]> git.sesse.net Git - ffmpeg/blob - libavformat/matroskadec.c
udp: Properly print error from getnameinfo
[ffmpeg] / libavformat / matroskadec.c
1 /*
2  * Matroska file demuxer
3  * Copyright (c) 2003-2008 The Libav Project
4  *
5  * This file is part of Libav.
6  *
7  * Libav is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * Libav is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with Libav; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21
22 /**
23  * @file
24  * Matroska file demuxer
25  * @author Ronald Bultje <rbultje@ronald.bitfreak.net>
26  * @author with a little help from Moritz Bunkus <moritz@bunkus.org>
27  * @author totally reworked by Aurelien Jacobs <aurel@gnuage.org>
28  * @see specs available on the Matroska project page: http://www.matroska.org/
29  */
30
31 #include <stdio.h>
32 #include "avformat.h"
33 #include "internal.h"
34 #include "avio_internal.h"
35 /* For ff_codec_get_id(). */
36 #include "riff.h"
37 #include "isom.h"
38 #include "rm.h"
39 #include "matroska.h"
40 #include "libavcodec/mpeg4audio.h"
41 #include "libavutil/intfloat.h"
42 #include "libavutil/intreadwrite.h"
43 #include "libavutil/avstring.h"
44 #include "libavutil/lzo.h"
45 #include "libavutil/dict.h"
46 #if CONFIG_ZLIB
47 #include <zlib.h>
48 #endif
49 #if CONFIG_BZLIB
50 #include <bzlib.h>
51 #endif
52
53 typedef enum {
54     EBML_NONE,
55     EBML_UINT,
56     EBML_FLOAT,
57     EBML_STR,
58     EBML_UTF8,
59     EBML_BIN,
60     EBML_NEST,
61     EBML_PASS,
62     EBML_STOP,
63     EBML_TYPE_COUNT
64 } EbmlType;
65
66 typedef const struct EbmlSyntax {
67     uint32_t id;
68     EbmlType type;
69     int list_elem_size;
70     int data_offset;
71     union {
72         uint64_t    u;
73         double      f;
74         const char *s;
75         const struct EbmlSyntax *n;
76     } def;
77 } EbmlSyntax;
78
79 typedef struct {
80     int nb_elem;
81     void *elem;
82 } EbmlList;
83
84 typedef struct {
85     int      size;
86     uint8_t *data;
87     int64_t  pos;
88 } EbmlBin;
89
90 typedef struct {
91     uint64_t version;
92     uint64_t max_size;
93     uint64_t id_length;
94     char    *doctype;
95     uint64_t doctype_version;
96 } Ebml;
97
98 typedef struct {
99     uint64_t algo;
100     EbmlBin  settings;
101 } MatroskaTrackCompression;
102
103 typedef struct {
104     uint64_t scope;
105     uint64_t type;
106     MatroskaTrackCompression compression;
107 } MatroskaTrackEncoding;
108
109 typedef struct {
110     double   frame_rate;
111     uint64_t display_width;
112     uint64_t display_height;
113     uint64_t pixel_width;
114     uint64_t pixel_height;
115     uint64_t fourcc;
116 } MatroskaTrackVideo;
117
118 typedef struct {
119     double   samplerate;
120     double   out_samplerate;
121     uint64_t bitdepth;
122     uint64_t channels;
123
124     /* real audio header (extracted from extradata) */
125     int      coded_framesize;
126     int      sub_packet_h;
127     int      frame_size;
128     int      sub_packet_size;
129     int      sub_packet_cnt;
130     int      pkt_cnt;
131     uint64_t buf_timecode;
132     uint8_t *buf;
133 } MatroskaTrackAudio;
134
135 typedef struct {
136     uint64_t num;
137     uint64_t uid;
138     uint64_t type;
139     char    *name;
140     char    *codec_id;
141     EbmlBin  codec_priv;
142     char    *language;
143     double time_scale;
144     uint64_t default_duration;
145     uint64_t flag_default;
146     uint64_t flag_forced;
147     MatroskaTrackVideo video;
148     MatroskaTrackAudio audio;
149     EbmlList encodings;
150
151     AVStream *stream;
152     int64_t end_timecode;
153     int ms_compat;
154 } MatroskaTrack;
155
156 typedef struct {
157     uint64_t uid;
158     char *filename;
159     char *mime;
160     EbmlBin bin;
161
162     AVStream *stream;
163 } MatroskaAttachement;
164
165 typedef struct {
166     uint64_t start;
167     uint64_t end;
168     uint64_t uid;
169     char    *title;
170
171     AVChapter *chapter;
172 } MatroskaChapter;
173
174 typedef struct {
175     uint64_t track;
176     uint64_t pos;
177 } MatroskaIndexPos;
178
179 typedef struct {
180     uint64_t time;
181     EbmlList pos;
182 } MatroskaIndex;
183
184 typedef struct {
185     char *name;
186     char *string;
187     char *lang;
188     uint64_t def;
189     EbmlList sub;
190 } MatroskaTag;
191
192 typedef struct {
193     char    *type;
194     uint64_t typevalue;
195     uint64_t trackuid;
196     uint64_t chapteruid;
197     uint64_t attachuid;
198 } MatroskaTagTarget;
199
200 typedef struct {
201     MatroskaTagTarget target;
202     EbmlList tag;
203 } MatroskaTags;
204
205 typedef struct {
206     uint64_t id;
207     uint64_t pos;
208 } MatroskaSeekhead;
209
210 typedef struct {
211     uint64_t start;
212     uint64_t length;
213 } MatroskaLevel;
214
215 typedef struct {
216     uint64_t timecode;
217     EbmlList blocks;
218 } MatroskaCluster;
219
220 typedef struct {
221     AVFormatContext *ctx;
222
223     /* EBML stuff */
224     int num_levels;
225     MatroskaLevel levels[EBML_MAX_DEPTH];
226     int level_up;
227     uint32_t current_id;
228
229     uint64_t time_scale;
230     double   duration;
231     char    *title;
232     EbmlList tracks;
233     EbmlList attachments;
234     EbmlList chapters;
235     EbmlList index;
236     EbmlList tags;
237     EbmlList seekhead;
238
239     /* byte position of the segment inside the stream */
240     int64_t segment_start;
241
242     /* the packet queue */
243     AVPacket **packets;
244     int num_packets;
245     AVPacket *prev_pkt;
246
247     int done;
248
249     /* What to skip before effectively reading a packet. */
250     int skip_to_keyframe;
251     uint64_t skip_to_timecode;
252
253     /* File has a CUES element, but we defer parsing until it is needed. */
254     int cues_parsing_deferred;
255
256     int current_cluster_num_blocks;
257     int64_t current_cluster_pos;
258     MatroskaCluster current_cluster;
259
260     /* File has SSA subtitles which prevent incremental cluster parsing. */
261     int contains_ssa;
262 } MatroskaDemuxContext;
263
264 typedef struct {
265     uint64_t duration;
266     int64_t  reference;
267     uint64_t non_simple;
268     EbmlBin  bin;
269 } MatroskaBlock;
270
271 static EbmlSyntax ebml_header[] = {
272     { EBML_ID_EBMLREADVERSION,        EBML_UINT, 0, offsetof(Ebml,version), {.u=EBML_VERSION} },
273     { EBML_ID_EBMLMAXSIZELENGTH,      EBML_UINT, 0, offsetof(Ebml,max_size), {.u=8} },
274     { EBML_ID_EBMLMAXIDLENGTH,        EBML_UINT, 0, offsetof(Ebml,id_length), {.u=4} },
275     { EBML_ID_DOCTYPE,                EBML_STR,  0, offsetof(Ebml,doctype), {.s="(none)"} },
276     { EBML_ID_DOCTYPEREADVERSION,     EBML_UINT, 0, offsetof(Ebml,doctype_version), {.u=1} },
277     { EBML_ID_EBMLVERSION,            EBML_NONE },
278     { EBML_ID_DOCTYPEVERSION,         EBML_NONE },
279     { 0 }
280 };
281
282 static EbmlSyntax ebml_syntax[] = {
283     { EBML_ID_HEADER,                 EBML_NEST, 0, 0, {.n=ebml_header} },
284     { 0 }
285 };
286
287 static EbmlSyntax matroska_info[] = {
288     { MATROSKA_ID_TIMECODESCALE,      EBML_UINT,  0, offsetof(MatroskaDemuxContext,time_scale), {.u=1000000} },
289     { MATROSKA_ID_DURATION,           EBML_FLOAT, 0, offsetof(MatroskaDemuxContext,duration) },
290     { MATROSKA_ID_TITLE,              EBML_UTF8,  0, offsetof(MatroskaDemuxContext,title) },
291     { MATROSKA_ID_WRITINGAPP,         EBML_NONE },
292     { MATROSKA_ID_MUXINGAPP,          EBML_NONE },
293     { MATROSKA_ID_DATEUTC,            EBML_NONE },
294     { MATROSKA_ID_SEGMENTUID,         EBML_NONE },
295     { 0 }
296 };
297
298 static EbmlSyntax matroska_track_video[] = {
299     { MATROSKA_ID_VIDEOFRAMERATE,     EBML_FLOAT,0, offsetof(MatroskaTrackVideo,frame_rate) },
300     { MATROSKA_ID_VIDEODISPLAYWIDTH,  EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_width) },
301     { MATROSKA_ID_VIDEODISPLAYHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_height) },
302     { MATROSKA_ID_VIDEOPIXELWIDTH,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_width) },
303     { MATROSKA_ID_VIDEOPIXELHEIGHT,   EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_height) },
304     { MATROSKA_ID_VIDEOCOLORSPACE,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,fourcc) },
305     { MATROSKA_ID_VIDEOPIXELCROPB,    EBML_NONE },
306     { MATROSKA_ID_VIDEOPIXELCROPT,    EBML_NONE },
307     { MATROSKA_ID_VIDEOPIXELCROPL,    EBML_NONE },
308     { MATROSKA_ID_VIDEOPIXELCROPR,    EBML_NONE },
309     { MATROSKA_ID_VIDEODISPLAYUNIT,   EBML_NONE },
310     { MATROSKA_ID_VIDEOFLAGINTERLACED,EBML_NONE },
311     { MATROSKA_ID_VIDEOSTEREOMODE,    EBML_NONE },
312     { MATROSKA_ID_VIDEOASPECTRATIO,   EBML_NONE },
313     { 0 }
314 };
315
316 static EbmlSyntax matroska_track_audio[] = {
317     { MATROSKA_ID_AUDIOSAMPLINGFREQ,  EBML_FLOAT,0, offsetof(MatroskaTrackAudio,samplerate), {.f=8000.0} },
318     { MATROSKA_ID_AUDIOOUTSAMPLINGFREQ,EBML_FLOAT,0,offsetof(MatroskaTrackAudio,out_samplerate) },
319     { MATROSKA_ID_AUDIOBITDEPTH,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,bitdepth) },
320     { MATROSKA_ID_AUDIOCHANNELS,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,channels), {.u=1} },
321     { 0 }
322 };
323
324 static EbmlSyntax matroska_track_encoding_compression[] = {
325     { MATROSKA_ID_ENCODINGCOMPALGO,   EBML_UINT, 0, offsetof(MatroskaTrackCompression,algo), {.u=0} },
326     { MATROSKA_ID_ENCODINGCOMPSETTINGS,EBML_BIN, 0, offsetof(MatroskaTrackCompression,settings) },
327     { 0 }
328 };
329
330 static EbmlSyntax matroska_track_encoding[] = {
331     { MATROSKA_ID_ENCODINGSCOPE,      EBML_UINT, 0, offsetof(MatroskaTrackEncoding,scope), {.u=1} },
332     { MATROSKA_ID_ENCODINGTYPE,       EBML_UINT, 0, offsetof(MatroskaTrackEncoding,type), {.u=0} },
333     { MATROSKA_ID_ENCODINGCOMPRESSION,EBML_NEST, 0, offsetof(MatroskaTrackEncoding,compression), {.n=matroska_track_encoding_compression} },
334     { MATROSKA_ID_ENCODINGORDER,      EBML_NONE },
335     { 0 }
336 };
337
338 static EbmlSyntax matroska_track_encodings[] = {
339     { MATROSKA_ID_TRACKCONTENTENCODING, EBML_NEST, sizeof(MatroskaTrackEncoding), offsetof(MatroskaTrack,encodings), {.n=matroska_track_encoding} },
340     { 0 }
341 };
342
343 static EbmlSyntax matroska_track[] = {
344     { MATROSKA_ID_TRACKNUMBER,          EBML_UINT, 0, offsetof(MatroskaTrack,num) },
345     { MATROSKA_ID_TRACKNAME,            EBML_UTF8, 0, offsetof(MatroskaTrack,name) },
346     { MATROSKA_ID_TRACKUID,             EBML_UINT, 0, offsetof(MatroskaTrack,uid) },
347     { MATROSKA_ID_TRACKTYPE,            EBML_UINT, 0, offsetof(MatroskaTrack,type) },
348     { MATROSKA_ID_CODECID,              EBML_STR,  0, offsetof(MatroskaTrack,codec_id) },
349     { MATROSKA_ID_CODECPRIVATE,         EBML_BIN,  0, offsetof(MatroskaTrack,codec_priv) },
350     { MATROSKA_ID_TRACKLANGUAGE,        EBML_UTF8, 0, offsetof(MatroskaTrack,language), {.s="eng"} },
351     { MATROSKA_ID_TRACKDEFAULTDURATION, EBML_UINT, 0, offsetof(MatroskaTrack,default_duration) },
352     { MATROSKA_ID_TRACKTIMECODESCALE,   EBML_FLOAT,0, offsetof(MatroskaTrack,time_scale), {.f=1.0} },
353     { MATROSKA_ID_TRACKFLAGDEFAULT,     EBML_UINT, 0, offsetof(MatroskaTrack,flag_default), {.u=1} },
354     { MATROSKA_ID_TRACKFLAGFORCED,      EBML_UINT, 0, offsetof(MatroskaTrack,flag_forced), {.u=0} },
355     { MATROSKA_ID_TRACKVIDEO,           EBML_NEST, 0, offsetof(MatroskaTrack,video), {.n=matroska_track_video} },
356     { MATROSKA_ID_TRACKAUDIO,           EBML_NEST, 0, offsetof(MatroskaTrack,audio), {.n=matroska_track_audio} },
357     { MATROSKA_ID_TRACKCONTENTENCODINGS,EBML_NEST, 0, 0, {.n=matroska_track_encodings} },
358     { MATROSKA_ID_TRACKFLAGENABLED,     EBML_NONE },
359     { MATROSKA_ID_TRACKFLAGLACING,      EBML_NONE },
360     { MATROSKA_ID_CODECNAME,            EBML_NONE },
361     { MATROSKA_ID_CODECDECODEALL,       EBML_NONE },
362     { MATROSKA_ID_CODECINFOURL,         EBML_NONE },
363     { MATROSKA_ID_CODECDOWNLOADURL,     EBML_NONE },
364     { MATROSKA_ID_TRACKMINCACHE,        EBML_NONE },
365     { MATROSKA_ID_TRACKMAXCACHE,        EBML_NONE },
366     { MATROSKA_ID_TRACKMAXBLKADDID,     EBML_NONE },
367     { 0 }
368 };
369
370 static EbmlSyntax matroska_tracks[] = {
371     { MATROSKA_ID_TRACKENTRY,         EBML_NEST, sizeof(MatroskaTrack), offsetof(MatroskaDemuxContext,tracks), {.n=matroska_track} },
372     { 0 }
373 };
374
375 static EbmlSyntax matroska_attachment[] = {
376     { MATROSKA_ID_FILEUID,            EBML_UINT, 0, offsetof(MatroskaAttachement,uid) },
377     { MATROSKA_ID_FILENAME,           EBML_UTF8, 0, offsetof(MatroskaAttachement,filename) },
378     { MATROSKA_ID_FILEMIMETYPE,       EBML_STR,  0, offsetof(MatroskaAttachement,mime) },
379     { MATROSKA_ID_FILEDATA,           EBML_BIN,  0, offsetof(MatroskaAttachement,bin) },
380     { MATROSKA_ID_FILEDESC,           EBML_NONE },
381     { 0 }
382 };
383
384 static EbmlSyntax matroska_attachments[] = {
385     { MATROSKA_ID_ATTACHEDFILE,       EBML_NEST, sizeof(MatroskaAttachement), offsetof(MatroskaDemuxContext,attachments), {.n=matroska_attachment} },
386     { 0 }
387 };
388
389 static EbmlSyntax matroska_chapter_display[] = {
390     { MATROSKA_ID_CHAPSTRING,         EBML_UTF8, 0, offsetof(MatroskaChapter,title) },
391     { MATROSKA_ID_CHAPLANG,           EBML_NONE },
392     { 0 }
393 };
394
395 static EbmlSyntax matroska_chapter_entry[] = {
396     { MATROSKA_ID_CHAPTERTIMESTART,   EBML_UINT, 0, offsetof(MatroskaChapter,start), {.u=AV_NOPTS_VALUE} },
397     { MATROSKA_ID_CHAPTERTIMEEND,     EBML_UINT, 0, offsetof(MatroskaChapter,end), {.u=AV_NOPTS_VALUE} },
398     { MATROSKA_ID_CHAPTERUID,         EBML_UINT, 0, offsetof(MatroskaChapter,uid) },
399     { MATROSKA_ID_CHAPTERDISPLAY,     EBML_NEST, 0, 0, {.n=matroska_chapter_display} },
400     { MATROSKA_ID_CHAPTERFLAGHIDDEN,  EBML_NONE },
401     { MATROSKA_ID_CHAPTERFLAGENABLED, EBML_NONE },
402     { MATROSKA_ID_CHAPTERPHYSEQUIV,   EBML_NONE },
403     { MATROSKA_ID_CHAPTERATOM,        EBML_NONE },
404     { 0 }
405 };
406
407 static EbmlSyntax matroska_chapter[] = {
408     { MATROSKA_ID_CHAPTERATOM,        EBML_NEST, sizeof(MatroskaChapter), offsetof(MatroskaDemuxContext,chapters), {.n=matroska_chapter_entry} },
409     { MATROSKA_ID_EDITIONUID,         EBML_NONE },
410     { MATROSKA_ID_EDITIONFLAGHIDDEN,  EBML_NONE },
411     { MATROSKA_ID_EDITIONFLAGDEFAULT, EBML_NONE },
412     { MATROSKA_ID_EDITIONFLAGORDERED, EBML_NONE },
413     { 0 }
414 };
415
416 static EbmlSyntax matroska_chapters[] = {
417     { MATROSKA_ID_EDITIONENTRY,       EBML_NEST, 0, 0, {.n=matroska_chapter} },
418     { 0 }
419 };
420
421 static EbmlSyntax matroska_index_pos[] = {
422     { MATROSKA_ID_CUETRACK,           EBML_UINT, 0, offsetof(MatroskaIndexPos,track) },
423     { MATROSKA_ID_CUECLUSTERPOSITION, EBML_UINT, 0, offsetof(MatroskaIndexPos,pos)   },
424     { MATROSKA_ID_CUEBLOCKNUMBER,     EBML_NONE },
425     { 0 }
426 };
427
428 static EbmlSyntax matroska_index_entry[] = {
429     { MATROSKA_ID_CUETIME,            EBML_UINT, 0, offsetof(MatroskaIndex,time) },
430     { MATROSKA_ID_CUETRACKPOSITION,   EBML_NEST, sizeof(MatroskaIndexPos), offsetof(MatroskaIndex,pos), {.n=matroska_index_pos} },
431     { 0 }
432 };
433
434 static EbmlSyntax matroska_index[] = {
435     { MATROSKA_ID_POINTENTRY,         EBML_NEST, sizeof(MatroskaIndex), offsetof(MatroskaDemuxContext,index), {.n=matroska_index_entry} },
436     { 0 }
437 };
438
439 static EbmlSyntax matroska_simpletag[] = {
440     { MATROSKA_ID_TAGNAME,            EBML_UTF8, 0, offsetof(MatroskaTag,name) },
441     { MATROSKA_ID_TAGSTRING,          EBML_UTF8, 0, offsetof(MatroskaTag,string) },
442     { MATROSKA_ID_TAGLANG,            EBML_STR,  0, offsetof(MatroskaTag,lang), {.s="und"} },
443     { MATROSKA_ID_TAGDEFAULT,         EBML_UINT, 0, offsetof(MatroskaTag,def) },
444     { MATROSKA_ID_TAGDEFAULT_BUG,     EBML_UINT, 0, offsetof(MatroskaTag,def) },
445     { MATROSKA_ID_SIMPLETAG,          EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTag,sub), {.n=matroska_simpletag} },
446     { 0 }
447 };
448
449 static EbmlSyntax matroska_tagtargets[] = {
450     { MATROSKA_ID_TAGTARGETS_TYPE,      EBML_STR,  0, offsetof(MatroskaTagTarget,type) },
451     { MATROSKA_ID_TAGTARGETS_TYPEVALUE, EBML_UINT, 0, offsetof(MatroskaTagTarget,typevalue), {.u=50} },
452     { MATROSKA_ID_TAGTARGETS_TRACKUID,  EBML_UINT, 0, offsetof(MatroskaTagTarget,trackuid) },
453     { MATROSKA_ID_TAGTARGETS_CHAPTERUID,EBML_UINT, 0, offsetof(MatroskaTagTarget,chapteruid) },
454     { MATROSKA_ID_TAGTARGETS_ATTACHUID, EBML_UINT, 0, offsetof(MatroskaTagTarget,attachuid) },
455     { 0 }
456 };
457
458 static EbmlSyntax matroska_tag[] = {
459     { MATROSKA_ID_SIMPLETAG,          EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTags,tag), {.n=matroska_simpletag} },
460     { MATROSKA_ID_TAGTARGETS,         EBML_NEST, 0, offsetof(MatroskaTags,target), {.n=matroska_tagtargets} },
461     { 0 }
462 };
463
464 static EbmlSyntax matroska_tags[] = {
465     { MATROSKA_ID_TAG,                EBML_NEST, sizeof(MatroskaTags), offsetof(MatroskaDemuxContext,tags), {.n=matroska_tag} },
466     { 0 }
467 };
468
469 static EbmlSyntax matroska_seekhead_entry[] = {
470     { MATROSKA_ID_SEEKID,             EBML_UINT, 0, offsetof(MatroskaSeekhead,id) },
471     { MATROSKA_ID_SEEKPOSITION,       EBML_UINT, 0, offsetof(MatroskaSeekhead,pos), {.u=-1} },
472     { 0 }
473 };
474
475 static EbmlSyntax matroska_seekhead[] = {
476     { MATROSKA_ID_SEEKENTRY,          EBML_NEST, sizeof(MatroskaSeekhead), offsetof(MatroskaDemuxContext,seekhead), {.n=matroska_seekhead_entry} },
477     { 0 }
478 };
479
480 static EbmlSyntax matroska_segment[] = {
481     { MATROSKA_ID_INFO,           EBML_NEST, 0, 0, {.n=matroska_info       } },
482     { MATROSKA_ID_TRACKS,         EBML_NEST, 0, 0, {.n=matroska_tracks     } },
483     { MATROSKA_ID_ATTACHMENTS,    EBML_NEST, 0, 0, {.n=matroska_attachments} },
484     { MATROSKA_ID_CHAPTERS,       EBML_NEST, 0, 0, {.n=matroska_chapters   } },
485     { MATROSKA_ID_CUES,           EBML_NEST, 0, 0, {.n=matroska_index      } },
486     { MATROSKA_ID_TAGS,           EBML_NEST, 0, 0, {.n=matroska_tags       } },
487     { MATROSKA_ID_SEEKHEAD,       EBML_NEST, 0, 0, {.n=matroska_seekhead   } },
488     { MATROSKA_ID_CLUSTER,        EBML_STOP },
489     { 0 }
490 };
491
492 static EbmlSyntax matroska_segments[] = {
493     { MATROSKA_ID_SEGMENT,        EBML_NEST, 0, 0, {.n=matroska_segment    } },
494     { 0 }
495 };
496
497 static EbmlSyntax matroska_blockgroup[] = {
498     { MATROSKA_ID_BLOCK,          EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
499     { MATROSKA_ID_SIMPLEBLOCK,    EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
500     { MATROSKA_ID_BLOCKDURATION,  EBML_UINT, 0, offsetof(MatroskaBlock,duration), {.u=AV_NOPTS_VALUE} },
501     { MATROSKA_ID_BLOCKREFERENCE, EBML_UINT, 0, offsetof(MatroskaBlock,reference) },
502     { 1,                          EBML_UINT, 0, offsetof(MatroskaBlock,non_simple), {.u=1} },
503     { 0 }
504 };
505
506 static EbmlSyntax matroska_cluster[] = {
507     { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
508     { MATROSKA_ID_BLOCKGROUP,     EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
509     { MATROSKA_ID_SIMPLEBLOCK,    EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
510     { MATROSKA_ID_CLUSTERPOSITION,EBML_NONE },
511     { MATROSKA_ID_CLUSTERPREVSIZE,EBML_NONE },
512     { 0 }
513 };
514
515 static EbmlSyntax matroska_clusters[] = {
516     { MATROSKA_ID_CLUSTER,        EBML_NEST, 0, 0, {.n=matroska_cluster} },
517     { MATROSKA_ID_INFO,           EBML_NONE },
518     { MATROSKA_ID_CUES,           EBML_NONE },
519     { MATROSKA_ID_TAGS,           EBML_NONE },
520     { MATROSKA_ID_SEEKHEAD,       EBML_NONE },
521     { 0 }
522 };
523
524 static EbmlSyntax matroska_cluster_incremental_parsing[] = {
525     { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
526     { MATROSKA_ID_BLOCKGROUP,     EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
527     { MATROSKA_ID_SIMPLEBLOCK,    EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
528     { MATROSKA_ID_CLUSTERPOSITION,EBML_NONE },
529     { MATROSKA_ID_CLUSTERPREVSIZE,EBML_NONE },
530     { MATROSKA_ID_INFO,           EBML_NONE },
531     { MATROSKA_ID_CUES,           EBML_NONE },
532     { MATROSKA_ID_TAGS,           EBML_NONE },
533     { MATROSKA_ID_SEEKHEAD,       EBML_NONE },
534     { MATROSKA_ID_CLUSTER,        EBML_STOP },
535     { 0 }
536 };
537
538 static EbmlSyntax matroska_cluster_incremental[] = {
539     { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
540     { MATROSKA_ID_BLOCKGROUP,     EBML_STOP },
541     { MATROSKA_ID_SIMPLEBLOCK,    EBML_STOP },
542     { MATROSKA_ID_CLUSTERPOSITION,EBML_NONE },
543     { MATROSKA_ID_CLUSTERPREVSIZE,EBML_NONE },
544     { 0 }
545 };
546
547 static EbmlSyntax matroska_clusters_incremental[] = {
548     { MATROSKA_ID_CLUSTER,        EBML_NEST, 0, 0, {.n=matroska_cluster_incremental} },
549     { MATROSKA_ID_INFO,           EBML_NONE },
550     { MATROSKA_ID_CUES,           EBML_NONE },
551     { MATROSKA_ID_TAGS,           EBML_NONE },
552     { MATROSKA_ID_SEEKHEAD,       EBML_NONE },
553     { 0 }
554 };
555
556 static const char *const matroska_doctypes[] = { "matroska", "webm" };
557
558 /*
559  * Return: Whether we reached the end of a level in the hierarchy or not.
560  */
561 static int ebml_level_end(MatroskaDemuxContext *matroska)
562 {
563     AVIOContext *pb = matroska->ctx->pb;
564     int64_t pos = avio_tell(pb);
565
566     if (matroska->num_levels > 0) {
567         MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
568         if (pos - level->start >= level->length || matroska->current_id) {
569             matroska->num_levels--;
570             return 1;
571         }
572     }
573     return 0;
574 }
575
576 /*
577  * Read: an "EBML number", which is defined as a variable-length
578  * array of bytes. The first byte indicates the length by giving a
579  * number of 0-bits followed by a one. The position of the first
580  * "one" bit inside the first byte indicates the length of this
581  * number.
582  * Returns: number of bytes read, < 0 on error
583  */
584 static int ebml_read_num(MatroskaDemuxContext *matroska, AVIOContext *pb,
585                          int max_size, uint64_t *number)
586 {
587     int read = 1, n = 1;
588     uint64_t total = 0;
589
590     /* The first byte tells us the length in bytes - avio_r8() can normally
591      * return 0, but since that's not a valid first ebmlID byte, we can
592      * use it safely here to catch EOS. */
593     if (!(total = avio_r8(pb))) {
594         /* we might encounter EOS here */
595         if (!pb->eof_reached) {
596             int64_t pos = avio_tell(pb);
597             av_log(matroska->ctx, AV_LOG_ERROR,
598                    "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
599                    pos, pos);
600         }
601         return AVERROR(EIO); /* EOS or actual I/O error */
602     }
603
604     /* get the length of the EBML number */
605     read = 8 - ff_log2_tab[total];
606     if (read > max_size) {
607         int64_t pos = avio_tell(pb) - 1;
608         av_log(matroska->ctx, AV_LOG_ERROR,
609                "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
610                (uint8_t) total, pos, pos);
611         return AVERROR_INVALIDDATA;
612     }
613
614     /* read out length */
615     total ^= 1 << ff_log2_tab[total];
616     while (n++ < read)
617         total = (total << 8) | avio_r8(pb);
618
619     *number = total;
620
621     return read;
622 }
623
624 /**
625  * Read a EBML length value.
626  * This needs special handling for the "unknown length" case which has multiple
627  * encodings.
628  */
629 static int ebml_read_length(MatroskaDemuxContext *matroska, AVIOContext *pb,
630                             uint64_t *number)
631 {
632     int res = ebml_read_num(matroska, pb, 8, number);
633     if (res > 0 && *number + 1 == 1ULL << (7 * res))
634         *number = 0xffffffffffffffULL;
635     return res;
636 }
637
638 /*
639  * Read the next element as an unsigned int.
640  * 0 is success, < 0 is failure.
641  */
642 static int ebml_read_uint(AVIOContext *pb, int size, uint64_t *num)
643 {
644     int n = 0;
645
646     if (size > 8)
647         return AVERROR_INVALIDDATA;
648
649     /* big-endian ordering; build up number */
650     *num = 0;
651     while (n++ < size)
652         *num = (*num << 8) | avio_r8(pb);
653
654     return 0;
655 }
656
657 /*
658  * Read the next element as a float.
659  * 0 is success, < 0 is failure.
660  */
661 static int ebml_read_float(AVIOContext *pb, int size, double *num)
662 {
663     if (size == 0) {
664         *num = 0;
665     } else if (size == 4) {
666         *num = av_int2float(avio_rb32(pb));
667     } else if (size == 8){
668         *num = av_int2double(avio_rb64(pb));
669     } else
670         return AVERROR_INVALIDDATA;
671
672     return 0;
673 }
674
675 /*
676  * Read the next element as an ASCII string.
677  * 0 is success, < 0 is failure.
678  */
679 static int ebml_read_ascii(AVIOContext *pb, int size, char **str)
680 {
681     char *res;
682
683     /* EBML strings are usually not 0-terminated, so we allocate one
684      * byte more, read the string and NULL-terminate it ourselves. */
685     if (!(res = av_malloc(size + 1)))
686         return AVERROR(ENOMEM);
687     if (avio_read(pb, (uint8_t *) res, size) != size) {
688         av_free(res);
689         return AVERROR(EIO);
690     }
691     (res)[size] = '\0';
692     av_free(*str);
693     *str = res;
694
695     return 0;
696 }
697
698 /*
699  * Read the next element as binary data.
700  * 0 is success, < 0 is failure.
701  */
702 static int ebml_read_binary(AVIOContext *pb, int length, EbmlBin *bin)
703 {
704     av_free(bin->data);
705     if (!(bin->data = av_malloc(length)))
706         return AVERROR(ENOMEM);
707
708     bin->size = length;
709     bin->pos  = avio_tell(pb);
710     if (avio_read(pb, bin->data, length) != length) {
711         av_freep(&bin->data);
712         return AVERROR(EIO);
713     }
714
715     return 0;
716 }
717
718 /*
719  * Read the next element, but only the header. The contents
720  * are supposed to be sub-elements which can be read separately.
721  * 0 is success, < 0 is failure.
722  */
723 static int ebml_read_master(MatroskaDemuxContext *matroska, uint64_t length)
724 {
725     AVIOContext *pb = matroska->ctx->pb;
726     MatroskaLevel *level;
727
728     if (matroska->num_levels >= EBML_MAX_DEPTH) {
729         av_log(matroska->ctx, AV_LOG_ERROR,
730                "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
731         return AVERROR(ENOSYS);
732     }
733
734     level = &matroska->levels[matroska->num_levels++];
735     level->start = avio_tell(pb);
736     level->length = length;
737
738     return 0;
739 }
740
741 /*
742  * Read signed/unsigned "EBML" numbers.
743  * Return: number of bytes processed, < 0 on error
744  */
745 static int matroska_ebmlnum_uint(MatroskaDemuxContext *matroska,
746                                  uint8_t *data, uint32_t size, uint64_t *num)
747 {
748     AVIOContext pb;
749     ffio_init_context(&pb, data, size, 0, NULL, NULL, NULL, NULL);
750     return ebml_read_num(matroska, &pb, FFMIN(size, 8), num);
751 }
752
753 /*
754  * Same as above, but signed.
755  */
756 static int matroska_ebmlnum_sint(MatroskaDemuxContext *matroska,
757                                  uint8_t *data, uint32_t size, int64_t *num)
758 {
759     uint64_t unum;
760     int res;
761
762     /* read as unsigned number first */
763     if ((res = matroska_ebmlnum_uint(matroska, data, size, &unum)) < 0)
764         return res;
765
766     /* make signed (weird way) */
767     *num = unum - ((1LL << (7*res - 1)) - 1);
768
769     return res;
770 }
771
772 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
773                            EbmlSyntax *syntax, void *data);
774
775 static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
776                          uint32_t id, void *data)
777 {
778     int i;
779     for (i=0; syntax[i].id; i++)
780         if (id == syntax[i].id)
781             break;
782     if (!syntax[i].id && id == MATROSKA_ID_CLUSTER &&
783         matroska->num_levels > 0 &&
784         matroska->levels[matroska->num_levels-1].length == 0xffffffffffffff)
785         return 0;  // we reached the end of an unknown size cluster
786     if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32)
787         av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%X\n", id);
788     return ebml_parse_elem(matroska, &syntax[i], data);
789 }
790
791 static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
792                       void *data)
793 {
794     if (!matroska->current_id) {
795         uint64_t id;
796         int res = ebml_read_num(matroska, matroska->ctx->pb, 4, &id);
797         if (res < 0)
798             return res;
799         matroska->current_id = id | 1 << 7*res;
800     }
801     return ebml_parse_id(matroska, syntax, matroska->current_id, data);
802 }
803
804 static int ebml_parse_nest(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
805                            void *data)
806 {
807     int i, res = 0;
808
809     for (i=0; syntax[i].id; i++)
810         switch (syntax[i].type) {
811         case EBML_UINT:
812             *(uint64_t *)((char *)data+syntax[i].data_offset) = syntax[i].def.u;
813             break;
814         case EBML_FLOAT:
815             *(double   *)((char *)data+syntax[i].data_offset) = syntax[i].def.f;
816             break;
817         case EBML_STR:
818         case EBML_UTF8:
819             *(char    **)((char *)data+syntax[i].data_offset) = av_strdup(syntax[i].def.s);
820             break;
821         }
822
823     while (!res && !ebml_level_end(matroska))
824         res = ebml_parse(matroska, syntax, data);
825
826     return res;
827 }
828
829 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
830                            EbmlSyntax *syntax, void *data)
831 {
832     static const uint64_t max_lengths[EBML_TYPE_COUNT] = {
833         [EBML_UINT]  = 8,
834         [EBML_FLOAT] = 8,
835         // max. 16 MB for strings
836         [EBML_STR]   = 0x1000000,
837         [EBML_UTF8]  = 0x1000000,
838         // max. 256 MB for binary data
839         [EBML_BIN]   = 0x10000000,
840         // no limits for anything else
841     };
842     AVIOContext *pb = matroska->ctx->pb;
843     uint32_t id = syntax->id;
844     uint64_t length;
845     int res;
846     void *newelem;
847
848     data = (char *)data + syntax->data_offset;
849     if (syntax->list_elem_size) {
850         EbmlList *list = data;
851         newelem = av_realloc(list->elem, (list->nb_elem+1)*syntax->list_elem_size);
852         if (!newelem)
853             return AVERROR(ENOMEM);
854         list->elem = newelem;
855         data = (char*)list->elem + list->nb_elem*syntax->list_elem_size;
856         memset(data, 0, syntax->list_elem_size);
857         list->nb_elem++;
858     }
859
860     if (syntax->type != EBML_PASS && syntax->type != EBML_STOP) {
861         matroska->current_id = 0;
862         if ((res = ebml_read_length(matroska, pb, &length)) < 0)
863             return res;
864         if (max_lengths[syntax->type] && length > max_lengths[syntax->type]) {
865             av_log(matroska->ctx, AV_LOG_ERROR,
866                    "Invalid length 0x%"PRIx64" > 0x%"PRIx64" for syntax element %i\n",
867                    length, max_lengths[syntax->type], syntax->type);
868             return AVERROR_INVALIDDATA;
869         }
870     }
871
872     switch (syntax->type) {
873     case EBML_UINT:  res = ebml_read_uint  (pb, length, data);  break;
874     case EBML_FLOAT: res = ebml_read_float (pb, length, data);  break;
875     case EBML_STR:
876     case EBML_UTF8:  res = ebml_read_ascii (pb, length, data);  break;
877     case EBML_BIN:   res = ebml_read_binary(pb, length, data);  break;
878     case EBML_NEST:  if ((res=ebml_read_master(matroska, length)) < 0)
879                          return res;
880                      if (id == MATROSKA_ID_SEGMENT)
881                          matroska->segment_start = avio_tell(matroska->ctx->pb);
882                      return ebml_parse_nest(matroska, syntax->def.n, data);
883     case EBML_PASS:  return ebml_parse_id(matroska, syntax->def.n, id, data);
884     case EBML_STOP:  return 1;
885     default:         return avio_skip(pb,length)<0 ? AVERROR(EIO) : 0;
886     }
887     if (res == AVERROR_INVALIDDATA)
888         av_log(matroska->ctx, AV_LOG_ERROR, "Invalid element\n");
889     else if (res == AVERROR(EIO))
890         av_log(matroska->ctx, AV_LOG_ERROR, "Read error\n");
891     return res;
892 }
893
894 static void ebml_free(EbmlSyntax *syntax, void *data)
895 {
896     int i, j;
897     for (i=0; syntax[i].id; i++) {
898         void *data_off = (char *)data + syntax[i].data_offset;
899         switch (syntax[i].type) {
900         case EBML_STR:
901         case EBML_UTF8:  av_freep(data_off);                      break;
902         case EBML_BIN:   av_freep(&((EbmlBin *)data_off)->data);  break;
903         case EBML_NEST:
904             if (syntax[i].list_elem_size) {
905                 EbmlList *list = data_off;
906                 char *ptr = list->elem;
907                 for (j=0; j<list->nb_elem; j++, ptr+=syntax[i].list_elem_size)
908                     ebml_free(syntax[i].def.n, ptr);
909                 av_free(list->elem);
910             } else
911                 ebml_free(syntax[i].def.n, data_off);
912         default:  break;
913         }
914     }
915 }
916
917
918 /*
919  * Autodetecting...
920  */
921 static int matroska_probe(AVProbeData *p)
922 {
923     uint64_t total = 0;
924     int len_mask = 0x80, size = 1, n = 1, i;
925
926     /* EBML header? */
927     if (AV_RB32(p->buf) != EBML_ID_HEADER)
928         return 0;
929
930     /* length of header */
931     total = p->buf[4];
932     while (size <= 8 && !(total & len_mask)) {
933         size++;
934         len_mask >>= 1;
935     }
936     if (size > 8)
937       return 0;
938     total &= (len_mask - 1);
939     while (n < size)
940         total = (total << 8) | p->buf[4 + n++];
941
942     /* Does the probe data contain the whole header? */
943     if (p->buf_size < 4 + size + total)
944       return 0;
945
946     /* The header should contain a known document type. For now,
947      * we don't parse the whole header but simply check for the
948      * availability of that array of characters inside the header.
949      * Not fully fool-proof, but good enough. */
950     for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++) {
951         int probelen = strlen(matroska_doctypes[i]);
952         if (total < probelen)
953             continue;
954         for (n = 4+size; n <= 4+size+total-probelen; n++)
955             if (!memcmp(p->buf+n, matroska_doctypes[i], probelen))
956                 return AVPROBE_SCORE_MAX;
957     }
958
959     // probably valid EBML header but no recognized doctype
960     return AVPROBE_SCORE_MAX/2;
961 }
962
963 static MatroskaTrack *matroska_find_track_by_num(MatroskaDemuxContext *matroska,
964                                                  int num)
965 {
966     MatroskaTrack *tracks = matroska->tracks.elem;
967     int i;
968
969     for (i=0; i < matroska->tracks.nb_elem; i++)
970         if (tracks[i].num == num)
971             return &tracks[i];
972
973     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid track number %d\n", num);
974     return NULL;
975 }
976
977 static int matroska_decode_buffer(uint8_t** buf, int* buf_size,
978                                   MatroskaTrack *track)
979 {
980     MatroskaTrackEncoding *encodings = track->encodings.elem;
981     uint8_t* data = *buf;
982     int isize = *buf_size;
983     uint8_t* pkt_data = NULL;
984     uint8_t av_unused *newpktdata;
985     int pkt_size = isize;
986     int result = 0;
987     int olen;
988
989     if (pkt_size >= 10000000)
990         return -1;
991
992     switch (encodings[0].compression.algo) {
993     case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
994         return encodings[0].compression.settings.size;
995     case MATROSKA_TRACK_ENCODING_COMP_LZO:
996         do {
997             olen = pkt_size *= 3;
998             pkt_data = av_realloc(pkt_data, pkt_size+AV_LZO_OUTPUT_PADDING);
999             result = av_lzo1x_decode(pkt_data, &olen, data, &isize);
1000         } while (result==AV_LZO_OUTPUT_FULL && pkt_size<10000000);
1001         if (result)
1002             goto failed;
1003         pkt_size -= olen;
1004         break;
1005 #if CONFIG_ZLIB
1006     case MATROSKA_TRACK_ENCODING_COMP_ZLIB: {
1007         z_stream zstream = {0};
1008         if (inflateInit(&zstream) != Z_OK)
1009             return -1;
1010         zstream.next_in = data;
1011         zstream.avail_in = isize;
1012         do {
1013             pkt_size *= 3;
1014             newpktdata = av_realloc(pkt_data, pkt_size);
1015             if (!newpktdata) {
1016                 inflateEnd(&zstream);
1017                 goto failed;
1018             }
1019             pkt_data = newpktdata;
1020             zstream.avail_out = pkt_size - zstream.total_out;
1021             zstream.next_out = pkt_data + zstream.total_out;
1022             result = inflate(&zstream, Z_NO_FLUSH);
1023         } while (result==Z_OK && pkt_size<10000000);
1024         pkt_size = zstream.total_out;
1025         inflateEnd(&zstream);
1026         if (result != Z_STREAM_END)
1027             goto failed;
1028         break;
1029     }
1030 #endif
1031 #if CONFIG_BZLIB
1032     case MATROSKA_TRACK_ENCODING_COMP_BZLIB: {
1033         bz_stream bzstream = {0};
1034         if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
1035             return -1;
1036         bzstream.next_in = data;
1037         bzstream.avail_in = isize;
1038         do {
1039             pkt_size *= 3;
1040             newpktdata = av_realloc(pkt_data, pkt_size);
1041             if (!newpktdata) {
1042                 BZ2_bzDecompressEnd(&bzstream);
1043                 goto failed;
1044             }
1045             pkt_data = newpktdata;
1046             bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
1047             bzstream.next_out = pkt_data + bzstream.total_out_lo32;
1048             result = BZ2_bzDecompress(&bzstream);
1049         } while (result==BZ_OK && pkt_size<10000000);
1050         pkt_size = bzstream.total_out_lo32;
1051         BZ2_bzDecompressEnd(&bzstream);
1052         if (result != BZ_STREAM_END)
1053             goto failed;
1054         break;
1055     }
1056 #endif
1057     default:
1058         return -1;
1059     }
1060
1061     *buf = pkt_data;
1062     *buf_size = pkt_size;
1063     return 0;
1064  failed:
1065     av_free(pkt_data);
1066     return -1;
1067 }
1068
1069 static void matroska_fix_ass_packet(MatroskaDemuxContext *matroska,
1070                                     AVPacket *pkt, uint64_t display_duration)
1071 {
1072     char *line, *layer, *ptr = pkt->data, *end = ptr+pkt->size;
1073     for (; *ptr!=',' && ptr<end-1; ptr++);
1074     if (*ptr == ',')
1075         layer = ++ptr;
1076     for (; *ptr!=',' && ptr<end-1; ptr++);
1077     if (*ptr == ',') {
1078         int64_t end_pts = pkt->pts + display_duration;
1079         int sc = matroska->time_scale * pkt->pts / 10000000;
1080         int ec = matroska->time_scale * end_pts  / 10000000;
1081         int sh, sm, ss, eh, em, es, len;
1082         sh = sc/360000;  sc -= 360000*sh;
1083         sm = sc/  6000;  sc -=   6000*sm;
1084         ss = sc/   100;  sc -=    100*ss;
1085         eh = ec/360000;  ec -= 360000*eh;
1086         em = ec/  6000;  ec -=   6000*em;
1087         es = ec/   100;  ec -=    100*es;
1088         *ptr++ = '\0';
1089         len = 50 + end-ptr + FF_INPUT_BUFFER_PADDING_SIZE;
1090         if (!(line = av_malloc(len)))
1091             return;
1092         snprintf(line,len,"Dialogue: %s,%d:%02d:%02d.%02d,%d:%02d:%02d.%02d,%s\r\n",
1093                  layer, sh, sm, ss, sc, eh, em, es, ec, ptr);
1094         av_free(pkt->data);
1095         pkt->data = line;
1096         pkt->size = strlen(line);
1097     }
1098 }
1099
1100 static int matroska_merge_packets(AVPacket *out, AVPacket *in)
1101 {
1102     void *newdata = av_realloc(out->data, out->size+in->size);
1103     if (!newdata)
1104         return AVERROR(ENOMEM);
1105     out->data = newdata;
1106     memcpy(out->data+out->size, in->data, in->size);
1107     out->size += in->size;
1108     av_destruct_packet(in);
1109     av_free(in);
1110     return 0;
1111 }
1112
1113 static void matroska_convert_tag(AVFormatContext *s, EbmlList *list,
1114                                  AVDictionary **metadata, char *prefix)
1115 {
1116     MatroskaTag *tags = list->elem;
1117     char key[1024];
1118     int i;
1119
1120     for (i=0; i < list->nb_elem; i++) {
1121         const char *lang = strcmp(tags[i].lang, "und") ? tags[i].lang : NULL;
1122
1123         if (!tags[i].name) {
1124             av_log(s, AV_LOG_WARNING, "Skipping invalid tag with no TagName.\n");
1125             continue;
1126         }
1127         if (prefix)  snprintf(key, sizeof(key), "%s/%s", prefix, tags[i].name);
1128         else         av_strlcpy(key, tags[i].name, sizeof(key));
1129         if (tags[i].def || !lang) {
1130         av_dict_set(metadata, key, tags[i].string, 0);
1131         if (tags[i].sub.nb_elem)
1132             matroska_convert_tag(s, &tags[i].sub, metadata, key);
1133         }
1134         if (lang) {
1135             av_strlcat(key, "-", sizeof(key));
1136             av_strlcat(key, lang, sizeof(key));
1137             av_dict_set(metadata, key, tags[i].string, 0);
1138             if (tags[i].sub.nb_elem)
1139                 matroska_convert_tag(s, &tags[i].sub, metadata, key);
1140         }
1141     }
1142     ff_metadata_conv(metadata, NULL, ff_mkv_metadata_conv);
1143 }
1144
1145 static void matroska_convert_tags(AVFormatContext *s)
1146 {
1147     MatroskaDemuxContext *matroska = s->priv_data;
1148     MatroskaTags *tags = matroska->tags.elem;
1149     int i, j;
1150
1151     for (i=0; i < matroska->tags.nb_elem; i++) {
1152         if (tags[i].target.attachuid) {
1153             MatroskaAttachement *attachment = matroska->attachments.elem;
1154             for (j=0; j<matroska->attachments.nb_elem; j++)
1155                 if (attachment[j].uid == tags[i].target.attachuid
1156                     && attachment[j].stream)
1157                     matroska_convert_tag(s, &tags[i].tag,
1158                                          &attachment[j].stream->metadata, NULL);
1159         } else if (tags[i].target.chapteruid) {
1160             MatroskaChapter *chapter = matroska->chapters.elem;
1161             for (j=0; j<matroska->chapters.nb_elem; j++)
1162                 if (chapter[j].uid == tags[i].target.chapteruid
1163                     && chapter[j].chapter)
1164                     matroska_convert_tag(s, &tags[i].tag,
1165                                          &chapter[j].chapter->metadata, NULL);
1166         } else if (tags[i].target.trackuid) {
1167             MatroskaTrack *track = matroska->tracks.elem;
1168             for (j=0; j<matroska->tracks.nb_elem; j++)
1169                 if (track[j].uid == tags[i].target.trackuid && track[j].stream)
1170                     matroska_convert_tag(s, &tags[i].tag,
1171                                          &track[j].stream->metadata, NULL);
1172         } else {
1173             matroska_convert_tag(s, &tags[i].tag, &s->metadata,
1174                                  tags[i].target.type);
1175         }
1176     }
1177 }
1178
1179 static int matroska_parse_seekhead_entry(MatroskaDemuxContext *matroska, int idx)
1180 {
1181     EbmlList *seekhead_list = &matroska->seekhead;
1182     MatroskaSeekhead *seekhead = seekhead_list->elem;
1183     uint32_t level_up = matroska->level_up;
1184     int64_t before_pos = avio_tell(matroska->ctx->pb);
1185     uint32_t saved_id = matroska->current_id;
1186     MatroskaLevel level;
1187     int64_t offset;
1188     int ret = 0;
1189
1190     if (idx >= seekhead_list->nb_elem
1191             || seekhead[idx].id == MATROSKA_ID_SEEKHEAD
1192             || seekhead[idx].id == MATROSKA_ID_CLUSTER)
1193         return 0;
1194
1195     /* seek */
1196     offset = seekhead[idx].pos + matroska->segment_start;
1197     if (avio_seek(matroska->ctx->pb, offset, SEEK_SET) == offset) {
1198         /* We don't want to lose our seekhead level, so we add
1199          * a dummy. This is a crude hack. */
1200         if (matroska->num_levels == EBML_MAX_DEPTH) {
1201             av_log(matroska->ctx, AV_LOG_INFO,
1202                    "Max EBML element depth (%d) reached, "
1203                    "cannot parse further.\n", EBML_MAX_DEPTH);
1204             ret = AVERROR_INVALIDDATA;
1205         } else {
1206             level.start = 0;
1207             level.length = (uint64_t)-1;
1208             matroska->levels[matroska->num_levels] = level;
1209             matroska->num_levels++;
1210             matroska->current_id = 0;
1211
1212             ret = ebml_parse(matroska, matroska_segment, matroska);
1213
1214             /* remove dummy level */
1215             while (matroska->num_levels) {
1216                 uint64_t length = matroska->levels[--matroska->num_levels].length;
1217                 if (length == (uint64_t)-1)
1218                     break;
1219             }
1220         }
1221     }
1222     /* seek back */
1223     avio_seek(matroska->ctx->pb, before_pos, SEEK_SET);
1224     matroska->level_up = level_up;
1225     matroska->current_id = saved_id;
1226
1227     return ret;
1228 }
1229
1230 static void matroska_execute_seekhead(MatroskaDemuxContext *matroska)
1231 {
1232     EbmlList *seekhead_list = &matroska->seekhead;
1233     int64_t before_pos = avio_tell(matroska->ctx->pb);
1234     int i;
1235
1236     // we should not do any seeking in the streaming case
1237     if (!matroska->ctx->pb->seekable ||
1238         (matroska->ctx->flags & AVFMT_FLAG_IGNIDX))
1239         return;
1240
1241     for (i = 0; i < seekhead_list->nb_elem; i++) {
1242         MatroskaSeekhead *seekhead = seekhead_list->elem;
1243         if (seekhead[i].pos <= before_pos)
1244             continue;
1245
1246         // defer cues parsing until we actually need cue data.
1247         if (seekhead[i].id == MATROSKA_ID_CUES) {
1248             matroska->cues_parsing_deferred = 1;
1249             continue;
1250         }
1251
1252         if (matroska_parse_seekhead_entry(matroska, i) < 0)
1253             break;
1254     }
1255 }
1256
1257 static void matroska_parse_cues(MatroskaDemuxContext *matroska) {
1258     EbmlList *seekhead_list = &matroska->seekhead;
1259     MatroskaSeekhead *seekhead = seekhead_list->elem;
1260     EbmlList *index_list;
1261     MatroskaIndex *index;
1262     int index_scale = 1;
1263     int i, j;
1264
1265     for (i = 0; i < seekhead_list->nb_elem; i++)
1266         if (seekhead[i].id == MATROSKA_ID_CUES)
1267             break;
1268     assert(i <= seekhead_list->nb_elem);
1269
1270     matroska_parse_seekhead_entry(matroska, i);
1271
1272     index_list = &matroska->index;
1273     index = index_list->elem;
1274     if (index_list->nb_elem
1275         && index[0].time > 1E14/matroska->time_scale) {
1276         av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n");
1277         index_scale = matroska->time_scale;
1278     }
1279     for (i = 0; i < index_list->nb_elem; i++) {
1280         EbmlList *pos_list = &index[i].pos;
1281         MatroskaIndexPos *pos = pos_list->elem;
1282         for (j = 0; j < pos_list->nb_elem; j++) {
1283             MatroskaTrack *track = matroska_find_track_by_num(matroska, pos[j].track);
1284             if (track && track->stream)
1285                 av_add_index_entry(track->stream,
1286                                    pos[j].pos + matroska->segment_start,
1287                                    index[i].time/index_scale, 0, 0,
1288                                    AVINDEX_KEYFRAME);
1289         }
1290     }
1291 }
1292
1293 static int matroska_aac_profile(char *codec_id)
1294 {
1295     static const char * const aac_profiles[] = { "MAIN", "LC", "SSR" };
1296     int profile;
1297
1298     for (profile=0; profile<FF_ARRAY_ELEMS(aac_profiles); profile++)
1299         if (strstr(codec_id, aac_profiles[profile]))
1300             break;
1301     return profile + 1;
1302 }
1303
1304 static int matroska_aac_sri(int samplerate)
1305 {
1306     int sri;
1307
1308     for (sri=0; sri<FF_ARRAY_ELEMS(avpriv_mpeg4audio_sample_rates); sri++)
1309         if (avpriv_mpeg4audio_sample_rates[sri] == samplerate)
1310             break;
1311     return sri;
1312 }
1313
1314 static int matroska_read_header(AVFormatContext *s)
1315 {
1316     MatroskaDemuxContext *matroska = s->priv_data;
1317     EbmlList *attachements_list = &matroska->attachments;
1318     MatroskaAttachement *attachements;
1319     EbmlList *chapters_list = &matroska->chapters;
1320     MatroskaChapter *chapters;
1321     MatroskaTrack *tracks;
1322     uint64_t max_start = 0;
1323     Ebml ebml = { 0 };
1324     AVStream *st;
1325     int i, j, res;
1326
1327     matroska->ctx = s;
1328
1329     /* First read the EBML header. */
1330     if (ebml_parse(matroska, ebml_syntax, &ebml)
1331         || ebml.version > EBML_VERSION       || ebml.max_size > sizeof(uint64_t)
1332         || ebml.id_length > sizeof(uint32_t) || ebml.doctype_version > 2) {
1333         av_log(matroska->ctx, AV_LOG_ERROR,
1334                "EBML header using unsupported features\n"
1335                "(EBML version %"PRIu64", doctype %s, doc version %"PRIu64")\n",
1336                ebml.version, ebml.doctype, ebml.doctype_version);
1337         ebml_free(ebml_syntax, &ebml);
1338         return AVERROR_PATCHWELCOME;
1339     }
1340     for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++)
1341         if (!strcmp(ebml.doctype, matroska_doctypes[i]))
1342             break;
1343     if (i >= FF_ARRAY_ELEMS(matroska_doctypes)) {
1344         av_log(s, AV_LOG_WARNING, "Unknown EBML doctype '%s'\n", ebml.doctype);
1345     }
1346     ebml_free(ebml_syntax, &ebml);
1347
1348     /* The next thing is a segment. */
1349     if ((res = ebml_parse(matroska, matroska_segments, matroska)) < 0)
1350         return res;
1351     matroska_execute_seekhead(matroska);
1352
1353     if (!matroska->time_scale)
1354         matroska->time_scale = 1000000;
1355     if (matroska->duration)
1356         matroska->ctx->duration = matroska->duration * matroska->time_scale
1357                                   * 1000 / AV_TIME_BASE;
1358     av_dict_set(&s->metadata, "title", matroska->title, 0);
1359
1360     tracks = matroska->tracks.elem;
1361     for (i=0; i < matroska->tracks.nb_elem; i++) {
1362         MatroskaTrack *track = &tracks[i];
1363         enum CodecID codec_id = CODEC_ID_NONE;
1364         EbmlList *encodings_list = &tracks->encodings;
1365         MatroskaTrackEncoding *encodings = encodings_list->elem;
1366         uint8_t *extradata = NULL;
1367         int extradata_size = 0;
1368         int extradata_offset = 0;
1369         AVIOContext b;
1370
1371         /* Apply some sanity checks. */
1372         if (track->type != MATROSKA_TRACK_TYPE_VIDEO &&
1373             track->type != MATROSKA_TRACK_TYPE_AUDIO &&
1374             track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1375             av_log(matroska->ctx, AV_LOG_INFO,
1376                    "Unknown or unsupported track type %"PRIu64"\n",
1377                    track->type);
1378             continue;
1379         }
1380         if (track->codec_id == NULL)
1381             continue;
1382
1383         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
1384             if (!track->default_duration && track->video.frame_rate > 0)
1385                 track->default_duration = 1000000000/track->video.frame_rate;
1386             if (!track->video.display_width)
1387                 track->video.display_width = track->video.pixel_width;
1388             if (!track->video.display_height)
1389                 track->video.display_height = track->video.pixel_height;
1390         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1391             if (!track->audio.out_samplerate)
1392                 track->audio.out_samplerate = track->audio.samplerate;
1393         }
1394         if (encodings_list->nb_elem > 1) {
1395             av_log(matroska->ctx, AV_LOG_ERROR,
1396                    "Multiple combined encodings not supported");
1397         } else if (encodings_list->nb_elem == 1) {
1398             if (encodings[0].type ||
1399                 (encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP &&
1400 #if CONFIG_ZLIB
1401                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
1402 #endif
1403 #if CONFIG_BZLIB
1404                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
1405 #endif
1406                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_LZO)) {
1407                 encodings[0].scope = 0;
1408                 av_log(matroska->ctx, AV_LOG_ERROR,
1409                        "Unsupported encoding type");
1410             } else if (track->codec_priv.size && encodings[0].scope&2) {
1411                 uint8_t *codec_priv = track->codec_priv.data;
1412                 int offset = matroska_decode_buffer(&track->codec_priv.data,
1413                                                     &track->codec_priv.size,
1414                                                     track);
1415                 if (offset < 0) {
1416                     track->codec_priv.data = NULL;
1417                     track->codec_priv.size = 0;
1418                     av_log(matroska->ctx, AV_LOG_ERROR,
1419                            "Failed to decode codec private data\n");
1420                 } else if (offset > 0) {
1421                     track->codec_priv.data = av_malloc(track->codec_priv.size + offset);
1422                     memcpy(track->codec_priv.data,
1423                            encodings[0].compression.settings.data, offset);
1424                     memcpy(track->codec_priv.data+offset, codec_priv,
1425                            track->codec_priv.size);
1426                     track->codec_priv.size += offset;
1427                 }
1428                 if (codec_priv != track->codec_priv.data)
1429                     av_free(codec_priv);
1430             }
1431         }
1432
1433         for(j=0; ff_mkv_codec_tags[j].id != CODEC_ID_NONE; j++){
1434             if(!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
1435                         strlen(ff_mkv_codec_tags[j].str))){
1436                 codec_id= ff_mkv_codec_tags[j].id;
1437                 break;
1438             }
1439         }
1440
1441         st = track->stream = avformat_new_stream(s, NULL);
1442         if (st == NULL)
1443             return AVERROR(ENOMEM);
1444
1445         if (!strcmp(track->codec_id, "V_MS/VFW/FOURCC")
1446             && track->codec_priv.size >= 40
1447             && track->codec_priv.data != NULL) {
1448             track->ms_compat = 1;
1449             track->video.fourcc = AV_RL32(track->codec_priv.data + 16);
1450             codec_id = ff_codec_get_id(ff_codec_bmp_tags, track->video.fourcc);
1451             extradata_offset = 40;
1452         } else if (!strcmp(track->codec_id, "A_MS/ACM")
1453                    && track->codec_priv.size >= 14
1454                    && track->codec_priv.data != NULL) {
1455             int ret;
1456             ffio_init_context(&b, track->codec_priv.data, track->codec_priv.size,
1457                           AVIO_FLAG_READ, NULL, NULL, NULL, NULL);
1458             ret = ff_get_wav_header(&b, st->codec, track->codec_priv.size);
1459             if (ret < 0)
1460                 return ret;
1461             codec_id = st->codec->codec_id;
1462             extradata_offset = FFMIN(track->codec_priv.size, 18);
1463         } else if (!strcmp(track->codec_id, "V_QUICKTIME")
1464                    && (track->codec_priv.size >= 86)
1465                    && (track->codec_priv.data != NULL)) {
1466             track->video.fourcc = AV_RL32(track->codec_priv.data);
1467             codec_id=ff_codec_get_id(ff_codec_movvideo_tags, track->video.fourcc);
1468         } else if (codec_id == CODEC_ID_PCM_S16BE) {
1469             switch (track->audio.bitdepth) {
1470             case  8:  codec_id = CODEC_ID_PCM_U8;     break;
1471             case 24:  codec_id = CODEC_ID_PCM_S24BE;  break;
1472             case 32:  codec_id = CODEC_ID_PCM_S32BE;  break;
1473             }
1474         } else if (codec_id == CODEC_ID_PCM_S16LE) {
1475             switch (track->audio.bitdepth) {
1476             case  8:  codec_id = CODEC_ID_PCM_U8;     break;
1477             case 24:  codec_id = CODEC_ID_PCM_S24LE;  break;
1478             case 32:  codec_id = CODEC_ID_PCM_S32LE;  break;
1479             }
1480         } else if (codec_id==CODEC_ID_PCM_F32LE && track->audio.bitdepth==64) {
1481             codec_id = CODEC_ID_PCM_F64LE;
1482         } else if (codec_id == CODEC_ID_AAC && !track->codec_priv.size) {
1483             int profile = matroska_aac_profile(track->codec_id);
1484             int sri = matroska_aac_sri(track->audio.samplerate);
1485             extradata = av_mallocz(5 + FF_INPUT_BUFFER_PADDING_SIZE);
1486             if (extradata == NULL)
1487                 return AVERROR(ENOMEM);
1488             extradata[0] = (profile << 3) | ((sri&0x0E) >> 1);
1489             extradata[1] = ((sri&0x01) << 7) | (track->audio.channels<<3);
1490             if (strstr(track->codec_id, "SBR")) {
1491                 sri = matroska_aac_sri(track->audio.out_samplerate);
1492                 extradata[2] = 0x56;
1493                 extradata[3] = 0xE5;
1494                 extradata[4] = 0x80 | (sri<<3);
1495                 extradata_size = 5;
1496             } else
1497                 extradata_size = 2;
1498         } else if (codec_id == CODEC_ID_TTA) {
1499             extradata_size = 30;
1500             extradata = av_mallocz(extradata_size);
1501             if (extradata == NULL)
1502                 return AVERROR(ENOMEM);
1503             ffio_init_context(&b, extradata, extradata_size, 1,
1504                           NULL, NULL, NULL, NULL);
1505             avio_write(&b, "TTA1", 4);
1506             avio_wl16(&b, 1);
1507             avio_wl16(&b, track->audio.channels);
1508             avio_wl16(&b, track->audio.bitdepth);
1509             avio_wl32(&b, track->audio.out_samplerate);
1510             avio_wl32(&b, matroska->ctx->duration * track->audio.out_samplerate);
1511         } else if (codec_id == CODEC_ID_RV10 || codec_id == CODEC_ID_RV20 ||
1512                    codec_id == CODEC_ID_RV30 || codec_id == CODEC_ID_RV40) {
1513             extradata_offset = 26;
1514         } else if (codec_id == CODEC_ID_RA_144) {
1515             track->audio.out_samplerate = 8000;
1516             track->audio.channels = 1;
1517         } else if (codec_id == CODEC_ID_RA_288 || codec_id == CODEC_ID_COOK ||
1518                    codec_id == CODEC_ID_ATRAC3 || codec_id == CODEC_ID_SIPR) {
1519             int flavor;
1520             ffio_init_context(&b, track->codec_priv.data,track->codec_priv.size,
1521                           0, NULL, NULL, NULL, NULL);
1522             avio_skip(&b, 22);
1523             flavor                       = avio_rb16(&b);
1524             track->audio.coded_framesize = avio_rb32(&b);
1525             avio_skip(&b, 12);
1526             track->audio.sub_packet_h    = avio_rb16(&b);
1527             track->audio.frame_size      = avio_rb16(&b);
1528             track->audio.sub_packet_size = avio_rb16(&b);
1529             track->audio.buf = av_malloc(track->audio.frame_size * track->audio.sub_packet_h);
1530             if (codec_id == CODEC_ID_RA_288) {
1531                 st->codec->block_align = track->audio.coded_framesize;
1532                 track->codec_priv.size = 0;
1533             } else {
1534                 if (codec_id == CODEC_ID_SIPR && flavor < 4) {
1535                     const int sipr_bit_rate[4] = { 6504, 8496, 5000, 16000 };
1536                     track->audio.sub_packet_size = ff_sipr_subpk_size[flavor];
1537                     st->codec->bit_rate = sipr_bit_rate[flavor];
1538                 }
1539                 st->codec->block_align = track->audio.sub_packet_size;
1540                 extradata_offset = 78;
1541             }
1542         }
1543         track->codec_priv.size -= extradata_offset;
1544
1545         if (codec_id == CODEC_ID_NONE)
1546             av_log(matroska->ctx, AV_LOG_INFO,
1547                    "Unknown/unsupported CodecID %s.\n", track->codec_id);
1548
1549         if (track->time_scale < 0.01)
1550             track->time_scale = 1.0;
1551         avpriv_set_pts_info(st, 64, matroska->time_scale*track->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
1552
1553         st->codec->codec_id = codec_id;
1554         st->start_time = 0;
1555         if (strcmp(track->language, "und"))
1556             av_dict_set(&st->metadata, "language", track->language, 0);
1557         av_dict_set(&st->metadata, "title", track->name, 0);
1558
1559         if (track->flag_default)
1560             st->disposition |= AV_DISPOSITION_DEFAULT;
1561         if (track->flag_forced)
1562             st->disposition |= AV_DISPOSITION_FORCED;
1563
1564         if (!st->codec->extradata) {
1565             if(extradata){
1566                 st->codec->extradata = extradata;
1567                 st->codec->extradata_size = extradata_size;
1568             } else if(track->codec_priv.data && track->codec_priv.size > 0){
1569                 st->codec->extradata = av_mallocz(track->codec_priv.size +
1570                                                   FF_INPUT_BUFFER_PADDING_SIZE);
1571                 if(st->codec->extradata == NULL)
1572                     return AVERROR(ENOMEM);
1573                 st->codec->extradata_size = track->codec_priv.size;
1574                 memcpy(st->codec->extradata,
1575                        track->codec_priv.data + extradata_offset,
1576                        track->codec_priv.size);
1577             }
1578         }
1579
1580         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
1581             st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
1582             st->codec->codec_tag  = track->video.fourcc;
1583             st->codec->width  = track->video.pixel_width;
1584             st->codec->height = track->video.pixel_height;
1585             av_reduce(&st->sample_aspect_ratio.num,
1586                       &st->sample_aspect_ratio.den,
1587                       st->codec->height * track->video.display_width,
1588                       st->codec-> width * track->video.display_height,
1589                       255);
1590             if (st->codec->codec_id != CODEC_ID_H264)
1591             st->need_parsing = AVSTREAM_PARSE_HEADERS;
1592             if (track->default_duration) {
1593                 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
1594                           1000000000, track->default_duration, 30000);
1595                 st->avg_frame_rate = st->r_frame_rate;
1596             }
1597         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1598             st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
1599             st->codec->sample_rate = track->audio.out_samplerate;
1600             st->codec->channels = track->audio.channels;
1601             if (st->codec->codec_id != CODEC_ID_AAC)
1602             st->need_parsing = AVSTREAM_PARSE_HEADERS;
1603         } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
1604             st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
1605             if (st->codec->codec_id == CODEC_ID_SSA)
1606                 matroska->contains_ssa = 1;
1607         }
1608     }
1609
1610     attachements = attachements_list->elem;
1611     for (j=0; j<attachements_list->nb_elem; j++) {
1612         if (!(attachements[j].filename && attachements[j].mime &&
1613               attachements[j].bin.data && attachements[j].bin.size > 0)) {
1614             av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
1615         } else {
1616             AVStream *st = avformat_new_stream(s, NULL);
1617             if (st == NULL)
1618                 break;
1619             av_dict_set(&st->metadata, "filename",attachements[j].filename, 0);
1620             av_dict_set(&st->metadata, "mimetype", attachements[j].mime, 0);
1621             st->codec->codec_id = CODEC_ID_NONE;
1622             st->codec->codec_type = AVMEDIA_TYPE_ATTACHMENT;
1623             st->codec->extradata  = av_malloc(attachements[j].bin.size);
1624             if(st->codec->extradata == NULL)
1625                 break;
1626             st->codec->extradata_size = attachements[j].bin.size;
1627             memcpy(st->codec->extradata, attachements[j].bin.data, attachements[j].bin.size);
1628
1629             for (i=0; ff_mkv_mime_tags[i].id != CODEC_ID_NONE; i++) {
1630                 if (!strncmp(ff_mkv_mime_tags[i].str, attachements[j].mime,
1631                              strlen(ff_mkv_mime_tags[i].str))) {
1632                     st->codec->codec_id = ff_mkv_mime_tags[i].id;
1633                     break;
1634                 }
1635             }
1636             attachements[j].stream = st;
1637         }
1638     }
1639
1640     chapters = chapters_list->elem;
1641     for (i=0; i<chapters_list->nb_elem; i++)
1642         if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid
1643             && (max_start==0 || chapters[i].start > max_start)) {
1644             chapters[i].chapter =
1645             avpriv_new_chapter(s, chapters[i].uid, (AVRational){1, 1000000000},
1646                            chapters[i].start, chapters[i].end,
1647                            chapters[i].title);
1648             av_dict_set(&chapters[i].chapter->metadata,
1649                              "title", chapters[i].title, 0);
1650             max_start = chapters[i].start;
1651         }
1652
1653     matroska_convert_tags(s);
1654
1655     return 0;
1656 }
1657
1658 /*
1659  * Put one packet in an application-supplied AVPacket struct.
1660  * Returns 0 on success or -1 on failure.
1661  */
1662 static int matroska_deliver_packet(MatroskaDemuxContext *matroska,
1663                                    AVPacket *pkt)
1664 {
1665     if (matroska->num_packets > 0) {
1666         memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
1667         av_free(matroska->packets[0]);
1668         if (matroska->num_packets > 1) {
1669             void *newpackets;
1670             memmove(&matroska->packets[0], &matroska->packets[1],
1671                     (matroska->num_packets - 1) * sizeof(AVPacket *));
1672             newpackets = av_realloc(matroska->packets,
1673                             (matroska->num_packets - 1) * sizeof(AVPacket *));
1674             if (newpackets)
1675                 matroska->packets = newpackets;
1676         } else {
1677             av_freep(&matroska->packets);
1678             matroska->prev_pkt = NULL;
1679         }
1680         matroska->num_packets--;
1681         return 0;
1682     }
1683
1684     return -1;
1685 }
1686
1687 /*
1688  * Free all packets in our internal queue.
1689  */
1690 static void matroska_clear_queue(MatroskaDemuxContext *matroska)
1691 {
1692     if (matroska->packets) {
1693         int n;
1694         for (n = 0; n < matroska->num_packets; n++) {
1695             av_free_packet(matroska->packets[n]);
1696             av_free(matroska->packets[n]);
1697         }
1698         av_freep(&matroska->packets);
1699         matroska->num_packets = 0;
1700     }
1701 }
1702
1703 static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data,
1704                                 int size, int64_t pos, uint64_t cluster_time,
1705                                 uint64_t duration, int is_keyframe,
1706                                 int64_t cluster_pos)
1707 {
1708     uint64_t timecode = AV_NOPTS_VALUE;
1709     MatroskaTrack *track;
1710     int res = 0;
1711     AVStream *st;
1712     AVPacket *pkt;
1713     int16_t block_time;
1714     uint32_t *lace_size = NULL;
1715     int n, flags, laces = 0;
1716     uint64_t num;
1717
1718     if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) {
1719         av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
1720         return n;
1721     }
1722     data += n;
1723     size -= n;
1724
1725     track = matroska_find_track_by_num(matroska, num);
1726     if (!track || !track->stream) {
1727         av_log(matroska->ctx, AV_LOG_INFO,
1728                "Invalid stream %"PRIu64" or size %u\n", num, size);
1729         return AVERROR_INVALIDDATA;
1730     } else if (size <= 3)
1731         return 0;
1732     st = track->stream;
1733     if (st->discard >= AVDISCARD_ALL)
1734         return res;
1735     if (duration == AV_NOPTS_VALUE)
1736         duration = track->default_duration / matroska->time_scale;
1737
1738     block_time = AV_RB16(data);
1739     data += 2;
1740     flags = *data++;
1741     size -= 3;
1742     if (is_keyframe == -1)
1743         is_keyframe = flags & 0x80 ? AV_PKT_FLAG_KEY : 0;
1744
1745     if (cluster_time != (uint64_t)-1
1746         && (block_time >= 0 || cluster_time >= -block_time)) {
1747         timecode = cluster_time + block_time;
1748         if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE
1749             && timecode < track->end_timecode)
1750             is_keyframe = 0;  /* overlapping subtitles are not key frame */
1751         if (is_keyframe)
1752             av_add_index_entry(st, cluster_pos, timecode, 0,0,AVINDEX_KEYFRAME);
1753         track->end_timecode = FFMAX(track->end_timecode, timecode+duration);
1754     }
1755
1756     if (matroska->skip_to_keyframe && track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1757         if (!is_keyframe || timecode < matroska->skip_to_timecode)
1758             return res;
1759         matroska->skip_to_keyframe = 0;
1760     }
1761
1762     switch ((flags & 0x06) >> 1) {
1763         case 0x0: /* no lacing */
1764             laces = 1;
1765             lace_size = av_mallocz(sizeof(int));
1766             lace_size[0] = size;
1767             break;
1768
1769         case 0x1: /* Xiph lacing */
1770         case 0x2: /* fixed-size lacing */
1771         case 0x3: /* EBML lacing */
1772             assert(size>0); // size <=3 is checked before size-=3 above
1773             laces = (*data) + 1;
1774             data += 1;
1775             size -= 1;
1776             lace_size = av_mallocz(laces * sizeof(int));
1777
1778             switch ((flags & 0x06) >> 1) {
1779                 case 0x1: /* Xiph lacing */ {
1780                     uint8_t temp;
1781                     uint32_t total = 0;
1782                     for (n = 0; res == 0 && n < laces - 1; n++) {
1783                         while (1) {
1784                             if (size == 0) {
1785                                 res = -1;
1786                                 break;
1787                             }
1788                             temp = *data;
1789                             lace_size[n] += temp;
1790                             data += 1;
1791                             size -= 1;
1792                             if (temp != 0xff)
1793                                 break;
1794                         }
1795                         total += lace_size[n];
1796                     }
1797                     lace_size[n] = size - total;
1798                     break;
1799                 }
1800
1801                 case 0x2: /* fixed-size lacing */
1802                     for (n = 0; n < laces; n++)
1803                         lace_size[n] = size / laces;
1804                     break;
1805
1806                 case 0x3: /* EBML lacing */ {
1807                     uint32_t total;
1808                     n = matroska_ebmlnum_uint(matroska, data, size, &num);
1809                     if (n < 0) {
1810                         av_log(matroska->ctx, AV_LOG_INFO,
1811                                "EBML block data error\n");
1812                         break;
1813                     }
1814                     data += n;
1815                     size -= n;
1816                     total = lace_size[0] = num;
1817                     for (n = 1; res == 0 && n < laces - 1; n++) {
1818                         int64_t snum;
1819                         int r;
1820                         r = matroska_ebmlnum_sint(matroska, data, size, &snum);
1821                         if (r < 0) {
1822                             av_log(matroska->ctx, AV_LOG_INFO,
1823                                    "EBML block data error\n");
1824                             break;
1825                         }
1826                         data += r;
1827                         size -= r;
1828                         lace_size[n] = lace_size[n - 1] + snum;
1829                         total += lace_size[n];
1830                     }
1831                     lace_size[laces - 1] = size - total;
1832                     break;
1833                 }
1834             }
1835             break;
1836     }
1837
1838     if (res == 0) {
1839         for (n = 0; n < laces; n++) {
1840             if ((st->codec->codec_id == CODEC_ID_RA_288 ||
1841                  st->codec->codec_id == CODEC_ID_COOK ||
1842                  st->codec->codec_id == CODEC_ID_SIPR ||
1843                  st->codec->codec_id == CODEC_ID_ATRAC3) &&
1844                  st->codec->block_align && track->audio.sub_packet_size) {
1845                 int a = st->codec->block_align;
1846                 int sps = track->audio.sub_packet_size;
1847                 int cfs = track->audio.coded_framesize;
1848                 int h = track->audio.sub_packet_h;
1849                 int y = track->audio.sub_packet_cnt;
1850                 int w = track->audio.frame_size;
1851                 int x;
1852
1853                 if (!track->audio.pkt_cnt) {
1854                     if (track->audio.sub_packet_cnt == 0)
1855                         track->audio.buf_timecode = timecode;
1856                     if (st->codec->codec_id == CODEC_ID_RA_288) {
1857                         if (size < cfs * h / 2) {
1858                             av_log(matroska->ctx, AV_LOG_ERROR,
1859                                    "Corrupt int4 RM-style audio packet size\n");
1860                             res = AVERROR_INVALIDDATA;
1861                             goto end;
1862                         }
1863                         for (x=0; x<h/2; x++)
1864                             memcpy(track->audio.buf+x*2*w+y*cfs,
1865                                    data+x*cfs, cfs);
1866                     } else if (st->codec->codec_id == CODEC_ID_SIPR) {
1867                         if (size < w) {
1868                             av_log(matroska->ctx, AV_LOG_ERROR,
1869                                    "Corrupt sipr RM-style audio packet size\n");
1870                             res = AVERROR_INVALIDDATA;
1871                             goto end;
1872                         }
1873                         memcpy(track->audio.buf + y*w, data, w);
1874                     } else {
1875                         if (size < sps * w / sps) {
1876                             av_log(matroska->ctx, AV_LOG_ERROR,
1877                                    "Corrupt generic RM-style audio packet size\n");
1878                             res = AVERROR_INVALIDDATA;
1879                             goto end;
1880                         }
1881                         for (x=0; x<w/sps; x++)
1882                             memcpy(track->audio.buf+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), data+x*sps, sps);
1883                     }
1884
1885                     if (++track->audio.sub_packet_cnt >= h) {
1886                         if (st->codec->codec_id == CODEC_ID_SIPR)
1887                             ff_rm_reorder_sipr_data(track->audio.buf, h, w);
1888                         track->audio.sub_packet_cnt = 0;
1889                         track->audio.pkt_cnt = h*w / a;
1890                     }
1891                 }
1892                 while (track->audio.pkt_cnt) {
1893                     pkt = av_mallocz(sizeof(AVPacket));
1894                     av_new_packet(pkt, a);
1895                     memcpy(pkt->data, track->audio.buf
1896                            + a * (h*w / a - track->audio.pkt_cnt--), a);
1897                     pkt->pts = track->audio.buf_timecode;
1898                     track->audio.buf_timecode = AV_NOPTS_VALUE;
1899                     pkt->pos = pos;
1900                     pkt->stream_index = st->index;
1901                     dynarray_add(&matroska->packets,&matroska->num_packets,pkt);
1902                 }
1903             } else {
1904                 MatroskaTrackEncoding *encodings = track->encodings.elem;
1905                 int offset = 0, pkt_size = lace_size[n];
1906                 uint8_t *pkt_data = data;
1907
1908                 if (pkt_size > size) {
1909                     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid packet size\n");
1910                     break;
1911                 }
1912
1913                 if (encodings && encodings->scope & 1) {
1914                     offset = matroska_decode_buffer(&pkt_data,&pkt_size, track);
1915                     if (offset < 0)
1916                         continue;
1917                 }
1918
1919                 pkt = av_mallocz(sizeof(AVPacket));
1920                 /* XXX: prevent data copy... */
1921                 if (av_new_packet(pkt, pkt_size+offset) < 0) {
1922                     av_free(pkt);
1923                     res = AVERROR(ENOMEM);
1924                     break;
1925                 }
1926                 if (offset)
1927                     memcpy (pkt->data, encodings->compression.settings.data, offset);
1928                 memcpy (pkt->data+offset, pkt_data, pkt_size);
1929
1930                 if (pkt_data != data)
1931                     av_free(pkt_data);
1932
1933                 if (n == 0)
1934                     pkt->flags = is_keyframe;
1935                 pkt->stream_index = st->index;
1936
1937                 if (track->ms_compat)
1938                     pkt->dts = timecode;
1939                 else
1940                     pkt->pts = timecode;
1941                 pkt->pos = pos;
1942                 if (st->codec->codec_id == CODEC_ID_TEXT)
1943                     pkt->convergence_duration = duration;
1944                 else if (track->type != MATROSKA_TRACK_TYPE_SUBTITLE)
1945                     pkt->duration = duration;
1946
1947                 if (st->codec->codec_id == CODEC_ID_SSA)
1948                     matroska_fix_ass_packet(matroska, pkt, duration);
1949
1950                 if (matroska->prev_pkt &&
1951                     timecode != AV_NOPTS_VALUE &&
1952                     matroska->prev_pkt->pts == timecode &&
1953                     matroska->prev_pkt->stream_index == st->index &&
1954                     st->codec->codec_id == CODEC_ID_SSA)
1955                     matroska_merge_packets(matroska->prev_pkt, pkt);
1956                 else {
1957                     dynarray_add(&matroska->packets,&matroska->num_packets,pkt);
1958                     matroska->prev_pkt = pkt;
1959                 }
1960             }
1961
1962             if (timecode != AV_NOPTS_VALUE)
1963                 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
1964             data += lace_size[n];
1965             size -= lace_size[n];
1966         }
1967     }
1968
1969 end:
1970     av_free(lace_size);
1971     return res;
1972 }
1973
1974 static int matroska_parse_cluster_incremental(MatroskaDemuxContext *matroska)
1975 {
1976     EbmlList *blocks_list;
1977     MatroskaBlock *blocks;
1978     int i, res;
1979     res = ebml_parse(matroska,
1980                      matroska_cluster_incremental_parsing,
1981                      &matroska->current_cluster);
1982     if (res == 1) {
1983         /* New Cluster */
1984         if (matroska->current_cluster_pos)
1985             ebml_level_end(matroska);
1986         ebml_free(matroska_cluster, &matroska->current_cluster);
1987         memset(&matroska->current_cluster, 0, sizeof(MatroskaCluster));
1988         matroska->current_cluster_num_blocks = 0;
1989         matroska->current_cluster_pos = avio_tell(matroska->ctx->pb);
1990         matroska->prev_pkt = NULL;
1991         /* sizeof the ID which was already read */
1992         if (matroska->current_id)
1993             matroska->current_cluster_pos -= 4;
1994         res = ebml_parse(matroska,
1995                          matroska_clusters_incremental,
1996                          &matroska->current_cluster);
1997         /* Try parsing the block again. */
1998         if (res == 1)
1999             res = ebml_parse(matroska,
2000                              matroska_cluster_incremental_parsing,
2001                              &matroska->current_cluster);
2002     }
2003
2004     if (!res &&
2005         matroska->current_cluster_num_blocks <
2006             matroska->current_cluster.blocks.nb_elem) {
2007         blocks_list = &matroska->current_cluster.blocks;
2008         blocks = blocks_list->elem;
2009
2010         matroska->current_cluster_num_blocks = blocks_list->nb_elem;
2011         i = blocks_list->nb_elem - 1;
2012         if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
2013             int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
2014             if (!blocks[i].non_simple)
2015                 blocks[i].duration = AV_NOPTS_VALUE;
2016             res = matroska_parse_block(matroska,
2017                                        blocks[i].bin.data, blocks[i].bin.size,
2018                                        blocks[i].bin.pos,
2019                                        matroska->current_cluster.timecode,
2020                                        blocks[i].duration, is_keyframe,
2021                                        matroska->current_cluster_pos);
2022         }
2023     }
2024
2025     if (res < 0)  matroska->done = 1;
2026     return res;
2027 }
2028
2029 static int matroska_parse_cluster(MatroskaDemuxContext *matroska)
2030 {
2031     MatroskaCluster cluster = { 0 };
2032     EbmlList *blocks_list;
2033     MatroskaBlock *blocks;
2034     int i, res;
2035     int64_t pos;
2036     if (!matroska->contains_ssa)
2037         return matroska_parse_cluster_incremental(matroska);
2038     pos = avio_tell(matroska->ctx->pb);
2039     matroska->prev_pkt = NULL;
2040     if (matroska->current_id)
2041         pos -= 4;  /* sizeof the ID which was already read */
2042     res = ebml_parse(matroska, matroska_clusters, &cluster);
2043     blocks_list = &cluster.blocks;
2044     blocks = blocks_list->elem;
2045     for (i=0; i<blocks_list->nb_elem && !res; i++)
2046         if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
2047             int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
2048             if (!blocks[i].non_simple)
2049                 blocks[i].duration = AV_NOPTS_VALUE;
2050             res=matroska_parse_block(matroska,
2051                                      blocks[i].bin.data, blocks[i].bin.size,
2052                                      blocks[i].bin.pos,  cluster.timecode,
2053                                      blocks[i].duration, is_keyframe,
2054                                      pos);
2055         }
2056     ebml_free(matroska_cluster, &cluster);
2057     if (res < 0)  matroska->done = 1;
2058     return res;
2059 }
2060
2061 static int matroska_read_packet(AVFormatContext *s, AVPacket *pkt)
2062 {
2063     MatroskaDemuxContext *matroska = s->priv_data;
2064     int ret = 0;
2065
2066     while (!ret && matroska_deliver_packet(matroska, pkt)) {
2067         if (matroska->done)
2068             return AVERROR_EOF;
2069         ret = matroska_parse_cluster(matroska);
2070     }
2071
2072     if (ret == AVERROR_INVALIDDATA) {
2073         pkt->flags |= AV_PKT_FLAG_CORRUPT;
2074         return 0;
2075     }
2076
2077     return ret;
2078 }
2079
2080 static int matroska_read_seek(AVFormatContext *s, int stream_index,
2081                               int64_t timestamp, int flags)
2082 {
2083     MatroskaDemuxContext *matroska = s->priv_data;
2084     MatroskaTrack *tracks = matroska->tracks.elem;
2085     AVStream *st = s->streams[stream_index];
2086     int i, index, index_sub, index_min;
2087
2088     /* Parse the CUES now since we need the index data to seek. */
2089     if (matroska->cues_parsing_deferred) {
2090         matroska_parse_cues(matroska);
2091         matroska->cues_parsing_deferred = 0;
2092     }
2093
2094     if (!st->nb_index_entries)
2095         return 0;
2096     timestamp = FFMAX(timestamp, st->index_entries[0].timestamp);
2097
2098     if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
2099         avio_seek(s->pb, st->index_entries[st->nb_index_entries-1].pos, SEEK_SET);
2100         matroska->current_id = 0;
2101         while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
2102             matroska->prev_pkt = NULL;
2103             matroska_clear_queue(matroska);
2104             if (matroska_parse_cluster(matroska) < 0)
2105                 break;
2106         }
2107     }
2108
2109     matroska_clear_queue(matroska);
2110     if (index < 0)
2111         return 0;
2112
2113     index_min = index;
2114     for (i=0; i < matroska->tracks.nb_elem; i++) {
2115         tracks[i].audio.pkt_cnt = 0;
2116         tracks[i].audio.sub_packet_cnt = 0;
2117         tracks[i].audio.buf_timecode = AV_NOPTS_VALUE;
2118         tracks[i].end_timecode = 0;
2119         if (tracks[i].type == MATROSKA_TRACK_TYPE_SUBTITLE
2120             && !tracks[i].stream->discard != AVDISCARD_ALL) {
2121             index_sub = av_index_search_timestamp(tracks[i].stream, st->index_entries[index].timestamp, AVSEEK_FLAG_BACKWARD);
2122             if (index_sub >= 0
2123                 && st->index_entries[index_sub].pos < st->index_entries[index_min].pos
2124                 && st->index_entries[index].timestamp - st->index_entries[index_sub].timestamp < 30000000000/matroska->time_scale)
2125                 index_min = index_sub;
2126         }
2127     }
2128
2129     avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET);
2130     matroska->current_id = 0;
2131     matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
2132     matroska->skip_to_timecode = st->index_entries[index].timestamp;
2133     matroska->done = 0;
2134     ff_update_cur_dts(s, st, st->index_entries[index].timestamp);
2135     return 0;
2136 }
2137
2138 static int matroska_read_close(AVFormatContext *s)
2139 {
2140     MatroskaDemuxContext *matroska = s->priv_data;
2141     MatroskaTrack *tracks = matroska->tracks.elem;
2142     int n;
2143
2144     matroska_clear_queue(matroska);
2145
2146     for (n=0; n < matroska->tracks.nb_elem; n++)
2147         if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
2148             av_free(tracks[n].audio.buf);
2149     ebml_free(matroska_cluster, &matroska->current_cluster);
2150     ebml_free(matroska_segment, matroska);
2151
2152     return 0;
2153 }
2154
2155 AVInputFormat ff_matroska_demuxer = {
2156     .name           = "matroska,webm",
2157     .long_name      = NULL_IF_CONFIG_SMALL("Matroska/WebM file format"),
2158     .priv_data_size = sizeof(MatroskaDemuxContext),
2159     .read_probe     = matroska_probe,
2160     .read_header    = matroska_read_header,
2161     .read_packet    = matroska_read_packet,
2162     .read_close     = matroska_read_close,
2163     .read_seek      = matroska_read_seek,
2164 };