blob: b168bdf9a63a2ba7b83963afc043a1e6a0d0d8bf [file] [log] [blame]
David Conradb061d892007-06-04 22:10:541/*
Aurelien Jacobsff33c5c2008-08-05 00:42:432 * Matroska file demuxer
Mans Rullgard2912e872011-03-18 17:35:103 * Copyright (c) 2003-2008 The Libav Project
David Conradb061d892007-06-04 22:10:544 *
Mans Rullgard2912e872011-03-18 17:35:105 * This file is part of Libav.
David Conradb061d892007-06-04 22:10:546 *
Mans Rullgard2912e872011-03-18 17:35:107 * Libav is free software; you can redistribute it and/or
David Conradb061d892007-06-04 22:10:548 * 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 *
Mans Rullgard2912e872011-03-18 17:35:1012 * Libav is distributed in the hope that it will be useful,
David Conradb061d892007-06-04 22:10:5413 * 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
Mans Rullgard2912e872011-03-18 17:35:1018 * License along with Libav; if not, write to the Free Software
David Conradb061d892007-06-04 22:10:5419 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22/**
Diego Biurrunba87f082010-04-20 14:45:3423 * @file
David Conradb061d892007-06-04 22:10:5424 * Matroska file demuxer
Diego Biurrune873c032011-10-30 18:10:5025 * @author Ronald Bultje <[email protected]>
26 * @author with a little help from Moritz Bunkus <[email protected]>
27 * @author totally reworked by Aurelien Jacobs <[email protected]>
28 * @see specs available on the Matroska project page: http://www.matroska.org/
David Conradb061d892007-06-04 22:10:5429 */
30
Keiji Costantini84cfce92014-03-01 16:28:1531#include "config.h"
32
Diego Biurrund92024f2014-03-10 14:35:5933#include <inttypes.h>
Aurelien Jacobs3eb9bfb2008-09-04 23:26:1234#include <stdio.h>
Aurelien Jacobsb250f9c2009-01-13 23:44:1635#if CONFIG_BZLIB
Aurelien Jacobs54dddf02008-05-15 23:12:4136#include <bzlib.h>
37#endif
Keiji Costantini84cfce92014-03-01 16:28:1538#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"
Anton Khirnoveb3b5502014-04-29 10:03:1347#include "libavutil/mathematics.h"
Keiji Costantini84cfce92014-03-01 16:28:1548
49#include "libavcodec/bytestream.h"
Anton Khirnov4efdadc2014-05-25 12:05:5150#include "libavcodec/flac.h"
Keiji Costantini84cfce92014-03-01 16:28:1551#include "libavcodec/mpeg4audio.h"
52
53#include "avformat.h"
54#include "avio_internal.h"
55#include "internal.h"
56#include "isom.h"
57#include "matroska.h"
Anton Khirnov23f741f2014-05-26 10:48:5658#include "oggdec.h"
Keiji Costantini84cfce92014-03-01 16:28:1559/* For ff_codec_get_id(). */
60#include "riff.h"
61#include "rmsipr.h"
David Conradb061d892007-06-04 22:10:5462
Aurelien Jacobs789ed102008-08-05 00:40:0063typedef 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,
Reimar Döffinger95ec3d42011-02-06 10:32:0373 EBML_TYPE_COUNT
Aurelien Jacobs789ed102008-08-05 00:40:0074} EbmlType;
75
76typedef const struct EbmlSyntax {
77 uint32_t id;
78 EbmlType type;
79 int list_elem_size;
80 int data_offset;
81 union {
82 uint64_t u;
83 double f;
84 const char *s;
85 const struct EbmlSyntax *n;
86 } def;
87} EbmlSyntax;
88
89typedef struct {
90 int nb_elem;
91 void *elem;
92} EbmlList;
93
94typedef struct {
95 int size;
96 uint8_t *data;
97 int64_t pos;
98} EbmlBin;
99
Aurelien Jacobs63511322008-08-05 00:40:02100typedef struct {
101 uint64_t version;
102 uint64_t max_size;
103 uint64_t id_length;
104 char *doctype;
105 uint64_t doctype_version;
106} Ebml;
107
Aurelien Jacobs2cbc8812008-08-05 00:40:31108typedef struct {
109 uint64_t algo;
110 EbmlBin settings;
111} MatroskaTrackCompression;
David Conradb061d892007-06-04 22:10:54112
Aurelien Jacobs2cbc8812008-08-05 00:40:31113typedef struct {
114 uint64_t scope;
115 uint64_t type;
116 MatroskaTrackCompression compression;
117} MatroskaTrackEncoding;
David Conradb061d892007-06-04 22:10:54118
Aurelien Jacobs2cbc8812008-08-05 00:40:31119typedef struct {
120 double frame_rate;
121 uint64_t display_width;
122 uint64_t display_height;
123 uint64_t pixel_width;
124 uint64_t pixel_height;
125 uint64_t fourcc;
126} MatroskaTrackVideo;
David Conradb061d892007-06-04 22:10:54127
Aurelien Jacobs2cbc8812008-08-05 00:40:31128typedef struct {
129 double samplerate;
130 double out_samplerate;
131 uint64_t bitdepth;
132 uint64_t channels;
David Conradb061d892007-06-04 22:10:54133
Aurelien Jacobs2cbc8812008-08-05 00:40:31134 /* real audio header (extracted from extradata) */
135 int coded_framesize;
136 int sub_packet_h;
137 int frame_size;
138 int sub_packet_size;
139 int sub_packet_cnt;
140 int pkt_cnt;
Reimar Döffinger46544202011-02-26 11:52:01141 uint64_t buf_timecode;
Aurelien Jacobs2cbc8812008-08-05 00:40:31142 uint8_t *buf;
143} MatroskaTrackAudio;
David Conradb061d892007-06-04 22:10:54144
Aurelien Jacobs2cbc8812008-08-05 00:40:31145typedef struct {
146 uint64_t num;
Aurelien Jacobs325ace32009-02-15 15:34:22147 uint64_t uid;
Aurelien Jacobs2cbc8812008-08-05 00:40:31148 uint64_t type;
Aurelien Jacobs38766e02009-02-15 15:29:09149 char *name;
Aurelien Jacobs2cbc8812008-08-05 00:40:31150 char *codec_id;
151 EbmlBin codec_priv;
152 char *language;
Anton Khirnov7ff97082008-06-01 13:54:11153 double time_scale;
David Conradb061d892007-06-04 22:10:54154 uint64_t default_duration;
Aurelien Jacobs4eff9742008-08-05 00:39:53155 uint64_t flag_default;
Aurelien Jacobs7a617a82010-07-02 16:38:44156 uint64_t flag_forced;
Aurelien Jacobs2cbc8812008-08-05 00:40:31157 MatroskaTrackVideo video;
158 MatroskaTrackAudio audio;
159 EbmlList encodings;
Anton Khirnoveb3b5502014-04-29 10:03:13160 uint64_t codec_delay;
Aurelien Jacobsfc4d3352008-08-05 00:40:06161
162 AVStream *stream;
Aurelien Jacobs82360e62008-09-09 12:07:10163 int64_t end_timecode;
Joakim Plate3e93c8e2010-03-03 21:46:43164 int ms_compat;
David Conradb061d892007-06-04 22:10:54165} MatroskaTrack;
166
Aurelien Jacobse5929fd2008-08-05 00:40:15167typedef struct {
Aurelien Jacobs325ace32009-02-15 15:34:22168 uint64_t uid;
Aurelien Jacobsb414cb82008-08-05 00:40:24169 char *filename;
170 char *mime;
171 EbmlBin bin;
Aurelien Jacobs929e9de2009-02-15 15:53:55172
173 AVStream *stream;
Diego Biurrunf69befe2014-03-07 12:54:18174} MatroskaAttachment;
Aurelien Jacobsb414cb82008-08-05 00:40:24175
176typedef struct {
Aurelien Jacobs6bbd7c72008-08-05 00:40:21177 uint64_t start;
178 uint64_t end;
179 uint64_t uid;
180 char *title;
Aurelien Jacobs6cb6e152009-02-15 15:25:14181
182 AVChapter *chapter;
Aurelien Jacobs6bbd7c72008-08-05 00:40:21183} MatroskaChapter;
184
185typedef struct {
Aurelien Jacobse5929fd2008-08-05 00:40:15186 uint64_t track;
187 uint64_t pos;
188} MatroskaIndexPos;
189
190typedef struct {
191 uint64_t time;
192 EbmlList pos;
193} MatroskaIndex;
194
Aurelien Jacobs13b350a2008-08-05 00:40:36195typedef struct {
Aurelien Jacobs44015c52008-08-08 23:50:38196 char *name;
197 char *string;
Aurelien Jacobsf702df32009-02-15 16:05:37198 char *lang;
199 uint64_t def;
Aurelien Jacobs44015c52008-08-08 23:50:38200 EbmlList sub;
201} MatroskaTag;
202
203typedef struct {
Aurelien Jacobs929e9de2009-02-15 15:53:55204 char *type;
205 uint64_t typevalue;
206 uint64_t trackuid;
207 uint64_t chapteruid;
208 uint64_t attachuid;
209} MatroskaTagTarget;
210
211typedef struct {
212 MatroskaTagTarget target;
213 EbmlList tag;
214} MatroskaTags;
215
216typedef struct {
Aurelien Jacobs13b350a2008-08-05 00:40:36217 uint64_t id;
218 uint64_t pos;
219} MatroskaSeekhead;
220
Aurelien Jacobsc171af92008-08-05 00:41:19221typedef struct {
Aurelien Jacobs8d75b5a2007-06-04 22:35:16222 uint64_t start;
223 uint64_t length;
David Conradb061d892007-06-04 22:10:54224} MatroskaLevel;
225
Aurelien Jacobsc171af92008-08-05 00:41:19226typedef struct {
Dale Curtis8336eb62012-04-19 18:12:24227 uint64_t timecode;
228 EbmlList blocks;
229} MatroskaCluster;
230
231typedef struct {
David Conradb061d892007-06-04 22:10:54232 AVFormatContext *ctx;
233
Diego Biurrun5968d2d2008-08-05 08:28:57234 /* EBML stuff */
David Conradb061d892007-06-04 22:10:54235 int num_levels;
236 MatroskaLevel levels[EBML_MAX_DEPTH];
237 int level_up;
Aurelien Jacobsc3ade622010-06-11 16:34:01238 uint32_t current_id;
David Conradb061d892007-06-04 22:10:54239
Aurelien Jacobs29708582008-08-05 00:40:27240 uint64_t time_scale;
241 double duration;
242 char *title;
Aurelien Jacobs2cbc8812008-08-05 00:40:31243 EbmlList tracks;
Aurelien Jacobsb414cb82008-08-05 00:40:24244 EbmlList attachments;
Aurelien Jacobs6bbd7c72008-08-05 00:40:21245 EbmlList chapters;
Aurelien Jacobse5929fd2008-08-05 00:40:15246 EbmlList index;
Aurelien Jacobs44015c52008-08-08 23:50:38247 EbmlList tags;
Aurelien Jacobs13b350a2008-08-05 00:40:36248 EbmlList seekhead;
David Conradb061d892007-06-04 22:10:54249
David Conradb061d892007-06-04 22:10:54250 /* byte position of the segment inside the stream */
Diego Biurrunbc5c9182008-10-03 10:16:29251 int64_t segment_start;
David Conradb061d892007-06-04 22:10:54252
Diego Biurrun5968d2d2008-08-05 08:28:57253 /* the packet queue */
David Conradb061d892007-06-04 22:10:54254 AVPacket **packets;
255 int num_packets;
Aurelien Jacobsd5e34dc2008-09-28 23:06:25256 AVPacket *prev_pkt;
David Conradb061d892007-06-04 22:10:54257
Aurelien Jacobs8d75b5a2007-06-04 22:35:16258 int done;
David Conradb061d892007-06-04 22:10:54259
David Conradb061d892007-06-04 22:10:54260 /* What to skip before effectively reading a packet. */
261 int skip_to_keyframe;
Aurelien Jacobs20f74662008-09-09 12:01:51262 uint64_t skip_to_timecode;
Aaron Colwell31ad14c2011-07-09 05:48:43263
264 /* File has a CUES element, but we defer parsing until it is needed. */
265 int cues_parsing_deferred;
Dale Curtis8336eb62012-04-19 18:12:24266
267 int current_cluster_num_blocks;
268 int64_t current_cluster_pos;
269 MatroskaCluster current_cluster;
270
271 /* File has SSA subtitles which prevent incremental cluster parsing. */
272 int contains_ssa;
David Conradb061d892007-06-04 22:10:54273} MatroskaDemuxContext;
274
Aurelien Jacobs209472b2008-08-05 00:41:05275typedef struct {
276 uint64_t duration;
277 int64_t reference;
Aurelien Jacobs194d4b42009-08-10 18:06:14278 uint64_t non_simple;
Aurelien Jacobs209472b2008-08-05 00:41:05279 EbmlBin bin;
280} MatroskaBlock;
281
Aurelien Jacobs63511322008-08-05 00:40:02282static EbmlSyntax ebml_header[] = {
Keiji Costantini84cfce92014-03-01 16:28:15283 { EBML_ID_EBMLREADVERSION, EBML_UINT, 0, offsetof(Ebml, version), { .u = EBML_VERSION } },
284 { EBML_ID_EBMLMAXSIZELENGTH, EBML_UINT, 0, offsetof(Ebml, max_size), { .u = 8 } },
285 { EBML_ID_EBMLMAXIDLENGTH, EBML_UINT, 0, offsetof(Ebml, id_length), { .u = 4 } },
286 { EBML_ID_DOCTYPE, EBML_STR, 0, offsetof(Ebml, doctype), { .s = "(none)" } },
287 { EBML_ID_DOCTYPEREADVERSION, EBML_UINT, 0, offsetof(Ebml, doctype_version), { .u = 1 } },
288 { EBML_ID_EBMLVERSION, EBML_NONE },
289 { EBML_ID_DOCTYPEVERSION, EBML_NONE },
Aurelien Jacobs63511322008-08-05 00:40:02290 { 0 }
291};
292
293static EbmlSyntax ebml_syntax[] = {
Keiji Costantini84cfce92014-03-01 16:28:15294 { EBML_ID_HEADER, EBML_NEST, 0, 0, { .n = ebml_header } },
Aurelien Jacobs63511322008-08-05 00:40:02295 { 0 }
296};
297
Aurelien Jacobs29708582008-08-05 00:40:27298static EbmlSyntax matroska_info[] = {
Keiji Costantini84cfce92014-03-01 16:28:15299 { MATROSKA_ID_TIMECODESCALE, EBML_UINT, 0, offsetof(MatroskaDemuxContext, time_scale), { .u = 1000000 } },
300 { MATROSKA_ID_DURATION, EBML_FLOAT, 0, offsetof(MatroskaDemuxContext, duration) },
301 { MATROSKA_ID_TITLE, EBML_UTF8, 0, offsetof(MatroskaDemuxContext, title) },
302 { MATROSKA_ID_WRITINGAPP, EBML_NONE },
303 { MATROSKA_ID_MUXINGAPP, EBML_NONE },
304 { MATROSKA_ID_DATEUTC, EBML_NONE },
305 { MATROSKA_ID_SEGMENTUID, EBML_NONE },
Aurelien Jacobs29708582008-08-05 00:40:27306 { 0 }
307};
308
Aurelien Jacobs2cbc8812008-08-05 00:40:31309static EbmlSyntax matroska_track_video[] = {
Keiji Costantini84cfce92014-03-01 16:28:15310 { MATROSKA_ID_VIDEOFRAMERATE, EBML_FLOAT, 0, offsetof(MatroskaTrackVideo, frame_rate) },
311 { MATROSKA_ID_VIDEODISPLAYWIDTH, EBML_UINT, 0, offsetof(MatroskaTrackVideo, display_width) },
312 { MATROSKA_ID_VIDEODISPLAYHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo, display_height) },
313 { MATROSKA_ID_VIDEOPIXELWIDTH, EBML_UINT, 0, offsetof(MatroskaTrackVideo, pixel_width) },
314 { MATROSKA_ID_VIDEOPIXELHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo, pixel_height) },
315 { MATROSKA_ID_VIDEOCOLORSPACE, EBML_UINT, 0, offsetof(MatroskaTrackVideo, fourcc) },
316 { MATROSKA_ID_VIDEOPIXELCROPB, EBML_NONE },
317 { MATROSKA_ID_VIDEOPIXELCROPT, EBML_NONE },
318 { MATROSKA_ID_VIDEOPIXELCROPL, EBML_NONE },
319 { MATROSKA_ID_VIDEOPIXELCROPR, EBML_NONE },
320 { MATROSKA_ID_VIDEODISPLAYUNIT, EBML_NONE },
321 { MATROSKA_ID_VIDEOFLAGINTERLACED, EBML_NONE },
322 { MATROSKA_ID_VIDEOSTEREOMODE, EBML_NONE },
323 { MATROSKA_ID_VIDEOASPECTRATIO, EBML_NONE },
Aurelien Jacobs2cbc8812008-08-05 00:40:31324 { 0 }
325};
326
327static EbmlSyntax matroska_track_audio[] = {
Keiji Costantini84cfce92014-03-01 16:28:15328 { MATROSKA_ID_AUDIOSAMPLINGFREQ, EBML_FLOAT, 0, offsetof(MatroskaTrackAudio, samplerate), { .f = 8000.0 } },
329 { MATROSKA_ID_AUDIOOUTSAMPLINGFREQ, EBML_FLOAT, 0, offsetof(MatroskaTrackAudio, out_samplerate) },
330 { MATROSKA_ID_AUDIOBITDEPTH, EBML_UINT, 0, offsetof(MatroskaTrackAudio, bitdepth) },
331 { MATROSKA_ID_AUDIOCHANNELS, EBML_UINT, 0, offsetof(MatroskaTrackAudio, channels), { .u = 1 } },
Aurelien Jacobs2cbc8812008-08-05 00:40:31332 { 0 }
333};
334
335static EbmlSyntax matroska_track_encoding_compression[] = {
Keiji Costantini84cfce92014-03-01 16:28:15336 { MATROSKA_ID_ENCODINGCOMPALGO, EBML_UINT, 0, offsetof(MatroskaTrackCompression, algo), { .u = 0 } },
337 { MATROSKA_ID_ENCODINGCOMPSETTINGS, EBML_BIN, 0, offsetof(MatroskaTrackCompression, settings) },
Aurelien Jacobs2cbc8812008-08-05 00:40:31338 { 0 }
339};
340
341static EbmlSyntax matroska_track_encoding[] = {
Keiji Costantini84cfce92014-03-01 16:28:15342 { MATROSKA_ID_ENCODINGSCOPE, EBML_UINT, 0, offsetof(MatroskaTrackEncoding, scope), { .u = 1 } },
343 { MATROSKA_ID_ENCODINGTYPE, EBML_UINT, 0, offsetof(MatroskaTrackEncoding, type), { .u = 0 } },
344 { MATROSKA_ID_ENCODINGCOMPRESSION, EBML_NEST, 0, offsetof(MatroskaTrackEncoding, compression), { .n = matroska_track_encoding_compression } },
345 { MATROSKA_ID_ENCODINGORDER, EBML_NONE },
Aurelien Jacobs2cbc8812008-08-05 00:40:31346 { 0 }
347};
348
349static EbmlSyntax matroska_track_encodings[] = {
Keiji Costantini84cfce92014-03-01 16:28:15350 { MATROSKA_ID_TRACKCONTENTENCODING, EBML_NEST, sizeof(MatroskaTrackEncoding), offsetof(MatroskaTrack, encodings), { .n = matroska_track_encoding } },
Aurelien Jacobs2cbc8812008-08-05 00:40:31351 { 0 }
352};
353
354static EbmlSyntax matroska_track[] = {
Keiji Costantini84cfce92014-03-01 16:28:15355 { MATROSKA_ID_TRACKNUMBER, EBML_UINT, 0, offsetof(MatroskaTrack, num) },
356 { MATROSKA_ID_TRACKNAME, EBML_UTF8, 0, offsetof(MatroskaTrack, name) },
357 { MATROSKA_ID_TRACKUID, EBML_UINT, 0, offsetof(MatroskaTrack, uid) },
358 { MATROSKA_ID_TRACKTYPE, EBML_UINT, 0, offsetof(MatroskaTrack, type) },
359 { MATROSKA_ID_CODECID, EBML_STR, 0, offsetof(MatroskaTrack, codec_id) },
360 { MATROSKA_ID_CODECPRIVATE, EBML_BIN, 0, offsetof(MatroskaTrack, codec_priv) },
Anton Khirnoveb3b5502014-04-29 10:03:13361 { MATROSKA_ID_CODECDELAY, EBML_UINT, 0, offsetof(MatroskaTrack, codec_delay) },
Keiji Costantini84cfce92014-03-01 16:28:15362 { MATROSKA_ID_TRACKLANGUAGE, EBML_UTF8, 0, offsetof(MatroskaTrack, language), { .s = "eng" } },
363 { MATROSKA_ID_TRACKDEFAULTDURATION, EBML_UINT, 0, offsetof(MatroskaTrack, default_duration) },
364 { MATROSKA_ID_TRACKTIMECODESCALE, EBML_FLOAT, 0, offsetof(MatroskaTrack, time_scale), { .f = 1.0 } },
365 { MATROSKA_ID_TRACKFLAGDEFAULT, EBML_UINT, 0, offsetof(MatroskaTrack, flag_default), { .u = 1 } },
366 { MATROSKA_ID_TRACKFLAGFORCED, EBML_UINT, 0, offsetof(MatroskaTrack, flag_forced), { .u = 0 } },
367 { MATROSKA_ID_TRACKVIDEO, EBML_NEST, 0, offsetof(MatroskaTrack, video), { .n = matroska_track_video } },
368 { MATROSKA_ID_TRACKAUDIO, EBML_NEST, 0, offsetof(MatroskaTrack, audio), { .n = matroska_track_audio } },
369 { MATROSKA_ID_TRACKCONTENTENCODINGS, EBML_NEST, 0, 0, { .n = matroska_track_encodings } },
370 { MATROSKA_ID_TRACKFLAGENABLED, EBML_NONE },
371 { MATROSKA_ID_TRACKFLAGLACING, EBML_NONE },
372 { MATROSKA_ID_CODECNAME, EBML_NONE },
373 { MATROSKA_ID_CODECDECODEALL, EBML_NONE },
374 { MATROSKA_ID_CODECINFOURL, EBML_NONE },
375 { MATROSKA_ID_CODECDOWNLOADURL, EBML_NONE },
376 { MATROSKA_ID_TRACKMINCACHE, EBML_NONE },
377 { MATROSKA_ID_TRACKMAXCACHE, EBML_NONE },
378 { MATROSKA_ID_TRACKMAXBLKADDID, EBML_NONE },
Aurelien Jacobs2cbc8812008-08-05 00:40:31379 { 0 }
380};
381
382static EbmlSyntax matroska_tracks[] = {
Keiji Costantini84cfce92014-03-01 16:28:15383 { MATROSKA_ID_TRACKENTRY, EBML_NEST, sizeof(MatroskaTrack), offsetof(MatroskaDemuxContext, tracks), { .n = matroska_track } },
Aurelien Jacobs2cbc8812008-08-05 00:40:31384 { 0 }
385};
386
Aurelien Jacobsb414cb82008-08-05 00:40:24387static EbmlSyntax matroska_attachment[] = {
Diego Biurrunf69befe2014-03-07 12:54:18388 { MATROSKA_ID_FILEUID, EBML_UINT, 0, offsetof(MatroskaAttachment, uid) },
389 { MATROSKA_ID_FILENAME, EBML_UTF8, 0, offsetof(MatroskaAttachment, filename) },
390 { MATROSKA_ID_FILEMIMETYPE, EBML_STR, 0, offsetof(MatroskaAttachment, mime) },
391 { MATROSKA_ID_FILEDATA, EBML_BIN, 0, offsetof(MatroskaAttachment, bin) },
Keiji Costantini84cfce92014-03-01 16:28:15392 { MATROSKA_ID_FILEDESC, EBML_NONE },
Aurelien Jacobsb414cb82008-08-05 00:40:24393 { 0 }
394};
395
396static EbmlSyntax matroska_attachments[] = {
Diego Biurrunf69befe2014-03-07 12:54:18397 { MATROSKA_ID_ATTACHEDFILE, EBML_NEST, sizeof(MatroskaAttachment), offsetof(MatroskaDemuxContext, attachments), { .n = matroska_attachment } },
Aurelien Jacobsb414cb82008-08-05 00:40:24398 { 0 }
399};
400
Aurelien Jacobs6bbd7c72008-08-05 00:40:21401static EbmlSyntax matroska_chapter_display[] = {
Keiji Costantini84cfce92014-03-01 16:28:15402 { MATROSKA_ID_CHAPSTRING, EBML_UTF8, 0, offsetof(MatroskaChapter, title) },
403 { MATROSKA_ID_CHAPLANG, EBML_NONE },
Aurelien Jacobs6bbd7c72008-08-05 00:40:21404 { 0 }
405};
406
407static EbmlSyntax matroska_chapter_entry[] = {
Keiji Costantini84cfce92014-03-01 16:28:15408 { MATROSKA_ID_CHAPTERTIMESTART, EBML_UINT, 0, offsetof(MatroskaChapter, start), { .u = AV_NOPTS_VALUE } },
409 { MATROSKA_ID_CHAPTERTIMEEND, EBML_UINT, 0, offsetof(MatroskaChapter, end), { .u = AV_NOPTS_VALUE } },
410 { MATROSKA_ID_CHAPTERUID, EBML_UINT, 0, offsetof(MatroskaChapter, uid) },
411 { MATROSKA_ID_CHAPTERDISPLAY, EBML_NEST, 0, 0, { .n = matroska_chapter_display } },
Aurelien Jacobs6bbd7c72008-08-05 00:40:21412 { MATROSKA_ID_CHAPTERFLAGHIDDEN, EBML_NONE },
Aurelien Jacobs5df3cc62008-08-13 21:15:15413 { MATROSKA_ID_CHAPTERFLAGENABLED, EBML_NONE },
414 { MATROSKA_ID_CHAPTERPHYSEQUIV, EBML_NONE },
415 { MATROSKA_ID_CHAPTERATOM, EBML_NONE },
Aurelien Jacobs6bbd7c72008-08-05 00:40:21416 { 0 }
417};
418
419static EbmlSyntax matroska_chapter[] = {
Keiji Costantini84cfce92014-03-01 16:28:15420 { MATROSKA_ID_CHAPTERATOM, EBML_NEST, sizeof(MatroskaChapter), offsetof(MatroskaDemuxContext, chapters), { .n = matroska_chapter_entry } },
Aurelien Jacobs6bbd7c72008-08-05 00:40:21421 { MATROSKA_ID_EDITIONUID, EBML_NONE },
422 { MATROSKA_ID_EDITIONFLAGHIDDEN, EBML_NONE },
423 { MATROSKA_ID_EDITIONFLAGDEFAULT, EBML_NONE },
Aurelien Jacobs5df3cc62008-08-13 21:15:15424 { MATROSKA_ID_EDITIONFLAGORDERED, EBML_NONE },
Aurelien Jacobs6bbd7c72008-08-05 00:40:21425 { 0 }
426};
427
428static EbmlSyntax matroska_chapters[] = {
Keiji Costantini84cfce92014-03-01 16:28:15429 { MATROSKA_ID_EDITIONENTRY, EBML_NEST, 0, 0, { .n = matroska_chapter } },
Aurelien Jacobs6bbd7c72008-08-05 00:40:21430 { 0 }
431};
432
Aurelien Jacobse5929fd2008-08-05 00:40:15433static EbmlSyntax matroska_index_pos[] = {
Keiji Costantini84cfce92014-03-01 16:28:15434 { MATROSKA_ID_CUETRACK, EBML_UINT, 0, offsetof(MatroskaIndexPos, track) },
435 { MATROSKA_ID_CUECLUSTERPOSITION, EBML_UINT, 0, offsetof(MatroskaIndexPos, pos) },
Aurelien Jacobs5df3cc62008-08-13 21:15:15436 { MATROSKA_ID_CUEBLOCKNUMBER, EBML_NONE },
Aurelien Jacobse5929fd2008-08-05 00:40:15437 { 0 }
438};
439
440static EbmlSyntax matroska_index_entry[] = {
Keiji Costantini84cfce92014-03-01 16:28:15441 { MATROSKA_ID_CUETIME, EBML_UINT, 0, offsetof(MatroskaIndex, time) },
442 { MATROSKA_ID_CUETRACKPOSITION, EBML_NEST, sizeof(MatroskaIndexPos), offsetof(MatroskaIndex, pos), { .n = matroska_index_pos } },
Aurelien Jacobse5929fd2008-08-05 00:40:15443 { 0 }
444};
445
446static EbmlSyntax matroska_index[] = {
Keiji Costantini84cfce92014-03-01 16:28:15447 { MATROSKA_ID_POINTENTRY, EBML_NEST, sizeof(MatroskaIndex), offsetof(MatroskaDemuxContext, index), { .n = matroska_index_entry } },
Aurelien Jacobse5929fd2008-08-05 00:40:15448 { 0 }
449};
450
Aurelien Jacobs44015c52008-08-08 23:50:38451static EbmlSyntax matroska_simpletag[] = {
Keiji Costantini84cfce92014-03-01 16:28:15452 { MATROSKA_ID_TAGNAME, EBML_UTF8, 0, offsetof(MatroskaTag, name) },
453 { MATROSKA_ID_TAGSTRING, EBML_UTF8, 0, offsetof(MatroskaTag, string) },
454 { MATROSKA_ID_TAGLANG, EBML_STR, 0, offsetof(MatroskaTag, lang), { .s = "und" } },
455 { MATROSKA_ID_TAGDEFAULT, EBML_UINT, 0, offsetof(MatroskaTag, def) },
456 { MATROSKA_ID_TAGDEFAULT_BUG, EBML_UINT, 0, offsetof(MatroskaTag, def) },
457 { MATROSKA_ID_SIMPLETAG, EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTag, sub), { .n = matroska_simpletag } },
Aurelien Jacobs44015c52008-08-08 23:50:38458 { 0 }
459};
460
Aurelien Jacobs929e9de2009-02-15 15:53:55461static EbmlSyntax matroska_tagtargets[] = {
Keiji Costantini84cfce92014-03-01 16:28:15462 { MATROSKA_ID_TAGTARGETS_TYPE, EBML_STR, 0, offsetof(MatroskaTagTarget, type) },
463 { MATROSKA_ID_TAGTARGETS_TYPEVALUE, EBML_UINT, 0, offsetof(MatroskaTagTarget, typevalue), { .u = 50 } },
464 { MATROSKA_ID_TAGTARGETS_TRACKUID, EBML_UINT, 0, offsetof(MatroskaTagTarget, trackuid) },
465 { MATROSKA_ID_TAGTARGETS_CHAPTERUID, EBML_UINT, 0, offsetof(MatroskaTagTarget, chapteruid) },
466 { MATROSKA_ID_TAGTARGETS_ATTACHUID, EBML_UINT, 0, offsetof(MatroskaTagTarget, attachuid) },
Aurelien Jacobs929e9de2009-02-15 15:53:55467 { 0 }
468};
469
Aurelien Jacobs44015c52008-08-08 23:50:38470static EbmlSyntax matroska_tag[] = {
Keiji Costantini84cfce92014-03-01 16:28:15471 { MATROSKA_ID_SIMPLETAG, EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTags, tag), { .n = matroska_simpletag } },
472 { MATROSKA_ID_TAGTARGETS, EBML_NEST, 0, offsetof(MatroskaTags, target), { .n = matroska_tagtargets } },
Aurelien Jacobs44015c52008-08-08 23:50:38473 { 0 }
474};
475
Aurelien Jacobs434d4962008-08-05 00:40:18476static EbmlSyntax matroska_tags[] = {
Keiji Costantini84cfce92014-03-01 16:28:15477 { MATROSKA_ID_TAG, EBML_NEST, sizeof(MatroskaTags), offsetof(MatroskaDemuxContext, tags), { .n = matroska_tag } },
Aurelien Jacobs434d4962008-08-05 00:40:18478 { 0 }
479};
480
Aurelien Jacobs13b350a2008-08-05 00:40:36481static EbmlSyntax matroska_seekhead_entry[] = {
Keiji Costantini84cfce92014-03-01 16:28:15482 { MATROSKA_ID_SEEKID, EBML_UINT, 0, offsetof(MatroskaSeekhead, id) },
483 { MATROSKA_ID_SEEKPOSITION, EBML_UINT, 0, offsetof(MatroskaSeekhead, pos), { .u = -1 } },
Aurelien Jacobs13b350a2008-08-05 00:40:36484 { 0 }
485};
486
487static EbmlSyntax matroska_seekhead[] = {
Keiji Costantini84cfce92014-03-01 16:28:15488 { MATROSKA_ID_SEEKENTRY, EBML_NEST, sizeof(MatroskaSeekhead), offsetof(MatroskaDemuxContext, seekhead), { .n = matroska_seekhead_entry } },
Aurelien Jacobs13b350a2008-08-05 00:40:36489 { 0 }
490};
491
Aurelien Jacobsce6f28b2008-08-05 00:40:58492static EbmlSyntax matroska_segment[] = {
Keiji Costantini84cfce92014-03-01 16:28:15493 { MATROSKA_ID_INFO, EBML_NEST, 0, 0, { .n = matroska_info } },
494 { MATROSKA_ID_TRACKS, EBML_NEST, 0, 0, { .n = matroska_tracks } },
495 { MATROSKA_ID_ATTACHMENTS, EBML_NEST, 0, 0, { .n = matroska_attachments } },
496 { MATROSKA_ID_CHAPTERS, EBML_NEST, 0, 0, { .n = matroska_chapters } },
497 { MATROSKA_ID_CUES, EBML_NEST, 0, 0, { .n = matroska_index } },
498 { MATROSKA_ID_TAGS, EBML_NEST, 0, 0, { .n = matroska_tags } },
499 { MATROSKA_ID_SEEKHEAD, EBML_NEST, 0, 0, { .n = matroska_seekhead } },
500 { MATROSKA_ID_CLUSTER, EBML_STOP },
Aurelien Jacobsce6f28b2008-08-05 00:40:58501 { 0 }
502};
503
504static EbmlSyntax matroska_segments[] = {
Keiji Costantini84cfce92014-03-01 16:28:15505 { MATROSKA_ID_SEGMENT, EBML_NEST, 0, 0, { .n = matroska_segment } },
Aurelien Jacobsce6f28b2008-08-05 00:40:58506 { 0 }
507};
508
Aurelien Jacobs209472b2008-08-05 00:41:05509static EbmlSyntax matroska_blockgroup[] = {
Keiji Costantini84cfce92014-03-01 16:28:15510 { MATROSKA_ID_BLOCK, EBML_BIN, 0, offsetof(MatroskaBlock, bin) },
511 { MATROSKA_ID_SIMPLEBLOCK, EBML_BIN, 0, offsetof(MatroskaBlock, bin) },
512 { MATROSKA_ID_BLOCKDURATION, EBML_UINT, 0, offsetof(MatroskaBlock, duration), { .u = AV_NOPTS_VALUE } },
513 { MATROSKA_ID_BLOCKREFERENCE, EBML_UINT, 0, offsetof(MatroskaBlock, reference) },
Anton Khirnov564b7e02013-05-15 13:48:15514 { MATROSKA_ID_CODECSTATE, EBML_NONE },
Keiji Costantini84cfce92014-03-01 16:28:15515 { 1, EBML_UINT, 0, offsetof(MatroskaBlock, non_simple), { .u = 1 } },
Aurelien Jacobs209472b2008-08-05 00:41:05516 { 0 }
517};
518
519static EbmlSyntax matroska_cluster[] = {
Keiji Costantini84cfce92014-03-01 16:28:15520 { MATROSKA_ID_CLUSTERTIMECODE, EBML_UINT, 0, offsetof(MatroskaCluster, timecode) },
521 { MATROSKA_ID_BLOCKGROUP, EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster, blocks), { .n = matroska_blockgroup } },
522 { MATROSKA_ID_SIMPLEBLOCK, EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster, blocks), { .n = matroska_blockgroup } },
523 { MATROSKA_ID_CLUSTERPOSITION, EBML_NONE },
524 { MATROSKA_ID_CLUSTERPREVSIZE, EBML_NONE },
Aurelien Jacobs209472b2008-08-05 00:41:05525 { 0 }
526};
527
528static EbmlSyntax matroska_clusters[] = {
Keiji Costantini84cfce92014-03-01 16:28:15529 { MATROSKA_ID_CLUSTER, EBML_NEST, 0, 0, { .n = matroska_cluster } },
530 { MATROSKA_ID_INFO, EBML_NONE },
531 { MATROSKA_ID_CUES, EBML_NONE },
532 { MATROSKA_ID_TAGS, EBML_NONE },
533 { MATROSKA_ID_SEEKHEAD, EBML_NONE },
Aurelien Jacobs209472b2008-08-05 00:41:05534 { 0 }
535};
536
Dale Curtis8336eb62012-04-19 18:12:24537static EbmlSyntax matroska_cluster_incremental_parsing[] = {
Keiji Costantini84cfce92014-03-01 16:28:15538 { MATROSKA_ID_CLUSTERTIMECODE, EBML_UINT, 0, offsetof(MatroskaCluster, timecode) },
539 { MATROSKA_ID_BLOCKGROUP, EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster, blocks), { .n = matroska_blockgroup } },
540 { MATROSKA_ID_SIMPLEBLOCK, EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster, blocks), { .n = matroska_blockgroup } },
541 { MATROSKA_ID_CLUSTERPOSITION, EBML_NONE },
542 { MATROSKA_ID_CLUSTERPREVSIZE, EBML_NONE },
543 { MATROSKA_ID_INFO, EBML_NONE },
544 { MATROSKA_ID_CUES, EBML_NONE },
545 { MATROSKA_ID_TAGS, EBML_NONE },
546 { MATROSKA_ID_SEEKHEAD, EBML_NONE },
547 { MATROSKA_ID_CLUSTER, EBML_STOP },
Dale Curtis8336eb62012-04-19 18:12:24548 { 0 }
549};
550
551static EbmlSyntax matroska_cluster_incremental[] = {
Keiji Costantini84cfce92014-03-01 16:28:15552 { MATROSKA_ID_CLUSTERTIMECODE, EBML_UINT, 0, offsetof(MatroskaCluster, timecode) },
553 { MATROSKA_ID_BLOCKGROUP, EBML_STOP },
554 { MATROSKA_ID_SIMPLEBLOCK, EBML_STOP },
555 { MATROSKA_ID_CLUSTERPOSITION, EBML_NONE },
556 { MATROSKA_ID_CLUSTERPREVSIZE, EBML_NONE },
Dale Curtis8336eb62012-04-19 18:12:24557 { 0 }
558};
559
560static EbmlSyntax matroska_clusters_incremental[] = {
Keiji Costantini84cfce92014-03-01 16:28:15561 { MATROSKA_ID_CLUSTER, EBML_NEST, 0, 0, { .n = matroska_cluster_incremental } },
562 { MATROSKA_ID_INFO, EBML_NONE },
563 { MATROSKA_ID_CUES, EBML_NONE },
564 { MATROSKA_ID_TAGS, EBML_NONE },
565 { MATROSKA_ID_SEEKHEAD, EBML_NONE },
Dale Curtis8336eb62012-04-19 18:12:24566 { 0 }
567};
568
Alex Converseb0f29db2012-02-20 08:42:33569static const char *const matroska_doctypes[] = { "matroska", "webm" };
James Zern470491f2010-05-22 01:41:32570
Sean McGovern8835c552013-05-27 22:11:50571static int matroska_resync(MatroskaDemuxContext *matroska, int64_t last_pos)
572{
573 AVIOContext *pb = matroska->ctx->pb;
574 uint32_t id;
575 matroska->current_id = 0;
576 matroska->num_levels = 0;
577
578 /* seek to next position to resync from */
579 if (avio_seek(pb, last_pos + 1, SEEK_SET) < 0)
580 goto eof;
581
582 id = avio_rb32(pb);
583
584 // try to find a toplevel element
585 while (!pb->eof_reached) {
586 if (id == MATROSKA_ID_INFO || id == MATROSKA_ID_TRACKS ||
587 id == MATROSKA_ID_CUES || id == MATROSKA_ID_TAGS ||
588 id == MATROSKA_ID_SEEKHEAD || id == MATROSKA_ID_ATTACHMENTS ||
589 id == MATROSKA_ID_CLUSTER || id == MATROSKA_ID_CHAPTERS) {
Keiji Costantini84cfce92014-03-01 16:28:15590 matroska->current_id = id;
591 return 0;
Sean McGovern8835c552013-05-27 22:11:50592 }
593 id = (id << 8) | avio_r8(pb);
594 }
Keiji Costantini84cfce92014-03-01 16:28:15595
Sean McGovern8835c552013-05-27 22:11:50596eof:
597 matroska->done = 1;
598 return AVERROR_EOF;
599}
600
David Conradb061d892007-06-04 22:10:54601/*
Diego Biurrun5968d2d2008-08-05 08:28:57602 * Return: Whether we reached the end of a level in the hierarchy or not.
David Conradb061d892007-06-04 22:10:54603 */
Aurelien Jacobs592110c2008-08-05 00:42:08604static int ebml_level_end(MatroskaDemuxContext *matroska)
David Conradb061d892007-06-04 22:10:54605{
Anton Khirnovae628ec2011-02-20 10:04:12606 AVIOContext *pb = matroska->ctx->pb;
Anton Khirnova2704c92011-03-03 19:11:45607 int64_t pos = avio_tell(pb);
David Conradb061d892007-06-04 22:10:54608
Aurelien Jacobs592110c2008-08-05 00:42:08609 if (matroska->num_levels > 0) {
David Conradb061d892007-06-04 22:10:54610 MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
Aurelien Jacobs8dbe48f2010-06-11 16:43:47611 if (pos - level->start >= level->length || matroska->current_id) {
David Conradb061d892007-06-04 22:10:54612 matroska->num_levels--;
Aurelien Jacobs592110c2008-08-05 00:42:08613 return 1;
David Conradb061d892007-06-04 22:10:54614 }
615 }
Aurelien Jacobs592110c2008-08-05 00:42:08616 return 0;
David Conradb061d892007-06-04 22:10:54617}
618
619/*
620 * Read: an "EBML number", which is defined as a variable-length
621 * array of bytes. The first byte indicates the length by giving a
622 * number of 0-bits followed by a one. The position of the first
623 * "one" bit inside the first byte indicates the length of this
624 * number.
Diego Biurrun5968d2d2008-08-05 08:28:57625 * Returns: number of bytes read, < 0 on error
David Conradb061d892007-06-04 22:10:54626 */
Anton Khirnovae628ec2011-02-20 10:04:12627static int ebml_read_num(MatroskaDemuxContext *matroska, AVIOContext *pb,
Aurelien Jacobsf7b96872008-08-05 00:42:05628 int max_size, uint64_t *number)
David Conradb061d892007-06-04 22:10:54629{
Reimar Döffingerff6a5fc2010-09-02 19:17:46630 int read = 1, n = 1;
631 uint64_t total = 0;
David Conradb061d892007-06-04 22:10:54632
Anton Khirnovb7effd42011-02-21 15:43:01633 /* The first byte tells us the length in bytes - avio_r8() can normally
David Conradb061d892007-06-04 22:10:54634 * return 0, but since that's not a valid first ebmlID byte, we can
635 * use it safely here to catch EOS. */
Anton Khirnovb7effd42011-02-21 15:43:01636 if (!(total = avio_r8(pb))) {
David Conradb061d892007-06-04 22:10:54637 /* we might encounter EOS here */
Anton Khirnov66e5b1d2011-03-07 20:50:25638 if (!pb->eof_reached) {
Anton Khirnova2704c92011-03-03 19:11:45639 int64_t pos = avio_tell(pb);
David Conradb061d892007-06-04 22:10:54640 av_log(matroska->ctx, AV_LOG_ERROR,
641 "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
642 pos, pos);
Anton Khirnov721113b2012-07-21 08:48:39643 return pb->error ? pb->error : AVERROR(EIO);
David Conradb061d892007-06-04 22:10:54644 }
Anton Khirnov721113b2012-07-21 08:48:39645 return AVERROR_EOF;
David Conradb061d892007-06-04 22:10:54646 }
647
648 /* get the length of the EBML number */
Reimar Döffingerff6a5fc2010-09-02 19:17:46649 read = 8 - ff_log2_tab[total];
David Conradb061d892007-06-04 22:10:54650 if (read > max_size) {
Anton Khirnova2704c92011-03-03 19:11:45651 int64_t pos = avio_tell(pb) - 1;
David Conradb061d892007-06-04 22:10:54652 av_log(matroska->ctx, AV_LOG_ERROR,
653 "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
654 (uint8_t) total, pos, pos);
655 return AVERROR_INVALIDDATA;
656 }
657
658 /* read out length */
Reimar Döffingerff6a5fc2010-09-02 19:17:46659 total ^= 1 << ff_log2_tab[total];
David Conradb061d892007-06-04 22:10:54660 while (n++ < read)
Anton Khirnovb7effd42011-02-21 15:43:01661 total = (total << 8) | avio_r8(pb);
David Conradb061d892007-06-04 22:10:54662
663 *number = total;
664
665 return read;
666}
667
Reimar Döffinger1b4d3272010-09-06 17:51:44668/**
669 * Read a EBML length value.
670 * This needs special handling for the "unknown length" case which has multiple
671 * encodings.
672 */
Anton Khirnovae628ec2011-02-20 10:04:12673static int ebml_read_length(MatroskaDemuxContext *matroska, AVIOContext *pb,
Reimar Döffinger1b4d3272010-09-06 17:51:44674 uint64_t *number)
675{
676 int res = ebml_read_num(matroska, pb, 8, number);
677 if (res > 0 && *number + 1 == 1ULL << (7 * res))
678 *number = 0xffffffffffffffULL;
679 return res;
680}
681
David Conradb061d892007-06-04 22:10:54682/*
David Conradb061d892007-06-04 22:10:54683 * Read the next element as an unsigned int.
684 * 0 is success, < 0 is failure.
685 */
Anton Khirnovae628ec2011-02-20 10:04:12686static int ebml_read_uint(AVIOContext *pb, int size, uint64_t *num)
David Conradb061d892007-06-04 22:10:54687{
Aurelien Jacobsc6cd2b32008-08-05 00:41:55688 int n = 0;
David Conradb061d892007-06-04 22:10:54689
Aurelien Jacobs4a194c82010-09-05 21:37:40690 if (size > 8)
David Conradb061d892007-06-04 22:10:54691 return AVERROR_INVALIDDATA;
David Conradb061d892007-06-04 22:10:54692
Diego Biurrun5968d2d2008-08-05 08:28:57693 /* big-endian ordering; build up number */
David Conradb061d892007-06-04 22:10:54694 *num = 0;
695 while (n++ < size)
Anton Khirnovb7effd42011-02-21 15:43:01696 *num = (*num << 8) | avio_r8(pb);
David Conradb061d892007-06-04 22:10:54697
698 return 0;
699}
700
701/*
David Conradb061d892007-06-04 22:10:54702 * Read the next element as a float.
703 * 0 is success, < 0 is failure.
704 */
Anton Khirnovae628ec2011-02-20 10:04:12705static int ebml_read_float(AVIOContext *pb, int size, double *num)
David Conradb061d892007-06-04 22:10:54706{
Keiji Costantini84cfce92014-03-01 16:28:15707 if (size == 0)
Aurelien Jacobs4a194c82010-09-05 21:37:40708 *num = 0;
Keiji Costantini84cfce92014-03-01 16:28:15709 else if (size == 4)
Mans Rullgard3383a532011-11-27 14:04:16710 *num = av_int2float(avio_rb32(pb));
Keiji Costantini84cfce92014-03-01 16:28:15711 else if (size == 8)
Mans Rullgard3383a532011-11-27 14:04:16712 *num = av_int2double(avio_rb64(pb));
Keiji Costantini84cfce92014-03-01 16:28:15713 else
David Conradb061d892007-06-04 22:10:54714 return AVERROR_INVALIDDATA;
David Conradb061d892007-06-04 22:10:54715
716 return 0;
717}
718
719/*
720 * Read the next element as an ASCII string.
721 * 0 is success, < 0 is failure.
722 */
Anton Khirnovae628ec2011-02-20 10:04:12723static int ebml_read_ascii(AVIOContext *pb, int size, char **str)
David Conradb061d892007-06-04 22:10:54724{
Ronald S. Bultjecd40c312012-02-25 00:12:18725 char *res;
726
Diego Biurrun5968d2d2008-08-05 08:28:57727 /* EBML strings are usually not 0-terminated, so we allocate one
David Conradb061d892007-06-04 22:10:54728 * byte more, read the string and NULL-terminate it ourselves. */
Ronald S. Bultjecd40c312012-02-25 00:12:18729 if (!(res = av_malloc(size + 1)))
Panagiotis Issaris769e10f2007-07-19 15:21:30730 return AVERROR(ENOMEM);
Ronald S. Bultjecd40c312012-02-25 00:12:18731 if (avio_read(pb, (uint8_t *) res, size) != size) {
732 av_free(res);
Panagiotis Issaris6f3e0b22007-07-19 15:23:32733 return AVERROR(EIO);
David Conradb061d892007-06-04 22:10:54734 }
Ronald S. Bultjecd40c312012-02-25 00:12:18735 (res)[size] = '\0';
736 av_free(*str);
737 *str = res;
David Conradb061d892007-06-04 22:10:54738
739 return 0;
740}
741
742/*
Aurelien Jacobs737c40d2008-08-05 00:42:39743 * Read the next element as binary data.
744 * 0 is success, < 0 is failure.
745 */
Anton Khirnovae628ec2011-02-20 10:04:12746static int ebml_read_binary(AVIOContext *pb, int length, EbmlBin *bin)
Aurelien Jacobs737c40d2008-08-05 00:42:39747{
748 av_free(bin->data);
Diego Biurrun73953df2014-06-19 22:04:36749 if (!(bin->data = av_mallocz(length + FF_INPUT_BUFFER_PADDING_SIZE)))
Aurelien Jacobs737c40d2008-08-05 00:42:39750 return AVERROR(ENOMEM);
751
Aurelien Jacobs737c40d2008-08-05 00:42:39752 bin->size = length;
Anton Khirnova2704c92011-03-03 19:11:45753 bin->pos = avio_tell(pb);
Anton Khirnovb7effd42011-02-21 15:43:01754 if (avio_read(pb, bin->data, length) != length) {
David Conrad5549aa62010-05-18 21:21:37755 av_freep(&bin->data);
Aurelien Jacobs737c40d2008-08-05 00:42:39756 return AVERROR(EIO);
David Conrad5549aa62010-05-18 21:21:37757 }
Aurelien Jacobs737c40d2008-08-05 00:42:39758
759 return 0;
760}
761
762/*
David Conradb061d892007-06-04 22:10:54763 * Read the next element, but only the header. The contents
764 * are supposed to be sub-elements which can be read separately.
765 * 0 is success, < 0 is failure.
766 */
Aurelien Jacobsbddd1d92010-06-11 17:16:43767static int ebml_read_master(MatroskaDemuxContext *matroska, uint64_t length)
David Conradb061d892007-06-04 22:10:54768{
Anton Khirnovae628ec2011-02-20 10:04:12769 AVIOContext *pb = matroska->ctx->pb;
David Conradb061d892007-06-04 22:10:54770 MatroskaLevel *level;
David Conradb061d892007-06-04 22:10:54771
David Conradb061d892007-06-04 22:10:54772 if (matroska->num_levels >= EBML_MAX_DEPTH) {
773 av_log(matroska->ctx, AV_LOG_ERROR,
774 "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
Panagiotis Issaris85565db2007-07-19 15:38:33775 return AVERROR(ENOSYS);
David Conradb061d892007-06-04 22:10:54776 }
777
Keiji Costantini84cfce92014-03-01 16:28:15778 level = &matroska->levels[matroska->num_levels++];
779 level->start = avio_tell(pb);
David Conradb061d892007-06-04 22:10:54780 level->length = length;
781
782 return 0;
783}
784
785/*
David Conradb061d892007-06-04 22:10:54786 * Read signed/unsigned "EBML" numbers.
Diego Biurrun5968d2d2008-08-05 08:28:57787 * Return: number of bytes processed, < 0 on error
David Conradb061d892007-06-04 22:10:54788 */
Aurelien Jacobsc1e01132008-08-05 00:42:52789static int matroska_ebmlnum_uint(MatroskaDemuxContext *matroska,
790 uint8_t *data, uint32_t size, uint64_t *num)
David Conradb061d892007-06-04 22:10:54791{
Anton Khirnovae628ec2011-02-20 10:04:12792 AVIOContext pb;
Anton Khirnove731b8d2011-02-20 10:04:13793 ffio_init_context(&pb, data, size, 0, NULL, NULL, NULL, NULL);
David Conrad465c28b2010-05-18 21:21:32794 return ebml_read_num(matroska, &pb, FFMIN(size, 8), num);
David Conradb061d892007-06-04 22:10:54795}
796
797/*
798 * Same as above, but signed.
799 */
Aurelien Jacobsc1e01132008-08-05 00:42:52800static int matroska_ebmlnum_sint(MatroskaDemuxContext *matroska,
801 uint8_t *data, uint32_t size, int64_t *num)
David Conradb061d892007-06-04 22:10:54802{
803 uint64_t unum;
804 int res;
805
806 /* read as unsigned number first */
Aurelien Jacobsc1e01132008-08-05 00:42:52807 if ((res = matroska_ebmlnum_uint(matroska, data, size, &unum)) < 0)
David Conradb061d892007-06-04 22:10:54808 return res;
809
810 /* make signed (weird way) */
Keiji Costantini84cfce92014-03-01 16:28:15811 *num = unum - ((1LL << (7 * res - 1)) - 1);
David Conradb061d892007-06-04 22:10:54812
813 return res;
814}
815
Aurelien Jacobs789ed102008-08-05 00:40:00816static int ebml_parse_elem(MatroskaDemuxContext *matroska,
Aurelien Jacobs737c40d2008-08-05 00:42:39817 EbmlSyntax *syntax, void *data);
Aurelien Jacobs789ed102008-08-05 00:40:00818
819static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
820 uint32_t id, void *data)
821{
822 int i;
Keiji Costantini84cfce92014-03-01 16:28:15823 for (i = 0; syntax[i].id; i++)
Aurelien Jacobs789ed102008-08-05 00:40:00824 if (id == syntax[i].id)
825 break;
Aurelien Jacobs8dbe48f2010-06-11 16:43:47826 if (!syntax[i].id && id == MATROSKA_ID_CLUSTER &&
Keiji Costantini84cfce92014-03-01 16:28:15827 matroska->num_levels > 0 &&
828 matroska->levels[matroska->num_levels - 1].length == 0xffffffffffffff)
Aurelien Jacobs8dbe48f2010-06-11 16:43:47829 return 0; // we reached the end of an unknown size cluster
Anton Khirnov5b7a88f2012-07-07 15:15:27830 if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32) {
Diego Biurrund92024f2014-03-10 14:35:59831 av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%"PRIX32"\n", id);
Anton Khirnov5b7a88f2012-07-07 15:15:27832 if (matroska->ctx->error_recognition & AV_EF_EXPLODE)
833 return AVERROR_INVALIDDATA;
834 }
Aurelien Jacobs789ed102008-08-05 00:40:00835 return ebml_parse_elem(matroska, &syntax[i], data);
836}
837
Aurelien Jacobs66a37e02008-08-05 00:42:17838static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
839 void *data)
840{
Aurelien Jacobsc3ade622010-06-11 16:34:01841 if (!matroska->current_id) {
Aurelien Jacobs73917812010-06-11 16:45:38842 uint64_t id;
843 int res = ebml_read_num(matroska, matroska->ctx->pb, 4, &id);
844 if (res < 0)
845 return res;
Keiji Costantini84cfce92014-03-01 16:28:15846 matroska->current_id = id | 1 << 7 * res;
Aurelien Jacobsc3ade622010-06-11 16:34:01847 }
848 return ebml_parse_id(matroska, syntax, matroska->current_id, data);
Aurelien Jacobs66a37e02008-08-05 00:42:17849}
850
Aurelien Jacobs9bcb92c2008-08-05 00:42:13851static int ebml_parse_nest(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
Aurelien Jacobs6314cca2008-08-05 00:42:23852 void *data)
Aurelien Jacobs789ed102008-08-05 00:40:00853{
Aurelien Jacobsc005b3f2008-08-05 00:42:10854 int i, res = 0;
Aurelien Jacobs789ed102008-08-05 00:40:00855
Keiji Costantini84cfce92014-03-01 16:28:15856 for (i = 0; syntax[i].id; i++)
Aurelien Jacobs789ed102008-08-05 00:40:00857 switch (syntax[i].type) {
858 case EBML_UINT:
Keiji Costantini84cfce92014-03-01 16:28:15859 *(uint64_t *) ((char *) data + syntax[i].data_offset) = syntax[i].def.u;
Aurelien Jacobs789ed102008-08-05 00:40:00860 break;
861 case EBML_FLOAT:
Keiji Costantini84cfce92014-03-01 16:28:15862 *(double *) ((char *) data + syntax[i].data_offset) = syntax[i].def.f;
Aurelien Jacobs789ed102008-08-05 00:40:00863 break;
864 case EBML_STR:
865 case EBML_UTF8:
Anton Khirnov668643b2013-09-04 06:55:17866 // the default may be NULL
867 if (syntax[i].def.s) {
Keiji Costantini84cfce92014-03-01 16:28:15868 uint8_t **dst = (uint8_t **) ((uint8_t *) data + syntax[i].data_offset);
Anton Khirnov668643b2013-09-04 06:55:17869 *dst = av_strdup(syntax[i].def.s);
870 if (!*dst)
871 return AVERROR(ENOMEM);
872 }
Aurelien Jacobs789ed102008-08-05 00:40:00873 break;
874 }
875
Aurelien Jacobs6314cca2008-08-05 00:42:23876 while (!res && !ebml_level_end(matroska))
Aurelien Jacobs66a37e02008-08-05 00:42:17877 res = ebml_parse(matroska, syntax, data);
Aurelien Jacobs789ed102008-08-05 00:40:00878
879 return res;
880}
881
Aurelien Jacobs737c40d2008-08-05 00:42:39882static int ebml_parse_elem(MatroskaDemuxContext *matroska,
883 EbmlSyntax *syntax, void *data)
884{
Reimar Döffinger95ec3d42011-02-06 10:32:03885 static const uint64_t max_lengths[EBML_TYPE_COUNT] = {
886 [EBML_UINT] = 8,
887 [EBML_FLOAT] = 8,
888 // max. 16 MB for strings
889 [EBML_STR] = 0x1000000,
890 [EBML_UTF8] = 0x1000000,
891 // max. 256 MB for binary data
892 [EBML_BIN] = 0x10000000,
893 // no limits for anything else
894 };
Anton Khirnovae628ec2011-02-20 10:04:12895 AVIOContext *pb = matroska->ctx->pb;
Aurelien Jacobs737c40d2008-08-05 00:42:39896 uint32_t id = syntax->id;
897 uint64_t length;
898 int res;
Aurelien Jacobs737c40d2008-08-05 00:42:39899
Keiji Costantini84cfce92014-03-01 16:28:15900 data = (char *) data + syntax->data_offset;
Aurelien Jacobs737c40d2008-08-05 00:42:39901 if (syntax->list_elem_size) {
902 EbmlList *list = data;
Alexandra Khirnovaf369b932013-09-10 09:57:35903 if ((res = av_reallocp_array(&list->elem,
904 list->nb_elem + 1,
905 syntax->list_elem_size)) < 0) {
906 list->nb_elem = 0;
907 return res;
908 }
Keiji Costantini84cfce92014-03-01 16:28:15909 data = (char *) list->elem + list->nb_elem * syntax->list_elem_size;
Aurelien Jacobs737c40d2008-08-05 00:42:39910 memset(data, 0, syntax->list_elem_size);
911 list->nb_elem++;
912 }
913
Aurelien Jacobsc3ade622010-06-11 16:34:01914 if (syntax->type != EBML_PASS && syntax->type != EBML_STOP) {
915 matroska->current_id = 0;
Reimar Döffinger1b4d3272010-09-06 17:51:44916 if ((res = ebml_read_length(matroska, pb, &length)) < 0)
Aurelien Jacobs737c40d2008-08-05 00:42:39917 return res;
Reimar Döffinger95ec3d42011-02-06 10:32:03918 if (max_lengths[syntax->type] && length > max_lengths[syntax->type]) {
919 av_log(matroska->ctx, AV_LOG_ERROR,
920 "Invalid length 0x%"PRIx64" > 0x%"PRIx64" for syntax element %i\n",
921 length, max_lengths[syntax->type], syntax->type);
922 return AVERROR_INVALIDDATA;
923 }
Aurelien Jacobsc3ade622010-06-11 16:34:01924 }
Aurelien Jacobs737c40d2008-08-05 00:42:39925
926 switch (syntax->type) {
Keiji Costantini84cfce92014-03-01 16:28:15927 case EBML_UINT:
928 res = ebml_read_uint(pb, length, data);
929 break;
930 case EBML_FLOAT:
931 res = ebml_read_float(pb, length, data);
932 break;
Aurelien Jacobs737c40d2008-08-05 00:42:39933 case EBML_STR:
Keiji Costantini84cfce92014-03-01 16:28:15934 case EBML_UTF8:
935 res = ebml_read_ascii(pb, length, data);
936 break;
937 case EBML_BIN:
938 res = ebml_read_binary(pb, length, data);
939 break;
940 case EBML_NEST:
941 if ((res = ebml_read_master(matroska, length)) < 0)
942 return res;
943 if (id == MATROSKA_ID_SEGMENT)
944 matroska->segment_start = avio_tell(matroska->ctx->pb);
945 return ebml_parse_nest(matroska, syntax->def.n, data);
946 case EBML_PASS:
947 return ebml_parse_id(matroska, syntax->def.n, id, data);
948 case EBML_STOP:
949 return 1;
950 default:
951 return avio_skip(pb, length) < 0 ? AVERROR(EIO) : 0;
Aurelien Jacobs737c40d2008-08-05 00:42:39952 }
953 if (res == AVERROR_INVALIDDATA)
954 av_log(matroska->ctx, AV_LOG_ERROR, "Invalid element\n");
955 else if (res == AVERROR(EIO))
956 av_log(matroska->ctx, AV_LOG_ERROR, "Read error\n");
957 return res;
958}
959
Aurelien Jacobs789ed102008-08-05 00:40:00960static void ebml_free(EbmlSyntax *syntax, void *data)
961{
962 int i, j;
Keiji Costantini84cfce92014-03-01 16:28:15963 for (i = 0; syntax[i].id; i++) {
964 void *data_off = (char *) data + syntax[i].data_offset;
Aurelien Jacobs789ed102008-08-05 00:40:00965 switch (syntax[i].type) {
966 case EBML_STR:
Keiji Costantini84cfce92014-03-01 16:28:15967 case EBML_UTF8:
968 av_freep(data_off);
969 break;
970 case EBML_BIN:
971 av_freep(&((EbmlBin *) data_off)->data);
972 break;
Aurelien Jacobs789ed102008-08-05 00:40:00973 case EBML_NEST:
974 if (syntax[i].list_elem_size) {
975 EbmlList *list = data_off;
976 char *ptr = list->elem;
Keiji Costantini84cfce92014-03-01 16:28:15977 for (j = 0; j < list->nb_elem;
978 j++, ptr += syntax[i].list_elem_size)
Aurelien Jacobs789ed102008-08-05 00:40:00979 ebml_free(syntax[i].def.n, ptr);
980 av_free(list->elem);
981 } else
982 ebml_free(syntax[i].def.n, data_off);
Keiji Costantini84cfce92014-03-01 16:28:15983 default:
984 break;
Aurelien Jacobs789ed102008-08-05 00:40:00985 }
986 }
987}
988
Aurelien Jacobs737c40d2008-08-05 00:42:39989/*
990 * Autodetecting...
991 */
992static int matroska_probe(AVProbeData *p)
993{
994 uint64_t total = 0;
James Zern470491f2010-05-22 01:41:32995 int len_mask = 0x80, size = 1, n = 1, i;
Aurelien Jacobs737c40d2008-08-05 00:42:39996
Diego Biurrun5968d2d2008-08-05 08:28:57997 /* EBML header? */
Aurelien Jacobs737c40d2008-08-05 00:42:39998 if (AV_RB32(p->buf) != EBML_ID_HEADER)
999 return 0;
1000
1001 /* length of header */
1002 total = p->buf[4];
1003 while (size <= 8 && !(total & len_mask)) {
1004 size++;
1005 len_mask >>= 1;
1006 }
1007 if (size > 8)
Keiji Costantini84cfce92014-03-01 16:28:151008 return 0;
Aurelien Jacobs737c40d2008-08-05 00:42:391009 total &= (len_mask - 1);
1010 while (n < size)
1011 total = (total << 8) | p->buf[4 + n++];
1012
Diego Biurrun5968d2d2008-08-05 08:28:571013 /* Does the probe data contain the whole header? */
Aurelien Jacobs737c40d2008-08-05 00:42:391014 if (p->buf_size < 4 + size + total)
Keiji Costantini84cfce92014-03-01 16:28:151015 return 0;
Aurelien Jacobs737c40d2008-08-05 00:42:391016
James Zern470491f2010-05-22 01:41:321017 /* The header should contain a known document type. For now,
Aurelien Jacobs737c40d2008-08-05 00:42:391018 * we don't parse the whole header but simply check for the
1019 * availability of that array of characters inside the header.
1020 * Not fully fool-proof, but good enough. */
James Zern470491f2010-05-22 01:41:321021 for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++) {
1022 int probelen = strlen(matroska_doctypes[i]);
Chris Evans69619a12011-07-20 00:51:481023 if (total < probelen)
1024 continue;
Keiji Costantini84cfce92014-03-01 16:28:151025 for (n = 4 + size; n <= 4 + size + total - probelen; n++)
1026 if (!memcmp(p->buf + n, matroska_doctypes[i], probelen))
James Zern470491f2010-05-22 01:41:321027 return AVPROBE_SCORE_MAX;
1028 }
Aurelien Jacobs737c40d2008-08-05 00:42:391029
David Conradc7b913c2010-05-22 01:41:351030 // probably valid EBML header but no recognized doctype
Diego Biurrune0f8be62013-03-25 15:12:511031 return AVPROBE_SCORE_EXTENSION;
Aurelien Jacobs737c40d2008-08-05 00:42:391032}
1033
1034static MatroskaTrack *matroska_find_track_by_num(MatroskaDemuxContext *matroska,
1035 int num)
1036{
1037 MatroskaTrack *tracks = matroska->tracks.elem;
1038 int i;
1039
Keiji Costantini84cfce92014-03-01 16:28:151040 for (i = 0; i < matroska->tracks.nb_elem; i++)
Aurelien Jacobs737c40d2008-08-05 00:42:391041 if (tracks[i].num == num)
1042 return &tracks[i];
1043
1044 av_log(matroska->ctx, AV_LOG_ERROR, "Invalid track number %d\n", num);
1045 return NULL;
1046}
1047
Keiji Costantini84cfce92014-03-01 16:28:151048static int matroska_decode_buffer(uint8_t **buf, int *buf_size,
Aurelien Jacobsf7b96872008-08-05 00:42:051049 MatroskaTrack *track)
Evgeniy Stepanov935ec5a2008-06-22 15:49:441050{
Aurelien Jacobs2cbc8812008-08-05 00:40:311051 MatroskaTrackEncoding *encodings = track->encodings.elem;
Keiji Costantini84cfce92014-03-01 16:28:151052 uint8_t *data = *buf;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441053 int isize = *buf_size;
Keiji Costantini84cfce92014-03-01 16:28:151054 uint8_t *pkt_data = NULL;
Diego Biurrun529506b2012-02-12 09:58:461055 uint8_t av_unused *newpktdata;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441056 int pkt_size = isize;
1057 int result = 0;
1058 int olen;
1059
Aurelien Jacobs4f906882010-08-17 14:05:231060 if (pkt_size >= 10000000)
Luca Barbatoc9a39ce2012-09-14 18:03:371061 return AVERROR_INVALIDDATA;
Aurelien Jacobs4f906882010-08-17 14:05:231062
Aurelien Jacobs2cbc8812008-08-05 00:40:311063 switch (encodings[0].compression.algo) {
Keiji Costantini84cfce92014-03-01 16:28:151064 case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
1065 {
Luca Barbato8d4dd552012-09-14 16:51:491066 int header_size = encodings[0].compression.settings.size;
1067 uint8_t *header = encodings[0].compression.settings.data;
1068
1069 if (!header_size)
1070 return 0;
1071
1072 pkt_size = isize + header_size;
1073 pkt_data = av_malloc(pkt_size);
1074 if (!pkt_data)
1075 return AVERROR(ENOMEM);
1076
1077 memcpy(pkt_data, header, header_size);
1078 memcpy(pkt_data + header_size, data, isize);
1079 break;
1080 }
Diego Biurrun2a91ada2012-10-18 17:48:271081#if CONFIG_LZO
Evgeniy Stepanov935ec5a2008-06-22 15:49:441082 case MATROSKA_TRACK_ENCODING_COMP_LZO:
1083 do {
Keiji Costantini84cfce92014-03-01 16:28:151084 olen = pkt_size *= 3;
Luca Barbato581281e2012-09-14 16:39:581085 newpktdata = av_realloc(pkt_data, pkt_size + AV_LZO_OUTPUT_PADDING);
1086 if (!newpktdata) {
Luca Barbatoc9a39ce2012-09-14 18:03:371087 result = AVERROR(ENOMEM);
Luca Barbato581281e2012-09-14 16:39:581088 goto failed;
1089 }
1090 pkt_data = newpktdata;
Keiji Costantini84cfce92014-03-01 16:28:151091 result = av_lzo1x_decode(pkt_data, &olen, data, &isize);
1092 } while (result == AV_LZO_OUTPUT_FULL && pkt_size < 10000000);
Luca Barbatoc9a39ce2012-09-14 18:03:371093 if (result) {
1094 result = AVERROR_INVALIDDATA;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441095 goto failed;
Luca Barbatoc9a39ce2012-09-14 18:03:371096 }
Evgeniy Stepanov935ec5a2008-06-22 15:49:441097 pkt_size -= olen;
1098 break;
Diego Biurrun2a91ada2012-10-18 17:48:271099#endif
Aurelien Jacobsb250f9c2009-01-13 23:44:161100#if CONFIG_ZLIB
Keiji Costantini84cfce92014-03-01 16:28:151101 case MATROSKA_TRACK_ENCODING_COMP_ZLIB:
1102 {
1103 z_stream zstream = { 0 };
Evgeniy Stepanov935ec5a2008-06-22 15:49:441104 if (inflateInit(&zstream) != Z_OK)
1105 return -1;
Keiji Costantini84cfce92014-03-01 16:28:151106 zstream.next_in = data;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441107 zstream.avail_in = isize;
1108 do {
Keiji Costantini84cfce92014-03-01 16:28:151109 pkt_size *= 3;
Michael Niedermayer77d2ef12011-07-28 12:59:541110 newpktdata = av_realloc(pkt_data, pkt_size);
1111 if (!newpktdata) {
1112 inflateEnd(&zstream);
1113 goto failed;
1114 }
Keiji Costantini84cfce92014-03-01 16:28:151115 pkt_data = newpktdata;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441116 zstream.avail_out = pkt_size - zstream.total_out;
Keiji Costantini84cfce92014-03-01 16:28:151117 zstream.next_out = pkt_data + zstream.total_out;
1118 result = inflate(&zstream, Z_NO_FLUSH);
1119 } while (result == Z_OK && pkt_size < 10000000);
Evgeniy Stepanov935ec5a2008-06-22 15:49:441120 pkt_size = zstream.total_out;
1121 inflateEnd(&zstream);
Luca Barbatoc9a39ce2012-09-14 18:03:371122 if (result != Z_STREAM_END) {
1123 if (result == Z_MEM_ERROR)
1124 result = AVERROR(ENOMEM);
1125 else
1126 result = AVERROR_INVALIDDATA;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441127 goto failed;
Luca Barbatoc9a39ce2012-09-14 18:03:371128 }
Evgeniy Stepanov935ec5a2008-06-22 15:49:441129 break;
1130 }
1131#endif
Aurelien Jacobsb250f9c2009-01-13 23:44:161132#if CONFIG_BZLIB
Keiji Costantini84cfce92014-03-01 16:28:151133 case MATROSKA_TRACK_ENCODING_COMP_BZLIB:
1134 {
1135 bz_stream bzstream = { 0 };
Evgeniy Stepanov935ec5a2008-06-22 15:49:441136 if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
1137 return -1;
Keiji Costantini84cfce92014-03-01 16:28:151138 bzstream.next_in = data;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441139 bzstream.avail_in = isize;
1140 do {
Keiji Costantini84cfce92014-03-01 16:28:151141 pkt_size *= 3;
Michael Niedermayer77d2ef12011-07-28 12:59:541142 newpktdata = av_realloc(pkt_data, pkt_size);
1143 if (!newpktdata) {
1144 BZ2_bzDecompressEnd(&bzstream);
1145 goto failed;
1146 }
Keiji Costantini84cfce92014-03-01 16:28:151147 pkt_data = newpktdata;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441148 bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
Keiji Costantini84cfce92014-03-01 16:28:151149 bzstream.next_out = pkt_data + bzstream.total_out_lo32;
1150 result = BZ2_bzDecompress(&bzstream);
1151 } while (result == BZ_OK && pkt_size < 10000000);
Evgeniy Stepanov935ec5a2008-06-22 15:49:441152 pkt_size = bzstream.total_out_lo32;
1153 BZ2_bzDecompressEnd(&bzstream);
Luca Barbatoc9a39ce2012-09-14 18:03:371154 if (result != BZ_STREAM_END) {
1155 if (result == BZ_MEM_ERROR)
1156 result = AVERROR(ENOMEM);
1157 else
1158 result = AVERROR_INVALIDDATA;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441159 goto failed;
Luca Barbatoc9a39ce2012-09-14 18:03:371160 }
Evgeniy Stepanov935ec5a2008-06-22 15:49:441161 break;
1162 }
1163#endif
Aurelien Jacobs28f27e02008-08-20 23:08:071164 default:
Luca Barbatoc9a39ce2012-09-14 18:03:371165 return AVERROR_INVALIDDATA;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441166 }
1167
Keiji Costantini84cfce92014-03-01 16:28:151168 *buf = pkt_data;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441169 *buf_size = pkt_size;
1170 return 0;
Keiji Costantini84cfce92014-03-01 16:28:151171
1172failed:
Evgeniy Stepanov935ec5a2008-06-22 15:49:441173 av_free(pkt_data);
Luca Barbatoc9a39ce2012-09-14 18:03:371174 return result;
Evgeniy Stepanov935ec5a2008-06-22 15:49:441175}
1176
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121177static void matroska_fix_ass_packet(MatroskaDemuxContext *matroska,
Aurelien Jacobse7d4b742008-09-28 22:55:281178 AVPacket *pkt, uint64_t display_duration)
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121179{
Anton Khirnov1afddbe2012-10-31 07:53:181180 AVBufferRef *line;
Keiji Costantini84cfce92014-03-01 16:28:151181 char *layer, *ptr = pkt->data, *end = ptr + pkt->size;
1182
1183 for (; *ptr != ',' && ptr < end - 1; ptr++)
1184 ;
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121185 if (*ptr == ',')
1186 layer = ++ptr;
Keiji Costantini84cfce92014-03-01 16:28:151187 for (; *ptr != ',' && ptr < end - 1; ptr++)
1188 ;
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121189 if (*ptr == ',') {
Aurelien Jacobse7d4b742008-09-28 22:55:281190 int64_t end_pts = pkt->pts + display_duration;
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121191 int sc = matroska->time_scale * pkt->pts / 10000000;
1192 int ec = matroska->time_scale * end_pts / 10000000;
1193 int sh, sm, ss, eh, em, es, len;
Keiji Costantini84cfce92014-03-01 16:28:151194 sh = sc / 360000;
1195 sc -= 360000 * sh;
1196 sm = sc / 6000;
1197 sc -= 6000 * sm;
1198 ss = sc / 100;
1199 sc -= 100 * ss;
1200 eh = ec / 360000;
1201 ec -= 360000 * eh;
1202 em = ec / 6000;
1203 ec -= 6000 * em;
1204 es = ec / 100;
1205 ec -= 100 * es;
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121206 *ptr++ = '\0';
Keiji Costantini84cfce92014-03-01 16:28:151207 len = 50 + end - ptr + FF_INPUT_BUFFER_PADDING_SIZE;
Anton Khirnov1afddbe2012-10-31 07:53:181208 if (!(line = av_buffer_alloc(len)))
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121209 return;
Keiji Costantini84cfce92014-03-01 16:28:151210 snprintf(line->data, len,
1211 "Dialogue: %s,%d:%02d:%02d.%02d,%d:%02d:%02d.%02d,%s\r\n",
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121212 layer, sh, sm, ss, sc, eh, em, es, ec, ptr);
Anton Khirnov1afddbe2012-10-31 07:53:181213 av_buffer_unref(&pkt->buf);
1214 pkt->buf = line;
1215 pkt->data = line->data;
1216 pkt->size = strlen(line->data);
Aurelien Jacobs3eb9bfb2008-09-04 23:26:121217 }
1218}
1219
Michael Niedermayer77d2ef12011-07-28 12:59:541220static int matroska_merge_packets(AVPacket *out, AVPacket *in)
Aurelien Jacobsd5e34dc2008-09-28 23:06:251221{
Anton Khirnov34871be2012-11-01 05:34:411222 int old_size = out->size;
1223 int ret = av_grow_packet(out, in->size);
1224 if (ret < 0)
1225 return ret;
1226
1227 memcpy(out->data + old_size, in->data, in->size);
1228
1229 av_free_packet(in);
Aurelien Jacobsd5e34dc2008-09-28 23:06:251230 av_free(in);
Michael Niedermayer77d2ef12011-07-28 12:59:541231 return 0;
Aurelien Jacobsd5e34dc2008-09-28 23:06:251232}
1233
Aurelien Jacobs929e9de2009-02-15 15:53:551234static void matroska_convert_tag(AVFormatContext *s, EbmlList *list,
Anton Khirnovd2d67e42011-05-22 10:46:291235 AVDictionary **metadata, char *prefix)
Aurelien Jacobs44015c52008-08-08 23:50:381236{
1237 MatroskaTag *tags = list->elem;
Aurelien Jacobs929e9de2009-02-15 15:53:551238 char key[1024];
1239 int i;
Aurelien Jacobs44015c52008-08-08 23:50:381240
Keiji Costantini84cfce92014-03-01 16:28:151241 for (i = 0; i < list->nb_elem; i++) {
1242 const char *lang = tags[i].lang &&
1243 strcmp(tags[i].lang, "und") ? tags[i].lang : NULL;
Anton Khirnovbf800c72010-11-03 06:29:041244
1245 if (!tags[i].name) {
1246 av_log(s, AV_LOG_WARNING, "Skipping invalid tag with no TagName.\n");
1247 continue;
1248 }
Keiji Costantini84cfce92014-03-01 16:28:151249 if (prefix)
1250 snprintf(key, sizeof(key), "%s/%s", prefix, tags[i].name);
1251 else
1252 av_strlcpy(key, tags[i].name, sizeof(key));
Aurelien Jacobsf702df32009-02-15 16:05:371253 if (tags[i].def || !lang) {
Keiji Costantini84cfce92014-03-01 16:28:151254 av_dict_set(metadata, key, tags[i].string, 0);
1255 if (tags[i].sub.nb_elem)
1256 matroska_convert_tag(s, &tags[i].sub, metadata, key);
Aurelien Jacobsf702df32009-02-15 16:05:371257 }
1258 if (lang) {
1259 av_strlcat(key, "-", sizeof(key));
1260 av_strlcat(key, lang, sizeof(key));
Anton Khirnovd2d67e42011-05-22 10:46:291261 av_dict_set(metadata, key, tags[i].string, 0);
Aurelien Jacobsf702df32009-02-15 16:05:371262 if (tags[i].sub.nb_elem)
1263 matroska_convert_tag(s, &tags[i].sub, metadata, key);
1264 }
Aurelien Jacobs929e9de2009-02-15 15:53:551265 }
Anton Khirnovad7768f2010-10-16 13:20:411266 ff_metadata_conv(metadata, NULL, ff_mkv_metadata_conv);
Aurelien Jacobs929e9de2009-02-15 15:53:551267}
1268
1269static void matroska_convert_tags(AVFormatContext *s)
1270{
1271 MatroskaDemuxContext *matroska = s->priv_data;
1272 MatroskaTags *tags = matroska->tags.elem;
1273 int i, j;
1274
Keiji Costantini84cfce92014-03-01 16:28:151275 for (i = 0; i < matroska->tags.nb_elem; i++) {
Aurelien Jacobs929e9de2009-02-15 15:53:551276 if (tags[i].target.attachuid) {
Diego Biurrunf69befe2014-03-07 12:54:181277 MatroskaAttachment *attachment = matroska->attachments.elem;
Keiji Costantini84cfce92014-03-01 16:28:151278 for (j = 0; j < matroska->attachments.nb_elem; j++)
1279 if (attachment[j].uid == tags[i].target.attachuid &&
1280 attachment[j].stream)
Aurelien Jacobs929e9de2009-02-15 15:53:551281 matroska_convert_tag(s, &tags[i].tag,
1282 &attachment[j].stream->metadata, NULL);
1283 } else if (tags[i].target.chapteruid) {
1284 MatroskaChapter *chapter = matroska->chapters.elem;
Keiji Costantini84cfce92014-03-01 16:28:151285 for (j = 0; j < matroska->chapters.nb_elem; j++)
1286 if (chapter[j].uid == tags[i].target.chapteruid &&
1287 chapter[j].chapter)
Aurelien Jacobs929e9de2009-02-15 15:53:551288 matroska_convert_tag(s, &tags[i].tag,
1289 &chapter[j].chapter->metadata, NULL);
1290 } else if (tags[i].target.trackuid) {
1291 MatroskaTrack *track = matroska->tracks.elem;
Keiji Costantini84cfce92014-03-01 16:28:151292 for (j = 0; j < matroska->tracks.nb_elem; j++)
Aurelien Jacobs9c569ef2011-03-23 23:28:191293 if (track[j].uid == tags[i].target.trackuid && track[j].stream)
Aurelien Jacobs929e9de2009-02-15 15:53:551294 matroska_convert_tag(s, &tags[i].tag,
1295 &track[j].stream->metadata, NULL);
1296 } else {
Aurelien Jacobs4f909c72009-06-13 22:29:381297 matroska_convert_tag(s, &tags[i].tag, &s->metadata,
1298 tags[i].target.type);
Aurelien Jacobs929e9de2009-02-15 15:53:551299 }
Aurelien Jacobs44015c52008-08-08 23:50:381300 }
1301}
1302
Keiji Costantini84cfce92014-03-01 16:28:151303static int matroska_parse_seekhead_entry(MatroskaDemuxContext *matroska,
1304 int idx)
Aurelien Jacobs13b350a2008-08-05 00:40:361305{
1306 EbmlList *seekhead_list = &matroska->seekhead;
Keiji Costantini84cfce92014-03-01 16:28:151307 uint32_t level_up = matroska->level_up;
1308 uint32_t saved_id = matroska->current_id;
Aurelien Jacobs13b350a2008-08-05 00:40:361309 MatroskaSeekhead *seekhead = seekhead_list->elem;
Anton Khirnova2704c92011-03-03 19:11:451310 int64_t before_pos = avio_tell(matroska->ctx->pb);
Aurelien Jacobs13b350a2008-08-05 00:40:361311 MatroskaLevel level;
Aaron Colwell31ad14c2011-07-09 05:48:431312 int64_t offset;
1313 int ret = 0;
David Conradb061d892007-06-04 22:10:541314
Keiji Costantini84cfce92014-03-01 16:28:151315 if (idx >= seekhead_list->nb_elem ||
1316 seekhead[idx].id == MATROSKA_ID_SEEKHEAD ||
1317 seekhead[idx].id == MATROSKA_ID_CLUSTER)
Aaron Colwell31ad14c2011-07-09 05:48:431318 return 0;
David Conradb061d892007-06-04 22:10:541319
Anton Khirnovf47ac3c2011-07-09 06:11:301320 /* seek */
Aaron Colwell31ad14c2011-07-09 05:48:431321 offset = seekhead[idx].pos + matroska->segment_start;
1322 if (avio_seek(matroska->ctx->pb, offset, SEEK_SET) == offset) {
Diego Biurrun5968d2d2008-08-05 08:28:571323 /* We don't want to lose our seekhead level, so we add
Aurelien Jacobs43485712008-08-05 00:40:431324 * a dummy. This is a crude hack. */
1325 if (matroska->num_levels == EBML_MAX_DEPTH) {
1326 av_log(matroska->ctx, AV_LOG_INFO,
1327 "Max EBML element depth (%d) reached, "
1328 "cannot parse further.\n", EBML_MAX_DEPTH);
Aaron Colwell31ad14c2011-07-09 05:48:431329 ret = AVERROR_INVALIDDATA;
1330 } else {
Keiji Costantini84cfce92014-03-01 16:28:151331 level.start = 0;
1332 level.length = (uint64_t) -1;
Anton Khirnovf47ac3c2011-07-09 06:11:301333 matroska->levels[matroska->num_levels] = level;
1334 matroska->num_levels++;
Keiji Costantini84cfce92014-03-01 16:28:151335 matroska->current_id = 0;
David Conradb061d892007-06-04 22:10:541336
Dustin Brody4a9628f2011-09-08 22:43:321337 ret = ebml_parse(matroska, matroska_segment, matroska);
David Conradb061d892007-06-04 22:10:541338
Anton Khirnovf47ac3c2011-07-09 06:11:301339 /* remove dummy level */
1340 while (matroska->num_levels) {
1341 uint64_t length = matroska->levels[--matroska->num_levels].length;
Keiji Costantini84cfce92014-03-01 16:28:151342 if (length == (uint64_t) -1)
Anton Khirnovf47ac3c2011-07-09 06:11:301343 break;
1344 }
Aurelien Jacobs43485712008-08-05 00:40:431345 }
David Conradb061d892007-06-04 22:10:541346 }
Aurelien Jacobs43485712008-08-05 00:40:431347 /* seek back */
Anton Khirnov6b4aa5d2011-02-28 13:57:541348 avio_seek(matroska->ctx->pb, before_pos, SEEK_SET);
Keiji Costantini84cfce92014-03-01 16:28:151349 matroska->level_up = level_up;
Aurelien Jacobsc3ade622010-06-11 16:34:011350 matroska->current_id = saved_id;
Aaron Colwell31ad14c2011-07-09 05:48:431351
1352 return ret;
1353}
1354
1355static void matroska_execute_seekhead(MatroskaDemuxContext *matroska)
1356{
1357 EbmlList *seekhead_list = &matroska->seekhead;
Aaron Colwell31ad14c2011-07-09 05:48:431358 int64_t before_pos = avio_tell(matroska->ctx->pb);
1359 int i;
1360
1361 // we should not do any seeking in the streaming case
1362 if (!matroska->ctx->pb->seekable ||
1363 (matroska->ctx->flags & AVFMT_FLAG_IGNIDX))
1364 return;
1365
1366 for (i = 0; i < seekhead_list->nb_elem; i++) {
Chris Evansfaaec462012-01-05 20:19:301367 MatroskaSeekhead *seekhead = seekhead_list->elem;
Aaron Colwell31ad14c2011-07-09 05:48:431368 if (seekhead[i].pos <= before_pos)
1369 continue;
1370
1371 // defer cues parsing until we actually need cue data.
1372 if (seekhead[i].id == MATROSKA_ID_CUES) {
1373 matroska->cues_parsing_deferred = 1;
1374 continue;
1375 }
1376
1377 if (matroska_parse_seekhead_entry(matroska, i) < 0)
1378 break;
1379 }
1380}
1381
Keiji Costantini84cfce92014-03-01 16:28:151382static void matroska_parse_cues(MatroskaDemuxContext *matroska)
1383{
Aaron Colwell31ad14c2011-07-09 05:48:431384 EbmlList *seekhead_list = &matroska->seekhead;
1385 MatroskaSeekhead *seekhead = seekhead_list->elem;
1386 EbmlList *index_list;
1387 MatroskaIndex *index;
1388 int index_scale = 1;
1389 int i, j;
1390
1391 for (i = 0; i < seekhead_list->nb_elem; i++)
Anton Khirnovfdb94442011-07-10 19:48:231392 if (seekhead[i].id == MATROSKA_ID_CUES)
Aaron Colwell31ad14c2011-07-09 05:48:431393 break;
1394 assert(i <= seekhead_list->nb_elem);
1395
1396 matroska_parse_seekhead_entry(matroska, i);
1397
1398 index_list = &matroska->index;
Keiji Costantini84cfce92014-03-01 16:28:151399 index = index_list->elem;
1400 if (index_list->nb_elem &&
1401 index[0].time > 1E14 / matroska->time_scale) {
Aaron Colwell31ad14c2011-07-09 05:48:431402 av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n");
1403 index_scale = matroska->time_scale;
1404 }
1405 for (i = 0; i < index_list->nb_elem; i++) {
Keiji Costantini84cfce92014-03-01 16:28:151406 EbmlList *pos_list = &index[i].pos;
Aaron Colwell31ad14c2011-07-09 05:48:431407 MatroskaIndexPos *pos = pos_list->elem;
1408 for (j = 0; j < pos_list->nb_elem; j++) {
Keiji Costantini84cfce92014-03-01 16:28:151409 MatroskaTrack *track = matroska_find_track_by_num(matroska,
1410 pos[j].track);
Aaron Colwell31ad14c2011-07-09 05:48:431411 if (track && track->stream)
1412 av_add_index_entry(track->stream,
1413 pos[j].pos + matroska->segment_start,
Keiji Costantini84cfce92014-03-01 16:28:151414 index[i].time / index_scale, 0, 0,
Aaron Colwell31ad14c2011-07-09 05:48:431415 AVINDEX_KEYFRAME);
1416 }
1417 }
David Conradb061d892007-06-04 22:10:541418}
1419
Aurelien Jacobsf7b96872008-08-05 00:42:051420static int matroska_aac_profile(char *codec_id)
David Conradb061d892007-06-04 22:10:541421{
Keiji Costantini84cfce92014-03-01 16:28:151422 static const char *const aac_profiles[] = { "MAIN", "LC", "SSR" };
David Conradb061d892007-06-04 22:10:541423 int profile;
1424
Keiji Costantini84cfce92014-03-01 16:28:151425 for (profile = 0; profile < FF_ARRAY_ELEMS(aac_profiles); profile++)
David Conradb061d892007-06-04 22:10:541426 if (strstr(codec_id, aac_profiles[profile]))
1427 break;
1428 return profile + 1;
1429}
1430
Aurelien Jacobsf7b96872008-08-05 00:42:051431static int matroska_aac_sri(int samplerate)
David Conradb061d892007-06-04 22:10:541432{
David Conradb061d892007-06-04 22:10:541433 int sri;
1434
Keiji Costantini84cfce92014-03-01 16:28:151435 for (sri = 0; sri < FF_ARRAY_ELEMS(avpriv_mpeg4audio_sample_rates); sri++)
Anton Khirnov59a9a232011-10-17 07:28:531436 if (avpriv_mpeg4audio_sample_rates[sri] == samplerate)
David Conradb061d892007-06-04 22:10:541437 break;
1438 return sri;
1439}
1440
Anton Khirnov4efdadc2014-05-25 12:05:511441static int matroska_parse_flac(AVFormatContext *s,
1442 MatroskaTrack *track,
1443 int *offset)
1444{
Anton Khirnov23f741f2014-05-26 10:48:561445 AVStream *st = track->stream;
Anton Khirnov4efdadc2014-05-25 12:05:511446 uint8_t *p = track->codec_priv.data;
1447 int size = track->codec_priv.size;
1448
1449 if (size < 8 + FLAC_STREAMINFO_SIZE || p[4] & 0x7f) {
1450 av_log(s, AV_LOG_WARNING, "Invalid FLAC private data\n");
1451 track->codec_priv.size = 0;
1452 return 0;
1453 }
1454 *offset = 8;
1455 track->codec_priv.size = 8 + FLAC_STREAMINFO_SIZE;
1456
Anton Khirnov23f741f2014-05-26 10:48:561457 p += track->codec_priv.size;
1458 size -= track->codec_priv.size;
1459
1460 /* parse the remaining metadata blocks if present */
1461 while (size >= 4) {
1462 int block_last, block_type, block_size;
1463
1464 flac_parse_block_header(p, &block_last, &block_type, &block_size);
1465
1466 p += 4;
1467 size -= 4;
1468 if (block_size > size)
1469 return 0;
1470
1471 /* check for the channel mask */
1472 if (block_type == FLAC_METADATA_TYPE_VORBIS_COMMENT) {
1473 AVDictionary *dict = NULL;
1474 AVDictionaryEntry *chmask;
1475
1476 ff_vorbis_comment(s, &dict, p, block_size, 0);
1477 chmask = av_dict_get(dict, "WAVEFORMATEXTENSIBLE_CHANNEL_MASK", NULL, 0);
1478 if (chmask) {
1479 uint64_t mask = strtol(chmask->value, NULL, 0);
1480 if (!mask || mask & ~0x3ffffULL) {
1481 av_log(s, AV_LOG_WARNING,
1482 "Invalid value of WAVEFORMATEXTENSIBLE_CHANNEL_MASK\n");
1483 } else
1484 st->codec->channel_layout = mask;
1485 }
1486 av_dict_free(&dict);
1487 }
1488
1489 p += block_size;
1490 size -= block_size;
1491 }
1492
Anton Khirnov4efdadc2014-05-25 12:05:511493 return 0;
1494}
1495
Anton Khirnov6df478b2014-05-25 07:07:321496static int matroska_parse_tracks(AVFormatContext *s)
David Conradb061d892007-06-04 22:10:541497{
1498 MatroskaDemuxContext *matroska = s->priv_data;
Anton Khirnov6df478b2014-05-25 07:07:321499 MatroskaTrack *tracks = matroska->tracks.elem;
Aurelien Jacobs9a9a3b02008-08-05 00:40:491500 AVStream *st;
Anton Khirnov4efdadc2014-05-25 12:05:511501 int i, j, ret;
David Conradb061d892007-06-04 22:10:541502
Keiji Costantini84cfce92014-03-01 16:28:151503 for (i = 0; i < matroska->tracks.nb_elem; i++) {
Aurelien Jacobsd88d8062008-08-05 00:40:521504 MatroskaTrack *track = &tracks[i];
Anton Khirnov36ef5362012-08-05 09:11:041505 enum AVCodecID codec_id = AV_CODEC_ID_NONE;
Aurelien Jacobs8b516f12011-08-21 14:03:131506 EbmlList *encodings_list = &track->encodings;
Aurelien Jacobs9c25baf2008-08-05 00:40:551507 MatroskaTrackEncoding *encodings = encodings_list->elem;
Aurelien Jacobsd88d8062008-08-05 00:40:521508 uint8_t *extradata = NULL;
1509 int extradata_size = 0;
1510 int extradata_offset = 0;
Anton Khirnovae628ec2011-02-20 10:04:121511 AVIOContext b;
David Conradb061d892007-06-04 22:10:541512
Aurelien Jacobsd88d8062008-08-05 00:40:521513 /* Apply some sanity checks. */
Aurelien Jacobs9c25baf2008-08-05 00:40:551514 if (track->type != MATROSKA_TRACK_TYPE_VIDEO &&
1515 track->type != MATROSKA_TRACK_TYPE_AUDIO &&
1516 track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1517 av_log(matroska->ctx, AV_LOG_INFO,
1518 "Unknown or unsupported track type %"PRIu64"\n",
1519 track->type);
1520 continue;
1521 }
Gabriel Dumef929ab02014-08-14 20:31:241522 if (!track->codec_id)
Aurelien Jacobsd88d8062008-08-05 00:40:521523 continue;
David Conradb061d892007-06-04 22:10:541524
Aurelien Jacobs9c25baf2008-08-05 00:40:551525 if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
Mans Rullgard3c583002012-04-18 12:48:201526 if (!track->default_duration && track->video.frame_rate > 0)
Keiji Costantini84cfce92014-03-01 16:28:151527 track->default_duration = 1000000000 / track->video.frame_rate;
Aurelien Jacobs9c25baf2008-08-05 00:40:551528 if (!track->video.display_width)
1529 track->video.display_width = track->video.pixel_width;
1530 if (!track->video.display_height)
1531 track->video.display_height = track->video.pixel_height;
1532 } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1533 if (!track->audio.out_samplerate)
1534 track->audio.out_samplerate = track->audio.samplerate;
1535 }
1536 if (encodings_list->nb_elem > 1) {
1537 av_log(matroska->ctx, AV_LOG_ERROR,
Dustin Brodyd7d2f0e2011-09-15 07:34:381538 "Multiple combined encodings not supported");
Aurelien Jacobs9c25baf2008-08-05 00:40:551539 } else if (encodings_list->nb_elem == 1) {
1540 if (encodings[0].type ||
Diego Biurrun2a91ada2012-10-18 17:48:271541 (
Aurelien Jacobsb250f9c2009-01-13 23:44:161542#if CONFIG_ZLIB
Keiji Costantini84cfce92014-03-01 16:28:151543 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
Aurelien Jacobs9c25baf2008-08-05 00:40:551544#endif
Aurelien Jacobsb250f9c2009-01-13 23:44:161545#if CONFIG_BZLIB
Aurelien Jacobs9c25baf2008-08-05 00:40:551546 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
1547#endif
Diego Biurrun2a91ada2012-10-18 17:48:271548#if CONFIG_LZO
Keiji Costantini84cfce92014-03-01 16:28:151549 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_LZO &&
Diego Biurrun2a91ada2012-10-18 17:48:271550#endif
1551 encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP)) {
Aurelien Jacobs9c25baf2008-08-05 00:40:551552 encodings[0].scope = 0;
1553 av_log(matroska->ctx, AV_LOG_ERROR,
1554 "Unsupported encoding type");
Keiji Costantini84cfce92014-03-01 16:28:151555 } else if (track->codec_priv.size && encodings[0].scope & 2) {
Aurelien Jacobs9c25baf2008-08-05 00:40:551556 uint8_t *codec_priv = track->codec_priv.data;
Luca Barbato8d4dd552012-09-14 16:51:491557 int ret = matroska_decode_buffer(&track->codec_priv.data,
1558 &track->codec_priv.size,
1559 track);
1560 if (ret < 0) {
Aurelien Jacobs9c25baf2008-08-05 00:40:551561 track->codec_priv.data = NULL;
1562 track->codec_priv.size = 0;
1563 av_log(matroska->ctx, AV_LOG_ERROR,
1564 "Failed to decode codec private data\n");
Aurelien Jacobs9c25baf2008-08-05 00:40:551565 }
Luca Barbato8d4dd552012-09-14 16:51:491566
Aurelien Jacobs9c25baf2008-08-05 00:40:551567 if (codec_priv != track->codec_priv.data)
1568 av_free(codec_priv);
1569 }
1570 }
1571
Keiji Costantini84cfce92014-03-01 16:28:151572 for (j = 0; ff_mkv_codec_tags[j].id != AV_CODEC_ID_NONE; j++) {
1573 if (!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
1574 strlen(ff_mkv_codec_tags[j].str))) {
1575 codec_id = ff_mkv_codec_tags[j].id;
Aurelien Jacobsd88d8062008-08-05 00:40:521576 break;
David Conradb061d892007-06-04 22:10:541577 }
Aurelien Jacobsd88d8062008-08-05 00:40:521578 }
David Conradb061d892007-06-04 22:10:541579
Anton Khirnov3b3bbdd2011-06-18 09:43:241580 st = track->stream = avformat_new_stream(s, NULL);
Gabriel Dumef929ab02014-08-14 20:31:241581 if (!st)
Aurelien Jacobsd88d8062008-08-05 00:40:521582 return AVERROR(ENOMEM);
1583
Keiji Costantini84cfce92014-03-01 16:28:151584 if (!strcmp(track->codec_id, "V_MS/VFW/FOURCC") &&
1585 track->codec_priv.size >= 40 &&
1586 track->codec_priv.data != NULL) {
1587 track->ms_compat = 1;
Aurelien Jacobsd88d8062008-08-05 00:40:521588 track->video.fourcc = AV_RL32(track->codec_priv.data + 16);
Keiji Costantini84cfce92014-03-01 16:28:151589 codec_id = ff_codec_get_id(ff_codec_bmp_tags,
1590 track->video.fourcc);
1591 extradata_offset = 40;
1592 } else if (!strcmp(track->codec_id, "A_MS/ACM") &&
1593 track->codec_priv.size >= 14 &&
1594 track->codec_priv.data != NULL) {
Max Hornca402f32011-04-12 15:44:201595 int ret;
Keiji Costantini84cfce92014-03-01 16:28:151596 ffio_init_context(&b, track->codec_priv.data,
1597 track->codec_priv.size,
Luca Barbato37cb3b12013-03-12 17:56:281598 0, NULL, NULL, NULL, NULL);
Max Hornca402f32011-04-12 15:44:201599 ret = ff_get_wav_header(&b, st->codec, track->codec_priv.size);
1600 if (ret < 0)
1601 return ret;
Keiji Costantini84cfce92014-03-01 16:28:151602 codec_id = st->codec->codec_id;
Aurelien Jacobs038146e2009-10-01 21:14:051603 extradata_offset = FFMIN(track->codec_priv.size, 18);
Keiji Costantini84cfce92014-03-01 16:28:151604 } else if (!strcmp(track->codec_id, "V_QUICKTIME") &&
1605 (track->codec_priv.size >= 86) &&
1606 (track->codec_priv.data != NULL)) {
Aurelien Jacobsd88d8062008-08-05 00:40:521607 track->video.fourcc = AV_RL32(track->codec_priv.data);
Keiji Costantini84cfce92014-03-01 16:28:151608 codec_id = ff_codec_get_id(ff_codec_movvideo_tags,
1609 track->video.fourcc);
Anton Khirnov36ef5362012-08-05 09:11:041610 } else if (codec_id == AV_CODEC_ID_PCM_S16BE) {
Aurelien Jacobseb9cf502008-08-20 00:49:451611 switch (track->audio.bitdepth) {
Keiji Costantini84cfce92014-03-01 16:28:151612 case 8:
1613 codec_id = AV_CODEC_ID_PCM_U8;
1614 break;
1615 case 24:
1616 codec_id = AV_CODEC_ID_PCM_S24BE;
1617 break;
1618 case 32:
1619 codec_id = AV_CODEC_ID_PCM_S32BE;
1620 break;
Aurelien Jacobseb9cf502008-08-20 00:49:451621 }
Anton Khirnov36ef5362012-08-05 09:11:041622 } else if (codec_id == AV_CODEC_ID_PCM_S16LE) {
Aurelien Jacobseb9cf502008-08-20 00:49:451623 switch (track->audio.bitdepth) {
Keiji Costantini84cfce92014-03-01 16:28:151624 case 8:
1625 codec_id = AV_CODEC_ID_PCM_U8;
1626 break;
1627 case 24:
1628 codec_id = AV_CODEC_ID_PCM_S24LE;
1629 break;
1630 case 32:
1631 codec_id = AV_CODEC_ID_PCM_S32LE;
1632 break;
Aurelien Jacobseb9cf502008-08-20 00:49:451633 }
Keiji Costantini84cfce92014-03-01 16:28:151634 } else if (codec_id == AV_CODEC_ID_PCM_F32LE &&
1635 track->audio.bitdepth == 64) {
Anton Khirnov36ef5362012-08-05 09:11:041636 codec_id = AV_CODEC_ID_PCM_F64LE;
1637 } else if (codec_id == AV_CODEC_ID_AAC && !track->codec_priv.size) {
Aurelien Jacobsd88d8062008-08-05 00:40:521638 int profile = matroska_aac_profile(track->codec_id);
Keiji Costantini84cfce92014-03-01 16:28:151639 int sri = matroska_aac_sri(track->audio.samplerate);
1640 extradata = av_mallocz(5 + FF_INPUT_BUFFER_PADDING_SIZE);
Gabriel Dumef929ab02014-08-14 20:31:241641 if (!extradata)
Aurelien Jacobs28f450a2008-08-05 00:40:091642 return AVERROR(ENOMEM);
Keiji Costantini84cfce92014-03-01 16:28:151643 extradata[0] = (profile << 3) | ((sri & 0x0E) >> 1);
1644 extradata[1] = ((sri & 0x01) << 7) | (track->audio.channels << 3);
Aurelien Jacobsd88d8062008-08-05 00:40:521645 if (strstr(track->codec_id, "SBR")) {
Keiji Costantini84cfce92014-03-01 16:28:151646 sri = matroska_aac_sri(track->audio.out_samplerate);
1647 extradata[2] = 0x56;
1648 extradata[3] = 0xE5;
1649 extradata[4] = 0x80 | (sri << 3);
Aurelien Jacobsd88d8062008-08-05 00:40:521650 extradata_size = 5;
Aurelien Jacobs16f97ab2008-08-05 00:41:101651 } else
Aurelien Jacobsd88d8062008-08-05 00:40:521652 extradata_size = 2;
Moritz Bunkus8071dca2012-09-14 20:26:141653 } else if (codec_id == AV_CODEC_ID_ALAC && track->codec_priv.size) {
1654 /* Only ALAC's magic cookie is stored in Matroska's track headers.
Keiji Costantini84cfce92014-03-01 16:28:151655 * Create the "atom size", "tag", and "tag version" fields the
1656 * decoder expects manually. */
Moritz Bunkus8071dca2012-09-14 20:26:141657 extradata_size = 12 + track->codec_priv.size;
Keiji Costantini84cfce92014-03-01 16:28:151658 extradata = av_mallocz(extradata_size +
1659 FF_INPUT_BUFFER_PADDING_SIZE);
Gabriel Dumef929ab02014-08-14 20:31:241660 if (!extradata)
Moritz Bunkus8071dca2012-09-14 20:26:141661 return AVERROR(ENOMEM);
1662 AV_WB32(extradata, extradata_size);
1663 memcpy(&extradata[4], "alac", 4);
1664 AV_WB32(&extradata[8], 0);
1665 memcpy(&extradata[12], track->codec_priv.data,
Keiji Costantini84cfce92014-03-01 16:28:151666 track->codec_priv.size);
Anton Khirnov36ef5362012-08-05 09:11:041667 } else if (codec_id == AV_CODEC_ID_TTA) {
Aurelien Jacobsd88d8062008-08-05 00:40:521668 extradata_size = 30;
Keiji Costantini84cfce92014-03-01 16:28:151669 extradata = av_mallocz(extradata_size);
Gabriel Dumef929ab02014-08-14 20:31:241670 if (!extradata)
Aurelien Jacobsd88d8062008-08-05 00:40:521671 return AVERROR(ENOMEM);
Anton Khirnove731b8d2011-02-20 10:04:131672 ffio_init_context(&b, extradata, extradata_size, 1,
Keiji Costantini84cfce92014-03-01 16:28:151673 NULL, NULL, NULL, NULL);
Anton Khirnov77eb5502011-02-21 18:28:171674 avio_write(&b, "TTA1", 4);
1675 avio_wl16(&b, 1);
1676 avio_wl16(&b, track->audio.channels);
1677 avio_wl16(&b, track->audio.bitdepth);
1678 avio_wl32(&b, track->audio.out_samplerate);
Keiji Costantini84cfce92014-03-01 16:28:151679 avio_wl32(&b, matroska->ctx->duration *
1680 track->audio.out_samplerate);
1681 } else if (codec_id == AV_CODEC_ID_RV10 ||
1682 codec_id == AV_CODEC_ID_RV20 ||
1683 codec_id == AV_CODEC_ID_RV30 ||
1684 codec_id == AV_CODEC_ID_RV40) {
Aurelien Jacobsd88d8062008-08-05 00:40:521685 extradata_offset = 26;
Anton Khirnov36ef5362012-08-05 09:11:041686 } else if (codec_id == AV_CODEC_ID_RA_144) {
Aurelien Jacobsd88d8062008-08-05 00:40:521687 track->audio.out_samplerate = 8000;
Keiji Costantini84cfce92014-03-01 16:28:151688 track->audio.channels = 1;
1689 } else if (codec_id == AV_CODEC_ID_RA_288 ||
1690 codec_id == AV_CODEC_ID_COOK ||
1691 codec_id == AV_CODEC_ID_ATRAC3 ||
1692 codec_id == AV_CODEC_ID_SIPR) {
Aurelien Jacobs6b102282010-03-12 23:49:061693 int flavor;
Keiji Costantini84cfce92014-03-01 16:28:151694 ffio_init_context(&b, track->codec_priv.data,
1695 track->codec_priv.size,
1696 0, NULL, NULL, NULL, NULL);
Anton Khirnov45a8a022011-03-15 08:14:381697 avio_skip(&b, 22);
Anton Khirnovb7effd42011-02-21 15:43:011698 flavor = avio_rb16(&b);
1699 track->audio.coded_framesize = avio_rb32(&b);
Anton Khirnov45a8a022011-03-15 08:14:381700 avio_skip(&b, 12);
Anton Khirnovb7effd42011-02-21 15:43:011701 track->audio.sub_packet_h = avio_rb16(&b);
1702 track->audio.frame_size = avio_rb16(&b);
1703 track->audio.sub_packet_size = avio_rb16(&b);
Keiji Costantini84cfce92014-03-01 16:28:151704 if (flavor <= 0 ||
1705 track->audio.coded_framesize <= 0 ||
1706 track->audio.sub_packet_h <= 0 ||
1707 track->audio.frame_size <= 0 ||
Martin Storsjö569d18a2013-09-16 12:36:241708 track->audio.sub_packet_size <= 0)
1709 return AVERROR_INVALIDDATA;
Keiji Costantini84cfce92014-03-01 16:28:151710 track->audio.buf = av_malloc(track->audio.frame_size *
1711 track->audio.sub_packet_h);
Anton Khirnov36ef5362012-08-05 09:11:041712 if (codec_id == AV_CODEC_ID_RA_288) {
Aurelien Jacobsd88d8062008-08-05 00:40:521713 st->codec->block_align = track->audio.coded_framesize;
1714 track->codec_priv.size = 0;
1715 } else {
Anton Khirnov36ef5362012-08-05 09:11:041716 if (codec_id == AV_CODEC_ID_SIPR && flavor < 4) {
Aurelien Jacobs6b102282010-03-12 23:49:061717 const int sipr_bit_rate[4] = { 6504, 8496, 5000, 16000 };
1718 track->audio.sub_packet_size = ff_sipr_subpk_size[flavor];
Keiji Costantini84cfce92014-03-01 16:28:151719 st->codec->bit_rate = sipr_bit_rate[flavor];
Aurelien Jacobs6b102282010-03-12 23:49:061720 }
Aurelien Jacobsd88d8062008-08-05 00:40:521721 st->codec->block_align = track->audio.sub_packet_size;
Keiji Costantini84cfce92014-03-01 16:28:151722 extradata_offset = 78;
David Conradb061d892007-06-04 22:10:541723 }
Anton Khirnov4efdadc2014-05-25 12:05:511724 } else if (codec_id == AV_CODEC_ID_FLAC && track->codec_priv.size) {
1725 ret = matroska_parse_flac(s, track, &extradata_offset);
1726 if (ret < 0)
1727 return ret;
David Conradb061d892007-06-04 22:10:541728 }
Aurelien Jacobse2644402009-08-24 13:40:301729 track->codec_priv.size -= extradata_offset;
Aurelien Jacobsd88d8062008-08-05 00:40:521730
Anton Khirnov36ef5362012-08-05 09:11:041731 if (codec_id == AV_CODEC_ID_NONE)
Aurelien Jacobsd88d8062008-08-05 00:40:521732 av_log(matroska->ctx, AV_LOG_INFO,
Anton Khirnov36ef5362012-08-05 09:11:041733 "Unknown/unsupported AVCodecID %s.\n", track->codec_id);
Aurelien Jacobsd88d8062008-08-05 00:40:521734
Aurelien Jacobs3fc9d7c2008-09-09 11:23:481735 if (track->time_scale < 0.01)
1736 track->time_scale = 1.0;
Keiji Costantini84cfce92014-03-01 16:28:151737 avpriv_set_pts_info(st, 64, matroska->time_scale * track->time_scale,
1738 1000 * 1000 * 1000); /* 64 bit pts in ns */
Aurelien Jacobsd88d8062008-08-05 00:40:521739
Anton Khirnoveb3b5502014-04-29 10:03:131740 /* convert the delay from ns to the track timebase */
1741 track->codec_delay = av_rescale_q(track->codec_delay,
1742 (AVRational){ 1, 1000000000 },
1743 st->time_base);
1744
Aurelien Jacobsd88d8062008-08-05 00:40:521745 st->codec->codec_id = codec_id;
Keiji Costantini84cfce92014-03-01 16:28:151746 st->start_time = 0;
Aurelien Jacobsd88d8062008-08-05 00:40:521747 if (strcmp(track->language, "und"))
Anton Khirnovd2d67e42011-05-22 10:46:291748 av_dict_set(&st->metadata, "language", track->language, 0);
1749 av_dict_set(&st->metadata, "title", track->name, 0);
Aurelien Jacobsd88d8062008-08-05 00:40:521750
1751 if (track->flag_default)
1752 st->disposition |= AV_DISPOSITION_DEFAULT;
Aurelien Jacobs7a617a82010-07-02 16:38:441753 if (track->flag_forced)
1754 st->disposition |= AV_DISPOSITION_FORCED;
Aurelien Jacobsd88d8062008-08-05 00:40:521755
Aurelien Jacobsff0d5a72009-10-01 21:14:461756 if (!st->codec->extradata) {
Keiji Costantini84cfce92014-03-01 16:28:151757 if (extradata) {
1758 st->codec->extradata = extradata;
Aurelien Jacobs553e9f72009-10-01 21:15:361759 st->codec->extradata_size = extradata_size;
Keiji Costantini84cfce92014-03-01 16:28:151760 } else if (track->codec_priv.data && track->codec_priv.size > 0) {
Aurelien Jacobs553e9f72009-10-01 21:15:361761 st->codec->extradata = av_mallocz(track->codec_priv.size +
1762 FF_INPUT_BUFFER_PADDING_SIZE);
Gabriel Dumef929ab02014-08-14 20:31:241763 if (!st->codec->extradata)
Aurelien Jacobs553e9f72009-10-01 21:15:361764 return AVERROR(ENOMEM);
1765 st->codec->extradata_size = track->codec_priv.size;
1766 memcpy(st->codec->extradata,
1767 track->codec_priv.data + extradata_offset,
1768 track->codec_priv.size);
1769 }
Aurelien Jacobsff0d5a72009-10-01 21:14:461770 }
Aurelien Jacobsd88d8062008-08-05 00:40:521771
1772 if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
Stefano Sabatini72415b22010-03-30 23:30:551773 st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
Aurelien Jacobsd88d8062008-08-05 00:40:521774 st->codec->codec_tag = track->video.fourcc;
Keiji Costantini84cfce92014-03-01 16:28:151775 st->codec->width = track->video.pixel_width;
1776 st->codec->height = track->video.pixel_height;
Aurelien Jacobs59729452008-08-23 23:43:201777 av_reduce(&st->sample_aspect_ratio.num,
1778 &st->sample_aspect_ratio.den,
Aurelien Jacobsd88d8062008-08-05 00:40:521779 st->codec->height * track->video.display_width,
Keiji Costantini84cfce92014-03-01 16:28:151780 st->codec->width * track->video.display_height,
Aurelien Jacobsd88d8062008-08-05 00:40:521781 255);
Yusuke Nakamura959bea12013-10-03 05:34:531782 if (st->codec->codec_id != AV_CODEC_ID_H264 &&
1783 st->codec->codec_id != AV_CODEC_ID_HEVC)
1784 st->need_parsing = AVSTREAM_PARSE_HEADERS;
Luca Barbatoac97d472012-04-17 23:32:071785 if (track->default_duration) {
Anton Khirnovaba232c2012-06-26 11:10:011786 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
Luca Barbatoac97d472012-04-17 23:32:071787 1000000000, track->default_duration, 30000);
Luca Barbatoac97d472012-04-17 23:32:071788 }
Aurelien Jacobsd88d8062008-08-05 00:40:521789 } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
Keiji Costantini84cfce92014-03-01 16:28:151790 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
Aurelien Jacobsd88d8062008-08-05 00:40:521791 st->codec->sample_rate = track->audio.out_samplerate;
Keiji Costantini84cfce92014-03-01 16:28:151792 st->codec->channels = track->audio.channels;
Anton Khirnov36ef5362012-08-05 09:11:041793 if (st->codec->codec_id != AV_CODEC_ID_AAC)
Keiji Costantini84cfce92014-03-01 16:28:151794 st->need_parsing = AVSTREAM_PARSE_HEADERS;
Aurelien Jacobsd88d8062008-08-05 00:40:521795 } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
Stefano Sabatini72415b22010-03-30 23:30:551796 st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
Anton Khirnov36ef5362012-08-05 09:11:041797 if (st->codec->codec_id == AV_CODEC_ID_SSA)
Dale Curtis8336eb62012-04-19 18:12:241798 matroska->contains_ssa = 1;
Aurelien Jacobsd88d8062008-08-05 00:40:521799 }
David Conradb061d892007-06-04 22:10:541800 }
1801
Anton Khirnov6df478b2014-05-25 07:07:321802 return 0;
1803}
1804
1805static int matroska_read_header(AVFormatContext *s)
1806{
1807 MatroskaDemuxContext *matroska = s->priv_data;
1808 EbmlList *attachments_list = &matroska->attachments;
1809 EbmlList *chapters_list = &matroska->chapters;
1810 MatroskaAttachment *attachments;
1811 MatroskaChapter *chapters;
1812 uint64_t max_start = 0;
1813 int64_t pos;
1814 Ebml ebml = { 0 };
1815 int i, j, res;
1816
1817 matroska->ctx = s;
1818
1819 /* First read the EBML header. */
1820 if (ebml_parse(matroska, ebml_syntax, &ebml) ||
1821 ebml.version > EBML_VERSION ||
1822 ebml.max_size > sizeof(uint64_t) ||
1823 ebml.id_length > sizeof(uint32_t) ||
1824 ebml.doctype_version > 2) {
1825 av_log(matroska->ctx, AV_LOG_ERROR,
1826 "EBML header using unsupported features\n"
1827 "(EBML version %"PRIu64", doctype %s, doc version %"PRIu64")\n",
1828 ebml.version, ebml.doctype, ebml.doctype_version);
1829 ebml_free(ebml_syntax, &ebml);
1830 return AVERROR_PATCHWELCOME;
1831 }
1832 for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++)
1833 if (!strcmp(ebml.doctype, matroska_doctypes[i]))
1834 break;
1835 if (i >= FF_ARRAY_ELEMS(matroska_doctypes)) {
1836 av_log(s, AV_LOG_WARNING, "Unknown EBML doctype '%s'\n", ebml.doctype);
1837 if (matroska->ctx->error_recognition & AV_EF_EXPLODE) {
1838 ebml_free(ebml_syntax, &ebml);
1839 return AVERROR_INVALIDDATA;
1840 }
1841 }
1842 ebml_free(ebml_syntax, &ebml);
1843
1844 /* The next thing is a segment. */
1845 pos = avio_tell(matroska->ctx->pb);
1846 res = ebml_parse(matroska, matroska_segments, matroska);
1847 // try resyncing until we find a EBML_STOP type element.
1848 while (res != 1) {
1849 res = matroska_resync(matroska, pos);
1850 if (res < 0)
1851 return res;
1852 pos = avio_tell(matroska->ctx->pb);
1853 res = ebml_parse(matroska, matroska_segment, matroska);
1854 }
1855 matroska_execute_seekhead(matroska);
1856
1857 if (!matroska->time_scale)
1858 matroska->time_scale = 1000000;
1859 if (matroska->duration)
1860 matroska->ctx->duration = matroska->duration * matroska->time_scale *
1861 1000 / AV_TIME_BASE;
1862 av_dict_set(&s->metadata, "title", matroska->title, 0);
1863
1864 res = matroska_parse_tracks(s);
1865 if (res < 0)
1866 return res;
1867
Diego Biurrunf69befe2014-03-07 12:54:181868 attachments = attachments_list->elem;
1869 for (j = 0; j < attachments_list->nb_elem; j++) {
1870 if (!(attachments[j].filename && attachments[j].mime &&
1871 attachments[j].bin.data && attachments[j].bin.size > 0)) {
Aurelien Jacobs9c25baf2008-08-05 00:40:551872 av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
1873 } else {
Anton Khirnov3b3bbdd2011-06-18 09:43:241874 AVStream *st = avformat_new_stream(s, NULL);
Gabriel Dumef929ab02014-08-14 20:31:241875 if (!st)
Aurelien Jacobs9c25baf2008-08-05 00:40:551876 break;
Diego Biurrunf69befe2014-03-07 12:54:181877 av_dict_set(&st->metadata, "filename", attachments[j].filename, 0);
1878 av_dict_set(&st->metadata, "mimetype", attachments[j].mime, 0);
Keiji Costantini84cfce92014-03-01 16:28:151879 st->codec->codec_id = AV_CODEC_ID_NONE;
Stefano Sabatini72415b22010-03-30 23:30:551880 st->codec->codec_type = AVMEDIA_TYPE_ATTACHMENT;
Diego Biurrunf69befe2014-03-07 12:54:181881 st->codec->extradata = av_malloc(attachments[j].bin.size);
Gabriel Dumef929ab02014-08-14 20:31:241882 if (!st->codec->extradata)
Aurelien Jacobs9c25baf2008-08-05 00:40:551883 break;
Diego Biurrunf69befe2014-03-07 12:54:181884 st->codec->extradata_size = attachments[j].bin.size;
1885 memcpy(st->codec->extradata, attachments[j].bin.data,
1886 attachments[j].bin.size);
Aurelien Jacobs9c25baf2008-08-05 00:40:551887
Keiji Costantini84cfce92014-03-01 16:28:151888 for (i = 0; ff_mkv_mime_tags[i].id != AV_CODEC_ID_NONE; i++) {
Diego Biurrunf69befe2014-03-07 12:54:181889 if (!strncmp(ff_mkv_mime_tags[i].str, attachments[j].mime,
Aurelien Jacobs9c25baf2008-08-05 00:40:551890 strlen(ff_mkv_mime_tags[i].str))) {
1891 st->codec->codec_id = ff_mkv_mime_tags[i].id;
1892 break;
1893 }
1894 }
Diego Biurrunf69befe2014-03-07 12:54:181895 attachments[j].stream = st;
Aurelien Jacobs9c25baf2008-08-05 00:40:551896 }
1897 }
1898
1899 chapters = chapters_list->elem;
Keiji Costantini84cfce92014-03-01 16:28:151900 for (i = 0; i < chapters_list->nb_elem; i++)
1901 if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid &&
1902 (max_start == 0 || chapters[i].start > max_start)) {
Aurelien Jacobs6cb6e152009-02-15 15:25:141903 chapters[i].chapter =
Keiji Costantini84cfce92014-03-01 16:28:151904 avpriv_new_chapter(s, chapters[i].uid,
1905 (AVRational) { 1, 1000000000 },
1906 chapters[i].start, chapters[i].end,
1907 chapters[i].title);
Anton Khirnovd2d67e42011-05-22 10:46:291908 av_dict_set(&chapters[i].chapter->metadata,
Keiji Costantini84cfce92014-03-01 16:28:151909 "title", chapters[i].title, 0);
Aurelien Jacobse0e4be52009-01-15 00:42:571910 max_start = chapters[i].start;
1911 }
Aurelien Jacobs9c25baf2008-08-05 00:40:551912
Aurelien Jacobs929e9de2009-02-15 15:53:551913 matroska_convert_tags(s);
1914
Aurelien Jacobsce6f28b2008-08-05 00:40:581915 return 0;
David Conradb061d892007-06-04 22:10:541916}
1917
Aurelien Jacobs737c40d2008-08-05 00:42:391918/*
Aurelien Jacobs737c40d2008-08-05 00:42:391919 * Put one packet in an application-supplied AVPacket struct.
1920 * Returns 0 on success or -1 on failure.
1921 */
1922static int matroska_deliver_packet(MatroskaDemuxContext *matroska,
1923 AVPacket *pkt)
1924{
1925 if (matroska->num_packets > 0) {
1926 memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
1927 av_free(matroska->packets[0]);
1928 if (matroska->num_packets > 1) {
Michael Niedermayer77d2ef12011-07-28 12:59:541929 void *newpackets;
Aurelien Jacobs737c40d2008-08-05 00:42:391930 memmove(&matroska->packets[0], &matroska->packets[1],
1931 (matroska->num_packets - 1) * sizeof(AVPacket *));
Michael Niedermayer77d2ef12011-07-28 12:59:541932 newpackets = av_realloc(matroska->packets,
Keiji Costantini84cfce92014-03-01 16:28:151933 (matroska->num_packets - 1) *
1934 sizeof(AVPacket *));
Michael Niedermayer77d2ef12011-07-28 12:59:541935 if (newpackets)
1936 matroska->packets = newpackets;
Aurelien Jacobs737c40d2008-08-05 00:42:391937 } else {
1938 av_freep(&matroska->packets);
Dale Curtis8336eb62012-04-19 18:12:241939 matroska->prev_pkt = NULL;
Aurelien Jacobs737c40d2008-08-05 00:42:391940 }
1941 matroska->num_packets--;
1942 return 0;
1943 }
1944
1945 return -1;
1946}
1947
1948/*
1949 * Free all packets in our internal queue.
1950 */
1951static void matroska_clear_queue(MatroskaDemuxContext *matroska)
1952{
Dale Curtisae3d4162013-01-10 19:05:291953 matroska->prev_pkt = NULL;
Aurelien Jacobs737c40d2008-08-05 00:42:391954 if (matroska->packets) {
1955 int n;
1956 for (n = 0; n < matroska->num_packets; n++) {
1957 av_free_packet(matroska->packets[n]);
1958 av_free(matroska->packets[n]);
1959 }
Aurelien Jacobs00a34312008-08-06 00:21:101960 av_freep(&matroska->packets);
Aurelien Jacobs737c40d2008-08-05 00:42:391961 matroska->num_packets = 0;
1962 }
1963}
1964
Luca Barbato2d0e7712012-09-16 23:58:321965static int matroska_parse_laces(MatroskaDemuxContext *matroska, uint8_t **buf,
Keiji Costantini84cfce92014-03-01 16:28:151966 int *buf_size, int type,
Luca Barbato2d0e7712012-09-16 23:58:321967 uint32_t **lace_buf, int *laces)
1968{
Dale Curtisfc43c192013-03-27 21:02:031969 int res = 0, n, size = *buf_size;
Luca Barbato2d0e7712012-09-16 23:58:321970 uint8_t *data = *buf;
1971 uint32_t *lace_size;
1972
1973 if (!type) {
Keiji Costantini84cfce92014-03-01 16:28:151974 *laces = 1;
Luca Barbato2d0e7712012-09-16 23:58:321975 *lace_buf = av_mallocz(sizeof(int));
1976 if (!*lace_buf)
1977 return AVERROR(ENOMEM);
1978
1979 *lace_buf[0] = size;
1980 return 0;
1981 }
1982
1983 assert(size > 0);
Keiji Costantini84cfce92014-03-01 16:28:151984 *laces = *data + 1;
1985 data += 1;
1986 size -= 1;
Luca Barbato2d0e7712012-09-16 23:58:321987 lace_size = av_mallocz(*laces * sizeof(int));
1988 if (!lace_size)
1989 return AVERROR(ENOMEM);
1990
1991 switch (type) {
Keiji Costantini84cfce92014-03-01 16:28:151992 case 0x1: /* Xiph lacing */
1993 {
Luca Barbato2d0e7712012-09-16 23:58:321994 uint8_t temp;
1995 uint32_t total = 0;
1996 for (n = 0; res == 0 && n < *laces - 1; n++) {
1997 while (1) {
1998 if (size == 0) {
1999 res = AVERROR_EOF;
2000 break;
2001 }
Keiji Costantini84cfce92014-03-01 16:28:152002 temp = *data;
Luca Barbato2d0e7712012-09-16 23:58:322003 lace_size[n] += temp;
Keiji Costantini84cfce92014-03-01 16:28:152004 data += 1;
2005 size -= 1;
Luca Barbato2d0e7712012-09-16 23:58:322006 if (temp != 0xff)
2007 break;
2008 }
2009 total += lace_size[n];
2010 }
2011 if (size <= total) {
2012 res = AVERROR_INVALIDDATA;
2013 break;
2014 }
2015
2016 lace_size[n] = size - total;
2017 break;
2018 }
2019
2020 case 0x2: /* fixed-size lacing */
Anton Khirnov87b017a2012-09-20 18:04:562021 if (size % (*laces)) {
Luca Barbato2d0e7712012-09-16 23:58:322022 res = AVERROR_INVALIDDATA;
2023 break;
2024 }
2025 for (n = 0; n < *laces; n++)
2026 lace_size[n] = size / *laces;
2027 break;
2028
Keiji Costantini84cfce92014-03-01 16:28:152029 case 0x3: /* EBML lacing */
2030 {
Luca Barbato2d0e7712012-09-16 23:58:322031 uint64_t num;
Luca Barbato8a96df72013-03-28 10:52:522032 uint64_t total;
Luca Barbato2d0e7712012-09-16 23:58:322033 n = matroska_ebmlnum_uint(matroska, data, size, &num);
2034 if (n < 0) {
2035 av_log(matroska->ctx, AV_LOG_INFO,
2036 "EBML block data error\n");
2037 res = n;
2038 break;
2039 }
2040 data += n;
2041 size -= n;
2042 total = lace_size[0] = num;
2043 for (n = 1; res == 0 && n < *laces - 1; n++) {
2044 int64_t snum;
2045 int r;
2046 r = matroska_ebmlnum_sint(matroska, data, size, &snum);
2047 if (r < 0) {
2048 av_log(matroska->ctx, AV_LOG_INFO,
2049 "EBML block data error\n");
2050 res = r;
2051 break;
2052 }
Keiji Costantini84cfce92014-03-01 16:28:152053 data += r;
2054 size -= r;
Luca Barbato2d0e7712012-09-16 23:58:322055 lace_size[n] = lace_size[n - 1] + snum;
Keiji Costantini84cfce92014-03-01 16:28:152056 total += lace_size[n];
Luca Barbato2d0e7712012-09-16 23:58:322057 }
2058 if (size <= total) {
2059 res = AVERROR_INVALIDDATA;
2060 break;
2061 }
2062 lace_size[*laces - 1] = size - total;
2063 break;
2064 }
2065 }
2066
2067 *buf = data;
2068 *lace_buf = lace_size;
Dale Curtisfc43c192013-03-27 21:02:032069 *buf_size = size;
Luca Barbato2d0e7712012-09-16 23:58:322070
2071 return res;
2072}
2073
Luca Barbatoc831ebf2012-09-16 23:28:132074static int matroska_parse_rm_audio(MatroskaDemuxContext *matroska,
Keiji Costantini84cfce92014-03-01 16:28:152075 MatroskaTrack *track, AVStream *st,
2076 uint8_t *data, int size, uint64_t timecode,
2077 uint64_t duration, int64_t pos)
Luca Barbatoc831ebf2012-09-16 23:28:132078{
2079 int a = st->codec->block_align;
2080 int sps = track->audio.sub_packet_size;
2081 int cfs = track->audio.coded_framesize;
Keiji Costantini84cfce92014-03-01 16:28:152082 int h = track->audio.sub_packet_h;
2083 int y = track->audio.sub_packet_cnt;
2084 int w = track->audio.frame_size;
Luca Barbatoc831ebf2012-09-16 23:28:132085 int x;
2086
2087 if (!track->audio.pkt_cnt) {
2088 if (track->audio.sub_packet_cnt == 0)
2089 track->audio.buf_timecode = timecode;
2090 if (st->codec->codec_id == AV_CODEC_ID_RA_288) {
2091 if (size < cfs * h / 2) {
2092 av_log(matroska->ctx, AV_LOG_ERROR,
2093 "Corrupt int4 RM-style audio packet size\n");
2094 return AVERROR_INVALIDDATA;
2095 }
Keiji Costantini84cfce92014-03-01 16:28:152096 for (x = 0; x < h / 2; x++)
2097 memcpy(track->audio.buf + x * 2 * w + y * cfs,
2098 data + x * cfs, cfs);
Luca Barbatoc831ebf2012-09-16 23:28:132099 } else if (st->codec->codec_id == AV_CODEC_ID_SIPR) {
2100 if (size < w) {
2101 av_log(matroska->ctx, AV_LOG_ERROR,
2102 "Corrupt sipr RM-style audio packet size\n");
2103 return AVERROR_INVALIDDATA;
2104 }
Keiji Costantini84cfce92014-03-01 16:28:152105 memcpy(track->audio.buf + y * w, data, w);
Luca Barbatoc831ebf2012-09-16 23:28:132106 } else {
2107 if (size < sps * w / sps) {
2108 av_log(matroska->ctx, AV_LOG_ERROR,
2109 "Corrupt generic RM-style audio packet size\n");
2110 return AVERROR_INVALIDDATA;
2111 }
Keiji Costantini84cfce92014-03-01 16:28:152112 for (x = 0; x < w / sps; x++)
2113 memcpy(track->audio.buf +
2114 sps * (h * x + ((h + 1) / 2) * (y & 1) + (y >> 1)),
2115 data + x * sps, sps);
Luca Barbatoc831ebf2012-09-16 23:28:132116 }
2117
2118 if (++track->audio.sub_packet_cnt >= h) {
2119 if (st->codec->codec_id == AV_CODEC_ID_SIPR)
2120 ff_rm_reorder_sipr_data(track->audio.buf, h, w);
2121 track->audio.sub_packet_cnt = 0;
Keiji Costantini84cfce92014-03-01 16:28:152122 track->audio.pkt_cnt = h * w / a;
Luca Barbatoc831ebf2012-09-16 23:28:132123 }
2124 }
2125
2126 while (track->audio.pkt_cnt) {
2127 AVPacket *pkt = av_mallocz(sizeof(AVPacket));
2128 av_new_packet(pkt, a);
Keiji Costantini84cfce92014-03-01 16:28:152129 memcpy(pkt->data,
2130 track->audio.buf + a * (h * w / a - track->audio.pkt_cnt--),
2131 a);
2132 pkt->pts = track->audio.buf_timecode;
Luca Barbatoc831ebf2012-09-16 23:28:132133 track->audio.buf_timecode = AV_NOPTS_VALUE;
Keiji Costantini84cfce92014-03-01 16:28:152134 pkt->pos = pos;
2135 pkt->stream_index = st->index;
2136 dynarray_add(&matroska->packets, &matroska->num_packets, pkt);
Luca Barbatoc831ebf2012-09-16 23:28:132137 }
2138
2139 return 0;
2140}
Anton Khirnov9b6f47c2013-05-27 07:44:272141
2142/* reconstruct full wavpack blocks from mangled matroska ones */
2143static int matroska_parse_wavpack(MatroskaTrack *track, uint8_t *src,
2144 uint8_t **pdst, int *size)
2145{
2146 uint8_t *dst = NULL;
2147 int dstlen = 0;
2148 int srclen = *size;
2149 uint32_t samples;
2150 uint16_t ver;
2151 int ret, offset = 0;
2152
2153 if (srclen < 12 || track->stream->codec->extradata_size < 2)
2154 return AVERROR_INVALIDDATA;
2155
2156 ver = AV_RL16(track->stream->codec->extradata);
2157
2158 samples = AV_RL32(src);
2159 src += 4;
2160 srclen -= 4;
2161
2162 while (srclen >= 8) {
2163 int multiblock;
2164 uint32_t blocksize;
2165 uint8_t *tmp;
2166
2167 uint32_t flags = AV_RL32(src);
2168 uint32_t crc = AV_RL32(src + 4);
2169 src += 8;
2170 srclen -= 8;
2171
2172 multiblock = (flags & 0x1800) != 0x1800;
2173 if (multiblock) {
2174 if (srclen < 4) {
2175 ret = AVERROR_INVALIDDATA;
2176 goto fail;
2177 }
2178 blocksize = AV_RL32(src);
Keiji Costantini84cfce92014-03-01 16:28:152179 src += 4;
2180 srclen -= 4;
Anton Khirnov9b6f47c2013-05-27 07:44:272181 } else
2182 blocksize = srclen;
2183
2184 if (blocksize > srclen) {
2185 ret = AVERROR_INVALIDDATA;
2186 goto fail;
2187 }
2188
2189 tmp = av_realloc(dst, dstlen + blocksize + 32);
2190 if (!tmp) {
2191 ret = AVERROR(ENOMEM);
2192 goto fail;
2193 }
2194 dst = tmp;
2195 dstlen += blocksize + 32;
2196
Keiji Costantini84cfce92014-03-01 16:28:152197 AV_WL32(dst + offset, MKTAG('w', 'v', 'p', 'k')); // tag
2198 AV_WL32(dst + offset + 4, blocksize + 24); // blocksize - 8
2199 AV_WL16(dst + offset + 8, ver); // version
2200 AV_WL16(dst + offset + 10, 0); // track/index_no
2201 AV_WL32(dst + offset + 12, 0); // total samples
2202 AV_WL32(dst + offset + 16, 0); // block index
2203 AV_WL32(dst + offset + 20, samples); // number of samples
2204 AV_WL32(dst + offset + 24, flags); // flags
2205 AV_WL32(dst + offset + 28, crc); // crc
2206 memcpy(dst + offset + 32, src, blocksize); // block data
Anton Khirnov9b6f47c2013-05-27 07:44:272207
2208 src += blocksize;
2209 srclen -= blocksize;
2210 offset += blocksize + 32;
2211 }
2212
2213 *pdst = dst;
2214 *size = dstlen;
2215
2216 return 0;
2217
2218fail:
2219 av_freep(&dst);
2220 return ret;
2221}
2222
Luca Barbatoc831ebf2012-09-16 23:28:132223static int matroska_parse_frame(MatroskaDemuxContext *matroska,
Keiji Costantini84cfce92014-03-01 16:28:152224 MatroskaTrack *track, AVStream *st,
Luca Barbatoc831ebf2012-09-16 23:28:132225 uint8_t *data, int pkt_size,
2226 uint64_t timecode, uint64_t duration,
2227 int64_t pos, int is_keyframe)
2228{
2229 MatroskaTrackEncoding *encodings = track->encodings.elem;
2230 uint8_t *pkt_data = data;
2231 int offset = 0, res;
2232 AVPacket *pkt;
2233
2234 if (encodings && encodings->scope & 1) {
2235 res = matroska_decode_buffer(&pkt_data, &pkt_size, track);
2236 if (res < 0)
2237 return res;
2238 }
2239
Anton Khirnov9b6f47c2013-05-27 07:44:272240 if (st->codec->codec_id == AV_CODEC_ID_WAVPACK) {
2241 uint8_t *wv_data;
2242 res = matroska_parse_wavpack(track, pkt_data, &wv_data, &pkt_size);
2243 if (res < 0) {
Keiji Costantini84cfce92014-03-01 16:28:152244 av_log(matroska->ctx, AV_LOG_ERROR,
2245 "Error parsing a wavpack block.\n");
Anton Khirnov9b6f47c2013-05-27 07:44:272246 goto fail;
2247 }
2248 if (pkt_data != data)
2249 av_freep(&pkt_data);
2250 pkt_data = wv_data;
2251 }
2252
Luca Barbatoc831ebf2012-09-16 23:28:132253 if (st->codec->codec_id == AV_CODEC_ID_PRORES)
2254 offset = 8;
2255
2256 pkt = av_mallocz(sizeof(AVPacket));
2257 /* XXX: prevent data copy... */
2258 if (av_new_packet(pkt, pkt_size + offset) < 0) {
2259 av_free(pkt);
2260 return AVERROR(ENOMEM);
2261 }
2262
2263 if (st->codec->codec_id == AV_CODEC_ID_PRORES) {
2264 uint8_t *buf = pkt->data;
2265 bytestream_put_be32(&buf, pkt_size);
2266 bytestream_put_be32(&buf, MKBETAG('i', 'c', 'p', 'f'));
2267 }
2268
2269 memcpy(pkt->data + offset, pkt_data, pkt_size);
2270
2271 if (pkt_data != data)
2272 av_free(pkt_data);
2273
Keiji Costantini84cfce92014-03-01 16:28:152274 pkt->flags = is_keyframe;
Luca Barbatoc831ebf2012-09-16 23:28:132275 pkt->stream_index = st->index;
2276
2277 if (track->ms_compat)
2278 pkt->dts = timecode;
2279 else
2280 pkt->pts = timecode;
2281 pkt->pos = pos;
2282 if (st->codec->codec_id == AV_CODEC_ID_TEXT)
2283 pkt->convergence_duration = duration;
2284 else if (track->type != MATROSKA_TRACK_TYPE_SUBTITLE)
2285 pkt->duration = duration;
2286
2287 if (st->codec->codec_id == AV_CODEC_ID_SSA)
2288 matroska_fix_ass_packet(matroska, pkt, duration);
2289
Keiji Costantini84cfce92014-03-01 16:28:152290 if (matroska->prev_pkt &&
2291 timecode != AV_NOPTS_VALUE &&
2292 matroska->prev_pkt->pts == timecode &&
2293 matroska->prev_pkt->stream_index == st->index &&
Luca Barbatoc831ebf2012-09-16 23:28:132294 st->codec->codec_id == AV_CODEC_ID_SSA)
2295 matroska_merge_packets(matroska->prev_pkt, pkt);
2296 else {
Keiji Costantini84cfce92014-03-01 16:28:152297 dynarray_add(&matroska->packets, &matroska->num_packets, pkt);
Luca Barbatoc831ebf2012-09-16 23:28:132298 matroska->prev_pkt = pkt;
2299 }
2300
2301 return 0;
Keiji Costantini84cfce92014-03-01 16:28:152302
Anton Khirnov9b6f47c2013-05-27 07:44:272303fail:
2304 if (pkt_data != data)
2305 av_freep(&pkt_data);
2306 return res;
Luca Barbatoc831ebf2012-09-16 23:28:132307}
2308
Aurelien Jacobsf7b96872008-08-05 00:42:052309static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data,
2310 int size, int64_t pos, uint64_t cluster_time,
Luca Barbato7d843102012-09-17 00:48:022311 uint64_t block_duration, int is_keyframe,
Aurelien Jacobs24c3da12008-09-06 23:39:592312 int64_t cluster_pos)
David Conradb061d892007-06-04 22:10:542313{
Aurelien Jacobsf14a2012008-09-09 11:54:352314 uint64_t timecode = AV_NOPTS_VALUE;
Aurelien Jacobs009ecd52008-08-05 00:40:122315 MatroskaTrack *track;
Aurelien Jacobsa3467f82008-09-06 23:44:292316 int res = 0;
David Conradb061d892007-06-04 22:10:542317 AVStream *st;
David Conradb061d892007-06-04 22:10:542318 int16_t block_time;
2319 uint32_t *lace_size = NULL;
2320 int n, flags, laces = 0;
Luca Barbato7d843102012-09-17 00:48:022321 uint64_t num, duration;
David Conradb061d892007-06-04 22:10:542322
Aurelien Jacobsc1e01132008-08-05 00:42:522323 if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) {
David Conradb061d892007-06-04 22:10:542324 av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
Luca Barbato721af292012-04-30 00:39:312325 return n;
David Conradb061d892007-06-04 22:10:542326 }
2327 data += n;
2328 size -= n;
2329
David Conradb061d892007-06-04 22:10:542330 track = matroska_find_track_by_num(matroska, num);
Ronald S. Bultjed31fb1a2011-10-29 23:17:512331 if (!track || !track->stream) {
David Conradb061d892007-06-04 22:10:542332 av_log(matroska->ctx, AV_LOG_INFO,
Aurelien Jacobs009ecd52008-08-05 00:40:122333 "Invalid stream %"PRIu64" or size %u\n", num, size);
Ami Fischman5dd514a2011-06-29 20:54:492334 return AVERROR_INVALIDDATA;
Ronald S. Bultjed31fb1a2011-10-29 23:17:512335 } else if (size <= 3)
2336 return 0;
Aurelien Jacobs009ecd52008-08-05 00:40:122337 st = track->stream;
Aurelien Jacobs16f97ab2008-08-05 00:41:102338 if (st->discard >= AVDISCARD_ALL)
David Conradb061d892007-06-04 22:10:542339 return res;
David Conradb061d892007-06-04 22:10:542340
Aurelien Jacobs2ce746c2007-06-23 12:32:192341 block_time = AV_RB16(data);
Keiji Costantini84cfce92014-03-01 16:28:152342 data += 2;
2343 flags = *data++;
2344 size -= 3;
David Conradb061d892007-06-04 22:10:542345 if (is_keyframe == -1)
Jean-Daniel Dupascc947f02010-03-31 12:29:582346 is_keyframe = flags & 0x80 ? AV_PKT_FLAG_KEY : 0;
David Conradb061d892007-06-04 22:10:542347
Keiji Costantini84cfce92014-03-01 16:28:152348 if (cluster_time != (uint64_t) -1 &&
2349 (block_time >= 0 || cluster_time >= -block_time)) {
Anton Khirnoveb3b5502014-04-29 10:03:132350 timecode = cluster_time + block_time - track->codec_delay;
Keiji Costantini84cfce92014-03-01 16:28:152351 if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE &&
2352 timecode < track->end_timecode)
Aurelien Jacobs82360e62008-09-09 12:07:102353 is_keyframe = 0; /* overlapping subtitles are not key frame */
Aurelien Jacobsa8fd7e72008-09-12 00:06:062354 if (is_keyframe)
Keiji Costantini84cfce92014-03-01 16:28:152355 av_add_index_entry(st, cluster_pos, timecode, 0, 0,
2356 AVINDEX_KEYFRAME);
Aurelien Jacobsf14a2012008-09-09 11:54:352357 }
2358
Keiji Costantini84cfce92014-03-01 16:28:152359 if (matroska->skip_to_keyframe &&
2360 track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
Aurelien Jacobs20f74662008-09-09 12:01:512361 if (!is_keyframe || timecode < matroska->skip_to_timecode)
David Conradb061d892007-06-04 22:10:542362 return res;
2363 matroska->skip_to_keyframe = 0;
2364 }
2365
Dale Curtisfc43c192013-03-27 21:02:032366 res = matroska_parse_laces(matroska, &data, &size, (flags & 0x06) >> 1,
Luca Barbato2d0e7712012-09-16 23:58:322367 &lace_size, &laces);
David Conradb061d892007-06-04 22:10:542368
Luca Barbatoc831ebf2012-09-16 23:28:132369 if (res)
2370 goto end;
Aurelien Jacobseabb8ba2007-06-04 22:19:172371
Luca Barbato7d843102012-09-17 00:48:022372 if (block_duration != AV_NOPTS_VALUE) {
2373 duration = block_duration / laces;
2374 if (block_duration != duration * laces) {
2375 av_log(matroska->ctx, AV_LOG_WARNING,
2376 "Incorrect block_duration, possibly corrupted container");
2377 }
2378 } else {
Keiji Costantini84cfce92014-03-01 16:28:152379 duration = track->default_duration / matroska->time_scale;
Luca Barbato7d843102012-09-17 00:48:022380 block_duration = duration * laces;
2381 }
2382
2383 if (timecode != AV_NOPTS_VALUE)
2384 track->end_timecode =
2385 FFMAX(track->end_timecode, timecode + block_duration);
2386
Luca Barbatoc831ebf2012-09-16 23:28:132387 for (n = 0; n < laces; n++) {
2388 if ((st->codec->codec_id == AV_CODEC_ID_RA_288 ||
Keiji Costantini84cfce92014-03-01 16:28:152389 st->codec->codec_id == AV_CODEC_ID_COOK ||
2390 st->codec->codec_id == AV_CODEC_ID_SIPR ||
Luca Barbatoc831ebf2012-09-16 23:28:132391 st->codec->codec_id == AV_CODEC_ID_ATRAC3) &&
Keiji Costantini84cfce92014-03-01 16:28:152392 st->codec->block_align && track->audio.sub_packet_size) {
Luca Barbato25a80a92013-03-29 11:51:512393 res = matroska_parse_rm_audio(matroska, track, st, data,
2394 lace_size[n],
Luca Barbatoc831ebf2012-09-16 23:28:132395 timecode, duration, pos);
2396 if (res)
2397 goto end;
Luca Barbatoc831ebf2012-09-16 23:28:132398 } else {
2399 res = matroska_parse_frame(matroska, track, st, data, lace_size[n],
Keiji Costantini84cfce92014-03-01 16:28:152400 timecode, duration, pos,
2401 !n ? is_keyframe : 0);
Luca Barbatoc831ebf2012-09-16 23:28:132402 if (res)
2403 goto end;
David Conradb061d892007-06-04 22:10:542404 }
Luca Barbatoc831ebf2012-09-16 23:28:132405
2406 if (timecode != AV_NOPTS_VALUE)
2407 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
2408 data += lace_size[n];
David Conradb061d892007-06-04 22:10:542409 }
2410
Dale Curtis31168582012-04-13 04:24:042411end:
David Conradb061d892007-06-04 22:10:542412 av_free(lace_size);
Aurelien Jacobsa3467f82008-09-06 23:44:292413 return res;
David Conradb061d892007-06-04 22:10:542414}
2415
Dale Curtis8336eb62012-04-19 18:12:242416static int matroska_parse_cluster_incremental(MatroskaDemuxContext *matroska)
2417{
2418 EbmlList *blocks_list;
2419 MatroskaBlock *blocks;
2420 int i, res;
2421 res = ebml_parse(matroska,
2422 matroska_cluster_incremental_parsing,
2423 &matroska->current_cluster);
2424 if (res == 1) {
2425 /* New Cluster */
2426 if (matroska->current_cluster_pos)
2427 ebml_level_end(matroska);
2428 ebml_free(matroska_cluster, &matroska->current_cluster);
2429 memset(&matroska->current_cluster, 0, sizeof(MatroskaCluster));
2430 matroska->current_cluster_num_blocks = 0;
Keiji Costantini84cfce92014-03-01 16:28:152431 matroska->current_cluster_pos = avio_tell(matroska->ctx->pb);
2432 matroska->prev_pkt = NULL;
Dale Curtis8336eb62012-04-19 18:12:242433 /* sizeof the ID which was already read */
2434 if (matroska->current_id)
2435 matroska->current_cluster_pos -= 4;
2436 res = ebml_parse(matroska,
2437 matroska_clusters_incremental,
2438 &matroska->current_cluster);
2439 /* Try parsing the block again. */
2440 if (res == 1)
2441 res = ebml_parse(matroska,
2442 matroska_cluster_incremental_parsing,
2443 &matroska->current_cluster);
2444 }
2445
2446 if (!res &&
2447 matroska->current_cluster_num_blocks <
Keiji Costantini84cfce92014-03-01 16:28:152448 matroska->current_cluster.blocks.nb_elem) {
Dale Curtis8336eb62012-04-19 18:12:242449 blocks_list = &matroska->current_cluster.blocks;
Keiji Costantini84cfce92014-03-01 16:28:152450 blocks = blocks_list->elem;
Dale Curtis8336eb62012-04-19 18:12:242451
2452 matroska->current_cluster_num_blocks = blocks_list->nb_elem;
Keiji Costantini84cfce92014-03-01 16:28:152453 i = blocks_list->nb_elem - 1;
Dale Curtis8336eb62012-04-19 18:12:242454 if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
2455 int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
2456 if (!blocks[i].non_simple)
2457 blocks[i].duration = AV_NOPTS_VALUE;
Keiji Costantini84cfce92014-03-01 16:28:152458 res = matroska_parse_block(matroska, blocks[i].bin.data,
2459 blocks[i].bin.size, blocks[i].bin.pos,
Dale Curtis8336eb62012-04-19 18:12:242460 matroska->current_cluster.timecode,
2461 blocks[i].duration, is_keyframe,
2462 matroska->current_cluster_pos);
2463 }
2464 }
2465
Keiji Costantini84cfce92014-03-01 16:28:152466 if (res < 0)
2467 matroska->done = 1;
Dale Curtis8336eb62012-04-19 18:12:242468 return res;
2469}
2470
Aurelien Jacobsf7b96872008-08-05 00:42:052471static int matroska_parse_cluster(MatroskaDemuxContext *matroska)
David Conradb061d892007-06-04 22:10:542472{
Aurelien Jacobs209472b2008-08-05 00:41:052473 MatroskaCluster cluster = { 0 };
2474 EbmlList *blocks_list;
2475 MatroskaBlock *blocks;
Aurelien Jacobs24c3da12008-09-06 23:39:592476 int i, res;
Dale Curtis8336eb62012-04-19 18:12:242477 int64_t pos;
Keiji Costantini84cfce92014-03-01 16:28:152478
Dale Curtis8336eb62012-04-19 18:12:242479 if (!matroska->contains_ssa)
2480 return matroska_parse_cluster_incremental(matroska);
2481 pos = avio_tell(matroska->ctx->pb);
Aurelien Jacobsd5e34dc2008-09-28 23:06:252482 matroska->prev_pkt = NULL;
Aurelien Jacobs80702032010-06-11 16:36:512483 if (matroska->current_id)
Aurelien Jacobs8bc98ba2008-08-25 00:09:082484 pos -= 4; /* sizeof the ID which was already read */
Keiji Costantini84cfce92014-03-01 16:28:152485 res = ebml_parse(matroska, matroska_clusters, &cluster);
Aurelien Jacobs209472b2008-08-05 00:41:052486 blocks_list = &cluster.blocks;
Keiji Costantini84cfce92014-03-01 16:28:152487 blocks = blocks_list->elem;
2488 for (i = 0; i < blocks_list->nb_elem && !res; i++)
Aurelien Jacobs37dd2352010-05-25 22:55:122489 if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
Aurelien Jacobs194d4b42009-08-10 18:06:142490 int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
John Stebbins98a7d562011-03-24 23:34:182491 if (!blocks[i].non_simple)
2492 blocks[i].duration = AV_NOPTS_VALUE;
Keiji Costantini84cfce92014-03-01 16:28:152493 res = matroska_parse_block(matroska, blocks[i].bin.data,
2494 blocks[i].bin.size, blocks[i].bin.pos,
2495 cluster.timecode, blocks[i].duration,
2496 is_keyframe, pos);
Aurelien Jacobs194d4b42009-08-10 18:06:142497 }
Aurelien Jacobs209472b2008-08-05 00:41:052498 ebml_free(matroska_cluster, &cluster);
David Conradb061d892007-06-04 22:10:542499 return res;
2500}
2501
Aurelien Jacobsf7b96872008-08-05 00:42:052502static int matroska_read_packet(AVFormatContext *s, AVPacket *pkt)
David Conradb061d892007-06-04 22:10:542503{
2504 MatroskaDemuxContext *matroska = s->priv_data;
Ami Fischman5dd514a2011-06-29 20:54:492505 int ret = 0;
David Conradb061d892007-06-04 22:10:542506
Ami Fischman5dd514a2011-06-29 20:54:492507 while (!ret && matroska_deliver_packet(matroska, pkt)) {
Sean McGovern8835c552013-05-27 22:11:502508 int64_t pos = avio_tell(matroska->ctx->pb);
David Conradb061d892007-06-04 22:10:542509 if (matroska->done)
Aurelien Jacobs9ebeea82009-02-19 21:01:452510 return AVERROR_EOF;
Sean McGovern8835c552013-05-27 22:11:502511 if (matroska_parse_cluster(matroska) < 0)
2512 ret = matroska_resync(matroska, pos);
David Conradb061d892007-06-04 22:10:542513 }
2514
Anton Khirnovbdb939a2012-09-20 18:01:432515 if (ret == AVERROR_INVALIDDATA && pkt->data) {
Luca Barbato0ca46422012-04-30 00:55:512516 pkt->flags |= AV_PKT_FLAG_CORRUPT;
2517 return 0;
2518 }
2519
Ami Fischman5dd514a2011-06-29 20:54:492520 return ret;
David Conradb061d892007-06-04 22:10:542521}
2522
Aurelien Jacobsf7b96872008-08-05 00:42:052523static int matroska_read_seek(AVFormatContext *s, int stream_index,
2524 int64_t timestamp, int flags)
David Conradb061d892007-06-04 22:10:542525{
2526 MatroskaDemuxContext *matroska = s->priv_data;
Aurelien Jacobsc1658252008-09-09 12:10:252527 MatroskaTrack *tracks = matroska->tracks.elem;
David Conradb061d892007-06-04 22:10:542528 AVStream *st = s->streams[stream_index];
Aurelien Jacobsc1658252008-09-09 12:10:252529 int i, index, index_sub, index_min;
David Conradb061d892007-06-04 22:10:542530
Aaron Colwell31ad14c2011-07-09 05:48:432531 /* Parse the CUES now since we need the index data to seek. */
2532 if (matroska->cues_parsing_deferred) {
2533 matroska_parse_cues(matroska);
2534 matroska->cues_parsing_deferred = 0;
2535 }
2536
Aurelien Jacobsa8fd7e72008-09-12 00:06:062537 if (!st->nb_index_entries)
2538 return 0;
2539 timestamp = FFMAX(timestamp, st->index_entries[0].timestamp);
Aurelien Jacobsdfbbbdc2008-08-24 23:57:292540
Aurelien Jacobs6bef5f92008-08-27 19:57:422541 if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
Keiji Costantini84cfce92014-03-01 16:28:152542 avio_seek(s->pb, st->index_entries[st->nb_index_entries - 1].pos,
2543 SEEK_SET);
John Stebbinscdc2c1c2011-07-01 15:57:422544 matroska->current_id = 0;
Aurelien Jacobs0dbddda2008-08-27 19:58:552545 while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
2546 matroska_clear_queue(matroska);
2547 if (matroska_parse_cluster(matroska) < 0)
2548 break;
2549 }
Aurelien Jacobs6bef5f92008-08-27 19:57:422550 }
David Conradb061d892007-06-04 22:10:542551
Aurelien Jacobs243cc4c2007-12-29 18:35:382552 matroska_clear_queue(matroska);
Aurelien Jacobs0f646a22008-08-25 00:15:492553 if (index < 0)
2554 return 0;
Aurelien Jacobs243cc4c2007-12-29 18:35:382555
Aurelien Jacobsc1658252008-09-09 12:10:252556 index_min = index;
Keiji Costantini84cfce92014-03-01 16:28:152557 for (i = 0; i < matroska->tracks.nb_elem; i++) {
2558 tracks[i].audio.pkt_cnt = 0;
Reimar Döffinger46544202011-02-26 11:52:012559 tracks[i].audio.sub_packet_cnt = 0;
Keiji Costantini84cfce92014-03-01 16:28:152560 tracks[i].audio.buf_timecode = AV_NOPTS_VALUE;
2561 tracks[i].end_timecode = 0;
2562 if (tracks[i].type == MATROSKA_TRACK_TYPE_SUBTITLE &&
Michael Niedermayer98569d82012-10-20 14:58:332563 tracks[i].stream->discard != AVDISCARD_ALL) {
Keiji Costantini84cfce92014-03-01 16:28:152564 index_sub = av_index_search_timestamp(
2565 tracks[i].stream, st->index_entries[index].timestamp,
2566 AVSEEK_FLAG_BACKWARD);
2567 if (index_sub >= 0 &&
2568 st->index_entries[index_sub].pos < st->index_entries[index_min].pos &&
2569 st->index_entries[index].timestamp -
2570 st->index_entries[index_sub].timestamp < 30000000000 / matroska->time_scale)
Aurelien Jacobsc1658252008-09-09 12:10:252571 index_min = index_sub;
2572 }
2573 }
2574
Anton Khirnov6b4aa5d2011-02-28 13:57:542575 avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET);
Keiji Costantini84cfce92014-03-01 16:28:152576 matroska->current_id = 0;
David Conradb061d892007-06-04 22:10:542577 matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
Aurelien Jacobs20f74662008-09-09 12:01:512578 matroska->skip_to_timecode = st->index_entries[index].timestamp;
Keiji Costantini84cfce92014-03-01 16:28:152579 matroska->done = 0;
Anton Khirnova2faa952011-10-16 13:03:302580 ff_update_cur_dts(s, st, st->index_entries[index].timestamp);
David Conradb061d892007-06-04 22:10:542581 return 0;
2582}
2583
Aurelien Jacobsf7b96872008-08-05 00:42:052584static int matroska_read_close(AVFormatContext *s)
David Conradb061d892007-06-04 22:10:542585{
2586 MatroskaDemuxContext *matroska = s->priv_data;
Aurelien Jacobs2cbc8812008-08-05 00:40:312587 MatroskaTrack *tracks = matroska->tracks.elem;
Aurelien Jacobs70109c02008-08-05 00:41:132588 int n;
David Conradb061d892007-06-04 22:10:542589
Aurelien Jacobs34c9c1b2007-12-29 18:32:472590 matroska_clear_queue(matroska);
David Conradb061d892007-06-04 22:10:542591
Keiji Costantini84cfce92014-03-01 16:28:152592 for (n = 0; n < matroska->tracks.nb_elem; n++)
Aurelien Jacobs2cbc8812008-08-05 00:40:312593 if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
2594 av_free(tracks[n].audio.buf);
Dale Curtis8336eb62012-04-19 18:12:242595 ebml_free(matroska_cluster, &matroska->current_cluster);
Aurelien Jacobsce6f28b2008-08-05 00:40:582596 ebml_free(matroska_segment, matroska);
David Conradb061d892007-06-04 22:10:542597
2598 return 0;
2599}
2600
Diego Elio Pettenòc6610a22011-01-25 22:03:282601AVInputFormat ff_matroska_demuxer = {
Anton Khirnovdfc2c4d2011-07-16 20:18:122602 .name = "matroska,webm",
Diego Biurrun67742472012-07-24 21:51:412603 .long_name = NULL_IF_CONFIG_SMALL("Matroska / WebM"),
Anton Khirnovdfc2c4d2011-07-16 20:18:122604 .priv_data_size = sizeof(MatroskaDemuxContext),
2605 .read_probe = matroska_probe,
2606 .read_header = matroska_read_header,
2607 .read_packet = matroska_read_packet,
2608 .read_close = matroska_read_close,
2609 .read_seek = matroska_read_seek,
Luca Barbatofa385732014-03-13 21:14:432610 .mime_type = "audio/webm,audio/x-matroska,video/webm,video/x-matroska"
David Conradb061d892007-06-04 22:10:542611};