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

   1 /*
   2  * Matroska file demuxer (no muxer yet)
   3  * Copyright (c) 2003-2004 The ffmpeg Project
   4  *
   5  * This file is part of FFmpeg.
   6  *
   7  * FFmpeg is free software; you can redistribute it and/or
   8  * modify it under the terms of the GNU Lesser General Public
   9  * License as published by the Free Software Foundation; either
  10  * version 2.1 of the License, or (at your option) any later version.
  11  *
  12  * FFmpeg is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  15  * Lesser General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Lesser General Public
  18  * License along with FFmpeg; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21
  22 /**
  23  * @file matroskadec.c
  24  * Matroska file demuxer
  25  * by Ronald Bultje <rbultje@ronald.bitfreak.net>
  26  * with a little help from Moritz Bunkus <moritz@bunkus.org>
  27  * Specs available on the matroska project page:
  28  * http://www.matroska.org/.
  29  */
  30
  31 #include "avformat.h"
  32 /* For codec_get_id(). */
  33 #include "riff.h"
  34 #include "intfloat_readwrite.h"
  35 #include "matroska.h"
  36
  37 typedef struct Track {
  38     MatroskaTrackType type;
  39
  40     /* Unique track number and track ID. stream_index is the index that
  41      * the calling app uses for this track. */
  42     uint32_t num;
  43     uint32_t uid;
  44     int stream_index;
  45
  46     char *name;
  47     char language[4];
  48
  49     char *codec_id;
  50     char *codec_name;
  51
  52     unsigned char *codec_priv;
  53     int codec_priv_size;
  54
  55     uint64_t default_duration;
  56     MatroskaTrackFlags flags;
  57 } MatroskaTrack;
  58
  59 typedef struct MatroskaVideoTrack {
  60     MatroskaTrack track;
  61
  62     int pixel_width;
  63     int pixel_height;
  64     int display_width;
  65     int display_height;
  66
  67     uint32_t fourcc;
  68
  69     MatroskaAspectRatioMode ar_mode;
  70     MatroskaEyeMode eye_mode;
  71
  72     //..
  73 } MatroskaVideoTrack;
  74
  75 typedef struct MatroskaAudioTrack {
  76     MatroskaTrack track;
  77
  78     int channels;
  79     int bitdepth;
  80     int internal_samplerate;
  81     int samplerate;
  82     int block_align;
  83
  84     /* real audio header */
  85     int coded_framesize;
  86     int sub_packet_h;
  87     int frame_size;
  88     int sub_packet_size;
  89     int sub_packet_cnt;
  90     int pkt_cnt;
  91     uint8_t *buf;
  92     //..
  93 } MatroskaAudioTrack;
  94
  95 typedef struct MatroskaSubtitleTrack {
  96     MatroskaTrack track;
  97     //..
  98 } MatroskaSubtitleTrack;
  99
 100 #define MAX_TRACK_SIZE (FFMAX(FFMAX(sizeof(MatroskaVideoTrack), \
 101                                     sizeof(MatroskaAudioTrack)), \
 102                                     sizeof(MatroskaSubtitleTrack)))
 103
 104 typedef struct MatroskaLevel {
 105     uint64_t start;
 106     uint64_t length;
 107 } MatroskaLevel;
 108
 109 typedef struct MatroskaDemuxIndex {
 110   uint64_t        pos;   /* of the corresponding *cluster*! */
 111   uint16_t        track; /* reference to 'num' */
 112   uint64_t        time;  /* in nanoseconds */
 113 } MatroskaDemuxIndex;
 114
 115 typedef struct MatroskaDemuxContext {
 116     AVFormatContext *ctx;
 117
 118     /* ebml stuff */
 119     int num_levels;
 120     MatroskaLevel levels[EBML_MAX_DEPTH];
 121     int level_up;
 122
 123     /* matroska stuff */
 124     char *writing_app;
 125     char *muxing_app;
 126     int64_t created;
 127
 128     /* timescale in the file */
 129     int64_t time_scale;
 130
 131     /* num_streams is the number of streams that av_new_stream() was called
 132      * for ( = that are available to the calling program). */
 133     int num_tracks;
 134     int num_streams;
 135     MatroskaTrack *tracks[MAX_STREAMS];
 136
 137     /* cache for ID peeking */
 138     uint32_t peek_id;
 139
 140     /* byte position of the segment inside the stream */
 141     offset_t segment_start;
 142
 143     /* The packet queue. */
 144     AVPacket **packets;
 145     int num_packets;
 146
 147     /* have we already parse metadata/cues/clusters? */
 148     int metadata_parsed;
 149     int index_parsed;
 150     int done;
 151
 152     /* The index for seeking. */
 153     int num_indexes;
 154     MatroskaDemuxIndex *index;
 155
 156     /* What to skip before effectively reading a packet. */
 157     int skip_to_keyframe;
 158     AVStream *skip_to_stream;
 159 } MatroskaDemuxContext;
 160
 161 /*
 162  * The first few functions handle EBML file parsing. The rest
 163  * is the document interpretation. Matroska really just is a
 164  * EBML file.
 165  */
 166
 167 /*
 168  * Return: the amount of levels in the hierarchy that the
 169  * current element lies higher than the previous one.
 170  * The opposite isn't done - that's auto-done using master
 171  * element reading.
 172  */
 173
 174 static int
 175 ebml_read_element_level_up (MatroskaDemuxContext *matroska)
 176 {
 177     ByteIOContext *pb = matroska->ctx->pb;
 178     offset_t pos = url_ftell(pb);
 179     int num = 0;
 180
 181     while (matroska->num_levels > 0) {
 182         MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
 183
 184         if (pos >= level->start + level->length) {
 185             matroska->num_levels--;
 186             num++;
 187         } else {
 188             break;
 189         }
 190     }
 191
 192     return num;
 193 }
 194
 195 /*
 196  * Read: an "EBML number", which is defined as a variable-length
 197  * array of bytes. The first byte indicates the length by giving a
 198  * number of 0-bits followed by a one. The position of the first
 199  * "one" bit inside the first byte indicates the length of this
 200  * number.
 201  * Returns: num. of bytes read. < 0 on error.
 202  */
 203
 204 static int
 205 ebml_read_num (MatroskaDemuxContext *matroska,
 206                int                   max_size,
 207                uint64_t             *number)
 208 {
 209     ByteIOContext *pb = matroska->ctx->pb;
 210     int len_mask = 0x80, read = 1, n = 1;
 211     int64_t total = 0;
 212
 213     /* the first byte tells us the length in bytes - get_byte() can normally
 214      * return 0, but since that's not a valid first ebmlID byte, we can
 215      * use it safely here to catch EOS. */
 216     if (!(total = get_byte(pb))) {
 217         /* we might encounter EOS here */
 218         if (!url_feof(pb)) {
 219             offset_t pos = url_ftell(pb);
 220             av_log(matroska->ctx, AV_LOG_ERROR,
 221                    "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
 222                    pos, pos);
 223         }
 224         return AVERROR(EIO); /* EOS or actual I/O error */
 225     }
 226
 227     /* get the length of the EBML number */
 228     while (read <= max_size && !(total & len_mask)) {
 229         read++;
 230         len_mask >>= 1;
 231     }
 232     if (read > max_size) {
 233         offset_t pos = url_ftell(pb) - 1;
 234         av_log(matroska->ctx, AV_LOG_ERROR,
 235                "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
 236                (uint8_t) total, pos, pos);
 237         return AVERROR_INVALIDDATA;
 238     }
 239
 240     /* read out length */
 241     total &= ~len_mask;
 242     while (n++ < read)
 243         total = (total << 8) | get_byte(pb);
 244
 245     *number = total;
 246
 247     return read;
 248 }
 249
 250 /*
 251  * Read: the element content data ID.
 252  * Return: the number of bytes read or < 0 on error.
 253  */
 254
 255 static int
 256 ebml_read_element_id (MatroskaDemuxContext *matroska,
 257                       uint32_t             *id,
 258                       int                  *level_up)
 259 {
 260     int read;
 261     uint64_t total;
 262
 263     /* if we re-call this, use our cached ID */
 264     if (matroska->peek_id != 0) {
 265         if (level_up)
 266             *level_up = 0;
 267         *id = matroska->peek_id;
 268         return 0;
 269     }
 270
 271     /* read out the "EBML number", include tag in ID */
 272     if ((read = ebml_read_num(matroska, 4, &total)) < 0)
 273         return read;
 274     *id = matroska->peek_id  = total | (1 << (read * 7));
 275
 276     /* level tracking */
 277     if (level_up)
 278         *level_up = ebml_read_element_level_up(matroska);
 279
 280     return read;
 281 }
 282
 283 /*
 284  * Read: element content length.
 285  * Return: the number of bytes read or < 0 on error.
 286  */
 287
 288 static int
 289 ebml_read_element_length (MatroskaDemuxContext *matroska,
 290                           uint64_t             *length)
 291 {
 292     /* clear cache since we're now beyond that data point */
 293     matroska->peek_id = 0;
 294
 295     /* read out the "EBML number", include tag in ID */
 296     return ebml_read_num(matroska, 8, length);
 297 }
 298
 299 /*
 300  * Return: the ID of the next element, or 0 on error.
 301  * Level_up contains the amount of levels that this
 302  * next element lies higher than the previous one.
 303  */
 304
 305 static uint32_t
 306 ebml_peek_id (MatroskaDemuxContext *matroska,
 307               int                  *level_up)
 308 {
 309     uint32_t id;
 310
 311     if (ebml_read_element_id(matroska, &id, level_up) < 0)
 312         return 0;
 313
 314     return id;
 315 }
 316
 317 /*
 318  * Seek to a given offset.
 319  * 0 is success, -1 is failure.
 320  */
 321
 322 static int
 323 ebml_read_seek (MatroskaDemuxContext *matroska,
 324                 offset_t              offset)
 325 {
 326     ByteIOContext *pb = matroska->ctx->pb;
 327
 328     /* clear ID cache, if any */
 329     matroska->peek_id = 0;
 330
 331     return (url_fseek(pb, offset, SEEK_SET) == offset) ? 0 : -1;
 332 }
 333
 334 /*
 335  * Skip the next element.
 336  * 0 is success, -1 is failure.
 337  */
 338
 339 static int
 340 ebml_read_skip (MatroskaDemuxContext *matroska)
 341 {
 342     ByteIOContext *pb = matroska->ctx->pb;
 343     uint32_t id;
 344     uint64_t length;
 345     int res;
 346
 347     if ((res = ebml_read_element_id(matroska, &id, NULL)) < 0 ||
 348         (res = ebml_read_element_length(matroska, &length)) < 0)
 349         return res;
 350
 351     url_fskip(pb, length);
 352
 353     return 0;
 354 }
 355
 356 /*
 357  * Read the next element as an unsigned int.
 358  * 0 is success, < 0 is failure.
 359  */
 360
 361 static int
 362 ebml_read_uint (MatroskaDemuxContext *matroska,
 363                 uint32_t             *id,
 364                 uint64_t             *num)
 365 {
 366     ByteIOContext *pb = matroska->ctx->pb;
 367     int n = 0, size, res;
 368     uint64_t rlength;
 369
 370     if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
 371         (res = ebml_read_element_length(matroska, &rlength)) < 0)
 372         return res;
 373     size = rlength;
 374     if (size < 1 || size > 8) {
 375         offset_t pos = url_ftell(pb);
 376         av_log(matroska->ctx, AV_LOG_ERROR,
 377                "Invalid uint element size %d at position %"PRId64" (0x%"PRIx64")\n",
 378                 size, pos, pos);
 379         return AVERROR_INVALIDDATA;
 380     }
 381
 382     /* big-endian ordening; build up number */
 383     *num = 0;
 384     while (n++ < size)
 385         *num = (*num << 8) | get_byte(pb);
 386
 387     return 0;
 388 }
 389
 390 /*
 391  * Read the next element as a signed int.
 392  * 0 is success, < 0 is failure.
 393  */
 394
 395 static int
 396 ebml_read_sint (MatroskaDemuxContext *matroska,
 397                 uint32_t             *id,
 398                 int64_t              *num)
 399 {
 400     ByteIOContext *pb = matroska->ctx->pb;
 401     int size, n = 1, negative = 0, res;
 402     uint64_t rlength;
 403
 404     if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
 405         (res = ebml_read_element_length(matroska, &rlength)) < 0)
 406         return res;
 407     size = rlength;
 408     if (size < 1 || size > 8) {
 409         offset_t pos = url_ftell(pb);
 410         av_log(matroska->ctx, AV_LOG_ERROR,
 411                "Invalid sint element size %d at position %"PRId64" (0x%"PRIx64")\n",
 412                 size, pos, pos);
 413         return AVERROR_INVALIDDATA;
 414     }
 415     if ((*num = get_byte(pb)) & 0x80) {
 416         negative = 1;
 417         *num &= ~0x80;
 418     }
 419     while (n++ < size)
 420         *num = (*num << 8) | get_byte(pb);
 421
 422     /* make signed */
 423     if (negative)
 424         *num = *num - (1LL << ((8 * size) - 1));
 425
 426     return 0;
 427 }
 428
 429 /*
 430  * Read the next element as a float.
 431  * 0 is success, < 0 is failure.
 432  */
 433
 434 static int
 435 ebml_read_float (MatroskaDemuxContext *matroska,
 436                  uint32_t             *id,
 437                  double               *num)
 438 {
 439     ByteIOContext *pb = matroska->ctx->pb;
 440     int size, res;
 441     uint64_t rlength;
 442
 443     if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
 444         (res = ebml_read_element_length(matroska, &rlength)) < 0)
 445         return res;
 446     size = rlength;
 447
 448     if (size == 4) {
 449         *num= av_int2flt(get_be32(pb));
 450     } else if(size==8){
 451         *num= av_int2dbl(get_be64(pb));
 452     } else{
 453         offset_t pos = url_ftell(pb);
 454         av_log(matroska->ctx, AV_LOG_ERROR,
 455                "Invalid float element size %d at position %"PRIu64" (0x%"PRIx64")\n",
 456                size, pos, pos);
 457         return AVERROR_INVALIDDATA;
 458     }
 459
 460     return 0;
 461 }
 462
 463 /*
 464  * Read the next element as an ASCII string.
 465  * 0 is success, < 0 is failure.
 466  */
 467
 468 static int
 469 ebml_read_ascii (MatroskaDemuxContext *matroska,
 470                  uint32_t             *id,
 471                  char                **str)
 472 {
 473     ByteIOContext *pb = matroska->ctx->pb;
 474     int size, res;
 475     uint64_t rlength;
 476
 477     if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
 478         (res = ebml_read_element_length(matroska, &rlength)) < 0)
 479         return res;
 480     size = rlength;
 481
 482     /* ebml strings are usually not 0-terminated, so we allocate one
 483      * byte more, read the string and NULL-terminate it ourselves. */
 484     if (size < 0 || !(*str = av_malloc(size + 1))) {
 485         av_log(matroska->ctx, AV_LOG_ERROR, "Memory allocation failed\n");
 486         return AVERROR(ENOMEM);
 487     }
 488     if (get_buffer(pb, (uint8_t *) *str, size) != size) {
 489         offset_t pos = url_ftell(pb);
 490         av_log(matroska->ctx, AV_LOG_ERROR,
 491                "Read error at pos. %"PRIu64" (0x%"PRIx64")\n", pos, pos);
 492         return AVERROR(EIO);
 493     }
 494     (*str)[size] = '\0';
 495
 496     return 0;
 497 }
 498
 499 /*
 500  * Read the next element as a UTF-8 string.
 501  * 0 is success, < 0 is failure.
 502  */
 503
 504 static int
 505 ebml_read_utf8 (MatroskaDemuxContext *matroska,
 506                 uint32_t             *id,
 507                 char                **str)
 508 {
 509   return ebml_read_ascii(matroska, id, str);
 510 }
 511
 512 /*
 513  * Read the next element as a date (nanoseconds since 1/1/2000).
 514  * 0 is success, < 0 is failure.
 515  */
 516
 517 static int
 518 ebml_read_date (MatroskaDemuxContext *matroska,
 519                 uint32_t             *id,
 520                 int64_t              *date)
 521 {
 522   return ebml_read_sint(matroska, id, date);
 523 }
 524
 525 /*
 526  * Read the next element, but only the header. The contents
 527  * are supposed to be sub-elements which can be read separately.
 528  * 0 is success, < 0 is failure.
 529  */
 530
 531 static int
 532 ebml_read_master (MatroskaDemuxContext *matroska,
 533                   uint32_t             *id)
 534 {
 535     ByteIOContext *pb = matroska->ctx->pb;
 536     uint64_t length;
 537     MatroskaLevel *level;
 538     int res;
 539
 540     if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
 541         (res = ebml_read_element_length(matroska, &length)) < 0)
 542         return res;
 543
 544     /* protect... (Heaven forbids that the '>' is true) */
 545     if (matroska->num_levels >= EBML_MAX_DEPTH) {
 546         av_log(matroska->ctx, AV_LOG_ERROR,
 547                "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
 548         return AVERROR(ENOSYS);
 549     }
 550
 551     /* remember level */
 552     level = &matroska->levels[matroska->num_levels++];
 553     level->start = url_ftell(pb);
 554     level->length = length;
 555
 556     return 0;
 557 }
 558
 559 /*
 560  * Read the next element as binary data.
 561  * 0 is success, < 0 is failure.
 562  */
 563
 564 static int
 565 ebml_read_binary (MatroskaDemuxContext *matroska,
 566                   uint32_t             *id,
 567                   uint8_t             **binary,
 568                   int                  *size)
 569 {
 570     ByteIOContext *pb = matroska->ctx->pb;
 571     uint64_t rlength;
 572     int res;
 573
 574     if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
 575         (res = ebml_read_element_length(matroska, &rlength)) < 0)
 576         return res;
 577     *size = rlength;
 578
 579     if (!(*binary = av_malloc(*size))) {
 580         av_log(matroska->ctx, AV_LOG_ERROR,
 581                "Memory allocation error\n");
 582         return AVERROR(ENOMEM);
 583     }
 584
 585     if (get_buffer(pb, *binary, *size) != *size) {
 586         offset_t pos = url_ftell(pb);
 587         av_log(matroska->ctx, AV_LOG_ERROR,
 588                "Read error at pos. %"PRIu64" (0x%"PRIx64")\n", pos, pos);
 589         return AVERROR(EIO);
 590     }
 591
 592     return 0;
 593 }
 594
 595 /*
 596  * Read signed/unsigned "EBML" numbers.
 597  * Return: number of bytes processed, < 0 on error.
 598  * XXX: use ebml_read_num().
 599  */
 600
 601 static int
 602 matroska_ebmlnum_uint (uint8_t  *data,
 603                        uint32_t  size,
 604                        uint64_t *num)
 605 {
 606     int len_mask = 0x80, read = 1, n = 1, num_ffs = 0;
 607     uint64_t total;
 608
 609     if (size <= 0)
 610         return AVERROR_INVALIDDATA;
 611
 612     total = data[0];
 613     while (read <= 8 && !(total & len_mask)) {
 614         read++;
 615         len_mask >>= 1;
 616     }
 617     if (read > 8)
 618         return AVERROR_INVALIDDATA;
 619
 620     if ((total &= (len_mask - 1)) == len_mask - 1)
 621         num_ffs++;
 622     if (size < read)
 623         return AVERROR_INVALIDDATA;
 624     while (n < read) {
 625         if (data[n] == 0xff)
 626             num_ffs++;
 627         total = (total << 8) | data[n];
 628         n++;
 629     }
 630
 631     if (read == num_ffs)
 632         *num = (uint64_t)-1;
 633     else
 634         *num = total;
 635
 636     return read;
 637 }
 638
 639 /*
 640  * Same as above, but signed.
 641  */
 642
 643 static int
 644 matroska_ebmlnum_sint (uint8_t  *data,
 645                        uint32_t  size,
 646                        int64_t  *num)
 647 {
 648     uint64_t unum;
 649     int res;
 650
 651     /* read as unsigned number first */
 652     if ((res = matroska_ebmlnum_uint(data, size, &unum)) < 0)
 653         return res;
 654
 655     /* make signed (weird way) */
 656     if (unum == (uint64_t)-1)
 657         *num = INT64_MAX;
 658     else
 659         *num = unum - ((1LL << ((7 * res) - 1)) - 1);
 660
 661     return res;
 662 }
 663
 664 /*
 665  * Read an EBML header.
 666  * 0 is success, < 0 is failure.
 667  */
 668
 669 static int
 670 ebml_read_header (MatroskaDemuxContext *matroska,
 671                   char                **doctype,
 672                   int                  *version)
 673 {
 674     uint32_t id;
 675     int level_up, res = 0;
 676
 677     /* default init */
 678     if (doctype)
 679         *doctype = NULL;
 680     if (version)
 681         *version = 1;
 682
 683     if (!(id = ebml_peek_id(matroska, &level_up)) ||
 684         level_up != 0 || id != EBML_ID_HEADER) {
 685         av_log(matroska->ctx, AV_LOG_ERROR,
 686                "This is not an EBML file (id=0x%x/0x%x)\n", id, EBML_ID_HEADER);
 687         return AVERROR_INVALIDDATA;
 688     }
 689     if ((res = ebml_read_master(matroska, &id)) < 0)
 690         return res;
 691
 692     while (res == 0) {
 693         if (!(id = ebml_peek_id(matroska, &level_up)))
 694             return AVERROR(EIO);
 695
 696         /* end-of-header */
 697         if (level_up)
 698             break;
 699
 700         switch (id) {
 701             /* is our read version uptodate? */
 702             case EBML_ID_EBMLREADVERSION: {
 703                 uint64_t num;
 704
 705                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
 706                     return res;
 707                 if (num > EBML_VERSION) {
 708                     av_log(matroska->ctx, AV_LOG_ERROR,
 709                            "EBML version %"PRIu64" (> %d) is not supported\n",
 710                            num, EBML_VERSION);
 711                     return AVERROR_INVALIDDATA;
 712                 }
 713                 break;
 714             }
 715
 716             /* we only handle 8 byte lengths at max */
 717             case EBML_ID_EBMLMAXSIZELENGTH: {
 718                 uint64_t num;
 719
 720                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
 721                     return res;
 722                 if (num > sizeof(uint64_t)) {
 723                     av_log(matroska->ctx, AV_LOG_ERROR,
 724                            "Integers of size %"PRIu64" (> %zd) not supported\n",
 725                            num, sizeof(uint64_t));
 726                     return AVERROR_INVALIDDATA;
 727                 }
 728                 break;
 729             }
 730
 731             /* we handle 4 byte IDs at max */
 732             case EBML_ID_EBMLMAXIDLENGTH: {
 733                 uint64_t num;
 734
 735                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
 736                     return res;
 737                 if (num > sizeof(uint32_t)) {
 738                     av_log(matroska->ctx, AV_LOG_ERROR,
 739                            "IDs of size %"PRIu64" (> %zu) not supported\n",
 740                             num, sizeof(uint32_t));
 741                     return AVERROR_INVALIDDATA;
 742                 }
 743                 break;
 744             }
 745
 746             case EBML_ID_DOCTYPE: {
 747                 char *text;
 748
 749                 if ((res = ebml_read_ascii(matroska, &id, &text)) < 0)
 750                     return res;
 751                 if (doctype) {
 752                     if (*doctype)
 753                         av_free(*doctype);
 754                     *doctype = text;
 755                 } else
 756                     av_free(text);
 757                 break;
 758             }
 759
 760             case EBML_ID_DOCTYPEREADVERSION: {
 761                 uint64_t num;
 762
 763                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
 764                     return res;
 765                 if (version)
 766                     *version = num;
 767                 break;
 768             }
 769
 770             default:
 771                 av_log(matroska->ctx, AV_LOG_INFO,
 772                        "Unknown data type 0x%x in EBML header", id);
 773                 /* pass-through */
 774
 775             case EBML_ID_VOID:
 776             /* we ignore these two, as they don't tell us anything we
 777              * care about */
 778             case EBML_ID_EBMLVERSION:
 779             case EBML_ID_DOCTYPEVERSION:
 780                 res = ebml_read_skip (matroska);
 781                 break;
 782         }
 783     }
 784
 785     return 0;
 786 }
 787
 788
 789 static int
 790 matroska_find_track_by_num (MatroskaDemuxContext *matroska,
 791                             int                   num)
 792 {
 793     int i;
 794
 795     for (i = 0; i < matroska->num_tracks; i++)
 796         if (matroska->tracks[i]->num == num)
 797             return i;
 798
 799     return -1;
 800 }
 801
 802
 803 /*
 804  * Put one packet in an application-supplied AVPacket struct.
 805  * Returns 0 on success or -1 on failure.
 806  */
 807
 808 static int
 809 matroska_deliver_packet (MatroskaDemuxContext *matroska,
 810                          AVPacket             *pkt)
 811 {
 812     if (matroska->num_packets > 0) {
 813         memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
 814         av_free(matroska->packets[0]);
 815         if (matroska->num_packets > 1) {
 816             memmove(&matroska->packets[0], &matroska->packets[1],
 817                     (matroska->num_packets - 1) * sizeof(AVPacket *));
 818             matroska->packets =
 819                 av_realloc(matroska->packets, (matroska->num_packets - 1) *
 820                            sizeof(AVPacket *));
 821         } else {
 822             av_freep(&matroska->packets);
 823         }
 824         matroska->num_packets--;
 825         return 0;
 826     }
 827
 828     return -1;
 829 }
 830
 831 /*
 832  * Put a packet into our internal queue. Will be delivered to the
 833  * user/application during the next get_packet() call.
 834  */
 835
 836 static void
 837 matroska_queue_packet (MatroskaDemuxContext *matroska,
 838                        AVPacket             *pkt)
 839 {
 840     matroska->packets =
 841         av_realloc(matroska->packets, (matroska->num_packets + 1) *
 842                    sizeof(AVPacket *));
 843     matroska->packets[matroska->num_packets] = pkt;
 844     matroska->num_packets++;
 845 }
 846
 847 /*
 848  * Free all packets in our internal queue.
 849  */
 850 static void
 851 matroska_clear_queue (MatroskaDemuxContext *matroska)
 852 {
 853     if (matroska->packets) {
 854         int n;
 855         for (n = 0; n < matroska->num_packets; n++) {
 856             av_free_packet(matroska->packets[n]);
 857             av_free(matroska->packets[n]);
 858         }
 859         av_free(matroska->packets);
 860         matroska->packets = NULL;
 861         matroska->num_packets = 0;
 862     }
 863 }
 864
 865
 866 /*
 867  * Autodetecting...
 868  */
 869
 870 static int
 871 matroska_probe (AVProbeData *p)
 872 {
 873     uint64_t total = 0;
 874     int len_mask = 0x80, size = 1, n = 1;
 875     uint8_t probe_data[] = { 'm', 'a', 't', 'r', 'o', 's', 'k', 'a' };
 876
 877     /* ebml header? */
 878     if (AV_RB32(p->buf) != EBML_ID_HEADER)
 879         return 0;
 880
 881     /* length of header */
 882     total = p->buf[4];
 883     while (size <= 8 && !(total & len_mask)) {
 884         size++;
 885         len_mask >>= 1;
 886     }
 887     if (size > 8)
 888       return 0;
 889     total &= (len_mask - 1);
 890     while (n < size)
 891         total = (total << 8) | p->buf[4 + n++];
 892
 893     /* does the probe data contain the whole header? */
 894     if (p->buf_size < 4 + size + total)
 895       return 0;
 896
 897     /* the header must contain the document type 'matroska'. For now,
 898      * we don't parse the whole header but simply check for the
 899      * availability of that array of characters inside the header.
 900      * Not fully fool-proof, but good enough. */
 901     for (n = 4 + size; n <= 4 + size + total - sizeof(probe_data); n++)
 902         if (!memcmp (&p->buf[n], probe_data, sizeof(probe_data)))
 903             return AVPROBE_SCORE_MAX;
 904
 905     return 0;
 906 }
 907
 908 /*
 909  * From here on, it's all XML-style DTD stuff... Needs no comments.
 910  */
 911
 912 static int
 913 matroska_parse_info (MatroskaDemuxContext *matroska)
 914 {
 915     int res = 0;
 916     uint32_t id;
 917
 918     av_log(matroska->ctx, AV_LOG_DEBUG, "Parsing info...\n");
 919
 920     while (res == 0) {
 921         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
 922             res = AVERROR(EIO);
 923             break;
 924         } else if (matroska->level_up) {
 925             matroska->level_up--;
 926             break;
 927         }
 928
 929         switch (id) {
 930             /* cluster timecode */
 931             case MATROSKA_ID_TIMECODESCALE: {
 932                 uint64_t num;
 933                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
 934                     break;
 935                 matroska->time_scale = num;
 936                 break;
 937             }
 938
 939             case MATROSKA_ID_DURATION: {
 940                 double num;
 941                 if ((res = ebml_read_float(matroska, &id, &num)) < 0)
 942                     break;
 943                 matroska->ctx->duration = num * matroska->time_scale * 1000 / AV_TIME_BASE;
 944                 break;
 945             }
 946
 947             case MATROSKA_ID_TITLE: {
 948                 char *text;
 949                 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
 950                     break;
 951                 strncpy(matroska->ctx->title, text,
 952                         sizeof(matroska->ctx->title)-1);
 953                 av_free(text);
 954                 break;
 955             }
 956
 957             case MATROSKA_ID_WRITINGAPP: {
 958                 char *text;
 959                 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
 960                     break;
 961                 matroska->writing_app = text;
 962                 break;
 963             }
 964
 965             case MATROSKA_ID_MUXINGAPP: {
 966                 char *text;
 967                 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
 968                     break;
 969                 matroska->muxing_app = text;
 970                 break;
 971             }
 972
 973             case MATROSKA_ID_DATEUTC: {
 974                 int64_t time;
 975                 if ((res = ebml_read_date(matroska, &id, &time)) < 0)
 976                     break;
 977                 matroska->created = time;
 978                 break;
 979             }
 980
 981             default:
 982                 av_log(matroska->ctx, AV_LOG_INFO,
 983                        "Unknown entry 0x%x in info header\n", id);
 984                 /* fall-through */
 985
 986             case EBML_ID_VOID:
 987                 res = ebml_read_skip(matroska);
 988                 break;
 989         }
 990
 991         if (matroska->level_up) {
 992             matroska->level_up--;
 993             break;
 994         }
 995     }
 996
 997     return res;
 998 }
 999
1000 static int
1001 matroska_add_stream (MatroskaDemuxContext *matroska)
1002 {
1003     int res = 0;
1004     uint32_t id;
1005     MatroskaTrack *track;
1006
1007     av_log(matroska->ctx, AV_LOG_DEBUG, "parsing track, adding stream..,\n");
1008
1009     /* Allocate a generic track. As soon as we know its type we'll realloc. */
1010     track = av_mallocz(MAX_TRACK_SIZE);
1011     matroska->num_tracks++;
1012     strcpy(track->language, "eng");
1013
1014     /* start with the master */
1015     if ((res = ebml_read_master(matroska, &id)) < 0)
1016         return res;
1017
1018     /* try reading the trackentry headers */
1019     while (res == 0) {
1020         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1021             res = AVERROR(EIO);
1022             break;
1023         } else if (matroska->level_up > 0) {
1024             matroska->level_up--;
1025             break;
1026         }
1027
1028         switch (id) {
1029             /* track number (unique stream ID) */
1030             case MATROSKA_ID_TRACKNUMBER: {
1031                 uint64_t num;
1032                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1033                     break;
1034                 track->num = num;
1035                 break;
1036             }
1037
1038             /* track UID (unique identifier) */
1039             case MATROSKA_ID_TRACKUID: {
1040                 uint64_t num;
1041                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1042                     break;
1043                 track->uid = num;
1044                 break;
1045             }
1046
1047             /* track type (video, audio, combined, subtitle, etc.) */
1048             case MATROSKA_ID_TRACKTYPE: {
1049                 uint64_t num;
1050                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1051                     break;
1052                 if (track->type && track->type != num) {
1053                     av_log(matroska->ctx, AV_LOG_INFO,
1054                            "More than one tracktype in an entry - skip\n");
1055                     break;
1056                 }
1057                 track->type = num;
1058
1059                 switch (track->type) {
1060                     case MATROSKA_TRACK_TYPE_VIDEO:
1061                     case MATROSKA_TRACK_TYPE_AUDIO:
1062                     case MATROSKA_TRACK_TYPE_SUBTITLE:
1063                         break;
1064                     case MATROSKA_TRACK_TYPE_COMPLEX:
1065                     case MATROSKA_TRACK_TYPE_LOGO:
1066                     case MATROSKA_TRACK_TYPE_CONTROL:
1067                     default:
1068                         av_log(matroska->ctx, AV_LOG_INFO,
1069                                "Unknown or unsupported track type 0x%x\n",
1070                                track->type);
1071                         track->type = 0;
1072                         break;
1073                 }
1074                 matroska->tracks[matroska->num_tracks - 1] = track;
1075                 break;
1076             }
1077
1078             /* tracktype specific stuff for video */
1079             case MATROSKA_ID_TRACKVIDEO: {
1080                 MatroskaVideoTrack *videotrack;
1081                 if (!track->type)
1082                     track->type = MATROSKA_TRACK_TYPE_VIDEO;
1083                 if (track->type != MATROSKA_TRACK_TYPE_VIDEO) {
1084                     av_log(matroska->ctx, AV_LOG_INFO,
1085                            "video data in non-video track - ignoring\n");
1086                     res = AVERROR_INVALIDDATA;
1087                     break;
1088                 } else if ((res = ebml_read_master(matroska, &id)) < 0)
1089                     break;
1090                 videotrack = (MatroskaVideoTrack *)track;
1091
1092                 while (res == 0) {
1093                     if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1094                         res = AVERROR(EIO);
1095                         break;
1096                     } else if (matroska->level_up > 0) {
1097                         matroska->level_up--;
1098                         break;
1099                     }
1100
1101                     switch (id) {
1102                         /* fixme, this should be one-up, but I get it here */
1103                         case MATROSKA_ID_TRACKDEFAULTDURATION: {
1104                             uint64_t num;
1105                             if ((res = ebml_read_uint (matroska, &id,
1106                                                        &num)) < 0)
1107                                 break;
1108                             track->default_duration = num;
1109                             break;
1110                         }
1111
1112                         /* video framerate */
1113                         case MATROSKA_ID_VIDEOFRAMERATE: {
1114                             double num;
1115                             if ((res = ebml_read_float(matroska, &id,
1116                                                        &num)) < 0)
1117                                 break;
1118                             if (!track->default_duration)
1119                                 track->default_duration = 1000000000/num;
1120                             break;
1121                         }
1122
1123                         /* width of the size to display the video at */
1124                         case MATROSKA_ID_VIDEODISPLAYWIDTH: {
1125                             uint64_t num;
1126                             if ((res = ebml_read_uint(matroska, &id,
1127                                                       &num)) < 0)
1128                                 break;
1129                             videotrack->display_width = num;
1130                             break;
1131                         }
1132
1133                         /* height of the size to display the video at */
1134                         case MATROSKA_ID_VIDEODISPLAYHEIGHT: {
1135                             uint64_t num;
1136                             if ((res = ebml_read_uint(matroska, &id,
1137                                                       &num)) < 0)
1138                                 break;
1139                             videotrack->display_height = num;
1140                             break;
1141                         }
1142
1143                         /* width of the video in the file */
1144                         case MATROSKA_ID_VIDEOPIXELWIDTH: {
1145                             uint64_t num;
1146                             if ((res = ebml_read_uint(matroska, &id,
1147                                                       &num)) < 0)
1148                                 break;
1149                             videotrack->pixel_width = num;
1150                             break;
1151                         }
1152
1153                         /* height of the video in the file */
1154                         case MATROSKA_ID_VIDEOPIXELHEIGHT: {
1155                             uint64_t num;
1156                             if ((res = ebml_read_uint(matroska, &id,
1157                                                       &num)) < 0)
1158                                 break;
1159                             videotrack->pixel_height = num;
1160                             break;
1161                         }
1162
1163                         /* whether the video is interlaced */
1164                         case MATROSKA_ID_VIDEOFLAGINTERLACED: {
1165                             uint64_t num;
1166                             if ((res = ebml_read_uint(matroska, &id,
1167                                                       &num)) < 0)
1168                                 break;
1169                             if (num)
1170                                 track->flags |=
1171                                     MATROSKA_VIDEOTRACK_INTERLACED;
1172                             else
1173                                 track->flags &=
1174                                     ~MATROSKA_VIDEOTRACK_INTERLACED;
1175                             break;
1176                         }
1177
1178                         /* stereo mode (whether the video has two streams,
1179                          * where one is for the left eye and the other for
1180                          * the right eye, which creates a 3D-like
1181                          * effect) */
1182                         case MATROSKA_ID_VIDEOSTEREOMODE: {
1183                             uint64_t num;
1184                             if ((res = ebml_read_uint(matroska, &id,
1185                                                       &num)) < 0)
1186                                 break;
1187                             if (num != MATROSKA_EYE_MODE_MONO &&
1188                                 num != MATROSKA_EYE_MODE_LEFT &&
1189                                 num != MATROSKA_EYE_MODE_RIGHT &&
1190                                 num != MATROSKA_EYE_MODE_BOTH) {
1191                                 av_log(matroska->ctx, AV_LOG_INFO,
1192                                        "Ignoring unknown eye mode 0x%x\n",
1193                                        (uint32_t) num);
1194                                 break;
1195                             }
1196                             videotrack->eye_mode = num;
1197                             break;
1198                         }
1199
1200                         /* aspect ratio behaviour */
1201                         case MATROSKA_ID_VIDEOASPECTRATIO: {
1202                             uint64_t num;
1203                             if ((res = ebml_read_uint(matroska, &id,
1204                                                       &num)) < 0)
1205                                 break;
1206                             if (num != MATROSKA_ASPECT_RATIO_MODE_FREE &&
1207                                 num != MATROSKA_ASPECT_RATIO_MODE_KEEP &&
1208                                 num != MATROSKA_ASPECT_RATIO_MODE_FIXED) {
1209                                 av_log(matroska->ctx, AV_LOG_INFO,
1210                                        "Ignoring unknown aspect ratio 0x%x\n",
1211                                        (uint32_t) num);
1212                                 break;
1213                             }
1214                             videotrack->ar_mode = num;
1215                             break;
1216                         }
1217
1218                         /* colorspace (only matters for raw video)
1219                          * fourcc */
1220                         case MATROSKA_ID_VIDEOCOLORSPACE: {
1221                             uint64_t num;
1222                             if ((res = ebml_read_uint(matroska, &id,
1223                                                       &num)) < 0)
1224                                 break;
1225                             videotrack->fourcc = num;
1226                             break;
1227                         }
1228
1229                         default:
1230                             av_log(matroska->ctx, AV_LOG_INFO,
1231                                    "Unknown video track header entry "
1232                                    "0x%x - ignoring\n", id);
1233                             /* pass-through */
1234
1235                         case EBML_ID_VOID:
1236                             res = ebml_read_skip(matroska);
1237                             break;
1238                     }
1239
1240                     if (matroska->level_up) {
1241                         matroska->level_up--;
1242                         break;
1243                     }
1244                 }
1245                 break;
1246             }
1247
1248             /* tracktype specific stuff for audio */
1249             case MATROSKA_ID_TRACKAUDIO: {
1250                 MatroskaAudioTrack *audiotrack;
1251                 if (!track->type)
1252                     track->type = MATROSKA_TRACK_TYPE_AUDIO;
1253                 if (track->type != MATROSKA_TRACK_TYPE_AUDIO) {
1254                     av_log(matroska->ctx, AV_LOG_INFO,
1255                            "audio data in non-audio track - ignoring\n");
1256                     res = AVERROR_INVALIDDATA;
1257                     break;
1258                 } else if ((res = ebml_read_master(matroska, &id)) < 0)
1259                     break;
1260                 audiotrack = (MatroskaAudioTrack *)track;
1261                 audiotrack->channels = 1;
1262                 audiotrack->samplerate = 8000;
1263
1264                 while (res == 0) {
1265                     if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1266                         res = AVERROR(EIO);
1267                         break;
1268                     } else if (matroska->level_up > 0) {
1269                         matroska->level_up--;
1270                         break;
1271                     }
1272
1273                     switch (id) {
1274                         /* samplerate */
1275                         case MATROSKA_ID_AUDIOSAMPLINGFREQ: {
1276                             double num;
1277                             if ((res = ebml_read_float(matroska, &id,
1278                                                        &num)) < 0)
1279                                 break;
1280                             audiotrack->internal_samplerate =
1281                             audiotrack->samplerate = num;
1282                             break;
1283                         }
1284
1285                         case MATROSKA_ID_AUDIOOUTSAMPLINGFREQ: {
1286                             double num;
1287                             if ((res = ebml_read_float(matroska, &id,
1288                                                        &num)) < 0)
1289                                 break;
1290                             audiotrack->samplerate = num;
1291                             break;
1292                         }
1293
1294                             /* bitdepth */
1295                         case MATROSKA_ID_AUDIOBITDEPTH: {
1296                             uint64_t num;
1297                             if ((res = ebml_read_uint(matroska, &id,
1298                                                       &num)) < 0)
1299                                 break;
1300                             audiotrack->bitdepth = num;
1301                             break;
1302                         }
1303
1304                             /* channels */
1305                         case MATROSKA_ID_AUDIOCHANNELS: {
1306                             uint64_t num;
1307                             if ((res = ebml_read_uint(matroska, &id,
1308                                                       &num)) < 0)
1309                                 break;
1310                             audiotrack->channels = num;
1311                             break;
1312                         }
1313
1314                         default:
1315                             av_log(matroska->ctx, AV_LOG_INFO,
1316                                    "Unknown audio track header entry "
1317                                    "0x%x - ignoring\n", id);
1318                             /* pass-through */
1319
1320                         case EBML_ID_VOID:
1321                             res = ebml_read_skip(matroska);
1322                             break;
1323                     }
1324
1325                     if (matroska->level_up) {
1326                         matroska->level_up--;
1327                         break;
1328                     }
1329                 }
1330                 break;
1331             }
1332
1333                 /* codec identifier */
1334             case MATROSKA_ID_CODECID: {
1335                 char *text;
1336                 if ((res = ebml_read_ascii(matroska, &id, &text)) < 0)
1337                     break;
1338                 track->codec_id = text;
1339                 break;
1340             }
1341
1342                 /* codec private data */
1343             case MATROSKA_ID_CODECPRIVATE: {
1344                 uint8_t *data;
1345                 int size;
1346                 if ((res = ebml_read_binary(matroska, &id, &data, &size) < 0))
1347                     break;
1348                 track->codec_priv = data;
1349                 track->codec_priv_size = size;
1350                 break;
1351             }
1352
1353                 /* name of the codec */
1354             case MATROSKA_ID_CODECNAME: {
1355                 char *text;
1356                 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
1357                     break;
1358                 track->codec_name = text;
1359                 break;
1360             }
1361
1362                 /* name of this track */
1363             case MATROSKA_ID_TRACKNAME: {
1364                 char *text;
1365                 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
1366                     break;
1367                 track->name = text;
1368                 break;
1369             }
1370
1371                 /* language (matters for audio/subtitles, mostly) */
1372             case MATROSKA_ID_TRACKLANGUAGE: {
1373                 char *text, *end;
1374                 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
1375                     break;
1376                 if ((end = strchr(text, '-')))
1377                     *end = '\0';
1378                 if (strlen(text) == 3)
1379                     strcpy(track->language, text);
1380                 av_free(text);
1381                 break;
1382             }
1383
1384                 /* whether this is actually used */
1385             case MATROSKA_ID_TRACKFLAGENABLED: {
1386                 uint64_t num;
1387                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1388                     break;
1389                 if (num)
1390                     track->flags |= MATROSKA_TRACK_ENABLED;
1391                 else
1392                     track->flags &= ~MATROSKA_TRACK_ENABLED;
1393                 break;
1394             }
1395
1396                 /* whether it's the default for this track type */
1397             case MATROSKA_ID_TRACKFLAGDEFAULT: {
1398                 uint64_t num;
1399                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1400                     break;
1401                 if (num)
1402                     track->flags |= MATROSKA_TRACK_DEFAULT;
1403                 else
1404                     track->flags &= ~MATROSKA_TRACK_DEFAULT;
1405                 break;
1406             }
1407
1408                 /* lacing (like MPEG, where blocks don't end/start on frame
1409                  * boundaries) */
1410             case MATROSKA_ID_TRACKFLAGLACING: {
1411                 uint64_t num;
1412                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1413                     break;
1414                 if (num)
1415                     track->flags |= MATROSKA_TRACK_LACING;
1416                 else
1417                     track->flags &= ~MATROSKA_TRACK_LACING;
1418                 break;
1419             }
1420
1421                 /* default length (in time) of one data block in this track */
1422             case MATROSKA_ID_TRACKDEFAULTDURATION: {
1423                 uint64_t num;
1424                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1425                     break;
1426                 track->default_duration = num;
1427                 break;
1428             }
1429
1430             default:
1431                 av_log(matroska->ctx, AV_LOG_INFO,
1432                        "Unknown track header entry 0x%x - ignoring\n", id);
1433                 /* pass-through */
1434
1435             case EBML_ID_VOID:
1436             /* we ignore these because they're nothing useful. */
1437             case MATROSKA_ID_CODECINFOURL:
1438             case MATROSKA_ID_CODECDOWNLOADURL:
1439             case MATROSKA_ID_TRACKMINCACHE:
1440             case MATROSKA_ID_TRACKMAXCACHE:
1441                 res = ebml_read_skip(matroska);
1442                 break;
1443         }
1444
1445         if (matroska->level_up) {
1446             matroska->level_up--;
1447             break;
1448         }
1449     }
1450
1451     return res;
1452 }
1453
1454 static int
1455 matroska_parse_tracks (MatroskaDemuxContext *matroska)
1456 {
1457     int res = 0;
1458     uint32_t id;
1459
1460     av_log(matroska->ctx, AV_LOG_DEBUG, "parsing tracks...\n");
1461
1462     while (res == 0) {
1463         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1464             res = AVERROR(EIO);
1465             break;
1466         } else if (matroska->level_up) {
1467             matroska->level_up--;
1468             break;
1469         }
1470
1471         switch (id) {
1472             /* one track within the "all-tracks" header */
1473             case MATROSKA_ID_TRACKENTRY:
1474                 res = matroska_add_stream(matroska);
1475                 break;
1476
1477             default:
1478                 av_log(matroska->ctx, AV_LOG_INFO,
1479                        "Unknown entry 0x%x in track header\n", id);
1480                 /* fall-through */
1481
1482             case EBML_ID_VOID:
1483                 res = ebml_read_skip(matroska);
1484                 break;
1485         }
1486
1487         if (matroska->level_up) {
1488             matroska->level_up--;
1489             break;
1490         }
1491     }
1492
1493     return res;
1494 }
1495
1496 static int
1497 matroska_parse_index (MatroskaDemuxContext *matroska)
1498 {
1499     int res = 0;
1500     uint32_t id;
1501     MatroskaDemuxIndex idx;
1502
1503     av_log(matroska->ctx, AV_LOG_DEBUG, "parsing index...\n");
1504
1505     while (res == 0) {
1506         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1507             res = AVERROR(EIO);
1508             break;
1509         } else if (matroska->level_up) {
1510             matroska->level_up--;
1511             break;
1512         }
1513
1514         switch (id) {
1515             /* one single index entry ('point') */
1516             case MATROSKA_ID_POINTENTRY:
1517                 if ((res = ebml_read_master(matroska, &id)) < 0)
1518                     break;
1519
1520                 /* in the end, we hope to fill one entry with a
1521                  * timestamp, a file position and a tracknum */
1522                 idx.pos   = (uint64_t) -1;
1523                 idx.time  = (uint64_t) -1;
1524                 idx.track = (uint16_t) -1;
1525
1526                 while (res == 0) {
1527                     if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1528                         res = AVERROR(EIO);
1529                         break;
1530                     } else if (matroska->level_up) {
1531                         matroska->level_up--;
1532                         break;
1533                     }
1534
1535                     switch (id) {
1536                         /* one single index entry ('point') */
1537                         case MATROSKA_ID_CUETIME: {
1538                             uint64_t time;
1539                             if ((res = ebml_read_uint(matroska, &id,
1540                                                       &time)) < 0)
1541                                 break;
1542                             idx.time = time * matroska->time_scale;
1543                             break;
1544                         }
1545
1546                         /* position in the file + track to which it
1547                          * belongs */
1548                         case MATROSKA_ID_CUETRACKPOSITION:
1549                             if ((res = ebml_read_master(matroska, &id)) < 0)
1550                                 break;
1551
1552                             while (res == 0) {
1553                                 if (!(id = ebml_peek_id (matroska,
1554                                                     &matroska->level_up))) {
1555                                     res = AVERROR(EIO);
1556                                     break;
1557                                 } else if (matroska->level_up) {
1558                                     matroska->level_up--;
1559                                     break;
1560                                 }
1561
1562                                 switch (id) {
1563                                     /* track number */
1564                                     case MATROSKA_ID_CUETRACK: {
1565                                         uint64_t num;
1566                                         if ((res = ebml_read_uint(matroska,
1567                                                           &id, &num)) < 0)
1568                                             break;
1569                                         idx.track = num;
1570                                         break;
1571                                     }
1572
1573                                         /* position in file */
1574                                     case MATROSKA_ID_CUECLUSTERPOSITION: {
1575                                         uint64_t num;
1576                                         if ((res = ebml_read_uint(matroska,
1577                                                           &id, &num)) < 0)
1578                                             break;
1579                                         idx.pos = num+matroska->segment_start;
1580                                         break;
1581                                     }
1582
1583                                     default:
1584                                         av_log(matroska->ctx, AV_LOG_INFO,
1585                                                "Unknown entry 0x%x in "
1586                                                "CuesTrackPositions\n", id);
1587                                         /* fall-through */
1588
1589                                     case EBML_ID_VOID:
1590                                         res = ebml_read_skip(matroska);
1591                                         break;
1592                                 }
1593
1594                                 if (matroska->level_up) {
1595                                     matroska->level_up--;
1596                                     break;
1597                                 }
1598                             }
1599
1600                             break;
1601
1602                         default:
1603                             av_log(matroska->ctx, AV_LOG_INFO,
1604                                    "Unknown entry 0x%x in cuespoint "
1605                                    "index\n", id);
1606                             /* fall-through */
1607
1608                         case EBML_ID_VOID:
1609                             res = ebml_read_skip(matroska);
1610                             break;
1611                     }
1612
1613                     if (matroska->level_up) {
1614                         matroska->level_up--;
1615                         break;
1616                     }
1617                 }
1618
1619                 /* so let's see if we got what we wanted */
1620                 if (idx.pos   != (uint64_t) -1 &&
1621                     idx.time  != (uint64_t) -1 &&
1622                     idx.track != (uint16_t) -1) {
1623                     if (matroska->num_indexes % 32 == 0) {
1624                         /* re-allocate bigger index */
1625                         matroska->index =
1626                             av_realloc(matroska->index,
1627                                        (matroska->num_indexes + 32) *
1628                                        sizeof(MatroskaDemuxIndex));
1629                     }
1630                     matroska->index[matroska->num_indexes] = idx;
1631                     matroska->num_indexes++;
1632                 }
1633                 break;
1634
1635             default:
1636                 av_log(matroska->ctx, AV_LOG_INFO,
1637                        "Unknown entry 0x%x in cues header\n", id);
1638                 /* fall-through */
1639
1640             case EBML_ID_VOID:
1641                 res = ebml_read_skip(matroska);
1642                 break;
1643         }
1644
1645         if (matroska->level_up) {
1646             matroska->level_up--;
1647             break;
1648         }
1649     }
1650
1651     return res;
1652 }
1653
1654 static int
1655 matroska_parse_metadata (MatroskaDemuxContext *matroska)
1656 {
1657     int res = 0;
1658     uint32_t id;
1659
1660     while (res == 0) {
1661         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1662             res = AVERROR(EIO);
1663             break;
1664         } else if (matroska->level_up) {
1665             matroska->level_up--;
1666             break;
1667         }
1668
1669         switch (id) {
1670             /* Hm, this is unsupported... */
1671             default:
1672                 av_log(matroska->ctx, AV_LOG_INFO,
1673                        "Unknown entry 0x%x in metadata header\n", id);
1674                 /* fall-through */
1675
1676             case EBML_ID_VOID:
1677                 res = ebml_read_skip(matroska);
1678                 break;
1679         }
1680
1681         if (matroska->level_up) {
1682             matroska->level_up--;
1683             break;
1684         }
1685     }
1686
1687     return res;
1688 }
1689
1690 static int
1691 matroska_parse_seekhead (MatroskaDemuxContext *matroska)
1692 {
1693     int res = 0;
1694     uint32_t id;
1695
1696     av_log(matroska->ctx, AV_LOG_DEBUG, "parsing seekhead...\n");
1697
1698     while (res == 0) {
1699         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1700             res = AVERROR(EIO);
1701             break;
1702         } else if (matroska->level_up) {
1703             matroska->level_up--;
1704             break;
1705         }
1706
1707         switch (id) {
1708             case MATROSKA_ID_SEEKENTRY: {
1709                 uint32_t seek_id = 0, peek_id_cache = 0;
1710                 uint64_t seek_pos = (uint64_t) -1, t;
1711
1712                 if ((res = ebml_read_master(matroska, &id)) < 0)
1713                     break;
1714
1715                 while (res == 0) {
1716                     if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1717                         res = AVERROR(EIO);
1718                         break;
1719                     } else if (matroska->level_up) {
1720                         matroska->level_up--;
1721                         break;
1722                     }
1723
1724                     switch (id) {
1725                         case MATROSKA_ID_SEEKID:
1726                             res = ebml_read_uint(matroska, &id, &t);
1727                             seek_id = t;
1728                             break;
1729
1730                         case MATROSKA_ID_SEEKPOSITION:
1731                             res = ebml_read_uint(matroska, &id, &seek_pos);
1732                             break;
1733
1734                         default:
1735                             av_log(matroska->ctx, AV_LOG_INFO,
1736                                    "Unknown seekhead ID 0x%x\n", id);
1737                             /* fall-through */
1738
1739                         case EBML_ID_VOID:
1740                             res = ebml_read_skip(matroska);
1741                             break;
1742                     }
1743
1744                     if (matroska->level_up) {
1745                         matroska->level_up--;
1746                         break;
1747                     }
1748                 }
1749
1750                 if (!seek_id || seek_pos == (uint64_t) -1) {
1751                     av_log(matroska->ctx, AV_LOG_INFO,
1752                            "Incomplete seekhead entry (0x%x/%"PRIu64")\n",
1753                            seek_id, seek_pos);
1754                     break;
1755                 }
1756
1757                 switch (seek_id) {
1758                     case MATROSKA_ID_CUES:
1759                     case MATROSKA_ID_TAGS: {
1760                         uint32_t level_up = matroska->level_up;
1761                         offset_t before_pos;
1762                         uint64_t length;
1763                         MatroskaLevel level;
1764
1765                         /* remember the peeked ID and the current position */
1766                         peek_id_cache = matroska->peek_id;
1767                         before_pos = url_ftell(matroska->ctx->pb);
1768
1769                         /* seek */
1770                         if ((res = ebml_read_seek(matroska, seek_pos +
1771                                                matroska->segment_start)) < 0)
1772                             return res;
1773
1774                         /* we don't want to lose our seekhead level, so we add
1775                          * a dummy. This is a crude hack. */
1776                         if (matroska->num_levels == EBML_MAX_DEPTH) {
1777                             av_log(matroska->ctx, AV_LOG_INFO,
1778                                    "Max EBML element depth (%d) reached, "
1779                                    "cannot parse further.\n", EBML_MAX_DEPTH);
1780                             return AVERROR_UNKNOWN;
1781                         }
1782
1783                         level.start = 0;
1784                         level.length = (uint64_t)-1;
1785                         matroska->levels[matroska->num_levels] = level;
1786                         matroska->num_levels++;
1787
1788                         /* check ID */
1789                         if (!(id = ebml_peek_id (matroska,
1790                                                  &matroska->level_up)))
1791                             goto finish;
1792                         if (id != seek_id) {
1793                             av_log(matroska->ctx, AV_LOG_INFO,
1794                                    "We looked for ID=0x%x but got "
1795                                    "ID=0x%x (pos=%"PRIu64")",
1796                                    seek_id, id, seek_pos +
1797                                    matroska->segment_start);
1798                             goto finish;
1799                         }
1800
1801                         /* read master + parse */
1802                         if ((res = ebml_read_master(matroska, &id)) < 0)
1803                             goto finish;
1804                         switch (id) {
1805                             case MATROSKA_ID_CUES:
1806                                 if (!(res = matroska_parse_index(matroska)) ||
1807                                     url_feof(matroska->ctx->pb)) {
1808                                     matroska->index_parsed = 1;
1809                                     res = 0;
1810                                 }
1811                                 break;
1812                             case MATROSKA_ID_TAGS:
1813                                 if (!(res = matroska_parse_metadata(matroska)) ||
1814                                    url_feof(matroska->ctx->pb)) {
1815                                     matroska->metadata_parsed = 1;
1816                                     res = 0;
1817                                 }
1818                                 break;
1819                         }
1820
1821                     finish:
1822                         /* remove dummy level */
1823                         while (matroska->num_levels) {
1824                             matroska->num_levels--;
1825                             length =
1826                                 matroska->levels[matroska->num_levels].length;
1827                             if (length == (uint64_t)-1)
1828                                 break;
1829                         }
1830
1831                         /* seek back */
1832                         if ((res = ebml_read_seek(matroska, before_pos)) < 0)
1833                             return res;
1834                         matroska->peek_id = peek_id_cache;
1835                         matroska->level_up = level_up;
1836                         break;
1837                     }
1838
1839                     default:
1840                         av_log(matroska->ctx, AV_LOG_INFO,
1841                                "Ignoring seekhead entry for ID=0x%x\n",
1842                                seek_id);
1843                         break;
1844                 }
1845
1846                 break;
1847             }
1848
1849             default:
1850                 av_log(matroska->ctx, AV_LOG_INFO,
1851                        "Unknown seekhead ID 0x%x\n", id);
1852                 /* fall-through */
1853
1854             case EBML_ID_VOID:
1855                 res = ebml_read_skip(matroska);
1856                 break;
1857         }
1858
1859         if (matroska->level_up) {
1860             matroska->level_up--;
1861             break;
1862         }
1863     }
1864
1865     return res;
1866 }
1867
1868 static int
1869 matroska_parse_attachments(AVFormatContext *s)
1870 {
1871     MatroskaDemuxContext *matroska = s->priv_data;
1872     int res = 0;
1873     uint32_t id;
1874
1875     av_log(matroska->ctx, AV_LOG_DEBUG, "parsing attachments...\n");
1876
1877     while (res == 0) {
1878         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1879             res = AVERROR(EIO);
1880             break;
1881         } else if (matroska->level_up) {
1882             matroska->level_up--;
1883             break;
1884         }
1885
1886         switch (id) {
1887         case MATROSKA_ID_ATTACHEDFILE: {
1888             char* name = NULL;
1889             char* mime = NULL;
1890             uint8_t* data = NULL;
1891             int i, data_size = 0;
1892             AVStream *st;
1893
1894             if ((res = ebml_read_master(matroska, &id)) < 0)
1895                 break;
1896
1897             while (res == 0) {
1898                 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1899                     res = AVERROR(EIO);
1900                     break;
1901                 } else if (matroska->level_up) {
1902                     matroska->level_up--;
1903                     break;
1904                 }
1905
1906                 switch (id) {
1907                 case MATROSKA_ID_FILENAME:
1908                     res = ebml_read_utf8 (matroska, &id, &name);
1909                     break;
1910
1911                 case MATROSKA_ID_FILEMIMETYPE:
1912                     res = ebml_read_ascii (matroska, &id, &mime);
1913                     break;
1914
1915                 case MATROSKA_ID_FILEDATA:
1916                     res = ebml_read_binary(matroska, &id, &data, &data_size);
1917                     break;
1918
1919                 default:
1920                     av_log(matroska->ctx, AV_LOG_INFO,
1921                            "Unknown attachedfile ID 0x%x\n", id);
1922                 case EBML_ID_VOID:
1923                     res = ebml_read_skip(matroska);
1924                     break;
1925                 }
1926
1927                 if (matroska->level_up) {
1928                     matroska->level_up--;
1929                     break;
1930                 }
1931             }
1932
1933             if (!(name && mime && data && data_size > 0)) {
1934                 av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
1935                 break;
1936             }
1937
1938             st = av_new_stream(s, matroska->num_streams++);
1939             if (st == NULL)
1940                 return AVERROR(ENOMEM);
1941             st->filename = av_strdup(name);
1942             st->codec->codec_id = CODEC_ID_NONE;
1943             st->codec->codec_type = CODEC_TYPE_ATTACHMENT;
1944             st->codec->extradata = av_malloc(data_size);
1945             if(st->codec->extradata == NULL)
1946                 return AVERROR(ENOMEM);
1947             st->codec->extradata_size = data_size;
1948             memcpy(st->codec->extradata, data, data_size);
1949
1950             for (i=0; ff_mkv_mime_tags[i].id != CODEC_ID_NONE; i++) {
1951                 if (!strncmp(ff_mkv_mime_tags[i].str, mime,
1952                              strlen(ff_mkv_mime_tags[i].str))) {
1953                     st->codec->codec_id = ff_mkv_mime_tags[i].id;
1954                     break;
1955                 }
1956             }
1957
1958             av_log(matroska->ctx, AV_LOG_DEBUG, "new attachment: %s, %s, size %d \n", name, mime, data_size);
1959             break;
1960         }
1961
1962         default:
1963             av_log(matroska->ctx, AV_LOG_INFO,
1964                    "Unknown attachments ID 0x%x\n", id);
1965             /* fall-through */
1966
1967         case EBML_ID_VOID:
1968             res = ebml_read_skip(matroska);
1969             break;
1970         }
1971
1972         if (matroska->level_up) {
1973             matroska->level_up--;
1974             break;
1975         }
1976     }
1977
1978     return res;
1979 }
1980
1981 #define ARRAY_SIZE(x)  (sizeof(x)/sizeof(*x))
1982
1983 static int
1984 matroska_aac_profile (char *codec_id)
1985 {
1986     static const char *aac_profiles[] = {
1987         "MAIN", "LC", "SSR"
1988     };
1989     int profile;
1990
1991     for (profile=0; profile<ARRAY_SIZE(aac_profiles); profile++)
1992         if (strstr(codec_id, aac_profiles[profile]))
1993             break;
1994     return profile + 1;
1995 }
1996
1997 static int
1998 matroska_aac_sri (int samplerate)
1999 {
2000     static const int aac_sample_rates[] = {
2001         96000, 88200, 64000, 48000, 44100, 32000,
2002         24000, 22050, 16000, 12000, 11025,  8000,
2003     };
2004     int sri;
2005
2006     for (sri=0; sri<ARRAY_SIZE(aac_sample_rates); sri++)
2007         if (aac_sample_rates[sri] == samplerate)
2008             break;
2009     return sri;
2010 }
2011
2012 static int
2013 matroska_read_header (AVFormatContext    *s,
2014                       AVFormatParameters *ap)
2015 {
2016     MatroskaDemuxContext *matroska = s->priv_data;
2017     char *doctype;
2018     int version, last_level, res = 0;
2019     uint32_t id;
2020
2021     matroska->ctx = s;
2022
2023     /* First read the EBML header. */
2024     doctype = NULL;
2025     if ((res = ebml_read_header(matroska, &doctype, &version)) < 0)
2026         return res;
2027     if ((doctype == NULL) || strcmp(doctype, "matroska")) {
2028         av_log(matroska->ctx, AV_LOG_ERROR,
2029                "Wrong EBML doctype ('%s' != 'matroska').\n",
2030                doctype ? doctype : "(none)");
2031         if (doctype)
2032             av_free(doctype);
2033         return AVERROR_NOFMT;
2034     }
2035     av_free(doctype);
2036     if (version > 2) {
2037         av_log(matroska->ctx, AV_LOG_ERROR,
2038                "Matroska demuxer version 2 too old for file version %d\n",
2039                version);
2040         return AVERROR_NOFMT;
2041     }
2042
2043     /* The next thing is a segment. */
2044     while (1) {
2045         if (!(id = ebml_peek_id(matroska, &last_level)))
2046             return AVERROR(EIO);
2047         if (id == MATROSKA_ID_SEGMENT)
2048             break;
2049
2050         /* oi! */
2051         av_log(matroska->ctx, AV_LOG_INFO,
2052                "Expected a Segment ID (0x%x), but received 0x%x!\n",
2053                MATROSKA_ID_SEGMENT, id);
2054         if ((res = ebml_read_skip(matroska)) < 0)
2055             return res;
2056     }
2057
2058     /* We now have a Matroska segment.
2059      * Seeks are from the beginning of the segment,
2060      * after the segment ID/length. */
2061     if ((res = ebml_read_master(matroska, &id)) < 0)
2062         return res;
2063     matroska->segment_start = url_ftell(s->pb);
2064
2065     matroska->time_scale = 1000000;
2066     /* we've found our segment, start reading the different contents in here */
2067     while (res == 0) {
2068         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2069             res = AVERROR(EIO);
2070             break;
2071         } else if (matroska->level_up) {
2072             matroska->level_up--;
2073             break;
2074         }
2075
2076         switch (id) {
2077             /* stream info */
2078             case MATROSKA_ID_INFO: {
2079                 if ((res = ebml_read_master(matroska, &id)) < 0)
2080                     break;
2081                 res = matroska_parse_info(matroska);
2082                 break;
2083             }
2084
2085             /* track info headers */
2086             case MATROSKA_ID_TRACKS: {
2087                 if ((res = ebml_read_master(matroska, &id)) < 0)
2088                     break;
2089                 res = matroska_parse_tracks(matroska);
2090                 break;
2091             }
2092
2093             /* stream index */
2094             case MATROSKA_ID_CUES: {
2095                 if (!matroska->index_parsed) {
2096                     if ((res = ebml_read_master(matroska, &id)) < 0)
2097                         break;
2098                     res = matroska_parse_index(matroska);
2099                 } else
2100                     res = ebml_read_skip(matroska);
2101                 break;
2102             }
2103
2104             /* metadata */
2105             case MATROSKA_ID_TAGS: {
2106                 if (!matroska->metadata_parsed) {
2107                     if ((res = ebml_read_master(matroska, &id)) < 0)
2108                         break;
2109                     res = matroska_parse_metadata(matroska);
2110                 } else
2111                     res = ebml_read_skip(matroska);
2112                 break;
2113             }
2114
2115             /* file index (if seekable, seek to Cues/Tags to parse it) */
2116             case MATROSKA_ID_SEEKHEAD: {
2117                 if ((res = ebml_read_master(matroska, &id)) < 0)
2118                     break;
2119                 res = matroska_parse_seekhead(matroska);
2120                 break;
2121             }
2122
2123             case MATROSKA_ID_ATTACHMENTS: {
2124                 if ((res = ebml_read_master(matroska, &id)) < 0)
2125                     break;
2126                 res = matroska_parse_attachments(s);
2127                 break;
2128             }
2129
2130             case MATROSKA_ID_CLUSTER: {
2131                 /* Do not read the master - this will be done in the next
2132                  * call to matroska_read_packet. */
2133                 res = 1;
2134                 break;
2135             }
2136
2137             default:
2138                 av_log(matroska->ctx, AV_LOG_INFO,
2139                        "Unknown matroska file header ID 0x%x\n", id);
2140             /* fall-through */
2141
2142             case EBML_ID_VOID:
2143                 res = ebml_read_skip(matroska);
2144                 break;
2145         }
2146
2147         if (matroska->level_up) {
2148             matroska->level_up--;
2149             break;
2150         }
2151     }
2152
2153     /* Have we found a cluster? */
2154     if (ebml_peek_id(matroska, NULL) == MATROSKA_ID_CLUSTER) {
2155         int i, j;
2156         MatroskaTrack *track;
2157         AVStream *st;
2158
2159         for (i = 0; i < matroska->num_tracks; i++) {
2160             enum CodecID codec_id = CODEC_ID_NONE;
2161             uint8_t *extradata = NULL;
2162             int extradata_size = 0;
2163             int extradata_offset = 0;
2164             track = matroska->tracks[i];
2165             track->stream_index = -1;
2166
2167             /* Apply some sanity checks. */
2168             if (track->codec_id == NULL)
2169                 continue;
2170
2171             for(j=0; ff_mkv_codec_tags[j].id != CODEC_ID_NONE; j++){
2172                 if(!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
2173                             strlen(ff_mkv_codec_tags[j].str))){
2174                     codec_id= ff_mkv_codec_tags[j].id;
2175                     break;
2176                 }
2177             }
2178
2179             /* Set the FourCC from the CodecID. */
2180             /* This is the MS compatibility mode which stores a
2181              * BITMAPINFOHEADER in the CodecPrivate. */
2182             if (!strcmp(track->codec_id,
2183                         MATROSKA_CODEC_ID_VIDEO_VFW_FOURCC) &&
2184                 (track->codec_priv_size >= 40) &&
2185                 (track->codec_priv != NULL)) {
2186                 MatroskaVideoTrack *vtrack = (MatroskaVideoTrack *) track;
2187
2188                 /* Offset of biCompression. Stored in LE. */
2189                 vtrack->fourcc = AV_RL32(track->codec_priv + 16);
2190                 codec_id = codec_get_id(codec_bmp_tags, vtrack->fourcc);
2191
2192             }
2193
2194             /* This is the MS compatibility mode which stores a
2195              * WAVEFORMATEX in the CodecPrivate. */
2196             else if (!strcmp(track->codec_id,
2197                              MATROSKA_CODEC_ID_AUDIO_ACM) &&
2198                 (track->codec_priv_size >= 18) &&
2199                 (track->codec_priv != NULL)) {
2200                 uint16_t tag;
2201
2202                 /* Offset of wFormatTag. Stored in LE. */
2203                 tag = AV_RL16(track->codec_priv);
2204                 codec_id = codec_get_id(codec_wav_tags, tag);
2205
2206             }
2207
2208             else if (codec_id == CODEC_ID_AAC && !track->codec_priv_size) {
2209                 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *) track;
2210                 int profile = matroska_aac_profile(track->codec_id);
2211                 int sri = matroska_aac_sri(audiotrack->internal_samplerate);
2212                 extradata = av_malloc(5);
2213                 if (extradata == NULL)
2214                     return AVERROR(ENOMEM);
2215                 extradata[0] = (profile << 3) | ((sri&0x0E) >> 1);
2216                 extradata[1] = ((sri&0x01) << 7) | (audiotrack->channels<<3);
2217                 if (strstr(track->codec_id, "SBR")) {
2218                     sri = matroska_aac_sri(audiotrack->samplerate);
2219                     extradata[2] = 0x56;
2220                     extradata[3] = 0xE5;
2221                     extradata[4] = 0x80 | (sri<<3);
2222                     extradata_size = 5;
2223                 } else {
2224                     extradata_size = 2;
2225                 }
2226             }
2227
2228             else if (codec_id == CODEC_ID_TTA) {
2229                 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *) track;
2230                 ByteIOContext b;
2231                 extradata_size = 30;
2232                 extradata = av_mallocz(extradata_size);
2233                 if (extradata == NULL)
2234                     return AVERROR(ENOMEM);
2235                 init_put_byte(&b, extradata, extradata_size, 1,
2236                               NULL, NULL, NULL, NULL);
2237                 put_buffer(&b, "TTA1", 4);
2238                 put_le16(&b, 1);
2239                 put_le16(&b, audiotrack->channels);
2240                 put_le16(&b, audiotrack->bitdepth);
2241                 put_le32(&b, audiotrack->samplerate);
2242                 put_le32(&b, matroska->ctx->duration * audiotrack->samplerate);
2243             }
2244
2245             else if (codec_id == CODEC_ID_RV10 || codec_id == CODEC_ID_RV20 ||
2246                      codec_id == CODEC_ID_RV30 || codec_id == CODEC_ID_RV40) {
2247                 extradata_offset = 26;
2248                 track->codec_priv_size -= extradata_offset;
2249             }
2250
2251             else if (codec_id == CODEC_ID_RA_144) {
2252                 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)track;
2253                 audiotrack->samplerate = 8000;
2254                 audiotrack->channels = 1;
2255             }
2256
2257             else if (codec_id == CODEC_ID_RA_288 ||
2258                      codec_id == CODEC_ID_COOK ||
2259                      codec_id == CODEC_ID_ATRAC3) {
2260                 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)track;
2261                 ByteIOContext b;
2262
2263                 init_put_byte(&b, track->codec_priv, track->codec_priv_size, 0,
2264                               NULL, NULL, NULL, NULL);
2265                 url_fskip(&b, 24);
2266                 audiotrack->coded_framesize = get_be32(&b);
2267                 url_fskip(&b, 12);
2268                 audiotrack->sub_packet_h    = get_be16(&b);
2269                 audiotrack->frame_size      = get_be16(&b);
2270                 audiotrack->sub_packet_size = get_be16(&b);
2271                 audiotrack->buf = av_malloc(audiotrack->frame_size * audiotrack->sub_packet_h);
2272                 if (codec_id == CODEC_ID_RA_288) {
2273                     audiotrack->block_align = audiotrack->coded_framesize;
2274                     track->codec_priv_size = 0;
2275                 } else {
2276                     audiotrack->block_align = audiotrack->sub_packet_size;
2277                     extradata_offset = 78;
2278                     track->codec_priv_size -= extradata_offset;
2279                 }
2280             }
2281
2282             if (codec_id == CODEC_ID_NONE) {
2283                 av_log(matroska->ctx, AV_LOG_INFO,
2284                        "Unknown/unsupported CodecID %s.\n",
2285                        track->codec_id);
2286             }
2287
2288             track->stream_index = matroska->num_streams;
2289
2290             matroska->num_streams++;
2291             st = av_new_stream(s, track->stream_index);
2292             if (st == NULL)
2293                 return AVERROR(ENOMEM);
2294             av_set_pts_info(st, 64, matroska->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
2295
2296             st->codec->codec_id = codec_id;
2297             st->start_time = 0;
2298             if (strcmp(track->language, "und"))
2299                 strcpy(st->language, track->language);
2300
2301             if (track->flags & MATROSKA_TRACK_DEFAULT)
2302                 st->disposition |= AV_DISPOSITION_DEFAULT;
2303
2304             if (track->default_duration)
2305                 av_reduce(&st->codec->time_base.num, &st->codec->time_base.den,
2306                           track->default_duration, 1000000000, 30000);
2307
2308             if(extradata){
2309                 st->codec->extradata = extradata;
2310                 st->codec->extradata_size = extradata_size;
2311             } else if(track->codec_priv && track->codec_priv_size > 0){
2312                 st->codec->extradata = av_malloc(track->codec_priv_size);
2313                 if(st->codec->extradata == NULL)
2314                     return AVERROR(ENOMEM);
2315                 st->codec->extradata_size = track->codec_priv_size;
2316                 memcpy(st->codec->extradata,track->codec_priv+extradata_offset,
2317                        track->codec_priv_size);
2318             }
2319
2320             if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
2321                 MatroskaVideoTrack *videotrack = (MatroskaVideoTrack *)track;
2322
2323                 st->codec->codec_type = CODEC_TYPE_VIDEO;
2324                 st->codec->codec_tag = videotrack->fourcc;
2325                 st->codec->width = videotrack->pixel_width;
2326                 st->codec->height = videotrack->pixel_height;
2327                 if (videotrack->display_width == 0)
2328                     videotrack->display_width= videotrack->pixel_width;
2329                 if (videotrack->display_height == 0)
2330                     videotrack->display_height= videotrack->pixel_height;
2331                 av_reduce(&st->codec->sample_aspect_ratio.num,
2332                           &st->codec->sample_aspect_ratio.den,
2333                           st->codec->height * videotrack->display_width,
2334                           st->codec-> width * videotrack->display_height,
2335                           255);
2336                 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2337             } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
2338                 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)track;
2339
2340                 st->codec->codec_type = CODEC_TYPE_AUDIO;
2341                 st->codec->sample_rate = audiotrack->samplerate;
2342                 st->codec->channels = audiotrack->channels;
2343                 st->codec->block_align = audiotrack->block_align;
2344             } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
2345                 st->codec->codec_type = CODEC_TYPE_SUBTITLE;
2346             }
2347
2348             /* What do we do with private data? E.g. for Vorbis. */
2349         }
2350         res = 0;
2351     }
2352
2353     if (matroska->index_parsed) {
2354         int i, track, stream;
2355         for (i=0; i<matroska->num_indexes; i++) {
2356             MatroskaDemuxIndex *idx = &matroska->index[i];
2357             track = matroska_find_track_by_num(matroska, idx->track);
2358             stream = matroska->tracks[track]->stream_index;
2359             if (stream >= 0 && stream < matroska->ctx->nb_streams)
2360                 av_add_index_entry(matroska->ctx->streams[stream],
2361                                    idx->pos, idx->time/matroska->time_scale,
2362                                    0, 0, AVINDEX_KEYFRAME);
2363         }
2364     }
2365
2366     return res;
2367 }
2368
2369 static int
2370 matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, int size,
2371                      int64_t pos, uint64_t cluster_time, uint64_t duration,
2372                      int is_keyframe, int is_bframe)
2373 {
2374     int res = 0;
2375     int track;
2376     AVStream *st;
2377     AVPacket *pkt;
2378     uint8_t *origdata = data;
2379     int16_t block_time;
2380     uint32_t *lace_size = NULL;
2381     int n, flags, laces = 0;
2382     uint64_t num;
2383     int stream_index;
2384
2385     /* first byte(s): tracknum */
2386     if ((n = matroska_ebmlnum_uint(data, size, &num)) < 0) {
2387         av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
2388         av_free(origdata);
2389         return res;
2390     }
2391     data += n;
2392     size -= n;
2393
2394     /* fetch track from num */
2395     track = matroska_find_track_by_num(matroska, num);
2396     if (size <= 3 || track < 0 || track >= matroska->num_tracks) {
2397         av_log(matroska->ctx, AV_LOG_INFO,
2398                "Invalid stream %d or size %u\n", track, size);
2399         av_free(origdata);
2400         return res;
2401     }
2402     stream_index = matroska->tracks[track]->stream_index;
2403     if (stream_index < 0 || stream_index >= matroska->ctx->nb_streams) {
2404         av_free(origdata);
2405         return res;
2406     }
2407     st = matroska->ctx->streams[stream_index];
2408     if (st->discard >= AVDISCARD_ALL) {
2409         av_free(origdata);
2410         return res;
2411     }
2412     if (duration == AV_NOPTS_VALUE)
2413         duration = matroska->tracks[track]->default_duration / matroska->time_scale;
2414
2415     /* block_time (relative to cluster time) */
2416     block_time = AV_RB16(data);
2417     data += 2;
2418     flags = *data++;
2419     size -= 3;
2420     if (is_keyframe == -1)
2421         is_keyframe = flags & 0x80 ? PKT_FLAG_KEY : 0;
2422
2423     if (matroska->skip_to_keyframe) {
2424         if (!is_keyframe || st != matroska->skip_to_stream) {
2425             av_free(origdata);
2426             return res;
2427         }
2428         matroska->skip_to_keyframe = 0;
2429     }
2430
2431     switch ((flags & 0x06) >> 1) {
2432         case 0x0: /* no lacing */
2433             laces = 1;
2434             lace_size = av_mallocz(sizeof(int));
2435             lace_size[0] = size;
2436             break;
2437
2438         case 0x1: /* xiph lacing */
2439         case 0x2: /* fixed-size lacing */
2440         case 0x3: /* EBML lacing */
2441             if (size == 0) {
2442                 res = -1;
2443                 break;
2444             }
2445             laces = (*data) + 1;
2446             data += 1;
2447             size -= 1;
2448             lace_size = av_mallocz(laces * sizeof(int));
2449
2450             switch ((flags & 0x06) >> 1) {
2451                 case 0x1: /* xiph lacing */ {
2452                     uint8_t temp;
2453                     uint32_t total = 0;
2454                     for (n = 0; res == 0 && n < laces - 1; n++) {
2455                         while (1) {
2456                             if (size == 0) {
2457                                 res = -1;
2458                                 break;
2459                             }
2460                             temp = *data;
2461                             lace_size[n] += temp;
2462                             data += 1;
2463                             size -= 1;
2464                             if (temp != 0xff)
2465                                 break;
2466                         }
2467                         total += lace_size[n];
2468                     }
2469                     lace_size[n] = size - total;
2470                     break;
2471                 }
2472
2473                 case 0x2: /* fixed-size lacing */
2474                     for (n = 0; n < laces; n++)
2475                         lace_size[n] = size / laces;
2476                     break;
2477
2478                 case 0x3: /* EBML lacing */ {
2479                     uint32_t total;
2480                     n = matroska_ebmlnum_uint(data, size, &num);
2481                     if (n < 0) {
2482                         av_log(matroska->ctx, AV_LOG_INFO,
2483                                "EBML block data error\n");
2484                         break;
2485                     }
2486                     data += n;
2487                     size -= n;
2488                     total = lace_size[0] = num;
2489                     for (n = 1; res == 0 && n < laces - 1; n++) {
2490                         int64_t snum;
2491                         int r;
2492                         r = matroska_ebmlnum_sint (data, size, &snum);
2493                         if (r < 0) {
2494                             av_log(matroska->ctx, AV_LOG_INFO,
2495                                    "EBML block data error\n");
2496                             break;
2497                         }
2498                         data += r;
2499                         size -= r;
2500                         lace_size[n] = lace_size[n - 1] + snum;
2501                         total += lace_size[n];
2502                     }
2503                     lace_size[n] = size - total;
2504                     break;
2505                 }
2506             }
2507             break;
2508     }
2509
2510     if (res == 0) {
2511         uint64_t timecode = AV_NOPTS_VALUE;
2512
2513         if (cluster_time != (uint64_t)-1
2514             && (block_time >= 0 || cluster_time >= -block_time))
2515             timecode = cluster_time + block_time;
2516
2517         for (n = 0; n < laces; n++) {
2518             if (st->codec->codec_id == CODEC_ID_RA_288 ||
2519                 st->codec->codec_id == CODEC_ID_COOK ||
2520                 st->codec->codec_id == CODEC_ID_ATRAC3) {
2521                 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)matroska->tracks[track];
2522                 int a = st->codec->block_align;
2523                 int sps = audiotrack->sub_packet_size;
2524                 int cfs = audiotrack->coded_framesize;
2525                 int h = audiotrack->sub_packet_h;
2526                 int y = audiotrack->sub_packet_cnt;
2527                 int w = audiotrack->frame_size;
2528                 int x;
2529
2530                 if (!audiotrack->pkt_cnt) {
2531                     if (st->codec->codec_id == CODEC_ID_RA_288)
2532                         for (x=0; x<h/2; x++)
2533                             memcpy(audiotrack->buf+x*2*w+y*cfs,
2534                                    data+x*cfs, cfs);
2535                     else
2536                         for (x=0; x<w/sps; x++)
2537                             memcpy(audiotrack->buf+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), data+x*sps, sps);
2538
2539                     if (++audiotrack->sub_packet_cnt >= h) {
2540                         audiotrack->sub_packet_cnt = 0;
2541                         audiotrack->pkt_cnt = h*w / a;
2542                     }
2543                 }
2544                 while (audiotrack->pkt_cnt) {
2545                     pkt = av_mallocz(sizeof(AVPacket));
2546                     av_new_packet(pkt, a);
2547                     memcpy(pkt->data, audiotrack->buf
2548                            + a * (h*w / a - audiotrack->pkt_cnt--), a);
2549                     pkt->pos = pos;
2550                     pkt->stream_index = stream_index;
2551                     matroska_queue_packet(matroska, pkt);
2552                 }
2553             } else {
2554                 int offset = 0;
2555
2556                 pkt = av_mallocz(sizeof(AVPacket));
2557                 /* XXX: prevent data copy... */
2558                 if (av_new_packet(pkt, lace_size[n]-offset) < 0) {
2559                     res = AVERROR(ENOMEM);
2560                     n = laces-1;
2561                     break;
2562                 }
2563                 memcpy (pkt->data, data+offset, lace_size[n]-offset);
2564
2565                 if (n == 0)
2566                     pkt->flags = is_keyframe;
2567                 pkt->stream_index = stream_index;
2568
2569                 pkt->pts = timecode;
2570                 pkt->pos = pos;
2571                 pkt->duration = duration;
2572
2573                 matroska_queue_packet(matroska, pkt);
2574             }
2575
2576             if (timecode != AV_NOPTS_VALUE)
2577                 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
2578             data += lace_size[n];
2579         }
2580     }
2581
2582     av_free(lace_size);
2583     av_free(origdata);
2584     return res;
2585 }
2586
2587 static int
2588 matroska_parse_blockgroup (MatroskaDemuxContext *matroska,
2589                            uint64_t              cluster_time)
2590 {
2591     int res = 0;
2592     uint32_t id;
2593     int is_bframe = 0;
2594     int is_keyframe = PKT_FLAG_KEY, last_num_packets = matroska->num_packets;
2595     uint64_t duration = AV_NOPTS_VALUE;
2596     uint8_t *data;
2597     int size = 0;
2598     int64_t pos = 0;
2599
2600     av_log(matroska->ctx, AV_LOG_DEBUG, "parsing blockgroup...\n");
2601
2602     while (res == 0) {
2603         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2604             res = AVERROR(EIO);
2605             break;
2606         } else if (matroska->level_up) {
2607             matroska->level_up--;
2608             break;
2609         }
2610
2611         switch (id) {
2612             /* one block inside the group. Note, block parsing is one
2613              * of the harder things, so this code is a bit complicated.
2614              * See http://www.matroska.org/ for documentation. */
2615             case MATROSKA_ID_BLOCK: {
2616                 pos = url_ftell(matroska->ctx->pb);
2617                 res = ebml_read_binary(matroska, &id, &data, &size);
2618                 break;
2619             }
2620
2621             case MATROSKA_ID_BLOCKDURATION: {
2622                 if ((res = ebml_read_uint(matroska, &id, &duration)) < 0)
2623                     break;
2624                 break;
2625             }
2626
2627             case MATROSKA_ID_BLOCKREFERENCE: {
2628                 int64_t num;
2629                 /* We've found a reference, so not even the first frame in
2630                  * the lace is a key frame. */
2631                 is_keyframe = 0;
2632                 if (last_num_packets != matroska->num_packets)
2633                     matroska->packets[last_num_packets]->flags = 0;
2634                 if ((res = ebml_read_sint(matroska, &id, &num)) < 0)
2635                     break;
2636                 if (num > 0)
2637                     is_bframe = 1;
2638                 break;
2639             }
2640
2641             default:
2642                 av_log(matroska->ctx, AV_LOG_INFO,
2643                        "Unknown entry 0x%x in blockgroup data\n", id);
2644                 /* fall-through */
2645
2646             case EBML_ID_VOID:
2647                 res = ebml_read_skip(matroska);
2648                 break;
2649         }
2650
2651         if (matroska->level_up) {
2652             matroska->level_up--;
2653             break;
2654         }
2655     }
2656
2657     if (res)
2658         return res;
2659
2660     if (size > 0)
2661         res = matroska_parse_block(matroska, data, size, pos, cluster_time,
2662                                    duration, is_keyframe, is_bframe);
2663
2664     return res;
2665 }
2666
2667 static int
2668 matroska_parse_cluster (MatroskaDemuxContext *matroska)
2669 {
2670     int res = 0;
2671     uint32_t id;
2672     uint64_t cluster_time = 0;
2673     uint8_t *data;
2674     int64_t pos;
2675     int size;
2676
2677     av_log(matroska->ctx, AV_LOG_DEBUG,
2678            "parsing cluster at %"PRId64"\n", url_ftell(matroska->ctx->pb));
2679
2680     while (res == 0) {
2681         if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2682             res = AVERROR(EIO);
2683             break;
2684         } else if (matroska->level_up) {
2685             matroska->level_up--;
2686             break;
2687         }
2688
2689         switch (id) {
2690             /* cluster timecode */
2691             case MATROSKA_ID_CLUSTERTIMECODE: {
2692                 uint64_t num;
2693                 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
2694                     break;
2695                 cluster_time = num;
2696                 break;
2697             }
2698
2699                 /* a group of blocks inside a cluster */
2700             case MATROSKA_ID_BLOCKGROUP:
2701                 if ((res = ebml_read_master(matroska, &id)) < 0)
2702                     break;
2703                 res = matroska_parse_blockgroup(matroska, cluster_time);
2704                 break;
2705
2706             case MATROSKA_ID_SIMPLEBLOCK:
2707                 pos = url_ftell(matroska->ctx->pb);
2708                 res = ebml_read_binary(matroska, &id, &data, &size);
2709                 if (res == 0)
2710                     res = matroska_parse_block(matroska, data, size, pos,
2711                                                cluster_time, AV_NOPTS_VALUE,
2712                                                -1, 0);
2713                 break;
2714
2715             default:
2716                 av_log(matroska->ctx, AV_LOG_INFO,
2717                        "Unknown entry 0x%x in cluster data\n", id);
2718                 /* fall-through */
2719
2720             case EBML_ID_VOID:
2721                 res = ebml_read_skip(matroska);
2722                 break;
2723         }
2724
2725         if (matroska->level_up) {
2726             matroska->level_up--;
2727             break;
2728         }
2729     }
2730
2731     return res;
2732 }
2733
2734 static int
2735 matroska_read_packet (AVFormatContext *s,
2736                       AVPacket        *pkt)
2737 {
2738     MatroskaDemuxContext *matroska = s->priv_data;
2739     int res;
2740     uint32_t id;
2741
2742     /* Read stream until we have a packet queued. */
2743     while (matroska_deliver_packet(matroska, pkt)) {
2744
2745         /* Have we already reached the end? */
2746         if (matroska->done)
2747             return AVERROR(EIO);
2748
2749         res = 0;
2750         while (res == 0) {
2751             if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2752                 return AVERROR(EIO);
2753             } else if (matroska->level_up) {
2754                 matroska->level_up--;
2755                 break;
2756             }
2757
2758             switch (id) {
2759                 case MATROSKA_ID_CLUSTER:
2760                     if ((res = ebml_read_master(matroska, &id)) < 0)
2761                         break;
2762                     if ((res = matroska_parse_cluster(matroska)) == 0)
2763                         res = 1; /* Parsed one cluster, let's get out. */
2764                     break;
2765
2766                 default:
2767                 case EBML_ID_VOID:
2768                     res = ebml_read_skip(matroska);
2769                     break;
2770             }
2771
2772             if (matroska->level_up) {
2773                 matroska->level_up--;
2774                 break;
2775             }
2776         }
2777
2778         if (res == -1)
2779             matroska->done = 1;
2780     }
2781
2782     return 0;
2783 }
2784
2785 static int
2786 matroska_read_seek (AVFormatContext *s, int stream_index, int64_t timestamp,
2787                     int flags)
2788 {
2789     MatroskaDemuxContext *matroska = s->priv_data;
2790     AVStream *st = s->streams[stream_index];
2791     int index;
2792
2793     /* find index entry */
2794     index = av_index_search_timestamp(st, timestamp, flags);
2795     if (index < 0)
2796         return 0;
2797
2798     matroska_clear_queue(matroska);
2799
2800     /* do the seek */
2801     url_fseek(s->pb, st->index_entries[index].pos, SEEK_SET);
2802     matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
2803     matroska->skip_to_stream = st;
2804     matroska->peek_id = 0;
2805     return 0;
2806 }
2807
2808 static int
2809 matroska_read_close (AVFormatContext *s)
2810 {
2811     MatroskaDemuxContext *matroska = s->priv_data;
2812     int n = 0;
2813
2814     av_free(matroska->writing_app);
2815     av_free(matroska->muxing_app);
2816     av_free(matroska->index);
2817
2818     matroska_clear_queue(matroska);
2819
2820     for (n = 0; n < matroska->num_tracks; n++) {
2821         MatroskaTrack *track = matroska->tracks[n];
2822         av_free(track->codec_id);
2823         av_free(track->codec_name);
2824         av_free(track->codec_priv);
2825         av_free(track->name);
2826
2827         if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
2828             MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)track;
2829             av_free(audiotrack->buf);
2830         }
2831
2832         av_free(track);
2833     }
2834
2835     return 0;
2836 }
2837
2838 AVInputFormat matroska_demuxer = {
2839     "matroska",
2840     "Matroska file format",
2841     sizeof(MatroskaDemuxContext),
2842     matroska_probe,
2843     matroska_read_header,
2844     matroska_read_packet,
2845     matroska_read_close,
2846     matroska_read_seek,
2847 };