
+ out = (stbi_uc *) stbi__malloc_mad4(s->img_n, s->img_x, s->img_y, ri->bits_per_channel / 8, 0);
if (!out) return stbi__errpuc("outofmem", "Out of memory");
- stbi__getn(s, out, s->img_n * s->img_x * s->img_y);
+ stbi__getn(s, out, s->img_n * s->img_x * s->img_y * (ri->bits_per_channel / 8));
if (req_comp && req_comp != s->img_n) {
out = stbi__convert_format(out, s->img_n, req_comp, s->img_x, s->img_y);
@@ -6874,11 +7522,19 @@ static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp)
stbi__pnm_skip_whitespace(s, &c);
maxv = stbi__pnm_getinteger(s, &c); // read max value
-
- if (maxv > 255)
- return stbi__err("max value > 255", "PPM image not 8-bit");
+ if (maxv > 65535)
+ return stbi__err("max value > 65535", "PPM image supports only 8-bit and 16-bit images");
+ else if (maxv > 255)
+ return 16;
else
- return 1;
+ return 8;
+}
+
+static int stbi__pnm_is16(stbi__context *s)
+{
+ if (stbi__pnm_info(s, NULL, NULL, NULL) == 16)
+ return 1;
+ return 0;
}
#endif
@@ -6924,6 +7580,22 @@ static int stbi__info_main(stbi__context *s, int *x, int *y, int *comp)
return stbi__err("unknown image type", "Image not of any known type, or corrupt");
}
+static int stbi__is_16_main(stbi__context *s)
+{
+ #ifndef STBI_NO_PNG
+ if (stbi__png_is16(s)) return 1;
+ #endif
+
+ #ifndef STBI_NO_PSD
+ if (stbi__psd_is16(s)) return 1;
+ #endif
+
+ #ifndef STBI_NO_PNM
+ if (stbi__pnm_is16(s)) return 1;
+ #endif
+ return 0;
+}
+
#ifndef STBI_NO_STDIO
STBIDEF int stbi_info(char const *filename, int *x, int *y, int *comp)
{
@@ -6945,6 +7617,27 @@ STBIDEF int stbi_info_from_file(FILE *f, int *x, int *y, int *comp)
fseek(f,pos,SEEK_SET);
return r;
}
+
+STBIDEF int stbi_is_16_bit(char const *filename)
+{
+ FILE *f = stbi__fopen(filename, "rb");
+ int result;
+ if (!f) return stbi__err("can't fopen", "Unable to open file");
+ result = stbi_is_16_bit_from_file(f);
+ fclose(f);
+ return result;
+}
+
+STBIDEF int stbi_is_16_bit_from_file(FILE *f)
+{
+ int r;
+ stbi__context s;
+ long pos = ftell(f);
+ stbi__start_file(&s, f);
+ r = stbi__is_16_main(&s);
+ fseek(f,pos,SEEK_SET);
+ return r;
+}
#endif // !STBI_NO_STDIO
STBIDEF int stbi_info_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp)
@@ -6961,10 +7654,38 @@ STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *c, void *user, int
return stbi__info_main(&s,x,y,comp);
}
+STBIDEF int stbi_is_16_bit_from_memory(stbi_uc const *buffer, int len)
+{
+ stbi__context s;
+ stbi__start_mem(&s,buffer,len);
+ return stbi__is_16_main(&s);
+}
+
+STBIDEF int stbi_is_16_bit_from_callbacks(stbi_io_callbacks const *c, void *user)
+{
+ stbi__context s;
+ stbi__start_callbacks(&s, (stbi_io_callbacks *) c, user);
+ return stbi__is_16_main(&s);
+}
+
#endif // STB_IMAGE_IMPLEMENTATION
/*
revision history:
+ 2.20 (2019-02-07) support utf8 filenames in Windows; fix warnings and platform ifdefs
+ 2.19 (2018-02-11) fix warning
+ 2.18 (2018-01-30) fix warnings
+ 2.17 (2018-01-29) change sbti__shiftsigned to avoid clang -O2 bug
+ 1-bit BMP
+ *_is_16_bit api
+ avoid warnings
+ 2.16 (2017-07-23) all functions have 16-bit variants;
+ STBI_NO_STDIO works again;
+ compilation fixes;
+ fix rounding in unpremultiply;
+ optimize vertical flip;
+ disable raw_len validation;
+ documentation fixes
2.15 (2017-03-18) fix png-1,2,4 bug; now all Imagenet JPGs decode;
warning fixes; disable run-time SSE detection on gcc;
uniform handling of optional "return" values;
diff --git a/ThirdParty/stb/stb_image_resize.h b/ThirdParty/stb/stb_image_resize.h
new file mode 100644
index 0000000..ef9e6fe
--- /dev/null
+++ b/ThirdParty/stb/stb_image_resize.h
@@ -0,0 +1,2634 @@
+/* stb_image_resize - v0.97 - public domain image resizing
+ by Jorge L Rodriguez (@VinoBS) - 2014
+ http://github.com/nothings/stb
+
+ Written with emphasis on usability, portability, and efficiency. (No
+ SIMD or threads, so it be easily outperformed by libs that use those.)
+ Only scaling and translation is supported, no rotations or shears.
+ Easy API downsamples w/Mitchell filter, upsamples w/cubic interpolation.
+
+ COMPILING & LINKING
+ In one C/C++ file that #includes this file, do this:
+ #define STB_IMAGE_RESIZE_IMPLEMENTATION
+ before the #include. That will create the implementation in that file.
+
+ QUICKSTART
+ stbir_resize_uint8( input_pixels , in_w , in_h , 0,
+ output_pixels, out_w, out_h, 0, num_channels)
+ stbir_resize_float(...)
+ stbir_resize_uint8_srgb( input_pixels , in_w , in_h , 0,
+ output_pixels, out_w, out_h, 0,
+ num_channels , alpha_chan , 0)
+ stbir_resize_uint8_srgb_edgemode(
+ input_pixels , in_w , in_h , 0,
+ output_pixels, out_w, out_h, 0,
+ num_channels , alpha_chan , 0, STBIR_EDGE_CLAMP)
+ // WRAP/REFLECT/ZERO
+
+ FULL API
+ See the "header file" section of the source for API documentation.
+
+ ADDITIONAL DOCUMENTATION
+
+ SRGB & FLOATING POINT REPRESENTATION
+ The sRGB functions presume IEEE floating point. If you do not have
+ IEEE floating point, define STBIR_NON_IEEE_FLOAT. This will use
+ a slower implementation.
+
+ MEMORY ALLOCATION
+ The resize functions here perform a single memory allocation using
+ malloc. To control the memory allocation, before the #include that
+ triggers the implementation, do:
+
+ #define STBIR_MALLOC(size,context) ...
+ #define STBIR_FREE(ptr,context) ...
+
+ Each resize function makes exactly one call to malloc/free, so to use
+ temp memory, store the temp memory in the context and return that.
+
+ ASSERT
+ Define STBIR_ASSERT(boolval) to override assert() and not use assert.h
+
+ OPTIMIZATION
+ Define STBIR_SATURATE_INT to compute clamp values in-range using
+ integer operations instead of float operations. This may be faster
+ on some platforms.
+
+ DEFAULT FILTERS
+ For functions which don't provide explicit control over what filters
+ to use, you can change the compile-time defaults with
+
+ #define STBIR_DEFAULT_FILTER_UPSAMPLE STBIR_FILTER_something
+ #define STBIR_DEFAULT_FILTER_DOWNSAMPLE STBIR_FILTER_something
+
+ See stbir_filter in the header-file section for the list of filters.
+
+ NEW FILTERS
+ A number of 1D filter kernels are used. For a list of
+ supported filters see the stbir_filter enum. To add a new filter,
+ write a filter function and add it to stbir__filter_info_table.
+
+ PROGRESS
+ For interactive use with slow resize operations, you can install
+ a progress-report callback:
+
+ #define STBIR_PROGRESS_REPORT(val) some_func(val)
+
+ The parameter val is a float which goes from 0 to 1 as progress is made.
+
+ For example:
+
+ static void my_progress_report(float progress);
+ #define STBIR_PROGRESS_REPORT(val) my_progress_report(val)
+
+ #define STB_IMAGE_RESIZE_IMPLEMENTATION
+ #include "stb_image_resize.h"
+
+ static void my_progress_report(float progress)
+ {
+ printf("Progress: %f%%\n", progress*100);
+ }
+
+ MAX CHANNELS
+ If your image has more than 64 channels, define STBIR_MAX_CHANNELS
+ to the max you'll have.
+
+ ALPHA CHANNEL
+ Most of the resizing functions provide the ability to control how
+ the alpha channel of an image is processed. The important things
+ to know about this:
+
+ 1. The best mathematically-behaved version of alpha to use is
+ called "premultiplied alpha", in which the other color channels
+ have had the alpha value multiplied in. If you use premultiplied
+ alpha, linear filtering (such as image resampling done by this
+ library, or performed in texture units on GPUs) does the "right
+ thing". While premultiplied alpha is standard in the movie CGI
+ industry, it is still uncommon in the videogame/real-time world.
+
+ If you linearly filter non-premultiplied alpha, strange effects
+ occur. (For example, the 50/50 average of 99% transparent bright green
+ and 1% transparent black produces 50% transparent dark green when
+ non-premultiplied, whereas premultiplied it produces 50%
+ transparent near-black. The former introduces green energy
+ that doesn't exist in the source image.)
+
+ 2. Artists should not edit premultiplied-alpha images; artists
+ want non-premultiplied alpha images. Thus, art tools generally output
+ non-premultiplied alpha images.
+
+ 3. You will get best results in most cases by converting images
+ to premultiplied alpha before processing them mathematically.
+
+ 4. If you pass the flag STBIR_FLAG_ALPHA_PREMULTIPLIED, the
+ resizer does not do anything special for the alpha channel;
+ it is resampled identically to other channels. This produces
+ the correct results for premultiplied-alpha images, but produces
+ less-than-ideal results for non-premultiplied-alpha images.
+
+ 5. If you do not pass the flag STBIR_FLAG_ALPHA_PREMULTIPLIED,
+ then the resizer weights the contribution of input pixels
+ based on their alpha values, or, equivalently, it multiplies
+ the alpha value into the color channels, resamples, then divides
+ by the resultant alpha value. Input pixels which have alpha=0 do
+ not contribute at all to output pixels unless _all_ of the input
+ pixels affecting that output pixel have alpha=0, in which case
+ the result for that pixel is the same as it would be without
+ STBIR_FLAG_ALPHA_PREMULTIPLIED. However, this is only true for
+ input images in integer formats. For input images in float format,
+ input pixels with alpha=0 have no effect, and output pixels
+ which have alpha=0 will be 0 in all channels. (For float images,
+ you can manually achieve the same result by adding a tiny epsilon
+ value to the alpha channel of every image, and then subtracting
+ or clamping it at the end.)
+
+ 6. You can suppress the behavior described in #5 and make
+ all-0-alpha pixels have 0 in all channels by #defining
+ STBIR_NO_ALPHA_EPSILON.
+
+ 7. You can separately control whether the alpha channel is
+ interpreted as linear or affected by the colorspace. By default
+ it is linear; you almost never want to apply the colorspace.
+ (For example, graphics hardware does not apply sRGB conversion
+ to the alpha channel.)
+
+ CONTRIBUTORS
+ Jorge L Rodriguez: Implementation
+ Sean Barrett: API design, optimizations
+ Aras Pranckevicius: bugfix
+ Nathan Reed: warning fixes
+
+ REVISIONS
+ 0.97 (2020-02-02) fixed warning
+ 0.96 (2019-03-04) fixed warnings
+ 0.95 (2017-07-23) fixed warnings
+ 0.94 (2017-03-18) fixed warnings
+ 0.93 (2017-03-03) fixed bug with certain combinations of heights
+ 0.92 (2017-01-02) fix integer overflow on large (>2GB) images
+ 0.91 (2016-04-02) fix warnings; fix handling of subpixel regions
+ 0.90 (2014-09-17) first released version
+
+ LICENSE
+ See end of file for license information.
+
+ TODO
+ Don't decode all of the image data when only processing a partial tile
+ Don't use full-width decode buffers when only processing a partial tile
+ When processing wide images, break processing into tiles so data fits in L1 cache
+ Installable filters?
+ Resize that respects alpha test coverage
+ (Reference code: FloatImage::alphaTestCoverage and FloatImage::scaleAlphaToCoverage:
+ https://code.google.com/p/nvidia-texture-tools/source/browse/trunk/src/nvimage/FloatImage.cpp )
+*/
+
+#ifndef STBIR_INCLUDE_STB_IMAGE_RESIZE_H
+#define STBIR_INCLUDE_STB_IMAGE_RESIZE_H
+
+#ifdef _MSC_VER
+typedef unsigned char stbir_uint8;
+typedef unsigned short stbir_uint16;
+typedef unsigned int stbir_uint32;
+#else
+#include <stdint.h>
+typedef uint8_t stbir_uint8;
+typedef uint16_t stbir_uint16;
+typedef uint32_t stbir_uint32;
+#endif
+
+#ifndef STBIRDEF
+#ifdef STB_IMAGE_RESIZE_STATIC
+#define STBIRDEF static
+#else
+#ifdef __cplusplus
+#define STBIRDEF extern "C"
+#else
+#define STBIRDEF extern
+#endif
+#endif
+#endif
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// Easy-to-use API:
+//
+// * "input pixels" points to an array of image data with 'num_channels' channels (e.g. RGB=3, RGBA=4)
+// * input_w is input image width (x-axis), input_h is input image height (y-axis)
+// * stride is the offset between successive rows of image data in memory, in bytes. you can
+// specify 0 to mean packed continuously in memory
+// * alpha channel is treated identically to other channels.
+// * colorspace is linear or sRGB as specified by function name
+// * returned result is 1 for success or 0 in case of an error.
+// #define STBIR_ASSERT() to trigger an assert on parameter validation errors.
+// * Memory required grows approximately linearly with input and output size, but with
+// discontinuities at input_w == output_w and input_h == output_h.
+// * These functions use a "default" resampling filter defined at compile time. To change the filter,
+// you can change the compile-time defaults by #defining STBIR_DEFAULT_FILTER_UPSAMPLE
+// and STBIR_DEFAULT_FILTER_DOWNSAMPLE, or you can use the medium-complexity API.
+
+STBIRDEF int stbir_resize_uint8( const unsigned char *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ unsigned char *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels);
+
+STBIRDEF int stbir_resize_float( const float *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ float *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels);
+
+
+// The following functions interpret image data as gamma-corrected sRGB.
+// Specify STBIR_ALPHA_CHANNEL_NONE if you have no alpha channel,
+// or otherwise provide the index of the alpha channel. Flags value
+// of 0 will probably do the right thing if you're not sure what
+// the flags mean.
+
+#define STBIR_ALPHA_CHANNEL_NONE -1
+
+// Set this flag if your texture has premultiplied alpha. Otherwise, stbir will
+// use alpha-weighted resampling (effectively premultiplying, resampling,
+// then unpremultiplying).
+#define STBIR_FLAG_ALPHA_PREMULTIPLIED (1 << 0)
+// The specified alpha channel should be handled as gamma-corrected value even
+// when doing sRGB operations.
+#define STBIR_FLAG_ALPHA_USES_COLORSPACE (1 << 1)
+
+STBIRDEF int stbir_resize_uint8_srgb(const unsigned char *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ unsigned char *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags);
+
+
+typedef enum
+{
+ STBIR_EDGE_CLAMP = 1,
+ STBIR_EDGE_REFLECT = 2,
+ STBIR_EDGE_WRAP = 3,
+ STBIR_EDGE_ZERO = 4,
+} stbir_edge;
+
+// This function adds the ability to specify how requests to sample off the edge of the image are handled.
+STBIRDEF int stbir_resize_uint8_srgb_edgemode(const unsigned char *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ unsigned char *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_wrap_mode);
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// Medium-complexity API
+//
+// This extends the easy-to-use API as follows:
+//
+// * Alpha-channel can be processed separately
+// * If alpha_channel is not STBIR_ALPHA_CHANNEL_NONE
+// * Alpha channel will not be gamma corrected (unless flags&STBIR_FLAG_GAMMA_CORRECT)
+// * Filters will be weighted by alpha channel (unless flags&STBIR_FLAG_ALPHA_PREMULTIPLIED)
+// * Filter can be selected explicitly
+// * uint16 image type
+// * sRGB colorspace available for all types
+// * context parameter for passing to STBIR_MALLOC
+
+typedef enum
+{
+ STBIR_FILTER_DEFAULT = 0, // use same filter type that easy-to-use API chooses
+ STBIR_FILTER_BOX = 1, // A trapezoid w/1-pixel wide ramps, same result as box for integer scale ratios
+ STBIR_FILTER_TRIANGLE = 2, // On upsampling, produces same results as bilinear texture filtering
+ STBIR_FILTER_CUBICBSPLINE = 3, // The cubic b-spline (aka Mitchell-Netrevalli with B=1,C=0), gaussian-esque
+ STBIR_FILTER_CATMULLROM = 4, // An interpolating cubic spline
+ STBIR_FILTER_MITCHELL = 5, // Mitchell-Netrevalli filter with B=1/3, C=1/3
+} stbir_filter;
+
+typedef enum
+{
+ STBIR_COLORSPACE_LINEAR,
+ STBIR_COLORSPACE_SRGB,
+
+ STBIR_MAX_COLORSPACES,
+} stbir_colorspace;
+
+// The following functions are all identical except for the type of the image data
+
+STBIRDEF int stbir_resize_uint8_generic( const unsigned char *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ unsigned char *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_wrap_mode, stbir_filter filter, stbir_colorspace space,
+ void *alloc_context);
+
+STBIRDEF int stbir_resize_uint16_generic(const stbir_uint16 *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ stbir_uint16 *output_pixels , int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_wrap_mode, stbir_filter filter, stbir_colorspace space,
+ void *alloc_context);
+
+STBIRDEF int stbir_resize_float_generic( const float *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ float *output_pixels , int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_wrap_mode, stbir_filter filter, stbir_colorspace space,
+ void *alloc_context);
+
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// Full-complexity API
+//
+// This extends the medium API as follows:
+//
+// * uint32 image type
+// * not typesafe
+// * separate filter types for each axis
+// * separate edge modes for each axis
+// * can specify scale explicitly for subpixel correctness
+// * can specify image source tile using texture coordinates
+
+typedef enum
+{
+ STBIR_TYPE_UINT8 ,
+ STBIR_TYPE_UINT16,
+ STBIR_TYPE_UINT32,
+ STBIR_TYPE_FLOAT ,
+
+ STBIR_MAX_TYPES
+} stbir_datatype;
+
+STBIRDEF int stbir_resize( const void *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ void *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ stbir_datatype datatype,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_mode_horizontal, stbir_edge edge_mode_vertical,
+ stbir_filter filter_horizontal, stbir_filter filter_vertical,
+ stbir_colorspace space, void *alloc_context);
+
+STBIRDEF int stbir_resize_subpixel(const void *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ void *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ stbir_datatype datatype,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_mode_horizontal, stbir_edge edge_mode_vertical,
+ stbir_filter filter_horizontal, stbir_filter filter_vertical,
+ stbir_colorspace space, void *alloc_context,
+ float x_scale, float y_scale,
+ float x_offset, float y_offset);
+
+STBIRDEF int stbir_resize_region( const void *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ void *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ stbir_datatype datatype,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_mode_horizontal, stbir_edge edge_mode_vertical,
+ stbir_filter filter_horizontal, stbir_filter filter_vertical,
+ stbir_colorspace space, void *alloc_context,
+ float s0, float t0, float s1, float t1);
+// (s0, t0) & (s1, t1) are the top-left and bottom right corner (uv addressing style: [0, 1]x[0, 1]) of a region of the input image to use.
+
+//
+//
+//// end header file /////////////////////////////////////////////////////
+#endif // STBIR_INCLUDE_STB_IMAGE_RESIZE_H
+
+
+
+
+
+#ifdef STB_IMAGE_RESIZE_IMPLEMENTATION
+
+#ifndef STBIR_ASSERT
+#include <assert.h>
+#define STBIR_ASSERT(x) assert(x)
+#endif
+
+// For memset
+#include <string.h>
+
+#include <math.h>
+
+#ifndef STBIR_MALLOC
+#include <stdlib.h>
+// use comma operator to evaluate c, to avoid "unused parameter" warnings
+#define STBIR_MALLOC(size,c) ((void)(c), malloc(size))
+#define STBIR_FREE(ptr,c) ((void)(c), free(ptr))
+#endif
+
+#ifndef _MSC_VER
+#ifdef __cplusplus
+#define stbir__inline inline
+#else
+#define stbir__inline
+#endif
+#else
+#define stbir__inline __forceinline
+#endif
+
+
+// should produce compiler error if size is wrong
+typedef unsigned char stbir__validate_uint32[sizeof(stbir_uint32) == 4 ? 1 : -1];
+
+#ifdef _MSC_VER
+#define STBIR__NOTUSED(v) (void)(v)
+#else
+#define STBIR__NOTUSED(v) (void)sizeof(v)
+#endif
+
+#define STBIR__ARRAY_SIZE(a) (sizeof((a))/sizeof((a)[0]))
+
+#ifndef STBIR_DEFAULT_FILTER_UPSAMPLE
+#define STBIR_DEFAULT_FILTER_UPSAMPLE STBIR_FILTER_CATMULLROM
+#endif
+
+#ifndef STBIR_DEFAULT_FILTER_DOWNSAMPLE
+#define STBIR_DEFAULT_FILTER_DOWNSAMPLE STBIR_FILTER_MITCHELL
+#endif
+
+#ifndef STBIR_PROGRESS_REPORT
+#define STBIR_PROGRESS_REPORT(float_0_to_1)
+#endif
+
+#ifndef STBIR_MAX_CHANNELS
+#define STBIR_MAX_CHANNELS 64
+#endif
+
+#if STBIR_MAX_CHANNELS > 65536
+#error "Too many channels; STBIR_MAX_CHANNELS must be no more than 65536."
+// because we store the indices in 16-bit variables
+#endif
+
+// This value is added to alpha just before premultiplication to avoid
+// zeroing out color values. It is equivalent to 2^-80. If you don't want
+// that behavior (it may interfere if you have floating point images with
+// very small alpha values) then you can define STBIR_NO_ALPHA_EPSILON to
+// disable it.
+#ifndef STBIR_ALPHA_EPSILON
+#define STBIR_ALPHA_EPSILON ((float)1 / (1 << 20) / (1 << 20) / (1 << 20) / (1 << 20))
+#endif
+
+
+
+#ifdef _MSC_VER
+#define STBIR__UNUSED_PARAM(v) (void)(v)
+#else
+#define STBIR__UNUSED_PARAM(v) (void)sizeof(v)
+#endif
+
+// must match stbir_datatype
+static unsigned char stbir__type_size[] = {
+ 1, // STBIR_TYPE_UINT8
+ 2, // STBIR_TYPE_UINT16
+ 4, // STBIR_TYPE_UINT32
+ 4, // STBIR_TYPE_FLOAT
+};
+
+// Kernel function centered at 0
+typedef float (stbir__kernel_fn)(float x, float scale);
+typedef float (stbir__support_fn)(float scale);
+
+typedef struct
+{
+ stbir__kernel_fn* kernel;
+ stbir__support_fn* support;
+} stbir__filter_info;
+
+// When upsampling, the contributors are which source pixels contribute.
+// When downsampling, the contributors are which destination pixels are contributed to.
+typedef struct
+{
+ int n0; // First contributing pixel
+ int n1; // Last contributing pixel
+} stbir__contributors;
+
+typedef struct
+{
+ const void* input_data;
+ int input_w;
+ int input_h;
+ int input_stride_bytes;
+
+ void* output_data;
+ int output_w;
+ int output_h;
+ int output_stride_bytes;
+
+ float s0, t0, s1, t1;
+
+ float horizontal_shift; // Units: output pixels
+ float vertical_shift; // Units: output pixels
+ float horizontal_scale;
+ float vertical_scale;
+
+ int channels;
+ int alpha_channel;
+ stbir_uint32 flags;
+ stbir_datatype type;
+ stbir_filter horizontal_filter;
+ stbir_filter vertical_filter;
+ stbir_edge edge_horizontal;
+ stbir_edge edge_vertical;
+ stbir_colorspace colorspace;
+
+ stbir__contributors* horizontal_contributors;
+ float* horizontal_coefficients;
+
+ stbir__contributors* vertical_contributors;
+ float* vertical_coefficients;
+
+ int decode_buffer_pixels;
+ float* decode_buffer;
+
+ float* horizontal_buffer;
+
+ // cache these because ceil/floor are inexplicably showing up in profile
+ int horizontal_coefficient_width;
+ int vertical_coefficient_width;
+ int horizontal_filter_pixel_width;
+ int vertical_filter_pixel_width;
+ int horizontal_filter_pixel_margin;
+ int vertical_filter_pixel_margin;
+ int horizontal_num_contributors;
+ int vertical_num_contributors;
+
+ int ring_buffer_length_bytes; // The length of an individual entry in the ring buffer. The total number of ring buffers is stbir__get_filter_pixel_width(filter)
+ int ring_buffer_num_entries; // Total number of entries in the ring buffer.
+ int ring_buffer_first_scanline;
+ int ring_buffer_last_scanline;
+ int ring_buffer_begin_index; // first_scanline is at this index in the ring buffer
+ float* ring_buffer;
+
+ float* encode_buffer; // A temporary buffer to store floats so we don't lose precision while we do multiply-adds.
+
+ int horizontal_contributors_size;
+ int horizontal_coefficients_size;
+ int vertical_contributors_size;
+ int vertical_coefficients_size;
+ int decode_buffer_size;
+ int horizontal_buffer_size;
+ int ring_buffer_size;
+ int encode_buffer_size;
+} stbir__info;
+
+
+static const float stbir__max_uint8_as_float = 255.0f;
+static const float stbir__max_uint16_as_float = 65535.0f;
+static const double stbir__max_uint32_as_float = 4294967295.0;
+
+
+static stbir__inline int stbir__min(int a, int b)
+{
+ return a < b ? a : b;
+}
+
+static stbir__inline float stbir__saturate(float x)
+{
+ if (x < 0)
+ return 0;
+
+ if (x > 1)
+ return 1;
+
+ return x;
+}
+
+#ifdef STBIR_SATURATE_INT
+static stbir__inline stbir_uint8 stbir__saturate8(int x)
+{
+ if ((unsigned int) x <= 255)
+ return x;
+
+ if (x < 0)
+ return 0;
+
+ return 255;
+}
+
+static stbir__inline stbir_uint16 stbir__saturate16(int x)
+{
+ if ((unsigned int) x <= 65535)
+ return x;
+
+ if (x < 0)
+ return 0;
+
+ return 65535;
+}
+#endif
+
+static float stbir__srgb_uchar_to_linear_float[256] = {
+ 0.000000f, 0.000304f, 0.000607f, 0.000911f, 0.001214f, 0.001518f, 0.001821f, 0.002125f, 0.002428f, 0.002732f, 0.003035f,
+ 0.003347f, 0.003677f, 0.004025f, 0.004391f, 0.004777f, 0.005182f, 0.005605f, 0.006049f, 0.006512f, 0.006995f, 0.007499f,
+ 0.008023f, 0.008568f, 0.009134f, 0.009721f, 0.010330f, 0.010960f, 0.011612f, 0.012286f, 0.012983f, 0.013702f, 0.014444f,
+ 0.015209f, 0.015996f, 0.016807f, 0.017642f, 0.018500f, 0.019382f, 0.020289f, 0.021219f, 0.022174f, 0.023153f, 0.024158f,
+ 0.025187f, 0.026241f, 0.027321f, 0.028426f, 0.029557f, 0.030713f, 0.031896f, 0.033105f, 0.034340f, 0.035601f, 0.036889f,
+ 0.038204f, 0.039546f, 0.040915f, 0.042311f, 0.043735f, 0.045186f, 0.046665f, 0.048172f, 0.049707f, 0.051269f, 0.052861f,
+ 0.054480f, 0.056128f, 0.057805f, 0.059511f, 0.061246f, 0.063010f, 0.064803f, 0.066626f, 0.068478f, 0.070360f, 0.072272f,
+ 0.074214f, 0.076185f, 0.078187f, 0.080220f, 0.082283f, 0.084376f, 0.086500f, 0.088656f, 0.090842f, 0.093059f, 0.095307f,
+ 0.097587f, 0.099899f, 0.102242f, 0.104616f, 0.107023f, 0.109462f, 0.111932f, 0.114435f, 0.116971f, 0.119538f, 0.122139f,
+ 0.124772f, 0.127438f, 0.130136f, 0.132868f, 0.135633f, 0.138432f, 0.141263f, 0.144128f, 0.147027f, 0.149960f, 0.152926f,
+ 0.155926f, 0.158961f, 0.162029f, 0.165132f, 0.168269f, 0.171441f, 0.174647f, 0.177888f, 0.181164f, 0.184475f, 0.187821f,
+ 0.191202f, 0.194618f, 0.198069f, 0.201556f, 0.205079f, 0.208637f, 0.212231f, 0.215861f, 0.219526f, 0.223228f, 0.226966f,
+ 0.230740f, 0.234551f, 0.238398f, 0.242281f, 0.246201f, 0.250158f, 0.254152f, 0.258183f, 0.262251f, 0.266356f, 0.270498f,
+ 0.274677f, 0.278894f, 0.283149f, 0.287441f, 0.291771f, 0.296138f, 0.300544f, 0.304987f, 0.309469f, 0.313989f, 0.318547f,
+ 0.323143f, 0.327778f, 0.332452f, 0.337164f, 0.341914f, 0.346704f, 0.351533f, 0.356400f, 0.361307f, 0.366253f, 0.371238f,
+ 0.376262f, 0.381326f, 0.386430f, 0.391573f, 0.396755f, 0.401978f, 0.407240f, 0.412543f, 0.417885f, 0.423268f, 0.428691f,
+ 0.434154f, 0.439657f, 0.445201f, 0.450786f, 0.456411f, 0.462077f, 0.467784f, 0.473532f, 0.479320f, 0.485150f, 0.491021f,
+ 0.496933f, 0.502887f, 0.508881f, 0.514918f, 0.520996f, 0.527115f, 0.533276f, 0.539480f, 0.545725f, 0.552011f, 0.558340f,
+ 0.564712f, 0.571125f, 0.577581f, 0.584078f, 0.590619f, 0.597202f, 0.603827f, 0.610496f, 0.617207f, 0.623960f, 0.630757f,
+ 0.637597f, 0.644480f, 0.651406f, 0.658375f, 0.665387f, 0.672443f, 0.679543f, 0.686685f, 0.693872f, 0.701102f, 0.708376f,
+ 0.715694f, 0.723055f, 0.730461f, 0.737911f, 0.745404f, 0.752942f, 0.760525f, 0.768151f, 0.775822f, 0.783538f, 0.791298f,
+ 0.799103f, 0.806952f, 0.814847f, 0.822786f, 0.830770f, 0.838799f, 0.846873f, 0.854993f, 0.863157f, 0.871367f, 0.879622f,
+ 0.887923f, 0.896269f, 0.904661f, 0.913099f, 0.921582f, 0.930111f, 0.938686f, 0.947307f, 0.955974f, 0.964686f, 0.973445f,
+ 0.982251f, 0.991102f, 1.0f
+};
+
+static float stbir__srgb_to_linear(float f)
+{
+ if (f <= 0.04045f)
+ return f / 12.92f;
+ else
+ return (float)pow((f + 0.055f) / 1.055f, 2.4f);
+}
+
+static float stbir__linear_to_srgb(float f)
+{
+ if (f <= 0.0031308f)
+ return f * 12.92f;
+ else
+ return 1.055f * (float)pow(f, 1 / 2.4f) - 0.055f;
+}
+
+#ifndef STBIR_NON_IEEE_FLOAT
+// From https://gist.github.com/rygorous/2203834
+
+typedef union
+{
+ stbir_uint32 u;
+ float f;
+} stbir__FP32;
+
+static const stbir_uint32 fp32_to_srgb8_tab4[104] = {
+ 0x0073000d, 0x007a000d, 0x0080000d, 0x0087000d, 0x008d000d, 0x0094000d, 0x009a000d, 0x00a1000d,
+ 0x00a7001a, 0x00b4001a, 0x00c1001a, 0x00ce001a, 0x00da001a, 0x00e7001a, 0x00f4001a, 0x0101001a,
+ 0x010e0033, 0x01280033, 0x01410033, 0x015b0033, 0x01750033, 0x018f0033, 0x01a80033, 0x01c20033,
+ 0x01dc0067, 0x020f0067, 0x02430067, 0x02760067, 0x02aa0067, 0x02dd0067, 0x03110067, 0x03440067,
+ 0x037800ce, 0x03df00ce, 0x044600ce, 0x04ad00ce, 0x051400ce, 0x057b00c5, 0x05dd00bc, 0x063b00b5,
+ 0x06970158, 0x07420142, 0x07e30130, 0x087b0120, 0x090b0112, 0x09940106, 0x0a1700fc, 0x0a9500f2,
+ 0x0b0f01cb, 0x0bf401ae, 0x0ccb0195, 0x0d950180, 0x0e56016e, 0x0f0d015e, 0x0fbc0150, 0x10630143,
+ 0x11070264, 0x1238023e, 0x1357021d, 0x14660201, 0x156601e9, 0x165a01d3, 0x174401c0, 0x182401af,
+ 0x18fe0331, 0x1a9602fe, 0x1c1502d2, 0x1d7e02ad, 0x1ed4028d, 0x201a0270, 0x21520256, 0x227d0240,
+ 0x239f0443, 0x25c003fe, 0x27bf03c4, 0x29a10392, 0x2b6a0367, 0x2d1d0341, 0x2ebe031f, 0x304d0300,
+ 0x31d105b0, 0x34a80555, 0x37520507, 0x39d504c5, 0x3c37048b, 0x3e7c0458, 0x40a8042a, 0x42bd0401,
+ 0x44c20798, 0x488e071e, 0x4c1c06b6, 0x4f76065d, 0x52a50610, 0x55ac05cc, 0x5892058f, 0x5b590559,
+ 0x5e0c0a23, 0x631c0980, 0x67db08f6, 0x6c55087f, 0x70940818, 0x74a007bd, 0x787d076c, 0x7c330723,
+};
+
+static stbir_uint8 stbir__linear_to_srgb_uchar(float in)
+{
+ static const stbir__FP32 almostone = { 0x3f7fffff }; // 1-eps
+ static const stbir__FP32 minval = { (127-13) << 23 };
+ stbir_uint32 tab,bias,scale,t;
+ stbir__FP32 f;
+
+ // Clamp to [2^(-13), 1-eps]; these two values map to 0 and 1, respectively.
+ // The tests are carefully written so that NaNs map to 0, same as in the reference
+ // implementation.
+ if (!(in > minval.f)) // written this way to catch NaNs
+ in = minval.f;
+ if (in > almostone.f)
+ in = almostone.f;
+
+ // Do the table lookup and unpack bias, scale
+ f.f = in;
+ tab = fp32_to_srgb8_tab4[(f.u - minval.u) >> 20];
+ bias = (tab >> 16) << 9;
+ scale = tab & 0xffff;
+
+ // Grab next-highest mantissa bits and perform linear interpolation
+ t = (f.u >> 12) & 0xff;
+ return (unsigned char) ((bias + scale*t) >> 16);
+}
+
+#else
+// sRGB transition values, scaled by 1<<28
+static int stbir__srgb_offset_to_linear_scaled[256] =
+{
+ 0, 40738, 122216, 203693, 285170, 366648, 448125, 529603,
+ 611080, 692557, 774035, 855852, 942009, 1033024, 1128971, 1229926,
+ 1335959, 1447142, 1563542, 1685229, 1812268, 1944725, 2082664, 2226148,
+ 2375238, 2529996, 2690481, 2856753, 3028870, 3206888, 3390865, 3580856,
+ 3776916, 3979100, 4187460, 4402049, 4622919, 4850123, 5083710, 5323731,
+ 5570236, 5823273, 6082892, 6349140, 6622065, 6901714, 7188133, 7481369,
+ 7781466, 8088471, 8402427, 8723380, 9051372, 9386448, 9728650, 10078021,
+ 10434603, 10798439, 11169569, 11548036, 11933879, 12327139, 12727857, 13136073,
+ 13551826, 13975156, 14406100, 14844697, 15290987, 15745007, 16206795, 16676389,
+ 17153826, 17639142, 18132374, 18633560, 19142734, 19659934, 20185196, 20718552,
+ 21260042, 21809696, 22367554, 22933648, 23508010, 24090680, 24681686, 25281066,
+ 25888850, 26505076, 27129772, 27762974, 28404716, 29055026, 29713942, 30381490,
+ 31057708, 31742624, 32436272, 33138682, 33849884, 34569912, 35298800, 36036568,
+ 36783260, 37538896, 38303512, 39077136, 39859796, 40651528, 41452360, 42262316,
+ 43081432, 43909732, 44747252, 45594016, 46450052, 47315392, 48190064, 49074096,
+ 49967516, 50870356, 51782636, 52704392, 53635648, 54576432, 55526772, 56486700,
+ 57456236, 58435408, 59424248, 60422780, 61431036, 62449032, 63476804, 64514376,
+ 65561776, 66619028, 67686160, 68763192, 69850160, 70947088, 72053992, 73170912,
+ 74297864, 75434880, 76581976, 77739184, 78906536, 80084040, 81271736, 82469648,
+ 83677792, 84896192, 86124888, 87363888, 88613232, 89872928, 91143016, 92423512,
+ 93714432, 95015816, 96327688, 97650056, 98982952, 100326408, 101680440, 103045072,
+ 104420320, 105806224, 107202800, 108610064, 110028048, 111456776, 112896264, 114346544,
+ 115807632, 117279552, 118762328, 120255976, 121760536, 123276016, 124802440, 126339832,
+ 127888216, 129447616, 131018048, 132599544, 134192112, 135795792, 137410592, 139036528,
+ 140673648, 142321952, 143981456, 145652208, 147334208, 149027488, 150732064, 152447968,
+ 154175200, 155913792, 157663776, 159425168, 161197984, 162982240, 164777968, 166585184,
+ 168403904, 170234160, 172075968, 173929344, 175794320, 177670896, 179559120, 181458992,
+ 183370528, 185293776, 187228736, 189175424, 191133888, 193104112, 195086128, 197079968,
+ 199085648, 201103184, 203132592, 205173888, 207227120, 209292272, 211369392, 213458480,
+ 215559568, 217672656, 219797792, 221934976, 224084240, 226245600, 228419056, 230604656,
+ 232802400, 235012320, 237234432, 239468736, 241715280, 243974080, 246245120, 248528464,
+ 250824112, 253132064, 255452368, 257785040, 260130080, 262487520, 264857376, 267239664,
+};
+
+static stbir_uint8 stbir__linear_to_srgb_uchar(float f)
+{
+ int x = (int) (f * (1 << 28)); // has headroom so you don't need to clamp
+ int v = 0;
+ int i;
+
+ // Refine the guess with a short binary search.
+ i = v + 128; if (x >= stbir__srgb_offset_to_linear_scaled[i]) v = i;
+ i = v + 64; if (x >= stbir__srgb_offset_to_linear_scaled[i]) v = i;
+ i = v + 32; if (x >= stbir__srgb_offset_to_linear_scaled[i]) v = i;
+ i = v + 16; if (x >= stbir__srgb_offset_to_linear_scaled[i]) v = i;
+ i = v + 8; if (x >= stbir__srgb_offset_to_linear_scaled[i]) v = i;
+ i = v + 4; if (x >= stbir__srgb_offset_to_linear_scaled[i]) v = i;
+ i = v + 2; if (x >= stbir__srgb_offset_to_linear_scaled[i]) v = i;
+ i = v + 1; if (x >= stbir__srgb_offset_to_linear_scaled[i]) v = i;
+
+ return (stbir_uint8) v;
+}
+#endif
+
+static float stbir__filter_trapezoid(float x, float scale)
+{
+ float halfscale = scale / 2;
+ float t = 0.5f + halfscale;
+ STBIR_ASSERT(scale <= 1);
+
+ x = (float)fabs(x);
+
+ if (x >= t)
+ return 0;
+ else
+ {
+ float r = 0.5f - halfscale;
+ if (x <= r)
+ return 1;
+ else
+ return (t - x) / scale;
+ }
+}
+
+static float stbir__support_trapezoid(float scale)
+{
+ STBIR_ASSERT(scale <= 1);
+ return 0.5f + scale / 2;
+}
+
+static float stbir__filter_triangle(float x, float s)
+{
+ STBIR__UNUSED_PARAM(s);
+
+ x = (float)fabs(x);
+
+ if (x <= 1.0f)
+ return 1 - x;
+ else
+ return 0;
+}
+
+static float stbir__filter_cubic(float x, float s)
+{
+ STBIR__UNUSED_PARAM(s);
+
+ x = (float)fabs(x);
+
+ if (x < 1.0f)
+ return (4 + x*x*(3*x - 6))/6;
+ else if (x < 2.0f)
+ return (8 + x*(-12 + x*(6 - x)))/6;
+
+ return (0.0f);
+}
+
+static float stbir__filter_catmullrom(float x, float s)
+{
+ STBIR__UNUSED_PARAM(s);
+
+ x = (float)fabs(x);
+
+ if (x < 1.0f)
+ return 1 - x*x*(2.5f - 1.5f*x);
+ else if (x < 2.0f)
+ return 2 - x*(4 + x*(0.5f*x - 2.5f));
+
+ return (0.0f);
+}
+
+static float stbir__filter_mitchell(float x, float s)
+{
+ STBIR__UNUSED_PARAM(s);
+
+ x = (float)fabs(x);
+
+ if (x < 1.0f)
+ return (16 + x*x*(21 * x - 36))/18;
+ else if (x < 2.0f)
+ return (32 + x*(-60 + x*(36 - 7*x)))/18;
+
+ return (0.0f);
+}
+
+static float stbir__support_zero(float s)
+{
+ STBIR__UNUSED_PARAM(s);
+ return 0;
+}
+
+static float stbir__support_one(float s)
+{
+ STBIR__UNUSED_PARAM(s);
+ return 1;
+}
+
+static float stbir__support_two(float s)
+{
+ STBIR__UNUSED_PARAM(s);
+ return 2;
+}
+
+static stbir__filter_info stbir__filter_info_table[] = {
+ { NULL, stbir__support_zero },
+ { stbir__filter_trapezoid, stbir__support_trapezoid },
+ { stbir__filter_triangle, stbir__support_one },
+ { stbir__filter_cubic, stbir__support_two },
+ { stbir__filter_catmullrom, stbir__support_two },
+ { stbir__filter_mitchell, stbir__support_two },
+};
+
+stbir__inline static int stbir__use_upsampling(float ratio)
+{
+ return ratio > 1;
+}
+
+stbir__inline static int stbir__use_width_upsampling(stbir__info* stbir_info)
+{
+ return stbir__use_upsampling(stbir_info->horizontal_scale);
+}
+
+stbir__inline static int stbir__use_height_upsampling(stbir__info* stbir_info)
+{
+ return stbir__use_upsampling(stbir_info->vertical_scale);
+}
+
+// This is the maximum number of input samples that can affect an output sample
+// with the given filter
+static int stbir__get_filter_pixel_width(stbir_filter filter, float scale)
+{
+ STBIR_ASSERT(filter != 0);
+ STBIR_ASSERT(filter < STBIR__ARRAY_SIZE(stbir__filter_info_table));
+
+ if (stbir__use_upsampling(scale))
+ return (int)ceil(stbir__filter_info_table[filter].support(1/scale) * 2);
+ else
+ return (int)ceil(stbir__filter_info_table[filter].support(scale) * 2 / scale);
+}
+
+// This is how much to expand buffers to account for filters seeking outside
+// the image boundaries.
+static int stbir__get_filter_pixel_margin(stbir_filter filter, float scale)
+{
+ return stbir__get_filter_pixel_width(filter, scale) / 2;
+}
+
+static int stbir__get_coefficient_width(stbir_filter filter, float scale)
+{
+ if (stbir__use_upsampling(scale))
+ return (int)ceil(stbir__filter_info_table[filter].support(1 / scale) * 2);
+ else
+ return (int)ceil(stbir__filter_info_table[filter].support(scale) * 2);
+}
+
+static int stbir__get_contributors(float scale, stbir_filter filter, int input_size, int output_size)
+{
+ if (stbir__use_upsampling(scale))
+ return output_size;
+ else
+ return (input_size + stbir__get_filter_pixel_margin(filter, scale) * 2);
+}
+
+static int stbir__get_total_horizontal_coefficients(stbir__info* info)
+{
+ return info->horizontal_num_contributors
+ * stbir__get_coefficient_width (info->horizontal_filter, info->horizontal_scale);
+}
+
+static int stbir__get_total_vertical_coefficients(stbir__info* info)
+{
+ return info->vertical_num_contributors
+ * stbir__get_coefficient_width (info->vertical_filter, info->vertical_scale);
+}
+
+static stbir__contributors* stbir__get_contributor(stbir__contributors* contributors, int n)
+{
+ return &contributors[n];
+}
+
+// For perf reasons this code is duplicated in stbir__resample_horizontal_upsample/downsample,
+// if you change it here change it there too.
+static float* stbir__get_coefficient(float* coefficients, stbir_filter filter, float scale, int n, int c)
+{
+ int width = stbir__get_coefficient_width(filter, scale);
+ return &coefficients[width*n + c];
+}
+
+static int stbir__edge_wrap_slow(stbir_edge edge, int n, int max)
+{
+ switch (edge)
+ {
+ case STBIR_EDGE_ZERO:
+ return 0; // we'll decode the wrong pixel here, and then overwrite with 0s later
+
+ case STBIR_EDGE_CLAMP:
+ if (n < 0)
+ return 0;
+
+ if (n >= max)
+ return max - 1;
+
+ return n; // NOTREACHED
+
+ case STBIR_EDGE_REFLECT:
+ {
+ if (n < 0)
+ {
+ if (n < max)
+ return -n;
+ else
+ return max - 1;
+ }
+
+ if (n >= max)
+ {
+ int max2 = max * 2;
+ if (n >= max2)
+ return 0;
+ else
+ return max2 - n - 1;
+ }
+
+ return n; // NOTREACHED
+ }
+
+ case STBIR_EDGE_WRAP:
+ if (n >= 0)
+ return (n % max);
+ else
+ {
+ int m = (-n) % max;
+
+ if (m != 0)
+ m = max - m;
+
+ return (m);
+ }
+ // NOTREACHED
+
+ default:
+ STBIR_ASSERT(!"Unimplemented edge type");
+ return 0;
+ }
+}
+
+stbir__inline static int stbir__edge_wrap(stbir_edge edge, int n, int max)
+{
+ // avoid per-pixel switch
+ if (n >= 0 && n < max)
+ return n;
+ return stbir__edge_wrap_slow(edge, n, max);
+}
+
+// What input pixels contribute to this output pixel?
+static void stbir__calculate_sample_range_upsample(int n, float out_filter_radius, float scale_ratio, float out_shift, int* in_first_pixel, int* in_last_pixel, float* in_center_of_out)
+{
+ float out_pixel_center = (float)n + 0.5f;
+ float out_pixel_influence_lowerbound = out_pixel_center - out_filter_radius;
+ float out_pixel_influence_upperbound = out_pixel_center + out_filter_radius;
+
+ float in_pixel_influence_lowerbound = (out_pixel_influence_lowerbound + out_shift) / scale_ratio;
+ float in_pixel_influence_upperbound = (out_pixel_influence_upperbound + out_shift) / scale_ratio;
+
+ *in_center_of_out = (out_pixel_center + out_shift) / scale_ratio;
+ *in_first_pixel = (int)(floor(in_pixel_influence_lowerbound + 0.5));
+ *in_last_pixel = (int)(floor(in_pixel_influence_upperbound - 0.5));
+}
+
+// What output pixels does this input pixel contribute to?
+static void stbir__calculate_sample_range_downsample(int n, float in_pixels_radius, float scale_ratio, float out_shift, int* out_first_pixel, int* out_last_pixel, float* out_center_of_in)
+{
+ float in_pixel_center = (float)n + 0.5f;
+ float in_pixel_influence_lowerbound = in_pixel_center - in_pixels_radius;
+ float in_pixel_influence_upperbound = in_pixel_center + in_pixels_radius;
+
+ float out_pixel_influence_lowerbound = in_pixel_influence_lowerbound * scale_ratio - out_shift;
+ float out_pixel_influence_upperbound = in_pixel_influence_upperbound * scale_ratio - out_shift;
+
+ *out_center_of_in = in_pixel_center * scale_ratio - out_shift;
+ *out_first_pixel = (int)(floor(out_pixel_influence_lowerbound + 0.5));
+ *out_last_pixel = (int)(floor(out_pixel_influence_upperbound - 0.5));
+}
+
+static void stbir__calculate_coefficients_upsample(stbir_filter filter, float scale, int in_first_pixel, int in_last_pixel, float in_center_of_out, stbir__contributors* contributor, float* coefficient_group)
+{
+ int i;
+ float total_filter = 0;
+ float filter_scale;
+
+ STBIR_ASSERT(in_last_pixel - in_first_pixel <= (int)ceil(stbir__filter_info_table[filter].support(1/scale) * 2)); // Taken directly from stbir__get_coefficient_width() which we can't call because we don't know if we're horizontal or vertical.
+
+ contributor->n0 = in_first_pixel;
+ contributor->n1 = in_last_pixel;
+
+ STBIR_ASSERT(contributor->n1 >= contributor->n0);
+
+ for (i = 0; i <= in_last_pixel - in_first_pixel; i++)
+ {
+ float in_pixel_center = (float)(i + in_first_pixel) + 0.5f;
+ coefficient_group[i] = stbir__filter_info_table[filter].kernel(in_center_of_out - in_pixel_center, 1 / scale);
+
+ // If the coefficient is zero, skip it. (Don't do the <0 check here, we want the influence of those outside pixels.)
+ if (i == 0 && !coefficient_group[i])
+ {
+ contributor->n0 = ++in_first_pixel;
+ i--;
+ continue;
+ }
+
+ total_filter += coefficient_group[i];
+ }
+
+ // NOTE(fg): Not actually true in general, nor is there any reason to expect it should be.
+ // It would be true in exact math but is at best approximately true in floating-point math,
+ // and it would not make sense to try and put actual bounds on this here because it depends
+ // on the image aspect ratio which can get pretty extreme.
+ //STBIR_ASSERT(stbir__filter_info_table[filter].kernel((float)(in_last_pixel + 1) + 0.5f - in_center_of_out, 1/scale) == 0);
+
+ STBIR_ASSERT(total_filter > 0.9);
+ STBIR_ASSERT(total_filter < 1.1f); // Make sure it's not way off.
+
+ // Make sure the sum of all coefficients is 1.
+ filter_scale = 1 / total_filter;
+
+ for (i = 0; i <= in_last_pixel - in_first_pixel; i++)
+ coefficient_group[i] *= filter_scale;
+
+ for (i = in_last_pixel - in_first_pixel; i >= 0; i--)
+ {
+ if (coefficient_group[i])
+ break;
+
+ // This line has no weight. We can skip it.
+ contributor->n1 = contributor->n0 + i - 1;
+ }
+}
+
+static void stbir__calculate_coefficients_downsample(stbir_filter filter, float scale_ratio, int out_first_pixel, int out_last_pixel, float out_center_of_in, stbir__contributors* contributor, float* coefficient_group)
+{
+ int i;
+
+ STBIR_ASSERT(out_last_pixel - out_first_pixel <= (int)ceil(stbir__filter_info_table[filter].support(scale_ratio) * 2)); // Taken directly from stbir__get_coefficient_width() which we can't call because we don't know if we're horizontal or vertical.
+
+ contributor->n0 = out_first_pixel;
+ contributor->n1 = out_last_pixel;
+
+ STBIR_ASSERT(contributor->n1 >= contributor->n0);
+
+ for (i = 0; i <= out_last_pixel - out_first_pixel; i++)
+ {
+ float out_pixel_center = (float)(i + out_first_pixel) + 0.5f;
+ float x = out_pixel_center - out_center_of_in;
+ coefficient_group[i] = stbir__filter_info_table[filter].kernel(x, scale_ratio) * scale_ratio;
+ }
+
+ // NOTE(fg): Not actually true in general, nor is there any reason to expect it should be.
+ // It would be true in exact math but is at best approximately true in floating-point math,
+ // and it would not make sense to try and put actual bounds on this here because it depends
+ // on the image aspect ratio which can get pretty extreme.
+ //STBIR_ASSERT(stbir__filter_info_table[filter].kernel((float)(out_last_pixel + 1) + 0.5f - out_center_of_in, scale_ratio) == 0);
+
+ for (i = out_last_pixel - out_first_pixel; i >= 0; i--)
+ {
+ if (coefficient_group[i])
+ break;
+
+ // This line has no weight. We can skip it.
+ contributor->n1 = contributor->n0 + i - 1;
+ }
+}
+
+static void stbir__normalize_downsample_coefficients(stbir__contributors* contributors, float* coefficients, stbir_filter filter, float scale_ratio, int input_size, int output_size)
+{
+ int num_contributors = stbir__get_contributors(scale_ratio, filter, input_size, output_size);
+ int num_coefficients = stbir__get_coefficient_width(filter, scale_ratio);
+ int i, j;
+ int skip;
+
+ for (i = 0; i < output_size; i++)
+ {
+ float scale;
+ float total = 0;
+
+ for (j = 0; j < num_contributors; j++)
+ {
+ if (i >= contributors[j].n0 && i <= contributors[j].n1)
+ {
+ float coefficient = *stbir__get_coefficient(coefficients, filter, scale_ratio, j, i - contributors[j].n0);
+ total += coefficient;
+ }
+ else if (i < contributors[j].n0)
+ break;
+ }
+
+ STBIR_ASSERT(total > 0.9f);
+ STBIR_ASSERT(total < 1.1f);
+
+ scale = 1 / total;
+
+ for (j = 0; j < num_contributors; j++)
+ {
+ if (i >= contributors[j].n0 && i <= contributors[j].n1)
+ *stbir__get_coefficient(coefficients, filter, scale_ratio, j, i - contributors[j].n0) *= scale;
+ else if (i < contributors[j].n0)
+ break;
+ }
+ }
+
+ // Optimize: Skip zero coefficients and contributions outside of image bounds.
+ // Do this after normalizing because normalization depends on the n0/n1 values.
+ for (j = 0; j < num_contributors; j++)
+ {
+ int range, max, width;
+
+ skip = 0;
+ while (*stbir__get_coefficient(coefficients, filter, scale_ratio, j, skip) == 0)
+ skip++;
+
+ contributors[j].n0 += skip;
+
+ while (contributors[j].n0 < 0)
+ {
+ contributors[j].n0++;
+ skip++;
+ }
+
+ range = contributors[j].n1 - contributors[j].n0 + 1;
+ max = stbir__min(num_coefficients, range);
+
+ width = stbir__get_coefficient_width(filter, scale_ratio);
+ for (i = 0; i < max; i++)
+ {
+ if (i + skip >= width)
+ break;
+
+ *stbir__get_coefficient(coefficients, filter, scale_ratio, j, i) = *stbir__get_coefficient(coefficients, filter, scale_ratio, j, i + skip);
+ }
+
+ continue;
+ }
+
+ // Using min to avoid writing into invalid pixels.
+ for (i = 0; i < num_contributors; i++)
+ contributors[i].n1 = stbir__min(contributors[i].n1, output_size - 1);
+}
+
+// Each scan line uses the same kernel values so we should calculate the kernel
+// values once and then we can use them for every scan line.
+static void stbir__calculate_filters(stbir__contributors* contributors, float* coefficients, stbir_filter filter, float scale_ratio, float shift, int input_size, int output_size)
+{
+ int n;
+ int total_contributors = stbir__get_contributors(scale_ratio, filter, input_size, output_size);
+
+ if (stbir__use_upsampling(scale_ratio))
+ {
+ float out_pixels_radius = stbir__filter_info_table[filter].support(1 / scale_ratio) * scale_ratio;
+
+ // Looping through out pixels
+ for (n = 0; n < total_contributors; n++)
+ {
+ float in_center_of_out; // Center of the current out pixel in the in pixel space
+ int in_first_pixel, in_last_pixel;
+
+ stbir__calculate_sample_range_upsample(n, out_pixels_radius, scale_ratio, shift, &in_first_pixel, &in_last_pixel, &in_center_of_out);
+
+ stbir__calculate_coefficients_upsample(filter, scale_ratio, in_first_pixel, in_last_pixel, in_center_of_out, stbir__get_contributor(contributors, n), stbir__get_coefficient(coefficients, filter, scale_ratio, n, 0));
+ }
+ }
+ else
+ {
+ float in_pixels_radius = stbir__filter_info_table[filter].support(scale_ratio) / scale_ratio;
+
+ // Looping through in pixels
+ for (n = 0; n < total_contributors; n++)
+ {
+ float out_center_of_in; // Center of the current out pixel in the in pixel space
+ int out_first_pixel, out_last_pixel;
+ int n_adjusted = n - stbir__get_filter_pixel_margin(filter, scale_ratio);
+
+ stbir__calculate_sample_range_downsample(n_adjusted, in_pixels_radius, scale_ratio, shift, &out_first_pixel, &out_last_pixel, &out_center_of_in);
+
+ stbir__calculate_coefficients_downsample(filter, scale_ratio, out_first_pixel, out_last_pixel, out_center_of_in, stbir__get_contributor(contributors, n), stbir__get_coefficient(coefficients, filter, scale_ratio, n, 0));
+ }
+
+ stbir__normalize_downsample_coefficients(contributors, coefficients, filter, scale_ratio, input_size, output_size);
+ }
+}
+
+static float* stbir__get_decode_buffer(stbir__info* stbir_info)
+{
+ // The 0 index of the decode buffer starts after the margin. This makes
+ // it okay to use negative indexes on the decode buffer.
+ return &stbir_info->decode_buffer[stbir_info->horizontal_filter_pixel_margin * stbir_info->channels];
+}
+
+#define STBIR__DECODE(type, colorspace) ((int)(type) * (STBIR_MAX_COLORSPACES) + (int)(colorspace))
+
+static void stbir__decode_scanline(stbir__info* stbir_info, int n)
+{
+ int c;
+ int channels = stbir_info->channels;
+ int alpha_channel = stbir_info->alpha_channel;
+ int type = stbir_info->type;
+ int colorspace = stbir_info->colorspace;
+ int input_w = stbir_info->input_w;
+ size_t input_stride_bytes = stbir_info->input_stride_bytes;
+ float* decode_buffer = stbir__get_decode_buffer(stbir_info);
+ stbir_edge edge_horizontal = stbir_info->edge_horizontal;
+ stbir_edge edge_vertical = stbir_info->edge_vertical;
+ size_t in_buffer_row_offset = stbir__edge_wrap(edge_vertical, n, stbir_info->input_h) * input_stride_bytes;
+ const void* input_data = (char *) stbir_info->input_data + in_buffer_row_offset;
+ int max_x = input_w + stbir_info->horizontal_filter_pixel_margin;
+ int decode = STBIR__DECODE(type, colorspace);
+
+ int x = -stbir_info->horizontal_filter_pixel_margin;
+
+ // special handling for STBIR_EDGE_ZERO because it needs to return an item that doesn't appear in the input,
+ // and we want to avoid paying overhead on every pixel if not STBIR_EDGE_ZERO
+ if (edge_vertical == STBIR_EDGE_ZERO && (n < 0 || n >= stbir_info->input_h))
+ {
+ for (; x < max_x; x++)
+ for (c = 0; c < channels; c++)
+ decode_buffer[x*channels + c] = 0;
+ return;
+ }
+
+ switch (decode)
+ {
+ case STBIR__DECODE(STBIR_TYPE_UINT8, STBIR_COLORSPACE_LINEAR):
+ for (; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+ int input_pixel_index = stbir__edge_wrap(edge_horizontal, x, input_w) * channels;
+ for (c = 0; c < channels; c++)
+ decode_buffer[decode_pixel_index + c] = ((float)((const unsigned char*)input_data)[input_pixel_index + c]) / stbir__max_uint8_as_float;
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT8, STBIR_COLORSPACE_SRGB):
+ for (; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+ int input_pixel_index = stbir__edge_wrap(edge_horizontal, x, input_w) * channels;
+ for (c = 0; c < channels; c++)
+ decode_buffer[decode_pixel_index + c] = stbir__srgb_uchar_to_linear_float[((const unsigned char*)input_data)[input_pixel_index + c]];
+
+ if (!(stbir_info->flags&STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ decode_buffer[decode_pixel_index + alpha_channel] = ((float)((const unsigned char*)input_data)[input_pixel_index + alpha_channel]) / stbir__max_uint8_as_float;
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT16, STBIR_COLORSPACE_LINEAR):
+ for (; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+ int input_pixel_index = stbir__edge_wrap(edge_horizontal, x, input_w) * channels;
+ for (c = 0; c < channels; c++)
+ decode_buffer[decode_pixel_index + c] = ((float)((const unsigned short*)input_data)[input_pixel_index + c]) / stbir__max_uint16_as_float;
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT16, STBIR_COLORSPACE_SRGB):
+ for (; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+ int input_pixel_index = stbir__edge_wrap(edge_horizontal, x, input_w) * channels;
+ for (c = 0; c < channels; c++)
+ decode_buffer[decode_pixel_index + c] = stbir__srgb_to_linear(((float)((const unsigned short*)input_data)[input_pixel_index + c]) / stbir__max_uint16_as_float);
+
+ if (!(stbir_info->flags&STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ decode_buffer[decode_pixel_index + alpha_channel] = ((float)((const unsigned short*)input_data)[input_pixel_index + alpha_channel]) / stbir__max_uint16_as_float;
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT32, STBIR_COLORSPACE_LINEAR):
+ for (; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+ int input_pixel_index = stbir__edge_wrap(edge_horizontal, x, input_w) * channels;
+ for (c = 0; c < channels; c++)
+ decode_buffer[decode_pixel_index + c] = (float)(((double)((const unsigned int*)input_data)[input_pixel_index + c]) / stbir__max_uint32_as_float);
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT32, STBIR_COLORSPACE_SRGB):
+ for (; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+ int input_pixel_index = stbir__edge_wrap(edge_horizontal, x, input_w) * channels;
+ for (c = 0; c < channels; c++)
+ decode_buffer[decode_pixel_index + c] = stbir__srgb_to_linear((float)(((double)((const unsigned int*)input_data)[input_pixel_index + c]) / stbir__max_uint32_as_float));
+
+ if (!(stbir_info->flags&STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ decode_buffer[decode_pixel_index + alpha_channel] = (float)(((double)((const unsigned int*)input_data)[input_pixel_index + alpha_channel]) / stbir__max_uint32_as_float);
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_FLOAT, STBIR_COLORSPACE_LINEAR):
+ for (; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+ int input_pixel_index = stbir__edge_wrap(edge_horizontal, x, input_w) * channels;
+ for (c = 0; c < channels; c++)
+ decode_buffer[decode_pixel_index + c] = ((const float*)input_data)[input_pixel_index + c];
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_FLOAT, STBIR_COLORSPACE_SRGB):
+ for (; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+ int input_pixel_index = stbir__edge_wrap(edge_horizontal, x, input_w) * channels;
+ for (c = 0; c < channels; c++)
+ decode_buffer[decode_pixel_index + c] = stbir__srgb_to_linear(((const float*)input_data)[input_pixel_index + c]);
+
+ if (!(stbir_info->flags&STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ decode_buffer[decode_pixel_index + alpha_channel] = ((const float*)input_data)[input_pixel_index + alpha_channel];
+ }
+
+ break;
+
+ default:
+ STBIR_ASSERT(!"Unknown type/colorspace/channels combination.");
+ break;
+ }
+
+ if (!(stbir_info->flags & STBIR_FLAG_ALPHA_PREMULTIPLIED))
+ {
+ for (x = -stbir_info->horizontal_filter_pixel_margin; x < max_x; x++)
+ {
+ int decode_pixel_index = x * channels;
+
+ // If the alpha value is 0 it will clobber the color values. Make sure it's not.
+ float alpha = decode_buffer[decode_pixel_index + alpha_channel];
+#ifndef STBIR_NO_ALPHA_EPSILON
+ if (stbir_info->type != STBIR_TYPE_FLOAT) {
+ alpha += STBIR_ALPHA_EPSILON;
+ decode_buffer[decode_pixel_index + alpha_channel] = alpha;
+ }
+#endif
+ for (c = 0; c < channels; c++)
+ {
+ if (c == alpha_channel)
+ continue;
+
+ decode_buffer[decode_pixel_index + c] *= alpha;
+ }
+ }
+ }
+
+ if (edge_horizontal == STBIR_EDGE_ZERO)
+ {
+ for (x = -stbir_info->horizontal_filter_pixel_margin; x < 0; x++)
+ {
+ for (c = 0; c < channels; c++)
+ decode_buffer[x*channels + c] = 0;
+ }
+ for (x = input_w; x < max_x; x++)
+ {
+ for (c = 0; c < channels; c++)
+ decode_buffer[x*channels + c] = 0;
+ }
+ }
+}
+
+static float* stbir__get_ring_buffer_entry(float* ring_buffer, int index, int ring_buffer_length)
+{
+ return &ring_buffer[index * ring_buffer_length];
+}
+
+static float* stbir__add_empty_ring_buffer_entry(stbir__info* stbir_info, int n)
+{
+ int ring_buffer_index;
+ float* ring_buffer;
+
+ stbir_info->ring_buffer_last_scanline = n;
+
+ if (stbir_info->ring_buffer_begin_index < 0)
+ {
+ ring_buffer_index = stbir_info->ring_buffer_begin_index = 0;
+ stbir_info->ring_buffer_first_scanline = n;
+ }
+ else
+ {
+ ring_buffer_index = (stbir_info->ring_buffer_begin_index + (stbir_info->ring_buffer_last_scanline - stbir_info->ring_buffer_first_scanline)) % stbir_info->ring_buffer_num_entries;
+ STBIR_ASSERT(ring_buffer_index != stbir_info->ring_buffer_begin_index);
+ }
+
+ ring_buffer = stbir__get_ring_buffer_entry(stbir_info->ring_buffer, ring_buffer_index, stbir_info->ring_buffer_length_bytes / sizeof(float));
+ memset(ring_buffer, 0, stbir_info->ring_buffer_length_bytes);
+
+ return ring_buffer;
+}
+
+
+static void stbir__resample_horizontal_upsample(stbir__info* stbir_info, float* output_buffer)
+{
+ int x, k;
+ int output_w = stbir_info->output_w;
+ int channels = stbir_info->channels;
+ float* decode_buffer = stbir__get_decode_buffer(stbir_info);
+ stbir__contributors* horizontal_contributors = stbir_info->horizontal_contributors;
+ float* horizontal_coefficients = stbir_info->horizontal_coefficients;
+ int coefficient_width = stbir_info->horizontal_coefficient_width;
+
+ for (x = 0; x < output_w; x++)
+ {
+ int n0 = horizontal_contributors[x].n0;
+ int n1 = horizontal_contributors[x].n1;
+
+ int out_pixel_index = x * channels;
+ int coefficient_group = coefficient_width * x;
+ int coefficient_counter = 0;
+
+ STBIR_ASSERT(n1 >= n0);
+ STBIR_ASSERT(n0 >= -stbir_info->horizontal_filter_pixel_margin);
+ STBIR_ASSERT(n1 >= -stbir_info->horizontal_filter_pixel_margin);
+ STBIR_ASSERT(n0 < stbir_info->input_w + stbir_info->horizontal_filter_pixel_margin);
+ STBIR_ASSERT(n1 < stbir_info->input_w + stbir_info->horizontal_filter_pixel_margin);
+
+ switch (channels) {
+ case 1:
+ for (k = n0; k <= n1; k++)
+ {
+ int in_pixel_index = k * 1;
+ float coefficient = horizontal_coefficients[coefficient_group + coefficient_counter++];
+ STBIR_ASSERT(coefficient != 0);
+ output_buffer[out_pixel_index + 0] += decode_buffer[in_pixel_index + 0] * coefficient;
+ }
+ break;
+ case 2:
+ for (k = n0; k <= n1; k++)
+ {
+ int in_pixel_index = k * 2;
+ float coefficient = horizontal_coefficients[coefficient_group + coefficient_counter++];
+ STBIR_ASSERT(coefficient != 0);
+ output_buffer[out_pixel_index + 0] += decode_buffer[in_pixel_index + 0] * coefficient;
+ output_buffer[out_pixel_index + 1] += decode_buffer[in_pixel_index + 1] * coefficient;
+ }
+ break;
+ case 3:
+ for (k = n0; k <= n1; k++)
+ {
+ int in_pixel_index = k * 3;
+ float coefficient = horizontal_coefficients[coefficient_group + coefficient_counter++];
+ STBIR_ASSERT(coefficient != 0);
+ output_buffer[out_pixel_index + 0] += decode_buffer[in_pixel_index + 0] * coefficient;
+ output_buffer[out_pixel_index + 1] += decode_buffer[in_pixel_index + 1] * coefficient;
+ output_buffer[out_pixel_index + 2] += decode_buffer[in_pixel_index + 2] * coefficient;
+ }
+ break;
+ case 4:
+ for (k = n0; k <= n1; k++)
+ {
+ int in_pixel_index = k * 4;
+ float coefficient = horizontal_coefficients[coefficient_group + coefficient_counter++];
+ STBIR_ASSERT(coefficient != 0);
+ output_buffer[out_pixel_index + 0] += decode_buffer[in_pixel_index + 0] * coefficient;
+ output_buffer[out_pixel_index + 1] += decode_buffer[in_pixel_index + 1] * coefficient;
+ output_buffer[out_pixel_index + 2] += decode_buffer[in_pixel_index + 2] * coefficient;
+ output_buffer[out_pixel_index + 3] += decode_buffer[in_pixel_index + 3] * coefficient;
+ }
+ break;
+ default:
+ for (k = n0; k <= n1; k++)
+ {
+ int in_pixel_index = k * channels;
+ float coefficient = horizontal_coefficients[coefficient_group + coefficient_counter++];
+ int c;
+ STBIR_ASSERT(coefficient != 0);
+ for (c = 0; c < channels; c++)
+ output_buffer[out_pixel_index + c] += decode_buffer[in_pixel_index + c] * coefficient;
+ }
+ break;
+ }
+ }
+}
+
+static void stbir__resample_horizontal_downsample(stbir__info* stbir_info, float* output_buffer)
+{
+ int x, k;
+ int input_w = stbir_info->input_w;
+ int channels = stbir_info->channels;
+ float* decode_buffer = stbir__get_decode_buffer(stbir_info);
+ stbir__contributors* horizontal_contributors = stbir_info->horizontal_contributors;
+ float* horizontal_coefficients = stbir_info->horizontal_coefficients;
+ int coefficient_width = stbir_info->horizontal_coefficient_width;
+ int filter_pixel_margin = stbir_info->horizontal_filter_pixel_margin;
+ int max_x = input_w + filter_pixel_margin * 2;
+
+ STBIR_ASSERT(!stbir__use_width_upsampling(stbir_info));
+
+ switch (channels) {
+ case 1:
+ for (x = 0; x < max_x; x++)
+ {
+ int n0 = horizontal_contributors[x].n0;
+ int n1 = horizontal_contributors[x].n1;
+
+ int in_x = x - filter_pixel_margin;
+ int in_pixel_index = in_x * 1;
+ int max_n = n1;
+ int coefficient_group = coefficient_width * x;
+
+ for (k = n0; k <= max_n; k++)
+ {
+ int out_pixel_index = k * 1;
+ float coefficient = horizontal_coefficients[coefficient_group + k - n0];
+ output_buffer[out_pixel_index + 0] += decode_buffer[in_pixel_index + 0] * coefficient;
+ }
+ }
+ break;
+
+ case 2:
+ for (x = 0; x < max_x; x++)
+ {
+ int n0 = horizontal_contributors[x].n0;
+ int n1 = horizontal_contributors[x].n1;
+
+ int in_x = x - filter_pixel_margin;
+ int in_pixel_index = in_x * 2;
+ int max_n = n1;
+ int coefficient_group = coefficient_width * x;
+
+ for (k = n0; k <= max_n; k++)
+ {
+ int out_pixel_index = k * 2;
+ float coefficient = horizontal_coefficients[coefficient_group + k - n0];
+ output_buffer[out_pixel_index + 0] += decode_buffer[in_pixel_index + 0] * coefficient;
+ output_buffer[out_pixel_index + 1] += decode_buffer[in_pixel_index + 1] * coefficient;
+ }
+ }
+ break;
+
+ case 3:
+ for (x = 0; x < max_x; x++)
+ {
+ int n0 = horizontal_contributors[x].n0;
+ int n1 = horizontal_contributors[x].n1;
+
+ int in_x = x - filter_pixel_margin;
+ int in_pixel_index = in_x * 3;
+ int max_n = n1;
+ int coefficient_group = coefficient_width * x;
+
+ for (k = n0; k <= max_n; k++)
+ {
+ int out_pixel_index = k * 3;
+ float coefficient = horizontal_coefficients[coefficient_group + k - n0];
+ output_buffer[out_pixel_index + 0] += decode_buffer[in_pixel_index + 0] * coefficient;
+ output_buffer[out_pixel_index + 1] += decode_buffer[in_pixel_index + 1] * coefficient;
+ output_buffer[out_pixel_index + 2] += decode_buffer[in_pixel_index + 2] * coefficient;
+ }
+ }
+ break;
+
+ case 4:
+ for (x = 0; x < max_x; x++)
+ {
+ int n0 = horizontal_contributors[x].n0;
+ int n1 = horizontal_contributors[x].n1;
+
+ int in_x = x - filter_pixel_margin;
+ int in_pixel_index = in_x * 4;
+ int max_n = n1;
+ int coefficient_group = coefficient_width * x;
+
+ for (k = n0; k <= max_n; k++)
+ {
+ int out_pixel_index = k * 4;
+ float coefficient = horizontal_coefficients[coefficient_group + k - n0];
+ output_buffer[out_pixel_index + 0] += decode_buffer[in_pixel_index + 0] * coefficient;
+ output_buffer[out_pixel_index + 1] += decode_buffer[in_pixel_index + 1] * coefficient;
+ output_buffer[out_pixel_index + 2] += decode_buffer[in_pixel_index + 2] * coefficient;
+ output_buffer[out_pixel_index + 3] += decode_buffer[in_pixel_index + 3] * coefficient;
+ }
+ }
+ break;
+
+ default:
+ for (x = 0; x < max_x; x++)
+ {
+ int n0 = horizontal_contributors[x].n0;
+ int n1 = horizontal_contributors[x].n1;
+
+ int in_x = x - filter_pixel_margin;
+ int in_pixel_index = in_x * channels;
+ int max_n = n1;
+ int coefficient_group = coefficient_width * x;
+
+ for (k = n0; k <= max_n; k++)
+ {
+ int c;
+ int out_pixel_index = k * channels;
+ float coefficient = horizontal_coefficients[coefficient_group + k - n0];
+ for (c = 0; c < channels; c++)
+ output_buffer[out_pixel_index + c] += decode_buffer[in_pixel_index + c] * coefficient;
+ }
+ }
+ break;
+ }
+}
+
+static void stbir__decode_and_resample_upsample(stbir__info* stbir_info, int n)
+{
+ // Decode the nth scanline from the source image into the decode buffer.
+ stbir__decode_scanline(stbir_info, n);
+
+ // Now resample it into the ring buffer.
+ if (stbir__use_width_upsampling(stbir_info))
+ stbir__resample_horizontal_upsample(stbir_info, stbir__add_empty_ring_buffer_entry(stbir_info, n));
+ else
+ stbir__resample_horizontal_downsample(stbir_info, stbir__add_empty_ring_buffer_entry(stbir_info, n));
+
+ // Now it's sitting in the ring buffer ready to be used as source for the vertical sampling.
+}
+
+static void stbir__decode_and_resample_downsample(stbir__info* stbir_info, int n)
+{
+ // Decode the nth scanline from the source image into the decode buffer.
+ stbir__decode_scanline(stbir_info, n);
+
+ memset(stbir_info->horizontal_buffer, 0, stbir_info->output_w * stbir_info->channels * sizeof(float));
+
+ // Now resample it into the horizontal buffer.
+ if (stbir__use_width_upsampling(stbir_info))
+ stbir__resample_horizontal_upsample(stbir_info, stbir_info->horizontal_buffer);
+ else
+ stbir__resample_horizontal_downsample(stbir_info, stbir_info->horizontal_buffer);
+
+ // Now it's sitting in the horizontal buffer ready to be distributed into the ring buffers.
+}
+
+// Get the specified scan line from the ring buffer.
+static float* stbir__get_ring_buffer_scanline(int get_scanline, float* ring_buffer, int begin_index, int first_scanline, int ring_buffer_num_entries, int ring_buffer_length)
+{
+ int ring_buffer_index = (begin_index + (get_scanline - first_scanline)) % ring_buffer_num_entries;
+ return stbir__get_ring_buffer_entry(ring_buffer, ring_buffer_index, ring_buffer_length);
+}
+
+
+static void stbir__encode_scanline(stbir__info* stbir_info, int num_pixels, void *output_buffer, float *encode_buffer, int channels, int alpha_channel, int decode)
+{
+ int x;
+ int n;
+ int num_nonalpha;
+ stbir_uint16 nonalpha[STBIR_MAX_CHANNELS];
+
+ if (!(stbir_info->flags&STBIR_FLAG_ALPHA_PREMULTIPLIED))
+ {
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ float alpha = encode_buffer[pixel_index + alpha_channel];
+ float reciprocal_alpha = alpha ? 1.0f / alpha : 0;
+
+ // unrolling this produced a 1% slowdown upscaling a large RGBA linear-space image on my machine - stb
+ for (n = 0; n < channels; n++)
+ if (n != alpha_channel)
+ encode_buffer[pixel_index + n] *= reciprocal_alpha;
+
+ // We added in a small epsilon to prevent the color channel from being deleted with zero alpha.
+ // Because we only add it for integer types, it will automatically be discarded on integer
+ // conversion, so we don't need to subtract it back out (which would be problematic for
+ // numeric precision reasons).
+ }
+ }
+
+ // build a table of all channels that need colorspace correction, so
+ // we don't perform colorspace correction on channels that don't need it.
+ for (x = 0, num_nonalpha = 0; x < channels; ++x)
+ {
+ if (x != alpha_channel || (stbir_info->flags & STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ {
+ nonalpha[num_nonalpha++] = (stbir_uint16)x;
+ }
+ }
+
+ #define STBIR__ROUND_INT(f) ((int) ((f)+0.5))
+ #define STBIR__ROUND_UINT(f) ((stbir_uint32) ((f)+0.5))
+
+ #ifdef STBIR__SATURATE_INT
+ #define STBIR__ENCODE_LINEAR8(f) stbir__saturate8 (STBIR__ROUND_INT((f) * stbir__max_uint8_as_float ))
+ #define STBIR__ENCODE_LINEAR16(f) stbir__saturate16(STBIR__ROUND_INT((f) * stbir__max_uint16_as_float))
+ #else
+ #define STBIR__ENCODE_LINEAR8(f) (unsigned char ) STBIR__ROUND_INT(stbir__saturate(f) * stbir__max_uint8_as_float )
+ #define STBIR__ENCODE_LINEAR16(f) (unsigned short) STBIR__ROUND_INT(stbir__saturate(f) * stbir__max_uint16_as_float)
+ #endif
+
+ switch (decode)
+ {
+ case STBIR__DECODE(STBIR_TYPE_UINT8, STBIR_COLORSPACE_LINEAR):
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ for (n = 0; n < channels; n++)
+ {
+ int index = pixel_index + n;
+ ((unsigned char*)output_buffer)[index] = STBIR__ENCODE_LINEAR8(encode_buffer[index]);
+ }
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT8, STBIR_COLORSPACE_SRGB):
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ for (n = 0; n < num_nonalpha; n++)
+ {
+ int index = pixel_index + nonalpha[n];
+ ((unsigned char*)output_buffer)[index] = stbir__linear_to_srgb_uchar(encode_buffer[index]);
+ }
+
+ if (!(stbir_info->flags & STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ ((unsigned char *)output_buffer)[pixel_index + alpha_channel] = STBIR__ENCODE_LINEAR8(encode_buffer[pixel_index+alpha_channel]);
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT16, STBIR_COLORSPACE_LINEAR):
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ for (n = 0; n < channels; n++)
+ {
+ int index = pixel_index + n;
+ ((unsigned short*)output_buffer)[index] = STBIR__ENCODE_LINEAR16(encode_buffer[index]);
+ }
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT16, STBIR_COLORSPACE_SRGB):
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ for (n = 0; n < num_nonalpha; n++)
+ {
+ int index = pixel_index + nonalpha[n];
+ ((unsigned short*)output_buffer)[index] = (unsigned short)STBIR__ROUND_INT(stbir__linear_to_srgb(stbir__saturate(encode_buffer[index])) * stbir__max_uint16_as_float);
+ }
+
+ if (!(stbir_info->flags&STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ ((unsigned short*)output_buffer)[pixel_index + alpha_channel] = STBIR__ENCODE_LINEAR16(encode_buffer[pixel_index + alpha_channel]);
+ }
+
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT32, STBIR_COLORSPACE_LINEAR):
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ for (n = 0; n < channels; n++)
+ {
+ int index = pixel_index + n;
+ ((unsigned int*)output_buffer)[index] = (unsigned int)STBIR__ROUND_UINT(((double)stbir__saturate(encode_buffer[index])) * stbir__max_uint32_as_float);
+ }
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_UINT32, STBIR_COLORSPACE_SRGB):
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ for (n = 0; n < num_nonalpha; n++)
+ {
+ int index = pixel_index + nonalpha[n];
+ ((unsigned int*)output_buffer)[index] = (unsigned int)STBIR__ROUND_UINT(((double)stbir__linear_to_srgb(stbir__saturate(encode_buffer[index]))) * stbir__max_uint32_as_float);
+ }
+
+ if (!(stbir_info->flags&STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ ((unsigned int*)output_buffer)[pixel_index + alpha_channel] = (unsigned int)STBIR__ROUND_INT(((double)stbir__saturate(encode_buffer[pixel_index + alpha_channel])) * stbir__max_uint32_as_float);
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_FLOAT, STBIR_COLORSPACE_LINEAR):
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ for (n = 0; n < channels; n++)
+ {
+ int index = pixel_index + n;
+ ((float*)output_buffer)[index] = encode_buffer[index];
+ }
+ }
+ break;
+
+ case STBIR__DECODE(STBIR_TYPE_FLOAT, STBIR_COLORSPACE_SRGB):
+ for (x=0; x < num_pixels; ++x)
+ {
+ int pixel_index = x*channels;
+
+ for (n = 0; n < num_nonalpha; n++)
+ {
+ int index = pixel_index + nonalpha[n];
+ ((float*)output_buffer)[index] = stbir__linear_to_srgb(encode_buffer[index]);
+ }
+
+ if (!(stbir_info->flags&STBIR_FLAG_ALPHA_USES_COLORSPACE))
+ ((float*)output_buffer)[pixel_index + alpha_channel] = encode_buffer[pixel_index + alpha_channel];
+ }
+ break;
+
+ default:
+ STBIR_ASSERT(!"Unknown type/colorspace/channels combination.");
+ break;
+ }
+}
+
+static void stbir__resample_vertical_upsample(stbir__info* stbir_info, int n)
+{
+ int x, k;
+ int output_w = stbir_info->output_w;
+ stbir__contributors* vertical_contributors = stbir_info->vertical_contributors;
+ float* vertical_coefficients = stbir_info->vertical_coefficients;
+ int channels = stbir_info->channels;
+ int alpha_channel = stbir_info->alpha_channel;
+ int type = stbir_info->type;
+ int colorspace = stbir_info->colorspace;
+ int ring_buffer_entries = stbir_info->ring_buffer_num_entries;
+ void* output_data = stbir_info->output_data;
+ float* encode_buffer = stbir_info->encode_buffer;
+ int decode = STBIR__DECODE(type, colorspace);
+ int coefficient_width = stbir_info->vertical_coefficient_width;
+ int coefficient_counter;
+ int contributor = n;
+
+ float* ring_buffer = stbir_info->ring_buffer;
+ int ring_buffer_begin_index = stbir_info->ring_buffer_begin_index;
+ int ring_buffer_first_scanline = stbir_info->ring_buffer_first_scanline;
+ int ring_buffer_length = stbir_info->ring_buffer_length_bytes/sizeof(float);
+
+ int n0,n1, output_row_start;
+ int coefficient_group = coefficient_width * contributor;
+
+ n0 = vertical_contributors[contributor].n0;
+ n1 = vertical_contributors[contributor].n1;
+
+ output_row_start = n * stbir_info->output_stride_bytes;
+
+ STBIR_ASSERT(stbir__use_height_upsampling(stbir_info));
+
+ memset(encode_buffer, 0, output_w * sizeof(float) * channels);
+
+ // I tried reblocking this for better cache usage of encode_buffer
+ // (using x_outer, k, x_inner), but it lost speed. -- stb
+
+ coefficient_counter = 0;
+ switch (channels) {
+ case 1:
+ for (k = n0; k <= n1; k++)
+ {
+ int coefficient_index = coefficient_counter++;
+ float* ring_buffer_entry = stbir__get_ring_buffer_scanline(k, ring_buffer, ring_buffer_begin_index, ring_buffer_first_scanline, ring_buffer_entries, ring_buffer_length);
+ float coefficient = vertical_coefficients[coefficient_group + coefficient_index];
+ for (x = 0; x < output_w; ++x)
+ {
+ int in_pixel_index = x * 1;
+ encode_buffer[in_pixel_index + 0] += ring_buffer_entry[in_pixel_index + 0] * coefficient;
+ }
+ }
+ break;
+ case 2:
+ for (k = n0; k <= n1; k++)
+ {
+ int coefficient_index = coefficient_counter++;
+ float* ring_buffer_entry = stbir__get_ring_buffer_scanline(k, ring_buffer, ring_buffer_begin_index, ring_buffer_first_scanline, ring_buffer_entries, ring_buffer_length);
+ float coefficient = vertical_coefficients[coefficient_group + coefficient_index];
+ for (x = 0; x < output_w; ++x)
+ {
+ int in_pixel_index = x * 2;
+ encode_buffer[in_pixel_index + 0] += ring_buffer_entry[in_pixel_index + 0] * coefficient;
+ encode_buffer[in_pixel_index + 1] += ring_buffer_entry[in_pixel_index + 1] * coefficient;
+ }
+ }
+ break;
+ case 3:
+ for (k = n0; k <= n1; k++)
+ {
+ int coefficient_index = coefficient_counter++;
+ float* ring_buffer_entry = stbir__get_ring_buffer_scanline(k, ring_buffer, ring_buffer_begin_index, ring_buffer_first_scanline, ring_buffer_entries, ring_buffer_length);
+ float coefficient = vertical_coefficients[coefficient_group + coefficient_index];
+ for (x = 0; x < output_w; ++x)
+ {
+ int in_pixel_index = x * 3;
+ encode_buffer[in_pixel_index + 0] += ring_buffer_entry[in_pixel_index + 0] * coefficient;
+ encode_buffer[in_pixel_index + 1] += ring_buffer_entry[in_pixel_index + 1] * coefficient;
+ encode_buffer[in_pixel_index + 2] += ring_buffer_entry[in_pixel_index + 2] * coefficient;
+ }
+ }
+ break;
+ case 4:
+ for (k = n0; k <= n1; k++)
+ {
+ int coefficient_index = coefficient_counter++;
+ float* ring_buffer_entry = stbir__get_ring_buffer_scanline(k, ring_buffer, ring_buffer_begin_index, ring_buffer_first_scanline, ring_buffer_entries, ring_buffer_length);
+ float coefficient = vertical_coefficients[coefficient_group + coefficient_index];
+ for (x = 0; x < output_w; ++x)
+ {
+ int in_pixel_index = x * 4;
+ encode_buffer[in_pixel_index + 0] += ring_buffer_entry[in_pixel_index + 0] * coefficient;
+ encode_buffer[in_pixel_index + 1] += ring_buffer_entry[in_pixel_index + 1] * coefficient;
+ encode_buffer[in_pixel_index + 2] += ring_buffer_entry[in_pixel_index + 2] * coefficient;
+ encode_buffer[in_pixel_index + 3] += ring_buffer_entry[in_pixel_index + 3] * coefficient;
+ }
+ }
+ break;
+ default:
+ for (k = n0; k <= n1; k++)
+ {
+ int coefficient_index = coefficient_counter++;
+ float* ring_buffer_entry = stbir__get_ring_buffer_scanline(k, ring_buffer, ring_buffer_begin_index, ring_buffer_first_scanline, ring_buffer_entries, ring_buffer_length);
+ float coefficient = vertical_coefficients[coefficient_group + coefficient_index];
+ for (x = 0; x < output_w; ++x)
+ {
+ int in_pixel_index = x * channels;
+ int c;
+ for (c = 0; c < channels; c++)
+ encode_buffer[in_pixel_index + c] += ring_buffer_entry[in_pixel_index + c] * coefficient;
+ }
+ }
+ break;
+ }
+ stbir__encode_scanline(stbir_info, output_w, (char *) output_data + output_row_start, encode_buffer, channels, alpha_channel, decode);
+}
+
+static void stbir__resample_vertical_downsample(stbir__info* stbir_info, int n)
+{
+ int x, k;
+ int output_w = stbir_info->output_w;
+ stbir__contributors* vertical_contributors = stbir_info->vertical_contributors;
+ float* vertical_coefficients = stbir_info->vertical_coefficients;
+ int channels = stbir_info->channels;
+ int ring_buffer_entries = stbir_info->ring_buffer_num_entries;
+ float* horizontal_buffer = stbir_info->horizontal_buffer;
+ int coefficient_width = stbir_info->vertical_coefficient_width;
+ int contributor = n + stbir_info->vertical_filter_pixel_margin;
+
+ float* ring_buffer = stbir_info->ring_buffer;
+ int ring_buffer_begin_index = stbir_info->ring_buffer_begin_index;
+ int ring_buffer_first_scanline = stbir_info->ring_buffer_first_scanline;
+ int ring_buffer_length = stbir_info->ring_buffer_length_bytes/sizeof(float);
+ int n0,n1;
+
+ n0 = vertical_contributors[contributor].n0;
+ n1 = vertical_contributors[contributor].n1;
+
+ STBIR_ASSERT(!stbir__use_height_upsampling(stbir_info));
+
+ for (k = n0; k <= n1; k++)
+ {
+ int coefficient_index = k - n0;
+ int coefficient_group = coefficient_width * contributor;
+ float coefficient = vertical_coefficients[coefficient_group + coefficient_index];
+
+ float* ring_buffer_entry = stbir__get_ring_buffer_scanline(k, ring_buffer, ring_buffer_begin_index, ring_buffer_first_scanline, ring_buffer_entries, ring_buffer_length);
+
+ switch (channels) {
+ case 1:
+ for (x = 0; x < output_w; x++)
+ {
+ int in_pixel_index = x * 1;
+ ring_buffer_entry[in_pixel_index + 0] += horizontal_buffer[in_pixel_index + 0] * coefficient;
+ }
+ break;
+ case 2:
+ for (x = 0; x < output_w; x++)
+ {
+ int in_pixel_index = x * 2;
+ ring_buffer_entry[in_pixel_index + 0] += horizontal_buffer[in_pixel_index + 0] * coefficient;
+ ring_buffer_entry[in_pixel_index + 1] += horizontal_buffer[in_pixel_index + 1] * coefficient;
+ }
+ break;
+ case 3:
+ for (x = 0; x < output_w; x++)
+ {
+ int in_pixel_index = x * 3;
+ ring_buffer_entry[in_pixel_index + 0] += horizontal_buffer[in_pixel_index + 0] * coefficient;
+ ring_buffer_entry[in_pixel_index + 1] += horizontal_buffer[in_pixel_index + 1] * coefficient;
+ ring_buffer_entry[in_pixel_index + 2] += horizontal_buffer[in_pixel_index + 2] * coefficient;
+ }
+ break;
+ case 4:
+ for (x = 0; x < output_w; x++)
+ {
+ int in_pixel_index = x * 4;
+ ring_buffer_entry[in_pixel_index + 0] += horizontal_buffer[in_pixel_index + 0] * coefficient;
+ ring_buffer_entry[in_pixel_index + 1] += horizontal_buffer[in_pixel_index + 1] * coefficient;
+ ring_buffer_entry[in_pixel_index + 2] += horizontal_buffer[in_pixel_index + 2] * coefficient;
+ ring_buffer_entry[in_pixel_index + 3] += horizontal_buffer[in_pixel_index + 3] * coefficient;
+ }
+ break;
+ default:
+ for (x = 0; x < output_w; x++)
+ {
+ int in_pixel_index = x * channels;
+
+ int c;
+ for (c = 0; c < channels; c++)
+ ring_buffer_entry[in_pixel_index + c] += horizontal_buffer[in_pixel_index + c] * coefficient;
+ }
+ break;
+ }
+ }
+}
+
+static void stbir__buffer_loop_upsample(stbir__info* stbir_info)
+{
+ int y;
+ float scale_ratio = stbir_info->vertical_scale;
+ float out_scanlines_radius = stbir__filter_info_table[stbir_info->vertical_filter].support(1/scale_ratio) * scale_ratio;
+
+ STBIR_ASSERT(stbir__use_height_upsampling(stbir_info));
+
+ for (y = 0; y < stbir_info->output_h; y++)
+ {
+ float in_center_of_out = 0; // Center of the current out scanline in the in scanline space
+ int in_first_scanline = 0, in_last_scanline = 0;
+
+ stbir__calculate_sample_range_upsample(y, out_scanlines_radius, scale_ratio, stbir_info->vertical_shift, &in_first_scanline, &in_last_scanline, &in_center_of_out);
+
+ STBIR_ASSERT(in_last_scanline - in_first_scanline + 1 <= stbir_info->ring_buffer_num_entries);
+
+ if (stbir_info->ring_buffer_begin_index >= 0)
+ {
+ // Get rid of whatever we don't need anymore.
+ while (in_first_scanline > stbir_info->ring_buffer_first_scanline)
+ {
+ if (stbir_info->ring_buffer_first_scanline == stbir_info->ring_buffer_last_scanline)
+ {
+ // We just popped the last scanline off the ring buffer.
+ // Reset it to the empty state.
+ stbir_info->ring_buffer_begin_index = -1;
+ stbir_info->ring_buffer_first_scanline = 0;
+ stbir_info->ring_buffer_last_scanline = 0;
+ break;
+ }
+ else
+ {
+ stbir_info->ring_buffer_first_scanline++;
+ stbir_info->ring_buffer_begin_index = (stbir_info->ring_buffer_begin_index + 1) % stbir_info->ring_buffer_num_entries;
+ }
+ }
+ }
+
+ // Load in new ones.
+ if (stbir_info->ring_buffer_begin_index < 0)
+ stbir__decode_and_resample_upsample(stbir_info, in_first_scanline);
+
+ while (in_last_scanline > stbir_info->ring_buffer_last_scanline)
+ stbir__decode_and_resample_upsample(stbir_info, stbir_info->ring_buffer_last_scanline + 1);
+
+ // Now all buffers should be ready to write a row of vertical sampling.
+ stbir__resample_vertical_upsample(stbir_info, y);
+
+ STBIR_PROGRESS_REPORT((float)y / stbir_info->output_h);
+ }
+}
+
+static void stbir__empty_ring_buffer(stbir__info* stbir_info, int first_necessary_scanline)
+{
+ int output_stride_bytes = stbir_info->output_stride_bytes;
+ int channels = stbir_info->channels;
+ int alpha_channel = stbir_info->alpha_channel;
+ int type = stbir_info->type;
+ int colorspace = stbir_info->colorspace;
+ int output_w = stbir_info->output_w;
+ void* output_data = stbir_info->output_data;
+ int decode = STBIR__DECODE(type, colorspace);
+
+ float* ring_buffer = stbir_info->ring_buffer;
+ int ring_buffer_length = stbir_info->ring_buffer_length_bytes/sizeof(float);
+
+ if (stbir_info->ring_buffer_begin_index >= 0)
+ {
+ // Get rid of whatever we don't need anymore.
+ while (first_necessary_scanline > stbir_info->ring_buffer_first_scanline)
+ {
+ if (stbir_info->ring_buffer_first_scanline >= 0 && stbir_info->ring_buffer_first_scanline < stbir_info->output_h)
+ {
+ int output_row_start = stbir_info->ring_buffer_first_scanline * output_stride_bytes;
+ float* ring_buffer_entry = stbir__get_ring_buffer_entry(ring_buffer, stbir_info->ring_buffer_begin_index, ring_buffer_length);
+ stbir__encode_scanline(stbir_info, output_w, (char *) output_data + output_row_start, ring_buffer_entry, channels, alpha_channel, decode);
+ STBIR_PROGRESS_REPORT((float)stbir_info->ring_buffer_first_scanline / stbir_info->output_h);
+ }
+
+ if (stbir_info->ring_buffer_first_scanline == stbir_info->ring_buffer_last_scanline)
+ {
+ // We just popped the last scanline off the ring buffer.
+ // Reset it to the empty state.
+ stbir_info->ring_buffer_begin_index = -1;
+ stbir_info->ring_buffer_first_scanline = 0;
+ stbir_info->ring_buffer_last_scanline = 0;
+ break;
+ }
+ else
+ {
+ stbir_info->ring_buffer_first_scanline++;
+ stbir_info->ring_buffer_begin_index = (stbir_info->ring_buffer_begin_index + 1) % stbir_info->ring_buffer_num_entries;
+ }
+ }
+ }
+}
+
+static void stbir__buffer_loop_downsample(stbir__info* stbir_info)
+{
+ int y;
+ float scale_ratio = stbir_info->vertical_scale;
+ int output_h = stbir_info->output_h;
+ float in_pixels_radius = stbir__filter_info_table[stbir_info->vertical_filter].support(scale_ratio) / scale_ratio;
+ int pixel_margin = stbir_info->vertical_filter_pixel_margin;
+ int max_y = stbir_info->input_h + pixel_margin;
+
+ STBIR_ASSERT(!stbir__use_height_upsampling(stbir_info));
+
+ for (y = -pixel_margin; y < max_y; y++)
+ {
+ float out_center_of_in; // Center of the current out scanline in the in scanline space
+ int out_first_scanline, out_last_scanline;
+
+ stbir__calculate_sample_range_downsample(y, in_pixels_radius, scale_ratio, stbir_info->vertical_shift, &out_first_scanline, &out_last_scanline, &out_center_of_in);
+
+ STBIR_ASSERT(out_last_scanline - out_first_scanline + 1 <= stbir_info->ring_buffer_num_entries);
+
+ if (out_last_scanline < 0 || out_first_scanline >= output_h)
+ continue;
+
+ stbir__empty_ring_buffer(stbir_info, out_first_scanline);
+
+ stbir__decode_and_resample_downsample(stbir_info, y);
+
+ // Load in new ones.
+ if (stbir_info->ring_buffer_begin_index < 0)
+ stbir__add_empty_ring_buffer_entry(stbir_info, out_first_scanline);
+
+ while (out_last_scanline > stbir_info->ring_buffer_last_scanline)
+ stbir__add_empty_ring_buffer_entry(stbir_info, stbir_info->ring_buffer_last_scanline + 1);
+
+ // Now the horizontal buffer is ready to write to all ring buffer rows.
+ stbir__resample_vertical_downsample(stbir_info, y);
+ }
+
+ stbir__empty_ring_buffer(stbir_info, stbir_info->output_h);
+}
+
+static void stbir__setup(stbir__info *info, int input_w, int input_h, int output_w, int output_h, int channels)
+{
+ info->input_w = input_w;
+ info->input_h = input_h;
+ info->output_w = output_w;
+ info->output_h = output_h;
+ info->channels = channels;
+}
+
+static void stbir__calculate_transform(stbir__info *info, float s0, float t0, float s1, float t1, float *transform)
+{
+ info->s0 = s0;
+ info->t0 = t0;
+ info->s1 = s1;
+ info->t1 = t1;
+
+ if (transform)
+ {
+ info->horizontal_scale = transform[0];
+ info->vertical_scale = transform[1];
+ info->horizontal_shift = transform[2];
+ info->vertical_shift = transform[3];
+ }
+ else
+ {
+ info->horizontal_scale = ((float)info->output_w / info->input_w) / (s1 - s0);
+ info->vertical_scale = ((float)info->output_h / info->input_h) / (t1 - t0);
+
+ info->horizontal_shift = s0 * info->output_w / (s1 - s0);
+ info->vertical_shift = t0 * info->output_h / (t1 - t0);
+ }
+}
+
+static void stbir__choose_filter(stbir__info *info, stbir_filter h_filter, stbir_filter v_filter)
+{
+ if (h_filter == 0)
+ h_filter = stbir__use_upsampling(info->horizontal_scale) ? STBIR_DEFAULT_FILTER_UPSAMPLE : STBIR_DEFAULT_FILTER_DOWNSAMPLE;
+ if (v_filter == 0)
+ v_filter = stbir__use_upsampling(info->vertical_scale) ? STBIR_DEFAULT_FILTER_UPSAMPLE : STBIR_DEFAULT_FILTER_DOWNSAMPLE;
+ info->horizontal_filter = h_filter;
+ info->vertical_filter = v_filter;
+}
+
+static stbir_uint32 stbir__calculate_memory(stbir__info *info)
+{
+ int pixel_margin = stbir__get_filter_pixel_margin(info->horizontal_filter, info->horizontal_scale);
+ int filter_height = stbir__get_filter_pixel_width(info->vertical_filter, info->vertical_scale);
+
+ info->horizontal_num_contributors = stbir__get_contributors(info->horizontal_scale, info->horizontal_filter, info->input_w, info->output_w);
+ info->vertical_num_contributors = stbir__get_contributors(info->vertical_scale , info->vertical_filter , info->input_h, info->output_h);
+
+ // One extra entry because floating point precision problems sometimes cause an extra to be necessary.
+ info->ring_buffer_num_entries = filter_height + 1;
+
+ info->horizontal_contributors_size = info->horizontal_num_contributors * sizeof(stbir__contributors);
+ info->horizontal_coefficients_size = stbir__get_total_horizontal_coefficients(info) * sizeof(float);
+ info->vertical_contributors_size = info->vertical_num_contributors * sizeof(stbir__contributors);
+ info->vertical_coefficients_size = stbir__get_total_vertical_coefficients(info) * sizeof(float);
+ info->decode_buffer_size = (info->input_w + pixel_margin * 2) * info->channels * sizeof(float);
+ info->horizontal_buffer_size = info->output_w * info->channels * sizeof(float);
+ info->ring_buffer_size = info->output_w * info->channels * info->ring_buffer_num_entries * sizeof(float);
+ info->encode_buffer_size = info->output_w * info->channels * sizeof(float);
+
+ STBIR_ASSERT(info->horizontal_filter != 0);
+ STBIR_ASSERT(info->horizontal_filter < STBIR__ARRAY_SIZE(stbir__filter_info_table)); // this now happens too late
+ STBIR_ASSERT(info->vertical_filter != 0);
+ STBIR_ASSERT(info->vertical_filter < STBIR__ARRAY_SIZE(stbir__filter_info_table)); // this now happens too late
+
+ if (stbir__use_height_upsampling(info))
+ // The horizontal buffer is for when we're downsampling the height and we
+ // can't output the result of sampling the decode buffer directly into the
+ // ring buffers.
+ info->horizontal_buffer_size = 0;
+ else
+ // The encode buffer is to retain precision in the height upsampling method
+ // and isn't used when height downsampling.
+ info->encode_buffer_size = 0;
+
+ return info->horizontal_contributors_size + info->horizontal_coefficients_size
+ + info->vertical_contributors_size + info->vertical_coefficients_size
+ + info->decode_buffer_size + info->horizontal_buffer_size
+ + info->ring_buffer_size + info->encode_buffer_size;
+}
+
+static int stbir__resize_allocated(stbir__info *info,
+ const void* input_data, int input_stride_in_bytes,
+ void* output_data, int output_stride_in_bytes,
+ int alpha_channel, stbir_uint32 flags, stbir_datatype type,
+ stbir_edge edge_horizontal, stbir_edge edge_vertical, stbir_colorspace colorspace,
+ void* tempmem, size_t tempmem_size_in_bytes)
+{
+ size_t memory_required = stbir__calculate_memory(info);
+
+ int width_stride_input = input_stride_in_bytes ? input_stride_in_bytes : info->channels * info->input_w * stbir__type_size[type];
+ int width_stride_output = output_stride_in_bytes ? output_stride_in_bytes : info->channels * info->output_w * stbir__type_size[type];
+
+#ifdef STBIR_DEBUG_OVERWRITE_TEST
+#define OVERWRITE_ARRAY_SIZE 8
+ unsigned char overwrite_output_before_pre[OVERWRITE_ARRAY_SIZE];
+ unsigned char overwrite_tempmem_before_pre[OVERWRITE_ARRAY_SIZE];
+ unsigned char overwrite_output_after_pre[OVERWRITE_ARRAY_SIZE];
+ unsigned char overwrite_tempmem_after_pre[OVERWRITE_ARRAY_SIZE];
+
+ size_t begin_forbidden = width_stride_output * (info->output_h - 1) + info->output_w * info->channels * stbir__type_size[type];
+ memcpy(overwrite_output_before_pre, &((unsigned char*)output_data)[-OVERWRITE_ARRAY_SIZE], OVERWRITE_ARRAY_SIZE);
+ memcpy(overwrite_output_after_pre, &((unsigned char*)output_data)[begin_forbidden], OVERWRITE_ARRAY_SIZE);
+ memcpy(overwrite_tempmem_before_pre, &((unsigned char*)tempmem)[-OVERWRITE_ARRAY_SIZE], OVERWRITE_ARRAY_SIZE);
+ memcpy(overwrite_tempmem_after_pre, &((unsigned char*)tempmem)[tempmem_size_in_bytes], OVERWRITE_ARRAY_SIZE);
+#endif
+
+ STBIR_ASSERT(info->channels >= 0);
+ STBIR_ASSERT(info->channels <= STBIR_MAX_CHANNELS);
+
+ if (info->channels < 0 || info->channels > STBIR_MAX_CHANNELS)
+ return 0;
+
+ STBIR_ASSERT(info->horizontal_filter < STBIR__ARRAY_SIZE(stbir__filter_info_table));
+ STBIR_ASSERT(info->vertical_filter < STBIR__ARRAY_SIZE(stbir__filter_info_table));
+
+ if (info->horizontal_filter >= STBIR__ARRAY_SIZE(stbir__filter_info_table))
+ return 0;
+ if (info->vertical_filter >= STBIR__ARRAY_SIZE(stbir__filter_info_table))
+ return 0;
+
+ if (alpha_channel < 0)
+ flags |= STBIR_FLAG_ALPHA_USES_COLORSPACE | STBIR_FLAG_ALPHA_PREMULTIPLIED;
+
+ if (!(flags&STBIR_FLAG_ALPHA_USES_COLORSPACE) || !(flags&STBIR_FLAG_ALPHA_PREMULTIPLIED)) {
+ STBIR_ASSERT(alpha_channel >= 0 && alpha_channel < info->channels);
+ }
+
+ if (alpha_channel >= info->channels)
+ return 0;
+
+ STBIR_ASSERT(tempmem);
+
+ if (!tempmem)
+ return 0;
+
+ STBIR_ASSERT(tempmem_size_in_bytes >= memory_required);
+
+ if (tempmem_size_in_bytes < memory_required)
+ return 0;
+
+ memset(tempmem, 0, tempmem_size_in_bytes);
+
+ info->input_data = input_data;
+ info->input_stride_bytes = width_stride_input;
+
+ info->output_data = output_data;
+ info->output_stride_bytes = width_stride_output;
+
+ info->alpha_channel = alpha_channel;
+ info->flags = flags;
+ info->type = type;
+ info->edge_horizontal = edge_horizontal;
+ info->edge_vertical = edge_vertical;
+ info->colorspace = colorspace;
+
+ info->horizontal_coefficient_width = stbir__get_coefficient_width (info->horizontal_filter, info->horizontal_scale);
+ info->vertical_coefficient_width = stbir__get_coefficient_width (info->vertical_filter , info->vertical_scale );
+ info->horizontal_filter_pixel_width = stbir__get_filter_pixel_width (info->horizontal_filter, info->horizontal_scale);
+ info->vertical_filter_pixel_width = stbir__get_filter_pixel_width (info->vertical_filter , info->vertical_scale );
+ info->horizontal_filter_pixel_margin = stbir__get_filter_pixel_margin(info->horizontal_filter, info->horizontal_scale);
+ info->vertical_filter_pixel_margin = stbir__get_filter_pixel_margin(info->vertical_filter , info->vertical_scale );
+
+ info->ring_buffer_length_bytes = info->output_w * info->channels * sizeof(float);
+ info->decode_buffer_pixels = info->input_w + info->horizontal_filter_pixel_margin * 2;
+
+#define STBIR__NEXT_MEMPTR(current, newtype) (newtype*)(((unsigned char*)current) + current##_size)
+
+ info->horizontal_contributors = (stbir__contributors *) tempmem;
+ info->horizontal_coefficients = STBIR__NEXT_MEMPTR(info->horizontal_contributors, float);
+ info->vertical_contributors = STBIR__NEXT_MEMPTR(info->horizontal_coefficients, stbir__contributors);
+ info->vertical_coefficients = STBIR__NEXT_MEMPTR(info->vertical_contributors, float);
+ info->decode_buffer = STBIR__NEXT_MEMPTR(info->vertical_coefficients, float);
+
+ if (stbir__use_height_upsampling(info))
+ {
+ info->horizontal_buffer = NULL;
+ info->ring_buffer = STBIR__NEXT_MEMPTR(info->decode_buffer, float);
+ info->encode_buffer = STBIR__NEXT_MEMPTR(info->ring_buffer, float);
+
+ STBIR_ASSERT((size_t)STBIR__NEXT_MEMPTR(info->encode_buffer, unsigned char) == (size_t)tempmem + tempmem_size_in_bytes);
+ }
+ else
+ {
+ info->horizontal_buffer = STBIR__NEXT_MEMPTR(info->decode_buffer, float);
+ info->ring_buffer = STBIR__NEXT_MEMPTR(info->horizontal_buffer, float);
+ info->encode_buffer = NULL;
+
+ STBIR_ASSERT((size_t)STBIR__NEXT_MEMPTR(info->ring_buffer, unsigned char) == (size_t)tempmem + tempmem_size_in_bytes);
+ }
+
+#undef STBIR__NEXT_MEMPTR
+
+ // This signals that the ring buffer is empty
+ info->ring_buffer_begin_index = -1;
+
+ stbir__calculate_filters(info->horizontal_contributors, info->horizontal_coefficients, info->horizontal_filter, info->horizontal_scale, info->horizontal_shift, info->input_w, info->output_w);
+ stbir__calculate_filters(info->vertical_contributors, info->vertical_coefficients, info->vertical_filter, info->vertical_scale, info->vertical_shift, info->input_h, info->output_h);
+
+ STBIR_PROGRESS_REPORT(0);
+
+ if (stbir__use_height_upsampling(info))
+ stbir__buffer_loop_upsample(info);
+ else
+ stbir__buffer_loop_downsample(info);
+
+ STBIR_PROGRESS_REPORT(1);
+
+#ifdef STBIR_DEBUG_OVERWRITE_TEST
+ STBIR_ASSERT(memcmp(overwrite_output_before_pre, &((unsigned char*)output_data)[-OVERWRITE_ARRAY_SIZE], OVERWRITE_ARRAY_SIZE) == 0);
+ STBIR_ASSERT(memcmp(overwrite_output_after_pre, &((unsigned char*)output_data)[begin_forbidden], OVERWRITE_ARRAY_SIZE) == 0);
+ STBIR_ASSERT(memcmp(overwrite_tempmem_before_pre, &((unsigned char*)tempmem)[-OVERWRITE_ARRAY_SIZE], OVERWRITE_ARRAY_SIZE) == 0);
+ STBIR_ASSERT(memcmp(overwrite_tempmem_after_pre, &((unsigned char*)tempmem)[tempmem_size_in_bytes], OVERWRITE_ARRAY_SIZE) == 0);
+#endif
+
+ return 1;
+}
+
+
+static int stbir__resize_arbitrary(
+ void *alloc_context,
+ const void* input_data, int input_w, int input_h, int input_stride_in_bytes,
+ void* output_data, int output_w, int output_h, int output_stride_in_bytes,
+ float s0, float t0, float s1, float t1, float *transform,
+ int channels, int alpha_channel, stbir_uint32 flags, stbir_datatype type,
+ stbir_filter h_filter, stbir_filter v_filter,
+ stbir_edge edge_horizontal, stbir_edge edge_vertical, stbir_colorspace colorspace)
+{
+ stbir__info info;
+ int result;
+ size_t memory_required;
+ void* extra_memory;
+
+ stbir__setup(&info, input_w, input_h, output_w, output_h, channels);
+ stbir__calculate_transform(&info, s0,t0,s1,t1,transform);
+ stbir__choose_filter(&info, h_filter, v_filter);
+ memory_required = stbir__calculate_memory(&info);
+ extra_memory = STBIR_MALLOC(memory_required, alloc_context);
+
+ if (!extra_memory)
+ return 0;
+
+ result = stbir__resize_allocated(&info, input_data, input_stride_in_bytes,
+ output_data, output_stride_in_bytes,
+ alpha_channel, flags, type,
+ edge_horizontal, edge_vertical,
+ colorspace, extra_memory, memory_required);
+
+ STBIR_FREE(extra_memory, alloc_context);
+
+ return result;
+}
+
+STBIRDEF int stbir_resize_uint8( const unsigned char *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ unsigned char *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels)
+{
+ return stbir__resize_arbitrary(NULL, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,NULL,num_channels,-1,0, STBIR_TYPE_UINT8, STBIR_FILTER_DEFAULT, STBIR_FILTER_DEFAULT,
+ STBIR_EDGE_CLAMP, STBIR_EDGE_CLAMP, STBIR_COLORSPACE_LINEAR);
+}
+
+STBIRDEF int stbir_resize_float( const float *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ float *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels)
+{
+ return stbir__resize_arbitrary(NULL, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,NULL,num_channels,-1,0, STBIR_TYPE_FLOAT, STBIR_FILTER_DEFAULT, STBIR_FILTER_DEFAULT,
+ STBIR_EDGE_CLAMP, STBIR_EDGE_CLAMP, STBIR_COLORSPACE_LINEAR);
+}
+
+STBIRDEF int stbir_resize_uint8_srgb(const unsigned char *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ unsigned char *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags)
+{
+ return stbir__resize_arbitrary(NULL, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,NULL,num_channels,alpha_channel,flags, STBIR_TYPE_UINT8, STBIR_FILTER_DEFAULT, STBIR_FILTER_DEFAULT,
+ STBIR_EDGE_CLAMP, STBIR_EDGE_CLAMP, STBIR_COLORSPACE_SRGB);
+}
+
+STBIRDEF int stbir_resize_uint8_srgb_edgemode(const unsigned char *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ unsigned char *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_wrap_mode)
+{
+ return stbir__resize_arbitrary(NULL, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,NULL,num_channels,alpha_channel,flags, STBIR_TYPE_UINT8, STBIR_FILTER_DEFAULT, STBIR_FILTER_DEFAULT,
+ edge_wrap_mode, edge_wrap_mode, STBIR_COLORSPACE_SRGB);
+}
+
+STBIRDEF int stbir_resize_uint8_generic( const unsigned char *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ unsigned char *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_wrap_mode, stbir_filter filter, stbir_colorspace space,
+ void *alloc_context)
+{
+ return stbir__resize_arbitrary(alloc_context, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,NULL,num_channels,alpha_channel,flags, STBIR_TYPE_UINT8, filter, filter,
+ edge_wrap_mode, edge_wrap_mode, space);
+}
+
+STBIRDEF int stbir_resize_uint16_generic(const stbir_uint16 *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ stbir_uint16 *output_pixels , int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_wrap_mode, stbir_filter filter, stbir_colorspace space,
+ void *alloc_context)
+{
+ return stbir__resize_arbitrary(alloc_context, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,NULL,num_channels,alpha_channel,flags, STBIR_TYPE_UINT16, filter, filter,
+ edge_wrap_mode, edge_wrap_mode, space);
+}
+
+
+STBIRDEF int stbir_resize_float_generic( const float *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ float *output_pixels , int output_w, int output_h, int output_stride_in_bytes,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_wrap_mode, stbir_filter filter, stbir_colorspace space,
+ void *alloc_context)
+{
+ return stbir__resize_arbitrary(alloc_context, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,NULL,num_channels,alpha_channel,flags, STBIR_TYPE_FLOAT, filter, filter,
+ edge_wrap_mode, edge_wrap_mode, space);
+}
+
+
+STBIRDEF int stbir_resize( const void *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ void *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ stbir_datatype datatype,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_mode_horizontal, stbir_edge edge_mode_vertical,
+ stbir_filter filter_horizontal, stbir_filter filter_vertical,
+ stbir_colorspace space, void *alloc_context)
+{
+ return stbir__resize_arbitrary(alloc_context, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,NULL,num_channels,alpha_channel,flags, datatype, filter_horizontal, filter_vertical,
+ edge_mode_horizontal, edge_mode_vertical, space);
+}
+
+
+STBIRDEF int stbir_resize_subpixel(const void *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ void *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ stbir_datatype datatype,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_mode_horizontal, stbir_edge edge_mode_vertical,
+ stbir_filter filter_horizontal, stbir_filter filter_vertical,
+ stbir_colorspace space, void *alloc_context,
+ float x_scale, float y_scale,
+ float x_offset, float y_offset)
+{
+ float transform[4];
+ transform[0] = x_scale;
+ transform[1] = y_scale;
+ transform[2] = x_offset;
+ transform[3] = y_offset;
+ return stbir__resize_arbitrary(alloc_context, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ 0,0,1,1,transform,num_channels,alpha_channel,flags, datatype, filter_horizontal, filter_vertical,
+ edge_mode_horizontal, edge_mode_vertical, space);
+}
+
+STBIRDEF int stbir_resize_region( const void *input_pixels , int input_w , int input_h , int input_stride_in_bytes,
+ void *output_pixels, int output_w, int output_h, int output_stride_in_bytes,
+ stbir_datatype datatype,
+ int num_channels, int alpha_channel, int flags,
+ stbir_edge edge_mode_horizontal, stbir_edge edge_mode_vertical,
+ stbir_filter filter_horizontal, stbir_filter filter_vertical,
+ stbir_colorspace space, void *alloc_context,
+ float s0, float t0, float s1, float t1)
+{
+ return stbir__resize_arbitrary(alloc_context, input_pixels, input_w, input_h, input_stride_in_bytes,
+ output_pixels, output_w, output_h, output_stride_in_bytes,
+ s0,t0,s1,t1,NULL,num_channels,alpha_channel,flags, datatype, filter_horizontal, filter_vertical,
+ edge_mode_horizontal, edge_mode_vertical, space);
+}
+
+#endif // STB_IMAGE_RESIZE_IMPLEMENTATION
+
+/*
+------------------------------------------------------------------------------
+This software is available under 2 licenses -- choose whichever you prefer.
+------------------------------------------------------------------------------
+ALTERNATIVE A - MIT License
+Copyright (c) 2017 Sean Barrett
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
+of the Software, and to permit persons to whom the Software is furnished to do
+so, subject to the following conditions:
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+------------------------------------------------------------------------------
+ALTERNATIVE B - Public Domain (www.unlicense.org)
+This is free and unencumbered software released into the public domain.
+Anyone is free to copy, modify, publish, use, compile, sell, or distribute this
+software, either in source code form or as a compiled binary, for any purpose,
+commercial or non-commercial, and by any means.
+In jurisdictions that recognize copyright laws, the author or authors of this
+software dedicate any and all copyright interest in the software to the public
+domain. We make this dedication for the benefit of the public at large and to
+the detriment of our heirs and successors. We intend this dedication to be an
+overt act of relinquishment in perpetuity of all present and future rights to
+this software under copyright law.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+------------------------------------------------------------------------------
+*/
diff --git a/ThirdParty/stb/stb_image_write.h b/ThirdParty/stb/stb_image_write.h
new file mode 100644
index 0000000..e4b32ed
--- /dev/null
+++ b/ThirdParty/stb/stb_image_write.h
@@ -0,0 +1,1724 @@
+/* stb_image_write - v1.16 - public domain - http://nothings.org/stb
+ writes out PNG/BMP/TGA/JPEG/HDR images to C stdio - Sean Barrett 2010-2015
+ no warranty implied; use at your own risk
+
+ Before #including,
+
+ #define STB_IMAGE_WRITE_IMPLEMENTATION
+
+ in the file that you want to have the implementation.
+
+ Will probably not work correctly with strict-aliasing optimizations.
+
+ABOUT:
+
+ This header file is a library for writing images to C stdio or a callback.
+
+ The PNG output is not optimal; it is 20-50% larger than the file
+ written by a decent optimizing implementation; though providing a custom
+ zlib compress function (see STBIW_ZLIB_COMPRESS) can mitigate that.
+ This library is designed for source code compactness and simplicity,
+ not optimal image file size or run-time performance.
+
+BUILDING:
+
+ You can #define STBIW_ASSERT(x) before the #include to avoid using assert.h.
+ You can #define STBIW_MALLOC(), STBIW_REALLOC(), and STBIW_FREE() to replace
+ malloc,realloc,free.
+ You can #define STBIW_MEMMOVE() to replace memmove()
+ You can #define STBIW_ZLIB_COMPRESS to use a custom zlib-style compress function
+ for PNG compression (instead of the builtin one), it must have the following signature:
+ unsigned char * my_compress(unsigned char *data, int data_len, int *out_len, int quality);
+ The returned data will be freed with STBIW_FREE() (free() by default),
+ so it must be heap allocated with STBIW_MALLOC() (malloc() by default),
+
+UNICODE:
+
+ If compiling for Windows and you wish to use Unicode filenames, compile
+ with
+ #define STBIW_WINDOWS_UTF8
+ and pass utf8-encoded filenames. Call stbiw_convert_wchar_to_utf8 to convert
+ Windows wchar_t filenames to utf8.
+
+USAGE:
+
+ There are five functions, one for each image file format:
+
+ int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes);
+ int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data);
+ int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data);
+ int stbi_write_jpg(char const *filename, int w, int h, int comp, const void *data, int quality);
+ int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data);
+
+ void stbi_flip_vertically_on_write(int flag); // flag is non-zero to flip data vertically
+
+ There are also five equivalent functions that use an arbitrary write function. You are
+ expected to open/close your file-equivalent before and after calling these:
+
+ int stbi_write_png_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data, int stride_in_bytes);
+ int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data);
+ int stbi_write_tga_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data);
+ int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const float *data);
+ int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality);
+
+ where the callback is:
+ void stbi_write_func(void *context, void *data, int size);
+
+ You can configure it with these global variables:
+ int stbi_write_tga_with_rle; // defaults to true; set to 0 to disable RLE
+ int stbi_write_png_compression_level; // defaults to 8; set to higher for more compression
+ int stbi_write_force_png_filter; // defaults to -1; set to 0..5 to force a filter mode
+
+
+ You can define STBI_WRITE_NO_STDIO to disable the file variant of these
+ functions, so the library will not use stdio.h at all. However, this will
+ also disable HDR writing, because it requires stdio for formatted output.
+
+ Each function returns 0 on failure and non-0 on success.
+
+ The functions create an image file defined by the parameters. The image
+ is a rectangle of pixels stored from left-to-right, top-to-bottom.
+ Each pixel contains 'comp' channels of data stored interleaved with 8-bits
+ per channel, in the following order: 1=Y, 2=YA, 3=RGB, 4=RGBA. (Y is
+ monochrome color.) The rectangle is 'w' pixels wide and 'h' pixels tall.
+ The *data pointer points to the first byte of the top-left-most pixel.
+ For PNG, "stride_in_bytes" is the distance in bytes from the first byte of
+ a row of pixels to the first byte of the next row of pixels.
+
+ PNG creates output files with the same number of components as the input.
+ The BMP format expands Y to RGB in the file format and does not
+ output alpha.
+
+ PNG supports writing rectangles of data even when the bytes storing rows of
+ data are not consecutive in memory (e.g. sub-rectangles of a larger image),
+ by supplying the stride between the beginning of adjacent rows. The other
+ formats do not. (Thus you cannot write a native-format BMP through the BMP
+ writer, both because it is in BGR order and because it may have padding
+ at the end of the line.)
+
+ PNG allows you to set the deflate compression level by setting the global
+ variable 'stbi_write_png_compression_level' (it defaults to 8).
+
+ HDR expects linear float data. Since the format is always 32-bit rgb(e)
+ data, alpha (if provided) is discarded, and for monochrome data it is
+ replicated across all three channels.
+
+ TGA supports RLE or non-RLE compressed data. To use non-RLE-compressed
+ data, set the global variable 'stbi_write_tga_with_rle' to 0.
+
+ JPEG does ignore alpha channels in input data; quality is between 1 and 100.
+ Higher quality looks better but results in a bigger image.
+ JPEG baseline (no JPEG progressive).
+
+CREDITS:
+
+
+ Sean Barrett - PNG/BMP/TGA
+ Baldur Karlsson - HDR
+ Jean-Sebastien Guay - TGA monochrome
+ Tim Kelsey - misc enhancements
+ Alan Hickman - TGA RLE
+ Emmanuel Julien - initial file IO callback implementation
+ Jon Olick - original jo_jpeg.cpp code
+ Daniel Gibson - integrate JPEG, allow external zlib
+ Aarni Koskela - allow choosing PNG filter
+
+ bugfixes:
+ github:Chribba
+ Guillaume Chereau
+ github:jry2
+ github:romigrou
+ Sergio Gonzalez
+ Jonas Karlsson
+ Filip Wasil
+ Thatcher Ulrich
+ github:poppolopoppo
+ Patrick Boettcher
+ github:xeekworx
+ Cap Petschulat
+ Simon Rodriguez
+ Ivan Tikhonov
+ github:ignotion
+ Adam Schackart
+ Andrew Kensler
+
+LICENSE
+
+ See end of file for license information.
+
+*/
+
+#ifndef INCLUDE_STB_IMAGE_WRITE_H
+#define INCLUDE_STB_IMAGE_WRITE_H
+
+#include <stdlib.h>
+
+// if STB_IMAGE_WRITE_STATIC causes problems, try defining STBIWDEF to 'inline' or 'static inline'
+#ifndef STBIWDEF
+#ifdef STB_IMAGE_WRITE_STATIC
+#define STBIWDEF static
+#else
+#ifdef __cplusplus
+#define STBIWDEF extern "C"
+#else
+#define STBIWDEF extern
+#endif
+#endif
+#endif
+
+#ifndef STB_IMAGE_WRITE_STATIC // C++ forbids static forward declarations
+STBIWDEF int stbi_write_tga_with_rle;
+STBIWDEF int stbi_write_png_compression_level;
+STBIWDEF int stbi_write_force_png_filter;
+#endif
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes);
+STBIWDEF int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data);
+STBIWDEF int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data);
+STBIWDEF int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data);
+STBIWDEF int stbi_write_jpg(char const *filename, int x, int y, int comp, const void *data, int quality);
+
+#ifdef STBIW_WINDOWS_UTF8
+STBIWDEF int stbiw_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input);
+#endif
+#endif
+
+typedef void stbi_write_func(void *context, void *data, int size);
+
+STBIWDEF int stbi_write_png_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data, int stride_in_bytes);
+STBIWDEF int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data);
+STBIWDEF int stbi_write_tga_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data);
+STBIWDEF int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const float *data);
+STBIWDEF int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality);
+
+STBIWDEF void stbi_flip_vertically_on_write(int flip_boolean);
+
+#endif//INCLUDE_STB_IMAGE_WRITE_H
+
+#ifdef STB_IMAGE_WRITE_IMPLEMENTATION
+
+#ifdef _WIN32
+ #ifndef _CRT_SECURE_NO_WARNINGS
+ #define _CRT_SECURE_NO_WARNINGS
+ #endif
+ #ifndef _CRT_NONSTDC_NO_DEPRECATE
+ #define _CRT_NONSTDC_NO_DEPRECATE
+ #endif
+#endif
+
+#ifndef STBI_WRITE_NO_STDIO
+#include <stdio.h>
+#endif // STBI_WRITE_NO_STDIO
+
+#include <stdarg.h>
+#include <stdlib.h>
+#include <string.h>
+#include <math.h>
+
+#if defined(STBIW_MALLOC) && defined(STBIW_FREE) && (defined(STBIW_REALLOC) || defined(STBIW_REALLOC_SIZED))
+// ok
+#elif !defined(STBIW_MALLOC) && !defined(STBIW_FREE) && !defined(STBIW_REALLOC) && !defined(STBIW_REALLOC_SIZED)
+// ok
+#else
+#error "Must define all or none of STBIW_MALLOC, STBIW_FREE, and STBIW_REALLOC (or STBIW_REALLOC_SIZED)."
+#endif
+
+#ifndef STBIW_MALLOC
+#define STBIW_MALLOC(sz) malloc(sz)
+#define STBIW_REALLOC(p,newsz) realloc(p,newsz)
+#define STBIW_FREE(p) free(p)
+#endif
+
+#ifndef STBIW_REALLOC_SIZED
+#define STBIW_REALLOC_SIZED(p,oldsz,newsz) STBIW_REALLOC(p,newsz)
+#endif
+
+
+#ifndef STBIW_MEMMOVE
+#define STBIW_MEMMOVE(a,b,sz) memmove(a,b,sz)
+#endif
+
+
+#ifndef STBIW_ASSERT
+#include <assert.h>
+#define STBIW_ASSERT(x) assert(x)
+#endif
+
+#define STBIW_UCHAR(x) (unsigned char) ((x) & 0xff)
+
+#ifdef STB_IMAGE_WRITE_STATIC
+static int stbi_write_png_compression_level = 8;
+static int stbi_write_tga_with_rle = 1;
+static int stbi_write_force_png_filter = -1;
+#else
+int stbi_write_png_compression_level = 8;
+int stbi_write_tga_with_rle = 1;
+int stbi_write_force_png_filter = -1;
+#endif
+
+static int stbi__flip_vertically_on_write = 0;
+
+STBIWDEF void stbi_flip_vertically_on_write(int flag)
+{
+ stbi__flip_vertically_on_write = flag;
+}
+
+typedef struct
+{
+ stbi_write_func *func;
+ void *context;
+ unsigned char buffer[64];
+ int buf_used;
+} stbi__write_context;
+
+// initialize a callback-based context
+static void stbi__start_write_callbacks(stbi__write_context *s, stbi_write_func *c, void *context)
+{
+ s->func = c;
+ s->context = context;
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+
+static void stbi__stdio_write(void *context, void *data, int size)
+{
+ fwrite(data,1,size,(FILE*) context);
+}
+
+#if defined(_WIN32) && defined(STBIW_WINDOWS_UTF8)
+#ifdef __cplusplus
+#define STBIW_EXTERN extern "C"
+#else
+#define STBIW_EXTERN extern
+#endif
+STBIW_EXTERN __declspec(dllimport) int __stdcall MultiByteToWideChar(unsigned int cp, unsigned long flags, const char *str, int cbmb, wchar_t *widestr, int cchwide);
+STBIW_EXTERN __declspec(dllimport) int __stdcall WideCharToMultiByte(unsigned int cp, unsigned long flags, const wchar_t *widestr, int cchwide, char *str, int cbmb, const char *defchar, int *used_default);
+
+STBIWDEF int stbiw_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input)
+{
+ return WideCharToMultiByte(65001 /* UTF8 */, 0, input, -1, buffer, (int) bufferlen, NULL, NULL);
+}
+#endif
+
+static FILE *stbiw__fopen(char const *filename, char const *mode)
+{
+ FILE *f;
+#if defined(_WIN32) && defined(STBIW_WINDOWS_UTF8)
+ wchar_t wMode[64];
+ wchar_t wFilename[1024];
+ if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, filename, -1, wFilename, sizeof(wFilename)/sizeof(*wFilename)))
+ return 0;
+
+ if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, mode, -1, wMode, sizeof(wMode)/sizeof(*wMode)))
+ return 0;
+
+#if defined(_MSC_VER) && _MSC_VER >= 1400
+ if (0 != _wfopen_s(&f, wFilename, wMode))
+ f = 0;
+#else
+ f = _wfopen(wFilename, wMode);
+#endif
+
+#elif defined(_MSC_VER) && _MSC_VER >= 1400
+ if (0 != fopen_s(&f, filename, mode))
+ f=0;
+#else
+ f = fopen(filename, mode);
+#endif
+ return f;
+}
+
+static int stbi__start_write_file(stbi__write_context *s, const char *filename)
+{
+ FILE *f = stbiw__fopen(filename, "wb");
+ stbi__start_write_callbacks(s, stbi__stdio_write, (void *) f);
+ return f != NULL;
+}
+
+static void stbi__end_write_file(stbi__write_context *s)
+{
+ fclose((FILE *)s->context);
+}
+
+#endif // !STBI_WRITE_NO_STDIO
+
+typedef unsigned int stbiw_uint32;
+typedef int stb_image_write_test[sizeof(stbiw_uint32)==4 ? 1 : -1];
+
+static void stbiw__writefv(stbi__write_context *s, const char *fmt, va_list v)
+{
+ while (*fmt) {
+ switch (*fmt++) {
+ case ' ': break;
+ case '1': { unsigned char x = STBIW_UCHAR(va_arg(v, int));
+ s->func(s->context,&x,1);
+ break; }
+ case '2': { int x = va_arg(v,int);
+ unsigned char b[2];
+ b[0] = STBIW_UCHAR(x);
+ b[1] = STBIW_UCHAR(x>>8);
+ s->func(s->context,b,2);
+ break; }
+ case '4': { stbiw_uint32 x = va_arg(v,int);
+ unsigned char b[4];
+ b[0]=STBIW_UCHAR(x);
+ b[1]=STBIW_UCHAR(x>>8);
+ b[2]=STBIW_UCHAR(x>>16);
+ b[3]=STBIW_UCHAR(x>>24);
+ s->func(s->context,b,4);
+ break; }
+ default:
+ STBIW_ASSERT(0);
+ return;
+ }
+ }
+}
+
+static void stbiw__writef(stbi__write_context *s, const char *fmt, ...)
+{
+ va_list v;
+ va_start(v, fmt);
+ stbiw__writefv(s, fmt, v);
+ va_end(v);
+}
+
+static void stbiw__write_flush(stbi__write_context *s)
+{
+ if (s->buf_used) {
+ s->func(s->context, &s->buffer, s->buf_used);
+ s->buf_used = 0;
+ }
+}
+
+static void stbiw__putc(stbi__write_context *s, unsigned char c)
+{
+ s->func(s->context, &c, 1);
+}
+
+static void stbiw__write1(stbi__write_context *s, unsigned char a)
+{
+ if ((size_t)s->buf_used + 1 > sizeof(s->buffer))
+ stbiw__write_flush(s);
+ s->buffer[s->buf_used++] = a;
+}
+
+static void stbiw__write3(stbi__write_context *s, unsigned char a, unsigned char b, unsigned char c)
+{
+ int n;
+ if ((size_t)s->buf_used + 3 > sizeof(s->buffer))
+ stbiw__write_flush(s);
+ n = s->buf_used;
+ s->buf_used = n+3;
+ s->buffer[n+0] = a;
+ s->buffer[n+1] = b;
+ s->buffer[n+2] = c;
+}
+
+static void stbiw__write_pixel(stbi__write_context *s, int rgb_dir, int comp, int write_alpha, int expand_mono, unsigned char *d)
+{
+ unsigned char bg[3] = { 255, 0, 255}, px[3];
+ int k;
+
+ if (write_alpha < 0)
+ stbiw__write1(s, d[comp - 1]);
+
+ switch (comp) {
+ case 2: // 2 pixels = mono + alpha, alpha is written separately, so same as 1-channel case
+ case 1:
+ if (expand_mono)
+ stbiw__write3(s, d[0], d[0], d[0]); // monochrome bmp
+ else
+ stbiw__write1(s, d[0]); // monochrome TGA
+ break;
+ case 4:
+ if (!write_alpha) {
+ // composite against pink background
+ for (k = 0; k < 3; ++k)
+ px[k] = bg[k] + ((d[k] - bg[k]) * d[3]) / 255;
+ stbiw__write3(s, px[1 - rgb_dir], px[1], px[1 + rgb_dir]);
+ break;
+ }
+ /* FALLTHROUGH */
+ case 3:
+ stbiw__write3(s, d[1 - rgb_dir], d[1], d[1 + rgb_dir]);
+ break;
+ }
+ if (write_alpha > 0)
+ stbiw__write1(s, d[comp - 1]);
+}
+
+static void stbiw__write_pixels(stbi__write_context *s, int rgb_dir, int vdir, int x, int y, int comp, void *data, int write_alpha, int scanline_pad, int expand_mono)
+{
+ stbiw_uint32 zero = 0;
+ int i,j, j_end;
+
+ if (y <= 0)
+ return;
+
+ if (stbi__flip_vertically_on_write)
+ vdir *= -1;
+
+ if (vdir < 0) {
+ j_end = -1; j = y-1;
+ } else {
+ j_end = y; j = 0;
+ }
+
+ for (; j != j_end; j += vdir) {
+ for (i=0; i < x; ++i) {
+ unsigned char *d = (unsigned char *) data + (j*x+i)*comp;
+ stbiw__write_pixel(s, rgb_dir, comp, write_alpha, expand_mono, d);
+ }
+ stbiw__write_flush(s);
+ s->func(s->context, &zero, scanline_pad);
+ }
+}
+
+static int stbiw__outfile(stbi__write_context *s, int rgb_dir, int vdir, int x, int y, int comp, int expand_mono, void *data, int alpha, int pad, const char *fmt, ...)
+{
+ if (y < 0 || x < 0) {
+ return 0;
+ } else {
+ va_list v;
+ va_start(v, fmt);
+ stbiw__writefv(s, fmt, v);
+ va_end(v);
+ stbiw__write_pixels(s,rgb_dir,vdir,x,y,comp,data,alpha,pad, expand_mono);
+ return 1;
+ }
+}
+
+static int stbi_write_bmp_core(stbi__write_context *s, int x, int y, int comp, const void *data)
+{
+ if (comp != 4) {
+ // write RGB bitmap
+ int pad = (-x*3) & 3;
+ return stbiw__outfile(s,-1,-1,x,y,comp,1,(void *) data,0,pad,
+ "11 4 22 4" "4 44 22 444444",
+ 'B', 'M', 14+40+(x*3+pad)*y, 0,0, 14+40, // file header
+ 40, x,y, 1,24, 0,0,0,0,0,0); // bitmap header
+ } else {
+ // RGBA bitmaps need a v4 header
+ // use BI_BITFIELDS mode with 32bpp and alpha mask
+ // (straight BI_RGB with alpha mask doesn't work in most readers)
+ return stbiw__outfile(s,-1,-1,x,y,comp,1,(void *)data,1,0,
+ "11 4 22 4" "4 44 22 444444 4444 4 444 444 444 444",
+ 'B', 'M', 14+108+x*y*4, 0, 0, 14+108, // file header
+ 108, x,y, 1,32, 3,0,0,0,0,0, 0xff0000,0xff00,0xff,0xff000000u, 0, 0,0,0, 0,0,0, 0,0,0, 0,0,0); // bitmap V4 header
+ }
+}
+
+STBIWDEF int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data)
+{
+ stbi__write_context s = { 0 };
+ stbi__start_write_callbacks(&s, func, context);
+ return stbi_write_bmp_core(&s, x, y, comp, data);
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_bmp(char const *filename, int x, int y, int comp, const void *data)
+{
+ stbi__write_context s = { 0 };
+ if (stbi__start_write_file(&s,filename)) {
+ int r = stbi_write_bmp_core(&s, x, y, comp, data);
+ stbi__end_write_file(&s);
+ return r;
+ } else
+ return 0;
+}
+#endif //!STBI_WRITE_NO_STDIO
+
+static int stbi_write_tga_core(stbi__write_context *s, int x, int y, int comp, void *data)
+{
+ int has_alpha = (comp == 2 || comp == 4);
+ int colorbytes = has_alpha ? comp-1 : comp;
+ int format = colorbytes < 2 ? 3 : 2; // 3 color channels (RGB/RGBA) = 2, 1 color channel (Y/YA) = 3
+
+ if (y < 0 || x < 0)
+ return 0;
+
+ if (!stbi_write_tga_with_rle) {
+ return stbiw__outfile(s, -1, -1, x, y, comp, 0, (void *) data, has_alpha, 0,
+ "111 221 2222 11", 0, 0, format, 0, 0, 0, 0, 0, x, y, (colorbytes + has_alpha) * 8, has_alpha * 8);
+ } else {
+ int i,j,k;
+ int jend, jdir;
+
+ stbiw__writef(s, "111 221 2222 11", 0,0,format+8, 0,0,0, 0,0,x,y, (colorbytes + has_alpha) * 8, has_alpha * 8);
+
+ if (stbi__flip_vertically_on_write) {
+ j = 0;
+ jend = y;
+ jdir = 1;
+ } else {
+ j = y-1;
+ jend = -1;
+ jdir = -1;
+ }
+ for (; j != jend; j += jdir) {
+ unsigned char *row = (unsigned char *) data + j * x * comp;
+ int len;
+
+ for (i = 0; i < x; i += len) {
+ unsigned char *begin = row + i * comp;
+ int diff = 1;
+ len = 1;
+
+ if (i < x - 1) {
+ ++len;
+ diff = memcmp(begin, row + (i + 1) * comp, comp);
+ if (diff) {
+ const unsigned char *prev = begin;
+ for (k = i + 2; k < x && len < 128; ++k) {
+ if (memcmp(prev, row + k * comp, comp)) {
+ prev += comp;
+ ++len;
+ } else {
+ --len;
+ break;
+ }
+ }
+ } else {
+ for (k = i + 2; k < x && len < 128; ++k) {
+ if (!memcmp(begin, row + k * comp, comp)) {
+ ++len;
+ } else {
+ break;
+ }
+ }
+ }
+ }
+
+ if (diff) {
+ unsigned char header = STBIW_UCHAR(len - 1);
+ stbiw__write1(s, header);
+ for (k = 0; k < len; ++k) {
+ stbiw__write_pixel(s, -1, comp, has_alpha, 0, begin + k * comp);
+ }
+ } else {
+ unsigned char header = STBIW_UCHAR(len - 129);
+ stbiw__write1(s, header);
+ stbiw__write_pixel(s, -1, comp, has_alpha, 0, begin);
+ }
+ }
+ }
+ stbiw__write_flush(s);
+ }
+ return 1;
+}
+
+STBIWDEF int stbi_write_tga_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data)
+{
+ stbi__write_context s = { 0 };
+ stbi__start_write_callbacks(&s, func, context);
+ return stbi_write_tga_core(&s, x, y, comp, (void *) data);
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_tga(char const *filename, int x, int y, int comp, const void *data)
+{
+ stbi__write_context s = { 0 };
+ if (stbi__start_write_file(&s,filename)) {
+ int r = stbi_write_tga_core(&s, x, y, comp, (void *) data);
+ stbi__end_write_file(&s);
+ return r;
+ } else
+ return 0;
+}
+#endif
+
+// *************************************************************************************************
+// Radiance RGBE HDR writer
+// by Baldur Karlsson
+
+#define stbiw__max(a, b) ((a) > (b) ? (a) : (b))
+
+#ifndef STBI_WRITE_NO_STDIO
+
+static void stbiw__linear_to_rgbe(unsigned char *rgbe, float *linear)
+{
+ int exponent;
+ float maxcomp = stbiw__max(linear[0], stbiw__max(linear[1], linear[2]));
+
+ if (maxcomp < 1e-32f) {
+ rgbe[0] = rgbe[1] = rgbe[2] = rgbe[3] = 0;
+ } else {
+ float normalize = (float) frexp(maxcomp, &exponent) * 256.0f/maxcomp;
+
+ rgbe[0] = (unsigned char)(linear[0] * normalize);
+ rgbe[1] = (unsigned char)(linear[1] * normalize);
+ rgbe[2] = (unsigned char)(linear[2] * normalize);
+ rgbe[3] = (unsigned char)(exponent + 128);
+ }
+}
+
+static void stbiw__write_run_data(stbi__write_context *s, int length, unsigned char databyte)
+{
+ unsigned char lengthbyte = STBIW_UCHAR(length+128);
+ STBIW_ASSERT(length+128 <= 255);
+ s->func(s->context, &lengthbyte, 1);
+ s->func(s->context, &databyte, 1);
+}
+
+static void stbiw__write_dump_data(stbi__write_context *s, int length, unsigned char *data)
+{
+ unsigned char lengthbyte = STBIW_UCHAR(length);
+ STBIW_ASSERT(length <= 128); // inconsistent with spec but consistent with official code
+ s->func(s->context, &lengthbyte, 1);
+ s->func(s->context, data, length);
+}
+
+static void stbiw__write_hdr_scanline(stbi__write_context *s, int width, int ncomp, unsigned char *scratch, float *scanline)
+{
+ unsigned char scanlineheader[4] = { 2, 2, 0, 0 };
+ unsigned char rgbe[4];
+ float linear[3];
+ int x;
+
+ scanlineheader[2] = (width&0xff00)>>8;
+ scanlineheader[3] = (width&0x00ff);
+
+ /* skip RLE for images too small or large */
+ if (width < 8 || width >= 32768) {
+ for (x=0; x < width; x++) {
+ switch (ncomp) {
+ case 4: /* fallthrough */
+ case 3: linear[2] = scanline[x*ncomp + 2];
+ linear[1] = scanline[x*ncomp + 1];
+ linear[0] = scanline[x*ncomp + 0];
+ break;
+ default:
+ linear[0] = linear[1] = linear[2] = scanline[x*ncomp + 0];
+ break;
+ }
+ stbiw__linear_to_rgbe(rgbe, linear);
+ s->func(s->context, rgbe, 4);
+ }
+ } else {
+ int c,r;
+ /* encode into scratch buffer */
+ for (x=0; x < width; x++) {
+ switch(ncomp) {
+ case 4: /* fallthrough */
+ case 3: linear[2] = scanline[x*ncomp + 2];
+ linear[1] = scanline[x*ncomp + 1];
+ linear[0] = scanline[x*ncomp + 0];
+ break;
+ default:
+ linear[0] = linear[1] = linear[2] = scanline[x*ncomp + 0];
+ break;
+ }
+ stbiw__linear_to_rgbe(rgbe, linear);
+ scratch[x + width*0] = rgbe[0];
+ scratch[x + width*1] = rgbe[1];
+ scratch[x + width*2] = rgbe[2];
+ scratch[x + width*3] = rgbe[3];
+ }
+
+ s->func(s->context, scanlineheader, 4);
+
+ /* RLE each component separately */
+ for (c=0; c < 4; c++) {
+ unsigned char *comp = &scratch[width*c];
+
+ x = 0;
+ while (x < width) {
+ // find first run
+ r = x;
+ while (r+2 < width) {
+ if (comp[r] == comp[r+1] && comp[r] == comp[r+2])
+ break;
+ ++r;
+ }
+ if (r+2 >= width)
+ r = width;
+ // dump up to first run
+ while (x < r) {
+ int len = r-x;
+ if (len > 128) len = 128;
+ stbiw__write_dump_data(s, len, &comp[x]);
+ x += len;
+ }
+ // if there's a run, output it
+ if (r+2 < width) { // same test as what we break out of in search loop, so only true if we break'd
+ // find next byte after run
+ while (r < width && comp[r] == comp[x])
+ ++r;
+ // output run up to r
+ while (x < r) {
+ int len = r-x;
+ if (len > 127) len = 127;
+ stbiw__write_run_data(s, len, comp[x]);
+ x += len;
+ }
+ }
+ }
+ }
+ }
+}
+
+static int stbi_write_hdr_core(stbi__write_context *s, int x, int y, int comp, float *data)
+{
+ if (y <= 0 || x <= 0 || data == NULL)
+ return 0;
+ else {
+ // Each component is stored separately. Allocate scratch space for full output scanline.
+ unsigned char *scratch = (unsigned char *) STBIW_MALLOC(x*4);
+ int i, len;
+ char buffer[128];
+ char header[] = "#?RADIANCE\n# Written by stb_image_write.h\nFORMAT=32-bit_rle_rgbe\n";
+ s->func(s->context, header, sizeof(header)-1);
+
+#ifdef __STDC_LIB_EXT1__
+ len = sprintf_s(buffer, sizeof(buffer), "EXPOSURE= 1.0000000000000\n\n-Y %d +X %d\n", y, x);
+#else
+ len = sprintf(buffer, "EXPOSURE= 1.0000000000000\n\n-Y %d +X %d\n", y, x);
+#endif
+ s->func(s->context, buffer, len);
+
+ for(i=0; i < y; i++)
+ stbiw__write_hdr_scanline(s, x, comp, scratch, data + comp*x*(stbi__flip_vertically_on_write ? y-1-i : i));
+ STBIW_FREE(scratch);
+ return 1;
+ }
+}
+
+STBIWDEF int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const float *data)
+{
+ stbi__write_context s = { 0 };
+ stbi__start_write_callbacks(&s, func, context);
+ return stbi_write_hdr_core(&s, x, y, comp, (float *) data);
+}
+
+STBIWDEF int stbi_write_hdr(char const *filename, int x, int y, int comp, const float *data)
+{
+ stbi__write_context s = { 0 };
+ if (stbi__start_write_file(&s,filename)) {
+ int r = stbi_write_hdr_core(&s, x, y, comp, (float *) data);
+ stbi__end_write_file(&s);
+ return r;
+ } else
+ return 0;
+}
+#endif // STBI_WRITE_NO_STDIO
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// PNG writer
+//
+
+#ifndef STBIW_ZLIB_COMPRESS
+// stretchy buffer; stbiw__sbpush() == vector<>::push_back() -- stbiw__sbcount() == vector<>::size()
+#define stbiw__sbraw(a) ((int *) (void *) (a) - 2)
+#define stbiw__sbm(a) stbiw__sbraw(a)[0]
+#define stbiw__sbn(a) stbiw__sbraw(a)[1]
+
+#define stbiw__sbneedgrow(a,n) ((a)==0 || stbiw__sbn(a)+n >= stbiw__sbm(a))
+#define stbiw__sbmaybegrow(a,n) (stbiw__sbneedgrow(a,(n)) ? stbiw__sbgrow(a,n) : 0)
+#define stbiw__sbgrow(a,n) stbiw__sbgrowf((void **) &(a), (n), sizeof(*(a)))
+
+#define stbiw__sbpush(a, v) (stbiw__sbmaybegrow(a,1), (a)[stbiw__sbn(a)++] = (v))
+#define stbiw__sbcount(a) ((a) ? stbiw__sbn(a) : 0)
+#define stbiw__sbfree(a) ((a) ? STBIW_FREE(stbiw__sbraw(a)),0 : 0)
+
+static void *stbiw__sbgrowf(void **arr, int increment, int itemsize)
+{
+ int m = *arr ? 2*stbiw__sbm(*arr)+increment : increment+1;
+ void *p = STBIW_REALLOC_SIZED(*arr ? stbiw__sbraw(*arr) : 0, *arr ? (stbiw__sbm(*arr)*itemsize + sizeof(int)*2) : 0, itemsize * m + sizeof(int)*2);
+ STBIW_ASSERT(p);
+ if (p) {
+ if (!*arr) ((int *) p)[1] = 0;
+ *arr = (void *) ((int *) p + 2);
+ stbiw__sbm(*arr) = m;
+ }
+ return *arr;
+}
+
+static unsigned char *stbiw__zlib_flushf(unsigned char *data, unsigned int *bitbuffer, int *bitcount)
+{
+ while (*bitcount >= 8) {
+ stbiw__sbpush(data, STBIW_UCHAR(*bitbuffer));
+ *bitbuffer >>= 8;
+ *bitcount -= 8;
+ }
+ return data;
+}
+
+static int stbiw__zlib_bitrev(int code, int codebits)
+{
+ int res=0;
+ while (codebits--) {
+ res = (res << 1) | (code & 1);
+ code >>= 1;
+ }
+ return res;
+}
+
+static unsigned int stbiw__zlib_countm(unsigned char *a, unsigned char *b, int limit)
+{
+ int i;
+ for (i=0; i < limit && i < 258; ++i)
+ if (a[i] != b[i]) break;
+ return i;
+}
+
+static unsigned int stbiw__zhash(unsigned char *data)
+{
+ stbiw_uint32 hash = data[0] + (data[1] << 8) + (data[2] << 16);
+ hash ^= hash << 3;
+ hash += hash >> 5;
+ hash ^= hash << 4;
+ hash += hash >> 17;
+ hash ^= hash << 25;
+ hash += hash >> 6;
+ return hash;
+}
+
+#define stbiw__zlib_flush() (out = stbiw__zlib_flushf(out, &bitbuf, &bitcount))
+#define stbiw__zlib_add(code,codebits) \
+ (bitbuf |= (code) << bitcount, bitcount += (codebits), stbiw__zlib_flush())
+#define stbiw__zlib_huffa(b,c) stbiw__zlib_add(stbiw__zlib_bitrev(b,c),c)
+// default huffman tables
+#define stbiw__zlib_huff1(n) stbiw__zlib_huffa(0x30 + (n), 8)
+#define stbiw__zlib_huff2(n) stbiw__zlib_huffa(0x190 + (n)-144, 9)
+#define stbiw__zlib_huff3(n) stbiw__zlib_huffa(0 + (n)-256,7)
+#define stbiw__zlib_huff4(n) stbiw__zlib_huffa(0xc0 + (n)-280,8)
+#define stbiw__zlib_huff(n) ((n) <= 143 ? stbiw__zlib_huff1(n) : (n) <= 255 ? stbiw__zlib_huff2(n) : (n) <= 279 ? stbiw__zlib_huff3(n) : stbiw__zlib_huff4(n))
+#define stbiw__zlib_huffb(n) ((n) <= 143 ? stbiw__zlib_huff1(n) : stbiw__zlib_huff2(n))
+
+#define stbiw__ZHASH 16384
+
+#endif // STBIW_ZLIB_COMPRESS
+
+STBIWDEF unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_len, int quality)
+{
+#ifdef STBIW_ZLIB_COMPRESS
+ // user provided a zlib compress implementation, use that
+ return STBIW_ZLIB_COMPRESS(data, data_len, out_len, quality);
+#else // use builtin
+ static unsigned short lengthc[] = { 3,4,5,6,7,8,9,10,11,13,15,17,19,23,27,31,35,43,51,59,67,83,99,115,131,163,195,227,258, 259 };
+ static unsigned char lengtheb[]= { 0,0,0,0,0,0,0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0 };
+ static unsigned short distc[] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577, 32768 };
+ static unsigned char disteb[] = { 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13 };
+ unsigned int bitbuf=0;
+ int i,j, bitcount=0;
+ unsigned char *out = NULL;
+ unsigned char ***hash_table = (unsigned char***) STBIW_MALLOC(stbiw__ZHASH * sizeof(unsigned char**));
+ if (hash_table == NULL)
+ return NULL;
+ if (quality < 5) quality = 5;
+
+ stbiw__sbpush(out, 0x78); // DEFLATE 32K window
+ stbiw__sbpush(out, 0x5e); // FLEVEL = 1
+ stbiw__zlib_add(1,1); // BFINAL = 1
+ stbiw__zlib_add(1,2); // BTYPE = 1 -- fixed huffman
+
+ for (i=0; i < stbiw__ZHASH; ++i)
+ hash_table[i] = NULL;
+
+ i=0;
+ while (i < data_len-3) {
+ // hash next 3 bytes of data to be compressed
+ int h = stbiw__zhash(data+i)&(stbiw__ZHASH-1), best=3;
+ unsigned char *bestloc = 0;
+ unsigned char **hlist = hash_table[h];
+ int n = stbiw__sbcount(hlist);
+ for (j=0; j < n; ++j) {
+ if (hlist[j]-data > i-32768) { // if entry lies within window
+ int d = stbiw__zlib_countm(hlist[j], data+i, data_len-i);
+ if (d >= best) { best=d; bestloc=hlist[j]; }
+ }
+ }
+ // when hash table entry is too long, delete half the entries
+ if (hash_table[h] && stbiw__sbn(hash_table[h]) == 2*quality) {
+ STBIW_MEMMOVE(hash_table[h], hash_table[h]+quality, sizeof(hash_table[h][0])*quality);
+ stbiw__sbn(hash_table[h]) = quality;
+ }
+ stbiw__sbpush(hash_table[h],data+i);
+
+ if (bestloc) {
+ // "lazy matching" - check match at *next* byte, and if it's better, do cur byte as literal
+ h = stbiw__zhash(data+i+1)&(stbiw__ZHASH-1);
+ hlist = hash_table[h];
+ n = stbiw__sbcount(hlist);
+ for (j=0; j < n; ++j) {
+ if (hlist[j]-data > i-32767) {
+ int e = stbiw__zlib_countm(hlist[j], data+i+1, data_len-i-1);
+ if (e > best) { // if next match is better, bail on current match
+ bestloc = NULL;
+ break;
+ }
+ }
+ }
+ }
+
+ if (bestloc) {
+ int d = (int) (data+i - bestloc); // distance back
+ STBIW_ASSERT(d <= 32767 && best <= 258);
+ for (j=0; best > lengthc[j+1]-1; ++j);
+ stbiw__zlib_huff(j+257);
+ if (lengtheb[j]) stbiw__zlib_add(best - lengthc[j], lengtheb[j]);
+ for (j=0; d > distc[j+1]-1; ++j);
+ stbiw__zlib_add(stbiw__zlib_bitrev(j,5),5);
+ if (disteb[j]) stbiw__zlib_add(d - distc[j], disteb[j]);
+ i += best;
+ } else {
+ stbiw__zlib_huffb(data[i]);
+ ++i;
+ }
+ }
+ // write out final bytes
+ for (;i < data_len; ++i)
+ stbiw__zlib_huffb(data[i]);
+ stbiw__zlib_huff(256); // end of block
+ // pad with 0 bits to byte boundary
+ while (bitcount)
+ stbiw__zlib_add(0,1);
+
+ for (i=0; i < stbiw__ZHASH; ++i)
+ (void) stbiw__sbfree(hash_table[i]);
+ STBIW_FREE(hash_table);
+
+ // store uncompressed instead if compression was worse
+ if (stbiw__sbn(out) > data_len + 2 + ((data_len+32766)/32767)*5) {
+ stbiw__sbn(out) = 2; // truncate to DEFLATE 32K window and FLEVEL = 1
+ for (j = 0; j < data_len;) {
+ int blocklen = data_len - j;
+ if (blocklen > 32767) blocklen = 32767;
+ stbiw__sbpush(out, data_len - j == blocklen); // BFINAL = ?, BTYPE = 0 -- no compression
+ stbiw__sbpush(out, STBIW_UCHAR(blocklen)); // LEN
+ stbiw__sbpush(out, STBIW_UCHAR(blocklen >> 8));
+ stbiw__sbpush(out, STBIW_UCHAR(~blocklen)); // NLEN
+ stbiw__sbpush(out, STBIW_UCHAR(~blocklen >> 8));
+ memcpy(out+stbiw__sbn(out), data+j, blocklen);
+ stbiw__sbn(out) += blocklen;
+ j += blocklen;
+ }
+ }
+
+ {
+ // compute adler32 on input
+ unsigned int s1=1, s2=0;
+ int blocklen = (int) (data_len % 5552);
+ j=0;
+ while (j < data_len) {
+ for (i=0; i < blocklen; ++i) { s1 += data[j+i]; s2 += s1; }
+ s1 %= 65521; s2 %= 65521;
+ j += blocklen;
+ blocklen = 5552;
+ }
+ stbiw__sbpush(out, STBIW_UCHAR(s2 >> 8));
+ stbiw__sbpush(out, STBIW_UCHAR(s2));
+ stbiw__sbpush(out, STBIW_UCHAR(s1 >> 8));
+ stbiw__sbpush(out, STBIW_UCHAR(s1));
+ }
+ *out_len = stbiw__sbn(out);
+ // make returned pointer freeable
+ STBIW_MEMMOVE(stbiw__sbraw(out), out, *out_len);
+ return (unsigned char *) stbiw__sbraw(out);
+#endif // STBIW_ZLIB_COMPRESS
+}
+
+static unsigned int stbiw__crc32(unsigned char *buffer, int len)
+{
+#ifdef STBIW_CRC32
+ return STBIW_CRC32(buffer, len);
+#else
+ static unsigned int crc_table[256] =
+ {
+ 0x00000000, 0x77073096, 0xEE0E612C, 0x990951BA, 0x076DC419, 0x706AF48F, 0xE963A535, 0x9E6495A3,
+ 0x0eDB8832, 0x79DCB8A4, 0xE0D5E91E, 0x97D2D988, 0x09B64C2B, 0x7EB17CBD, 0xE7B82D07, 0x90BF1D91,
+ 0x1DB71064, 0x6AB020F2, 0xF3B97148, 0x84BE41DE, 0x1ADAD47D, 0x6DDDE4EB, 0xF4D4B551, 0x83D385C7,
+ 0x136C9856, 0x646BA8C0, 0xFD62F97A, 0x8A65C9EC, 0x14015C4F, 0x63066CD9, 0xFA0F3D63, 0x8D080DF5,
+ 0x3B6E20C8, 0x4C69105E, 0xD56041E4, 0xA2677172, 0x3C03E4D1, 0x4B04D447, 0xD20D85FD, 0xA50AB56B,
+ 0x35B5A8FA, 0x42B2986C, 0xDBBBC9D6, 0xACBCF940, 0x32D86CE3, 0x45DF5C75, 0xDCD60DCF, 0xABD13D59,
+ 0x26D930AC, 0x51DE003A, 0xC8D75180, 0xBFD06116, 0x21B4F4B5, 0x56B3C423, 0xCFBA9599, 0xB8BDA50F,
+ 0x2802B89E, 0x5F058808, 0xC60CD9B2, 0xB10BE924, 0x2F6F7C87, 0x58684C11, 0xC1611DAB, 0xB6662D3D,
+ 0x76DC4190, 0x01DB7106, 0x98D220BC, 0xEFD5102A, 0x71B18589, 0x06B6B51F, 0x9FBFE4A5, 0xE8B8D433,
+ 0x7807C9A2, 0x0F00F934, 0x9609A88E, 0xE10E9818, 0x7F6A0DBB, 0x086D3D2D, 0x91646C97, 0xE6635C01,
+ 0x6B6B51F4, 0x1C6C6162, 0x856530D8, 0xF262004E, 0x6C0695ED, 0x1B01A57B, 0x8208F4C1, 0xF50FC457,
+ 0x65B0D9C6, 0x12B7E950, 0x8BBEB8EA, 0xFCB9887C, 0x62DD1DDF, 0x15DA2D49, 0x8CD37CF3, 0xFBD44C65,
+ 0x4DB26158, 0x3AB551CE, 0xA3BC0074, 0xD4BB30E2, 0x4ADFA541, 0x3DD895D7, 0xA4D1C46D, 0xD3D6F4FB,
+ 0x4369E96A, 0x346ED9FC, 0xAD678846, 0xDA60B8D0, 0x44042D73, 0x33031DE5, 0xAA0A4C5F, 0xDD0D7CC9,
+ 0x5005713C, 0x270241AA, 0xBE0B1010, 0xC90C2086, 0x5768B525, 0x206F85B3, 0xB966D409, 0xCE61E49F,
+ 0x5EDEF90E, 0x29D9C998, 0xB0D09822, 0xC7D7A8B4, 0x59B33D17, 0x2EB40D81, 0xB7BD5C3B, 0xC0BA6CAD,
+ 0xEDB88320, 0x9ABFB3B6, 0x03B6E20C, 0x74B1D29A, 0xEAD54739, 0x9DD277AF, 0x04DB2615, 0x73DC1683,
+ 0xE3630B12, 0x94643B84, 0x0D6D6A3E, 0x7A6A5AA8, 0xE40ECF0B, 0x9309FF9D, 0x0A00AE27, 0x7D079EB1,
+ 0xF00F9344, 0x8708A3D2, 0x1E01F268, 0x6906C2FE, 0xF762575D, 0x806567CB, 0x196C3671, 0x6E6B06E7,
+ 0xFED41B76, 0x89D32BE0, 0x10DA7A5A, 0x67DD4ACC, 0xF9B9DF6F, 0x8EBEEFF9, 0x17B7BE43, 0x60B08ED5,
+ 0xD6D6A3E8, 0xA1D1937E, 0x38D8C2C4, 0x4FDFF252, 0xD1BB67F1, 0xA6BC5767, 0x3FB506DD, 0x48B2364B,
+ 0xD80D2BDA, 0xAF0A1B4C, 0x36034AF6, 0x41047A60, 0xDF60EFC3, 0xA867DF55, 0x316E8EEF, 0x4669BE79,
+ 0xCB61B38C, 0xBC66831A, 0x256FD2A0, 0x5268E236, 0xCC0C7795, 0xBB0B4703, 0x220216B9, 0x5505262F,
+ 0xC5BA3BBE, 0xB2BD0B28, 0x2BB45A92, 0x5CB36A04, 0xC2D7FFA7, 0xB5D0CF31, 0x2CD99E8B, 0x5BDEAE1D,
+ 0x9B64C2B0, 0xEC63F226, 0x756AA39C, 0x026D930A, 0x9C0906A9, 0xEB0E363F, 0x72076785, 0x05005713,
+ 0x95BF4A82, 0xE2B87A14, 0x7BB12BAE, 0x0CB61B38, 0x92D28E9B, 0xE5D5BE0D, 0x7CDCEFB7, 0x0BDBDF21,
+ 0x86D3D2D4, 0xF1D4E242, 0x68DDB3F8, 0x1FDA836E, 0x81BE16CD, 0xF6B9265B, 0x6FB077E1, 0x18B74777,
+ 0x88085AE6, 0xFF0F6A70, 0x66063BCA, 0x11010B5C, 0x8F659EFF, 0xF862AE69, 0x616BFFD3, 0x166CCF45,
+ 0xA00AE278, 0xD70DD2EE, 0x4E048354, 0x3903B3C2, 0xA7672661, 0xD06016F7, 0x4969474D, 0x3E6E77DB,
+ 0xAED16A4A, 0xD9D65ADC, 0x40DF0B66, 0x37D83BF0, 0xA9BCAE53, 0xDEBB9EC5, 0x47B2CF7F, 0x30B5FFE9,
+ 0xBDBDF21C, 0xCABAC28A, 0x53B39330, 0x24B4A3A6, 0xBAD03605, 0xCDD70693, 0x54DE5729, 0x23D967BF,
+ 0xB3667A2E, 0xC4614AB8, 0x5D681B02, 0x2A6F2B94, 0xB40BBE37, 0xC30C8EA1, 0x5A05DF1B, 0x2D02EF8D
+ };
+
+ unsigned int crc = ~0u;
+ int i;
+ for (i=0; i < len; ++i)
+ crc = (crc >> 8) ^ crc_table[buffer[i] ^ (crc & 0xff)];
+ return ~crc;
+#endif
+}
+
+#define stbiw__wpng4(o,a,b,c,d) ((o)[0]=STBIW_UCHAR(a),(o)[1]=STBIW_UCHAR(b),(o)[2]=STBIW_UCHAR(c),(o)[3]=STBIW_UCHAR(d),(o)+=4)
+#define stbiw__wp32(data,v) stbiw__wpng4(data, (v)>>24,(v)>>16,(v)>>8,(v));
+#define stbiw__wptag(data,s) stbiw__wpng4(data, s[0],s[1],s[2],s[3])
+
+static void stbiw__wpcrc(unsigned char **data, int len)
+{
+ unsigned int crc = stbiw__crc32(*data - len - 4, len+4);
+ stbiw__wp32(*data, crc);
+}
+
+static unsigned char stbiw__paeth(int a, int b, int c)
+{
+ int p = a + b - c, pa = abs(p-a), pb = abs(p-b), pc = abs(p-c);
+ if (pa <= pb && pa <= pc) return STBIW_UCHAR(a);
+ if (pb <= pc) return STBIW_UCHAR(b);
+ return STBIW_UCHAR(c);
+}
+
+// @OPTIMIZE: provide an option that always forces left-predict or paeth predict
+static void stbiw__encode_png_line(unsigned char *pixels, int stride_bytes, int width, int height, int y, int n, int filter_type, signed char *line_buffer)
+{
+ static int mapping[] = { 0,1,2,3,4 };
+ static int firstmap[] = { 0,1,0,5,6 };
+ int *mymap = (y != 0) ? mapping : firstmap;
+ int i;
+ int type = mymap[filter_type];
+ unsigned char *z = pixels + stride_bytes * (stbi__flip_vertically_on_write ? height-1-y : y);
+ int signed_stride = stbi__flip_vertically_on_write ? -stride_bytes : stride_bytes;
+
+ if (type==0) {
+ memcpy(line_buffer, z, width*n);
+ return;
+ }
+
+ // first loop isn't optimized since it's just one pixel
+ for (i = 0; i < n; ++i) {
+ switch (type) {
+ case 1: line_buffer[i] = z[i]; break;
+ case 2: line_buffer[i] = z[i] - z[i-signed_stride]; break;
+ case 3: line_buffer[i] = z[i] - (z[i-signed_stride]>>1); break;
+ case 4: line_buffer[i] = (signed char) (z[i] - stbiw__paeth(0,z[i-signed_stride],0)); break;
+ case 5: line_buffer[i] = z[i]; break;
+ case 6: line_buffer[i] = z[i]; break;
+ }
+ }
+ switch (type) {
+ case 1: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - z[i-n]; break;
+ case 2: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - z[i-signed_stride]; break;
+ case 3: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - ((z[i-n] + z[i-signed_stride])>>1); break;
+ case 4: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - stbiw__paeth(z[i-n], z[i-signed_stride], z[i-signed_stride-n]); break;
+ case 5: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - (z[i-n]>>1); break;
+ case 6: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - stbiw__paeth(z[i-n], 0,0); break;
+ }
+}
+
+STBIWDEF unsigned char *stbi_write_png_to_mem(const unsigned char *pixels, int stride_bytes, int x, int y, int n, int *out_len)
+{
+ int force_filter = stbi_write_force_png_filter;
+ int ctype[5] = { -1, 0, 4, 2, 6 };
+ unsigned char sig[8] = { 137,80,78,71,13,10,26,10 };
+ unsigned char *out,*o, *filt, *zlib;
+ signed char *line_buffer;
+ int j,zlen;
+
+ if (stride_bytes == 0)
+ stride_bytes = x * n;
+
+ if (force_filter >= 5) {
+ force_filter = -1;
+ }
+
+ filt = (unsigned char *) STBIW_MALLOC((x*n+1) * y); if (!filt) return 0;
+ line_buffer = (signed char *) STBIW_MALLOC(x * n); if (!line_buffer) { STBIW_FREE(filt); return 0; }
+ for (j=0; j < y; ++j) {
+ int filter_type;
+ if (force_filter > -1) {
+ filter_type = force_filter;
+ stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, force_filter, line_buffer);
+ } else { // Estimate the best filter by running through all of them:
+ int best_filter = 0, best_filter_val = 0x7fffffff, est, i;
+ for (filter_type = 0; filter_type < 5; filter_type++) {
+ stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, filter_type, line_buffer);
+
+ // Estimate the entropy of the line using this filter; the less, the better.
+ est = 0;
+ for (i = 0; i < x*n; ++i) {
+ est += abs((signed char) line_buffer[i]);
+ }
+ if (est < best_filter_val) {
+ best_filter_val = est;
+ best_filter = filter_type;
+ }
+ }
+ if (filter_type != best_filter) { // If the last iteration already got us the best filter, don't redo it
+ stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, best_filter, line_buffer);
+ filter_type = best_filter;
+ }
+ }
+ // when we get here, filter_type contains the filter type, and line_buffer contains the data
+ filt[j*(x*n+1)] = (unsigned char) filter_type;
+ STBIW_MEMMOVE(filt+j*(x*n+1)+1, line_buffer, x*n);
+ }
+ STBIW_FREE(line_buffer);
+ zlib = stbi_zlib_compress(filt, y*( x*n+1), &zlen, stbi_write_png_compression_level);
+ STBIW_FREE(filt);
+ if (!zlib) return 0;
+
+ // each tag requires 12 bytes of overhead
+ out = (unsigned char *) STBIW_MALLOC(8 + 12+13 + 12+zlen + 12);
+ if (!out) return 0;
+ *out_len = 8 + 12+13 + 12+zlen + 12;
+
+ o=out;
+ STBIW_MEMMOVE(o,sig,8); o+= 8;
+ stbiw__wp32(o, 13); // header length
+ stbiw__wptag(o, "IHDR");
+ stbiw__wp32(o, x);
+ stbiw__wp32(o, y);
+ *o++ = 8;
+ *o++ = STBIW_UCHAR(ctype[n]);
+ *o++ = 0;
+ *o++ = 0;
+ *o++ = 0;
+ stbiw__wpcrc(&o,13);
+
+ stbiw__wp32(o, zlen);
+ stbiw__wptag(o, "IDAT");
+ STBIW_MEMMOVE(o, zlib, zlen);
+ o += zlen;
+ STBIW_FREE(zlib);
+ stbiw__wpcrc(&o, zlen);
+
+ stbiw__wp32(o,0);
+ stbiw__wptag(o, "IEND");
+ stbiw__wpcrc(&o,0);
+
+ STBIW_ASSERT(o == out + *out_len);
+
+ return out;
+}
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_png(char const *filename, int x, int y, int comp, const void *data, int stride_bytes)
+{
+ FILE *f;
+ int len;
+ unsigned char *png = stbi_write_png_to_mem((const unsigned char *) data, stride_bytes, x, y, comp, &len);
+ if (png == NULL) return 0;
+
+ f = stbiw__fopen(filename, "wb");
+ if (!f) { STBIW_FREE(png); return 0; }
+ fwrite(png, 1, len, f);
+ fclose(f);
+ STBIW_FREE(png);
+ return 1;
+}
+#endif
+
+STBIWDEF int stbi_write_png_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int stride_bytes)
+{
+ int len;
+ unsigned char *png = stbi_write_png_to_mem((const unsigned char *) data, stride_bytes, x, y, comp, &len);
+ if (png == NULL) return 0;
+ func(context, png, len);
+ STBIW_FREE(png);
+ return 1;
+}
+
+
+/* ***************************************************************************
+ *
+ * JPEG writer
+ *
+ * This is based on Jon Olick's jo_jpeg.cpp:
+ * public domain Simple, Minimalistic JPEG writer - http://www.jonolick.com/code.html
+ */
+
+static const unsigned char stbiw__jpg_ZigZag[] = { 0,1,5,6,14,15,27,28,2,4,7,13,16,26,29,42,3,8,12,17,25,30,41,43,9,11,18,
+ 24,31,40,44,53,10,19,23,32,39,45,52,54,20,22,33,38,46,51,55,60,21,34,37,47,50,56,59,61,35,36,48,49,57,58,62,63 };
+
+static void stbiw__jpg_writeBits(stbi__write_context *s, int *bitBufP, int *bitCntP, const unsigned short *bs) {
+ int bitBuf = *bitBufP, bitCnt = *bitCntP;
+ bitCnt += bs[1];
+ bitBuf |= bs[0] << (24 - bitCnt);
+ while(bitCnt >= 8) {
+ unsigned char c = (bitBuf >> 16) & 255;
+ stbiw__putc(s, c);
+ if(c == 255) {
+ stbiw__putc(s, 0);
+ }
+ bitBuf <<= 8;
+ bitCnt -= 8;
+ }
+ *bitBufP = bitBuf;
+ *bitCntP = bitCnt;
+}
+
+static void stbiw__jpg_DCT(float *d0p, float *d1p, float *d2p, float *d3p, float *d4p, float *d5p, float *d6p, float *d7p) {
+ float d0 = *d0p, d1 = *d1p, d2 = *d2p, d3 = *d3p, d4 = *d4p, d5 = *d5p, d6 = *d6p, d7 = *d7p;
+ float z1, z2, z3, z4, z5, z11, z13;
+
+ float tmp0 = d0 + d7;
+ float tmp7 = d0 - d7;
+ float tmp1 = d1 + d6;
+ float tmp6 = d1 - d6;
+ float tmp2 = d2 + d5;
+ float tmp5 = d2 - d5;
+ float tmp3 = d3 + d4;
+ float tmp4 = d3 - d4;
+
+ // Even part
+ float tmp10 = tmp0 + tmp3; // phase 2
+ float tmp13 = tmp0 - tmp3;
+ float tmp11 = tmp1 + tmp2;
+ float tmp12 = tmp1 - tmp2;
+
+ d0 = tmp10 + tmp11; // phase 3
+ d4 = tmp10 - tmp11;
+
+ z1 = (tmp12 + tmp13) * 0.707106781f; // c4
+ d2 = tmp13 + z1; // phase 5
+ d6 = tmp13 - z1;
+
+ // Odd part
+ tmp10 = tmp4 + tmp5; // phase 2
+ tmp11 = tmp5 + tmp6;
+ tmp12 = tmp6 + tmp7;
+
+ // The rotator is modified from fig 4-8 to avoid extra negations.
+ z5 = (tmp10 - tmp12) * 0.382683433f; // c6
+ z2 = tmp10 * 0.541196100f + z5; // c2-c6
+ z4 = tmp12 * 1.306562965f + z5; // c2+c6
+ z3 = tmp11 * 0.707106781f; // c4
+
+ z11 = tmp7 + z3; // phase 5
+ z13 = tmp7 - z3;
+
+ *d5p = z13 + z2; // phase 6
+ *d3p = z13 - z2;
+ *d1p = z11 + z4;
+ *d7p = z11 - z4;
+
+ *d0p = d0; *d2p = d2; *d4p = d4; *d6p = d6;
+}
+
+static void stbiw__jpg_calcBits(int val, unsigned short bits[2]) {
+ int tmp1 = val < 0 ? -val : val;
+ val = val < 0 ? val-1 : val;
+ bits[1] = 1;
+ while(tmp1 >>= 1) {
+ ++bits[1];
+ }
+ bits[0] = val & ((1<<bits[1])-1);
+}
+
+static int stbiw__jpg_processDU(stbi__write_context *s, int *bitBuf, int *bitCnt, float *CDU, int du_stride, float *fdtbl, int DC, const unsigned short HTDC[256][2], const unsigned short HTAC[256][2]) {
+ const unsigned short EOB[2] = { HTAC[0x00][0], HTAC[0x00][1] };
+ const unsigned short M16zeroes[2] = { HTAC[0xF0][0], HTAC[0xF0][1] };
+ int dataOff, i, j, n, diff, end0pos, x, y;
+ int DU[64];
+
+ // DCT rows
+ for(dataOff=0, n=du_stride*8; dataOff<n; dataOff+=du_stride) {
+ stbiw__jpg_DCT(&CDU[dataOff], &CDU[dataOff+1], &CDU[dataOff+2], &CDU[dataOff+3], &CDU[dataOff+4], &CDU[dataOff+5], &CDU[dataOff+6], &CDU[dataOff+7]);
+ }
+ // DCT columns
+ for(dataOff=0; dataOff<8; ++dataOff) {
+ stbiw__jpg_DCT(&CDU[dataOff], &CDU[dataOff+du_stride], &CDU[dataOff+du_stride*2], &CDU[dataOff+du_stride*3], &CDU[dataOff+du_stride*4],
+ &CDU[dataOff+du_stride*5], &CDU[dataOff+du_stride*6], &CDU[dataOff+du_stride*7]);
+ }
+ // Quantize/descale/zigzag the coefficients
+ for(y = 0, j=0; y < 8; ++y) {
+ for(x = 0; x < 8; ++x,++j) {
+ float v;
+ i = y*du_stride+x;
+ v = CDU[i]*fdtbl[j];
+ // DU[stbiw__jpg_ZigZag[j]] = (int)(v < 0 ? ceilf(v - 0.5f) : floorf(v + 0.5f));
+ // ceilf() and floorf() are C99, not C89, but I /think/ they're not needed here anyway?
+ DU[stbiw__jpg_ZigZag[j]] = (int)(v < 0 ? v - 0.5f : v + 0.5f);
+ }
+ }
+
+ // Encode DC
+ diff = DU[0] - DC;
+ if (diff == 0) {
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, HTDC[0]);
+ } else {
+ unsigned short bits[2];
+ stbiw__jpg_calcBits(diff, bits);
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, HTDC[bits[1]]);
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, bits);
+ }
+ // Encode ACs
+ end0pos = 63;
+ for(; (end0pos>0)&&(DU[end0pos]==0); --end0pos) {
+ }
+ // end0pos = first element in reverse order !=0
+ if(end0pos == 0) {
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, EOB);
+ return DU[0];
+ }
+ for(i = 1; i <= end0pos; ++i) {
+ int startpos = i;
+ int nrzeroes;
+ unsigned short bits[2];
+ for (; DU[i]==0 && i<=end0pos; ++i) {
+ }
+ nrzeroes = i-startpos;
+ if ( nrzeroes >= 16 ) {
+ int lng = nrzeroes>>4;
+ int nrmarker;
+ for (nrmarker=1; nrmarker <= lng; ++nrmarker)
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, M16zeroes);
+ nrzeroes &= 15;
+ }
+ stbiw__jpg_calcBits(DU[i], bits);
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, HTAC[(nrzeroes<<4)+bits[1]]);
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, bits);
+ }
+ if(end0pos != 63) {
+ stbiw__jpg_writeBits(s, bitBuf, bitCnt, EOB);
+ }
+ return DU[0];
+}
+
+static int stbi_write_jpg_core(stbi__write_context *s, int width, int height, int comp, const void* data, int quality) {
+ // Constants that don't pollute global namespace
+ static const unsigned char std_dc_luminance_nrcodes[] = {0,0,1,5,1,1,1,1,1,1,0,0,0,0,0,0,0};
+ static const unsigned char std_dc_luminance_values[] = {0,1,2,3,4,5,6,7,8,9,10,11};
+ static const unsigned char std_ac_luminance_nrcodes[] = {0,0,2,1,3,3,2,4,3,5,5,4,4,0,0,1,0x7d};
+ static const unsigned char std_ac_luminance_values[] = {
+ 0x01,0x02,0x03,0x00,0x04,0x11,0x05,0x12,0x21,0x31,0x41,0x06,0x13,0x51,0x61,0x07,0x22,0x71,0x14,0x32,0x81,0x91,0xa1,0x08,
+ 0x23,0x42,0xb1,0xc1,0x15,0x52,0xd1,0xf0,0x24,0x33,0x62,0x72,0x82,0x09,0x0a,0x16,0x17,0x18,0x19,0x1a,0x25,0x26,0x27,0x28,
+ 0x29,0x2a,0x34,0x35,0x36,0x37,0x38,0x39,0x3a,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x53,0x54,0x55,0x56,0x57,0x58,0x59,
+ 0x5a,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x83,0x84,0x85,0x86,0x87,0x88,0x89,
+ 0x8a,0x92,0x93,0x94,0x95,0x96,0x97,0x98,0x99,0x9a,0xa2,0xa3,0xa4,0xa5,0xa6,0xa7,0xa8,0xa9,0xaa,0xb2,0xb3,0xb4,0xb5,0xb6,
+ 0xb7,0xb8,0xb9,0xba,0xc2,0xc3,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xd2,0xd3,0xd4,0xd5,0xd6,0xd7,0xd8,0xd9,0xda,0xe1,0xe2,
+ 0xe3,0xe4,0xe5,0xe6,0xe7,0xe8,0xe9,0xea,0xf1,0xf2,0xf3,0xf4,0xf5,0xf6,0xf7,0xf8,0xf9,0xfa
+ };
+ static const unsigned char std_dc_chrominance_nrcodes[] = {0,0,3,1,1,1,1,1,1,1,1,1,0,0,0,0,0};
+ static const unsigned char std_dc_chrominance_values[] = {0,1,2,3,4,5,6,7,8,9,10,11};
+ static const unsigned char std_ac_chrominance_nrcodes[] = {0,0,2,1,2,4,4,3,4,7,5,4,4,0,1,2,0x77};
+ static const unsigned char std_ac_chrominance_values[] = {
+ 0x00,0x01,0x02,0x03,0x11,0x04,0x05,0x21,0x31,0x06,0x12,0x41,0x51,0x07,0x61,0x71,0x13,0x22,0x32,0x81,0x08,0x14,0x42,0x91,
+ 0xa1,0xb1,0xc1,0x09,0x23,0x33,0x52,0xf0,0x15,0x62,0x72,0xd1,0x0a,0x16,0x24,0x34,0xe1,0x25,0xf1,0x17,0x18,0x19,0x1a,0x26,
+ 0x27,0x28,0x29,0x2a,0x35,0x36,0x37,0x38,0x39,0x3a,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x53,0x54,0x55,0x56,0x57,0x58,
+ 0x59,0x5a,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x82,0x83,0x84,0x85,0x86,0x87,
+ 0x88,0x89,0x8a,0x92,0x93,0x94,0x95,0x96,0x97,0x98,0x99,0x9a,0xa2,0xa3,0xa4,0xa5,0xa6,0xa7,0xa8,0xa9,0xaa,0xb2,0xb3,0xb4,
+ 0xb5,0xb6,0xb7,0xb8,0xb9,0xba,0xc2,0xc3,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xd2,0xd3,0xd4,0xd5,0xd6,0xd7,0xd8,0xd9,0xda,
+ 0xe2,0xe3,0xe4,0xe5,0xe6,0xe7,0xe8,0xe9,0xea,0xf2,0xf3,0xf4,0xf5,0xf6,0xf7,0xf8,0xf9,0xfa
+ };
+ // Huffman tables
+ static const unsigned short YDC_HT[256][2] = { {0,2},{2,3},{3,3},{4,3},{5,3},{6,3},{14,4},{30,5},{62,6},{126,7},{254,8},{510,9}};
+ static const unsigned short UVDC_HT[256][2] = { {0,2},{1,2},{2,2},{6,3},{14,4},{30,5},{62,6},{126,7},{254,8},{510,9},{1022,10},{2046,11}};
+ static const unsigned short YAC_HT[256][2] = {
+ {10,4},{0,2},{1,2},{4,3},{11,4},{26,5},{120,7},{248,8},{1014,10},{65410,16},{65411,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {12,4},{27,5},{121,7},{502,9},{2038,11},{65412,16},{65413,16},{65414,16},{65415,16},{65416,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {28,5},{249,8},{1015,10},{4084,12},{65417,16},{65418,16},{65419,16},{65420,16},{65421,16},{65422,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {58,6},{503,9},{4085,12},{65423,16},{65424,16},{65425,16},{65426,16},{65427,16},{65428,16},{65429,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {59,6},{1016,10},{65430,16},{65431,16},{65432,16},{65433,16},{65434,16},{65435,16},{65436,16},{65437,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {122,7},{2039,11},{65438,16},{65439,16},{65440,16},{65441,16},{65442,16},{65443,16},{65444,16},{65445,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {123,7},{4086,12},{65446,16},{65447,16},{65448,16},{65449,16},{65450,16},{65451,16},{65452,16},{65453,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {250,8},{4087,12},{65454,16},{65455,16},{65456,16},{65457,16},{65458,16},{65459,16},{65460,16},{65461,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {504,9},{32704,15},{65462,16},{65463,16},{65464,16},{65465,16},{65466,16},{65467,16},{65468,16},{65469,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {505,9},{65470,16},{65471,16},{65472,16},{65473,16},{65474,16},{65475,16},{65476,16},{65477,16},{65478,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {506,9},{65479,16},{65480,16},{65481,16},{65482,16},{65483,16},{65484,16},{65485,16},{65486,16},{65487,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {1017,10},{65488,16},{65489,16},{65490,16},{65491,16},{65492,16},{65493,16},{65494,16},{65495,16},{65496,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {1018,10},{65497,16},{65498,16},{65499,16},{65500,16},{65501,16},{65502,16},{65503,16},{65504,16},{65505,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {2040,11},{65506,16},{65507,16},{65508,16},{65509,16},{65510,16},{65511,16},{65512,16},{65513,16},{65514,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {65515,16},{65516,16},{65517,16},{65518,16},{65519,16},{65520,16},{65521,16},{65522,16},{65523,16},{65524,16},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {2041,11},{65525,16},{65526,16},{65527,16},{65528,16},{65529,16},{65530,16},{65531,16},{65532,16},{65533,16},{65534,16},{0,0},{0,0},{0,0},{0,0},{0,0}
+ };
+ static const unsigned short UVAC_HT[256][2] = {
+ {0,2},{1,2},{4,3},{10,4},{24,5},{25,5},{56,6},{120,7},{500,9},{1014,10},{4084,12},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {11,4},{57,6},{246,8},{501,9},{2038,11},{4085,12},{65416,16},{65417,16},{65418,16},{65419,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {26,5},{247,8},{1015,10},{4086,12},{32706,15},{65420,16},{65421,16},{65422,16},{65423,16},{65424,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {27,5},{248,8},{1016,10},{4087,12},{65425,16},{65426,16},{65427,16},{65428,16},{65429,16},{65430,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {58,6},{502,9},{65431,16},{65432,16},{65433,16},{65434,16},{65435,16},{65436,16},{65437,16},{65438,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {59,6},{1017,10},{65439,16},{65440,16},{65441,16},{65442,16},{65443,16},{65444,16},{65445,16},{65446,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {121,7},{2039,11},{65447,16},{65448,16},{65449,16},{65450,16},{65451,16},{65452,16},{65453,16},{65454,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {122,7},{2040,11},{65455,16},{65456,16},{65457,16},{65458,16},{65459,16},{65460,16},{65461,16},{65462,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {249,8},{65463,16},{65464,16},{65465,16},{65466,16},{65467,16},{65468,16},{65469,16},{65470,16},{65471,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {503,9},{65472,16},{65473,16},{65474,16},{65475,16},{65476,16},{65477,16},{65478,16},{65479,16},{65480,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {504,9},{65481,16},{65482,16},{65483,16},{65484,16},{65485,16},{65486,16},{65487,16},{65488,16},{65489,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {505,9},{65490,16},{65491,16},{65492,16},{65493,16},{65494,16},{65495,16},{65496,16},{65497,16},{65498,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {506,9},{65499,16},{65500,16},{65501,16},{65502,16},{65503,16},{65504,16},{65505,16},{65506,16},{65507,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {2041,11},{65508,16},{65509,16},{65510,16},{65511,16},{65512,16},{65513,16},{65514,16},{65515,16},{65516,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {16352,14},{65517,16},{65518,16},{65519,16},{65520,16},{65521,16},{65522,16},{65523,16},{65524,16},{65525,16},{0,0},{0,0},{0,0},{0,0},{0,0},
+ {1018,10},{32707,15},{65526,16},{65527,16},{65528,16},{65529,16},{65530,16},{65531,16},{65532,16},{65533,16},{65534,16},{0,0},{0,0},{0,0},{0,0},{0,0}
+ };
+ static const int YQT[] = {16,11,10,16,24,40,51,61,12,12,14,19,26,58,60,55,14,13,16,24,40,57,69,56,14,17,22,29,51,87,80,62,18,22,
+ 37,56,68,109,103,77,24,35,55,64,81,104,113,92,49,64,78,87,103,121,120,101,72,92,95,98,112,100,103,99};
+ static const int UVQT[] = {17,18,24,47,99,99,99,99,18,21,26,66,99,99,99,99,24,26,56,99,99,99,99,99,47,66,99,99,99,99,99,99,
+ 99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99};
+ static const float aasf[] = { 1.0f * 2.828427125f, 1.387039845f * 2.828427125f, 1.306562965f * 2.828427125f, 1.175875602f * 2.828427125f,
+ 1.0f * 2.828427125f, 0.785694958f * 2.828427125f, 0.541196100f * 2.828427125f, 0.275899379f * 2.828427125f };
+
+ int row, col, i, k, subsample;
+ float fdtbl_Y[64], fdtbl_UV[64];
+ unsigned char YTable[64], UVTable[64];
+
+ if(!data || !width || !height || comp > 4 || comp < 1) {
+ return 0;
+ }
+
+ quality = quality ? quality : 90;
+ subsample = quality <= 90 ? 1 : 0;
+ quality = quality < 1 ? 1 : quality > 100 ? 100 : quality;
+ quality = quality < 50 ? 5000 / quality : 200 - quality * 2;
+
+ for(i = 0; i < 64; ++i) {
+ int uvti, yti = (YQT[i]*quality+50)/100;
+ YTable[stbiw__jpg_ZigZag[i]] = (unsigned char) (yti < 1 ? 1 : yti > 255 ? 255 : yti);
+ uvti = (UVQT[i]*quality+50)/100;
+ UVTable[stbiw__jpg_ZigZag[i]] = (unsigned char) (uvti < 1 ? 1 : uvti > 255 ? 255 : uvti);
+ }
+
+ for(row = 0, k = 0; row < 8; ++row) {
+ for(col = 0; col < 8; ++col, ++k) {
+ fdtbl_Y[k] = 1 / (YTable [stbiw__jpg_ZigZag[k]] * aasf[row] * aasf[col]);
+ fdtbl_UV[k] = 1 / (UVTable[stbiw__jpg_ZigZag[k]] * aasf[row] * aasf[col]);
+ }
+ }
+
+ // Write Headers
+ {
+ static const unsigned char head0[] = { 0xFF,0xD8,0xFF,0xE0,0,0x10,'J','F','I','F',0,1,1,0,0,1,0,1,0,0,0xFF,0xDB,0,0x84,0 };
+ static const unsigned char head2[] = { 0xFF,0xDA,0,0xC,3,1,0,2,0x11,3,0x11,0,0x3F,0 };
+ const unsigned char head1[] = { 0xFF,0xC0,0,0x11,8,(unsigned char)(height>>8),STBIW_UCHAR(height),(unsigned char)(width>>8),STBIW_UCHAR(width),
+ 3,1,(unsigned char)(subsample?0x22:0x11),0,2,0x11,1,3,0x11,1,0xFF,0xC4,0x01,0xA2,0 };
+ s->func(s->context, (void*)head0, sizeof(head0));
+ s->func(s->context, (void*)YTable, sizeof(YTable));
+ stbiw__putc(s, 1);
+ s->func(s->context, UVTable, sizeof(UVTable));
+ s->func(s->context, (void*)head1, sizeof(head1));
+ s->func(s->context, (void*)(std_dc_luminance_nrcodes+1), sizeof(std_dc_luminance_nrcodes)-1);
+ s->func(s->context, (void*)std_dc_luminance_values, sizeof(std_dc_luminance_values));
+ stbiw__putc(s, 0x10); // HTYACinfo
+ s->func(s->context, (void*)(std_ac_luminance_nrcodes+1), sizeof(std_ac_luminance_nrcodes)-1);
+ s->func(s->context, (void*)std_ac_luminance_values, sizeof(std_ac_luminance_values));
+ stbiw__putc(s, 1); // HTUDCinfo
+ s->func(s->context, (void*)(std_dc_chrominance_nrcodes+1), sizeof(std_dc_chrominance_nrcodes)-1);
+ s->func(s->context, (void*)std_dc_chrominance_values, sizeof(std_dc_chrominance_values));
+ stbiw__putc(s, 0x11); // HTUACinfo
+ s->func(s->context, (void*)(std_ac_chrominance_nrcodes+1), sizeof(std_ac_chrominance_nrcodes)-1);
+ s->func(s->context, (void*)std_ac_chrominance_values, sizeof(std_ac_chrominance_values));
+ s->func(s->context, (void*)head2, sizeof(head2));
+ }
+
+ // Encode 8x8 macroblocks
+ {
+ static const unsigned short fillBits[] = {0x7F, 7};
+ int DCY=0, DCU=0, DCV=0;
+ int bitBuf=0, bitCnt=0;
+ // comp == 2 is grey+alpha (alpha is ignored)
+ int ofsG = comp > 2 ? 1 : 0, ofsB = comp > 2 ? 2 : 0;
+ const unsigned char *dataR = (const unsigned char *)data;
+ const unsigned char *dataG = dataR + ofsG;
+ const unsigned char *dataB = dataR + ofsB;
+ int x, y, pos;
+ if(subsample) {
+ for(y = 0; y < height; y += 16) {
+ for(x = 0; x < width; x += 16) {
+ float Y[256], U[256], V[256];
+ for(row = y, pos = 0; row < y+16; ++row) {
+ // row >= height => use last input row
+ int clamped_row = (row < height) ? row : height - 1;
+ int base_p = (stbi__flip_vertically_on_write ? (height-1-clamped_row) : clamped_row)*width*comp;
+ for(col = x; col < x+16; ++col, ++pos) {
+ // if col >= width => use pixel from last input column
+ int p = base_p + ((col < width) ? col : (width-1))*comp;
+ float r = dataR[p], g = dataG[p], b = dataB[p];
+ Y[pos]= +0.29900f*r + 0.58700f*g + 0.11400f*b - 128;
+ U[pos]= -0.16874f*r - 0.33126f*g + 0.50000f*b;
+ V[pos]= +0.50000f*r - 0.41869f*g - 0.08131f*b;
+ }
+ }
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+0, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+8, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+128, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+136, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+
+ // subsample U,V
+ {
+ float subU[64], subV[64];
+ int yy, xx;
+ for(yy = 0, pos = 0; yy < 8; ++yy) {
+ for(xx = 0; xx < 8; ++xx, ++pos) {
+ int j = yy*32+xx*2;
+ subU[pos] = (U[j+0] + U[j+1] + U[j+16] + U[j+17]) * 0.25f;
+ subV[pos] = (V[j+0] + V[j+1] + V[j+16] + V[j+17]) * 0.25f;
+ }
+ }
+ DCU = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, subU, 8, fdtbl_UV, DCU, UVDC_HT, UVAC_HT);
+ DCV = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, subV, 8, fdtbl_UV, DCV, UVDC_HT, UVAC_HT);
+ }
+ }
+ }
+ } else {
+ for(y = 0; y < height; y += 8) {
+ for(x = 0; x < width; x += 8) {
+ float Y[64], U[64], V[64];
+ for(row = y, pos = 0; row < y+8; ++row) {
+ // row >= height => use last input row
+ int clamped_row = (row < height) ? row : height - 1;
+ int base_p = (stbi__flip_vertically_on_write ? (height-1-clamped_row) : clamped_row)*width*comp;
+ for(col = x; col < x+8; ++col, ++pos) {
+ // if col >= width => use pixel from last input column
+ int p = base_p + ((col < width) ? col : (width-1))*comp;
+ float r = dataR[p], g = dataG[p], b = dataB[p];
+ Y[pos]= +0.29900f*r + 0.58700f*g + 0.11400f*b - 128;
+ U[pos]= -0.16874f*r - 0.33126f*g + 0.50000f*b;
+ V[pos]= +0.50000f*r - 0.41869f*g - 0.08131f*b;
+ }
+ }
+
+ DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y, 8, fdtbl_Y, DCY, YDC_HT, YAC_HT);
+ DCU = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, U, 8, fdtbl_UV, DCU, UVDC_HT, UVAC_HT);
+ DCV = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, V, 8, fdtbl_UV, DCV, UVDC_HT, UVAC_HT);
+ }
+ }
+ }
+
+ // Do the bit alignment of the EOI marker
+ stbiw__jpg_writeBits(s, &bitBuf, &bitCnt, fillBits);
+ }
+
+ // EOI
+ stbiw__putc(s, 0xFF);
+ stbiw__putc(s, 0xD9);
+
+ return 1;
+}
+
+STBIWDEF int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality)
+{
+ stbi__write_context s = { 0 };
+ stbi__start_write_callbacks(&s, func, context);
+ return stbi_write_jpg_core(&s, x, y, comp, (void *) data, quality);
+}
+
+
+#ifndef STBI_WRITE_NO_STDIO
+STBIWDEF int stbi_write_jpg(char const *filename, int x, int y, int comp, const void *data, int quality)
+{
+ stbi__write_context s = { 0 };
+ if (stbi__start_write_file(&s,filename)) {
+ int r = stbi_write_jpg_core(&s, x, y, comp, data, quality);
+ stbi__end_write_file(&s);
+ return r;
+ } else
+ return 0;
+}
+#endif
+
+#endif // STB_IMAGE_WRITE_IMPLEMENTATION
+
+/* Revision history
+ 1.16 (2021-07-11)
+ make Deflate code emit uncompressed blocks when it would otherwise expand
+ support writing BMPs with alpha channel
+ 1.15 (2020-07-13) unknown
+ 1.14 (2020-02-02) updated JPEG writer to downsample chroma channels
+ 1.13
+ 1.12
+ 1.11 (2019-08-11)
+
+ 1.10 (2019-02-07)
+ support utf8 filenames in Windows; fix warnings and platform ifdefs
+ 1.09 (2018-02-11)
+ fix typo in zlib quality API, improve STB_I_W_STATIC in C++
+ 1.08 (2018-01-29)
+ add stbi__flip_vertically_on_write, external zlib, zlib quality, choose PNG filter
+ 1.07 (2017-07-24)
+ doc fix
+ 1.06 (2017-07-23)
+ writing JPEG (using Jon Olick's code)
+ 1.05 ???
+ 1.04 (2017-03-03)
+ monochrome BMP expansion
+ 1.03 ???
+ 1.02 (2016-04-02)
+ avoid allocating large structures on the stack
+ 1.01 (2016-01-16)
+ STBIW_REALLOC_SIZED: support allocators with no realloc support
+ avoid race-condition in crc initialization
+ minor compile issues
+ 1.00 (2015-09-14)
+ installable file IO function
+ 0.99 (2015-09-13)
+ warning fixes; TGA rle support
+ 0.98 (2015-04-08)
+ added STBIW_MALLOC, STBIW_ASSERT etc
+ 0.97 (2015-01-18)
+ fixed HDR asserts, rewrote HDR rle logic
+ 0.96 (2015-01-17)
+ add HDR output
+ fix monochrome BMP
+ 0.95 (2014-08-17)
+ add monochrome TGA output
+ 0.94 (2014-05-31)
+ rename private functions to avoid conflicts with stb_image.h
+ 0.93 (2014-05-27)
+ warning fixes
+ 0.92 (2010-08-01)
+ casts to unsigned char to fix warnings
+ 0.91 (2010-07-17)
+ first public release
+ 0.90 first internal release
+*/
+
+/*
+------------------------------------------------------------------------------
+This software is available under 2 licenses -- choose whichever you prefer.
+------------------------------------------------------------------------------
+ALTERNATIVE A - MIT License
+Copyright (c) 2017 Sean Barrett
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
+of the Software, and to permit persons to whom the Software is furnished to do
+so, subject to the following conditions:
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+------------------------------------------------------------------------------
+ALTERNATIVE B - Public Domain (www.unlicense.org)
+This is free and unencumbered software released into the public domain.
+Anyone is free to copy, modify, publish, use, compile, sell, or distribute this
+software, either in source code form or as a compiled binary, for any purpose,
+commercial or non-commercial, and by any means.
+In jurisdictions that recognize copyright laws, the author or authors of this
+software dedicate any and all copyright interest in the software to the public
+domain. We make this dedication for the benefit of the public at large and to
+the detriment of our heirs and successors. We intend this dedication to be an
+overt act of relinquishment in perpetuity of all present and future rights to
+this software under copyright law.
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+------------------------------------------------------------------------------
+*/
diff --git a/ThirdParty/stb/stb_truetype.h b/ThirdParty/stb/stb_truetype.h
index 927ee21..bbf2284 100644
--- a/ThirdParty/stb/stb_truetype.h
+++ b/ThirdParty/stb/stb_truetype.h
@@ -1,5 +1,14 @@
-// stb_truetype.h - v1.19 - public domain
-// authored from 2009-2016 by Sean Barrett / RAD Game Tools
+// stb_truetype.h - v1.26 - public domain
+// authored from 2009-2021 by Sean Barrett / RAD Game Tools
+//
+// =======================================================================
+//
+// NO SECURITY GUARANTEE -- DO NOT USE THIS ON UNTRUSTED FONT FILES
+//
+// This library does no range checking of the offsets found in the file,
+// meaning an attacker can use it to read arbitrary memory.
+//
+// =======================================================================
//
// This library processes TrueType files:
// parse files
@@ -32,11 +41,11 @@
// Daniel Ribeiro Maciel
//
// Bug/warning reports/fixes:
-// "Zer" on mollyrocket Fabian "ryg" Giesen
-// Cass Everitt Martins Mozeiko
-// stoiko (Haemimont Games) Cap Petschulat
-// Brian Hook Omar Cornut
-// Walter van Niftrik github:aloucks
+// "Zer" on mollyrocket Fabian "ryg" Giesen github:NiLuJe
+// Cass Everitt Martins Mozeiko github:aloucks
+// stoiko (Haemimont Games) Cap Petschulat github:oyvindjam
+// Brian Hook Omar Cornut github:vassvik
+// Walter van Niftrik Ryan Griege
// David Gow Peter LaValle
// David Given Sergey Popov
// Ivan-Assen Ivanov Giumo X. Clanjor
@@ -44,11 +53,18 @@
// Johan Duparc Thomas Fields
// Hou Qiming Derek Vinyard
// Rob Loach Cort Stratton
-// Kenney Phillis Jr. github:oyvindjam
-// Brian Costabile github:vassvik
-//
+// Kenney Phillis Jr. Brian Costabile
+// Ken Voskuil (kaesve)
+//
// VERSION HISTORY
//
+// 1.26 (2021-08-28) fix broken rasterizer
+// 1.25 (2021-07-11) many fixes
+// 1.24 (2020-02-05) fix warning
+// 1.23 (2020-02-02) query SVG data for glyphs; query whole kerning table (but only kern not GPOS)
+// 1.22 (2019-08-11) minimize missing-glyph duplication; fix kerning if both 'GPOS' and 'kern' are defined
+// 1.21 (2019-02-25) fix warning
+// 1.20 (2019-02-07) PackFontRange skips missing codepoints; GetScaleFontVMetrics()
// 1.19 (2018-02-11) GPOS kerning, STBTT_fmod
// 1.18 (2018-01-29) add missing function
// 1.17 (2017-07-23) make more arguments const; doc fix
@@ -75,7 +91,7 @@
//
// USAGE
//
-// Include this file in whatever places neeed to refer to it. In ONE C/C++
+// Include this file in whatever places need to refer to it. In ONE C/C++
// file, write:
// #define STB_TRUETYPE_IMPLEMENTATION
// before the #include of this file. This expands out the actual
@@ -206,7 +222,7 @@
//
// Advancing for the next character:
// Call GlyphHMetrics, and compute 'current_point += SF * advance'.
-//
+//
//
// ADVANCED USAGE
//
@@ -242,19 +258,6 @@
// recommend it.
//
//
-// SOURCE STATISTICS (based on v0.6c, 2050 LOC)
-//
-// Documentation & header file 520 LOC \___ 660 LOC documentation
-// Sample code 140 LOC /
-// Truetype parsing 620 LOC ---- 620 LOC TrueType
-// Software rasterization 240 LOC \ .
-// Curve tesselation 120 LOC \__ 550 LOC Bitmap creation
-// Bitmap management 100 LOC /
-// Baked bitmap interface 70 LOC /
-// Font name matching & access 150 LOC ---- 150
-// C runtime library abstraction 60 LOC ---- 60
-//
-//
// PERFORMANCE MEASUREMENTS FOR 1.06:
//
// 32-bit 64-bit
@@ -269,48 +272,50 @@
//// SAMPLE PROGRAMS
////
//
-// Incomplete text-in-3d-api example, which draws quads properly aligned to be lossless
-//
+// Incomplete text-in-3d-api example, which draws quads properly aligned to be lossless.
+// See "tests/truetype_demo_win32.c" for a complete version.
#if 0
#define STB_TRUETYPE_IMPLEMENTATION // force following include to generate implementation
#include "stb_truetype.h"
-unsigned char ttf_buffer[1 << 20];
-unsigned char temp_bitmap[512 * 512];
+unsigned char ttf_buffer[1<<20];
+unsigned char temp_bitmap[512*512];
stbtt_bakedchar cdata[96]; // ASCII 32..126 is 95 glyphs
GLuint ftex;
void my_stbtt_initfont(void)
{
- fread(ttf_buffer, 1, 1 << 20, fopen("c:/windows/fonts/times.ttf", "rb"));
- stbtt_BakeFontBitmap(ttf_buffer, 0, 32.0, temp_bitmap, 512, 512, 32, 96, cdata); // no guarantee this fits!
- // can free ttf_buffer at this point
- glGenTextures(1, &ftex);
- glBindTexture(GL_TEXTURE_2D, ftex);
- glTexImage2D(GL_TEXTURE_2D, 0, GL_ALPHA, 512, 512, 0, GL_ALPHA, GL_UNSIGNED_BYTE, temp_bitmap);
- // can free temp_bitmap at this point
- glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+ fread(ttf_buffer, 1, 1<<20, fopen("c:/windows/fonts/times.ttf", "rb"));
+ stbtt_BakeFontBitmap(ttf_buffer,0, 32.0, temp_bitmap,512,512, 32,96, cdata); // no guarantee this fits!
+ // can free ttf_buffer at this point
+ glGenTextures(1, &ftex);
+ glBindTexture(GL_TEXTURE_2D, ftex);
+ glTexImage2D(GL_TEXTURE_2D, 0, GL_ALPHA, 512,512, 0, GL_ALPHA, GL_UNSIGNED_BYTE, temp_bitmap);
+ // can free temp_bitmap at this point
+ glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
}
void my_stbtt_print(float x, float y, char *text)
{
- // assume orthographic projection with units = screen pixels, origin at top left
- glEnable(GL_TEXTURE_2D);
- glBindTexture(GL_TEXTURE_2D, ftex);
- glBegin(GL_QUADS);
- while (*text) {
- if (*text >= 32 && *text < 128) {
- stbtt_aligned_quad q;
- stbtt_GetBakedQuad(cdata, 512, 512, *text - 32, &x, &y, &q, 1);//1=opengl & d3d10+,0=d3d9
- glTexCoord2f(q.s0, q.t1); glVertex2f(q.x0, q.y0);
- glTexCoord2f(q.s1, q.t1); glVertex2f(q.x1, q.y0);
- glTexCoord2f(q.s1, q.t0); glVertex2f(q.x1, q.y1);
- glTexCoord2f(q.s0, q.t0); glVertex2f(q.x0, q.y1);
- }
- ++text;
- }
- glEnd();
+ // assume orthographic projection with units = screen pixels, origin at top left
+ glEnable(GL_BLEND);
+ glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA);
+ glEnable(GL_TEXTURE_2D);
+ glBindTexture(GL_TEXTURE_2D, ftex);
+ glBegin(GL_QUADS);
+ while (*text) {
+ if (*text >= 32 && *text < 128) {
+ stbtt_aligned_quad q;
+ stbtt_GetBakedQuad(cdata, 512,512, *text-32, &x,&y,&q,1);//1=opengl & d3d10+,0=d3d9
+ glTexCoord2f(q.s0,q.t0); glVertex2f(q.x0,q.y0);
+ glTexCoord2f(q.s1,q.t0); glVertex2f(q.x1,q.y0);
+ glTexCoord2f(q.s1,q.t1); glVertex2f(q.x1,q.y1);
+ glTexCoord2f(q.s0,q.t1); glVertex2f(q.x0,q.y1);
+ }
+ ++text;
+ }
+ glEnd();
}
#endif
//
@@ -324,27 +329,27 @@ void my_stbtt_print(float x, float y, char *text)
#define STB_TRUETYPE_IMPLEMENTATION // force following include to generate implementation
#include "stb_truetype.h"
-char ttf_buffer[1 << 25];
+char ttf_buffer[1<<25];
int main(int argc, char **argv)
{
- stbtt_fontinfo font;
- unsigned char *bitmap;
- int w, h, i, j, c = (argc > 1 ? atoi(argv[1]) : 'a'), s = (argc > 2 ? atoi(argv[2]) : 20);
+ stbtt_fontinfo font;
+ unsigned char *bitmap;
+ int w,h,i,j,c = (argc > 1 ? atoi(argv[1]) : 'a'), s = (argc > 2 ? atoi(argv[2]) : 20);
- fread(ttf_buffer, 1, 1 << 25, fopen(argc > 3 ? argv[3] : "c:/windows/fonts/arialbd.ttf", "rb"));
+ fread(ttf_buffer, 1, 1<<25, fopen(argc > 3 ? argv[3] : "c:/windows/fonts/arialbd.ttf", "rb"));
- stbtt_InitFont(&font, ttf_buffer, stbtt_GetFontOffsetForIndex(ttf_buffer, 0));
- bitmap = stbtt_GetCodepointBitmap(&font, 0, stbtt_ScaleForPixelHeight(&font, s), c, &w, &h, 0, 0);
+ stbtt_InitFont(&font, ttf_buffer, stbtt_GetFontOffsetForIndex(ttf_buffer,0));
+ bitmap = stbtt_GetCodepointBitmap(&font, 0,stbtt_ScaleForPixelHeight(&font, s), c, &w, &h, 0,0);
- for (j = 0; j < h; ++j) {
- for (i = 0; i < w; ++i)
- putchar(" .:ioVM@"[bitmap[j*w + i] >> 5]);
- putchar('\n');
- }
- return 0;
+ for (j=0; j < h; ++j) {
+ for (i=0; i < w; ++i)
+ putchar(" .:ioVM@"[bitmap[j*w+i]>>5]);
+ putchar('\n');
+ }
+ return 0;
}
-#endif
+#endif
//
// Output:
//
@@ -358,52 +363,52 @@ int main(int argc, char **argv)
// :@@. M@M
// @@@o@@@@
// :M@@V:@@.
-//
+//
//////////////////////////////////////////////////////////////////////////////
-//
+//
// Complete program: print "Hello World!" banner, with bugs
//
#if 0
-char buffer[24 << 20];
+char buffer[24<<20];
unsigned char screen[20][79];
int main(int arg, char **argv)
{
- stbtt_fontinfo font;
- int i, j, ascent, baseline, ch = 0;
- float scale, xpos = 2; // leave a little padding in case the character extends left
- char *text = "Heljo World!"; // intentionally misspelled to show 'lj' brokenness
-
- fread(buffer, 1, 1000000, fopen("c:/windows/fonts/arialbd.ttf", "rb"));
- stbtt_InitFont(&font, buffer, 0);
-
- scale = stbtt_ScaleForPixelHeight(&font, 15);
- stbtt_GetFontVMetrics(&font, &ascent, 0, 0);
- baseline = (int)(ascent*scale);
-
- while (text[ch]) {
- int advance, lsb, x0, y0, x1, y1;
- float x_shift = xpos - (float)floor(xpos);
- stbtt_GetCodepointHMetrics(&font, text[ch], &advance, &lsb);
- stbtt_GetCodepointBitmapBoxSubpixel(&font, text[ch], scale, scale, x_shift, 0, &x0, &y0, &x1, &y1);
- stbtt_MakeCodepointBitmapSubpixel(&font, &screen[baseline + y0][(int)xpos + x0], x1 - x0, y1 - y0, 79, scale, scale, x_shift, 0, text[ch]);
- // note that this stomps the old data, so where character boxes overlap (e.g. 'lj') it's wrong
- // because this API is really for baking character bitmaps into textures. if you want to render
- // a sequence of characters, you really need to render each bitmap to a temp buffer, then
- // "alpha blend" that into the working buffer
- xpos += (advance * scale);
- if (text[ch + 1])
- xpos += scale * stbtt_GetCodepointKernAdvance(&font, text[ch], text[ch + 1]);
- ++ch;
- }
-
- for (j = 0; j < 20; ++j) {
- for (i = 0; i < 78; ++i)
- putchar(" .:ioVM@"[screen[j][i] >> 5]);
- putchar('\n');
- }
-
- return 0;
+ stbtt_fontinfo font;
+ int i,j,ascent,baseline,ch=0;
+ float scale, xpos=2; // leave a little padding in case the character extends left
+ char *text = "Heljo World!"; // intentionally misspelled to show 'lj' brokenness
+
+ fread(buffer, 1, 1000000, fopen("c:/windows/fonts/arialbd.ttf", "rb"));
+ stbtt_InitFont(&font, buffer, 0);
+
+ scale = stbtt_ScaleForPixelHeight(&font, 15);
+ stbtt_GetFontVMetrics(&font, &ascent,0,0);
+ baseline = (int) (ascent*scale);
+
+ while (text[ch]) {
+ int advance,lsb,x0,y0,x1,y1;
+ float x_shift = xpos - (float) floor(xpos);
+ stbtt_GetCodepointHMetrics(&font, text[ch], &advance, &lsb);
+ stbtt_GetCodepointBitmapBoxSubpixel(&font, text[ch], scale,scale,x_shift,0, &x0,&y0,&x1,&y1);
+ stbtt_MakeCodepointBitmapSubpixel(&font, &screen[baseline + y0][(int) xpos + x0], x1-x0,y1-y0, 79, scale,scale,x_shift,0, text[ch]);
+ // note that this stomps the old data, so where character boxes overlap (e.g. 'lj') it's wrong
+ // because this API is really for baking character bitmaps into textures. if you want to render
+ // a sequence of characters, you really need to render each bitmap to a temp buffer, then
+ // "alpha blend" that into the working buffer
+ xpos += (advance * scale);
+ if (text[ch+1])
+ xpos += scale*stbtt_GetCodepointKernAdvance(&font, text[ch],text[ch+1]);
+ ++ch;
+ }
+
+ for (j=0; j < 20; ++j) {
+ for (i=0; i < 78; ++i)
+ putchar(" .:ioVM@"[screen[j][i]>>5]);
+ putchar('\n');
+ }
+
+ return 0;
}
#endif
@@ -418,70 +423,70 @@ int main(int arg, char **argv)
//// link with the C runtime library.
#ifdef STB_TRUETYPE_IMPLEMENTATION
-// #define your own (u)stbtt_int8/16/32 before including to override this
-#ifndef stbtt_uint8
-typedef unsigned char stbtt_uint8;
-typedef signed char stbtt_int8;
-typedef unsigned short stbtt_uint16;
-typedef signed short stbtt_int16;
-typedef unsigned int stbtt_uint32;
-typedef signed int stbtt_int32;
-#endif
-
-typedef char stbtt__check_size32[sizeof(stbtt_int32) == 4 ? 1 : -1];
-typedef char stbtt__check_size16[sizeof(stbtt_int16) == 2 ? 1 : -1];
-
-// e.g. #define your own STBTT_ifloor/STBTT_iceil() to avoid math.h
-#ifndef STBTT_ifloor
-#include <math.h>
-#define STBTT_ifloor(x) ((int) floor(x))
-#define STBTT_iceil(x) ((int) ceil(x))
-#endif
-
-#ifndef STBTT_sqrt
-#include <math.h>
-#define STBTT_sqrt(x) sqrt(x)
-#define STBTT_pow(x,y) pow(x,y)
-#endif
-
-#ifndef STBTT_fmod
-#include <math.h>
-#define STBTT_fmod(x,y) fmod(x,y)
-#endif
-
-#ifndef STBTT_cos
-#include <math.h>
-#define STBTT_cos(x) cos(x)
-#define STBTT_acos(x) acos(x)
-#endif
-
-#ifndef STBTT_fabs
-#include <math.h>
-#define STBTT_fabs(x) fabs(x)
-#endif
-
-// #define your own functions "STBTT_malloc" / "STBTT_free" to avoid malloc.h
-#ifndef STBTT_malloc
-#include <stdlib.h>
-#define STBTT_malloc(x,u) ((void)(u),malloc(x))
-#define STBTT_free(x,u) ((void)(u),free(x))
-#endif
-
-#ifndef STBTT_assert
-#include <assert.h>
-#define STBTT_assert(x) assert(x)
-#endif
-
-#ifndef STBTT_strlen
-#include <string.h>
-#define STBTT_strlen(x) strlen(x)
-#endif
-
-#ifndef STBTT_memcpy
-#include <string.h>
-#define STBTT_memcpy memcpy
-#define STBTT_memset memset
-#endif
+ // #define your own (u)stbtt_int8/16/32 before including to override this
+ #ifndef stbtt_uint8
+ typedef unsigned char stbtt_uint8;
+ typedef signed char stbtt_int8;
+ typedef unsigned short stbtt_uint16;
+ typedef signed short stbtt_int16;
+ typedef unsigned int stbtt_uint32;
+ typedef signed int stbtt_int32;
+ #endif
+
+ typedef char stbtt__check_size32[sizeof(stbtt_int32)==4 ? 1 : -1];
+ typedef char stbtt__check_size16[sizeof(stbtt_int16)==2 ? 1 : -1];
+
+ // e.g. #define your own STBTT_ifloor/STBTT_iceil() to avoid math.h
+ #ifndef STBTT_ifloor
+ #include <math.h>
+ #define STBTT_ifloor(x) ((int) floor(x))
+ #define STBTT_iceil(x) ((int) ceil(x))
+ #endif
+
+ #ifndef STBTT_sqrt
+ #include <math.h>
+ #define STBTT_sqrt(x) sqrt(x)
+ #define STBTT_pow(x,y) pow(x,y)
+ #endif
+
+ #ifndef STBTT_fmod
+ #include <math.h>
+ #define STBTT_fmod(x,y) fmod(x,y)
+ #endif
+
+ #ifndef STBTT_cos
+ #include <math.h>
+ #define STBTT_cos(x) cos(x)
+ #define STBTT_acos(x) acos(x)
+ #endif
+
+ #ifndef STBTT_fabs
+ #include <math.h>
+ #define STBTT_fabs(x) fabs(x)
+ #endif
+
+ // #define your own functions "STBTT_malloc" / "STBTT_free" to avoid malloc.h
+ #ifndef STBTT_malloc
+ #include <stdlib.h>
+ #define STBTT_malloc(x,u) ((void)(u),malloc(x))
+ #define STBTT_free(x,u) ((void)(u),free(x))
+ #endif
+
+ #ifndef STBTT_assert
+ #include <assert.h>
+ #define STBTT_assert(x) assert(x)
+ #endif
+
+ #ifndef STBTT_strlen
+ #include <string.h>
+ #define STBTT_strlen(x) strlen(x)
+ #endif
+
+ #ifndef STBTT_memcpy
+ #include <string.h>
+ #define STBTT_memcpy memcpy
+ #define STBTT_memset memset
+ #endif
#endif
///////////////////////////////////////////////////////////////////////////////
@@ -504,558 +509,586 @@ typedef char stbtt__check_size16[sizeof(stbtt_int16) == 2 ? 1 : -1];
extern "C" {
#endif
- // private structure
- typedef struct
- {
- unsigned char *data;
- int cursor;
- int size;
- } stbtt__buf;
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // TEXTURE BAKING API
- //
- // If you use this API, you only have to call two functions ever.
- //
-
- typedef struct
- {
- unsigned short x0, y0, x1, y1; // coordinates of bbox in bitmap
- float xoff, yoff, xadvance;
- } stbtt_bakedchar;
-
- STBTT_DEF int stbtt_BakeFontBitmap(const unsigned char *data, int offset, // font location (use offset=0 for plain .ttf)
- float pixel_height, // height of font in pixels
- unsigned char *pixels, int pw, int ph, // bitmap to be filled in
- int first_char, int num_chars, // characters to bake
- stbtt_bakedchar *chardata); // you allocate this, it's num_chars long
- // if return is positive, the first unused row of the bitmap
- // if return is negative, returns the negative of the number of characters that fit
- // if return is 0, no characters fit and no rows were used
- // This uses a very crappy packing.
-
- typedef struct
- {
- float x0, y0, s0, t0; // top-left
- float x1, y1, s1, t1; // bottom-right
- } stbtt_aligned_quad;
-
- STBTT_DEF void stbtt_GetBakedQuad(const stbtt_bakedchar *chardata, int pw, int ph, // same data as above
- int char_index, // character to display
- float *xpos, float *ypos, // pointers to current position in screen pixel space
- stbtt_aligned_quad *q, // output: quad to draw
- int opengl_fillrule); // true if opengl fill rule; false if DX9 or earlier
- // Call GetBakedQuad with char_index = 'character - first_char', and it
- // creates the quad you need to draw and advances the current position.
- //
- // The coordinate system used assumes y increases downwards.
- //
- // Characters will extend both above and below the current position;
- // see discussion of "BASELINE" above.
- //
- // It's inefficient; you might want to c&p it and optimize it.
-
-
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // NEW TEXTURE BAKING API
- //
- // This provides options for packing multiple fonts into one atlas, not
- // perfectly but better than nothing.
-
- typedef struct
- {
- unsigned short x0, y0, x1, y1; // coordinates of bbox in bitmap
- float xoff, yoff, xadvance;
- float xoff2, yoff2;
- } stbtt_packedchar;
-
- typedef struct stbtt_pack_context stbtt_pack_context;
- typedef struct stbtt_fontinfo stbtt_fontinfo;
+// private structure
+typedef struct
+{
+ unsigned char *data;
+ int cursor;
+ int size;
+} stbtt__buf;
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// TEXTURE BAKING API
+//
+// If you use this API, you only have to call two functions ever.
+//
+
+typedef struct
+{
+ unsigned short x0,y0,x1,y1; // coordinates of bbox in bitmap
+ float xoff,yoff,xadvance;
+} stbtt_bakedchar;
+
+STBTT_DEF int stbtt_BakeFontBitmap(const unsigned char *data, int offset, // font location (use offset=0 for plain .ttf)
+ float pixel_height, // height of font in pixels
+ unsigned char *pixels, int pw, int ph, // bitmap to be filled in
+ int first_char, int num_chars, // characters to bake
+ stbtt_bakedchar *chardata); // you allocate this, it's num_chars long
+// if return is positive, the first unused row of the bitmap
+// if return is negative, returns the negative of the number of characters that fit
+// if return is 0, no characters fit and no rows were used
+// This uses a very crappy packing.
+
+typedef struct
+{
+ float x0,y0,s0,t0; // top-left
+ float x1,y1,s1,t1; // bottom-right
+} stbtt_aligned_quad;
+
+STBTT_DEF void stbtt_GetBakedQuad(const stbtt_bakedchar *chardata, int pw, int ph, // same data as above
+ int char_index, // character to display
+ float *xpos, float *ypos, // pointers to current position in screen pixel space
+ stbtt_aligned_quad *q, // output: quad to draw
+ int opengl_fillrule); // true if opengl fill rule; false if DX9 or earlier
+// Call GetBakedQuad with char_index = 'character - first_char', and it
+// creates the quad you need to draw and advances the current position.
+//
+// The coordinate system used assumes y increases downwards.
+//
+// Characters will extend both above and below the current position;
+// see discussion of "BASELINE" above.
+//
+// It's inefficient; you might want to c&p it and optimize it.
+
+STBTT_DEF void stbtt_GetScaledFontVMetrics(const unsigned char *fontdata, int index, float size, float *ascent, float *descent, float *lineGap);
+// Query the font vertical metrics without having to create a font first.
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// NEW TEXTURE BAKING API
+//
+// This provides options for packing multiple fonts into one atlas, not
+// perfectly but better than nothing.
+
+typedef struct
+{
+ unsigned short x0,y0,x1,y1; // coordinates of bbox in bitmap
+ float xoff,yoff,xadvance;
+ float xoff2,yoff2;
+} stbtt_packedchar;
+
+typedef struct stbtt_pack_context stbtt_pack_context;
+typedef struct stbtt_fontinfo stbtt_fontinfo;
#ifndef STB_RECT_PACK_VERSION
- typedef struct stbrp_rect stbrp_rect;
+typedef struct stbrp_rect stbrp_rect;
#endif
- STBTT_DEF int stbtt_PackBegin(stbtt_pack_context *spc, unsigned char *pixels, int width, int height, int stride_in_bytes, int padding, void *alloc_context);
- // Initializes a packing context stored in the passed-in stbtt_pack_context.
- // Future calls using this context will pack characters into the bitmap passed
- // in here: a 1-channel bitmap that is width * height. stride_in_bytes is
- // the distance from one row to the next (or 0 to mean they are packed tightly
- // together). "padding" is the amount of padding to leave between each
- // character (normally you want '1' for bitmaps you'll use as textures with
- // bilinear filtering).
- //
- // Returns 0 on failure, 1 on success.
+STBTT_DEF int stbtt_PackBegin(stbtt_pack_context *spc, unsigned char *pixels, int width, int height, int stride_in_bytes, int padding, void *alloc_context);
+// Initializes a packing context stored in the passed-in stbtt_pack_context.
+// Future calls using this context will pack characters into the bitmap passed
+// in here: a 1-channel bitmap that is width * height. stride_in_bytes is
+// the distance from one row to the next (or 0 to mean they are packed tightly
+// together). "padding" is the amount of padding to leave between each
+// character (normally you want '1' for bitmaps you'll use as textures with
+// bilinear filtering).
+//
+// Returns 0 on failure, 1 on success.
- STBTT_DEF void stbtt_PackEnd(stbtt_pack_context *spc);
- // Cleans up the packing context and frees all memory.
+STBTT_DEF void stbtt_PackEnd (stbtt_pack_context *spc);
+// Cleans up the packing context and frees all memory.
#define STBTT_POINT_SIZE(x) (-(x))
- STBTT_DEF int stbtt_PackFontRange(stbtt_pack_context *spc, const unsigned char *fontdata, int font_index, float font_size,
- int first_unicode_char_in_range, int num_chars_in_range, stbtt_packedchar *chardata_for_range);
- // Creates character bitmaps from the font_index'th font found in fontdata (use
- // font_index=0 if you don't know what that is). It creates num_chars_in_range
- // bitmaps for characters with unicode values starting at first_unicode_char_in_range
- // and increasing. Data for how to render them is stored in chardata_for_range;
- // pass these to stbtt_GetPackedQuad to get back renderable quads.
- //
- // font_size is the full height of the character from ascender to descender,
- // as computed by stbtt_ScaleForPixelHeight. To use a point size as computed
- // by stbtt_ScaleForMappingEmToPixels, wrap the point size in STBTT_POINT_SIZE()
- // and pass that result as 'font_size':
- // ..., 20 , ... // font max minus min y is 20 pixels tall
- // ..., STBTT_POINT_SIZE(20), ... // 'M' is 20 pixels tall
-
- typedef struct
- {
- float font_size;
- int first_unicode_codepoint_in_range; // if non-zero, then the chars are continuous, and this is the first codepoint
- int *array_of_unicode_codepoints; // if non-zero, then this is an array of unicode codepoints
- int num_chars;
- stbtt_packedchar *chardata_for_range; // output
- unsigned char h_oversample, v_oversample; // don't set these, they're used internally
- } stbtt_pack_range;
-
- STBTT_DEF int stbtt_PackFontRanges(stbtt_pack_context *spc, const unsigned char *fontdata, int font_index, stbtt_pack_range *ranges, int num_ranges);
- // Creates character bitmaps from multiple ranges of characters stored in
- // ranges. This will usually create a better-packed bitmap than multiple
- // calls to stbtt_PackFontRange. Note that you can call this multiple
- // times within a single PackBegin/PackEnd.
-
- STBTT_DEF void stbtt_PackSetOversampling(stbtt_pack_context *spc, unsigned int h_oversample, unsigned int v_oversample);
- // Oversampling a font increases the quality by allowing higher-quality subpixel
- // positioning, and is especially valuable at smaller text sizes.
- //
- // This function sets the amount of oversampling for all following calls to
- // stbtt_PackFontRange(s) or stbtt_PackFontRangesGatherRects for a given
- // pack context. The default (no oversampling) is achieved by h_oversample=1
- // and v_oversample=1. The total number of pixels required is
- // h_oversample*v_oversample larger than the default; for example, 2x2
- // oversampling requires 4x the storage of 1x1. For best results, render
- // oversampled textures with bilinear filtering. Look at the readme in
- // stb/tests/oversample for information about oversampled fonts
- //
- // To use with PackFontRangesGather etc., you must set it before calls
- // call to PackFontRangesGatherRects.
-
- STBTT_DEF void stbtt_GetPackedQuad(const stbtt_packedchar *chardata, int pw, int ph, // same data as above
- int char_index, // character to display
- float *xpos, float *ypos, // pointers to current position in screen pixel space
- stbtt_aligned_quad *q, // output: quad to draw
- int align_to_integer);
-
- STBTT_DEF int stbtt_PackFontRangesGatherRects(stbtt_pack_context *spc, const stbtt_fontinfo *info, stbtt_pack_range *ranges, int num_ranges, stbrp_rect *rects);
- STBTT_DEF void stbtt_PackFontRangesPackRects(stbtt_pack_context *spc, stbrp_rect *rects, int num_rects);
- STBTT_DEF int stbtt_PackFontRangesRenderIntoRects(stbtt_pack_context *spc, const stbtt_fontinfo *info, stbtt_pack_range *ranges, int num_ranges, stbrp_rect *rects);
- // Calling these functions in sequence is roughly equivalent to calling
- // stbtt_PackFontRanges(). If you more control over the packing of multiple
- // fonts, or if you want to pack custom data into a font texture, take a look
- // at the source to of stbtt_PackFontRanges() and create a custom version
- // using these functions, e.g. call GatherRects multiple times,
- // building up a single array of rects, then call PackRects once,
- // then call RenderIntoRects repeatedly. This may result in a
- // better packing than calling PackFontRanges multiple times
- // (or it may not).
-
- // this is an opaque structure that you shouldn't mess with which holds
- // all the context needed from PackBegin to PackEnd.
- struct stbtt_pack_context {
- void *user_allocator_context;
- void *pack_info;
- int width;
- int height;
- int stride_in_bytes;
- int padding;
- unsigned int h_oversample, v_oversample;
- unsigned char *pixels;
- void *nodes;
- };
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // FONT LOADING
- //
- //
-
- STBTT_DEF int stbtt_GetNumberOfFonts(const unsigned char *data);
- // This function will determine the number of fonts in a font file. TrueType
- // collection (.ttc) files may contain multiple fonts, while TrueType font
- // (.ttf) files only contain one font. The number of fonts can be used for
- // indexing with the previous function where the index is between zero and one
- // less than the total fonts. If an error occurs, -1 is returned.
-
- STBTT_DEF int stbtt_GetFontOffsetForIndex(const unsigned char *data, int index);
- // Each .ttf/.ttc file may have more than one font. Each font has a sequential
- // index number starting from 0. Call this function to get the font offset for
- // a given index; it returns -1 if the index is out of range. A regular .ttf
- // file will only define one font and it always be at offset 0, so it will
- // return '0' for index 0, and -1 for all other indices.
-
- // The following structure is defined publically so you can declare one on
- // the stack or as a global or etc, but you should treat it as opaque.
- struct stbtt_fontinfo
- {
- void * userdata;
- unsigned char * data; // pointer to .ttf file
- int fontstart; // offset of start of font
-
- int numGlyphs; // number of glyphs, needed for range checking
-
- int loca, head, glyf, hhea, hmtx, kern, gpos; // table locations as offset from start of .ttf
- int index_map; // a cmap mapping for our chosen character encoding
- int indexToLocFormat; // format needed to map from glyph index to glyph
-
- stbtt__buf cff; // cff font data
- stbtt__buf charstrings; // the charstring index
- stbtt__buf gsubrs; // global charstring subroutines index
- stbtt__buf subrs; // private charstring subroutines index
- stbtt__buf fontdicts; // array of font dicts
- stbtt__buf fdselect; // map from glyph to fontdict
- };
-
- STBTT_DEF int stbtt_InitFont(stbtt_fontinfo *info, const unsigned char *data, int offset);
- // Given an offset into the file that defines a font, this function builds
- // the necessary cached info for the rest of the system. You must allocate
- // the stbtt_fontinfo yourself, and stbtt_InitFont will fill it out. You don't
- // need to do anything special to free it, because the contents are pure
- // value data with no additional data structures. Returns 0 on failure.
-
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // CHARACTER TO GLYPH-INDEX CONVERSIOn
-
- STBTT_DEF int stbtt_FindGlyphIndex(const stbtt_fontinfo *info, int unicode_codepoint);
- // If you're going to perform multiple operations on the same character
- // and you want a speed-up, call this function with the character you're
- // going to process, then use glyph-based functions instead of the
- // codepoint-based functions.
-
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // CHARACTER PROPERTIES
- //
-
- STBTT_DEF float stbtt_ScaleForPixelHeight(const stbtt_fontinfo *info, float pixels);
- // computes a scale factor to produce a font whose "height" is 'pixels' tall.
- // Height is measured as the distance from the highest ascender to the lowest
- // descender; in other words, it's equivalent to calling stbtt_GetFontVMetrics
- // and computing:
- // scale = pixels / (ascent - descent)
- // so if you prefer to measure height by the ascent only, use a similar calculation.
-
- STBTT_DEF float stbtt_ScaleForMappingEmToPixels(const stbtt_fontinfo *info, float pixels);
- // computes a scale factor to produce a font whose EM size is mapped to
- // 'pixels' tall. This is probably what traditional APIs compute, but
- // I'm not positive.
-
- STBTT_DEF void stbtt_GetFontVMetrics(const stbtt_fontinfo *info, int *ascent, int *descent, int *lineGap);
- // ascent is the coordinate above the baseline the font extends; descent
- // is the coordinate below the baseline the font extends (i.e. it is typically negative)
- // lineGap is the spacing between one row's descent and the next row's ascent...
- // so you should advance the vertical position by "*ascent - *descent + *lineGap"
- // these are expressed in unscaled coordinates, so you must multiply by
- // the scale factor for a given size
-
- STBTT_DEF int stbtt_GetFontVMetricsOS2(const stbtt_fontinfo *info, int *typoAscent, int *typoDescent, int *typoLineGap);
- // analogous to GetFontVMetrics, but returns the "typographic" values from the OS/2
- // table (specific to MS/Windows TTF files).
- //
- // Returns 1 on success (table present), 0 on failure.
-
- STBTT_DEF void stbtt_GetFontBoundingBox(const stbtt_fontinfo *info, int *x0, int *y0, int *x1, int *y1);
- // the bounding box around all possible characters
-
- STBTT_DEF void stbtt_GetCodepointHMetrics(const stbtt_fontinfo *info, int codepoint, int *advanceWidth, int *leftSideBearing);
- // leftSideBearing is the offset from the current horizontal position to the left edge of the character
- // advanceWidth is the offset from the current horizontal position to the next horizontal position
- // these are expressed in unscaled coordinates
-
- STBTT_DEF int stbtt_GetCodepointKernAdvance(const stbtt_fontinfo *info, int ch1, int ch2);
- // an additional amount to add to the 'advance' value between ch1 and ch2
-
- STBTT_DEF int stbtt_GetCodepointBox(const stbtt_fontinfo *info, int codepoint, int *x0, int *y0, int *x1, int *y1);
- // Gets the bounding box of the visible part of the glyph, in unscaled coordinates
-
- STBTT_DEF void stbtt_GetGlyphHMetrics(const stbtt_fontinfo *info, int glyph_index, int *advanceWidth, int *leftSideBearing);
- STBTT_DEF int stbtt_GetGlyphKernAdvance(const stbtt_fontinfo *info, int glyph1, int glyph2);
- STBTT_DEF int stbtt_GetGlyphBox(const stbtt_fontinfo *info, int glyph_index, int *x0, int *y0, int *x1, int *y1);
- // as above, but takes one or more glyph indices for greater efficiency
-
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // GLYPH SHAPES (you probably don't need these, but they have to go before
- // the bitmaps for C declaration-order reasons)
- //
+STBTT_DEF int stbtt_PackFontRange(stbtt_pack_context *spc, const unsigned char *fontdata, int font_index, float font_size,
+ int first_unicode_char_in_range, int num_chars_in_range, stbtt_packedchar *chardata_for_range);
+// Creates character bitmaps from the font_index'th font found in fontdata (use
+// font_index=0 if you don't know what that is). It creates num_chars_in_range
+// bitmaps for characters with unicode values starting at first_unicode_char_in_range
+// and increasing. Data for how to render them is stored in chardata_for_range;
+// pass these to stbtt_GetPackedQuad to get back renderable quads.
+//
+// font_size is the full height of the character from ascender to descender,
+// as computed by stbtt_ScaleForPixelHeight. To use a point size as computed
+// by stbtt_ScaleForMappingEmToPixels, wrap the point size in STBTT_POINT_SIZE()
+// and pass that result as 'font_size':
+// ..., 20 , ... // font max minus min y is 20 pixels tall
+// ..., STBTT_POINT_SIZE(20), ... // 'M' is 20 pixels tall
+
+typedef struct
+{
+ float font_size;
+ int first_unicode_codepoint_in_range; // if non-zero, then the chars are continuous, and this is the first codepoint
+ int *array_of_unicode_codepoints; // if non-zero, then this is an array of unicode codepoints
+ int num_chars;
+ stbtt_packedchar *chardata_for_range; // output
+ unsigned char h_oversample, v_oversample; // don't set these, they're used internally
+} stbtt_pack_range;
+
+STBTT_DEF int stbtt_PackFontRanges(stbtt_pack_context *spc, const unsigned char *fontdata, int font_index, stbtt_pack_range *ranges, int num_ranges);
+// Creates character bitmaps from multiple ranges of characters stored in
+// ranges. This will usually create a better-packed bitmap than multiple
+// calls to stbtt_PackFontRange. Note that you can call this multiple
+// times within a single PackBegin/PackEnd.
+
+STBTT_DEF void stbtt_PackSetOversampling(stbtt_pack_context *spc, unsigned int h_oversample, unsigned int v_oversample);
+// Oversampling a font increases the quality by allowing higher-quality subpixel
+// positioning, and is especially valuable at smaller text sizes.
+//
+// This function sets the amount of oversampling for all following calls to
+// stbtt_PackFontRange(s) or stbtt_PackFontRangesGatherRects for a given
+// pack context. The default (no oversampling) is achieved by h_oversample=1
+// and v_oversample=1. The total number of pixels required is
+// h_oversample*v_oversample larger than the default; for example, 2x2
+// oversampling requires 4x the storage of 1x1. For best results, render
+// oversampled textures with bilinear filtering. Look at the readme in
+// stb/tests/oversample for information about oversampled fonts
+//
+// To use with PackFontRangesGather etc., you must set it before calls
+// call to PackFontRangesGatherRects.
+
+STBTT_DEF void stbtt_PackSetSkipMissingCodepoints(stbtt_pack_context *spc, int skip);
+// If skip != 0, this tells stb_truetype to skip any codepoints for which
+// there is no corresponding glyph. If skip=0, which is the default, then
+// codepoints without a glyph recived the font's "missing character" glyph,
+// typically an empty box by convention.
+
+STBTT_DEF void stbtt_GetPackedQuad(const stbtt_packedchar *chardata, int pw, int ph, // same data as above
+ int char_index, // character to display
+ float *xpos, float *ypos, // pointers to current position in screen pixel space
+ stbtt_aligned_quad *q, // output: quad to draw
+ int align_to_integer);
+
+STBTT_DEF int stbtt_PackFontRangesGatherRects(stbtt_pack_context *spc, const stbtt_fontinfo *info, stbtt_pack_range *ranges, int num_ranges, stbrp_rect *rects);
+STBTT_DEF void stbtt_PackFontRangesPackRects(stbtt_pack_context *spc, stbrp_rect *rects, int num_rects);
+STBTT_DEF int stbtt_PackFontRangesRenderIntoRects(stbtt_pack_context *spc, const stbtt_fontinfo *info, stbtt_pack_range *ranges, int num_ranges, stbrp_rect *rects);
+// Calling these functions in sequence is roughly equivalent to calling
+// stbtt_PackFontRanges(). If you more control over the packing of multiple
+// fonts, or if you want to pack custom data into a font texture, take a look
+// at the source to of stbtt_PackFontRanges() and create a custom version
+// using these functions, e.g. call GatherRects multiple times,
+// building up a single array of rects, then call PackRects once,
+// then call RenderIntoRects repeatedly. This may result in a
+// better packing than calling PackFontRanges multiple times
+// (or it may not).
+
+// this is an opaque structure that you shouldn't mess with which holds
+// all the context needed from PackBegin to PackEnd.
+struct stbtt_pack_context {
+ void *user_allocator_context;
+ void *pack_info;
+ int width;
+ int height;
+ int stride_in_bytes;
+ int padding;
+ int skip_missing;
+ unsigned int h_oversample, v_oversample;
+ unsigned char *pixels;
+ void *nodes;
+};
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// FONT LOADING
+//
+//
+
+STBTT_DEF int stbtt_GetNumberOfFonts(const unsigned char *data);
+// This function will determine the number of fonts in a font file. TrueType
+// collection (.ttc) files may contain multiple fonts, while TrueType font
+// (.ttf) files only contain one font. The number of fonts can be used for
+// indexing with the previous function where the index is between zero and one
+// less than the total fonts. If an error occurs, -1 is returned.
+
+STBTT_DEF int stbtt_GetFontOffsetForIndex(const unsigned char *data, int index);
+// Each .ttf/.ttc file may have more than one font. Each font has a sequential
+// index number starting from 0. Call this function to get the font offset for
+// a given index; it returns -1 if the index is out of range. A regular .ttf
+// file will only define one font and it always be at offset 0, so it will
+// return '0' for index 0, and -1 for all other indices.
+
+// The following structure is defined publicly so you can declare one on
+// the stack or as a global or etc, but you should treat it as opaque.
+struct stbtt_fontinfo
+{
+ void * userdata;
+ unsigned char * data; // pointer to .ttf file
+ int fontstart; // offset of start of font
+
+ int numGlyphs; // number of glyphs, needed for range checking
+
+ int loca,head,glyf,hhea,hmtx,kern,gpos,svg; // table locations as offset from start of .ttf
+ int index_map; // a cmap mapping for our chosen character encoding
+ int indexToLocFormat; // format needed to map from glyph index to glyph
+
+ stbtt__buf cff; // cff font data
+ stbtt__buf charstrings; // the charstring index
+ stbtt__buf gsubrs; // global charstring subroutines index
+ stbtt__buf subrs; // private charstring subroutines index
+ stbtt__buf fontdicts; // array of font dicts
+ stbtt__buf fdselect; // map from glyph to fontdict
+};
+
+STBTT_DEF int stbtt_InitFont(stbtt_fontinfo *info, const unsigned char *data, int offset);
+// Given an offset into the file that defines a font, this function builds
+// the necessary cached info for the rest of the system. You must allocate
+// the stbtt_fontinfo yourself, and stbtt_InitFont will fill it out. You don't
+// need to do anything special to free it, because the contents are pure
+// value data with no additional data structures. Returns 0 on failure.
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// CHARACTER TO GLYPH-INDEX CONVERSIOn
+
+STBTT_DEF int stbtt_FindGlyphIndex(const stbtt_fontinfo *info, int unicode_codepoint);
+// If you're going to perform multiple operations on the same character
+// and you want a speed-up, call this function with the character you're
+// going to process, then use glyph-based functions instead of the
+// codepoint-based functions.
+// Returns 0 if the character codepoint is not defined in the font.
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// CHARACTER PROPERTIES
+//
+
+STBTT_DEF float stbtt_ScaleForPixelHeight(const stbtt_fontinfo *info, float pixels);
+// computes a scale factor to produce a font whose "height" is 'pixels' tall.
+// Height is measured as the distance from the highest ascender to the lowest
+// descender; in other words, it's equivalent to calling stbtt_GetFontVMetrics
+// and computing:
+// scale = pixels / (ascent - descent)
+// so if you prefer to measure height by the ascent only, use a similar calculation.
+
+STBTT_DEF float stbtt_ScaleForMappingEmToPixels(const stbtt_fontinfo *info, float pixels);
+// computes a scale factor to produce a font whose EM size is mapped to
+// 'pixels' tall. This is probably what traditional APIs compute, but
+// I'm not positive.
+
+STBTT_DEF void stbtt_GetFontVMetrics(const stbtt_fontinfo *info, int *ascent, int *descent, int *lineGap);
+// ascent is the coordinate above the baseline the font extends; descent
+// is the coordinate below the baseline the font extends (i.e. it is typically negative)
+// lineGap is the spacing between one row's descent and the next row's ascent...
+// so you should advance the vertical position by "*ascent - *descent + *lineGap"
+// these are expressed in unscaled coordinates, so you must multiply by
+// the scale factor for a given size
+
+STBTT_DEF int stbtt_GetFontVMetricsOS2(const stbtt_fontinfo *info, int *typoAscent, int *typoDescent, int *typoLineGap);
+// analogous to GetFontVMetrics, but returns the "typographic" values from the OS/2
+// table (specific to MS/Windows TTF files).
+//
+// Returns 1 on success (table present), 0 on failure.
+
+STBTT_DEF void stbtt_GetFontBoundingBox(const stbtt_fontinfo *info, int *x0, int *y0, int *x1, int *y1);
+// the bounding box around all possible characters
+
+STBTT_DEF void stbtt_GetCodepointHMetrics(const stbtt_fontinfo *info, int codepoint, int *advanceWidth, int *leftSideBearing);
+// leftSideBearing is the offset from the current horizontal position to the left edge of the character
+// advanceWidth is the offset from the current horizontal position to the next horizontal position
+// these are expressed in unscaled coordinates
+
+STBTT_DEF int stbtt_GetCodepointKernAdvance(const stbtt_fontinfo *info, int ch1, int ch2);
+// an additional amount to add to the 'advance' value between ch1 and ch2
+
+STBTT_DEF int stbtt_GetCodepointBox(const stbtt_fontinfo *info, int codepoint, int *x0, int *y0, int *x1, int *y1);
+// Gets the bounding box of the visible part of the glyph, in unscaled coordinates
+
+STBTT_DEF void stbtt_GetGlyphHMetrics(const stbtt_fontinfo *info, int glyph_index, int *advanceWidth, int *leftSideBearing);
+STBTT_DEF int stbtt_GetGlyphKernAdvance(const stbtt_fontinfo *info, int glyph1, int glyph2);
+STBTT_DEF int stbtt_GetGlyphBox(const stbtt_fontinfo *info, int glyph_index, int *x0, int *y0, int *x1, int *y1);
+// as above, but takes one or more glyph indices for greater efficiency
+
+typedef struct stbtt_kerningentry
+{
+ int glyph1; // use stbtt_FindGlyphIndex
+ int glyph2;
+ int advance;
+} stbtt_kerningentry;
+
+STBTT_DEF int stbtt_GetKerningTableLength(const stbtt_fontinfo *info);
+STBTT_DEF int stbtt_GetKerningTable(const stbtt_fontinfo *info, stbtt_kerningentry* table, int table_length);
+// Retrieves a complete list of all of the kerning pairs provided by the font
+// stbtt_GetKerningTable never writes more than table_length entries and returns how many entries it did write.
+// The table will be sorted by (a.glyph1 == b.glyph1)?(a.glyph2 < b.glyph2):(a.glyph1 < b.glyph1)
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// GLYPH SHAPES (you probably don't need these, but they have to go before
+// the bitmaps for C declaration-order reasons)
+//
#ifndef STBTT_vmove // you can predefine these to use different values (but why?)
- enum {
- STBTT_vmove = 1,
- STBTT_vline,
- STBTT_vcurve,
- STBTT_vcubic
- };
+ enum {
+ STBTT_vmove=1,
+ STBTT_vline,
+ STBTT_vcurve,
+ STBTT_vcubic
+ };
#endif
#ifndef stbtt_vertex // you can predefine this to use different values
- // (we share this with other code at RAD)
-#define stbtt_vertex_type short // can't use stbtt_int16 because that's not visible in the header file
- typedef struct
- {
- stbtt_vertex_type x, y, cx, cy, cx1, cy1;
- unsigned char type, padding;
- } stbtt_vertex;
+ // (we share this with other code at RAD)
+ #define stbtt_vertex_type short // can't use stbtt_int16 because that's not visible in the header file
+ typedef struct
+ {
+ stbtt_vertex_type x,y,cx,cy,cx1,cy1;
+ unsigned char type,padding;
+ } stbtt_vertex;
#endif
- STBTT_DEF int stbtt_IsGlyphEmpty(const stbtt_fontinfo *info, int glyph_index);
- // returns non-zero if nothing is drawn for this glyph
-
- STBTT_DEF int stbtt_GetCodepointShape(const stbtt_fontinfo *info, int unicode_codepoint, stbtt_vertex **vertices);
- STBTT_DEF int stbtt_GetGlyphShape(const stbtt_fontinfo *info, int glyph_index, stbtt_vertex **vertices);
- // returns # of vertices and fills *vertices with the pointer to them
- // these are expressed in "unscaled" coordinates
- //
- // The shape is a series of countours. Each one starts with
- // a STBTT_moveto, then consists of a series of mixed
- // STBTT_lineto and STBTT_curveto segments. A lineto
- // draws a line from previous endpoint to its x,y; a curveto
- // draws a quadratic bezier from previous endpoint to
- // its x,y, using cx,cy as the bezier control point.
-
- STBTT_DEF void stbtt_FreeShape(const stbtt_fontinfo *info, stbtt_vertex *vertices);
- // frees the data allocated above
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // BITMAP RENDERING
- //
-
- STBTT_DEF void stbtt_FreeBitmap(unsigned char *bitmap, void *userdata);
- // frees the bitmap allocated below
-
- STBTT_DEF unsigned char *stbtt_GetCodepointBitmap(const stbtt_fontinfo *info, float scale_x, float scale_y, int codepoint, int *width, int *height, int *xoff, int *yoff);
- // allocates a large-enough single-channel 8bpp bitmap and renders the
- // specified character/glyph at the specified scale into it, with
- // antialiasing. 0 is no coverage (transparent), 255 is fully covered (opaque).
- // *width & *height are filled out with the width & height of the bitmap,
- // which is stored left-to-right, top-to-bottom.
- //
- // xoff/yoff are the offset it pixel space from the glyph origin to the top-left of the bitmap
-
- STBTT_DEF unsigned char *stbtt_GetCodepointBitmapSubpixel(const stbtt_fontinfo *info, float scale_x, float scale_y, float shift_x, float shift_y, int codepoint, int *width, int *height, int *xoff, int *yoff);
- // the same as stbtt_GetCodepoitnBitmap, but you can specify a subpixel
- // shift for the character
-
- STBTT_DEF void stbtt_MakeCodepointBitmap(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, int codepoint);
- // the same as stbtt_GetCodepointBitmap, but you pass in storage for the bitmap
- // in the form of 'output', with row spacing of 'out_stride' bytes. the bitmap
- // is clipped to out_w/out_h bytes. Call stbtt_GetCodepointBitmapBox to get the
- // width and height and positioning info for it first.
-
- STBTT_DEF void stbtt_MakeCodepointBitmapSubpixel(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int codepoint);
- // same as stbtt_MakeCodepointBitmap, but you can specify a subpixel
- // shift for the character
-
- STBTT_DEF void stbtt_MakeCodepointBitmapSubpixelPrefilter(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int oversample_x, int oversample_y, float *sub_x, float *sub_y, int codepoint);
- // same as stbtt_MakeCodepointBitmapSubpixel, but prefiltering
- // is performed (see stbtt_PackSetOversampling)
-
- STBTT_DEF void stbtt_GetCodepointBitmapBox(const stbtt_fontinfo *font, int codepoint, float scale_x, float scale_y, int *ix0, int *iy0, int *ix1, int *iy1);
- // get the bbox of the bitmap centered around the glyph origin; so the
- // bitmap width is ix1-ix0, height is iy1-iy0, and location to place
- // the bitmap top left is (leftSideBearing*scale,iy0).
- // (Note that the bitmap uses y-increases-down, but the shape uses
- // y-increases-up, so CodepointBitmapBox and CodepointBox are inverted.)
-
- STBTT_DEF void stbtt_GetCodepointBitmapBoxSubpixel(const stbtt_fontinfo *font, int codepoint, float scale_x, float scale_y, float shift_x, float shift_y, int *ix0, int *iy0, int *ix1, int *iy1);
- // same as stbtt_GetCodepointBitmapBox, but you can specify a subpixel
- // shift for the character
-
- // the following functions are equivalent to the above functions, but operate
- // on glyph indices instead of Unicode codepoints (for efficiency)
- STBTT_DEF unsigned char *stbtt_GetGlyphBitmap(const stbtt_fontinfo *info, float scale_x, float scale_y, int glyph, int *width, int *height, int *xoff, int *yoff);
- STBTT_DEF unsigned char *stbtt_GetGlyphBitmapSubpixel(const stbtt_fontinfo *info, float scale_x, float scale_y, float shift_x, float shift_y, int glyph, int *width, int *height, int *xoff, int *yoff);
- STBTT_DEF void stbtt_MakeGlyphBitmap(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, int glyph);
- STBTT_DEF void stbtt_MakeGlyphBitmapSubpixel(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int glyph);
- STBTT_DEF void stbtt_MakeGlyphBitmapSubpixelPrefilter(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int oversample_x, int oversample_y, float *sub_x, float *sub_y, int glyph);
- STBTT_DEF void stbtt_GetGlyphBitmapBox(const stbtt_fontinfo *font, int glyph, float scale_x, float scale_y, int *ix0, int *iy0, int *ix1, int *iy1);
- STBTT_DEF void stbtt_GetGlyphBitmapBoxSubpixel(const stbtt_fontinfo *font, int glyph, float scale_x, float scale_y, float shift_x, float shift_y, int *ix0, int *iy0, int *ix1, int *iy1);
-
-
- // @TODO: don't expose this structure
- typedef struct
- {
- int w, h, stride;
- unsigned char *pixels;
- } stbtt__bitmap;
-
- // rasterize a shape with quadratic beziers into a bitmap
- STBTT_DEF void stbtt_Rasterize(stbtt__bitmap *result, // 1-channel bitmap to draw into
- float flatness_in_pixels, // allowable error of curve in pixels
- stbtt_vertex *vertices, // array of vertices defining shape
- int num_verts, // number of vertices in above array
- float scale_x, float scale_y, // scale applied to input vertices
- float shift_x, float shift_y, // translation applied to input vertices
- int x_off, int y_off, // another translation applied to input
- int invert, // if non-zero, vertically flip shape
- void *userdata); // context for to STBTT_MALLOC
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // Signed Distance Function (or Field) rendering
-
- STBTT_DEF void stbtt_FreeSDF(unsigned char *bitmap, void *userdata);
- // frees the SDF bitmap allocated below
-
- STBTT_DEF unsigned char * stbtt_GetGlyphSDF(const stbtt_fontinfo *info, float scale, int glyph, int padding, unsigned char onedge_value, float pixel_dist_scale, int *width, int *height, int *xoff, int *yoff);
- STBTT_DEF unsigned char * stbtt_GetCodepointSDF(const stbtt_fontinfo *info, float scale, int codepoint, int padding, unsigned char onedge_value, float pixel_dist_scale, int *width, int *height, int *xoff, int *yoff);
- // These functions compute a discretized SDF field for a single character, suitable for storing
- // in a single-channel texture, sampling with bilinear filtering, and testing against
- // larger than some threshhold to produce scalable fonts.
- // info -- the font
- // scale -- controls the size of the resulting SDF bitmap, same as it would be creating a regular bitmap
- // glyph/codepoint -- the character to generate the SDF for
- // padding -- extra "pixels" around the character which are filled with the distance to the character (not 0),
- // which allows effects like bit outlines
- // onedge_value -- value 0-255 to test the SDF against to reconstruct the character (i.e. the isocontour of the character)
- // pixel_dist_scale -- what value the SDF should increase by when moving one SDF "pixel" away from the edge (on the 0..255 scale)
- // if positive, > onedge_value is inside; if negative, < onedge_value is inside
- // width,height -- output height & width of the SDF bitmap (including padding)
- // xoff,yoff -- output origin of the character
- // return value -- a 2D array of bytes 0..255, width*height in size
- //
- // pixel_dist_scale & onedge_value are a scale & bias that allows you to make
- // optimal use of the limited 0..255 for your application, trading off precision
- // and special effects. SDF values outside the range 0..255 are clamped to 0..255.
- //
- // Example:
- // scale = stbtt_ScaleForPixelHeight(22)
- // padding = 5
- // onedge_value = 180
- // pixel_dist_scale = 180/5.0 = 36.0
- //
- // This will create an SDF bitmap in which the character is about 22 pixels
- // high but the whole bitmap is about 22+5+5=32 pixels high. To produce a filled
- // shape, sample the SDF at each pixel and fill the pixel if the SDF value
- // is greater than or equal to 180/255. (You'll actually want to antialias,
- // which is beyond the scope of this example.) Additionally, you can compute
- // offset outlines (e.g. to stroke the character border inside & outside,
- // or only outside). For example, to fill outside the character up to 3 SDF
- // pixels, you would compare against (180-36.0*3)/255 = 72/255. The above
- // choice of variables maps a range from 5 pixels outside the shape to
- // 2 pixels inside the shape to 0..255; this is intended primarily for apply
- // outside effects only (the interior range is needed to allow proper
- // antialiasing of the font at *smaller* sizes)
- //
- // The function computes the SDF analytically at each SDF pixel, not by e.g.
- // building a higher-res bitmap and approximating it. In theory the quality
- // should be as high as possible for an SDF of this size & representation, but
- // unclear if this is true in practice (perhaps building a higher-res bitmap
- // and computing from that can allow drop-out prevention).
- //
- // The algorithm has not been optimized at all, so expect it to be slow
- // if computing lots of characters or very large sizes.
-
-
-
- //////////////////////////////////////////////////////////////////////////////
- //
- // Finding the right font...
- //
- // You should really just solve this offline, keep your own tables
- // of what font is what, and don't try to get it out of the .ttf file.
- // That's because getting it out of the .ttf file is really hard, because
- // the names in the file can appear in many possible encodings, in many
- // possible languages, and e.g. if you need a case-insensitive comparison,
- // the details of that depend on the encoding & language in a complex way
- // (actually underspecified in truetype, but also gigantic).
- //
- // But you can use the provided functions in two possible ways:
- // stbtt_FindMatchingFont() will use *case-sensitive* comparisons on
- // unicode-encoded names to try to find the font you want;
- // you can run this before calling stbtt_InitFont()
- //
- // stbtt_GetFontNameString() lets you get any of the various strings
- // from the file yourself and do your own comparisons on them.
- // You have to have called stbtt_InitFont() first.
-
-
- STBTT_DEF int stbtt_FindMatchingFont(const unsigned char *fontdata, const char *name, int flags);
- // returns the offset (not index) of the font that matches, or -1 if none
- // if you use STBTT_MACSTYLE_DONTCARE, use a font name like "Arial Bold".
- // if you use any other flag, use a font name like "Arial"; this checks
- // the 'macStyle' header field; i don't know if fonts set this consistently
+STBTT_DEF int stbtt_IsGlyphEmpty(const stbtt_fontinfo *info, int glyph_index);
+// returns non-zero if nothing is drawn for this glyph
+
+STBTT_DEF int stbtt_GetCodepointShape(const stbtt_fontinfo *info, int unicode_codepoint, stbtt_vertex **vertices);
+STBTT_DEF int stbtt_GetGlyphShape(const stbtt_fontinfo *info, int glyph_index, stbtt_vertex **vertices);
+// returns # of vertices and fills *vertices with the pointer to them
+// these are expressed in "unscaled" coordinates
+//
+// The shape is a series of contours. Each one starts with
+// a STBTT_moveto, then consists of a series of mixed
+// STBTT_lineto and STBTT_curveto segments. A lineto
+// draws a line from previous endpoint to its x,y; a curveto
+// draws a quadratic bezier from previous endpoint to
+// its x,y, using cx,cy as the bezier control point.
+
+STBTT_DEF void stbtt_FreeShape(const stbtt_fontinfo *info, stbtt_vertex *vertices);
+// frees the data allocated above
+
+STBTT_DEF unsigned char *stbtt_FindSVGDoc(const stbtt_fontinfo *info, int gl);
+STBTT_DEF int stbtt_GetCodepointSVG(const stbtt_fontinfo *info, int unicode_codepoint, const char **svg);
+STBTT_DEF int stbtt_GetGlyphSVG(const stbtt_fontinfo *info, int gl, const char **svg);
+// fills svg with the character's SVG data.
+// returns data size or 0 if SVG not found.
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// BITMAP RENDERING
+//
+
+STBTT_DEF void stbtt_FreeBitmap(unsigned char *bitmap, void *userdata);
+// frees the bitmap allocated below
+
+STBTT_DEF unsigned char *stbtt_GetCodepointBitmap(const stbtt_fontinfo *info, float scale_x, float scale_y, int codepoint, int *width, int *height, int *xoff, int *yoff);
+// allocates a large-enough single-channel 8bpp bitmap and renders the
+// specified character/glyph at the specified scale into it, with
+// antialiasing. 0 is no coverage (transparent), 255 is fully covered (opaque).
+// *width & *height are filled out with the width & height of the bitmap,
+// which is stored left-to-right, top-to-bottom.
+//
+// xoff/yoff are the offset it pixel space from the glyph origin to the top-left of the bitmap
+
+STBTT_DEF unsigned char *stbtt_GetCodepointBitmapSubpixel(const stbtt_fontinfo *info, float scale_x, float scale_y, float shift_x, float shift_y, int codepoint, int *width, int *height, int *xoff, int *yoff);
+// the same as stbtt_GetCodepoitnBitmap, but you can specify a subpixel
+// shift for the character
+
+STBTT_DEF void stbtt_MakeCodepointBitmap(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, int codepoint);
+// the same as stbtt_GetCodepointBitmap, but you pass in storage for the bitmap
+// in the form of 'output', with row spacing of 'out_stride' bytes. the bitmap
+// is clipped to out_w/out_h bytes. Call stbtt_GetCodepointBitmapBox to get the
+// width and height and positioning info for it first.
+
+STBTT_DEF void stbtt_MakeCodepointBitmapSubpixel(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int codepoint);
+// same as stbtt_MakeCodepointBitmap, but you can specify a subpixel
+// shift for the character
+
+STBTT_DEF void stbtt_MakeCodepointBitmapSubpixelPrefilter(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int oversample_x, int oversample_y, float *sub_x, float *sub_y, int codepoint);
+// same as stbtt_MakeCodepointBitmapSubpixel, but prefiltering
+// is performed (see stbtt_PackSetOversampling)
+
+STBTT_DEF void stbtt_GetCodepointBitmapBox(const stbtt_fontinfo *font, int codepoint, float scale_x, float scale_y, int *ix0, int *iy0, int *ix1, int *iy1);
+// get the bbox of the bitmap centered around the glyph origin; so the
+// bitmap width is ix1-ix0, height is iy1-iy0, and location to place
+// the bitmap top left is (leftSideBearing*scale,iy0).
+// (Note that the bitmap uses y-increases-down, but the shape uses
+// y-increases-up, so CodepointBitmapBox and CodepointBox are inverted.)
+
+STBTT_DEF void stbtt_GetCodepointBitmapBoxSubpixel(const stbtt_fontinfo *font, int codepoint, float scale_x, float scale_y, float shift_x, float shift_y, int *ix0, int *iy0, int *ix1, int *iy1);
+// same as stbtt_GetCodepointBitmapBox, but you can specify a subpixel
+// shift for the character
+
+// the following functions are equivalent to the above functions, but operate
+// on glyph indices instead of Unicode codepoints (for efficiency)
+STBTT_DEF unsigned char *stbtt_GetGlyphBitmap(const stbtt_fontinfo *info, float scale_x, float scale_y, int glyph, int *width, int *height, int *xoff, int *yoff);
+STBTT_DEF unsigned char *stbtt_GetGlyphBitmapSubpixel(const stbtt_fontinfo *info, float scale_x, float scale_y, float shift_x, float shift_y, int glyph, int *width, int *height, int *xoff, int *yoff);
+STBTT_DEF void stbtt_MakeGlyphBitmap(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, int glyph);
+STBTT_DEF void stbtt_MakeGlyphBitmapSubpixel(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int glyph);
+STBTT_DEF void stbtt_MakeGlyphBitmapSubpixelPrefilter(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int oversample_x, int oversample_y, float *sub_x, float *sub_y, int glyph);
+STBTT_DEF void stbtt_GetGlyphBitmapBox(const stbtt_fontinfo *font, int glyph, float scale_x, float scale_y, int *ix0, int *iy0, int *ix1, int *iy1);
+STBTT_DEF void stbtt_GetGlyphBitmapBoxSubpixel(const stbtt_fontinfo *font, int glyph, float scale_x, float scale_y,float shift_x, float shift_y, int *ix0, int *iy0, int *ix1, int *iy1);
+
+
+// @TODO: don't expose this structure
+typedef struct
+{
+ int w,h,stride;
+ unsigned char *pixels;
+} stbtt__bitmap;
+
+// rasterize a shape with quadratic beziers into a bitmap
+STBTT_DEF void stbtt_Rasterize(stbtt__bitmap *result, // 1-channel bitmap to draw into
+ float flatness_in_pixels, // allowable error of curve in pixels
+ stbtt_vertex *vertices, // array of vertices defining shape
+ int num_verts, // number of vertices in above array
+ float scale_x, float scale_y, // scale applied to input vertices
+ float shift_x, float shift_y, // translation applied to input vertices
+ int x_off, int y_off, // another translation applied to input
+ int invert, // if non-zero, vertically flip shape
+ void *userdata); // context for to STBTT_MALLOC
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// Signed Distance Function (or Field) rendering
+
+STBTT_DEF void stbtt_FreeSDF(unsigned char *bitmap, void *userdata);
+// frees the SDF bitmap allocated below
+
+STBTT_DEF unsigned char * stbtt_GetGlyphSDF(const stbtt_fontinfo *info, float scale, int glyph, int padding, unsigned char onedge_value, float pixel_dist_scale, int *width, int *height, int *xoff, int *yoff);
+STBTT_DEF unsigned char * stbtt_GetCodepointSDF(const stbtt_fontinfo *info, float scale, int codepoint, int padding, unsigned char onedge_value, float pixel_dist_scale, int *width, int *height, int *xoff, int *yoff);
+// These functions compute a discretized SDF field for a single character, suitable for storing
+// in a single-channel texture, sampling with bilinear filtering, and testing against
+// larger than some threshold to produce scalable fonts.
+// info -- the font
+// scale -- controls the size of the resulting SDF bitmap, same as it would be creating a regular bitmap
+// glyph/codepoint -- the character to generate the SDF for
+// padding -- extra "pixels" around the character which are filled with the distance to the character (not 0),
+// which allows effects like bit outlines
+// onedge_value -- value 0-255 to test the SDF against to reconstruct the character (i.e. the isocontour of the character)
+// pixel_dist_scale -- what value the SDF should increase by when moving one SDF "pixel" away from the edge (on the 0..255 scale)
+// if positive, > onedge_value is inside; if negative, < onedge_value is inside
+// width,height -- output height & width of the SDF bitmap (including padding)
+// xoff,yoff -- output origin of the character
+// return value -- a 2D array of bytes 0..255, width*height in size
+//
+// pixel_dist_scale & onedge_value are a scale & bias that allows you to make
+// optimal use of the limited 0..255 for your application, trading off precision
+// and special effects. SDF values outside the range 0..255 are clamped to 0..255.
+//
+// Example:
+// scale = stbtt_ScaleForPixelHeight(22)
+// padding = 5
+// onedge_value = 180
+// pixel_dist_scale = 180/5.0 = 36.0
+//
+// This will create an SDF bitmap in which the character is about 22 pixels
+// high but the whole bitmap is about 22+5+5=32 pixels high. To produce a filled
+// shape, sample the SDF at each pixel and fill the pixel if the SDF value
+// is greater than or equal to 180/255. (You'll actually want to antialias,
+// which is beyond the scope of this example.) Additionally, you can compute
+// offset outlines (e.g. to stroke the character border inside & outside,
+// or only outside). For example, to fill outside the character up to 3 SDF
+// pixels, you would compare against (180-36.0*3)/255 = 72/255. The above
+// choice of variables maps a range from 5 pixels outside the shape to
+// 2 pixels inside the shape to 0..255; this is intended primarily for apply
+// outside effects only (the interior range is needed to allow proper
+// antialiasing of the font at *smaller* sizes)
+//
+// The function computes the SDF analytically at each SDF pixel, not by e.g.
+// building a higher-res bitmap and approximating it. In theory the quality
+// should be as high as possible for an SDF of this size & representation, but
+// unclear if this is true in practice (perhaps building a higher-res bitmap
+// and computing from that can allow drop-out prevention).
+//
+// The algorithm has not been optimized at all, so expect it to be slow
+// if computing lots of characters or very large sizes.
+
+
+
+//////////////////////////////////////////////////////////////////////////////
+//
+// Finding the right font...
+//
+// You should really just solve this offline, keep your own tables
+// of what font is what, and don't try to get it out of the .ttf file.
+// That's because getting it out of the .ttf file is really hard, because
+// the names in the file can appear in many possible encodings, in many
+// possible languages, and e.g. if you need a case-insensitive comparison,
+// the details of that depend on the encoding & language in a complex way
+// (actually underspecified in truetype, but also gigantic).
+//
+// But you can use the provided functions in two possible ways:
+// stbtt_FindMatchingFont() will use *case-sensitive* comparisons on
+// unicode-encoded names to try to find the font you want;
+// you can run this before calling stbtt_InitFont()
+//
+// stbtt_GetFontNameString() lets you get any of the various strings
+// from the file yourself and do your own comparisons on them.
+// You have to have called stbtt_InitFont() first.
+
+
+STBTT_DEF int stbtt_FindMatchingFont(const unsigned char *fontdata, const char *name, int flags);
+// returns the offset (not index) of the font that matches, or -1 if none
+// if you use STBTT_MACSTYLE_DONTCARE, use a font name like "Arial Bold".
+// if you use any other flag, use a font name like "Arial"; this checks
+// the 'macStyle' header field; i don't know if fonts set this consistently
#define STBTT_MACSTYLE_DONTCARE 0
#define STBTT_MACSTYLE_BOLD 1
#define STBTT_MACSTYLE_ITALIC 2
#define STBTT_MACSTYLE_UNDERSCORE 4
#define STBTT_MACSTYLE_NONE 8 // <= not same as 0, this makes us check the bitfield is 0
- STBTT_DEF int stbtt_CompareUTF8toUTF16_bigendian(const char *s1, int len1, const char *s2, int len2);
- // returns 1/0 whether the first string interpreted as utf8 is identical to
- // the second string interpreted as big-endian utf16... useful for strings from next func
-
- STBTT_DEF const char *stbtt_GetFontNameString(const stbtt_fontinfo *font, int *length, int platformID, int encodingID, int languageID, int nameID);
- // returns the string (which may be big-endian double byte, e.g. for unicode)
- // and puts the length in bytes in *length.
- //
- // some of the values for the IDs are below; for more see the truetype spec:
- // http://developer.apple.com/textfonts/TTRefMan/RM06/Chap6name.html
- // http://www.microsoft.com/typography/otspec/name.htm
-
- enum { // platformID
- STBTT_PLATFORM_ID_UNICODE = 0,
- STBTT_PLATFORM_ID_MAC = 1,
- STBTT_PLATFORM_ID_ISO = 2,
- STBTT_PLATFORM_ID_MICROSOFT = 3
- };
-
- enum { // encodingID for STBTT_PLATFORM_ID_UNICODE
- STBTT_UNICODE_EID_UNICODE_1_0 = 0,
- STBTT_UNICODE_EID_UNICODE_1_1 = 1,
- STBTT_UNICODE_EID_ISO_10646 = 2,
- STBTT_UNICODE_EID_UNICODE_2_0_BMP = 3,
- STBTT_UNICODE_EID_UNICODE_2_0_FULL = 4
- };
-
- enum { // encodingID for STBTT_PLATFORM_ID_MICROSOFT
- STBTT_MS_EID_SYMBOL = 0,
- STBTT_MS_EID_UNICODE_BMP = 1,
- STBTT_MS_EID_SHIFTJIS = 2,
- STBTT_MS_EID_UNICODE_FULL = 10
- };
-
- enum { // encodingID for STBTT_PLATFORM_ID_MAC; same as Script Manager codes
- STBTT_MAC_EID_ROMAN = 0, STBTT_MAC_EID_ARABIC = 4,
- STBTT_MAC_EID_JAPANESE = 1, STBTT_MAC_EID_HEBREW = 5,
- STBTT_MAC_EID_CHINESE_TRAD = 2, STBTT_MAC_EID_GREEK = 6,
- STBTT_MAC_EID_KOREAN = 3, STBTT_MAC_EID_RUSSIAN = 7
- };
-
- enum { // languageID for STBTT_PLATFORM_ID_MICROSOFT; same as LCID...
- // problematic because there are e.g. 16 english LCIDs and 16 arabic LCIDs
- STBTT_MS_LANG_ENGLISH = 0x0409, STBTT_MS_LANG_ITALIAN = 0x0410,
- STBTT_MS_LANG_CHINESE = 0x0804, STBTT_MS_LANG_JAPANESE = 0x0411,
- STBTT_MS_LANG_DUTCH = 0x0413, STBTT_MS_LANG_KOREAN = 0x0412,
- STBTT_MS_LANG_FRENCH = 0x040c, STBTT_MS_LANG_RUSSIAN = 0x0419,
- STBTT_MS_LANG_GERMAN = 0x0407, STBTT_MS_LANG_SPANISH = 0x0409,
- STBTT_MS_LANG_HEBREW = 0x040d, STBTT_MS_LANG_SWEDISH = 0x041D
- };
-
- enum { // languageID for STBTT_PLATFORM_ID_MAC
- STBTT_MAC_LANG_ENGLISH = 0, STBTT_MAC_LANG_JAPANESE = 11,
- STBTT_MAC_LANG_ARABIC = 12, STBTT_MAC_LANG_KOREAN = 23,
- STBTT_MAC_LANG_DUTCH = 4, STBTT_MAC_LANG_RUSSIAN = 32,
- STBTT_MAC_LANG_FRENCH = 1, STBTT_MAC_LANG_SPANISH = 6,
- STBTT_MAC_LANG_GERMAN = 2, STBTT_MAC_LANG_SWEDISH = 5,
- STBTT_MAC_LANG_HEBREW = 10, STBTT_MAC_LANG_CHINESE_SIMPLIFIED = 33,
- STBTT_MAC_LANG_ITALIAN = 3, STBTT_MAC_LANG_CHINESE_TRAD = 19
- };
+STBTT_DEF int stbtt_CompareUTF8toUTF16_bigendian(const char *s1, int len1, const char *s2, int len2);
+// returns 1/0 whether the first string interpreted as utf8 is identical to
+// the second string interpreted as big-endian utf16... useful for strings from next func
+
+STBTT_DEF const char *stbtt_GetFontNameString(const stbtt_fontinfo *font, int *length, int platformID, int encodingID, int languageID, int nameID);
+// returns the string (which may be big-endian double byte, e.g. for unicode)
+// and puts the length in bytes in *length.
+//
+// some of the values for the IDs are below; for more see the truetype spec:
+// http://developer.apple.com/textfonts/TTRefMan/RM06/Chap6name.html
+// http://www.microsoft.com/typography/otspec/name.htm
+
+enum { // platformID
+ STBTT_PLATFORM_ID_UNICODE =0,
+ STBTT_PLATFORM_ID_MAC =1,
+ STBTT_PLATFORM_ID_ISO =2,
+ STBTT_PLATFORM_ID_MICROSOFT =3
+};
+
+enum { // encodingID for STBTT_PLATFORM_ID_UNICODE
+ STBTT_UNICODE_EID_UNICODE_1_0 =0,
+ STBTT_UNICODE_EID_UNICODE_1_1 =1,
+ STBTT_UNICODE_EID_ISO_10646 =2,
+ STBTT_UNICODE_EID_UNICODE_2_0_BMP=3,
+ STBTT_UNICODE_EID_UNICODE_2_0_FULL=4
+};
+
+enum { // encodingID for STBTT_PLATFORM_ID_MICROSOFT
+ STBTT_MS_EID_SYMBOL =0,
+ STBTT_MS_EID_UNICODE_BMP =1,
+ STBTT_MS_EID_SHIFTJIS =2,
+ STBTT_MS_EID_UNICODE_FULL =10
+};
+
+enum { // encodingID for STBTT_PLATFORM_ID_MAC; same as Script Manager codes
+ STBTT_MAC_EID_ROMAN =0, STBTT_MAC_EID_ARABIC =4,
+ STBTT_MAC_EID_JAPANESE =1, STBTT_MAC_EID_HEBREW =5,
+ STBTT_MAC_EID_CHINESE_TRAD =2, STBTT_MAC_EID_GREEK =6,
+ STBTT_MAC_EID_KOREAN =3, STBTT_MAC_EID_RUSSIAN =7
+};
+
+enum { // languageID for STBTT_PLATFORM_ID_MICROSOFT; same as LCID...
+ // problematic because there are e.g. 16 english LCIDs and 16 arabic LCIDs
+ STBTT_MS_LANG_ENGLISH =0x0409, STBTT_MS_LANG_ITALIAN =0x0410,
+ STBTT_MS_LANG_CHINESE =0x0804, STBTT_MS_LANG_JAPANESE =0x0411,
+ STBTT_MS_LANG_DUTCH =0x0413, STBTT_MS_LANG_KOREAN =0x0412,
+ STBTT_MS_LANG_FRENCH =0x040c, STBTT_MS_LANG_RUSSIAN =0x0419,
+ STBTT_MS_LANG_GERMAN =0x0407, STBTT_MS_LANG_SPANISH =0x0409,
+ STBTT_MS_LANG_HEBREW =0x040d, STBTT_MS_LANG_SWEDISH =0x041D
+};
+
+enum { // languageID for STBTT_PLATFORM_ID_MAC
+ STBTT_MAC_LANG_ENGLISH =0 , STBTT_MAC_LANG_JAPANESE =11,
+ STBTT_MAC_LANG_ARABIC =12, STBTT_MAC_LANG_KOREAN =23,
+ STBTT_MAC_LANG_DUTCH =4 , STBTT_MAC_LANG_RUSSIAN =32,
+ STBTT_MAC_LANG_FRENCH =1 , STBTT_MAC_LANG_SPANISH =6 ,
+ STBTT_MAC_LANG_GERMAN =2 , STBTT_MAC_LANG_SWEDISH =5 ,
+ STBTT_MAC_LANG_HEBREW =10, STBTT_MAC_LANG_CHINESE_SIMPLIFIED =33,
+ STBTT_MAC_LANG_ITALIAN =3 , STBTT_MAC_LANG_CHINESE_TRAD =19
+};
#ifdef __cplusplus
}
@@ -1080,7 +1113,7 @@ extern "C" {
#error "STBTT_MAX_OVERSAMPLE cannot be > 255"
#endif
-typedef int stbtt__test_oversample_pow2[(STBTT_MAX_OVERSAMPLE & (STBTT_MAX_OVERSAMPLE - 1)) == 0 ? 1 : -1];
+typedef int stbtt__test_oversample_pow2[(STBTT_MAX_OVERSAMPLE & (STBTT_MAX_OVERSAMPLE-1)) == 0 ? 1 : -1];
#ifndef STBTT_RASTERIZER_VERSION
#define STBTT_RASTERIZER_VERSION 2
@@ -1099,47 +1132,47 @@ typedef int stbtt__test_oversample_pow2[(STBTT_MAX_OVERSAMPLE & (STBTT_MAX_OVERS
static stbtt_uint8 stbtt__buf_get8(stbtt__buf *b)
{
- if (b->cursor >= b->size)
- return 0;
- return b->data[b->cursor++];
+ if (b->cursor >= b->size)
+ return 0;
+ return b->data[b->cursor++];
}
static stbtt_uint8 stbtt__buf_peek8(stbtt__buf *b)
{
- if (b->cursor >= b->size)
- return 0;
- return b->data[b->cursor];
+ if (b->cursor >= b->size)
+ return 0;
+ return b->data[b->cursor];
}
static void stbtt__buf_seek(stbtt__buf *b, int o)
{
- STBTT_assert(!(o > b->size || o < 0));
- b->cursor = (o > b->size || o < 0) ? b->size : o;
+ STBTT_assert(!(o > b->size || o < 0));
+ b->cursor = (o > b->size || o < 0) ? b->size : o;
}
static void stbtt__buf_skip(stbtt__buf *b, int o)
{
- stbtt__buf_seek(b, b->cursor + o);
+ stbtt__buf_seek(b, b->cursor + o);
}
static stbtt_uint32 stbtt__buf_get(stbtt__buf *b, int n)
{
- stbtt_uint32 v = 0;
- int i;
- STBTT_assert(n >= 1 && n <= 4);
- for (i = 0; i < n; i++)
- v = (v << 8) | stbtt__buf_get8(b);
- return v;
+ stbtt_uint32 v = 0;
+ int i;
+ STBTT_assert(n >= 1 && n <= 4);
+ for (i = 0; i < n; i++)
+ v = (v << 8) | stbtt__buf_get8(b);
+ return v;
}
static stbtt__buf stbtt__new_buf(const void *p, size_t size)
{
- stbtt__buf r;
- STBTT_assert(size < 0x40000000);
- r.data = (stbtt_uint8*)p;
- r.size = (int)size;
- r.cursor = 0;
- return r;
+ stbtt__buf r;
+ STBTT_assert(size < 0x40000000);
+ r.data = (stbtt_uint8*) p;
+ r.size = (int) size;
+ r.cursor = 0;
+ return r;
}
#define stbtt__buf_get16(b) stbtt__buf_get((b), 2)
@@ -1147,96 +1180,95 @@ static stbtt__buf stbtt__new_buf(const void *p, size_t size)
static stbtt__buf stbtt__buf_range(const stbtt__buf *b, int o, int s)
{
- stbtt__buf r = stbtt__new_buf(NULL, 0);
- if (o < 0 || s < 0 || o > b->size || s > b->size - o) return r;
- r.data = b->data + o;
- r.size = s;
- return r;
+ stbtt__buf r = stbtt__new_buf(NULL, 0);
+ if (o < 0 || s < 0 || o > b->size || s > b->size - o) return r;
+ r.data = b->data + o;
+ r.size = s;
+ return r;
}
static stbtt__buf stbtt__cff_get_index(stbtt__buf *b)
{
- int count, start, offsize;
- start = b->cursor;
- count = stbtt__buf_get16(b);
- if (count) {
- offsize = stbtt__buf_get8(b);
- STBTT_assert(offsize >= 1 && offsize <= 4);
- stbtt__buf_skip(b, offsize * count);
- stbtt__buf_skip(b, stbtt__buf_get(b, offsize) - 1);
- }
- return stbtt__buf_range(b, start, b->cursor - start);
+ int count, start, offsize;
+ start = b->cursor;
+ count = stbtt__buf_get16(b);
+ if (count) {
+ offsize = stbtt__buf_get8(b);
+ STBTT_assert(offsize >= 1 && offsize <= 4);
+ stbtt__buf_skip(b, offsize * count);
+ stbtt__buf_skip(b, stbtt__buf_get(b, offsize) - 1);
+ }
+ return stbtt__buf_range(b, start, b->cursor - start);
}
static stbtt_uint32 stbtt__cff_int(stbtt__buf *b)
{
- int b0 = stbtt__buf_get8(b);
- if (b0 >= 32 && b0 <= 246) return b0 - 139;
- else if (b0 >= 247 && b0 <= 250) return (b0 - 247) * 256 + stbtt__buf_get8(b) + 108;
- else if (b0 >= 251 && b0 <= 254) return -(b0 - 251) * 256 - stbtt__buf_get8(b) - 108;
- else if (b0 == 28) return stbtt__buf_get16(b);
- else if (b0 == 29) return stbtt__buf_get32(b);
- STBTT_assert(0);
- return 0;
+ int b0 = stbtt__buf_get8(b);
+ if (b0 >= 32 && b0 <= 246) return b0 - 139;
+ else if (b0 >= 247 && b0 <= 250) return (b0 - 247)*256 + stbtt__buf_get8(b) + 108;
+ else if (b0 >= 251 && b0 <= 254) return -(b0 - 251)*256 - stbtt__buf_get8(b) - 108;
+ else if (b0 == 28) return stbtt__buf_get16(b);
+ else if (b0 == 29) return stbtt__buf_get32(b);
+ STBTT_assert(0);
+ return 0;
}
static void stbtt__cff_skip_operand(stbtt__buf *b) {
- int v, b0 = stbtt__buf_peek8(b);
- STBTT_assert(b0 >= 28);
- if (b0 == 30) {
- stbtt__buf_skip(b, 1);
- while (b->cursor < b->size) {
- v = stbtt__buf_get8(b);
- if ((v & 0xF) == 0xF || (v >> 4) == 0xF)
- break;
- }
- }
- else {
- stbtt__cff_int(b);
- }
+ int v, b0 = stbtt__buf_peek8(b);
+ STBTT_assert(b0 >= 28);
+ if (b0 == 30) {
+ stbtt__buf_skip(b, 1);
+ while (b->cursor < b->size) {
+ v = stbtt__buf_get8(b);
+ if ((v & 0xF) == 0xF || (v >> 4) == 0xF)
+ break;
+ }
+ } else {
+ stbtt__cff_int(b);
+ }
}
static stbtt__buf stbtt__dict_get(stbtt__buf *b, int key)
{
- stbtt__buf_seek(b, 0);
- while (b->cursor < b->size) {
- int start = b->cursor, end, op;
- while (stbtt__buf_peek8(b) >= 28)
- stbtt__cff_skip_operand(b);
- end = b->cursor;
- op = stbtt__buf_get8(b);
- if (op == 12) op = stbtt__buf_get8(b) | 0x100;
- if (op == key) return stbtt__buf_range(b, start, end - start);
- }
- return stbtt__buf_range(b, 0, 0);
+ stbtt__buf_seek(b, 0);
+ while (b->cursor < b->size) {
+ int start = b->cursor, end, op;
+ while (stbtt__buf_peek8(b) >= 28)
+ stbtt__cff_skip_operand(b);
+ end = b->cursor;
+ op = stbtt__buf_get8(b);
+ if (op == 12) op = stbtt__buf_get8(b) | 0x100;
+ if (op == key) return stbtt__buf_range(b, start, end-start);
+ }
+ return stbtt__buf_range(b, 0, 0);
}
static void stbtt__dict_get_ints(stbtt__buf *b, int key, int outcount, stbtt_uint32 *out)
{
- int i;
- stbtt__buf operands = stbtt__dict_get(b, key);
- for (i = 0; i < outcount && operands.cursor < operands.size; i++)
- out[i] = stbtt__cff_int(&operands);
+ int i;
+ stbtt__buf operands = stbtt__dict_get(b, key);
+ for (i = 0; i < outcount && operands.cursor < operands.size; i++)
+ out[i] = stbtt__cff_int(&operands);
}
static int stbtt__cff_index_count(stbtt__buf *b)
{
- stbtt__buf_seek(b, 0);
- return stbtt__buf_get16(b);
+ stbtt__buf_seek(b, 0);
+ return stbtt__buf_get16(b);
}
static stbtt__buf stbtt__cff_index_get(stbtt__buf b, int i)
{
- int count, offsize, start, end;
- stbtt__buf_seek(&b, 0);
- count = stbtt__buf_get16(&b);
- offsize = stbtt__buf_get8(&b);
- STBTT_assert(i >= 0 && i < count);
- STBTT_assert(offsize >= 1 && offsize <= 4);
- stbtt__buf_skip(&b, i*offsize);
- start = stbtt__buf_get(&b, offsize);
- end = stbtt__buf_get(&b, offsize);
- return stbtt__buf_range(&b, 2 + (count + 1)*offsize + start, end - start);
+ int count, offsize, start, end;
+ stbtt__buf_seek(&b, 0);
+ count = stbtt__buf_get16(&b);
+ offsize = stbtt__buf_get8(&b);
+ STBTT_assert(i >= 0 && i < count);
+ STBTT_assert(offsize >= 1 && offsize <= 4);
+ stbtt__buf_skip(&b, i*offsize);
+ start = stbtt__buf_get(&b, offsize);
+ end = stbtt__buf_get(&b, offsize);
+ return stbtt__buf_range(&b, 2+(count+1)*offsize+start, end - start);
}
//////////////////////////////////////////////////////////////////////////
@@ -1251,1381 +1283,1434 @@ static stbtt__buf stbtt__cff_index_get(stbtt__buf b, int i)
#define ttCHAR(p) (* (stbtt_int8 *) (p))
#define ttFixed(p) ttLONG(p)
-static stbtt_uint16 ttUSHORT(stbtt_uint8 *p) { return p[0] * 256 + p[1]; }
-static stbtt_int16 ttSHORT(stbtt_uint8 *p) { return p[0] * 256 + p[1]; }
-static stbtt_uint32 ttULONG(stbtt_uint8 *p) { return (p[0] << 24) + (p[1] << 16) + (p[2] << 8) + p[3]; }
-static stbtt_int32 ttLONG(stbtt_uint8 *p) { return (p[0] << 24) + (p[1] << 16) + (p[2] << 8) + p[3]; }
+static stbtt_uint16 ttUSHORT(stbtt_uint8 *p) { return p[0]*256 + p[1]; }
+static stbtt_int16 ttSHORT(stbtt_uint8 *p) { return p[0]*256 + p[1]; }
+static stbtt_uint32 ttULONG(stbtt_uint8 *p) { return (p[0]<<24) + (p[1]<<16) + (p[2]<<8) + p[3]; }
+static stbtt_int32 ttLONG(stbtt_uint8 *p) { return (p[0]<<24) + (p[1]<<16) + (p[2]<<8) + p[3]; }
#define stbtt_tag4(p,c0,c1,c2,c3) ((p)[0] == (c0) && (p)[1] == (c1) && (p)[2] == (c2) && (p)[3] == (c3))
#define stbtt_tag(p,str) stbtt_tag4(p,str[0],str[1],str[2],str[3])
static int stbtt__isfont(stbtt_uint8 *font)
{
- // check the version number
- if (stbtt_tag4(font, '1', 0, 0, 0)) return 1; // TrueType 1
- if (stbtt_tag(font, "typ1")) return 1; // TrueType with type 1 font -- we don't support this!
- if (stbtt_tag(font, "OTTO")) return 1; // OpenType with CFF
- if (stbtt_tag4(font, 0, 1, 0, 0)) return 1; // OpenType 1.0
- if (stbtt_tag(font, "true")) return 1; // Apple specification for TrueType fonts
- return 0;
+ // check the version number
+ if (stbtt_tag4(font, '1',0,0,0)) return 1; // TrueType 1
+ if (stbtt_tag(font, "typ1")) return 1; // TrueType with type 1 font -- we don't support this!
+ if (stbtt_tag(font, "OTTO")) return 1; // OpenType with CFF
+ if (stbtt_tag4(font, 0,1,0,0)) return 1; // OpenType 1.0
+ if (stbtt_tag(font, "true")) return 1; // Apple specification for TrueType fonts
+ return 0;
}
// @OPTIMIZE: binary search
static stbtt_uint32 stbtt__find_table(stbtt_uint8 *data, stbtt_uint32 fontstart, const char *tag)
{
- stbtt_int32 num_tables = ttUSHORT(data + fontstart + 4);
- stbtt_uint32 tabledir = fontstart + 12;
- stbtt_int32 i;
- for (i = 0; i < num_tables; ++i) {
- stbtt_uint32 loc = tabledir + 16 * i;
- if (stbtt_tag(data + loc + 0, tag))
- return ttULONG(data + loc + 8);
- }
- return 0;
+ stbtt_int32 num_tables = ttUSHORT(data+fontstart+4);
+ stbtt_uint32 tabledir = fontstart + 12;
+ stbtt_int32 i;
+ for (i=0; i < num_tables; ++i) {
+ stbtt_uint32 loc = tabledir + 16*i;
+ if (stbtt_tag(data+loc+0, tag))
+ return ttULONG(data+loc+8);
+ }
+ return 0;
}
static int stbtt_GetFontOffsetForIndex_internal(unsigned char *font_collection, int index)
{
- // if it's just a font, there's only one valid index
- if (stbtt__isfont(font_collection))
- return index == 0 ? 0 : -1;
+ // if it's just a font, there's only one valid index
+ if (stbtt__isfont(font_collection))
+ return index == 0 ? 0 : -1;
- // check if it's a TTC
- if (stbtt_tag(font_collection, "ttcf")) {
- // version 1?
- if (ttULONG(font_collection + 4) == 0x00010000 || ttULONG(font_collection + 4) == 0x00020000) {
- stbtt_int32 n = ttLONG(font_collection + 8);
- if (index >= n)
- return -1;
- return ttULONG(font_collection + 12 + index * 4);
- }
- }
- return -1;
+ // check if it's a TTC
+ if (stbtt_tag(font_collection, "ttcf")) {
+ // version 1?
+ if (ttULONG(font_collection+4) == 0x00010000 || ttULONG(font_collection+4) == 0x00020000) {
+ stbtt_int32 n = ttLONG(font_collection+8);
+ if (index >= n)
+ return -1;
+ return ttULONG(font_collection+12+index*4);
+ }
+ }
+ return -1;
}
static int stbtt_GetNumberOfFonts_internal(unsigned char *font_collection)
{
- // if it's just a font, there's only one valid font
- if (stbtt__isfont(font_collection))
- return 1;
+ // if it's just a font, there's only one valid font
+ if (stbtt__isfont(font_collection))
+ return 1;
- // check if it's a TTC
- if (stbtt_tag(font_collection, "ttcf")) {
- // version 1?
- if (ttULONG(font_collection + 4) == 0x00010000 || ttULONG(font_collection + 4) == 0x00020000) {
- return ttLONG(font_collection + 8);
- }
- }
- return 0;
+ // check if it's a TTC
+ if (stbtt_tag(font_collection, "ttcf")) {
+ // version 1?
+ if (ttULONG(font_collection+4) == 0x00010000 || ttULONG(font_collection+4) == 0x00020000) {
+ return ttLONG(font_collection+8);
+ }
+ }
+ return 0;
}
static stbtt__buf stbtt__get_subrs(stbtt__buf cff, stbtt__buf fontdict)
{
- stbtt_uint32 subrsoff = 0, private_loc[2] = { 0, 0 };
- stbtt__buf pdict;
- stbtt__dict_get_ints(&fontdict, 18, 2, private_loc);
- if (!private_loc[1] || !private_loc[0]) return stbtt__new_buf(NULL, 0);
- pdict = stbtt__buf_range(&cff, private_loc[1], private_loc[0]);
- stbtt__dict_get_ints(&pdict, 19, 1, &subrsoff);
- if (!subrsoff) return stbtt__new_buf(NULL, 0);
- stbtt__buf_seek(&cff, private_loc[1] + subrsoff);
- return stbtt__cff_get_index(&cff);
+ stbtt_uint32 subrsoff = 0, private_loc[2] = { 0, 0 };
+ stbtt__buf pdict;
+ stbtt__dict_get_ints(&fontdict, 18, 2, private_loc);
+ if (!private_loc[1] || !private_loc[0]) return stbtt__new_buf(NULL, 0);
+ pdict = stbtt__buf_range(&cff, private_loc[1], private_loc[0]);
+ stbtt__dict_get_ints(&pdict, 19, 1, &subrsoff);
+ if (!subrsoff) return stbtt__new_buf(NULL, 0);
+ stbtt__buf_seek(&cff, private_loc[1]+subrsoff);
+ return stbtt__cff_get_index(&cff);
+}
+
+// since most people won't use this, find this table the first time it's needed
+static int stbtt__get_svg(stbtt_fontinfo *info)
+{
+ stbtt_uint32 t;
+ if (info->svg < 0) {
+ t = stbtt__find_table(info->data, info->fontstart, "SVG ");
+ if (t) {
+ stbtt_uint32 offset = ttULONG(info->data + t + 2);
+ info->svg = t + offset;
+ } else {
+ info->svg = 0;
+ }
+ }
+ return info->svg;
}
static int stbtt_InitFont_internal(stbtt_fontinfo *info, unsigned char *data, int fontstart)
{
- stbtt_uint32 cmap, t;
- stbtt_int32 i, numTables;
-
- info->data = data;
- info->fontstart = fontstart;
- info->cff = stbtt__new_buf(NULL, 0);
-
- cmap = stbtt__find_table(data, fontstart, "cmap"); // required
- info->loca = stbtt__find_table(data, fontstart, "loca"); // required
- info->head = stbtt__find_table(data, fontstart, "head"); // required
- info->glyf = stbtt__find_table(data, fontstart, "glyf"); // required
- info->hhea = stbtt__find_table(data, fontstart, "hhea"); // required
- info->hmtx = stbtt__find_table(data, fontstart, "hmtx"); // required
- info->kern = stbtt__find_table(data, fontstart, "kern"); // not required
- info->gpos = stbtt__find_table(data, fontstart, "GPOS"); // not required
-
- if (!cmap || !info->head || !info->hhea || !info->hmtx)
- return 0;
- if (info->glyf) {
- // required for truetype
- if (!info->loca) return 0;
- }
- else {
- // initialization for CFF / Type2 fonts (OTF)
- stbtt__buf b, topdict, topdictidx;
- stbtt_uint32 cstype = 2, charstrings = 0, fdarrayoff = 0, fdselectoff = 0;
- stbtt_uint32 cff;
-
- cff = stbtt__find_table(data, fontstart, "CFF ");
- if (!cff) return 0;
-
- info->fontdicts = stbtt__new_buf(NULL, 0);
- info->fdselect = stbtt__new_buf(NULL, 0);
-
- // @TODO this should use size from table (not 512MB)
- info->cff = stbtt__new_buf(data + cff, 512 * 1024 * 1024);
- b = info->cff;
-
- // read the header
- stbtt__buf_skip(&b, 2);
- stbtt__buf_seek(&b, stbtt__buf_get8(&b)); // hdrsize
-
- // @TODO the name INDEX could list multiple fonts,
- // but we just use the first one.
- stbtt__cff_get_index(&b); // name INDEX
- topdictidx = stbtt__cff_get_index(&b);
- topdict = stbtt__cff_index_get(topdictidx, 0);
- stbtt__cff_get_index(&b); // string INDEX
- info->gsubrs = stbtt__cff_get_index(&b);
-
- stbtt__dict_get_ints(&topdict, 17, 1, &charstrings);
- stbtt__dict_get_ints(&topdict, 0x100 | 6, 1, &cstype);
- stbtt__dict_get_ints(&topdict, 0x100 | 36, 1, &fdarrayoff);
- stbtt__dict_get_ints(&topdict, 0x100 | 37, 1, &fdselectoff);
- info->subrs = stbtt__get_subrs(b, topdict);
-
- // we only support Type 2 charstrings
- if (cstype != 2) return 0;
- if (charstrings == 0) return 0;
-
- if (fdarrayoff) {
- // looks like a CID font
- if (!fdselectoff) return 0;
- stbtt__buf_seek(&b, fdarrayoff);
- info->fontdicts = stbtt__cff_get_index(&b);
- info->fdselect = stbtt__buf_range(&b, fdselectoff, b.size - fdselectoff);
- }
-
- stbtt__buf_seek(&b, charstrings);
- info->charstrings = stbtt__cff_get_index(&b);
- }
-
- t = stbtt__find_table(data, fontstart, "maxp");
- if (t)
- info->numGlyphs = ttUSHORT(data + t + 4);
- else
- info->numGlyphs = 0xffff;
-
- // find a cmap encoding table we understand *now* to avoid searching
- // later. (todo: could make this installable)
- // the same regardless of glyph.
- numTables = ttUSHORT(data + cmap + 2);
- info->index_map = 0;
- for (i = 0; i < numTables; ++i) {
- stbtt_uint32 encoding_record = cmap + 4 + 8 * i;
- // find an encoding we understand:
- switch (ttUSHORT(data + encoding_record)) {
- case STBTT_PLATFORM_ID_MICROSOFT:
- switch (ttUSHORT(data + encoding_record + 2)) {
- case STBTT_MS_EID_UNICODE_BMP:
- case STBTT_MS_EID_UNICODE_FULL:
- // MS/Unicode
- info->index_map = cmap + ttULONG(data + encoding_record + 4);
- break;
+ stbtt_uint32 cmap, t;
+ stbtt_int32 i,numTables;
+
+ info->data = data;
+ info->fontstart = fontstart;
+ info->cff = stbtt__new_buf(NULL, 0);
+
+ cmap = stbtt__find_table(data, fontstart, "cmap"); // required
+ info->loca = stbtt__find_table(data, fontstart, "loca"); // required
+ info->head = stbtt__find_table(data, fontstart, "head"); // required
+ info->glyf = stbtt__find_table(data, fontstart, "glyf"); // required
+ info->hhea = stbtt__find_table(data, fontstart, "hhea"); // required
+ info->hmtx = stbtt__find_table(data, fontstart, "hmtx"); // required
+ info->kern = stbtt__find_table(data, fontstart, "kern"); // not required
+ info->gpos = stbtt__find_table(data, fontstart, "GPOS"); // not required
+
+ if (!cmap || !info->head || !info->hhea || !info->hmtx)
+ return 0;
+ if (info->glyf) {
+ // required for truetype
+ if (!info->loca) return 0;
+ } else {
+ // initialization for CFF / Type2 fonts (OTF)
+ stbtt__buf b, topdict, topdictidx;
+ stbtt_uint32 cstype = 2, charstrings = 0, fdarrayoff = 0, fdselectoff = 0;
+ stbtt_uint32 cff;
+
+ cff = stbtt__find_table(data, fontstart, "CFF ");
+ if (!cff) return 0;
+
+ info->fontdicts = stbtt__new_buf(NULL, 0);
+ info->fdselect = stbtt__new_buf(NULL, 0);
+
+ // @TODO this should use size from table (not 512MB)
+ info->cff = stbtt__new_buf(data+cff, 512*1024*1024);
+ b = info->cff;
+
+ // read the header
+ stbtt__buf_skip(&b, 2);
+ stbtt__buf_seek(&b, stbtt__buf_get8(&b)); // hdrsize
+
+ // @TODO the name INDEX could list multiple fonts,
+ // but we just use the first one.
+ stbtt__cff_get_index(&b); // name INDEX
+ topdictidx = stbtt__cff_get_index(&b);
+ topdict = stbtt__cff_index_get(topdictidx, 0);
+ stbtt__cff_get_index(&b); // string INDEX
+ info->gsubrs = stbtt__cff_get_index(&b);
+
+ stbtt__dict_get_ints(&topdict, 17, 1, &charstrings);
+ stbtt__dict_get_ints(&topdict, 0x100 | 6, 1, &cstype);
+ stbtt__dict_get_ints(&topdict, 0x100 | 36, 1, &fdarrayoff);
+ stbtt__dict_get_ints(&topdict, 0x100 | 37, 1, &fdselectoff);
+ info->subrs = stbtt__get_subrs(b, topdict);
+
+ // we only support Type 2 charstrings
+ if (cstype != 2) return 0;
+ if (charstrings == 0) return 0;
+
+ if (fdarrayoff) {
+ // looks like a CID font
+ if (!fdselectoff) return 0;
+ stbtt__buf_seek(&b, fdarrayoff);
+ info->fontdicts = stbtt__cff_get_index(&b);
+ info->fdselect = stbtt__buf_range(&b, fdselectoff, b.size-fdselectoff);
+ }
+
+ stbtt__buf_seek(&b, charstrings);
+ info->charstrings = stbtt__cff_get_index(&b);
+ }
+
+ t = stbtt__find_table(data, fontstart, "maxp");
+ if (t)
+ info->numGlyphs = ttUSHORT(data+t+4);
+ else
+ info->numGlyphs = 0xffff;
+
+ info->svg = -1;
+
+ // find a cmap encoding table we understand *now* to avoid searching
+ // later. (todo: could make this installable)
+ // the same regardless of glyph.
+ numTables = ttUSHORT(data + cmap + 2);
+ info->index_map = 0;
+ for (i=0; i < numTables; ++i) {
+ stbtt_uint32 encoding_record = cmap + 4 + 8 * i;
+ // find an encoding we understand:
+ switch(ttUSHORT(data+encoding_record)) {
+ case STBTT_PLATFORM_ID_MICROSOFT:
+ switch (ttUSHORT(data+encoding_record+2)) {
+ case STBTT_MS_EID_UNICODE_BMP:
+ case STBTT_MS_EID_UNICODE_FULL:
+ // MS/Unicode
+ info->index_map = cmap + ttULONG(data+encoding_record+4);
+ break;
}
break;
case STBTT_PLATFORM_ID_UNICODE:
// Mac/iOS has these
// all the encodingIDs are unicode, so we don't bother to check it
- info->index_map = cmap + ttULONG(data + encoding_record + 4);
+ info->index_map = cmap + ttULONG(data+encoding_record+4);
break;
- }
- }
- if (info->index_map == 0)
- return 0;
+ }
+ }
+ if (info->index_map == 0)
+ return 0;
- info->indexToLocFormat = ttUSHORT(data + info->head + 50);
- return 1;
+ info->indexToLocFormat = ttUSHORT(data+info->head + 50);
+ return 1;
}
STBTT_DEF int stbtt_FindGlyphIndex(const stbtt_fontinfo *info, int unicode_codepoint)
{
- stbtt_uint8 *data = info->data;
- stbtt_uint32 index_map = info->index_map;
-
- stbtt_uint16 format = ttUSHORT(data + index_map + 0);
- if (format == 0) { // apple byte encoding
- stbtt_int32 bytes = ttUSHORT(data + index_map + 2);
- if (unicode_codepoint < bytes - 6)
- return ttBYTE(data + index_map + 6 + unicode_codepoint);
- return 0;
- }
- else if (format == 6) {
- stbtt_uint32 first = ttUSHORT(data + index_map + 6);
- stbtt_uint32 count = ttUSHORT(data + index_map + 8);
- if ((stbtt_uint32)unicode_codepoint >= first && (stbtt_uint32)unicode_codepoint < first + count)
- return ttUSHORT(data + index_map + 10 + (unicode_codepoint - first) * 2);
- return 0;
- }
- else if (format == 2) {
- STBTT_assert(0); // @TODO: high-byte mapping for japanese/chinese/korean
- return 0;
- }
- else if (format == 4) { // standard mapping for windows fonts: binary search collection of ranges
- stbtt_uint16 segcount = ttUSHORT(data + index_map + 6) >> 1;
- stbtt_uint16 searchRange = ttUSHORT(data + index_map + 8) >> 1;
- stbtt_uint16 entrySelector = ttUSHORT(data + index_map + 10);
- stbtt_uint16 rangeShift = ttUSHORT(data + index_map + 12) >> 1;
-
- // do a binary search of the segments
- stbtt_uint32 endCount = index_map + 14;
- stbtt_uint32 search = endCount;
-
- if (unicode_codepoint > 0xffff)
+ stbtt_uint8 *data = info->data;
+ stbtt_uint32 index_map = info->index_map;
+
+ stbtt_uint16 format = ttUSHORT(data + index_map + 0);
+ if (format == 0) { // apple byte encoding
+ stbtt_int32 bytes = ttUSHORT(data + index_map + 2);
+ if (unicode_codepoint < bytes-6)
+ return ttBYTE(data + index_map + 6 + unicode_codepoint);
+ return 0;
+ } else if (format == 6) {
+ stbtt_uint32 first = ttUSHORT(data + index_map + 6);
+ stbtt_uint32 count = ttUSHORT(data + index_map + 8);
+ if ((stbtt_uint32) unicode_codepoint >= first && (stbtt_uint32) unicode_codepoint < first+count)
+ return ttUSHORT(data + index_map + 10 + (unicode_codepoint - first)*2);
+ return 0;
+ } else if (format == 2) {
+ STBTT_assert(0); // @TODO: high-byte mapping for japanese/chinese/korean
+ return 0;
+ } else if (format == 4) { // standard mapping for windows fonts: binary search collection of ranges
+ stbtt_uint16 segcount = ttUSHORT(data+index_map+6) >> 1;
+ stbtt_uint16 searchRange = ttUSHORT(data+index_map+8) >> 1;
+ stbtt_uint16 entrySelector = ttUSHORT(data+index_map+10);
+ stbtt_uint16 rangeShift = ttUSHORT(data+index_map+12) >> 1;
+
+ // do a binary search of the segments
+ stbtt_uint32 endCount = index_map + 14;
+ stbtt_uint32 search = endCount;
+
+ if (unicode_codepoint > 0xffff)
+ return 0;
+
+ // they lie from endCount .. endCount + segCount
+ // but searchRange is the nearest power of two, so...
+ if (unicode_codepoint >= ttUSHORT(data + search + rangeShift*2))
+ search += rangeShift*2;
+
+ // now decrement to bias correctly to find smallest
+ search -= 2;
+ while (entrySelector) {
+ stbtt_uint16 end;
+ searchRange >>= 1;
+ end = ttUSHORT(data + search + searchRange*2);
+ if (unicode_codepoint > end)
+ search += searchRange*2;
+ --entrySelector;
+ }
+ search += 2;
+
+ {
+ stbtt_uint16 offset, start, last;
+ stbtt_uint16 item = (stbtt_uint16) ((search - endCount) >> 1);
+
+ start = ttUSHORT(data + index_map + 14 + segcount*2 + 2 + 2*item);
+ last = ttUSHORT(data + endCount + 2*item);
+ if (unicode_codepoint < start || unicode_codepoint > last)
return 0;
- // they lie from endCount .. endCount + segCount
- // but searchRange is the nearest power of two, so...
- if (unicode_codepoint >= ttUSHORT(data + search + rangeShift * 2))
- search += rangeShift * 2;
-
- // now decrement to bias correctly to find smallest
- search -= 2;
- while (entrySelector) {
- stbtt_uint16 end;
- searchRange >>= 1;
- end = ttUSHORT(data + search + searchRange * 2);
- if (unicode_codepoint > end)
- search += searchRange * 2;
- --entrySelector;
- }
- search += 2;
-
- {
- stbtt_uint16 offset, start;
- stbtt_uint16 item = (stbtt_uint16)((search - endCount) >> 1);
-
- STBTT_assert(unicode_codepoint <= ttUSHORT(data + endCount + 2 * item));
- start = ttUSHORT(data + index_map + 14 + segcount * 2 + 2 + 2 * item);
- if (unicode_codepoint < start)
- return 0;
-
- offset = ttUSHORT(data + index_map + 14 + segcount * 6 + 2 + 2 * item);
- if (offset == 0)
- return (stbtt_uint16)(unicode_codepoint + ttSHORT(data + index_map + 14 + segcount * 4 + 2 + 2 * item));
-
- return ttUSHORT(data + offset + (unicode_codepoint - start) * 2 + index_map + 14 + segcount * 6 + 2 + 2 * item);
- }
- }
- else if (format == 12 || format == 13) {
- stbtt_uint32 ngroups = ttULONG(data + index_map + 12);
- stbtt_int32 low, high;
- low = 0; high = (stbtt_int32)ngroups;
- // Binary search the right group.
- while (low < high) {
- stbtt_int32 mid = low + ((high - low) >> 1); // rounds down, so low <= mid < high
- stbtt_uint32 start_char = ttULONG(data + index_map + 16 + mid * 12);
- stbtt_uint32 end_char = ttULONG(data + index_map + 16 + mid * 12 + 4);
- if ((stbtt_uint32)unicode_codepoint < start_char)
- high = mid;
- else if ((stbtt_uint32)unicode_codepoint > end_char)
- low = mid + 1;
- else {
- stbtt_uint32 start_glyph = ttULONG(data + index_map + 16 + mid * 12 + 8);
- if (format == 12)
- return start_glyph + unicode_codepoint - start_char;
- else // format == 13
- return start_glyph;
- }
- }
- return 0; // not found
- }
- // @TODO
- STBTT_assert(0);
- return 0;
+ offset = ttUSHORT(data + index_map + 14 + segcount*6 + 2 + 2*item);
+ if (offset == 0)
+ return (stbtt_uint16) (unicode_codepoint + ttSHORT(data + index_map + 14 + segcount*4 + 2 + 2*item));
+
+ return ttUSHORT(data + offset + (unicode_codepoint-start)*2 + index_map + 14 + segcount*6 + 2 + 2*item);
+ }
+ } else if (format == 12 || format == 13) {
+ stbtt_uint32 ngroups = ttULONG(data+index_map+12);
+ stbtt_int32 low,high;
+ low = 0; high = (stbtt_int32)ngroups;
+ // Binary search the right group.
+ while (low < high) {
+ stbtt_int32 mid = low + ((high-low) >> 1); // rounds down, so low <= mid < high
+ stbtt_uint32 start_char = ttULONG(data+index_map+16+mid*12);
+ stbtt_uint32 end_char = ttULONG(data+index_map+16+mid*12+4);
+ if ((stbtt_uint32) unicode_codepoint < start_char)
+ high = mid;
+ else if ((stbtt_uint32) unicode_codepoint > end_char)
+ low = mid+1;
+ else {
+ stbtt_uint32 start_glyph = ttULONG(data+index_map+16+mid*12+8);
+ if (format == 12)
+ return start_glyph + unicode_codepoint-start_char;
+ else // format == 13
+ return start_glyph;
+ }
+ }
+ return 0; // not found
+ }
+ // @TODO
+ STBTT_assert(0);
+ return 0;
}
STBTT_DEF int stbtt_GetCodepointShape(const stbtt_fontinfo *info, int unicode_codepoint, stbtt_vertex **vertices)
{
- return stbtt_GetGlyphShape(info, stbtt_FindGlyphIndex(info, unicode_codepoint), vertices);
+ return stbtt_GetGlyphShape(info, stbtt_FindGlyphIndex(info, unicode_codepoint), vertices);
}
static void stbtt_setvertex(stbtt_vertex *v, stbtt_uint8 type, stbtt_int32 x, stbtt_int32 y, stbtt_int32 cx, stbtt_int32 cy)
{
- v->type = type;
- v->x = (stbtt_int16)x;
- v->y = (stbtt_int16)y;
- v->cx = (stbtt_int16)cx;
- v->cy = (stbtt_int16)cy;
+ v->type = type;
+ v->x = (stbtt_int16) x;
+ v->y = (stbtt_int16) y;
+ v->cx = (stbtt_int16) cx;
+ v->cy = (stbtt_int16) cy;
}
static int stbtt__GetGlyfOffset(const stbtt_fontinfo *info, int glyph_index)
{
- int g1, g2;
+ int g1,g2;
- STBTT_assert(!info->cff.size);
+ STBTT_assert(!info->cff.size);
- if (glyph_index >= info->numGlyphs) return -1; // glyph index out of range
- if (info->indexToLocFormat >= 2) return -1; // unknown index->glyph map format
+ if (glyph_index >= info->numGlyphs) return -1; // glyph index out of range
+ if (info->indexToLocFormat >= 2) return -1; // unknown index->glyph map format
- if (info->indexToLocFormat == 0) {
- g1 = info->glyf + ttUSHORT(info->data + info->loca + glyph_index * 2) * 2;
- g2 = info->glyf + ttUSHORT(info->data + info->loca + glyph_index * 2 + 2) * 2;
- }
- else {
- g1 = info->glyf + ttULONG(info->data + info->loca + glyph_index * 4);
- g2 = info->glyf + ttULONG(info->data + info->loca + glyph_index * 4 + 4);
- }
+ if (info->indexToLocFormat == 0) {
+ g1 = info->glyf + ttUSHORT(info->data + info->loca + glyph_index * 2) * 2;
+ g2 = info->glyf + ttUSHORT(info->data + info->loca + glyph_index * 2 + 2) * 2;
+ } else {
+ g1 = info->glyf + ttULONG (info->data + info->loca + glyph_index * 4);
+ g2 = info->glyf + ttULONG (info->data + info->loca + glyph_index * 4 + 4);
+ }
- return g1 == g2 ? -1 : g1; // if length is 0, return -1
+ return g1==g2 ? -1 : g1; // if length is 0, return -1
}
static int stbtt__GetGlyphInfoT2(const stbtt_fontinfo *info, int glyph_index, int *x0, int *y0, int *x1, int *y1);
STBTT_DEF int stbtt_GetGlyphBox(const stbtt_fontinfo *info, int glyph_index, int *x0, int *y0, int *x1, int *y1)
{
- if (info->cff.size) {
- stbtt__GetGlyphInfoT2(info, glyph_index, x0, y0, x1, y1);
- }
- else {
- int g = stbtt__GetGlyfOffset(info, glyph_index);
- if (g < 0) return 0;
+ if (info->cff.size) {
+ stbtt__GetGlyphInfoT2(info, glyph_index, x0, y0, x1, y1);
+ } else {
+ int g = stbtt__GetGlyfOffset(info, glyph_index);
+ if (g < 0) return 0;
- if (x0) *x0 = ttSHORT(info->data + g + 2);
- if (y0) *y0 = ttSHORT(info->data + g + 4);
- if (x1) *x1 = ttSHORT(info->data + g + 6);
- if (y1) *y1 = ttSHORT(info->data + g + 8);
- }
- return 1;
+ if (x0) *x0 = ttSHORT(info->data + g + 2);
+ if (y0) *y0 = ttSHORT(info->data + g + 4);
+ if (x1) *x1 = ttSHORT(info->data + g + 6);
+ if (y1) *y1 = ttSHORT(info->data + g + 8);
+ }
+ return 1;
}
STBTT_DEF int stbtt_GetCodepointBox(const stbtt_fontinfo *info, int codepoint, int *x0, int *y0, int *x1, int *y1)
{
- return stbtt_GetGlyphBox(info, stbtt_FindGlyphIndex(info, codepoint), x0, y0, x1, y1);
+ return stbtt_GetGlyphBox(info, stbtt_FindGlyphIndex(info,codepoint), x0,y0,x1,y1);
}
STBTT_DEF int stbtt_IsGlyphEmpty(const stbtt_fontinfo *info, int glyph_index)
{
- stbtt_int16 numberOfContours;
- int g;
- if (info->cff.size)
- return stbtt__GetGlyphInfoT2(info, glyph_index, NULL, NULL, NULL, NULL) == 0;
- g = stbtt__GetGlyfOffset(info, glyph_index);
- if (g < 0) return 1;
- numberOfContours = ttSHORT(info->data + g);
- return numberOfContours == 0;
+ stbtt_int16 numberOfContours;
+ int g;
+ if (info->cff.size)
+ return stbtt__GetGlyphInfoT2(info, glyph_index, NULL, NULL, NULL, NULL) == 0;
+ g = stbtt__GetGlyfOffset(info, glyph_index);
+ if (g < 0) return 1;
+ numberOfContours = ttSHORT(info->data + g);
+ return numberOfContours == 0;
}
static int stbtt__close_shape(stbtt_vertex *vertices, int num_vertices, int was_off, int start_off,
stbtt_int32 sx, stbtt_int32 sy, stbtt_int32 scx, stbtt_int32 scy, stbtt_int32 cx, stbtt_int32 cy)
{
- if (start_off) {
- if (was_off)
- stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, (cx + scx) >> 1, (cy + scy) >> 1, cx, cy);
- stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, sx, sy, scx, scy);
- }
- else {
- if (was_off)
- stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, sx, sy, cx, cy);
- else
- stbtt_setvertex(&vertices[num_vertices++], STBTT_vline, sx, sy, 0, 0);
- }
- return num_vertices;
+ if (start_off) {
+ if (was_off)
+ stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, (cx+scx)>>1, (cy+scy)>>1, cx,cy);
+ stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, sx,sy,scx,scy);
+ } else {
+ if (was_off)
+ stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve,sx,sy,cx,cy);
+ else
+ stbtt_setvertex(&vertices[num_vertices++], STBTT_vline,sx,sy,0,0);
+ }
+ return num_vertices;
}
static int stbtt__GetGlyphShapeTT(const stbtt_fontinfo *info, int glyph_index, stbtt_vertex **pvertices)
{
- stbtt_int16 numberOfContours;
- stbtt_uint8 *endPtsOfContours;
- stbtt_uint8 *data = info->data;
- stbtt_vertex *vertices = 0;
- int num_vertices = 0;
- int g = stbtt__GetGlyfOffset(info, glyph_index);
-
- *pvertices = NULL;
-
- if (g < 0) return 0;
-
- numberOfContours = ttSHORT(data + g);
-
- if (numberOfContours > 0) {
- stbtt_uint8 flags = 0, flagcount;
- stbtt_int32 ins, i, j = 0, m, n, next_move, was_off = 0, off, start_off = 0;
- stbtt_int32 x, y, cx, cy, sx, sy, scx, scy;
- stbtt_uint8 *points;
- endPtsOfContours = (data + g + 10);
- ins = ttUSHORT(data + g + 10 + numberOfContours * 2);
- points = data + g + 10 + numberOfContours * 2 + 2 + ins;
-
- n = 1 + ttUSHORT(endPtsOfContours + numberOfContours * 2 - 2);
-
- m = n + 2 * numberOfContours; // a loose bound on how many vertices we might need
- vertices = (stbtt_vertex *)STBTT_malloc(m * sizeof(vertices[0]), info->userdata);
- if (vertices == 0)
- return 0;
-
- next_move = 0;
- flagcount = 0;
-
- // in first pass, we load uninterpreted data into the allocated array
- // above, shifted to the end of the array so we won't overwrite it when
- // we create our final data starting from the front
-
- off = m - n; // starting offset for uninterpreted data, regardless of how m ends up being calculated
-
- // first load flags
-
- for (i = 0; i < n; ++i) {
- if (flagcount == 0) {
- flags = *points++;
- if (flags & 8)
- flagcount = *points++;
+ stbtt_int16 numberOfContours;
+ stbtt_uint8 *endPtsOfContours;
+ stbtt_uint8 *data = info->data;
+ stbtt_vertex *vertices=0;
+ int num_vertices=0;
+ int g = stbtt__GetGlyfOffset(info, glyph_index);
+
+ *pvertices = NULL;
+
+ if (g < 0) return 0;
+
+ numberOfContours = ttSHORT(data + g);
+
+ if (numberOfContours > 0) {
+ stbtt_uint8 flags=0,flagcount;
+ stbtt_int32 ins, i,j=0,m,n, next_move, was_off=0, off, start_off=0;
+ stbtt_int32 x,y,cx,cy,sx,sy, scx,scy;
+ stbtt_uint8 *points;
+ endPtsOfContours = (data + g + 10);
+ ins = ttUSHORT(data + g + 10 + numberOfContours * 2);
+ points = data + g + 10 + numberOfContours * 2 + 2 + ins;
+
+ n = 1+ttUSHORT(endPtsOfContours + numberOfContours*2-2);
+
+ m = n + 2*numberOfContours; // a loose bound on how many vertices we might need
+ vertices = (stbtt_vertex *) STBTT_malloc(m * sizeof(vertices[0]), info->userdata);
+ if (vertices == 0)
+ return 0;
+
+ next_move = 0;
+ flagcount=0;
+
+ // in first pass, we load uninterpreted data into the allocated array
+ // above, shifted to the end of the array so we won't overwrite it when
+ // we create our final data starting from the front
+
+ off = m - n; // starting offset for uninterpreted data, regardless of how m ends up being calculated
+
+ // first load flags
+
+ for (i=0; i < n; ++i) {
+ if (flagcount == 0) {
+ flags = *points++;
+ if (flags & 8)
+ flagcount = *points++;
+ } else
+ --flagcount;
+ vertices[off+i].type = flags;
+ }
+
+ // now load x coordinates
+ x=0;
+ for (i=0; i < n; ++i) {
+ flags = vertices[off+i].type;
+ if (flags & 2) {
+ stbtt_int16 dx = *points++;
+ x += (flags & 16) ? dx : -dx; // ???
+ } else {
+ if (!(flags & 16)) {
+ x = x + (stbtt_int16) (points[0]*256 + points[1]);
+ points += 2;
}
- else
- --flagcount;
- vertices[off + i].type = flags;
- }
-
- // now load x coordinates
- x = 0;
- for (i = 0; i < n; ++i) {
- flags = vertices[off + i].type;
- if (flags & 2) {
- stbtt_int16 dx = *points++;
- x += (flags & 16) ? dx : -dx; // ???
+ }
+ vertices[off+i].x = (stbtt_int16) x;
+ }
+
+ // now load y coordinates
+ y=0;
+ for (i=0; i < n; ++i) {
+ flags = vertices[off+i].type;
+ if (flags & 4) {
+ stbtt_int16 dy = *points++;
+ y += (flags & 32) ? dy : -dy; // ???
+ } else {
+ if (!(flags & 32)) {
+ y = y + (stbtt_int16) (points[0]*256 + points[1]);
+ points += 2;
}
- else {
- if (!(flags & 16)) {
- x = x + (stbtt_int16)(points[0] * 256 + points[1]);
- points += 2;
- }
+ }
+ vertices[off+i].y = (stbtt_int16) y;
+ }
+
+ // now convert them to our format
+ num_vertices=0;
+ sx = sy = cx = cy = scx = scy = 0;
+ for (i=0; i < n; ++i) {
+ flags = vertices[off+i].type;
+ x = (stbtt_int16) vertices[off+i].x;
+ y = (stbtt_int16) vertices[off+i].y;
+
+ if (next_move == i) {
+ if (i != 0)
+ num_vertices = stbtt__close_shape(vertices, num_vertices, was_off, start_off, sx,sy,scx,scy,cx,cy);
+
+ // now start the new one
+ start_off = !(flags & 1);
+ if (start_off) {
+ // if we start off with an off-curve point, then when we need to find a point on the curve
+ // where we can start, and we need to save some state for when we wraparound.
+ scx = x;
+ scy = y;
+ if (!(vertices[off+i+1].type & 1)) {
+ // next point is also a curve point, so interpolate an on-point curve
+ sx = (x + (stbtt_int32) vertices[off+i+1].x) >> 1;
+ sy = (y + (stbtt_int32) vertices[off+i+1].y) >> 1;
+ } else {
+ // otherwise just use the next point as our start point
+ sx = (stbtt_int32) vertices[off+i+1].x;
+ sy = (stbtt_int32) vertices[off+i+1].y;
+ ++i; // we're using point i+1 as the starting point, so skip it
+ }
+ } else {
+ sx = x;
+ sy = y;
}
- vertices[off + i].x = (stbtt_int16)x;
- }
-
- // now load y coordinates
- y = 0;
- for (i = 0; i < n; ++i) {
- flags = vertices[off + i].type;
- if (flags & 4) {
- stbtt_int16 dy = *points++;
- y += (flags & 32) ? dy : -dy; // ???
+ stbtt_setvertex(&vertices[num_vertices++], STBTT_vmove,sx,sy,0,0);
+ was_off = 0;
+ next_move = 1 + ttUSHORT(endPtsOfContours+j*2);
+ ++j;
+ } else {
+ if (!(flags & 1)) { // if it's a curve
+ if (was_off) // two off-curve control points in a row means interpolate an on-curve midpoint
+ stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, (cx+x)>>1, (cy+y)>>1, cx, cy);
+ cx = x;
+ cy = y;
+ was_off = 1;
+ } else {
+ if (was_off)
+ stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, x,y, cx, cy);
+ else
+ stbtt_setvertex(&vertices[num_vertices++], STBTT_vline, x,y,0,0);
+ was_off = 0;
}
- else {
- if (!(flags & 32)) {
- y = y + (stbtt_int16)(points[0] * 256 + points[1]);
- points += 2;
- }
+ }
+ }
+ num_vertices = stbtt__close_shape(vertices, num_vertices, was_off, start_off, sx,sy,scx,scy,cx,cy);
+ } else if (numberOfContours < 0) {
+ // Compound shapes.
+ int more = 1;
+ stbtt_uint8 *comp = data + g + 10;
+ num_vertices = 0;
+ vertices = 0;
+ while (more) {
+ stbtt_uint16 flags, gidx;
+ int comp_num_verts = 0, i;
+ stbtt_vertex *comp_verts = 0, *tmp = 0;
+ float mtx[6] = {1,0,0,1,0,0}, m, n;
+
+ flags = ttSHORT(comp); comp+=2;
+ gidx = ttSHORT(comp); comp+=2;
+
+ if (flags & 2) { // XY values
+ if (flags & 1) { // shorts
+ mtx[4] = ttSHORT(comp); comp+=2;
+ mtx[5] = ttSHORT(comp); comp+=2;
+ } else {
+ mtx[4] = ttCHAR(comp); comp+=1;
+ mtx[5] = ttCHAR(comp); comp+=1;
}
- vertices[off + i].y = (stbtt_int16)y;
- }
-
- // now convert them to our format
- num_vertices = 0;
- sx = sy = cx = cy = scx = scy = 0;
- for (i = 0; i < n; ++i) {
- flags = vertices[off + i].type;
- x = (stbtt_int16)vertices[off + i].x;
- y = (stbtt_int16)vertices[off + i].y;
-
- if (next_move == i) {
- if (i != 0)
- num_vertices = stbtt__close_shape(vertices, num_vertices, was_off, start_off, sx, sy, scx, scy, cx, cy);
-
- // now start the new one
- start_off = !(flags & 1);
- if (start_off) {
- // if we start off with an off-curve point, then when we need to find a point on the curve
- // where we can start, and we need to save some state for when we wraparound.
- scx = x;
- scy = y;
- if (!(vertices[off + i + 1].type & 1)) {
- // next point is also a curve point, so interpolate an on-point curve
- sx = (x + (stbtt_int32)vertices[off + i + 1].x) >> 1;
- sy = (y + (stbtt_int32)vertices[off + i + 1].y) >> 1;
- }
- else {
- // otherwise just use the next point as our start point
- sx = (stbtt_int32)vertices[off + i + 1].x;
- sy = (stbtt_int32)vertices[off + i + 1].y;
- ++i; // we're using point i+1 as the starting point, so skip it
- }
- }
- else {
- sx = x;
- sy = y;
- }
- stbtt_setvertex(&vertices[num_vertices++], STBTT_vmove, sx, sy, 0, 0);
- was_off = 0;
- next_move = 1 + ttUSHORT(endPtsOfContours + j * 2);
- ++j;
+ }
+ else {
+ // @TODO handle matching point
+ STBTT_assert(0);
+ }
+ if (flags & (1<<3)) { // WE_HAVE_A_SCALE
+ mtx[0] = mtx[3] = ttSHORT(comp)/16384.0f; comp+=2;
+ mtx[1] = mtx[2] = 0;
+ } else if (flags & (1<<6)) { // WE_HAVE_AN_X_AND_YSCALE
+ mtx[0] = ttSHORT(comp)/16384.0f; comp+=2;
+ mtx[1] = mtx[2] = 0;
+ mtx[3] = ttSHORT(comp)/16384.0f; comp+=2;
+ } else if (flags & (1<<7)) { // WE_HAVE_A_TWO_BY_TWO
+ mtx[0] = ttSHORT(comp)/16384.0f; comp+=2;
+ mtx[1] = ttSHORT(comp)/16384.0f; comp+=2;
+ mtx[2] = ttSHORT(comp)/16384.0f; comp+=2;
+ mtx[3] = ttSHORT(comp)/16384.0f; comp+=2;
+ }
+
+ // Find transformation scales.
+ m = (float) STBTT_sqrt(mtx[0]*mtx[0] + mtx[1]*mtx[1]);
+ n = (float) STBTT_sqrt(mtx[2]*mtx[2] + mtx[3]*mtx[3]);
+
+ // Get indexed glyph.
+ comp_num_verts = stbtt_GetGlyphShape(info, gidx, &comp_verts);
+ if (comp_num_verts > 0) {
+ // Transform vertices.
+ for (i = 0; i < comp_num_verts; ++i) {
+ stbtt_vertex* v = &comp_verts[i];
+ stbtt_vertex_type x,y;
+ x=v->x; y=v->y;
+ v->x = (stbtt_vertex_type)(m * (mtx[0]*x + mtx[2]*y + mtx[4]));
+ v->y = (stbtt_vertex_type)(n * (mtx[1]*x + mtx[3]*y + mtx[5]));
+ x=v->cx; y=v->cy;
+ v->cx = (stbtt_vertex_type)(m * (mtx[0]*x + mtx[2]*y + mtx[4]));
+ v->cy = (stbtt_vertex_type)(n * (mtx[1]*x + mtx[3]*y + mtx[5]));
}
- else {
- if (!(flags & 1)) { // if it's a curve
- if (was_off) // two off-curve control points in a row means interpolate an on-curve midpoint
- stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, (cx + x) >> 1, (cy + y) >> 1, cx, cy);
- cx = x;
- cy = y;
- was_off = 1;
- }
- else {
- if (was_off)
- stbtt_setvertex(&vertices[num_vertices++], STBTT_vcurve, x, y, cx, cy);
- else
- stbtt_setvertex(&vertices[num_vertices++], STBTT_vline, x, y, 0, 0);
- was_off = 0;
- }
- }
- }
- num_vertices = stbtt__close_shape(vertices, num_vertices, was_off, start_off, sx, sy, scx, scy, cx, cy);
- }
- else if (numberOfContours == -1) {
- // Compound shapes.
- int more = 1;
- stbtt_uint8 *comp = data + g + 10;
- num_vertices = 0;
- vertices = 0;
- while (more) {
- stbtt_uint16 flags, gidx;
- int comp_num_verts = 0, i;
- stbtt_vertex *comp_verts = 0, *tmp = 0;
- float mtx[6] = { 1,0,0,1,0,0 }, m, n;
-
- flags = ttSHORT(comp); comp += 2;
- gidx = ttSHORT(comp); comp += 2;
-
- if (flags & 2) { // XY values
- if (flags & 1) { // shorts
- mtx[4] = ttSHORT(comp); comp += 2;
- mtx[5] = ttSHORT(comp); comp += 2;
- }
- else {
- mtx[4] = ttCHAR(comp); comp += 1;
- mtx[5] = ttCHAR(comp); comp += 1;
- }
- }
- else {
- // @TODO handle matching point
- STBTT_assert(0);
- }
- if (flags & (1 << 3)) { // WE_HAVE_A_SCALE
- mtx[0] = mtx[3] = ttSHORT(comp) / 16384.0f; comp += 2;
- mtx[1] = mtx[2] = 0;
+ // Append vertices.
+ tmp = (stbtt_vertex*)STBTT_malloc((num_vertices+comp_num_verts)*sizeof(stbtt_vertex), info->userdata);
+ if (!tmp) {
+ if (vertices) STBTT_free(vertices, info->userdata);
+ if (comp_verts) STBTT_free(comp_verts, info->userdata);
+ return 0;
}
- else if (flags & (1 << 6)) { // WE_HAVE_AN_X_AND_YSCALE
- mtx[0] = ttSHORT(comp) / 16384.0f; comp += 2;
- mtx[1] = mtx[2] = 0;
- mtx[3] = ttSHORT(comp) / 16384.0f; comp += 2;
- }
- else if (flags & (1 << 7)) { // WE_HAVE_A_TWO_BY_TWO
- mtx[0] = ttSHORT(comp) / 16384.0f; comp += 2;
- mtx[1] = ttSHORT(comp) / 16384.0f; comp += 2;
- mtx[2] = ttSHORT(comp) / 16384.0f; comp += 2;
- mtx[3] = ttSHORT(comp) / 16384.0f; comp += 2;
- }
-
- // Find transformation scales.
- m = (float)STBTT_sqrt(mtx[0] * mtx[0] + mtx[1] * mtx[1]);
- n = (float)STBTT_sqrt(mtx[2] * mtx[2] + mtx[3] * mtx[3]);
-
- // Get indexed glyph.
- comp_num_verts = stbtt_GetGlyphShape(info, gidx, &comp_verts);
- if (comp_num_verts > 0) {
- // Transform vertices.
- for (i = 0; i < comp_num_verts; ++i) {
- stbtt_vertex* v = &comp_verts[i];
- stbtt_vertex_type x, y;
- x = v->x; y = v->y;
- v->x = (stbtt_vertex_type)(m * (mtx[0] * x + mtx[2] * y + mtx[4]));
- v->y = (stbtt_vertex_type)(n * (mtx[1] * x + mtx[3] * y + mtx[5]));
- x = v->cx; y = v->cy;
- v->cx = (stbtt_vertex_type)(m * (mtx[0] * x + mtx[2] * y + mtx[4]));
- v->cy = (stbtt_vertex_type)(n * (mtx[1] * x + mtx[3] * y + mtx[5]));
- }
- // Append vertices.
- tmp = (stbtt_vertex*)STBTT_malloc((num_vertices + comp_num_verts) * sizeof(stbtt_vertex), info->userdata);
- if (!tmp) {
- if (vertices) STBTT_free(vertices, info->userdata);
- if (comp_verts) STBTT_free(comp_verts, info->userdata);
- return 0;
- }
- if (num_vertices > 0) STBTT_memcpy(tmp, vertices, num_vertices * sizeof(stbtt_vertex));
- STBTT_memcpy(tmp + num_vertices, comp_verts, comp_num_verts * sizeof(stbtt_vertex));
- if (vertices) STBTT_free(vertices, info->userdata);
- vertices = tmp;
- STBTT_free(comp_verts, info->userdata);
- num_vertices += comp_num_verts;
- }
- // More components ?
- more = flags & (1 << 5);
- }
- }
- else if (numberOfContours < 0) {
- // @TODO other compound variations?
- STBTT_assert(0);
- }
- else {
- // numberOfCounters == 0, do nothing
- }
-
- *pvertices = vertices;
- return num_vertices;
+ if (num_vertices > 0 && vertices) STBTT_memcpy(tmp, vertices, num_vertices*sizeof(stbtt_vertex));
+ STBTT_memcpy(tmp+num_vertices, comp_verts, comp_num_verts*sizeof(stbtt_vertex));
+ if (vertices) STBTT_free(vertices, info->userdata);
+ vertices = tmp;
+ STBTT_free(comp_verts, info->userdata);
+ num_vertices += comp_num_verts;
+ }
+ // More components ?
+ more = flags & (1<<5);
+ }
+ } else {
+ // numberOfCounters == 0, do nothing
+ }
+
+ *pvertices = vertices;
+ return num_vertices;
}
typedef struct
{
- int bounds;
- int started;
- float first_x, first_y;
- float x, y;
- stbtt_int32 min_x, max_x, min_y, max_y;
+ int bounds;
+ int started;
+ float first_x, first_y;
+ float x, y;
+ stbtt_int32 min_x, max_x, min_y, max_y;
- stbtt_vertex *pvertices;
- int num_vertices;
+ stbtt_vertex *pvertices;
+ int num_vertices;
} stbtt__csctx;
#define STBTT__CSCTX_INIT(bounds) {bounds,0, 0,0, 0,0, 0,0,0,0, NULL, 0}
static void stbtt__track_vertex(stbtt__csctx *c, stbtt_int32 x, stbtt_int32 y)
{
- if (x > c->max_x || !c->started) c->max_x = x;
- if (y > c->max_y || !c->started) c->max_y = y;
- if (x < c->min_x || !c->started) c->min_x = x;
- if (y < c->min_y || !c->started) c->min_y = y;
- c->started = 1;
+ if (x > c->max_x || !c->started) c->max_x = x;
+ if (y > c->max_y || !c->started) c->max_y = y;
+ if (x < c->min_x || !c->started) c->min_x = x;
+ if (y < c->min_y || !c->started) c->min_y = y;
+ c->started = 1;
}
static void stbtt__csctx_v(stbtt__csctx *c, stbtt_uint8 type, stbtt_int32 x, stbtt_int32 y, stbtt_int32 cx, stbtt_int32 cy, stbtt_int32 cx1, stbtt_int32 cy1)
{
- if (c->bounds) {
- stbtt__track_vertex(c, x, y);
- if (type == STBTT_vcubic) {
- stbtt__track_vertex(c, cx, cy);
- stbtt__track_vertex(c, cx1, cy1);
- }
- }
- else {
- stbtt_setvertex(&c->pvertices[c->num_vertices], type, x, y, cx, cy);
- c->pvertices[c->num_vertices].cx1 = (stbtt_int16)cx1;
- c->pvertices[c->num_vertices].cy1 = (stbtt_int16)cy1;
- }
- c->num_vertices++;
+ if (c->bounds) {
+ stbtt__track_vertex(c, x, y);
+ if (type == STBTT_vcubic) {
+ stbtt__track_vertex(c, cx, cy);
+ stbtt__track_vertex(c, cx1, cy1);
+ }
+ } else {
+ stbtt_setvertex(&c->pvertices[c->num_vertices], type, x, y, cx, cy);
+ c->pvertices[c->num_vertices].cx1 = (stbtt_int16) cx1;
+ c->pvertices[c->num_vertices].cy1 = (stbtt_int16) cy1;
+ }
+ c->num_vertices++;
}
static void stbtt__csctx_close_shape(stbtt__csctx *ctx)
{
- if (ctx->first_x != ctx->x || ctx->first_y != ctx->y)
- stbtt__csctx_v(ctx, STBTT_vline, (int)ctx->first_x, (int)ctx->first_y, 0, 0, 0, 0);
+ if (ctx->first_x != ctx->x || ctx->first_y != ctx->y)
+ stbtt__csctx_v(ctx, STBTT_vline, (int)ctx->first_x, (int)ctx->first_y, 0, 0, 0, 0);
}
static void stbtt__csctx_rmove_to(stbtt__csctx *ctx, float dx, float dy)
{
- stbtt__csctx_close_shape(ctx);
- ctx->first_x = ctx->x = ctx->x + dx;
- ctx->first_y = ctx->y = ctx->y + dy;
- stbtt__csctx_v(ctx, STBTT_vmove, (int)ctx->x, (int)ctx->y, 0, 0, 0, 0);
+ stbtt__csctx_close_shape(ctx);
+ ctx->first_x = ctx->x = ctx->x + dx;
+ ctx->first_y = ctx->y = ctx->y + dy;
+ stbtt__csctx_v(ctx, STBTT_vmove, (int)ctx->x, (int)ctx->y, 0, 0, 0, 0);
}
static void stbtt__csctx_rline_to(stbtt__csctx *ctx, float dx, float dy)
{
- ctx->x += dx;
- ctx->y += dy;
- stbtt__csctx_v(ctx, STBTT_vline, (int)ctx->x, (int)ctx->y, 0, 0, 0, 0);
+ ctx->x += dx;
+ ctx->y += dy;
+ stbtt__csctx_v(ctx, STBTT_vline, (int)ctx->x, (int)ctx->y, 0, 0, 0, 0);
}
static void stbtt__csctx_rccurve_to(stbtt__csctx *ctx, float dx1, float dy1, float dx2, float dy2, float dx3, float dy3)
{
- float cx1 = ctx->x + dx1;
- float cy1 = ctx->y + dy1;
- float cx2 = cx1 + dx2;
- float cy2 = cy1 + dy2;
- ctx->x = cx2 + dx3;
- ctx->y = cy2 + dy3;
- stbtt__csctx_v(ctx, STBTT_vcubic, (int)ctx->x, (int)ctx->y, (int)cx1, (int)cy1, (int)cx2, (int)cy2);
+ float cx1 = ctx->x + dx1;
+ float cy1 = ctx->y + dy1;
+ float cx2 = cx1 + dx2;
+ float cy2 = cy1 + dy2;
+ ctx->x = cx2 + dx3;
+ ctx->y = cy2 + dy3;
+ stbtt__csctx_v(ctx, STBTT_vcubic, (int)ctx->x, (int)ctx->y, (int)cx1, (int)cy1, (int)cx2, (int)cy2);
}
static stbtt__buf stbtt__get_subr(stbtt__buf idx, int n)
{
- int count = stbtt__cff_index_count(&idx);
- int bias = 107;
- if (count >= 33900)
- bias = 32768;
- else if (count >= 1240)
- bias = 1131;
- n += bias;
- if (n < 0 || n >= count)
- return stbtt__new_buf(NULL, 0);
- return stbtt__cff_index_get(idx, n);
+ int count = stbtt__cff_index_count(&idx);
+ int bias = 107;
+ if (count >= 33900)
+ bias = 32768;
+ else if (count >= 1240)
+ bias = 1131;
+ n += bias;
+ if (n < 0 || n >= count)
+ return stbtt__new_buf(NULL, 0);
+ return stbtt__cff_index_get(idx, n);
}
static stbtt__buf stbtt__cid_get_glyph_subrs(const stbtt_fontinfo *info, int glyph_index)
{
- stbtt__buf fdselect = info->fdselect;
- int nranges, start, end, v, fmt, fdselector = -1, i;
-
- stbtt__buf_seek(&fdselect, 0);
- fmt = stbtt__buf_get8(&fdselect);
- if (fmt == 0) {
- // untested
- stbtt__buf_skip(&fdselect, glyph_index);
- fdselector = stbtt__buf_get8(&fdselect);
- }
- else if (fmt == 3) {
- nranges = stbtt__buf_get16(&fdselect);
- start = stbtt__buf_get16(&fdselect);
- for (i = 0; i < nranges; i++) {
- v = stbtt__buf_get8(&fdselect);
- end = stbtt__buf_get16(&fdselect);
- if (glyph_index >= start && glyph_index < end) {
- fdselector = v;
- break;
- }
- start = end;
- }
- }
- if (fdselector == -1) stbtt__new_buf(NULL, 0);
- return stbtt__get_subrs(info->cff, stbtt__cff_index_get(info->fontdicts, fdselector));
+ stbtt__buf fdselect = info->fdselect;
+ int nranges, start, end, v, fmt, fdselector = -1, i;
+
+ stbtt__buf_seek(&fdselect, 0);
+ fmt = stbtt__buf_get8(&fdselect);
+ if (fmt == 0) {
+ // untested
+ stbtt__buf_skip(&fdselect, glyph_index);
+ fdselector = stbtt__buf_get8(&fdselect);
+ } else if (fmt == 3) {
+ nranges = stbtt__buf_get16(&fdselect);
+ start = stbtt__buf_get16(&fdselect);
+ for (i = 0; i < nranges; i++) {
+ v = stbtt__buf_get8(&fdselect);
+ end = stbtt__buf_get16(&fdselect);
+ if (glyph_index >= start && glyph_index < end) {
+ fdselector = v;
+ break;
+ }
+ start = end;
+ }
+ }
+ if (fdselector == -1) stbtt__new_buf(NULL, 0);
+ return stbtt__get_subrs(info->cff, stbtt__cff_index_get(info->fontdicts, fdselector));
}
static int stbtt__run_charstring(const stbtt_fontinfo *info, int glyph_index, stbtt__csctx *c)
{
- int in_header = 1, maskbits = 0, subr_stack_height = 0, sp = 0, v, i, b0;
- int has_subrs = 0, clear_stack;
- float s[48];
- stbtt__buf subr_stack[10], subrs = info->subrs, b;
- float f;
+ int in_header = 1, maskbits = 0, subr_stack_height = 0, sp = 0, v, i, b0;
+ int has_subrs = 0, clear_stack;
+ float s[48];
+ stbtt__buf subr_stack[10], subrs = info->subrs, b;
+ float f;
#define STBTT__CSERR(s) (0)
- // this currently ignores the initial width value, which isn't needed if we have hmtx
- b = stbtt__cff_index_get(info->charstrings, glyph_index);
- while (b.cursor < b.size) {
- i = 0;
- clear_stack = 1;
- b0 = stbtt__buf_get8(&b);
- switch (b0) {
- // @TODO implement hinting
- case 0x13: // hintmask
- case 0x14: // cntrmask
- if (in_header)
- maskbits += (sp / 2); // implicit "vstem"
- in_header = 0;
- stbtt__buf_skip(&b, (maskbits + 7) / 8);
- break;
-
- case 0x01: // hstem
- case 0x03: // vstem
- case 0x12: // hstemhm
- case 0x17: // vstemhm
- maskbits += (sp / 2);
- break;
-
- case 0x15: // rmoveto
- in_header = 0;
- if (sp < 2) return STBTT__CSERR("rmoveto stack");
- stbtt__csctx_rmove_to(c, s[sp - 2], s[sp - 1]);
- break;
- case 0x04: // vmoveto
- in_header = 0;
- if (sp < 1) return STBTT__CSERR("vmoveto stack");
- stbtt__csctx_rmove_to(c, 0, s[sp - 1]);
- break;
- case 0x16: // hmoveto
- in_header = 0;
- if (sp < 1) return STBTT__CSERR("hmoveto stack");
- stbtt__csctx_rmove_to(c, s[sp - 1], 0);
- break;
-
- case 0x05: // rlineto
- if (sp < 2) return STBTT__CSERR("rlineto stack");
- for (; i + 1 < sp; i += 2)
- stbtt__csctx_rline_to(c, s[i], s[i + 1]);
- break;
-
- // hlineto/vlineto and vhcurveto/hvcurveto alternate horizontal and vertical
- // starting from a different place.
-
- case 0x07: // vlineto
- if (sp < 1) return STBTT__CSERR("vlineto stack");
- goto vlineto;
- case 0x06: // hlineto
- if (sp < 1) return STBTT__CSERR("hlineto stack");
- for (;;) {
- if (i >= sp) break;
- stbtt__csctx_rline_to(c, s[i], 0);
- i++;
- vlineto:
- if (i >= sp) break;
- stbtt__csctx_rline_to(c, 0, s[i]);
- i++;
- }
- break;
-
- case 0x1F: // hvcurveto
- if (sp < 4) return STBTT__CSERR("hvcurveto stack");
- goto hvcurveto;
- case 0x1E: // vhcurveto
- if (sp < 4) return STBTT__CSERR("vhcurveto stack");
- for (;;) {
- if (i + 3 >= sp) break;
- stbtt__csctx_rccurve_to(c, 0, s[i], s[i + 1], s[i + 2], s[i + 3], (sp - i == 5) ? s[i + 4] : 0.0f);
- i += 4;
- hvcurveto:
- if (i + 3 >= sp) break;
- stbtt__csctx_rccurve_to(c, s[i], 0, s[i + 1], s[i + 2], (sp - i == 5) ? s[i + 4] : 0.0f, s[i + 3]);
- i += 4;
- }
- break;
-
- case 0x08: // rrcurveto
- if (sp < 6) return STBTT__CSERR("rcurveline stack");
- for (; i + 5 < sp; i += 6)
- stbtt__csctx_rccurve_to(c, s[i], s[i + 1], s[i + 2], s[i + 3], s[i + 4], s[i + 5]);
- break;
-
- case 0x18: // rcurveline
- if (sp < 8) return STBTT__CSERR("rcurveline stack");
- for (; i + 5 < sp - 2; i += 6)
- stbtt__csctx_rccurve_to(c, s[i], s[i + 1], s[i + 2], s[i + 3], s[i + 4], s[i + 5]);
- if (i + 1 >= sp) return STBTT__CSERR("rcurveline stack");
- stbtt__csctx_rline_to(c, s[i], s[i + 1]);
- break;
-
- case 0x19: // rlinecurve
- if (sp < 8) return STBTT__CSERR("rlinecurve stack");
- for (; i + 1 < sp - 6; i += 2)
- stbtt__csctx_rline_to(c, s[i], s[i + 1]);
- if (i + 5 >= sp) return STBTT__CSERR("rlinecurve stack");
- stbtt__csctx_rccurve_to(c, s[i], s[i + 1], s[i + 2], s[i + 3], s[i + 4], s[i + 5]);
- break;
-
- case 0x1A: // vvcurveto
- case 0x1B: // hhcurveto
- if (sp < 4) return STBTT__CSERR("(vv|hh)curveto stack");
+ // this currently ignores the initial width value, which isn't needed if we have hmtx
+ b = stbtt__cff_index_get(info->charstrings, glyph_index);
+ while (b.cursor < b.size) {
+ i = 0;
+ clear_stack = 1;
+ b0 = stbtt__buf_get8(&b);
+ switch (b0) {
+ // @TODO implement hinting
+ case 0x13: // hintmask
+ case 0x14: // cntrmask
+ if (in_header)
+ maskbits += (sp / 2); // implicit "vstem"
+ in_header = 0;
+ stbtt__buf_skip(&b, (maskbits + 7) / 8);
+ break;
+
+ case 0x01: // hstem
+ case 0x03: // vstem
+ case 0x12: // hstemhm
+ case 0x17: // vstemhm
+ maskbits += (sp / 2);
+ break;
+
+ case 0x15: // rmoveto
+ in_header = 0;
+ if (sp < 2) return STBTT__CSERR("rmoveto stack");
+ stbtt__csctx_rmove_to(c, s[sp-2], s[sp-1]);
+ break;
+ case 0x04: // vmoveto
+ in_header = 0;
+ if (sp < 1) return STBTT__CSERR("vmoveto stack");
+ stbtt__csctx_rmove_to(c, 0, s[sp-1]);
+ break;
+ case 0x16: // hmoveto
+ in_header = 0;
+ if (sp < 1) return STBTT__CSERR("hmoveto stack");
+ stbtt__csctx_rmove_to(c, s[sp-1], 0);
+ break;
+
+ case 0x05: // rlineto
+ if (sp < 2) return STBTT__CSERR("rlineto stack");
+ for (; i + 1 < sp; i += 2)
+ stbtt__csctx_rline_to(c, s[i], s[i+1]);
+ break;
+
+ // hlineto/vlineto and vhcurveto/hvcurveto alternate horizontal and vertical
+ // starting from a different place.
+
+ case 0x07: // vlineto
+ if (sp < 1) return STBTT__CSERR("vlineto stack");
+ goto vlineto;
+ case 0x06: // hlineto
+ if (sp < 1) return STBTT__CSERR("hlineto stack");
+ for (;;) {
+ if (i >= sp) break;
+ stbtt__csctx_rline_to(c, s[i], 0);
+ i++;
+ vlineto:
+ if (i >= sp) break;
+ stbtt__csctx_rline_to(c, 0, s[i]);
+ i++;
+ }
+ break;
+
+ case 0x1F: // hvcurveto
+ if (sp < 4) return STBTT__CSERR("hvcurveto stack");
+ goto hvcurveto;
+ case 0x1E: // vhcurveto
+ if (sp < 4) return STBTT__CSERR("vhcurveto stack");
+ for (;;) {
+ if (i + 3 >= sp) break;
+ stbtt__csctx_rccurve_to(c, 0, s[i], s[i+1], s[i+2], s[i+3], (sp - i == 5) ? s[i + 4] : 0.0f);
+ i += 4;
+ hvcurveto:
+ if (i + 3 >= sp) break;
+ stbtt__csctx_rccurve_to(c, s[i], 0, s[i+1], s[i+2], (sp - i == 5) ? s[i+4] : 0.0f, s[i+3]);
+ i += 4;
+ }
+ break;
+
+ case 0x08: // rrcurveto
+ if (sp < 6) return STBTT__CSERR("rcurveline stack");
+ for (; i + 5 < sp; i += 6)
+ stbtt__csctx_rccurve_to(c, s[i], s[i+1], s[i+2], s[i+3], s[i+4], s[i+5]);
+ break;
+
+ case 0x18: // rcurveline
+ if (sp < 8) return STBTT__CSERR("rcurveline stack");
+ for (; i + 5 < sp - 2; i += 6)
+ stbtt__csctx_rccurve_to(c, s[i], s[i+1], s[i+2], s[i+3], s[i+4], s[i+5]);
+ if (i + 1 >= sp) return STBTT__CSERR("rcurveline stack");
+ stbtt__csctx_rline_to(c, s[i], s[i+1]);
+ break;
+
+ case 0x19: // rlinecurve
+ if (sp < 8) return STBTT__CSERR("rlinecurve stack");
+ for (; i + 1 < sp - 6; i += 2)
+ stbtt__csctx_rline_to(c, s[i], s[i+1]);
+ if (i + 5 >= sp) return STBTT__CSERR("rlinecurve stack");
+ stbtt__csctx_rccurve_to(c, s[i], s[i+1], s[i+2], s[i+3], s[i+4], s[i+5]);
+ break;
+
+ case 0x1A: // vvcurveto
+ case 0x1B: // hhcurveto
+ if (sp < 4) return STBTT__CSERR("(vv|hh)curveto stack");
+ f = 0.0;
+ if (sp & 1) { f = s[i]; i++; }
+ for (; i + 3 < sp; i += 4) {
+ if (b0 == 0x1B)
+ stbtt__csctx_rccurve_to(c, s[i], f, s[i+1], s[i+2], s[i+3], 0.0);
+ else
+ stbtt__csctx_rccurve_to(c, f, s[i], s[i+1], s[i+2], 0.0, s[i+3]);
f = 0.0;
- if (sp & 1) { f = s[i]; i++; }
- for (; i + 3 < sp; i += 4) {
- if (b0 == 0x1B)
- stbtt__csctx_rccurve_to(c, s[i], f, s[i + 1], s[i + 2], s[i + 3], 0.0);
- else
- stbtt__csctx_rccurve_to(c, f, s[i], s[i + 1], s[i + 2], 0.0, s[i + 3]);
- f = 0.0;
- }
+ }
+ break;
+
+ case 0x0A: // callsubr
+ if (!has_subrs) {
+ if (info->fdselect.size)
+ subrs = stbtt__cid_get_glyph_subrs(info, glyph_index);
+ has_subrs = 1;
+ }
+ // FALLTHROUGH
+ case 0x1D: // callgsubr
+ if (sp < 1) return STBTT__CSERR("call(g|)subr stack");
+ v = (int) s[--sp];
+ if (subr_stack_height >= 10) return STBTT__CSERR("recursion limit");
+ subr_stack[subr_stack_height++] = b;
+ b = stbtt__get_subr(b0 == 0x0A ? subrs : info->gsubrs, v);
+ if (b.size == 0) return STBTT__CSERR("subr not found");
+ b.cursor = 0;
+ clear_stack = 0;
+ break;
+
+ case 0x0B: // return
+ if (subr_stack_height <= 0) return STBTT__CSERR("return outside subr");
+ b = subr_stack[--subr_stack_height];
+ clear_stack = 0;
+ break;
+
+ case 0x0E: // endchar
+ stbtt__csctx_close_shape(c);
+ return 1;
+
+ case 0x0C: { // two-byte escape
+ float dx1, dx2, dx3, dx4, dx5, dx6, dy1, dy2, dy3, dy4, dy5, dy6;
+ float dx, dy;
+ int b1 = stbtt__buf_get8(&b);
+ switch (b1) {
+ // @TODO These "flex" implementations ignore the flex-depth and resolution,
+ // and always draw beziers.
+ case 0x22: // hflex
+ if (sp < 7) return STBTT__CSERR("hflex stack");
+ dx1 = s[0];
+ dx2 = s[1];
+ dy2 = s[2];
+ dx3 = s[3];
+ dx4 = s[4];
+ dx5 = s[5];
+ dx6 = s[6];
+ stbtt__csctx_rccurve_to(c, dx1, 0, dx2, dy2, dx3, 0);
+ stbtt__csctx_rccurve_to(c, dx4, 0, dx5, -dy2, dx6, 0);
break;
- case 0x0A: // callsubr
- if (!has_subrs) {
- if (info->fdselect.size)
- subrs = stbtt__cid_get_glyph_subrs(info, glyph_index);
- has_subrs = 1;
- }
- // fallthrough
- case 0x1D: // callgsubr
- if (sp < 1) return STBTT__CSERR("call(g|)subr stack");
- v = (int)s[--sp];
- if (subr_stack_height >= 10) return STBTT__CSERR("recursion limit");
- subr_stack[subr_stack_height++] = b;
- b = stbtt__get_subr(b0 == 0x0A ? subrs : info->gsubrs, v);
- if (b.size == 0) return STBTT__CSERR("subr not found");
- b.cursor = 0;
- clear_stack = 0;
+ case 0x23: // flex
+ if (sp < 13) return STBTT__CSERR("flex stack");
+ dx1 = s[0];
+ dy1 = s[1];
+ dx2 = s[2];
+ dy2 = s[3];
+ dx3 = s[4];
+ dy3 = s[5];
+ dx4 = s[6];
+ dy4 = s[7];
+ dx5 = s[8];
+ dy5 = s[9];
+ dx6 = s[10];
+ dy6 = s[11];
+ //fd is s[12]
+ stbtt__csctx_rccurve_to(c, dx1, dy1, dx2, dy2, dx3, dy3);
+ stbtt__csctx_rccurve_to(c, dx4, dy4, dx5, dy5, dx6, dy6);
break;
- case 0x0B: // return
- if (subr_stack_height <= 0) return STBTT__CSERR("return outside subr");
- b = subr_stack[--subr_stack_height];
- clear_stack = 0;
+ case 0x24: // hflex1
+ if (sp < 9) return STBTT__CSERR("hflex1 stack");
+ dx1 = s[0];
+ dy1 = s[1];
+ dx2 = s[2];
+ dy2 = s[3];
+ dx3 = s[4];
+ dx4 = s[5];
+ dx5 = s[6];
+ dy5 = s[7];
+ dx6 = s[8];
+ stbtt__csctx_rccurve_to(c, dx1, dy1, dx2, dy2, dx3, 0);
+ stbtt__csctx_rccurve_to(c, dx4, 0, dx5, dy5, dx6, -(dy1+dy2+dy5));
break;
- case 0x0E: // endchar
- stbtt__csctx_close_shape(c);
- return 1;
-
- case 0x0C: { // two-byte escape
- float dx1, dx2, dx3, dx4, dx5, dx6, dy1, dy2, dy3, dy4, dy5, dy6;
- float dx, dy;
- int b1 = stbtt__buf_get8(&b);
- switch (b1) {
- // @TODO These "flex" implementations ignore the flex-depth and resolution,
- // and always draw beziers.
- case 0x22: // hflex
- if (sp < 7) return STBTT__CSERR("hflex stack");
- dx1 = s[0];
- dx2 = s[1];
- dy2 = s[2];
- dx3 = s[3];
- dx4 = s[4];
- dx5 = s[5];
- dx6 = s[6];
- stbtt__csctx_rccurve_to(c, dx1, 0, dx2, dy2, dx3, 0);
- stbtt__csctx_rccurve_to(c, dx4, 0, dx5, -dy2, dx6, 0);
- break;
-
- case 0x23: // flex
- if (sp < 13) return STBTT__CSERR("flex stack");
- dx1 = s[0];
- dy1 = s[1];
- dx2 = s[2];
- dy2 = s[3];
- dx3 = s[4];
- dy3 = s[5];
- dx4 = s[6];
- dy4 = s[7];
- dx5 = s[8];
- dy5 = s[9];
- dx6 = s[10];
- dy6 = s[11];
- //fd is s[12]
- stbtt__csctx_rccurve_to(c, dx1, dy1, dx2, dy2, dx3, dy3);
- stbtt__csctx_rccurve_to(c, dx4, dy4, dx5, dy5, dx6, dy6);
- break;
-
- case 0x24: // hflex1
- if (sp < 9) return STBTT__CSERR("hflex1 stack");
- dx1 = s[0];
- dy1 = s[1];
- dx2 = s[2];
- dy2 = s[3];
- dx3 = s[4];
- dx4 = s[5];
- dx5 = s[6];
- dy5 = s[7];
- dx6 = s[8];
- stbtt__csctx_rccurve_to(c, dx1, dy1, dx2, dy2, dx3, 0);
- stbtt__csctx_rccurve_to(c, dx4, 0, dx5, dy5, dx6, -(dy1 + dy2 + dy5));
- break;
-
- case 0x25: // flex1
- if (sp < 11) return STBTT__CSERR("flex1 stack");
- dx1 = s[0];
- dy1 = s[1];
- dx2 = s[2];
- dy2 = s[3];
- dx3 = s[4];
- dy3 = s[5];
- dx4 = s[6];
- dy4 = s[7];
- dx5 = s[8];
- dy5 = s[9];
- dx6 = dy6 = s[10];
- dx = dx1 + dx2 + dx3 + dx4 + dx5;
- dy = dy1 + dy2 + dy3 + dy4 + dy5;
- if (STBTT_fabs(dx) > STBTT_fabs(dy))
- dy6 = -dy;
- else
- dx6 = -dx;
- stbtt__csctx_rccurve_to(c, dx1, dy1, dx2, dy2, dx3, dy3);
- stbtt__csctx_rccurve_to(c, dx4, dy4, dx5, dy5, dx6, dy6);
- break;
-
- default:
- return STBTT__CSERR("unimplemented");
- }
- } break;
-
- default:
- if (b0 != 255 && b0 != 28 && (b0 < 32 || b0 > 254))
- return STBTT__CSERR("reserved operator");
-
- // push immediate
- if (b0 == 255) {
- f = (float)(stbtt_int32)stbtt__buf_get32(&b) / 0x10000;
- }
- else {
- stbtt__buf_skip(&b, -1);
- f = (float)(stbtt_int16)stbtt__cff_int(&b);
- }
- if (sp >= 48) return STBTT__CSERR("push stack overflow");
- s[sp++] = f;
- clear_stack = 0;
+ case 0x25: // flex1
+ if (sp < 11) return STBTT__CSERR("flex1 stack");
+ dx1 = s[0];
+ dy1 = s[1];
+ dx2 = s[2];
+ dy2 = s[3];
+ dx3 = s[4];
+ dy3 = s[5];
+ dx4 = s[6];
+ dy4 = s[7];
+ dx5 = s[8];
+ dy5 = s[9];
+ dx6 = dy6 = s[10];
+ dx = dx1+dx2+dx3+dx4+dx5;
+ dy = dy1+dy2+dy3+dy4+dy5;
+ if (STBTT_fabs(dx) > STBTT_fabs(dy))
+ dy6 = -dy;
+ else
+ dx6 = -dx;
+ stbtt__csctx_rccurve_to(c, dx1, dy1, dx2, dy2, dx3, dy3);
+ stbtt__csctx_rccurve_to(c, dx4, dy4, dx5, dy5, dx6, dy6);
break;
- }
- if (clear_stack) sp = 0;
- }
- return STBTT__CSERR("no endchar");
+
+ default:
+ return STBTT__CSERR("unimplemented");
+ }
+ } break;
+
+ default:
+ if (b0 != 255 && b0 != 28 && b0 < 32)
+ return STBTT__CSERR("reserved operator");
+
+ // push immediate
+ if (b0 == 255) {
+ f = (float)(stbtt_int32)stbtt__buf_get32(&b) / 0x10000;
+ } else {
+ stbtt__buf_skip(&b, -1);
+ f = (float)(stbtt_int16)stbtt__cff_int(&b);
+ }
+ if (sp >= 48) return STBTT__CSERR("push stack overflow");
+ s[sp++] = f;
+ clear_stack = 0;
+ break;
+ }
+ if (clear_stack) sp = 0;
+ }
+ return STBTT__CSERR("no endchar");
#undef STBTT__CSERR
}
static int stbtt__GetGlyphShapeT2(const stbtt_fontinfo *info, int glyph_index, stbtt_vertex **pvertices)
{
- // runs the charstring twice, once to count and once to output (to avoid realloc)
- stbtt__csctx count_ctx = STBTT__CSCTX_INIT(1);
- stbtt__csctx output_ctx = STBTT__CSCTX_INIT(0);
- if (stbtt__run_charstring(info, glyph_index, &count_ctx)) {
- *pvertices = (stbtt_vertex*)STBTT_malloc(count_ctx.num_vertices * sizeof(stbtt_vertex), info->userdata);
- output_ctx.pvertices = *pvertices;
- if (stbtt__run_charstring(info, glyph_index, &output_ctx)) {
- STBTT_assert(output_ctx.num_vertices == count_ctx.num_vertices);
- return output_ctx.num_vertices;
- }
- }
- *pvertices = NULL;
- return 0;
+ // runs the charstring twice, once to count and once to output (to avoid realloc)
+ stbtt__csctx count_ctx = STBTT__CSCTX_INIT(1);
+ stbtt__csctx output_ctx = STBTT__CSCTX_INIT(0);
+ if (stbtt__run_charstring(info, glyph_index, &count_ctx)) {
+ *pvertices = (stbtt_vertex*)STBTT_malloc(count_ctx.num_vertices*sizeof(stbtt_vertex), info->userdata);
+ output_ctx.pvertices = *pvertices;
+ if (stbtt__run_charstring(info, glyph_index, &output_ctx)) {
+ STBTT_assert(output_ctx.num_vertices == count_ctx.num_vertices);
+ return output_ctx.num_vertices;
+ }
+ }
+ *pvertices = NULL;
+ return 0;
}
static int stbtt__GetGlyphInfoT2(const stbtt_fontinfo *info, int glyph_index, int *x0, int *y0, int *x1, int *y1)
{
- stbtt__csctx c = STBTT__CSCTX_INIT(1);
- int r = stbtt__run_charstring(info, glyph_index, &c);
- if (x0) *x0 = r ? c.min_x : 0;
- if (y0) *y0 = r ? c.min_y : 0;
- if (x1) *x1 = r ? c.max_x : 0;
- if (y1) *y1 = r ? c.max_y : 0;
- return r ? c.num_vertices : 0;
+ stbtt__csctx c = STBTT__CSCTX_INIT(1);
+ int r = stbtt__run_charstring(info, glyph_index, &c);
+ if (x0) *x0 = r ? c.min_x : 0;
+ if (y0) *y0 = r ? c.min_y : 0;
+ if (x1) *x1 = r ? c.max_x : 0;
+ if (y1) *y1 = r ? c.max_y : 0;
+ return r ? c.num_vertices : 0;
}
STBTT_DEF int stbtt_GetGlyphShape(const stbtt_fontinfo *info, int glyph_index, stbtt_vertex **pvertices)
{
- if (!info->cff.size)
- return stbtt__GetGlyphShapeTT(info, glyph_index, pvertices);
- else
- return stbtt__GetGlyphShapeT2(info, glyph_index, pvertices);
+ if (!info->cff.size)
+ return stbtt__GetGlyphShapeTT(info, glyph_index, pvertices);
+ else
+ return stbtt__GetGlyphShapeT2(info, glyph_index, pvertices);
}
STBTT_DEF void stbtt_GetGlyphHMetrics(const stbtt_fontinfo *info, int glyph_index, int *advanceWidth, int *leftSideBearing)
{
- stbtt_uint16 numOfLongHorMetrics = ttUSHORT(info->data + info->hhea + 34);
- if (glyph_index < numOfLongHorMetrics) {
- if (advanceWidth) *advanceWidth = ttSHORT(info->data + info->hmtx + 4 * glyph_index);
- if (leftSideBearing) *leftSideBearing = ttSHORT(info->data + info->hmtx + 4 * glyph_index + 2);
- }
- else {
- if (advanceWidth) *advanceWidth = ttSHORT(info->data + info->hmtx + 4 * (numOfLongHorMetrics - 1));
- if (leftSideBearing) *leftSideBearing = ttSHORT(info->data + info->hmtx + 4 * numOfLongHorMetrics + 2 * (glyph_index - numOfLongHorMetrics));
- }
-}
-
-static int stbtt__GetGlyphKernInfoAdvance(const stbtt_fontinfo *info, int glyph1, int glyph2)
-{
- stbtt_uint8 *data = info->data + info->kern;
- stbtt_uint32 needle, straw;
- int l, r, m;
-
- // we only look at the first table. it must be 'horizontal' and format 0.
- if (!info->kern)
- return 0;
- if (ttUSHORT(data + 2) < 1) // number of tables, need at least 1
- return 0;
- if (ttUSHORT(data + 8) != 1) // horizontal flag must be set in format
- return 0;
-
- l = 0;
- r = ttUSHORT(data + 10) - 1;
- needle = glyph1 << 16 | glyph2;
- while (l <= r) {
- m = (l + r) >> 1;
- straw = ttULONG(data + 18 + (m * 6)); // note: unaligned read
- if (needle < straw)
- r = m - 1;
- else if (needle > straw)
- l = m + 1;
- else
- return ttSHORT(data + 22 + (m * 6));
- }
- return 0;
-}
-
-static stbtt_int32 stbtt__GetCoverageIndex(stbtt_uint8 *coverageTable, int glyph)
-{
- stbtt_uint16 coverageFormat = ttUSHORT(coverageTable);
- switch (coverageFormat) {
- case 1: {
- stbtt_uint16 glyphCount = ttUSHORT(coverageTable + 2);
-
- // Binary search.
- stbtt_int32 l = 0, r = glyphCount - 1, m;
- int straw, needle = glyph;
- while (l <= r) {
+ stbtt_uint16 numOfLongHorMetrics = ttUSHORT(info->data+info->hhea + 34);
+ if (glyph_index < numOfLongHorMetrics) {
+ if (advanceWidth) *advanceWidth = ttSHORT(info->data + info->hmtx + 4*glyph_index);
+ if (leftSideBearing) *leftSideBearing = ttSHORT(info->data + info->hmtx + 4*glyph_index + 2);
+ } else {
+ if (advanceWidth) *advanceWidth = ttSHORT(info->data + info->hmtx + 4*(numOfLongHorMetrics-1));
+ if (leftSideBearing) *leftSideBearing = ttSHORT(info->data + info->hmtx + 4*numOfLongHorMetrics + 2*(glyph_index - numOfLongHorMetrics));
+ }
+}
+
+STBTT_DEF int stbtt_GetKerningTableLength(const stbtt_fontinfo *info)
+{
+ stbtt_uint8 *data = info->data + info->kern;
+
+ // we only look at the first table. it must be 'horizontal' and format 0.
+ if (!info->kern)
+ return 0;
+ if (ttUSHORT(data+2) < 1) // number of tables, need at least 1
+ return 0;
+ if (ttUSHORT(data+8) != 1) // horizontal flag must be set in format
+ return 0;
+
+ return ttUSHORT(data+10);
+}
+
+STBTT_DEF int stbtt_GetKerningTable(const stbtt_fontinfo *info, stbtt_kerningentry* table, int table_length)
+{
+ stbtt_uint8 *data = info->data + info->kern;
+ int k, length;
+
+ // we only look at the first table. it must be 'horizontal' and format 0.
+ if (!info->kern)
+ return 0;
+ if (ttUSHORT(data+2) < 1) // number of tables, need at least 1
+ return 0;
+ if (ttUSHORT(data+8) != 1) // horizontal flag must be set in format
+ return 0;
+
+ length = ttUSHORT(data+10);
+ if (table_length < length)
+ length = table_length;
+
+ for (k = 0; k < length; k++)
+ {
+ table[k].glyph1 = ttUSHORT(data+18+(k*6));
+ table[k].glyph2 = ttUSHORT(data+20+(k*6));
+ table[k].advance = ttSHORT(data+22+(k*6));
+ }
+
+ return length;
+}
+
+static int stbtt__GetGlyphKernInfoAdvance(const stbtt_fontinfo *info, int glyph1, int glyph2)
+{
+ stbtt_uint8 *data = info->data + info->kern;
+ stbtt_uint32 needle, straw;
+ int l, r, m;
+
+ // we only look at the first table. it must be 'horizontal' and format 0.
+ if (!info->kern)
+ return 0;
+ if (ttUSHORT(data+2) < 1) // number of tables, need at least 1
+ return 0;
+ if (ttUSHORT(data+8) != 1) // horizontal flag must be set in format
+ return 0;
+
+ l = 0;
+ r = ttUSHORT(data+10) - 1;
+ needle = glyph1 << 16 | glyph2;
+ while (l <= r) {
+ m = (l + r) >> 1;
+ straw = ttULONG(data+18+(m*6)); // note: unaligned read
+ if (needle < straw)
+ r = m - 1;
+ else if (needle > straw)
+ l = m + 1;
+ else
+ return ttSHORT(data+22+(m*6));
+ }
+ return 0;
+}
+
+static stbtt_int32 stbtt__GetCoverageIndex(stbtt_uint8 *coverageTable, int glyph)
+{
+ stbtt_uint16 coverageFormat = ttUSHORT(coverageTable);
+ switch (coverageFormat) {
+ case 1: {
+ stbtt_uint16 glyphCount = ttUSHORT(coverageTable + 2);
+
+ // Binary search.
+ stbtt_int32 l=0, r=glyphCount-1, m;
+ int straw, needle=glyph;
+ while (l <= r) {
stbtt_uint8 *glyphArray = coverageTable + 4;
stbtt_uint16 glyphID;
m = (l + r) >> 1;
glyphID = ttUSHORT(glyphArray + 2 * m);
straw = glyphID;
if (needle < straw)
- r = m - 1;
+ r = m - 1;
else if (needle > straw)
- l = m + 1;
+ l = m + 1;
else {
- return m;
+ return m;
}
- }
- } break;
-
- case 2: {
- stbtt_uint16 rangeCount = ttUSHORT(coverageTable + 2);
- stbtt_uint8 *rangeArray = coverageTable + 4;
-
- // Binary search.
- stbtt_int32 l = 0, r = rangeCount - 1, m;
- int strawStart, strawEnd, needle = glyph;
- while (l <= r) {
+ }
+ break;
+ }
+
+ case 2: {
+ stbtt_uint16 rangeCount = ttUSHORT(coverageTable + 2);
+ stbtt_uint8 *rangeArray = coverageTable + 4;
+
+ // Binary search.
+ stbtt_int32 l=0, r=rangeCount-1, m;
+ int strawStart, strawEnd, needle=glyph;
+ while (l <= r) {
stbtt_uint8 *rangeRecord;
m = (l + r) >> 1;
rangeRecord = rangeArray + 6 * m;
strawStart = ttUSHORT(rangeRecord);
strawEnd = ttUSHORT(rangeRecord + 2);
if (needle < strawStart)
- r = m - 1;
+ r = m - 1;
else if (needle > strawEnd)
- l = m + 1;
+ l = m + 1;
else {
- stbtt_uint16 startCoverageIndex = ttUSHORT(rangeRecord + 4);
- return startCoverageIndex + glyph - strawStart;
+ stbtt_uint16 startCoverageIndex = ttUSHORT(rangeRecord + 4);
+ return startCoverageIndex + glyph - strawStart;
}
- }
- } break;
+ }
+ break;
+ }
- default: {
- // There are no other cases.
- STBTT_assert(0);
- } break;
- }
+ default: return -1; // unsupported
+ }
- return -1;
+ return -1;
}
static stbtt_int32 stbtt__GetGlyphClass(stbtt_uint8 *classDefTable, int glyph)
{
- stbtt_uint16 classDefFormat = ttUSHORT(classDefTable);
- switch (classDefFormat)
- {
- case 1: {
- stbtt_uint16 startGlyphID = ttUSHORT(classDefTable + 2);
- stbtt_uint16 glyphCount = ttUSHORT(classDefTable + 4);
- stbtt_uint8 *classDef1ValueArray = classDefTable + 6;
+ stbtt_uint16 classDefFormat = ttUSHORT(classDefTable);
+ switch (classDefFormat)
+ {
+ case 1: {
+ stbtt_uint16 startGlyphID = ttUSHORT(classDefTable + 2);
+ stbtt_uint16 glyphCount = ttUSHORT(classDefTable + 4);
+ stbtt_uint8 *classDef1ValueArray = classDefTable + 6;
- if (glyph >= startGlyphID && glyph < startGlyphID + glyphCount)
+ if (glyph >= startGlyphID && glyph < startGlyphID + glyphCount)
return (stbtt_int32)ttUSHORT(classDef1ValueArray + 2 * (glyph - startGlyphID));
+ break;
+ }
- classDefTable = classDef1ValueArray + 2 * glyphCount;
- } break;
+ case 2: {
+ stbtt_uint16 classRangeCount = ttUSHORT(classDefTable + 2);
+ stbtt_uint8 *classRangeRecords = classDefTable + 4;
- case 2: {
- stbtt_uint16 classRangeCount = ttUSHORT(classDefTable + 2);
- stbtt_uint8 *classRangeRecords = classDefTable + 4;
-
- // Binary search.
- stbtt_int32 l = 0, r = classRangeCount - 1, m;
- int strawStart, strawEnd, needle = glyph;
- while (l <= r) {
+ // Binary search.
+ stbtt_int32 l=0, r=classRangeCount-1, m;
+ int strawStart, strawEnd, needle=glyph;
+ while (l <= r) {
stbtt_uint8 *classRangeRecord;
m = (l + r) >> 1;
classRangeRecord = classRangeRecords + 6 * m;
strawStart = ttUSHORT(classRangeRecord);
strawEnd = ttUSHORT(classRangeRecord + 2);
if (needle < strawStart)
- r = m - 1;
+ r = m - 1;
else if (needle > strawEnd)
- l = m + 1;
+ l = m + 1;
else
- return (stbtt_int32)ttUSHORT(classRangeRecord + 4);
- }
-
- classDefTable = classRangeRecords + 6 * classRangeCount;
- } break;
+ return (stbtt_int32)ttUSHORT(classRangeRecord + 4);
+ }
+ break;
+ }
- default: {
- // There are no other cases.
- STBTT_assert(0);
- } break;
- }
+ default:
+ return -1; // Unsupported definition type, return an error.
+ }
- return -1;
+ // "All glyphs not assigned to a class fall into class 0". (OpenType spec)
+ return 0;
}
// Define to STBTT_assert(x) if you want to break on unimplemented formats.
#define STBTT_GPOS_TODO_assert(x)
-static stbtt_int32 stbtt__GetGlyphGPOSInfoAdvance(const stbtt_fontinfo *info, int glyph1, int glyph2)
-{
- stbtt_uint16 lookupListOffset;
- stbtt_uint8 *lookupList;
- stbtt_uint16 lookupCount;
- stbtt_uint8 *data;
- stbtt_int32 i;
-
- if (!info->gpos) return 0;
-
- data = info->data + info->gpos;
-
- if (ttUSHORT(data + 0) != 1) return 0; // Major version 1
- if (ttUSHORT(data + 2) != 0) return 0; // Minor version 0
-
- lookupListOffset = ttUSHORT(data + 8);
- lookupList = data + lookupListOffset;
- lookupCount = ttUSHORT(lookupList);
-
- for (i = 0; i<lookupCount; ++i) {
- stbtt_uint16 lookupOffset = ttUSHORT(lookupList + 2 + 2 * i);
- stbtt_uint8 *lookupTable = lookupList + lookupOffset;
-
- stbtt_uint16 lookupType = ttUSHORT(lookupTable);
- stbtt_uint16 subTableCount = ttUSHORT(lookupTable + 4);
- stbtt_uint8 *subTableOffsets = lookupTable + 6;
- switch (lookupType) {
- case 2: { // Pair Adjustment Positioning Subtable
- stbtt_int32 sti;
- for (sti = 0; sti<subTableCount; sti++) {
- stbtt_uint16 subtableOffset = ttUSHORT(subTableOffsets + 2 * sti);
- stbtt_uint8 *table = lookupTable + subtableOffset;
- stbtt_uint16 posFormat = ttUSHORT(table);
- stbtt_uint16 coverageOffset = ttUSHORT(table + 2);
- stbtt_int32 coverageIndex = stbtt__GetCoverageIndex(table + coverageOffset, glyph1);
- if (coverageIndex == -1) continue;
-
- switch (posFormat) {
- case 1: {
- stbtt_int32 l, r, m;
- int straw, needle;
- stbtt_uint16 valueFormat1 = ttUSHORT(table + 4);
- stbtt_uint16 valueFormat2 = ttUSHORT(table + 6);
- stbtt_int32 valueRecordPairSizeInBytes = 2;
- stbtt_uint16 pairSetCount = ttUSHORT(table + 8);
- stbtt_uint16 pairPosOffset = ttUSHORT(table + 10 + 2 * coverageIndex);
- stbtt_uint8 *pairValueTable = table + pairPosOffset;
- stbtt_uint16 pairValueCount = ttUSHORT(pairValueTable);
- stbtt_uint8 *pairValueArray = pairValueTable + 2;
- // TODO: Support more formats.
- STBTT_GPOS_TODO_assert(valueFormat1 == 4);
- if (valueFormat1 != 4) return 0;
- STBTT_GPOS_TODO_assert(valueFormat2 == 0);
- if (valueFormat2 != 0) return 0;
-
- STBTT_assert(coverageIndex < pairSetCount);
-
- needle = glyph2;
- r = pairValueCount - 1;
- l = 0;
-
- // Binary search.
- while (l <= r) {
- stbtt_uint16 secondGlyph;
- stbtt_uint8 *pairValue;
- m = (l + r) >> 1;
- pairValue = pairValueArray + (2 + valueRecordPairSizeInBytes) * m;
- secondGlyph = ttUSHORT(pairValue);
- straw = secondGlyph;
- if (needle < straw)
- r = m - 1;
- else if (needle > straw)
- l = m + 1;
- else {
- stbtt_int16 xAdvance = ttSHORT(pairValue + 2);
- return xAdvance;
- }
- }
- } break;
-
- case 2: {
- stbtt_uint16 valueFormat1 = ttUSHORT(table + 4);
- stbtt_uint16 valueFormat2 = ttUSHORT(table + 6);
-
- stbtt_uint16 classDef1Offset = ttUSHORT(table + 8);
- stbtt_uint16 classDef2Offset = ttUSHORT(table + 10);
- int glyph1class = stbtt__GetGlyphClass(table + classDef1Offset, glyph1);
- int glyph2class = stbtt__GetGlyphClass(table + classDef2Offset, glyph2);
-
- stbtt_uint16 class1Count = ttUSHORT(table + 12);
- stbtt_uint16 class2Count = ttUSHORT(table + 14);
- STBTT_assert(glyph1class < class1Count);
- STBTT_assert(glyph2class < class2Count);
-
- // TODO: Support more formats.
- STBTT_GPOS_TODO_assert(valueFormat1 == 4);
- if (valueFormat1 != 4) return 0;
- STBTT_GPOS_TODO_assert(valueFormat2 == 0);
- if (valueFormat2 != 0) return 0;
-
- if (glyph1class >= 0 && glyph1class < class1Count && glyph2class >= 0 && glyph2class < class2Count) {
- stbtt_uint8 *class1Records = table + 16;
- stbtt_uint8 *class2Records = class1Records + 2 * (glyph1class * class2Count);
- stbtt_int16 xAdvance = ttSHORT(class2Records + 2 * glyph2class);
+static stbtt_int32 stbtt__GetGlyphGPOSInfoAdvance(const stbtt_fontinfo *info, int glyph1, int glyph2)
+{
+ stbtt_uint16 lookupListOffset;
+ stbtt_uint8 *lookupList;
+ stbtt_uint16 lookupCount;
+ stbtt_uint8 *data;
+ stbtt_int32 i, sti;
+
+ if (!info->gpos) return 0;
+
+ data = info->data + info->gpos;
+
+ if (ttUSHORT(data+0) != 1) return 0; // Major version 1
+ if (ttUSHORT(data+2) != 0) return 0; // Minor version 0
+
+ lookupListOffset = ttUSHORT(data+8);
+ lookupList = data + lookupListOffset;
+ lookupCount = ttUSHORT(lookupList);
+
+ for (i=0; i<lookupCount; ++i) {
+ stbtt_uint16 lookupOffset = ttUSHORT(lookupList + 2 + 2 * i);
+ stbtt_uint8 *lookupTable = lookupList + lookupOffset;
+
+ stbtt_uint16 lookupType = ttUSHORT(lookupTable);
+ stbtt_uint16 subTableCount = ttUSHORT(lookupTable + 4);
+ stbtt_uint8 *subTableOffsets = lookupTable + 6;
+ if (lookupType != 2) // Pair Adjustment Positioning Subtable
+ continue;
+
+ for (sti=0; sti<subTableCount; sti++) {
+ stbtt_uint16 subtableOffset = ttUSHORT(subTableOffsets + 2 * sti);
+ stbtt_uint8 *table = lookupTable + subtableOffset;
+ stbtt_uint16 posFormat = ttUSHORT(table);
+ stbtt_uint16 coverageOffset = ttUSHORT(table + 2);
+ stbtt_int32 coverageIndex = stbtt__GetCoverageIndex(table + coverageOffset, glyph1);
+ if (coverageIndex == -1) continue;
+
+ switch (posFormat) {
+ case 1: {
+ stbtt_int32 l, r, m;
+ int straw, needle;
+ stbtt_uint16 valueFormat1 = ttUSHORT(table + 4);
+ stbtt_uint16 valueFormat2 = ttUSHORT(table + 6);
+ if (valueFormat1 == 4 && valueFormat2 == 0) { // Support more formats?
+ stbtt_int32 valueRecordPairSizeInBytes = 2;
+ stbtt_uint16 pairSetCount = ttUSHORT(table + 8);
+ stbtt_uint16 pairPosOffset = ttUSHORT(table + 10 + 2 * coverageIndex);
+ stbtt_uint8 *pairValueTable = table + pairPosOffset;
+ stbtt_uint16 pairValueCount = ttUSHORT(pairValueTable);
+ stbtt_uint8 *pairValueArray = pairValueTable + 2;
+
+ if (coverageIndex >= pairSetCount) return 0;
+
+ needle=glyph2;
+ r=pairValueCount-1;
+ l=0;
+
+ // Binary search.
+ while (l <= r) {
+ stbtt_uint16 secondGlyph;
+ stbtt_uint8 *pairValue;
+ m = (l + r) >> 1;
+ pairValue = pairValueArray + (2 + valueRecordPairSizeInBytes) * m;
+ secondGlyph = ttUSHORT(pairValue);
+ straw = secondGlyph;
+ if (needle < straw)
+ r = m - 1;
+ else if (needle > straw)
+ l = m + 1;
+ else {
+ stbtt_int16 xAdvance = ttSHORT(pairValue + 2);
return xAdvance;
- }
- } break;
-
- default: {
- // There are no other cases.
- STBTT_assert(0);
- break;
- };
- }
+ }
+ }
+ } else
+ return 0;
+ break;
}
- break;
- };
- default:
- // TODO: Implement other stuff.
- break;
- }
- }
+ case 2: {
+ stbtt_uint16 valueFormat1 = ttUSHORT(table + 4);
+ stbtt_uint16 valueFormat2 = ttUSHORT(table + 6);
+ if (valueFormat1 == 4 && valueFormat2 == 0) { // Support more formats?
+ stbtt_uint16 classDef1Offset = ttUSHORT(table + 8);
+ stbtt_uint16 classDef2Offset = ttUSHORT(table + 10);
+ int glyph1class = stbtt__GetGlyphClass(table + classDef1Offset, glyph1);
+ int glyph2class = stbtt__GetGlyphClass(table + classDef2Offset, glyph2);
+
+ stbtt_uint16 class1Count = ttUSHORT(table + 12);
+ stbtt_uint16 class2Count = ttUSHORT(table + 14);
+ stbtt_uint8 *class1Records, *class2Records;
+ stbtt_int16 xAdvance;
+
+ if (glyph1class < 0 || glyph1class >= class1Count) return 0; // malformed
+ if (glyph2class < 0 || glyph2class >= class2Count) return 0; // malformed
+
+ class1Records = table + 16;
+ class2Records = class1Records + 2 * (glyph1class * class2Count);
+ xAdvance = ttSHORT(class2Records + 2 * glyph2class);
+ return xAdvance;
+ } else
+ return 0;
+ break;
+ }
+
+ default:
+ return 0; // Unsupported position format
+ }
+ }
+ }
- return 0;
+ return 0;
}
STBTT_DEF int stbtt_GetGlyphKernAdvance(const stbtt_fontinfo *info, int g1, int g2)
{
- int xAdvance = 0;
-
- if (info->gpos)
- xAdvance += stbtt__GetGlyphGPOSInfoAdvance(info, g1, g2);
+ int xAdvance = 0;
- if (info->kern)
- xAdvance += stbtt__GetGlyphKernInfoAdvance(info, g1, g2);
+ if (info->gpos)
+ xAdvance += stbtt__GetGlyphGPOSInfoAdvance(info, g1, g2);
+ else if (info->kern)
+ xAdvance += stbtt__GetGlyphKernInfoAdvance(info, g1, g2);
- return xAdvance;
+ return xAdvance;
}
STBTT_DEF int stbtt_GetCodepointKernAdvance(const stbtt_fontinfo *info, int ch1, int ch2)
{
- if (!info->kern && !info->gpos) // if no kerning table, don't waste time looking up both codepoint->glyphs
- return 0;
- return stbtt_GetGlyphKernAdvance(info, stbtt_FindGlyphIndex(info, ch1), stbtt_FindGlyphIndex(info, ch2));
+ if (!info->kern && !info->gpos) // if no kerning table, don't waste time looking up both codepoint->glyphs
+ return 0;
+ return stbtt_GetGlyphKernAdvance(info, stbtt_FindGlyphIndex(info,ch1), stbtt_FindGlyphIndex(info,ch2));
}
STBTT_DEF void stbtt_GetCodepointHMetrics(const stbtt_fontinfo *info, int codepoint, int *advanceWidth, int *leftSideBearing)
{
- stbtt_GetGlyphHMetrics(info, stbtt_FindGlyphIndex(info, codepoint), advanceWidth, leftSideBearing);
+ stbtt_GetGlyphHMetrics(info, stbtt_FindGlyphIndex(info,codepoint), advanceWidth, leftSideBearing);
}
STBTT_DEF void stbtt_GetFontVMetrics(const stbtt_fontinfo *info, int *ascent, int *descent, int *lineGap)
{
- if (ascent) *ascent = ttSHORT(info->data + info->hhea + 4);
- if (descent) *descent = ttSHORT(info->data + info->hhea + 6);
- if (lineGap) *lineGap = ttSHORT(info->data + info->hhea + 8);
+ if (ascent ) *ascent = ttSHORT(info->data+info->hhea + 4);
+ if (descent) *descent = ttSHORT(info->data+info->hhea + 6);
+ if (lineGap) *lineGap = ttSHORT(info->data+info->hhea + 8);
}
STBTT_DEF int stbtt_GetFontVMetricsOS2(const stbtt_fontinfo *info, int *typoAscent, int *typoDescent, int *typoLineGap)
{
- int tab = stbtt__find_table(info->data, info->fontstart, "OS/2");
- if (!tab)
- return 0;
- if (typoAscent) *typoAscent = ttSHORT(info->data + tab + 68);
- if (typoDescent) *typoDescent = ttSHORT(info->data + tab + 70);
- if (typoLineGap) *typoLineGap = ttSHORT(info->data + tab + 72);
- return 1;
+ int tab = stbtt__find_table(info->data, info->fontstart, "OS/2");
+ if (!tab)
+ return 0;
+ if (typoAscent ) *typoAscent = ttSHORT(info->data+tab + 68);
+ if (typoDescent) *typoDescent = ttSHORT(info->data+tab + 70);
+ if (typoLineGap) *typoLineGap = ttSHORT(info->data+tab + 72);
+ return 1;
}
STBTT_DEF void stbtt_GetFontBoundingBox(const stbtt_fontinfo *info, int *x0, int *y0, int *x1, int *y1)
{
- *x0 = ttSHORT(info->data + info->head + 36);
- *y0 = ttSHORT(info->data + info->head + 38);
- *x1 = ttSHORT(info->data + info->head + 40);
- *y1 = ttSHORT(info->data + info->head + 42);
+ *x0 = ttSHORT(info->data + info->head + 36);
+ *y0 = ttSHORT(info->data + info->head + 38);
+ *x1 = ttSHORT(info->data + info->head + 40);
+ *y1 = ttSHORT(info->data + info->head + 42);
}
STBTT_DEF float stbtt_ScaleForPixelHeight(const stbtt_fontinfo *info, float height)
{
- int fheight = ttSHORT(info->data + info->hhea + 4) - ttSHORT(info->data + info->hhea + 6);
- return (float)height / fheight;
+ int fheight = ttSHORT(info->data + info->hhea + 4) - ttSHORT(info->data + info->hhea + 6);
+ return (float) height / fheight;
}
STBTT_DEF float stbtt_ScaleForMappingEmToPixels(const stbtt_fontinfo *info, float pixels)
{
- int unitsPerEm = ttUSHORT(info->data + info->head + 18);
- return pixels / unitsPerEm;
+ int unitsPerEm = ttUSHORT(info->data + info->head + 18);
+ return pixels / unitsPerEm;
}
STBTT_DEF void stbtt_FreeShape(const stbtt_fontinfo *info, stbtt_vertex *v)
{
- STBTT_free(v, info->userdata);
+ STBTT_free(v, info->userdata);
+}
+
+STBTT_DEF stbtt_uint8 *stbtt_FindSVGDoc(const stbtt_fontinfo *info, int gl)
+{
+ int i;
+ stbtt_uint8 *data = info->data;
+ stbtt_uint8 *svg_doc_list = data + stbtt__get_svg((stbtt_fontinfo *) info);
+
+ int numEntries = ttUSHORT(svg_doc_list);
+ stbtt_uint8 *svg_docs = svg_doc_list + 2;
+
+ for(i=0; i<numEntries; i++) {
+ stbtt_uint8 *svg_doc = svg_docs + (12 * i);
+ if ((gl >= ttUSHORT(svg_doc)) && (gl <= ttUSHORT(svg_doc + 2)))
+ return svg_doc;
+ }
+ return 0;
+}
+
+STBTT_DEF int stbtt_GetGlyphSVG(const stbtt_fontinfo *info, int gl, const char **svg)
+{
+ stbtt_uint8 *data = info->data;
+ stbtt_uint8 *svg_doc;
+
+ if (info->svg == 0)
+ return 0;
+
+ svg_doc = stbtt_FindSVGDoc(info, gl);
+ if (svg_doc != NULL) {
+ *svg = (char *) data + info->svg + ttULONG(svg_doc + 4);
+ return ttULONG(svg_doc + 8);
+ } else {
+ return 0;
+ }
+}
+
+STBTT_DEF int stbtt_GetCodepointSVG(const stbtt_fontinfo *info, int unicode_codepoint, const char **svg)
+{
+ return stbtt_GetGlyphSVG(info, stbtt_FindGlyphIndex(info, unicode_codepoint), svg);
}
//////////////////////////////////////////////////////////////////////////////
@@ -2633,38 +2718,37 @@ STBTT_DEF void stbtt_FreeShape(const stbtt_fontinfo *info, stbtt_vertex *v)
// antialiasing software rasterizer
//
-STBTT_DEF void stbtt_GetGlyphBitmapBoxSubpixel(const stbtt_fontinfo *font, int glyph, float scale_x, float scale_y, float shift_x, float shift_y, int *ix0, int *iy0, int *ix1, int *iy1)
-{
- int x0 = 0, y0 = 0, x1, y1; // =0 suppresses compiler warning
- if (!stbtt_GetGlyphBox(font, glyph, &x0, &y0, &x1, &y1)) {
- // e.g. space character
- if (ix0) *ix0 = 0;
- if (iy0) *iy0 = 0;
- if (ix1) *ix1 = 0;
- if (iy1) *iy1 = 0;
- }
- else {
- // move to integral bboxes (treating pixels as little squares, what pixels get touched)?
- if (ix0) *ix0 = STBTT_ifloor(x0 * scale_x + shift_x);
- if (iy0) *iy0 = STBTT_ifloor(-y1 * scale_y + shift_y);
- if (ix1) *ix1 = STBTT_iceil(x1 * scale_x + shift_x);
- if (iy1) *iy1 = STBTT_iceil(-y0 * scale_y + shift_y);
- }
+STBTT_DEF void stbtt_GetGlyphBitmapBoxSubpixel(const stbtt_fontinfo *font, int glyph, float scale_x, float scale_y,float shift_x, float shift_y, int *ix0, int *iy0, int *ix1, int *iy1)
+{
+ int x0=0,y0=0,x1,y1; // =0 suppresses compiler warning
+ if (!stbtt_GetGlyphBox(font, glyph, &x0,&y0,&x1,&y1)) {
+ // e.g. space character
+ if (ix0) *ix0 = 0;
+ if (iy0) *iy0 = 0;
+ if (ix1) *ix1 = 0;
+ if (iy1) *iy1 = 0;
+ } else {
+ // move to integral bboxes (treating pixels as little squares, what pixels get touched)?
+ if (ix0) *ix0 = STBTT_ifloor( x0 * scale_x + shift_x);
+ if (iy0) *iy0 = STBTT_ifloor(-y1 * scale_y + shift_y);
+ if (ix1) *ix1 = STBTT_iceil ( x1 * scale_x + shift_x);
+ if (iy1) *iy1 = STBTT_iceil (-y0 * scale_y + shift_y);
+ }
}
STBTT_DEF void stbtt_GetGlyphBitmapBox(const stbtt_fontinfo *font, int glyph, float scale_x, float scale_y, int *ix0, int *iy0, int *ix1, int *iy1)
{
- stbtt_GetGlyphBitmapBoxSubpixel(font, glyph, scale_x, scale_y, 0.0f, 0.0f, ix0, iy0, ix1, iy1);
+ stbtt_GetGlyphBitmapBoxSubpixel(font, glyph, scale_x, scale_y,0.0f,0.0f, ix0, iy0, ix1, iy1);
}
STBTT_DEF void stbtt_GetCodepointBitmapBoxSubpixel(const stbtt_fontinfo *font, int codepoint, float scale_x, float scale_y, float shift_x, float shift_y, int *ix0, int *iy0, int *ix1, int *iy1)
{
- stbtt_GetGlyphBitmapBoxSubpixel(font, stbtt_FindGlyphIndex(font, codepoint), scale_x, scale_y, shift_x, shift_y, ix0, iy0, ix1, iy1);
+ stbtt_GetGlyphBitmapBoxSubpixel(font, stbtt_FindGlyphIndex(font,codepoint), scale_x, scale_y,shift_x,shift_y, ix0,iy0,ix1,iy1);
}
STBTT_DEF void stbtt_GetCodepointBitmapBox(const stbtt_fontinfo *font, int codepoint, float scale_x, float scale_y, int *ix0, int *iy0, int *ix1, int *iy1)
{
- stbtt_GetCodepointBitmapBoxSubpixel(font, codepoint, scale_x, scale_y, 0.0f, 0.0f, ix0, iy0, ix1, iy1);
+ stbtt_GetCodepointBitmapBoxSubpixel(font, codepoint, scale_x, scale_y,0.0f,0.0f, ix0,iy0,ix1,iy1);
}
//////////////////////////////////////////////////////////////////////////////
@@ -2673,75 +2757,74 @@ STBTT_DEF void stbtt_GetCodepointBitmapBox(const stbtt_fontinfo *font, int codep
typedef struct stbtt__hheap_chunk
{
- struct stbtt__hheap_chunk *next;
+ struct stbtt__hheap_chunk *next;
} stbtt__hheap_chunk;
typedef struct stbtt__hheap
{
- struct stbtt__hheap_chunk *head;
- void *first_free;
- int num_remaining_in_head_chunk;
+ struct stbtt__hheap_chunk *head;
+ void *first_free;
+ int num_remaining_in_head_chunk;
} stbtt__hheap;
static void *stbtt__hheap_alloc(stbtt__hheap *hh, size_t size, void *userdata)
{
- if (hh->first_free) {
- void *p = hh->first_free;
- hh->first_free = *(void **)p;
- return p;
- }
- else {
- if (hh->num_remaining_in_head_chunk == 0) {
- int count = (size < 32 ? 2000 : size < 128 ? 800 : 100);
- stbtt__hheap_chunk *c = (stbtt__hheap_chunk *)STBTT_malloc(sizeof(stbtt__hheap_chunk) + size * count, userdata);
- if (c == NULL)
- return NULL;
- c->next = hh->head;
- hh->head = c;
- hh->num_remaining_in_head_chunk = count;
- }
- --hh->num_remaining_in_head_chunk;
- return (char *)(hh->head) + sizeof(stbtt__hheap_chunk) + size * hh->num_remaining_in_head_chunk;
- }
+ if (hh->first_free) {
+ void *p = hh->first_free;
+ hh->first_free = * (void **) p;
+ return p;
+ } else {
+ if (hh->num_remaining_in_head_chunk == 0) {
+ int count = (size < 32 ? 2000 : size < 128 ? 800 : 100);
+ stbtt__hheap_chunk *c = (stbtt__hheap_chunk *) STBTT_malloc(sizeof(stbtt__hheap_chunk) + size * count, userdata);
+ if (c == NULL)
+ return NULL;
+ c->next = hh->head;
+ hh->head = c;
+ hh->num_remaining_in_head_chunk = count;
+ }
+ --hh->num_remaining_in_head_chunk;
+ return (char *) (hh->head) + sizeof(stbtt__hheap_chunk) + size * hh->num_remaining_in_head_chunk;
+ }
}
static void stbtt__hheap_free(stbtt__hheap *hh, void *p)
{
- *(void **)p = hh->first_free;
- hh->first_free = p;
+ *(void **) p = hh->first_free;
+ hh->first_free = p;
}
static void stbtt__hheap_cleanup(stbtt__hheap *hh, void *userdata)
{
- stbtt__hheap_chunk *c = hh->head;
- while (c) {
- stbtt__hheap_chunk *n = c->next;
- STBTT_free(c, userdata);
- c = n;
- }
+ stbtt__hheap_chunk *c = hh->head;
+ while (c) {
+ stbtt__hheap_chunk *n = c->next;
+ STBTT_free(c, userdata);
+ c = n;
+ }
}
typedef struct stbtt__edge {
- float x0, y0, x1, y1;
- int invert;
+ float x0,y0, x1,y1;
+ int invert;
} stbtt__edge;
typedef struct stbtt__active_edge
{
- struct stbtt__active_edge *next;
-#if STBTT_RASTERIZER_VERSION==1
- int x, dx;
- float ey;
- int direction;
-#elif STBTT_RASTERIZER_VERSION==2
- float fx, fdx, fdy;
- float direction;
- float sy;
- float ey;
-#else
-#error "Unrecognized value of STBTT_RASTERIZER_VERSION"
-#endif
+ struct stbtt__active_edge *next;
+ #if STBTT_RASTERIZER_VERSION==1
+ int x,dx;
+ float ey;
+ int direction;
+ #elif STBTT_RASTERIZER_VERSION==2
+ float fx,fdx,fdy;
+ float direction;
+ float sy;
+ float ey;
+ #else
+ #error "Unrecognized value of STBTT_RASTERIZER_VERSION"
+ #endif
} stbtt__active_edge;
#if STBTT_RASTERIZER_VERSION == 1
@@ -2751,42 +2834,42 @@ typedef struct stbtt__active_edge
static stbtt__active_edge *stbtt__new_active(stbtt__hheap *hh, stbtt__edge *e, int off_x, float start_point, void *userdata)
{
- stbtt__active_edge *z = (stbtt__active_edge *)stbtt__hheap_alloc(hh, sizeof(*z), userdata);
- float dxdy = (e->x1 - e->x0) / (e->y1 - e->y0);
- STBTT_assert(z != NULL);
- if (!z) return z;
+ stbtt__active_edge *z = (stbtt__active_edge *) stbtt__hheap_alloc(hh, sizeof(*z), userdata);
+ float dxdy = (e->x1 - e->x0) / (e->y1 - e->y0);
+ STBTT_assert(z != NULL);
+ if (!z) return z;
- // round dx down to avoid overshooting
- if (dxdy < 0)
- z->dx = -STBTT_ifloor(STBTT_FIX * -dxdy);
- else
- z->dx = STBTT_ifloor(STBTT_FIX * dxdy);
+ // round dx down to avoid overshooting
+ if (dxdy < 0)
+ z->dx = -STBTT_ifloor(STBTT_FIX * -dxdy);
+ else
+ z->dx = STBTT_ifloor(STBTT_FIX * dxdy);
- z->x = STBTT_ifloor(STBTT_FIX * e->x0 + z->dx * (start_point - e->y0)); // use z->dx so when we offset later it's by the same amount
- z->x -= off_x * STBTT_FIX;
+ z->x = STBTT_ifloor(STBTT_FIX * e->x0 + z->dx * (start_point - e->y0)); // use z->dx so when we offset later it's by the same amount
+ z->x -= off_x * STBTT_FIX;
- z->ey = e->y1;
- z->next = 0;
- z->direction = e->invert ? 1 : -1;
- return z;
+ z->ey = e->y1;
+ z->next = 0;
+ z->direction = e->invert ? 1 : -1;
+ return z;
}
#elif STBTT_RASTERIZER_VERSION == 2
static stbtt__active_edge *stbtt__new_active(stbtt__hheap *hh, stbtt__edge *e, int off_x, float start_point, void *userdata)
{
- stbtt__active_edge *z = (stbtt__active_edge *)stbtt__hheap_alloc(hh, sizeof(*z), userdata);
- float dxdy = (e->x1 - e->x0) / (e->y1 - e->y0);
- STBTT_assert(z != NULL);
- //STBTT_assert(e->y0 <= start_point);
- if (!z) return z;
- z->fdx = dxdy;
- z->fdy = dxdy != 0.0f ? (1.0f / dxdy) : 0.0f;
- z->fx = e->x0 + dxdy * (start_point - e->y0);
- z->fx -= off_x;
- z->direction = e->invert ? 1.0f : -1.0f;
- z->sy = e->y0;
- z->ey = e->y1;
- z->next = 0;
- return z;
+ stbtt__active_edge *z = (stbtt__active_edge *) stbtt__hheap_alloc(hh, sizeof(*z), userdata);
+ float dxdy = (e->x1 - e->x0) / (e->y1 - e->y0);
+ STBTT_assert(z != NULL);
+ //STBTT_assert(e->y0 <= start_point);
+ if (!z) return z;
+ z->fdx = dxdy;
+ z->fdy = dxdy != 0.0f ? (1.0f/dxdy) : 0.0f;
+ z->fx = e->x0 + dxdy * (start_point - e->y0);
+ z->fx -= off_x;
+ z->direction = e->invert ? 1.0f : -1.0f;
+ z->sy = e->y0;
+ z->ey = e->y1;
+ z->next = 0;
+ return z;
}
#else
#error "Unrecognized value of STBTT_RASTERIZER_VERSION"
@@ -2798,148 +2881,144 @@ static stbtt__active_edge *stbtt__new_active(stbtt__hheap *hh, stbtt__edge *e, i
// are wrong, or if the user supplies a too-small bitmap
static void stbtt__fill_active_edges(unsigned char *scanline, int len, stbtt__active_edge *e, int max_weight)
{
- // non-zero winding fill
- int x0 = 0, w = 0;
-
- while (e) {
- if (w == 0) {
- // if we're currently at zero, we need to record the edge start point
- x0 = e->x; w += e->direction;
- }
- else {
- int x1 = e->x; w += e->direction;
- // if we went to zero, we need to draw
- if (w == 0) {
- int i = x0 >> STBTT_FIXSHIFT;
- int j = x1 >> STBTT_FIXSHIFT;
-
- if (i < len && j >= 0) {
- if (i == j) {
- // x0,x1 are the same pixel, so compute combined coverage
- scanline[i] = scanline[i] + (stbtt_uint8)((x1 - x0) * max_weight >> STBTT_FIXSHIFT);
- }
- else {
- if (i >= 0) // add antialiasing for x0
- scanline[i] = scanline[i] + (stbtt_uint8)(((STBTT_FIX - (x0 & STBTT_FIXMASK)) * max_weight) >> STBTT_FIXSHIFT);
- else
- i = -1; // clip
-
- if (j < len) // add antialiasing for x1
- scanline[j] = scanline[j] + (stbtt_uint8)(((x1 & STBTT_FIXMASK) * max_weight) >> STBTT_FIXSHIFT);
- else
- j = len; // clip
-
- for (++i; i < j; ++i) // fill pixels between x0 and x1
- scanline[i] = scanline[i] + (stbtt_uint8)max_weight;
- }
- }
+ // non-zero winding fill
+ int x0=0, w=0;
+
+ while (e) {
+ if (w == 0) {
+ // if we're currently at zero, we need to record the edge start point
+ x0 = e->x; w += e->direction;
+ } else {
+ int x1 = e->x; w += e->direction;
+ // if we went to zero, we need to draw
+ if (w == 0) {
+ int i = x0 >> STBTT_FIXSHIFT;
+ int j = x1 >> STBTT_FIXSHIFT;
+
+ if (i < len && j >= 0) {
+ if (i == j) {
+ // x0,x1 are the same pixel, so compute combined coverage
+ scanline[i] = scanline[i] + (stbtt_uint8) ((x1 - x0) * max_weight >> STBTT_FIXSHIFT);
+ } else {
+ if (i >= 0) // add antialiasing for x0
+ scanline[i] = scanline[i] + (stbtt_uint8) (((STBTT_FIX - (x0 & STBTT_FIXMASK)) * max_weight) >> STBTT_FIXSHIFT);
+ else
+ i = -1; // clip
+
+ if (j < len) // add antialiasing for x1
+ scanline[j] = scanline[j] + (stbtt_uint8) (((x1 & STBTT_FIXMASK) * max_weight) >> STBTT_FIXSHIFT);
+ else
+ j = len; // clip
+
+ for (++i; i < j; ++i) // fill pixels between x0 and x1
+ scanline[i] = scanline[i] + (stbtt_uint8) max_weight;
+ }
}
- }
+ }
+ }
- e = e->next;
- }
+ e = e->next;
+ }
}
static void stbtt__rasterize_sorted_edges(stbtt__bitmap *result, stbtt__edge *e, int n, int vsubsample, int off_x, int off_y, void *userdata)
{
- stbtt__hheap hh = { 0, 0, 0 };
- stbtt__active_edge *active = NULL;
- int y, j = 0;
- int max_weight = (255 / vsubsample); // weight per vertical scanline
- int s; // vertical subsample index
- unsigned char scanline_data[512], *scanline;
-
- if (result->w > 512)
- scanline = (unsigned char *)STBTT_malloc(result->w, userdata);
- else
- scanline = scanline_data;
-
- y = off_y * vsubsample;
- e[n].y0 = (off_y + result->h) * (float)vsubsample + 1;
-
- while (j < result->h) {
- STBTT_memset(scanline, 0, result->w);
- for (s = 0; s < vsubsample; ++s) {
- // find center of pixel for this scanline
- float scan_y = y + 0.5f;
- stbtt__active_edge **step = &active;
-
- // update all active edges;
- // remove all active edges that terminate before the center of this scanline
- while (*step) {
- stbtt__active_edge * z = *step;
- if (z->ey <= scan_y) {
- *step = z->next; // delete from list
- STBTT_assert(z->direction);
- z->direction = 0;
- stbtt__hheap_free(&hh, z);
- }
- else {
- z->x += z->dx; // advance to position for current scanline
- step = &((*step)->next); // advance through list
- }
+ stbtt__hheap hh = { 0, 0, 0 };
+ stbtt__active_edge *active = NULL;
+ int y,j=0;
+ int max_weight = (255 / vsubsample); // weight per vertical scanline
+ int s; // vertical subsample index
+ unsigned char scanline_data[512], *scanline;
+
+ if (result->w > 512)
+ scanline = (unsigned char *) STBTT_malloc(result->w, userdata);
+ else
+ scanline = scanline_data;
+
+ y = off_y * vsubsample;
+ e[n].y0 = (off_y + result->h) * (float) vsubsample + 1;
+
+ while (j < result->h) {
+ STBTT_memset(scanline, 0, result->w);
+ for (s=0; s < vsubsample; ++s) {
+ // find center of pixel for this scanline
+ float scan_y = y + 0.5f;
+ stbtt__active_edge **step = &active;
+
+ // update all active edges;
+ // remove all active edges that terminate before the center of this scanline
+ while (*step) {
+ stbtt__active_edge * z = *step;
+ if (z->ey <= scan_y) {
+ *step = z->next; // delete from list
+ STBTT_assert(z->direction);
+ z->direction = 0;
+ stbtt__hheap_free(&hh, z);
+ } else {
+ z->x += z->dx; // advance to position for current scanline
+ step = &((*step)->next); // advance through list
}
-
- // resort the list if needed
- for (;;) {
- int changed = 0;
- step = &active;
- while (*step && (*step)->next) {
- if ((*step)->x > (*step)->next->x) {
- stbtt__active_edge *t = *step;
- stbtt__active_edge *q = t->next;
-
- t->next = q->next;
- q->next = t;
- *step = q;
- changed = 1;
- }
- step = &(*step)->next;
- }
- if (!changed) break;
+ }
+
+ // resort the list if needed
+ for(;;) {
+ int changed=0;
+ step = &active;
+ while (*step && (*step)->next) {
+ if ((*step)->x > (*step)->next->x) {
+ stbtt__active_edge *t = *step;
+ stbtt__active_edge *q = t->next;
+
+ t->next = q->next;
+ q->next = t;
+ *step = q;
+ changed = 1;
+ }
+ step = &(*step)->next;
}
-
- // insert all edges that start before the center of this scanline -- omit ones that also end on this scanline
- while (e->y0 <= scan_y) {
- if (e->y1 > scan_y) {
- stbtt__active_edge *z = stbtt__new_active(&hh, e, off_x, scan_y, userdata);
- if (z != NULL) {
- // find insertion point
- if (active == NULL)
- active = z;
- else if (z->x < active->x) {
- // insert at front
- z->next = active;
- active = z;
- }
- else {
- // find thing to insert AFTER
- stbtt__active_edge *p = active;
- while (p->next && p->next->x < z->x)
- p = p->next;
- // at this point, p->next->x is NOT < z->x
- z->next = p->next;
- p->next = z;
- }
- }
- }
- ++e;
+ if (!changed) break;
+ }
+
+ // insert all edges that start before the center of this scanline -- omit ones that also end on this scanline
+ while (e->y0 <= scan_y) {
+ if (e->y1 > scan_y) {
+ stbtt__active_edge *z = stbtt__new_active(&hh, e, off_x, scan_y, userdata);
+ if (z != NULL) {
+ // find insertion point
+ if (active == NULL)
+ active = z;
+ else if (z->x < active->x) {
+ // insert at front
+ z->next = active;
+ active = z;
+ } else {
+ // find thing to insert AFTER
+ stbtt__active_edge *p = active;
+ while (p->next && p->next->x < z->x)
+ p = p->next;
+ // at this point, p->next->x is NOT < z->x
+ z->next = p->next;
+ p->next = z;
+ }
+ }
}
+ ++e;
+ }
- // now process all active edges in XOR fashion
- if (active)
- stbtt__fill_active_edges(scanline, result->w, active, max_weight);
+ // now process all active edges in XOR fashion
+ if (active)
+ stbtt__fill_active_edges(scanline, result->w, active, max_weight);
- ++y;
- }
- STBTT_memcpy(result->pixels + j * result->stride, scanline, result->w);
- ++j;
- }
+ ++y;
+ }
+ STBTT_memcpy(result->pixels + j * result->stride, scanline, result->w);
+ ++j;
+ }
- stbtt__hheap_cleanup(&hh, userdata);
+ stbtt__hheap_cleanup(&hh, userdata);
- if (scanline != scanline_data)
- STBTT_free(scanline, userdata);
+ if (scanline != scanline_data)
+ STBTT_free(scanline, userdata);
}
#elif STBTT_RASTERIZER_VERSION == 2
@@ -2948,305 +3027,368 @@ static void stbtt__rasterize_sorted_edges(stbtt__bitmap *result, stbtt__edge *e,
// (i.e. it has already been clipped to those)
static void stbtt__handle_clipped_edge(float *scanline, int x, stbtt__active_edge *e, float x0, float y0, float x1, float y1)
{
- if (y0 == y1) return;
- STBTT_assert(y0 < y1);
- STBTT_assert(e->sy <= e->ey);
- if (y0 > e->ey) return;
- if (y1 < e->sy) return;
- if (y0 < e->sy) {
- x0 += (x1 - x0) * (e->sy - y0) / (y1 - y0);
- y0 = e->sy;
- }
- if (y1 > e->ey) {
- x1 += (x1 - x0) * (e->ey - y1) / (y1 - y0);
- y1 = e->ey;
- }
-
- if (x0 == x)
- STBTT_assert(x1 <= x + 1);
- else if (x0 == x + 1)
- STBTT_assert(x1 >= x);
- else if (x0 <= x)
- STBTT_assert(x1 <= x);
- else if (x0 >= x + 1)
- STBTT_assert(x1 >= x + 1);
- else
- STBTT_assert(x1 >= x && x1 <= x + 1);
-
- if (x0 <= x && x1 <= x)
- scanline[x] += e->direction * (y1 - y0);
- else if (x0 >= x + 1 && x1 >= x + 1)
- ;
- else {
- STBTT_assert(x0 >= x && x0 <= x + 1 && x1 >= x && x1 <= x + 1);
- scanline[x] += e->direction * (y1 - y0) * (1 - ((x0 - x) + (x1 - x)) / 2); // coverage = 1 - average x position
- }
+ if (y0 == y1) return;
+ STBTT_assert(y0 < y1);
+ STBTT_assert(e->sy <= e->ey);
+ if (y0 > e->ey) return;
+ if (y1 < e->sy) return;
+ if (y0 < e->sy) {
+ x0 += (x1-x0) * (e->sy - y0) / (y1-y0);
+ y0 = e->sy;
+ }
+ if (y1 > e->ey) {
+ x1 += (x1-x0) * (e->ey - y1) / (y1-y0);
+ y1 = e->ey;
+ }
+
+ if (x0 == x)
+ STBTT_assert(x1 <= x+1);
+ else if (x0 == x+1)
+ STBTT_assert(x1 >= x);
+ else if (x0 <= x)
+ STBTT_assert(x1 <= x);
+ else if (x0 >= x+1)
+ STBTT_assert(x1 >= x+1);
+ else
+ STBTT_assert(x1 >= x && x1 <= x+1);
+
+ if (x0 <= x && x1 <= x)
+ scanline[x] += e->direction * (y1-y0);
+ else if (x0 >= x+1 && x1 >= x+1)
+ ;
+ else {
+ STBTT_assert(x0 >= x && x0 <= x+1 && x1 >= x && x1 <= x+1);
+ scanline[x] += e->direction * (y1-y0) * (1-((x0-x)+(x1-x))/2); // coverage = 1 - average x position
+ }
+}
+
+static float stbtt__sized_trapezoid_area(float height, float top_width, float bottom_width)
+{
+ STBTT_assert(top_width >= 0);
+ STBTT_assert(bottom_width >= 0);
+ return (top_width + bottom_width) / 2.0f * height;
+}
+
+static float stbtt__position_trapezoid_area(float height, float tx0, float tx1, float bx0, float bx1)
+{
+ return stbtt__sized_trapezoid_area(height, tx1 - tx0, bx1 - bx0);
+}
+
+static float stbtt__sized_triangle_area(float height, float width)
+{
+ return height * width / 2;
}
static void stbtt__fill_active_edges_new(float *scanline, float *scanline_fill, int len, stbtt__active_edge *e, float y_top)
{
- float y_bottom = y_top + 1;
+ float y_bottom = y_top+1;
- while (e) {
- // brute force every pixel
+ while (e) {
+ // brute force every pixel
- // compute intersection points with top & bottom
- STBTT_assert(e->ey >= y_top);
+ // compute intersection points with top & bottom
+ STBTT_assert(e->ey >= y_top);
- if (e->fdx == 0) {
- float x0 = e->fx;
- if (x0 < len) {
- if (x0 >= 0) {
- stbtt__handle_clipped_edge(scanline, (int)x0, e, x0, y_top, x0, y_bottom);
- stbtt__handle_clipped_edge(scanline_fill - 1, (int)x0 + 1, e, x0, y_top, x0, y_bottom);
- }
- else {
- stbtt__handle_clipped_edge(scanline_fill - 1, 0, e, x0, y_top, x0, y_bottom);
- }
- }
- }
- else {
- float x0 = e->fx;
- float dx = e->fdx;
- float xb = x0 + dx;
- float x_top, x_bottom;
- float sy0, sy1;
- float dy = e->fdy;
- STBTT_assert(e->sy <= y_bottom && e->ey >= y_top);
-
- // compute endpoints of line segment clipped to this scanline (if the
- // line segment starts on this scanline. x0 is the intersection of the
- // line with y_top, but that may be off the line segment.
- if (e->sy > y_top) {
- x_top = x0 + dx * (e->sy - y_top);
- sy0 = e->sy;
- }
- else {
- x_top = x0;
- sy0 = y_top;
+ if (e->fdx == 0) {
+ float x0 = e->fx;
+ if (x0 < len) {
+ if (x0 >= 0) {
+ stbtt__handle_clipped_edge(scanline,(int) x0,e, x0,y_top, x0,y_bottom);
+ stbtt__handle_clipped_edge(scanline_fill-1,(int) x0+1,e, x0,y_top, x0,y_bottom);
+ } else {
+ stbtt__handle_clipped_edge(scanline_fill-1,0,e, x0,y_top, x0,y_bottom);
}
- if (e->ey < y_bottom) {
- x_bottom = x0 + dx * (e->ey - y_top);
- sy1 = e->ey;
+ }
+ } else {
+ float x0 = e->fx;
+ float dx = e->fdx;
+ float xb = x0 + dx;
+ float x_top, x_bottom;
+ float sy0,sy1;
+ float dy = e->fdy;
+ STBTT_assert(e->sy <= y_bottom && e->ey >= y_top);
+
+ // compute endpoints of line segment clipped to this scanline (if the
+ // line segment starts on this scanline. x0 is the intersection of the
+ // line with y_top, but that may be off the line segment.
+ if (e->sy > y_top) {
+ x_top = x0 + dx * (e->sy - y_top);
+ sy0 = e->sy;
+ } else {
+ x_top = x0;
+ sy0 = y_top;
+ }
+ if (e->ey < y_bottom) {
+ x_bottom = x0 + dx * (e->ey - y_top);
+ sy1 = e->ey;
+ } else {
+ x_bottom = xb;
+ sy1 = y_bottom;
+ }
+
+ if (x_top >= 0 && x_bottom >= 0 && x_top < len && x_bottom < len) {
+ // from here on, we don't have to range check x values
+
+ if ((int) x_top == (int) x_bottom) {
+ float height;
+ // simple case, only spans one pixel
+ int x = (int) x_top;
+ height = (sy1 - sy0) * e->direction;
+ STBTT_assert(x >= 0 && x < len);
+ scanline[x] += stbtt__position_trapezoid_area(height, x_top, x+1.0f, x_bottom, x+1.0f);
+ scanline_fill[x] += height; // everything right of this pixel is filled
+ } else {
+ int x,x1,x2;
+ float y_crossing, y_final, step, sign, area;
+ // covers 2+ pixels
+ if (x_top > x_bottom) {
+ // flip scanline vertically; signed area is the same
+ float t;
+ sy0 = y_bottom - (sy0 - y_top);
+ sy1 = y_bottom - (sy1 - y_top);
+ t = sy0, sy0 = sy1, sy1 = t;
+ t = x_bottom, x_bottom = x_top, x_top = t;
+ dx = -dx;
+ dy = -dy;
+ t = x0, x0 = xb, xb = t;
+ }
+ STBTT_assert(dy >= 0);
+ STBTT_assert(dx >= 0);
+
+ x1 = (int) x_top;
+ x2 = (int) x_bottom;
+ // compute intersection with y axis at x1+1
+ y_crossing = y_top + dy * (x1+1 - x0);
+
+ // compute intersection with y axis at x2
+ y_final = y_top + dy * (x2 - x0);
+
+ // x1 x_top x2 x_bottom
+ // y_top +------|-----+------------+------------+--------|---+------------+
+ // | | | | | |
+ // | | | | | |
+ // sy0 | Txxxxx|............|............|............|............|
+ // y_crossing | *xxxxx.......|............|............|............|
+ // | | xxxxx..|............|............|............|
+ // | | /- xx*xxxx........|............|............|
+ // | | dy < | xxxxxx..|............|............|
+ // y_final | | \- | xx*xxx.........|............|
+ // sy1 | | | | xxxxxB...|............|
+ // | | | | | |
+ // | | | | | |
+ // y_bottom +------------+------------+------------+------------+------------+
+ //
+ // goal is to measure the area covered by '.' in each pixel
+
+ // if x2 is right at the right edge of x1, y_crossing can blow up, github #1057
+ // @TODO: maybe test against sy1 rather than y_bottom?
+ if (y_crossing > y_bottom)
+ y_crossing = y_bottom;
+
+ sign = e->direction;
+
+ // area of the rectangle covered from sy0..y_crossing
+ area = sign * (y_crossing-sy0);
+
+ // area of the triangle (x_top,sy0), (x1+1,sy0), (x1+1,y_crossing)
+ scanline[x1] += stbtt__sized_triangle_area(area, x1+1 - x_top);
+
+ // check if final y_crossing is blown up; no test case for this
+ if (y_final > y_bottom) {
+ y_final = y_bottom;
+ dy = (y_final - y_crossing ) / (x2 - (x1+1)); // if denom=0, y_final = y_crossing, so y_final <= y_bottom
+ }
+
+ // in second pixel, area covered by line segment found in first pixel
+ // is always a rectangle 1 wide * the height of that line segment; this
+ // is exactly what the variable 'area' stores. it also gets a contribution
+ // from the line segment within it. the THIRD pixel will get the first
+ // pixel's rectangle contribution, the second pixel's rectangle contribution,
+ // and its own contribution. the 'own contribution' is the same in every pixel except
+ // the leftmost and rightmost, a trapezoid that slides down in each pixel.
+ // the second pixel's contribution to the third pixel will be the
+ // rectangle 1 wide times the height change in the second pixel, which is dy.
+
+ step = sign * dy * 1; // dy is dy/dx, change in y for every 1 change in x,
+ // which multiplied by 1-pixel-width is how much pixel area changes for each step in x
+ // so the area advances by 'step' every time
+
+ for (x = x1+1; x < x2; ++x) {
+ scanline[x] += area + step/2; // area of trapezoid is 1*step/2
+ area += step;
+ }
+ STBTT_assert(STBTT_fabs(area) <= 1.01f); // accumulated error from area += step unless we round step down
+ STBTT_assert(sy1 > y_final-0.01f);
+
+ // area covered in the last pixel is the rectangle from all the pixels to the left,
+ // plus the trapezoid filled by the line segment in this pixel all the way to the right edge
+ scanline[x2] += area + sign * stbtt__position_trapezoid_area(sy1-y_final, (float) x2, x2+1.0f, x_bottom, x2+1.0f);
+
+ // the rest of the line is filled based on the total height of the line segment in this pixel
+ scanline_fill[x2] += sign * (sy1-sy0);
}
- else {
- x_bottom = xb;
- sy1 = y_bottom;
- }
-
- if (x_top >= 0 && x_bottom >= 0 && x_top < len && x_bottom < len) {
- // from here on, we don't have to range check x values
-
- if ((int)x_top == (int)x_bottom) {
- float height;
- // simple case, only spans one pixel
- int x = (int)x_top;
- height = sy1 - sy0;
- STBTT_assert(x >= 0 && x < len);
- scanline[x] += e->direction * (1 - ((x_top - x) + (x_bottom - x)) / 2) * height;
- scanline_fill[x] += e->direction * height; // everything right of this pixel is filled
- }
- else {
- int x, x1, x2;
- float y_crossing, step, sign, area;
- // covers 2+ pixels
- if (x_top > x_bottom) {
- // flip scanline vertically; signed area is the same
- float t;
- sy0 = y_bottom - (sy0 - y_top);
- sy1 = y_bottom - (sy1 - y_top);
- t = sy0, sy0 = sy1, sy1 = t;
- t = x_bottom, x_bottom = x_top, x_top = t;
- dx = -dx;
- dy = -dy;
- t = x0, x0 = xb, xb = t;
- }
-
- x1 = (int)x_top;
- x2 = (int)x_bottom;
- // compute intersection with y axis at x1+1
- y_crossing = (x1 + 1 - x0) * dy + y_top;
-
- sign = e->direction;
- // area of the rectangle covered from y0..y_crossing
- area = sign * (y_crossing - sy0);
- // area of the triangle (x_top,y0), (x+1,y0), (x+1,y_crossing)
- scanline[x1] += area * (1 - ((x_top - x1) + (x1 + 1 - x1)) / 2);
-
- step = sign * dy;
- for (x = x1 + 1; x < x2; ++x) {
- scanline[x] += area + step / 2;
- area += step;
- }
- y_crossing += dy * (x2 - (x1 + 1));
-
- STBTT_assert(STBTT_fabs(area) <= 1.01f);
-
- scanline[x2] += area + sign * (1 - ((x2 - x2) + (x_bottom - x2)) / 2) * (sy1 - y_crossing);
-
- scanline_fill[x2] += sign * (sy1 - sy0);
- }
- }
- else {
- // if edge goes outside of box we're drawing, we require
- // clipping logic. since this does not match the intended use
- // of this library, we use a different, very slow brute
- // force implementation
- int x;
- for (x = 0; x < len; ++x) {
- // cases:
- //
- // there can be up to two intersections with the pixel. any intersection
- // with left or right edges can be handled by splitting into two (or three)
- // regions. intersections with top & bottom do not necessitate case-wise logic.
- //
- // the old way of doing this found the intersections with the left & right edges,
- // then used some simple logic to produce up to three segments in sorted order
- // from top-to-bottom. however, this had a problem: if an x edge was epsilon
- // across the x border, then the corresponding y position might not be distinct
- // from the other y segment, and it might ignored as an empty segment. to avoid
- // that, we need to explicitly produce segments based on x positions.
-
- // rename variables to clearly-defined pairs
- float y0 = y_top;
- float x1 = (float)(x);
- float x2 = (float)(x + 1);
- float x3 = xb;
- float y3 = y_bottom;
-
- // x = e->x + e->dx * (y-y_top)
- // (y-y_top) = (x - e->x) / e->dx
- // y = (x - e->x) / e->dx + y_top
- float y1 = (x - x0) / dx + y_top;
- float y2 = (x + 1 - x0) / dx + y_top;
-
- if (x0 < x1 && x3 > x2) { // three segments descending down-right
- stbtt__handle_clipped_edge(scanline, x, e, x0, y0, x1, y1);
- stbtt__handle_clipped_edge(scanline, x, e, x1, y1, x2, y2);
- stbtt__handle_clipped_edge(scanline, x, e, x2, y2, x3, y3);
- }
- else if (x3 < x1 && x0 > x2) { // three segments descending down-left
- stbtt__handle_clipped_edge(scanline, x, e, x0, y0, x2, y2);
- stbtt__handle_clipped_edge(scanline, x, e, x2, y2, x1, y1);
- stbtt__handle_clipped_edge(scanline, x, e, x1, y1, x3, y3);
- }
- else if (x0 < x1 && x3 > x1) { // two segments across x, down-right
- stbtt__handle_clipped_edge(scanline, x, e, x0, y0, x1, y1);
- stbtt__handle_clipped_edge(scanline, x, e, x1, y1, x3, y3);
- }
- else if (x3 < x1 && x0 > x1) { // two segments across x, down-left
- stbtt__handle_clipped_edge(scanline, x, e, x0, y0, x1, y1);
- stbtt__handle_clipped_edge(scanline, x, e, x1, y1, x3, y3);
- }
- else if (x0 < x2 && x3 > x2) { // two segments across x+1, down-right
- stbtt__handle_clipped_edge(scanline, x, e, x0, y0, x2, y2);
- stbtt__handle_clipped_edge(scanline, x, e, x2, y2, x3, y3);
- }
- else if (x3 < x2 && x0 > x2) { // two segments across x+1, down-left
- stbtt__handle_clipped_edge(scanline, x, e, x0, y0, x2, y2);
- stbtt__handle_clipped_edge(scanline, x, e, x2, y2, x3, y3);
- }
- else { // one segment
- stbtt__handle_clipped_edge(scanline, x, e, x0, y0, x3, y3);
- }
- }
+ } else {
+ // if edge goes outside of box we're drawing, we require
+ // clipping logic. since this does not match the intended use
+ // of this library, we use a different, very slow brute
+ // force implementation
+ // note though that this does happen some of the time because
+ // x_top and x_bottom can be extrapolated at the top & bottom of
+ // the shape and actually lie outside the bounding box
+ int x;
+ for (x=0; x < len; ++x) {
+ // cases:
+ //
+ // there can be up to two intersections with the pixel. any intersection
+ // with left or right edges can be handled by splitting into two (or three)
+ // regions. intersections with top & bottom do not necessitate case-wise logic.
+ //
+ // the old way of doing this found the intersections with the left & right edges,
+ // then used some simple logic to produce up to three segments in sorted order
+ // from top-to-bottom. however, this had a problem: if an x edge was epsilon
+ // across the x border, then the corresponding y position might not be distinct
+ // from the other y segment, and it might ignored as an empty segment. to avoid
+ // that, we need to explicitly produce segments based on x positions.
+
+ // rename variables to clearly-defined pairs
+ float y0 = y_top;
+ float x1 = (float) (x);
+ float x2 = (float) (x+1);
+ float x3 = xb;
+ float y3 = y_bottom;
+
+ // x = e->x + e->dx * (y-y_top)
+ // (y-y_top) = (x - e->x) / e->dx
+ // y = (x - e->x) / e->dx + y_top
+ float y1 = (x - x0) / dx + y_top;
+ float y2 = (x+1 - x0) / dx + y_top;
+
+ if (x0 < x1 && x3 > x2) { // three segments descending down-right
+ stbtt__handle_clipped_edge(scanline,x,e, x0,y0, x1,y1);
+ stbtt__handle_clipped_edge(scanline,x,e, x1,y1, x2,y2);
+ stbtt__handle_clipped_edge(scanline,x,e, x2,y2, x3,y3);
+ } else if (x3 < x1 && x0 > x2) { // three segments descending down-left
+ stbtt__handle_clipped_edge(scanline,x,e, x0,y0, x2,y2);
+ stbtt__handle_clipped_edge(scanline,x,e, x2,y2, x1,y1);
+ stbtt__handle_clipped_edge(scanline,x,e, x1,y1, x3,y3);
+ } else if (x0 < x1 && x3 > x1) { // two segments across x, down-right
+ stbtt__handle_clipped_edge(scanline,x,e, x0,y0, x1,y1);
+ stbtt__handle_clipped_edge(scanline,x,e, x1,y1, x3,y3);
+ } else if (x3 < x1 && x0 > x1) { // two segments across x, down-left
+ stbtt__handle_clipped_edge(scanline,x,e, x0,y0, x1,y1);
+ stbtt__handle_clipped_edge(scanline,x,e, x1,y1, x3,y3);
+ } else if (x0 < x2 && x3 > x2) { // two segments across x+1, down-right
+ stbtt__handle_clipped_edge(scanline,x,e, x0,y0, x2,y2);
+ stbtt__handle_clipped_edge(scanline,x,e, x2,y2, x3,y3);
+ } else if (x3 < x2 && x0 > x2) { // two segments across x+1, down-left
+ stbtt__handle_clipped_edge(scanline,x,e, x0,y0, x2,y2);
+ stbtt__handle_clipped_edge(scanline,x,e, x2,y2, x3,y3);
+ } else { // one segment
+ stbtt__handle_clipped_edge(scanline,x,e, x0,y0, x3,y3);
+ }
}
- }
- e = e->next;
- }
+ }
+ }
+ e = e->next;
+ }
}
// directly AA rasterize edges w/o supersampling
static void stbtt__rasterize_sorted_edges(stbtt__bitmap *result, stbtt__edge *e, int n, int vsubsample, int off_x, int off_y, void *userdata)
{
- stbtt__hheap hh = { 0, 0, 0 };
- stbtt__active_edge *active = NULL;
- int y, j = 0, i;
- float scanline_data[129], *scanline, *scanline2;
-
- STBTT__NOTUSED(vsubsample);
-
- if (result->w > 64)
- scanline = (float *)STBTT_malloc((result->w * 2 + 1) * sizeof(float), userdata);
- else
- scanline = scanline_data;
-
- scanline2 = scanline + result->w;
-
- y = off_y;
- e[n].y0 = (float)(off_y + result->h) + 1;
-
- while (j < result->h) {
- // find center of pixel for this scanline
- float scan_y_top = y + 0.0f;
- float scan_y_bottom = y + 1.0f;
- stbtt__active_edge **step = &active;
-
- STBTT_memset(scanline, 0, result->w * sizeof(scanline[0]));
- STBTT_memset(scanline2, 0, (result->w + 1) * sizeof(scanline[0]));
-
- // update all active edges;
- // remove all active edges that terminate before the top of this scanline
- while (*step) {
- stbtt__active_edge * z = *step;
- if (z->ey <= scan_y_top) {
- *step = z->next; // delete from list
- STBTT_assert(z->direction);
- z->direction = 0;
- stbtt__hheap_free(&hh, z);
- }
- else {
- step = &((*step)->next); // advance through list
- }
- }
-
- // insert all edges that start before the bottom of this scanline
- while (e->y0 <= scan_y_bottom) {
- if (e->y0 != e->y1) {
- stbtt__active_edge *z = stbtt__new_active(&hh, e, off_x, scan_y_top, userdata);
- if (z != NULL) {
- STBTT_assert(z->ey >= scan_y_top);
- // insert at front
- z->next = active;
- active = z;
- }
- }
- ++e;
- }
-
- // now process all active edges
- if (active)
- stbtt__fill_active_edges_new(scanline, scanline2 + 1, result->w, active, scan_y_top);
-
- {
- float sum = 0;
- for (i = 0; i < result->w; ++i) {
- float k;
- int m;
- sum += scanline2[i];
- k = scanline[i] + sum;
- k = (float)STBTT_fabs(k) * 255 + 0.5f;
- m = (int)k;
- if (m > 255) m = 255;
- result->pixels[j*result->stride + i] = (unsigned char)m;
- }
- }
- // advance all the edges
- step = &active;
- while (*step) {
- stbtt__active_edge *z = *step;
- z->fx += z->fdx; // advance to position for current scanline
+ stbtt__hheap hh = { 0, 0, 0 };
+ stbtt__active_edge *active = NULL;
+ int y,j=0, i;
+ float scanline_data[129], *scanline, *scanline2;
+
+ STBTT__NOTUSED(vsubsample);
+
+ if (result->w > 64)
+ scanline = (float *) STBTT_malloc((result->w*2+1) * sizeof(float), userdata);
+ else
+ scanline = scanline_data;
+
+ scanline2 = scanline + result->w;
+
+ y = off_y;
+ e[n].y0 = (float) (off_y + result->h) + 1;
+
+ while (j < result->h) {
+ // find center of pixel for this scanline
+ float scan_y_top = y + 0.0f;
+ float scan_y_bottom = y + 1.0f;
+ stbtt__active_edge **step = &active;
+
+ STBTT_memset(scanline , 0, result->w*sizeof(scanline[0]));
+ STBTT_memset(scanline2, 0, (result->w+1)*sizeof(scanline[0]));
+
+ // update all active edges;
+ // remove all active edges that terminate before the top of this scanline
+ while (*step) {
+ stbtt__active_edge * z = *step;
+ if (z->ey <= scan_y_top) {
+ *step = z->next; // delete from list
+ STBTT_assert(z->direction);
+ z->direction = 0;
+ stbtt__hheap_free(&hh, z);
+ } else {
step = &((*step)->next); // advance through list
- }
-
- ++y;
- ++j;
- }
-
- stbtt__hheap_cleanup(&hh, userdata);
-
- if (scanline != scanline_data)
- STBTT_free(scanline, userdata);
+ }
+ }
+
+ // insert all edges that start before the bottom of this scanline
+ while (e->y0 <= scan_y_bottom) {
+ if (e->y0 != e->y1) {
+ stbtt__active_edge *z = stbtt__new_active(&hh, e, off_x, scan_y_top, userdata);
+ if (z != NULL) {
+ if (j == 0 && off_y != 0) {
+ if (z->ey < scan_y_top) {
+ // this can happen due to subpixel positioning and some kind of fp rounding error i think
+ z->ey = scan_y_top;
+ }
+ }
+ STBTT_assert(z->ey >= scan_y_top); // if we get really unlucky a tiny bit of an edge can be out of bounds
+ // insert at front
+ z->next = active;
+ active = z;
+ }
+ }
+ ++e;
+ }
+
+ // now process all active edges
+ if (active)
+ stbtt__fill_active_edges_new(scanline, scanline2+1, result->w, active, scan_y_top);
+
+ {
+ float sum = 0;
+ for (i=0; i < result->w; ++i) {
+ float k;
+ int m;
+ sum += scanline2[i];
+ k = scanline[i] + sum;
+ k = (float) STBTT_fabs(k)*255 + 0.5f;
+ m = (int) k;
+ if (m > 255) m = 255;
+ result->pixels[j*result->stride + i] = (unsigned char) m;
+ }
+ }
+ // advance all the edges
+ step = &active;
+ while (*step) {
+ stbtt__active_edge *z = *step;
+ z->fx += z->fdx; // advance to position for current scanline
+ step = &((*step)->next); // advance through list
+ }
+
+ ++y;
+ ++j;
+ }
+
+ stbtt__hheap_cleanup(&hh, userdata);
+
+ if (scanline != scanline_data)
+ STBTT_free(scanline, userdata);
}
#else
#error "Unrecognized value of STBTT_RASTERIZER_VERSION"
@@ -3256,413 +3398,410 @@ static void stbtt__rasterize_sorted_edges(stbtt__bitmap *result, stbtt__edge *e,
static void stbtt__sort_edges_ins_sort(stbtt__edge *p, int n)
{
- int i, j;
- for (i = 1; i < n; ++i) {
- stbtt__edge t = p[i], *a = &t;
- j = i;
- while (j > 0) {
- stbtt__edge *b = &p[j - 1];
- int c = STBTT__COMPARE(a, b);
- if (!c) break;
- p[j] = p[j - 1];
- --j;
- }
- if (i != j)
- p[j] = t;
- }
+ int i,j;
+ for (i=1; i < n; ++i) {
+ stbtt__edge t = p[i], *a = &t;
+ j = i;
+ while (j > 0) {
+ stbtt__edge *b = &p[j-1];
+ int c = STBTT__COMPARE(a,b);
+ if (!c) break;
+ p[j] = p[j-1];
+ --j;
+ }
+ if (i != j)
+ p[j] = t;
+ }
}
static void stbtt__sort_edges_quicksort(stbtt__edge *p, int n)
{
- /* threshhold for transitioning to insertion sort */
- while (n > 12) {
- stbtt__edge t;
- int c01, c12, c, m, i, j;
-
- /* compute median of three */
- m = n >> 1;
- c01 = STBTT__COMPARE(&p[0], &p[m]);
- c12 = STBTT__COMPARE(&p[m], &p[n - 1]);
- /* if 0 >= mid >= end, or 0 < mid < end, then use mid */
- if (c01 != c12) {
- /* otherwise, we'll need to swap something else to middle */
- int z;
- c = STBTT__COMPARE(&p[0], &p[n - 1]);
- /* 0>mid && mid<n: 0>n => n; 0<n => 0 */
- /* 0<mid && mid>n: 0>n => 0; 0<n => n */
- z = (c == c12) ? 0 : n - 1;
- t = p[z];
- p[z] = p[m];
- p[m] = t;
- }
- /* now p[m] is the median-of-three */
- /* swap it to the beginning so it won't move around */
- t = p[0];
- p[0] = p[m];
- p[m] = t;
-
- /* partition loop */
- i = 1;
- j = n - 1;
- for (;;) {
- /* handling of equality is crucial here */
- /* for sentinels & efficiency with duplicates */
- for (;; ++i) {
- if (!STBTT__COMPARE(&p[i], &p[0])) break;
- }
- for (;; --j) {
- if (!STBTT__COMPARE(&p[0], &p[j])) break;
- }
- /* make sure we haven't crossed */
- if (i >= j) break;
- t = p[i];
- p[i] = p[j];
- p[j] = t;
-
- ++i;
- --j;
- }
- /* recurse on smaller side, iterate on larger */
- if (j < (n - i)) {
- stbtt__sort_edges_quicksort(p, j);
- p = p + i;
- n = n - i;
- }
- else {
- stbtt__sort_edges_quicksort(p + i, n - i);
- n = j;
- }
- }
+ /* threshold for transitioning to insertion sort */
+ while (n > 12) {
+ stbtt__edge t;
+ int c01,c12,c,m,i,j;
+
+ /* compute median of three */
+ m = n >> 1;
+ c01 = STBTT__COMPARE(&p[0],&p[m]);
+ c12 = STBTT__COMPARE(&p[m],&p[n-1]);
+ /* if 0 >= mid >= end, or 0 < mid < end, then use mid */
+ if (c01 != c12) {
+ /* otherwise, we'll need to swap something else to middle */
+ int z;
+ c = STBTT__COMPARE(&p[0],&p[n-1]);
+ /* 0>mid && mid<n: 0>n => n; 0<n => 0 */
+ /* 0<mid && mid>n: 0>n => 0; 0<n => n */
+ z = (c == c12) ? 0 : n-1;
+ t = p[z];
+ p[z] = p[m];
+ p[m] = t;
+ }
+ /* now p[m] is the median-of-three */
+ /* swap it to the beginning so it won't move around */
+ t = p[0];
+ p[0] = p[m];
+ p[m] = t;
+
+ /* partition loop */
+ i=1;
+ j=n-1;
+ for(;;) {
+ /* handling of equality is crucial here */
+ /* for sentinels & efficiency with duplicates */
+ for (;;++i) {
+ if (!STBTT__COMPARE(&p[i], &p[0])) break;
+ }
+ for (;;--j) {
+ if (!STBTT__COMPARE(&p[0], &p[j])) break;
+ }
+ /* make sure we haven't crossed */
+ if (i >= j) break;
+ t = p[i];
+ p[i] = p[j];
+ p[j] = t;
+
+ ++i;
+ --j;
+ }
+ /* recurse on smaller side, iterate on larger */
+ if (j < (n-i)) {
+ stbtt__sort_edges_quicksort(p,j);
+ p = p+i;
+ n = n-i;
+ } else {
+ stbtt__sort_edges_quicksort(p+i, n-i);
+ n = j;
+ }
+ }
}
static void stbtt__sort_edges(stbtt__edge *p, int n)
{
- stbtt__sort_edges_quicksort(p, n);
- stbtt__sort_edges_ins_sort(p, n);
+ stbtt__sort_edges_quicksort(p, n);
+ stbtt__sort_edges_ins_sort(p, n);
}
typedef struct
{
- float x, y;
+ float x,y;
} stbtt__point;
static void stbtt__rasterize(stbtt__bitmap *result, stbtt__point *pts, int *wcount, int windings, float scale_x, float scale_y, float shift_x, float shift_y, int off_x, int off_y, int invert, void *userdata)
{
- float y_scale_inv = invert ? -scale_y : scale_y;
- stbtt__edge *e;
- int n, i, j, k, m;
+ float y_scale_inv = invert ? -scale_y : scale_y;
+ stbtt__edge *e;
+ int n,i,j,k,m;
#if STBTT_RASTERIZER_VERSION == 1
- int vsubsample = result->h < 8 ? 15 : 5;
+ int vsubsample = result->h < 8 ? 15 : 5;
#elif STBTT_RASTERIZER_VERSION == 2
- int vsubsample = 1;
+ int vsubsample = 1;
#else
-#error "Unrecognized value of STBTT_RASTERIZER_VERSION"
+ #error "Unrecognized value of STBTT_RASTERIZER_VERSION"
#endif
- // vsubsample should divide 255 evenly; otherwise we won't reach full opacity
-
- // now we have to blow out the windings into explicit edge lists
- n = 0;
- for (i = 0; i < windings; ++i)
- n += wcount[i];
-
- e = (stbtt__edge *)STBTT_malloc(sizeof(*e) * (n + 1), userdata); // add an extra one as a sentinel
- if (e == 0) return;
- n = 0;
-
- m = 0;
- for (i = 0; i < windings; ++i) {
- stbtt__point *p = pts + m;
- m += wcount[i];
- j = wcount[i] - 1;
- for (k = 0; k < wcount[i]; j = k++) {
- int a = k, b = j;
- // skip the edge if horizontal
- if (p[j].y == p[k].y)
- continue;
- // add edge from j to k to the list
- e[n].invert = 0;
- if (invert ? p[j].y > p[k].y : p[j].y < p[k].y) {
- e[n].invert = 1;
- a = j, b = k;
- }
- e[n].x0 = p[a].x * scale_x + shift_x;
- e[n].y0 = (p[a].y * y_scale_inv + shift_y) * vsubsample;
- e[n].x1 = p[b].x * scale_x + shift_x;
- e[n].y1 = (p[b].y * y_scale_inv + shift_y) * vsubsample;
- ++n;
- }
- }
-
- // now sort the edges by their highest point (should snap to integer, and then by x)
- //STBTT_sort(e, n, sizeof(e[0]), stbtt__edge_compare);
- stbtt__sort_edges(e, n);
-
- // now, traverse the scanlines and find the intersections on each scanline, use xor winding rule
- stbtt__rasterize_sorted_edges(result, e, n, vsubsample, off_x, off_y, userdata);
-
- STBTT_free(e, userdata);
+ // vsubsample should divide 255 evenly; otherwise we won't reach full opacity
+
+ // now we have to blow out the windings into explicit edge lists
+ n = 0;
+ for (i=0; i < windings; ++i)
+ n += wcount[i];
+
+ e = (stbtt__edge *) STBTT_malloc(sizeof(*e) * (n+1), userdata); // add an extra one as a sentinel
+ if (e == 0) return;
+ n = 0;
+
+ m=0;
+ for (i=0; i < windings; ++i) {
+ stbtt__point *p = pts + m;
+ m += wcount[i];
+ j = wcount[i]-1;
+ for (k=0; k < wcount[i]; j=k++) {
+ int a=k,b=j;
+ // skip the edge if horizontal
+ if (p[j].y == p[k].y)
+ continue;
+ // add edge from j to k to the list
+ e[n].invert = 0;
+ if (invert ? p[j].y > p[k].y : p[j].y < p[k].y) {
+ e[n].invert = 1;
+ a=j,b=k;
+ }
+ e[n].x0 = p[a].x * scale_x + shift_x;
+ e[n].y0 = (p[a].y * y_scale_inv + shift_y) * vsubsample;
+ e[n].x1 = p[b].x * scale_x + shift_x;
+ e[n].y1 = (p[b].y * y_scale_inv + shift_y) * vsubsample;
+ ++n;
+ }
+ }
+
+ // now sort the edges by their highest point (should snap to integer, and then by x)
+ //STBTT_sort(e, n, sizeof(e[0]), stbtt__edge_compare);
+ stbtt__sort_edges(e, n);
+
+ // now, traverse the scanlines and find the intersections on each scanline, use xor winding rule
+ stbtt__rasterize_sorted_edges(result, e, n, vsubsample, off_x, off_y, userdata);
+
+ STBTT_free(e, userdata);
}
static void stbtt__add_point(stbtt__point *points, int n, float x, float y)
{
- if (!points) return; // during first pass, it's unallocated
- points[n].x = x;
- points[n].y = y;
+ if (!points) return; // during first pass, it's unallocated
+ points[n].x = x;
+ points[n].y = y;
}
-// tesselate until threshhold p is happy... @TODO warped to compensate for non-linear stretching
+// tessellate until threshold p is happy... @TODO warped to compensate for non-linear stretching
static int stbtt__tesselate_curve(stbtt__point *points, int *num_points, float x0, float y0, float x1, float y1, float x2, float y2, float objspace_flatness_squared, int n)
{
- // midpoint
- float mx = (x0 + 2 * x1 + x2) / 4;
- float my = (y0 + 2 * y1 + y2) / 4;
- // versus directly drawn line
- float dx = (x0 + x2) / 2 - mx;
- float dy = (y0 + y2) / 2 - my;
- if (n > 16) // 65536 segments on one curve better be enough!
- return 1;
- if (dx*dx + dy * dy > objspace_flatness_squared) { // half-pixel error allowed... need to be smaller if AA
- stbtt__tesselate_curve(points, num_points, x0, y0, (x0 + x1) / 2.0f, (y0 + y1) / 2.0f, mx, my, objspace_flatness_squared, n + 1);
- stbtt__tesselate_curve(points, num_points, mx, my, (x1 + x2) / 2.0f, (y1 + y2) / 2.0f, x2, y2, objspace_flatness_squared, n + 1);
- }
- else {
- stbtt__add_point(points, *num_points, x2, y2);
- *num_points = *num_points + 1;
- }
- return 1;
+ // midpoint
+ float mx = (x0 + 2*x1 + x2)/4;
+ float my = (y0 + 2*y1 + y2)/4;
+ // versus directly drawn line
+ float dx = (x0+x2)/2 - mx;
+ float dy = (y0+y2)/2 - my;
+ if (n > 16) // 65536 segments on one curve better be enough!
+ return 1;
+ if (dx*dx+dy*dy > objspace_flatness_squared) { // half-pixel error allowed... need to be smaller if AA
+ stbtt__tesselate_curve(points, num_points, x0,y0, (x0+x1)/2.0f,(y0+y1)/2.0f, mx,my, objspace_flatness_squared,n+1);
+ stbtt__tesselate_curve(points, num_points, mx,my, (x1+x2)/2.0f,(y1+y2)/2.0f, x2,y2, objspace_flatness_squared,n+1);
+ } else {
+ stbtt__add_point(points, *num_points,x2,y2);
+ *num_points = *num_points+1;
+ }
+ return 1;
}
static void stbtt__tesselate_cubic(stbtt__point *points, int *num_points, float x0, float y0, float x1, float y1, float x2, float y2, float x3, float y3, float objspace_flatness_squared, int n)
{
- // @TODO this "flatness" calculation is just made-up nonsense that seems to work well enough
- float dx0 = x1 - x0;
- float dy0 = y1 - y0;
- float dx1 = x2 - x1;
- float dy1 = y2 - y1;
- float dx2 = x3 - x2;
- float dy2 = y3 - y2;
- float dx = x3 - x0;
- float dy = y3 - y0;
- float longlen = (float)(STBTT_sqrt(dx0*dx0 + dy0 * dy0) + STBTT_sqrt(dx1*dx1 + dy1 * dy1) + STBTT_sqrt(dx2*dx2 + dy2 * dy2));
- float shortlen = (float)STBTT_sqrt(dx*dx + dy * dy);
- float flatness_squared = longlen * longlen - shortlen * shortlen;
-
- if (n > 16) // 65536 segments on one curve better be enough!
- return;
-
- if (flatness_squared > objspace_flatness_squared) {
- float x01 = (x0 + x1) / 2;
- float y01 = (y0 + y1) / 2;
- float x12 = (x1 + x2) / 2;
- float y12 = (y1 + y2) / 2;
- float x23 = (x2 + x3) / 2;
- float y23 = (y2 + y3) / 2;
-
- float xa = (x01 + x12) / 2;
- float ya = (y01 + y12) / 2;
- float xb = (x12 + x23) / 2;
- float yb = (y12 + y23) / 2;
-
- float mx = (xa + xb) / 2;
- float my = (ya + yb) / 2;
-
- stbtt__tesselate_cubic(points, num_points, x0, y0, x01, y01, xa, ya, mx, my, objspace_flatness_squared, n + 1);
- stbtt__tesselate_cubic(points, num_points, mx, my, xb, yb, x23, y23, x3, y3, objspace_flatness_squared, n + 1);
- }
- else {
- stbtt__add_point(points, *num_points, x3, y3);
- *num_points = *num_points + 1;
- }
+ // @TODO this "flatness" calculation is just made-up nonsense that seems to work well enough
+ float dx0 = x1-x0;
+ float dy0 = y1-y0;
+ float dx1 = x2-x1;
+ float dy1 = y2-y1;
+ float dx2 = x3-x2;
+ float dy2 = y3-y2;
+ float dx = x3-x0;
+ float dy = y3-y0;
+ float longlen = (float) (STBTT_sqrt(dx0*dx0+dy0*dy0)+STBTT_sqrt(dx1*dx1+dy1*dy1)+STBTT_sqrt(dx2*dx2+dy2*dy2));
+ float shortlen = (float) STBTT_sqrt(dx*dx+dy*dy);
+ float flatness_squared = longlen*longlen-shortlen*shortlen;
+
+ if (n > 16) // 65536 segments on one curve better be enough!
+ return;
+
+ if (flatness_squared > objspace_flatness_squared) {
+ float x01 = (x0+x1)/2;
+ float y01 = (y0+y1)/2;
+ float x12 = (x1+x2)/2;
+ float y12 = (y1+y2)/2;
+ float x23 = (x2+x3)/2;
+ float y23 = (y2+y3)/2;
+
+ float xa = (x01+x12)/2;
+ float ya = (y01+y12)/2;
+ float xb = (x12+x23)/2;
+ float yb = (y12+y23)/2;
+
+ float mx = (xa+xb)/2;
+ float my = (ya+yb)/2;
+
+ stbtt__tesselate_cubic(points, num_points, x0,y0, x01,y01, xa,ya, mx,my, objspace_flatness_squared,n+1);
+ stbtt__tesselate_cubic(points, num_points, mx,my, xb,yb, x23,y23, x3,y3, objspace_flatness_squared,n+1);
+ } else {
+ stbtt__add_point(points, *num_points,x3,y3);
+ *num_points = *num_points+1;
+ }
}
// returns number of contours
static stbtt__point *stbtt_FlattenCurves(stbtt_vertex *vertices, int num_verts, float objspace_flatness, int **contour_lengths, int *num_contours, void *userdata)
{
- stbtt__point *points = 0;
- int num_points = 0;
-
- float objspace_flatness_squared = objspace_flatness * objspace_flatness;
- int i, n = 0, start = 0, pass;
-
- // count how many "moves" there are to get the contour count
- for (i = 0; i < num_verts; ++i)
- if (vertices[i].type == STBTT_vmove)
- ++n;
-
- *num_contours = n;
- if (n == 0) return 0;
-
- *contour_lengths = (int *)STBTT_malloc(sizeof(**contour_lengths) * n, userdata);
-
- if (*contour_lengths == 0) {
- *num_contours = 0;
- return 0;
- }
-
- // make two passes through the points so we don't need to realloc
- for (pass = 0; pass < 2; ++pass) {
- float x = 0, y = 0;
- if (pass == 1) {
- points = (stbtt__point *)STBTT_malloc(num_points * sizeof(points[0]), userdata);
- if (points == NULL) goto error;
- }
- num_points = 0;
- n = -1;
- for (i = 0; i < num_verts; ++i) {
- switch (vertices[i].type) {
+ stbtt__point *points=0;
+ int num_points=0;
+
+ float objspace_flatness_squared = objspace_flatness * objspace_flatness;
+ int i,n=0,start=0, pass;
+
+ // count how many "moves" there are to get the contour count
+ for (i=0; i < num_verts; ++i)
+ if (vertices[i].type == STBTT_vmove)
+ ++n;
+
+ *num_contours = n;
+ if (n == 0) return 0;
+
+ *contour_lengths = (int *) STBTT_malloc(sizeof(**contour_lengths) * n, userdata);
+
+ if (*contour_lengths == 0) {
+ *num_contours = 0;
+ return 0;
+ }
+
+ // make two passes through the points so we don't need to realloc
+ for (pass=0; pass < 2; ++pass) {
+ float x=0,y=0;
+ if (pass == 1) {
+ points = (stbtt__point *) STBTT_malloc(num_points * sizeof(points[0]), userdata);
+ if (points == NULL) goto error;
+ }
+ num_points = 0;
+ n= -1;
+ for (i=0; i < num_verts; ++i) {
+ switch (vertices[i].type) {
case STBTT_vmove:
- // start the next contour
- if (n >= 0)
- (*contour_lengths)[n] = num_points - start;
- ++n;
- start = num_points;
-
- x = vertices[i].x, y = vertices[i].y;
- stbtt__add_point(points, num_points++, x, y);
- break;
+ // start the next contour
+ if (n >= 0)
+ (*contour_lengths)[n] = num_points - start;
+ ++n;
+ start = num_points;
+
+ x = vertices[i].x, y = vertices[i].y;
+ stbtt__add_point(points, num_points++, x,y);
+ break;
case STBTT_vline:
- x = vertices[i].x, y = vertices[i].y;
- stbtt__add_point(points, num_points++, x, y);
- break;
+ x = vertices[i].x, y = vertices[i].y;
+ stbtt__add_point(points, num_points++, x, y);
+ break;
case STBTT_vcurve:
- stbtt__tesselate_curve(points, &num_points, x, y,
- vertices[i].cx, vertices[i].cy,
- vertices[i].x, vertices[i].y,
- objspace_flatness_squared, 0);
- x = vertices[i].x, y = vertices[i].y;
- break;
+ stbtt__tesselate_curve(points, &num_points, x,y,
+ vertices[i].cx, vertices[i].cy,
+ vertices[i].x, vertices[i].y,
+ objspace_flatness_squared, 0);
+ x = vertices[i].x, y = vertices[i].y;
+ break;
case STBTT_vcubic:
- stbtt__tesselate_cubic(points, &num_points, x, y,
- vertices[i].cx, vertices[i].cy,
- vertices[i].cx1, vertices[i].cy1,
- vertices[i].x, vertices[i].y,
- objspace_flatness_squared, 0);
- x = vertices[i].x, y = vertices[i].y;
- break;
- }
- }
- (*contour_lengths)[n] = num_points - start;
- }
-
- return points;
+ stbtt__tesselate_cubic(points, &num_points, x,y,
+ vertices[i].cx, vertices[i].cy,
+ vertices[i].cx1, vertices[i].cy1,
+ vertices[i].x, vertices[i].y,
+ objspace_flatness_squared, 0);
+ x = vertices[i].x, y = vertices[i].y;
+ break;
+ }
+ }
+ (*contour_lengths)[n] = num_points - start;
+ }
+
+ return points;
error:
- STBTT_free(points, userdata);
- STBTT_free(*contour_lengths, userdata);
- *contour_lengths = 0;
- *num_contours = 0;
- return NULL;
+ STBTT_free(points, userdata);
+ STBTT_free(*contour_lengths, userdata);
+ *contour_lengths = 0;
+ *num_contours = 0;
+ return NULL;
}
STBTT_DEF void stbtt_Rasterize(stbtt__bitmap *result, float flatness_in_pixels, stbtt_vertex *vertices, int num_verts, float scale_x, float scale_y, float shift_x, float shift_y, int x_off, int y_off, int invert, void *userdata)
{
- float scale = scale_x > scale_y ? scale_y : scale_x;
- int winding_count = 0;
- int *winding_lengths = NULL;
- stbtt__point *windings = stbtt_FlattenCurves(vertices, num_verts, flatness_in_pixels / scale, &winding_lengths, &winding_count, userdata);
- if (windings) {
- stbtt__rasterize(result, windings, winding_lengths, winding_count, scale_x, scale_y, shift_x, shift_y, x_off, y_off, invert, userdata);
- STBTT_free(winding_lengths, userdata);
- STBTT_free(windings, userdata);
- }
+ float scale = scale_x > scale_y ? scale_y : scale_x;
+ int winding_count = 0;
+ int *winding_lengths = NULL;
+ stbtt__point *windings = stbtt_FlattenCurves(vertices, num_verts, flatness_in_pixels / scale, &winding_lengths, &winding_count, userdata);
+ if (windings) {
+ stbtt__rasterize(result, windings, winding_lengths, winding_count, scale_x, scale_y, shift_x, shift_y, x_off, y_off, invert, userdata);
+ STBTT_free(winding_lengths, userdata);
+ STBTT_free(windings, userdata);
+ }
}
STBTT_DEF void stbtt_FreeBitmap(unsigned char *bitmap, void *userdata)
{
- STBTT_free(bitmap, userdata);
+ STBTT_free(bitmap, userdata);
}
STBTT_DEF unsigned char *stbtt_GetGlyphBitmapSubpixel(const stbtt_fontinfo *info, float scale_x, float scale_y, float shift_x, float shift_y, int glyph, int *width, int *height, int *xoff, int *yoff)
{
- int ix0, iy0, ix1, iy1;
- stbtt__bitmap gbm;
- stbtt_vertex *vertices;
- int num_verts = stbtt_GetGlyphShape(info, glyph, &vertices);
+ int ix0,iy0,ix1,iy1;
+ stbtt__bitmap gbm;
+ stbtt_vertex *vertices;
+ int num_verts = stbtt_GetGlyphShape(info, glyph, &vertices);
- if (scale_x == 0) scale_x = scale_y;
- if (scale_y == 0) {
- if (scale_x == 0) {
- STBTT_free(vertices, info->userdata);
- return NULL;
- }
- scale_y = scale_x;
- }
+ if (scale_x == 0) scale_x = scale_y;
+ if (scale_y == 0) {
+ if (scale_x == 0) {
+ STBTT_free(vertices, info->userdata);
+ return NULL;
+ }
+ scale_y = scale_x;
+ }
- stbtt_GetGlyphBitmapBoxSubpixel(info, glyph, scale_x, scale_y, shift_x, shift_y, &ix0, &iy0, &ix1, &iy1);
+ stbtt_GetGlyphBitmapBoxSubpixel(info, glyph, scale_x, scale_y, shift_x, shift_y, &ix0,&iy0,&ix1,&iy1);
- // now we get the size
- gbm.w = (ix1 - ix0);
- gbm.h = (iy1 - iy0);
- gbm.pixels = NULL; // in case we error
+ // now we get the size
+ gbm.w = (ix1 - ix0);
+ gbm.h = (iy1 - iy0);
+ gbm.pixels = NULL; // in case we error
- if (width) *width = gbm.w;
- if (height) *height = gbm.h;
- if (xoff) *xoff = ix0;
- if (yoff) *yoff = iy0;
+ if (width ) *width = gbm.w;
+ if (height) *height = gbm.h;
+ if (xoff ) *xoff = ix0;
+ if (yoff ) *yoff = iy0;
- if (gbm.w && gbm.h) {
- gbm.pixels = (unsigned char *)STBTT_malloc(gbm.w * gbm.h, info->userdata);
- if (gbm.pixels) {
- gbm.stride = gbm.w;
+ if (gbm.w && gbm.h) {
+ gbm.pixels = (unsigned char *) STBTT_malloc(gbm.w * gbm.h, info->userdata);
+ if (gbm.pixels) {
+ gbm.stride = gbm.w;
- stbtt_Rasterize(&gbm, 0.35f, vertices, num_verts, scale_x, scale_y, shift_x, shift_y, ix0, iy0, 1, info->userdata);
- }
- }
- STBTT_free(vertices, info->userdata);
- return gbm.pixels;
+ stbtt_Rasterize(&gbm, 0.35f, vertices, num_verts, scale_x, scale_y, shift_x, shift_y, ix0, iy0, 1, info->userdata);
+ }
+ }
+ STBTT_free(vertices, info->userdata);
+ return gbm.pixels;
}
STBTT_DEF unsigned char *stbtt_GetGlyphBitmap(const stbtt_fontinfo *info, float scale_x, float scale_y, int glyph, int *width, int *height, int *xoff, int *yoff)
{
- return stbtt_GetGlyphBitmapSubpixel(info, scale_x, scale_y, 0.0f, 0.0f, glyph, width, height, xoff, yoff);
+ return stbtt_GetGlyphBitmapSubpixel(info, scale_x, scale_y, 0.0f, 0.0f, glyph, width, height, xoff, yoff);
}
STBTT_DEF void stbtt_MakeGlyphBitmapSubpixel(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int glyph)
{
- int ix0, iy0;
- stbtt_vertex *vertices;
- int num_verts = stbtt_GetGlyphShape(info, glyph, &vertices);
- stbtt__bitmap gbm;
+ int ix0,iy0;
+ stbtt_vertex *vertices;
+ int num_verts = stbtt_GetGlyphShape(info, glyph, &vertices);
+ stbtt__bitmap gbm;
- stbtt_GetGlyphBitmapBoxSubpixel(info, glyph, scale_x, scale_y, shift_x, shift_y, &ix0, &iy0, 0, 0);
- gbm.pixels = output;
- gbm.w = out_w;
- gbm.h = out_h;
- gbm.stride = out_stride;
+ stbtt_GetGlyphBitmapBoxSubpixel(info, glyph, scale_x, scale_y, shift_x, shift_y, &ix0,&iy0,0,0);
+ gbm.pixels = output;
+ gbm.w = out_w;
+ gbm.h = out_h;
+ gbm.stride = out_stride;
- if (gbm.w && gbm.h)
- stbtt_Rasterize(&gbm, 0.35f, vertices, num_verts, scale_x, scale_y, shift_x, shift_y, ix0, iy0, 1, info->userdata);
+ if (gbm.w && gbm.h)
+ stbtt_Rasterize(&gbm, 0.35f, vertices, num_verts, scale_x, scale_y, shift_x, shift_y, ix0,iy0, 1, info->userdata);
- STBTT_free(vertices, info->userdata);
+ STBTT_free(vertices, info->userdata);
}
STBTT_DEF void stbtt_MakeGlyphBitmap(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, int glyph)
{
- stbtt_MakeGlyphBitmapSubpixel(info, output, out_w, out_h, out_stride, scale_x, scale_y, 0.0f, 0.0f, glyph);
+ stbtt_MakeGlyphBitmapSubpixel(info, output, out_w, out_h, out_stride, scale_x, scale_y, 0.0f,0.0f, glyph);
}
STBTT_DEF unsigned char *stbtt_GetCodepointBitmapSubpixel(const stbtt_fontinfo *info, float scale_x, float scale_y, float shift_x, float shift_y, int codepoint, int *width, int *height, int *xoff, int *yoff)
{
- return stbtt_GetGlyphBitmapSubpixel(info, scale_x, scale_y, shift_x, shift_y, stbtt_FindGlyphIndex(info, codepoint), width, height, xoff, yoff);
+ return stbtt_GetGlyphBitmapSubpixel(info, scale_x, scale_y,shift_x,shift_y, stbtt_FindGlyphIndex(info,codepoint), width,height,xoff,yoff);
}
STBTT_DEF void stbtt_MakeCodepointBitmapSubpixelPrefilter(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int oversample_x, int oversample_y, float *sub_x, float *sub_y, int codepoint)
{
- stbtt_MakeGlyphBitmapSubpixelPrefilter(info, output, out_w, out_h, out_stride, scale_x, scale_y, shift_x, shift_y, oversample_x, oversample_y, sub_x, sub_y, stbtt_FindGlyphIndex(info, codepoint));
+ stbtt_MakeGlyphBitmapSubpixelPrefilter(info, output, out_w, out_h, out_stride, scale_x, scale_y, shift_x, shift_y, oversample_x, oversample_y, sub_x, sub_y, stbtt_FindGlyphIndex(info,codepoint));
}
STBTT_DEF void stbtt_MakeCodepointBitmapSubpixel(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int codepoint)
{
- stbtt_MakeGlyphBitmapSubpixel(info, output, out_w, out_h, out_stride, scale_x, scale_y, shift_x, shift_y, stbtt_FindGlyphIndex(info, codepoint));
+ stbtt_MakeGlyphBitmapSubpixel(info, output, out_w, out_h, out_stride, scale_x, scale_y, shift_x, shift_y, stbtt_FindGlyphIndex(info,codepoint));
}
STBTT_DEF unsigned char *stbtt_GetCodepointBitmap(const stbtt_fontinfo *info, float scale_x, float scale_y, int codepoint, int *width, int *height, int *xoff, int *yoff)
{
- return stbtt_GetCodepointBitmapSubpixel(info, scale_x, scale_y, 0.0f, 0.0f, codepoint, width, height, xoff, yoff);
+ return stbtt_GetCodepointBitmapSubpixel(info, scale_x, scale_y, 0.0f,0.0f, codepoint, width,height,xoff,yoff);
}
STBTT_DEF void stbtt_MakeCodepointBitmap(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, int codepoint)
{
- stbtt_MakeCodepointBitmapSubpixel(info, output, out_w, out_h, out_stride, scale_x, scale_y, 0.0f, 0.0f, codepoint);
+ stbtt_MakeCodepointBitmapSubpixel(info, output, out_w, out_h, out_stride, scale_x, scale_y, 0.0f,0.0f, codepoint);
}
//////////////////////////////////////////////////////////////////////////////
@@ -3672,70 +3811,70 @@ STBTT_DEF void stbtt_MakeCodepointBitmap(const stbtt_fontinfo *info, unsigned ch
// This is SUPER-CRAPPY packing to keep source code small
static int stbtt_BakeFontBitmap_internal(unsigned char *data, int offset, // font location (use offset=0 for plain .ttf)
- float pixel_height, // height of font in pixels
- unsigned char *pixels, int pw, int ph, // bitmap to be filled in
- int first_char, int num_chars, // characters to bake
- stbtt_bakedchar *chardata)
-{
- float scale;
- int x, y, bottom_y, i;
- stbtt_fontinfo f;
- f.userdata = NULL;
- if (!stbtt_InitFont(&f, data, offset))
- return -1;
- STBTT_memset(pixels, 0, pw*ph); // background of 0 around pixels
- x = y = 1;
- bottom_y = 1;
-
- scale = stbtt_ScaleForPixelHeight(&f, pixel_height);
-
- for (i = 0; i < num_chars; ++i) {
- int advance, lsb, x0, y0, x1, y1, gw, gh;
- int g = stbtt_FindGlyphIndex(&f, first_char + i);
- stbtt_GetGlyphHMetrics(&f, g, &advance, &lsb);
- stbtt_GetGlyphBitmapBox(&f, g, scale, scale, &x0, &y0, &x1, &y1);
- gw = x1 - x0;
- gh = y1 - y0;
- if (x + gw + 1 >= pw)
- y = bottom_y, x = 1; // advance to next row
- if (y + gh + 1 >= ph) // check if it fits vertically AFTER potentially moving to next row
- return -i;
- STBTT_assert(x + gw < pw);
- STBTT_assert(y + gh < ph);
- stbtt_MakeGlyphBitmap(&f, pixels + x + y * pw, gw, gh, pw, scale, scale, g);
- chardata[i].x0 = (stbtt_int16)x;
- chardata[i].y0 = (stbtt_int16)y;
- chardata[i].x1 = (stbtt_int16)(x + gw);
- chardata[i].y1 = (stbtt_int16)(y + gh);
- chardata[i].xadvance = scale * advance;
- chardata[i].xoff = (float)x0;
- chardata[i].yoff = (float)y0;
- x = x + gw + 1;
- if (y + gh + 1 > bottom_y)
- bottom_y = y + gh + 1;
- }
- return bottom_y;
+ float pixel_height, // height of font in pixels
+ unsigned char *pixels, int pw, int ph, // bitmap to be filled in
+ int first_char, int num_chars, // characters to bake
+ stbtt_bakedchar *chardata)
+{
+ float scale;
+ int x,y,bottom_y, i;
+ stbtt_fontinfo f;
+ f.userdata = NULL;
+ if (!stbtt_InitFont(&f, data, offset))
+ return -1;
+ STBTT_memset(pixels, 0, pw*ph); // background of 0 around pixels
+ x=y=1;
+ bottom_y = 1;
+
+ scale = stbtt_ScaleForPixelHeight(&f, pixel_height);
+
+ for (i=0; i < num_chars; ++i) {
+ int advance, lsb, x0,y0,x1,y1,gw,gh;
+ int g = stbtt_FindGlyphIndex(&f, first_char + i);
+ stbtt_GetGlyphHMetrics(&f, g, &advance, &lsb);
+ stbtt_GetGlyphBitmapBox(&f, g, scale,scale, &x0,&y0,&x1,&y1);
+ gw = x1-x0;
+ gh = y1-y0;
+ if (x + gw + 1 >= pw)
+ y = bottom_y, x = 1; // advance to next row
+ if (y + gh + 1 >= ph) // check if it fits vertically AFTER potentially moving to next row
+ return -i;
+ STBTT_assert(x+gw < pw);
+ STBTT_assert(y+gh < ph);
+ stbtt_MakeGlyphBitmap(&f, pixels+x+y*pw, gw,gh,pw, scale,scale, g);
+ chardata[i].x0 = (stbtt_int16) x;
+ chardata[i].y0 = (stbtt_int16) y;
+ chardata[i].x1 = (stbtt_int16) (x + gw);
+ chardata[i].y1 = (stbtt_int16) (y + gh);
+ chardata[i].xadvance = scale * advance;
+ chardata[i].xoff = (float) x0;
+ chardata[i].yoff = (float) y0;
+ x = x + gw + 1;
+ if (y+gh+1 > bottom_y)
+ bottom_y = y+gh+1;
+ }
+ return bottom_y;
}
STBTT_DEF void stbtt_GetBakedQuad(const stbtt_bakedchar *chardata, int pw, int ph, int char_index, float *xpos, float *ypos, stbtt_aligned_quad *q, int opengl_fillrule)
{
- float d3d_bias = opengl_fillrule ? 0 : -0.5f;
- float ipw = 1.0f / pw, iph = 1.0f / ph;
- const stbtt_bakedchar *b = chardata + char_index;
- int round_x = STBTT_ifloor((*xpos + b->xoff) + 0.5f);
- int round_y = STBTT_ifloor((*ypos + b->yoff) + 0.5f);
+ float d3d_bias = opengl_fillrule ? 0 : -0.5f;
+ float ipw = 1.0f / pw, iph = 1.0f / ph;
+ const stbtt_bakedchar *b = chardata + char_index;
+ int round_x = STBTT_ifloor((*xpos + b->xoff) + 0.5f);
+ int round_y = STBTT_ifloor((*ypos + b->yoff) + 0.5f);
- q->x0 = round_x + d3d_bias;
- q->y0 = round_y + d3d_bias;
- q->x1 = round_x + b->x1 - b->x0 + d3d_bias;
- q->y1 = round_y + b->y1 - b->y0 + d3d_bias;
+ q->x0 = round_x + d3d_bias;
+ q->y0 = round_y + d3d_bias;
+ q->x1 = round_x + b->x1 - b->x0 + d3d_bias;
+ q->y1 = round_y + b->y1 - b->y0 + d3d_bias;
- q->s0 = b->x0 * ipw;
- q->t0 = b->y0 * iph;
- q->s1 = b->x1 * ipw;
- q->t1 = b->y1 * iph;
+ q->s0 = b->x0 * ipw;
+ q->t0 = b->y0 * iph;
+ q->s1 = b->x1 * ipw;
+ q->t1 = b->y1 * iph;
- *xpos += b->xadvance;
+ *xpos += b->xadvance;
}
//////////////////////////////////////////////////////////////////////////////
@@ -3760,51 +3899,51 @@ typedef int stbrp_coord;
typedef struct
{
- int width, height;
- int x, y, bottom_y;
+ int width,height;
+ int x,y,bottom_y;
} stbrp_context;
typedef struct
{
- unsigned char x;
+ unsigned char x;
} stbrp_node;
struct stbrp_rect
{
- stbrp_coord x, y;
- int id, w, h, was_packed;
+ stbrp_coord x,y;
+ int id,w,h,was_packed;
};
static void stbrp_init_target(stbrp_context *con, int pw, int ph, stbrp_node *nodes, int num_nodes)
{
- con->width = pw;
- con->height = ph;
- con->x = 0;
- con->y = 0;
- con->bottom_y = 0;
- STBTT__NOTUSED(nodes);
- STBTT__NOTUSED(num_nodes);
+ con->width = pw;
+ con->height = ph;
+ con->x = 0;
+ con->y = 0;
+ con->bottom_y = 0;
+ STBTT__NOTUSED(nodes);
+ STBTT__NOTUSED(num_nodes);
}
static void stbrp_pack_rects(stbrp_context *con, stbrp_rect *rects, int num_rects)
{
- int i;
- for (i = 0; i < num_rects; ++i) {
- if (con->x + rects[i].w > con->width) {
- con->x = 0;
- con->y = con->bottom_y;
- }
- if (con->y + rects[i].h > con->height)
- break;
- rects[i].x = con->x;
- rects[i].y = con->y;
- rects[i].was_packed = 1;
- con->x += rects[i].w;
- if (con->y + rects[i].h > con->bottom_y)
- con->bottom_y = con->y + rects[i].h;
- }
- for (; i < num_rects; ++i)
- rects[i].was_packed = 0;
+ int i;
+ for (i=0; i < num_rects; ++i) {
+ if (con->x + rects[i].w > con->width) {
+ con->x = 0;
+ con->y = con->bottom_y;
+ }
+ if (con->y + rects[i].h > con->height)
+ break;
+ rects[i].x = con->x;
+ rects[i].y = con->y;
+ rects[i].was_packed = 1;
+ con->x += rects[i].w;
+ if (con->y + rects[i].h > con->bottom_y)
+ con->bottom_y = con->y + rects[i].h;
+ }
+ for ( ; i < num_rects; ++i)
+ rects[i].was_packed = 0;
}
#endif
@@ -3817,404 +3956,435 @@ static void stbrp_pack_rects(stbrp_context *con, stbrp_rect *rects, int num_rect
STBTT_DEF int stbtt_PackBegin(stbtt_pack_context *spc, unsigned char *pixels, int pw, int ph, int stride_in_bytes, int padding, void *alloc_context)
{
- stbrp_context *context = (stbrp_context *)STBTT_malloc(sizeof(*context), alloc_context);
- int num_nodes = pw - padding;
- stbrp_node *nodes = (stbrp_node *)STBTT_malloc(sizeof(*nodes) * num_nodes, alloc_context);
+ stbrp_context *context = (stbrp_context *) STBTT_malloc(sizeof(*context) ,alloc_context);
+ int num_nodes = pw - padding;
+ stbrp_node *nodes = (stbrp_node *) STBTT_malloc(sizeof(*nodes ) * num_nodes,alloc_context);
- if (context == NULL || nodes == NULL) {
- if (context != NULL) STBTT_free(context, alloc_context);
- if (nodes != NULL) STBTT_free(nodes, alloc_context);
- return 0;
- }
+ if (context == NULL || nodes == NULL) {
+ if (context != NULL) STBTT_free(context, alloc_context);
+ if (nodes != NULL) STBTT_free(nodes , alloc_context);
+ return 0;
+ }
- spc->user_allocator_context = alloc_context;
- spc->width = pw;
- spc->height = ph;
- spc->pixels = pixels;
- spc->pack_info = context;
- spc->nodes = nodes;
- spc->padding = padding;
- spc->stride_in_bytes = stride_in_bytes != 0 ? stride_in_bytes : pw;
- spc->h_oversample = 1;
- spc->v_oversample = 1;
+ spc->user_allocator_context = alloc_context;
+ spc->width = pw;
+ spc->height = ph;
+ spc->pixels = pixels;
+ spc->pack_info = context;
+ spc->nodes = nodes;
+ spc->padding = padding;
+ spc->stride_in_bytes = stride_in_bytes != 0 ? stride_in_bytes : pw;
+ spc->h_oversample = 1;
+ spc->v_oversample = 1;
+ spc->skip_missing = 0;
- stbrp_init_target(context, pw - padding, ph - padding, nodes, num_nodes);
+ stbrp_init_target(context, pw-padding, ph-padding, nodes, num_nodes);
- if (pixels)
- STBTT_memset(pixels, 0, pw*ph); // background of 0 around pixels
+ if (pixels)
+ STBTT_memset(pixels, 0, pw*ph); // background of 0 around pixels
- return 1;
+ return 1;
}
-STBTT_DEF void stbtt_PackEnd(stbtt_pack_context *spc)
+STBTT_DEF void stbtt_PackEnd (stbtt_pack_context *spc)
{
- STBTT_free(spc->nodes, spc->user_allocator_context);
- STBTT_free(spc->pack_info, spc->user_allocator_context);
+ STBTT_free(spc->nodes , spc->user_allocator_context);
+ STBTT_free(spc->pack_info, spc->user_allocator_context);
}
STBTT_DEF void stbtt_PackSetOversampling(stbtt_pack_context *spc, unsigned int h_oversample, unsigned int v_oversample)
{
- STBTT_assert(h_oversample <= STBTT_MAX_OVERSAMPLE);
- STBTT_assert(v_oversample <= STBTT_MAX_OVERSAMPLE);
- if (h_oversample <= STBTT_MAX_OVERSAMPLE)
- spc->h_oversample = h_oversample;
- if (v_oversample <= STBTT_MAX_OVERSAMPLE)
- spc->v_oversample = v_oversample;
+ STBTT_assert(h_oversample <= STBTT_MAX_OVERSAMPLE);
+ STBTT_assert(v_oversample <= STBTT_MAX_OVERSAMPLE);
+ if (h_oversample <= STBTT_MAX_OVERSAMPLE)
+ spc->h_oversample = h_oversample;
+ if (v_oversample <= STBTT_MAX_OVERSAMPLE)
+ spc->v_oversample = v_oversample;
+}
+
+STBTT_DEF void stbtt_PackSetSkipMissingCodepoints(stbtt_pack_context *spc, int skip)
+{
+ spc->skip_missing = skip;
}
#define STBTT__OVER_MASK (STBTT_MAX_OVERSAMPLE-1)
static void stbtt__h_prefilter(unsigned char *pixels, int w, int h, int stride_in_bytes, unsigned int kernel_width)
{
- unsigned char buffer[STBTT_MAX_OVERSAMPLE];
- int safe_w = w - kernel_width;
- int j;
- STBTT_memset(buffer, 0, STBTT_MAX_OVERSAMPLE); // suppress bogus warning from VS2013 -analyze
- for (j = 0; j < h; ++j) {
- int i;
- unsigned int total;
- STBTT_memset(buffer, 0, kernel_width);
-
- total = 0;
-
- // make kernel_width a constant in common cases so compiler can optimize out the divide
- switch (kernel_width) {
- case 2:
- for (i = 0; i <= safe_w; ++i) {
- total += pixels[i] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i];
- pixels[i] = (unsigned char)(total / 2);
+ unsigned char buffer[STBTT_MAX_OVERSAMPLE];
+ int safe_w = w - kernel_width;
+ int j;
+ STBTT_memset(buffer, 0, STBTT_MAX_OVERSAMPLE); // suppress bogus warning from VS2013 -analyze
+ for (j=0; j < h; ++j) {
+ int i;
+ unsigned int total;
+ STBTT_memset(buffer, 0, kernel_width);
+
+ total = 0;
+
+ // make kernel_width a constant in common cases so compiler can optimize out the divide
+ switch (kernel_width) {
+ case 2:
+ for (i=0; i <= safe_w; ++i) {
+ total += pixels[i] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i];
+ pixels[i] = (unsigned char) (total / 2);
}
break;
- case 3:
- for (i = 0; i <= safe_w; ++i) {
- total += pixels[i] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i];
- pixels[i] = (unsigned char)(total / 3);
+ case 3:
+ for (i=0; i <= safe_w; ++i) {
+ total += pixels[i] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i];
+ pixels[i] = (unsigned char) (total / 3);
}
break;
- case 4:
- for (i = 0; i <= safe_w; ++i) {
- total += pixels[i] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i];
- pixels[i] = (unsigned char)(total / 4);
+ case 4:
+ for (i=0; i <= safe_w; ++i) {
+ total += pixels[i] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i];
+ pixels[i] = (unsigned char) (total / 4);
}
break;
- case 5:
- for (i = 0; i <= safe_w; ++i) {
- total += pixels[i] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i];
- pixels[i] = (unsigned char)(total / 5);
+ case 5:
+ for (i=0; i <= safe_w; ++i) {
+ total += pixels[i] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i];
+ pixels[i] = (unsigned char) (total / 5);
}
break;
- default:
- for (i = 0; i <= safe_w; ++i) {
- total += pixels[i] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i];
- pixels[i] = (unsigned char)(total / kernel_width);
+ default:
+ for (i=0; i <= safe_w; ++i) {
+ total += pixels[i] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i];
+ pixels[i] = (unsigned char) (total / kernel_width);
}
break;
- }
+ }
- for (; i < w; ++i) {
- STBTT_assert(pixels[i] == 0);
- total -= buffer[i & STBTT__OVER_MASK];
- pixels[i] = (unsigned char)(total / kernel_width);
- }
+ for (; i < w; ++i) {
+ STBTT_assert(pixels[i] == 0);
+ total -= buffer[i & STBTT__OVER_MASK];
+ pixels[i] = (unsigned char) (total / kernel_width);
+ }
- pixels += stride_in_bytes;
- }
+ pixels += stride_in_bytes;
+ }
}
static void stbtt__v_prefilter(unsigned char *pixels, int w, int h, int stride_in_bytes, unsigned int kernel_width)
{
- unsigned char buffer[STBTT_MAX_OVERSAMPLE];
- int safe_h = h - kernel_width;
- int j;
- STBTT_memset(buffer, 0, STBTT_MAX_OVERSAMPLE); // suppress bogus warning from VS2013 -analyze
- for (j = 0; j < w; ++j) {
- int i;
- unsigned int total;
- STBTT_memset(buffer, 0, kernel_width);
-
- total = 0;
-
- // make kernel_width a constant in common cases so compiler can optimize out the divide
- switch (kernel_width) {
- case 2:
- for (i = 0; i <= safe_h; ++i) {
- total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
- pixels[i*stride_in_bytes] = (unsigned char)(total / 2);
+ unsigned char buffer[STBTT_MAX_OVERSAMPLE];
+ int safe_h = h - kernel_width;
+ int j;
+ STBTT_memset(buffer, 0, STBTT_MAX_OVERSAMPLE); // suppress bogus warning from VS2013 -analyze
+ for (j=0; j < w; ++j) {
+ int i;
+ unsigned int total;
+ STBTT_memset(buffer, 0, kernel_width);
+
+ total = 0;
+
+ // make kernel_width a constant in common cases so compiler can optimize out the divide
+ switch (kernel_width) {
+ case 2:
+ for (i=0; i <= safe_h; ++i) {
+ total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
+ pixels[i*stride_in_bytes] = (unsigned char) (total / 2);
}
break;
- case 3:
- for (i = 0; i <= safe_h; ++i) {
- total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
- pixels[i*stride_in_bytes] = (unsigned char)(total / 3);
+ case 3:
+ for (i=0; i <= safe_h; ++i) {
+ total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
+ pixels[i*stride_in_bytes] = (unsigned char) (total / 3);
}
break;
- case 4:
- for (i = 0; i <= safe_h; ++i) {
- total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
- pixels[i*stride_in_bytes] = (unsigned char)(total / 4);
+ case 4:
+ for (i=0; i <= safe_h; ++i) {
+ total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
+ pixels[i*stride_in_bytes] = (unsigned char) (total / 4);
}
break;
- case 5:
- for (i = 0; i <= safe_h; ++i) {
- total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
- pixels[i*stride_in_bytes] = (unsigned char)(total / 5);
+ case 5:
+ for (i=0; i <= safe_h; ++i) {
+ total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
+ pixels[i*stride_in_bytes] = (unsigned char) (total / 5);
}
break;
- default:
- for (i = 0; i <= safe_h; ++i) {
- total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
- buffer[(i + kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
- pixels[i*stride_in_bytes] = (unsigned char)(total / kernel_width);
+ default:
+ for (i=0; i <= safe_h; ++i) {
+ total += pixels[i*stride_in_bytes] - buffer[i & STBTT__OVER_MASK];
+ buffer[(i+kernel_width) & STBTT__OVER_MASK] = pixels[i*stride_in_bytes];
+ pixels[i*stride_in_bytes] = (unsigned char) (total / kernel_width);
}
break;
- }
+ }
- for (; i < h; ++i) {
- STBTT_assert(pixels[i*stride_in_bytes] == 0);
- total -= buffer[i & STBTT__OVER_MASK];
- pixels[i*stride_in_bytes] = (unsigned char)(total / kernel_width);
- }
+ for (; i < h; ++i) {
+ STBTT_assert(pixels[i*stride_in_bytes] == 0);
+ total -= buffer[i & STBTT__OVER_MASK];
+ pixels[i*stride_in_bytes] = (unsigned char) (total / kernel_width);
+ }
- pixels += 1;
- }
+ pixels += 1;
+ }
}
static float stbtt__oversample_shift(int oversample)
{
- if (!oversample)
- return 0.0f;
+ if (!oversample)
+ return 0.0f;
- // The prefilter is a box filter of width "oversample",
- // which shifts phase by (oversample - 1)/2 pixels in
- // oversampled space. We want to shift in the opposite
- // direction to counter this.
- return (float)-(oversample - 1) / (2.0f * (float)oversample);
+ // The prefilter is a box filter of width "oversample",
+ // which shifts phase by (oversample - 1)/2 pixels in
+ // oversampled space. We want to shift in the opposite
+ // direction to counter this.
+ return (float)-(oversample - 1) / (2.0f * (float)oversample);
}
// rects array must be big enough to accommodate all characters in the given ranges
STBTT_DEF int stbtt_PackFontRangesGatherRects(stbtt_pack_context *spc, const stbtt_fontinfo *info, stbtt_pack_range *ranges, int num_ranges, stbrp_rect *rects)
{
- int i, j, k;
-
- k = 0;
- for (i = 0; i < num_ranges; ++i) {
- float fh = ranges[i].font_size;
- float scale = fh > 0 ? stbtt_ScaleForPixelHeight(info, fh) : stbtt_ScaleForMappingEmToPixels(info, -fh);
- ranges[i].h_oversample = (unsigned char)spc->h_oversample;
- ranges[i].v_oversample = (unsigned char)spc->v_oversample;
- for (j = 0; j < ranges[i].num_chars; ++j) {
- int x0, y0, x1, y1;
- int codepoint = ranges[i].array_of_unicode_codepoints == NULL ? ranges[i].first_unicode_codepoint_in_range + j : ranges[i].array_of_unicode_codepoints[j];
- int glyph = stbtt_FindGlyphIndex(info, codepoint);
- stbtt_GetGlyphBitmapBoxSubpixel(info, glyph,
- scale * spc->h_oversample,
- scale * spc->v_oversample,
- 0, 0,
- &x0, &y0, &x1, &y1);
- rects[k].w = (stbrp_coord)(x1 - x0 + spc->padding + spc->h_oversample - 1);
- rects[k].h = (stbrp_coord)(y1 - y0 + spc->padding + spc->v_oversample - 1);
- ++k;
- }
- }
-
- return k;
+ int i,j,k;
+ int missing_glyph_added = 0;
+
+ k=0;
+ for (i=0; i < num_ranges; ++i) {
+ float fh = ranges[i].font_size;
+ float scale = fh > 0 ? stbtt_ScaleForPixelHeight(info, fh) : stbtt_ScaleForMappingEmToPixels(info, -fh);
+ ranges[i].h_oversample = (unsigned char) spc->h_oversample;
+ ranges[i].v_oversample = (unsigned char) spc->v_oversample;
+ for (j=0; j < ranges[i].num_chars; ++j) {
+ int x0,y0,x1,y1;
+ int codepoint = ranges[i].array_of_unicode_codepoints == NULL ? ranges[i].first_unicode_codepoint_in_range + j : ranges[i].array_of_unicode_codepoints[j];
+ int glyph = stbtt_FindGlyphIndex(info, codepoint);
+ if (glyph == 0 && (spc->skip_missing || missing_glyph_added)) {
+ rects[k].w = rects[k].h = 0;
+ } else {
+ stbtt_GetGlyphBitmapBoxSubpixel(info,glyph,
+ scale * spc->h_oversample,
+ scale * spc->v_oversample,
+ 0,0,
+ &x0,&y0,&x1,&y1);
+ rects[k].w = (stbrp_coord) (x1-x0 + spc->padding + spc->h_oversample-1);
+ rects[k].h = (stbrp_coord) (y1-y0 + spc->padding + spc->v_oversample-1);
+ if (glyph == 0)
+ missing_glyph_added = 1;
+ }
+ ++k;
+ }
+ }
+
+ return k;
}
STBTT_DEF void stbtt_MakeGlyphBitmapSubpixelPrefilter(const stbtt_fontinfo *info, unsigned char *output, int out_w, int out_h, int out_stride, float scale_x, float scale_y, float shift_x, float shift_y, int prefilter_x, int prefilter_y, float *sub_x, float *sub_y, int glyph)
{
- stbtt_MakeGlyphBitmapSubpixel(info,
- output,
- out_w - (prefilter_x - 1),
- out_h - (prefilter_y - 1),
- out_stride,
- scale_x,
- scale_y,
- shift_x,
- shift_y,
- glyph);
+ stbtt_MakeGlyphBitmapSubpixel(info,
+ output,
+ out_w - (prefilter_x - 1),
+ out_h - (prefilter_y - 1),
+ out_stride,
+ scale_x,
+ scale_y,
+ shift_x,
+ shift_y,
+ glyph);
- if (prefilter_x > 1)
- stbtt__h_prefilter(output, out_w, out_h, out_stride, prefilter_x);
+ if (prefilter_x > 1)
+ stbtt__h_prefilter(output, out_w, out_h, out_stride, prefilter_x);
- if (prefilter_y > 1)
- stbtt__v_prefilter(output, out_w, out_h, out_stride, prefilter_y);
+ if (prefilter_y > 1)
+ stbtt__v_prefilter(output, out_w, out_h, out_stride, prefilter_y);
- *sub_x = stbtt__oversample_shift(prefilter_x);
- *sub_y = stbtt__oversample_shift(prefilter_y);
+ *sub_x = stbtt__oversample_shift(prefilter_x);
+ *sub_y = stbtt__oversample_shift(prefilter_y);
}
// rects array must be big enough to accommodate all characters in the given ranges
STBTT_DEF int stbtt_PackFontRangesRenderIntoRects(stbtt_pack_context *spc, const stbtt_fontinfo *info, stbtt_pack_range *ranges, int num_ranges, stbrp_rect *rects)
{
- int i, j, k, return_value = 1;
-
- // save current values
- int old_h_over = spc->h_oversample;
- int old_v_over = spc->v_oversample;
-
- k = 0;
- for (i = 0; i < num_ranges; ++i) {
- float fh = ranges[i].font_size;
- float scale = fh > 0 ? stbtt_ScaleForPixelHeight(info, fh) : stbtt_ScaleForMappingEmToPixels(info, -fh);
- float recip_h, recip_v, sub_x, sub_y;
- spc->h_oversample = ranges[i].h_oversample;
- spc->v_oversample = ranges[i].v_oversample;
- recip_h = 1.0f / spc->h_oversample;
- recip_v = 1.0f / spc->v_oversample;
- sub_x = stbtt__oversample_shift(spc->h_oversample);
- sub_y = stbtt__oversample_shift(spc->v_oversample);
- for (j = 0; j < ranges[i].num_chars; ++j) {
- stbrp_rect *r = &rects[k];
- if (r->was_packed) {
- stbtt_packedchar *bc = &ranges[i].chardata_for_range[j];
- int advance, lsb, x0, y0, x1, y1;
- int codepoint = ranges[i].array_of_unicode_codepoints == NULL ? ranges[i].first_unicode_codepoint_in_range + j : ranges[i].array_of_unicode_codepoints[j];
- int glyph = stbtt_FindGlyphIndex(info, codepoint);
- stbrp_coord pad = (stbrp_coord)spc->padding;
-
- // pad on left and top
- r->x += pad;
- r->y += pad;
- r->w -= pad;
- r->h -= pad;
- stbtt_GetGlyphHMetrics(info, glyph, &advance, &lsb);
- stbtt_GetGlyphBitmapBox(info, glyph,
- scale * spc->h_oversample,
- scale * spc->v_oversample,
- &x0, &y0, &x1, &y1);
- stbtt_MakeGlyphBitmapSubpixel(info,
- spc->pixels + r->x + r->y*spc->stride_in_bytes,
- r->w - spc->h_oversample + 1,
- r->h - spc->v_oversample + 1,
- spc->stride_in_bytes,
- scale * spc->h_oversample,
- scale * spc->v_oversample,
- 0, 0,
- glyph);
-
- if (spc->h_oversample > 1)
- stbtt__h_prefilter(spc->pixels + r->x + r->y*spc->stride_in_bytes,
- r->w, r->h, spc->stride_in_bytes,
- spc->h_oversample);
-
- if (spc->v_oversample > 1)
- stbtt__v_prefilter(spc->pixels + r->x + r->y*spc->stride_in_bytes,
- r->w, r->h, spc->stride_in_bytes,
- spc->v_oversample);
-
- bc->x0 = (stbtt_int16)r->x;
- bc->y0 = (stbtt_int16)r->y;
- bc->x1 = (stbtt_int16)(r->x + r->w);
- bc->y1 = (stbtt_int16)(r->y + r->h);
- bc->xadvance = scale * advance;
- bc->xoff = (float)x0 * recip_h + sub_x;
- bc->yoff = (float)y0 * recip_v + sub_y;
- bc->xoff2 = (x0 + r->w) * recip_h + sub_x;
- bc->yoff2 = (y0 + r->h) * recip_v + sub_y;
- }
- else {
- return_value = 0; // if any fail, report failure
- }
-
- ++k;
- }
- }
-
- // restore original values
- spc->h_oversample = old_h_over;
- spc->v_oversample = old_v_over;
-
- return return_value;
+ int i,j,k, missing_glyph = -1, return_value = 1;
+
+ // save current values
+ int old_h_over = spc->h_oversample;
+ int old_v_over = spc->v_oversample;
+
+ k = 0;
+ for (i=0; i < num_ranges; ++i) {
+ float fh = ranges[i].font_size;
+ float scale = fh > 0 ? stbtt_ScaleForPixelHeight(info, fh) : stbtt_ScaleForMappingEmToPixels(info, -fh);
+ float recip_h,recip_v,sub_x,sub_y;
+ spc->h_oversample = ranges[i].h_oversample;
+ spc->v_oversample = ranges[i].v_oversample;
+ recip_h = 1.0f / spc->h_oversample;
+ recip_v = 1.0f / spc->v_oversample;
+ sub_x = stbtt__oversample_shift(spc->h_oversample);
+ sub_y = stbtt__oversample_shift(spc->v_oversample);
+ for (j=0; j < ranges[i].num_chars; ++j) {
+ stbrp_rect *r = &rects[k];
+ if (r->was_packed && r->w != 0 && r->h != 0) {
+ stbtt_packedchar *bc = &ranges[i].chardata_for_range[j];
+ int advance, lsb, x0,y0,x1,y1;
+ int codepoint = ranges[i].array_of_unicode_codepoints == NULL ? ranges[i].first_unicode_codepoint_in_range + j : ranges[i].array_of_unicode_codepoints[j];
+ int glyph = stbtt_FindGlyphIndex(info, codepoint);
+ stbrp_coord pad = (stbrp_coord) spc->padding;
+
+ // pad on left and top
+ r->x += pad;
+ r->y += pad;
+ r->w -= pad;
+ r->h -= pad;
+ stbtt_GetGlyphHMetrics(info, glyph, &advance, &lsb);
+ stbtt_GetGlyphBitmapBox(info, glyph,
+ scale * spc->h_oversample,
+ scale * spc->v_oversample,
+ &x0,&y0,&x1,&y1);
+ stbtt_MakeGlyphBitmapSubpixel(info,
+ spc->pixels + r->x + r->y*spc->stride_in_bytes,
+ r->w - spc->h_oversample+1,
+ r->h - spc->v_oversample+1,
+ spc->stride_in_bytes,
+ scale * spc->h_oversample,
+ scale * spc->v_oversample,
+ 0,0,
+ glyph);
+
+ if (spc->h_oversample > 1)
+ stbtt__h_prefilter(spc->pixels + r->x + r->y*spc->stride_in_bytes,
+ r->w, r->h, spc->stride_in_bytes,
+ spc->h_oversample);
+
+ if (spc->v_oversample > 1)
+ stbtt__v_prefilter(spc->pixels + r->x + r->y*spc->stride_in_bytes,
+ r->w, r->h, spc->stride_in_bytes,
+ spc->v_oversample);
+
+ bc->x0 = (stbtt_int16) r->x;
+ bc->y0 = (stbtt_int16) r->y;
+ bc->x1 = (stbtt_int16) (r->x + r->w);
+ bc->y1 = (stbtt_int16) (r->y + r->h);
+ bc->xadvance = scale * advance;
+ bc->xoff = (float) x0 * recip_h + sub_x;
+ bc->yoff = (float) y0 * recip_v + sub_y;
+ bc->xoff2 = (x0 + r->w) * recip_h + sub_x;
+ bc->yoff2 = (y0 + r->h) * recip_v + sub_y;
+
+ if (glyph == 0)
+ missing_glyph = j;
+ } else if (spc->skip_missing) {
+ return_value = 0;
+ } else if (r->was_packed && r->w == 0 && r->h == 0 && missing_glyph >= 0) {
+ ranges[i].chardata_for_range[j] = ranges[i].chardata_for_range[missing_glyph];
+ } else {
+ return_value = 0; // if any fail, report failure
+ }
+
+ ++k;
+ }
+ }
+
+ // restore original values
+ spc->h_oversample = old_h_over;
+ spc->v_oversample = old_v_over;
+
+ return return_value;
}
STBTT_DEF void stbtt_PackFontRangesPackRects(stbtt_pack_context *spc, stbrp_rect *rects, int num_rects)
{
- stbrp_pack_rects((stbrp_context *)spc->pack_info, rects, num_rects);
+ stbrp_pack_rects((stbrp_context *) spc->pack_info, rects, num_rects);
}
STBTT_DEF int stbtt_PackFontRanges(stbtt_pack_context *spc, const unsigned char *fontdata, int font_index, stbtt_pack_range *ranges, int num_ranges)
{
- stbtt_fontinfo info;
- int i, j, n, return_value = 1;
- //stbrp_context *context = (stbrp_context *) spc->pack_info;
- stbrp_rect *rects;
+ stbtt_fontinfo info;
+ int i,j,n, return_value = 1;
+ //stbrp_context *context = (stbrp_context *) spc->pack_info;
+ stbrp_rect *rects;
- // flag all characters as NOT packed
- for (i = 0; i < num_ranges; ++i)
- for (j = 0; j < ranges[i].num_chars; ++j)
- ranges[i].chardata_for_range[j].x0 =
- ranges[i].chardata_for_range[j].y0 =
- ranges[i].chardata_for_range[j].x1 =
- ranges[i].chardata_for_range[j].y1 = 0;
+ // flag all characters as NOT packed
+ for (i=0; i < num_ranges; ++i)
+ for (j=0; j < ranges[i].num_chars; ++j)
+ ranges[i].chardata_for_range[j].x0 =
+ ranges[i].chardata_for_range[j].y0 =
+ ranges[i].chardata_for_range[j].x1 =
+ ranges[i].chardata_for_range[j].y1 = 0;
- n = 0;
- for (i = 0; i < num_ranges; ++i)
- n += ranges[i].num_chars;
+ n = 0;
+ for (i=0; i < num_ranges; ++i)
+ n += ranges[i].num_chars;
- rects = (stbrp_rect *)STBTT_malloc(sizeof(*rects) * n, spc->user_allocator_context);
- if (rects == NULL)
- return 0;
+ rects = (stbrp_rect *) STBTT_malloc(sizeof(*rects) * n, spc->user_allocator_context);
+ if (rects == NULL)
+ return 0;
- info.userdata = spc->user_allocator_context;
- stbtt_InitFont(&info, fontdata, stbtt_GetFontOffsetForIndex(fontdata, font_index));
+ info.userdata = spc->user_allocator_context;
+ stbtt_InitFont(&info, fontdata, stbtt_GetFontOffsetForIndex(fontdata,font_index));
- n = stbtt_PackFontRangesGatherRects(spc, &info, ranges, num_ranges, rects);
+ n = stbtt_PackFontRangesGatherRects(spc, &info, ranges, num_ranges, rects);
- stbtt_PackFontRangesPackRects(spc, rects, n);
+ stbtt_PackFontRangesPackRects(spc, rects, n);
- return_value = stbtt_PackFontRangesRenderIntoRects(spc, &info, ranges, num_ranges, rects);
+ return_value = stbtt_PackFontRangesRenderIntoRects(spc, &info, ranges, num_ranges, rects);
- STBTT_free(rects, spc->user_allocator_context);
- return return_value;
+ STBTT_free(rects, spc->user_allocator_context);
+ return return_value;
}
STBTT_DEF int stbtt_PackFontRange(stbtt_pack_context *spc, const unsigned char *fontdata, int font_index, float font_size,
- int first_unicode_codepoint_in_range, int num_chars_in_range, stbtt_packedchar *chardata_for_range)
+ int first_unicode_codepoint_in_range, int num_chars_in_range, stbtt_packedchar *chardata_for_range)
+{
+ stbtt_pack_range range;
+ range.first_unicode_codepoint_in_range = first_unicode_codepoint_in_range;
+ range.array_of_unicode_codepoints = NULL;
+ range.num_chars = num_chars_in_range;
+ range.chardata_for_range = chardata_for_range;
+ range.font_size = font_size;
+ return stbtt_PackFontRanges(spc, fontdata, font_index, &range, 1);
+}
+
+STBTT_DEF void stbtt_GetScaledFontVMetrics(const unsigned char *fontdata, int index, float size, float *ascent, float *descent, float *lineGap)
{
- stbtt_pack_range range;
- range.first_unicode_codepoint_in_range = first_unicode_codepoint_in_range;
- range.array_of_unicode_codepoints = NULL;
- range.num_chars = num_chars_in_range;
- range.chardata_for_range = chardata_for_range;
- range.font_size = font_size;
- return stbtt_PackFontRanges(spc, fontdata, font_index, &range, 1);
+ int i_ascent, i_descent, i_lineGap;
+ float scale;
+ stbtt_fontinfo info;
+ stbtt_InitFont(&info, fontdata, stbtt_GetFontOffsetForIndex(fontdata, index));
+ scale = size > 0 ? stbtt_ScaleForPixelHeight(&info, size) : stbtt_ScaleForMappingEmToPixels(&info, -size);
+ stbtt_GetFontVMetrics(&info, &i_ascent, &i_descent, &i_lineGap);
+ *ascent = (float) i_ascent * scale;
+ *descent = (float) i_descent * scale;
+ *lineGap = (float) i_lineGap * scale;
}
STBTT_DEF void stbtt_GetPackedQuad(const stbtt_packedchar *chardata, int pw, int ph, int char_index, float *xpos, float *ypos, stbtt_aligned_quad *q, int align_to_integer)
{
- float ipw = 1.0f / pw, iph = 1.0f / ph;
- const stbtt_packedchar *b = chardata + char_index;
+ float ipw = 1.0f / pw, iph = 1.0f / ph;
+ const stbtt_packedchar *b = chardata + char_index;
- if (align_to_integer) {
- float x = (float)STBTT_ifloor((*xpos + b->xoff) + 0.5f);
- float y = (float)STBTT_ifloor((*ypos + b->yoff) + 0.5f);
- q->x0 = x;
- q->y0 = y;
- q->x1 = x + b->xoff2 - b->xoff;
- q->y1 = y + b->yoff2 - b->yoff;
- }
- else {
- q->x0 = *xpos + b->xoff;
- q->y0 = *ypos + b->yoff;
- q->x1 = *xpos + b->xoff2;
- q->y1 = *ypos + b->yoff2;
- }
+ if (align_to_integer) {
+ float x = (float) STBTT_ifloor((*xpos + b->xoff) + 0.5f);
+ float y = (float) STBTT_ifloor((*ypos + b->yoff) + 0.5f);
+ q->x0 = x;
+ q->y0 = y;
+ q->x1 = x + b->xoff2 - b->xoff;
+ q->y1 = y + b->yoff2 - b->yoff;
+ } else {
+ q->x0 = *xpos + b->xoff;
+ q->y0 = *ypos + b->yoff;
+ q->x1 = *xpos + b->xoff2;
+ q->y1 = *ypos + b->yoff2;
+ }
- q->s0 = b->x0 * ipw;
- q->t0 = b->y0 * iph;
- q->s1 = b->x1 * ipw;
- q->t1 = b->y1 * iph;
+ q->s0 = b->x0 * ipw;
+ q->t0 = b->y0 * iph;
+ q->s1 = b->x1 * ipw;
+ q->t1 = b->y1 * iph;
- *xpos += b->xadvance;
+ *xpos += b->xadvance;
}
//////////////////////////////////////////////////////////////////////////////
@@ -4227,385 +4397,374 @@ STBTT_DEF void stbtt_GetPackedQuad(const stbtt_packedchar *chardata, int pw, int
static int stbtt__ray_intersect_bezier(float orig[2], float ray[2], float q0[2], float q1[2], float q2[2], float hits[2][2])
{
- float q0perp = q0[1] * ray[0] - q0[0] * ray[1];
- float q1perp = q1[1] * ray[0] - q1[0] * ray[1];
- float q2perp = q2[1] * ray[0] - q2[0] * ray[1];
- float roperp = orig[1] * ray[0] - orig[0] * ray[1];
-
- float a = q0perp - 2 * q1perp + q2perp;
- float b = q1perp - q0perp;
- float c = q0perp - roperp;
-
- float s0 = 0., s1 = 0.;
- int num_s = 0;
-
- if (a != 0.0) {
- float discr = b * b - a * c;
- if (discr > 0.0) {
- float rcpna = -1 / a;
- float d = (float)STBTT_sqrt(discr);
- s0 = (b + d) * rcpna;
- s1 = (b - d) * rcpna;
- if (s0 >= 0.0 && s0 <= 1.0)
- num_s = 1;
- if (d > 0.0 && s1 >= 0.0 && s1 <= 1.0) {
- if (num_s == 0) s0 = s1;
- ++num_s;
- }
- }
- }
- else {
- // 2*b*s + c = 0
- // s = -c / (2*b)
- s0 = c / (-2 * b);
- if (s0 >= 0.0 && s0 <= 1.0)
+ float q0perp = q0[1]*ray[0] - q0[0]*ray[1];
+ float q1perp = q1[1]*ray[0] - q1[0]*ray[1];
+ float q2perp = q2[1]*ray[0] - q2[0]*ray[1];
+ float roperp = orig[1]*ray[0] - orig[0]*ray[1];
+
+ float a = q0perp - 2*q1perp + q2perp;
+ float b = q1perp - q0perp;
+ float c = q0perp - roperp;
+
+ float s0 = 0., s1 = 0.;
+ int num_s = 0;
+
+ if (a != 0.0) {
+ float discr = b*b - a*c;
+ if (discr > 0.0) {
+ float rcpna = -1 / a;
+ float d = (float) STBTT_sqrt(discr);
+ s0 = (b+d) * rcpna;
+ s1 = (b-d) * rcpna;
+ if (s0 >= 0.0 && s0 <= 1.0)
num_s = 1;
- }
-
- if (num_s == 0)
- return 0;
- else {
- float rcp_len2 = 1 / (ray[0] * ray[0] + ray[1] * ray[1]);
- float rayn_x = ray[0] * rcp_len2, rayn_y = ray[1] * rcp_len2;
-
- float q0d = q0[0] * rayn_x + q0[1] * rayn_y;
- float q1d = q1[0] * rayn_x + q1[1] * rayn_y;
- float q2d = q2[0] * rayn_x + q2[1] * rayn_y;
- float rod = orig[0] * rayn_x + orig[1] * rayn_y;
-
- float q10d = q1d - q0d;
- float q20d = q2d - q0d;
- float q0rd = q0d - rod;
-
- hits[0][0] = q0rd + s0 * (2.0f - 2.0f*s0)*q10d + s0 * s0*q20d;
- hits[0][1] = a * s0 + b;
-
- if (num_s > 1) {
- hits[1][0] = q0rd + s1 * (2.0f - 2.0f*s1)*q10d + s1 * s1*q20d;
- hits[1][1] = a * s1 + b;
- return 2;
- }
- else {
- return 1;
- }
- }
+ if (d > 0.0 && s1 >= 0.0 && s1 <= 1.0) {
+ if (num_s == 0) s0 = s1;
+ ++num_s;
+ }
+ }
+ } else {
+ // 2*b*s + c = 0
+ // s = -c / (2*b)
+ s0 = c / (-2 * b);
+ if (s0 >= 0.0 && s0 <= 1.0)
+ num_s = 1;
+ }
+
+ if (num_s == 0)
+ return 0;
+ else {
+ float rcp_len2 = 1 / (ray[0]*ray[0] + ray[1]*ray[1]);
+ float rayn_x = ray[0] * rcp_len2, rayn_y = ray[1] * rcp_len2;
+
+ float q0d = q0[0]*rayn_x + q0[1]*rayn_y;
+ float q1d = q1[0]*rayn_x + q1[1]*rayn_y;
+ float q2d = q2[0]*rayn_x + q2[1]*rayn_y;
+ float rod = orig[0]*rayn_x + orig[1]*rayn_y;
+
+ float q10d = q1d - q0d;
+ float q20d = q2d - q0d;
+ float q0rd = q0d - rod;
+
+ hits[0][0] = q0rd + s0*(2.0f - 2.0f*s0)*q10d + s0*s0*q20d;
+ hits[0][1] = a*s0+b;
+
+ if (num_s > 1) {
+ hits[1][0] = q0rd + s1*(2.0f - 2.0f*s1)*q10d + s1*s1*q20d;
+ hits[1][1] = a*s1+b;
+ return 2;
+ } else {
+ return 1;
+ }
+ }
}
static int equal(float *a, float *b)
{
- return (a[0] == b[0] && a[1] == b[1]);
+ return (a[0] == b[0] && a[1] == b[1]);
}
static int stbtt__compute_crossings_x(float x, float y, int nverts, stbtt_vertex *verts)
{
- int i;
- float orig[2], ray[2] = { 1, 0 };
- float y_frac;
- int winding = 0;
-
- orig[0] = x;
- orig[1] = y;
-
- // make sure y never passes through a vertex of the shape
- y_frac = (float)STBTT_fmod(y, 1.0f);
- if (y_frac < 0.01f)
- y += 0.01f;
- else if (y_frac > 0.99f)
- y -= 0.01f;
- orig[1] = y;
-
- // test a ray from (-infinity,y) to (x,y)
- for (i = 0; i < nverts; ++i) {
- if (verts[i].type == STBTT_vline) {
- int x0 = (int)verts[i - 1].x, y0 = (int)verts[i - 1].y;
- int x1 = (int)verts[i].x, y1 = (int)verts[i].y;
- if (y > STBTT_min(y0, y1) && y < STBTT_max(y0, y1) && x > STBTT_min(x0, x1)) {
- float x_inter = (y - y0) / (y1 - y0) * (x1 - x0) + x0;
- if (x_inter < x)
- winding += (y0 < y1) ? 1 : -1;
+ int i;
+ float orig[2], ray[2] = { 1, 0 };
+ float y_frac;
+ int winding = 0;
+
+ // make sure y never passes through a vertex of the shape
+ y_frac = (float) STBTT_fmod(y, 1.0f);
+ if (y_frac < 0.01f)
+ y += 0.01f;
+ else if (y_frac > 0.99f)
+ y -= 0.01f;
+
+ orig[0] = x;
+ orig[1] = y;
+
+ // test a ray from (-infinity,y) to (x,y)
+ for (i=0; i < nverts; ++i) {
+ if (verts[i].type == STBTT_vline) {
+ int x0 = (int) verts[i-1].x, y0 = (int) verts[i-1].y;
+ int x1 = (int) verts[i ].x, y1 = (int) verts[i ].y;
+ if (y > STBTT_min(y0,y1) && y < STBTT_max(y0,y1) && x > STBTT_min(x0,x1)) {
+ float x_inter = (y - y0) / (y1 - y0) * (x1-x0) + x0;
+ if (x_inter < x)
+ winding += (y0 < y1) ? 1 : -1;
+ }
+ }
+ if (verts[i].type == STBTT_vcurve) {
+ int x0 = (int) verts[i-1].x , y0 = (int) verts[i-1].y ;
+ int x1 = (int) verts[i ].cx, y1 = (int) verts[i ].cy;
+ int x2 = (int) verts[i ].x , y2 = (int) verts[i ].y ;
+ int ax = STBTT_min(x0,STBTT_min(x1,x2)), ay = STBTT_min(y0,STBTT_min(y1,y2));
+ int by = STBTT_max(y0,STBTT_max(y1,y2));
+ if (y > ay && y < by && x > ax) {
+ float q0[2],q1[2],q2[2];
+ float hits[2][2];
+ q0[0] = (float)x0;
+ q0[1] = (float)y0;
+ q1[0] = (float)x1;
+ q1[1] = (float)y1;
+ q2[0] = (float)x2;
+ q2[1] = (float)y2;
+ if (equal(q0,q1) || equal(q1,q2)) {
+ x0 = (int)verts[i-1].x;
+ y0 = (int)verts[i-1].y;
+ x1 = (int)verts[i ].x;
+ y1 = (int)verts[i ].y;
+ if (y > STBTT_min(y0,y1) && y < STBTT_max(y0,y1) && x > STBTT_min(x0,x1)) {
+ float x_inter = (y - y0) / (y1 - y0) * (x1-x0) + x0;
+ if (x_inter < x)
+ winding += (y0 < y1) ? 1 : -1;
+ }
+ } else {
+ int num_hits = stbtt__ray_intersect_bezier(orig, ray, q0, q1, q2, hits);
+ if (num_hits >= 1)
+ if (hits[0][0] < 0)
+ winding += (hits[0][1] < 0 ? -1 : 1);
+ if (num_hits >= 2)
+ if (hits[1][0] < 0)
+ winding += (hits[1][1] < 0 ? -1 : 1);
}
- }
- if (verts[i].type == STBTT_vcurve) {
- int x0 = (int)verts[i - 1].x, y0 = (int)verts[i - 1].y;
- int x1 = (int)verts[i].cx, y1 = (int)verts[i].cy;
- int x2 = (int)verts[i].x, y2 = (int)verts[i].y;
- int ax = STBTT_min(x0, STBTT_min(x1, x2)), ay = STBTT_min(y0, STBTT_min(y1, y2));
- int by = STBTT_max(y0, STBTT_max(y1, y2));
- if (y > ay && y < by && x > ax) {
- float q0[2], q1[2], q2[2];
- float hits[2][2];
- q0[0] = (float)x0;
- q0[1] = (float)y0;
- q1[0] = (float)x1;
- q1[1] = (float)y1;
- q2[0] = (float)x2;
- q2[1] = (float)y2;
- if (equal(q0, q1) || equal(q1, q2)) {
- x0 = (int)verts[i - 1].x;
- y0 = (int)verts[i - 1].y;
- x1 = (int)verts[i].x;
- y1 = (int)verts[i].y;
- if (y > STBTT_min(y0, y1) && y < STBTT_max(y0, y1) && x > STBTT_min(x0, x1)) {
- float x_inter = (y - y0) / (y1 - y0) * (x1 - x0) + x0;
- if (x_inter < x)
- winding += (y0 < y1) ? 1 : -1;
- }
- }
- else {
- int num_hits = stbtt__ray_intersect_bezier(orig, ray, q0, q1, q2, hits);
- if (num_hits >= 1)
- if (hits[0][0] < 0)
- winding += (hits[0][1] < 0 ? -1 : 1);
- if (num_hits >= 2)
- if (hits[1][0] < 0)
- winding += (hits[1][1] < 0 ? -1 : 1);
- }
- }
- }
- }
- return winding;
+ }
+ }
+ }
+ return winding;
}
-static float stbtt__cuberoot(float x)
+static float stbtt__cuberoot( float x )
{
- if (x<0)
- return -(float)STBTT_pow(-x, 1.0f / 3.0f);
- else
- return (float)STBTT_pow(x, 1.0f / 3.0f);
+ if (x<0)
+ return -(float) STBTT_pow(-x,1.0f/3.0f);
+ else
+ return (float) STBTT_pow( x,1.0f/3.0f);
}
-// x^3 + c*x^2 + b*x + a = 0
+// x^3 + a*x^2 + b*x + c = 0
static int stbtt__solve_cubic(float a, float b, float c, float* r)
{
- float s = -a / 3;
- float p = b - a * a / 3;
- float q = a * (2 * a*a - 9 * b) / 27 + c;
- float p3 = p * p*p;
- float d = q * q + 4 * p3 / 27;
- if (d >= 0) {
- float z = (float)STBTT_sqrt(d);
- float u = (-q + z) / 2;
- float v = (-q - z) / 2;
- u = stbtt__cuberoot(u);
- v = stbtt__cuberoot(v);
- r[0] = s + u + v;
- return 1;
- }
- else {
- float u = (float)STBTT_sqrt(-p / 3);
- float v = (float)STBTT_acos(-STBTT_sqrt(-27 / p3) * q / 2) / 3; // p3 must be negative, since d is negative
- float m = (float)STBTT_cos(v);
- float n = (float)STBTT_cos(v - 3.141592 / 2)*1.732050808f;
- r[0] = s + u * 2 * m;
- r[1] = s - u * (m + n);
- r[2] = s - u * (m - n);
-
- //STBTT_assert( STBTT_fabs(((r[0]+a)*r[0]+b)*r[0]+c) < 0.05f); // these asserts may not be safe at all scales, though they're in bezier t parameter units so maybe?
- //STBTT_assert( STBTT_fabs(((r[1]+a)*r[1]+b)*r[1]+c) < 0.05f);
- //STBTT_assert( STBTT_fabs(((r[2]+a)*r[2]+b)*r[2]+c) < 0.05f);
- return 3;
- }
+ float s = -a / 3;
+ float p = b - a*a / 3;
+ float q = a * (2*a*a - 9*b) / 27 + c;
+ float p3 = p*p*p;
+ float d = q*q + 4*p3 / 27;
+ if (d >= 0) {
+ float z = (float) STBTT_sqrt(d);
+ float u = (-q + z) / 2;
+ float v = (-q - z) / 2;
+ u = stbtt__cuberoot(u);
+ v = stbtt__cuberoot(v);
+ r[0] = s + u + v;
+ return 1;
+ } else {
+ float u = (float) STBTT_sqrt(-p/3);
+ float v = (float) STBTT_acos(-STBTT_sqrt(-27/p3) * q / 2) / 3; // p3 must be negative, since d is negative
+ float m = (float) STBTT_cos(v);
+ float n = (float) STBTT_cos(v-3.141592/2)*1.732050808f;
+ r[0] = s + u * 2 * m;
+ r[1] = s - u * (m + n);
+ r[2] = s - u * (m - n);
+
+ //STBTT_assert( STBTT_fabs(((r[0]+a)*r[0]+b)*r[0]+c) < 0.05f); // these asserts may not be safe at all scales, though they're in bezier t parameter units so maybe?
+ //STBTT_assert( STBTT_fabs(((r[1]+a)*r[1]+b)*r[1]+c) < 0.05f);
+ //STBTT_assert( STBTT_fabs(((r[2]+a)*r[2]+b)*r[2]+c) < 0.05f);
+ return 3;
+ }
}
STBTT_DEF unsigned char * stbtt_GetGlyphSDF(const stbtt_fontinfo *info, float scale, int glyph, int padding, unsigned char onedge_value, float pixel_dist_scale, int *width, int *height, int *xoff, int *yoff)
{
- float scale_x = scale, scale_y = scale;
- int ix0, iy0, ix1, iy1;
- int w, h;
- unsigned char *data;
-
- // if one scale is 0, use same scale for both
- if (scale_x == 0) scale_x = scale_y;
- if (scale_y == 0) {
- if (scale_x == 0) return NULL; // if both scales are 0, return NULL
- scale_y = scale_x;
- }
-
- stbtt_GetGlyphBitmapBoxSubpixel(info, glyph, scale, scale, 0.0f, 0.0f, &ix0, &iy0, &ix1, &iy1);
-
- // if empty, return NULL
- if (ix0 == ix1 || iy0 == iy1)
- return NULL;
-
- ix0 -= padding;
- iy0 -= padding;
- ix1 += padding;
- iy1 += padding;
-
- w = (ix1 - ix0);
- h = (iy1 - iy0);
-
- if (width) *width = w;
- if (height) *height = h;
- if (xoff) *xoff = ix0;
- if (yoff) *yoff = iy0;
-
- // invert for y-downwards bitmaps
- scale_y = -scale_y;
-
- {
- int x, y, i, j;
- float *precompute;
- stbtt_vertex *verts;
- int num_verts = stbtt_GetGlyphShape(info, glyph, &verts);
- data = (unsigned char *)STBTT_malloc(w * h, info->userdata);
- precompute = (float *)STBTT_malloc(num_verts * sizeof(float), info->userdata);
-
- for (i = 0, j = num_verts - 1; i < num_verts; j = i++) {
- if (verts[i].type == STBTT_vline) {
- float x0 = verts[i].x*scale_x, y0 = verts[i].y*scale_y;
- float x1 = verts[j].x*scale_x, y1 = verts[j].y*scale_y;
- float dist = (float)STBTT_sqrt((x1 - x0)*(x1 - x0) + (y1 - y0)*(y1 - y0));
- precompute[i] = (dist == 0) ? 0.0f : 1.0f / dist;
- }
- else if (verts[i].type == STBTT_vcurve) {
- float x2 = verts[j].x *scale_x, y2 = verts[j].y *scale_y;
- float x1 = verts[i].cx*scale_x, y1 = verts[i].cy*scale_y;
- float x0 = verts[i].x *scale_x, y0 = verts[i].y *scale_y;
- float bx = x0 - 2 * x1 + x2, by = y0 - 2 * y1 + y2;
- float len2 = bx * bx + by * by;
- if (len2 != 0.0f)
- precompute[i] = 1.0f / (bx*bx + by * by);
- else
- precompute[i] = 0.0f;
- }
+ float scale_x = scale, scale_y = scale;
+ int ix0,iy0,ix1,iy1;
+ int w,h;
+ unsigned char *data;
+
+ if (scale == 0) return NULL;
+
+ stbtt_GetGlyphBitmapBoxSubpixel(info, glyph, scale, scale, 0.0f,0.0f, &ix0,&iy0,&ix1,&iy1);
+
+ // if empty, return NULL
+ if (ix0 == ix1 || iy0 == iy1)
+ return NULL;
+
+ ix0 -= padding;
+ iy0 -= padding;
+ ix1 += padding;
+ iy1 += padding;
+
+ w = (ix1 - ix0);
+ h = (iy1 - iy0);
+
+ if (width ) *width = w;
+ if (height) *height = h;
+ if (xoff ) *xoff = ix0;
+ if (yoff ) *yoff = iy0;
+
+ // invert for y-downwards bitmaps
+ scale_y = -scale_y;
+
+ {
+ int x,y,i,j;
+ float *precompute;
+ stbtt_vertex *verts;
+ int num_verts = stbtt_GetGlyphShape(info, glyph, &verts);
+ data = (unsigned char *) STBTT_malloc(w * h, info->userdata);
+ precompute = (float *) STBTT_malloc(num_verts * sizeof(float), info->userdata);
+
+ for (i=0,j=num_verts-1; i < num_verts; j=i++) {
+ if (verts[i].type == STBTT_vline) {
+ float x0 = verts[i].x*scale_x, y0 = verts[i].y*scale_y;
+ float x1 = verts[j].x*scale_x, y1 = verts[j].y*scale_y;
+ float dist = (float) STBTT_sqrt((x1-x0)*(x1-x0) + (y1-y0)*(y1-y0));
+ precompute[i] = (dist == 0) ? 0.0f : 1.0f / dist;
+ } else if (verts[i].type == STBTT_vcurve) {
+ float x2 = verts[j].x *scale_x, y2 = verts[j].y *scale_y;
+ float x1 = verts[i].cx*scale_x, y1 = verts[i].cy*scale_y;
+ float x0 = verts[i].x *scale_x, y0 = verts[i].y *scale_y;
+ float bx = x0 - 2*x1 + x2, by = y0 - 2*y1 + y2;
+ float len2 = bx*bx + by*by;
+ if (len2 != 0.0f)
+ precompute[i] = 1.0f / (bx*bx + by*by);
else
- precompute[i] = 0.0f;
- }
-
- for (y = iy0; y < iy1; ++y) {
- for (x = ix0; x < ix1; ++x) {
- float val;
- float min_dist = 999999.0f;
- float sx = (float)x + 0.5f;
- float sy = (float)y + 0.5f;
- float x_gspace = (sx / scale_x);
- float y_gspace = (sy / scale_y);
-
- int winding = stbtt__compute_crossings_x(x_gspace, y_gspace, num_verts, verts); // @OPTIMIZE: this could just be a rasterization, but needs to be line vs. non-tesselated curves so a new path
-
- for (i = 0; i < num_verts; ++i) {
- float x0 = verts[i].x*scale_x, y0 = verts[i].y*scale_y;
-
- // check against every point here rather than inside line/curve primitives -- @TODO: wrong if multiple 'moves' in a row produce a garbage point, and given culling, probably more efficient to do within line/curve
- float dist2 = (x0 - sx)*(x0 - sx) + (y0 - sy)*(y0 - sy);
- if (dist2 < min_dist*min_dist)
- min_dist = (float)STBTT_sqrt(dist2);
-
- if (verts[i].type == STBTT_vline) {
- float x1 = verts[i - 1].x*scale_x, y1 = verts[i - 1].y*scale_y;
-
- // coarse culling against bbox
- //if (sx > STBTT_min(x0,x1)-min_dist && sx < STBTT_max(x0,x1)+min_dist &&
- // sy > STBTT_min(y0,y1)-min_dist && sy < STBTT_max(y0,y1)+min_dist)
- float dist = (float)STBTT_fabs((x1 - x0)*(y0 - sy) - (y1 - y0)*(x0 - sx)) * precompute[i];
- STBTT_assert(i != 0);
- if (dist < min_dist) {
- // check position along line
- // x' = x0 + t*(x1-x0), y' = y0 + t*(y1-y0)
- // minimize (x'-sx)*(x'-sx)+(y'-sy)*(y'-sy)
- float dx = x1 - x0, dy = y1 - y0;
- float px = x0 - sx, py = y0 - sy;
- // minimize (px+t*dx)^2 + (py+t*dy)^2 = px*px + 2*px*dx*t + t^2*dx*dx + py*py + 2*py*dy*t + t^2*dy*dy
- // derivative: 2*px*dx + 2*py*dy + (2*dx*dx+2*dy*dy)*t, set to 0 and solve
- float t = -(px*dx + py * dy) / (dx*dx + dy * dy);
- if (t >= 0.0f && t <= 1.0f)
- min_dist = dist;
- }
- }
- else if (verts[i].type == STBTT_vcurve) {
- float x2 = verts[i - 1].x *scale_x, y2 = verts[i - 1].y *scale_y;
- float x1 = verts[i].cx*scale_x, y1 = verts[i].cy*scale_y;
- float box_x0 = STBTT_min(STBTT_min(x0, x1), x2);
- float box_y0 = STBTT_min(STBTT_min(y0, y1), y2);
- float box_x1 = STBTT_max(STBTT_max(x0, x1), x2);
- float box_y1 = STBTT_max(STBTT_max(y0, y1), y2);
- // coarse culling against bbox to avoid computing cubic unnecessarily
- if (sx > box_x0 - min_dist && sx < box_x1 + min_dist && sy > box_y0 - min_dist && sy < box_y1 + min_dist) {
- int num = 0;
- float ax = x1 - x0, ay = y1 - y0;
- float bx = x0 - 2 * x1 + x2, by = y0 - 2 * y1 + y2;
- float mx = x0 - sx, my = y0 - sy;
- float res[3], px, py, t, it;
- float a_inv = precompute[i];
- if (a_inv == 0.0) { // if a_inv is 0, it's 2nd degree so use quadratic formula
- float a = 3 * (ax*bx + ay * by);
- float b = 2 * (ax*ax + ay * ay) + (mx*bx + my * by);
- float c = mx * ax + my * ay;
- if (a == 0.0) { // if a is 0, it's linear
- if (b != 0.0) {
- res[num++] = -c / b;
- }
- }
- else {
- float discriminant = b * b - 4 * a*c;
- if (discriminant < 0)
- num = 0;
- else {
- float root = (float)STBTT_sqrt(discriminant);
- res[0] = (-b - root) / (2 * a);
- res[1] = (-b + root) / (2 * a);
- num = 2; // don't bother distinguishing 1-solution case, as code below will still work
- }
- }
- }
- else {
- float b = 3 * (ax*bx + ay * by) * a_inv; // could precompute this as it doesn't depend on sample point
- float c = (2 * (ax*ax + ay * ay) + (mx*bx + my * by)) * a_inv;
- float d = (mx*ax + my * ay) * a_inv;
- num = stbtt__solve_cubic(b, c, d, res);
- }
- if (num >= 1 && res[0] >= 0.0f && res[0] <= 1.0f) {
- t = res[0], it = 1.0f - t;
- px = it * it*x0 + 2 * t*it*x1 + t * t*x2;
- py = it * it*y0 + 2 * t*it*y1 + t * t*y2;
- dist2 = (px - sx)*(px - sx) + (py - sy)*(py - sy);
- if (dist2 < min_dist * min_dist)
- min_dist = (float)STBTT_sqrt(dist2);
- }
- if (num >= 2 && res[1] >= 0.0f && res[1] <= 1.0f) {
- t = res[1], it = 1.0f - t;
- px = it * it*x0 + 2 * t*it*x1 + t * t*x2;
- py = it * it*y0 + 2 * t*it*y1 + t * t*y2;
- dist2 = (px - sx)*(px - sx) + (py - sy)*(py - sy);
- if (dist2 < min_dist * min_dist)
- min_dist = (float)STBTT_sqrt(dist2);
- }
- if (num >= 3 && res[2] >= 0.0f && res[2] <= 1.0f) {
- t = res[2], it = 1.0f - t;
- px = it * it*x0 + 2 * t*it*x1 + t * t*x2;
- py = it * it*y0 + 2 * t*it*y1 + t * t*y2;
- dist2 = (px - sx)*(px - sx) + (py - sy)*(py - sy);
- if (dist2 < min_dist * min_dist)
- min_dist = (float)STBTT_sqrt(dist2);
- }
+ precompute[i] = 0.0f;
+ } else
+ precompute[i] = 0.0f;
+ }
+
+ for (y=iy0; y < iy1; ++y) {
+ for (x=ix0; x < ix1; ++x) {
+ float val;
+ float min_dist = 999999.0f;
+ float sx = (float) x + 0.5f;
+ float sy = (float) y + 0.5f;
+ float x_gspace = (sx / scale_x);
+ float y_gspace = (sy / scale_y);
+
+ int winding = stbtt__compute_crossings_x(x_gspace, y_gspace, num_verts, verts); // @OPTIMIZE: this could just be a rasterization, but needs to be line vs. non-tesselated curves so a new path
+
+ for (i=0; i < num_verts; ++i) {
+ float x0 = verts[i].x*scale_x, y0 = verts[i].y*scale_y;
+
+ if (verts[i].type == STBTT_vline && precompute[i] != 0.0f) {
+ float x1 = verts[i-1].x*scale_x, y1 = verts[i-1].y*scale_y;
+
+ float dist,dist2 = (x0-sx)*(x0-sx) + (y0-sy)*(y0-sy);
+ if (dist2 < min_dist*min_dist)
+ min_dist = (float) STBTT_sqrt(dist2);
+
+ // coarse culling against bbox
+ //if (sx > STBTT_min(x0,x1)-min_dist && sx < STBTT_max(x0,x1)+min_dist &&
+ // sy > STBTT_min(y0,y1)-min_dist && sy < STBTT_max(y0,y1)+min_dist)
+ dist = (float) STBTT_fabs((x1-x0)*(y0-sy) - (y1-y0)*(x0-sx)) * precompute[i];
+ STBTT_assert(i != 0);
+ if (dist < min_dist) {
+ // check position along line
+ // x' = x0 + t*(x1-x0), y' = y0 + t*(y1-y0)
+ // minimize (x'-sx)*(x'-sx)+(y'-sy)*(y'-sy)
+ float dx = x1-x0, dy = y1-y0;
+ float px = x0-sx, py = y0-sy;
+ // minimize (px+t*dx)^2 + (py+t*dy)^2 = px*px + 2*px*dx*t + t^2*dx*dx + py*py + 2*py*dy*t + t^2*dy*dy
+ // derivative: 2*px*dx + 2*py*dy + (2*dx*dx+2*dy*dy)*t, set to 0 and solve
+ float t = -(px*dx + py*dy) / (dx*dx + dy*dy);
+ if (t >= 0.0f && t <= 1.0f)
+ min_dist = dist;
+ }
+ } else if (verts[i].type == STBTT_vcurve) {
+ float x2 = verts[i-1].x *scale_x, y2 = verts[i-1].y *scale_y;
+ float x1 = verts[i ].cx*scale_x, y1 = verts[i ].cy*scale_y;
+ float box_x0 = STBTT_min(STBTT_min(x0,x1),x2);
+ float box_y0 = STBTT_min(STBTT_min(y0,y1),y2);
+ float box_x1 = STBTT_max(STBTT_max(x0,x1),x2);
+ float box_y1 = STBTT_max(STBTT_max(y0,y1),y2);
+ // coarse culling against bbox to avoid computing cubic unnecessarily
+ if (sx > box_x0-min_dist && sx < box_x1+min_dist && sy > box_y0-min_dist && sy < box_y1+min_dist) {
+ int num=0;
+ float ax = x1-x0, ay = y1-y0;
+ float bx = x0 - 2*x1 + x2, by = y0 - 2*y1 + y2;
+ float mx = x0 - sx, my = y0 - sy;
+ float res[3] = {0.f,0.f,0.f};
+ float px,py,t,it,dist2;
+ float a_inv = precompute[i];
+ if (a_inv == 0.0) { // if a_inv is 0, it's 2nd degree so use quadratic formula
+ float a = 3*(ax*bx + ay*by);
+ float b = 2*(ax*ax + ay*ay) + (mx*bx+my*by);
+ float c = mx*ax+my*ay;
+ if (a == 0.0) { // if a is 0, it's linear
+ if (b != 0.0) {
+ res[num++] = -c/b;
+ }
+ } else {
+ float discriminant = b*b - 4*a*c;
+ if (discriminant < 0)
+ num = 0;
+ else {
+ float root = (float) STBTT_sqrt(discriminant);
+ res[0] = (-b - root)/(2*a);
+ res[1] = (-b + root)/(2*a);
+ num = 2; // don't bother distinguishing 1-solution case, as code below will still work
+ }
}
- }
- }
- if (winding == 0)
- min_dist = -min_dist; // if outside the shape, value is negative
- val = onedge_value + pixel_dist_scale * min_dist;
- if (val < 0)
- val = 0;
- else if (val > 255)
- val = 255;
- data[(y - iy0)*w + (x - ix0)] = (unsigned char)val;
+ } else {
+ float b = 3*(ax*bx + ay*by) * a_inv; // could precompute this as it doesn't depend on sample point
+ float c = (2*(ax*ax + ay*ay) + (mx*bx+my*by)) * a_inv;
+ float d = (mx*ax+my*ay) * a_inv;
+ num = stbtt__solve_cubic(b, c, d, res);
+ }
+ dist2 = (x0-sx)*(x0-sx) + (y0-sy)*(y0-sy);
+ if (dist2 < min_dist*min_dist)
+ min_dist = (float) STBTT_sqrt(dist2);
+
+ if (num >= 1 && res[0] >= 0.0f && res[0] <= 1.0f) {
+ t = res[0], it = 1.0f - t;
+ px = it*it*x0 + 2*t*it*x1 + t*t*x2;
+ py = it*it*y0 + 2*t*it*y1 + t*t*y2;
+ dist2 = (px-sx)*(px-sx) + (py-sy)*(py-sy);
+ if (dist2 < min_dist * min_dist)
+ min_dist = (float) STBTT_sqrt(dist2);
+ }
+ if (num >= 2 && res[1] >= 0.0f && res[1] <= 1.0f) {
+ t = res[1], it = 1.0f - t;
+ px = it*it*x0 + 2*t*it*x1 + t*t*x2;
+ py = it*it*y0 + 2*t*it*y1 + t*t*y2;
+ dist2 = (px-sx)*(px-sx) + (py-sy)*(py-sy);
+ if (dist2 < min_dist * min_dist)
+ min_dist = (float) STBTT_sqrt(dist2);
+ }
+ if (num >= 3 && res[2] >= 0.0f && res[2] <= 1.0f) {
+ t = res[2], it = 1.0f - t;
+ px = it*it*x0 + 2*t*it*x1 + t*t*x2;
+ py = it*it*y0 + 2*t*it*y1 + t*t*y2;
+ dist2 = (px-sx)*(px-sx) + (py-sy)*(py-sy);
+ if (dist2 < min_dist * min_dist)
+ min_dist = (float) STBTT_sqrt(dist2);
+ }
+ }
+ }
}
- }
- STBTT_free(precompute, info->userdata);
- STBTT_free(verts, info->userdata);
- }
- return data;
+ if (winding == 0)
+ min_dist = -min_dist; // if outside the shape, value is negative
+ val = onedge_value + pixel_dist_scale * min_dist;
+ if (val < 0)
+ val = 0;
+ else if (val > 255)
+ val = 255;
+ data[(y-iy0)*w+(x-ix0)] = (unsigned char) val;
+ }
+ }
+ STBTT_free(precompute, info->userdata);
+ STBTT_free(verts, info->userdata);
+ }
+ return data;
}
STBTT_DEF unsigned char * stbtt_GetCodepointSDF(const stbtt_fontinfo *info, float scale, int codepoint, int padding, unsigned char onedge_value, float pixel_dist_scale, int *width, int *height, int *xoff, int *yoff)
{
- return stbtt_GetGlyphSDF(info, scale, stbtt_FindGlyphIndex(info, codepoint), padding, onedge_value, pixel_dist_scale, width, height, xoff, yoff);
+ return stbtt_GetGlyphSDF(info, scale, stbtt_FindGlyphIndex(info, codepoint), padding, onedge_value, pixel_dist_scale, width, height, xoff, yoff);
}
STBTT_DEF void stbtt_FreeSDF(unsigned char *bitmap, void *userdata)
{
- STBTT_free(bitmap, userdata);
+ STBTT_free(bitmap, userdata);
}
//////////////////////////////////////////////////////////////////////////////
@@ -4616,163 +4775,156 @@ STBTT_DEF void stbtt_FreeSDF(unsigned char *bitmap, void *userdata)
// check if a utf8 string contains a prefix which is the utf16 string; if so return length of matching utf8 string
static stbtt_int32 stbtt__CompareUTF8toUTF16_bigendian_prefix(stbtt_uint8 *s1, stbtt_int32 len1, stbtt_uint8 *s2, stbtt_int32 len2)
{
- stbtt_int32 i = 0;
-
- // convert utf16 to utf8 and compare the results while converting
- while (len2) {
- stbtt_uint16 ch = s2[0] * 256 + s2[1];
- if (ch < 0x80) {
- if (i >= len1) return -1;
- if (s1[i++] != ch) return -1;
- }
- else if (ch < 0x800) {
- if (i + 1 >= len1) return -1;
- if (s1[i++] != 0xc0 + (ch >> 6)) return -1;
- if (s1[i++] != 0x80 + (ch & 0x3f)) return -1;
- }
- else if (ch >= 0xd800 && ch < 0xdc00) {
- stbtt_uint32 c;
- stbtt_uint16 ch2 = s2[2] * 256 + s2[3];
- if (i + 3 >= len1) return -1;
- c = ((ch - 0xd800) << 10) + (ch2 - 0xdc00) + 0x10000;
- if (s1[i++] != 0xf0 + (c >> 18)) return -1;
- if (s1[i++] != 0x80 + ((c >> 12) & 0x3f)) return -1;
- if (s1[i++] != 0x80 + ((c >> 6) & 0x3f)) return -1;
- if (s1[i++] != 0x80 + ((c) & 0x3f)) return -1;
- s2 += 2; // plus another 2 below
- len2 -= 2;
- }
- else if (ch >= 0xdc00 && ch < 0xe000) {
- return -1;
- }
- else {
- if (i + 2 >= len1) return -1;
- if (s1[i++] != 0xe0 + (ch >> 12)) return -1;
- if (s1[i++] != 0x80 + ((ch >> 6) & 0x3f)) return -1;
- if (s1[i++] != 0x80 + ((ch) & 0x3f)) return -1;
- }
- s2 += 2;
- len2 -= 2;
- }
- return i;
+ stbtt_int32 i=0;
+
+ // convert utf16 to utf8 and compare the results while converting
+ while (len2) {
+ stbtt_uint16 ch = s2[0]*256 + s2[1];
+ if (ch < 0x80) {
+ if (i >= len1) return -1;
+ if (s1[i++] != ch) return -1;
+ } else if (ch < 0x800) {
+ if (i+1 >= len1) return -1;
+ if (s1[i++] != 0xc0 + (ch >> 6)) return -1;
+ if (s1[i++] != 0x80 + (ch & 0x3f)) return -1;
+ } else if (ch >= 0xd800 && ch < 0xdc00) {
+ stbtt_uint32 c;
+ stbtt_uint16 ch2 = s2[2]*256 + s2[3];
+ if (i+3 >= len1) return -1;
+ c = ((ch - 0xd800) << 10) + (ch2 - 0xdc00) + 0x10000;
+ if (s1[i++] != 0xf0 + (c >> 18)) return -1;
+ if (s1[i++] != 0x80 + ((c >> 12) & 0x3f)) return -1;
+ if (s1[i++] != 0x80 + ((c >> 6) & 0x3f)) return -1;
+ if (s1[i++] != 0x80 + ((c ) & 0x3f)) return -1;
+ s2 += 2; // plus another 2 below
+ len2 -= 2;
+ } else if (ch >= 0xdc00 && ch < 0xe000) {
+ return -1;
+ } else {
+ if (i+2 >= len1) return -1;
+ if (s1[i++] != 0xe0 + (ch >> 12)) return -1;
+ if (s1[i++] != 0x80 + ((ch >> 6) & 0x3f)) return -1;
+ if (s1[i++] != 0x80 + ((ch ) & 0x3f)) return -1;
+ }
+ s2 += 2;
+ len2 -= 2;
+ }
+ return i;
}
static int stbtt_CompareUTF8toUTF16_bigendian_internal(char *s1, int len1, char *s2, int len2)
{
- return len1 == stbtt__CompareUTF8toUTF16_bigendian_prefix((stbtt_uint8*)s1, len1, (stbtt_uint8*)s2, len2);
+ return len1 == stbtt__CompareUTF8toUTF16_bigendian_prefix((stbtt_uint8*) s1, len1, (stbtt_uint8*) s2, len2);
}
// returns results in whatever encoding you request... but note that 2-byte encodings
// will be BIG-ENDIAN... use stbtt_CompareUTF8toUTF16_bigendian() to compare
STBTT_DEF const char *stbtt_GetFontNameString(const stbtt_fontinfo *font, int *length, int platformID, int encodingID, int languageID, int nameID)
{
- stbtt_int32 i, count, stringOffset;
- stbtt_uint8 *fc = font->data;
- stbtt_uint32 offset = font->fontstart;
- stbtt_uint32 nm = stbtt__find_table(fc, offset, "name");
- if (!nm) return NULL;
-
- count = ttUSHORT(fc + nm + 2);
- stringOffset = nm + ttUSHORT(fc + nm + 4);
- for (i = 0; i < count; ++i) {
- stbtt_uint32 loc = nm + 6 + 12 * i;
- if (platformID == ttUSHORT(fc + loc + 0) && encodingID == ttUSHORT(fc + loc + 2)
- && languageID == ttUSHORT(fc + loc + 4) && nameID == ttUSHORT(fc + loc + 6)) {
- *length = ttUSHORT(fc + loc + 8);
- return (const char *)(fc + stringOffset + ttUSHORT(fc + loc + 10));
- }
- }
- return NULL;
+ stbtt_int32 i,count,stringOffset;
+ stbtt_uint8 *fc = font->data;
+ stbtt_uint32 offset = font->fontstart;
+ stbtt_uint32 nm = stbtt__find_table(fc, offset, "name");
+ if (!nm) return NULL;
+
+ count = ttUSHORT(fc+nm+2);
+ stringOffset = nm + ttUSHORT(fc+nm+4);
+ for (i=0; i < count; ++i) {
+ stbtt_uint32 loc = nm + 6 + 12 * i;
+ if (platformID == ttUSHORT(fc+loc+0) && encodingID == ttUSHORT(fc+loc+2)
+ && languageID == ttUSHORT(fc+loc+4) && nameID == ttUSHORT(fc+loc+6)) {
+ *length = ttUSHORT(fc+loc+8);
+ return (const char *) (fc+stringOffset+ttUSHORT(fc+loc+10));
+ }
+ }
+ return NULL;
}
static int stbtt__matchpair(stbtt_uint8 *fc, stbtt_uint32 nm, stbtt_uint8 *name, stbtt_int32 nlen, stbtt_int32 target_id, stbtt_int32 next_id)
{
- stbtt_int32 i;
- stbtt_int32 count = ttUSHORT(fc + nm + 2);
- stbtt_int32 stringOffset = nm + ttUSHORT(fc + nm + 4);
-
- for (i = 0; i < count; ++i) {
- stbtt_uint32 loc = nm + 6 + 12 * i;
- stbtt_int32 id = ttUSHORT(fc + loc + 6);
- if (id == target_id) {
- // find the encoding
- stbtt_int32 platform = ttUSHORT(fc + loc + 0), encoding = ttUSHORT(fc + loc + 2), language = ttUSHORT(fc + loc + 4);
-
- // is this a Unicode encoding?
- if (platform == 0 || (platform == 3 && encoding == 1) || (platform == 3 && encoding == 10)) {
- stbtt_int32 slen = ttUSHORT(fc + loc + 8);
- stbtt_int32 off = ttUSHORT(fc + loc + 10);
-
- // check if there's a prefix match
- stbtt_int32 matchlen = stbtt__CompareUTF8toUTF16_bigendian_prefix(name, nlen, fc + stringOffset + off, slen);
- if (matchlen >= 0) {
- // check for target_id+1 immediately following, with same encoding & language
- if (i + 1 < count && ttUSHORT(fc + loc + 12 + 6) == next_id && ttUSHORT(fc + loc + 12) == platform && ttUSHORT(fc + loc + 12 + 2) == encoding && ttUSHORT(fc + loc + 12 + 4) == language) {
- slen = ttUSHORT(fc + loc + 12 + 8);
- off = ttUSHORT(fc + loc + 12 + 10);
- if (slen == 0) {
- if (matchlen == nlen)
- return 1;
- }
- else if (matchlen < nlen && name[matchlen] == ' ') {
- ++matchlen;
- if (stbtt_CompareUTF8toUTF16_bigendian_internal((char*)(name + matchlen), nlen - matchlen, (char*)(fc + stringOffset + off), slen))
- return 1;
- }
- }
- else {
- // if nothing immediately following
- if (matchlen == nlen)
- return 1;
- }
- }
+ stbtt_int32 i;
+ stbtt_int32 count = ttUSHORT(fc+nm+2);
+ stbtt_int32 stringOffset = nm + ttUSHORT(fc+nm+4);
+
+ for (i=0; i < count; ++i) {
+ stbtt_uint32 loc = nm + 6 + 12 * i;
+ stbtt_int32 id = ttUSHORT(fc+loc+6);
+ if (id == target_id) {
+ // find the encoding
+ stbtt_int32 platform = ttUSHORT(fc+loc+0), encoding = ttUSHORT(fc+loc+2), language = ttUSHORT(fc+loc+4);
+
+ // is this a Unicode encoding?
+ if (platform == 0 || (platform == 3 && encoding == 1) || (platform == 3 && encoding == 10)) {
+ stbtt_int32 slen = ttUSHORT(fc+loc+8);
+ stbtt_int32 off = ttUSHORT(fc+loc+10);
+
+ // check if there's a prefix match
+ stbtt_int32 matchlen = stbtt__CompareUTF8toUTF16_bigendian_prefix(name, nlen, fc+stringOffset+off,slen);
+ if (matchlen >= 0) {
+ // check for target_id+1 immediately following, with same encoding & language
+ if (i+1 < count && ttUSHORT(fc+loc+12+6) == next_id && ttUSHORT(fc+loc+12) == platform && ttUSHORT(fc+loc+12+2) == encoding && ttUSHORT(fc+loc+12+4) == language) {
+ slen = ttUSHORT(fc+loc+12+8);
+ off = ttUSHORT(fc+loc+12+10);
+ if (slen == 0) {
+ if (matchlen == nlen)
+ return 1;
+ } else if (matchlen < nlen && name[matchlen] == ' ') {
+ ++matchlen;
+ if (stbtt_CompareUTF8toUTF16_bigendian_internal((char*) (name+matchlen), nlen-matchlen, (char*)(fc+stringOffset+off),slen))
+ return 1;
+ }
+ } else {
+ // if nothing immediately following
+ if (matchlen == nlen)
+ return 1;
+ }
}
+ }
- // @TODO handle other encodings
- }
- }
- return 0;
+ // @TODO handle other encodings
+ }
+ }
+ return 0;
}
static int stbtt__matches(stbtt_uint8 *fc, stbtt_uint32 offset, stbtt_uint8 *name, stbtt_int32 flags)
{
- stbtt_int32 nlen = (stbtt_int32)STBTT_strlen((char *)name);
- stbtt_uint32 nm, hd;
- if (!stbtt__isfont(fc + offset)) return 0;
+ stbtt_int32 nlen = (stbtt_int32) STBTT_strlen((char *) name);
+ stbtt_uint32 nm,hd;
+ if (!stbtt__isfont(fc+offset)) return 0;
- // check italics/bold/underline flags in macStyle...
- if (flags) {
- hd = stbtt__find_table(fc, offset, "head");
- if ((ttUSHORT(fc + hd + 44) & 7) != (flags & 7)) return 0;
- }
+ // check italics/bold/underline flags in macStyle...
+ if (flags) {
+ hd = stbtt__find_table(fc, offset, "head");
+ if ((ttUSHORT(fc+hd+44) & 7) != (flags & 7)) return 0;
+ }
- nm = stbtt__find_table(fc, offset, "name");
- if (!nm) return 0;
+ nm = stbtt__find_table(fc, offset, "name");
+ if (!nm) return 0;
- if (flags) {
- // if we checked the macStyle flags, then just check the family and ignore the subfamily
- if (stbtt__matchpair(fc, nm, name, nlen, 16, -1)) return 1;
- if (stbtt__matchpair(fc, nm, name, nlen, 1, -1)) return 1;
- if (stbtt__matchpair(fc, nm, name, nlen, 3, -1)) return 1;
- }
- else {
- if (stbtt__matchpair(fc, nm, name, nlen, 16, 17)) return 1;
- if (stbtt__matchpair(fc, nm, name, nlen, 1, 2)) return 1;
- if (stbtt__matchpair(fc, nm, name, nlen, 3, -1)) return 1;
- }
+ if (flags) {
+ // if we checked the macStyle flags, then just check the family and ignore the subfamily
+ if (stbtt__matchpair(fc, nm, name, nlen, 16, -1)) return 1;
+ if (stbtt__matchpair(fc, nm, name, nlen, 1, -1)) return 1;
+ if (stbtt__matchpair(fc, nm, name, nlen, 3, -1)) return 1;
+ } else {
+ if (stbtt__matchpair(fc, nm, name, nlen, 16, 17)) return 1;
+ if (stbtt__matchpair(fc, nm, name, nlen, 1, 2)) return 1;
+ if (stbtt__matchpair(fc, nm, name, nlen, 3, -1)) return 1;
+ }
- return 0;
+ return 0;
}
static int stbtt_FindMatchingFont_internal(unsigned char *font_collection, char *name_utf8, stbtt_int32 flags)
{
- stbtt_int32 i;
- for (i = 0;; ++i) {
- stbtt_int32 off = stbtt_GetFontOffsetForIndex(font_collection, i);
- if (off < 0) return off;
- if (stbtt__matches((stbtt_uint8 *)font_collection, off, (stbtt_uint8*)name_utf8, flags))
- return off;
- }
+ stbtt_int32 i;
+ for (i=0;;++i) {
+ stbtt_int32 off = stbtt_GetFontOffsetForIndex(font_collection, i);
+ if (off < 0) return off;
+ if (stbtt__matches((stbtt_uint8 *) font_collection, off, (stbtt_uint8*) name_utf8, flags))
+ return off;
+ }
}
#if defined(__GNUC__) || defined(__clang__)
@@ -4781,35 +4933,35 @@ static int stbtt_FindMatchingFont_internal(unsigned char *font_collection, char
#endif
STBTT_DEF int stbtt_BakeFontBitmap(const unsigned char *data, int offset,
- float pixel_height, unsigned char *pixels, int pw, int ph,
- int first_char, int num_chars, stbtt_bakedchar *chardata)
+ float pixel_height, unsigned char *pixels, int pw, int ph,
+ int first_char, int num_chars, stbtt_bakedchar *chardata)
{
- return stbtt_BakeFontBitmap_internal((unsigned char *)data, offset, pixel_height, pixels, pw, ph, first_char, num_chars, chardata);
+ return stbtt_BakeFontBitmap_internal((unsigned char *) data, offset, pixel_height, pixels, pw, ph, first_char, num_chars, chardata);
}
STBTT_DEF int stbtt_GetFontOffsetForIndex(const unsigned char *data, int index)
{
- return stbtt_GetFontOffsetForIndex_internal((unsigned char *)data, index);
+ return stbtt_GetFontOffsetForIndex_internal((unsigned char *) data, index);
}
STBTT_DEF int stbtt_GetNumberOfFonts(const unsigned char *data)
{
- return stbtt_GetNumberOfFonts_internal((unsigned char *)data);
+ return stbtt_GetNumberOfFonts_internal((unsigned char *) data);
}
STBTT_DEF int stbtt_InitFont(stbtt_fontinfo *info, const unsigned char *data, int offset)
{
- return stbtt_InitFont_internal(info, (unsigned char *)data, offset);
+ return stbtt_InitFont_internal(info, (unsigned char *) data, offset);
}
STBTT_DEF int stbtt_FindMatchingFont(const unsigned char *fontdata, const char *name, int flags)
{
- return stbtt_FindMatchingFont_internal((unsigned char *)fontdata, (char *)name, flags);
+ return stbtt_FindMatchingFont_internal((unsigned char *) fontdata, (char *) name, flags);
}
STBTT_DEF int stbtt_CompareUTF8toUTF16_bigendian(const char *s1, int len1, const char *s2, int len2)
{
- return stbtt_CompareUTF8toUTF16_bigendian_internal((char *)s1, len1, (char *)s2, len2);
+ return stbtt_CompareUTF8toUTF16_bigendian_internal((char *) s1, len1, (char *) s2, len2);
}
#if defined(__GNUC__) || defined(__clang__)
@@ -4821,6 +4973,12 @@ STBTT_DEF int stbtt_CompareUTF8toUTF16_bigendian(const char *s1, int len1, const
// FULL VERSION HISTORY
//
+// 1.25 (2021-07-11) many fixes
+// 1.24 (2020-02-05) fix warning
+// 1.23 (2020-02-02) query SVG data for glyphs; query whole kerning table (but only kern not GPOS)
+// 1.22 (2019-08-11) minimize missing-glyph duplication; fix kerning if both 'GPOS' and 'kern' are defined
+// 1.21 (2019-02-25) fix warning
+// 1.20 (2019-02-07) PackFontRange skips missing codepoints; GetScaleFontVMetrics()
// 1.19 (2018-02-11) OpenType GPOS kerning (horizontal only), STBTT_fmod
// 1.18 (2018-01-29) add missing function
// 1.17 (2017-07-23) make more arguments const; doc fix
@@ -4916,4 +5074,4 @@ AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
------------------------------------------------------------------------------
-*/ \ No newline at end of file
+*/
diff --git a/ThirdParty/stb/stb_vorbis.c b/ThirdParty/stb/stb_vorbis.c
index a863042..3e5c250 100644
--- a/ThirdParty/stb/stb_vorbis.c
+++ b/ThirdParty/stb/stb_vorbis.c
@@ -1,4 +1,4 @@
-// Ogg Vorbis audio decoder - v1.14 - public domain
+// Ogg Vorbis audio decoder - v1.22 - public domain
// http://nothings.org/stb_vorbis/
//
// Original version written by Sean Barrett in 2007.
@@ -26,17 +26,28 @@
// Terje Mathisen Niklas Frykholm Andy Hill
// Casey Muratori John Bolton Gargaj
// Laurent Gomila Marc LeBlanc Ronny Chevalier
-// Bernhard Wodo Evan Balster alxprd@github
+// Bernhard Wodo Evan Balster github:alxprd
// Tom Beaumont Ingo Leitgeb Nicolas Guillemot
// Phillip Bennefall Rohit Thiago Goulart
-// manxorist@github saga musix github:infatum
-// Timur Gagiev
+// github:manxorist Saga Musix github:infatum
+// Timur Gagiev Maxwell Koo Peter Waller
+// github:audinowho Dougall Johnson David Reid
+// github:Clownacy Pedro J. Estebanez Remi Verschelde
+// AnthoFoxo github:morlat Gabriel Ravier
//
// Partial history:
+// 1.22 - 2021-07-11 - various small fixes
+// 1.21 - 2021-07-02 - fix bug for files with no comments
+// 1.20 - 2020-07-11 - several small fixes
+// 1.19 - 2020-02-05 - warnings
+// 1.18 - 2020-02-02 - fix seek bugs; parse header comments; misc warnings etc.
+// 1.17 - 2019-07-08 - fix CVE-2019-13217..CVE-2019-13223 (by ForAllSecure)
+// 1.16 - 2019-03-04 - fix warnings
+// 1.15 - 2019-02-07 - explicit failure if Ogg Skeleton data is found
// 1.14 - 2018-02-11 - delete bogus dealloca usage
// 1.13 - 2018-01-29 - fix truncation of last frame (hopefully)
// 1.12 - 2017-11-21 - limit residue begin/end to blocksize/2 to avoid large temp allocs in bad/corrupt files
-// 1.11 - 2017-07-23 - fix MinGW compilation
+// 1.11 - 2017-07-23 - fix MinGW compilation
// 1.10 - 2017-03-03 - more robust seeking; fix negative ilog(); clear error in open_memory
// 1.09 - 2016-04-04 - back out 'truncation of last frame' fix from previous version
// 1.08 - 2016-04-02 - warnings; setup memory leaks; truncation of last frame
@@ -75,307 +86,325 @@
extern "C" {
#endif
- /////////// THREAD SAFETY
+/////////// THREAD SAFETY
- // Individual stb_vorbis* handles are not thread-safe; you cannot decode from
- // them from multiple threads at the same time. However, you can have multiple
- // stb_vorbis* handles and decode from them independently in multiple thrads.
+// Individual stb_vorbis* handles are not thread-safe; you cannot decode from
+// them from multiple threads at the same time. However, you can have multiple
+// stb_vorbis* handles and decode from them independently in multiple thrads.
- /////////// MEMORY ALLOCATION
+/////////// MEMORY ALLOCATION
- // normally stb_vorbis uses malloc() to allocate memory at startup,
- // and alloca() to allocate temporary memory during a frame on the
- // stack. (Memory consumption will depend on the amount of setup
- // data in the file and how you set the compile flags for speed
- // vs. size. In my test files the maximal-size usage is ~150KB.)
- //
- // You can modify the wrapper functions in the source (setup_malloc,
- // setup_temp_malloc, temp_malloc) to change this behavior, or you
- // can use a simpler allocation model: you pass in a buffer from
- // which stb_vorbis will allocate _all_ its memory (including the
- // temp memory). "open" may fail with a VORBIS_outofmem if you
- // do not pass in enough data; there is no way to determine how
- // much you do need except to succeed (at which point you can
- // query get_info to find the exact amount required. yes I know
- // this is lame).
- //
- // If you pass in a non-NULL buffer of the type below, allocation
- // will occur from it as described above. Otherwise just pass NULL
- // to use malloc()/alloca()
+// normally stb_vorbis uses malloc() to allocate memory at startup,
+// and alloca() to allocate temporary memory during a frame on the
+// stack. (Memory consumption will depend on the amount of setup
+// data in the file and how you set the compile flags for speed
+// vs. size. In my test files the maximal-size usage is ~150KB.)
+//
+// You can modify the wrapper functions in the source (setup_malloc,
+// setup_temp_malloc, temp_malloc) to change this behavior, or you
+// can use a simpler allocation model: you pass in a buffer from
+// which stb_vorbis will allocate _all_ its memory (including the
+// temp memory). "open" may fail with a VORBIS_outofmem if you
+// do not pass in enough data; there is no way to determine how
+// much you do need except to succeed (at which point you can
+// query get_info to find the exact amount required. yes I know
+// this is lame).
+//
+// If you pass in a non-NULL buffer of the type below, allocation
+// will occur from it as described above. Otherwise just pass NULL
+// to use malloc()/alloca()
- typedef struct
- {
- char *alloc_buffer;
- int alloc_buffer_length_in_bytes;
- } stb_vorbis_alloc;
+typedef struct
+{
+ char *alloc_buffer;
+ int alloc_buffer_length_in_bytes;
+} stb_vorbis_alloc;
- /////////// FUNCTIONS USEABLE WITH ALL INPUT MODES
+/////////// FUNCTIONS USEABLE WITH ALL INPUT MODES
- typedef struct stb_vorbis stb_vorbis;
+typedef struct stb_vorbis stb_vorbis;
- typedef struct
- {
- unsigned int sample_rate;
- int channels;
+typedef struct
+{
+ unsigned int sample_rate;
+ int channels;
- unsigned int setup_memory_required;
- unsigned int setup_temp_memory_required;
- unsigned int temp_memory_required;
+ unsigned int setup_memory_required;
+ unsigned int setup_temp_memory_required;
+ unsigned int temp_memory_required;
- int max_frame_size;
- } stb_vorbis_info;
+ int max_frame_size;
+} stb_vorbis_info;
- // get general information about the file
- extern stb_vorbis_info stb_vorbis_get_info(stb_vorbis *f);
+typedef struct
+{
+ char *vendor;
- // get the last error detected (clears it, too)
- extern int stb_vorbis_get_error(stb_vorbis *f);
+ int comment_list_length;
+ char **comment_list;
+} stb_vorbis_comment;
- // close an ogg vorbis file and free all memory in use
- extern void stb_vorbis_close(stb_vorbis *f);
+// get general information about the file
+extern stb_vorbis_info stb_vorbis_get_info(stb_vorbis *f);
- // this function returns the offset (in samples) from the beginning of the
- // file that will be returned by the next decode, if it is known, or -1
- // otherwise. after a flush_pushdata() call, this may take a while before
- // it becomes valid again.
- // NOT WORKING YET after a seek with PULLDATA API
- extern int stb_vorbis_get_sample_offset(stb_vorbis *f);
+// get ogg comments
+extern stb_vorbis_comment stb_vorbis_get_comment(stb_vorbis *f);
- // returns the current seek point within the file, or offset from the beginning
- // of the memory buffer. In pushdata mode it returns 0.
- extern unsigned int stb_vorbis_get_file_offset(stb_vorbis *f);
+// get the last error detected (clears it, too)
+extern int stb_vorbis_get_error(stb_vorbis *f);
- /////////// PUSHDATA API
+// close an ogg vorbis file and free all memory in use
+extern void stb_vorbis_close(stb_vorbis *f);
+
+// this function returns the offset (in samples) from the beginning of the
+// file that will be returned by the next decode, if it is known, or -1
+// otherwise. after a flush_pushdata() call, this may take a while before
+// it becomes valid again.
+// NOT WORKING YET after a seek with PULLDATA API
+extern int stb_vorbis_get_sample_offset(stb_vorbis *f);
+
+// returns the current seek point within the file, or offset from the beginning
+// of the memory buffer. In pushdata mode it returns 0.
+extern unsigned int stb_vorbis_get_file_offset(stb_vorbis *f);
+
+/////////// PUSHDATA API
#ifndef STB_VORBIS_NO_PUSHDATA_API
- // this API allows you to get blocks of data from any source and hand
- // them to stb_vorbis. you have to buffer them; stb_vorbis will tell
- // you how much it used, and you have to give it the rest next time;
- // and stb_vorbis may not have enough data to work with and you will
- // need to give it the same data again PLUS more. Note that the Vorbis
- // specification does not bound the size of an individual frame.
-
- extern stb_vorbis *stb_vorbis_open_pushdata(
- const unsigned char * datablock, int datablock_length_in_bytes,
- int *datablock_memory_consumed_in_bytes,
- int *error,
- const stb_vorbis_alloc *alloc_buffer);
- // create a vorbis decoder by passing in the initial data block containing
- // the ogg&vorbis headers (you don't need to do parse them, just provide
- // the first N bytes of the file--you're told if it's not enough, see below)
- // on success, returns an stb_vorbis *, does not set error, returns the amount of
- // data parsed/consumed on this call in *datablock_memory_consumed_in_bytes;
- // on failure, returns NULL on error and sets *error, does not change *datablock_memory_consumed
- // if returns NULL and *error is VORBIS_need_more_data, then the input block was
- // incomplete and you need to pass in a larger block from the start of the file
-
- extern int stb_vorbis_decode_frame_pushdata(
- stb_vorbis *f,
- const unsigned char *datablock, int datablock_length_in_bytes,
- int *channels, // place to write number of float * buffers
- float ***output, // place to write float ** array of float * buffers
- int *samples // place to write number of output samples
- );
- // decode a frame of audio sample data if possible from the passed-in data block
- //
- // return value: number of bytes we used from datablock
- //
- // possible cases:
- // 0 bytes used, 0 samples output (need more data)
- // N bytes used, 0 samples output (resynching the stream, keep going)
- // N bytes used, M samples output (one frame of data)
- // note that after opening a file, you will ALWAYS get one N-bytes,0-sample
- // frame, because Vorbis always "discards" the first frame.
- //
- // Note that on resynch, stb_vorbis will rarely consume all of the buffer,
- // instead only datablock_length_in_bytes-3 or less. This is because it wants
- // to avoid missing parts of a page header if they cross a datablock boundary,
- // without writing state-machiney code to record a partial detection.
- //
- // The number of channels returned are stored in *channels (which can be
- // NULL--it is always the same as the number of channels reported by
- // get_info). *output will contain an array of float* buffers, one per
- // channel. In other words, (*output)[0][0] contains the first sample from
- // the first channel, and (*output)[1][0] contains the first sample from
- // the second channel.
-
- extern void stb_vorbis_flush_pushdata(stb_vorbis *f);
- // inform stb_vorbis that your next datablock will not be contiguous with
- // previous ones (e.g. you've seeked in the data); future attempts to decode
- // frames will cause stb_vorbis to resynchronize (as noted above), and
- // once it sees a valid Ogg page (typically 4-8KB, as large as 64KB), it
- // will begin decoding the _next_ frame.
- //
- // if you want to seek using pushdata, you need to seek in your file, then
- // call stb_vorbis_flush_pushdata(), then start calling decoding, then once
- // decoding is returning you data, call stb_vorbis_get_sample_offset, and
- // if you don't like the result, seek your file again and repeat.
+// this API allows you to get blocks of data from any source and hand
+// them to stb_vorbis. you have to buffer them; stb_vorbis will tell
+// you how much it used, and you have to give it the rest next time;
+// and stb_vorbis may not have enough data to work with and you will
+// need to give it the same data again PLUS more. Note that the Vorbis
+// specification does not bound the size of an individual frame.
+
+extern stb_vorbis *stb_vorbis_open_pushdata(
+ const unsigned char * datablock, int datablock_length_in_bytes,
+ int *datablock_memory_consumed_in_bytes,
+ int *error,
+ const stb_vorbis_alloc *alloc_buffer);
+// create a vorbis decoder by passing in the initial data block containing
+// the ogg&vorbis headers (you don't need to do parse them, just provide
+// the first N bytes of the file--you're told if it's not enough, see below)
+// on success, returns an stb_vorbis *, does not set error, returns the amount of
+// data parsed/consumed on this call in *datablock_memory_consumed_in_bytes;
+// on failure, returns NULL on error and sets *error, does not change *datablock_memory_consumed
+// if returns NULL and *error is VORBIS_need_more_data, then the input block was
+// incomplete and you need to pass in a larger block from the start of the file
+
+extern int stb_vorbis_decode_frame_pushdata(
+ stb_vorbis *f,
+ const unsigned char *datablock, int datablock_length_in_bytes,
+ int *channels, // place to write number of float * buffers
+ float ***output, // place to write float ** array of float * buffers
+ int *samples // place to write number of output samples
+ );
+// decode a frame of audio sample data if possible from the passed-in data block
+//
+// return value: number of bytes we used from datablock
+//
+// possible cases:
+// 0 bytes used, 0 samples output (need more data)
+// N bytes used, 0 samples output (resynching the stream, keep going)
+// N bytes used, M samples output (one frame of data)
+// note that after opening a file, you will ALWAYS get one N-bytes,0-sample
+// frame, because Vorbis always "discards" the first frame.
+//
+// Note that on resynch, stb_vorbis will rarely consume all of the buffer,
+// instead only datablock_length_in_bytes-3 or less. This is because it wants
+// to avoid missing parts of a page header if they cross a datablock boundary,
+// without writing state-machiney code to record a partial detection.
+//
+// The number of channels returned are stored in *channels (which can be
+// NULL--it is always the same as the number of channels reported by
+// get_info). *output will contain an array of float* buffers, one per
+// channel. In other words, (*output)[0][0] contains the first sample from
+// the first channel, and (*output)[1][0] contains the first sample from
+// the second channel.
+//
+// *output points into stb_vorbis's internal output buffer storage; these
+// buffers are owned by stb_vorbis and application code should not free
+// them or modify their contents. They are transient and will be overwritten
+// once you ask for more data to get decoded, so be sure to grab any data
+// you need before then.
+
+extern void stb_vorbis_flush_pushdata(stb_vorbis *f);
+// inform stb_vorbis that your next datablock will not be contiguous with
+// previous ones (e.g. you've seeked in the data); future attempts to decode
+// frames will cause stb_vorbis to resynchronize (as noted above), and
+// once it sees a valid Ogg page (typically 4-8KB, as large as 64KB), it
+// will begin decoding the _next_ frame.
+//
+// if you want to seek using pushdata, you need to seek in your file, then
+// call stb_vorbis_flush_pushdata(), then start calling decoding, then once
+// decoding is returning you data, call stb_vorbis_get_sample_offset, and
+// if you don't like the result, seek your file again and repeat.
#endif
- ////////// PULLING INPUT API
+////////// PULLING INPUT API
#ifndef STB_VORBIS_NO_PULLDATA_API
- // This API assumes stb_vorbis is allowed to pull data from a source--
- // either a block of memory containing the _entire_ vorbis stream, or a
- // FILE * that you or it create, or possibly some other reading mechanism
- // if you go modify the source to replace the FILE * case with some kind
- // of callback to your code. (But if you don't support seeking, you may
- // just want to go ahead and use pushdata.)
+// This API assumes stb_vorbis is allowed to pull data from a source--
+// either a block of memory containing the _entire_ vorbis stream, or a
+// FILE * that you or it create, or possibly some other reading mechanism
+// if you go modify the source to replace the FILE * case with some kind
+// of callback to your code. (But if you don't support seeking, you may
+// just want to go ahead and use pushdata.)
#if !defined(STB_VORBIS_NO_STDIO) && !defined(STB_VORBIS_NO_INTEGER_CONVERSION)
- extern int stb_vorbis_decode_filename(const char *filename, int *channels, int *sample_rate, short **output);
+extern int stb_vorbis_decode_filename(const char *filename, int *channels, int *sample_rate, short **output);
#endif
#if !defined(STB_VORBIS_NO_INTEGER_CONVERSION)
- extern int stb_vorbis_decode_memory(const unsigned char *mem, int len, int *channels, int *sample_rate, short **output);
+extern int stb_vorbis_decode_memory(const unsigned char *mem, int len, int *channels, int *sample_rate, short **output);
#endif
- // decode an entire file and output the data interleaved into a malloc()ed
- // buffer stored in *output. The return value is the number of samples
- // decoded, or -1 if the file could not be opened or was not an ogg vorbis file.
- // When you're done with it, just free() the pointer returned in *output.
+// decode an entire file and output the data interleaved into a malloc()ed
+// buffer stored in *output. The return value is the number of samples
+// decoded, or -1 if the file could not be opened or was not an ogg vorbis file.
+// When you're done with it, just free() the pointer returned in *output.
- extern stb_vorbis * stb_vorbis_open_memory(const unsigned char *data, int len,
- int *error, const stb_vorbis_alloc *alloc_buffer);
- // create an ogg vorbis decoder from an ogg vorbis stream in memory (note
- // this must be the entire stream!). on failure, returns NULL and sets *error
+extern stb_vorbis * stb_vorbis_open_memory(const unsigned char *data, int len,
+ int *error, const stb_vorbis_alloc *alloc_buffer);
+// create an ogg vorbis decoder from an ogg vorbis stream in memory (note
+// this must be the entire stream!). on failure, returns NULL and sets *error
#ifndef STB_VORBIS_NO_STDIO
- extern stb_vorbis * stb_vorbis_open_filename(const char *filename,
- int *error, const stb_vorbis_alloc *alloc_buffer);
- // create an ogg vorbis decoder from a filename via fopen(). on failure,
- // returns NULL and sets *error (possibly to VORBIS_file_open_failure).
-
- extern stb_vorbis * stb_vorbis_open_file(FILE *f, int close_handle_on_close,
- int *error, const stb_vorbis_alloc *alloc_buffer);
- // create an ogg vorbis decoder from an open FILE *, looking for a stream at
- // the _current_ seek point (ftell). on failure, returns NULL and sets *error.
- // note that stb_vorbis must "own" this stream; if you seek it in between
- // calls to stb_vorbis, it will become confused. Morever, if you attempt to
- // perform stb_vorbis_seek_*() operations on this file, it will assume it
- // owns the _entire_ rest of the file after the start point. Use the next
- // function, stb_vorbis_open_file_section(), to limit it.
-
- extern stb_vorbis * stb_vorbis_open_file_section(FILE *f, int close_handle_on_close,
- int *error, const stb_vorbis_alloc *alloc_buffer, unsigned int len);
- // create an ogg vorbis decoder from an open FILE *, looking for a stream at
- // the _current_ seek point (ftell); the stream will be of length 'len' bytes.
- // on failure, returns NULL and sets *error. note that stb_vorbis must "own"
- // this stream; if you seek it in between calls to stb_vorbis, it will become
- // confused.
+extern stb_vorbis * stb_vorbis_open_filename(const char *filename,
+ int *error, const stb_vorbis_alloc *alloc_buffer);
+// create an ogg vorbis decoder from a filename via fopen(). on failure,
+// returns NULL and sets *error (possibly to VORBIS_file_open_failure).
+
+extern stb_vorbis * stb_vorbis_open_file(FILE *f, int close_handle_on_close,
+ int *error, const stb_vorbis_alloc *alloc_buffer);
+// create an ogg vorbis decoder from an open FILE *, looking for a stream at
+// the _current_ seek point (ftell). on failure, returns NULL and sets *error.
+// note that stb_vorbis must "own" this stream; if you seek it in between
+// calls to stb_vorbis, it will become confused. Moreover, if you attempt to
+// perform stb_vorbis_seek_*() operations on this file, it will assume it
+// owns the _entire_ rest of the file after the start point. Use the next
+// function, stb_vorbis_open_file_section(), to limit it.
+
+extern stb_vorbis * stb_vorbis_open_file_section(FILE *f, int close_handle_on_close,
+ int *error, const stb_vorbis_alloc *alloc_buffer, unsigned int len);
+// create an ogg vorbis decoder from an open FILE *, looking for a stream at
+// the _current_ seek point (ftell); the stream will be of length 'len' bytes.
+// on failure, returns NULL and sets *error. note that stb_vorbis must "own"
+// this stream; if you seek it in between calls to stb_vorbis, it will become
+// confused.
#endif
- extern int stb_vorbis_seek_frame(stb_vorbis *f, unsigned int sample_number);
- extern int stb_vorbis_seek(stb_vorbis *f, unsigned int sample_number);
- // these functions seek in the Vorbis file to (approximately) 'sample_number'.
- // after calling seek_frame(), the next call to get_frame_*() will include
- // the specified sample. after calling stb_vorbis_seek(), the next call to
- // stb_vorbis_get_samples_* will start with the specified sample. If you
- // do not need to seek to EXACTLY the target sample when using get_samples_*,
- // you can also use seek_frame().
-
- extern int stb_vorbis_seek_start(stb_vorbis *f);
- // this function is equivalent to stb_vorbis_seek(f,0)
-
- extern unsigned int stb_vorbis_stream_length_in_samples(stb_vorbis *f);
- extern float stb_vorbis_stream_length_in_seconds(stb_vorbis *f);
- // these functions return the total length of the vorbis stream
-
- extern int stb_vorbis_get_frame_float(stb_vorbis *f, int *channels, float ***output);
- // decode the next frame and return the number of samples. the number of
- // channels returned are stored in *channels (which can be NULL--it is always
- // the same as the number of channels reported by get_info). *output will
- // contain an array of float* buffers, one per channel. These outputs will
- // be overwritten on the next call to stb_vorbis_get_frame_*.
- //
- // You generally should not intermix calls to stb_vorbis_get_frame_*()
- // and stb_vorbis_get_samples_*(), since the latter calls the former.
+extern int stb_vorbis_seek_frame(stb_vorbis *f, unsigned int sample_number);
+extern int stb_vorbis_seek(stb_vorbis *f, unsigned int sample_number);
+// these functions seek in the Vorbis file to (approximately) 'sample_number'.
+// after calling seek_frame(), the next call to get_frame_*() will include
+// the specified sample. after calling stb_vorbis_seek(), the next call to
+// stb_vorbis_get_samples_* will start with the specified sample. If you
+// do not need to seek to EXACTLY the target sample when using get_samples_*,
+// you can also use seek_frame().
+
+extern int stb_vorbis_seek_start(stb_vorbis *f);
+// this function is equivalent to stb_vorbis_seek(f,0)
+
+extern unsigned int stb_vorbis_stream_length_in_samples(stb_vorbis *f);
+extern float stb_vorbis_stream_length_in_seconds(stb_vorbis *f);
+// these functions return the total length of the vorbis stream
+
+extern int stb_vorbis_get_frame_float(stb_vorbis *f, int *channels, float ***output);
+// decode the next frame and return the number of samples. the number of
+// channels returned are stored in *channels (which can be NULL--it is always
+// the same as the number of channels reported by get_info). *output will
+// contain an array of float* buffers, one per channel. These outputs will
+// be overwritten on the next call to stb_vorbis_get_frame_*.
+//
+// You generally should not intermix calls to stb_vorbis_get_frame_*()
+// and stb_vorbis_get_samples_*(), since the latter calls the former.
#ifndef STB_VORBIS_NO_INTEGER_CONVERSION
- extern int stb_vorbis_get_frame_short_interleaved(stb_vorbis *f, int num_c, short *buffer, int num_shorts);
- extern int stb_vorbis_get_frame_short(stb_vorbis *f, int num_c, short **buffer, int num_samples);
+extern int stb_vorbis_get_frame_short_interleaved(stb_vorbis *f, int num_c, short *buffer, int num_shorts);
+extern int stb_vorbis_get_frame_short (stb_vorbis *f, int num_c, short **buffer, int num_samples);
#endif
- // decode the next frame and return the number of *samples* per channel.
- // Note that for interleaved data, you pass in the number of shorts (the
- // size of your array), but the return value is the number of samples per
- // channel, not the total number of samples.
- //
- // The data is coerced to the number of channels you request according to the
- // channel coercion rules (see below). You must pass in the size of your
- // buffer(s) so that stb_vorbis will not overwrite the end of the buffer.
- // The maximum buffer size needed can be gotten from get_info(); however,
- // the Vorbis I specification implies an absolute maximum of 4096 samples
- // per channel.
-
- // Channel coercion rules:
- // Let M be the number of channels requested, and N the number of channels present,
- // and Cn be the nth channel; let stereo L be the sum of all L and center channels,
- // and stereo R be the sum of all R and center channels (channel assignment from the
- // vorbis spec).
- // M N output
- // 1 k sum(Ck) for all k
- // 2 * stereo L, stereo R
- // k l k > l, the first l channels, then 0s
- // k l k <= l, the first k channels
- // Note that this is not _good_ surround etc. mixing at all! It's just so
- // you get something useful.
-
- extern int stb_vorbis_get_samples_float_interleaved(stb_vorbis *f, int channels, float *buffer, int num_floats);
- extern int stb_vorbis_get_samples_float(stb_vorbis *f, int channels, float **buffer, int num_samples);
- // gets num_samples samples, not necessarily on a frame boundary--this requires
- // buffering so you have to supply the buffers. DOES NOT APPLY THE COERCION RULES.
- // Returns the number of samples stored per channel; it may be less than requested
- // at the end of the file. If there are no more samples in the file, returns 0.
+// decode the next frame and return the number of *samples* per channel.
+// Note that for interleaved data, you pass in the number of shorts (the
+// size of your array), but the return value is the number of samples per
+// channel, not the total number of samples.
+//
+// The data is coerced to the number of channels you request according to the
+// channel coercion rules (see below). You must pass in the size of your
+// buffer(s) so that stb_vorbis will not overwrite the end of the buffer.
+// The maximum buffer size needed can be gotten from get_info(); however,
+// the Vorbis I specification implies an absolute maximum of 4096 samples
+// per channel.
+
+// Channel coercion rules:
+// Let M be the number of channels requested, and N the number of channels present,
+// and Cn be the nth channel; let stereo L be the sum of all L and center channels,
+// and stereo R be the sum of all R and center channels (channel assignment from the
+// vorbis spec).
+// M N output
+// 1 k sum(Ck) for all k
+// 2 * stereo L, stereo R
+// k l k > l, the first l channels, then 0s
+// k l k <= l, the first k channels
+// Note that this is not _good_ surround etc. mixing at all! It's just so
+// you get something useful.
+
+extern int stb_vorbis_get_samples_float_interleaved(stb_vorbis *f, int channels, float *buffer, int num_floats);
+extern int stb_vorbis_get_samples_float(stb_vorbis *f, int channels, float **buffer, int num_samples);
+// gets num_samples samples, not necessarily on a frame boundary--this requires
+// buffering so you have to supply the buffers. DOES NOT APPLY THE COERCION RULES.
+// Returns the number of samples stored per channel; it may be less than requested
+// at the end of the file. If there are no more samples in the file, returns 0.
#ifndef STB_VORBIS_NO_INTEGER_CONVERSION
- extern int stb_vorbis_get_samples_short_interleaved(stb_vorbis *f, int channels, short *buffer, int num_shorts);
- extern int stb_vorbis_get_samples_short(stb_vorbis *f, int channels, short **buffer, int num_samples);
+extern int stb_vorbis_get_samples_short_interleaved(stb_vorbis *f, int channels, short *buffer, int num_shorts);
+extern int stb_vorbis_get_samples_short(stb_vorbis *f, int channels, short **buffer, int num_samples);
#endif
- // gets num_samples samples, not necessarily on a frame boundary--this requires
- // buffering so you have to supply the buffers. Applies the coercion rules above
- // to produce 'channels' channels. Returns the number of samples stored per channel;
- // it may be less than requested at the end of the file. If there are no more
- // samples in the file, returns 0.
+// gets num_samples samples, not necessarily on a frame boundary--this requires
+// buffering so you have to supply the buffers. Applies the coercion rules above
+// to produce 'channels' channels. Returns the number of samples stored per channel;
+// it may be less than requested at the end of the file. If there are no more
+// samples in the file, returns 0.
#endif
- //////// ERROR CODES
+//////// ERROR CODES
- enum STBVorbisError
- {
- VORBIS__no_error,
+enum STBVorbisError
+{
+ VORBIS__no_error,
- VORBIS_need_more_data = 1, // not a real error
+ VORBIS_need_more_data=1, // not a real error
- VORBIS_invalid_api_mixing, // can't mix API modes
- VORBIS_outofmem, // not enough memory
- VORBIS_feature_not_supported, // uses floor 0
- VORBIS_too_many_channels, // STB_VORBIS_MAX_CHANNELS is too small
- VORBIS_file_open_failure, // fopen() failed
- VORBIS_seek_without_length, // can't seek in unknown-length file
+ VORBIS_invalid_api_mixing, // can't mix API modes
+ VORBIS_outofmem, // not enough memory
+ VORBIS_feature_not_supported, // uses floor 0
+ VORBIS_too_many_channels, // STB_VORBIS_MAX_CHANNELS is too small
+ VORBIS_file_open_failure, // fopen() failed
+ VORBIS_seek_without_length, // can't seek in unknown-length file
- VORBIS_unexpected_eof = 10, // file is truncated?
- VORBIS_seek_invalid, // seek past EOF
+ VORBIS_unexpected_eof=10, // file is truncated?
+ VORBIS_seek_invalid, // seek past EOF
- // decoding errors (corrupt/invalid stream) -- you probably
- // don't care about the exact details of these
+ // decoding errors (corrupt/invalid stream) -- you probably
+ // don't care about the exact details of these
- // vorbis errors:
- VORBIS_invalid_setup = 20,
- VORBIS_invalid_stream,
+ // vorbis errors:
+ VORBIS_invalid_setup=20,
+ VORBIS_invalid_stream,
- // ogg errors:
- VORBIS_missing_capture_pattern = 30,
- VORBIS_invalid_stream_structure_version,
- VORBIS_continued_packet_flag_invalid,
- VORBIS_incorrect_stream_serial_number,
- VORBIS_invalid_first_page,
- VORBIS_bad_packet_type,
- VORBIS_cant_find_last_page,
- VORBIS_seek_failed
- };
+ // ogg errors:
+ VORBIS_missing_capture_pattern=30,
+ VORBIS_invalid_stream_structure_version,
+ VORBIS_continued_packet_flag_invalid,
+ VORBIS_incorrect_stream_serial_number,
+ VORBIS_invalid_first_page,
+ VORBIS_bad_packet_type,
+ VORBIS_cant_find_last_page,
+ VORBIS_seek_failed,
+ VORBIS_ogg_skeleton_not_supported
+};
#ifdef __cplusplus
@@ -521,25 +550,25 @@ extern "C" {
//////////////////////////////////////////////////////////////////////////////
#ifdef STB_VORBIS_NO_PULLDATA_API
-#define STB_VORBIS_NO_INTEGER_CONVERSION
-#define STB_VORBIS_NO_STDIO
+ #define STB_VORBIS_NO_INTEGER_CONVERSION
+ #define STB_VORBIS_NO_STDIO
#endif
#if defined(STB_VORBIS_NO_CRT) && !defined(STB_VORBIS_NO_STDIO)
-#define STB_VORBIS_NO_STDIO 1
+ #define STB_VORBIS_NO_STDIO 1
#endif
#ifndef STB_VORBIS_NO_INTEGER_CONVERSION
#ifndef STB_VORBIS_NO_FAST_SCALED_FLOAT
-// only need endianness for fast-float-to-int, which we don't
-// use for pushdata
+ // only need endianness for fast-float-to-int, which we don't
+ // use for pushdata
-#ifndef STB_VORBIS_BIG_ENDIAN
-#define STB_VORBIS_ENDIAN 0
-#else
-#define STB_VORBIS_ENDIAN 1
-#endif
+ #ifndef STB_VORBIS_BIG_ENDIAN
+ #define STB_VORBIS_ENDIAN 0
+ #else
+ #define STB_VORBIS_ENDIAN 1
+ #endif
#endif
#endif
@@ -550,44 +579,46 @@ extern "C" {
#endif
#ifndef STB_VORBIS_NO_CRT
-#include <stdlib.h>
-#include <string.h>
-#include <assert.h>
-#include <math.h>
-
-// find definition of alloca if it's not in stdlib.h:
-#if defined(_MSC_VER) || defined(__MINGW32__)
-#include <malloc.h>
-#endif
-#if defined(__linux__) || defined(__linux) || defined(__EMSCRIPTEN__)
-#include <alloca.h>
-#endif
+ #include <stdlib.h>
+ #include <string.h>
+ #include <assert.h>
+ #include <math.h>
+
+ // find definition of alloca if it's not in stdlib.h:
+ #if defined(_MSC_VER) || defined(__MINGW32__)
+ #include <malloc.h>
+ #endif
+ #if defined(__linux__) || defined(__linux) || defined(__sun__) || defined(__EMSCRIPTEN__) || defined(__NEWLIB__)
+ #include <alloca.h>
+ #endif
#else // STB_VORBIS_NO_CRT
-#define NULL 0
-#define malloc(s) 0
-#define free(s) ((void) 0)
-#define realloc(s) 0
+ #define NULL 0
+ #define malloc(s) 0
+ #define free(s) ((void) 0)
+ #define realloc(s) 0
#endif // STB_VORBIS_NO_CRT
#include <limits.h>
#ifdef __MINGW32__
-// eff you mingw:
-// "fixed":
-// http://sourceforge.net/p/mingw-w64/mailman/message/32882927/
-// "no that broke the build, reverted, who cares about C":
-// http://sourceforge.net/p/mingw-w64/mailman/message/32890381/
-#ifdef __forceinline
-#undef __forceinline
-#endif
-#define __forceinline
-#define alloca __builtin_alloca
+ // eff you mingw:
+ // "fixed":
+ // http://sourceforge.net/p/mingw-w64/mailman/message/32882927/
+ // "no that broke the build, reverted, who cares about C":
+ // http://sourceforge.net/p/mingw-w64/mailman/message/32890381/
+ #ifdef __forceinline
+ #undef __forceinline
+ #endif
+ #define __forceinline
+ #ifndef alloca
+ #define alloca __builtin_alloca
+ #endif
#elif !defined(_MSC_VER)
-#if __GNUC__
-#define __forceinline inline
-#else
-#define __forceinline
-#endif
+ #if __GNUC__
+ #define __forceinline inline
+ #else
+ #define __forceinline
+ #endif
#endif
#if STB_VORBIS_MAX_CHANNELS > 256
@@ -624,6 +655,12 @@ typedef signed int int32;
typedef float codetype;
+#ifdef _MSC_VER
+#define STBV_NOTUSED(v) (void)(v)
+#else
+#define STBV_NOTUSED(v) (void)sizeof(v)
+#endif
+
// @NOTE
//
// Some arrays below are tagged "//varies", which means it's actually
@@ -642,237 +679,244 @@ typedef float codetype;
typedef struct
{
- int dimensions, entries;
- uint8 *codeword_lengths;
- float minimum_value;
- float delta_value;
- uint8 value_bits;
- uint8 lookup_type;
- uint8 sequence_p;
- uint8 sparse;
- uint32 lookup_values;
- codetype *multiplicands;
- uint32 *codewords;
-#ifdef STB_VORBIS_FAST_HUFFMAN_SHORT
+ int dimensions, entries;
+ uint8 *codeword_lengths;
+ float minimum_value;
+ float delta_value;
+ uint8 value_bits;
+ uint8 lookup_type;
+ uint8 sequence_p;
+ uint8 sparse;
+ uint32 lookup_values;
+ codetype *multiplicands;
+ uint32 *codewords;
+ #ifdef STB_VORBIS_FAST_HUFFMAN_SHORT
int16 fast_huffman[FAST_HUFFMAN_TABLE_SIZE];
-#else
+ #else
int32 fast_huffman[FAST_HUFFMAN_TABLE_SIZE];
-#endif
- uint32 *sorted_codewords;
- int *sorted_values;
- int sorted_entries;
+ #endif
+ uint32 *sorted_codewords;
+ int *sorted_values;
+ int sorted_entries;
} Codebook;
typedef struct
{
- uint8 order;
- uint16 rate;
- uint16 bark_map_size;
- uint8 amplitude_bits;
- uint8 amplitude_offset;
- uint8 number_of_books;
- uint8 book_list[16]; // varies
+ uint8 order;
+ uint16 rate;
+ uint16 bark_map_size;
+ uint8 amplitude_bits;
+ uint8 amplitude_offset;
+ uint8 number_of_books;
+ uint8 book_list[16]; // varies
} Floor0;
typedef struct
{
- uint8 partitions;
- uint8 partition_class_list[32]; // varies
- uint8 class_dimensions[16]; // varies
- uint8 class_subclasses[16]; // varies
- uint8 class_masterbooks[16]; // varies
- int16 subclass_books[16][8]; // varies
- uint16 Xlist[31 * 8 + 2]; // varies
- uint8 sorted_order[31 * 8 + 2];
- uint8 neighbors[31 * 8 + 2][2];
- uint8 floor1_multiplier;
- uint8 rangebits;
- int values;
+ uint8 partitions;
+ uint8 partition_class_list[32]; // varies
+ uint8 class_dimensions[16]; // varies
+ uint8 class_subclasses[16]; // varies
+ uint8 class_masterbooks[16]; // varies
+ int16 subclass_books[16][8]; // varies
+ uint16 Xlist[31*8+2]; // varies
+ uint8 sorted_order[31*8+2];
+ uint8 neighbors[31*8+2][2];
+ uint8 floor1_multiplier;
+ uint8 rangebits;
+ int values;
} Floor1;
typedef union
{
- Floor0 floor0;
- Floor1 floor1;
+ Floor0 floor0;
+ Floor1 floor1;
} Floor;
typedef struct
{
- uint32 begin, end;
- uint32 part_size;
- uint8 classifications;
- uint8 classbook;
- uint8 **classdata;
- int16(*residue_books)[8];
+ uint32 begin, end;
+ uint32 part_size;
+ uint8 classifications;
+ uint8 classbook;
+ uint8 **classdata;
+ int16 (*residue_books)[8];
} Residue;
typedef struct
{
- uint8 magnitude;
- uint8 angle;
- uint8 mux;
+ uint8 magnitude;
+ uint8 angle;
+ uint8 mux;
} MappingChannel;
typedef struct
{
- uint16 coupling_steps;
- MappingChannel *chan;
- uint8 submaps;
- uint8 submap_floor[15]; // varies
- uint8 submap_residue[15]; // varies
+ uint16 coupling_steps;
+ MappingChannel *chan;
+ uint8 submaps;
+ uint8 submap_floor[15]; // varies
+ uint8 submap_residue[15]; // varies
} Mapping;
typedef struct
{
- uint8 blockflag;
- uint8 mapping;
- uint16 windowtype;
- uint16 transformtype;
+ uint8 blockflag;
+ uint8 mapping;
+ uint16 windowtype;
+ uint16 transformtype;
} Mode;
typedef struct
{
- uint32 goal_crc; // expected crc if match
- int bytes_left; // bytes left in packet
- uint32 crc_so_far; // running crc
- int bytes_done; // bytes processed in _current_ chunk
- uint32 sample_loc; // granule pos encoded in page
+ uint32 goal_crc; // expected crc if match
+ int bytes_left; // bytes left in packet
+ uint32 crc_so_far; // running crc
+ int bytes_done; // bytes processed in _current_ chunk
+ uint32 sample_loc; // granule pos encoded in page
} CRCscan;
typedef struct
{
- uint32 page_start, page_end;
- uint32 last_decoded_sample;
+ uint32 page_start, page_end;
+ uint32 last_decoded_sample;
} ProbedPage;
struct stb_vorbis
{
- // user-accessible info
- unsigned int sample_rate;
- int channels;
-
- unsigned int setup_memory_required;
- unsigned int temp_memory_required;
- unsigned int setup_temp_memory_required;
-
- // input config
-#ifndef STB_VORBIS_NO_STDIO
- FILE *f;
- uint32 f_start;
- int close_on_free;
-#endif
-
- uint8 *stream;
- uint8 *stream_start;
- uint8 *stream_end;
-
- uint32 stream_len;
-
- uint8 push_mode;
-
- uint32 first_audio_page_offset;
+ // user-accessible info
+ unsigned int sample_rate;
+ int channels;
- ProbedPage p_first, p_last;
+ unsigned int setup_memory_required;
+ unsigned int temp_memory_required;
+ unsigned int setup_temp_memory_required;
- // memory management
- stb_vorbis_alloc alloc;
- int setup_offset;
- int temp_offset;
+ char *vendor;
+ int comment_list_length;
+ char **comment_list;
- // run-time results
- int eof;
- enum STBVorbisError error;
-
- // user-useful data
-
- // header info
- int blocksize[2];
- int blocksize_0, blocksize_1;
- int codebook_count;
- Codebook *codebooks;
- int floor_count;
- uint16 floor_types[64]; // varies
- Floor *floor_config;
- int residue_count;
- uint16 residue_types[64]; // varies
- Residue *residue_config;
- int mapping_count;
- Mapping *mapping;
- int mode_count;
- Mode mode_config[64]; // varies
-
- uint32 total_samples;
-
- // decode buffer
- float *channel_buffers[STB_VORBIS_MAX_CHANNELS];
- float *outputs[STB_VORBIS_MAX_CHANNELS];
-
- float *previous_window[STB_VORBIS_MAX_CHANNELS];
- int previous_length;
-
-#ifndef STB_VORBIS_NO_DEFER_FLOOR
- int16 *finalY[STB_VORBIS_MAX_CHANNELS];
-#else
- float *floor_buffers[STB_VORBIS_MAX_CHANNELS];
+ // input config
+#ifndef STB_VORBIS_NO_STDIO
+ FILE *f;
+ uint32 f_start;
+ int close_on_free;
#endif
- uint32 current_loc; // sample location of next frame to decode
- int current_loc_valid;
-
- // per-blocksize precomputed data
-
- // twiddle factors
- float *A[2], *B[2], *C[2];
- float *window[2];
- uint16 *bit_reverse[2];
-
- // current page/packet/segment streaming info
- uint32 serial; // stream serial number for verification
- int last_page;
- int segment_count;
- uint8 segments[255];
- uint8 page_flag;
- uint8 bytes_in_seg;
- uint8 first_decode;
- int next_seg;
- int last_seg; // flag that we're on the last segment
- int last_seg_which; // what was the segment number of the last seg?
- uint32 acc;
- int valid_bits;
- int packet_bytes;
- int end_seg_with_known_loc;
- uint32 known_loc_for_packet;
- int discard_samples_deferred;
- uint32 samples_output;
-
- // push mode scanning
- int page_crc_tests; // only in push_mode: number of tests active; -1 if not searching
+ uint8 *stream;
+ uint8 *stream_start;
+ uint8 *stream_end;
+
+ uint32 stream_len;
+
+ uint8 push_mode;
+
+ // the page to seek to when seeking to start, may be zero
+ uint32 first_audio_page_offset;
+
+ // p_first is the page on which the first audio packet ends
+ // (but not necessarily the page on which it starts)
+ ProbedPage p_first, p_last;
+
+ // memory management
+ stb_vorbis_alloc alloc;
+ int setup_offset;
+ int temp_offset;
+
+ // run-time results
+ int eof;
+ enum STBVorbisError error;
+
+ // user-useful data
+
+ // header info
+ int blocksize[2];
+ int blocksize_0, blocksize_1;
+ int codebook_count;
+ Codebook *codebooks;
+ int floor_count;
+ uint16 floor_types[64]; // varies
+ Floor *floor_config;
+ int residue_count;
+ uint16 residue_types[64]; // varies
+ Residue *residue_config;
+ int mapping_count;
+ Mapping *mapping;
+ int mode_count;
+ Mode mode_config[64]; // varies
+
+ uint32 total_samples;
+
+ // decode buffer
+ float *channel_buffers[STB_VORBIS_MAX_CHANNELS];
+ float *outputs [STB_VORBIS_MAX_CHANNELS];
+
+ float *previous_window[STB_VORBIS_MAX_CHANNELS];
+ int previous_length;
+
+ #ifndef STB_VORBIS_NO_DEFER_FLOOR
+ int16 *finalY[STB_VORBIS_MAX_CHANNELS];
+ #else
+ float *floor_buffers[STB_VORBIS_MAX_CHANNELS];
+ #endif
+
+ uint32 current_loc; // sample location of next frame to decode
+ int current_loc_valid;
+
+ // per-blocksize precomputed data
+
+ // twiddle factors
+ float *A[2],*B[2],*C[2];
+ float *window[2];
+ uint16 *bit_reverse[2];
+
+ // current page/packet/segment streaming info
+ uint32 serial; // stream serial number for verification
+ int last_page;
+ int segment_count;
+ uint8 segments[255];
+ uint8 page_flag;
+ uint8 bytes_in_seg;
+ uint8 first_decode;
+ int next_seg;
+ int last_seg; // flag that we're on the last segment
+ int last_seg_which; // what was the segment number of the last seg?
+ uint32 acc;
+ int valid_bits;
+ int packet_bytes;
+ int end_seg_with_known_loc;
+ uint32 known_loc_for_packet;
+ int discard_samples_deferred;
+ uint32 samples_output;
+
+ // push mode scanning
+ int page_crc_tests; // only in push_mode: number of tests active; -1 if not searching
#ifndef STB_VORBIS_NO_PUSHDATA_API
- CRCscan scan[STB_VORBIS_PUSHDATA_CRC_COUNT];
+ CRCscan scan[STB_VORBIS_PUSHDATA_CRC_COUNT];
#endif
- // sample-access
- int channel_buffer_start;
- int channel_buffer_end;
+ // sample-access
+ int channel_buffer_start;
+ int channel_buffer_end;
};
#if defined(STB_VORBIS_NO_PUSHDATA_API)
-#define IS_PUSH_MODE(f) FALSE
+ #define IS_PUSH_MODE(f) FALSE
#elif defined(STB_VORBIS_NO_PULLDATA_API)
-#define IS_PUSH_MODE(f) TRUE
+ #define IS_PUSH_MODE(f) TRUE
#else
-#define IS_PUSH_MODE(f) ((f)->push_mode)
+ #define IS_PUSH_MODE(f) ((f)->push_mode)
#endif
typedef struct stb_vorbis vorb;
static int error(vorb *f, enum STBVorbisError e)
{
- f->error = e;
- if (!f->eof && e != VORBIS_need_more_data) {
- f->error = e; // breakpoint for debugging
- }
- return 0;
+ f->error = e;
+ if (!f->eof && e != VORBIS_need_more_data) {
+ f->error=e; // breakpoint for debugging
+ }
+ return 0;
}
@@ -884,7 +928,7 @@ static int error(vorb *f, enum STBVorbisError e)
#define array_size_required(count,size) (count*(sizeof(void *)+(size)))
#define temp_alloc(f,size) (f->alloc.alloc_buffer ? setup_temp_malloc(f,size) : alloca(size))
-#define temp_free(f,p) 0
+#define temp_free(f,p) (void)0
#define temp_alloc_save(f) ((f)->temp_offset)
#define temp_alloc_restore(f,p) ((f)->temp_offset = (p))
@@ -893,53 +937,53 @@ static int error(vorb *f, enum STBVorbisError e)
// given a sufficiently large block of memory, make an array of pointers to subblocks of it
static void *make_block_array(void *mem, int count, int size)
{
- int i;
- void ** p = (void **)mem;
- char *q = (char *)(p + count);
- for (i = 0; i < count; ++i) {
- p[i] = q;
- q += size;
- }
- return p;
+ int i;
+ void ** p = (void **) mem;
+ char *q = (char *) (p + count);
+ for (i=0; i < count; ++i) {
+ p[i] = q;
+ q += size;
+ }
+ return p;
}
static void *setup_malloc(vorb *f, int sz)
{
- sz = (sz + 3) & ~3;
- f->setup_memory_required += sz;
- if (f->alloc.alloc_buffer) {
- void *p = (char *)f->alloc.alloc_buffer + f->setup_offset;
- if (f->setup_offset + sz > f->temp_offset) return NULL;
- f->setup_offset += sz;
- return p;
- }
- return sz ? malloc(sz) : NULL;
+ sz = (sz+7) & ~7; // round up to nearest 8 for alignment of future allocs.
+ f->setup_memory_required += sz;
+ if (f->alloc.alloc_buffer) {
+ void *p = (char *) f->alloc.alloc_buffer + f->setup_offset;
+ if (f->setup_offset + sz > f->temp_offset) return NULL;
+ f->setup_offset += sz;
+ return p;
+ }
+ return sz ? malloc(sz) : NULL;
}
static void setup_free(vorb *f, void *p)
{
- if (f->alloc.alloc_buffer) return; // do nothing; setup mem is a stack
- free(p);
+ if (f->alloc.alloc_buffer) return; // do nothing; setup mem is a stack
+ free(p);
}
static void *setup_temp_malloc(vorb *f, int sz)
{
- sz = (sz + 3) & ~3;
- if (f->alloc.alloc_buffer) {
- if (f->temp_offset - sz < f->setup_offset) return NULL;
- f->temp_offset -= sz;
- return (char *)f->alloc.alloc_buffer + f->temp_offset;
- }
- return malloc(sz);
+ sz = (sz+7) & ~7; // round up to nearest 8 for alignment of future allocs.
+ if (f->alloc.alloc_buffer) {
+ if (f->temp_offset - sz < f->setup_offset) return NULL;
+ f->temp_offset -= sz;
+ return (char *) f->alloc.alloc_buffer + f->temp_offset;
+ }
+ return malloc(sz);
}
static void setup_temp_free(vorb *f, void *p, int sz)
{
- if (f->alloc.alloc_buffer) {
- f->temp_offset += (sz + 3)&~3;
- return;
- }
- free(p);
+ if (f->alloc.alloc_buffer) {
+ f->temp_offset += (sz+7)&~7;
+ return;
+ }
+ free(p);
}
#define CRC32_POLY 0x04c11db7 // from spec
@@ -947,34 +991,34 @@ static void setup_temp_free(vorb *f, void *p, int sz)
static uint32 crc_table[256];
static void crc32_init(void)
{
- int i, j;
- uint32 s;
- for (i = 0; i < 256; i++) {
- for (s = (uint32)i << 24, j = 0; j < 8; ++j)
- s = (s << 1) ^ (s >= (1U << 31) ? CRC32_POLY : 0);
- crc_table[i] = s;
- }
+ int i,j;
+ uint32 s;
+ for(i=0; i < 256; i++) {
+ for (s=(uint32) i << 24, j=0; j < 8; ++j)
+ s = (s << 1) ^ (s >= (1U<<31) ? CRC32_POLY : 0);
+ crc_table[i] = s;
+ }
}
static __forceinline uint32 crc32_update(uint32 crc, uint8 byte)
{
- return (crc << 8) ^ crc_table[byte ^ (crc >> 24)];
+ return (crc << 8) ^ crc_table[byte ^ (crc >> 24)];
}
// used in setup, and for huffman that doesn't go fast path
static unsigned int bit_reverse(unsigned int n)
{
- n = ((n & 0xAAAAAAAA) >> 1) | ((n & 0x55555555) << 1);
- n = ((n & 0xCCCCCCCC) >> 2) | ((n & 0x33333333) << 2);
- n = ((n & 0xF0F0F0F0) >> 4) | ((n & 0x0F0F0F0F) << 4);
- n = ((n & 0xFF00FF00) >> 8) | ((n & 0x00FF00FF) << 8);
- return (n >> 16) | (n << 16);
+ n = ((n & 0xAAAAAAAA) >> 1) | ((n & 0x55555555) << 1);
+ n = ((n & 0xCCCCCCCC) >> 2) | ((n & 0x33333333) << 2);
+ n = ((n & 0xF0F0F0F0) >> 4) | ((n & 0x0F0F0F0F) << 4);
+ n = ((n & 0xFF00FF00) >> 8) | ((n & 0x00FF00FF) << 8);
+ return (n >> 16) | (n << 16);
}
static float square(float x)
{
- return x*x;
+ return x*x;
}
// this is a weird definition of log2() for which log2(1) = 1, log2(2) = 2, log2(4) = 3
@@ -982,24 +1026,24 @@ static float square(float x)
// @OPTIMIZE: called multiple times per-packet with "constants"; move to setup
static int ilog(int32 n)
{
- static signed char log2_4[16] = { 0,1,2,2,3,3,3,3,4,4,4,4,4,4,4,4 };
+ static signed char log2_4[16] = { 0,1,2,2,3,3,3,3,4,4,4,4,4,4,4,4 };
- if (n < 0) return 0; // signed n returns 0
+ if (n < 0) return 0; // signed n returns 0
- // 2 compares if n < 16, 3 compares otherwise (4 if signed or n > 1<<29)
- if (n < (1 << 14))
- if (n < (1 << 4)) return 0 + log2_4[n];
- else if (n < (1 << 9)) return 5 + log2_4[n >> 5];
- else return 10 + log2_4[n >> 10];
- else if (n < (1 << 24))
- if (n < (1 << 19)) return 15 + log2_4[n >> 15];
- else return 20 + log2_4[n >> 20];
- else if (n < (1 << 29)) return 25 + log2_4[n >> 25];
- else return 30 + log2_4[n >> 30];
+ // 2 compares if n < 16, 3 compares otherwise (4 if signed or n > 1<<29)
+ if (n < (1 << 14))
+ if (n < (1 << 4)) return 0 + log2_4[n ];
+ else if (n < (1 << 9)) return 5 + log2_4[n >> 5];
+ else return 10 + log2_4[n >> 10];
+ else if (n < (1 << 24))
+ if (n < (1 << 19)) return 15 + log2_4[n >> 15];
+ else return 20 + log2_4[n >> 20];
+ else if (n < (1 << 29)) return 25 + log2_4[n >> 25];
+ else return 30 + log2_4[n >> 30];
}
#ifndef M_PI
-#define M_PI 3.14159265358979323846264f // from CRC
+ #define M_PI 3.14159265358979323846264f // from CRC
#endif
// code length assigned to a value with no huffman encoding
@@ -1012,12 +1056,12 @@ static int ilog(int32 n)
static float float32_unpack(uint32 x)
{
- // from the specification
- uint32 mantissa = x & 0x1fffff;
- uint32 sign = x & 0x80000000;
- uint32 exp = (x & 0x7fe00000) >> 21;
- double res = sign ? -(double)mantissa : (double)mantissa;
- return (float)ldexp((float)res, exp - 788);
+ // from the specification
+ uint32 mantissa = x & 0x1fffff;
+ uint32 sign = x & 0x80000000;
+ uint32 exp = (x & 0x7fe00000) >> 21;
+ double res = sign ? -(double)mantissa : (double)mantissa;
+ return (float) ldexp((float)res, (int)exp-788);
}
@@ -1030,84 +1074,83 @@ static float float32_unpack(uint32 x)
// order do not map to huffman codes "in order".
static void add_entry(Codebook *c, uint32 huff_code, int symbol, int count, int len, uint32 *values)
{
- if (!c->sparse) {
- c->codewords[symbol] = huff_code;
- }
- else {
- c->codewords[count] = huff_code;
- c->codeword_lengths[count] = len;
- values[count] = symbol;
- }
+ if (!c->sparse) {
+ c->codewords [symbol] = huff_code;
+ } else {
+ c->codewords [count] = huff_code;
+ c->codeword_lengths[count] = len;
+ values [count] = symbol;
+ }
}
static int compute_codewords(Codebook *c, uint8 *len, int n, uint32 *values)
{
- int i, k, m = 0;
- uint32 available[32];
-
- memset(available, 0, sizeof(available));
- // find the first entry
- for (k = 0; k < n; ++k) if (len[k] < NO_CODE) break;
- if (k == n) { assert(c->sorted_entries == 0); return TRUE; }
- // add to the list
- add_entry(c, 0, k, m++, len[k], values);
- // add all available leaves
- for (i = 1; i <= len[k]; ++i)
- available[i] = 1U << (32 - i);
- // note that the above code treats the first case specially,
- // but it's really the same as the following code, so they
- // could probably be combined (except the initial code is 0,
- // and I use 0 in available[] to mean 'empty')
- for (i = k + 1; i < n; ++i) {
- uint32 res;
- int z = len[i], y;
- if (z == NO_CODE) continue;
- // find lowest available leaf (should always be earliest,
- // which is what the specification calls for)
- // note that this property, and the fact we can never have
- // more than one free leaf at a given level, isn't totally
- // trivial to prove, but it seems true and the assert never
- // fires, so!
- while (z > 0 && !available[z]) --z;
- if (z == 0) { return FALSE; }
- res = available[z];
- assert(z >= 0 && z < 32);
- available[z] = 0;
- add_entry(c, bit_reverse(res), i, m++, len[i], values);
- // propogate availability up the tree
- if (z != len[i]) {
- assert(len[i] >= 0 && len[i] < 32);
- for (y = len[i]; y > z; --y) {
- assert(available[y] == 0);
- available[y] = res + (1 << (32 - y));
- }
- }
- }
- return TRUE;
+ int i,k,m=0;
+ uint32 available[32];
+
+ memset(available, 0, sizeof(available));
+ // find the first entry
+ for (k=0; k < n; ++k) if (len[k] < NO_CODE) break;
+ if (k == n) { assert(c->sorted_entries == 0); return TRUE; }
+ assert(len[k] < 32); // no error return required, code reading lens checks this
+ // add to the list
+ add_entry(c, 0, k, m++, len[k], values);
+ // add all available leaves
+ for (i=1; i <= len[k]; ++i)
+ available[i] = 1U << (32-i);
+ // note that the above code treats the first case specially,
+ // but it's really the same as the following code, so they
+ // could probably be combined (except the initial code is 0,
+ // and I use 0 in available[] to mean 'empty')
+ for (i=k+1; i < n; ++i) {
+ uint32 res;
+ int z = len[i], y;
+ if (z == NO_CODE) continue;
+ assert(z < 32); // no error return required, code reading lens checks this
+ // find lowest available leaf (should always be earliest,
+ // which is what the specification calls for)
+ // note that this property, and the fact we can never have
+ // more than one free leaf at a given level, isn't totally
+ // trivial to prove, but it seems true and the assert never
+ // fires, so!
+ while (z > 0 && !available[z]) --z;
+ if (z == 0) { return FALSE; }
+ res = available[z];
+ available[z] = 0;
+ add_entry(c, bit_reverse(res), i, m++, len[i], values);
+ // propagate availability up the tree
+ if (z != len[i]) {
+ for (y=len[i]; y > z; --y) {
+ assert(available[y] == 0);
+ available[y] = res + (1 << (32-y));
+ }
+ }
+ }
+ return TRUE;
}
// accelerated huffman table allows fast O(1) match of all symbols
// of length <= STB_VORBIS_FAST_HUFFMAN_LENGTH
static void compute_accelerated_huffman(Codebook *c)
{
- int i, len;
- for (i = 0; i < FAST_HUFFMAN_TABLE_SIZE; ++i)
- c->fast_huffman[i] = -1;
-
- len = c->sparse ? c->sorted_entries : c->entries;
-#ifdef STB_VORBIS_FAST_HUFFMAN_SHORT
- if (len > 32767) len = 32767; // largest possible value we can encode!
-#endif
- for (i = 0; i < len; ++i) {
- if (c->codeword_lengths[i] <= STB_VORBIS_FAST_HUFFMAN_LENGTH) {
- uint32 z = c->sparse ? bit_reverse(c->sorted_codewords[i]) : c->codewords[i];
- // set table entries for all bit combinations in the higher bits
- while (z < FAST_HUFFMAN_TABLE_SIZE) {
- c->fast_huffman[z] = i;
- z += 1 << c->codeword_lengths[i];
- }
- }
- }
+ int i, len;
+ for (i=0; i < FAST_HUFFMAN_TABLE_SIZE; ++i)
+ c->fast_huffman[i] = -1;
+
+ len = c->sparse ? c->sorted_entries : c->entries;
+ #ifdef STB_VORBIS_FAST_HUFFMAN_SHORT
+ if (len > 32767) len = 32767; // largest possible value we can encode!
+ #endif
+ for (i=0; i < len; ++i) {
+ if (c->codeword_lengths[i] <= STB_VORBIS_FAST_HUFFMAN_LENGTH) {
+ uint32 z = c->sparse ? bit_reverse(c->sorted_codewords[i]) : c->codewords[i];
+ // set table entries for all bit combinations in the higher bits
+ while (z < FAST_HUFFMAN_TABLE_SIZE) {
+ c->fast_huffman[z] = i;
+ z += 1 << c->codeword_lengths[i];
+ }
+ }
+ }
}
#ifdef _MSC_VER
@@ -1118,168 +1161,167 @@ static void compute_accelerated_huffman(Codebook *c)
static int STBV_CDECL uint32_compare(const void *p, const void *q)
{
- uint32 x = *(uint32 *)p;
- uint32 y = *(uint32 *)q;
- return x < y ? -1 : x > y;
+ uint32 x = * (uint32 *) p;
+ uint32 y = * (uint32 *) q;
+ return x < y ? -1 : x > y;
}
static int include_in_sort(Codebook *c, uint8 len)
{
- if (c->sparse) { assert(len != NO_CODE); return TRUE; }
- if (len == NO_CODE) return FALSE;
- if (len > STB_VORBIS_FAST_HUFFMAN_LENGTH) return TRUE;
- return FALSE;
+ if (c->sparse) { assert(len != NO_CODE); return TRUE; }
+ if (len == NO_CODE) return FALSE;
+ if (len > STB_VORBIS_FAST_HUFFMAN_LENGTH) return TRUE;
+ return FALSE;
}
// if the fast table above doesn't work, we want to binary
// search them... need to reverse the bits
static void compute_sorted_huffman(Codebook *c, uint8 *lengths, uint32 *values)
{
- int i, len;
- // build a list of all the entries
- // OPTIMIZATION: don't include the short ones, since they'll be caught by FAST_HUFFMAN.
- // this is kind of a frivolous optimization--I don't see any performance improvement,
- // but it's like 4 extra lines of code, so.
- if (!c->sparse) {
- int k = 0;
- for (i = 0; i < c->entries; ++i)
- if (include_in_sort(c, lengths[i]))
- c->sorted_codewords[k++] = bit_reverse(c->codewords[i]);
- assert(k == c->sorted_entries);
- }
- else {
- for (i = 0; i < c->sorted_entries; ++i)
- c->sorted_codewords[i] = bit_reverse(c->codewords[i]);
- }
-
- qsort(c->sorted_codewords, c->sorted_entries, sizeof(c->sorted_codewords[0]), uint32_compare);
- c->sorted_codewords[c->sorted_entries] = 0xffffffff;
-
- len = c->sparse ? c->sorted_entries : c->entries;
- // now we need to indicate how they correspond; we could either
- // #1: sort a different data structure that says who they correspond to
- // #2: for each sorted entry, search the original list to find who corresponds
- // #3: for each original entry, find the sorted entry
- // #1 requires extra storage, #2 is slow, #3 can use binary search!
- for (i = 0; i < len; ++i) {
- int huff_len = c->sparse ? lengths[values[i]] : lengths[i];
- if (include_in_sort(c, huff_len)) {
- uint32 code = bit_reverse(c->codewords[i]);
- int x = 0, n = c->sorted_entries;
- while (n > 1) {
- // invariant: sc[x] <= code < sc[x+n]
- int m = x + (n >> 1);
- if (c->sorted_codewords[m] <= code) {
- x = m;
- n -= (n >> 1);
- }
- else {
- n >>= 1;
- }
- }
- assert(c->sorted_codewords[x] == code);
- if (c->sparse) {
- c->sorted_values[x] = values[i];
- c->codeword_lengths[x] = huff_len;
- }
- else {
- c->sorted_values[x] = i;
+ int i, len;
+ // build a list of all the entries
+ // OPTIMIZATION: don't include the short ones, since they'll be caught by FAST_HUFFMAN.
+ // this is kind of a frivolous optimization--I don't see any performance improvement,
+ // but it's like 4 extra lines of code, so.
+ if (!c->sparse) {
+ int k = 0;
+ for (i=0; i < c->entries; ++i)
+ if (include_in_sort(c, lengths[i]))
+ c->sorted_codewords[k++] = bit_reverse(c->codewords[i]);
+ assert(k == c->sorted_entries);
+ } else {
+ for (i=0; i < c->sorted_entries; ++i)
+ c->sorted_codewords[i] = bit_reverse(c->codewords[i]);
+ }
+
+ qsort(c->sorted_codewords, c->sorted_entries, sizeof(c->sorted_codewords[0]), uint32_compare);
+ c->sorted_codewords[c->sorted_entries] = 0xffffffff;
+
+ len = c->sparse ? c->sorted_entries : c->entries;
+ // now we need to indicate how they correspond; we could either
+ // #1: sort a different data structure that says who they correspond to
+ // #2: for each sorted entry, search the original list to find who corresponds
+ // #3: for each original entry, find the sorted entry
+ // #1 requires extra storage, #2 is slow, #3 can use binary search!
+ for (i=0; i < len; ++i) {
+ int huff_len = c->sparse ? lengths[values[i]] : lengths[i];
+ if (include_in_sort(c,huff_len)) {
+ uint32 code = bit_reverse(c->codewords[i]);
+ int x=0, n=c->sorted_entries;
+ while (n > 1) {
+ // invariant: sc[x] <= code < sc[x+n]
+ int m = x + (n >> 1);
+ if (c->sorted_codewords[m] <= code) {
+ x = m;
+ n -= (n>>1);
+ } else {
+ n >>= 1;
}
- }
- }
+ }
+ assert(c->sorted_codewords[x] == code);
+ if (c->sparse) {
+ c->sorted_values[x] = values[i];
+ c->codeword_lengths[x] = huff_len;
+ } else {
+ c->sorted_values[x] = i;
+ }
+ }
+ }
}
// only run while parsing the header (3 times)
static int vorbis_validate(uint8 *data)
{
- static uint8 vorbis[6] = { 'v', 'o', 'r', 'b', 'i', 's' };
- return memcmp(data, vorbis, 6) == 0;
+ static uint8 vorbis[6] = { 'v', 'o', 'r', 'b', 'i', 's' };
+ return memcmp(data, vorbis, 6) == 0;
}
// called from setup only, once per code book
// (formula implied by specification)
static int lookup1_values(int entries, int dim)
{
- int r = (int)floor(exp((float)log((float)entries) / dim));
- if ((int)floor(pow((float)r + 1, dim)) <= entries) // (int) cast for MinGW warning;
- ++r; // floor() to avoid _ftol() when non-CRT
- assert(pow((float)r + 1, dim) > entries);
- assert((int)floor(pow((float)r, dim)) <= entries); // (int),floor() as above
- return r;
+ int r = (int) floor(exp((float) log((float) entries) / dim));
+ if ((int) floor(pow((float) r+1, dim)) <= entries) // (int) cast for MinGW warning;
+ ++r; // floor() to avoid _ftol() when non-CRT
+ if (pow((float) r+1, dim) <= entries)
+ return -1;
+ if ((int) floor(pow((float) r, dim)) > entries)
+ return -1;
+ return r;
}
// called twice per file
static void compute_twiddle_factors(int n, float *A, float *B, float *C)
{
- int n4 = n >> 2, n8 = n >> 3;
- int k, k2;
-
- for (k = k2 = 0; k < n4; ++k, k2 += 2) {
- A[k2] = (float)cos(4 * k*M_PI / n);
- A[k2 + 1] = (float)-sin(4 * k*M_PI / n);
- B[k2] = (float)cos((k2 + 1)*M_PI / n / 2) * 0.5f;
- B[k2 + 1] = (float)sin((k2 + 1)*M_PI / n / 2) * 0.5f;
- }
- for (k = k2 = 0; k < n8; ++k, k2 += 2) {
- C[k2] = (float)cos(2 * (k2 + 1)*M_PI / n);
- C[k2 + 1] = (float)-sin(2 * (k2 + 1)*M_PI / n);
- }
+ int n4 = n >> 2, n8 = n >> 3;
+ int k,k2;
+
+ for (k=k2=0; k < n4; ++k,k2+=2) {
+ A[k2 ] = (float) cos(4*k*M_PI/n);
+ A[k2+1] = (float) -sin(4*k*M_PI/n);
+ B[k2 ] = (float) cos((k2+1)*M_PI/n/2) * 0.5f;
+ B[k2+1] = (float) sin((k2+1)*M_PI/n/2) * 0.5f;
+ }
+ for (k=k2=0; k < n8; ++k,k2+=2) {
+ C[k2 ] = (float) cos(2*(k2+1)*M_PI/n);
+ C[k2+1] = (float) -sin(2*(k2+1)*M_PI/n);
+ }
}
static void compute_window(int n, float *window)
{
- int n2 = n >> 1, i;
- for (i = 0; i < n2; ++i)
- window[i] = (float)sin(0.5 * M_PI * square((float)sin((i - 0 + 0.5) / n2 * 0.5 * M_PI)));
+ int n2 = n >> 1, i;
+ for (i=0; i < n2; ++i)
+ window[i] = (float) sin(0.5 * M_PI * square((float) sin((i - 0 + 0.5) / n2 * 0.5 * M_PI)));
}
static void compute_bitreverse(int n, uint16 *rev)
{
- int ld = ilog(n) - 1; // ilog is off-by-one from normal definitions
- int i, n8 = n >> 3;
- for (i = 0; i < n8; ++i)
- rev[i] = (bit_reverse(i) >> (32 - ld + 3)) << 2;
+ int ld = ilog(n) - 1; // ilog is off-by-one from normal definitions
+ int i, n8 = n >> 3;
+ for (i=0; i < n8; ++i)
+ rev[i] = (bit_reverse(i) >> (32-ld+3)) << 2;
}
static int init_blocksize(vorb *f, int b, int n)
{
- int n2 = n >> 1, n4 = n >> 2, n8 = n >> 3;
- f->A[b] = (float *)setup_malloc(f, sizeof(float) * n2);
- f->B[b] = (float *)setup_malloc(f, sizeof(float) * n2);
- f->C[b] = (float *)setup_malloc(f, sizeof(float) * n4);
- if (!f->A[b] || !f->B[b] || !f->C[b]) return error(f, VORBIS_outofmem);
- compute_twiddle_factors(n, f->A[b], f->B[b], f->C[b]);
- f->window[b] = (float *)setup_malloc(f, sizeof(float) * n2);
- if (!f->window[b]) return error(f, VORBIS_outofmem);
- compute_window(n, f->window[b]);
- f->bit_reverse[b] = (uint16 *)setup_malloc(f, sizeof(uint16) * n8);
- if (!f->bit_reverse[b]) return error(f, VORBIS_outofmem);
- compute_bitreverse(n, f->bit_reverse[b]);
- return TRUE;
+ int n2 = n >> 1, n4 = n >> 2, n8 = n >> 3;
+ f->A[b] = (float *) setup_malloc(f, sizeof(float) * n2);
+ f->B[b] = (float *) setup_malloc(f, sizeof(float) * n2);
+ f->C[b] = (float *) setup_malloc(f, sizeof(float) * n4);
+ if (!f->A[b] || !f->B[b] || !f->C[b]) return error(f, VORBIS_outofmem);
+ compute_twiddle_factors(n, f->A[b], f->B[b], f->C[b]);
+ f->window[b] = (float *) setup_malloc(f, sizeof(float) * n2);
+ if (!f->window[b]) return error(f, VORBIS_outofmem);
+ compute_window(n, f->window[b]);
+ f->bit_reverse[b] = (uint16 *) setup_malloc(f, sizeof(uint16) * n8);
+ if (!f->bit_reverse[b]) return error(f, VORBIS_outofmem);
+ compute_bitreverse(n, f->bit_reverse[b]);
+ return TRUE;
}
static void neighbors(uint16 *x, int n, int *plow, int *phigh)
{
- int low = -1;
- int high = 65536;
- int i;
- for (i = 0; i < n; ++i) {
- if (x[i] > low && x[i] < x[n]) { *plow = i; low = x[i]; }
- if (x[i] < high && x[i] > x[n]) { *phigh = i; high = x[i]; }
- }
+ int low = -1;
+ int high = 65536;
+ int i;
+ for (i=0; i < n; ++i) {
+ if (x[i] > low && x[i] < x[n]) { *plow = i; low = x[i]; }
+ if (x[i] < high && x[i] > x[n]) { *phigh = i; high = x[i]; }
+ }
}
// this has been repurposed so y is now the original index instead of y
typedef struct
{
- uint16 x, id;
+ uint16 x,id;
} stbv__floor_ordering;
static int STBV_CDECL point_compare(const void *p, const void *q)
{
- stbv__floor_ordering *a = (stbv__floor_ordering *)p;
- stbv__floor_ordering *b = (stbv__floor_ordering *)q;
- return a->x < b->x ? -1 : a->x > b->x;
+ stbv__floor_ordering *a = (stbv__floor_ordering *) p;
+ stbv__floor_ordering *b = (stbv__floor_ordering *) q;
+ return a->x < b->x ? -1 : a->x > b->x;
}
//
@@ -1287,102 +1329,100 @@ static int STBV_CDECL point_compare(const void *p, const void *q)
#if defined(STB_VORBIS_NO_STDIO)
-#define USE_MEMORY(z) TRUE
+ #define USE_MEMORY(z) TRUE
#else
-#define USE_MEMORY(z) ((z)->stream)
+ #define USE_MEMORY(z) ((z)->stream)
#endif
static uint8 get8(vorb *z)
{
- if (USE_MEMORY(z)) {
- if (z->stream >= z->stream_end) { z->eof = TRUE; return 0; }
- return *z->stream++;
- }
+ if (USE_MEMORY(z)) {
+ if (z->stream >= z->stream_end) { z->eof = TRUE; return 0; }
+ return *z->stream++;
+ }
-#ifndef STB_VORBIS_NO_STDIO
- {
- int c = fgetc(z->f);
- if (c == EOF) { z->eof = TRUE; return 0; }
- return c;
- }
-#endif
+ #ifndef STB_VORBIS_NO_STDIO
+ {
+ int c = fgetc(z->f);
+ if (c == EOF) { z->eof = TRUE; return 0; }
+ return c;
+ }
+ #endif
}
static uint32 get32(vorb *f)
{
- uint32 x;
- x = get8(f);
- x += get8(f) << 8;
- x += get8(f) << 16;
- x += (uint32)get8(f) << 24;
- return x;
+ uint32 x;
+ x = get8(f);
+ x += get8(f) << 8;
+ x += get8(f) << 16;
+ x += (uint32) get8(f) << 24;
+ return x;
}
static int getn(vorb *z, uint8 *data, int n)
{
- if (USE_MEMORY(z)) {
- if (z->stream + n > z->stream_end) { z->eof = 1; return 0; }
- memcpy(data, z->stream, n);
- z->stream += n;
- return 1;
- }
-
-#ifndef STB_VORBIS_NO_STDIO
- if (fread(data, n, 1, z->f) == 1)
- return 1;
- else {
- z->eof = 1;
- return 0;
- }
-#endif
+ if (USE_MEMORY(z)) {
+ if (z->stream+n > z->stream_end) { z->eof = 1; return 0; }
+ memcpy(data, z->stream, n);
+ z->stream += n;
+ return 1;
+ }
+
+ #ifndef STB_VORBIS_NO_STDIO
+ if (fread(data, n, 1, z->f) == 1)
+ return 1;
+ else {
+ z->eof = 1;
+ return 0;
+ }
+ #endif
}
static void skip(vorb *z, int n)
{
- if (USE_MEMORY(z)) {
- z->stream += n;
- if (z->stream >= z->stream_end) z->eof = 1;
- return;
- }
-#ifndef STB_VORBIS_NO_STDIO
- {
- long x = ftell(z->f);
- fseek(z->f, x + n, SEEK_SET);
- }
-#endif
+ if (USE_MEMORY(z)) {
+ z->stream += n;
+ if (z->stream >= z->stream_end) z->eof = 1;
+ return;
+ }
+ #ifndef STB_VORBIS_NO_STDIO
+ {
+ long x = ftell(z->f);
+ fseek(z->f, x+n, SEEK_SET);
+ }
+ #endif
}
static int set_file_offset(stb_vorbis *f, unsigned int loc)
{
-#ifndef STB_VORBIS_NO_PUSHDATA_API
- if (f->push_mode) return 0;
-#endif
- f->eof = 0;
- if (USE_MEMORY(f)) {
- if (f->stream_start + loc >= f->stream_end || f->stream_start + loc < f->stream_start) {
- f->stream = f->stream_end;
- f->eof = 1;
- return 0;
- }
- else {
- f->stream = f->stream_start + loc;
- return 1;
- }
- }
-#ifndef STB_VORBIS_NO_STDIO
- if (loc + f->f_start < loc || loc >= 0x80000000) {
- loc = 0x7fffffff;
- f->eof = 1;
- }
- else {
- loc += f->f_start;
- }
- if (!fseek(f->f, loc, SEEK_SET))
- return 1;
- f->eof = 1;
- fseek(f->f, f->f_start, SEEK_END);
- return 0;
-#endif
+ #ifndef STB_VORBIS_NO_PUSHDATA_API
+ if (f->push_mode) return 0;
+ #endif
+ f->eof = 0;
+ if (USE_MEMORY(f)) {
+ if (f->stream_start + loc >= f->stream_end || f->stream_start + loc < f->stream_start) {
+ f->stream = f->stream_end;
+ f->eof = 1;
+ return 0;
+ } else {
+ f->stream = f->stream_start + loc;
+ return 1;
+ }
+ }
+ #ifndef STB_VORBIS_NO_STDIO
+ if (loc + f->f_start < loc || loc >= 0x80000000) {
+ loc = 0x7fffffff;
+ f->eof = 1;
+ } else {
+ loc += f->f_start;
+ }
+ if (!fseek(f->f, loc, SEEK_SET))
+ return 1;
+ f->eof = 1;
+ fseek(f->f, f->f_start, SEEK_END);
+ return 0;
+ #endif
}
@@ -1390,11 +1430,11 @@ static uint8 ogg_page_header[4] = { 0x4f, 0x67, 0x67, 0x53 };
static int capture_pattern(vorb *f)
{
- if (0x4f != get8(f)) return FALSE;
- if (0x67 != get8(f)) return FALSE;
- if (0x67 != get8(f)) return FALSE;
- if (0x53 != get8(f)) return FALSE;
- return TRUE;
+ if (0x4f != get8(f)) return FALSE;
+ if (0x67 != get8(f)) return FALSE;
+ if (0x67 != get8(f)) return FALSE;
+ if (0x53 != get8(f)) return FALSE;
+ return TRUE;
}
#define PAGEFLAG_continued_packet 1
@@ -1403,118 +1443,118 @@ static int capture_pattern(vorb *f)
static int start_page_no_capturepattern(vorb *f)
{
- uint32 loc0, loc1, n;
- // stream structure version
- if (0 != get8(f)) return error(f, VORBIS_invalid_stream_structure_version);
- // header flag
- f->page_flag = get8(f);
- // absolute granule position
- loc0 = get32(f);
- loc1 = get32(f);
- // @TODO: validate loc0,loc1 as valid positions?
- // stream serial number -- vorbis doesn't interleave, so discard
- get32(f);
- //if (f->serial != get32(f)) return error(f, VORBIS_incorrect_stream_serial_number);
- // page sequence number
- n = get32(f);
- f->last_page = n;
- // CRC32
- get32(f);
- // page_segments
- f->segment_count = get8(f);
- if (!getn(f, f->segments, f->segment_count))
- return error(f, VORBIS_unexpected_eof);
- // assume we _don't_ know any the sample position of any segments
- f->end_seg_with_known_loc = -2;
- if (loc0 != ~0U || loc1 != ~0U) {
- int i;
- // determine which packet is the last one that will complete
- for (i = f->segment_count - 1; i >= 0; --i)
- if (f->segments[i] < 255)
- break;
- // 'i' is now the index of the _last_ segment of a packet that ends
- if (i >= 0) {
- f->end_seg_with_known_loc = i;
- f->known_loc_for_packet = loc0;
- }
- }
- if (f->first_decode) {
- int i, len;
- ProbedPage p;
- len = 0;
- for (i = 0; i < f->segment_count; ++i)
- len += f->segments[i];
- len += 27 + f->segment_count;
- p.page_start = f->first_audio_page_offset;
- p.page_end = p.page_start + len;
- p.last_decoded_sample = loc0;
- f->p_first = p;
- }
- f->next_seg = 0;
- return TRUE;
+ uint32 loc0,loc1,n;
+ if (f->first_decode && !IS_PUSH_MODE(f)) {
+ f->p_first.page_start = stb_vorbis_get_file_offset(f) - 4;
+ }
+ // stream structure version
+ if (0 != get8(f)) return error(f, VORBIS_invalid_stream_structure_version);
+ // header flag
+ f->page_flag = get8(f);
+ // absolute granule position
+ loc0 = get32(f);
+ loc1 = get32(f);
+ // @TODO: validate loc0,loc1 as valid positions?
+ // stream serial number -- vorbis doesn't interleave, so discard
+ get32(f);
+ //if (f->serial != get32(f)) return error(f, VORBIS_incorrect_stream_serial_number);
+ // page sequence number
+ n = get32(f);
+ f->last_page = n;
+ // CRC32
+ get32(f);
+ // page_segments
+ f->segment_count = get8(f);
+ if (!getn(f, f->segments, f->segment_count))
+ return error(f, VORBIS_unexpected_eof);
+ // assume we _don't_ know any the sample position of any segments
+ f->end_seg_with_known_loc = -2;
+ if (loc0 != ~0U || loc1 != ~0U) {
+ int i;
+ // determine which packet is the last one that will complete
+ for (i=f->segment_count-1; i >= 0; --i)
+ if (f->segments[i] < 255)
+ break;
+ // 'i' is now the index of the _last_ segment of a packet that ends
+ if (i >= 0) {
+ f->end_seg_with_known_loc = i;
+ f->known_loc_for_packet = loc0;
+ }
+ }
+ if (f->first_decode) {
+ int i,len;
+ len = 0;
+ for (i=0; i < f->segment_count; ++i)
+ len += f->segments[i];
+ len += 27 + f->segment_count;
+ f->p_first.page_end = f->p_first.page_start + len;
+ f->p_first.last_decoded_sample = loc0;
+ }
+ f->next_seg = 0;
+ return TRUE;
}
static int start_page(vorb *f)
{
- if (!capture_pattern(f)) return error(f, VORBIS_missing_capture_pattern);
- return start_page_no_capturepattern(f);
+ if (!capture_pattern(f)) return error(f, VORBIS_missing_capture_pattern);
+ return start_page_no_capturepattern(f);
}
static int start_packet(vorb *f)
{
- while (f->next_seg == -1) {
- if (!start_page(f)) return FALSE;
- if (f->page_flag & PAGEFLAG_continued_packet)
- return error(f, VORBIS_continued_packet_flag_invalid);
- }
- f->last_seg = FALSE;
- f->valid_bits = 0;
- f->packet_bytes = 0;
- f->bytes_in_seg = 0;
- // f->next_seg is now valid
- return TRUE;
+ while (f->next_seg == -1) {
+ if (!start_page(f)) return FALSE;
+ if (f->page_flag & PAGEFLAG_continued_packet)
+ return error(f, VORBIS_continued_packet_flag_invalid);
+ }
+ f->last_seg = FALSE;
+ f->valid_bits = 0;
+ f->packet_bytes = 0;
+ f->bytes_in_seg = 0;
+ // f->next_seg is now valid
+ return TRUE;
}
static int maybe_start_packet(vorb *f)
{
- if (f->next_seg == -1) {
- int x = get8(f);
- if (f->eof) return FALSE; // EOF at page boundary is not an error!
- if (0x4f != x) return error(f, VORBIS_missing_capture_pattern);
- if (0x67 != get8(f)) return error(f, VORBIS_missing_capture_pattern);
- if (0x67 != get8(f)) return error(f, VORBIS_missing_capture_pattern);
- if (0x53 != get8(f)) return error(f, VORBIS_missing_capture_pattern);
- if (!start_page_no_capturepattern(f)) return FALSE;
- if (f->page_flag & PAGEFLAG_continued_packet) {
- // set up enough state that we can read this packet if we want,
- // e.g. during recovery
- f->last_seg = FALSE;
- f->bytes_in_seg = 0;
- return error(f, VORBIS_continued_packet_flag_invalid);
- }
- }
- return start_packet(f);
+ if (f->next_seg == -1) {
+ int x = get8(f);
+ if (f->eof) return FALSE; // EOF at page boundary is not an error!
+ if (0x4f != x ) return error(f, VORBIS_missing_capture_pattern);
+ if (0x67 != get8(f)) return error(f, VORBIS_missing_capture_pattern);
+ if (0x67 != get8(f)) return error(f, VORBIS_missing_capture_pattern);
+ if (0x53 != get8(f)) return error(f, VORBIS_missing_capture_pattern);
+ if (!start_page_no_capturepattern(f)) return FALSE;
+ if (f->page_flag & PAGEFLAG_continued_packet) {
+ // set up enough state that we can read this packet if we want,
+ // e.g. during recovery
+ f->last_seg = FALSE;
+ f->bytes_in_seg = 0;
+ return error(f, VORBIS_continued_packet_flag_invalid);
+ }
+ }
+ return start_packet(f);
}
static int next_segment(vorb *f)
{
- int len;
- if (f->last_seg) return 0;
- if (f->next_seg == -1) {
- f->last_seg_which = f->segment_count - 1; // in case start_page fails
- if (!start_page(f)) { f->last_seg = 1; return 0; }
- if (!(f->page_flag & PAGEFLAG_continued_packet)) return error(f, VORBIS_continued_packet_flag_invalid);
- }
- len = f->segments[f->next_seg++];
- if (len < 255) {
- f->last_seg = TRUE;
- f->last_seg_which = f->next_seg - 1;
- }
- if (f->next_seg >= f->segment_count)
- f->next_seg = -1;
- assert(f->bytes_in_seg == 0);
- f->bytes_in_seg = len;
- return len;
+ int len;
+ if (f->last_seg) return 0;
+ if (f->next_seg == -1) {
+ f->last_seg_which = f->segment_count-1; // in case start_page fails
+ if (!start_page(f)) { f->last_seg = 1; return 0; }
+ if (!(f->page_flag & PAGEFLAG_continued_packet)) return error(f, VORBIS_continued_packet_flag_invalid);
+ }
+ len = f->segments[f->next_seg++];
+ if (len < 255) {
+ f->last_seg = TRUE;
+ f->last_seg_which = f->next_seg-1;
+ }
+ if (f->next_seg >= f->segment_count)
+ f->next_seg = -1;
+ assert(f->bytes_in_seg == 0);
+ f->bytes_in_seg = len;
+ return len;
}
#define EOP (-1)
@@ -1522,58 +1562,69 @@ static int next_segment(vorb *f)
static int get8_packet_raw(vorb *f)
{
- if (!f->bytes_in_seg) { // CLANG!
- if (f->last_seg) return EOP;
- else if (!next_segment(f)) return EOP;
- }
- assert(f->bytes_in_seg > 0);
- --f->bytes_in_seg;
- ++f->packet_bytes;
- return get8(f);
+ if (!f->bytes_in_seg) { // CLANG!
+ if (f->last_seg) return EOP;
+ else if (!next_segment(f)) return EOP;
+ }
+ assert(f->bytes_in_seg > 0);
+ --f->bytes_in_seg;
+ ++f->packet_bytes;
+ return get8(f);
}
static int get8_packet(vorb *f)
{
- int x = get8_packet_raw(f);
- f->valid_bits = 0;
- return x;
+ int x = get8_packet_raw(f);
+ f->valid_bits = 0;
+ return x;
+}
+
+static int get32_packet(vorb *f)
+{
+ uint32 x;
+ x = get8_packet(f);
+ x += get8_packet(f) << 8;
+ x += get8_packet(f) << 16;
+ x += (uint32) get8_packet(f) << 24;
+ return x;
}
static void flush_packet(vorb *f)
{
- while (get8_packet_raw(f) != EOP);
+ while (get8_packet_raw(f) != EOP);
}
// @OPTIMIZE: this is the secondary bit decoder, so it's probably not as important
// as the huffman decoder?
static uint32 get_bits(vorb *f, int n)
{
- uint32 z;
-
- if (f->valid_bits < 0) return 0;
- if (f->valid_bits < n) {
- if (n > 24) {
- // the accumulator technique below would not work correctly in this case
- z = get_bits(f, 24);
- z += get_bits(f, n - 24) << 24;
- return z;
- }
- if (f->valid_bits == 0) f->acc = 0;
- while (f->valid_bits < n) {
- int z = get8_packet_raw(f);
- if (z == EOP) {
- f->valid_bits = INVALID_BITS;
- return 0;
- }
- f->acc += z << f->valid_bits;
- f->valid_bits += 8;
- }
- }
- if (f->valid_bits < 0) return 0;
- z = f->acc & ((1 << n) - 1);
- f->acc >>= n;
- f->valid_bits -= n;
- return z;
+ uint32 z;
+
+ if (f->valid_bits < 0) return 0;
+ if (f->valid_bits < n) {
+ if (n > 24) {
+ // the accumulator technique below would not work correctly in this case
+ z = get_bits(f, 24);
+ z += get_bits(f, n-24) << 24;
+ return z;
+ }
+ if (f->valid_bits == 0) f->acc = 0;
+ while (f->valid_bits < n) {
+ int z = get8_packet_raw(f);
+ if (z == EOP) {
+ f->valid_bits = INVALID_BITS;
+ return 0;
+ }
+ f->acc += z << f->valid_bits;
+ f->valid_bits += 8;
+ }
+ }
+
+ assert(f->valid_bits >= n);
+ z = f->acc & ((1 << n)-1);
+ f->acc >>= n;
+ f->valid_bits -= n;
+ return z;
}
// @OPTIMIZE: primary accumulator for huffman
@@ -1582,84 +1633,83 @@ static uint32 get_bits(vorb *f, int n)
// e.g. cache them locally and decode locally
static __forceinline void prep_huffman(vorb *f)
{
- if (f->valid_bits <= 24) {
- if (f->valid_bits == 0) f->acc = 0;
- do {
- int z;
- if (f->last_seg && !f->bytes_in_seg) return;
- z = get8_packet_raw(f);
- if (z == EOP) return;
- f->acc += (unsigned)z << f->valid_bits;
- f->valid_bits += 8;
- } while (f->valid_bits <= 24);
- }
+ if (f->valid_bits <= 24) {
+ if (f->valid_bits == 0) f->acc = 0;
+ do {
+ int z;
+ if (f->last_seg && !f->bytes_in_seg) return;
+ z = get8_packet_raw(f);
+ if (z == EOP) return;
+ f->acc += (unsigned) z << f->valid_bits;
+ f->valid_bits += 8;
+ } while (f->valid_bits <= 24);
+ }
}
enum
{
- VORBIS_packet_id = 1,
- VORBIS_packet_comment = 3,
- VORBIS_packet_setup = 5
+ VORBIS_packet_id = 1,
+ VORBIS_packet_comment = 3,
+ VORBIS_packet_setup = 5
};
static int codebook_decode_scalar_raw(vorb *f, Codebook *c)
{
- int i;
- prep_huffman(f);
-
- if (c->codewords == NULL && c->sorted_codewords == NULL)
- return -1;
+ int i;
+ prep_huffman(f);
+
+ if (c->codewords == NULL && c->sorted_codewords == NULL)
+ return -1;
+
+ // cases to use binary search: sorted_codewords && !c->codewords
+ // sorted_codewords && c->entries > 8
+ if (c->entries > 8 ? c->sorted_codewords!=NULL : !c->codewords) {
+ // binary search
+ uint32 code = bit_reverse(f->acc);
+ int x=0, n=c->sorted_entries, len;
+
+ while (n > 1) {
+ // invariant: sc[x] <= code < sc[x+n]
+ int m = x + (n >> 1);
+ if (c->sorted_codewords[m] <= code) {
+ x = m;
+ n -= (n>>1);
+ } else {
+ n >>= 1;
+ }
+ }
+ // x is now the sorted index
+ if (!c->sparse) x = c->sorted_values[x];
+ // x is now sorted index if sparse, or symbol otherwise
+ len = c->codeword_lengths[x];
+ if (f->valid_bits >= len) {
+ f->acc >>= len;
+ f->valid_bits -= len;
+ return x;
+ }
+
+ f->valid_bits = 0;
+ return -1;
+ }
- // cases to use binary search: sorted_codewords && !c->codewords
- // sorted_codewords && c->entries > 8
- if (c->entries > 8 ? c->sorted_codewords != NULL : !c->codewords) {
- // binary search
- uint32 code = bit_reverse(f->acc);
- int x = 0, n = c->sorted_entries, len;
+ // if small, linear search
+ assert(!c->sparse);
+ for (i=0; i < c->entries; ++i) {
+ if (c->codeword_lengths[i] == NO_CODE) continue;
+ if (c->codewords[i] == (f->acc & ((1 << c->codeword_lengths[i])-1))) {
+ if (f->valid_bits >= c->codeword_lengths[i]) {
+ f->acc >>= c->codeword_lengths[i];
+ f->valid_bits -= c->codeword_lengths[i];
+ return i;
+ }
+ f->valid_bits = 0;
+ return -1;
+ }
+ }
- while (n > 1) {
- // invariant: sc[x] <= code < sc[x+n]
- int m = x + (n >> 1);
- if (c->sorted_codewords[m] <= code) {
- x = m;
- n -= (n >> 1);
- }
- else {
- n >>= 1;
- }
- }
- // x is now the sorted index
- if (!c->sparse) x = c->sorted_values[x];
- // x is now sorted index if sparse, or symbol otherwise
- len = c->codeword_lengths[x];
- if (f->valid_bits >= len) {
- f->acc >>= len;
- f->valid_bits -= len;
- return x;
- }
-
- f->valid_bits = 0;
- return -1;
- }
-
- // if small, linear search
- assert(!c->sparse);
- for (i = 0; i < c->entries; ++i) {
- if (c->codeword_lengths[i] == NO_CODE) continue;
- if (c->codewords[i] == (f->acc & ((1 << c->codeword_lengths[i]) - 1))) {
- if (f->valid_bits >= c->codeword_lengths[i]) {
- f->acc >>= c->codeword_lengths[i];
- f->valid_bits -= c->codeword_lengths[i];
- return i;
- }
- f->valid_bits = 0;
- return -1;
- }
- }
-
- error(f, VORBIS_invalid_stream);
- f->valid_bits = 0;
- return -1;
+ error(f, VORBIS_invalid_stream);
+ f->valid_bits = 0;
+ return -1;
}
#ifndef STB_VORBIS_NO_INLINE_DECODE
@@ -1682,19 +1732,19 @@ static int codebook_decode_scalar_raw(vorb *f, Codebook *c)
static int codebook_decode_scalar(vorb *f, Codebook *c)
{
- int i;
- if (f->valid_bits < STB_VORBIS_FAST_HUFFMAN_LENGTH)
- prep_huffman(f);
- // fast huffman table lookup
- i = f->acc & FAST_HUFFMAN_TABLE_MASK;
- i = c->fast_huffman[i];
- if (i >= 0) {
- f->acc >>= c->codeword_lengths[i];
- f->valid_bits -= c->codeword_lengths[i];
- if (f->valid_bits < 0) { f->valid_bits = 0; return -1; }
- return i;
- }
- return codebook_decode_scalar_raw(f, c);
+ int i;
+ if (f->valid_bits < STB_VORBIS_FAST_HUFFMAN_LENGTH)
+ prep_huffman(f);
+ // fast huffman table lookup
+ i = f->acc & FAST_HUFFMAN_TABLE_MASK;
+ i = c->fast_huffman[i];
+ if (i >= 0) {
+ f->acc >>= c->codeword_lengths[i];
+ f->valid_bits -= c->codeword_lengths[i];
+ if (f->valid_bits < 0) { f->valid_bits = 0; return -1; }
+ return i;
+ }
+ return codebook_decode_scalar_raw(f,c);
}
#define DECODE_RAW(var,f,c) var = codebook_decode_scalar(f,c);
@@ -1706,9 +1756,9 @@ static int codebook_decode_scalar(vorb *f, Codebook *c)
if (c->sparse) var = c->sorted_values[var];
#ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK
-#define DECODE_VQ(var,f,c) DECODE_RAW(var,f,c)
+ #define DECODE_VQ(var,f,c) DECODE_RAW(var,f,c)
#else
-#define DECODE_VQ(var,f,c) DECODE(var,f,c)
+ #define DECODE_VQ(var,f,c) DECODE(var,f,c)
#endif
@@ -1724,244 +1774,241 @@ static int codebook_decode_scalar(vorb *f, Codebook *c)
static int codebook_decode_start(vorb *f, Codebook *c)
{
- int z = -1;
-
- // type 0 is only legal in a scalar context
- if (c->lookup_type == 0)
- error(f, VORBIS_invalid_stream);
- else {
- DECODE_VQ(z, f, c);
- if (c->sparse) assert(z < c->sorted_entries);
- if (z < 0) { // check for EOP
- if (!f->bytes_in_seg)
- if (f->last_seg)
- return z;
- error(f, VORBIS_invalid_stream);
- }
- }
- return z;
+ int z = -1;
+
+ // type 0 is only legal in a scalar context
+ if (c->lookup_type == 0)
+ error(f, VORBIS_invalid_stream);
+ else {
+ DECODE_VQ(z,f,c);
+ if (c->sparse) assert(z < c->sorted_entries);
+ if (z < 0) { // check for EOP
+ if (!f->bytes_in_seg)
+ if (f->last_seg)
+ return z;
+ error(f, VORBIS_invalid_stream);
+ }
+ }
+ return z;
}
static int codebook_decode(vorb *f, Codebook *c, float *output, int len)
{
- int i, z = codebook_decode_start(f, c);
- if (z < 0) return FALSE;
- if (len > c->dimensions) len = c->dimensions;
+ int i,z = codebook_decode_start(f,c);
+ if (z < 0) return FALSE;
+ if (len > c->dimensions) len = c->dimensions;
#ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK
- if (c->lookup_type == 1) {
- float last = CODEBOOK_ELEMENT_BASE(c);
- int div = 1;
- for (i = 0; i < len; ++i) {
- int off = (z / div) % c->lookup_values;
- float val = CODEBOOK_ELEMENT_FAST(c, off) + last;
- output[i] += val;
- if (c->sequence_p) last = val + c->minimum_value;
- div *= c->lookup_values;
- }
- return TRUE;
- }
+ if (c->lookup_type == 1) {
+ float last = CODEBOOK_ELEMENT_BASE(c);
+ int div = 1;
+ for (i=0; i < len; ++i) {
+ int off = (z / div) % c->lookup_values;
+ float val = CODEBOOK_ELEMENT_FAST(c,off) + last;
+ output[i] += val;
+ if (c->sequence_p) last = val + c->minimum_value;
+ div *= c->lookup_values;
+ }
+ return TRUE;
+ }
#endif
- z *= c->dimensions;
- if (c->sequence_p) {
- float last = CODEBOOK_ELEMENT_BASE(c);
- for (i = 0; i < len; ++i) {
- float val = CODEBOOK_ELEMENT_FAST(c, z + i) + last;
- output[i] += val;
- last = val + c->minimum_value;
- }
- }
- else {
- float last = CODEBOOK_ELEMENT_BASE(c);
- for (i = 0; i < len; ++i) {
- output[i] += CODEBOOK_ELEMENT_FAST(c, z + i) + last;
- }
- }
-
- return TRUE;
+ z *= c->dimensions;
+ if (c->sequence_p) {
+ float last = CODEBOOK_ELEMENT_BASE(c);
+ for (i=0; i < len; ++i) {
+ float val = CODEBOOK_ELEMENT_FAST(c,z+i) + last;
+ output[i] += val;
+ last = val + c->minimum_value;
+ }
+ } else {
+ float last = CODEBOOK_ELEMENT_BASE(c);
+ for (i=0; i < len; ++i) {
+ output[i] += CODEBOOK_ELEMENT_FAST(c,z+i) + last;
+ }
+ }
+
+ return TRUE;
}
static int codebook_decode_step(vorb *f, Codebook *c, float *output, int len, int step)
{
- int i, z = codebook_decode_start(f, c);
- float last = CODEBOOK_ELEMENT_BASE(c);
- if (z < 0) return FALSE;
- if (len > c->dimensions) len = c->dimensions;
+ int i,z = codebook_decode_start(f,c);
+ float last = CODEBOOK_ELEMENT_BASE(c);
+ if (z < 0) return FALSE;
+ if (len > c->dimensions) len = c->dimensions;
#ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK
- if (c->lookup_type == 1) {
- int div = 1;
- for (i = 0; i < len; ++i) {
- int off = (z / div) % c->lookup_values;
- float val = CODEBOOK_ELEMENT_FAST(c, off) + last;
- output[i*step] += val;
- if (c->sequence_p) last = val;
- div *= c->lookup_values;
- }
- return TRUE;
- }
+ if (c->lookup_type == 1) {
+ int div = 1;
+ for (i=0; i < len; ++i) {
+ int off = (z / div) % c->lookup_values;
+ float val = CODEBOOK_ELEMENT_FAST(c,off) + last;
+ output[i*step] += val;
+ if (c->sequence_p) last = val;
+ div *= c->lookup_values;
+ }
+ return TRUE;
+ }
#endif
- z *= c->dimensions;
- for (i = 0; i < len; ++i) {
- float val = CODEBOOK_ELEMENT_FAST(c, z + i) + last;
- output[i*step] += val;
- if (c->sequence_p) last = val;
- }
+ z *= c->dimensions;
+ for (i=0; i < len; ++i) {
+ float val = CODEBOOK_ELEMENT_FAST(c,z+i) + last;
+ output[i*step] += val;
+ if (c->sequence_p) last = val;
+ }
- return TRUE;
+ return TRUE;
}
static int codebook_decode_deinterleave_repeat(vorb *f, Codebook *c, float **outputs, int ch, int *c_inter_p, int *p_inter_p, int len, int total_decode)
{
- int c_inter = *c_inter_p;
- int p_inter = *p_inter_p;
- int i, z, effective = c->dimensions;
-
- // type 0 is only legal in a scalar context
- if (c->lookup_type == 0) return error(f, VORBIS_invalid_stream);
-
- while (total_decode > 0) {
- float last = CODEBOOK_ELEMENT_BASE(c);
- DECODE_VQ(z, f, c);
-#ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK
- assert(!c->sparse || z < c->sorted_entries);
-#endif
- if (z < 0) {
- if (!f->bytes_in_seg)
- if (f->last_seg) return FALSE;
- return error(f, VORBIS_invalid_stream);
- }
-
- // if this will take us off the end of the buffers, stop short!
- // we check by computing the length of the virtual interleaved
- // buffer (len*ch), our current offset within it (p_inter*ch)+(c_inter),
- // and the length we'll be using (effective)
- if (c_inter + p_inter*ch + effective > len * ch) {
- effective = len*ch - (p_inter*ch - c_inter);
- }
-
-#ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK
- if (c->lookup_type == 1) {
- int div = 1;
- for (i = 0; i < effective; ++i) {
- int off = (z / div) % c->lookup_values;
- float val = CODEBOOK_ELEMENT_FAST(c, off) + last;
- if (outputs[c_inter])
- outputs[c_inter][p_inter] += val;
- if (++c_inter == ch) { c_inter = 0; ++p_inter; }
- if (c->sequence_p) last = val;
- div *= c->lookup_values;
- }
- }
- else
-#endif
- {
- z *= c->dimensions;
- if (c->sequence_p) {
- for (i = 0; i < effective; ++i) {
- float val = CODEBOOK_ELEMENT_FAST(c, z + i) + last;
- if (outputs[c_inter])
- outputs[c_inter][p_inter] += val;
- if (++c_inter == ch) { c_inter = 0; ++p_inter; }
- last = val;
- }
+ int c_inter = *c_inter_p;
+ int p_inter = *p_inter_p;
+ int i,z, effective = c->dimensions;
+
+ // type 0 is only legal in a scalar context
+ if (c->lookup_type == 0) return error(f, VORBIS_invalid_stream);
+
+ while (total_decode > 0) {
+ float last = CODEBOOK_ELEMENT_BASE(c);
+ DECODE_VQ(z,f,c);
+ #ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK
+ assert(!c->sparse || z < c->sorted_entries);
+ #endif
+ if (z < 0) {
+ if (!f->bytes_in_seg)
+ if (f->last_seg) return FALSE;
+ return error(f, VORBIS_invalid_stream);
+ }
+
+ // if this will take us off the end of the buffers, stop short!
+ // we check by computing the length of the virtual interleaved
+ // buffer (len*ch), our current offset within it (p_inter*ch)+(c_inter),
+ // and the length we'll be using (effective)
+ if (c_inter + p_inter*ch + effective > len * ch) {
+ effective = len*ch - (p_inter*ch - c_inter);
+ }
+
+ #ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK
+ if (c->lookup_type == 1) {
+ int div = 1;
+ for (i=0; i < effective; ++i) {
+ int off = (z / div) % c->lookup_values;
+ float val = CODEBOOK_ELEMENT_FAST(c,off) + last;
+ if (outputs[c_inter])
+ outputs[c_inter][p_inter] += val;
+ if (++c_inter == ch) { c_inter = 0; ++p_inter; }
+ if (c->sequence_p) last = val;
+ div *= c->lookup_values;
+ }
+ } else
+ #endif
+ {
+ z *= c->dimensions;
+ if (c->sequence_p) {
+ for (i=0; i < effective; ++i) {
+ float val = CODEBOOK_ELEMENT_FAST(c,z+i) + last;
+ if (outputs[c_inter])
+ outputs[c_inter][p_inter] += val;
+ if (++c_inter == ch) { c_inter = 0; ++p_inter; }
+ last = val;
}
- else {
- for (i = 0; i < effective; ++i) {
- float val = CODEBOOK_ELEMENT_FAST(c, z + i) + last;
- if (outputs[c_inter])
- outputs[c_inter][p_inter] += val;
- if (++c_inter == ch) { c_inter = 0; ++p_inter; }
- }
+ } else {
+ for (i=0; i < effective; ++i) {
+ float val = CODEBOOK_ELEMENT_FAST(c,z+i) + last;
+ if (outputs[c_inter])
+ outputs[c_inter][p_inter] += val;
+ if (++c_inter == ch) { c_inter = 0; ++p_inter; }
}
- }
+ }
+ }
- total_decode -= effective;
- }
- *c_inter_p = c_inter;
- *p_inter_p = p_inter;
- return TRUE;
+ total_decode -= effective;
+ }
+ *c_inter_p = c_inter;
+ *p_inter_p = p_inter;
+ return TRUE;
}
static int predict_point(int x, int x0, int x1, int y0, int y1)
{
- int dy = y1 - y0;
- int adx = x1 - x0;
- // @OPTIMIZE: force int division to round in the right direction... is this necessary on x86?
- int err = abs(dy) * (x - x0);
- int off = err / adx;
- return dy < 0 ? y0 - off : y0 + off;
+ int dy = y1 - y0;
+ int adx = x1 - x0;
+ // @OPTIMIZE: force int division to round in the right direction... is this necessary on x86?
+ int err = abs(dy) * (x - x0);
+ int off = err / adx;
+ return dy < 0 ? y0 - off : y0 + off;
}
// the following table is block-copied from the specification
static float inverse_db_table[256] =
{
- 1.0649863e-07f, 1.1341951e-07f, 1.2079015e-07f, 1.2863978e-07f,
- 1.3699951e-07f, 1.4590251e-07f, 1.5538408e-07f, 1.6548181e-07f,
- 1.7623575e-07f, 1.8768855e-07f, 1.9988561e-07f, 2.1287530e-07f,
- 2.2670913e-07f, 2.4144197e-07f, 2.5713223e-07f, 2.7384213e-07f,
- 2.9163793e-07f, 3.1059021e-07f, 3.3077411e-07f, 3.5226968e-07f,
- 3.7516214e-07f, 3.9954229e-07f, 4.2550680e-07f, 4.5315863e-07f,
- 4.8260743e-07f, 5.1396998e-07f, 5.4737065e-07f, 5.8294187e-07f,
- 6.2082472e-07f, 6.6116941e-07f, 7.0413592e-07f, 7.4989464e-07f,
- 7.9862701e-07f, 8.5052630e-07f, 9.0579828e-07f, 9.6466216e-07f,
- 1.0273513e-06f, 1.0941144e-06f, 1.1652161e-06f, 1.2409384e-06f,
- 1.3215816e-06f, 1.4074654e-06f, 1.4989305e-06f, 1.5963394e-06f,
- 1.7000785e-06f, 1.8105592e-06f, 1.9282195e-06f, 2.0535261e-06f,
- 2.1869758e-06f, 2.3290978e-06f, 2.4804557e-06f, 2.6416497e-06f,
- 2.8133190e-06f, 2.9961443e-06f, 3.1908506e-06f, 3.3982101e-06f,
- 3.6190449e-06f, 3.8542308e-06f, 4.1047004e-06f, 4.3714470e-06f,
- 4.6555282e-06f, 4.9580707e-06f, 5.2802740e-06f, 5.6234160e-06f,
- 5.9888572e-06f, 6.3780469e-06f, 6.7925283e-06f, 7.2339451e-06f,
- 7.7040476e-06f, 8.2047000e-06f, 8.7378876e-06f, 9.3057248e-06f,
- 9.9104632e-06f, 1.0554501e-05f, 1.1240392e-05f, 1.1970856e-05f,
- 1.2748789e-05f, 1.3577278e-05f, 1.4459606e-05f, 1.5399272e-05f,
- 1.6400004e-05f, 1.7465768e-05f, 1.8600792e-05f, 1.9809576e-05f,
- 2.1096914e-05f, 2.2467911e-05f, 2.3928002e-05f, 2.5482978e-05f,
- 2.7139006e-05f, 2.8902651e-05f, 3.0780908e-05f, 3.2781225e-05f,
- 3.4911534e-05f, 3.7180282e-05f, 3.9596466e-05f, 4.2169667e-05f,
- 4.4910090e-05f, 4.7828601e-05f, 5.0936773e-05f, 5.4246931e-05f,
- 5.7772202e-05f, 6.1526565e-05f, 6.5524908e-05f, 6.9783085e-05f,
- 7.4317983e-05f, 7.9147585e-05f, 8.4291040e-05f, 8.9768747e-05f,
- 9.5602426e-05f, 0.00010181521f, 0.00010843174f, 0.00011547824f,
- 0.00012298267f, 0.00013097477f, 0.00013948625f, 0.00014855085f,
- 0.00015820453f, 0.00016848555f, 0.00017943469f, 0.00019109536f,
- 0.00020351382f, 0.00021673929f, 0.00023082423f, 0.00024582449f,
- 0.00026179955f, 0.00027881276f, 0.00029693158f, 0.00031622787f,
- 0.00033677814f, 0.00035866388f, 0.00038197188f, 0.00040679456f,
- 0.00043323036f, 0.00046138411f, 0.00049136745f, 0.00052329927f,
- 0.00055730621f, 0.00059352311f, 0.00063209358f, 0.00067317058f,
- 0.00071691700f, 0.00076350630f, 0.00081312324f, 0.00086596457f,
- 0.00092223983f, 0.00098217216f, 0.0010459992f, 0.0011139742f,
- 0.0011863665f, 0.0012634633f, 0.0013455702f, 0.0014330129f,
- 0.0015261382f, 0.0016253153f, 0.0017309374f, 0.0018434235f,
- 0.0019632195f, 0.0020908006f, 0.0022266726f, 0.0023713743f,
- 0.0025254795f, 0.0026895994f, 0.0028643847f, 0.0030505286f,
- 0.0032487691f, 0.0034598925f, 0.0036847358f, 0.0039241906f,
- 0.0041792066f, 0.0044507950f, 0.0047400328f, 0.0050480668f,
- 0.0053761186f, 0.0057254891f, 0.0060975636f, 0.0064938176f,
- 0.0069158225f, 0.0073652516f, 0.0078438871f, 0.0083536271f,
- 0.0088964928f, 0.009474637f, 0.010090352f, 0.010746080f,
- 0.011444421f, 0.012188144f, 0.012980198f, 0.013823725f,
- 0.014722068f, 0.015678791f, 0.016697687f, 0.017782797f,
- 0.018938423f, 0.020169149f, 0.021479854f, 0.022875735f,
- 0.024362330f, 0.025945531f, 0.027631618f, 0.029427276f,
- 0.031339626f, 0.033376252f, 0.035545228f, 0.037855157f,
- 0.040315199f, 0.042935108f, 0.045725273f, 0.048696758f,
- 0.051861348f, 0.055231591f, 0.058820850f, 0.062643361f,
- 0.066714279f, 0.071049749f, 0.075666962f, 0.080584227f,
- 0.085821044f, 0.091398179f, 0.097337747f, 0.10366330f,
- 0.11039993f, 0.11757434f, 0.12521498f, 0.13335215f,
- 0.14201813f, 0.15124727f, 0.16107617f, 0.17154380f,
- 0.18269168f, 0.19456402f, 0.20720788f, 0.22067342f,
- 0.23501402f, 0.25028656f, 0.26655159f, 0.28387361f,
- 0.30232132f, 0.32196786f, 0.34289114f, 0.36517414f,
- 0.38890521f, 0.41417847f, 0.44109412f, 0.46975890f,
- 0.50028648f, 0.53279791f, 0.56742212f, 0.60429640f,
- 0.64356699f, 0.68538959f, 0.72993007f, 0.77736504f,
- 0.82788260f, 0.88168307f, 0.9389798f, 1.0f
+ 1.0649863e-07f, 1.1341951e-07f, 1.2079015e-07f, 1.2863978e-07f,
+ 1.3699951e-07f, 1.4590251e-07f, 1.5538408e-07f, 1.6548181e-07f,
+ 1.7623575e-07f, 1.8768855e-07f, 1.9988561e-07f, 2.1287530e-07f,
+ 2.2670913e-07f, 2.4144197e-07f, 2.5713223e-07f, 2.7384213e-07f,
+ 2.9163793e-07f, 3.1059021e-07f, 3.3077411e-07f, 3.5226968e-07f,
+ 3.7516214e-07f, 3.9954229e-07f, 4.2550680e-07f, 4.5315863e-07f,
+ 4.8260743e-07f, 5.1396998e-07f, 5.4737065e-07f, 5.8294187e-07f,
+ 6.2082472e-07f, 6.6116941e-07f, 7.0413592e-07f, 7.4989464e-07f,
+ 7.9862701e-07f, 8.5052630e-07f, 9.0579828e-07f, 9.6466216e-07f,
+ 1.0273513e-06f, 1.0941144e-06f, 1.1652161e-06f, 1.2409384e-06f,
+ 1.3215816e-06f, 1.4074654e-06f, 1.4989305e-06f, 1.5963394e-06f,
+ 1.7000785e-06f, 1.8105592e-06f, 1.9282195e-06f, 2.0535261e-06f,
+ 2.1869758e-06f, 2.3290978e-06f, 2.4804557e-06f, 2.6416497e-06f,
+ 2.8133190e-06f, 2.9961443e-06f, 3.1908506e-06f, 3.3982101e-06f,
+ 3.6190449e-06f, 3.8542308e-06f, 4.1047004e-06f, 4.3714470e-06f,
+ 4.6555282e-06f, 4.9580707e-06f, 5.2802740e-06f, 5.6234160e-06f,
+ 5.9888572e-06f, 6.3780469e-06f, 6.7925283e-06f, 7.2339451e-06f,
+ 7.7040476e-06f, 8.2047000e-06f, 8.7378876e-06f, 9.3057248e-06f,
+ 9.9104632e-06f, 1.0554501e-05f, 1.1240392e-05f, 1.1970856e-05f,
+ 1.2748789e-05f, 1.3577278e-05f, 1.4459606e-05f, 1.5399272e-05f,
+ 1.6400004e-05f, 1.7465768e-05f, 1.8600792e-05f, 1.9809576e-05f,
+ 2.1096914e-05f, 2.2467911e-05f, 2.3928002e-05f, 2.5482978e-05f,
+ 2.7139006e-05f, 2.8902651e-05f, 3.0780908e-05f, 3.2781225e-05f,
+ 3.4911534e-05f, 3.7180282e-05f, 3.9596466e-05f, 4.2169667e-05f,
+ 4.4910090e-05f, 4.7828601e-05f, 5.0936773e-05f, 5.4246931e-05f,
+ 5.7772202e-05f, 6.1526565e-05f, 6.5524908e-05f, 6.9783085e-05f,
+ 7.4317983e-05f, 7.9147585e-05f, 8.4291040e-05f, 8.9768747e-05f,
+ 9.5602426e-05f, 0.00010181521f, 0.00010843174f, 0.00011547824f,
+ 0.00012298267f, 0.00013097477f, 0.00013948625f, 0.00014855085f,
+ 0.00015820453f, 0.00016848555f, 0.00017943469f, 0.00019109536f,
+ 0.00020351382f, 0.00021673929f, 0.00023082423f, 0.00024582449f,
+ 0.00026179955f, 0.00027881276f, 0.00029693158f, 0.00031622787f,
+ 0.00033677814f, 0.00035866388f, 0.00038197188f, 0.00040679456f,
+ 0.00043323036f, 0.00046138411f, 0.00049136745f, 0.00052329927f,
+ 0.00055730621f, 0.00059352311f, 0.00063209358f, 0.00067317058f,
+ 0.00071691700f, 0.00076350630f, 0.00081312324f, 0.00086596457f,
+ 0.00092223983f, 0.00098217216f, 0.0010459992f, 0.0011139742f,
+ 0.0011863665f, 0.0012634633f, 0.0013455702f, 0.0014330129f,
+ 0.0015261382f, 0.0016253153f, 0.0017309374f, 0.0018434235f,
+ 0.0019632195f, 0.0020908006f, 0.0022266726f, 0.0023713743f,
+ 0.0025254795f, 0.0026895994f, 0.0028643847f, 0.0030505286f,
+ 0.0032487691f, 0.0034598925f, 0.0036847358f, 0.0039241906f,
+ 0.0041792066f, 0.0044507950f, 0.0047400328f, 0.0050480668f,
+ 0.0053761186f, 0.0057254891f, 0.0060975636f, 0.0064938176f,
+ 0.0069158225f, 0.0073652516f, 0.0078438871f, 0.0083536271f,
+ 0.0088964928f, 0.009474637f, 0.010090352f, 0.010746080f,
+ 0.011444421f, 0.012188144f, 0.012980198f, 0.013823725f,
+ 0.014722068f, 0.015678791f, 0.016697687f, 0.017782797f,
+ 0.018938423f, 0.020169149f, 0.021479854f, 0.022875735f,
+ 0.024362330f, 0.025945531f, 0.027631618f, 0.029427276f,
+ 0.031339626f, 0.033376252f, 0.035545228f, 0.037855157f,
+ 0.040315199f, 0.042935108f, 0.045725273f, 0.048696758f,
+ 0.051861348f, 0.055231591f, 0.058820850f, 0.062643361f,
+ 0.066714279f, 0.071049749f, 0.075666962f, 0.080584227f,
+ 0.085821044f, 0.091398179f, 0.097337747f, 0.10366330f,
+ 0.11039993f, 0.11757434f, 0.12521498f, 0.13335215f,
+ 0.14201813f, 0.15124727f, 0.16107617f, 0.17154380f,
+ 0.18269168f, 0.19456402f, 0.20720788f, 0.22067342f,
+ 0.23501402f, 0.25028656f, 0.26655159f, 0.28387361f,
+ 0.30232132f, 0.32196786f, 0.34289114f, 0.36517414f,
+ 0.38890521f, 0.41417847f, 0.44109412f, 0.46975890f,
+ 0.50028648f, 0.53279791f, 0.56742212f, 0.60429640f,
+ 0.64356699f, 0.68538959f, 0.72993007f, 0.77736504f,
+ 0.82788260f, 0.88168307f, 0.9389798f, 1.0f
};
@@ -1986,303 +2033,254 @@ int8 integer_divide_table[DIVTAB_NUMER][DIVTAB_DENOM]; // 2KB
static __forceinline void draw_line(float *output, int x0, int y0, int x1, int y1, int n)
{
- int dy = y1 - y0;
- int adx = x1 - x0;
- int ady = abs(dy);
- int base;
- int x = x0, y = y0;
- int err = 0;
- int sy;
+ int dy = y1 - y0;
+ int adx = x1 - x0;
+ int ady = abs(dy);
+ int base;
+ int x=x0,y=y0;
+ int err = 0;
+ int sy;
#ifdef STB_VORBIS_DIVIDE_TABLE
- if (adx < DIVTAB_DENOM && ady < DIVTAB_NUMER) {
- if (dy < 0) {
- base = -integer_divide_table[ady][adx];
- sy = base - 1;
- }
- else {
- base = integer_divide_table[ady][adx];
- sy = base + 1;
- }
- }
- else {
- base = dy / adx;
- if (dy < 0)
- sy = base - 1;
- else
- sy = base + 1;
- }
+ if (adx < DIVTAB_DENOM && ady < DIVTAB_NUMER) {
+ if (dy < 0) {
+ base = -integer_divide_table[ady][adx];
+ sy = base-1;
+ } else {
+ base = integer_divide_table[ady][adx];
+ sy = base+1;
+ }
+ } else {
+ base = dy / adx;
+ if (dy < 0)
+ sy = base - 1;
+ else
+ sy = base+1;
+ }
#else
- base = dy / adx;
- if (dy < 0)
- sy = base - 1;
- else
- sy = base + 1;
+ base = dy / adx;
+ if (dy < 0)
+ sy = base - 1;
+ else
+ sy = base+1;
#endif
- ady -= abs(base) * adx;
- if (x1 > n) x1 = n;
- if (x < x1) {
- LINE_OP(output[x], inverse_db_table[y]);
- for (++x; x < x1; ++x) {
- err += ady;
- if (err >= adx) {
- err -= adx;
- y += sy;
- }
- else
- y += base;
- LINE_OP(output[x], inverse_db_table[y]);
- }
- }
+ ady -= abs(base) * adx;
+ if (x1 > n) x1 = n;
+ if (x < x1) {
+ LINE_OP(output[x], inverse_db_table[y&255]);
+ for (++x; x < x1; ++x) {
+ err += ady;
+ if (err >= adx) {
+ err -= adx;
+ y += sy;
+ } else
+ y += base;
+ LINE_OP(output[x], inverse_db_table[y&255]);
+ }
+ }
}
static int residue_decode(vorb *f, Codebook *book, float *target, int offset, int n, int rtype)
{
- int k;
- if (rtype == 0) {
- int step = n / book->dimensions;
- for (k = 0; k < step; ++k)
- if (!codebook_decode_step(f, book, target + offset + k, n - offset - k, step))
- return FALSE;
- }
- else {
- for (k = 0; k < n; ) {
- if (!codebook_decode(f, book, target + offset, n - k))
- return FALSE;
- k += book->dimensions;
- offset += book->dimensions;
- }
- }
- return TRUE;
+ int k;
+ if (rtype == 0) {
+ int step = n / book->dimensions;
+ for (k=0; k < step; ++k)
+ if (!codebook_decode_step(f, book, target+offset+k, n-offset-k, step))
+ return FALSE;
+ } else {
+ for (k=0; k < n; ) {
+ if (!codebook_decode(f, book, target+offset, n-k))
+ return FALSE;
+ k += book->dimensions;
+ offset += book->dimensions;
+ }
+ }
+ return TRUE;
}
// n is 1/2 of the blocksize --
// specification: "Correct per-vector decode length is [n]/2"
static void decode_residue(vorb *f, float *residue_buffers[], int ch, int n, int rn, uint8 *do_not_decode)
{
- int i, j, pass;
- Residue *r = f->residue_config + rn;
- int rtype = f->residue_types[rn];
- int c = r->classbook;
- int classwords = f->codebooks[c].dimensions;
- unsigned int actual_size = rtype == 2 ? n * 2 : n;
- unsigned int limit_r_begin = (r->begin < actual_size ? r->begin : actual_size);
- unsigned int limit_r_end = (r->end < actual_size ? r->end : actual_size);
- int n_read = limit_r_end - limit_r_begin;
- int part_read = n_read / r->part_size;
- int temp_alloc_point = temp_alloc_save(f);
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- uint8 ***part_classdata = (uint8 ***)temp_block_array(f, f->channels, part_read * sizeof(**part_classdata));
-#else
- int **classifications = (int **)temp_block_array(f, f->channels, part_read * sizeof(**classifications));
-#endif
-
- CHECK(f);
-
- for (i = 0; i < ch; ++i)
- if (!do_not_decode[i])
- memset(residue_buffers[i], 0, sizeof(float) * n);
-
- if (rtype == 2 && ch != 1) {
- for (j = 0; j < ch; ++j)
- if (!do_not_decode[j])
- break;
- if (j == ch)
- goto done;
-
- for (pass = 0; pass < 8; ++pass) {
- int pcount = 0, class_set = 0;
- if (ch == 2) {
- while (pcount < part_read) {
- int z = r->begin + pcount*r->part_size;
- int c_inter = (z & 1), p_inter = z >> 1;
- if (pass == 0) {
- Codebook *c = f->codebooks + r->classbook;
- int q;
- DECODE(q, f, c);
- if (q == EOP) goto done;
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- part_classdata[0][class_set] = r->classdata[q];
-#else
- for (i = classwords - 1; i >= 0; --i) {
- classifications[0][i + pcount] = q % r->classifications;
- q /= r->classifications;
- }
-#endif
- }
- for (i = 0; i < classwords && pcount < part_read; ++i, ++pcount) {
- int z = r->begin + pcount*r->part_size;
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- int c = part_classdata[0][class_set][i];
-#else
- int c = classifications[0][pcount];
-#endif
- int b = r->residue_books[c][pass];
- if (b >= 0) {
- Codebook *book = f->codebooks + b;
-#ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK
- if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size))
- goto done;
-#else
- // saves 1%
- if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size))
- goto done;
-#endif
- }
- else {
- z += r->part_size;
- c_inter = z & 1;
- p_inter = z >> 1;
- }
- }
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- ++class_set;
-#endif
- }
- }
- else if (ch == 1) {
- while (pcount < part_read) {
- int z = r->begin + pcount*r->part_size;
- int c_inter = 0, p_inter = z;
- if (pass == 0) {
- Codebook *c = f->codebooks + r->classbook;
- int q;
- DECODE(q, f, c);
- if (q == EOP) goto done;
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- part_classdata[0][class_set] = r->classdata[q];
-#else
- for (i = classwords - 1; i >= 0; --i) {
- classifications[0][i + pcount] = q % r->classifications;
- q /= r->classifications;
- }
-#endif
- }
- for (i = 0; i < classwords && pcount < part_read; ++i, ++pcount) {
- int z = r->begin + pcount*r->part_size;
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- int c = part_classdata[0][class_set][i];
-#else
- int c = classifications[0][pcount];
-#endif
- int b = r->residue_books[c][pass];
- if (b >= 0) {
- Codebook *book = f->codebooks + b;
- if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size))
- goto done;
- }
- else {
- z += r->part_size;
- c_inter = 0;
- p_inter = z;
- }
- }
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- ++class_set;
-#endif
- }
+ int i,j,pass;
+ Residue *r = f->residue_config + rn;
+ int rtype = f->residue_types[rn];
+ int c = r->classbook;
+ int classwords = f->codebooks[c].dimensions;
+ unsigned int actual_size = rtype == 2 ? n*2 : n;
+ unsigned int limit_r_begin = (r->begin < actual_size ? r->begin : actual_size);
+ unsigned int limit_r_end = (r->end < actual_size ? r->end : actual_size);
+ int n_read = limit_r_end - limit_r_begin;
+ int part_read = n_read / r->part_size;
+ int temp_alloc_point = temp_alloc_save(f);
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ uint8 ***part_classdata = (uint8 ***) temp_block_array(f,f->channels, part_read * sizeof(**part_classdata));
+ #else
+ int **classifications = (int **) temp_block_array(f,f->channels, part_read * sizeof(**classifications));
+ #endif
+
+ CHECK(f);
+
+ for (i=0; i < ch; ++i)
+ if (!do_not_decode[i])
+ memset(residue_buffers[i], 0, sizeof(float) * n);
+
+ if (rtype == 2 && ch != 1) {
+ for (j=0; j < ch; ++j)
+ if (!do_not_decode[j])
+ break;
+ if (j == ch)
+ goto done;
+
+ for (pass=0; pass < 8; ++pass) {
+ int pcount = 0, class_set = 0;
+ if (ch == 2) {
+ while (pcount < part_read) {
+ int z = r->begin + pcount*r->part_size;
+ int c_inter = (z & 1), p_inter = z>>1;
+ if (pass == 0) {
+ Codebook *c = f->codebooks+r->classbook;
+ int q;
+ DECODE(q,f,c);
+ if (q == EOP) goto done;
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ part_classdata[0][class_set] = r->classdata[q];
+ #else
+ for (i=classwords-1; i >= 0; --i) {
+ classifications[0][i+pcount] = q % r->classifications;
+ q /= r->classifications;
+ }
+ #endif
+ }
+ for (i=0; i < classwords && pcount < part_read; ++i, ++pcount) {
+ int z = r->begin + pcount*r->part_size;
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ int c = part_classdata[0][class_set][i];
+ #else
+ int c = classifications[0][pcount];
+ #endif
+ int b = r->residue_books[c][pass];
+ if (b >= 0) {
+ Codebook *book = f->codebooks + b;
+ #ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK
+ if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size))
+ goto done;
+ #else
+ // saves 1%
+ if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size))
+ goto done;
+ #endif
+ } else {
+ z += r->part_size;
+ c_inter = z & 1;
+ p_inter = z >> 1;
+ }
+ }
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ ++class_set;
+ #endif
}
- else {
- while (pcount < part_read) {
- int z = r->begin + pcount*r->part_size;
- int c_inter = z % ch, p_inter = z / ch;
- if (pass == 0) {
- Codebook *c = f->codebooks + r->classbook;
- int q;
- DECODE(q, f, c);
- if (q == EOP) goto done;
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- part_classdata[0][class_set] = r->classdata[q];
-#else
- for (i = classwords - 1; i >= 0; --i) {
- classifications[0][i + pcount] = q % r->classifications;
- q /= r->classifications;
- }
-#endif
- }
- for (i = 0; i < classwords && pcount < part_read; ++i, ++pcount) {
- int z = r->begin + pcount*r->part_size;
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- int c = part_classdata[0][class_set][i];
-#else
- int c = classifications[0][pcount];
-#endif
- int b = r->residue_books[c][pass];
- if (b >= 0) {
- Codebook *book = f->codebooks + b;
- if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size))
- goto done;
- }
- else {
- z += r->part_size;
- c_inter = z % ch;
- p_inter = z / ch;
- }
- }
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- ++class_set;
-#endif
- }
+ } else if (ch > 2) {
+ while (pcount < part_read) {
+ int z = r->begin + pcount*r->part_size;
+ int c_inter = z % ch, p_inter = z/ch;
+ if (pass == 0) {
+ Codebook *c = f->codebooks+r->classbook;
+ int q;
+ DECODE(q,f,c);
+ if (q == EOP) goto done;
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ part_classdata[0][class_set] = r->classdata[q];
+ #else
+ for (i=classwords-1; i >= 0; --i) {
+ classifications[0][i+pcount] = q % r->classifications;
+ q /= r->classifications;
+ }
+ #endif
+ }
+ for (i=0; i < classwords && pcount < part_read; ++i, ++pcount) {
+ int z = r->begin + pcount*r->part_size;
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ int c = part_classdata[0][class_set][i];
+ #else
+ int c = classifications[0][pcount];
+ #endif
+ int b = r->residue_books[c][pass];
+ if (b >= 0) {
+ Codebook *book = f->codebooks + b;
+ if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size))
+ goto done;
+ } else {
+ z += r->part_size;
+ c_inter = z % ch;
+ p_inter = z / ch;
+ }
+ }
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ ++class_set;
+ #endif
}
- }
- goto done;
- }
- CHECK(f);
-
- for (pass = 0; pass < 8; ++pass) {
- int pcount = 0, class_set = 0;
- while (pcount < part_read) {
- if (pass == 0) {
- for (j = 0; j < ch; ++j) {
- if (!do_not_decode[j]) {
- Codebook *c = f->codebooks + r->classbook;
- int temp;
- DECODE(temp, f, c);
- if (temp == EOP) goto done;
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- part_classdata[j][class_set] = r->classdata[temp];
-#else
- for (i = classwords - 1; i >= 0; --i) {
- classifications[j][i + pcount] = temp % r->classifications;
- temp /= r->classifications;
- }
-#endif
- }
- }
+ }
+ }
+ goto done;
+ }
+ CHECK(f);
+
+ for (pass=0; pass < 8; ++pass) {
+ int pcount = 0, class_set=0;
+ while (pcount < part_read) {
+ if (pass == 0) {
+ for (j=0; j < ch; ++j) {
+ if (!do_not_decode[j]) {
+ Codebook *c = f->codebooks+r->classbook;
+ int temp;
+ DECODE(temp,f,c);
+ if (temp == EOP) goto done;
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ part_classdata[j][class_set] = r->classdata[temp];
+ #else
+ for (i=classwords-1; i >= 0; --i) {
+ classifications[j][i+pcount] = temp % r->classifications;
+ temp /= r->classifications;
+ }
+ #endif
+ }
}
- for (i = 0; i < classwords && pcount < part_read; ++i, ++pcount) {
- for (j = 0; j < ch; ++j) {
- if (!do_not_decode[j]) {
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- int c = part_classdata[j][class_set][i];
-#else
- int c = classifications[j][pcount];
-#endif
- int b = r->residue_books[c][pass];
- if (b >= 0) {
- float *target = residue_buffers[j];
- int offset = r->begin + pcount * r->part_size;
- int n = r->part_size;
- Codebook *book = f->codebooks + b;
- if (!residue_decode(f, book, target, offset, n, rtype))
- goto done;
- }
- }
- }
+ }
+ for (i=0; i < classwords && pcount < part_read; ++i, ++pcount) {
+ for (j=0; j < ch; ++j) {
+ if (!do_not_decode[j]) {
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ int c = part_classdata[j][class_set][i];
+ #else
+ int c = classifications[j][pcount];
+ #endif
+ int b = r->residue_books[c][pass];
+ if (b >= 0) {
+ float *target = residue_buffers[j];
+ int offset = r->begin + pcount * r->part_size;
+ int n = r->part_size;
+ Codebook *book = f->codebooks + b;
+ if (!residue_decode(f, book, target, offset, n, rtype))
+ goto done;
+ }
+ }
}
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- ++class_set;
-#endif
- }
- }
-done:
- CHECK(f);
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- temp_free(f, part_classdata);
-#else
- temp_free(f, classifications);
-#endif
- temp_alloc_restore(f, temp_alloc_point);
+ }
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ ++class_set;
+ #endif
+ }
+ }
+ done:
+ CHECK(f);
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ temp_free(f,part_classdata);
+ #else
+ temp_free(f,classifications);
+ #endif
+ temp_alloc_restore(f,temp_alloc_point);
}
@@ -2290,76 +2288,76 @@ done:
// slow way for debugging
void inverse_mdct_slow(float *buffer, int n)
{
- int i, j;
- int n2 = n >> 1;
- float *x = (float *)malloc(sizeof(*x) * n2);
- memcpy(x, buffer, sizeof(*x) * n2);
- for (i = 0; i < n; ++i) {
- float acc = 0;
- for (j = 0; j < n2; ++j)
- // formula from paper:
- //acc += n/4.0f * x[j] * (float) cos(M_PI / 2 / n * (2 * i + 1 + n/2.0)*(2*j+1));
- // formula from wikipedia
- //acc += 2.0f / n2 * x[j] * (float) cos(M_PI/n2 * (i + 0.5 + n2/2)*(j + 0.5));
- // these are equivalent, except the formula from the paper inverts the multiplier!
- // however, what actually works is NO MULTIPLIER!?!
- //acc += 64 * 2.0f / n2 * x[j] * (float) cos(M_PI/n2 * (i + 0.5 + n2/2)*(j + 0.5));
- acc += x[j] * (float)cos(M_PI / 2 / n * (2 * i + 1 + n / 2.0)*(2 * j + 1));
- buffer[i] = acc;
- }
- free(x);
+ int i,j;
+ int n2 = n >> 1;
+ float *x = (float *) malloc(sizeof(*x) * n2);
+ memcpy(x, buffer, sizeof(*x) * n2);
+ for (i=0; i < n; ++i) {
+ float acc = 0;
+ for (j=0; j < n2; ++j)
+ // formula from paper:
+ //acc += n/4.0f * x[j] * (float) cos(M_PI / 2 / n * (2 * i + 1 + n/2.0)*(2*j+1));
+ // formula from wikipedia
+ //acc += 2.0f / n2 * x[j] * (float) cos(M_PI/n2 * (i + 0.5 + n2/2)*(j + 0.5));
+ // these are equivalent, except the formula from the paper inverts the multiplier!
+ // however, what actually works is NO MULTIPLIER!?!
+ //acc += 64 * 2.0f / n2 * x[j] * (float) cos(M_PI/n2 * (i + 0.5 + n2/2)*(j + 0.5));
+ acc += x[j] * (float) cos(M_PI / 2 / n * (2 * i + 1 + n/2.0)*(2*j+1));
+ buffer[i] = acc;
+ }
+ free(x);
}
#elif 0
// same as above, but just barely able to run in real time on modern machines
void inverse_mdct_slow(float *buffer, int n, vorb *f, int blocktype)
{
- float mcos[16384];
- int i, j;
- int n2 = n >> 1, nmask = (n << 2) - 1;
- float *x = (float *)malloc(sizeof(*x) * n2);
- memcpy(x, buffer, sizeof(*x) * n2);
- for (i = 0; i < 4 * n; ++i)
- mcos[i] = (float)cos(M_PI / 2 * i / n);
-
- for (i = 0; i < n; ++i) {
- float acc = 0;
- for (j = 0; j < n2; ++j)
- acc += x[j] * mcos[(2 * i + 1 + n2)*(2 * j + 1) & nmask];
- buffer[i] = acc;
- }
- free(x);
+ float mcos[16384];
+ int i,j;
+ int n2 = n >> 1, nmask = (n << 2) -1;
+ float *x = (float *) malloc(sizeof(*x) * n2);
+ memcpy(x, buffer, sizeof(*x) * n2);
+ for (i=0; i < 4*n; ++i)
+ mcos[i] = (float) cos(M_PI / 2 * i / n);
+
+ for (i=0; i < n; ++i) {
+ float acc = 0;
+ for (j=0; j < n2; ++j)
+ acc += x[j] * mcos[(2 * i + 1 + n2)*(2*j+1) & nmask];
+ buffer[i] = acc;
+ }
+ free(x);
}
#elif 0
// transform to use a slow dct-iv; this is STILL basically trivial,
// but only requires half as many ops
void dct_iv_slow(float *buffer, int n)
{
- float mcos[16384];
- float x[2048];
- int i, j;
- int n2 = n >> 1, nmask = (n << 3) - 1;
- memcpy(x, buffer, sizeof(*x) * n);
- for (i = 0; i < 8 * n; ++i)
- mcos[i] = (float)cos(M_PI / 4 * i / n);
- for (i = 0; i < n; ++i) {
- float acc = 0;
- for (j = 0; j < n; ++j)
- acc += x[j] * mcos[((2 * i + 1)*(2 * j + 1)) & nmask];
- buffer[i] = acc;
- }
+ float mcos[16384];
+ float x[2048];
+ int i,j;
+ int n2 = n >> 1, nmask = (n << 3) - 1;
+ memcpy(x, buffer, sizeof(*x) * n);
+ for (i=0; i < 8*n; ++i)
+ mcos[i] = (float) cos(M_PI / 4 * i / n);
+ for (i=0; i < n; ++i) {
+ float acc = 0;
+ for (j=0; j < n; ++j)
+ acc += x[j] * mcos[((2 * i + 1)*(2*j+1)) & nmask];
+ buffer[i] = acc;
+ }
}
void inverse_mdct_slow(float *buffer, int n, vorb *f, int blocktype)
{
- int i, n4 = n >> 2, n2 = n >> 1, n3_4 = n - n4;
- float temp[4096];
+ int i, n4 = n >> 2, n2 = n >> 1, n3_4 = n - n4;
+ float temp[4096];
- memcpy(temp, buffer, n2 * sizeof(float));
- dct_iv_slow(temp, n2); // returns -c'-d, a-b'
+ memcpy(temp, buffer, n2 * sizeof(float));
+ dct_iv_slow(temp, n2); // returns -c'-d, a-b'
- for (i = 0; i < n4; ++i) buffer[i] = temp[i + n4]; // a-b'
- for (; i < n3_4; ++i) buffer[i] = -temp[n3_4 - i - 1]; // b-a', c+d'
- for (; i < n; ++i) buffer[i] = -temp[i - n3_4]; // c'+d
+ for (i=0; i < n4 ; ++i) buffer[i] = temp[i+n4]; // a-b'
+ for ( ; i < n3_4; ++i) buffer[i] = -temp[n3_4 - i - 1]; // b-a', c+d'
+ for ( ; i < n ; ++i) buffer[i] = -temp[i - n3_4]; // c'+d
}
#endif
@@ -2372,34 +2370,34 @@ void inverse_mdct_slow(float *buffer, int n, vorb *f, int blocktype)
// by Jeff Roberts... useful for performance comparison
typedef struct
{
- int n;
- int log2n;
+ int n;
+ int log2n;
- float *trig;
- int *bitrev;
+ float *trig;
+ int *bitrev;
- float scale;
+ float scale;
} mdct_lookup;
extern void mdct_init(mdct_lookup *lookup, int n);
extern void mdct_clear(mdct_lookup *l);
extern void mdct_backward(mdct_lookup *init, float *in, float *out);
-mdct_lookup M1, M2;
+mdct_lookup M1,M2;
void inverse_mdct(float *buffer, int n, vorb *f, int blocktype)
{
- mdct_lookup *M;
- if (M1.n == n) M = &M1;
- else if (M2.n == n) M = &M2;
- else if (M1.n == 0) { mdct_init(&M1, n); M = &M1; }
- else {
- if (M2.n) __asm int 3;
- mdct_init(&M2, n);
- M = &M2;
- }
-
- mdct_backward(M, buffer, buffer);
+ mdct_lookup *M;
+ if (M1.n == n) M = &M1;
+ else if (M2.n == n) M = &M2;
+ else if (M1.n == 0) { mdct_init(&M1, n); M = &M1; }
+ else {
+ if (M2.n) __asm int 3;
+ mdct_init(&M2, n);
+ M = &M2;
+ }
+
+ mdct_backward(M, buffer, buffer);
}
#endif
@@ -2409,664 +2407,661 @@ void inverse_mdct(float *buffer, int n, vorb *f, int blocktype)
// they're probably already being inlined.
static void imdct_step3_iter0_loop(int n, float *e, int i_off, int k_off, float *A)
{
- float *ee0 = e + i_off;
- float *ee2 = ee0 + k_off;
- int i;
-
- assert((n & 3) == 0);
- for (i = (n >> 2); i > 0; --i) {
- float k00_20, k01_21;
- k00_20 = ee0[0] - ee2[0];
- k01_21 = ee0[-1] - ee2[-1];
- ee0[0] += ee2[0];//ee0[ 0] = ee0[ 0] + ee2[ 0];
- ee0[-1] += ee2[-1];//ee0[-1] = ee0[-1] + ee2[-1];
- ee2[0] = k00_20 * A[0] - k01_21 * A[1];
- ee2[-1] = k01_21 * A[0] + k00_20 * A[1];
- A += 8;
-
- k00_20 = ee0[-2] - ee2[-2];
- k01_21 = ee0[-3] - ee2[-3];
- ee0[-2] += ee2[-2];//ee0[-2] = ee0[-2] + ee2[-2];
- ee0[-3] += ee2[-3];//ee0[-3] = ee0[-3] + ee2[-3];
- ee2[-2] = k00_20 * A[0] - k01_21 * A[1];
- ee2[-3] = k01_21 * A[0] + k00_20 * A[1];
- A += 8;
-
- k00_20 = ee0[-4] - ee2[-4];
- k01_21 = ee0[-5] - ee2[-5];
- ee0[-4] += ee2[-4];//ee0[-4] = ee0[-4] + ee2[-4];
- ee0[-5] += ee2[-5];//ee0[-5] = ee0[-5] + ee2[-5];
- ee2[-4] = k00_20 * A[0] - k01_21 * A[1];
- ee2[-5] = k01_21 * A[0] + k00_20 * A[1];
- A += 8;
-
- k00_20 = ee0[-6] - ee2[-6];
- k01_21 = ee0[-7] - ee2[-7];
- ee0[-6] += ee2[-6];//ee0[-6] = ee0[-6] + ee2[-6];
- ee0[-7] += ee2[-7];//ee0[-7] = ee0[-7] + ee2[-7];
- ee2[-6] = k00_20 * A[0] - k01_21 * A[1];
- ee2[-7] = k01_21 * A[0] + k00_20 * A[1];
- A += 8;
- ee0 -= 8;
- ee2 -= 8;
- }
+ float *ee0 = e + i_off;
+ float *ee2 = ee0 + k_off;
+ int i;
+
+ assert((n & 3) == 0);
+ for (i=(n>>2); i > 0; --i) {
+ float k00_20, k01_21;
+ k00_20 = ee0[ 0] - ee2[ 0];
+ k01_21 = ee0[-1] - ee2[-1];
+ ee0[ 0] += ee2[ 0];//ee0[ 0] = ee0[ 0] + ee2[ 0];
+ ee0[-1] += ee2[-1];//ee0[-1] = ee0[-1] + ee2[-1];
+ ee2[ 0] = k00_20 * A[0] - k01_21 * A[1];
+ ee2[-1] = k01_21 * A[0] + k00_20 * A[1];
+ A += 8;
+
+ k00_20 = ee0[-2] - ee2[-2];
+ k01_21 = ee0[-3] - ee2[-3];
+ ee0[-2] += ee2[-2];//ee0[-2] = ee0[-2] + ee2[-2];
+ ee0[-3] += ee2[-3];//ee0[-3] = ee0[-3] + ee2[-3];
+ ee2[-2] = k00_20 * A[0] - k01_21 * A[1];
+ ee2[-3] = k01_21 * A[0] + k00_20 * A[1];
+ A += 8;
+
+ k00_20 = ee0[-4] - ee2[-4];
+ k01_21 = ee0[-5] - ee2[-5];
+ ee0[-4] += ee2[-4];//ee0[-4] = ee0[-4] + ee2[-4];
+ ee0[-5] += ee2[-5];//ee0[-5] = ee0[-5] + ee2[-5];
+ ee2[-4] = k00_20 * A[0] - k01_21 * A[1];
+ ee2[-5] = k01_21 * A[0] + k00_20 * A[1];
+ A += 8;
+
+ k00_20 = ee0[-6] - ee2[-6];
+ k01_21 = ee0[-7] - ee2[-7];
+ ee0[-6] += ee2[-6];//ee0[-6] = ee0[-6] + ee2[-6];
+ ee0[-7] += ee2[-7];//ee0[-7] = ee0[-7] + ee2[-7];
+ ee2[-6] = k00_20 * A[0] - k01_21 * A[1];
+ ee2[-7] = k01_21 * A[0] + k00_20 * A[1];
+ A += 8;
+ ee0 -= 8;
+ ee2 -= 8;
+ }
}
static void imdct_step3_inner_r_loop(int lim, float *e, int d0, int k_off, float *A, int k1)
{
- int i;
- float k00_20, k01_21;
+ int i;
+ float k00_20, k01_21;
- float *e0 = e + d0;
- float *e2 = e0 + k_off;
+ float *e0 = e + d0;
+ float *e2 = e0 + k_off;
- for (i = lim >> 2; i > 0; --i) {
- k00_20 = e0[-0] - e2[-0];
- k01_21 = e0[-1] - e2[-1];
- e0[-0] += e2[-0];//e0[-0] = e0[-0] + e2[-0];
- e0[-1] += e2[-1];//e0[-1] = e0[-1] + e2[-1];
- e2[-0] = (k00_20)*A[0] - (k01_21)* A[1];
- e2[-1] = (k01_21)*A[0] + (k00_20)* A[1];
+ for (i=lim >> 2; i > 0; --i) {
+ k00_20 = e0[-0] - e2[-0];
+ k01_21 = e0[-1] - e2[-1];
+ e0[-0] += e2[-0];//e0[-0] = e0[-0] + e2[-0];
+ e0[-1] += e2[-1];//e0[-1] = e0[-1] + e2[-1];
+ e2[-0] = (k00_20)*A[0] - (k01_21) * A[1];
+ e2[-1] = (k01_21)*A[0] + (k00_20) * A[1];
- A += k1;
+ A += k1;
- k00_20 = e0[-2] - e2[-2];
- k01_21 = e0[-3] - e2[-3];
- e0[-2] += e2[-2];//e0[-2] = e0[-2] + e2[-2];
- e0[-3] += e2[-3];//e0[-3] = e0[-3] + e2[-3];
- e2[-2] = (k00_20)*A[0] - (k01_21)* A[1];
- e2[-3] = (k01_21)*A[0] + (k00_20)* A[1];
+ k00_20 = e0[-2] - e2[-2];
+ k01_21 = e0[-3] - e2[-3];
+ e0[-2] += e2[-2];//e0[-2] = e0[-2] + e2[-2];
+ e0[-3] += e2[-3];//e0[-3] = e0[-3] + e2[-3];
+ e2[-2] = (k00_20)*A[0] - (k01_21) * A[1];
+ e2[-3] = (k01_21)*A[0] + (k00_20) * A[1];
- A += k1;
+ A += k1;
- k00_20 = e0[-4] - e2[-4];
- k01_21 = e0[-5] - e2[-5];
- e0[-4] += e2[-4];//e0[-4] = e0[-4] + e2[-4];
- e0[-5] += e2[-5];//e0[-5] = e0[-5] + e2[-5];
- e2[-4] = (k00_20)*A[0] - (k01_21)* A[1];
- e2[-5] = (k01_21)*A[0] + (k00_20)* A[1];
+ k00_20 = e0[-4] - e2[-4];
+ k01_21 = e0[-5] - e2[-5];
+ e0[-4] += e2[-4];//e0[-4] = e0[-4] + e2[-4];
+ e0[-5] += e2[-5];//e0[-5] = e0[-5] + e2[-5];
+ e2[-4] = (k00_20)*A[0] - (k01_21) * A[1];
+ e2[-5] = (k01_21)*A[0] + (k00_20) * A[1];
- A += k1;
+ A += k1;
- k00_20 = e0[-6] - e2[-6];
- k01_21 = e0[-7] - e2[-7];
- e0[-6] += e2[-6];//e0[-6] = e0[-6] + e2[-6];
- e0[-7] += e2[-7];//e0[-7] = e0[-7] + e2[-7];
- e2[-6] = (k00_20)*A[0] - (k01_21)* A[1];
- e2[-7] = (k01_21)*A[0] + (k00_20)* A[1];
+ k00_20 = e0[-6] - e2[-6];
+ k01_21 = e0[-7] - e2[-7];
+ e0[-6] += e2[-6];//e0[-6] = e0[-6] + e2[-6];
+ e0[-7] += e2[-7];//e0[-7] = e0[-7] + e2[-7];
+ e2[-6] = (k00_20)*A[0] - (k01_21) * A[1];
+ e2[-7] = (k01_21)*A[0] + (k00_20) * A[1];
- e0 -= 8;
- e2 -= 8;
+ e0 -= 8;
+ e2 -= 8;
- A += k1;
- }
+ A += k1;
+ }
}
static void imdct_step3_inner_s_loop(int n, float *e, int i_off, int k_off, float *A, int a_off, int k0)
{
- int i;
- float A0 = A[0];
- float A1 = A[0 + 1];
- float A2 = A[0 + a_off];
- float A3 = A[0 + a_off + 1];
- float A4 = A[0 + a_off * 2 + 0];
- float A5 = A[0 + a_off * 2 + 1];
- float A6 = A[0 + a_off * 3 + 0];
- float A7 = A[0 + a_off * 3 + 1];
-
- float k00, k11;
-
- float *ee0 = e + i_off;
- float *ee2 = ee0 + k_off;
-
- for (i = n; i > 0; --i) {
- k00 = ee0[0] - ee2[0];
- k11 = ee0[-1] - ee2[-1];
- ee0[0] = ee0[0] + ee2[0];
- ee0[-1] = ee0[-1] + ee2[-1];
- ee2[0] = (k00)* A0 - (k11)* A1;
- ee2[-1] = (k11)* A0 + (k00)* A1;
-
- k00 = ee0[-2] - ee2[-2];
- k11 = ee0[-3] - ee2[-3];
- ee0[-2] = ee0[-2] + ee2[-2];
- ee0[-3] = ee0[-3] + ee2[-3];
- ee2[-2] = (k00)* A2 - (k11)* A3;
- ee2[-3] = (k11)* A2 + (k00)* A3;
-
- k00 = ee0[-4] - ee2[-4];
- k11 = ee0[-5] - ee2[-5];
- ee0[-4] = ee0[-4] + ee2[-4];
- ee0[-5] = ee0[-5] + ee2[-5];
- ee2[-4] = (k00)* A4 - (k11)* A5;
- ee2[-5] = (k11)* A4 + (k00)* A5;
-
- k00 = ee0[-6] - ee2[-6];
- k11 = ee0[-7] - ee2[-7];
- ee0[-6] = ee0[-6] + ee2[-6];
- ee0[-7] = ee0[-7] + ee2[-7];
- ee2[-6] = (k00)* A6 - (k11)* A7;
- ee2[-7] = (k11)* A6 + (k00)* A7;
-
- ee0 -= k0;
- ee2 -= k0;
- }
+ int i;
+ float A0 = A[0];
+ float A1 = A[0+1];
+ float A2 = A[0+a_off];
+ float A3 = A[0+a_off+1];
+ float A4 = A[0+a_off*2+0];
+ float A5 = A[0+a_off*2+1];
+ float A6 = A[0+a_off*3+0];
+ float A7 = A[0+a_off*3+1];
+
+ float k00,k11;
+
+ float *ee0 = e +i_off;
+ float *ee2 = ee0+k_off;
+
+ for (i=n; i > 0; --i) {
+ k00 = ee0[ 0] - ee2[ 0];
+ k11 = ee0[-1] - ee2[-1];
+ ee0[ 0] = ee0[ 0] + ee2[ 0];
+ ee0[-1] = ee0[-1] + ee2[-1];
+ ee2[ 0] = (k00) * A0 - (k11) * A1;
+ ee2[-1] = (k11) * A0 + (k00) * A1;
+
+ k00 = ee0[-2] - ee2[-2];
+ k11 = ee0[-3] - ee2[-3];
+ ee0[-2] = ee0[-2] + ee2[-2];
+ ee0[-3] = ee0[-3] + ee2[-3];
+ ee2[-2] = (k00) * A2 - (k11) * A3;
+ ee2[-3] = (k11) * A2 + (k00) * A3;
+
+ k00 = ee0[-4] - ee2[-4];
+ k11 = ee0[-5] - ee2[-5];
+ ee0[-4] = ee0[-4] + ee2[-4];
+ ee0[-5] = ee0[-5] + ee2[-5];
+ ee2[-4] = (k00) * A4 - (k11) * A5;
+ ee2[-5] = (k11) * A4 + (k00) * A5;
+
+ k00 = ee0[-6] - ee2[-6];
+ k11 = ee0[-7] - ee2[-7];
+ ee0[-6] = ee0[-6] + ee2[-6];
+ ee0[-7] = ee0[-7] + ee2[-7];
+ ee2[-6] = (k00) * A6 - (k11) * A7;
+ ee2[-7] = (k11) * A6 + (k00) * A7;
+
+ ee0 -= k0;
+ ee2 -= k0;
+ }
}
static __forceinline void iter_54(float *z)
{
- float k00, k11, k22, k33;
- float y0, y1, y2, y3;
+ float k00,k11,k22,k33;
+ float y0,y1,y2,y3;
- k00 = z[0] - z[-4];
- y0 = z[0] + z[-4];
- y2 = z[-2] + z[-6];
- k22 = z[-2] - z[-6];
+ k00 = z[ 0] - z[-4];
+ y0 = z[ 0] + z[-4];
+ y2 = z[-2] + z[-6];
+ k22 = z[-2] - z[-6];
- z[-0] = y0 + y2; // z0 + z4 + z2 + z6
- z[-2] = y0 - y2; // z0 + z4 - z2 - z6
+ z[-0] = y0 + y2; // z0 + z4 + z2 + z6
+ z[-2] = y0 - y2; // z0 + z4 - z2 - z6
- // done with y0,y2
+ // done with y0,y2
- k33 = z[-3] - z[-7];
+ k33 = z[-3] - z[-7];
- z[-4] = k00 + k33; // z0 - z4 + z3 - z7
- z[-6] = k00 - k33; // z0 - z4 - z3 + z7
+ z[-4] = k00 + k33; // z0 - z4 + z3 - z7
+ z[-6] = k00 - k33; // z0 - z4 - z3 + z7
- // done with k33
+ // done with k33
- k11 = z[-1] - z[-5];
- y1 = z[-1] + z[-5];
- y3 = z[-3] + z[-7];
+ k11 = z[-1] - z[-5];
+ y1 = z[-1] + z[-5];
+ y3 = z[-3] + z[-7];
- z[-1] = y1 + y3; // z1 + z5 + z3 + z7
- z[-3] = y1 - y3; // z1 + z5 - z3 - z7
- z[-5] = k11 - k22; // z1 - z5 + z2 - z6
- z[-7] = k11 + k22; // z1 - z5 - z2 + z6
+ z[-1] = y1 + y3; // z1 + z5 + z3 + z7
+ z[-3] = y1 - y3; // z1 + z5 - z3 - z7
+ z[-5] = k11 - k22; // z1 - z5 + z2 - z6
+ z[-7] = k11 + k22; // z1 - z5 - z2 + z6
}
static void imdct_step3_inner_s_loop_ld654(int n, float *e, int i_off, float *A, int base_n)
{
- int a_off = base_n >> 3;
- float A2 = A[0 + a_off];
- float *z = e + i_off;
- float *base = z - 16 * n;
-
- while (z > base) {
- float k00, k11;
-
- k00 = z[-0] - z[-8];
- k11 = z[-1] - z[-9];
- z[-0] = z[-0] + z[-8];
- z[-1] = z[-1] + z[-9];
- z[-8] = k00;
- z[-9] = k11;
-
- k00 = z[-2] - z[-10];
- k11 = z[-3] - z[-11];
- z[-2] = z[-2] + z[-10];
- z[-3] = z[-3] + z[-11];
- z[-10] = (k00 + k11) * A2;
- z[-11] = (k11 - k00) * A2;
-
- k00 = z[-12] - z[-4]; // reverse to avoid a unary negation
- k11 = z[-5] - z[-13];
- z[-4] = z[-4] + z[-12];
- z[-5] = z[-5] + z[-13];
- z[-12] = k11;
- z[-13] = k00;
-
- k00 = z[-14] - z[-6]; // reverse to avoid a unary negation
- k11 = z[-7] - z[-15];
- z[-6] = z[-6] + z[-14];
- z[-7] = z[-7] + z[-15];
- z[-14] = (k00 + k11) * A2;
- z[-15] = (k00 - k11) * A2;
-
- iter_54(z);
- iter_54(z - 8);
- z -= 16;
- }
+ int a_off = base_n >> 3;
+ float A2 = A[0+a_off];
+ float *z = e + i_off;
+ float *base = z - 16 * n;
+
+ while (z > base) {
+ float k00,k11;
+ float l00,l11;
+
+ k00 = z[-0] - z[ -8];
+ k11 = z[-1] - z[ -9];
+ l00 = z[-2] - z[-10];
+ l11 = z[-3] - z[-11];
+ z[ -0] = z[-0] + z[ -8];
+ z[ -1] = z[-1] + z[ -9];
+ z[ -2] = z[-2] + z[-10];
+ z[ -3] = z[-3] + z[-11];
+ z[ -8] = k00;
+ z[ -9] = k11;
+ z[-10] = (l00+l11) * A2;
+ z[-11] = (l11-l00) * A2;
+
+ k00 = z[ -4] - z[-12];
+ k11 = z[ -5] - z[-13];
+ l00 = z[ -6] - z[-14];
+ l11 = z[ -7] - z[-15];
+ z[ -4] = z[ -4] + z[-12];
+ z[ -5] = z[ -5] + z[-13];
+ z[ -6] = z[ -6] + z[-14];
+ z[ -7] = z[ -7] + z[-15];
+ z[-12] = k11;
+ z[-13] = -k00;
+ z[-14] = (l11-l00) * A2;
+ z[-15] = (l00+l11) * -A2;
+
+ iter_54(z);
+ iter_54(z-8);
+ z -= 16;
+ }
}
static void inverse_mdct(float *buffer, int n, vorb *f, int blocktype)
{
- int n2 = n >> 1, n4 = n >> 2, n8 = n >> 3, l;
- int ld;
- // @OPTIMIZE: reduce register pressure by using fewer variables?
- int save_point = temp_alloc_save(f);
- float *buf2 = (float *)temp_alloc(f, n2 * sizeof(*buf2));
- float *u = NULL, *v = NULL;
- // twiddle factors
- float *A = f->A[blocktype];
-
- // IMDCT algorithm from "The use of multirate filter banks for coding of high quality digital audio"
- // See notes about bugs in that paper in less-optimal implementation 'inverse_mdct_old' after this function.
-
- // kernel from paper
-
-
- // merged:
- // copy and reflect spectral data
- // step 0
-
- // note that it turns out that the items added together during
- // this step are, in fact, being added to themselves (as reflected
- // by step 0). inexplicable inefficiency! this became obvious
- // once I combined the passes.
-
- // so there's a missing 'times 2' here (for adding X to itself).
- // this propogates through linearly to the end, where the numbers
- // are 1/2 too small, and need to be compensated for.
-
- {
- float *d, *e, *AA, *e_stop;
- d = &buf2[n2 - 2];
- AA = A;
- e = &buffer[0];
- e_stop = &buffer[n2];
- while (e != e_stop) {
- d[1] = (e[0] * AA[0] - e[2] * AA[1]);
- d[0] = (e[0] * AA[1] + e[2] * AA[0]);
- d -= 2;
- AA += 2;
- e += 4;
- }
-
- e = &buffer[n2 - 3];
- while (d >= buf2) {
- d[1] = (-e[2] * AA[0] - -e[0] * AA[1]);
- d[0] = (-e[2] * AA[1] + -e[0] * AA[0]);
- d -= 2;
- AA += 2;
- e -= 4;
- }
- }
-
- // now we use symbolic names for these, so that we can
- // possibly swap their meaning as we change which operations
- // are in place
-
- u = buffer;
- v = buf2;
-
- // step 2 (paper output is w, now u)
- // this could be in place, but the data ends up in the wrong
- // place... _somebody_'s got to swap it, so this is nominated
- {
- float *AA = &A[n2 - 8];
- float *d0, *d1, *e0, *e1;
-
- e0 = &v[n4];
- e1 = &v[0];
-
- d0 = &u[n4];
- d1 = &u[0];
-
- while (AA >= A) {
- float v40_20, v41_21;
-
- v41_21 = e0[1] - e1[1];
- v40_20 = e0[0] - e1[0];
- d0[1] = e0[1] + e1[1];
- d0[0] = e0[0] + e1[0];
- d1[1] = v41_21*AA[4] - v40_20*AA[5];
- d1[0] = v40_20*AA[4] + v41_21*AA[5];
-
- v41_21 = e0[3] - e1[3];
- v40_20 = e0[2] - e1[2];
- d0[3] = e0[3] + e1[3];
- d0[2] = e0[2] + e1[2];
- d1[3] = v41_21*AA[0] - v40_20*AA[1];
- d1[2] = v40_20*AA[0] + v41_21*AA[1];
-
- AA -= 8;
-
- d0 += 4;
- d1 += 4;
- e0 += 4;
- e1 += 4;
- }
- }
-
- // step 3
- ld = ilog(n) - 1; // ilog is off-by-one from normal definitions
-
- // optimized step 3:
-
- // the original step3 loop can be nested r inside s or s inside r;
- // it's written originally as s inside r, but this is dumb when r
- // iterates many times, and s few. So I have two copies of it and
- // switch between them halfway.
-
- // this is iteration 0 of step 3
- imdct_step3_iter0_loop(n >> 4, u, n2 - 1 - n4 * 0, -(n >> 3), A);
- imdct_step3_iter0_loop(n >> 4, u, n2 - 1 - n4 * 1, -(n >> 3), A);
-
- // this is iteration 1 of step 3
- imdct_step3_inner_r_loop(n >> 5, u, n2 - 1 - n8 * 0, -(n >> 4), A, 16);
- imdct_step3_inner_r_loop(n >> 5, u, n2 - 1 - n8 * 1, -(n >> 4), A, 16);
- imdct_step3_inner_r_loop(n >> 5, u, n2 - 1 - n8 * 2, -(n >> 4), A, 16);
- imdct_step3_inner_r_loop(n >> 5, u, n2 - 1 - n8 * 3, -(n >> 4), A, 16);
-
- l = 2;
- for (; l < (ld - 3) >> 1; ++l) {
- int k0 = n >> (l + 2), k0_2 = k0 >> 1;
- int lim = 1 << (l + 1);
- int i;
- for (i = 0; i < lim; ++i)
- imdct_step3_inner_r_loop(n >> (l + 4), u, n2 - 1 - k0*i, -k0_2, A, 1 << (l + 3));
- }
-
- for (; l < ld - 6; ++l) {
- int k0 = n >> (l + 2), k1 = 1 << (l + 3), k0_2 = k0 >> 1;
- int rlim = n >> (l + 6), r;
- int lim = 1 << (l + 1);
- int i_off;
- float *A0 = A;
- i_off = n2 - 1;
- for (r = rlim; r > 0; --r) {
- imdct_step3_inner_s_loop(lim, u, i_off, -k0_2, A0, k1, k0);
- A0 += k1 * 4;
- i_off -= 8;
- }
- }
-
- // iterations with count:
- // ld-6,-5,-4 all interleaved together
- // the big win comes from getting rid of needless flops
- // due to the constants on pass 5 & 4 being all 1 and 0;
- // combining them to be simultaneous to improve cache made little difference
- imdct_step3_inner_s_loop_ld654(n >> 5, u, n2 - 1, A, n);
-
- // output is u
-
- // step 4, 5, and 6
- // cannot be in-place because of step 5
- {
- uint16 *bitrev = f->bit_reverse[blocktype];
- // weirdly, I'd have thought reading sequentially and writing
- // erratically would have been better than vice-versa, but in
- // fact that's not what my testing showed. (That is, with
- // j = bitreverse(i), do you read i and write j, or read j and write i.)
-
- float *d0 = &v[n4 - 4];
- float *d1 = &v[n2 - 4];
- while (d0 >= v) {
- int k4;
-
- k4 = bitrev[0];
- d1[3] = u[k4 + 0];
- d1[2] = u[k4 + 1];
- d0[3] = u[k4 + 2];
- d0[2] = u[k4 + 3];
-
- k4 = bitrev[1];
- d1[1] = u[k4 + 0];
- d1[0] = u[k4 + 1];
- d0[1] = u[k4 + 2];
- d0[0] = u[k4 + 3];
-
- d0 -= 4;
- d1 -= 4;
- bitrev += 2;
- }
- }
- // (paper output is u, now v)
-
-
- // data must be in buf2
- assert(v == buf2);
-
- // step 7 (paper output is v, now v)
- // this is now in place
- {
- float *C = f->C[blocktype];
- float *d, *e;
-
- d = v;
- e = v + n2 - 4;
-
- while (d < e) {
- float a02, a11, b0, b1, b2, b3;
-
- a02 = d[0] - e[2];
- a11 = d[1] + e[3];
-
- b0 = C[1] * a02 + C[0] * a11;
- b1 = C[1] * a11 - C[0] * a02;
-
- b2 = d[0] + e[2];
- b3 = d[1] - e[3];
-
- d[0] = b2 + b0;
- d[1] = b3 + b1;
- e[2] = b2 - b0;
- e[3] = b1 - b3;
-
- a02 = d[2] - e[0];
- a11 = d[3] + e[1];
-
- b0 = C[3] * a02 + C[2] * a11;
- b1 = C[3] * a11 - C[2] * a02;
-
- b2 = d[2] + e[0];
- b3 = d[3] - e[1];
-
- d[2] = b2 + b0;
- d[3] = b3 + b1;
- e[0] = b2 - b0;
- e[1] = b1 - b3;
-
- C += 4;
- d += 4;
- e -= 4;
- }
- }
-
- // data must be in buf2
-
-
- // step 8+decode (paper output is X, now buffer)
- // this generates pairs of data a la 8 and pushes them directly through
- // the decode kernel (pushing rather than pulling) to avoid having
- // to make another pass later
-
- // this cannot POSSIBLY be in place, so we refer to the buffers directly
-
- {
- float *d0, *d1, *d2, *d3;
-
- float *B = f->B[blocktype] + n2 - 8;
- float *e = buf2 + n2 - 8;
- d0 = &buffer[0];
- d1 = &buffer[n2 - 4];
- d2 = &buffer[n2];
- d3 = &buffer[n - 4];
- while (e >= v) {
- float p0, p1, p2, p3;
-
- p3 = e[6] * B[7] - e[7] * B[6];
- p2 = -e[6] * B[6] - e[7] * B[7];
-
- d0[0] = p3;
- d1[3] = -p3;
- d2[0] = p2;
- d3[3] = p2;
-
- p1 = e[4] * B[5] - e[5] * B[4];
- p0 = -e[4] * B[4] - e[5] * B[5];
-
- d0[1] = p1;
- d1[2] = -p1;
- d2[1] = p0;
- d3[2] = p0;
-
- p3 = e[2] * B[3] - e[3] * B[2];
- p2 = -e[2] * B[2] - e[3] * B[3];
+ int n2 = n >> 1, n4 = n >> 2, n8 = n >> 3, l;
+ int ld;
+ // @OPTIMIZE: reduce register pressure by using fewer variables?
+ int save_point = temp_alloc_save(f);
+ float *buf2 = (float *) temp_alloc(f, n2 * sizeof(*buf2));
+ float *u=NULL,*v=NULL;
+ // twiddle factors
+ float *A = f->A[blocktype];
+
+ // IMDCT algorithm from "The use of multirate filter banks for coding of high quality digital audio"
+ // See notes about bugs in that paper in less-optimal implementation 'inverse_mdct_old' after this function.
+
+ // kernel from paper
+
+
+ // merged:
+ // copy and reflect spectral data
+ // step 0
+
+ // note that it turns out that the items added together during
+ // this step are, in fact, being added to themselves (as reflected
+ // by step 0). inexplicable inefficiency! this became obvious
+ // once I combined the passes.
+
+ // so there's a missing 'times 2' here (for adding X to itself).
+ // this propagates through linearly to the end, where the numbers
+ // are 1/2 too small, and need to be compensated for.
+
+ {
+ float *d,*e, *AA, *e_stop;
+ d = &buf2[n2-2];
+ AA = A;
+ e = &buffer[0];
+ e_stop = &buffer[n2];
+ while (e != e_stop) {
+ d[1] = (e[0] * AA[0] - e[2]*AA[1]);
+ d[0] = (e[0] * AA[1] + e[2]*AA[0]);
+ d -= 2;
+ AA += 2;
+ e += 4;
+ }
+
+ e = &buffer[n2-3];
+ while (d >= buf2) {
+ d[1] = (-e[2] * AA[0] - -e[0]*AA[1]);
+ d[0] = (-e[2] * AA[1] + -e[0]*AA[0]);
+ d -= 2;
+ AA += 2;
+ e -= 4;
+ }
+ }
+
+ // now we use symbolic names for these, so that we can
+ // possibly swap their meaning as we change which operations
+ // are in place
+
+ u = buffer;
+ v = buf2;
+
+ // step 2 (paper output is w, now u)
+ // this could be in place, but the data ends up in the wrong
+ // place... _somebody_'s got to swap it, so this is nominated
+ {
+ float *AA = &A[n2-8];
+ float *d0,*d1, *e0, *e1;
+
+ e0 = &v[n4];
+ e1 = &v[0];
+
+ d0 = &u[n4];
+ d1 = &u[0];
+
+ while (AA >= A) {
+ float v40_20, v41_21;
+
+ v41_21 = e0[1] - e1[1];
+ v40_20 = e0[0] - e1[0];
+ d0[1] = e0[1] + e1[1];
+ d0[0] = e0[0] + e1[0];
+ d1[1] = v41_21*AA[4] - v40_20*AA[5];
+ d1[0] = v40_20*AA[4] + v41_21*AA[5];
+
+ v41_21 = e0[3] - e1[3];
+ v40_20 = e0[2] - e1[2];
+ d0[3] = e0[3] + e1[3];
+ d0[2] = e0[2] + e1[2];
+ d1[3] = v41_21*AA[0] - v40_20*AA[1];
+ d1[2] = v40_20*AA[0] + v41_21*AA[1];
+
+ AA -= 8;
+
+ d0 += 4;
+ d1 += 4;
+ e0 += 4;
+ e1 += 4;
+ }
+ }
+
+ // step 3
+ ld = ilog(n) - 1; // ilog is off-by-one from normal definitions
+
+ // optimized step 3:
+
+ // the original step3 loop can be nested r inside s or s inside r;
+ // it's written originally as s inside r, but this is dumb when r
+ // iterates many times, and s few. So I have two copies of it and
+ // switch between them halfway.
+
+ // this is iteration 0 of step 3
+ imdct_step3_iter0_loop(n >> 4, u, n2-1-n4*0, -(n >> 3), A);
+ imdct_step3_iter0_loop(n >> 4, u, n2-1-n4*1, -(n >> 3), A);
+
+ // this is iteration 1 of step 3
+ imdct_step3_inner_r_loop(n >> 5, u, n2-1 - n8*0, -(n >> 4), A, 16);
+ imdct_step3_inner_r_loop(n >> 5, u, n2-1 - n8*1, -(n >> 4), A, 16);
+ imdct_step3_inner_r_loop(n >> 5, u, n2-1 - n8*2, -(n >> 4), A, 16);
+ imdct_step3_inner_r_loop(n >> 5, u, n2-1 - n8*3, -(n >> 4), A, 16);
+
+ l=2;
+ for (; l < (ld-3)>>1; ++l) {
+ int k0 = n >> (l+2), k0_2 = k0>>1;
+ int lim = 1 << (l+1);
+ int i;
+ for (i=0; i < lim; ++i)
+ imdct_step3_inner_r_loop(n >> (l+4), u, n2-1 - k0*i, -k0_2, A, 1 << (l+3));
+ }
+
+ for (; l < ld-6; ++l) {
+ int k0 = n >> (l+2), k1 = 1 << (l+3), k0_2 = k0>>1;
+ int rlim = n >> (l+6), r;
+ int lim = 1 << (l+1);
+ int i_off;
+ float *A0 = A;
+ i_off = n2-1;
+ for (r=rlim; r > 0; --r) {
+ imdct_step3_inner_s_loop(lim, u, i_off, -k0_2, A0, k1, k0);
+ A0 += k1*4;
+ i_off -= 8;
+ }
+ }
+
+ // iterations with count:
+ // ld-6,-5,-4 all interleaved together
+ // the big win comes from getting rid of needless flops
+ // due to the constants on pass 5 & 4 being all 1 and 0;
+ // combining them to be simultaneous to improve cache made little difference
+ imdct_step3_inner_s_loop_ld654(n >> 5, u, n2-1, A, n);
+
+ // output is u
+
+ // step 4, 5, and 6
+ // cannot be in-place because of step 5
+ {
+ uint16 *bitrev = f->bit_reverse[blocktype];
+ // weirdly, I'd have thought reading sequentially and writing
+ // erratically would have been better than vice-versa, but in
+ // fact that's not what my testing showed. (That is, with
+ // j = bitreverse(i), do you read i and write j, or read j and write i.)
+
+ float *d0 = &v[n4-4];
+ float *d1 = &v[n2-4];
+ while (d0 >= v) {
+ int k4;
+
+ k4 = bitrev[0];
+ d1[3] = u[k4+0];
+ d1[2] = u[k4+1];
+ d0[3] = u[k4+2];
+ d0[2] = u[k4+3];
+
+ k4 = bitrev[1];
+ d1[1] = u[k4+0];
+ d1[0] = u[k4+1];
+ d0[1] = u[k4+2];
+ d0[0] = u[k4+3];
+
+ d0 -= 4;
+ d1 -= 4;
+ bitrev += 2;
+ }
+ }
+ // (paper output is u, now v)
+
+
+ // data must be in buf2
+ assert(v == buf2);
+
+ // step 7 (paper output is v, now v)
+ // this is now in place
+ {
+ float *C = f->C[blocktype];
+ float *d, *e;
+
+ d = v;
+ e = v + n2 - 4;
+
+ while (d < e) {
+ float a02,a11,b0,b1,b2,b3;
+
+ a02 = d[0] - e[2];
+ a11 = d[1] + e[3];
+
+ b0 = C[1]*a02 + C[0]*a11;
+ b1 = C[1]*a11 - C[0]*a02;
+
+ b2 = d[0] + e[ 2];
+ b3 = d[1] - e[ 3];
+
+ d[0] = b2 + b0;
+ d[1] = b3 + b1;
+ e[2] = b2 - b0;
+ e[3] = b1 - b3;
+
+ a02 = d[2] - e[0];
+ a11 = d[3] + e[1];
+
+ b0 = C[3]*a02 + C[2]*a11;
+ b1 = C[3]*a11 - C[2]*a02;
+
+ b2 = d[2] + e[ 0];
+ b3 = d[3] - e[ 1];
+
+ d[2] = b2 + b0;
+ d[3] = b3 + b1;
+ e[0] = b2 - b0;
+ e[1] = b1 - b3;
+
+ C += 4;
+ d += 4;
+ e -= 4;
+ }
+ }
+
+ // data must be in buf2
+
+
+ // step 8+decode (paper output is X, now buffer)
+ // this generates pairs of data a la 8 and pushes them directly through
+ // the decode kernel (pushing rather than pulling) to avoid having
+ // to make another pass later
+
+ // this cannot POSSIBLY be in place, so we refer to the buffers directly
+
+ {
+ float *d0,*d1,*d2,*d3;
+
+ float *B = f->B[blocktype] + n2 - 8;
+ float *e = buf2 + n2 - 8;
+ d0 = &buffer[0];
+ d1 = &buffer[n2-4];
+ d2 = &buffer[n2];
+ d3 = &buffer[n-4];
+ while (e >= v) {
+ float p0,p1,p2,p3;
+
+ p3 = e[6]*B[7] - e[7]*B[6];
+ p2 = -e[6]*B[6] - e[7]*B[7];
+
+ d0[0] = p3;
+ d1[3] = - p3;
+ d2[0] = p2;
+ d3[3] = p2;
+
+ p1 = e[4]*B[5] - e[5]*B[4];
+ p0 = -e[4]*B[4] - e[5]*B[5];
+
+ d0[1] = p1;
+ d1[2] = - p1;
+ d2[1] = p0;
+ d3[2] = p0;
- d0[2] = p3;
- d1[1] = -p3;
- d2[2] = p2;
- d3[1] = p2;
+ p3 = e[2]*B[3] - e[3]*B[2];
+ p2 = -e[2]*B[2] - e[3]*B[3];
- p1 = e[0] * B[1] - e[1] * B[0];
- p0 = -e[0] * B[0] - e[1] * B[1];
+ d0[2] = p3;
+ d1[1] = - p3;
+ d2[2] = p2;
+ d3[1] = p2;
- d0[3] = p1;
- d1[0] = -p1;
- d2[3] = p0;
- d3[0] = p0;
+ p1 = e[0]*B[1] - e[1]*B[0];
+ p0 = -e[0]*B[0] - e[1]*B[1];
- B -= 8;
- e -= 8;
- d0 += 4;
- d2 += 4;
- d1 -= 4;
- d3 -= 4;
- }
- }
+ d0[3] = p1;
+ d1[0] = - p1;
+ d2[3] = p0;
+ d3[0] = p0;
- temp_free(f, buf2);
- temp_alloc_restore(f, save_point);
+ B -= 8;
+ e -= 8;
+ d0 += 4;
+ d2 += 4;
+ d1 -= 4;
+ d3 -= 4;
+ }
+ }
+
+ temp_free(f,buf2);
+ temp_alloc_restore(f,save_point);
}
#if 0
// this is the original version of the above code, if you want to optimize it from scratch
void inverse_mdct_naive(float *buffer, int n)
{
- float s;
- float A[1 << 12], B[1 << 12], C[1 << 11];
- int i, k, k2, k4, n2 = n >> 1, n4 = n >> 2, n8 = n >> 3, l;
- int n3_4 = n - n4, ld;
- // how can they claim this only uses N words?!
- // oh, because they're only used sparsely, whoops
- float u[1 << 13], X[1 << 13], v[1 << 13], w[1 << 13];
- // set up twiddle factors
-
- for (k = k2 = 0; k < n4; ++k, k2 += 2) {
- A[k2] = (float)cos(4 * k*M_PI / n);
- A[k2 + 1] = (float)-sin(4 * k*M_PI / n);
- B[k2] = (float)cos((k2 + 1)*M_PI / n / 2);
- B[k2 + 1] = (float)sin((k2 + 1)*M_PI / n / 2);
- }
- for (k = k2 = 0; k < n8; ++k, k2 += 2) {
- C[k2] = (float)cos(2 * (k2 + 1)*M_PI / n);
- C[k2 + 1] = (float)-sin(2 * (k2 + 1)*M_PI / n);
- }
-
- // IMDCT algorithm from "The use of multirate filter banks for coding of high quality digital audio"
- // Note there are bugs in that pseudocode, presumably due to them attempting
- // to rename the arrays nicely rather than representing the way their actual
- // implementation bounces buffers back and forth. As a result, even in the
- // "some formulars corrected" version, a direct implementation fails. These
- // are noted below as "paper bug".
-
- // copy and reflect spectral data
- for (k = 0; k < n2; ++k) u[k] = buffer[k];
- for (; k < n; ++k) u[k] = -buffer[n - k - 1];
- // kernel from paper
- // step 1
- for (k = k2 = k4 = 0; k < n4; k += 1, k2 += 2, k4 += 4) {
- v[n - k4 - 1] = (u[k4] - u[n - k4 - 1]) * A[k2] - (u[k4 + 2] - u[n - k4 - 3])*A[k2 + 1];
- v[n - k4 - 3] = (u[k4] - u[n - k4 - 1]) * A[k2 + 1] + (u[k4 + 2] - u[n - k4 - 3])*A[k2];
- }
- // step 2
- for (k = k4 = 0; k < n8; k += 1, k4 += 4) {
- w[n2 + 3 + k4] = v[n2 + 3 + k4] + v[k4 + 3];
- w[n2 + 1 + k4] = v[n2 + 1 + k4] + v[k4 + 1];
- w[k4 + 3] = (v[n2 + 3 + k4] - v[k4 + 3])*A[n2 - 4 - k4] - (v[n2 + 1 + k4] - v[k4 + 1])*A[n2 - 3 - k4];
- w[k4 + 1] = (v[n2 + 1 + k4] - v[k4 + 1])*A[n2 - 4 - k4] + (v[n2 + 3 + k4] - v[k4 + 3])*A[n2 - 3 - k4];
- }
- // step 3
- ld = ilog(n) - 1; // ilog is off-by-one from normal definitions
- for (l = 0; l < ld - 3; ++l) {
- int k0 = n >> (l + 2), k1 = 1 << (l + 3);
- int rlim = n >> (l + 4), r4, r;
- int s2lim = 1 << (l + 2), s2;
- for (r = r4 = 0; r < rlim; r4 += 4, ++r) {
- for (s2 = 0; s2 < s2lim; s2 += 2) {
- u[n - 1 - k0*s2 - r4] = w[n - 1 - k0*s2 - r4] + w[n - 1 - k0*(s2 + 1) - r4];
- u[n - 3 - k0*s2 - r4] = w[n - 3 - k0*s2 - r4] + w[n - 3 - k0*(s2 + 1) - r4];
- u[n - 1 - k0*(s2 + 1) - r4] = (w[n - 1 - k0*s2 - r4] - w[n - 1 - k0*(s2 + 1) - r4]) * A[r*k1]
- - (w[n - 3 - k0*s2 - r4] - w[n - 3 - k0*(s2 + 1) - r4]) * A[r*k1 + 1];
- u[n - 3 - k0*(s2 + 1) - r4] = (w[n - 3 - k0*s2 - r4] - w[n - 3 - k0*(s2 + 1) - r4]) * A[r*k1]
- + (w[n - 1 - k0*s2 - r4] - w[n - 1 - k0*(s2 + 1) - r4]) * A[r*k1 + 1];
- }
- }
- if (l + 1 < ld - 3) {
- // paper bug: ping-ponging of u&w here is omitted
- memcpy(w, u, sizeof(u));
- }
- }
-
- // step 4
- for (i = 0; i < n8; ++i) {
- int j = bit_reverse(i) >> (32 - ld + 3);
- assert(j < n8);
- if (i == j) {
- // paper bug: original code probably swapped in place; if copying,
- // need to directly copy in this case
- int i8 = i << 3;
- v[i8 + 1] = u[i8 + 1];
- v[i8 + 3] = u[i8 + 3];
- v[i8 + 5] = u[i8 + 5];
- v[i8 + 7] = u[i8 + 7];
- }
- else if (i < j) {
- int i8 = i << 3, j8 = j << 3;
- v[j8 + 1] = u[i8 + 1], v[i8 + 1] = u[j8 + 1];
- v[j8 + 3] = u[i8 + 3], v[i8 + 3] = u[j8 + 3];
- v[j8 + 5] = u[i8 + 5], v[i8 + 5] = u[j8 + 5];
- v[j8 + 7] = u[i8 + 7], v[i8 + 7] = u[j8 + 7];
- }
- }
- // step 5
- for (k = 0; k < n2; ++k) {
- w[k] = v[k * 2 + 1];
- }
- // step 6
- for (k = k2 = k4 = 0; k < n8; ++k, k2 += 2, k4 += 4) {
- u[n - 1 - k2] = w[k4];
- u[n - 2 - k2] = w[k4 + 1];
- u[n3_4 - 1 - k2] = w[k4 + 2];
- u[n3_4 - 2 - k2] = w[k4 + 3];
- }
- // step 7
- for (k = k2 = 0; k < n8; ++k, k2 += 2) {
- v[n2 + k2] = (u[n2 + k2] + u[n - 2 - k2] + C[k2 + 1] * (u[n2 + k2] - u[n - 2 - k2]) + C[k2] * (u[n2 + k2 + 1] + u[n - 2 - k2 + 1])) / 2;
- v[n - 2 - k2] = (u[n2 + k2] + u[n - 2 - k2] - C[k2 + 1] * (u[n2 + k2] - u[n - 2 - k2]) - C[k2] * (u[n2 + k2 + 1] + u[n - 2 - k2 + 1])) / 2;
- v[n2 + 1 + k2] = (u[n2 + 1 + k2] - u[n - 1 - k2] + C[k2 + 1] * (u[n2 + 1 + k2] + u[n - 1 - k2]) - C[k2] * (u[n2 + k2] - u[n - 2 - k2])) / 2;
- v[n - 1 - k2] = (-u[n2 + 1 + k2] + u[n - 1 - k2] + C[k2 + 1] * (u[n2 + 1 + k2] + u[n - 1 - k2]) - C[k2] * (u[n2 + k2] - u[n - 2 - k2])) / 2;
- }
- // step 8
- for (k = k2 = 0; k < n4; ++k, k2 += 2) {
- X[k] = v[k2 + n2] * B[k2] + v[k2 + 1 + n2] * B[k2 + 1];
- X[n2 - 1 - k] = v[k2 + n2] * B[k2 + 1] - v[k2 + 1 + n2] * B[k2];
- }
-
- // decode kernel to output
- // determined the following value experimentally
- // (by first figuring out what made inverse_mdct_slow work); then matching that here
- // (probably vorbis encoder premultiplies by n or n/2, to save it on the decoder?)
- s = 0.5; // theoretically would be n4
-
- // [[[ note! the s value of 0.5 is compensated for by the B[] in the current code,
- // so it needs to use the "old" B values to behave correctly, or else
- // set s to 1.0 ]]]
- for (i = 0; i < n4; ++i) buffer[i] = s * X[i + n4];
- for (; i < n3_4; ++i) buffer[i] = -s * X[n3_4 - i - 1];
- for (; i < n; ++i) buffer[i] = -s * X[i - n3_4];
+ float s;
+ float A[1 << 12], B[1 << 12], C[1 << 11];
+ int i,k,k2,k4, n2 = n >> 1, n4 = n >> 2, n8 = n >> 3, l;
+ int n3_4 = n - n4, ld;
+ // how can they claim this only uses N words?!
+ // oh, because they're only used sparsely, whoops
+ float u[1 << 13], X[1 << 13], v[1 << 13], w[1 << 13];
+ // set up twiddle factors
+
+ for (k=k2=0; k < n4; ++k,k2+=2) {
+ A[k2 ] = (float) cos(4*k*M_PI/n);
+ A[k2+1] = (float) -sin(4*k*M_PI/n);
+ B[k2 ] = (float) cos((k2+1)*M_PI/n/2);
+ B[k2+1] = (float) sin((k2+1)*M_PI/n/2);
+ }
+ for (k=k2=0; k < n8; ++k,k2+=2) {
+ C[k2 ] = (float) cos(2*(k2+1)*M_PI/n);
+ C[k2+1] = (float) -sin(2*(k2+1)*M_PI/n);
+ }
+
+ // IMDCT algorithm from "The use of multirate filter banks for coding of high quality digital audio"
+ // Note there are bugs in that pseudocode, presumably due to them attempting
+ // to rename the arrays nicely rather than representing the way their actual
+ // implementation bounces buffers back and forth. As a result, even in the
+ // "some formulars corrected" version, a direct implementation fails. These
+ // are noted below as "paper bug".
+
+ // copy and reflect spectral data
+ for (k=0; k < n2; ++k) u[k] = buffer[k];
+ for ( ; k < n ; ++k) u[k] = -buffer[n - k - 1];
+ // kernel from paper
+ // step 1
+ for (k=k2=k4=0; k < n4; k+=1, k2+=2, k4+=4) {
+ v[n-k4-1] = (u[k4] - u[n-k4-1]) * A[k2] - (u[k4+2] - u[n-k4-3])*A[k2+1];
+ v[n-k4-3] = (u[k4] - u[n-k4-1]) * A[k2+1] + (u[k4+2] - u[n-k4-3])*A[k2];
+ }
+ // step 2
+ for (k=k4=0; k < n8; k+=1, k4+=4) {
+ w[n2+3+k4] = v[n2+3+k4] + v[k4+3];
+ w[n2+1+k4] = v[n2+1+k4] + v[k4+1];
+ w[k4+3] = (v[n2+3+k4] - v[k4+3])*A[n2-4-k4] - (v[n2+1+k4]-v[k4+1])*A[n2-3-k4];
+ w[k4+1] = (v[n2+1+k4] - v[k4+1])*A[n2-4-k4] + (v[n2+3+k4]-v[k4+3])*A[n2-3-k4];
+ }
+ // step 3
+ ld = ilog(n) - 1; // ilog is off-by-one from normal definitions
+ for (l=0; l < ld-3; ++l) {
+ int k0 = n >> (l+2), k1 = 1 << (l+3);
+ int rlim = n >> (l+4), r4, r;
+ int s2lim = 1 << (l+2), s2;
+ for (r=r4=0; r < rlim; r4+=4,++r) {
+ for (s2=0; s2 < s2lim; s2+=2) {
+ u[n-1-k0*s2-r4] = w[n-1-k0*s2-r4] + w[n-1-k0*(s2+1)-r4];
+ u[n-3-k0*s2-r4] = w[n-3-k0*s2-r4] + w[n-3-k0*(s2+1)-r4];
+ u[n-1-k0*(s2+1)-r4] = (w[n-1-k0*s2-r4] - w[n-1-k0*(s2+1)-r4]) * A[r*k1]
+ - (w[n-3-k0*s2-r4] - w[n-3-k0*(s2+1)-r4]) * A[r*k1+1];
+ u[n-3-k0*(s2+1)-r4] = (w[n-3-k0*s2-r4] - w[n-3-k0*(s2+1)-r4]) * A[r*k1]
+ + (w[n-1-k0*s2-r4] - w[n-1-k0*(s2+1)-r4]) * A[r*k1+1];
+ }
+ }
+ if (l+1 < ld-3) {
+ // paper bug: ping-ponging of u&w here is omitted
+ memcpy(w, u, sizeof(u));
+ }
+ }
+
+ // step 4
+ for (i=0; i < n8; ++i) {
+ int j = bit_reverse(i) >> (32-ld+3);
+ assert(j < n8);
+ if (i == j) {
+ // paper bug: original code probably swapped in place; if copying,
+ // need to directly copy in this case
+ int i8 = i << 3;
+ v[i8+1] = u[i8+1];
+ v[i8+3] = u[i8+3];
+ v[i8+5] = u[i8+5];
+ v[i8+7] = u[i8+7];
+ } else if (i < j) {
+ int i8 = i << 3, j8 = j << 3;
+ v[j8+1] = u[i8+1], v[i8+1] = u[j8 + 1];
+ v[j8+3] = u[i8+3], v[i8+3] = u[j8 + 3];
+ v[j8+5] = u[i8+5], v[i8+5] = u[j8 + 5];
+ v[j8+7] = u[i8+7], v[i8+7] = u[j8 + 7];
+ }
+ }
+ // step 5
+ for (k=0; k < n2; ++k) {
+ w[k] = v[k*2+1];
+ }
+ // step 6
+ for (k=k2=k4=0; k < n8; ++k, k2 += 2, k4 += 4) {
+ u[n-1-k2] = w[k4];
+ u[n-2-k2] = w[k4+1];
+ u[n3_4 - 1 - k2] = w[k4+2];
+ u[n3_4 - 2 - k2] = w[k4+3];
+ }
+ // step 7
+ for (k=k2=0; k < n8; ++k, k2 += 2) {
+ v[n2 + k2 ] = ( u[n2 + k2] + u[n-2-k2] + C[k2+1]*(u[n2+k2]-u[n-2-k2]) + C[k2]*(u[n2+k2+1]+u[n-2-k2+1]))/2;
+ v[n-2 - k2] = ( u[n2 + k2] + u[n-2-k2] - C[k2+1]*(u[n2+k2]-u[n-2-k2]) - C[k2]*(u[n2+k2+1]+u[n-2-k2+1]))/2;
+ v[n2+1+ k2] = ( u[n2+1+k2] - u[n-1-k2] + C[k2+1]*(u[n2+1+k2]+u[n-1-k2]) - C[k2]*(u[n2+k2]-u[n-2-k2]))/2;
+ v[n-1 - k2] = (-u[n2+1+k2] + u[n-1-k2] + C[k2+1]*(u[n2+1+k2]+u[n-1-k2]) - C[k2]*(u[n2+k2]-u[n-2-k2]))/2;
+ }
+ // step 8
+ for (k=k2=0; k < n4; ++k,k2 += 2) {
+ X[k] = v[k2+n2]*B[k2 ] + v[k2+1+n2]*B[k2+1];
+ X[n2-1-k] = v[k2+n2]*B[k2+1] - v[k2+1+n2]*B[k2 ];
+ }
+
+ // decode kernel to output
+ // determined the following value experimentally
+ // (by first figuring out what made inverse_mdct_slow work); then matching that here
+ // (probably vorbis encoder premultiplies by n or n/2, to save it on the decoder?)
+ s = 0.5; // theoretically would be n4
+
+ // [[[ note! the s value of 0.5 is compensated for by the B[] in the current code,
+ // so it needs to use the "old" B values to behave correctly, or else
+ // set s to 1.0 ]]]
+ for (i=0; i < n4 ; ++i) buffer[i] = s * X[i+n4];
+ for ( ; i < n3_4; ++i) buffer[i] = -s * X[n3_4 - i - 1];
+ for ( ; i < n ; ++i) buffer[i] = -s * X[i - n3_4];
}
#endif
static float *get_window(vorb *f, int len)
{
- len <<= 1;
- if (len == f->blocksize_0) return f->window[0];
- if (len == f->blocksize_1) return f->window[1];
- assert(0);
- return NULL;
+ len <<= 1;
+ if (len == f->blocksize_0) return f->window[0];
+ if (len == f->blocksize_1) return f->window[1];
+ return NULL;
}
#ifndef STB_VORBIS_NO_DEFER_FLOOR
@@ -3076,40 +3071,40 @@ typedef int YTYPE;
#endif
static int do_floor(vorb *f, Mapping *map, int i, int n, float *target, YTYPE *finalY, uint8 *step2_flag)
{
- int n2 = n >> 1;
- int s = map->chan[i].mux, floor;
- floor = map->submap_floor[s];
- if (f->floor_types[floor] == 0) {
- return error(f, VORBIS_invalid_stream);
- }
- else {
- Floor1 *g = &f->floor_config[floor].floor1;
- int j, q;
- int lx = 0, ly = finalY[0] * g->floor1_multiplier;
- for (q = 1; q < g->values; ++q) {
- j = g->sorted_order[q];
-#ifndef STB_VORBIS_NO_DEFER_FLOOR
- if (finalY[j] >= 0)
-#else
- if (step2_flag[j])
-#endif
- {
- int hy = finalY[j] * g->floor1_multiplier;
- int hx = g->Xlist[j];
- if (lx != hx)
- draw_line(target, lx, ly, hx, hy, n2);
- CHECK(f);
- lx = hx, ly = hy;
- }
- }
- if (lx < n2) {
- // optimization of: draw_line(target, lx,ly, n,ly, n2);
- for (j = lx; j < n2; ++j)
- LINE_OP(target[j], inverse_db_table[ly]);
+ int n2 = n >> 1;
+ int s = map->chan[i].mux, floor;
+ floor = map->submap_floor[s];
+ if (f->floor_types[floor] == 0) {
+ return error(f, VORBIS_invalid_stream);
+ } else {
+ Floor1 *g = &f->floor_config[floor].floor1;
+ int j,q;
+ int lx = 0, ly = finalY[0] * g->floor1_multiplier;
+ for (q=1; q < g->values; ++q) {
+ j = g->sorted_order[q];
+ #ifndef STB_VORBIS_NO_DEFER_FLOOR
+ STBV_NOTUSED(step2_flag);
+ if (finalY[j] >= 0)
+ #else
+ if (step2_flag[j])
+ #endif
+ {
+ int hy = finalY[j] * g->floor1_multiplier;
+ int hx = g->Xlist[j];
+ if (lx != hx)
+ draw_line(target, lx,ly, hx,hy, n2);
CHECK(f);
- }
- }
- return TRUE;
+ lx = hx, ly = hy;
+ }
+ }
+ if (lx < n2) {
+ // optimization of: draw_line(target, lx,ly, n,ly, n2);
+ for (j=lx; j < n2; ++j)
+ LINE_OP(target[j], inverse_db_table[ly]);
+ CHECK(f);
+ }
+ }
+ return TRUE;
}
// The meaning of "left" and "right"
@@ -3128,1373 +3123,1434 @@ static int do_floor(vorb *f, Mapping *map, int i, int n, float *target, YTYPE *f
static int vorbis_decode_initial(vorb *f, int *p_left_start, int *p_left_end, int *p_right_start, int *p_right_end, int *mode)
{
- Mode *m;
- int i, n, prev, next, window_center;
- f->channel_buffer_start = f->channel_buffer_end = 0;
-
-retry:
- if (f->eof) return FALSE;
- if (!maybe_start_packet(f))
- return FALSE;
- // check packet type
- if (get_bits(f, 1) != 0) {
- if (IS_PUSH_MODE(f))
- return error(f, VORBIS_bad_packet_type);
- while (EOP != get8_packet(f));
- goto retry;
- }
-
- if (f->alloc.alloc_buffer)
- assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset);
-
- i = get_bits(f, ilog(f->mode_count - 1));
- if (i == EOP) return FALSE;
- if (i >= f->mode_count) return FALSE;
- *mode = i;
- m = f->mode_config + i;
- if (m->blockflag) {
- n = f->blocksize_1;
- prev = get_bits(f, 1);
- next = get_bits(f, 1);
- }
- else {
- prev = next = 0;
- n = f->blocksize_0;
- }
-
- // WINDOWING
-
- window_center = n >> 1;
- if (m->blockflag && !prev) {
- *p_left_start = (n - f->blocksize_0) >> 2;
- *p_left_end = (n + f->blocksize_0) >> 2;
- }
- else {
- *p_left_start = 0;
- *p_left_end = window_center;
- }
- if (m->blockflag && !next) {
- *p_right_start = (n * 3 - f->blocksize_0) >> 2;
- *p_right_end = (n * 3 + f->blocksize_0) >> 2;
- }
- else {
- *p_right_start = window_center;
- *p_right_end = n;
- }
-
- return TRUE;
+ Mode *m;
+ int i, n, prev, next, window_center;
+ f->channel_buffer_start = f->channel_buffer_end = 0;
+
+ retry:
+ if (f->eof) return FALSE;
+ if (!maybe_start_packet(f))
+ return FALSE;
+ // check packet type
+ if (get_bits(f,1) != 0) {
+ if (IS_PUSH_MODE(f))
+ return error(f,VORBIS_bad_packet_type);
+ while (EOP != get8_packet(f));
+ goto retry;
+ }
+
+ if (f->alloc.alloc_buffer)
+ assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset);
+
+ i = get_bits(f, ilog(f->mode_count-1));
+ if (i == EOP) return FALSE;
+ if (i >= f->mode_count) return FALSE;
+ *mode = i;
+ m = f->mode_config + i;
+ if (m->blockflag) {
+ n = f->blocksize_1;
+ prev = get_bits(f,1);
+ next = get_bits(f,1);
+ } else {
+ prev = next = 0;
+ n = f->blocksize_0;
+ }
+
+// WINDOWING
+
+ window_center = n >> 1;
+ if (m->blockflag && !prev) {
+ *p_left_start = (n - f->blocksize_0) >> 2;
+ *p_left_end = (n + f->blocksize_0) >> 2;
+ } else {
+ *p_left_start = 0;
+ *p_left_end = window_center;
+ }
+ if (m->blockflag && !next) {
+ *p_right_start = (n*3 - f->blocksize_0) >> 2;
+ *p_right_end = (n*3 + f->blocksize_0) >> 2;
+ } else {
+ *p_right_start = window_center;
+ *p_right_end = n;
+ }
+
+ return TRUE;
}
static int vorbis_decode_packet_rest(vorb *f, int *len, Mode *m, int left_start, int left_end, int right_start, int right_end, int *p_left)
{
- Mapping *map;
- int i, j, k, n, n2;
- int zero_channel[256];
- int really_zero_channel[256];
-
- // WINDOWING
-
- n = f->blocksize[m->blockflag];
- map = &f->mapping[m->mapping];
-
- // FLOORS
- n2 = n >> 1;
-
- CHECK(f);
-
- for (i = 0; i < f->channels; ++i) {
- int s = map->chan[i].mux, floor;
- zero_channel[i] = FALSE;
- floor = map->submap_floor[s];
- if (f->floor_types[floor] == 0) {
- return error(f, VORBIS_invalid_stream);
- }
- else {
- Floor1 *g = &f->floor_config[floor].floor1;
- if (get_bits(f, 1)) {
- short *finalY;
- uint8 step2_flag[256];
- static int range_list[4] = { 256, 128, 86, 64 };
- int range = range_list[g->floor1_multiplier - 1];
- int offset = 2;
- finalY = f->finalY[i];
- finalY[0] = get_bits(f, ilog(range) - 1);
- finalY[1] = get_bits(f, ilog(range) - 1);
- for (j = 0; j < g->partitions; ++j) {
- int pclass = g->partition_class_list[j];
- int cdim = g->class_dimensions[pclass];
- int cbits = g->class_subclasses[pclass];
- int csub = (1 << cbits) - 1;
- int cval = 0;
- if (cbits) {
- Codebook *c = f->codebooks + g->class_masterbooks[pclass];
- DECODE(cval, f, c);
- }
- for (k = 0; k < cdim; ++k) {
- int book = g->subclass_books[pclass][cval & csub];
- cval = cval >> cbits;
- if (book >= 0) {
- int temp;
- Codebook *c = f->codebooks + book;
- DECODE(temp, f, c);
- finalY[offset++] = temp;
- }
- else
- finalY[offset++] = 0;
- }
- }
- if (f->valid_bits == INVALID_BITS) goto error; // behavior according to spec
- step2_flag[0] = step2_flag[1] = 1;
- for (j = 2; j < g->values; ++j) {
- int low, high, pred, highroom, lowroom, room, val;
- low = g->neighbors[j][0];
- high = g->neighbors[j][1];
- //neighbors(g->Xlist, j, &low, &high);
- pred = predict_point(g->Xlist[j], g->Xlist[low], g->Xlist[high], finalY[low], finalY[high]);
- val = finalY[j];
- highroom = range - pred;
- lowroom = pred;
- if (highroom < lowroom)
- room = highroom * 2;
- else
- room = lowroom * 2;
- if (val) {
- step2_flag[low] = step2_flag[high] = 1;
- step2_flag[j] = 1;
- if (val >= room)
- if (highroom > lowroom)
- finalY[j] = val - lowroom + pred;
- else
- finalY[j] = pred - val + highroom - 1;
- else
- if (val & 1)
- finalY[j] = pred - ((val + 1) >> 1);
- else
- finalY[j] = pred + (val >> 1);
- }
- else {
- step2_flag[j] = 0;
- finalY[j] = pred;
- }
- }
+ Mapping *map;
+ int i,j,k,n,n2;
+ int zero_channel[256];
+ int really_zero_channel[256];
+
+// WINDOWING
+
+ STBV_NOTUSED(left_end);
+ n = f->blocksize[m->blockflag];
+ map = &f->mapping[m->mapping];
+
+// FLOORS
+ n2 = n >> 1;
+
+ CHECK(f);
+
+ for (i=0; i < f->channels; ++i) {
+ int s = map->chan[i].mux, floor;
+ zero_channel[i] = FALSE;
+ floor = map->submap_floor[s];
+ if (f->floor_types[floor] == 0) {
+ return error(f, VORBIS_invalid_stream);
+ } else {
+ Floor1 *g = &f->floor_config[floor].floor1;
+ if (get_bits(f, 1)) {
+ short *finalY;
+ uint8 step2_flag[256];
+ static int range_list[4] = { 256, 128, 86, 64 };
+ int range = range_list[g->floor1_multiplier-1];
+ int offset = 2;
+ finalY = f->finalY[i];
+ finalY[0] = get_bits(f, ilog(range)-1);
+ finalY[1] = get_bits(f, ilog(range)-1);
+ for (j=0; j < g->partitions; ++j) {
+ int pclass = g->partition_class_list[j];
+ int cdim = g->class_dimensions[pclass];
+ int cbits = g->class_subclasses[pclass];
+ int csub = (1 << cbits)-1;
+ int cval = 0;
+ if (cbits) {
+ Codebook *c = f->codebooks + g->class_masterbooks[pclass];
+ DECODE(cval,f,c);
+ }
+ for (k=0; k < cdim; ++k) {
+ int book = g->subclass_books[pclass][cval & csub];
+ cval = cval >> cbits;
+ if (book >= 0) {
+ int temp;
+ Codebook *c = f->codebooks + book;
+ DECODE(temp,f,c);
+ finalY[offset++] = temp;
+ } else
+ finalY[offset++] = 0;
+ }
+ }
+ if (f->valid_bits == INVALID_BITS) goto error; // behavior according to spec
+ step2_flag[0] = step2_flag[1] = 1;
+ for (j=2; j < g->values; ++j) {
+ int low, high, pred, highroom, lowroom, room, val;
+ low = g->neighbors[j][0];
+ high = g->neighbors[j][1];
+ //neighbors(g->Xlist, j, &low, &high);
+ pred = predict_point(g->Xlist[j], g->Xlist[low], g->Xlist[high], finalY[low], finalY[high]);
+ val = finalY[j];
+ highroom = range - pred;
+ lowroom = pred;
+ if (highroom < lowroom)
+ room = highroom * 2;
+ else
+ room = lowroom * 2;
+ if (val) {
+ step2_flag[low] = step2_flag[high] = 1;
+ step2_flag[j] = 1;
+ if (val >= room)
+ if (highroom > lowroom)
+ finalY[j] = val - lowroom + pred;
+ else
+ finalY[j] = pred - val + highroom - 1;
+ else
+ if (val & 1)
+ finalY[j] = pred - ((val+1)>>1);
+ else
+ finalY[j] = pred + (val>>1);
+ } else {
+ step2_flag[j] = 0;
+ finalY[j] = pred;
+ }
+ }
#ifdef STB_VORBIS_NO_DEFER_FLOOR
- do_floor(f, map, i, n, f->floor_buffers[i], finalY, step2_flag);
+ do_floor(f, map, i, n, f->floor_buffers[i], finalY, step2_flag);
#else
- // defer final floor computation until _after_ residue
- for (j = 0; j < g->values; ++j) {
- if (!step2_flag[j])
- finalY[j] = -1;
- }
-#endif
- }
- else {
- error:
- zero_channel[i] = TRUE;
+ // defer final floor computation until _after_ residue
+ for (j=0; j < g->values; ++j) {
+ if (!step2_flag[j])
+ finalY[j] = -1;
}
- // So we just defer everything else to later
-
- // at this point we've decoded the floor into buffer
- }
- }
- CHECK(f);
- // at this point we've decoded all floors
-
- if (f->alloc.alloc_buffer)
- assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset);
-
- // re-enable coupled channels if necessary
- memcpy(really_zero_channel, zero_channel, sizeof(really_zero_channel[0]) * f->channels);
- for (i = 0; i < map->coupling_steps; ++i)
- if (!zero_channel[map->chan[i].magnitude] || !zero_channel[map->chan[i].angle]) {
- zero_channel[map->chan[i].magnitude] = zero_channel[map->chan[i].angle] = FALSE;
- }
-
- CHECK(f);
- // RESIDUE DECODE
- for (i = 0; i < map->submaps; ++i) {
- float *residue_buffers[STB_VORBIS_MAX_CHANNELS];
- int r;
- uint8 do_not_decode[256];
- int ch = 0;
- for (j = 0; j < f->channels; ++j) {
- if (map->chan[j].mux == i) {
- if (zero_channel[j]) {
- do_not_decode[ch] = TRUE;
- residue_buffers[ch] = NULL;
- }
- else {
- do_not_decode[ch] = FALSE;
- residue_buffers[ch] = f->channel_buffers[j];
- }
- ++ch;
+#endif
+ } else {
+ error:
+ zero_channel[i] = TRUE;
+ }
+ // So we just defer everything else to later
+
+ // at this point we've decoded the floor into buffer
+ }
+ }
+ CHECK(f);
+ // at this point we've decoded all floors
+
+ if (f->alloc.alloc_buffer)
+ assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset);
+
+ // re-enable coupled channels if necessary
+ memcpy(really_zero_channel, zero_channel, sizeof(really_zero_channel[0]) * f->channels);
+ for (i=0; i < map->coupling_steps; ++i)
+ if (!zero_channel[map->chan[i].magnitude] || !zero_channel[map->chan[i].angle]) {
+ zero_channel[map->chan[i].magnitude] = zero_channel[map->chan[i].angle] = FALSE;
+ }
+
+ CHECK(f);
+// RESIDUE DECODE
+ for (i=0; i < map->submaps; ++i) {
+ float *residue_buffers[STB_VORBIS_MAX_CHANNELS];
+ int r;
+ uint8 do_not_decode[256];
+ int ch = 0;
+ for (j=0; j < f->channels; ++j) {
+ if (map->chan[j].mux == i) {
+ if (zero_channel[j]) {
+ do_not_decode[ch] = TRUE;
+ residue_buffers[ch] = NULL;
+ } else {
+ do_not_decode[ch] = FALSE;
+ residue_buffers[ch] = f->channel_buffers[j];
}
- }
- r = map->submap_residue[i];
- decode_residue(f, residue_buffers, ch, n2, r, do_not_decode);
- }
-
- if (f->alloc.alloc_buffer)
- assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset);
- CHECK(f);
-
- // INVERSE COUPLING
- for (i = map->coupling_steps - 1; i >= 0; --i) {
- int n2 = n >> 1;
- float *m = f->channel_buffers[map->chan[i].magnitude];
- float *a = f->channel_buffers[map->chan[i].angle];
- for (j = 0; j < n2; ++j) {
- float a2, m2;
- if (m[j] > 0)
- if (a[j] > 0)
- m2 = m[j], a2 = m[j] - a[j];
- else
- a2 = m[j], m2 = m[j] + a[j];
+ ++ch;
+ }
+ }
+ r = map->submap_residue[i];
+ decode_residue(f, residue_buffers, ch, n2, r, do_not_decode);
+ }
+
+ if (f->alloc.alloc_buffer)
+ assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset);
+ CHECK(f);
+
+// INVERSE COUPLING
+ for (i = map->coupling_steps-1; i >= 0; --i) {
+ int n2 = n >> 1;
+ float *m = f->channel_buffers[map->chan[i].magnitude];
+ float *a = f->channel_buffers[map->chan[i].angle ];
+ for (j=0; j < n2; ++j) {
+ float a2,m2;
+ if (m[j] > 0)
+ if (a[j] > 0)
+ m2 = m[j], a2 = m[j] - a[j];
+ else
+ a2 = m[j], m2 = m[j] + a[j];
+ else
+ if (a[j] > 0)
+ m2 = m[j], a2 = m[j] + a[j];
else
- if (a[j] > 0)
- m2 = m[j], a2 = m[j] + a[j];
- else
- a2 = m[j], m2 = m[j] - a[j];
- m[j] = m2;
- a[j] = a2;
- }
- }
- CHECK(f);
-
- // finish decoding the floors
+ a2 = m[j], m2 = m[j] - a[j];
+ m[j] = m2;
+ a[j] = a2;
+ }
+ }
+ CHECK(f);
+
+ // finish decoding the floors
#ifndef STB_VORBIS_NO_DEFER_FLOOR
- for (i = 0; i < f->channels; ++i) {
- if (really_zero_channel[i]) {
- memset(f->channel_buffers[i], 0, sizeof(*f->channel_buffers[i]) * n2);
- }
- else {
- do_floor(f, map, i, n, f->channel_buffers[i], f->finalY[i], NULL);
- }
- }
+ for (i=0; i < f->channels; ++i) {
+ if (really_zero_channel[i]) {
+ memset(f->channel_buffers[i], 0, sizeof(*f->channel_buffers[i]) * n2);
+ } else {
+ do_floor(f, map, i, n, f->channel_buffers[i], f->finalY[i], NULL);
+ }
+ }
#else
- for (i = 0; i < f->channels; ++i) {
- if (really_zero_channel[i]) {
- memset(f->channel_buffers[i], 0, sizeof(*f->channel_buffers[i]) * n2);
- }
- else {
- for (j = 0; j < n2; ++j)
- f->channel_buffers[i][j] *= f->floor_buffers[i][j];
- }
- }
+ for (i=0; i < f->channels; ++i) {
+ if (really_zero_channel[i]) {
+ memset(f->channel_buffers[i], 0, sizeof(*f->channel_buffers[i]) * n2);
+ } else {
+ for (j=0; j < n2; ++j)
+ f->channel_buffers[i][j] *= f->floor_buffers[i][j];
+ }
+ }
#endif
- // INVERSE MDCT
- CHECK(f);
- for (i = 0; i < f->channels; ++i)
- inverse_mdct(f->channel_buffers[i], n, f, m->blockflag);
- CHECK(f);
-
- // this shouldn't be necessary, unless we exited on an error
- // and want to flush to get to the next packet
- flush_packet(f);
-
- if (f->first_decode) {
- // assume we start so first non-discarded sample is sample 0
- // this isn't to spec, but spec would require us to read ahead
- // and decode the size of all current frames--could be done,
- // but presumably it's not a commonly used feature
- f->current_loc = -n2; // start of first frame is positioned for discard
- // we might have to discard samples "from" the next frame too,
- // if we're lapping a large block then a small at the start?
- f->discard_samples_deferred = n - right_end;
- f->current_loc_valid = TRUE;
- f->first_decode = FALSE;
- }
- else if (f->discard_samples_deferred) {
- if (f->discard_samples_deferred >= right_start - left_start) {
- f->discard_samples_deferred -= (right_start - left_start);
- left_start = right_start;
- *p_left = left_start;
- }
- else {
- left_start += f->discard_samples_deferred;
- *p_left = left_start;
- f->discard_samples_deferred = 0;
- }
- }
- else if (f->previous_length == 0 && f->current_loc_valid) {
- // we're recovering from a seek... that means we're going to discard
- // the samples from this packet even though we know our position from
- // the last page header, so we need to update the position based on
- // the discarded samples here
- // but wait, the code below is going to add this in itself even
- // on a discard, so we don't need to do it here...
- }
-
- // check if we have ogg information about the sample # for this packet
- if (f->last_seg_which == f->end_seg_with_known_loc) {
- // if we have a valid current loc, and this is final:
- if (f->current_loc_valid && (f->page_flag & PAGEFLAG_last_page)) {
- uint32 current_end = f->known_loc_for_packet;
- // then let's infer the size of the (probably) short final frame
- if (current_end < f->current_loc + (right_end - left_start)) {
- if (current_end < f->current_loc) {
- // negative truncation, that's impossible!
- *len = 0;
- }
- else {
- *len = current_end - f->current_loc;
- }
- *len += left_start; // this doesn't seem right, but has no ill effect on my test files
- if (*len > right_end) *len = right_end; // this should never happen
- f->current_loc += *len;
- return TRUE;
+// INVERSE MDCT
+ CHECK(f);
+ for (i=0; i < f->channels; ++i)
+ inverse_mdct(f->channel_buffers[i], n, f, m->blockflag);
+ CHECK(f);
+
+ // this shouldn't be necessary, unless we exited on an error
+ // and want to flush to get to the next packet
+ flush_packet(f);
+
+ if (f->first_decode) {
+ // assume we start so first non-discarded sample is sample 0
+ // this isn't to spec, but spec would require us to read ahead
+ // and decode the size of all current frames--could be done,
+ // but presumably it's not a commonly used feature
+ f->current_loc = 0u - n2; // start of first frame is positioned for discard (NB this is an intentional unsigned overflow/wrap-around)
+ // we might have to discard samples "from" the next frame too,
+ // if we're lapping a large block then a small at the start?
+ f->discard_samples_deferred = n - right_end;
+ f->current_loc_valid = TRUE;
+ f->first_decode = FALSE;
+ } else if (f->discard_samples_deferred) {
+ if (f->discard_samples_deferred >= right_start - left_start) {
+ f->discard_samples_deferred -= (right_start - left_start);
+ left_start = right_start;
+ *p_left = left_start;
+ } else {
+ left_start += f->discard_samples_deferred;
+ *p_left = left_start;
+ f->discard_samples_deferred = 0;
+ }
+ } else if (f->previous_length == 0 && f->current_loc_valid) {
+ // we're recovering from a seek... that means we're going to discard
+ // the samples from this packet even though we know our position from
+ // the last page header, so we need to update the position based on
+ // the discarded samples here
+ // but wait, the code below is going to add this in itself even
+ // on a discard, so we don't need to do it here...
+ }
+
+ // check if we have ogg information about the sample # for this packet
+ if (f->last_seg_which == f->end_seg_with_known_loc) {
+ // if we have a valid current loc, and this is final:
+ if (f->current_loc_valid && (f->page_flag & PAGEFLAG_last_page)) {
+ uint32 current_end = f->known_loc_for_packet;
+ // then let's infer the size of the (probably) short final frame
+ if (current_end < f->current_loc + (right_end-left_start)) {
+ if (current_end < f->current_loc) {
+ // negative truncation, that's impossible!
+ *len = 0;
+ } else {
+ *len = current_end - f->current_loc;
}
- }
- // otherwise, just set our sample loc
- // guess that the ogg granule pos refers to the _middle_ of the
- // last frame?
- // set f->current_loc to the position of left_start
- f->current_loc = f->known_loc_for_packet - (n2 - left_start);
- f->current_loc_valid = TRUE;
- }
- if (f->current_loc_valid)
- f->current_loc += (right_start - left_start);
-
- if (f->alloc.alloc_buffer)
- assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset);
- *len = right_end; // ignore samples after the window goes to 0
- CHECK(f);
-
- return TRUE;
+ *len += left_start; // this doesn't seem right, but has no ill effect on my test files
+ if (*len > right_end) *len = right_end; // this should never happen
+ f->current_loc += *len;
+ return TRUE;
+ }
+ }
+ // otherwise, just set our sample loc
+ // guess that the ogg granule pos refers to the _middle_ of the
+ // last frame?
+ // set f->current_loc to the position of left_start
+ f->current_loc = f->known_loc_for_packet - (n2-left_start);
+ f->current_loc_valid = TRUE;
+ }
+ if (f->current_loc_valid)
+ f->current_loc += (right_start - left_start);
+
+ if (f->alloc.alloc_buffer)
+ assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset);
+ *len = right_end; // ignore samples after the window goes to 0
+ CHECK(f);
+
+ return TRUE;
}
static int vorbis_decode_packet(vorb *f, int *len, int *p_left, int *p_right)
{
- int mode, left_end, right_end;
- if (!vorbis_decode_initial(f, p_left, &left_end, p_right, &right_end, &mode)) return 0;
- return vorbis_decode_packet_rest(f, len, f->mode_config + mode, *p_left, left_end, *p_right, right_end, p_left);
+ int mode, left_end, right_end;
+ if (!vorbis_decode_initial(f, p_left, &left_end, p_right, &right_end, &mode)) return 0;
+ return vorbis_decode_packet_rest(f, len, f->mode_config + mode, *p_left, left_end, *p_right, right_end, p_left);
}
static int vorbis_finish_frame(stb_vorbis *f, int len, int left, int right)
{
- int prev, i, j;
- // we use right&left (the start of the right- and left-window sin()-regions)
- // to determine how much to return, rather than inferring from the rules
- // (same result, clearer code); 'left' indicates where our sin() window
- // starts, therefore where the previous window's right edge starts, and
- // therefore where to start mixing from the previous buffer. 'right'
- // indicates where our sin() ending-window starts, therefore that's where
- // we start saving, and where our returned-data ends.
-
- // mixin from previous window
- if (f->previous_length) {
- int i, j, n = f->previous_length;
- float *w = get_window(f, n);
- for (i = 0; i < f->channels; ++i) {
- for (j = 0; j < n; ++j)
- f->channel_buffers[i][left + j] =
- f->channel_buffers[i][left + j] * w[j] +
- f->previous_window[i][j] * w[n - 1 - j];
- }
- }
-
- prev = f->previous_length;
-
- // last half of this data becomes previous window
- f->previous_length = len - right;
-
- // @OPTIMIZE: could avoid this copy by double-buffering the
- // output (flipping previous_window with channel_buffers), but
- // then previous_window would have to be 2x as large, and
- // channel_buffers couldn't be temp mem (although they're NOT
- // currently temp mem, they could be (unless we want to level
- // performance by spreading out the computation))
- for (i = 0; i < f->channels; ++i)
- for (j = 0; right + j < len; ++j)
- f->previous_window[i][j] = f->channel_buffers[i][right + j];
-
- if (!prev)
- // there was no previous packet, so this data isn't valid...
- // this isn't entirely true, only the would-have-overlapped data
- // isn't valid, but this seems to be what the spec requires
- return 0;
-
- // truncate a short frame
- if (len < right) right = len;
-
- f->samples_output += right - left;
-
- return right - left;
+ int prev,i,j;
+ // we use right&left (the start of the right- and left-window sin()-regions)
+ // to determine how much to return, rather than inferring from the rules
+ // (same result, clearer code); 'left' indicates where our sin() window
+ // starts, therefore where the previous window's right edge starts, and
+ // therefore where to start mixing from the previous buffer. 'right'
+ // indicates where our sin() ending-window starts, therefore that's where
+ // we start saving, and where our returned-data ends.
+
+ // mixin from previous window
+ if (f->previous_length) {
+ int i,j, n = f->previous_length;
+ float *w = get_window(f, n);
+ if (w == NULL) return 0;
+ for (i=0; i < f->channels; ++i) {
+ for (j=0; j < n; ++j)
+ f->channel_buffers[i][left+j] =
+ f->channel_buffers[i][left+j]*w[ j] +
+ f->previous_window[i][ j]*w[n-1-j];
+ }
+ }
+
+ prev = f->previous_length;
+
+ // last half of this data becomes previous window
+ f->previous_length = len - right;
+
+ // @OPTIMIZE: could avoid this copy by double-buffering the
+ // output (flipping previous_window with channel_buffers), but
+ // then previous_window would have to be 2x as large, and
+ // channel_buffers couldn't be temp mem (although they're NOT
+ // currently temp mem, they could be (unless we want to level
+ // performance by spreading out the computation))
+ for (i=0; i < f->channels; ++i)
+ for (j=0; right+j < len; ++j)
+ f->previous_window[i][j] = f->channel_buffers[i][right+j];
+
+ if (!prev)
+ // there was no previous packet, so this data isn't valid...
+ // this isn't entirely true, only the would-have-overlapped data
+ // isn't valid, but this seems to be what the spec requires
+ return 0;
+
+ // truncate a short frame
+ if (len < right) right = len;
+
+ f->samples_output += right-left;
+
+ return right - left;
}
static int vorbis_pump_first_frame(stb_vorbis *f)
{
- int len, right, left, res;
- res = vorbis_decode_packet(f, &len, &left, &right);
- if (res)
- vorbis_finish_frame(f, len, left, right);
- return res;
+ int len, right, left, res;
+ res = vorbis_decode_packet(f, &len, &left, &right);
+ if (res)
+ vorbis_finish_frame(f, len, left, right);
+ return res;
}
#ifndef STB_VORBIS_NO_PUSHDATA_API
-static int is_whole_packet_present(stb_vorbis *f, int end_page)
-{
- // make sure that we have the packet available before continuing...
- // this requires a full ogg parse, but we know we can fetch from f->stream
-
- // instead of coding this out explicitly, we could save the current read state,
- // read the next packet with get8() until end-of-packet, check f->eof, then
- // reset the state? but that would be slower, esp. since we'd have over 256 bytes
- // of state to restore (primarily the page segment table)
-
- int s = f->next_seg, first = TRUE;
- uint8 *p = f->stream;
-
- if (s != -1) { // if we're not starting the packet with a 'continue on next page' flag
- for (; s < f->segment_count; ++s) {
- p += f->segments[s];
- if (f->segments[s] < 255) // stop at first short segment
- break;
- }
- // either this continues, or it ends it...
- if (end_page)
- if (s < f->segment_count - 1) return error(f, VORBIS_invalid_stream);
- if (s == f->segment_count)
- s = -1; // set 'crosses page' flag
- if (p > f->stream_end) return error(f, VORBIS_need_more_data);
- first = FALSE;
- }
- for (; s == -1;) {
- uint8 *q;
- int n;
-
- // check that we have the page header ready
- if (p + 26 >= f->stream_end) return error(f, VORBIS_need_more_data);
- // validate the page
- if (memcmp(p, ogg_page_header, 4)) return error(f, VORBIS_invalid_stream);
- if (p[4] != 0) return error(f, VORBIS_invalid_stream);
- if (first) { // the first segment must NOT have 'continued_packet', later ones MUST
- if (f->previous_length)
- if ((p[5] & PAGEFLAG_continued_packet)) return error(f, VORBIS_invalid_stream);
- // if no previous length, we're resynching, so we can come in on a continued-packet,
- // which we'll just drop
- }
- else {
- if (!(p[5] & PAGEFLAG_continued_packet)) return error(f, VORBIS_invalid_stream);
- }
- n = p[26]; // segment counts
- q = p + 27; // q points to segment table
- p = q + n; // advance past header
- // make sure we've read the segment table
- if (p > f->stream_end) return error(f, VORBIS_need_more_data);
- for (s = 0; s < n; ++s) {
- p += q[s];
- if (q[s] < 255)
- break;
- }
- if (end_page)
- if (s < n - 1) return error(f, VORBIS_invalid_stream);
- if (s == n)
- s = -1; // set 'crosses page' flag
- if (p > f->stream_end) return error(f, VORBIS_need_more_data);
- first = FALSE;
- }
- return TRUE;
+static int is_whole_packet_present(stb_vorbis *f)
+{
+ // make sure that we have the packet available before continuing...
+ // this requires a full ogg parse, but we know we can fetch from f->stream
+
+ // instead of coding this out explicitly, we could save the current read state,
+ // read the next packet with get8() until end-of-packet, check f->eof, then
+ // reset the state? but that would be slower, esp. since we'd have over 256 bytes
+ // of state to restore (primarily the page segment table)
+
+ int s = f->next_seg, first = TRUE;
+ uint8 *p = f->stream;
+
+ if (s != -1) { // if we're not starting the packet with a 'continue on next page' flag
+ for (; s < f->segment_count; ++s) {
+ p += f->segments[s];
+ if (f->segments[s] < 255) // stop at first short segment
+ break;
+ }
+ // either this continues, or it ends it...
+ if (s == f->segment_count)
+ s = -1; // set 'crosses page' flag
+ if (p > f->stream_end) return error(f, VORBIS_need_more_data);
+ first = FALSE;
+ }
+ for (; s == -1;) {
+ uint8 *q;
+ int n;
+
+ // check that we have the page header ready
+ if (p + 26 >= f->stream_end) return error(f, VORBIS_need_more_data);
+ // validate the page
+ if (memcmp(p, ogg_page_header, 4)) return error(f, VORBIS_invalid_stream);
+ if (p[4] != 0) return error(f, VORBIS_invalid_stream);
+ if (first) { // the first segment must NOT have 'continued_packet', later ones MUST
+ if (f->previous_length)
+ if ((p[5] & PAGEFLAG_continued_packet)) return error(f, VORBIS_invalid_stream);
+ // if no previous length, we're resynching, so we can come in on a continued-packet,
+ // which we'll just drop
+ } else {
+ if (!(p[5] & PAGEFLAG_continued_packet)) return error(f, VORBIS_invalid_stream);
+ }
+ n = p[26]; // segment counts
+ q = p+27; // q points to segment table
+ p = q + n; // advance past header
+ // make sure we've read the segment table
+ if (p > f->stream_end) return error(f, VORBIS_need_more_data);
+ for (s=0; s < n; ++s) {
+ p += q[s];
+ if (q[s] < 255)
+ break;
+ }
+ if (s == n)
+ s = -1; // set 'crosses page' flag
+ if (p > f->stream_end) return error(f, VORBIS_need_more_data);
+ first = FALSE;
+ }
+ return TRUE;
}
#endif // !STB_VORBIS_NO_PUSHDATA_API
static int start_decoder(vorb *f)
{
- uint8 header[6], x, y;
- int len, i, j, k, max_submaps = 0;
- int longest_floorlist = 0;
-
- // first page, first packet
-
- if (!start_page(f)) return FALSE;
- // validate page flag
- if (!(f->page_flag & PAGEFLAG_first_page)) return error(f, VORBIS_invalid_first_page);
- if (f->page_flag & PAGEFLAG_last_page) return error(f, VORBIS_invalid_first_page);
- if (f->page_flag & PAGEFLAG_continued_packet) return error(f, VORBIS_invalid_first_page);
- // check for expected packet length
- if (f->segment_count != 1) return error(f, VORBIS_invalid_first_page);
- if (f->segments[0] != 30) return error(f, VORBIS_invalid_first_page);
- // read packet
- // check packet header
- if (get8(f) != VORBIS_packet_id) return error(f, VORBIS_invalid_first_page);
- if (!getn(f, header, 6)) return error(f, VORBIS_unexpected_eof);
- if (!vorbis_validate(header)) return error(f, VORBIS_invalid_first_page);
- // vorbis_version
- if (get32(f) != 0) return error(f, VORBIS_invalid_first_page);
- f->channels = get8(f); if (!f->channels) return error(f, VORBIS_invalid_first_page);
- if (f->channels > STB_VORBIS_MAX_CHANNELS) return error(f, VORBIS_too_many_channels);
- f->sample_rate = get32(f); if (!f->sample_rate) return error(f, VORBIS_invalid_first_page);
- get32(f); // bitrate_maximum
- get32(f); // bitrate_nominal
- get32(f); // bitrate_minimum
- x = get8(f);
- {
- int log0, log1;
- log0 = x & 15;
- log1 = x >> 4;
- f->blocksize_0 = 1 << log0;
- f->blocksize_1 = 1 << log1;
- if (log0 < 6 || log0 > 13) return error(f, VORBIS_invalid_setup);
- if (log1 < 6 || log1 > 13) return error(f, VORBIS_invalid_setup);
- if (log0 > log1) return error(f, VORBIS_invalid_setup);
- }
-
- // framing_flag
- x = get8(f);
- if (!(x & 1)) return error(f, VORBIS_invalid_first_page);
-
- // second packet!
- if (!start_page(f)) return FALSE;
-
- if (!start_packet(f)) return FALSE;
- do {
- len = next_segment(f);
- skip(f, len);
- f->bytes_in_seg = 0;
- } while (len);
-
- // third packet!
- if (!start_packet(f)) return FALSE;
+ uint8 header[6], x,y;
+ int len,i,j,k, max_submaps = 0;
+ int longest_floorlist=0;
+
+ // first page, first packet
+ f->first_decode = TRUE;
+
+ if (!start_page(f)) return FALSE;
+ // validate page flag
+ if (!(f->page_flag & PAGEFLAG_first_page)) return error(f, VORBIS_invalid_first_page);
+ if (f->page_flag & PAGEFLAG_last_page) return error(f, VORBIS_invalid_first_page);
+ if (f->page_flag & PAGEFLAG_continued_packet) return error(f, VORBIS_invalid_first_page);
+ // check for expected packet length
+ if (f->segment_count != 1) return error(f, VORBIS_invalid_first_page);
+ if (f->segments[0] != 30) {
+ // check for the Ogg skeleton fishead identifying header to refine our error
+ if (f->segments[0] == 64 &&
+ getn(f, header, 6) &&
+ header[0] == 'f' &&
+ header[1] == 'i' &&
+ header[2] == 's' &&
+ header[3] == 'h' &&
+ header[4] == 'e' &&
+ header[5] == 'a' &&
+ get8(f) == 'd' &&
+ get8(f) == '\0') return error(f, VORBIS_ogg_skeleton_not_supported);
+ else
+ return error(f, VORBIS_invalid_first_page);
+ }
-#ifndef STB_VORBIS_NO_PUSHDATA_API
- if (IS_PUSH_MODE(f)) {
- if (!is_whole_packet_present(f, TRUE)) {
- // convert error in ogg header to write type
- if (f->error == VORBIS_invalid_stream)
- f->error = VORBIS_invalid_setup;
- return FALSE;
- }
- }
-#endif
+ // read packet
+ // check packet header
+ if (get8(f) != VORBIS_packet_id) return error(f, VORBIS_invalid_first_page);
+ if (!getn(f, header, 6)) return error(f, VORBIS_unexpected_eof);
+ if (!vorbis_validate(header)) return error(f, VORBIS_invalid_first_page);
+ // vorbis_version
+ if (get32(f) != 0) return error(f, VORBIS_invalid_first_page);
+ f->channels = get8(f); if (!f->channels) return error(f, VORBIS_invalid_first_page);
+ if (f->channels > STB_VORBIS_MAX_CHANNELS) return error(f, VORBIS_too_many_channels);
+ f->sample_rate = get32(f); if (!f->sample_rate) return error(f, VORBIS_invalid_first_page);
+ get32(f); // bitrate_maximum
+ get32(f); // bitrate_nominal
+ get32(f); // bitrate_minimum
+ x = get8(f);
+ {
+ int log0,log1;
+ log0 = x & 15;
+ log1 = x >> 4;
+ f->blocksize_0 = 1 << log0;
+ f->blocksize_1 = 1 << log1;
+ if (log0 < 6 || log0 > 13) return error(f, VORBIS_invalid_setup);
+ if (log1 < 6 || log1 > 13) return error(f, VORBIS_invalid_setup);
+ if (log0 > log1) return error(f, VORBIS_invalid_setup);
+ }
- crc32_init(); // always init it, to avoid multithread race conditions
-
- if (get8_packet(f) != VORBIS_packet_setup) return error(f, VORBIS_invalid_setup);
- for (i = 0; i < 6; ++i) header[i] = get8_packet(f);
- if (!vorbis_validate(header)) return error(f, VORBIS_invalid_setup);
-
- // codebooks
-
- f->codebook_count = get_bits(f, 8) + 1;
- f->codebooks = (Codebook *)setup_malloc(f, sizeof(*f->codebooks) * f->codebook_count);
- if (f->codebooks == NULL) return error(f, VORBIS_outofmem);
- memset(f->codebooks, 0, sizeof(*f->codebooks) * f->codebook_count);
- for (i = 0; i < f->codebook_count; ++i) {
- uint32 *values;
- int ordered, sorted_count;
- int total = 0;
- uint8 *lengths;
- Codebook *c = f->codebooks + i;
- CHECK(f);
- x = get_bits(f, 8); if (x != 0x42) return error(f, VORBIS_invalid_setup);
- x = get_bits(f, 8); if (x != 0x43) return error(f, VORBIS_invalid_setup);
- x = get_bits(f, 8); if (x != 0x56) return error(f, VORBIS_invalid_setup);
- x = get_bits(f, 8);
- c->dimensions = (get_bits(f, 8) << 8) + x;
- x = get_bits(f, 8);
- y = get_bits(f, 8);
- c->entries = (get_bits(f, 8) << 16) + (y << 8) + x;
- ordered = get_bits(f, 1);
- c->sparse = ordered ? 0 : get_bits(f, 1);
-
- if (c->dimensions == 0 && c->entries != 0) return error(f, VORBIS_invalid_setup);
-
- if (c->sparse)
- lengths = (uint8 *)setup_temp_malloc(f, c->entries);
- else
- lengths = c->codeword_lengths = (uint8 *)setup_malloc(f, c->entries);
-
- if (!lengths) return error(f, VORBIS_outofmem);
-
- if (ordered) {
- int current_entry = 0;
- int current_length = get_bits(f, 5) + 1;
- while (current_entry < c->entries) {
- int limit = c->entries - current_entry;
- int n = get_bits(f, ilog(limit));
- if (current_entry + n >(int) c->entries) { return error(f, VORBIS_invalid_setup); }
- memset(lengths + current_entry, current_length, n);
- current_entry += n;
- ++current_length;
- }
- }
- else {
- for (j = 0; j < c->entries; ++j) {
- int present = c->sparse ? get_bits(f, 1) : 1;
- if (present) {
- lengths[j] = get_bits(f, 5) + 1;
- ++total;
- if (lengths[j] == 32)
- return error(f, VORBIS_invalid_setup);
- }
- else {
- lengths[j] = NO_CODE;
- }
- }
- }
-
- if (c->sparse && total >= c->entries >> 2) {
- // convert sparse items to non-sparse!
- if (c->entries > (int)f->setup_temp_memory_required)
- f->setup_temp_memory_required = c->entries;
-
- c->codeword_lengths = (uint8 *)setup_malloc(f, c->entries);
- if (c->codeword_lengths == NULL) return error(f, VORBIS_outofmem);
- memcpy(c->codeword_lengths, lengths, c->entries);
- setup_temp_free(f, lengths, c->entries); // note this is only safe if there have been no intervening temp mallocs!
- lengths = c->codeword_lengths;
- c->sparse = 0;
- }
-
- // compute the size of the sorted tables
- if (c->sparse) {
- sorted_count = total;
- }
- else {
- sorted_count = 0;
-#ifndef STB_VORBIS_NO_HUFFMAN_BINARY_SEARCH
- for (j = 0; j < c->entries; ++j)
- if (lengths[j] > STB_VORBIS_FAST_HUFFMAN_LENGTH && lengths[j] != NO_CODE)
- ++sorted_count;
-#endif
- }
+ // framing_flag
+ x = get8(f);
+ if (!(x & 1)) return error(f, VORBIS_invalid_first_page);
- c->sorted_entries = sorted_count;
- values = NULL;
+ // second packet!
+ if (!start_page(f)) return FALSE;
- CHECK(f);
- if (!c->sparse) {
- c->codewords = (uint32 *)setup_malloc(f, sizeof(c->codewords[0]) * c->entries);
- if (!c->codewords) return error(f, VORBIS_outofmem);
- }
- else {
- unsigned int size;
- if (c->sorted_entries) {
- c->codeword_lengths = (uint8 *)setup_malloc(f, c->sorted_entries);
- if (!c->codeword_lengths) return error(f, VORBIS_outofmem);
- c->codewords = (uint32 *)setup_temp_malloc(f, sizeof(*c->codewords) * c->sorted_entries);
- if (!c->codewords) return error(f, VORBIS_outofmem);
- values = (uint32 *)setup_temp_malloc(f, sizeof(*values) * c->sorted_entries);
- if (!values) return error(f, VORBIS_outofmem);
- }
- size = c->entries + (sizeof(*c->codewords) + sizeof(*values)) * c->sorted_entries;
- if (size > f->setup_temp_memory_required)
- f->setup_temp_memory_required = size;
- }
-
- if (!compute_codewords(c, lengths, c->entries, values)) {
- if (c->sparse) setup_temp_free(f, values, 0);
- return error(f, VORBIS_invalid_setup);
- }
-
- if (c->sorted_entries) {
- // allocate an extra slot for sentinels
- c->sorted_codewords = (uint32 *)setup_malloc(f, sizeof(*c->sorted_codewords) * (c->sorted_entries + 1));
- if (c->sorted_codewords == NULL) return error(f, VORBIS_outofmem);
- // allocate an extra slot at the front so that c->sorted_values[-1] is defined
- // so that we can catch that case without an extra if
- c->sorted_values = (int *)setup_malloc(f, sizeof(*c->sorted_values) * (c->sorted_entries + 1));
- if (c->sorted_values == NULL) return error(f, VORBIS_outofmem);
- ++c->sorted_values;
- c->sorted_values[-1] = -1;
- compute_sorted_huffman(c, lengths, values);
- }
-
- if (c->sparse) {
- setup_temp_free(f, values, sizeof(*values)*c->sorted_entries);
- setup_temp_free(f, c->codewords, sizeof(*c->codewords)*c->sorted_entries);
- setup_temp_free(f, lengths, c->entries);
- c->codewords = NULL;
- }
-
- compute_accelerated_huffman(c);
-
- CHECK(f);
- c->lookup_type = get_bits(f, 4);
- if (c->lookup_type > 2) return error(f, VORBIS_invalid_setup);
- if (c->lookup_type > 0) {
- uint16 *mults;
- c->minimum_value = float32_unpack(get_bits(f, 32));
- c->delta_value = float32_unpack(get_bits(f, 32));
- c->value_bits = get_bits(f, 4) + 1;
- c->sequence_p = get_bits(f, 1);
- if (c->lookup_type == 1) {
- c->lookup_values = lookup1_values(c->entries, c->dimensions);
- }
- else {
- c->lookup_values = c->entries * c->dimensions;
- }
- if (c->lookup_values == 0) return error(f, VORBIS_invalid_setup);
- mults = (uint16 *)setup_temp_malloc(f, sizeof(mults[0]) * c->lookup_values);
- if (mults == NULL) return error(f, VORBIS_outofmem);
- for (j = 0; j < (int)c->lookup_values; ++j) {
- int q = get_bits(f, c->value_bits);
- if (q == EOP) { setup_temp_free(f, mults, sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_invalid_setup); }
- mults[j] = q;
+ if (!start_packet(f)) return FALSE;
+
+ if (!next_segment(f)) return FALSE;
+
+ if (get8_packet(f) != VORBIS_packet_comment) return error(f, VORBIS_invalid_setup);
+ for (i=0; i < 6; ++i) header[i] = get8_packet(f);
+ if (!vorbis_validate(header)) return error(f, VORBIS_invalid_setup);
+ //file vendor
+ len = get32_packet(f);
+ f->vendor = (char*)setup_malloc(f, sizeof(char) * (len+1));
+ if (f->vendor == NULL) return error(f, VORBIS_outofmem);
+ for(i=0; i < len; ++i) {
+ f->vendor[i] = get8_packet(f);
+ }
+ f->vendor[len] = (char)'\0';
+ //user comments
+ f->comment_list_length = get32_packet(f);
+ f->comment_list = NULL;
+ if (f->comment_list_length > 0)
+ {
+ f->comment_list = (char**) setup_malloc(f, sizeof(char*) * (f->comment_list_length));
+ if (f->comment_list == NULL) return error(f, VORBIS_outofmem);
+ }
+
+ for(i=0; i < f->comment_list_length; ++i) {
+ len = get32_packet(f);
+ f->comment_list[i] = (char*)setup_malloc(f, sizeof(char) * (len+1));
+ if (f->comment_list[i] == NULL) return error(f, VORBIS_outofmem);
+
+ for(j=0; j < len; ++j) {
+ f->comment_list[i][j] = get8_packet(f);
+ }
+ f->comment_list[i][len] = (char)'\0';
+ }
+
+ // framing_flag
+ x = get8_packet(f);
+ if (!(x & 1)) return error(f, VORBIS_invalid_setup);
+
+
+ skip(f, f->bytes_in_seg);
+ f->bytes_in_seg = 0;
+
+ do {
+ len = next_segment(f);
+ skip(f, len);
+ f->bytes_in_seg = 0;
+ } while (len);
+
+ // third packet!
+ if (!start_packet(f)) return FALSE;
+
+ #ifndef STB_VORBIS_NO_PUSHDATA_API
+ if (IS_PUSH_MODE(f)) {
+ if (!is_whole_packet_present(f)) {
+ // convert error in ogg header to write type
+ if (f->error == VORBIS_invalid_stream)
+ f->error = VORBIS_invalid_setup;
+ return FALSE;
+ }
+ }
+ #endif
+
+ crc32_init(); // always init it, to avoid multithread race conditions
+
+ if (get8_packet(f) != VORBIS_packet_setup) return error(f, VORBIS_invalid_setup);
+ for (i=0; i < 6; ++i) header[i] = get8_packet(f);
+ if (!vorbis_validate(header)) return error(f, VORBIS_invalid_setup);
+
+ // codebooks
+
+ f->codebook_count = get_bits(f,8) + 1;
+ f->codebooks = (Codebook *) setup_malloc(f, sizeof(*f->codebooks) * f->codebook_count);
+ if (f->codebooks == NULL) return error(f, VORBIS_outofmem);
+ memset(f->codebooks, 0, sizeof(*f->codebooks) * f->codebook_count);
+ for (i=0; i < f->codebook_count; ++i) {
+ uint32 *values;
+ int ordered, sorted_count;
+ int total=0;
+ uint8 *lengths;
+ Codebook *c = f->codebooks+i;
+ CHECK(f);
+ x = get_bits(f, 8); if (x != 0x42) return error(f, VORBIS_invalid_setup);
+ x = get_bits(f, 8); if (x != 0x43) return error(f, VORBIS_invalid_setup);
+ x = get_bits(f, 8); if (x != 0x56) return error(f, VORBIS_invalid_setup);
+ x = get_bits(f, 8);
+ c->dimensions = (get_bits(f, 8)<<8) + x;
+ x = get_bits(f, 8);
+ y = get_bits(f, 8);
+ c->entries = (get_bits(f, 8)<<16) + (y<<8) + x;
+ ordered = get_bits(f,1);
+ c->sparse = ordered ? 0 : get_bits(f,1);
+
+ if (c->dimensions == 0 && c->entries != 0) return error(f, VORBIS_invalid_setup);
+
+ if (c->sparse)
+ lengths = (uint8 *) setup_temp_malloc(f, c->entries);
+ else
+ lengths = c->codeword_lengths = (uint8 *) setup_malloc(f, c->entries);
+
+ if (!lengths) return error(f, VORBIS_outofmem);
+
+ if (ordered) {
+ int current_entry = 0;
+ int current_length = get_bits(f,5) + 1;
+ while (current_entry < c->entries) {
+ int limit = c->entries - current_entry;
+ int n = get_bits(f, ilog(limit));
+ if (current_length >= 32) return error(f, VORBIS_invalid_setup);
+ if (current_entry + n > (int) c->entries) { return error(f, VORBIS_invalid_setup); }
+ memset(lengths + current_entry, current_length, n);
+ current_entry += n;
+ ++current_length;
+ }
+ } else {
+ for (j=0; j < c->entries; ++j) {
+ int present = c->sparse ? get_bits(f,1) : 1;
+ if (present) {
+ lengths[j] = get_bits(f, 5) + 1;
+ ++total;
+ if (lengths[j] == 32)
+ return error(f, VORBIS_invalid_setup);
+ } else {
+ lengths[j] = NO_CODE;
}
+ }
+ }
+
+ if (c->sparse && total >= c->entries >> 2) {
+ // convert sparse items to non-sparse!
+ if (c->entries > (int) f->setup_temp_memory_required)
+ f->setup_temp_memory_required = c->entries;
+
+ c->codeword_lengths = (uint8 *) setup_malloc(f, c->entries);
+ if (c->codeword_lengths == NULL) return error(f, VORBIS_outofmem);
+ memcpy(c->codeword_lengths, lengths, c->entries);
+ setup_temp_free(f, lengths, c->entries); // note this is only safe if there have been no intervening temp mallocs!
+ lengths = c->codeword_lengths;
+ c->sparse = 0;
+ }
+
+ // compute the size of the sorted tables
+ if (c->sparse) {
+ sorted_count = total;
+ } else {
+ sorted_count = 0;
+ #ifndef STB_VORBIS_NO_HUFFMAN_BINARY_SEARCH
+ for (j=0; j < c->entries; ++j)
+ if (lengths[j] > STB_VORBIS_FAST_HUFFMAN_LENGTH && lengths[j] != NO_CODE)
+ ++sorted_count;
+ #endif
+ }
+
+ c->sorted_entries = sorted_count;
+ values = NULL;
+
+ CHECK(f);
+ if (!c->sparse) {
+ c->codewords = (uint32 *) setup_malloc(f, sizeof(c->codewords[0]) * c->entries);
+ if (!c->codewords) return error(f, VORBIS_outofmem);
+ } else {
+ unsigned int size;
+ if (c->sorted_entries) {
+ c->codeword_lengths = (uint8 *) setup_malloc(f, c->sorted_entries);
+ if (!c->codeword_lengths) return error(f, VORBIS_outofmem);
+ c->codewords = (uint32 *) setup_temp_malloc(f, sizeof(*c->codewords) * c->sorted_entries);
+ if (!c->codewords) return error(f, VORBIS_outofmem);
+ values = (uint32 *) setup_temp_malloc(f, sizeof(*values) * c->sorted_entries);
+ if (!values) return error(f, VORBIS_outofmem);
+ }
+ size = c->entries + (sizeof(*c->codewords) + sizeof(*values)) * c->sorted_entries;
+ if (size > f->setup_temp_memory_required)
+ f->setup_temp_memory_required = size;
+ }
+
+ if (!compute_codewords(c, lengths, c->entries, values)) {
+ if (c->sparse) setup_temp_free(f, values, 0);
+ return error(f, VORBIS_invalid_setup);
+ }
+
+ if (c->sorted_entries) {
+ // allocate an extra slot for sentinels
+ c->sorted_codewords = (uint32 *) setup_malloc(f, sizeof(*c->sorted_codewords) * (c->sorted_entries+1));
+ if (c->sorted_codewords == NULL) return error(f, VORBIS_outofmem);
+ // allocate an extra slot at the front so that c->sorted_values[-1] is defined
+ // so that we can catch that case without an extra if
+ c->sorted_values = ( int *) setup_malloc(f, sizeof(*c->sorted_values ) * (c->sorted_entries+1));
+ if (c->sorted_values == NULL) return error(f, VORBIS_outofmem);
+ ++c->sorted_values;
+ c->sorted_values[-1] = -1;
+ compute_sorted_huffman(c, lengths, values);
+ }
+
+ if (c->sparse) {
+ setup_temp_free(f, values, sizeof(*values)*c->sorted_entries);
+ setup_temp_free(f, c->codewords, sizeof(*c->codewords)*c->sorted_entries);
+ setup_temp_free(f, lengths, c->entries);
+ c->codewords = NULL;
+ }
+
+ compute_accelerated_huffman(c);
+
+ CHECK(f);
+ c->lookup_type = get_bits(f, 4);
+ if (c->lookup_type > 2) return error(f, VORBIS_invalid_setup);
+ if (c->lookup_type > 0) {
+ uint16 *mults;
+ c->minimum_value = float32_unpack(get_bits(f, 32));
+ c->delta_value = float32_unpack(get_bits(f, 32));
+ c->value_bits = get_bits(f, 4)+1;
+ c->sequence_p = get_bits(f,1);
+ if (c->lookup_type == 1) {
+ int values = lookup1_values(c->entries, c->dimensions);
+ if (values < 0) return error(f, VORBIS_invalid_setup);
+ c->lookup_values = (uint32) values;
+ } else {
+ c->lookup_values = c->entries * c->dimensions;
+ }
+ if (c->lookup_values == 0) return error(f, VORBIS_invalid_setup);
+ mults = (uint16 *) setup_temp_malloc(f, sizeof(mults[0]) * c->lookup_values);
+ if (mults == NULL) return error(f, VORBIS_outofmem);
+ for (j=0; j < (int) c->lookup_values; ++j) {
+ int q = get_bits(f, c->value_bits);
+ if (q == EOP) { setup_temp_free(f,mults,sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_invalid_setup); }
+ mults[j] = q;
+ }
#ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK
- if (c->lookup_type == 1) {
- int len, sparse = c->sparse;
- float last = 0;
- // pre-expand the lookup1-style multiplicands, to avoid a divide in the inner loop
- if (sparse) {
- if (c->sorted_entries == 0) goto skip;
- c->multiplicands = (codetype *)setup_malloc(f, sizeof(c->multiplicands[0]) * c->sorted_entries * c->dimensions);
- }
- else
- c->multiplicands = (codetype *)setup_malloc(f, sizeof(c->multiplicands[0]) * c->entries * c->dimensions);
- if (c->multiplicands == NULL) { setup_temp_free(f, mults, sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_outofmem); }
- len = sparse ? c->sorted_entries : c->entries;
- for (j = 0; j < len; ++j) {
- unsigned int z = sparse ? c->sorted_values[j] : j;
- unsigned int div = 1;
- for (k = 0; k < c->dimensions; ++k) {
- int off = (z / div) % c->lookup_values;
- float val = mults[off];
- val = mults[off] * c->delta_value + c->minimum_value + last;
- c->multiplicands[j*c->dimensions + k] = val;
- if (c->sequence_p)
- last = val;
- if (k + 1 < c->dimensions) {
- if (div > UINT_MAX / (unsigned int)c->lookup_values) {
- setup_temp_free(f, mults, sizeof(mults[0])*c->lookup_values);
- return error(f, VORBIS_invalid_setup);
- }
- div *= c->lookup_values;
- }
- }
- }
- c->lookup_type = 2;
+ if (c->lookup_type == 1) {
+ int len, sparse = c->sparse;
+ float last=0;
+ // pre-expand the lookup1-style multiplicands, to avoid a divide in the inner loop
+ if (sparse) {
+ if (c->sorted_entries == 0) goto skip;
+ c->multiplicands = (codetype *) setup_malloc(f, sizeof(c->multiplicands[0]) * c->sorted_entries * c->dimensions);
+ } else
+ c->multiplicands = (codetype *) setup_malloc(f, sizeof(c->multiplicands[0]) * c->entries * c->dimensions);
+ if (c->multiplicands == NULL) { setup_temp_free(f,mults,sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_outofmem); }
+ len = sparse ? c->sorted_entries : c->entries;
+ for (j=0; j < len; ++j) {
+ unsigned int z = sparse ? c->sorted_values[j] : j;
+ unsigned int div=1;
+ for (k=0; k < c->dimensions; ++k) {
+ int off = (z / div) % c->lookup_values;
+ float val = mults[off]*c->delta_value + c->minimum_value + last;
+ c->multiplicands[j*c->dimensions + k] = val;
+ if (c->sequence_p)
+ last = val;
+ if (k+1 < c->dimensions) {
+ if (div > UINT_MAX / (unsigned int) c->lookup_values) {
+ setup_temp_free(f, mults,sizeof(mults[0])*c->lookup_values);
+ return error(f, VORBIS_invalid_setup);
+ }
+ div *= c->lookup_values;
+ }
+ }
}
- else
+ c->lookup_type = 2;
+ }
+ else
#endif
- {
- float last = 0;
- CHECK(f);
- c->multiplicands = (codetype *)setup_malloc(f, sizeof(c->multiplicands[0]) * c->lookup_values);
- if (c->multiplicands == NULL) { setup_temp_free(f, mults, sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_outofmem); }
- for (j = 0; j < (int)c->lookup_values; ++j) {
- float val = mults[j] * c->delta_value + c->minimum_value + last;
- c->multiplicands[j] = val;
- if (c->sequence_p)
- last = val;
- }
+ {
+ float last=0;
+ CHECK(f);
+ c->multiplicands = (codetype *) setup_malloc(f, sizeof(c->multiplicands[0]) * c->lookup_values);
+ if (c->multiplicands == NULL) { setup_temp_free(f, mults,sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_outofmem); }
+ for (j=0; j < (int) c->lookup_values; ++j) {
+ float val = mults[j] * c->delta_value + c->minimum_value + last;
+ c->multiplicands[j] = val;
+ if (c->sequence_p)
+ last = val;
}
+ }
#ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK
- skip : ;
+ skip:;
#endif
- setup_temp_free(f, mults, sizeof(mults[0])*c->lookup_values);
-
- CHECK(f);
- }
- CHECK(f);
- }
-
- // time domain transfers (notused)
-
- x = get_bits(f, 6) + 1;
- for (i = 0; i < x; ++i) {
- uint32 z = get_bits(f, 16);
- if (z != 0) return error(f, VORBIS_invalid_setup);
- }
-
- // Floors
- f->floor_count = get_bits(f, 6) + 1;
- f->floor_config = (Floor *)setup_malloc(f, f->floor_count * sizeof(*f->floor_config));
- if (f->floor_config == NULL) return error(f, VORBIS_outofmem);
- for (i = 0; i < f->floor_count; ++i) {
- f->floor_types[i] = get_bits(f, 16);
- if (f->floor_types[i] > 1) return error(f, VORBIS_invalid_setup);
- if (f->floor_types[i] == 0) {
- Floor0 *g = &f->floor_config[i].floor0;
- g->order = get_bits(f, 8);
- g->rate = get_bits(f, 16);
- g->bark_map_size = get_bits(f, 16);
- g->amplitude_bits = get_bits(f, 6);
- g->amplitude_offset = get_bits(f, 8);
- g->number_of_books = get_bits(f, 4) + 1;
- for (j = 0; j < g->number_of_books; ++j)
- g->book_list[j] = get_bits(f, 8);
- return error(f, VORBIS_feature_not_supported);
- }
- else {
- stbv__floor_ordering p[31 * 8 + 2];
- Floor1 *g = &f->floor_config[i].floor1;
- int max_class = -1;
- g->partitions = get_bits(f, 5);
- for (j = 0; j < g->partitions; ++j) {
- g->partition_class_list[j] = get_bits(f, 4);
- if (g->partition_class_list[j] > max_class)
- max_class = g->partition_class_list[j];
- }
- for (j = 0; j <= max_class; ++j) {
- g->class_dimensions[j] = get_bits(f, 3) + 1;
- g->class_subclasses[j] = get_bits(f, 2);
- if (g->class_subclasses[j]) {
- g->class_masterbooks[j] = get_bits(f, 8);
- if (g->class_masterbooks[j] >= f->codebook_count) return error(f, VORBIS_invalid_setup);
- }
- for (k = 0; k < 1 << g->class_subclasses[j]; ++k) {
- g->subclass_books[j][k] = get_bits(f, 8) - 1;
- if (g->subclass_books[j][k] >= f->codebook_count) return error(f, VORBIS_invalid_setup);
- }
- }
- g->floor1_multiplier = get_bits(f, 2) + 1;
- g->rangebits = get_bits(f, 4);
- g->Xlist[0] = 0;
- g->Xlist[1] = 1 << g->rangebits;
- g->values = 2;
- for (j = 0; j < g->partitions; ++j) {
- int c = g->partition_class_list[j];
- for (k = 0; k < g->class_dimensions[c]; ++k) {
- g->Xlist[g->values] = get_bits(f, g->rangebits);
- ++g->values;
- }
- }
- // precompute the sorting
- for (j = 0; j < g->values; ++j) {
- p[j].x = g->Xlist[j];
- p[j].id = j;
- }
- qsort(p, g->values, sizeof(p[0]), point_compare);
- for (j = 0; j < g->values; ++j)
- g->sorted_order[j] = (uint8)p[j].id;
- // precompute the neighbors
- for (j = 2; j < g->values; ++j) {
- int low, hi;
- neighbors(g->Xlist, j, &low, &hi);
- g->neighbors[j][0] = low;
- g->neighbors[j][1] = hi;
- }
+ setup_temp_free(f, mults, sizeof(mults[0])*c->lookup_values);
+
+ CHECK(f);
+ }
+ CHECK(f);
+ }
+
+ // time domain transfers (notused)
+
+ x = get_bits(f, 6) + 1;
+ for (i=0; i < x; ++i) {
+ uint32 z = get_bits(f, 16);
+ if (z != 0) return error(f, VORBIS_invalid_setup);
+ }
- if (g->values > longest_floorlist)
- longest_floorlist = g->values;
- }
- }
-
- // Residue
- f->residue_count = get_bits(f, 6) + 1;
- f->residue_config = (Residue *)setup_malloc(f, f->residue_count * sizeof(f->residue_config[0]));
- if (f->residue_config == NULL) return error(f, VORBIS_outofmem);
- memset(f->residue_config, 0, f->residue_count * sizeof(f->residue_config[0]));
- for (i = 0; i < f->residue_count; ++i) {
- uint8 residue_cascade[64];
- Residue *r = f->residue_config + i;
- f->residue_types[i] = get_bits(f, 16);
- if (f->residue_types[i] > 2) return error(f, VORBIS_invalid_setup);
- r->begin = get_bits(f, 24);
- r->end = get_bits(f, 24);
- if (r->end < r->begin) return error(f, VORBIS_invalid_setup);
- r->part_size = get_bits(f, 24) + 1;
- r->classifications = get_bits(f, 6) + 1;
- r->classbook = get_bits(f, 8);
- if (r->classbook >= f->codebook_count) return error(f, VORBIS_invalid_setup);
- for (j = 0; j < r->classifications; ++j) {
- uint8 high_bits = 0;
- uint8 low_bits = get_bits(f, 3);
- if (get_bits(f, 1))
- high_bits = get_bits(f, 5);
- residue_cascade[j] = high_bits * 8 + low_bits;
- }
- r->residue_books = (short(*)[8]) setup_malloc(f, sizeof(r->residue_books[0]) * r->classifications);
- if (r->residue_books == NULL) return error(f, VORBIS_outofmem);
- for (j = 0; j < r->classifications; ++j) {
- for (k = 0; k < 8; ++k) {
- if (residue_cascade[j] & (1 << k)) {
- r->residue_books[j][k] = get_bits(f, 8);
- if (r->residue_books[j][k] >= f->codebook_count) return error(f, VORBIS_invalid_setup);
- }
- else {
- r->residue_books[j][k] = -1;
- }
+ // Floors
+ f->floor_count = get_bits(f, 6)+1;
+ f->floor_config = (Floor *) setup_malloc(f, f->floor_count * sizeof(*f->floor_config));
+ if (f->floor_config == NULL) return error(f, VORBIS_outofmem);
+ for (i=0; i < f->floor_count; ++i) {
+ f->floor_types[i] = get_bits(f, 16);
+ if (f->floor_types[i] > 1) return error(f, VORBIS_invalid_setup);
+ if (f->floor_types[i] == 0) {
+ Floor0 *g = &f->floor_config[i].floor0;
+ g->order = get_bits(f,8);
+ g->rate = get_bits(f,16);
+ g->bark_map_size = get_bits(f,16);
+ g->amplitude_bits = get_bits(f,6);
+ g->amplitude_offset = get_bits(f,8);
+ g->number_of_books = get_bits(f,4) + 1;
+ for (j=0; j < g->number_of_books; ++j)
+ g->book_list[j] = get_bits(f,8);
+ return error(f, VORBIS_feature_not_supported);
+ } else {
+ stbv__floor_ordering p[31*8+2];
+ Floor1 *g = &f->floor_config[i].floor1;
+ int max_class = -1;
+ g->partitions = get_bits(f, 5);
+ for (j=0; j < g->partitions; ++j) {
+ g->partition_class_list[j] = get_bits(f, 4);
+ if (g->partition_class_list[j] > max_class)
+ max_class = g->partition_class_list[j];
+ }
+ for (j=0; j <= max_class; ++j) {
+ g->class_dimensions[j] = get_bits(f, 3)+1;
+ g->class_subclasses[j] = get_bits(f, 2);
+ if (g->class_subclasses[j]) {
+ g->class_masterbooks[j] = get_bits(f, 8);
+ if (g->class_masterbooks[j] >= f->codebook_count) return error(f, VORBIS_invalid_setup);
}
- }
- // precompute the classifications[] array to avoid inner-loop mod/divide
- // call it 'classdata' since we already have r->classifications
- r->classdata = (uint8 **)setup_malloc(f, sizeof(*r->classdata) * f->codebooks[r->classbook].entries);
- if (!r->classdata) return error(f, VORBIS_outofmem);
- memset(r->classdata, 0, sizeof(*r->classdata) * f->codebooks[r->classbook].entries);
- for (j = 0; j < f->codebooks[r->classbook].entries; ++j) {
- int classwords = f->codebooks[r->classbook].dimensions;
- int temp = j;
- r->classdata[j] = (uint8 *)setup_malloc(f, sizeof(r->classdata[j][0]) * classwords);
- if (r->classdata[j] == NULL) return error(f, VORBIS_outofmem);
- for (k = classwords - 1; k >= 0; --k) {
- r->classdata[j][k] = temp % r->classifications;
- temp /= r->classifications;
+ for (k=0; k < 1 << g->class_subclasses[j]; ++k) {
+ g->subclass_books[j][k] = (int16)get_bits(f,8)-1;
+ if (g->subclass_books[j][k] >= f->codebook_count) return error(f, VORBIS_invalid_setup);
}
- }
- }
-
- f->mapping_count = get_bits(f, 6) + 1;
- f->mapping = (Mapping *)setup_malloc(f, f->mapping_count * sizeof(*f->mapping));
- if (f->mapping == NULL) return error(f, VORBIS_outofmem);
- memset(f->mapping, 0, f->mapping_count * sizeof(*f->mapping));
- for (i = 0; i < f->mapping_count; ++i) {
- Mapping *m = f->mapping + i;
- int mapping_type = get_bits(f, 16);
- if (mapping_type != 0) return error(f, VORBIS_invalid_setup);
- m->chan = (MappingChannel *)setup_malloc(f, f->channels * sizeof(*m->chan));
- if (m->chan == NULL) return error(f, VORBIS_outofmem);
- if (get_bits(f, 1))
- m->submaps = get_bits(f, 4) + 1;
- else
- m->submaps = 1;
- if (m->submaps > max_submaps)
- max_submaps = m->submaps;
- if (get_bits(f, 1)) {
- m->coupling_steps = get_bits(f, 8) + 1;
- for (k = 0; k < m->coupling_steps; ++k) {
- m->chan[k].magnitude = get_bits(f, ilog(f->channels - 1));
- m->chan[k].angle = get_bits(f, ilog(f->channels - 1));
- if (m->chan[k].magnitude >= f->channels) return error(f, VORBIS_invalid_setup);
- if (m->chan[k].angle >= f->channels) return error(f, VORBIS_invalid_setup);
- if (m->chan[k].magnitude == m->chan[k].angle) return error(f, VORBIS_invalid_setup);
+ }
+ g->floor1_multiplier = get_bits(f,2)+1;
+ g->rangebits = get_bits(f,4);
+ g->Xlist[0] = 0;
+ g->Xlist[1] = 1 << g->rangebits;
+ g->values = 2;
+ for (j=0; j < g->partitions; ++j) {
+ int c = g->partition_class_list[j];
+ for (k=0; k < g->class_dimensions[c]; ++k) {
+ g->Xlist[g->values] = get_bits(f, g->rangebits);
+ ++g->values;
}
- }
- else
- m->coupling_steps = 0;
-
- // reserved field
- if (get_bits(f, 2)) return error(f, VORBIS_invalid_setup);
- if (m->submaps > 1) {
- for (j = 0; j < f->channels; ++j) {
- m->chan[j].mux = get_bits(f, 4);
- if (m->chan[j].mux >= m->submaps) return error(f, VORBIS_invalid_setup);
+ }
+ // precompute the sorting
+ for (j=0; j < g->values; ++j) {
+ p[j].x = g->Xlist[j];
+ p[j].id = j;
+ }
+ qsort(p, g->values, sizeof(p[0]), point_compare);
+ for (j=0; j < g->values-1; ++j)
+ if (p[j].x == p[j+1].x)
+ return error(f, VORBIS_invalid_setup);
+ for (j=0; j < g->values; ++j)
+ g->sorted_order[j] = (uint8) p[j].id;
+ // precompute the neighbors
+ for (j=2; j < g->values; ++j) {
+ int low = 0,hi = 0;
+ neighbors(g->Xlist, j, &low,&hi);
+ g->neighbors[j][0] = low;
+ g->neighbors[j][1] = hi;
+ }
+
+ if (g->values > longest_floorlist)
+ longest_floorlist = g->values;
+ }
+ }
+
+ // Residue
+ f->residue_count = get_bits(f, 6)+1;
+ f->residue_config = (Residue *) setup_malloc(f, f->residue_count * sizeof(f->residue_config[0]));
+ if (f->residue_config == NULL) return error(f, VORBIS_outofmem);
+ memset(f->residue_config, 0, f->residue_count * sizeof(f->residue_config[0]));
+ for (i=0; i < f->residue_count; ++i) {
+ uint8 residue_cascade[64];
+ Residue *r = f->residue_config+i;
+ f->residue_types[i] = get_bits(f, 16);
+ if (f->residue_types[i] > 2) return error(f, VORBIS_invalid_setup);
+ r->begin = get_bits(f, 24);
+ r->end = get_bits(f, 24);
+ if (r->end < r->begin) return error(f, VORBIS_invalid_setup);
+ r->part_size = get_bits(f,24)+1;
+ r->classifications = get_bits(f,6)+1;
+ r->classbook = get_bits(f,8);
+ if (r->classbook >= f->codebook_count) return error(f, VORBIS_invalid_setup);
+ for (j=0; j < r->classifications; ++j) {
+ uint8 high_bits=0;
+ uint8 low_bits=get_bits(f,3);
+ if (get_bits(f,1))
+ high_bits = get_bits(f,5);
+ residue_cascade[j] = high_bits*8 + low_bits;
+ }
+ r->residue_books = (short (*)[8]) setup_malloc(f, sizeof(r->residue_books[0]) * r->classifications);
+ if (r->residue_books == NULL) return error(f, VORBIS_outofmem);
+ for (j=0; j < r->classifications; ++j) {
+ for (k=0; k < 8; ++k) {
+ if (residue_cascade[j] & (1 << k)) {
+ r->residue_books[j][k] = get_bits(f, 8);
+ if (r->residue_books[j][k] >= f->codebook_count) return error(f, VORBIS_invalid_setup);
+ } else {
+ r->residue_books[j][k] = -1;
}
- }
- else
- // @SPECIFICATION: this case is missing from the spec
- for (j = 0; j < f->channels; ++j)
- m->chan[j].mux = 0;
-
- for (j = 0; j < m->submaps; ++j) {
- get_bits(f, 8); // discard
- m->submap_floor[j] = get_bits(f, 8);
- m->submap_residue[j] = get_bits(f, 8);
- if (m->submap_floor[j] >= f->floor_count) return error(f, VORBIS_invalid_setup);
- if (m->submap_residue[j] >= f->residue_count) return error(f, VORBIS_invalid_setup);
- }
- }
-
- // Modes
- f->mode_count = get_bits(f, 6) + 1;
- for (i = 0; i < f->mode_count; ++i) {
- Mode *m = f->mode_config + i;
- m->blockflag = get_bits(f, 1);
- m->windowtype = get_bits(f, 16);
- m->transformtype = get_bits(f, 16);
- m->mapping = get_bits(f, 8);
- if (m->windowtype != 0) return error(f, VORBIS_invalid_setup);
- if (m->transformtype != 0) return error(f, VORBIS_invalid_setup);
- if (m->mapping >= f->mapping_count) return error(f, VORBIS_invalid_setup);
- }
-
- flush_packet(f);
-
- f->previous_length = 0;
-
- for (i = 0; i < f->channels; ++i) {
- f->channel_buffers[i] = (float *)setup_malloc(f, sizeof(float) * f->blocksize_1);
- f->previous_window[i] = (float *)setup_malloc(f, sizeof(float) * f->blocksize_1 / 2);
- f->finalY[i] = (int16 *)setup_malloc(f, sizeof(int16) * longest_floorlist);
- if (f->channel_buffers[i] == NULL || f->previous_window[i] == NULL || f->finalY[i] == NULL) return error(f, VORBIS_outofmem);
- memset(f->channel_buffers[i], 0, sizeof(float) * f->blocksize_1);
-#ifdef STB_VORBIS_NO_DEFER_FLOOR
- f->floor_buffers[i] = (float *)setup_malloc(f, sizeof(float) * f->blocksize_1 / 2);
- if (f->floor_buffers[i] == NULL) return error(f, VORBIS_outofmem);
-#endif
- }
+ }
+ }
+ // precompute the classifications[] array to avoid inner-loop mod/divide
+ // call it 'classdata' since we already have r->classifications
+ r->classdata = (uint8 **) setup_malloc(f, sizeof(*r->classdata) * f->codebooks[r->classbook].entries);
+ if (!r->classdata) return error(f, VORBIS_outofmem);
+ memset(r->classdata, 0, sizeof(*r->classdata) * f->codebooks[r->classbook].entries);
+ for (j=0; j < f->codebooks[r->classbook].entries; ++j) {
+ int classwords = f->codebooks[r->classbook].dimensions;
+ int temp = j;
+ r->classdata[j] = (uint8 *) setup_malloc(f, sizeof(r->classdata[j][0]) * classwords);
+ if (r->classdata[j] == NULL) return error(f, VORBIS_outofmem);
+ for (k=classwords-1; k >= 0; --k) {
+ r->classdata[j][k] = temp % r->classifications;
+ temp /= r->classifications;
+ }
+ }
+ }
- if (!init_blocksize(f, 0, f->blocksize_0)) return FALSE;
- if (!init_blocksize(f, 1, f->blocksize_1)) return FALSE;
- f->blocksize[0] = f->blocksize_0;
- f->blocksize[1] = f->blocksize_1;
+ f->mapping_count = get_bits(f,6)+1;
+ f->mapping = (Mapping *) setup_malloc(f, f->mapping_count * sizeof(*f->mapping));
+ if (f->mapping == NULL) return error(f, VORBIS_outofmem);
+ memset(f->mapping, 0, f->mapping_count * sizeof(*f->mapping));
+ for (i=0; i < f->mapping_count; ++i) {
+ Mapping *m = f->mapping + i;
+ int mapping_type = get_bits(f,16);
+ if (mapping_type != 0) return error(f, VORBIS_invalid_setup);
+ m->chan = (MappingChannel *) setup_malloc(f, f->channels * sizeof(*m->chan));
+ if (m->chan == NULL) return error(f, VORBIS_outofmem);
+ if (get_bits(f,1))
+ m->submaps = get_bits(f,4)+1;
+ else
+ m->submaps = 1;
+ if (m->submaps > max_submaps)
+ max_submaps = m->submaps;
+ if (get_bits(f,1)) {
+ m->coupling_steps = get_bits(f,8)+1;
+ if (m->coupling_steps > f->channels) return error(f, VORBIS_invalid_setup);
+ for (k=0; k < m->coupling_steps; ++k) {
+ m->chan[k].magnitude = get_bits(f, ilog(f->channels-1));
+ m->chan[k].angle = get_bits(f, ilog(f->channels-1));
+ if (m->chan[k].magnitude >= f->channels) return error(f, VORBIS_invalid_setup);
+ if (m->chan[k].angle >= f->channels) return error(f, VORBIS_invalid_setup);
+ if (m->chan[k].magnitude == m->chan[k].angle) return error(f, VORBIS_invalid_setup);
+ }
+ } else
+ m->coupling_steps = 0;
+
+ // reserved field
+ if (get_bits(f,2)) return error(f, VORBIS_invalid_setup);
+ if (m->submaps > 1) {
+ for (j=0; j < f->channels; ++j) {
+ m->chan[j].mux = get_bits(f, 4);
+ if (m->chan[j].mux >= m->submaps) return error(f, VORBIS_invalid_setup);
+ }
+ } else
+ // @SPECIFICATION: this case is missing from the spec
+ for (j=0; j < f->channels; ++j)
+ m->chan[j].mux = 0;
+
+ for (j=0; j < m->submaps; ++j) {
+ get_bits(f,8); // discard
+ m->submap_floor[j] = get_bits(f,8);
+ m->submap_residue[j] = get_bits(f,8);
+ if (m->submap_floor[j] >= f->floor_count) return error(f, VORBIS_invalid_setup);
+ if (m->submap_residue[j] >= f->residue_count) return error(f, VORBIS_invalid_setup);
+ }
+ }
-#ifdef STB_VORBIS_DIVIDE_TABLE
- if (integer_divide_table[1][1] == 0)
- for (i = 0; i < DIVTAB_NUMER; ++i)
- for (j = 1; j < DIVTAB_DENOM; ++j)
- integer_divide_table[i][j] = i / j;
-#endif
+ // Modes
+ f->mode_count = get_bits(f, 6)+1;
+ for (i=0; i < f->mode_count; ++i) {
+ Mode *m = f->mode_config+i;
+ m->blockflag = get_bits(f,1);
+ m->windowtype = get_bits(f,16);
+ m->transformtype = get_bits(f,16);
+ m->mapping = get_bits(f,8);
+ if (m->windowtype != 0) return error(f, VORBIS_invalid_setup);
+ if (m->transformtype != 0) return error(f, VORBIS_invalid_setup);
+ if (m->mapping >= f->mapping_count) return error(f, VORBIS_invalid_setup);
+ }
- // compute how much temporary memory is needed
-
- // 1.
- {
- uint32 imdct_mem = (f->blocksize_1 * sizeof(float) >> 1);
- uint32 classify_mem;
- int i, max_part_read = 0;
- for (i = 0; i < f->residue_count; ++i) {
- Residue *r = f->residue_config + i;
- unsigned int actual_size = f->blocksize_1 / 2;
- unsigned int limit_r_begin = r->begin < actual_size ? r->begin : actual_size;
- unsigned int limit_r_end = r->end < actual_size ? r->end : actual_size;
- int n_read = limit_r_end - limit_r_begin;
- int part_read = n_read / r->part_size;
- if (part_read > max_part_read)
- max_part_read = part_read;
- }
-#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
- classify_mem = f->channels * (sizeof(void*) + max_part_read * sizeof(uint8 *));
-#else
- classify_mem = f->channels * (sizeof(void*) + max_part_read * sizeof(int *));
-#endif
+ flush_packet(f);
+
+ f->previous_length = 0;
+
+ for (i=0; i < f->channels; ++i) {
+ f->channel_buffers[i] = (float *) setup_malloc(f, sizeof(float) * f->blocksize_1);
+ f->previous_window[i] = (float *) setup_malloc(f, sizeof(float) * f->blocksize_1/2);
+ f->finalY[i] = (int16 *) setup_malloc(f, sizeof(int16) * longest_floorlist);
+ if (f->channel_buffers[i] == NULL || f->previous_window[i] == NULL || f->finalY[i] == NULL) return error(f, VORBIS_outofmem);
+ memset(f->channel_buffers[i], 0, sizeof(float) * f->blocksize_1);
+ #ifdef STB_VORBIS_NO_DEFER_FLOOR
+ f->floor_buffers[i] = (float *) setup_malloc(f, sizeof(float) * f->blocksize_1/2);
+ if (f->floor_buffers[i] == NULL) return error(f, VORBIS_outofmem);
+ #endif
+ }
- // maximum reasonable partition size is f->blocksize_1
+ if (!init_blocksize(f, 0, f->blocksize_0)) return FALSE;
+ if (!init_blocksize(f, 1, f->blocksize_1)) return FALSE;
+ f->blocksize[0] = f->blocksize_0;
+ f->blocksize[1] = f->blocksize_1;
- f->temp_memory_required = classify_mem;
- if (imdct_mem > f->temp_memory_required)
- f->temp_memory_required = imdct_mem;
- }
+#ifdef STB_VORBIS_DIVIDE_TABLE
+ if (integer_divide_table[1][1]==0)
+ for (i=0; i < DIVTAB_NUMER; ++i)
+ for (j=1; j < DIVTAB_DENOM; ++j)
+ integer_divide_table[i][j] = i / j;
+#endif
- f->first_decode = TRUE;
+ // compute how much temporary memory is needed
+
+ // 1.
+ {
+ uint32 imdct_mem = (f->blocksize_1 * sizeof(float) >> 1);
+ uint32 classify_mem;
+ int i,max_part_read=0;
+ for (i=0; i < f->residue_count; ++i) {
+ Residue *r = f->residue_config + i;
+ unsigned int actual_size = f->blocksize_1 / 2;
+ unsigned int limit_r_begin = r->begin < actual_size ? r->begin : actual_size;
+ unsigned int limit_r_end = r->end < actual_size ? r->end : actual_size;
+ int n_read = limit_r_end - limit_r_begin;
+ int part_read = n_read / r->part_size;
+ if (part_read > max_part_read)
+ max_part_read = part_read;
+ }
+ #ifndef STB_VORBIS_DIVIDES_IN_RESIDUE
+ classify_mem = f->channels * (sizeof(void*) + max_part_read * sizeof(uint8 *));
+ #else
+ classify_mem = f->channels * (sizeof(void*) + max_part_read * sizeof(int *));
+ #endif
+
+ // maximum reasonable partition size is f->blocksize_1
+
+ f->temp_memory_required = classify_mem;
+ if (imdct_mem > f->temp_memory_required)
+ f->temp_memory_required = imdct_mem;
+ }
- if (f->alloc.alloc_buffer) {
- assert(f->temp_offset == f->alloc.alloc_buffer_length_in_bytes);
- // check if there's enough temp memory so we don't error later
- if (f->setup_offset + sizeof(*f) + f->temp_memory_required > (unsigned)f->temp_offset)
- return error(f, VORBIS_outofmem);
- }
- f->first_audio_page_offset = stb_vorbis_get_file_offset(f);
+ if (f->alloc.alloc_buffer) {
+ assert(f->temp_offset == f->alloc.alloc_buffer_length_in_bytes);
+ // check if there's enough temp memory so we don't error later
+ if (f->setup_offset + sizeof(*f) + f->temp_memory_required > (unsigned) f->temp_offset)
+ return error(f, VORBIS_outofmem);
+ }
- return TRUE;
+ // @TODO: stb_vorbis_seek_start expects first_audio_page_offset to point to a page
+ // without PAGEFLAG_continued_packet, so this either points to the first page, or
+ // the page after the end of the headers. It might be cleaner to point to a page
+ // in the middle of the headers, when that's the page where the first audio packet
+ // starts, but we'd have to also correctly skip the end of any continued packet in
+ // stb_vorbis_seek_start.
+ if (f->next_seg == -1) {
+ f->first_audio_page_offset = stb_vorbis_get_file_offset(f);
+ } else {
+ f->first_audio_page_offset = 0;
+ }
+
+ return TRUE;
}
static void vorbis_deinit(stb_vorbis *p)
{
- int i, j;
- if (p->residue_config) {
- for (i = 0; i < p->residue_count; ++i) {
- Residue *r = p->residue_config + i;
- if (r->classdata) {
- for (j = 0; j < p->codebooks[r->classbook].entries; ++j)
- setup_free(p, r->classdata[j]);
- setup_free(p, r->classdata);
- }
- setup_free(p, r->residue_books);
- }
- }
-
- if (p->codebooks) {
- CHECK(p);
- for (i = 0; i < p->codebook_count; ++i) {
- Codebook *c = p->codebooks + i;
- setup_free(p, c->codeword_lengths);
- setup_free(p, c->multiplicands);
- setup_free(p, c->codewords);
- setup_free(p, c->sorted_codewords);
- // c->sorted_values[-1] is the first entry in the array
- setup_free(p, c->sorted_values ? c->sorted_values - 1 : NULL);
- }
- setup_free(p, p->codebooks);
- }
- setup_free(p, p->floor_config);
- setup_free(p, p->residue_config);
- if (p->mapping) {
- for (i = 0; i < p->mapping_count; ++i)
- setup_free(p, p->mapping[i].chan);
- setup_free(p, p->mapping);
- }
- CHECK(p);
- for (i = 0; i < p->channels && i < STB_VORBIS_MAX_CHANNELS; ++i) {
- setup_free(p, p->channel_buffers[i]);
- setup_free(p, p->previous_window[i]);
-#ifdef STB_VORBIS_NO_DEFER_FLOOR
- setup_free(p, p->floor_buffers[i]);
-#endif
- setup_free(p, p->finalY[i]);
- }
- for (i = 0; i < 2; ++i) {
- setup_free(p, p->A[i]);
- setup_free(p, p->B[i]);
- setup_free(p, p->C[i]);
- setup_free(p, p->window[i]);
- setup_free(p, p->bit_reverse[i]);
- }
-#ifndef STB_VORBIS_NO_STDIO
- if (p->close_on_free) fclose(p->f);
-#endif
+ int i,j;
+
+ setup_free(p, p->vendor);
+ for (i=0; i < p->comment_list_length; ++i) {
+ setup_free(p, p->comment_list[i]);
+ }
+ setup_free(p, p->comment_list);
+
+ if (p->residue_config) {
+ for (i=0; i < p->residue_count; ++i) {
+ Residue *r = p->residue_config+i;
+ if (r->classdata) {
+ for (j=0; j < p->codebooks[r->classbook].entries; ++j)
+ setup_free(p, r->classdata[j]);
+ setup_free(p, r->classdata);
+ }
+ setup_free(p, r->residue_books);
+ }
+ }
+
+ if (p->codebooks) {
+ CHECK(p);
+ for (i=0; i < p->codebook_count; ++i) {
+ Codebook *c = p->codebooks + i;
+ setup_free(p, c->codeword_lengths);
+ setup_free(p, c->multiplicands);
+ setup_free(p, c->codewords);
+ setup_free(p, c->sorted_codewords);
+ // c->sorted_values[-1] is the first entry in the array
+ setup_free(p, c->sorted_values ? c->sorted_values-1 : NULL);
+ }
+ setup_free(p, p->codebooks);
+ }
+ setup_free(p, p->floor_config);
+ setup_free(p, p->residue_config);
+ if (p->mapping) {
+ for (i=0; i < p->mapping_count; ++i)
+ setup_free(p, p->mapping[i].chan);
+ setup_free(p, p->mapping);
+ }
+ CHECK(p);
+ for (i=0; i < p->channels && i < STB_VORBIS_MAX_CHANNELS; ++i) {
+ setup_free(p, p->channel_buffers[i]);
+ setup_free(p, p->previous_window[i]);
+ #ifdef STB_VORBIS_NO_DEFER_FLOOR
+ setup_free(p, p->floor_buffers[i]);
+ #endif
+ setup_free(p, p->finalY[i]);
+ }
+ for (i=0; i < 2; ++i) {
+ setup_free(p, p->A[i]);
+ setup_free(p, p->B[i]);
+ setup_free(p, p->C[i]);
+ setup_free(p, p->window[i]);
+ setup_free(p, p->bit_reverse[i]);
+ }
+ #ifndef STB_VORBIS_NO_STDIO
+ if (p->close_on_free) fclose(p->f);
+ #endif
}
void stb_vorbis_close(stb_vorbis *p)
{
- if (p == NULL) return;
- vorbis_deinit(p);
- setup_free(p, p);
+ if (p == NULL) return;
+ vorbis_deinit(p);
+ setup_free(p,p);
}
static void vorbis_init(stb_vorbis *p, const stb_vorbis_alloc *z)
{
- memset(p, 0, sizeof(*p)); // NULL out all malloc'd pointers to start
- if (z) {
- p->alloc = *z;
- p->alloc.alloc_buffer_length_in_bytes = (p->alloc.alloc_buffer_length_in_bytes + 3) & ~3;
- p->temp_offset = p->alloc.alloc_buffer_length_in_bytes;
- }
- p->eof = 0;
- p->error = VORBIS__no_error;
- p->stream = NULL;
- p->codebooks = NULL;
- p->page_crc_tests = -1;
-#ifndef STB_VORBIS_NO_STDIO
- p->close_on_free = FALSE;
- p->f = NULL;
-#endif
+ memset(p, 0, sizeof(*p)); // NULL out all malloc'd pointers to start
+ if (z) {
+ p->alloc = *z;
+ p->alloc.alloc_buffer_length_in_bytes &= ~7;
+ p->temp_offset = p->alloc.alloc_buffer_length_in_bytes;
+ }
+ p->eof = 0;
+ p->error = VORBIS__no_error;
+ p->stream = NULL;
+ p->codebooks = NULL;
+ p->page_crc_tests = -1;
+ #ifndef STB_VORBIS_NO_STDIO
+ p->close_on_free = FALSE;
+ p->f = NULL;
+ #endif
}
int stb_vorbis_get_sample_offset(stb_vorbis *f)
{
- if (f->current_loc_valid)
- return f->current_loc;
- else
- return -1;
+ if (f->current_loc_valid)
+ return f->current_loc;
+ else
+ return -1;
}
stb_vorbis_info stb_vorbis_get_info(stb_vorbis *f)
{
- stb_vorbis_info d;
- d.channels = f->channels;
- d.sample_rate = f->sample_rate;
- d.setup_memory_required = f->setup_memory_required;
- d.setup_temp_memory_required = f->setup_temp_memory_required;
- d.temp_memory_required = f->temp_memory_required;
- d.max_frame_size = f->blocksize_1 >> 1;
- return d;
+ stb_vorbis_info d;
+ d.channels = f->channels;
+ d.sample_rate = f->sample_rate;
+ d.setup_memory_required = f->setup_memory_required;
+ d.setup_temp_memory_required = f->setup_temp_memory_required;
+ d.temp_memory_required = f->temp_memory_required;
+ d.max_frame_size = f->blocksize_1 >> 1;
+ return d;
+}
+
+stb_vorbis_comment stb_vorbis_get_comment(stb_vorbis *f)
+{
+ stb_vorbis_comment d;
+ d.vendor = f->vendor;
+ d.comment_list_length = f->comment_list_length;
+ d.comment_list = f->comment_list;
+ return d;
}
int stb_vorbis_get_error(stb_vorbis *f)
{
- int e = f->error;
- f->error = VORBIS__no_error;
- return e;
+ int e = f->error;
+ f->error = VORBIS__no_error;
+ return e;
}
static stb_vorbis * vorbis_alloc(stb_vorbis *f)
{
- stb_vorbis *p = (stb_vorbis *)setup_malloc(f, sizeof(*p));
- return p;
+ stb_vorbis *p = (stb_vorbis *) setup_malloc(f, sizeof(*p));
+ return p;
}
#ifndef STB_VORBIS_NO_PUSHDATA_API
void stb_vorbis_flush_pushdata(stb_vorbis *f)
{
- f->previous_length = 0;
- f->page_crc_tests = 0;
- f->discard_samples_deferred = 0;
- f->current_loc_valid = FALSE;
- f->first_decode = FALSE;
- f->samples_output = 0;
- f->channel_buffer_start = 0;
- f->channel_buffer_end = 0;
+ f->previous_length = 0;
+ f->page_crc_tests = 0;
+ f->discard_samples_deferred = 0;
+ f->current_loc_valid = FALSE;
+ f->first_decode = FALSE;
+ f->samples_output = 0;
+ f->channel_buffer_start = 0;
+ f->channel_buffer_end = 0;
}
static int vorbis_search_for_page_pushdata(vorb *f, uint8 *data, int data_len)
{
- int i, n;
- for (i = 0; i < f->page_crc_tests; ++i)
- f->scan[i].bytes_done = 0;
-
- // if we have room for more scans, search for them first, because
- // they may cause us to stop early if their header is incomplete
- if (f->page_crc_tests < STB_VORBIS_PUSHDATA_CRC_COUNT) {
- if (data_len < 4) return 0;
- data_len -= 3; // need to look for 4-byte sequence, so don't miss
- // one that straddles a boundary
- for (i = 0; i < data_len; ++i) {
- if (data[i] == 0x4f) {
- if (0 == memcmp(data + i, ogg_page_header, 4)) {
- int j, len;
- uint32 crc;
- // make sure we have the whole page header
- if (i + 26 >= data_len || i + 27 + data[i + 26] >= data_len) {
- // only read up to this page start, so hopefully we'll
- // have the whole page header start next time
- data_len = i;
- break;
- }
- // ok, we have it all; compute the length of the page
- len = 27 + data[i + 26];
- for (j = 0; j < data[i + 26]; ++j)
- len += data[i + 27 + j];
- // scan everything up to the embedded crc (which we must 0)
- crc = 0;
- for (j = 0; j < 22; ++j)
- crc = crc32_update(crc, data[i + j]);
- // now process 4 0-bytes
- for (; j < 26; ++j)
- crc = crc32_update(crc, 0);
- // len is the total number of bytes we need to scan
- n = f->page_crc_tests++;
- f->scan[n].bytes_left = len - j;
- f->scan[n].crc_so_far = crc;
- f->scan[n].goal_crc = data[i + 22] + (data[i + 23] << 8) + (data[i + 24] << 16) + (data[i + 25] << 24);
- // if the last frame on a page is continued to the next, then
- // we can't recover the sample_loc immediately
- if (data[i + 27 + data[i + 26] - 1] == 255)
- f->scan[n].sample_loc = ~0;
- else
- f->scan[n].sample_loc = data[i + 6] + (data[i + 7] << 8) + (data[i + 8] << 16) + (data[i + 9] << 24);
- f->scan[n].bytes_done = i + j;
- if (f->page_crc_tests == STB_VORBIS_PUSHDATA_CRC_COUNT)
- break;
- // keep going if we still have room for more
- }
+ int i,n;
+ for (i=0; i < f->page_crc_tests; ++i)
+ f->scan[i].bytes_done = 0;
+
+ // if we have room for more scans, search for them first, because
+ // they may cause us to stop early if their header is incomplete
+ if (f->page_crc_tests < STB_VORBIS_PUSHDATA_CRC_COUNT) {
+ if (data_len < 4) return 0;
+ data_len -= 3; // need to look for 4-byte sequence, so don't miss
+ // one that straddles a boundary
+ for (i=0; i < data_len; ++i) {
+ if (data[i] == 0x4f) {
+ if (0==memcmp(data+i, ogg_page_header, 4)) {
+ int j,len;
+ uint32 crc;
+ // make sure we have the whole page header
+ if (i+26 >= data_len || i+27+data[i+26] >= data_len) {
+ // only read up to this page start, so hopefully we'll
+ // have the whole page header start next time
+ data_len = i;
+ break;
+ }
+ // ok, we have it all; compute the length of the page
+ len = 27 + data[i+26];
+ for (j=0; j < data[i+26]; ++j)
+ len += data[i+27+j];
+ // scan everything up to the embedded crc (which we must 0)
+ crc = 0;
+ for (j=0; j < 22; ++j)
+ crc = crc32_update(crc, data[i+j]);
+ // now process 4 0-bytes
+ for ( ; j < 26; ++j)
+ crc = crc32_update(crc, 0);
+ // len is the total number of bytes we need to scan
+ n = f->page_crc_tests++;
+ f->scan[n].bytes_left = len-j;
+ f->scan[n].crc_so_far = crc;
+ f->scan[n].goal_crc = data[i+22] + (data[i+23] << 8) + (data[i+24]<<16) + (data[i+25]<<24);
+ // if the last frame on a page is continued to the next, then
+ // we can't recover the sample_loc immediately
+ if (data[i+27+data[i+26]-1] == 255)
+ f->scan[n].sample_loc = ~0;
+ else
+ f->scan[n].sample_loc = data[i+6] + (data[i+7] << 8) + (data[i+ 8]<<16) + (data[i+ 9]<<24);
+ f->scan[n].bytes_done = i+j;
+ if (f->page_crc_tests == STB_VORBIS_PUSHDATA_CRC_COUNT)
+ break;
+ // keep going if we still have room for more
}
- }
- }
-
- for (i = 0; i < f->page_crc_tests;) {
- uint32 crc;
- int j;
- int n = f->scan[i].bytes_done;
- int m = f->scan[i].bytes_left;
- if (m > data_len - n) m = data_len - n;
- // m is the bytes to scan in the current chunk
- crc = f->scan[i].crc_so_far;
- for (j = 0; j < m; ++j)
- crc = crc32_update(crc, data[n + j]);
- f->scan[i].bytes_left -= m;
- f->scan[i].crc_so_far = crc;
- if (f->scan[i].bytes_left == 0) {
- // does it match?
- if (f->scan[i].crc_so_far == f->scan[i].goal_crc) {
- // Houston, we have page
- data_len = n + m; // consumption amount is wherever that scan ended
- f->page_crc_tests = -1; // drop out of page scan mode
- f->previous_length = 0; // decode-but-don't-output one frame
- f->next_seg = -1; // start a new page
- f->current_loc = f->scan[i].sample_loc; // set the current sample location
- // to the amount we'd have decoded had we decoded this page
- f->current_loc_valid = f->current_loc != ~0U;
- return data_len;
- }
- // delete entry
- f->scan[i] = f->scan[--f->page_crc_tests];
- }
- else {
- ++i;
- }
- }
-
- return data_len;
+ }
+ }
+ }
+
+ for (i=0; i < f->page_crc_tests;) {
+ uint32 crc;
+ int j;
+ int n = f->scan[i].bytes_done;
+ int m = f->scan[i].bytes_left;
+ if (m > data_len - n) m = data_len - n;
+ // m is the bytes to scan in the current chunk
+ crc = f->scan[i].crc_so_far;
+ for (j=0; j < m; ++j)
+ crc = crc32_update(crc, data[n+j]);
+ f->scan[i].bytes_left -= m;
+ f->scan[i].crc_so_far = crc;
+ if (f->scan[i].bytes_left == 0) {
+ // does it match?
+ if (f->scan[i].crc_so_far == f->scan[i].goal_crc) {
+ // Houston, we have page
+ data_len = n+m; // consumption amount is wherever that scan ended
+ f->page_crc_tests = -1; // drop out of page scan mode
+ f->previous_length = 0; // decode-but-don't-output one frame
+ f->next_seg = -1; // start a new page
+ f->current_loc = f->scan[i].sample_loc; // set the current sample location
+ // to the amount we'd have decoded had we decoded this page
+ f->current_loc_valid = f->current_loc != ~0U;
+ return data_len;
+ }
+ // delete entry
+ f->scan[i] = f->scan[--f->page_crc_tests];
+ } else {
+ ++i;
+ }
+ }
+
+ return data_len;
}
// return value: number of bytes we used
int stb_vorbis_decode_frame_pushdata(
- stb_vorbis *f, // the file we're decoding
- const uint8 *data, int data_len, // the memory available for decoding
- int *channels, // place to write number of float * buffers
- float ***output, // place to write float ** array of float * buffers
- int *samples // place to write number of output samples
-)
-{
- int i;
- int len, right, left;
-
- if (!IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing);
-
- if (f->page_crc_tests >= 0) {
- *samples = 0;
- return vorbis_search_for_page_pushdata(f, (uint8 *)data, data_len);
- }
-
- f->stream = (uint8 *)data;
- f->stream_end = (uint8 *)data + data_len;
- f->error = VORBIS__no_error;
-
- // check that we have the entire packet in memory
- if (!is_whole_packet_present(f, FALSE)) {
- *samples = 0;
- return 0;
- }
-
- if (!vorbis_decode_packet(f, &len, &left, &right)) {
- // save the actual error we encountered
- enum STBVorbisError error = f->error;
- if (error == VORBIS_bad_packet_type) {
- // flush and resynch
+ stb_vorbis *f, // the file we're decoding
+ const uint8 *data, int data_len, // the memory available for decoding
+ int *channels, // place to write number of float * buffers
+ float ***output, // place to write float ** array of float * buffers
+ int *samples // place to write number of output samples
+ )
+{
+ int i;
+ int len,right,left;
+
+ if (!IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing);
+
+ if (f->page_crc_tests >= 0) {
+ *samples = 0;
+ return vorbis_search_for_page_pushdata(f, (uint8 *) data, data_len);
+ }
+
+ f->stream = (uint8 *) data;
+ f->stream_end = (uint8 *) data + data_len;
+ f->error = VORBIS__no_error;
+
+ // check that we have the entire packet in memory
+ if (!is_whole_packet_present(f)) {
+ *samples = 0;
+ return 0;
+ }
+
+ if (!vorbis_decode_packet(f, &len, &left, &right)) {
+ // save the actual error we encountered
+ enum STBVorbisError error = f->error;
+ if (error == VORBIS_bad_packet_type) {
+ // flush and resynch
+ f->error = VORBIS__no_error;
+ while (get8_packet(f) != EOP)
+ if (f->eof) break;
+ *samples = 0;
+ return (int) (f->stream - data);
+ }
+ if (error == VORBIS_continued_packet_flag_invalid) {
+ if (f->previous_length == 0) {
+ // we may be resynching, in which case it's ok to hit one
+ // of these; just discard the packet
f->error = VORBIS__no_error;
while (get8_packet(f) != EOP)
- if (f->eof) break;
+ if (f->eof) break;
*samples = 0;
- return (int)(f->stream - data);
- }
- if (error == VORBIS_continued_packet_flag_invalid) {
- if (f->previous_length == 0) {
- // we may be resynching, in which case it's ok to hit one
- // of these; just discard the packet
- f->error = VORBIS__no_error;
- while (get8_packet(f) != EOP)
- if (f->eof) break;
- *samples = 0;
- return (int)(f->stream - data);
- }
- }
- // if we get an error while parsing, what to do?
- // well, it DEFINITELY won't work to continue from where we are!
- stb_vorbis_flush_pushdata(f);
- // restore the error that actually made us bail
- f->error = error;
- *samples = 0;
- return 1;
- }
-
- // success!
- len = vorbis_finish_frame(f, len, left, right);
- for (i = 0; i < f->channels; ++i)
- f->outputs[i] = f->channel_buffers[i] + left;
-
- if (channels) *channels = f->channels;
- *samples = len;
- *output = f->outputs;
- return (int)(f->stream - data);
+ return (int) (f->stream - data);
+ }
+ }
+ // if we get an error while parsing, what to do?
+ // well, it DEFINITELY won't work to continue from where we are!
+ stb_vorbis_flush_pushdata(f);
+ // restore the error that actually made us bail
+ f->error = error;
+ *samples = 0;
+ return 1;
+ }
+
+ // success!
+ len = vorbis_finish_frame(f, len, left, right);
+ for (i=0; i < f->channels; ++i)
+ f->outputs[i] = f->channel_buffers[i] + left;
+
+ if (channels) *channels = f->channels;
+ *samples = len;
+ *output = f->outputs;
+ return (int) (f->stream - data);
}
stb_vorbis *stb_vorbis_open_pushdata(
- const unsigned char *data, int data_len, // the memory available for decoding
- int *data_used, // only defined if result is not NULL
- int *error, const stb_vorbis_alloc *alloc)
-{
- stb_vorbis *f, p;
- vorbis_init(&p, alloc);
- p.stream = (uint8 *)data;
- p.stream_end = (uint8 *)data + data_len;
- p.push_mode = TRUE;
- if (!start_decoder(&p)) {
- if (p.eof)
- *error = VORBIS_need_more_data;
- else
- *error = p.error;
- return NULL;
- }
- f = vorbis_alloc(&p);
- if (f) {
- *f = p;
- *data_used = (int)(f->stream - data);
- *error = 0;
- return f;
- }
- else {
- vorbis_deinit(&p);
- return NULL;
- }
+ const unsigned char *data, int data_len, // the memory available for decoding
+ int *data_used, // only defined if result is not NULL
+ int *error, const stb_vorbis_alloc *alloc)
+{
+ stb_vorbis *f, p;
+ vorbis_init(&p, alloc);
+ p.stream = (uint8 *) data;
+ p.stream_end = (uint8 *) data + data_len;
+ p.push_mode = TRUE;
+ if (!start_decoder(&p)) {
+ if (p.eof)
+ *error = VORBIS_need_more_data;
+ else
+ *error = p.error;
+ vorbis_deinit(&p);
+ return NULL;
+ }
+ f = vorbis_alloc(&p);
+ if (f) {
+ *f = p;
+ *data_used = (int) (f->stream - data);
+ *error = 0;
+ return f;
+ } else {
+ vorbis_deinit(&p);
+ return NULL;
+ }
}
#endif // STB_VORBIS_NO_PUSHDATA_API
unsigned int stb_vorbis_get_file_offset(stb_vorbis *f)
{
-#ifndef STB_VORBIS_NO_PUSHDATA_API
- if (f->push_mode) return 0;
-#endif
- if (USE_MEMORY(f)) return (unsigned int)(f->stream - f->stream_start);
-#ifndef STB_VORBIS_NO_STDIO
- return (unsigned int)(ftell(f->f) - f->f_start);
-#endif
+ #ifndef STB_VORBIS_NO_PUSHDATA_API
+ if (f->push_mode) return 0;
+ #endif
+ if (USE_MEMORY(f)) return (unsigned int) (f->stream - f->stream_start);
+ #ifndef STB_VORBIS_NO_STDIO
+ return (unsigned int) (ftell(f->f) - f->f_start);
+ #endif
}
#ifndef STB_VORBIS_NO_PULLDATA_API
@@ -4504,72 +4560,72 @@ unsigned int stb_vorbis_get_file_offset(stb_vorbis *f)
static uint32 vorbis_find_page(stb_vorbis *f, uint32 *end, uint32 *last)
{
- for (;;) {
- int n;
- if (f->eof) return 0;
- n = get8(f);
- if (n == 0x4f) { // page header candidate
- unsigned int retry_loc = stb_vorbis_get_file_offset(f);
- int i;
- // check if we're off the end of a file_section stream
- if (retry_loc - 25 > f->stream_len)
- return 0;
- // check the rest of the header
- for (i = 1; i < 4; ++i)
- if (get8(f) != ogg_page_header[i])
- break;
+ for(;;) {
+ int n;
+ if (f->eof) return 0;
+ n = get8(f);
+ if (n == 0x4f) { // page header candidate
+ unsigned int retry_loc = stb_vorbis_get_file_offset(f);
+ int i;
+ // check if we're off the end of a file_section stream
+ if (retry_loc - 25 > f->stream_len)
+ return 0;
+ // check the rest of the header
+ for (i=1; i < 4; ++i)
+ if (get8(f) != ogg_page_header[i])
+ break;
+ if (f->eof) return 0;
+ if (i == 4) {
+ uint8 header[27];
+ uint32 i, crc, goal, len;
+ for (i=0; i < 4; ++i)
+ header[i] = ogg_page_header[i];
+ for (; i < 27; ++i)
+ header[i] = get8(f);
if (f->eof) return 0;
- if (i == 4) {
- uint8 header[27];
- uint32 i, crc, goal, len;
- for (i = 0; i < 4; ++i)
- header[i] = ogg_page_header[i];
- for (; i < 27; ++i)
- header[i] = get8(f);
- if (f->eof) return 0;
- if (header[4] != 0) goto invalid;
- goal = header[22] + (header[23] << 8) + (header[24] << 16) + (header[25] << 24);
- for (i = 22; i < 26; ++i)
- header[i] = 0;
- crc = 0;
- for (i = 0; i < 27; ++i)
- crc = crc32_update(crc, header[i]);
- len = 0;
- for (i = 0; i < header[26]; ++i) {
- int s = get8(f);
- crc = crc32_update(crc, s);
- len += s;
- }
- if (len && f->eof) return 0;
- for (i = 0; i < len; ++i)
- crc = crc32_update(crc, get8(f));
- // finished parsing probable page
- if (crc == goal) {
- // we could now check that it's either got the last
- // page flag set, OR it's followed by the capture
- // pattern, but I guess TECHNICALLY you could have
- // a file with garbage between each ogg page and recover
- // from it automatically? So even though that paranoia
- // might decrease the chance of an invalid decode by
- // another 2^32, not worth it since it would hose those
- // invalid-but-useful files?
- if (end)
- *end = stb_vorbis_get_file_offset(f);
- if (last) {
- if (header[5] & 0x04)
- *last = 1;
- else
- *last = 0;
- }
- set_file_offset(f, retry_loc - 1);
- return 1;
- }
+ if (header[4] != 0) goto invalid;
+ goal = header[22] + (header[23] << 8) + (header[24]<<16) + ((uint32)header[25]<<24);
+ for (i=22; i < 26; ++i)
+ header[i] = 0;
+ crc = 0;
+ for (i=0; i < 27; ++i)
+ crc = crc32_update(crc, header[i]);
+ len = 0;
+ for (i=0; i < header[26]; ++i) {
+ int s = get8(f);
+ crc = crc32_update(crc, s);
+ len += s;
+ }
+ if (len && f->eof) return 0;
+ for (i=0; i < len; ++i)
+ crc = crc32_update(crc, get8(f));
+ // finished parsing probable page
+ if (crc == goal) {
+ // we could now check that it's either got the last
+ // page flag set, OR it's followed by the capture
+ // pattern, but I guess TECHNICALLY you could have
+ // a file with garbage between each ogg page and recover
+ // from it automatically? So even though that paranoia
+ // might decrease the chance of an invalid decode by
+ // another 2^32, not worth it since it would hose those
+ // invalid-but-useful files?
+ if (end)
+ *end = stb_vorbis_get_file_offset(f);
+ if (last) {
+ if (header[5] & 0x04)
+ *last = 1;
+ else
+ *last = 0;
+ }
+ set_file_offset(f, retry_loc-1);
+ return 1;
}
+ }
invalid:
- // not a valid page, so rewind and look for next one
- set_file_offset(f, retry_loc);
- }
- }
+ // not a valid page, so rewind and look for next one
+ set_file_offset(f, retry_loc);
+ }
+ }
}
@@ -4586,55 +4642,55 @@ static uint32 vorbis_find_page(stb_vorbis *f, uint32 *end, uint32 *last)
static int get_seek_page_info(stb_vorbis *f, ProbedPage *z)
{
- uint8 header[27], lacing[255];
- int i, len;
+ uint8 header[27], lacing[255];
+ int i,len;
- // record where the page starts
- z->page_start = stb_vorbis_get_file_offset(f);
+ // record where the page starts
+ z->page_start = stb_vorbis_get_file_offset(f);
- // parse the header
- getn(f, header, 27);
- if (header[0] != 'O' || header[1] != 'g' || header[2] != 'g' || header[3] != 'S')
- return 0;
- getn(f, lacing, header[26]);
+ // parse the header
+ getn(f, header, 27);
+ if (header[0] != 'O' || header[1] != 'g' || header[2] != 'g' || header[3] != 'S')
+ return 0;
+ getn(f, lacing, header[26]);
- // determine the length of the payload
- len = 0;
- for (i = 0; i < header[26]; ++i)
- len += lacing[i];
+ // determine the length of the payload
+ len = 0;
+ for (i=0; i < header[26]; ++i)
+ len += lacing[i];
- // this implies where the page ends
- z->page_end = z->page_start + 27 + header[26] + len;
+ // this implies where the page ends
+ z->page_end = z->page_start + 27 + header[26] + len;
- // read the last-decoded sample out of the data
- z->last_decoded_sample = header[6] + (header[7] << 8) + (header[8] << 16) + (header[9] << 24);
+ // read the last-decoded sample out of the data
+ z->last_decoded_sample = header[6] + (header[7] << 8) + (header[8] << 16) + (header[9] << 24);
- // restore file state to where we were
- set_file_offset(f, z->page_start);
- return 1;
+ // restore file state to where we were
+ set_file_offset(f, z->page_start);
+ return 1;
}
-// rarely used function to seek back to the preceeding page while finding the
+// rarely used function to seek back to the preceding page while finding the
// start of a packet
static int go_to_page_before(stb_vorbis *f, unsigned int limit_offset)
{
- unsigned int previous_safe, end;
+ unsigned int previous_safe, end;
- // now we want to seek back 64K from the limit
- if (limit_offset >= 65536 && limit_offset - 65536 >= f->first_audio_page_offset)
- previous_safe = limit_offset - 65536;
- else
- previous_safe = f->first_audio_page_offset;
+ // now we want to seek back 64K from the limit
+ if (limit_offset >= 65536 && limit_offset-65536 >= f->first_audio_page_offset)
+ previous_safe = limit_offset - 65536;
+ else
+ previous_safe = f->first_audio_page_offset;
- set_file_offset(f, previous_safe);
+ set_file_offset(f, previous_safe);
- while (vorbis_find_page(f, &end, NULL)) {
- if (end >= limit_offset && stb_vorbis_get_file_offset(f) < limit_offset)
- return 1;
- set_file_offset(f, end);
- }
+ while (vorbis_find_page(f, &end, NULL)) {
+ if (end >= limit_offset && stb_vorbis_get_file_offset(f) < limit_offset)
+ return 1;
+ set_file_offset(f, end);
+ }
- return 0;
+ return 0;
}
// implements the search logic for finding a page and starting decoding. if
@@ -4643,419 +4699,428 @@ static int go_to_page_before(stb_vorbis *f, unsigned int limit_offset)
// better).
static int seek_to_sample_coarse(stb_vorbis *f, uint32 sample_number)
{
- ProbedPage left, right, mid;
- int i, start_seg_with_known_loc, end_pos, page_start;
- uint32 delta, stream_length, padding;
- double offset, bytes_per_sample;
- int probe = 0;
-
- // find the last page and validate the target sample
- stream_length = stb_vorbis_stream_length_in_samples(f);
- if (stream_length == 0) return error(f, VORBIS_seek_without_length);
- if (sample_number > stream_length) return error(f, VORBIS_seek_invalid);
-
- // this is the maximum difference between the window-center (which is the
- // actual granule position value), and the right-start (which the spec
- // indicates should be the granule position (give or take one)).
- padding = ((f->blocksize_1 - f->blocksize_0) >> 2);
- if (sample_number < padding)
- sample_number = 0;
- else
- sample_number -= padding;
-
- left = f->p_first;
- while (left.last_decoded_sample == ~0U) {
- // (untested) the first page does not have a 'last_decoded_sample'
- set_file_offset(f, left.page_end);
- if (!get_seek_page_info(f, &left)) goto error;
- }
-
- right = f->p_last;
- assert(right.last_decoded_sample != ~0U);
-
- // starting from the start is handled differently
- if (sample_number <= left.last_decoded_sample) {
- if (stb_vorbis_seek_start(f))
- return 1;
- return 0;
- }
-
- while (left.page_end != right.page_start) {
- assert(left.page_end < right.page_start);
- // search range in bytes
- delta = right.page_start - left.page_end;
- if (delta <= 65536) {
- // there's only 64K left to search - handle it linearly
- set_file_offset(f, left.page_end);
- }
- else {
- if (probe < 2) {
- if (probe == 0) {
- // first probe (interpolate)
- double data_bytes = right.page_end - left.page_start;
- bytes_per_sample = data_bytes / right.last_decoded_sample;
- offset = left.page_start + bytes_per_sample * (sample_number - left.last_decoded_sample);
- }
- else {
- // second probe (try to bound the other side)
- double error = ((double)sample_number - mid.last_decoded_sample) * bytes_per_sample;
- if (error >= 0 && error < 8000) error = 8000;
- if (error < 0 && error > -8000) error = -8000;
- offset += error * 2;
- }
-
- // ensure the offset is valid
- if (offset < left.page_end)
- offset = left.page_end;
- if (offset > right.page_start - 65536)
- offset = right.page_start - 65536;
-
- set_file_offset(f, (unsigned int)offset);
- }
- else {
- // binary search for large ranges (offset by 32K to ensure
- // we don't hit the right page)
- set_file_offset(f, left.page_end + (delta / 2) - 32768);
- }
+ ProbedPage left, right, mid;
+ int i, start_seg_with_known_loc, end_pos, page_start;
+ uint32 delta, stream_length, padding, last_sample_limit;
+ double offset = 0.0, bytes_per_sample = 0.0;
+ int probe = 0;
+
+ // find the last page and validate the target sample
+ stream_length = stb_vorbis_stream_length_in_samples(f);
+ if (stream_length == 0) return error(f, VORBIS_seek_without_length);
+ if (sample_number > stream_length) return error(f, VORBIS_seek_invalid);
+
+ // this is the maximum difference between the window-center (which is the
+ // actual granule position value), and the right-start (which the spec
+ // indicates should be the granule position (give or take one)).
+ padding = ((f->blocksize_1 - f->blocksize_0) >> 2);
+ if (sample_number < padding)
+ last_sample_limit = 0;
+ else
+ last_sample_limit = sample_number - padding;
+
+ left = f->p_first;
+ while (left.last_decoded_sample == ~0U) {
+ // (untested) the first page does not have a 'last_decoded_sample'
+ set_file_offset(f, left.page_end);
+ if (!get_seek_page_info(f, &left)) goto error;
+ }
- if (!vorbis_find_page(f, NULL, NULL)) goto error;
- }
+ right = f->p_last;
+ assert(right.last_decoded_sample != ~0U);
+
+ // starting from the start is handled differently
+ if (last_sample_limit <= left.last_decoded_sample) {
+ if (stb_vorbis_seek_start(f)) {
+ if (f->current_loc > sample_number)
+ return error(f, VORBIS_seek_failed);
+ return 1;
+ }
+ return 0;
+ }
- for (;;) {
- if (!get_seek_page_info(f, &mid)) goto error;
- if (mid.last_decoded_sample != ~0U) break;
- // (untested) no frames end on this page
- set_file_offset(f, mid.page_end);
- assert(mid.page_start < right.page_start);
- }
+ while (left.page_end != right.page_start) {
+ assert(left.page_end < right.page_start);
+ // search range in bytes
+ delta = right.page_start - left.page_end;
+ if (delta <= 65536) {
+ // there's only 64K left to search - handle it linearly
+ set_file_offset(f, left.page_end);
+ } else {
+ if (probe < 2) {
+ if (probe == 0) {
+ // first probe (interpolate)
+ double data_bytes = right.page_end - left.page_start;
+ bytes_per_sample = data_bytes / right.last_decoded_sample;
+ offset = left.page_start + bytes_per_sample * (last_sample_limit - left.last_decoded_sample);
+ } else {
+ // second probe (try to bound the other side)
+ double error = ((double) last_sample_limit - mid.last_decoded_sample) * bytes_per_sample;
+ if (error >= 0 && error < 8000) error = 8000;
+ if (error < 0 && error > -8000) error = -8000;
+ offset += error * 2;
+ }
- // if we've just found the last page again then we're in a tricky file,
- // and we're close enough.
- if (mid.page_start == right.page_start)
+ // ensure the offset is valid
+ if (offset < left.page_end)
+ offset = left.page_end;
+ if (offset > right.page_start - 65536)
+ offset = right.page_start - 65536;
+
+ set_file_offset(f, (unsigned int) offset);
+ } else {
+ // binary search for large ranges (offset by 32K to ensure
+ // we don't hit the right page)
+ set_file_offset(f, left.page_end + (delta / 2) - 32768);
+ }
+
+ if (!vorbis_find_page(f, NULL, NULL)) goto error;
+ }
+
+ for (;;) {
+ if (!get_seek_page_info(f, &mid)) goto error;
+ if (mid.last_decoded_sample != ~0U) break;
+ // (untested) no frames end on this page
+ set_file_offset(f, mid.page_end);
+ assert(mid.page_start < right.page_start);
+ }
+
+ // if we've just found the last page again then we're in a tricky file,
+ // and we're close enough (if it wasn't an interpolation probe).
+ if (mid.page_start == right.page_start) {
+ if (probe >= 2 || delta <= 65536)
break;
-
- if (sample_number < mid.last_decoded_sample)
+ } else {
+ if (last_sample_limit < mid.last_decoded_sample)
right = mid;
- else
+ else
left = mid;
+ }
- ++probe;
- }
+ ++probe;
+ }
- // seek back to start of the last packet
- page_start = left.page_start;
- set_file_offset(f, page_start);
- if (!start_page(f)) return error(f, VORBIS_seek_failed);
- end_pos = f->end_seg_with_known_loc;
- assert(end_pos >= 0);
+ // seek back to start of the last packet
+ page_start = left.page_start;
+ set_file_offset(f, page_start);
+ if (!start_page(f)) return error(f, VORBIS_seek_failed);
+ end_pos = f->end_seg_with_known_loc;
+ assert(end_pos >= 0);
- for (;;) {
- for (i = end_pos; i > 0; --i)
- if (f->segments[i - 1] != 255)
- break;
+ for (;;) {
+ for (i = end_pos; i > 0; --i)
+ if (f->segments[i-1] != 255)
+ break;
- start_seg_with_known_loc = i;
+ start_seg_with_known_loc = i;
- if (start_seg_with_known_loc > 0 || !(f->page_flag & PAGEFLAG_continued_packet))
- break;
+ if (start_seg_with_known_loc > 0 || !(f->page_flag & PAGEFLAG_continued_packet))
+ break;
- // (untested) the final packet begins on an earlier page
- if (!go_to_page_before(f, page_start))
- goto error;
-
- page_start = stb_vorbis_get_file_offset(f);
- if (!start_page(f)) goto error;
- end_pos = f->segment_count - 1;
- }
-
- // prepare to start decoding
- f->current_loc_valid = FALSE;
- f->last_seg = FALSE;
- f->valid_bits = 0;
- f->packet_bytes = 0;
- f->bytes_in_seg = 0;
- f->previous_length = 0;
- f->next_seg = start_seg_with_known_loc;
-
- for (i = 0; i < start_seg_with_known_loc; i++)
- skip(f, f->segments[i]);
-
- // start decoding (optimizable - this frame is generally discarded)
- if (!vorbis_pump_first_frame(f))
- return 0;
- if (f->current_loc > sample_number)
- return error(f, VORBIS_seek_failed);
- return 1;
+ // (untested) the final packet begins on an earlier page
+ if (!go_to_page_before(f, page_start))
+ goto error;
+
+ page_start = stb_vorbis_get_file_offset(f);
+ if (!start_page(f)) goto error;
+ end_pos = f->segment_count - 1;
+ }
+
+ // prepare to start decoding
+ f->current_loc_valid = FALSE;
+ f->last_seg = FALSE;
+ f->valid_bits = 0;
+ f->packet_bytes = 0;
+ f->bytes_in_seg = 0;
+ f->previous_length = 0;
+ f->next_seg = start_seg_with_known_loc;
+
+ for (i = 0; i < start_seg_with_known_loc; i++)
+ skip(f, f->segments[i]);
+
+ // start decoding (optimizable - this frame is generally discarded)
+ if (!vorbis_pump_first_frame(f))
+ return 0;
+ if (f->current_loc > sample_number)
+ return error(f, VORBIS_seek_failed);
+ return 1;
error:
- // try to restore the file to a valid state
- stb_vorbis_seek_start(f);
- return error(f, VORBIS_seek_failed);
+ // try to restore the file to a valid state
+ stb_vorbis_seek_start(f);
+ return error(f, VORBIS_seek_failed);
}
// the same as vorbis_decode_initial, but without advancing
static int peek_decode_initial(vorb *f, int *p_left_start, int *p_left_end, int *p_right_start, int *p_right_end, int *mode)
{
- int bits_read, bytes_read;
+ int bits_read, bytes_read;
- if (!vorbis_decode_initial(f, p_left_start, p_left_end, p_right_start, p_right_end, mode))
- return 0;
+ if (!vorbis_decode_initial(f, p_left_start, p_left_end, p_right_start, p_right_end, mode))
+ return 0;
- // either 1 or 2 bytes were read, figure out which so we can rewind
- bits_read = 1 + ilog(f->mode_count - 1);
- if (f->mode_config[*mode].blockflag)
- bits_read += 2;
- bytes_read = (bits_read + 7) / 8;
+ // either 1 or 2 bytes were read, figure out which so we can rewind
+ bits_read = 1 + ilog(f->mode_count-1);
+ if (f->mode_config[*mode].blockflag)
+ bits_read += 2;
+ bytes_read = (bits_read + 7) / 8;
- f->bytes_in_seg += bytes_read;
- f->packet_bytes -= bytes_read;
- skip(f, -bytes_read);
- if (f->next_seg == -1)
- f->next_seg = f->segment_count - 1;
- else
- f->next_seg--;
- f->valid_bits = 0;
+ f->bytes_in_seg += bytes_read;
+ f->packet_bytes -= bytes_read;
+ skip(f, -bytes_read);
+ if (f->next_seg == -1)
+ f->next_seg = f->segment_count - 1;
+ else
+ f->next_seg--;
+ f->valid_bits = 0;
- return 1;
+ return 1;
}
int stb_vorbis_seek_frame(stb_vorbis *f, unsigned int sample_number)
{
- uint32 max_frame_samples;
-
- if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing);
-
- // fast page-level search
- if (!seek_to_sample_coarse(f, sample_number))
- return 0;
-
- assert(f->current_loc_valid);
- assert(f->current_loc <= sample_number);
-
- // linear search for the relevant packet
- max_frame_samples = (f->blocksize_1 * 3 - f->blocksize_0) >> 2;
- while (f->current_loc < sample_number) {
- int left_start, left_end, right_start, right_end, mode, frame_samples;
- if (!peek_decode_initial(f, &left_start, &left_end, &right_start, &right_end, &mode))
- return error(f, VORBIS_seek_failed);
- // calculate the number of samples returned by the next frame
- frame_samples = right_start - left_start;
- if (f->current_loc + frame_samples > sample_number) {
- return 1; // the next frame will contain the sample
- }
- else if (f->current_loc + frame_samples + max_frame_samples > sample_number) {
- // there's a chance the frame after this could contain the sample
- vorbis_pump_first_frame(f);
- }
- else {
- // this frame is too early to be relevant
- f->current_loc += frame_samples;
- f->previous_length = 0;
- maybe_start_packet(f);
- flush_packet(f);
- }
- }
- // the next frame will start with the sample
- assert(f->current_loc == sample_number);
- return 1;
+ uint32 max_frame_samples;
+
+ if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing);
+
+ // fast page-level search
+ if (!seek_to_sample_coarse(f, sample_number))
+ return 0;
+
+ assert(f->current_loc_valid);
+ assert(f->current_loc <= sample_number);
+
+ // linear search for the relevant packet
+ max_frame_samples = (f->blocksize_1*3 - f->blocksize_0) >> 2;
+ while (f->current_loc < sample_number) {
+ int left_start, left_end, right_start, right_end, mode, frame_samples;
+ if (!peek_decode_initial(f, &left_start, &left_end, &right_start, &right_end, &mode))
+ return error(f, VORBIS_seek_failed);
+ // calculate the number of samples returned by the next frame
+ frame_samples = right_start - left_start;
+ if (f->current_loc + frame_samples > sample_number) {
+ return 1; // the next frame will contain the sample
+ } else if (f->current_loc + frame_samples + max_frame_samples > sample_number) {
+ // there's a chance the frame after this could contain the sample
+ vorbis_pump_first_frame(f);
+ } else {
+ // this frame is too early to be relevant
+ f->current_loc += frame_samples;
+ f->previous_length = 0;
+ maybe_start_packet(f);
+ flush_packet(f);
+ }
+ }
+ // the next frame should start with the sample
+ if (f->current_loc != sample_number) return error(f, VORBIS_seek_failed);
+ return 1;
}
int stb_vorbis_seek(stb_vorbis *f, unsigned int sample_number)
{
- if (!stb_vorbis_seek_frame(f, sample_number))
- return 0;
+ if (!stb_vorbis_seek_frame(f, sample_number))
+ return 0;
- if (sample_number != f->current_loc) {
- int n;
- uint32 frame_start = f->current_loc;
- stb_vorbis_get_frame_float(f, &n, NULL);
- assert(sample_number > frame_start);
- assert(f->channel_buffer_start + (int)(sample_number - frame_start) <= f->channel_buffer_end);
- f->channel_buffer_start += (sample_number - frame_start);
- }
+ if (sample_number != f->current_loc) {
+ int n;
+ uint32 frame_start = f->current_loc;
+ stb_vorbis_get_frame_float(f, &n, NULL);
+ assert(sample_number > frame_start);
+ assert(f->channel_buffer_start + (int) (sample_number-frame_start) <= f->channel_buffer_end);
+ f->channel_buffer_start += (sample_number - frame_start);
+ }
- return 1;
+ return 1;
}
int stb_vorbis_seek_start(stb_vorbis *f)
{
- if (IS_PUSH_MODE(f)) { return error(f, VORBIS_invalid_api_mixing); }
- set_file_offset(f, f->first_audio_page_offset);
- f->previous_length = 0;
- f->first_decode = TRUE;
- f->next_seg = -1;
- return vorbis_pump_first_frame(f);
+ if (IS_PUSH_MODE(f)) { return error(f, VORBIS_invalid_api_mixing); }
+ set_file_offset(f, f->first_audio_page_offset);
+ f->previous_length = 0;
+ f->first_decode = TRUE;
+ f->next_seg = -1;
+ return vorbis_pump_first_frame(f);
}
unsigned int stb_vorbis_stream_length_in_samples(stb_vorbis *f)
{
- unsigned int restore_offset, previous_safe;
- unsigned int end, last_page_loc;
-
- if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing);
- if (!f->total_samples) {
- unsigned int last;
- uint32 lo, hi;
- char header[6];
-
- // first, store the current decode position so we can restore it
- restore_offset = stb_vorbis_get_file_offset(f);
-
- // now we want to seek back 64K from the end (the last page must
- // be at most a little less than 64K, but let's allow a little slop)
- if (f->stream_len >= 65536 && f->stream_len - 65536 >= f->first_audio_page_offset)
- previous_safe = f->stream_len - 65536;
- else
- previous_safe = f->first_audio_page_offset;
-
- set_file_offset(f, previous_safe);
- // previous_safe is now our candidate 'earliest known place that seeking
- // to will lead to the final page'
-
- if (!vorbis_find_page(f, &end, &last)) {
- // if we can't find a page, we're hosed!
- f->error = VORBIS_cant_find_last_page;
- f->total_samples = 0xffffffff;
- goto done;
- }
-
- // check if there are more pages
- last_page_loc = stb_vorbis_get_file_offset(f);
-
- // stop when the last_page flag is set, not when we reach eof;
- // this allows us to stop short of a 'file_section' end without
- // explicitly checking the length of the section
- while (!last) {
- set_file_offset(f, end);
- if (!vorbis_find_page(f, &end, &last)) {
- // the last page we found didn't have the 'last page' flag
- // set. whoops!
- break;
- }
- previous_safe = last_page_loc + 1;
- last_page_loc = stb_vorbis_get_file_offset(f);
- }
-
- set_file_offset(f, last_page_loc);
-
- // parse the header
- getn(f, (unsigned char *)header, 6);
- // extract the absolute granule position
- lo = get32(f);
- hi = get32(f);
- if (lo == 0xffffffff && hi == 0xffffffff) {
- f->error = VORBIS_cant_find_last_page;
- f->total_samples = SAMPLE_unknown;
- goto done;
- }
- if (hi)
- lo = 0xfffffffe; // saturate
- f->total_samples = lo;
-
- f->p_last.page_start = last_page_loc;
- f->p_last.page_end = end;
- f->p_last.last_decoded_sample = lo;
-
- done:
- set_file_offset(f, restore_offset);
- }
- return f->total_samples == SAMPLE_unknown ? 0 : f->total_samples;
+ unsigned int restore_offset, previous_safe;
+ unsigned int end, last_page_loc;
+
+ if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing);
+ if (!f->total_samples) {
+ unsigned int last;
+ uint32 lo,hi;
+ char header[6];
+
+ // first, store the current decode position so we can restore it
+ restore_offset = stb_vorbis_get_file_offset(f);
+
+ // now we want to seek back 64K from the end (the last page must
+ // be at most a little less than 64K, but let's allow a little slop)
+ if (f->stream_len >= 65536 && f->stream_len-65536 >= f->first_audio_page_offset)
+ previous_safe = f->stream_len - 65536;
+ else
+ previous_safe = f->first_audio_page_offset;
+
+ set_file_offset(f, previous_safe);
+ // previous_safe is now our candidate 'earliest known place that seeking
+ // to will lead to the final page'
+
+ if (!vorbis_find_page(f, &end, &last)) {
+ // if we can't find a page, we're hosed!
+ f->error = VORBIS_cant_find_last_page;
+ f->total_samples = 0xffffffff;
+ goto done;
+ }
+
+ // check if there are more pages
+ last_page_loc = stb_vorbis_get_file_offset(f);
+
+ // stop when the last_page flag is set, not when we reach eof;
+ // this allows us to stop short of a 'file_section' end without
+ // explicitly checking the length of the section
+ while (!last) {
+ set_file_offset(f, end);
+ if (!vorbis_find_page(f, &end, &last)) {
+ // the last page we found didn't have the 'last page' flag
+ // set. whoops!
+ break;
+ }
+ //previous_safe = last_page_loc+1; // NOTE: not used after this point, but note for debugging
+ last_page_loc = stb_vorbis_get_file_offset(f);
+ }
+
+ set_file_offset(f, last_page_loc);
+
+ // parse the header
+ getn(f, (unsigned char *)header, 6);
+ // extract the absolute granule position
+ lo = get32(f);
+ hi = get32(f);
+ if (lo == 0xffffffff && hi == 0xffffffff) {
+ f->error = VORBIS_cant_find_last_page;
+ f->total_samples = SAMPLE_unknown;
+ goto done;
+ }
+ if (hi)
+ lo = 0xfffffffe; // saturate
+ f->total_samples = lo;
+
+ f->p_last.page_start = last_page_loc;
+ f->p_last.page_end = end;
+ f->p_last.last_decoded_sample = lo;
+
+ done:
+ set_file_offset(f, restore_offset);
+ }
+ return f->total_samples == SAMPLE_unknown ? 0 : f->total_samples;
}
float stb_vorbis_stream_length_in_seconds(stb_vorbis *f)
{
- return stb_vorbis_stream_length_in_samples(f) / (float)f->sample_rate;
+ return stb_vorbis_stream_length_in_samples(f) / (float) f->sample_rate;
}
int stb_vorbis_get_frame_float(stb_vorbis *f, int *channels, float ***output)
{
- int len, right, left, i;
- if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing);
+ int len, right,left,i;
+ if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing);
- if (!vorbis_decode_packet(f, &len, &left, &right)) {
- f->channel_buffer_start = f->channel_buffer_end = 0;
- return 0;
- }
+ if (!vorbis_decode_packet(f, &len, &left, &right)) {
+ f->channel_buffer_start = f->channel_buffer_end = 0;
+ return 0;
+ }
- len = vorbis_finish_frame(f, len, left, right);
- for (i = 0; i < f->channels; ++i)
- f->outputs[i] = f->channel_buffers[i] + left;
+ len = vorbis_finish_frame(f, len, left, right);
+ for (i=0; i < f->channels; ++i)
+ f->outputs[i] = f->channel_buffers[i] + left;
- f->channel_buffer_start = left;
- f->channel_buffer_end = left + len;
+ f->channel_buffer_start = left;
+ f->channel_buffer_end = left+len;
- if (channels) *channels = f->channels;
- if (output) *output = f->outputs;
- return len;
+ if (channels) *channels = f->channels;
+ if (output) *output = f->outputs;
+ return len;
}
#ifndef STB_VORBIS_NO_STDIO
stb_vorbis * stb_vorbis_open_file_section(FILE *file, int close_on_free, int *error, const stb_vorbis_alloc *alloc, unsigned int length)
{
- stb_vorbis *f, p;
- vorbis_init(&p, alloc);
- p.f = file;
- p.f_start = (uint32)ftell(file);
- p.stream_len = length;
- p.close_on_free = close_on_free;
- if (start_decoder(&p)) {
- f = vorbis_alloc(&p);
- if (f) {
- *f = p;
- vorbis_pump_first_frame(f);
- return f;
- }
- }
- if (error) *error = p.error;
- vorbis_deinit(&p);
- return NULL;
+ stb_vorbis *f, p;
+ vorbis_init(&p, alloc);
+ p.f = file;
+ p.f_start = (uint32) ftell(file);
+ p.stream_len = length;
+ p.close_on_free = close_on_free;
+ if (start_decoder(&p)) {
+ f = vorbis_alloc(&p);
+ if (f) {
+ *f = p;
+ vorbis_pump_first_frame(f);
+ return f;
+ }
+ }
+ if (error) *error = p.error;
+ vorbis_deinit(&p);
+ return NULL;
}
stb_vorbis * stb_vorbis_open_file(FILE *file, int close_on_free, int *error, const stb_vorbis_alloc *alloc)
{
- unsigned int len, start;
- start = (unsigned int)ftell(file);
- fseek(file, 0, SEEK_END);
- len = (unsigned int)(ftell(file) - start);
- fseek(file, start, SEEK_SET);
- return stb_vorbis_open_file_section(file, close_on_free, error, alloc, len);
+ unsigned int len, start;
+ start = (unsigned int) ftell(file);
+ fseek(file, 0, SEEK_END);
+ len = (unsigned int) (ftell(file) - start);
+ fseek(file, start, SEEK_SET);
+ return stb_vorbis_open_file_section(file, close_on_free, error, alloc, len);
}
stb_vorbis * stb_vorbis_open_filename(const char *filename, int *error, const stb_vorbis_alloc *alloc)
{
- FILE *f = fopen(filename, "rb");
- if (f)
- return stb_vorbis_open_file(f, TRUE, error, alloc);
- if (error) *error = VORBIS_file_open_failure;
- return NULL;
+ FILE *f;
+#if defined(_WIN32) && defined(__STDC_WANT_SECURE_LIB__)
+ if (0 != fopen_s(&f, filename, "rb"))
+ f = NULL;
+#else
+ f = fopen(filename, "rb");
+#endif
+ if (f)
+ return stb_vorbis_open_file(f, TRUE, error, alloc);
+ if (error) *error = VORBIS_file_open_failure;
+ return NULL;
}
#endif // STB_VORBIS_NO_STDIO
stb_vorbis * stb_vorbis_open_memory(const unsigned char *data, int len, int *error, const stb_vorbis_alloc *alloc)
{
- stb_vorbis *f, p;
- if (data == NULL) return NULL;
- vorbis_init(&p, alloc);
- p.stream = (uint8 *)data;
- p.stream_end = (uint8 *)data + len;
- p.stream_start = (uint8 *)p.stream;
- p.stream_len = len;
- p.push_mode = FALSE;
- if (start_decoder(&p)) {
- f = vorbis_alloc(&p);
- if (f) {
- *f = p;
- vorbis_pump_first_frame(f);
- if (error) *error = VORBIS__no_error;
- return f;
- }
- }
- if (error) *error = p.error;
- vorbis_deinit(&p);
- return NULL;
+ stb_vorbis *f, p;
+ if (!data) {
+ if (error) *error = VORBIS_unexpected_eof;
+ return NULL;
+ }
+ vorbis_init(&p, alloc);
+ p.stream = (uint8 *) data;
+ p.stream_end = (uint8 *) data + len;
+ p.stream_start = (uint8 *) p.stream;
+ p.stream_len = len;
+ p.push_mode = FALSE;
+ if (start_decoder(&p)) {
+ f = vorbis_alloc(&p);
+ if (f) {
+ *f = p;
+ vorbis_pump_first_frame(f);
+ if (error) *error = VORBIS__no_error;
+ return f;
+ }
+ }
+ if (error) *error = p.error;
+ vorbis_deinit(&p);
+ return NULL;
}
#ifndef STB_VORBIS_NO_INTEGER_CONVERSION
@@ -5069,408 +5134,408 @@ stb_vorbis * stb_vorbis_open_memory(const unsigned char *data, int len, int *err
static int8 channel_position[7][6] =
{
- { 0 },
- { C },
- { L, R },
- { L, C, R },
- { L, R, L, R },
- { L, C, R, L, R },
- { L, C, R, L, R, C },
+ { 0 },
+ { C },
+ { L, R },
+ { L, C, R },
+ { L, R, L, R },
+ { L, C, R, L, R },
+ { L, C, R, L, R, C },
};
#ifndef STB_VORBIS_NO_FAST_SCALED_FLOAT
-typedef union {
- float f;
- int i;
-} float_conv;
-typedef char stb_vorbis_float_size_test[sizeof(float) == 4 && sizeof(int) == 4];
-#define FASTDEF(x) float_conv x
-// add (1<<23) to convert to int, then divide by 2^SHIFT, then add 0.5/2^SHIFT to round
-#define MAGIC(SHIFT) (1.5f * (1 << (23-SHIFT)) + 0.5f/(1 << SHIFT))
-#define ADDEND(SHIFT) (((150-SHIFT) << 23) + (1 << 22))
-#define FAST_SCALED_FLOAT_TO_INT(temp,x,s) (temp.f = (x) + MAGIC(s), temp.i - ADDEND(s))
-#define check_endianness()
+ typedef union {
+ float f;
+ int i;
+ } float_conv;
+ typedef char stb_vorbis_float_size_test[sizeof(float)==4 && sizeof(int) == 4];
+ #define FASTDEF(x) float_conv x
+ // add (1<<23) to convert to int, then divide by 2^SHIFT, then add 0.5/2^SHIFT to round
+ #define MAGIC(SHIFT) (1.5f * (1 << (23-SHIFT)) + 0.5f/(1 << SHIFT))
+ #define ADDEND(SHIFT) (((150-SHIFT) << 23) + (1 << 22))
+ #define FAST_SCALED_FLOAT_TO_INT(temp,x,s) (temp.f = (x) + MAGIC(s), temp.i - ADDEND(s))
+ #define check_endianness()
#else
-#define FAST_SCALED_FLOAT_TO_INT(temp,x,s) ((int) ((x) * (1 << (s))))
-#define check_endianness()
-#define FASTDEF(x)
+ #define FAST_SCALED_FLOAT_TO_INT(temp,x,s) ((int) ((x) * (1 << (s))))
+ #define check_endianness()
+ #define FASTDEF(x)
#endif
static void copy_samples(short *dest, float *src, int len)
{
- int i;
- check_endianness();
- for (i = 0; i < len; ++i) {
- FASTDEF(temp);
- int v = FAST_SCALED_FLOAT_TO_INT(temp, src[i], 15);
- if ((unsigned int)(v + 32768) > 65535)
- v = v < 0 ? -32768 : 32767;
- dest[i] = v;
- }
+ int i;
+ check_endianness();
+ for (i=0; i < len; ++i) {
+ FASTDEF(temp);
+ int v = FAST_SCALED_FLOAT_TO_INT(temp, src[i],15);
+ if ((unsigned int) (v + 32768) > 65535)
+ v = v < 0 ? -32768 : 32767;
+ dest[i] = v;
+ }
}
static void compute_samples(int mask, short *output, int num_c, float **data, int d_offset, int len)
{
-#define BUFFER_SIZE 32
- float buffer[BUFFER_SIZE];
- int i, j, o, n = BUFFER_SIZE;
- check_endianness();
- for (o = 0; o < len; o += BUFFER_SIZE) {
- memset(buffer, 0, sizeof(buffer));
- if (o + n > len) n = len - o;
- for (j = 0; j < num_c; ++j) {
- if (channel_position[num_c][j] & mask) {
- for (i = 0; i < n; ++i)
- buffer[i] += data[j][d_offset + o + i];
- }
- }
- for (i = 0; i < n; ++i) {
- FASTDEF(temp);
- int v = FAST_SCALED_FLOAT_TO_INT(temp, buffer[i], 15);
- if ((unsigned int)(v + 32768) > 65535)
- v = v < 0 ? -32768 : 32767;
- output[o + i] = v;
- }
- }
+ #define STB_BUFFER_SIZE 32
+ float buffer[STB_BUFFER_SIZE];
+ int i,j,o,n = STB_BUFFER_SIZE;
+ check_endianness();
+ for (o = 0; o < len; o += STB_BUFFER_SIZE) {
+ memset(buffer, 0, sizeof(buffer));
+ if (o + n > len) n = len - o;
+ for (j=0; j < num_c; ++j) {
+ if (channel_position[num_c][j] & mask) {
+ for (i=0; i < n; ++i)
+ buffer[i] += data[j][d_offset+o+i];
+ }
+ }
+ for (i=0; i < n; ++i) {
+ FASTDEF(temp);
+ int v = FAST_SCALED_FLOAT_TO_INT(temp,buffer[i],15);
+ if ((unsigned int) (v + 32768) > 65535)
+ v = v < 0 ? -32768 : 32767;
+ output[o+i] = v;
+ }
+ }
+ #undef STB_BUFFER_SIZE
}
static void compute_stereo_samples(short *output, int num_c, float **data, int d_offset, int len)
{
-#define BUFFER_SIZE 32
- float buffer[BUFFER_SIZE];
- int i, j, o, n = BUFFER_SIZE >> 1;
- // o is the offset in the source data
- check_endianness();
- for (o = 0; o < len; o += BUFFER_SIZE >> 1) {
- // o2 is the offset in the output data
- int o2 = o << 1;
- memset(buffer, 0, sizeof(buffer));
- if (o + n > len) n = len - o;
- for (j = 0; j < num_c; ++j) {
- int m = channel_position[num_c][j] & (PLAYBACK_LEFT | PLAYBACK_RIGHT);
- if (m == (PLAYBACK_LEFT | PLAYBACK_RIGHT)) {
- for (i = 0; i < n; ++i) {
- buffer[i * 2 + 0] += data[j][d_offset + o + i];
- buffer[i * 2 + 1] += data[j][d_offset + o + i];
- }
+ #define STB_BUFFER_SIZE 32
+ float buffer[STB_BUFFER_SIZE];
+ int i,j,o,n = STB_BUFFER_SIZE >> 1;
+ // o is the offset in the source data
+ check_endianness();
+ for (o = 0; o < len; o += STB_BUFFER_SIZE >> 1) {
+ // o2 is the offset in the output data
+ int o2 = o << 1;
+ memset(buffer, 0, sizeof(buffer));
+ if (o + n > len) n = len - o;
+ for (j=0; j < num_c; ++j) {
+ int m = channel_position[num_c][j] & (PLAYBACK_LEFT | PLAYBACK_RIGHT);
+ if (m == (PLAYBACK_LEFT | PLAYBACK_RIGHT)) {
+ for (i=0; i < n; ++i) {
+ buffer[i*2+0] += data[j][d_offset+o+i];
+ buffer[i*2+1] += data[j][d_offset+o+i];
}
- else if (m == PLAYBACK_LEFT) {
- for (i = 0; i < n; ++i) {
- buffer[i * 2 + 0] += data[j][d_offset + o + i];
- }
+ } else if (m == PLAYBACK_LEFT) {
+ for (i=0; i < n; ++i) {
+ buffer[i*2+0] += data[j][d_offset+o+i];
}
- else if (m == PLAYBACK_RIGHT) {
- for (i = 0; i < n; ++i) {
- buffer[i * 2 + 1] += data[j][d_offset + o + i];
- }
+ } else if (m == PLAYBACK_RIGHT) {
+ for (i=0; i < n; ++i) {
+ buffer[i*2+1] += data[j][d_offset+o+i];
}
- }
- for (i = 0; i < (n << 1); ++i) {
- FASTDEF(temp);
- int v = FAST_SCALED_FLOAT_TO_INT(temp, buffer[i], 15);
- if ((unsigned int)(v + 32768) > 65535)
- v = v < 0 ? -32768 : 32767;
- output[o2 + i] = v;
- }
- }
+ }
+ }
+ for (i=0; i < (n<<1); ++i) {
+ FASTDEF(temp);
+ int v = FAST_SCALED_FLOAT_TO_INT(temp,buffer[i],15);
+ if ((unsigned int) (v + 32768) > 65535)
+ v = v < 0 ? -32768 : 32767;
+ output[o2+i] = v;
+ }
+ }
+ #undef STB_BUFFER_SIZE
}
static void convert_samples_short(int buf_c, short **buffer, int b_offset, int data_c, float **data, int d_offset, int samples)
{
- int i;
- if (buf_c != data_c && buf_c <= 2 && data_c <= 6) {
- static int channel_selector[3][2] = { { 0 },{ PLAYBACK_MONO },{ PLAYBACK_LEFT, PLAYBACK_RIGHT } };
- for (i = 0; i < buf_c; ++i)
- compute_samples(channel_selector[buf_c][i], buffer[i] + b_offset, data_c, data, d_offset, samples);
- }
- else {
- int limit = buf_c < data_c ? buf_c : data_c;
- for (i = 0; i < limit; ++i)
- copy_samples(buffer[i] + b_offset, data[i] + d_offset, samples);
- for (; i < buf_c; ++i)
- memset(buffer[i] + b_offset, 0, sizeof(short) * samples);
- }
+ int i;
+ if (buf_c != data_c && buf_c <= 2 && data_c <= 6) {
+ static int channel_selector[3][2] = { {0}, {PLAYBACK_MONO}, {PLAYBACK_LEFT, PLAYBACK_RIGHT} };
+ for (i=0; i < buf_c; ++i)
+ compute_samples(channel_selector[buf_c][i], buffer[i]+b_offset, data_c, data, d_offset, samples);
+ } else {
+ int limit = buf_c < data_c ? buf_c : data_c;
+ for (i=0; i < limit; ++i)
+ copy_samples(buffer[i]+b_offset, data[i]+d_offset, samples);
+ for ( ; i < buf_c; ++i)
+ memset(buffer[i]+b_offset, 0, sizeof(short) * samples);
+ }
}
int stb_vorbis_get_frame_short(stb_vorbis *f, int num_c, short **buffer, int num_samples)
{
- float **output;
- int len = stb_vorbis_get_frame_float(f, NULL, &output);
- if (len > num_samples) len = num_samples;
- if (len)
- convert_samples_short(num_c, buffer, 0, f->channels, output, 0, len);
- return len;
+ float **output = NULL;
+ int len = stb_vorbis_get_frame_float(f, NULL, &output);
+ if (len > num_samples) len = num_samples;
+ if (len)
+ convert_samples_short(num_c, buffer, 0, f->channels, output, 0, len);
+ return len;
}
static void convert_channels_short_interleaved(int buf_c, short *buffer, int data_c, float **data, int d_offset, int len)
{
- int i;
- check_endianness();
- if (buf_c != data_c && buf_c <= 2 && data_c <= 6) {
- assert(buf_c == 2);
- for (i = 0; i < buf_c; ++i)
- compute_stereo_samples(buffer, data_c, data, d_offset, len);
- }
- else {
- int limit = buf_c < data_c ? buf_c : data_c;
- int j;
- for (j = 0; j < len; ++j) {
- for (i = 0; i < limit; ++i) {
- FASTDEF(temp);
- float f = data[i][d_offset + j];
- int v = FAST_SCALED_FLOAT_TO_INT(temp, f, 15);//data[i][d_offset+j],15);
- if ((unsigned int)(v + 32768) > 65535)
- v = v < 0 ? -32768 : 32767;
- *buffer++ = v;
- }
- for (; i < buf_c; ++i)
- *buffer++ = 0;
- }
- }
+ int i;
+ check_endianness();
+ if (buf_c != data_c && buf_c <= 2 && data_c <= 6) {
+ assert(buf_c == 2);
+ for (i=0; i < buf_c; ++i)
+ compute_stereo_samples(buffer, data_c, data, d_offset, len);
+ } else {
+ int limit = buf_c < data_c ? buf_c : data_c;
+ int j;
+ for (j=0; j < len; ++j) {
+ for (i=0; i < limit; ++i) {
+ FASTDEF(temp);
+ float f = data[i][d_offset+j];
+ int v = FAST_SCALED_FLOAT_TO_INT(temp, f,15);//data[i][d_offset+j],15);
+ if ((unsigned int) (v + 32768) > 65535)
+ v = v < 0 ? -32768 : 32767;
+ *buffer++ = v;
+ }
+ for ( ; i < buf_c; ++i)
+ *buffer++ = 0;
+ }
+ }
}
int stb_vorbis_get_frame_short_interleaved(stb_vorbis *f, int num_c, short *buffer, int num_shorts)
{
- float **output;
- int len;
- if (num_c == 1) return stb_vorbis_get_frame_short(f, num_c, &buffer, num_shorts);
- len = stb_vorbis_get_frame_float(f, NULL, &output);
- if (len) {
- if (len*num_c > num_shorts) len = num_shorts / num_c;
- convert_channels_short_interleaved(num_c, buffer, f->channels, output, 0, len);
- }
- return len;
+ float **output;
+ int len;
+ if (num_c == 1) return stb_vorbis_get_frame_short(f,num_c,&buffer, num_shorts);
+ len = stb_vorbis_get_frame_float(f, NULL, &output);
+ if (len) {
+ if (len*num_c > num_shorts) len = num_shorts / num_c;
+ convert_channels_short_interleaved(num_c, buffer, f->channels, output, 0, len);
+ }
+ return len;
}
int stb_vorbis_get_samples_short_interleaved(stb_vorbis *f, int channels, short *buffer, int num_shorts)
{
- float **outputs;
- int len = num_shorts / channels;
- int n = 0;
- int z = f->channels;
- if (z > channels) z = channels;
- while (n < len) {
- int k = f->channel_buffer_end - f->channel_buffer_start;
- if (n + k >= len) k = len - n;
- if (k)
- convert_channels_short_interleaved(channels, buffer, f->channels, f->channel_buffers, f->channel_buffer_start, k);
- buffer += k*channels;
- n += k;
- f->channel_buffer_start += k;
- if (n == len) break;
- if (!stb_vorbis_get_frame_float(f, NULL, &outputs)) break;
- }
- return n;
+ float **outputs;
+ int len = num_shorts / channels;
+ int n=0;
+ while (n < len) {
+ int k = f->channel_buffer_end - f->channel_buffer_start;
+ if (n+k >= len) k = len - n;
+ if (k)
+ convert_channels_short_interleaved(channels, buffer, f->channels, f->channel_buffers, f->channel_buffer_start, k);
+ buffer += k*channels;
+ n += k;
+ f->channel_buffer_start += k;
+ if (n == len) break;
+ if (!stb_vorbis_get_frame_float(f, NULL, &outputs)) break;
+ }
+ return n;
}
int stb_vorbis_get_samples_short(stb_vorbis *f, int channels, short **buffer, int len)
{
- float **outputs;
- int n = 0;
- int z = f->channels;
- if (z > channels) z = channels;
- while (n < len) {
- int k = f->channel_buffer_end - f->channel_buffer_start;
- if (n + k >= len) k = len - n;
- if (k)
- convert_samples_short(channels, buffer, n, f->channels, f->channel_buffers, f->channel_buffer_start, k);
- n += k;
- f->channel_buffer_start += k;
- if (n == len) break;
- if (!stb_vorbis_get_frame_float(f, NULL, &outputs)) break;
- }
- return n;
+ float **outputs;
+ int n=0;
+ while (n < len) {
+ int k = f->channel_buffer_end - f->channel_buffer_start;
+ if (n+k >= len) k = len - n;
+ if (k)
+ convert_samples_short(channels, buffer, n, f->channels, f->channel_buffers, f->channel_buffer_start, k);
+ n += k;
+ f->channel_buffer_start += k;
+ if (n == len) break;
+ if (!stb_vorbis_get_frame_float(f, NULL, &outputs)) break;
+ }
+ return n;
}
#ifndef STB_VORBIS_NO_STDIO
int stb_vorbis_decode_filename(const char *filename, int *channels, int *sample_rate, short **output)
{
- int data_len, offset, total, limit, error;
- short *data;
- stb_vorbis *v = stb_vorbis_open_filename(filename, &error, NULL);
- if (v == NULL) return -1;
- limit = v->channels * 4096;
- *channels = v->channels;
- if (sample_rate)
- *sample_rate = v->sample_rate;
- offset = data_len = 0;
- total = limit;
- data = (short *)malloc(total * sizeof(*data));
- if (data == NULL) {
- stb_vorbis_close(v);
- return -2;
- }
- for (;;) {
- int n = stb_vorbis_get_frame_short_interleaved(v, v->channels, data + offset, total - offset);
- if (n == 0) break;
- data_len += n;
- offset += n * v->channels;
- if (offset + limit > total) {
- short *data2;
- total *= 2;
- data2 = (short *)realloc(data, total * sizeof(*data));
- if (data2 == NULL) {
- free(data);
- stb_vorbis_close(v);
- return -2;
- }
- data = data2;
- }
- }
- *output = data;
- stb_vorbis_close(v);
- return data_len;
+ int data_len, offset, total, limit, error;
+ short *data;
+ stb_vorbis *v = stb_vorbis_open_filename(filename, &error, NULL);
+ if (v == NULL) return -1;
+ limit = v->channels * 4096;
+ *channels = v->channels;
+ if (sample_rate)
+ *sample_rate = v->sample_rate;
+ offset = data_len = 0;
+ total = limit;
+ data = (short *) malloc(total * sizeof(*data));
+ if (data == NULL) {
+ stb_vorbis_close(v);
+ return -2;
+ }
+ for (;;) {
+ int n = stb_vorbis_get_frame_short_interleaved(v, v->channels, data+offset, total-offset);
+ if (n == 0) break;
+ data_len += n;
+ offset += n * v->channels;
+ if (offset + limit > total) {
+ short *data2;
+ total *= 2;
+ data2 = (short *) realloc(data, total * sizeof(*data));
+ if (data2 == NULL) {
+ free(data);
+ stb_vorbis_close(v);
+ return -2;
+ }
+ data = data2;
+ }
+ }
+ *output = data;
+ stb_vorbis_close(v);
+ return data_len;
}
#endif // NO_STDIO
int stb_vorbis_decode_memory(const uint8 *mem, int len, int *channels, int *sample_rate, short **output)
{
- int data_len, offset, total, limit, error;
- short *data;
- stb_vorbis *v = stb_vorbis_open_memory(mem, len, &error, NULL);
- if (v == NULL) return -1;
- limit = v->channels * 4096;
- *channels = v->channels;
- if (sample_rate)
- *sample_rate = v->sample_rate;
- offset = data_len = 0;
- total = limit;
- data = (short *)malloc(total * sizeof(*data));
- if (data == NULL) {
- stb_vorbis_close(v);
- return -2;
- }
- for (;;) {
- int n = stb_vorbis_get_frame_short_interleaved(v, v->channels, data + offset, total - offset);
- if (n == 0) break;
- data_len += n;
- offset += n * v->channels;
- if (offset + limit > total) {
- short *data2;
- total *= 2;
- data2 = (short *)realloc(data, total * sizeof(*data));
- if (data2 == NULL) {
- free(data);
- stb_vorbis_close(v);
- return -2;
- }
- data = data2;
- }
- }
- *output = data;
- stb_vorbis_close(v);
- return data_len;
+ int data_len, offset, total, limit, error;
+ short *data;
+ stb_vorbis *v = stb_vorbis_open_memory(mem, len, &error, NULL);
+ if (v == NULL) return -1;
+ limit = v->channels * 4096;
+ *channels = v->channels;
+ if (sample_rate)
+ *sample_rate = v->sample_rate;
+ offset = data_len = 0;
+ total = limit;
+ data = (short *) malloc(total * sizeof(*data));
+ if (data == NULL) {
+ stb_vorbis_close(v);
+ return -2;
+ }
+ for (;;) {
+ int n = stb_vorbis_get_frame_short_interleaved(v, v->channels, data+offset, total-offset);
+ if (n == 0) break;
+ data_len += n;
+ offset += n * v->channels;
+ if (offset + limit > total) {
+ short *data2;
+ total *= 2;
+ data2 = (short *) realloc(data, total * sizeof(*data));
+ if (data2 == NULL) {
+ free(data);
+ stb_vorbis_close(v);
+ return -2;
+ }
+ data = data2;
+ }
+ }
+ *output = data;
+ stb_vorbis_close(v);
+ return data_len;
}
#endif // STB_VORBIS_NO_INTEGER_CONVERSION
int stb_vorbis_get_samples_float_interleaved(stb_vorbis *f, int channels, float *buffer, int num_floats)
{
- float **outputs;
- int len = num_floats / channels;
- int n = 0;
- int z = f->channels;
- if (z > channels) z = channels;
- while (n < len) {
- int i, j;
- int k = f->channel_buffer_end - f->channel_buffer_start;
- if (n + k >= len) k = len - n;
- for (j = 0; j < k; ++j) {
- for (i = 0; i < z; ++i)
- *buffer++ = f->channel_buffers[i][f->channel_buffer_start + j];
- for (; i < channels; ++i)
- *buffer++ = 0;
- }
- n += k;
- f->channel_buffer_start += k;
- if (n == len)
- break;
- if (!stb_vorbis_get_frame_float(f, NULL, &outputs))
- break;
- }
- return n;
+ float **outputs;
+ int len = num_floats / channels;
+ int n=0;
+ int z = f->channels;
+ if (z > channels) z = channels;
+ while (n < len) {
+ int i,j;
+ int k = f->channel_buffer_end - f->channel_buffer_start;
+ if (n+k >= len) k = len - n;
+ for (j=0; j < k; ++j) {
+ for (i=0; i < z; ++i)
+ *buffer++ = f->channel_buffers[i][f->channel_buffer_start+j];
+ for ( ; i < channels; ++i)
+ *buffer++ = 0;
+ }
+ n += k;
+ f->channel_buffer_start += k;
+ if (n == len)
+ break;
+ if (!stb_vorbis_get_frame_float(f, NULL, &outputs))
+ break;
+ }
+ return n;
}
int stb_vorbis_get_samples_float(stb_vorbis *f, int channels, float **buffer, int num_samples)
{
- float **outputs;
- int n = 0;
- int z = f->channels;
- if (z > channels) z = channels;
- while (n < num_samples) {
- int i;
- int k = f->channel_buffer_end - f->channel_buffer_start;
- if (n + k >= num_samples) k = num_samples - n;
- if (k) {
- for (i = 0; i < z; ++i)
- memcpy(buffer[i] + n, f->channel_buffers[i] + f->channel_buffer_start, sizeof(float)*k);
- for (; i < channels; ++i)
- memset(buffer[i] + n, 0, sizeof(float) * k);
- }
- n += k;
- f->channel_buffer_start += k;
- if (n == num_samples)
- break;
- if (!stb_vorbis_get_frame_float(f, NULL, &outputs))
- break;
- }
- return n;
+ float **outputs;
+ int n=0;
+ int z = f->channels;
+ if (z > channels) z = channels;
+ while (n < num_samples) {
+ int i;
+ int k = f->channel_buffer_end - f->channel_buffer_start;
+ if (n+k >= num_samples) k = num_samples - n;
+ if (k) {
+ for (i=0; i < z; ++i)
+ memcpy(buffer[i]+n, f->channel_buffers[i]+f->channel_buffer_start, sizeof(float)*k);
+ for ( ; i < channels; ++i)
+ memset(buffer[i]+n, 0, sizeof(float) * k);
+ }
+ n += k;
+ f->channel_buffer_start += k;
+ if (n == num_samples)
+ break;
+ if (!stb_vorbis_get_frame_float(f, NULL, &outputs))
+ break;
+ }
+ return n;
}
#endif // STB_VORBIS_NO_PULLDATA_API
/* Version history
-1.12 - 2017-11-21 - limit residue begin/end to blocksize/2 to avoid large temp allocs in bad/corrupt files
-1.11 - 2017-07-23 - fix MinGW compilation
-1.10 - 2017-03-03 - more robust seeking; fix negative ilog(); clear error in open_memory
-1.09 - 2016-04-04 - back out 'avoid discarding last frame' fix from previous version
-1.08 - 2016-04-02 - fixed multiple warnings; fix setup memory leaks;
-avoid discarding last frame of audio data
-1.07 - 2015-01-16 - fixed some warnings, fix mingw, const-correct API
-some more crash fixes when out of memory or with corrupt files
-1.06 - 2015-08-31 - full, correct support for seeking API (Dougall Johnson)
-some crash fixes when out of memory or with corrupt files
-1.05 - 2015-04-19 - don't define __forceinline if it's redundant
-1.04 - 2014-08-27 - fix missing const-correct case in API
-1.03 - 2014-08-07 - Warning fixes
-1.02 - 2014-07-09 - Declare qsort compare function _cdecl on windows
-1.01 - 2014-06-18 - fix stb_vorbis_get_samples_float
-1.0 - 2014-05-26 - fix memory leaks; fix warnings; fix bugs in multichannel
-(API change) report sample rate for decode-full-file funcs
-0.99996 - bracket #include <malloc.h> for macintosh compilation by Laurent Gomila
-0.99995 - use union instead of pointer-cast for fast-float-to-int to avoid alias-optimization problem
-0.99994 - change fast-float-to-int to work in single-precision FPU mode, remove endian-dependence
-0.99993 - remove assert that fired on legal files with empty tables
-0.99992 - rewind-to-start
-0.99991 - bugfix to stb_vorbis_get_samples_short by Bernhard Wodo
-0.9999 - (should have been 0.99990) fix no-CRT support, compiling as C++
-0.9998 - add a full-decode function with a memory source
-0.9997 - fix a bug in the read-from-FILE case in 0.9996 addition
-0.9996 - query length of vorbis stream in samples/seconds
-0.9995 - bugfix to another optimization that only happened in certain files
-0.9994 - bugfix to one of the optimizations that caused significant (but inaudible?) errors
-0.9993 - performance improvements; runs in 99% to 104% of time of reference implementation
-0.9992 - performance improvement of IMDCT; now performs close to reference implementation
-0.9991 - performance improvement of IMDCT
-0.999 - (should have been 0.9990) performance improvement of IMDCT
-0.998 - no-CRT support from Casey Muratori
-0.997 - bugfixes for bugs found by Terje Mathisen
-0.996 - bugfix: fast-huffman decode initialized incorrectly for sparse codebooks; fixing gives 10% speedup - found by Terje Mathisen
-0.995 - bugfix: fix to 'effective' overrun detection - found by Terje Mathisen
-0.994 - bugfix: garbage decode on final VQ symbol of a non-multiple - found by Terje Mathisen
-0.993 - bugfix: pushdata API required 1 extra byte for empty page (failed to consume final page if empty) - found by Terje Mathisen
-0.992 - fixes for MinGW warning
-0.991 - turn fast-float-conversion on by default
-0.990 - fix push-mode seek recovery if you seek into the headers
-0.98b - fix to bad release of 0.98
-0.98 - fix push-mode seek recovery; robustify float-to-int and support non-fast mode
-0.97 - builds under c++ (typecasting, don't use 'class' keyword)
-0.96 - somehow MY 0.95 was right, but the web one was wrong, so here's my 0.95 rereleased as 0.96, fixes a typo in the clamping code
-0.95 - clamping code for 16-bit functions
-0.94 - not publically released
-0.93 - fixed all-zero-floor case (was decoding garbage)
-0.92 - fixed a memory leak
-0.91 - conditional compiles to omit parts of the API and the infrastructure to support them: STB_VORBIS_NO_PULLDATA_API, STB_VORBIS_NO_PUSHDATA_API, STB_VORBIS_NO_STDIO, STB_VORBIS_NO_INTEGER_CONVERSION
-0.90 - first public release
+ 1.17 - 2019-07-08 - fix CVE-2019-13217, -13218, -13219, -13220, -13221, -13222, -13223
+ found with Mayhem by ForAllSecure
+ 1.16 - 2019-03-04 - fix warnings
+ 1.15 - 2019-02-07 - explicit failure if Ogg Skeleton data is found
+ 1.14 - 2018-02-11 - delete bogus dealloca usage
+ 1.13 - 2018-01-29 - fix truncation of last frame (hopefully)
+ 1.12 - 2017-11-21 - limit residue begin/end to blocksize/2 to avoid large temp allocs in bad/corrupt files
+ 1.11 - 2017-07-23 - fix MinGW compilation
+ 1.10 - 2017-03-03 - more robust seeking; fix negative ilog(); clear error in open_memory
+ 1.09 - 2016-04-04 - back out 'avoid discarding last frame' fix from previous version
+ 1.08 - 2016-04-02 - fixed multiple warnings; fix setup memory leaks;
+ avoid discarding last frame of audio data
+ 1.07 - 2015-01-16 - fixed some warnings, fix mingw, const-correct API
+ some more crash fixes when out of memory or with corrupt files
+ 1.06 - 2015-08-31 - full, correct support for seeking API (Dougall Johnson)
+ some crash fixes when out of memory or with corrupt files
+ 1.05 - 2015-04-19 - don't define __forceinline if it's redundant
+ 1.04 - 2014-08-27 - fix missing const-correct case in API
+ 1.03 - 2014-08-07 - Warning fixes
+ 1.02 - 2014-07-09 - Declare qsort compare function _cdecl on windows
+ 1.01 - 2014-06-18 - fix stb_vorbis_get_samples_float
+ 1.0 - 2014-05-26 - fix memory leaks; fix warnings; fix bugs in multichannel
+ (API change) report sample rate for decode-full-file funcs
+ 0.99996 - bracket #include <malloc.h> for macintosh compilation by Laurent Gomila
+ 0.99995 - use union instead of pointer-cast for fast-float-to-int to avoid alias-optimization problem
+ 0.99994 - change fast-float-to-int to work in single-precision FPU mode, remove endian-dependence
+ 0.99993 - remove assert that fired on legal files with empty tables
+ 0.99992 - rewind-to-start
+ 0.99991 - bugfix to stb_vorbis_get_samples_short by Bernhard Wodo
+ 0.9999 - (should have been 0.99990) fix no-CRT support, compiling as C++
+ 0.9998 - add a full-decode function with a memory source
+ 0.9997 - fix a bug in the read-from-FILE case in 0.9996 addition
+ 0.9996 - query length of vorbis stream in samples/seconds
+ 0.9995 - bugfix to another optimization that only happened in certain files
+ 0.9994 - bugfix to one of the optimizations that caused significant (but inaudible?) errors
+ 0.9993 - performance improvements; runs in 99% to 104% of time of reference implementation
+ 0.9992 - performance improvement of IMDCT; now performs close to reference implementation
+ 0.9991 - performance improvement of IMDCT
+ 0.999 - (should have been 0.9990) performance improvement of IMDCT
+ 0.998 - no-CRT support from Casey Muratori
+ 0.997 - bugfixes for bugs found by Terje Mathisen
+ 0.996 - bugfix: fast-huffman decode initialized incorrectly for sparse codebooks; fixing gives 10% speedup - found by Terje Mathisen
+ 0.995 - bugfix: fix to 'effective' overrun detection - found by Terje Mathisen
+ 0.994 - bugfix: garbage decode on final VQ symbol of a non-multiple - found by Terje Mathisen
+ 0.993 - bugfix: pushdata API required 1 extra byte for empty page (failed to consume final page if empty) - found by Terje Mathisen
+ 0.992 - fixes for MinGW warning
+ 0.991 - turn fast-float-conversion on by default
+ 0.990 - fix push-mode seek recovery if you seek into the headers
+ 0.98b - fix to bad release of 0.98
+ 0.98 - fix push-mode seek recovery; robustify float-to-int and support non-fast mode
+ 0.97 - builds under c++ (typecasting, don't use 'class' keyword)
+ 0.96 - somehow MY 0.95 was right, but the web one was wrong, so here's my 0.95 rereleased as 0.96, fixes a typo in the clamping code
+ 0.95 - clamping code for 16-bit functions
+ 0.94 - not publically released
+ 0.93 - fixed all-zero-floor case (was decoding garbage)
+ 0.92 - fixed a memory leak
+ 0.91 - conditional compiles to omit parts of the API and the infrastructure to support them: STB_VORBIS_NO_PULLDATA_API, STB_VORBIS_NO_PUSHDATA_API, STB_VORBIS_NO_STDIO, STB_VORBIS_NO_INTEGER_CONVERSION
+ 0.90 - first public release
*/
#endif // STB_VORBIS_HEADER_ONLY
generated by cgit v1.1-26-g67d0 at 2025-05-18 21:44:46 +0000