3 * Copyright (c) 2001 Fabrice Bellard
4 * Copyright (c) 2009 Baptiste Coudurier <baptiste dot coudurier at gmail dot com>
6 * This file is part of FFmpeg.
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
13 * FFmpeg is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
26 //#define DEBUG_METADATA
27 //#define MOV_EXPORT_ALL_METADATA
29 #include "libavutil/intreadwrite.h"
30 #include "libavutil/avstring.h"
34 #include "libavcodec/mpeg4audio.h"
35 #include "libavcodec/mpegaudiodata.h"
36 #include "libavcodec/get_bits.h"
43 * First version by Francois Revol revol@free.fr
44 * Seek function by Gael Chardon gael.dev@4now.net
46 * Features and limitations:
47 * - reads most of the QT files I have (at least the structure),
48 * Sample QuickTime files with mp3 audio can be found at: http://www.3ivx.com/showcase.html
49 * - the code is quite ugly... maybe I won't do it recursive next time :-)
51 * Funny I didn't know about http://sourceforge.net/projects/qt-ffmpeg/
52 * when coding this :) (it's a writer anyway)
54 * Reference documents:
55 * http://www.geocities.com/xhelmboyx/quicktime/formats/qtm-layout.txt
57 * http://developer.apple.com/documentation/QuickTime/QTFF/
58 * http://developer.apple.com/documentation/QuickTime/QTFF/qtff.pdf
59 * QuickTime is a trademark of Apple (AFAIK :))
62 #include "qtpalette.h"
68 /* XXX: it's the first time I make a recursive parser I think... sorry if it's ugly :P */
70 /* those functions parse an atom */
72 0: continue to parse next atom
73 <0: error occurred, exit
75 /* links atom IDs to parse functions */
76 typedef struct MOVParseTableEntry {
78 int (*parse)(MOVContext *ctx, ByteIOContext *pb, MOVAtom atom);
81 static const MOVParseTableEntry mov_default_parse_table[];
83 static int mov_metadata_trkn(MOVContext *c, ByteIOContext *pb, unsigned len)
87 get_be16(pb); // unknown
88 snprintf(buf, sizeof(buf), "%d", get_be16(pb));
89 av_metadata_set(&c->fc->metadata, "track", buf);
91 get_be16(pb); // total tracks
96 static int mov_read_udta_string(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
98 #ifdef MOV_EXPORT_ALL_METADATA
101 char str[1024], key2[16], language[4] = {0};
102 const char *key = NULL;
104 int (*parse)(MOVContext*, ByteIOContext*, unsigned) = NULL;
107 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
108 case MKTAG(0xa9,'a','u','t'):
109 case MKTAG(0xa9,'A','R','T'): key = "author"; break;
110 case MKTAG(0xa9,'w','r','t'): key = "composer"; break;
111 case MKTAG( 'c','p','r','t'):
112 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
113 case MKTAG(0xa9,'c','m','t'):
114 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
115 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
116 case MKTAG(0xa9,'d','a','y'): key = "year"; break;
117 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
118 case MKTAG(0xa9,'t','o','o'):
119 case MKTAG(0xa9,'e','n','c'): key = "encoder"; break;
120 case MKTAG( 'd','e','s','c'): key = "description";break;
121 case MKTAG( 'l','d','e','s'): key = "synopsis"; break;
122 case MKTAG( 't','v','s','h'): key = "show"; break;
123 case MKTAG( 't','v','e','n'): key = "episode_id";break;
124 case MKTAG( 't','v','n','n'): key = "network"; break;
125 case MKTAG( 't','r','k','n'): key = "track";
126 parse = mov_metadata_trkn; break;
129 if (c->itunes_metadata && atom.size > 8) {
130 int data_size = get_be32(pb);
131 int tag = get_le32(pb);
132 if (tag == MKTAG('d','a','t','a')) {
133 get_be32(pb); // type
134 get_be32(pb); // unknown
135 str_size = data_size - 16;
138 } else if (atom.size > 4 && key && !c->itunes_metadata) {
139 str_size = get_be16(pb); // string length
140 ff_mov_lang_to_iso639(get_be16(pb), language);
143 str_size = atom.size;
145 #ifdef MOV_EXPORT_ALL_METADATA
147 snprintf(tmp_key, 5, "%.4s", (char*)&atom.type);
157 str_size = FFMIN3(sizeof(str)-1, str_size, atom.size);
160 parse(c, pb, str_size);
162 get_buffer(pb, str, str_size);
164 av_metadata_set(&c->fc->metadata, key, str);
165 if (*language && strcmp(language, "und")) {
166 snprintf(key2, sizeof(key2), "%s-%s", key, language);
167 av_metadata_set(&c->fc->metadata, key2, str);
170 #ifdef DEBUG_METADATA
171 av_log(c->fc, AV_LOG_DEBUG, "lang \"%3s\" ", language);
172 av_log(c->fc, AV_LOG_DEBUG, "tag \"%s\" value \"%s\" atom \"%.4s\" %d %lld\n",
173 key, str, (char*)&atom.type, str_size, atom.size);
179 static int mov_read_default(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
181 int64_t total_size = 0;
186 atom.size = INT64_MAX;
187 while (total_size + 8 < atom.size && !url_feof(pb)) {
188 int (*parse)(MOVContext*, ByteIOContext*, MOVAtom) = NULL;
192 a.size = get_be32(pb);
193 a.type = get_le32(pb);
196 dprintf(c->fc, "type: %08x %.4s sz: %"PRIx64" %"PRIx64" %"PRIx64"\n",
197 a.type, (char*)&a.type, a.size, atom.size, total_size);
198 if (a.size == 1) { /* 64 bit extended size */
199 a.size = get_be64(pb) - 8;
203 a.size = atom.size - total_size;
210 a.size = FFMIN(a.size, atom.size - total_size);
212 for (i = 0; mov_default_parse_table[i].type; i++)
213 if (mov_default_parse_table[i].type == a.type) {
214 parse = mov_default_parse_table[i].parse;
218 // container is user data
219 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
220 atom.type == MKTAG('i','l','s','t')))
221 parse = mov_read_udta_string;
223 if (!parse) { /* skip leaf atoms data */
224 url_fskip(pb, a.size);
226 int64_t start_pos = url_ftell(pb);
228 int err = parse(c, pb, a);
231 if (c->found_moov && c->found_mdat &&
232 (url_is_streamed(pb) || start_pos + a.size == url_fsize(pb)))
234 left = a.size - url_ftell(pb) + start_pos;
235 if (left > 0) /* skip garbage at atom end */
239 total_size += a.size;
242 if (total_size < atom.size && atom.size < 0x7ffff)
243 url_fskip(pb, atom.size - total_size);
248 static int mov_read_dref(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
251 MOVStreamContext *sc;
254 if (c->fc->nb_streams < 1)
256 st = c->fc->streams[c->fc->nb_streams-1];
259 get_be32(pb); // version + flags
260 entries = get_be32(pb);
261 if (entries >= UINT_MAX / sizeof(*sc->drefs))
263 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
265 return AVERROR(ENOMEM);
266 sc->drefs_count = entries;
268 for (i = 0; i < sc->drefs_count; i++) {
269 MOVDref *dref = &sc->drefs[i];
270 uint32_t size = get_be32(pb);
271 int64_t next = url_ftell(pb) + size - 4;
273 dref->type = get_le32(pb);
274 get_be32(pb); // version + flags
275 dprintf(c->fc, "type %.4s size %d\n", (char*)&dref->type, size);
277 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
278 /* macintosh alias record */
279 uint16_t volume_len, len;
284 volume_len = get_byte(pb);
285 volume_len = FFMIN(volume_len, 27);
286 get_buffer(pb, dref->volume, 27);
287 dref->volume[volume_len] = 0;
288 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", dref->volume, volume_len);
293 len = FFMIN(len, 63);
294 get_buffer(pb, dref->filename, 63);
295 dref->filename[len] = 0;
296 av_log(c->fc, AV_LOG_DEBUG, "filename %s, len %d\n", dref->filename, len);
300 /* read next level up_from_alias/down_to_target */
301 dref->nlvl_from = get_be16(pb);
302 dref->nlvl_to = get_be16(pb);
303 av_log(c->fc, AV_LOG_DEBUG, "nlvl from %d, nlvl to %d\n",
304 dref->nlvl_from, dref->nlvl_to);
308 for (type = 0; type != -1 && url_ftell(pb) < next; ) {
311 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
314 if (type == 2) { // absolute path
316 dref->path = av_mallocz(len+1);
318 return AVERROR(ENOMEM);
319 get_buffer(pb, dref->path, len);
320 if (len > volume_len && !strncmp(dref->path, dref->volume, volume_len)) {
322 memmove(dref->path, dref->path+volume_len, len);
325 for (j = 0; j < len; j++)
326 if (dref->path[j] == ':')
328 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
329 } else if (type == 0) { // directory name
331 dref->dir = av_malloc(len+1);
333 return AVERROR(ENOMEM);
334 get_buffer(pb, dref->dir, len);
336 for (j = 0; j < len; j++)
337 if (dref->dir[j] == ':')
339 av_log(c->fc, AV_LOG_DEBUG, "dir %s\n", dref->dir);
344 url_fseek(pb, next, SEEK_SET);
349 static int mov_read_hdlr(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
355 if (c->fc->nb_streams < 1) // meta before first trak
358 st = c->fc->streams[c->fc->nb_streams-1];
360 get_byte(pb); /* version */
361 get_be24(pb); /* flags */
364 ctype = get_le32(pb);
365 type = get_le32(pb); /* component subtype */
367 dprintf(c->fc, "ctype= %.4s (0x%08x)\n", (char*)&ctype, ctype);
368 dprintf(c->fc, "stype= %.4s\n", (char*)&type);
370 if (type == MKTAG('v','i','d','e'))
371 st->codec->codec_type = CODEC_TYPE_VIDEO;
372 else if(type == MKTAG('s','o','u','n'))
373 st->codec->codec_type = CODEC_TYPE_AUDIO;
374 else if(type == MKTAG('m','1','a',' '))
375 st->codec->codec_id = CODEC_ID_MP2;
376 else if(type == MKTAG('s','u','b','p'))
377 st->codec->codec_type = CODEC_TYPE_SUBTITLE;
379 get_be32(pb); /* component manufacture */
380 get_be32(pb); /* component flags */
381 get_be32(pb); /* component flags mask */
386 int ff_mp4_read_descr_len(ByteIOContext *pb)
391 int c = get_byte(pb);
392 len = (len << 7) | (c & 0x7f);
399 int mp4_read_descr(AVFormatContext *fc, ByteIOContext *pb, int *tag)
403 len = ff_mp4_read_descr_len(pb);
404 dprintf(fc, "MPEG4 description: tag=0x%02x len=%d\n", *tag, len);
408 #define MP4ESDescrTag 0x03
409 #define MP4DecConfigDescrTag 0x04
410 #define MP4DecSpecificDescrTag 0x05
412 static const AVCodecTag mp4_audio_types[] = {
413 { CODEC_ID_MP3ON4, AOT_PS }, /* old mp3on4 draft */
414 { CODEC_ID_MP3ON4, AOT_L1 }, /* layer 1 */
415 { CODEC_ID_MP3ON4, AOT_L2 }, /* layer 2 */
416 { CODEC_ID_MP3ON4, AOT_L3 }, /* layer 3 */
417 { CODEC_ID_MP4ALS, AOT_ALS }, /* MPEG-4 ALS */
418 { CODEC_ID_NONE, AOT_NULL },
421 int ff_mov_read_esds(AVFormatContext *fc, ByteIOContext *pb, MOVAtom atom)
426 if (fc->nb_streams < 1)
428 st = fc->streams[fc->nb_streams-1];
430 get_be32(pb); /* version + flags */
431 len = mp4_read_descr(fc, pb, &tag);
432 if (tag == MP4ESDescrTag) {
433 get_be16(pb); /* ID */
434 get_byte(pb); /* priority */
436 get_be16(pb); /* ID */
438 len = mp4_read_descr(fc, pb, &tag);
439 if (tag == MP4DecConfigDescrTag) {
440 int object_type_id = get_byte(pb);
441 get_byte(pb); /* stream type */
442 get_be24(pb); /* buffer size db */
443 get_be32(pb); /* max bitrate */
444 get_be32(pb); /* avg bitrate */
446 st->codec->codec_id= ff_codec_get_id(ff_mp4_obj_type, object_type_id);
447 dprintf(fc, "esds object type id 0x%02x\n", object_type_id);
448 len = mp4_read_descr(fc, pb, &tag);
449 if (tag == MP4DecSpecificDescrTag) {
450 dprintf(fc, "Specific MPEG4 header len=%d\n", len);
451 if((uint64_t)len > (1<<30))
453 st->codec->extradata = av_mallocz(len + FF_INPUT_BUFFER_PADDING_SIZE);
454 if (!st->codec->extradata)
455 return AVERROR(ENOMEM);
456 get_buffer(pb, st->codec->extradata, len);
457 st->codec->extradata_size = len;
458 if (st->codec->codec_id == CODEC_ID_AAC) {
459 MPEG4AudioConfig cfg;
460 ff_mpeg4audio_get_config(&cfg, st->codec->extradata,
461 st->codec->extradata_size);
462 st->codec->channels = cfg.channels;
463 if (cfg.object_type == 29 && cfg.sampling_index < 3) // old mp3on4
464 st->codec->sample_rate = ff_mpa_freq_tab[cfg.sampling_index];
466 st->codec->sample_rate = cfg.sample_rate; // ext sample rate ?
467 dprintf(fc, "mp4a config channels %d obj %d ext obj %d "
468 "sample rate %d ext sample rate %d\n", st->codec->channels,
469 cfg.object_type, cfg.ext_object_type,
470 cfg.sample_rate, cfg.ext_sample_rate);
471 if (!(st->codec->codec_id = ff_codec_get_id(mp4_audio_types,
473 st->codec->codec_id = CODEC_ID_AAC;
480 static int mov_read_esds(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
482 return ff_mov_read_esds(c->fc, pb, atom);
485 static int mov_read_pasp(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
487 const int num = get_be32(pb);
488 const int den = get_be32(pb);
491 if (c->fc->nb_streams < 1)
493 st = c->fc->streams[c->fc->nb_streams-1];
496 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
497 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num))
498 av_log(c->fc, AV_LOG_WARNING,
499 "sample aspect ratio already set to %d:%d, overriding by 'pasp' atom\n",
500 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den);
501 st->sample_aspect_ratio.num = num;
502 st->sample_aspect_ratio.den = den;
507 /* this atom contains actual media data */
508 static int mov_read_mdat(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
510 if(atom.size == 0) /* wrong one (MP4) */
513 return 0; /* now go for moov */
516 /* read major brand, minor version and compatible brands and store them as metadata */
517 static int mov_read_ftyp(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
521 char minor_ver_str[11]; /* 32 bit integer -> 10 digits + null */
522 char* comp_brands_str;
523 uint8_t type[5] = {0};
525 get_buffer(pb, type, 4);
526 if (strcmp(type, "qt "))
528 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
529 av_metadata_set(&c->fc->metadata, "major_brand", type);
530 minor_ver = get_be32(pb); /* minor version */
531 snprintf(minor_ver_str, sizeof(minor_ver_str), "%d", minor_ver);
532 av_metadata_set(&c->fc->metadata, "minor_version", minor_ver_str);
534 comp_brand_size = atom.size - 8;
535 if (comp_brand_size < 0)
537 comp_brands_str = av_malloc(comp_brand_size + 1); /* Add null terminator */
538 if (!comp_brands_str)
539 return AVERROR(ENOMEM);
540 get_buffer(pb, comp_brands_str, comp_brand_size);
541 comp_brands_str[comp_brand_size] = 0;
542 av_metadata_set(&c->fc->metadata, "compatible_brands", comp_brands_str);
543 av_freep(&comp_brands_str);
548 /* this atom should contain all header atoms */
549 static int mov_read_moov(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
551 if (mov_read_default(c, pb, atom) < 0)
553 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
554 /* so we don't parse the whole file if over a network */
556 return 0; /* now go for mdat */
559 static int mov_read_moof(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
561 c->fragment.moof_offset = url_ftell(pb) - 8;
562 dprintf(c->fc, "moof offset %llx\n", c->fragment.moof_offset);
563 return mov_read_default(c, pb, atom);
566 static int mov_read_mdhd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
569 MOVStreamContext *sc;
571 char language[4] = {0};
574 if (c->fc->nb_streams < 1)
576 st = c->fc->streams[c->fc->nb_streams-1];
579 version = get_byte(pb);
581 return -1; /* unsupported */
583 get_be24(pb); /* flags */
588 get_be32(pb); /* creation time */
589 get_be32(pb); /* modification time */
592 sc->time_scale = get_be32(pb);
593 st->duration = (version == 1) ? get_be64(pb) : get_be32(pb); /* duration */
595 lang = get_be16(pb); /* language */
596 if (ff_mov_lang_to_iso639(lang, language))
597 av_metadata_set(&st->metadata, "language", language);
598 get_be16(pb); /* quality */
603 static int mov_read_mvhd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
605 int version = get_byte(pb); /* version */
606 get_be24(pb); /* flags */
612 get_be32(pb); /* creation time */
613 get_be32(pb); /* modification time */
615 c->time_scale = get_be32(pb); /* time scale */
617 dprintf(c->fc, "time scale = %i\n", c->time_scale);
619 c->duration = (version == 1) ? get_be64(pb) : get_be32(pb); /* duration */
620 get_be32(pb); /* preferred scale */
622 get_be16(pb); /* preferred volume */
624 url_fskip(pb, 10); /* reserved */
626 url_fskip(pb, 36); /* display matrix */
628 get_be32(pb); /* preview time */
629 get_be32(pb); /* preview duration */
630 get_be32(pb); /* poster time */
631 get_be32(pb); /* selection time */
632 get_be32(pb); /* selection duration */
633 get_be32(pb); /* current time */
634 get_be32(pb); /* next track ID */
639 static int mov_read_smi(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
643 if (c->fc->nb_streams < 1)
645 st = c->fc->streams[c->fc->nb_streams-1];
647 if((uint64_t)atom.size > (1<<30))
650 // currently SVQ3 decoder expect full STSD header - so let's fake it
651 // this should be fixed and just SMI header should be passed
652 av_free(st->codec->extradata);
653 st->codec->extradata = av_mallocz(atom.size + 0x5a + FF_INPUT_BUFFER_PADDING_SIZE);
654 if (!st->codec->extradata)
655 return AVERROR(ENOMEM);
656 st->codec->extradata_size = 0x5a + atom.size;
657 memcpy(st->codec->extradata, "SVQ3", 4); // fake
658 get_buffer(pb, st->codec->extradata + 0x5a, atom.size);
659 dprintf(c->fc, "Reading SMI %"PRId64" %s\n", atom.size, st->codec->extradata + 0x5a);
663 static int mov_read_enda(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
668 if (c->fc->nb_streams < 1)
670 st = c->fc->streams[c->fc->nb_streams-1];
672 little_endian = get_be16(pb);
673 dprintf(c->fc, "enda %d\n", little_endian);
674 if (little_endian == 1) {
675 switch (st->codec->codec_id) {
676 case CODEC_ID_PCM_S24BE:
677 st->codec->codec_id = CODEC_ID_PCM_S24LE;
679 case CODEC_ID_PCM_S32BE:
680 st->codec->codec_id = CODEC_ID_PCM_S32LE;
682 case CODEC_ID_PCM_F32BE:
683 st->codec->codec_id = CODEC_ID_PCM_F32LE;
685 case CODEC_ID_PCM_F64BE:
686 st->codec->codec_id = CODEC_ID_PCM_F64LE;
695 /* FIXME modify qdm2/svq3/h264 decoders to take full atom as extradata */
696 static int mov_read_extradata(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
702 if (c->fc->nb_streams < 1) // will happen with jp2 files
704 st= c->fc->streams[c->fc->nb_streams-1];
705 size= (uint64_t)st->codec->extradata_size + atom.size + 8 + FF_INPUT_BUFFER_PADDING_SIZE;
706 if(size > INT_MAX || (uint64_t)atom.size > INT_MAX)
708 buf= av_realloc(st->codec->extradata, size);
711 st->codec->extradata= buf;
712 buf+= st->codec->extradata_size;
713 st->codec->extradata_size= size - FF_INPUT_BUFFER_PADDING_SIZE;
714 AV_WB32( buf , atom.size + 8);
715 AV_WL32( buf + 4, atom.type);
716 get_buffer(pb, buf + 8, atom.size);
720 static int mov_read_wave(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
724 if (c->fc->nb_streams < 1)
726 st = c->fc->streams[c->fc->nb_streams-1];
728 if((uint64_t)atom.size > (1<<30))
731 if (st->codec->codec_id == CODEC_ID_QDM2) {
732 // pass all frma atom to codec, needed at least for QDM2
733 av_free(st->codec->extradata);
734 st->codec->extradata = av_mallocz(atom.size + FF_INPUT_BUFFER_PADDING_SIZE);
735 if (!st->codec->extradata)
736 return AVERROR(ENOMEM);
737 st->codec->extradata_size = atom.size;
738 get_buffer(pb, st->codec->extradata, atom.size);
739 } else if (atom.size > 8) { /* to read frma, esds atoms */
740 if (mov_read_default(c, pb, atom) < 0)
743 url_fskip(pb, atom.size);
748 * This function reads atom content and puts data in extradata without tag
749 * nor size unlike mov_read_extradata.
751 static int mov_read_glbl(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
755 if (c->fc->nb_streams < 1)
757 st = c->fc->streams[c->fc->nb_streams-1];
759 if((uint64_t)atom.size > (1<<30))
762 av_free(st->codec->extradata);
763 st->codec->extradata = av_mallocz(atom.size + FF_INPUT_BUFFER_PADDING_SIZE);
764 if (!st->codec->extradata)
765 return AVERROR(ENOMEM);
766 st->codec->extradata_size = atom.size;
767 get_buffer(pb, st->codec->extradata, atom.size);
771 static int mov_read_stco(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
774 MOVStreamContext *sc;
775 unsigned int i, entries;
777 if (c->fc->nb_streams < 1)
779 st = c->fc->streams[c->fc->nb_streams-1];
782 get_byte(pb); /* version */
783 get_be24(pb); /* flags */
785 entries = get_be32(pb);
787 if(entries >= UINT_MAX/sizeof(int64_t))
790 sc->chunk_offsets = av_malloc(entries * sizeof(int64_t));
791 if (!sc->chunk_offsets)
792 return AVERROR(ENOMEM);
793 sc->chunk_count = entries;
795 if (atom.type == MKTAG('s','t','c','o'))
796 for(i=0; i<entries; i++)
797 sc->chunk_offsets[i] = get_be32(pb);
798 else if (atom.type == MKTAG('c','o','6','4'))
799 for(i=0; i<entries; i++)
800 sc->chunk_offsets[i] = get_be64(pb);
808 * Compute codec id for 'lpcm' tag.
809 * See CoreAudioTypes and AudioStreamBasicDescription at Apple.
811 enum CodecID ff_mov_get_lpcm_codec_id(int bps, int flags)
813 if (flags & 1) { // floating point
814 if (flags & 2) { // big endian
815 if (bps == 32) return CODEC_ID_PCM_F32BE;
816 else if (bps == 64) return CODEC_ID_PCM_F64BE;
818 if (bps == 32) return CODEC_ID_PCM_F32LE;
819 else if (bps == 64) return CODEC_ID_PCM_F64LE;
825 if (flags & 4) return CODEC_ID_PCM_S8;
826 else return CODEC_ID_PCM_U8;
827 else if (bps == 16) return CODEC_ID_PCM_S16BE;
828 else if (bps == 24) return CODEC_ID_PCM_S24BE;
829 else if (bps == 32) return CODEC_ID_PCM_S32BE;
832 if (flags & 4) return CODEC_ID_PCM_S8;
833 else return CODEC_ID_PCM_U8;
834 else if (bps == 16) return CODEC_ID_PCM_S16LE;
835 else if (bps == 24) return CODEC_ID_PCM_S24LE;
836 else if (bps == 32) return CODEC_ID_PCM_S32LE;
839 return CODEC_ID_NONE;
842 static int mov_read_stsd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
845 MOVStreamContext *sc;
846 int j, entries, pseudo_stream_id;
848 if (c->fc->nb_streams < 1)
850 st = c->fc->streams[c->fc->nb_streams-1];
853 get_byte(pb); /* version */
854 get_be24(pb); /* flags */
856 entries = get_be32(pb);
858 for(pseudo_stream_id=0; pseudo_stream_id<entries; pseudo_stream_id++) {
859 //Parsing Sample description table
863 int64_t start_pos = url_ftell(pb);
864 int size = get_be32(pb); /* size */
865 uint32_t format = get_le32(pb); /* data format */
868 get_be32(pb); /* reserved */
869 get_be16(pb); /* reserved */
870 dref_id = get_be16(pb);
873 if (st->codec->codec_tag &&
874 st->codec->codec_tag != format &&
875 (c->fc->video_codec_id ? ff_codec_get_id(codec_movvideo_tags, format) != c->fc->video_codec_id
876 : st->codec->codec_tag != MKTAG('j','p','e','g'))
878 /* Multiple fourcc, we skip JPEG. This is not correct, we should
879 * export it as a separate AVStream but this needs a few changes
880 * in the MOV demuxer, patch welcome. */
881 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
882 url_fskip(pb, size - (url_ftell(pb) - start_pos));
885 sc->pseudo_stream_id = st->codec->codec_tag ? -1 : pseudo_stream_id;
886 sc->dref_id= dref_id;
888 st->codec->codec_tag = format;
889 id = ff_codec_get_id(codec_movaudio_tags, format);
890 if (id<=0 && ((format&0xFFFF) == 'm'+('s'<<8) || (format&0xFFFF) == 'T'+('S'<<8)))
891 id = ff_codec_get_id(ff_codec_wav_tags, bswap_32(format)&0xFFFF);
893 if (st->codec->codec_type != CODEC_TYPE_VIDEO && id > 0) {
894 st->codec->codec_type = CODEC_TYPE_AUDIO;
895 } else if (st->codec->codec_type != CODEC_TYPE_AUDIO && /* do not overwrite codec type */
896 format && format != MKTAG('m','p','4','s')) { /* skip old asf mpeg4 tag */
897 id = ff_codec_get_id(codec_movvideo_tags, format);
899 id = ff_codec_get_id(ff_codec_bmp_tags, format);
901 st->codec->codec_type = CODEC_TYPE_VIDEO;
902 else if(st->codec->codec_type == CODEC_TYPE_DATA){
903 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
905 st->codec->codec_type = CODEC_TYPE_SUBTITLE;
909 dprintf(c->fc, "size=%d 4CC= %c%c%c%c codec_type=%d\n", size,
910 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff,
911 (format >> 24) & 0xff, st->codec->codec_type);
913 if(st->codec->codec_type==CODEC_TYPE_VIDEO) {
914 uint8_t codec_name[32];
915 unsigned int color_depth;
918 st->codec->codec_id = id;
919 get_be16(pb); /* version */
920 get_be16(pb); /* revision level */
921 get_be32(pb); /* vendor */
922 get_be32(pb); /* temporal quality */
923 get_be32(pb); /* spatial quality */
925 st->codec->width = get_be16(pb); /* width */
926 st->codec->height = get_be16(pb); /* height */
928 get_be32(pb); /* horiz resolution */
929 get_be32(pb); /* vert resolution */
930 get_be32(pb); /* data size, always 0 */
931 get_be16(pb); /* frames per samples */
933 get_buffer(pb, codec_name, 32); /* codec name, pascal string */
934 if (codec_name[0] <= 31) {
937 for (i = 0; i < codec_name[0] && pos < sizeof(st->codec->codec_name) - 3; i++) {
939 PUT_UTF8(codec_name[i+1], tmp, st->codec->codec_name[pos++] = tmp;)
941 st->codec->codec_name[pos] = 0;
942 /* codec_tag YV12 triggers an UV swap in rawdec.c */
943 if (!memcmp(st->codec->codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25))
944 st->codec->codec_tag=MKTAG('I', '4', '2', '0');
947 st->codec->bits_per_coded_sample = get_be16(pb); /* depth */
948 st->codec->color_table_id = get_be16(pb); /* colortable id */
949 dprintf(c->fc, "depth %d, ctab id %d\n",
950 st->codec->bits_per_coded_sample, st->codec->color_table_id);
951 /* figure out the palette situation */
952 color_depth = st->codec->bits_per_coded_sample & 0x1F;
953 color_greyscale = st->codec->bits_per_coded_sample & 0x20;
955 /* if the depth is 2, 4, or 8 bpp, file is palettized */
956 if ((color_depth == 2) || (color_depth == 4) ||
957 (color_depth == 8)) {
958 /* for palette traversal */
959 unsigned int color_start, color_count, color_end;
960 unsigned char r, g, b;
962 st->codec->palctrl = av_malloc(sizeof(*st->codec->palctrl));
963 if (color_greyscale) {
964 int color_index, color_dec;
965 /* compute the greyscale palette */
966 st->codec->bits_per_coded_sample = color_depth;
967 color_count = 1 << color_depth;
969 color_dec = 256 / (color_count - 1);
970 for (j = 0; j < color_count; j++) {
971 r = g = b = color_index;
972 st->codec->palctrl->palette[j] =
973 (r << 16) | (g << 8) | (b);
974 color_index -= color_dec;
978 } else if (st->codec->color_table_id) {
979 const uint8_t *color_table;
980 /* if flag bit 3 is set, use the default palette */
981 color_count = 1 << color_depth;
982 if (color_depth == 2)
983 color_table = ff_qt_default_palette_4;
984 else if (color_depth == 4)
985 color_table = ff_qt_default_palette_16;
987 color_table = ff_qt_default_palette_256;
989 for (j = 0; j < color_count; j++) {
990 r = color_table[j * 3 + 0];
991 g = color_table[j * 3 + 1];
992 b = color_table[j * 3 + 2];
993 st->codec->palctrl->palette[j] =
994 (r << 16) | (g << 8) | (b);
997 /* load the palette from the file */
998 color_start = get_be32(pb);
999 color_count = get_be16(pb);
1000 color_end = get_be16(pb);
1001 if ((color_start <= 255) &&
1002 (color_end <= 255)) {
1003 for (j = color_start; j <= color_end; j++) {
1004 /* each R, G, or B component is 16 bits;
1005 * only use the top 8 bits; skip alpha bytes
1015 st->codec->palctrl->palette[j] =
1016 (r << 16) | (g << 8) | (b);
1020 st->codec->palctrl->palette_changed = 1;
1022 } else if(st->codec->codec_type==CODEC_TYPE_AUDIO) {
1023 int bits_per_sample, flags;
1024 uint16_t version = get_be16(pb);
1026 st->codec->codec_id = id;
1027 get_be16(pb); /* revision level */
1028 get_be32(pb); /* vendor */
1030 st->codec->channels = get_be16(pb); /* channel count */
1031 dprintf(c->fc, "audio channels %d\n", st->codec->channels);
1032 st->codec->bits_per_coded_sample = get_be16(pb); /* sample size */
1034 sc->audio_cid = get_be16(pb);
1035 get_be16(pb); /* packet size = 0 */
1037 st->codec->sample_rate = ((get_be32(pb) >> 16));
1039 //Read QT version 1 fields. In version 0 these do not exist.
1040 dprintf(c->fc, "version =%d, isom =%d\n",version,c->isom);
1043 sc->samples_per_frame = get_be32(pb);
1044 get_be32(pb); /* bytes per packet */
1045 sc->bytes_per_frame = get_be32(pb);
1046 get_be32(pb); /* bytes per sample */
1047 } else if(version==2) {
1048 get_be32(pb); /* sizeof struct only */
1049 st->codec->sample_rate = av_int2dbl(get_be64(pb)); /* float 64 */
1050 st->codec->channels = get_be32(pb);
1051 get_be32(pb); /* always 0x7F000000 */
1052 st->codec->bits_per_coded_sample = get_be32(pb); /* bits per channel if sound is uncompressed */
1053 flags = get_be32(pb); /* lpcm format specific flag */
1054 sc->bytes_per_frame = get_be32(pb); /* bytes per audio packet if constant */
1055 sc->samples_per_frame = get_be32(pb); /* lpcm frames per audio packet if constant */
1056 if (format == MKTAG('l','p','c','m'))
1057 st->codec->codec_id = ff_mov_get_lpcm_codec_id(st->codec->bits_per_coded_sample, flags);
1061 switch (st->codec->codec_id) {
1062 case CODEC_ID_PCM_S8:
1063 case CODEC_ID_PCM_U8:
1064 if (st->codec->bits_per_coded_sample == 16)
1065 st->codec->codec_id = CODEC_ID_PCM_S16BE;
1067 case CODEC_ID_PCM_S16LE:
1068 case CODEC_ID_PCM_S16BE:
1069 if (st->codec->bits_per_coded_sample == 8)
1070 st->codec->codec_id = CODEC_ID_PCM_S8;
1071 else if (st->codec->bits_per_coded_sample == 24)
1072 st->codec->codec_id =
1073 st->codec->codec_id == CODEC_ID_PCM_S16BE ?
1074 CODEC_ID_PCM_S24BE : CODEC_ID_PCM_S24LE;
1076 /* set values for old format before stsd version 1 appeared */
1077 case CODEC_ID_MACE3:
1078 sc->samples_per_frame = 6;
1079 sc->bytes_per_frame = 2*st->codec->channels;
1081 case CODEC_ID_MACE6:
1082 sc->samples_per_frame = 6;
1083 sc->bytes_per_frame = 1*st->codec->channels;
1085 case CODEC_ID_ADPCM_IMA_QT:
1086 sc->samples_per_frame = 64;
1087 sc->bytes_per_frame = 34*st->codec->channels;
1090 sc->samples_per_frame = 160;
1091 sc->bytes_per_frame = 33;
1097 bits_per_sample = av_get_bits_per_sample(st->codec->codec_id);
1098 if (bits_per_sample) {
1099 st->codec->bits_per_coded_sample = bits_per_sample;
1100 sc->sample_size = (bits_per_sample >> 3) * st->codec->channels;
1102 } else if(st->codec->codec_type==CODEC_TYPE_SUBTITLE){
1103 // ttxt stsd contains display flags, justification, background
1104 // color, fonts, and default styles, so fake an atom to read it
1105 MOVAtom fake_atom = { .size = size - (url_ftell(pb) - start_pos) };
1106 if (format != AV_RL32("mp4s")) // mp4s contains a regular esds atom
1107 mov_read_glbl(c, pb, fake_atom);
1108 st->codec->codec_id= id;
1109 st->codec->width = sc->width;
1110 st->codec->height = sc->height;
1112 /* other codec type, just skip (rtp, mp4s, tmcd ...) */
1113 url_fskip(pb, size - (url_ftell(pb) - start_pos));
1115 /* this will read extra atoms at the end (wave, alac, damr, avcC, SMI ...) */
1116 a.size = size - (url_ftell(pb) - start_pos);
1118 if (mov_read_default(c, pb, a) < 0)
1120 } else if (a.size > 0)
1121 url_fskip(pb, a.size);
1124 if(st->codec->codec_type==CODEC_TYPE_AUDIO && st->codec->sample_rate==0 && sc->time_scale>1)
1125 st->codec->sample_rate= sc->time_scale;
1127 /* special codec parameters handling */
1128 switch (st->codec->codec_id) {
1129 #if CONFIG_DV_DEMUXER
1130 case CODEC_ID_DVAUDIO:
1131 c->dv_fctx = avformat_alloc_context();
1132 c->dv_demux = dv_init_demux(c->dv_fctx);
1134 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
1137 sc->dv_audio_container = 1;
1138 st->codec->codec_id = CODEC_ID_PCM_S16LE;
1141 /* no ifdef since parameters are always those */
1142 case CODEC_ID_QCELP:
1143 // force sample rate for qcelp when not stored in mov
1144 if (st->codec->codec_tag != MKTAG('Q','c','l','p'))
1145 st->codec->sample_rate = 8000;
1146 st->codec->frame_size= 160;
1147 st->codec->channels= 1; /* really needed */
1149 case CODEC_ID_AMR_NB:
1150 case CODEC_ID_AMR_WB:
1151 st->codec->frame_size= sc->samples_per_frame;
1152 st->codec->channels= 1; /* really needed */
1153 /* force sample rate for amr, stsd in 3gp does not store sample rate */
1154 if (st->codec->codec_id == CODEC_ID_AMR_NB)
1155 st->codec->sample_rate = 8000;
1156 else if (st->codec->codec_id == CODEC_ID_AMR_WB)
1157 st->codec->sample_rate = 16000;
1161 st->codec->codec_type = CODEC_TYPE_AUDIO; /* force type after stsd for m1a hdlr */
1162 st->need_parsing = AVSTREAM_PARSE_FULL;
1165 case CODEC_ID_ADPCM_MS:
1166 case CODEC_ID_ADPCM_IMA_WAV:
1167 st->codec->block_align = sc->bytes_per_frame;
1170 if (st->codec->extradata_size == 36) {
1171 st->codec->frame_size = AV_RB32(st->codec->extradata+12);
1172 st->codec->channels = AV_RB8 (st->codec->extradata+21);
1182 static int mov_read_stsc(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1185 MOVStreamContext *sc;
1186 unsigned int i, entries;
1188 if (c->fc->nb_streams < 1)
1190 st = c->fc->streams[c->fc->nb_streams-1];
1193 get_byte(pb); /* version */
1194 get_be24(pb); /* flags */
1196 entries = get_be32(pb);
1198 dprintf(c->fc, "track[%i].stsc.entries = %i\n", c->fc->nb_streams-1, entries);
1200 if(entries >= UINT_MAX / sizeof(*sc->stsc_data))
1202 sc->stsc_data = av_malloc(entries * sizeof(*sc->stsc_data));
1204 return AVERROR(ENOMEM);
1205 sc->stsc_count = entries;
1207 for(i=0; i<entries; i++) {
1208 sc->stsc_data[i].first = get_be32(pb);
1209 sc->stsc_data[i].count = get_be32(pb);
1210 sc->stsc_data[i].id = get_be32(pb);
1215 static int mov_read_stps(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1218 MOVStreamContext *sc;
1219 unsigned i, entries;
1221 if (c->fc->nb_streams < 1)
1223 st = c->fc->streams[c->fc->nb_streams-1];
1226 get_be32(pb); // version + flags
1228 entries = get_be32(pb);
1229 if (entries >= UINT_MAX / sizeof(*sc->stps_data))
1231 sc->stps_data = av_malloc(entries * sizeof(*sc->stps_data));
1233 return AVERROR(ENOMEM);
1234 sc->stps_count = entries;
1236 for (i = 0; i < entries; i++) {
1237 sc->stps_data[i] = get_be32(pb);
1238 //dprintf(c->fc, "stps %d\n", sc->stps_data[i]);
1244 static int mov_read_stss(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1247 MOVStreamContext *sc;
1248 unsigned int i, entries;
1250 if (c->fc->nb_streams < 1)
1252 st = c->fc->streams[c->fc->nb_streams-1];
1255 get_byte(pb); /* version */
1256 get_be24(pb); /* flags */
1258 entries = get_be32(pb);
1260 dprintf(c->fc, "keyframe_count = %d\n", entries);
1262 if(entries >= UINT_MAX / sizeof(int))
1264 sc->keyframes = av_malloc(entries * sizeof(int));
1266 return AVERROR(ENOMEM);
1267 sc->keyframe_count = entries;
1269 for(i=0; i<entries; i++) {
1270 sc->keyframes[i] = get_be32(pb);
1271 //dprintf(c->fc, "keyframes[]=%d\n", sc->keyframes[i]);
1276 static int mov_read_stsz(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1279 MOVStreamContext *sc;
1280 unsigned int i, entries, sample_size, field_size, num_bytes;
1284 if (c->fc->nb_streams < 1)
1286 st = c->fc->streams[c->fc->nb_streams-1];
1289 get_byte(pb); /* version */
1290 get_be24(pb); /* flags */
1292 if (atom.type == MKTAG('s','t','s','z')) {
1293 sample_size = get_be32(pb);
1294 if (!sc->sample_size) /* do not overwrite value computed in stsd */
1295 sc->sample_size = sample_size;
1299 get_be24(pb); /* reserved */
1300 field_size = get_byte(pb);
1302 entries = get_be32(pb);
1304 dprintf(c->fc, "sample_size = %d sample_count = %d\n", sc->sample_size, entries);
1306 sc->sample_count = entries;
1310 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
1311 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %d\n", field_size);
1315 if (entries >= UINT_MAX / sizeof(int) || entries >= (UINT_MAX - 4) / field_size)
1317 sc->sample_sizes = av_malloc(entries * sizeof(int));
1318 if (!sc->sample_sizes)
1319 return AVERROR(ENOMEM);
1321 num_bytes = (entries*field_size+4)>>3;
1323 buf = av_malloc(num_bytes+FF_INPUT_BUFFER_PADDING_SIZE);
1325 av_freep(&sc->sample_sizes);
1326 return AVERROR(ENOMEM);
1329 if (get_buffer(pb, buf, num_bytes) < num_bytes) {
1330 av_freep(&sc->sample_sizes);
1335 init_get_bits(&gb, buf, 8*num_bytes);
1337 for(i=0; i<entries; i++)
1338 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
1344 static int mov_read_stts(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1347 MOVStreamContext *sc;
1348 unsigned int i, entries;
1350 int64_t total_sample_count=0;
1352 if (c->fc->nb_streams < 1)
1354 st = c->fc->streams[c->fc->nb_streams-1];
1357 get_byte(pb); /* version */
1358 get_be24(pb); /* flags */
1359 entries = get_be32(pb);
1361 dprintf(c->fc, "track[%i].stts.entries = %i\n", c->fc->nb_streams-1, entries);
1363 if(entries >= UINT_MAX / sizeof(*sc->stts_data))
1365 sc->stts_data = av_malloc(entries * sizeof(*sc->stts_data));
1367 return AVERROR(ENOMEM);
1368 sc->stts_count = entries;
1370 for(i=0; i<entries; i++) {
1371 int sample_duration;
1374 sample_count=get_be32(pb);
1375 sample_duration = get_be32(pb);
1376 sc->stts_data[i].count= sample_count;
1377 sc->stts_data[i].duration= sample_duration;
1379 dprintf(c->fc, "sample_count=%d, sample_duration=%d\n",sample_count,sample_duration);
1381 duration+=(int64_t)sample_duration*sample_count;
1382 total_sample_count+=sample_count;
1385 st->nb_frames= total_sample_count;
1387 st->duration= duration;
1391 static int mov_read_ctts(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1394 MOVStreamContext *sc;
1395 unsigned int i, entries;
1397 if (c->fc->nb_streams < 1)
1399 st = c->fc->streams[c->fc->nb_streams-1];
1402 get_byte(pb); /* version */
1403 get_be24(pb); /* flags */
1404 entries = get_be32(pb);
1406 dprintf(c->fc, "track[%i].ctts.entries = %i\n", c->fc->nb_streams-1, entries);
1408 if(entries >= UINT_MAX / sizeof(*sc->ctts_data))
1410 sc->ctts_data = av_malloc(entries * sizeof(*sc->ctts_data));
1412 return AVERROR(ENOMEM);
1413 sc->ctts_count = entries;
1415 for(i=0; i<entries; i++) {
1416 int count =get_be32(pb);
1417 int duration =get_be32(pb);
1419 sc->ctts_data[i].count = count;
1420 sc->ctts_data[i].duration= duration;
1422 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
1425 dprintf(c->fc, "dts shift %d\n", sc->dts_shift);
1430 static void mov_build_index(MOVContext *mov, AVStream *st)
1432 MOVStreamContext *sc = st->priv_data;
1433 int64_t current_offset;
1434 int64_t current_dts = 0;
1435 unsigned int stts_index = 0;
1436 unsigned int stsc_index = 0;
1437 unsigned int stss_index = 0;
1438 unsigned int stps_index = 0;
1440 uint64_t stream_size = 0;
1442 /* adjust first dts according to edit list */
1443 if (sc->time_offset) {
1444 int rescaled = sc->time_offset < 0 ? av_rescale(sc->time_offset, sc->time_scale, mov->time_scale) : sc->time_offset;
1445 current_dts = -rescaled;
1446 if (sc->ctts_data && sc->ctts_data[0].duration / sc->stts_data[0].duration > 16) {
1447 /* more than 16 frames delay, dts are likely wrong
1448 this happens with files created by iMovie */
1450 st->codec->has_b_frames = 1;
1454 /* only use old uncompressed audio chunk demuxing when stts specifies it */
1455 if (!(st->codec->codec_type == CODEC_TYPE_AUDIO &&
1456 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
1457 unsigned int current_sample = 0;
1458 unsigned int stts_sample = 0;
1459 unsigned int sample_size;
1460 unsigned int distance = 0;
1461 int key_off = sc->keyframes && sc->keyframes[0] == 1;
1463 current_dts -= sc->dts_shift;
1465 for (i = 0; i < sc->chunk_count; i++) {
1466 current_offset = sc->chunk_offsets[i];
1467 if (stsc_index + 1 < sc->stsc_count &&
1468 i + 1 == sc->stsc_data[stsc_index + 1].first)
1470 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
1472 if (current_sample >= sc->sample_count) {
1473 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
1477 if (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index]) {
1479 if (stss_index + 1 < sc->keyframe_count)
1481 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
1483 if (stps_index + 1 < sc->stps_count)
1488 sample_size = sc->sample_size > 0 ? sc->sample_size : sc->sample_sizes[current_sample];
1489 if(sc->pseudo_stream_id == -1 ||
1490 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
1491 av_add_index_entry(st, current_offset, current_dts, sample_size, distance,
1492 keyframe ? AVINDEX_KEYFRAME : 0);
1493 dprintf(mov->fc, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
1494 "size %d, distance %d, keyframe %d\n", st->index, current_sample,
1495 current_offset, current_dts, sample_size, distance, keyframe);
1498 current_offset += sample_size;
1499 stream_size += sample_size;
1500 current_dts += sc->stts_data[stts_index].duration;
1504 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
1510 if (st->duration > 0)
1511 st->codec->bit_rate = stream_size*8*sc->time_scale/st->duration;
1513 for (i = 0; i < sc->chunk_count; i++) {
1514 unsigned chunk_samples;
1516 current_offset = sc->chunk_offsets[i];
1517 if (stsc_index + 1 < sc->stsc_count &&
1518 i + 1 == sc->stsc_data[stsc_index + 1].first)
1520 chunk_samples = sc->stsc_data[stsc_index].count;
1522 if (sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
1523 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
1527 while (chunk_samples > 0) {
1528 unsigned size, samples;
1530 if (sc->samples_per_frame >= 160) { // gsm
1531 samples = sc->samples_per_frame;
1532 size = sc->bytes_per_frame;
1534 if (sc->samples_per_frame > 1) {
1535 samples = FFMIN((1024 / sc->samples_per_frame)*
1536 sc->samples_per_frame, chunk_samples);
1537 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
1539 samples = FFMIN(1024, chunk_samples);
1540 size = samples * sc->sample_size;
1544 av_add_index_entry(st, current_offset, current_dts, size, 0, AVINDEX_KEYFRAME);
1545 dprintf(mov->fc, "AVIndex stream %d, chunk %d, offset %"PRIx64", dts %"PRId64", "
1546 "size %d, duration %d\n", st->index, i, current_offset, current_dts,
1549 current_offset += size;
1550 current_dts += samples;
1551 chunk_samples -= samples;
1557 static int mov_open_dref(ByteIOContext **pb, char *src, MOVDref *ref)
1559 /* try absolute path */
1560 if (!url_fopen(pb, ref->path, URL_RDONLY))
1563 /* try relative path */
1564 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
1565 char filename[1024];
1569 /* find a source dir */
1570 src_path = strrchr(src, '/');
1576 /* find a next level down to target */
1577 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
1578 if (ref->path[l] == '/') {
1579 if (i == ref->nlvl_to - 1)
1585 /* compose filename if next level down to target was found */
1586 if (i == ref->nlvl_to - 1) {
1587 memcpy(filename, src, src_path - src);
1588 filename[src_path - src] = 0;
1590 for (i = 1; i < ref->nlvl_from; i++)
1591 av_strlcat(filename, "../", 1024);
1593 av_strlcat(filename, ref->path + l + 1, 1024);
1595 if (!url_fopen(pb, filename, URL_RDONLY))
1600 return AVERROR(ENOENT);
1603 static int mov_read_trak(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1606 MOVStreamContext *sc;
1609 st = av_new_stream(c->fc, c->fc->nb_streams);
1610 if (!st) return AVERROR(ENOMEM);
1611 sc = av_mallocz(sizeof(MOVStreamContext));
1612 if (!sc) return AVERROR(ENOMEM);
1615 st->codec->codec_type = CODEC_TYPE_DATA;
1616 sc->ffindex = st->index;
1618 if ((ret = mov_read_default(c, pb, atom)) < 0)
1622 if (sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
1623 (!sc->sample_size && !sc->sample_count))) {
1624 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
1629 if (!sc->time_scale) {
1630 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", st->index);
1631 sc->time_scale = c->time_scale;
1632 if (!sc->time_scale)
1636 av_set_pts_info(st, 64, 1, sc->time_scale);
1638 if (st->codec->codec_type == CODEC_TYPE_AUDIO &&
1639 !st->codec->frame_size && sc->stts_count == 1) {
1640 st->codec->frame_size = av_rescale(sc->stts_data[0].duration,
1641 st->codec->sample_rate, sc->time_scale);
1642 dprintf(c->fc, "frame size %d\n", st->codec->frame_size);
1645 mov_build_index(c, st);
1647 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
1648 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
1649 if (mov_open_dref(&sc->pb, c->fc->filename, dref) < 0)
1650 av_log(c->fc, AV_LOG_ERROR,
1651 "stream %d, error opening alias: path='%s', dir='%s', "
1652 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
1653 st->index, dref->path, dref->dir, dref->filename,
1654 dref->volume, dref->nlvl_from, dref->nlvl_to);
1658 if (st->codec->codec_type == CODEC_TYPE_VIDEO) {
1659 if (st->codec->width != sc->width || st->codec->height != sc->height) {
1660 AVRational r = av_d2q(((double)st->codec->height * sc->width) /
1661 ((double)st->codec->width * sc->height), INT_MAX);
1662 if (st->sample_aspect_ratio.num)
1663 st->sample_aspect_ratio = av_mul_q(st->sample_aspect_ratio, r);
1665 st->sample_aspect_ratio = r;
1668 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
1669 sc->time_scale*st->nb_frames, st->duration, INT_MAX);
1672 switch (st->codec->codec_id) {
1673 #if CONFIG_H261_DECODER
1676 #if CONFIG_H263_DECODER
1679 #if CONFIG_H264_DECODER
1682 #if CONFIG_MPEG4_DECODER
1683 case CODEC_ID_MPEG4:
1685 st->codec->width = 0; /* let decoder init width/height */
1686 st->codec->height= 0;
1690 /* Do not need those anymore. */
1691 av_freep(&sc->chunk_offsets);
1692 av_freep(&sc->stsc_data);
1693 av_freep(&sc->sample_sizes);
1694 av_freep(&sc->keyframes);
1695 av_freep(&sc->stts_data);
1696 av_freep(&sc->stps_data);
1701 static int mov_read_ilst(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1704 c->itunes_metadata = 1;
1705 ret = mov_read_default(c, pb, atom);
1706 c->itunes_metadata = 0;
1710 static int mov_read_meta(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1712 while (atom.size > 8) {
1713 uint32_t tag = get_le32(pb);
1715 if (tag == MKTAG('h','d','l','r')) {
1716 url_fseek(pb, -8, SEEK_CUR);
1718 return mov_read_default(c, pb, atom);
1724 static int mov_read_tkhd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1729 int64_t disp_transform[2];
1730 int display_matrix[3][2];
1732 MOVStreamContext *sc;
1735 if (c->fc->nb_streams < 1)
1737 st = c->fc->streams[c->fc->nb_streams-1];
1740 version = get_byte(pb);
1741 get_be24(pb); /* flags */
1743 MOV_TRACK_ENABLED 0x0001
1744 MOV_TRACK_IN_MOVIE 0x0002
1745 MOV_TRACK_IN_PREVIEW 0x0004
1746 MOV_TRACK_IN_POSTER 0x0008
1753 get_be32(pb); /* creation time */
1754 get_be32(pb); /* modification time */
1756 st->id = (int)get_be32(pb); /* track id (NOT 0 !)*/
1757 get_be32(pb); /* reserved */
1759 /* highlevel (considering edits) duration in movie timebase */
1760 (version == 1) ? get_be64(pb) : get_be32(pb);
1761 get_be32(pb); /* reserved */
1762 get_be32(pb); /* reserved */
1764 get_be16(pb); /* layer */
1765 get_be16(pb); /* alternate group */
1766 get_be16(pb); /* volume */
1767 get_be16(pb); /* reserved */
1769 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
1770 // they're kept in fixed point format through all calculations
1771 // ignore u,v,z b/c we don't need the scale factor to calc aspect ratio
1772 for (i = 0; i < 3; i++) {
1773 display_matrix[i][0] = get_be32(pb); // 16.16 fixed point
1774 display_matrix[i][1] = get_be32(pb); // 16.16 fixed point
1775 get_be32(pb); // 2.30 fixed point (not used)
1778 width = get_be32(pb); // 16.16 fixed point track width
1779 height = get_be32(pb); // 16.16 fixed point track height
1780 sc->width = width >> 16;
1781 sc->height = height >> 16;
1783 // transform the display width/height according to the matrix
1784 // skip this if the display matrix is the default identity matrix
1785 // or if it is rotating the picture, ex iPhone 3GS
1786 // to keep the same scale, use [width height 1<<16]
1787 if (width && height &&
1788 ((display_matrix[0][0] != 65536 ||
1789 display_matrix[1][1] != 65536) &&
1790 !display_matrix[0][1] &&
1791 !display_matrix[1][0] &&
1792 !display_matrix[2][0] && !display_matrix[2][1])) {
1793 for (i = 0; i < 2; i++)
1795 (int64_t) width * display_matrix[0][i] +
1796 (int64_t) height * display_matrix[1][i] +
1797 ((int64_t) display_matrix[2][i] << 16);
1799 //sample aspect ratio is new width/height divided by old width/height
1800 st->sample_aspect_ratio = av_d2q(
1801 ((double) disp_transform[0] * height) /
1802 ((double) disp_transform[1] * width), INT_MAX);
1807 static int mov_read_tfhd(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1809 MOVFragment *frag = &c->fragment;
1810 MOVTrackExt *trex = NULL;
1811 int flags, track_id, i;
1813 get_byte(pb); /* version */
1814 flags = get_be24(pb);
1816 track_id = get_be32(pb);
1819 frag->track_id = track_id;
1820 for (i = 0; i < c->trex_count; i++)
1821 if (c->trex_data[i].track_id == frag->track_id) {
1822 trex = &c->trex_data[i];
1826 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
1830 if (flags & 0x01) frag->base_data_offset = get_be64(pb);
1831 else frag->base_data_offset = frag->moof_offset;
1832 if (flags & 0x02) frag->stsd_id = get_be32(pb);
1833 else frag->stsd_id = trex->stsd_id;
1835 frag->duration = flags & 0x08 ? get_be32(pb) : trex->duration;
1836 frag->size = flags & 0x10 ? get_be32(pb) : trex->size;
1837 frag->flags = flags & 0x20 ? get_be32(pb) : trex->flags;
1838 dprintf(c->fc, "frag flags 0x%x\n", frag->flags);
1842 static int mov_read_trex(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1846 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
1848 trex = av_realloc(c->trex_data, (c->trex_count+1)*sizeof(*c->trex_data));
1850 return AVERROR(ENOMEM);
1851 c->trex_data = trex;
1852 trex = &c->trex_data[c->trex_count++];
1853 get_byte(pb); /* version */
1854 get_be24(pb); /* flags */
1855 trex->track_id = get_be32(pb);
1856 trex->stsd_id = get_be32(pb);
1857 trex->duration = get_be32(pb);
1858 trex->size = get_be32(pb);
1859 trex->flags = get_be32(pb);
1863 static int mov_read_trun(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1865 MOVFragment *frag = &c->fragment;
1866 AVStream *st = NULL;
1867 MOVStreamContext *sc;
1870 int data_offset = 0;
1871 unsigned entries, first_sample_flags = frag->flags;
1872 int flags, distance, i;
1874 for (i = 0; i < c->fc->nb_streams; i++) {
1875 if (c->fc->streams[i]->id == frag->track_id) {
1876 st = c->fc->streams[i];
1881 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
1885 if (sc->pseudo_stream_id+1 != frag->stsd_id)
1887 get_byte(pb); /* version */
1888 flags = get_be24(pb);
1889 entries = get_be32(pb);
1890 dprintf(c->fc, "flags 0x%x entries %d\n", flags, entries);
1891 if (flags & 0x001) data_offset = get_be32(pb);
1892 if (flags & 0x004) first_sample_flags = get_be32(pb);
1893 if (flags & 0x800) {
1895 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
1897 ctts_data = av_realloc(sc->ctts_data,
1898 (entries+sc->ctts_count)*sizeof(*sc->ctts_data));
1900 return AVERROR(ENOMEM);
1901 sc->ctts_data = ctts_data;
1904 offset = frag->base_data_offset + data_offset;
1906 dprintf(c->fc, "first sample flags 0x%x\n", first_sample_flags);
1907 for (i = 0; i < entries; i++) {
1908 unsigned sample_size = frag->size;
1909 int sample_flags = i ? frag->flags : first_sample_flags;
1910 unsigned sample_duration = frag->duration;
1913 if (flags & 0x100) sample_duration = get_be32(pb);
1914 if (flags & 0x200) sample_size = get_be32(pb);
1915 if (flags & 0x400) sample_flags = get_be32(pb);
1916 if (flags & 0x800) {
1917 sc->ctts_data[sc->ctts_count].count = 1;
1918 sc->ctts_data[sc->ctts_count].duration = get_be32(pb);
1921 if ((keyframe = st->codec->codec_type == CODEC_TYPE_AUDIO ||
1922 (flags & 0x004 && !i && !sample_flags) || sample_flags & 0x2000000))
1924 av_add_index_entry(st, offset, dts, sample_size, distance,
1925 keyframe ? AVINDEX_KEYFRAME : 0);
1926 dprintf(c->fc, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
1927 "size %d, distance %d, keyframe %d\n", st->index, sc->sample_count+i,
1928 offset, dts, sample_size, distance, keyframe);
1930 dts += sample_duration;
1931 offset += sample_size;
1933 frag->moof_offset = offset;
1938 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
1939 /* like the files created with Adobe Premiere 5.0, for samples see */
1940 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
1941 static int mov_read_wide(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1946 return 0; /* continue */
1947 if (get_be32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
1948 url_fskip(pb, atom.size - 4);
1951 atom.type = get_le32(pb);
1953 if (atom.type != MKTAG('m','d','a','t')) {
1954 url_fskip(pb, atom.size);
1957 err = mov_read_mdat(c, pb, atom);
1961 static int mov_read_cmov(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
1966 uint8_t *moov_data; /* uncompressed data */
1967 long cmov_len, moov_len;
1970 get_be32(pb); /* dcom atom */
1971 if (get_le32(pb) != MKTAG('d','c','o','m'))
1973 if (get_le32(pb) != MKTAG('z','l','i','b')) {
1974 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !");
1977 get_be32(pb); /* cmvd atom */
1978 if (get_le32(pb) != MKTAG('c','m','v','d'))
1980 moov_len = get_be32(pb); /* uncompressed size */
1981 cmov_len = atom.size - 6 * 4;
1983 cmov_data = av_malloc(cmov_len);
1985 return AVERROR(ENOMEM);
1986 moov_data = av_malloc(moov_len);
1989 return AVERROR(ENOMEM);
1991 get_buffer(pb, cmov_data, cmov_len);
1992 if(uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
1993 goto free_and_return;
1994 if(init_put_byte(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
1995 goto free_and_return;
1996 atom.type = MKTAG('m','o','o','v');
1997 atom.size = moov_len;
1999 // { int fd = open("/tmp/uncompheader.mov", O_WRONLY | O_CREAT); write(fd, moov_data, moov_len); close(fd); }
2001 ret = mov_read_default(c, &ctx, atom);
2007 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
2012 /* edit list atom */
2013 static int mov_read_elst(MOVContext *c, ByteIOContext *pb, MOVAtom atom)
2015 MOVStreamContext *sc;
2018 if (c->fc->nb_streams < 1)
2020 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
2022 get_byte(pb); /* version */
2023 get_be24(pb); /* flags */
2024 edit_count = get_be32(pb); /* entries */
2026 if((uint64_t)edit_count*12+8 > atom.size)
2029 for(i=0; i<edit_count; i++){
2031 int duration = get_be32(pb); /* Track duration */
2032 time = get_be32(pb); /* Media time */
2033 get_be32(pb); /* Media rate */
2034 if (i == 0 && time >= -1) {
2035 sc->time_offset = time != -1 ? time : -duration;
2040 av_log(c->fc, AV_LOG_WARNING, "multiple edit list entries, "
2041 "a/v desync might occur, patch welcome\n");
2043 dprintf(c->fc, "track[%i].edit_count = %i\n", c->fc->nb_streams-1, edit_count);
2047 static const MOVParseTableEntry mov_default_parse_table[] = {
2048 { MKTAG('a','v','s','s'), mov_read_extradata },
2049 { MKTAG('c','o','6','4'), mov_read_stco },
2050 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
2051 { MKTAG('d','i','n','f'), mov_read_default },
2052 { MKTAG('d','r','e','f'), mov_read_dref },
2053 { MKTAG('e','d','t','s'), mov_read_default },
2054 { MKTAG('e','l','s','t'), mov_read_elst },
2055 { MKTAG('e','n','d','a'), mov_read_enda },
2056 { MKTAG('f','i','e','l'), mov_read_extradata },
2057 { MKTAG('f','t','y','p'), mov_read_ftyp },
2058 { MKTAG('g','l','b','l'), mov_read_glbl },
2059 { MKTAG('h','d','l','r'), mov_read_hdlr },
2060 { MKTAG('i','l','s','t'), mov_read_ilst },
2061 { MKTAG('j','p','2','h'), mov_read_extradata },
2062 { MKTAG('m','d','a','t'), mov_read_mdat },
2063 { MKTAG('m','d','h','d'), mov_read_mdhd },
2064 { MKTAG('m','d','i','a'), mov_read_default },
2065 { MKTAG('m','e','t','a'), mov_read_meta },
2066 { MKTAG('m','i','n','f'), mov_read_default },
2067 { MKTAG('m','o','o','f'), mov_read_moof },
2068 { MKTAG('m','o','o','v'), mov_read_moov },
2069 { MKTAG('m','v','e','x'), mov_read_default },
2070 { MKTAG('m','v','h','d'), mov_read_mvhd },
2071 { MKTAG('S','M','I',' '), mov_read_smi }, /* Sorenson extension ??? */
2072 { MKTAG('a','l','a','c'), mov_read_extradata }, /* alac specific atom */
2073 { MKTAG('a','v','c','C'), mov_read_glbl },
2074 { MKTAG('p','a','s','p'), mov_read_pasp },
2075 { MKTAG('s','t','b','l'), mov_read_default },
2076 { MKTAG('s','t','c','o'), mov_read_stco },
2077 { MKTAG('s','t','p','s'), mov_read_stps },
2078 { MKTAG('s','t','s','c'), mov_read_stsc },
2079 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
2080 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
2081 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
2082 { MKTAG('s','t','t','s'), mov_read_stts },
2083 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
2084 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
2085 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
2086 { MKTAG('t','r','a','k'), mov_read_trak },
2087 { MKTAG('t','r','a','f'), mov_read_default },
2088 { MKTAG('t','r','e','x'), mov_read_trex },
2089 { MKTAG('t','r','u','n'), mov_read_trun },
2090 { MKTAG('u','d','t','a'), mov_read_default },
2091 { MKTAG('w','a','v','e'), mov_read_wave },
2092 { MKTAG('e','s','d','s'), mov_read_esds },
2093 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
2094 { MKTAG('c','m','o','v'), mov_read_cmov },
2098 static int mov_probe(AVProbeData *p)
2100 unsigned int offset;
2104 /* check file header */
2107 /* ignore invalid offset */
2108 if ((offset + 8) > (unsigned int)p->buf_size)
2110 tag = AV_RL32(p->buf + offset + 4);
2112 /* check for obvious tags */
2113 case MKTAG('j','P',' ',' '): /* jpeg 2000 signature */
2114 case MKTAG('m','o','o','v'):
2115 case MKTAG('m','d','a','t'):
2116 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
2117 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
2118 case MKTAG('f','t','y','p'):
2119 return AVPROBE_SCORE_MAX;
2120 /* those are more common words, so rate then a bit less */
2121 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
2122 case MKTAG('w','i','d','e'):
2123 case MKTAG('f','r','e','e'):
2124 case MKTAG('j','u','n','k'):
2125 case MKTAG('p','i','c','t'):
2126 return AVPROBE_SCORE_MAX - 5;
2127 case MKTAG(0x82,0x82,0x7f,0x7d):
2128 case MKTAG('s','k','i','p'):
2129 case MKTAG('u','u','i','d'):
2130 case MKTAG('p','r','f','l'):
2131 offset = AV_RB32(p->buf+offset) + offset;
2132 /* if we only find those cause probedata is too small at least rate them */
2133 score = AVPROBE_SCORE_MAX - 50;
2136 /* unrecognized tag */
2143 static int mov_read_header(AVFormatContext *s, AVFormatParameters *ap)
2145 MOVContext *mov = s->priv_data;
2146 ByteIOContext *pb = s->pb;
2148 MOVAtom atom = { 0 };
2151 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
2152 if(!url_is_streamed(pb))
2153 atom.size = url_fsize(pb);
2155 atom.size = INT64_MAX;
2157 /* check MOV header */
2158 if ((err = mov_read_default(mov, pb, atom)) < 0) {
2159 av_log(s, AV_LOG_ERROR, "error reading header: %d\n", err);
2162 if (!mov->found_moov) {
2163 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
2166 dprintf(mov->fc, "on_parse_exit_offset=%lld\n", url_ftell(pb));
2171 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
2173 AVIndexEntry *sample = NULL;
2174 int64_t best_dts = INT64_MAX;
2176 for (i = 0; i < s->nb_streams; i++) {
2177 AVStream *avst = s->streams[i];
2178 MOVStreamContext *msc = avst->priv_data;
2179 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
2180 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
2181 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
2182 dprintf(s, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
2183 if (!sample || (url_is_streamed(s->pb) && current_sample->pos < sample->pos) ||
2184 (!url_is_streamed(s->pb) &&
2185 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
2186 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
2187 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
2188 sample = current_sample;
2197 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
2199 MOVContext *mov = s->priv_data;
2200 MOVStreamContext *sc;
2201 AVIndexEntry *sample;
2202 AVStream *st = NULL;
2205 sample = mov_find_next_sample(s, &st);
2207 mov->found_mdat = 0;
2208 if (!url_is_streamed(s->pb) ||
2209 mov_read_default(mov, s->pb, (MOVAtom){ 0, INT64_MAX }) < 0 ||
2212 dprintf(s, "read fragments, offset 0x%llx\n", url_ftell(s->pb));
2216 /* must be done just before reading, to avoid infinite loop on sample */
2217 sc->current_sample++;
2219 if (st->discard != AVDISCARD_ALL) {
2220 if (url_fseek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
2221 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
2222 sc->ffindex, sample->pos);
2225 ret = av_get_packet(sc->pb, pkt, sample->size);
2228 #if CONFIG_DV_DEMUXER
2229 if (mov->dv_demux && sc->dv_audio_container) {
2230 dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size);
2233 ret = dv_get_packet(mov->dv_demux, pkt);
2240 pkt->stream_index = sc->ffindex;
2241 pkt->dts = sample->timestamp;
2242 if (sc->ctts_data) {
2243 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
2244 /* update ctts context */
2246 if (sc->ctts_index < sc->ctts_count &&
2247 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
2249 sc->ctts_sample = 0;
2252 pkt->dts = AV_NOPTS_VALUE;
2254 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
2255 st->index_entries[sc->current_sample].timestamp : st->duration;
2256 pkt->duration = next_dts - pkt->dts;
2257 pkt->pts = pkt->dts;
2259 if (st->discard == AVDISCARD_ALL)
2261 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? PKT_FLAG_KEY : 0;
2262 pkt->pos = sample->pos;
2263 dprintf(s, "stream %d, pts %"PRId64", dts %"PRId64", pos 0x%"PRIx64", duration %d\n",
2264 pkt->stream_index, pkt->pts, pkt->dts, pkt->pos, pkt->duration);
2268 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
2270 MOVStreamContext *sc = st->priv_data;
2271 int sample, time_sample;
2274 sample = av_index_search_timestamp(st, timestamp, flags);
2275 dprintf(s, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
2276 if (sample < 0) /* not sure what to do */
2278 sc->current_sample = sample;
2279 dprintf(s, "stream %d, found sample %d\n", st->index, sc->current_sample);
2280 /* adjust ctts index */
2281 if (sc->ctts_data) {
2283 for (i = 0; i < sc->ctts_count; i++) {
2284 int next = time_sample + sc->ctts_data[i].count;
2285 if (next > sc->current_sample) {
2287 sc->ctts_sample = sc->current_sample - time_sample;
2296 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
2299 int64_t seek_timestamp, timestamp;
2303 if (stream_index >= s->nb_streams)
2305 if (sample_time < 0)
2308 st = s->streams[stream_index];
2309 sample = mov_seek_stream(s, st, sample_time, flags);
2313 /* adjust seek timestamp to found sample timestamp */
2314 seek_timestamp = st->index_entries[sample].timestamp;
2316 for (i = 0; i < s->nb_streams; i++) {
2318 if (stream_index == i)
2321 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
2322 mov_seek_stream(s, st, timestamp, flags);
2327 static int mov_read_close(AVFormatContext *s)
2329 MOVContext *mov = s->priv_data;
2332 for (i = 0; i < s->nb_streams; i++) {
2333 AVStream *st = s->streams[i];
2334 MOVStreamContext *sc = st->priv_data;
2336 av_freep(&sc->ctts_data);
2337 for (j = 0; j < sc->drefs_count; j++) {
2338 av_freep(&sc->drefs[j].path);
2339 av_freep(&sc->drefs[j].dir);
2341 av_freep(&sc->drefs);
2342 if (sc->pb && sc->pb != s->pb)
2345 av_freep(&st->codec->palctrl);
2348 if (mov->dv_demux) {
2349 for(i = 0; i < mov->dv_fctx->nb_streams; i++) {
2350 av_freep(&mov->dv_fctx->streams[i]->codec);
2351 av_freep(&mov->dv_fctx->streams[i]);
2353 av_freep(&mov->dv_fctx);
2354 av_freep(&mov->dv_demux);
2357 av_freep(&mov->trex_data);
2362 AVInputFormat mov_demuxer = {
2363 "mov,mp4,m4a,3gp,3g2,mj2",
2364 NULL_IF_CONFIG_SMALL("QuickTime/MPEG-4/Motion JPEG 2000 format"),