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 #include "avio_internal.h"
  35 /* For ff_codec_get_id(). */
  36 #include "riff.h"
  37 #include "isom.h"
  38 #include "rm.h"
  39 #include "matroska.h"
  40 #include "libavcodec/mpeg4audio.h"
  41 #include "libavutil/intfloat_readwrite.h"
  42 #include "libavutil/intreadwrite.h"
  43 #include "libavutil/avstring.h"
  44 #include "libavutil/lzo.h"
  45 #if CONFIG_ZLIB
  46 #include <zlib.h>
  47 #endif
  48 #if CONFIG_BZLIB
  49 #include <bzlib.h>
  50 #endif
  51
  52 typedef enum {
  53     EBML_NONE,
  54     EBML_UINT,
  55     EBML_FLOAT,
  56     EBML_STR,
  57     EBML_UTF8,
  58     EBML_BIN,
  59     EBML_NEST,
  60     EBML_PASS,
  61     EBML_STOP,
  62     EBML_TYPE_COUNT
  63 } EbmlType;
  64
  65 typedef const struct EbmlSyntax {
  66     uint32_t id;
  67     EbmlType type;
  68     int list_elem_size;
  69     int data_offset;
  70     union {
  71         uint64_t    u;
  72         double      f;
  73         const char *s;
  74         const struct EbmlSyntax *n;
  75     } def;
  76 } EbmlSyntax;
  77
  78 typedef struct {
  79     int nb_elem;
  80     void *elem;
  81 } EbmlList;
  82
  83 typedef struct {
  84     int      size;
  85     uint8_t *data;
  86     int64_t  pos;
  87 } EbmlBin;
  88
  89 typedef struct {
  90     uint64_t version;
  91     uint64_t max_size;
  92     uint64_t id_length;
  93     char    *doctype;
  94     uint64_t doctype_version;
  95 } Ebml;
  96
  97 typedef struct {
  98     uint64_t algo;
  99     EbmlBin  settings;
 100 } MatroskaTrackCompression;
 101
 102 typedef struct {
 103     uint64_t scope;
 104     uint64_t type;
 105     MatroskaTrackCompression compression;
 106 } MatroskaTrackEncoding;
 107
 108 typedef struct {
 109     double   frame_rate;
 110     uint64_t display_width;
 111     uint64_t display_height;
 112     uint64_t pixel_width;
 113     uint64_t pixel_height;
 114     uint64_t fourcc;
 115 } MatroskaTrackVideo;
 116
 117 typedef struct {
 118     double   samplerate;
 119     double   out_samplerate;
 120     uint64_t bitdepth;
 121     uint64_t channels;
 122
 123     /* real audio header (extracted from extradata) */
 124     int      coded_framesize;
 125     int      sub_packet_h;
 126     int      frame_size;
 127     int      sub_packet_size;
 128     int      sub_packet_cnt;
 129     int      pkt_cnt;
 130     uint64_t buf_timecode;
 131     uint8_t *buf;
 132 } MatroskaTrackAudio;
 133
 134 typedef struct {
 135     uint64_t num;
 136     uint64_t uid;
 137     uint64_t type;
 138     char    *name;
 139     char    *codec_id;
 140     EbmlBin  codec_priv;
 141     char    *language;
 142     double time_scale;
 143     uint64_t default_duration;
 144     uint64_t flag_default;
 145     uint64_t flag_forced;
 146     MatroskaTrackVideo video;
 147     MatroskaTrackAudio audio;
 148     EbmlList encodings;
 149
 150     AVStream *stream;
 151     int64_t end_timecode;
 152     int ms_compat;
 153 } MatroskaTrack;
 154
 155 typedef struct {
 156     uint64_t uid;
 157     char *filename;
 158     char *mime;
 159     EbmlBin bin;
 160
 161     AVStream *stream;
 162 } MatroskaAttachement;
 163
 164 typedef struct {
 165     uint64_t start;
 166     uint64_t end;
 167     uint64_t uid;
 168     char    *title;
 169
 170     AVChapter *chapter;
 171 } MatroskaChapter;
 172
 173 typedef struct {
 174     uint64_t track;
 175     uint64_t pos;
 176 } MatroskaIndexPos;
 177
 178 typedef struct {
 179     uint64_t time;
 180     EbmlList pos;
 181 } MatroskaIndex;
 182
 183 typedef struct {
 184     char *name;
 185     char *string;
 186     char *lang;
 187     uint64_t def;
 188     EbmlList sub;
 189 } MatroskaTag;
 190
 191 typedef struct {
 192     char    *type;
 193     uint64_t typevalue;
 194     uint64_t trackuid;
 195     uint64_t chapteruid;
 196     uint64_t attachuid;
 197 } MatroskaTagTarget;
 198
 199 typedef struct {
 200     MatroskaTagTarget target;
 201     EbmlList tag;
 202 } MatroskaTags;
 203
 204 typedef struct {
 205     uint64_t id;
 206     uint64_t pos;
 207 } MatroskaSeekhead;
 208
 209 typedef struct {
 210     uint64_t start;
 211     uint64_t length;
 212 } MatroskaLevel;
 213
 214 typedef struct {
 215     AVFormatContext *ctx;
 216
 217     /* EBML stuff */
 218     int num_levels;
 219     MatroskaLevel levels[EBML_MAX_DEPTH];
 220     int level_up;
 221     uint32_t current_id;
 222
 223     uint64_t time_scale;
 224     double   duration;
 225     char    *title;
 226     EbmlList tracks;
 227     EbmlList attachments;
 228     EbmlList chapters;
 229     EbmlList index;
 230     EbmlList tags;
 231     EbmlList seekhead;
 232
 233     /* byte position of the segment inside the stream */
 234     int64_t segment_start;
 235
 236     /* the packet queue */
 237     AVPacket **packets;
 238     int num_packets;
 239     AVPacket *prev_pkt;
 240
 241     int done;
 242
 243     /* What to skip before effectively reading a packet. */
 244     int skip_to_keyframe;
 245     uint64_t skip_to_timecode;
 246 } MatroskaDemuxContext;
 247
 248 typedef struct {
 249     uint64_t duration;
 250     int64_t  reference;
 251     uint64_t non_simple;
 252     EbmlBin  bin;
 253 } MatroskaBlock;
 254
 255 typedef struct {
 256     uint64_t timecode;
 257     EbmlList blocks;
 258 } MatroskaCluster;
 259
 260 static EbmlSyntax ebml_header[] = {
 261     { EBML_ID_EBMLREADVERSION,        EBML_UINT, 0, offsetof(Ebml,version), {.u=EBML_VERSION} },
 262     { EBML_ID_EBMLMAXSIZELENGTH,      EBML_UINT, 0, offsetof(Ebml,max_size), {.u=8} },
 263     { EBML_ID_EBMLMAXIDLENGTH,        EBML_UINT, 0, offsetof(Ebml,id_length), {.u=4} },
 264     { EBML_ID_DOCTYPE,                EBML_STR,  0, offsetof(Ebml,doctype), {.s="(none)"} },
 265     { EBML_ID_DOCTYPEREADVERSION,     EBML_UINT, 0, offsetof(Ebml,doctype_version), {.u=1} },
 266     { EBML_ID_EBMLVERSION,            EBML_NONE },
 267     { EBML_ID_DOCTYPEVERSION,         EBML_NONE },
 268     { 0 }
 269 };
 270
 271 static EbmlSyntax ebml_syntax[] = {
 272     { EBML_ID_HEADER,                 EBML_NEST, 0, 0, {.n=ebml_header} },
 273     { 0 }
 274 };
 275
 276 static EbmlSyntax matroska_info[] = {
 277     { MATROSKA_ID_TIMECODESCALE,      EBML_UINT,  0, offsetof(MatroskaDemuxContext,time_scale), {.u=1000000} },
 278     { MATROSKA_ID_DURATION,           EBML_FLOAT, 0, offsetof(MatroskaDemuxContext,duration) },
 279     { MATROSKA_ID_TITLE,              EBML_UTF8,  0, offsetof(MatroskaDemuxContext,title) },
 280     { MATROSKA_ID_WRITINGAPP,         EBML_NONE },
 281     { MATROSKA_ID_MUXINGAPP,          EBML_NONE },
 282     { MATROSKA_ID_DATEUTC,            EBML_NONE },
 283     { MATROSKA_ID_SEGMENTUID,         EBML_NONE },
 284     { 0 }
 285 };
 286
 287 static EbmlSyntax matroska_track_video[] = {
 288     { MATROSKA_ID_VIDEOFRAMERATE,     EBML_FLOAT,0, offsetof(MatroskaTrackVideo,frame_rate) },
 289     { MATROSKA_ID_VIDEODISPLAYWIDTH,  EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_width) },
 290     { MATROSKA_ID_VIDEODISPLAYHEIGHT, EBML_UINT, 0, offsetof(MatroskaTrackVideo,display_height) },
 291     { MATROSKA_ID_VIDEOPIXELWIDTH,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_width) },
 292     { MATROSKA_ID_VIDEOPIXELHEIGHT,   EBML_UINT, 0, offsetof(MatroskaTrackVideo,pixel_height) },
 293     { MATROSKA_ID_VIDEOCOLORSPACE,    EBML_UINT, 0, offsetof(MatroskaTrackVideo,fourcc) },
 294     { MATROSKA_ID_VIDEOPIXELCROPB,    EBML_NONE },
 295     { MATROSKA_ID_VIDEOPIXELCROPT,    EBML_NONE },
 296     { MATROSKA_ID_VIDEOPIXELCROPL,    EBML_NONE },
 297     { MATROSKA_ID_VIDEOPIXELCROPR,    EBML_NONE },
 298     { MATROSKA_ID_VIDEODISPLAYUNIT,   EBML_NONE },
 299     { MATROSKA_ID_VIDEOFLAGINTERLACED,EBML_NONE },
 300     { MATROSKA_ID_VIDEOSTEREOMODE,    EBML_NONE },
 301     { MATROSKA_ID_VIDEOASPECTRATIO,   EBML_NONE },
 302     { 0 }
 303 };
 304
 305 static EbmlSyntax matroska_track_audio[] = {
 306     { MATROSKA_ID_AUDIOSAMPLINGFREQ,  EBML_FLOAT,0, offsetof(MatroskaTrackAudio,samplerate), {.f=8000.0} },
 307     { MATROSKA_ID_AUDIOOUTSAMPLINGFREQ,EBML_FLOAT,0,offsetof(MatroskaTrackAudio,out_samplerate) },
 308     { MATROSKA_ID_AUDIOBITDEPTH,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,bitdepth) },
 309     { MATROSKA_ID_AUDIOCHANNELS,      EBML_UINT, 0, offsetof(MatroskaTrackAudio,channels), {.u=1} },
 310     { 0 }
 311 };
 312
 313 static EbmlSyntax matroska_track_encoding_compression[] = {
 314     { MATROSKA_ID_ENCODINGCOMPALGO,   EBML_UINT, 0, offsetof(MatroskaTrackCompression,algo), {.u=0} },
 315     { MATROSKA_ID_ENCODINGCOMPSETTINGS,EBML_BIN, 0, offsetof(MatroskaTrackCompression,settings) },
 316     { 0 }
 317 };
 318
 319 static EbmlSyntax matroska_track_encoding[] = {
 320     { MATROSKA_ID_ENCODINGSCOPE,      EBML_UINT, 0, offsetof(MatroskaTrackEncoding,scope), {.u=1} },
 321     { MATROSKA_ID_ENCODINGTYPE,       EBML_UINT, 0, offsetof(MatroskaTrackEncoding,type), {.u=0} },
 322     { MATROSKA_ID_ENCODINGCOMPRESSION,EBML_NEST, 0, offsetof(MatroskaTrackEncoding,compression), {.n=matroska_track_encoding_compression} },
 323     { MATROSKA_ID_ENCODINGORDER,      EBML_NONE },
 324     { 0 }
 325 };
 326
 327 static EbmlSyntax matroska_track_encodings[] = {
 328     { MATROSKA_ID_TRACKCONTENTENCODING, EBML_NEST, sizeof(MatroskaTrackEncoding), offsetof(MatroskaTrack,encodings), {.n=matroska_track_encoding} },
 329     { 0 }
 330 };
 331
 332 static EbmlSyntax matroska_track[] = {
 333     { MATROSKA_ID_TRACKNUMBER,          EBML_UINT, 0, offsetof(MatroskaTrack,num) },
 334     { MATROSKA_ID_TRACKNAME,            EBML_UTF8, 0, offsetof(MatroskaTrack,name) },
 335     { MATROSKA_ID_TRACKUID,             EBML_UINT, 0, offsetof(MatroskaTrack,uid) },
 336     { MATROSKA_ID_TRACKTYPE,            EBML_UINT, 0, offsetof(MatroskaTrack,type) },
 337     { MATROSKA_ID_CODECID,              EBML_STR,  0, offsetof(MatroskaTrack,codec_id) },
 338     { MATROSKA_ID_CODECPRIVATE,         EBML_BIN,  0, offsetof(MatroskaTrack,codec_priv) },
 339     { MATROSKA_ID_TRACKLANGUAGE,        EBML_UTF8, 0, offsetof(MatroskaTrack,language), {.s="eng"} },
 340     { MATROSKA_ID_TRACKDEFAULTDURATION, EBML_UINT, 0, offsetof(MatroskaTrack,default_duration) },
 341     { MATROSKA_ID_TRACKTIMECODESCALE,   EBML_FLOAT,0, offsetof(MatroskaTrack,time_scale), {.f=1.0} },
 342     { MATROSKA_ID_TRACKFLAGDEFAULT,     EBML_UINT, 0, offsetof(MatroskaTrack,flag_default), {.u=1} },
 343     { MATROSKA_ID_TRACKFLAGFORCED,      EBML_UINT, 0, offsetof(MatroskaTrack,flag_forced), {.u=0} },
 344     { MATROSKA_ID_TRACKVIDEO,           EBML_NEST, 0, offsetof(MatroskaTrack,video), {.n=matroska_track_video} },
 345     { MATROSKA_ID_TRACKAUDIO,           EBML_NEST, 0, offsetof(MatroskaTrack,audio), {.n=matroska_track_audio} },
 346     { MATROSKA_ID_TRACKCONTENTENCODINGS,EBML_NEST, 0, 0, {.n=matroska_track_encodings} },
 347     { MATROSKA_ID_TRACKFLAGENABLED,     EBML_NONE },
 348     { MATROSKA_ID_TRACKFLAGLACING,      EBML_NONE },
 349     { MATROSKA_ID_CODECNAME,            EBML_NONE },
 350     { MATROSKA_ID_CODECDECODEALL,       EBML_NONE },
 351     { MATROSKA_ID_CODECINFOURL,         EBML_NONE },
 352     { MATROSKA_ID_CODECDOWNLOADURL,     EBML_NONE },
 353     { MATROSKA_ID_TRACKMINCACHE,        EBML_NONE },
 354     { MATROSKA_ID_TRACKMAXCACHE,        EBML_NONE },
 355     { MATROSKA_ID_TRACKMAXBLKADDID,     EBML_NONE },
 356     { 0 }
 357 };
 358
 359 static EbmlSyntax matroska_tracks[] = {
 360     { MATROSKA_ID_TRACKENTRY,         EBML_NEST, sizeof(MatroskaTrack), offsetof(MatroskaDemuxContext,tracks), {.n=matroska_track} },
 361     { 0 }
 362 };
 363
 364 static EbmlSyntax matroska_attachment[] = {
 365     { MATROSKA_ID_FILEUID,            EBML_UINT, 0, offsetof(MatroskaAttachement,uid) },
 366     { MATROSKA_ID_FILENAME,           EBML_UTF8, 0, offsetof(MatroskaAttachement,filename) },
 367     { MATROSKA_ID_FILEMIMETYPE,       EBML_STR,  0, offsetof(MatroskaAttachement,mime) },
 368     { MATROSKA_ID_FILEDATA,           EBML_BIN,  0, offsetof(MatroskaAttachement,bin) },
 369     { MATROSKA_ID_FILEDESC,           EBML_NONE },
 370     { 0 }
 371 };
 372
 373 static EbmlSyntax matroska_attachments[] = {
 374     { MATROSKA_ID_ATTACHEDFILE,       EBML_NEST, sizeof(MatroskaAttachement), offsetof(MatroskaDemuxContext,attachments), {.n=matroska_attachment} },
 375     { 0 }
 376 };
 377
 378 static EbmlSyntax matroska_chapter_display[] = {
 379     { MATROSKA_ID_CHAPSTRING,         EBML_UTF8, 0, offsetof(MatroskaChapter,title) },
 380     { MATROSKA_ID_CHAPLANG,           EBML_NONE },
 381     { 0 }
 382 };
 383
 384 static EbmlSyntax matroska_chapter_entry[] = {
 385     { MATROSKA_ID_CHAPTERTIMESTART,   EBML_UINT, 0, offsetof(MatroskaChapter,start), {.u=AV_NOPTS_VALUE} },
 386     { MATROSKA_ID_CHAPTERTIMEEND,     EBML_UINT, 0, offsetof(MatroskaChapter,end), {.u=AV_NOPTS_VALUE} },
 387     { MATROSKA_ID_CHAPTERUID,         EBML_UINT, 0, offsetof(MatroskaChapter,uid) },
 388     { MATROSKA_ID_CHAPTERDISPLAY,     EBML_NEST, 0, 0, {.n=matroska_chapter_display} },
 389     { MATROSKA_ID_CHAPTERFLAGHIDDEN,  EBML_NONE },
 390     { MATROSKA_ID_CHAPTERFLAGENABLED, EBML_NONE },
 391     { MATROSKA_ID_CHAPTERPHYSEQUIV,   EBML_NONE },
 392     { MATROSKA_ID_CHAPTERATOM,        EBML_NONE },
 393     { 0 }
 394 };
 395
 396 static EbmlSyntax matroska_chapter[] = {
 397     { MATROSKA_ID_CHAPTERATOM,        EBML_NEST, sizeof(MatroskaChapter), offsetof(MatroskaDemuxContext,chapters), {.n=matroska_chapter_entry} },
 398     { MATROSKA_ID_EDITIONUID,         EBML_NONE },
 399     { MATROSKA_ID_EDITIONFLAGHIDDEN,  EBML_NONE },
 400     { MATROSKA_ID_EDITIONFLAGDEFAULT, EBML_NONE },
 401     { MATROSKA_ID_EDITIONFLAGORDERED, EBML_NONE },
 402     { 0 }
 403 };
 404
 405 static EbmlSyntax matroska_chapters[] = {
 406     { MATROSKA_ID_EDITIONENTRY,       EBML_NEST, 0, 0, {.n=matroska_chapter} },
 407     { 0 }
 408 };
 409
 410 static EbmlSyntax matroska_index_pos[] = {
 411     { MATROSKA_ID_CUETRACK,           EBML_UINT, 0, offsetof(MatroskaIndexPos,track) },
 412     { MATROSKA_ID_CUECLUSTERPOSITION, EBML_UINT, 0, offsetof(MatroskaIndexPos,pos)   },
 413     { MATROSKA_ID_CUEBLOCKNUMBER,     EBML_NONE },
 414     { 0 }
 415 };
 416
 417 static EbmlSyntax matroska_index_entry[] = {
 418     { MATROSKA_ID_CUETIME,            EBML_UINT, 0, offsetof(MatroskaIndex,time) },
 419     { MATROSKA_ID_CUETRACKPOSITION,   EBML_NEST, sizeof(MatroskaIndexPos), offsetof(MatroskaIndex,pos), {.n=matroska_index_pos} },
 420     { 0 }
 421 };
 422
 423 static EbmlSyntax matroska_index[] = {
 424     { MATROSKA_ID_POINTENTRY,         EBML_NEST, sizeof(MatroskaIndex), offsetof(MatroskaDemuxContext,index), {.n=matroska_index_entry} },
 425     { 0 }
 426 };
 427
 428 static EbmlSyntax matroska_simpletag[] = {
 429     { MATROSKA_ID_TAGNAME,            EBML_UTF8, 0, offsetof(MatroskaTag,name) },
 430     { MATROSKA_ID_TAGSTRING,          EBML_UTF8, 0, offsetof(MatroskaTag,string) },
 431     { MATROSKA_ID_TAGLANG,            EBML_STR,  0, offsetof(MatroskaTag,lang), {.s="und"} },
 432     { MATROSKA_ID_TAGDEFAULT,         EBML_UINT, 0, offsetof(MatroskaTag,def) },
 433     { MATROSKA_ID_TAGDEFAULT_BUG,     EBML_UINT, 0, offsetof(MatroskaTag,def) },
 434     { MATROSKA_ID_SIMPLETAG,          EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTag,sub), {.n=matroska_simpletag} },
 435     { 0 }
 436 };
 437
 438 static EbmlSyntax matroska_tagtargets[] = {
 439     { MATROSKA_ID_TAGTARGETS_TYPE,      EBML_STR,  0, offsetof(MatroskaTagTarget,type) },
 440     { MATROSKA_ID_TAGTARGETS_TYPEVALUE, EBML_UINT, 0, offsetof(MatroskaTagTarget,typevalue), {.u=50} },
 441     { MATROSKA_ID_TAGTARGETS_TRACKUID,  EBML_UINT, 0, offsetof(MatroskaTagTarget,trackuid) },
 442     { MATROSKA_ID_TAGTARGETS_CHAPTERUID,EBML_UINT, 0, offsetof(MatroskaTagTarget,chapteruid) },
 443     { MATROSKA_ID_TAGTARGETS_ATTACHUID, EBML_UINT, 0, offsetof(MatroskaTagTarget,attachuid) },
 444     { 0 }
 445 };
 446
 447 static EbmlSyntax matroska_tag[] = {
 448     { MATROSKA_ID_SIMPLETAG,          EBML_NEST, sizeof(MatroskaTag), offsetof(MatroskaTags,tag), {.n=matroska_simpletag} },
 449     { MATROSKA_ID_TAGTARGETS,         EBML_NEST, 0, offsetof(MatroskaTags,target), {.n=matroska_tagtargets} },
 450     { 0 }
 451 };
 452
 453 static EbmlSyntax matroska_tags[] = {
 454     { MATROSKA_ID_TAG,                EBML_NEST, sizeof(MatroskaTags), offsetof(MatroskaDemuxContext,tags), {.n=matroska_tag} },
 455     { 0 }
 456 };
 457
 458 static EbmlSyntax matroska_seekhead_entry[] = {
 459     { MATROSKA_ID_SEEKID,             EBML_UINT, 0, offsetof(MatroskaSeekhead,id) },
 460     { MATROSKA_ID_SEEKPOSITION,       EBML_UINT, 0, offsetof(MatroskaSeekhead,pos), {.u=-1} },
 461     { 0 }
 462 };
 463
 464 static EbmlSyntax matroska_seekhead[] = {
 465     { MATROSKA_ID_SEEKENTRY,          EBML_NEST, sizeof(MatroskaSeekhead), offsetof(MatroskaDemuxContext,seekhead), {.n=matroska_seekhead_entry} },
 466     { 0 }
 467 };
 468
 469 static EbmlSyntax matroska_segment[] = {
 470     { MATROSKA_ID_INFO,           EBML_NEST, 0, 0, {.n=matroska_info       } },
 471     { MATROSKA_ID_TRACKS,         EBML_NEST, 0, 0, {.n=matroska_tracks     } },
 472     { MATROSKA_ID_ATTACHMENTS,    EBML_NEST, 0, 0, {.n=matroska_attachments} },
 473     { MATROSKA_ID_CHAPTERS,       EBML_NEST, 0, 0, {.n=matroska_chapters   } },
 474     { MATROSKA_ID_CUES,           EBML_NEST, 0, 0, {.n=matroska_index      } },
 475     { MATROSKA_ID_TAGS,           EBML_NEST, 0, 0, {.n=matroska_tags       } },
 476     { MATROSKA_ID_SEEKHEAD,       EBML_NEST, 0, 0, {.n=matroska_seekhead   } },
 477     { MATROSKA_ID_CLUSTER,        EBML_STOP },
 478     { 0 }
 479 };
 480
 481 static EbmlSyntax matroska_segments[] = {
 482     { MATROSKA_ID_SEGMENT,        EBML_NEST, 0, 0, {.n=matroska_segment    } },
 483     { 0 }
 484 };
 485
 486 static EbmlSyntax matroska_blockgroup[] = {
 487     { MATROSKA_ID_BLOCK,          EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
 488     { MATROSKA_ID_SIMPLEBLOCK,    EBML_BIN,  0, offsetof(MatroskaBlock,bin) },
 489     { MATROSKA_ID_BLOCKDURATION,  EBML_UINT, 0, offsetof(MatroskaBlock,duration) },
 490     { MATROSKA_ID_BLOCKREFERENCE, EBML_UINT, 0, offsetof(MatroskaBlock,reference) },
 491     { 1,                          EBML_UINT, 0, offsetof(MatroskaBlock,non_simple), {.u=1} },
 492     { 0 }
 493 };
 494
 495 static EbmlSyntax matroska_cluster[] = {
 496     { MATROSKA_ID_CLUSTERTIMECODE,EBML_UINT,0, offsetof(MatroskaCluster,timecode) },
 497     { MATROSKA_ID_BLOCKGROUP,     EBML_NEST, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
 498     { MATROSKA_ID_SIMPLEBLOCK,    EBML_PASS, sizeof(MatroskaBlock), offsetof(MatroskaCluster,blocks), {.n=matroska_blockgroup} },
 499     { MATROSKA_ID_CLUSTERPOSITION,EBML_NONE },
 500     { MATROSKA_ID_CLUSTERPREVSIZE,EBML_NONE },
 501     { 0 }
 502 };
 503
 504 static EbmlSyntax matroska_clusters[] = {
 505     { MATROSKA_ID_CLUSTER,        EBML_NEST, 0, 0, {.n=matroska_cluster} },
 506     { MATROSKA_ID_INFO,           EBML_NONE },
 507     { MATROSKA_ID_CUES,           EBML_NONE },
 508     { MATROSKA_ID_TAGS,           EBML_NONE },
 509     { MATROSKA_ID_SEEKHEAD,       EBML_NONE },
 510     { 0 }
 511 };
 512
 513 static const char *matroska_doctypes[] = { "matroska", "webm" };
 514
 515 /*
 516  * Return: Whether we reached the end of a level in the hierarchy or not.
 517  */
 518 static int ebml_level_end(MatroskaDemuxContext *matroska)
 519 {
 520     AVIOContext *pb = matroska->ctx->pb;
 521     int64_t pos = avio_tell(pb);
 522
 523     if (matroska->num_levels > 0) {
 524         MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
 525         if (pos - level->start >= level->length || matroska->current_id) {
 526             matroska->num_levels--;
 527             return 1;
 528         }
 529     }
 530     return 0;
 531 }
 532
 533 /*
 534  * Read: an "EBML number", which is defined as a variable-length
 535  * array of bytes. The first byte indicates the length by giving a
 536  * number of 0-bits followed by a one. The position of the first
 537  * "one" bit inside the first byte indicates the length of this
 538  * number.
 539  * Returns: number of bytes read, < 0 on error
 540  */
 541 static int ebml_read_num(MatroskaDemuxContext *matroska, AVIOContext *pb,
 542                          int max_size, uint64_t *number)
 543 {
 544     int read = 1, n = 1;
 545     uint64_t total = 0;
 546
 547     /* The first byte tells us the length in bytes - avio_r8() can normally
 548      * return 0, but since that's not a valid first ebmlID byte, we can
 549      * use it safely here to catch EOS. */
 550     if (!(total = avio_r8(pb))) {
 551         /* we might encounter EOS here */
 552         if (!url_feof(pb)) {
 553             int64_t pos = avio_tell(pb);
 554             av_log(matroska->ctx, AV_LOG_ERROR,
 555                    "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
 556                    pos, pos);
 557         }
 558         return AVERROR(EIO); /* EOS or actual I/O error */
 559     }
 560
 561     /* get the length of the EBML number */
 562     read = 8 - ff_log2_tab[total];
 563     if (read > max_size) {
 564         int64_t pos = avio_tell(pb) - 1;
 565         av_log(matroska->ctx, AV_LOG_ERROR,
 566                "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
 567                (uint8_t) total, pos, pos);
 568         return AVERROR_INVALIDDATA;
 569     }
 570
 571     /* read out length */
 572     total ^= 1 << ff_log2_tab[total];
 573     while (n++ < read)
 574         total = (total << 8) | avio_r8(pb);
 575
 576     *number = total;
 577
 578     return read;
 579 }
 580
 581 /**
 582  * Read a EBML length value.
 583  * This needs special handling for the "unknown length" case which has multiple
 584  * encodings.
 585  */
 586 static int ebml_read_length(MatroskaDemuxContext *matroska, AVIOContext *pb,
 587                             uint64_t *number)
 588 {
 589     int res = ebml_read_num(matroska, pb, 8, number);
 590     if (res > 0 && *number + 1 == 1ULL << (7 * res))
 591         *number = 0xffffffffffffffULL;
 592     return res;
 593 }
 594
 595 /*
 596  * Read the next element as an unsigned int.
 597  * 0 is success, < 0 is failure.
 598  */
 599 static int ebml_read_uint(AVIOContext *pb, int size, uint64_t *num)
 600 {
 601     int n = 0;
 602
 603     if (size > 8)
 604         return AVERROR_INVALIDDATA;
 605
 606     /* big-endian ordering; build up number */
 607     *num = 0;
 608     while (n++ < size)
 609         *num = (*num << 8) | avio_r8(pb);
 610
 611     return 0;
 612 }
 613
 614 /*
 615  * Read the next element as a float.
 616  * 0 is success, < 0 is failure.
 617  */
 618 static int ebml_read_float(AVIOContext *pb, int size, double *num)
 619 {
 620     if (size == 0) {
 621         *num = 0;
 622     } else if (size == 4) {
 623         *num= av_int2flt(avio_rb32(pb));
 624     } else if(size==8){
 625         *num= av_int2dbl(avio_rb64(pb));
 626     } else
 627         return AVERROR_INVALIDDATA;
 628
 629     return 0;
 630 }
 631
 632 /*
 633  * Read the next element as an ASCII string.
 634  * 0 is success, < 0 is failure.
 635  */
 636 static int ebml_read_ascii(AVIOContext *pb, int size, char **str)
 637 {
 638     av_free(*str);
 639     /* EBML strings are usually not 0-terminated, so we allocate one
 640      * byte more, read the string and NULL-terminate it ourselves. */
 641     if (!(*str = av_malloc(size + 1)))
 642         return AVERROR(ENOMEM);
 643     if (avio_read(pb, (uint8_t *) *str, size) != size) {
 644         av_freep(str);
 645         return AVERROR(EIO);
 646     }
 647     (*str)[size] = '\0';
 648
 649     return 0;
 650 }
 651
 652 /*
 653  * Read the next element as binary data.
 654  * 0 is success, < 0 is failure.
 655  */
 656 static int ebml_read_binary(AVIOContext *pb, int length, EbmlBin *bin)
 657 {
 658     av_free(bin->data);
 659     if (!(bin->data = av_malloc(length)))
 660         return AVERROR(ENOMEM);
 661
 662     bin->size = length;
 663     bin->pos  = avio_tell(pb);
 664     if (avio_read(pb, bin->data, length) != length) {
 665         av_freep(&bin->data);
 666         return AVERROR(EIO);
 667     }
 668
 669     return 0;
 670 }
 671
 672 /*
 673  * Read the next element, but only the header. The contents
 674  * are supposed to be sub-elements which can be read separately.
 675  * 0 is success, < 0 is failure.
 676  */
 677 static int ebml_read_master(MatroskaDemuxContext *matroska, uint64_t length)
 678 {
 679     AVIOContext *pb = matroska->ctx->pb;
 680     MatroskaLevel *level;
 681
 682     if (matroska->num_levels >= EBML_MAX_DEPTH) {
 683         av_log(matroska->ctx, AV_LOG_ERROR,
 684                "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
 685         return AVERROR(ENOSYS);
 686     }
 687
 688     level = &matroska->levels[matroska->num_levels++];
 689     level->start = avio_tell(pb);
 690     level->length = length;
 691
 692     return 0;
 693 }
 694
 695 /*
 696  * Read signed/unsigned "EBML" numbers.
 697  * Return: number of bytes processed, < 0 on error
 698  */
 699 static int matroska_ebmlnum_uint(MatroskaDemuxContext *matroska,
 700                                  uint8_t *data, uint32_t size, uint64_t *num)
 701 {
 702     AVIOContext pb;
 703     ffio_init_context(&pb, data, size, 0, NULL, NULL, NULL, NULL);
 704     return ebml_read_num(matroska, &pb, FFMIN(size, 8), num);
 705 }
 706
 707 /*
 708  * Same as above, but signed.
 709  */
 710 static int matroska_ebmlnum_sint(MatroskaDemuxContext *matroska,
 711                                  uint8_t *data, uint32_t size, int64_t *num)
 712 {
 713     uint64_t unum;
 714     int res;
 715
 716     /* read as unsigned number first */
 717     if ((res = matroska_ebmlnum_uint(matroska, data, size, &unum)) < 0)
 718         return res;
 719
 720     /* make signed (weird way) */
 721     *num = unum - ((1LL << (7*res - 1)) - 1);
 722
 723     return res;
 724 }
 725
 726 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
 727                            EbmlSyntax *syntax, void *data);
 728
 729 static int ebml_parse_id(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
 730                          uint32_t id, void *data)
 731 {
 732     int i;
 733     for (i=0; syntax[i].id; i++)
 734         if (id == syntax[i].id)
 735             break;
 736     if (!syntax[i].id && id == MATROSKA_ID_CLUSTER &&
 737         matroska->num_levels > 0 &&
 738         matroska->levels[matroska->num_levels-1].length == 0xffffffffffffff)
 739         return 0;  // we reached the end of an unknown size cluster
 740     if (!syntax[i].id && id != EBML_ID_VOID && id != EBML_ID_CRC32)
 741         av_log(matroska->ctx, AV_LOG_INFO, "Unknown entry 0x%X\n", id);
 742     return ebml_parse_elem(matroska, &syntax[i], data);
 743 }
 744
 745 static int ebml_parse(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
 746                       void *data)
 747 {
 748     if (!matroska->current_id) {
 749         uint64_t id;
 750         int res = ebml_read_num(matroska, matroska->ctx->pb, 4, &id);
 751         if (res < 0)
 752             return res;
 753         matroska->current_id = id | 1 << 7*res;
 754     }
 755     return ebml_parse_id(matroska, syntax, matroska->current_id, data);
 756 }
 757
 758 static int ebml_parse_nest(MatroskaDemuxContext *matroska, EbmlSyntax *syntax,
 759                            void *data)
 760 {
 761     int i, res = 0;
 762
 763     for (i=0; syntax[i].id; i++)
 764         switch (syntax[i].type) {
 765         case EBML_UINT:
 766             *(uint64_t *)((char *)data+syntax[i].data_offset) = syntax[i].def.u;
 767             break;
 768         case EBML_FLOAT:
 769             *(double   *)((char *)data+syntax[i].data_offset) = syntax[i].def.f;
 770             break;
 771         case EBML_STR:
 772         case EBML_UTF8:
 773             *(char    **)((char *)data+syntax[i].data_offset) = av_strdup(syntax[i].def.s);
 774             break;
 775         }
 776
 777     while (!res && !ebml_level_end(matroska))
 778         res = ebml_parse(matroska, syntax, data);
 779
 780     return res;
 781 }
 782
 783 static int ebml_parse_elem(MatroskaDemuxContext *matroska,
 784                            EbmlSyntax *syntax, void *data)
 785 {
 786     static const uint64_t max_lengths[EBML_TYPE_COUNT] = {
 787         [EBML_UINT]  = 8,
 788         [EBML_FLOAT] = 8,
 789         // max. 16 MB for strings
 790         [EBML_STR]   = 0x1000000,
 791         [EBML_UTF8]  = 0x1000000,
 792         // max. 256 MB for binary data
 793         [EBML_BIN]   = 0x10000000,
 794         // no limits for anything else
 795     };
 796     AVIOContext *pb = matroska->ctx->pb;
 797     uint32_t id = syntax->id;
 798     uint64_t length;
 799     int res;
 800
 801     data = (char *)data + syntax->data_offset;
 802     if (syntax->list_elem_size) {
 803         EbmlList *list = data;
 804         list->elem = av_realloc(list->elem, (list->nb_elem+1)*syntax->list_elem_size);
 805         data = (char*)list->elem + list->nb_elem*syntax->list_elem_size;
 806         memset(data, 0, syntax->list_elem_size);
 807         list->nb_elem++;
 808     }
 809
 810     if (syntax->type != EBML_PASS && syntax->type != EBML_STOP) {
 811         matroska->current_id = 0;
 812         if ((res = ebml_read_length(matroska, pb, &length)) < 0)
 813             return res;
 814         if (max_lengths[syntax->type] && length > max_lengths[syntax->type]) {
 815             av_log(matroska->ctx, AV_LOG_ERROR,
 816                    "Invalid length 0x%"PRIx64" > 0x%"PRIx64" for syntax element %i\n",
 817                    length, max_lengths[syntax->type], syntax->type);
 818             return AVERROR_INVALIDDATA;
 819         }
 820     }
 821
 822     switch (syntax->type) {
 823     case EBML_UINT:  res = ebml_read_uint  (pb, length, data);  break;
 824     case EBML_FLOAT: res = ebml_read_float (pb, length, data);  break;
 825     case EBML_STR:
 826     case EBML_UTF8:  res = ebml_read_ascii (pb, length, data);  break;
 827     case EBML_BIN:   res = ebml_read_binary(pb, length, data);  break;
 828     case EBML_NEST:  if ((res=ebml_read_master(matroska, length)) < 0)
 829                          return res;
 830                      if (id == MATROSKA_ID_SEGMENT)
 831                          matroska->segment_start = avio_tell(matroska->ctx->pb);
 832                      return ebml_parse_nest(matroska, syntax->def.n, data);
 833     case EBML_PASS:  return ebml_parse_id(matroska, syntax->def.n, id, data);
 834     case EBML_STOP:  return 1;
 835     default:         return avio_skip(pb,length)<0 ? AVERROR(EIO) : 0;
 836     }
 837     if (res == AVERROR_INVALIDDATA)
 838         av_log(matroska->ctx, AV_LOG_ERROR, "Invalid element\n");
 839     else if (res == AVERROR(EIO))
 840         av_log(matroska->ctx, AV_LOG_ERROR, "Read error\n");
 841     return res;
 842 }
 843
 844 static void ebml_free(EbmlSyntax *syntax, void *data)
 845 {
 846     int i, j;
 847     for (i=0; syntax[i].id; i++) {
 848         void *data_off = (char *)data + syntax[i].data_offset;
 849         switch (syntax[i].type) {
 850         case EBML_STR:
 851         case EBML_UTF8:  av_freep(data_off);                      break;
 852         case EBML_BIN:   av_freep(&((EbmlBin *)data_off)->data);  break;
 853         case EBML_NEST:
 854             if (syntax[i].list_elem_size) {
 855                 EbmlList *list = data_off;
 856                 char *ptr = list->elem;
 857                 for (j=0; j<list->nb_elem; j++, ptr+=syntax[i].list_elem_size)
 858                     ebml_free(syntax[i].def.n, ptr);
 859                 av_free(list->elem);
 860             } else
 861                 ebml_free(syntax[i].def.n, data_off);
 862         default:  break;
 863         }
 864     }
 865 }
 866
 867
 868 /*
 869  * Autodetecting...
 870  */
 871 static int matroska_probe(AVProbeData *p)
 872 {
 873     uint64_t total = 0;
 874     int len_mask = 0x80, size = 1, n = 1, i;
 875
 876     /* EBML header? */
 877     if (AV_RB32(p->buf) != EBML_ID_HEADER)
 878         return 0;
 879
 880     /* length of header */
 881     total = p->buf[4];
 882     while (size <= 8 && !(total & len_mask)) {
 883         size++;
 884         len_mask >>= 1;
 885     }
 886     if (size > 8)
 887       return 0;
 888     total &= (len_mask - 1);
 889     while (n < size)
 890         total = (total << 8) | p->buf[4 + n++];
 891
 892     /* Does the probe data contain the whole header? */
 893     if (p->buf_size < 4 + size + total)
 894       return 0;
 895
 896     /* The header should contain a known document type. For now,
 897      * we don't parse the whole header but simply check for the
 898      * availability of that array of characters inside the header.
 899      * Not fully fool-proof, but good enough. */
 900     for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++) {
 901         int probelen = strlen(matroska_doctypes[i]);
 902         for (n = 4+size; n <= 4+size+total-probelen; n++)
 903             if (!memcmp(p->buf+n, matroska_doctypes[i], probelen))
 904                 return AVPROBE_SCORE_MAX;
 905     }
 906
 907     // probably valid EBML header but no recognized doctype
 908     return AVPROBE_SCORE_MAX/2;
 909 }
 910
 911 static MatroskaTrack *matroska_find_track_by_num(MatroskaDemuxContext *matroska,
 912                                                  int num)
 913 {
 914     MatroskaTrack *tracks = matroska->tracks.elem;
 915     int i;
 916
 917     for (i=0; i < matroska->tracks.nb_elem; i++)
 918         if (tracks[i].num == num)
 919             return &tracks[i];
 920
 921     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid track number %d\n", num);
 922     return NULL;
 923 }
 924
 925 static int matroska_decode_buffer(uint8_t** buf, int* buf_size,
 926                                   MatroskaTrack *track)
 927 {
 928     MatroskaTrackEncoding *encodings = track->encodings.elem;
 929     uint8_t* data = *buf;
 930     int isize = *buf_size;
 931     uint8_t* pkt_data = NULL;
 932     int pkt_size = isize;
 933     int result = 0;
 934     int olen;
 935
 936     if (pkt_size >= 10000000)
 937         return -1;
 938
 939     switch (encodings[0].compression.algo) {
 940     case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
 941         return encodings[0].compression.settings.size;
 942     case MATROSKA_TRACK_ENCODING_COMP_LZO:
 943         do {
 944             olen = pkt_size *= 3;
 945             pkt_data = av_realloc(pkt_data, pkt_size+AV_LZO_OUTPUT_PADDING);
 946             result = av_lzo1x_decode(pkt_data, &olen, data, &isize);
 947         } while (result==AV_LZO_OUTPUT_FULL && pkt_size<10000000);
 948         if (result)
 949             goto failed;
 950         pkt_size -= olen;
 951         break;
 952 #if CONFIG_ZLIB
 953     case MATROSKA_TRACK_ENCODING_COMP_ZLIB: {
 954         z_stream zstream = {0};
 955         if (inflateInit(&zstream) != Z_OK)
 956             return -1;
 957         zstream.next_in = data;
 958         zstream.avail_in = isize;
 959         do {
 960             pkt_size *= 3;
 961             pkt_data = av_realloc(pkt_data, pkt_size);
 962             zstream.avail_out = pkt_size - zstream.total_out;
 963             zstream.next_out = pkt_data + zstream.total_out;
 964             result = inflate(&zstream, Z_NO_FLUSH);
 965         } while (result==Z_OK && pkt_size<10000000);
 966         pkt_size = zstream.total_out;
 967         inflateEnd(&zstream);
 968         if (result != Z_STREAM_END)
 969             goto failed;
 970         break;
 971     }
 972 #endif
 973 #if CONFIG_BZLIB
 974     case MATROSKA_TRACK_ENCODING_COMP_BZLIB: {
 975         bz_stream bzstream = {0};
 976         if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
 977             return -1;
 978         bzstream.next_in = data;
 979         bzstream.avail_in = isize;
 980         do {
 981             pkt_size *= 3;
 982             pkt_data = av_realloc(pkt_data, pkt_size);
 983             bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
 984             bzstream.next_out = pkt_data + bzstream.total_out_lo32;
 985             result = BZ2_bzDecompress(&bzstream);
 986         } while (result==BZ_OK && pkt_size<10000000);
 987         pkt_size = bzstream.total_out_lo32;
 988         BZ2_bzDecompressEnd(&bzstream);
 989         if (result != BZ_STREAM_END)
 990             goto failed;
 991         break;
 992     }
 993 #endif
 994     default:
 995         return -1;
 996     }
 997
 998     *buf = pkt_data;
 999     *buf_size = pkt_size;
1000     return 0;
1001  failed:
1002     av_free(pkt_data);
1003     return -1;
1004 }
1005
1006 static void matroska_fix_ass_packet(MatroskaDemuxContext *matroska,
1007                                     AVPacket *pkt, uint64_t display_duration)
1008 {
1009     char *line, *layer, *ptr = pkt->data, *end = ptr+pkt->size;
1010     for (; *ptr!=',' && ptr<end-1; ptr++);
1011     if (*ptr == ',')
1012         layer = ++ptr;
1013     for (; *ptr!=',' && ptr<end-1; ptr++);
1014     if (*ptr == ',') {
1015         int64_t end_pts = pkt->pts + display_duration;
1016         int sc = matroska->time_scale * pkt->pts / 10000000;
1017         int ec = matroska->time_scale * end_pts  / 10000000;
1018         int sh, sm, ss, eh, em, es, len;
1019         sh = sc/360000;  sc -= 360000*sh;
1020         sm = sc/  6000;  sc -=   6000*sm;
1021         ss = sc/   100;  sc -=    100*ss;
1022         eh = ec/360000;  ec -= 360000*eh;
1023         em = ec/  6000;  ec -=   6000*em;
1024         es = ec/   100;  ec -=    100*es;
1025         *ptr++ = '\0';
1026         len = 50 + end-ptr + FF_INPUT_BUFFER_PADDING_SIZE;
1027         if (!(line = av_malloc(len)))
1028             return;
1029         snprintf(line,len,"Dialogue: %s,%d:%02d:%02d.%02d,%d:%02d:%02d.%02d,%s\r\n",
1030                  layer, sh, sm, ss, sc, eh, em, es, ec, ptr);
1031         av_free(pkt->data);
1032         pkt->data = line;
1033         pkt->size = strlen(line);
1034     }
1035 }
1036
1037 static void matroska_merge_packets(AVPacket *out, AVPacket *in)
1038 {
1039     out->data = av_realloc(out->data, out->size+in->size);
1040     memcpy(out->data+out->size, in->data, in->size);
1041     out->size += in->size;
1042     av_destruct_packet(in);
1043     av_free(in);
1044 }
1045
1046 static void matroska_convert_tag(AVFormatContext *s, EbmlList *list,
1047                                  AVMetadata **metadata, char *prefix)
1048 {
1049     MatroskaTag *tags = list->elem;
1050     char key[1024];
1051     int i;
1052
1053     for (i=0; i < list->nb_elem; i++) {
1054         const char *lang = strcmp(tags[i].lang, "und") ? tags[i].lang : NULL;
1055
1056         if (!tags[i].name) {
1057             av_log(s, AV_LOG_WARNING, "Skipping invalid tag with no TagName.\n");
1058             continue;
1059         }
1060         if (prefix)  snprintf(key, sizeof(key), "%s/%s", prefix, tags[i].name);
1061         else         av_strlcpy(key, tags[i].name, sizeof(key));
1062         if (tags[i].def || !lang) {
1063         av_metadata_set2(metadata, key, tags[i].string, 0);
1064         if (tags[i].sub.nb_elem)
1065             matroska_convert_tag(s, &tags[i].sub, metadata, key);
1066         }
1067         if (lang) {
1068             av_strlcat(key, "-", sizeof(key));
1069             av_strlcat(key, lang, sizeof(key));
1070             av_metadata_set2(metadata, key, tags[i].string, 0);
1071             if (tags[i].sub.nb_elem)
1072                 matroska_convert_tag(s, &tags[i].sub, metadata, key);
1073         }
1074     }
1075     ff_metadata_conv(metadata, NULL, ff_mkv_metadata_conv);
1076 }
1077
1078 static void matroska_convert_tags(AVFormatContext *s)
1079 {
1080     MatroskaDemuxContext *matroska = s->priv_data;
1081     MatroskaTags *tags = matroska->tags.elem;
1082     int i, j;
1083
1084     for (i=0; i < matroska->tags.nb_elem; i++) {
1085         if (tags[i].target.attachuid) {
1086             MatroskaAttachement *attachment = matroska->attachments.elem;
1087             for (j=0; j<matroska->attachments.nb_elem; j++)
1088                 if (attachment[j].uid == tags[i].target.attachuid)
1089                     matroska_convert_tag(s, &tags[i].tag,
1090                                          &attachment[j].stream->metadata, NULL);
1091         } else if (tags[i].target.chapteruid) {
1092             MatroskaChapter *chapter = matroska->chapters.elem;
1093             for (j=0; j<matroska->chapters.nb_elem; j++)
1094                 if (chapter[j].uid == tags[i].target.chapteruid)
1095                     matroska_convert_tag(s, &tags[i].tag,
1096                                          &chapter[j].chapter->metadata, NULL);
1097         } else if (tags[i].target.trackuid) {
1098             MatroskaTrack *track = matroska->tracks.elem;
1099             for (j=0; j<matroska->tracks.nb_elem; j++)
1100                 if (track[j].uid == tags[i].target.trackuid)
1101                     matroska_convert_tag(s, &tags[i].tag,
1102                                          &track[j].stream->metadata, NULL);
1103         } else {
1104             matroska_convert_tag(s, &tags[i].tag, &s->metadata,
1105                                  tags[i].target.type);
1106         }
1107     }
1108 }
1109
1110 static void matroska_execute_seekhead(MatroskaDemuxContext *matroska)
1111 {
1112     EbmlList *seekhead_list = &matroska->seekhead;
1113     MatroskaSeekhead *seekhead = seekhead_list->elem;
1114     uint32_t level_up = matroska->level_up;
1115     int64_t before_pos = avio_tell(matroska->ctx->pb);
1116     uint32_t saved_id = matroska->current_id;
1117     MatroskaLevel level;
1118     int i;
1119
1120     // we should not do any seeking in the streaming case
1121     if (url_is_streamed(matroska->ctx->pb) ||
1122         (matroska->ctx->flags & AVFMT_FLAG_IGNIDX))
1123         return;
1124
1125     for (i=0; i<seekhead_list->nb_elem; i++) {
1126         int64_t offset = seekhead[i].pos + matroska->segment_start;
1127
1128         if (seekhead[i].pos <= before_pos
1129             || seekhead[i].id == MATROSKA_ID_SEEKHEAD
1130             || seekhead[i].id == MATROSKA_ID_CLUSTER)
1131             continue;
1132
1133         /* seek */
1134         if (avio_seek(matroska->ctx->pb, offset, SEEK_SET) != offset)
1135             continue;
1136
1137         /* We don't want to lose our seekhead level, so we add
1138          * a dummy. This is a crude hack. */
1139         if (matroska->num_levels == EBML_MAX_DEPTH) {
1140             av_log(matroska->ctx, AV_LOG_INFO,
1141                    "Max EBML element depth (%d) reached, "
1142                    "cannot parse further.\n", EBML_MAX_DEPTH);
1143             break;
1144         }
1145
1146         level.start = 0;
1147         level.length = (uint64_t)-1;
1148         matroska->levels[matroska->num_levels] = level;
1149         matroska->num_levels++;
1150         matroska->current_id = 0;
1151
1152         ebml_parse(matroska, matroska_segment, matroska);
1153
1154         /* remove dummy level */
1155         while (matroska->num_levels) {
1156             uint64_t length = matroska->levels[--matroska->num_levels].length;
1157             if (length == (uint64_t)-1)
1158                 break;
1159         }
1160     }
1161
1162     /* seek back */
1163     avio_seek(matroska->ctx->pb, before_pos, SEEK_SET);
1164     matroska->level_up = level_up;
1165     matroska->current_id = saved_id;
1166 }
1167
1168 static int matroska_aac_profile(char *codec_id)
1169 {
1170     static const char * const aac_profiles[] = { "MAIN", "LC", "SSR" };
1171     int profile;
1172
1173     for (profile=0; profile<FF_ARRAY_ELEMS(aac_profiles); profile++)
1174         if (strstr(codec_id, aac_profiles[profile]))
1175             break;
1176     return profile + 1;
1177 }
1178
1179 static int matroska_aac_sri(int samplerate)
1180 {
1181     int sri;
1182
1183     for (sri=0; sri<FF_ARRAY_ELEMS(ff_mpeg4audio_sample_rates); sri++)
1184         if (ff_mpeg4audio_sample_rates[sri] == samplerate)
1185             break;
1186     return sri;
1187 }
1188
1189 static int matroska_read_header(AVFormatContext *s, AVFormatParameters *ap)
1190 {
1191     MatroskaDemuxContext *matroska = s->priv_data;
1192     EbmlList *attachements_list = &matroska->attachments;
1193     MatroskaAttachement *attachements;
1194     EbmlList *chapters_list = &matroska->chapters;
1195     MatroskaChapter *chapters;
1196     MatroskaTrack *tracks;
1197     EbmlList *index_list;
1198     MatroskaIndex *index;
1199     int index_scale = 1;
1200     uint64_t max_start = 0;
1201     Ebml ebml = { 0 };
1202     AVStream *st;
1203     int i, j, res;
1204
1205     matroska->ctx = s;
1206
1207     /* First read the EBML header. */
1208     if (ebml_parse(matroska, ebml_syntax, &ebml)
1209         || ebml.version > EBML_VERSION       || ebml.max_size > sizeof(uint64_t)
1210         || ebml.id_length > sizeof(uint32_t) || ebml.doctype_version > 2) {
1211         av_log(matroska->ctx, AV_LOG_ERROR,
1212                "EBML header using unsupported features\n"
1213                "(EBML version %"PRIu64", doctype %s, doc version %"PRIu64")\n",
1214                ebml.version, ebml.doctype, ebml.doctype_version);
1215         ebml_free(ebml_syntax, &ebml);
1216         return AVERROR_PATCHWELCOME;
1217     }
1218     for (i = 0; i < FF_ARRAY_ELEMS(matroska_doctypes); i++)
1219         if (!strcmp(ebml.doctype, matroska_doctypes[i]))
1220             break;
1221     if (i >= FF_ARRAY_ELEMS(matroska_doctypes)) {
1222         av_log(s, AV_LOG_WARNING, "Unknown EBML doctype '%s'\n", ebml.doctype);
1223     }
1224     ebml_free(ebml_syntax, &ebml);
1225
1226     /* The next thing is a segment. */
1227     if ((res = ebml_parse(matroska, matroska_segments, matroska)) < 0)
1228         return res;
1229     matroska_execute_seekhead(matroska);
1230
1231     if (!matroska->time_scale)
1232         matroska->time_scale = 1000000;
1233     if (matroska->duration)
1234         matroska->ctx->duration = matroska->duration * matroska->time_scale
1235                                   * 1000 / AV_TIME_BASE;
1236     av_metadata_set2(&s->metadata, "title", matroska->title, 0);
1237
1238     tracks = matroska->tracks.elem;
1239     for (i=0; i < matroska->tracks.nb_elem; i++) {
1240         MatroskaTrack *track = &tracks[i];
1241         enum CodecID codec_id = CODEC_ID_NONE;
1242         EbmlList *encodings_list = &tracks->encodings;
1243         MatroskaTrackEncoding *encodings = encodings_list->elem;
1244         uint8_t *extradata = NULL;
1245         int extradata_size = 0;
1246         int extradata_offset = 0;
1247         AVIOContext b;
1248
1249         /* Apply some sanity checks. */
1250         if (track->type != MATROSKA_TRACK_TYPE_VIDEO &&
1251             track->type != MATROSKA_TRACK_TYPE_AUDIO &&
1252             track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1253             av_log(matroska->ctx, AV_LOG_INFO,
1254                    "Unknown or unsupported track type %"PRIu64"\n",
1255                    track->type);
1256             continue;
1257         }
1258         if (track->codec_id == NULL)
1259             continue;
1260
1261         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
1262             if (!track->default_duration)
1263                 track->default_duration = 1000000000/track->video.frame_rate;
1264             if (!track->video.display_width)
1265                 track->video.display_width = track->video.pixel_width;
1266             if (!track->video.display_height)
1267                 track->video.display_height = track->video.pixel_height;
1268         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1269             if (!track->audio.out_samplerate)
1270                 track->audio.out_samplerate = track->audio.samplerate;
1271         }
1272         if (encodings_list->nb_elem > 1) {
1273             av_log(matroska->ctx, AV_LOG_ERROR,
1274                    "Multiple combined encodings no supported");
1275         } else if (encodings_list->nb_elem == 1) {
1276             if (encodings[0].type ||
1277                 (encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP &&
1278 #if CONFIG_ZLIB
1279                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
1280 #endif
1281 #if CONFIG_BZLIB
1282                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
1283 #endif
1284                  encodings[0].compression.algo != MATROSKA_TRACK_ENCODING_COMP_LZO)) {
1285                 encodings[0].scope = 0;
1286                 av_log(matroska->ctx, AV_LOG_ERROR,
1287                        "Unsupported encoding type");
1288             } else if (track->codec_priv.size && encodings[0].scope&2) {
1289                 uint8_t *codec_priv = track->codec_priv.data;
1290                 int offset = matroska_decode_buffer(&track->codec_priv.data,
1291                                                     &track->codec_priv.size,
1292                                                     track);
1293                 if (offset < 0) {
1294                     track->codec_priv.data = NULL;
1295                     track->codec_priv.size = 0;
1296                     av_log(matroska->ctx, AV_LOG_ERROR,
1297                            "Failed to decode codec private data\n");
1298                 } else if (offset > 0) {
1299                     track->codec_priv.data = av_malloc(track->codec_priv.size + offset);
1300                     memcpy(track->codec_priv.data,
1301                            encodings[0].compression.settings.data, offset);
1302                     memcpy(track->codec_priv.data+offset, codec_priv,
1303                            track->codec_priv.size);
1304                     track->codec_priv.size += offset;
1305                 }
1306                 if (codec_priv != track->codec_priv.data)
1307                     av_free(codec_priv);
1308             }
1309         }
1310
1311         for(j=0; ff_mkv_codec_tags[j].id != CODEC_ID_NONE; j++){
1312             if(!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
1313                         strlen(ff_mkv_codec_tags[j].str))){
1314                 codec_id= ff_mkv_codec_tags[j].id;
1315                 break;
1316             }
1317         }
1318
1319         st = track->stream = av_new_stream(s, 0);
1320         if (st == NULL)
1321             return AVERROR(ENOMEM);
1322
1323         if (!strcmp(track->codec_id, "V_MS/VFW/FOURCC")
1324             && track->codec_priv.size >= 40
1325             && track->codec_priv.data != NULL) {
1326             track->ms_compat = 1;
1327             track->video.fourcc = AV_RL32(track->codec_priv.data + 16);
1328             codec_id = ff_codec_get_id(ff_codec_bmp_tags, track->video.fourcc);
1329             extradata_offset = 40;
1330         } else if (!strcmp(track->codec_id, "A_MS/ACM")
1331                    && track->codec_priv.size >= 14
1332                    && track->codec_priv.data != NULL) {
1333             ffio_init_context(&b, track->codec_priv.data, track->codec_priv.size,
1334                           URL_RDONLY, NULL, NULL, NULL, NULL);
1335             ff_get_wav_header(&b, st->codec, track->codec_priv.size);
1336             codec_id = st->codec->codec_id;
1337             extradata_offset = FFMIN(track->codec_priv.size, 18);
1338         } else if (!strcmp(track->codec_id, "V_QUICKTIME")
1339                    && (track->codec_priv.size >= 86)
1340                    && (track->codec_priv.data != NULL)) {
1341             track->video.fourcc = AV_RL32(track->codec_priv.data);
1342             codec_id=ff_codec_get_id(codec_movvideo_tags, track->video.fourcc);
1343         } else if (codec_id == CODEC_ID_PCM_S16BE) {
1344             switch (track->audio.bitdepth) {
1345             case  8:  codec_id = CODEC_ID_PCM_U8;     break;
1346             case 24:  codec_id = CODEC_ID_PCM_S24BE;  break;
1347             case 32:  codec_id = CODEC_ID_PCM_S32BE;  break;
1348             }
1349         } else if (codec_id == CODEC_ID_PCM_S16LE) {
1350             switch (track->audio.bitdepth) {
1351             case  8:  codec_id = CODEC_ID_PCM_U8;     break;
1352             case 24:  codec_id = CODEC_ID_PCM_S24LE;  break;
1353             case 32:  codec_id = CODEC_ID_PCM_S32LE;  break;
1354             }
1355         } else if (codec_id==CODEC_ID_PCM_F32LE && track->audio.bitdepth==64) {
1356             codec_id = CODEC_ID_PCM_F64LE;
1357         } else if (codec_id == CODEC_ID_AAC && !track->codec_priv.size) {
1358             int profile = matroska_aac_profile(track->codec_id);
1359             int sri = matroska_aac_sri(track->audio.samplerate);
1360             extradata = av_malloc(5);
1361             if (extradata == NULL)
1362                 return AVERROR(ENOMEM);
1363             extradata[0] = (profile << 3) | ((sri&0x0E) >> 1);
1364             extradata[1] = ((sri&0x01) << 7) | (track->audio.channels<<3);
1365             if (strstr(track->codec_id, "SBR")) {
1366                 sri = matroska_aac_sri(track->audio.out_samplerate);
1367                 extradata[2] = 0x56;
1368                 extradata[3] = 0xE5;
1369                 extradata[4] = 0x80 | (sri<<3);
1370                 extradata_size = 5;
1371             } else
1372                 extradata_size = 2;
1373         } else if (codec_id == CODEC_ID_TTA) {
1374             extradata_size = 30;
1375             extradata = av_mallocz(extradata_size);
1376             if (extradata == NULL)
1377                 return AVERROR(ENOMEM);
1378             ffio_init_context(&b, extradata, extradata_size, 1,
1379                           NULL, NULL, NULL, NULL);
1380             avio_write(&b, "TTA1", 4);
1381             avio_wl16(&b, 1);
1382             avio_wl16(&b, track->audio.channels);
1383             avio_wl16(&b, track->audio.bitdepth);
1384             avio_wl32(&b, track->audio.out_samplerate);
1385             avio_wl32(&b, matroska->ctx->duration * track->audio.out_samplerate);
1386         } else if (codec_id == CODEC_ID_RV10 || codec_id == CODEC_ID_RV20 ||
1387                    codec_id == CODEC_ID_RV30 || codec_id == CODEC_ID_RV40) {
1388             extradata_offset = 26;
1389         } else if (codec_id == CODEC_ID_RA_144) {
1390             track->audio.out_samplerate = 8000;
1391             track->audio.channels = 1;
1392         } else if (codec_id == CODEC_ID_RA_288 || codec_id == CODEC_ID_COOK ||
1393                    codec_id == CODEC_ID_ATRAC3 || codec_id == CODEC_ID_SIPR) {
1394             int flavor;
1395             ffio_init_context(&b, track->codec_priv.data,track->codec_priv.size,
1396                           0, NULL, NULL, NULL, NULL);
1397             avio_skip(&b, 22);
1398             flavor                       = avio_rb16(&b);
1399             track->audio.coded_framesize = avio_rb32(&b);
1400             avio_skip(&b, 12);
1401             track->audio.sub_packet_h    = avio_rb16(&b);
1402             track->audio.frame_size      = avio_rb16(&b);
1403             track->audio.sub_packet_size = avio_rb16(&b);
1404             track->audio.buf = av_malloc(track->audio.frame_size * track->audio.sub_packet_h);
1405             if (codec_id == CODEC_ID_RA_288) {
1406                 st->codec->block_align = track->audio.coded_framesize;
1407                 track->codec_priv.size = 0;
1408             } else {
1409                 if (codec_id == CODEC_ID_SIPR && flavor < 4) {
1410                     const int sipr_bit_rate[4] = { 6504, 8496, 5000, 16000 };
1411                     track->audio.sub_packet_size = ff_sipr_subpk_size[flavor];
1412                     st->codec->bit_rate = sipr_bit_rate[flavor];
1413                 }
1414                 st->codec->block_align = track->audio.sub_packet_size;
1415                 extradata_offset = 78;
1416             }
1417         }
1418         track->codec_priv.size -= extradata_offset;
1419
1420         if (codec_id == CODEC_ID_NONE)
1421             av_log(matroska->ctx, AV_LOG_INFO,
1422                    "Unknown/unsupported CodecID %s.\n", track->codec_id);
1423
1424         if (track->time_scale < 0.01)
1425             track->time_scale = 1.0;
1426         av_set_pts_info(st, 64, matroska->time_scale*track->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
1427
1428         st->codec->codec_id = codec_id;
1429         st->start_time = 0;
1430         if (strcmp(track->language, "und"))
1431             av_metadata_set2(&st->metadata, "language", track->language, 0);
1432         av_metadata_set2(&st->metadata, "title", track->name, 0);
1433
1434         if (track->flag_default)
1435             st->disposition |= AV_DISPOSITION_DEFAULT;
1436         if (track->flag_forced)
1437             st->disposition |= AV_DISPOSITION_FORCED;
1438
1439         if (track->default_duration)
1440             av_reduce(&st->codec->time_base.num, &st->codec->time_base.den,
1441                       track->default_duration, 1000000000, 30000);
1442
1443         if (!st->codec->extradata) {
1444             if(extradata){
1445                 st->codec->extradata = extradata;
1446                 st->codec->extradata_size = extradata_size;
1447             } else if(track->codec_priv.data && track->codec_priv.size > 0){
1448                 st->codec->extradata = av_mallocz(track->codec_priv.size +
1449                                                   FF_INPUT_BUFFER_PADDING_SIZE);
1450                 if(st->codec->extradata == NULL)
1451                     return AVERROR(ENOMEM);
1452                 st->codec->extradata_size = track->codec_priv.size;
1453                 memcpy(st->codec->extradata,
1454                        track->codec_priv.data + extradata_offset,
1455                        track->codec_priv.size);
1456             }
1457         }
1458
1459         if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
1460             st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
1461             st->codec->codec_tag  = track->video.fourcc;
1462             st->codec->width  = track->video.pixel_width;
1463             st->codec->height = track->video.pixel_height;
1464             av_reduce(&st->sample_aspect_ratio.num,
1465                       &st->sample_aspect_ratio.den,
1466                       st->codec->height * track->video.display_width,
1467                       st->codec-> width * track->video.display_height,
1468                       255);
1469             if (st->codec->codec_id != CODEC_ID_H264)
1470             st->need_parsing = AVSTREAM_PARSE_HEADERS;
1471             if (track->default_duration)
1472                 st->avg_frame_rate = av_d2q(1000000000.0/track->default_duration, INT_MAX);
1473         } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
1474             st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
1475             st->codec->sample_rate = track->audio.out_samplerate;
1476             st->codec->channels = track->audio.channels;
1477             if (st->codec->codec_id != CODEC_ID_AAC)
1478             st->need_parsing = AVSTREAM_PARSE_HEADERS;
1479         } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
1480             st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
1481         }
1482     }
1483
1484     attachements = attachements_list->elem;
1485     for (j=0; j<attachements_list->nb_elem; j++) {
1486         if (!(attachements[j].filename && attachements[j].mime &&
1487               attachements[j].bin.data && attachements[j].bin.size > 0)) {
1488             av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
1489         } else {
1490             AVStream *st = av_new_stream(s, 0);
1491             if (st == NULL)
1492                 break;
1493             av_metadata_set2(&st->metadata, "filename",attachements[j].filename, 0);
1494             st->codec->codec_id = CODEC_ID_NONE;
1495             st->codec->codec_type = AVMEDIA_TYPE_ATTACHMENT;
1496             st->codec->extradata  = av_malloc(attachements[j].bin.size);
1497             if(st->codec->extradata == NULL)
1498                 break;
1499             st->codec->extradata_size = attachements[j].bin.size;
1500             memcpy(st->codec->extradata, attachements[j].bin.data, attachements[j].bin.size);
1501
1502             for (i=0; ff_mkv_mime_tags[i].id != CODEC_ID_NONE; i++) {
1503                 if (!strncmp(ff_mkv_mime_tags[i].str, attachements[j].mime,
1504                              strlen(ff_mkv_mime_tags[i].str))) {
1505                     st->codec->codec_id = ff_mkv_mime_tags[i].id;
1506                     break;
1507                 }
1508             }
1509             attachements[j].stream = st;
1510         }
1511     }
1512
1513     chapters = chapters_list->elem;
1514     for (i=0; i<chapters_list->nb_elem; i++)
1515         if (chapters[i].start != AV_NOPTS_VALUE && chapters[i].uid
1516             && (max_start==0 || chapters[i].start > max_start)) {
1517             chapters[i].chapter =
1518             ff_new_chapter(s, chapters[i].uid, (AVRational){1, 1000000000},
1519                            chapters[i].start, chapters[i].end,
1520                            chapters[i].title);
1521             av_metadata_set2(&chapters[i].chapter->metadata,
1522                              "title", chapters[i].title, 0);
1523             max_start = chapters[i].start;
1524         }
1525
1526     index_list = &matroska->index;
1527     index = index_list->elem;
1528     if (index_list->nb_elem
1529         && index[0].time > 100000000000000/matroska->time_scale) {
1530         av_log(matroska->ctx, AV_LOG_WARNING, "Working around broken index.\n");
1531         index_scale = matroska->time_scale;
1532     }
1533     for (i=0; i<index_list->nb_elem; i++) {
1534         EbmlList *pos_list = &index[i].pos;
1535         MatroskaIndexPos *pos = pos_list->elem;
1536         for (j=0; j<pos_list->nb_elem; j++) {
1537             MatroskaTrack *track = matroska_find_track_by_num(matroska,
1538                                                               pos[j].track);
1539             if (track && track->stream)
1540                 av_add_index_entry(track->stream,
1541                                    pos[j].pos + matroska->segment_start,
1542                                    index[i].time/index_scale, 0, 0,
1543                                    AVINDEX_KEYFRAME);
1544         }
1545     }
1546
1547     matroska_convert_tags(s);
1548
1549     return 0;
1550 }
1551
1552 /*
1553  * Put one packet in an application-supplied AVPacket struct.
1554  * Returns 0 on success or -1 on failure.
1555  */
1556 static int matroska_deliver_packet(MatroskaDemuxContext *matroska,
1557                                    AVPacket *pkt)
1558 {
1559     if (matroska->num_packets > 0) {
1560         memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
1561         av_free(matroska->packets[0]);
1562         if (matroska->num_packets > 1) {
1563             memmove(&matroska->packets[0], &matroska->packets[1],
1564                     (matroska->num_packets - 1) * sizeof(AVPacket *));
1565             matroska->packets =
1566                 av_realloc(matroska->packets, (matroska->num_packets - 1) *
1567                            sizeof(AVPacket *));
1568         } else {
1569             av_freep(&matroska->packets);
1570         }
1571         matroska->num_packets--;
1572         return 0;
1573     }
1574
1575     return -1;
1576 }
1577
1578 /*
1579  * Free all packets in our internal queue.
1580  */
1581 static void matroska_clear_queue(MatroskaDemuxContext *matroska)
1582 {
1583     if (matroska->packets) {
1584         int n;
1585         for (n = 0; n < matroska->num_packets; n++) {
1586             av_free_packet(matroska->packets[n]);
1587             av_free(matroska->packets[n]);
1588         }
1589         av_freep(&matroska->packets);
1590         matroska->num_packets = 0;
1591     }
1592 }
1593
1594 static int matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data,
1595                                 int size, int64_t pos, uint64_t cluster_time,
1596                                 uint64_t duration, int is_keyframe,
1597                                 int64_t cluster_pos)
1598 {
1599     uint64_t timecode = AV_NOPTS_VALUE;
1600     MatroskaTrack *track;
1601     int res = 0;
1602     AVStream *st;
1603     AVPacket *pkt;
1604     int16_t block_time;
1605     uint32_t *lace_size = NULL;
1606     int n, flags, laces = 0;
1607     uint64_t num;
1608
1609     if ((n = matroska_ebmlnum_uint(matroska, data, size, &num)) < 0) {
1610         av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
1611         return res;
1612     }
1613     data += n;
1614     size -= n;
1615
1616     track = matroska_find_track_by_num(matroska, num);
1617     if (size <= 3 || !track || !track->stream) {
1618         av_log(matroska->ctx, AV_LOG_INFO,
1619                "Invalid stream %"PRIu64" or size %u\n", num, size);
1620         return res;
1621     }
1622     st = track->stream;
1623     if (st->discard >= AVDISCARD_ALL)
1624         return res;
1625     if (!duration)
1626         duration = track->default_duration / matroska->time_scale;
1627
1628     block_time = AV_RB16(data);
1629     data += 2;
1630     flags = *data++;
1631     size -= 3;
1632     if (is_keyframe == -1)
1633         is_keyframe = flags & 0x80 ? AV_PKT_FLAG_KEY : 0;
1634
1635     if (cluster_time != (uint64_t)-1
1636         && (block_time >= 0 || cluster_time >= -block_time)) {
1637         timecode = cluster_time + block_time;
1638         if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE
1639             && timecode < track->end_timecode)
1640             is_keyframe = 0;  /* overlapping subtitles are not key frame */
1641         if (is_keyframe)
1642             av_add_index_entry(st, cluster_pos, timecode, 0,0,AVINDEX_KEYFRAME);
1643         track->end_timecode = FFMAX(track->end_timecode, timecode+duration);
1644     }
1645
1646     if (matroska->skip_to_keyframe && track->type != MATROSKA_TRACK_TYPE_SUBTITLE) {
1647         if (!is_keyframe || timecode < matroska->skip_to_timecode)
1648             return res;
1649         matroska->skip_to_keyframe = 0;
1650     }
1651
1652     switch ((flags & 0x06) >> 1) {
1653         case 0x0: /* no lacing */
1654             laces = 1;
1655             lace_size = av_mallocz(sizeof(int));
1656             lace_size[0] = size;
1657             break;
1658
1659         case 0x1: /* Xiph lacing */
1660         case 0x2: /* fixed-size lacing */
1661         case 0x3: /* EBML lacing */
1662             assert(size>0); // size <=3 is checked before size-=3 above
1663             laces = (*data) + 1;
1664             data += 1;
1665             size -= 1;
1666             lace_size = av_mallocz(laces * sizeof(int));
1667
1668             switch ((flags & 0x06) >> 1) {
1669                 case 0x1: /* Xiph lacing */ {
1670                     uint8_t temp;
1671                     uint32_t total = 0;
1672                     for (n = 0; res == 0 && n < laces - 1; n++) {
1673                         while (1) {
1674                             if (size == 0) {
1675                                 res = -1;
1676                                 break;
1677                             }
1678                             temp = *data;
1679                             lace_size[n] += temp;
1680                             data += 1;
1681                             size -= 1;
1682                             if (temp != 0xff)
1683                                 break;
1684                         }
1685                         total += lace_size[n];
1686                     }
1687                     lace_size[n] = size - total;
1688                     break;
1689                 }
1690
1691                 case 0x2: /* fixed-size lacing */
1692                     for (n = 0; n < laces; n++)
1693                         lace_size[n] = size / laces;
1694                     break;
1695
1696                 case 0x3: /* EBML lacing */ {
1697                     uint32_t total;
1698                     n = matroska_ebmlnum_uint(matroska, data, size, &num);
1699                     if (n < 0) {
1700                         av_log(matroska->ctx, AV_LOG_INFO,
1701                                "EBML block data error\n");
1702                         break;
1703                     }
1704                     data += n;
1705                     size -= n;
1706                     total = lace_size[0] = num;
1707                     for (n = 1; res == 0 && n < laces - 1; n++) {
1708                         int64_t snum;
1709                         int r;
1710                         r = matroska_ebmlnum_sint(matroska, data, size, &snum);
1711                         if (r < 0) {
1712                             av_log(matroska->ctx, AV_LOG_INFO,
1713                                    "EBML block data error\n");
1714                             break;
1715                         }
1716                         data += r;
1717                         size -= r;
1718                         lace_size[n] = lace_size[n - 1] + snum;
1719                         total += lace_size[n];
1720                     }
1721                     lace_size[n] = size - total;
1722                     break;
1723                 }
1724             }
1725             break;
1726     }
1727
1728     if (res == 0) {
1729         for (n = 0; n < laces; n++) {
1730             if ((st->codec->codec_id == CODEC_ID_RA_288 ||
1731                  st->codec->codec_id == CODEC_ID_COOK ||
1732                  st->codec->codec_id == CODEC_ID_SIPR ||
1733                  st->codec->codec_id == CODEC_ID_ATRAC3) &&
1734                  st->codec->block_align && track->audio.sub_packet_size) {
1735                 int a = st->codec->block_align;
1736                 int sps = track->audio.sub_packet_size;
1737                 int cfs = track->audio.coded_framesize;
1738                 int h = track->audio.sub_packet_h;
1739                 int y = track->audio.sub_packet_cnt;
1740                 int w = track->audio.frame_size;
1741                 int x;
1742
1743                 if (!track->audio.pkt_cnt) {
1744                     if (track->audio.sub_packet_cnt == 0)
1745                         track->audio.buf_timecode = timecode;
1746                     if (st->codec->codec_id == CODEC_ID_RA_288)
1747                         for (x=0; x<h/2; x++)
1748                             memcpy(track->audio.buf+x*2*w+y*cfs,
1749                                    data+x*cfs, cfs);
1750                     else if (st->codec->codec_id == CODEC_ID_SIPR)
1751                         memcpy(track->audio.buf + y*w, data, w);
1752                     else
1753                         for (x=0; x<w/sps; x++)
1754                             memcpy(track->audio.buf+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), data+x*sps, sps);
1755
1756                     if (++track->audio.sub_packet_cnt >= h) {
1757                         if (st->codec->codec_id == CODEC_ID_SIPR)
1758                             ff_rm_reorder_sipr_data(track->audio.buf, h, w);
1759                         track->audio.sub_packet_cnt = 0;
1760                         track->audio.pkt_cnt = h*w / a;
1761                     }
1762                 }
1763                 while (track->audio.pkt_cnt) {
1764                     pkt = av_mallocz(sizeof(AVPacket));
1765                     av_new_packet(pkt, a);
1766                     memcpy(pkt->data, track->audio.buf
1767                            + a * (h*w / a - track->audio.pkt_cnt--), a);
1768                     pkt->pts = track->audio.buf_timecode;
1769                     track->audio.buf_timecode = AV_NOPTS_VALUE;
1770                     pkt->pos = pos;
1771                     pkt->stream_index = st->index;
1772                     dynarray_add(&matroska->packets,&matroska->num_packets,pkt);
1773                 }
1774             } else {
1775                 MatroskaTrackEncoding *encodings = track->encodings.elem;
1776                 int offset = 0, pkt_size = lace_size[n];
1777                 uint8_t *pkt_data = data;
1778
1779                 if (pkt_size > size) {
1780                     av_log(matroska->ctx, AV_LOG_ERROR, "Invalid packet size\n");
1781                     break;
1782                 }
1783
1784                 if (encodings && encodings->scope & 1) {
1785                     offset = matroska_decode_buffer(&pkt_data,&pkt_size, track);
1786                     if (offset < 0)
1787                         continue;
1788                 }
1789
1790                 pkt = av_mallocz(sizeof(AVPacket));
1791                 /* XXX: prevent data copy... */
1792                 if (av_new_packet(pkt, pkt_size+offset) < 0) {
1793                     av_free(pkt);
1794                     res = AVERROR(ENOMEM);
1795                     break;
1796                 }
1797                 if (offset)
1798                     memcpy (pkt->data, encodings->compression.settings.data, offset);
1799                 memcpy (pkt->data+offset, pkt_data, pkt_size);
1800
1801                 if (pkt_data != data)
1802                     av_free(pkt_data);
1803
1804                 if (n == 0)
1805                     pkt->flags = is_keyframe;
1806                 pkt->stream_index = st->index;
1807
1808                 if (track->ms_compat)
1809                     pkt->dts = timecode;
1810                 else
1811                     pkt->pts = timecode;
1812                 pkt->pos = pos;
1813                 if (st->codec->codec_id == CODEC_ID_TEXT)
1814                     pkt->convergence_duration = duration;
1815                 else if (track->type != MATROSKA_TRACK_TYPE_SUBTITLE)
1816                     pkt->duration = duration;
1817
1818                 if (st->codec->codec_id == CODEC_ID_SSA)
1819                     matroska_fix_ass_packet(matroska, pkt, duration);
1820
1821                 if (matroska->prev_pkt &&
1822                     timecode != AV_NOPTS_VALUE &&
1823                     matroska->prev_pkt->pts == timecode &&
1824                     matroska->prev_pkt->stream_index == st->index &&
1825                     st->codec->codec_id == CODEC_ID_SSA)
1826                     matroska_merge_packets(matroska->prev_pkt, pkt);
1827                 else {
1828                     dynarray_add(&matroska->packets,&matroska->num_packets,pkt);
1829                     matroska->prev_pkt = pkt;
1830                 }
1831             }
1832
1833             if (timecode != AV_NOPTS_VALUE)
1834                 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
1835             data += lace_size[n];
1836             size -= lace_size[n];
1837         }
1838     }
1839
1840     av_free(lace_size);
1841     return res;
1842 }
1843
1844 static int matroska_parse_cluster(MatroskaDemuxContext *matroska)
1845 {
1846     MatroskaCluster cluster = { 0 };
1847     EbmlList *blocks_list;
1848     MatroskaBlock *blocks;
1849     int i, res;
1850     int64_t pos = avio_tell(matroska->ctx->pb);
1851     matroska->prev_pkt = NULL;
1852     if (matroska->current_id)
1853         pos -= 4;  /* sizeof the ID which was already read */
1854     res = ebml_parse(matroska, matroska_clusters, &cluster);
1855     blocks_list = &cluster.blocks;
1856     blocks = blocks_list->elem;
1857     for (i=0; i<blocks_list->nb_elem; i++)
1858         if (blocks[i].bin.size > 0 && blocks[i].bin.data) {
1859             int is_keyframe = blocks[i].non_simple ? !blocks[i].reference : -1;
1860             res=matroska_parse_block(matroska,
1861                                      blocks[i].bin.data, blocks[i].bin.size,
1862                                      blocks[i].bin.pos,  cluster.timecode,
1863                                      blocks[i].duration, is_keyframe,
1864                                      pos);
1865         }
1866     ebml_free(matroska_cluster, &cluster);
1867     if (res < 0)  matroska->done = 1;
1868     return res;
1869 }
1870
1871 static int matroska_read_packet(AVFormatContext *s, AVPacket *pkt)
1872 {
1873     MatroskaDemuxContext *matroska = s->priv_data;
1874
1875     while (matroska_deliver_packet(matroska, pkt)) {
1876         if (matroska->done)
1877             return AVERROR_EOF;
1878         matroska_parse_cluster(matroska);
1879     }
1880
1881     return 0;
1882 }
1883
1884 static int matroska_read_seek(AVFormatContext *s, int stream_index,
1885                               int64_t timestamp, int flags)
1886 {
1887     MatroskaDemuxContext *matroska = s->priv_data;
1888     MatroskaTrack *tracks = matroska->tracks.elem;
1889     AVStream *st = s->streams[stream_index];
1890     int i, index, index_sub, index_min;
1891
1892     if (!st->nb_index_entries)
1893         return 0;
1894     timestamp = FFMAX(timestamp, st->index_entries[0].timestamp);
1895
1896     if ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
1897         avio_seek(s->pb, st->index_entries[st->nb_index_entries-1].pos, SEEK_SET);
1898         while ((index = av_index_search_timestamp(st, timestamp, flags)) < 0) {
1899             matroska_clear_queue(matroska);
1900             if (matroska_parse_cluster(matroska) < 0)
1901                 break;
1902         }
1903     }
1904
1905     matroska_clear_queue(matroska);
1906     if (index < 0)
1907         return 0;
1908
1909     index_min = index;
1910     for (i=0; i < matroska->tracks.nb_elem; i++) {
1911         tracks[i].audio.pkt_cnt = 0;
1912         tracks[i].audio.sub_packet_cnt = 0;
1913         tracks[i].audio.buf_timecode = AV_NOPTS_VALUE;
1914         tracks[i].end_timecode = 0;
1915         if (tracks[i].type == MATROSKA_TRACK_TYPE_SUBTITLE
1916             && !tracks[i].stream->discard != AVDISCARD_ALL) {
1917             index_sub = av_index_search_timestamp(tracks[i].stream, st->index_entries[index].timestamp, AVSEEK_FLAG_BACKWARD);
1918             if (index_sub >= 0
1919                 && st->index_entries[index_sub].pos < st->index_entries[index_min].pos
1920                 && st->index_entries[index].timestamp - st->index_entries[index_sub].timestamp < 30000000000/matroska->time_scale)
1921                 index_min = index_sub;
1922         }
1923     }
1924
1925     avio_seek(s->pb, st->index_entries[index_min].pos, SEEK_SET);
1926     matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
1927     matroska->skip_to_timecode = st->index_entries[index].timestamp;
1928     matroska->done = 0;
1929     av_update_cur_dts(s, st, st->index_entries[index].timestamp);
1930     return 0;
1931 }
1932
1933 static int matroska_read_close(AVFormatContext *s)
1934 {
1935     MatroskaDemuxContext *matroska = s->priv_data;
1936     MatroskaTrack *tracks = matroska->tracks.elem;
1937     int n;
1938
1939     matroska_clear_queue(matroska);
1940
1941     for (n=0; n < matroska->tracks.nb_elem; n++)
1942         if (tracks[n].type == MATROSKA_TRACK_TYPE_AUDIO)
1943             av_free(tracks[n].audio.buf);
1944     ebml_free(matroska_segment, matroska);
1945
1946     return 0;
1947 }
1948
1949 AVInputFormat ff_matroska_demuxer = {
1950     "matroska,webm",
1951     NULL_IF_CONFIG_SMALL("Matroska/WebM file format"),
1952     sizeof(MatroskaDemuxContext),
1953     matroska_probe,
1954     matroska_read_header,
1955     matroska_read_packet,
1956     matroska_read_close,
1957     matroska_read_seek,
1958 };