matroskadec: allow RealAudio/Cook/Sipro streams of flavor 0
[libav.git] / libavformat / matroskadec.c
CommitLineData
b061d892 1/*
ff33c5c5 2 * Matroska file demuxer
2912e87a 3 * Copyright (c) 2003-2008 The Libav Project
b061d892 4 *
2912e87a 5 * This file is part of Libav.
b061d892 6 *
2912e87a 7 * Libav is free software; you can redistribute it and/or
b061d892
DC
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
2912e87a 12 * Libav is distributed in the hope that it will be useful,
b061d892
DC
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
2912e87a 18 * License along with Libav; if not, write to the Free Software
b061d892
DC
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22/**
ba87f080 23 * @file
b061d892 24 * Matroska file demuxer
e873c03a
DB
25 * @author Ronald Bultje <rbultje@ronald.bitfreak.net>
26 * @author with a little help from Moritz Bunkus <moritz@bunkus.org>
27 * @author totally reworked by Aurelien Jacobs <aurel@gnuage.org>
28 * @see specs available on the Matroska project page: http://www.matroska.org/
b061d892
DC
29 */
30
84cfce9f
KC
31#include "config.h"
32
d92024f1 33#include <inttypes.h>
3eb9bfbb 34#include <stdio.h>
84cfce9f
KC
35#if CONFIG_BZLIB
36#include <bzlib.h>
37#endif
38#if CONFIG_ZLIB
39#include <zlib.h>
40#endif
41
42#include "libavutil/avstring.h"
43#include "libavutil/dict.h"
44#include "libavutil/intfloat.h"
45#include "libavutil/intreadwrite.h"
46#include "libavutil/lzo.h"
eb3b5501 47#include "libavutil/mathematics.h"
84cfce9f
KC
48
49#include "libavcodec/bytestream.h"
4efdadc8 50#include "libavcodec/flac.h"
84cfce9f
KC
51#include "libavcodec/mpeg4audio.h"
52
b061d892 53#include "avformat.h"
e731b8d8 54#include "avio_internal.h"
84cfce9f
KC
55#include "internal.h"
56#include "isom.h"
57#include "matroska.h"
23f741f7 58#include "oggdec.h"
1a40491e 59/* For ff_codec_get_id(). */
b061d892 60#include "riff.h"
7e68c91e 61#include "rmsipr.h"
b061d892 62
789ed100
AJ
63typedef enum {
64 EBML_NONE,
65 EBML_UINT,
66 EBML_FLOAT,
67 EBML_STR,
68 EBML_UTF8,
69 EBML_BIN,
70 EBML_NEST,
71 EBML_PASS,
72 EBML_STOP,
95ec3d4c 73 EBML_TYPE_COUNT
789ed100
AJ
74} EbmlType;
75
76typedef const struct EbmlSyntax {
77 uint32_t id;
78 EbmlType type;
79 int list_elem_size;
80 int data_offset;
81 union {
82 uint64_t u;
83 double f;
84 const char *s;
85 const struct EbmlSyntax *n;
86 } def;
87} EbmlSyntax;
88
89typedef struct {
90 int nb_elem;
91 void *elem;
92} EbmlList;
93
94typedef struct {
95 int size;
96 uint8_t *data;
97 int64_t pos;
98} EbmlBin;
99
63511324
AJ
100typedef struct {
101 uint64_t version;
102 uint64_t max_size;
103 uint64_t id_length;
104 char *doctype;
105 uint64_t doctype_version;
106} Ebml;
107
2cbc8811
AJ
108typedef struct {
109 uint64_t algo;
110 EbmlBin settings;
111} MatroskaTrackCompression;
b061d892 112
2cbc8811
AJ
113typedef struct {
114 uint64_t scope;
115 uint64_t type;
116 MatroskaTrackCompression compression;
117} MatroskaTrackEncoding;
b061d892 118
2cbc8811
AJ
119typedef struct {
120 double frame_rate;
121 uint64_t display_width;
122 uint64_t display_height;
123 uint64_t pixel_width;
124 uint64_t pixel_height;
125 uint64_t fourcc;
152e09fd 126 uint64_t stereo_mode;
2cbc8811 127} MatroskaTrackVideo;
b061d892 128
2cbc8811
AJ
129typedef struct {
130 double samplerate;
131 double out_samplerate;
132 uint64_t bitdepth;
133 uint64_t channels;
134
135 /* real audio header (extracted from extradata) */
136 int coded_framesize;
137 int sub_packet_h;
138 int frame_size;
139 int sub_packet_size;
140 int sub_packet_cnt;
141 int pkt_cnt;
46544204 142 uint64_t buf_timecode;
2cbc8811
AJ
143 uint8_t *buf;
144} MatroskaTrackAudio;
b061d892 145
2cbc8811
AJ
146typedef struct {
147 uint64_t num;
325ace3e 148 uint64_t uid;
2cbc8811 149 uint64_t type;
38766e08 150 char *name;
2cbc8811
AJ
151 char *codec_id;
152 EbmlBin codec_priv;
153 char *language;
7ff97085 154 double time_scale;
b061d892 155 uint64_t default_duration;
4eff9743 156 uint64_t flag_default;
7a617a86 157 uint64_t flag_forced;
2cbc8811
AJ
158 MatroskaTrackVideo video;
159 MatroskaTrackAudio audio;
160 EbmlList encodings;
eb3b5501 161 uint64_t codec_delay;
fc4d335f
AJ
162
163 AVStream *stream;
82360e67 164 int64_t end_timecode;
3e93c8ed 165 int ms_compat;
b061d892
DC
166} MatroskaTrack;
167
e5929fdf 168typedef struct {
325ace3e 169 uint64_t uid;
b414cb89
AJ
170 char *filename;
171 char *mime;
172 EbmlBin bin;
929e9de7
AJ
173
174 AVStream *stream;
f69befe5 175} MatroskaAttachment;
b414cb89
AJ
176
177typedef struct {
6bbd7c7b
AJ
178 uint64_t start;
179 uint64_t end;
180 uint64_t uid;
181 char *title;
6cb6e159
AJ
182
183 AVChapter *chapter;
6bbd7c7b
AJ
184} MatroskaChapter;
185
186typedef struct {
e5929fdf
AJ
187 uint64_t track;
188 uint64_t pos;
189} MatroskaIndexPos;
190
191typedef struct {
192 uint64_t time;
193 EbmlList pos;
194} MatroskaIndex;
195
13b350a3 196typedef struct {
44015c56
AJ
197 char *name;
198 char *string;
f702df39
AJ
199 char *lang;
200 uint64_t def;
44015c56
AJ
201 EbmlList sub;
202} MatroskaTag;
203
204typedef struct {
929e9de7
AJ
205 char *type;
206 uint64_t typevalue;
207 uint64_t trackuid;
208 uint64_t chapteruid;
209 uint64_t attachuid;
210} MatroskaTagTarget;
211
212typedef struct {
213 MatroskaTagTarget target;
214 EbmlList tag;
215} MatroskaTags;
216
217typedef struct {
13b350a3
AJ
218 uint64_t id;
219 uint64_t pos;
220} MatroskaSeekhead;
221
c171af9b 222typedef struct {
8d75b5a2
AJ
223 uint64_t start;
224 uint64_t length;
b061d892
DC
225} MatroskaLevel;
226
c171af9b 227typedef struct {
8336eb6f
DC
228 uint64_t timecode;
229 EbmlList blocks;
230} MatroskaCluster;
231
232typedef struct {
b061d892
DC
233 AVFormatContext *ctx;
234
5968d2dd 235 /* EBML stuff */
b061d892
DC
236 int num_levels;
237 MatroskaLevel levels[EBML_MAX_DEPTH];
238 int level_up;
c3ade62c 239 uint32_t current_id;
b061d892 240
29708581
AJ
241 uint64_t time_scale;
242 double duration;
243 char *title;
2cbc8811 244 EbmlList tracks;
b414cb89 245 EbmlList attachments;
6bbd7c7b 246 EbmlList chapters;
e5929fdf 247 EbmlList index;
44015c56 248 EbmlList tags;
13b350a3 249 EbmlList seekhead;
b061d892 250
b061d892 251 /* byte position of the segment inside the stream */
bc5c918e 252 int64_t segment_start;
b061d892 253
5968d2dd 254 /* the packet queue */
b061d892
DC
255 AVPacket **packets;
256 int num_packets;
d5e34dcb 257 AVPacket *prev_pkt;
b061d892 258
8d75b5a2 259 int done;
b061d892 260
b061d892
DC
261 /* What to skip before effectively reading a packet. */
262 int skip_to_keyframe;
20f74665 263 uint64_t skip_to_timecode;
31ad14c2
AC
264
265 /* File has a CUES element, but we defer parsing until it is needed. */
266 int cues_parsing_deferred;
8336eb6f
DC
267
268 int current_cluster_num_blocks;
269 int64_t current_cluster_pos;
270 MatroskaCluster current_cluster;
271
272 /* File has SSA subtitles which prevent incremental cluster parsing. */
273 int contains_ssa;
b061d892
DC
274} MatroskaDemuxContext;
275
209472b4
AJ
276typedef struct {
277 uint64_t duration;
278 int64_t reference;
194d4b49 279 uint64_t non_simple;
209472b4
AJ
280 EbmlBin bin;
281} MatroskaBlock;
282
63511324 283static EbmlSyntax ebml_header[] = {
84cfce9f
KC
284 { EBML_ID_EBMLREADVERSION, EBML_UINT, 0, offsetof(Ebml, version), { .u = EBML_VERSION } },
285 { EBML_ID_EBMLMAXSIZELENGTH, EBML_UINT, 0, offsetof(Ebml, max_size), { .u = 8 } },
286 { EBML_ID_EBMLMAXIDLENGTH, EBML_UINT, 0, offsetof(Ebml, id_length), { .u = 4 } },
287 { EBML_ID_DOCTYPE, EBML_STR, 0, offsetof(Ebml, doctype), { .s = "(none)" } },
288 { EBML_ID_DOCTYPEREADVERSION, EBML_UINT, 0, offsetof(Ebml, doctype_version), { .u = 1 } },
289 { EBML_ID_EBMLVERSION, EBML_NONE },
290 { EBML_ID_DOCTYPEVERSION, EBML_NONE },
63511324
AJ
291 { 0 }
292};
293
294static EbmlSyntax ebml_syntax[] = {
84cfce9f 295 { EBML_ID_HEADER, EBML_NEST, 0, 0, { .n = ebml_header } },
63511324
AJ
296 { 0 }
297};
298
29708581 299static EbmlSyntax matroska_info[] = {
84cfce9f
KC
300 { MATROSKA_ID_TIMECODESCALE, EBML_UINT, 0, offsetof(MatroskaDemuxContext, time_scale), { .u = 1000000 } },
301 { MATROSKA_ID_DURATION, EBML_FLOAT, 0, offsetof(MatroskaDemuxContext, duration) },
302 { MATROSKA_ID_TITLE, EBML_UTF8, 0, offsetof(MatroskaDemuxContext, title) },
303 { MATROSKA_ID_WRITINGAPP, EBML_NONE },
304 { MATROSKA_ID_MUXINGAPP, EBML_NONE },
305 { MATROSKA_ID_DATEUTC, EBML_NONE },
306 { MATROSKA_ID_SEGMENTUID, EBML_NONE },
29708581
AJ
307 { 0 }
308};
309
2cbc8811 310static EbmlSyntax matroska_track_video[] = {
84cfce9f
KC
311 { MATROSKA_ID_VIDEOFRAMERATE, EBML_FLOAT, 0, offsetof(MatroskaTrackVideo, frame_rate) },
312 { MATROSKA_ID_VIDEODISPLAYWIDTH, EBML_UINT, 0, offsetof(MatroskaTrackVideo, display_width) },
313 { MATROSKA_ID_VIDEODISPLAYHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo, display_height) },
314 { MATROSKA_ID_VIDEOPIXELWIDTH, EBML_UINT, 0, offsetof(MatroskaTrackVideo, pixel_width) },
315 { MATROSKA_ID_VIDEOPIXELHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo, pixel_height) },
316 { MATROSKA_ID_VIDEOCOLORSPACE, EBML_UINT, 0, offsetof(MatroskaTrackVideo, fourcc) },
317 { MATROSKA_ID_VIDEOPIXELCROPB, EBML_NONE },
318 { MATROSKA_ID_VIDEOPIXELCROPT, EBML_NONE },
319 { MATROSKA_ID_VIDEOPIXELCROPL, EBML_NONE },
320 { MATROSKA_ID_VIDEOPIXELCROPR, EBML_NONE },
321 { MATROSKA_ID_VIDEODISPLAYUNIT, EBML_NONE },
322 { MATROSKA_ID_VIDEOFLAGINTERLACED, EBML_NONE },
152e09fd 323 { MATROSKA_ID_VIDEOSTEREOMODE, EBML_UINT, 0, offsetof(MatroskaTrackVideo, stereo_mode), { .u = MATROSKA_VIDEO_STEREOMODE_TYPE_NB } },
84cfce9f 324 { MATROSKA_ID_VIDEOASPECTRATIO, EBML_NONE },
2cbc8811
AJ
325 { 0 }
326};
327
328static EbmlSyntax matroska_track_audio[] = {
84cfce9f
KC
329 { MATROSKA_ID_AUDIOSAMPLINGFREQ, EBML_FLOAT, 0, offsetof(MatroskaTrackAudio, samplerate), { .f = 8000.0 } },
330 { MATROSKA_ID_AUDIOOUTSAMPLINGFREQ, EBML_FLOAT, 0, offsetof(MatroskaTrackAudio, out_samplerate) },
331 { MATROSKA_ID_AUDIOBITDEPTH, EBML_UINT, 0, offsetof(MatroskaTrackAudio, bitdepth) },
332 { MATROSKA_ID_AUDIOCHANNELS, EBML_UINT, 0, offsetof(MatroskaTrackAudio, channels), { .u = 1 } },
2cbc8811
AJ
333 { 0 }
334};
335
336static EbmlSyntax matroska_track_encoding_compression[] = {
84cfce9f
KC
337 { MATROSKA_ID_ENCODINGCOMPALGO, EBML_UINT, 0, offsetof(MatroskaTrackCompression, algo), { .u = 0 } },
338 { MATROSKA_ID_ENCODINGCOMPSETTINGS, EBML_BIN, 0, offsetof(MatroskaTrackCompression, settings) },
2cbc8811
AJ
339 { 0 }
340};
341
342static EbmlSyntax matroska_track_encoding[] = {
84cfce9f
KC
343 { MATROSKA_ID_ENCODINGSCOPE, EBML_UINT, 0, offsetof(MatroskaTrackEncoding, scope), { .u = 1 } },
344 { MATROSKA_ID_ENCODINGTYPE, EBML_UINT, 0, offsetof(MatroskaTrackEncoding, type), { .u = 0 } },
345 { MATROSKA_ID_ENCODINGCOMPRESSION, EBML_NEST, 0, offsetof(MatroskaTrackEncoding, compression), { .n = matroska_track_encoding_compression } },
346 { MATROSKA_ID_ENCODINGORDER, EBML_NONE },
2cbc8811
AJ
347 { 0 }
348};
349
350static EbmlSyntax matroska_track_encodings[] = {
84cfce9f 351 { MATROSKA_ID_TRACKCONTENTENCODING, EBML_NEST, sizeof(MatroskaTrackEncoding), offsetof(MatroskaTrack, encodings), { .n = matroska_track_encoding } },
2cbc8811
AJ
352 { 0 }
353};
354
355static EbmlSyntax matroska_track[] = {
84cfce9f
KC
356 { MATROSKA_ID_TRACKNUMBER, EBML_UINT, 0, offsetof(MatroskaTrack, num) },
357 { MATROSKA_ID_TRACKNAME, EBML_UTF8, 0, offsetof(MatroskaTrack, name) },
358 { MATROSKA_ID_TRACKUID, EBML_UINT, 0, offsetof(MatroskaTrack, uid) },
359 { MATROSKA_ID_TRACKTYPE, EBML_UINT, 0, offsetof(MatroskaTrack, type) },
360 { MATROSKA_ID_CODECID, EBML_STR, 0, offsetof(MatroskaTrack, codec_id) },
361 { MATROSKA_ID_CODECPRIVATE, EBML_BIN, 0, offsetof(MatroskaTrack, codec_priv) },
eb3b5501 362 { MATROSKA_ID_CODECDELAY, EBML_UINT, 0, offsetof(MatroskaTrack, codec_delay) },
84cfce9f
KC
363 { MATROSKA_ID_TRACKLANGUAGE, EBML_UTF8, 0, offsetof(MatroskaTrack, language), { .s = "eng" } },
364 { MATROSKA_ID_TRACKDEFAULTDURATION, EBML_UINT, 0, offsetof(MatroskaTrack, default_duration) },
365 { MATROSKA_ID_TRACKTIMECODESCALE, EBML_FLOAT, 0, offsetof(MatroskaTrack, time_scale), { .f = 1.0 } },
366 { MATROSKA_ID_TRACKFLAGDEFAULT, EBML_UINT, 0, offsetof(MatroskaTrack, flag_default), { .u = 1 } },
367 { MATROSKA_ID_TRACKFLAGFORCED, EBML_UINT, 0, offsetof(MatroskaTrack, flag_forced), { .u = 0 } },
368 { MATROSKA_ID_TRACKVIDEO, EBML_NEST, 0, offsetof(MatroskaTrack, video), { .n = matroska_track_video } },
369 { MATROSKA_ID_TRACKAUDIO, EBML_NEST, 0, offsetof(MatroskaTrack, audio), { .n = matroska_track_audio } },
370 { MATROSKA_ID_TRACKCONTENTENCODINGS, EBML_NEST, 0, 0, { .n = matroska_track_encodings } },
371 { MATROSKA_ID_TRACKFLAGENABLED, EBML_NONE },
372 { MATROSKA_ID_TRACKFLAGLACING, EBML_NONE },
373 { MATROSKA_ID_CODECNAME, EBML_NONE },
374 { MATROSKA_ID_CODECDECODEALL, EBML_NONE },
375 { MATROSKA_ID_CODECINFOURL, EBML_NONE },
376 { MATROSKA_ID_CODECDOWNLOADURL, EBML_NONE },
377 { MATROSKA_ID_TRACKMINCACHE, EBML_NONE },
378 { MATROSKA_ID_TRACKMAXCACHE, EBML_NONE },
379 { MATROSKA_ID_TRACKMAXBLKADDID, EBML_NONE },
2cbc8811
AJ
380 { 0 }
381};
382
383static EbmlSyntax matroska_tracks[] = {
84cfce9f 384 { MATROSKA_ID_TRACKENTRY, EBML_NEST, sizeof(MatroskaTrack), offsetof(MatroskaDemuxContext, tracks), { .n = matroska_track } },
2cbc8811
AJ
385 { 0 }
386};
387
b414cb89 388static EbmlSyntax matroska_attachment[] = {
f69befe5
DB
389 { MATROSKA_ID_FILEUID, EBML_UINT, 0, offsetof(MatroskaAttachment, uid) },
390 { MATROSKA_ID_FILENAME, EBML_UTF8, 0, offsetof(MatroskaAttachment, filename) },
391 { MATROSKA_ID_FILEMIMETYPE, EBML_STR, 0, offsetof(MatroskaAttachment, mime) },
392 { MATROSKA_ID_FILEDATA, EBML_BIN, 0, offsetof(MatroskaAttachment, bin) },
84cfce9f 393 { MATROSKA_ID_FILEDESC, EBML_NONE },
b414cb89
AJ
394 { 0 }
395};
396
397static EbmlSyntax matroska_attachments[] = {
f69befe5 398 { MATROSKA_ID_ATTACHEDFILE, EBML_NEST, sizeof(MatroskaAttachment), offsetof(MatroskaDemuxContext, attachments), { .n = matroska_attachment } },
b414cb89
AJ
399 { 0 }
400};
401
6bbd7c7b 402static EbmlSyntax matroska_chapter_display[] = {
84cfce9f
KC
403 { MATROSKA_ID_CHAPSTRING, EBML_UTF8, 0, offsetof(MatroskaChapter, title) },
404 { MATROSKA_ID_CHAPLANG, EBML_NONE },
6bbd7c7b
AJ
405 { 0 }
406};
407
408static EbmlSyntax matroska_chapter_entry[] = {
84cfce9f
KC
409 { MATROSKA_ID_CHAPTERTIMESTART, EBML_UINT, 0, offsetof(MatroskaChapter, start), { .u = AV_NOPTS_VALUE } },
410 { MATROSKA_ID_CHAPTERTIMEEND, EBML_UINT, 0, offsetof(MatroskaChapter, end), { .u = AV_NOPTS_VALUE } },
411 { MATROSKA_ID_CHAPTERUID, EBML_UINT, 0, offsetof(MatroskaChapter, uid) },
412 { MATROSKA_ID_CHAPTERDISPLAY, EBML_NEST, 0, 0, { .n = matroska_chapter_display } },
6bbd7c7b 413 { MATROSKA_ID_CHAPTERFLAGHIDDEN, EBML_NONE },
5df3cc6f
AJ
414 { MATROSKA_ID_CHAPTERFLAGENABLED, EBML_NONE },
415 { MATROSKA_ID_CHAPTERPHYSEQUIV, EBML_NONE },
416 { MATROSKA_ID_CHAPTERATOM, EBML_NONE },
6bbd7c7b
AJ
417 { 0 }
418};
419
420static EbmlSyntax matroska_chapter[] = {
84cfce9f 421 { MATROSKA_ID_CHAPTERATOM, EBML_NEST, sizeof(MatroskaChapter), offsetof(MatroskaDemuxContext, chapters), { .n = matroska_chapter_entry } },
6bbd7c7b
AJ
422 { MATROSKA_ID_EDITIONUID, EBML_NONE },
423 { MATROSKA_ID_EDITIONFLAGHIDDEN, EBML_NONE },
424 { MATROSKA_ID_EDITIONFLAGDEFAULT, EBML_NONE },
5df3cc6f 425 { MATROSKA_ID_EDITIONFLAGORDERED, EBML_NONE },
6bbd7c7b
AJ
426 { 0 }
427};
428
429static EbmlSyntax matroska_chapters[] = {
84cfce9f 430 { MATROSKA_ID_EDITIONENTRY, EBML_NEST, 0, 0, { .n = matroska_chapter } },
6bbd7c7b
AJ
431 { 0 }
432};
433
e5929fdf 434static EbmlSyntax matroska_index_pos[] = {
84cfce9f
KC
435 { MATROSKA_ID_CUETRACK, EBML_UINT, 0, offsetof(MatroskaIndexPos, track) },
436 { MATROSKA_ID_CUECLUSTERPOSITION, EBML_UINT, 0, offsetof(MatroskaIndexPos, pos) },
5df3cc6f 437 { MATROSKA_ID_CUEBLOCKNUMBER, EBML_NONE },
e5929fdf
AJ
438 { 0 }
439};
440
441static EbmlSyntax matroska_index_entry[] = {
84cfce9f
KC
442 { MATROSKA_ID_CUETIME, EBML_UINT, 0, offsetof(MatroskaIndex, time) },
443 { MATROSKA_ID_CUETRACKPOSITION, EBML_NEST, sizeof(MatroskaIndexPos), offsetof(MatroskaIndex, pos), { .n = matroska_index_pos } },
e5929fdf
AJ
444 { 0 }
445};
446
447static EbmlSyntax matroska_index[] = {
84cfce9f 448 { MATROSKA_ID_POINTENTRY, EBML_NEST, sizeof(MatroskaIndex), offsetof(MatroskaDemuxContext, index), { .n = matroska_index_entry } },
e5929fdf
AJ
449 { 0 }
450};
451
44015c56 452static EbmlSyntax matroska_simpletag[] = {
84cfce9f
KC
453 { MATROSKA_ID_TAGNAME, EBML_UTF8, 0, offsetof(MatroskaTag, name) },
454 { MATROSKA_ID_TAGSTRING, EBML_UTF8, 0, offsetof(MatroskaTag, string) },
455 { MATROSKA_ID_TAGLANG, EBML_STR, 0, offsetof(MatroskaTag, lang), { .s = "und" } },
456 { MATROSKA_ID_TAGDEFAULT, EBML_UINT, 0, offsetof(MatroskaTag, def) },
457 { MATROSKA_ID_TAGDEFAULT_BUG, EBML_UINT, 0, offsetof(MatroskaTag, def) },
458 { MATROSKA_ID_SIMPLETAG, EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTag, sub), { .n = matroska_simpletag } },
44015c56
AJ
459 { 0 }
460};
461
929e9de7 462static EbmlSyntax matroska_tagtargets[] = {
84cfce9f
KC
463 { MATROSKA_ID_TAGTARGETS_TYPE, EBML_STR, 0, offsetof(MatroskaTagTarget, type) },
464 { MATROSKA_ID_TAGTARGETS_TYPEVALUE, EBML_UINT, 0, offsetof(MatroskaTagTarget, typevalue), { .u = 50 } },
465 { MATROSKA_ID_TAGTARGETS_TRACKUID, EBML_UINT, 0, offsetof(MatroskaTagTarget, trackuid) },
466 { MATROSKA_ID_TAGTARGETS_CHAPTERUID, EBML_UINT, 0, offsetof(MatroskaTagTarget, chapteruid) },
467 { MATROSKA_ID_TAGTARGETS_ATTACHUID, EBML_UINT, 0, offsetof(MatroskaTagTarget, attachuid) },
929e9de7
AJ
468 { 0 }
469};
470
44015c56 471static EbmlSyntax matroska_tag[] = {
84cfce9f
KC
472 { MATROSKA_ID_SIMPLETAG, EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTags, tag), { .n = matroska_simpletag } },
473 { MATROSKA_ID_TAGTARGETS, EBML_NEST, 0, offsetof(MatroskaTags, target), { .n = matroska_tagtargets } },
44015c56
AJ
474 { 0 }
475};
476
434d496a 477static EbmlSyntax matroska_tags[] = {
84cfce9f 478 { MATROSKA_ID_TAG, EBML_NEST, sizeof(MatroskaTags), offsetof(MatroskaDemuxContext, tags), { .n = matroska_tag } },
434d496a
AJ
479 { 0 }
480};
481
13b350a3 482static EbmlSyntax matroska_seekhead_entry[] = {
84cfce9f
KC
483 { MATROSKA_ID_SEEKID, EBML_UINT, 0, offsetof(MatroskaSeekhead, id) },
484 { MATROSKA_ID_SEEKPOSITION, EBML_UINT, 0, offsetof(MatroskaSeekhead, pos), { .u = -1 } },
13b350a3
AJ
485 { 0 }
486};
487
488static EbmlSyntax matroska_seekhead[] = {
84cfce9f 489 { MATROSKA_ID_SEEKENTRY, EBML_NEST, sizeof(MatroskaSeekhead), offsetof(MatroskaDemuxContext, seekhead), { .n = matroska_seekhead_entry } },
13b350a3
AJ
490 { 0 }
491};
492
ce6f28bd 493static EbmlSyntax matroska_segment[] = {
84cfce9f
KC
494 { MATROSKA_ID_INFO, EBML_NEST, 0, 0, { .n = matroska_info } },
495 { MATROSKA_ID_TRACKS, EBML_NEST, 0, 0, { .n = matroska_tracks } },
496 { MATROSKA_ID_ATTACHMENTS, EBML_NEST, 0, 0, { .n = matroska_attachments } },
497 { MATROSKA_ID_CHAPTERS, EBML_NEST, 0, 0, { .n = matroska_chapters } },
498 { MATROSKA_ID_CUES, EBML_NEST, 0, 0, { .n = matroska_index } },
499 { MATROSKA_ID_TAGS, EBML_NEST, 0, 0, { .n = matroska_tags } },
500 { MATROSKA_ID_SEEKHEAD, EBML_NEST, 0, 0, { .n = matroska_seekhead } },
501 { MATROSKA_ID_CLUSTER, EBML_STOP },
ce6f28bd
AJ
502 { 0 }
503};
504
505static EbmlSyntax matroska_segments[] = {
84cfce9f 506 { MATROSKA_ID_SEGMENT, EBML_NEST, 0, 0, { .n = matroska_segment } },
ce6f28bd
AJ
507 { 0 }
508};
509
209472b4 510static EbmlSyntax matroska_blockgroup[] = {
84cfce9f
KC
511 { MATROSKA_ID_BLOCK, EBML_BIN, 0, offsetof(MatroskaBlock, bin) },
512 { MATROSKA_ID_SIMPLEBLOCK, EBML_BIN, 0, offsetof(MatroskaBlock, bin) },
513 { MATROSKA_ID_BLOCKDURATION, EBML_UINT, 0, offsetof(MatroskaBlock, duration), { .u = AV_NOPTS_VALUE } },
514 { MATROSKA_ID_BLOCKREFERENCE, EBML_UINT, 0, offsetof(MatroskaBlock, reference) },
564b7e0c 515 { MATROSKA_ID_CODECSTATE, EBML_NONE },
84cfce9f 516 { 1, EBML_UINT, 0, offsetof(MatroskaBlock, non_simple), { .u = 1 } },
209472b4
AJ
517 { 0 }
518};
519
520static EbmlSyntax matroska_cluster[] = {
84cfce9f
KC
521 { MATROSKA_ID_CLUSTERTIMECODE, EBML_UINT, 0, offsetof(MatroskaCluster, timecode) },
522 { MATROSKA_ID_BLOCKGROUP, EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster, blocks), { .n = matroska_blockgroup } },
523 { MATROSKA_ID_SIMPLEBLOCK, EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster, blocks), { .n = matroska_blockgroup } },
524 { MATROSKA_ID_CLUSTERPOSITION, EBML_NONE },
525 { MATROSKA_ID_CLUSTERPREVSIZE, EBML_NONE },
209472b4
AJ
526 { 0 }
527};
528
529static EbmlSyntax matroska_clusters[] = {
84cfce9f
KC
530 { MATROSKA_ID_CLUSTER, EBML_NEST, 0, 0, { .n = matroska_cluster } },
531 { MATROSKA_ID_INFO, EBML_NONE },
532 { MATROSKA_ID_CUES, EBML_NONE },
533 { MATROSKA_ID_TAGS, EBML_NONE },
534 { MATROSKA_ID_SEEKHEAD, EBML_NONE },
209472b4
AJ
535 { 0 }
536};
537
8336eb6f 538static EbmlSyntax matroska_cluster_incremental_parsing[] = {
84cfce9f
KC
539 { MATROSKA_ID_CLUSTERTIMECODE, EBML_UINT, 0, offsetof(MatroskaCluster, timecode) },
540 { MATROSKA_ID_BLOCKGROUP, EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster, blocks), { .n = matroska_blockgroup } },
541 { MATROSKA_ID_SIMPLEBLOCK, EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster, blocks), { .n = matroska_blockgroup } },
542 { MATROSKA_ID_CLUSTERPOSITION, EBML_NONE },
543 { MATROSKA_ID_CLUSTERPREVSIZE, EBML_NONE },
544 { MATROSKA_ID_INFO, EBML_NONE },
545 { MATROSKA_ID_CUES, EBML_NONE },
546 { MATROSKA_ID_TAGS, EBML_NONE },
547 { MATROSKA_ID_SEEKHEAD, EBML_NONE },
548 { MATROSKA_ID_CLUSTER, EBML_STOP },
8336eb6f
DC
549 { 0 }
550};
551
552static EbmlSyntax matroska_cluster_incremental[] = {
84cfce9f
KC
553 { MATROSKA_ID_CLUSTERTIMECODE, EBML_UINT, 0, offsetof(MatroskaCluster, timecode) },
554 { MATROSKA_ID_BLOCKGROUP, EBML_STOP },
555 { MATROSKA_ID_SIMPLEBLOCK, EBML_STOP },
556 { MATROSKA_ID_CLUSTERPOSITION, EBML_NONE },
557 { MATROSKA_ID_CLUSTERPREVSIZE, EBML_NONE },
8336eb6f
DC
558 { 0 }
559};
560
561static EbmlSyntax matroska_clusters_incremental[] = {
84cfce9f
KC
562 { MATROSKA_ID_CLUSTER, EBML_NEST, 0, 0, { .n = matroska_cluster_incremental } },
563 { MATROSKA_ID_INFO, EBML_NONE },
564 { MATROSKA_ID_CUES, EBML_NONE },
565 { MATROSKA_ID_TAGS, EBML_NONE },
566 { MATROSKA_ID_SEEKHEAD, EBML_NONE },
8336eb6f
DC
567 { 0 }
568};
569
b0f29db5 570static const char *const matroska_doctypes[] = { "matroska", "webm" };
470491fa 571
8835c554
SM
572static int matroska_resync(MatroskaDemuxContext *matroska, int64_t last_pos)
573{
574 AVIOContext *pb = matroska->ctx->pb;
575 uint32_t id;
576 matroska->current_id = 0;
577 matroska->num_levels = 0;
578
579 /* seek to next position to resync from */
580 if (avio_seek(pb, last_pos + 1, SEEK_SET) < 0)
581 goto eof;
582
583 id = avio_rb32(pb);
584
585 // try to find a toplevel element
586 while (!pb->eof_reached) {
587 if (id == MATROSKA_ID_INFO || id == MATROSKA_ID_TRACKS ||
588 id == MATROSKA_ID_CUES || id == MATROSKA_ID_TAGS ||
589 id == MATROSKA_ID_SEEKHEAD || id == MATROSKA_ID_ATTACHMENTS ||
590 id == MATROSKA_ID_CLUSTER || id == MATROSKA_ID_CHAPTERS) {
84cfce9f
KC
591 matroska->current_id = id;
592 return 0;
8835c554
SM
593 }
594 id = (id << 8) | avio_r8(pb);
595 }
84cfce9f 596
8835c554
SM
597eof:
598 matroska->done = 1;
599 return AVERROR_EOF;
600}
601
b061d892 602/*
5968d2dd 603 * Return: Whether we reached the end of a level in the hierarchy or not.
b061d892 604 */
592110c2 605static int ebml_level_end(MatroskaDemuxContext *matroska)
b061d892 606{
ae628ec1 607 AVIOContext *pb = matroska->ctx->pb;
a2704c97 608 int64_t pos = avio_tell(pb);
b061d892 609
592110c2 610 if (matroska->num_levels > 0) {
b061d892 611 MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
8dbe48fc 612 if (pos - level->start >= level->length || matroska->current_id) {
b061d892 613 matroska->num_levels--;
592110c2 614 return 1;
b061d892
DC
615 }
616 }
592110c2 617 return 0;
b061d892
DC
618}
619
620/*
621 * Read: an "EBML number", which is defined as a variable-length
622 * array of bytes. The first byte indicates the length by giving a
623 * number of 0-bits followed by a one. The position of the first
624 * "one" bit inside the first byte indicates the length of this
625 * number.
5968d2dd 626 * Returns: number of bytes read, < 0 on error
b061d892 627 */
ae628ec1 628static int ebml_read_num(MatroskaDemuxContext *matroska, AVIOContext *pb,
f7b9687c 629 int max_size, uint64_t *number)
b061d892 630{
ff6a5fc1
RD
631 int read = 1, n = 1;
632 uint64_t total = 0;
b061d892 633
b7effd4e 634 /* The first byte tells us the length in bytes - avio_r8() can normally
b061d892
DC
635 * return 0, but since that's not a valid first ebmlID byte, we can
636 * use it safely here to catch EOS. */
b7effd4e 637 if (!(total = avio_r8(pb))) {
b061d892 638 /* we might encounter EOS here */
66e5b1df 639 if (!pb->eof_reached) {
a2704c97 640 int64_t pos = avio_tell(pb);
b061d892
DC
641 av_log(matroska->ctx, AV_LOG_ERROR,
642 "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
643 pos, pos);
721113be 644 return pb->error ? pb->error : AVERROR(EIO);
b061d892 645 }
721113be 646 return AVERROR_EOF;
b061d892
DC
647 }
648
649 /* get the length of the EBML number */
ff6a5fc1 650 read = 8 - ff_log2_tab[total];
b061d892 651 if (read > max_size) {
a2704c97 652 int64_t pos = avio_tell(pb) - 1;
b061d892
DC
653 av_log(matroska->ctx, AV_LOG_ERROR,
654 "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
655 (uint8_t) total, pos, pos);
656 return AVERROR_INVALIDDATA;
657 }
658
659 /* read out length */
ff6a5fc1 660 total ^= 1 << ff_log2_tab[total];
b061d892 661 while (n++ < read)
b7effd4e 662 total = (total << 8) | avio_r8(pb);
b061d892
DC
663
664 *number = total;
665
666 return read;
667}
668
1b4d327b
RD
669/**
670 * Read a EBML length value.
671 * This needs special handling for the "unknown length" case which has multiple
672 * encodings.
673 */
ae628ec1 674static int ebml_read_length(MatroskaDemuxContext *matroska, AVIOContext *pb,
1b4d327b
RD
675 uint64_t *number)
676{
677 int res = ebml_read_num(matroska, pb, 8, number);
678 if (res > 0 && *number + 1 == 1ULL << (7 * res))
679 *number = 0xffffffffffffffULL;
680 return res;
681}
682
b061d892 683/*
b061d892
DC
684 * Read the next element as an unsigned int.
685 * 0 is success, < 0 is failure.
686 */
ae628ec1 687static int ebml_read_uint(AVIOContext *pb, int size, uint64_t *num)
b061d892 688{
c6cd2b3d 689 int n = 0;
b061d892 690
4a194c8f 691 if (size > 8)
b061d892 692 return AVERROR_INVALIDDATA;
b061d892 693
5968d2dd 694 /* big-endian ordering; build up number */
b061d892
DC
695 *num = 0;
696 while (n++ < size)
b7effd4e 697 *num = (*num << 8) | avio_r8(pb);
b061d892
DC
698
699 return 0;
700}
701
702/*
b061d892
DC
703 * Read the next element as a float.
704 * 0 is success, < 0 is failure.
705 */
ae628ec1 706static int ebml_read_float(AVIOContext *pb, int size, double *num)
b061d892 707{
84cfce9f 708 if (size == 0)
4a194c8f 709 *num = 0;
84cfce9f 710 else if (size == 4)
3383a53e 711 *num = av_int2float(avio_rb32(pb));
84cfce9f 712 else if (size == 8)
3383a53e 713 *num = av_int2double(avio_rb64(pb));
84cfce9f 714 else
b061d892 715 return AVERROR_INVALIDDATA;
b061d892
DC
716
717 return 0;
718}
719
720/*
721 * Read the next element as an ASCII string.
722 * 0 is success, < 0 is failure.
723 */
ae628ec1 724static int ebml_read_ascii(AVIOContext *pb, int size, char **str)
b061d892 725{
cd40c31e
RB
726 char *res;
727
5968d2dd 728 /* EBML strings are usually not 0-terminated, so we allocate one
b061d892 729 * byte more, read the string and NULL-terminate it ourselves. */
cd40c31e 730 if (!(res = av_malloc(size + 1)))
769e10f0 731 return AVERROR(ENOMEM);
cd40c31e
RB
732 if (avio_read(pb, (uint8_t *) res, size) != size) {
733 av_free(res);
6f3e0b21 734 return AVERROR(EIO);
b061d892 735 }
cd40c31e
RB
736 (res)[size] = '\0';
737 av_free(*str);
738 *str = res;
b061d892
DC
739
740 return 0;
741}
742
743/*
737c40da
AJ
744 * Read the next element as binary data.
745 * 0 is success, < 0 is failure.
746 */
ae628ec1 747static int ebml_read_binary(AVIOContext *pb, int length, EbmlBin *bin)
737c40da
AJ
748{
749 av_free(bin->data);
80bd3e5c
AK
750 bin->size = 0;
751
73953df7 752 if (!(bin->data = av_mallocz(length + FF_INPUT_BUFFER_PADDING_SIZE)))
737c40da
AJ
753 return AVERROR(ENOMEM);
754
a2704c97 755 bin->pos = avio_tell(pb);
b7effd4e 756 if (avio_read(pb, bin->data, length) != length) {
5549aa6d 757 av_freep(&bin->data);
737c40da 758 return AVERROR(EIO);
5549aa6d 759 }
737c40da 760
80bd3e5c
AK
761 bin->size = length;
762
737c40da
AJ
763 return 0;
764}
765
766/*
b061d892
DC
767 * Read the next element, but only the header. The contents
768 * are supposed to be sub-elements which can be read separately.
769 * 0 is success, < 0 is failure.
770 */
bddd1d93 771static int ebml_read_master(MatroskaDemuxContext *matroska, uint64_t length)
b061d892 772{
ae628ec1 773 AVIOContext *pb = matroska->ctx->pb;
b061d892 774 MatroskaLevel *level;
b061d892 775
b061d892
DC
776 if (matroska->num_levels >= EBML_MAX_DEPTH) {
777 av_log(matroska->ctx, AV_LOG_ERROR,
778 "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
85565db0 779 return AVERROR(ENOSYS);
b061d892
DC
780 }
781
84cfce9f
KC
782 level = &matroska->levels[matroska->num_levels++];
783 level->start = avio_tell(pb);
b061d892
DC
784 level->length = length;
785
786 return 0;
787}
788
789/*
b061d892 790 * Read signed/unsigned "EBML" numbers.
5968d2dd 791 * Return: number of bytes processed, < 0 on error
b061d892 792 */
c1e01133
AJ
793static int matroska_ebmlnum_uint(MatroskaDemuxContext *matroska,
794 uint8_t *data, uint32_t size, uint64_t *num)
b061d892 795{
ae628ec1 796 AVIOContext pb;
e731b8d8 797 ffio_init_context(&pb, data, size, 0, NULL, NULL, NULL, NULL);
465c28b6 798 return ebml_read_num(matroska, &pb, FFMIN(size, 8), num);
b061d892
DC
799}
800
801/*
802 * Same as above, but signed.
803 */
c1e01133
AJ
804static int matroska_ebmlnum_sint(MatroskaDemuxContext *matroska,
805 uint8_t *data, uint32_t size, int64_t *num)
b061d892
DC
806{
807 uint64_t unum;
808 int res;
809
810 /* read as unsigned number first */
c1e01133 811 if ((res = matroska_ebmlnum_uint(matroska, data, size, &unum)) < 0)
b061d892
DC
812 return res;
813
814 /* make signed (weird way) */
84cfce9f 815 *num = unum - ((1LL << (7 * res - 1)) - 1);
b061d892
DC
816
817 return res;
818}
819
737c40da
AJ
820static int ebml_parse_elem(MatroskaDemuxContext *matroska,
821 EbmlSyntax *syntax, void *data);
b061d892 822
737c40da
AJ
823static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
824 uint32_t id, void *data)
b061d892
DC
825{
826 int i;
84cfce9f 827 for (i = 0; syntax[i].id; i++)
737c40da
AJ
828 if (id == syntax[i].id)
829 break;
8dbe48fc 830 if (!syntax[i].id && id == MATROSKA_ID_CLUSTER &&
84cfce9f
KC
831 matroska->num_levels > 0 &&
832 matroska->levels[matroska->num_levels - 1].length == 0xffffffffffffff)
8dbe48fc 833 return 0; // we reached the end of an unknown size cluster
5b7a88f0 834 if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32) {
d92024f1 835 av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%"PRIX32"\n", id);
5b7a88f0
AK
836 if (matroska->ctx->error_recognition & AV_EF_EXPLODE)
837 return AVERROR_INVALIDDATA;
838 }
737c40da 839 return ebml_parse_elem(matroska, &syntax[i], data);
b061d892
DC
840}
841
737c40da
AJ
842static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
843 void *data)
b061d892 844{
c3ade62c 845 if (!matroska->current_id) {
73917810
AJ
846 uint64_t id;
847 int res = ebml_read_num(matroska, matroska->ctx->pb, 4, &id);
848 if (res < 0)
849 return res;
84cfce9f 850 matroska->current_id = id | 1 << 7 * res;
c3ade62c
AJ
851 }
852 return ebml_parse_id(matroska, syntax, matroska->current_id, data);
b061d892
DC
853}
854
737c40da
AJ
855static int ebml_parse_nest(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
856 void *data)
b061d892 857{
737c40da 858 int i, res = 0;
b061d892 859
84cfce9f 860 for (i = 0; syntax[i].id; i++)
737c40da
AJ
861 switch (syntax[i].type) {
862 case EBML_UINT:
84cfce9f 863 *(uint64_t *) ((char *) data + syntax[i].data_offset) = syntax[i].def.u;
737c40da
AJ
864 break;
865 case EBML_FLOAT:
84cfce9f 866 *(double *) ((char *) data + syntax[i].data_offset) = syntax[i].def.f;
737c40da
AJ
867 break;
868 case EBML_STR:
869 case EBML_UTF8:
668643b9
AK
870 // the default may be NULL
871 if (syntax[i].def.s) {
84cfce9f 872 uint8_t **dst = (uint8_t **) ((uint8_t *) data + syntax[i].data_offset);
668643b9
AK
873 *dst = av_strdup(syntax[i].def.s);
874 if (!*dst)
875 return AVERROR(ENOMEM);
876 }
737c40da 877 break;
34c9c1ba 878 }
b061d892 879
737c40da
AJ
880 while (!res && !ebml_level_end(matroska))
881 res = ebml_parse(matroska, syntax, data);
b061d892 882
737c40da 883 return res;
b061d892
DC
884}
885
789ed100
AJ
886static int ebml_parse_elem(MatroskaDemuxContext *matroska,
887 EbmlSyntax *syntax, void *data)
888{
95ec3d4c
RD
889 static const uint64_t max_lengths[EBML_TYPE_COUNT] = {
890 [EBML_UINT] = 8,
891 [EBML_FLOAT] = 8,
892 // max. 16 MB for strings
893 [EBML_STR] = 0x1000000,
894 [EBML_UTF8] = 0x1000000,
895 // max. 256 MB for binary data
896 [EBML_BIN] = 0x10000000,
897 // no limits for anything else
898 };
ae628ec1 899 AVIOContext *pb = matroska->ctx->pb;
789ed100 900 uint32_t id = syntax->id;
c6cd2b3d 901 uint64_t length;
789ed100
AJ
902 int res;
903
84cfce9f 904 data = (char *) data + syntax->data_offset;
789ed100
AJ
905 if (syntax->list_elem_size) {
906 EbmlList *list = data;
f369b935
AK
907 if ((res = av_reallocp_array(&list->elem,
908 list->nb_elem + 1,
909 syntax->list_elem_size)) < 0) {
910 list->nb_elem = 0;
911 return res;
912 }
84cfce9f 913 data = (char *) list->elem + list->nb_elem * syntax->list_elem_size;
789ed100
AJ
914 memset(data, 0, syntax->list_elem_size);
915 list->nb_elem++;
916 }
789ed100 917
c3ade62c
AJ
918 if (syntax->type != EBML_PASS && syntax->type != EBML_STOP) {
919 matroska->current_id = 0;
1b4d327b 920 if ((res = ebml_read_length(matroska, pb, &length)) < 0)
c6cd2b3d 921 return res;
95ec3d4c
RD
922 if (max_lengths[syntax->type] && length > max_lengths[syntax->type]) {
923 av_log(matroska->ctx, AV_LOG_ERROR,
924 "Invalid length 0x%"PRIx64" > 0x%"PRIx64" for syntax element %i\n",
925 length, max_lengths[syntax->type], syntax->type);
926 return AVERROR_INVALIDDATA;
927 }
c3ade62c 928 }
c6cd2b3d 929
789ed100 930 switch (syntax->type) {
84cfce9f
KC
931 case EBML_UINT:
932 res = ebml_read_uint(pb, length, data);
933 break;
934 case EBML_FLOAT:
935 res = ebml_read_float(pb, length, data);
936 break;
789ed100 937 case EBML_STR:
84cfce9f
KC
938 case EBML_UTF8:
939 res = ebml_read_ascii(pb, length, data);
940 break;
941 case EBML_BIN:
942 res = ebml_read_binary(pb, length, data);
943 break;
944 case EBML_NEST:
945 if ((res = ebml_read_master(matroska, length)) < 0)
946 return res;
947 if (id == MATROSKA_ID_SEGMENT)
948 matroska->segment_start = avio_tell(matroska->ctx->pb);
949 return ebml_parse_nest(matroska, syntax->def.n, data);
950 case EBML_PASS:
951 return ebml_parse_id(matroska, syntax->def.n, id, data);
952 case EBML_STOP:
953 return 1;
954 default:
955 return avio_skip(pb, length) < 0 ? AVERROR(EIO) : 0;
789ed100 956 }
ba5a1f99
AJ
957 if (res == AVERROR_INVALIDDATA)
958 av_log(matroska->ctx, AV_LOG_ERROR, "Invalid element\n");
959 else if (res == AVERROR(EIO))
960 av_log(matroska->ctx, AV_LOG_ERROR, "Read error\n");
961 return res;
789ed100
AJ
962}
963
789ed100
AJ
964static void ebml_free(EbmlSyntax *syntax, void *data)
965{
966 int i, j;
84cfce9f
KC
967 for (i = 0; syntax[i].id; i++) {
968 void *data_off = (char *) data + syntax[i].data_offset;
789ed100
AJ
969 switch (syntax[i].type) {
970 case EBML_STR:
84cfce9f
KC
971 case EBML_UTF8:
972 av_freep(data_off);
973 break;
974 case EBML_BIN:
975 av_freep(&((EbmlBin *) data_off)->data);
976 break;
789ed100
AJ
977 case EBML_NEST:
978 if (syntax[i].list_elem_size) {
979 EbmlList *list = data_off;
980 char *ptr = list->elem;
84cfce9f
KC
981 for (j = 0; j < list->nb_elem;
982 j++, ptr += syntax[i].list_elem_size)
789ed100
AJ
983 ebml_free(syntax[i].def.n, ptr);
984 av_free(list->elem);
985 } else
986 ebml_free(syntax[i].def.n, data_off);
84cfce9f
KC
987 default:
988 break;
789ed100
AJ
989 }
990 }
991}
992
737c40da
AJ
993/*
994 * Autodetecting...
995 */
996static int matroska_probe(AVProbeData *p)
997{
998 uint64_t total = 0;
470491fa 999 int len_mask = 0x80, size = 1, n = 1, i;
737c40da 1000
5968d2dd 1001 /* EBML header? */
737c40da
AJ
1002 if (AV_RB32(p->buf) != EBML_ID_HEADER)
1003 return 0;
1004
1005 /* length of header */
1006 total = p->buf[4];
1007 while (size <= 8 && !(total & len_mask)) {
1008 size++;
1009 len_mask >>= 1;
1010 }
1011 if (size > 8)
84cfce9f 1012 return 0;
737c40da
AJ
1013 total &= (len_mask - 1);
1014 while (n < size)
1015 total = (total << 8) | p->buf[4 + n++];
1016
5968d2dd 1017 /* Does the probe data contain the whole header? */
737c40da 1018 if (p->buf_size < 4 + size + total)
84cfce9f 1019 return 0;
737c40da 1020
470491fa 1021 /* The header should contain a known document type. For now,
737c40da
AJ
1022 * we don't parse the whole header but simply check for the
1023 * availability of that array of characters inside the header.
1024 * Not fully fool-proof, but good enough. */
470491fa
JZ
1025 for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++) {
1026 int probelen = strlen(matroska_doctypes[i]);
69619a13
CE
1027 if (total < probelen)
1028 continue;
84cfce9f
KC
1029 for (n = 4 + size; n <= 4 + size + total - probelen; n++)
1030 if (!memcmp(p->buf + n, matroska_doctypes[i], probelen))
470491fa
JZ
1031 return AVPROBE_SCORE_MAX;
1032 }
737c40da 1033
c7b913c8 1034 // probably valid EBML header but no recognized doctype
e0f8be64 1035 return AVPROBE_SCORE_EXTENSION;
737c40da
AJ
1036}
1037
1038static MatroskaTrack *matroska_find_track_by_num(MatroskaDemuxContext *matroska,
1039 int num)
1040{
1041 MatroskaTrack *tracks = matroska->tracks.elem;
1042 int i;
1043
84cfce9f 1044 for (i = 0; i < matroska->tracks.nb_elem; i++)
737c40da
AJ
1045 if (tracks[i].num == num)
1046 return &tracks[i];
1047
1048 av_log(matroska->ctx, AV_LOG_ERROR, "Invalid track number %d\n", num);
1049 return NULL;
1050}
1051
84cfce9f 1052static int matroska_decode_buffer(uint8_t **buf, int *buf_size,
f7b9687c 1053 MatroskaTrack *track)
935ec5a1 1054{
2cbc8811 1055 MatroskaTrackEncoding *encodings = track->encodings.elem;
84cfce9f 1056 uint8_t *data = *buf;
935ec5a1 1057 int isize = *buf_size;
84cfce9f 1058 uint8_t *pkt_data = NULL;
529506b5 1059 uint8_t av_unused *newpktdata;
935ec5a1
ES
1060 int pkt_size = isize;
1061 int result = 0;
1062 int olen;
1063
4f90688b 1064 if (pkt_size >= 10000000)
c9a39cec 1065 return AVERROR_INVALIDDATA;
4f90688b 1066
2cbc8811 1067 switch (encodings[0].compression.algo) {
84cfce9f
KC
1068 case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
1069 {
8d4dd55c
LB
1070 int header_size = encodings[0].compression.settings.size;
1071 uint8_t *header = encodings[0].compression.settings.data;
1072
1073 if (!header_size)
1074 return 0;
1075
1076 pkt_size = isize + header_size;
1077 pkt_data = av_malloc(pkt_size);
1078 if (!pkt_data)
1079 return AVERROR(ENOMEM);
1080
1081 memcpy(pkt_data, header, header_size);
1082 memcpy(pkt_data + header_size, data, isize);
1083 break;
1084 }
2a91ada8 1085#if CONFIG_LZO
935ec5a1
ES
1086 case MATROSKA_TRACK_ENCODING_COMP_LZO:
1087 do {
84cfce9f 1088 olen = pkt_size *= 3;
581281e2
LB
1089 newpktdata = av_realloc(pkt_data, pkt_size + AV_LZO_OUTPUT_PADDING);
1090 if (!newpktdata) {
c9a39cec 1091 result = AVERROR(ENOMEM);
581281e2
LB
1092 goto failed;
1093 }
1094 pkt_data = newpktdata;
84cfce9f
KC
1095 result = av_lzo1x_decode(pkt_data, &olen, data, &isize);
1096 } while (result == AV_LZO_OUTPUT_FULL && pkt_size < 10000000);
c9a39cec
LB
1097 if (result) {
1098 result = AVERROR_INVALIDDATA;
935ec5a1 1099 goto failed;
c9a39cec 1100 }
935ec5a1
ES
1101 pkt_size -= olen;
1102 break;
2a91ada8 1103#endif
b250f9c6 1104#if CONFIG_ZLIB
84cfce9f
KC
1105 case MATROSKA_TRACK_ENCODING_COMP_ZLIB:
1106 {
1107 z_stream zstream = { 0 };
935ec5a1
ES
1108 if (inflateInit(&zstream) != Z_OK)
1109 return -1;
84cfce9f 1110 zstream.next_in = data;
935ec5a1
ES
1111 zstream.avail_in = isize;
1112 do {
84cfce9f 1113 pkt_size *= 3;
77d2ef13
MN
1114 newpktdata = av_realloc(pkt_data, pkt_size);
1115 if (!newpktdata) {
1116 inflateEnd(&zstream);
1117 goto failed;
1118 }
84cfce9f 1119 pkt_data = newpktdata;
935ec5a1 1120 zstream.avail_out = pkt_size - zstream.total_out;
84cfce9f
KC
1121 zstream.next_out = pkt_data + zstream.total_out;
1122 result = inflate(&zstream, Z_NO_FLUSH);
1123 } while (result == Z_OK && pkt_size < 10000000);
935ec5a1
ES
1124 pkt_size = zstream.total_out;
1125 inflateEnd(&zstream);
c9a39cec
LB
1126 if (result != Z_STREAM_END) {
1127 if (result == Z_MEM_ERROR)
1128 result = AVERROR(ENOMEM);
1129 else
1130 result = AVERROR_INVALIDDATA;
935ec5a1 1131 goto failed;
c9a39cec 1132 }
935ec5a1
ES
1133 break;
1134 }
1135#endif
b250f9c6 1136#if CONFIG_BZLIB
84cfce9f
KC
1137 case MATROSKA_TRACK_ENCODING_COMP_BZLIB:
1138 {
1139 bz_stream bzstream = { 0 };
935ec5a1
ES
1140 if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
1141 return -1;
84cfce9f 1142 bzstream.next_in = data;
935ec5a1
ES
1143 bzstream.avail_in = isize;
1144 do {
84cfce9f 1145 pkt_size *= 3;
77d2ef13
MN
1146 newpktdata = av_realloc(pkt_data, pkt_size);
1147 if (!newpktdata) {
1148 BZ2_bzDecompressEnd(&bzstream);
1149 goto failed;
1150 }
84cfce9f 1151 pkt_data = newpktdata;
935ec5a1 1152 bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
84cfce9f
KC
1153 bzstream.next_out = pkt_data + bzstream.total_out_lo32;
1154 result = BZ2_bzDecompress(&bzstream);
1155 } while (result == BZ_OK && pkt_size < 10000000);
935ec5a1
ES
1156 pkt_size = bzstream.total_out_lo32;
1157 BZ2_bzDecompressEnd(&bzstream);
c9a39cec
LB
1158 if (result != BZ_STREAM_END) {
1159 if (result == BZ_MEM_ERROR)
1160 result = AVERROR(ENOMEM);
1161 else
1162 result = AVERROR_INVALIDDATA;
935ec5a1 1163 goto failed;
c9a39cec 1164 }
935ec5a1
ES
1165 break;
1166 }
1167#endif
28f27e0c 1168 default:
c9a39cec 1169 return AVERROR_INVALIDDATA;
935ec5a1
ES
1170 }
1171
84cfce9f 1172 *buf = pkt_data;
935ec5a1
ES
1173 *buf_size = pkt_size;
1174 return 0;
84cfce9f
KC
1175
1176failed:
935ec5a1 1177 av_free(pkt_data);
c9a39cec 1178 return result;
935ec5a1
ES
1179}
1180
3eb9bfbb 1181static void matroska_fix_ass_packet(MatroskaDemuxContext *matroska,
e7d4b743 1182 AVPacket *pkt, uint64_t display_duration)
3eb9bfbb 1183{
1afddbe5 1184 AVBufferRef *line;
84cfce9f
KC
1185 char *layer, *ptr = pkt->data, *end = ptr + pkt->size;
1186
1187 for (; *ptr != ',' && ptr < end - 1; ptr++)
1188 ;
3eb9bfbb
AJ
1189 if (*ptr == ',')
1190 layer = ++ptr;
84cfce9f
KC
1191 for (; *ptr != ',' && ptr < end - 1; ptr++)
1192 ;
3eb9bfbb 1193 if (*ptr == ',') {
e7d4b743 1194 int64_t end_pts = pkt->pts + display_duration;
3eb9bfbb
AJ
1195 int sc = matroska->time_scale * pkt->pts / 10000000;
1196 int ec = matroska->time_scale * end_pts / 10000000;
1197 int sh, sm, ss, eh, em, es, len;
84cfce9f
KC
1198 sh = sc / 360000;
1199 sc -= 360000 * sh;
1200 sm = sc / 6000;
1201 sc -= 6000 * sm;
1202 ss = sc / 100;
1203 sc -= 100 * ss;
1204 eh = ec / 360000;
1205 ec -= 360000 * eh;
1206 em = ec / 6000;
1207 ec -= 6000 * em;
1208 es = ec / 100;
1209 ec -= 100 * es;
3eb9bfbb 1210 *ptr++ = '\0';
84cfce9f 1211 len = 50 + end - ptr + FF_INPUT_BUFFER_PADDING_SIZE;
1afddbe5 1212 if (!(line = av_buffer_alloc(len)))
3eb9bfbb 1213 return;
84cfce9f
KC
1214 snprintf(line->data, len,
1215 "Dialogue: %s,%d:%02d:%02d.%02d,%d:%02d:%02d.%02d,%s\r\n",
3eb9bfbb 1216 layer, sh, sm, ss, sc, eh, em, es, ec, ptr);
1afddbe5
AK
1217 av_buffer_unref(&pkt->buf);
1218 pkt->buf = line;
1219 pkt->data = line->data;
1220 pkt->size = strlen(line->data);
3eb9bfbb
AJ
1221 }
1222}
1223
77d2ef13 1224static int matroska_merge_packets(AVPacket *out, AVPacket *in)
d5e34dcb 1225{
34871beb
AK
1226 int old_size = out->size;
1227 int ret = av_grow_packet(out, in->size);
1228 if (ret < 0)
1229 return ret;
1230
1231 memcpy(out->data + old_size, in->data, in->size);
1232
1233 av_free_packet(in);
d5e34dcb 1234 av_free(in);
77d2ef13 1235 return 0;
d5e34dcb
AJ
1236}
1237
929e9de7 1238static void matroska_convert_tag(AVFormatContext *s, EbmlList *list,
d2d67e42 1239 AVDictionary **metadata, char *prefix)
44015c56
AJ
1240{
1241 MatroskaTag *tags = list->elem;
929e9de7
AJ
1242 char key[1024];
1243 int i;
44015c56 1244
84cfce9f
KC
1245 for (i = 0; i < list->nb_elem; i++) {
1246 const char *lang = tags[i].lang &&
1247 strcmp(tags[i].lang, "und") ? tags[i].lang : NULL;
bf800c7d
AK
1248
1249 if (!tags[i].name) {
1250 av_log(s, AV_LOG_WARNING, "Skipping invalid tag with no TagName.\n");
1251 continue;
1252 }
84cfce9f
KC
1253 if (prefix)
1254 snprintf(key, sizeof(key), "%s/%s", prefix, tags[i].name);
1255 else
1256 av_strlcpy(key, tags[i].name, sizeof(key));
f702df39 1257 if (tags[i].def || !lang) {
84cfce9f
KC
1258 av_dict_set(metadata, key, tags[i].string, 0);
1259 if (tags[i].sub.nb_elem)
1260 matroska_convert_tag(s, &tags[i].sub, metadata, key);
f702df39
AJ
1261 }
1262 if (lang) {
1263 av_strlcat(key, "-", sizeof(key));
1264 av_strlcat(key, lang, sizeof(key));
d2d67e42 1265 av_dict_set(metadata, key, tags[i].string, 0);
f702df39
AJ
1266 if (tags[i].sub.nb_elem)
1267 matroska_convert_tag(s, &tags[i].sub, metadata, key);
1268 }
929e9de7 1269 }
ad7768f4 1270 ff_metadata_conv(metadata, NULL, ff_mkv_metadata_conv);
929e9de7
AJ
1271}
1272
1273static void matroska_convert_tags(AVFormatContext *s)
1274{
1275 MatroskaDemuxContext *matroska = s->priv_data;
1276 MatroskaTags *tags = matroska->tags.elem;
1277 int i, j;
1278
84cfce9f 1279 for (i = 0; i < matroska->tags.nb_elem; i++) {
929e9de7 1280 if (tags[i].target.attachuid) {
f69befe5 1281 MatroskaAttachment *attachment = matroska->attachments.elem;
84cfce9f
KC
1282 for (j = 0; j < matroska->attachments.nb_elem; j++)
1283 if (attachment[j].uid == tags[i].target.attachuid &&
1284 attachment[j].stream)
929e9de7
AJ
1285 matroska_convert_tag(s, &tags[i].tag,
1286 &attachment[j].stream->metadata, NULL);
1287 } else if (tags[i].target.chapteruid) {
1288 MatroskaChapter *chapter = matroska->chapters.elem;
84cfce9f
KC
1289 for (j = 0; j < matroska->chapters.nb_elem; j++)
1290 if (chapter[j].uid == tags[i].target.chapteruid &&
1291 chapter[j].chapter)
929e9de7
AJ
1292 matroska_convert_tag(s, &tags[i].tag,
1293 &chapter[j].chapter->metadata, NULL);
1294 } else if (tags[i].target.trackuid) {
1295 MatroskaTrack *track = matroska->tracks.elem;
84cfce9f 1296 for (j = 0; j < matroska->tracks.nb_elem; j++)
9c569ef7 1297 if (track[j].uid == tags[i].target.trackuid && track[j].stream)
929e9de7
AJ
1298 matroska_convert_tag(s, &tags[i].tag,
1299 &track[j].stream->metadata, NULL);
1300 } else {
4f909c76
AJ
1301 matroska_convert_tag(s, &tags[i].tag, &s->metadata,
1302 tags[i].target.type);
929e9de7 1303 }
44015c56
AJ
1304 }
1305}
1306
84cfce9f
KC
1307static int matroska_parse_seekhead_entry(MatroskaDemuxContext *matroska,
1308 int idx)
13b350a3
AJ
1309{
1310 EbmlList *seekhead_list = &matroska->seekhead;
84cfce9f
KC
1311 uint32_t level_up = matroska->level_up;
1312 uint32_t saved_id = matroska->current_id;
13b350a3 1313 MatroskaSeekhead *seekhead = seekhead_list->elem;
a2704c97 1314 int64_t before_pos = avio_tell(matroska->ctx->pb);
13b350a3 1315 MatroskaLevel level;
31ad14c2
AC
1316 int64_t offset;
1317 int ret = 0;
66cfc385 1318
84cfce9f
KC
1319 if (idx >= seekhead_list->nb_elem ||
1320 seekhead[idx].id == MATROSKA_ID_SEEKHEAD ||
1321 seekhead[idx].id == MATROSKA_ID_CLUSTER)
31ad14c2 1322 return 0;
b061d892 1323
f47ac3c6 1324 /* seek */
31ad14c2
AC
1325 offset = seekhead[idx].pos + matroska->segment_start;
1326 if (avio_seek(matroska->ctx->pb, offset, SEEK_SET) == offset) {
5968d2dd 1327 /* We don't want to lose our seekhead level, so we add
4348571b
AJ
1328 * a dummy. This is a crude hack. */
1329 if (matroska->num_levels == EBML_MAX_DEPTH) {
1330 av_log(matroska->ctx, AV_LOG_INFO,
1331 "Max EBML element depth (%d) reached, "
1332 "cannot parse further.\n", EBML_MAX_DEPTH);
31ad14c2
AC
1333 ret = AVERROR_INVALIDDATA;
1334 } else {
84cfce9f
KC
1335 level.start = 0;
1336 level.length = (uint64_t) -1;
f47ac3c6
AK
1337 matroska->levels[matroska->num_levels] = level;
1338 matroska->num_levels++;
84cfce9f 1339 matroska->current_id = 0;
f47ac3c6 1340
4a9628f2 1341 ret = ebml_parse(matroska, matroska_segment, matroska);
f47ac3c6
AK
1342
1343 /* remove dummy level */
1344 while (matroska->num_levels) {
1345 uint64_t length = matroska->levels[--matroska->num_levels].length;
84cfce9f 1346 if (length == (uint64_t) -1)
f47ac3c6
AK
1347 break;
1348 }
4348571b 1349 }
13b350a3 1350 }
4348571b 1351 /* seek back */
6b4aa5da 1352 avio_seek(matroska->ctx->pb, before_pos, SEEK_SET);
84cfce9f 1353 matroska->level_up = level_up;
c3ade62c 1354 matroska->current_id = saved_id;
31ad14c2
AC
1355
1356 return ret;
1357}
1358
1359static void matroska_execute_seekhead(MatroskaDemuxContext *matroska)
1360{
1361 EbmlList *seekhead_list = &matroska->seekhead;
31ad14c2
AC
1362 int64_t before_pos = avio_tell(matroska->ctx->pb);
1363 int i;
1364
1365 // we should not do any seeking in the streaming case
1366 if (!matroska->ctx->pb->seekable ||
1367 (matroska->ctx->flags & AVFMT_FLAG_IGNIDX))
1368 return;
1369
1370 for (i = 0; i < seekhead_list->nb_elem; i++) {
faaec467 1371 MatroskaSeekhead *seekhead = seekhead_list->elem;
31ad14c2
AC
1372 if (seekhead[i].pos <= before_pos)
1373 continue;
1374
1375 // defer cues parsing until we actually need cue data.
1376 if (seekhead[i].id == MATROSKA_ID_CUES) {
1377 matroska->cues_parsing_deferred = 1;
1378 continue;
1379 }
1380
1381 if (matroska_parse_seekhead_entry(matroska, i) < 0)
1382 break;
1383 }
1384}
1385
84cfce9f
KC
1386static void matroska_parse_cues(MatroskaDemuxContext *matroska)
1387{
31ad14c2
AC
1388 EbmlList *seekhead_list = &matroska->seekhead;
1389 MatroskaSeekhead *seekhead = seekhead_list->elem;
1390 EbmlList *index_list;
1391 MatroskaIndex *index;
1392 int index_scale = 1;
1393 int i, j;
1394
1395 for (i = 0; i < seekhead_list->nb_elem; i++)
fdb94444 1396 if (seekhead[i].id == MATROSKA_ID_CUES)
31ad14c2
AC
1397 break;
1398 assert(i <= seekhead_list->nb_elem);
1399
1400 matroska_parse_seekhead_entry(matroska, i);
1401
1402 index_list = &matroska->index;
84cfce9f
KC
1403 index = index_list->elem;
1404 if (index_list->nb_elem &&
1405 index[0].time > 1E14 / matroska->time_scale) {
31ad14c2
AC
1406 av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n");
1407 index_scale = matroska->time_scale;
1408 }
1409 for (i = 0; i < index_list->nb_elem; i++) {
84cfce9f 1410 EbmlList *pos_list = &index[i].pos;
31ad14c2
AC
1411 MatroskaIndexPos *pos = pos_list->elem;
1412 for (j = 0; j < pos_list->nb_elem; j++) {
84cfce9f
KC
1413 MatroskaTrack *track = matroska_find_track_by_num(matroska,
1414 pos[j].track);
31ad14c2
AC
1415 if (track && track->stream)
1416 av_add_index_entry(track->stream,
1417 pos[j].pos + matroska->segment_start,
84cfce9f 1418 index[i].time / index_scale, 0, 0,
31ad14c2
AC
1419 AVINDEX_KEYFRAME);
1420 }
1421 }
b061d892
DC
1422}
1423
f7b9687c 1424static int matroska_aac_profile(char *codec_id)
b061d892 1425{
84cfce9f 1426 static const char *const aac_profiles[] = { "MAIN", "LC", "SSR" };
b061d892
DC
1427 int profile;
1428
84cfce9f 1429 for (profile = 0; profile < FF_ARRAY_ELEMS(aac_profiles); profile++)
b061d892
DC
1430 if (strstr(codec_id, aac_profiles[profile]))
1431 break;
1432 return profile + 1;
1433}
1434
f7b9687c 1435static int matroska_aac_sri(int samplerate)
b061d892 1436{
b061d892
DC
1437 int sri;
1438
84cfce9f 1439 for (sri = 0; sri < FF_ARRAY_ELEMS(avpriv_mpeg4audio_sample_rates); sri++)
59a9a235 1440 if (avpriv_mpeg4audio_sample_rates[sri] == samplerate)
b061d892
DC
1441 break;
1442 return sri;
1443}
1444
4efdadc8
AK
1445static int matroska_parse_flac(AVFormatContext *s,
1446 MatroskaTrack *track,
1447 int *offset)
1448{
23f741f7 1449 AVStream *st = track->stream;
4efdadc8
AK
1450 uint8_t *p = track->codec_priv.data;
1451 int size = track->codec_priv.size;
1452
1453 if (size < 8 + FLAC_STREAMINFO_SIZE || p[4] & 0x7f) {
1454 av_log(s, AV_LOG_WARNING, "Invalid FLAC private data\n");
1455 track->codec_priv.size = 0;
1456 return 0;
1457 }
1458 *offset = 8;
1459 track->codec_priv.size = 8 + FLAC_STREAMINFO_SIZE;
1460
23f741f7
AK
1461 p += track->codec_priv.size;
1462 size -= track->codec_priv.size;
1463
1464 /* parse the remaining metadata blocks if present */
1465 while (size >= 4) {
1466 int block_last, block_type, block_size;
1467
1468 flac_parse_block_header(p, &block_last, &block_type, &block_size);
1469
1470 p += 4;
1471 size -= 4;
1472 if (block_size > size)
1473 return 0;
1474
1475 /* check for the channel mask */
1476 if (block_type == FLAC_METADATA_TYPE_VORBIS_COMMENT) {
1477 AVDictionary *dict = NULL;
1478 AVDictionaryEntry *chmask;
1479
1480 ff_vorbis_comment(s, &dict, p, block_size, 0);
1481 chmask = av_dict_get(dict, "WAVEFORMATEXTENSIBLE_CHANNEL_MASK", NULL, 0);
1482 if (chmask) {
1483 uint64_t mask = strtol(chmask->value, NULL, 0);
1484 if (!mask || mask & ~0x3ffffULL) {
1485 av_log(s, AV_LOG_WARNING,
1486 "Invalid value of WAVEFORMATEXTENSIBLE_CHANNEL_MASK\n");
1487 } else
1488 st->codec->channel_layout = mask;
1489 }
1490 av_dict_free(&dict);
1491 }
1492
1493 p += block_size;
1494 size -= block_size;
1495 }
1496
4efdadc8
AK
1497 return 0;
1498}
1499
6df478bf 1500static int matroska_parse_tracks(AVFormatContext *s)
b061d892
DC
1501{
1502 MatroskaDemuxContext *matroska = s->priv_data;
6df478bf 1503 MatroskaTrack *tracks = matroska->tracks.elem;
9a9a3b03 1504 AVStream *st;
4efdadc8 1505 int i, j, ret;
9c25bafa 1506
84cfce9f 1507 for (i = 0; i < matroska->tracks.nb_elem; i++) {
d88d806b 1508 MatroskaTrack *track = &tracks[i];
36ef5369 1509 enum AVCodecID codec_id = AV_CODEC_ID_NONE;
8b516f15 1510 EbmlList *encodings_list = &track->encodings;
9c25bafa 1511 MatroskaTrackEncoding *encodings = encodings_list->elem;
d88d806b
AJ
1512 uint8_t *extradata = NULL;
1513 int extradata_size = 0;
1514 int extradata_offset = 0;
ae628ec1 1515 AVIOContext b;
d88d806b
AJ
1516
1517 /* Apply some sanity checks. */
9c25bafa
AJ
1518 if (track->type != MATROSKA_TRACK_TYPE_VIDEO &&
1519 track->type != MATROSKA_TRACK_TYPE_AUDIO &&
1520 track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1521 av_log(matroska->ctx, AV_LOG_INFO,
1522 "Unknown or unsupported track type %"PRIu64"\n",
1523 track->type);
1524 continue;
1525 }
f929ab05 1526 if (!track->codec_id)
d88d806b
AJ
1527 continue;
1528
9c25bafa 1529 if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
3c583002 1530 if (!track->default_duration && track->video.frame_rate > 0)
84cfce9f 1531 track->default_duration = 1000000000 / track->video.frame_rate;
9c25bafa
AJ
1532 if (!track->video.display_width)
1533 track->video.display_width = track->video.pixel_width;
1534 if (!track->video.display_height)
1535 track->video.display_height = track->video.pixel_height;
1536 } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1537 if (!track->audio.out_samplerate)
1538 track->audio.out_samplerate = track->audio.samplerate;
1539 }
1540 if (encodings_list->nb_elem > 1) {
1541 av_log(matroska->ctx, AV_LOG_ERROR,
d7d2f0e6 1542 "Multiple combined encodings not supported");
9c25bafa
AJ
1543 } else if (encodings_list->nb_elem == 1) {
1544 if (encodings[0].type ||
2a91ada8 1545 (
b250f9c6 1546#if CONFIG_ZLIB
84cfce9f 1547 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
9c25bafa 1548#endif
b250f9c6 1549#if CONFIG_BZLIB
9c25bafa
AJ
1550 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
1551#endif
2a91ada8 1552#if CONFIG_LZO
84cfce9f 1553 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_LZO &&
2a91ada8
DB
1554#endif
1555 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP)) {
9c25bafa
AJ
1556 encodings[0].scope = 0;
1557 av_log(matroska->ctx, AV_LOG_ERROR,
1558 "Unsupported encoding type");
84cfce9f 1559 } else if (track->codec_priv.size && encodings[0].scope & 2) {
9c25bafa 1560 uint8_t *codec_priv = track->codec_priv.data;
8d4dd55c
LB
1561 int ret = matroska_decode_buffer(&track->codec_priv.data,
1562 &track->codec_priv.size,
1563 track);
1564 if (ret < 0) {
9c25bafa
AJ
1565 track->codec_priv.data = NULL;
1566 track->codec_priv.size = 0;
1567 av_log(matroska->ctx, AV_LOG_ERROR,
1568 "Failed to decode codec private data\n");
9c25bafa 1569 }
8d4dd55c 1570
9c25bafa
AJ
1571 if (codec_priv != track->codec_priv.data)
1572 av_free(codec_priv);
1573 }
1574 }
1575
84cfce9f
KC
1576 for (j = 0; ff_mkv_codec_tags[j].id != AV_CODEC_ID_NONE; j++) {
1577 if (!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
1578 strlen(ff_mkv_codec_tags[j].str))) {
1579 codec_id = ff_mkv_codec_tags[j].id;
d88d806b 1580 break;
b061d892 1581 }
d88d806b 1582 }
b061d892 1583
3b3bbdd3 1584 st = track->stream = avformat_new_stream(s, NULL);
f929ab05 1585 if (!st)
d88d806b
AJ
1586 return AVERROR(ENOMEM);
1587
84cfce9f
KC
1588 if (!strcmp(track->codec_id, "V_MS/VFW/FOURCC") &&
1589 track->codec_priv.size >= 40 &&
4b1f5e50 1590 track->codec_priv.data) {
84cfce9f 1591 track->ms_compat = 1;
d88d806b 1592 track->video.fourcc = AV_RL32(track->codec_priv.data + 16);
84cfce9f
KC
1593 codec_id = ff_codec_get_id(ff_codec_bmp_tags,
1594 track->video.fourcc);
1595 extradata_offset = 40;
1596 } else if (!strcmp(track->codec_id, "A_MS/ACM") &&
1597 track->codec_priv.size >= 14 &&
4b1f5e50 1598 track->codec_priv.data) {
ca402f32 1599 int ret;
84cfce9f
KC
1600 ffio_init_context(&b, track->codec_priv.data,
1601 track->codec_priv.size,
37cb3b18 1602 0, NULL, NULL, NULL, NULL);
ca402f32
MH
1603 ret = ff_get_wav_header(&b, st->codec, track->codec_priv.size);
1604 if (ret < 0)
1605 return ret;
84cfce9f 1606 codec_id = st->codec->codec_id;
038146e9 1607 extradata_offset = FFMIN(track->codec_priv.size, 18);
84cfce9f
KC
1608 } else if (!strcmp(track->codec_id, "V_QUICKTIME") &&
1609 (track->codec_priv.size >= 86) &&
4b1f5e50 1610 (track->codec_priv.data)) {
8286ae42
LB
1611 if (track->codec_priv.size == AV_RB32(track->codec_priv.data)) {
1612 track->video.fourcc = AV_RL32(track->codec_priv.data + 4);
1613 codec_id = ff_codec_get_id(ff_codec_movvideo_tags,
1614 track->video.fourcc);
1615 }
1616 if (codec_id == AV_CODEC_ID_NONE) {
1617 track->video.fourcc = AV_RL32(track->codec_priv.data);
1618 codec_id = ff_codec_get_id(ff_codec_movvideo_tags,
1619 track->video.fourcc);
1620 }
36ef5369 1621 } else if (codec_id == AV_CODEC_ID_PCM_S16BE) {
eb9cf50a 1622 switch (track->audio.bitdepth) {
84cfce9f
KC
1623 case 8:
1624 codec_id = AV_CODEC_ID_PCM_U8;
1625 break;
1626 case 24:
1627 codec_id = AV_CODEC_ID_PCM_S24BE;
1628 break;
1629 case 32:
1630 codec_id = AV_CODEC_ID_PCM_S32BE;
1631 break;
eb9cf50a 1632 }
36ef5369 1633 } else if (codec_id == AV_CODEC_ID_PCM_S16LE) {
eb9cf50a 1634 switch (track->audio.bitdepth) {
84cfce9f
KC
1635 case 8:
1636 codec_id = AV_CODEC_ID_PCM_U8;
1637 break;
1638 case 24:
1639 codec_id = AV_CODEC_ID_PCM_S24LE;
1640 break;
1641 case 32:
1642 codec_id = AV_CODEC_ID_PCM_S32LE;
1643 break;
eb9cf50a 1644 }
84cfce9f
KC
1645 } else if (codec_id == AV_CODEC_ID_PCM_F32LE &&
1646 track->audio.bitdepth == 64) {
36ef5369
AK
1647 codec_id = AV_CODEC_ID_PCM_F64LE;
1648 } else if (codec_id == AV_CODEC_ID_AAC && !track->codec_priv.size) {
d88d806b 1649 int profile = matroska_aac_profile(track->codec_id);
84cfce9f
KC
1650 int sri = matroska_aac_sri(track->audio.samplerate);
1651 extradata = av_mallocz(5 + FF_INPUT_BUFFER_PADDING_SIZE);
f929ab05 1652 if (!extradata)
d88d806b 1653 return AVERROR(ENOMEM);
84cfce9f
KC
1654 extradata[0] = (profile << 3) | ((sri & 0x0E) >> 1);
1655 extradata[1] = ((sri & 0x01) << 7) | (track->audio.channels << 3);
d88d806b 1656 if (strstr(track->codec_id, "SBR")) {
84cfce9f
KC
1657 sri = matroska_aac_sri(track->audio.out_samplerate);
1658 extradata[2] = 0x56;
1659 extradata[3] = 0xE5;
1660 extradata[4] = 0x80 | (sri << 3);
d88d806b 1661 extradata_size = 5;
16f97ab0 1662 } else
d88d806b 1663 extradata_size = 2;
8071dca3
MB
1664 } else if (codec_id == AV_CODEC_ID_ALAC && track->codec_priv.size) {
1665 /* Only ALAC's magic cookie is stored in Matroska's track headers.
84cfce9f
KC
1666 * Create the "atom size", "tag", and "tag version" fields the
1667 * decoder expects manually. */
8071dca3 1668 extradata_size = 12 + track->codec_priv.size;
84cfce9f
KC
1669 extradata = av_mallocz(extradata_size +
1670 FF_INPUT_BUFFER_PADDING_SIZE);
f929ab05 1671 if (!extradata)
8071dca3
MB
1672 return AVERROR(ENOMEM);
1673 AV_WB32(extradata, extradata_size);
1674 memcpy(&extradata[4], "alac", 4);
1675 AV_WB32(&extradata[8], 0);
1676 memcpy(&extradata[12], track->codec_priv.data,
84cfce9f 1677 track->codec_priv.size);
36ef5369 1678 } else if (codec_id == AV_CODEC_ID_TTA) {
d88d806b 1679 extradata_size = 30;
84cfce9f 1680 extradata = av_mallocz(extradata_size);
f929ab05 1681 if (!extradata)
d88d806b 1682 return AVERROR(ENOMEM);
e731b8d8 1683 ffio_init_context(&b, extradata, extradata_size, 1,
84cfce9f 1684 NULL, NULL, NULL, NULL);
77eb5504
AK
1685 avio_write(&b, "TTA1", 4);
1686 avio_wl16(&b, 1);
1687 avio_wl16(&b, track->audio.channels);
1688 avio_wl16(&b, track->audio.bitdepth);
1689 avio_wl32(&b, track->audio.out_samplerate);
84cfce9f
KC
1690 avio_wl32(&b, matroska->ctx->duration *
1691 track->audio.out_samplerate);
1692 } else if (codec_id == AV_CODEC_ID_RV10 ||
1693 codec_id == AV_CODEC_ID_RV20 ||
1694 codec_id == AV_CODEC_ID_RV30 ||
1695 codec_id == AV_CODEC_ID_RV40) {
d88d806b 1696 extradata_offset = 26;
36ef5369 1697 } else if (codec_id == AV_CODEC_ID_RA_144) {
d88d806b 1698 track->audio.out_samplerate = 8000;
84cfce9f
KC
1699 track->audio.channels = 1;
1700 } else if (codec_id == AV_CODEC_ID_RA_288 ||
1701 codec_id == AV_CODEC_ID_COOK ||
1702 codec_id == AV_CODEC_ID_ATRAC3 ||
1703 codec_id == AV_CODEC_ID_SIPR) {
6b10228a 1704 int flavor;
84cfce9f
KC
1705 ffio_init_context(&b, track->codec_priv.data,
1706 track->codec_priv.size,
1707 0, NULL, NULL, NULL, NULL);
45a8a02a 1708 avio_skip(&b, 22);
b7effd4e
AK
1709 flavor = avio_rb16(&b);
1710 track->audio.coded_framesize = avio_rb32(&b);
45a8a02a 1711 avio_skip(&b, 12);
b7effd4e
AK
1712 track->audio.sub_packet_h = avio_rb16(&b);
1713 track->audio.frame_size = avio_rb16(&b);
1714 track->audio.sub_packet_size = avio_rb16(&b);
0eab1f64 1715 if (flavor < 0 ||
84cfce9f
KC
1716 track->audio.coded_framesize <= 0 ||
1717 track->audio.sub_packet_h <= 0 ||
1718 track->audio.frame_size <= 0 ||
569d18aa
MS
1719 track->audio.sub_packet_size <= 0)
1720 return AVERROR_INVALIDDATA;
84cfce9f
KC
1721 track->audio.buf = av_malloc(track->audio.frame_size *
1722 track->audio.sub_packet_h);
36ef5369 1723 if (codec_id == AV_CODEC_ID_RA_288) {
d88d806b
AJ
1724 st->codec->block_align = track->audio.coded_framesize;
1725 track->codec_priv.size = 0;
1726 } else {
36ef5369 1727 if (codec_id == AV_CODEC_ID_SIPR && flavor < 4) {
6b10228a
AJ
1728 const int sipr_bit_rate[4] = { 6504, 8496, 5000, 16000 };
1729 track->audio.sub_packet_size = ff_sipr_subpk_size[flavor];
84cfce9f 1730 st->codec->bit_rate = sipr_bit_rate[flavor];
6b10228a 1731 }
d88d806b 1732 st->codec->block_align = track->audio.sub_packet_size;
84cfce9f 1733 extradata_offset = 78;
b061d892 1734 }
4efdadc8
AK
1735 } else if (codec_id == AV_CODEC_ID_FLAC && track->codec_priv.size) {
1736 ret = matroska_parse_flac(s, track, &extradata_offset);
1737 if (ret < 0)
1738 return ret;
d88d806b 1739 }
e2644407 1740 track->codec_priv.size -= extradata_offset;
b061d892 1741
36ef5369 1742 if (codec_id == AV_CODEC_ID_NONE)
d88d806b 1743 av_log(matroska->ctx, AV_LOG_INFO,
36ef5369 1744 "Unknown/unsupported AVCodecID %s.\n", track->codec_id);
eabb8ba4 1745
3fc9d7ce
AJ
1746 if (track->time_scale < 0.01)
1747 track->time_scale = 1.0;
84cfce9f
KC
1748 avpriv_set_pts_info(st, 64, matroska->time_scale * track->time_scale,
1749 1000 * 1000 * 1000); /* 64 bit pts in ns */
eabb8ba4 1750
eb3b5501
AK
1751 /* convert the delay from ns to the track timebase */
1752 track->codec_delay = av_rescale_q(track->codec_delay,
1753 (AVRational){ 1, 1000000000 },
1754 st->time_base);
1755
d88d806b 1756 st->codec->codec_id = codec_id;
84cfce9f 1757 st->start_time = 0;
d88d806b 1758 if (strcmp(track->language, "und"))
d2d67e42
AK
1759 av_dict_set(&st->metadata, "language", track->language, 0);
1760 av_dict_set(&st->metadata, "title", track->name, 0);
b061d892 1761
d88d806b
AJ
1762 if (track->flag_default)
1763 st->disposition |= AV_DISPOSITION_DEFAULT;
7a617a86
AJ
1764 if (track->flag_forced)
1765 st->disposition |= AV_DISPOSITION_FORCED;
b061d892 1766
ff0d5a74 1767 if (!st->codec->extradata) {
84cfce9f
KC
1768 if (extradata) {
1769 st->codec->extradata = extradata;
553e9f75 1770 st->codec->extradata_size = extradata_size;
84cfce9f 1771 } else if (track->codec_priv.data && track->codec_priv.size > 0) {
553e9f75
AJ
1772 st->codec->extradata = av_mallocz(track->codec_priv.size +
1773 FF_INPUT_BUFFER_PADDING_SIZE);
f929ab05 1774 if (!st->codec->extradata)
553e9f75
AJ
1775 return AVERROR(ENOMEM);
1776 st->codec->extradata_size = track->codec_priv.size;
1777 memcpy(st->codec->extradata,
1778 track->codec_priv.data + extradata_offset,
1779 track->codec_priv.size);
1780 }
ff0d5a74 1781 }
d88d806b
AJ
1782
1783 if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
72415b2a 1784 st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
d88d806b 1785 st->codec->codec_tag = track->video.fourcc;
84cfce9f
KC
1786 st->codec->width = track->video.pixel_width;
1787 st->codec->height = track->video.pixel_height;
59729451
AJ
1788 av_reduce(&st->sample_aspect_ratio.num,
1789 &st->sample_aspect_ratio.den,
d88d806b 1790 st->codec->height * track->video.display_width,
84cfce9f 1791 st->codec->width * track->video.display_height,
d88d806b 1792 255);
959bea13
YN
1793 if (st->codec->codec_id != AV_CODEC_ID_H264 &&
1794 st->codec->codec_id != AV_CODEC_ID_HEVC)
1795 st->need_parsing = AVSTREAM_PARSE_HEADERS;
ac97d47d 1796 if (track->default_duration) {
aba232cf 1797 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
ac97d47d 1798 1000000000, track->default_duration, 30000);
ac97d47d 1799 }
152e09fd
VG
1800 // add stream level stereo3d side data if it is a supported format
1801 if (track->video.stereo_mode < MATROSKA_VIDEO_STEREOMODE_TYPE_NB &&
1802 track->video.stereo_mode != 10 && track->video.stereo_mode != 12) {
1803 int ret = ff_mkv_stereo3d_conv(st, track->video.stereo_mode);
1804 if (ret < 0)
1805 return ret;
1806 }
d88d806b 1807 } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
84cfce9f 1808 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
d88d806b 1809 st->codec->sample_rate = track->audio.out_samplerate;
84cfce9f 1810 st->codec->channels = track->audio.channels;
36ef5369 1811 if (st->codec->codec_id != AV_CODEC_ID_AAC)
84cfce9f 1812 st->need_parsing = AVSTREAM_PARSE_HEADERS;
d45ccd2b
LB
1813 if (st->codec->codec_id == AV_CODEC_ID_MP3)
1814 st->need_parsing = AVSTREAM_PARSE_FULL;
d88d806b 1815 } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
72415b2a 1816 st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
36ef5369 1817 if (st->codec->codec_id == AV_CODEC_ID_SSA)
8336eb6f 1818 matroska->contains_ssa = 1;
d88d806b 1819 }
b061d892
DC
1820 }
1821
6df478bf
AK
1822 return 0;
1823}
1824
1825static int matroska_read_header(AVFormatContext *s)
1826{
1827 MatroskaDemuxContext *matroska = s->priv_data;
1828 EbmlList *attachments_list = &matroska->attachments;
1829 EbmlList *chapters_list = &matroska->chapters;
1830 MatroskaAttachment *attachments;
1831 MatroskaChapter *chapters;
1832 uint64_t max_start = 0;
1833 int64_t pos;
1834 Ebml ebml = { 0 };
1835 int i, j, res;
1836
1837 matroska->ctx = s;
1838
1839 /* First read the EBML header. */
3e1c9da3
TG
1840 if (ebml_parse(matroska, ebml_syntax, &ebml) || !ebml.doctype) {
1841 av_log(matroska->ctx, AV_LOG_ERROR, "EBML header parsing failed\n");
1842 ebml_free(ebml_syntax, &ebml);
1843 return AVERROR_INVALIDDATA;
1844 }
1845 if (ebml.version > EBML_VERSION ||
6df478bf
AK
1846 ebml.max_size > sizeof(uint64_t) ||
1847 ebml.id_length > sizeof(uint32_t) ||
152e09fd 1848 ebml.doctype_version > 3) {
6df478bf
AK
1849 av_log(matroska->ctx, AV_LOG_ERROR,
1850 "EBML header using unsupported features\n"
1851 "(EBML version %"PRIu64", doctype %s, doc version %"PRIu64")\n",
1852 ebml.version, ebml.doctype, ebml.doctype_version);
1853 ebml_free(ebml_syntax, &ebml);
1854 return AVERROR_PATCHWELCOME;
1855 }
1856 for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++)
1857 if (!strcmp(ebml.doctype, matroska_doctypes[i]))
1858 break;
1859 if (i >= FF_ARRAY_ELEMS(matroska_doctypes)) {
1860 av_log(s, AV_LOG_WARNING, "Unknown EBML doctype '%s'\n", ebml.doctype);
1861 if (matroska->ctx->error_recognition & AV_EF_EXPLODE) {
1862 ebml_free(ebml_syntax, &ebml);
1863 return AVERROR_INVALIDDATA;
1864 }
1865 }
1866 ebml_free(ebml_syntax, &ebml);
1867
1868 /* The next thing is a segment. */
1869 pos = avio_tell(matroska->ctx->pb);
1870 res = ebml_parse(matroska, matroska_segments, matroska);
1871 // try resyncing until we find a EBML_STOP type element.
1872 while (res != 1) {
1873 res = matroska_resync(matroska, pos);
1874 if (res < 0)
1875 return res;
1876 pos = avio_tell(matroska->ctx->pb);
1877 res = ebml_parse(matroska, matroska_segment, matroska);
1878 }
1879 matroska_execute_seekhead(matroska);
1880
1881 if (!matroska->time_scale)
1882 matroska->time_scale = 1000000;
1883 if (matroska->duration)
1884 matroska->ctx->duration = matroska->duration * matroska->time_scale *
1885 1000 / AV_TIME_BASE;
1886 av_dict_set(&s->metadata, "title", matroska->title, 0);
1887
1888 res = matroska_parse_tracks(s);
1889 if (res < 0)
1890 return res;
1891
f69befe5
DB
1892 attachments = attachments_list->elem;
1893 for (j = 0; j < attachments_list->nb_elem; j++) {
1894 if (!(attachments[j].filename && attachments[j].mime &&
1895 attachments[j].bin.data && attachments[j].bin.size > 0)) {
9c25bafa
AJ
1896 av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
1897 } else {
3b3bbdd3 1898 AVStream *st = avformat_new_stream(s, NULL);
f929ab05 1899 if (!st)
9c25bafa 1900 break;
f69befe5
DB
1901 av_dict_set(&st->metadata, "filename", attachments[j].filename, 0);
1902 av_dict_set(&st->metadata, "mimetype", attachments[j].mime, 0);
84cfce9f 1903 st->codec->codec_id = AV_CODEC_ID_NONE;
72415b2a 1904 st->codec->codec_type = AVMEDIA_TYPE_ATTACHMENT;
f69befe5 1905 st->codec->extradata = av_malloc(attachments[j].bin.size);
f929ab05 1906 if (!st->codec->extradata)
9c25bafa 1907 break;
f69befe5
DB
1908 st->codec->extradata_size = attachments[j].bin.size;
1909 memcpy(st->codec->extradata, attachments[j].bin.data,
1910 attachments[j].bin.size);
9c25bafa 1911
84cfce9f 1912 for (i = 0; ff_mkv_mime_tags[i].id != AV_CODEC_ID_NONE; i++) {
f69befe5 1913 if (!strncmp(ff_mkv_mime_tags[i].str, attachments[j].mime,
9c25bafa
AJ
1914 strlen(ff_mkv_mime_tags[i].str))) {
1915 st->codec->codec_id = ff_mkv_mime_tags[i].id;
1916 break;
1917 }
1918 }
f69befe5 1919 attachments[j].stream = st;
9c25bafa
AJ
1920 }
1921 }
1922
1923 chapters = chapters_list->elem;
84cfce9f
KC
1924 for (i = 0; i < chapters_list->nb_elem; i++)
1925 if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid &&
1926 (max_start == 0 || chapters[i].start > max_start)) {
6cb6e159 1927 chapters[i].chapter =
84cfce9f
KC
1928 avpriv_new_chapter(s, chapters[i].uid,
1929 (AVRational) { 1, 1000000000 },
1930 chapters[i].start, chapters[i].end,
1931 chapters[i].title);
d2d67e42 1932 av_dict_set(&chapters[i].chapter->metadata,
84cfce9f 1933 "title", chapters[i].title, 0);
e0e4be59
AJ
1934 max_start = chapters[i].start;
1935 }
9c25bafa 1936
929e9de7
AJ
1937 matroska_convert_tags(s);
1938
ce6f28bd 1939 return 0;
b061d892
DC
1940}
1941
737c40da 1942/*
737c40da
AJ
1943 * Put one packet in an application-supplied AVPacket struct.
1944 * Returns 0 on success or -1 on failure.
1945 */
1946static int matroska_deliver_packet(MatroskaDemuxContext *matroska,
1947 AVPacket *pkt)
1948{
1949 if (matroska->num_packets > 0) {
1950 memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
1951 av_free(matroska->packets[0]);
1952 if (matroska->num_packets > 1) {
77d2ef13 1953 void *newpackets;
737c40da
AJ
1954 memmove(&matroska->packets[0], &matroska->packets[1],
1955 (matroska->num_packets - 1) * sizeof(AVPacket *));
77d2ef13 1956 newpackets = av_realloc(matroska->packets,
84cfce9f
KC
1957 (matroska->num_packets - 1) *
1958 sizeof(AVPacket *));
77d2ef13
MN
1959 if (newpackets)
1960 matroska->packets = newpackets;
737c40da
AJ
1961 } else {
1962 av_freep(&matroska->packets);
8336eb6f 1963 matroska->prev_pkt = NULL;
737c40da
AJ
1964 }
1965 matroska->num_packets--;
1966 return 0;
1967 }
1968
1969 return -1;
1970}
1971
1972/*
1973 * Free all packets in our internal queue.
1974 */
1975static void matroska_clear_queue(MatroskaDemuxContext *matroska)
1976{
ae3d4163 1977 matroska->prev_pkt = NULL;
737c40da
AJ
1978 if (matroska->packets) {
1979 int n;
1980 for (n = 0; n < matroska->num_packets; n++) {
1981 av_free_packet(matroska->packets[n]);
1982 av_free(matroska->packets[n]);
1983 }
00a3431c 1984 av_freep(&matroska->packets);
737c40da
AJ
1985 matroska->num_packets = 0;
1986 }
1987}
1988
2d0e7713 1989static int matroska_parse_laces(MatroskaDemuxContext *matroska, uint8_t **buf,
84cfce9f 1990 int *buf_size, int type,
2d0e7713
LB
1991 uint32_t **lace_buf, int *laces)
1992{
fc43c19a 1993 int res = 0, n, size = *buf_size;
2d0e7713
LB
1994 uint8_t *data = *buf;
1995 uint32_t *lace_size;
1996
1997 if (!type) {
84cfce9f 1998 *laces = 1;
2d0e7713
LB
1999 *lace_buf = av_mallocz(sizeof(int));
2000 if (!*lace_buf)
2001 return AVERROR(ENOMEM);
2002
2003 *lace_buf[0] = size;
2004 return 0;
2005 }
2006
2007 assert(size > 0);
84cfce9f
KC
2008 *laces = *data + 1;
2009 data += 1;
2010 size -= 1;
2d0e7713
LB
2011 lace_size = av_mallocz(*laces * sizeof(int));
2012 if (!lace_size)
2013 return AVERROR(ENOMEM);
2014
2015 switch (type) {
84cfce9f
KC
2016 case 0x1: /* Xiph lacing */
2017 {
2d0e7713
LB
2018 uint8_t temp;
2019 uint32_t total = 0;
2020 for (n = 0; res == 0 && n < *laces - 1; n++) {
2021 while (1) {
2022 if (size == 0) {
2023 res = AVERROR_EOF;
2024 break;
2025 }
84cfce9f 2026 temp = *data;
2d0e7713 2027 lace_size[n] += temp;
84cfce9f
KC
2028 data += 1;
2029 size -= 1;
2d0e7713
LB
2030 if (temp != 0xff)
2031 break;
2032 }
2033 total += lace_size[n];
2034 }
2035 if (size <= total) {
2036 res = AVERROR_INVALIDDATA;
2037 break;
2038 }
2039
2040 lace_size[n] = size - total;
2041 break;
2042 }
2043
2044 case 0x2: /* fixed-size lacing */
87b017a2 2045 if (size % (*laces)) {
2d0e7713
LB
2046 res = AVERROR_INVALIDDATA;
2047 break;
2048 }
2049 for (n = 0; n < *laces; n++)
2050 lace_size[n] = size / *laces;
2051 break;
2052
84cfce9f
KC
2053 case 0x3: /* EBML lacing */
2054 {
2d0e7713 2055 uint64_t num;
8a96df7b 2056 uint64_t total;
2d0e7713
LB
2057 n = matroska_ebmlnum_uint(matroska, data, size, &num);
2058 if (n < 0) {
2059 av_log(matroska->ctx, AV_LOG_INFO,
2060 "EBML block data error\n");
2061 res = n;
2062 break;
2063 }
2064 data += n;
2065 size -= n;
2066 total = lace_size[0] = num;
2067 for (n = 1; res == 0 && n < *laces - 1; n++) {
2068 int64_t snum;
2069 int r;
2070 r = matroska_ebmlnum_sint(matroska, data, size, &snum);
2071 if (r < 0) {
2072 av_log(matroska->ctx, AV_LOG_INFO,
2073 "EBML block data error\n");
2074 res = r;
2075 break;
2076 }
84cfce9f
KC
2077 data += r;
2078 size -= r;
2d0e7713 2079 lace_size[n] = lace_size[n - 1] + snum;
84cfce9f 2080 total += lace_size[n];
2d0e7713
LB
2081 }
2082 if (size <= total) {
2083 res = AVERROR_INVALIDDATA;
2084 break;
2085 }
2086 lace_size[*laces - 1] = size - total;
2087 break;
2088 }
2089 }
2090
2091 *buf = data;
2092 *lace_buf = lace_size;
fc43c19a 2093 *buf_size = size;
2d0e7713
LB
2094
2095 return res;
2096}
2097
c831ebf6 2098static int matroska_parse_rm_audio(MatroskaDemuxContext *matroska,
84cfce9f
KC
2099 MatroskaTrack *track, AVStream *st,
2100 uint8_t *data, int size, uint64_t timecode,
2101 uint64_t duration, int64_t pos)
c831ebf6
LB
2102{
2103 int a = st->codec->block_align;
2104 int sps = track->audio.sub_packet_size;
2105 int cfs = track->audio.coded_framesize;
84cfce9f
KC
2106 int h = track->audio.sub_packet_h;
2107 int y = track->audio.sub_packet_cnt;
2108 int w = track->audio.frame_size;
c831ebf6
LB
2109 int x;
2110
2111 if (!track->audio.pkt_cnt) {
2112 if (track->audio.sub_packet_cnt == 0)
2113 track->audio.buf_timecode = timecode;
2114 if (st->codec->codec_id == AV_CODEC_ID_RA_288) {
2115 if (size < cfs * h / 2) {
2116 av_log(matroska->ctx, AV_LOG_ERROR,
2117 "Corrupt int4 RM-style audio packet size\n");
2118 return AVERROR_INVALIDDATA;
2119 }
84cfce9f
KC
2120 for (x = 0; x < h / 2; x++)
2121 memcpy(track->audio.buf + x * 2 * w + y * cfs,
2122 data + x * cfs, cfs);
c831ebf6
LB
2123 } else if (st->codec->codec_id == AV_CODEC_ID_SIPR) {
2124 if (size < w) {
2125 av_log(matroska->ctx, AV_LOG_ERROR,
2126 "Corrupt sipr RM-style audio packet size\n");
2127 return AVERROR_INVALIDDATA;
2128 }
84cfce9f 2129 memcpy(track->audio.buf + y * w, data, w);
c831ebf6
LB
2130 } else {
2131 if (size < sps * w / sps) {
2132 av_log(matroska->ctx, AV_LOG_ERROR,
2133 "Corrupt generic RM-style audio packet size\n");
2134 return AVERROR_INVALIDDATA;
2135 }
84cfce9f
KC
2136 for (x = 0; x < w / sps; x++)
2137 memcpy(track->audio.buf +
2138 sps * (h * x + ((h + 1) / 2) * (y & 1) + (y >> 1)),
2139 data + x * sps, sps);
c831ebf6
LB
2140 }
2141
2142 if (++track->audio.sub_packet_cnt >= h) {
2143 if (st->codec->codec_id == AV_CODEC_ID_SIPR)
2144 ff_rm_reorder_sipr_data(track->audio.buf, h, w);
2145 track->audio.sub_packet_cnt = 0;
84cfce9f 2146 track->audio.pkt_cnt = h * w / a;
c831ebf6
LB
2147 }
2148 }
2149
2150 while (track->audio.pkt_cnt) {
2151 AVPacket *pkt = av_mallocz(sizeof(AVPacket));
2152 av_new_packet(pkt, a);
84cfce9f
KC
2153 memcpy(pkt->data,
2154 track->audio.buf + a * (h * w / a - track->audio.pkt_cnt--),
2155 a);
2156 pkt->pts = track->audio.buf_timecode;
c831ebf6 2157 track->audio.buf_timecode = AV_NOPTS_VALUE;
84cfce9f
KC
2158 pkt->pos = pos;
2159 pkt->stream_index = st->index;
2160 dynarray_add(&matroska->packets, &matroska->num_packets, pkt);
c831ebf6
LB
2161 }
2162
2163 return 0;
2164}
9b6f47c4
AK
2165
2166/* reconstruct full wavpack blocks from mangled matroska ones */
2167static int matroska_parse_wavpack(MatroskaTrack *track, uint8_t *src,
2168 uint8_t **pdst, int *size)
2169{
2170 uint8_t *dst = NULL;
2171 int dstlen = 0;
2172 int srclen = *size;
2173 uint32_t samples;
2174 uint16_t ver;
2175 int ret, offset = 0;
2176
2177 if (srclen < 12 || track->stream->codec->extradata_size < 2)
2178 return AVERROR_INVALIDDATA;
2179
2180 ver = AV_RL16(track->stream->codec->extradata);
2181
2182 samples = AV_RL32(src);
2183 src += 4;
2184 srclen -= 4;
2185
2186 while (srclen >= 8) {
2187 int multiblock;
2188 uint32_t blocksize;
2189 uint8_t *tmp;
2190
2191 uint32_t flags = AV_RL32(src);
2192 uint32_t crc = AV_RL32(src + 4);
2193 src += 8;
2194 srclen -= 8;
2195
2196 multiblock = (flags & 0x1800) != 0x1800;
2197 if (multiblock) {
2198 if (srclen < 4) {
2199 ret = AVERROR_INVALIDDATA;
2200 goto fail;
2201 }
2202 blocksize = AV_RL32(src);
84cfce9f
KC
2203 src += 4;
2204 srclen -= 4;
9b6f47c4
AK
2205 } else
2206 blocksize = srclen;
2207
2208 if (blocksize > srclen) {
2209 ret = AVERROR_INVALIDDATA;
2210 goto fail;
2211 }
2212
2213 tmp = av_realloc(dst, dstlen + blocksize + 32);
2214 if (!tmp) {
2215 ret = AVERROR(ENOMEM);
2216 goto fail;
2217 }
2218 dst = tmp;
2219 dstlen += blocksize + 32;
2220
84cfce9f
KC
2221 AV_WL32(dst + offset, MKTAG('w', 'v', 'p', 'k')); // tag
2222 AV_WL32(dst + offset + 4, blocksize + 24); // blocksize - 8
2223 AV_WL16(dst + offset + 8, ver); // version
2224 AV_WL16(dst + offset + 10, 0); // track/index_no
2225 AV_WL32(dst + offset + 12, 0); // total samples
2226 AV_WL32(dst + offset + 16, 0); // block index
2227 AV_WL32(dst + offset + 20, samples); // number of samples
2228 AV_WL32(dst + offset + 24, flags); // flags
2229 AV_WL32(dst + offset + 28, crc); // crc
2230 memcpy(dst + offset + 32, src, blocksize); // block data
9b6f47c4
AK
2231
2232 src += blocksize;
2233 srclen -= blocksize;
2234 offset += blocksize + 32;
2235 }
2236
2237 *pdst = dst;
2238 *size = dstlen;
2239
2240 return 0;
2241
2242fail:
2243 av_freep(&dst);
2244 return ret;
2245}
2246
c831ebf6 2247static int matroska_parse_frame(MatroskaDemuxContext *matroska,
84cfce9f 2248 MatroskaTrack *track, AVStream *st,
c831ebf6
LB
2249 uint8_t *data, int pkt_size,
2250 uint64_t timecode, uint64_t duration,
2251 int64_t pos, int is_keyframe)
2252{
2253 MatroskaTrackEncoding *encodings = track->encodings.elem;
2254 uint8_t *pkt_data = data;
2255 int offset = 0, res;
2256 AVPacket *pkt;
2257
2258 if (encodings && encodings->scope & 1) {
2259 res = matroska_decode_buffer(&pkt_data, &pkt_size, track);
2260 if (res < 0)
2261 return res;
2262 }
2263
9b6f47c4
AK
2264 if (st->codec->codec_id == AV_CODEC_ID_WAVPACK) {
2265 uint8_t *wv_data;
2266 res = matroska_parse_wavpack(track, pkt_data, &wv_data, &pkt_size);
2267 if (res < 0) {
84cfce9f
KC
2268 av_log(matroska->ctx, AV_LOG_ERROR,
2269 "Error parsing a wavpack block.\n");
9b6f47c4
AK
2270 goto fail;
2271 }
2272 if (pkt_data != data)
2273 av_freep(&pkt_data);
2274 pkt_data = wv_data;
2275 }
2276
c831ebf6
LB
2277 if (st->codec->codec_id == AV_CODEC_ID_PRORES)
2278 offset = 8;
2279
2280 pkt = av_mallocz(sizeof(AVPacket));
2281 /* XXX: prevent data copy... */
2282 if (av_new_packet(pkt, pkt_size + offset) < 0) {
2283 av_free(pkt);
2284 return AVERROR(ENOMEM);
2285 }
2286
2287 if (st->codec->codec_id == AV_CODEC_ID_PRORES) {
2288 uint8_t *buf = pkt->data;
2289 bytestream_put_be32(&buf, pkt_size);
2290 bytestream_put_be32(&buf, MKBETAG('i', 'c', 'p', 'f'));
2291 }
2292
2293 memcpy(pkt->data + offset, pkt_data, pkt_size);
2294
2295 if (pkt_data != data)
2296 av_free(pkt_data);
2297
84cfce9f 2298 pkt->flags = is_keyframe;
c831ebf6
LB
2299 pkt->stream_index = st->index;
2300
2301 if (track->ms_compat)
2302 pkt->dts = timecode;
2303 else
2304 pkt->pts = timecode;
2305 pkt->pos = pos;
2306 if (st->codec->codec_id == AV_CODEC_ID_TEXT)
2307 pkt->convergence_duration = duration;
2308 else if (track->type != MATROSKA_TRACK_TYPE_SUBTITLE)
2309 pkt->duration = duration;
2310
2311 if (st->codec->codec_id == AV_CODEC_ID_SSA)
2312 matroska_fix_ass_packet(matroska, pkt, duration);
2313
84cfce9f
KC
2314 if (matroska->prev_pkt &&
2315 timecode != AV_NOPTS_VALUE &&
2316 matroska->prev_pkt->pts == timecode &&
2317 matroska->prev_pkt->stream_index == st->index &&
c831ebf6
LB
2318 st->codec->codec_id == AV_CODEC_ID_SSA)
2319 matroska_merge_packets(matroska->prev_pkt, pkt);
2320 else {
84cfce9f 2321 dynarray_add(&matroska->packets, &matroska->num_packets, pkt);
c831ebf6
LB
2322 matroska->prev_pkt = pkt;
2323 }
2324
2325 return 0;
84cfce9f 2326
9b6f47c4
AK
2327fail:
2328 if (pkt_data != data)
2329 av_freep(&pkt_data);
2330 return res;
c831ebf6
LB
2331}
2332
f7b9687c
AJ
2333static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data,
2334 int size, int64_t pos, uint64_t cluster_time,
7d843100 2335 uint64_t block_duration, int is_keyframe,
24c3da16 2336 int64_t cluster_pos)
b061d892 2337{
f14a201b 2338 uint64_t timecode = AV_NOPTS_VALUE;
009ecd50 2339 MatroskaTrack *track;
a3467f86 2340 int res = 0;
b061d892 2341 AVStream *st;
b061d892
DC
2342 int16_t block_time;
2343 uint32_t *lace_size = NULL;
2344 int n, flags, laces = 0;
7d843100 2345 uint64_t num, duration;
b061d892 2346
c1e01133 2347 if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) {
b061d892 2348 av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
721af294 2349 return n;
b061d892
DC
2350 }
2351 data += n;
2352 size -= n;
2353
b061d892 2354 track = matroska_find_track_by_num(matroska, num);
d31fb1a9 2355 if (!track || !track->stream) {
b061d892 2356 av_log(matroska->ctx, AV_LOG_INFO,
009ecd50 2357 "Invalid stream %"PRIu64" or size %u\n", num, size);
5dd514af 2358 return AVERROR_INVALIDDATA;
d31fb1a9
RB
2359 } else if (size <= 3)
2360 return 0;
009ecd50 2361 st = track->stream;
16f97ab0 2362 if (st->discard >= AVDISCARD_ALL)
b061d892 2363 return res;
b061d892 2364
2ce746c6 2365 block_time = AV_RB16(data);
84cfce9f
KC
2366 data += 2;
2367 flags = *data++;
2368 size -= 3;
b061d892 2369 if (is_keyframe == -1)
cc947f04 2370 is_keyframe = flags & 0x80 ? AV_PKT_FLAG_KEY : 0;
b061d892 2371
84cfce9f
KC
2372 if (cluster_time != (uint64_t) -1 &&
2373 (block_time >= 0 || cluster_time >= -block_time)) {
eb3b5501 2374 timecode = cluster_time + block_time - track->codec_delay;
84cfce9f
KC
2375 if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE &&
2376 timecode < track->end_timecode)
82360e67 2377 is_keyframe = 0; /* overlapping subtitles are not key frame */
a8fd7e76 2378 if (is_keyframe)
84cfce9f
KC
2379 av_add_index_entry(st, cluster_pos, timecode, 0, 0,
2380 AVINDEX_KEYFRAME);
f14a201b
AJ
2381 }
2382
84cfce9f
KC
2383 if (matroska->skip_to_keyframe &&
2384 track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
20f74665 2385 if (!is_keyframe || timecode < matroska->skip_to_timecode)
b061d892
DC
2386 return res;
2387 matroska->skip_to_keyframe = 0;
2388 }
2389
fc43c19a 2390 res = matroska_parse_laces(matroska, &data, &size, (flags & 0x06) >> 1,
2d0e7713 2391 &lace_size, &laces);
b061d892 2392
c831ebf6
LB
2393 if (res)
2394 goto end;
8d4dd55c 2395
7d843100
LB
2396 if (block_duration != AV_NOPTS_VALUE) {
2397 duration = block_duration / laces;
2398 if (block_duration != duration * laces) {
2399 av_log(matroska->ctx, AV_LOG_WARNING,
2400 "Incorrect block_duration, possibly corrupted container");
2401 }
2402 } else {
84cfce9f 2403 duration = track->default_duration / matroska->time_scale;
7d843100
LB
2404 block_duration = duration * laces;
2405 }
2406
2407 if (timecode != AV_NOPTS_VALUE)
2408 track->end_timecode =
2409 FFMAX(track->end_timecode, timecode + block_duration);
2410
c831ebf6
LB
2411 for (n = 0; n < laces; n++) {
2412 if ((st->codec->codec_id == AV_CODEC_ID_RA_288 ||
84cfce9f
KC
2413 st->codec->codec_id == AV_CODEC_ID_COOK ||
2414 st->codec->codec_id == AV_CODEC_ID_SIPR ||
c831ebf6 2415 st->codec->codec_id == AV_CODEC_ID_ATRAC3) &&
84cfce9f 2416 st->codec->block_align && track->audio.sub_packet_size) {
25a80a93
LB
2417 res = matroska_parse_rm_audio(matroska, track, st, data,
2418 lace_size[n],
c831ebf6
LB
2419 timecode, duration, pos);
2420 if (res)
2421 goto end;
c831ebf6
LB
2422 } else {
2423 res = matroska_parse_frame(matroska, track, st, data, lace_size[n],
84cfce9f
KC
2424 timecode, duration, pos,
2425 !n ? is_keyframe : 0);
c831ebf6
LB
2426 if (res)
2427 goto end;
b061d892 2428 }
c831ebf6
LB
2429
2430 if (timecode != AV_NOPTS_VALUE)
2431 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
2432 data += lace_size[n];
b061d892
DC
2433 }
2434
31168588 2435end:
b061d892 2436 av_free(lace_size);
a3467f86 2437 return res;
b061d892
DC
2438}
2439
8336eb6f
DC
2440static int matroska_parse_cluster_incremental(MatroskaDemuxContext *matroska)
2441{
2442 EbmlList *blocks_list;
2443 MatroskaBlock *blocks;
2444 int i, res;
2445 res = ebml_parse(matroska,
2446 matroska_cluster_incremental_parsing,
2447 &matroska->current_cluster);
2448 if (res == 1) {
2449 /* New Cluster */
2450 if (matroska->current_cluster_pos)
2451 ebml_level_end(matroska);
2452 ebml_free(matroska_cluster, &matroska->current_cluster);
2453 memset(&matroska->current_cluster, 0, sizeof(MatroskaCluster));
2454 matroska->current_cluster_num_blocks = 0;
84cfce9f
KC
2455 matroska->current_cluster_pos = avio_tell(matroska->ctx->pb);
2456 matroska->prev_pkt = NULL;
8336eb6f
DC
2457 /* sizeof the ID which was already read */
2458 if (matroska->current_id)
2459 matroska->current_cluster_pos -= 4;
2460 res = ebml_parse(matroska,
2461 matroska_clusters_incremental,
2462 &matroska->current_cluster);
2463 /* Try parsing the block again. */
2464 if (res == 1)
2465 res = ebml_parse(matroska,
2466 matroska_cluster_incremental_parsing,
2467 &matroska->current_cluster);
2468 }
2469
2470 if (!res &&
2471 matroska->current_cluster_num_blocks <
84cfce9f 2472 matroska->current_cluster.blocks.nb_elem) {
8336eb6f 2473 blocks_list = &matroska->current_cluster.blocks;
84cfce9f 2474 blocks = blocks_list->elem;
8336eb6f
DC
2475
2476 matroska->current_cluster_num_blocks = blocks_list->nb_elem;
84cfce9f 2477 i = blocks_list->nb_elem - 1;
8336eb6f
DC
2478 if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
2479 int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
2480 if (!blocks[i].non_simple)
2481 blocks[i].duration = AV_NOPTS_VALUE;
84cfce9f
KC
2482 res = matroska_parse_block(matroska, blocks[i].bin.data,
2483 blocks[i].bin.size, blocks[i].bin.pos,
8336eb6f
DC
2484 matroska->current_cluster.timecode,
2485 blocks[i].duration, is_keyframe,
2486 matroska->current_cluster_pos);
2487 }
2488 }
2489
84cfce9f
KC
2490 if (res < 0)
2491 matroska->done = 1;
8336eb6f
DC
2492 return res;
2493}
2494
f7b9687c 2495static int matroska_parse_cluster(MatroskaDemuxContext *matroska)
b061d892 2496{
209472b4
AJ
2497 MatroskaCluster cluster = { 0 };
2498 EbmlList *blocks_list;
2499 MatroskaBlock *blocks;
24c3da16 2500 int i, res;
8336eb6f 2501 int64_t pos;
84cfce9f 2502
8336eb6f
DC
2503 if (!matroska->contains_ssa)
2504 return matroska_parse_cluster_incremental(matroska);
2505 pos = avio_tell(matroska->ctx->pb);
d5e34dcb 2506 matroska->prev_pkt = NULL;
80702035 2507 if (matroska->current_id)
8bc98ba9 2508 pos -= 4; /* sizeof the ID which was already read */
84cfce9f 2509 res = ebml_parse(matroska, matroska_clusters, &cluster);
209472b4 2510 blocks_list = &cluster.blocks;
84cfce9f
KC
2511 blocks = blocks_list->elem;
2512 for (i = 0; i < blocks_list->nb_elem && !res; i++)
37dd2356 2513 if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
194d4b49 2514 int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
98a7d560
JS
2515 if (!blocks[i].non_simple)
2516 blocks[i].duration = AV_NOPTS_VALUE;
84cfce9f
KC
2517 res = matroska_parse_block(matroska, blocks[i].bin.data,
2518 blocks[i].bin.size, blocks[i].bin.pos,
2519 cluster.timecode, blocks[i].duration,
2520 is_keyframe, pos);
194d4b49 2521 }
209472b4 2522 ebml_free(matroska_cluster, &cluster);
b061d892
DC
2523 return res;
2524}
2525
f7b9687c 2526static int matroska_read_packet(AVFormatContext *s, AVPacket *pkt)
b061d892
DC
2527{
2528 MatroskaDemuxContext *matroska = s->priv_data;
5dd514af 2529 int ret = 0;
b061d892 2530
5dd514af 2531 while (!ret && matroska_deliver_packet(matroska, pkt)) {
8835c554 2532 int64_t pos = avio_tell(matroska->ctx->pb);
b061d892 2533 if (matroska->done)
9ebeea82 2534 return AVERROR_EOF;
8835c554
SM
2535 if (matroska_parse_cluster(matroska) < 0)
2536 ret = matroska_resync(matroska, pos);
b061d892
DC
2537 }
2538
bdb939ad 2539 if (ret == AVERROR_INVALIDDATA && pkt->data) {
0ca4642e
LB
2540 pkt->flags |= AV_PKT_FLAG_CORRUPT;
2541 return 0;
2542 }
2543
5dd514af 2544 return ret;
b061d892
DC
2545}
2546
f7b9687c
AJ
2547static int matroska_read_seek(AVFormatContext *s, int stream_index,
2548 int64_t timestamp, int flags)
b061d892
DC
2549{
2550 MatroskaDemuxContext *matroska = s->priv_data;
532c96a2 2551 MatroskaTrack *tracks = NULL;
b061d892 2552 AVStream *st = s->streams[stream_index];
c1658257 2553 int i, index, index_sub, index_min;
b061d892 2554
31ad14c2
AC
2555 /* Parse the CUES now since we need the index data to seek. */
2556 if (matroska->cues_parsing_deferred) {
2557 matroska_parse_cues(matroska);
2558 matroska->cues_parsing_deferred = 0;
2559 }
2560
a8fd7e76
AJ
2561 if (!st->nb_index_entries)
2562 return 0;
2563 timestamp = FFMAX(timestamp, st->index_entries[0].timestamp);
dfbbbdc0 2564
6bef5f9b 2565 if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
84cfce9f
KC
2566 avio_seek(s->pb, st->index_entries[st->nb_index_entries - 1].pos,
2567 SEEK_SET);
cdc2c1c5 2568 matroska->current_id = 0;
0dbddda3
AJ
2569 while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
2570 matroska_clear_queue(matroska);
2571 if (matroska_parse_cluster(matroska) < 0)
2572 break;
2573 }
6bef5f9b 2574 }
b061d892 2575
243cc4c3 2576 matroska_clear_queue(matroska);
0f646a22
AJ
2577 if (index < 0)
2578 return 0;
243cc4c3 2579
c1658257 2580 index_min = index;
532c96a2 2581 tracks = matroska->tracks.elem;
84cfce9f
KC
2582 for (i = 0; i < matroska->tracks.nb_elem; i++) {
2583 tracks[i].audio.pkt_cnt = 0;
46544204 2584 tracks[i].audio.sub_packet_cnt = 0;
84cfce9f
KC
2585 tracks[i].audio.buf_timecode = AV_NOPTS_VALUE;
2586 tracks[i].end_timecode = 0;
2587 if (tracks[i].type == MATROSKA_TRACK_TYPE_SUBTITLE &&
98569d89 2588 tracks[i].stream->discard != AVDISCARD_ALL) {
84cfce9f
KC
2589 index_sub = av_index_search_timestamp(
2590 tracks[i].stream, st->index_entries[index].timestamp,
2591 AVSEEK_FLAG_BACKWARD);
2592 if (index_sub >= 0 &&
2593 st->index_entries[index_sub].pos < st->index_entries[index_min].pos &&
2594 st->index_entries[index].timestamp -
2595 st->index_entries[index_sub].timestamp < 30000000000 / matroska->time_scale)
c1658257
AJ
2596 index_min = index_sub;
2597 }
2598 }
2599
6b4aa5da 2600 avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET);
84cfce9f 2601 matroska->current_id = 0;
b061d892 2602 matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
20f74665 2603 matroska->skip_to_timecode = st->index_entries[index].timestamp;
84cfce9f 2604 matroska->done = 0;
a2faa951 2605 ff_update_cur_dts(s, st, st->index_entries[index].timestamp);
b061d892
DC
2606 return 0;
2607}
2608
f7b9687c 2609static int matroska_read_close(AVFormatContext *s)
b061d892
DC
2610{
2611 MatroskaDemuxContext *matroska = s->priv_data;
2cbc8811 2612 MatroskaTrack *tracks = matroska->tracks.elem;
70109c0d 2613 int n;
b061d892 2614
34c9c1ba 2615 matroska_clear_queue(matroska);
b061d892 2616
84cfce9f 2617 for (n = 0; n < matroska->tracks.nb_elem; n++)
2cbc8811
AJ
2618 if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
2619 av_free(tracks[n].audio.buf);
8336eb6f 2620 ebml_free(matroska_cluster, &matroska->current_cluster);
ce6f28bd 2621 ebml_free(matroska_segment, matroska);
b061d892
DC
2622
2623 return 0;
2624}
2625
c6610a21 2626AVInputFormat ff_matroska_demuxer = {
dfc2c4d9 2627 .name = "matroska,webm",
6774247a 2628 .long_name = NULL_IF_CONFIG_SMALL("Matroska / WebM"),
5694831e 2629 .extensions = "mkv,mk3d,mka,mks",
dfc2c4d9
AK
2630 .priv_data_size = sizeof(MatroskaDemuxContext),
2631 .read_probe = matroska_probe,
2632 .read_header = matroska_read_header,
2633 .read_packet = matroska_read_packet,
2634 .read_close = matroska_read_close,
2635 .read_seek = matroska_read_seek,
fa38573c 2636 .mime_type = "audio/webm,audio/x-matroska,video/webm,video/x-matroska"
b061d892 2637};