3 * Copyright (c) 2001 Fabrice Bellard
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
26 #include "libavutil/intreadwrite.h"
27 #include "libavutil/avstring.h"
31 #include "libavcodec/mpeg4audio.h"
32 #include "libavcodec/mpegaudiodata.h"
39 * First version by Francois Revol revol@free.fr
40 * Seek function by Gael Chardon gael.dev@4now.net
42 * Features and limitations:
43 * - reads most of the QT files I have (at least the structure),
44 * Sample QuickTime files with mp3 audio can be found at: http://www.3ivx.com/showcase.html
45 * - the code is quite ugly... maybe I won't do it recursive next time :-)
47 * Funny I didn't know about http://sourceforge.net/projects/qt-ffmpeg/
48 * when coding this :) (it's a writer anyway)
50 * Reference documents:
51 * http://www.geocities.com/xhelmboyx/quicktime/formats/qtm-layout.txt
53 * http://developer.apple.com/documentation/QuickTime/QTFF/
54 * http://developer.apple.com/documentation/QuickTime/QTFF/qtff.pdf
55 * QuickTime is a trademark of Apple (AFAIK :))
58 #include "qtpalette.h"
64 /* XXX: it's the first time I make a recursive parser I think... sorry if it's ugly :P */
66 /* those functions parse an atom */
68 0: continue to parse next atom
69 <0: error occurred, exit
71 /* links atom IDs to parse functions */
72 typedef struct MOVParseTableEntry {
74 int (*parse)(MOVContext *ctx, ByteIOContext *pb, MOVAtom atom);
77 static const MOVParseTableEntry mov_default_parse_table[];
79 static int mov_read_default(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
81 int64_t total_size = 0;
86 a.offset = atom.offset;
89 atom.size = INT64_MAX;
90 while(((total_size + 8) < atom.size) && !url_feof(pb) && !err) {
94 a.size = get_be32(pb);
95 a.type = get_le32(pb);
99 dprintf(c->fc, "type: %08x %.4s sz: %"PRIx64" %"PRIx64" %"PRIx64"\n",
100 a.type, (char*)&a.type, a.size, atom.size, total_size);
101 if (a.size == 1) { /* 64 bit extended size */
102 a.size = get_be64(pb) - 8;
107 a.size = atom.size - total_size;
114 a.size = FFMIN(a.size, atom.size - total_size);
116 for (i = 0; mov_default_parse_table[i].type != 0
117 && mov_default_parse_table[i].type != a.type; i++)
120 if (mov_default_parse_table[i].type == 0) { /* skip leaf atoms data */
121 url_fskip(pb, a.size);
123 int64_t start_pos = url_ftell(pb);
125 err = mov_default_parse_table[i].parse(c, pb, a);
126 if (url_is_streamed(pb) && c->found_moov && c->found_mdat)
128 left = a.size - url_ftell(pb) + start_pos;
129 if (left > 0) /* skip garbage at atom end */
134 total_size += a.size;
137 if (!err && total_size < atom.size && atom.size < 0x7ffff)
138 url_fskip(pb, atom.size - total_size);
143 static int mov_read_dref(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
145 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
146 MOVStreamContext *sc = st->priv_data;
149 get_be32(pb); // version + flags
150 entries = get_be32(pb);
151 if (entries >= UINT_MAX / sizeof(*sc->drefs))
153 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
155 return AVERROR(ENOMEM);
156 sc->drefs_count = entries;
158 for (i = 0; i < sc->drefs_count; i++) {
159 MOVDref *dref = &sc->drefs[i];
160 uint32_t size = get_be32(pb);
161 int64_t next = url_ftell(pb) + size - 4;
163 dref->type = get_le32(pb);
164 get_be32(pb); // version + flags
165 dprintf(c->fc, "type %.4s size %d\n", (char*)&dref->type, size);
167 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
168 /* macintosh alias record */
169 uint16_t volume_len, len;
175 volume_len = get_byte(pb);
176 volume_len = FFMIN(volume_len, 27);
177 get_buffer(pb, volume, 27);
178 volume[volume_len] = 0;
179 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", volume, volume_len);
183 for (type = 0; type != -1 && url_ftell(pb) < next; ) {
186 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
189 if (type == 2) { // absolute path
191 dref->path = av_mallocz(len+1);
193 return AVERROR(ENOMEM);
194 get_buffer(pb, dref->path, len);
195 if (len > volume_len && !strncmp(dref->path, volume, volume_len)) {
197 memmove(dref->path, dref->path+volume_len, len);
200 for (j = 0; j < len; j++)
201 if (dref->path[j] == ':')
203 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
208 url_fseek(pb, next, SEEK_SET);
213 static int mov_read_hdlr(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
219 if (c->fc->nb_streams < 1) // meta before first trak
222 st = c->fc->streams[c->fc->nb_streams-1];
224 get_byte(pb); /* version */
225 get_be24(pb); /* flags */
228 ctype = get_le32(pb);
229 type = get_le32(pb); /* component subtype */
231 dprintf(c->fc, "ctype= %c%c%c%c (0x%08x)\n", *((char *)&ctype), ((char *)&ctype)[1],
232 ((char *)&ctype)[2], ((char *)&ctype)[3], (int) ctype);
233 dprintf(c->fc, "stype= %c%c%c%c\n",
234 *((char *)&type), ((char *)&type)[1], ((char *)&type)[2], ((char *)&type)[3]);
237 if (type == MKTAG('v','i','d','e'))
238 st->codec->codec_type = CODEC_TYPE_VIDEO;
239 else if(type == MKTAG('s','o','u','n'))
240 st->codec->codec_type = CODEC_TYPE_AUDIO;
241 else if(type == MKTAG('m','1','a',' '))
242 st->codec->codec_id = CODEC_ID_MP2;
243 else if(type == MKTAG('s','u','b','p')) {
244 st->codec->codec_type = CODEC_TYPE_SUBTITLE;
246 get_be32(pb); /* component manufacture */
247 get_be32(pb); /* component flags */
248 get_be32(pb); /* component flags mask */
251 return 0; /* nothing left to read */
253 url_fskip(pb, atom.size - (url_ftell(pb) - atom.offset));
257 static int mp4_read_descr_len(ByteIOContext *pb)
262 int c = get_byte(pb);
263 len = (len << 7) | (c & 0x7f);
270 static int mp4_read_descr(MOVContext *c, ByteIOContext *pb, int *tag)
274 len = mp4_read_descr_len(pb);
275 dprintf(c->fc, "MPEG4 description: tag=0x%02x len=%d\n", *tag, len);
279 #define MP4ESDescrTag 0x03
280 #define MP4DecConfigDescrTag 0x04
281 #define MP4DecSpecificDescrTag 0x05
283 static const AVCodecTag mp4_audio_types[] = {
284 { CODEC_ID_MP3ON4, 29 }, /* old mp3on4 draft */
285 { CODEC_ID_MP3ON4, 32 }, /* layer 1 */
286 { CODEC_ID_MP3ON4, 33 }, /* layer 2 */
287 { CODEC_ID_MP3ON4, 34 }, /* layer 3 */
288 { CODEC_ID_NONE, 0 },
291 static int mov_read_esds(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
293 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
296 get_be32(pb); /* version + flags */
297 len = mp4_read_descr(c, pb, &tag);
298 if (tag == MP4ESDescrTag) {
299 get_be16(pb); /* ID */
300 get_byte(pb); /* priority */
302 get_be16(pb); /* ID */
304 len = mp4_read_descr(c, pb, &tag);
305 if (tag == MP4DecConfigDescrTag) {
306 int object_type_id = get_byte(pb);
307 get_byte(pb); /* stream type */
308 get_be24(pb); /* buffer size db */
309 get_be32(pb); /* max bitrate */
310 get_be32(pb); /* avg bitrate */
312 st->codec->codec_id= codec_get_id(ff_mp4_obj_type, object_type_id);
313 dprintf(c->fc, "esds object type id %d\n", object_type_id);
314 len = mp4_read_descr(c, pb, &tag);
315 if (tag == MP4DecSpecificDescrTag) {
316 dprintf(c->fc, "Specific MPEG4 header len=%d\n", len);
317 if((uint64_t)len > (1<<30))
319 st->codec->extradata = av_mallocz(len + FF_INPUT_BUFFER_PADDING_SIZE);
320 if (!st->codec->extradata)
321 return AVERROR(ENOMEM);
322 get_buffer(pb, st->codec->extradata, len);
323 st->codec->extradata_size = len;
324 if (st->codec->codec_id == CODEC_ID_AAC) {
325 MPEG4AudioConfig cfg;
326 ff_mpeg4audio_get_config(&cfg, st->codec->extradata,
327 st->codec->extradata_size);
328 if (cfg.chan_config > 7)
330 st->codec->channels = ff_mpeg4audio_channels[cfg.chan_config];
331 if (cfg.object_type == 29 && cfg.sampling_index < 3) // old mp3on4
332 st->codec->sample_rate = ff_mpa_freq_tab[cfg.sampling_index];
334 st->codec->sample_rate = cfg.sample_rate; // ext sample rate ?
335 dprintf(c->fc, "mp4a config channels %d obj %d ext obj %d "
336 "sample rate %d ext sample rate %d\n", st->codec->channels,
337 cfg.object_type, cfg.ext_object_type,
338 cfg.sample_rate, cfg.ext_sample_rate);
339 if (!(st->codec->codec_id = codec_get_id(mp4_audio_types,
341 st->codec->codec_id = CODEC_ID_AAC;
348 static int mov_read_pasp(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
350 const int num = get_be32(pb);
351 const int den = get_be32(pb);
352 AVStream * const st = c->fc->streams[c->fc->nb_streams-1];
354 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
355 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num))
356 av_log(c->fc, AV_LOG_WARNING,
357 "sample aspect ratio already set to %d:%d, overriding by 'pasp' atom\n",
358 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den);
359 st->sample_aspect_ratio.num = num;
360 st->sample_aspect_ratio.den = den;
365 /* this atom contains actual media data */
366 static int mov_read_mdat(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
368 if(atom.size == 0) /* wrong one (MP4) */
371 return 0; /* now go for moov */
374 static int mov_read_ftyp(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
376 uint32_t type = get_le32(pb);
378 if (type != MKTAG('q','t',' ',' '))
380 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
381 get_be32(pb); /* minor version */
382 url_fskip(pb, atom.size - 8);
386 /* this atom should contain all header atoms */
387 static int mov_read_moov(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
389 if (mov_read_default(c, pb, atom) < 0)
391 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
392 /* so we don't parse the whole file if over a network */
394 return 0; /* now go for mdat */
397 static int mov_read_moof(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
399 c->fragment.moof_offset = url_ftell(pb) - 8;
400 dprintf(c->fc, "moof offset %llx\n", c->fragment.moof_offset);
401 return mov_read_default(c, pb, atom);
404 static int mov_read_mdhd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
406 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
407 MOVStreamContext *sc = st->priv_data;
408 int version = get_byte(pb);
409 char language[4] = {0};
413 return -1; /* unsupported */
415 get_be24(pb); /* flags */
420 get_be32(pb); /* creation time */
421 get_be32(pb); /* modification time */
424 sc->time_scale = get_be32(pb);
425 st->duration = (version == 1) ? get_be64(pb) : get_be32(pb); /* duration */
427 lang = get_be16(pb); /* language */
428 if (ff_mov_lang_to_iso639(lang, language))
429 av_metadata_set(&st->metadata, "language", language);
430 get_be16(pb); /* quality */
435 static int mov_read_mvhd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
437 int version = get_byte(pb); /* version */
438 get_be24(pb); /* flags */
444 get_be32(pb); /* creation time */
445 get_be32(pb); /* modification time */
447 c->time_scale = get_be32(pb); /* time scale */
449 dprintf(c->fc, "time scale = %i\n", c->time_scale);
451 c->duration = (version == 1) ? get_be64(pb) : get_be32(pb); /* duration */
452 get_be32(pb); /* preferred scale */
454 get_be16(pb); /* preferred volume */
456 url_fskip(pb, 10); /* reserved */
458 url_fskip(pb, 36); /* display matrix */
460 get_be32(pb); /* preview time */
461 get_be32(pb); /* preview duration */
462 get_be32(pb); /* poster time */
463 get_be32(pb); /* selection time */
464 get_be32(pb); /* selection duration */
465 get_be32(pb); /* current time */
466 get_be32(pb); /* next track ID */
471 static int mov_read_smi(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
473 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
475 if((uint64_t)atom.size > (1<<30))
478 // currently SVQ3 decoder expect full STSD header - so let's fake it
479 // this should be fixed and just SMI header should be passed
480 av_free(st->codec->extradata);
481 st->codec->extradata = av_mallocz(atom.size + 0x5a + FF_INPUT_BUFFER_PADDING_SIZE);
482 if (!st->codec->extradata)
483 return AVERROR(ENOMEM);
484 st->codec->extradata_size = 0x5a + atom.size;
485 memcpy(st->codec->extradata, "SVQ3", 4); // fake
486 get_buffer(pb, st->codec->extradata + 0x5a, atom.size);
487 dprintf(c->fc, "Reading SMI %"PRId64" %s\n", atom.size, st->codec->extradata + 0x5a);
491 static int mov_read_enda(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
493 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
494 int little_endian = get_be16(pb);
496 dprintf(c->fc, "enda %d\n", little_endian);
497 if (little_endian == 1) {
498 switch (st->codec->codec_id) {
499 case CODEC_ID_PCM_S24BE:
500 st->codec->codec_id = CODEC_ID_PCM_S24LE;
502 case CODEC_ID_PCM_S32BE:
503 st->codec->codec_id = CODEC_ID_PCM_S32LE;
505 case CODEC_ID_PCM_F32BE:
506 st->codec->codec_id = CODEC_ID_PCM_F32LE;
508 case CODEC_ID_PCM_F64BE:
509 st->codec->codec_id = CODEC_ID_PCM_F64LE;
518 /* FIXME modify qdm2/svq3/h264 decoders to take full atom as extradata */
519 static int mov_read_extradata(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
525 if (c->fc->nb_streams < 1) // will happen with jp2 files
527 st= c->fc->streams[c->fc->nb_streams-1];
528 size= (uint64_t)st->codec->extradata_size + atom.size + 8 + FF_INPUT_BUFFER_PADDING_SIZE;
529 if(size > INT_MAX || (uint64_t)atom.size > INT_MAX)
531 buf= av_realloc(st->codec->extradata, size);
534 st->codec->extradata= buf;
535 buf+= st->codec->extradata_size;
536 st->codec->extradata_size= size - FF_INPUT_BUFFER_PADDING_SIZE;
537 AV_WB32( buf , atom.size + 8);
538 AV_WL32( buf + 4, atom.type);
539 get_buffer(pb, buf + 8, atom.size);
543 static int mov_read_wave(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
545 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
547 if((uint64_t)atom.size > (1<<30))
550 if (st->codec->codec_id == CODEC_ID_QDM2) {
551 // pass all frma atom to codec, needed at least for QDM2
552 av_free(st->codec->extradata);
553 st->codec->extradata = av_mallocz(atom.size + FF_INPUT_BUFFER_PADDING_SIZE);
554 if (!st->codec->extradata)
555 return AVERROR(ENOMEM);
556 st->codec->extradata_size = atom.size;
557 get_buffer(pb, st->codec->extradata, atom.size);
558 } else if (atom.size > 8) { /* to read frma, esds atoms */
559 if (mov_read_default(c, pb, atom) < 0)
562 url_fskip(pb, atom.size);
567 * This function reads atom content and puts data in extradata without tag
568 * nor size unlike mov_read_extradata.
570 static int mov_read_glbl(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
572 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
574 if((uint64_t)atom.size > (1<<30))
577 av_free(st->codec->extradata);
578 st->codec->extradata = av_mallocz(atom.size + FF_INPUT_BUFFER_PADDING_SIZE);
579 if (!st->codec->extradata)
580 return AVERROR(ENOMEM);
581 st->codec->extradata_size = atom.size;
582 get_buffer(pb, st->codec->extradata, atom.size);
586 static int mov_read_stco(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
588 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
589 MOVStreamContext *sc = st->priv_data;
590 unsigned int i, entries;
592 get_byte(pb); /* version */
593 get_be24(pb); /* flags */
595 entries = get_be32(pb);
597 if(entries >= UINT_MAX/sizeof(int64_t))
600 sc->chunk_offsets = av_malloc(entries * sizeof(int64_t));
601 if (!sc->chunk_offsets)
602 return AVERROR(ENOMEM);
603 sc->chunk_count = entries;
605 if (atom.type == MKTAG('s','t','c','o'))
606 for(i=0; i<entries; i++)
607 sc->chunk_offsets[i] = get_be32(pb);
608 else if (atom.type == MKTAG('c','o','6','4'))
609 for(i=0; i<entries; i++)
610 sc->chunk_offsets[i] = get_be64(pb);
618 * Compute codec id for 'lpcm' tag.
619 * See CoreAudioTypes and AudioStreamBasicDescription at Apple.
621 static enum CodecID mov_get_lpcm_codec_id(int bps, int flags)
623 if (flags & 1) { // floating point
624 if (flags & 2) { // big endian
625 if (bps == 32) return CODEC_ID_PCM_F32BE;
626 else if (bps == 64) return CODEC_ID_PCM_F64BE;
628 if (bps == 32) return CODEC_ID_PCM_F32LE;
629 else if (bps == 64) return CODEC_ID_PCM_F64LE;
635 if (flags & 4) return CODEC_ID_PCM_S8;
636 else return CODEC_ID_PCM_U8;
637 else if (bps == 16) return CODEC_ID_PCM_S16BE;
638 else if (bps == 24) return CODEC_ID_PCM_S24BE;
639 else if (bps == 32) return CODEC_ID_PCM_S32BE;
642 if (flags & 4) return CODEC_ID_PCM_S8;
643 else return CODEC_ID_PCM_U8;
644 else if (bps == 16) return CODEC_ID_PCM_S16LE;
645 else if (bps == 24) return CODEC_ID_PCM_S24LE;
646 else if (bps == 32) return CODEC_ID_PCM_S32LE;
649 return CODEC_ID_NONE;
652 static int mov_read_stsd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
654 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
655 MOVStreamContext *sc = st->priv_data;
656 int j, entries, pseudo_stream_id;
658 get_byte(pb); /* version */
659 get_be24(pb); /* flags */
661 entries = get_be32(pb);
663 for(pseudo_stream_id=0; pseudo_stream_id<entries; pseudo_stream_id++) {
664 //Parsing Sample description table
667 MOVAtom a = { 0, 0, 0 };
668 int64_t start_pos = url_ftell(pb);
669 int size = get_be32(pb); /* size */
670 uint32_t format = get_le32(pb); /* data format */
672 get_be32(pb); /* reserved */
673 get_be16(pb); /* reserved */
674 dref_id = get_be16(pb);
676 if (st->codec->codec_tag &&
677 st->codec->codec_tag != format &&
678 (c->fc->video_codec_id ? codec_get_id(codec_movvideo_tags, format) != c->fc->video_codec_id
679 : st->codec->codec_tag != MKTAG('j','p','e','g'))
681 /* Multiple fourcc, we skip JPEG. This is not correct, we should
682 * export it as a separate AVStream but this needs a few changes
683 * in the MOV demuxer, patch welcome. */
684 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
685 url_fskip(pb, size - (url_ftell(pb) - start_pos));
688 sc->pseudo_stream_id = st->codec->codec_tag ? -1 : pseudo_stream_id;
689 sc->dref_id= dref_id;
691 st->codec->codec_tag = format;
692 id = codec_get_id(codec_movaudio_tags, format);
693 if (id<=0 && (format&0xFFFF) == 'm'+('s'<<8))
694 id = codec_get_id(codec_wav_tags, bswap_32(format)&0xFFFF);
696 if (st->codec->codec_type != CODEC_TYPE_VIDEO && id > 0) {
697 st->codec->codec_type = CODEC_TYPE_AUDIO;
698 } else if (st->codec->codec_type != CODEC_TYPE_AUDIO && /* do not overwrite codec type */
699 format && format != MKTAG('m','p','4','s')) { /* skip old asf mpeg4 tag */
700 id = codec_get_id(codec_movvideo_tags, format);
702 id = codec_get_id(codec_bmp_tags, format);
704 st->codec->codec_type = CODEC_TYPE_VIDEO;
705 else if(st->codec->codec_type == CODEC_TYPE_DATA){
706 id = codec_get_id(ff_codec_movsubtitle_tags, format);
708 st->codec->codec_type = CODEC_TYPE_SUBTITLE;
712 dprintf(c->fc, "size=%d 4CC= %c%c%c%c codec_type=%d\n", size,
713 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff,
714 (format >> 24) & 0xff, st->codec->codec_type);
716 if(st->codec->codec_type==CODEC_TYPE_VIDEO) {
717 uint8_t codec_name[32];
718 unsigned int color_depth;
721 st->codec->codec_id = id;
722 get_be16(pb); /* version */
723 get_be16(pb); /* revision level */
724 get_be32(pb); /* vendor */
725 get_be32(pb); /* temporal quality */
726 get_be32(pb); /* spatial quality */
728 st->codec->width = get_be16(pb); /* width */
729 st->codec->height = get_be16(pb); /* height */
731 get_be32(pb); /* horiz resolution */
732 get_be32(pb); /* vert resolution */
733 get_be32(pb); /* data size, always 0 */
734 get_be16(pb); /* frames per samples */
736 get_buffer(pb, codec_name, 32); /* codec name, pascal string */
737 if (codec_name[0] <= 31) {
738 memcpy(st->codec->codec_name, &codec_name[1],codec_name[0]);
739 st->codec->codec_name[codec_name[0]] = 0;
742 st->codec->bits_per_coded_sample = get_be16(pb); /* depth */
743 st->codec->color_table_id = get_be16(pb); /* colortable id */
744 dprintf(c->fc, "depth %d, ctab id %d\n",
745 st->codec->bits_per_coded_sample, st->codec->color_table_id);
746 /* figure out the palette situation */
747 color_depth = st->codec->bits_per_coded_sample & 0x1F;
748 color_greyscale = st->codec->bits_per_coded_sample & 0x20;
750 /* if the depth is 2, 4, or 8 bpp, file is palettized */
751 if ((color_depth == 2) || (color_depth == 4) ||
752 (color_depth == 8)) {
753 /* for palette traversal */
754 unsigned int color_start, color_count, color_end;
755 unsigned char r, g, b;
757 if (color_greyscale) {
758 int color_index, color_dec;
759 /* compute the greyscale palette */
760 st->codec->bits_per_coded_sample = color_depth;
761 color_count = 1 << color_depth;
763 color_dec = 256 / (color_count - 1);
764 for (j = 0; j < color_count; j++) {
765 r = g = b = color_index;
766 c->palette_control.palette[j] =
767 (r << 16) | (g << 8) | (b);
768 color_index -= color_dec;
772 } else if (st->codec->color_table_id) {
773 const uint8_t *color_table;
774 /* if flag bit 3 is set, use the default palette */
775 color_count = 1 << color_depth;
776 if (color_depth == 2)
777 color_table = ff_qt_default_palette_4;
778 else if (color_depth == 4)
779 color_table = ff_qt_default_palette_16;
781 color_table = ff_qt_default_palette_256;
783 for (j = 0; j < color_count; j++) {
784 r = color_table[j * 4 + 0];
785 g = color_table[j * 4 + 1];
786 b = color_table[j * 4 + 2];
787 c->palette_control.palette[j] =
788 (r << 16) | (g << 8) | (b);
791 /* load the palette from the file */
792 color_start = get_be32(pb);
793 color_count = get_be16(pb);
794 color_end = get_be16(pb);
795 if ((color_start <= 255) &&
796 (color_end <= 255)) {
797 for (j = color_start; j <= color_end; j++) {
798 /* each R, G, or B component is 16 bits;
799 * only use the top 8 bits; skip alpha bytes
809 c->palette_control.palette[j] =
810 (r << 16) | (g << 8) | (b);
814 st->codec->palctrl = &c->palette_control;
815 st->codec->palctrl->palette_changed = 1;
817 st->codec->palctrl = NULL;
818 } else if(st->codec->codec_type==CODEC_TYPE_AUDIO) {
819 int bits_per_sample, flags;
820 uint16_t version = get_be16(pb);
822 st->codec->codec_id = id;
823 get_be16(pb); /* revision level */
824 get_be32(pb); /* vendor */
826 st->codec->channels = get_be16(pb); /* channel count */
827 dprintf(c->fc, "audio channels %d\n", st->codec->channels);
828 st->codec->bits_per_coded_sample = get_be16(pb); /* sample size */
830 sc->audio_cid = get_be16(pb);
831 get_be16(pb); /* packet size = 0 */
833 st->codec->sample_rate = ((get_be32(pb) >> 16));
835 //Read QT version 1 fields. In version 0 these do not exist.
836 dprintf(c->fc, "version =%d, isom =%d\n",version,c->isom);
839 sc->samples_per_frame = get_be32(pb);
840 get_be32(pb); /* bytes per packet */
841 sc->bytes_per_frame = get_be32(pb);
842 get_be32(pb); /* bytes per sample */
843 } else if(version==2) {
844 get_be32(pb); /* sizeof struct only */
845 st->codec->sample_rate = av_int2dbl(get_be64(pb)); /* float 64 */
846 st->codec->channels = get_be32(pb);
847 get_be32(pb); /* always 0x7F000000 */
848 st->codec->bits_per_coded_sample = get_be32(pb); /* bits per channel if sound is uncompressed */
849 flags = get_be32(pb); /* lcpm format specific flag */
850 sc->bytes_per_frame = get_be32(pb); /* bytes per audio packet if constant */
851 sc->samples_per_frame = get_be32(pb); /* lpcm frames per audio packet if constant */
852 if (format == MKTAG('l','p','c','m'))
853 st->codec->codec_id = mov_get_lpcm_codec_id(st->codec->bits_per_coded_sample, flags);
857 switch (st->codec->codec_id) {
858 case CODEC_ID_PCM_S8:
859 case CODEC_ID_PCM_U8:
860 if (st->codec->bits_per_coded_sample == 16)
861 st->codec->codec_id = CODEC_ID_PCM_S16BE;
863 case CODEC_ID_PCM_S16LE:
864 case CODEC_ID_PCM_S16BE:
865 if (st->codec->bits_per_coded_sample == 8)
866 st->codec->codec_id = CODEC_ID_PCM_S8;
867 else if (st->codec->bits_per_coded_sample == 24)
868 st->codec->codec_id =
869 st->codec->codec_id == CODEC_ID_PCM_S16BE ?
870 CODEC_ID_PCM_S24BE : CODEC_ID_PCM_S24LE;
872 /* set values for old format before stsd version 1 appeared */
874 sc->samples_per_frame = 6;
875 sc->bytes_per_frame = 2*st->codec->channels;
878 sc->samples_per_frame = 6;
879 sc->bytes_per_frame = 1*st->codec->channels;
881 case CODEC_ID_ADPCM_IMA_QT:
882 sc->samples_per_frame = 64;
883 sc->bytes_per_frame = 34*st->codec->channels;
886 sc->samples_per_frame = 160;
887 sc->bytes_per_frame = 33;
893 bits_per_sample = av_get_bits_per_sample(st->codec->codec_id);
894 if (bits_per_sample) {
895 st->codec->bits_per_coded_sample = bits_per_sample;
896 sc->sample_size = (bits_per_sample >> 3) * st->codec->channels;
898 } else if(st->codec->codec_type==CODEC_TYPE_SUBTITLE){
899 // ttxt stsd contains display flags, justification, background
900 // color, fonts, and default styles, so fake an atom to read it
901 MOVAtom fake_atom = { .size = size - (url_ftell(pb) - start_pos) };
902 mov_read_glbl(c, pb, fake_atom);
903 st->codec->codec_id= id;
904 st->codec->width = sc->width;
905 st->codec->height = sc->height;
907 /* other codec type, just skip (rtp, mp4s, tmcd ...) */
908 url_fskip(pb, size - (url_ftell(pb) - start_pos));
910 /* this will read extra atoms at the end (wave, alac, damr, avcC, SMI ...) */
911 a.size = size - (url_ftell(pb) - start_pos);
913 if (mov_read_default(c, pb, a) < 0)
915 } else if (a.size > 0)
916 url_fskip(pb, a.size);
919 if(st->codec->codec_type==CODEC_TYPE_AUDIO && st->codec->sample_rate==0 && sc->time_scale>1)
920 st->codec->sample_rate= sc->time_scale;
922 /* special codec parameters handling */
923 switch (st->codec->codec_id) {
924 #if CONFIG_DV_DEMUXER
925 case CODEC_ID_DVAUDIO:
926 c->dv_fctx = avformat_alloc_context();
927 c->dv_demux = dv_init_demux(c->dv_fctx);
929 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
932 sc->dv_audio_container = 1;
933 st->codec->codec_id = CODEC_ID_PCM_S16LE;
936 /* no ifdef since parameters are always those */
938 st->codec->frame_size= 160;
939 st->codec->channels= 1; /* really needed */
941 case CODEC_ID_AMR_NB:
942 case CODEC_ID_AMR_WB:
943 st->codec->frame_size= sc->samples_per_frame;
944 st->codec->channels= 1; /* really needed */
945 /* force sample rate for amr, stsd in 3gp does not store sample rate */
946 if (st->codec->codec_id == CODEC_ID_AMR_NB)
947 st->codec->sample_rate = 8000;
948 else if (st->codec->codec_id == CODEC_ID_AMR_WB)
949 st->codec->sample_rate = 16000;
953 st->codec->codec_type = CODEC_TYPE_AUDIO; /* force type after stsd for m1a hdlr */
954 st->need_parsing = AVSTREAM_PARSE_FULL;
957 case CODEC_ID_ADPCM_MS:
958 case CODEC_ID_ADPCM_IMA_WAV:
959 st->codec->block_align = sc->bytes_per_frame;
962 if (st->codec->extradata_size == 36) {
963 st->codec->frame_size = AV_RB32(st->codec->extradata+12);
964 st->codec->channels = AV_RB8 (st->codec->extradata+21);
974 static int mov_read_stsc(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
976 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
977 MOVStreamContext *sc = st->priv_data;
978 unsigned int i, entries;
980 get_byte(pb); /* version */
981 get_be24(pb); /* flags */
983 entries = get_be32(pb);
985 dprintf(c->fc, "track[%i].stsc.entries = %i\n", c->fc->nb_streams-1, entries);
987 if(entries >= UINT_MAX / sizeof(*sc->stsc_data))
989 sc->stsc_data = av_malloc(entries * sizeof(*sc->stsc_data));
991 return AVERROR(ENOMEM);
992 sc->stsc_count = entries;
994 for(i=0; i<entries; i++) {
995 sc->stsc_data[i].first = get_be32(pb);
996 sc->stsc_data[i].count = get_be32(pb);
997 sc->stsc_data[i].id = get_be32(pb);
1002 static int mov_read_stss(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1004 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1005 MOVStreamContext *sc = st->priv_data;
1006 unsigned int i, entries;
1008 get_byte(pb); /* version */
1009 get_be24(pb); /* flags */
1011 entries = get_be32(pb);
1013 dprintf(c->fc, "keyframe_count = %d\n", entries);
1015 if(entries >= UINT_MAX / sizeof(int))
1017 sc->keyframes = av_malloc(entries * sizeof(int));
1019 return AVERROR(ENOMEM);
1020 sc->keyframe_count = entries;
1022 for(i=0; i<entries; i++) {
1023 sc->keyframes[i] = get_be32(pb);
1024 //dprintf(c->fc, "keyframes[]=%d\n", sc->keyframes[i]);
1029 static int mov_read_stsz(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1031 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1032 MOVStreamContext *sc = st->priv_data;
1033 unsigned int i, entries, sample_size;
1035 get_byte(pb); /* version */
1036 get_be24(pb); /* flags */
1038 sample_size = get_be32(pb);
1039 if (!sc->sample_size) /* do not overwrite value computed in stsd */
1040 sc->sample_size = sample_size;
1041 entries = get_be32(pb);
1043 dprintf(c->fc, "sample_size = %d sample_count = %d\n", sc->sample_size, entries);
1045 sc->sample_count = entries;
1049 if(entries >= UINT_MAX / sizeof(int))
1051 sc->sample_sizes = av_malloc(entries * sizeof(int));
1052 if (!sc->sample_sizes)
1053 return AVERROR(ENOMEM);
1055 for(i=0; i<entries; i++)
1056 sc->sample_sizes[i] = get_be32(pb);
1060 static int mov_read_stts(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1062 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1063 MOVStreamContext *sc = st->priv_data;
1064 unsigned int i, entries;
1066 int64_t total_sample_count=0;
1068 get_byte(pb); /* version */
1069 get_be24(pb); /* flags */
1070 entries = get_be32(pb);
1072 dprintf(c->fc, "track[%i].stts.entries = %i\n", c->fc->nb_streams-1, entries);
1074 if(entries >= UINT_MAX / sizeof(*sc->stts_data))
1076 sc->stts_data = av_malloc(entries * sizeof(*sc->stts_data));
1078 return AVERROR(ENOMEM);
1079 sc->stts_count = entries;
1081 for(i=0; i<entries; i++) {
1082 int sample_duration;
1085 sample_count=get_be32(pb);
1086 sample_duration = get_be32(pb);
1087 sc->stts_data[i].count= sample_count;
1088 sc->stts_data[i].duration= sample_duration;
1090 sc->time_rate= av_gcd(sc->time_rate, sample_duration);
1092 dprintf(c->fc, "sample_count=%d, sample_duration=%d\n",sample_count,sample_duration);
1094 duration+=(int64_t)sample_duration*sample_count;
1095 total_sample_count+=sample_count;
1098 st->nb_frames= total_sample_count;
1100 st->duration= duration;
1104 static int mov_read_ctts(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1106 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1107 MOVStreamContext *sc = st->priv_data;
1108 unsigned int i, entries;
1110 get_byte(pb); /* version */
1111 get_be24(pb); /* flags */
1112 entries = get_be32(pb);
1114 dprintf(c->fc, "track[%i].ctts.entries = %i\n", c->fc->nb_streams-1, entries);
1116 if(entries >= UINT_MAX / sizeof(*sc->ctts_data))
1118 sc->ctts_data = av_malloc(entries * sizeof(*sc->ctts_data));
1120 return AVERROR(ENOMEM);
1121 sc->ctts_count = entries;
1123 for(i=0; i<entries; i++) {
1124 int count =get_be32(pb);
1125 int duration =get_be32(pb);
1129 st->codec->has_b_frames = 1;
1131 sc->ctts_data[i].count = count;
1132 sc->ctts_data[i].duration= duration;
1134 sc->time_rate= av_gcd(sc->time_rate, FFABS(duration));
1139 static void mov_build_index(MOVContext *mov, AVStream *st)
1141 MOVStreamContext *sc = st->priv_data;
1142 int64_t current_offset;
1143 int64_t current_dts = 0;
1144 unsigned int stts_index = 0;
1145 unsigned int stsc_index = 0;
1146 unsigned int stss_index = 0;
1149 /* adjust first dts according to edit list */
1150 if (sc->time_offset) {
1151 assert(sc->time_offset % sc->time_rate == 0);
1152 current_dts = - (sc->time_offset / sc->time_rate);
1155 /* only use old uncompressed audio chunk demuxing when stts specifies it */
1156 if (!(st->codec->codec_type == CODEC_TYPE_AUDIO &&
1157 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
1158 unsigned int current_sample = 0;
1159 unsigned int stts_sample = 0;
1160 unsigned int keyframe, sample_size;
1161 unsigned int distance = 0;
1162 int key_off = sc->keyframes && sc->keyframes[0] == 1;
1164 st->nb_frames = sc->sample_count;
1165 for (i = 0; i < sc->chunk_count; i++) {
1166 current_offset = sc->chunk_offsets[i];
1167 if (stsc_index + 1 < sc->stsc_count &&
1168 i + 1 == sc->stsc_data[stsc_index + 1].first)
1170 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
1171 if (current_sample >= sc->sample_count) {
1172 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
1175 keyframe = !sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index];
1178 if (stss_index + 1 < sc->keyframe_count)
1181 sample_size = sc->sample_size > 0 ? sc->sample_size : sc->sample_sizes[current_sample];
1182 if(sc->pseudo_stream_id == -1 ||
1183 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
1184 av_add_index_entry(st, current_offset, current_dts, sample_size, distance,
1185 keyframe ? AVINDEX_KEYFRAME : 0);
1186 dprintf(mov->fc, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
1187 "size %d, distance %d, keyframe %d\n", st->index, current_sample,
1188 current_offset, current_dts, sample_size, distance, keyframe);
1190 current_offset += sample_size;
1191 assert(sc->stts_data[stts_index].duration % sc->time_rate == 0);
1192 current_dts += sc->stts_data[stts_index].duration / sc->time_rate;
1196 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
1202 } else { /* read whole chunk */
1203 unsigned int chunk_samples, chunk_size, chunk_duration;
1204 unsigned int frames = 1;
1205 for (i = 0; i < sc->chunk_count; i++) {
1206 current_offset = sc->chunk_offsets[i];
1207 if (stsc_index + 1 < sc->stsc_count &&
1208 i + 1 == sc->stsc_data[stsc_index + 1].first)
1210 chunk_samples = sc->stsc_data[stsc_index].count;
1211 /* get chunk size, beware of alaw/ulaw/mace */
1212 if (sc->samples_per_frame > 0 &&
1213 (chunk_samples * sc->bytes_per_frame % sc->samples_per_frame == 0)) {
1214 if (sc->samples_per_frame < 160)
1215 chunk_size = chunk_samples * sc->bytes_per_frame / sc->samples_per_frame;
1217 chunk_size = sc->bytes_per_frame;
1218 frames = chunk_samples / sc->samples_per_frame;
1219 chunk_samples = sc->samples_per_frame;
1222 chunk_size = chunk_samples * sc->sample_size;
1223 for (j = 0; j < frames; j++) {
1224 av_add_index_entry(st, current_offset, current_dts, chunk_size, 0, AVINDEX_KEYFRAME);
1225 /* get chunk duration */
1227 while (chunk_samples > 0) {
1228 if (chunk_samples < sc->stts_data[stts_index].count) {
1229 chunk_duration += sc->stts_data[stts_index].duration * chunk_samples;
1230 sc->stts_data[stts_index].count -= chunk_samples;
1233 chunk_duration += sc->stts_data[stts_index].duration * chunk_samples;
1234 chunk_samples -= sc->stts_data[stts_index].count;
1235 if (stts_index + 1 < sc->stts_count)
1239 current_offset += sc->bytes_per_frame;
1240 dprintf(mov->fc, "AVIndex stream %d, chunk %d, offset %"PRIx64", dts %"PRId64", "
1241 "size %d, duration %d\n", st->index, i, current_offset, current_dts,
1242 chunk_size, chunk_duration);
1243 assert(chunk_duration % sc->time_rate == 0);
1244 current_dts += chunk_duration / sc->time_rate;
1250 static int mov_read_trak(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1253 MOVStreamContext *sc;
1256 st = av_new_stream(c->fc, c->fc->nb_streams);
1257 if (!st) return AVERROR(ENOMEM);
1258 sc = av_mallocz(sizeof(MOVStreamContext));
1259 if (!sc) return AVERROR(ENOMEM);
1262 st->codec->codec_type = CODEC_TYPE_DATA;
1263 sc->ffindex = st->index;
1265 if ((ret = mov_read_default(c, pb, atom)) < 0)
1269 if(sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
1270 (!sc->sample_size && !sc->sample_count))){
1271 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
1278 sc->time_scale= c->time_scale;
1279 av_set_pts_info(st, 64, sc->time_rate, sc->time_scale);
1281 if (st->codec->codec_type == CODEC_TYPE_AUDIO &&
1282 !st->codec->frame_size && sc->stts_count == 1) {
1283 st->codec->frame_size = av_rescale(sc->stts_data[0].duration,
1284 st->codec->sample_rate, sc->time_scale);
1285 dprintf(c->fc, "frame size %d\n", st->codec->frame_size);
1288 if(st->duration != AV_NOPTS_VALUE){
1289 assert(st->duration % sc->time_rate == 0);
1290 st->duration /= sc->time_rate;
1293 mov_build_index(c, st);
1295 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
1296 if (url_fopen(&sc->pb, sc->drefs[sc->dref_id-1].path, URL_RDONLY) < 0)
1297 av_log(c->fc, AV_LOG_ERROR, "stream %d, error opening file %s: %s\n",
1298 st->index, sc->drefs[sc->dref_id-1].path, strerror(errno));
1302 switch (st->codec->codec_id) {
1303 #if CONFIG_H261_DECODER
1306 #if CONFIG_H263_DECODER
1309 #if CONFIG_MPEG4_DECODER
1310 case CODEC_ID_MPEG4:
1312 st->codec->width= 0; /* let decoder init width/height */
1313 st->codec->height= 0;
1317 /* Do not need those anymore. */
1318 av_freep(&sc->chunk_offsets);
1319 av_freep(&sc->stsc_data);
1320 av_freep(&sc->sample_sizes);
1321 av_freep(&sc->keyframes);
1322 av_freep(&sc->stts_data);
1327 static int mov_read_ilst(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1330 c->itunes_metadata = 1;
1331 ret = mov_read_default(c, pb, atom);
1332 c->itunes_metadata = 0;
1336 static int mov_read_meta(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1338 url_fskip(pb, 4); // version + flags
1340 return mov_read_default(c, pb, atom);
1343 static int mov_read_trkn(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1346 get_be32(pb); // type
1347 get_be32(pb); // unknown
1348 snprintf(track, sizeof(track), "%d", get_be32(pb));
1349 av_metadata_set(&c->fc->metadata, "track", track);
1350 dprintf(c->fc, "%.4s %s\n", (char*)&atom.type, track);
1354 static int mov_read_udta_string(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1356 char str[1024], key2[16], language[4] = {0};
1357 const char *key = NULL;
1360 if (c->itunes_metadata) {
1361 int data_size = get_be32(pb);
1362 int tag = get_le32(pb);
1363 if (tag == MKTAG('d','a','t','a')) {
1364 get_be32(pb); // type
1365 get_be32(pb); // unknown
1366 str_size = data_size - 16;
1370 str_size = get_be16(pb); // string length
1371 ff_mov_lang_to_iso639(get_be16(pb), language);
1374 switch (atom.type) {
1375 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
1376 case MKTAG(0xa9,'a','u','t'):
1377 case MKTAG(0xa9,'A','R','T'):
1378 case MKTAG(0xa9,'w','r','t'): key = "author"; break;
1379 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
1380 case MKTAG(0xa9,'c','m','t'):
1381 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
1382 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
1383 case MKTAG(0xa9,'d','a','y'): key = "year"; break;
1384 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
1385 case MKTAG(0xa9,'t','o','o'):
1386 case MKTAG(0xa9,'e','n','c'): key = "muxer"; break;
1393 str_size = FFMIN3(sizeof(str)-1, str_size, atom.size);
1394 get_buffer(pb, str, str_size);
1396 av_metadata_set(&c->fc->metadata, key, str);
1397 if (*language && strcmp(language, "und")) {
1398 snprintf(key2, sizeof(key2), "%s-%s", key, language);
1399 av_metadata_set(&c->fc->metadata, key2, str);
1401 dprintf(c->fc, "%.4s %s %d %lld\n", (char*)&atom.type, str, str_size, atom.size);
1405 static int mov_read_tkhd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1410 int64_t disp_transform[2];
1411 int display_matrix[3][2];
1412 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1413 MOVStreamContext *sc = st->priv_data;
1414 int version = get_byte(pb);
1416 get_be24(pb); /* flags */
1418 MOV_TRACK_ENABLED 0x0001
1419 MOV_TRACK_IN_MOVIE 0x0002
1420 MOV_TRACK_IN_PREVIEW 0x0004
1421 MOV_TRACK_IN_POSTER 0x0008
1428 get_be32(pb); /* creation time */
1429 get_be32(pb); /* modification time */
1431 st->id = (int)get_be32(pb); /* track id (NOT 0 !)*/
1432 get_be32(pb); /* reserved */
1434 /* highlevel (considering edits) duration in movie timebase */
1435 (version == 1) ? get_be64(pb) : get_be32(pb);
1436 get_be32(pb); /* reserved */
1437 get_be32(pb); /* reserved */
1439 get_be16(pb); /* layer */
1440 get_be16(pb); /* alternate group */
1441 get_be16(pb); /* volume */
1442 get_be16(pb); /* reserved */
1444 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
1445 // they're kept in fixed point format through all calculations
1446 // ignore u,v,z b/c we don't need the scale factor to calc aspect ratio
1447 for (i = 0; i < 3; i++) {
1448 display_matrix[i][0] = get_be32(pb); // 16.16 fixed point
1449 display_matrix[i][1] = get_be32(pb); // 16.16 fixed point
1450 get_be32(pb); // 2.30 fixed point (not used)
1453 width = get_be32(pb); // 16.16 fixed point track width
1454 height = get_be32(pb); // 16.16 fixed point track height
1455 sc->width = width >> 16;
1456 sc->height = height >> 16;
1458 //transform the display width/height according to the matrix
1459 // skip this if the display matrix is the default identity matrix
1460 // to keep the same scale, use [width height 1<<16]
1461 if (width && height &&
1462 (display_matrix[0][0] != 65536 || display_matrix[0][1] ||
1463 display_matrix[1][0] || display_matrix[1][1] != 65536 ||
1464 display_matrix[2][0] || display_matrix[2][1])) {
1465 for (i = 0; i < 2; i++)
1467 (int64_t) width * display_matrix[0][i] +
1468 (int64_t) height * display_matrix[1][i] +
1469 ((int64_t) display_matrix[2][i] << 16);
1471 //sample aspect ratio is new width/height divided by old width/height
1472 st->sample_aspect_ratio = av_d2q(
1473 ((double) disp_transform[0] * height) /
1474 ((double) disp_transform[1] * width), INT_MAX);
1479 static int mov_read_tfhd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1481 MOVFragment *frag = &c->fragment;
1482 MOVTrackExt *trex = NULL;
1483 int flags, track_id, i;
1485 get_byte(pb); /* version */
1486 flags = get_be24(pb);
1488 track_id = get_be32(pb);
1491 frag->track_id = track_id;
1492 for (i = 0; i < c->trex_count; i++)
1493 if (c->trex_data[i].track_id == frag->track_id) {
1494 trex = &c->trex_data[i];
1498 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
1502 if (flags & 0x01) frag->base_data_offset = get_be64(pb);
1503 else frag->base_data_offset = frag->moof_offset;
1504 if (flags & 0x02) frag->stsd_id = get_be32(pb);
1505 else frag->stsd_id = trex->stsd_id;
1507 frag->duration = flags & 0x08 ? get_be32(pb) : trex->duration;
1508 frag->size = flags & 0x10 ? get_be32(pb) : trex->size;
1509 frag->flags = flags & 0x20 ? get_be32(pb) : trex->flags;
1510 dprintf(c->fc, "frag flags 0x%x\n", frag->flags);
1514 static int mov_read_trex(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1518 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
1520 trex = av_realloc(c->trex_data, (c->trex_count+1)*sizeof(*c->trex_data));
1522 return AVERROR(ENOMEM);
1523 c->trex_data = trex;
1524 trex = &c->trex_data[c->trex_count++];
1525 get_byte(pb); /* version */
1526 get_be24(pb); /* flags */
1527 trex->track_id = get_be32(pb);
1528 trex->stsd_id = get_be32(pb);
1529 trex->duration = get_be32(pb);
1530 trex->size = get_be32(pb);
1531 trex->flags = get_be32(pb);
1535 static int mov_read_trun(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1537 MOVFragment *frag = &c->fragment;
1538 AVStream *st = NULL;
1539 MOVStreamContext *sc;
1542 int data_offset = 0;
1543 unsigned entries, first_sample_flags = frag->flags;
1544 int flags, distance, i;
1546 for (i = 0; i < c->fc->nb_streams; i++) {
1547 if (c->fc->streams[i]->id == frag->track_id) {
1548 st = c->fc->streams[i];
1553 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
1557 if (sc->pseudo_stream_id+1 != frag->stsd_id)
1559 get_byte(pb); /* version */
1560 flags = get_be24(pb);
1561 entries = get_be32(pb);
1562 dprintf(c->fc, "flags 0x%x entries %d\n", flags, entries);
1563 if (flags & 0x001) data_offset = get_be32(pb);
1564 if (flags & 0x004) first_sample_flags = get_be32(pb);
1565 if (flags & 0x800) {
1567 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
1569 ctts_data = av_realloc(sc->ctts_data,
1570 (entries+sc->ctts_count)*sizeof(*sc->ctts_data));
1572 return AVERROR(ENOMEM);
1573 sc->ctts_data = ctts_data;
1576 offset = frag->base_data_offset + data_offset;
1578 dprintf(c->fc, "first sample flags 0x%x\n", first_sample_flags);
1579 for (i = 0; i < entries; i++) {
1580 unsigned sample_size = frag->size;
1581 int sample_flags = i ? frag->flags : first_sample_flags;
1582 unsigned sample_duration = frag->duration;
1585 if (flags & 0x100) sample_duration = get_be32(pb);
1586 if (flags & 0x200) sample_size = get_be32(pb);
1587 if (flags & 0x400) sample_flags = get_be32(pb);
1588 if (flags & 0x800) {
1589 sc->ctts_data[sc->ctts_count].count = 1;
1590 sc->ctts_data[sc->ctts_count].duration = get_be32(pb);
1593 if ((keyframe = st->codec->codec_type == CODEC_TYPE_AUDIO ||
1594 (flags & 0x004 && !i && !sample_flags) || sample_flags & 0x2000000))
1596 av_add_index_entry(st, offset, dts, sample_size, distance,
1597 keyframe ? AVINDEX_KEYFRAME : 0);
1598 dprintf(c->fc, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
1599 "size %d, distance %d, keyframe %d\n", st->index, sc->sample_count+i,
1600 offset, dts, sample_size, distance, keyframe);
1602 assert(sample_duration % sc->time_rate == 0);
1603 dts += sample_duration / sc->time_rate;
1604 offset += sample_size;
1606 frag->moof_offset = offset;
1611 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
1612 /* like the files created with Adobe Premiere 5.0, for samples see */
1613 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
1614 static int mov_read_wide(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1619 return 0; /* continue */
1620 if (get_be32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
1621 url_fskip(pb, atom.size - 4);
1624 atom.type = get_le32(pb);
1627 if (atom.type != MKTAG('m','d','a','t')) {
1628 url_fskip(pb, atom.size);
1631 err = mov_read_mdat(c, pb, atom);
1635 static int mov_read_cmov(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1640 uint8_t *moov_data; /* uncompressed data */
1641 long cmov_len, moov_len;
1644 get_be32(pb); /* dcom atom */
1645 if (get_le32(pb) != MKTAG('d','c','o','m'))
1647 if (get_le32(pb) != MKTAG('z','l','i','b')) {
1648 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !");
1651 get_be32(pb); /* cmvd atom */
1652 if (get_le32(pb) != MKTAG('c','m','v','d'))
1654 moov_len = get_be32(pb); /* uncompressed size */
1655 cmov_len = atom.size - 6 * 4;
1657 cmov_data = av_malloc(cmov_len);
1659 return AVERROR(ENOMEM);
1660 moov_data = av_malloc(moov_len);
1663 return AVERROR(ENOMEM);
1665 get_buffer(pb, cmov_data, cmov_len);
1666 if(uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
1667 goto free_and_return;
1668 if(init_put_byte(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
1669 goto free_and_return;
1670 atom.type = MKTAG('m','o','o','v');
1672 atom.size = moov_len;
1674 // { int fd = open("/tmp/uncompheader.mov", O_WRONLY | O_CREAT); write(fd, moov_data, moov_len); close(fd); }
1676 ret = mov_read_default(c, &ctx, atom);
1682 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
1687 /* edit list atom */
1688 static int mov_read_elst(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1690 MOVStreamContext *sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
1693 get_byte(pb); /* version */
1694 get_be24(pb); /* flags */
1695 edit_count = get_be32(pb); /* entries */
1697 for(i=0; i<edit_count; i++){
1699 get_be32(pb); /* Track duration */
1700 time = get_be32(pb); /* Media time */
1701 get_be32(pb); /* Media rate */
1702 if (i == 0 && time != -1) {
1703 sc->time_offset = time;
1704 sc->time_rate = av_gcd(sc->time_rate, time);
1709 av_log(c->fc, AV_LOG_WARNING, "multiple edit list entries, "
1710 "a/v desync might occur, patch welcome\n");
1712 dprintf(c->fc, "track[%i].edit_count = %i\n", c->fc->nb_streams-1, edit_count);
1716 static const MOVParseTableEntry mov_default_parse_table[] = {
1717 { MKTAG('a','v','s','s'), mov_read_extradata },
1718 { MKTAG('c','o','6','4'), mov_read_stco },
1719 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
1720 { MKTAG('d','i','n','f'), mov_read_default },
1721 { MKTAG('d','r','e','f'), mov_read_dref },
1722 { MKTAG('e','d','t','s'), mov_read_default },
1723 { MKTAG('e','l','s','t'), mov_read_elst },
1724 { MKTAG('e','n','d','a'), mov_read_enda },
1725 { MKTAG('f','i','e','l'), mov_read_extradata },
1726 { MKTAG('f','t','y','p'), mov_read_ftyp },
1727 { MKTAG('g','l','b','l'), mov_read_glbl },
1728 { MKTAG('h','d','l','r'), mov_read_hdlr },
1729 { MKTAG('i','l','s','t'), mov_read_ilst },
1730 { MKTAG('j','p','2','h'), mov_read_extradata },
1731 { MKTAG('m','d','a','t'), mov_read_mdat },
1732 { MKTAG('m','d','h','d'), mov_read_mdhd },
1733 { MKTAG('m','d','i','a'), mov_read_default },
1734 { MKTAG('m','e','t','a'), mov_read_meta },
1735 { MKTAG('m','i','n','f'), mov_read_default },
1736 { MKTAG('m','o','o','f'), mov_read_moof },
1737 { MKTAG('m','o','o','v'), mov_read_moov },
1738 { MKTAG('m','v','e','x'), mov_read_default },
1739 { MKTAG('m','v','h','d'), mov_read_mvhd },
1740 { MKTAG('S','M','I',' '), mov_read_smi }, /* Sorenson extension ??? */
1741 { MKTAG('a','l','a','c'), mov_read_extradata }, /* alac specific atom */
1742 { MKTAG('a','v','c','C'), mov_read_glbl },
1743 { MKTAG('p','a','s','p'), mov_read_pasp },
1744 { MKTAG('s','t','b','l'), mov_read_default },
1745 { MKTAG('s','t','c','o'), mov_read_stco },
1746 { MKTAG('s','t','s','c'), mov_read_stsc },
1747 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
1748 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
1749 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
1750 { MKTAG('s','t','t','s'), mov_read_stts },
1751 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
1752 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
1753 { MKTAG('t','r','a','k'), mov_read_trak },
1754 { MKTAG('t','r','a','f'), mov_read_default },
1755 { MKTAG('t','r','e','x'), mov_read_trex },
1756 { MKTAG('t','r','k','n'), mov_read_trkn },
1757 { MKTAG('t','r','u','n'), mov_read_trun },
1758 { MKTAG('u','d','t','a'), mov_read_default },
1759 { MKTAG('w','a','v','e'), mov_read_wave },
1760 { MKTAG('e','s','d','s'), mov_read_esds },
1761 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
1762 { MKTAG('c','m','o','v'), mov_read_cmov },
1763 { MKTAG(0xa9,'n','a','m'), mov_read_udta_string },
1764 { MKTAG(0xa9,'w','r','t'), mov_read_udta_string },
1765 { MKTAG(0xa9,'c','p','y'), mov_read_udta_string },
1766 { MKTAG(0xa9,'i','n','f'), mov_read_udta_string },
1767 { MKTAG(0xa9,'i','n','f'), mov_read_udta_string },
1768 { MKTAG(0xa9,'A','R','T'), mov_read_udta_string },
1769 { MKTAG(0xa9,'a','l','b'), mov_read_udta_string },
1770 { MKTAG(0xa9,'c','m','t'), mov_read_udta_string },
1771 { MKTAG(0xa9,'a','u','t'), mov_read_udta_string },
1772 { MKTAG(0xa9,'d','a','y'), mov_read_udta_string },
1773 { MKTAG(0xa9,'g','e','n'), mov_read_udta_string },
1774 { MKTAG(0xa9,'e','n','c'), mov_read_udta_string },
1775 { MKTAG(0xa9,'t','o','o'), mov_read_udta_string },
1779 static int mov_probe(AVProbeData *p)
1781 unsigned int offset;
1785 /* check file header */
1788 /* ignore invalid offset */
1789 if ((offset + 8) > (unsigned int)p->buf_size)
1791 tag = AV_RL32(p->buf + offset + 4);
1793 /* check for obvious tags */
1794 case MKTAG('j','P',' ',' '): /* jpeg 2000 signature */
1795 case MKTAG('m','o','o','v'):
1796 case MKTAG('m','d','a','t'):
1797 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
1798 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
1799 case MKTAG('f','t','y','p'):
1800 return AVPROBE_SCORE_MAX;
1801 /* those are more common words, so rate then a bit less */
1802 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
1803 case MKTAG('w','i','d','e'):
1804 case MKTAG('f','r','e','e'):
1805 case MKTAG('j','u','n','k'):
1806 case MKTAG('p','i','c','t'):
1807 return AVPROBE_SCORE_MAX - 5;
1808 case MKTAG(0x82,0x82,0x7f,0x7d):
1809 case MKTAG('s','k','i','p'):
1810 case MKTAG('u','u','i','d'):
1811 case MKTAG('p','r','f','l'):
1812 offset = AV_RB32(p->buf+offset) + offset;
1813 /* if we only find those cause probedata is too small at least rate them */
1814 score = AVPROBE_SCORE_MAX - 50;
1817 /* unrecognized tag */
1824 static int mov_read_header(AVFormatContext *s, AVFormatParameters *ap)
1826 MOVContext *mov = s->priv_data;
1827 ByteIOContext *pb = s->pb;
1829 MOVAtom atom = { 0, 0, 0 };
1832 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
1833 if(!url_is_streamed(pb))
1834 atom.size = url_fsize(pb);
1836 atom.size = INT64_MAX;
1838 /* check MOV header */
1839 if ((err = mov_read_default(mov, pb, atom)) < 0) {
1840 av_log(s, AV_LOG_ERROR, "error reading header: %d\n", err);
1843 if (!mov->found_moov) {
1844 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
1847 dprintf(mov->fc, "on_parse_exit_offset=%lld\n", url_ftell(pb));
1852 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
1854 MOVContext *mov = s->priv_data;
1855 MOVStreamContext *sc = 0;
1856 AVIndexEntry *sample = 0;
1857 int64_t best_dts = INT64_MAX;
1860 for (i = 0; i < s->nb_streams; i++) {
1861 AVStream *st = s->streams[i];
1862 MOVStreamContext *msc = st->priv_data;
1863 if (st->discard != AVDISCARD_ALL && msc->pb && msc->current_sample < st->nb_index_entries) {
1864 AVIndexEntry *current_sample = &st->index_entries[msc->current_sample];
1865 int64_t dts = av_rescale(current_sample->timestamp * (int64_t)msc->time_rate,
1866 AV_TIME_BASE, msc->time_scale);
1867 dprintf(s, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
1868 if (!sample || (url_is_streamed(s->pb) && current_sample->pos < sample->pos) ||
1869 (!url_is_streamed(s->pb) &&
1870 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
1871 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
1872 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
1873 sample = current_sample;
1880 mov->found_mdat = 0;
1881 if (!url_is_streamed(s->pb) ||
1882 mov_read_default(mov, s->pb, (MOVAtom){ 0, 0, INT64_MAX }) < 0 ||
1885 dprintf(s, "read fragments, offset 0x%llx\n", url_ftell(s->pb));
1888 /* must be done just before reading, to avoid infinite loop on sample */
1889 sc->current_sample++;
1890 if (url_fseek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
1891 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
1892 sc->ffindex, sample->pos);
1895 ret = av_get_packet(sc->pb, pkt, sample->size);
1898 #if CONFIG_DV_DEMUXER
1899 if (mov->dv_demux && sc->dv_audio_container) {
1900 dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size);
1903 if (dv_get_packet(mov->dv_demux, pkt) < 0)
1907 pkt->stream_index = sc->ffindex;
1908 pkt->dts = sample->timestamp;
1909 if (sc->ctts_data) {
1910 assert(sc->ctts_data[sc->ctts_index].duration % sc->time_rate == 0);
1911 pkt->pts = pkt->dts + sc->ctts_data[sc->ctts_index].duration / sc->time_rate;
1912 /* update ctts context */
1914 if (sc->ctts_index < sc->ctts_count &&
1915 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
1917 sc->ctts_sample = 0;
1920 pkt->dts = AV_NOPTS_VALUE;
1922 AVStream *st = s->streams[sc->ffindex];
1923 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
1924 st->index_entries[sc->current_sample].timestamp : st->duration;
1925 pkt->duration = next_dts - pkt->dts;
1926 pkt->pts = pkt->dts;
1928 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? PKT_FLAG_KEY : 0;
1929 pkt->pos = sample->pos;
1930 dprintf(s, "stream %d, pts %"PRId64", dts %"PRId64", pos 0x%"PRIx64", duration %d\n",
1931 pkt->stream_index, pkt->pts, pkt->dts, pkt->pos, pkt->duration);
1935 static int mov_seek_stream(AVStream *st, int64_t timestamp, int flags)
1937 MOVStreamContext *sc = st->priv_data;
1938 int sample, time_sample;
1941 sample = av_index_search_timestamp(st, timestamp, flags);
1942 dprintf(st->codec, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
1943 if (sample < 0) /* not sure what to do */
1945 sc->current_sample = sample;
1946 dprintf(st->codec, "stream %d, found sample %d\n", st->index, sc->current_sample);
1947 /* adjust ctts index */
1948 if (sc->ctts_data) {
1950 for (i = 0; i < sc->ctts_count; i++) {
1951 int next = time_sample + sc->ctts_data[i].count;
1952 if (next > sc->current_sample) {
1954 sc->ctts_sample = sc->current_sample - time_sample;
1963 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
1966 int64_t seek_timestamp, timestamp;
1970 if (stream_index >= s->nb_streams)
1972 if (sample_time < 0)
1975 st = s->streams[stream_index];
1976 sample = mov_seek_stream(st, sample_time, flags);
1980 /* adjust seek timestamp to found sample timestamp */
1981 seek_timestamp = st->index_entries[sample].timestamp;
1983 for (i = 0; i < s->nb_streams; i++) {
1985 if (stream_index == i || st->discard == AVDISCARD_ALL)
1988 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
1989 mov_seek_stream(st, timestamp, flags);
1994 static int mov_read_close(AVFormatContext *s)
1997 MOVContext *mov = s->priv_data;
1998 for(i=0; i<s->nb_streams; i++) {
1999 MOVStreamContext *sc = s->streams[i]->priv_data;
2000 av_freep(&sc->ctts_data);
2001 for (j=0; j<sc->drefs_count; j++)
2002 av_freep(&sc->drefs[j].path);
2003 av_freep(&sc->drefs);
2004 if (sc->pb && sc->pb != s->pb)
2008 for(i=0; i<mov->dv_fctx->nb_streams; i++){
2009 av_freep(&mov->dv_fctx->streams[i]->codec);
2010 av_freep(&mov->dv_fctx->streams[i]);
2012 av_freep(&mov->dv_fctx);
2013 av_freep(&mov->dv_demux);
2015 av_freep(&mov->trex_data);
2019 AVInputFormat mov_demuxer = {
2020 "mov,mp4,m4a,3gp,3g2,mj2",
2021 NULL_IF_CONFIG_SMALL("QuickTime/MPEG-4/Motion JPEG 2000 format"),