git.sesse.net Git - ffmpeg/blob - libavformat/matroskadec.c

   1 /*
   2  * Matroska file demuxer
   3  * Copyright (c) 2003-2008 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
  24  * Matroska file demuxer
  25  * by Ronald Bultje <rbultje@ronald.bitfreak.net>
  26  * with a little help from Moritz Bunkus <moritz@bunkus.org>
  27  * totally reworked by Aurelien Jacobs <aurel@gnuage.org>
  28  * 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 /* For ff_codec_get_id(). */
  35 #include "riff.h"
  36 #include "isom.h"
  37 #include "rm.h"
  38 #include "matroska.h"
  39 #include "libavcodec/mpeg4audio.h"
  40 #include "libavutil/intfloat_readwrite.h"
  41 #include "libavutil/intreadwrite.h"
  42 #include "libavutil/avstring.h"
  43 #include "libavutil/lzo.h"
  44 #if CONFIG_ZLIB
  45 #include <zlib.h>
  46 #endif
  47 #if CONFIG_BZLIB
  48 #include <bzlib.h>
  49 #endif
  50
  51 typedef enum {
  52     EBML_NONE,
  53     EBML_UINT,
  54     EBML_FLOAT,
  55     EBML_STR,
  56     EBML_UTF8,
  57     EBML_BIN,
  58     EBML_NEST,
  59     EBML_PASS,
  60     EBML_STOP,
  61 } EbmlType;
  62
  63 typedef const struct EbmlSyntax {
  64     uint32_t id;
  65     EbmlType type;
  66     int list_elem_size;
  67     int data_offset;
  68     union {
  69         uint64_t    u;
  70         double      f;
  71         const char *s;
  72         const struct EbmlSyntax *n;
  73     } def;
  74 } EbmlSyntax;
  75
  76 typedef struct {
  77     int nb_elem;
  78     void *elem;
  79 } EbmlList;
  80
  81 typedef struct {
  82     int      size;
  83     uint8_t *data;
  84     int64_t  pos;
  85 } EbmlBin;
  86
  87 typedef struct {
  88     uint64_t version;
  89     uint64_t max_size;
  90     uint64_t id_length;
  91     char    *doctype;
  92     uint64_t doctype_version;
  93 } Ebml;
  94
  95 typedef struct {
  96     uint64_t algo;
  97     EbmlBin  settings;
  98 } MatroskaTrackCompression;
  99
 100 typedef struct {
 101     uint64_t scope;
 102     uint64_t type;
 103     MatroskaTrackCompression compression;
 104 } MatroskaTrackEncoding;
 105
 106 typedef struct {
 107     double   frame_rate;
 108     uint64_t display_width;
 109     uint64_t display_height;
 110     uint64_t pixel_width;
 111     uint64_t pixel_height;
 112     uint64_t fourcc;
 113 } MatroskaTrackVideo;
 114
 115 typedef struct {
 116     double   samplerate;
 117     double   out_samplerate;
 118     uint64_t bitdepth;
 119     uint64_t channels;
 120
 121     /* real audio header (extracted from extradata) */
 122     int      coded_framesize;
 123     int      sub_packet_h;
 124     int      frame_size;
 125     int      sub_packet_size;
 126     int      sub_packet_cnt;
 127     int      pkt_cnt;
 128     uint8_t *buf;
 129 } MatroskaTrackAudio;
 130
 131 typedef struct {
 132     uint64_t num;
 133     uint64_t uid;
 134     uint64_t type;
 135     char    *name;
 136     char    *codec_id;
 137     EbmlBin  codec_priv;
 138     char    *language;
 139     double time_scale;
 140     uint64_t default_duration;
 141     uint64_t flag_default;
 142     MatroskaTrackVideo video;
 143     MatroskaTrackAudio audio;
 144     EbmlList encodings;
 145
 146     AVStream *stream;
 147     int64_t end_timecode;
 148     int ms_compat;
 149 } MatroskaTrack;
 150
 151 typedef struct {
 152     uint64_t uid;
 153     char *filename;
 154     char *mime;
 155     EbmlBin bin;
 156
 157     AVStream *stream;
 158 } MatroskaAttachement;
 159
 160 typedef struct {
 161     uint64_t start;
 162     uint64_t end;
 163     uint64_t uid;
 164     char    *title;
 165
 166     AVChapter *chapter;
 167 } MatroskaChapter;
 168
 169 typedef struct {
 170     uint64_t track;
 171     uint64_t pos;
 172 } MatroskaIndexPos;
 173
 174 typedef struct {
 175     uint64_t time;
 176     EbmlList pos;
 177 } MatroskaIndex;
 178
 179 typedef struct {
 180     char *name;
 181     char *string;
 182     char *lang;
 183     uint64_t def;
 184     EbmlList sub;
 185 } MatroskaTag;
 186
 187 typedef struct {
 188     char    *type;
 189     uint64_t typevalue;
 190     uint64_t trackuid;
 191     uint64_t chapteruid;
 192     uint64_t attachuid;
 193 } MatroskaTagTarget;
 194
 195 typedef struct {
 196     MatroskaTagTarget target;
 197     EbmlList tag;
 198 } MatroskaTags;
 199
 200 typedef struct {
 201     uint64_t id;
 202     uint64_t pos;
 203 } MatroskaSeekhead;
 204
 205 typedef struct {
 206     uint64_t start;
 207     uint64_t length;
 208 } MatroskaLevel;
 209
 210 typedef struct {
 211     AVFormatContext *ctx;
 212
 213     /* EBML stuff */
 214     int num_levels;
 215     MatroskaLevel levels[EBML_MAX_DEPTH];
 216     int level_up;
 217
 218     uint64_t time_scale;
 219     double   duration;
 220     char    *title;
 221     EbmlList tracks;
 222     EbmlList attachments;
 223     EbmlList chapters;
 224     EbmlList index;
 225     EbmlList tags;
 226     EbmlList seekhead;
 227
 228     /* byte position of the segment inside the stream */
 229     int64_t segment_start;
 230
 231     /* the packet queue */
 232     AVPacket **packets;
 233     int num_packets;
 234     AVPacket *prev_pkt;
 235
 236     int done;
 237     int has_cluster_id;
 238
 239     /* What to skip before effectively reading a packet. */
 240     int skip_to_keyframe;
 241     uint64_t skip_to_timecode;
 242 } MatroskaDemuxContext;
 243
 244 typedef struct {
 245     uint64_t duration;
 246     int64_t  reference;
 247     uint64_t non_simple;
 248     EbmlBin  bin;
 249 } MatroskaBlock;
 250
 251 typedef struct {
 252     uint64_t timecode;
 253     EbmlList blocks;
 254 } MatroskaCluster;
 255
 256 static EbmlSyntax ebml_header[] = {
 257     { EBML_ID_EBMLREADVERSION,        EBML_UINT, 0, offsetof(Ebml,version), {.u=EBML_VERSION} },
 258     { EBML_ID_EBMLMAXSIZELENGTH,      EBML_UINT, 0, offsetof(Ebml,max_size), {.u=8} },
 259     { EBML_ID_EBMLMAXIDLENGTH,        EBML_UINT, 0, offsetof(Ebml,id_length), {.u=4} },
 260     { EBML_ID_DOCTYPE,                EBML_STR,  0, offsetof(Ebml,doctype), {.s="(none)"} },
 261     { EBML_ID_DOCTYPEREADVERSION,     EBML_UINT, 0, offsetof(Ebml,doctype_version), {.u=1} },
 262     { EBML_ID_EBMLVERSION,            EBML_NONE },
 263     { EBML_ID_DOCTYPEVERSION,         EBML_NONE },
 264     { 0 }
 265 };
 266
 267 static EbmlSyntax ebml_syntax[] = {
 268     { EBML_ID_HEADER,                 EBML_NEST, 0, 0, {.n=ebml_header} },
 269     { 0 }
 270 };
 271
 272 static EbmlSyntax matroska_info[] = {
 273     { MATROSKA_ID_TIMECODESCALE,      EBML_UINT,  0, offsetof(MatroskaDemuxContext,time_scale), {.u=1000000} },
 274     { MATROSKA_ID_DURATION,           EBML_FLOAT, 0, offsetof(MatroskaDemuxContext,duration) },
 275     { MATROSKA_ID_TITLE,              EBML_UTF8,  0, offsetof(MatroskaDemuxContext,title) },
 276     { MATROSKA_ID_WRITINGAPP,         EBML_NONE },
 277     { MATROSKA_ID_MUXINGAPP,          EBML_NONE },
 278     { MATROSKA_ID_DATEUTC,            EBML_NONE },
 279     { MATROSKA_ID_SEGMENTUID,         EBML_NONE },
 280     { 0 }
 281 };
 282
 283 static EbmlSyntax matroska_track_video[] = {
 284     { MATROSKA_ID_VIDEOFRAMERATE,     EBML_FLOAT,0, offsetof(MatroskaTrackVideo,frame_rate) },
 285     { MATROSKA_ID_VIDEODISPLAYWIDTH,  EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_width) },
 286     { MATROSKA_ID_VIDEODISPLAYHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_height) },
 287     { MATROSKA_ID_VIDEOPIXELWIDTH,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_width) },
 288     { MATROSKA_ID_VIDEOPIXELHEIGHT,   EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_height) },
 289     { MATROSKA_ID_VIDEOCOLORSPACE,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,fourcc) },
 290     { MATROSKA_ID_VIDEOPIXELCROPB,    EBML_NONE },
 291     { MATROSKA_ID_VIDEOPIXELCROPT,    EBML_NONE },
 292     { MATROSKA_ID_VIDEOPIXELCROPL,    EBML_NONE },
 293     { MATROSKA_ID_VIDEOPIXELCROPR,    EBML_NONE },
 294     { MATROSKA_ID_VIDEODISPLAYUNIT,   EBML_NONE },
 295     { MATROSKA_ID_VIDEOFLAGINTERLACED,EBML_NONE },
 296     { MATROSKA_ID_VIDEOSTEREOMODE,    EBML_NONE },
 297     { MATROSKA_ID_VIDEOASPECTRATIO,   EBML_NONE },
 298     { 0 }
 299 };
 300
 301 static EbmlSyntax matroska_track_audio[] = {
 302     { MATROSKA_ID_AUDIOSAMPLINGFREQ,  EBML_FLOAT,0, offsetof(MatroskaTrackAudio,samplerate), {.f=8000.0} },
 303     { MATROSKA_ID_AUDIOOUTSAMPLINGFREQ,EBML_FLOAT,0,offsetof(MatroskaTrackAudio,out_samplerate) },
 304     { MATROSKA_ID_AUDIOBITDEPTH,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,bitdepth) },
 305     { MATROSKA_ID_AUDIOCHANNELS,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,channels), {.u=1} },
 306     { 0 }
 307 };
 308
 309 static EbmlSyntax matroska_track_encoding_compression[] = {
 310     { MATROSKA_ID_ENCODINGCOMPALGO,   EBML_UINT, 0, offsetof(MatroskaTrackCompression,algo), {.u=0} },
 311     { MATROSKA_ID_ENCODINGCOMPSETTINGS,EBML_BIN, 0, offsetof(MatroskaTrackCompression,settings) },
 312     { 0 }
 313 };
 314
 315 static EbmlSyntax matroska_track_encoding[] = {
 316     { MATROSKA_ID_ENCODINGSCOPE,      EBML_UINT, 0, offsetof(MatroskaTrackEncoding,scope), {.u=1} },
 317     { MATROSKA_ID_ENCODINGTYPE,       EBML_UINT, 0, offsetof(MatroskaTrackEncoding,type), {.u=0} },
 318     { MATROSKA_ID_ENCODINGCOMPRESSION,EBML_NEST, 0, offsetof(MatroskaTrackEncoding,compression), {.n=matroska_track_encoding_compression} },
 319     { MATROSKA_ID_ENCODINGORDER,      EBML_NONE },
 320     { 0 }
 321 };
 322
 323 static EbmlSyntax matroska_track_encodings[] = {
 324     { MATROSKA_ID_TRACKCONTENTENCODING, EBML_NEST, sizeof(MatroskaTrackEncoding), offsetof(MatroskaTrack,encodings), {.n=matroska_track_encoding} },
 325     { 0 }
 326 };
 327
 328 static EbmlSyntax matroska_track[] = {
 329     { MATROSKA_ID_TRACKNUMBER,          EBML_UINT, 0, offsetof(MatroskaTrack,num) },
 330     { MATROSKA_ID_TRACKNAME,            EBML_UTF8, 0, offsetof(MatroskaTrack,name) },
 331     { MATROSKA_ID_TRACKUID,             EBML_UINT, 0, offsetof(MatroskaTrack,uid) },
 332     { MATROSKA_ID_TRACKTYPE,            EBML_UINT, 0, offsetof(MatroskaTrack,type) },
 333     { MATROSKA_ID_CODECID,              EBML_STR,  0, offsetof(MatroskaTrack,codec_id) },
 334     { MATROSKA_ID_CODECPRIVATE,         EBML_BIN,  0, offsetof(MatroskaTrack,codec_priv) },
 335     { MATROSKA_ID_TRACKLANGUAGE,        EBML_UTF8, 0, offsetof(MatroskaTrack,language), {.s="eng"} },
 336     { MATROSKA_ID_TRACKDEFAULTDURATION, EBML_UINT, 0, offsetof(MatroskaTrack,default_duration) },
 337     { MATROSKA_ID_TRACKTIMECODESCALE,   EBML_FLOAT,0, offsetof(MatroskaTrack,time_scale), {.f=1.0} },
 338     { MATROSKA_ID_TRACKFLAGDEFAULT,     EBML_UINT, 0, offsetof(MatroskaTrack,flag_default), {.u=1} },
 339     { MATROSKA_ID_TRACKVIDEO,           EBML_NEST, 0, offsetof(MatroskaTrack,video), {.n=matroska_track_video} },
 340     { MATROSKA_ID_TRACKAUDIO,           EBML_NEST, 0, offsetof(MatroskaTrack,audio), {.n=matroska_track_audio} },
 341     { MATROSKA_ID_TRACKCONTENTENCODINGS,EBML_NEST, 0, 0, {.n=matroska_track_encodings} },
 342     { MATROSKA_ID_TRACKFLAGENABLED,     EBML_NONE },
 343     { MATROSKA_ID_TRACKFLAGFORCED,      EBML_NONE },
 344     { MATROSKA_ID_TRACKFLAGLACING,      EBML_NONE },
 345     { MATROSKA_ID_CODECNAME,            EBML_NONE },
 346     { MATROSKA_ID_CODECDECODEALL,       EBML_NONE },
 347     { MATROSKA_ID_CODECINFOURL,         EBML_NONE },
 348     { MATROSKA_ID_CODECDOWNLOADURL,     EBML_NONE },
 349     { MATROSKA_ID_TRACKMINCACHE,        EBML_NONE },
 350     { MATROSKA_ID_TRACKMAXCACHE,        EBML_NONE },
 351     { MATROSKA_ID_TRACKMAXBLKADDID,     EBML_NONE },
 352     { 0 }
 353 };
 354
 355 static EbmlSyntax matroska_tracks[] = {
 356     { MATROSKA_ID_TRACKENTRY,         EBML_NEST, sizeof(MatroskaTrack), offsetof(MatroskaDemuxContext,tracks), {.n=matroska_track} },
 357     { 0 }
 358 };
 359
 360 static EbmlSyntax matroska_attachment[] = {
 361     { MATROSKA_ID_FILEUID,            EBML_UINT, 0, offsetof(MatroskaAttachement,uid) },
 362     { MATROSKA_ID_FILENAME,           EBML_UTF8, 0, offsetof(MatroskaAttachement,filename) },
 363     { MATROSKA_ID_FILEMIMETYPE,       EBML_STR,  0, offsetof(MatroskaAttachement,mime) },
 364     { MATROSKA_ID_FILEDATA,           EBML_BIN,  0, offsetof(MatroskaAttachement,bin) },
 365     { MATROSKA_ID_FILEDESC,           EBML_NONE },
 366     { 0 }
 367 };
 368
 369 static EbmlSyntax matroska_attachments[] = {
 370     { MATROSKA_ID_ATTACHEDFILE,       EBML_NEST, sizeof(MatroskaAttachement), offsetof(MatroskaDemuxContext,attachments), {.n=matroska_attachment} },
 371     { 0 }
 372 };
 373
 374 static EbmlSyntax matroska_chapter_display[] = {
 375     { MATROSKA_ID_CHAPSTRING,         EBML_UTF8, 0, offsetof(MatroskaChapter,title) },
 376     { MATROSKA_ID_CHAPLANG,           EBML_NONE },
 377     { 0 }
 378 };
 379
 380 static EbmlSyntax matroska_chapter_entry[] = {
 381     { MATROSKA_ID_CHAPTERTIMESTART,   EBML_UINT, 0, offsetof(MatroskaChapter,start), {.u=AV_NOPTS_VALUE} },
 382     { MATROSKA_ID_CHAPTERTIMEEND,     EBML_UINT, 0, offsetof(MatroskaChapter,end), {.u=AV_NOPTS_VALUE} },
 383     { MATROSKA_ID_CHAPTERUID,         EBML_UINT, 0, offsetof(MatroskaChapter,uid) },
 384     { MATROSKA_ID_CHAPTERDISPLAY,     EBML_NEST, 0, 0, {.n=matroska_chapter_display} },
 385     { MATROSKA_ID_CHAPTERFLAGHIDDEN,  EBML_NONE },
 386     { MATROSKA_ID_CHAPTERFLAGENABLED, EBML_NONE },
 387     { MATROSKA_ID_CHAPTERPHYSEQUIV,   EBML_NONE },
 388     { MATROSKA_ID_CHAPTERATOM,        EBML_NONE },
 389     { 0 }
 390 };
 391
 392 static EbmlSyntax matroska_chapter[] = {
 393     { MATROSKA_ID_CHAPTERATOM,        EBML_NEST, sizeof(MatroskaChapter), offsetof(MatroskaDemuxContext,chapters), {.n=matroska_chapter_entry} },
 394     { MATROSKA_ID_EDITIONUID,         EBML_NONE },
 395     { MATROSKA_ID_EDITIONFLAGHIDDEN,  EBML_NONE },
 396     { MATROSKA_ID_EDITIONFLAGDEFAULT, EBML_NONE },
 397     { MATROSKA_ID_EDITIONFLAGORDERED, EBML_NONE },
 398     { 0 }
 399 };
 400
 401 static EbmlSyntax matroska_chapters[] = {
 402     { MATROSKA_ID_EDITIONENTRY,       EBML_NEST, 0, 0, {.n=matroska_chapter} },
 403     { 0 }
 404 };
 405
 406 static EbmlSyntax matroska_index_pos[] = {
 407     { MATROSKA_ID_CUETRACK,           EBML_UINT, 0, offsetof(MatroskaIndexPos,track) },
 408     { MATROSKA_ID_CUECLUSTERPOSITION, EBML_UINT, 0, offsetof(MatroskaIndexPos,pos)   },
 409     { MATROSKA_ID_CUEBLOCKNUMBER,     EBML_NONE },
 410     { 0 }
 411 };
 412
 413 static EbmlSyntax matroska_index_entry[] = {
 414     { MATROSKA_ID_CUETIME,            EBML_UINT, 0, offsetof(MatroskaIndex,time) },
 415     { MATROSKA_ID_CUETRACKPOSITION,   EBML_NEST, sizeof(MatroskaIndexPos), offsetof(MatroskaIndex,pos), {.n=matroska_index_pos} },
 416     { 0 }
 417 };
 418
 419 static EbmlSyntax matroska_index[] = {
 420     { MATROSKA_ID_POINTENTRY,         EBML_NEST, sizeof(MatroskaIndex), offsetof(MatroskaDemuxContext,index), {.n=matroska_index_entry} },
 421     { 0 }
 422 };
 423
 424 static EbmlSyntax matroska_simpletag[] = {
 425     { MATROSKA_ID_TAGNAME,            EBML_UTF8, 0, offsetof(MatroskaTag,name) },
 426     { MATROSKA_ID_TAGSTRING,          EBML_UTF8, 0, offsetof(MatroskaTag,string) },
 427     { MATROSKA_ID_TAGLANG,            EBML_STR,  0, offsetof(MatroskaTag,lang), {.s="und"} },
 428     { MATROSKA_ID_TAGDEFAULT,         EBML_UINT, 0, offsetof(MatroskaTag,def) },
 429     { MATROSKA_ID_SIMPLETAG,          EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTag,sub), {.n=matroska_simpletag} },
 430     { 0 }
 431 };
 432
 433 static EbmlSyntax matroska_tagtargets[] = {
 434     { MATROSKA_ID_TAGTARGETS_TYPE,      EBML_STR,  0, offsetof(MatroskaTagTarget,type) },
 435     { MATROSKA_ID_TAGTARGETS_TYPEVALUE, EBML_UINT, 0, offsetof(MatroskaTagTarget,typevalue), {.u=50} },
 436     { MATROSKA_ID_TAGTARGETS_TRACKUID,  EBML_UINT, 0, offsetof(MatroskaTagTarget,trackuid) },
 437     { MATROSKA_ID_TAGTARGETS_CHAPTERUID,EBML_UINT, 0, offsetof(MatroskaTagTarget,chapteruid) },
 438     { MATROSKA_ID_TAGTARGETS_ATTACHUID, EBML_UINT, 0, offsetof(MatroskaTagTarget,attachuid) },
 439     { 0 }
 440 };
 441
 442 static EbmlSyntax matroska_tag[] = {
 443     { MATROSKA_ID_SIMPLETAG,          EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTags,tag), {.n=matroska_simpletag} },
 444     { MATROSKA_ID_TAGTARGETS,         EBML_NEST, 0, offsetof(MatroskaTags,target), {.n=matroska_tagtargets} },
 445     { 0 }
 446 };
 447
 448 static EbmlSyntax matroska_tags[] = {
 449     { MATROSKA_ID_TAG,                EBML_NEST, sizeof(MatroskaTags), offsetof(MatroskaDemuxContext,tags), {.n=matroska_tag} },
 450     { 0 }
 451 };
 452
 453 static EbmlSyntax matroska_seekhead_entry[] = {
 454     { MATROSKA_ID_SEEKID,             EBML_UINT, 0, offsetof(MatroskaSeekhead,id) },
 455     { MATROSKA_ID_SEEKPOSITION,       EBML_UINT, 0, offsetof(MatroskaSeekhead,pos), {.u=-1} },
 456     { 0 }
 457 };
 458
 459 static EbmlSyntax matroska_seekhead[] = {
 460     { MATROSKA_ID_SEEKENTRY,          EBML_NEST, sizeof(MatroskaSeekhead), offsetof(MatroskaDemuxContext,seekhead), {.n=matroska_seekhead_entry} },
 461     { 0 }
 462 };
 463
 464 static EbmlSyntax matroska_segment[] = {
 465     { MATROSKA_ID_INFO,           EBML_NEST, 0, 0, {.n=matroska_info       } },
 466     { MATROSKA_ID_TRACKS,         EBML_NEST, 0, 0, {.n=matroska_tracks     } },
 467     { MATROSKA_ID_ATTACHMENTS,    EBML_NEST, 0, 0, {.n=matroska_attachments} },
 468     { MATROSKA_ID_CHAPTERS,       EBML_NEST, 0, 0, {.n=matroska_chapters   } },
 469     { MATROSKA_ID_CUES,           EBML_NEST, 0, 0, {.n=matroska_index      } },
 470     { MATROSKA_ID_TAGS,           EBML_NEST, 0, 0, {.n=matroska_tags       } },
 471     { MATROSKA_ID_SEEKHEAD,       EBML_NEST, 0, 0, {.n=matroska_seekhead   } },
 472     { MATROSKA_ID_CLUSTER,        EBML_STOP, 0, offsetof(MatroskaDemuxContext,has_cluster_id) },
 473     { 0 }
 474 };
 475
 476 static EbmlSyntax matroska_segments[] = {
 477     { MATROSKA_ID_SEGMENT,        EBML_NEST, 0, 0, {.n=matroska_segment    } },
 478     { 0 }
 479 };
 480
 481 static EbmlSyntax matroska_blockgroup[] = {
 482     { MATROSKA_ID_BLOCK,          EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
 483     { MATROSKA_ID_SIMPLEBLOCK,    EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
 484     { MATROSKA_ID_BLOCKDURATION,  EBML_UINT, 0, offsetof(MatroskaBlock,duration), {.u=AV_NOPTS_VALUE} },
 485     { MATROSKA_ID_BLOCKREFERENCE, EBML_UINT, 0, offsetof(MatroskaBlock,reference) },
 486     { 1,                          EBML_UINT, 0, offsetof(MatroskaBlock,non_simple), {.u=1} },
 487     { 0 }
 488 };
 489
 490 static EbmlSyntax matroska_cluster[] = {
 491     { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
 492     { MATROSKA_ID_BLOCKGROUP,     EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
 493     { MATROSKA_ID_SIMPLEBLOCK,    EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
 494     { MATROSKA_ID_CLUSTERPOSITION,EBML_NONE },
 495     { MATROSKA_ID_CLUSTERPREVSIZE,EBML_NONE },
 496     { 0 }
 497 };
 498
 499 static EbmlSyntax matroska_clusters[] = {
 500     { MATROSKA_ID_CLUSTER,        EBML_NEST, 0, 0, {.n=matroska_cluster} },
 501     { MATROSKA_ID_INFO,           EBML_NONE },
 502     { MATROSKA_ID_CUES,           EBML_NONE },
 503     { MATROSKA_ID_TAGS,           EBML_NONE },
 504     { MATROSKA_ID_SEEKHEAD,       EBML_NONE },
 505     { 0 }
 506 };
 507
 508 static const char *matroska_doctypes[] = { "matroska", "webm" };
 509
 510 /*
 511  * Return: Whether we reached the end of a level in the hierarchy or not.
 512  */
 513 static int ebml_level_end(MatroskaDemuxContext *matroska)
 514 {
 515     ByteIOContext *pb = matroska->ctx->pb;
 516     int64_t pos = url_ftell(pb);
 517
 518     if (matroska->num_levels > 0) {
 519         MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
 520         if (pos - level->start >= level->length) {
 521             matroska->num_levels--;
 522             return 1;
 523         }
 524     }
 525     return 0;
 526 }
 527
 528 /*
 529  * Read: an "EBML number", which is defined as a variable-length
 530  * array of bytes. The first byte indicates the length by giving a
 531  * number of 0-bits followed by a one. The position of the first
 532  * "one" bit inside the first byte indicates the length of this
 533  * number.
 534  * Returns: number of bytes read, < 0 on error
 535  */
 536 static int ebml_read_num(MatroskaDemuxContext *matroska, ByteIOContext *pb,
 537                          int max_size, uint64_t *number)
 538 {
 539     int len_mask = 0x80, read = 1, n = 1;
 540     int64_t total = 0;
 541
 542     /* The first byte tells us the length in bytes - get_byte() can normally
 543      * return 0, but since that's not a valid first ebmlID byte, we can
 544      * use it safely here to catch EOS. */
 545     if (!(total = get_byte(pb))) {
 546         /* we might encounter EOS here */
 547         if (!url_feof(pb)) {
 548             int64_t pos = url_ftell(pb);
 549             av_log(matroska->ctx, AV_LOG_ERROR,
 550                    "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
 551                    pos, pos);
 552         }
 553         return AVERROR(EIO); /* EOS or actual I/O error */
 554     }
 555
 556     /* get the length of the EBML number */
 557     while (read <= max_size && !(total & len_mask)) {
 558         read++;
 559         len_mask >>= 1;
 560     }
 561     if (read > max_size) {
 562         int64_t pos = url_ftell(pb) - 1;
 563         av_log(matroska->ctx, AV_LOG_ERROR,
 564                "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
 565                (uint8_t) total, pos, pos);
 566         return AVERROR_INVALIDDATA;
 567     }
 568
 569     /* read out length */
 570     total &= ~len_mask;
 571     while (n++ < read)
 572         total = (total << 8) | get_byte(pb);
 573
 574     *number = total;
 575
 576     return read;
 577 }
 578
 579 /*
 580  * Read the next element as an unsigned int.
 581  * 0 is success, < 0 is failure.
 582  */
 583 static int ebml_read_uint(ByteIOContext *pb, int size, uint64_t *num)
 584 {
 585     int n = 0;
 586
 587     if (size < 1 || size > 8)
 588         return AVERROR_INVALIDDATA;
 589
 590     /* big-endian ordering; build up number */
 591     *num = 0;
 592     while (n++ < size)
 593         *num = (*num << 8) | get_byte(pb);
 594
 595     return 0;
 596 }
 597
 598 /*
 599  * Read the next element as a float.
 600  * 0 is success, < 0 is failure.
 601  */
 602 static int ebml_read_float(ByteIOContext *pb, int size, double *num)
 603 {
 604     if (size == 4) {
 605         *num= av_int2flt(get_be32(pb));
 606     } else if(size==8){
 607         *num= av_int2dbl(get_be64(pb));
 608     } else
 609         return AVERROR_INVALIDDATA;
 610
 611     return 0;
 612 }
 613
 614 /*
 615  * Read the next element as an ASCII string.
 616  * 0 is success, < 0 is failure.
 617  */
 618 static int ebml_read_ascii(ByteIOContext *pb, int size, char **str)
 619 {
 620     av_free(*str);
 621     /* EBML strings are usually not 0-terminated, so we allocate one
 622      * byte more, read the string and NULL-terminate it ourselves. */
 623     if (!(*str = av_malloc(size + 1)))
 624         return AVERROR(ENOMEM);
 625     if (get_buffer(pb, (uint8_t *) *str, size) != size) {
 626         av_freep(str);
 627         return AVERROR(EIO);
 628     }
 629     (*str)[size] = '\0';
 630
 631     return 0;
 632 }
 633
 634 /*
 635  * Read the next element as binary data.
 636  * 0 is success, < 0 is failure.
 637  */
 638 static int ebml_read_binary(ByteIOContext *pb, int length, EbmlBin *bin)
 639 {
 640     av_free(bin->data);
 641     if (!(bin->data = av_malloc(length)))
 642         return AVERROR(ENOMEM);
 643
 644     bin->size = length;
 645     bin->pos  = url_ftell(pb);
 646     if (get_buffer(pb, bin->data, length) != length) {
 647         av_freep(&bin->data);
 648         return AVERROR(EIO);
 649     }
 650
 651     return 0;
 652 }
 653
 654 /*
 655  * Read the next element, but only the header. The contents
 656  * are supposed to be sub-elements which can be read separately.
 657  * 0 is success, < 0 is failure.
 658  */
 659 static int ebml_read_master(MatroskaDemuxContext *matroska, int length)
 660 {
 661     ByteIOContext *pb = matroska->ctx->pb;
 662     MatroskaLevel *level;
 663
 664     if (matroska->num_levels >= EBML_MAX_DEPTH) {
 665         av_log(matroska->ctx, AV_LOG_ERROR,
 666                "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
 667         return AVERROR(ENOSYS);
 668     }
 669
 670     level = &matroska->levels[matroska->num_levels++];
 671     level->start = url_ftell(pb);
 672     level->length = length;
 673
 674     return 0;
 675 }
 676
 677 /*
 678  * Read signed/unsigned "EBML" numbers.
 679  * Return: number of bytes processed, < 0 on error
 680  */
 681 static int matroska_ebmlnum_uint(MatroskaDemuxContext *matroska,
 682                                  uint8_t *data, uint32_t size, uint64_t *num)
 683 {
 684     ByteIOContext pb;
 685     init_put_byte(&pb, data, size, 0, NULL, NULL, NULL, NULL);
 686     return ebml_read_num(matroska, &pb, FFMIN(size, 8), num);
 687 }
 688
 689 /*
 690  * Same as above, but signed.
 691  */
 692 static int matroska_ebmlnum_sint(MatroskaDemuxContext *matroska,
 693                                  uint8_t *data, uint32_t size, int64_t *num)
 694 {
 695     uint64_t unum;
 696     int res;
 697
 698     /* read as unsigned number first */
 699     if ((res = matroska_ebmlnum_uint(matroska, data, size, &unum)) < 0)
 700         return res;
 701
 702     /* make signed (weird way) */
 703     *num = unum - ((1LL << (7*res - 1)) - 1);
 704
 705     return res;
 706 }
 707
 708 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
 709                            EbmlSyntax *syntax, void *data);
 710
 711 static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
 712                          uint32_t id, void *data)
 713 {
 714     int i;
 715     for (i=0; syntax[i].id; i++)
 716         if (id == syntax[i].id)
 717             break;
 718     if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32)
 719         av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%X\n", id);
 720     return ebml_parse_elem(matroska, &syntax[i], data);
 721 }
 722
 723 static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
 724                       void *data)
 725 {
 726     uint64_t id;
 727     int res = ebml_read_num(matroska, matroska->ctx->pb, 4, &id);
 728     id |= 1 << 7*res;
 729     return res < 0 ? res : ebml_parse_id(matroska, syntax, id, data);
 730 }
 731
 732 static int ebml_parse_nest(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
 733                            void *data)
 734 {
 735     int i, res = 0;
 736
 737     for (i=0; syntax[i].id; i++)
 738         switch (syntax[i].type) {
 739         case EBML_UINT:
 740             *(uint64_t *)((char *)data+syntax[i].data_offset) = syntax[i].def.u;
 741             break;
 742         case EBML_FLOAT:
 743             *(double   *)((char *)data+syntax[i].data_offset) = syntax[i].def.f;
 744             break;
 745         case EBML_STR:
 746         case EBML_UTF8:
 747             *(char    **)((char *)data+syntax[i].data_offset) = av_strdup(syntax[i].def.s);
 748             break;
 749         }
 750
 751     while (!res && !ebml_level_end(matroska))
 752         res = ebml_parse(matroska, syntax, data);
 753
 754     return res;
 755 }
 756
 757 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
 758                            EbmlSyntax *syntax, void *data)
 759 {
 760     ByteIOContext *pb = matroska->ctx->pb;
 761     uint32_t id = syntax->id;
 762     uint64_t length;
 763     int res;
 764
 765     data = (char *)data + syntax->data_offset;
 766     if (syntax->list_elem_size) {
 767         EbmlList *list = data;
 768         list->elem = av_realloc(list->elem, (list->nb_elem+1)*syntax->list_elem_size);
 769         data = (char*)list->elem + list->nb_elem*syntax->list_elem_size;
 770         memset(data, 0, syntax->list_elem_size);
 771         list->nb_elem++;
 772     }
 773
 774     if (syntax->type != EBML_PASS && syntax->type != EBML_STOP)
 775         if ((res = ebml_read_num(matroska, pb, 8, &length)) < 0)
 776             return res;
 777
 778     switch (syntax->type) {
 779     case EBML_UINT:  res = ebml_read_uint  (pb, length, data);  break;
 780     case EBML_FLOAT: res = ebml_read_float (pb, length, data);  break;
 781     case EBML_STR:
 782     case EBML_UTF8:  res = ebml_read_ascii (pb, length, data);  break;
 783     case EBML_BIN:   res = ebml_read_binary(pb, length, data);  break;
 784     case EBML_NEST:  if ((res=ebml_read_master(matroska, length)) < 0)
 785                          return res;
 786                      if (id == MATROSKA_ID_SEGMENT)
 787                          matroska->segment_start = url_ftell(matroska->ctx->pb);
 788                      return ebml_parse_nest(matroska, syntax->def.n, data);
 789     case EBML_PASS:  return ebml_parse_id(matroska, syntax->def.n, id, data);
 790     case EBML_STOP:  *(int *)data = 1;      return 1;
 791     default:         return url_fseek(pb,length,SEEK_CUR)<0 ? AVERROR(EIO) : 0;
 792     }
 793     if (res == AVERROR_INVALIDDATA)
 794         av_log(matroska->ctx, AV_LOG_ERROR, "Invalid element\n");
 795     else if (res == AVERROR(EIO))
 796         av_log(matroska->ctx, AV_LOG_ERROR, "Read error\n");
 797     return res;
 798 }
 799
 800 static void ebml_free(EbmlSyntax *syntax, void *data)
 801 {
 802     int i, j;
 803     for (i=0; syntax[i].id; i++) {
 804         void *data_off = (char *)data + syntax[i].data_offset;
 805         switch (syntax[i].type) {
 806         case EBML_STR:
 807         case EBML_UTF8:  av_freep(data_off);                      break;
 808         case EBML_BIN:   av_freep(&((EbmlBin *)data_off)->data);  break;
 809         case EBML_NEST:
 810             if (syntax[i].list_elem_size) {
 811                 EbmlList *list = data_off;
 812                 char *ptr = list->elem;
 813                 for (j=0; j<list->nb_elem; j++, ptr+=syntax[i].list_elem_size)
 814                     ebml_free(syntax[i].def.n, ptr);
 815                 av_free(list->elem);
 816             } else
 817                 ebml_free(syntax[i].def.n, data_off);
 818         default:  break;
 819         }
 820     }
 821 }
 822
 823
 824 /*
 825  * Autodetecting...
 826  */
 827 static int matroska_probe(AVProbeData *p)
 828 {
 829     uint64_t total = 0;
 830     int len_mask = 0x80, size = 1, n = 1, i;
 831
 832     /* EBML header? */
 833     if (AV_RB32(p->buf) != EBML_ID_HEADER)
 834         return 0;
 835
 836     /* length of header */
 837     total = p->buf[4];
 838     while (size <= 8 && !(total & len_mask)) {
 839         size++;
 840         len_mask >>= 1;
 841     }
 842     if (size > 8)
 843       return 0;
 844     total &= (len_mask - 1);
 845     while (n < size)
 846         total = (total << 8) | p->buf[4 + n++];
 847
 848     /* Does the probe data contain the whole header? */
 849     if (p->buf_size < 4 + size + total)
 850       return 0;
 851
 852     /* The header should contain a known document type. For now,
 853      * we don't parse the whole header but simply check for the
 854      * availability of that array of characters inside the header.
 855      * Not fully fool-proof, but good enough. */
 856     for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++) {
 857         int probelen = strlen(matroska_doctypes[i]);
 858         for (n = 4+size; n <= 4+size+total-probelen; n++)
 859             if (!memcmp(p->buf+n, matroska_doctypes[i], probelen))
 860                 return AVPROBE_SCORE_MAX;
 861     }
 862
 863     // probably valid EBML header but no recognized doctype
 864     return AVPROBE_SCORE_MAX/2;
 865 }
 866
 867 static MatroskaTrack *matroska_find_track_by_num(MatroskaDemuxContext *matroska,
 868                                                  int num)
 869 {
 870     MatroskaTrack *tracks = matroska->tracks.elem;
 871     int i;
 872
 873     for (i=0; i < matroska->tracks.nb_elem; i++)
 874         if (tracks[i].num == num)
 875             return &tracks[i];
 876
 877     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid track number %d\n", num);
 878     return NULL;
 879 }
 880
 881 static int matroska_decode_buffer(uint8_t** buf, int* buf_size,
 882                                   MatroskaTrack *track)
 883 {
 884     MatroskaTrackEncoding *encodings = track->encodings.elem;
 885     uint8_t* data = *buf;
 886     int isize = *buf_size;
 887     uint8_t* pkt_data = NULL;
 888     int pkt_size = isize;
 889     int result = 0;
 890     int olen;
 891
 892     switch (encodings[0].compression.algo) {
 893     case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
 894         return encodings[0].compression.settings.size;
 895     case MATROSKA_TRACK_ENCODING_COMP_LZO:
 896         do {
 897             olen = pkt_size *= 3;
 898             pkt_data = av_realloc(pkt_data, pkt_size+AV_LZO_OUTPUT_PADDING);
 899             result = av_lzo1x_decode(pkt_data, &olen, data, &isize);
 900         } while (result==AV_LZO_OUTPUT_FULL && pkt_size<10000000);
 901         if (result)
 902             goto failed;
 903         pkt_size -= olen;
 904         break;
 905 #if CONFIG_ZLIB
 906     case MATROSKA_TRACK_ENCODING_COMP_ZLIB: {
 907         z_stream zstream = {0};
 908         if (inflateInit(&zstream) != Z_OK)
 909             return -1;
 910         zstream.next_in = data;
 911         zstream.avail_in = isize;
 912         do {
 913             pkt_size *= 3;
 914             pkt_data = av_realloc(pkt_data, pkt_size);
 915             zstream.avail_out = pkt_size - zstream.total_out;
 916             zstream.next_out = pkt_data + zstream.total_out;
 917             result = inflate(&zstream, Z_NO_FLUSH);
 918         } while (result==Z_OK && pkt_size<10000000);
 919         pkt_size = zstream.total_out;
 920         inflateEnd(&zstream);
 921         if (result != Z_STREAM_END)
 922             goto failed;
 923         break;
 924     }
 925 #endif
 926 #if CONFIG_BZLIB
 927     case MATROSKA_TRACK_ENCODING_COMP_BZLIB: {
 928         bz_stream bzstream = {0};
 929         if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
 930             return -1;
 931         bzstream.next_in = data;
 932         bzstream.avail_in = isize;
 933         do {
 934             pkt_size *= 3;
 935             pkt_data = av_realloc(pkt_data, pkt_size);
 936             bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
 937             bzstream.next_out = pkt_data + bzstream.total_out_lo32;
 938             result = BZ2_bzDecompress(&bzstream);
 939         } while (result==BZ_OK && pkt_size<10000000);
 940         pkt_size = bzstream.total_out_lo32;
 941         BZ2_bzDecompressEnd(&bzstream);
 942         if (result != BZ_STREAM_END)
 943             goto failed;
 944         break;
 945     }
 946 #endif
 947     default:
 948         return -1;
 949     }
 950
 951     *buf = pkt_data;
 952     *buf_size = pkt_size;
 953     return 0;
 954  failed:
 955     av_free(pkt_data);
 956     return -1;
 957 }
 958
 959 static void matroska_fix_ass_packet(MatroskaDemuxContext *matroska,
 960                                     AVPacket *pkt, uint64_t display_duration)
 961 {
 962     char *line, *layer, *ptr = pkt->data, *end = ptr+pkt->size;
 963     for (; *ptr!=',' && ptr<end-1; ptr++);
 964     if (*ptr == ',')
 965         layer = ++ptr;
 966     for (; *ptr!=',' && ptr<end-1; ptr++);
 967     if (*ptr == ',') {
 968         int64_t end_pts = pkt->pts + display_duration;
 969         int sc = matroska->time_scale * pkt->pts / 10000000;
 970         int ec = matroska->time_scale * end_pts  / 10000000;
 971         int sh, sm, ss, eh, em, es, len;
 972         sh = sc/360000;  sc -= 360000*sh;
 973         sm = sc/  6000;  sc -=   6000*sm;
 974         ss = sc/   100;  sc -=    100*ss;
 975         eh = ec/360000;  ec -= 360000*eh;
 976         em = ec/  6000;  ec -=   6000*em;
 977         es = ec/   100;  ec -=    100*es;
 978         *ptr++ = '\0';
 979         len = 50 + end-ptr + FF_INPUT_BUFFER_PADDING_SIZE;
 980         if (!(line = av_malloc(len)))
 981             return;
 982         snprintf(line,len,"Dialogue: %s,%d:%02d:%02d.%02d,%d:%02d:%02d.%02d,%s\r\n",
 983                  layer, sh, sm, ss, sc, eh, em, es, ec, ptr);
 984         av_free(pkt->data);
 985         pkt->data = line;
 986         pkt->size = strlen(line);
 987     }
 988 }
 989
 990 static void matroska_merge_packets(AVPacket *out, AVPacket *in)
 991 {
 992     out->data = av_realloc(out->data, out->size+in->size);
 993     memcpy(out->data+out->size, in->data, in->size);
 994     out->size += in->size;
 995     av_destruct_packet(in);
 996     av_free(in);
 997 }
 998
 999 static void matroska_convert_tag(AVFormatContext *s, EbmlList *list,
1000                                  AVMetadata **metadata, char *prefix)
1001 {
1002     MatroskaTag *tags = list->elem;
1003     char key[1024];
1004     int i;
1005
1006     for (i=0; i < list->nb_elem; i++) {
1007         const char *lang = strcmp(tags[i].lang, "und") ? tags[i].lang : NULL;
1008         if (prefix)  snprintf(key, sizeof(key), "%s/%s", prefix, tags[i].name);
1009         else         av_strlcpy(key, tags[i].name, sizeof(key));
1010         if (tags[i].def || !lang) {
1011         av_metadata_set2(metadata, key, tags[i].string, 0);
1012         if (tags[i].sub.nb_elem)
1013             matroska_convert_tag(s, &tags[i].sub, metadata, key);
1014         }
1015         if (lang) {
1016             av_strlcat(key, "-", sizeof(key));
1017             av_strlcat(key, lang, sizeof(key));
1018             av_metadata_set2(metadata, key, tags[i].string, 0);
1019             if (tags[i].sub.nb_elem)
1020                 matroska_convert_tag(s, &tags[i].sub, metadata, key);
1021         }
1022     }
1023 }
1024
1025 static void matroska_convert_tags(AVFormatContext *s)
1026 {
1027     MatroskaDemuxContext *matroska = s->priv_data;
1028     MatroskaTags *tags = matroska->tags.elem;
1029     int i, j;
1030
1031     for (i=0; i < matroska->tags.nb_elem; i++) {
1032         if (tags[i].target.attachuid) {
1033             MatroskaAttachement *attachment = matroska->attachments.elem;
1034             for (j=0; j<matroska->attachments.nb_elem; j++)
1035                 if (attachment[j].uid == tags[i].target.attachuid)
1036                     matroska_convert_tag(s, &tags[i].tag,
1037                                          &attachment[j].stream->metadata, NULL);
1038         } else if (tags[i].target.chapteruid) {
1039             MatroskaChapter *chapter = matroska->chapters.elem;
1040             for (j=0; j<matroska->chapters.nb_elem; j++)
1041                 if (chapter[j].uid == tags[i].target.chapteruid)
1042                     matroska_convert_tag(s, &tags[i].tag,
1043                                          &chapter[j].chapter->metadata, NULL);
1044         } else if (tags[i].target.trackuid) {
1045             MatroskaTrack *track = matroska->tracks.elem;
1046             for (j=0; j<matroska->tracks.nb_elem; j++)
1047                 if (track[j].uid == tags[i].target.trackuid)
1048                     matroska_convert_tag(s, &tags[i].tag,
1049                                          &track[j].stream->metadata, NULL);
1050         } else {
1051             matroska_convert_tag(s, &tags[i].tag, &s->metadata,
1052                                  tags[i].target.type);
1053         }
1054     }
1055 }
1056
1057 static void matroska_execute_seekhead(MatroskaDemuxContext *matroska)
1058 {
1059     EbmlList *seekhead_list = &matroska->seekhead;
1060     MatroskaSeekhead *seekhead = seekhead_list->elem;
1061     uint32_t level_up = matroska->level_up;
1062     int64_t before_pos = url_ftell(matroska->ctx->pb);
1063     MatroskaLevel level;
1064     int i;
1065
1066     for (i=0; i<seekhead_list->nb_elem; i++) {
1067         int64_t offset = seekhead[i].pos + matroska->segment_start;
1068
1069         if (seekhead[i].pos <= before_pos
1070             || seekhead[i].id == MATROSKA_ID_SEEKHEAD
1071             || seekhead[i].id == MATROSKA_ID_CLUSTER)
1072             continue;
1073
1074         /* seek */
1075         if (url_fseek(matroska->ctx->pb, offset, SEEK_SET) != offset)
1076             continue;
1077
1078         /* We don't want to lose our seekhead level, so we add
1079          * a dummy. This is a crude hack. */
1080         if (matroska->num_levels == EBML_MAX_DEPTH) {
1081             av_log(matroska->ctx, AV_LOG_INFO,
1082                    "Max EBML element depth (%d) reached, "
1083                    "cannot parse further.\n", EBML_MAX_DEPTH);
1084             break;
1085         }
1086
1087         level.start = 0;
1088         level.length = (uint64_t)-1;
1089         matroska->levels[matroska->num_levels] = level;
1090         matroska->num_levels++;
1091
1092         ebml_parse(matroska, matroska_segment, matroska);
1093
1094         /* remove dummy level */
1095         while (matroska->num_levels) {
1096             uint64_t length = matroska->levels[--matroska->num_levels].length;
1097             if (length == (uint64_t)-1)
1098                 break;
1099         }
1100     }
1101
1102     /* seek back */
1103     url_fseek(matroska->ctx->pb, before_pos, SEEK_SET);
1104     matroska->level_up = level_up;
1105 }
1106
1107 static int matroska_aac_profile(char *codec_id)
1108 {
1109     static const char * const aac_profiles[] = { "MAIN", "LC", "SSR" };
1110     int profile;
1111
1112     for (profile=0; profile<FF_ARRAY_ELEMS(aac_profiles); profile++)
1113         if (strstr(codec_id, aac_profiles[profile]))
1114             break;
1115     return profile + 1;
1116 }
1117
1118 static int matroska_aac_sri(int samplerate)
1119 {
1120     int sri;
1121
1122     for (sri=0; sri<FF_ARRAY_ELEMS(ff_mpeg4audio_sample_rates); sri++)
1123         if (ff_mpeg4audio_sample_rates[sri] == samplerate)
1124             break;
1125     return sri;
1126 }
1127
1128 static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap)
1129 {
1130     MatroskaDemuxContext *matroska = s->priv_data;
1131     EbmlList *attachements_list = &matroska->attachments;
1132     MatroskaAttachement *attachements;
1133     EbmlList *chapters_list = &matroska->chapters;
1134     MatroskaChapter *chapters;
1135     MatroskaTrack *tracks;
1136     EbmlList *index_list;
1137     MatroskaIndex *index;
1138     int index_scale = 1;
1139     uint64_t max_start = 0;
1140     Ebml ebml = { 0 };
1141     AVStream *st;
1142     int i, j;
1143
1144     matroska->ctx = s;
1145
1146     /* First read the EBML header. */
1147     if (ebml_parse(matroska, ebml_syntax, &ebml)
1148         || ebml.version > EBML_VERSION       || ebml.max_size > sizeof(uint64_t)
1149         || ebml.id_length > sizeof(uint32_t) || ebml.doctype_version > 2) {
1150         av_log(matroska->ctx, AV_LOG_ERROR,
1151                "EBML header using unsupported features\n"
1152                "(EBML version %"PRIu64", doctype %s, doc version %"PRIu64")\n",
1153                ebml.version, ebml.doctype, ebml.doctype_version);
1154         ebml_free(ebml_syntax, &ebml);
1155         return AVERROR_PATCHWELCOME;
1156     }
1157     for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++)
1158         if (!strcmp(ebml.doctype, matroska_doctypes[i]))
1159             break;
1160     if (i >= FF_ARRAY_ELEMS(matroska_doctypes)) {
1161         av_log(s, AV_LOG_WARNING, "Unknown EBML doctype '%s'\n", ebml.doctype);
1162     }
1163     ebml_free(ebml_syntax, &ebml);
1164
1165     /* The next thing is a segment. */
1166     if (ebml_parse(matroska, matroska_segments, matroska) < 0)
1167         return -1;
1168     matroska_execute_seekhead(matroska);
1169
1170     if (!matroska->time_scale)
1171         matroska->time_scale = 1000000;
1172     if (matroska->duration)
1173         matroska->ctx->duration = matroska->duration * matroska->time_scale
1174                                   * 1000 / AV_TIME_BASE;
1175     av_metadata_set2(&s->metadata, "title", matroska->title, 0);
1176
1177     tracks = matroska->tracks.elem;
1178     for (i=0; i < matroska->tracks.nb_elem; i++) {
1179         MatroskaTrack *track = &tracks[i];
1180         enum CodecID codec_id = CODEC_ID_NONE;
1181         EbmlList *encodings_list = &tracks->encodings;
1182         MatroskaTrackEncoding *encodings = encodings_list->elem;
1183         uint8_t *extradata = NULL;
1184         int extradata_size = 0;
1185         int extradata_offset = 0;
1186         ByteIOContext b;
1187
1188         /* Apply some sanity checks. */
1189         if (track->type != MATROSKA_TRACK_TYPE_VIDEO &&
1190             track->type != MATROSKA_TRACK_TYPE_AUDIO &&
1191             track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1192             av_log(matroska->ctx, AV_LOG_INFO,
1193                    "Unknown or unsupported track type %"PRIu64"\n",
1194                    track->type);
1195             continue;
1196         }
1197         if (track->codec_id == NULL)
1198             continue;
1199
1200         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
1201             if (!track->default_duration)
1202                 track->default_duration = 1000000000/track->video.frame_rate;
1203             if (!track->video.display_width)
1204                 track->video.display_width = track->video.pixel_width;
1205             if (!track->video.display_height)
1206                 track->video.display_height = track->video.pixel_height;
1207         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1208             if (!track->audio.out_samplerate)
1209                 track->audio.out_samplerate = track->audio.samplerate;
1210         }
1211         if (encodings_list->nb_elem > 1) {
1212             av_log(matroska->ctx, AV_LOG_ERROR,
1213                    "Multiple combined encodings no supported");
1214         } else if (encodings_list->nb_elem == 1) {
1215             if (encodings[0].type ||
1216                 (encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP &&
1217 #if CONFIG_ZLIB
1218                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
1219 #endif
1220 #if CONFIG_BZLIB
1221                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
1222 #endif
1223                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_LZO)) {
1224                 encodings[0].scope = 0;
1225                 av_log(matroska->ctx, AV_LOG_ERROR,
1226                        "Unsupported encoding type");
1227             } else if (track->codec_priv.size && encodings[0].scope&2) {
1228                 uint8_t *codec_priv = track->codec_priv.data;
1229                 int offset = matroska_decode_buffer(&track->codec_priv.data,
1230                                                     &track->codec_priv.size,
1231                                                     track);
1232                 if (offset < 0) {
1233                     track->codec_priv.data = NULL;
1234                     track->codec_priv.size = 0;
1235                     av_log(matroska->ctx, AV_LOG_ERROR,
1236                            "Failed to decode codec private data\n");
1237                 } else if (offset > 0) {
1238                     track->codec_priv.data = av_malloc(track->codec_priv.size + offset);
1239                     memcpy(track->codec_priv.data,
1240                            encodings[0].compression.settings.data, offset);
1241                     memcpy(track->codec_priv.data+offset, codec_priv,
1242                            track->codec_priv.size);
1243                     track->codec_priv.size += offset;
1244                 }
1245                 if (codec_priv != track->codec_priv.data)
1246                     av_free(codec_priv);
1247             }
1248         }
1249
1250         for(j=0; ff_mkv_codec_tags[j].id != CODEC_ID_NONE; j++){
1251             if(!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
1252                         strlen(ff_mkv_codec_tags[j].str))){
1253                 codec_id= ff_mkv_codec_tags[j].id;
1254                 break;
1255             }
1256         }
1257
1258         st = track->stream = av_new_stream(s, 0);
1259         if (st == NULL)
1260             return AVERROR(ENOMEM);
1261
1262         if (!strcmp(track->codec_id, "V_MS/VFW/FOURCC")
1263             && track->codec_priv.size >= 40
1264             && track->codec_priv.data != NULL) {
1265             track->ms_compat = 1;
1266             track->video.fourcc = AV_RL32(track->codec_priv.data + 16);
1267             codec_id = ff_codec_get_id(ff_codec_bmp_tags, track->video.fourcc);
1268             extradata_offset = 40;
1269         } else if (!strcmp(track->codec_id, "A_MS/ACM")
1270                    && track->codec_priv.size >= 14
1271                    && track->codec_priv.data != NULL) {
1272             init_put_byte(&b, track->codec_priv.data, track->codec_priv.size,
1273                           URL_RDONLY, NULL, NULL, NULL, NULL);
1274             ff_get_wav_header(&b, st->codec, track->codec_priv.size);
1275             codec_id = st->codec->codec_id;
1276             extradata_offset = FFMIN(track->codec_priv.size, 18);
1277         } else if (!strcmp(track->codec_id, "V_QUICKTIME")
1278                    && (track->codec_priv.size >= 86)
1279                    && (track->codec_priv.data != NULL)) {
1280             track->video.fourcc = AV_RL32(track->codec_priv.data);
1281             codec_id=ff_codec_get_id(codec_movvideo_tags, track->video.fourcc);
1282         } else if (codec_id == CODEC_ID_PCM_S16BE) {
1283             switch (track->audio.bitdepth) {
1284             case  8:  codec_id = CODEC_ID_PCM_U8;     break;
1285             case 24:  codec_id = CODEC_ID_PCM_S24BE;  break;
1286             case 32:  codec_id = CODEC_ID_PCM_S32BE;  break;
1287             }
1288         } else if (codec_id == CODEC_ID_PCM_S16LE) {
1289             switch (track->audio.bitdepth) {
1290             case  8:  codec_id = CODEC_ID_PCM_U8;     break;
1291             case 24:  codec_id = CODEC_ID_PCM_S24LE;  break;
1292             case 32:  codec_id = CODEC_ID_PCM_S32LE;  break;
1293             }
1294         } else if (codec_id==CODEC_ID_PCM_F32LE && track->audio.bitdepth==64) {
1295             codec_id = CODEC_ID_PCM_F64LE;
1296         } else if (codec_id == CODEC_ID_AAC && !track->codec_priv.size) {
1297             int profile = matroska_aac_profile(track->codec_id);
1298             int sri = matroska_aac_sri(track->audio.samplerate);
1299             extradata = av_malloc(5);
1300             if (extradata == NULL)
1301                 return AVERROR(ENOMEM);
1302             extradata[0] = (profile << 3) | ((sri&0x0E) >> 1);
1303             extradata[1] = ((sri&0x01) << 7) | (track->audio.channels<<3);
1304             if (strstr(track->codec_id, "SBR")) {
1305                 sri = matroska_aac_sri(track->audio.out_samplerate);
1306                 extradata[2] = 0x56;
1307                 extradata[3] = 0xE5;
1308                 extradata[4] = 0x80 | (sri<<3);
1309                 extradata_size = 5;
1310             } else
1311                 extradata_size = 2;
1312         } else if (codec_id == CODEC_ID_TTA) {
1313             extradata_size = 30;
1314             extradata = av_mallocz(extradata_size);
1315             if (extradata == NULL)
1316                 return AVERROR(ENOMEM);
1317             init_put_byte(&b, extradata, extradata_size, 1,
1318                           NULL, NULL, NULL, NULL);
1319             put_buffer(&b, "TTA1", 4);
1320             put_le16(&b, 1);
1321             put_le16(&b, track->audio.channels);
1322             put_le16(&b, track->audio.bitdepth);
1323             put_le32(&b, track->audio.out_samplerate);
1324             put_le32(&b, matroska->ctx->duration * track->audio.out_samplerate);
1325         } else if (codec_id == CODEC_ID_RV10 || codec_id == CODEC_ID_RV20 ||
1326                    codec_id == CODEC_ID_RV30 || codec_id == CODEC_ID_RV40) {
1327             extradata_offset = 26;
1328         } else if (codec_id == CODEC_ID_RA_144) {
1329             track->audio.out_samplerate = 8000;
1330             track->audio.channels = 1;
1331         } else if (codec_id == CODEC_ID_RA_288 || codec_id == CODEC_ID_COOK ||
1332                    codec_id == CODEC_ID_ATRAC3 || codec_id == CODEC_ID_SIPR) {
1333             int flavor;
1334             init_put_byte(&b, track->codec_priv.data,track->codec_priv.size,
1335                           0, NULL, NULL, NULL, NULL);
1336             url_fskip(&b, 22);
1337             flavor                       = get_be16(&b);
1338             track->audio.coded_framesize = get_be32(&b);
1339             url_fskip(&b, 12);
1340             track->audio.sub_packet_h    = get_be16(&b);
1341             track->audio.frame_size      = get_be16(&b);
1342             track->audio.sub_packet_size = get_be16(&b);
1343             track->audio.buf = av_malloc(track->audio.frame_size * track->audio.sub_packet_h);
1344             if (codec_id == CODEC_ID_RA_288) {
1345                 st->codec->block_align = track->audio.coded_framesize;
1346                 track->codec_priv.size = 0;
1347             } else {
1348                 if (codec_id == CODEC_ID_SIPR && flavor < 4) {
1349                     const int sipr_bit_rate[4] = { 6504, 8496, 5000, 16000 };
1350                     track->audio.sub_packet_size = ff_sipr_subpk_size[flavor];
1351                     st->codec->bit_rate = sipr_bit_rate[flavor];
1352                 }
1353                 st->codec->block_align = track->audio.sub_packet_size;
1354                 extradata_offset = 78;
1355             }
1356         }
1357         track->codec_priv.size -= extradata_offset;
1358
1359         if (codec_id == CODEC_ID_NONE)
1360             av_log(matroska->ctx, AV_LOG_INFO,
1361                    "Unknown/unsupported CodecID %s.\n", track->codec_id);
1362
1363         if (track->time_scale < 0.01)
1364             track->time_scale = 1.0;
1365         av_set_pts_info(st, 64, matroska->time_scale*track->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
1366
1367         st->codec->codec_id = codec_id;
1368         st->start_time = 0;
1369         if (strcmp(track->language, "und"))
1370             av_metadata_set2(&st->metadata, "language", track->language, 0);
1371         av_metadata_set2(&st->metadata, "title", track->name, 0);
1372
1373         if (track->flag_default)
1374             st->disposition |= AV_DISPOSITION_DEFAULT;
1375
1376         if (track->default_duration)
1377             av_reduce(&st->codec->time_base.num, &st->codec->time_base.den,
1378                       track->default_duration, 1000000000, 30000);
1379
1380         if (!st->codec->extradata) {
1381             if(extradata){
1382                 st->codec->extradata = extradata;
1383                 st->codec->extradata_size = extradata_size;
1384             } else if(track->codec_priv.data && track->codec_priv.size > 0){
1385                 st->codec->extradata = av_mallocz(track->codec_priv.size +
1386                                                   FF_INPUT_BUFFER_PADDING_SIZE);
1387                 if(st->codec->extradata == NULL)
1388                     return AVERROR(ENOMEM);
1389                 st->codec->extradata_size = track->codec_priv.size;
1390                 memcpy(st->codec->extradata,
1391                        track->codec_priv.data + extradata_offset,
1392                        track->codec_priv.size);
1393             }
1394         }
1395
1396         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
1397             st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
1398             st->codec->codec_tag  = track->video.fourcc;
1399             st->codec->width  = track->video.pixel_width;
1400             st->codec->height = track->video.pixel_height;
1401             av_reduce(&st->sample_aspect_ratio.num,
1402                       &st->sample_aspect_ratio.den,
1403                       st->codec->height * track->video.display_width,
1404                       st->codec-> width * track->video.display_height,
1405                       255);
1406             if (st->codec->codec_id != CODEC_ID_H264)
1407             st->need_parsing = AVSTREAM_PARSE_HEADERS;
1408             if (track->default_duration)
1409                 st->avg_frame_rate = av_d2q(1000000000.0/track->default_duration, INT_MAX);
1410         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1411             st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
1412             st->codec->sample_rate = track->audio.out_samplerate;
1413             st->codec->channels = track->audio.channels;
1414             if (st->codec->codec_id != CODEC_ID_AAC)
1415             st->need_parsing = AVSTREAM_PARSE_HEADERS;
1416         } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
1417             st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
1418         }
1419     }
1420
1421     attachements = attachements_list->elem;
1422     for (j=0; j<attachements_list->nb_elem; j++) {
1423         if (!(attachements[j].filename && attachements[j].mime &&
1424               attachements[j].bin.data && attachements[j].bin.size > 0)) {
1425             av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
1426         } else {
1427             AVStream *st = av_new_stream(s, 0);
1428             if (st == NULL)
1429                 break;
1430             av_metadata_set2(&st->metadata, "filename",attachements[j].filename, 0);
1431             st->codec->codec_id = CODEC_ID_NONE;
1432             st->codec->codec_type = AVMEDIA_TYPE_ATTACHMENT;
1433             st->codec->extradata  = av_malloc(attachements[j].bin.size);
1434             if(st->codec->extradata == NULL)
1435                 break;
1436             st->codec->extradata_size = attachements[j].bin.size;
1437             memcpy(st->codec->extradata, attachements[j].bin.data, attachements[j].bin.size);
1438
1439             for (i=0; ff_mkv_mime_tags[i].id != CODEC_ID_NONE; i++) {
1440                 if (!strncmp(ff_mkv_mime_tags[i].str, attachements[j].mime,
1441                              strlen(ff_mkv_mime_tags[i].str))) {
1442                     st->codec->codec_id = ff_mkv_mime_tags[i].id;
1443                     break;
1444                 }
1445             }
1446             attachements[j].stream = st;
1447         }
1448     }
1449
1450     chapters = chapters_list->elem;
1451     for (i=0; i<chapters_list->nb_elem; i++)
1452         if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid
1453             && (max_start==0 || chapters[i].start > max_start)) {
1454             chapters[i].chapter =
1455             ff_new_chapter(s, chapters[i].uid, (AVRational){1, 1000000000},
1456                            chapters[i].start, chapters[i].end,
1457                            chapters[i].title);
1458             av_metadata_set2(&chapters[i].chapter->metadata,
1459                              "title", chapters[i].title, 0);
1460             max_start = chapters[i].start;
1461         }
1462
1463     index_list = &matroska->index;
1464     index = index_list->elem;
1465     if (index_list->nb_elem
1466         && index[0].time > 100000000000000/matroska->time_scale) {
1467         av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n");
1468         index_scale = matroska->time_scale;
1469     }
1470     for (i=0; i<index_list->nb_elem; i++) {
1471         EbmlList *pos_list = &index[i].pos;
1472         MatroskaIndexPos *pos = pos_list->elem;
1473         for (j=0; j<pos_list->nb_elem; j++) {
1474             MatroskaTrack *track = matroska_find_track_by_num(matroska,
1475                                                               pos[j].track);
1476             if (track && track->stream)
1477                 av_add_index_entry(track->stream,
1478                                    pos[j].pos + matroska->segment_start,
1479                                    index[i].time/index_scale, 0, 0,
1480                                    AVINDEX_KEYFRAME);
1481         }
1482     }
1483
1484     matroska_convert_tags(s);
1485
1486     return 0;
1487 }
1488
1489 /*
1490  * Put one packet in an application-supplied AVPacket struct.
1491  * Returns 0 on success or -1 on failure.
1492  */
1493 static int matroska_deliver_packet(MatroskaDemuxContext *matroska,
1494                                    AVPacket *pkt)
1495 {
1496     if (matroska->num_packets > 0) {
1497         memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
1498         av_free(matroska->packets[0]);
1499         if (matroska->num_packets > 1) {
1500             memmove(&matroska->packets[0], &matroska->packets[1],
1501                     (matroska->num_packets - 1) * sizeof(AVPacket *));
1502             matroska->packets =
1503                 av_realloc(matroska->packets, (matroska->num_packets - 1) *
1504                            sizeof(AVPacket *));
1505         } else {
1506             av_freep(&matroska->packets);
1507         }
1508         matroska->num_packets--;
1509         return 0;
1510     }
1511
1512     return -1;
1513 }
1514
1515 /*
1516  * Free all packets in our internal queue.
1517  */
1518 static void matroska_clear_queue(MatroskaDemuxContext *matroska)
1519 {
1520     if (matroska->packets) {
1521         int n;
1522         for (n = 0; n < matroska->num_packets; n++) {
1523             av_free_packet(matroska->packets[n]);
1524             av_free(matroska->packets[n]);
1525         }
1526         av_freep(&matroska->packets);
1527         matroska->num_packets = 0;
1528     }
1529 }
1530
1531 static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data,
1532                                 int size, int64_t pos, uint64_t cluster_time,
1533                                 uint64_t duration, int is_keyframe,
1534                                 int64_t cluster_pos)
1535 {
1536     uint64_t timecode = AV_NOPTS_VALUE;
1537     MatroskaTrack *track;
1538     int res = 0;
1539     AVStream *st;
1540     AVPacket *pkt;
1541     int16_t block_time;
1542     uint32_t *lace_size = NULL;
1543     int n, flags, laces = 0;
1544     uint64_t num;
1545
1546     if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) {
1547         av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
1548         return res;
1549     }
1550     data += n;
1551     size -= n;
1552
1553     track = matroska_find_track_by_num(matroska, num);
1554     if (size <= 3 || !track || !track->stream) {
1555         av_log(matroska->ctx, AV_LOG_INFO,
1556                "Invalid stream %"PRIu64" or size %u\n", num, size);
1557         return res;
1558     }
1559     st = track->stream;
1560     if (st->discard >= AVDISCARD_ALL)
1561         return res;
1562     if (duration == AV_NOPTS_VALUE)
1563         duration = track->default_duration / matroska->time_scale;
1564
1565     block_time = AV_RB16(data);
1566     data += 2;
1567     flags = *data++;
1568     size -= 3;
1569     if (is_keyframe == -1)
1570         is_keyframe = flags & 0x80 ? AV_PKT_FLAG_KEY : 0;
1571
1572     if (cluster_time != (uint64_t)-1
1573         && (block_time >= 0 || cluster_time >= -block_time)) {
1574         timecode = cluster_time + block_time;
1575         if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE
1576             && timecode < track->end_timecode)
1577             is_keyframe = 0;  /* overlapping subtitles are not key frame */
1578         if (is_keyframe)
1579             av_add_index_entry(st, cluster_pos, timecode, 0,0,AVINDEX_KEYFRAME);
1580         track->end_timecode = FFMAX(track->end_timecode, timecode+duration);
1581     }
1582
1583     if (matroska->skip_to_keyframe && track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1584         if (!is_keyframe || timecode < matroska->skip_to_timecode)
1585             return res;
1586         matroska->skip_to_keyframe = 0;
1587     }
1588
1589     switch ((flags & 0x06) >> 1) {
1590         case 0x0: /* no lacing */
1591             laces = 1;
1592             lace_size = av_mallocz(sizeof(int));
1593             lace_size[0] = size;
1594             break;
1595
1596         case 0x1: /* Xiph lacing */
1597         case 0x2: /* fixed-size lacing */
1598         case 0x3: /* EBML lacing */
1599             assert(size>0); // size <=3 is checked before size-=3 above
1600             laces = (*data) + 1;
1601             data += 1;
1602             size -= 1;
1603             lace_size = av_mallocz(laces * sizeof(int));
1604
1605             switch ((flags & 0x06) >> 1) {
1606                 case 0x1: /* Xiph lacing */ {
1607                     uint8_t temp;
1608                     uint32_t total = 0;
1609                     for (n = 0; res == 0 && n < laces - 1; n++) {
1610                         while (1) {
1611                             if (size == 0) {
1612                                 res = -1;
1613                                 break;
1614                             }
1615                             temp = *data;
1616                             lace_size[n] += temp;
1617                             data += 1;
1618                             size -= 1;
1619                             if (temp != 0xff)
1620                                 break;
1621                         }
1622                         total += lace_size[n];
1623                     }
1624                     lace_size[n] = size - total;
1625                     break;
1626                 }
1627
1628                 case 0x2: /* fixed-size lacing */
1629                     for (n = 0; n < laces; n++)
1630                         lace_size[n] = size / laces;
1631                     break;
1632
1633                 case 0x3: /* EBML lacing */ {
1634                     uint32_t total;
1635                     n = matroska_ebmlnum_uint(matroska, data, size, &num);
1636                     if (n < 0) {
1637                         av_log(matroska->ctx, AV_LOG_INFO,
1638                                "EBML block data error\n");
1639                         break;
1640                     }
1641                     data += n;
1642                     size -= n;
1643                     total = lace_size[0] = num;
1644                     for (n = 1; res == 0 && n < laces - 1; n++) {
1645                         int64_t snum;
1646                         int r;
1647                         r = matroska_ebmlnum_sint(matroska, data, size, &snum);
1648                         if (r < 0) {
1649                             av_log(matroska->ctx, AV_LOG_INFO,
1650                                    "EBML block data error\n");
1651                             break;
1652                         }
1653                         data += r;
1654                         size -= r;
1655                         lace_size[n] = lace_size[n - 1] + snum;
1656                         total += lace_size[n];
1657                     }
1658                     lace_size[n] = size - total;
1659                     break;
1660                 }
1661             }
1662             break;
1663     }
1664
1665     if (res == 0) {
1666         for (n = 0; n < laces; n++) {
1667             if ((st->codec->codec_id == CODEC_ID_RA_288 ||
1668                  st->codec->codec_id == CODEC_ID_COOK ||
1669                  st->codec->codec_id == CODEC_ID_SIPR ||
1670                  st->codec->codec_id == CODEC_ID_ATRAC3) &&
1671                  st->codec->block_align && track->audio.sub_packet_size) {
1672                 int a = st->codec->block_align;
1673                 int sps = track->audio.sub_packet_size;
1674                 int cfs = track->audio.coded_framesize;
1675                 int h = track->audio.sub_packet_h;
1676                 int y = track->audio.sub_packet_cnt;
1677                 int w = track->audio.frame_size;
1678                 int x;
1679
1680                 if (!track->audio.pkt_cnt) {
1681                     if (st->codec->codec_id == CODEC_ID_RA_288)
1682                         for (x=0; x<h/2; x++)
1683                             memcpy(track->audio.buf+x*2*w+y*cfs,
1684                                    data+x*cfs, cfs);
1685                     else if (st->codec->codec_id == CODEC_ID_SIPR)
1686                         memcpy(track->audio.buf + y*w, data, w);
1687                     else
1688                         for (x=0; x<w/sps; x++)
1689                             memcpy(track->audio.buf+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), data+x*sps, sps);
1690
1691                     if (++track->audio.sub_packet_cnt >= h) {
1692                         if (st->codec->codec_id == CODEC_ID_SIPR)
1693                             ff_rm_reorder_sipr_data(track->audio.buf, h, w);
1694                         track->audio.sub_packet_cnt = 0;
1695                         track->audio.pkt_cnt = h*w / a;
1696                     }
1697                 }
1698                 while (track->audio.pkt_cnt) {
1699                     pkt = av_mallocz(sizeof(AVPacket));
1700                     av_new_packet(pkt, a);
1701                     memcpy(pkt->data, track->audio.buf
1702                            + a * (h*w / a - track->audio.pkt_cnt--), a);
1703                     pkt->pos = pos;
1704                     pkt->stream_index = st->index;
1705                     dynarray_add(&matroska->packets,&matroska->num_packets,pkt);
1706                 }
1707             } else {
1708                 MatroskaTrackEncoding *encodings = track->encodings.elem;
1709                 int offset = 0, pkt_size = lace_size[n];
1710                 uint8_t *pkt_data = data;
1711
1712                 if (lace_size[n] > size) {
1713                     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid packet size\n");
1714                     break;
1715                 }
1716
1717                 if (encodings && encodings->scope & 1) {
1718                     offset = matroska_decode_buffer(&pkt_data,&pkt_size, track);
1719                     if (offset < 0)
1720                         continue;
1721                 }
1722
1723                 pkt = av_mallocz(sizeof(AVPacket));
1724                 /* XXX: prevent data copy... */
1725                 if (av_new_packet(pkt, pkt_size+offset) < 0) {
1726                     av_free(pkt);
1727                     res = AVERROR(ENOMEM);
1728                     break;
1729                 }
1730                 if (offset)
1731                     memcpy (pkt->data, encodings->compression.settings.data, offset);
1732                 memcpy (pkt->data+offset, pkt_data, pkt_size);
1733
1734                 if (pkt_data != data)
1735                     av_free(pkt_data);
1736
1737                 if (n == 0)
1738                     pkt->flags = is_keyframe;
1739                 pkt->stream_index = st->index;
1740
1741                 if (track->ms_compat)
1742                     pkt->dts = timecode;
1743                 else
1744                     pkt->pts = timecode;
1745                 pkt->pos = pos;
1746                 if (st->codec->codec_id == CODEC_ID_TEXT)
1747                     pkt->convergence_duration = duration;
1748                 else if (track->type != MATROSKA_TRACK_TYPE_SUBTITLE)
1749                     pkt->duration = duration;
1750
1751                 if (st->codec->codec_id == CODEC_ID_SSA)
1752                     matroska_fix_ass_packet(matroska, pkt, duration);
1753
1754                 if (matroska->prev_pkt &&
1755                     timecode != AV_NOPTS_VALUE &&
1756                     matroska->prev_pkt->pts == timecode &&
1757                     matroska->prev_pkt->stream_index == st->index)
1758                     matroska_merge_packets(matroska->prev_pkt, pkt);
1759                 else {
1760                     dynarray_add(&matroska->packets,&matroska->num_packets,pkt);
1761                     matroska->prev_pkt = pkt;
1762                 }
1763             }
1764
1765             if (timecode != AV_NOPTS_VALUE)
1766                 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
1767             data += lace_size[n];
1768             size -= lace_size[n];
1769         }
1770     }
1771
1772     av_free(lace_size);
1773     return res;
1774 }
1775
1776 static int matroska_parse_cluster(MatroskaDemuxContext *matroska)
1777 {
1778     MatroskaCluster cluster = { 0 };
1779     EbmlList *blocks_list;
1780     MatroskaBlock *blocks;
1781     int i, res;
1782     int64_t pos = url_ftell(matroska->ctx->pb);
1783     matroska->prev_pkt = NULL;
1784     if (matroska->has_cluster_id){
1785         /* For the first cluster we parse, its ID was already read as
1786            part of matroska_read_header(), so don't read it again */
1787         res = ebml_parse_id(matroska, matroska_clusters,
1788                             MATROSKA_ID_CLUSTER, &cluster);
1789         pos -= 4;  /* sizeof the ID which was already read */
1790         matroska->has_cluster_id = 0;
1791     } else
1792         res = ebml_parse(matroska, matroska_clusters, &cluster);
1793     blocks_list = &cluster.blocks;
1794     blocks = blocks_list->elem;
1795     for (i=0; i<blocks_list->nb_elem; i++)
1796         if (blocks[i].bin.size > 0) {
1797             int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
1798             res=matroska_parse_block(matroska,
1799                                      blocks[i].bin.data, blocks[i].bin.size,
1800                                      blocks[i].bin.pos,  cluster.timecode,
1801                                      blocks[i].duration, is_keyframe,
1802                                      pos);
1803         }
1804     ebml_free(matroska_cluster, &cluster);
1805     if (res < 0)  matroska->done = 1;
1806     return res;
1807 }
1808
1809 static int matroska_read_packet(AVFormatContext *s, AVPacket *pkt)
1810 {
1811     MatroskaDemuxContext *matroska = s->priv_data;
1812
1813     while (matroska_deliver_packet(matroska, pkt)) {
1814         if (matroska->done)
1815             return AVERROR_EOF;
1816         matroska_parse_cluster(matroska);
1817     }
1818
1819     return 0;
1820 }
1821
1822 static int matroska_read_seek(AVFormatContext *s, int stream_index,
1823                               int64_t timestamp, int flags)
1824 {
1825     MatroskaDemuxContext *matroska = s->priv_data;
1826     MatroskaTrack *tracks = matroska->tracks.elem;
1827     AVStream *st = s->streams[stream_index];
1828     int i, index, index_sub, index_min;
1829
1830     if (!st->nb_index_entries)
1831         return 0;
1832     timestamp = FFMAX(timestamp, st->index_entries[0].timestamp);
1833
1834     if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
1835         url_fseek(s->pb, st->index_entries[st->nb_index_entries-1].pos, SEEK_SET);
1836         while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
1837             matroska_clear_queue(matroska);
1838             if (matroska_parse_cluster(matroska) < 0)
1839                 break;
1840         }
1841     }
1842
1843     matroska_clear_queue(matroska);
1844     if (index < 0)
1845         return 0;
1846
1847     index_min = index;
1848     for (i=0; i < matroska->tracks.nb_elem; i++) {
1849         tracks[i].end_timecode = 0;
1850         if (tracks[i].type == MATROSKA_TRACK_TYPE_SUBTITLE
1851             && !tracks[i].stream->discard != AVDISCARD_ALL) {
1852             index_sub = av_index_search_timestamp(tracks[i].stream, st->index_entries[index].timestamp, AVSEEK_FLAG_BACKWARD);
1853             if (index_sub >= 0
1854                 && st->index_entries[index_sub].pos < st->index_entries[index_min].pos
1855                 && st->index_entries[index].timestamp - st->index_entries[index_sub].timestamp < 30000000000/matroska->time_scale)
1856                 index_min = index_sub;
1857         }
1858     }
1859
1860     url_fseek(s->pb, st->index_entries[index_min].pos, SEEK_SET);
1861     matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
1862     matroska->skip_to_timecode = st->index_entries[index].timestamp;
1863     matroska->done = 0;
1864     av_update_cur_dts(s, st, st->index_entries[index].timestamp);
1865     return 0;
1866 }
1867
1868 static int matroska_read_close(AVFormatContext *s)
1869 {
1870     MatroskaDemuxContext *matroska = s->priv_data;
1871     MatroskaTrack *tracks = matroska->tracks.elem;
1872     int n;
1873
1874     matroska_clear_queue(matroska);
1875
1876     for (n=0; n < matroska->tracks.nb_elem; n++)
1877         if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
1878             av_free(tracks[n].audio.buf);
1879     ebml_free(matroska_segment, matroska);
1880
1881     return 0;
1882 }
1883
1884 AVInputFormat matroska_demuxer = {
1885     "matroska",
1886     NULL_IF_CONFIG_SMALL("Matroska file format"),
1887     sizeof(MatroskaDemuxContext),
1888     matroska_probe,
1889     matroska_read_header,
1890     matroska_read_packet,
1891     matroska_read_close,
1892     matroska_read_seek,
1893     .metadata_conv = ff_mkv_metadata_conv,
1894 };