matroska: Support V_QUICKTIME as written in the specification
[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
daf8cf35 89typedef struct EbmlList {
789ed100
AJ
90 int nb_elem;
91 void *elem;
92} EbmlList;
93
daf8cf35 94typedef struct EbmlBin {
789ed100
AJ
95 int size;
96 uint8_t *data;
97 int64_t pos;
98} EbmlBin;
99
daf8cf35 100typedef struct Ebml {
63511324
AJ
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
daf8cf35 108typedef struct MatroskaTrackCompression {
2cbc8811
AJ
109 uint64_t algo;
110 EbmlBin settings;
111} MatroskaTrackCompression;
b061d892 112
daf8cf35 113typedef struct MatroskaTrackEncoding {
2cbc8811
AJ
114 uint64_t scope;
115 uint64_t type;
116 MatroskaTrackCompression compression;
117} MatroskaTrackEncoding;
b061d892 118
daf8cf35 119typedef struct MatroskaTrackVideo {
2cbc8811
AJ
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;
d4ae8ac9 126 uint64_t stereo_mode;
2cbc8811 127} MatroskaTrackVideo;
b061d892 128
daf8cf35 129typedef struct MatroskaTrackAudio {
2cbc8811
AJ
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
daf8cf35 146typedef struct MatroskaTrack {
2cbc8811 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
daf8cf35 168typedef struct MatroskaAttachment {
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 176
daf8cf35 177typedef struct MatroskaChapter {
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
daf8cf35 186typedef struct MatroskaIndexPos {
e5929fdf
AJ
187 uint64_t track;
188 uint64_t pos;
189} MatroskaIndexPos;
190
daf8cf35 191typedef struct MatroskaIndex {
e5929fdf
AJ
192 uint64_t time;
193 EbmlList pos;
194} MatroskaIndex;
195
daf8cf35 196typedef struct MatroskaTag {
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
daf8cf35 204typedef struct MatroskaTagTarget {
929e9de7
AJ
205 char *type;
206 uint64_t typevalue;
207 uint64_t trackuid;
208 uint64_t chapteruid;
209 uint64_t attachuid;
210} MatroskaTagTarget;
211
daf8cf35 212typedef struct MatroskaTags {
929e9de7
AJ
213 MatroskaTagTarget target;
214 EbmlList tag;
215} MatroskaTags;
216
daf8cf35 217typedef struct MatroskaSeekhead {
13b350a3
AJ
218 uint64_t id;
219 uint64_t pos;
220} MatroskaSeekhead;
221
daf8cf35 222typedef struct MatroskaLevel {
8d75b5a2
AJ
223 uint64_t start;
224 uint64_t length;
b061d892
DC
225} MatroskaLevel;
226
daf8cf35 227typedef struct MatroskaCluster {
8336eb6f
DC
228 uint64_t timecode;
229 EbmlList blocks;
230} MatroskaCluster;
231
daf8cf35 232typedef struct MatroskaDemuxContext {
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
daf8cf35 276typedef struct MatroskaBlock {
209472b4
AJ
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 },
d4ae8ac9 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);
059a9348 750 if (!(bin->data = av_mallocz(length + AV_INPUT_BUFFER_PADDING_SIZE)))
737c40da
AJ
751 return AVERROR(ENOMEM);
752
753 bin->size = length;
a2704c97 754 bin->pos = avio_tell(pb);
b7effd4e 755 if (avio_read(pb, bin->data, length) != length) {
5549aa6d 756 av_freep(&bin->data);
737c40da 757 return AVERROR(EIO);
5549aa6d 758 }
737c40da
AJ
759
760 return 0;
761}
762
763/*
b061d892
DC
764 * Read the next element, but only the header. The contents
765 * are supposed to be sub-elements which can be read separately.
766 * 0 is success, < 0 is failure.
767 */
bddd1d93 768static int ebml_read_master(MatroskaDemuxContext *matroska, uint64_t length)
b061d892 769{
ae628ec1 770 AVIOContext *pb = matroska->ctx->pb;
b061d892 771 MatroskaLevel *level;
b061d892 772
b061d892
DC
773 if (matroska->num_levels >= EBML_MAX_DEPTH) {
774 av_log(matroska->ctx, AV_LOG_ERROR,
775 "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
85565db0 776 return AVERROR(ENOSYS);
b061d892
DC
777 }
778
84cfce9f
KC
779 level = &matroska->levels[matroska->num_levels++];
780 level->start = avio_tell(pb);
b061d892
DC
781 level->length = length;
782
783 return 0;
784}
785
786/*
b061d892 787 * Read signed/unsigned "EBML" numbers.
5968d2dd 788 * Return: number of bytes processed, < 0 on error
b061d892 789 */
c1e01133
AJ
790static int matroska_ebmlnum_uint(MatroskaDemuxContext *matroska,
791 uint8_t *data, uint32_t size, uint64_t *num)
b061d892 792{
ae628ec1 793 AVIOContext pb;
e731b8d8 794 ffio_init_context(&pb, data, size, 0, NULL, NULL, NULL, NULL);
465c28b6 795 return ebml_read_num(matroska, &pb, FFMIN(size, 8), num);
b061d892
DC
796}
797
798/*
799 * Same as above, but signed.
800 */
c1e01133
AJ
801static int matroska_ebmlnum_sint(MatroskaDemuxContext *matroska,
802 uint8_t *data, uint32_t size, int64_t *num)
b061d892
DC
803{
804 uint64_t unum;
805 int res;
806
807 /* read as unsigned number first */
c1e01133 808 if ((res = matroska_ebmlnum_uint(matroska, data, size, &unum)) < 0)
b061d892
DC
809 return res;
810
811 /* make signed (weird way) */
84cfce9f 812 *num = unum - ((1LL << (7 * res - 1)) - 1);
b061d892
DC
813
814 return res;
815}
816
737c40da
AJ
817static int ebml_parse_elem(MatroskaDemuxContext *matroska,
818 EbmlSyntax *syntax, void *data);
b061d892 819
737c40da
AJ
820static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
821 uint32_t id, void *data)
b061d892
DC
822{
823 int i;
84cfce9f 824 for (i = 0; syntax[i].id; i++)
737c40da
AJ
825 if (id == syntax[i].id)
826 break;
8dbe48fc 827 if (!syntax[i].id && id == MATROSKA_ID_CLUSTER &&
84cfce9f
KC
828 matroska->num_levels > 0 &&
829 matroska->levels[matroska->num_levels - 1].length == 0xffffffffffffff)
8dbe48fc 830 return 0; // we reached the end of an unknown size cluster
5b7a88f0 831 if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32) {
d92024f1 832 av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%"PRIX32"\n", id);
5b7a88f0
AK
833 if (matroska->ctx->error_recognition & AV_EF_EXPLODE)
834 return AVERROR_INVALIDDATA;
835 }
737c40da 836 return ebml_parse_elem(matroska, &syntax[i], data);
b061d892
DC
837}
838
737c40da
AJ
839static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
840 void *data)
b061d892 841{
c3ade62c 842 if (!matroska->current_id) {
73917810
AJ
843 uint64_t id;
844 int res = ebml_read_num(matroska, matroska->ctx->pb, 4, &id);
845 if (res < 0)
846 return res;
84cfce9f 847 matroska->current_id = id | 1 << 7 * res;
c3ade62c
AJ
848 }
849 return ebml_parse_id(matroska, syntax, matroska->current_id, data);
b061d892
DC
850}
851
737c40da
AJ
852static int ebml_parse_nest(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
853 void *data)
b061d892 854{
737c40da 855 int i, res = 0;
b061d892 856
84cfce9f 857 for (i = 0; syntax[i].id; i++)
737c40da
AJ
858 switch (syntax[i].type) {
859 case EBML_UINT:
84cfce9f 860 *(uint64_t *) ((char *) data + syntax[i].data_offset) = syntax[i].def.u;
737c40da
AJ
861 break;
862 case EBML_FLOAT:
84cfce9f 863 *(double *) ((char *) data + syntax[i].data_offset) = syntax[i].def.f;
737c40da
AJ
864 break;
865 case EBML_STR:
866 case EBML_UTF8:
668643b9
AK
867 // the default may be NULL
868 if (syntax[i].def.s) {
84cfce9f 869 uint8_t **dst = (uint8_t **) ((uint8_t *) data + syntax[i].data_offset);
668643b9
AK
870 *dst = av_strdup(syntax[i].def.s);
871 if (!*dst)
872 return AVERROR(ENOMEM);
873 }
737c40da 874 break;
34c9c1ba 875 }
b061d892 876
737c40da
AJ
877 while (!res && !ebml_level_end(matroska))
878 res = ebml_parse(matroska, syntax, data);
b061d892 879
737c40da 880 return res;
b061d892
DC
881}
882
789ed100
AJ
883static int ebml_parse_elem(MatroskaDemuxContext *matroska,
884 EbmlSyntax *syntax, void *data)
885{
95ec3d4c
RD
886 static const uint64_t max_lengths[EBML_TYPE_COUNT] = {
887 [EBML_UINT] = 8,
888 [EBML_FLOAT] = 8,
889 // max. 16 MB for strings
890 [EBML_STR] = 0x1000000,
891 [EBML_UTF8] = 0x1000000,
892 // max. 256 MB for binary data
893 [EBML_BIN] = 0x10000000,
894 // no limits for anything else
895 };
ae628ec1 896 AVIOContext *pb = matroska->ctx->pb;
789ed100 897 uint32_t id = syntax->id;
c6cd2b3d 898 uint64_t length;
789ed100
AJ
899 int res;
900
84cfce9f 901 data = (char *) data + syntax->data_offset;
789ed100
AJ
902 if (syntax->list_elem_size) {
903 EbmlList *list = data;
f369b935
AK
904 if ((res = av_reallocp_array(&list->elem,
905 list->nb_elem + 1,
906 syntax->list_elem_size)) < 0) {
907 list->nb_elem = 0;
908 return res;
909 }
84cfce9f 910 data = (char *) list->elem + list->nb_elem * syntax->list_elem_size;
789ed100
AJ
911 memset(data, 0, syntax->list_elem_size);
912 list->nb_elem++;
913 }
789ed100 914
c3ade62c
AJ
915 if (syntax->type != EBML_PASS && syntax->type != EBML_STOP) {
916 matroska->current_id = 0;
1b4d327b 917 if ((res = ebml_read_length(matroska, pb, &length)) < 0)
c6cd2b3d 918 return res;
95ec3d4c
RD
919 if (max_lengths[syntax->type] && length > max_lengths[syntax->type]) {
920 av_log(matroska->ctx, AV_LOG_ERROR,
921 "Invalid length 0x%"PRIx64" > 0x%"PRIx64" for syntax element %i\n",
922 length, max_lengths[syntax->type], syntax->type);
923 return AVERROR_INVALIDDATA;
924 }
c3ade62c 925 }
c6cd2b3d 926
789ed100 927 switch (syntax->type) {
84cfce9f
KC
928 case EBML_UINT:
929 res = ebml_read_uint(pb, length, data);
930 break;
931 case EBML_FLOAT:
932 res = ebml_read_float(pb, length, data);
933 break;
789ed100 934 case EBML_STR:
84cfce9f
KC
935 case EBML_UTF8:
936 res = ebml_read_ascii(pb, length, data);
937 break;
938 case EBML_BIN:
939 res = ebml_read_binary(pb, length, data);
940 break;
941 case EBML_NEST:
942 if ((res = ebml_read_master(matroska, length)) < 0)
943 return res;
944 if (id == MATROSKA_ID_SEGMENT)
945 matroska->segment_start = avio_tell(matroska->ctx->pb);
946 return ebml_parse_nest(matroska, syntax->def.n, data);
947 case EBML_PASS:
948 return ebml_parse_id(matroska, syntax->def.n, id, data);
949 case EBML_STOP:
950 return 1;
951 default:
952 return avio_skip(pb, length) < 0 ? AVERROR(EIO) : 0;
789ed100 953 }
ba5a1f99
AJ
954 if (res == AVERROR_INVALIDDATA)
955 av_log(matroska->ctx, AV_LOG_ERROR, "Invalid element\n");
956 else if (res == AVERROR(EIO))
957 av_log(matroska->ctx, AV_LOG_ERROR, "Read error\n");
958 return res;
789ed100
AJ
959}
960
789ed100
AJ
961static void ebml_free(EbmlSyntax *syntax, void *data)
962{
963 int i, j;
84cfce9f
KC
964 for (i = 0; syntax[i].id; i++) {
965 void *data_off = (char *) data + syntax[i].data_offset;
789ed100
AJ
966 switch (syntax[i].type) {
967 case EBML_STR:
84cfce9f
KC
968 case EBML_UTF8:
969 av_freep(data_off);
970 break;
971 case EBML_BIN:
972 av_freep(&((EbmlBin *) data_off)->data);
973 break;
789ed100
AJ
974 case EBML_NEST:
975 if (syntax[i].list_elem_size) {
976 EbmlList *list = data_off;
977 char *ptr = list->elem;
84cfce9f
KC
978 for (j = 0; j < list->nb_elem;
979 j++, ptr += syntax[i].list_elem_size)
789ed100
AJ
980 ebml_free(syntax[i].def.n, ptr);
981 av_free(list->elem);
982 } else
983 ebml_free(syntax[i].def.n, data_off);
84cfce9f
KC
984 default:
985 break;
789ed100
AJ
986 }
987 }
988}
989
737c40da
AJ
990/*
991 * Autodetecting...
992 */
993static int matroska_probe(AVProbeData *p)
994{
995 uint64_t total = 0;
470491fa 996 int len_mask = 0x80, size = 1, n = 1, i;
737c40da 997
5968d2dd 998 /* EBML header? */
737c40da
AJ
999 if (AV_RB32(p->buf) != EBML_ID_HEADER)
1000 return 0;
1001
1002 /* length of header */
1003 total = p->buf[4];
1004 while (size <= 8 && !(total & len_mask)) {
1005 size++;
1006 len_mask >>= 1;
1007 }
1008 if (size > 8)
84cfce9f 1009 return 0;
737c40da
AJ
1010 total &= (len_mask - 1);
1011 while (n < size)
1012 total = (total << 8) | p->buf[4 + n++];
1013
5968d2dd 1014 /* Does the probe data contain the whole header? */
737c40da 1015 if (p->buf_size < 4 + size + total)
84cfce9f 1016 return 0;
737c40da 1017
470491fa 1018 /* The header should contain a known document type. For now,
737c40da
AJ
1019 * we don't parse the whole header but simply check for the
1020 * availability of that array of characters inside the header.
1021 * Not fully fool-proof, but good enough. */
470491fa
JZ
1022 for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++) {
1023 int probelen = strlen(matroska_doctypes[i]);
69619a13
CE
1024 if (total < probelen)
1025 continue;
84cfce9f
KC
1026 for (n = 4 + size; n <= 4 + size + total - probelen; n++)
1027 if (!memcmp(p->buf + n, matroska_doctypes[i], probelen))
470491fa
JZ
1028 return AVPROBE_SCORE_MAX;
1029 }
737c40da 1030
c7b913c8 1031 // probably valid EBML header but no recognized doctype
e0f8be64 1032 return AVPROBE_SCORE_EXTENSION;
737c40da
AJ
1033}
1034
1035static MatroskaTrack *matroska_find_track_by_num(MatroskaDemuxContext *matroska,
1036 int num)
1037{
1038 MatroskaTrack *tracks = matroska->tracks.elem;
1039 int i;
1040
84cfce9f 1041 for (i = 0; i < matroska->tracks.nb_elem; i++)
737c40da
AJ
1042 if (tracks[i].num == num)
1043 return &tracks[i];
1044
1045 av_log(matroska->ctx, AV_LOG_ERROR, "Invalid track number %d\n", num);
1046 return NULL;
1047}
1048
84cfce9f 1049static int matroska_decode_buffer(uint8_t **buf, int *buf_size,
f7b9687c 1050 MatroskaTrack *track)
935ec5a1 1051{
2cbc8811 1052 MatroskaTrackEncoding *encodings = track->encodings.elem;
84cfce9f 1053 uint8_t *data = *buf;
935ec5a1 1054 int isize = *buf_size;
84cfce9f 1055 uint8_t *pkt_data = NULL;
529506b5 1056 uint8_t av_unused *newpktdata;
935ec5a1
ES
1057 int pkt_size = isize;
1058 int result = 0;
1059 int olen;
1060
4f90688b 1061 if (pkt_size >= 10000000)
c9a39cec 1062 return AVERROR_INVALIDDATA;
4f90688b 1063
2cbc8811 1064 switch (encodings[0].compression.algo) {
84cfce9f
KC
1065 case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
1066 {
8d4dd55c
LB
1067 int header_size = encodings[0].compression.settings.size;
1068 uint8_t *header = encodings[0].compression.settings.data;
1069
1070 if (!header_size)
1071 return 0;
1072
1073 pkt_size = isize + header_size;
1074 pkt_data = av_malloc(pkt_size);
1075 if (!pkt_data)
1076 return AVERROR(ENOMEM);
1077
1078 memcpy(pkt_data, header, header_size);
1079 memcpy(pkt_data + header_size, data, isize);
1080 break;
1081 }
2a91ada8 1082#if CONFIG_LZO
935ec5a1
ES
1083 case MATROSKA_TRACK_ENCODING_COMP_LZO:
1084 do {
84cfce9f 1085 olen = pkt_size *= 3;
581281e2
LB
1086 newpktdata = av_realloc(pkt_data, pkt_size + AV_LZO_OUTPUT_PADDING);
1087 if (!newpktdata) {
c9a39cec 1088 result = AVERROR(ENOMEM);
581281e2
LB
1089 goto failed;
1090 }
1091 pkt_data = newpktdata;
84cfce9f
KC
1092 result = av_lzo1x_decode(pkt_data, &olen, data, &isize);
1093 } while (result == AV_LZO_OUTPUT_FULL && pkt_size < 10000000);
c9a39cec
LB
1094 if (result) {
1095 result = AVERROR_INVALIDDATA;
935ec5a1 1096 goto failed;
c9a39cec 1097 }
935ec5a1
ES
1098 pkt_size -= olen;
1099 break;
2a91ada8 1100#endif
b250f9c6 1101#if CONFIG_ZLIB
84cfce9f
KC
1102 case MATROSKA_TRACK_ENCODING_COMP_ZLIB:
1103 {
1104 z_stream zstream = { 0 };
935ec5a1
ES
1105 if (inflateInit(&zstream) != Z_OK)
1106 return -1;
84cfce9f 1107 zstream.next_in = data;
935ec5a1
ES
1108 zstream.avail_in = isize;
1109 do {
84cfce9f 1110 pkt_size *= 3;
77d2ef13
MN
1111 newpktdata = av_realloc(pkt_data, pkt_size);
1112 if (!newpktdata) {
1113 inflateEnd(&zstream);
1114 goto failed;
1115 }
84cfce9f 1116 pkt_data = newpktdata;
935ec5a1 1117 zstream.avail_out = pkt_size - zstream.total_out;
84cfce9f
KC
1118 zstream.next_out = pkt_data + zstream.total_out;
1119 result = inflate(&zstream, Z_NO_FLUSH);
1120 } while (result == Z_OK && pkt_size < 10000000);
935ec5a1
ES
1121 pkt_size = zstream.total_out;
1122 inflateEnd(&zstream);
c9a39cec
LB
1123 if (result != Z_STREAM_END) {
1124 if (result == Z_MEM_ERROR)
1125 result = AVERROR(ENOMEM);
1126 else
1127 result = AVERROR_INVALIDDATA;
935ec5a1 1128 goto failed;
c9a39cec 1129 }
935ec5a1
ES
1130 break;
1131 }
1132#endif
b250f9c6 1133#if CONFIG_BZLIB
84cfce9f
KC
1134 case MATROSKA_TRACK_ENCODING_COMP_BZLIB:
1135 {
1136 bz_stream bzstream = { 0 };
935ec5a1
ES
1137 if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
1138 return -1;
84cfce9f 1139 bzstream.next_in = data;
935ec5a1
ES
1140 bzstream.avail_in = isize;
1141 do {
84cfce9f 1142 pkt_size *= 3;
77d2ef13
MN
1143 newpktdata = av_realloc(pkt_data, pkt_size);
1144 if (!newpktdata) {
1145 BZ2_bzDecompressEnd(&bzstream);
1146 goto failed;
1147 }
84cfce9f 1148 pkt_data = newpktdata;
935ec5a1 1149 bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
84cfce9f
KC
1150 bzstream.next_out = pkt_data + bzstream.total_out_lo32;
1151 result = BZ2_bzDecompress(&bzstream);
1152 } while (result == BZ_OK && pkt_size < 10000000);
935ec5a1
ES
1153 pkt_size = bzstream.total_out_lo32;
1154 BZ2_bzDecompressEnd(&bzstream);
c9a39cec
LB
1155 if (result != BZ_STREAM_END) {
1156 if (result == BZ_MEM_ERROR)
1157 result = AVERROR(ENOMEM);
1158 else
1159 result = AVERROR_INVALIDDATA;
935ec5a1 1160 goto failed;
c9a39cec 1161 }
935ec5a1
ES
1162 break;
1163 }
1164#endif
28f27e0c 1165 default:
c9a39cec 1166 return AVERROR_INVALIDDATA;
935ec5a1
ES
1167 }
1168
84cfce9f 1169 *buf = pkt_data;
935ec5a1
ES
1170 *buf_size = pkt_size;
1171 return 0;
84cfce9f
KC
1172
1173failed:
935ec5a1 1174 av_free(pkt_data);
c9a39cec 1175 return result;
935ec5a1
ES
1176}
1177
3eb9bfbb 1178static void matroska_fix_ass_packet(MatroskaDemuxContext *matroska,
e7d4b743 1179 AVPacket *pkt, uint64_t display_duration)
3eb9bfbb 1180{
1afddbe5 1181 AVBufferRef *line;
84cfce9f
KC
1182 char *layer, *ptr = pkt->data, *end = ptr + pkt->size;
1183
1184 for (; *ptr != ',' && ptr < end - 1; ptr++)
1185 ;
3eb9bfbb
AJ
1186 if (*ptr == ',')
1187 layer = ++ptr;
84cfce9f
KC
1188 for (; *ptr != ',' && ptr < end - 1; ptr++)
1189 ;
3eb9bfbb 1190 if (*ptr == ',') {
e7d4b743 1191 int64_t end_pts = pkt->pts + display_duration;
3eb9bfbb
AJ
1192 int sc = matroska->time_scale * pkt->pts / 10000000;
1193 int ec = matroska->time_scale * end_pts / 10000000;
1194 int sh, sm, ss, eh, em, es, len;
84cfce9f
KC
1195 sh = sc / 360000;
1196 sc -= 360000 * sh;
1197 sm = sc / 6000;
1198 sc -= 6000 * sm;
1199 ss = sc / 100;
1200 sc -= 100 * ss;
1201 eh = ec / 360000;
1202 ec -= 360000 * eh;
1203 em = ec / 6000;
1204 ec -= 6000 * em;
1205 es = ec / 100;
1206 ec -= 100 * es;
3eb9bfbb 1207 *ptr++ = '\0';
059a9348 1208 len = 50 + end - ptr + AV_INPUT_BUFFER_PADDING_SIZE;
1afddbe5 1209 if (!(line = av_buffer_alloc(len)))
3eb9bfbb 1210 return;
84cfce9f
KC
1211 snprintf(line->data, len,
1212 "Dialogue: %s,%d:%02d:%02d.%02d,%d:%02d:%02d.%02d,%s\r\n",
3eb9bfbb 1213 layer, sh, sm, ss, sc, eh, em, es, ec, ptr);
1afddbe5
AK
1214 av_buffer_unref(&pkt->buf);
1215 pkt->buf = line;
1216 pkt->data = line->data;
1217 pkt->size = strlen(line->data);
3eb9bfbb
AJ
1218 }
1219}
1220
77d2ef13 1221static int matroska_merge_packets(AVPacket *out, AVPacket *in)
d5e34dcb 1222{
34871beb
AK
1223 int old_size = out->size;
1224 int ret = av_grow_packet(out, in->size);
1225 if (ret < 0)
1226 return ret;
1227
1228 memcpy(out->data + old_size, in->data, in->size);
1229
ce70f28a 1230 av_packet_unref(in);
d5e34dcb 1231 av_free(in);
77d2ef13 1232 return 0;
d5e34dcb
AJ
1233}
1234
929e9de7 1235static void matroska_convert_tag(AVFormatContext *s, EbmlList *list,
d2d67e42 1236 AVDictionary **metadata, char *prefix)
44015c56
AJ
1237{
1238 MatroskaTag *tags = list->elem;
929e9de7
AJ
1239 char key[1024];
1240 int i;
44015c56 1241
84cfce9f
KC
1242 for (i = 0; i < list->nb_elem; i++) {
1243 const char *lang = tags[i].lang &&
1244 strcmp(tags[i].lang, "und") ? tags[i].lang : NULL;
bf800c7d
AK
1245
1246 if (!tags[i].name) {
1247 av_log(s, AV_LOG_WARNING, "Skipping invalid tag with no TagName.\n");
1248 continue;
1249 }
84cfce9f
KC
1250 if (prefix)
1251 snprintf(key, sizeof(key), "%s/%s", prefix, tags[i].name);
1252 else
1253 av_strlcpy(key, tags[i].name, sizeof(key));
f702df39 1254 if (tags[i].def || !lang) {
84cfce9f
KC
1255 av_dict_set(metadata, key, tags[i].string, 0);
1256 if (tags[i].sub.nb_elem)
1257 matroska_convert_tag(s, &tags[i].sub, metadata, key);
f702df39
AJ
1258 }
1259 if (lang) {
1260 av_strlcat(key, "-", sizeof(key));
1261 av_strlcat(key, lang, sizeof(key));
d2d67e42 1262 av_dict_set(metadata, key, tags[i].string, 0);
f702df39
AJ
1263 if (tags[i].sub.nb_elem)
1264 matroska_convert_tag(s, &tags[i].sub, metadata, key);
1265 }
929e9de7 1266 }
ad7768f4 1267 ff_metadata_conv(metadata, NULL, ff_mkv_metadata_conv);
929e9de7
AJ
1268}
1269
1270static void matroska_convert_tags(AVFormatContext *s)
1271{
1272 MatroskaDemuxContext *matroska = s->priv_data;
1273 MatroskaTags *tags = matroska->tags.elem;
1274 int i, j;
1275
84cfce9f 1276 for (i = 0; i < matroska->tags.nb_elem; i++) {
929e9de7 1277 if (tags[i].target.attachuid) {
f69befe5 1278 MatroskaAttachment *attachment = matroska->attachments.elem;
a0fa6d06
LB
1279 int found = 0;
1280 for (j = 0; j < matroska->attachments.nb_elem; j++) {
84cfce9f 1281 if (attachment[j].uid == tags[i].target.attachuid &&
a0fa6d06 1282 attachment[j].stream) {
929e9de7
AJ
1283 matroska_convert_tag(s, &tags[i].tag,
1284 &attachment[j].stream->metadata, NULL);
a0fa6d06
LB
1285 found = 1;
1286 }
1287 }
1288 if (!found) {
1289 av_log(NULL, AV_LOG_WARNING,
1290 "The tags at index %d refer to a "
1291 "non-existent attachment %"PRId64".\n",
1292 i, tags[i].target.attachuid);
1293 }
929e9de7
AJ
1294 } else if (tags[i].target.chapteruid) {
1295 MatroskaChapter *chapter = matroska->chapters.elem;
a0fa6d06
LB
1296 int found = 0;
1297 for (j = 0; j < matroska->chapters.nb_elem; j++) {
84cfce9f 1298 if (chapter[j].uid == tags[i].target.chapteruid &&
a0fa6d06 1299 chapter[j].chapter) {
929e9de7
AJ
1300 matroska_convert_tag(s, &tags[i].tag,
1301 &chapter[j].chapter->metadata, NULL);
a0fa6d06
LB
1302 found = 1;
1303 }
1304 }
1305 if (!found) {
1306 av_log(NULL, AV_LOG_WARNING,
1307 "The tags at index %d refer to a non-existent chapter "
1308 "%"PRId64".\n",
1309 i, tags[i].target.chapteruid);
1310 }
929e9de7
AJ
1311 } else if (tags[i].target.trackuid) {
1312 MatroskaTrack *track = matroska->tracks.elem;
a0fa6d06
LB
1313 int found = 0;
1314 for (j = 0; j < matroska->tracks.nb_elem; j++) {
1315 if (track[j].uid == tags[i].target.trackuid &&
1316 track[j].stream) {
929e9de7
AJ
1317 matroska_convert_tag(s, &tags[i].tag,
1318 &track[j].stream->metadata, NULL);
a0fa6d06
LB
1319 found = 1;
1320 }
1321 }
1322 if (!found) {
1323 av_log(NULL, AV_LOG_WARNING,
1324 "The tags at index %d refer to a non-existent track "
1325 "%"PRId64".\n",
1326 i, tags[i].target.trackuid);
1327 }
929e9de7 1328 } else {
4f909c76
AJ
1329 matroska_convert_tag(s, &tags[i].tag, &s->metadata,
1330 tags[i].target.type);
929e9de7 1331 }
44015c56
AJ
1332 }
1333}
1334
84cfce9f
KC
1335static int matroska_parse_seekhead_entry(MatroskaDemuxContext *matroska,
1336 int idx)
13b350a3
AJ
1337{
1338 EbmlList *seekhead_list = &matroska->seekhead;
84cfce9f
KC
1339 uint32_t level_up = matroska->level_up;
1340 uint32_t saved_id = matroska->current_id;
13b350a3 1341 MatroskaSeekhead *seekhead = seekhead_list->elem;
a2704c97 1342 int64_t before_pos = avio_tell(matroska->ctx->pb);
13b350a3 1343 MatroskaLevel level;
31ad14c2
AC
1344 int64_t offset;
1345 int ret = 0;
66cfc385 1346
84cfce9f
KC
1347 if (idx >= seekhead_list->nb_elem ||
1348 seekhead[idx].id == MATROSKA_ID_SEEKHEAD ||
1349 seekhead[idx].id == MATROSKA_ID_CLUSTER)
31ad14c2 1350 return 0;
b061d892 1351
f47ac3c6 1352 /* seek */
31ad14c2
AC
1353 offset = seekhead[idx].pos + matroska->segment_start;
1354 if (avio_seek(matroska->ctx->pb, offset, SEEK_SET) == offset) {
5968d2dd 1355 /* We don't want to lose our seekhead level, so we add
4348571b
AJ
1356 * a dummy. This is a crude hack. */
1357 if (matroska->num_levels == EBML_MAX_DEPTH) {
1358 av_log(matroska->ctx, AV_LOG_INFO,
1359 "Max EBML element depth (%d) reached, "
1360 "cannot parse further.\n", EBML_MAX_DEPTH);
31ad14c2
AC
1361 ret = AVERROR_INVALIDDATA;
1362 } else {
84cfce9f
KC
1363 level.start = 0;
1364 level.length = (uint64_t) -1;
f47ac3c6
AK
1365 matroska->levels[matroska->num_levels] = level;
1366 matroska->num_levels++;
84cfce9f 1367 matroska->current_id = 0;
f47ac3c6 1368
4a9628f2 1369 ret = ebml_parse(matroska, matroska_segment, matroska);
f47ac3c6
AK
1370
1371 /* remove dummy level */
1372 while (matroska->num_levels) {
1373 uint64_t length = matroska->levels[--matroska->num_levels].length;
84cfce9f 1374 if (length == (uint64_t) -1)
f47ac3c6
AK
1375 break;
1376 }
4348571b 1377 }
13b350a3 1378 }
4348571b 1379 /* seek back */
6b4aa5da 1380 avio_seek(matroska->ctx->pb, before_pos, SEEK_SET);
84cfce9f 1381 matroska->level_up = level_up;
c3ade62c 1382 matroska->current_id = saved_id;
31ad14c2
AC
1383
1384 return ret;
1385}
1386
1387static void matroska_execute_seekhead(MatroskaDemuxContext *matroska)
1388{
1389 EbmlList *seekhead_list = &matroska->seekhead;
31ad14c2
AC
1390 int64_t before_pos = avio_tell(matroska->ctx->pb);
1391 int i;
1392
1393 // we should not do any seeking in the streaming case
1394 if (!matroska->ctx->pb->seekable ||
1395 (matroska->ctx->flags & AVFMT_FLAG_IGNIDX))
1396 return;
1397
1398 for (i = 0; i < seekhead_list->nb_elem; i++) {
faaec467 1399 MatroskaSeekhead *seekhead = seekhead_list->elem;
31ad14c2
AC
1400 if (seekhead[i].pos <= before_pos)
1401 continue;
1402
1403 // defer cues parsing until we actually need cue data.
1404 if (seekhead[i].id == MATROSKA_ID_CUES) {
1405 matroska->cues_parsing_deferred = 1;
1406 continue;
1407 }
1408
1409 if (matroska_parse_seekhead_entry(matroska, i) < 0)
1410 break;
1411 }
1412}
1413
84cfce9f
KC
1414static void matroska_parse_cues(MatroskaDemuxContext *matroska)
1415{
31ad14c2
AC
1416 EbmlList *seekhead_list = &matroska->seekhead;
1417 MatroskaSeekhead *seekhead = seekhead_list->elem;
1418 EbmlList *index_list;
1419 MatroskaIndex *index;
1420 int index_scale = 1;
1421 int i, j;
1422
1423 for (i = 0; i < seekhead_list->nb_elem; i++)
fdb94444 1424 if (seekhead[i].id == MATROSKA_ID_CUES)
31ad14c2
AC
1425 break;
1426 assert(i <= seekhead_list->nb_elem);
1427
1428 matroska_parse_seekhead_entry(matroska, i);
1429
1430 index_list = &matroska->index;
84cfce9f
KC
1431 index = index_list->elem;
1432 if (index_list->nb_elem &&
1433 index[0].time > 1E14 / matroska->time_scale) {
31ad14c2
AC
1434 av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n");
1435 index_scale = matroska->time_scale;
1436 }
1437 for (i = 0; i < index_list->nb_elem; i++) {
84cfce9f 1438 EbmlList *pos_list = &index[i].pos;
31ad14c2
AC
1439 MatroskaIndexPos *pos = pos_list->elem;
1440 for (j = 0; j < pos_list->nb_elem; j++) {
84cfce9f
KC
1441 MatroskaTrack *track = matroska_find_track_by_num(matroska,
1442 pos[j].track);
31ad14c2
AC
1443 if (track && track->stream)
1444 av_add_index_entry(track->stream,
1445 pos[j].pos + matroska->segment_start,
84cfce9f 1446 index[i].time / index_scale, 0, 0,
31ad14c2
AC
1447 AVINDEX_KEYFRAME);
1448 }
1449 }
b061d892
DC
1450}
1451
f7b9687c 1452static int matroska_aac_profile(char *codec_id)
b061d892 1453{
84cfce9f 1454 static const char *const aac_profiles[] = { "MAIN", "LC", "SSR" };
b061d892
DC
1455 int profile;
1456
84cfce9f 1457 for (profile = 0; profile < FF_ARRAY_ELEMS(aac_profiles); profile++)
b061d892
DC
1458 if (strstr(codec_id, aac_profiles[profile]))
1459 break;
1460 return profile + 1;
1461}
1462
f7b9687c 1463static int matroska_aac_sri(int samplerate)
b061d892 1464{
b061d892
DC
1465 int sri;
1466
84cfce9f 1467 for (sri = 0; sri < FF_ARRAY_ELEMS(avpriv_mpeg4audio_sample_rates); sri++)
59a9a235 1468 if (avpriv_mpeg4audio_sample_rates[sri] == samplerate)
b061d892
DC
1469 break;
1470 return sri;
1471}
1472
4efdadc8
AK
1473static int matroska_parse_flac(AVFormatContext *s,
1474 MatroskaTrack *track,
1475 int *offset)
1476{
23f741f7 1477 AVStream *st = track->stream;
4efdadc8
AK
1478 uint8_t *p = track->codec_priv.data;
1479 int size = track->codec_priv.size;
1480
1481 if (size < 8 + FLAC_STREAMINFO_SIZE || p[4] & 0x7f) {
1482 av_log(s, AV_LOG_WARNING, "Invalid FLAC private data\n");
1483 track->codec_priv.size = 0;
1484 return 0;
1485 }
1486 *offset = 8;
1487 track->codec_priv.size = 8 + FLAC_STREAMINFO_SIZE;
1488
23f741f7
AK
1489 p += track->codec_priv.size;
1490 size -= track->codec_priv.size;
1491
1492 /* parse the remaining metadata blocks if present */
1493 while (size >= 4) {
1494 int block_last, block_type, block_size;
1495
1496 flac_parse_block_header(p, &block_last, &block_type, &block_size);
1497
1498 p += 4;
1499 size -= 4;
1500 if (block_size > size)
1501 return 0;
1502
1503 /* check for the channel mask */
1504 if (block_type == FLAC_METADATA_TYPE_VORBIS_COMMENT) {
1505 AVDictionary *dict = NULL;
1506 AVDictionaryEntry *chmask;
1507
1508 ff_vorbis_comment(s, &dict, p, block_size, 0);
1509 chmask = av_dict_get(dict, "WAVEFORMATEXTENSIBLE_CHANNEL_MASK", NULL, 0);
1510 if (chmask) {
1511 uint64_t mask = strtol(chmask->value, NULL, 0);
1512 if (!mask || mask & ~0x3ffffULL) {
1513 av_log(s, AV_LOG_WARNING,
1514 "Invalid value of WAVEFORMATEXTENSIBLE_CHANNEL_MASK\n");
1515 } else
9200514a 1516 st->codecpar->channel_layout = mask;
23f741f7
AK
1517 }
1518 av_dict_free(&dict);
1519 }
1520
1521 p += block_size;
1522 size -= block_size;
1523 }
1524
4efdadc8
AK
1525 return 0;
1526}
1527
febfb49a
AC
1528static void mkv_stereo_mode_display_mul(int stereo_mode,
1529 int *h_width, int *h_height)
1530{
1531 switch (stereo_mode) {
1532 case MATROSKA_VIDEO_STEREOMODE_TYPE_MONO:
1533 case MATROSKA_VIDEO_STEREOMODE_TYPE_CHECKERBOARD_RL:
1534 case MATROSKA_VIDEO_STEREOMODE_TYPE_CHECKERBOARD_LR:
1535 case MATROSKA_VIDEO_STEREOMODE_TYPE_BOTH_EYES_BLOCK_RL:
1536 case MATROSKA_VIDEO_STEREOMODE_TYPE_BOTH_EYES_BLOCK_LR:
1537 break;
1538 case MATROSKA_VIDEO_STEREOMODE_TYPE_RIGHT_LEFT:
1539 case MATROSKA_VIDEO_STEREOMODE_TYPE_LEFT_RIGHT:
1540 case MATROSKA_VIDEO_STEREOMODE_TYPE_COL_INTERLEAVED_RL:
1541 case MATROSKA_VIDEO_STEREOMODE_TYPE_COL_INTERLEAVED_LR:
1542 *h_width = 2;
1543 break;
1544 case MATROSKA_VIDEO_STEREOMODE_TYPE_BOTTOM_TOP:
1545 case MATROSKA_VIDEO_STEREOMODE_TYPE_TOP_BOTTOM:
1546 case MATROSKA_VIDEO_STEREOMODE_TYPE_ROW_INTERLEAVED_RL:
1547 case MATROSKA_VIDEO_STEREOMODE_TYPE_ROW_INTERLEAVED_LR:
1548 *h_height = 2;
1549 break;
1550 }
1551}
1552
6df478bf 1553static int matroska_parse_tracks(AVFormatContext *s)
b061d892
DC
1554{
1555 MatroskaDemuxContext *matroska = s->priv_data;
6df478bf 1556 MatroskaTrack *tracks = matroska->tracks.elem;
9a9a3b03 1557 AVStream *st;
4efdadc8 1558 int i, j, ret;
9c25bafa 1559
84cfce9f 1560 for (i = 0; i < matroska->tracks.nb_elem; i++) {
d88d806b 1561 MatroskaTrack *track = &tracks[i];
36ef5369 1562 enum AVCodecID codec_id = AV_CODEC_ID_NONE;
8b516f15 1563 EbmlList *encodings_list = &track->encodings;
9c25bafa 1564 MatroskaTrackEncoding *encodings = encodings_list->elem;
d88d806b
AJ
1565 uint8_t *extradata = NULL;
1566 int extradata_size = 0;
1567 int extradata_offset = 0;
ae628ec1 1568 AVIOContext b;
d88d806b
AJ
1569
1570 /* Apply some sanity checks. */
9c25bafa
AJ
1571 if (track->type != MATROSKA_TRACK_TYPE_VIDEO &&
1572 track->type != MATROSKA_TRACK_TYPE_AUDIO &&
1573 track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1574 av_log(matroska->ctx, AV_LOG_INFO,
1575 "Unknown or unsupported track type %"PRIu64"\n",
1576 track->type);
1577 continue;
1578 }
f929ab05 1579 if (!track->codec_id)
d88d806b
AJ
1580 continue;
1581
9c25bafa 1582 if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
3c583002 1583 if (!track->default_duration && track->video.frame_rate > 0)
84cfce9f 1584 track->default_duration = 1000000000 / track->video.frame_rate;
9c25bafa
AJ
1585 if (!track->video.display_width)
1586 track->video.display_width = track->video.pixel_width;
1587 if (!track->video.display_height)
1588 track->video.display_height = track->video.pixel_height;
1589 } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1590 if (!track->audio.out_samplerate)
1591 track->audio.out_samplerate = track->audio.samplerate;
1592 }
1593 if (encodings_list->nb_elem > 1) {
1594 av_log(matroska->ctx, AV_LOG_ERROR,
d7d2f0e6 1595 "Multiple combined encodings not supported");
9c25bafa
AJ
1596 } else if (encodings_list->nb_elem == 1) {
1597 if (encodings[0].type ||
2a91ada8 1598 (
b250f9c6 1599#if CONFIG_ZLIB
84cfce9f 1600 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
9c25bafa 1601#endif
b250f9c6 1602#if CONFIG_BZLIB
9c25bafa
AJ
1603 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
1604#endif
2a91ada8 1605#if CONFIG_LZO
84cfce9f 1606 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_LZO &&
2a91ada8
DB
1607#endif
1608 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP)) {
9c25bafa
AJ
1609 encodings[0].scope = 0;
1610 av_log(matroska->ctx, AV_LOG_ERROR,
1611 "Unsupported encoding type");
84cfce9f 1612 } else if (track->codec_priv.size && encodings[0].scope & 2) {
9c25bafa 1613 uint8_t *codec_priv = track->codec_priv.data;
8d4dd55c
LB
1614 int ret = matroska_decode_buffer(&track->codec_priv.data,
1615 &track->codec_priv.size,
1616 track);
1617 if (ret < 0) {
9c25bafa
AJ
1618 track->codec_priv.data = NULL;
1619 track->codec_priv.size = 0;
1620 av_log(matroska->ctx, AV_LOG_ERROR,
1621 "Failed to decode codec private data\n");
9c25bafa 1622 }
8d4dd55c 1623
9c25bafa
AJ
1624 if (codec_priv != track->codec_priv.data)
1625 av_free(codec_priv);
1626 }
1627 }
1628
84cfce9f
KC
1629 for (j = 0; ff_mkv_codec_tags[j].id != AV_CODEC_ID_NONE; j++) {
1630 if (!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
1631 strlen(ff_mkv_codec_tags[j].str))) {
1632 codec_id = ff_mkv_codec_tags[j].id;
d88d806b 1633 break;
b061d892 1634 }
d88d806b 1635 }
b061d892 1636
3b3bbdd3 1637 st = track->stream = avformat_new_stream(s, NULL);
f929ab05 1638 if (!st)
d88d806b
AJ
1639 return AVERROR(ENOMEM);
1640
84cfce9f
KC
1641 if (!strcmp(track->codec_id, "V_MS/VFW/FOURCC") &&
1642 track->codec_priv.size >= 40 &&
4b1f5e50 1643 track->codec_priv.data) {
84cfce9f 1644 track->ms_compat = 1;
d88d806b 1645 track->video.fourcc = AV_RL32(track->codec_priv.data + 16);
84cfce9f
KC
1646 codec_id = ff_codec_get_id(ff_codec_bmp_tags,
1647 track->video.fourcc);
1648 extradata_offset = 40;
1649 } else if (!strcmp(track->codec_id, "A_MS/ACM") &&
1650 track->codec_priv.size >= 14 &&
4b1f5e50 1651 track->codec_priv.data) {
ca402f32 1652 int ret;
84cfce9f
KC
1653 ffio_init_context(&b, track->codec_priv.data,
1654 track->codec_priv.size,
37cb3b18 1655 0, NULL, NULL, NULL, NULL);
9200514a 1656 ret = ff_get_wav_header(s, &b, st->codecpar, track->codec_priv.size);
ca402f32
MH
1657 if (ret < 0)
1658 return ret;
9200514a 1659 codec_id = st->codecpar->codec_id;
038146e9 1660 extradata_offset = FFMIN(track->codec_priv.size, 18);
84cfce9f
KC
1661 } else if (!strcmp(track->codec_id, "V_QUICKTIME") &&
1662 (track->codec_priv.size >= 86) &&
4b1f5e50 1663 (track->codec_priv.data)) {
8b4b1c1e
LB
1664 if (track->codec_priv.size == AV_RB32(track->codec_priv.data)) {
1665 track->video.fourcc = AV_RL32(track->codec_priv.data + 4);
1666 codec_id = ff_codec_get_id(ff_codec_movvideo_tags,
1667 track->video.fourcc);
1668 }
1669 if (codec_id == AV_CODEC_ID_NONE) {
1670 track->video.fourcc = AV_RL32(track->codec_priv.data);
1671 codec_id = ff_codec_get_id(ff_codec_movvideo_tags,
1672 track->video.fourcc);
1673 }
8fc11abe
VG
1674 if (codec_id == AV_CODEC_ID_NONE) {
1675 char buf[32];
1676 av_get_codec_tag_string(buf, sizeof(buf), track->video.fourcc);
1677 av_log(matroska->ctx, AV_LOG_ERROR,
1678 "mov FourCC not found %s.\n", buf);
1679 }
36ef5369 1680 } else if (codec_id == AV_CODEC_ID_PCM_S16BE) {
eb9cf50a 1681 switch (track->audio.bitdepth) {
84cfce9f
KC
1682 case 8:
1683 codec_id = AV_CODEC_ID_PCM_U8;
1684 break;
1685 case 24:
1686 codec_id = AV_CODEC_ID_PCM_S24BE;
1687 break;
1688 case 32:
1689 codec_id = AV_CODEC_ID_PCM_S32BE;
1690 break;
eb9cf50a 1691 }
36ef5369 1692 } else if (codec_id == AV_CODEC_ID_PCM_S16LE) {
eb9cf50a 1693 switch (track->audio.bitdepth) {
84cfce9f
KC
1694 case 8:
1695 codec_id = AV_CODEC_ID_PCM_U8;
1696 break;
1697 case 24:
1698 codec_id = AV_CODEC_ID_PCM_S24LE;
1699 break;
1700 case 32:
1701 codec_id = AV_CODEC_ID_PCM_S32LE;
1702 break;
eb9cf50a 1703 }
84cfce9f
KC
1704 } else if (codec_id == AV_CODEC_ID_PCM_F32LE &&
1705 track->audio.bitdepth == 64) {
36ef5369
AK
1706 codec_id = AV_CODEC_ID_PCM_F64LE;
1707 } else if (codec_id == AV_CODEC_ID_AAC && !track->codec_priv.size) {
d88d806b 1708 int profile = matroska_aac_profile(track->codec_id);
84cfce9f 1709 int sri = matroska_aac_sri(track->audio.samplerate);
059a9348 1710 extradata = av_mallocz(5 + AV_INPUT_BUFFER_PADDING_SIZE);
f929ab05 1711 if (!extradata)
d88d806b 1712 return AVERROR(ENOMEM);
84cfce9f
KC
1713 extradata[0] = (profile << 3) | ((sri & 0x0E) >> 1);
1714 extradata[1] = ((sri & 0x01) << 7) | (track->audio.channels << 3);
d88d806b 1715 if (strstr(track->codec_id, "SBR")) {
84cfce9f
KC
1716 sri = matroska_aac_sri(track->audio.out_samplerate);
1717 extradata[2] = 0x56;
1718 extradata[3] = 0xE5;
1719 extradata[4] = 0x80 | (sri << 3);
d88d806b 1720 extradata_size = 5;
16f97ab0 1721 } else
d88d806b 1722 extradata_size = 2;
8071dca3
MB
1723 } else if (codec_id == AV_CODEC_ID_ALAC && track->codec_priv.size) {
1724 /* Only ALAC's magic cookie is stored in Matroska's track headers.
84cfce9f
KC
1725 * Create the "atom size", "tag", and "tag version" fields the
1726 * decoder expects manually. */
8071dca3 1727 extradata_size = 12 + track->codec_priv.size;
84cfce9f 1728 extradata = av_mallocz(extradata_size +
059a9348 1729 AV_INPUT_BUFFER_PADDING_SIZE);
f929ab05 1730 if (!extradata)
8071dca3
MB
1731 return AVERROR(ENOMEM);
1732 AV_WB32(extradata, extradata_size);
1733 memcpy(&extradata[4], "alac", 4);
1734 AV_WB32(&extradata[8], 0);
1735 memcpy(&extradata[12], track->codec_priv.data,
84cfce9f 1736 track->codec_priv.size);
36ef5369 1737 } else if (codec_id == AV_CODEC_ID_TTA) {
d88d806b 1738 extradata_size = 30;
84cfce9f 1739 extradata = av_mallocz(extradata_size);
f929ab05 1740 if (!extradata)
d88d806b 1741 return AVERROR(ENOMEM);
e731b8d8 1742 ffio_init_context(&b, extradata, extradata_size, 1,
84cfce9f 1743 NULL, NULL, NULL, NULL);
77eb5504
AK
1744 avio_write(&b, "TTA1", 4);
1745 avio_wl16(&b, 1);
1746 avio_wl16(&b, track->audio.channels);
1747 avio_wl16(&b, track->audio.bitdepth);
1748 avio_wl32(&b, track->audio.out_samplerate);
84cfce9f
KC
1749 avio_wl32(&b, matroska->ctx->duration *
1750 track->audio.out_samplerate);
1751 } else if (codec_id == AV_CODEC_ID_RV10 ||
1752 codec_id == AV_CODEC_ID_RV20 ||
1753 codec_id == AV_CODEC_ID_RV30 ||
1754 codec_id == AV_CODEC_ID_RV40) {
d88d806b 1755 extradata_offset = 26;
36ef5369 1756 } else if (codec_id == AV_CODEC_ID_RA_144) {
d88d806b 1757 track->audio.out_samplerate = 8000;
84cfce9f
KC
1758 track->audio.channels = 1;
1759 } else if (codec_id == AV_CODEC_ID_RA_288 ||
1760 codec_id == AV_CODEC_ID_COOK ||
1761 codec_id == AV_CODEC_ID_ATRAC3 ||
1762 codec_id == AV_CODEC_ID_SIPR) {
6b10228a 1763 int flavor;
84cfce9f
KC
1764 ffio_init_context(&b, track->codec_priv.data,
1765 track->codec_priv.size,
1766 0, NULL, NULL, NULL, NULL);
45a8a02a 1767 avio_skip(&b, 22);
b7effd4e
AK
1768 flavor = avio_rb16(&b);
1769 track->audio.coded_framesize = avio_rb32(&b);
45a8a02a 1770 avio_skip(&b, 12);
b7effd4e
AK
1771 track->audio.sub_packet_h = avio_rb16(&b);
1772 track->audio.frame_size = avio_rb16(&b);
1773 track->audio.sub_packet_size = avio_rb16(&b);
84cfce9f
KC
1774 if (flavor <= 0 ||
1775 track->audio.coded_framesize <= 0 ||
1776 track->audio.sub_packet_h <= 0 ||
1777 track->audio.frame_size <= 0 ||
569d18aa
MS
1778 track->audio.sub_packet_size <= 0)
1779 return AVERROR_INVALIDDATA;
84cfce9f
KC
1780 track->audio.buf = av_malloc(track->audio.frame_size *
1781 track->audio.sub_packet_h);
6bdae41d
VG
1782 if (!track->audio.buf)
1783 return AVERROR(ENOMEM);
36ef5369 1784 if (codec_id == AV_CODEC_ID_RA_288) {
9200514a 1785 st->codecpar->block_align = track->audio.coded_framesize;
d88d806b
AJ
1786 track->codec_priv.size = 0;
1787 } else {
36ef5369 1788 if (codec_id == AV_CODEC_ID_SIPR && flavor < 4) {
6b10228a
AJ
1789 const int sipr_bit_rate[4] = { 6504, 8496, 5000, 16000 };
1790 track->audio.sub_packet_size = ff_sipr_subpk_size[flavor];
9200514a 1791 st->codecpar->bit_rate = sipr_bit_rate[flavor];
6b10228a 1792 }
9200514a 1793 st->codecpar->block_align = track->audio.sub_packet_size;
84cfce9f 1794 extradata_offset = 78;
b061d892 1795 }
4efdadc8
AK
1796 } else if (codec_id == AV_CODEC_ID_FLAC && track->codec_priv.size) {
1797 ret = matroska_parse_flac(s, track, &extradata_offset);
1798 if (ret < 0)
1799 return ret;
d88d806b 1800 }
e2644407 1801 track->codec_priv.size -= extradata_offset;
b061d892 1802
36ef5369 1803 if (codec_id == AV_CODEC_ID_NONE)
d88d806b 1804 av_log(matroska->ctx, AV_LOG_INFO,
36ef5369 1805 "Unknown/unsupported AVCodecID %s.\n", track->codec_id);
eabb8ba4 1806
3fc9d7ce
AJ
1807 if (track->time_scale < 0.01)
1808 track->time_scale = 1.0;
84cfce9f
KC
1809 avpriv_set_pts_info(st, 64, matroska->time_scale * track->time_scale,
1810 1000 * 1000 * 1000); /* 64 bit pts in ns */
eabb8ba4 1811
eb3b5501
AK
1812 /* convert the delay from ns to the track timebase */
1813 track->codec_delay = av_rescale_q(track->codec_delay,
1814 (AVRational){ 1, 1000000000 },
1815 st->time_base);
1816
9200514a 1817 st->codecpar->codec_id = codec_id;
84cfce9f 1818 st->start_time = 0;
d88d806b 1819 if (strcmp(track->language, "und"))
d2d67e42
AK
1820 av_dict_set(&st->metadata, "language", track->language, 0);
1821 av_dict_set(&st->metadata, "title", track->name, 0);
b061d892 1822
d88d806b
AJ
1823 if (track->flag_default)
1824 st->disposition |= AV_DISPOSITION_DEFAULT;
7a617a86
AJ
1825 if (track->flag_forced)
1826 st->disposition |= AV_DISPOSITION_FORCED;
b061d892 1827
9200514a 1828 if (!st->codecpar->extradata) {
84cfce9f 1829 if (extradata) {
9200514a
AK
1830 st->codecpar->extradata = extradata;
1831 st->codecpar->extradata_size = extradata_size;
84cfce9f 1832 } else if (track->codec_priv.data && track->codec_priv.size > 0) {
9200514a
AK
1833 st->codecpar->extradata = av_mallocz(track->codec_priv.size +
1834 AV_INPUT_BUFFER_PADDING_SIZE);
1835 if (!st->codecpar->extradata)
553e9f75 1836 return AVERROR(ENOMEM);
9200514a
AK
1837 st->codecpar->extradata_size = track->codec_priv.size;
1838 memcpy(st->codecpar->extradata,
553e9f75
AJ
1839 track->codec_priv.data + extradata_offset,
1840 track->codec_priv.size);
1841 }
ff0d5a74 1842 }
d88d806b
AJ
1843
1844 if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
febfb49a
AC
1845 int display_width_mul = 1;
1846 int display_height_mul = 1;
1847
9200514a
AK
1848 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
1849 st->codecpar->codec_tag = track->video.fourcc;
1850 st->codecpar->width = track->video.pixel_width;
1851 st->codecpar->height = track->video.pixel_height;
febfb49a
AC
1852
1853 if (track->video.stereo_mode && track->video.stereo_mode < MATROSKA_VIDEO_STEREOMODE_TYPE_NB)
1854 mkv_stereo_mode_display_mul(track->video.stereo_mode, &display_width_mul, &display_height_mul);
1855
59729451
AJ
1856 av_reduce(&st->sample_aspect_ratio.num,
1857 &st->sample_aspect_ratio.den,
9200514a
AK
1858 st->codecpar->height * track->video.display_width * display_width_mul,
1859 st->codecpar->width * track->video.display_height * display_height_mul,
d88d806b 1860 255);
9200514a
AK
1861 if (st->codecpar->codec_id != AV_CODEC_ID_H264 &&
1862 st->codecpar->codec_id != AV_CODEC_ID_HEVC)
959bea13 1863 st->need_parsing = AVSTREAM_PARSE_HEADERS;
ac97d47d 1864 if (track->default_duration) {
aba232cf 1865 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
ac97d47d 1866 1000000000, track->default_duration, 30000);
ac97d47d 1867 }
d4ae8ac9
VG
1868 // add stream level stereo3d side data if it is a supported format
1869 if (track->video.stereo_mode < MATROSKA_VIDEO_STEREOMODE_TYPE_NB &&
1870 track->video.stereo_mode != 10 && track->video.stereo_mode != 12) {
1871 int ret = ff_mkv_stereo3d_conv(st, track->video.stereo_mode);
1872 if (ret < 0)
1873 return ret;
1874 }
d88d806b 1875 } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
9200514a
AK
1876 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
1877 st->codecpar->sample_rate = track->audio.out_samplerate;
1878 st->codecpar->channels = track->audio.channels;
1879 if (st->codecpar->codec_id != AV_CODEC_ID_AAC)
84cfce9f 1880 st->need_parsing = AVSTREAM_PARSE_HEADERS;
9200514a 1881 if (st->codecpar->codec_id == AV_CODEC_ID_MP3)
f273f7fb 1882 st->need_parsing = AVSTREAM_PARSE_FULL;
d88d806b 1883 } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
9200514a
AK
1884 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
1885 if (st->codecpar->codec_id == AV_CODEC_ID_SSA)
8336eb6f 1886 matroska->contains_ssa = 1;
d88d806b 1887 }
b061d892
DC
1888 }
1889
6df478bf
AK
1890 return 0;
1891}
1892
1893static int matroska_read_header(AVFormatContext *s)
1894{
1895 MatroskaDemuxContext *matroska = s->priv_data;
1896 EbmlList *attachments_list = &matroska->attachments;
1897 EbmlList *chapters_list = &matroska->chapters;
1898 MatroskaAttachment *attachments;
1899 MatroskaChapter *chapters;
1900 uint64_t max_start = 0;
1901 int64_t pos;
1902 Ebml ebml = { 0 };
1903 int i, j, res;
1904
1905 matroska->ctx = s;
1906
1907 /* First read the EBML header. */
b8d7f318
TG
1908 if (ebml_parse(matroska, ebml_syntax, &ebml) || !ebml.doctype) {
1909 av_log(matroska->ctx, AV_LOG_ERROR, "EBML header parsing failed\n");
1910 ebml_free(ebml_syntax, &ebml);
1911 return AVERROR_INVALIDDATA;
1912 }
1913 if (ebml.version > EBML_VERSION ||
6df478bf
AK
1914 ebml.max_size > sizeof(uint64_t) ||
1915 ebml.id_length > sizeof(uint32_t) ||
d4ae8ac9 1916 ebml.doctype_version > 3) {
6df478bf
AK
1917 av_log(matroska->ctx, AV_LOG_ERROR,
1918 "EBML header using unsupported features\n"
1919 "(EBML version %"PRIu64", doctype %s, doc version %"PRIu64")\n",
1920 ebml.version, ebml.doctype, ebml.doctype_version);
1921 ebml_free(ebml_syntax, &ebml);
1922 return AVERROR_PATCHWELCOME;
1923 }
1924 for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++)
1925 if (!strcmp(ebml.doctype, matroska_doctypes[i]))
1926 break;
1927 if (i >= FF_ARRAY_ELEMS(matroska_doctypes)) {
1928 av_log(s, AV_LOG_WARNING, "Unknown EBML doctype '%s'\n", ebml.doctype);
1929 if (matroska->ctx->error_recognition & AV_EF_EXPLODE) {
1930 ebml_free(ebml_syntax, &ebml);
1931 return AVERROR_INVALIDDATA;
1932 }
1933 }
1934 ebml_free(ebml_syntax, &ebml);
1935
1936 /* The next thing is a segment. */
1937 pos = avio_tell(matroska->ctx->pb);
1938 res = ebml_parse(matroska, matroska_segments, matroska);
1939 // try resyncing until we find a EBML_STOP type element.
1940 while (res != 1) {
1941 res = matroska_resync(matroska, pos);
1942 if (res < 0)
1943 return res;
1944 pos = avio_tell(matroska->ctx->pb);
1945 res = ebml_parse(matroska, matroska_segment, matroska);
1946 }
1947 matroska_execute_seekhead(matroska);
1948
1949 if (!matroska->time_scale)
1950 matroska->time_scale = 1000000;
1951 if (matroska->duration)
1952 matroska->ctx->duration = matroska->duration * matroska->time_scale *
1953 1000 / AV_TIME_BASE;
1954 av_dict_set(&s->metadata, "title", matroska->title, 0);
1955
1956 res = matroska_parse_tracks(s);
1957 if (res < 0)
1958 return res;
1959
f69befe5
DB
1960 attachments = attachments_list->elem;
1961 for (j = 0; j < attachments_list->nb_elem; j++) {
1962 if (!(attachments[j].filename && attachments[j].mime &&
1963 attachments[j].bin.data && attachments[j].bin.size > 0)) {
9c25bafa
AJ
1964 av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
1965 } else {
3b3bbdd3 1966 AVStream *st = avformat_new_stream(s, NULL);
f929ab05 1967 if (!st)
9c25bafa 1968 break;
f69befe5
DB
1969 av_dict_set(&st->metadata, "filename", attachments[j].filename, 0);
1970 av_dict_set(&st->metadata, "mimetype", attachments[j].mime, 0);
9200514a 1971 st->codecpar->codec_id = AV_CODEC_ID_NONE;
c4d37cd9 1972
1973 for (i = 0; ff_mkv_image_mime_tags[i].id != AV_CODEC_ID_NONE; i++) {
1974 if (!strncmp(ff_mkv_image_mime_tags[i].str, attachments[j].mime,
1975 strlen(ff_mkv_image_mime_tags[i].str))) {
9200514a 1976 st->codecpar->codec_id = ff_mkv_image_mime_tags[i].id;
9c25bafa
AJ
1977 break;
1978 }
1979 }
c4d37cd9 1980
f69befe5 1981 attachments[j].stream = st;
c4d37cd9 1982
9200514a
AK
1983 if (st->codecpar->codec_id != AV_CODEC_ID_NONE) {
1984 st->disposition |= AV_DISPOSITION_ATTACHED_PIC;
1985 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
c4d37cd9 1986
1987 av_init_packet(&st->attached_pic);
1988 if ((res = av_new_packet(&st->attached_pic, attachments[j].bin.size)) < 0)
1989 return res;
1990 memcpy(st->attached_pic.data, attachments[j].bin.data, attachments[j].bin.size);
1991 st->attached_pic.stream_index = st->index;
1992 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
1993 } else {
9200514a
AK
1994 st->codecpar->codec_type = AVMEDIA_TYPE_ATTACHMENT;
1995 st->codecpar->extradata = av_malloc(attachments[j].bin.size);
1996 if (!st->codecpar->extradata)
c4d37cd9 1997 break;
1998
9200514a
AK
1999 st->codecpar->extradata_size = attachments[j].bin.size;
2000 memcpy(st->codecpar->extradata, attachments[j].bin.data,
c4d37cd9 2001 attachments[j].bin.size);
2002
2003 for (i = 0; ff_mkv_mime_tags[i].id != AV_CODEC_ID_NONE; i++) {
2004 if (!strncmp(ff_mkv_mime_tags[i].str, attachments[j].mime,
2005 strlen(ff_mkv_mime_tags[i].str))) {
9200514a 2006 st->codecpar->codec_id = ff_mkv_mime_tags[i].id;
c4d37cd9 2007 break;
2008 }
2009 }
2010 }
9c25bafa
AJ
2011 }
2012 }
2013
2014 chapters = chapters_list->elem;
84cfce9f
KC
2015 for (i = 0; i < chapters_list->nb_elem; i++)
2016 if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid &&
2017 (max_start == 0 || chapters[i].start > max_start)) {
6cb6e159 2018 chapters[i].chapter =
84cfce9f
KC
2019 avpriv_new_chapter(s, chapters[i].uid,
2020 (AVRational) { 1, 1000000000 },
2021 chapters[i].start, chapters[i].end,
2022 chapters[i].title);
d2d67e42 2023 av_dict_set(&chapters[i].chapter->metadata,
84cfce9f 2024 "title", chapters[i].title, 0);
e0e4be59
AJ
2025 max_start = chapters[i].start;
2026 }
9c25bafa 2027
929e9de7
AJ
2028 matroska_convert_tags(s);
2029
ce6f28bd 2030 return 0;
b061d892
DC
2031}
2032
737c40da 2033/*
737c40da
AJ
2034 * Put one packet in an application-supplied AVPacket struct.
2035 * Returns 0 on success or -1 on failure.
2036 */
2037static int matroska_deliver_packet(MatroskaDemuxContext *matroska,
2038 AVPacket *pkt)
2039{
2040 if (matroska->num_packets > 0) {
2041 memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
2042 av_free(matroska->packets[0]);
2043 if (matroska->num_packets > 1) {
77d2ef13 2044 void *newpackets;
737c40da
AJ
2045 memmove(&matroska->packets[0], &matroska->packets[1],
2046 (matroska->num_packets - 1) * sizeof(AVPacket *));
77d2ef13 2047 newpackets = av_realloc(matroska->packets,
84cfce9f
KC
2048 (matroska->num_packets - 1) *
2049 sizeof(AVPacket *));
77d2ef13
MN
2050 if (newpackets)
2051 matroska->packets = newpackets;
737c40da
AJ
2052 } else {
2053 av_freep(&matroska->packets);
8336eb6f 2054 matroska->prev_pkt = NULL;
737c40da
AJ
2055 }
2056 matroska->num_packets--;
2057 return 0;
2058 }
2059
2060 return -1;
2061}
2062
2063/*
2064 * Free all packets in our internal queue.
2065 */
2066static void matroska_clear_queue(MatroskaDemuxContext *matroska)
2067{
ae3d4163 2068 matroska->prev_pkt = NULL;
737c40da
AJ
2069 if (matroska->packets) {
2070 int n;
2071 for (n = 0; n < matroska->num_packets; n++) {
ce70f28a 2072 av_packet_unref(matroska->packets[n]);
737c40da
AJ
2073 av_free(matroska->packets[n]);
2074 }
00a3431c 2075 av_freep(&matroska->packets);
737c40da
AJ
2076 matroska->num_packets = 0;
2077 }
2078}
2079
2d0e7713 2080static int matroska_parse_laces(MatroskaDemuxContext *matroska, uint8_t **buf,
84cfce9f 2081 int *buf_size, int type,
2d0e7713
LB
2082 uint32_t **lace_buf, int *laces)
2083{
fc43c19a 2084 int res = 0, n, size = *buf_size;
2d0e7713
LB
2085 uint8_t *data = *buf;
2086 uint32_t *lace_size;
2087
2088 if (!type) {
84cfce9f 2089 *laces = 1;
2d0e7713
LB
2090 *lace_buf = av_mallocz(sizeof(int));
2091 if (!*lace_buf)
2092 return AVERROR(ENOMEM);
2093
2094 *lace_buf[0] = size;
2095 return 0;
2096 }
2097
2098 assert(size > 0);
84cfce9f
KC
2099 *laces = *data + 1;
2100 data += 1;
2101 size -= 1;
2d0e7713
LB
2102 lace_size = av_mallocz(*laces * sizeof(int));
2103 if (!lace_size)
2104 return AVERROR(ENOMEM);
2105
2106 switch (type) {
84cfce9f
KC
2107 case 0x1: /* Xiph lacing */
2108 {
2d0e7713
LB
2109 uint8_t temp;
2110 uint32_t total = 0;
2111 for (n = 0; res == 0 && n < *laces - 1; n++) {
2112 while (1) {
2113 if (size == 0) {
2114 res = AVERROR_EOF;
2115 break;
2116 }
84cfce9f 2117 temp = *data;
2d0e7713 2118 lace_size[n] += temp;
84cfce9f
KC
2119 data += 1;
2120 size -= 1;
2d0e7713
LB
2121 if (temp != 0xff)
2122 break;
2123 }
2124 total += lace_size[n];
2125 }
2126 if (size <= total) {
2127 res = AVERROR_INVALIDDATA;
2128 break;
2129 }
2130
2131 lace_size[n] = size - total;
2132 break;
2133 }
2134
2135 case 0x2: /* fixed-size lacing */
87b017a2 2136 if (size % (*laces)) {
2d0e7713
LB
2137 res = AVERROR_INVALIDDATA;
2138 break;
2139 }
2140 for (n = 0; n < *laces; n++)
2141 lace_size[n] = size / *laces;
2142 break;
2143
84cfce9f
KC
2144 case 0x3: /* EBML lacing */
2145 {
2d0e7713 2146 uint64_t num;
8a96df7b 2147 uint64_t total;
2d0e7713
LB
2148 n = matroska_ebmlnum_uint(matroska, data, size, &num);
2149 if (n < 0) {
2150 av_log(matroska->ctx, AV_LOG_INFO,
2151 "EBML block data error\n");
2152 res = n;
2153 break;
2154 }
2155 data += n;
2156 size -= n;
2157 total = lace_size[0] = num;
2158 for (n = 1; res == 0 && n < *laces - 1; n++) {
2159 int64_t snum;
2160 int r;
2161 r = matroska_ebmlnum_sint(matroska, data, size, &snum);
2162 if (r < 0) {
2163 av_log(matroska->ctx, AV_LOG_INFO,
2164 "EBML block data error\n");
2165 res = r;
2166 break;
2167 }
84cfce9f
KC
2168 data += r;
2169 size -= r;
2d0e7713 2170 lace_size[n] = lace_size[n - 1] + snum;
84cfce9f 2171 total += lace_size[n];
2d0e7713
LB
2172 }
2173 if (size <= total) {
2174 res = AVERROR_INVALIDDATA;
2175 break;
2176 }
2177 lace_size[*laces - 1] = size - total;
2178 break;
2179 }
2180 }
2181
2182 *buf = data;
2183 *lace_buf = lace_size;
fc43c19a 2184 *buf_size = size;
2d0e7713
LB
2185
2186 return res;
2187}
2188
c831ebf6 2189static int matroska_parse_rm_audio(MatroskaDemuxContext *matroska,
84cfce9f
KC
2190 MatroskaTrack *track, AVStream *st,
2191 uint8_t *data, int size, uint64_t timecode,
2192 uint64_t duration, int64_t pos)
c831ebf6 2193{
9200514a 2194 int a = st->codecpar->block_align;
c831ebf6
LB
2195 int sps = track->audio.sub_packet_size;
2196 int cfs = track->audio.coded_framesize;
84cfce9f
KC
2197 int h = track->audio.sub_packet_h;
2198 int y = track->audio.sub_packet_cnt;
2199 int w = track->audio.frame_size;
c831ebf6
LB
2200 int x;
2201
2202 if (!track->audio.pkt_cnt) {
2203 if (track->audio.sub_packet_cnt == 0)
2204 track->audio.buf_timecode = timecode;
9200514a 2205 if (st->codecpar->codec_id == AV_CODEC_ID_RA_288) {
c831ebf6
LB
2206 if (size < cfs * h / 2) {
2207 av_log(matroska->ctx, AV_LOG_ERROR,
2208 "Corrupt int4 RM-style audio packet size\n");
2209 return AVERROR_INVALIDDATA;
2210 }
84cfce9f
KC
2211 for (x = 0; x < h / 2; x++)
2212 memcpy(track->audio.buf + x * 2 * w + y * cfs,
2213 data + x * cfs, cfs);
9200514a 2214 } else if (st->codecpar->codec_id == AV_CODEC_ID_SIPR) {
c831ebf6
LB
2215 if (size < w) {
2216 av_log(matroska->ctx, AV_LOG_ERROR,
2217 "Corrupt sipr RM-style audio packet size\n");
2218 return AVERROR_INVALIDDATA;
2219 }
84cfce9f 2220 memcpy(track->audio.buf + y * w, data, w);
c831ebf6
LB
2221 } else {
2222 if (size < sps * w / sps) {
2223 av_log(matroska->ctx, AV_LOG_ERROR,
2224 "Corrupt generic RM-style audio packet size\n");
2225 return AVERROR_INVALIDDATA;
2226 }
84cfce9f
KC
2227 for (x = 0; x < w / sps; x++)
2228 memcpy(track->audio.buf +
2229 sps * (h * x + ((h + 1) / 2) * (y & 1) + (y >> 1)),
2230 data + x * sps, sps);
c831ebf6
LB
2231 }
2232
2233 if (++track->audio.sub_packet_cnt >= h) {
9200514a 2234 if (st->codecpar->codec_id == AV_CODEC_ID_SIPR)
c831ebf6
LB
2235 ff_rm_reorder_sipr_data(track->audio.buf, h, w);
2236 track->audio.sub_packet_cnt = 0;
84cfce9f 2237 track->audio.pkt_cnt = h * w / a;
c831ebf6
LB
2238 }
2239 }
2240
2241 while (track->audio.pkt_cnt) {
e0caa1eb 2242 int ret;
c831ebf6 2243 AVPacket *pkt = av_mallocz(sizeof(AVPacket));
e0caa1eb
VG
2244 if (!pkt)
2245 return AVERROR(ENOMEM);
2246
2247 ret = av_new_packet(pkt, a);
2248 if (ret < 0) {
2249 av_free(pkt);
2250 return ret;
2251 }
84cfce9f
KC
2252 memcpy(pkt->data,
2253 track->audio.buf + a * (h * w / a - track->audio.pkt_cnt--),
2254 a);
2255 pkt->pts = track->audio.buf_timecode;
c831ebf6 2256 track->audio.buf_timecode = AV_NOPTS_VALUE;
84cfce9f
KC
2257 pkt->pos = pos;
2258 pkt->stream_index = st->index;
2259 dynarray_add(&matroska->packets, &matroska->num_packets, pkt);
c831ebf6
LB
2260 }
2261
2262 return 0;
2263}
9b6f47c4
AK
2264
2265/* reconstruct full wavpack blocks from mangled matroska ones */
2266static int matroska_parse_wavpack(MatroskaTrack *track, uint8_t *src,
2267 uint8_t **pdst, int *size)
2268{
2269 uint8_t *dst = NULL;
2270 int dstlen = 0;
2271 int srclen = *size;
2272 uint32_t samples;
2273 uint16_t ver;
2274 int ret, offset = 0;
2275
9200514a 2276 if (srclen < 12 || track->stream->codecpar->extradata_size < 2)
9b6f47c4
AK
2277 return AVERROR_INVALIDDATA;
2278
9200514a 2279 ver = AV_RL16(track->stream->codecpar->extradata);
9b6f47c4
AK
2280
2281 samples = AV_RL32(src);
2282 src += 4;
2283 srclen -= 4;
2284
2285 while (srclen >= 8) {
2286 int multiblock;
2287 uint32_t blocksize;
2288 uint8_t *tmp;
2289
2290 uint32_t flags = AV_RL32(src);
2291 uint32_t crc = AV_RL32(src + 4);
2292 src += 8;
2293 srclen -= 8;
2294
2295 multiblock = (flags & 0x1800) != 0x1800;
2296 if (multiblock) {
2297 if (srclen < 4) {
2298 ret = AVERROR_INVALIDDATA;
2299 goto fail;
2300 }
2301 blocksize = AV_RL32(src);
84cfce9f
KC
2302 src += 4;
2303 srclen -= 4;
9b6f47c4
AK
2304 } else
2305 blocksize = srclen;
2306
2307 if (blocksize > srclen) {
2308 ret = AVERROR_INVALIDDATA;
2309 goto fail;
2310 }
2311
2312 tmp = av_realloc(dst, dstlen + blocksize + 32);
2313 if (!tmp) {
2314 ret = AVERROR(ENOMEM);
2315 goto fail;
2316 }
2317 dst = tmp;
2318 dstlen += blocksize + 32;
2319
84cfce9f
KC
2320 AV_WL32(dst + offset, MKTAG('w', 'v', 'p', 'k')); // tag
2321 AV_WL32(dst + offset + 4, blocksize + 24); // blocksize - 8
2322 AV_WL16(dst + offset + 8, ver); // version
2323 AV_WL16(dst + offset + 10, 0); // track/index_no
2324 AV_WL32(dst + offset + 12, 0); // total samples
2325 AV_WL32(dst + offset + 16, 0); // block index
2326 AV_WL32(dst + offset + 20, samples); // number of samples
2327 AV_WL32(dst + offset + 24, flags); // flags
2328 AV_WL32(dst + offset + 28, crc); // crc
2329 memcpy(dst + offset + 32, src, blocksize); // block data
9b6f47c4
AK
2330
2331 src += blocksize;
2332 srclen -= blocksize;
2333 offset += blocksize + 32;
2334 }
2335
2336 *pdst = dst;
2337 *size = dstlen;
2338
2339 return 0;
2340
2341fail:
2342 av_freep(&dst);
2343 return ret;
2344}
2345
c831ebf6 2346static int matroska_parse_frame(MatroskaDemuxContext *matroska,
84cfce9f 2347 MatroskaTrack *track, AVStream *st,
c831ebf6
LB
2348 uint8_t *data, int pkt_size,
2349 uint64_t timecode, uint64_t duration,
2350 int64_t pos, int is_keyframe)
2351{
2352 MatroskaTrackEncoding *encodings = track->encodings.elem;
2353 uint8_t *pkt_data = data;
2354 int offset = 0, res;
2355 AVPacket *pkt;
2356
2357 if (encodings && encodings->scope & 1) {
2358 res = matroska_decode_buffer(&pkt_data, &pkt_size, track);
2359 if (res < 0)
2360 return res;
2361 }
2362
9200514a 2363 if (st->codecpar->codec_id == AV_CODEC_ID_WAVPACK) {
9b6f47c4
AK
2364 uint8_t *wv_data;
2365 res = matroska_parse_wavpack(track, pkt_data, &wv_data, &pkt_size);
2366 if (res < 0) {
84cfce9f
KC
2367 av_log(matroska->ctx, AV_LOG_ERROR,
2368 "Error parsing a wavpack block.\n");
9b6f47c4
AK
2369 goto fail;
2370 }
2371 if (pkt_data != data)
2372 av_freep(&pkt_data);
2373 pkt_data = wv_data;
2374 }
2375
9200514a 2376 if (st->codecpar->codec_id == AV_CODEC_ID_PRORES)
c831ebf6
LB
2377 offset = 8;
2378
2379 pkt = av_mallocz(sizeof(AVPacket));
6bdae41d
VG
2380 if (!pkt) {
2381 av_freep(&pkt_data);
2382 return AVERROR(ENOMEM);
2383 }
c831ebf6
LB
2384 /* XXX: prevent data copy... */
2385 if (av_new_packet(pkt, pkt_size + offset) < 0) {
2386 av_free(pkt);
3c1199c3 2387 av_freep(&pkt_data);
c831ebf6
LB
2388 return AVERROR(ENOMEM);
2389 }
2390
9200514a 2391 if (st->codecpar->codec_id == AV_CODEC_ID_PRORES) {
c831ebf6
LB
2392 uint8_t *buf = pkt->data;
2393 bytestream_put_be32(&buf, pkt_size);
2394 bytestream_put_be32(&buf, MKBETAG('i', 'c', 'p', 'f'));
2395 }
2396
2397 memcpy(pkt->data + offset, pkt_data, pkt_size);
2398
2399 if (pkt_data != data)
2400 av_free(pkt_data);
2401
84cfce9f 2402 pkt->flags = is_keyframe;
c831ebf6
LB
2403 pkt->stream_index = st->index;
2404
2405 if (track->ms_compat)
2406 pkt->dts = timecode;
2407 else
2408 pkt->pts = timecode;
2409 pkt->pos = pos;
9200514a 2410 if (track->type != MATROSKA_TRACK_TYPE_SUBTITLE || st->codecpar->codec_id == AV_CODEC_ID_TEXT)
948f3c19 2411 pkt->duration = duration;
2412#if FF_API_CONVERGENCE_DURATION
2413FF_DISABLE_DEPRECATION_WARNINGS
9200514a 2414 if (st->codecpar->codec_id == AV_CODEC_ID_TEXT)
c831ebf6 2415 pkt->convergence_duration = duration;
948f3c19 2416FF_ENABLE_DEPRECATION_WARNINGS
2417#endif
c831ebf6 2418
9200514a 2419 if (st->codecpar->codec_id == AV_CODEC_ID_SSA)
c831ebf6
LB
2420 matroska_fix_ass_packet(matroska, pkt, duration);
2421
84cfce9f
KC
2422 if (matroska->prev_pkt &&
2423 timecode != AV_NOPTS_VALUE &&
2424 matroska->prev_pkt->pts == timecode &&
2425 matroska->prev_pkt->stream_index == st->index &&
9200514a 2426 st->codecpar->codec_id == AV_CODEC_ID_SSA)
c831ebf6
LB
2427 matroska_merge_packets(matroska->prev_pkt, pkt);
2428 else {
84cfce9f 2429 dynarray_add(&matroska->packets, &matroska->num_packets, pkt);
c831ebf6
LB
2430 matroska->prev_pkt = pkt;
2431 }
2432
2433 return 0;
84cfce9f 2434
9b6f47c4
AK
2435fail:
2436 if (pkt_data != data)
2437 av_freep(&pkt_data);
2438 return res;
c831ebf6
LB
2439}
2440
f7b9687c
AJ
2441static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data,
2442 int size, int64_t pos, uint64_t cluster_time,
7d843100 2443 uint64_t block_duration, int is_keyframe,
24c3da16 2444 int64_t cluster_pos)
b061d892 2445{
f14a201b 2446 uint64_t timecode = AV_NOPTS_VALUE;
009ecd50 2447 MatroskaTrack *track;
a3467f86 2448 int res = 0;
b061d892 2449 AVStream *st;
b061d892
DC
2450 int16_t block_time;
2451 uint32_t *lace_size = NULL;
2452 int n, flags, laces = 0;
7d843100 2453 uint64_t num, duration;
b061d892 2454
c1e01133 2455 if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) {
b061d892 2456 av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
721af294 2457 return n;
b061d892
DC
2458 }
2459 data += n;
2460 size -= n;
2461
b061d892 2462 track = matroska_find_track_by_num(matroska, num);
d31fb1a9 2463 if (!track || !track->stream) {
b061d892 2464 av_log(matroska->ctx, AV_LOG_INFO,
009ecd50 2465 "Invalid stream %"PRIu64" or size %u\n", num, size);
5dd514af 2466 return AVERROR_INVALIDDATA;
d31fb1a9
RB
2467 } else if (size <= 3)
2468 return 0;
009ecd50 2469 st = track->stream;
16f97ab0 2470 if (st->discard >= AVDISCARD_ALL)
b061d892 2471 return res;
b061d892 2472
2ce746c6 2473 block_time = AV_RB16(data);
84cfce9f
KC
2474 data += 2;
2475 flags = *data++;
2476 size -= 3;
b061d892 2477 if (is_keyframe == -1)
cc947f04 2478 is_keyframe = flags & 0x80 ? AV_PKT_FLAG_KEY : 0;
b061d892 2479
84cfce9f
KC
2480 if (cluster_time != (uint64_t) -1 &&
2481 (block_time >= 0 || cluster_time >= -block_time)) {
eb3b5501 2482 timecode = cluster_time + block_time - track->codec_delay;
84cfce9f
KC
2483 if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE &&
2484 timecode < track->end_timecode)
82360e67 2485 is_keyframe = 0; /* overlapping subtitles are not key frame */
a8fd7e76 2486 if (is_keyframe)
84cfce9f
KC
2487 av_add_index_entry(st, cluster_pos, timecode, 0, 0,
2488 AVINDEX_KEYFRAME);
f14a201b
AJ
2489 }
2490
84cfce9f
KC
2491 if (matroska->skip_to_keyframe &&
2492 track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
20f74665 2493 if (!is_keyframe || timecode < matroska->skip_to_timecode)
b061d892
DC
2494 return res;
2495 matroska->skip_to_keyframe = 0;
2496 }
2497
fc43c19a 2498 res = matroska_parse_laces(matroska, &data, &size, (flags & 0x06) >> 1,
2d0e7713 2499 &lace_size, &laces);
b061d892 2500
c831ebf6
LB
2501 if (res)
2502 goto end;
8d4dd55c 2503
7d843100
LB
2504 if (block_duration != AV_NOPTS_VALUE) {
2505 duration = block_duration / laces;
2506 if (block_duration != duration * laces) {
2507 av_log(matroska->ctx, AV_LOG_WARNING,
2508 "Incorrect block_duration, possibly corrupted container");
2509 }
2510 } else {
84cfce9f 2511 duration = track->default_duration / matroska->time_scale;
7d843100
LB
2512 block_duration = duration * laces;
2513 }
2514
2515 if (timecode != AV_NOPTS_VALUE)
2516 track->end_timecode =
2517 FFMAX(track->end_timecode, timecode + block_duration);
2518
c831ebf6 2519 for (n = 0; n < laces; n++) {
9200514a
AK
2520 if ((st->codecpar->codec_id == AV_CODEC_ID_RA_288 ||
2521 st->codecpar->codec_id == AV_CODEC_ID_COOK ||
2522 st->codecpar->codec_id == AV_CODEC_ID_SIPR ||
2523 st->codecpar->codec_id == AV_CODEC_ID_ATRAC3) &&
2524 st->codecpar->block_align && track->audio.sub_packet_size) {
25a80a93
LB
2525 res = matroska_parse_rm_audio(matroska, track, st, data,
2526 lace_size[n],
c831ebf6
LB
2527 timecode, duration, pos);
2528 if (res)
2529 goto end;
c831ebf6
LB
2530 } else {
2531 res = matroska_parse_frame(matroska, track, st, data, lace_size[n],
84cfce9f
KC
2532 timecode, duration, pos,
2533 !n ? is_keyframe : 0);
c831ebf6
LB
2534 if (res)
2535 goto end;
b061d892 2536 }
c831ebf6
LB
2537
2538 if (timecode != AV_NOPTS_VALUE)
2539 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
2540 data += lace_size[n];
b061d892
DC
2541 }
2542
31168588 2543end:
b061d892 2544 av_free(lace_size);
a3467f86 2545 return res;
b061d892
DC
2546}
2547
8336eb6f
DC
2548static int matroska_parse_cluster_incremental(MatroskaDemuxContext *matroska)
2549{
2550 EbmlList *blocks_list;
2551 MatroskaBlock *blocks;
2552 int i, res;
2553 res = ebml_parse(matroska,
2554 matroska_cluster_incremental_parsing,
2555 &matroska->current_cluster);
2556 if (res == 1) {
2557 /* New Cluster */
2558 if (matroska->current_cluster_pos)
2559 ebml_level_end(matroska);
2560 ebml_free(matroska_cluster, &matroska->current_cluster);
2561 memset(&matroska->current_cluster, 0, sizeof(MatroskaCluster));
2562 matroska->current_cluster_num_blocks = 0;
84cfce9f
KC
2563 matroska->current_cluster_pos = avio_tell(matroska->ctx->pb);
2564 matroska->prev_pkt = NULL;
8336eb6f
DC
2565 /* sizeof the ID which was already read */
2566 if (matroska->current_id)
2567 matroska->current_cluster_pos -= 4;
2568 res = ebml_parse(matroska,
2569 matroska_clusters_incremental,
2570 &matroska->current_cluster);
2571 /* Try parsing the block again. */
2572 if (res == 1)
2573 res = ebml_parse(matroska,
2574 matroska_cluster_incremental_parsing,
2575 &matroska->current_cluster);
2576 }
2577
2578 if (!res &&
2579 matroska->current_cluster_num_blocks <
84cfce9f 2580 matroska->current_cluster.blocks.nb_elem) {
8336eb6f 2581 blocks_list = &matroska->current_cluster.blocks;
84cfce9f 2582 blocks = blocks_list->elem;
8336eb6f
DC
2583
2584 matroska->current_cluster_num_blocks = blocks_list->nb_elem;
84cfce9f 2585 i = blocks_list->nb_elem - 1;
8336eb6f
DC
2586 if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
2587 int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
2588 if (!blocks[i].non_simple)
2589 blocks[i].duration = AV_NOPTS_VALUE;
84cfce9f
KC
2590 res = matroska_parse_block(matroska, blocks[i].bin.data,
2591 blocks[i].bin.size, blocks[i].bin.pos,
8336eb6f
DC
2592 matroska->current_cluster.timecode,
2593 blocks[i].duration, is_keyframe,
2594 matroska->current_cluster_pos);
2595 }
2596 }
2597
84cfce9f
KC
2598 if (res < 0)
2599 matroska->done = 1;
8336eb6f
DC
2600 return res;
2601}
2602
f7b9687c 2603static int matroska_parse_cluster(MatroskaDemuxContext *matroska)
b061d892 2604{
209472b4
AJ
2605 MatroskaCluster cluster = { 0 };
2606 EbmlList *blocks_list;
2607 MatroskaBlock *blocks;
24c3da16 2608 int i, res;
8336eb6f 2609 int64_t pos;
84cfce9f 2610
8336eb6f
DC
2611 if (!matroska->contains_ssa)
2612 return matroska_parse_cluster_incremental(matroska);
2613 pos = avio_tell(matroska->ctx->pb);
d5e34dcb 2614 matroska->prev_pkt = NULL;
80702035 2615 if (matroska->current_id)
8bc98ba9 2616 pos -= 4; /* sizeof the ID which was already read */
84cfce9f 2617 res = ebml_parse(matroska, matroska_clusters, &cluster);
209472b4 2618 blocks_list = &cluster.blocks;
84cfce9f
KC
2619 blocks = blocks_list->elem;
2620 for (i = 0; i < blocks_list->nb_elem && !res; i++)
37dd2356 2621 if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
194d4b49 2622 int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
98a7d560
JS
2623 if (!blocks[i].non_simple)
2624 blocks[i].duration = AV_NOPTS_VALUE;
84cfce9f
KC
2625 res = matroska_parse_block(matroska, blocks[i].bin.data,
2626 blocks[i].bin.size, blocks[i].bin.pos,
2627 cluster.timecode, blocks[i].duration,
2628 is_keyframe, pos);
194d4b49 2629 }
209472b4 2630 ebml_free(matroska_cluster, &cluster);
b061d892
DC
2631 return res;
2632}
2633
f7b9687c 2634static int matroska_read_packet(AVFormatContext *s, AVPacket *pkt)
b061d892
DC
2635{
2636 MatroskaDemuxContext *matroska = s->priv_data;
5dd514af 2637 int ret = 0;
b061d892 2638
5dd514af 2639 while (!ret && matroska_deliver_packet(matroska, pkt)) {
8835c554 2640 int64_t pos = avio_tell(matroska->ctx->pb);
b061d892 2641 if (matroska->done)
9ebeea82 2642 return AVERROR_EOF;
8835c554
SM
2643 if (matroska_parse_cluster(matroska) < 0)
2644 ret = matroska_resync(matroska, pos);
b061d892
DC
2645 }
2646
bdb939ad 2647 if (ret == AVERROR_INVALIDDATA && pkt->data) {
0ca4642e
LB
2648 pkt->flags |= AV_PKT_FLAG_CORRUPT;
2649 return 0;
2650 }
2651
5dd514af 2652 return ret;
b061d892
DC
2653}
2654
f7b9687c
AJ
2655static int matroska_read_seek(AVFormatContext *s, int stream_index,
2656 int64_t timestamp, int flags)
b061d892
DC
2657{
2658 MatroskaDemuxContext *matroska = s->priv_data;
490a3ebf 2659 MatroskaTrack *tracks = NULL;
b061d892 2660 AVStream *st = s->streams[stream_index];
c1658257 2661 int i, index, index_sub, index_min;
b061d892 2662
31ad14c2
AC
2663 /* Parse the CUES now since we need the index data to seek. */
2664 if (matroska->cues_parsing_deferred) {
2665 matroska_parse_cues(matroska);
2666 matroska->cues_parsing_deferred = 0;
2667 }
2668
a8fd7e76
AJ
2669 if (!st->nb_index_entries)
2670 return 0;
2671 timestamp = FFMAX(timestamp, st->index_entries[0].timestamp);
dfbbbdc0 2672
6bef5f9b 2673 if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
84cfce9f
KC
2674 avio_seek(s->pb, st->index_entries[st->nb_index_entries - 1].pos,
2675 SEEK_SET);
cdc2c1c5 2676 matroska->current_id = 0;
0dbddda3
AJ
2677 while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
2678 matroska_clear_queue(matroska);
2679 if (matroska_parse_cluster(matroska) < 0)
2680 break;
2681 }
6bef5f9b 2682 }
b061d892 2683
243cc4c3 2684 matroska_clear_queue(matroska);
0f646a22
AJ
2685 if (index < 0)
2686 return 0;
243cc4c3 2687
c1658257 2688 index_min = index;
490a3ebf 2689 tracks = matroska->tracks.elem;
84cfce9f
KC
2690 for (i = 0; i < matroska->tracks.nb_elem; i++) {
2691 tracks[i].audio.pkt_cnt = 0;
46544204 2692 tracks[i].audio.sub_packet_cnt = 0;
84cfce9f
KC
2693 tracks[i].audio.buf_timecode = AV_NOPTS_VALUE;
2694 tracks[i].end_timecode = 0;
2695 if (tracks[i].type == MATROSKA_TRACK_TYPE_SUBTITLE &&
98569d89 2696 tracks[i].stream->discard != AVDISCARD_ALL) {
84cfce9f
KC
2697 index_sub = av_index_search_timestamp(
2698 tracks[i].stream, st->index_entries[index].timestamp,
2699 AVSEEK_FLAG_BACKWARD);
2700 if (index_sub >= 0 &&
2701 st->index_entries[index_sub].pos < st->index_entries[index_min].pos &&
2702 st->index_entries[index].timestamp -
2703 st->index_entries[index_sub].timestamp < 30000000000 / matroska->time_scale)
c1658257
AJ
2704 index_min = index_sub;
2705 }
2706 }
2707
6b4aa5da 2708 avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET);
84cfce9f 2709 matroska->current_id = 0;
b061d892 2710 matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
20f74665 2711 matroska->skip_to_timecode = st->index_entries[index].timestamp;
84cfce9f 2712 matroska->done = 0;
a2faa951 2713 ff_update_cur_dts(s, st, st->index_entries[index].timestamp);
b061d892
DC
2714 return 0;
2715}
2716
f7b9687c 2717static int matroska_read_close(AVFormatContext *s)
b061d892
DC
2718{
2719 MatroskaDemuxContext *matroska = s->priv_data;
2cbc8811 2720 MatroskaTrack *tracks = matroska->tracks.elem;
70109c0d 2721 int n;
b061d892 2722
34c9c1ba 2723 matroska_clear_queue(matroska);
b061d892 2724
84cfce9f 2725 for (n = 0; n < matroska->tracks.nb_elem; n++)
2cbc8811
AJ
2726 if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
2727 av_free(tracks[n].audio.buf);
8336eb6f 2728 ebml_free(matroska_cluster, &matroska->current_cluster);
ce6f28bd 2729 ebml_free(matroska_segment, matroska);
b061d892
DC
2730
2731 return 0;
2732}
2733
c6610a21 2734AVInputFormat ff_matroska_demuxer = {
dfc2c4d9 2735 .name = "matroska,webm",
6774247a 2736 .long_name = NULL_IF_CONFIG_SMALL("Matroska / WebM"),
f2583bc8 2737 .extensions = "mkv,mk3d,mka,mks",
dfc2c4d9
AK
2738 .priv_data_size = sizeof(MatroskaDemuxContext),
2739 .read_probe = matroska_probe,
2740 .read_header = matroska_read_header,
2741 .read_packet = matroska_read_packet,
2742 .read_close = matroska_read_close,
2743 .read_seek = matroska_read_seek,
fa38573c 2744 .mime_type = "audio/webm,audio/x-matroska,video/webm,video/x-matroska"
b061d892 2745};