Line |
Branch |
Exec |
Source |
1 |
|
|
// Ogg Vorbis audio decoder - v1.22 - public domain |
2 |
|
|
// http://nothings.org/stb_vorbis/ |
3 |
|
|
// |
4 |
|
|
// Original version written by Sean Barrett in 2007. |
5 |
|
|
// |
6 |
|
|
// Originally sponsored by RAD Game Tools. Seeking implementation |
7 |
|
|
// sponsored by Phillip Bennefall, Marc Andersen, Aaron Baker, |
8 |
|
|
// Elias Software, Aras Pranckevicius, and Sean Barrett. |
9 |
|
|
// |
10 |
|
|
// LICENSE |
11 |
|
|
// |
12 |
|
|
// See end of file for license information. |
13 |
|
|
// |
14 |
|
|
// Limitations: |
15 |
|
|
// |
16 |
|
|
// - floor 0 not supported (used in old ogg vorbis files pre-2004) |
17 |
|
|
// - lossless sample-truncation at beginning ignored |
18 |
|
|
// - cannot concatenate multiple vorbis streams |
19 |
|
|
// - sample positions are 32-bit, limiting seekable 192Khz |
20 |
|
|
// files to around 6 hours (Ogg supports 64-bit) |
21 |
|
|
// |
22 |
|
|
// Feature contributors: |
23 |
|
|
// Dougall Johnson (sample-exact seeking) |
24 |
|
|
// |
25 |
|
|
// Bugfix/warning contributors: |
26 |
|
|
// Terje Mathisen Niklas Frykholm Andy Hill |
27 |
|
|
// Casey Muratori John Bolton Gargaj |
28 |
|
|
// Laurent Gomila Marc LeBlanc Ronny Chevalier |
29 |
|
|
// Bernhard Wodo Evan Balster github:alxprd |
30 |
|
|
// Tom Beaumont Ingo Leitgeb Nicolas Guillemot |
31 |
|
|
// Phillip Bennefall Rohit Thiago Goulart |
32 |
|
|
// github:manxorist Saga Musix github:infatum |
33 |
|
|
// Timur Gagiev Maxwell Koo Peter Waller |
34 |
|
|
// github:audinowho Dougall Johnson David Reid |
35 |
|
|
// github:Clownacy Pedro J. Estebanez Remi Verschelde |
36 |
|
|
// AnthoFoxo github:morlat Gabriel Ravier |
37 |
|
|
// |
38 |
|
|
// Partial history: |
39 |
|
|
// 1.22 - 2021-07-11 - various small fixes |
40 |
|
|
// 1.21 - 2021-07-02 - fix bug for files with no comments |
41 |
|
|
// 1.20 - 2020-07-11 - several small fixes |
42 |
|
|
// 1.19 - 2020-02-05 - warnings |
43 |
|
|
// 1.18 - 2020-02-02 - fix seek bugs; parse header comments; misc warnings etc. |
44 |
|
|
// 1.17 - 2019-07-08 - fix CVE-2019-13217..CVE-2019-13223 (by ForAllSecure) |
45 |
|
|
// 1.16 - 2019-03-04 - fix warnings |
46 |
|
|
// 1.15 - 2019-02-07 - explicit failure if Ogg Skeleton data is found |
47 |
|
|
// 1.14 - 2018-02-11 - delete bogus dealloca usage |
48 |
|
|
// 1.13 - 2018-01-29 - fix truncation of last frame (hopefully) |
49 |
|
|
// 1.12 - 2017-11-21 - limit residue begin/end to blocksize/2 to avoid large temp allocs in bad/corrupt files |
50 |
|
|
// 1.11 - 2017-07-23 - fix MinGW compilation |
51 |
|
|
// 1.10 - 2017-03-03 - more robust seeking; fix negative ilog(); clear error in open_memory |
52 |
|
|
// 1.09 - 2016-04-04 - back out 'truncation of last frame' fix from previous version |
53 |
|
|
// 1.08 - 2016-04-02 - warnings; setup memory leaks; truncation of last frame |
54 |
|
|
// 1.07 - 2015-01-16 - fixes for crashes on invalid files; warning fixes; const |
55 |
|
|
// 1.06 - 2015-08-31 - full, correct support for seeking API (Dougall Johnson) |
56 |
|
|
// some crash fixes when out of memory or with corrupt files |
57 |
|
|
// fix some inappropriately signed shifts |
58 |
|
|
// 1.05 - 2015-04-19 - don't define __forceinline if it's redundant |
59 |
|
|
// 1.04 - 2014-08-27 - fix missing const-correct case in API |
60 |
|
|
// 1.03 - 2014-08-07 - warning fixes |
61 |
|
|
// 1.02 - 2014-07-09 - declare qsort comparison as explicitly _cdecl in Windows |
62 |
|
|
// 1.01 - 2014-06-18 - fix stb_vorbis_get_samples_float (interleaved was correct) |
63 |
|
|
// 1.0 - 2014-05-26 - fix memory leaks; fix warnings; fix bugs in >2-channel; |
64 |
|
|
// (API change) report sample rate for decode-full-file funcs |
65 |
|
|
// |
66 |
|
|
// See end of file for full version history. |
67 |
|
|
|
68 |
|
|
|
69 |
|
|
////////////////////////////////////////////////////////////////////////////// |
70 |
|
|
// |
71 |
|
|
// HEADER BEGINS HERE |
72 |
|
|
// |
73 |
|
|
|
74 |
|
|
#ifndef STB_VORBIS_INCLUDE_STB_VORBIS_H |
75 |
|
|
#define STB_VORBIS_INCLUDE_STB_VORBIS_H |
76 |
|
|
|
77 |
|
|
#if defined(STB_VORBIS_NO_CRT) && !defined(STB_VORBIS_NO_STDIO) |
78 |
|
|
#define STB_VORBIS_NO_STDIO 1 |
79 |
|
|
#endif |
80 |
|
|
|
81 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
82 |
|
|
#include <stdio.h> |
83 |
|
|
#endif |
84 |
|
|
|
85 |
|
|
#ifdef __cplusplus |
86 |
|
|
extern "C" { |
87 |
|
|
#endif |
88 |
|
|
|
89 |
|
|
/////////// THREAD SAFETY |
90 |
|
|
|
91 |
|
|
// Individual stb_vorbis* handles are not thread-safe; you cannot decode from |
92 |
|
|
// them from multiple threads at the same time. However, you can have multiple |
93 |
|
|
// stb_vorbis* handles and decode from them independently in multiple thrads. |
94 |
|
|
|
95 |
|
|
|
96 |
|
|
/////////// MEMORY ALLOCATION |
97 |
|
|
|
98 |
|
|
// normally stb_vorbis uses malloc() to allocate memory at startup, |
99 |
|
|
// and alloca() to allocate temporary memory during a frame on the |
100 |
|
|
// stack. (Memory consumption will depend on the amount of setup |
101 |
|
|
// data in the file and how you set the compile flags for speed |
102 |
|
|
// vs. size. In my test files the maximal-size usage is ~150KB.) |
103 |
|
|
// |
104 |
|
|
// You can modify the wrapper functions in the source (setup_malloc, |
105 |
|
|
// setup_temp_malloc, temp_malloc) to change this behavior, or you |
106 |
|
|
// can use a simpler allocation model: you pass in a buffer from |
107 |
|
|
// which stb_vorbis will allocate _all_ its memory (including the |
108 |
|
|
// temp memory). "open" may fail with a VORBIS_outofmem if you |
109 |
|
|
// do not pass in enough data; there is no way to determine how |
110 |
|
|
// much you do need except to succeed (at which point you can |
111 |
|
|
// query get_info to find the exact amount required. yes I know |
112 |
|
|
// this is lame). |
113 |
|
|
// |
114 |
|
|
// If you pass in a non-NULL buffer of the type below, allocation |
115 |
|
|
// will occur from it as described above. Otherwise just pass NULL |
116 |
|
|
// to use malloc()/alloca() |
117 |
|
|
|
118 |
|
|
typedef struct |
119 |
|
|
{ |
120 |
|
|
char *alloc_buffer; |
121 |
|
|
int alloc_buffer_length_in_bytes; |
122 |
|
|
} stb_vorbis_alloc; |
123 |
|
|
|
124 |
|
|
|
125 |
|
|
/////////// FUNCTIONS USEABLE WITH ALL INPUT MODES |
126 |
|
|
|
127 |
|
|
typedef struct stb_vorbis stb_vorbis; |
128 |
|
|
|
129 |
|
|
typedef struct |
130 |
|
|
{ |
131 |
|
|
unsigned int sample_rate; |
132 |
|
|
int channels; |
133 |
|
|
|
134 |
|
|
unsigned int setup_memory_required; |
135 |
|
|
unsigned int setup_temp_memory_required; |
136 |
|
|
unsigned int temp_memory_required; |
137 |
|
|
|
138 |
|
|
int max_frame_size; |
139 |
|
|
} stb_vorbis_info; |
140 |
|
|
|
141 |
|
|
typedef struct |
142 |
|
|
{ |
143 |
|
|
char *vendor; |
144 |
|
|
|
145 |
|
|
int comment_list_length; |
146 |
|
|
char **comment_list; |
147 |
|
|
} stb_vorbis_comment; |
148 |
|
|
|
149 |
|
|
// get general information about the file |
150 |
|
|
extern stb_vorbis_info stb_vorbis_get_info(stb_vorbis *f); |
151 |
|
|
|
152 |
|
|
// get ogg comments |
153 |
|
|
extern stb_vorbis_comment stb_vorbis_get_comment(stb_vorbis *f); |
154 |
|
|
|
155 |
|
|
// get the last error detected (clears it, too) |
156 |
|
|
extern int stb_vorbis_get_error(stb_vorbis *f); |
157 |
|
|
|
158 |
|
|
// close an ogg vorbis file and free all memory in use |
159 |
|
|
extern void stb_vorbis_close(stb_vorbis *f); |
160 |
|
|
|
161 |
|
|
// this function returns the offset (in samples) from the beginning of the |
162 |
|
|
// file that will be returned by the next decode, if it is known, or -1 |
163 |
|
|
// otherwise. after a flush_pushdata() call, this may take a while before |
164 |
|
|
// it becomes valid again. |
165 |
|
|
// NOT WORKING YET after a seek with PULLDATA API |
166 |
|
|
extern int stb_vorbis_get_sample_offset(stb_vorbis *f); |
167 |
|
|
|
168 |
|
|
// returns the current seek point within the file, or offset from the beginning |
169 |
|
|
// of the memory buffer. In pushdata mode it returns 0. |
170 |
|
|
extern unsigned int stb_vorbis_get_file_offset(stb_vorbis *f); |
171 |
|
|
|
172 |
|
|
/////////// PUSHDATA API |
173 |
|
|
|
174 |
|
|
#ifndef STB_VORBIS_NO_PUSHDATA_API |
175 |
|
|
|
176 |
|
|
// this API allows you to get blocks of data from any source and hand |
177 |
|
|
// them to stb_vorbis. you have to buffer them; stb_vorbis will tell |
178 |
|
|
// you how much it used, and you have to give it the rest next time; |
179 |
|
|
// and stb_vorbis may not have enough data to work with and you will |
180 |
|
|
// need to give it the same data again PLUS more. Note that the Vorbis |
181 |
|
|
// specification does not bound the size of an individual frame. |
182 |
|
|
|
183 |
|
|
extern stb_vorbis *stb_vorbis_open_pushdata( |
184 |
|
|
const unsigned char * datablock, int datablock_length_in_bytes, |
185 |
|
|
int *datablock_memory_consumed_in_bytes, |
186 |
|
|
int *error, |
187 |
|
|
const stb_vorbis_alloc *alloc_buffer); |
188 |
|
|
// create a vorbis decoder by passing in the initial data block containing |
189 |
|
|
// the ogg&vorbis headers (you don't need to do parse them, just provide |
190 |
|
|
// the first N bytes of the file--you're told if it's not enough, see below) |
191 |
|
|
// on success, returns an stb_vorbis *, does not set error, returns the amount of |
192 |
|
|
// data parsed/consumed on this call in *datablock_memory_consumed_in_bytes; |
193 |
|
|
// on failure, returns NULL on error and sets *error, does not change *datablock_memory_consumed |
194 |
|
|
// if returns NULL and *error is VORBIS_need_more_data, then the input block was |
195 |
|
|
// incomplete and you need to pass in a larger block from the start of the file |
196 |
|
|
|
197 |
|
|
extern int stb_vorbis_decode_frame_pushdata( |
198 |
|
|
stb_vorbis *f, |
199 |
|
|
const unsigned char *datablock, int datablock_length_in_bytes, |
200 |
|
|
int *channels, // place to write number of float * buffers |
201 |
|
|
float ***output, // place to write float ** array of float * buffers |
202 |
|
|
int *samples // place to write number of output samples |
203 |
|
|
); |
204 |
|
|
// decode a frame of audio sample data if possible from the passed-in data block |
205 |
|
|
// |
206 |
|
|
// return value: number of bytes we used from datablock |
207 |
|
|
// |
208 |
|
|
// possible cases: |
209 |
|
|
// 0 bytes used, 0 samples output (need more data) |
210 |
|
|
// N bytes used, 0 samples output (resynching the stream, keep going) |
211 |
|
|
// N bytes used, M samples output (one frame of data) |
212 |
|
|
// note that after opening a file, you will ALWAYS get one N-bytes,0-sample |
213 |
|
|
// frame, because Vorbis always "discards" the first frame. |
214 |
|
|
// |
215 |
|
|
// Note that on resynch, stb_vorbis will rarely consume all of the buffer, |
216 |
|
|
// instead only datablock_length_in_bytes-3 or less. This is because it wants |
217 |
|
|
// to avoid missing parts of a page header if they cross a datablock boundary, |
218 |
|
|
// without writing state-machiney code to record a partial detection. |
219 |
|
|
// |
220 |
|
|
// The number of channels returned are stored in *channels (which can be |
221 |
|
|
// NULL--it is always the same as the number of channels reported by |
222 |
|
|
// get_info). *output will contain an array of float* buffers, one per |
223 |
|
|
// channel. In other words, (*output)[0][0] contains the first sample from |
224 |
|
|
// the first channel, and (*output)[1][0] contains the first sample from |
225 |
|
|
// the second channel. |
226 |
|
|
// |
227 |
|
|
// *output points into stb_vorbis's internal output buffer storage; these |
228 |
|
|
// buffers are owned by stb_vorbis and application code should not free |
229 |
|
|
// them or modify their contents. They are transient and will be overwritten |
230 |
|
|
// once you ask for more data to get decoded, so be sure to grab any data |
231 |
|
|
// you need before then. |
232 |
|
|
|
233 |
|
|
extern void stb_vorbis_flush_pushdata(stb_vorbis *f); |
234 |
|
|
// inform stb_vorbis that your next datablock will not be contiguous with |
235 |
|
|
// previous ones (e.g. you've seeked in the data); future attempts to decode |
236 |
|
|
// frames will cause stb_vorbis to resynchronize (as noted above), and |
237 |
|
|
// once it sees a valid Ogg page (typically 4-8KB, as large as 64KB), it |
238 |
|
|
// will begin decoding the _next_ frame. |
239 |
|
|
// |
240 |
|
|
// if you want to seek using pushdata, you need to seek in your file, then |
241 |
|
|
// call stb_vorbis_flush_pushdata(), then start calling decoding, then once |
242 |
|
|
// decoding is returning you data, call stb_vorbis_get_sample_offset, and |
243 |
|
|
// if you don't like the result, seek your file again and repeat. |
244 |
|
|
#endif |
245 |
|
|
|
246 |
|
|
|
247 |
|
|
////////// PULLING INPUT API |
248 |
|
|
|
249 |
|
|
#ifndef STB_VORBIS_NO_PULLDATA_API |
250 |
|
|
// This API assumes stb_vorbis is allowed to pull data from a source-- |
251 |
|
|
// either a block of memory containing the _entire_ vorbis stream, or a |
252 |
|
|
// FILE * that you or it create, or possibly some other reading mechanism |
253 |
|
|
// if you go modify the source to replace the FILE * case with some kind |
254 |
|
|
// of callback to your code. (But if you don't support seeking, you may |
255 |
|
|
// just want to go ahead and use pushdata.) |
256 |
|
|
|
257 |
|
|
#if !defined(STB_VORBIS_NO_STDIO) && !defined(STB_VORBIS_NO_INTEGER_CONVERSION) |
258 |
|
|
extern int stb_vorbis_decode_filename(const char *filename, int *channels, int *sample_rate, short **output); |
259 |
|
|
#endif |
260 |
|
|
#if !defined(STB_VORBIS_NO_INTEGER_CONVERSION) |
261 |
|
|
extern int stb_vorbis_decode_memory(const unsigned char *mem, int len, int *channels, int *sample_rate, short **output); |
262 |
|
|
#endif |
263 |
|
|
// decode an entire file and output the data interleaved into a malloc()ed |
264 |
|
|
// buffer stored in *output. The return value is the number of samples |
265 |
|
|
// decoded, or -1 if the file could not be opened or was not an ogg vorbis file. |
266 |
|
|
// When you're done with it, just free() the pointer returned in *output. |
267 |
|
|
|
268 |
|
|
extern stb_vorbis * stb_vorbis_open_memory(const unsigned char *data, int len, |
269 |
|
|
int *error, const stb_vorbis_alloc *alloc_buffer); |
270 |
|
|
// create an ogg vorbis decoder from an ogg vorbis stream in memory (note |
271 |
|
|
// this must be the entire stream!). on failure, returns NULL and sets *error |
272 |
|
|
|
273 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
274 |
|
|
extern stb_vorbis * stb_vorbis_open_filename(const char *filename, |
275 |
|
|
int *error, const stb_vorbis_alloc *alloc_buffer); |
276 |
|
|
// create an ogg vorbis decoder from a filename via fopen(). on failure, |
277 |
|
|
// returns NULL and sets *error (possibly to VORBIS_file_open_failure). |
278 |
|
|
|
279 |
|
|
extern stb_vorbis * stb_vorbis_open_file(FILE *f, int close_handle_on_close, |
280 |
|
|
int *error, const stb_vorbis_alloc *alloc_buffer); |
281 |
|
|
// create an ogg vorbis decoder from an open FILE *, looking for a stream at |
282 |
|
|
// the _current_ seek point (ftell). on failure, returns NULL and sets *error. |
283 |
|
|
// note that stb_vorbis must "own" this stream; if you seek it in between |
284 |
|
|
// calls to stb_vorbis, it will become confused. Moreover, if you attempt to |
285 |
|
|
// perform stb_vorbis_seek_*() operations on this file, it will assume it |
286 |
|
|
// owns the _entire_ rest of the file after the start point. Use the next |
287 |
|
|
// function, stb_vorbis_open_file_section(), to limit it. |
288 |
|
|
|
289 |
|
|
extern stb_vorbis * stb_vorbis_open_file_section(FILE *f, int close_handle_on_close, |
290 |
|
|
int *error, const stb_vorbis_alloc *alloc_buffer, unsigned int len); |
291 |
|
|
// create an ogg vorbis decoder from an open FILE *, looking for a stream at |
292 |
|
|
// the _current_ seek point (ftell); the stream will be of length 'len' bytes. |
293 |
|
|
// on failure, returns NULL and sets *error. note that stb_vorbis must "own" |
294 |
|
|
// this stream; if you seek it in between calls to stb_vorbis, it will become |
295 |
|
|
// confused. |
296 |
|
|
#endif |
297 |
|
|
|
298 |
|
|
extern int stb_vorbis_seek_frame(stb_vorbis *f, unsigned int sample_number); |
299 |
|
|
extern int stb_vorbis_seek(stb_vorbis *f, unsigned int sample_number); |
300 |
|
|
// these functions seek in the Vorbis file to (approximately) 'sample_number'. |
301 |
|
|
// after calling seek_frame(), the next call to get_frame_*() will include |
302 |
|
|
// the specified sample. after calling stb_vorbis_seek(), the next call to |
303 |
|
|
// stb_vorbis_get_samples_* will start with the specified sample. If you |
304 |
|
|
// do not need to seek to EXACTLY the target sample when using get_samples_*, |
305 |
|
|
// you can also use seek_frame(). |
306 |
|
|
|
307 |
|
|
extern int stb_vorbis_seek_start(stb_vorbis *f); |
308 |
|
|
// this function is equivalent to stb_vorbis_seek(f,0) |
309 |
|
|
|
310 |
|
|
extern unsigned int stb_vorbis_stream_length_in_samples(stb_vorbis *f); |
311 |
|
|
extern float stb_vorbis_stream_length_in_seconds(stb_vorbis *f); |
312 |
|
|
// these functions return the total length of the vorbis stream |
313 |
|
|
|
314 |
|
|
extern int stb_vorbis_get_frame_float(stb_vorbis *f, int *channels, float ***output); |
315 |
|
|
// decode the next frame and return the number of samples. the number of |
316 |
|
|
// channels returned are stored in *channels (which can be NULL--it is always |
317 |
|
|
// the same as the number of channels reported by get_info). *output will |
318 |
|
|
// contain an array of float* buffers, one per channel. These outputs will |
319 |
|
|
// be overwritten on the next call to stb_vorbis_get_frame_*. |
320 |
|
|
// |
321 |
|
|
// You generally should not intermix calls to stb_vorbis_get_frame_*() |
322 |
|
|
// and stb_vorbis_get_samples_*(), since the latter calls the former. |
323 |
|
|
|
324 |
|
|
#ifndef STB_VORBIS_NO_INTEGER_CONVERSION |
325 |
|
|
extern int stb_vorbis_get_frame_short_interleaved(stb_vorbis *f, int num_c, short *buffer, int num_shorts); |
326 |
|
|
extern int stb_vorbis_get_frame_short (stb_vorbis *f, int num_c, short **buffer, int num_samples); |
327 |
|
|
#endif |
328 |
|
|
// decode the next frame and return the number of *samples* per channel. |
329 |
|
|
// Note that for interleaved data, you pass in the number of shorts (the |
330 |
|
|
// size of your array), but the return value is the number of samples per |
331 |
|
|
// channel, not the total number of samples. |
332 |
|
|
// |
333 |
|
|
// The data is coerced to the number of channels you request according to the |
334 |
|
|
// channel coercion rules (see below). You must pass in the size of your |
335 |
|
|
// buffer(s) so that stb_vorbis will not overwrite the end of the buffer. |
336 |
|
|
// The maximum buffer size needed can be gotten from get_info(); however, |
337 |
|
|
// the Vorbis I specification implies an absolute maximum of 4096 samples |
338 |
|
|
// per channel. |
339 |
|
|
|
340 |
|
|
// Channel coercion rules: |
341 |
|
|
// Let M be the number of channels requested, and N the number of channels present, |
342 |
|
|
// and Cn be the nth channel; let stereo L be the sum of all L and center channels, |
343 |
|
|
// and stereo R be the sum of all R and center channels (channel assignment from the |
344 |
|
|
// vorbis spec). |
345 |
|
|
// M N output |
346 |
|
|
// 1 k sum(Ck) for all k |
347 |
|
|
// 2 * stereo L, stereo R |
348 |
|
|
// k l k > l, the first l channels, then 0s |
349 |
|
|
// k l k <= l, the first k channels |
350 |
|
|
// Note that this is not _good_ surround etc. mixing at all! It's just so |
351 |
|
|
// you get something useful. |
352 |
|
|
|
353 |
|
|
extern int stb_vorbis_get_samples_float_interleaved(stb_vorbis *f, int channels, float *buffer, int num_floats); |
354 |
|
|
extern int stb_vorbis_get_samples_float(stb_vorbis *f, int channels, float **buffer, int num_samples); |
355 |
|
|
// gets num_samples samples, not necessarily on a frame boundary--this requires |
356 |
|
|
// buffering so you have to supply the buffers. DOES NOT APPLY THE COERCION RULES. |
357 |
|
|
// Returns the number of samples stored per channel; it may be less than requested |
358 |
|
|
// at the end of the file. If there are no more samples in the file, returns 0. |
359 |
|
|
|
360 |
|
|
#ifndef STB_VORBIS_NO_INTEGER_CONVERSION |
361 |
|
|
extern int stb_vorbis_get_samples_short_interleaved(stb_vorbis *f, int channels, short *buffer, int num_shorts); |
362 |
|
|
extern int stb_vorbis_get_samples_short(stb_vorbis *f, int channels, short **buffer, int num_samples); |
363 |
|
|
#endif |
364 |
|
|
// gets num_samples samples, not necessarily on a frame boundary--this requires |
365 |
|
|
// buffering so you have to supply the buffers. Applies the coercion rules above |
366 |
|
|
// to produce 'channels' channels. Returns the number of samples stored per channel; |
367 |
|
|
// it may be less than requested at the end of the file. If there are no more |
368 |
|
|
// samples in the file, returns 0. |
369 |
|
|
|
370 |
|
|
#endif |
371 |
|
|
|
372 |
|
|
//////// ERROR CODES |
373 |
|
|
|
374 |
|
|
enum STBVorbisError |
375 |
|
|
{ |
376 |
|
|
VORBIS__no_error, |
377 |
|
|
|
378 |
|
|
VORBIS_need_more_data=1, // not a real error |
379 |
|
|
|
380 |
|
|
VORBIS_invalid_api_mixing, // can't mix API modes |
381 |
|
|
VORBIS_outofmem, // not enough memory |
382 |
|
|
VORBIS_feature_not_supported, // uses floor 0 |
383 |
|
|
VORBIS_too_many_channels, // STB_VORBIS_MAX_CHANNELS is too small |
384 |
|
|
VORBIS_file_open_failure, // fopen() failed |
385 |
|
|
VORBIS_seek_without_length, // can't seek in unknown-length file |
386 |
|
|
|
387 |
|
|
VORBIS_unexpected_eof=10, // file is truncated? |
388 |
|
|
VORBIS_seek_invalid, // seek past EOF |
389 |
|
|
|
390 |
|
|
// decoding errors (corrupt/invalid stream) -- you probably |
391 |
|
|
// don't care about the exact details of these |
392 |
|
|
|
393 |
|
|
// vorbis errors: |
394 |
|
|
VORBIS_invalid_setup=20, |
395 |
|
|
VORBIS_invalid_stream, |
396 |
|
|
|
397 |
|
|
// ogg errors: |
398 |
|
|
VORBIS_missing_capture_pattern=30, |
399 |
|
|
VORBIS_invalid_stream_structure_version, |
400 |
|
|
VORBIS_continued_packet_flag_invalid, |
401 |
|
|
VORBIS_incorrect_stream_serial_number, |
402 |
|
|
VORBIS_invalid_first_page, |
403 |
|
|
VORBIS_bad_packet_type, |
404 |
|
|
VORBIS_cant_find_last_page, |
405 |
|
|
VORBIS_seek_failed, |
406 |
|
|
VORBIS_ogg_skeleton_not_supported |
407 |
|
|
}; |
408 |
|
|
|
409 |
|
|
|
410 |
|
|
#ifdef __cplusplus |
411 |
|
|
} |
412 |
|
|
#endif |
413 |
|
|
|
414 |
|
|
#endif // STB_VORBIS_INCLUDE_STB_VORBIS_H |
415 |
|
|
// |
416 |
|
|
// HEADER ENDS HERE |
417 |
|
|
// |
418 |
|
|
////////////////////////////////////////////////////////////////////////////// |
419 |
|
|
|
420 |
|
|
#ifndef STB_VORBIS_HEADER_ONLY |
421 |
|
|
|
422 |
|
|
// global configuration settings (e.g. set these in the project/makefile), |
423 |
|
|
// or just set them in this file at the top (although ideally the first few |
424 |
|
|
// should be visible when the header file is compiled too, although it's not |
425 |
|
|
// crucial) |
426 |
|
|
|
427 |
|
|
// STB_VORBIS_NO_PUSHDATA_API |
428 |
|
|
// does not compile the code for the various stb_vorbis_*_pushdata() |
429 |
|
|
// functions |
430 |
|
|
// #define STB_VORBIS_NO_PUSHDATA_API |
431 |
|
|
|
432 |
|
|
// STB_VORBIS_NO_PULLDATA_API |
433 |
|
|
// does not compile the code for the non-pushdata APIs |
434 |
|
|
// #define STB_VORBIS_NO_PULLDATA_API |
435 |
|
|
|
436 |
|
|
// STB_VORBIS_NO_STDIO |
437 |
|
|
// does not compile the code for the APIs that use FILE *s internally |
438 |
|
|
// or externally (implied by STB_VORBIS_NO_PULLDATA_API) |
439 |
|
|
// #define STB_VORBIS_NO_STDIO |
440 |
|
|
|
441 |
|
|
// STB_VORBIS_NO_INTEGER_CONVERSION |
442 |
|
|
// does not compile the code for converting audio sample data from |
443 |
|
|
// float to integer (implied by STB_VORBIS_NO_PULLDATA_API) |
444 |
|
|
// #define STB_VORBIS_NO_INTEGER_CONVERSION |
445 |
|
|
|
446 |
|
|
// STB_VORBIS_NO_FAST_SCALED_FLOAT |
447 |
|
|
// does not use a fast float-to-int trick to accelerate float-to-int on |
448 |
|
|
// most platforms which requires endianness be defined correctly. |
449 |
|
|
//#define STB_VORBIS_NO_FAST_SCALED_FLOAT |
450 |
|
|
|
451 |
|
|
|
452 |
|
|
// STB_VORBIS_MAX_CHANNELS [number] |
453 |
|
|
// globally define this to the maximum number of channels you need. |
454 |
|
|
// The spec does not put a restriction on channels except that |
455 |
|
|
// the count is stored in a byte, so 255 is the hard limit. |
456 |
|
|
// Reducing this saves about 16 bytes per value, so using 16 saves |
457 |
|
|
// (255-16)*16 or around 4KB. Plus anything other memory usage |
458 |
|
|
// I forgot to account for. Can probably go as low as 8 (7.1 audio), |
459 |
|
|
// 6 (5.1 audio), or 2 (stereo only). |
460 |
|
|
#ifndef STB_VORBIS_MAX_CHANNELS |
461 |
|
|
#define STB_VORBIS_MAX_CHANNELS 16 // enough for anyone? |
462 |
|
|
#endif |
463 |
|
|
|
464 |
|
|
// STB_VORBIS_PUSHDATA_CRC_COUNT [number] |
465 |
|
|
// after a flush_pushdata(), stb_vorbis begins scanning for the |
466 |
|
|
// next valid page, without backtracking. when it finds something |
467 |
|
|
// that looks like a page, it streams through it and verifies its |
468 |
|
|
// CRC32. Should that validation fail, it keeps scanning. But it's |
469 |
|
|
// possible that _while_ streaming through to check the CRC32 of |
470 |
|
|
// one candidate page, it sees another candidate page. This #define |
471 |
|
|
// determines how many "overlapping" candidate pages it can search |
472 |
|
|
// at once. Note that "real" pages are typically ~4KB to ~8KB, whereas |
473 |
|
|
// garbage pages could be as big as 64KB, but probably average ~16KB. |
474 |
|
|
// So don't hose ourselves by scanning an apparent 64KB page and |
475 |
|
|
// missing a ton of real ones in the interim; so minimum of 2 |
476 |
|
|
#ifndef STB_VORBIS_PUSHDATA_CRC_COUNT |
477 |
|
|
#define STB_VORBIS_PUSHDATA_CRC_COUNT 4 |
478 |
|
|
#endif |
479 |
|
|
|
480 |
|
|
// STB_VORBIS_FAST_HUFFMAN_LENGTH [number] |
481 |
|
|
// sets the log size of the huffman-acceleration table. Maximum |
482 |
|
|
// supported value is 24. with larger numbers, more decodings are O(1), |
483 |
|
|
// but the table size is larger so worse cache missing, so you'll have |
484 |
|
|
// to probe (and try multiple ogg vorbis files) to find the sweet spot. |
485 |
|
|
#ifndef STB_VORBIS_FAST_HUFFMAN_LENGTH |
486 |
|
|
#define STB_VORBIS_FAST_HUFFMAN_LENGTH 10 |
487 |
|
|
#endif |
488 |
|
|
|
489 |
|
|
// STB_VORBIS_FAST_BINARY_LENGTH [number] |
490 |
|
|
// sets the log size of the binary-search acceleration table. this |
491 |
|
|
// is used in similar fashion to the fast-huffman size to set initial |
492 |
|
|
// parameters for the binary search |
493 |
|
|
|
494 |
|
|
// STB_VORBIS_FAST_HUFFMAN_INT |
495 |
|
|
// The fast huffman tables are much more efficient if they can be |
496 |
|
|
// stored as 16-bit results instead of 32-bit results. This restricts |
497 |
|
|
// the codebooks to having only 65535 possible outcomes, though. |
498 |
|
|
// (At least, accelerated by the huffman table.) |
499 |
|
|
#ifndef STB_VORBIS_FAST_HUFFMAN_INT |
500 |
|
|
#define STB_VORBIS_FAST_HUFFMAN_SHORT |
501 |
|
|
#endif |
502 |
|
|
|
503 |
|
|
// STB_VORBIS_NO_HUFFMAN_BINARY_SEARCH |
504 |
|
|
// If the 'fast huffman' search doesn't succeed, then stb_vorbis falls |
505 |
|
|
// back on binary searching for the correct one. This requires storing |
506 |
|
|
// extra tables with the huffman codes in sorted order. Defining this |
507 |
|
|
// symbol trades off space for speed by forcing a linear search in the |
508 |
|
|
// non-fast case, except for "sparse" codebooks. |
509 |
|
|
// #define STB_VORBIS_NO_HUFFMAN_BINARY_SEARCH |
510 |
|
|
|
511 |
|
|
// STB_VORBIS_DIVIDES_IN_RESIDUE |
512 |
|
|
// stb_vorbis precomputes the result of the scalar residue decoding |
513 |
|
|
// that would otherwise require a divide per chunk. you can trade off |
514 |
|
|
// space for time by defining this symbol. |
515 |
|
|
// #define STB_VORBIS_DIVIDES_IN_RESIDUE |
516 |
|
|
|
517 |
|
|
// STB_VORBIS_DIVIDES_IN_CODEBOOK |
518 |
|
|
// vorbis VQ codebooks can be encoded two ways: with every case explicitly |
519 |
|
|
// stored, or with all elements being chosen from a small range of values, |
520 |
|
|
// and all values possible in all elements. By default, stb_vorbis expands |
521 |
|
|
// this latter kind out to look like the former kind for ease of decoding, |
522 |
|
|
// because otherwise an integer divide-per-vector-element is required to |
523 |
|
|
// unpack the index. If you define STB_VORBIS_DIVIDES_IN_CODEBOOK, you can |
524 |
|
|
// trade off storage for speed. |
525 |
|
|
//#define STB_VORBIS_DIVIDES_IN_CODEBOOK |
526 |
|
|
|
527 |
|
|
#ifdef STB_VORBIS_CODEBOOK_SHORTS |
528 |
|
|
#error "STB_VORBIS_CODEBOOK_SHORTS is no longer supported as it produced incorrect results for some input formats" |
529 |
|
|
#endif |
530 |
|
|
|
531 |
|
|
// STB_VORBIS_DIVIDE_TABLE |
532 |
|
|
// this replaces small integer divides in the floor decode loop with |
533 |
|
|
// table lookups. made less than 1% difference, so disabled by default. |
534 |
|
|
|
535 |
|
|
// STB_VORBIS_NO_INLINE_DECODE |
536 |
|
|
// disables the inlining of the scalar codebook fast-huffman decode. |
537 |
|
|
// might save a little codespace; useful for debugging |
538 |
|
|
// #define STB_VORBIS_NO_INLINE_DECODE |
539 |
|
|
|
540 |
|
|
// STB_VORBIS_NO_DEFER_FLOOR |
541 |
|
|
// Normally we only decode the floor without synthesizing the actual |
542 |
|
|
// full curve. We can instead synthesize the curve immediately. This |
543 |
|
|
// requires more memory and is very likely slower, so I don't think |
544 |
|
|
// you'd ever want to do it except for debugging. |
545 |
|
|
// #define STB_VORBIS_NO_DEFER_FLOOR |
546 |
|
|
|
547 |
|
|
|
548 |
|
|
|
549 |
|
|
|
550 |
|
|
////////////////////////////////////////////////////////////////////////////// |
551 |
|
|
|
552 |
|
|
#ifdef STB_VORBIS_NO_PULLDATA_API |
553 |
|
|
#define STB_VORBIS_NO_INTEGER_CONVERSION |
554 |
|
|
#define STB_VORBIS_NO_STDIO |
555 |
|
|
#endif |
556 |
|
|
|
557 |
|
|
#if defined(STB_VORBIS_NO_CRT) && !defined(STB_VORBIS_NO_STDIO) |
558 |
|
|
#define STB_VORBIS_NO_STDIO 1 |
559 |
|
|
#endif |
560 |
|
|
|
561 |
|
|
#ifndef STB_VORBIS_NO_INTEGER_CONVERSION |
562 |
|
|
#ifndef STB_VORBIS_NO_FAST_SCALED_FLOAT |
563 |
|
|
|
564 |
|
|
// only need endianness for fast-float-to-int, which we don't |
565 |
|
|
// use for pushdata |
566 |
|
|
|
567 |
|
|
#ifndef STB_VORBIS_BIG_ENDIAN |
568 |
|
|
#define STB_VORBIS_ENDIAN 0 |
569 |
|
|
#else |
570 |
|
|
#define STB_VORBIS_ENDIAN 1 |
571 |
|
|
#endif |
572 |
|
|
|
573 |
|
|
#endif |
574 |
|
|
#endif |
575 |
|
|
|
576 |
|
|
|
577 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
578 |
|
|
#include <stdio.h> |
579 |
|
|
#endif |
580 |
|
|
|
581 |
|
|
#ifndef STB_VORBIS_NO_CRT |
582 |
|
|
#include <stdlib.h> |
583 |
|
|
#include <string.h> |
584 |
|
|
#include <assert.h> |
585 |
|
|
#include <math.h> |
586 |
|
|
|
587 |
|
|
// find definition of alloca if it's not in stdlib.h: |
588 |
|
|
#if defined(_MSC_VER) || defined(__MINGW32__) |
589 |
|
|
#include <malloc.h> |
590 |
|
|
#endif |
591 |
|
|
#if defined(__linux__) || defined(__linux) || defined(__sun__) || defined(__EMSCRIPTEN__) || defined(__NEWLIB__) |
592 |
|
|
#include <alloca.h> |
593 |
|
|
#endif |
594 |
|
|
#else // STB_VORBIS_NO_CRT |
595 |
|
|
#define NULL 0 |
596 |
|
|
#define malloc(s) 0 |
597 |
|
|
#define free(s) ((void) 0) |
598 |
|
|
#define realloc(s) 0 |
599 |
|
|
#endif // STB_VORBIS_NO_CRT |
600 |
|
|
|
601 |
|
|
#include <limits.h> |
602 |
|
|
|
603 |
|
|
#ifdef __MINGW32__ |
604 |
|
|
// eff you mingw: |
605 |
|
|
// "fixed": |
606 |
|
|
// http://sourceforge.net/p/mingw-w64/mailman/message/32882927/ |
607 |
|
|
// "no that broke the build, reverted, who cares about C": |
608 |
|
|
// http://sourceforge.net/p/mingw-w64/mailman/message/32890381/ |
609 |
|
|
#ifdef __forceinline |
610 |
|
|
#undef __forceinline |
611 |
|
|
#endif |
612 |
|
|
#define __forceinline |
613 |
|
|
#ifndef alloca |
614 |
|
|
#define alloca __builtin_alloca |
615 |
|
|
#endif |
616 |
|
|
#elif !defined(_MSC_VER) |
617 |
|
|
#if __GNUC__ |
618 |
|
|
#define __forceinline inline |
619 |
|
|
#else |
620 |
|
|
#define __forceinline |
621 |
|
|
#endif |
622 |
|
|
#endif |
623 |
|
|
|
624 |
|
|
#if STB_VORBIS_MAX_CHANNELS > 256 |
625 |
|
|
#error "Value of STB_VORBIS_MAX_CHANNELS outside of allowed range" |
626 |
|
|
#endif |
627 |
|
|
|
628 |
|
|
#if STB_VORBIS_FAST_HUFFMAN_LENGTH > 24 |
629 |
|
|
#error "Value of STB_VORBIS_FAST_HUFFMAN_LENGTH outside of allowed range" |
630 |
|
|
#endif |
631 |
|
|
|
632 |
|
|
|
633 |
|
|
#if 0 |
634 |
|
|
#include <crtdbg.h> |
635 |
|
|
#define CHECK(f) _CrtIsValidHeapPointer(f->channel_buffers[1]) |
636 |
|
|
#else |
637 |
|
|
#define CHECK(f) ((void) 0) |
638 |
|
|
#endif |
639 |
|
|
|
640 |
|
|
#define MAX_BLOCKSIZE_LOG 13 // from specification |
641 |
|
|
#define MAX_BLOCKSIZE (1 << MAX_BLOCKSIZE_LOG) |
642 |
|
|
|
643 |
|
|
|
644 |
|
|
typedef unsigned char uint8; |
645 |
|
|
typedef signed char int8; |
646 |
|
|
typedef unsigned short uint16; |
647 |
|
|
typedef signed short int16; |
648 |
|
|
typedef unsigned int uint32; |
649 |
|
|
typedef signed int int32; |
650 |
|
|
|
651 |
|
|
#ifndef TRUE |
652 |
|
|
#define TRUE 1 |
653 |
|
|
#define FALSE 0 |
654 |
|
|
#endif |
655 |
|
|
|
656 |
|
|
typedef float codetype; |
657 |
|
|
|
658 |
|
|
#ifdef _MSC_VER |
659 |
|
|
#define STBV_NOTUSED(v) (void)(v) |
660 |
|
|
#else |
661 |
|
|
#define STBV_NOTUSED(v) (void)sizeof(v) |
662 |
|
|
#endif |
663 |
|
|
|
664 |
|
|
// @NOTE |
665 |
|
|
// |
666 |
|
|
// Some arrays below are tagged "//varies", which means it's actually |
667 |
|
|
// a variable-sized piece of data, but rather than malloc I assume it's |
668 |
|
|
// small enough it's better to just allocate it all together with the |
669 |
|
|
// main thing |
670 |
|
|
// |
671 |
|
|
// Most of the variables are specified with the smallest size I could pack |
672 |
|
|
// them into. It might give better performance to make them all full-sized |
673 |
|
|
// integers. It should be safe to freely rearrange the structures or change |
674 |
|
|
// the sizes larger--nothing relies on silently truncating etc., nor the |
675 |
|
|
// order of variables. |
676 |
|
|
|
677 |
|
|
#define FAST_HUFFMAN_TABLE_SIZE (1 << STB_VORBIS_FAST_HUFFMAN_LENGTH) |
678 |
|
|
#define FAST_HUFFMAN_TABLE_MASK (FAST_HUFFMAN_TABLE_SIZE - 1) |
679 |
|
|
|
680 |
|
|
typedef struct |
681 |
|
|
{ |
682 |
|
|
int dimensions, entries; |
683 |
|
|
uint8 *codeword_lengths; |
684 |
|
|
float minimum_value; |
685 |
|
|
float delta_value; |
686 |
|
|
uint8 value_bits; |
687 |
|
|
uint8 lookup_type; |
688 |
|
|
uint8 sequence_p; |
689 |
|
|
uint8 sparse; |
690 |
|
|
uint32 lookup_values; |
691 |
|
|
codetype *multiplicands; |
692 |
|
|
uint32 *codewords; |
693 |
|
|
#ifdef STB_VORBIS_FAST_HUFFMAN_SHORT |
694 |
|
|
int16 fast_huffman[FAST_HUFFMAN_TABLE_SIZE]; |
695 |
|
|
#else |
696 |
|
|
int32 fast_huffman[FAST_HUFFMAN_TABLE_SIZE]; |
697 |
|
|
#endif |
698 |
|
|
uint32 *sorted_codewords; |
699 |
|
|
int *sorted_values; |
700 |
|
|
int sorted_entries; |
701 |
|
|
} Codebook; |
702 |
|
|
|
703 |
|
|
typedef struct |
704 |
|
|
{ |
705 |
|
|
uint8 order; |
706 |
|
|
uint16 rate; |
707 |
|
|
uint16 bark_map_size; |
708 |
|
|
uint8 amplitude_bits; |
709 |
|
|
uint8 amplitude_offset; |
710 |
|
|
uint8 number_of_books; |
711 |
|
|
uint8 book_list[16]; // varies |
712 |
|
|
} Floor0; |
713 |
|
|
|
714 |
|
|
typedef struct |
715 |
|
|
{ |
716 |
|
|
uint8 partitions; |
717 |
|
|
uint8 partition_class_list[32]; // varies |
718 |
|
|
uint8 class_dimensions[16]; // varies |
719 |
|
|
uint8 class_subclasses[16]; // varies |
720 |
|
|
uint8 class_masterbooks[16]; // varies |
721 |
|
|
int16 subclass_books[16][8]; // varies |
722 |
|
|
uint16 Xlist[31*8+2]; // varies |
723 |
|
|
uint8 sorted_order[31*8+2]; |
724 |
|
|
uint8 neighbors[31*8+2][2]; |
725 |
|
|
uint8 floor1_multiplier; |
726 |
|
|
uint8 rangebits; |
727 |
|
|
int values; |
728 |
|
|
} Floor1; |
729 |
|
|
|
730 |
|
|
typedef union |
731 |
|
|
{ |
732 |
|
|
Floor0 floor0; |
733 |
|
|
Floor1 floor1; |
734 |
|
|
} Floor; |
735 |
|
|
|
736 |
|
|
typedef struct |
737 |
|
|
{ |
738 |
|
|
uint32 begin, end; |
739 |
|
|
uint32 part_size; |
740 |
|
|
uint8 classifications; |
741 |
|
|
uint8 classbook; |
742 |
|
|
uint8 **classdata; |
743 |
|
|
int16 (*residue_books)[8]; |
744 |
|
|
} Residue; |
745 |
|
|
|
746 |
|
|
typedef struct |
747 |
|
|
{ |
748 |
|
|
uint8 magnitude; |
749 |
|
|
uint8 angle; |
750 |
|
|
uint8 mux; |
751 |
|
|
} MappingChannel; |
752 |
|
|
|
753 |
|
|
typedef struct |
754 |
|
|
{ |
755 |
|
|
uint16 coupling_steps; |
756 |
|
|
MappingChannel *chan; |
757 |
|
|
uint8 submaps; |
758 |
|
|
uint8 submap_floor[15]; // varies |
759 |
|
|
uint8 submap_residue[15]; // varies |
760 |
|
|
} Mapping; |
761 |
|
|
|
762 |
|
|
typedef struct |
763 |
|
|
{ |
764 |
|
|
uint8 blockflag; |
765 |
|
|
uint8 mapping; |
766 |
|
|
uint16 windowtype; |
767 |
|
|
uint16 transformtype; |
768 |
|
|
} Mode; |
769 |
|
|
|
770 |
|
|
typedef struct |
771 |
|
|
{ |
772 |
|
|
uint32 goal_crc; // expected crc if match |
773 |
|
|
int bytes_left; // bytes left in packet |
774 |
|
|
uint32 crc_so_far; // running crc |
775 |
|
|
int bytes_done; // bytes processed in _current_ chunk |
776 |
|
|
uint32 sample_loc; // granule pos encoded in page |
777 |
|
|
} CRCscan; |
778 |
|
|
|
779 |
|
|
typedef struct |
780 |
|
|
{ |
781 |
|
|
uint32 page_start, page_end; |
782 |
|
|
uint32 last_decoded_sample; |
783 |
|
|
} ProbedPage; |
784 |
|
|
|
785 |
|
|
struct stb_vorbis |
786 |
|
|
{ |
787 |
|
|
// user-accessible info |
788 |
|
|
unsigned int sample_rate; |
789 |
|
|
int channels; |
790 |
|
|
|
791 |
|
|
unsigned int setup_memory_required; |
792 |
|
|
unsigned int temp_memory_required; |
793 |
|
|
unsigned int setup_temp_memory_required; |
794 |
|
|
|
795 |
|
|
char *vendor; |
796 |
|
|
int comment_list_length; |
797 |
|
|
char **comment_list; |
798 |
|
|
|
799 |
|
|
// input config |
800 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
801 |
|
|
FILE *f; |
802 |
|
|
uint32 f_start; |
803 |
|
|
int close_on_free; |
804 |
|
|
#endif |
805 |
|
|
|
806 |
|
|
uint8 *stream; |
807 |
|
|
uint8 *stream_start; |
808 |
|
|
uint8 *stream_end; |
809 |
|
|
|
810 |
|
|
uint32 stream_len; |
811 |
|
|
|
812 |
|
|
uint8 push_mode; |
813 |
|
|
|
814 |
|
|
// the page to seek to when seeking to start, may be zero |
815 |
|
|
uint32 first_audio_page_offset; |
816 |
|
|
|
817 |
|
|
// p_first is the page on which the first audio packet ends |
818 |
|
|
// (but not necessarily the page on which it starts) |
819 |
|
|
ProbedPage p_first, p_last; |
820 |
|
|
|
821 |
|
|
// memory management |
822 |
|
|
stb_vorbis_alloc alloc; |
823 |
|
|
int setup_offset; |
824 |
|
|
int temp_offset; |
825 |
|
|
|
826 |
|
|
// run-time results |
827 |
|
|
int eof; |
828 |
|
|
enum STBVorbisError error; |
829 |
|
|
|
830 |
|
|
// user-useful data |
831 |
|
|
|
832 |
|
|
// header info |
833 |
|
|
int blocksize[2]; |
834 |
|
|
int blocksize_0, blocksize_1; |
835 |
|
|
int codebook_count; |
836 |
|
|
Codebook *codebooks; |
837 |
|
|
int floor_count; |
838 |
|
|
uint16 floor_types[64]; // varies |
839 |
|
|
Floor *floor_config; |
840 |
|
|
int residue_count; |
841 |
|
|
uint16 residue_types[64]; // varies |
842 |
|
|
Residue *residue_config; |
843 |
|
|
int mapping_count; |
844 |
|
|
Mapping *mapping; |
845 |
|
|
int mode_count; |
846 |
|
|
Mode mode_config[64]; // varies |
847 |
|
|
|
848 |
|
|
uint32 total_samples; |
849 |
|
|
|
850 |
|
|
// decode buffer |
851 |
|
|
float *channel_buffers[STB_VORBIS_MAX_CHANNELS]; |
852 |
|
|
float *outputs [STB_VORBIS_MAX_CHANNELS]; |
853 |
|
|
|
854 |
|
|
float *previous_window[STB_VORBIS_MAX_CHANNELS]; |
855 |
|
|
int previous_length; |
856 |
|
|
|
857 |
|
|
#ifndef STB_VORBIS_NO_DEFER_FLOOR |
858 |
|
|
int16 *finalY[STB_VORBIS_MAX_CHANNELS]; |
859 |
|
|
#else |
860 |
|
|
float *floor_buffers[STB_VORBIS_MAX_CHANNELS]; |
861 |
|
|
#endif |
862 |
|
|
|
863 |
|
|
uint32 current_loc; // sample location of next frame to decode |
864 |
|
|
int current_loc_valid; |
865 |
|
|
|
866 |
|
|
// per-blocksize precomputed data |
867 |
|
|
|
868 |
|
|
// twiddle factors |
869 |
|
|
float *A[2],*B[2],*C[2]; |
870 |
|
|
float *window[2]; |
871 |
|
|
uint16 *bit_reverse[2]; |
872 |
|
|
|
873 |
|
|
// current page/packet/segment streaming info |
874 |
|
|
uint32 serial; // stream serial number for verification |
875 |
|
|
int last_page; |
876 |
|
|
int segment_count; |
877 |
|
|
uint8 segments[255]; |
878 |
|
|
uint8 page_flag; |
879 |
|
|
uint8 bytes_in_seg; |
880 |
|
|
uint8 first_decode; |
881 |
|
|
int next_seg; |
882 |
|
|
int last_seg; // flag that we're on the last segment |
883 |
|
|
int last_seg_which; // what was the segment number of the last seg? |
884 |
|
|
uint32 acc; |
885 |
|
|
int valid_bits; |
886 |
|
|
int packet_bytes; |
887 |
|
|
int end_seg_with_known_loc; |
888 |
|
|
uint32 known_loc_for_packet; |
889 |
|
|
int discard_samples_deferred; |
890 |
|
|
uint32 samples_output; |
891 |
|
|
|
892 |
|
|
// push mode scanning |
893 |
|
|
int page_crc_tests; // only in push_mode: number of tests active; -1 if not searching |
894 |
|
|
#ifndef STB_VORBIS_NO_PUSHDATA_API |
895 |
|
|
CRCscan scan[STB_VORBIS_PUSHDATA_CRC_COUNT]; |
896 |
|
|
#endif |
897 |
|
|
|
898 |
|
|
// sample-access |
899 |
|
|
int channel_buffer_start; |
900 |
|
|
int channel_buffer_end; |
901 |
|
|
}; |
902 |
|
|
|
903 |
|
|
#if defined(STB_VORBIS_NO_PUSHDATA_API) |
904 |
|
|
#define IS_PUSH_MODE(f) FALSE |
905 |
|
|
#elif defined(STB_VORBIS_NO_PULLDATA_API) |
906 |
|
|
#define IS_PUSH_MODE(f) TRUE |
907 |
|
|
#else |
908 |
|
|
#define IS_PUSH_MODE(f) ((f)->push_mode) |
909 |
|
|
#endif |
910 |
|
|
|
911 |
|
|
typedef struct stb_vorbis vorb; |
912 |
|
|
|
913 |
|
|
static int error(vorb *f, enum STBVorbisError e) |
914 |
|
|
{ |
915 |
|
✗ |
f->error = e; |
916 |
|
|
if (!f->eof && e != VORBIS_need_more_data) { |
917 |
|
|
f->error=e; // breakpoint for debugging |
918 |
|
|
} |
919 |
|
✗ |
return 0; |
920 |
|
|
} |
921 |
|
|
|
922 |
|
|
|
923 |
|
|
// these functions are used for allocating temporary memory |
924 |
|
|
// while decoding. if you can afford the stack space, use |
925 |
|
|
// alloca(); otherwise, provide a temp buffer and it will |
926 |
|
|
// allocate out of those. |
927 |
|
|
|
928 |
|
|
#define array_size_required(count,size) (count*(sizeof(void *)+(size))) |
929 |
|
|
|
930 |
|
|
#define temp_alloc(f,size) (f->alloc.alloc_buffer ? setup_temp_malloc(f,size) : alloca(size)) |
931 |
|
|
#define temp_free(f,p) (void)0 |
932 |
|
|
#define temp_alloc_save(f) ((f)->temp_offset) |
933 |
|
|
#define temp_alloc_restore(f,p) ((f)->temp_offset = (p)) |
934 |
|
|
|
935 |
|
|
#define temp_block_array(f,count,size) make_block_array(temp_alloc(f,array_size_required(count,size)), count, size) |
936 |
|
|
|
937 |
|
|
// given a sufficiently large block of memory, make an array of pointers to subblocks of it |
938 |
|
|
static void *make_block_array(void *mem, int count, int size) |
939 |
|
|
{ |
940 |
|
|
int i; |
941 |
|
|
void ** p = (void **) mem; |
942 |
|
✗ |
char *q = (char *) (p + count); |
943 |
|
✗ |
for (i=0; i < count; ++i) { |
944 |
|
✗ |
p[i] = q; |
945 |
|
✗ |
q += size; |
946 |
|
|
} |
947 |
|
|
return p; |
948 |
|
|
} |
949 |
|
|
|
950 |
|
✗ |
static void *setup_malloc(vorb *f, int sz) |
951 |
|
|
{ |
952 |
|
✗ |
sz = (sz+7) & ~7; // round up to nearest 8 for alignment of future allocs. |
953 |
|
✗ |
f->setup_memory_required += sz; |
954 |
|
✗ |
if (f->alloc.alloc_buffer) { |
955 |
|
✗ |
void *p = (char *) f->alloc.alloc_buffer + f->setup_offset; |
956 |
|
✗ |
if (f->setup_offset + sz > f->temp_offset) return NULL; |
957 |
|
✗ |
f->setup_offset += sz; |
958 |
|
✗ |
return p; |
959 |
|
|
} |
960 |
|
✗ |
return sz ? malloc(sz) : NULL; |
961 |
|
|
} |
962 |
|
|
|
963 |
|
|
static void setup_free(vorb *f, void *p) |
964 |
|
|
{ |
965 |
|
✗ |
if (f->alloc.alloc_buffer) return; // do nothing; setup mem is a stack |
966 |
|
✗ |
free(p); |
967 |
|
|
} |
968 |
|
|
|
969 |
|
✗ |
static void *setup_temp_malloc(vorb *f, int sz) |
970 |
|
|
{ |
971 |
|
✗ |
sz = (sz+7) & ~7; // round up to nearest 8 for alignment of future allocs. |
972 |
|
✗ |
if (f->alloc.alloc_buffer) { |
973 |
|
✗ |
if (f->temp_offset - sz < f->setup_offset) return NULL; |
974 |
|
✗ |
f->temp_offset -= sz; |
975 |
|
✗ |
return (char *) f->alloc.alloc_buffer + f->temp_offset; |
976 |
|
|
} |
977 |
|
✗ |
return malloc(sz); |
978 |
|
|
} |
979 |
|
|
|
980 |
|
|
static void setup_temp_free(vorb *f, void *p, int sz) |
981 |
|
|
{ |
982 |
|
✗ |
if (f->alloc.alloc_buffer) { |
983 |
|
✗ |
f->temp_offset += (sz+7)&~7; |
984 |
|
✗ |
return; |
985 |
|
|
} |
986 |
|
✗ |
free(p); |
987 |
|
|
} |
988 |
|
|
|
989 |
|
|
#define CRC32_POLY 0x04c11db7 // from spec |
990 |
|
|
|
991 |
|
|
static uint32 crc_table[256]; |
992 |
|
|
static void crc32_init(void) |
993 |
|
|
{ |
994 |
|
|
int i,j; |
995 |
|
|
uint32 s; |
996 |
|
✗ |
for(i=0; i < 256; i++) { |
997 |
|
✗ |
for (s=(uint32) i << 24, j=0; j < 8; ++j) |
998 |
|
✗ |
s = (s << 1) ^ (s >= (1U<<31) ? CRC32_POLY : 0); |
999 |
|
✗ |
crc_table[i] = s; |
1000 |
|
|
} |
1001 |
|
|
} |
1002 |
|
|
|
1003 |
|
|
static __forceinline uint32 crc32_update(uint32 crc, uint8 byte) |
1004 |
|
|
{ |
1005 |
|
✗ |
return (crc << 8) ^ crc_table[byte ^ (crc >> 24)]; |
1006 |
|
|
} |
1007 |
|
|
|
1008 |
|
|
|
1009 |
|
|
// used in setup, and for huffman that doesn't go fast path |
1010 |
|
✗ |
static unsigned int bit_reverse(unsigned int n) |
1011 |
|
|
{ |
1012 |
|
✗ |
n = ((n & 0xAAAAAAAA) >> 1) | ((n & 0x55555555) << 1); |
1013 |
|
✗ |
n = ((n & 0xCCCCCCCC) >> 2) | ((n & 0x33333333) << 2); |
1014 |
|
✗ |
n = ((n & 0xF0F0F0F0) >> 4) | ((n & 0x0F0F0F0F) << 4); |
1015 |
|
✗ |
n = ((n & 0xFF00FF00) >> 8) | ((n & 0x00FF00FF) << 8); |
1016 |
|
✗ |
return (n >> 16) | (n << 16); |
1017 |
|
|
} |
1018 |
|
|
|
1019 |
|
|
static float square(float x) |
1020 |
|
|
{ |
1021 |
|
✗ |
return x*x; |
1022 |
|
|
} |
1023 |
|
|
|
1024 |
|
|
// this is a weird definition of log2() for which log2(1) = 1, log2(2) = 2, log2(4) = 3 |
1025 |
|
|
// as required by the specification. fast(?) implementation from stb.h |
1026 |
|
|
// @OPTIMIZE: called multiple times per-packet with "constants"; move to setup |
1027 |
|
✗ |
static int ilog(int32 n) |
1028 |
|
|
{ |
1029 |
|
|
static signed char log2_4[16] = { 0,1,2,2,3,3,3,3,4,4,4,4,4,4,4,4 }; |
1030 |
|
|
|
1031 |
|
✗ |
if (n < 0) return 0; // signed n returns 0 |
1032 |
|
|
|
1033 |
|
|
// 2 compares if n < 16, 3 compares otherwise (4 if signed or n > 1<<29) |
1034 |
|
✗ |
if (n < (1 << 14)) |
1035 |
|
✗ |
if (n < (1 << 4)) return 0 + log2_4[n ]; |
1036 |
|
✗ |
else if (n < (1 << 9)) return 5 + log2_4[n >> 5]; |
1037 |
|
✗ |
else return 10 + log2_4[n >> 10]; |
1038 |
|
✗ |
else if (n < (1 << 24)) |
1039 |
|
✗ |
if (n < (1 << 19)) return 15 + log2_4[n >> 15]; |
1040 |
|
✗ |
else return 20 + log2_4[n >> 20]; |
1041 |
|
✗ |
else if (n < (1 << 29)) return 25 + log2_4[n >> 25]; |
1042 |
|
✗ |
else return 30 + log2_4[n >> 30]; |
1043 |
|
|
} |
1044 |
|
|
|
1045 |
|
|
#ifndef M_PI |
1046 |
|
|
#define M_PI 3.14159265358979323846264f // from CRC |
1047 |
|
|
#endif |
1048 |
|
|
|
1049 |
|
|
// code length assigned to a value with no huffman encoding |
1050 |
|
|
#define NO_CODE 255 |
1051 |
|
|
|
1052 |
|
|
/////////////////////// LEAF SETUP FUNCTIONS ////////////////////////// |
1053 |
|
|
// |
1054 |
|
|
// these functions are only called at setup, and only a few times |
1055 |
|
|
// per file |
1056 |
|
|
|
1057 |
|
✗ |
static float float32_unpack(uint32 x) |
1058 |
|
|
{ |
1059 |
|
|
// from the specification |
1060 |
|
✗ |
uint32 mantissa = x & 0x1fffff; |
1061 |
|
|
uint32 sign = x & 0x80000000; |
1062 |
|
✗ |
uint32 exp = (x & 0x7fe00000) >> 21; |
1063 |
|
✗ |
double res = sign ? -(double)mantissa : (double)mantissa; |
1064 |
|
✗ |
return (float) ldexp((float)res, (int)exp-788); |
1065 |
|
|
} |
1066 |
|
|
|
1067 |
|
|
|
1068 |
|
|
// zlib & jpeg huffman tables assume that the output symbols |
1069 |
|
|
// can either be arbitrarily arranged, or have monotonically |
1070 |
|
|
// increasing frequencies--they rely on the lengths being sorted; |
1071 |
|
|
// this makes for a very simple generation algorithm. |
1072 |
|
|
// vorbis allows a huffman table with non-sorted lengths. This |
1073 |
|
|
// requires a more sophisticated construction, since symbols in |
1074 |
|
|
// order do not map to huffman codes "in order". |
1075 |
|
|
static void add_entry(Codebook *c, uint32 huff_code, int symbol, int count, int len, uint32 *values) |
1076 |
|
|
{ |
1077 |
|
✗ |
if (!c->sparse) { |
1078 |
|
✗ |
c->codewords [symbol] = huff_code; |
1079 |
|
|
} else { |
1080 |
|
✗ |
c->codewords [count] = huff_code; |
1081 |
|
✗ |
c->codeword_lengths[count] = len; |
1082 |
|
✗ |
values [count] = symbol; |
1083 |
|
|
} |
1084 |
|
|
} |
1085 |
|
|
|
1086 |
|
✗ |
static int compute_codewords(Codebook *c, uint8 *len, int n, uint32 *values) |
1087 |
|
|
{ |
1088 |
|
|
int i,k,m=0; |
1089 |
|
|
uint32 available[32]; |
1090 |
|
|
|
1091 |
|
|
memset(available, 0, sizeof(available)); |
1092 |
|
|
// find the first entry |
1093 |
|
✗ |
for (k=0; k < n; ++k) if (len[k] < NO_CODE) break; |
1094 |
|
✗ |
if (k == n) { assert(c->sorted_entries == 0); return TRUE; } |
1095 |
|
|
assert(len[k] < 32); // no error return required, code reading lens checks this |
1096 |
|
|
// add to the list |
1097 |
|
✗ |
add_entry(c, 0, k, m++, len[k], values); |
1098 |
|
|
// add all available leaves |
1099 |
|
✗ |
for (i=1; i <= len[k]; ++i) |
1100 |
|
✗ |
available[i] = 1U << (32-i); |
1101 |
|
|
// note that the above code treats the first case specially, |
1102 |
|
|
// but it's really the same as the following code, so they |
1103 |
|
|
// could probably be combined (except the initial code is 0, |
1104 |
|
|
// and I use 0 in available[] to mean 'empty') |
1105 |
|
✗ |
for (i=k+1; i < n; ++i) { |
1106 |
|
|
uint32 res; |
1107 |
|
✗ |
int z = len[i], y; |
1108 |
|
✗ |
if (z == NO_CODE) continue; |
1109 |
|
|
assert(z < 32); // no error return required, code reading lens checks this |
1110 |
|
|
// find lowest available leaf (should always be earliest, |
1111 |
|
|
// which is what the specification calls for) |
1112 |
|
|
// note that this property, and the fact we can never have |
1113 |
|
|
// more than one free leaf at a given level, isn't totally |
1114 |
|
|
// trivial to prove, but it seems true and the assert never |
1115 |
|
|
// fires, so! |
1116 |
|
✗ |
while (z > 0 && !available[z]) --z; |
1117 |
|
✗ |
if (z == 0) { return FALSE; } |
1118 |
|
✗ |
res = available[z]; |
1119 |
|
✗ |
available[z] = 0; |
1120 |
|
✗ |
add_entry(c, bit_reverse(res), i, m++, len[i], values); |
1121 |
|
|
// propagate availability up the tree |
1122 |
|
✗ |
if (z != len[i]) { |
1123 |
|
✗ |
for (y=len[i]; y > z; --y) { |
1124 |
|
|
assert(available[y] == 0); |
1125 |
|
✗ |
available[y] = res + (1 << (32-y)); |
1126 |
|
|
} |
1127 |
|
|
} |
1128 |
|
|
} |
1129 |
|
|
return TRUE; |
1130 |
|
|
} |
1131 |
|
|
|
1132 |
|
|
// accelerated huffman table allows fast O(1) match of all symbols |
1133 |
|
|
// of length <= STB_VORBIS_FAST_HUFFMAN_LENGTH |
1134 |
|
✗ |
static void compute_accelerated_huffman(Codebook *c) |
1135 |
|
|
{ |
1136 |
|
|
int i, len; |
1137 |
|
✗ |
for (i=0; i < FAST_HUFFMAN_TABLE_SIZE; ++i) |
1138 |
|
✗ |
c->fast_huffman[i] = -1; |
1139 |
|
|
|
1140 |
|
✗ |
len = c->sparse ? c->sorted_entries : c->entries; |
1141 |
|
|
#ifdef STB_VORBIS_FAST_HUFFMAN_SHORT |
1142 |
|
|
if (len > 32767) len = 32767; // largest possible value we can encode! |
1143 |
|
|
#endif |
1144 |
|
✗ |
for (i=0; i < len; ++i) { |
1145 |
|
✗ |
if (c->codeword_lengths[i] <= STB_VORBIS_FAST_HUFFMAN_LENGTH) { |
1146 |
|
✗ |
uint32 z = c->sparse ? bit_reverse(c->sorted_codewords[i]) : c->codewords[i]; |
1147 |
|
|
// set table entries for all bit combinations in the higher bits |
1148 |
|
✗ |
while (z < FAST_HUFFMAN_TABLE_SIZE) { |
1149 |
|
✗ |
c->fast_huffman[z] = i; |
1150 |
|
✗ |
z += 1 << c->codeword_lengths[i]; |
1151 |
|
|
} |
1152 |
|
|
} |
1153 |
|
|
} |
1154 |
|
|
} |
1155 |
|
|
|
1156 |
|
|
#ifdef _MSC_VER |
1157 |
|
|
#define STBV_CDECL __cdecl |
1158 |
|
|
#else |
1159 |
|
|
#define STBV_CDECL |
1160 |
|
|
#endif |
1161 |
|
|
|
1162 |
|
✗ |
static int STBV_CDECL uint32_compare(const void *p, const void *q) |
1163 |
|
|
{ |
1164 |
|
✗ |
uint32 x = * (uint32 *) p; |
1165 |
|
✗ |
uint32 y = * (uint32 *) q; |
1166 |
|
✗ |
return x < y ? -1 : x > y; |
1167 |
|
|
} |
1168 |
|
|
|
1169 |
|
|
static int include_in_sort(Codebook *c, uint8 len) |
1170 |
|
|
{ |
1171 |
|
✗ |
if (c->sparse) { assert(len != NO_CODE); return TRUE; } |
1172 |
|
✗ |
if (len == NO_CODE) return FALSE; |
1173 |
|
✗ |
if (len > STB_VORBIS_FAST_HUFFMAN_LENGTH) return TRUE; |
1174 |
|
|
return FALSE; |
1175 |
|
|
} |
1176 |
|
|
|
1177 |
|
|
// if the fast table above doesn't work, we want to binary |
1178 |
|
|
// search them... need to reverse the bits |
1179 |
|
✗ |
static void compute_sorted_huffman(Codebook *c, uint8 *lengths, uint32 *values) |
1180 |
|
|
{ |
1181 |
|
|
int i, len; |
1182 |
|
|
// build a list of all the entries |
1183 |
|
|
// OPTIMIZATION: don't include the short ones, since they'll be caught by FAST_HUFFMAN. |
1184 |
|
|
// this is kind of a frivolous optimization--I don't see any performance improvement, |
1185 |
|
|
// but it's like 4 extra lines of code, so. |
1186 |
|
✗ |
if (!c->sparse) { |
1187 |
|
|
int k = 0; |
1188 |
|
✗ |
for (i=0; i < c->entries; ++i) |
1189 |
|
✗ |
if (include_in_sort(c, lengths[i])) |
1190 |
|
✗ |
c->sorted_codewords[k++] = bit_reverse(c->codewords[i]); |
1191 |
|
|
assert(k == c->sorted_entries); |
1192 |
|
|
} else { |
1193 |
|
✗ |
for (i=0; i < c->sorted_entries; ++i) |
1194 |
|
✗ |
c->sorted_codewords[i] = bit_reverse(c->codewords[i]); |
1195 |
|
|
} |
1196 |
|
|
|
1197 |
|
✗ |
qsort(c->sorted_codewords, c->sorted_entries, sizeof(c->sorted_codewords[0]), uint32_compare); |
1198 |
|
✗ |
c->sorted_codewords[c->sorted_entries] = 0xffffffff; |
1199 |
|
|
|
1200 |
|
✗ |
len = c->sparse ? c->sorted_entries : c->entries; |
1201 |
|
|
// now we need to indicate how they correspond; we could either |
1202 |
|
|
// #1: sort a different data structure that says who they correspond to |
1203 |
|
|
// #2: for each sorted entry, search the original list to find who corresponds |
1204 |
|
|
// #3: for each original entry, find the sorted entry |
1205 |
|
|
// #1 requires extra storage, #2 is slow, #3 can use binary search! |
1206 |
|
✗ |
for (i=0; i < len; ++i) { |
1207 |
|
✗ |
int huff_len = c->sparse ? lengths[values[i]] : lengths[i]; |
1208 |
|
✗ |
if (include_in_sort(c,huff_len)) { |
1209 |
|
✗ |
uint32 code = bit_reverse(c->codewords[i]); |
1210 |
|
✗ |
int x=0, n=c->sorted_entries; |
1211 |
|
✗ |
while (n > 1) { |
1212 |
|
|
// invariant: sc[x] <= code < sc[x+n] |
1213 |
|
✗ |
int m = x + (n >> 1); |
1214 |
|
✗ |
if (c->sorted_codewords[m] <= code) { |
1215 |
|
|
x = m; |
1216 |
|
✗ |
n -= (n>>1); |
1217 |
|
|
} else { |
1218 |
|
|
n >>= 1; |
1219 |
|
|
} |
1220 |
|
|
} |
1221 |
|
|
assert(c->sorted_codewords[x] == code); |
1222 |
|
✗ |
if (c->sparse) { |
1223 |
|
✗ |
c->sorted_values[x] = values[i]; |
1224 |
|
✗ |
c->codeword_lengths[x] = huff_len; |
1225 |
|
|
} else { |
1226 |
|
✗ |
c->sorted_values[x] = i; |
1227 |
|
|
} |
1228 |
|
|
} |
1229 |
|
|
} |
1230 |
|
|
} |
1231 |
|
|
|
1232 |
|
|
// only run while parsing the header (3 times) |
1233 |
|
|
static int vorbis_validate(uint8 *data) |
1234 |
|
|
{ |
1235 |
|
|
static uint8 vorbis[6] = { 'v', 'o', 'r', 'b', 'i', 's' }; |
1236 |
|
✗ |
return memcmp(data, vorbis, 6) == 0; |
1237 |
|
|
} |
1238 |
|
|
|
1239 |
|
|
// called from setup only, once per code book |
1240 |
|
|
// (formula implied by specification) |
1241 |
|
✗ |
static int lookup1_values(int entries, int dim) |
1242 |
|
|
{ |
1243 |
|
✗ |
int r = (int) floor(exp((float) log((float) entries) / dim)); |
1244 |
|
✗ |
if ((int) floor(pow((float) r+1, dim)) <= entries) // (int) cast for MinGW warning; |
1245 |
|
✗ |
++r; // floor() to avoid _ftol() when non-CRT |
1246 |
|
✗ |
if (pow((float) r+1, dim) <= entries) |
1247 |
|
|
return -1; |
1248 |
|
✗ |
if ((int) floor(pow((float) r, dim)) > entries) |
1249 |
|
✗ |
return -1; |
1250 |
|
|
return r; |
1251 |
|
|
} |
1252 |
|
|
|
1253 |
|
|
// called twice per file |
1254 |
|
✗ |
static void compute_twiddle_factors(int n, float *A, float *B, float *C) |
1255 |
|
|
{ |
1256 |
|
✗ |
int n4 = n >> 2, n8 = n >> 3; |
1257 |
|
|
int k,k2; |
1258 |
|
|
|
1259 |
|
✗ |
for (k=k2=0; k < n4; ++k,k2+=2) { |
1260 |
|
✗ |
A[k2 ] = (float) cos(4*k*M_PI/n); |
1261 |
|
✗ |
A[k2+1] = (float) -sin(4*k*M_PI/n); |
1262 |
|
✗ |
B[k2 ] = (float) cos((k2+1)*M_PI/n/2) * 0.5f; |
1263 |
|
✗ |
B[k2+1] = (float) sin((k2+1)*M_PI/n/2) * 0.5f; |
1264 |
|
|
} |
1265 |
|
✗ |
for (k=k2=0; k < n8; ++k,k2+=2) { |
1266 |
|
✗ |
C[k2 ] = (float) cos(2*(k2+1)*M_PI/n); |
1267 |
|
✗ |
C[k2+1] = (float) -sin(2*(k2+1)*M_PI/n); |
1268 |
|
|
} |
1269 |
|
|
} |
1270 |
|
|
|
1271 |
|
✗ |
static void compute_window(int n, float *window) |
1272 |
|
|
{ |
1273 |
|
✗ |
int n2 = n >> 1, i; |
1274 |
|
✗ |
for (i=0; i < n2; ++i) |
1275 |
|
✗ |
window[i] = (float) sin(0.5 * M_PI * square((float) sin((i - 0 + 0.5) / n2 * 0.5 * M_PI))); |
1276 |
|
|
} |
1277 |
|
|
|
1278 |
|
✗ |
static void compute_bitreverse(int n, uint16 *rev) |
1279 |
|
|
{ |
1280 |
|
✗ |
int ld = ilog(n) - 1; // ilog is off-by-one from normal definitions |
1281 |
|
✗ |
int i, n8 = n >> 3; |
1282 |
|
✗ |
for (i=0; i < n8; ++i) |
1283 |
|
✗ |
rev[i] = (bit_reverse(i) >> (32-ld+3)) << 2; |
1284 |
|
|
} |
1285 |
|
|
|
1286 |
|
✗ |
static int init_blocksize(vorb *f, int b, int n) |
1287 |
|
|
{ |
1288 |
|
✗ |
int n2 = n >> 1, n4 = n >> 2, n8 = n >> 3; |
1289 |
|
✗ |
f->A[b] = (float *) setup_malloc(f, sizeof(float) * n2); |
1290 |
|
✗ |
f->B[b] = (float *) setup_malloc(f, sizeof(float) * n2); |
1291 |
|
✗ |
f->C[b] = (float *) setup_malloc(f, sizeof(float) * n4); |
1292 |
|
✗ |
if (!f->A[b] || !f->B[b] || !f->C[b]) return error(f, VORBIS_outofmem); |
1293 |
|
✗ |
compute_twiddle_factors(n, f->A[b], f->B[b], f->C[b]); |
1294 |
|
✗ |
f->window[b] = (float *) setup_malloc(f, sizeof(float) * n2); |
1295 |
|
✗ |
if (!f->window[b]) return error(f, VORBIS_outofmem); |
1296 |
|
✗ |
compute_window(n, f->window[b]); |
1297 |
|
✗ |
f->bit_reverse[b] = (uint16 *) setup_malloc(f, sizeof(uint16) * n8); |
1298 |
|
✗ |
if (!f->bit_reverse[b]) return error(f, VORBIS_outofmem); |
1299 |
|
✗ |
compute_bitreverse(n, f->bit_reverse[b]); |
1300 |
|
✗ |
return TRUE; |
1301 |
|
|
} |
1302 |
|
|
|
1303 |
|
✗ |
static void neighbors(uint16 *x, int n, int *plow, int *phigh) |
1304 |
|
|
{ |
1305 |
|
|
int low = -1; |
1306 |
|
|
int high = 65536; |
1307 |
|
|
int i; |
1308 |
|
✗ |
for (i=0; i < n; ++i) { |
1309 |
|
✗ |
if (x[i] > low && x[i] < x[n]) { *plow = i; low = x[i]; } |
1310 |
|
✗ |
if (x[i] < high && x[i] > x[n]) { *phigh = i; high = x[i]; } |
1311 |
|
|
} |
1312 |
|
|
} |
1313 |
|
|
|
1314 |
|
|
// this has been repurposed so y is now the original index instead of y |
1315 |
|
|
typedef struct |
1316 |
|
|
{ |
1317 |
|
|
uint16 x,id; |
1318 |
|
|
} stbv__floor_ordering; |
1319 |
|
|
|
1320 |
|
✗ |
static int STBV_CDECL point_compare(const void *p, const void *q) |
1321 |
|
|
{ |
1322 |
|
|
stbv__floor_ordering *a = (stbv__floor_ordering *) p; |
1323 |
|
|
stbv__floor_ordering *b = (stbv__floor_ordering *) q; |
1324 |
|
✗ |
return a->x < b->x ? -1 : a->x > b->x; |
1325 |
|
|
} |
1326 |
|
|
|
1327 |
|
|
// |
1328 |
|
|
/////////////////////// END LEAF SETUP FUNCTIONS ////////////////////////// |
1329 |
|
|
|
1330 |
|
|
|
1331 |
|
|
#if defined(STB_VORBIS_NO_STDIO) |
1332 |
|
|
#define USE_MEMORY(z) TRUE |
1333 |
|
|
#else |
1334 |
|
|
#define USE_MEMORY(z) ((z)->stream) |
1335 |
|
|
#endif |
1336 |
|
|
|
1337 |
|
✗ |
static uint8 get8(vorb *z) |
1338 |
|
|
{ |
1339 |
|
✗ |
if (USE_MEMORY(z)) { |
1340 |
|
✗ |
if (z->stream >= z->stream_end) { z->eof = TRUE; return 0; } |
1341 |
|
✗ |
return *z->stream++; |
1342 |
|
|
} |
1343 |
|
|
|
1344 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
1345 |
|
|
{ |
1346 |
|
✗ |
int c = fgetc(z->f); |
1347 |
|
✗ |
if (c == EOF) { z->eof = TRUE; return 0; } |
1348 |
|
✗ |
return c; |
1349 |
|
|
} |
1350 |
|
|
#endif |
1351 |
|
|
} |
1352 |
|
|
|
1353 |
|
✗ |
static uint32 get32(vorb *f) |
1354 |
|
|
{ |
1355 |
|
|
uint32 x; |
1356 |
|
✗ |
x = get8(f); |
1357 |
|
✗ |
x += get8(f) << 8; |
1358 |
|
✗ |
x += get8(f) << 16; |
1359 |
|
✗ |
x += (uint32) get8(f) << 24; |
1360 |
|
✗ |
return x; |
1361 |
|
|
} |
1362 |
|
|
|
1363 |
|
✗ |
static int getn(vorb *z, uint8 *data, int n) |
1364 |
|
|
{ |
1365 |
|
✗ |
if (USE_MEMORY(z)) { |
1366 |
|
✗ |
if (z->stream+n > z->stream_end) { z->eof = 1; return 0; } |
1367 |
|
|
memcpy(data, z->stream, n); |
1368 |
|
✗ |
z->stream += n; |
1369 |
|
✗ |
return 1; |
1370 |
|
|
} |
1371 |
|
|
|
1372 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
1373 |
|
✗ |
if (fread(data, n, 1, z->f) == 1) |
1374 |
|
|
return 1; |
1375 |
|
|
else { |
1376 |
|
✗ |
z->eof = 1; |
1377 |
|
✗ |
return 0; |
1378 |
|
|
} |
1379 |
|
|
#endif |
1380 |
|
|
} |
1381 |
|
|
|
1382 |
|
✗ |
static void skip(vorb *z, int n) |
1383 |
|
|
{ |
1384 |
|
✗ |
if (USE_MEMORY(z)) { |
1385 |
|
✗ |
z->stream += n; |
1386 |
|
✗ |
if (z->stream >= z->stream_end) z->eof = 1; |
1387 |
|
✗ |
return; |
1388 |
|
|
} |
1389 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
1390 |
|
|
{ |
1391 |
|
✗ |
long x = ftell(z->f); |
1392 |
|
✗ |
fseek(z->f, x+n, SEEK_SET); |
1393 |
|
|
} |
1394 |
|
|
#endif |
1395 |
|
|
} |
1396 |
|
|
|
1397 |
|
✗ |
static int set_file_offset(stb_vorbis *f, unsigned int loc) |
1398 |
|
|
{ |
1399 |
|
|
#ifndef STB_VORBIS_NO_PUSHDATA_API |
1400 |
|
✗ |
if (f->push_mode) return 0; |
1401 |
|
|
#endif |
1402 |
|
✗ |
f->eof = 0; |
1403 |
|
✗ |
if (USE_MEMORY(f)) { |
1404 |
|
✗ |
if (f->stream_start + loc >= f->stream_end || f->stream_start + loc < f->stream_start) { |
1405 |
|
✗ |
f->stream = f->stream_end; |
1406 |
|
✗ |
f->eof = 1; |
1407 |
|
✗ |
return 0; |
1408 |
|
|
} else { |
1409 |
|
✗ |
f->stream = f->stream_start + loc; |
1410 |
|
✗ |
return 1; |
1411 |
|
|
} |
1412 |
|
|
} |
1413 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
1414 |
|
✗ |
if (loc + f->f_start < loc || loc >= 0x80000000) { |
1415 |
|
|
loc = 0x7fffffff; |
1416 |
|
✗ |
f->eof = 1; |
1417 |
|
|
} else { |
1418 |
|
|
loc += f->f_start; |
1419 |
|
|
} |
1420 |
|
✗ |
if (!fseek(f->f, loc, SEEK_SET)) |
1421 |
|
|
return 1; |
1422 |
|
✗ |
f->eof = 1; |
1423 |
|
✗ |
fseek(f->f, f->f_start, SEEK_END); |
1424 |
|
✗ |
return 0; |
1425 |
|
|
#endif |
1426 |
|
|
} |
1427 |
|
|
|
1428 |
|
|
|
1429 |
|
|
static uint8 ogg_page_header[4] = { 0x4f, 0x67, 0x67, 0x53 }; |
1430 |
|
|
|
1431 |
|
✗ |
static int capture_pattern(vorb *f) |
1432 |
|
|
{ |
1433 |
|
✗ |
if (0x4f != get8(f)) return FALSE; |
1434 |
|
✗ |
if (0x67 != get8(f)) return FALSE; |
1435 |
|
✗ |
if (0x67 != get8(f)) return FALSE; |
1436 |
|
✗ |
if (0x53 != get8(f)) return FALSE; |
1437 |
|
|
return TRUE; |
1438 |
|
|
} |
1439 |
|
|
|
1440 |
|
|
#define PAGEFLAG_continued_packet 1 |
1441 |
|
|
#define PAGEFLAG_first_page 2 |
1442 |
|
|
#define PAGEFLAG_last_page 4 |
1443 |
|
|
|
1444 |
|
✗ |
static int start_page_no_capturepattern(vorb *f) |
1445 |
|
|
{ |
1446 |
|
|
uint32 loc0,loc1,n; |
1447 |
|
✗ |
if (f->first_decode && !IS_PUSH_MODE(f)) { |
1448 |
|
✗ |
f->p_first.page_start = stb_vorbis_get_file_offset(f) - 4; |
1449 |
|
|
} |
1450 |
|
|
// stream structure version |
1451 |
|
✗ |
if (0 != get8(f)) return error(f, VORBIS_invalid_stream_structure_version); |
1452 |
|
|
// header flag |
1453 |
|
✗ |
f->page_flag = get8(f); |
1454 |
|
|
// absolute granule position |
1455 |
|
✗ |
loc0 = get32(f); |
1456 |
|
✗ |
loc1 = get32(f); |
1457 |
|
|
// @TODO: validate loc0,loc1 as valid positions? |
1458 |
|
|
// stream serial number -- vorbis doesn't interleave, so discard |
1459 |
|
✗ |
get32(f); |
1460 |
|
|
//if (f->serial != get32(f)) return error(f, VORBIS_incorrect_stream_serial_number); |
1461 |
|
|
// page sequence number |
1462 |
|
✗ |
n = get32(f); |
1463 |
|
✗ |
f->last_page = n; |
1464 |
|
|
// CRC32 |
1465 |
|
✗ |
get32(f); |
1466 |
|
|
// page_segments |
1467 |
|
✗ |
f->segment_count = get8(f); |
1468 |
|
✗ |
if (!getn(f, f->segments, f->segment_count)) |
1469 |
|
✗ |
return error(f, VORBIS_unexpected_eof); |
1470 |
|
|
// assume we _don't_ know any the sample position of any segments |
1471 |
|
✗ |
f->end_seg_with_known_loc = -2; |
1472 |
|
✗ |
if (loc0 != ~0U || loc1 != ~0U) { |
1473 |
|
|
int i; |
1474 |
|
|
// determine which packet is the last one that will complete |
1475 |
|
✗ |
for (i=f->segment_count-1; i >= 0; --i) |
1476 |
|
✗ |
if (f->segments[i] < 255) |
1477 |
|
|
break; |
1478 |
|
|
// 'i' is now the index of the _last_ segment of a packet that ends |
1479 |
|
✗ |
if (i >= 0) { |
1480 |
|
✗ |
f->end_seg_with_known_loc = i; |
1481 |
|
✗ |
f->known_loc_for_packet = loc0; |
1482 |
|
|
} |
1483 |
|
|
} |
1484 |
|
✗ |
if (f->first_decode) { |
1485 |
|
|
int i,len; |
1486 |
|
|
len = 0; |
1487 |
|
✗ |
for (i=0; i < f->segment_count; ++i) |
1488 |
|
✗ |
len += f->segments[i]; |
1489 |
|
✗ |
len += 27 + f->segment_count; |
1490 |
|
✗ |
f->p_first.page_end = f->p_first.page_start + len; |
1491 |
|
✗ |
f->p_first.last_decoded_sample = loc0; |
1492 |
|
|
} |
1493 |
|
✗ |
f->next_seg = 0; |
1494 |
|
✗ |
return TRUE; |
1495 |
|
|
} |
1496 |
|
|
|
1497 |
|
✗ |
static int start_page(vorb *f) |
1498 |
|
|
{ |
1499 |
|
✗ |
if (!capture_pattern(f)) return error(f, VORBIS_missing_capture_pattern); |
1500 |
|
✗ |
return start_page_no_capturepattern(f); |
1501 |
|
|
} |
1502 |
|
|
|
1503 |
|
✗ |
static int start_packet(vorb *f) |
1504 |
|
|
{ |
1505 |
|
✗ |
while (f->next_seg == -1) { |
1506 |
|
✗ |
if (!start_page(f)) return FALSE; |
1507 |
|
✗ |
if (f->page_flag & PAGEFLAG_continued_packet) |
1508 |
|
✗ |
return error(f, VORBIS_continued_packet_flag_invalid); |
1509 |
|
|
} |
1510 |
|
✗ |
f->last_seg = FALSE; |
1511 |
|
✗ |
f->valid_bits = 0; |
1512 |
|
✗ |
f->packet_bytes = 0; |
1513 |
|
✗ |
f->bytes_in_seg = 0; |
1514 |
|
|
// f->next_seg is now valid |
1515 |
|
✗ |
return TRUE; |
1516 |
|
|
} |
1517 |
|
|
|
1518 |
|
✗ |
static int maybe_start_packet(vorb *f) |
1519 |
|
|
{ |
1520 |
|
✗ |
if (f->next_seg == -1) { |
1521 |
|
✗ |
int x = get8(f); |
1522 |
|
✗ |
if (f->eof) return FALSE; // EOF at page boundary is not an error! |
1523 |
|
✗ |
if (0x4f != x ) return error(f, VORBIS_missing_capture_pattern); |
1524 |
|
✗ |
if (0x67 != get8(f)) return error(f, VORBIS_missing_capture_pattern); |
1525 |
|
✗ |
if (0x67 != get8(f)) return error(f, VORBIS_missing_capture_pattern); |
1526 |
|
✗ |
if (0x53 != get8(f)) return error(f, VORBIS_missing_capture_pattern); |
1527 |
|
✗ |
if (!start_page_no_capturepattern(f)) return FALSE; |
1528 |
|
✗ |
if (f->page_flag & PAGEFLAG_continued_packet) { |
1529 |
|
|
// set up enough state that we can read this packet if we want, |
1530 |
|
|
// e.g. during recovery |
1531 |
|
✗ |
f->last_seg = FALSE; |
1532 |
|
✗ |
f->bytes_in_seg = 0; |
1533 |
|
✗ |
return error(f, VORBIS_continued_packet_flag_invalid); |
1534 |
|
|
} |
1535 |
|
|
} |
1536 |
|
✗ |
return start_packet(f); |
1537 |
|
|
} |
1538 |
|
|
|
1539 |
|
✗ |
static int next_segment(vorb *f) |
1540 |
|
|
{ |
1541 |
|
|
int len; |
1542 |
|
✗ |
if (f->last_seg) return 0; |
1543 |
|
✗ |
if (f->next_seg == -1) { |
1544 |
|
✗ |
f->last_seg_which = f->segment_count-1; // in case start_page fails |
1545 |
|
✗ |
if (!start_page(f)) { f->last_seg = 1; return 0; } |
1546 |
|
✗ |
if (!(f->page_flag & PAGEFLAG_continued_packet)) return error(f, VORBIS_continued_packet_flag_invalid); |
1547 |
|
|
} |
1548 |
|
✗ |
len = f->segments[f->next_seg++]; |
1549 |
|
✗ |
if (len < 255) { |
1550 |
|
✗ |
f->last_seg = TRUE; |
1551 |
|
✗ |
f->last_seg_which = f->next_seg-1; |
1552 |
|
|
} |
1553 |
|
✗ |
if (f->next_seg >= f->segment_count) |
1554 |
|
✗ |
f->next_seg = -1; |
1555 |
|
|
assert(f->bytes_in_seg == 0); |
1556 |
|
✗ |
f->bytes_in_seg = len; |
1557 |
|
✗ |
return len; |
1558 |
|
|
} |
1559 |
|
|
|
1560 |
|
|
#define EOP (-1) |
1561 |
|
|
#define INVALID_BITS (-1) |
1562 |
|
|
|
1563 |
|
✗ |
static int get8_packet_raw(vorb *f) |
1564 |
|
|
{ |
1565 |
|
✗ |
if (!f->bytes_in_seg) { // CLANG! |
1566 |
|
✗ |
if (f->last_seg) return EOP; |
1567 |
|
✗ |
else if (!next_segment(f)) return EOP; |
1568 |
|
|
} |
1569 |
|
|
assert(f->bytes_in_seg > 0); |
1570 |
|
✗ |
--f->bytes_in_seg; |
1571 |
|
✗ |
++f->packet_bytes; |
1572 |
|
✗ |
return get8(f); |
1573 |
|
|
} |
1574 |
|
|
|
1575 |
|
|
static int get8_packet(vorb *f) |
1576 |
|
|
{ |
1577 |
|
✗ |
int x = get8_packet_raw(f); |
1578 |
|
✗ |
f->valid_bits = 0; |
1579 |
|
|
return x; |
1580 |
|
|
} |
1581 |
|
|
|
1582 |
|
✗ |
static int get32_packet(vorb *f) |
1583 |
|
|
{ |
1584 |
|
|
uint32 x; |
1585 |
|
✗ |
x = get8_packet(f); |
1586 |
|
✗ |
x += get8_packet(f) << 8; |
1587 |
|
✗ |
x += get8_packet(f) << 16; |
1588 |
|
✗ |
x += (uint32) get8_packet(f) << 24; |
1589 |
|
✗ |
return x; |
1590 |
|
|
} |
1591 |
|
|
|
1592 |
|
|
static void flush_packet(vorb *f) |
1593 |
|
|
{ |
1594 |
|
✗ |
while (get8_packet_raw(f) != EOP); |
1595 |
|
|
} |
1596 |
|
|
|
1597 |
|
|
// @OPTIMIZE: this is the secondary bit decoder, so it's probably not as important |
1598 |
|
|
// as the huffman decoder? |
1599 |
|
✗ |
static uint32 get_bits(vorb *f, int n) |
1600 |
|
|
{ |
1601 |
|
|
uint32 z; |
1602 |
|
|
|
1603 |
|
✗ |
if (f->valid_bits < 0) return 0; |
1604 |
|
✗ |
if (f->valid_bits < n) { |
1605 |
|
✗ |
if (n > 24) { |
1606 |
|
|
// the accumulator technique below would not work correctly in this case |
1607 |
|
✗ |
z = get_bits(f, 24); |
1608 |
|
✗ |
z += get_bits(f, n-24) << 24; |
1609 |
|
✗ |
return z; |
1610 |
|
|
} |
1611 |
|
✗ |
if (f->valid_bits == 0) f->acc = 0; |
1612 |
|
✗ |
while (f->valid_bits < n) { |
1613 |
|
✗ |
int z = get8_packet_raw(f); |
1614 |
|
✗ |
if (z == EOP) { |
1615 |
|
✗ |
f->valid_bits = INVALID_BITS; |
1616 |
|
✗ |
return 0; |
1617 |
|
|
} |
1618 |
|
✗ |
f->acc += z << f->valid_bits; |
1619 |
|
✗ |
f->valid_bits += 8; |
1620 |
|
|
} |
1621 |
|
|
} |
1622 |
|
|
|
1623 |
|
|
assert(f->valid_bits >= n); |
1624 |
|
✗ |
z = f->acc & ((1 << n)-1); |
1625 |
|
✗ |
f->acc >>= n; |
1626 |
|
✗ |
f->valid_bits -= n; |
1627 |
|
✗ |
return z; |
1628 |
|
|
} |
1629 |
|
|
|
1630 |
|
|
// @OPTIMIZE: primary accumulator for huffman |
1631 |
|
|
// expand the buffer to as many bits as possible without reading off end of packet |
1632 |
|
|
// it might be nice to allow f->valid_bits and f->acc to be stored in registers, |
1633 |
|
|
// e.g. cache them locally and decode locally |
1634 |
|
✗ |
static __forceinline void prep_huffman(vorb *f) |
1635 |
|
|
{ |
1636 |
|
✗ |
if (f->valid_bits <= 24) { |
1637 |
|
✗ |
if (f->valid_bits == 0) f->acc = 0; |
1638 |
|
|
do { |
1639 |
|
|
int z; |
1640 |
|
✗ |
if (f->last_seg && !f->bytes_in_seg) return; |
1641 |
|
✗ |
z = get8_packet_raw(f); |
1642 |
|
✗ |
if (z == EOP) return; |
1643 |
|
✗ |
f->acc += (unsigned) z << f->valid_bits; |
1644 |
|
✗ |
f->valid_bits += 8; |
1645 |
|
✗ |
} while (f->valid_bits <= 24); |
1646 |
|
|
} |
1647 |
|
|
} |
1648 |
|
|
|
1649 |
|
|
enum |
1650 |
|
|
{ |
1651 |
|
|
VORBIS_packet_id = 1, |
1652 |
|
|
VORBIS_packet_comment = 3, |
1653 |
|
|
VORBIS_packet_setup = 5 |
1654 |
|
|
}; |
1655 |
|
|
|
1656 |
|
✗ |
static int codebook_decode_scalar_raw(vorb *f, Codebook *c) |
1657 |
|
|
{ |
1658 |
|
|
int i; |
1659 |
|
✗ |
prep_huffman(f); |
1660 |
|
|
|
1661 |
|
✗ |
if (c->codewords == NULL && c->sorted_codewords == NULL) |
1662 |
|
|
return -1; |
1663 |
|
|
|
1664 |
|
|
// cases to use binary search: sorted_codewords && !c->codewords |
1665 |
|
|
// sorted_codewords && c->entries > 8 |
1666 |
|
✗ |
if (c->entries > 8 ? c->sorted_codewords!=NULL : !c->codewords) { |
1667 |
|
|
// binary search |
1668 |
|
✗ |
uint32 code = bit_reverse(f->acc); |
1669 |
|
✗ |
int x=0, n=c->sorted_entries, len; |
1670 |
|
|
|
1671 |
|
✗ |
while (n > 1) { |
1672 |
|
|
// invariant: sc[x] <= code < sc[x+n] |
1673 |
|
✗ |
int m = x + (n >> 1); |
1674 |
|
✗ |
if (c->sorted_codewords[m] <= code) { |
1675 |
|
|
x = m; |
1676 |
|
✗ |
n -= (n>>1); |
1677 |
|
|
} else { |
1678 |
|
|
n >>= 1; |
1679 |
|
|
} |
1680 |
|
|
} |
1681 |
|
|
// x is now the sorted index |
1682 |
|
✗ |
if (!c->sparse) x = c->sorted_values[x]; |
1683 |
|
|
// x is now sorted index if sparse, or symbol otherwise |
1684 |
|
✗ |
len = c->codeword_lengths[x]; |
1685 |
|
✗ |
if (f->valid_bits >= len) { |
1686 |
|
✗ |
f->acc >>= len; |
1687 |
|
✗ |
f->valid_bits -= len; |
1688 |
|
✗ |
return x; |
1689 |
|
|
} |
1690 |
|
|
|
1691 |
|
✗ |
f->valid_bits = 0; |
1692 |
|
✗ |
return -1; |
1693 |
|
|
} |
1694 |
|
|
|
1695 |
|
|
// if small, linear search |
1696 |
|
|
assert(!c->sparse); |
1697 |
|
✗ |
for (i=0; i < c->entries; ++i) { |
1698 |
|
✗ |
if (c->codeword_lengths[i] == NO_CODE) continue; |
1699 |
|
✗ |
if (c->codewords[i] == (f->acc & ((1 << c->codeword_lengths[i])-1))) { |
1700 |
|
✗ |
if (f->valid_bits >= c->codeword_lengths[i]) { |
1701 |
|
✗ |
f->acc >>= c->codeword_lengths[i]; |
1702 |
|
✗ |
f->valid_bits -= c->codeword_lengths[i]; |
1703 |
|
✗ |
return i; |
1704 |
|
|
} |
1705 |
|
✗ |
f->valid_bits = 0; |
1706 |
|
✗ |
return -1; |
1707 |
|
|
} |
1708 |
|
|
} |
1709 |
|
|
|
1710 |
|
|
error(f, VORBIS_invalid_stream); |
1711 |
|
✗ |
f->valid_bits = 0; |
1712 |
|
✗ |
return -1; |
1713 |
|
|
} |
1714 |
|
|
|
1715 |
|
|
#ifndef STB_VORBIS_NO_INLINE_DECODE |
1716 |
|
|
|
1717 |
|
|
#define DECODE_RAW(var, f,c) \ |
1718 |
|
|
if (f->valid_bits < STB_VORBIS_FAST_HUFFMAN_LENGTH) \ |
1719 |
|
|
prep_huffman(f); \ |
1720 |
|
|
var = f->acc & FAST_HUFFMAN_TABLE_MASK; \ |
1721 |
|
|
var = c->fast_huffman[var]; \ |
1722 |
|
|
if (var >= 0) { \ |
1723 |
|
|
int n = c->codeword_lengths[var]; \ |
1724 |
|
|
f->acc >>= n; \ |
1725 |
|
|
f->valid_bits -= n; \ |
1726 |
|
|
if (f->valid_bits < 0) { f->valid_bits = 0; var = -1; } \ |
1727 |
|
|
} else { \ |
1728 |
|
|
var = codebook_decode_scalar_raw(f,c); \ |
1729 |
|
|
} |
1730 |
|
|
|
1731 |
|
|
#else |
1732 |
|
|
|
1733 |
|
|
static int codebook_decode_scalar(vorb *f, Codebook *c) |
1734 |
|
|
{ |
1735 |
|
|
int i; |
1736 |
|
|
if (f->valid_bits < STB_VORBIS_FAST_HUFFMAN_LENGTH) |
1737 |
|
|
prep_huffman(f); |
1738 |
|
|
// fast huffman table lookup |
1739 |
|
|
i = f->acc & FAST_HUFFMAN_TABLE_MASK; |
1740 |
|
|
i = c->fast_huffman[i]; |
1741 |
|
|
if (i >= 0) { |
1742 |
|
|
f->acc >>= c->codeword_lengths[i]; |
1743 |
|
|
f->valid_bits -= c->codeword_lengths[i]; |
1744 |
|
|
if (f->valid_bits < 0) { f->valid_bits = 0; return -1; } |
1745 |
|
|
return i; |
1746 |
|
|
} |
1747 |
|
|
return codebook_decode_scalar_raw(f,c); |
1748 |
|
|
} |
1749 |
|
|
|
1750 |
|
|
#define DECODE_RAW(var,f,c) var = codebook_decode_scalar(f,c); |
1751 |
|
|
|
1752 |
|
|
#endif |
1753 |
|
|
|
1754 |
|
|
#define DECODE(var,f,c) \ |
1755 |
|
|
DECODE_RAW(var,f,c) \ |
1756 |
|
|
if (c->sparse) var = c->sorted_values[var]; |
1757 |
|
|
|
1758 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK |
1759 |
|
|
#define DECODE_VQ(var,f,c) DECODE_RAW(var,f,c) |
1760 |
|
|
#else |
1761 |
|
|
#define DECODE_VQ(var,f,c) DECODE(var,f,c) |
1762 |
|
|
#endif |
1763 |
|
|
|
1764 |
|
|
|
1765 |
|
|
|
1766 |
|
|
|
1767 |
|
|
|
1768 |
|
|
|
1769 |
|
|
// CODEBOOK_ELEMENT_FAST is an optimization for the CODEBOOK_FLOATS case |
1770 |
|
|
// where we avoid one addition |
1771 |
|
|
#define CODEBOOK_ELEMENT(c,off) (c->multiplicands[off]) |
1772 |
|
|
#define CODEBOOK_ELEMENT_FAST(c,off) (c->multiplicands[off]) |
1773 |
|
|
#define CODEBOOK_ELEMENT_BASE(c) (0) |
1774 |
|
|
|
1775 |
|
✗ |
static int codebook_decode_start(vorb *f, Codebook *c) |
1776 |
|
|
{ |
1777 |
|
|
int z = -1; |
1778 |
|
|
|
1779 |
|
|
// type 0 is only legal in a scalar context |
1780 |
|
✗ |
if (c->lookup_type == 0) |
1781 |
|
|
error(f, VORBIS_invalid_stream); |
1782 |
|
|
else { |
1783 |
|
✗ |
DECODE_VQ(z,f,c); |
1784 |
|
|
if (c->sparse) assert(z < c->sorted_entries); |
1785 |
|
✗ |
if (z < 0) { // check for EOP |
1786 |
|
✗ |
if (!f->bytes_in_seg) |
1787 |
|
✗ |
if (f->last_seg) |
1788 |
|
|
return z; |
1789 |
|
|
error(f, VORBIS_invalid_stream); |
1790 |
|
|
} |
1791 |
|
|
} |
1792 |
|
|
return z; |
1793 |
|
|
} |
1794 |
|
|
|
1795 |
|
✗ |
static int codebook_decode(vorb *f, Codebook *c, float *output, int len) |
1796 |
|
|
{ |
1797 |
|
✗ |
int i,z = codebook_decode_start(f,c); |
1798 |
|
✗ |
if (z < 0) return FALSE; |
1799 |
|
✗ |
if (len > c->dimensions) len = c->dimensions; |
1800 |
|
|
|
1801 |
|
|
#ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK |
1802 |
|
|
if (c->lookup_type == 1) { |
1803 |
|
|
float last = CODEBOOK_ELEMENT_BASE(c); |
1804 |
|
|
int div = 1; |
1805 |
|
|
for (i=0; i < len; ++i) { |
1806 |
|
|
int off = (z / div) % c->lookup_values; |
1807 |
|
|
float val = CODEBOOK_ELEMENT_FAST(c,off) + last; |
1808 |
|
|
output[i] += val; |
1809 |
|
|
if (c->sequence_p) last = val + c->minimum_value; |
1810 |
|
|
div *= c->lookup_values; |
1811 |
|
|
} |
1812 |
|
|
return TRUE; |
1813 |
|
|
} |
1814 |
|
|
#endif |
1815 |
|
|
|
1816 |
|
✗ |
z *= c->dimensions; |
1817 |
|
✗ |
if (c->sequence_p) { |
1818 |
|
|
float last = CODEBOOK_ELEMENT_BASE(c); |
1819 |
|
✗ |
for (i=0; i < len; ++i) { |
1820 |
|
✗ |
float val = CODEBOOK_ELEMENT_FAST(c,z+i) + last; |
1821 |
|
✗ |
output[i] += val; |
1822 |
|
✗ |
last = val + c->minimum_value; |
1823 |
|
|
} |
1824 |
|
|
} else { |
1825 |
|
|
float last = CODEBOOK_ELEMENT_BASE(c); |
1826 |
|
✗ |
for (i=0; i < len; ++i) { |
1827 |
|
✗ |
output[i] += CODEBOOK_ELEMENT_FAST(c,z+i) + last; |
1828 |
|
|
} |
1829 |
|
|
} |
1830 |
|
|
|
1831 |
|
|
return TRUE; |
1832 |
|
|
} |
1833 |
|
|
|
1834 |
|
✗ |
static int codebook_decode_step(vorb *f, Codebook *c, float *output, int len, int step) |
1835 |
|
|
{ |
1836 |
|
✗ |
int i,z = codebook_decode_start(f,c); |
1837 |
|
|
float last = CODEBOOK_ELEMENT_BASE(c); |
1838 |
|
✗ |
if (z < 0) return FALSE; |
1839 |
|
✗ |
if (len > c->dimensions) len = c->dimensions; |
1840 |
|
|
|
1841 |
|
|
#ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK |
1842 |
|
|
if (c->lookup_type == 1) { |
1843 |
|
|
int div = 1; |
1844 |
|
|
for (i=0; i < len; ++i) { |
1845 |
|
|
int off = (z / div) % c->lookup_values; |
1846 |
|
|
float val = CODEBOOK_ELEMENT_FAST(c,off) + last; |
1847 |
|
|
output[i*step] += val; |
1848 |
|
|
if (c->sequence_p) last = val; |
1849 |
|
|
div *= c->lookup_values; |
1850 |
|
|
} |
1851 |
|
|
return TRUE; |
1852 |
|
|
} |
1853 |
|
|
#endif |
1854 |
|
|
|
1855 |
|
✗ |
z *= c->dimensions; |
1856 |
|
✗ |
for (i=0; i < len; ++i) { |
1857 |
|
✗ |
float val = CODEBOOK_ELEMENT_FAST(c,z+i) + last; |
1858 |
|
✗ |
output[i*step] += val; |
1859 |
|
✗ |
if (c->sequence_p) last = val; |
1860 |
|
|
} |
1861 |
|
|
|
1862 |
|
|
return TRUE; |
1863 |
|
|
} |
1864 |
|
|
|
1865 |
|
✗ |
static int codebook_decode_deinterleave_repeat(vorb *f, Codebook *c, float **outputs, int ch, int *c_inter_p, int *p_inter_p, int len, int total_decode) |
1866 |
|
|
{ |
1867 |
|
✗ |
int c_inter = *c_inter_p; |
1868 |
|
✗ |
int p_inter = *p_inter_p; |
1869 |
|
✗ |
int i,z, effective = c->dimensions; |
1870 |
|
|
|
1871 |
|
|
// type 0 is only legal in a scalar context |
1872 |
|
✗ |
if (c->lookup_type == 0) return error(f, VORBIS_invalid_stream); |
1873 |
|
|
|
1874 |
|
✗ |
while (total_decode > 0) { |
1875 |
|
|
float last = CODEBOOK_ELEMENT_BASE(c); |
1876 |
|
✗ |
DECODE_VQ(z,f,c); |
1877 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK |
1878 |
|
|
assert(!c->sparse || z < c->sorted_entries); |
1879 |
|
|
#endif |
1880 |
|
✗ |
if (z < 0) { |
1881 |
|
✗ |
if (!f->bytes_in_seg) |
1882 |
|
✗ |
if (f->last_seg) return FALSE; |
1883 |
|
✗ |
return error(f, VORBIS_invalid_stream); |
1884 |
|
|
} |
1885 |
|
|
|
1886 |
|
|
// if this will take us off the end of the buffers, stop short! |
1887 |
|
|
// we check by computing the length of the virtual interleaved |
1888 |
|
|
// buffer (len*ch), our current offset within it (p_inter*ch)+(c_inter), |
1889 |
|
|
// and the length we'll be using (effective) |
1890 |
|
✗ |
if (c_inter + p_inter*ch + effective > len * ch) { |
1891 |
|
✗ |
effective = len*ch - (p_inter*ch - c_inter); |
1892 |
|
|
} |
1893 |
|
|
|
1894 |
|
|
#ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK |
1895 |
|
|
if (c->lookup_type == 1) { |
1896 |
|
|
int div = 1; |
1897 |
|
|
for (i=0; i < effective; ++i) { |
1898 |
|
|
int off = (z / div) % c->lookup_values; |
1899 |
|
|
float val = CODEBOOK_ELEMENT_FAST(c,off) + last; |
1900 |
|
|
if (outputs[c_inter]) |
1901 |
|
|
outputs[c_inter][p_inter] += val; |
1902 |
|
|
if (++c_inter == ch) { c_inter = 0; ++p_inter; } |
1903 |
|
|
if (c->sequence_p) last = val; |
1904 |
|
|
div *= c->lookup_values; |
1905 |
|
|
} |
1906 |
|
|
} else |
1907 |
|
|
#endif |
1908 |
|
|
{ |
1909 |
|
✗ |
z *= c->dimensions; |
1910 |
|
✗ |
if (c->sequence_p) { |
1911 |
|
✗ |
for (i=0; i < effective; ++i) { |
1912 |
|
✗ |
float val = CODEBOOK_ELEMENT_FAST(c,z+i) + last; |
1913 |
|
✗ |
if (outputs[c_inter]) |
1914 |
|
✗ |
outputs[c_inter][p_inter] += val; |
1915 |
|
✗ |
if (++c_inter == ch) { c_inter = 0; ++p_inter; } |
1916 |
|
|
last = val; |
1917 |
|
|
} |
1918 |
|
|
} else { |
1919 |
|
✗ |
for (i=0; i < effective; ++i) { |
1920 |
|
✗ |
float val = CODEBOOK_ELEMENT_FAST(c,z+i) + last; |
1921 |
|
✗ |
if (outputs[c_inter]) |
1922 |
|
✗ |
outputs[c_inter][p_inter] += val; |
1923 |
|
✗ |
if (++c_inter == ch) { c_inter = 0; ++p_inter; } |
1924 |
|
|
} |
1925 |
|
|
} |
1926 |
|
|
} |
1927 |
|
|
|
1928 |
|
✗ |
total_decode -= effective; |
1929 |
|
|
} |
1930 |
|
✗ |
*c_inter_p = c_inter; |
1931 |
|
✗ |
*p_inter_p = p_inter; |
1932 |
|
✗ |
return TRUE; |
1933 |
|
|
} |
1934 |
|
|
|
1935 |
|
|
static int predict_point(int x, int x0, int x1, int y0, int y1) |
1936 |
|
|
{ |
1937 |
|
✗ |
int dy = y1 - y0; |
1938 |
|
✗ |
int adx = x1 - x0; |
1939 |
|
|
// @OPTIMIZE: force int division to round in the right direction... is this necessary on x86? |
1940 |
|
✗ |
int err = abs(dy) * (x - x0); |
1941 |
|
✗ |
int off = err / adx; |
1942 |
|
✗ |
return dy < 0 ? y0 - off : y0 + off; |
1943 |
|
|
} |
1944 |
|
|
|
1945 |
|
|
// the following table is block-copied from the specification |
1946 |
|
|
static float inverse_db_table[256] = |
1947 |
|
|
{ |
1948 |
|
|
1.0649863e-07f, 1.1341951e-07f, 1.2079015e-07f, 1.2863978e-07f, |
1949 |
|
|
1.3699951e-07f, 1.4590251e-07f, 1.5538408e-07f, 1.6548181e-07f, |
1950 |
|
|
1.7623575e-07f, 1.8768855e-07f, 1.9988561e-07f, 2.1287530e-07f, |
1951 |
|
|
2.2670913e-07f, 2.4144197e-07f, 2.5713223e-07f, 2.7384213e-07f, |
1952 |
|
|
2.9163793e-07f, 3.1059021e-07f, 3.3077411e-07f, 3.5226968e-07f, |
1953 |
|
|
3.7516214e-07f, 3.9954229e-07f, 4.2550680e-07f, 4.5315863e-07f, |
1954 |
|
|
4.8260743e-07f, 5.1396998e-07f, 5.4737065e-07f, 5.8294187e-07f, |
1955 |
|
|
6.2082472e-07f, 6.6116941e-07f, 7.0413592e-07f, 7.4989464e-07f, |
1956 |
|
|
7.9862701e-07f, 8.5052630e-07f, 9.0579828e-07f, 9.6466216e-07f, |
1957 |
|
|
1.0273513e-06f, 1.0941144e-06f, 1.1652161e-06f, 1.2409384e-06f, |
1958 |
|
|
1.3215816e-06f, 1.4074654e-06f, 1.4989305e-06f, 1.5963394e-06f, |
1959 |
|
|
1.7000785e-06f, 1.8105592e-06f, 1.9282195e-06f, 2.0535261e-06f, |
1960 |
|
|
2.1869758e-06f, 2.3290978e-06f, 2.4804557e-06f, 2.6416497e-06f, |
1961 |
|
|
2.8133190e-06f, 2.9961443e-06f, 3.1908506e-06f, 3.3982101e-06f, |
1962 |
|
|
3.6190449e-06f, 3.8542308e-06f, 4.1047004e-06f, 4.3714470e-06f, |
1963 |
|
|
4.6555282e-06f, 4.9580707e-06f, 5.2802740e-06f, 5.6234160e-06f, |
1964 |
|
|
5.9888572e-06f, 6.3780469e-06f, 6.7925283e-06f, 7.2339451e-06f, |
1965 |
|
|
7.7040476e-06f, 8.2047000e-06f, 8.7378876e-06f, 9.3057248e-06f, |
1966 |
|
|
9.9104632e-06f, 1.0554501e-05f, 1.1240392e-05f, 1.1970856e-05f, |
1967 |
|
|
1.2748789e-05f, 1.3577278e-05f, 1.4459606e-05f, 1.5399272e-05f, |
1968 |
|
|
1.6400004e-05f, 1.7465768e-05f, 1.8600792e-05f, 1.9809576e-05f, |
1969 |
|
|
2.1096914e-05f, 2.2467911e-05f, 2.3928002e-05f, 2.5482978e-05f, |
1970 |
|
|
2.7139006e-05f, 2.8902651e-05f, 3.0780908e-05f, 3.2781225e-05f, |
1971 |
|
|
3.4911534e-05f, 3.7180282e-05f, 3.9596466e-05f, 4.2169667e-05f, |
1972 |
|
|
4.4910090e-05f, 4.7828601e-05f, 5.0936773e-05f, 5.4246931e-05f, |
1973 |
|
|
5.7772202e-05f, 6.1526565e-05f, 6.5524908e-05f, 6.9783085e-05f, |
1974 |
|
|
7.4317983e-05f, 7.9147585e-05f, 8.4291040e-05f, 8.9768747e-05f, |
1975 |
|
|
9.5602426e-05f, 0.00010181521f, 0.00010843174f, 0.00011547824f, |
1976 |
|
|
0.00012298267f, 0.00013097477f, 0.00013948625f, 0.00014855085f, |
1977 |
|
|
0.00015820453f, 0.00016848555f, 0.00017943469f, 0.00019109536f, |
1978 |
|
|
0.00020351382f, 0.00021673929f, 0.00023082423f, 0.00024582449f, |
1979 |
|
|
0.00026179955f, 0.00027881276f, 0.00029693158f, 0.00031622787f, |
1980 |
|
|
0.00033677814f, 0.00035866388f, 0.00038197188f, 0.00040679456f, |
1981 |
|
|
0.00043323036f, 0.00046138411f, 0.00049136745f, 0.00052329927f, |
1982 |
|
|
0.00055730621f, 0.00059352311f, 0.00063209358f, 0.00067317058f, |
1983 |
|
|
0.00071691700f, 0.00076350630f, 0.00081312324f, 0.00086596457f, |
1984 |
|
|
0.00092223983f, 0.00098217216f, 0.0010459992f, 0.0011139742f, |
1985 |
|
|
0.0011863665f, 0.0012634633f, 0.0013455702f, 0.0014330129f, |
1986 |
|
|
0.0015261382f, 0.0016253153f, 0.0017309374f, 0.0018434235f, |
1987 |
|
|
0.0019632195f, 0.0020908006f, 0.0022266726f, 0.0023713743f, |
1988 |
|
|
0.0025254795f, 0.0026895994f, 0.0028643847f, 0.0030505286f, |
1989 |
|
|
0.0032487691f, 0.0034598925f, 0.0036847358f, 0.0039241906f, |
1990 |
|
|
0.0041792066f, 0.0044507950f, 0.0047400328f, 0.0050480668f, |
1991 |
|
|
0.0053761186f, 0.0057254891f, 0.0060975636f, 0.0064938176f, |
1992 |
|
|
0.0069158225f, 0.0073652516f, 0.0078438871f, 0.0083536271f, |
1993 |
|
|
0.0088964928f, 0.009474637f, 0.010090352f, 0.010746080f, |
1994 |
|
|
0.011444421f, 0.012188144f, 0.012980198f, 0.013823725f, |
1995 |
|
|
0.014722068f, 0.015678791f, 0.016697687f, 0.017782797f, |
1996 |
|
|
0.018938423f, 0.020169149f, 0.021479854f, 0.022875735f, |
1997 |
|
|
0.024362330f, 0.025945531f, 0.027631618f, 0.029427276f, |
1998 |
|
|
0.031339626f, 0.033376252f, 0.035545228f, 0.037855157f, |
1999 |
|
|
0.040315199f, 0.042935108f, 0.045725273f, 0.048696758f, |
2000 |
|
|
0.051861348f, 0.055231591f, 0.058820850f, 0.062643361f, |
2001 |
|
|
0.066714279f, 0.071049749f, 0.075666962f, 0.080584227f, |
2002 |
|
|
0.085821044f, 0.091398179f, 0.097337747f, 0.10366330f, |
2003 |
|
|
0.11039993f, 0.11757434f, 0.12521498f, 0.13335215f, |
2004 |
|
|
0.14201813f, 0.15124727f, 0.16107617f, 0.17154380f, |
2005 |
|
|
0.18269168f, 0.19456402f, 0.20720788f, 0.22067342f, |
2006 |
|
|
0.23501402f, 0.25028656f, 0.26655159f, 0.28387361f, |
2007 |
|
|
0.30232132f, 0.32196786f, 0.34289114f, 0.36517414f, |
2008 |
|
|
0.38890521f, 0.41417847f, 0.44109412f, 0.46975890f, |
2009 |
|
|
0.50028648f, 0.53279791f, 0.56742212f, 0.60429640f, |
2010 |
|
|
0.64356699f, 0.68538959f, 0.72993007f, 0.77736504f, |
2011 |
|
|
0.82788260f, 0.88168307f, 0.9389798f, 1.0f |
2012 |
|
|
}; |
2013 |
|
|
|
2014 |
|
|
|
2015 |
|
|
// @OPTIMIZE: if you want to replace this bresenham line-drawing routine, |
2016 |
|
|
// note that you must produce bit-identical output to decode correctly; |
2017 |
|
|
// this specific sequence of operations is specified in the spec (it's |
2018 |
|
|
// drawing integer-quantized frequency-space lines that the encoder |
2019 |
|
|
// expects to be exactly the same) |
2020 |
|
|
// ... also, isn't the whole point of Bresenham's algorithm to NOT |
2021 |
|
|
// have to divide in the setup? sigh. |
2022 |
|
|
#ifndef STB_VORBIS_NO_DEFER_FLOOR |
2023 |
|
|
#define LINE_OP(a,b) a *= b |
2024 |
|
|
#else |
2025 |
|
|
#define LINE_OP(a,b) a = b |
2026 |
|
|
#endif |
2027 |
|
|
|
2028 |
|
|
#ifdef STB_VORBIS_DIVIDE_TABLE |
2029 |
|
|
#define DIVTAB_NUMER 32 |
2030 |
|
|
#define DIVTAB_DENOM 64 |
2031 |
|
|
int8 integer_divide_table[DIVTAB_NUMER][DIVTAB_DENOM]; // 2KB |
2032 |
|
|
#endif |
2033 |
|
|
|
2034 |
|
✗ |
static __forceinline void draw_line(float *output, int x0, int y0, int x1, int y1, int n) |
2035 |
|
|
{ |
2036 |
|
✗ |
int dy = y1 - y0; |
2037 |
|
✗ |
int adx = x1 - x0; |
2038 |
|
✗ |
int ady = abs(dy); |
2039 |
|
|
int base; |
2040 |
|
|
int x=x0,y=y0; |
2041 |
|
|
int err = 0; |
2042 |
|
|
int sy; |
2043 |
|
|
|
2044 |
|
|
#ifdef STB_VORBIS_DIVIDE_TABLE |
2045 |
|
|
if (adx < DIVTAB_DENOM && ady < DIVTAB_NUMER) { |
2046 |
|
|
if (dy < 0) { |
2047 |
|
|
base = -integer_divide_table[ady][adx]; |
2048 |
|
|
sy = base-1; |
2049 |
|
|
} else { |
2050 |
|
|
base = integer_divide_table[ady][adx]; |
2051 |
|
|
sy = base+1; |
2052 |
|
|
} |
2053 |
|
|
} else { |
2054 |
|
|
base = dy / adx; |
2055 |
|
|
if (dy < 0) |
2056 |
|
|
sy = base - 1; |
2057 |
|
|
else |
2058 |
|
|
sy = base+1; |
2059 |
|
|
} |
2060 |
|
|
#else |
2061 |
|
✗ |
base = dy / adx; |
2062 |
|
✗ |
if (dy < 0) |
2063 |
|
✗ |
sy = base - 1; |
2064 |
|
|
else |
2065 |
|
✗ |
sy = base+1; |
2066 |
|
|
#endif |
2067 |
|
✗ |
ady -= abs(base) * adx; |
2068 |
|
|
if (x1 > n) x1 = n; |
2069 |
|
✗ |
if (x < x1) { |
2070 |
|
✗ |
LINE_OP(output[x], inverse_db_table[y&255]); |
2071 |
|
✗ |
for (++x; x < x1; ++x) { |
2072 |
|
✗ |
err += ady; |
2073 |
|
✗ |
if (err >= adx) { |
2074 |
|
✗ |
err -= adx; |
2075 |
|
✗ |
y += sy; |
2076 |
|
|
} else |
2077 |
|
✗ |
y += base; |
2078 |
|
✗ |
LINE_OP(output[x], inverse_db_table[y&255]); |
2079 |
|
|
} |
2080 |
|
|
} |
2081 |
|
|
} |
2082 |
|
|
|
2083 |
|
✗ |
static int residue_decode(vorb *f, Codebook *book, float *target, int offset, int n, int rtype) |
2084 |
|
|
{ |
2085 |
|
|
int k; |
2086 |
|
✗ |
if (rtype == 0) { |
2087 |
|
✗ |
int step = n / book->dimensions; |
2088 |
|
✗ |
for (k=0; k < step; ++k) |
2089 |
|
✗ |
if (!codebook_decode_step(f, book, target+offset+k, n-offset-k, step)) |
2090 |
|
|
return FALSE; |
2091 |
|
|
} else { |
2092 |
|
✗ |
for (k=0; k < n; ) { |
2093 |
|
✗ |
if (!codebook_decode(f, book, target+offset, n-k)) |
2094 |
|
|
return FALSE; |
2095 |
|
✗ |
k += book->dimensions; |
2096 |
|
✗ |
offset += book->dimensions; |
2097 |
|
|
} |
2098 |
|
|
} |
2099 |
|
|
return TRUE; |
2100 |
|
|
} |
2101 |
|
|
|
2102 |
|
|
// n is 1/2 of the blocksize -- |
2103 |
|
|
// specification: "Correct per-vector decode length is [n]/2" |
2104 |
|
✗ |
static void decode_residue(vorb *f, float *residue_buffers[], int ch, int n, int rn, uint8 *do_not_decode) |
2105 |
|
|
{ |
2106 |
|
|
int i,j,pass; |
2107 |
|
✗ |
Residue *r = f->residue_config + rn; |
2108 |
|
✗ |
int rtype = f->residue_types[rn]; |
2109 |
|
✗ |
int c = r->classbook; |
2110 |
|
✗ |
int classwords = f->codebooks[c].dimensions; |
2111 |
|
✗ |
unsigned int actual_size = rtype == 2 ? n*2 : n; |
2112 |
|
✗ |
unsigned int limit_r_begin = (r->begin < actual_size ? r->begin : actual_size); |
2113 |
|
✗ |
unsigned int limit_r_end = (r->end < actual_size ? r->end : actual_size); |
2114 |
|
✗ |
int n_read = limit_r_end - limit_r_begin; |
2115 |
|
✗ |
int part_read = n_read / r->part_size; |
2116 |
|
✗ |
int temp_alloc_point = temp_alloc_save(f); |
2117 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2118 |
|
✗ |
uint8 ***part_classdata = (uint8 ***) temp_block_array(f,f->channels, part_read * sizeof(**part_classdata)); |
2119 |
|
|
#else |
2120 |
|
|
int **classifications = (int **) temp_block_array(f,f->channels, part_read * sizeof(**classifications)); |
2121 |
|
|
#endif |
2122 |
|
|
|
2123 |
|
|
CHECK(f); |
2124 |
|
|
|
2125 |
|
✗ |
for (i=0; i < ch; ++i) |
2126 |
|
✗ |
if (!do_not_decode[i]) |
2127 |
|
✗ |
memset(residue_buffers[i], 0, sizeof(float) * n); |
2128 |
|
|
|
2129 |
|
✗ |
if (rtype == 2 && ch != 1) { |
2130 |
|
✗ |
for (j=0; j < ch; ++j) |
2131 |
|
✗ |
if (!do_not_decode[j]) |
2132 |
|
|
break; |
2133 |
|
✗ |
if (j == ch) |
2134 |
|
✗ |
goto done; |
2135 |
|
|
|
2136 |
|
✗ |
for (pass=0; pass < 8; ++pass) { |
2137 |
|
|
int pcount = 0, class_set = 0; |
2138 |
|
✗ |
if (ch == 2) { |
2139 |
|
✗ |
while (pcount < part_read) { |
2140 |
|
✗ |
int z = r->begin + pcount*r->part_size; |
2141 |
|
✗ |
int c_inter = (z & 1), p_inter = z>>1; |
2142 |
|
✗ |
if (pass == 0) { |
2143 |
|
✗ |
Codebook *c = f->codebooks+r->classbook; |
2144 |
|
|
int q; |
2145 |
|
✗ |
DECODE(q,f,c); |
2146 |
|
✗ |
if (q == EOP) goto done; |
2147 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2148 |
|
✗ |
part_classdata[0][class_set] = r->classdata[q]; |
2149 |
|
|
#else |
2150 |
|
|
for (i=classwords-1; i >= 0; --i) { |
2151 |
|
|
classifications[0][i+pcount] = q % r->classifications; |
2152 |
|
|
q /= r->classifications; |
2153 |
|
|
} |
2154 |
|
|
#endif |
2155 |
|
|
} |
2156 |
|
✗ |
for (i=0; i < classwords && pcount < part_read; ++i, ++pcount) { |
2157 |
|
✗ |
int z = r->begin + pcount*r->part_size; |
2158 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2159 |
|
✗ |
int c = part_classdata[0][class_set][i]; |
2160 |
|
|
#else |
2161 |
|
|
int c = classifications[0][pcount]; |
2162 |
|
|
#endif |
2163 |
|
✗ |
int b = r->residue_books[c][pass]; |
2164 |
|
✗ |
if (b >= 0) { |
2165 |
|
✗ |
Codebook *book = f->codebooks + b; |
2166 |
|
|
#ifdef STB_VORBIS_DIVIDES_IN_CODEBOOK |
2167 |
|
|
if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size)) |
2168 |
|
|
goto done; |
2169 |
|
|
#else |
2170 |
|
|
// saves 1% |
2171 |
|
✗ |
if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size)) |
2172 |
|
✗ |
goto done; |
2173 |
|
|
#endif |
2174 |
|
|
} else { |
2175 |
|
✗ |
z += r->part_size; |
2176 |
|
✗ |
c_inter = z & 1; |
2177 |
|
✗ |
p_inter = z >> 1; |
2178 |
|
|
} |
2179 |
|
|
} |
2180 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2181 |
|
✗ |
++class_set; |
2182 |
|
|
#endif |
2183 |
|
|
} |
2184 |
|
✗ |
} else if (ch > 2) { |
2185 |
|
✗ |
while (pcount < part_read) { |
2186 |
|
✗ |
int z = r->begin + pcount*r->part_size; |
2187 |
|
✗ |
int c_inter = z % ch, p_inter = z/ch; |
2188 |
|
✗ |
if (pass == 0) { |
2189 |
|
✗ |
Codebook *c = f->codebooks+r->classbook; |
2190 |
|
|
int q; |
2191 |
|
✗ |
DECODE(q,f,c); |
2192 |
|
✗ |
if (q == EOP) goto done; |
2193 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2194 |
|
✗ |
part_classdata[0][class_set] = r->classdata[q]; |
2195 |
|
|
#else |
2196 |
|
|
for (i=classwords-1; i >= 0; --i) { |
2197 |
|
|
classifications[0][i+pcount] = q % r->classifications; |
2198 |
|
|
q /= r->classifications; |
2199 |
|
|
} |
2200 |
|
|
#endif |
2201 |
|
|
} |
2202 |
|
✗ |
for (i=0; i < classwords && pcount < part_read; ++i, ++pcount) { |
2203 |
|
✗ |
int z = r->begin + pcount*r->part_size; |
2204 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2205 |
|
✗ |
int c = part_classdata[0][class_set][i]; |
2206 |
|
|
#else |
2207 |
|
|
int c = classifications[0][pcount]; |
2208 |
|
|
#endif |
2209 |
|
✗ |
int b = r->residue_books[c][pass]; |
2210 |
|
✗ |
if (b >= 0) { |
2211 |
|
✗ |
Codebook *book = f->codebooks + b; |
2212 |
|
✗ |
if (!codebook_decode_deinterleave_repeat(f, book, residue_buffers, ch, &c_inter, &p_inter, n, r->part_size)) |
2213 |
|
✗ |
goto done; |
2214 |
|
|
} else { |
2215 |
|
✗ |
z += r->part_size; |
2216 |
|
✗ |
c_inter = z % ch; |
2217 |
|
✗ |
p_inter = z / ch; |
2218 |
|
|
} |
2219 |
|
|
} |
2220 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2221 |
|
✗ |
++class_set; |
2222 |
|
|
#endif |
2223 |
|
|
} |
2224 |
|
|
} |
2225 |
|
|
} |
2226 |
|
✗ |
goto done; |
2227 |
|
|
} |
2228 |
|
|
CHECK(f); |
2229 |
|
|
|
2230 |
|
✗ |
for (pass=0; pass < 8; ++pass) { |
2231 |
|
|
int pcount = 0, class_set=0; |
2232 |
|
✗ |
while (pcount < part_read) { |
2233 |
|
✗ |
if (pass == 0) { |
2234 |
|
✗ |
for (j=0; j < ch; ++j) { |
2235 |
|
✗ |
if (!do_not_decode[j]) { |
2236 |
|
✗ |
Codebook *c = f->codebooks+r->classbook; |
2237 |
|
|
int temp; |
2238 |
|
✗ |
DECODE(temp,f,c); |
2239 |
|
✗ |
if (temp == EOP) goto done; |
2240 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2241 |
|
✗ |
part_classdata[j][class_set] = r->classdata[temp]; |
2242 |
|
|
#else |
2243 |
|
|
for (i=classwords-1; i >= 0; --i) { |
2244 |
|
|
classifications[j][i+pcount] = temp % r->classifications; |
2245 |
|
|
temp /= r->classifications; |
2246 |
|
|
} |
2247 |
|
|
#endif |
2248 |
|
|
} |
2249 |
|
|
} |
2250 |
|
|
} |
2251 |
|
✗ |
for (i=0; i < classwords && pcount < part_read; ++i, ++pcount) { |
2252 |
|
✗ |
for (j=0; j < ch; ++j) { |
2253 |
|
✗ |
if (!do_not_decode[j]) { |
2254 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2255 |
|
✗ |
int c = part_classdata[j][class_set][i]; |
2256 |
|
|
#else |
2257 |
|
|
int c = classifications[j][pcount]; |
2258 |
|
|
#endif |
2259 |
|
✗ |
int b = r->residue_books[c][pass]; |
2260 |
|
✗ |
if (b >= 0) { |
2261 |
|
✗ |
float *target = residue_buffers[j]; |
2262 |
|
✗ |
int offset = r->begin + pcount * r->part_size; |
2263 |
|
✗ |
int n = r->part_size; |
2264 |
|
✗ |
Codebook *book = f->codebooks + b; |
2265 |
|
✗ |
if (!residue_decode(f, book, target, offset, n, rtype)) |
2266 |
|
✗ |
goto done; |
2267 |
|
|
} |
2268 |
|
|
} |
2269 |
|
|
} |
2270 |
|
|
} |
2271 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2272 |
|
✗ |
++class_set; |
2273 |
|
|
#endif |
2274 |
|
|
} |
2275 |
|
|
} |
2276 |
|
✗ |
done: |
2277 |
|
|
CHECK(f); |
2278 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
2279 |
|
|
temp_free(f,part_classdata); |
2280 |
|
|
#else |
2281 |
|
|
temp_free(f,classifications); |
2282 |
|
|
#endif |
2283 |
|
✗ |
temp_alloc_restore(f,temp_alloc_point); |
2284 |
|
|
} |
2285 |
|
|
|
2286 |
|
|
|
2287 |
|
|
#if 0 |
2288 |
|
|
// slow way for debugging |
2289 |
|
|
void inverse_mdct_slow(float *buffer, int n) |
2290 |
|
|
{ |
2291 |
|
|
int i,j; |
2292 |
|
|
int n2 = n >> 1; |
2293 |
|
|
float *x = (float *) malloc(sizeof(*x) * n2); |
2294 |
|
|
memcpy(x, buffer, sizeof(*x) * n2); |
2295 |
|
|
for (i=0; i < n; ++i) { |
2296 |
|
|
float acc = 0; |
2297 |
|
|
for (j=0; j < n2; ++j) |
2298 |
|
|
// formula from paper: |
2299 |
|
|
//acc += n/4.0f * x[j] * (float) cos(M_PI / 2 / n * (2 * i + 1 + n/2.0)*(2*j+1)); |
2300 |
|
|
// formula from wikipedia |
2301 |
|
|
//acc += 2.0f / n2 * x[j] * (float) cos(M_PI/n2 * (i + 0.5 + n2/2)*(j + 0.5)); |
2302 |
|
|
// these are equivalent, except the formula from the paper inverts the multiplier! |
2303 |
|
|
// however, what actually works is NO MULTIPLIER!?! |
2304 |
|
|
//acc += 64 * 2.0f / n2 * x[j] * (float) cos(M_PI/n2 * (i + 0.5 + n2/2)*(j + 0.5)); |
2305 |
|
|
acc += x[j] * (float) cos(M_PI / 2 / n * (2 * i + 1 + n/2.0)*(2*j+1)); |
2306 |
|
|
buffer[i] = acc; |
2307 |
|
|
} |
2308 |
|
|
free(x); |
2309 |
|
|
} |
2310 |
|
|
#elif 0 |
2311 |
|
|
// same as above, but just barely able to run in real time on modern machines |
2312 |
|
|
void inverse_mdct_slow(float *buffer, int n, vorb *f, int blocktype) |
2313 |
|
|
{ |
2314 |
|
|
float mcos[16384]; |
2315 |
|
|
int i,j; |
2316 |
|
|
int n2 = n >> 1, nmask = (n << 2) -1; |
2317 |
|
|
float *x = (float *) malloc(sizeof(*x) * n2); |
2318 |
|
|
memcpy(x, buffer, sizeof(*x) * n2); |
2319 |
|
|
for (i=0; i < 4*n; ++i) |
2320 |
|
|
mcos[i] = (float) cos(M_PI / 2 * i / n); |
2321 |
|
|
|
2322 |
|
|
for (i=0; i < n; ++i) { |
2323 |
|
|
float acc = 0; |
2324 |
|
|
for (j=0; j < n2; ++j) |
2325 |
|
|
acc += x[j] * mcos[(2 * i + 1 + n2)*(2*j+1) & nmask]; |
2326 |
|
|
buffer[i] = acc; |
2327 |
|
|
} |
2328 |
|
|
free(x); |
2329 |
|
|
} |
2330 |
|
|
#elif 0 |
2331 |
|
|
// transform to use a slow dct-iv; this is STILL basically trivial, |
2332 |
|
|
// but only requires half as many ops |
2333 |
|
|
void dct_iv_slow(float *buffer, int n) |
2334 |
|
|
{ |
2335 |
|
|
float mcos[16384]; |
2336 |
|
|
float x[2048]; |
2337 |
|
|
int i,j; |
2338 |
|
|
int n2 = n >> 1, nmask = (n << 3) - 1; |
2339 |
|
|
memcpy(x, buffer, sizeof(*x) * n); |
2340 |
|
|
for (i=0; i < 8*n; ++i) |
2341 |
|
|
mcos[i] = (float) cos(M_PI / 4 * i / n); |
2342 |
|
|
for (i=0; i < n; ++i) { |
2343 |
|
|
float acc = 0; |
2344 |
|
|
for (j=0; j < n; ++j) |
2345 |
|
|
acc += x[j] * mcos[((2 * i + 1)*(2*j+1)) & nmask]; |
2346 |
|
|
buffer[i] = acc; |
2347 |
|
|
} |
2348 |
|
|
} |
2349 |
|
|
|
2350 |
|
|
void inverse_mdct_slow(float *buffer, int n, vorb *f, int blocktype) |
2351 |
|
|
{ |
2352 |
|
|
int i, n4 = n >> 2, n2 = n >> 1, n3_4 = n - n4; |
2353 |
|
|
float temp[4096]; |
2354 |
|
|
|
2355 |
|
|
memcpy(temp, buffer, n2 * sizeof(float)); |
2356 |
|
|
dct_iv_slow(temp, n2); // returns -c'-d, a-b' |
2357 |
|
|
|
2358 |
|
|
for (i=0; i < n4 ; ++i) buffer[i] = temp[i+n4]; // a-b' |
2359 |
|
|
for ( ; i < n3_4; ++i) buffer[i] = -temp[n3_4 - i - 1]; // b-a', c+d' |
2360 |
|
|
for ( ; i < n ; ++i) buffer[i] = -temp[i - n3_4]; // c'+d |
2361 |
|
|
} |
2362 |
|
|
#endif |
2363 |
|
|
|
2364 |
|
|
#ifndef LIBVORBIS_MDCT |
2365 |
|
|
#define LIBVORBIS_MDCT 0 |
2366 |
|
|
#endif |
2367 |
|
|
|
2368 |
|
|
#if LIBVORBIS_MDCT |
2369 |
|
|
// directly call the vorbis MDCT using an interface documented |
2370 |
|
|
// by Jeff Roberts... useful for performance comparison |
2371 |
|
|
typedef struct |
2372 |
|
|
{ |
2373 |
|
|
int n; |
2374 |
|
|
int log2n; |
2375 |
|
|
|
2376 |
|
|
float *trig; |
2377 |
|
|
int *bitrev; |
2378 |
|
|
|
2379 |
|
|
float scale; |
2380 |
|
|
} mdct_lookup; |
2381 |
|
|
|
2382 |
|
|
extern void mdct_init(mdct_lookup *lookup, int n); |
2383 |
|
|
extern void mdct_clear(mdct_lookup *l); |
2384 |
|
|
extern void mdct_backward(mdct_lookup *init, float *in, float *out); |
2385 |
|
|
|
2386 |
|
|
mdct_lookup M1,M2; |
2387 |
|
|
|
2388 |
|
|
void inverse_mdct(float *buffer, int n, vorb *f, int blocktype) |
2389 |
|
|
{ |
2390 |
|
|
mdct_lookup *M; |
2391 |
|
|
if (M1.n == n) M = &M1; |
2392 |
|
|
else if (M2.n == n) M = &M2; |
2393 |
|
|
else if (M1.n == 0) { mdct_init(&M1, n); M = &M1; } |
2394 |
|
|
else { |
2395 |
|
|
if (M2.n) __asm int 3; |
2396 |
|
|
mdct_init(&M2, n); |
2397 |
|
|
M = &M2; |
2398 |
|
|
} |
2399 |
|
|
|
2400 |
|
|
mdct_backward(M, buffer, buffer); |
2401 |
|
|
} |
2402 |
|
|
#endif |
2403 |
|
|
|
2404 |
|
|
|
2405 |
|
|
// the following were split out into separate functions while optimizing; |
2406 |
|
|
// they could be pushed back up but eh. __forceinline showed no change; |
2407 |
|
|
// they're probably already being inlined. |
2408 |
|
✗ |
static void imdct_step3_iter0_loop(int n, float *e, int i_off, int k_off, float *A) |
2409 |
|
|
{ |
2410 |
|
✗ |
float *ee0 = e + i_off; |
2411 |
|
✗ |
float *ee2 = ee0 + k_off; |
2412 |
|
|
int i; |
2413 |
|
|
|
2414 |
|
|
assert((n & 3) == 0); |
2415 |
|
✗ |
for (i=(n>>2); i > 0; --i) { |
2416 |
|
|
float k00_20, k01_21; |
2417 |
|
✗ |
k00_20 = ee0[ 0] - ee2[ 0]; |
2418 |
|
✗ |
k01_21 = ee0[-1] - ee2[-1]; |
2419 |
|
✗ |
ee0[ 0] += ee2[ 0];//ee0[ 0] = ee0[ 0] + ee2[ 0]; |
2420 |
|
✗ |
ee0[-1] += ee2[-1];//ee0[-1] = ee0[-1] + ee2[-1]; |
2421 |
|
✗ |
ee2[ 0] = k00_20 * A[0] - k01_21 * A[1]; |
2422 |
|
✗ |
ee2[-1] = k01_21 * A[0] + k00_20 * A[1]; |
2423 |
|
|
A += 8; |
2424 |
|
|
|
2425 |
|
✗ |
k00_20 = ee0[-2] - ee2[-2]; |
2426 |
|
✗ |
k01_21 = ee0[-3] - ee2[-3]; |
2427 |
|
✗ |
ee0[-2] += ee2[-2];//ee0[-2] = ee0[-2] + ee2[-2]; |
2428 |
|
✗ |
ee0[-3] += ee2[-3];//ee0[-3] = ee0[-3] + ee2[-3]; |
2429 |
|
✗ |
ee2[-2] = k00_20 * A[0] - k01_21 * A[1]; |
2430 |
|
✗ |
ee2[-3] = k01_21 * A[0] + k00_20 * A[1]; |
2431 |
|
|
A += 8; |
2432 |
|
|
|
2433 |
|
✗ |
k00_20 = ee0[-4] - ee2[-4]; |
2434 |
|
✗ |
k01_21 = ee0[-5] - ee2[-5]; |
2435 |
|
✗ |
ee0[-4] += ee2[-4];//ee0[-4] = ee0[-4] + ee2[-4]; |
2436 |
|
✗ |
ee0[-5] += ee2[-5];//ee0[-5] = ee0[-5] + ee2[-5]; |
2437 |
|
✗ |
ee2[-4] = k00_20 * A[0] - k01_21 * A[1]; |
2438 |
|
✗ |
ee2[-5] = k01_21 * A[0] + k00_20 * A[1]; |
2439 |
|
|
A += 8; |
2440 |
|
|
|
2441 |
|
✗ |
k00_20 = ee0[-6] - ee2[-6]; |
2442 |
|
✗ |
k01_21 = ee0[-7] - ee2[-7]; |
2443 |
|
✗ |
ee0[-6] += ee2[-6];//ee0[-6] = ee0[-6] + ee2[-6]; |
2444 |
|
✗ |
ee0[-7] += ee2[-7];//ee0[-7] = ee0[-7] + ee2[-7]; |
2445 |
|
✗ |
ee2[-6] = k00_20 * A[0] - k01_21 * A[1]; |
2446 |
|
✗ |
ee2[-7] = k01_21 * A[0] + k00_20 * A[1]; |
2447 |
|
✗ |
A += 8; |
2448 |
|
✗ |
ee0 -= 8; |
2449 |
|
✗ |
ee2 -= 8; |
2450 |
|
|
} |
2451 |
|
|
} |
2452 |
|
|
|
2453 |
|
✗ |
static void imdct_step3_inner_r_loop(int lim, float *e, int d0, int k_off, float *A, int k1) |
2454 |
|
|
{ |
2455 |
|
|
int i; |
2456 |
|
|
float k00_20, k01_21; |
2457 |
|
|
|
2458 |
|
✗ |
float *e0 = e + d0; |
2459 |
|
✗ |
float *e2 = e0 + k_off; |
2460 |
|
|
|
2461 |
|
✗ |
for (i=lim >> 2; i > 0; --i) { |
2462 |
|
✗ |
k00_20 = e0[-0] - e2[-0]; |
2463 |
|
✗ |
k01_21 = e0[-1] - e2[-1]; |
2464 |
|
✗ |
e0[-0] += e2[-0];//e0[-0] = e0[-0] + e2[-0]; |
2465 |
|
✗ |
e0[-1] += e2[-1];//e0[-1] = e0[-1] + e2[-1]; |
2466 |
|
✗ |
e2[-0] = (k00_20)*A[0] - (k01_21) * A[1]; |
2467 |
|
✗ |
e2[-1] = (k01_21)*A[0] + (k00_20) * A[1]; |
2468 |
|
|
|
2469 |
|
✗ |
A += k1; |
2470 |
|
|
|
2471 |
|
✗ |
k00_20 = e0[-2] - e2[-2]; |
2472 |
|
✗ |
k01_21 = e0[-3] - e2[-3]; |
2473 |
|
✗ |
e0[-2] += e2[-2];//e0[-2] = e0[-2] + e2[-2]; |
2474 |
|
✗ |
e0[-3] += e2[-3];//e0[-3] = e0[-3] + e2[-3]; |
2475 |
|
✗ |
e2[-2] = (k00_20)*A[0] - (k01_21) * A[1]; |
2476 |
|
✗ |
e2[-3] = (k01_21)*A[0] + (k00_20) * A[1]; |
2477 |
|
|
|
2478 |
|
✗ |
A += k1; |
2479 |
|
|
|
2480 |
|
✗ |
k00_20 = e0[-4] - e2[-4]; |
2481 |
|
✗ |
k01_21 = e0[-5] - e2[-5]; |
2482 |
|
✗ |
e0[-4] += e2[-4];//e0[-4] = e0[-4] + e2[-4]; |
2483 |
|
✗ |
e0[-5] += e2[-5];//e0[-5] = e0[-5] + e2[-5]; |
2484 |
|
✗ |
e2[-4] = (k00_20)*A[0] - (k01_21) * A[1]; |
2485 |
|
✗ |
e2[-5] = (k01_21)*A[0] + (k00_20) * A[1]; |
2486 |
|
|
|
2487 |
|
✗ |
A += k1; |
2488 |
|
|
|
2489 |
|
✗ |
k00_20 = e0[-6] - e2[-6]; |
2490 |
|
✗ |
k01_21 = e0[-7] - e2[-7]; |
2491 |
|
✗ |
e0[-6] += e2[-6];//e0[-6] = e0[-6] + e2[-6]; |
2492 |
|
✗ |
e0[-7] += e2[-7];//e0[-7] = e0[-7] + e2[-7]; |
2493 |
|
✗ |
e2[-6] = (k00_20)*A[0] - (k01_21) * A[1]; |
2494 |
|
✗ |
e2[-7] = (k01_21)*A[0] + (k00_20) * A[1]; |
2495 |
|
|
|
2496 |
|
✗ |
e0 -= 8; |
2497 |
|
✗ |
e2 -= 8; |
2498 |
|
|
|
2499 |
|
✗ |
A += k1; |
2500 |
|
|
} |
2501 |
|
|
} |
2502 |
|
|
|
2503 |
|
✗ |
static void imdct_step3_inner_s_loop(int n, float *e, int i_off, int k_off, float *A, int a_off, int k0) |
2504 |
|
|
{ |
2505 |
|
|
int i; |
2506 |
|
✗ |
float A0 = A[0]; |
2507 |
|
✗ |
float A1 = A[0+1]; |
2508 |
|
✗ |
float A2 = A[0+a_off]; |
2509 |
|
✗ |
float A3 = A[0+a_off+1]; |
2510 |
|
✗ |
float A4 = A[0+a_off*2+0]; |
2511 |
|
✗ |
float A5 = A[0+a_off*2+1]; |
2512 |
|
✗ |
float A6 = A[0+a_off*3+0]; |
2513 |
|
✗ |
float A7 = A[0+a_off*3+1]; |
2514 |
|
|
|
2515 |
|
|
float k00,k11; |
2516 |
|
|
|
2517 |
|
✗ |
float *ee0 = e +i_off; |
2518 |
|
✗ |
float *ee2 = ee0+k_off; |
2519 |
|
|
|
2520 |
|
✗ |
for (i=n; i > 0; --i) { |
2521 |
|
✗ |
k00 = ee0[ 0] - ee2[ 0]; |
2522 |
|
✗ |
k11 = ee0[-1] - ee2[-1]; |
2523 |
|
✗ |
ee0[ 0] = ee0[ 0] + ee2[ 0]; |
2524 |
|
✗ |
ee0[-1] = ee0[-1] + ee2[-1]; |
2525 |
|
✗ |
ee2[ 0] = (k00) * A0 - (k11) * A1; |
2526 |
|
✗ |
ee2[-1] = (k11) * A0 + (k00) * A1; |
2527 |
|
|
|
2528 |
|
✗ |
k00 = ee0[-2] - ee2[-2]; |
2529 |
|
✗ |
k11 = ee0[-3] - ee2[-3]; |
2530 |
|
✗ |
ee0[-2] = ee0[-2] + ee2[-2]; |
2531 |
|
✗ |
ee0[-3] = ee0[-3] + ee2[-3]; |
2532 |
|
✗ |
ee2[-2] = (k00) * A2 - (k11) * A3; |
2533 |
|
✗ |
ee2[-3] = (k11) * A2 + (k00) * A3; |
2534 |
|
|
|
2535 |
|
✗ |
k00 = ee0[-4] - ee2[-4]; |
2536 |
|
✗ |
k11 = ee0[-5] - ee2[-5]; |
2537 |
|
✗ |
ee0[-4] = ee0[-4] + ee2[-4]; |
2538 |
|
✗ |
ee0[-5] = ee0[-5] + ee2[-5]; |
2539 |
|
✗ |
ee2[-4] = (k00) * A4 - (k11) * A5; |
2540 |
|
✗ |
ee2[-5] = (k11) * A4 + (k00) * A5; |
2541 |
|
|
|
2542 |
|
✗ |
k00 = ee0[-6] - ee2[-6]; |
2543 |
|
✗ |
k11 = ee0[-7] - ee2[-7]; |
2544 |
|
✗ |
ee0[-6] = ee0[-6] + ee2[-6]; |
2545 |
|
✗ |
ee0[-7] = ee0[-7] + ee2[-7]; |
2546 |
|
✗ |
ee2[-6] = (k00) * A6 - (k11) * A7; |
2547 |
|
✗ |
ee2[-7] = (k11) * A6 + (k00) * A7; |
2548 |
|
|
|
2549 |
|
✗ |
ee0 -= k0; |
2550 |
|
✗ |
ee2 -= k0; |
2551 |
|
|
} |
2552 |
|
|
} |
2553 |
|
|
|
2554 |
|
✗ |
static __forceinline void iter_54(float *z) |
2555 |
|
|
{ |
2556 |
|
|
float k00,k11,k22,k33; |
2557 |
|
|
float y0,y1,y2,y3; |
2558 |
|
|
|
2559 |
|
✗ |
k00 = z[ 0] - z[-4]; |
2560 |
|
✗ |
y0 = z[ 0] + z[-4]; |
2561 |
|
✗ |
y2 = z[-2] + z[-6]; |
2562 |
|
✗ |
k22 = z[-2] - z[-6]; |
2563 |
|
|
|
2564 |
|
✗ |
z[-0] = y0 + y2; // z0 + z4 + z2 + z6 |
2565 |
|
✗ |
z[-2] = y0 - y2; // z0 + z4 - z2 - z6 |
2566 |
|
|
|
2567 |
|
|
// done with y0,y2 |
2568 |
|
|
|
2569 |
|
✗ |
k33 = z[-3] - z[-7]; |
2570 |
|
|
|
2571 |
|
✗ |
z[-4] = k00 + k33; // z0 - z4 + z3 - z7 |
2572 |
|
✗ |
z[-6] = k00 - k33; // z0 - z4 - z3 + z7 |
2573 |
|
|
|
2574 |
|
|
// done with k33 |
2575 |
|
|
|
2576 |
|
✗ |
k11 = z[-1] - z[-5]; |
2577 |
|
✗ |
y1 = z[-1] + z[-5]; |
2578 |
|
✗ |
y3 = z[-3] + z[-7]; |
2579 |
|
|
|
2580 |
|
✗ |
z[-1] = y1 + y3; // z1 + z5 + z3 + z7 |
2581 |
|
✗ |
z[-3] = y1 - y3; // z1 + z5 - z3 - z7 |
2582 |
|
✗ |
z[-5] = k11 - k22; // z1 - z5 + z2 - z6 |
2583 |
|
✗ |
z[-7] = k11 + k22; // z1 - z5 - z2 + z6 |
2584 |
|
|
} |
2585 |
|
|
|
2586 |
|
✗ |
static void imdct_step3_inner_s_loop_ld654(int n, float *e, int i_off, float *A, int base_n) |
2587 |
|
|
{ |
2588 |
|
✗ |
int a_off = base_n >> 3; |
2589 |
|
✗ |
float A2 = A[0+a_off]; |
2590 |
|
✗ |
float *z = e + i_off; |
2591 |
|
✗ |
float *base = z - 16 * n; |
2592 |
|
|
|
2593 |
|
✗ |
while (z > base) { |
2594 |
|
|
float k00,k11; |
2595 |
|
|
float l00,l11; |
2596 |
|
|
|
2597 |
|
✗ |
k00 = z[-0] - z[ -8]; |
2598 |
|
✗ |
k11 = z[-1] - z[ -9]; |
2599 |
|
✗ |
l00 = z[-2] - z[-10]; |
2600 |
|
✗ |
l11 = z[-3] - z[-11]; |
2601 |
|
✗ |
z[ -0] = z[-0] + z[ -8]; |
2602 |
|
✗ |
z[ -1] = z[-1] + z[ -9]; |
2603 |
|
✗ |
z[ -2] = z[-2] + z[-10]; |
2604 |
|
✗ |
z[ -3] = z[-3] + z[-11]; |
2605 |
|
✗ |
z[ -8] = k00; |
2606 |
|
✗ |
z[ -9] = k11; |
2607 |
|
✗ |
z[-10] = (l00+l11) * A2; |
2608 |
|
✗ |
z[-11] = (l11-l00) * A2; |
2609 |
|
|
|
2610 |
|
✗ |
k00 = z[ -4] - z[-12]; |
2611 |
|
✗ |
k11 = z[ -5] - z[-13]; |
2612 |
|
✗ |
l00 = z[ -6] - z[-14]; |
2613 |
|
✗ |
l11 = z[ -7] - z[-15]; |
2614 |
|
✗ |
z[ -4] = z[ -4] + z[-12]; |
2615 |
|
✗ |
z[ -5] = z[ -5] + z[-13]; |
2616 |
|
✗ |
z[ -6] = z[ -6] + z[-14]; |
2617 |
|
✗ |
z[ -7] = z[ -7] + z[-15]; |
2618 |
|
✗ |
z[-12] = k11; |
2619 |
|
✗ |
z[-13] = -k00; |
2620 |
|
✗ |
z[-14] = (l11-l00) * A2; |
2621 |
|
✗ |
z[-15] = (l00+l11) * -A2; |
2622 |
|
|
|
2623 |
|
✗ |
iter_54(z); |
2624 |
|
✗ |
iter_54(z-8); |
2625 |
|
✗ |
z -= 16; |
2626 |
|
|
} |
2627 |
|
|
} |
2628 |
|
|
|
2629 |
|
✗ |
static void inverse_mdct(float *buffer, int n, vorb *f, int blocktype) |
2630 |
|
|
{ |
2631 |
|
✗ |
int n2 = n >> 1, n4 = n >> 2, n8 = n >> 3, l; |
2632 |
|
|
int ld; |
2633 |
|
|
// @OPTIMIZE: reduce register pressure by using fewer variables? |
2634 |
|
✗ |
int save_point = temp_alloc_save(f); |
2635 |
|
✗ |
float *buf2 = (float *) temp_alloc(f, n2 * sizeof(*buf2)); |
2636 |
|
|
float *u=NULL,*v=NULL; |
2637 |
|
|
// twiddle factors |
2638 |
|
✗ |
float *A = f->A[blocktype]; |
2639 |
|
|
|
2640 |
|
|
// IMDCT algorithm from "The use of multirate filter banks for coding of high quality digital audio" |
2641 |
|
|
// See notes about bugs in that paper in less-optimal implementation 'inverse_mdct_old' after this function. |
2642 |
|
|
|
2643 |
|
|
// kernel from paper |
2644 |
|
|
|
2645 |
|
|
|
2646 |
|
|
// merged: |
2647 |
|
|
// copy and reflect spectral data |
2648 |
|
|
// step 0 |
2649 |
|
|
|
2650 |
|
|
// note that it turns out that the items added together during |
2651 |
|
|
// this step are, in fact, being added to themselves (as reflected |
2652 |
|
|
// by step 0). inexplicable inefficiency! this became obvious |
2653 |
|
|
// once I combined the passes. |
2654 |
|
|
|
2655 |
|
|
// so there's a missing 'times 2' here (for adding X to itself). |
2656 |
|
|
// this propagates through linearly to the end, where the numbers |
2657 |
|
|
// are 1/2 too small, and need to be compensated for. |
2658 |
|
|
|
2659 |
|
|
{ |
2660 |
|
|
float *d,*e, *AA, *e_stop; |
2661 |
|
✗ |
d = &buf2[n2-2]; |
2662 |
|
|
AA = A; |
2663 |
|
|
e = &buffer[0]; |
2664 |
|
✗ |
e_stop = &buffer[n2]; |
2665 |
|
✗ |
while (e != e_stop) { |
2666 |
|
✗ |
d[1] = (e[0] * AA[0] - e[2]*AA[1]); |
2667 |
|
✗ |
d[0] = (e[0] * AA[1] + e[2]*AA[0]); |
2668 |
|
✗ |
d -= 2; |
2669 |
|
✗ |
AA += 2; |
2670 |
|
✗ |
e += 4; |
2671 |
|
|
} |
2672 |
|
|
|
2673 |
|
✗ |
e = &buffer[n2-3]; |
2674 |
|
✗ |
while (d >= buf2) { |
2675 |
|
✗ |
d[1] = (-e[2] * AA[0] - -e[0]*AA[1]); |
2676 |
|
✗ |
d[0] = (-e[2] * AA[1] + -e[0]*AA[0]); |
2677 |
|
✗ |
d -= 2; |
2678 |
|
✗ |
AA += 2; |
2679 |
|
✗ |
e -= 4; |
2680 |
|
|
} |
2681 |
|
|
} |
2682 |
|
|
|
2683 |
|
|
// now we use symbolic names for these, so that we can |
2684 |
|
|
// possibly swap their meaning as we change which operations |
2685 |
|
|
// are in place |
2686 |
|
|
|
2687 |
|
|
u = buffer; |
2688 |
|
|
v = buf2; |
2689 |
|
|
|
2690 |
|
|
// step 2 (paper output is w, now u) |
2691 |
|
|
// this could be in place, but the data ends up in the wrong |
2692 |
|
|
// place... _somebody_'s got to swap it, so this is nominated |
2693 |
|
|
{ |
2694 |
|
✗ |
float *AA = &A[n2-8]; |
2695 |
|
|
float *d0,*d1, *e0, *e1; |
2696 |
|
|
|
2697 |
|
✗ |
e0 = &v[n4]; |
2698 |
|
|
e1 = &v[0]; |
2699 |
|
|
|
2700 |
|
✗ |
d0 = &u[n4]; |
2701 |
|
|
d1 = &u[0]; |
2702 |
|
|
|
2703 |
|
✗ |
while (AA >= A) { |
2704 |
|
|
float v40_20, v41_21; |
2705 |
|
|
|
2706 |
|
✗ |
v41_21 = e0[1] - e1[1]; |
2707 |
|
✗ |
v40_20 = e0[0] - e1[0]; |
2708 |
|
✗ |
d0[1] = e0[1] + e1[1]; |
2709 |
|
✗ |
d0[0] = e0[0] + e1[0]; |
2710 |
|
✗ |
d1[1] = v41_21*AA[4] - v40_20*AA[5]; |
2711 |
|
✗ |
d1[0] = v40_20*AA[4] + v41_21*AA[5]; |
2712 |
|
|
|
2713 |
|
✗ |
v41_21 = e0[3] - e1[3]; |
2714 |
|
✗ |
v40_20 = e0[2] - e1[2]; |
2715 |
|
✗ |
d0[3] = e0[3] + e1[3]; |
2716 |
|
✗ |
d0[2] = e0[2] + e1[2]; |
2717 |
|
✗ |
d1[3] = v41_21*AA[0] - v40_20*AA[1]; |
2718 |
|
✗ |
d1[2] = v40_20*AA[0] + v41_21*AA[1]; |
2719 |
|
|
|
2720 |
|
✗ |
AA -= 8; |
2721 |
|
|
|
2722 |
|
✗ |
d0 += 4; |
2723 |
|
✗ |
d1 += 4; |
2724 |
|
✗ |
e0 += 4; |
2725 |
|
✗ |
e1 += 4; |
2726 |
|
|
} |
2727 |
|
|
} |
2728 |
|
|
|
2729 |
|
|
// step 3 |
2730 |
|
✗ |
ld = ilog(n) - 1; // ilog is off-by-one from normal definitions |
2731 |
|
|
|
2732 |
|
|
// optimized step 3: |
2733 |
|
|
|
2734 |
|
|
// the original step3 loop can be nested r inside s or s inside r; |
2735 |
|
|
// it's written originally as s inside r, but this is dumb when r |
2736 |
|
|
// iterates many times, and s few. So I have two copies of it and |
2737 |
|
|
// switch between them halfway. |
2738 |
|
|
|
2739 |
|
|
// this is iteration 0 of step 3 |
2740 |
|
✗ |
imdct_step3_iter0_loop(n >> 4, u, n2-1-n4*0, -(n >> 3), A); |
2741 |
|
✗ |
imdct_step3_iter0_loop(n >> 4, u, n2-1-n4*1, -(n >> 3), A); |
2742 |
|
|
|
2743 |
|
|
// this is iteration 1 of step 3 |
2744 |
|
✗ |
imdct_step3_inner_r_loop(n >> 5, u, n2-1 - n8*0, -(n >> 4), A, 16); |
2745 |
|
✗ |
imdct_step3_inner_r_loop(n >> 5, u, n2-1 - n8*1, -(n >> 4), A, 16); |
2746 |
|
✗ |
imdct_step3_inner_r_loop(n >> 5, u, n2-1 - n8*2, -(n >> 4), A, 16); |
2747 |
|
✗ |
imdct_step3_inner_r_loop(n >> 5, u, n2-1 - n8*3, -(n >> 4), A, 16); |
2748 |
|
|
|
2749 |
|
|
l=2; |
2750 |
|
✗ |
for (; l < (ld-3)>>1; ++l) { |
2751 |
|
✗ |
int k0 = n >> (l+2), k0_2 = k0>>1; |
2752 |
|
✗ |
int lim = 1 << (l+1); |
2753 |
|
|
int i; |
2754 |
|
✗ |
for (i=0; i < lim; ++i) |
2755 |
|
✗ |
imdct_step3_inner_r_loop(n >> (l+4), u, n2-1 - k0*i, -k0_2, A, 1 << (l+3)); |
2756 |
|
|
} |
2757 |
|
|
|
2758 |
|
✗ |
for (; l < ld-6; ++l) { |
2759 |
|
✗ |
int k0 = n >> (l+2), k1 = 1 << (l+3), k0_2 = k0>>1; |
2760 |
|
✗ |
int rlim = n >> (l+6), r; |
2761 |
|
✗ |
int lim = 1 << (l+1); |
2762 |
|
|
int i_off; |
2763 |
|
|
float *A0 = A; |
2764 |
|
|
i_off = n2-1; |
2765 |
|
✗ |
for (r=rlim; r > 0; --r) { |
2766 |
|
✗ |
imdct_step3_inner_s_loop(lim, u, i_off, -k0_2, A0, k1, k0); |
2767 |
|
✗ |
A0 += k1*4; |
2768 |
|
✗ |
i_off -= 8; |
2769 |
|
|
} |
2770 |
|
|
} |
2771 |
|
|
|
2772 |
|
|
// iterations with count: |
2773 |
|
|
// ld-6,-5,-4 all interleaved together |
2774 |
|
|
// the big win comes from getting rid of needless flops |
2775 |
|
|
// due to the constants on pass 5 & 4 being all 1 and 0; |
2776 |
|
|
// combining them to be simultaneous to improve cache made little difference |
2777 |
|
✗ |
imdct_step3_inner_s_loop_ld654(n >> 5, u, n2-1, A, n); |
2778 |
|
|
|
2779 |
|
|
// output is u |
2780 |
|
|
|
2781 |
|
|
// step 4, 5, and 6 |
2782 |
|
|
// cannot be in-place because of step 5 |
2783 |
|
|
{ |
2784 |
|
✗ |
uint16 *bitrev = f->bit_reverse[blocktype]; |
2785 |
|
|
// weirdly, I'd have thought reading sequentially and writing |
2786 |
|
|
// erratically would have been better than vice-versa, but in |
2787 |
|
|
// fact that's not what my testing showed. (That is, with |
2788 |
|
|
// j = bitreverse(i), do you read i and write j, or read j and write i.) |
2789 |
|
|
|
2790 |
|
✗ |
float *d0 = &v[n4-4]; |
2791 |
|
✗ |
float *d1 = &v[n2-4]; |
2792 |
|
✗ |
while (d0 >= v) { |
2793 |
|
|
int k4; |
2794 |
|
|
|
2795 |
|
✗ |
k4 = bitrev[0]; |
2796 |
|
✗ |
d1[3] = u[k4+0]; |
2797 |
|
✗ |
d1[2] = u[k4+1]; |
2798 |
|
✗ |
d0[3] = u[k4+2]; |
2799 |
|
✗ |
d0[2] = u[k4+3]; |
2800 |
|
|
|
2801 |
|
✗ |
k4 = bitrev[1]; |
2802 |
|
✗ |
d1[1] = u[k4+0]; |
2803 |
|
✗ |
d1[0] = u[k4+1]; |
2804 |
|
✗ |
d0[1] = u[k4+2]; |
2805 |
|
✗ |
d0[0] = u[k4+3]; |
2806 |
|
|
|
2807 |
|
✗ |
d0 -= 4; |
2808 |
|
✗ |
d1 -= 4; |
2809 |
|
✗ |
bitrev += 2; |
2810 |
|
|
} |
2811 |
|
|
} |
2812 |
|
|
// (paper output is u, now v) |
2813 |
|
|
|
2814 |
|
|
|
2815 |
|
|
// data must be in buf2 |
2816 |
|
|
assert(v == buf2); |
2817 |
|
|
|
2818 |
|
|
// step 7 (paper output is v, now v) |
2819 |
|
|
// this is now in place |
2820 |
|
|
{ |
2821 |
|
✗ |
float *C = f->C[blocktype]; |
2822 |
|
|
float *d, *e; |
2823 |
|
|
|
2824 |
|
|
d = v; |
2825 |
|
|
e = v + n2 - 4; |
2826 |
|
|
|
2827 |
|
✗ |
while (d < e) { |
2828 |
|
|
float a02,a11,b0,b1,b2,b3; |
2829 |
|
|
|
2830 |
|
✗ |
a02 = d[0] - e[2]; |
2831 |
|
✗ |
a11 = d[1] + e[3]; |
2832 |
|
|
|
2833 |
|
✗ |
b0 = C[1]*a02 + C[0]*a11; |
2834 |
|
✗ |
b1 = C[1]*a11 - C[0]*a02; |
2835 |
|
|
|
2836 |
|
✗ |
b2 = d[0] + e[ 2]; |
2837 |
|
✗ |
b3 = d[1] - e[ 3]; |
2838 |
|
|
|
2839 |
|
✗ |
d[0] = b2 + b0; |
2840 |
|
✗ |
d[1] = b3 + b1; |
2841 |
|
✗ |
e[2] = b2 - b0; |
2842 |
|
✗ |
e[3] = b1 - b3; |
2843 |
|
|
|
2844 |
|
✗ |
a02 = d[2] - e[0]; |
2845 |
|
✗ |
a11 = d[3] + e[1]; |
2846 |
|
|
|
2847 |
|
✗ |
b0 = C[3]*a02 + C[2]*a11; |
2848 |
|
✗ |
b1 = C[3]*a11 - C[2]*a02; |
2849 |
|
|
|
2850 |
|
✗ |
b2 = d[2] + e[ 0]; |
2851 |
|
✗ |
b3 = d[3] - e[ 1]; |
2852 |
|
|
|
2853 |
|
✗ |
d[2] = b2 + b0; |
2854 |
|
✗ |
d[3] = b3 + b1; |
2855 |
|
✗ |
e[0] = b2 - b0; |
2856 |
|
✗ |
e[1] = b1 - b3; |
2857 |
|
|
|
2858 |
|
✗ |
C += 4; |
2859 |
|
✗ |
d += 4; |
2860 |
|
✗ |
e -= 4; |
2861 |
|
|
} |
2862 |
|
|
} |
2863 |
|
|
|
2864 |
|
|
// data must be in buf2 |
2865 |
|
|
|
2866 |
|
|
|
2867 |
|
|
// step 8+decode (paper output is X, now buffer) |
2868 |
|
|
// this generates pairs of data a la 8 and pushes them directly through |
2869 |
|
|
// the decode kernel (pushing rather than pulling) to avoid having |
2870 |
|
|
// to make another pass later |
2871 |
|
|
|
2872 |
|
|
// this cannot POSSIBLY be in place, so we refer to the buffers directly |
2873 |
|
|
|
2874 |
|
|
{ |
2875 |
|
|
float *d0,*d1,*d2,*d3; |
2876 |
|
|
|
2877 |
|
✗ |
float *B = f->B[blocktype] + n2 - 8; |
2878 |
|
✗ |
float *e = buf2 + n2 - 8; |
2879 |
|
|
d0 = &buffer[0]; |
2880 |
|
✗ |
d1 = &buffer[n2-4]; |
2881 |
|
|
d2 = &buffer[n2]; |
2882 |
|
✗ |
d3 = &buffer[n-4]; |
2883 |
|
✗ |
while (e >= v) { |
2884 |
|
|
float p0,p1,p2,p3; |
2885 |
|
|
|
2886 |
|
✗ |
p3 = e[6]*B[7] - e[7]*B[6]; |
2887 |
|
✗ |
p2 = -e[6]*B[6] - e[7]*B[7]; |
2888 |
|
|
|
2889 |
|
✗ |
d0[0] = p3; |
2890 |
|
✗ |
d1[3] = - p3; |
2891 |
|
✗ |
d2[0] = p2; |
2892 |
|
✗ |
d3[3] = p2; |
2893 |
|
|
|
2894 |
|
✗ |
p1 = e[4]*B[5] - e[5]*B[4]; |
2895 |
|
✗ |
p0 = -e[4]*B[4] - e[5]*B[5]; |
2896 |
|
|
|
2897 |
|
✗ |
d0[1] = p1; |
2898 |
|
✗ |
d1[2] = - p1; |
2899 |
|
✗ |
d2[1] = p0; |
2900 |
|
✗ |
d3[2] = p0; |
2901 |
|
|
|
2902 |
|
✗ |
p3 = e[2]*B[3] - e[3]*B[2]; |
2903 |
|
✗ |
p2 = -e[2]*B[2] - e[3]*B[3]; |
2904 |
|
|
|
2905 |
|
✗ |
d0[2] = p3; |
2906 |
|
✗ |
d1[1] = - p3; |
2907 |
|
✗ |
d2[2] = p2; |
2908 |
|
✗ |
d3[1] = p2; |
2909 |
|
|
|
2910 |
|
✗ |
p1 = e[0]*B[1] - e[1]*B[0]; |
2911 |
|
✗ |
p0 = -e[0]*B[0] - e[1]*B[1]; |
2912 |
|
|
|
2913 |
|
✗ |
d0[3] = p1; |
2914 |
|
✗ |
d1[0] = - p1; |
2915 |
|
✗ |
d2[3] = p0; |
2916 |
|
✗ |
d3[0] = p0; |
2917 |
|
|
|
2918 |
|
✗ |
B -= 8; |
2919 |
|
✗ |
e -= 8; |
2920 |
|
✗ |
d0 += 4; |
2921 |
|
✗ |
d2 += 4; |
2922 |
|
✗ |
d1 -= 4; |
2923 |
|
✗ |
d3 -= 4; |
2924 |
|
|
} |
2925 |
|
|
} |
2926 |
|
|
|
2927 |
|
|
temp_free(f,buf2); |
2928 |
|
✗ |
temp_alloc_restore(f,save_point); |
2929 |
|
|
} |
2930 |
|
|
|
2931 |
|
|
#if 0 |
2932 |
|
|
// this is the original version of the above code, if you want to optimize it from scratch |
2933 |
|
|
void inverse_mdct_naive(float *buffer, int n) |
2934 |
|
|
{ |
2935 |
|
|
float s; |
2936 |
|
|
float A[1 << 12], B[1 << 12], C[1 << 11]; |
2937 |
|
|
int i,k,k2,k4, n2 = n >> 1, n4 = n >> 2, n8 = n >> 3, l; |
2938 |
|
|
int n3_4 = n - n4, ld; |
2939 |
|
|
// how can they claim this only uses N words?! |
2940 |
|
|
// oh, because they're only used sparsely, whoops |
2941 |
|
|
float u[1 << 13], X[1 << 13], v[1 << 13], w[1 << 13]; |
2942 |
|
|
// set up twiddle factors |
2943 |
|
|
|
2944 |
|
|
for (k=k2=0; k < n4; ++k,k2+=2) { |
2945 |
|
|
A[k2 ] = (float) cos(4*k*M_PI/n); |
2946 |
|
|
A[k2+1] = (float) -sin(4*k*M_PI/n); |
2947 |
|
|
B[k2 ] = (float) cos((k2+1)*M_PI/n/2); |
2948 |
|
|
B[k2+1] = (float) sin((k2+1)*M_PI/n/2); |
2949 |
|
|
} |
2950 |
|
|
for (k=k2=0; k < n8; ++k,k2+=2) { |
2951 |
|
|
C[k2 ] = (float) cos(2*(k2+1)*M_PI/n); |
2952 |
|
|
C[k2+1] = (float) -sin(2*(k2+1)*M_PI/n); |
2953 |
|
|
} |
2954 |
|
|
|
2955 |
|
|
// IMDCT algorithm from "The use of multirate filter banks for coding of high quality digital audio" |
2956 |
|
|
// Note there are bugs in that pseudocode, presumably due to them attempting |
2957 |
|
|
// to rename the arrays nicely rather than representing the way their actual |
2958 |
|
|
// implementation bounces buffers back and forth. As a result, even in the |
2959 |
|
|
// "some formulars corrected" version, a direct implementation fails. These |
2960 |
|
|
// are noted below as "paper bug". |
2961 |
|
|
|
2962 |
|
|
// copy and reflect spectral data |
2963 |
|
|
for (k=0; k < n2; ++k) u[k] = buffer[k]; |
2964 |
|
|
for ( ; k < n ; ++k) u[k] = -buffer[n - k - 1]; |
2965 |
|
|
// kernel from paper |
2966 |
|
|
// step 1 |
2967 |
|
|
for (k=k2=k4=0; k < n4; k+=1, k2+=2, k4+=4) { |
2968 |
|
|
v[n-k4-1] = (u[k4] - u[n-k4-1]) * A[k2] - (u[k4+2] - u[n-k4-3])*A[k2+1]; |
2969 |
|
|
v[n-k4-3] = (u[k4] - u[n-k4-1]) * A[k2+1] + (u[k4+2] - u[n-k4-3])*A[k2]; |
2970 |
|
|
} |
2971 |
|
|
// step 2 |
2972 |
|
|
for (k=k4=0; k < n8; k+=1, k4+=4) { |
2973 |
|
|
w[n2+3+k4] = v[n2+3+k4] + v[k4+3]; |
2974 |
|
|
w[n2+1+k4] = v[n2+1+k4] + v[k4+1]; |
2975 |
|
|
w[k4+3] = (v[n2+3+k4] - v[k4+3])*A[n2-4-k4] - (v[n2+1+k4]-v[k4+1])*A[n2-3-k4]; |
2976 |
|
|
w[k4+1] = (v[n2+1+k4] - v[k4+1])*A[n2-4-k4] + (v[n2+3+k4]-v[k4+3])*A[n2-3-k4]; |
2977 |
|
|
} |
2978 |
|
|
// step 3 |
2979 |
|
|
ld = ilog(n) - 1; // ilog is off-by-one from normal definitions |
2980 |
|
|
for (l=0; l < ld-3; ++l) { |
2981 |
|
|
int k0 = n >> (l+2), k1 = 1 << (l+3); |
2982 |
|
|
int rlim = n >> (l+4), r4, r; |
2983 |
|
|
int s2lim = 1 << (l+2), s2; |
2984 |
|
|
for (r=r4=0; r < rlim; r4+=4,++r) { |
2985 |
|
|
for (s2=0; s2 < s2lim; s2+=2) { |
2986 |
|
|
u[n-1-k0*s2-r4] = w[n-1-k0*s2-r4] + w[n-1-k0*(s2+1)-r4]; |
2987 |
|
|
u[n-3-k0*s2-r4] = w[n-3-k0*s2-r4] + w[n-3-k0*(s2+1)-r4]; |
2988 |
|
|
u[n-1-k0*(s2+1)-r4] = (w[n-1-k0*s2-r4] - w[n-1-k0*(s2+1)-r4]) * A[r*k1] |
2989 |
|
|
- (w[n-3-k0*s2-r4] - w[n-3-k0*(s2+1)-r4]) * A[r*k1+1]; |
2990 |
|
|
u[n-3-k0*(s2+1)-r4] = (w[n-3-k0*s2-r4] - w[n-3-k0*(s2+1)-r4]) * A[r*k1] |
2991 |
|
|
+ (w[n-1-k0*s2-r4] - w[n-1-k0*(s2+1)-r4]) * A[r*k1+1]; |
2992 |
|
|
} |
2993 |
|
|
} |
2994 |
|
|
if (l+1 < ld-3) { |
2995 |
|
|
// paper bug: ping-ponging of u&w here is omitted |
2996 |
|
|
memcpy(w, u, sizeof(u)); |
2997 |
|
|
} |
2998 |
|
|
} |
2999 |
|
|
|
3000 |
|
|
// step 4 |
3001 |
|
|
for (i=0; i < n8; ++i) { |
3002 |
|
|
int j = bit_reverse(i) >> (32-ld+3); |
3003 |
|
|
assert(j < n8); |
3004 |
|
|
if (i == j) { |
3005 |
|
|
// paper bug: original code probably swapped in place; if copying, |
3006 |
|
|
// need to directly copy in this case |
3007 |
|
|
int i8 = i << 3; |
3008 |
|
|
v[i8+1] = u[i8+1]; |
3009 |
|
|
v[i8+3] = u[i8+3]; |
3010 |
|
|
v[i8+5] = u[i8+5]; |
3011 |
|
|
v[i8+7] = u[i8+7]; |
3012 |
|
|
} else if (i < j) { |
3013 |
|
|
int i8 = i << 3, j8 = j << 3; |
3014 |
|
|
v[j8+1] = u[i8+1], v[i8+1] = u[j8 + 1]; |
3015 |
|
|
v[j8+3] = u[i8+3], v[i8+3] = u[j8 + 3]; |
3016 |
|
|
v[j8+5] = u[i8+5], v[i8+5] = u[j8 + 5]; |
3017 |
|
|
v[j8+7] = u[i8+7], v[i8+7] = u[j8 + 7]; |
3018 |
|
|
} |
3019 |
|
|
} |
3020 |
|
|
// step 5 |
3021 |
|
|
for (k=0; k < n2; ++k) { |
3022 |
|
|
w[k] = v[k*2+1]; |
3023 |
|
|
} |
3024 |
|
|
// step 6 |
3025 |
|
|
for (k=k2=k4=0; k < n8; ++k, k2 += 2, k4 += 4) { |
3026 |
|
|
u[n-1-k2] = w[k4]; |
3027 |
|
|
u[n-2-k2] = w[k4+1]; |
3028 |
|
|
u[n3_4 - 1 - k2] = w[k4+2]; |
3029 |
|
|
u[n3_4 - 2 - k2] = w[k4+3]; |
3030 |
|
|
} |
3031 |
|
|
// step 7 |
3032 |
|
|
for (k=k2=0; k < n8; ++k, k2 += 2) { |
3033 |
|
|
v[n2 + k2 ] = ( u[n2 + k2] + u[n-2-k2] + C[k2+1]*(u[n2+k2]-u[n-2-k2]) + C[k2]*(u[n2+k2+1]+u[n-2-k2+1]))/2; |
3034 |
|
|
v[n-2 - k2] = ( u[n2 + k2] + u[n-2-k2] - C[k2+1]*(u[n2+k2]-u[n-2-k2]) - C[k2]*(u[n2+k2+1]+u[n-2-k2+1]))/2; |
3035 |
|
|
v[n2+1+ k2] = ( u[n2+1+k2] - u[n-1-k2] + C[k2+1]*(u[n2+1+k2]+u[n-1-k2]) - C[k2]*(u[n2+k2]-u[n-2-k2]))/2; |
3036 |
|
|
v[n-1 - k2] = (-u[n2+1+k2] + u[n-1-k2] + C[k2+1]*(u[n2+1+k2]+u[n-1-k2]) - C[k2]*(u[n2+k2]-u[n-2-k2]))/2; |
3037 |
|
|
} |
3038 |
|
|
// step 8 |
3039 |
|
|
for (k=k2=0; k < n4; ++k,k2 += 2) { |
3040 |
|
|
X[k] = v[k2+n2]*B[k2 ] + v[k2+1+n2]*B[k2+1]; |
3041 |
|
|
X[n2-1-k] = v[k2+n2]*B[k2+1] - v[k2+1+n2]*B[k2 ]; |
3042 |
|
|
} |
3043 |
|
|
|
3044 |
|
|
// decode kernel to output |
3045 |
|
|
// determined the following value experimentally |
3046 |
|
|
// (by first figuring out what made inverse_mdct_slow work); then matching that here |
3047 |
|
|
// (probably vorbis encoder premultiplies by n or n/2, to save it on the decoder?) |
3048 |
|
|
s = 0.5; // theoretically would be n4 |
3049 |
|
|
|
3050 |
|
|
// [[[ note! the s value of 0.5 is compensated for by the B[] in the current code, |
3051 |
|
|
// so it needs to use the "old" B values to behave correctly, or else |
3052 |
|
|
// set s to 1.0 ]]] |
3053 |
|
|
for (i=0; i < n4 ; ++i) buffer[i] = s * X[i+n4]; |
3054 |
|
|
for ( ; i < n3_4; ++i) buffer[i] = -s * X[n3_4 - i - 1]; |
3055 |
|
|
for ( ; i < n ; ++i) buffer[i] = -s * X[i - n3_4]; |
3056 |
|
|
} |
3057 |
|
|
#endif |
3058 |
|
|
|
3059 |
|
|
static float *get_window(vorb *f, int len) |
3060 |
|
|
{ |
3061 |
|
✗ |
len <<= 1; |
3062 |
|
✗ |
if (len == f->blocksize_0) return f->window[0]; |
3063 |
|
✗ |
if (len == f->blocksize_1) return f->window[1]; |
3064 |
|
|
return NULL; |
3065 |
|
|
} |
3066 |
|
|
|
3067 |
|
|
#ifndef STB_VORBIS_NO_DEFER_FLOOR |
3068 |
|
|
typedef int16 YTYPE; |
3069 |
|
|
#else |
3070 |
|
|
typedef int YTYPE; |
3071 |
|
|
#endif |
3072 |
|
✗ |
static int do_floor(vorb *f, Mapping *map, int i, int n, float *target, YTYPE *finalY, uint8 *step2_flag) |
3073 |
|
|
{ |
3074 |
|
✗ |
int n2 = n >> 1; |
3075 |
|
✗ |
int s = map->chan[i].mux, floor; |
3076 |
|
✗ |
floor = map->submap_floor[s]; |
3077 |
|
✗ |
if (f->floor_types[floor] == 0) { |
3078 |
|
✗ |
return error(f, VORBIS_invalid_stream); |
3079 |
|
|
} else { |
3080 |
|
✗ |
Floor1 *g = &f->floor_config[floor].floor1; |
3081 |
|
|
int j,q; |
3082 |
|
✗ |
int lx = 0, ly = finalY[0] * g->floor1_multiplier; |
3083 |
|
✗ |
for (q=1; q < g->values; ++q) { |
3084 |
|
✗ |
j = g->sorted_order[q]; |
3085 |
|
|
#ifndef STB_VORBIS_NO_DEFER_FLOOR |
3086 |
|
|
STBV_NOTUSED(step2_flag); |
3087 |
|
✗ |
if (finalY[j] >= 0) |
3088 |
|
|
#else |
3089 |
|
|
if (step2_flag[j]) |
3090 |
|
|
#endif |
3091 |
|
|
{ |
3092 |
|
✗ |
int hy = finalY[j] * g->floor1_multiplier; |
3093 |
|
✗ |
int hx = g->Xlist[j]; |
3094 |
|
✗ |
if (lx != hx) |
3095 |
|
✗ |
draw_line(target, lx,ly, hx,hy, n2); |
3096 |
|
|
CHECK(f); |
3097 |
|
|
lx = hx, ly = hy; |
3098 |
|
|
} |
3099 |
|
|
} |
3100 |
|
✗ |
if (lx < n2) { |
3101 |
|
|
// optimization of: draw_line(target, lx,ly, n,ly, n2); |
3102 |
|
✗ |
for (j=lx; j < n2; ++j) |
3103 |
|
✗ |
LINE_OP(target[j], inverse_db_table[ly]); |
3104 |
|
|
CHECK(f); |
3105 |
|
|
} |
3106 |
|
|
} |
3107 |
|
|
return TRUE; |
3108 |
|
|
} |
3109 |
|
|
|
3110 |
|
|
// The meaning of "left" and "right" |
3111 |
|
|
// |
3112 |
|
|
// For a given frame: |
3113 |
|
|
// we compute samples from 0..n |
3114 |
|
|
// window_center is n/2 |
3115 |
|
|
// we'll window and mix the samples from left_start to left_end with data from the previous frame |
3116 |
|
|
// all of the samples from left_end to right_start can be output without mixing; however, |
3117 |
|
|
// this interval is 0-length except when transitioning between short and long frames |
3118 |
|
|
// all of the samples from right_start to right_end need to be mixed with the next frame, |
3119 |
|
|
// which we don't have, so those get saved in a buffer |
3120 |
|
|
// frame N's right_end-right_start, the number of samples to mix with the next frame, |
3121 |
|
|
// has to be the same as frame N+1's left_end-left_start (which they are by |
3122 |
|
|
// construction) |
3123 |
|
|
|
3124 |
|
✗ |
static int vorbis_decode_initial(vorb *f, int *p_left_start, int *p_left_end, int *p_right_start, int *p_right_end, int *mode) |
3125 |
|
|
{ |
3126 |
|
|
Mode *m; |
3127 |
|
|
int i, n, prev, next, window_center; |
3128 |
|
✗ |
f->channel_buffer_start = f->channel_buffer_end = 0; |
3129 |
|
|
|
3130 |
|
✗ |
retry: |
3131 |
|
✗ |
if (f->eof) return FALSE; |
3132 |
|
✗ |
if (!maybe_start_packet(f)) |
3133 |
|
|
return FALSE; |
3134 |
|
|
// check packet type |
3135 |
|
✗ |
if (get_bits(f,1) != 0) { |
3136 |
|
✗ |
if (IS_PUSH_MODE(f)) |
3137 |
|
✗ |
return error(f,VORBIS_bad_packet_type); |
3138 |
|
✗ |
while (EOP != get8_packet(f)); |
3139 |
|
✗ |
goto retry; |
3140 |
|
|
} |
3141 |
|
|
|
3142 |
|
|
if (f->alloc.alloc_buffer) |
3143 |
|
|
assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset); |
3144 |
|
|
|
3145 |
|
✗ |
i = get_bits(f, ilog(f->mode_count-1)); |
3146 |
|
✗ |
if (i == EOP) return FALSE; |
3147 |
|
✗ |
if (i >= f->mode_count) return FALSE; |
3148 |
|
✗ |
*mode = i; |
3149 |
|
✗ |
m = f->mode_config + i; |
3150 |
|
✗ |
if (m->blockflag) { |
3151 |
|
✗ |
n = f->blocksize_1; |
3152 |
|
✗ |
prev = get_bits(f,1); |
3153 |
|
✗ |
next = get_bits(f,1); |
3154 |
|
|
} else { |
3155 |
|
|
prev = next = 0; |
3156 |
|
✗ |
n = f->blocksize_0; |
3157 |
|
|
} |
3158 |
|
|
|
3159 |
|
|
// WINDOWING |
3160 |
|
|
|
3161 |
|
✗ |
window_center = n >> 1; |
3162 |
|
✗ |
if (m->blockflag && !prev) { |
3163 |
|
✗ |
*p_left_start = (n - f->blocksize_0) >> 2; |
3164 |
|
✗ |
*p_left_end = (n + f->blocksize_0) >> 2; |
3165 |
|
|
} else { |
3166 |
|
✗ |
*p_left_start = 0; |
3167 |
|
✗ |
*p_left_end = window_center; |
3168 |
|
|
} |
3169 |
|
✗ |
if (m->blockflag && !next) { |
3170 |
|
✗ |
*p_right_start = (n*3 - f->blocksize_0) >> 2; |
3171 |
|
✗ |
*p_right_end = (n*3 + f->blocksize_0) >> 2; |
3172 |
|
|
} else { |
3173 |
|
✗ |
*p_right_start = window_center; |
3174 |
|
✗ |
*p_right_end = n; |
3175 |
|
|
} |
3176 |
|
|
|
3177 |
|
|
return TRUE; |
3178 |
|
|
} |
3179 |
|
|
|
3180 |
|
✗ |
static int vorbis_decode_packet_rest(vorb *f, int *len, Mode *m, int left_start, int left_end, int right_start, int right_end, int *p_left) |
3181 |
|
|
{ |
3182 |
|
|
Mapping *map; |
3183 |
|
|
int i,j,k,n,n2; |
3184 |
|
|
int zero_channel[256]; |
3185 |
|
|
int really_zero_channel[256]; |
3186 |
|
|
|
3187 |
|
|
// WINDOWING |
3188 |
|
|
|
3189 |
|
|
STBV_NOTUSED(left_end); |
3190 |
|
✗ |
n = f->blocksize[m->blockflag]; |
3191 |
|
✗ |
map = &f->mapping[m->mapping]; |
3192 |
|
|
|
3193 |
|
|
// FLOORS |
3194 |
|
✗ |
n2 = n >> 1; |
3195 |
|
|
|
3196 |
|
|
CHECK(f); |
3197 |
|
|
|
3198 |
|
✗ |
for (i=0; i < f->channels; ++i) { |
3199 |
|
✗ |
int s = map->chan[i].mux, floor; |
3200 |
|
✗ |
zero_channel[i] = FALSE; |
3201 |
|
✗ |
floor = map->submap_floor[s]; |
3202 |
|
✗ |
if (f->floor_types[floor] == 0) { |
3203 |
|
✗ |
return error(f, VORBIS_invalid_stream); |
3204 |
|
|
} else { |
3205 |
|
✗ |
Floor1 *g = &f->floor_config[floor].floor1; |
3206 |
|
✗ |
if (get_bits(f, 1)) { |
3207 |
|
|
short *finalY; |
3208 |
|
|
uint8 step2_flag[256]; |
3209 |
|
|
static int range_list[4] = { 256, 128, 86, 64 }; |
3210 |
|
✗ |
int range = range_list[g->floor1_multiplier-1]; |
3211 |
|
|
int offset = 2; |
3212 |
|
✗ |
finalY = f->finalY[i]; |
3213 |
|
✗ |
finalY[0] = get_bits(f, ilog(range)-1); |
3214 |
|
✗ |
finalY[1] = get_bits(f, ilog(range)-1); |
3215 |
|
✗ |
for (j=0; j < g->partitions; ++j) { |
3216 |
|
✗ |
int pclass = g->partition_class_list[j]; |
3217 |
|
✗ |
int cdim = g->class_dimensions[pclass]; |
3218 |
|
✗ |
int cbits = g->class_subclasses[pclass]; |
3219 |
|
✗ |
int csub = (1 << cbits)-1; |
3220 |
|
|
int cval = 0; |
3221 |
|
✗ |
if (cbits) { |
3222 |
|
✗ |
Codebook *c = f->codebooks + g->class_masterbooks[pclass]; |
3223 |
|
✗ |
DECODE(cval,f,c); |
3224 |
|
|
} |
3225 |
|
✗ |
for (k=0; k < cdim; ++k) { |
3226 |
|
✗ |
int book = g->subclass_books[pclass][cval & csub]; |
3227 |
|
✗ |
cval = cval >> cbits; |
3228 |
|
✗ |
if (book >= 0) { |
3229 |
|
|
int temp; |
3230 |
|
✗ |
Codebook *c = f->codebooks + book; |
3231 |
|
✗ |
DECODE(temp,f,c); |
3232 |
|
✗ |
finalY[offset++] = temp; |
3233 |
|
|
} else |
3234 |
|
✗ |
finalY[offset++] = 0; |
3235 |
|
|
} |
3236 |
|
|
} |
3237 |
|
✗ |
if (f->valid_bits == INVALID_BITS) goto error; // behavior according to spec |
3238 |
|
✗ |
step2_flag[0] = step2_flag[1] = 1; |
3239 |
|
✗ |
for (j=2; j < g->values; ++j) { |
3240 |
|
|
int low, high, pred, highroom, lowroom, room, val; |
3241 |
|
✗ |
low = g->neighbors[j][0]; |
3242 |
|
✗ |
high = g->neighbors[j][1]; |
3243 |
|
|
//neighbors(g->Xlist, j, &low, &high); |
3244 |
|
✗ |
pred = predict_point(g->Xlist[j], g->Xlist[low], g->Xlist[high], finalY[low], finalY[high]); |
3245 |
|
✗ |
val = finalY[j]; |
3246 |
|
✗ |
highroom = range - pred; |
3247 |
|
|
lowroom = pred; |
3248 |
|
✗ |
if (highroom < lowroom) |
3249 |
|
✗ |
room = highroom * 2; |
3250 |
|
|
else |
3251 |
|
✗ |
room = lowroom * 2; |
3252 |
|
✗ |
if (val) { |
3253 |
|
✗ |
step2_flag[low] = step2_flag[high] = 1; |
3254 |
|
✗ |
step2_flag[j] = 1; |
3255 |
|
✗ |
if (val >= room) |
3256 |
|
✗ |
if (highroom > lowroom) |
3257 |
|
|
finalY[j] = val - lowroom + pred; |
3258 |
|
|
else |
3259 |
|
✗ |
finalY[j] = pred - val + highroom - 1; |
3260 |
|
|
else |
3261 |
|
✗ |
if (val & 1) |
3262 |
|
✗ |
finalY[j] = pred - ((val+1)>>1); |
3263 |
|
|
else |
3264 |
|
✗ |
finalY[j] = pred + (val>>1); |
3265 |
|
|
} else { |
3266 |
|
✗ |
step2_flag[j] = 0; |
3267 |
|
✗ |
finalY[j] = pred; |
3268 |
|
|
} |
3269 |
|
|
} |
3270 |
|
|
|
3271 |
|
|
#ifdef STB_VORBIS_NO_DEFER_FLOOR |
3272 |
|
|
do_floor(f, map, i, n, f->floor_buffers[i], finalY, step2_flag); |
3273 |
|
|
#else |
3274 |
|
|
// defer final floor computation until _after_ residue |
3275 |
|
✗ |
for (j=0; j < g->values; ++j) { |
3276 |
|
✗ |
if (!step2_flag[j]) |
3277 |
|
✗ |
finalY[j] = -1; |
3278 |
|
|
} |
3279 |
|
|
#endif |
3280 |
|
|
} else { |
3281 |
|
✗ |
error: |
3282 |
|
✗ |
zero_channel[i] = TRUE; |
3283 |
|
|
} |
3284 |
|
|
// So we just defer everything else to later |
3285 |
|
|
|
3286 |
|
|
// at this point we've decoded the floor into buffer |
3287 |
|
|
} |
3288 |
|
|
} |
3289 |
|
|
CHECK(f); |
3290 |
|
|
// at this point we've decoded all floors |
3291 |
|
|
|
3292 |
|
|
if (f->alloc.alloc_buffer) |
3293 |
|
|
assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset); |
3294 |
|
|
|
3295 |
|
|
// re-enable coupled channels if necessary |
3296 |
|
✗ |
memcpy(really_zero_channel, zero_channel, sizeof(really_zero_channel[0]) * f->channels); |
3297 |
|
✗ |
for (i=0; i < map->coupling_steps; ++i) |
3298 |
|
✗ |
if (!zero_channel[map->chan[i].magnitude] || !zero_channel[map->chan[i].angle]) { |
3299 |
|
✗ |
zero_channel[map->chan[i].magnitude] = zero_channel[map->chan[i].angle] = FALSE; |
3300 |
|
|
} |
3301 |
|
|
|
3302 |
|
|
CHECK(f); |
3303 |
|
|
// RESIDUE DECODE |
3304 |
|
✗ |
for (i=0; i < map->submaps; ++i) { |
3305 |
|
|
float *residue_buffers[STB_VORBIS_MAX_CHANNELS]; |
3306 |
|
|
int r; |
3307 |
|
|
uint8 do_not_decode[256]; |
3308 |
|
|
int ch = 0; |
3309 |
|
✗ |
for (j=0; j < f->channels; ++j) { |
3310 |
|
✗ |
if (map->chan[j].mux == i) { |
3311 |
|
✗ |
if (zero_channel[j]) { |
3312 |
|
✗ |
do_not_decode[ch] = TRUE; |
3313 |
|
✗ |
residue_buffers[ch] = NULL; |
3314 |
|
|
} else { |
3315 |
|
✗ |
do_not_decode[ch] = FALSE; |
3316 |
|
✗ |
residue_buffers[ch] = f->channel_buffers[j]; |
3317 |
|
|
} |
3318 |
|
✗ |
++ch; |
3319 |
|
|
} |
3320 |
|
|
} |
3321 |
|
✗ |
r = map->submap_residue[i]; |
3322 |
|
✗ |
decode_residue(f, residue_buffers, ch, n2, r, do_not_decode); |
3323 |
|
|
} |
3324 |
|
|
|
3325 |
|
|
if (f->alloc.alloc_buffer) |
3326 |
|
|
assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset); |
3327 |
|
|
CHECK(f); |
3328 |
|
|
|
3329 |
|
|
// INVERSE COUPLING |
3330 |
|
✗ |
for (i = map->coupling_steps-1; i >= 0; --i) { |
3331 |
|
|
int n2 = n >> 1; |
3332 |
|
✗ |
float *m = f->channel_buffers[map->chan[i].magnitude]; |
3333 |
|
✗ |
float *a = f->channel_buffers[map->chan[i].angle ]; |
3334 |
|
✗ |
for (j=0; j < n2; ++j) { |
3335 |
|
|
float a2,m2; |
3336 |
|
✗ |
if (m[j] > 0) |
3337 |
|
✗ |
if (a[j] > 0) |
3338 |
|
✗ |
m2 = m[j], a2 = m[j] - a[j]; |
3339 |
|
|
else |
3340 |
|
✗ |
a2 = m[j], m2 = m[j] + a[j]; |
3341 |
|
|
else |
3342 |
|
✗ |
if (a[j] > 0) |
3343 |
|
✗ |
m2 = m[j], a2 = m[j] + a[j]; |
3344 |
|
|
else |
3345 |
|
✗ |
a2 = m[j], m2 = m[j] - a[j]; |
3346 |
|
✗ |
m[j] = m2; |
3347 |
|
✗ |
a[j] = a2; |
3348 |
|
|
} |
3349 |
|
|
} |
3350 |
|
|
CHECK(f); |
3351 |
|
|
|
3352 |
|
|
// finish decoding the floors |
3353 |
|
|
#ifndef STB_VORBIS_NO_DEFER_FLOOR |
3354 |
|
✗ |
for (i=0; i < f->channels; ++i) { |
3355 |
|
✗ |
if (really_zero_channel[i]) { |
3356 |
|
✗ |
memset(f->channel_buffers[i], 0, sizeof(*f->channel_buffers[i]) * n2); |
3357 |
|
|
} else { |
3358 |
|
✗ |
do_floor(f, map, i, n, f->channel_buffers[i], f->finalY[i], NULL); |
3359 |
|
|
} |
3360 |
|
|
} |
3361 |
|
|
#else |
3362 |
|
|
for (i=0; i < f->channels; ++i) { |
3363 |
|
|
if (really_zero_channel[i]) { |
3364 |
|
|
memset(f->channel_buffers[i], 0, sizeof(*f->channel_buffers[i]) * n2); |
3365 |
|
|
} else { |
3366 |
|
|
for (j=0; j < n2; ++j) |
3367 |
|
|
f->channel_buffers[i][j] *= f->floor_buffers[i][j]; |
3368 |
|
|
} |
3369 |
|
|
} |
3370 |
|
|
#endif |
3371 |
|
|
|
3372 |
|
|
// INVERSE MDCT |
3373 |
|
|
CHECK(f); |
3374 |
|
✗ |
for (i=0; i < f->channels; ++i) |
3375 |
|
✗ |
inverse_mdct(f->channel_buffers[i], n, f, m->blockflag); |
3376 |
|
|
CHECK(f); |
3377 |
|
|
|
3378 |
|
|
// this shouldn't be necessary, unless we exited on an error |
3379 |
|
|
// and want to flush to get to the next packet |
3380 |
|
|
flush_packet(f); |
3381 |
|
|
|
3382 |
|
✗ |
if (f->first_decode) { |
3383 |
|
|
// assume we start so first non-discarded sample is sample 0 |
3384 |
|
|
// this isn't to spec, but spec would require us to read ahead |
3385 |
|
|
// and decode the size of all current frames--could be done, |
3386 |
|
|
// but presumably it's not a commonly used feature |
3387 |
|
✗ |
f->current_loc = 0u - n2; // start of first frame is positioned for discard (NB this is an intentional unsigned overflow/wrap-around) |
3388 |
|
|
// we might have to discard samples "from" the next frame too, |
3389 |
|
|
// if we're lapping a large block then a small at the start? |
3390 |
|
✗ |
f->discard_samples_deferred = n - right_end; |
3391 |
|
✗ |
f->current_loc_valid = TRUE; |
3392 |
|
✗ |
f->first_decode = FALSE; |
3393 |
|
✗ |
} else if (f->discard_samples_deferred) { |
3394 |
|
✗ |
if (f->discard_samples_deferred >= right_start - left_start) { |
3395 |
|
✗ |
f->discard_samples_deferred -= (right_start - left_start); |
3396 |
|
|
left_start = right_start; |
3397 |
|
✗ |
*p_left = left_start; |
3398 |
|
|
} else { |
3399 |
|
✗ |
left_start += f->discard_samples_deferred; |
3400 |
|
✗ |
*p_left = left_start; |
3401 |
|
✗ |
f->discard_samples_deferred = 0; |
3402 |
|
|
} |
3403 |
|
✗ |
} else if (f->previous_length == 0 && f->current_loc_valid) { |
3404 |
|
|
// we're recovering from a seek... that means we're going to discard |
3405 |
|
|
// the samples from this packet even though we know our position from |
3406 |
|
|
// the last page header, so we need to update the position based on |
3407 |
|
|
// the discarded samples here |
3408 |
|
|
// but wait, the code below is going to add this in itself even |
3409 |
|
|
// on a discard, so we don't need to do it here... |
3410 |
|
|
} |
3411 |
|
|
|
3412 |
|
|
// check if we have ogg information about the sample # for this packet |
3413 |
|
✗ |
if (f->last_seg_which == f->end_seg_with_known_loc) { |
3414 |
|
|
// if we have a valid current loc, and this is final: |
3415 |
|
✗ |
if (f->current_loc_valid && (f->page_flag & PAGEFLAG_last_page)) { |
3416 |
|
✗ |
uint32 current_end = f->known_loc_for_packet; |
3417 |
|
|
// then let's infer the size of the (probably) short final frame |
3418 |
|
✗ |
if (current_end < f->current_loc + (right_end-left_start)) { |
3419 |
|
✗ |
if (current_end < f->current_loc) { |
3420 |
|
|
// negative truncation, that's impossible! |
3421 |
|
✗ |
*len = 0; |
3422 |
|
|
} else { |
3423 |
|
✗ |
*len = current_end - f->current_loc; |
3424 |
|
|
} |
3425 |
|
✗ |
*len += left_start; // this doesn't seem right, but has no ill effect on my test files |
3426 |
|
✗ |
if (*len > right_end) *len = right_end; // this should never happen |
3427 |
|
✗ |
f->current_loc += *len; |
3428 |
|
✗ |
return TRUE; |
3429 |
|
|
} |
3430 |
|
|
} |
3431 |
|
|
// otherwise, just set our sample loc |
3432 |
|
|
// guess that the ogg granule pos refers to the _middle_ of the |
3433 |
|
|
// last frame? |
3434 |
|
|
// set f->current_loc to the position of left_start |
3435 |
|
✗ |
f->current_loc = f->known_loc_for_packet - (n2-left_start); |
3436 |
|
✗ |
f->current_loc_valid = TRUE; |
3437 |
|
|
} |
3438 |
|
✗ |
if (f->current_loc_valid) |
3439 |
|
✗ |
f->current_loc += (right_start - left_start); |
3440 |
|
|
|
3441 |
|
|
if (f->alloc.alloc_buffer) |
3442 |
|
|
assert(f->alloc.alloc_buffer_length_in_bytes == f->temp_offset); |
3443 |
|
✗ |
*len = right_end; // ignore samples after the window goes to 0 |
3444 |
|
|
CHECK(f); |
3445 |
|
|
|
3446 |
|
✗ |
return TRUE; |
3447 |
|
|
} |
3448 |
|
|
|
3449 |
|
✗ |
static int vorbis_decode_packet(vorb *f, int *len, int *p_left, int *p_right) |
3450 |
|
|
{ |
3451 |
|
|
int mode, left_end, right_end; |
3452 |
|
✗ |
if (!vorbis_decode_initial(f, p_left, &left_end, p_right, &right_end, &mode)) return 0; |
3453 |
|
✗ |
return vorbis_decode_packet_rest(f, len, f->mode_config + mode, *p_left, left_end, *p_right, right_end, p_left); |
3454 |
|
|
} |
3455 |
|
|
|
3456 |
|
✗ |
static int vorbis_finish_frame(stb_vorbis *f, int len, int left, int right) |
3457 |
|
|
{ |
3458 |
|
|
int prev,i,j; |
3459 |
|
|
// we use right&left (the start of the right- and left-window sin()-regions) |
3460 |
|
|
// to determine how much to return, rather than inferring from the rules |
3461 |
|
|
// (same result, clearer code); 'left' indicates where our sin() window |
3462 |
|
|
// starts, therefore where the previous window's right edge starts, and |
3463 |
|
|
// therefore where to start mixing from the previous buffer. 'right' |
3464 |
|
|
// indicates where our sin() ending-window starts, therefore that's where |
3465 |
|
|
// we start saving, and where our returned-data ends. |
3466 |
|
|
|
3467 |
|
|
// mixin from previous window |
3468 |
|
✗ |
if (f->previous_length) { |
3469 |
|
|
int i,j, n = f->previous_length; |
3470 |
|
|
float *w = get_window(f, n); |
3471 |
|
✗ |
if (w == NULL) return 0; |
3472 |
|
✗ |
for (i=0; i < f->channels; ++i) { |
3473 |
|
✗ |
for (j=0; j < n; ++j) |
3474 |
|
✗ |
f->channel_buffers[i][left+j] = |
3475 |
|
✗ |
f->channel_buffers[i][left+j]*w[ j] + |
3476 |
|
✗ |
f->previous_window[i][ j]*w[n-1-j]; |
3477 |
|
|
} |
3478 |
|
|
} |
3479 |
|
|
|
3480 |
|
✗ |
prev = f->previous_length; |
3481 |
|
|
|
3482 |
|
|
// last half of this data becomes previous window |
3483 |
|
✗ |
f->previous_length = len - right; |
3484 |
|
|
|
3485 |
|
|
// @OPTIMIZE: could avoid this copy by double-buffering the |
3486 |
|
|
// output (flipping previous_window with channel_buffers), but |
3487 |
|
|
// then previous_window would have to be 2x as large, and |
3488 |
|
|
// channel_buffers couldn't be temp mem (although they're NOT |
3489 |
|
|
// currently temp mem, they could be (unless we want to level |
3490 |
|
|
// performance by spreading out the computation)) |
3491 |
|
✗ |
for (i=0; i < f->channels; ++i) |
3492 |
|
✗ |
for (j=0; right+j < len; ++j) |
3493 |
|
✗ |
f->previous_window[i][j] = f->channel_buffers[i][right+j]; |
3494 |
|
|
|
3495 |
|
✗ |
if (!prev) |
3496 |
|
|
// there was no previous packet, so this data isn't valid... |
3497 |
|
|
// this isn't entirely true, only the would-have-overlapped data |
3498 |
|
|
// isn't valid, but this seems to be what the spec requires |
3499 |
|
|
return 0; |
3500 |
|
|
|
3501 |
|
|
// truncate a short frame |
3502 |
|
|
if (len < right) right = len; |
3503 |
|
|
|
3504 |
|
✗ |
f->samples_output += right-left; |
3505 |
|
|
|
3506 |
|
✗ |
return right - left; |
3507 |
|
|
} |
3508 |
|
|
|
3509 |
|
✗ |
static int vorbis_pump_first_frame(stb_vorbis *f) |
3510 |
|
|
{ |
3511 |
|
|
int len, right, left, res; |
3512 |
|
✗ |
res = vorbis_decode_packet(f, &len, &left, &right); |
3513 |
|
✗ |
if (res) |
3514 |
|
✗ |
vorbis_finish_frame(f, len, left, right); |
3515 |
|
✗ |
return res; |
3516 |
|
|
} |
3517 |
|
|
|
3518 |
|
|
#ifndef STB_VORBIS_NO_PUSHDATA_API |
3519 |
|
✗ |
static int is_whole_packet_present(stb_vorbis *f) |
3520 |
|
|
{ |
3521 |
|
|
// make sure that we have the packet available before continuing... |
3522 |
|
|
// this requires a full ogg parse, but we know we can fetch from f->stream |
3523 |
|
|
|
3524 |
|
|
// instead of coding this out explicitly, we could save the current read state, |
3525 |
|
|
// read the next packet with get8() until end-of-packet, check f->eof, then |
3526 |
|
|
// reset the state? but that would be slower, esp. since we'd have over 256 bytes |
3527 |
|
|
// of state to restore (primarily the page segment table) |
3528 |
|
|
|
3529 |
|
✗ |
int s = f->next_seg, first = TRUE; |
3530 |
|
✗ |
uint8 *p = f->stream; |
3531 |
|
|
|
3532 |
|
✗ |
if (s != -1) { // if we're not starting the packet with a 'continue on next page' flag |
3533 |
|
✗ |
for (; s < f->segment_count; ++s) { |
3534 |
|
✗ |
p += f->segments[s]; |
3535 |
|
✗ |
if (f->segments[s] < 255) // stop at first short segment |
3536 |
|
|
break; |
3537 |
|
|
} |
3538 |
|
|
// either this continues, or it ends it... |
3539 |
|
✗ |
if (s == f->segment_count) |
3540 |
|
|
s = -1; // set 'crosses page' flag |
3541 |
|
✗ |
if (p > f->stream_end) return error(f, VORBIS_need_more_data); |
3542 |
|
|
first = FALSE; |
3543 |
|
|
} |
3544 |
|
✗ |
for (; s == -1;) { |
3545 |
|
|
uint8 *q; |
3546 |
|
|
int n; |
3547 |
|
|
|
3548 |
|
|
// check that we have the page header ready |
3549 |
|
✗ |
if (p + 26 >= f->stream_end) return error(f, VORBIS_need_more_data); |
3550 |
|
|
// validate the page |
3551 |
|
✗ |
if (memcmp(p, ogg_page_header, 4)) return error(f, VORBIS_invalid_stream); |
3552 |
|
✗ |
if (p[4] != 0) return error(f, VORBIS_invalid_stream); |
3553 |
|
✗ |
if (first) { // the first segment must NOT have 'continued_packet', later ones MUST |
3554 |
|
✗ |
if (f->previous_length) |
3555 |
|
✗ |
if ((p[5] & PAGEFLAG_continued_packet)) return error(f, VORBIS_invalid_stream); |
3556 |
|
|
// if no previous length, we're resynching, so we can come in on a continued-packet, |
3557 |
|
|
// which we'll just drop |
3558 |
|
|
} else { |
3559 |
|
✗ |
if (!(p[5] & PAGEFLAG_continued_packet)) return error(f, VORBIS_invalid_stream); |
3560 |
|
|
} |
3561 |
|
✗ |
n = p[26]; // segment counts |
3562 |
|
✗ |
q = p+27; // q points to segment table |
3563 |
|
✗ |
p = q + n; // advance past header |
3564 |
|
|
// make sure we've read the segment table |
3565 |
|
✗ |
if (p > f->stream_end) return error(f, VORBIS_need_more_data); |
3566 |
|
✗ |
for (s=0; s < n; ++s) { |
3567 |
|
✗ |
p += q[s]; |
3568 |
|
✗ |
if (q[s] < 255) |
3569 |
|
|
break; |
3570 |
|
|
} |
3571 |
|
✗ |
if (s == n) |
3572 |
|
|
s = -1; // set 'crosses page' flag |
3573 |
|
✗ |
if (p > f->stream_end) return error(f, VORBIS_need_more_data); |
3574 |
|
|
first = FALSE; |
3575 |
|
|
} |
3576 |
|
|
return TRUE; |
3577 |
|
|
} |
3578 |
|
|
#endif // !STB_VORBIS_NO_PUSHDATA_API |
3579 |
|
|
|
3580 |
|
✗ |
static int start_decoder(vorb *f) |
3581 |
|
|
{ |
3582 |
|
|
uint8 header[6], x,y; |
3583 |
|
|
int len,i,j,k, max_submaps = 0; |
3584 |
|
|
int longest_floorlist=0; |
3585 |
|
|
|
3586 |
|
|
// first page, first packet |
3587 |
|
✗ |
f->first_decode = TRUE; |
3588 |
|
|
|
3589 |
|
✗ |
if (!start_page(f)) return FALSE; |
3590 |
|
|
// validate page flag |
3591 |
|
✗ |
if (!(f->page_flag & PAGEFLAG_first_page)) return error(f, VORBIS_invalid_first_page); |
3592 |
|
✗ |
if (f->page_flag & PAGEFLAG_last_page) return error(f, VORBIS_invalid_first_page); |
3593 |
|
✗ |
if (f->page_flag & PAGEFLAG_continued_packet) return error(f, VORBIS_invalid_first_page); |
3594 |
|
|
// check for expected packet length |
3595 |
|
✗ |
if (f->segment_count != 1) return error(f, VORBIS_invalid_first_page); |
3596 |
|
✗ |
if (f->segments[0] != 30) { |
3597 |
|
|
// check for the Ogg skeleton fishead identifying header to refine our error |
3598 |
|
✗ |
if (f->segments[0] == 64 && |
3599 |
|
✗ |
getn(f, header, 6) && |
3600 |
|
✗ |
header[0] == 'f' && |
3601 |
|
✗ |
header[1] == 'i' && |
3602 |
|
✗ |
header[2] == 's' && |
3603 |
|
✗ |
header[3] == 'h' && |
3604 |
|
✗ |
header[4] == 'e' && |
3605 |
|
✗ |
header[5] == 'a' && |
3606 |
|
✗ |
get8(f) == 'd' && |
3607 |
|
✗ |
get8(f) == '\0') return error(f, VORBIS_ogg_skeleton_not_supported); |
3608 |
|
|
else |
3609 |
|
✗ |
return error(f, VORBIS_invalid_first_page); |
3610 |
|
|
} |
3611 |
|
|
|
3612 |
|
|
// read packet |
3613 |
|
|
// check packet header |
3614 |
|
✗ |
if (get8(f) != VORBIS_packet_id) return error(f, VORBIS_invalid_first_page); |
3615 |
|
✗ |
if (!getn(f, header, 6)) return error(f, VORBIS_unexpected_eof); |
3616 |
|
✗ |
if (!vorbis_validate(header)) return error(f, VORBIS_invalid_first_page); |
3617 |
|
|
// vorbis_version |
3618 |
|
✗ |
if (get32(f) != 0) return error(f, VORBIS_invalid_first_page); |
3619 |
|
✗ |
f->channels = get8(f); if (!f->channels) return error(f, VORBIS_invalid_first_page); |
3620 |
|
✗ |
if (f->channels > STB_VORBIS_MAX_CHANNELS) return error(f, VORBIS_too_many_channels); |
3621 |
|
✗ |
f->sample_rate = get32(f); if (!f->sample_rate) return error(f, VORBIS_invalid_first_page); |
3622 |
|
✗ |
get32(f); // bitrate_maximum |
3623 |
|
✗ |
get32(f); // bitrate_nominal |
3624 |
|
✗ |
get32(f); // bitrate_minimum |
3625 |
|
✗ |
x = get8(f); |
3626 |
|
|
{ |
3627 |
|
|
int log0,log1; |
3628 |
|
✗ |
log0 = x & 15; |
3629 |
|
✗ |
log1 = x >> 4; |
3630 |
|
✗ |
f->blocksize_0 = 1 << log0; |
3631 |
|
✗ |
f->blocksize_1 = 1 << log1; |
3632 |
|
✗ |
if (log0 < 6 || log0 > 13) return error(f, VORBIS_invalid_setup); |
3633 |
|
✗ |
if (log1 < 6 || log1 > 13) return error(f, VORBIS_invalid_setup); |
3634 |
|
✗ |
if (log0 > log1) return error(f, VORBIS_invalid_setup); |
3635 |
|
|
} |
3636 |
|
|
|
3637 |
|
|
// framing_flag |
3638 |
|
✗ |
x = get8(f); |
3639 |
|
✗ |
if (!(x & 1)) return error(f, VORBIS_invalid_first_page); |
3640 |
|
|
|
3641 |
|
|
// second packet! |
3642 |
|
✗ |
if (!start_page(f)) return FALSE; |
3643 |
|
|
|
3644 |
|
✗ |
if (!start_packet(f)) return FALSE; |
3645 |
|
|
|
3646 |
|
✗ |
if (!next_segment(f)) return FALSE; |
3647 |
|
|
|
3648 |
|
✗ |
if (get8_packet(f) != VORBIS_packet_comment) return error(f, VORBIS_invalid_setup); |
3649 |
|
✗ |
for (i=0; i < 6; ++i) header[i] = get8_packet(f); |
3650 |
|
✗ |
if (!vorbis_validate(header)) return error(f, VORBIS_invalid_setup); |
3651 |
|
|
//file vendor |
3652 |
|
✗ |
len = get32_packet(f); |
3653 |
|
✗ |
f->vendor = (char*)setup_malloc(f, sizeof(char) * (len+1)); |
3654 |
|
✗ |
if (f->vendor == NULL) return error(f, VORBIS_outofmem); |
3655 |
|
✗ |
for(i=0; i < len; ++i) { |
3656 |
|
✗ |
f->vendor[i] = get8_packet(f); |
3657 |
|
|
} |
3658 |
|
✗ |
f->vendor[len] = (char)'\0'; |
3659 |
|
|
//user comments |
3660 |
|
✗ |
f->comment_list_length = get32_packet(f); |
3661 |
|
✗ |
f->comment_list = NULL; |
3662 |
|
✗ |
if (f->comment_list_length > 0) |
3663 |
|
|
{ |
3664 |
|
✗ |
f->comment_list = (char**) setup_malloc(f, sizeof(char*) * (f->comment_list_length)); |
3665 |
|
✗ |
if (f->comment_list == NULL) return error(f, VORBIS_outofmem); |
3666 |
|
|
} |
3667 |
|
|
|
3668 |
|
✗ |
for(i=0; i < f->comment_list_length; ++i) { |
3669 |
|
✗ |
len = get32_packet(f); |
3670 |
|
✗ |
f->comment_list[i] = (char*)setup_malloc(f, sizeof(char) * (len+1)); |
3671 |
|
✗ |
if (f->comment_list[i] == NULL) return error(f, VORBIS_outofmem); |
3672 |
|
|
|
3673 |
|
✗ |
for(j=0; j < len; ++j) { |
3674 |
|
✗ |
f->comment_list[i][j] = get8_packet(f); |
3675 |
|
|
} |
3676 |
|
✗ |
f->comment_list[i][len] = (char)'\0'; |
3677 |
|
|
} |
3678 |
|
|
|
3679 |
|
|
// framing_flag |
3680 |
|
|
x = get8_packet(f); |
3681 |
|
✗ |
if (!(x & 1)) return error(f, VORBIS_invalid_setup); |
3682 |
|
|
|
3683 |
|
|
|
3684 |
|
✗ |
skip(f, f->bytes_in_seg); |
3685 |
|
✗ |
f->bytes_in_seg = 0; |
3686 |
|
|
|
3687 |
|
|
do { |
3688 |
|
✗ |
len = next_segment(f); |
3689 |
|
✗ |
skip(f, len); |
3690 |
|
✗ |
f->bytes_in_seg = 0; |
3691 |
|
✗ |
} while (len); |
3692 |
|
|
|
3693 |
|
|
// third packet! |
3694 |
|
✗ |
if (!start_packet(f)) return FALSE; |
3695 |
|
|
|
3696 |
|
|
#ifndef STB_VORBIS_NO_PUSHDATA_API |
3697 |
|
✗ |
if (IS_PUSH_MODE(f)) { |
3698 |
|
✗ |
if (!is_whole_packet_present(f)) { |
3699 |
|
|
// convert error in ogg header to write type |
3700 |
|
✗ |
if (f->error == VORBIS_invalid_stream) |
3701 |
|
✗ |
f->error = VORBIS_invalid_setup; |
3702 |
|
✗ |
return FALSE; |
3703 |
|
|
} |
3704 |
|
|
} |
3705 |
|
|
#endif |
3706 |
|
|
|
3707 |
|
|
crc32_init(); // always init it, to avoid multithread race conditions |
3708 |
|
|
|
3709 |
|
✗ |
if (get8_packet(f) != VORBIS_packet_setup) return error(f, VORBIS_invalid_setup); |
3710 |
|
✗ |
for (i=0; i < 6; ++i) header[i] = get8_packet(f); |
3711 |
|
✗ |
if (!vorbis_validate(header)) return error(f, VORBIS_invalid_setup); |
3712 |
|
|
|
3713 |
|
|
// codebooks |
3714 |
|
|
|
3715 |
|
✗ |
f->codebook_count = get_bits(f,8) + 1; |
3716 |
|
✗ |
f->codebooks = (Codebook *) setup_malloc(f, sizeof(*f->codebooks) * f->codebook_count); |
3717 |
|
✗ |
if (f->codebooks == NULL) return error(f, VORBIS_outofmem); |
3718 |
|
✗ |
memset(f->codebooks, 0, sizeof(*f->codebooks) * f->codebook_count); |
3719 |
|
✗ |
for (i=0; i < f->codebook_count; ++i) { |
3720 |
|
|
uint32 *values; |
3721 |
|
|
int ordered, sorted_count; |
3722 |
|
|
int total=0; |
3723 |
|
|
uint8 *lengths; |
3724 |
|
✗ |
Codebook *c = f->codebooks+i; |
3725 |
|
|
CHECK(f); |
3726 |
|
✗ |
x = get_bits(f, 8); if (x != 0x42) return error(f, VORBIS_invalid_setup); |
3727 |
|
✗ |
x = get_bits(f, 8); if (x != 0x43) return error(f, VORBIS_invalid_setup); |
3728 |
|
✗ |
x = get_bits(f, 8); if (x != 0x56) return error(f, VORBIS_invalid_setup); |
3729 |
|
✗ |
x = get_bits(f, 8); |
3730 |
|
✗ |
c->dimensions = (get_bits(f, 8)<<8) + x; |
3731 |
|
✗ |
x = get_bits(f, 8); |
3732 |
|
✗ |
y = get_bits(f, 8); |
3733 |
|
✗ |
c->entries = (get_bits(f, 8)<<16) + (y<<8) + x; |
3734 |
|
✗ |
ordered = get_bits(f,1); |
3735 |
|
✗ |
c->sparse = ordered ? 0 : get_bits(f,1); |
3736 |
|
|
|
3737 |
|
✗ |
if (c->dimensions == 0 && c->entries != 0) return error(f, VORBIS_invalid_setup); |
3738 |
|
|
|
3739 |
|
✗ |
if (c->sparse) |
3740 |
|
✗ |
lengths = (uint8 *) setup_temp_malloc(f, c->entries); |
3741 |
|
|
else |
3742 |
|
✗ |
lengths = c->codeword_lengths = (uint8 *) setup_malloc(f, c->entries); |
3743 |
|
|
|
3744 |
|
✗ |
if (!lengths) return error(f, VORBIS_outofmem); |
3745 |
|
|
|
3746 |
|
✗ |
if (ordered) { |
3747 |
|
|
int current_entry = 0; |
3748 |
|
✗ |
int current_length = get_bits(f,5) + 1; |
3749 |
|
✗ |
while (current_entry < c->entries) { |
3750 |
|
✗ |
int limit = c->entries - current_entry; |
3751 |
|
✗ |
int n = get_bits(f, ilog(limit)); |
3752 |
|
✗ |
if (current_length >= 32) return error(f, VORBIS_invalid_setup); |
3753 |
|
✗ |
if (current_entry + n > (int) c->entries) { return error(f, VORBIS_invalid_setup); } |
3754 |
|
✗ |
memset(lengths + current_entry, current_length, n); |
3755 |
|
|
current_entry += n; |
3756 |
|
✗ |
++current_length; |
3757 |
|
|
} |
3758 |
|
|
} else { |
3759 |
|
✗ |
for (j=0; j < c->entries; ++j) { |
3760 |
|
✗ |
int present = c->sparse ? get_bits(f,1) : 1; |
3761 |
|
✗ |
if (present) { |
3762 |
|
✗ |
lengths[j] = get_bits(f, 5) + 1; |
3763 |
|
✗ |
++total; |
3764 |
|
✗ |
if (lengths[j] == 32) |
3765 |
|
✗ |
return error(f, VORBIS_invalid_setup); |
3766 |
|
|
} else { |
3767 |
|
✗ |
lengths[j] = NO_CODE; |
3768 |
|
|
} |
3769 |
|
|
} |
3770 |
|
|
} |
3771 |
|
|
|
3772 |
|
✗ |
if (c->sparse && total >= c->entries >> 2) { |
3773 |
|
|
// convert sparse items to non-sparse! |
3774 |
|
✗ |
if (c->entries > (int) f->setup_temp_memory_required) |
3775 |
|
✗ |
f->setup_temp_memory_required = c->entries; |
3776 |
|
|
|
3777 |
|
✗ |
c->codeword_lengths = (uint8 *) setup_malloc(f, c->entries); |
3778 |
|
✗ |
if (c->codeword_lengths == NULL) return error(f, VORBIS_outofmem); |
3779 |
|
✗ |
memcpy(c->codeword_lengths, lengths, c->entries); |
3780 |
|
✗ |
setup_temp_free(f, lengths, c->entries); // note this is only safe if there have been no intervening temp mallocs! |
3781 |
|
✗ |
lengths = c->codeword_lengths; |
3782 |
|
✗ |
c->sparse = 0; |
3783 |
|
|
} |
3784 |
|
|
|
3785 |
|
|
// compute the size of the sorted tables |
3786 |
|
✗ |
if (c->sparse) { |
3787 |
|
|
sorted_count = total; |
3788 |
|
|
} else { |
3789 |
|
|
sorted_count = 0; |
3790 |
|
|
#ifndef STB_VORBIS_NO_HUFFMAN_BINARY_SEARCH |
3791 |
|
✗ |
for (j=0; j < c->entries; ++j) |
3792 |
|
✗ |
if (lengths[j] > STB_VORBIS_FAST_HUFFMAN_LENGTH && lengths[j] != NO_CODE) |
3793 |
|
✗ |
++sorted_count; |
3794 |
|
|
#endif |
3795 |
|
|
} |
3796 |
|
|
|
3797 |
|
✗ |
c->sorted_entries = sorted_count; |
3798 |
|
|
values = NULL; |
3799 |
|
|
|
3800 |
|
|
CHECK(f); |
3801 |
|
✗ |
if (!c->sparse) { |
3802 |
|
✗ |
c->codewords = (uint32 *) setup_malloc(f, sizeof(c->codewords[0]) * c->entries); |
3803 |
|
✗ |
if (!c->codewords) return error(f, VORBIS_outofmem); |
3804 |
|
|
} else { |
3805 |
|
|
unsigned int size; |
3806 |
|
✗ |
if (c->sorted_entries) { |
3807 |
|
✗ |
c->codeword_lengths = (uint8 *) setup_malloc(f, c->sorted_entries); |
3808 |
|
✗ |
if (!c->codeword_lengths) return error(f, VORBIS_outofmem); |
3809 |
|
✗ |
c->codewords = (uint32 *) setup_temp_malloc(f, sizeof(*c->codewords) * c->sorted_entries); |
3810 |
|
✗ |
if (!c->codewords) return error(f, VORBIS_outofmem); |
3811 |
|
✗ |
values = (uint32 *) setup_temp_malloc(f, sizeof(*values) * c->sorted_entries); |
3812 |
|
✗ |
if (!values) return error(f, VORBIS_outofmem); |
3813 |
|
|
} |
3814 |
|
✗ |
size = c->entries + (sizeof(*c->codewords) + sizeof(*values)) * c->sorted_entries; |
3815 |
|
✗ |
if (size > f->setup_temp_memory_required) |
3816 |
|
✗ |
f->setup_temp_memory_required = size; |
3817 |
|
|
} |
3818 |
|
|
|
3819 |
|
✗ |
if (!compute_codewords(c, lengths, c->entries, values)) { |
3820 |
|
✗ |
if (c->sparse) setup_temp_free(f, values, 0); |
3821 |
|
✗ |
return error(f, VORBIS_invalid_setup); |
3822 |
|
|
} |
3823 |
|
|
|
3824 |
|
✗ |
if (c->sorted_entries) { |
3825 |
|
|
// allocate an extra slot for sentinels |
3826 |
|
✗ |
c->sorted_codewords = (uint32 *) setup_malloc(f, sizeof(*c->sorted_codewords) * (c->sorted_entries+1)); |
3827 |
|
✗ |
if (c->sorted_codewords == NULL) return error(f, VORBIS_outofmem); |
3828 |
|
|
// allocate an extra slot at the front so that c->sorted_values[-1] is defined |
3829 |
|
|
// so that we can catch that case without an extra if |
3830 |
|
✗ |
c->sorted_values = ( int *) setup_malloc(f, sizeof(*c->sorted_values ) * (c->sorted_entries+1)); |
3831 |
|
✗ |
if (c->sorted_values == NULL) return error(f, VORBIS_outofmem); |
3832 |
|
✗ |
++c->sorted_values; |
3833 |
|
✗ |
c->sorted_values[-1] = -1; |
3834 |
|
✗ |
compute_sorted_huffman(c, lengths, values); |
3835 |
|
|
} |
3836 |
|
|
|
3837 |
|
✗ |
if (c->sparse) { |
3838 |
|
✗ |
setup_temp_free(f, values, sizeof(*values)*c->sorted_entries); |
3839 |
|
✗ |
setup_temp_free(f, c->codewords, sizeof(*c->codewords)*c->sorted_entries); |
3840 |
|
✗ |
setup_temp_free(f, lengths, c->entries); |
3841 |
|
✗ |
c->codewords = NULL; |
3842 |
|
|
} |
3843 |
|
|
|
3844 |
|
✗ |
compute_accelerated_huffman(c); |
3845 |
|
|
|
3846 |
|
|
CHECK(f); |
3847 |
|
✗ |
c->lookup_type = get_bits(f, 4); |
3848 |
|
✗ |
if (c->lookup_type > 2) return error(f, VORBIS_invalid_setup); |
3849 |
|
✗ |
if (c->lookup_type > 0) { |
3850 |
|
|
uint16 *mults; |
3851 |
|
✗ |
c->minimum_value = float32_unpack(get_bits(f, 32)); |
3852 |
|
✗ |
c->delta_value = float32_unpack(get_bits(f, 32)); |
3853 |
|
✗ |
c->value_bits = get_bits(f, 4)+1; |
3854 |
|
✗ |
c->sequence_p = get_bits(f,1); |
3855 |
|
✗ |
if (c->lookup_type == 1) { |
3856 |
|
✗ |
int values = lookup1_values(c->entries, c->dimensions); |
3857 |
|
✗ |
if (values < 0) return error(f, VORBIS_invalid_setup); |
3858 |
|
✗ |
c->lookup_values = (uint32) values; |
3859 |
|
|
} else { |
3860 |
|
✗ |
c->lookup_values = c->entries * c->dimensions; |
3861 |
|
|
} |
3862 |
|
✗ |
if (c->lookup_values == 0) return error(f, VORBIS_invalid_setup); |
3863 |
|
✗ |
mults = (uint16 *) setup_temp_malloc(f, sizeof(mults[0]) * c->lookup_values); |
3864 |
|
✗ |
if (mults == NULL) return error(f, VORBIS_outofmem); |
3865 |
|
✗ |
for (j=0; j < (int) c->lookup_values; ++j) { |
3866 |
|
✗ |
int q = get_bits(f, c->value_bits); |
3867 |
|
✗ |
if (q == EOP) { setup_temp_free(f,mults,sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_invalid_setup); } |
3868 |
|
✗ |
mults[j] = q; |
3869 |
|
|
} |
3870 |
|
|
|
3871 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK |
3872 |
|
✗ |
if (c->lookup_type == 1) { |
3873 |
|
✗ |
int len, sparse = c->sparse; |
3874 |
|
|
float last=0; |
3875 |
|
|
// pre-expand the lookup1-style multiplicands, to avoid a divide in the inner loop |
3876 |
|
✗ |
if (sparse) { |
3877 |
|
✗ |
if (c->sorted_entries == 0) goto skip; |
3878 |
|
✗ |
c->multiplicands = (codetype *) setup_malloc(f, sizeof(c->multiplicands[0]) * c->sorted_entries * c->dimensions); |
3879 |
|
|
} else |
3880 |
|
✗ |
c->multiplicands = (codetype *) setup_malloc(f, sizeof(c->multiplicands[0]) * c->entries * c->dimensions); |
3881 |
|
✗ |
if (c->multiplicands == NULL) { setup_temp_free(f,mults,sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_outofmem); } |
3882 |
|
✗ |
len = sparse ? c->sorted_entries : c->entries; |
3883 |
|
✗ |
for (j=0; j < len; ++j) { |
3884 |
|
✗ |
unsigned int z = sparse ? c->sorted_values[j] : j; |
3885 |
|
|
unsigned int div=1; |
3886 |
|
✗ |
for (k=0; k < c->dimensions; ++k) { |
3887 |
|
✗ |
int off = (z / div) % c->lookup_values; |
3888 |
|
✗ |
float val = mults[off]*c->delta_value + c->minimum_value + last; |
3889 |
|
✗ |
c->multiplicands[j*c->dimensions + k] = val; |
3890 |
|
✗ |
if (c->sequence_p) |
3891 |
|
|
last = val; |
3892 |
|
✗ |
if (k+1 < c->dimensions) { |
3893 |
|
✗ |
if (div > UINT_MAX / (unsigned int) c->lookup_values) { |
3894 |
|
✗ |
setup_temp_free(f, mults,sizeof(mults[0])*c->lookup_values); |
3895 |
|
✗ |
return error(f, VORBIS_invalid_setup); |
3896 |
|
|
} |
3897 |
|
✗ |
div *= c->lookup_values; |
3898 |
|
|
} |
3899 |
|
|
} |
3900 |
|
|
} |
3901 |
|
✗ |
c->lookup_type = 2; |
3902 |
|
|
} |
3903 |
|
|
else |
3904 |
|
|
#endif |
3905 |
|
|
{ |
3906 |
|
|
float last=0; |
3907 |
|
|
CHECK(f); |
3908 |
|
✗ |
c->multiplicands = (codetype *) setup_malloc(f, sizeof(c->multiplicands[0]) * c->lookup_values); |
3909 |
|
✗ |
if (c->multiplicands == NULL) { setup_temp_free(f, mults,sizeof(mults[0])*c->lookup_values); return error(f, VORBIS_outofmem); } |
3910 |
|
✗ |
for (j=0; j < (int) c->lookup_values; ++j) { |
3911 |
|
✗ |
float val = mults[j] * c->delta_value + c->minimum_value + last; |
3912 |
|
✗ |
c->multiplicands[j] = val; |
3913 |
|
✗ |
if (c->sequence_p) |
3914 |
|
|
last = val; |
3915 |
|
|
} |
3916 |
|
|
} |
3917 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_CODEBOOK |
3918 |
|
✗ |
skip:; |
3919 |
|
|
#endif |
3920 |
|
✗ |
setup_temp_free(f, mults, sizeof(mults[0])*c->lookup_values); |
3921 |
|
|
|
3922 |
|
|
CHECK(f); |
3923 |
|
|
} |
3924 |
|
|
CHECK(f); |
3925 |
|
|
} |
3926 |
|
|
|
3927 |
|
|
// time domain transfers (notused) |
3928 |
|
|
|
3929 |
|
✗ |
x = get_bits(f, 6) + 1; |
3930 |
|
✗ |
for (i=0; i < x; ++i) { |
3931 |
|
✗ |
uint32 z = get_bits(f, 16); |
3932 |
|
✗ |
if (z != 0) return error(f, VORBIS_invalid_setup); |
3933 |
|
|
} |
3934 |
|
|
|
3935 |
|
|
// Floors |
3936 |
|
✗ |
f->floor_count = get_bits(f, 6)+1; |
3937 |
|
✗ |
f->floor_config = (Floor *) setup_malloc(f, f->floor_count * sizeof(*f->floor_config)); |
3938 |
|
✗ |
if (f->floor_config == NULL) return error(f, VORBIS_outofmem); |
3939 |
|
✗ |
for (i=0; i < f->floor_count; ++i) { |
3940 |
|
✗ |
f->floor_types[i] = get_bits(f, 16); |
3941 |
|
✗ |
if (f->floor_types[i] > 1) return error(f, VORBIS_invalid_setup); |
3942 |
|
✗ |
if (f->floor_types[i] == 0) { |
3943 |
|
✗ |
Floor0 *g = &f->floor_config[i].floor0; |
3944 |
|
✗ |
g->order = get_bits(f,8); |
3945 |
|
✗ |
g->rate = get_bits(f,16); |
3946 |
|
✗ |
g->bark_map_size = get_bits(f,16); |
3947 |
|
✗ |
g->amplitude_bits = get_bits(f,6); |
3948 |
|
✗ |
g->amplitude_offset = get_bits(f,8); |
3949 |
|
✗ |
g->number_of_books = get_bits(f,4) + 1; |
3950 |
|
✗ |
for (j=0; j < g->number_of_books; ++j) |
3951 |
|
✗ |
g->book_list[j] = get_bits(f,8); |
3952 |
|
✗ |
return error(f, VORBIS_feature_not_supported); |
3953 |
|
|
} else { |
3954 |
|
|
stbv__floor_ordering p[31*8+2]; |
3955 |
|
✗ |
Floor1 *g = &f->floor_config[i].floor1; |
3956 |
|
|
int max_class = -1; |
3957 |
|
✗ |
g->partitions = get_bits(f, 5); |
3958 |
|
✗ |
for (j=0; j < g->partitions; ++j) { |
3959 |
|
✗ |
g->partition_class_list[j] = get_bits(f, 4); |
3960 |
|
✗ |
if (g->partition_class_list[j] > max_class) |
3961 |
|
|
max_class = g->partition_class_list[j]; |
3962 |
|
|
} |
3963 |
|
✗ |
for (j=0; j <= max_class; ++j) { |
3964 |
|
✗ |
g->class_dimensions[j] = get_bits(f, 3)+1; |
3965 |
|
✗ |
g->class_subclasses[j] = get_bits(f, 2); |
3966 |
|
✗ |
if (g->class_subclasses[j]) { |
3967 |
|
✗ |
g->class_masterbooks[j] = get_bits(f, 8); |
3968 |
|
✗ |
if (g->class_masterbooks[j] >= f->codebook_count) return error(f, VORBIS_invalid_setup); |
3969 |
|
|
} |
3970 |
|
✗ |
for (k=0; k < 1 << g->class_subclasses[j]; ++k) { |
3971 |
|
✗ |
g->subclass_books[j][k] = (int16)get_bits(f,8)-1; |
3972 |
|
✗ |
if (g->subclass_books[j][k] >= f->codebook_count) return error(f, VORBIS_invalid_setup); |
3973 |
|
|
} |
3974 |
|
|
} |
3975 |
|
✗ |
g->floor1_multiplier = get_bits(f,2)+1; |
3976 |
|
✗ |
g->rangebits = get_bits(f,4); |
3977 |
|
✗ |
g->Xlist[0] = 0; |
3978 |
|
✗ |
g->Xlist[1] = 1 << g->rangebits; |
3979 |
|
✗ |
g->values = 2; |
3980 |
|
✗ |
for (j=0; j < g->partitions; ++j) { |
3981 |
|
✗ |
int c = g->partition_class_list[j]; |
3982 |
|
✗ |
for (k=0; k < g->class_dimensions[c]; ++k) { |
3983 |
|
✗ |
g->Xlist[g->values] = get_bits(f, g->rangebits); |
3984 |
|
✗ |
++g->values; |
3985 |
|
|
} |
3986 |
|
|
} |
3987 |
|
|
// precompute the sorting |
3988 |
|
✗ |
for (j=0; j < g->values; ++j) { |
3989 |
|
✗ |
p[j].x = g->Xlist[j]; |
3990 |
|
✗ |
p[j].id = j; |
3991 |
|
|
} |
3992 |
|
✗ |
qsort(p, g->values, sizeof(p[0]), point_compare); |
3993 |
|
✗ |
for (j=0; j < g->values-1; ++j) |
3994 |
|
✗ |
if (p[j].x == p[j+1].x) |
3995 |
|
✗ |
return error(f, VORBIS_invalid_setup); |
3996 |
|
✗ |
for (j=0; j < g->values; ++j) |
3997 |
|
✗ |
g->sorted_order[j] = (uint8) p[j].id; |
3998 |
|
|
// precompute the neighbors |
3999 |
|
✗ |
for (j=2; j < g->values; ++j) { |
4000 |
|
✗ |
int low = 0,hi = 0; |
4001 |
|
✗ |
neighbors(g->Xlist, j, &low,&hi); |
4002 |
|
✗ |
g->neighbors[j][0] = low; |
4003 |
|
✗ |
g->neighbors[j][1] = hi; |
4004 |
|
|
} |
4005 |
|
|
|
4006 |
|
|
if (g->values > longest_floorlist) |
4007 |
|
|
longest_floorlist = g->values; |
4008 |
|
|
} |
4009 |
|
|
} |
4010 |
|
|
|
4011 |
|
|
// Residue |
4012 |
|
✗ |
f->residue_count = get_bits(f, 6)+1; |
4013 |
|
✗ |
f->residue_config = (Residue *) setup_malloc(f, f->residue_count * sizeof(f->residue_config[0])); |
4014 |
|
✗ |
if (f->residue_config == NULL) return error(f, VORBIS_outofmem); |
4015 |
|
✗ |
memset(f->residue_config, 0, f->residue_count * sizeof(f->residue_config[0])); |
4016 |
|
✗ |
for (i=0; i < f->residue_count; ++i) { |
4017 |
|
|
uint8 residue_cascade[64]; |
4018 |
|
✗ |
Residue *r = f->residue_config+i; |
4019 |
|
✗ |
f->residue_types[i] = get_bits(f, 16); |
4020 |
|
✗ |
if (f->residue_types[i] > 2) return error(f, VORBIS_invalid_setup); |
4021 |
|
✗ |
r->begin = get_bits(f, 24); |
4022 |
|
✗ |
r->end = get_bits(f, 24); |
4023 |
|
✗ |
if (r->end < r->begin) return error(f, VORBIS_invalid_setup); |
4024 |
|
✗ |
r->part_size = get_bits(f,24)+1; |
4025 |
|
✗ |
r->classifications = get_bits(f,6)+1; |
4026 |
|
✗ |
r->classbook = get_bits(f,8); |
4027 |
|
✗ |
if (r->classbook >= f->codebook_count) return error(f, VORBIS_invalid_setup); |
4028 |
|
✗ |
for (j=0; j < r->classifications; ++j) { |
4029 |
|
|
uint8 high_bits=0; |
4030 |
|
✗ |
uint8 low_bits=get_bits(f,3); |
4031 |
|
✗ |
if (get_bits(f,1)) |
4032 |
|
✗ |
high_bits = get_bits(f,5); |
4033 |
|
✗ |
residue_cascade[j] = high_bits*8 + low_bits; |
4034 |
|
|
} |
4035 |
|
✗ |
r->residue_books = (short (*)[8]) setup_malloc(f, sizeof(r->residue_books[0]) * r->classifications); |
4036 |
|
✗ |
if (r->residue_books == NULL) return error(f, VORBIS_outofmem); |
4037 |
|
✗ |
for (j=0; j < r->classifications; ++j) { |
4038 |
|
✗ |
for (k=0; k < 8; ++k) { |
4039 |
|
✗ |
if (residue_cascade[j] & (1 << k)) { |
4040 |
|
✗ |
r->residue_books[j][k] = get_bits(f, 8); |
4041 |
|
✗ |
if (r->residue_books[j][k] >= f->codebook_count) return error(f, VORBIS_invalid_setup); |
4042 |
|
|
} else { |
4043 |
|
✗ |
r->residue_books[j][k] = -1; |
4044 |
|
|
} |
4045 |
|
|
} |
4046 |
|
|
} |
4047 |
|
|
// precompute the classifications[] array to avoid inner-loop mod/divide |
4048 |
|
|
// call it 'classdata' since we already have r->classifications |
4049 |
|
✗ |
r->classdata = (uint8 **) setup_malloc(f, sizeof(*r->classdata) * f->codebooks[r->classbook].entries); |
4050 |
|
✗ |
if (!r->classdata) return error(f, VORBIS_outofmem); |
4051 |
|
✗ |
memset(r->classdata, 0, sizeof(*r->classdata) * f->codebooks[r->classbook].entries); |
4052 |
|
✗ |
for (j=0; j < f->codebooks[r->classbook].entries; ++j) { |
4053 |
|
✗ |
int classwords = f->codebooks[r->classbook].dimensions; |
4054 |
|
|
int temp = j; |
4055 |
|
✗ |
r->classdata[j] = (uint8 *) setup_malloc(f, sizeof(r->classdata[j][0]) * classwords); |
4056 |
|
✗ |
if (r->classdata[j] == NULL) return error(f, VORBIS_outofmem); |
4057 |
|
✗ |
for (k=classwords-1; k >= 0; --k) { |
4058 |
|
✗ |
r->classdata[j][k] = temp % r->classifications; |
4059 |
|
✗ |
temp /= r->classifications; |
4060 |
|
|
} |
4061 |
|
|
} |
4062 |
|
|
} |
4063 |
|
|
|
4064 |
|
✗ |
f->mapping_count = get_bits(f,6)+1; |
4065 |
|
✗ |
f->mapping = (Mapping *) setup_malloc(f, f->mapping_count * sizeof(*f->mapping)); |
4066 |
|
✗ |
if (f->mapping == NULL) return error(f, VORBIS_outofmem); |
4067 |
|
✗ |
memset(f->mapping, 0, f->mapping_count * sizeof(*f->mapping)); |
4068 |
|
✗ |
for (i=0; i < f->mapping_count; ++i) { |
4069 |
|
✗ |
Mapping *m = f->mapping + i; |
4070 |
|
✗ |
int mapping_type = get_bits(f,16); |
4071 |
|
✗ |
if (mapping_type != 0) return error(f, VORBIS_invalid_setup); |
4072 |
|
✗ |
m->chan = (MappingChannel *) setup_malloc(f, f->channels * sizeof(*m->chan)); |
4073 |
|
✗ |
if (m->chan == NULL) return error(f, VORBIS_outofmem); |
4074 |
|
✗ |
if (get_bits(f,1)) |
4075 |
|
✗ |
m->submaps = get_bits(f,4)+1; |
4076 |
|
|
else |
4077 |
|
✗ |
m->submaps = 1; |
4078 |
|
|
if (m->submaps > max_submaps) |
4079 |
|
|
max_submaps = m->submaps; |
4080 |
|
✗ |
if (get_bits(f,1)) { |
4081 |
|
✗ |
m->coupling_steps = get_bits(f,8)+1; |
4082 |
|
✗ |
if (m->coupling_steps > f->channels) return error(f, VORBIS_invalid_setup); |
4083 |
|
✗ |
for (k=0; k < m->coupling_steps; ++k) { |
4084 |
|
✗ |
m->chan[k].magnitude = get_bits(f, ilog(f->channels-1)); |
4085 |
|
✗ |
m->chan[k].angle = get_bits(f, ilog(f->channels-1)); |
4086 |
|
✗ |
if (m->chan[k].magnitude >= f->channels) return error(f, VORBIS_invalid_setup); |
4087 |
|
✗ |
if (m->chan[k].angle >= f->channels) return error(f, VORBIS_invalid_setup); |
4088 |
|
✗ |
if (m->chan[k].magnitude == m->chan[k].angle) return error(f, VORBIS_invalid_setup); |
4089 |
|
|
} |
4090 |
|
|
} else |
4091 |
|
✗ |
m->coupling_steps = 0; |
4092 |
|
|
|
4093 |
|
|
// reserved field |
4094 |
|
✗ |
if (get_bits(f,2)) return error(f, VORBIS_invalid_setup); |
4095 |
|
✗ |
if (m->submaps > 1) { |
4096 |
|
✗ |
for (j=0; j < f->channels; ++j) { |
4097 |
|
✗ |
m->chan[j].mux = get_bits(f, 4); |
4098 |
|
✗ |
if (m->chan[j].mux >= m->submaps) return error(f, VORBIS_invalid_setup); |
4099 |
|
|
} |
4100 |
|
|
} else |
4101 |
|
|
// @SPECIFICATION: this case is missing from the spec |
4102 |
|
✗ |
for (j=0; j < f->channels; ++j) |
4103 |
|
✗ |
m->chan[j].mux = 0; |
4104 |
|
|
|
4105 |
|
✗ |
for (j=0; j < m->submaps; ++j) { |
4106 |
|
✗ |
get_bits(f,8); // discard |
4107 |
|
✗ |
m->submap_floor[j] = get_bits(f,8); |
4108 |
|
✗ |
m->submap_residue[j] = get_bits(f,8); |
4109 |
|
✗ |
if (m->submap_floor[j] >= f->floor_count) return error(f, VORBIS_invalid_setup); |
4110 |
|
✗ |
if (m->submap_residue[j] >= f->residue_count) return error(f, VORBIS_invalid_setup); |
4111 |
|
|
} |
4112 |
|
|
} |
4113 |
|
|
|
4114 |
|
|
// Modes |
4115 |
|
✗ |
f->mode_count = get_bits(f, 6)+1; |
4116 |
|
✗ |
for (i=0; i < f->mode_count; ++i) { |
4117 |
|
✗ |
Mode *m = f->mode_config+i; |
4118 |
|
✗ |
m->blockflag = get_bits(f,1); |
4119 |
|
✗ |
m->windowtype = get_bits(f,16); |
4120 |
|
✗ |
m->transformtype = get_bits(f,16); |
4121 |
|
✗ |
m->mapping = get_bits(f,8); |
4122 |
|
✗ |
if (m->windowtype != 0) return error(f, VORBIS_invalid_setup); |
4123 |
|
✗ |
if (m->transformtype != 0) return error(f, VORBIS_invalid_setup); |
4124 |
|
✗ |
if (m->mapping >= f->mapping_count) return error(f, VORBIS_invalid_setup); |
4125 |
|
|
} |
4126 |
|
|
|
4127 |
|
|
flush_packet(f); |
4128 |
|
|
|
4129 |
|
✗ |
f->previous_length = 0; |
4130 |
|
|
|
4131 |
|
✗ |
for (i=0; i < f->channels; ++i) { |
4132 |
|
✗ |
f->channel_buffers[i] = (float *) setup_malloc(f, sizeof(float) * f->blocksize_1); |
4133 |
|
✗ |
f->previous_window[i] = (float *) setup_malloc(f, sizeof(float) * f->blocksize_1/2); |
4134 |
|
✗ |
f->finalY[i] = (int16 *) setup_malloc(f, sizeof(int16) * longest_floorlist); |
4135 |
|
✗ |
if (f->channel_buffers[i] == NULL || f->previous_window[i] == NULL || f->finalY[i] == NULL) return error(f, VORBIS_outofmem); |
4136 |
|
|
memset(f->channel_buffers[i], 0, sizeof(float) * f->blocksize_1); |
4137 |
|
|
#ifdef STB_VORBIS_NO_DEFER_FLOOR |
4138 |
|
|
f->floor_buffers[i] = (float *) setup_malloc(f, sizeof(float) * f->blocksize_1/2); |
4139 |
|
|
if (f->floor_buffers[i] == NULL) return error(f, VORBIS_outofmem); |
4140 |
|
|
#endif |
4141 |
|
|
} |
4142 |
|
|
|
4143 |
|
✗ |
if (!init_blocksize(f, 0, f->blocksize_0)) return FALSE; |
4144 |
|
✗ |
if (!init_blocksize(f, 1, f->blocksize_1)) return FALSE; |
4145 |
|
✗ |
f->blocksize[0] = f->blocksize_0; |
4146 |
|
✗ |
f->blocksize[1] = f->blocksize_1; |
4147 |
|
|
|
4148 |
|
|
#ifdef STB_VORBIS_DIVIDE_TABLE |
4149 |
|
|
if (integer_divide_table[1][1]==0) |
4150 |
|
|
for (i=0; i < DIVTAB_NUMER; ++i) |
4151 |
|
|
for (j=1; j < DIVTAB_DENOM; ++j) |
4152 |
|
|
integer_divide_table[i][j] = i / j; |
4153 |
|
|
#endif |
4154 |
|
|
|
4155 |
|
|
// compute how much temporary memory is needed |
4156 |
|
|
|
4157 |
|
|
// 1. |
4158 |
|
|
{ |
4159 |
|
✗ |
uint32 imdct_mem = (f->blocksize_1 * sizeof(float) >> 1); |
4160 |
|
|
uint32 classify_mem; |
4161 |
|
|
int i,max_part_read=0; |
4162 |
|
✗ |
for (i=0; i < f->residue_count; ++i) { |
4163 |
|
✗ |
Residue *r = f->residue_config + i; |
4164 |
|
✗ |
unsigned int actual_size = f->blocksize_1 / 2; |
4165 |
|
✗ |
unsigned int limit_r_begin = r->begin < actual_size ? r->begin : actual_size; |
4166 |
|
✗ |
unsigned int limit_r_end = r->end < actual_size ? r->end : actual_size; |
4167 |
|
✗ |
int n_read = limit_r_end - limit_r_begin; |
4168 |
|
✗ |
int part_read = n_read / r->part_size; |
4169 |
|
|
if (part_read > max_part_read) |
4170 |
|
|
max_part_read = part_read; |
4171 |
|
|
} |
4172 |
|
|
#ifndef STB_VORBIS_DIVIDES_IN_RESIDUE |
4173 |
|
✗ |
classify_mem = f->channels * (sizeof(void*) + max_part_read * sizeof(uint8 *)); |
4174 |
|
|
#else |
4175 |
|
|
classify_mem = f->channels * (sizeof(void*) + max_part_read * sizeof(int *)); |
4176 |
|
|
#endif |
4177 |
|
|
|
4178 |
|
|
// maximum reasonable partition size is f->blocksize_1 |
4179 |
|
|
|
4180 |
|
✗ |
f->temp_memory_required = classify_mem; |
4181 |
|
✗ |
if (imdct_mem > f->temp_memory_required) |
4182 |
|
✗ |
f->temp_memory_required = imdct_mem; |
4183 |
|
|
} |
4184 |
|
|
|
4185 |
|
|
|
4186 |
|
✗ |
if (f->alloc.alloc_buffer) { |
4187 |
|
|
assert(f->temp_offset == f->alloc.alloc_buffer_length_in_bytes); |
4188 |
|
|
// check if there's enough temp memory so we don't error later |
4189 |
|
✗ |
if (f->setup_offset + sizeof(*f) + f->temp_memory_required > (unsigned) f->temp_offset) |
4190 |
|
✗ |
return error(f, VORBIS_outofmem); |
4191 |
|
|
} |
4192 |
|
|
|
4193 |
|
|
// @TODO: stb_vorbis_seek_start expects first_audio_page_offset to point to a page |
4194 |
|
|
// without PAGEFLAG_continued_packet, so this either points to the first page, or |
4195 |
|
|
// the page after the end of the headers. It might be cleaner to point to a page |
4196 |
|
|
// in the middle of the headers, when that's the page where the first audio packet |
4197 |
|
|
// starts, but we'd have to also correctly skip the end of any continued packet in |
4198 |
|
|
// stb_vorbis_seek_start. |
4199 |
|
✗ |
if (f->next_seg == -1) { |
4200 |
|
✗ |
f->first_audio_page_offset = stb_vorbis_get_file_offset(f); |
4201 |
|
|
} else { |
4202 |
|
✗ |
f->first_audio_page_offset = 0; |
4203 |
|
|
} |
4204 |
|
|
|
4205 |
|
|
return TRUE; |
4206 |
|
|
} |
4207 |
|
|
|
4208 |
|
✗ |
static void vorbis_deinit(stb_vorbis *p) |
4209 |
|
|
{ |
4210 |
|
|
int i,j; |
4211 |
|
|
|
4212 |
|
✗ |
setup_free(p, p->vendor); |
4213 |
|
✗ |
for (i=0; i < p->comment_list_length; ++i) { |
4214 |
|
✗ |
setup_free(p, p->comment_list[i]); |
4215 |
|
|
} |
4216 |
|
✗ |
setup_free(p, p->comment_list); |
4217 |
|
|
|
4218 |
|
✗ |
if (p->residue_config) { |
4219 |
|
✗ |
for (i=0; i < p->residue_count; ++i) { |
4220 |
|
✗ |
Residue *r = p->residue_config+i; |
4221 |
|
✗ |
if (r->classdata) { |
4222 |
|
✗ |
for (j=0; j < p->codebooks[r->classbook].entries; ++j) |
4223 |
|
✗ |
setup_free(p, r->classdata[j]); |
4224 |
|
✗ |
setup_free(p, r->classdata); |
4225 |
|
|
} |
4226 |
|
✗ |
setup_free(p, r->residue_books); |
4227 |
|
|
} |
4228 |
|
|
} |
4229 |
|
|
|
4230 |
|
✗ |
if (p->codebooks) { |
4231 |
|
|
CHECK(p); |
4232 |
|
✗ |
for (i=0; i < p->codebook_count; ++i) { |
4233 |
|
✗ |
Codebook *c = p->codebooks + i; |
4234 |
|
✗ |
setup_free(p, c->codeword_lengths); |
4235 |
|
✗ |
setup_free(p, c->multiplicands); |
4236 |
|
✗ |
setup_free(p, c->codewords); |
4237 |
|
✗ |
setup_free(p, c->sorted_codewords); |
4238 |
|
|
// c->sorted_values[-1] is the first entry in the array |
4239 |
|
✗ |
setup_free(p, c->sorted_values ? c->sorted_values-1 : NULL); |
4240 |
|
|
} |
4241 |
|
✗ |
setup_free(p, p->codebooks); |
4242 |
|
|
} |
4243 |
|
✗ |
setup_free(p, p->floor_config); |
4244 |
|
✗ |
setup_free(p, p->residue_config); |
4245 |
|
✗ |
if (p->mapping) { |
4246 |
|
✗ |
for (i=0; i < p->mapping_count; ++i) |
4247 |
|
✗ |
setup_free(p, p->mapping[i].chan); |
4248 |
|
✗ |
setup_free(p, p->mapping); |
4249 |
|
|
} |
4250 |
|
|
CHECK(p); |
4251 |
|
✗ |
for (i=0; i < p->channels && i < STB_VORBIS_MAX_CHANNELS; ++i) { |
4252 |
|
✗ |
setup_free(p, p->channel_buffers[i]); |
4253 |
|
✗ |
setup_free(p, p->previous_window[i]); |
4254 |
|
|
#ifdef STB_VORBIS_NO_DEFER_FLOOR |
4255 |
|
|
setup_free(p, p->floor_buffers[i]); |
4256 |
|
|
#endif |
4257 |
|
✗ |
setup_free(p, p->finalY[i]); |
4258 |
|
|
} |
4259 |
|
✗ |
for (i=0; i < 2; ++i) { |
4260 |
|
✗ |
setup_free(p, p->A[i]); |
4261 |
|
✗ |
setup_free(p, p->B[i]); |
4262 |
|
✗ |
setup_free(p, p->C[i]); |
4263 |
|
✗ |
setup_free(p, p->window[i]); |
4264 |
|
✗ |
setup_free(p, p->bit_reverse[i]); |
4265 |
|
|
} |
4266 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
4267 |
|
✗ |
if (p->close_on_free) fclose(p->f); |
4268 |
|
|
#endif |
4269 |
|
|
} |
4270 |
|
|
|
4271 |
|
✗ |
void stb_vorbis_close(stb_vorbis *p) |
4272 |
|
|
{ |
4273 |
|
✗ |
if (p == NULL) return; |
4274 |
|
✗ |
vorbis_deinit(p); |
4275 |
|
|
setup_free(p,p); |
4276 |
|
|
} |
4277 |
|
|
|
4278 |
|
✗ |
static void vorbis_init(stb_vorbis *p, const stb_vorbis_alloc *z) |
4279 |
|
|
{ |
4280 |
|
|
memset(p, 0, sizeof(*p)); // NULL out all malloc'd pointers to start |
4281 |
|
✗ |
if (z) { |
4282 |
|
✗ |
p->alloc = *z; |
4283 |
|
✗ |
p->alloc.alloc_buffer_length_in_bytes &= ~7; |
4284 |
|
✗ |
p->temp_offset = p->alloc.alloc_buffer_length_in_bytes; |
4285 |
|
|
} |
4286 |
|
|
p->eof = 0; |
4287 |
|
|
p->error = VORBIS__no_error; |
4288 |
|
|
p->stream = NULL; |
4289 |
|
|
p->codebooks = NULL; |
4290 |
|
✗ |
p->page_crc_tests = -1; |
4291 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
4292 |
|
|
p->close_on_free = FALSE; |
4293 |
|
|
p->f = NULL; |
4294 |
|
|
#endif |
4295 |
|
|
} |
4296 |
|
|
|
4297 |
|
✗ |
int stb_vorbis_get_sample_offset(stb_vorbis *f) |
4298 |
|
|
{ |
4299 |
|
✗ |
if (f->current_loc_valid) |
4300 |
|
✗ |
return f->current_loc; |
4301 |
|
|
else |
4302 |
|
|
return -1; |
4303 |
|
|
} |
4304 |
|
|
|
4305 |
|
✗ |
stb_vorbis_info stb_vorbis_get_info(stb_vorbis *f) |
4306 |
|
|
{ |
4307 |
|
|
stb_vorbis_info d; |
4308 |
|
✗ |
d.channels = f->channels; |
4309 |
|
✗ |
d.sample_rate = f->sample_rate; |
4310 |
|
✗ |
d.setup_memory_required = f->setup_memory_required; |
4311 |
|
✗ |
d.setup_temp_memory_required = f->setup_temp_memory_required; |
4312 |
|
✗ |
d.temp_memory_required = f->temp_memory_required; |
4313 |
|
✗ |
d.max_frame_size = f->blocksize_1 >> 1; |
4314 |
|
✗ |
return d; |
4315 |
|
|
} |
4316 |
|
|
|
4317 |
|
✗ |
stb_vorbis_comment stb_vorbis_get_comment(stb_vorbis *f) |
4318 |
|
|
{ |
4319 |
|
|
stb_vorbis_comment d; |
4320 |
|
✗ |
d.vendor = f->vendor; |
4321 |
|
✗ |
d.comment_list_length = f->comment_list_length; |
4322 |
|
✗ |
d.comment_list = f->comment_list; |
4323 |
|
✗ |
return d; |
4324 |
|
|
} |
4325 |
|
|
|
4326 |
|
✗ |
int stb_vorbis_get_error(stb_vorbis *f) |
4327 |
|
|
{ |
4328 |
|
✗ |
int e = f->error; |
4329 |
|
✗ |
f->error = VORBIS__no_error; |
4330 |
|
✗ |
return e; |
4331 |
|
|
} |
4332 |
|
|
|
4333 |
|
|
static stb_vorbis * vorbis_alloc(stb_vorbis *f) |
4334 |
|
|
{ |
4335 |
|
✗ |
stb_vorbis *p = (stb_vorbis *) setup_malloc(f, sizeof(*p)); |
4336 |
|
|
return p; |
4337 |
|
|
} |
4338 |
|
|
|
4339 |
|
|
#ifndef STB_VORBIS_NO_PUSHDATA_API |
4340 |
|
|
|
4341 |
|
✗ |
void stb_vorbis_flush_pushdata(stb_vorbis *f) |
4342 |
|
|
{ |
4343 |
|
✗ |
f->previous_length = 0; |
4344 |
|
✗ |
f->page_crc_tests = 0; |
4345 |
|
✗ |
f->discard_samples_deferred = 0; |
4346 |
|
✗ |
f->current_loc_valid = FALSE; |
4347 |
|
✗ |
f->first_decode = FALSE; |
4348 |
|
✗ |
f->samples_output = 0; |
4349 |
|
✗ |
f->channel_buffer_start = 0; |
4350 |
|
✗ |
f->channel_buffer_end = 0; |
4351 |
|
|
} |
4352 |
|
|
|
4353 |
|
✗ |
static int vorbis_search_for_page_pushdata(vorb *f, uint8 *data, int data_len) |
4354 |
|
|
{ |
4355 |
|
|
int i,n; |
4356 |
|
✗ |
for (i=0; i < f->page_crc_tests; ++i) |
4357 |
|
✗ |
f->scan[i].bytes_done = 0; |
4358 |
|
|
|
4359 |
|
|
// if we have room for more scans, search for them first, because |
4360 |
|
|
// they may cause us to stop early if their header is incomplete |
4361 |
|
✗ |
if (f->page_crc_tests < STB_VORBIS_PUSHDATA_CRC_COUNT) { |
4362 |
|
✗ |
if (data_len < 4) return 0; |
4363 |
|
✗ |
data_len -= 3; // need to look for 4-byte sequence, so don't miss |
4364 |
|
|
// one that straddles a boundary |
4365 |
|
✗ |
for (i=0; i < data_len; ++i) { |
4366 |
|
✗ |
if (data[i] == 0x4f) { |
4367 |
|
✗ |
if (0==memcmp(data+i, ogg_page_header, 4)) { |
4368 |
|
|
int j,len; |
4369 |
|
|
uint32 crc; |
4370 |
|
|
// make sure we have the whole page header |
4371 |
|
✗ |
if (i+26 >= data_len || i+27+data[i+26] >= data_len) { |
4372 |
|
|
// only read up to this page start, so hopefully we'll |
4373 |
|
|
// have the whole page header start next time |
4374 |
|
|
data_len = i; |
4375 |
|
|
break; |
4376 |
|
|
} |
4377 |
|
|
// ok, we have it all; compute the length of the page |
4378 |
|
✗ |
len = 27 + data[i+26]; |
4379 |
|
✗ |
for (j=0; j < data[i+26]; ++j) |
4380 |
|
✗ |
len += data[i+27+j]; |
4381 |
|
|
// scan everything up to the embedded crc (which we must 0) |
4382 |
|
|
crc = 0; |
4383 |
|
✗ |
for (j=0; j < 22; ++j) |
4384 |
|
✗ |
crc = crc32_update(crc, data[i+j]); |
4385 |
|
|
// now process 4 0-bytes |
4386 |
|
✗ |
for ( ; j < 26; ++j) |
4387 |
|
|
crc = crc32_update(crc, 0); |
4388 |
|
|
// len is the total number of bytes we need to scan |
4389 |
|
✗ |
n = f->page_crc_tests++; |
4390 |
|
✗ |
f->scan[n].bytes_left = len-j; |
4391 |
|
✗ |
f->scan[n].crc_so_far = crc; |
4392 |
|
✗ |
f->scan[n].goal_crc = data[i+22] + (data[i+23] << 8) + (data[i+24]<<16) + (data[i+25]<<24); |
4393 |
|
|
// if the last frame on a page is continued to the next, then |
4394 |
|
|
// we can't recover the sample_loc immediately |
4395 |
|
✗ |
if (data[i+27+data[i+26]-1] == 255) |
4396 |
|
✗ |
f->scan[n].sample_loc = ~0; |
4397 |
|
|
else |
4398 |
|
✗ |
f->scan[n].sample_loc = data[i+6] + (data[i+7] << 8) + (data[i+ 8]<<16) + (data[i+ 9]<<24); |
4399 |
|
✗ |
f->scan[n].bytes_done = i+j; |
4400 |
|
✗ |
if (f->page_crc_tests == STB_VORBIS_PUSHDATA_CRC_COUNT) |
4401 |
|
|
break; |
4402 |
|
|
// keep going if we still have room for more |
4403 |
|
|
} |
4404 |
|
|
} |
4405 |
|
|
} |
4406 |
|
|
} |
4407 |
|
|
|
4408 |
|
✗ |
for (i=0; i < f->page_crc_tests;) { |
4409 |
|
|
uint32 crc; |
4410 |
|
|
int j; |
4411 |
|
✗ |
int n = f->scan[i].bytes_done; |
4412 |
|
✗ |
int m = f->scan[i].bytes_left; |
4413 |
|
✗ |
if (m > data_len - n) m = data_len - n; |
4414 |
|
|
// m is the bytes to scan in the current chunk |
4415 |
|
✗ |
crc = f->scan[i].crc_so_far; |
4416 |
|
✗ |
for (j=0; j < m; ++j) |
4417 |
|
✗ |
crc = crc32_update(crc, data[n+j]); |
4418 |
|
✗ |
f->scan[i].bytes_left -= m; |
4419 |
|
✗ |
f->scan[i].crc_so_far = crc; |
4420 |
|
✗ |
if (f->scan[i].bytes_left == 0) { |
4421 |
|
|
// does it match? |
4422 |
|
✗ |
if (f->scan[i].crc_so_far == f->scan[i].goal_crc) { |
4423 |
|
|
// Houston, we have page |
4424 |
|
✗ |
data_len = n+m; // consumption amount is wherever that scan ended |
4425 |
|
✗ |
f->page_crc_tests = -1; // drop out of page scan mode |
4426 |
|
✗ |
f->previous_length = 0; // decode-but-don't-output one frame |
4427 |
|
✗ |
f->next_seg = -1; // start a new page |
4428 |
|
✗ |
f->current_loc = f->scan[i].sample_loc; // set the current sample location |
4429 |
|
|
// to the amount we'd have decoded had we decoded this page |
4430 |
|
✗ |
f->current_loc_valid = f->current_loc != ~0U; |
4431 |
|
✗ |
return data_len; |
4432 |
|
|
} |
4433 |
|
|
// delete entry |
4434 |
|
✗ |
f->scan[i] = f->scan[--f->page_crc_tests]; |
4435 |
|
|
} else { |
4436 |
|
✗ |
++i; |
4437 |
|
|
} |
4438 |
|
|
} |
4439 |
|
|
|
4440 |
|
|
return data_len; |
4441 |
|
|
} |
4442 |
|
|
|
4443 |
|
|
// return value: number of bytes we used |
4444 |
|
✗ |
int stb_vorbis_decode_frame_pushdata( |
4445 |
|
|
stb_vorbis *f, // the file we're decoding |
4446 |
|
|
const uint8 *data, int data_len, // the memory available for decoding |
4447 |
|
|
int *channels, // place to write number of float * buffers |
4448 |
|
|
float ***output, // place to write float ** array of float * buffers |
4449 |
|
|
int *samples // place to write number of output samples |
4450 |
|
|
) |
4451 |
|
|
{ |
4452 |
|
|
int i; |
4453 |
|
|
int len,right,left; |
4454 |
|
|
|
4455 |
|
✗ |
if (!IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing); |
4456 |
|
|
|
4457 |
|
✗ |
if (f->page_crc_tests >= 0) { |
4458 |
|
✗ |
*samples = 0; |
4459 |
|
✗ |
return vorbis_search_for_page_pushdata(f, (uint8 *) data, data_len); |
4460 |
|
|
} |
4461 |
|
|
|
4462 |
|
✗ |
f->stream = (uint8 *) data; |
4463 |
|
✗ |
f->stream_end = (uint8 *) data + data_len; |
4464 |
|
✗ |
f->error = VORBIS__no_error; |
4465 |
|
|
|
4466 |
|
|
// check that we have the entire packet in memory |
4467 |
|
✗ |
if (!is_whole_packet_present(f)) { |
4468 |
|
✗ |
*samples = 0; |
4469 |
|
✗ |
return 0; |
4470 |
|
|
} |
4471 |
|
|
|
4472 |
|
✗ |
if (!vorbis_decode_packet(f, &len, &left, &right)) { |
4473 |
|
|
// save the actual error we encountered |
4474 |
|
✗ |
enum STBVorbisError error = f->error; |
4475 |
|
✗ |
if (error == VORBIS_bad_packet_type) { |
4476 |
|
|
// flush and resynch |
4477 |
|
✗ |
f->error = VORBIS__no_error; |
4478 |
|
✗ |
while (get8_packet(f) != EOP) |
4479 |
|
✗ |
if (f->eof) break; |
4480 |
|
✗ |
*samples = 0; |
4481 |
|
✗ |
return (int) (f->stream - data); |
4482 |
|
|
} |
4483 |
|
✗ |
if (error == VORBIS_continued_packet_flag_invalid) { |
4484 |
|
✗ |
if (f->previous_length == 0) { |
4485 |
|
|
// we may be resynching, in which case it's ok to hit one |
4486 |
|
|
// of these; just discard the packet |
4487 |
|
✗ |
f->error = VORBIS__no_error; |
4488 |
|
✗ |
while (get8_packet(f) != EOP) |
4489 |
|
✗ |
if (f->eof) break; |
4490 |
|
✗ |
*samples = 0; |
4491 |
|
✗ |
return (int) (f->stream - data); |
4492 |
|
|
} |
4493 |
|
|
} |
4494 |
|
|
// if we get an error while parsing, what to do? |
4495 |
|
|
// well, it DEFINITELY won't work to continue from where we are! |
4496 |
|
✗ |
stb_vorbis_flush_pushdata(f); |
4497 |
|
|
// restore the error that actually made us bail |
4498 |
|
✗ |
f->error = error; |
4499 |
|
✗ |
*samples = 0; |
4500 |
|
✗ |
return 1; |
4501 |
|
|
} |
4502 |
|
|
|
4503 |
|
|
// success! |
4504 |
|
✗ |
len = vorbis_finish_frame(f, len, left, right); |
4505 |
|
✗ |
for (i=0; i < f->channels; ++i) |
4506 |
|
✗ |
f->outputs[i] = f->channel_buffers[i] + left; |
4507 |
|
|
|
4508 |
|
✗ |
if (channels) *channels = f->channels; |
4509 |
|
✗ |
*samples = len; |
4510 |
|
✗ |
*output = f->outputs; |
4511 |
|
✗ |
return (int) (f->stream - data); |
4512 |
|
|
} |
4513 |
|
|
|
4514 |
|
✗ |
stb_vorbis *stb_vorbis_open_pushdata( |
4515 |
|
|
const unsigned char *data, int data_len, // the memory available for decoding |
4516 |
|
|
int *data_used, // only defined if result is not NULL |
4517 |
|
|
int *error, const stb_vorbis_alloc *alloc) |
4518 |
|
|
{ |
4519 |
|
|
stb_vorbis *f, p; |
4520 |
|
✗ |
vorbis_init(&p, alloc); |
4521 |
|
✗ |
p.stream = (uint8 *) data; |
4522 |
|
✗ |
p.stream_end = (uint8 *) data + data_len; |
4523 |
|
✗ |
p.push_mode = TRUE; |
4524 |
|
✗ |
if (!start_decoder(&p)) { |
4525 |
|
✗ |
if (p.eof) |
4526 |
|
✗ |
*error = VORBIS_need_more_data; |
4527 |
|
|
else |
4528 |
|
✗ |
*error = p.error; |
4529 |
|
✗ |
vorbis_deinit(&p); |
4530 |
|
✗ |
return NULL; |
4531 |
|
|
} |
4532 |
|
|
f = vorbis_alloc(&p); |
4533 |
|
✗ |
if (f) { |
4534 |
|
✗ |
*f = p; |
4535 |
|
✗ |
*data_used = (int) (f->stream - data); |
4536 |
|
✗ |
*error = 0; |
4537 |
|
✗ |
return f; |
4538 |
|
|
} else { |
4539 |
|
✗ |
vorbis_deinit(&p); |
4540 |
|
✗ |
return NULL; |
4541 |
|
|
} |
4542 |
|
|
} |
4543 |
|
|
#endif // STB_VORBIS_NO_PUSHDATA_API |
4544 |
|
|
|
4545 |
|
✗ |
unsigned int stb_vorbis_get_file_offset(stb_vorbis *f) |
4546 |
|
|
{ |
4547 |
|
|
#ifndef STB_VORBIS_NO_PUSHDATA_API |
4548 |
|
✗ |
if (f->push_mode) return 0; |
4549 |
|
|
#endif |
4550 |
|
✗ |
if (USE_MEMORY(f)) return (unsigned int) (f->stream - f->stream_start); |
4551 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
4552 |
|
✗ |
return (unsigned int) (ftell(f->f) - f->f_start); |
4553 |
|
|
#endif |
4554 |
|
|
} |
4555 |
|
|
|
4556 |
|
|
#ifndef STB_VORBIS_NO_PULLDATA_API |
4557 |
|
|
// |
4558 |
|
|
// DATA-PULLING API |
4559 |
|
|
// |
4560 |
|
|
|
4561 |
|
✗ |
static uint32 vorbis_find_page(stb_vorbis *f, uint32 *end, uint32 *last) |
4562 |
|
|
{ |
4563 |
|
|
for(;;) { |
4564 |
|
|
int n; |
4565 |
|
✗ |
if (f->eof) return 0; |
4566 |
|
✗ |
n = get8(f); |
4567 |
|
✗ |
if (n == 0x4f) { // page header candidate |
4568 |
|
✗ |
unsigned int retry_loc = stb_vorbis_get_file_offset(f); |
4569 |
|
|
int i; |
4570 |
|
|
// check if we're off the end of a file_section stream |
4571 |
|
✗ |
if (retry_loc - 25 > f->stream_len) |
4572 |
|
|
return 0; |
4573 |
|
|
// check the rest of the header |
4574 |
|
✗ |
for (i=1; i < 4; ++i) |
4575 |
|
✗ |
if (get8(f) != ogg_page_header[i]) |
4576 |
|
|
break; |
4577 |
|
✗ |
if (f->eof) return 0; |
4578 |
|
✗ |
if (i == 4) { |
4579 |
|
|
uint8 header[27]; |
4580 |
|
|
uint32 i, crc, goal, len; |
4581 |
|
✗ |
for (i=0; i < 4; ++i) |
4582 |
|
✗ |
header[i] = ogg_page_header[i]; |
4583 |
|
✗ |
for (; i < 27; ++i) |
4584 |
|
✗ |
header[i] = get8(f); |
4585 |
|
✗ |
if (f->eof) return 0; |
4586 |
|
✗ |
if (header[4] != 0) goto invalid; |
4587 |
|
✗ |
goal = header[22] + (header[23] << 8) + (header[24]<<16) + ((uint32)header[25]<<24); |
4588 |
|
✗ |
for (i=22; i < 26; ++i) |
4589 |
|
✗ |
header[i] = 0; |
4590 |
|
|
crc = 0; |
4591 |
|
✗ |
for (i=0; i < 27; ++i) |
4592 |
|
✗ |
crc = crc32_update(crc, header[i]); |
4593 |
|
|
len = 0; |
4594 |
|
✗ |
for (i=0; i < header[26]; ++i) { |
4595 |
|
✗ |
int s = get8(f); |
4596 |
|
|
crc = crc32_update(crc, s); |
4597 |
|
✗ |
len += s; |
4598 |
|
|
} |
4599 |
|
✗ |
if (len && f->eof) return 0; |
4600 |
|
✗ |
for (i=0; i < len; ++i) |
4601 |
|
✗ |
crc = crc32_update(crc, get8(f)); |
4602 |
|
|
// finished parsing probable page |
4603 |
|
✗ |
if (crc == goal) { |
4604 |
|
|
// we could now check that it's either got the last |
4605 |
|
|
// page flag set, OR it's followed by the capture |
4606 |
|
|
// pattern, but I guess TECHNICALLY you could have |
4607 |
|
|
// a file with garbage between each ogg page and recover |
4608 |
|
|
// from it automatically? So even though that paranoia |
4609 |
|
|
// might decrease the chance of an invalid decode by |
4610 |
|
|
// another 2^32, not worth it since it would hose those |
4611 |
|
|
// invalid-but-useful files? |
4612 |
|
✗ |
if (end) |
4613 |
|
✗ |
*end = stb_vorbis_get_file_offset(f); |
4614 |
|
✗ |
if (last) { |
4615 |
|
✗ |
if (header[5] & 0x04) |
4616 |
|
✗ |
*last = 1; |
4617 |
|
|
else |
4618 |
|
✗ |
*last = 0; |
4619 |
|
|
} |
4620 |
|
✗ |
set_file_offset(f, retry_loc-1); |
4621 |
|
✗ |
return 1; |
4622 |
|
|
} |
4623 |
|
|
} |
4624 |
|
✗ |
invalid: |
4625 |
|
|
// not a valid page, so rewind and look for next one |
4626 |
|
✗ |
set_file_offset(f, retry_loc); |
4627 |
|
|
} |
4628 |
|
|
} |
4629 |
|
|
} |
4630 |
|
|
|
4631 |
|
|
|
4632 |
|
|
#define SAMPLE_unknown 0xffffffff |
4633 |
|
|
|
4634 |
|
|
// seeking is implemented with a binary search, which narrows down the range to |
4635 |
|
|
// 64K, before using a linear search (because finding the synchronization |
4636 |
|
|
// pattern can be expensive, and the chance we'd find the end page again is |
4637 |
|
|
// relatively high for small ranges) |
4638 |
|
|
// |
4639 |
|
|
// two initial interpolation-style probes are used at the start of the search |
4640 |
|
|
// to try to bound either side of the binary search sensibly, while still |
4641 |
|
|
// working in O(log n) time if they fail. |
4642 |
|
|
|
4643 |
|
✗ |
static int get_seek_page_info(stb_vorbis *f, ProbedPage *z) |
4644 |
|
|
{ |
4645 |
|
|
uint8 header[27], lacing[255]; |
4646 |
|
|
int i,len; |
4647 |
|
|
|
4648 |
|
|
// record where the page starts |
4649 |
|
✗ |
z->page_start = stb_vorbis_get_file_offset(f); |
4650 |
|
|
|
4651 |
|
|
// parse the header |
4652 |
|
✗ |
getn(f, header, 27); |
4653 |
|
✗ |
if (header[0] != 'O' || header[1] != 'g' || header[2] != 'g' || header[3] != 'S') |
4654 |
|
|
return 0; |
4655 |
|
✗ |
getn(f, lacing, header[26]); |
4656 |
|
|
|
4657 |
|
|
// determine the length of the payload |
4658 |
|
|
len = 0; |
4659 |
|
✗ |
for (i=0; i < header[26]; ++i) |
4660 |
|
✗ |
len += lacing[i]; |
4661 |
|
|
|
4662 |
|
|
// this implies where the page ends |
4663 |
|
✗ |
z->page_end = z->page_start + 27 + header[26] + len; |
4664 |
|
|
|
4665 |
|
|
// read the last-decoded sample out of the data |
4666 |
|
✗ |
z->last_decoded_sample = header[6] + (header[7] << 8) + (header[8] << 16) + (header[9] << 24); |
4667 |
|
|
|
4668 |
|
|
// restore file state to where we were |
4669 |
|
✗ |
set_file_offset(f, z->page_start); |
4670 |
|
✗ |
return 1; |
4671 |
|
|
} |
4672 |
|
|
|
4673 |
|
|
// rarely used function to seek back to the preceding page while finding the |
4674 |
|
|
// start of a packet |
4675 |
|
✗ |
static int go_to_page_before(stb_vorbis *f, unsigned int limit_offset) |
4676 |
|
|
{ |
4677 |
|
|
unsigned int previous_safe, end; |
4678 |
|
|
|
4679 |
|
|
// now we want to seek back 64K from the limit |
4680 |
|
✗ |
if (limit_offset >= 65536 && limit_offset-65536 >= f->first_audio_page_offset) |
4681 |
|
|
previous_safe = limit_offset - 65536; |
4682 |
|
|
else |
4683 |
|
✗ |
previous_safe = f->first_audio_page_offset; |
4684 |
|
|
|
4685 |
|
✗ |
set_file_offset(f, previous_safe); |
4686 |
|
|
|
4687 |
|
✗ |
while (vorbis_find_page(f, &end, NULL)) { |
4688 |
|
✗ |
if (end >= limit_offset && stb_vorbis_get_file_offset(f) < limit_offset) |
4689 |
|
|
return 1; |
4690 |
|
✗ |
set_file_offset(f, end); |
4691 |
|
|
} |
4692 |
|
|
|
4693 |
|
|
return 0; |
4694 |
|
|
} |
4695 |
|
|
|
4696 |
|
|
// implements the search logic for finding a page and starting decoding. if |
4697 |
|
|
// the function succeeds, current_loc_valid will be true and current_loc will |
4698 |
|
|
// be less than or equal to the provided sample number (the closer the |
4699 |
|
|
// better). |
4700 |
|
✗ |
static int seek_to_sample_coarse(stb_vorbis *f, uint32 sample_number) |
4701 |
|
|
{ |
4702 |
|
|
ProbedPage left, right, mid; |
4703 |
|
|
int i, start_seg_with_known_loc, end_pos, page_start; |
4704 |
|
|
uint32 delta, stream_length, padding, last_sample_limit; |
4705 |
|
|
double offset = 0.0, bytes_per_sample = 0.0; |
4706 |
|
|
int probe = 0; |
4707 |
|
|
|
4708 |
|
|
// find the last page and validate the target sample |
4709 |
|
✗ |
stream_length = stb_vorbis_stream_length_in_samples(f); |
4710 |
|
✗ |
if (stream_length == 0) return error(f, VORBIS_seek_without_length); |
4711 |
|
✗ |
if (sample_number > stream_length) return error(f, VORBIS_seek_invalid); |
4712 |
|
|
|
4713 |
|
|
// this is the maximum difference between the window-center (which is the |
4714 |
|
|
// actual granule position value), and the right-start (which the spec |
4715 |
|
|
// indicates should be the granule position (give or take one)). |
4716 |
|
✗ |
padding = ((f->blocksize_1 - f->blocksize_0) >> 2); |
4717 |
|
✗ |
if (sample_number < padding) |
4718 |
|
|
last_sample_limit = 0; |
4719 |
|
|
else |
4720 |
|
✗ |
last_sample_limit = sample_number - padding; |
4721 |
|
|
|
4722 |
|
✗ |
left = f->p_first; |
4723 |
|
✗ |
while (left.last_decoded_sample == ~0U) { |
4724 |
|
|
// (untested) the first page does not have a 'last_decoded_sample' |
4725 |
|
✗ |
set_file_offset(f, left.page_end); |
4726 |
|
✗ |
if (!get_seek_page_info(f, &left)) goto error; |
4727 |
|
|
} |
4728 |
|
|
|
4729 |
|
✗ |
right = f->p_last; |
4730 |
|
|
assert(right.last_decoded_sample != ~0U); |
4731 |
|
|
|
4732 |
|
|
// starting from the start is handled differently |
4733 |
|
✗ |
if (last_sample_limit <= left.last_decoded_sample) { |
4734 |
|
✗ |
if (stb_vorbis_seek_start(f)) { |
4735 |
|
✗ |
if (f->current_loc > sample_number) |
4736 |
|
✗ |
return error(f, VORBIS_seek_failed); |
4737 |
|
|
return 1; |
4738 |
|
|
} |
4739 |
|
|
return 0; |
4740 |
|
|
} |
4741 |
|
|
|
4742 |
|
✗ |
while (left.page_end != right.page_start) { |
4743 |
|
|
assert(left.page_end < right.page_start); |
4744 |
|
|
// search range in bytes |
4745 |
|
✗ |
delta = right.page_start - left.page_end; |
4746 |
|
✗ |
if (delta <= 65536) { |
4747 |
|
|
// there's only 64K left to search - handle it linearly |
4748 |
|
✗ |
set_file_offset(f, left.page_end); |
4749 |
|
|
} else { |
4750 |
|
✗ |
if (probe < 2) { |
4751 |
|
✗ |
if (probe == 0) { |
4752 |
|
|
// first probe (interpolate) |
4753 |
|
✗ |
double data_bytes = right.page_end - left.page_start; |
4754 |
|
✗ |
bytes_per_sample = data_bytes / right.last_decoded_sample; |
4755 |
|
✗ |
offset = left.page_start + bytes_per_sample * (last_sample_limit - left.last_decoded_sample); |
4756 |
|
|
} else { |
4757 |
|
|
// second probe (try to bound the other side) |
4758 |
|
✗ |
double error = ((double) last_sample_limit - mid.last_decoded_sample) * bytes_per_sample; |
4759 |
|
✗ |
if (error >= 0 && error < 8000) error = 8000; |
4760 |
|
✗ |
if (error < 0 && error > -8000) error = -8000; |
4761 |
|
✗ |
offset += error * 2; |
4762 |
|
|
} |
4763 |
|
|
|
4764 |
|
|
// ensure the offset is valid |
4765 |
|
✗ |
if (offset < left.page_end) |
4766 |
|
|
offset = left.page_end; |
4767 |
|
✗ |
if (offset > right.page_start - 65536) |
4768 |
|
|
offset = right.page_start - 65536; |
4769 |
|
|
|
4770 |
|
✗ |
set_file_offset(f, (unsigned int) offset); |
4771 |
|
|
} else { |
4772 |
|
|
// binary search for large ranges (offset by 32K to ensure |
4773 |
|
|
// we don't hit the right page) |
4774 |
|
✗ |
set_file_offset(f, left.page_end + (delta / 2) - 32768); |
4775 |
|
|
} |
4776 |
|
|
|
4777 |
|
✗ |
if (!vorbis_find_page(f, NULL, NULL)) goto error; |
4778 |
|
|
} |
4779 |
|
|
|
4780 |
|
|
for (;;) { |
4781 |
|
✗ |
if (!get_seek_page_info(f, &mid)) goto error; |
4782 |
|
✗ |
if (mid.last_decoded_sample != ~0U) break; |
4783 |
|
|
// (untested) no frames end on this page |
4784 |
|
✗ |
set_file_offset(f, mid.page_end); |
4785 |
|
|
assert(mid.page_start < right.page_start); |
4786 |
|
|
} |
4787 |
|
|
|
4788 |
|
|
// if we've just found the last page again then we're in a tricky file, |
4789 |
|
|
// and we're close enough (if it wasn't an interpolation probe). |
4790 |
|
✗ |
if (mid.page_start == right.page_start) { |
4791 |
|
✗ |
if (probe >= 2 || delta <= 65536) |
4792 |
|
|
break; |
4793 |
|
|
} else { |
4794 |
|
✗ |
if (last_sample_limit < mid.last_decoded_sample) |
4795 |
|
✗ |
right = mid; |
4796 |
|
|
else |
4797 |
|
✗ |
left = mid; |
4798 |
|
|
} |
4799 |
|
|
|
4800 |
|
✗ |
++probe; |
4801 |
|
|
} |
4802 |
|
|
|
4803 |
|
|
// seek back to start of the last packet |
4804 |
|
✗ |
page_start = left.page_start; |
4805 |
|
✗ |
set_file_offset(f, page_start); |
4806 |
|
✗ |
if (!start_page(f)) return error(f, VORBIS_seek_failed); |
4807 |
|
✗ |
end_pos = f->end_seg_with_known_loc; |
4808 |
|
|
assert(end_pos >= 0); |
4809 |
|
|
|
4810 |
|
|
for (;;) { |
4811 |
|
✗ |
for (i = end_pos; i > 0; --i) |
4812 |
|
✗ |
if (f->segments[i-1] != 255) |
4813 |
|
|
break; |
4814 |
|
|
|
4815 |
|
|
start_seg_with_known_loc = i; |
4816 |
|
|
|
4817 |
|
✗ |
if (start_seg_with_known_loc > 0 || !(f->page_flag & PAGEFLAG_continued_packet)) |
4818 |
|
|
break; |
4819 |
|
|
|
4820 |
|
|
// (untested) the final packet begins on an earlier page |
4821 |
|
✗ |
if (!go_to_page_before(f, page_start)) |
4822 |
|
✗ |
goto error; |
4823 |
|
|
|
4824 |
|
✗ |
page_start = stb_vorbis_get_file_offset(f); |
4825 |
|
✗ |
if (!start_page(f)) goto error; |
4826 |
|
✗ |
end_pos = f->segment_count - 1; |
4827 |
|
|
} |
4828 |
|
|
|
4829 |
|
|
// prepare to start decoding |
4830 |
|
✗ |
f->current_loc_valid = FALSE; |
4831 |
|
✗ |
f->last_seg = FALSE; |
4832 |
|
✗ |
f->valid_bits = 0; |
4833 |
|
✗ |
f->packet_bytes = 0; |
4834 |
|
✗ |
f->bytes_in_seg = 0; |
4835 |
|
✗ |
f->previous_length = 0; |
4836 |
|
✗ |
f->next_seg = start_seg_with_known_loc; |
4837 |
|
|
|
4838 |
|
✗ |
for (i = 0; i < start_seg_with_known_loc; i++) |
4839 |
|
✗ |
skip(f, f->segments[i]); |
4840 |
|
|
|
4841 |
|
|
// start decoding (optimizable - this frame is generally discarded) |
4842 |
|
✗ |
if (!vorbis_pump_first_frame(f)) |
4843 |
|
|
return 0; |
4844 |
|
✗ |
if (f->current_loc > sample_number) |
4845 |
|
✗ |
return error(f, VORBIS_seek_failed); |
4846 |
|
|
return 1; |
4847 |
|
|
|
4848 |
|
✗ |
error: |
4849 |
|
|
// try to restore the file to a valid state |
4850 |
|
✗ |
stb_vorbis_seek_start(f); |
4851 |
|
✗ |
return error(f, VORBIS_seek_failed); |
4852 |
|
|
} |
4853 |
|
|
|
4854 |
|
|
// the same as vorbis_decode_initial, but without advancing |
4855 |
|
✗ |
static int peek_decode_initial(vorb *f, int *p_left_start, int *p_left_end, int *p_right_start, int *p_right_end, int *mode) |
4856 |
|
|
{ |
4857 |
|
|
int bits_read, bytes_read; |
4858 |
|
|
|
4859 |
|
✗ |
if (!vorbis_decode_initial(f, p_left_start, p_left_end, p_right_start, p_right_end, mode)) |
4860 |
|
|
return 0; |
4861 |
|
|
|
4862 |
|
|
// either 1 or 2 bytes were read, figure out which so we can rewind |
4863 |
|
✗ |
bits_read = 1 + ilog(f->mode_count-1); |
4864 |
|
✗ |
if (f->mode_config[*mode].blockflag) |
4865 |
|
✗ |
bits_read += 2; |
4866 |
|
✗ |
bytes_read = (bits_read + 7) / 8; |
4867 |
|
|
|
4868 |
|
✗ |
f->bytes_in_seg += bytes_read; |
4869 |
|
✗ |
f->packet_bytes -= bytes_read; |
4870 |
|
✗ |
skip(f, -bytes_read); |
4871 |
|
✗ |
if (f->next_seg == -1) |
4872 |
|
✗ |
f->next_seg = f->segment_count - 1; |
4873 |
|
|
else |
4874 |
|
✗ |
f->next_seg--; |
4875 |
|
✗ |
f->valid_bits = 0; |
4876 |
|
|
|
4877 |
|
✗ |
return 1; |
4878 |
|
|
} |
4879 |
|
|
|
4880 |
|
✗ |
int stb_vorbis_seek_frame(stb_vorbis *f, unsigned int sample_number) |
4881 |
|
|
{ |
4882 |
|
|
uint32 max_frame_samples; |
4883 |
|
|
|
4884 |
|
✗ |
if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing); |
4885 |
|
|
|
4886 |
|
|
// fast page-level search |
4887 |
|
✗ |
if (!seek_to_sample_coarse(f, sample_number)) |
4888 |
|
|
return 0; |
4889 |
|
|
|
4890 |
|
|
assert(f->current_loc_valid); |
4891 |
|
|
assert(f->current_loc <= sample_number); |
4892 |
|
|
|
4893 |
|
|
// linear search for the relevant packet |
4894 |
|
✗ |
max_frame_samples = (f->blocksize_1*3 - f->blocksize_0) >> 2; |
4895 |
|
✗ |
while (f->current_loc < sample_number) { |
4896 |
|
|
int left_start, left_end, right_start, right_end, mode, frame_samples; |
4897 |
|
✗ |
if (!peek_decode_initial(f, &left_start, &left_end, &right_start, &right_end, &mode)) |
4898 |
|
✗ |
return error(f, VORBIS_seek_failed); |
4899 |
|
|
// calculate the number of samples returned by the next frame |
4900 |
|
✗ |
frame_samples = right_start - left_start; |
4901 |
|
✗ |
if (f->current_loc + frame_samples > sample_number) { |
4902 |
|
|
return 1; // the next frame will contain the sample |
4903 |
|
✗ |
} else if (f->current_loc + frame_samples + max_frame_samples > sample_number) { |
4904 |
|
|
// there's a chance the frame after this could contain the sample |
4905 |
|
✗ |
vorbis_pump_first_frame(f); |
4906 |
|
|
} else { |
4907 |
|
|
// this frame is too early to be relevant |
4908 |
|
✗ |
f->current_loc += frame_samples; |
4909 |
|
✗ |
f->previous_length = 0; |
4910 |
|
✗ |
maybe_start_packet(f); |
4911 |
|
|
flush_packet(f); |
4912 |
|
|
} |
4913 |
|
|
} |
4914 |
|
|
// the next frame should start with the sample |
4915 |
|
✗ |
if (f->current_loc != sample_number) return error(f, VORBIS_seek_failed); |
4916 |
|
|
return 1; |
4917 |
|
|
} |
4918 |
|
|
|
4919 |
|
✗ |
int stb_vorbis_seek(stb_vorbis *f, unsigned int sample_number) |
4920 |
|
|
{ |
4921 |
|
✗ |
if (!stb_vorbis_seek_frame(f, sample_number)) |
4922 |
|
|
return 0; |
4923 |
|
|
|
4924 |
|
✗ |
if (sample_number != f->current_loc) { |
4925 |
|
|
int n; |
4926 |
|
|
uint32 frame_start = f->current_loc; |
4927 |
|
✗ |
stb_vorbis_get_frame_float(f, &n, NULL); |
4928 |
|
|
assert(sample_number > frame_start); |
4929 |
|
|
assert(f->channel_buffer_start + (int) (sample_number-frame_start) <= f->channel_buffer_end); |
4930 |
|
✗ |
f->channel_buffer_start += (sample_number - frame_start); |
4931 |
|
|
} |
4932 |
|
|
|
4933 |
|
|
return 1; |
4934 |
|
|
} |
4935 |
|
|
|
4936 |
|
✗ |
int stb_vorbis_seek_start(stb_vorbis *f) |
4937 |
|
|
{ |
4938 |
|
✗ |
if (IS_PUSH_MODE(f)) { return error(f, VORBIS_invalid_api_mixing); } |
4939 |
|
✗ |
set_file_offset(f, f->first_audio_page_offset); |
4940 |
|
✗ |
f->previous_length = 0; |
4941 |
|
✗ |
f->first_decode = TRUE; |
4942 |
|
✗ |
f->next_seg = -1; |
4943 |
|
✗ |
return vorbis_pump_first_frame(f); |
4944 |
|
|
} |
4945 |
|
|
|
4946 |
|
✗ |
unsigned int stb_vorbis_stream_length_in_samples(stb_vorbis *f) |
4947 |
|
|
{ |
4948 |
|
|
unsigned int restore_offset, previous_safe; |
4949 |
|
|
unsigned int end, last_page_loc; |
4950 |
|
|
|
4951 |
|
✗ |
if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing); |
4952 |
|
✗ |
if (!f->total_samples) { |
4953 |
|
|
unsigned int last; |
4954 |
|
|
uint32 lo,hi; |
4955 |
|
|
char header[6]; |
4956 |
|
|
|
4957 |
|
|
// first, store the current decode position so we can restore it |
4958 |
|
✗ |
restore_offset = stb_vorbis_get_file_offset(f); |
4959 |
|
|
|
4960 |
|
|
// now we want to seek back 64K from the end (the last page must |
4961 |
|
|
// be at most a little less than 64K, but let's allow a little slop) |
4962 |
|
✗ |
if (f->stream_len >= 65536 && f->stream_len-65536 >= f->first_audio_page_offset) |
4963 |
|
|
previous_safe = f->stream_len - 65536; |
4964 |
|
|
else |
4965 |
|
✗ |
previous_safe = f->first_audio_page_offset; |
4966 |
|
|
|
4967 |
|
✗ |
set_file_offset(f, previous_safe); |
4968 |
|
|
// previous_safe is now our candidate 'earliest known place that seeking |
4969 |
|
|
// to will lead to the final page' |
4970 |
|
|
|
4971 |
|
✗ |
if (!vorbis_find_page(f, &end, &last)) { |
4972 |
|
|
// if we can't find a page, we're hosed! |
4973 |
|
✗ |
f->error = VORBIS_cant_find_last_page; |
4974 |
|
✗ |
f->total_samples = 0xffffffff; |
4975 |
|
✗ |
goto done; |
4976 |
|
|
} |
4977 |
|
|
|
4978 |
|
|
// check if there are more pages |
4979 |
|
✗ |
last_page_loc = stb_vorbis_get_file_offset(f); |
4980 |
|
|
|
4981 |
|
|
// stop when the last_page flag is set, not when we reach eof; |
4982 |
|
|
// this allows us to stop short of a 'file_section' end without |
4983 |
|
|
// explicitly checking the length of the section |
4984 |
|
✗ |
while (!last) { |
4985 |
|
✗ |
set_file_offset(f, end); |
4986 |
|
✗ |
if (!vorbis_find_page(f, &end, &last)) { |
4987 |
|
|
// the last page we found didn't have the 'last page' flag |
4988 |
|
|
// set. whoops! |
4989 |
|
|
break; |
4990 |
|
|
} |
4991 |
|
|
//previous_safe = last_page_loc+1; // NOTE: not used after this point, but note for debugging |
4992 |
|
✗ |
last_page_loc = stb_vorbis_get_file_offset(f); |
4993 |
|
|
} |
4994 |
|
|
|
4995 |
|
✗ |
set_file_offset(f, last_page_loc); |
4996 |
|
|
|
4997 |
|
|
// parse the header |
4998 |
|
✗ |
getn(f, (unsigned char *)header, 6); |
4999 |
|
|
// extract the absolute granule position |
5000 |
|
✗ |
lo = get32(f); |
5001 |
|
✗ |
hi = get32(f); |
5002 |
|
✗ |
if (lo == 0xffffffff && hi == 0xffffffff) { |
5003 |
|
✗ |
f->error = VORBIS_cant_find_last_page; |
5004 |
|
✗ |
f->total_samples = SAMPLE_unknown; |
5005 |
|
✗ |
goto done; |
5006 |
|
|
} |
5007 |
|
✗ |
if (hi) |
5008 |
|
|
lo = 0xfffffffe; // saturate |
5009 |
|
✗ |
f->total_samples = lo; |
5010 |
|
|
|
5011 |
|
✗ |
f->p_last.page_start = last_page_loc; |
5012 |
|
✗ |
f->p_last.page_end = end; |
5013 |
|
✗ |
f->p_last.last_decoded_sample = lo; |
5014 |
|
|
|
5015 |
|
✗ |
done: |
5016 |
|
✗ |
set_file_offset(f, restore_offset); |
5017 |
|
|
} |
5018 |
|
✗ |
return f->total_samples == SAMPLE_unknown ? 0 : f->total_samples; |
5019 |
|
|
} |
5020 |
|
|
|
5021 |
|
✗ |
float stb_vorbis_stream_length_in_seconds(stb_vorbis *f) |
5022 |
|
|
{ |
5023 |
|
✗ |
return stb_vorbis_stream_length_in_samples(f) / (float) f->sample_rate; |
5024 |
|
|
} |
5025 |
|
|
|
5026 |
|
|
|
5027 |
|
|
|
5028 |
|
✗ |
int stb_vorbis_get_frame_float(stb_vorbis *f, int *channels, float ***output) |
5029 |
|
|
{ |
5030 |
|
|
int len, right,left,i; |
5031 |
|
✗ |
if (IS_PUSH_MODE(f)) return error(f, VORBIS_invalid_api_mixing); |
5032 |
|
|
|
5033 |
|
✗ |
if (!vorbis_decode_packet(f, &len, &left, &right)) { |
5034 |
|
✗ |
f->channel_buffer_start = f->channel_buffer_end = 0; |
5035 |
|
✗ |
return 0; |
5036 |
|
|
} |
5037 |
|
|
|
5038 |
|
✗ |
len = vorbis_finish_frame(f, len, left, right); |
5039 |
|
✗ |
for (i=0; i < f->channels; ++i) |
5040 |
|
✗ |
f->outputs[i] = f->channel_buffers[i] + left; |
5041 |
|
|
|
5042 |
|
✗ |
f->channel_buffer_start = left; |
5043 |
|
✗ |
f->channel_buffer_end = left+len; |
5044 |
|
|
|
5045 |
|
✗ |
if (channels) *channels = f->channels; |
5046 |
|
✗ |
if (output) *output = f->outputs; |
5047 |
|
|
return len; |
5048 |
|
|
} |
5049 |
|
|
|
5050 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
5051 |
|
|
|
5052 |
|
✗ |
stb_vorbis * stb_vorbis_open_file_section(FILE *file, int close_on_free, int *error, const stb_vorbis_alloc *alloc, unsigned int length) |
5053 |
|
|
{ |
5054 |
|
|
stb_vorbis *f, p; |
5055 |
|
✗ |
vorbis_init(&p, alloc); |
5056 |
|
✗ |
p.f = file; |
5057 |
|
✗ |
p.f_start = (uint32) ftell(file); |
5058 |
|
✗ |
p.stream_len = length; |
5059 |
|
✗ |
p.close_on_free = close_on_free; |
5060 |
|
✗ |
if (start_decoder(&p)) { |
5061 |
|
|
f = vorbis_alloc(&p); |
5062 |
|
✗ |
if (f) { |
5063 |
|
✗ |
*f = p; |
5064 |
|
✗ |
vorbis_pump_first_frame(f); |
5065 |
|
✗ |
return f; |
5066 |
|
|
} |
5067 |
|
|
} |
5068 |
|
✗ |
if (error) *error = p.error; |
5069 |
|
✗ |
vorbis_deinit(&p); |
5070 |
|
✗ |
return NULL; |
5071 |
|
|
} |
5072 |
|
|
|
5073 |
|
✗ |
stb_vorbis * stb_vorbis_open_file(FILE *file, int close_on_free, int *error, const stb_vorbis_alloc *alloc) |
5074 |
|
|
{ |
5075 |
|
|
unsigned int len, start; |
5076 |
|
✗ |
start = (unsigned int) ftell(file); |
5077 |
|
✗ |
fseek(file, 0, SEEK_END); |
5078 |
|
✗ |
len = (unsigned int) (ftell(file) - start); |
5079 |
|
✗ |
fseek(file, start, SEEK_SET); |
5080 |
|
✗ |
return stb_vorbis_open_file_section(file, close_on_free, error, alloc, len); |
5081 |
|
|
} |
5082 |
|
|
|
5083 |
|
✗ |
stb_vorbis * stb_vorbis_open_filename(const char *filename, int *error, const stb_vorbis_alloc *alloc) |
5084 |
|
|
{ |
5085 |
|
|
FILE *f; |
5086 |
|
|
#if defined(_WIN32) && defined(__STDC_WANT_SECURE_LIB__) |
5087 |
|
|
if (0 != fopen_s(&f, filename, "rb")) |
5088 |
|
|
f = NULL; |
5089 |
|
|
#else |
5090 |
|
✗ |
f = fopen(filename, "rb"); |
5091 |
|
|
#endif |
5092 |
|
✗ |
if (f) |
5093 |
|
✗ |
return stb_vorbis_open_file(f, TRUE, error, alloc); |
5094 |
|
✗ |
if (error) *error = VORBIS_file_open_failure; |
5095 |
|
|
return NULL; |
5096 |
|
|
} |
5097 |
|
|
#endif // STB_VORBIS_NO_STDIO |
5098 |
|
|
|
5099 |
|
✗ |
stb_vorbis * stb_vorbis_open_memory(const unsigned char *data, int len, int *error, const stb_vorbis_alloc *alloc) |
5100 |
|
|
{ |
5101 |
|
|
stb_vorbis *f, p; |
5102 |
|
✗ |
if (!data) { |
5103 |
|
✗ |
if (error) *error = VORBIS_unexpected_eof; |
5104 |
|
✗ |
return NULL; |
5105 |
|
|
} |
5106 |
|
✗ |
vorbis_init(&p, alloc); |
5107 |
|
✗ |
p.stream = (uint8 *) data; |
5108 |
|
✗ |
p.stream_end = (uint8 *) data + len; |
5109 |
|
✗ |
p.stream_start = (uint8 *) p.stream; |
5110 |
|
✗ |
p.stream_len = len; |
5111 |
|
✗ |
p.push_mode = FALSE; |
5112 |
|
✗ |
if (start_decoder(&p)) { |
5113 |
|
|
f = vorbis_alloc(&p); |
5114 |
|
✗ |
if (f) { |
5115 |
|
✗ |
*f = p; |
5116 |
|
✗ |
vorbis_pump_first_frame(f); |
5117 |
|
✗ |
if (error) *error = VORBIS__no_error; |
5118 |
|
✗ |
return f; |
5119 |
|
|
} |
5120 |
|
|
} |
5121 |
|
✗ |
if (error) *error = p.error; |
5122 |
|
✗ |
vorbis_deinit(&p); |
5123 |
|
✗ |
return NULL; |
5124 |
|
|
} |
5125 |
|
|
|
5126 |
|
|
#ifndef STB_VORBIS_NO_INTEGER_CONVERSION |
5127 |
|
|
#define PLAYBACK_MONO 1 |
5128 |
|
|
#define PLAYBACK_LEFT 2 |
5129 |
|
|
#define PLAYBACK_RIGHT 4 |
5130 |
|
|
|
5131 |
|
|
#define L (PLAYBACK_LEFT | PLAYBACK_MONO) |
5132 |
|
|
#define C (PLAYBACK_LEFT | PLAYBACK_RIGHT | PLAYBACK_MONO) |
5133 |
|
|
#define R (PLAYBACK_RIGHT | PLAYBACK_MONO) |
5134 |
|
|
|
5135 |
|
|
static int8 channel_position[7][6] = |
5136 |
|
|
{ |
5137 |
|
|
{ 0 }, |
5138 |
|
|
{ C }, |
5139 |
|
|
{ L, R }, |
5140 |
|
|
{ L, C, R }, |
5141 |
|
|
{ L, R, L, R }, |
5142 |
|
|
{ L, C, R, L, R }, |
5143 |
|
|
{ L, C, R, L, R, C }, |
5144 |
|
|
}; |
5145 |
|
|
|
5146 |
|
|
|
5147 |
|
|
#ifndef STB_VORBIS_NO_FAST_SCALED_FLOAT |
5148 |
|
|
typedef union { |
5149 |
|
|
float f; |
5150 |
|
|
int i; |
5151 |
|
|
} float_conv; |
5152 |
|
|
typedef char stb_vorbis_float_size_test[sizeof(float)==4 && sizeof(int) == 4]; |
5153 |
|
|
#define FASTDEF(x) float_conv x |
5154 |
|
|
// add (1<<23) to convert to int, then divide by 2^SHIFT, then add 0.5/2^SHIFT to round |
5155 |
|
|
#define MAGIC(SHIFT) (1.5f * (1 << (23-SHIFT)) + 0.5f/(1 << SHIFT)) |
5156 |
|
|
#define ADDEND(SHIFT) (((150-SHIFT) << 23) + (1 << 22)) |
5157 |
|
|
#define FAST_SCALED_FLOAT_TO_INT(temp,x,s) (temp.f = (x) + MAGIC(s), temp.i - ADDEND(s)) |
5158 |
|
|
#define check_endianness() |
5159 |
|
|
#else |
5160 |
|
|
#define FAST_SCALED_FLOAT_TO_INT(temp,x,s) ((int) ((x) * (1 << (s)))) |
5161 |
|
|
#define check_endianness() |
5162 |
|
|
#define FASTDEF(x) |
5163 |
|
|
#endif |
5164 |
|
|
|
5165 |
|
|
static void copy_samples(short *dest, float *src, int len) |
5166 |
|
|
{ |
5167 |
|
|
int i; |
5168 |
|
|
check_endianness(); |
5169 |
|
✗ |
for (i=0; i < len; ++i) { |
5170 |
|
|
FASTDEF(temp); |
5171 |
|
✗ |
int v = FAST_SCALED_FLOAT_TO_INT(temp, src[i],15); |
5172 |
|
✗ |
if ((unsigned int) (v + 32768) > 65535) |
5173 |
|
✗ |
v = v < 0 ? -32768 : 32767; |
5174 |
|
✗ |
dest[i] = v; |
5175 |
|
|
} |
5176 |
|
|
} |
5177 |
|
|
|
5178 |
|
✗ |
static void compute_samples(int mask, short *output, int num_c, float **data, int d_offset, int len) |
5179 |
|
|
{ |
5180 |
|
|
#define STB_BUFFER_SIZE 32 |
5181 |
|
|
float buffer[STB_BUFFER_SIZE]; |
5182 |
|
|
int i,j,o,n = STB_BUFFER_SIZE; |
5183 |
|
|
check_endianness(); |
5184 |
|
✗ |
for (o = 0; o < len; o += STB_BUFFER_SIZE) { |
5185 |
|
|
memset(buffer, 0, sizeof(buffer)); |
5186 |
|
✗ |
if (o + n > len) n = len - o; |
5187 |
|
✗ |
for (j=0; j < num_c; ++j) { |
5188 |
|
✗ |
if (channel_position[num_c][j] & mask) { |
5189 |
|
✗ |
for (i=0; i < n; ++i) |
5190 |
|
✗ |
buffer[i] += data[j][d_offset+o+i]; |
5191 |
|
|
} |
5192 |
|
|
} |
5193 |
|
✗ |
for (i=0; i < n; ++i) { |
5194 |
|
|
FASTDEF(temp); |
5195 |
|
✗ |
int v = FAST_SCALED_FLOAT_TO_INT(temp,buffer[i],15); |
5196 |
|
✗ |
if ((unsigned int) (v + 32768) > 65535) |
5197 |
|
✗ |
v = v < 0 ? -32768 : 32767; |
5198 |
|
✗ |
output[o+i] = v; |
5199 |
|
|
} |
5200 |
|
|
} |
5201 |
|
|
#undef STB_BUFFER_SIZE |
5202 |
|
|
} |
5203 |
|
|
|
5204 |
|
✗ |
static void compute_stereo_samples(short *output, int num_c, float **data, int d_offset, int len) |
5205 |
|
|
{ |
5206 |
|
|
#define STB_BUFFER_SIZE 32 |
5207 |
|
|
float buffer[STB_BUFFER_SIZE]; |
5208 |
|
|
int i,j,o,n = STB_BUFFER_SIZE >> 1; |
5209 |
|
|
// o is the offset in the source data |
5210 |
|
|
check_endianness(); |
5211 |
|
✗ |
for (o = 0; o < len; o += STB_BUFFER_SIZE >> 1) { |
5212 |
|
|
// o2 is the offset in the output data |
5213 |
|
✗ |
int o2 = o << 1; |
5214 |
|
|
memset(buffer, 0, sizeof(buffer)); |
5215 |
|
✗ |
if (o + n > len) n = len - o; |
5216 |
|
✗ |
for (j=0; j < num_c; ++j) { |
5217 |
|
✗ |
int m = channel_position[num_c][j] & (PLAYBACK_LEFT | PLAYBACK_RIGHT); |
5218 |
|
✗ |
if (m == (PLAYBACK_LEFT | PLAYBACK_RIGHT)) { |
5219 |
|
✗ |
for (i=0; i < n; ++i) { |
5220 |
|
✗ |
buffer[i*2+0] += data[j][d_offset+o+i]; |
5221 |
|
✗ |
buffer[i*2+1] += data[j][d_offset+o+i]; |
5222 |
|
|
} |
5223 |
|
✗ |
} else if (m == PLAYBACK_LEFT) { |
5224 |
|
✗ |
for (i=0; i < n; ++i) { |
5225 |
|
✗ |
buffer[i*2+0] += data[j][d_offset+o+i]; |
5226 |
|
|
} |
5227 |
|
✗ |
} else if (m == PLAYBACK_RIGHT) { |
5228 |
|
✗ |
for (i=0; i < n; ++i) { |
5229 |
|
✗ |
buffer[i*2+1] += data[j][d_offset+o+i]; |
5230 |
|
|
} |
5231 |
|
|
} |
5232 |
|
|
} |
5233 |
|
✗ |
for (i=0; i < (n<<1); ++i) { |
5234 |
|
|
FASTDEF(temp); |
5235 |
|
✗ |
int v = FAST_SCALED_FLOAT_TO_INT(temp,buffer[i],15); |
5236 |
|
✗ |
if ((unsigned int) (v + 32768) > 65535) |
5237 |
|
✗ |
v = v < 0 ? -32768 : 32767; |
5238 |
|
✗ |
output[o2+i] = v; |
5239 |
|
|
} |
5240 |
|
|
} |
5241 |
|
|
#undef STB_BUFFER_SIZE |
5242 |
|
|
} |
5243 |
|
|
|
5244 |
|
✗ |
static void convert_samples_short(int buf_c, short **buffer, int b_offset, int data_c, float **data, int d_offset, int samples) |
5245 |
|
|
{ |
5246 |
|
|
int i; |
5247 |
|
✗ |
if (buf_c != data_c && buf_c <= 2 && data_c <= 6) { |
5248 |
|
|
static int channel_selector[3][2] = { {0}, {PLAYBACK_MONO}, {PLAYBACK_LEFT, PLAYBACK_RIGHT} }; |
5249 |
|
✗ |
for (i=0; i < buf_c; ++i) |
5250 |
|
✗ |
compute_samples(channel_selector[buf_c][i], buffer[i]+b_offset, data_c, data, d_offset, samples); |
5251 |
|
|
} else { |
5252 |
|
✗ |
int limit = buf_c < data_c ? buf_c : data_c; |
5253 |
|
✗ |
for (i=0; i < limit; ++i) |
5254 |
|
✗ |
copy_samples(buffer[i]+b_offset, data[i]+d_offset, samples); |
5255 |
|
✗ |
for ( ; i < buf_c; ++i) |
5256 |
|
✗ |
memset(buffer[i]+b_offset, 0, sizeof(short) * samples); |
5257 |
|
|
} |
5258 |
|
|
} |
5259 |
|
|
|
5260 |
|
✗ |
int stb_vorbis_get_frame_short(stb_vorbis *f, int num_c, short **buffer, int num_samples) |
5261 |
|
|
{ |
5262 |
|
✗ |
float **output = NULL; |
5263 |
|
✗ |
int len = stb_vorbis_get_frame_float(f, NULL, &output); |
5264 |
|
|
if (len > num_samples) len = num_samples; |
5265 |
|
✗ |
if (len) |
5266 |
|
✗ |
convert_samples_short(num_c, buffer, 0, f->channels, output, 0, len); |
5267 |
|
✗ |
return len; |
5268 |
|
|
} |
5269 |
|
|
|
5270 |
|
✗ |
static void convert_channels_short_interleaved(int buf_c, short *buffer, int data_c, float **data, int d_offset, int len) |
5271 |
|
|
{ |
5272 |
|
|
int i; |
5273 |
|
|
check_endianness(); |
5274 |
|
✗ |
if (buf_c != data_c && buf_c <= 2 && data_c <= 6) { |
5275 |
|
|
assert(buf_c == 2); |
5276 |
|
✗ |
for (i=0; i < buf_c; ++i) |
5277 |
|
✗ |
compute_stereo_samples(buffer, data_c, data, d_offset, len); |
5278 |
|
|
} else { |
5279 |
|
✗ |
int limit = buf_c < data_c ? buf_c : data_c; |
5280 |
|
|
int j; |
5281 |
|
✗ |
for (j=0; j < len; ++j) { |
5282 |
|
✗ |
for (i=0; i < limit; ++i) { |
5283 |
|
|
FASTDEF(temp); |
5284 |
|
✗ |
float f = data[i][d_offset+j]; |
5285 |
|
✗ |
int v = FAST_SCALED_FLOAT_TO_INT(temp, f,15);//data[i][d_offset+j],15); |
5286 |
|
✗ |
if ((unsigned int) (v + 32768) > 65535) |
5287 |
|
✗ |
v = v < 0 ? -32768 : 32767; |
5288 |
|
✗ |
*buffer++ = v; |
5289 |
|
|
} |
5290 |
|
✗ |
for ( ; i < buf_c; ++i) |
5291 |
|
✗ |
*buffer++ = 0; |
5292 |
|
|
} |
5293 |
|
|
} |
5294 |
|
|
} |
5295 |
|
|
|
5296 |
|
✗ |
int stb_vorbis_get_frame_short_interleaved(stb_vorbis *f, int num_c, short *buffer, int num_shorts) |
5297 |
|
|
{ |
5298 |
|
|
float **output; |
5299 |
|
|
int len; |
5300 |
|
✗ |
if (num_c == 1) return stb_vorbis_get_frame_short(f,num_c,&buffer, num_shorts); |
5301 |
|
✗ |
len = stb_vorbis_get_frame_float(f, NULL, &output); |
5302 |
|
✗ |
if (len) { |
5303 |
|
✗ |
if (len*num_c > num_shorts) len = num_shorts / num_c; |
5304 |
|
✗ |
convert_channels_short_interleaved(num_c, buffer, f->channels, output, 0, len); |
5305 |
|
|
} |
5306 |
|
|
return len; |
5307 |
|
|
} |
5308 |
|
|
|
5309 |
|
✗ |
int stb_vorbis_get_samples_short_interleaved(stb_vorbis *f, int channels, short *buffer, int num_shorts) |
5310 |
|
|
{ |
5311 |
|
|
float **outputs; |
5312 |
|
✗ |
int len = num_shorts / channels; |
5313 |
|
|
int n=0; |
5314 |
|
✗ |
while (n < len) { |
5315 |
|
✗ |
int k = f->channel_buffer_end - f->channel_buffer_start; |
5316 |
|
✗ |
if (n+k >= len) k = len - n; |
5317 |
|
✗ |
if (k) |
5318 |
|
✗ |
convert_channels_short_interleaved(channels, buffer, f->channels, f->channel_buffers, f->channel_buffer_start, k); |
5319 |
|
✗ |
buffer += k*channels; |
5320 |
|
✗ |
n += k; |
5321 |
|
✗ |
f->channel_buffer_start += k; |
5322 |
|
✗ |
if (n == len) break; |
5323 |
|
✗ |
if (!stb_vorbis_get_frame_float(f, NULL, &outputs)) break; |
5324 |
|
|
} |
5325 |
|
✗ |
return n; |
5326 |
|
|
} |
5327 |
|
|
|
5328 |
|
✗ |
int stb_vorbis_get_samples_short(stb_vorbis *f, int channels, short **buffer, int len) |
5329 |
|
|
{ |
5330 |
|
|
float **outputs; |
5331 |
|
|
int n=0; |
5332 |
|
✗ |
while (n < len) { |
5333 |
|
✗ |
int k = f->channel_buffer_end - f->channel_buffer_start; |
5334 |
|
✗ |
if (n+k >= len) k = len - n; |
5335 |
|
✗ |
if (k) |
5336 |
|
✗ |
convert_samples_short(channels, buffer, n, f->channels, f->channel_buffers, f->channel_buffer_start, k); |
5337 |
|
✗ |
n += k; |
5338 |
|
✗ |
f->channel_buffer_start += k; |
5339 |
|
✗ |
if (n == len) break; |
5340 |
|
✗ |
if (!stb_vorbis_get_frame_float(f, NULL, &outputs)) break; |
5341 |
|
|
} |
5342 |
|
✗ |
return n; |
5343 |
|
|
} |
5344 |
|
|
|
5345 |
|
|
#ifndef STB_VORBIS_NO_STDIO |
5346 |
|
✗ |
int stb_vorbis_decode_filename(const char *filename, int *channels, int *sample_rate, short **output) |
5347 |
|
|
{ |
5348 |
|
|
int data_len, offset, total, limit, error; |
5349 |
|
|
short *data; |
5350 |
|
✗ |
stb_vorbis *v = stb_vorbis_open_filename(filename, &error, NULL); |
5351 |
|
✗ |
if (v == NULL) return -1; |
5352 |
|
✗ |
limit = v->channels * 4096; |
5353 |
|
✗ |
*channels = v->channels; |
5354 |
|
✗ |
if (sample_rate) |
5355 |
|
✗ |
*sample_rate = v->sample_rate; |
5356 |
|
|
offset = data_len = 0; |
5357 |
|
|
total = limit; |
5358 |
|
✗ |
data = (short *) malloc(total * sizeof(*data)); |
5359 |
|
✗ |
if (data == NULL) { |
5360 |
|
✗ |
stb_vorbis_close(v); |
5361 |
|
✗ |
return -2; |
5362 |
|
|
} |
5363 |
|
|
for (;;) { |
5364 |
|
✗ |
int n = stb_vorbis_get_frame_short_interleaved(v, v->channels, data+offset, total-offset); |
5365 |
|
✗ |
if (n == 0) break; |
5366 |
|
✗ |
data_len += n; |
5367 |
|
✗ |
offset += n * v->channels; |
5368 |
|
✗ |
if (offset + limit > total) { |
5369 |
|
|
short *data2; |
5370 |
|
✗ |
total *= 2; |
5371 |
|
✗ |
data2 = (short *) realloc(data, total * sizeof(*data)); |
5372 |
|
✗ |
if (data2 == NULL) { |
5373 |
|
✗ |
free(data); |
5374 |
|
✗ |
stb_vorbis_close(v); |
5375 |
|
✗ |
return -2; |
5376 |
|
|
} |
5377 |
|
|
data = data2; |
5378 |
|
|
} |
5379 |
|
|
} |
5380 |
|
✗ |
*output = data; |
5381 |
|
✗ |
stb_vorbis_close(v); |
5382 |
|
✗ |
return data_len; |
5383 |
|
|
} |
5384 |
|
|
#endif // NO_STDIO |
5385 |
|
|
|
5386 |
|
✗ |
int stb_vorbis_decode_memory(const uint8 *mem, int len, int *channels, int *sample_rate, short **output) |
5387 |
|
|
{ |
5388 |
|
|
int data_len, offset, total, limit, error; |
5389 |
|
|
short *data; |
5390 |
|
✗ |
stb_vorbis *v = stb_vorbis_open_memory(mem, len, &error, NULL); |
5391 |
|
✗ |
if (v == NULL) return -1; |
5392 |
|
✗ |
limit = v->channels * 4096; |
5393 |
|
✗ |
*channels = v->channels; |
5394 |
|
✗ |
if (sample_rate) |
5395 |
|
✗ |
*sample_rate = v->sample_rate; |
5396 |
|
|
offset = data_len = 0; |
5397 |
|
|
total = limit; |
5398 |
|
✗ |
data = (short *) malloc(total * sizeof(*data)); |
5399 |
|
✗ |
if (data == NULL) { |
5400 |
|
✗ |
stb_vorbis_close(v); |
5401 |
|
✗ |
return -2; |
5402 |
|
|
} |
5403 |
|
|
for (;;) { |
5404 |
|
✗ |
int n = stb_vorbis_get_frame_short_interleaved(v, v->channels, data+offset, total-offset); |
5405 |
|
✗ |
if (n == 0) break; |
5406 |
|
✗ |
data_len += n; |
5407 |
|
✗ |
offset += n * v->channels; |
5408 |
|
✗ |
if (offset + limit > total) { |
5409 |
|
|
short *data2; |
5410 |
|
✗ |
total *= 2; |
5411 |
|
✗ |
data2 = (short *) realloc(data, total * sizeof(*data)); |
5412 |
|
✗ |
if (data2 == NULL) { |
5413 |
|
✗ |
free(data); |
5414 |
|
✗ |
stb_vorbis_close(v); |
5415 |
|
✗ |
return -2; |
5416 |
|
|
} |
5417 |
|
|
data = data2; |
5418 |
|
|
} |
5419 |
|
|
} |
5420 |
|
✗ |
*output = data; |
5421 |
|
✗ |
stb_vorbis_close(v); |
5422 |
|
✗ |
return data_len; |
5423 |
|
|
} |
5424 |
|
|
#endif // STB_VORBIS_NO_INTEGER_CONVERSION |
5425 |
|
|
|
5426 |
|
✗ |
int stb_vorbis_get_samples_float_interleaved(stb_vorbis *f, int channels, float *buffer, int num_floats) |
5427 |
|
|
{ |
5428 |
|
|
float **outputs; |
5429 |
|
✗ |
int len = num_floats / channels; |
5430 |
|
|
int n=0; |
5431 |
|
✗ |
int z = f->channels; |
5432 |
|
|
if (z > channels) z = channels; |
5433 |
|
✗ |
while (n < len) { |
5434 |
|
|
int i,j; |
5435 |
|
✗ |
int k = f->channel_buffer_end - f->channel_buffer_start; |
5436 |
|
✗ |
if (n+k >= len) k = len - n; |
5437 |
|
✗ |
for (j=0; j < k; ++j) { |
5438 |
|
✗ |
for (i=0; i < z; ++i) |
5439 |
|
✗ |
*buffer++ = f->channel_buffers[i][f->channel_buffer_start+j]; |
5440 |
|
✗ |
for ( ; i < channels; ++i) |
5441 |
|
✗ |
*buffer++ = 0; |
5442 |
|
|
} |
5443 |
|
✗ |
n += k; |
5444 |
|
✗ |
f->channel_buffer_start += k; |
5445 |
|
✗ |
if (n == len) |
5446 |
|
|
break; |
5447 |
|
✗ |
if (!stb_vorbis_get_frame_float(f, NULL, &outputs)) |
5448 |
|
|
break; |
5449 |
|
|
} |
5450 |
|
✗ |
return n; |
5451 |
|
|
} |
5452 |
|
|
|
5453 |
|
✗ |
int stb_vorbis_get_samples_float(stb_vorbis *f, int channels, float **buffer, int num_samples) |
5454 |
|
|
{ |
5455 |
|
|
float **outputs; |
5456 |
|
|
int n=0; |
5457 |
|
✗ |
int z = f->channels; |
5458 |
|
|
if (z > channels) z = channels; |
5459 |
|
✗ |
while (n < num_samples) { |
5460 |
|
|
int i; |
5461 |
|
✗ |
int k = f->channel_buffer_end - f->channel_buffer_start; |
5462 |
|
✗ |
if (n+k >= num_samples) k = num_samples - n; |
5463 |
|
✗ |
if (k) { |
5464 |
|
✗ |
for (i=0; i < z; ++i) |
5465 |
|
✗ |
memcpy(buffer[i]+n, f->channel_buffers[i]+f->channel_buffer_start, sizeof(float)*k); |
5466 |
|
✗ |
for ( ; i < channels; ++i) |
5467 |
|
✗ |
memset(buffer[i]+n, 0, sizeof(float) * k); |
5468 |
|
|
} |
5469 |
|
✗ |
n += k; |
5470 |
|
✗ |
f->channel_buffer_start += k; |
5471 |
|
✗ |
if (n == num_samples) |
5472 |
|
|
break; |
5473 |
|
✗ |
if (!stb_vorbis_get_frame_float(f, NULL, &outputs)) |
5474 |
|
|
break; |
5475 |
|
|
} |
5476 |
|
✗ |
return n; |
5477 |
|
|
} |
5478 |
|
|
#endif // STB_VORBIS_NO_PULLDATA_API |
5479 |
|
|
|
5480 |
|
|
/* Version history |
5481 |
|
|
1.17 - 2019-07-08 - fix CVE-2019-13217, -13218, -13219, -13220, -13221, -13222, -13223 |
5482 |
|
|
found with Mayhem by ForAllSecure |
5483 |
|
|
1.16 - 2019-03-04 - fix warnings |
5484 |
|
|
1.15 - 2019-02-07 - explicit failure if Ogg Skeleton data is found |
5485 |
|
|
1.14 - 2018-02-11 - delete bogus dealloca usage |
5486 |
|
|
1.13 - 2018-01-29 - fix truncation of last frame (hopefully) |
5487 |
|
|
1.12 - 2017-11-21 - limit residue begin/end to blocksize/2 to avoid large temp allocs in bad/corrupt files |
5488 |
|
|
1.11 - 2017-07-23 - fix MinGW compilation |
5489 |
|
|
1.10 - 2017-03-03 - more robust seeking; fix negative ilog(); clear error in open_memory |
5490 |
|
|
1.09 - 2016-04-04 - back out 'avoid discarding last frame' fix from previous version |
5491 |
|
|
1.08 - 2016-04-02 - fixed multiple warnings; fix setup memory leaks; |
5492 |
|
|
avoid discarding last frame of audio data |
5493 |
|
|
1.07 - 2015-01-16 - fixed some warnings, fix mingw, const-correct API |
5494 |
|
|
some more crash fixes when out of memory or with corrupt files |
5495 |
|
|
1.06 - 2015-08-31 - full, correct support for seeking API (Dougall Johnson) |
5496 |
|
|
some crash fixes when out of memory or with corrupt files |
5497 |
|
|
1.05 - 2015-04-19 - don't define __forceinline if it's redundant |
5498 |
|
|
1.04 - 2014-08-27 - fix missing const-correct case in API |
5499 |
|
|
1.03 - 2014-08-07 - Warning fixes |
5500 |
|
|
1.02 - 2014-07-09 - Declare qsort compare function _cdecl on windows |
5501 |
|
|
1.01 - 2014-06-18 - fix stb_vorbis_get_samples_float |
5502 |
|
|
1.0 - 2014-05-26 - fix memory leaks; fix warnings; fix bugs in multichannel |
5503 |
|
|
(API change) report sample rate for decode-full-file funcs |
5504 |
|
|
0.99996 - bracket #include <malloc.h> for macintosh compilation by Laurent Gomila |
5505 |
|
|
0.99995 - use union instead of pointer-cast for fast-float-to-int to avoid alias-optimization problem |
5506 |
|
|
0.99994 - change fast-float-to-int to work in single-precision FPU mode, remove endian-dependence |
5507 |
|
|
0.99993 - remove assert that fired on legal files with empty tables |
5508 |
|
|
0.99992 - rewind-to-start |
5509 |
|
|
0.99991 - bugfix to stb_vorbis_get_samples_short by Bernhard Wodo |
5510 |
|
|
0.9999 - (should have been 0.99990) fix no-CRT support, compiling as C++ |
5511 |
|
|
0.9998 - add a full-decode function with a memory source |
5512 |
|
|
0.9997 - fix a bug in the read-from-FILE case in 0.9996 addition |
5513 |
|
|
0.9996 - query length of vorbis stream in samples/seconds |
5514 |
|
|
0.9995 - bugfix to another optimization that only happened in certain files |
5515 |
|
|
0.9994 - bugfix to one of the optimizations that caused significant (but inaudible?) errors |
5516 |
|
|
0.9993 - performance improvements; runs in 99% to 104% of time of reference implementation |
5517 |
|
|
0.9992 - performance improvement of IMDCT; now performs close to reference implementation |
5518 |
|
|
0.9991 - performance improvement of IMDCT |
5519 |
|
|
0.999 - (should have been 0.9990) performance improvement of IMDCT |
5520 |
|
|
0.998 - no-CRT support from Casey Muratori |
5521 |
|
|
0.997 - bugfixes for bugs found by Terje Mathisen |
5522 |
|
|
0.996 - bugfix: fast-huffman decode initialized incorrectly for sparse codebooks; fixing gives 10% speedup - found by Terje Mathisen |
5523 |
|
|
0.995 - bugfix: fix to 'effective' overrun detection - found by Terje Mathisen |
5524 |
|
|
0.994 - bugfix: garbage decode on final VQ symbol of a non-multiple - found by Terje Mathisen |
5525 |
|
|
0.993 - bugfix: pushdata API required 1 extra byte for empty page (failed to consume final page if empty) - found by Terje Mathisen |
5526 |
|
|
0.992 - fixes for MinGW warning |
5527 |
|
|
0.991 - turn fast-float-conversion on by default |
5528 |
|
|
0.990 - fix push-mode seek recovery if you seek into the headers |
5529 |
|
|
0.98b - fix to bad release of 0.98 |
5530 |
|
|
0.98 - fix push-mode seek recovery; robustify float-to-int and support non-fast mode |
5531 |
|
|
0.97 - builds under c++ (typecasting, don't use 'class' keyword) |
5532 |
|
|
0.96 - somehow MY 0.95 was right, but the web one was wrong, so here's my 0.95 rereleased as 0.96, fixes a typo in the clamping code |
5533 |
|
|
0.95 - clamping code for 16-bit functions |
5534 |
|
|
0.94 - not publically released |
5535 |
|
|
0.93 - fixed all-zero-floor case (was decoding garbage) |
5536 |
|
|
0.92 - fixed a memory leak |
5537 |
|
|
0.91 - conditional compiles to omit parts of the API and the infrastructure to support them: STB_VORBIS_NO_PULLDATA_API, STB_VORBIS_NO_PUSHDATA_API, STB_VORBIS_NO_STDIO, STB_VORBIS_NO_INTEGER_CONVERSION |
5538 |
|
|
0.90 - first public release |
5539 |
|
|
*/ |
5540 |
|
|
|
5541 |
|
|
#endif // STB_VORBIS_HEADER_ONLY |
5542 |
|
|
|
5543 |
|
|
|
5544 |
|
|
/* |
5545 |
|
|
------------------------------------------------------------------------------ |
5546 |
|
|
This software is available under 2 licenses -- choose whichever you prefer. |
5547 |
|
|
------------------------------------------------------------------------------ |
5548 |
|
|
ALTERNATIVE A - MIT License |
5549 |
|
|
Copyright (c) 2017 Sean Barrett |
5550 |
|
|
Permission is hereby granted, free of charge, to any person obtaining a copy of |
5551 |
|
|
this software and associated documentation files (the "Software"), to deal in |
5552 |
|
|
the Software without restriction, including without limitation the rights to |
5553 |
|
|
use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies |
5554 |
|
|
of the Software, and to permit persons to whom the Software is furnished to do |
5555 |
|
|
so, subject to the following conditions: |
5556 |
|
|
The above copyright notice and this permission notice shall be included in all |
5557 |
|
|
copies or substantial portions of the Software. |
5558 |
|
|
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
5559 |
|
|
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
5560 |
|
|
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
5561 |
|
|
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
5562 |
|
|
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, |
5563 |
|
|
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE |
5564 |
|
|
SOFTWARE. |
5565 |
|
|
------------------------------------------------------------------------------ |
5566 |
|
|
ALTERNATIVE B - Public Domain (www.unlicense.org) |
5567 |
|
|
This is free and unencumbered software released into the public domain. |
5568 |
|
|
Anyone is free to copy, modify, publish, use, compile, sell, or distribute this |
5569 |
|
|
software, either in source code form or as a compiled binary, for any purpose, |
5570 |
|
|
commercial or non-commercial, and by any means. |
5571 |
|
|
In jurisdictions that recognize copyright laws, the author or authors of this |
5572 |
|
|
software dedicate any and all copyright interest in the software to the public |
5573 |
|
|
domain. We make this dedication for the benefit of the public at large and to |
5574 |
|
|
the detriment of our heirs and successors. We intend this dedication to be an |
5575 |
|
|
overt act of relinquishment in perpetuity of all present and future rights to |
5576 |
|
|
this software under copyright law. |
5577 |
|
|
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
5578 |
|
|
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
5579 |
|
|
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
5580 |
|
|
AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN |
5581 |
|
|
ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION |
5582 |
|
|
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. |
5583 |
|
|
------------------------------------------------------------------------------ |
5584 |
|
|
*/ |
5585 |
|
|
|