]> git.sesse.net Git - ffmpeg/blob - libavformat/matroskadec.c
matroskadec: use an EbmlBin as parameter to ebml_read_binary()
[ffmpeg] / libavformat / matroskadec.c
1 /*
2  * Matroska file demuxer (no muxer yet)
3  * Copyright (c) 2003-2004 The ffmpeg Project
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg 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  * FFmpeg 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 FFmpeg; 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 matroskadec.c
24  * Matroska file demuxer
25  * by Ronald Bultje <rbultje@ronald.bitfreak.net>
26  * with a little help from Moritz Bunkus <moritz@bunkus.org>
27  * Specs available on the matroska project page:
28  * http://www.matroska.org/.
29  */
30
31 #include "avformat.h"
32 /* For codec_get_id(). */
33 #include "riff.h"
34 #include "isom.h"
35 #include "matroska.h"
36 #include "libavcodec/mpeg4audio.h"
37 #include "libavutil/intfloat_readwrite.h"
38 #include "libavutil/avstring.h"
39 #include "libavutil/lzo.h"
40 #ifdef CONFIG_ZLIB
41 #include <zlib.h>
42 #endif
43 #ifdef CONFIG_BZLIB
44 #include <bzlib.h>
45 #endif
46
47 typedef enum {
48     EBML_NONE,
49     EBML_UINT,
50     EBML_FLOAT,
51     EBML_STR,
52     EBML_UTF8,
53     EBML_BIN,
54     EBML_NEST,
55     EBML_PASS,
56     EBML_STOP,
57 } EbmlType;
58
59 typedef const struct EbmlSyntax {
60     uint32_t id;
61     EbmlType type;
62     int list_elem_size;
63     int data_offset;
64     union {
65         uint64_t    u;
66         double      f;
67         const char *s;
68         const struct EbmlSyntax *n;
69     } def;
70 } EbmlSyntax;
71
72 typedef struct {
73     int nb_elem;
74     void *elem;
75 } EbmlList;
76
77 typedef struct {
78     int      size;
79     uint8_t *data;
80     int64_t  pos;
81 } EbmlBin;
82
83 typedef struct {
84     uint64_t version;
85     uint64_t max_size;
86     uint64_t id_length;
87     char    *doctype;
88     uint64_t doctype_version;
89 } Ebml;
90
91 typedef struct {
92     uint64_t algo;
93     EbmlBin  settings;
94 } MatroskaTrackCompression;
95
96 typedef struct {
97     uint64_t scope;
98     uint64_t type;
99     MatroskaTrackCompression compression;
100 } MatroskaTrackEncoding;
101
102 typedef struct {
103     double   frame_rate;
104     uint64_t display_width;
105     uint64_t display_height;
106     uint64_t pixel_width;
107     uint64_t pixel_height;
108     uint64_t fourcc;
109 } MatroskaTrackVideo;
110
111 typedef struct {
112     double   samplerate;
113     double   out_samplerate;
114     uint64_t bitdepth;
115     uint64_t channels;
116
117     /* real audio header (extracted from extradata) */
118     int      coded_framesize;
119     int      sub_packet_h;
120     int      frame_size;
121     int      sub_packet_size;
122     int      sub_packet_cnt;
123     int      pkt_cnt;
124     uint8_t *buf;
125 } MatroskaTrackAudio;
126
127 typedef struct {
128     uint64_t num;
129     uint64_t type;
130     char    *codec_id;
131     EbmlBin  codec_priv;
132     char    *language;
133     double time_scale;
134     uint64_t default_duration;
135     uint64_t flag_default;
136     MatroskaTrackVideo video;
137     MatroskaTrackAudio audio;
138     EbmlList encodings;
139
140     AVStream *stream;
141 } MatroskaTrack;
142
143 typedef struct {
144     char *filename;
145     char *mime;
146     EbmlBin bin;
147 } MatroskaAttachement;
148
149 typedef struct {
150     uint64_t start;
151     uint64_t end;
152     uint64_t uid;
153     char    *title;
154 } MatroskaChapter;
155
156 typedef struct {
157     uint64_t track;
158     uint64_t pos;
159 } MatroskaIndexPos;
160
161 typedef struct {
162     uint64_t time;
163     EbmlList pos;
164 } MatroskaIndex;
165
166 typedef struct {
167     uint64_t id;
168     uint64_t pos;
169 } MatroskaSeekhead;
170
171 typedef struct {
172     uint64_t start;
173     uint64_t length;
174 } MatroskaLevel;
175
176 typedef struct {
177     AVFormatContext *ctx;
178
179     /* ebml stuff */
180     int num_levels;
181     MatroskaLevel levels[EBML_MAX_DEPTH];
182     int level_up;
183
184     uint64_t time_scale;
185     double   duration;
186     char    *title;
187     EbmlList tracks;
188     EbmlList attachments;
189     EbmlList chapters;
190     EbmlList index;
191     EbmlList seekhead;
192
193     /* num_streams is the number of streams that av_new_stream() was called
194      * for ( = that are available to the calling program). */
195     int num_streams;
196
197     /* cache for ID peeking */
198     uint32_t peek_id;
199
200     /* byte position of the segment inside the stream */
201     offset_t segment_start;
202
203     /* The packet queue. */
204     AVPacket **packets;
205     int num_packets;
206
207     int done;
208     int has_cluster_id;
209
210     /* What to skip before effectively reading a packet. */
211     int skip_to_keyframe;
212     AVStream *skip_to_stream;
213 } MatroskaDemuxContext;
214
215 typedef struct {
216     uint64_t duration;
217     int64_t  reference;
218     EbmlBin  bin;
219 } MatroskaBlock;
220
221 typedef struct {
222     uint64_t timecode;
223     EbmlList blocks;
224 } MatroskaCluster;
225
226 #define ARRAY_SIZE(x)  (sizeof(x)/sizeof(*x))
227
228 static EbmlSyntax ebml_header[] = {
229     { EBML_ID_EBMLREADVERSION,        EBML_UINT, 0, offsetof(Ebml,version), {.u=EBML_VERSION} },
230     { EBML_ID_EBMLMAXSIZELENGTH,      EBML_UINT, 0, offsetof(Ebml,max_size), {.u=8} },
231     { EBML_ID_EBMLMAXIDLENGTH,        EBML_UINT, 0, offsetof(Ebml,id_length), {.u=4} },
232     { EBML_ID_DOCTYPE,                EBML_STR,  0, offsetof(Ebml,doctype), {.s="(none)"} },
233     { EBML_ID_DOCTYPEREADVERSION,     EBML_UINT, 0, offsetof(Ebml,doctype_version), {.u=1} },
234     { EBML_ID_EBMLVERSION,            EBML_NONE },
235     { EBML_ID_DOCTYPEVERSION,         EBML_NONE },
236     { EBML_ID_VOID,                   EBML_NONE },
237     { 0 }
238 };
239
240 static EbmlSyntax ebml_syntax[] = {
241     { EBML_ID_HEADER,                 EBML_NEST, 0, 0, {.n=ebml_header} },
242     { 0 }
243 };
244
245 static EbmlSyntax matroska_info[] = {
246     { MATROSKA_ID_TIMECODESCALE,      EBML_UINT,  0, offsetof(MatroskaDemuxContext,time_scale), {.u=1000000} },
247     { MATROSKA_ID_DURATION,           EBML_FLOAT, 0, offsetof(MatroskaDemuxContext,duration) },
248     { MATROSKA_ID_TITLE,              EBML_UTF8,  0, offsetof(MatroskaDemuxContext,title) },
249     { MATROSKA_ID_WRITINGAPP,         EBML_NONE },
250     { MATROSKA_ID_MUXINGAPP,          EBML_NONE },
251     { MATROSKA_ID_DATEUTC,            EBML_NONE },
252     { MATROSKA_ID_SEGMENTUID,         EBML_NONE },
253     { EBML_ID_VOID,                   EBML_NONE },
254     { 0 }
255 };
256
257 static EbmlSyntax matroska_track_video[] = {
258     { MATROSKA_ID_VIDEOFRAMERATE,     EBML_FLOAT,0, offsetof(MatroskaTrackVideo,frame_rate) },
259     { MATROSKA_ID_VIDEODISPLAYWIDTH,  EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_width) },
260     { MATROSKA_ID_VIDEODISPLAYHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_height) },
261     { MATROSKA_ID_VIDEOPIXELWIDTH,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_width) },
262     { MATROSKA_ID_VIDEOPIXELHEIGHT,   EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_height) },
263     { MATROSKA_ID_VIDEOCOLORSPACE,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,fourcc) },
264     { MATROSKA_ID_VIDEOFLAGINTERLACED,EBML_NONE },
265     { MATROSKA_ID_VIDEOSTEREOMODE,    EBML_NONE },
266     { MATROSKA_ID_VIDEOASPECTRATIO,   EBML_NONE },
267     { EBML_ID_VOID,                   EBML_NONE },
268     { 0 }
269 };
270
271 static EbmlSyntax matroska_track_audio[] = {
272     { MATROSKA_ID_AUDIOSAMPLINGFREQ,  EBML_FLOAT,0, offsetof(MatroskaTrackAudio,samplerate), {.f=8000.0} },
273     { MATROSKA_ID_AUDIOOUTSAMPLINGFREQ,EBML_FLOAT,0,offsetof(MatroskaTrackAudio,out_samplerate) },
274     { MATROSKA_ID_AUDIOBITDEPTH,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,bitdepth) },
275     { MATROSKA_ID_AUDIOCHANNELS,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,channels), {.u=1} },
276     { EBML_ID_VOID,                   EBML_NONE },
277     { 0 }
278 };
279
280 static EbmlSyntax matroska_track_encoding_compression[] = {
281     { MATROSKA_ID_ENCODINGCOMPALGO,   EBML_UINT, 0, offsetof(MatroskaTrackCompression,algo), {.u=0} },
282     { MATROSKA_ID_ENCODINGCOMPSETTINGS,EBML_BIN, 0, offsetof(MatroskaTrackCompression,settings) },
283     { EBML_ID_VOID,                   EBML_NONE },
284     { 0 }
285 };
286
287 static EbmlSyntax matroska_track_encoding[] = {
288     { MATROSKA_ID_ENCODINGSCOPE,      EBML_UINT, 0, offsetof(MatroskaTrackEncoding,scope), {.u=1} },
289     { MATROSKA_ID_ENCODINGTYPE,       EBML_UINT, 0, offsetof(MatroskaTrackEncoding,type), {.u=0} },
290     { MATROSKA_ID_ENCODINGCOMPRESSION,EBML_NEST, 0, offsetof(MatroskaTrackEncoding,compression), {.n=matroska_track_encoding_compression} },
291     { EBML_ID_VOID,                   EBML_NONE },
292     { 0 }
293 };
294
295 static EbmlSyntax matroska_track_encodings[] = {
296     { MATROSKA_ID_TRACKCONTENTENCODING, EBML_NEST, sizeof(MatroskaTrackEncoding), offsetof(MatroskaTrack,encodings), {.n=matroska_track_encoding} },
297     { EBML_ID_VOID,                   EBML_NONE },
298     { 0 }
299 };
300
301 static EbmlSyntax matroska_track[] = {
302     { MATROSKA_ID_TRACKNUMBER,          EBML_UINT, 0, offsetof(MatroskaTrack,num) },
303     { MATROSKA_ID_TRACKTYPE,            EBML_UINT, 0, offsetof(MatroskaTrack,type) },
304     { MATROSKA_ID_CODECID,              EBML_STR,  0, offsetof(MatroskaTrack,codec_id) },
305     { MATROSKA_ID_CODECPRIVATE,         EBML_BIN,  0, offsetof(MatroskaTrack,codec_priv) },
306     { MATROSKA_ID_TRACKLANGUAGE,        EBML_UTF8, 0, offsetof(MatroskaTrack,language), {.s="eng"} },
307     { MATROSKA_ID_TRACKDEFAULTDURATION, EBML_UINT, 0, offsetof(MatroskaTrack,default_duration) },
308     { MATROSKA_ID_TRACKTIMECODESCALE,   EBML_FLOAT,0, offsetof(MatroskaTrack,time_scale), {.f=1.0} },
309     { MATROSKA_ID_TRACKFLAGDEFAULT,     EBML_UINT, 0, offsetof(MatroskaTrack,flag_default), {.u=1} },
310     { MATROSKA_ID_TRACKVIDEO,           EBML_NEST, 0, offsetof(MatroskaTrack,video), {.n=matroska_track_video} },
311     { MATROSKA_ID_TRACKAUDIO,           EBML_NEST, 0, offsetof(MatroskaTrack,audio), {.n=matroska_track_audio} },
312     { MATROSKA_ID_TRACKCONTENTENCODINGS,EBML_NEST, 0, 0, {.n=matroska_track_encodings} },
313     { MATROSKA_ID_TRACKUID,             EBML_NONE },
314     { MATROSKA_ID_TRACKNAME,            EBML_NONE },
315     { MATROSKA_ID_TRACKFLAGENABLED,     EBML_NONE },
316     { MATROSKA_ID_TRACKFLAGFORCED,      EBML_NONE },
317     { MATROSKA_ID_TRACKFLAGLACING,      EBML_NONE },
318     { MATROSKA_ID_CODECNAME,            EBML_NONE },
319     { MATROSKA_ID_CODECDECODEALL,       EBML_NONE },
320     { MATROSKA_ID_CODECINFOURL,         EBML_NONE },
321     { MATROSKA_ID_CODECDOWNLOADURL,     EBML_NONE },
322     { MATROSKA_ID_TRACKMINCACHE,        EBML_NONE },
323     { MATROSKA_ID_TRACKMAXCACHE,        EBML_NONE },
324     { EBML_ID_VOID,                     EBML_NONE },
325     { 0 }
326 };
327
328 static EbmlSyntax matroska_tracks[] = {
329     { MATROSKA_ID_TRACKENTRY,         EBML_NEST, sizeof(MatroskaTrack), offsetof(MatroskaDemuxContext,tracks), {.n=matroska_track} },
330     { EBML_ID_VOID,                   EBML_NONE },
331     { 0 }
332 };
333
334 static EbmlSyntax matroska_attachment[] = {
335     { MATROSKA_ID_FILENAME,           EBML_UTF8, 0, offsetof(MatroskaAttachement,filename) },
336     { MATROSKA_ID_FILEMIMETYPE,       EBML_STR,  0, offsetof(MatroskaAttachement,mime) },
337     { MATROSKA_ID_FILEDATA,           EBML_BIN,  0, offsetof(MatroskaAttachement,bin) },
338     { MATROSKA_ID_FILEUID,            EBML_NONE },
339     { EBML_ID_VOID,                   EBML_NONE },
340     { 0 }
341 };
342
343 static EbmlSyntax matroska_attachments[] = {
344     { MATROSKA_ID_ATTACHEDFILE,       EBML_NEST, sizeof(MatroskaAttachement), offsetof(MatroskaDemuxContext,attachments), {.n=matroska_attachment} },
345     { EBML_ID_VOID,                   EBML_NONE },
346     { 0 }
347 };
348
349 static EbmlSyntax matroska_chapter_display[] = {
350     { MATROSKA_ID_CHAPSTRING,         EBML_UTF8, 0, offsetof(MatroskaChapter,title) },
351     { EBML_ID_VOID,                   EBML_NONE },
352     { 0 }
353 };
354
355 static EbmlSyntax matroska_chapter_entry[] = {
356     { MATROSKA_ID_CHAPTERTIMESTART,   EBML_UINT, 0, offsetof(MatroskaChapter,start), {.u=AV_NOPTS_VALUE} },
357     { MATROSKA_ID_CHAPTERTIMEEND,     EBML_UINT, 0, offsetof(MatroskaChapter,end), {.u=AV_NOPTS_VALUE} },
358     { MATROSKA_ID_CHAPTERUID,         EBML_UINT, 0, offsetof(MatroskaChapter,uid) },
359     { MATROSKA_ID_CHAPTERDISPLAY,     EBML_NEST, 0, 0, {.n=matroska_chapter_display} },
360     { MATROSKA_ID_CHAPTERFLAGHIDDEN,  EBML_NONE },
361     { EBML_ID_VOID,                   EBML_NONE },
362     { 0 }
363 };
364
365 static EbmlSyntax matroska_chapter[] = {
366     { MATROSKA_ID_CHAPTERATOM,        EBML_NEST, sizeof(MatroskaChapter), offsetof(MatroskaDemuxContext,chapters), {.n=matroska_chapter_entry} },
367     { MATROSKA_ID_EDITIONUID,         EBML_NONE },
368     { MATROSKA_ID_EDITIONFLAGHIDDEN,  EBML_NONE },
369     { MATROSKA_ID_EDITIONFLAGDEFAULT, EBML_NONE },
370     { EBML_ID_VOID,                   EBML_NONE },
371     { 0 }
372 };
373
374 static EbmlSyntax matroska_chapters[] = {
375     { MATROSKA_ID_EDITIONENTRY,       EBML_NEST, 0, 0, {.n=matroska_chapter} },
376     { EBML_ID_VOID,                   EBML_NONE },
377     { 0 }
378 };
379
380 static EbmlSyntax matroska_index_pos[] = {
381     { MATROSKA_ID_CUETRACK,           EBML_UINT, 0, offsetof(MatroskaIndexPos,track) },
382     { MATROSKA_ID_CUECLUSTERPOSITION, EBML_UINT, 0, offsetof(MatroskaIndexPos,pos)   },
383     { EBML_ID_VOID,                   EBML_NONE },
384     { 0 }
385 };
386
387 static EbmlSyntax matroska_index_entry[] = {
388     { MATROSKA_ID_CUETIME,            EBML_UINT, 0, offsetof(MatroskaIndex,time) },
389     { MATROSKA_ID_CUETRACKPOSITION,   EBML_NEST, sizeof(MatroskaIndexPos), offsetof(MatroskaIndex,pos), {.n=matroska_index_pos} },
390     { EBML_ID_VOID,                   EBML_NONE },
391     { 0 }
392 };
393
394 static EbmlSyntax matroska_index[] = {
395     { MATROSKA_ID_POINTENTRY,         EBML_NEST, sizeof(MatroskaIndex), offsetof(MatroskaDemuxContext,index), {.n=matroska_index_entry} },
396     { EBML_ID_VOID,                   EBML_NONE },
397     { 0 }
398 };
399
400 static EbmlSyntax matroska_tags[] = {
401     { EBML_ID_VOID,                   EBML_NONE },
402     { 0 }
403 };
404
405 static EbmlSyntax matroska_seekhead_entry[] = {
406     { MATROSKA_ID_SEEKID,             EBML_UINT, 0, offsetof(MatroskaSeekhead,id) },
407     { MATROSKA_ID_SEEKPOSITION,       EBML_UINT, 0, offsetof(MatroskaSeekhead,pos), {.u=-1} },
408     { EBML_ID_VOID,                   EBML_NONE },
409     { 0 }
410 };
411
412 static EbmlSyntax matroska_seekhead[] = {
413     { MATROSKA_ID_SEEKENTRY,          EBML_NEST, sizeof(MatroskaSeekhead), offsetof(MatroskaDemuxContext,seekhead), {.n=matroska_seekhead_entry} },
414     { EBML_ID_VOID,                   EBML_NONE },
415     { 0 }
416 };
417
418 static EbmlSyntax matroska_segment[] = {
419     { MATROSKA_ID_INFO,           EBML_NEST, 0, 0, {.n=matroska_info       } },
420     { MATROSKA_ID_TRACKS,         EBML_NEST, 0, 0, {.n=matroska_tracks     } },
421     { MATROSKA_ID_ATTACHMENTS,    EBML_NEST, 0, 0, {.n=matroska_attachments} },
422     { MATROSKA_ID_CHAPTERS,       EBML_NEST, 0, 0, {.n=matroska_chapters   } },
423     { MATROSKA_ID_CUES,           EBML_NEST, 0, 0, {.n=matroska_index      } },
424     { MATROSKA_ID_TAGS,           EBML_NEST, 0, 0, {.n=matroska_tags       } },
425     { MATROSKA_ID_SEEKHEAD,       EBML_NEST, 0, 0, {.n=matroska_seekhead   } },
426     { MATROSKA_ID_CLUSTER,        EBML_STOP, 0, offsetof(MatroskaDemuxContext,has_cluster_id) },
427     { EBML_ID_VOID,               EBML_NONE },
428     { 0 }
429 };
430
431 static EbmlSyntax matroska_segments[] = {
432     { MATROSKA_ID_SEGMENT,        EBML_NEST, 0, 0, {.n=matroska_segment    } },
433     { 0 }
434 };
435
436 static EbmlSyntax matroska_blockgroup[] = {
437     { MATROSKA_ID_BLOCK,          EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
438     { MATROSKA_ID_SIMPLEBLOCK,    EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
439     { MATROSKA_ID_BLOCKDURATION,  EBML_UINT, 0, offsetof(MatroskaBlock,duration), {.u=AV_NOPTS_VALUE} },
440     { MATROSKA_ID_BLOCKREFERENCE, EBML_UINT, 0, offsetof(MatroskaBlock,reference) },
441     { EBML_ID_VOID,               EBML_NONE },
442     { 0 }
443 };
444
445 static EbmlSyntax matroska_cluster[] = {
446     { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
447     { MATROSKA_ID_BLOCKGROUP,     EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
448     { MATROSKA_ID_SIMPLEBLOCK,    EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
449     { EBML_ID_VOID,               EBML_NONE },
450     { 0 }
451 };
452
453 static EbmlSyntax matroska_clusters[] = {
454     { MATROSKA_ID_CLUSTER,        EBML_NEST, 0, 0, {.n=matroska_cluster} },
455     { 0 }
456 };
457
458 /*
459  * Return: the amount of levels in the hierarchy that the
460  * current element lies higher than the previous one.
461  * The opposite isn't done - that's auto-done using master
462  * element reading.
463  */
464 static int
465 ebml_read_element_level_up (MatroskaDemuxContext *matroska)
466 {
467     ByteIOContext *pb = matroska->ctx->pb;
468     offset_t pos = url_ftell(pb);
469     int num = 0;
470
471     while (matroska->num_levels > 0) {
472         MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
473
474         if (pos >= level->start + level->length) {
475             matroska->num_levels--;
476             num++;
477         } else {
478             break;
479         }
480     }
481
482     return num;
483 }
484
485 /*
486  * Read: an "EBML number", which is defined as a variable-length
487  * array of bytes. The first byte indicates the length by giving a
488  * number of 0-bits followed by a one. The position of the first
489  * "one" bit inside the first byte indicates the length of this
490  * number.
491  * Returns: num. of bytes read. < 0 on error.
492  */
493 static int
494 ebml_read_num (MatroskaDemuxContext *matroska,
495                int                   max_size,
496                uint64_t             *number)
497 {
498     ByteIOContext *pb = matroska->ctx->pb;
499     int len_mask = 0x80, read = 1, n = 1;
500     int64_t total = 0;
501
502     /* the first byte tells us the length in bytes - get_byte() can normally
503      * return 0, but since that's not a valid first ebmlID byte, we can
504      * use it safely here to catch EOS. */
505     if (!(total = get_byte(pb))) {
506         /* we might encounter EOS here */
507         if (!url_feof(pb)) {
508             offset_t pos = url_ftell(pb);
509             av_log(matroska->ctx, AV_LOG_ERROR,
510                    "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
511                    pos, pos);
512         }
513         return AVERROR(EIO); /* EOS or actual I/O error */
514     }
515
516     /* get the length of the EBML number */
517     while (read <= max_size && !(total & len_mask)) {
518         read++;
519         len_mask >>= 1;
520     }
521     if (read > max_size) {
522         offset_t pos = url_ftell(pb) - 1;
523         av_log(matroska->ctx, AV_LOG_ERROR,
524                "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
525                (uint8_t) total, pos, pos);
526         return AVERROR_INVALIDDATA;
527     }
528
529     /* read out length */
530     total &= ~len_mask;
531     while (n++ < read)
532         total = (total << 8) | get_byte(pb);
533
534     *number = total;
535
536     return read;
537 }
538
539 /*
540  * Read: the element content data ID.
541  * Return: the number of bytes read or < 0 on error.
542  */
543 static int
544 ebml_read_element_id (MatroskaDemuxContext *matroska,
545                       uint32_t             *id)
546 {
547     int read;
548     uint64_t total;
549
550     /* if we re-call this, use our cached ID */
551     if (matroska->peek_id != 0) {
552         *id = matroska->peek_id;
553         return 0;
554     }
555
556     /* read out the "EBML number", include tag in ID */
557     if ((read = ebml_read_num(matroska, 4, &total)) < 0)
558         return read;
559     *id = matroska->peek_id  = total | (1 << (read * 7));
560
561     return read;
562 }
563
564 /*
565  * Read: element content length.
566  * Return: the number of bytes read or < 0 on error.
567  */
568 static int
569 ebml_read_element_length (MatroskaDemuxContext *matroska,
570                           uint64_t             *length)
571 {
572     /* clear cache since we're now beyond that data point */
573     matroska->peek_id = 0;
574
575     /* read out the "EBML number", include tag in ID */
576     return ebml_read_num(matroska, 8, length);
577 }
578
579 /*
580  * Seek to a given offset.
581  * 0 is success, -1 is failure.
582  */
583 static int
584 ebml_read_seek (MatroskaDemuxContext *matroska,
585                 offset_t              offset)
586 {
587     ByteIOContext *pb = matroska->ctx->pb;
588
589     /* clear ID cache, if any */
590     matroska->peek_id = 0;
591
592     return (url_fseek(pb, offset, SEEK_SET) == offset) ? 0 : -1;
593 }
594
595 /*
596  * Read the next element as an unsigned int.
597  * 0 is success, < 0 is failure.
598  */
599 static int
600 ebml_read_uint (ByteIOContext *pb,
601                 int            size,
602                 uint64_t             *num)
603 {
604     int n = 0;
605
606     if (size < 1 || size > 8)
607         return AVERROR_INVALIDDATA;
608
609     /* big-endian ordening; build up number */
610     *num = 0;
611     while (n++ < size)
612         *num = (*num << 8) | get_byte(pb);
613
614     return 0;
615 }
616
617 /*
618  * Read the next element as a float.
619  * 0 is success, < 0 is failure.
620  */
621 static int
622 ebml_read_float (ByteIOContext *pb,
623                  int            size,
624                  double               *num)
625 {
626     if (size == 4) {
627         *num= av_int2flt(get_be32(pb));
628     } else if(size==8){
629         *num= av_int2dbl(get_be64(pb));
630     } else
631         return AVERROR_INVALIDDATA;
632
633     return 0;
634 }
635
636 /*
637  * Read the next element as an ASCII string.
638  * 0 is success, < 0 is failure.
639  */
640 static int
641 ebml_read_ascii (ByteIOContext *pb,
642                  int            size,
643                  char                **str)
644 {
645     av_free(*str);
646     /* ebml strings are usually not 0-terminated, so we allocate one
647      * byte more, read the string and NULL-terminate it ourselves. */
648     if (!(*str = av_malloc(size + 1)))
649         return AVERROR(ENOMEM);
650     if (get_buffer(pb, (uint8_t *) *str, size) != size) {
651         av_free(*str);
652         return AVERROR(EIO);
653     }
654     (*str)[size] = '\0';
655
656     return 0;
657 }
658
659 /*
660  * Read the next element, but only the header. The contents
661  * are supposed to be sub-elements which can be read separately.
662  * 0 is success, < 0 is failure.
663  */
664 static int
665 ebml_read_master (MatroskaDemuxContext *matroska,
666                   int                   length)
667 {
668     ByteIOContext *pb = matroska->ctx->pb;
669     MatroskaLevel *level;
670
671     if (matroska->num_levels >= EBML_MAX_DEPTH) {
672         av_log(matroska->ctx, AV_LOG_ERROR,
673                "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
674         return AVERROR(ENOSYS);
675     }
676
677     level = &matroska->levels[matroska->num_levels++];
678     level->start = url_ftell(pb);
679     level->length = length;
680
681     return 0;
682 }
683
684 /*
685  * Read the next element as binary data.
686  * 0 is success, < 0 is failure.
687  */
688 static int
689 ebml_read_binary (ByteIOContext *pb,
690                   int            length,
691                   EbmlBin       *bin)
692 {
693     av_free(bin->data);
694     if (!(bin->data = av_malloc(length)))
695         return AVERROR(ENOMEM);
696
697     bin->size = length;
698     bin->pos  = url_ftell(pb);
699     if (get_buffer(pb, bin->data, length) != length)
700         return AVERROR(EIO);
701
702     return 0;
703 }
704
705 /*
706  * Read signed/unsigned "EBML" numbers.
707  * Return: number of bytes processed, < 0 on error.
708  * XXX: use ebml_read_num().
709  */
710 static int
711 matroska_ebmlnum_uint (uint8_t  *data,
712                        uint32_t  size,
713                        uint64_t *num)
714 {
715     int len_mask = 0x80, read = 1, n = 1, num_ffs = 0;
716     uint64_t total;
717
718     if (size <= 0)
719         return AVERROR_INVALIDDATA;
720
721     total = data[0];
722     while (read <= 8 && !(total & len_mask)) {
723         read++;
724         len_mask >>= 1;
725     }
726     if (read > 8)
727         return AVERROR_INVALIDDATA;
728
729     if ((total &= (len_mask - 1)) == len_mask - 1)
730         num_ffs++;
731     if (size < read)
732         return AVERROR_INVALIDDATA;
733     while (n < read) {
734         if (data[n] == 0xff)
735             num_ffs++;
736         total = (total << 8) | data[n];
737         n++;
738     }
739
740     if (read == num_ffs)
741         *num = (uint64_t)-1;
742     else
743         *num = total;
744
745     return read;
746 }
747
748 /*
749  * Same as above, but signed.
750  */
751 static int
752 matroska_ebmlnum_sint (uint8_t  *data,
753                        uint32_t  size,
754                        int64_t  *num)
755 {
756     uint64_t unum;
757     int res;
758
759     /* read as unsigned number first */
760     if ((res = matroska_ebmlnum_uint(data, size, &unum)) < 0)
761         return res;
762
763     /* make signed (weird way) */
764     if (unum == (uint64_t)-1)
765         *num = INT64_MAX;
766     else
767         *num = unum - ((1LL << ((7 * res) - 1)) - 1);
768
769     return res;
770 }
771
772
773 static MatroskaTrack *
774 matroska_find_track_by_num (MatroskaDemuxContext *matroska,
775                             int                   num)
776 {
777     MatroskaTrack *tracks = matroska->tracks.elem;
778     int i;
779
780     for (i=0; i < matroska->tracks.nb_elem; i++)
781         if (tracks[i].num == num)
782             return &tracks[i];
783
784     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid track number %d\n", num);
785     return NULL;
786 }
787
788
789 /*
790  * Put one packet in an application-supplied AVPacket struct.
791  * Returns 0 on success or -1 on failure.
792  */
793 static int
794 matroska_deliver_packet (MatroskaDemuxContext *matroska,
795                          AVPacket             *pkt)
796 {
797     if (matroska->num_packets > 0) {
798         memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
799         av_free(matroska->packets[0]);
800         if (matroska->num_packets > 1) {
801             memmove(&matroska->packets[0], &matroska->packets[1],
802                     (matroska->num_packets - 1) * sizeof(AVPacket *));
803             matroska->packets =
804                 av_realloc(matroska->packets, (matroska->num_packets - 1) *
805                            sizeof(AVPacket *));
806         } else {
807             av_freep(&matroska->packets);
808         }
809         matroska->num_packets--;
810         return 0;
811     }
812
813     return -1;
814 }
815
816 /*
817  * Put a packet into our internal queue. Will be delivered to the
818  * user/application during the next get_packet() call.
819  */
820 static void
821 matroska_queue_packet (MatroskaDemuxContext *matroska,
822                        AVPacket             *pkt)
823 {
824     matroska->packets =
825         av_realloc(matroska->packets, (matroska->num_packets + 1) *
826                    sizeof(AVPacket *));
827     matroska->packets[matroska->num_packets] = pkt;
828     matroska->num_packets++;
829 }
830
831 /*
832  * Free all packets in our internal queue.
833  */
834 static void
835 matroska_clear_queue (MatroskaDemuxContext *matroska)
836 {
837     if (matroska->packets) {
838         int n;
839         for (n = 0; n < matroska->num_packets; n++) {
840             av_free_packet(matroska->packets[n]);
841             av_free(matroska->packets[n]);
842         }
843         av_free(matroska->packets);
844         matroska->packets = NULL;
845         matroska->num_packets = 0;
846     }
847 }
848
849
850 /*
851  * Autodetecting...
852  */
853 static int
854 matroska_probe (AVProbeData *p)
855 {
856     uint64_t total = 0;
857     int len_mask = 0x80, size = 1, n = 1;
858     uint8_t probe_data[] = { 'm', 'a', 't', 'r', 'o', 's', 'k', 'a' };
859
860     /* ebml header? */
861     if (AV_RB32(p->buf) != EBML_ID_HEADER)
862         return 0;
863
864     /* length of header */
865     total = p->buf[4];
866     while (size <= 8 && !(total & len_mask)) {
867         size++;
868         len_mask >>= 1;
869     }
870     if (size > 8)
871       return 0;
872     total &= (len_mask - 1);
873     while (n < size)
874         total = (total << 8) | p->buf[4 + n++];
875
876     /* does the probe data contain the whole header? */
877     if (p->buf_size < 4 + size + total)
878       return 0;
879
880     /* the header must contain the document type 'matroska'. For now,
881      * we don't parse the whole header but simply check for the
882      * availability of that array of characters inside the header.
883      * Not fully fool-proof, but good enough. */
884     for (n = 4 + size; n <= 4 + size + total - sizeof(probe_data); n++)
885         if (!memcmp (&p->buf[n], probe_data, sizeof(probe_data)))
886             return AVPROBE_SCORE_MAX;
887
888     return 0;
889 }
890
891 static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
892                       void *data, int once);
893
894 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
895                            EbmlSyntax *syntax, void *data)
896 {
897     ByteIOContext *pb = matroska->ctx->pb;
898     uint32_t id = syntax->id;
899     uint64_t length;
900     int res;
901
902     data = (char *)data + syntax->data_offset;
903     if (syntax->list_elem_size) {
904         EbmlList *list = data;
905         list->elem = av_realloc(list->elem, (list->nb_elem+1)*syntax->list_elem_size);
906         data = (char*)list->elem + list->nb_elem*syntax->list_elem_size;
907         memset(data, 0, syntax->list_elem_size);
908         list->nb_elem++;
909     }
910
911     if (syntax->type != EBML_PASS && syntax->type != EBML_STOP)
912         if ((res = ebml_read_element_id(matroska, &id)) < 0 ||
913             (res = ebml_read_element_length(matroska, &length)) < 0)
914             return res;
915
916     switch (syntax->type) {
917     case EBML_UINT:  res = ebml_read_uint  (pb, length, data);  break;
918     case EBML_FLOAT: res = ebml_read_float (pb, length, data);  break;
919     case EBML_STR:
920     case EBML_UTF8:  res = ebml_read_ascii (pb, length, data);  break;
921     case EBML_BIN:   res = ebml_read_binary(pb, length, data);  break;
922     case EBML_NEST:  if ((res=ebml_read_master(matroska, length)) < 0)
923                          return res;
924                      if (id == MATROSKA_ID_SEGMENT)
925                          matroska->segment_start = url_ftell(matroska->ctx->pb);
926                      return ebml_parse(matroska, syntax->def.n, data, 0);
927     case EBML_PASS:  return ebml_parse(matroska, syntax->def.n, data, 1);
928     case EBML_STOP:  *(int *)data = 1;      return 1;
929     default:         url_fskip(pb, length); return 0;
930     }
931     if (res == AVERROR_INVALIDDATA)
932         av_log(matroska->ctx, AV_LOG_ERROR, "Invalid element\n");
933     else if (res == AVERROR(EIO))
934         av_log(matroska->ctx, AV_LOG_ERROR, "Read error\n");
935     return res;
936 }
937
938 static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
939                          uint32_t id, void *data)
940 {
941     int i;
942     for (i=0; syntax[i].id; i++)
943         if (id == syntax[i].id)
944             break;
945     if (!syntax[i].id)
946         av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%X\n", id);
947     return ebml_parse_elem(matroska, &syntax[i], data);
948 }
949
950 static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
951                       void *data, int once)
952 {
953     int i, res = 0, res2;
954     uint32_t id = 0;
955
956     for (i=0; syntax[i].id; i++)
957         switch (syntax[i].type) {
958         case EBML_UINT:
959             *(uint64_t *)((char *)data+syntax[i].data_offset) = syntax[i].def.u;
960             break;
961         case EBML_FLOAT:
962             *(double   *)((char *)data+syntax[i].data_offset) = syntax[i].def.f;
963             break;
964         case EBML_STR:
965         case EBML_UTF8:
966             *(char    **)((char *)data+syntax[i].data_offset) = av_strdup(syntax[i].def.s);
967             break;
968         }
969
970     while (!res) {
971         res2 = ebml_read_element_id(matroska, &id);
972         if (res2 < 0)
973             break;
974         if (res2 > 0)
975             matroska->level_up = ebml_read_element_level_up(matroska);
976         if (matroska->level_up) {
977             matroska->level_up--;
978             break;
979         }
980
981         res = ebml_parse_id(matroska, syntax, id, data);
982         if (once)
983             break;
984     }
985
986     return res;
987 }
988
989 static void ebml_free(EbmlSyntax *syntax, void *data)
990 {
991     int i, j;
992     for (i=0; syntax[i].id; i++) {
993         void *data_off = (char *)data + syntax[i].data_offset;
994         switch (syntax[i].type) {
995         case EBML_STR:
996         case EBML_UTF8:  av_freep(data_off);                      break;
997         case EBML_BIN:   av_freep(&((EbmlBin *)data_off)->data);  break;
998         case EBML_NEST:
999             if (syntax[i].list_elem_size) {
1000                 EbmlList *list = data_off;
1001                 char *ptr = list->elem;
1002                 for (j=0; j<list->nb_elem; j++, ptr+=syntax[i].list_elem_size)
1003                     ebml_free(syntax[i].def.n, ptr);
1004                 av_free(list->elem);
1005             } else
1006                 ebml_free(syntax[i].def.n, data_off);
1007         default:  break;
1008         }
1009     }
1010 }
1011
1012 static int
1013 matroska_decode_buffer(uint8_t** buf, int* buf_size, MatroskaTrack *track)
1014 {
1015     MatroskaTrackEncoding *encodings = track->encodings.elem;
1016     uint8_t* data = *buf;
1017     int isize = *buf_size;
1018     uint8_t* pkt_data = NULL;
1019     int pkt_size = isize;
1020     int result = 0;
1021     int olen;
1022
1023     switch (encodings[0].compression.algo) {
1024     case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
1025         return encodings[0].compression.settings.size;
1026     case MATROSKA_TRACK_ENCODING_COMP_LZO:
1027         do {
1028             olen = pkt_size *= 3;
1029             pkt_data = av_realloc(pkt_data,
1030                                   pkt_size+LZO_OUTPUT_PADDING);
1031             result = lzo1x_decode(pkt_data, &olen, data, &isize);
1032         } while (result==LZO_OUTPUT_FULL && pkt_size<10000000);
1033         if (result)
1034             goto failed;
1035         pkt_size -= olen;
1036         break;
1037 #ifdef CONFIG_ZLIB
1038     case MATROSKA_TRACK_ENCODING_COMP_ZLIB: {
1039         z_stream zstream = {0};
1040         if (inflateInit(&zstream) != Z_OK)
1041             return -1;
1042         zstream.next_in = data;
1043         zstream.avail_in = isize;
1044         do {
1045             pkt_size *= 3;
1046             pkt_data = av_realloc(pkt_data, pkt_size);
1047             zstream.avail_out = pkt_size - zstream.total_out;
1048             zstream.next_out = pkt_data + zstream.total_out;
1049             result = inflate(&zstream, Z_NO_FLUSH);
1050         } while (result==Z_OK && pkt_size<10000000);
1051         pkt_size = zstream.total_out;
1052         inflateEnd(&zstream);
1053         if (result != Z_STREAM_END)
1054             goto failed;
1055         break;
1056     }
1057 #endif
1058 #ifdef CONFIG_BZLIB
1059     case MATROSKA_TRACK_ENCODING_COMP_BZLIB: {
1060         bz_stream bzstream = {0};
1061         if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
1062             return -1;
1063         bzstream.next_in = data;
1064         bzstream.avail_in = isize;
1065         do {
1066             pkt_size *= 3;
1067             pkt_data = av_realloc(pkt_data, pkt_size);
1068             bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
1069             bzstream.next_out = pkt_data + bzstream.total_out_lo32;
1070             result = BZ2_bzDecompress(&bzstream);
1071         } while (result==BZ_OK && pkt_size<10000000);
1072         pkt_size = bzstream.total_out_lo32;
1073         BZ2_bzDecompressEnd(&bzstream);
1074         if (result != BZ_STREAM_END)
1075             goto failed;
1076         break;
1077     }
1078 #endif
1079     }
1080
1081     *buf = pkt_data;
1082     *buf_size = pkt_size;
1083     return 0;
1084  failed:
1085     av_free(pkt_data);
1086     return -1;
1087 }
1088
1089 static void
1090 matroska_execute_seekhead(MatroskaDemuxContext *matroska)
1091 {
1092     EbmlList *seekhead_list = &matroska->seekhead;
1093     MatroskaSeekhead *seekhead = seekhead_list->elem;
1094     uint32_t peek_id_cache = matroska->peek_id;
1095     uint32_t level_up = matroska->level_up;
1096     offset_t before_pos = url_ftell(matroska->ctx->pb);
1097     MatroskaLevel level;
1098     int i;
1099
1100     for (i=0; i<seekhead_list->nb_elem; i++) {
1101         if (seekhead[i].pos <= before_pos
1102             || seekhead[i].id == MATROSKA_ID_SEEKHEAD
1103             || seekhead[i].id == MATROSKA_ID_CLUSTER)
1104             continue;
1105
1106         /* seek */
1107         if (ebml_read_seek(matroska,
1108                            seekhead[i].pos+matroska->segment_start) < 0)
1109             continue;
1110
1111         /* we don't want to lose our seekhead level, so we add
1112          * a dummy. This is a crude hack. */
1113         if (matroska->num_levels == EBML_MAX_DEPTH) {
1114             av_log(matroska->ctx, AV_LOG_INFO,
1115                    "Max EBML element depth (%d) reached, "
1116                    "cannot parse further.\n", EBML_MAX_DEPTH);
1117             break;
1118         }
1119
1120         level.start = 0;
1121         level.length = (uint64_t)-1;
1122         matroska->levels[matroska->num_levels] = level;
1123         matroska->num_levels++;
1124
1125         ebml_parse_id(matroska, matroska_segment, seekhead[i].id, matroska);
1126
1127         /* remove dummy level */
1128         while (matroska->num_levels) {
1129             uint64_t length = matroska->levels[--matroska->num_levels].length;
1130             if (length == (uint64_t)-1)
1131                 break;
1132         }
1133     }
1134
1135     /* seek back */
1136     ebml_read_seek(matroska, before_pos);
1137     matroska->peek_id = peek_id_cache;
1138     matroska->level_up = level_up;
1139 }
1140
1141 static int
1142 matroska_aac_profile (char *codec_id)
1143 {
1144     static const char *aac_profiles[] = { "MAIN", "LC", "SSR" };
1145     int profile;
1146
1147     for (profile=0; profile<ARRAY_SIZE(aac_profiles); profile++)
1148         if (strstr(codec_id, aac_profiles[profile]))
1149             break;
1150     return profile + 1;
1151 }
1152
1153 static int
1154 matroska_aac_sri (int samplerate)
1155 {
1156     int sri;
1157
1158     for (sri=0; sri<ARRAY_SIZE(ff_mpeg4audio_sample_rates); sri++)
1159         if (ff_mpeg4audio_sample_rates[sri] == samplerate)
1160             break;
1161     return sri;
1162 }
1163
1164 static int
1165 matroska_read_header (AVFormatContext    *s,
1166                       AVFormatParameters *ap)
1167 {
1168     MatroskaDemuxContext *matroska = s->priv_data;
1169     EbmlList *attachements_list = &matroska->attachments;
1170     MatroskaAttachement *attachements;
1171     EbmlList *chapters_list = &matroska->chapters;
1172     MatroskaChapter *chapters;
1173     MatroskaTrack *tracks;
1174     EbmlList *index_list;
1175     MatroskaIndex *index;
1176     Ebml ebml = { 0 };
1177     AVStream *st;
1178     int i, j;
1179
1180     matroska->ctx = s;
1181
1182     /* First read the EBML header. */
1183     if (ebml_parse(matroska, ebml_syntax, &ebml, 1)
1184         || ebml.version > EBML_VERSION       || ebml.max_size > sizeof(uint64_t)
1185         || ebml.id_length > sizeof(uint32_t) || strcmp(ebml.doctype, "matroska")
1186         || ebml.doctype_version > 2) {
1187         av_log(matroska->ctx, AV_LOG_ERROR,
1188                "EBML header using unsupported features\n"
1189                "(EBML version %"PRIu64", doctype %s, doc version %"PRIu64")\n",
1190                ebml.version, ebml.doctype, ebml.doctype_version);
1191         return AVERROR_NOFMT;
1192     }
1193     ebml_free(ebml_syntax, &ebml);
1194
1195     /* The next thing is a segment. */
1196     if (ebml_parse(matroska, matroska_segments, matroska, 1) < 0)
1197         return -1;
1198     matroska_execute_seekhead(matroska);
1199
1200     if (matroska->duration)
1201         matroska->ctx->duration = matroska->duration * matroska->time_scale
1202                                   * 1000 / AV_TIME_BASE;
1203     if (matroska->title)
1204         strncpy(matroska->ctx->title, matroska->title,
1205                 sizeof(matroska->ctx->title)-1);
1206
1207     tracks = matroska->tracks.elem;
1208     for (i=0; i < matroska->tracks.nb_elem; i++) {
1209         MatroskaTrack *track = &tracks[i];
1210         enum CodecID codec_id = CODEC_ID_NONE;
1211         EbmlList *encodings_list = &tracks->encodings;
1212         MatroskaTrackEncoding *encodings = encodings_list->elem;
1213         uint8_t *extradata = NULL;
1214         int extradata_size = 0;
1215         int extradata_offset = 0;
1216
1217         /* Apply some sanity checks. */
1218         if (track->type != MATROSKA_TRACK_TYPE_VIDEO &&
1219             track->type != MATROSKA_TRACK_TYPE_AUDIO &&
1220             track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1221             av_log(matroska->ctx, AV_LOG_INFO,
1222                    "Unknown or unsupported track type %"PRIu64"\n",
1223                    track->type);
1224             continue;
1225         }
1226         if (track->codec_id == NULL)
1227             continue;
1228
1229         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
1230             if (!track->default_duration)
1231                 track->default_duration = 1000000000/track->video.frame_rate;
1232             if (!track->video.display_width)
1233                 track->video.display_width = track->video.pixel_width;
1234             if (!track->video.display_height)
1235                 track->video.display_height = track->video.pixel_height;
1236         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1237             if (!track->audio.out_samplerate)
1238                 track->audio.out_samplerate = track->audio.samplerate;
1239         }
1240         if (encodings_list->nb_elem > 1) {
1241             av_log(matroska->ctx, AV_LOG_ERROR,
1242                    "Multiple combined encodings no supported");
1243         } else if (encodings_list->nb_elem == 1) {
1244             if (encodings[0].type ||
1245                 (encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP &&
1246 #ifdef CONFIG_ZLIB
1247                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
1248 #endif
1249 #ifdef CONFIG_BZLIB
1250                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
1251 #endif
1252                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_LZO)) {
1253                 encodings[0].scope = 0;
1254                 av_log(matroska->ctx, AV_LOG_ERROR,
1255                        "Unsupported encoding type");
1256             } else if (track->codec_priv.size && encodings[0].scope&2) {
1257                 uint8_t *codec_priv = track->codec_priv.data;
1258                 int offset = matroska_decode_buffer(&track->codec_priv.data,
1259                                                     &track->codec_priv.size,
1260                                                     track);
1261                 if (offset < 0) {
1262                     track->codec_priv.data = NULL;
1263                     track->codec_priv.size = 0;
1264                     av_log(matroska->ctx, AV_LOG_ERROR,
1265                            "Failed to decode codec private data\n");
1266                 } else if (offset > 0) {
1267                     track->codec_priv.data = av_malloc(track->codec_priv.size + offset);
1268                     memcpy(track->codec_priv.data,
1269                            encodings[0].compression.settings.data, offset);
1270                     memcpy(track->codec_priv.data+offset, codec_priv,
1271                            track->codec_priv.size);
1272                     track->codec_priv.size += offset;
1273                 }
1274                 if (codec_priv != track->codec_priv.data)
1275                     av_free(codec_priv);
1276             }
1277         }
1278
1279         for(j=0; ff_mkv_codec_tags[j].id != CODEC_ID_NONE; j++){
1280             if(!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
1281                         strlen(ff_mkv_codec_tags[j].str))){
1282                 codec_id= ff_mkv_codec_tags[j].id;
1283                 break;
1284             }
1285         }
1286
1287         st = track->stream = av_new_stream(s, matroska->num_streams++);
1288         if (st == NULL)
1289             return AVERROR(ENOMEM);
1290
1291         if (!strcmp(track->codec_id, MATROSKA_CODEC_ID_VIDEO_VFW_FOURCC)
1292             && track->codec_priv.size >= 40
1293             && track->codec_priv.data != NULL) {
1294             track->video.fourcc = AV_RL32(track->codec_priv.data + 16);
1295             codec_id = codec_get_id(codec_bmp_tags, track->video.fourcc);
1296         } else if (!strcmp(track->codec_id, MATROSKA_CODEC_ID_AUDIO_ACM)
1297                    && track->codec_priv.size >= 18
1298                    && track->codec_priv.data != NULL) {
1299             uint16_t tag = AV_RL16(track->codec_priv.data);
1300             codec_id = codec_get_id(codec_wav_tags, tag);
1301         } else if (!strcmp(track->codec_id, "V_QUICKTIME")
1302                    && (track->codec_priv.size >= 86)
1303                    && (track->codec_priv.data != NULL)) {
1304             track->video.fourcc = AV_RL32(track->codec_priv.data);
1305             codec_id=codec_get_id(codec_movvideo_tags, track->video.fourcc);
1306         } else if (codec_id == CODEC_ID_AAC && !track->codec_priv.size) {
1307             int profile = matroska_aac_profile(track->codec_id);
1308             int sri = matroska_aac_sri(track->audio.samplerate);
1309             extradata = av_malloc(5);
1310             if (extradata == NULL)
1311                 return AVERROR(ENOMEM);
1312             extradata[0] = (profile << 3) | ((sri&0x0E) >> 1);
1313             extradata[1] = ((sri&0x01) << 7) | (track->audio.channels<<3);
1314             if (strstr(track->codec_id, "SBR")) {
1315                 sri = matroska_aac_sri(track->audio.out_samplerate);
1316                 extradata[2] = 0x56;
1317                 extradata[3] = 0xE5;
1318                 extradata[4] = 0x80 | (sri<<3);
1319                 extradata_size = 5;
1320             } else
1321                 extradata_size = 2;
1322         } else if (codec_id == CODEC_ID_TTA) {
1323             ByteIOContext b;
1324             extradata_size = 30;
1325             extradata = av_mallocz(extradata_size);
1326             if (extradata == NULL)
1327                 return AVERROR(ENOMEM);
1328             init_put_byte(&b, extradata, extradata_size, 1,
1329                           NULL, NULL, NULL, NULL);
1330             put_buffer(&b, "TTA1", 4);
1331             put_le16(&b, 1);
1332             put_le16(&b, track->audio.channels);
1333             put_le16(&b, track->audio.bitdepth);
1334             put_le32(&b, track->audio.out_samplerate);
1335             put_le32(&b, matroska->ctx->duration * track->audio.out_samplerate);
1336         } else if (codec_id == CODEC_ID_RV10 || codec_id == CODEC_ID_RV20 ||
1337                    codec_id == CODEC_ID_RV30 || codec_id == CODEC_ID_RV40) {
1338             extradata_offset = 26;
1339             track->codec_priv.size -= extradata_offset;
1340         } else if (codec_id == CODEC_ID_RA_144) {
1341             track->audio.out_samplerate = 8000;
1342             track->audio.channels = 1;
1343         } else if (codec_id == CODEC_ID_RA_288 || codec_id == CODEC_ID_COOK ||
1344                    codec_id == CODEC_ID_ATRAC3) {
1345             ByteIOContext b;
1346
1347             init_put_byte(&b, track->codec_priv.data,track->codec_priv.size,
1348                           0, NULL, NULL, NULL, NULL);
1349             url_fskip(&b, 24);
1350             track->audio.coded_framesize = get_be32(&b);
1351             url_fskip(&b, 12);
1352             track->audio.sub_packet_h    = get_be16(&b);
1353             track->audio.frame_size      = get_be16(&b);
1354             track->audio.sub_packet_size = get_be16(&b);
1355             track->audio.buf = av_malloc(track->audio.frame_size * track->audio.sub_packet_h);
1356             if (codec_id == CODEC_ID_RA_288) {
1357                 st->codec->block_align = track->audio.coded_framesize;
1358                 track->codec_priv.size = 0;
1359             } else {
1360                 st->codec->block_align = track->audio.sub_packet_size;
1361                 extradata_offset = 78;
1362                 track->codec_priv.size -= extradata_offset;
1363             }
1364         }
1365
1366         if (codec_id == CODEC_ID_NONE)
1367             av_log(matroska->ctx, AV_LOG_INFO,
1368                    "Unknown/unsupported CodecID %s.\n", track->codec_id);
1369
1370         av_set_pts_info(st, 64, matroska->time_scale*track->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
1371
1372         st->codec->codec_id = codec_id;
1373         st->start_time = 0;
1374         if (strcmp(track->language, "und"))
1375             av_strlcpy(st->language, track->language, 4);
1376
1377         if (track->flag_default)
1378             st->disposition |= AV_DISPOSITION_DEFAULT;
1379
1380         if (track->default_duration)
1381             av_reduce(&st->codec->time_base.num, &st->codec->time_base.den,
1382                       track->default_duration, 1000000000, 30000);
1383
1384         if(extradata){
1385             st->codec->extradata = extradata;
1386             st->codec->extradata_size = extradata_size;
1387         } else if(track->codec_priv.data && track->codec_priv.size > 0){
1388             st->codec->extradata = av_malloc(track->codec_priv.size);
1389             if(st->codec->extradata == NULL)
1390                 return AVERROR(ENOMEM);
1391             st->codec->extradata_size = track->codec_priv.size;
1392             memcpy(st->codec->extradata,
1393                    track->codec_priv.data + extradata_offset,
1394                    track->codec_priv.size);
1395         }
1396
1397         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
1398             st->codec->codec_type = CODEC_TYPE_VIDEO;
1399             st->codec->codec_tag  = track->video.fourcc;
1400             st->codec->width  = track->video.pixel_width;
1401             st->codec->height = track->video.pixel_height;
1402             av_reduce(&st->codec->sample_aspect_ratio.num,
1403                       &st->codec->sample_aspect_ratio.den,
1404                       st->codec->height * track->video.display_width,
1405                       st->codec-> width * track->video.display_height,
1406                       255);
1407             st->need_parsing = AVSTREAM_PARSE_HEADERS;
1408         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1409             st->codec->codec_type = CODEC_TYPE_AUDIO;
1410             st->codec->sample_rate = track->audio.out_samplerate;
1411             st->codec->channels = track->audio.channels;
1412         } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
1413             st->codec->codec_type = CODEC_TYPE_SUBTITLE;
1414         }
1415     }
1416
1417     attachements = attachements_list->elem;
1418     for (j=0; j<attachements_list->nb_elem; j++) {
1419         if (!(attachements[j].filename && attachements[j].mime &&
1420               attachements[j].bin.data && attachements[j].bin.size > 0)) {
1421             av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
1422         } else {
1423             AVStream *st = av_new_stream(s, matroska->num_streams++);
1424             if (st == NULL)
1425                 break;
1426             st->filename          = av_strdup(attachements[j].filename);
1427             st->codec->codec_id = CODEC_ID_NONE;
1428             st->codec->codec_type = CODEC_TYPE_ATTACHMENT;
1429             st->codec->extradata  = av_malloc(attachements[j].bin.size);
1430             if(st->codec->extradata == NULL)
1431                 break;
1432             st->codec->extradata_size = attachements[j].bin.size;
1433             memcpy(st->codec->extradata, attachements[j].bin.data, attachements[j].bin.size);
1434
1435             for (i=0; ff_mkv_mime_tags[i].id != CODEC_ID_NONE; i++) {
1436                 if (!strncmp(ff_mkv_mime_tags[i].str, attachements[j].mime,
1437                              strlen(ff_mkv_mime_tags[i].str))) {
1438                     st->codec->codec_id = ff_mkv_mime_tags[i].id;
1439                     break;
1440                 }
1441             }
1442         }
1443     }
1444
1445     chapters = chapters_list->elem;
1446     for (i=0; i<chapters_list->nb_elem; i++)
1447         if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid)
1448             ff_new_chapter(s, chapters[i].uid, (AVRational){1, 1000000000},
1449                            chapters[i].start, chapters[i].end,
1450                            chapters[i].title);
1451
1452     index_list = &matroska->index;
1453     index = index_list->elem;
1454     for (i=0; i<index_list->nb_elem; i++) {
1455         EbmlList *pos_list = &index[i].pos;
1456         MatroskaIndexPos *pos = pos_list->elem;
1457         for (j=0; j<pos_list->nb_elem; j++) {
1458             MatroskaTrack *track = matroska_find_track_by_num(matroska,
1459                                                               pos[j].track);
1460             if (track && track->stream)
1461                 av_add_index_entry(track->stream,
1462                                    pos[j].pos + matroska->segment_start,
1463                                    index[i].time*matroska->time_scale/AV_TIME_BASE,
1464                                    0, 0, AVINDEX_KEYFRAME);
1465         }
1466     }
1467
1468     return 0;
1469 }
1470
1471 static int
1472 matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, int size,
1473                      int64_t pos, uint64_t cluster_time, uint64_t duration,
1474                      int is_keyframe)
1475 {
1476     MatroskaTrack *track;
1477     int res = 0;
1478     AVStream *st;
1479     AVPacket *pkt;
1480     int16_t block_time;
1481     uint32_t *lace_size = NULL;
1482     int n, flags, laces = 0;
1483     uint64_t num;
1484
1485     if ((n = matroska_ebmlnum_uint(data, size, &num)) < 0) {
1486         av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
1487         return res;
1488     }
1489     data += n;
1490     size -= n;
1491
1492     track = matroska_find_track_by_num(matroska, num);
1493     if (size <= 3 || !track || !track->stream) {
1494         av_log(matroska->ctx, AV_LOG_INFO,
1495                "Invalid stream %"PRIu64" or size %u\n", num, size);
1496         return res;
1497     }
1498     st = track->stream;
1499     if (st->discard >= AVDISCARD_ALL)
1500         return res;
1501     if (duration == AV_NOPTS_VALUE)
1502         duration = track->default_duration / matroska->time_scale;
1503
1504     block_time = AV_RB16(data);
1505     data += 2;
1506     flags = *data++;
1507     size -= 3;
1508     if (is_keyframe == -1)
1509         is_keyframe = flags & 0x80 ? PKT_FLAG_KEY : 0;
1510
1511     if (matroska->skip_to_keyframe) {
1512         if (!is_keyframe || st != matroska->skip_to_stream)
1513             return res;
1514         matroska->skip_to_keyframe = 0;
1515     }
1516
1517     switch ((flags & 0x06) >> 1) {
1518         case 0x0: /* no lacing */
1519             laces = 1;
1520             lace_size = av_mallocz(sizeof(int));
1521             lace_size[0] = size;
1522             break;
1523
1524         case 0x1: /* xiph lacing */
1525         case 0x2: /* fixed-size lacing */
1526         case 0x3: /* EBML lacing */
1527             assert(size>0); // size <=3 is checked before size-=3 above
1528             laces = (*data) + 1;
1529             data += 1;
1530             size -= 1;
1531             lace_size = av_mallocz(laces * sizeof(int));
1532
1533             switch ((flags & 0x06) >> 1) {
1534                 case 0x1: /* xiph lacing */ {
1535                     uint8_t temp;
1536                     uint32_t total = 0;
1537                     for (n = 0; res == 0 && n < laces - 1; n++) {
1538                         while (1) {
1539                             if (size == 0) {
1540                                 res = -1;
1541                                 break;
1542                             }
1543                             temp = *data;
1544                             lace_size[n] += temp;
1545                             data += 1;
1546                             size -= 1;
1547                             if (temp != 0xff)
1548                                 break;
1549                         }
1550                         total += lace_size[n];
1551                     }
1552                     lace_size[n] = size - total;
1553                     break;
1554                 }
1555
1556                 case 0x2: /* fixed-size lacing */
1557                     for (n = 0; n < laces; n++)
1558                         lace_size[n] = size / laces;
1559                     break;
1560
1561                 case 0x3: /* EBML lacing */ {
1562                     uint32_t total;
1563                     n = matroska_ebmlnum_uint(data, size, &num);
1564                     if (n < 0) {
1565                         av_log(matroska->ctx, AV_LOG_INFO,
1566                                "EBML block data error\n");
1567                         break;
1568                     }
1569                     data += n;
1570                     size -= n;
1571                     total = lace_size[0] = num;
1572                     for (n = 1; res == 0 && n < laces - 1; n++) {
1573                         int64_t snum;
1574                         int r;
1575                         r = matroska_ebmlnum_sint (data, size, &snum);
1576                         if (r < 0) {
1577                             av_log(matroska->ctx, AV_LOG_INFO,
1578                                    "EBML block data error\n");
1579                             break;
1580                         }
1581                         data += r;
1582                         size -= r;
1583                         lace_size[n] = lace_size[n - 1] + snum;
1584                         total += lace_size[n];
1585                     }
1586                     lace_size[n] = size - total;
1587                     break;
1588                 }
1589             }
1590             break;
1591     }
1592
1593     if (res == 0) {
1594         uint64_t timecode = AV_NOPTS_VALUE;
1595
1596         if (cluster_time != (uint64_t)-1
1597             && (block_time >= 0 || cluster_time >= -block_time))
1598             timecode = cluster_time + block_time;
1599
1600         for (n = 0; n < laces; n++) {
1601             if (st->codec->codec_id == CODEC_ID_RA_288 ||
1602                 st->codec->codec_id == CODEC_ID_COOK ||
1603                 st->codec->codec_id == CODEC_ID_ATRAC3) {
1604                 int a = st->codec->block_align;
1605                 int sps = track->audio.sub_packet_size;
1606                 int cfs = track->audio.coded_framesize;
1607                 int h = track->audio.sub_packet_h;
1608                 int y = track->audio.sub_packet_cnt;
1609                 int w = track->audio.frame_size;
1610                 int x;
1611
1612                 if (!track->audio.pkt_cnt) {
1613                     if (st->codec->codec_id == CODEC_ID_RA_288)
1614                         for (x=0; x<h/2; x++)
1615                             memcpy(track->audio.buf+x*2*w+y*cfs,
1616                                    data+x*cfs, cfs);
1617                     else
1618                         for (x=0; x<w/sps; x++)
1619                             memcpy(track->audio.buf+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), data+x*sps, sps);
1620
1621                     if (++track->audio.sub_packet_cnt >= h) {
1622                         track->audio.sub_packet_cnt = 0;
1623                         track->audio.pkt_cnt = h*w / a;
1624                     }
1625                 }
1626                 while (track->audio.pkt_cnt) {
1627                     pkt = av_mallocz(sizeof(AVPacket));
1628                     av_new_packet(pkt, a);
1629                     memcpy(pkt->data, track->audio.buf
1630                            + a * (h*w / a - track->audio.pkt_cnt--), a);
1631                     pkt->pos = pos;
1632                     pkt->stream_index = st->index;
1633                     matroska_queue_packet(matroska, pkt);
1634                 }
1635             } else {
1636                 MatroskaTrackEncoding *encodings = track->encodings.elem;
1637                 int offset = 0, pkt_size = lace_size[n];
1638                 uint8_t *pkt_data = data;
1639
1640                 if (encodings && encodings->scope & 1) {
1641                     offset = matroska_decode_buffer(&pkt_data,&pkt_size, track);
1642                     if (offset < 0)
1643                         continue;
1644                 }
1645
1646                 pkt = av_mallocz(sizeof(AVPacket));
1647                 /* XXX: prevent data copy... */
1648                 if (av_new_packet(pkt, pkt_size+offset) < 0) {
1649                     av_free(pkt);
1650                     res = AVERROR(ENOMEM);
1651                     n = laces-1;
1652                     break;
1653                 }
1654                 if (offset)
1655                     memcpy (pkt->data, encodings->compression.settings.data, offset);
1656                 memcpy (pkt->data+offset, pkt_data, pkt_size);
1657
1658                 if (pkt_data != data)
1659                     av_free(pkt_data);
1660
1661                 if (n == 0)
1662                     pkt->flags = is_keyframe;
1663                 pkt->stream_index = st->index;
1664
1665                 pkt->pts = timecode;
1666                 pkt->pos = pos;
1667                 pkt->duration = duration;
1668
1669                 matroska_queue_packet(matroska, pkt);
1670             }
1671
1672             if (timecode != AV_NOPTS_VALUE)
1673                 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
1674             data += lace_size[n];
1675         }
1676     }
1677
1678     av_free(lace_size);
1679     return res;
1680 }
1681
1682 static int
1683 matroska_parse_cluster (MatroskaDemuxContext *matroska)
1684 {
1685     MatroskaCluster cluster = { 0 };
1686     EbmlList *blocks_list;
1687     MatroskaBlock *blocks;
1688     int i, res = ebml_parse(matroska, matroska_clusters, &cluster, 1);
1689     blocks_list = &cluster.blocks;
1690     blocks = blocks_list->elem;
1691     for (i=0; !res && i<blocks_list->nb_elem; i++)
1692         if (blocks[i].bin.size > 0)
1693             res=matroska_parse_block(matroska,
1694                                      blocks[i].bin.data, blocks[i].bin.size,
1695                                      blocks[i].bin.pos,  cluster.timecode,
1696                                      blocks[i].duration, !blocks[i].reference);
1697     ebml_free(matroska_cluster, &cluster);
1698     return res;
1699 }
1700
1701 static int
1702 matroska_read_packet (AVFormatContext *s,
1703                       AVPacket        *pkt)
1704 {
1705     MatroskaDemuxContext *matroska = s->priv_data;
1706
1707     while (matroska_deliver_packet(matroska, pkt)) {
1708         if (matroska->done)
1709             return AVERROR(EIO);
1710         if (matroska_parse_cluster(matroska) < 0)
1711             matroska->done = 1;
1712     }
1713
1714     return 0;
1715 }
1716
1717 static int
1718 matroska_read_seek (AVFormatContext *s, int stream_index, int64_t timestamp,
1719                     int flags)
1720 {
1721     MatroskaDemuxContext *matroska = s->priv_data;
1722     AVStream *st = s->streams[stream_index];
1723     int index;
1724
1725     index = av_index_search_timestamp(st, timestamp, flags);
1726     if (index < 0)
1727         return 0;
1728
1729     matroska_clear_queue(matroska);
1730
1731     url_fseek(s->pb, st->index_entries[index].pos, SEEK_SET);
1732     matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
1733     matroska->skip_to_stream = st;
1734     matroska->peek_id = 0;
1735     av_update_cur_dts(s, st, st->index_entries[index].timestamp);
1736     return 0;
1737 }
1738
1739 static int
1740 matroska_read_close (AVFormatContext *s)
1741 {
1742     MatroskaDemuxContext *matroska = s->priv_data;
1743     MatroskaTrack *tracks = matroska->tracks.elem;
1744     int n;
1745
1746     matroska_clear_queue(matroska);
1747
1748     for (n=0; n < matroska->tracks.nb_elem; n++)
1749         if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
1750             av_free(tracks[n].audio.buf);
1751     ebml_free(matroska_segment, matroska);
1752
1753     return 0;
1754 }
1755
1756 AVInputFormat matroska_demuxer = {
1757     "matroska",
1758     NULL_IF_CONFIG_SMALL("Matroska file format"),
1759     sizeof(MatroskaDemuxContext),
1760     matroska_probe,
1761     matroska_read_header,
1762     matroska_read_packet,
1763     matroska_read_close,
1764     matroska_read_seek,
1765 };