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 MOV_EXPORT_ALL_METADATA
28 #include "libavutil/intreadwrite.h"
29 #include "libavutil/intfloat_readwrite.h"
30 #include "libavutil/mathematics.h"
31 #include "libavutil/avstring.h"
32 #include "libavutil/dict.h"
34 #include "avio_internal.h"
37 #include "libavcodec/get_bits.h"
44 * First version by Francois Revol revol@free.fr
45 * Seek function by Gael Chardon gael.dev@4now.net
47 * Features and limitations:
48 * - reads most of the QT files I have (at least the structure),
49 * Sample QuickTime files with mp3 audio can be found at: http://www.3ivx.com/showcase.html
50 * - the code is quite ugly... maybe I won't do it recursive next time :-)
52 * Funny I didn't know about http://sourceforge.net/projects/qt-ffmpeg/
53 * when coding this :) (it's a writer anyway)
55 * Reference documents:
56 * http://www.geocities.com/xhelmboyx/quicktime/formats/qtm-layout.txt
58 * http://developer.apple.com/documentation/QuickTime/QTFF/
59 * http://developer.apple.com/documentation/QuickTime/QTFF/qtff.pdf
60 * QuickTime is a trademark of Apple (AFAIK :))
63 #include "qtpalette.h"
69 /* XXX: it's the first time I make a recursive parser I think... sorry if it's ugly :P */
71 /* those functions parse an atom */
73 0: continue to parse next atom
74 <0: error occurred, exit
76 /* links atom IDs to parse functions */
77 typedef struct MOVParseTableEntry {
79 int (*parse)(MOVContext *ctx, AVIOContext *pb, MOVAtom atom);
82 static const MOVParseTableEntry mov_default_parse_table[];
84 static int mov_metadata_track_or_disc_number(MOVContext *c, AVIOContext *pb,
85 unsigned len, const char *key)
90 avio_rb16(pb); // unknown
91 current = avio_rb16(pb);
92 total = avio_rb16(pb);
94 snprintf(buf, sizeof(buf), "%d", current);
96 snprintf(buf, sizeof(buf), "%d/%d", current, total);
97 av_dict_set(&c->fc->metadata, key, buf, 0);
102 static const uint32_t mac_to_unicode[128] = {
103 0x00C4,0x00C5,0x00C7,0x00C9,0x00D1,0x00D6,0x00DC,0x00E1,
104 0x00E0,0x00E2,0x00E4,0x00E3,0x00E5,0x00E7,0x00E9,0x00E8,
105 0x00EA,0x00EB,0x00ED,0x00EC,0x00EE,0x00EF,0x00F1,0x00F3,
106 0x00F2,0x00F4,0x00F6,0x00F5,0x00FA,0x00F9,0x00FB,0x00FC,
107 0x2020,0x00B0,0x00A2,0x00A3,0x00A7,0x2022,0x00B6,0x00DF,
108 0x00AE,0x00A9,0x2122,0x00B4,0x00A8,0x2260,0x00C6,0x00D8,
109 0x221E,0x00B1,0x2264,0x2265,0x00A5,0x00B5,0x2202,0x2211,
110 0x220F,0x03C0,0x222B,0x00AA,0x00BA,0x03A9,0x00E6,0x00F8,
111 0x00BF,0x00A1,0x00AC,0x221A,0x0192,0x2248,0x2206,0x00AB,
112 0x00BB,0x2026,0x00A0,0x00C0,0x00C3,0x00D5,0x0152,0x0153,
113 0x2013,0x2014,0x201C,0x201D,0x2018,0x2019,0x00F7,0x25CA,
114 0x00FF,0x0178,0x2044,0x20AC,0x2039,0x203A,0xFB01,0xFB02,
115 0x2021,0x00B7,0x201A,0x201E,0x2030,0x00C2,0x00CA,0x00C1,
116 0x00CB,0x00C8,0x00CD,0x00CE,0x00CF,0x00CC,0x00D3,0x00D4,
117 0xF8FF,0x00D2,0x00DA,0x00DB,0x00D9,0x0131,0x02C6,0x02DC,
118 0x00AF,0x02D8,0x02D9,0x02DA,0x00B8,0x02DD,0x02DB,0x02C7,
121 static int mov_read_mac_string(MOVContext *c, AVIOContext *pb, int len,
122 char *dst, int dstlen)
125 char *end = dst+dstlen-1;
128 for (i = 0; i < len; i++) {
129 uint8_t t, c = avio_r8(pb);
130 if (c < 0x80 && p < end)
133 PUT_UTF8(mac_to_unicode[c-0x80], t, if (p < end) *p++ = t;);
139 static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom)
141 #ifdef MOV_EXPORT_ALL_METADATA
144 char str[1024], key2[16], language[4] = {0};
145 const char *key = NULL;
146 uint16_t str_size, langcode = 0;
147 uint32_t data_type = 0;
148 int (*parse)(MOVContext*, AVIOContext*, unsigned, const char*) = NULL;
151 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
152 case MKTAG(0xa9,'a','u','t'):
153 case MKTAG(0xa9,'A','R','T'): key = "artist"; break;
154 case MKTAG( 'a','A','R','T'): key = "album_artist"; break;
155 case MKTAG(0xa9,'w','r','t'): key = "composer"; break;
156 case MKTAG( 'c','p','r','t'):
157 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
158 case MKTAG(0xa9,'g','r','p'): key = "grouping"; break;
159 case MKTAG(0xa9,'l','y','r'): key = "lyrics"; break;
160 case MKTAG(0xa9,'c','m','t'):
161 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
162 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
163 case MKTAG(0xa9,'d','a','y'): key = "date"; break;
164 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
165 case MKTAG(0xa9,'t','o','o'):
166 case MKTAG(0xa9,'s','w','r'): key = "encoder"; break;
167 case MKTAG(0xa9,'e','n','c'): key = "encoder"; break;
168 case MKTAG( 'd','e','s','c'): key = "description";break;
169 case MKTAG( 'l','d','e','s'): key = "synopsis"; break;
170 case MKTAG( 't','v','s','h'): key = "show"; break;
171 case MKTAG( 't','v','e','n'): key = "episode_id";break;
172 case MKTAG( 't','v','n','n'): key = "network"; break;
173 case MKTAG( 't','r','k','n'): key = "track";
174 parse = mov_metadata_track_or_disc_number; break;
175 case MKTAG( 'd','i','s','k'): key = "disc";
176 parse = mov_metadata_track_or_disc_number; break;
179 if (c->itunes_metadata && atom.size > 8) {
180 int data_size = avio_rb32(pb);
181 int tag = avio_rl32(pb);
182 if (tag == MKTAG('d','a','t','a')) {
183 data_type = avio_rb32(pb); // type
184 avio_rb32(pb); // unknown
185 str_size = data_size - 16;
188 } else if (atom.size > 4 && key && !c->itunes_metadata) {
189 str_size = avio_rb16(pb); // string length
190 langcode = avio_rb16(pb);
191 ff_mov_lang_to_iso639(langcode, language);
194 str_size = atom.size;
196 #ifdef MOV_EXPORT_ALL_METADATA
198 snprintf(tmp_key, 5, "%.4s", (char*)&atom.type);
208 str_size = FFMIN3(sizeof(str)-1, str_size, atom.size);
211 parse(c, pb, str_size, key);
213 if (data_type == 3 || (data_type == 0 && langcode < 0x800)) { // MAC Encoded
214 mov_read_mac_string(c, pb, str_size, str, sizeof(str));
216 avio_read(pb, str, str_size);
219 av_dict_set(&c->fc->metadata, key, str, 0);
220 if (*language && strcmp(language, "und")) {
221 snprintf(key2, sizeof(key2), "%s-%s", key, language);
222 av_dict_set(&c->fc->metadata, key2, str, 0);
225 av_dlog(c->fc, "lang \"%3s\" ", language);
226 av_dlog(c->fc, "tag \"%s\" value \"%s\" atom \"%.4s\" %d %"PRId64"\n",
227 key, str, (char*)&atom.type, str_size, atom.size);
232 static int mov_read_chpl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
235 int i, nb_chapters, str_len, version;
238 if ((atom.size -= 5) < 0)
241 version = avio_r8(pb);
244 avio_rb32(pb); // ???
245 nb_chapters = avio_r8(pb);
247 for (i = 0; i < nb_chapters; i++) {
251 start = avio_rb64(pb);
252 str_len = avio_r8(pb);
254 if ((atom.size -= 9+str_len) < 0)
257 avio_read(pb, str, str_len);
259 ff_new_chapter(c->fc, i, (AVRational){1,10000000}, start, AV_NOPTS_VALUE, str);
264 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom)
266 int64_t total_size = 0;
271 atom.size = INT64_MAX;
272 while (total_size + 8 < atom.size && !url_feof(pb)) {
273 int (*parse)(MOVContext*, AVIOContext*, MOVAtom) = NULL;
276 if (atom.size >= 8) {
277 a.size = avio_rb32(pb);
278 a.type = avio_rl32(pb);
280 av_dlog(c->fc, "type: %08x '%.4s' parent:'%.4s' sz: %"PRId64" %"PRId64" %"PRId64"\n",
281 a.type, (char*)&a.type, (char*)&atom.type, a.size, total_size, atom.size);
283 if (a.size == 1) { /* 64 bit extended size */
284 a.size = avio_rb64(pb) - 8;
288 a.size = atom.size - total_size;
295 a.size = FFMIN(a.size, atom.size - total_size);
297 for (i = 0; mov_default_parse_table[i].type; i++)
298 if (mov_default_parse_table[i].type == a.type) {
299 parse = mov_default_parse_table[i].parse;
303 // container is user data
304 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
305 atom.type == MKTAG('i','l','s','t')))
306 parse = mov_read_udta_string;
308 if (!parse) { /* skip leaf atoms data */
309 avio_skip(pb, a.size);
311 int64_t start_pos = avio_tell(pb);
313 int err = parse(c, pb, a);
316 if (c->found_moov && c->found_mdat &&
317 (!pb->seekable || start_pos + a.size == avio_size(pb)))
319 left = a.size - avio_tell(pb) + start_pos;
320 if (left > 0) /* skip garbage at atom end */
324 total_size += a.size;
327 if (total_size < atom.size && atom.size < 0x7ffff)
328 avio_skip(pb, atom.size - total_size);
333 static int mov_read_dref(MOVContext *c, AVIOContext *pb, MOVAtom atom)
336 MOVStreamContext *sc;
339 if (c->fc->nb_streams < 1)
341 st = c->fc->streams[c->fc->nb_streams-1];
344 avio_rb32(pb); // version + flags
345 entries = avio_rb32(pb);
346 if (entries >= UINT_MAX / sizeof(*sc->drefs))
348 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
350 return AVERROR(ENOMEM);
351 sc->drefs_count = entries;
353 for (i = 0; i < sc->drefs_count; i++) {
354 MOVDref *dref = &sc->drefs[i];
355 uint32_t size = avio_rb32(pb);
356 int64_t next = avio_tell(pb) + size - 4;
361 dref->type = avio_rl32(pb);
362 avio_rb32(pb); // version + flags
363 av_dlog(c->fc, "type %.4s size %d\n", (char*)&dref->type, size);
365 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
366 /* macintosh alias record */
367 uint16_t volume_len, len;
372 volume_len = avio_r8(pb);
373 volume_len = FFMIN(volume_len, 27);
374 avio_read(pb, dref->volume, 27);
375 dref->volume[volume_len] = 0;
376 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", dref->volume, volume_len);
381 len = FFMIN(len, 63);
382 avio_read(pb, dref->filename, 63);
383 dref->filename[len] = 0;
384 av_log(c->fc, AV_LOG_DEBUG, "filename %s, len %d\n", dref->filename, len);
388 /* read next level up_from_alias/down_to_target */
389 dref->nlvl_from = avio_rb16(pb);
390 dref->nlvl_to = avio_rb16(pb);
391 av_log(c->fc, AV_LOG_DEBUG, "nlvl from %d, nlvl to %d\n",
392 dref->nlvl_from, dref->nlvl_to);
396 for (type = 0; type != -1 && avio_tell(pb) < next; ) {
397 type = avio_rb16(pb);
399 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
402 if (type == 2) { // absolute path
404 dref->path = av_mallocz(len+1);
406 return AVERROR(ENOMEM);
407 avio_read(pb, dref->path, len);
408 if (len > volume_len && !strncmp(dref->path, dref->volume, volume_len)) {
410 memmove(dref->path, dref->path+volume_len, len);
413 for (j = 0; j < len; j++)
414 if (dref->path[j] == ':')
416 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
417 } else if (type == 0) { // directory name
419 dref->dir = av_malloc(len+1);
421 return AVERROR(ENOMEM);
422 avio_read(pb, dref->dir, len);
424 for (j = 0; j < len; j++)
425 if (dref->dir[j] == ':')
427 av_log(c->fc, AV_LOG_DEBUG, "dir %s\n", dref->dir);
432 avio_seek(pb, next, SEEK_SET);
437 static int mov_read_hdlr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
441 uint32_t av_unused ctype;
443 if (c->fc->nb_streams < 1) // meta before first trak
446 st = c->fc->streams[c->fc->nb_streams-1];
448 avio_r8(pb); /* version */
449 avio_rb24(pb); /* flags */
452 ctype = avio_rl32(pb);
453 type = avio_rl32(pb); /* component subtype */
455 av_dlog(c->fc, "ctype= %.4s (0x%08x)\n", (char*)&ctype, ctype);
456 av_dlog(c->fc, "stype= %.4s\n", (char*)&type);
458 if (type == MKTAG('v','i','d','e'))
459 st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
460 else if (type == MKTAG('s','o','u','n'))
461 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
462 else if (type == MKTAG('m','1','a',' '))
463 st->codec->codec_id = CODEC_ID_MP2;
464 else if ((type == MKTAG('s','u','b','p')) || (type == MKTAG('c','l','c','p')))
465 st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
467 avio_rb32(pb); /* component manufacture */
468 avio_rb32(pb); /* component flags */
469 avio_rb32(pb); /* component flags mask */
474 int ff_mov_read_esds(AVFormatContext *fc, AVIOContext *pb, MOVAtom atom)
479 if (fc->nb_streams < 1)
481 st = fc->streams[fc->nb_streams-1];
483 avio_rb32(pb); /* version + flags */
484 ff_mp4_read_descr(fc, pb, &tag);
485 if (tag == MP4ESDescrTag) {
486 ff_mp4_parse_es_descr(pb, NULL);
488 avio_rb16(pb); /* ID */
490 ff_mp4_read_descr(fc, pb, &tag);
491 if (tag == MP4DecConfigDescrTag)
492 ff_mp4_read_dec_config_descr(fc, st, pb);
496 static int mov_read_esds(MOVContext *c, AVIOContext *pb, MOVAtom atom)
498 return ff_mov_read_esds(c->fc, pb, atom);
501 static int mov_read_dac3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
504 int ac3info, acmod, lfeon, bsmod;
506 if (c->fc->nb_streams < 1)
508 st = c->fc->streams[c->fc->nb_streams-1];
510 ac3info = avio_rb24(pb);
511 bsmod = (ac3info >> 14) & 0x7;
512 acmod = (ac3info >> 11) & 0x7;
513 lfeon = (ac3info >> 10) & 0x1;
514 st->codec->channels = ((int[]){2,1,2,3,3,4,4,5})[acmod] + lfeon;
515 st->codec->audio_service_type = bsmod;
516 if (st->codec->channels > 1 && bsmod == 0x7)
517 st->codec->audio_service_type = AV_AUDIO_SERVICE_TYPE_KARAOKE;
522 static int mov_read_wfex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
526 if (c->fc->nb_streams < 1)
528 st = c->fc->streams[c->fc->nb_streams-1];
530 ff_get_wav_header(pb, st->codec, atom.size);
535 static int mov_read_pasp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
537 const int num = avio_rb32(pb);
538 const int den = avio_rb32(pb);
541 if (c->fc->nb_streams < 1)
543 st = c->fc->streams[c->fc->nb_streams-1];
545 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
546 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num)) {
547 av_log(c->fc, AV_LOG_WARNING,
548 "sample aspect ratio already set to %d:%d, ignoring 'pasp' atom (%d:%d)\n",
549 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den,
551 } else if (den != 0) {
552 st->sample_aspect_ratio.num = num;
553 st->sample_aspect_ratio.den = den;
558 /* this atom contains actual media data */
559 static int mov_read_mdat(MOVContext *c, AVIOContext *pb, MOVAtom atom)
561 if (atom.size == 0) /* wrong one (MP4) */
564 return 0; /* now go for moov */
567 /* read major brand, minor version and compatible brands and store them as metadata */
568 static int mov_read_ftyp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
572 char minor_ver_str[11]; /* 32 bit integer -> 10 digits + null */
573 char* comp_brands_str;
574 uint8_t type[5] = {0};
576 avio_read(pb, type, 4);
577 if (strcmp(type, "qt "))
579 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
580 av_dict_set(&c->fc->metadata, "major_brand", type, 0);
581 minor_ver = avio_rb32(pb); /* minor version */
582 snprintf(minor_ver_str, sizeof(minor_ver_str), "%d", minor_ver);
583 av_dict_set(&c->fc->metadata, "minor_version", minor_ver_str, 0);
585 comp_brand_size = atom.size - 8;
586 if (comp_brand_size < 0)
588 comp_brands_str = av_malloc(comp_brand_size + 1); /* Add null terminator */
589 if (!comp_brands_str)
590 return AVERROR(ENOMEM);
591 avio_read(pb, comp_brands_str, comp_brand_size);
592 comp_brands_str[comp_brand_size] = 0;
593 av_dict_set(&c->fc->metadata, "compatible_brands", comp_brands_str, 0);
594 av_freep(&comp_brands_str);
599 /* this atom should contain all header atoms */
600 static int mov_read_moov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
602 if (mov_read_default(c, pb, atom) < 0)
604 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
605 /* so we don't parse the whole file if over a network */
607 return 0; /* now go for mdat */
610 static int mov_read_moof(MOVContext *c, AVIOContext *pb, MOVAtom atom)
612 c->fragment.moof_offset = avio_tell(pb) - 8;
613 av_dlog(c->fc, "moof offset %"PRIx64"\n", c->fragment.moof_offset);
614 return mov_read_default(c, pb, atom);
617 static void mov_metadata_creation_time(AVDictionary **metadata, time_t time)
622 time -= 2082844800; /* seconds between 1904-01-01 and Epoch */
625 strftime(buffer, sizeof(buffer), "%Y-%m-%d %H:%M:%S", ptm);
626 av_dict_set(metadata, "creation_time", buffer, 0);
630 static int mov_read_mdhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
633 MOVStreamContext *sc;
635 char language[4] = {0};
637 time_t creation_time;
639 if (c->fc->nb_streams < 1)
641 st = c->fc->streams[c->fc->nb_streams-1];
644 version = avio_r8(pb);
646 return -1; /* unsupported */
648 avio_rb24(pb); /* flags */
650 creation_time = avio_rb64(pb);
653 creation_time = avio_rb32(pb);
654 avio_rb32(pb); /* modification time */
656 mov_metadata_creation_time(&st->metadata, creation_time);
658 sc->time_scale = avio_rb32(pb);
659 st->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
661 lang = avio_rb16(pb); /* language */
662 if (ff_mov_lang_to_iso639(lang, language))
663 av_dict_set(&st->metadata, "language", language, 0);
664 avio_rb16(pb); /* quality */
669 static int mov_read_mvhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
671 time_t creation_time;
672 int version = avio_r8(pb); /* version */
673 avio_rb24(pb); /* flags */
676 creation_time = avio_rb64(pb);
679 creation_time = avio_rb32(pb);
680 avio_rb32(pb); /* modification time */
682 mov_metadata_creation_time(&c->fc->metadata, creation_time);
683 c->time_scale = avio_rb32(pb); /* time scale */
685 av_dlog(c->fc, "time scale = %i\n", c->time_scale);
687 c->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
688 avio_rb32(pb); /* preferred scale */
690 avio_rb16(pb); /* preferred volume */
692 avio_skip(pb, 10); /* reserved */
694 avio_skip(pb, 36); /* display matrix */
696 avio_rb32(pb); /* preview time */
697 avio_rb32(pb); /* preview duration */
698 avio_rb32(pb); /* poster time */
699 avio_rb32(pb); /* selection time */
700 avio_rb32(pb); /* selection duration */
701 avio_rb32(pb); /* current time */
702 avio_rb32(pb); /* next track ID */
707 static int mov_read_smi(MOVContext *c, AVIOContext *pb, MOVAtom atom)
711 if (c->fc->nb_streams < 1)
713 st = c->fc->streams[c->fc->nb_streams-1];
715 if ((uint64_t)atom.size > (1<<30))
718 // currently SVQ3 decoder expect full STSD header - so let's fake it
719 // this should be fixed and just SMI header should be passed
720 av_free(st->codec->extradata);
721 st->codec->extradata = av_mallocz(atom.size + 0x5a + FF_INPUT_BUFFER_PADDING_SIZE);
722 if (!st->codec->extradata)
723 return AVERROR(ENOMEM);
724 st->codec->extradata_size = 0x5a + atom.size;
725 memcpy(st->codec->extradata, "SVQ3", 4); // fake
726 avio_read(pb, st->codec->extradata + 0x5a, atom.size);
727 av_dlog(c->fc, "Reading SMI %"PRId64" %s\n", atom.size, st->codec->extradata + 0x5a);
731 static int mov_read_enda(MOVContext *c, AVIOContext *pb, MOVAtom atom)
736 if (c->fc->nb_streams < 1)
738 st = c->fc->streams[c->fc->nb_streams-1];
740 little_endian = avio_rb16(pb) & 0xFF;
741 av_dlog(c->fc, "enda %d\n", little_endian);
742 if (little_endian == 1) {
743 switch (st->codec->codec_id) {
744 case CODEC_ID_PCM_S24BE:
745 st->codec->codec_id = CODEC_ID_PCM_S24LE;
747 case CODEC_ID_PCM_S32BE:
748 st->codec->codec_id = CODEC_ID_PCM_S32LE;
750 case CODEC_ID_PCM_F32BE:
751 st->codec->codec_id = CODEC_ID_PCM_F32LE;
753 case CODEC_ID_PCM_F64BE:
754 st->codec->codec_id = CODEC_ID_PCM_F64LE;
763 /* FIXME modify qdm2/svq3/h264 decoders to take full atom as extradata */
764 static int mov_read_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
765 enum CodecID codec_id)
771 if (c->fc->nb_streams < 1) // will happen with jp2 files
773 st= c->fc->streams[c->fc->nb_streams-1];
775 if (st->codec->codec_id != codec_id)
776 return 0; /* unexpected codec_id - don't mess with extradata */
778 size= (uint64_t)st->codec->extradata_size + atom.size + 8 + FF_INPUT_BUFFER_PADDING_SIZE;
779 if (size > INT_MAX || (uint64_t)atom.size > INT_MAX)
781 buf= av_realloc(st->codec->extradata, size);
784 st->codec->extradata= buf;
785 buf+= st->codec->extradata_size;
786 st->codec->extradata_size= size - FF_INPUT_BUFFER_PADDING_SIZE;
787 AV_WB32( buf , atom.size + 8);
788 AV_WL32( buf + 4, atom.type);
789 avio_read(pb, buf + 8, atom.size);
793 /* wrapper functions for reading ALAC/AVS/MJPEG/MJPEG2000 extradata atoms only for those codecs */
794 static int mov_read_alac(MOVContext *c, AVIOContext *pb, MOVAtom atom)
796 return mov_read_extradata(c, pb, atom, CODEC_ID_ALAC);
799 static int mov_read_avss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
801 return mov_read_extradata(c, pb, atom, CODEC_ID_AVS);
804 static int mov_read_fiel(MOVContext *c, AVIOContext *pb, MOVAtom atom)
806 return mov_read_extradata(c, pb, atom, CODEC_ID_MJPEG);
809 static int mov_read_jp2h(MOVContext *c, AVIOContext *pb, MOVAtom atom)
811 return mov_read_extradata(c, pb, atom, CODEC_ID_JPEG2000);
814 static int mov_read_wave(MOVContext *c, AVIOContext *pb, MOVAtom atom)
818 if (c->fc->nb_streams < 1)
820 st = c->fc->streams[c->fc->nb_streams-1];
822 if ((uint64_t)atom.size > (1<<30))
825 if (st->codec->codec_id == CODEC_ID_QDM2 || st->codec->codec_id == CODEC_ID_QDMC) {
826 // pass all frma atom to codec, needed at least for QDMC and QDM2
827 av_free(st->codec->extradata);
828 st->codec->extradata = av_mallocz(atom.size + FF_INPUT_BUFFER_PADDING_SIZE);
829 if (!st->codec->extradata)
830 return AVERROR(ENOMEM);
831 st->codec->extradata_size = atom.size;
832 avio_read(pb, st->codec->extradata, atom.size);
833 } else if (atom.size > 8) { /* to read frma, esds atoms */
834 if (mov_read_default(c, pb, atom) < 0)
837 avio_skip(pb, atom.size);
842 * This function reads atom content and puts data in extradata without tag
843 * nor size unlike mov_read_extradata.
845 static int mov_read_glbl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
849 if (c->fc->nb_streams < 1)
851 st = c->fc->streams[c->fc->nb_streams-1];
853 if ((uint64_t)atom.size > (1<<30))
856 av_free(st->codec->extradata);
857 st->codec->extradata = av_mallocz(atom.size + FF_INPUT_BUFFER_PADDING_SIZE);
858 if (!st->codec->extradata)
859 return AVERROR(ENOMEM);
860 st->codec->extradata_size = atom.size;
861 avio_read(pb, st->codec->extradata, atom.size);
866 * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
867 * but can have extradata appended at the end after the 40 bytes belonging
870 static int mov_read_strf(MOVContext *c, AVIOContext *pb, MOVAtom atom)
874 if (c->fc->nb_streams < 1)
878 st = c->fc->streams[c->fc->nb_streams-1];
880 if ((uint64_t)atom.size > (1<<30))
883 av_free(st->codec->extradata);
884 st->codec->extradata = av_mallocz(atom.size - 40 + FF_INPUT_BUFFER_PADDING_SIZE);
885 if (!st->codec->extradata)
886 return AVERROR(ENOMEM);
887 st->codec->extradata_size = atom.size - 40;
889 avio_read(pb, st->codec->extradata, atom.size - 40);
893 static int mov_read_stco(MOVContext *c, AVIOContext *pb, MOVAtom atom)
896 MOVStreamContext *sc;
897 unsigned int i, entries;
899 if (c->fc->nb_streams < 1)
901 st = c->fc->streams[c->fc->nb_streams-1];
904 avio_r8(pb); /* version */
905 avio_rb24(pb); /* flags */
907 entries = avio_rb32(pb);
909 if (entries >= UINT_MAX/sizeof(int64_t))
912 sc->chunk_offsets = av_malloc(entries * sizeof(int64_t));
913 if (!sc->chunk_offsets)
914 return AVERROR(ENOMEM);
915 sc->chunk_count = entries;
917 if (atom.type == MKTAG('s','t','c','o'))
918 for (i=0; i<entries; i++)
919 sc->chunk_offsets[i] = avio_rb32(pb);
920 else if (atom.type == MKTAG('c','o','6','4'))
921 for (i=0; i<entries; i++)
922 sc->chunk_offsets[i] = avio_rb64(pb);
930 * Compute codec id for 'lpcm' tag.
931 * See CoreAudioTypes and AudioStreamBasicDescription at Apple.
933 enum CodecID ff_mov_get_lpcm_codec_id(int bps, int flags)
935 if (flags & 1) { // floating point
936 if (flags & 2) { // big endian
937 if (bps == 32) return CODEC_ID_PCM_F32BE;
938 else if (bps == 64) return CODEC_ID_PCM_F64BE;
940 if (bps == 32) return CODEC_ID_PCM_F32LE;
941 else if (bps == 64) return CODEC_ID_PCM_F64LE;
947 if (flags & 4) return CODEC_ID_PCM_S8;
948 else return CODEC_ID_PCM_U8;
949 else if (bps == 16) return CODEC_ID_PCM_S16BE;
950 else if (bps == 24) return CODEC_ID_PCM_S24BE;
951 else if (bps == 32) return CODEC_ID_PCM_S32BE;
954 if (flags & 4) return CODEC_ID_PCM_S8;
955 else return CODEC_ID_PCM_U8;
956 else if (bps == 16) return CODEC_ID_PCM_S16LE;
957 else if (bps == 24) return CODEC_ID_PCM_S24LE;
958 else if (bps == 32) return CODEC_ID_PCM_S32LE;
961 return CODEC_ID_NONE;
964 int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries)
967 MOVStreamContext *sc;
968 int j, pseudo_stream_id;
970 if (c->fc->nb_streams < 1)
972 st = c->fc->streams[c->fc->nb_streams-1];
975 for (pseudo_stream_id=0; pseudo_stream_id<entries; pseudo_stream_id++) {
976 //Parsing Sample description table
979 MOVAtom a = { AV_RL32("stsd") };
980 int64_t start_pos = avio_tell(pb);
981 int size = avio_rb32(pb); /* size */
982 uint32_t format = avio_rl32(pb); /* data format */
985 avio_rb32(pb); /* reserved */
986 avio_rb16(pb); /* reserved */
987 dref_id = avio_rb16(pb);
990 if (st->codec->codec_tag &&
991 st->codec->codec_tag != format &&
992 (c->fc->video_codec_id ? ff_codec_get_id(codec_movvideo_tags, format) != c->fc->video_codec_id
993 : st->codec->codec_tag != MKTAG('j','p','e','g'))
995 /* Multiple fourcc, we skip JPEG. This is not correct, we should
996 * export it as a separate AVStream but this needs a few changes
997 * in the MOV demuxer, patch welcome. */
999 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
1000 avio_skip(pb, size - (avio_tell(pb) - start_pos));
1003 /* we cannot demux concatenated h264 streams because of different extradata */
1004 if (st->codec->codec_tag && st->codec->codec_tag == AV_RL32("avc1"))
1006 sc->pseudo_stream_id = st->codec->codec_tag ? -1 : pseudo_stream_id;
1007 sc->dref_id= dref_id;
1009 st->codec->codec_tag = format;
1010 id = ff_codec_get_id(codec_movaudio_tags, format);
1011 if (id<=0 && ((format&0xFFFF) == 'm'+('s'<<8) || (format&0xFFFF) == 'T'+('S'<<8)))
1012 id = ff_codec_get_id(ff_codec_wav_tags, av_bswap32(format)&0xFFFF);
1014 if (st->codec->codec_type != AVMEDIA_TYPE_VIDEO && id > 0) {
1015 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
1016 } else if (st->codec->codec_type != AVMEDIA_TYPE_AUDIO && /* do not overwrite codec type */
1017 format && format != MKTAG('m','p','4','s')) { /* skip old asf mpeg4 tag */
1018 id = ff_codec_get_id(codec_movvideo_tags, format);
1020 id = ff_codec_get_id(ff_codec_bmp_tags, format);
1022 st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
1023 else if (st->codec->codec_type == AVMEDIA_TYPE_DATA){
1024 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
1026 st->codec->codec_type = AVMEDIA_TYPE_SUBTITLE;
1030 av_dlog(c->fc, "size=%d 4CC= %c%c%c%c codec_type=%d\n", size,
1031 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff,
1032 (format >> 24) & 0xff, st->codec->codec_type);
1034 if (st->codec->codec_type==AVMEDIA_TYPE_VIDEO) {
1035 unsigned int color_depth, len;
1036 int color_greyscale;
1038 st->codec->codec_id = id;
1039 avio_rb16(pb); /* version */
1040 avio_rb16(pb); /* revision level */
1041 avio_rb32(pb); /* vendor */
1042 avio_rb32(pb); /* temporal quality */
1043 avio_rb32(pb); /* spatial quality */
1045 st->codec->width = avio_rb16(pb); /* width */
1046 st->codec->height = avio_rb16(pb); /* height */
1048 avio_rb32(pb); /* horiz resolution */
1049 avio_rb32(pb); /* vert resolution */
1050 avio_rb32(pb); /* data size, always 0 */
1051 avio_rb16(pb); /* frames per samples */
1053 len = avio_r8(pb); /* codec name, pascal string */
1056 mov_read_mac_string(c, pb, len, st->codec->codec_name, 32);
1058 avio_skip(pb, 31 - len);
1059 /* codec_tag YV12 triggers an UV swap in rawdec.c */
1060 if (!memcmp(st->codec->codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25))
1061 st->codec->codec_tag=MKTAG('I', '4', '2', '0');
1063 st->codec->bits_per_coded_sample = avio_rb16(pb); /* depth */
1064 st->codec->color_table_id = avio_rb16(pb); /* colortable id */
1065 av_dlog(c->fc, "depth %d, ctab id %d\n",
1066 st->codec->bits_per_coded_sample, st->codec->color_table_id);
1067 /* figure out the palette situation */
1068 color_depth = st->codec->bits_per_coded_sample & 0x1F;
1069 color_greyscale = st->codec->bits_per_coded_sample & 0x20;
1071 /* if the depth is 2, 4, or 8 bpp, file is palettized */
1072 if ((color_depth == 2) || (color_depth == 4) ||
1073 (color_depth == 8)) {
1074 /* for palette traversal */
1075 unsigned int color_start, color_count, color_end;
1076 unsigned char r, g, b;
1078 if (color_greyscale) {
1079 int color_index, color_dec;
1080 /* compute the greyscale palette */
1081 st->codec->bits_per_coded_sample = color_depth;
1082 color_count = 1 << color_depth;
1084 color_dec = 256 / (color_count - 1);
1085 for (j = 0; j < color_count; j++) {
1086 if (id == CODEC_ID_CINEPAK){
1087 r = g = b = color_count - 1 - color_index;
1089 r = g = b = color_index;
1091 (r << 16) | (g << 8) | (b);
1092 color_index -= color_dec;
1093 if (color_index < 0)
1096 } else if (st->codec->color_table_id) {
1097 const uint8_t *color_table;
1098 /* if flag bit 3 is set, use the default palette */
1099 color_count = 1 << color_depth;
1100 if (color_depth == 2)
1101 color_table = ff_qt_default_palette_4;
1102 else if (color_depth == 4)
1103 color_table = ff_qt_default_palette_16;
1105 color_table = ff_qt_default_palette_256;
1107 for (j = 0; j < color_count; j++) {
1108 r = color_table[j * 3 + 0];
1109 g = color_table[j * 3 + 1];
1110 b = color_table[j * 3 + 2];
1112 (r << 16) | (g << 8) | (b);
1115 /* load the palette from the file */
1116 color_start = avio_rb32(pb);
1117 color_count = avio_rb16(pb);
1118 color_end = avio_rb16(pb);
1119 if ((color_start <= 255) &&
1120 (color_end <= 255)) {
1121 for (j = color_start; j <= color_end; j++) {
1122 /* each R, G, or B component is 16 bits;
1123 * only use the top 8 bits; skip alpha bytes
1134 (r << 16) | (g << 8) | (b);
1138 sc->has_palette = 1;
1140 } else if (st->codec->codec_type==AVMEDIA_TYPE_AUDIO) {
1141 int bits_per_sample, flags;
1142 uint16_t version = avio_rb16(pb);
1144 st->codec->codec_id = id;
1145 avio_rb16(pb); /* revision level */
1146 avio_rb32(pb); /* vendor */
1148 st->codec->channels = avio_rb16(pb); /* channel count */
1149 av_dlog(c->fc, "audio channels %d\n", st->codec->channels);
1150 st->codec->bits_per_coded_sample = avio_rb16(pb); /* sample size */
1152 sc->audio_cid = avio_rb16(pb);
1153 avio_rb16(pb); /* packet size = 0 */
1155 st->codec->sample_rate = ((avio_rb32(pb) >> 16));
1157 //Read QT version 1 fields. In version 0 these do not exist.
1158 av_dlog(c->fc, "version =%d, isom =%d\n",version,c->isom);
1161 sc->samples_per_frame = avio_rb32(pb);
1162 avio_rb32(pb); /* bytes per packet */
1163 sc->bytes_per_frame = avio_rb32(pb);
1164 avio_rb32(pb); /* bytes per sample */
1165 } else if (version==2) {
1166 avio_rb32(pb); /* sizeof struct only */
1167 st->codec->sample_rate = av_int2dbl(avio_rb64(pb)); /* float 64 */
1168 st->codec->channels = avio_rb32(pb);
1169 avio_rb32(pb); /* always 0x7F000000 */
1170 st->codec->bits_per_coded_sample = avio_rb32(pb); /* bits per channel if sound is uncompressed */
1171 flags = avio_rb32(pb); /* lpcm format specific flag */
1172 sc->bytes_per_frame = avio_rb32(pb); /* bytes per audio packet if constant */
1173 sc->samples_per_frame = avio_rb32(pb); /* lpcm frames per audio packet if constant */
1174 if (format == MKTAG('l','p','c','m'))
1175 st->codec->codec_id = ff_mov_get_lpcm_codec_id(st->codec->bits_per_coded_sample, flags);
1179 switch (st->codec->codec_id) {
1180 case CODEC_ID_PCM_S8:
1181 case CODEC_ID_PCM_U8:
1182 if (st->codec->bits_per_coded_sample == 16)
1183 st->codec->codec_id = CODEC_ID_PCM_S16BE;
1185 case CODEC_ID_PCM_S16LE:
1186 case CODEC_ID_PCM_S16BE:
1187 if (st->codec->bits_per_coded_sample == 8)
1188 st->codec->codec_id = CODEC_ID_PCM_S8;
1189 else if (st->codec->bits_per_coded_sample == 24)
1190 st->codec->codec_id =
1191 st->codec->codec_id == CODEC_ID_PCM_S16BE ?
1192 CODEC_ID_PCM_S24BE : CODEC_ID_PCM_S24LE;
1194 /* set values for old format before stsd version 1 appeared */
1195 case CODEC_ID_MACE3:
1196 sc->samples_per_frame = 6;
1197 sc->bytes_per_frame = 2*st->codec->channels;
1199 case CODEC_ID_MACE6:
1200 sc->samples_per_frame = 6;
1201 sc->bytes_per_frame = 1*st->codec->channels;
1203 case CODEC_ID_ADPCM_IMA_QT:
1204 sc->samples_per_frame = 64;
1205 sc->bytes_per_frame = 34*st->codec->channels;
1208 sc->samples_per_frame = 160;
1209 sc->bytes_per_frame = 33;
1215 bits_per_sample = av_get_bits_per_sample(st->codec->codec_id);
1216 if (bits_per_sample) {
1217 st->codec->bits_per_coded_sample = bits_per_sample;
1218 sc->sample_size = (bits_per_sample >> 3) * st->codec->channels;
1220 } else if (st->codec->codec_type==AVMEDIA_TYPE_SUBTITLE){
1221 // ttxt stsd contains display flags, justification, background
1222 // color, fonts, and default styles, so fake an atom to read it
1223 MOVAtom fake_atom = { .size = size - (avio_tell(pb) - start_pos) };
1224 if (format != AV_RL32("mp4s")) // mp4s contains a regular esds atom
1225 mov_read_glbl(c, pb, fake_atom);
1226 st->codec->codec_id= id;
1227 st->codec->width = sc->width;
1228 st->codec->height = sc->height;
1230 if (st->codec->codec_tag == MKTAG('t','m','c','d')) {
1232 avio_rb32(pb); /* reserved */
1233 val = avio_rb32(pb); /* flags */
1235 st->codec->flags2 |= CODEC_FLAG2_DROP_FRAME_TIMECODE;
1238 st->codec->time_base.den = get_byte(pb);
1239 st->codec->time_base.num = 1;
1241 /* other codec type, just skip (rtp, mp4s, ...) */
1242 avio_skip(pb, size - (avio_tell(pb) - start_pos));
1244 /* this will read extra atoms at the end (wave, alac, damr, avcC, SMI ...) */
1245 a.size = size - (avio_tell(pb) - start_pos);
1247 if (mov_read_default(c, pb, a) < 0)
1249 } else if (a.size > 0)
1250 avio_skip(pb, a.size);
1253 if (st->codec->codec_type==AVMEDIA_TYPE_AUDIO && st->codec->sample_rate==0 && sc->time_scale>1)
1254 st->codec->sample_rate= sc->time_scale;
1256 /* special codec parameters handling */
1257 switch (st->codec->codec_id) {
1258 #if CONFIG_DV_DEMUXER
1259 case CODEC_ID_DVAUDIO:
1260 c->dv_fctx = avformat_alloc_context();
1261 c->dv_demux = dv_init_demux(c->dv_fctx);
1263 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
1266 sc->dv_audio_container = 1;
1267 st->codec->codec_id = CODEC_ID_PCM_S16LE;
1270 /* no ifdef since parameters are always those */
1271 case CODEC_ID_QCELP:
1272 // force sample rate for qcelp when not stored in mov
1273 if (st->codec->codec_tag != MKTAG('Q','c','l','p'))
1274 st->codec->sample_rate = 8000;
1275 st->codec->frame_size= 160;
1276 st->codec->channels= 1; /* really needed */
1278 case CODEC_ID_AMR_NB:
1279 case CODEC_ID_AMR_WB:
1280 st->codec->frame_size= sc->samples_per_frame;
1281 st->codec->channels= 1; /* really needed */
1282 /* force sample rate for amr, stsd in 3gp does not store sample rate */
1283 if (st->codec->codec_id == CODEC_ID_AMR_NB)
1284 st->codec->sample_rate = 8000;
1285 else if (st->codec->codec_id == CODEC_ID_AMR_WB)
1286 st->codec->sample_rate = 16000;
1290 st->codec->codec_type = AVMEDIA_TYPE_AUDIO; /* force type after stsd for m1a hdlr */
1291 st->need_parsing = AVSTREAM_PARSE_FULL;
1294 case CODEC_ID_ADPCM_MS:
1295 case CODEC_ID_ADPCM_IMA_WAV:
1296 st->codec->frame_size = sc->samples_per_frame;
1297 st->codec->block_align = sc->bytes_per_frame;
1300 if (st->codec->extradata_size == 36) {
1301 st->codec->frame_size = AV_RB32(st->codec->extradata+12);
1302 st->codec->channels = AV_RB8 (st->codec->extradata+21);
1303 st->codec->sample_rate = AV_RB32(st->codec->extradata+32);
1307 st->need_parsing = AVSTREAM_PARSE_FULL;
1316 static int mov_read_stsd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1320 avio_r8(pb); /* version */
1321 avio_rb24(pb); /* flags */
1322 entries = avio_rb32(pb);
1324 return ff_mov_read_stsd_entries(c, pb, entries);
1327 static int mov_read_stsc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1330 MOVStreamContext *sc;
1331 unsigned int i, entries;
1333 if (c->fc->nb_streams < 1)
1335 st = c->fc->streams[c->fc->nb_streams-1];
1338 avio_r8(pb); /* version */
1339 avio_rb24(pb); /* flags */
1341 entries = avio_rb32(pb);
1343 av_dlog(c->fc, "track[%i].stsc.entries = %i\n", c->fc->nb_streams-1, entries);
1345 if (entries >= UINT_MAX / sizeof(*sc->stsc_data))
1347 sc->stsc_data = av_malloc(entries * sizeof(*sc->stsc_data));
1349 return AVERROR(ENOMEM);
1350 sc->stsc_count = entries;
1352 for (i=0; i<entries; i++) {
1353 sc->stsc_data[i].first = avio_rb32(pb);
1354 sc->stsc_data[i].count = avio_rb32(pb);
1355 sc->stsc_data[i].id = avio_rb32(pb);
1360 static int mov_read_stps(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1363 MOVStreamContext *sc;
1364 unsigned i, entries;
1366 if (c->fc->nb_streams < 1)
1368 st = c->fc->streams[c->fc->nb_streams-1];
1371 avio_rb32(pb); // version + flags
1373 entries = avio_rb32(pb);
1374 if (entries >= UINT_MAX / sizeof(*sc->stps_data))
1376 sc->stps_data = av_malloc(entries * sizeof(*sc->stps_data));
1378 return AVERROR(ENOMEM);
1379 sc->stps_count = entries;
1381 for (i = 0; i < entries; i++) {
1382 sc->stps_data[i] = avio_rb32(pb);
1383 //av_dlog(c->fc, "stps %d\n", sc->stps_data[i]);
1389 static int mov_read_stss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1392 MOVStreamContext *sc;
1393 unsigned int i, entries;
1395 if (c->fc->nb_streams < 1)
1397 st = c->fc->streams[c->fc->nb_streams-1];
1400 avio_r8(pb); /* version */
1401 avio_rb24(pb); /* flags */
1403 entries = avio_rb32(pb);
1405 av_dlog(c->fc, "keyframe_count = %d\n", entries);
1407 if (entries >= UINT_MAX / sizeof(int))
1409 sc->keyframes = av_malloc(entries * sizeof(int));
1411 return AVERROR(ENOMEM);
1412 sc->keyframe_count = entries;
1414 for (i=0; i<entries; i++) {
1415 sc->keyframes[i] = avio_rb32(pb);
1416 //av_dlog(c->fc, "keyframes[]=%d\n", sc->keyframes[i]);
1421 static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1424 MOVStreamContext *sc;
1425 unsigned int i, entries, sample_size, field_size, num_bytes;
1429 if (c->fc->nb_streams < 1)
1431 st = c->fc->streams[c->fc->nb_streams-1];
1434 avio_r8(pb); /* version */
1435 avio_rb24(pb); /* flags */
1437 if (atom.type == MKTAG('s','t','s','z')) {
1438 sample_size = avio_rb32(pb);
1439 if (!sc->sample_size) /* do not overwrite value computed in stsd */
1440 sc->sample_size = sample_size;
1444 avio_rb24(pb); /* reserved */
1445 field_size = avio_r8(pb);
1447 entries = avio_rb32(pb);
1449 av_dlog(c->fc, "sample_size = %d sample_count = %d\n", sc->sample_size, entries);
1451 sc->sample_count = entries;
1455 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
1456 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %d\n", field_size);
1460 if (entries >= UINT_MAX / sizeof(int) || entries >= (UINT_MAX - 4) / field_size)
1462 sc->sample_sizes = av_malloc(entries * sizeof(int));
1463 if (!sc->sample_sizes)
1464 return AVERROR(ENOMEM);
1466 num_bytes = (entries*field_size+4)>>3;
1468 buf = av_malloc(num_bytes+FF_INPUT_BUFFER_PADDING_SIZE);
1470 av_freep(&sc->sample_sizes);
1471 return AVERROR(ENOMEM);
1474 if (avio_read(pb, buf, num_bytes) < num_bytes) {
1475 av_freep(&sc->sample_sizes);
1480 init_get_bits(&gb, buf, 8*num_bytes);
1482 for (i=0; i<entries; i++)
1483 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
1489 static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1492 MOVStreamContext *sc;
1493 unsigned int i, entries;
1495 int64_t total_sample_count=0;
1497 if (c->fc->nb_streams < 1)
1499 st = c->fc->streams[c->fc->nb_streams-1];
1502 avio_r8(pb); /* version */
1503 avio_rb24(pb); /* flags */
1504 entries = avio_rb32(pb);
1506 av_dlog(c->fc, "track[%i].stts.entries = %i\n",
1507 c->fc->nb_streams-1, entries);
1509 if (!entries || entries >= UINT_MAX / sizeof(*sc->stts_data))
1510 return AVERROR(EINVAL);
1512 sc->stts_data = av_malloc(entries * sizeof(*sc->stts_data));
1514 return AVERROR(ENOMEM);
1516 sc->stts_count = entries;
1518 for (i=0; i<entries; i++) {
1519 int sample_duration;
1522 sample_count=avio_rb32(pb);
1523 sample_duration = avio_rb32(pb);
1524 /* sample_duration < 0 is invalid based on the spec */
1525 if (sample_duration < 0) {
1526 av_log(c->fc, AV_LOG_ERROR, "Invalid SampleDelta in STTS %d", sample_duration);
1527 sample_duration = 1;
1529 sc->stts_data[i].count= sample_count;
1530 sc->stts_data[i].duration= sample_duration;
1532 av_dlog(c->fc, "sample_count=%d, sample_duration=%d\n",
1533 sample_count, sample_duration);
1535 duration+=(int64_t)sample_duration*sample_count;
1536 total_sample_count+=sample_count;
1539 st->nb_frames= total_sample_count;
1541 st->duration= duration;
1545 static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1548 MOVStreamContext *sc;
1549 unsigned int i, entries;
1551 if (c->fc->nb_streams < 1)
1553 st = c->fc->streams[c->fc->nb_streams-1];
1556 avio_r8(pb); /* version */
1557 avio_rb24(pb); /* flags */
1558 entries = avio_rb32(pb);
1560 av_dlog(c->fc, "track[%i].ctts.entries = %i\n", c->fc->nb_streams-1, entries);
1562 if (entries >= UINT_MAX / sizeof(*sc->ctts_data))
1564 sc->ctts_data = av_malloc(entries * sizeof(*sc->ctts_data));
1566 return AVERROR(ENOMEM);
1567 sc->ctts_count = entries;
1569 for (i=0; i<entries; i++) {
1570 int count =avio_rb32(pb);
1571 int duration =avio_rb32(pb);
1573 sc->ctts_data[i].count = count;
1574 sc->ctts_data[i].duration= duration;
1575 if (duration < 0 && i+1<entries)
1576 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
1579 av_dlog(c->fc, "dts shift %d\n", sc->dts_shift);
1584 static void mov_build_index(MOVContext *mov, AVStream *st)
1586 MOVStreamContext *sc = st->priv_data;
1587 int64_t current_offset;
1588 int64_t current_dts = 0;
1589 unsigned int stts_index = 0;
1590 unsigned int stsc_index = 0;
1591 unsigned int stss_index = 0;
1592 unsigned int stps_index = 0;
1594 uint64_t stream_size = 0;
1596 /* adjust first dts according to edit list */
1597 if (sc->time_offset && mov->time_scale > 0) {
1598 if (sc->time_offset < 0)
1599 sc->time_offset = av_rescale(sc->time_offset, sc->time_scale, mov->time_scale);
1600 current_dts = -sc->time_offset;
1601 if (sc->ctts_data && sc->stts_data &&
1602 sc->ctts_data[0].duration / FFMAX(sc->stts_data[0].duration, 1) > 16) {
1603 /* more than 16 frames delay, dts are likely wrong
1604 this happens with files created by iMovie */
1606 st->codec->has_b_frames = 1;
1610 /* only use old uncompressed audio chunk demuxing when stts specifies it */
1611 if (!(st->codec->codec_type == AVMEDIA_TYPE_AUDIO &&
1612 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
1613 unsigned int current_sample = 0;
1614 unsigned int stts_sample = 0;
1615 unsigned int sample_size;
1616 unsigned int distance = 0;
1617 int key_off = sc->keyframes && sc->keyframes[0] == 1;
1619 current_dts -= sc->dts_shift;
1621 if (sc->sample_count >= UINT_MAX / sizeof(*st->index_entries))
1623 st->index_entries = av_malloc(sc->sample_count*sizeof(*st->index_entries));
1624 if (!st->index_entries)
1626 st->index_entries_allocated_size = sc->sample_count*sizeof(*st->index_entries);
1628 for (i = 0; i < sc->chunk_count; i++) {
1629 current_offset = sc->chunk_offsets[i];
1630 while (stsc_index + 1 < sc->stsc_count &&
1631 i + 1 == sc->stsc_data[stsc_index + 1].first)
1633 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
1635 if (current_sample >= sc->sample_count) {
1636 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
1640 if (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index]) {
1642 if (stss_index + 1 < sc->keyframe_count)
1644 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
1646 if (stps_index + 1 < sc->stps_count)
1651 sample_size = sc->sample_size > 0 ? sc->sample_size : sc->sample_sizes[current_sample];
1652 if (sc->pseudo_stream_id == -1 ||
1653 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
1654 AVIndexEntry *e = &st->index_entries[st->nb_index_entries++];
1655 e->pos = current_offset;
1656 e->timestamp = current_dts;
1657 e->size = sample_size;
1658 e->min_distance = distance;
1659 e->flags = keyframe ? AVINDEX_KEYFRAME : 0;
1660 av_dlog(mov->fc, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
1661 "size %d, distance %d, keyframe %d\n", st->index, current_sample,
1662 current_offset, current_dts, sample_size, distance, keyframe);
1665 current_offset += sample_size;
1666 stream_size += sample_size;
1667 current_dts += sc->stts_data[stts_index].duration;
1671 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
1677 if (st->duration > 0)
1678 st->codec->bit_rate = stream_size*8*sc->time_scale/st->duration;
1680 unsigned chunk_samples, total = 0;
1682 // compute total chunk count
1683 for (i = 0; i < sc->stsc_count; i++) {
1684 unsigned count, chunk_count;
1686 chunk_samples = sc->stsc_data[i].count;
1687 if (sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
1688 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
1692 if (sc->samples_per_frame >= 160) { // gsm
1693 count = chunk_samples / sc->samples_per_frame;
1694 } else if (sc->samples_per_frame > 1) {
1695 unsigned samples = (1024/sc->samples_per_frame)*sc->samples_per_frame;
1696 count = (chunk_samples+samples-1) / samples;
1698 count = (chunk_samples+1023) / 1024;
1701 if (i < sc->stsc_count - 1)
1702 chunk_count = sc->stsc_data[i+1].first - sc->stsc_data[i].first;
1704 chunk_count = sc->chunk_count - (sc->stsc_data[i].first - 1);
1705 total += chunk_count * count;
1708 av_dlog(mov->fc, "chunk count %d\n", total);
1709 if (total >= UINT_MAX / sizeof(*st->index_entries))
1711 st->index_entries = av_malloc(total*sizeof(*st->index_entries));
1712 if (!st->index_entries)
1714 st->index_entries_allocated_size = total*sizeof(*st->index_entries);
1717 for (i = 0; i < sc->chunk_count; i++) {
1718 current_offset = sc->chunk_offsets[i];
1719 if (stsc_index + 1 < sc->stsc_count &&
1720 i + 1 == sc->stsc_data[stsc_index + 1].first)
1722 chunk_samples = sc->stsc_data[stsc_index].count;
1724 while (chunk_samples > 0) {
1726 unsigned size, samples;
1728 if (sc->samples_per_frame >= 160) { // gsm
1729 samples = sc->samples_per_frame;
1730 size = sc->bytes_per_frame;
1732 if (sc->samples_per_frame > 1) {
1733 samples = FFMIN((1024 / sc->samples_per_frame)*
1734 sc->samples_per_frame, chunk_samples);
1735 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
1737 samples = FFMIN(1024, chunk_samples);
1738 size = samples * sc->sample_size;
1742 if (st->nb_index_entries >= total) {
1743 av_log(mov->fc, AV_LOG_ERROR, "wrong chunk count %d\n", total);
1746 e = &st->index_entries[st->nb_index_entries++];
1747 e->pos = current_offset;
1748 e->timestamp = current_dts;
1750 e->min_distance = 0;
1751 e->flags = AVINDEX_KEYFRAME;
1752 av_dlog(mov->fc, "AVIndex stream %d, chunk %d, offset %"PRIx64", dts %"PRId64", "
1753 "size %d, duration %d\n", st->index, i, current_offset, current_dts,
1756 current_offset += size;
1757 current_dts += samples;
1758 chunk_samples -= samples;
1764 static int mov_open_dref(AVIOContext **pb, const char *src, MOVDref *ref)
1766 /* try relative path, we do not try the absolute because it can leak information about our
1767 system to an attacker */
1768 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
1769 char filename[1024];
1770 const char *src_path;
1773 /* find a source dir */
1774 src_path = strrchr(src, '/');
1780 /* find a next level down to target */
1781 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
1782 if (ref->path[l] == '/') {
1783 if (i == ref->nlvl_to - 1)
1789 /* compose filename if next level down to target was found */
1790 if (i == ref->nlvl_to - 1 && src_path - src < sizeof(filename)) {
1791 memcpy(filename, src, src_path - src);
1792 filename[src_path - src] = 0;
1794 for (i = 1; i < ref->nlvl_from; i++)
1795 av_strlcat(filename, "../", 1024);
1797 av_strlcat(filename, ref->path + l + 1, 1024);
1799 if (!avio_open(pb, filename, AVIO_FLAG_READ))
1804 return AVERROR(ENOENT);
1807 static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1810 MOVStreamContext *sc;
1813 st = av_new_stream(c->fc, c->fc->nb_streams);
1814 if (!st) return AVERROR(ENOMEM);
1815 sc = av_mallocz(sizeof(MOVStreamContext));
1816 if (!sc) return AVERROR(ENOMEM);
1819 st->codec->codec_type = AVMEDIA_TYPE_DATA;
1820 sc->ffindex = st->index;
1822 if ((ret = mov_read_default(c, pb, atom)) < 0)
1826 if (sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
1827 (!sc->sample_size && !sc->sample_count))) {
1828 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
1833 if (sc->time_scale <= 0) {
1834 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", st->index);
1835 sc->time_scale = c->time_scale;
1836 if (sc->time_scale <= 0)
1840 av_set_pts_info(st, 64, 1, sc->time_scale);
1842 if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO &&
1843 !st->codec->frame_size && sc->stts_count == 1) {
1844 st->codec->frame_size = av_rescale(sc->stts_data[0].duration,
1845 st->codec->sample_rate, sc->time_scale);
1846 av_dlog(c->fc, "frame size %d\n", st->codec->frame_size);
1849 mov_build_index(c, st);
1851 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
1852 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
1853 if (mov_open_dref(&sc->pb, c->fc->filename, dref) < 0)
1854 av_log(c->fc, AV_LOG_ERROR,
1855 "stream %d, error opening alias: path='%s', dir='%s', "
1856 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
1857 st->index, dref->path, dref->dir, dref->filename,
1858 dref->volume, dref->nlvl_from, dref->nlvl_to);
1862 if (st->codec->codec_type == AVMEDIA_TYPE_VIDEO) {
1863 if (!st->sample_aspect_ratio.num &&
1864 (st->codec->width != sc->width || st->codec->height != sc->height)) {
1865 st->sample_aspect_ratio = av_d2q(((double)st->codec->height * sc->width) /
1866 ((double)st->codec->width * sc->height), INT_MAX);
1869 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
1870 sc->time_scale*st->nb_frames, st->duration, INT_MAX);
1872 if (sc->stts_count == 1 || (sc->stts_count == 2 && sc->stts_data[1].count == 1))
1873 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
1874 sc->time_scale, sc->stts_data[0].duration, INT_MAX);
1877 switch (st->codec->codec_id) {
1878 #if CONFIG_H261_DECODER
1881 #if CONFIG_H263_DECODER
1884 #if CONFIG_H264_DECODER
1887 #if CONFIG_MPEG4_DECODER
1888 case CODEC_ID_MPEG4:
1890 st->codec->width = 0; /* let decoder init width/height */
1891 st->codec->height= 0;
1895 /* Do not need those anymore. */
1896 av_freep(&sc->chunk_offsets);
1897 av_freep(&sc->stsc_data);
1898 av_freep(&sc->sample_sizes);
1899 av_freep(&sc->keyframes);
1900 av_freep(&sc->stts_data);
1901 av_freep(&sc->stps_data);
1906 static int mov_read_ilst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1909 c->itunes_metadata = 1;
1910 ret = mov_read_default(c, pb, atom);
1911 c->itunes_metadata = 0;
1915 static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1917 while (atom.size > 8) {
1918 uint32_t tag = avio_rl32(pb);
1920 if (tag == MKTAG('h','d','l','r')) {
1921 avio_seek(pb, -8, SEEK_CUR);
1923 return mov_read_default(c, pb, atom);
1929 static int mov_read_tkhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1934 int64_t disp_transform[2];
1935 int display_matrix[3][2];
1937 MOVStreamContext *sc;
1940 if (c->fc->nb_streams < 1)
1942 st = c->fc->streams[c->fc->nb_streams-1];
1945 version = avio_r8(pb);
1946 avio_rb24(pb); /* flags */
1948 MOV_TRACK_ENABLED 0x0001
1949 MOV_TRACK_IN_MOVIE 0x0002
1950 MOV_TRACK_IN_PREVIEW 0x0004
1951 MOV_TRACK_IN_POSTER 0x0008
1958 avio_rb32(pb); /* creation time */
1959 avio_rb32(pb); /* modification time */
1961 st->id = (int)avio_rb32(pb); /* track id (NOT 0 !)*/
1962 avio_rb32(pb); /* reserved */
1964 /* highlevel (considering edits) duration in movie timebase */
1965 (version == 1) ? avio_rb64(pb) : avio_rb32(pb);
1966 avio_rb32(pb); /* reserved */
1967 avio_rb32(pb); /* reserved */
1969 avio_rb16(pb); /* layer */
1970 avio_rb16(pb); /* alternate group */
1971 avio_rb16(pb); /* volume */
1972 avio_rb16(pb); /* reserved */
1974 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
1975 // they're kept in fixed point format through all calculations
1976 // ignore u,v,z b/c we don't need the scale factor to calc aspect ratio
1977 for (i = 0; i < 3; i++) {
1978 display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
1979 display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
1980 avio_rb32(pb); // 2.30 fixed point (not used)
1983 width = avio_rb32(pb); // 16.16 fixed point track width
1984 height = avio_rb32(pb); // 16.16 fixed point track height
1985 sc->width = width >> 16;
1986 sc->height = height >> 16;
1988 if (display_matrix[0][0] == -65536 && display_matrix[1][1] == -65536) {
1989 av_dict_set(&st->metadata, "rotate", "180", 0);
1992 // transform the display width/height according to the matrix
1993 // skip this if the display matrix is the default identity matrix
1994 // or if it is rotating the picture, ex iPhone 3GS
1995 // to keep the same scale, use [width height 1<<16]
1996 if (width && height &&
1997 ((display_matrix[0][0] != 65536 ||
1998 display_matrix[1][1] != 65536) &&
1999 !display_matrix[0][1] &&
2000 !display_matrix[1][0] &&
2001 !display_matrix[2][0] && !display_matrix[2][1])) {
2002 for (i = 0; i < 2; i++)
2004 (int64_t) width * display_matrix[0][i] +
2005 (int64_t) height * display_matrix[1][i] +
2006 ((int64_t) display_matrix[2][i] << 16);
2008 //sample aspect ratio is new width/height divided by old width/height
2009 st->sample_aspect_ratio = av_d2q(
2010 ((double) disp_transform[0] * height) /
2011 ((double) disp_transform[1] * width), INT_MAX);
2016 static int mov_read_tfhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2018 MOVFragment *frag = &c->fragment;
2019 MOVTrackExt *trex = NULL;
2020 int flags, track_id, i;
2022 avio_r8(pb); /* version */
2023 flags = avio_rb24(pb);
2025 track_id = avio_rb32(pb);
2028 frag->track_id = track_id;
2029 for (i = 0; i < c->trex_count; i++)
2030 if (c->trex_data[i].track_id == frag->track_id) {
2031 trex = &c->trex_data[i];
2035 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
2039 if (flags & 0x01) frag->base_data_offset = avio_rb64(pb);
2040 else frag->base_data_offset = frag->moof_offset;
2041 if (flags & 0x02) frag->stsd_id = avio_rb32(pb);
2042 else frag->stsd_id = trex->stsd_id;
2044 frag->duration = flags & 0x08 ? avio_rb32(pb) : trex->duration;
2045 frag->size = flags & 0x10 ? avio_rb32(pb) : trex->size;
2046 frag->flags = flags & 0x20 ? avio_rb32(pb) : trex->flags;
2047 av_dlog(c->fc, "frag flags 0x%x\n", frag->flags);
2051 static int mov_read_chap(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2053 c->chapter_track = avio_rb32(pb);
2057 static int mov_read_trex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2061 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
2063 trex = av_realloc(c->trex_data, (c->trex_count+1)*sizeof(*c->trex_data));
2065 return AVERROR(ENOMEM);
2066 c->trex_data = trex;
2067 trex = &c->trex_data[c->trex_count++];
2068 avio_r8(pb); /* version */
2069 avio_rb24(pb); /* flags */
2070 trex->track_id = avio_rb32(pb);
2071 trex->stsd_id = avio_rb32(pb);
2072 trex->duration = avio_rb32(pb);
2073 trex->size = avio_rb32(pb);
2074 trex->flags = avio_rb32(pb);
2078 static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2080 MOVFragment *frag = &c->fragment;
2081 AVStream *st = NULL;
2082 MOVStreamContext *sc;
2086 int data_offset = 0;
2087 unsigned entries, first_sample_flags = frag->flags;
2088 int flags, distance, i;
2090 for (i = 0; i < c->fc->nb_streams; i++) {
2091 if (c->fc->streams[i]->id == frag->track_id) {
2092 st = c->fc->streams[i];
2097 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
2101 if (sc->pseudo_stream_id+1 != frag->stsd_id)
2103 avio_r8(pb); /* version */
2104 flags = avio_rb24(pb);
2105 entries = avio_rb32(pb);
2106 av_dlog(c->fc, "flags 0x%x entries %d\n", flags, entries);
2108 /* Always assume the presence of composition time offsets.
2109 * Without this assumption, for instance, we cannot deal with a track in fragmented movies that meet the following.
2110 * 1) in the initial movie, there are no samples.
2111 * 2) in the first movie fragment, there is only one sample without composition time offset.
2112 * 3) in the subsequent movie fragments, there are samples with composition time offset. */
2113 if (!sc->ctts_count && sc->sample_count)
2115 /* Complement ctts table if moov atom doesn't have ctts atom. */
2116 ctts_data = av_malloc(sizeof(*sc->ctts_data));
2118 return AVERROR(ENOMEM);
2119 sc->ctts_data = ctts_data;
2120 sc->ctts_data[sc->ctts_count].count = sc->sample_count;
2121 sc->ctts_data[sc->ctts_count].duration = 0;
2124 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
2126 ctts_data = av_realloc(sc->ctts_data,
2127 (entries+sc->ctts_count)*sizeof(*sc->ctts_data));
2129 return AVERROR(ENOMEM);
2130 sc->ctts_data = ctts_data;
2132 if (flags & 0x001) data_offset = avio_rb32(pb);
2133 if (flags & 0x004) first_sample_flags = avio_rb32(pb);
2134 dts = st->duration - sc->time_offset;
2135 offset = frag->base_data_offset + data_offset;
2137 av_dlog(c->fc, "first sample flags 0x%x\n", first_sample_flags);
2138 for (i = 0; i < entries; i++) {
2139 unsigned sample_size = frag->size;
2140 int sample_flags = i ? frag->flags : first_sample_flags;
2141 unsigned sample_duration = frag->duration;
2144 if (flags & 0x100) sample_duration = avio_rb32(pb);
2145 if (flags & 0x200) sample_size = avio_rb32(pb);
2146 if (flags & 0x400) sample_flags = avio_rb32(pb);
2147 sc->ctts_data[sc->ctts_count].count = 1;
2148 sc->ctts_data[sc->ctts_count].duration = (flags & 0x800) ? avio_rb32(pb) : 0;
2150 if ((keyframe = st->codec->codec_type == AVMEDIA_TYPE_AUDIO ||
2151 (flags & 0x004 && !i && !sample_flags) || sample_flags & 0x2000000))
2153 av_add_index_entry(st, offset, dts, sample_size, distance,
2154 keyframe ? AVINDEX_KEYFRAME : 0);
2155 av_dlog(c->fc, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
2156 "size %d, distance %d, keyframe %d\n", st->index, sc->sample_count+i,
2157 offset, dts, sample_size, distance, keyframe);
2159 dts += sample_duration;
2160 offset += sample_size;
2162 frag->moof_offset = offset;
2163 st->duration = dts + sc->time_offset;
2167 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
2168 /* like the files created with Adobe Premiere 5.0, for samples see */
2169 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
2170 static int mov_read_wide(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2175 return 0; /* continue */
2176 if (avio_rb32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
2177 avio_skip(pb, atom.size - 4);
2180 atom.type = avio_rl32(pb);
2182 if (atom.type != MKTAG('m','d','a','t')) {
2183 avio_skip(pb, atom.size);
2186 err = mov_read_mdat(c, pb, atom);
2190 static int mov_read_cmov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2195 uint8_t *moov_data; /* uncompressed data */
2196 long cmov_len, moov_len;
2199 avio_rb32(pb); /* dcom atom */
2200 if (avio_rl32(pb) != MKTAG('d','c','o','m'))
2202 if (avio_rl32(pb) != MKTAG('z','l','i','b')) {
2203 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !");
2206 avio_rb32(pb); /* cmvd atom */
2207 if (avio_rl32(pb) != MKTAG('c','m','v','d'))
2209 moov_len = avio_rb32(pb); /* uncompressed size */
2210 cmov_len = atom.size - 6 * 4;
2212 cmov_data = av_malloc(cmov_len);
2214 return AVERROR(ENOMEM);
2215 moov_data = av_malloc(moov_len);
2218 return AVERROR(ENOMEM);
2220 avio_read(pb, cmov_data, cmov_len);
2221 if (uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
2222 goto free_and_return;
2223 if (ffio_init_context(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
2224 goto free_and_return;
2225 atom.type = MKTAG('m','o','o','v');
2226 atom.size = moov_len;
2227 ret = mov_read_default(c, &ctx, atom);
2233 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
2238 /* edit list atom */
2239 static int mov_read_elst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2241 MOVStreamContext *sc;
2242 int i, edit_count, version;
2244 if (c->fc->nb_streams < 1)
2246 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
2248 version = avio_r8(pb); /* version */
2249 avio_rb24(pb); /* flags */
2250 edit_count = avio_rb32(pb); /* entries */
2252 if ((uint64_t)edit_count*12+8 > atom.size)
2255 for (i=0; i<edit_count; i++){
2259 duration = avio_rb64(pb);
2260 time = avio_rb64(pb);
2262 duration = avio_rb32(pb); /* segment duration */
2263 time = (int32_t)avio_rb32(pb); /* media time */
2265 avio_rb32(pb); /* Media rate */
2266 if (i == 0 && time >= -1) {
2267 sc->time_offset = time != -1 ? time : -duration;
2272 av_log(c->fc, AV_LOG_WARNING, "multiple edit list entries, "
2273 "a/v desync might occur, patch welcome\n");
2275 av_dlog(c->fc, "track[%i].edit_count = %i\n", c->fc->nb_streams-1, edit_count);
2279 static int mov_read_chan(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2282 return AVERROR_INVALIDDATA;
2284 ff_mov_read_chan(c->fc, atom.size - 4, c->fc->streams[0]->codec);
2288 static const MOVParseTableEntry mov_default_parse_table[] = {
2289 { MKTAG('a','v','s','s'), mov_read_avss },
2290 { MKTAG('c','h','p','l'), mov_read_chpl },
2291 { MKTAG('c','o','6','4'), mov_read_stco },
2292 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
2293 { MKTAG('d','i','n','f'), mov_read_default },
2294 { MKTAG('d','r','e','f'), mov_read_dref },
2295 { MKTAG('e','d','t','s'), mov_read_default },
2296 { MKTAG('e','l','s','t'), mov_read_elst },
2297 { MKTAG('e','n','d','a'), mov_read_enda },
2298 { MKTAG('f','i','e','l'), mov_read_fiel },
2299 { MKTAG('f','t','y','p'), mov_read_ftyp },
2300 { MKTAG('g','l','b','l'), mov_read_glbl },
2301 { MKTAG('h','d','l','r'), mov_read_hdlr },
2302 { MKTAG('i','l','s','t'), mov_read_ilst },
2303 { MKTAG('j','p','2','h'), mov_read_jp2h },
2304 { MKTAG('m','d','a','t'), mov_read_mdat },
2305 { MKTAG('m','d','h','d'), mov_read_mdhd },
2306 { MKTAG('m','d','i','a'), mov_read_default },
2307 { MKTAG('m','e','t','a'), mov_read_meta },
2308 { MKTAG('m','i','n','f'), mov_read_default },
2309 { MKTAG('m','o','o','f'), mov_read_moof },
2310 { MKTAG('m','o','o','v'), mov_read_moov },
2311 { MKTAG('m','v','e','x'), mov_read_default },
2312 { MKTAG('m','v','h','d'), mov_read_mvhd },
2313 { MKTAG('S','M','I',' '), mov_read_smi }, /* Sorenson extension ??? */
2314 { MKTAG('a','l','a','c'), mov_read_alac }, /* alac specific atom */
2315 { MKTAG('a','v','c','C'), mov_read_glbl },
2316 { MKTAG('p','a','s','p'), mov_read_pasp },
2317 { MKTAG('s','t','b','l'), mov_read_default },
2318 { MKTAG('s','t','c','o'), mov_read_stco },
2319 { MKTAG('s','t','p','s'), mov_read_stps },
2320 { MKTAG('s','t','r','f'), mov_read_strf },
2321 { MKTAG('s','t','s','c'), mov_read_stsc },
2322 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
2323 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
2324 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
2325 { MKTAG('s','t','t','s'), mov_read_stts },
2326 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
2327 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
2328 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
2329 { MKTAG('t','r','a','k'), mov_read_trak },
2330 { MKTAG('t','r','a','f'), mov_read_default },
2331 { MKTAG('t','r','e','f'), mov_read_default },
2332 { MKTAG('c','h','a','p'), mov_read_chap },
2333 { MKTAG('t','r','e','x'), mov_read_trex },
2334 { MKTAG('t','r','u','n'), mov_read_trun },
2335 { MKTAG('u','d','t','a'), mov_read_default },
2336 { MKTAG('w','a','v','e'), mov_read_wave },
2337 { MKTAG('e','s','d','s'), mov_read_esds },
2338 { MKTAG('d','a','c','3'), mov_read_dac3 }, /* AC-3 info */
2339 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
2340 { MKTAG('w','f','e','x'), mov_read_wfex },
2341 { MKTAG('c','m','o','v'), mov_read_cmov },
2342 { MKTAG('c','h','a','n'), mov_read_chan },
2346 static int mov_probe(AVProbeData *p)
2348 unsigned int offset;
2352 /* check file header */
2355 /* ignore invalid offset */
2356 if ((offset + 8) > (unsigned int)p->buf_size)
2358 tag = AV_RL32(p->buf + offset + 4);
2360 /* check for obvious tags */
2361 case MKTAG('j','P',' ',' '): /* jpeg 2000 signature */
2362 case MKTAG('m','o','o','v'):
2363 case MKTAG('m','d','a','t'):
2364 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
2365 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
2366 case MKTAG('f','t','y','p'):
2367 return AVPROBE_SCORE_MAX;
2368 /* those are more common words, so rate then a bit less */
2369 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
2370 case MKTAG('w','i','d','e'):
2371 case MKTAG('f','r','e','e'):
2372 case MKTAG('j','u','n','k'):
2373 case MKTAG('p','i','c','t'):
2374 return AVPROBE_SCORE_MAX - 5;
2375 case MKTAG(0x82,0x82,0x7f,0x7d):
2376 case MKTAG('s','k','i','p'):
2377 case MKTAG('u','u','i','d'):
2378 case MKTAG('p','r','f','l'):
2379 offset = AV_RB32(p->buf+offset) + offset;
2380 /* if we only find those cause probedata is too small at least rate them */
2381 score = AVPROBE_SCORE_MAX - 50;
2384 /* unrecognized tag */
2390 // must be done after parsing all trak because there's no order requirement
2391 static void mov_read_chapters(AVFormatContext *s)
2393 MOVContext *mov = s->priv_data;
2394 AVStream *st = NULL;
2395 MOVStreamContext *sc;
2399 for (i = 0; i < s->nb_streams; i++)
2400 if (s->streams[i]->id == mov->chapter_track) {
2405 av_log(s, AV_LOG_ERROR, "Referenced QT chapter track not found\n");
2409 st->discard = AVDISCARD_ALL;
2411 cur_pos = avio_tell(sc->pb);
2413 for (i = 0; i < st->nb_index_entries; i++) {
2414 AVIndexEntry *sample = &st->index_entries[i];
2415 int64_t end = i+1 < st->nb_index_entries ? st->index_entries[i+1].timestamp : st->duration;
2420 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
2421 av_log(s, AV_LOG_ERROR, "Chapter %d not found in file\n", i);
2425 // the first two bytes are the length of the title
2426 len = avio_rb16(sc->pb);
2427 if (len > sample->size-2)
2429 title_len = 2*len + 1;
2430 if (!(title = av_mallocz(title_len)))
2433 // The samples could theoretically be in any encoding if there's an encd
2434 // atom following, but in practice are only utf-8 or utf-16, distinguished
2435 // instead by the presence of a BOM
2439 ch = avio_rb16(sc->pb);
2441 avio_get_str16be(sc->pb, len, title, title_len);
2442 else if (ch == 0xfffe)
2443 avio_get_str16le(sc->pb, len, title, title_len);
2446 if (len == 1 || len == 2)
2449 get_strz(sc->pb, title + 2, len - 1);
2453 ff_new_chapter(s, i, st->time_base, sample->timestamp, end, title);
2457 avio_seek(sc->pb, cur_pos, SEEK_SET);
2460 static int mov_read_header(AVFormatContext *s, AVFormatParameters *ap)
2462 MOVContext *mov = s->priv_data;
2463 AVIOContext *pb = s->pb;
2465 MOVAtom atom = { AV_RL32("root") };
2468 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
2470 atom.size = avio_size(pb);
2472 atom.size = INT64_MAX;
2474 /* check MOV header */
2475 if ((err = mov_read_default(mov, pb, atom)) < 0) {
2476 av_log(s, AV_LOG_ERROR, "error reading header: %d\n", err);
2479 if (!mov->found_moov) {
2480 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
2483 av_dlog(mov->fc, "on_parse_exit_offset=%"PRId64"\n", avio_tell(pb));
2485 if (pb->seekable && mov->chapter_track > 0)
2486 mov_read_chapters(s);
2491 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
2493 AVIndexEntry *sample = NULL;
2494 int64_t best_dts = INT64_MAX;
2496 for (i = 0; i < s->nb_streams; i++) {
2497 AVStream *avst = s->streams[i];
2498 MOVStreamContext *msc = avst->priv_data;
2499 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
2500 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
2501 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
2502 av_dlog(s, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
2503 if (!sample || (!s->pb->seekable && current_sample->pos < sample->pos) ||
2505 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
2506 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
2507 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
2508 sample = current_sample;
2517 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
2519 MOVContext *mov = s->priv_data;
2520 MOVStreamContext *sc;
2521 AVIndexEntry *sample;
2522 AVStream *st = NULL;
2525 sample = mov_find_next_sample(s, &st);
2527 mov->found_mdat = 0;
2528 if (s->pb->seekable||
2529 mov_read_default(mov, s->pb, (MOVAtom){ AV_RL32("root"), INT64_MAX }) < 0 ||
2532 av_dlog(s, "read fragments, offset 0x%"PRIx64"\n", avio_tell(s->pb));
2536 /* must be done just before reading, to avoid infinite loop on sample */
2537 sc->current_sample++;
2539 if (st->discard != AVDISCARD_ALL) {
2540 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
2541 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
2542 sc->ffindex, sample->pos);
2545 ret = av_get_packet(sc->pb, pkt, sample->size);
2548 if (sc->has_palette) {
2551 pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, AVPALETTE_SIZE);
2553 av_log(mov->fc, AV_LOG_ERROR, "Cannot append palette to packet\n");
2555 memcpy(pal, sc->palette, AVPALETTE_SIZE);
2556 sc->has_palette = 0;
2559 #if CONFIG_DV_DEMUXER
2560 if (mov->dv_demux && sc->dv_audio_container) {
2561 dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size, pkt->pos);
2564 ret = dv_get_packet(mov->dv_demux, pkt);
2571 pkt->stream_index = sc->ffindex;
2572 pkt->dts = sample->timestamp;
2573 if (sc->ctts_data) {
2574 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
2575 /* update ctts context */
2577 if (sc->ctts_index < sc->ctts_count &&
2578 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
2580 sc->ctts_sample = 0;
2583 pkt->dts = AV_NOPTS_VALUE;
2585 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
2586 st->index_entries[sc->current_sample].timestamp : st->duration;
2587 pkt->duration = next_dts - pkt->dts;
2588 pkt->pts = pkt->dts;
2590 if (st->discard == AVDISCARD_ALL)
2592 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? AV_PKT_FLAG_KEY : 0;
2593 pkt->pos = sample->pos;
2594 av_dlog(s, "stream %d, pts %"PRId64", dts %"PRId64", pos 0x%"PRIx64", duration %d\n",
2595 pkt->stream_index, pkt->pts, pkt->dts, pkt->pos, pkt->duration);
2599 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
2601 MOVStreamContext *sc = st->priv_data;
2602 int sample, time_sample;
2605 sample = av_index_search_timestamp(st, timestamp, flags);
2606 av_dlog(s, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
2607 if (sample < 0 && st->nb_index_entries && timestamp < st->index_entries[0].timestamp)
2609 if (sample < 0) /* not sure what to do */
2611 sc->current_sample = sample;
2612 av_dlog(s, "stream %d, found sample %d\n", st->index, sc->current_sample);
2613 /* adjust ctts index */
2614 if (sc->ctts_data) {
2616 for (i = 0; i < sc->ctts_count; i++) {
2617 int next = time_sample + sc->ctts_data[i].count;
2618 if (next > sc->current_sample) {
2620 sc->ctts_sample = sc->current_sample - time_sample;
2629 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
2632 int64_t seek_timestamp, timestamp;
2636 if (stream_index >= s->nb_streams)
2638 if (sample_time < 0)
2641 st = s->streams[stream_index];
2642 sample = mov_seek_stream(s, st, sample_time, flags);
2646 /* adjust seek timestamp to found sample timestamp */
2647 seek_timestamp = st->index_entries[sample].timestamp;
2649 for (i = 0; i < s->nb_streams; i++) {
2651 if (stream_index == i)
2654 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
2655 mov_seek_stream(s, st, timestamp, flags);
2660 static int mov_read_close(AVFormatContext *s)
2662 MOVContext *mov = s->priv_data;
2665 for (i = 0; i < s->nb_streams; i++) {
2666 AVStream *st = s->streams[i];
2667 MOVStreamContext *sc = st->priv_data;
2669 av_freep(&sc->ctts_data);
2670 for (j = 0; j < sc->drefs_count; j++) {
2671 av_freep(&sc->drefs[j].path);
2672 av_freep(&sc->drefs[j].dir);
2674 av_freep(&sc->drefs);
2675 if (sc->pb && sc->pb != s->pb)
2679 if (mov->dv_demux) {
2680 for (i = 0; i < mov->dv_fctx->nb_streams; i++) {
2681 av_freep(&mov->dv_fctx->streams[i]->codec);
2682 av_freep(&mov->dv_fctx->streams[i]);
2684 av_freep(&mov->dv_fctx);
2685 av_freep(&mov->dv_demux);
2688 av_freep(&mov->trex_data);
2693 AVInputFormat ff_mov_demuxer = {
2694 .name = "mov,mp4,m4a,3gp,3g2,mj2",
2695 .long_name = NULL_IF_CONFIG_SMALL("QuickTime/MPEG-4/Motion JPEG 2000 format"),
2696 .priv_data_size = sizeof(MOVContext),
2697 .read_probe = mov_probe,
2698 .read_header = mov_read_header,
2699 .read_packet = mov_read_packet,
2700 .read_close = mov_read_close,
2701 .read_seek = mov_read_seek,