Simplify implementation and use of dec2()
[FFMpeg-mirror/DVCPRO-HD.git] / libavformat / matroskadec.c
blob6f6caa169de87a88bd51d2b48dd442299fb39b94
1 /*
2 * Matroska file demuxer (no muxer yet)
3 * Copyright (c) 2003-2004 The ffmpeg Project
5 * This file is part of FFmpeg.
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.
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.
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
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/.
31 #include "avformat.h"
32 /* For codec_get_id(). */
33 #include "riff.h"
34 #include "matroska.h"
35 #include "libavcodec/mpeg4audio.h"
36 #include "libavutil/intfloat_readwrite.h"
37 #include "libavutil/lzo.h"
38 #ifdef CONFIG_ZLIB
39 #include <zlib.h>
40 #endif
41 #ifdef CONFIG_BZLIB
42 #include <bzlib.h>
43 #endif
45 typedef struct Track {
46 MatroskaTrackType type;
48 /* Unique track number and track ID. stream_index is the index that
49 * the calling app uses for this track. */
50 uint32_t num;
51 uint32_t uid;
52 int stream_index;
54 char *name;
55 char language[4];
57 char *codec_id;
58 char *codec_name;
60 unsigned char *codec_priv;
61 int codec_priv_size;
63 uint64_t default_duration;
64 MatroskaTrackFlags flags;
66 int encoding_scope;
67 MatroskaTrackEncodingCompAlgo encoding_algo;
68 uint8_t *encoding_settings;
69 int encoding_settings_len;
70 } MatroskaTrack;
72 typedef struct MatroskaVideoTrack {
73 MatroskaTrack track;
75 int pixel_width;
76 int pixel_height;
77 int display_width;
78 int display_height;
80 uint32_t fourcc;
82 MatroskaAspectRatioMode ar_mode;
83 MatroskaEyeMode eye_mode;
85 //..
86 } MatroskaVideoTrack;
88 typedef struct MatroskaAudioTrack {
89 MatroskaTrack track;
91 int channels;
92 int bitdepth;
93 int internal_samplerate;
94 int samplerate;
95 int block_align;
97 /* real audio header */
98 int coded_framesize;
99 int sub_packet_h;
100 int frame_size;
101 int sub_packet_size;
102 int sub_packet_cnt;
103 int pkt_cnt;
104 uint8_t *buf;
105 //..
106 } MatroskaAudioTrack;
108 typedef struct MatroskaSubtitleTrack {
109 MatroskaTrack track;
110 //..
111 } MatroskaSubtitleTrack;
113 #define MAX_TRACK_SIZE (FFMAX3(sizeof(MatroskaVideoTrack), \
114 sizeof(MatroskaAudioTrack), \
115 sizeof(MatroskaSubtitleTrack)))
117 typedef struct MatroskaLevel {
118 uint64_t start;
119 uint64_t length;
120 } MatroskaLevel;
122 typedef struct MatroskaDemuxIndex {
123 uint64_t pos; /* of the corresponding *cluster*! */
124 uint16_t track; /* reference to 'num' */
125 uint64_t time; /* in nanoseconds */
126 } MatroskaDemuxIndex;
128 typedef struct MatroskaDemuxContext {
129 AVFormatContext *ctx;
131 /* ebml stuff */
132 int num_levels;
133 MatroskaLevel levels[EBML_MAX_DEPTH];
134 int level_up;
136 /* matroska stuff */
137 char *writing_app;
138 char *muxing_app;
139 int64_t created;
141 /* timescale in the file */
142 int64_t time_scale;
144 /* num_streams is the number of streams that av_new_stream() was called
145 * for ( = that are available to the calling program). */
146 int num_tracks;
147 int num_streams;
148 MatroskaTrack *tracks[MAX_STREAMS];
150 /* cache for ID peeking */
151 uint32_t peek_id;
153 /* byte position of the segment inside the stream */
154 offset_t segment_start;
156 /* The packet queue. */
157 AVPacket **packets;
158 int num_packets;
160 /* have we already parse metadata/cues/clusters? */
161 int metadata_parsed;
162 int index_parsed;
163 int done;
165 /* The index for seeking. */
166 int num_indexes;
167 MatroskaDemuxIndex *index;
169 /* What to skip before effectively reading a packet. */
170 int skip_to_keyframe;
171 AVStream *skip_to_stream;
172 } MatroskaDemuxContext;
175 * The first few functions handle EBML file parsing. The rest
176 * is the document interpretation. Matroska really just is a
177 * EBML file.
181 * Return: the amount of levels in the hierarchy that the
182 * current element lies higher than the previous one.
183 * The opposite isn't done - that's auto-done using master
184 * element reading.
187 static int
188 ebml_read_element_level_up (MatroskaDemuxContext *matroska)
190 ByteIOContext *pb = matroska->ctx->pb;
191 offset_t pos = url_ftell(pb);
192 int num = 0;
194 while (matroska->num_levels > 0) {
195 MatroskaLevel *level = &matroska->levels[matroska->num_levels - 1];
197 if (pos >= level->start + level->length) {
198 matroska->num_levels--;
199 num++;
200 } else {
201 break;
205 return num;
209 * Read: an "EBML number", which is defined as a variable-length
210 * array of bytes. The first byte indicates the length by giving a
211 * number of 0-bits followed by a one. The position of the first
212 * "one" bit inside the first byte indicates the length of this
213 * number.
214 * Returns: num. of bytes read. < 0 on error.
217 static int
218 ebml_read_num (MatroskaDemuxContext *matroska,
219 int max_size,
220 uint64_t *number)
222 ByteIOContext *pb = matroska->ctx->pb;
223 int len_mask = 0x80, read = 1, n = 1;
224 int64_t total = 0;
226 /* the first byte tells us the length in bytes - get_byte() can normally
227 * return 0, but since that's not a valid first ebmlID byte, we can
228 * use it safely here to catch EOS. */
229 if (!(total = get_byte(pb))) {
230 /* we might encounter EOS here */
231 if (!url_feof(pb)) {
232 offset_t pos = url_ftell(pb);
233 av_log(matroska->ctx, AV_LOG_ERROR,
234 "Read error at pos. %"PRIu64" (0x%"PRIx64")\n",
235 pos, pos);
237 return AVERROR(EIO); /* EOS or actual I/O error */
240 /* get the length of the EBML number */
241 while (read <= max_size && !(total & len_mask)) {
242 read++;
243 len_mask >>= 1;
245 if (read > max_size) {
246 offset_t pos = url_ftell(pb) - 1;
247 av_log(matroska->ctx, AV_LOG_ERROR,
248 "Invalid EBML number size tag 0x%02x at pos %"PRIu64" (0x%"PRIx64")\n",
249 (uint8_t) total, pos, pos);
250 return AVERROR_INVALIDDATA;
253 /* read out length */
254 total &= ~len_mask;
255 while (n++ < read)
256 total = (total << 8) | get_byte(pb);
258 *number = total;
260 return read;
264 * Read: the element content data ID.
265 * Return: the number of bytes read or < 0 on error.
268 static int
269 ebml_read_element_id (MatroskaDemuxContext *matroska,
270 uint32_t *id,
271 int *level_up)
273 int read;
274 uint64_t total;
276 /* if we re-call this, use our cached ID */
277 if (matroska->peek_id != 0) {
278 if (level_up)
279 *level_up = 0;
280 *id = matroska->peek_id;
281 return 0;
284 /* read out the "EBML number", include tag in ID */
285 if ((read = ebml_read_num(matroska, 4, &total)) < 0)
286 return read;
287 *id = matroska->peek_id = total | (1 << (read * 7));
289 /* level tracking */
290 if (level_up)
291 *level_up = ebml_read_element_level_up(matroska);
293 return read;
297 * Read: element content length.
298 * Return: the number of bytes read or < 0 on error.
301 static int
302 ebml_read_element_length (MatroskaDemuxContext *matroska,
303 uint64_t *length)
305 /* clear cache since we're now beyond that data point */
306 matroska->peek_id = 0;
308 /* read out the "EBML number", include tag in ID */
309 return ebml_read_num(matroska, 8, length);
313 * Return: the ID of the next element, or 0 on error.
314 * Level_up contains the amount of levels that this
315 * next element lies higher than the previous one.
318 static uint32_t
319 ebml_peek_id (MatroskaDemuxContext *matroska,
320 int *level_up)
322 uint32_t id;
324 if (ebml_read_element_id(matroska, &id, level_up) < 0)
325 return 0;
327 return id;
331 * Seek to a given offset.
332 * 0 is success, -1 is failure.
335 static int
336 ebml_read_seek (MatroskaDemuxContext *matroska,
337 offset_t offset)
339 ByteIOContext *pb = matroska->ctx->pb;
341 /* clear ID cache, if any */
342 matroska->peek_id = 0;
344 return (url_fseek(pb, offset, SEEK_SET) == offset) ? 0 : -1;
348 * Skip the next element.
349 * 0 is success, -1 is failure.
352 static int
353 ebml_read_skip (MatroskaDemuxContext *matroska)
355 ByteIOContext *pb = matroska->ctx->pb;
356 uint32_t id;
357 uint64_t length;
358 int res;
360 if ((res = ebml_read_element_id(matroska, &id, NULL)) < 0 ||
361 (res = ebml_read_element_length(matroska, &length)) < 0)
362 return res;
364 url_fskip(pb, length);
366 return 0;
370 * Read the next element as an unsigned int.
371 * 0 is success, < 0 is failure.
374 static int
375 ebml_read_uint (MatroskaDemuxContext *matroska,
376 uint32_t *id,
377 uint64_t *num)
379 ByteIOContext *pb = matroska->ctx->pb;
380 int n = 0, size, res;
381 uint64_t rlength;
383 if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
384 (res = ebml_read_element_length(matroska, &rlength)) < 0)
385 return res;
386 size = rlength;
387 if (size < 1 || size > 8) {
388 offset_t pos = url_ftell(pb);
389 av_log(matroska->ctx, AV_LOG_ERROR,
390 "Invalid uint element size %d at position %"PRId64" (0x%"PRIx64")\n",
391 size, pos, pos);
392 return AVERROR_INVALIDDATA;
395 /* big-endian ordening; build up number */
396 *num = 0;
397 while (n++ < size)
398 *num = (*num << 8) | get_byte(pb);
400 return 0;
404 * Read the next element as a signed int.
405 * 0 is success, < 0 is failure.
408 static int
409 ebml_read_sint (MatroskaDemuxContext *matroska,
410 uint32_t *id,
411 int64_t *num)
413 ByteIOContext *pb = matroska->ctx->pb;
414 int size, n = 1, negative = 0, res;
415 uint64_t rlength;
417 if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
418 (res = ebml_read_element_length(matroska, &rlength)) < 0)
419 return res;
420 size = rlength;
421 if (size < 1 || size > 8) {
422 offset_t pos = url_ftell(pb);
423 av_log(matroska->ctx, AV_LOG_ERROR,
424 "Invalid sint element size %d at position %"PRId64" (0x%"PRIx64")\n",
425 size, pos, pos);
426 return AVERROR_INVALIDDATA;
428 if ((*num = get_byte(pb)) & 0x80) {
429 negative = 1;
430 *num &= ~0x80;
432 while (n++ < size)
433 *num = (*num << 8) | get_byte(pb);
435 /* make signed */
436 if (negative)
437 *num = *num - (1LL << ((8 * size) - 1));
439 return 0;
443 * Read the next element as a float.
444 * 0 is success, < 0 is failure.
447 static int
448 ebml_read_float (MatroskaDemuxContext *matroska,
449 uint32_t *id,
450 double *num)
452 ByteIOContext *pb = matroska->ctx->pb;
453 int size, res;
454 uint64_t rlength;
456 if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
457 (res = ebml_read_element_length(matroska, &rlength)) < 0)
458 return res;
459 size = rlength;
461 if (size == 4) {
462 *num= av_int2flt(get_be32(pb));
463 } else if(size==8){
464 *num= av_int2dbl(get_be64(pb));
465 } else{
466 offset_t pos = url_ftell(pb);
467 av_log(matroska->ctx, AV_LOG_ERROR,
468 "Invalid float element size %d at position %"PRIu64" (0x%"PRIx64")\n",
469 size, pos, pos);
470 return AVERROR_INVALIDDATA;
473 return 0;
477 * Read the next element as an ASCII string.
478 * 0 is success, < 0 is failure.
481 static int
482 ebml_read_ascii (MatroskaDemuxContext *matroska,
483 uint32_t *id,
484 char **str)
486 ByteIOContext *pb = matroska->ctx->pb;
487 int size, res;
488 uint64_t rlength;
490 if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
491 (res = ebml_read_element_length(matroska, &rlength)) < 0)
492 return res;
493 size = rlength;
495 /* ebml strings are usually not 0-terminated, so we allocate one
496 * byte more, read the string and NULL-terminate it ourselves. */
497 if (size < 0 || !(*str = av_malloc(size + 1))) {
498 av_log(matroska->ctx, AV_LOG_ERROR, "Memory allocation failed\n");
499 return AVERROR(ENOMEM);
501 if (get_buffer(pb, (uint8_t *) *str, size) != size) {
502 offset_t pos = url_ftell(pb);
503 av_log(matroska->ctx, AV_LOG_ERROR,
504 "Read error at pos. %"PRIu64" (0x%"PRIx64")\n", pos, pos);
505 return AVERROR(EIO);
507 (*str)[size] = '\0';
509 return 0;
513 * Read the next element as a UTF-8 string.
514 * 0 is success, < 0 is failure.
517 static int
518 ebml_read_utf8 (MatroskaDemuxContext *matroska,
519 uint32_t *id,
520 char **str)
522 return ebml_read_ascii(matroska, id, str);
526 * Read the next element as a date (nanoseconds since 1/1/2000).
527 * 0 is success, < 0 is failure.
530 static int
531 ebml_read_date (MatroskaDemuxContext *matroska,
532 uint32_t *id,
533 int64_t *date)
535 return ebml_read_sint(matroska, id, date);
539 * Read the next element, but only the header. The contents
540 * are supposed to be sub-elements which can be read separately.
541 * 0 is success, < 0 is failure.
544 static int
545 ebml_read_master (MatroskaDemuxContext *matroska,
546 uint32_t *id)
548 ByteIOContext *pb = matroska->ctx->pb;
549 uint64_t length;
550 MatroskaLevel *level;
551 int res;
553 if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
554 (res = ebml_read_element_length(matroska, &length)) < 0)
555 return res;
557 /* protect... (Heaven forbids that the '>' is true) */
558 if (matroska->num_levels >= EBML_MAX_DEPTH) {
559 av_log(matroska->ctx, AV_LOG_ERROR,
560 "File moves beyond max. allowed depth (%d)\n", EBML_MAX_DEPTH);
561 return AVERROR(ENOSYS);
564 /* remember level */
565 level = &matroska->levels[matroska->num_levels++];
566 level->start = url_ftell(pb);
567 level->length = length;
569 return 0;
573 * Read the next element as binary data.
574 * 0 is success, < 0 is failure.
577 static int
578 ebml_read_binary (MatroskaDemuxContext *matroska,
579 uint32_t *id,
580 uint8_t **binary,
581 int *size)
583 ByteIOContext *pb = matroska->ctx->pb;
584 uint64_t rlength;
585 int res;
587 if ((res = ebml_read_element_id(matroska, id, NULL)) < 0 ||
588 (res = ebml_read_element_length(matroska, &rlength)) < 0)
589 return res;
590 *size = rlength;
592 if (!(*binary = av_malloc(*size))) {
593 av_log(matroska->ctx, AV_LOG_ERROR,
594 "Memory allocation error\n");
595 return AVERROR(ENOMEM);
598 if (get_buffer(pb, *binary, *size) != *size) {
599 offset_t pos = url_ftell(pb);
600 av_log(matroska->ctx, AV_LOG_ERROR,
601 "Read error at pos. %"PRIu64" (0x%"PRIx64")\n", pos, pos);
602 return AVERROR(EIO);
605 return 0;
609 * Read signed/unsigned "EBML" numbers.
610 * Return: number of bytes processed, < 0 on error.
611 * XXX: use ebml_read_num().
614 static int
615 matroska_ebmlnum_uint (uint8_t *data,
616 uint32_t size,
617 uint64_t *num)
619 int len_mask = 0x80, read = 1, n = 1, num_ffs = 0;
620 uint64_t total;
622 if (size <= 0)
623 return AVERROR_INVALIDDATA;
625 total = data[0];
626 while (read <= 8 && !(total & len_mask)) {
627 read++;
628 len_mask >>= 1;
630 if (read > 8)
631 return AVERROR_INVALIDDATA;
633 if ((total &= (len_mask - 1)) == len_mask - 1)
634 num_ffs++;
635 if (size < read)
636 return AVERROR_INVALIDDATA;
637 while (n < read) {
638 if (data[n] == 0xff)
639 num_ffs++;
640 total = (total << 8) | data[n];
641 n++;
644 if (read == num_ffs)
645 *num = (uint64_t)-1;
646 else
647 *num = total;
649 return read;
653 * Same as above, but signed.
656 static int
657 matroska_ebmlnum_sint (uint8_t *data,
658 uint32_t size,
659 int64_t *num)
661 uint64_t unum;
662 int res;
664 /* read as unsigned number first */
665 if ((res = matroska_ebmlnum_uint(data, size, &unum)) < 0)
666 return res;
668 /* make signed (weird way) */
669 if (unum == (uint64_t)-1)
670 *num = INT64_MAX;
671 else
672 *num = unum - ((1LL << ((7 * res) - 1)) - 1);
674 return res;
678 * Read an EBML header.
679 * 0 is success, < 0 is failure.
682 static int
683 ebml_read_header (MatroskaDemuxContext *matroska,
684 char **doctype,
685 int *version)
687 uint32_t id;
688 int level_up, res = 0;
690 /* default init */
691 if (doctype)
692 *doctype = NULL;
693 if (version)
694 *version = 1;
696 if (!(id = ebml_peek_id(matroska, &level_up)) ||
697 level_up != 0 || id != EBML_ID_HEADER) {
698 av_log(matroska->ctx, AV_LOG_ERROR,
699 "This is not an EBML file (id=0x%x/0x%x)\n", id, EBML_ID_HEADER);
700 return AVERROR_INVALIDDATA;
702 if ((res = ebml_read_master(matroska, &id)) < 0)
703 return res;
705 while (res == 0) {
706 if (!(id = ebml_peek_id(matroska, &level_up)))
707 return AVERROR(EIO);
709 /* end-of-header */
710 if (level_up)
711 break;
713 switch (id) {
714 /* is our read version uptodate? */
715 case EBML_ID_EBMLREADVERSION: {
716 uint64_t num;
718 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
719 return res;
720 if (num > EBML_VERSION) {
721 av_log(matroska->ctx, AV_LOG_ERROR,
722 "EBML version %"PRIu64" (> %d) is not supported\n",
723 num, EBML_VERSION);
724 return AVERROR_INVALIDDATA;
726 break;
729 /* we only handle 8 byte lengths at max */
730 case EBML_ID_EBMLMAXSIZELENGTH: {
731 uint64_t num;
733 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
734 return res;
735 if (num > sizeof(uint64_t)) {
736 av_log(matroska->ctx, AV_LOG_ERROR,
737 "Integers of size %"PRIu64" (> %zd) not supported\n",
738 num, sizeof(uint64_t));
739 return AVERROR_INVALIDDATA;
741 break;
744 /* we handle 4 byte IDs at max */
745 case EBML_ID_EBMLMAXIDLENGTH: {
746 uint64_t num;
748 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
749 return res;
750 if (num > sizeof(uint32_t)) {
751 av_log(matroska->ctx, AV_LOG_ERROR,
752 "IDs of size %"PRIu64" (> %zu) not supported\n",
753 num, sizeof(uint32_t));
754 return AVERROR_INVALIDDATA;
756 break;
759 case EBML_ID_DOCTYPE: {
760 char *text;
762 if ((res = ebml_read_ascii(matroska, &id, &text)) < 0)
763 return res;
764 if (doctype) {
765 if (*doctype)
766 av_free(*doctype);
767 *doctype = text;
768 } else
769 av_free(text);
770 break;
773 case EBML_ID_DOCTYPEREADVERSION: {
774 uint64_t num;
776 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
777 return res;
778 if (version)
779 *version = num;
780 break;
783 default:
784 av_log(matroska->ctx, AV_LOG_INFO,
785 "Unknown data type 0x%x in EBML header", id);
786 /* pass-through */
788 case EBML_ID_VOID:
789 /* we ignore these two, as they don't tell us anything we
790 * care about */
791 case EBML_ID_EBMLVERSION:
792 case EBML_ID_DOCTYPEVERSION:
793 res = ebml_read_skip (matroska);
794 break;
798 return 0;
802 static int
803 matroska_find_track_by_num (MatroskaDemuxContext *matroska,
804 int num)
806 int i;
808 for (i = 0; i < matroska->num_tracks; i++)
809 if (matroska->tracks[i]->num == num)
810 return i;
812 return -1;
817 * Put one packet in an application-supplied AVPacket struct.
818 * Returns 0 on success or -1 on failure.
821 static int
822 matroska_deliver_packet (MatroskaDemuxContext *matroska,
823 AVPacket *pkt)
825 if (matroska->num_packets > 0) {
826 memcpy(pkt, matroska->packets[0], sizeof(AVPacket));
827 av_free(matroska->packets[0]);
828 if (matroska->num_packets > 1) {
829 memmove(&matroska->packets[0], &matroska->packets[1],
830 (matroska->num_packets - 1) * sizeof(AVPacket *));
831 matroska->packets =
832 av_realloc(matroska->packets, (matroska->num_packets - 1) *
833 sizeof(AVPacket *));
834 } else {
835 av_freep(&matroska->packets);
837 matroska->num_packets--;
838 return 0;
841 return -1;
845 * Put a packet into our internal queue. Will be delivered to the
846 * user/application during the next get_packet() call.
849 static void
850 matroska_queue_packet (MatroskaDemuxContext *matroska,
851 AVPacket *pkt)
853 matroska->packets =
854 av_realloc(matroska->packets, (matroska->num_packets + 1) *
855 sizeof(AVPacket *));
856 matroska->packets[matroska->num_packets] = pkt;
857 matroska->num_packets++;
861 * Free all packets in our internal queue.
863 static void
864 matroska_clear_queue (MatroskaDemuxContext *matroska)
866 if (matroska->packets) {
867 int n;
868 for (n = 0; n < matroska->num_packets; n++) {
869 av_free_packet(matroska->packets[n]);
870 av_free(matroska->packets[n]);
872 av_free(matroska->packets);
873 matroska->packets = NULL;
874 matroska->num_packets = 0;
880 * Autodetecting...
883 static int
884 matroska_probe (AVProbeData *p)
886 uint64_t total = 0;
887 int len_mask = 0x80, size = 1, n = 1;
888 uint8_t probe_data[] = { 'm', 'a', 't', 'r', 'o', 's', 'k', 'a' };
890 /* ebml header? */
891 if (AV_RB32(p->buf) != EBML_ID_HEADER)
892 return 0;
894 /* length of header */
895 total = p->buf[4];
896 while (size <= 8 && !(total & len_mask)) {
897 size++;
898 len_mask >>= 1;
900 if (size > 8)
901 return 0;
902 total &= (len_mask - 1);
903 while (n < size)
904 total = (total << 8) | p->buf[4 + n++];
906 /* does the probe data contain the whole header? */
907 if (p->buf_size < 4 + size + total)
908 return 0;
910 /* the header must contain the document type 'matroska'. For now,
911 * we don't parse the whole header but simply check for the
912 * availability of that array of characters inside the header.
913 * Not fully fool-proof, but good enough. */
914 for (n = 4 + size; n <= 4 + size + total - sizeof(probe_data); n++)
915 if (!memcmp (&p->buf[n], probe_data, sizeof(probe_data)))
916 return AVPROBE_SCORE_MAX;
918 return 0;
922 * From here on, it's all XML-style DTD stuff... Needs no comments.
925 static int
926 matroska_parse_info (MatroskaDemuxContext *matroska)
928 int res = 0;
929 uint32_t id;
931 av_log(matroska->ctx, AV_LOG_DEBUG, "Parsing info...\n");
933 while (res == 0) {
934 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
935 res = AVERROR(EIO);
936 break;
937 } else if (matroska->level_up) {
938 matroska->level_up--;
939 break;
942 switch (id) {
943 /* cluster timecode */
944 case MATROSKA_ID_TIMECODESCALE: {
945 uint64_t num;
946 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
947 break;
948 matroska->time_scale = num;
949 break;
952 case MATROSKA_ID_DURATION: {
953 double num;
954 if ((res = ebml_read_float(matroska, &id, &num)) < 0)
955 break;
956 matroska->ctx->duration = num * matroska->time_scale * 1000 / AV_TIME_BASE;
957 break;
960 case MATROSKA_ID_TITLE: {
961 char *text;
962 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
963 break;
964 strncpy(matroska->ctx->title, text,
965 sizeof(matroska->ctx->title)-1);
966 av_free(text);
967 break;
970 case MATROSKA_ID_WRITINGAPP: {
971 char *text;
972 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
973 break;
974 matroska->writing_app = text;
975 break;
978 case MATROSKA_ID_MUXINGAPP: {
979 char *text;
980 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
981 break;
982 matroska->muxing_app = text;
983 break;
986 case MATROSKA_ID_DATEUTC: {
987 int64_t time;
988 if ((res = ebml_read_date(matroska, &id, &time)) < 0)
989 break;
990 matroska->created = time;
991 break;
994 default:
995 av_log(matroska->ctx, AV_LOG_INFO,
996 "Unknown entry 0x%x in info header\n", id);
997 /* fall-through */
999 case EBML_ID_VOID:
1000 res = ebml_read_skip(matroska);
1001 break;
1004 if (matroska->level_up) {
1005 matroska->level_up--;
1006 break;
1010 return res;
1013 static int
1014 matroska_add_stream (MatroskaDemuxContext *matroska)
1016 int res = 0;
1017 uint32_t id;
1018 MatroskaTrack *track;
1020 av_log(matroska->ctx, AV_LOG_DEBUG, "parsing track, adding stream..,\n");
1022 /* Allocate a generic track. As soon as we know its type we'll realloc. */
1023 track = av_mallocz(MAX_TRACK_SIZE);
1024 matroska->num_tracks++;
1025 strcpy(track->language, "eng");
1027 /* start with the master */
1028 if ((res = ebml_read_master(matroska, &id)) < 0)
1029 return res;
1031 /* try reading the trackentry headers */
1032 while (res == 0) {
1033 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1034 res = AVERROR(EIO);
1035 break;
1036 } else if (matroska->level_up > 0) {
1037 matroska->level_up--;
1038 break;
1041 switch (id) {
1042 /* track number (unique stream ID) */
1043 case MATROSKA_ID_TRACKNUMBER: {
1044 uint64_t num;
1045 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1046 break;
1047 track->num = num;
1048 break;
1051 /* track UID (unique identifier) */
1052 case MATROSKA_ID_TRACKUID: {
1053 uint64_t num;
1054 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1055 break;
1056 track->uid = num;
1057 break;
1060 /* track type (video, audio, combined, subtitle, etc.) */
1061 case MATROSKA_ID_TRACKTYPE: {
1062 uint64_t num;
1063 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1064 break;
1065 if (track->type && track->type != num) {
1066 av_log(matroska->ctx, AV_LOG_INFO,
1067 "More than one tracktype in an entry - skip\n");
1068 break;
1070 track->type = num;
1072 switch (track->type) {
1073 case MATROSKA_TRACK_TYPE_VIDEO:
1074 case MATROSKA_TRACK_TYPE_AUDIO:
1075 case MATROSKA_TRACK_TYPE_SUBTITLE:
1076 break;
1077 case MATROSKA_TRACK_TYPE_COMPLEX:
1078 case MATROSKA_TRACK_TYPE_LOGO:
1079 case MATROSKA_TRACK_TYPE_CONTROL:
1080 default:
1081 av_log(matroska->ctx, AV_LOG_INFO,
1082 "Unknown or unsupported track type 0x%x\n",
1083 track->type);
1084 track->type = MATROSKA_TRACK_TYPE_NONE;
1085 break;
1087 matroska->tracks[matroska->num_tracks - 1] = track;
1088 break;
1091 /* tracktype specific stuff for video */
1092 case MATROSKA_ID_TRACKVIDEO: {
1093 MatroskaVideoTrack *videotrack;
1094 if (!track->type)
1095 track->type = MATROSKA_TRACK_TYPE_VIDEO;
1096 if (track->type != MATROSKA_TRACK_TYPE_VIDEO) {
1097 av_log(matroska->ctx, AV_LOG_INFO,
1098 "video data in non-video track - ignoring\n");
1099 res = AVERROR_INVALIDDATA;
1100 break;
1101 } else if ((res = ebml_read_master(matroska, &id)) < 0)
1102 break;
1103 videotrack = (MatroskaVideoTrack *)track;
1105 while (res == 0) {
1106 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1107 res = AVERROR(EIO);
1108 break;
1109 } else if (matroska->level_up > 0) {
1110 matroska->level_up--;
1111 break;
1114 switch (id) {
1115 /* fixme, this should be one-up, but I get it here */
1116 case MATROSKA_ID_TRACKDEFAULTDURATION: {
1117 uint64_t num;
1118 if ((res = ebml_read_uint (matroska, &id,
1119 &num)) < 0)
1120 break;
1121 track->default_duration = num;
1122 break;
1125 /* video framerate */
1126 case MATROSKA_ID_VIDEOFRAMERATE: {
1127 double num;
1128 if ((res = ebml_read_float(matroska, &id,
1129 &num)) < 0)
1130 break;
1131 if (!track->default_duration)
1132 track->default_duration = 1000000000/num;
1133 break;
1136 /* width of the size to display the video at */
1137 case MATROSKA_ID_VIDEODISPLAYWIDTH: {
1138 uint64_t num;
1139 if ((res = ebml_read_uint(matroska, &id,
1140 &num)) < 0)
1141 break;
1142 videotrack->display_width = num;
1143 break;
1146 /* height of the size to display the video at */
1147 case MATROSKA_ID_VIDEODISPLAYHEIGHT: {
1148 uint64_t num;
1149 if ((res = ebml_read_uint(matroska, &id,
1150 &num)) < 0)
1151 break;
1152 videotrack->display_height = num;
1153 break;
1156 /* width of the video in the file */
1157 case MATROSKA_ID_VIDEOPIXELWIDTH: {
1158 uint64_t num;
1159 if ((res = ebml_read_uint(matroska, &id,
1160 &num)) < 0)
1161 break;
1162 videotrack->pixel_width = num;
1163 break;
1166 /* height of the video in the file */
1167 case MATROSKA_ID_VIDEOPIXELHEIGHT: {
1168 uint64_t num;
1169 if ((res = ebml_read_uint(matroska, &id,
1170 &num)) < 0)
1171 break;
1172 videotrack->pixel_height = num;
1173 break;
1176 /* whether the video is interlaced */
1177 case MATROSKA_ID_VIDEOFLAGINTERLACED: {
1178 uint64_t num;
1179 if ((res = ebml_read_uint(matroska, &id,
1180 &num)) < 0)
1181 break;
1182 if (num)
1183 track->flags |=
1184 MATROSKA_VIDEOTRACK_INTERLACED;
1185 else
1186 track->flags &=
1187 ~MATROSKA_VIDEOTRACK_INTERLACED;
1188 break;
1191 /* stereo mode (whether the video has two streams,
1192 * where one is for the left eye and the other for
1193 * the right eye, which creates a 3D-like
1194 * effect) */
1195 case MATROSKA_ID_VIDEOSTEREOMODE: {
1196 uint64_t num;
1197 if ((res = ebml_read_uint(matroska, &id,
1198 &num)) < 0)
1199 break;
1200 if (num != MATROSKA_EYE_MODE_MONO &&
1201 num != MATROSKA_EYE_MODE_LEFT &&
1202 num != MATROSKA_EYE_MODE_RIGHT &&
1203 num != MATROSKA_EYE_MODE_BOTH) {
1204 av_log(matroska->ctx, AV_LOG_INFO,
1205 "Ignoring unknown eye mode 0x%x\n",
1206 (uint32_t) num);
1207 break;
1209 videotrack->eye_mode = num;
1210 break;
1213 /* aspect ratio behaviour */
1214 case MATROSKA_ID_VIDEOASPECTRATIO: {
1215 uint64_t num;
1216 if ((res = ebml_read_uint(matroska, &id,
1217 &num)) < 0)
1218 break;
1219 if (num != MATROSKA_ASPECT_RATIO_MODE_FREE &&
1220 num != MATROSKA_ASPECT_RATIO_MODE_KEEP &&
1221 num != MATROSKA_ASPECT_RATIO_MODE_FIXED) {
1222 av_log(matroska->ctx, AV_LOG_INFO,
1223 "Ignoring unknown aspect ratio 0x%x\n",
1224 (uint32_t) num);
1225 break;
1227 videotrack->ar_mode = num;
1228 break;
1231 /* colorspace (only matters for raw video)
1232 * fourcc */
1233 case MATROSKA_ID_VIDEOCOLORSPACE: {
1234 uint64_t num;
1235 if ((res = ebml_read_uint(matroska, &id,
1236 &num)) < 0)
1237 break;
1238 videotrack->fourcc = num;
1239 break;
1242 default:
1243 av_log(matroska->ctx, AV_LOG_INFO,
1244 "Unknown video track header entry "
1245 "0x%x - ignoring\n", id);
1246 /* pass-through */
1248 case EBML_ID_VOID:
1249 res = ebml_read_skip(matroska);
1250 break;
1253 if (matroska->level_up) {
1254 matroska->level_up--;
1255 break;
1258 break;
1261 /* tracktype specific stuff for audio */
1262 case MATROSKA_ID_TRACKAUDIO: {
1263 MatroskaAudioTrack *audiotrack;
1264 if (!track->type)
1265 track->type = MATROSKA_TRACK_TYPE_AUDIO;
1266 if (track->type != MATROSKA_TRACK_TYPE_AUDIO) {
1267 av_log(matroska->ctx, AV_LOG_INFO,
1268 "audio data in non-audio track - ignoring\n");
1269 res = AVERROR_INVALIDDATA;
1270 break;
1271 } else if ((res = ebml_read_master(matroska, &id)) < 0)
1272 break;
1273 audiotrack = (MatroskaAudioTrack *)track;
1274 audiotrack->channels = 1;
1275 audiotrack->samplerate = 8000;
1277 while (res == 0) {
1278 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1279 res = AVERROR(EIO);
1280 break;
1281 } else if (matroska->level_up > 0) {
1282 matroska->level_up--;
1283 break;
1286 switch (id) {
1287 /* samplerate */
1288 case MATROSKA_ID_AUDIOSAMPLINGFREQ: {
1289 double num;
1290 if ((res = ebml_read_float(matroska, &id,
1291 &num)) < 0)
1292 break;
1293 audiotrack->internal_samplerate =
1294 audiotrack->samplerate = num;
1295 break;
1298 case MATROSKA_ID_AUDIOOUTSAMPLINGFREQ: {
1299 double num;
1300 if ((res = ebml_read_float(matroska, &id,
1301 &num)) < 0)
1302 break;
1303 audiotrack->samplerate = num;
1304 break;
1307 /* bitdepth */
1308 case MATROSKA_ID_AUDIOBITDEPTH: {
1309 uint64_t num;
1310 if ((res = ebml_read_uint(matroska, &id,
1311 &num)) < 0)
1312 break;
1313 audiotrack->bitdepth = num;
1314 break;
1317 /* channels */
1318 case MATROSKA_ID_AUDIOCHANNELS: {
1319 uint64_t num;
1320 if ((res = ebml_read_uint(matroska, &id,
1321 &num)) < 0)
1322 break;
1323 audiotrack->channels = num;
1324 break;
1327 default:
1328 av_log(matroska->ctx, AV_LOG_INFO,
1329 "Unknown audio track header entry "
1330 "0x%x - ignoring\n", id);
1331 /* pass-through */
1333 case EBML_ID_VOID:
1334 res = ebml_read_skip(matroska);
1335 break;
1338 if (matroska->level_up) {
1339 matroska->level_up--;
1340 break;
1343 break;
1346 /* codec identifier */
1347 case MATROSKA_ID_CODECID: {
1348 char *text;
1349 if ((res = ebml_read_ascii(matroska, &id, &text)) < 0)
1350 break;
1351 track->codec_id = text;
1352 break;
1355 /* codec private data */
1356 case MATROSKA_ID_CODECPRIVATE: {
1357 uint8_t *data;
1358 int size;
1359 if ((res = ebml_read_binary(matroska, &id, &data, &size) < 0))
1360 break;
1361 track->codec_priv = data;
1362 track->codec_priv_size = size;
1363 break;
1366 /* name of the codec */
1367 case MATROSKA_ID_CODECNAME: {
1368 char *text;
1369 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
1370 break;
1371 track->codec_name = text;
1372 break;
1375 /* name of this track */
1376 case MATROSKA_ID_TRACKNAME: {
1377 char *text;
1378 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
1379 break;
1380 track->name = text;
1381 break;
1384 /* language (matters for audio/subtitles, mostly) */
1385 case MATROSKA_ID_TRACKLANGUAGE: {
1386 char *text, *end;
1387 if ((res = ebml_read_utf8(matroska, &id, &text)) < 0)
1388 break;
1389 if ((end = strchr(text, '-')))
1390 *end = '\0';
1391 if (strlen(text) == 3)
1392 strcpy(track->language, text);
1393 av_free(text);
1394 break;
1397 /* whether this is actually used */
1398 case MATROSKA_ID_TRACKFLAGENABLED: {
1399 uint64_t num;
1400 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1401 break;
1402 if (num)
1403 track->flags |= MATROSKA_TRACK_ENABLED;
1404 else
1405 track->flags &= ~MATROSKA_TRACK_ENABLED;
1406 break;
1409 /* whether it's the default for this track type */
1410 case MATROSKA_ID_TRACKFLAGDEFAULT: {
1411 uint64_t num;
1412 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1413 break;
1414 if (num)
1415 track->flags |= MATROSKA_TRACK_DEFAULT;
1416 else
1417 track->flags &= ~MATROSKA_TRACK_DEFAULT;
1418 break;
1421 /* lacing (like MPEG, where blocks don't end/start on frame
1422 * boundaries) */
1423 case MATROSKA_ID_TRACKFLAGLACING: {
1424 uint64_t num;
1425 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1426 break;
1427 if (num)
1428 track->flags |= MATROSKA_TRACK_LACING;
1429 else
1430 track->flags &= ~MATROSKA_TRACK_LACING;
1431 break;
1434 /* default length (in time) of one data block in this track */
1435 case MATROSKA_ID_TRACKDEFAULTDURATION: {
1436 uint64_t num;
1437 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1438 break;
1439 track->default_duration = num;
1440 break;
1443 case MATROSKA_ID_TRACKCONTENTENCODINGS: {
1444 if ((res = ebml_read_master(matroska, &id)) < 0)
1445 break;
1447 while (res == 0) {
1448 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1449 res = AVERROR(EIO);
1450 break;
1451 } else if (matroska->level_up > 0) {
1452 matroska->level_up--;
1453 break;
1456 switch (id) {
1457 case MATROSKA_ID_TRACKCONTENTENCODING: {
1458 int encoding_scope = 1;
1459 if ((res = ebml_read_master(matroska, &id)) < 0)
1460 break;
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 > 0) {
1467 matroska->level_up--;
1468 break;
1471 switch (id) {
1472 case MATROSKA_ID_ENCODINGSCOPE: {
1473 uint64_t num;
1474 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1475 break;
1476 encoding_scope = num;
1477 break;
1480 case MATROSKA_ID_ENCODINGTYPE: {
1481 uint64_t num;
1482 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1483 break;
1484 if (num)
1485 av_log(matroska->ctx, AV_LOG_ERROR,
1486 "Unsupported encoding type");
1487 break;
1490 case MATROSKA_ID_ENCODINGCOMPRESSION: {
1491 if ((res = ebml_read_master(matroska, &id)) < 0)
1492 break;
1494 while (res == 0) {
1495 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1496 res = AVERROR(EIO);
1497 break;
1498 } else if (matroska->level_up > 0) {
1499 matroska->level_up--;
1500 break;
1503 switch (id) {
1504 case MATROSKA_ID_ENCODINGCOMPALGO: {
1505 uint64_t num;
1506 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
1507 break;
1508 if (num != MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP &&
1509 #ifdef CONFIG_ZLIB
1510 num != MATROSKA_TRACK_ENCODING_COMP_ZLIB &&
1511 #endif
1512 #ifdef CONFIG_BZLIB
1513 num != MATROSKA_TRACK_ENCODING_COMP_BZLIB &&
1514 #endif
1515 num != MATROSKA_TRACK_ENCODING_COMP_LZO)
1516 av_log(matroska->ctx, AV_LOG_ERROR,
1517 "Unsupported compression algo\n");
1518 track->encoding_algo = num;
1519 break;
1522 case MATROSKA_ID_ENCODINGCOMPSETTINGS: {
1523 uint8_t *data;
1524 int size;
1525 if ((res = ebml_read_binary(matroska, &id, &data, &size) < 0))
1526 break;
1527 track->encoding_settings = data;
1528 track->encoding_settings_len = size;
1529 break;
1532 default:
1533 av_log(matroska->ctx, AV_LOG_INFO,
1534 "Unknown compression header entry "
1535 "0x%x - ignoring\n", id);
1536 /* pass-through */
1538 case EBML_ID_VOID:
1539 res = ebml_read_skip(matroska);
1540 break;
1543 if (matroska->level_up) {
1544 matroska->level_up--;
1545 break;
1548 break;
1551 default:
1552 av_log(matroska->ctx, AV_LOG_INFO,
1553 "Unknown content encoding header entry "
1554 "0x%x - ignoring\n", id);
1555 /* pass-through */
1557 case EBML_ID_VOID:
1558 res = ebml_read_skip(matroska);
1559 break;
1562 if (matroska->level_up) {
1563 matroska->level_up--;
1564 break;
1568 track->encoding_scope = encoding_scope;
1569 break;
1572 default:
1573 av_log(matroska->ctx, AV_LOG_INFO,
1574 "Unknown content encodings header entry "
1575 "0x%x - ignoring\n", id);
1576 /* pass-through */
1578 case EBML_ID_VOID:
1579 res = ebml_read_skip(matroska);
1580 break;
1583 if (matroska->level_up) {
1584 matroska->level_up--;
1585 break;
1588 break;
1591 default:
1592 av_log(matroska->ctx, AV_LOG_INFO,
1593 "Unknown track header entry 0x%x - ignoring\n", id);
1594 /* pass-through */
1596 case EBML_ID_VOID:
1597 /* we ignore these because they're nothing useful. */
1598 case MATROSKA_ID_CODECINFOURL:
1599 case MATROSKA_ID_CODECDOWNLOADURL:
1600 case MATROSKA_ID_TRACKMINCACHE:
1601 case MATROSKA_ID_TRACKMAXCACHE:
1602 res = ebml_read_skip(matroska);
1603 break;
1606 if (matroska->level_up) {
1607 matroska->level_up--;
1608 break;
1612 return res;
1615 static int
1616 matroska_parse_tracks (MatroskaDemuxContext *matroska)
1618 int res = 0;
1619 uint32_t id;
1621 av_log(matroska->ctx, AV_LOG_DEBUG, "parsing tracks...\n");
1623 while (res == 0) {
1624 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1625 res = AVERROR(EIO);
1626 break;
1627 } else if (matroska->level_up) {
1628 matroska->level_up--;
1629 break;
1632 switch (id) {
1633 /* one track within the "all-tracks" header */
1634 case MATROSKA_ID_TRACKENTRY:
1635 res = matroska_add_stream(matroska);
1636 break;
1638 default:
1639 av_log(matroska->ctx, AV_LOG_INFO,
1640 "Unknown entry 0x%x in track header\n", id);
1641 /* fall-through */
1643 case EBML_ID_VOID:
1644 res = ebml_read_skip(matroska);
1645 break;
1648 if (matroska->level_up) {
1649 matroska->level_up--;
1650 break;
1654 return res;
1657 static int
1658 matroska_parse_index (MatroskaDemuxContext *matroska)
1660 int res = 0;
1661 uint32_t id;
1662 MatroskaDemuxIndex idx;
1664 av_log(matroska->ctx, AV_LOG_DEBUG, "parsing index...\n");
1666 while (res == 0) {
1667 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1668 res = AVERROR(EIO);
1669 break;
1670 } else if (matroska->level_up) {
1671 matroska->level_up--;
1672 break;
1675 switch (id) {
1676 /* one single index entry ('point') */
1677 case MATROSKA_ID_POINTENTRY:
1678 if ((res = ebml_read_master(matroska, &id)) < 0)
1679 break;
1681 /* in the end, we hope to fill one entry with a
1682 * timestamp, a file position and a tracknum */
1683 idx.pos = (uint64_t) -1;
1684 idx.time = (uint64_t) -1;
1685 idx.track = (uint16_t) -1;
1687 while (res == 0) {
1688 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1689 res = AVERROR(EIO);
1690 break;
1691 } else if (matroska->level_up) {
1692 matroska->level_up--;
1693 break;
1696 switch (id) {
1697 /* one single index entry ('point') */
1698 case MATROSKA_ID_CUETIME: {
1699 uint64_t time;
1700 if ((res = ebml_read_uint(matroska, &id,
1701 &time)) < 0)
1702 break;
1703 idx.time = time * matroska->time_scale;
1704 break;
1707 /* position in the file + track to which it
1708 * belongs */
1709 case MATROSKA_ID_CUETRACKPOSITION:
1710 if ((res = ebml_read_master(matroska, &id)) < 0)
1711 break;
1713 while (res == 0) {
1714 if (!(id = ebml_peek_id (matroska,
1715 &matroska->level_up))) {
1716 res = AVERROR(EIO);
1717 break;
1718 } else if (matroska->level_up) {
1719 matroska->level_up--;
1720 break;
1723 switch (id) {
1724 /* track number */
1725 case MATROSKA_ID_CUETRACK: {
1726 uint64_t num;
1727 if ((res = ebml_read_uint(matroska,
1728 &id, &num)) < 0)
1729 break;
1730 idx.track = num;
1731 break;
1734 /* position in file */
1735 case MATROSKA_ID_CUECLUSTERPOSITION: {
1736 uint64_t num;
1737 if ((res = ebml_read_uint(matroska,
1738 &id, &num)) < 0)
1739 break;
1740 idx.pos = num+matroska->segment_start;
1741 break;
1744 default:
1745 av_log(matroska->ctx, AV_LOG_INFO,
1746 "Unknown entry 0x%x in "
1747 "CuesTrackPositions\n", id);
1748 /* fall-through */
1750 case EBML_ID_VOID:
1751 res = ebml_read_skip(matroska);
1752 break;
1755 if (matroska->level_up) {
1756 matroska->level_up--;
1757 break;
1761 break;
1763 default:
1764 av_log(matroska->ctx, AV_LOG_INFO,
1765 "Unknown entry 0x%x in cuespoint "
1766 "index\n", id);
1767 /* fall-through */
1769 case EBML_ID_VOID:
1770 res = ebml_read_skip(matroska);
1771 break;
1774 if (matroska->level_up) {
1775 matroska->level_up--;
1776 break;
1780 /* so let's see if we got what we wanted */
1781 if (idx.pos != (uint64_t) -1 &&
1782 idx.time != (uint64_t) -1 &&
1783 idx.track != (uint16_t) -1) {
1784 if (matroska->num_indexes % 32 == 0) {
1785 /* re-allocate bigger index */
1786 matroska->index =
1787 av_realloc(matroska->index,
1788 (matroska->num_indexes + 32) *
1789 sizeof(MatroskaDemuxIndex));
1791 matroska->index[matroska->num_indexes] = idx;
1792 matroska->num_indexes++;
1794 break;
1796 default:
1797 av_log(matroska->ctx, AV_LOG_INFO,
1798 "Unknown entry 0x%x in cues header\n", id);
1799 /* fall-through */
1801 case EBML_ID_VOID:
1802 res = ebml_read_skip(matroska);
1803 break;
1806 if (matroska->level_up) {
1807 matroska->level_up--;
1808 break;
1812 return res;
1815 static int
1816 matroska_parse_metadata (MatroskaDemuxContext *matroska)
1818 int res = 0;
1819 uint32_t id;
1821 while (res == 0) {
1822 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1823 res = AVERROR(EIO);
1824 break;
1825 } else if (matroska->level_up) {
1826 matroska->level_up--;
1827 break;
1830 switch (id) {
1831 /* Hm, this is unsupported... */
1832 default:
1833 av_log(matroska->ctx, AV_LOG_INFO,
1834 "Unknown entry 0x%x in metadata header\n", id);
1835 /* fall-through */
1837 case EBML_ID_VOID:
1838 res = ebml_read_skip(matroska);
1839 break;
1842 if (matroska->level_up) {
1843 matroska->level_up--;
1844 break;
1848 return res;
1851 static int
1852 matroska_parse_seekhead (MatroskaDemuxContext *matroska)
1854 int res = 0;
1855 uint32_t id;
1857 av_log(matroska->ctx, AV_LOG_DEBUG, "parsing seekhead...\n");
1859 while (res == 0) {
1860 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1861 res = AVERROR(EIO);
1862 break;
1863 } else if (matroska->level_up) {
1864 matroska->level_up--;
1865 break;
1868 switch (id) {
1869 case MATROSKA_ID_SEEKENTRY: {
1870 uint32_t seek_id = 0, peek_id_cache = 0;
1871 uint64_t seek_pos = (uint64_t) -1, t;
1873 if ((res = ebml_read_master(matroska, &id)) < 0)
1874 break;
1876 while (res == 0) {
1877 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
1878 res = AVERROR(EIO);
1879 break;
1880 } else if (matroska->level_up) {
1881 matroska->level_up--;
1882 break;
1885 switch (id) {
1886 case MATROSKA_ID_SEEKID:
1887 res = ebml_read_uint(matroska, &id, &t);
1888 seek_id = t;
1889 break;
1891 case MATROSKA_ID_SEEKPOSITION:
1892 res = ebml_read_uint(matroska, &id, &seek_pos);
1893 break;
1895 default:
1896 av_log(matroska->ctx, AV_LOG_INFO,
1897 "Unknown seekhead ID 0x%x\n", id);
1898 /* fall-through */
1900 case EBML_ID_VOID:
1901 res = ebml_read_skip(matroska);
1902 break;
1905 if (matroska->level_up) {
1906 matroska->level_up--;
1907 break;
1911 if (!seek_id || seek_pos == (uint64_t) -1) {
1912 av_log(matroska->ctx, AV_LOG_INFO,
1913 "Incomplete seekhead entry (0x%x/%"PRIu64")\n",
1914 seek_id, seek_pos);
1915 break;
1918 switch (seek_id) {
1919 case MATROSKA_ID_CUES:
1920 case MATROSKA_ID_TAGS: {
1921 uint32_t level_up = matroska->level_up;
1922 offset_t before_pos;
1923 uint64_t length;
1924 MatroskaLevel level;
1926 /* remember the peeked ID and the current position */
1927 peek_id_cache = matroska->peek_id;
1928 before_pos = url_ftell(matroska->ctx->pb);
1930 /* seek */
1931 if ((res = ebml_read_seek(matroska, seek_pos +
1932 matroska->segment_start)) < 0)
1933 goto finish;
1935 /* we don't want to lose our seekhead level, so we add
1936 * a dummy. This is a crude hack. */
1937 if (matroska->num_levels == EBML_MAX_DEPTH) {
1938 av_log(matroska->ctx, AV_LOG_INFO,
1939 "Max EBML element depth (%d) reached, "
1940 "cannot parse further.\n", EBML_MAX_DEPTH);
1941 return AVERROR_UNKNOWN;
1944 level.start = 0;
1945 level.length = (uint64_t)-1;
1946 matroska->levels[matroska->num_levels] = level;
1947 matroska->num_levels++;
1949 /* check ID */
1950 if (!(id = ebml_peek_id (matroska,
1951 &matroska->level_up)))
1952 goto finish;
1953 if (id != seek_id) {
1954 av_log(matroska->ctx, AV_LOG_INFO,
1955 "We looked for ID=0x%x but got "
1956 "ID=0x%x (pos=%"PRIu64")",
1957 seek_id, id, seek_pos +
1958 matroska->segment_start);
1959 goto finish;
1962 /* read master + parse */
1963 if ((res = ebml_read_master(matroska, &id)) < 0)
1964 goto finish;
1965 switch (id) {
1966 case MATROSKA_ID_CUES:
1967 if (!(res = matroska_parse_index(matroska)) ||
1968 url_feof(matroska->ctx->pb)) {
1969 matroska->index_parsed = 1;
1970 res = 0;
1972 break;
1973 case MATROSKA_ID_TAGS:
1974 if (!(res = matroska_parse_metadata(matroska)) ||
1975 url_feof(matroska->ctx->pb)) {
1976 matroska->metadata_parsed = 1;
1977 res = 0;
1979 break;
1982 finish:
1983 /* remove dummy level */
1984 while (matroska->num_levels) {
1985 matroska->num_levels--;
1986 length =
1987 matroska->levels[matroska->num_levels].length;
1988 if (length == (uint64_t)-1)
1989 break;
1992 /* seek back */
1993 if ((res = ebml_read_seek(matroska, before_pos)) < 0)
1994 return res;
1995 matroska->peek_id = peek_id_cache;
1996 matroska->level_up = level_up;
1997 break;
2000 default:
2001 av_log(matroska->ctx, AV_LOG_INFO,
2002 "Ignoring seekhead entry for ID=0x%x\n",
2003 seek_id);
2004 break;
2007 break;
2010 default:
2011 av_log(matroska->ctx, AV_LOG_INFO,
2012 "Unknown seekhead ID 0x%x\n", id);
2013 /* fall-through */
2015 case EBML_ID_VOID:
2016 res = ebml_read_skip(matroska);
2017 break;
2020 if (matroska->level_up) {
2021 matroska->level_up--;
2022 break;
2026 return res;
2029 static int
2030 matroska_parse_attachments(AVFormatContext *s)
2032 MatroskaDemuxContext *matroska = s->priv_data;
2033 int res = 0;
2034 uint32_t id;
2036 av_log(matroska->ctx, AV_LOG_DEBUG, "parsing attachments...\n");
2038 while (res == 0) {
2039 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2040 res = AVERROR(EIO);
2041 break;
2042 } else if (matroska->level_up) {
2043 matroska->level_up--;
2044 break;
2047 switch (id) {
2048 case MATROSKA_ID_ATTACHEDFILE: {
2049 char* name = NULL;
2050 char* mime = NULL;
2051 uint8_t* data = NULL;
2052 int i, data_size = 0;
2053 AVStream *st;
2055 if ((res = ebml_read_master(matroska, &id)) < 0)
2056 break;
2058 while (res == 0) {
2059 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2060 res = AVERROR(EIO);
2061 break;
2062 } else if (matroska->level_up) {
2063 matroska->level_up--;
2064 break;
2067 switch (id) {
2068 case MATROSKA_ID_FILENAME:
2069 res = ebml_read_utf8 (matroska, &id, &name);
2070 break;
2072 case MATROSKA_ID_FILEMIMETYPE:
2073 res = ebml_read_ascii (matroska, &id, &mime);
2074 break;
2076 case MATROSKA_ID_FILEDATA:
2077 res = ebml_read_binary(matroska, &id, &data, &data_size);
2078 break;
2080 default:
2081 av_log(matroska->ctx, AV_LOG_INFO,
2082 "Unknown attachedfile ID 0x%x\n", id);
2083 case EBML_ID_VOID:
2084 res = ebml_read_skip(matroska);
2085 break;
2088 if (matroska->level_up) {
2089 matroska->level_up--;
2090 break;
2094 if (!(name && mime && data && data_size > 0)) {
2095 av_log(matroska->ctx, AV_LOG_ERROR, "incomplete attachment\n");
2096 break;
2099 st = av_new_stream(s, matroska->num_streams++);
2100 if (st == NULL)
2101 return AVERROR(ENOMEM);
2102 st->filename = av_strdup(name);
2103 st->codec->codec_id = CODEC_ID_NONE;
2104 st->codec->codec_type = CODEC_TYPE_ATTACHMENT;
2105 st->codec->extradata = av_malloc(data_size);
2106 if(st->codec->extradata == NULL)
2107 return AVERROR(ENOMEM);
2108 st->codec->extradata_size = data_size;
2109 memcpy(st->codec->extradata, data, data_size);
2111 for (i=0; ff_mkv_mime_tags[i].id != CODEC_ID_NONE; i++) {
2112 if (!strncmp(ff_mkv_mime_tags[i].str, mime,
2113 strlen(ff_mkv_mime_tags[i].str))) {
2114 st->codec->codec_id = ff_mkv_mime_tags[i].id;
2115 break;
2119 av_log(matroska->ctx, AV_LOG_DEBUG, "new attachment: %s, %s, size %d \n", name, mime, data_size);
2120 break;
2123 default:
2124 av_log(matroska->ctx, AV_LOG_INFO,
2125 "Unknown attachments ID 0x%x\n", id);
2126 /* fall-through */
2128 case EBML_ID_VOID:
2129 res = ebml_read_skip(matroska);
2130 break;
2133 if (matroska->level_up) {
2134 matroska->level_up--;
2135 break;
2139 return res;
2142 static int
2143 matroska_parse_chapters(AVFormatContext *s)
2145 MatroskaDemuxContext *matroska = s->priv_data;
2146 int res = 0;
2147 uint32_t id;
2149 av_log(s, AV_LOG_DEBUG, "parsing chapters...\n");
2151 while (res == 0) {
2152 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2153 res = AVERROR(EIO);
2154 break;
2155 } else if (matroska->level_up) {
2156 matroska->level_up--;
2157 break;
2160 switch (id) {
2161 case MATROSKA_ID_EDITIONENTRY: {
2162 uint64_t end = AV_NOPTS_VALUE, start = AV_NOPTS_VALUE;
2163 int64_t uid= -1;
2164 char* title = NULL;
2165 /* if there is more than one chapter edition
2166 we take only the first one */
2167 if(s->chapters) {
2168 ebml_read_skip(matroska);
2169 break;
2172 if ((res = ebml_read_master(matroska, &id)) < 0)
2173 break;
2175 while (res == 0) {
2176 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2177 res = AVERROR(EIO);
2178 break;
2179 } else if (matroska->level_up) {
2180 matroska->level_up--;
2181 break;
2184 switch (id) {
2185 case MATROSKA_ID_CHAPTERATOM:
2186 if ((res = ebml_read_master(matroska, &id)) < 0)
2187 break;
2189 while (res == 0) {
2190 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2191 res = AVERROR(EIO);
2192 break;
2193 } else if (matroska->level_up) {
2194 matroska->level_up--;
2195 break;
2198 switch (id) {
2199 case MATROSKA_ID_CHAPTERTIMEEND:
2200 res = ebml_read_uint(matroska, &id, &end);
2201 break;
2203 case MATROSKA_ID_CHAPTERTIMESTART:
2204 res = ebml_read_uint(matroska, &id, &start);
2205 break;
2207 case MATROSKA_ID_CHAPTERDISPLAY:
2208 if ((res = ebml_read_master(matroska, &id)) < 0)
2209 break;
2211 while (res == 0) {
2212 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2213 res = AVERROR(EIO);
2214 break;
2215 } else if (matroska->level_up) {
2216 matroska->level_up--;
2217 break;
2220 switch (id) {
2221 case MATROSKA_ID_CHAPSTRING:
2222 res = ebml_read_utf8(matroska, &id, &title);
2223 break;
2225 default:
2226 av_log(s, AV_LOG_INFO, "Ignoring unknown Chapter display ID 0x%x\n", id);
2227 case EBML_ID_VOID:
2228 res = ebml_read_skip(matroska);
2229 break;
2232 if (matroska->level_up) {
2233 matroska->level_up--;
2234 break;
2237 break;
2239 case MATROSKA_ID_CHAPTERUID:
2240 res = ebml_read_uint(matroska, &id, &uid);
2241 break;
2242 default:
2243 av_log(s, AV_LOG_INFO, "Ignoring unknown Chapter atom ID 0x%x\n", id);
2244 case MATROSKA_ID_CHAPTERFLAGHIDDEN:
2245 case EBML_ID_VOID:
2246 res = ebml_read_skip(matroska);
2247 break;
2250 if (matroska->level_up) {
2251 matroska->level_up--;
2252 break;
2256 if (start != AV_NOPTS_VALUE && uid != -1) {
2257 if(!ff_new_chapter(s, uid, (AVRational){1, 1000000000}, start, end, title))
2258 res= AVERROR(ENOMEM);
2260 av_free(title);
2261 break;
2263 default:
2264 av_log(s, AV_LOG_INFO, "Ignoring unknown Edition entry ID 0x%x\n", id);
2265 case MATROSKA_ID_EDITIONUID:
2266 case MATROSKA_ID_EDITIONFLAGHIDDEN:
2267 case EBML_ID_VOID:
2268 res = ebml_read_skip(matroska);
2269 break;
2273 if (matroska->level_up) {
2274 matroska->level_up--;
2275 break;
2278 break;
2281 default:
2282 av_log(s, AV_LOG_INFO, "Expected an Edition entry (0x%x), but found 0x%x\n", MATROSKA_ID_EDITIONENTRY, id);
2283 case EBML_ID_VOID:
2284 res = ebml_read_skip(matroska);
2285 break;
2288 if (matroska->level_up) {
2289 matroska->level_up--;
2290 break;
2294 return res;
2297 #define ARRAY_SIZE(x) (sizeof(x)/sizeof(*x))
2299 static int
2300 matroska_aac_profile (char *codec_id)
2302 static const char *aac_profiles[] = {
2303 "MAIN", "LC", "SSR"
2305 int profile;
2307 for (profile=0; profile<ARRAY_SIZE(aac_profiles); profile++)
2308 if (strstr(codec_id, aac_profiles[profile]))
2309 break;
2310 return profile + 1;
2313 static int
2314 matroska_aac_sri (int samplerate)
2316 int sri;
2318 for (sri=0; sri<ARRAY_SIZE(ff_mpeg4audio_sample_rates); sri++)
2319 if (ff_mpeg4audio_sample_rates[sri] == samplerate)
2320 break;
2321 return sri;
2324 static int
2325 matroska_read_header (AVFormatContext *s,
2326 AVFormatParameters *ap)
2328 MatroskaDemuxContext *matroska = s->priv_data;
2329 char *doctype;
2330 int version, last_level, res = 0;
2331 uint32_t id;
2333 matroska->ctx = s;
2335 /* First read the EBML header. */
2336 doctype = NULL;
2337 if ((res = ebml_read_header(matroska, &doctype, &version)) < 0)
2338 return res;
2339 if ((doctype == NULL) || strcmp(doctype, "matroska")) {
2340 av_log(matroska->ctx, AV_LOG_ERROR,
2341 "Wrong EBML doctype ('%s' != 'matroska').\n",
2342 doctype ? doctype : "(none)");
2343 if (doctype)
2344 av_free(doctype);
2345 return AVERROR_NOFMT;
2347 av_free(doctype);
2348 if (version > 2) {
2349 av_log(matroska->ctx, AV_LOG_ERROR,
2350 "Matroska demuxer version 2 too old for file version %d\n",
2351 version);
2352 return AVERROR_NOFMT;
2355 /* The next thing is a segment. */
2356 while (1) {
2357 if (!(id = ebml_peek_id(matroska, &last_level)))
2358 return AVERROR(EIO);
2359 if (id == MATROSKA_ID_SEGMENT)
2360 break;
2362 /* oi! */
2363 av_log(matroska->ctx, AV_LOG_INFO,
2364 "Expected a Segment ID (0x%x), but received 0x%x!\n",
2365 MATROSKA_ID_SEGMENT, id);
2366 if ((res = ebml_read_skip(matroska)) < 0)
2367 return res;
2370 /* We now have a Matroska segment.
2371 * Seeks are from the beginning of the segment,
2372 * after the segment ID/length. */
2373 if ((res = ebml_read_master(matroska, &id)) < 0)
2374 return res;
2375 matroska->segment_start = url_ftell(s->pb);
2377 matroska->time_scale = 1000000;
2378 /* we've found our segment, start reading the different contents in here */
2379 while (res == 0) {
2380 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2381 res = AVERROR(EIO);
2382 break;
2383 } else if (matroska->level_up) {
2384 matroska->level_up--;
2385 break;
2388 switch (id) {
2389 /* stream info */
2390 case MATROSKA_ID_INFO: {
2391 if ((res = ebml_read_master(matroska, &id)) < 0)
2392 break;
2393 res = matroska_parse_info(matroska);
2394 break;
2397 /* track info headers */
2398 case MATROSKA_ID_TRACKS: {
2399 if ((res = ebml_read_master(matroska, &id)) < 0)
2400 break;
2401 res = matroska_parse_tracks(matroska);
2402 break;
2405 /* stream index */
2406 case MATROSKA_ID_CUES: {
2407 if (!matroska->index_parsed) {
2408 if ((res = ebml_read_master(matroska, &id)) < 0)
2409 break;
2410 res = matroska_parse_index(matroska);
2411 } else
2412 res = ebml_read_skip(matroska);
2413 break;
2416 /* metadata */
2417 case MATROSKA_ID_TAGS: {
2418 if (!matroska->metadata_parsed) {
2419 if ((res = ebml_read_master(matroska, &id)) < 0)
2420 break;
2421 res = matroska_parse_metadata(matroska);
2422 } else
2423 res = ebml_read_skip(matroska);
2424 break;
2427 /* file index (if seekable, seek to Cues/Tags to parse it) */
2428 case MATROSKA_ID_SEEKHEAD: {
2429 if ((res = ebml_read_master(matroska, &id)) < 0)
2430 break;
2431 res = matroska_parse_seekhead(matroska);
2432 break;
2435 case MATROSKA_ID_ATTACHMENTS: {
2436 if ((res = ebml_read_master(matroska, &id)) < 0)
2437 break;
2438 res = matroska_parse_attachments(s);
2439 break;
2442 case MATROSKA_ID_CLUSTER: {
2443 /* Do not read the master - this will be done in the next
2444 * call to matroska_read_packet. */
2445 res = 1;
2446 break;
2449 case MATROSKA_ID_CHAPTERS: {
2450 if ((res = ebml_read_master(matroska, &id)) < 0)
2451 return res;
2452 res = matroska_parse_chapters(s);
2453 break;
2456 default:
2457 av_log(matroska->ctx, AV_LOG_INFO,
2458 "Unknown matroska file header ID 0x%x\n", id);
2459 /* fall-through */
2461 case EBML_ID_VOID:
2462 res = ebml_read_skip(matroska);
2463 break;
2466 if (matroska->level_up) {
2467 matroska->level_up--;
2468 break;
2472 /* Have we found a cluster? */
2473 if (ebml_peek_id(matroska, NULL) == MATROSKA_ID_CLUSTER) {
2474 int i, j;
2475 MatroskaTrack *track;
2476 AVStream *st;
2478 for (i = 0; i < matroska->num_tracks; i++) {
2479 enum CodecID codec_id = CODEC_ID_NONE;
2480 uint8_t *extradata = NULL;
2481 int extradata_size = 0;
2482 int extradata_offset = 0;
2483 track = matroska->tracks[i];
2484 track->stream_index = -1;
2486 /* Apply some sanity checks. */
2487 if (track->codec_id == NULL)
2488 continue;
2490 for(j=0; ff_mkv_codec_tags[j].id != CODEC_ID_NONE; j++){
2491 if(!strncmp(ff_mkv_codec_tags[j].str, track->codec_id,
2492 strlen(ff_mkv_codec_tags[j].str))){
2493 codec_id= ff_mkv_codec_tags[j].id;
2494 break;
2498 /* Set the FourCC from the CodecID. */
2499 /* This is the MS compatibility mode which stores a
2500 * BITMAPINFOHEADER in the CodecPrivate. */
2501 if (!strcmp(track->codec_id,
2502 MATROSKA_CODEC_ID_VIDEO_VFW_FOURCC) &&
2503 (track->codec_priv_size >= 40) &&
2504 (track->codec_priv != NULL)) {
2505 MatroskaVideoTrack *vtrack = (MatroskaVideoTrack *) track;
2507 /* Offset of biCompression. Stored in LE. */
2508 vtrack->fourcc = AV_RL32(track->codec_priv + 16);
2509 codec_id = codec_get_id(codec_bmp_tags, vtrack->fourcc);
2513 /* This is the MS compatibility mode which stores a
2514 * WAVEFORMATEX in the CodecPrivate. */
2515 else if (!strcmp(track->codec_id,
2516 MATROSKA_CODEC_ID_AUDIO_ACM) &&
2517 (track->codec_priv_size >= 18) &&
2518 (track->codec_priv != NULL)) {
2519 uint16_t tag;
2521 /* Offset of wFormatTag. Stored in LE. */
2522 tag = AV_RL16(track->codec_priv);
2523 codec_id = codec_get_id(codec_wav_tags, tag);
2527 else if (codec_id == CODEC_ID_AAC && !track->codec_priv_size) {
2528 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *) track;
2529 int profile = matroska_aac_profile(track->codec_id);
2530 int sri = matroska_aac_sri(audiotrack->internal_samplerate);
2531 extradata = av_malloc(5);
2532 if (extradata == NULL)
2533 return AVERROR(ENOMEM);
2534 extradata[0] = (profile << 3) | ((sri&0x0E) >> 1);
2535 extradata[1] = ((sri&0x01) << 7) | (audiotrack->channels<<3);
2536 if (strstr(track->codec_id, "SBR")) {
2537 sri = matroska_aac_sri(audiotrack->samplerate);
2538 extradata[2] = 0x56;
2539 extradata[3] = 0xE5;
2540 extradata[4] = 0x80 | (sri<<3);
2541 extradata_size = 5;
2542 } else {
2543 extradata_size = 2;
2547 else if (codec_id == CODEC_ID_TTA) {
2548 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *) track;
2549 ByteIOContext b;
2550 extradata_size = 30;
2551 extradata = av_mallocz(extradata_size);
2552 if (extradata == NULL)
2553 return AVERROR(ENOMEM);
2554 init_put_byte(&b, extradata, extradata_size, 1,
2555 NULL, NULL, NULL, NULL);
2556 put_buffer(&b, "TTA1", 4);
2557 put_le16(&b, 1);
2558 put_le16(&b, audiotrack->channels);
2559 put_le16(&b, audiotrack->bitdepth);
2560 put_le32(&b, audiotrack->samplerate);
2561 put_le32(&b, matroska->ctx->duration * audiotrack->samplerate);
2564 else if (codec_id == CODEC_ID_RV10 || codec_id == CODEC_ID_RV20 ||
2565 codec_id == CODEC_ID_RV30 || codec_id == CODEC_ID_RV40) {
2566 extradata_offset = 26;
2567 track->codec_priv_size -= extradata_offset;
2570 else if (codec_id == CODEC_ID_RA_144) {
2571 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)track;
2572 audiotrack->samplerate = 8000;
2573 audiotrack->channels = 1;
2576 else if (codec_id == CODEC_ID_RA_288 ||
2577 codec_id == CODEC_ID_COOK ||
2578 codec_id == CODEC_ID_ATRAC3) {
2579 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)track;
2580 ByteIOContext b;
2582 init_put_byte(&b, track->codec_priv, track->codec_priv_size, 0,
2583 NULL, NULL, NULL, NULL);
2584 url_fskip(&b, 24);
2585 audiotrack->coded_framesize = get_be32(&b);
2586 url_fskip(&b, 12);
2587 audiotrack->sub_packet_h = get_be16(&b);
2588 audiotrack->frame_size = get_be16(&b);
2589 audiotrack->sub_packet_size = get_be16(&b);
2590 audiotrack->buf = av_malloc(audiotrack->frame_size * audiotrack->sub_packet_h);
2591 if (codec_id == CODEC_ID_RA_288) {
2592 audiotrack->block_align = audiotrack->coded_framesize;
2593 track->codec_priv_size = 0;
2594 } else {
2595 audiotrack->block_align = audiotrack->sub_packet_size;
2596 extradata_offset = 78;
2597 track->codec_priv_size -= extradata_offset;
2601 if (codec_id == CODEC_ID_NONE) {
2602 av_log(matroska->ctx, AV_LOG_INFO,
2603 "Unknown/unsupported CodecID %s.\n",
2604 track->codec_id);
2607 track->stream_index = matroska->num_streams;
2609 matroska->num_streams++;
2610 st = av_new_stream(s, track->stream_index);
2611 if (st == NULL)
2612 return AVERROR(ENOMEM);
2613 av_set_pts_info(st, 64, matroska->time_scale, 1000*1000*1000); /* 64 bit pts in ns */
2615 st->codec->codec_id = codec_id;
2616 st->start_time = 0;
2617 if (strcmp(track->language, "und"))
2618 strcpy(st->language, track->language);
2620 if (track->flags & MATROSKA_TRACK_DEFAULT)
2621 st->disposition |= AV_DISPOSITION_DEFAULT;
2623 if (track->default_duration)
2624 av_reduce(&st->codec->time_base.num, &st->codec->time_base.den,
2625 track->default_duration, 1000000000, 30000);
2627 if(extradata){
2628 st->codec->extradata = extradata;
2629 st->codec->extradata_size = extradata_size;
2630 } else if(track->codec_priv && track->codec_priv_size > 0){
2631 st->codec->extradata = av_malloc(track->codec_priv_size);
2632 if(st->codec->extradata == NULL)
2633 return AVERROR(ENOMEM);
2634 st->codec->extradata_size = track->codec_priv_size;
2635 memcpy(st->codec->extradata,track->codec_priv+extradata_offset,
2636 track->codec_priv_size);
2639 if (track->type == MATROSKA_TRACK_TYPE_VIDEO) {
2640 MatroskaVideoTrack *videotrack = (MatroskaVideoTrack *)track;
2642 st->codec->codec_type = CODEC_TYPE_VIDEO;
2643 st->codec->codec_tag = videotrack->fourcc;
2644 st->codec->width = videotrack->pixel_width;
2645 st->codec->height = videotrack->pixel_height;
2646 if (videotrack->display_width == 0)
2647 videotrack->display_width= videotrack->pixel_width;
2648 if (videotrack->display_height == 0)
2649 videotrack->display_height= videotrack->pixel_height;
2650 av_reduce(&st->codec->sample_aspect_ratio.num,
2651 &st->codec->sample_aspect_ratio.den,
2652 st->codec->height * videotrack->display_width,
2653 st->codec-> width * videotrack->display_height,
2654 255);
2655 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2656 } else if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
2657 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)track;
2659 st->codec->codec_type = CODEC_TYPE_AUDIO;
2660 st->codec->sample_rate = audiotrack->samplerate;
2661 st->codec->channels = audiotrack->channels;
2662 st->codec->block_align = audiotrack->block_align;
2663 } else if (track->type == MATROSKA_TRACK_TYPE_SUBTITLE) {
2664 st->codec->codec_type = CODEC_TYPE_SUBTITLE;
2667 /* What do we do with private data? E.g. for Vorbis. */
2669 res = 0;
2672 if (matroska->index_parsed) {
2673 int i, track, stream;
2674 for (i=0; i<matroska->num_indexes; i++) {
2675 MatroskaDemuxIndex *idx = &matroska->index[i];
2676 track = matroska_find_track_by_num(matroska, idx->track);
2677 if (track < 0) continue;
2678 stream = matroska->tracks[track]->stream_index;
2679 if (stream >= 0 && stream < matroska->ctx->nb_streams)
2680 av_add_index_entry(matroska->ctx->streams[stream],
2681 idx->pos, idx->time/matroska->time_scale,
2682 0, 0, AVINDEX_KEYFRAME);
2686 return res;
2689 static int
2690 matroska_parse_block(MatroskaDemuxContext *matroska, uint8_t *data, int size,
2691 int64_t pos, uint64_t cluster_time, uint64_t duration,
2692 int is_keyframe, int is_bframe)
2694 int res = 0;
2695 int track;
2696 AVStream *st;
2697 AVPacket *pkt;
2698 uint8_t *origdata = data;
2699 int16_t block_time;
2700 uint32_t *lace_size = NULL;
2701 int n, flags, laces = 0;
2702 uint64_t num;
2703 int stream_index;
2705 /* first byte(s): tracknum */
2706 if ((n = matroska_ebmlnum_uint(data, size, &num)) < 0) {
2707 av_log(matroska->ctx, AV_LOG_ERROR, "EBML block data error\n");
2708 av_free(origdata);
2709 return res;
2711 data += n;
2712 size -= n;
2714 /* fetch track from num */
2715 track = matroska_find_track_by_num(matroska, num);
2716 if (size <= 3 || track < 0 || track >= matroska->num_tracks) {
2717 av_log(matroska->ctx, AV_LOG_INFO,
2718 "Invalid stream %d or size %u\n", track, size);
2719 av_free(origdata);
2720 return res;
2722 stream_index = matroska->tracks[track]->stream_index;
2723 if (stream_index < 0 || stream_index >= matroska->ctx->nb_streams) {
2724 av_free(origdata);
2725 return res;
2727 st = matroska->ctx->streams[stream_index];
2728 if (st->discard >= AVDISCARD_ALL) {
2729 av_free(origdata);
2730 return res;
2732 if (duration == AV_NOPTS_VALUE)
2733 duration = matroska->tracks[track]->default_duration / matroska->time_scale;
2735 /* block_time (relative to cluster time) */
2736 block_time = AV_RB16(data);
2737 data += 2;
2738 flags = *data++;
2739 size -= 3;
2740 if (is_keyframe == -1)
2741 is_keyframe = flags & 0x80 ? PKT_FLAG_KEY : 0;
2743 if (matroska->skip_to_keyframe) {
2744 if (!is_keyframe || st != matroska->skip_to_stream) {
2745 av_free(origdata);
2746 return res;
2748 matroska->skip_to_keyframe = 0;
2751 switch ((flags & 0x06) >> 1) {
2752 case 0x0: /* no lacing */
2753 laces = 1;
2754 lace_size = av_mallocz(sizeof(int));
2755 lace_size[0] = size;
2756 break;
2758 case 0x1: /* xiph lacing */
2759 case 0x2: /* fixed-size lacing */
2760 case 0x3: /* EBML lacing */
2761 if (size == 0) {
2762 res = -1;
2763 break;
2765 laces = (*data) + 1;
2766 data += 1;
2767 size -= 1;
2768 lace_size = av_mallocz(laces * sizeof(int));
2770 switch ((flags & 0x06) >> 1) {
2771 case 0x1: /* xiph lacing */ {
2772 uint8_t temp;
2773 uint32_t total = 0;
2774 for (n = 0; res == 0 && n < laces - 1; n++) {
2775 while (1) {
2776 if (size == 0) {
2777 res = -1;
2778 break;
2780 temp = *data;
2781 lace_size[n] += temp;
2782 data += 1;
2783 size -= 1;
2784 if (temp != 0xff)
2785 break;
2787 total += lace_size[n];
2789 lace_size[n] = size - total;
2790 break;
2793 case 0x2: /* fixed-size lacing */
2794 for (n = 0; n < laces; n++)
2795 lace_size[n] = size / laces;
2796 break;
2798 case 0x3: /* EBML lacing */ {
2799 uint32_t total;
2800 n = matroska_ebmlnum_uint(data, size, &num);
2801 if (n < 0) {
2802 av_log(matroska->ctx, AV_LOG_INFO,
2803 "EBML block data error\n");
2804 break;
2806 data += n;
2807 size -= n;
2808 total = lace_size[0] = num;
2809 for (n = 1; res == 0 && n < laces - 1; n++) {
2810 int64_t snum;
2811 int r;
2812 r = matroska_ebmlnum_sint (data, size, &snum);
2813 if (r < 0) {
2814 av_log(matroska->ctx, AV_LOG_INFO,
2815 "EBML block data error\n");
2816 break;
2818 data += r;
2819 size -= r;
2820 lace_size[n] = lace_size[n - 1] + snum;
2821 total += lace_size[n];
2823 lace_size[n] = size - total;
2824 break;
2827 break;
2830 if (res == 0) {
2831 uint64_t timecode = AV_NOPTS_VALUE;
2833 if (cluster_time != (uint64_t)-1
2834 && (block_time >= 0 || cluster_time >= -block_time))
2835 timecode = cluster_time + block_time;
2837 for (n = 0; n < laces; n++) {
2838 if (st->codec->codec_id == CODEC_ID_RA_288 ||
2839 st->codec->codec_id == CODEC_ID_COOK ||
2840 st->codec->codec_id == CODEC_ID_ATRAC3) {
2841 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)matroska->tracks[track];
2842 int a = st->codec->block_align;
2843 int sps = audiotrack->sub_packet_size;
2844 int cfs = audiotrack->coded_framesize;
2845 int h = audiotrack->sub_packet_h;
2846 int y = audiotrack->sub_packet_cnt;
2847 int w = audiotrack->frame_size;
2848 int x;
2850 if (!audiotrack->pkt_cnt) {
2851 if (st->codec->codec_id == CODEC_ID_RA_288)
2852 for (x=0; x<h/2; x++)
2853 memcpy(audiotrack->buf+x*2*w+y*cfs,
2854 data+x*cfs, cfs);
2855 else
2856 for (x=0; x<w/sps; x++)
2857 memcpy(audiotrack->buf+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), data+x*sps, sps);
2859 if (++audiotrack->sub_packet_cnt >= h) {
2860 audiotrack->sub_packet_cnt = 0;
2861 audiotrack->pkt_cnt = h*w / a;
2864 while (audiotrack->pkt_cnt) {
2865 pkt = av_mallocz(sizeof(AVPacket));
2866 av_new_packet(pkt, a);
2867 memcpy(pkt->data, audiotrack->buf
2868 + a * (h*w / a - audiotrack->pkt_cnt--), a);
2869 pkt->pos = pos;
2870 pkt->stream_index = stream_index;
2871 matroska_queue_packet(matroska, pkt);
2873 } else {
2874 int result, offset = 0, ilen, olen, pkt_size = lace_size[n];
2875 uint8_t *pkt_data = data;
2877 if (matroska->tracks[track]->encoding_scope & 1) {
2878 switch (matroska->tracks[track]->encoding_algo) {
2879 case MATROSKA_TRACK_ENCODING_COMP_HEADERSTRIP:
2880 offset = matroska->tracks[track]->encoding_settings_len;
2881 break;
2882 case MATROSKA_TRACK_ENCODING_COMP_LZO:
2883 pkt_data = NULL;
2884 do {
2885 ilen = lace_size[n];
2886 olen = pkt_size *= 3;
2887 pkt_data = av_realloc(pkt_data,
2888 pkt_size+LZO_OUTPUT_PADDING);
2889 result = lzo1x_decode(pkt_data, &olen, data, &ilen);
2890 } while (result==LZO_OUTPUT_FULL && pkt_size<10000000);
2891 if (result) {
2892 av_free(pkt_data);
2893 continue;
2895 pkt_size -= olen;
2896 break;
2897 #ifdef CONFIG_ZLIB
2898 case MATROSKA_TRACK_ENCODING_COMP_ZLIB: {
2899 z_stream zstream = {0};
2900 pkt_data = NULL;
2901 if (inflateInit(&zstream) != Z_OK)
2902 continue;
2903 zstream.next_in = data;
2904 zstream.avail_in = lace_size[n];
2905 do {
2906 pkt_size *= 3;
2907 pkt_data = av_realloc(pkt_data, pkt_size);
2908 zstream.avail_out = pkt_size - zstream.total_out;
2909 zstream.next_out = pkt_data + zstream.total_out;
2910 result = inflate(&zstream, Z_NO_FLUSH);
2911 } while (result==Z_OK && pkt_size<10000000);
2912 pkt_size = zstream.total_out;
2913 inflateEnd(&zstream);
2914 if (result != Z_STREAM_END) {
2915 av_free(pkt_data);
2916 continue;
2918 break;
2920 #endif
2921 #ifdef CONFIG_BZLIB
2922 case MATROSKA_TRACK_ENCODING_COMP_BZLIB: {
2923 bz_stream bzstream = {0};
2924 pkt_data = NULL;
2925 if (BZ2_bzDecompressInit(&bzstream, 0, 0) != BZ_OK)
2926 continue;
2927 bzstream.next_in = data;
2928 bzstream.avail_in = lace_size[n];
2929 do {
2930 pkt_size *= 3;
2931 pkt_data = av_realloc(pkt_data, pkt_size);
2932 bzstream.avail_out = pkt_size - bzstream.total_out_lo32;
2933 bzstream.next_out = pkt_data + bzstream.total_out_lo32;
2934 result = BZ2_bzDecompress(&bzstream);
2935 } while (result==BZ_OK && pkt_size<10000000);
2936 pkt_size = bzstream.total_out_lo32;
2937 BZ2_bzDecompressEnd(&bzstream);
2938 if (result != BZ_STREAM_END) {
2939 av_free(pkt_data);
2940 continue;
2942 break;
2944 #endif
2948 pkt = av_mallocz(sizeof(AVPacket));
2949 /* XXX: prevent data copy... */
2950 if (av_new_packet(pkt, pkt_size+offset) < 0) {
2951 res = AVERROR(ENOMEM);
2952 n = laces-1;
2953 break;
2955 if (offset)
2956 memcpy (pkt->data, matroska->tracks[track]->encoding_settings, offset);
2957 memcpy (pkt->data+offset, pkt_data, pkt_size);
2959 if (n == 0)
2960 pkt->flags = is_keyframe;
2961 pkt->stream_index = stream_index;
2963 pkt->pts = timecode;
2964 pkt->pos = pos;
2965 pkt->duration = duration;
2967 matroska_queue_packet(matroska, pkt);
2970 if (timecode != AV_NOPTS_VALUE)
2971 timecode = duration ? timecode + duration : AV_NOPTS_VALUE;
2972 data += lace_size[n];
2976 av_free(lace_size);
2977 av_free(origdata);
2978 return res;
2981 static int
2982 matroska_parse_blockgroup (MatroskaDemuxContext *matroska,
2983 uint64_t cluster_time)
2985 int res = 0;
2986 uint32_t id;
2987 int is_bframe = 0;
2988 int is_keyframe = PKT_FLAG_KEY, last_num_packets = matroska->num_packets;
2989 uint64_t duration = AV_NOPTS_VALUE;
2990 uint8_t *data;
2991 int size = 0;
2992 int64_t pos = 0;
2994 av_log(matroska->ctx, AV_LOG_DEBUG, "parsing blockgroup...\n");
2996 while (res == 0) {
2997 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
2998 res = AVERROR(EIO);
2999 break;
3000 } else if (matroska->level_up) {
3001 matroska->level_up--;
3002 break;
3005 switch (id) {
3006 /* one block inside the group. Note, block parsing is one
3007 * of the harder things, so this code is a bit complicated.
3008 * See http://www.matroska.org/ for documentation. */
3009 case MATROSKA_ID_BLOCK: {
3010 pos = url_ftell(matroska->ctx->pb);
3011 res = ebml_read_binary(matroska, &id, &data, &size);
3012 break;
3015 case MATROSKA_ID_BLOCKDURATION: {
3016 if ((res = ebml_read_uint(matroska, &id, &duration)) < 0)
3017 break;
3018 break;
3021 case MATROSKA_ID_BLOCKREFERENCE: {
3022 int64_t num;
3023 /* We've found a reference, so not even the first frame in
3024 * the lace is a key frame. */
3025 is_keyframe = 0;
3026 if (last_num_packets != matroska->num_packets)
3027 matroska->packets[last_num_packets]->flags = 0;
3028 if ((res = ebml_read_sint(matroska, &id, &num)) < 0)
3029 break;
3030 if (num > 0)
3031 is_bframe = 1;
3032 break;
3035 default:
3036 av_log(matroska->ctx, AV_LOG_INFO,
3037 "Unknown entry 0x%x in blockgroup data\n", id);
3038 /* fall-through */
3040 case EBML_ID_VOID:
3041 res = ebml_read_skip(matroska);
3042 break;
3045 if (matroska->level_up) {
3046 matroska->level_up--;
3047 break;
3051 if (res)
3052 return res;
3054 if (size > 0)
3055 res = matroska_parse_block(matroska, data, size, pos, cluster_time,
3056 duration, is_keyframe, is_bframe);
3058 return res;
3061 static int
3062 matroska_parse_cluster (MatroskaDemuxContext *matroska)
3064 int res = 0;
3065 uint32_t id;
3066 uint64_t cluster_time = 0;
3067 uint8_t *data;
3068 int64_t pos;
3069 int size;
3071 av_log(matroska->ctx, AV_LOG_DEBUG,
3072 "parsing cluster at %"PRId64"\n", url_ftell(matroska->ctx->pb));
3074 while (res == 0) {
3075 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
3076 res = AVERROR(EIO);
3077 break;
3078 } else if (matroska->level_up) {
3079 matroska->level_up--;
3080 break;
3083 switch (id) {
3084 /* cluster timecode */
3085 case MATROSKA_ID_CLUSTERTIMECODE: {
3086 uint64_t num;
3087 if ((res = ebml_read_uint(matroska, &id, &num)) < 0)
3088 break;
3089 cluster_time = num;
3090 break;
3093 /* a group of blocks inside a cluster */
3094 case MATROSKA_ID_BLOCKGROUP:
3095 if ((res = ebml_read_master(matroska, &id)) < 0)
3096 break;
3097 res = matroska_parse_blockgroup(matroska, cluster_time);
3098 break;
3100 case MATROSKA_ID_SIMPLEBLOCK:
3101 pos = url_ftell(matroska->ctx->pb);
3102 res = ebml_read_binary(matroska, &id, &data, &size);
3103 if (res == 0)
3104 res = matroska_parse_block(matroska, data, size, pos,
3105 cluster_time, AV_NOPTS_VALUE,
3106 -1, 0);
3107 break;
3109 default:
3110 av_log(matroska->ctx, AV_LOG_INFO,
3111 "Unknown entry 0x%x in cluster data\n", id);
3112 /* fall-through */
3114 case EBML_ID_VOID:
3115 res = ebml_read_skip(matroska);
3116 break;
3119 if (matroska->level_up) {
3120 matroska->level_up--;
3121 break;
3125 return res;
3128 static int
3129 matroska_read_packet (AVFormatContext *s,
3130 AVPacket *pkt)
3132 MatroskaDemuxContext *matroska = s->priv_data;
3133 int res;
3134 uint32_t id;
3136 /* Read stream until we have a packet queued. */
3137 while (matroska_deliver_packet(matroska, pkt)) {
3139 /* Have we already reached the end? */
3140 if (matroska->done)
3141 return AVERROR(EIO);
3143 res = 0;
3144 while (res == 0) {
3145 if (!(id = ebml_peek_id(matroska, &matroska->level_up))) {
3146 return AVERROR(EIO);
3147 } else if (matroska->level_up) {
3148 matroska->level_up--;
3149 break;
3152 switch (id) {
3153 case MATROSKA_ID_CLUSTER:
3154 if ((res = ebml_read_master(matroska, &id)) < 0)
3155 break;
3156 if ((res = matroska_parse_cluster(matroska)) == 0)
3157 res = 1; /* Parsed one cluster, let's get out. */
3158 break;
3160 default:
3161 case EBML_ID_VOID:
3162 res = ebml_read_skip(matroska);
3163 break;
3166 if (matroska->level_up) {
3167 matroska->level_up--;
3168 break;
3172 if (res == -1)
3173 matroska->done = 1;
3176 return 0;
3179 static int
3180 matroska_read_seek (AVFormatContext *s, int stream_index, int64_t timestamp,
3181 int flags)
3183 MatroskaDemuxContext *matroska = s->priv_data;
3184 AVStream *st = s->streams[stream_index];
3185 int index;
3187 /* find index entry */
3188 index = av_index_search_timestamp(st, timestamp, flags);
3189 if (index < 0)
3190 return 0;
3192 matroska_clear_queue(matroska);
3194 /* do the seek */
3195 url_fseek(s->pb, st->index_entries[index].pos, SEEK_SET);
3196 matroska->skip_to_keyframe = !(flags & AVSEEK_FLAG_ANY);
3197 matroska->skip_to_stream = st;
3198 matroska->peek_id = 0;
3199 return 0;
3202 static int
3203 matroska_read_close (AVFormatContext *s)
3205 MatroskaDemuxContext *matroska = s->priv_data;
3206 int n = 0;
3208 av_free(matroska->writing_app);
3209 av_free(matroska->muxing_app);
3210 av_free(matroska->index);
3212 matroska_clear_queue(matroska);
3214 for (n = 0; n < matroska->num_tracks; n++) {
3215 MatroskaTrack *track = matroska->tracks[n];
3216 av_free(track->codec_id);
3217 av_free(track->codec_name);
3218 av_free(track->codec_priv);
3219 av_free(track->name);
3221 if (track->type == MATROSKA_TRACK_TYPE_AUDIO) {
3222 MatroskaAudioTrack *audiotrack = (MatroskaAudioTrack *)track;
3223 av_free(audiotrack->buf);
3226 av_free(track);
3229 return 0;
3232 AVInputFormat matroska_demuxer = {
3233 "matroska",
3234 "Matroska file format",
3235 sizeof(MatroskaDemuxContext),
3236 matroska_probe,
3237 matroska_read_header,
3238 matroska_read_packet,
3239 matroska_read_close,
3240 matroska_read_seek,