From 67edc97ff17ab2c52886a2547866a4cdc1d2a436 Mon Sep 17 00:00:00 2001 From: twinaphex Date: Sun, 28 Jun 2015 17:56:02 +0200 Subject: [PATCH] Update stb_image.h --- cores/image_core.c | 2 +- cores/stb_image.h | 3702 ++++++++++++++++++++++++++++++++++++++------ 2 files changed, 3219 insertions(+), 485 deletions(-) diff --git a/cores/image_core.c b/cores/image_core.c index 38ca564033..6c19c26740 100644 --- a/cores/image_core.c +++ b/cores/image_core.c @@ -8,12 +8,12 @@ #define STB_IMAGE_IMPLEMENTATION #ifdef RARCH_INTERNAL -#define STB_IMAGE_STATIC #define STBI_NO_PSD #define STBI_NO_GIF #define STBI_NO_HDR #define STBI_NO_PIC #define STBI_NO_PNM +#define STBI_SUPPORT_ZLIB #endif #include "stb_image.h" diff --git a/cores/stb_image.h b/cores/stb_image.h index bde3de20db..f1ca600639 100644 --- a/cores/stb_image.h +++ b/cores/stb_image.h @@ -1,13 +1,398 @@ +/* stb_image - v2.06 - public domain image loader - http://nothings.org/stb_image.h + no warranty implied; use at your own risk + + Do this: + #define STB_IMAGE_IMPLEMENTATION + before you include this file in *one* C or C++ file to create the implementation. + + // i.e. it should look like this: + #include ... + #include ... + #include ... + #define STB_IMAGE_IMPLEMENTATION + #include "stb_image.h" + + You can #define STBI_ASSERT(x) before the #include to avoid using assert.h. + And #define STBI_MALLOC, STBI_REALLOC, and STBI_FREE to avoid using malloc,realloc,free + + + QUICK NOTES: + Primarily of interest to game developers and other people who can + avoid problematic images and only need the trivial interface + + JPEG baseline & progressive (12 bpc/arithmetic not supported, same as stock IJG lib) + PNG 1/2/4/8-bit-per-channel (16 bpc not supported) + + TGA (not sure what subset, if a subset) + BMP non-1bpp, non-RLE + PSD (composited view only, no extra channels) + + GIF (*comp always reports as 4-channel) + HDR (radiance rgbE format) + PIC (Softimage PIC) + PNM (PPM and PGM binary only) + + - decode from memory or through FILE (define STBI_NO_STDIO to remove code) + - decode from arbitrary I/O callbacks + - SIMD acceleration on x86/x64 (SSE2) and ARM (NEON) + + Full documentation under "DOCUMENTATION" below. + + + Revision 2.00 release notes: + + - Progressive JPEG is now supported. + + - PPM and PGM binary formats are now supported, thanks to Ken Miller. + + - x86 platforms now make use of SSE2 SIMD instructions for + JPEG decoding, and ARM platforms can use NEON SIMD if requested. + This work was done by Fabian "ryg" Giesen. SSE2 is used by + default, but NEON must be enabled explicitly; see docs. + + With other JPEG optimizations included in this version, we see + 2x speedup on a JPEG on an x86 machine, and a 1.5x speedup + on a JPEG on an ARM machine, relative to previous versions of this + library. The same results will not obtain for all JPGs and for all + x86/ARM machines. (Note that progressive JPEGs are significantly + slower to decode than regular JPEGs.) This doesn't mean that this + is the fastest JPEG decoder in the land; rather, it brings it + closer to parity with standard libraries. If you want the fastest + decode, look elsewhere. (See "Philosophy" section of docs below.) + + See final bullet items below for more info on SIMD. + + - Added STBI_MALLOC, STBI_REALLOC, and STBI_FREE macros for replacing + the memory allocator. Unlike other STBI libraries, these macros don't + support a context parameter, so if you need to pass a context in to + the allocator, you'll have to store it in a global or a thread-local + variable. + + - Split existing STBI_NO_HDR flag into two flags, STBI_NO_HDR and + STBI_NO_LINEAR. + STBI_NO_HDR: suppress implementation of .hdr reader format + STBI_NO_LINEAR: suppress high-dynamic-range light-linear float API + + - You can suppress implementation of any of the decoders to reduce + your code footprint by #defining one or more of the following + symbols before creating the implementation. + + STBI_NO_JPEG + STBI_NO_PNG + STBI_NO_BMP + STBI_NO_PSD + STBI_NO_TGA + STBI_NO_GIF + STBI_NO_HDR + STBI_NO_PIC + STBI_NO_PNM (.ppm and .pgm) + + - You can request *only* certain decoders and suppress all other ones + (this will be more forward-compatible, as addition of new decoders + doesn't require you to disable them explicitly): + + STBI_ONLY_JPEG + STBI_ONLY_PNG + STBI_ONLY_BMP + STBI_ONLY_PSD + STBI_ONLY_TGA + STBI_ONLY_GIF + STBI_ONLY_HDR + STBI_ONLY_PIC + STBI_ONLY_PNM (.ppm and .pgm) + + Note that you can define multiples of these, and you will get all + of them ("only x" and "only y" is interpreted to mean "only x&y"). + + - If you use STBI_NO_PNG (or _ONLY_ without PNG), and you still + want the zlib decoder to be available, #define STBI_SUPPORT_ZLIB + + - Compilation of all SIMD code can be suppressed with + #define STBI_NO_SIMD + It should not be necessary to disable SIMD unless you have issues + compiling (e.g. using an x86 compiler which doesn't support SSE + intrinsics or that doesn't support the method used to detect + SSE2 support at run-time), and even those can be reported as + bugs so I can refine the built-in compile-time checking to be + smarter. + + - The old STBI_SIMD system which allowed installing a user-defined + IDCT etc. has been removed. If you need this, don't upgrade. My + assumption is that almost nobody was doing this, and those who + were will find the built-in SIMD more satisfactory anyway. + + - RGB values computed for JPEG images are slightly different from + previous versions of stb_image. (This is due to using less + integer precision in SIMD.) The C code has been adjusted so + that the same RGB values will be computed regardless of whether + SIMD support is available, so your app should always produce + consistent results. But these results are slightly different from + previous versions. (Specifically, about 3% of available YCbCr values + will compute different RGB results from pre-1.49 versions by +-1; + most of the deviating values are one smaller in the G channel.) + + - If you must produce consistent results with previous versions of + stb_image, #define STBI_JPEG_OLD and you will get the same results + you used to; however, you will not get the SIMD speedups for + the YCbCr-to-RGB conversion step (although you should still see + significant JPEG speedup from the other changes). + + Please note that STBI_JPEG_OLD is a temporary feature; it will be + removed in future versions of the library. It is only intended for + near-term back-compatibility use. + + + Latest revision history: + 2.06 (2015-04-19) fix bug where PSD returns wrong '*comp' value + 2.05 (2015-04-19) fix bug in progressive JPEG handling, fix warning + 2.04 (2015-04-15) try to re-enable SIMD on MinGW 64-bit + 2.03 (2015-04-12) additional corruption checking + stbi_set_flip_vertically_on_load + fix NEON support; fix mingw support + 2.02 (2015-01-19) fix incorrect assert, fix warning + 2.01 (2015-01-17) fix various warnings + 2.00b (2014-12-25) fix STBI_MALLOC in progressive JPEG + 2.00 (2014-12-25) optimize JPEG, including x86 SSE2 & ARM NEON SIMD + progressive JPEG + PGM/PPM support + STBI_MALLOC,STBI_REALLOC,STBI_FREE + STBI_NO_*, STBI_ONLY_* + GIF bugfix + 1.48 (2014-12-14) fix incorrectly-named assert() + 1.47 (2014-12-14) 1/2/4-bit PNG support (both grayscale and paletted) + optimize PNG + fix bug in interlaced PNG with user-specified channel count + + See end of file for full revision history. + + + ============================ Contributors ========================= + + Image formats Bug fixes & warning fixes + Sean Barrett (jpeg, png, bmp) Marc LeBlanc + Nicolas Schulz (hdr, psd) Christpher Lloyd + Jonathan Dummer (tga) Dave Moore + Jean-Marc Lienher (gif) Won Chun + Tom Seddon (pic) the Horde3D community + Thatcher Ulrich (psd) Janez Zemva + Ken Miller (pgm, ppm) Jonathan Blow + Laurent Gomila + Aruelien Pocheville + Extensions, features Ryamond Barbiero + Jetro Lauha (stbi_info) David Woo + Martin "SpartanJ" Golini (stbi_info) Martin Golini + James "moose2000" Brown (iPhone PNG) Roy Eltham + Ben "Disch" Wenger (io callbacks) Luke Graham + Omar Cornut (1/2/4-bit PNG) Thomas Ruf + Nicolas Guillemot (vertical flip) John Bartholomew + Ken Hamada + Optimizations & bugfixes Cort Stratton + Fabian "ryg" Giesen Blazej Dariusz Roszkowski + Arseny Kapoulkine Thibault Reuille + Paul Du Bois + Guillaume George + If your name should be here but Jerry Jansson + isn't, let Sean know. Hayaki Saito + Johan Duparc + Ronny Chevalier + Michal Cichon + Tero Hanninen + Sergio Gonzalez + Cass Everitt + Engin Manap + Martins Mozeiko + Joseph Thomson + Phil Jordan + +License: + This software is in the public domain. Where that dedication is not + recognized, you are granted a perpetual, irrevocable license to copy + and modify this file however you want. + +*/ + #ifndef STBI_INCLUDE_STB_IMAGE_H #define STBI_INCLUDE_STB_IMAGE_H -#include +// 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): +// int x,y,n; +// unsigned char *data = stbi_load(filename, &x, &y, &n, 0); +// // ... process data if not NULL ... +// // ... x = width, y = height, n = # 8-bit components per pixel ... +// // ... replace '0' with '1'..'4' to force that many components per pixel +// // ... but 'n' will always be the number that it would have been if you said 0 +// stbi_image_free(data) +// +// Standard parameters: +// int *x -- outputs image width in pixels +// int *y -- outputs image height in pixels +// int *comp -- outputs # of image components in image file +// int req_comp -- if non-zero, # of image components requested in result +// +// The return value from an image loader is an 'unsigned char *' which points +// to the pixel data, or NULL on an allocation failure or if the image is +// corrupt or invalid. The pixel data consists of *y scanlines of *x pixels, +// 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. +// +// An output image with N components has the following components interleaved +// in this order in each pixel: +// +// N=#comp components +// 1 grey +// 2 grey, alpha +// 3 red, green, blue +// 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 +// more user-friendly ones. +// +// Paletted PNG, BMP, GIF, and PIC images are automatically depalettized. +// +// =========================================================================== +// +// Philosophy +// +// stb libraries are designed with the following priorities: +// +// 1. easy to use +// 2. easy to maintain +// 3. good performance +// +// 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 +// 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. +// +// - Portable ("ease of use") +// - Small footprint ("easy to maintain") +// - No dependencies ("ease of use") +// +// =========================================================================== +// +// I/O callbacks +// +// I/O callbacks allow you to read from arbitrary sources, like packaged +// files or some other source. Data read from callbacks are processed +// through a small internal buffer (currently 128 bytes) to try to reduce +// overhead. +// +// The three functions you must define are "read" (reads some bytes of data), +// "skip" (skips some bytes of data), "eof" (reports if the stream is at the end). +// +// =========================================================================== +// +// SIMD support +// +// The JPEG decoder will try to automatically use SIMD kernels on x86 when +// supported by the compiler. For ARM Neon support, you must explicitly +// request it. +// +// (The old do-it-yourself SIMD API is no longer supported in the current +// code.) +// +// On x86, SSE2 will automatically be used when available based on a run-time +// test; if not, the generic C versions are used as a fall-back. On ARM targets, +// the typical path is to have separate builds for NEON and non-NEON devices +// (at least this is true for iOS and Android). Therefore, the NEON support is +// toggled by a build flag: define STBI_NEON to get NEON loops. +// +// The output of the JPEG decoder is slightly different from versions where +// SIMD support was introduced (that is, for versions before 1.49). The +// difference is only +-1 in the 8-bit RGB channels, and only on a small +// fraction of pixels. You can force the pre-1.49 behavior by defining +// STBI_JPEG_OLD, but this will disable some of the SIMD decoding path +// and hence cost some performance. +// +// If for some reason you do not want to use any of SIMD code, or if +// you have issues compiling it, you can disable it entirely by +// defining STBI_NO_SIMD. +// +// =========================================================================== +// +// 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; +// both of these constants can be reconfigured through this interface: +// +// stbi_hdr_to_ldr_gamma(2.2f); +// stbi_hdr_to_ldr_scale(1.0f); +// +// (note, do not use _inverse_ constants; stbi_image will invert them +// appropriately). +// +// Additionally, there is a new, parallel interface for loading files as +// (linear) floats to preserve the full dynamic range: +// +// float *data = stbi_loadf(filename, &x, &y, &n, 0); +// +// If you load LDR images through this interface, those images will +// be promoted to floating point values, run through the inverse of +// constants corresponding to the above: +// +// stbi_ldr_to_hdr_scale(1.0f); +// stbi_ldr_to_hdr_gamma(2.2f); +// +// Finally, given a filename (or an open file or memory block--see header +// file for details) containing image data, you can query for the "most +// appropriate" interface to use (that is, whether the image is HDR or +// not), using: +// +// stbi_is_hdr(char *filename); +// +// =========================================================================== +// +// 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). +// +// 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 +// says there's premultiplied data (currently only happens in iPhone images, +// and only if iPhone convert-to-rgb processing is on). +// + + +#ifndef STBI_NO_STDIO +#include +#endif // STBI_NO_STDIO #define STBI_VERSION 1 enum { - STBI_default = 0, + STBI_default = 0, // only used for req_comp STBI_grey = 1, STBI_grey_alpha = 2, @@ -15,6 +400,8 @@ enum STBI_rgb_alpha = 4 }; +typedef unsigned char stbi_uc; + #ifdef __cplusplus extern "C" { #endif @@ -25,26 +412,110 @@ extern "C" { #define STBIDEF extern #endif +////////////////////////////////////////////////////////////////////////////// +// +// PRIMARY API - works on images of any type +// + +// +// load image by filename, open file, or memory buffer +// + typedef struct { - int (*read) (void *user,char *data,int size); - void (*skip) (void *user,int n); - int (*eof) (void *user); + int (*read) (void *user,char *data,int size); // fill 'data' with 'size' bytes. return number of bytes actually read + void (*skip) (void *user,int n); // skip the next 'n' bytes, or 'unget' the last -n bytes if negative + int (*eof) (void *user); // returns nonzero if we are at end of file/data } stbi_io_callbacks; -STBIDEF uint8_t *stbi_load (char const *filename, int *x, int *y, int *comp, int req_comp); +STBIDEF stbi_uc *stbi_load (char const *filename, int *x, int *y, int *comp, int req_comp); +STBIDEF stbi_uc *stbi_load_from_memory (stbi_uc const *buffer, int len , int *x, int *y, int *comp, int req_comp); +STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk , void *user, int *x, int *y, int *comp, int req_comp); #ifndef STBI_NO_STDIO -STBIDEF uint8_t *stbi_load_from_file (FILE *f, int *x, int *y, int *comp, int req_comp); +STBIDEF stbi_uc *stbi_load_from_file (FILE *f, int *x, int *y, int *comp, int req_comp); +// for stbi_load_from_file, file pointer is left pointing immediately after image #endif +#ifndef STBI_NO_LINEAR + STBIDEF float *stbi_loadf (char const *filename, int *x, int *y, int *comp, int req_comp); + STBIDEF float *stbi_loadf_from_memory (stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp); + STBIDEF float *stbi_loadf_from_callbacks (stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp); + + #ifndef STBI_NO_STDIO + STBIDEF float *stbi_loadf_from_file (FILE *f, int *x, int *y, int *comp, int req_comp); + #endif +#endif + +#ifndef STBI_NO_HDR + STBIDEF void stbi_hdr_to_ldr_gamma(float gamma); + STBIDEF void stbi_hdr_to_ldr_scale(float scale); +#endif + +#ifndef STBI_NO_LINEAR + STBIDEF void stbi_ldr_to_hdr_gamma(float gamma); + STBIDEF void stbi_ldr_to_hdr_scale(float scale); +#endif // STBI_NO_HDR + +// stbi_is_hdr is always defined, but always returns false if STBI_NO_HDR +STBIDEF int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user); +STBIDEF int stbi_is_hdr_from_memory(stbi_uc const *buffer, int len); +#ifndef STBI_NO_STDIO +STBIDEF int stbi_is_hdr (char const *filename); +STBIDEF int stbi_is_hdr_from_file(FILE *f); +#endif // STBI_NO_STDIO + + +// get a VERY brief reason for failure +// NOT THREADSAFE +STBIDEF const char *stbi_failure_reason (void); + +// free the loaded image -- this is just free() +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); + +#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); + +#endif + + + +// for image formats that explicitly notate that they have premultiplied alpha, +// we just return the colors as stored in the file. set this flag to force +// unpremultiplication. results are undefined if the unpremultiply overflow. +STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply); + +// indicate whether we should process iphone images back to canonical format, +// or just pass them through "as-is" +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); + +// 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); STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, int len, int initial_size, int *outlen, int parse_header); +STBIDEF char *stbi_zlib_decode_malloc(const char *buffer, int len, int *outlen); +STBIDEF int stbi_zlib_decode_buffer(char *obuffer, int olen, const char *ibuffer, int ilen); + +STBIDEF char *stbi_zlib_decode_noheader_malloc(const char *buffer, int len, int *outlen); +STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char *ibuffer, int ilen); + #ifdef __cplusplus } #endif -#endif +// +// +//// end header file ///////////////////////////////////////////////////// +#endif // STBI_INCLUDE_STB_IMAGE_H #ifdef STB_IMAGE_IMPLEMENTATION @@ -61,9 +532,24 @@ STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, i #ifndef STBI_ONLY_BMP #define STBI_NO_BMP #endif + #ifndef STBI_ONLY_PSD + #define STBI_NO_PSD + #endif #ifndef STBI_ONLY_TGA #define STBI_NO_TGA #endif + #ifndef STBI_ONLY_GIF + #define STBI_NO_GIF + #endif + #ifndef STBI_ONLY_HDR + #define STBI_NO_HDR + #endif + #ifndef STBI_ONLY_PIC + #define STBI_NO_PIC + #endif + #ifndef STBI_ONLY_PNM + #define STBI_NO_PNM + #endif #endif #if defined(STBI_NO_PNG) && !defined(STBI_SUPPORT_ZLIB) && !defined(STBI_NO_ZLIB) @@ -72,12 +558,12 @@ STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, i #include -#include +#include // ptrdiff_t on osx #include #include #if !defined(STBI_NO_LINEAR) || !defined(STBI_NO_HDR) -#include +#include // ldexp #endif #ifndef STBI_NO_STDIO @@ -89,9 +575,33 @@ STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, i #define STBI_ASSERT(x) assert(x) #endif -#include -typedef unsigned char validate_uint32[sizeof(uint32_t)==4 ? 1 : -1]; +#ifndef _MSC_VER + #ifdef __cplusplus + #define stbi_inline inline + #else + #define stbi_inline + #endif +#else + #define stbi_inline __forceinline +#endif + + +#ifdef _MSC_VER +typedef unsigned short stbi__uint16; +typedef signed short stbi__int16; +typedef unsigned int stbi__uint32; +typedef signed int stbi__int32; +#else +#include +typedef uint16_t stbi__uint16; +typedef int16_t stbi__int16; +typedef uint32_t stbi__uint32; +typedef int32_t stbi__int32; +#endif + +// should produce compiler error if size is wrong +typedef unsigned char validate_uint32[sizeof(stbi__uint32)==4 ? 1 : -1]; #ifdef _MSC_VER #define STBI_NOTUSED(v) (void)(v) @@ -109,6 +619,21 @@ typedef unsigned char validate_uint32[sizeof(uint32_t)==4 ? 1 : -1]; #define stbi_lrot(x,y) (((x) << (y)) | ((x) >> (32 - (y)))) #endif +#if defined(STBI_MALLOC) && defined(STBI_FREE) && defined(STBI_REALLOC) +// ok +#elif !defined(STBI_MALLOC) && !defined(STBI_FREE) && !defined(STBI_REALLOC) +// ok +#else +#error "Must define all or none of STBI_MALLOC, STBI_FREE, and STBI_REALLOC." +#endif + +#ifndef STBI_MALLOC +#define STBI_MALLOC(sz) malloc(sz) +#define STBI_REALLOC(p,sz) realloc(p,sz) +#define STBI_FREE(p) free(p) +#endif + +// x86/x64 detection #if defined(__x86_64__) || defined(_M_X64) #define STBI__X64_TARGET #elif defined(__i386) || defined(_M_IX86) @@ -116,10 +641,26 @@ typedef unsigned char validate_uint32[sizeof(uint32_t)==4 ? 1 : -1]; #endif #if defined(__GNUC__) && (defined(STBI__X86_TARGET) || defined(STBI__X64_TARGET)) && !defined(__SSE2__) && !defined(STBI_NO_SIMD) +// NOTE: not clear do we actually need this for the 64-bit path? +// gcc doesn't support sse2 intrinsics unless you compile with -msse2, +// (but compiling with -msse2 allows the compiler to use SSE2 everywhere; +// this is just broken and gcc are jerks for not fixing it properly +// http://www.virtualdub.org/blog/pivot/entry.php?id=363 ) #define STBI_NO_SIMD #endif #if defined(__MINGW32__) && defined(STBI__X86_TARGET) && !defined(STBI_MINGW_ENABLE_SSE2) && !defined(STBI_NO_SIMD) +// Note that __MINGW32__ doesn't actually mean 32-bit, so we have to avoid STBI__X64_TARGET +// +// 32-bit MinGW wants ESP to be 16-byte aligned, but this is not in the +// Windows ABI and VC++ as well as Windows DLLs don't maintain that invariant. +// As a result, enabling SSE2 on 32-bit MinGW is dangerous when not +// simultaneously enabling "-mstackrealign". +// +// See https://github.com/nothings/stb/issues/81 for more information. +// +// So default to no SSE2 on 32-bit MinGW. If you've read this far and added +// -mstackrealign to your build settings, feel free to #define STBI_MINGW_ENABLE_SSE2. #define STBI_NO_SIMD #endif @@ -129,8 +670,8 @@ typedef unsigned char validate_uint32[sizeof(uint32_t)==4 ? 1 : -1]; #ifdef _MSC_VER -#if _MSC_VER >= 1400 -#include +#if _MSC_VER >= 1400 // not VC6 +#include // __cpuid static int stbi__cpuid3(void) { int info[4]; @@ -157,26 +698,31 @@ static int stbi__sse2_available() int info3 = stbi__cpuid3(); return ((info3 >> 26) & 1) != 0; } -#else +#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(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__) >= 408 +#if defined(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__) >= 408 // GCC 4.8 or later + // GCC 4.8+ has a nice way to do this return __builtin_cpu_supports("sse2"); #else + // portable way to do this, preferably without using GCC inline ASM? + // just bail for now. return 0; #endif } #endif #endif +// ARM NEON #if defined(STBI_NO_SIMD) && defined(STBI_NEON) #undef STBI_NEON #endif #ifdef STBI_NEON #include +// assume GCC or Clang on ARM targets #define STBI_SIMD_ALIGN(type, name) type name __attribute__((aligned(16))) #endif @@ -184,9 +730,15 @@ static int stbi__sse2_available() #define STBI_SIMD_ALIGN(type, name) type name #endif +/////////////////////////////////////////////// +// +// stbi__context struct and start_xxx functions + +// stbi__context structure is our basic context used by all images, so it +// contains all the IO context, plus some basic image information typedef struct { - uint32_t img_x, img_y; + stbi__uint32 img_x, img_y; int img_n, img_out_n; stbi_io_callbacks io; @@ -194,15 +746,25 @@ typedef struct int read_from_callbacks; int buflen; - uint8_t buffer_start[128]; + stbi_uc buffer_start[128]; - uint8_t *img_buffer, *img_buffer_end; - uint8_t *img_buffer_original; + stbi_uc *img_buffer, *img_buffer_end; + stbi_uc *img_buffer_original; } stbi__context; static void stbi__refill_buffer(stbi__context *s); +// initialize a memory-decode context +static void stbi__start_mem(stbi__context *s, stbi_uc const *buffer, int len) +{ + s->io.read = NULL; + s->read_from_callbacks = 0; + s->img_buffer = s->img_buffer_original = (stbi_uc *) buffer; + s->img_buffer_end = (stbi_uc *) buffer+len; +} + +// initialize a callback-based context static void stbi__start_callbacks(stbi__context *s, stbi_io_callbacks *c, void *user) { s->io = *c; @@ -242,41 +804,95 @@ static void stbi__start_file(stbi__context *s, FILE *f) stbi__start_callbacks(s, &stbi__stdio_callbacks, (void *) f); } -#endif +//static void stop_file(stbi__context *s) { } + +#endif // !STBI_NO_STDIO static void stbi__rewind(stbi__context *s) { + // conceptually rewind SHOULD rewind to the beginning of the stream, + // but we just rewind to the beginning of the initial buffer, because + // we only use it after doing 'test', which only ever looks at at most 92 bytes s->img_buffer = s->img_buffer_original; } #ifndef STBI_NO_JPEG static int stbi__jpeg_test(stbi__context *s); -static uint8_t *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static stbi_uc *stbi__jpeg_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp); #endif #ifndef STBI_NO_PNG static int stbi__png_test(stbi__context *s); -static uint8_t *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static stbi_uc *stbi__png_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp); #endif #ifndef STBI_NO_BMP static int stbi__bmp_test(stbi__context *s); -static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp); #endif #ifndef STBI_NO_TGA static int stbi__tga_test(stbi__context *s); -static uint8_t *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp); #endif +#ifndef STBI_NO_PSD +static int stbi__psd_test(stbi__context *s); +static stbi_uc *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_HDR +static int stbi__hdr_test(stbi__context *s); +static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_PIC +static int stbi__pic_test(stbi__context *s); +static stbi_uc *stbi__pic_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_GIF +static int stbi__gif_test(stbi__context *s); +static stbi_uc *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +#ifndef STBI_NO_PNM +static int stbi__pnm_test(stbi__context *s); +static stbi_uc *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp); +static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp); +#endif + +// this is not threadsafe static const char *stbi__g_failure_reason; +STBIDEF const char *stbi_failure_reason(void) +{ + return stbi__g_failure_reason; +} + static int stbi__err(const char *str) { stbi__g_failure_reason = str; return 0; } +static void *stbi__malloc(size_t size) +{ + return STBI_MALLOC(size); +} + +// stbi__err - error +// stbi__errpf - error returning pointer to float +// stbi__errpuc - error returning pointer to unsigned char + #ifdef STBI_NO_FAILURE_STRINGS #define stbi__err(x,y) 0 #elif defined(STBI_FAILURE_USERMSG) @@ -288,8 +904,26 @@ static int stbi__err(const char *str) #define stbi__errpf(x,y) ((float *) (stbi__err(x,y)?NULL:NULL)) #define stbi__errpuc(x,y) ((unsigned char *) (stbi__err(x,y)?NULL:NULL)) +STBIDEF void stbi_image_free(void *retval_from_stbi_load) +{ + STBI_FREE(retval_from_stbi_load); +} + +#ifndef STBI_NO_LINEAR +static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp); +#endif + +#ifndef STBI_NO_HDR +static stbi_uc *stbi__hdr_to_ldr(float *data, int x, int y, int comp); +#endif + static int stbi__vertically_flip_on_load = 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; +} + static unsigned char *stbi__load_main(stbi__context *s, int *x, int *y, int *comp, int req_comp) { #ifndef STBI_NO_JPEG @@ -301,8 +935,28 @@ static unsigned char *stbi__load_main(stbi__context *s, int *x, int *y, int *com #ifndef STBI_NO_BMP if (stbi__bmp_test(s)) return stbi__bmp_load(s,x,y,comp,req_comp); #endif + #ifndef STBI_NO_GIF + if (stbi__gif_test(s)) return stbi__gif_load(s,x,y,comp,req_comp); + #endif + #ifndef STBI_NO_PSD + if (stbi__psd_test(s)) return stbi__psd_load(s,x,y,comp,req_comp); + #endif + #ifndef STBI_NO_PIC + if (stbi__pic_test(s)) return stbi__pic_load(s,x,y,comp,req_comp); + #endif + #ifndef STBI_NO_PNM + if (stbi__pnm_test(s)) return stbi__pnm_load(s,x,y,comp,req_comp); + #endif + + #ifndef STBI_NO_HDR + if (stbi__hdr_test(s)) { + float *hdr = stbi__hdr_load(s, x,y,comp,req_comp); + return stbi__hdr_to_ldr(hdr, *x, *y, req_comp ? req_comp : *comp); + } + #endif #ifndef STBI_NO_TGA + // test tga last because it's a crappy test! if (stbi__tga_test(s)) return stbi__tga_load(s,x,y,comp,req_comp); #endif @@ -314,19 +968,16 @@ static unsigned char *stbi__load_flip(stbi__context *s, int *x, int *y, int *com { unsigned char *result = stbi__load_main(s, x, y, comp, req_comp); - if (stbi__vertically_flip_on_load && result != NULL) - { - int row,col,z; - uint8_t temp; + if (stbi__vertically_flip_on_load && result != NULL) { int w = *x, h = *y; int depth = req_comp ? req_comp : *comp; + int row,col,z; + stbi_uc temp; - for (row = 0; row < (h>>1); row++) - { - for (col = 0; col < w; col++) - { - for (z = 0; z < depth; z++) - { + // @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; @@ -338,6 +989,30 @@ static unsigned char *stbi__load_flip(stbi__context *s, int *x, int *y, int *com return result; } +#ifndef STBI_NO_HDR +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; + } + } + } + } +} +#endif + + #ifndef STBI_NO_STDIO static FILE *stbi__fopen(char const *filename, char const *mode) @@ -353,32 +1028,166 @@ static FILE *stbi__fopen(char const *filename, char const *mode) } -STBIDEF uint8_t *stbi_load(char const *filename, int *x, int *y, int *comp, int req_comp) +STBIDEF stbi_uc *stbi_load(char const *filename, int *x, int *y, int *comp, int req_comp) { - unsigned char *result; FILE *f = stbi__fopen(filename, "rb"); - if (!f) - return stbi__errpuc("can't fopen", "Unable to open file"); + unsigned char *result; + if (!f) return stbi__errpuc("can't fopen", "Unable to open file"); result = stbi_load_from_file(f,x,y,comp,req_comp); fclose(f); return result; } -STBIDEF uint8_t *stbi_load_from_file(FILE *f, int *x, int *y, int *comp, int req_comp) +STBIDEF stbi_uc *stbi_load_from_file(FILE *f, int *x, int *y, int *comp, int req_comp) { unsigned char *result; stbi__context s; stbi__start_file(&s,f); result = stbi__load_flip(&s,x,y,comp,req_comp); - if (result) + if (result) { + // need to 'unget' all the characters in the IO buffer fseek(f, - (int) (s.img_buffer_end - s.img_buffer), SEEK_CUR); + } return result; } -#endif +#endif //!STBI_NO_STDIO + +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; + stbi__start_mem(&s,buffer,len); + return stbi__load_flip(&s,x,y,comp,req_comp); +} + +STBIDEF stbi_uc *stbi_load_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user); + return stbi__load_flip(&s,x,y,comp,req_comp); +} + +#ifndef STBI_NO_LINEAR +static float *stbi__loadf_main(stbi__context *s, int *x, int *y, int *comp, int req_comp) +{ + unsigned char *data; + #ifndef STBI_NO_HDR + if (stbi__hdr_test(s)) { + float *hdr_data = stbi__hdr_load(s,x,y,comp,req_comp); + if (hdr_data) + stbi__float_postprocess(hdr_data,x,y,comp,req_comp); + return hdr_data; + } + #endif + data = stbi__load_flip(s, x, y, comp, req_comp); + if (data) + return stbi__ldr_to_hdr(data, *x, *y, req_comp ? req_comp : *comp); + return stbi__errpf("unknown image type", "Image not of any known type, or corrupt"); +} + +STBIDEF float *stbi_loadf_from_memory(stbi_uc const *buffer, int len, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__loadf_main(&s,x,y,comp,req_comp); +} + +STBIDEF float *stbi_loadf_from_callbacks(stbi_io_callbacks const *clbk, void *user, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user); + return stbi__loadf_main(&s,x,y,comp,req_comp); +} + +#ifndef STBI_NO_STDIO +STBIDEF float *stbi_loadf(char const *filename, int *x, int *y, int *comp, int req_comp) +{ + float *result; + FILE *f = stbi__fopen(filename, "rb"); + if (!f) return stbi__errpf("can't fopen", "Unable to open file"); + result = stbi_loadf_from_file(f,x,y,comp,req_comp); + fclose(f); + return result; +} + +STBIDEF float *stbi_loadf_from_file(FILE *f, int *x, int *y, int *comp, int req_comp) +{ + stbi__context s; + stbi__start_file(&s,f); + return stbi__loadf_main(&s,x,y,comp,req_comp); +} +#endif // !STBI_NO_STDIO + +#endif // !STBI_NO_LINEAR + +// these is-hdr-or-not is defined independent of whether STBI_NO_LINEAR is +// defined, for API simplicity; if STBI_NO_LINEAR is defined, it always +// reports false! + +STBIDEF int stbi_is_hdr_from_memory(stbi_uc const *buffer, int len) +{ + #ifndef STBI_NO_HDR + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__hdr_test(&s); + #else + STBI_NOTUSED(buffer); + STBI_NOTUSED(len); + return 0; + #endif +} + +#ifndef STBI_NO_STDIO +STBIDEF int stbi_is_hdr (char const *filename) +{ + FILE *f = stbi__fopen(filename, "rb"); + int result=0; + if (f) { + result = stbi_is_hdr_from_file(f); + fclose(f); + } + return result; +} + +STBIDEF int stbi_is_hdr_from_file(FILE *f) +{ + #ifndef STBI_NO_HDR + stbi__context s; + stbi__start_file(&s,f); + return stbi__hdr_test(&s); + #else + return 0; + #endif +} +#endif // !STBI_NO_STDIO + +STBIDEF int stbi_is_hdr_from_callbacks(stbi_io_callbacks const *clbk, void *user) +{ + #ifndef STBI_NO_HDR + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) clbk, user); + return stbi__hdr_test(&s); + #else + return 0; + #endif +} static float stbi__h2l_gamma_i=1.0f/2.2f, stbi__h2l_scale_i=1.0f; static float stbi__l2h_gamma=2.2f, stbi__l2h_scale=1.0f; +#ifndef STBI_NO_LINEAR +STBIDEF void stbi_ldr_to_hdr_gamma(float gamma) { stbi__l2h_gamma = gamma; } +STBIDEF void stbi_ldr_to_hdr_scale(float scale) { stbi__l2h_scale = scale; } +#endif + +STBIDEF void stbi_hdr_to_ldr_gamma(float gamma) { stbi__h2l_gamma_i = 1/gamma; } +STBIDEF void stbi_hdr_to_ldr_scale(float scale) { stbi__h2l_scale_i = 1/scale; } + + +////////////////////////////////////////////////////////////////////////////// +// +// Common code used by all image loaders +// + enum { STBI__SCAN_load=0, @@ -389,8 +1198,9 @@ enum static void stbi__refill_buffer(stbi__context *s) { int n = (s->io.read)(s->io_user_data,(char*)s->buffer_start,s->buflen); - if (n == 0) - { + 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 s->read_from_callbacks = 0; s->img_buffer = s->buffer_start; s->img_buffer_end = s->buffer_start+1; @@ -401,7 +1211,7 @@ static void stbi__refill_buffer(stbi__context *s) } } -static INLINE uint8_t stbi__get8(stbi__context *s) +stbi_inline static stbi_uc stbi__get8(stbi__context *s) { if (s->img_buffer < s->img_buffer_end) return *s->img_buffer++; @@ -412,12 +1222,12 @@ static INLINE uint8_t stbi__get8(stbi__context *s) return 0; } -static INLINE int stbi__at_eof(stbi__context *s) +stbi_inline static int stbi__at_eof(stbi__context *s) { - if (s->io.read) - { - if (!(s->io.eof)(s->io_user_data)) - return 0; + if (s->io.read) { + if (!(s->io.eof)(s->io_user_data)) return 0; + // if feof() is true, check if buffer = end + // special case: we've only got the special 0 character at the end if (s->read_from_callbacks == 0) return 1; } @@ -441,7 +1251,7 @@ static void stbi__skip(stbi__context *s, int n) s->img_buffer += n; } -static int stbi__getn(stbi__context *s, uint8_t *buffer, int n) +static int stbi__getn(stbi__context *s, stbi_uc *buffer, int n) { if (s->io.read) { int blen = (int) (s->img_buffer_end - s->img_buffer); @@ -471,9 +1281,9 @@ static int stbi__get16be(stbi__context *s) return (z << 8) + stbi__get8(s); } -static uint32_t stbi__get32be(stbi__context *s) +static stbi__uint32 stbi__get32be(stbi__context *s) { - uint32_t z = stbi__get16be(s); + stbi__uint32 z = stbi__get16be(s); return (z << 16) + stbi__get16be(s); } @@ -483,17 +1293,29 @@ static int stbi__get16le(stbi__context *s) return z + (stbi__get8(s) << 8); } -static uint32_t stbi__get32le(stbi__context *s) +static stbi__uint32 stbi__get32le(stbi__context *s) { - uint32_t z = stbi__get16le(s); + stbi__uint32 z = stbi__get16le(s); return z + (stbi__get16le(s) << 16); } -#define STBI__BYTECAST(x) ((uint8_t) ((x) & 255)) +#define STBI__BYTECAST(x) ((stbi_uc) ((x) & 255)) // truncate int to byte without warnings -static uint8_t stbi__compute_y(int r, int g, int b) + +////////////////////////////////////////////////////////////////////////////// +// +// generic converter from built-in img_n to req_comp +// individual types do this automatically as much as possible (e.g. jpeg +// does all cases internally since it needs to colorspace convert anyway, +// and it never has alpha, so very few cases ). png can automatically +// interleave an alpha=255 channel, but falls back to this for other cases +// +// assume data buffer is malloced, so malloc a new one and free that one +// only failure mode is malloc failing + +static stbi_uc stbi__compute_y(int r, int g, int b) { - return (uint8_t) (((r*77) + (g*150) + (29*b)) >> 8); + return (stbi_uc) (((r*77) + (g*150) + (29*b)) >> 8); } static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int req_comp, unsigned int x, unsigned int y) @@ -504,11 +1326,9 @@ static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int r if (req_comp == img_n) return data; STBI_ASSERT(req_comp >= 1 && req_comp <= 4); - good = (unsigned char *) malloc(req_comp * x * y); - - if (good == NULL) - { - free(data); + good = (unsigned char *) stbi__malloc(req_comp * x * y); + if (good == NULL) { + STBI_FREE(data); return stbi__errpuc("outofmem", "Out of memory"); } @@ -518,8 +1338,9 @@ static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int r #define COMBO(a,b) ((a)*8+(b)) #define CASE(a,b) case COMBO(a,b): for(i=x-1; i >= 0; --i, src += a, dest += b) - switch (COMBO(img_n, req_comp)) - { + // 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 (COMBO(img_n, req_comp)) { CASE(1,2) dest[0]=src[0], dest[1]=255; break; CASE(1,3) dest[0]=dest[1]=dest[2]=src[0]; break; CASE(1,4) dest[0]=dest[1]=dest[2]=src[0], dest[3]=255; break; @@ -537,22 +1358,92 @@ static unsigned char *stbi__convert_format(unsigned char *data, int img_n, int r #undef CASE } - free(data); + STBI_FREE(data); return good; } +#ifndef STBI_NO_LINEAR +static float *stbi__ldr_to_hdr(stbi_uc *data, int x, int y, int comp) +{ + int i,k,n; + float *output = (float *) stbi__malloc(x * y * comp * sizeof(float)); + if (output == NULL) { STBI_FREE(data); return stbi__errpf("outofmem", "Out of memory"); } + // compute number of non-alpha components + if (comp & 1) n = comp; else n = comp-1; + for (i=0; i < x*y; ++i) { + 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; + } + STBI_FREE(data); + return output; +} +#endif + +#ifndef STBI_NO_HDR +#define stbi__float2int(x) ((int) (x)) +static stbi_uc *stbi__hdr_to_ldr(float *data, int x, int y, int comp) +{ + int i,k,n; + stbi_uc *output = (stbi_uc *) stbi__malloc(x * y * comp); + if (output == NULL) { STBI_FREE(data); return stbi__errpuc("outofmem", "Out of memory"); } + // compute number of non-alpha components + if (comp & 1) n = comp; else n = comp-1; + for (i=0; i < x*y; ++i) { + for (k=0; k < n; ++k) { + float z = (float) pow(data[i*comp+k]*stbi__h2l_scale_i, stbi__h2l_gamma_i) * 255 + 0.5f; + if (z < 0) z = 0; + if (z > 255) z = 255; + output[i*comp + k] = (stbi_uc) stbi__float2int(z); + } + if (k < comp) { + float z = data[i*comp+k] * 255 + 0.5f; + if (z < 0) z = 0; + if (z > 255) z = 255; + output[i*comp + k] = (stbi_uc) stbi__float2int(z); + } + } + STBI_FREE(data); + return output; +} +#endif + +////////////////////////////////////////////////////////////////////////////// +// +// "baseline" JPEG/JFIF decoder +// +// simple implementation +// - doesn't support delayed output of y-dimension +// - simple interface (only one output format: 8-bit interleaved RGB) +// - doesn't try to recover corrupt jpegs +// - doesn't allow partial loading, loading multiple at once +// - still fast on x86 (copying globals into locals doesn't help x86) +// - allocates lots of intermediate memory (full size of all components) +// - non-interleaved case requires this anyway +// - allows good upsampling (see next) +// high-quality +// - upsampled channels are bilinearly interpolated, even across blocks +// - quality integer IDCT derived from IJG's 'slow' +// performance +// - fast huffman; reasonable integer IDCT +// - some SIMD kernels for common paths on targets with SSE2/NEON +// - uses a lot of intermediate memory, could cache poorly + #ifndef STBI_NO_JPEG -#define FAST_BITS 9 +// huffman decoding acceleration +#define FAST_BITS 9 // larger handles more cases; smaller stomps less cache typedef struct { - uint8_t fast[1 << FAST_BITS]; - uint16_t code[256]; - uint8_t values[256]; - uint8_t size[257]; + stbi_uc fast[1 << FAST_BITS]; + // weirdly, repacking this into AoS is a 10% speed loss, instead of a win + stbi__uint16 code[256]; + stbi_uc values[256]; + stbi_uc size[257]; unsigned int maxcode[18]; - int delta[17]; + int delta[17]; // old 'firstsymbol' - old 'firstcode' } stbi__huffman; typedef struct @@ -560,13 +1451,15 @@ typedef struct stbi__context *s; stbi__huffman huff_dc[4]; stbi__huffman huff_ac[4]; - uint8_t dequant[4][64]; - int16_t fast_ac[4][1 << FAST_BITS]; + stbi_uc dequant[4][64]; + stbi__int16 fast_ac[4][1 << FAST_BITS]; +// sizes for components, interleaved MCUs int img_h_max, img_v_max; int img_mcu_x, img_mcu_y; int img_mcu_w, img_mcu_h; +// definition of jpeg image component struct { int id; @@ -576,17 +1469,17 @@ typedef struct int dc_pred; int x,y,w2,h2; - uint8_t *data; + stbi_uc *data; void *raw_data, *raw_coeff; - uint8_t *linebuf; - short *coeff; - int coeff_w, coeff_h; + stbi_uc *linebuf; + short *coeff; // progressive only + int coeff_w, coeff_h; // number of 8x8 coefficient blocks } img_comp[4]; - uint32_t code_buffer; - int code_bits; - unsigned char marker; - int nomore; + stbi__uint32 code_buffer; // jpeg entropy-coded buffer + int code_bits; // number of valid bits + unsigned char marker; // marker seen while filling entropy buffer + int nomore; // flag if we saw a marker so must stop int progressive; int spec_start; @@ -598,34 +1491,39 @@ typedef struct int scan_n, order[4]; int restart_interval, todo; - void (*idct_block_kernel)(uint8_t *out, int out_stride, short data[64]); - void (*YCbCr_to_RGB_kernel)(uint8_t *out, const uint8_t *y, const uint8_t *pcb, const uint8_t *pcr, int count, int step); - uint8_t *(*resample_row_hv_2_kernel)(uint8_t *out, uint8_t *in_near, uint8_t *in_far, int w, int hs); +// kernels + void (*idct_block_kernel)(stbi_uc *out, int out_stride, short data[64]); + void (*YCbCr_to_RGB_kernel)(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step); + stbi_uc *(*resample_row_hv_2_kernel)(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs); } stbi__jpeg; static int stbi__build_huffman(stbi__huffman *h, int *count) { int i,j,k=0,code; - + // build size list for each symbol (from JPEG spec) for (i=0; i < 16; ++i) for (j=0; j < count[i]; ++j) - h->size[k++] = (uint8_t) (i+1); + h->size[k++] = (stbi_uc) (i+1); h->size[k] = 0; + // compute actual symbols (from jpeg spec) code = 0; k = 0; for(j=1; j <= 16; ++j) { + // compute delta to add to code to compute symbol id h->delta[j] = k - code; if (h->size[k] == j) { while (h->size[k] == j) - h->code[k++] = (uint16_t) (code++); + h->code[k++] = (stbi__uint16) (code++); if (code-1 >= (1 << 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); code <<= 1; } h->maxcode[j] = 0xffffffff; + // build non-spec acceleration table; 255 is flag for not-accelerated memset(h->fast, 255, 1 << FAST_BITS); for (i=0; i < k; ++i) { int s = h->size[i]; @@ -633,18 +1531,20 @@ static int stbi__build_huffman(stbi__huffman *h, int *count) int c = h->code[i] << (FAST_BITS-s); int m = 1 << (FAST_BITS-s); for (j=0; j < m; ++j) { - h->fast[c+j] = (uint8_t) i; + h->fast[c+j] = (stbi_uc) i; } } } return 1; } -static void stbi__build_fast_ac(int16_t *fast_ac, stbi__huffman *h) +// build a table that decodes both magnitude and value of small ACs in +// one go. +static void stbi__build_fast_ac(stbi__int16 *fast_ac, stbi__huffman *h) { int i; for (i=0; i < (1 << FAST_BITS); ++i) { - uint8_t fast = h->fast[i]; + stbi_uc fast = h->fast[i]; fast_ac[i] = 0; if (fast < 255) { int rs = h->values[fast]; @@ -652,13 +1552,14 @@ static void stbi__build_fast_ac(int16_t *fast_ac, stbi__huffman *h) int magbits = rs & 15; int len = h->size[fast]; - if (magbits && len + magbits <= FAST_BITS) - { + if (magbits && len + magbits <= FAST_BITS) { + // magnitude code followed by receive_extend code int k = ((i << len) & ((1 << FAST_BITS) - 1)) >> (FAST_BITS - magbits); int m = 1 << (magbits - 1); if (k < m) k += (-1 << magbits) + 1; + // if the result is small enough, we can fit it in fast_ac table if (k >= -128 && k <= 127) - fast_ac[i] = (int16_t) ((k << 8) + (run << 4) + (len + magbits)); + fast_ac[i] = (stbi__int16) ((k << 8) + (run << 4) + (len + magbits)); } } } @@ -681,15 +1582,19 @@ static void stbi__grow_buffer_unsafe(stbi__jpeg *j) } while (j->code_bits <= 24); } -static uint32_t stbi__bmask[17]={0,1,3,7,15,31,63,127,255,511,1023,2047,4095,8191,16383,32767,65535}; +// (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 INLINE int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h) +// decode a jpeg huffman value from the bitstream +stbi_inline static int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h) { unsigned int temp; int c,k; if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); + // look at the top FAST_BITS and determine what symbol ID it is, + // if the code is <= FAST_BITS c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1); k = h->fast[c]; if (k < 255) { @@ -701,11 +1606,18 @@ static INLINE int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h) return h->values[k]; } + // naive test is to shift the code_buffer down so k bits are + // valid, then test against maxcode. To speed this up, we've + // preshifted maxcode left so that it has (16-k) 0s at the + // end; in other words, regardless of the number of bits, it + // wants to be compared against something shifted to have 16; + // that way we don't need to shift inside the loop. temp = j->code_buffer >> 16; for (k=FAST_BITS+1 ; ; ++k) if (temp < h->maxcode[k]) break; if (k == 17) { + // error! code not found j->code_bits -= 16; return -1; } @@ -713,32 +1625,38 @@ static INLINE int stbi__jpeg_huff_decode(stbi__jpeg *j, stbi__huffman *h) if (k > j->code_bits) return -1; + // convert the huffman code to the symbol id c = ((j->code_buffer >> (32 - k)) & stbi__bmask[k]) + h->delta[k]; STBI_ASSERT((((j->code_buffer) >> (32 - h->size[c])) & stbi__bmask[h->size[c]]) == h->code[c]); + // convert the id to a symbol j->code_bits -= k; j->code_buffer <<= k; return h->values[c]; } +// bias[n] = (-1<code_bits < n) stbi__grow_buffer_unsafe(j); - sgn = (int32_t)j->code_buffer >> 31; + sgn = (stbi__int32)j->code_buffer >> 31; // sign bit is always in MSB k = stbi_lrot(j->code_buffer, n); - STBI_ASSERT((n >= 0) && ((unsigned)n < sizeof(stbi__bmask)/sizeof(*stbi__bmask))); + 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); } -static INLINE int stbi__jpeg_get_bits(stbi__jpeg *j, int n) +// get some unsigned bits +stbi_inline static int stbi__jpeg_get_bits(stbi__jpeg *j, int n) { unsigned int k; if (j->code_bits < n) stbi__grow_buffer_unsafe(j); @@ -749,7 +1667,7 @@ static INLINE int stbi__jpeg_get_bits(stbi__jpeg *j, int n) return k; } -static INLINE int stbi__jpeg_get_bit(stbi__jpeg *j) +stbi_inline static int stbi__jpeg_get_bit(stbi__jpeg *j) { unsigned int k; if (j->code_bits < 1) stbi__grow_buffer_unsafe(j); @@ -759,7 +1677,9 @@ static INLINE int stbi__jpeg_get_bit(stbi__jpeg *j) return k & 0x80000000; } -static uint8_t stbi__jpeg_dezigzag[64+15] = +// 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] = { 0, 1, 8, 16, 9, 2, 3, 10, 17, 24, 32, 25, 18, 11, 4, 5, @@ -769,11 +1689,13 @@ static uint8_t stbi__jpeg_dezigzag[64+15] = 29, 22, 15, 23, 30, 37, 44, 51, 58, 59, 52, 45, 38, 31, 39, 46, 53, 60, 61, 54, 47, 55, 62, 63, + // let corrupt input sample past end 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63, 63 }; -static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman *hdc, stbi__huffman *hac, int16_t *fac, int b, uint8_t *dequant) +// decode one 64-entry block-- +static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman *hdc, stbi__huffman *hac, stbi__int16 *fac, int b, stbi_uc *dequant) { int diff,dc,k; int t; @@ -782,6 +1704,7 @@ static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman t = stbi__jpeg_huff_decode(j, hdc); if (t < 0) 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])); diff = t ? stbi__extend_receive(j, t) : 0; @@ -789,6 +1712,7 @@ static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman j->img_comp[b].dc_pred = dc; data[0] = (short) (dc * dequant[0]); + // decode AC components, see JPEG spec k = 1; do { unsigned int zig; @@ -796,12 +1720,12 @@ static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1); r = fac[c]; - if (r) - { - k += (r >> 4) & 15; - s = r & 15; + if (r) { // fast-AC path + k += (r >> 4) & 15; // run + s = r & 15; // combined length j->code_buffer <<= s; j->code_bits -= s; + // decode into unzigzag'd location zig = stbi__jpeg_dezigzag[k++]; data[zig] = (short) ((r >> 8) * dequant[zig]); } else { @@ -810,10 +1734,11 @@ static int stbi__jpeg_decode_block(stbi__jpeg *j, short data[64], stbi__huffman s = rs & 15; r = rs >> 4; if (s == 0) { - if (rs != 0xf0) break; + if (rs != 0xf0) break; // end block k += 16; } else { k += r; + // decode into unzigzag'd location zig = stbi__jpeg_dezigzag[k++]; data[zig] = (short) (stbi__extend_receive(j,s) * dequant[zig]); } @@ -831,7 +1756,8 @@ static int stbi__jpeg_decode_block_prog_dc(stbi__jpeg *j, short data[64], stbi__ if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); if (j->succ_high == 0) { - memset(data,0,64*sizeof(data[0])); + // 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); diff = t ? stbi__extend_receive(j, t) : 0; @@ -839,13 +1765,16 @@ static int stbi__jpeg_decode_block_prog_dc(stbi__jpeg *j, short data[64], stbi__ j->img_comp[b].dc_pred = dc; data[0] = (short) (dc << j->succ_low); } else { + // refinement scan for DC coefficient if (stbi__jpeg_get_bit(j)) data[0] += (short) (1 << j->succ_low); } return 1; } -static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__huffman *hac, int16_t *fac) +// @OPTIMIZE: store non-zigzagged during the decode passes, +// and only de-zigzag when dequantizing +static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__huffman *hac, stbi__int16 *fac) { int k; if (j->spec_start == 0) return stbi__err("can't merge dc and ac", "Corrupt JPEG"); @@ -865,10 +1794,9 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__ if (j->code_bits < 16) stbi__grow_buffer_unsafe(j); c = (j->code_buffer >> (32 - FAST_BITS)) & ((1 << FAST_BITS)-1); r = fac[c]; - if (r) - { - k += (r >> 4) & 15; - s = r & 15; + if (r) { // fast-AC path + k += (r >> 4) & 15; // run + s = r & 15; // combined length j->code_buffer <<= s; j->code_bits -= s; zig = stbi__jpeg_dezigzag[k++]; @@ -894,9 +1822,9 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__ } } } while (k <= j->spec_end); - } - else - { + } else { + // refinement scan for these AC coefficients + short bit = (short) (1 << j->succ_low); if (j->eob_run) { @@ -916,7 +1844,7 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__ k = j->spec_start; do { int r,s; - int rs = stbi__jpeg_huff_decode(j, hac); + int rs = stbi__jpeg_huff_decode(j, hac); // @OPTIMIZE see if we can use the fast path here, advance-by-r is so slow, eh if (rs < 0) return stbi__err("bad huffman code","Corrupt JPEG"); s = rs & 15; r = rs >> 4; @@ -925,19 +1853,24 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__ j->eob_run = (1 << r) - 1; if (r) j->eob_run += stbi__jpeg_get_bits(j, r); - r = 64; - } else - r = 16; + r = 64; // force end of block + } else { + // r=15 s=0 should write 16 0s, so we just do + // a run of 15 0s and then write s (which is 0), + // so we don't have to do anything special here + } } else { if (s != 1) return stbi__err("bad huffman code", "Corrupt JPEG"); + // sign bit if (stbi__jpeg_get_bit(j)) s = bit; else s = -bit; } + // advance by r while (k <= j->spec_end) { - short *p = &data[stbi__jpeg_dezigzag[k]]; + short *p = &data[stbi__jpeg_dezigzag[k++]]; if (*p != 0) { if (stbi__jpeg_get_bit(j)) if ((*p & bit)==0) { @@ -946,15 +1879,12 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__ else *p -= bit; } - ++k; } else { if (r == 0) { - if (s) - data[stbi__jpeg_dezigzag[k++]] = (short) s; + *p = (short) s; break; } --r; - ++k; } } } while (k <= j->spec_end); @@ -963,18 +1893,21 @@ static int stbi__jpeg_decode_block_prog_ac(stbi__jpeg *j, short data[64], stbi__ return 1; } -static INLINE uint8_t stbi__clamp(int x) +// take a -128..127 value and stbi__clamp it and convert to 0..255 +stbi_inline static stbi_uc stbi__clamp(int x) { + // trick to use a single test to catch both cases if ((unsigned int) x > 255) { if (x < 0) return 0; if (x > 255) return 255; } - return (uint8_t) x; + return (stbi_uc) x; } #define stbi__f2f(x) ((int) (((x) * 4096 + 0.5))) #define stbi__fsh(x) ((x) << 12) +// derived from jidctint -- DCT_ISLOW #define STBI__IDCT_1D(s0,s1,s2,s3,s4,s5,s6,s7) \ int t0,t1,t2,t3,p1,p2,p3,p4,p5,x0,x1,x2,x3; \ p2 = s2; \ @@ -1012,19 +1945,27 @@ static INLINE uint8_t stbi__clamp(int x) t1 += p2+p4; \ t0 += p1+p3; -static void stbi__idct_block(uint8_t *out, int out_stride, short data[64]) +static void stbi__idct_block(stbi_uc *out, int out_stride, short data[64]) { int i,val[64],*v=val; - uint8_t *o; + stbi_uc *o; short *d = data; + // columns for (i=0; i < 8; ++i,++d, ++v) { + // if all zeroes, shortcut -- this avoids dequantizing 0s and IDCTing if (d[ 8]==0 && d[16]==0 && d[24]==0 && d[32]==0 && d[40]==0 && d[48]==0 && d[56]==0) { + // no shortcut 0 seconds + // (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; 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]) + // constants scaled things up by 1<<12; let's bring them back + // down, but keep 2 extra bits of precision x0 += 512; x1 += 512; x2 += 512; x3 += 512; v[ 0] = (x0+t3) >> 10; v[56] = (x0-t3) >> 10; @@ -1038,11 +1979,20 @@ static void stbi__idct_block(uint8_t *out, int out_stride, short data[64]) } for (i=0, v=val, o=out; i < 8; ++i,v+=8,o+=out_stride) { + // no fast case since the first 1D IDCT spread components out STBI__IDCT_1D(v[0],v[1],v[2],v[3],v[4],v[5],v[6],v[7]) + // constants scaled things up by 1<<12, plus we had 1<<2 from first + // loop, plus horizontal and vertical each scale by sqrt(8) so together + // we've got an extra 1<<3, so 1<<17 total we need to remove. + // so we want to round that, which means adding 0.5 * 1<<17, + // aka 65536. Also, we'll end up with -128 to 127 that we want + // to encode as 0..255 by adding 128, so we'll add that before the shift x0 += 65536 + (128<<17); x1 += 65536 + (128<<17); x2 += 65536 + (128<<17); x3 += 65536 + (128<<17); + // tried computing the shifts into temps, or'ing the temps to see + // if any were out of range, but that was slower o[0] = stbi__clamp((x0+t3) >> 17); o[7] = stbi__clamp((x0-t3) >> 17); o[1] = stbi__clamp((x1+t2) >> 17); @@ -1055,13 +2005,20 @@ static void stbi__idct_block(uint8_t *out, int out_stride, short data[64]) } #ifdef STBI_SSE2 -static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) +// sse2 integer IDCT. not the fastest possible implementation but it +// produces bit-identical results to the generic C version so it's +// fully "transparent". +static void stbi__idct_simd(stbi_uc *out, int out_stride, short data[64]) { + // This is constructed to match our regular (generic) integer IDCT exactly. __m128i row0, row1, row2, row3, row4, row5, row6, row7; __m128i tmp; + // dot product constant: even elems=x, odd elems=y #define dct_const(x,y) _mm_setr_epi16((x),(y),(x),(y),(x),(y),(x),(y)) + // out(0) = c0[even]*x + c0[odd]*y (c0, x, y 16-bit, out 32-bit) + // out(1) = c1[even]*x + c1[odd]*y #define dct_rot(out0,out1, x,y,c0,c1) \ __m128i c0##lo = _mm_unpacklo_epi16((x),(y)); \ __m128i c0##hi = _mm_unpackhi_epi16((x),(y)); \ @@ -1070,18 +2027,22 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) __m128i out1##_l = _mm_madd_epi16(c0##lo, c1); \ __m128i out1##_h = _mm_madd_epi16(c0##hi, c1) + // out = in << 12 (in 16-bit, out 32-bit) #define dct_widen(out, in) \ __m128i out##_l = _mm_srai_epi32(_mm_unpacklo_epi16(_mm_setzero_si128(), (in)), 4); \ __m128i out##_h = _mm_srai_epi32(_mm_unpackhi_epi16(_mm_setzero_si128(), (in)), 4) + // wide add #define dct_wadd(out, a, b) \ __m128i out##_l = _mm_add_epi32(a##_l, b##_l); \ __m128i out##_h = _mm_add_epi32(a##_h, b##_h) + // wide sub #define dct_wsub(out, a, b) \ __m128i out##_l = _mm_sub_epi32(a##_l, b##_l); \ __m128i out##_h = _mm_sub_epi32(a##_h, b##_h) + // butterfly a/b, add bias, then shift by "s" and pack #define dct_bfly32o(out0, out1, a,b,bias,s) \ { \ __m128i abiased_l = _mm_add_epi32(a##_l, bias); \ @@ -1092,11 +2053,13 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) out1 = _mm_packs_epi32(_mm_srai_epi32(dif_l, s), _mm_srai_epi32(dif_h, s)); \ } + // 8-bit interleave step (for transposes) #define dct_interleave8(a, b) \ tmp = a; \ a = _mm_unpacklo_epi8(a, b); \ b = _mm_unpackhi_epi8(tmp, b) + // 16-bit interleave step (for transposes) #define dct_interleave16(a, b) \ tmp = a; \ a = _mm_unpacklo_epi16(a, b); \ @@ -1139,9 +2102,11 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) __m128i rot3_0 = dct_const(stbi__f2f(-0.390180644f) + stbi__f2f( 2.053119869f), stbi__f2f(-0.390180644f)); __m128i rot3_1 = dct_const(stbi__f2f(-0.390180644f), stbi__f2f(-0.390180644f) + stbi__f2f( 1.501321110f)); + // rounding biases in column/row passes, see stbi__idct_block for explanation. __m128i bias_0 = _mm_set1_epi32(512); __m128i bias_1 = _mm_set1_epi32(65536 + (128<<17)); + // load row0 = _mm_load_si128((const __m128i *) (data + 0*8)); row1 = _mm_load_si128((const __m128i *) (data + 1*8)); row2 = _mm_load_si128((const __m128i *) (data + 2*8)); @@ -1151,42 +2116,52 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) row6 = _mm_load_si128((const __m128i *) (data + 6*8)); row7 = _mm_load_si128((const __m128i *) (data + 7*8)); + // column pass dct_pass(bias_0, 10); { + // 16bit 8x8 transpose pass 1 dct_interleave16(row0, row4); dct_interleave16(row1, row5); dct_interleave16(row2, row6); dct_interleave16(row3, row7); + // transpose pass 2 dct_interleave16(row0, row2); dct_interleave16(row1, row3); dct_interleave16(row4, row6); dct_interleave16(row5, row7); + // transpose pass 3 dct_interleave16(row0, row1); dct_interleave16(row2, row3); dct_interleave16(row4, row5); dct_interleave16(row6, row7); } + // row pass dct_pass(bias_1, 17); { - __m128i p0 = _mm_packus_epi16(row0, row1); + // pack + __m128i p0 = _mm_packus_epi16(row0, row1); // a0a1a2a3...a7b0b1b2b3...b7 __m128i p1 = _mm_packus_epi16(row2, row3); __m128i p2 = _mm_packus_epi16(row4, row5); __m128i p3 = _mm_packus_epi16(row6, row7); - dct_interleave8(p0, p2); - dct_interleave8(p1, p3); + // 8bit 8x8 transpose pass 1 + dct_interleave8(p0, p2); // a0e0a1e1... + dct_interleave8(p1, p3); // c0g0c1g1... - dct_interleave8(p0, p1); - dct_interleave8(p2, p3); + // transpose pass 2 + dct_interleave8(p0, p1); // a0c0e0g0... + dct_interleave8(p2, p3); // b0d0f0h0... - dct_interleave8(p0, p2); - dct_interleave8(p1, p3); + // transpose pass 3 + dct_interleave8(p0, p2); // a0b0c0d0... + dct_interleave8(p1, p3); // a4b4c4d4... + // store _mm_storel_epi64((__m128i *) out, p0); out += out_stride; _mm_storel_epi64((__m128i *) out, _mm_shuffle_epi32(p0, 0x4e)); out += out_stride; _mm_storel_epi64((__m128i *) out, p2); out += out_stride; @@ -1208,11 +2183,13 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) #undef dct_pass } -#endif +#endif // STBI_SSE2 #ifdef STBI_NEON -static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) +// NEON integer IDCT. should produce bit-identical +// results to the generic C version. +static void stbi__idct_simd(stbi_uc *out, int out_stride, short data[64]) { int16x8_t row0, row1, row2, row3, row4, row5, row6, row7; @@ -1241,14 +2218,17 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) int32x4_t out##_l = vshll_n_s16(vget_low_s16(inq), 12); \ int32x4_t out##_h = vshll_n_s16(vget_high_s16(inq), 12) +// wide add #define dct_wadd(out, a, b) \ int32x4_t out##_l = vaddq_s32(a##_l, b##_l); \ int32x4_t out##_h = vaddq_s32(a##_h, b##_h) +// wide sub #define dct_wsub(out, a, b) \ int32x4_t out##_l = vsubq_s32(a##_l, b##_l); \ int32x4_t out##_h = vsubq_s32(a##_h, b##_h) +// butterfly a/b, then shift using "shiftop" by "s" and pack #define dct_bfly32o(out0,out1, a,b,shiftop,s) \ { \ dct_wadd(sum, a, b); \ @@ -1297,6 +2277,7 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) dct_bfly32o(row3,row4, x3,x4,shiftop,shift); \ } + // load row0 = vld1q_s16(data + 0*8); row1 = vld1q_s16(data + 1*8); row2 = vld1q_s16(data + 2*8); @@ -1306,26 +2287,34 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) row6 = vld1q_s16(data + 6*8); row7 = vld1q_s16(data + 7*8); + // add DC bias row0 = vaddq_s16(row0, vsetq_lane_s16(1024, vdupq_n_s16(0), 0)); + // column pass dct_pass(vrshrn_n_s32, 10); + // 16bit 8x8 transpose { +// these three map to a single VTRN.16, VTRN.32, and VSWP, respectively. +// whether compilers actually get this is another story, sadly. #define dct_trn16(x, y) { int16x8x2_t t = vtrnq_s16(x, y); x = t.val[0]; y = t.val[1]; } #define dct_trn32(x, y) { int32x4x2_t t = vtrnq_s32(vreinterpretq_s32_s16(x), vreinterpretq_s32_s16(y)); x = vreinterpretq_s16_s32(t.val[0]); y = vreinterpretq_s16_s32(t.val[1]); } #define dct_trn64(x, y) { int16x8_t x0 = x; int16x8_t y0 = y; x = vcombine_s16(vget_low_s16(x0), vget_low_s16(y0)); y = vcombine_s16(vget_high_s16(x0), vget_high_s16(y0)); } - dct_trn16(row0, row1); + // pass 1 + dct_trn16(row0, row1); // a0b0a2b2a4b4a6b6 dct_trn16(row2, row3); dct_trn16(row4, row5); dct_trn16(row6, row7); - dct_trn32(row0, row2); + // pass 2 + dct_trn32(row0, row2); // a0b0c0d0a4b4c4d4 dct_trn32(row1, row3); dct_trn32(row4, row6); dct_trn32(row5, row7); - dct_trn64(row0, row4); + // pass 3 + dct_trn64(row0, row4); // a0b0c0d0e0f0g0h0 dct_trn64(row1, row5); dct_trn64(row2, row6); dct_trn64(row3, row7); @@ -1335,9 +2324,14 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) #undef dct_trn64 } + // row pass + // vrshrn_n_s32 only supports shifts up to 16, we need + // 17. so do a non-rounding shift of 16 first then follow + // up with a rounding shift by 1. dct_pass(vshrn_n_s32, 16); { + // pack and round uint8x8_t p0 = vqrshrun_n_s16(row0, 1); uint8x8_t p1 = vqrshrun_n_s16(row1, 1); uint8x8_t p2 = vqrshrun_n_s16(row2, 1); @@ -1347,25 +2341,33 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) uint8x8_t p6 = vqrshrun_n_s16(row6, 1); uint8x8_t p7 = vqrshrun_n_s16(row7, 1); + // again, these can translate into one instruction, but often don't. #define dct_trn8_8(x, y) { uint8x8x2_t t = vtrn_u8(x, y); x = t.val[0]; y = t.val[1]; } #define dct_trn8_16(x, y) { uint16x4x2_t t = vtrn_u16(vreinterpret_u16_u8(x), vreinterpret_u16_u8(y)); x = vreinterpret_u8_u16(t.val[0]); y = vreinterpret_u8_u16(t.val[1]); } #define dct_trn8_32(x, y) { uint32x2x2_t t = vtrn_u32(vreinterpret_u32_u8(x), vreinterpret_u32_u8(y)); x = vreinterpret_u8_u32(t.val[0]); y = vreinterpret_u8_u32(t.val[1]); } + // sadly can't use interleaved stores here since we only write + // 8 bytes to each scan line! + + // 8x8 8-bit transpose pass 1 dct_trn8_8(p0, p1); dct_trn8_8(p2, p3); dct_trn8_8(p4, p5); dct_trn8_8(p6, p7); + // pass 2 dct_trn8_16(p0, p2); dct_trn8_16(p1, p3); dct_trn8_16(p4, p6); dct_trn8_16(p5, p7); + // pass 3 dct_trn8_32(p0, p4); dct_trn8_32(p1, p5); dct_trn8_32(p2, p6); dct_trn8_32(p3, p7); + // store vst1_u8(out, p0); out += out_stride; vst1_u8(out, p1); out += out_stride; vst1_u8(out, p2); out += out_stride; @@ -1389,12 +2391,15 @@ static void stbi__idct_simd(uint8_t *out, int out_stride, short data[64]) #undef dct_pass } -#endif +#endif // STBI_NEON #define STBI__MARKER_none 0xff -static uint8_t stbi__get_marker(stbi__jpeg *j) +// if there's a pending marker from the entropy stream, return that +// otherwise, fetch from the stream and get a marker. if there's no +// marker, return 0xff, which is never a valid marker value +static stbi_uc stbi__get_marker(stbi__jpeg *j) { - uint8_t x; + stbi_uc x; if (j->marker != STBI__MARKER_none) { x = j->marker; j->marker = STBI__MARKER_none; return x; } x = stbi__get8(j->s); if (x != 0xff) return STBI__MARKER_none; @@ -1403,8 +2408,12 @@ static uint8_t stbi__get_marker(stbi__jpeg *j) return x; } +// in each scan, we'll have scan_n components, and the order +// of the components is specified by order[] #define STBI__RESTART(x) ((x) >= 0xd0 && (x) <= 0xd7) +// after a restart interval, stbi__jpeg_reset the entropy decoder and +// the dc prediction static void stbi__jpeg_reset(stbi__jpeg *j) { j->code_bits = 0; @@ -1414,18 +2423,22 @@ static void stbi__jpeg_reset(stbi__jpeg *j) j->marker = STBI__MARKER_none; j->todo = j->restart_interval ? j->restart_interval : 0x7fffffff; j->eob_run = 0; + // no more than 1<<31 MCUs if no restart_interal? that's plenty safe, + // since we don't even allow 1<<30 pixels } static int stbi__parse_entropy_coded_data(stbi__jpeg *z) { stbi__jpeg_reset(z); - if (!z->progressive) - { - if (z->scan_n == 1) - { + if (!z->progressive) { + if (z->scan_n == 1) { int i,j; STBI_SIMD_ALIGN(short, data[64]); int n = z->order[0]; + // non-interleaved data, we just need to process one block at a time, + // in trivial scanline order + // number of blocks to do just depends on how many actual "pixels" this + // component has, independent of interleaved MCU blocking and such int w = (z->img_comp[n].x+7) >> 3; int h = (z->img_comp[n].y+7) >> 3; for (j=0; j < h; ++j) { @@ -1433,26 +2446,27 @@ static int stbi__parse_entropy_coded_data(stbi__jpeg *z) int ha = z->img_comp[n].ha; if (!stbi__jpeg_decode_block(z, data, z->huff_dc+z->img_comp[n].hd, z->huff_ac+ha, z->fast_ac[ha], n, z->dequant[z->img_comp[n].tq])) return 0; z->idct_block_kernel(z->img_comp[n].data+z->img_comp[n].w2*j*8+i*8, z->img_comp[n].w2, data); + // every data block is an MCU, so countdown the restart interval if (--z->todo <= 0) { if (z->code_bits < 24) stbi__grow_buffer_unsafe(z); + // if it's NOT a restart, then just bail, so we get corrupt data + // rather than no data if (!STBI__RESTART(z->marker)) return 1; stbi__jpeg_reset(z); } } } return 1; - } - else - { + } else { // interleaved int i,j,k,x,y; STBI_SIMD_ALIGN(short, data[64]); - for (j=0; j < z->img_mcu_y; ++j) - { - for (i=0; i < z->img_mcu_x; ++i) - { - for (k=0; k < z->scan_n; ++k) - { + for (j=0; j < z->img_mcu_y; ++j) { + for (i=0; i < z->img_mcu_x; ++i) { + // scan an interleaved mcu... process scan_n components in order + for (k=0; k < z->scan_n; ++k) { int n = z->order[k]; + // scan out an mcu's worth of this component; that's just determined + // by the basic H and V specified for the component for (y=0; y < z->img_comp[n].v; ++y) { for (x=0; x < z->img_comp[n].h; ++x) { int x2 = (i*z->img_comp[n].h + x)*8; @@ -1463,7 +2477,8 @@ static int stbi__parse_entropy_coded_data(stbi__jpeg *z) } } } - + // after all interleaved components, that's an interleaved MCU, + // so now count down the restart interval if (--z->todo <= 0) { if (z->code_bits < 24) stbi__grow_buffer_unsafe(z); if (!STBI__RESTART(z->marker)) return 1; @@ -1473,13 +2488,14 @@ static int stbi__parse_entropy_coded_data(stbi__jpeg *z) } return 1; } - } - else - { - if (z->scan_n == 1) - { + } else { + if (z->scan_n == 1) { int i,j; int n = z->order[0]; + // non-interleaved data, we just need to process one block at a time, + // in trivial scanline order + // number of blocks to do just depends on how many actual "pixels" this + // component has, independent of interleaved MCU blocking and such int w = (z->img_comp[n].x+7) >> 3; int h = (z->img_comp[n].y+7) >> 3; for (j=0; j < h; ++j) { @@ -1493,6 +2509,7 @@ static int stbi__parse_entropy_coded_data(stbi__jpeg *z) if (!stbi__jpeg_decode_block_prog_ac(z, data, &z->huff_ac[ha], z->fast_ac[ha])) return 0; } + // every data block is an MCU, so countdown the restart interval if (--z->todo <= 0) { if (z->code_bits < 24) stbi__grow_buffer_unsafe(z); if (!STBI__RESTART(z->marker)) return 1; @@ -1501,14 +2518,15 @@ static int stbi__parse_entropy_coded_data(stbi__jpeg *z) } } return 1; - } - else - { + } else { // interleaved int i,j,k,x,y; for (j=0; j < z->img_mcu_y; ++j) { for (i=0; i < z->img_mcu_x; ++i) { + // scan an interleaved mcu... process scan_n components in order for (k=0; k < z->scan_n; ++k) { int n = z->order[k]; + // scan out an mcu's worth of this component; that's just determined + // by the basic H and V specified for the component for (y=0; y < z->img_comp[n].v; ++y) { for (x=0; x < z->img_comp[n].h; ++x) { int x2 = (i*z->img_comp[n].h + x); @@ -1519,6 +2537,8 @@ static int stbi__parse_entropy_coded_data(stbi__jpeg *z) } } } + // after all interleaved components, that's an interleaved MCU, + // so now count down the restart interval if (--z->todo <= 0) { if (z->code_bits < 24) stbi__grow_buffer_unsafe(z); if (!STBI__RESTART(z->marker)) return 1; @@ -1531,7 +2551,7 @@ static int stbi__parse_entropy_coded_data(stbi__jpeg *z) } } -static void stbi__jpeg_dequantize(short *data, uint8_t *dequant) +static void stbi__jpeg_dequantize(short *data, stbi_uc *dequant) { int i; for (i=0; i < 64; ++i) @@ -1541,6 +2561,7 @@ static void stbi__jpeg_dequantize(short *data, uint8_t *dequant) static void stbi__jpeg_finish(stbi__jpeg *z) { if (z->progressive) { + // dequantize and idct the data int i,j,n; for (n=0; n < z->s->img_n; ++n) { int w = (z->img_comp[n].x+7) >> 3; @@ -1560,15 +2581,15 @@ static int stbi__process_marker(stbi__jpeg *z, int m) { int L; switch (m) { - case STBI__MARKER_none: + case STBI__MARKER_none: // no marker found return stbi__err("expected marker","Corrupt JPEG"); - case 0xDD: + case 0xDD: // DRI - specify restart interval if (stbi__get16be(z->s) != 4) return stbi__err("bad DRI len","Corrupt JPEG"); z->restart_interval = stbi__get16be(z->s); return 1; - case 0xDB: + case 0xDB: // DQT - define quantization table L = stbi__get16be(z->s)-2; while (L > 0) { int q = stbi__get8(z->s); @@ -1582,10 +2603,10 @@ static int stbi__process_marker(stbi__jpeg *z, int m) } return L==0; - case 0xC4: + case 0xC4: // DHT - define huffman table L = stbi__get16be(z->s)-2; while (L > 0) { - uint8_t *v; + stbi_uc *v; int sizes[16],i,n=0; int q = stbi__get8(z->s); int tc = q >> 4; @@ -1611,6 +2632,7 @@ static int stbi__process_marker(stbi__jpeg *z, int m) } return L==0; } + // check for comment block or APP blocks if ((m >= 0xE0 && m <= 0xEF) || m == 0xFE) { stbi__skip(z->s, stbi__get16be(z->s)-2); return 1; @@ -1618,6 +2640,7 @@ static int stbi__process_marker(stbi__jpeg *z, int m) return 0; } +// after we see SOS static int stbi__process_scan_header(stbi__jpeg *z) { int i; @@ -1631,7 +2654,7 @@ static int stbi__process_scan_header(stbi__jpeg *z) for (which = 0; which < z->s->img_n; ++which) if (z->img_comp[which].id == id) break; - if (which == z->s->img_n) return 0; + if (which == z->s->img_n) return 0; // no match z->img_comp[which].hd = q >> 4; if (z->img_comp[which].hd > 3) return stbi__err("bad DC huff","Corrupt JPEG"); z->img_comp[which].ha = q & 15; if (z->img_comp[which].ha > 3) return stbi__err("bad AC huff","Corrupt JPEG"); z->order[i] = which; @@ -1640,7 +2663,7 @@ static int stbi__process_scan_header(stbi__jpeg *z) { int aa; z->spec_start = stbi__get8(z->s); - z->spec_end = stbi__get8(z->s); + z->spec_end = stbi__get8(z->s); // should be 63, but might be 0 aa = stbi__get8(z->s); z->succ_high = (aa >> 4); z->succ_low = (aa & 15); @@ -1661,12 +2684,12 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan) { stbi__context *s = z->s; int Lf,p,i,q, h_max=1,v_max=1,c; - Lf = stbi__get16be(s); if (Lf < 11) return stbi__err("bad SOF len","Corrupt JPEG"); - p = stbi__get8(s); if (p != 8) return stbi__err("only 8-bit","JPEG format not supported: 8-bit only"); - s->img_y = stbi__get16be(s); if (s->img_y == 0) return stbi__err("no header height", "JPEG format not supported: delayed height"); - s->img_x = stbi__get16be(s); if (s->img_x == 0) return stbi__err("0 width","Corrupt JPEG"); + Lf = stbi__get16be(s); if (Lf < 11) return stbi__err("bad SOF len","Corrupt JPEG"); // JPEG + 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 c = stbi__get8(s); - if (c != 3 && c != 1) return stbi__err("bad component count","Corrupt JPEG"); + if (c != 3 && c != 1) return stbi__err("bad component count","Corrupt JPEG"); // JFIF requires s->img_n = c; for (i=0; i < c; ++i) { z->img_comp[i].data = NULL; @@ -1677,8 +2700,8 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan) for (i=0; i < s->img_n; ++i) { z->img_comp[i].id = stbi__get8(s); - if (z->img_comp[i].id != i+1) - if (z->img_comp[i].id != i) + if (z->img_comp[i].id != i+1) // JFIF requires + if (z->img_comp[i].id != i) // some version of jpegtran outputs non-JFIF-compliant files! return stbi__err("bad component ID","Corrupt JPEG"); q = stbi__get8(s); z->img_comp[i].h = (q >> 4); if (!z->img_comp[i].h || z->img_comp[i].h > 4) return stbi__err("bad H","Corrupt JPEG"); @@ -1695,6 +2718,7 @@ 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; } + // compute interleaved mcu info z->img_h_max = h_max; z->img_v_max = v_max; z->img_mcu_w = h_max * 8; @@ -1703,25 +2727,31 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan) z->img_mcu_y = (s->img_y + z->img_mcu_h-1) / z->img_mcu_h; for (i=0; i < s->img_n; ++i) { + // number of effective pixels (e.g. for non-interleaved MCU) z->img_comp[i].x = (s->img_x * z->img_comp[i].h + h_max-1) / h_max; z->img_comp[i].y = (s->img_y * z->img_comp[i].v + v_max-1) / v_max; + // to simplify generation, we'll allocate enough memory to decode + // the bogus oversized data from using interleaved MCUs and their + // big blocks (e.g. a 16x16 iMCU on an image of width 33); we won't + // discard the extra data until colorspace conversion z->img_comp[i].w2 = z->img_mcu_x * z->img_comp[i].h * 8; z->img_comp[i].h2 = z->img_mcu_y * z->img_comp[i].v * 8; - z->img_comp[i].raw_data = malloc(z->img_comp[i].w2 * z->img_comp[i].h2+15); + z->img_comp[i].raw_data = stbi__malloc(z->img_comp[i].w2 * z->img_comp[i].h2+15); if (z->img_comp[i].raw_data == NULL) { for(--i; i >= 0; --i) { - free(z->img_comp[i].raw_data); + STBI_FREE(z->img_comp[i].raw_data); z->img_comp[i].data = NULL; } return stbi__err("outofmem", "Out of memory"); } - z->img_comp[i].data = (uint8_t*) (((size_t) z->img_comp[i].raw_data + 15) & ~15); + // align blocks for idct using mmx/sse + z->img_comp[i].data = (stbi_uc*) (((size_t) z->img_comp[i].raw_data + 15) & ~15); z->img_comp[i].linebuf = NULL; if (z->progressive) { z->img_comp[i].coeff_w = (z->img_comp[i].w2 + 7) >> 3; z->img_comp[i].coeff_h = (z->img_comp[i].h2 + 7) >> 3; - z->img_comp[i].raw_coeff = malloc(z->img_comp[i].coeff_w * z->img_comp[i].coeff_h * 64 * sizeof(short) + 15); + z->img_comp[i].raw_coeff = STBI_MALLOC(z->img_comp[i].coeff_w * z->img_comp[i].coeff_h * 64 * sizeof(short) + 15); z->img_comp[i].coeff = (short*) (((size_t) z->img_comp[i].raw_coeff + 15) & ~15); } else { z->img_comp[i].coeff = 0; @@ -1732,6 +2762,7 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan) return 1; } +// use comparisons since in some cases we handle more than one case (e.g. SOF) #define stbi__DNL(x) ((x) == 0xdc) #define stbi__SOI(x) ((x) == 0xd8) #define stbi__EOI(x) ((x) == 0xd9) @@ -1743,17 +2774,16 @@ static int stbi__process_frame_header(stbi__jpeg *z, int scan) static int stbi__decode_jpeg_header(stbi__jpeg *z, int scan) { int m; - z->marker = STBI__MARKER_none; + z->marker = STBI__MARKER_none; // initialize cached marker to empty m = stbi__get_marker(z); if (!stbi__SOI(m)) return stbi__err("no SOI","Corrupt JPEG"); if (scan == STBI__SCAN_type) return 1; m = stbi__get_marker(z); - while (!stbi__SOF(m)) - { + while (!stbi__SOF(m)) { if (!stbi__process_marker(z,m)) return 0; m = stbi__get_marker(z); - while (m == STBI__MARKER_none) - { + while (m == STBI__MARKER_none) { + // some files have extra padding after their blocks, so ok, we'll scan if (stbi__at_eof(z->s)) return stbi__err("no SOF", "Corrupt JPEG"); m = stbi__get_marker(z); } @@ -1763,6 +2793,7 @@ static int stbi__decode_jpeg_header(stbi__jpeg *z, int scan) return 1; } +// decode image to YCbCr format static int stbi__decode_jpeg_image(stbi__jpeg *j) { int m; @@ -1778,6 +2809,7 @@ static int stbi__decode_jpeg_image(stbi__jpeg *j) if (!stbi__process_scan_header(j)) return 0; if (!stbi__parse_entropy_coded_data(j)) return 0; if (j->marker == STBI__MARKER_none ) { + // handle 0s at the end of image data from IP Kamera 9060 while (!stbi__at_eof(j->s)) { int x = stbi__get8(j->s); if (x == 255) { @@ -1787,6 +2819,7 @@ static int stbi__decode_jpeg_image(stbi__jpeg *j) return stbi__err("junk before marker", "Corrupt JPEG"); } } + // if we reach eof without hitting a marker, stbi__get_marker() below will fail and we'll eventually return 0 } } else { if (!stbi__process_marker(j, m)) return 0; @@ -1798,13 +2831,14 @@ static int stbi__decode_jpeg_image(stbi__jpeg *j) return 1; } +// static jfif-centered resampling (across block boundaries) -typedef uint8_t *(*resample_row_func)(uint8_t *out, uint8_t *in0, uint8_t *in1, +typedef stbi_uc *(*resample_row_func)(stbi_uc *out, stbi_uc *in0, stbi_uc *in1, int w, int hs); -#define stbi__div4(x) ((uint8_t) ((x) >> 2)) +#define stbi__div4(x) ((stbi_uc) ((x) >> 2)) -static uint8_t *resample_row_1(uint8_t *out, uint8_t *in_near, uint8_t *in_far, int w, int hs) +static stbi_uc *resample_row_1(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) { STBI_NOTUSED(out); STBI_NOTUSED(in_far); @@ -1813,8 +2847,9 @@ static uint8_t *resample_row_1(uint8_t *out, uint8_t *in_near, uint8_t *in_far, return in_near; } -static uint8_t* stbi__resample_row_v_2(uint8_t *out, uint8_t *in_near, uint8_t *in_far, int w, int hs) +static stbi_uc* stbi__resample_row_v_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) { + // need to generate two samples vertically for every one in input int i; STBI_NOTUSED(hs); for (i=0; i < w; ++i) @@ -1822,12 +2857,14 @@ static uint8_t* stbi__resample_row_v_2(uint8_t *out, uint8_t *in_near, uint8_t * return out; } -static uint8_t* stbi__resample_row_h_2(uint8_t *out, uint8_t *in_near, uint8_t *in_far, int w, int hs) +static stbi_uc* stbi__resample_row_h_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) { + // need to generate two samples horizontally for every one in input int i; - uint8_t *input = in_near; + stbi_uc *input = in_near; if (w == 1) { + // if only one sample, can't do any interpolation out[0] = out[1] = input[0]; return out; } @@ -1848,10 +2885,11 @@ static uint8_t* stbi__resample_row_h_2(uint8_t *out, uint8_t *in_near, uint8_t return out; } -#define stbi__div16(x) ((uint8_t) ((x) >> 4)) +#define stbi__div16(x) ((stbi_uc) ((x) >> 4)) -static uint8_t *stbi__resample_row_hv_2(uint8_t *out, uint8_t *in_near, uint8_t *in_far, int w, int hs) +static stbi_uc *stbi__resample_row_hv_2(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) { + // need to generate 2x2 samples for every one in input int i,t0,t1; if (w == 1) { out[0] = out[1] = stbi__div4(3*in_near[0] + in_far[0] + 2); @@ -1874,8 +2912,9 @@ static uint8_t *stbi__resample_row_hv_2(uint8_t *out, uint8_t *in_near, uint8_t } #if defined(STBI_SSE2) || defined(STBI_NEON) -static uint8_t *stbi__resample_row_hv_2_simd(uint8_t *out, uint8_t *in_near, uint8_t *in_far, int w, int hs) +static stbi_uc *stbi__resample_row_hv_2_simd(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) { + // need to generate 2x2 samples for every one in input int i=0,t0,t1; if (w == 1) { @@ -1884,9 +2923,13 @@ static uint8_t *stbi__resample_row_hv_2_simd(uint8_t *out, uint8_t *in_near, uin } t1 = 3*in_near[0] + in_far[0]; - for (; i < ((w-1) & ~7); i += 8) - { + // process groups of 8 pixels for as long as we can. + // note we can't handle the last pixel in a row in this loop + // because we need to handle the filter boundary conditions. + for (; i < ((w-1) & ~7); i += 8) { #if defined(STBI_SSE2) + // load and perform the vertical filtering pass + // this uses 3*x + y = 4*x + (y - x) __m128i zero = _mm_setzero_si128(); __m128i farb = _mm_loadl_epi64((__m128i *) (in_far + i)); __m128i nearb = _mm_loadl_epi64((__m128i *) (in_near + i)); @@ -1894,11 +2937,22 @@ static uint8_t *stbi__resample_row_hv_2_simd(uint8_t *out, uint8_t *in_near, uin __m128i nearw = _mm_unpacklo_epi8(nearb, zero); __m128i diff = _mm_sub_epi16(farw, nearw); __m128i nears = _mm_slli_epi16(nearw, 2); - __m128i curr = _mm_add_epi16(nears, diff); + __m128i curr = _mm_add_epi16(nears, diff); // current row + + // horizontal filter works the same based on shifted vers of current + // row. "prev" is current row shifted right by 1 pixel; we need to + // insert the previous pixel value (from t1). + // "next" is current row shifted left by 1 pixel, with first pixel + // of next block of 8 pixels added in. __m128i prv0 = _mm_slli_si128(curr, 2); __m128i nxt0 = _mm_srli_si128(curr, 2); __m128i prev = _mm_insert_epi16(prv0, t1, 0); __m128i next = _mm_insert_epi16(nxt0, 3*in_near[i+8] + in_far[i+8], 7); + + // horizontal filter, polyphase implementation since it's convenient: + // even pixels = 3*cur + prev = cur*4 + (prev - cur) + // odd pixels = 3*cur + next = cur*4 + (next - cur) + // note the shared term. __m128i bias = _mm_set1_epi16(8); __m128i curs = _mm_slli_epi16(curr, 2); __m128i prvd = _mm_sub_epi16(prev, curr); @@ -1906,33 +2960,53 @@ static uint8_t *stbi__resample_row_hv_2_simd(uint8_t *out, uint8_t *in_near, uin __m128i curb = _mm_add_epi16(curs, bias); __m128i even = _mm_add_epi16(prvd, curb); __m128i odd = _mm_add_epi16(nxtd, curb); + + // interleave even and odd pixels, then undo scaling. __m128i int0 = _mm_unpacklo_epi16(even, odd); __m128i int1 = _mm_unpackhi_epi16(even, odd); __m128i de0 = _mm_srli_epi16(int0, 4); __m128i de1 = _mm_srli_epi16(int1, 4); + + // pack and write output __m128i outv = _mm_packus_epi16(de0, de1); _mm_storeu_si128((__m128i *) (out + i*2), outv); #elif defined(STBI_NEON) - uint8x8x2_t o; + // load and perform the vertical filtering pass + // this uses 3*x + y = 4*x + (y - x) uint8x8_t farb = vld1_u8(in_far + i); uint8x8_t nearb = vld1_u8(in_near + i); int16x8_t diff = vreinterpretq_s16_u16(vsubl_u8(farb, nearb)); int16x8_t nears = vreinterpretq_s16_u16(vshll_n_u8(nearb, 2)); - int16x8_t curr = vaddq_s16(nears, diff); + int16x8_t curr = vaddq_s16(nears, diff); // current row + + // horizontal filter works the same based on shifted vers of current + // row. "prev" is current row shifted right by 1 pixel; we need to + // insert the previous pixel value (from t1). + // "next" is current row shifted left by 1 pixel, with first pixel + // of next block of 8 pixels added in. int16x8_t prv0 = vextq_s16(curr, curr, 7); int16x8_t nxt0 = vextq_s16(curr, curr, 1); int16x8_t prev = vsetq_lane_s16(t1, prv0, 0); int16x8_t next = vsetq_lane_s16(3*in_near[i+8] + in_far[i+8], nxt0, 7); + + // horizontal filter, polyphase implementation since it's convenient: + // even pixels = 3*cur + prev = cur*4 + (prev - cur) + // odd pixels = 3*cur + next = cur*4 + (next - cur) + // note the shared term. int16x8_t curs = vshlq_n_s16(curr, 2); int16x8_t prvd = vsubq_s16(prev, curr); int16x8_t nxtd = vsubq_s16(next, curr); int16x8_t even = vaddq_s16(curs, prvd); int16x8_t odd = vaddq_s16(curs, nxtd); + + // undo scaling and round, then store with even/odd phases interleaved + uint8x8x2_t o; o.val[0] = vqrshrun_n_s16(even, 4); o.val[1] = vqrshrun_n_s16(odd, 4); vst2_u8(out + i*2, o); #endif + // "previous" value for next iter t1 = 3*in_near[i+7] + in_far[i+7]; } @@ -1954,8 +3028,9 @@ static uint8_t *stbi__resample_row_hv_2_simd(uint8_t *out, uint8_t *in_near, uin } #endif -static uint8_t *stbi__resample_row_generic(uint8_t *out, uint8_t *in_near, uint8_t *in_far, int w, int hs) +static stbi_uc *stbi__resample_row_generic(stbi_uc *out, stbi_uc *in_near, stbi_uc *in_far, int w, int hs) { + // resample with nearest-neighbor int i,j; STBI_NOTUSED(in_far); for (i=0; i < w; ++i) @@ -1965,52 +3040,57 @@ static uint8_t *stbi__resample_row_generic(uint8_t *out, uint8_t *in_near, uint8 } #ifdef STBI_JPEG_OLD +// this is the same YCbCr-to-RGB calculation that stb_image has used +// historically before the algorithm changes in 1.49 #define float2fixed(x) ((int) ((x) * 65536 + 0.5)) -static void stbi__YCbCr_to_RGB_row(uint8_t *out, const uint8_t *y, const uint8_t *pcb, const uint8_t *pcr, int count, int step) +static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step) { int i; - for (i=0; i < count; ++i) - { - int y_fixed = (y[i] << 16) + 32768; + for (i=0; i < count; ++i) { + int y_fixed = (y[i] << 16) + 32768; // rounding + int r,g,b; int cr = pcr[i] - 128; int cb = pcb[i] - 128; - int r = y_fixed + cr*float2fixed(1.40200f); - int g = y_fixed - cr*float2fixed(0.71414f) - cb*float2fixed(0.34414f); - int b = y_fixed + cb*float2fixed(1.77200f); + r = y_fixed + cr*float2fixed(1.40200f); + g = y_fixed - cr*float2fixed(0.71414f) - cb*float2fixed(0.34414f); + b = y_fixed + cb*float2fixed(1.77200f); r >>= 16; g >>= 16; b >>= 16; if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; } if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; } if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; } - out[0] = (uint8_t)r; - out[1] = (uint8_t)g; - out[2] = (uint8_t)b; + out[0] = (stbi_uc)r; + out[1] = (stbi_uc)g; + out[2] = (stbi_uc)b; out[3] = 255; out += step; } } #else +// this is a reduced-precision calculation of YCbCr-to-RGB introduced +// to make sure the code produces the same results in both SIMD and scalar #define float2fixed(x) (((int) ((x) * 4096.0f + 0.5f)) << 8) -static void stbi__YCbCr_to_RGB_row(uint8_t *out, const uint8_t *y, const uint8_t *pcb, const uint8_t *pcr, int count, int step) +static void stbi__YCbCr_to_RGB_row(stbi_uc *out, const stbi_uc *y, const stbi_uc *pcb, const stbi_uc *pcr, int count, int step) { int i; for (i=0; i < count; ++i) { - int y_fixed = (y[i] << 20) + (1<<19); + int y_fixed = (y[i] << 20) + (1<<19); // rounding + int r,g,b; int cr = pcr[i] - 128; int cb = pcb[i] - 128; - int r = y_fixed + cr* float2fixed(1.40200f); - int g = y_fixed + (cr*-float2fixed(0.71414f)) + ((cb*-float2fixed(0.34414f)) & 0xffff0000); - int b = y_fixed + cb* float2fixed(1.77200f); + r = y_fixed + cr* float2fixed(1.40200f); + g = y_fixed + (cr*-float2fixed(0.71414f)) + ((cb*-float2fixed(0.34414f)) & 0xffff0000); + b = y_fixed + cb* float2fixed(1.77200f); r >>= 20; g >>= 20; b >>= 20; if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; } if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; } if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; } - out[0] = (uint8_t)r; - out[1] = (uint8_t)g; - out[2] = (uint8_t)b; + out[0] = (stbi_uc)r; + out[1] = (stbi_uc)g; + out[2] = (stbi_uc)b; out[3] = 255; out += step; } @@ -2018,32 +3098,38 @@ static void stbi__YCbCr_to_RGB_row(uint8_t *out, const uint8_t *y, const uint8_t #endif #if defined(STBI_SSE2) || defined(STBI_NEON) -static void stbi__YCbCr_to_RGB_simd(uint8_t *out, uint8_t const *y, uint8_t const *pcb, uint8_t const *pcr, int count, int step) +static void stbi__YCbCr_to_RGB_simd(stbi_uc *out, stbi_uc const *y, stbi_uc const *pcb, stbi_uc const *pcr, int count, int step) { int i = 0; #ifdef STBI_SSE2 - if (step == 4) - { + // step == 3 is pretty ugly on the final interleave, and i'm not convinced + // it's useful in practice (you wouldn't use it for textures, for example). + // so just accelerate step == 4 case. + if (step == 4) { + // this is a fairly straightforward implementation and not super-optimized. __m128i signflip = _mm_set1_epi8(-0x80); __m128i cr_const0 = _mm_set1_epi16( (short) ( 1.40200f*4096.0f+0.5f)); __m128i cr_const1 = _mm_set1_epi16( - (short) ( 0.71414f*4096.0f+0.5f)); __m128i cb_const0 = _mm_set1_epi16( - (short) ( 0.34414f*4096.0f+0.5f)); __m128i cb_const1 = _mm_set1_epi16( (short) ( 1.77200f*4096.0f+0.5f)); __m128i y_bias = _mm_set1_epi8((char) (unsigned char) 128); - __m128i xw = _mm_set1_epi16(255); + __m128i xw = _mm_set1_epi16(255); // alpha channel for (; i+7 < count; i += 8) { + // load __m128i y_bytes = _mm_loadl_epi64((__m128i *) (y+i)); __m128i cr_bytes = _mm_loadl_epi64((__m128i *) (pcr+i)); __m128i cb_bytes = _mm_loadl_epi64((__m128i *) (pcb+i)); - __m128i cr_biased = _mm_xor_si128(cr_bytes, signflip); - __m128i cb_biased = _mm_xor_si128(cb_bytes, signflip); + __m128i cr_biased = _mm_xor_si128(cr_bytes, signflip); // -128 + __m128i cb_biased = _mm_xor_si128(cb_bytes, signflip); // -128 + // unpack to short (and left-shift cr, cb by 8) __m128i yw = _mm_unpacklo_epi8(y_bias, y_bytes); __m128i crw = _mm_unpacklo_epi8(_mm_setzero_si128(), cr_biased); __m128i cbw = _mm_unpacklo_epi8(_mm_setzero_si128(), cb_biased); + // color transform __m128i yws = _mm_srli_epi16(yw, 4); __m128i cr0 = _mm_mulhi_epi16(cr_const0, crw); __m128i cb0 = _mm_mulhi_epi16(cb_const0, cbw); @@ -2054,18 +3140,22 @@ static void stbi__YCbCr_to_RGB_simd(uint8_t *out, uint8_t const *y, uint8_t cons __m128i bws = _mm_add_epi16(yws, cb1); __m128i gws = _mm_add_epi16(gwt, cr1); + // descale __m128i rw = _mm_srai_epi16(rws, 4); __m128i bw = _mm_srai_epi16(bws, 4); __m128i gw = _mm_srai_epi16(gws, 4); + // back to byte, set up for transpose __m128i brb = _mm_packus_epi16(rw, bw); __m128i gxb = _mm_packus_epi16(gw, xw); + // transpose to interleave channels __m128i t0 = _mm_unpacklo_epi8(brb, gxb); __m128i t1 = _mm_unpackhi_epi8(brb, gxb); __m128i o0 = _mm_unpacklo_epi16(t0, t1); __m128i o1 = _mm_unpackhi_epi16(t0, t1); + // store _mm_storeu_si128((__m128i *) (out + 0), o0); _mm_storeu_si128((__m128i *) (out + 16), o1); out += 32; @@ -2074,27 +3164,29 @@ static void stbi__YCbCr_to_RGB_simd(uint8_t *out, uint8_t const *y, uint8_t cons #endif #ifdef STBI_NEON - if (step == 4) - { + // in this version, step=3 support would be easy to add. but is there demand? + if (step == 4) { + // this is a fairly straightforward implementation and not super-optimized. uint8x8_t signflip = vdup_n_u8(0x80); int16x8_t cr_const0 = vdupq_n_s16( (short) ( 1.40200f*4096.0f+0.5f)); int16x8_t cr_const1 = vdupq_n_s16( - (short) ( 0.71414f*4096.0f+0.5f)); int16x8_t cb_const0 = vdupq_n_s16( - (short) ( 0.34414f*4096.0f+0.5f)); int16x8_t cb_const1 = vdupq_n_s16( (short) ( 1.77200f*4096.0f+0.5f)); - for (; i+7 < count; i += 8) - { - uint8x8x4_t o; + for (; i+7 < count; i += 8) { + // load uint8x8_t y_bytes = vld1_u8(y + i); uint8x8_t cr_bytes = vld1_u8(pcr + i); uint8x8_t cb_bytes = vld1_u8(pcb + i); int8x8_t cr_biased = vreinterpret_s8_u8(vsub_u8(cr_bytes, signflip)); int8x8_t cb_biased = vreinterpret_s8_u8(vsub_u8(cb_bytes, signflip)); + // expand to s16 int16x8_t yws = vreinterpretq_s16_u16(vshll_n_u8(y_bytes, 4)); int16x8_t crw = vshll_n_s8(cr_biased, 7); int16x8_t cbw = vshll_n_s8(cb_biased, 7); + // color transform int16x8_t cr0 = vqdmulhq_s16(crw, cr_const0); int16x8_t cb0 = vqdmulhq_s16(cbw, cb_const0); int16x8_t cr1 = vqdmulhq_s16(crw, cr_const1); @@ -2103,40 +3195,44 @@ static void stbi__YCbCr_to_RGB_simd(uint8_t *out, uint8_t const *y, uint8_t cons int16x8_t gws = vaddq_s16(vaddq_s16(yws, cb0), cr1); int16x8_t bws = vaddq_s16(yws, cb1); + // undo scaling, round, convert to byte + uint8x8x4_t o; o.val[0] = vqrshrun_n_s16(rws, 4); o.val[1] = vqrshrun_n_s16(gws, 4); o.val[2] = vqrshrun_n_s16(bws, 4); o.val[3] = vdup_n_u8(255); + // store, interleaving r/g/b/a vst4_u8(out, o); out += 8*4; } } #endif - for (; i < count; ++i) - { - int y_fixed = (y[i] << 20) + (1<<19); + for (; i < count; ++i) { + int y_fixed = (y[i] << 20) + (1<<19); // rounding + int r,g,b; int cr = pcr[i] - 128; int cb = pcb[i] - 128; - int r = y_fixed + cr* float2fixed(1.40200f); - int g = y_fixed + cr*-float2fixed(0.71414f) + ((cb*-float2fixed(0.34414f)) & 0xffff0000); - int b = y_fixed + cb* float2fixed(1.77200f); + r = y_fixed + cr* float2fixed(1.40200f); + g = y_fixed + cr*-float2fixed(0.71414f) + ((cb*-float2fixed(0.34414f)) & 0xffff0000); + b = y_fixed + cb* float2fixed(1.77200f); r >>= 20; g >>= 20; b >>= 20; if ((unsigned) r > 255) { if (r < 0) r = 0; else r = 255; } if ((unsigned) g > 255) { if (g < 0) g = 0; else g = 255; } if ((unsigned) b > 255) { if (b < 0) b = 0; else b = 255; } - out[0] = (uint8_t)r; - out[1] = (uint8_t)g; - out[2] = (uint8_t)b; + out[0] = (stbi_uc)r; + out[1] = (stbi_uc)g; + out[2] = (stbi_uc)b; out[3] = 255; out += step; } } #endif +// set up the kernels static void stbi__setup_jpeg(stbi__jpeg *j) { j->idct_block_kernel = stbi__idct_block; @@ -2144,45 +3240,41 @@ static void stbi__setup_jpeg(stbi__jpeg *j) j->resample_row_hv_2_kernel = stbi__resample_row_hv_2; #ifdef STBI_SSE2 - if (stbi__sse2_available()) - { + if (stbi__sse2_available()) { j->idct_block_kernel = stbi__idct_simd; -#ifndef STBI_JPEG_OLD + #ifndef STBI_JPEG_OLD j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd; -#endif + #endif j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd; } #endif #ifdef STBI_NEON j->idct_block_kernel = stbi__idct_simd; -#ifndef STBI_JPEG_OLD + #ifndef STBI_JPEG_OLD j->YCbCr_to_RGB_kernel = stbi__YCbCr_to_RGB_simd; -#endif + #endif j->resample_row_hv_2_kernel = stbi__resample_row_hv_2_simd; #endif } +// clean up the temporary component buffers static void stbi__cleanup_jpeg(stbi__jpeg *j) { int i; - for (i=0; i < j->s->img_n; ++i) - { - if (j->img_comp[i].raw_data) - { - free(j->img_comp[i].raw_data); + for (i=0; i < j->s->img_n; ++i) { + if (j->img_comp[i].raw_data) { + STBI_FREE(j->img_comp[i].raw_data); j->img_comp[i].raw_data = NULL; j->img_comp[i].data = NULL; } - if (j->img_comp[i].raw_coeff) - { - free(j->img_comp[i].raw_coeff); + if (j->img_comp[i].raw_coeff) { + STBI_FREE(j->img_comp[i].raw_coeff); j->img_comp[i].raw_coeff = 0; j->img_comp[i].coeff = 0; } - if (j->img_comp[i].linebuf) - { - free(j->img_comp[i].linebuf); + if (j->img_comp[i].linebuf) { + STBI_FREE(j->img_comp[i].linebuf); j->img_comp[i].linebuf = NULL; } } @@ -2191,23 +3283,25 @@ static void stbi__cleanup_jpeg(stbi__jpeg *j) typedef struct { resample_row_func resample; - uint8_t *line0,*line1; - int hs,vs; - int w_lores; - int ystep; - int ypos; + stbi_uc *line0,*line1; + int hs,vs; // expansion factor in each axis + int w_lores; // horizontal pixels pre-expansion + int ystep; // how far through vertical expansion we are + int ypos; // which pre-expansion row we're on } stbi__resample; -static uint8_t *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp, int req_comp) +static stbi_uc *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp, int req_comp) { int n, decode_n; - z->s->img_n = 0; + z->s->img_n = 0; // make stbi__cleanup_jpeg safe - if (req_comp < 0 || req_comp > 4) - return stbi__errpuc("bad req_comp", "Internal error"); + // validate req_comp + if (req_comp < 0 || req_comp > 4) return stbi__errpuc("bad req_comp", "Internal error"); + // load a jpeg image from whichever source, but leave in YCbCr format if (!stbi__decode_jpeg_image(z)) { stbi__cleanup_jpeg(z); return NULL; } + // determine actual number of components to generate n = req_comp ? req_comp : z->s->img_n; if (z->s->img_n == 3 && n < 3) @@ -2215,18 +3309,21 @@ static uint8_t *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp else decode_n = z->s->img_n; + // resample and color-convert { int k; unsigned int i,j; - uint8_t *output; - uint8_t *coutput[4]; + stbi_uc *output; + stbi_uc *coutput[4]; stbi__resample res_comp[4]; for (k=0; k < decode_n; ++k) { stbi__resample *r = &res_comp[k]; - z->img_comp[k].linebuf = (uint8_t *) malloc(z->s->img_x + 3); + // allocate line buffer big enough for upsampling off the edges + // with upsample factor of 4 + z->img_comp[k].linebuf = (stbi_uc *) stbi__malloc(z->s->img_x + 3); if (!z->img_comp[k].linebuf) { stbi__cleanup_jpeg(z); return stbi__errpuc("outofmem", "Out of memory"); } r->hs = z->img_h_max / z->img_comp[k].h; @@ -2243,18 +3340,20 @@ static uint8_t *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp else r->resample = stbi__resample_row_generic; } - output = (uint8_t *) malloc(n * z->s->img_x * z->s->img_y + 1); + // can't error after this so, this is safe + output = (stbi_uc *) stbi__malloc(n * z->s->img_x * z->s->img_y + 1); if (!output) { stbi__cleanup_jpeg(z); return stbi__errpuc("outofmem", "Out of memory"); } + // now go ahead and resample for (j=0; j < z->s->img_y; ++j) { - uint8_t *out = output + n * z->s->img_x * j; + stbi_uc *out = output + n * z->s->img_x * j; for (k=0; k < decode_n; ++k) { stbi__resample *r = &res_comp[k]; int y_bot = r->ystep >= (r->vs >> 1); coutput[k] = r->resample(z->img_comp[k].linebuf, - y_bot ? r->line1 : r->line0, - y_bot ? r->line0 : r->line1, - r->w_lores, r->hs); + y_bot ? r->line1 : r->line0, + y_bot ? r->line0 : r->line1, + r->w_lores, r->hs); if (++r->ystep >= r->vs) { r->ystep = 0; r->line0 = r->line1; @@ -2263,17 +3362,17 @@ static uint8_t *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp } } if (n >= 3) { - uint8_t *y = coutput[0]; + stbi_uc *y = coutput[0]; if (z->s->img_n == 3) { z->YCbCr_to_RGB_kernel(out, y, coutput[1], coutput[2], z->s->img_x, n); } else for (i=0; i < z->s->img_x; ++i) { out[0] = out[1] = out[2] = y[i]; - out[3] = 255; + out[3] = 255; // not used if n==3 out += n; } } else { - uint8_t *y = coutput[0]; + stbi_uc *y = coutput[0]; if (n == 1) for (i=0; i < z->s->img_x; ++i) out[i] = y[i]; else @@ -2283,7 +3382,7 @@ static uint8_t *load_jpeg_image(stbi__jpeg *z, int *out_x, int *out_y, int *comp stbi__cleanup_jpeg(z); *out_x = z->s->img_x; *out_y = z->s->img_y; - if (comp) *comp = z->s->img_n; + if (comp) *comp = z->s->img_n; // report original components, not output return output; } } @@ -2306,24 +3405,53 @@ static int stbi__jpeg_test(stbi__context *s) stbi__rewind(s); return r; } + +static int stbi__jpeg_info_raw(stbi__jpeg *j, int *x, int *y, int *comp) +{ + if (!stbi__decode_jpeg_header(j, STBI__SCAN_header)) { + stbi__rewind( j->s ); + return 0; + } + if (x) *x = j->s->img_x; + if (y) *y = j->s->img_y; + if (comp) *comp = j->s->img_n; + return 1; +} + +static int stbi__jpeg_info(stbi__context *s, int *x, int *y, int *comp) +{ + stbi__jpeg j; + j.s = s; + return stbi__jpeg_info_raw(&j, x, y, comp); +} #endif +// public domain zlib decode v0.2 Sean Barrett 2006-11-18 +// simple implementation +// - all input must be provided in an upfront buffer +// - all output is written to a single output buffer (can malloc/realloc) +// performance +// - fast huffman + #ifndef STBI_NO_ZLIB -#define STBI__ZFAST_BITS 9 +// 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) +// zlib-style huffman encoding +// (jpegs packs from left, zlib from right, so can't share code) typedef struct { - uint16_t fast[1 << STBI__ZFAST_BITS]; - uint16_t firstcode[16]; + stbi__uint16 fast[1 << STBI__ZFAST_BITS]; + stbi__uint16 firstcode[16]; int maxcode[17]; - uint16_t firstsymbol[16]; - uint8_t size[288]; - uint16_t value[288]; + stbi__uint16 firstsymbol[16]; + stbi_uc size[288]; + stbi__uint16 value[288]; } stbi__zhuffman; -static INLINE int stbi__bitreverse16(int n) +stbi_inline static int stbi__bitreverse16(int n) { n = ((n & 0xAAAA) >> 1) | ((n & 0x5555) << 1); n = ((n & 0xCCCC) >> 2) | ((n & 0x3333) << 2); @@ -2332,17 +3460,20 @@ static INLINE int stbi__bitreverse16(int n) return n; } -static INLINE int stbi__bit_reverse(int v, int bits) +stbi_inline static int stbi__bit_reverse(int v, int bits) { STBI_ASSERT(bits <= 16); + // to bit reverse n bits, reverse 16 and shift + // e.g. 11 bits, bit reverse and shift away 5 return stbi__bitreverse16(v) >> (16-bits); } -static int stbi__zbuild_huffman(stbi__zhuffman *z, uint8_t *sizelist, int num) +static int stbi__zbuild_huffman(stbi__zhuffman *z, stbi_uc *sizelist, int num) { int i,k=0; int code, next_code[16], sizes[17]; + // DEFLATE spec for generating codes memset(sizes, 0, sizeof(sizes)); memset(z->fast, 0, sizeof(z->fast)); for (i=0; i < num; ++i) @@ -2354,25 +3485,23 @@ static int stbi__zbuild_huffman(stbi__zhuffman *z, uint8_t *sizelist, int num) code = 0; for (i=1; i < 16; ++i) { next_code[i] = code; - z->firstcode[i] = (uint16_t) code; - z->firstsymbol[i] = (uint16_t) k; + z->firstcode[i] = (stbi__uint16) code; + z->firstsymbol[i] = (stbi__uint16) k; code = (code + sizes[i]); if (sizes[i]) if (code-1 >= (1 << i)) return stbi__err("bad codelengths","Corrupt PNG"); - z->maxcode[i] = code << (16-i); + z->maxcode[i] = code << (16-i); // preshift for inner loop code <<= 1; k += sizes[i]; } - z->maxcode[16] = 0x10000; - for (i=0; i < num; ++i) - { + z->maxcode[16] = 0x10000; // sentinel + for (i=0; i < num; ++i) { int s = sizelist[i]; - if (s) - { + if (s) { int c = next_code[s] - z->firstcode[s] + z->firstsymbol[s]; - uint16_t fastv = (uint16_t) ((s << 9) | i); - z->size [c] = (uint8_t ) s; - z->value[c] = (uint16_t) i; + stbi__uint16 fastv = (stbi__uint16) ((s << 9) | i); + z->size [c] = (stbi_uc ) s; + z->value[c] = (stbi__uint16) i; if (s <= STBI__ZFAST_BITS) { int k = stbi__bit_reverse(next_code[s],s); while (k < (1 << STBI__ZFAST_BITS)) { @@ -2386,11 +3515,17 @@ static int stbi__zbuild_huffman(stbi__zhuffman *z, uint8_t *sizelist, int num) return 1; } +// zlib-from-memory implementation for PNG reading +// because PNG allows splitting the zlib stream arbitrarily, +// and it's annoying structurally to have PNG call ZLIB call PNG, +// we require PNG read all the IDATs and combine them into a single +// memory buffer + typedef struct { - uint8_t *zbuffer, *zbuffer_end; + stbi_uc *zbuffer, *zbuffer_end; int num_bits; - uint32_t code_buffer; + stbi__uint32 code_buffer; char *zout; char *zout_start; @@ -2400,7 +3535,7 @@ typedef struct stbi__zhuffman z_length, z_distance; } stbi__zbuf; -static INLINE uint8_t stbi__zget8(stbi__zbuf *z) +stbi_inline static stbi_uc stbi__zget8(stbi__zbuf *z) { if (z->zbuffer >= z->zbuffer_end) return 0; return *z->zbuffer++; @@ -2415,7 +3550,7 @@ static void stbi__fill_bits(stbi__zbuf *z) } while (z->num_bits <= 24); } -static INLINE unsigned int stbi__zreceive(stbi__zbuf *z, int n) +stbi_inline static unsigned int stbi__zreceive(stbi__zbuf *z, int n) { unsigned int k; if (z->num_bits < n) stbi__fill_bits(z); @@ -2428,11 +3563,14 @@ static INLINE unsigned int stbi__zreceive(stbi__zbuf *z, int n) static int stbi__zhuffman_decode_slowpath(stbi__zbuf *a, stbi__zhuffman *z) { int b,s,k; + // not resolved by fast table, so compute it the slow way + // use jpeg approach, which requires MSbits at top k = stbi__bit_reverse(a->code_buffer, 16); for (s=STBI__ZFAST_BITS+1; ; ++s) if (k < z->maxcode[s]) break; - if (s == 16) return -1; + 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); a->code_buffer >>= s; @@ -2440,7 +3578,7 @@ static int stbi__zhuffman_decode_slowpath(stbi__zbuf *a, stbi__zhuffman *z) return z->value[b]; } -static INLINE int stbi__zhuffman_decode(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); @@ -2454,7 +3592,7 @@ static INLINE int stbi__zhuffman_decode(stbi__zbuf *a, stbi__zhuffman *z) return stbi__zhuffman_decode_slowpath(a, z); } -static int stbi__zexpand(stbi__zbuf *z, char *zout, int n) +static int stbi__zexpand(stbi__zbuf *z, char *zout, int n) // need to make room for n bytes { char *q; int cur, limit; @@ -2464,7 +3602,7 @@ static int stbi__zexpand(stbi__zbuf *z, char *zout, int n) limit = (int) (z->zout_end - z->zout_start); while (cur + n > limit) limit *= 2; - q = (char *) realloc(z->zout_start, limit); + q = (char *) STBI_REALLOC(z->zout_start, limit); if (q == NULL) return stbi__err("outofmem", "Out of memory"); z->zout_start = q; z->zout = q + cur; @@ -2492,14 +3630,14 @@ static int stbi__parse_huffman_block(stbi__zbuf *a) for(;;) { int z = stbi__zhuffman_decode(a, &a->z_length); if (z < 256) { - if (z < 0) return stbi__err("bad huffman code","Corrupt PNG"); + if (z < 0) return stbi__err("bad huffman code","Corrupt PNG"); // error in huffman codes if (zout >= a->zout_end) { if (!stbi__zexpand(a, zout, 1)) return 0; zout = a->zout; } *zout++ = (char) z; } else { - uint8_t *p; + stbi_uc *p; int len,dist; if (z == 256) { a->zout = zout; @@ -2517,10 +3655,9 @@ static int stbi__parse_huffman_block(stbi__zbuf *a) if (!stbi__zexpand(a, zout, len)) return 0; zout = a->zout; } - p = (uint8_t *) (zout - dist); - if (dist == 1) - { - uint8_t v = *p; + p = (stbi_uc *) (zout - dist); + if (dist == 1) { // run of one byte; common in images. + stbi_uc v = *p; if (len) { do *zout++ = v; while (--len); } } else { if (len) { do *zout++ = *p++; while (--len); } @@ -2531,10 +3668,10 @@ static int stbi__parse_huffman_block(stbi__zbuf *a) static int stbi__compute_huffman_codes(stbi__zbuf *a) { - static uint8_t length_dezigzag[19] = { 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 }; + 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 }; stbi__zhuffman z_codelength; - uint8_t lencodes[286+32+137]; - uint8_t codelength_sizes[19]; + stbi_uc lencodes[286+32+137];//padding for maximum single op + stbi_uc codelength_sizes[19]; int i,n; int hlit = stbi__zreceive(a,5) + 257; @@ -2544,7 +3681,7 @@ static int stbi__compute_huffman_codes(stbi__zbuf *a) memset(codelength_sizes, 0, sizeof(codelength_sizes)); for (i=0; i < hclen; ++i) { int s = stbi__zreceive(a,3); - codelength_sizes[length_dezigzag[i]] = (uint8_t) s; + codelength_sizes[length_dezigzag[i]] = (stbi_uc) s; } if (!stbi__zbuild_huffman(&z_codelength, codelength_sizes, 19)) return 0; @@ -2553,7 +3690,7 @@ static int stbi__compute_huffman_codes(stbi__zbuf *a) int c = stbi__zhuffman_decode(a, &z_codelength); if (c < 0 || c >= 19) return stbi__err("bad codelengths", "Corrupt PNG"); if (c < 16) - lencodes[n++] = (uint8_t) c; + lencodes[n++] = (stbi_uc) c; else if (c == 16) { c = stbi__zreceive(a,2)+3; memset(lencodes+n, lencodes[n-1], c); @@ -2577,17 +3714,19 @@ static int stbi__compute_huffman_codes(stbi__zbuf *a) static int stbi__parse_uncomperssed_block(stbi__zbuf *a) { - uint8_t header[4]; + stbi_uc header[4]; int len,nlen,k; if (a->num_bits & 7) - stbi__zreceive(a, a->num_bits & 7); + stbi__zreceive(a, a->num_bits & 7); // discard + // drain the bit-packed data into header k = 0; while (a->num_bits > 0) { - header[k++] = (uint8_t) (a->code_buffer & 255); + header[k++] = (stbi_uc) (a->code_buffer & 255); // suppress MSVC run-time check a->code_buffer >>= 8; a->num_bits -= 8; } STBI_ASSERT(a->num_bits == 0); + // now fill header the normal way while (k < 4) header[k++] = stbi__zget8(a); len = header[1] * 256 + header[0]; @@ -2608,16 +3747,18 @@ static int stbi__parse_zlib_header(stbi__zbuf *a) int cm = cmf & 15; /* int cinfo = cmf >> 4; */ int flg = stbi__zget8(a); - if ((cmf*256+flg) % 31 != 0) return stbi__err("bad zlib header","Corrupt PNG"); - if (flg & 32) return stbi__err("no preset dict","Corrupt PNG"); - if (cm != 8) return stbi__err("bad compression","Corrupt PNG"); + 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 + // window = 1 << (8 + cinfo)... but who cares, we fully buffer output return 1; } -static uint8_t stbi__zdefault_length[288], stbi__zdefault_distance[32]; +// @TODO: should statically initialize these for optimal thread safety +static stbi_uc stbi__zdefault_length[288], stbi__zdefault_distance[32]; static void stbi__init_zdefaults(void) { - int i; + int i; // use <= to match clearly with spec for (i=0; i <= 143; ++i) stbi__zdefault_length[i] = 8; for ( ; i <= 255; ++i) stbi__zdefault_length[i] = 9; for ( ; i <= 279; ++i) stbi__zdefault_length[i] = 7; @@ -2642,6 +3783,7 @@ static int stbi__parse_zlib(stbi__zbuf *a, int parse_header) return 0; } else { if (type == 1) { + // use fixed code lengths if (!stbi__zdefault_distance[31]) stbi__init_zdefaults(); if (!stbi__zbuild_huffman(&a->z_length , stbi__zdefault_length , 288)) return 0; if (!stbi__zbuild_huffman(&a->z_distance, stbi__zdefault_distance, 32)) return 0; @@ -2664,28 +3806,97 @@ static int stbi__do_zlib(stbi__zbuf *a, char *obuf, int olen, int exp, int parse return stbi__parse_zlib(a, parse_header); } +STBIDEF char *stbi_zlib_decode_malloc_guesssize(const char *buffer, int len, int initial_size, int *outlen) +{ + stbi__zbuf a; + char *p = (char *) stbi__malloc(initial_size); + if (p == NULL) return NULL; + a.zbuffer = (stbi_uc *) buffer; + a.zbuffer_end = (stbi_uc *) buffer + len; + if (stbi__do_zlib(&a, p, initial_size, 1, 1)) { + if (outlen) *outlen = (int) (a.zout - a.zout_start); + return a.zout_start; + } else { + STBI_FREE(a.zout_start); + return NULL; + } +} + +STBIDEF char *stbi_zlib_decode_malloc(char const *buffer, int len, int *outlen) +{ + return stbi_zlib_decode_malloc_guesssize(buffer, len, 16384, outlen); +} + STBIDEF char *stbi_zlib_decode_malloc_guesssize_headerflag(const char *buffer, int len, int initial_size, int *outlen, int parse_header) { stbi__zbuf a; - char *p = (char *) malloc(initial_size); + char *p = (char *) stbi__malloc(initial_size); if (p == NULL) return NULL; - a.zbuffer = (uint8_t *) buffer; - a.zbuffer_end = (uint8_t *) buffer + len; + a.zbuffer = (stbi_uc *) buffer; + a.zbuffer_end = (stbi_uc *) buffer + len; if (stbi__do_zlib(&a, p, initial_size, 1, parse_header)) { if (outlen) *outlen = (int) (a.zout - a.zout_start); return a.zout_start; } else { - free(a.zout_start); + STBI_FREE(a.zout_start); return NULL; } } + +STBIDEF int stbi_zlib_decode_buffer(char *obuffer, int olen, char const *ibuffer, int ilen) +{ + stbi__zbuf a; + a.zbuffer = (stbi_uc *) ibuffer; + a.zbuffer_end = (stbi_uc *) ibuffer + ilen; + if (stbi__do_zlib(&a, obuffer, olen, 0, 1)) + return (int) (a.zout - a.zout_start); + else + return -1; +} + +STBIDEF char *stbi_zlib_decode_noheader_malloc(char const *buffer, int len, int *outlen) +{ + stbi__zbuf a; + char *p = (char *) stbi__malloc(16384); + if (p == NULL) return NULL; + a.zbuffer = (stbi_uc *) buffer; + a.zbuffer_end = (stbi_uc *) buffer+len; + if (stbi__do_zlib(&a, p, 16384, 1, 0)) { + if (outlen) *outlen = (int) (a.zout - a.zout_start); + return a.zout_start; + } else { + STBI_FREE(a.zout_start); + return NULL; + } +} + +STBIDEF int stbi_zlib_decode_noheader_buffer(char *obuffer, int olen, const char *ibuffer, int ilen) +{ + stbi__zbuf a; + a.zbuffer = (stbi_uc *) ibuffer; + a.zbuffer_end = (stbi_uc *) ibuffer + ilen; + if (stbi__do_zlib(&a, obuffer, olen, 0, 0)) + return (int) (a.zout - a.zout_start); + else + return -1; +} #endif +// public domain "baseline" PNG decoder v0.10 Sean Barrett 2006-11-18 +// simple implementation +// - only 8-bit samples +// - no CRC checking +// - allocates lots of intermediate memory +// - avoids problem of streaming data between subsystems +// - avoids explicit window management +// performance +// - uses stb_zlib, a PD zlib implementation with fast huffman decoding + #ifndef STBI_NO_PNG typedef struct { - uint32_t length; - uint32_t type; + stbi__uint32 length; + stbi__uint32 type; } stbi__pngchunk; static stbi__pngchunk stbi__get_chunk_header(stbi__context *s) @@ -2698,7 +3909,7 @@ static stbi__pngchunk stbi__get_chunk_header(stbi__context *s) static int stbi__check_png_header(stbi__context *s) { - static uint8_t png_sig[8] = { 137,80,78,71,13,10,26,10 }; + static 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"); @@ -2708,7 +3919,7 @@ static int stbi__check_png_header(stbi__context *s) typedef struct { stbi__context *s; - uint8_t *idata, *expanded, *out; + stbi_uc *idata, *expanded, *out; } stbi__png; @@ -2718,11 +3929,12 @@ enum { STBI__F_up=2, STBI__F_avg=3, STBI__F_paeth=4, + // synthetic filters used for first scanline to avoid needing a dummy row of 0s STBI__F_avg_first, STBI__F_paeth_first }; -static uint8_t first_row_filter[5] = +static stbi_uc first_row_filter[5] = { STBI__F_none, STBI__F_sub, @@ -2742,34 +3954,32 @@ static int stbi__paeth(int a, int b, int c) return c; } -static uint8_t stbi__depth_scale_table[9] = { 0, 0xff, 0x55, 0, 0x11, 0,0,0, 0x01 }; +static stbi_uc stbi__depth_scale_table[9] = { 0, 0xff, 0x55, 0, 0x11, 0,0,0, 0x01 }; -static int stbi__create_png_image_raw(stbi__png *a, uint8_t *raw, uint32_t raw_len, int out_n, uint32_t x, uint32_t y, int depth, int color) +// 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) { - int k; stbi__context *s = a->s; - uint32_t i,j,stride = x*out_n; - uint32_t img_len, img_width_bytes; - int img_n = s->img_n; + stbi__uint32 i,j,stride = x*out_n; + stbi__uint32 img_len, img_width_bytes; + int k; + int img_n = s->img_n; // copy it into a local for later STBI_ASSERT(out_n == s->img_n || out_n == s->img_n+1); - a->out = (uint8_t *) malloc(x * y * out_n); + a->out = (stbi_uc *) stbi__malloc(x * y * out_n); // extra bytes to write off the end into if (!a->out) return stbi__err("outofmem", "Out of memory"); 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 (s->img_x == x && s->img_y == y) { if (raw_len != img_len) return stbi__err("not enough pixels","Corrupt PNG"); - } - else - { + } else { // interlaced: if (raw_len < img_len) return stbi__err("not enough pixels","Corrupt PNG"); } for (j=0; j < y; ++j) { - uint8_t *cur = a->out + stride*j; - uint8_t *prior = cur - stride; + stbi_uc *cur = a->out + stride*j; + stbi_uc *prior = cur - stride; int filter = *raw++; int filter_bytes = img_n; int width = x; @@ -2778,17 +3988,17 @@ static int stbi__create_png_image_raw(stbi__png *a, uint8_t *raw, uint32_t raw_l if (depth < 8) { STBI_ASSERT(img_width_bytes <= x); - cur += x*out_n - img_width_bytes; + 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; } + // if first row, use special filter that doesn't sample previous row if (j == 0) filter = first_row_filter[filter]; - for (k=0; k < filter_bytes; ++k) - { - switch (filter) - { + // handle first byte explicitly + for (k=0; k < filter_bytes; ++k) { + switch (filter) { case STBI__F_none : cur[k] = raw[k]; break; case STBI__F_sub : cur[k] = raw[k]; break; case STBI__F_up : cur[k] = STBI__BYTECAST(raw[k] + prior[k]); break; @@ -2801,7 +4011,7 @@ static int stbi__create_png_image_raw(stbi__png *a, uint8_t *raw, uint32_t raw_l if (depth == 8) { if (img_n != out_n) - cur[img_n] = 255; + cur[img_n] = 255; // first pixel raw += img_n; cur += out_n; prior += out_n; @@ -2811,13 +4021,14 @@ static int stbi__create_png_image_raw(stbi__png *a, uint8_t *raw, uint32_t raw_l prior += 1; } - if (depth < 8 || img_n == out_n) - { + // this is a little gross, so that we don't switch per-pixel or per-component + if (depth < 8 || img_n == out_n) { int nk = (width - 1)*img_n; #define CASE(f) \ case f: \ for (k=0; k < nk; ++k) switch (filter) { + // "none" filter turns into a memcpy here; make that explicit. case STBI__F_none: memcpy(cur, raw, nk); break; CASE(STBI__F_sub) cur[k] = STBI__BYTECAST(raw[k] + cur[k-filter_bytes]); break; CASE(STBI__F_up) cur[k] = STBI__BYTECAST(raw[k] + prior[k]); break; @@ -2847,16 +4058,24 @@ static int stbi__create_png_image_raw(stbi__png *a, uint8_t *raw, uint32_t raw_l } } - if (depth < 8) - { - for (j=0; j < y; ++j) - { - uint8_t *cur = a->out + stride*j; - uint8_t *in = a->out + stride*j + x*out_n - img_width_bytes; - uint8_t scale = (color == 0) ? stbi__depth_scale_table[depth] : 1; + // we make a separate pass to expand bits to pixels; for performance, + // this could run two scanlines behind the above code, so it won't + // intefere with filtering but will still be in the cache. + if (depth < 8) { + for (j=0; j < y; ++j) { + stbi_uc *cur = a->out + stride*j; + stbi_uc *in = a->out + stride*j + x*out_n - img_width_bytes; + // unpack 1/2/4-bit into a 8-bit buffer. allows us to keep the common 8-bit path optimal at minimal cost for 1/2/4-bit + // png guarante byte alignment, if width is not multiple of 8/4/2 we'll decode dummy trailing data that will be skipped in the later loop + stbi_uc scale = (color == 0) ? stbi__depth_scale_table[depth] : 1; // scale grayscale values to 0..255 range - if (depth == 4) - { + // note that the final byte might overshoot and write more data than desired. + // we can allocate enough data that this never writes out of memory, but it + // could also overwrite the next scanline. can it overwrite non-empty data + // on the next scanline? yes, consider 1-pixel-wide scanlines with 1-bit-per-pixel. + // so we need to explicitly clamp the final ones + + if (depth == 4) { for (k=x*img_n; k >= 2; k-=2, ++in) { *cur++ = scale * ((*in >> 4) ); *cur++ = scale * ((*in ) & 0x0f); @@ -2892,7 +4111,8 @@ static int stbi__create_png_image_raw(stbi__png *a, uint8_t *raw, uint32_t raw_l if (k > 6) *cur++ = scale * ((*in >> 1) & 0x01); } if (img_n != out_n) { - uint8_t *cur = a->out + stride*j; + // insert alpha = 255 + stbi_uc *cur = a->out + stride*j; int i; if (img_n == 1) { for (i=x-1; i >= 0; --i) { @@ -2915,26 +4135,28 @@ static int stbi__create_png_image_raw(stbi__png *a, uint8_t *raw, uint32_t raw_l return 1; } -static int stbi__create_png_image(stbi__png *a, uint8_t *image_data, uint32_t image_data_len, int out_n, int depth, int color, int interlaced) +static int stbi__create_png_image(stbi__png *a, stbi_uc *image_data, stbi__uint32 image_data_len, int out_n, int depth, int color, int interlaced) { - uint8_t *final; + stbi_uc *final; int p; if (!interlaced) return stbi__create_png_image_raw(a, image_data, image_data_len, out_n, a->s->img_x, a->s->img_y, depth, color); - final = (uint8_t *) malloc(a->s->img_x * a->s->img_y * out_n); + // de-interlacing + final = (stbi_uc *) stbi__malloc(a->s->img_x * a->s->img_y * out_n); for (p=0; p < 7; ++p) { int xorig[] = { 0,4,0,2,0,1,0 }; int yorig[] = { 0,0,4,0,2,0,1 }; int xspc[] = { 8,8,4,4,2,2,1 }; int yspc[] = { 8,8,8,4,4,2,2 }; int i,j,x,y; + // pass1_x[4] = 0, pass1_x[5] = 1, pass1_x[12] = 1 x = (a->s->img_x - xorig[p] + xspc[p]-1) / xspc[p]; y = (a->s->img_y - yorig[p] + yspc[p]-1) / yspc[p]; if (x && y) { - uint32_t img_len = ((((a->s->img_n * x * depth) + 7) >> 3) + 1) * y; + stbi__uint32 img_len = ((((a->s->img_n * x * depth) + 7) >> 3) + 1) * y; if (!stbi__create_png_image_raw(a, image_data, image_data_len, out_n, x, y, depth, color)) { - free(final); + STBI_FREE(final); return 0; } for (j=0; j < y; ++j) { @@ -2942,10 +4164,10 @@ static int stbi__create_png_image(stbi__png *a, uint8_t *image_data, uint32_t im int out_y = j*yspc[p]+yorig[p]; int out_x = i*xspc[p]+xorig[p]; memcpy(final + out_y*a->s->img_x*out_n + out_x*out_n, - a->out + (j*x+i)*out_n, out_n); + a->out + (j*x+i)*out_n, out_n); } } - free(a->out); + STBI_FREE(a->out); image_data += img_len; image_data_len -= img_len; } @@ -2955,12 +4177,14 @@ static int stbi__create_png_image(stbi__png *a, uint8_t *image_data, uint32_t im return 1; } -static int stbi__compute_transparency(stbi__png *z, uint8_t tc[3], int out_n) +static int stbi__compute_transparency(stbi__png *z, stbi_uc tc[3], int out_n) { stbi__context *s = z->s; - uint32_t i, pixel_count = s->img_x * s->img_y; - uint8_t *p = z->out; + stbi__uint32 i, pixel_count = s->img_x * s->img_y; + stbi_uc *p = z->out; + // compute color-based transparency, assuming we've + // already got 255 as the alpha value in the output STBI_ASSERT(out_n == 2 || out_n == 4); if (out_n == 2) { @@ -2978,14 +4202,15 @@ static int stbi__compute_transparency(stbi__png *z, uint8_t tc[3], int out_n) return 1; } -static int stbi__expand_png_palette(stbi__png *a, uint8_t *palette, int len, int pal_img_n) +static int stbi__expand_png_palette(stbi__png *a, stbi_uc *palette, int len, int pal_img_n) { - uint32_t i, pixel_count = a->s->img_x * a->s->img_y; - uint8_t *p, *temp_out, *orig = a->out; + stbi__uint32 i, pixel_count = a->s->img_x * a->s->img_y; + stbi_uc *p, *temp_out, *orig = a->out; - p = (uint8_t *) malloc(pixel_count * pal_img_n); + p = (stbi_uc *) stbi__malloc(pixel_count * pal_img_n); if (p == NULL) return stbi__err("outofmem", "Out of memory"); + // between here and free(out) below, exitting would leak temp_out = p; if (pal_img_n == 3) { @@ -3006,7 +4231,7 @@ static int stbi__expand_png_palette(stbi__png *a, uint8_t *palette, int len, int p += 4; } } - free(a->out); + STBI_FREE(a->out); a->out = temp_out; STBI_NOTUSED(len); @@ -3017,29 +4242,36 @@ static int stbi__expand_png_palette(stbi__png *a, uint8_t *palette, int len, int static int stbi__unpremultiply_on_load = 0; static int stbi__de_iphone_flag = 0; +STBIDEF void stbi_set_unpremultiply_on_load(int flag_true_if_should_unpremultiply) +{ + stbi__unpremultiply_on_load = 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; +} + static void stbi__de_iphone(stbi__png *z) { stbi__context *s = z->s; - uint32_t i, pixel_count = s->img_x * s->img_y; - uint8_t *p = z->out; + stbi__uint32 i, pixel_count = s->img_x * s->img_y; + stbi_uc *p = z->out; - if (s->img_out_n == 3) - { + if (s->img_out_n == 3) { // convert bgr to rgb for (i=0; i < pixel_count; ++i) { - uint8_t t = p[0]; + stbi_uc t = p[0]; p[0] = p[2]; p[2] = t; p += 3; } - } - else - { + } else { STBI_ASSERT(s->img_out_n == 4); - if (stbi__unpremultiply_on_load) - { + if (stbi__unpremultiply_on_load) { + // convert bgr to rgb and unpremultiply for (i=0; i < pixel_count; ++i) { - uint8_t a = p[3]; - uint8_t t = p[0]; + stbi_uc a = p[3]; + stbi_uc t = p[0]; if (a) { p[0] = p[2] * 255 / a; p[1] = p[1] * 255 / a; @@ -3051,8 +4283,9 @@ static void stbi__de_iphone(stbi__png *z) p += 4; } } else { + // convert bgr to rgb for (i=0; i < pixel_count; ++i) { - uint8_t t = p[0]; + stbi_uc t = p[0]; p[0] = p[2]; p[2] = t; p += 4; @@ -3065,9 +4298,9 @@ static void stbi__de_iphone(stbi__png *z) static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp) { - uint8_t palette[1024], pal_img_n=0; - uint8_t has_trans=0, tc[3]; - uint32_t ioff=0, idata_limit=0, i, pal_len=0; + stbi_uc palette[1024], pal_img_n=0; + stbi_uc has_trans=0, tc[3]; + stbi__uint32 ioff=0, idata_limit=0, i, pal_len=0; int first=1,k,interlace=0, color=0, depth=0, is_iphone=0; stbi__context *s = z->s; @@ -3105,8 +4338,11 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp) if ((1 << 30) / s->img_x / s->img_n < s->img_y) return stbi__err("too large", "Image too large to decode"); if (scan == STBI__SCAN_header) return 1; } else { + // if paletted, then pal_n is our final components, and + // img_n is # components to decompress/filter. s->img_n = 1; if ((1 << 30) / s->img_x / 4 < s->img_y) return stbi__err("too large","Corrupt PNG"); + // if SCAN_header, have to scan to see if we have a tRNS } break; } @@ -3137,10 +4373,10 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp) palette[i*4+3] = stbi__get8(s); } else { if (!(s->img_n & 1)) return stbi__err("tRNS with alpha","Corrupt PNG"); - if (c.length != (uint32_t) s->img_n*2) return stbi__err("bad tRNS len","Corrupt PNG"); + if (c.length != (stbi__uint32) s->img_n*2) return stbi__err("bad tRNS len","Corrupt PNG"); has_trans = 1; for (k=0; k < s->img_n; ++k) - tc[k] = (uint8_t) (stbi__get16be(s) & 255) * stbi__depth_scale_table[depth]; + tc[k] = (stbi_uc) (stbi__get16be(s) & 255) * stbi__depth_scale_table[depth]; // non 8-bit images will be larger } break; } @@ -3151,11 +4387,11 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp) if (scan == STBI__SCAN_header) { s->img_n = pal_img_n; return 1; } if ((int)(ioff + c.length) < (int)ioff) return 0; if (ioff + c.length > idata_limit) { - uint8_t *p; + stbi_uc *p; if (idata_limit == 0) idata_limit = c.length > 4096 ? c.length : 4096; while (ioff + c.length > idata_limit) idata_limit *= 2; - p = (uint8_t *) realloc(z->idata, idata_limit); if (p == NULL) return stbi__err("outofmem", "Out of memory"); + p = (stbi_uc *) STBI_REALLOC(z->idata, idata_limit); if (p == NULL) return stbi__err("outofmem", "Out of memory"); z->idata = p; } if (!stbi__getn(s, z->idata+ioff,c.length)) return stbi__err("outofdata","Corrupt PNG"); @@ -3164,15 +4400,16 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp) } case STBI__PNG_TYPE('I','E','N','D'): { - uint32_t raw_len, bpl; + stbi__uint32 raw_len, bpl; if (first) return stbi__err("first not IHDR", "Corrupt PNG"); if (scan != STBI__SCAN_load) return 1; if (z->idata == NULL) return stbi__err("no IDAT","Corrupt PNG"); - bpl = (s->img_x * depth + 7) / 8; + // initial guess for decoded data size to avoid unnecessary reallocs + bpl = (s->img_x * depth + 7) / 8; // bytes per line, per component raw_len = bpl * s->img_y * s->img_n /* pixels */ + s->img_y /* filter mode per row */; - z->expanded = (uint8_t *) stbi_zlib_decode_malloc_guesssize_headerflag((char *) z->idata, ioff, raw_len, (int *) &raw_len, !is_iphone); - if (z->expanded == NULL) return 0; - free(z->idata); z->idata = NULL; + z->expanded = (stbi_uc *) stbi_zlib_decode_malloc_guesssize_headerflag((char *) z->idata, ioff, raw_len, (int *) &raw_len, !is_iphone); + if (z->expanded == NULL) return 0; // zlib should set error + STBI_FREE(z->idata); z->idata = NULL; if ((req_comp == s->img_n+1 && req_comp != 3 && !pal_img_n) || has_trans) s->img_out_n = s->img_n+1; else @@ -3183,20 +4420,23 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp) if (is_iphone && stbi__de_iphone_flag && s->img_out_n > 2) stbi__de_iphone(z); if (pal_img_n) { - s->img_n = pal_img_n; + // pal_img_n == 3 or 4 + s->img_n = pal_img_n; // record the actual colors we had s->img_out_n = pal_img_n; 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; } - free(z->expanded); z->expanded = NULL; + STBI_FREE(z->expanded); z->expanded = NULL; return 1; } default: + // if critical, fail if (first) return stbi__err("first not IHDR", "Corrupt PNG"); if ((c.type & (1 << 29)) == 0) { #ifndef STBI_NO_FAILURE_STRINGS + // not threadsafe static char invalid_chunk[] = "XXXX PNG chunk not known"; invalid_chunk[0] = STBI__BYTECAST(c.type >> 24); invalid_chunk[1] = STBI__BYTECAST(c.type >> 16); @@ -3208,6 +4448,7 @@ static int stbi__parse_png_file(stbi__png *z, int scan, int req_comp) stbi__skip(s, c.length); break; } + // end of PNG chunk, read and skip CRC stbi__get32be(s); } } @@ -3228,9 +4469,9 @@ static unsigned char *stbi__do_png(stbi__png *p, int *x, int *y, int *n, int req *y = p->s->img_y; if (n) *n = p->s->img_out_n; } - free(p->out); p->out = NULL; - free(p->expanded); p->expanded = NULL; - free(p->idata); p->idata = NULL; + STBI_FREE(p->out); p->out = NULL; + STBI_FREE(p->expanded); p->expanded = NULL; + STBI_FREE(p->idata); p->idata = NULL; return result; } @@ -3249,8 +4490,29 @@ static int stbi__png_test(stbi__context *s) stbi__rewind(s); return r; } + +static int stbi__png_info_raw(stbi__png *p, int *x, int *y, int *comp) +{ + if (!stbi__parse_png_file(p, STBI__SCAN_header, 0)) { + stbi__rewind( p->s ); + return 0; + } + if (x) *x = p->s->img_x; + if (y) *y = p->s->img_y; + if (comp) *comp = p->s->img_n; + return 1; +} + +static int stbi__png_info(stbi__context *s, int *x, int *y, int *comp) +{ + stbi__png p; + p.s = s; + return stbi__png_info_raw(&p, x, y, comp); +} #endif +// Microsoft/Windows BMP image + #ifndef STBI_NO_BMP static int stbi__bmp_test_raw(stbi__context *s) { @@ -3258,10 +4520,10 @@ static int stbi__bmp_test_raw(stbi__context *s) int sz; if (stbi__get8(s) != 'B') return 0; if (stbi__get8(s) != 'M') return 0; - stbi__get32le(s); - stbi__get16le(s); - stbi__get16le(s); - stbi__get32le(s); + stbi__get32le(s); // discard filesize + stbi__get16le(s); // discard reserved + stbi__get16le(s); // discard reserved + stbi__get32le(s); // discard data offset sz = stbi__get32le(s); r = (sz == 12 || sz == 40 || sz == 56 || sz == 108 || sz == 124); return r; @@ -3275,6 +4537,7 @@ static int stbi__bmp_test(stbi__context *s) } +// returns 0..31 for the highest set bit static int stbi__high_bit(unsigned int z) { int n=0; @@ -3289,11 +4552,11 @@ static int stbi__high_bit(unsigned int z) static int stbi__bitcount(unsigned int a) { - a = (a & 0x55555555) + ((a >> 1) & 0x55555555); - a = (a & 0x33333333) + ((a >> 2) & 0x33333333); - a = (a + (a >> 4)) & 0x0f0f0f0f; - a = (a + (a >> 8)); - a = (a + (a >> 16)); + a = (a & 0x55555555) + ((a >> 1) & 0x55555555); // max 2 + a = (a & 0x33333333) + ((a >> 2) & 0x33333333); // max 4 + a = (a + (a >> 4)) & 0x0f0f0f0f; // max 8 per 4, now 8 bits + a = (a + (a >> 8)); // max 16 per 8 bits + a = (a + (a >> 16)); // max 32 per 8 bits return a & 0xff; } @@ -3314,17 +4577,17 @@ static int stbi__shiftsigned(int v, int shift, int bits) return result; } -static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp) +static stbi_uc *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int req_comp) { - uint8_t *out; + stbi_uc *out; unsigned int mr=0,mg=0,mb=0,ma=0, fake_a=0; - uint8_t pal[256][4]; + stbi_uc pal[256][4]; int psize=0,i,j,compress=0,width; int bpp, flip_vertically, pad, target, offset, hsz; if (stbi__get8(s) != 'B' || stbi__get8(s) != 'M') return stbi__errpuc("not BMP", "Corrupt BMP"); - stbi__get32le(s); - stbi__get16le(s); - stbi__get16le(s); + stbi__get32le(s); // discard filesize + stbi__get16le(s); // discard reserved + stbi__get16le(s); // discard reserved offset = stbi__get32le(s); hsz = stbi__get32le(s); if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) return stbi__errpuc("unknown BMP", "BMP type not supported: unknown"); @@ -3346,11 +4609,11 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int } else { compress = stbi__get32le(s); if (compress == 1 || compress == 2) return stbi__errpuc("BMP RLE", "BMP type not supported: RLE"); - stbi__get32le(s); - stbi__get32le(s); - stbi__get32le(s); - stbi__get32le(s); - stbi__get32le(s); + stbi__get32le(s); // discard sizeof + stbi__get32le(s); // discard hres + stbi__get32le(s); // discard vres + stbi__get32le(s); // discard colorsused + stbi__get32le(s); // discard max important if (hsz == 40 || hsz == 56) { if (hsz == 56) { stbi__get32le(s); @@ -3366,7 +4629,7 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int mg = 0xffu << 8; mb = 0xffu << 0; ma = 0xffu << 24; - fake_a = 1; + fake_a = 1; // @TODO: check for cases like alpha value is all 0 and switch it to 255 STBI_NOTUSED(fake_a); } else { mr = 31u << 10; @@ -3377,7 +4640,9 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int mr = stbi__get32le(s); mg = stbi__get32le(s); mb = stbi__get32le(s); + // not documented, but generated by photoshop and handled by mspaint if (mr == mg && mg == mb) { + // ?!?!? return stbi__errpuc("bad BMP", "bad BMP"); } } else @@ -3389,29 +4654,29 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int mg = stbi__get32le(s); mb = stbi__get32le(s); ma = stbi__get32le(s); - stbi__get32le(s); + stbi__get32le(s); // discard color space for (i=0; i < 12; ++i) - stbi__get32le(s); + stbi__get32le(s); // discard color space parameters if (hsz == 124) { - stbi__get32le(s); - stbi__get32le(s); - stbi__get32le(s); - stbi__get32le(s); + stbi__get32le(s); // discard rendering intent + stbi__get32le(s); // discard offset of profile data + stbi__get32le(s); // discard size of profile data + stbi__get32le(s); // discard reserved } } if (bpp < 16) psize = (offset - 14 - hsz) >> 2; } s->img_n = ma ? 4 : 3; - if (req_comp && req_comp >= 3) + if (req_comp && req_comp >= 3) // we can directly decode 3 or 4 target = req_comp; else - target = s->img_n; - out = (uint8_t *) malloc(target * s->img_x * s->img_y); + target = s->img_n; // if they want monochrome, we'll post-convert + out = (stbi_uc *) stbi__malloc(target * s->img_x * s->img_y); if (!out) return stbi__errpuc("outofmem", "Out of memory"); if (bpp < 16) { int z=0; - if (psize == 0 || psize > 256) { free(out); return stbi__errpuc("invalid", "Corrupt BMP"); } + if (psize == 0 || psize > 256) { STBI_FREE(out); return stbi__errpuc("invalid", "Corrupt BMP"); } for (i=0; i < psize; ++i) { pal[i][2] = stbi__get8(s); pal[i][1] = stbi__get8(s); @@ -3422,7 +4687,7 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int stbi__skip(s, offset - 14 - hsz - psize * (hsz == 12 ? 3 : 4)); if (bpp == 4) width = (s->img_x + 1) >> 1; else if (bpp == 8) width = s->img_x; - else { free(out); return stbi__errpuc("bad bpp", "Corrupt BMP"); } + 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) { @@ -3460,7 +4725,8 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int easy = 2; } if (!easy) { - if (!mr || !mg || !mb) { free(out); return stbi__errpuc("bad masks", "Corrupt BMP"); } + if (!mr || !mg || !mb) { STBI_FREE(out); return stbi__errpuc("bad masks", "Corrupt BMP"); } + // right shift amt to put high bit in position #7 rshift = stbi__high_bit(mr)-7; rcount = stbi__bitcount(mr); gshift = stbi__high_bit(mg)-7; gcount = stbi__bitcount(mg); bshift = stbi__high_bit(mb)-7; bcount = stbi__bitcount(mb); @@ -3479,7 +4745,7 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int } } else { for (i=0; i < (int) s->img_x; ++i) { - uint32_t v = (bpp == 16 ? (uint32_t) stbi__get16le(s) : stbi__get32le(s)); + stbi__uint32 v = (bpp == 16 ? (stbi__uint32) stbi__get16le(s) : stbi__get32le(s)); int a; out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mr, rshift, rcount)); out[z++] = STBI__BYTECAST(stbi__shiftsigned(v & mg, gshift, gcount)); @@ -3492,10 +4758,10 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int } } if (flip_vertically) { - uint8_t t; + stbi_uc t; for (j=0; j < (int) s->img_y>>1; ++j) { - uint8_t *p1 = out + j *s->img_x*target; - uint8_t *p2 = out + (s->img_y-1-j)*s->img_x*target; + 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; } @@ -3504,7 +4770,7 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int if (req_comp && req_comp != target) { out = stbi__convert_format(out, target, req_comp, s->img_x, s->img_y); - if (out == NULL) return out; + if (out == NULL) return out; // stbi__convert_format frees input on failure } *x = s->img_x; @@ -3514,24 +4780,63 @@ static uint8_t *stbi__bmp_load(stbi__context *s, int *x, int *y, int *comp, int } #endif +// Targa Truevision - TGA +// by Jonathan Dummer #ifndef STBI_NO_TGA +static int stbi__tga_info(stbi__context *s, int *x, int *y, int *comp) +{ + int tga_w, tga_h, tga_comp; + int sz; + stbi__get8(s); // discard Offset + sz = stbi__get8(s); // color type + if( sz > 1 ) { + stbi__rewind(s); + return 0; // only RGB or indexed allowed + } + sz = stbi__get8(s); // image type + // only RGB or grey allowed, +/- RLE + if ((sz != 1) && (sz != 2) && (sz != 3) && (sz != 9) && (sz != 10) && (sz != 11)) return 0; + stbi__skip(s,9); + tga_w = stbi__get16le(s); + if( tga_w < 1 ) { + stbi__rewind(s); + return 0; // test width + } + tga_h = stbi__get16le(s); + if( tga_h < 1 ) { + stbi__rewind(s); + return 0; // test height + } + sz = stbi__get8(s); // bits per pixel + // only RGB or RGBA or grey allowed + if ((sz != 8) && (sz != 16) && (sz != 24) && (sz != 32)) { + stbi__rewind(s); + return 0; + } + tga_comp = sz; + if (x) *x = tga_w; + if (y) *y = tga_h; + if (comp) *comp = tga_comp / 8; + return 1; // seems to have passed everything +} + static int stbi__tga_test(stbi__context *s) { int res; int sz; - stbi__get8(s); - sz = stbi__get8(s); - if ( sz > 1 ) return 0; - sz = stbi__get8(s); - if ( (sz != 1) && (sz != 2) && (sz != 3) && (sz != 9) && (sz != 10) && (sz != 11) ) return 0; - stbi__get16be(s); - stbi__get16be(s); - stbi__get8(s); - stbi__get16be(s); - stbi__get16be(s); - if ( stbi__get16be(s) < 1 ) return 0; - if ( stbi__get16be(s) < 1 ) return 0; - sz = stbi__get8(s); + stbi__get8(s); // discard Offset + sz = stbi__get8(s); // color type + if ( sz > 1 ) return 0; // only RGB or indexed allowed + sz = stbi__get8(s); // image type + if ( (sz != 1) && (sz != 2) && (sz != 3) && (sz != 9) && (sz != 10) && (sz != 11) ) return 0; // only RGB or grey allowed, +/- RLE + stbi__get16be(s); // discard palette start + stbi__get16be(s); // discard palette length + stbi__get8(s); // discard bits per palette color entry + stbi__get16be(s); // discard x origin + stbi__get16be(s); // discard y origin + if ( stbi__get16be(s) < 1 ) return 0; // test width + if ( stbi__get16be(s) < 1 ) return 0; // test height + sz = stbi__get8(s); // bits per pixel if ( (sz != 8) && (sz != 16) && (sz != 24) && (sz != 32) ) res = 0; else @@ -3540,10 +4845,9 @@ static int stbi__tga_test(stbi__context *s) return res; } -static uint8_t *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp) +static stbi_uc *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int req_comp) { - unsigned char *tga_data; - int i, j; + // read in the TGA header stuff int tga_offset = stbi__get8(s); int tga_indexed = stbi__get8(s); int tga_image_type = stbi__get8(s); @@ -3558,68 +4862,85 @@ static uint8_t *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int int tga_bits_per_pixel = stbi__get8(s); int tga_comp = tga_bits_per_pixel / 8; int tga_inverted = stbi__get8(s); + // image data + unsigned char *tga_data; unsigned char *tga_palette = NULL; + int i, j; unsigned char raw_data[4] = {0}; int RLE_count = 0; int RLE_repeating = 0; int read_next_pixel = 1; + // do a tiny bit of precessing if ( tga_image_type >= 8 ) { tga_image_type -= 8; tga_is_RLE = 1; } - + /* int tga_alpha_bits = tga_inverted & 15; */ tga_inverted = 1 - ((tga_inverted >> 5) & 1); - if ( (tga_width < 1) || (tga_height < 1) || + // error check + if ( //(tga_indexed) || + (tga_width < 1) || (tga_height < 1) || (tga_image_type < 1) || (tga_image_type > 3) || ((tga_bits_per_pixel != 8) && (tga_bits_per_pixel != 16) && (tga_bits_per_pixel != 24) && (tga_bits_per_pixel != 32)) ) - return NULL; + { + return NULL; // we don't report this as a bad TGA because we don't even know if it's TGA + } + // If I'm paletted, then I'll use the number of bits from the palette if ( tga_indexed ) + { tga_comp = tga_palette_bits / 8; + } + // tga info *x = tga_width; *y = tga_height; if (comp) *comp = tga_comp; - tga_data = (unsigned char*)malloc( (size_t)tga_width * tga_height * tga_comp ); + tga_data = (unsigned char*)stbi__malloc( (size_t)tga_width * tga_height * tga_comp ); if (!tga_data) return stbi__errpuc("outofmem", "Out of memory"); + // skip to the data's starting position (offset usually = 0) stbi__skip(s, tga_offset ); if ( !tga_indexed && !tga_is_RLE) { for (i=0; i < tga_height; ++i) { int y = tga_inverted ? tga_height -i - 1 : i; - uint8_t *tga_row = tga_data + y*tga_width*tga_comp; + stbi_uc *tga_row = tga_data + y*tga_width*tga_comp; stbi__getn(s, tga_row, tga_width * tga_comp); } - } - else - { + } else { + // do I need to load a palette? if ( tga_indexed) { + // any data to skip? (offset usually = 0) stbi__skip(s, tga_palette_start ); - tga_palette = (unsigned char*)malloc( tga_palette_len * tga_palette_bits / 8 ); + // load the palette + tga_palette = (unsigned char*)stbi__malloc( tga_palette_len * tga_palette_bits / 8 ); if (!tga_palette) { - free(tga_data); + STBI_FREE(tga_data); return stbi__errpuc("outofmem", "Out of memory"); } if (!stbi__getn(s, tga_palette, tga_palette_len * tga_palette_bits / 8 )) { - free(tga_data); - free(tga_palette); + STBI_FREE(tga_data); + STBI_FREE(tga_palette); return stbi__errpuc("bad palette", "Corrupt TGA"); } } + // load the data for (i=0; i < tga_width * tga_height; ++i) { + // if I'm in RLE mode, do I need to get a RLE stbi__pngchunk? if ( tga_is_RLE ) { if ( RLE_count == 0 ) { + // yep, get the next byte as a RLE command int RLE_cmd = stbi__get8(s); RLE_count = 1 + (RLE_cmd & 127); RLE_repeating = RLE_cmd >> 7; @@ -3632,13 +4953,17 @@ static uint8_t *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int { read_next_pixel = 1; } + // OK, if I need to read a pixel, do it now if ( read_next_pixel ) { + // load however much data we did have if ( tga_indexed ) { + // read in 1 byte, then perform the lookup int pal_idx = stbi__get8(s); if ( pal_idx >= tga_palette_len ) { + // invalid index pal_idx = 0; } pal_idx *= tga_bits_per_pixel / 8; @@ -3648,17 +4973,24 @@ static uint8_t *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int } } else { + // read in the data raw for (j = 0; j*8 < tga_bits_per_pixel; ++j) + { raw_data[j] = stbi__get8(s); + } } + // clear the reading flag for the next pixel read_next_pixel = 0; - } + } // end of reading a pixel + // copy data for (j = 0; j < tga_comp; ++j) tga_data[i*tga_comp+j] = raw_data[j]; + // in case we're in RLE mode, keep counting down --RLE_count; } + // do I need to invert the image? if ( tga_inverted ) { for (j = 0; j*2 < tga_height; ++j) @@ -3675,12 +5007,14 @@ static uint8_t *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int } } } + // clear my palette, if I had one if ( tga_palette != NULL ) { - free( tga_palette ); + STBI_FREE( tga_palette ); } } + // swap RGB if (tga_comp >= 3) { unsigned char* tga_pixel = tga_data; @@ -3693,13 +5027,1413 @@ static uint8_t *stbi__tga_load(stbi__context *s, int *x, int *y, int *comp, int } } + // convert to target component count if (req_comp && req_comp != tga_comp) tga_data = stbi__convert_format(tga_data, tga_comp, req_comp, tga_width, tga_height); + // the things I do to get rid of an error message, and yet keep + // Microsoft's C compilers happy... [8^( tga_palette_start = tga_palette_len = tga_palette_bits = tga_x_origin = tga_y_origin = 0; + // OK, done return tga_data; } #endif +// ************************************************************************************************* +// Photoshop PSD loader -- PD by Thatcher Ulrich, integration by Nicolas Schulz, tweaked by STB + +#ifndef STBI_NO_PSD +static int stbi__psd_test(stbi__context *s) +{ + int r = (stbi__get32be(s) == 0x38425053); + stbi__rewind(s); + return r; +} + +static stbi_uc *stbi__psd_load(stbi__context *s, int *x, int *y, int *comp, int req_comp) +{ + int pixelCount; + int channelCount, compression; + int channel, i, count, len; + int w,h; + stbi_uc *out; + + // Check identifier + if (stbi__get32be(s) != 0x38425053) // "8BPS" + return stbi__errpuc("not PSD", "Corrupt PSD image"); + + // Check file type version. + if (stbi__get16be(s) != 1) + return stbi__errpuc("wrong version", "Unsupported version of PSD image"); + + // Skip 6 reserved bytes. + stbi__skip(s, 6 ); + + // Read the number of channels (R, G, B, A, etc). + channelCount = stbi__get16be(s); + if (channelCount < 0 || channelCount > 16) + return stbi__errpuc("wrong channel count", "Unsupported number of channels in PSD image"); + + // Read the rows and columns of the image. + h = stbi__get32be(s); + w = stbi__get32be(s); + + // Make sure the depth is 8 bits. + if (stbi__get16be(s) != 8) + return stbi__errpuc("unsupported bit depth", "PSD bit depth is not 8 bit"); + + // Make sure the color mode is RGB. + // Valid options are: + // 0: Bitmap + // 1: Grayscale + // 2: Indexed color + // 3: RGB color + // 4: CMYK color + // 7: Multichannel + // 8: Duotone + // 9: Lab color + if (stbi__get16be(s) != 3) + return stbi__errpuc("wrong color format", "PSD is not in RGB color format"); + + // Skip the Mode Data. (It's the palette for indexed color; other info for other modes.) + stbi__skip(s,stbi__get32be(s) ); + + // Skip the image resources. (resolution, pen tool paths, etc) + stbi__skip(s, stbi__get32be(s) ); + + // Skip the reserved data. + stbi__skip(s, stbi__get32be(s) ); + + // Find out if the data is compressed. + // Known values: + // 0: no compression + // 1: RLE compressed + compression = stbi__get16be(s); + if (compression > 1) + return stbi__errpuc("bad compression", "PSD has an unknown compression format"); + + // Create the destination image. + out = (stbi_uc *) stbi__malloc(4 * w*h); + if (!out) return stbi__errpuc("outofmem", "Out of memory"); + pixelCount = w*h; + + // Initialize the data to zero. + //memset( out, 0, pixelCount * 4 ); + + // Finally, the image data. + if (compression) { + // RLE as used by .PSD and .TIFF + // Loop until you get the number of unpacked bytes you are expecting: + // Read the next source byte into n. + // If n is between 0 and 127 inclusive, copy the next n+1 bytes literally. + // Else if n is between -127 and -1 inclusive, copy the next byte -n+1 times. + // 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, + // which we're going to just skip. + stbi__skip(s, h * channelCount * 2 ); + + // Read the RLE data by channel. + for (channel = 0; channel < 4; channel++) { + stbi_uc *p; + + p = out+channel; + if (channel >= channelCount) { + // Fill this channel with default data. + for (i = 0; i < pixelCount; i++, p += 4) + *p = (channel == 3 ? 255 : 0); + } else { + // Read the RLE data. + count = 0; + while (count < pixelCount) { + len = stbi__get8(s); + if (len == 128) { + // No-op. + } else if (len < 128) { + // Copy next len+1 bytes literally. + len++; + count += len; + while (len) { + *p = stbi__get8(s); + p += 4; + len--; + } + } else if (len > 128) { + stbi_uc val; + // Next -len+1 bytes in the dest are replicated from next source byte. + // (Interpret len as a negative 8-bit int.) + len ^= 0x0FF; + len += 2; + val = stbi__get8(s); + count += len; + while (len) { + *p = val; + p += 4; + len--; + } + } + } + } + } + + } else { + // We're at the raw image data. It's each channel in order (Red, Green, Blue, Alpha, ...) + // where each channel consists of an 8-bit value for each pixel in the image. + + // Read the data by channel. + for (channel = 0; channel < 4; channel++) { + stbi_uc *p; + + p = out + channel; + if (channel > channelCount) { + // Fill this channel with default data. + for (i = 0; i < pixelCount; i++, p += 4) + *p = channel == 3 ? 255 : 0; + } else { + // Read the data. + for (i = 0; i < pixelCount; i++, p += 4) + *p = stbi__get8(s); + } + } + } + + if (req_comp && req_comp != 4) { + out = stbi__convert_format(out, 4, req_comp, w, h); + if (out == NULL) return out; // stbi__convert_format frees input on failure + } + + if (comp) *comp = 4; + *y = h; + *x = w; + + return out; +} #endif + +// ************************************************************************************************* +// Softimage PIC loader +// by Tom Seddon +// +// See http://softimage.wiki.softimage.com/index.php/INFO:_PIC_file_format +// See http://ozviz.wasp.uwa.edu.au/~pbourke/dataformats/softimagepic/ + +#ifndef STBI_NO_PIC +static int stbi__pic_is4(stbi__context *s,const char *str) +{ + int i; + for (i=0; i<4; ++i) + if (stbi__get8(s) != (stbi_uc)str[i]) + return 0; + + return 1; +} + +static int stbi__pic_test_core(stbi__context *s) +{ + int i; + + if (!stbi__pic_is4(s,"\x53\x80\xF6\x34")) + return 0; + + for(i=0;i<84;++i) + stbi__get8(s); + + if (!stbi__pic_is4(s,"PICT")) + return 0; + + return 1; +} + +typedef struct +{ + stbi_uc size,type,channel; +} stbi__pic_packet; + +static stbi_uc *stbi__readval(stbi__context *s, int channel, stbi_uc *dest) +{ + int mask=0x80, i; + + for (i=0; i<4; ++i, mask>>=1) { + if (channel & mask) { + if (stbi__at_eof(s)) return stbi__errpuc("bad file","PIC file too short"); + dest[i]=stbi__get8(s); + } + } + + return dest; +} + +static void stbi__copyval(int channel,stbi_uc *dest,const stbi_uc *src) +{ + int mask=0x80,i; + + for (i=0;i<4; ++i, mask>>=1) + if (channel&mask) + dest[i]=src[i]; +} + +static stbi_uc *stbi__pic_load_core(stbi__context *s,int width,int height,int *comp, stbi_uc *result) +{ + int act_comp=0,num_packets=0,y,chained; + stbi__pic_packet packets[10]; + + // this will (should...) cater for even some bizarre stuff like having data + // for the same channel in multiple packets. + do { + stbi__pic_packet *packet; + + if (num_packets==sizeof(packets)/sizeof(packets[0])) + return stbi__errpuc("bad format","too many packets"); + + packet = &packets[num_packets++]; + + chained = stbi__get8(s); + packet->size = stbi__get8(s); + packet->type = stbi__get8(s); + packet->channel = stbi__get8(s); + + act_comp |= packet->channel; + + if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (reading packets)"); + if (packet->size != 8) return stbi__errpuc("bad format","packet isn't 8bpp"); + } while (chained); + + *comp = (act_comp & 0x10 ? 4 : 3); // has alpha channel? + + for(y=0; ytype) { + default: + return stbi__errpuc("bad format","packet has bad compression type"); + + case 0: {//uncompressed + int x; + + for(x=0;xchannel,dest)) + return 0; + break; + } + + case 1://Pure RLE + { + int left=width, i; + + while (left>0) { + stbi_uc count,value[4]; + + count=stbi__get8(s); + if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (pure read count)"); + + if (count > left) + count = (stbi_uc) left; + + if (!stbi__readval(s,packet->channel,value)) return 0; + + for(i=0; ichannel,dest,value); + left -= count; + } + } + break; + + case 2: {//Mixed RLE + int left=width; + while (left>0) { + int count = stbi__get8(s), i; + if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (mixed read count)"); + + if (count >= 128) { // Repeated + stbi_uc value[4]; + int i; + + if (count==128) + count = stbi__get16be(s); + else + count -= 127; + if (count > left) + return stbi__errpuc("bad file","scanline overrun"); + + if (!stbi__readval(s,packet->channel,value)) + return 0; + + for(i=0;ichannel,dest,value); + } else { // Raw + ++count; + if (count>left) return stbi__errpuc("bad file","scanline overrun"); + + for(i=0;ichannel,dest)) + return 0; + } + left-=count; + } + break; + } + } + } + } + + return result; +} + +static stbi_uc *stbi__pic_load(stbi__context *s,int *px,int *py,int *comp,int req_comp) +{ + stbi_uc *result; + int i, x,y; + + for (i=0; i<92; ++i) + stbi__get8(s); + + x = stbi__get16be(s); + y = stbi__get16be(s); + if (stbi__at_eof(s)) return stbi__errpuc("bad file","file too short (pic header)"); + if ((1 << 28) / x < y) return stbi__errpuc("too large", "Image too large to decode"); + + stbi__get32be(s); //skip `ratio' + stbi__get16be(s); //skip `fields' + stbi__get16be(s); //skip `pad' + + // intermediate buffer is RGBA + result = (stbi_uc *) stbi__malloc(x*y*4); + memset(result, 0xff, x*y*4); + + if (!stbi__pic_load_core(s,x,y,comp, result)) { + STBI_FREE(result); + result=0; + } + *px = x; + *py = y; + if (req_comp == 0) req_comp = *comp; + result=stbi__convert_format(result,4,req_comp,x,y); + + return result; +} + +static int stbi__pic_test(stbi__context *s) +{ + int r = stbi__pic_test_core(s); + stbi__rewind(s); + return r; +} +#endif + +// ************************************************************************************************* +// GIF loader -- public domain by Jean-Marc Lienher -- simplified/shrunk by stb + +#ifndef STBI_NO_GIF +typedef struct +{ + stbi__int16 prefix; + stbi_uc first; + stbi_uc suffix; +} stbi__gif_lzw; + +typedef struct +{ + int w,h; + stbi_uc *out; // output buffer (always 4 components) + int flags, bgindex, ratio, transparent, eflags; + stbi_uc pal[256][4]; + stbi_uc lpal[256][4]; + stbi__gif_lzw codes[4096]; + stbi_uc *color_table; + int parse, step; + int lflags; + int start_x, start_y; + int max_x, max_y; + int cur_x, cur_y; + int line_size; +} stbi__gif; + +static int stbi__gif_test_raw(stbi__context *s) +{ + int sz; + if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') return 0; + sz = stbi__get8(s); + if (sz != '9' && sz != '7') return 0; + if (stbi__get8(s) != 'a') return 0; + return 1; +} + +static int stbi__gif_test(stbi__context *s) +{ + int r = stbi__gif_test_raw(s); + stbi__rewind(s); + return r; +} + +static void stbi__gif_parse_colortable(stbi__context *s, stbi_uc pal[256][4], int num_entries, int transp) +{ + int i; + for (i=0; i < num_entries; ++i) { + pal[i][2] = stbi__get8(s); + pal[i][1] = stbi__get8(s); + pal[i][0] = stbi__get8(s); + pal[i][3] = transp == i ? 0 : 255; + } +} + +static int stbi__gif_header(stbi__context *s, stbi__gif *g, int *comp, int is_info) +{ + stbi_uc version; + if (stbi__get8(s) != 'G' || stbi__get8(s) != 'I' || stbi__get8(s) != 'F' || stbi__get8(s) != '8') + return stbi__err("not GIF", "Corrupt GIF"); + + version = stbi__get8(s); + if (version != '7' && version != '9') return stbi__err("not GIF", "Corrupt GIF"); + if (stbi__get8(s) != 'a') return stbi__err("not GIF", "Corrupt GIF"); + + stbi__g_failure_reason = ""; + g->w = stbi__get16le(s); + g->h = stbi__get16le(s); + g->flags = stbi__get8(s); + g->bgindex = stbi__get8(s); + g->ratio = stbi__get8(s); + g->transparent = -1; + + 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; + + if (g->flags & 0x80) + stbi__gif_parse_colortable(s,g->pal, 2 << (g->flags & 7), -1); + + return 1; +} + +static int stbi__gif_info_raw(stbi__context *s, int *x, int *y, int *comp) +{ + stbi__gif g; + if (!stbi__gif_header(s, &g, comp, 1)) { + stbi__rewind( s ); + return 0; + } + if (x) *x = g.w; + if (y) *y = g.h; + return 1; +} + +static void stbi__out_gif_code(stbi__gif *g, stbi__uint16 code) +{ + stbi_uc *p, *c; + + // recurse to decode the prefixes, since the linked-list is backwards, + // and working backwards through an interleaved image would be nasty + if (g->codes[code].prefix >= 0) + stbi__out_gif_code(g, g->codes[code].prefix); + + 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]; + + if (c[3] >= 128) { + p[0] = c[2]; + p[1] = c[1]; + p[2] = c[0]; + p[3] = c[3]; + } + g->cur_x += 4; + + if (g->cur_x >= g->max_x) { + g->cur_x = g->start_x; + g->cur_y += g->step; + + while (g->cur_y >= g->max_y && g->parse > 0) { + g->step = (1 << g->parse) * g->line_size; + g->cur_y = g->start_y + (g->step >> 1); + --g->parse; + } + } +} + +static stbi_uc *stbi__process_gif_raster(stbi__context *s, stbi__gif *g) +{ + stbi_uc lzw_cs; + stbi__int32 len, code; + stbi__uint32 first; + stbi__int32 codesize, codemask, avail, oldcode, bits, valid_bits, clear; + stbi__gif_lzw *p; + + lzw_cs = stbi__get8(s); + if (lzw_cs > 12) return NULL; + clear = 1 << lzw_cs; + first = 1; + codesize = lzw_cs + 1; + codemask = (1 << codesize) - 1; + bits = 0; + valid_bits = 0; + for (code = 0; code < clear; code++) { + g->codes[code].prefix = -1; + g->codes[code].first = (stbi_uc) code; + g->codes[code].suffix = (stbi_uc) code; + } + + // support no starting clear code + avail = clear+2; + oldcode = -1; + + len = 0; + for(;;) { + if (valid_bits < codesize) { + if (len == 0) { + len = stbi__get8(s); // start new block + if (len == 0) + return g->out; + } + --len; + bits |= (stbi__int32) stbi__get8(s) << valid_bits; + valid_bits += 8; + } else { + stbi__int32 code = bits & codemask; + bits >>= codesize; + valid_bits -= codesize; + // @OPTIMIZE: is there some way we can accelerate the non-clear path? + if (code == clear) { // clear code + codesize = lzw_cs + 1; + codemask = (1 << codesize) - 1; + avail = clear + 2; + oldcode = -1; + first = 0; + } else if (code == clear + 1) { // end of stream code + stbi__skip(s, len); + while ((len = stbi__get8(s)) > 0) + stbi__skip(s,len); + return g->out; + } else if (code <= avail) { + 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"); + p->prefix = (stbi__int16) oldcode; + p->first = g->codes[oldcode].first; + p->suffix = (code == avail) ? p->first : g->codes[code].first; + } else if (code == avail) + return stbi__errpuc("illegal code in raster", "Corrupt GIF"); + + stbi__out_gif_code(g, (stbi__uint16) code); + + if ((avail & codemask) == 0 && avail <= 0x0FFF) { + codesize++; + codemask = (1 << codesize) - 1; + } + + oldcode = code; + } else { + return stbi__errpuc("illegal code in raster", "Corrupt GIF"); + } + } + } +} + +static void stbi__fill_gif_background(stbi__gif *g) +{ + int i; + stbi_uc *c = g->pal[g->bgindex]; + // @OPTIMIZE: write a dword at a time + for (i = 0; i < g->w * g->h * 4; i += 4) { + stbi_uc *p = &g->out[i]; + p[0] = c[2]; + p[1] = c[1]; + p[2] = c[0]; + p[3] = c[3]; + } +} + +// 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) +{ + int i; + stbi_uc *old_out = 0; + + if (g->out == 0) { + if (!stbi__gif_header(s, g, comp,0)) return 0; // stbi__g_failure_reason set by stbi__gif_header + g->out = (stbi_uc *) stbi__malloc(4 * g->w * g->h); + if (g->out == 0) return stbi__errpuc("outofmem", "Out of memory"); + stbi__fill_gif_background(g); + } else { + // animated-gif-only path + if (((g->eflags & 0x1C) >> 2) == 3) { + old_out = g->out; + g->out = (stbi_uc *) stbi__malloc(4 * g->w * g->h); + if (g->out == 0) return stbi__errpuc("outofmem", "Out of memory"); + memcpy(g->out, old_out, g->w*g->h*4); + } + } + + for (;;) { + switch (stbi__get8(s)) { + case 0x2C: /* Image Descriptor */ + { + stbi__int32 x, y, w, h; + stbi_uc *o; + + x = stbi__get16le(s); + y = stbi__get16le(s); + w = stbi__get16le(s); + h = stbi__get16le(s); + if (((x + w) > (g->w)) || ((y + h) > (g->h))) + return stbi__errpuc("bad Image Descriptor", "Corrupt GIF"); + + g->line_size = g->w * 4; + g->start_x = x * 4; + g->start_y = y * g->line_size; + g->max_x = g->start_x + w * 4; + g->max_y = g->start_y + h * g->line_size; + g->cur_x = g->start_x; + g->cur_y = g->start_y; + + g->lflags = stbi__get8(s); + + if (g->lflags & 0x40) { + g->step = 8 * g->line_size; // first interlaced spacing + g->parse = 3; + } else { + g->step = g->line_size; + g->parse = 0; + } + + if (g->lflags & 0x80) { + 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) { + for (i=0; i < 256; ++i) // @OPTIMIZE: stbi__jpeg_reset only the previous transparent + g->pal[i][3] = 255; + if (g->transparent >= 0 && (g->eflags & 0x01)) + 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 (req_comp && req_comp != 4) + o = stbi__convert_format(o, 4, req_comp, g->w, g->h); + return o; + } + + case 0x21: // Comment Extension. + { + int len; + if (stbi__get8(s) == 0xF9) { // Graphic Control Extension. + len = stbi__get8(s); + if (len == 4) { + g->eflags = stbi__get8(s); + stbi__get16le(s); // delay + g->transparent = stbi__get8(s); + } else { + stbi__skip(s, len); + break; + } + } + while ((len = stbi__get8(s)) != 0) + stbi__skip(s, len); + break; + } + + case 0x3B: // gif stream termination code + return (stbi_uc *) s; // using '1' causes warning on some compilers + + default: + return stbi__errpuc("unknown code", "Corrupt GIF"); + } + } +} + +static stbi_uc *stbi__gif_load(stbi__context *s, int *x, int *y, int *comp, int req_comp) +{ + stbi_uc *u = 0; + stbi__gif g; + memset(&g, 0, sizeof(g)); + + u = stbi__gif_load_next(s, &g, comp, req_comp); + if (u == (stbi_uc *) s) u = 0; // end of animated gif marker + if (u) { + *x = g.w; + *y = g.h; + } + + return u; +} + +static int stbi__gif_info(stbi__context *s, int *x, int *y, int *comp) +{ + return stbi__gif_info_raw(s,x,y,comp); +} +#endif + +// ************************************************************************************************* +// Radiance RGBE HDR loader +// originally by Nicolas Schulz +#ifndef STBI_NO_HDR +static int stbi__hdr_test_core(stbi__context *s) +{ + const char *signature = "#?RADIANCE\n"; + int i; + for (i=0; signature[i]; ++i) + if (stbi__get8(s) != signature[i]) + return 0; + return 1; +} + +static int stbi__hdr_test(stbi__context* s) +{ + int r = stbi__hdr_test_core(s); + stbi__rewind(s); + return r; +} + +#define STBI__HDR_BUFLEN 1024 +static char *stbi__hdr_gettoken(stbi__context *z, char *buffer) +{ + int len=0; + char c = '\0'; + + c = (char) stbi__get8(z); + + while (!stbi__at_eof(z) && c != '\n') { + buffer[len++] = c; + if (len == STBI__HDR_BUFLEN-1) { + // flush to end of line + while (!stbi__at_eof(z) && stbi__get8(z) != '\n') + ; + break; + } + c = (char) stbi__get8(z); + } + + buffer[len] = 0; + return buffer; +} + +static void stbi__hdr_convert(float *output, stbi_uc *input, int req_comp) +{ + if ( input[3] != 0 ) { + float f1; + // Exponent + f1 = (float) ldexp(1.0f, input[3] - (int)(128 + 8)); + if (req_comp <= 2) + output[0] = (input[0] + input[1] + input[2]) * f1 / 3; + else { + output[0] = input[0] * f1; + output[1] = input[1] * f1; + output[2] = input[2] * f1; + } + if (req_comp == 2) output[1] = 1; + if (req_comp == 4) output[3] = 1; + } else { + switch (req_comp) { + case 4: output[3] = 1; /* fallthrough */ + case 3: output[0] = output[1] = output[2] = 0; + break; + case 2: output[1] = 1; /* fallthrough */ + case 1: output[0] = 0; + break; + } + } +} + +static float *stbi__hdr_load(stbi__context *s, int *x, int *y, int *comp, int req_comp) +{ + char buffer[STBI__HDR_BUFLEN]; + char *token; + int valid = 0; + int width, height; + stbi_uc *scanline; + float *hdr_data; + int len; + unsigned char count, value; + int i, j, k, c1,c2, z; + + + // Check identifier + if (strcmp(stbi__hdr_gettoken(s,buffer), "#?RADIANCE") != 0) + return stbi__errpf("not HDR", "Corrupt HDR image"); + + // Parse header + for(;;) { + token = stbi__hdr_gettoken(s,buffer); + if (token[0] == 0) break; + if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) valid = 1; + } + + if (!valid) return stbi__errpf("unsupported format", "Unsupported HDR format"); + + // Parse width and height + // can't use sscanf() if we're not using stdio! + token = stbi__hdr_gettoken(s,buffer); + if (strncmp(token, "-Y ", 3)) return stbi__errpf("unsupported data layout", "Unsupported HDR format"); + token += 3; + height = (int) strtol(token, &token, 10); + while (*token == ' ') ++token; + if (strncmp(token, "+X ", 3)) return stbi__errpf("unsupported data layout", "Unsupported HDR format"); + token += 3; + width = (int) strtol(token, NULL, 10); + + *x = width; + *y = height; + + if (comp) *comp = 3; + if (req_comp == 0) req_comp = 3; + + // Read data + hdr_data = (float *) stbi__malloc(height * width * req_comp * sizeof(float)); + + // Load image data + // image data is stored as some number of sca + if ( width < 8 || width >= 32768) { + // Read flat data + for (j=0; j < height; ++j) { + for (i=0; i < width; ++i) { + stbi_uc rgbe[4]; + main_decode_loop: + stbi__getn(s, rgbe, 4); + stbi__hdr_convert(hdr_data + j * width * req_comp + i * req_comp, rgbe, req_comp); + } + } + } else { + // Read RLE-encoded data + scanline = NULL; + + for (j = 0; j < height; ++j) { + c1 = stbi__get8(s); + c2 = stbi__get8(s); + len = stbi__get8(s); + if (c1 != 2 || c2 != 2 || (len & 0x80)) { + // not run-length encoded, so we have to actually use THIS data as a decoded + // pixel (note this can't be a valid pixel--one of RGB must be >= 128) + stbi_uc rgbe[4]; + rgbe[0] = (stbi_uc) c1; + rgbe[1] = (stbi_uc) c2; + rgbe[2] = (stbi_uc) len; + rgbe[3] = (stbi_uc) stbi__get8(s); + stbi__hdr_convert(hdr_data, rgbe, req_comp); + i = 1; + j = 0; + STBI_FREE(scanline); + goto main_decode_loop; // yes, this makes no sense + } + len <<= 8; + len |= stbi__get8(s); + if (len != width) { STBI_FREE(hdr_data); STBI_FREE(scanline); return stbi__errpf("invalid decoded scanline length", "corrupt HDR"); } + if (scanline == NULL) scanline = (stbi_uc *) stbi__malloc(width * 4); + + for (k = 0; k < 4; ++k) { + i = 0; + while (i < width) { + count = stbi__get8(s); + if (count > 128) { + // Run + value = stbi__get8(s); + count -= 128; + for (z = 0; z < count; ++z) + scanline[i++ * 4 + k] = value; + } else { + // Dump + for (z = 0; z < count; ++z) + scanline[i++ * 4 + k] = stbi__get8(s); + } + } + } + for (i=0; i < width; ++i) + stbi__hdr_convert(hdr_data+(j*width + i)*req_comp, scanline + i*4, req_comp); + } + STBI_FREE(scanline); + } + + return hdr_data; +} + +static int stbi__hdr_info(stbi__context *s, int *x, int *y, int *comp) +{ + char buffer[STBI__HDR_BUFLEN]; + char *token; + int valid = 0; + + if (strcmp(stbi__hdr_gettoken(s,buffer), "#?RADIANCE") != 0) { + stbi__rewind( s ); + return 0; + } + + for(;;) { + token = stbi__hdr_gettoken(s,buffer); + if (token[0] == 0) break; + if (strcmp(token, "FORMAT=32-bit_rle_rgbe") == 0) valid = 1; + } + + if (!valid) { + stbi__rewind( s ); + return 0; + } + token = stbi__hdr_gettoken(s,buffer); + if (strncmp(token, "-Y ", 3)) { + stbi__rewind( s ); + return 0; + } + token += 3; + *y = (int) strtol(token, &token, 10); + while (*token == ' ') ++token; + if (strncmp(token, "+X ", 3)) { + stbi__rewind( s ); + return 0; + } + token += 3; + *x = (int) strtol(token, NULL, 10); + *comp = 3; + return 1; +} +#endif // STBI_NO_HDR + +#ifndef STBI_NO_BMP +static int stbi__bmp_info(stbi__context *s, int *x, int *y, int *comp) +{ + int hsz; + if (stbi__get8(s) != 'B' || stbi__get8(s) != 'M') { + stbi__rewind( s ); + return 0; + } + stbi__skip(s,12); + hsz = stbi__get32le(s); + if (hsz != 12 && hsz != 40 && hsz != 56 && hsz != 108 && hsz != 124) { + stbi__rewind( s ); + return 0; + } + if (hsz == 12) { + *x = stbi__get16le(s); + *y = stbi__get16le(s); + } else { + *x = stbi__get32le(s); + *y = stbi__get32le(s); + } + if (stbi__get16le(s) != 1) { + stbi__rewind( s ); + return 0; + } + *comp = stbi__get16le(s) / 8; + return 1; +} +#endif + +#ifndef STBI_NO_PSD +static int stbi__psd_info(stbi__context *s, int *x, int *y, int *comp) +{ + int channelCount; + 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; + } + *y = stbi__get32be(s); + *x = stbi__get32be(s); + if (stbi__get16be(s) != 8) { + stbi__rewind( s ); + return 0; + } + if (stbi__get16be(s) != 3) { + stbi__rewind( s ); + return 0; + } + *comp = 4; + return 1; +} +#endif + +#ifndef STBI_NO_PIC +static int stbi__pic_info(stbi__context *s, int *x, int *y, int *comp) +{ + int act_comp=0,num_packets=0,chained; + stbi__pic_packet packets[10]; + + stbi__skip(s, 92); + + *x = stbi__get16be(s); + *y = stbi__get16be(s); + if (stbi__at_eof(s)) return 0; + if ( (*x) != 0 && (1 << 28) / (*x) < (*y)) { + stbi__rewind( s ); + return 0; + } + + stbi__skip(s, 8); + + do { + stbi__pic_packet *packet; + + if (num_packets==sizeof(packets)/sizeof(packets[0])) + return 0; + + packet = &packets[num_packets++]; + chained = stbi__get8(s); + packet->size = stbi__get8(s); + packet->type = stbi__get8(s); + packet->channel = stbi__get8(s); + act_comp |= packet->channel; + + if (stbi__at_eof(s)) { + stbi__rewind( s ); + return 0; + } + if (packet->size != 8) { + stbi__rewind( s ); + return 0; + } + } while (chained); + + *comp = (act_comp & 0x10 ? 4 : 3); + + return 1; +} +#endif + +// ************************************************************************************************* +// Portable Gray Map and Portable Pixel Map loader +// by Ken Miller +// +// PGM: http://netpbm.sourceforge.net/doc/pgm.html +// PPM: http://netpbm.sourceforge.net/doc/ppm.html +// +// 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 + +static int stbi__pnm_test(stbi__context *s) +{ + char p, t; + p = (char) stbi__get8(s); + t = (char) stbi__get8(s); + if (p != 'P' || (t != '5' && t != '6')) { + stbi__rewind( s ); + return 0; + } + return 1; +} + +static stbi_uc *stbi__pnm_load(stbi__context *s, int *x, int *y, int *comp, int req_comp) +{ + stbi_uc *out; + if (!stbi__pnm_info(s, (int *)&s->img_x, (int *)&s->img_y, (int *)&s->img_n)) + return 0; + *x = s->img_x; + *y = s->img_y; + *comp = s->img_n; + + out = (stbi_uc *) stbi__malloc(s->img_n * s->img_x * s->img_y); + if (!out) return stbi__errpuc("outofmem", "Out of memory"); + stbi__getn(s, out, s->img_n * s->img_x * s->img_y); + + if (req_comp && req_comp != s->img_n) { + out = stbi__convert_format(out, s->img_n, req_comp, s->img_x, s->img_y); + if (out == NULL) return out; // stbi__convert_format frees input on failure + } + return out; +} + +static int stbi__pnm_isspace(char c) +{ + return c == ' ' || c == '\t' || c == '\n' || c == '\v' || c == '\f' || c == '\r'; +} + +static void stbi__pnm_skip_whitespace(stbi__context *s, char *c) +{ + while (!stbi__at_eof(s) && stbi__pnm_isspace(*c)) + *c = (char) stbi__get8(s); +} + +static int stbi__pnm_isdigit(char c) +{ + return c >= '0' && c <= '9'; +} + +static int stbi__pnm_getinteger(stbi__context *s, char *c) +{ + int value = 0; + + while (!stbi__at_eof(s) && stbi__pnm_isdigit(*c)) { + value = value*10 + (*c - '0'); + *c = (char) stbi__get8(s); + } + + return value; +} + +static int stbi__pnm_info(stbi__context *s, int *x, int *y, int *comp) +{ + int maxv; + char c, p, t; + + stbi__rewind( s ); + + // Get identifier + p = (char) stbi__get8(s); + t = (char) stbi__get8(s); + if (p != 'P' || (t != '5' && t != '6')) { + stbi__rewind( s ); + return 0; + } + + *comp = (t == '6') ? 3 : 1; // '5' is 1-component .pgm; '6' is 3-component .ppm + + c = (char) stbi__get8(s); + stbi__pnm_skip_whitespace(s, &c); + + *x = stbi__pnm_getinteger(s, &c); // read width + stbi__pnm_skip_whitespace(s, &c); + + *y = stbi__pnm_getinteger(s, &c); // read height + 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"); + else + return 1; +} +#endif + +static int stbi__info_main(stbi__context *s, int *x, int *y, int *comp) +{ + #ifndef STBI_NO_JPEG + if (stbi__jpeg_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_PNG + if (stbi__png_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_GIF + if (stbi__gif_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_BMP + if (stbi__bmp_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_PSD + if (stbi__psd_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_PIC + if (stbi__pic_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_PNM + if (stbi__pnm_info(s, x, y, comp)) return 1; + #endif + + #ifndef STBI_NO_HDR + if (stbi__hdr_info(s, x, y, comp)) return 1; + #endif + + // test tga last because it's a crappy test! + #ifndef STBI_NO_TGA + if (stbi__tga_info(s, x, y, comp)) + return 1; + #endif + return stbi__err("unknown image type", "Image not of any known type, or corrupt"); +} + +#ifndef STBI_NO_STDIO +STBIDEF int stbi_info(char const *filename, int *x, int *y, int *comp) +{ + FILE *f = stbi__fopen(filename, "rb"); + int result; + if (!f) return stbi__err("can't fopen", "Unable to open file"); + result = stbi_info_from_file(f, x, y, comp); + fclose(f); + return result; +} + +STBIDEF int stbi_info_from_file(FILE *f, int *x, int *y, int *comp) +{ + int r; + stbi__context s; + long pos = ftell(f); + stbi__start_file(&s, f); + r = stbi__info_main(&s,x,y,comp); + 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) +{ + stbi__context s; + stbi__start_mem(&s,buffer,len); + return stbi__info_main(&s,x,y,comp); +} + +STBIDEF int stbi_info_from_callbacks(stbi_io_callbacks const *c, void *user, int *x, int *y, int *comp) +{ + stbi__context s; + stbi__start_callbacks(&s, (stbi_io_callbacks *) c, user); + return stbi__info_main(&s,x,y,comp); +} + +#endif // STB_IMAGE_IMPLEMENTATION + +/* + revision history: + 2.06 (2015-04-19) fix bug where PSD returns wrong '*comp' value + 2.05 (2015-04-19) fix bug in progressive JPEG handling, fix warning + 2.04 (2015-04-15) try to re-enable SIMD on MinGW 64-bit + 2.03 (2015-04-12) extra corruption checking (mmozeiko) + stbi_set_flip_vertically_on_load (nguillemot) + fix NEON support; fix mingw support + 2.02 (2015-01-19) fix incorrect assert, fix warning + 2.01 (2015-01-17) fix various warnings; suppress SIMD on gcc 32-bit without -msse2 + 2.00b (2014-12-25) fix STBI_MALLOC in progressive JPEG + 2.00 (2014-12-25) optimize JPG, including x86 SSE2 & NEON SIMD (ryg) + progressive JPEG (stb) + PGM/PPM support (Ken Miller) + STBI_MALLOC,STBI_REALLOC,STBI_FREE + GIF bugfix -- seemingly never worked + STBI_NO_*, STBI_ONLY_* + 1.48 (2014-12-14) fix incorrectly-named assert() + 1.47 (2014-12-14) 1/2/4-bit PNG support, both direct and paletted (Omar Cornut & stb) + optimize PNG (ryg) + fix bug in interlaced PNG with user-specified channel count (stb) + 1.46 (2014-08-26) + fix broken tRNS chunk (colorkey-style transparency) in non-paletted PNG + 1.45 (2014-08-16) + fix MSVC-ARM internal compiler error by wrapping malloc + 1.44 (2014-08-07) + various warning fixes from Ronny Chevalier + 1.43 (2014-07-15) + fix MSVC-only compiler problem in code changed in 1.42 + 1.42 (2014-07-09) + don't define _CRT_SECURE_NO_WARNINGS (affects user code) + fixes to stbi__cleanup_jpeg path + added STBI_ASSERT to avoid requiring assert.h + 1.41 (2014-06-25) + fix search&replace from 1.36 that messed up comments/error messages + 1.40 (2014-06-22) + fix gcc struct-initialization warning + 1.39 (2014-06-15) + fix to TGA optimization when req_comp != number of components in TGA; + fix to GIF loading because BMP wasn't rewinding (whoops, no GIFs in my test suite) + add support for BMP version 5 (more ignored fields) + 1.38 (2014-06-06) + suppress MSVC warnings on integer casts truncating values + fix accidental rename of 'skip' field of I/O + 1.37 (2014-06-04) + remove duplicate typedef + 1.36 (2014-06-03) + convert to header file single-file library + if de-iphone isn't set, load iphone images color-swapped instead of returning NULL + 1.35 (2014-05-27) + various warnings + fix broken STBI_SIMD path + fix bug where stbi_load_from_file no longer left file pointer in correct place + fix broken non-easy path for 32-bit BMP (possibly never used) + TGA optimization by Arseny Kapoulkine + 1.34 (unknown) + use STBI_NOTUSED in stbi__resample_row_generic(), fix one more leak in tga failure case + 1.33 (2011-07-14) + make stbi_is_hdr work in STBI_NO_HDR (as specified), minor compiler-friendly improvements + 1.32 (2011-07-13) + support for "info" function for all supported filetypes (SpartanJ) + 1.31 (2011-06-20) + a few more leak fixes, bug in PNG handling (SpartanJ) + 1.30 (2011-06-11) + added ability to load files via callbacks to accomidate custom input streams (Ben Wenger) + removed deprecated format-specific test/load functions + removed support for installable file formats (stbi_loader) -- would have been broken for IO callbacks anyway + error cases in bmp and tga give messages and don't leak (Raymond Barbiero, grisha) + fix inefficiency in decoding 32-bit BMP (David Woo) + 1.29 (2010-08-16) + various warning fixes from Aurelien Pocheville + 1.28 (2010-08-01) + fix bug in GIF palette transparency (SpartanJ) + 1.27 (2010-08-01) + cast-to-stbi_uc to fix warnings + 1.26 (2010-07-24) + fix bug in file buffering for PNG reported by SpartanJ + 1.25 (2010-07-17) + refix trans_data warning (Won Chun) + 1.24 (2010-07-12) + perf improvements reading from files on platforms with lock-heavy fgetc() + minor perf improvements for jpeg + deprecated type-specific functions so we'll get feedback if they're needed + attempt to fix trans_data warning (Won Chun) + 1.23 fixed bug in iPhone support + 1.22 (2010-07-10) + removed image *writing* support + stbi_info support from Jetro Lauha + GIF support from Jean-Marc Lienher + iPhone PNG-extensions from James Brown + warning-fixes from Nicolas Schulz and Janez Zemva (i.stbi__err. Janez (U+017D)emva) + 1.21 fix use of 'stbi_uc' in header (reported by jon blow) + 1.20 added support for Softimage PIC, by Tom Seddon + 1.19 bug in interlaced PNG corruption check (found by ryg) + 1.18 (2008-08-02) + fix a threading bug (local mutable static) + 1.17 support interlaced PNG + 1.16 major bugfix - stbi__convert_format converted one too many pixels + 1.15 initialize some fields for thread safety + 1.14 fix threadsafe conversion bug + header-file-only version (#define STBI_HEADER_FILE_ONLY before including) + 1.13 threadsafe + 1.12 const qualifiers in the API + 1.11 Support installable IDCT, colorspace conversion routines + 1.10 Fixes for 64-bit (don't use "unsigned long") + optimized upsampling by Fabian "ryg" Giesen + 1.09 Fix format-conversion for PSD code (bad global variables!) + 1.08 Thatcher Ulrich's PSD code integrated by Nicolas Schulz + 1.07 attempt to fix C++ warning/errors again + 1.06 attempt to fix C++ warning/errors again + 1.05 fix TGA loading to return correct *comp and use good luminance calc + 1.04 default float alpha is 1, not 255; use 'void *' for stbi_image_free + 1.03 bugfixes to STBI_NO_STDIO, STBI_NO_HDR + 1.02 support for (subset of) HDR files, float interface for preferred access to them + 1.01 fix bug: possible bug in handling right-side up bmps... not sure + fix bug: the stbi__bmp_load() and stbi__tga_load() functions didn't work at all + 1.00 interface to zlib that skips zlib header + 0.99 correct handling of alpha in palette + 0.98 TGA loader by lonesock; dynamically add loaders (untested) + 0.97 jpeg errors on too large a file; also catch another malloc failure + 0.96 fix detection of invalid v value - particleman@mollyrocket forum + 0.95 during header scan, seek to markers in case of padding + 0.94 STBI_NO_STDIO to disable stdio usage; rename all #defines the same + 0.93 handle jpegtran output; verbose errors + 0.92 read 4,8,16,24,32-bit BMP files of several formats + 0.91 output 24-bit Windows 3.0 BMP files + 0.90 fix a few more warnings; bump version number to approach 1.0 + 0.61 bugfixes due to Marc LeBlanc, Christopher Lloyd + 0.60 fix compiling as c++ + 0.59 fix warnings: merge Dave Moore's -Wall fixes + 0.58 fix bug: zlib uncompressed mode len/nlen was wrong endian + 0.57 fix bug: jpg last huffman symbol before marker was >9 bits but less than 16 available + 0.56 fix bug: zlib uncompressed mode len vs. nlen + 0.55 fix bug: restart_interval not initialized to 0 + 0.54 allow NULL for 'int *comp' + 0.53 fix bug in png 3->4; speedup png decoding + 0.52 png handles req_comp=3,4 directly; minor cleanup; jpeg comments + 0.51 obey req_comp requests, 1-component jpegs return as 1-component, + on 'test' only check type, not whether we support this variant + 0.50 (2006-11-19) + first released version +*/