3 * Copyright (c) 2001 Fabrice Bellard
4 * Copyright (c) 2009 Baptiste Coudurier <baptiste dot coudurier at gmail dot com>
6 * first version by Francois Revol <revol@free.fr>
7 * seek function by Gael Chardon <gael.dev@4now.net>
9 * This file is part of FFmpeg.
11 * FFmpeg is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public
13 * License as published by the Free Software Foundation; either
14 * version 2.1 of the License, or (at your option) any later version.
16 * FFmpeg is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with FFmpeg; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
30 #include "libavutil/attributes.h"
31 #include "libavutil/channel_layout.h"
32 #include "libavutil/internal.h"
33 #include "libavutil/intreadwrite.h"
34 #include "libavutil/intfloat.h"
35 #include "libavutil/mathematics.h"
36 #include "libavutil/time_internal.h"
37 #include "libavutil/avassert.h"
38 #include "libavutil/avstring.h"
39 #include "libavutil/dict.h"
40 #include "libavutil/display.h"
41 #include "libavutil/opt.h"
42 #include "libavutil/aes.h"
43 #include "libavutil/aes_ctr.h"
44 #include "libavutil/pixdesc.h"
45 #include "libavutil/sha.h"
46 #include "libavutil/spherical.h"
47 #include "libavutil/stereo3d.h"
48 #include "libavutil/timecode.h"
49 #include "libavcodec/ac3tab.h"
50 #include "libavcodec/flac.h"
51 #include "libavcodec/mpegaudiodecheader.h"
54 #include "avio_internal.h"
57 #include "libavcodec/get_bits.h"
60 #include "replaygain.h"
66 #include "qtpalette.h"
68 /* those functions parse an atom */
69 /* links atom IDs to parse functions */
70 typedef struct MOVParseTableEntry {
72 int (*parse)(MOVContext *ctx, AVIOContext *pb, MOVAtom atom);
75 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom);
76 static int mov_read_mfra(MOVContext *c, AVIOContext *f);
77 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
78 int count, int duration);
80 static int mov_metadata_track_or_disc_number(MOVContext *c, AVIOContext *pb,
81 unsigned len, const char *key)
85 short current, total = 0;
86 avio_rb16(pb); // unknown
87 current = avio_rb16(pb);
89 total = avio_rb16(pb);
91 snprintf(buf, sizeof(buf), "%d", current);
93 snprintf(buf, sizeof(buf), "%d/%d", current, total);
94 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
95 av_dict_set(&c->fc->metadata, key, buf, 0);
100 static int mov_metadata_int8_bypass_padding(MOVContext *c, AVIOContext *pb,
101 unsigned len, const char *key)
103 /* bypass padding bytes */
108 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
109 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
114 static int mov_metadata_int8_no_padding(MOVContext *c, AVIOContext *pb,
115 unsigned len, const char *key)
117 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
118 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
123 static int mov_metadata_gnre(MOVContext *c, AVIOContext *pb,
124 unsigned len, const char *key)
128 avio_r8(pb); // unknown
131 if (genre < 1 || genre > ID3v1_GENRE_MAX)
133 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
134 av_dict_set(&c->fc->metadata, key, ff_id3v1_genre_str[genre-1], 0);
139 static const uint32_t mac_to_unicode[128] = {
140 0x00C4,0x00C5,0x00C7,0x00C9,0x00D1,0x00D6,0x00DC,0x00E1,
141 0x00E0,0x00E2,0x00E4,0x00E3,0x00E5,0x00E7,0x00E9,0x00E8,
142 0x00EA,0x00EB,0x00ED,0x00EC,0x00EE,0x00EF,0x00F1,0x00F3,
143 0x00F2,0x00F4,0x00F6,0x00F5,0x00FA,0x00F9,0x00FB,0x00FC,
144 0x2020,0x00B0,0x00A2,0x00A3,0x00A7,0x2022,0x00B6,0x00DF,
145 0x00AE,0x00A9,0x2122,0x00B4,0x00A8,0x2260,0x00C6,0x00D8,
146 0x221E,0x00B1,0x2264,0x2265,0x00A5,0x00B5,0x2202,0x2211,
147 0x220F,0x03C0,0x222B,0x00AA,0x00BA,0x03A9,0x00E6,0x00F8,
148 0x00BF,0x00A1,0x00AC,0x221A,0x0192,0x2248,0x2206,0x00AB,
149 0x00BB,0x2026,0x00A0,0x00C0,0x00C3,0x00D5,0x0152,0x0153,
150 0x2013,0x2014,0x201C,0x201D,0x2018,0x2019,0x00F7,0x25CA,
151 0x00FF,0x0178,0x2044,0x20AC,0x2039,0x203A,0xFB01,0xFB02,
152 0x2021,0x00B7,0x201A,0x201E,0x2030,0x00C2,0x00CA,0x00C1,
153 0x00CB,0x00C8,0x00CD,0x00CE,0x00CF,0x00CC,0x00D3,0x00D4,
154 0xF8FF,0x00D2,0x00DA,0x00DB,0x00D9,0x0131,0x02C6,0x02DC,
155 0x00AF,0x02D8,0x02D9,0x02DA,0x00B8,0x02DD,0x02DB,0x02C7,
158 static int mov_read_mac_string(MOVContext *c, AVIOContext *pb, int len,
159 char *dst, int dstlen)
162 char *end = dst+dstlen-1;
165 for (i = 0; i < len; i++) {
166 uint8_t t, c = avio_r8(pb);
174 PUT_UTF8(mac_to_unicode[c-0x80], t, if (p < end) *p++ = t;);
180 static int mov_read_covr(MOVContext *c, AVIOContext *pb, int type, int len)
184 MOVStreamContext *sc;
189 case 0xd: id = AV_CODEC_ID_MJPEG; break;
190 case 0xe: id = AV_CODEC_ID_PNG; break;
191 case 0x1b: id = AV_CODEC_ID_BMP; break;
193 av_log(c->fc, AV_LOG_WARNING, "Unknown cover type: 0x%x.\n", type);
198 st = avformat_new_stream(c->fc, NULL);
200 return AVERROR(ENOMEM);
201 sc = av_mallocz(sizeof(*sc));
203 return AVERROR(ENOMEM);
206 ret = av_get_packet(pb, &pkt, len);
210 if (pkt.size >= 8 && id != AV_CODEC_ID_BMP) {
211 if (AV_RB64(pkt.data) == 0x89504e470d0a1a0a) {
212 id = AV_CODEC_ID_PNG;
214 id = AV_CODEC_ID_MJPEG;
218 st->disposition |= AV_DISPOSITION_ATTACHED_PIC;
220 st->attached_pic = pkt;
221 st->attached_pic.stream_index = st->index;
222 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
224 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
225 st->codecpar->codec_id = id;
231 static int mov_metadata_loci(MOVContext *c, AVIOContext *pb, unsigned len)
233 char language[4] = { 0 };
234 char buf[200], place[100];
235 uint16_t langcode = 0;
236 double longitude, latitude, altitude;
237 const char *key = "location";
239 if (len < 4 + 2 + 1 + 1 + 4 + 4 + 4) {
240 av_log(c->fc, AV_LOG_ERROR, "loci too short\n");
241 return AVERROR_INVALIDDATA;
244 avio_skip(pb, 4); // version+flags
245 langcode = avio_rb16(pb);
246 ff_mov_lang_to_iso639(langcode, language);
249 len -= avio_get_str(pb, len, place, sizeof(place));
251 av_log(c->fc, AV_LOG_ERROR, "place name too long\n");
252 return AVERROR_INVALIDDATA;
254 avio_skip(pb, 1); // role
258 av_log(c->fc, AV_LOG_ERROR,
259 "loci too short (%u bytes left, need at least %d)\n", len, 12);
260 return AVERROR_INVALIDDATA;
262 longitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
263 latitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
264 altitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
266 // Try to output in the same format as the ?xyz field
267 snprintf(buf, sizeof(buf), "%+08.4f%+09.4f", latitude, longitude);
269 av_strlcatf(buf, sizeof(buf), "%+f", altitude);
270 av_strlcatf(buf, sizeof(buf), "/%s", place);
272 if (*language && strcmp(language, "und")) {
274 snprintf(key2, sizeof(key2), "%s-%s", key, language);
275 av_dict_set(&c->fc->metadata, key2, buf, 0);
277 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
278 return av_dict_set(&c->fc->metadata, key, buf, 0);
281 static int mov_metadata_hmmt(MOVContext *c, AVIOContext *pb, unsigned len)
287 if (c->ignore_chapters)
290 n_hmmt = avio_rb32(pb);
291 for (i = 0; i < n_hmmt && !pb->eof_reached; i++) {
292 int moment_time = avio_rb32(pb);
293 avpriv_new_chapter(c->fc, i, av_make_q(1, 1000), moment_time, AV_NOPTS_VALUE, NULL);
298 static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom)
301 char key2[32], language[4] = {0};
303 const char *key = NULL;
304 uint16_t langcode = 0;
305 uint32_t data_type = 0, str_size, str_size_alloc;
306 int (*parse)(MOVContext*, AVIOContext*, unsigned, const char*) = NULL;
311 case MKTAG( '@','P','R','M'): key = "premiere_version"; raw = 1; break;
312 case MKTAG( '@','P','R','Q'): key = "quicktime_version"; raw = 1; break;
313 case MKTAG( 'X','M','P','_'):
314 if (c->export_xmp) { key = "xmp"; raw = 1; } break;
315 case MKTAG( 'a','A','R','T'): key = "album_artist"; break;
316 case MKTAG( 'a','k','I','D'): key = "account_type";
317 parse = mov_metadata_int8_no_padding; break;
318 case MKTAG( 'a','p','I','D'): key = "account_id"; break;
319 case MKTAG( 'c','a','t','g'): key = "category"; break;
320 case MKTAG( 'c','p','i','l'): key = "compilation";
321 parse = mov_metadata_int8_no_padding; break;
322 case MKTAG( 'c','p','r','t'): key = "copyright"; break;
323 case MKTAG( 'd','e','s','c'): key = "description"; break;
324 case MKTAG( 'd','i','s','k'): key = "disc";
325 parse = mov_metadata_track_or_disc_number; break;
326 case MKTAG( 'e','g','i','d'): key = "episode_uid";
327 parse = mov_metadata_int8_no_padding; break;
328 case MKTAG( 'F','I','R','M'): key = "firmware"; raw = 1; break;
329 case MKTAG( 'g','n','r','e'): key = "genre";
330 parse = mov_metadata_gnre; break;
331 case MKTAG( 'h','d','v','d'): key = "hd_video";
332 parse = mov_metadata_int8_no_padding; break;
333 case MKTAG( 'H','M','M','T'):
334 return mov_metadata_hmmt(c, pb, atom.size);
335 case MKTAG( 'k','e','y','w'): key = "keywords"; break;
336 case MKTAG( 'l','d','e','s'): key = "synopsis"; break;
337 case MKTAG( 'l','o','c','i'):
338 return mov_metadata_loci(c, pb, atom.size);
339 case MKTAG( 'p','c','s','t'): key = "podcast";
340 parse = mov_metadata_int8_no_padding; break;
341 case MKTAG( 'p','g','a','p'): key = "gapless_playback";
342 parse = mov_metadata_int8_no_padding; break;
343 case MKTAG( 'p','u','r','d'): key = "purchase_date"; break;
344 case MKTAG( 'r','t','n','g'): key = "rating";
345 parse = mov_metadata_int8_no_padding; break;
346 case MKTAG( 's','o','a','a'): key = "sort_album_artist"; break;
347 case MKTAG( 's','o','a','l'): key = "sort_album"; break;
348 case MKTAG( 's','o','a','r'): key = "sort_artist"; break;
349 case MKTAG( 's','o','c','o'): key = "sort_composer"; break;
350 case MKTAG( 's','o','n','m'): key = "sort_name"; break;
351 case MKTAG( 's','o','s','n'): key = "sort_show"; break;
352 case MKTAG( 's','t','i','k'): key = "media_type";
353 parse = mov_metadata_int8_no_padding; break;
354 case MKTAG( 't','r','k','n'): key = "track";
355 parse = mov_metadata_track_or_disc_number; break;
356 case MKTAG( 't','v','e','n'): key = "episode_id"; break;
357 case MKTAG( 't','v','e','s'): key = "episode_sort";
358 parse = mov_metadata_int8_bypass_padding; break;
359 case MKTAG( 't','v','n','n'): key = "network"; break;
360 case MKTAG( 't','v','s','h'): key = "show"; break;
361 case MKTAG( 't','v','s','n'): key = "season_number";
362 parse = mov_metadata_int8_bypass_padding; break;
363 case MKTAG(0xa9,'A','R','T'): key = "artist"; break;
364 case MKTAG(0xa9,'P','R','D'): key = "producer"; break;
365 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
366 case MKTAG(0xa9,'a','u','t'): key = "artist"; break;
367 case MKTAG(0xa9,'c','h','p'): key = "chapter"; break;
368 case MKTAG(0xa9,'c','m','t'): key = "comment"; break;
369 case MKTAG(0xa9,'c','o','m'): key = "composer"; break;
370 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
371 case MKTAG(0xa9,'d','a','y'): key = "date"; break;
372 case MKTAG(0xa9,'d','i','r'): key = "director"; break;
373 case MKTAG(0xa9,'d','i','s'): key = "disclaimer"; break;
374 case MKTAG(0xa9,'e','d','1'): key = "edit_date"; break;
375 case MKTAG(0xa9,'e','n','c'): key = "encoder"; break;
376 case MKTAG(0xa9,'f','m','t'): key = "original_format"; break;
377 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
378 case MKTAG(0xa9,'g','r','p'): key = "grouping"; break;
379 case MKTAG(0xa9,'h','s','t'): key = "host_computer"; break;
380 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
381 case MKTAG(0xa9,'l','y','r'): key = "lyrics"; break;
382 case MKTAG(0xa9,'m','a','k'): key = "make"; break;
383 case MKTAG(0xa9,'m','o','d'): key = "model"; break;
384 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
385 case MKTAG(0xa9,'o','p','e'): key = "original_artist"; break;
386 case MKTAG(0xa9,'p','r','d'): key = "producer"; break;
387 case MKTAG(0xa9,'p','r','f'): key = "performers"; break;
388 case MKTAG(0xa9,'r','e','q'): key = "playback_requirements"; break;
389 case MKTAG(0xa9,'s','r','c'): key = "original_source"; break;
390 case MKTAG(0xa9,'s','t','3'): key = "subtitle"; break;
391 case MKTAG(0xa9,'s','w','r'): key = "encoder"; break;
392 case MKTAG(0xa9,'t','o','o'): key = "encoder"; break;
393 case MKTAG(0xa9,'t','r','k'): key = "track"; break;
394 case MKTAG(0xa9,'u','r','l'): key = "URL"; break;
395 case MKTAG(0xa9,'w','r','n'): key = "warning"; break;
396 case MKTAG(0xa9,'w','r','t'): key = "composer"; break;
397 case MKTAG(0xa9,'x','y','z'): key = "location"; break;
400 if (c->itunes_metadata && atom.size > 8) {
401 int data_size = avio_rb32(pb);
402 int tag = avio_rl32(pb);
403 if (tag == MKTAG('d','a','t','a') && data_size <= atom.size) {
404 data_type = avio_rb32(pb); // type
405 avio_rb32(pb); // unknown
406 str_size = data_size - 16;
409 if (atom.type == MKTAG('c', 'o', 'v', 'r')) {
410 int ret = mov_read_covr(c, pb, data_type, str_size);
412 av_log(c->fc, AV_LOG_ERROR, "Error parsing cover art.\n");
415 } else if (!key && c->found_hdlr_mdta && c->meta_keys) {
416 uint32_t index = AV_RB32(&atom.type);
417 if (index < c->meta_keys_count && index > 0) {
418 key = c->meta_keys[index];
420 av_log(c->fc, AV_LOG_WARNING,
421 "The index of 'data' is out of range: %"PRId32" < 1 or >= %d.\n",
422 index, c->meta_keys_count);
426 } else if (atom.size > 4 && key && !c->itunes_metadata && !raw) {
427 str_size = avio_rb16(pb); // string length
428 if (str_size > atom.size) {
430 avio_seek(pb, -2, SEEK_CUR);
431 av_log(c->fc, AV_LOG_WARNING, "UDTA parsing failed retrying raw\n");
434 langcode = avio_rb16(pb);
435 ff_mov_lang_to_iso639(langcode, language);
438 str_size = atom.size;
440 if (c->export_all && !key) {
441 snprintf(tmp_key, 5, "%.4s", (char*)&atom.type);
447 if (atom.size < 0 || str_size >= INT_MAX/2)
448 return AVERROR_INVALIDDATA;
450 // Allocates enough space if data_type is a int32 or float32 number, otherwise
451 // worst-case requirement for output string in case of utf8 coded input
452 num = (data_type >= 21 && data_type <= 23);
453 str_size_alloc = (num ? 512 : (raw ? str_size : str_size * 2)) + 1;
454 str = av_mallocz(str_size_alloc);
456 return AVERROR(ENOMEM);
459 parse(c, pb, str_size, key);
461 if (!raw && (data_type == 3 || (data_type == 0 && (langcode < 0x400 || langcode == 0x7fff)))) { // MAC Encoded
462 mov_read_mac_string(c, pb, str_size, str, str_size_alloc);
463 } else if (data_type == 21) { // BE signed integer, variable size
466 val = (int8_t)avio_r8(pb);
467 else if (str_size == 2)
468 val = (int16_t)avio_rb16(pb);
469 else if (str_size == 3)
470 val = ((int32_t)(avio_rb24(pb)<<8))>>8;
471 else if (str_size == 4)
472 val = (int32_t)avio_rb32(pb);
473 if (snprintf(str, str_size_alloc, "%d", val) >= str_size_alloc) {
474 av_log(c->fc, AV_LOG_ERROR,
475 "Failed to store the number (%d) in string.\n", val);
477 return AVERROR_INVALIDDATA;
479 } else if (data_type == 22) { // BE unsigned integer, variable size
480 unsigned int val = 0;
483 else if (str_size == 2)
485 else if (str_size == 3)
487 else if (str_size == 4)
489 if (snprintf(str, str_size_alloc, "%u", val) >= str_size_alloc) {
490 av_log(c->fc, AV_LOG_ERROR,
491 "Failed to store the number (%u) in string.\n", val);
493 return AVERROR_INVALIDDATA;
495 } else if (data_type == 23 && str_size >= 4) { // BE float32
496 float val = av_int2float(avio_rb32(pb));
497 if (snprintf(str, str_size_alloc, "%f", val) >= str_size_alloc) {
498 av_log(c->fc, AV_LOG_ERROR,
499 "Failed to store the float32 number (%f) in string.\n", val);
501 return AVERROR_INVALIDDATA;
504 int ret = ffio_read_size(pb, str, str_size);
511 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
512 av_dict_set(&c->fc->metadata, key, str, 0);
513 if (*language && strcmp(language, "und")) {
514 snprintf(key2, sizeof(key2), "%s-%s", key, language);
515 av_dict_set(&c->fc->metadata, key2, str, 0);
517 if (!strcmp(key, "encoder")) {
518 int major, minor, micro;
519 if (sscanf(str, "HandBrake %d.%d.%d", &major, &minor, µ) == 3) {
520 c->handbrake_version = 1000000*major + 1000*minor + micro;
529 static int mov_read_chpl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
532 int i, nb_chapters, str_len, version;
536 if (c->ignore_chapters)
539 if ((atom.size -= 5) < 0)
542 version = avio_r8(pb);
545 avio_rb32(pb); // ???
546 nb_chapters = avio_r8(pb);
548 for (i = 0; i < nb_chapters; i++) {
552 start = avio_rb64(pb);
553 str_len = avio_r8(pb);
555 if ((atom.size -= 9+str_len) < 0)
558 ret = ffio_read_size(pb, str, str_len);
562 avpriv_new_chapter(c->fc, i, (AVRational){1,10000000}, start, AV_NOPTS_VALUE, str);
567 #define MIN_DATA_ENTRY_BOX_SIZE 12
568 static int mov_read_dref(MOVContext *c, AVIOContext *pb, MOVAtom atom)
571 MOVStreamContext *sc;
574 if (c->fc->nb_streams < 1)
576 st = c->fc->streams[c->fc->nb_streams-1];
579 avio_rb32(pb); // version + flags
580 entries = avio_rb32(pb);
582 entries > (atom.size - 1) / MIN_DATA_ENTRY_BOX_SIZE + 1 ||
583 entries >= UINT_MAX / sizeof(*sc->drefs))
584 return AVERROR_INVALIDDATA;
588 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
590 return AVERROR(ENOMEM);
591 sc->drefs_count = entries;
593 for (i = 0; i < entries; i++) {
594 MOVDref *dref = &sc->drefs[i];
595 uint32_t size = avio_rb32(pb);
596 int64_t next = avio_tell(pb) + size - 4;
599 return AVERROR_INVALIDDATA;
601 dref->type = avio_rl32(pb);
602 avio_rb32(pb); // version + flags
604 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
605 /* macintosh alias record */
606 uint16_t volume_len, len;
612 volume_len = avio_r8(pb);
613 volume_len = FFMIN(volume_len, 27);
614 ret = ffio_read_size(pb, dref->volume, 27);
617 dref->volume[volume_len] = 0;
618 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", dref->volume, volume_len);
623 len = FFMIN(len, 63);
624 ret = ffio_read_size(pb, dref->filename, 63);
627 dref->filename[len] = 0;
628 av_log(c->fc, AV_LOG_DEBUG, "filename %s, len %d\n", dref->filename, len);
632 /* read next level up_from_alias/down_to_target */
633 dref->nlvl_from = avio_rb16(pb);
634 dref->nlvl_to = avio_rb16(pb);
635 av_log(c->fc, AV_LOG_DEBUG, "nlvl from %d, nlvl to %d\n",
636 dref->nlvl_from, dref->nlvl_to);
640 for (type = 0; type != -1 && avio_tell(pb) < next; ) {
643 type = avio_rb16(pb);
645 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
648 if (type == 2) { // absolute path
650 dref->path = av_mallocz(len+1);
652 return AVERROR(ENOMEM);
654 ret = ffio_read_size(pb, dref->path, len);
656 av_freep(&dref->path);
659 if (len > volume_len && !strncmp(dref->path, dref->volume, volume_len)) {
661 memmove(dref->path, dref->path+volume_len, len);
664 // trim string of any ending zeros
665 for (j = len - 1; j >= 0; j--) {
666 if (dref->path[j] == 0)
671 for (j = 0; j < len; j++)
672 if (dref->path[j] == ':' || dref->path[j] == 0)
674 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
675 } else if (type == 0) { // directory name
677 dref->dir = av_malloc(len+1);
679 return AVERROR(ENOMEM);
681 ret = ffio_read_size(pb, dref->dir, len);
683 av_freep(&dref->dir);
687 for (j = 0; j < len; j++)
688 if (dref->dir[j] == ':')
690 av_log(c->fc, AV_LOG_DEBUG, "dir %s\n", dref->dir);
695 av_log(c->fc, AV_LOG_DEBUG, "Unknown dref type 0x%08"PRIx32" size %"PRIu32"\n",
700 avio_seek(pb, next, SEEK_SET);
705 static int mov_read_hdlr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
714 avio_r8(pb); /* version */
715 avio_rb24(pb); /* flags */
718 ctype = avio_rl32(pb);
719 type = avio_rl32(pb); /* component subtype */
721 av_log(c->fc, AV_LOG_TRACE, "ctype=%s\n", av_fourcc2str(ctype));
722 av_log(c->fc, AV_LOG_TRACE, "stype=%s\n", av_fourcc2str(type));
724 if (c->trak_index < 0) { // meta not inside a trak
725 if (type == MKTAG('m','d','t','a')) {
726 c->found_hdlr_mdta = 1;
731 st = c->fc->streams[c->fc->nb_streams-1];
733 if (type == MKTAG('v','i','d','e'))
734 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
735 else if (type == MKTAG('s','o','u','n'))
736 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
737 else if (type == MKTAG('m','1','a',' '))
738 st->codecpar->codec_id = AV_CODEC_ID_MP2;
739 else if ((type == MKTAG('s','u','b','p')) || (type == MKTAG('c','l','c','p')))
740 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
742 avio_rb32(pb); /* component manufacture */
743 avio_rb32(pb); /* component flags */
744 avio_rb32(pb); /* component flags mask */
746 title_size = atom.size - 24;
747 if (title_size > 0) {
748 if (title_size > FFMIN(INT_MAX, SIZE_MAX-1))
749 return AVERROR_INVALIDDATA;
750 title_str = av_malloc(title_size + 1); /* Add null terminator */
752 return AVERROR(ENOMEM);
754 ret = ffio_read_size(pb, title_str, title_size);
756 av_freep(&title_str);
759 title_str[title_size] = 0;
761 int off = (!c->isom && title_str[0] == title_size - 1);
762 av_dict_set(&st->metadata, "handler_name", title_str + off, 0);
764 av_freep(&title_str);
770 static int mov_read_esds(MOVContext *c, AVIOContext *pb, MOVAtom atom)
772 return ff_mov_read_esds(c->fc, pb);
775 static int mov_read_dac3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
778 enum AVAudioServiceType *ast;
779 int ac3info, acmod, lfeon, bsmod;
781 if (c->fc->nb_streams < 1)
783 st = c->fc->streams[c->fc->nb_streams-1];
785 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
788 return AVERROR(ENOMEM);
790 ac3info = avio_rb24(pb);
791 bsmod = (ac3info >> 14) & 0x7;
792 acmod = (ac3info >> 11) & 0x7;
793 lfeon = (ac3info >> 10) & 0x1;
794 st->codecpar->channels = ((int[]){2,1,2,3,3,4,4,5})[acmod] + lfeon;
795 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
797 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
799 if (st->codecpar->channels > 1 && bsmod == 0x7)
800 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
802 #if FF_API_LAVF_AVCTX
803 FF_DISABLE_DEPRECATION_WARNINGS
804 st->codec->audio_service_type = *ast;
805 FF_ENABLE_DEPRECATION_WARNINGS
811 static int mov_read_dec3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
814 enum AVAudioServiceType *ast;
815 int eac3info, acmod, lfeon, bsmod;
817 if (c->fc->nb_streams < 1)
819 st = c->fc->streams[c->fc->nb_streams-1];
821 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
824 return AVERROR(ENOMEM);
826 /* No need to parse fields for additional independent substreams and its
827 * associated dependent substreams since libavcodec's E-AC-3 decoder
828 * does not support them yet. */
829 avio_rb16(pb); /* data_rate and num_ind_sub */
830 eac3info = avio_rb24(pb);
831 bsmod = (eac3info >> 12) & 0x1f;
832 acmod = (eac3info >> 9) & 0x7;
833 lfeon = (eac3info >> 8) & 0x1;
834 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
836 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
837 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
839 if (st->codecpar->channels > 1 && bsmod == 0x7)
840 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
842 #if FF_API_LAVF_AVCTX
843 FF_DISABLE_DEPRECATION_WARNINGS
844 st->codec->audio_service_type = *ast;
845 FF_ENABLE_DEPRECATION_WARNINGS
851 static int mov_read_ddts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
853 const uint32_t ddts_size = 20;
856 uint32_t frame_duration_code = 0;
857 uint32_t channel_layout_code = 0;
860 buf = av_malloc(ddts_size + AV_INPUT_BUFFER_PADDING_SIZE);
862 return AVERROR(ENOMEM);
864 if (avio_read(pb, buf, ddts_size) < ddts_size) {
866 return AVERROR_INVALIDDATA;
869 init_get_bits(&gb, buf, 8*ddts_size);
871 if (c->fc->nb_streams < 1) {
875 st = c->fc->streams[c->fc->nb_streams-1];
877 st->codecpar->sample_rate = get_bits_long(&gb, 32);
878 if (st->codecpar->sample_rate <= 0) {
879 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
881 return AVERROR_INVALIDDATA;
883 skip_bits_long(&gb, 32); /* max bitrate */
884 st->codecpar->bit_rate = get_bits_long(&gb, 32);
885 st->codecpar->bits_per_coded_sample = get_bits(&gb, 8);
886 frame_duration_code = get_bits(&gb, 2);
887 skip_bits(&gb, 30); /* various fields */
888 channel_layout_code = get_bits(&gb, 16);
890 st->codecpar->frame_size =
891 (frame_duration_code == 0) ? 512 :
892 (frame_duration_code == 1) ? 1024 :
893 (frame_duration_code == 2) ? 2048 :
894 (frame_duration_code == 3) ? 4096 : 0;
896 if (channel_layout_code > 0xff) {
897 av_log(c->fc, AV_LOG_WARNING, "Unsupported DTS audio channel layout");
899 st->codecpar->channel_layout =
900 ((channel_layout_code & 0x1) ? AV_CH_FRONT_CENTER : 0) |
901 ((channel_layout_code & 0x2) ? AV_CH_FRONT_LEFT : 0) |
902 ((channel_layout_code & 0x2) ? AV_CH_FRONT_RIGHT : 0) |
903 ((channel_layout_code & 0x4) ? AV_CH_SIDE_LEFT : 0) |
904 ((channel_layout_code & 0x4) ? AV_CH_SIDE_RIGHT : 0) |
905 ((channel_layout_code & 0x8) ? AV_CH_LOW_FREQUENCY : 0);
907 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
913 static int mov_read_chan(MOVContext *c, AVIOContext *pb, MOVAtom atom)
917 if (c->fc->nb_streams < 1)
919 st = c->fc->streams[c->fc->nb_streams-1];
924 /* skip version and flags */
927 ff_mov_read_chan(c->fc, pb, st, atom.size - 4);
932 static int mov_read_wfex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
937 if (c->fc->nb_streams < 1)
939 st = c->fc->streams[c->fc->nb_streams-1];
941 if ((ret = ff_get_wav_header(c->fc, pb, st->codecpar, atom.size, 0)) < 0)
942 av_log(c->fc, AV_LOG_WARNING, "get_wav_header failed\n");
947 static int mov_read_pasp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
949 const int num = avio_rb32(pb);
950 const int den = avio_rb32(pb);
953 if (c->fc->nb_streams < 1)
955 st = c->fc->streams[c->fc->nb_streams-1];
957 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
958 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num)) {
959 av_log(c->fc, AV_LOG_WARNING,
960 "sample aspect ratio already set to %d:%d, ignoring 'pasp' atom (%d:%d)\n",
961 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den,
963 } else if (den != 0) {
964 av_reduce(&st->sample_aspect_ratio.num, &st->sample_aspect_ratio.den,
970 /* this atom contains actual media data */
971 static int mov_read_mdat(MOVContext *c, AVIOContext *pb, MOVAtom atom)
973 if (atom.size == 0) /* wrong one (MP4) */
976 return 0; /* now go for moov */
979 #define DRM_BLOB_SIZE 56
981 static int mov_read_adrm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
983 uint8_t intermediate_key[20];
984 uint8_t intermediate_iv[20];
987 uint8_t file_checksum[20];
988 uint8_t calculated_checksum[20];
992 uint8_t *activation_bytes = c->activation_bytes;
993 uint8_t *fixed_key = c->audible_fixed_key;
997 sha = av_sha_alloc();
999 return AVERROR(ENOMEM);
1000 c->aes_decrypt = av_aes_alloc();
1001 if (!c->aes_decrypt) {
1002 ret = AVERROR(ENOMEM);
1006 /* drm blob processing */
1007 avio_read(pb, output, 8); // go to offset 8, absolute position 0x251
1008 avio_read(pb, input, DRM_BLOB_SIZE);
1009 avio_read(pb, output, 4); // go to offset 4, absolute position 0x28d
1010 avio_read(pb, file_checksum, 20);
1012 av_log(c->fc, AV_LOG_INFO, "[aax] file checksum == "); // required by external tools
1013 for (i = 0; i < 20; i++)
1014 av_log(c->fc, AV_LOG_INFO, "%02x", file_checksum[i]);
1015 av_log(c->fc, AV_LOG_INFO, "\n");
1017 /* verify activation data */
1018 if (!activation_bytes) {
1019 av_log(c->fc, AV_LOG_WARNING, "[aax] activation_bytes option is missing!\n");
1020 ret = 0; /* allow ffprobe to continue working on .aax files */
1023 if (c->activation_bytes_size != 4) {
1024 av_log(c->fc, AV_LOG_FATAL, "[aax] activation_bytes value needs to be 4 bytes!\n");
1025 ret = AVERROR(EINVAL);
1029 /* verify fixed key */
1030 if (c->audible_fixed_key_size != 16) {
1031 av_log(c->fc, AV_LOG_FATAL, "[aax] audible_fixed_key value needs to be 16 bytes!\n");
1032 ret = AVERROR(EINVAL);
1036 /* AAX (and AAX+) key derivation */
1037 av_sha_init(sha, 160);
1038 av_sha_update(sha, fixed_key, 16);
1039 av_sha_update(sha, activation_bytes, 4);
1040 av_sha_final(sha, intermediate_key);
1041 av_sha_init(sha, 160);
1042 av_sha_update(sha, fixed_key, 16);
1043 av_sha_update(sha, intermediate_key, 20);
1044 av_sha_update(sha, activation_bytes, 4);
1045 av_sha_final(sha, intermediate_iv);
1046 av_sha_init(sha, 160);
1047 av_sha_update(sha, intermediate_key, 16);
1048 av_sha_update(sha, intermediate_iv, 16);
1049 av_sha_final(sha, calculated_checksum);
1050 if (memcmp(calculated_checksum, file_checksum, 20)) { // critical error
1051 av_log(c->fc, AV_LOG_ERROR, "[aax] mismatch in checksums!\n");
1052 ret = AVERROR_INVALIDDATA;
1055 av_aes_init(c->aes_decrypt, intermediate_key, 128, 1);
1056 av_aes_crypt(c->aes_decrypt, output, input, DRM_BLOB_SIZE >> 4, intermediate_iv, 1);
1057 for (i = 0; i < 4; i++) {
1058 // file data (in output) is stored in big-endian mode
1059 if (activation_bytes[i] != output[3 - i]) { // critical error
1060 av_log(c->fc, AV_LOG_ERROR, "[aax] error in drm blob decryption!\n");
1061 ret = AVERROR_INVALIDDATA;
1065 memcpy(c->file_key, output + 8, 16);
1066 memcpy(input, output + 26, 16);
1067 av_sha_init(sha, 160);
1068 av_sha_update(sha, input, 16);
1069 av_sha_update(sha, c->file_key, 16);
1070 av_sha_update(sha, fixed_key, 16);
1071 av_sha_final(sha, c->file_iv);
1079 // Audible AAX (and AAX+) bytestream decryption
1080 static int aax_filter(uint8_t *input, int size, MOVContext *c)
1083 unsigned char iv[16];
1085 memcpy(iv, c->file_iv, 16); // iv is overwritten
1086 blocks = size >> 4; // trailing bytes are not encrypted!
1087 av_aes_init(c->aes_decrypt, c->file_key, 128, 1);
1088 av_aes_crypt(c->aes_decrypt, input, input, blocks, iv, 1);
1093 /* read major brand, minor version and compatible brands and store them as metadata */
1094 static int mov_read_ftyp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1097 int comp_brand_size;
1098 char* comp_brands_str;
1099 uint8_t type[5] = {0};
1100 int ret = ffio_read_size(pb, type, 4);
1104 if (strcmp(type, "qt "))
1106 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
1107 av_dict_set(&c->fc->metadata, "major_brand", type, 0);
1108 minor_ver = avio_rb32(pb); /* minor version */
1109 av_dict_set_int(&c->fc->metadata, "minor_version", minor_ver, 0);
1111 comp_brand_size = atom.size - 8;
1112 if (comp_brand_size < 0)
1113 return AVERROR_INVALIDDATA;
1114 comp_brands_str = av_malloc(comp_brand_size + 1); /* Add null terminator */
1115 if (!comp_brands_str)
1116 return AVERROR(ENOMEM);
1118 ret = ffio_read_size(pb, comp_brands_str, comp_brand_size);
1120 av_freep(&comp_brands_str);
1123 comp_brands_str[comp_brand_size] = 0;
1124 av_dict_set(&c->fc->metadata, "compatible_brands", comp_brands_str, 0);
1125 av_freep(&comp_brands_str);
1130 /* this atom should contain all header atoms */
1131 static int mov_read_moov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1135 if (c->found_moov) {
1136 av_log(c->fc, AV_LOG_WARNING, "Found duplicated MOOV Atom. Skipped it\n");
1137 avio_skip(pb, atom.size);
1141 if ((ret = mov_read_default(c, pb, atom)) < 0)
1143 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
1144 /* so we don't parse the whole file if over a network */
1146 return 0; /* now go for mdat */
1149 static MOVFragmentStreamInfo * get_frag_stream_info(
1150 MOVFragmentIndex *frag_index,
1155 MOVFragmentIndexItem * item;
1157 if (index < 0 || index >= frag_index->nb_items)
1159 item = &frag_index->item[index];
1160 for (i = 0; i < item->nb_stream_info; i++)
1161 if (item->stream_info[i].id == id)
1162 return &item->stream_info[i];
1164 // This shouldn't happen
1168 static void set_frag_stream(MOVFragmentIndex *frag_index, int id)
1171 MOVFragmentIndexItem * item;
1173 if (frag_index->current < 0 ||
1174 frag_index->current >= frag_index->nb_items)
1177 item = &frag_index->item[frag_index->current];
1178 for (i = 0; i < item->nb_stream_info; i++)
1179 if (item->stream_info[i].id == id) {
1184 // id not found. This shouldn't happen.
1188 static MOVFragmentStreamInfo * get_current_frag_stream_info(
1189 MOVFragmentIndex *frag_index)
1191 MOVFragmentIndexItem * item = &frag_index->item[frag_index->current];
1192 if (item->current >= 0 && item->current < item->nb_stream_info)
1193 return &item->stream_info[item->current];
1195 // This shouldn't happen
1199 static int search_frag_moof_offset(MOVFragmentIndex *frag_index, int64_t offset)
1202 int64_t moof_offset;
1204 // Optimize for appending new entries
1205 if (!frag_index->nb_items ||
1206 frag_index->item[frag_index->nb_items - 1].moof_offset < offset)
1207 return frag_index->nb_items;
1210 b = frag_index->nb_items;
1214 moof_offset = frag_index->item[m].moof_offset;
1215 if (moof_offset >= offset)
1217 if (moof_offset <= offset)
1223 static int64_t get_stream_info_time(MOVFragmentStreamInfo * frag_stream_info)
1226 if (frag_stream_info) {
1227 if (frag_stream_info->sidx_pts != AV_NOPTS_VALUE)
1228 return frag_stream_info->sidx_pts;
1229 if (frag_stream_info->first_tfra_pts != AV_NOPTS_VALUE)
1230 return frag_stream_info->first_tfra_pts;
1231 if (frag_stream_info->tfdt_dts != AV_NOPTS_VALUE)
1232 return frag_stream_info->tfdt_dts;
1234 return AV_NOPTS_VALUE;
1237 static int64_t get_frag_time(MOVFragmentIndex *frag_index,
1238 int index, int track_id)
1240 MOVFragmentStreamInfo * frag_stream_info;
1244 if (track_id >= 0) {
1245 frag_stream_info = get_frag_stream_info(frag_index, index, track_id);
1246 return frag_stream_info->sidx_pts;
1249 for (i = 0; i < frag_index->item[index].nb_stream_info; i++) {
1250 frag_stream_info = &frag_index->item[index].stream_info[i];
1251 timestamp = get_stream_info_time(frag_stream_info);
1252 if (timestamp != AV_NOPTS_VALUE)
1255 return AV_NOPTS_VALUE;
1258 static int search_frag_timestamp(MOVFragmentIndex *frag_index,
1259 AVStream *st, int64_t timestamp)
1266 // If the stream is referenced by any sidx, limit the search
1267 // to fragments that referenced this stream in the sidx
1268 MOVStreamContext *sc = st->priv_data;
1274 b = frag_index->nb_items;
1278 frag_time = get_frag_time(frag_index, m, id);
1279 if (frag_time != AV_NOPTS_VALUE) {
1280 if (frag_time >= timestamp)
1282 if (frag_time <= timestamp)
1289 static int update_frag_index(MOVContext *c, int64_t offset)
1292 MOVFragmentIndexItem * item;
1293 MOVFragmentStreamInfo * frag_stream_info;
1295 // If moof_offset already exists in frag_index, return index to it
1296 index = search_frag_moof_offset(&c->frag_index, offset);
1297 if (index < c->frag_index.nb_items &&
1298 c->frag_index.item[index].moof_offset == offset)
1301 // offset is not yet in frag index.
1302 // Insert new item at index (sorted by moof offset)
1303 item = av_fast_realloc(c->frag_index.item,
1304 &c->frag_index.allocated_size,
1305 (c->frag_index.nb_items + 1) *
1306 sizeof(*c->frag_index.item));
1309 c->frag_index.item = item;
1311 frag_stream_info = av_realloc_array(NULL, c->fc->nb_streams,
1312 sizeof(*item->stream_info));
1313 if (!frag_stream_info)
1316 for (i = 0; i < c->fc->nb_streams; i++) {
1317 frag_stream_info[i].id = c->fc->streams[i]->id;
1318 frag_stream_info[i].sidx_pts = AV_NOPTS_VALUE;
1319 frag_stream_info[i].tfdt_dts = AV_NOPTS_VALUE;
1320 frag_stream_info[i].first_tfra_pts = AV_NOPTS_VALUE;
1321 frag_stream_info[i].index_entry = -1;
1324 if (index < c->frag_index.nb_items)
1325 memmove(c->frag_index.item + index + 1, c->frag_index.item + index,
1326 (c->frag_index.nb_items - index) * sizeof(*c->frag_index.item));
1328 item = &c->frag_index.item[index];
1329 item->headers_read = 0;
1331 item->nb_stream_info = c->fc->nb_streams;
1332 item->moof_offset = offset;
1333 item->stream_info = frag_stream_info;
1334 c->frag_index.nb_items++;
1339 static void fix_frag_index_entries(MOVFragmentIndex *frag_index, int index,
1340 int id, int entries)
1343 MOVFragmentStreamInfo * frag_stream_info;
1347 for (i = index; i < frag_index->nb_items; i++) {
1348 frag_stream_info = get_frag_stream_info(frag_index, i, id);
1349 if (frag_stream_info && frag_stream_info->index_entry >= 0)
1350 frag_stream_info->index_entry += entries;
1354 static int mov_read_moof(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1356 if (!c->has_looked_for_mfra && c->use_mfra_for > 0) {
1357 c->has_looked_for_mfra = 1;
1358 if (pb->seekable & AVIO_SEEKABLE_NORMAL) {
1360 av_log(c->fc, AV_LOG_VERBOSE, "stream has moof boxes, will look "
1362 if ((ret = mov_read_mfra(c, pb)) < 0) {
1363 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but failed to "
1364 "read the mfra (may be a live ismv)\n");
1367 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but stream is not "
1368 "seekable, can not look for mfra\n");
1371 c->fragment.moof_offset = c->fragment.implicit_offset = avio_tell(pb) - 8;
1372 av_log(c->fc, AV_LOG_TRACE, "moof offset %"PRIx64"\n", c->fragment.moof_offset);
1373 c->frag_index.current = update_frag_index(c, c->fragment.moof_offset);
1374 return mov_read_default(c, pb, atom);
1377 static void mov_metadata_creation_time(AVDictionary **metadata, int64_t time)
1380 if(time >= 2082844800)
1381 time -= 2082844800; /* seconds between 1904-01-01 and Epoch */
1383 if ((int64_t)(time * 1000000ULL) / 1000000 != time) {
1384 av_log(NULL, AV_LOG_DEBUG, "creation_time is not representable\n");
1388 avpriv_dict_set_timestamp(metadata, "creation_time", time * 1000000);
1392 static int mov_read_mdhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1395 MOVStreamContext *sc;
1397 char language[4] = {0};
1399 int64_t creation_time;
1401 if (c->fc->nb_streams < 1)
1403 st = c->fc->streams[c->fc->nb_streams-1];
1406 if (sc->time_scale) {
1407 av_log(c->fc, AV_LOG_ERROR, "Multiple mdhd?\n");
1408 return AVERROR_INVALIDDATA;
1411 version = avio_r8(pb);
1413 avpriv_request_sample(c->fc, "Version %d", version);
1414 return AVERROR_PATCHWELCOME;
1416 avio_rb24(pb); /* flags */
1418 creation_time = avio_rb64(pb);
1421 creation_time = avio_rb32(pb);
1422 avio_rb32(pb); /* modification time */
1424 mov_metadata_creation_time(&st->metadata, creation_time);
1426 sc->time_scale = avio_rb32(pb);
1427 if (sc->time_scale <= 0) {
1428 av_log(c->fc, AV_LOG_ERROR, "Invalid mdhd time scale %d, defaulting to 1\n", sc->time_scale);
1431 st->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1433 lang = avio_rb16(pb); /* language */
1434 if (ff_mov_lang_to_iso639(lang, language))
1435 av_dict_set(&st->metadata, "language", language, 0);
1436 avio_rb16(pb); /* quality */
1441 static int mov_read_mvhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1444 int64_t creation_time;
1445 int version = avio_r8(pb); /* version */
1446 avio_rb24(pb); /* flags */
1449 creation_time = avio_rb64(pb);
1452 creation_time = avio_rb32(pb);
1453 avio_rb32(pb); /* modification time */
1455 mov_metadata_creation_time(&c->fc->metadata, creation_time);
1456 c->time_scale = avio_rb32(pb); /* time scale */
1457 if (c->time_scale <= 0) {
1458 av_log(c->fc, AV_LOG_ERROR, "Invalid mvhd time scale %d, defaulting to 1\n", c->time_scale);
1461 av_log(c->fc, AV_LOG_TRACE, "time scale = %i\n", c->time_scale);
1463 c->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1464 // set the AVCodecContext duration because the duration of individual tracks
1465 // may be inaccurate
1466 if (c->time_scale > 0 && !c->trex_data)
1467 c->fc->duration = av_rescale(c->duration, AV_TIME_BASE, c->time_scale);
1468 avio_rb32(pb); /* preferred scale */
1470 avio_rb16(pb); /* preferred volume */
1472 avio_skip(pb, 10); /* reserved */
1474 /* movie display matrix, store it in main context and use it later on */
1475 for (i = 0; i < 3; i++) {
1476 c->movie_display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
1477 c->movie_display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
1478 c->movie_display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
1481 avio_rb32(pb); /* preview time */
1482 avio_rb32(pb); /* preview duration */
1483 avio_rb32(pb); /* poster time */
1484 avio_rb32(pb); /* selection time */
1485 avio_rb32(pb); /* selection duration */
1486 avio_rb32(pb); /* current time */
1487 avio_rb32(pb); /* next track ID */
1492 static int mov_read_enda(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1497 if (c->fc->nb_streams < 1)
1499 st = c->fc->streams[c->fc->nb_streams-1];
1501 little_endian = avio_rb16(pb) & 0xFF;
1502 av_log(c->fc, AV_LOG_TRACE, "enda %d\n", little_endian);
1503 if (little_endian == 1) {
1504 switch (st->codecpar->codec_id) {
1505 case AV_CODEC_ID_PCM_S24BE:
1506 st->codecpar->codec_id = AV_CODEC_ID_PCM_S24LE;
1508 case AV_CODEC_ID_PCM_S32BE:
1509 st->codecpar->codec_id = AV_CODEC_ID_PCM_S32LE;
1511 case AV_CODEC_ID_PCM_F32BE:
1512 st->codecpar->codec_id = AV_CODEC_ID_PCM_F32LE;
1514 case AV_CODEC_ID_PCM_F64BE:
1515 st->codecpar->codec_id = AV_CODEC_ID_PCM_F64LE;
1524 static int mov_read_colr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1527 char color_parameter_type[5] = { 0 };
1528 uint16_t color_primaries, color_trc, color_matrix;
1531 if (c->fc->nb_streams < 1)
1533 st = c->fc->streams[c->fc->nb_streams - 1];
1535 ret = ffio_read_size(pb, color_parameter_type, 4);
1538 if (strncmp(color_parameter_type, "nclx", 4) &&
1539 strncmp(color_parameter_type, "nclc", 4)) {
1540 av_log(c->fc, AV_LOG_WARNING, "unsupported color_parameter_type %s\n",
1541 color_parameter_type);
1545 color_primaries = avio_rb16(pb);
1546 color_trc = avio_rb16(pb);
1547 color_matrix = avio_rb16(pb);
1549 av_log(c->fc, AV_LOG_TRACE,
1550 "%s: pri %d trc %d matrix %d",
1551 color_parameter_type, color_primaries, color_trc, color_matrix);
1553 if (!strncmp(color_parameter_type, "nclx", 4)) {
1554 uint8_t color_range = avio_r8(pb) >> 7;
1555 av_log(c->fc, AV_LOG_TRACE, " full %"PRIu8"", color_range);
1557 st->codecpar->color_range = AVCOL_RANGE_JPEG;
1559 st->codecpar->color_range = AVCOL_RANGE_MPEG;
1562 if (!av_color_primaries_name(color_primaries))
1563 color_primaries = AVCOL_PRI_UNSPECIFIED;
1564 if (!av_color_transfer_name(color_trc))
1565 color_trc = AVCOL_TRC_UNSPECIFIED;
1566 if (!av_color_space_name(color_matrix))
1567 color_matrix = AVCOL_SPC_UNSPECIFIED;
1569 st->codecpar->color_primaries = color_primaries;
1570 st->codecpar->color_trc = color_trc;
1571 st->codecpar->color_space = color_matrix;
1572 av_log(c->fc, AV_LOG_TRACE, "\n");
1577 static int mov_read_fiel(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1580 unsigned mov_field_order;
1581 enum AVFieldOrder decoded_field_order = AV_FIELD_UNKNOWN;
1583 if (c->fc->nb_streams < 1) // will happen with jp2 files
1585 st = c->fc->streams[c->fc->nb_streams-1];
1587 return AVERROR_INVALIDDATA;
1588 mov_field_order = avio_rb16(pb);
1589 if ((mov_field_order & 0xFF00) == 0x0100)
1590 decoded_field_order = AV_FIELD_PROGRESSIVE;
1591 else if ((mov_field_order & 0xFF00) == 0x0200) {
1592 switch (mov_field_order & 0xFF) {
1593 case 0x01: decoded_field_order = AV_FIELD_TT;
1595 case 0x06: decoded_field_order = AV_FIELD_BB;
1597 case 0x09: decoded_field_order = AV_FIELD_TB;
1599 case 0x0E: decoded_field_order = AV_FIELD_BT;
1603 if (decoded_field_order == AV_FIELD_UNKNOWN && mov_field_order) {
1604 av_log(NULL, AV_LOG_ERROR, "Unknown MOV field order 0x%04x\n", mov_field_order);
1606 st->codecpar->field_order = decoded_field_order;
1611 static int mov_realloc_extradata(AVCodecParameters *par, MOVAtom atom)
1614 uint64_t size = (uint64_t)par->extradata_size + atom.size + 8 + AV_INPUT_BUFFER_PADDING_SIZE;
1615 if (size > INT_MAX || (uint64_t)atom.size > INT_MAX)
1616 return AVERROR_INVALIDDATA;
1617 if ((err = av_reallocp(&par->extradata, size)) < 0) {
1618 par->extradata_size = 0;
1621 par->extradata_size = size - AV_INPUT_BUFFER_PADDING_SIZE;
1625 /* Read a whole atom into the extradata return the size of the atom read, possibly truncated if != atom.size */
1626 static int64_t mov_read_atom_into_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1627 AVCodecParameters *par, uint8_t *buf)
1629 int64_t result = atom.size;
1632 AV_WB32(buf , atom.size + 8);
1633 AV_WL32(buf + 4, atom.type);
1634 err = ffio_read_size(pb, buf + 8, atom.size);
1636 par->extradata_size -= atom.size;
1638 } else if (err < atom.size) {
1639 av_log(c->fc, AV_LOG_WARNING, "truncated extradata\n");
1640 par->extradata_size -= atom.size - err;
1643 memset(buf + 8 + err, 0, AV_INPUT_BUFFER_PADDING_SIZE);
1647 /* FIXME modify QDM2/SVQ3/H.264 decoders to take full atom as extradata */
1648 static int mov_read_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1649 enum AVCodecID codec_id)
1652 uint64_t original_size;
1655 if (c->fc->nb_streams < 1) // will happen with jp2 files
1657 st = c->fc->streams[c->fc->nb_streams-1];
1659 if (st->codecpar->codec_id != codec_id)
1660 return 0; /* unexpected codec_id - don't mess with extradata */
1662 original_size = st->codecpar->extradata_size;
1663 err = mov_realloc_extradata(st->codecpar, atom);
1667 err = mov_read_atom_into_extradata(c, pb, atom, st->codecpar, st->codecpar->extradata + original_size);
1670 return 0; // Note: this is the original behavior to ignore truncation.
1673 /* wrapper functions for reading ALAC/AVS/MJPEG/MJPEG2000 extradata atoms only for those codecs */
1674 static int mov_read_alac(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1676 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_ALAC);
1679 static int mov_read_avss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1681 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVS);
1684 static int mov_read_jp2h(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1686 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_JPEG2000);
1689 static int mov_read_dpxe(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1691 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_R10K);
1694 static int mov_read_avid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1696 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVUI);
1698 ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_DNXHD);
1702 static int mov_read_targa_y216(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1704 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_TARGA_Y216);
1706 if (!ret && c->fc->nb_streams >= 1) {
1707 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1708 if (par->extradata_size >= 40) {
1709 par->height = AV_RB16(&par->extradata[36]);
1710 par->width = AV_RB16(&par->extradata[38]);
1716 static int mov_read_ares(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1718 if (c->fc->nb_streams >= 1) {
1719 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1720 if (par->codec_tag == MKTAG('A', 'V', 'i', 'n') &&
1721 par->codec_id == AV_CODEC_ID_H264 &&
1725 cid = avio_rb16(pb);
1726 /* For AVID AVCI50, force width of 1440 to be able to select the correct SPS and PPS */
1727 if (cid == 0xd4d || cid == 0xd4e)
1730 } else if ((par->codec_tag == MKTAG('A', 'V', 'd', '1') ||
1731 par->codec_tag == MKTAG('A', 'V', 'd', 'n')) &&
1735 num = avio_rb32(pb);
1736 den = avio_rb32(pb);
1737 if (num <= 0 || den <= 0)
1739 switch (avio_rb32(pb)) {
1741 if (den >= INT_MAX / 2)
1745 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.num = num;
1746 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.den = den;
1753 return mov_read_avid(c, pb, atom);
1756 static int mov_read_aclr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1760 uint64_t original_size;
1761 if (c->fc->nb_streams >= 1) {
1762 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1763 if (par->codec_id == AV_CODEC_ID_H264)
1765 if (atom.size == 16) {
1766 original_size = par->extradata_size;
1767 ret = mov_realloc_extradata(par, atom);
1769 length = mov_read_atom_into_extradata(c, pb, atom, par, par->extradata + original_size);
1770 if (length == atom.size) {
1771 const uint8_t range_value = par->extradata[original_size + 19];
1772 switch (range_value) {
1774 par->color_range = AVCOL_RANGE_MPEG;
1777 par->color_range = AVCOL_RANGE_JPEG;
1780 av_log(c, AV_LOG_WARNING, "ignored unknown aclr value (%d)\n", range_value);
1783 ff_dlog(c, "color_range: %d\n", par->color_range);
1785 /* For some reason the whole atom was not added to the extradata */
1786 av_log(c, AV_LOG_ERROR, "aclr not decoded - incomplete atom\n");
1789 av_log(c, AV_LOG_ERROR, "aclr not decoded - unable to add atom to extradata\n");
1792 av_log(c, AV_LOG_WARNING, "aclr not decoded - unexpected size %"PRId64"\n", atom.size);
1799 static int mov_read_svq3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1801 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_SVQ3);
1804 static int mov_read_wave(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1809 if (c->fc->nb_streams < 1)
1811 st = c->fc->streams[c->fc->nb_streams-1];
1813 if ((uint64_t)atom.size > (1<<30))
1814 return AVERROR_INVALIDDATA;
1816 if (st->codecpar->codec_id == AV_CODEC_ID_QDM2 ||
1817 st->codecpar->codec_id == AV_CODEC_ID_QDMC ||
1818 st->codecpar->codec_id == AV_CODEC_ID_SPEEX) {
1819 // pass all frma atom to codec, needed at least for QDMC and QDM2
1820 av_freep(&st->codecpar->extradata);
1821 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1824 } else if (atom.size > 8) { /* to read frma, esds atoms */
1825 if (st->codecpar->codec_id == AV_CODEC_ID_ALAC && atom.size >= 24) {
1827 ret = ffio_ensure_seekback(pb, 8);
1830 buffer = avio_rb64(pb);
1832 if ( (buffer & 0xFFFFFFFF) == MKBETAG('f','r','m','a')
1833 && buffer >> 32 <= atom.size
1834 && buffer >> 32 >= 8) {
1837 } else if (!st->codecpar->extradata_size) {
1838 #define ALAC_EXTRADATA_SIZE 36
1839 st->codecpar->extradata = av_mallocz(ALAC_EXTRADATA_SIZE + AV_INPUT_BUFFER_PADDING_SIZE);
1840 if (!st->codecpar->extradata)
1841 return AVERROR(ENOMEM);
1842 st->codecpar->extradata_size = ALAC_EXTRADATA_SIZE;
1843 AV_WB32(st->codecpar->extradata , ALAC_EXTRADATA_SIZE);
1844 AV_WB32(st->codecpar->extradata + 4, MKTAG('a','l','a','c'));
1845 AV_WB64(st->codecpar->extradata + 12, buffer);
1846 avio_read(pb, st->codecpar->extradata + 20, 16);
1847 avio_skip(pb, atom.size - 24);
1851 if ((ret = mov_read_default(c, pb, atom)) < 0)
1854 avio_skip(pb, atom.size);
1859 * This function reads atom content and puts data in extradata without tag
1860 * nor size unlike mov_read_extradata.
1862 static int mov_read_glbl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1867 if (c->fc->nb_streams < 1)
1869 st = c->fc->streams[c->fc->nb_streams-1];
1871 if ((uint64_t)atom.size > (1<<30))
1872 return AVERROR_INVALIDDATA;
1874 if (atom.size >= 10) {
1875 // Broken files created by legacy versions of libavformat will
1876 // wrap a whole fiel atom inside of a glbl atom.
1877 unsigned size = avio_rb32(pb);
1878 unsigned type = avio_rl32(pb);
1879 avio_seek(pb, -8, SEEK_CUR);
1880 if (type == MKTAG('f','i','e','l') && size == atom.size)
1881 return mov_read_default(c, pb, atom);
1883 if (st->codecpar->extradata_size > 1 && st->codecpar->extradata) {
1884 av_log(c, AV_LOG_WARNING, "ignoring multiple glbl\n");
1887 av_freep(&st->codecpar->extradata);
1888 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1895 static int mov_read_dvc1(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1898 uint8_t profile_level;
1901 if (c->fc->nb_streams < 1)
1903 st = c->fc->streams[c->fc->nb_streams-1];
1905 if (atom.size >= (1<<28) || atom.size < 7)
1906 return AVERROR_INVALIDDATA;
1908 profile_level = avio_r8(pb);
1909 if ((profile_level & 0xf0) != 0xc0)
1912 avio_seek(pb, 6, SEEK_CUR);
1913 av_freep(&st->codecpar->extradata);
1914 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 7);
1922 * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
1923 * but can have extradata appended at the end after the 40 bytes belonging
1926 static int mov_read_strf(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1931 if (c->fc->nb_streams < 1)
1933 if (atom.size <= 40)
1935 st = c->fc->streams[c->fc->nb_streams-1];
1937 if ((uint64_t)atom.size > (1<<30))
1938 return AVERROR_INVALIDDATA;
1941 av_freep(&st->codecpar->extradata);
1942 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 40);
1949 static int mov_read_stco(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1952 MOVStreamContext *sc;
1953 unsigned int i, entries;
1955 if (c->fc->nb_streams < 1)
1957 st = c->fc->streams[c->fc->nb_streams-1];
1960 avio_r8(pb); /* version */
1961 avio_rb24(pb); /* flags */
1963 entries = avio_rb32(pb);
1968 if (sc->chunk_offsets)
1969 av_log(c->fc, AV_LOG_WARNING, "Duplicated STCO atom\n");
1970 av_free(sc->chunk_offsets);
1971 sc->chunk_count = 0;
1972 sc->chunk_offsets = av_malloc_array(entries, sizeof(*sc->chunk_offsets));
1973 if (!sc->chunk_offsets)
1974 return AVERROR(ENOMEM);
1975 sc->chunk_count = entries;
1977 if (atom.type == MKTAG('s','t','c','o'))
1978 for (i = 0; i < entries && !pb->eof_reached; i++)
1979 sc->chunk_offsets[i] = avio_rb32(pb);
1980 else if (atom.type == MKTAG('c','o','6','4'))
1981 for (i = 0; i < entries && !pb->eof_reached; i++)
1982 sc->chunk_offsets[i] = avio_rb64(pb);
1984 return AVERROR_INVALIDDATA;
1986 sc->chunk_count = i;
1988 if (pb->eof_reached)
1994 static int mov_codec_id(AVStream *st, uint32_t format)
1996 int id = ff_codec_get_id(ff_codec_movaudio_tags, format);
1999 ((format & 0xFFFF) == 'm' + ('s' << 8) ||
2000 (format & 0xFFFF) == 'T' + ('S' << 8)))
2001 id = ff_codec_get_id(ff_codec_wav_tags, av_bswap32(format) & 0xFFFF);
2003 if (st->codecpar->codec_type != AVMEDIA_TYPE_VIDEO && id > 0) {
2004 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2005 } else if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO &&
2006 /* skip old ASF MPEG-4 tag */
2007 format && format != MKTAG('m','p','4','s')) {
2008 id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2010 id = ff_codec_get_id(ff_codec_bmp_tags, format);
2012 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
2013 else if (st->codecpar->codec_type == AVMEDIA_TYPE_DATA ||
2014 (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE &&
2015 st->codecpar->codec_id == AV_CODEC_ID_NONE)) {
2016 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
2018 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
2022 st->codecpar->codec_tag = format;
2027 static void mov_parse_stsd_video(MOVContext *c, AVIOContext *pb,
2028 AVStream *st, MOVStreamContext *sc)
2030 uint8_t codec_name[32] = { 0 };
2034 /* The first 16 bytes of the video sample description are already
2035 * read in ff_mov_read_stsd_entries() */
2036 stsd_start = avio_tell(pb) - 16;
2038 avio_rb16(pb); /* version */
2039 avio_rb16(pb); /* revision level */
2040 avio_rb32(pb); /* vendor */
2041 avio_rb32(pb); /* temporal quality */
2042 avio_rb32(pb); /* spatial quality */
2044 st->codecpar->width = avio_rb16(pb); /* width */
2045 st->codecpar->height = avio_rb16(pb); /* height */
2047 avio_rb32(pb); /* horiz resolution */
2048 avio_rb32(pb); /* vert resolution */
2049 avio_rb32(pb); /* data size, always 0 */
2050 avio_rb16(pb); /* frames per samples */
2052 len = avio_r8(pb); /* codec name, pascal string */
2055 mov_read_mac_string(c, pb, len, codec_name, sizeof(codec_name));
2057 avio_skip(pb, 31 - len);
2060 av_dict_set(&st->metadata, "encoder", codec_name, 0);
2062 /* codec_tag YV12 triggers an UV swap in rawdec.c */
2063 if (!strncmp(codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25)) {
2064 st->codecpar->codec_tag = MKTAG('I', '4', '2', '0');
2065 st->codecpar->width &= ~1;
2066 st->codecpar->height &= ~1;
2068 /* Flash Media Server uses tag H.263 with Sorenson Spark */
2069 if (st->codecpar->codec_tag == MKTAG('H','2','6','3') &&
2070 !strncmp(codec_name, "Sorenson H263", 13))
2071 st->codecpar->codec_id = AV_CODEC_ID_FLV1;
2073 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* depth */
2075 avio_seek(pb, stsd_start, SEEK_SET);
2077 if (ff_get_qtpalette(st->codecpar->codec_id, pb, sc->palette)) {
2078 st->codecpar->bits_per_coded_sample &= 0x1F;
2079 sc->has_palette = 1;
2083 static void mov_parse_stsd_audio(MOVContext *c, AVIOContext *pb,
2084 AVStream *st, MOVStreamContext *sc)
2086 int bits_per_sample, flags;
2087 uint16_t version = avio_rb16(pb);
2088 AVDictionaryEntry *compatible_brands = av_dict_get(c->fc->metadata, "compatible_brands", NULL, AV_DICT_MATCH_CASE);
2090 avio_rb16(pb); /* revision level */
2091 avio_rb32(pb); /* vendor */
2093 st->codecpar->channels = avio_rb16(pb); /* channel count */
2094 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* sample size */
2095 av_log(c->fc, AV_LOG_TRACE, "audio channels %d\n", st->codecpar->channels);
2097 sc->audio_cid = avio_rb16(pb);
2098 avio_rb16(pb); /* packet size = 0 */
2100 st->codecpar->sample_rate = ((avio_rb32(pb) >> 16));
2102 // Read QT version 1 fields. In version 0 these do not exist.
2103 av_log(c->fc, AV_LOG_TRACE, "version =%d, isom =%d\n", version, c->isom);
2105 (compatible_brands && strstr(compatible_brands->value, "qt "))) {
2108 sc->samples_per_frame = avio_rb32(pb);
2109 avio_rb32(pb); /* bytes per packet */
2110 sc->bytes_per_frame = avio_rb32(pb);
2111 avio_rb32(pb); /* bytes per sample */
2112 } else if (version == 2) {
2113 avio_rb32(pb); /* sizeof struct only */
2114 st->codecpar->sample_rate = av_int2double(avio_rb64(pb));
2115 st->codecpar->channels = avio_rb32(pb);
2116 avio_rb32(pb); /* always 0x7F000000 */
2117 st->codecpar->bits_per_coded_sample = avio_rb32(pb);
2119 flags = avio_rb32(pb); /* lpcm format specific flag */
2120 sc->bytes_per_frame = avio_rb32(pb);
2121 sc->samples_per_frame = avio_rb32(pb);
2122 if (st->codecpar->codec_tag == MKTAG('l','p','c','m'))
2123 st->codecpar->codec_id =
2124 ff_mov_get_lpcm_codec_id(st->codecpar->bits_per_coded_sample,
2127 if (version == 0 || (version == 1 && sc->audio_cid != -2)) {
2128 /* can't correctly handle variable sized packet as audio unit */
2129 switch (st->codecpar->codec_id) {
2130 case AV_CODEC_ID_MP2:
2131 case AV_CODEC_ID_MP3:
2132 st->need_parsing = AVSTREAM_PARSE_FULL;
2138 if (sc->format == 0) {
2139 if (st->codecpar->bits_per_coded_sample == 8)
2140 st->codecpar->codec_id = mov_codec_id(st, MKTAG('r','a','w',' '));
2141 else if (st->codecpar->bits_per_coded_sample == 16)
2142 st->codecpar->codec_id = mov_codec_id(st, MKTAG('t','w','o','s'));
2145 switch (st->codecpar->codec_id) {
2146 case AV_CODEC_ID_PCM_S8:
2147 case AV_CODEC_ID_PCM_U8:
2148 if (st->codecpar->bits_per_coded_sample == 16)
2149 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16BE;
2151 case AV_CODEC_ID_PCM_S16LE:
2152 case AV_CODEC_ID_PCM_S16BE:
2153 if (st->codecpar->bits_per_coded_sample == 8)
2154 st->codecpar->codec_id = AV_CODEC_ID_PCM_S8;
2155 else if (st->codecpar->bits_per_coded_sample == 24)
2156 st->codecpar->codec_id =
2157 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
2158 AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
2159 else if (st->codecpar->bits_per_coded_sample == 32)
2160 st->codecpar->codec_id =
2161 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
2162 AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
2164 /* set values for old format before stsd version 1 appeared */
2165 case AV_CODEC_ID_MACE3:
2166 sc->samples_per_frame = 6;
2167 sc->bytes_per_frame = 2 * st->codecpar->channels;
2169 case AV_CODEC_ID_MACE6:
2170 sc->samples_per_frame = 6;
2171 sc->bytes_per_frame = 1 * st->codecpar->channels;
2173 case AV_CODEC_ID_ADPCM_IMA_QT:
2174 sc->samples_per_frame = 64;
2175 sc->bytes_per_frame = 34 * st->codecpar->channels;
2177 case AV_CODEC_ID_GSM:
2178 sc->samples_per_frame = 160;
2179 sc->bytes_per_frame = 33;
2185 bits_per_sample = av_get_bits_per_sample(st->codecpar->codec_id);
2186 if (bits_per_sample) {
2187 st->codecpar->bits_per_coded_sample = bits_per_sample;
2188 sc->sample_size = (bits_per_sample >> 3) * st->codecpar->channels;
2192 static void mov_parse_stsd_subtitle(MOVContext *c, AVIOContext *pb,
2193 AVStream *st, MOVStreamContext *sc,
2196 // ttxt stsd contains display flags, justification, background
2197 // color, fonts, and default styles, so fake an atom to read it
2198 MOVAtom fake_atom = { .size = size };
2199 // mp4s contains a regular esds atom
2200 if (st->codecpar->codec_tag != AV_RL32("mp4s"))
2201 mov_read_glbl(c, pb, fake_atom);
2202 st->codecpar->width = sc->width;
2203 st->codecpar->height = sc->height;
2206 static uint32_t yuv_to_rgba(uint32_t ycbcr)
2211 y = (ycbcr >> 16) & 0xFF;
2212 cr = (ycbcr >> 8) & 0xFF;
2215 b = av_clip_uint8((1164 * (y - 16) + 2018 * (cb - 128)) / 1000);
2216 g = av_clip_uint8((1164 * (y - 16) - 813 * (cr - 128) - 391 * (cb - 128)) / 1000);
2217 r = av_clip_uint8((1164 * (y - 16) + 1596 * (cr - 128) ) / 1000);
2219 return (r << 16) | (g << 8) | b;
2222 static int mov_rewrite_dvd_sub_extradata(AVStream *st)
2224 char buf[256] = {0};
2225 uint8_t *src = st->codecpar->extradata;
2228 if (st->codecpar->extradata_size != 64)
2231 if (st->codecpar->width > 0 && st->codecpar->height > 0)
2232 snprintf(buf, sizeof(buf), "size: %dx%d\n",
2233 st->codecpar->width, st->codecpar->height);
2234 av_strlcat(buf, "palette: ", sizeof(buf));
2236 for (i = 0; i < 16; i++) {
2237 uint32_t yuv = AV_RB32(src + i * 4);
2238 uint32_t rgba = yuv_to_rgba(yuv);
2240 av_strlcatf(buf, sizeof(buf), "%06"PRIx32"%s", rgba, i != 15 ? ", " : "");
2243 if (av_strlcat(buf, "\n", sizeof(buf)) >= sizeof(buf))
2246 av_freep(&st->codecpar->extradata);
2247 st->codecpar->extradata_size = 0;
2248 st->codecpar->extradata = av_mallocz(strlen(buf) + AV_INPUT_BUFFER_PADDING_SIZE);
2249 if (!st->codecpar->extradata)
2250 return AVERROR(ENOMEM);
2251 st->codecpar->extradata_size = strlen(buf);
2252 memcpy(st->codecpar->extradata, buf, st->codecpar->extradata_size);
2257 static int mov_parse_stsd_data(MOVContext *c, AVIOContext *pb,
2258 AVStream *st, MOVStreamContext *sc,
2263 if (st->codecpar->codec_tag == MKTAG('t','m','c','d')) {
2264 if ((int)size != size)
2265 return AVERROR(ENOMEM);
2267 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
2271 MOVStreamContext *tmcd_ctx = st->priv_data;
2273 val = AV_RB32(st->codecpar->extradata + 4);
2274 tmcd_ctx->tmcd_flags = val;
2275 st->avg_frame_rate.num = st->codecpar->extradata[16]; /* number of frame */
2276 st->avg_frame_rate.den = 1;
2277 #if FF_API_LAVF_AVCTX
2278 FF_DISABLE_DEPRECATION_WARNINGS
2279 st->codec->time_base = av_inv_q(st->avg_frame_rate);
2280 FF_ENABLE_DEPRECATION_WARNINGS
2282 /* adjust for per frame dur in counter mode */
2283 if (tmcd_ctx->tmcd_flags & 0x0008) {
2284 int timescale = AV_RB32(st->codecpar->extradata + 8);
2285 int framedur = AV_RB32(st->codecpar->extradata + 12);
2286 st->avg_frame_rate.num *= timescale;
2287 st->avg_frame_rate.den *= framedur;
2288 #if FF_API_LAVF_AVCTX
2289 FF_DISABLE_DEPRECATION_WARNINGS
2290 st->codec->time_base.den *= timescale;
2291 st->codec->time_base.num *= framedur;
2292 FF_ENABLE_DEPRECATION_WARNINGS
2296 uint32_t len = AV_RB32(st->codecpar->extradata + 18); /* name atom length */
2297 uint32_t format = AV_RB32(st->codecpar->extradata + 22);
2298 if (format == AV_RB32("name") && (int64_t)size >= (int64_t)len + 18) {
2299 uint16_t str_size = AV_RB16(st->codecpar->extradata + 26); /* string length */
2300 if (str_size > 0 && size >= (int)str_size + 26) {
2301 char *reel_name = av_malloc(str_size + 1);
2303 return AVERROR(ENOMEM);
2304 memcpy(reel_name, st->codecpar->extradata + 30, str_size);
2305 reel_name[str_size] = 0; /* Add null terminator */
2306 /* don't add reel_name if emtpy string */
2307 if (*reel_name == 0) {
2310 av_dict_set(&st->metadata, "reel_name", reel_name, AV_DICT_DONT_STRDUP_VAL);
2317 /* other codec type, just skip (rtp, mp4s ...) */
2318 avio_skip(pb, size);
2323 static int mov_finalize_stsd_codec(MOVContext *c, AVIOContext *pb,
2324 AVStream *st, MOVStreamContext *sc)
2326 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
2327 !st->codecpar->sample_rate && sc->time_scale > 1)
2328 st->codecpar->sample_rate = sc->time_scale;
2330 /* special codec parameters handling */
2331 switch (st->codecpar->codec_id) {
2332 #if CONFIG_DV_DEMUXER
2333 case AV_CODEC_ID_DVAUDIO:
2334 c->dv_fctx = avformat_alloc_context();
2336 av_log(c->fc, AV_LOG_ERROR, "dv demux context alloc error\n");
2337 return AVERROR(ENOMEM);
2339 c->dv_demux = avpriv_dv_init_demux(c->dv_fctx);
2341 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
2342 return AVERROR(ENOMEM);
2344 sc->dv_audio_container = 1;
2345 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16LE;
2348 /* no ifdef since parameters are always those */
2349 case AV_CODEC_ID_QCELP:
2350 st->codecpar->channels = 1;
2351 // force sample rate for qcelp when not stored in mov
2352 if (st->codecpar->codec_tag != MKTAG('Q','c','l','p'))
2353 st->codecpar->sample_rate = 8000;
2354 // FIXME: Why is the following needed for some files?
2355 sc->samples_per_frame = 160;
2356 if (!sc->bytes_per_frame)
2357 sc->bytes_per_frame = 35;
2359 case AV_CODEC_ID_AMR_NB:
2360 st->codecpar->channels = 1;
2361 /* force sample rate for amr, stsd in 3gp does not store sample rate */
2362 st->codecpar->sample_rate = 8000;
2364 case AV_CODEC_ID_AMR_WB:
2365 st->codecpar->channels = 1;
2366 st->codecpar->sample_rate = 16000;
2368 case AV_CODEC_ID_MP2:
2369 case AV_CODEC_ID_MP3:
2370 /* force type after stsd for m1a hdlr */
2371 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2373 case AV_CODEC_ID_GSM:
2374 case AV_CODEC_ID_ADPCM_MS:
2375 case AV_CODEC_ID_ADPCM_IMA_WAV:
2376 case AV_CODEC_ID_ILBC:
2377 case AV_CODEC_ID_MACE3:
2378 case AV_CODEC_ID_MACE6:
2379 case AV_CODEC_ID_QDM2:
2380 st->codecpar->block_align = sc->bytes_per_frame;
2382 case AV_CODEC_ID_ALAC:
2383 if (st->codecpar->extradata_size == 36) {
2384 st->codecpar->channels = AV_RB8 (st->codecpar->extradata + 21);
2385 st->codecpar->sample_rate = AV_RB32(st->codecpar->extradata + 32);
2388 case AV_CODEC_ID_AC3:
2389 case AV_CODEC_ID_EAC3:
2390 case AV_CODEC_ID_MPEG1VIDEO:
2391 case AV_CODEC_ID_VC1:
2392 case AV_CODEC_ID_VP9:
2393 st->need_parsing = AVSTREAM_PARSE_FULL;
2401 static int mov_skip_multiple_stsd(MOVContext *c, AVIOContext *pb,
2402 int codec_tag, int format,
2405 int video_codec_id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2408 (codec_tag != format &&
2409 // AVID 1:1 samples with differing data format and codec tag exist
2410 (codec_tag != AV_RL32("AV1x") || format != AV_RL32("AVup")) &&
2411 // prores is allowed to have differing data format and codec tag
2412 codec_tag != AV_RL32("apcn") && codec_tag != AV_RL32("apch") &&
2414 codec_tag != AV_RL32("dvpp") && codec_tag != AV_RL32("dvcp") &&
2415 (c->fc->video_codec_id ? video_codec_id != c->fc->video_codec_id
2416 : codec_tag != MKTAG('j','p','e','g')))) {
2417 /* Multiple fourcc, we skip JPEG. This is not correct, we should
2418 * export it as a separate AVStream but this needs a few changes
2419 * in the MOV demuxer, patch welcome. */
2421 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
2422 avio_skip(pb, size);
2429 int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries)
2432 MOVStreamContext *sc;
2433 int pseudo_stream_id;
2435 if (c->fc->nb_streams < 1)
2437 st = c->fc->streams[c->fc->nb_streams-1];
2440 for (pseudo_stream_id = 0;
2441 pseudo_stream_id < entries && !pb->eof_reached;
2442 pseudo_stream_id++) {
2443 //Parsing Sample description table
2445 int ret, dref_id = 1;
2446 MOVAtom a = { AV_RL32("stsd") };
2447 int64_t start_pos = avio_tell(pb);
2448 int64_t size = avio_rb32(pb); /* size */
2449 uint32_t format = avio_rl32(pb); /* data format */
2452 avio_rb32(pb); /* reserved */
2453 avio_rb16(pb); /* reserved */
2454 dref_id = avio_rb16(pb);
2455 } else if (size <= 7) {
2456 av_log(c->fc, AV_LOG_ERROR,
2457 "invalid size %"PRId64" in stsd\n", size);
2458 return AVERROR_INVALIDDATA;
2461 if (mov_skip_multiple_stsd(c, pb, st->codecpar->codec_tag, format,
2462 size - (avio_tell(pb) - start_pos)))
2465 sc->pseudo_stream_id = st->codecpar->codec_tag ? -1 : pseudo_stream_id;
2466 sc->dref_id= dref_id;
2467 sc->format = format;
2469 id = mov_codec_id(st, format);
2471 av_log(c->fc, AV_LOG_TRACE,
2472 "size=%"PRId64" 4CC=%s codec_type=%d\n", size,
2473 av_fourcc2str(format), st->codecpar->codec_type);
2475 if (st->codecpar->codec_type==AVMEDIA_TYPE_VIDEO) {
2476 st->codecpar->codec_id = id;
2477 mov_parse_stsd_video(c, pb, st, sc);
2478 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_AUDIO) {
2479 st->codecpar->codec_id = id;
2480 mov_parse_stsd_audio(c, pb, st, sc);
2481 if (st->codecpar->sample_rate < 0) {
2482 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
2483 return AVERROR_INVALIDDATA;
2485 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_SUBTITLE){
2486 st->codecpar->codec_id = id;
2487 mov_parse_stsd_subtitle(c, pb, st, sc,
2488 size - (avio_tell(pb) - start_pos));
2490 ret = mov_parse_stsd_data(c, pb, st, sc,
2491 size - (avio_tell(pb) - start_pos));
2495 /* this will read extra atoms at the end (wave, alac, damr, avcC, hvcC, SMI ...) */
2496 a.size = size - (avio_tell(pb) - start_pos);
2498 if ((ret = mov_read_default(c, pb, a)) < 0)
2500 } else if (a.size > 0)
2501 avio_skip(pb, a.size);
2503 if (sc->extradata && st->codecpar->extradata) {
2504 int extra_size = st->codecpar->extradata_size;
2506 /* Move the current stream extradata to the stream context one. */
2507 sc->extradata_size[pseudo_stream_id] = extra_size;
2508 sc->extradata[pseudo_stream_id] = av_malloc(extra_size + AV_INPUT_BUFFER_PADDING_SIZE);
2509 if (!sc->extradata[pseudo_stream_id])
2510 return AVERROR(ENOMEM);
2511 memcpy(sc->extradata[pseudo_stream_id], st->codecpar->extradata, extra_size);
2512 av_freep(&st->codecpar->extradata);
2513 st->codecpar->extradata_size = 0;
2517 if (pb->eof_reached)
2523 static int mov_read_stsd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2526 MOVStreamContext *sc;
2529 if (c->fc->nb_streams < 1)
2531 st = c->fc->streams[c->fc->nb_streams - 1];
2534 avio_r8(pb); /* version */
2535 avio_rb24(pb); /* flags */
2536 entries = avio_rb32(pb);
2539 av_log(c->fc, AV_LOG_ERROR, "invalid STSD entries %d\n", entries);
2540 return AVERROR_INVALIDDATA;
2543 if (sc->extradata) {
2544 av_log(c->fc, AV_LOG_ERROR,
2545 "Duplicate stsd found in this track.\n");
2546 return AVERROR_INVALIDDATA;
2549 /* Prepare space for hosting multiple extradata. */
2550 sc->extradata = av_mallocz_array(entries, sizeof(*sc->extradata));
2552 return AVERROR(ENOMEM);
2554 sc->extradata_size = av_mallocz_array(entries, sizeof(*sc->extradata_size));
2555 if (!sc->extradata_size) {
2556 ret = AVERROR(ENOMEM);
2560 ret = ff_mov_read_stsd_entries(c, pb, entries);
2564 sc->stsd_count = entries;
2566 /* Restore back the primary extradata. */
2567 av_freep(&st->codecpar->extradata);
2568 st->codecpar->extradata_size = sc->extradata_size[0];
2569 if (sc->extradata_size[0]) {
2570 st->codecpar->extradata = av_mallocz(sc->extradata_size[0] + AV_INPUT_BUFFER_PADDING_SIZE);
2571 if (!st->codecpar->extradata)
2572 return AVERROR(ENOMEM);
2573 memcpy(st->codecpar->extradata, sc->extradata[0], sc->extradata_size[0]);
2576 return mov_finalize_stsd_codec(c, pb, st, sc);
2578 av_freep(&sc->extradata);
2579 av_freep(&sc->extradata_size);
2583 static int mov_read_stsc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2586 MOVStreamContext *sc;
2587 unsigned int i, entries;
2589 if (c->fc->nb_streams < 1)
2591 st = c->fc->streams[c->fc->nb_streams-1];
2594 avio_r8(pb); /* version */
2595 avio_rb24(pb); /* flags */
2597 entries = avio_rb32(pb);
2599 av_log(c->fc, AV_LOG_TRACE, "track[%u].stsc.entries = %u\n", c->fc->nb_streams - 1, entries);
2604 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSC atom\n");
2605 av_free(sc->stsc_data);
2607 sc->stsc_data = av_malloc_array(entries, sizeof(*sc->stsc_data));
2609 return AVERROR(ENOMEM);
2611 for (i = 0; i < entries && !pb->eof_reached; i++) {
2612 sc->stsc_data[i].first = avio_rb32(pb);
2613 sc->stsc_data[i].count = avio_rb32(pb);
2614 sc->stsc_data[i].id = avio_rb32(pb);
2619 if (pb->eof_reached)
2625 static inline int mov_stsc_index_valid(unsigned int index, unsigned int count)
2627 return index < count - 1;
2630 /* Compute the samples value for the stsc entry at the given index. */
2631 static inline int mov_get_stsc_samples(MOVStreamContext *sc, unsigned int index)
2635 if (mov_stsc_index_valid(index, sc->stsc_count))
2636 chunk_count = sc->stsc_data[index + 1].first - sc->stsc_data[index].first;
2638 chunk_count = sc->chunk_count - (sc->stsc_data[index].first - 1);
2640 return sc->stsc_data[index].count * chunk_count;
2643 static int mov_read_stps(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2646 MOVStreamContext *sc;
2647 unsigned i, entries;
2649 if (c->fc->nb_streams < 1)
2651 st = c->fc->streams[c->fc->nb_streams-1];
2654 avio_rb32(pb); // version + flags
2656 entries = avio_rb32(pb);
2658 av_log(c->fc, AV_LOG_WARNING, "Duplicated STPS atom\n");
2659 av_free(sc->stps_data);
2661 sc->stps_data = av_malloc_array(entries, sizeof(*sc->stps_data));
2663 return AVERROR(ENOMEM);
2665 for (i = 0; i < entries && !pb->eof_reached; i++) {
2666 sc->stps_data[i] = avio_rb32(pb);
2671 if (pb->eof_reached)
2677 static int mov_read_stss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2680 MOVStreamContext *sc;
2681 unsigned int i, entries;
2683 if (c->fc->nb_streams < 1)
2685 st = c->fc->streams[c->fc->nb_streams-1];
2688 avio_r8(pb); /* version */
2689 avio_rb24(pb); /* flags */
2691 entries = avio_rb32(pb);
2693 av_log(c->fc, AV_LOG_TRACE, "keyframe_count = %u\n", entries);
2697 sc->keyframe_absent = 1;
2698 if (!st->need_parsing && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
2699 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2703 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSS atom\n");
2704 if (entries >= UINT_MAX / sizeof(int))
2705 return AVERROR_INVALIDDATA;
2706 av_freep(&sc->keyframes);
2707 sc->keyframe_count = 0;
2708 sc->keyframes = av_malloc_array(entries, sizeof(*sc->keyframes));
2710 return AVERROR(ENOMEM);
2712 for (i = 0; i < entries && !pb->eof_reached; i++) {
2713 sc->keyframes[i] = avio_rb32(pb);
2716 sc->keyframe_count = i;
2718 if (pb->eof_reached)
2724 static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2727 MOVStreamContext *sc;
2728 unsigned int i, entries, sample_size, field_size, num_bytes;
2733 if (c->fc->nb_streams < 1)
2735 st = c->fc->streams[c->fc->nb_streams-1];
2738 avio_r8(pb); /* version */
2739 avio_rb24(pb); /* flags */
2741 if (atom.type == MKTAG('s','t','s','z')) {
2742 sample_size = avio_rb32(pb);
2743 if (!sc->sample_size) /* do not overwrite value computed in stsd */
2744 sc->sample_size = sample_size;
2745 sc->stsz_sample_size = sample_size;
2749 avio_rb24(pb); /* reserved */
2750 field_size = avio_r8(pb);
2752 entries = avio_rb32(pb);
2754 av_log(c->fc, AV_LOG_TRACE, "sample_size = %u sample_count = %u\n", sc->sample_size, entries);
2756 sc->sample_count = entries;
2760 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
2761 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %u\n", field_size);
2762 return AVERROR_INVALIDDATA;
2767 if (entries >= (UINT_MAX - 4) / field_size)
2768 return AVERROR_INVALIDDATA;
2769 if (sc->sample_sizes)
2770 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSZ atom\n");
2771 av_free(sc->sample_sizes);
2772 sc->sample_count = 0;
2773 sc->sample_sizes = av_malloc_array(entries, sizeof(*sc->sample_sizes));
2774 if (!sc->sample_sizes)
2775 return AVERROR(ENOMEM);
2777 num_bytes = (entries*field_size+4)>>3;
2779 buf = av_malloc(num_bytes+AV_INPUT_BUFFER_PADDING_SIZE);
2781 av_freep(&sc->sample_sizes);
2782 return AVERROR(ENOMEM);
2785 ret = ffio_read_size(pb, buf, num_bytes);
2787 av_freep(&sc->sample_sizes);
2792 init_get_bits(&gb, buf, 8*num_bytes);
2794 for (i = 0; i < entries && !pb->eof_reached; i++) {
2795 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
2796 sc->data_size += sc->sample_sizes[i];
2799 sc->sample_count = i;
2803 if (pb->eof_reached)
2809 static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2812 MOVStreamContext *sc;
2813 unsigned int i, entries;
2815 int64_t total_sample_count=0;
2817 if (c->fc->nb_streams < 1)
2819 st = c->fc->streams[c->fc->nb_streams-1];
2822 avio_r8(pb); /* version */
2823 avio_rb24(pb); /* flags */
2824 entries = avio_rb32(pb);
2826 av_log(c->fc, AV_LOG_TRACE, "track[%u].stts.entries = %u\n",
2827 c->fc->nb_streams-1, entries);
2830 av_log(c->fc, AV_LOG_WARNING, "Duplicated STTS atom\n");
2831 av_free(sc->stts_data);
2833 sc->stts_data = av_malloc_array(entries, sizeof(*sc->stts_data));
2835 return AVERROR(ENOMEM);
2837 for (i = 0; i < entries && !pb->eof_reached; i++) {
2838 int sample_duration;
2839 unsigned int sample_count;
2841 sample_count=avio_rb32(pb);
2842 sample_duration = avio_rb32(pb);
2844 sc->stts_data[i].count= sample_count;
2845 sc->stts_data[i].duration= sample_duration;
2847 av_log(c->fc, AV_LOG_TRACE, "sample_count=%d, sample_duration=%d\n",
2848 sample_count, sample_duration);
2852 && sample_count == 1
2853 && total_sample_count > 100
2854 && sample_duration/10 > duration / total_sample_count)
2855 sample_duration = duration / total_sample_count;
2856 duration+=(int64_t)sample_duration*sample_count;
2857 total_sample_count+=sample_count;
2862 sc->duration_for_fps += duration;
2863 sc->nb_frames_for_fps += total_sample_count;
2865 if (pb->eof_reached)
2868 st->nb_frames= total_sample_count;
2870 st->duration= duration;
2871 sc->track_end = duration;
2875 static void mov_update_dts_shift(MOVStreamContext *sc, int duration)
2878 if (duration == INT_MIN) {
2879 av_log(NULL, AV_LOG_WARNING, "mov_update_dts_shift(): dts_shift set to %d\n", INT_MAX);
2882 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
2886 static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2889 MOVStreamContext *sc;
2890 unsigned int i, j, entries, ctts_count = 0;
2892 if (c->fc->nb_streams < 1)
2894 st = c->fc->streams[c->fc->nb_streams-1];
2897 avio_r8(pb); /* version */
2898 avio_rb24(pb); /* flags */
2899 entries = avio_rb32(pb);
2901 av_log(c->fc, AV_LOG_TRACE, "track[%u].ctts.entries = %u\n", c->fc->nb_streams - 1, entries);
2905 if (entries >= UINT_MAX / sizeof(*sc->ctts_data))
2906 return AVERROR_INVALIDDATA;
2907 av_freep(&sc->ctts_data);
2908 sc->ctts_data = av_fast_realloc(NULL, &sc->ctts_allocated_size, entries * sizeof(*sc->ctts_data));
2910 return AVERROR(ENOMEM);
2912 for (i = 0; i < entries && !pb->eof_reached; i++) {
2913 int count =avio_rb32(pb);
2914 int duration =avio_rb32(pb);
2917 av_log(c->fc, AV_LOG_TRACE,
2918 "ignoring CTTS entry with count=%d duration=%d\n",
2923 /* Expand entries such that we have a 1-1 mapping with samples. */
2924 for (j = 0; j < count; j++)
2925 add_ctts_entry(&sc->ctts_data, &ctts_count, &sc->ctts_allocated_size, 1, duration);
2927 av_log(c->fc, AV_LOG_TRACE, "count=%d, duration=%d\n",
2930 if (FFNABS(duration) < -(1<<28) && i+2<entries) {
2931 av_log(c->fc, AV_LOG_WARNING, "CTTS invalid\n");
2932 av_freep(&sc->ctts_data);
2938 mov_update_dts_shift(sc, duration);
2941 sc->ctts_count = ctts_count;
2943 if (pb->eof_reached)
2946 av_log(c->fc, AV_LOG_TRACE, "dts shift %d\n", sc->dts_shift);
2951 static int mov_read_sbgp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2954 MOVStreamContext *sc;
2955 unsigned int i, entries;
2957 uint32_t grouping_type;
2959 if (c->fc->nb_streams < 1)
2961 st = c->fc->streams[c->fc->nb_streams-1];
2964 version = avio_r8(pb); /* version */
2965 avio_rb24(pb); /* flags */
2966 grouping_type = avio_rl32(pb);
2967 if (grouping_type != MKTAG( 'r','a','p',' '))
2968 return 0; /* only support 'rap ' grouping */
2970 avio_rb32(pb); /* grouping_type_parameter */
2972 entries = avio_rb32(pb);
2976 av_log(c->fc, AV_LOG_WARNING, "Duplicated SBGP atom\n");
2977 av_free(sc->rap_group);
2978 sc->rap_group_count = 0;
2979 sc->rap_group = av_malloc_array(entries, sizeof(*sc->rap_group));
2981 return AVERROR(ENOMEM);
2983 for (i = 0; i < entries && !pb->eof_reached; i++) {
2984 sc->rap_group[i].count = avio_rb32(pb); /* sample_count */
2985 sc->rap_group[i].index = avio_rb32(pb); /* group_description_index */
2988 sc->rap_group_count = i;
2990 return pb->eof_reached ? AVERROR_EOF : 0;
2994 * Get ith edit list entry (media time, duration).
2996 static int get_edit_list_entry(MOVContext *mov,
2997 const MOVStreamContext *msc,
2998 unsigned int edit_list_index,
2999 int64_t *edit_list_media_time,
3000 int64_t *edit_list_duration,
3001 int64_t global_timescale)
3003 if (edit_list_index == msc->elst_count) {
3006 *edit_list_media_time = msc->elst_data[edit_list_index].time;
3007 *edit_list_duration = msc->elst_data[edit_list_index].duration;
3009 /* duration is in global timescale units;convert to msc timescale */
3010 if (global_timescale == 0) {
3011 avpriv_request_sample(mov->fc, "Support for mvhd.timescale = 0 with editlists");
3014 *edit_list_duration = av_rescale(*edit_list_duration, msc->time_scale,
3020 * Find the closest previous frame to the timestamp_pts, in e_old index
3021 * entries. Searching for just any frame / just key frames can be controlled by
3022 * last argument 'flag'.
3023 * Note that if ctts_data is not NULL, we will always search for a key frame
3024 * irrespective of the value of 'flag'. If we don't find any keyframe, we will
3025 * return the first frame of the video.
3027 * Here the timestamp_pts is considered to be a presentation timestamp and
3028 * the timestamp of index entries are considered to be decoding timestamps.
3030 * Returns 0 if successful in finding a frame, else returns -1.
3031 * Places the found index corresponding output arg.
3033 * If ctts_old is not NULL, then refines the searched entry by searching
3034 * backwards from the found timestamp, to find the frame with correct PTS.
3036 * Places the found ctts_index and ctts_sample in corresponding output args.
3038 static int find_prev_closest_index(AVStream *st,
3039 AVIndexEntry *e_old,
3043 int64_t timestamp_pts,
3046 int64_t* ctts_index,
3047 int64_t* ctts_sample)
3049 MOVStreamContext *msc = st->priv_data;
3050 AVIndexEntry *e_keep = st->index_entries;
3051 int nb_keep = st->nb_index_entries;
3053 int64_t index_ctts_count;
3057 // If dts_shift > 0, then all the index timestamps will have to be offset by
3058 // at least dts_shift amount to obtain PTS.
3059 // Hence we decrement the searched timestamp_pts by dts_shift to find the closest index element.
3060 if (msc->dts_shift > 0) {
3061 timestamp_pts -= msc->dts_shift;
3064 st->index_entries = e_old;
3065 st->nb_index_entries = nb_old;
3066 *index = av_index_search_timestamp(st, timestamp_pts, flag | AVSEEK_FLAG_BACKWARD);
3068 // Keep going backwards in the index entries until the timestamp is the same.
3070 for (i = *index; i > 0 && e_old[i].timestamp == e_old[i - 1].timestamp;
3072 if ((flag & AVSEEK_FLAG_ANY) ||
3073 (e_old[i - 1].flags & AVINDEX_KEYFRAME)) {
3079 // If we have CTTS then refine the search, by searching backwards over PTS
3080 // computed by adding corresponding CTTS durations to index timestamps.
3081 if (ctts_data && *index >= 0) {
3082 av_assert0(ctts_index);
3083 av_assert0(ctts_sample);
3084 // Find out the ctts_index for the found frame.
3087 for (index_ctts_count = 0; index_ctts_count < *index; index_ctts_count++) {
3088 if (*ctts_index < ctts_count) {
3090 if (ctts_data[*ctts_index].count == *ctts_sample) {
3097 while (*index >= 0 && (*ctts_index) >= 0) {
3098 // Find a "key frame" with PTS <= timestamp_pts (So that we can decode B-frames correctly).
3099 // No need to add dts_shift to the timestamp here becase timestamp_pts has already been
3100 // compensated by dts_shift above.
3101 if ((e_old[*index].timestamp + ctts_data[*ctts_index].duration) <= timestamp_pts &&
3102 (e_old[*index].flags & AVINDEX_KEYFRAME)) {
3107 if (*ctts_sample == 0) {
3109 if (*ctts_index >= 0)
3110 *ctts_sample = ctts_data[*ctts_index].count - 1;
3117 /* restore AVStream state*/
3118 st->index_entries = e_keep;
3119 st->nb_index_entries = nb_keep;
3120 return *index >= 0 ? 0 : -1;
3124 * Add index entry with the given values, to the end of st->index_entries.
3125 * Returns the new size st->index_entries if successful, else returns -1.
3127 * This function is similar to ff_add_index_entry in libavformat/utils.c
3128 * except that here we are always unconditionally adding an index entry to
3129 * the end, instead of searching the entries list and skipping the add if
3130 * there is an existing entry with the same timestamp.
3131 * This is needed because the mov_fix_index calls this func with the same
3132 * unincremented timestamp for successive discarded frames.
3134 static int64_t add_index_entry(AVStream *st, int64_t pos, int64_t timestamp,
3135 int size, int distance, int flags)
3137 AVIndexEntry *entries, *ie;
3139 const size_t min_size_needed = (st->nb_index_entries + 1) * sizeof(AVIndexEntry);
3141 // Double the allocation each time, to lower memory fragmentation.
3142 // Another difference from ff_add_index_entry function.
3143 const size_t requested_size =
3144 min_size_needed > st->index_entries_allocated_size ?
3145 FFMAX(min_size_needed, 2 * st->index_entries_allocated_size) :
3148 if((unsigned)st->nb_index_entries + 1 >= UINT_MAX / sizeof(AVIndexEntry))
3151 entries = av_fast_realloc(st->index_entries,
3152 &st->index_entries_allocated_size,
3157 st->index_entries= entries;
3159 index= st->nb_index_entries++;
3160 ie= &entries[index];
3163 ie->timestamp = timestamp;
3164 ie->min_distance= distance;
3171 * Rewrite timestamps of index entries in the range [end_index - frame_duration_buffer_size, end_index)
3172 * by subtracting end_ts successively by the amounts given in frame_duration_buffer.
3174 static void fix_index_entry_timestamps(AVStream* st, int end_index, int64_t end_ts,
3175 int64_t* frame_duration_buffer,
3176 int frame_duration_buffer_size) {
3178 av_assert0(end_index >= 0 && end_index <= st->nb_index_entries);
3179 for (i = 0; i < frame_duration_buffer_size; i++) {
3180 end_ts -= frame_duration_buffer[frame_duration_buffer_size - 1 - i];
3181 st->index_entries[end_index - 1 - i].timestamp = end_ts;
3186 * Append a new ctts entry to ctts_data.
3187 * Returns the new ctts_count if successful, else returns -1.
3189 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
3190 int count, int duration)
3192 MOVStts *ctts_buf_new;
3193 const size_t min_size_needed = (*ctts_count + 1) * sizeof(MOVStts);
3194 const size_t requested_size =
3195 min_size_needed > *allocated_size ?
3196 FFMAX(min_size_needed, 2 * (*allocated_size)) :
3199 if((unsigned)(*ctts_count) + 1 >= UINT_MAX / sizeof(MOVStts))
3202 ctts_buf_new = av_fast_realloc(*ctts_data, allocated_size, requested_size);
3207 *ctts_data = ctts_buf_new;
3209 ctts_buf_new[*ctts_count].count = count;
3210 ctts_buf_new[*ctts_count].duration = duration;
3212 *ctts_count = (*ctts_count) + 1;
3216 static void mov_current_sample_inc(MOVStreamContext *sc)
3218 sc->current_sample++;
3219 sc->current_index++;
3220 if (sc->index_ranges &&
3221 sc->current_index >= sc->current_index_range->end &&
3222 sc->current_index_range->end) {
3223 sc->current_index_range++;
3224 sc->current_index = sc->current_index_range->start;
3228 static void mov_current_sample_dec(MOVStreamContext *sc)
3230 sc->current_sample--;
3231 sc->current_index--;
3232 if (sc->index_ranges &&
3233 sc->current_index < sc->current_index_range->start &&
3234 sc->current_index_range > sc->index_ranges) {
3235 sc->current_index_range--;
3236 sc->current_index = sc->current_index_range->end - 1;
3240 static void mov_current_sample_set(MOVStreamContext *sc, int current_sample)
3244 sc->current_sample = current_sample;
3245 sc->current_index = current_sample;
3246 if (!sc->index_ranges) {
3250 for (sc->current_index_range = sc->index_ranges;
3251 sc->current_index_range->end;
3252 sc->current_index_range++) {
3253 range_size = sc->current_index_range->end - sc->current_index_range->start;
3254 if (range_size > current_sample) {
3255 sc->current_index = sc->current_index_range->start + current_sample;
3258 current_sample -= range_size;
3263 * Fix st->index_entries, so that it contains only the entries (and the entries
3264 * which are needed to decode them) that fall in the edit list time ranges.
3265 * Also fixes the timestamps of the index entries to match the timeline
3266 * specified the edit lists.
3268 static void mov_fix_index(MOVContext *mov, AVStream *st)
3270 MOVStreamContext *msc = st->priv_data;
3271 AVIndexEntry *e_old = st->index_entries;
3272 int nb_old = st->nb_index_entries;
3273 const AVIndexEntry *e_old_end = e_old + nb_old;
3274 const AVIndexEntry *current = NULL;
3275 MOVStts *ctts_data_old = msc->ctts_data;
3276 int64_t ctts_index_old = 0;
3277 int64_t ctts_sample_old = 0;
3278 int64_t ctts_count_old = msc->ctts_count;
3279 int64_t edit_list_media_time = 0;
3280 int64_t edit_list_duration = 0;
3281 int64_t frame_duration = 0;
3282 int64_t edit_list_dts_counter = 0;
3283 int64_t edit_list_dts_entry_end = 0;
3284 int64_t edit_list_start_ctts_sample = 0;
3286 int64_t curr_ctts = 0;
3287 int64_t min_corrected_pts = -1;
3288 int64_t empty_edits_sum_duration = 0;
3289 int64_t edit_list_index = 0;
3292 int64_t start_dts = 0;
3293 int64_t edit_list_start_encountered = 0;
3294 int64_t search_timestamp = 0;
3295 int64_t* frame_duration_buffer = NULL;
3296 int num_discarded_begin = 0;
3297 int first_non_zero_audio_edit = -1;
3298 int packet_skip_samples = 0;
3299 MOVIndexRange *current_index_range;
3302 if (!msc->elst_data || msc->elst_count <= 0 || nb_old <= 0) {
3306 // allocate the index ranges array
3307 msc->index_ranges = av_malloc((msc->elst_count + 1) * sizeof(msc->index_ranges[0]));
3308 if (!msc->index_ranges) {
3309 av_log(mov->fc, AV_LOG_ERROR, "Cannot allocate index ranges buffer\n");
3312 msc->current_index_range = msc->index_ranges;
3313 current_index_range = msc->index_ranges - 1;
3315 // Clean AVStream from traces of old index
3316 st->index_entries = NULL;
3317 st->index_entries_allocated_size = 0;
3318 st->nb_index_entries = 0;
3320 // Clean ctts fields of MOVStreamContext
3321 msc->ctts_data = NULL;
3322 msc->ctts_count = 0;
3323 msc->ctts_index = 0;
3324 msc->ctts_sample = 0;
3325 msc->ctts_allocated_size = 0;
3327 // If the dts_shift is positive (in case of negative ctts values in mov),
3328 // then negate the DTS by dts_shift
3329 if (msc->dts_shift > 0) {
3330 edit_list_dts_entry_end -= msc->dts_shift;
3331 av_log(mov->fc, AV_LOG_DEBUG, "Shifting DTS by %d because of negative CTTS.\n", msc->dts_shift);
3334 start_dts = edit_list_dts_entry_end;
3336 while (get_edit_list_entry(mov, msc, edit_list_index, &edit_list_media_time,
3337 &edit_list_duration, mov->time_scale)) {
3338 av_log(mov->fc, AV_LOG_DEBUG, "Processing st: %d, edit list %"PRId64" - media time: %"PRId64", duration: %"PRId64"\n",
3339 st->index, edit_list_index, edit_list_media_time, edit_list_duration);
3341 edit_list_dts_counter = edit_list_dts_entry_end;
3342 edit_list_dts_entry_end += edit_list_duration;
3343 num_discarded_begin = 0;
3344 if (edit_list_media_time == -1) {
3345 empty_edits_sum_duration += edit_list_duration;
3349 // If we encounter a non-negative edit list reset the skip_samples/start_pad fields and set them
3350 // according to the edit list below.
3351 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3352 if (first_non_zero_audio_edit < 0) {
3353 first_non_zero_audio_edit = 1;
3355 first_non_zero_audio_edit = 0;
3358 if (first_non_zero_audio_edit > 0)
3359 st->skip_samples = msc->start_pad = 0;
3362 // While reordering frame index according to edit list we must handle properly
3363 // the scenario when edit list entry starts from none key frame.
3364 // We find closest previous key frame and preserve it and consequent frames in index.
3365 // All frames which are outside edit list entry time boundaries will be dropped after decoding.
3366 search_timestamp = edit_list_media_time;
3367 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3368 // Audio decoders like AAC need need a decoder delay samples previous to the current sample,
3369 // to correctly decode this frame. Hence for audio we seek to a frame 1 sec. before the
3370 // edit_list_media_time to cover the decoder delay.
3371 search_timestamp = FFMAX(search_timestamp - msc->time_scale, e_old[0].timestamp);
3374 if (find_prev_closest_index(st, e_old, nb_old, ctts_data_old, ctts_count_old, search_timestamp, 0,
3375 &index, &ctts_index_old, &ctts_sample_old) < 0) {
3376 av_log(mov->fc, AV_LOG_WARNING,
3377 "st: %d edit list: %"PRId64" Missing key frame while searching for timestamp: %"PRId64"\n",
3378 st->index, edit_list_index, search_timestamp);
3379 if (find_prev_closest_index(st, e_old, nb_old, ctts_data_old, ctts_count_old, search_timestamp, AVSEEK_FLAG_ANY,
3380 &index, &ctts_index_old, &ctts_sample_old) < 0) {
3381 av_log(mov->fc, AV_LOG_WARNING,
3382 "st: %d edit list %"PRId64" Cannot find an index entry before timestamp: %"PRId64".\n"
3383 "Rounding edit list media time to zero.\n",
3384 st->index, edit_list_index, search_timestamp);
3387 ctts_sample_old = 0;
3388 edit_list_media_time = 0;
3391 current = e_old + index;
3392 edit_list_start_ctts_sample = ctts_sample_old;
3394 // Iterate over index and arrange it according to edit list
3395 edit_list_start_encountered = 0;
3396 for (; current < e_old_end; current++, index++) {
3397 // check if frame outside edit list mark it for discard
3398 frame_duration = (current + 1 < e_old_end) ?
3399 ((current + 1)->timestamp - current->timestamp) : edit_list_duration;
3401 flags = current->flags;
3403 // frames (pts) before or after edit list
3404 curr_cts = current->timestamp + msc->dts_shift;
3407 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3408 curr_ctts = ctts_data_old[ctts_index_old].duration;
3409 av_log(mov->fc, AV_LOG_DEBUG, "stts: %"PRId64" ctts: %"PRId64", ctts_index: %"PRId64", ctts_count: %"PRId64"\n",
3410 curr_cts, curr_ctts, ctts_index_old, ctts_count_old);
3411 curr_cts += curr_ctts;
3413 if (ctts_sample_old == ctts_data_old[ctts_index_old].count) {
3414 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3415 &msc->ctts_allocated_size,
3416 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3417 ctts_data_old[ctts_index_old].duration) == -1) {
3418 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3420 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3421 ctts_data_old[ctts_index_old].duration);
3425 ctts_sample_old = 0;
3426 edit_list_start_ctts_sample = 0;
3430 if (curr_cts < edit_list_media_time || curr_cts >= (edit_list_duration + edit_list_media_time)) {
3431 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id != AV_CODEC_ID_VORBIS &&
3432 curr_cts < edit_list_media_time && curr_cts + frame_duration > edit_list_media_time &&
3433 first_non_zero_audio_edit > 0) {
3434 packet_skip_samples = edit_list_media_time - curr_cts;
3435 st->skip_samples += packet_skip_samples;
3437 // Shift the index entry timestamp by packet_skip_samples to be correct.
3438 edit_list_dts_counter -= packet_skip_samples;
3439 if (edit_list_start_encountered == 0) {
3440 edit_list_start_encountered = 1;
3441 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3442 // discarded packets.
3443 if (frame_duration_buffer) {
3444 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3445 frame_duration_buffer, num_discarded_begin);
3446 av_freep(&frame_duration_buffer);
3450 av_log(mov->fc, AV_LOG_DEBUG, "skip %d audio samples from curr_cts: %"PRId64"\n", packet_skip_samples, curr_cts);
3452 flags |= AVINDEX_DISCARD_FRAME;
3453 av_log(mov->fc, AV_LOG_DEBUG, "drop a frame at curr_cts: %"PRId64" @ %"PRId64"\n", curr_cts, index);
3455 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && edit_list_start_encountered == 0) {
3456 num_discarded_begin++;
3457 frame_duration_buffer = av_realloc(frame_duration_buffer,
3458 num_discarded_begin * sizeof(int64_t));
3459 if (!frame_duration_buffer) {
3460 av_log(mov->fc, AV_LOG_ERROR, "Cannot reallocate frame duration buffer\n");
3463 frame_duration_buffer[num_discarded_begin - 1] = frame_duration;
3465 // Increment skip_samples for the first non-zero audio edit list
3466 if (first_non_zero_audio_edit > 0 && st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
3467 st->skip_samples += frame_duration;
3472 if (min_corrected_pts < 0) {
3473 min_corrected_pts = edit_list_dts_counter + curr_ctts + msc->dts_shift;
3475 min_corrected_pts = FFMIN(min_corrected_pts, edit_list_dts_counter + curr_ctts + msc->dts_shift);
3477 if (edit_list_start_encountered == 0) {
3478 edit_list_start_encountered = 1;
3479 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3480 // discarded packets.
3481 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && frame_duration_buffer) {
3482 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3483 frame_duration_buffer, num_discarded_begin);
3484 av_freep(&frame_duration_buffer);
3489 if (add_index_entry(st, current->pos, edit_list_dts_counter, current->size,
3490 current->min_distance, flags) == -1) {
3491 av_log(mov->fc, AV_LOG_ERROR, "Cannot add index entry\n");
3495 // Update the index ranges array
3496 if (current_index_range < msc->index_ranges || index != current_index_range->end) {
3497 current_index_range++;
3498 current_index_range->start = index;
3500 current_index_range->end = index + 1;
3502 // Only start incrementing DTS in frame_duration amounts, when we encounter a frame in edit list.
3503 if (edit_list_start_encountered > 0) {
3504 edit_list_dts_counter = edit_list_dts_counter + frame_duration;
3507 // Break when found first key frame after edit entry completion
3508 if (((curr_cts + frame_duration) >= (edit_list_duration + edit_list_media_time)) &&
3509 ((flags & AVINDEX_KEYFRAME) || ((st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)))) {
3511 if (ctts_data_old && ctts_sample_old != 0) {
3512 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3513 &msc->ctts_allocated_size,
3514 ctts_sample_old - edit_list_start_ctts_sample,
3515 ctts_data_old[ctts_index_old].duration) == -1) {
3516 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3517 ctts_index_old, ctts_sample_old - edit_list_start_ctts_sample,
3518 ctts_data_old[ctts_index_old].duration);
3526 // If there are empty edits, then min_corrected_pts might be positive intentionally. So we subtract the
3527 // sum duration of emtpy edits here.
3528 min_corrected_pts -= empty_edits_sum_duration;
3530 // If the minimum pts turns out to be greater than zero after fixing the index, then we subtract the
3531 // dts by that amount to make the first pts zero.
3532 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && min_corrected_pts > 0) {
3533 av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by %"PRId64" to make first pts zero.\n", min_corrected_pts);
3534 for (i = 0; i < st->nb_index_entries; ++i) {
3535 st->index_entries[i].timestamp -= min_corrected_pts;
3539 // Update av stream length
3540 st->duration = edit_list_dts_entry_end - start_dts;
3541 msc->start_pad = st->skip_samples;
3543 // Free the old index and the old CTTS structures
3545 av_free(ctts_data_old);
3547 // Null terminate the index ranges array
3548 current_index_range++;
3549 current_index_range->start = 0;
3550 current_index_range->end = 0;
3551 msc->current_index = msc->index_ranges[0].start;
3554 static void mov_build_index(MOVContext *mov, AVStream *st)
3556 MOVStreamContext *sc = st->priv_data;
3557 int64_t current_offset;
3558 int64_t current_dts = 0;
3559 unsigned int stts_index = 0;
3560 unsigned int stsc_index = 0;
3561 unsigned int stss_index = 0;
3562 unsigned int stps_index = 0;
3564 uint64_t stream_size = 0;
3566 if (sc->elst_count) {
3567 int i, edit_start_index = 0, multiple_edits = 0;
3568 int64_t empty_duration = 0; // empty duration of the first edit list entry
3569 int64_t start_time = 0; // start time of the media
3571 for (i = 0; i < sc->elst_count; i++) {
3572 const MOVElst *e = &sc->elst_data[i];
3573 if (i == 0 && e->time == -1) {
3574 /* if empty, the first entry is the start time of the stream
3575 * relative to the presentation itself */
3576 empty_duration = e->duration;
3577 edit_start_index = 1;
3578 } else if (i == edit_start_index && e->time >= 0) {
3579 start_time = e->time;
3585 if (multiple_edits && !mov->advanced_editlist)
3586 av_log(mov->fc, AV_LOG_WARNING, "multiple edit list entries, "
3587 "Use -advanced_editlist to correctly decode otherwise "
3588 "a/v desync might occur\n");
3590 /* adjust first dts according to edit list */
3591 if ((empty_duration || start_time) && mov->time_scale > 0) {
3593 empty_duration = av_rescale(empty_duration, sc->time_scale, mov->time_scale);
3594 sc->time_offset = start_time - empty_duration;
3595 if (!mov->advanced_editlist)
3596 current_dts = -sc->time_offset;
3599 if (!multiple_edits && !mov->advanced_editlist &&
3600 st->codecpar->codec_id == AV_CODEC_ID_AAC && start_time > 0)
3601 sc->start_pad = start_time;
3604 /* only use old uncompressed audio chunk demuxing when stts specifies it */
3605 if (!(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3606 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
3607 unsigned int current_sample = 0;
3608 unsigned int stts_sample = 0;
3609 unsigned int sample_size;
3610 unsigned int distance = 0;
3611 unsigned int rap_group_index = 0;
3612 unsigned int rap_group_sample = 0;
3613 int64_t last_dts = 0;
3614 int64_t dts_correction = 0;
3615 int rap_group_present = sc->rap_group_count && sc->rap_group;
3616 int key_off = (sc->keyframe_count && sc->keyframes[0] > 0) || (sc->stps_count && sc->stps_data[0] > 0);
3618 current_dts -= sc->dts_shift;
3619 last_dts = current_dts;
3621 if (!sc->sample_count || st->nb_index_entries)
3623 if (sc->sample_count >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3625 if (av_reallocp_array(&st->index_entries,
3626 st->nb_index_entries + sc->sample_count,
3627 sizeof(*st->index_entries)) < 0) {
3628 st->nb_index_entries = 0;
3631 st->index_entries_allocated_size = (st->nb_index_entries + sc->sample_count) * sizeof(*st->index_entries);
3633 for (i = 0; i < sc->chunk_count; i++) {
3634 int64_t next_offset = i+1 < sc->chunk_count ? sc->chunk_offsets[i+1] : INT64_MAX;
3635 current_offset = sc->chunk_offsets[i];
3636 while (mov_stsc_index_valid(stsc_index, sc->stsc_count) &&
3637 i + 1 == sc->stsc_data[stsc_index + 1].first)
3640 if (next_offset > current_offset && sc->sample_size>0 && sc->sample_size < sc->stsz_sample_size &&
3641 sc->stsc_data[stsc_index].count * (int64_t)sc->stsz_sample_size > next_offset - current_offset) {
3642 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too large), ignoring\n", sc->stsz_sample_size);
3643 sc->stsz_sample_size = sc->sample_size;
3645 if (sc->stsz_sample_size>0 && sc->stsz_sample_size < sc->sample_size) {
3646 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too small), ignoring\n", sc->stsz_sample_size);
3647 sc->stsz_sample_size = sc->sample_size;
3650 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
3652 if (current_sample >= sc->sample_count) {
3653 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
3657 if (!sc->keyframe_absent && (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index])) {
3659 if (stss_index + 1 < sc->keyframe_count)
3661 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
3663 if (stps_index + 1 < sc->stps_count)
3666 if (rap_group_present && rap_group_index < sc->rap_group_count) {
3667 if (sc->rap_group[rap_group_index].index > 0)
3669 if (++rap_group_sample == sc->rap_group[rap_group_index].count) {
3670 rap_group_sample = 0;
3674 if (sc->keyframe_absent
3676 && !rap_group_present
3677 && (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO || (i==0 && j==0)))
3681 sample_size = sc->stsz_sample_size > 0 ? sc->stsz_sample_size : sc->sample_sizes[current_sample];
3682 if (sc->pseudo_stream_id == -1 ||
3683 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
3685 if (sample_size > 0x3FFFFFFF) {
3686 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", sample_size);
3689 e = &st->index_entries[st->nb_index_entries++];
3690 e->pos = current_offset;
3691 e->timestamp = current_dts;
3692 e->size = sample_size;
3693 e->min_distance = distance;
3694 e->flags = keyframe ? AVINDEX_KEYFRAME : 0;
3695 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %u, offset %"PRIx64", dts %"PRId64", "
3696 "size %u, distance %u, keyframe %d\n", st->index, current_sample,
3697 current_offset, current_dts, sample_size, distance, keyframe);
3698 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries < 100)
3699 ff_rfps_add_frame(mov->fc, st, current_dts);
3702 current_offset += sample_size;
3703 stream_size += sample_size;
3705 /* A negative sample duration is invalid based on the spec,
3706 * but some samples need it to correct the DTS. */
3707 if (sc->stts_data[stts_index].duration < 0) {
3708 av_log(mov->fc, AV_LOG_WARNING,
3709 "Invalid SampleDelta %d in STTS, at %d st:%d\n",
3710 sc->stts_data[stts_index].duration, stts_index,
3712 dts_correction += sc->stts_data[stts_index].duration - 1;
3713 sc->stts_data[stts_index].duration = 1;
3715 current_dts += sc->stts_data[stts_index].duration;
3716 if (!dts_correction || current_dts + dts_correction > last_dts) {
3717 current_dts += dts_correction;
3720 /* Avoid creating non-monotonous DTS */
3721 dts_correction += current_dts - last_dts - 1;
3722 current_dts = last_dts + 1;
3724 last_dts = current_dts;
3728 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
3734 if (st->duration > 0)
3735 st->codecpar->bit_rate = stream_size*8*sc->time_scale/st->duration;
3737 unsigned chunk_samples, total = 0;
3739 // compute total chunk count
3740 for (i = 0; i < sc->stsc_count; i++) {
3741 unsigned count, chunk_count;
3743 chunk_samples = sc->stsc_data[i].count;
3744 if (i != sc->stsc_count - 1 &&
3745 sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
3746 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
3750 if (sc->samples_per_frame >= 160) { // gsm
3751 count = chunk_samples / sc->samples_per_frame;
3752 } else if (sc->samples_per_frame > 1) {
3753 unsigned samples = (1024/sc->samples_per_frame)*sc->samples_per_frame;
3754 count = (chunk_samples+samples-1) / samples;
3756 count = (chunk_samples+1023) / 1024;
3759 if (mov_stsc_index_valid(i, sc->stsc_count))
3760 chunk_count = sc->stsc_data[i+1].first - sc->stsc_data[i].first;
3762 chunk_count = sc->chunk_count - (sc->stsc_data[i].first - 1);
3763 total += chunk_count * count;
3766 av_log(mov->fc, AV_LOG_TRACE, "chunk count %u\n", total);
3767 if (total >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3769 if (av_reallocp_array(&st->index_entries,
3770 st->nb_index_entries + total,
3771 sizeof(*st->index_entries)) < 0) {
3772 st->nb_index_entries = 0;
3775 st->index_entries_allocated_size = (st->nb_index_entries + total) * sizeof(*st->index_entries);
3778 for (i = 0; i < sc->chunk_count; i++) {
3779 current_offset = sc->chunk_offsets[i];
3780 if (mov_stsc_index_valid(stsc_index, sc->stsc_count) &&
3781 i + 1 == sc->stsc_data[stsc_index + 1].first)
3783 chunk_samples = sc->stsc_data[stsc_index].count;
3785 while (chunk_samples > 0) {
3787 unsigned size, samples;
3789 if (sc->samples_per_frame > 1 && !sc->bytes_per_frame) {
3790 avpriv_request_sample(mov->fc,
3791 "Zero bytes per frame, but %d samples per frame",
3792 sc->samples_per_frame);
3796 if (sc->samples_per_frame >= 160) { // gsm
3797 samples = sc->samples_per_frame;
3798 size = sc->bytes_per_frame;
3800 if (sc->samples_per_frame > 1) {
3801 samples = FFMIN((1024 / sc->samples_per_frame)*
3802 sc->samples_per_frame, chunk_samples);
3803 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
3805 samples = FFMIN(1024, chunk_samples);
3806 size = samples * sc->sample_size;
3810 if (st->nb_index_entries >= total) {
3811 av_log(mov->fc, AV_LOG_ERROR, "wrong chunk count %u\n", total);
3814 if (size > 0x3FFFFFFF) {
3815 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", size);
3818 e = &st->index_entries[st->nb_index_entries++];
3819 e->pos = current_offset;
3820 e->timestamp = current_dts;
3822 e->min_distance = 0;
3823 e->flags = AVINDEX_KEYFRAME;
3824 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, chunk %u, offset %"PRIx64", dts %"PRId64", "
3825 "size %u, duration %u\n", st->index, i, current_offset, current_dts,
3828 current_offset += size;
3829 current_dts += samples;
3830 chunk_samples -= samples;
3835 if (!mov->ignore_editlist && mov->advanced_editlist) {
3836 // Fix index according to edit lists.
3837 mov_fix_index(mov, st);
3841 static int test_same_origin(const char *src, const char *ref) {
3851 av_url_split(src_proto, sizeof(src_proto), src_auth, sizeof(src_auth), src_host, sizeof(src_host), &src_port, NULL, 0, src);
3852 av_url_split(ref_proto, sizeof(ref_proto), ref_auth, sizeof(ref_auth), ref_host, sizeof(ref_host), &ref_port, NULL, 0, ref);
3854 if (strlen(src) == 0) {
3856 } else if (strlen(src_auth) + 1 >= sizeof(src_auth) ||
3857 strlen(ref_auth) + 1 >= sizeof(ref_auth) ||
3858 strlen(src_host) + 1 >= sizeof(src_host) ||
3859 strlen(ref_host) + 1 >= sizeof(ref_host)) {
3861 } else if (strcmp(src_proto, ref_proto) ||
3862 strcmp(src_auth, ref_auth) ||
3863 strcmp(src_host, ref_host) ||
3864 src_port != ref_port) {
3870 static int mov_open_dref(MOVContext *c, AVIOContext **pb, const char *src, MOVDref *ref)
3872 /* try relative path, we do not try the absolute because it can leak information about our
3873 system to an attacker */
3874 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
3875 char filename[1025];
3876 const char *src_path;
3879 /* find a source dir */
3880 src_path = strrchr(src, '/');
3886 /* find a next level down to target */
3887 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
3888 if (ref->path[l] == '/') {
3889 if (i == ref->nlvl_to - 1)
3895 /* compose filename if next level down to target was found */
3896 if (i == ref->nlvl_to - 1 && src_path - src < sizeof(filename)) {
3897 memcpy(filename, src, src_path - src);
3898 filename[src_path - src] = 0;
3900 for (i = 1; i < ref->nlvl_from; i++)
3901 av_strlcat(filename, "../", sizeof(filename));
3903 av_strlcat(filename, ref->path + l + 1, sizeof(filename));
3904 if (!c->use_absolute_path) {
3905 int same_origin = test_same_origin(src, filename);
3908 av_log(c->fc, AV_LOG_ERROR,
3909 "Reference with mismatching origin, %s not tried for security reasons, "
3910 "set demuxer option use_absolute_path to allow it anyway\n",
3912 return AVERROR(ENOENT);
3915 if(strstr(ref->path + l + 1, "..") ||
3916 strstr(ref->path + l + 1, ":") ||
3917 (ref->nlvl_from > 1 && same_origin < 0) ||
3918 (filename[0] == '/' && src_path == src))
3919 return AVERROR(ENOENT);
3922 if (strlen(filename) + 1 == sizeof(filename))
3923 return AVERROR(ENOENT);
3924 if (!c->fc->io_open(c->fc, pb, filename, AVIO_FLAG_READ, NULL))
3927 } else if (c->use_absolute_path) {
3928 av_log(c->fc, AV_LOG_WARNING, "Using absolute path on user request, "
3929 "this is a possible security issue\n");
3930 if (!c->fc->io_open(c->fc, pb, ref->path, AVIO_FLAG_READ, NULL))
3933 av_log(c->fc, AV_LOG_ERROR,
3934 "Absolute path %s not tried for security reasons, "
3935 "set demuxer option use_absolute_path to allow absolute paths\n",
3939 return AVERROR(ENOENT);
3942 static void fix_timescale(MOVContext *c, MOVStreamContext *sc)
3944 if (sc->time_scale <= 0) {
3945 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", sc->ffindex);
3946 sc->time_scale = c->time_scale;
3947 if (sc->time_scale <= 0)
3952 static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3955 MOVStreamContext *sc;
3958 st = avformat_new_stream(c->fc, NULL);
3959 if (!st) return AVERROR(ENOMEM);
3960 st->id = c->fc->nb_streams;
3961 sc = av_mallocz(sizeof(MOVStreamContext));
3962 if (!sc) return AVERROR(ENOMEM);
3965 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
3966 sc->ffindex = st->index;
3967 c->trak_index = st->index;
3969 if ((ret = mov_read_default(c, pb, atom)) < 0)
3975 if ((sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
3976 (!sc->sample_size && !sc->sample_count))) ||
3977 (!sc->chunk_count && sc->sample_count)) {
3978 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
3983 fix_timescale(c, sc);
3985 avpriv_set_pts_info(st, 64, 1, sc->time_scale);
3987 mov_build_index(c, st);
3989 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
3990 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
3991 if (c->enable_drefs) {
3992 if (mov_open_dref(c, &sc->pb, c->fc->filename, dref) < 0)
3993 av_log(c->fc, AV_LOG_ERROR,
3994 "stream %d, error opening alias: path='%s', dir='%s', "
3995 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
3996 st->index, dref->path, dref->dir, dref->filename,
3997 dref->volume, dref->nlvl_from, dref->nlvl_to);
3999 av_log(c->fc, AV_LOG_WARNING,
4000 "Skipped opening external track: "
4001 "stream %d, alias: path='%s', dir='%s', "
4002 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d."
4003 "Set enable_drefs to allow this.\n",
4004 st->index, dref->path, dref->dir, dref->filename,
4005 dref->volume, dref->nlvl_from, dref->nlvl_to);
4009 sc->pb_is_copied = 1;
4012 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
4013 if (!st->sample_aspect_ratio.num && st->codecpar->width && st->codecpar->height &&
4014 sc->height && sc->width &&
4015 (st->codecpar->width != sc->width || st->codecpar->height != sc->height)) {
4016 st->sample_aspect_ratio = av_d2q(((double)st->codecpar->height * sc->width) /
4017 ((double)st->codecpar->width * sc->height), INT_MAX);
4020 #if FF_API_R_FRAME_RATE
4021 if (sc->stts_count == 1 || (sc->stts_count == 2 && sc->stts_data[1].count == 1))
4022 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
4023 sc->time_scale, sc->stts_data[0].duration, INT_MAX);
4027 // done for ai5q, ai52, ai55, ai1q, ai12 and ai15.
4028 if (!st->codecpar->extradata_size && st->codecpar->codec_id == AV_CODEC_ID_H264 &&
4029 TAG_IS_AVCI(st->codecpar->codec_tag)) {
4030 ret = ff_generate_avci_extradata(st);
4035 switch (st->codecpar->codec_id) {
4036 #if CONFIG_H261_DECODER
4037 case AV_CODEC_ID_H261:
4039 #if CONFIG_H263_DECODER
4040 case AV_CODEC_ID_H263:
4042 #if CONFIG_MPEG4_DECODER
4043 case AV_CODEC_ID_MPEG4:
4045 st->codecpar->width = 0; /* let decoder init width/height */
4046 st->codecpar->height= 0;
4050 // If the duration of the mp3 packets is not constant, then they could need a parser
4051 if (st->codecpar->codec_id == AV_CODEC_ID_MP3
4052 && sc->stts_count > 3
4053 && sc->stts_count*10 > st->nb_frames
4054 && sc->time_scale == st->codecpar->sample_rate) {
4055 st->need_parsing = AVSTREAM_PARSE_FULL;
4057 /* Do not need those anymore. */
4058 av_freep(&sc->chunk_offsets);
4059 av_freep(&sc->sample_sizes);
4060 av_freep(&sc->keyframes);
4061 av_freep(&sc->stts_data);
4062 av_freep(&sc->stps_data);
4063 av_freep(&sc->elst_data);
4064 av_freep(&sc->rap_group);
4069 static int mov_read_ilst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4072 c->itunes_metadata = 1;
4073 ret = mov_read_default(c, pb, atom);
4074 c->itunes_metadata = 0;
4078 static int mov_read_keys(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4087 count = avio_rb32(pb);
4088 if (count > UINT_MAX / sizeof(*c->meta_keys) - 1) {
4089 av_log(c->fc, AV_LOG_ERROR,
4090 "The 'keys' atom with the invalid key count: %"PRIu32"\n", count);
4091 return AVERROR_INVALIDDATA;
4094 c->meta_keys_count = count + 1;
4095 c->meta_keys = av_mallocz(c->meta_keys_count * sizeof(*c->meta_keys));
4097 return AVERROR(ENOMEM);
4099 for (i = 1; i <= count; ++i) {
4100 uint32_t key_size = avio_rb32(pb);
4101 uint32_t type = avio_rl32(pb);
4103 av_log(c->fc, AV_LOG_ERROR,
4104 "The key# %"PRIu32" in meta has invalid size:"
4105 "%"PRIu32"\n", i, key_size);
4106 return AVERROR_INVALIDDATA;
4109 if (type != MKTAG('m','d','t','a')) {
4110 avio_skip(pb, key_size);
4112 c->meta_keys[i] = av_mallocz(key_size + 1);
4113 if (!c->meta_keys[i])
4114 return AVERROR(ENOMEM);
4115 avio_read(pb, c->meta_keys[i], key_size);
4121 static int mov_read_custom(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4123 int64_t end = avio_tell(pb) + atom.size;
4124 uint8_t *key = NULL, *val = NULL, *mean = NULL;
4128 MOVStreamContext *sc;
4130 if (c->fc->nb_streams < 1)
4132 st = c->fc->streams[c->fc->nb_streams-1];
4135 for (i = 0; i < 3; i++) {
4139 if (end - avio_tell(pb) <= 12)
4142 len = avio_rb32(pb);
4143 tag = avio_rl32(pb);
4144 avio_skip(pb, 4); // flags
4146 if (len < 12 || len - 12 > end - avio_tell(pb))
4150 if (tag == MKTAG('m', 'e', 'a', 'n'))
4152 else if (tag == MKTAG('n', 'a', 'm', 'e'))
4154 else if (tag == MKTAG('d', 'a', 't', 'a') && len > 4) {
4161 *p = av_malloc(len + 1);
4164 ret = ffio_read_size(pb, *p, len);
4172 if (mean && key && val) {
4173 if (strcmp(key, "iTunSMPB") == 0) {
4174 int priming, remainder, samples;
4175 if(sscanf(val, "%*X %X %X %X", &priming, &remainder, &samples) == 3){
4176 if(priming>0 && priming<16384)
4177 sc->start_pad = priming;
4180 if (strcmp(key, "cdec") != 0) {
4181 av_dict_set(&c->fc->metadata, key, val,
4182 AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL);
4186 av_log(c->fc, AV_LOG_VERBOSE,
4187 "Unhandled or malformed custom metadata of size %"PRId64"\n", atom.size);
4190 avio_seek(pb, end, SEEK_SET);
4197 static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4199 while (atom.size > 8) {
4200 uint32_t tag = avio_rl32(pb);
4202 if (tag == MKTAG('h','d','l','r')) {
4203 avio_seek(pb, -8, SEEK_CUR);
4205 return mov_read_default(c, pb, atom);
4211 // return 1 when matrix is identity, 0 otherwise
4212 #define IS_MATRIX_IDENT(matrix) \
4213 ( (matrix)[0][0] == (1 << 16) && \
4214 (matrix)[1][1] == (1 << 16) && \
4215 (matrix)[2][2] == (1 << 30) && \
4216 !(matrix)[0][1] && !(matrix)[0][2] && \
4217 !(matrix)[1][0] && !(matrix)[1][2] && \
4218 !(matrix)[2][0] && !(matrix)[2][1])
4220 static int mov_read_tkhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4225 int display_matrix[3][3];
4226 int res_display_matrix[3][3] = { { 0 } };
4228 MOVStreamContext *sc;
4232 if (c->fc->nb_streams < 1)
4234 st = c->fc->streams[c->fc->nb_streams-1];
4237 version = avio_r8(pb);
4238 flags = avio_rb24(pb);
4239 st->disposition |= (flags & MOV_TKHD_FLAG_ENABLED) ? AV_DISPOSITION_DEFAULT : 0;
4245 avio_rb32(pb); /* creation time */
4246 avio_rb32(pb); /* modification time */
4248 st->id = (int)avio_rb32(pb); /* track id (NOT 0 !)*/
4249 avio_rb32(pb); /* reserved */
4251 /* highlevel (considering edits) duration in movie timebase */
4252 (version == 1) ? avio_rb64(pb) : avio_rb32(pb);
4253 avio_rb32(pb); /* reserved */
4254 avio_rb32(pb); /* reserved */
4256 avio_rb16(pb); /* layer */
4257 avio_rb16(pb); /* alternate group */
4258 avio_rb16(pb); /* volume */
4259 avio_rb16(pb); /* reserved */
4261 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
4262 // they're kept in fixed point format through all calculations
4263 // save u,v,z to store the whole matrix in the AV_PKT_DATA_DISPLAYMATRIX
4264 // side data, but the scale factor is not needed to calculate aspect ratio
4265 for (i = 0; i < 3; i++) {
4266 display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
4267 display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
4268 display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
4271 width = avio_rb32(pb); // 16.16 fixed point track width
4272 height = avio_rb32(pb); // 16.16 fixed point track height
4273 sc->width = width >> 16;
4274 sc->height = height >> 16;
4276 // apply the moov display matrix (after the tkhd one)
4277 for (i = 0; i < 3; i++) {
4278 const int sh[3] = { 16, 16, 30 };
4279 for (j = 0; j < 3; j++) {
4280 for (e = 0; e < 3; e++) {
4281 res_display_matrix[i][j] +=
4282 ((int64_t) display_matrix[i][e] *
4283 c->movie_display_matrix[e][j]) >> sh[e];
4288 // save the matrix when it is not the default identity
4289 if (!IS_MATRIX_IDENT(res_display_matrix)) {
4292 av_freep(&sc->display_matrix);
4293 sc->display_matrix = av_malloc(sizeof(int32_t) * 9);
4294 if (!sc->display_matrix)
4295 return AVERROR(ENOMEM);
4297 for (i = 0; i < 3; i++)
4298 for (j = 0; j < 3; j++)
4299 sc->display_matrix[i * 3 + j] = res_display_matrix[i][j];
4301 #if FF_API_OLD_ROTATE_API
4302 rotate = av_display_rotation_get(sc->display_matrix);
4303 if (!isnan(rotate)) {
4304 char rotate_buf[64];
4306 if (rotate < 0) // for backward compatibility
4308 snprintf(rotate_buf, sizeof(rotate_buf), "%g", rotate);
4309 av_dict_set(&st->metadata, "rotate", rotate_buf, 0);
4314 // transform the display width/height according to the matrix
4315 // to keep the same scale, use [width height 1<<16]
4316 if (width && height && sc->display_matrix) {
4317 double disp_transform[2];
4319 for (i = 0; i < 2; i++)
4320 disp_transform[i] = hypot(sc->display_matrix[0 + i],
4321 sc->display_matrix[3 + i]);
4323 if (disp_transform[0] > 0 && disp_transform[1] > 0 &&
4324 disp_transform[0] < (1<<24) && disp_transform[1] < (1<<24) &&
4325 fabs((disp_transform[0] / disp_transform[1]) - 1.0) > 0.01)
4326 st->sample_aspect_ratio = av_d2q(
4327 disp_transform[0] / disp_transform[1],
4333 static int mov_read_tfhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4335 MOVFragment *frag = &c->fragment;
4336 MOVTrackExt *trex = NULL;
4337 int flags, track_id, i;
4339 avio_r8(pb); /* version */
4340 flags = avio_rb24(pb);
4342 track_id = avio_rb32(pb);
4344 return AVERROR_INVALIDDATA;
4345 frag->track_id = track_id;
4346 set_frag_stream(&c->frag_index, track_id);
4347 for (i = 0; i < c->trex_count; i++)
4348 if (c->trex_data[i].track_id == frag->track_id) {
4349 trex = &c->trex_data[i];
4353 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
4354 return AVERROR_INVALIDDATA;
4357 frag->base_data_offset = flags & MOV_TFHD_BASE_DATA_OFFSET ?
4358 avio_rb64(pb) : flags & MOV_TFHD_DEFAULT_BASE_IS_MOOF ?
4359 frag->moof_offset : frag->implicit_offset;
4360 frag->stsd_id = flags & MOV_TFHD_STSD_ID ? avio_rb32(pb) : trex->stsd_id;
4362 frag->duration = flags & MOV_TFHD_DEFAULT_DURATION ?
4363 avio_rb32(pb) : trex->duration;
4364 frag->size = flags & MOV_TFHD_DEFAULT_SIZE ?
4365 avio_rb32(pb) : trex->size;
4366 frag->flags = flags & MOV_TFHD_DEFAULT_FLAGS ?
4367 avio_rb32(pb) : trex->flags;
4368 av_log(c->fc, AV_LOG_TRACE, "frag flags 0x%x\n", frag->flags);
4373 static int mov_read_chap(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4378 num = atom.size / 4;
4379 if (!(new_tracks = av_malloc_array(num, sizeof(int))))
4380 return AVERROR(ENOMEM);
4382 av_free(c->chapter_tracks);
4383 c->chapter_tracks = new_tracks;
4384 c->nb_chapter_tracks = num;
4386 for (i = 0; i < num && !pb->eof_reached; i++)
4387 c->chapter_tracks[i] = avio_rb32(pb);
4392 static int mov_read_trex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4397 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
4398 return AVERROR_INVALIDDATA;
4399 if ((err = av_reallocp_array(&c->trex_data, c->trex_count + 1,
4400 sizeof(*c->trex_data))) < 0) {
4405 c->fc->duration = AV_NOPTS_VALUE; // the duration from mvhd is not representing the whole file when fragments are used.
4407 trex = &c->trex_data[c->trex_count++];
4408 avio_r8(pb); /* version */
4409 avio_rb24(pb); /* flags */
4410 trex->track_id = avio_rb32(pb);
4411 trex->stsd_id = avio_rb32(pb);
4412 trex->duration = avio_rb32(pb);
4413 trex->size = avio_rb32(pb);
4414 trex->flags = avio_rb32(pb);
4418 static int mov_read_tfdt(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4420 MOVFragment *frag = &c->fragment;
4421 AVStream *st = NULL;
4422 MOVStreamContext *sc;
4424 MOVFragmentStreamInfo * frag_stream_info;
4425 int64_t base_media_decode_time;
4427 for (i = 0; i < c->fc->nb_streams; i++) {
4428 if (c->fc->streams[i]->id == frag->track_id) {
4429 st = c->fc->streams[i];
4434 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %u\n", frag->track_id);
4435 return AVERROR_INVALIDDATA;
4438 if (sc->pseudo_stream_id + 1 != frag->stsd_id)
4440 version = avio_r8(pb);
4441 avio_rb24(pb); /* flags */
4443 base_media_decode_time = avio_rb64(pb);
4445 base_media_decode_time = avio_rb32(pb);
4448 frag_stream_info = get_current_frag_stream_info(&c->frag_index);
4449 if (frag_stream_info)
4450 frag_stream_info->tfdt_dts = base_media_decode_time;
4451 sc->track_end = base_media_decode_time;
4456 static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4458 MOVFragment *frag = &c->fragment;
4459 AVStream *st = NULL;
4460 MOVStreamContext *sc;
4463 int64_t dts, pts = AV_NOPTS_VALUE;
4464 int data_offset = 0;
4465 unsigned entries, first_sample_flags = frag->flags;
4466 int flags, distance, i;
4467 int64_t prev_dts = AV_NOPTS_VALUE;
4468 int next_frag_index = -1, index_entry_pos;
4469 size_t requested_size;
4470 AVIndexEntry *new_entries;
4471 MOVFragmentStreamInfo * frag_stream_info;
4473 for (i = 0; i < c->fc->nb_streams; i++) {
4474 if (c->fc->streams[i]->id == frag->track_id) {
4475 st = c->fc->streams[i];
4480 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %u\n", frag->track_id);
4481 return AVERROR_INVALIDDATA;
4484 if (sc->pseudo_stream_id+1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4487 // Find the next frag_index index that has a valid index_entry for
4488 // the current track_id.
4490 // A valid index_entry means the trun for the fragment was read
4491 // and it's samples are in index_entries at the given position.
4492 // New index entries will be inserted before the index_entry found.
4493 index_entry_pos = st->nb_index_entries;
4494 for (i = c->frag_index.current + 1; i < c->frag_index.nb_items; i++) {
4495 frag_stream_info = get_frag_stream_info(&c->frag_index, i, frag->track_id);
4496 if (frag_stream_info && frag_stream_info->index_entry >= 0) {
4497 next_frag_index = i;
4498 index_entry_pos = frag_stream_info->index_entry;
4503 avio_r8(pb); /* version */
4504 flags = avio_rb24(pb);
4505 entries = avio_rb32(pb);
4506 av_log(c->fc, AV_LOG_TRACE, "flags 0x%x entries %u\n", flags, entries);
4508 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
4509 return AVERROR_INVALIDDATA;
4510 if (flags & MOV_TRUN_DATA_OFFSET) data_offset = avio_rb32(pb);
4511 if (flags & MOV_TRUN_FIRST_SAMPLE_FLAGS) first_sample_flags = avio_rb32(pb);
4513 frag_stream_info = get_current_frag_stream_info(&c->frag_index);
4514 if (frag_stream_info)
4516 if (frag_stream_info->first_tfra_pts != AV_NOPTS_VALUE &&
4517 c->use_mfra_for == FF_MOV_FLAG_MFRA_PTS) {
4518 pts = frag_stream_info->first_tfra_pts;
4519 av_log(c->fc, AV_LOG_DEBUG, "found mfra time %"PRId64
4520 ", using it for pts\n", pts);
4521 } else if (frag_stream_info->sidx_pts != AV_NOPTS_VALUE) {
4522 // FIXME: sidx earliest_presentation_time is *PTS*, s.b.
4523 // pts = frag_stream_info->sidx_pts;
4524 dts = frag_stream_info->sidx_pts - sc->time_offset;
4525 av_log(c->fc, AV_LOG_DEBUG, "found sidx time %"PRId64
4526 ", using it for pts\n", pts);
4527 } else if (frag_stream_info->tfdt_dts != AV_NOPTS_VALUE) {
4528 dts = frag_stream_info->tfdt_dts - sc->time_offset;
4529 av_log(c->fc, AV_LOG_DEBUG, "found tfdt time %"PRId64
4530 ", using it for dts\n", dts);
4532 dts = sc->track_end - sc->time_offset;
4533 av_log(c->fc, AV_LOG_DEBUG, "found track end time %"PRId64
4534 ", using it for dts\n", dts);
4537 dts = sc->track_end - sc->time_offset;
4538 av_log(c->fc, AV_LOG_DEBUG, "found track end time %"PRId64
4539 ", using it for dts\n", dts);
4541 offset = frag->base_data_offset + data_offset;
4543 av_log(c->fc, AV_LOG_TRACE, "first sample flags 0x%x\n", first_sample_flags);
4545 // realloc space for new index entries
4546 if((unsigned)st->nb_index_entries + entries >= UINT_MAX / sizeof(AVIndexEntry)) {
4547 entries = UINT_MAX / sizeof(AVIndexEntry) - st->nb_index_entries;
4548 av_log(c->fc, AV_LOG_ERROR, "Failed to add index entry\n");
4553 requested_size = (st->nb_index_entries + entries) * sizeof(AVIndexEntry);
4554 new_entries = av_fast_realloc(st->index_entries,
4555 &st->index_entries_allocated_size,
4558 return AVERROR(ENOMEM);
4559 st->index_entries= new_entries;
4561 requested_size = (st->nb_index_entries + entries) * sizeof(*sc->ctts_data);
4562 ctts_data = av_fast_realloc(sc->ctts_data, &sc->ctts_allocated_size,
4565 return AVERROR(ENOMEM);
4566 sc->ctts_data = ctts_data;
4568 // In case there were samples without ctts entries, ensure they get
4569 // zero valued entries. This ensures clips which mix boxes with and
4570 // without ctts entries don't pickup uninitialized data.
4571 memset(sc->ctts_data + sc->ctts_count, 0,
4572 (st->nb_index_entries - sc->ctts_count) * sizeof(*sc->ctts_data));
4574 if (index_entry_pos < st->nb_index_entries) {
4575 // Make hole in index_entries and ctts_data for new samples
4576 memmove(st->index_entries + index_entry_pos + entries,
4577 st->index_entries + index_entry_pos,
4578 sizeof(*st->index_entries) *
4579 (st->nb_index_entries - index_entry_pos));
4580 memmove(sc->ctts_data + index_entry_pos + entries,
4581 sc->ctts_data + index_entry_pos,
4582 sizeof(*sc->ctts_data) * (sc->ctts_count - index_entry_pos));
4583 if (index_entry_pos < sc->current_sample) {
4584 sc->current_sample += entries;
4588 st->nb_index_entries += entries;
4589 sc->ctts_count = st->nb_index_entries;
4591 // Record the index_entry position in frag_index of this fragment
4592 if (frag_stream_info)
4593 frag_stream_info->index_entry = index_entry_pos;
4595 if (index_entry_pos > 0)
4596 prev_dts = st->index_entries[index_entry_pos-1].timestamp;
4598 for (i = 0; i < entries && !pb->eof_reached; i++) {
4599 unsigned sample_size = frag->size;
4600 int sample_flags = i ? frag->flags : first_sample_flags;
4601 unsigned sample_duration = frag->duration;
4602 unsigned ctts_duration = 0;
4604 int index_entry_flags = 0;
4606 if (flags & MOV_TRUN_SAMPLE_DURATION) sample_duration = avio_rb32(pb);
4607 if (flags & MOV_TRUN_SAMPLE_SIZE) sample_size = avio_rb32(pb);
4608 if (flags & MOV_TRUN_SAMPLE_FLAGS) sample_flags = avio_rb32(pb);
4609 if (flags & MOV_TRUN_SAMPLE_CTS) ctts_duration = avio_rb32(pb);
4611 mov_update_dts_shift(sc, ctts_duration);
4612 if (pts != AV_NOPTS_VALUE) {
4613 dts = pts - sc->dts_shift;
4614 if (flags & MOV_TRUN_SAMPLE_CTS) {
4615 dts -= ctts_duration;
4617 dts -= sc->time_offset;
4619 av_log(c->fc, AV_LOG_DEBUG,
4620 "pts %"PRId64" calculated dts %"PRId64
4621 " sc->dts_shift %d ctts.duration %d"
4622 " sc->time_offset %"PRId64
4623 " flags & MOV_TRUN_SAMPLE_CTS %d\n",
4625 sc->dts_shift, ctts_duration,
4626 sc->time_offset, flags & MOV_TRUN_SAMPLE_CTS);
4627 pts = AV_NOPTS_VALUE;
4630 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
4634 !(sample_flags & (MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC |
4635 MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES));
4638 index_entry_flags |= AVINDEX_KEYFRAME;
4640 // Fragments can overlap in time. Discard overlapping frames after
4642 if (prev_dts >= dts)
4643 index_entry_flags |= AVINDEX_DISCARD_FRAME;
4645 st->index_entries[index_entry_pos].pos = offset;
4646 st->index_entries[index_entry_pos].timestamp = dts;
4647 st->index_entries[index_entry_pos].size= sample_size;
4648 st->index_entries[index_entry_pos].min_distance= distance;
4649 st->index_entries[index_entry_pos].flags = index_entry_flags;
4651 sc->ctts_data[index_entry_pos].count = 1;
4652 sc->ctts_data[index_entry_pos].duration = ctts_duration;
4655 av_log(c->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
4656 "size %u, distance %d, keyframe %d\n", st->index,
4657 index_entry_pos, offset, dts, sample_size, distance, keyframe);
4659 dts += sample_duration;
4660 offset += sample_size;
4661 sc->data_size += sample_size;
4662 sc->duration_for_fps += sample_duration;
4663 sc->nb_frames_for_fps ++;
4666 // EOF found before reading all entries. Fix the hole this would
4667 // leave in index_entries and ctts_data
4668 int gap = entries - i;
4669 memmove(st->index_entries + index_entry_pos,
4670 st->index_entries + index_entry_pos + gap,
4671 sizeof(*st->index_entries) *
4672 (st->nb_index_entries - (index_entry_pos + gap)));
4673 memmove(sc->ctts_data + index_entry_pos,
4674 sc->ctts_data + index_entry_pos + gap,
4675 sizeof(*sc->ctts_data) *
4676 (sc->ctts_count - (index_entry_pos + gap)));
4678 st->nb_index_entries -= gap;
4679 sc->ctts_count -= gap;
4680 if (index_entry_pos < sc->current_sample) {
4681 sc->current_sample -= gap;
4686 // The end of this new fragment may overlap in time with the start
4687 // of the next fragment in index_entries. Mark the samples in the next
4688 // fragment that overlap with AVINDEX_DISCARD_FRAME
4689 prev_dts = AV_NOPTS_VALUE;
4690 if (index_entry_pos > 0)
4691 prev_dts = st->index_entries[index_entry_pos-1].timestamp;
4692 for (i = index_entry_pos; i < st->nb_index_entries; i++) {
4693 if (prev_dts < st->index_entries[i].timestamp)
4695 st->index_entries[i].flags |= AVINDEX_DISCARD_FRAME;
4698 // If a hole was created to insert the new index_entries into,
4699 // the index_entry recorded for all subsequent moof must
4700 // be incremented by the number of entries inserted.
4701 fix_frag_index_entries(&c->frag_index, next_frag_index,
4702 frag->track_id, entries);
4704 if (pb->eof_reached)
4707 frag->implicit_offset = offset;
4709 sc->track_end = dts + sc->time_offset;
4710 if (st->duration < sc->track_end)
4711 st->duration = sc->track_end;
4716 static int mov_read_sidx(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4718 int64_t offset = avio_tell(pb) + atom.size, pts, timestamp;
4720 unsigned i, j, track_id, item_count;
4721 AVStream *st = NULL;
4722 AVStream *ref_st = NULL;
4723 MOVStreamContext *sc, *ref_sc = NULL;
4724 AVRational timescale;
4726 version = avio_r8(pb);
4728 avpriv_request_sample(c->fc, "sidx version %u", version);
4732 avio_rb24(pb); // flags
4734 track_id = avio_rb32(pb); // Reference ID
4735 for (i = 0; i < c->fc->nb_streams; i++) {
4736 if (c->fc->streams[i]->id == track_id) {
4737 st = c->fc->streams[i];
4742 av_log(c->fc, AV_LOG_WARNING, "could not find corresponding track id %d\n", track_id);
4748 timescale = av_make_q(1, avio_rb32(pb));
4750 if (timescale.den <= 0) {
4751 av_log(c->fc, AV_LOG_ERROR, "Invalid sidx timescale 1/%d\n", timescale.den);
4752 return AVERROR_INVALIDDATA;
4756 pts = avio_rb32(pb);
4757 offset += avio_rb32(pb);
4759 pts = avio_rb64(pb);
4760 offset += avio_rb64(pb);
4763 avio_rb16(pb); // reserved
4765 item_count = avio_rb16(pb);
4767 for (i = 0; i < item_count; i++) {
4769 MOVFragmentStreamInfo * frag_stream_info;
4770 uint32_t size = avio_rb32(pb);
4771 uint32_t duration = avio_rb32(pb);
4772 if (size & 0x80000000) {
4773 avpriv_request_sample(c->fc, "sidx reference_type 1");
4774 return AVERROR_PATCHWELCOME;
4776 avio_rb32(pb); // sap_flags
4777 timestamp = av_rescale_q(pts, st->time_base, timescale);
4779 index = update_frag_index(c, offset);
4780 frag_stream_info = get_frag_stream_info(&c->frag_index, index, track_id);
4781 if (frag_stream_info)
4782 frag_stream_info->sidx_pts = timestamp;
4788 st->duration = sc->track_end = pts;
4792 if (offset == avio_size(pb)) {
4793 // Find first entry in fragment index that came from an sidx.
4794 // This will pretty much always be the first entry.
4795 for (i = 0; i < c->frag_index.nb_items; i++) {
4796 MOVFragmentIndexItem * item = &c->frag_index.item[i];
4797 for (j = 0; ref_st == NULL && j < item->nb_stream_info; j++) {
4798 MOVFragmentStreamInfo * si;
4799 si = &item->stream_info[j];
4800 if (si->sidx_pts != AV_NOPTS_VALUE) {
4801 ref_st = c->fc->streams[i];
4802 ref_sc = ref_st->priv_data;
4807 for (i = 0; i < c->fc->nb_streams; i++) {
4808 st = c->fc->streams[i];
4810 if (!sc->has_sidx) {
4811 st->duration = sc->track_end = av_rescale(ref_st->duration, sc->time_scale, ref_sc->time_scale);
4815 c->frag_index.complete = 1;
4821 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
4822 /* like the files created with Adobe Premiere 5.0, for samples see */
4823 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
4824 static int mov_read_wide(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4829 return 0; /* continue */
4830 if (avio_rb32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
4831 avio_skip(pb, atom.size - 4);
4834 atom.type = avio_rl32(pb);
4836 if (atom.type != MKTAG('m','d','a','t')) {
4837 avio_skip(pb, atom.size);
4840 err = mov_read_mdat(c, pb, atom);
4844 static int mov_read_cmov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4849 uint8_t *moov_data; /* uncompressed data */
4850 long cmov_len, moov_len;
4853 avio_rb32(pb); /* dcom atom */
4854 if (avio_rl32(pb) != MKTAG('d','c','o','m'))
4855 return AVERROR_INVALIDDATA;
4856 if (avio_rl32(pb) != MKTAG('z','l','i','b')) {
4857 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !\n");
4858 return AVERROR_INVALIDDATA;
4860 avio_rb32(pb); /* cmvd atom */
4861 if (avio_rl32(pb) != MKTAG('c','m','v','d'))
4862 return AVERROR_INVALIDDATA;
4863 moov_len = avio_rb32(pb); /* uncompressed size */
4864 cmov_len = atom.size - 6 * 4;
4866 cmov_data = av_malloc(cmov_len);
4868 return AVERROR(ENOMEM);
4869 moov_data = av_malloc(moov_len);
4872 return AVERROR(ENOMEM);
4874 ret = ffio_read_size(pb, cmov_data, cmov_len);
4876 goto free_and_return;
4878 if (uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
4879 goto free_and_return;
4880 if (ffio_init_context(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
4881 goto free_and_return;
4882 ctx.seekable = AVIO_SEEKABLE_NORMAL;
4883 atom.type = MKTAG('m','o','o','v');
4884 atom.size = moov_len;
4885 ret = mov_read_default(c, &ctx, atom);
4891 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
4892 return AVERROR(ENOSYS);
4896 /* edit list atom */
4897 static int mov_read_elst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4899 MOVStreamContext *sc;
4900 int i, edit_count, version;
4901 int64_t elst_entry_size;
4903 if (c->fc->nb_streams < 1 || c->ignore_editlist)
4905 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
4907 version = avio_r8(pb); /* version */
4908 avio_rb24(pb); /* flags */
4909 edit_count = avio_rb32(pb); /* entries */
4912 elst_entry_size = version == 1 ? 20 : 12;
4913 if (atom.size != edit_count * elst_entry_size) {
4914 if (c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
4915 av_log(c->fc, AV_LOG_ERROR, "Invalid edit list entry_count: %d for elst atom of size: %"PRId64" bytes.\n",
4916 edit_count, atom.size + 8);
4917 return AVERROR_INVALIDDATA;
4919 edit_count = atom.size / elst_entry_size;
4920 if (edit_count * elst_entry_size != atom.size) {
4921 av_log(c->fc, AV_LOG_WARNING, "ELST atom of %"PRId64" bytes, bigger than %d entries.", atom.size, edit_count);
4929 av_log(c->fc, AV_LOG_WARNING, "Duplicated ELST atom\n");
4930 av_free(sc->elst_data);
4932 sc->elst_data = av_malloc_array(edit_count, sizeof(*sc->elst_data));
4934 return AVERROR(ENOMEM);
4936 av_log(c->fc, AV_LOG_TRACE, "track[%u].edit_count = %i\n", c->fc->nb_streams - 1, edit_count);
4937 for (i = 0; i < edit_count && atom.size > 0 && !pb->eof_reached; i++) {
4938 MOVElst *e = &sc->elst_data[i];
4941 e->duration = avio_rb64(pb);
4942 e->time = avio_rb64(pb);
4945 e->duration = avio_rb32(pb); /* segment duration */
4946 e->time = (int32_t)avio_rb32(pb); /* media time */
4949 e->rate = avio_rb32(pb) / 65536.0;
4951 av_log(c->fc, AV_LOG_TRACE, "duration=%"PRId64" time=%"PRId64" rate=%f\n",
4952 e->duration, e->time, e->rate);
4954 if (e->time < 0 && e->time != -1 &&
4955 c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
4956 av_log(c->fc, AV_LOG_ERROR, "Track %d, edit %d: Invalid edit list media time=%"PRId64"\n",
4957 c->fc->nb_streams-1, i, e->time);
4958 return AVERROR_INVALIDDATA;
4966 static int mov_read_tmcd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4968 MOVStreamContext *sc;
4970 if (c->fc->nb_streams < 1)
4971 return AVERROR_INVALIDDATA;
4972 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
4973 sc->timecode_track = avio_rb32(pb);
4977 static int mov_read_vpcc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4980 int version, color_range, color_primaries, color_trc, color_space;
4982 if (c->fc->nb_streams < 1)
4984 st = c->fc->streams[c->fc->nb_streams - 1];
4986 if (atom.size < 5) {
4987 av_log(c->fc, AV_LOG_ERROR, "Empty VP Codec Configuration box\n");
4988 return AVERROR_INVALIDDATA;
4991 version = avio_r8(pb);
4993 av_log(c->fc, AV_LOG_WARNING, "Unsupported VP Codec Configuration box version %d\n", version);
4996 avio_skip(pb, 3); /* flags */
4998 avio_skip(pb, 2); /* profile + level */
4999 color_range = avio_r8(pb); /* bitDepth, chromaSubsampling, videoFullRangeFlag */
5000 color_primaries = avio_r8(pb);
5001 color_trc = avio_r8(pb);
5002 color_space = avio_r8(pb);
5003 if (avio_rb16(pb)) /* codecIntializationDataSize */
5004 return AVERROR_INVALIDDATA;
5006 if (!av_color_primaries_name(color_primaries))
5007 color_primaries = AVCOL_PRI_UNSPECIFIED;
5008 if (!av_color_transfer_name(color_trc))
5009 color_trc = AVCOL_TRC_UNSPECIFIED;
5010 if (!av_color_space_name(color_space))
5011 color_space = AVCOL_SPC_UNSPECIFIED;
5013 st->codecpar->color_range = (color_range & 1) ? AVCOL_RANGE_JPEG : AVCOL_RANGE_MPEG;
5014 st->codecpar->color_primaries = color_primaries;
5015 st->codecpar->color_trc = color_trc;
5016 st->codecpar->color_space = color_space;
5021 static int mov_read_smdm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5023 MOVStreamContext *sc;
5024 const int chroma_den = 50000;
5025 const int luma_den = 10000;
5028 if (c->fc->nb_streams < 1)
5029 return AVERROR_INVALIDDATA;
5031 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5033 if (atom.size < 5) {
5034 av_log(c->fc, AV_LOG_ERROR, "Empty Mastering Display Metadata box\n");
5035 return AVERROR_INVALIDDATA;
5038 version = avio_r8(pb);
5040 av_log(c->fc, AV_LOG_WARNING, "Unsupported Mastering Display Metadata box version %d\n", version);
5043 avio_skip(pb, 3); /* flags */
5045 sc->mastering = av_mastering_display_metadata_alloc();
5047 return AVERROR(ENOMEM);
5049 for (i = 0; i < 3; i++)
5050 for (j = 0; j < 2; j++)
5051 sc->mastering->display_primaries[i][j] =
5052 av_make_q(lrint(((double)avio_rb16(pb) / (1 << 16)) * chroma_den), chroma_den);
5053 for (i = 0; i < 2; i++)
5054 sc->mastering->white_point[i] =
5055 av_make_q(lrint(((double)avio_rb16(pb) / (1 << 16)) * chroma_den), chroma_den);
5056 sc->mastering->max_luminance =
5057 av_make_q(lrint(((double)avio_rb32(pb) / (1 << 8)) * luma_den), luma_den);
5058 sc->mastering->min_luminance =
5059 av_make_q(lrint(((double)avio_rb32(pb) / (1 << 14)) * luma_den), luma_den);
5061 sc->mastering->has_primaries = 1;
5062 sc->mastering->has_luminance = 1;
5067 static int mov_read_coll(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5069 MOVStreamContext *sc;
5072 if (c->fc->nb_streams < 1)
5073 return AVERROR_INVALIDDATA;
5075 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5077 if (atom.size < 5) {
5078 av_log(c->fc, AV_LOG_ERROR, "Empty Content Light Level box\n");
5079 return AVERROR_INVALIDDATA;
5082 version = avio_r8(pb);
5084 av_log(c->fc, AV_LOG_WARNING, "Unsupported Content Light Level box version %d\n", version);
5087 avio_skip(pb, 3); /* flags */
5089 sc->coll = av_content_light_metadata_alloc(&sc->coll_size);
5091 return AVERROR(ENOMEM);
5093 sc->coll->MaxCLL = avio_rb16(pb);
5094 sc->coll->MaxFALL = avio_rb16(pb);
5099 static int mov_read_st3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5102 MOVStreamContext *sc;
5103 enum AVStereo3DType type;
5106 if (c->fc->nb_streams < 1)
5109 st = c->fc->streams[c->fc->nb_streams - 1];
5112 if (atom.size < 5) {
5113 av_log(c->fc, AV_LOG_ERROR, "Empty stereoscopic video box\n");
5114 return AVERROR_INVALIDDATA;
5116 avio_skip(pb, 4); /* version + flags */
5121 type = AV_STEREO3D_2D;
5124 type = AV_STEREO3D_TOPBOTTOM;
5127 type = AV_STEREO3D_SIDEBYSIDE;
5130 av_log(c->fc, AV_LOG_WARNING, "Unknown st3d mode value %d\n", mode);
5134 sc->stereo3d = av_stereo3d_alloc();
5136 return AVERROR(ENOMEM);
5138 sc->stereo3d->type = type;
5142 static int mov_read_sv3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5145 MOVStreamContext *sc;
5146 int size, version, layout;
5147 int32_t yaw, pitch, roll;
5148 uint32_t l = 0, t = 0, r = 0, b = 0;
5149 uint32_t tag, padding = 0;
5150 enum AVSphericalProjection projection;
5152 if (c->fc->nb_streams < 1)
5155 st = c->fc->streams[c->fc->nb_streams - 1];
5158 if (atom.size < 8) {
5159 av_log(c->fc, AV_LOG_ERROR, "Empty spherical video box\n");
5160 return AVERROR_INVALIDDATA;
5163 size = avio_rb32(pb);
5164 if (size <= 12 || size > atom.size)
5165 return AVERROR_INVALIDDATA;
5167 tag = avio_rl32(pb);
5168 if (tag != MKTAG('s','v','h','d')) {
5169 av_log(c->fc, AV_LOG_ERROR, "Missing spherical video header\n");
5172 version = avio_r8(pb);
5174 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5178 avio_skip(pb, 3); /* flags */
5179 avio_skip(pb, size - 12); /* metadata_source */
5181 size = avio_rb32(pb);
5182 if (size > atom.size)
5183 return AVERROR_INVALIDDATA;
5185 tag = avio_rl32(pb);
5186 if (tag != MKTAG('p','r','o','j')) {
5187 av_log(c->fc, AV_LOG_ERROR, "Missing projection box\n");
5191 size = avio_rb32(pb);
5192 if (size > atom.size)
5193 return AVERROR_INVALIDDATA;
5195 tag = avio_rl32(pb);
5196 if (tag != MKTAG('p','r','h','d')) {
5197 av_log(c->fc, AV_LOG_ERROR, "Missing projection header box\n");
5200 version = avio_r8(pb);
5202 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5206 avio_skip(pb, 3); /* flags */
5208 /* 16.16 fixed point */
5209 yaw = avio_rb32(pb);
5210 pitch = avio_rb32(pb);
5211 roll = avio_rb32(pb);
5213 size = avio_rb32(pb);
5214 if (size > atom.size)
5215 return AVERROR_INVALIDDATA;
5217 tag = avio_rl32(pb);
5218 version = avio_r8(pb);
5220 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5224 avio_skip(pb, 3); /* flags */
5226 case MKTAG('c','b','m','p'):
5227 layout = avio_rb32(pb);
5229 av_log(c->fc, AV_LOG_WARNING,
5230 "Unsupported cubemap layout %d\n", layout);
5233 projection = AV_SPHERICAL_CUBEMAP;
5234 padding = avio_rb32(pb);
5236 case MKTAG('e','q','u','i'):
5242 if (b >= UINT_MAX - t || r >= UINT_MAX - l) {
5243 av_log(c->fc, AV_LOG_ERROR,
5244 "Invalid bounding rectangle coordinates "
5245 "%"PRIu32",%"PRIu32",%"PRIu32",%"PRIu32"\n", l, t, r, b);
5246 return AVERROR_INVALIDDATA;
5249 if (l || t || r || b)
5250 projection = AV_SPHERICAL_EQUIRECTANGULAR_TILE;
5252 projection = AV_SPHERICAL_EQUIRECTANGULAR;
5255 av_log(c->fc, AV_LOG_ERROR, "Unknown projection type\n");
5259 sc->spherical = av_spherical_alloc(&sc->spherical_size);
5261 return AVERROR(ENOMEM);
5263 sc->spherical->projection = projection;
5265 sc->spherical->yaw = yaw;
5266 sc->spherical->pitch = pitch;
5267 sc->spherical->roll = roll;
5269 sc->spherical->padding = padding;
5271 sc->spherical->bound_left = l;
5272 sc->spherical->bound_top = t;
5273 sc->spherical->bound_right = r;
5274 sc->spherical->bound_bottom = b;
5279 static int mov_parse_uuid_spherical(MOVStreamContext *sc, AVIOContext *pb, size_t len)
5282 uint8_t *buffer = av_malloc(len + 1);
5286 return AVERROR(ENOMEM);
5289 ret = ffio_read_size(pb, buffer, len);
5293 /* Check for mandatory keys and values, try to support XML as best-effort */
5294 if (!sc->spherical &&
5295 av_stristr(buffer, "<GSpherical:StitchingSoftware>") &&
5296 (val = av_stristr(buffer, "<GSpherical:Spherical>")) &&
5297 av_stristr(val, "true") &&
5298 (val = av_stristr(buffer, "<GSpherical:Stitched>")) &&
5299 av_stristr(val, "true") &&
5300 (val = av_stristr(buffer, "<GSpherical:ProjectionType>")) &&
5301 av_stristr(val, "equirectangular")) {
5302 sc->spherical = av_spherical_alloc(&sc->spherical_size);
5306 sc->spherical->projection = AV_SPHERICAL_EQUIRECTANGULAR;
5308 if (av_stristr(buffer, "<GSpherical:StereoMode>") && !sc->stereo3d) {
5309 enum AVStereo3DType mode;
5311 if (av_stristr(buffer, "left-right"))
5312 mode = AV_STEREO3D_SIDEBYSIDE;
5313 else if (av_stristr(buffer, "top-bottom"))
5314 mode = AV_STEREO3D_TOPBOTTOM;
5316 mode = AV_STEREO3D_2D;
5318 sc->stereo3d = av_stereo3d_alloc();
5322 sc->stereo3d->type = mode;
5326 val = av_stristr(buffer, "<GSpherical:InitialViewHeadingDegrees>");
5328 sc->spherical->yaw = strtol(val, NULL, 10) * (1 << 16);
5329 val = av_stristr(buffer, "<GSpherical:InitialViewPitchDegrees>");
5331 sc->spherical->pitch = strtol(val, NULL, 10) * (1 << 16);
5332 val = av_stristr(buffer, "<GSpherical:InitialViewRollDegrees>");
5334 sc->spherical->roll = strtol(val, NULL, 10) * (1 << 16);
5342 static int mov_read_uuid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5345 MOVStreamContext *sc;
5348 static const uint8_t uuid_isml_manifest[] = {
5349 0xa5, 0xd4, 0x0b, 0x30, 0xe8, 0x14, 0x11, 0xdd,
5350 0xba, 0x2f, 0x08, 0x00, 0x20, 0x0c, 0x9a, 0x66
5352 static const uint8_t uuid_xmp[] = {
5353 0xbe, 0x7a, 0xcf, 0xcb, 0x97, 0xa9, 0x42, 0xe8,
5354 0x9c, 0x71, 0x99, 0x94, 0x91, 0xe3, 0xaf, 0xac
5356 static const uint8_t uuid_spherical[] = {
5357 0xff, 0xcc, 0x82, 0x63, 0xf8, 0x55, 0x4a, 0x93,
5358 0x88, 0x14, 0x58, 0x7a, 0x02, 0x52, 0x1f, 0xdd,
5361 if (atom.size < sizeof(uuid) || atom.size >= FFMIN(INT_MAX, SIZE_MAX))
5362 return AVERROR_INVALIDDATA;
5364 if (c->fc->nb_streams < 1)
5366 st = c->fc->streams[c->fc->nb_streams - 1];
5369 ret = avio_read(pb, uuid, sizeof(uuid));
5372 } else if (ret != sizeof(uuid)) {
5373 return AVERROR_INVALIDDATA;
5375 if (!memcmp(uuid, uuid_isml_manifest, sizeof(uuid))) {
5376 uint8_t *buffer, *ptr;
5378 size_t len = atom.size - sizeof(uuid);
5381 return AVERROR_INVALIDDATA;
5383 ret = avio_skip(pb, 4); // zeroes
5386 buffer = av_mallocz(len + 1);
5388 return AVERROR(ENOMEM);
5390 ret = avio_read(pb, buffer, len);
5394 } else if (ret != len) {
5396 return AVERROR_INVALIDDATA;
5400 while ((ptr = av_stristr(ptr, "systemBitrate=\""))) {
5401 ptr += sizeof("systemBitrate=\"") - 1;
5402 c->bitrates_count++;
5403 c->bitrates = av_realloc_f(c->bitrates, c->bitrates_count, sizeof(*c->bitrates));
5405 c->bitrates_count = 0;
5407 return AVERROR(ENOMEM);
5410 ret = strtol(ptr, &endptr, 10);
5411 if (ret < 0 || errno || *endptr != '"') {
5412 c->bitrates[c->bitrates_count - 1] = 0;
5414 c->bitrates[c->bitrates_count - 1] = ret;
5419 } else if (!memcmp(uuid, uuid_xmp, sizeof(uuid))) {
5421 size_t len = atom.size - sizeof(uuid);
5422 if (c->export_xmp) {
5423 buffer = av_mallocz(len + 1);
5425 return AVERROR(ENOMEM);
5427 ret = avio_read(pb, buffer, len);
5431 } else if (ret != len) {
5433 return AVERROR_INVALIDDATA;
5436 av_dict_set(&c->fc->metadata, "xmp", buffer, 0);
5439 // skip all uuid atom, which makes it fast for long uuid-xmp file
5440 ret = avio_skip(pb, len);
5444 } else if (!memcmp(uuid, uuid_spherical, sizeof(uuid))) {
5445 size_t len = atom.size - sizeof(uuid);
5446 ret = mov_parse_uuid_spherical(sc, pb, len);
5450 av_log(c->fc, AV_LOG_WARNING, "Invalid spherical metadata found\n"); }
5455 static int mov_read_free(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5458 uint8_t content[16];
5463 ret = avio_read(pb, content, FFMIN(sizeof(content), atom.size));
5469 && !memcmp(content, "Anevia\x1A\x1A", 8)
5470 && c->use_mfra_for == FF_MOV_FLAG_MFRA_AUTO) {
5471 c->use_mfra_for = FF_MOV_FLAG_MFRA_PTS;
5477 static int mov_read_frma(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5479 uint32_t format = avio_rl32(pb);
5480 MOVStreamContext *sc;
5484 if (c->fc->nb_streams < 1)
5486 st = c->fc->streams[c->fc->nb_streams - 1];
5491 case MKTAG('e','n','c','v'): // encrypted video
5492 case MKTAG('e','n','c','a'): // encrypted audio
5493 id = mov_codec_id(st, format);
5494 if (st->codecpar->codec_id != AV_CODEC_ID_NONE &&
5495 st->codecpar->codec_id != id) {
5496 av_log(c->fc, AV_LOG_WARNING,
5497 "ignoring 'frma' atom of '%.4s', stream has codec id %d\n",
5498 (char*)&format, st->codecpar->codec_id);
5502 st->codecpar->codec_id = id;
5503 sc->format = format;
5507 if (format != sc->format) {
5508 av_log(c->fc, AV_LOG_WARNING,
5509 "ignoring 'frma' atom of '%.4s', stream format is '%.4s'\n",
5510 (char*)&format, (char*)&sc->format);
5518 static int mov_read_senc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5521 MOVStreamContext *sc;
5522 size_t auxiliary_info_size;
5524 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
5527 st = c->fc->streams[c->fc->nb_streams - 1];
5530 if (sc->cenc.aes_ctr) {
5531 av_log(c->fc, AV_LOG_ERROR, "duplicate senc atom\n");
5532 return AVERROR_INVALIDDATA;
5535 avio_r8(pb); /* version */
5536 sc->cenc.use_subsamples = avio_rb24(pb) & 0x02; /* flags */
5538 avio_rb32(pb); /* entries */
5540 if (atom.size < 8 || atom.size > FFMIN(INT_MAX, SIZE_MAX)) {
5541 av_log(c->fc, AV_LOG_ERROR, "senc atom size %"PRId64" invalid\n", atom.size);
5542 return AVERROR_INVALIDDATA;
5545 /* save the auxiliary info as is */
5546 auxiliary_info_size = atom.size - 8;
5548 sc->cenc.auxiliary_info = av_malloc(auxiliary_info_size);
5549 if (!sc->cenc.auxiliary_info) {
5550 return AVERROR(ENOMEM);
5553 sc->cenc.auxiliary_info_end = sc->cenc.auxiliary_info + auxiliary_info_size;
5554 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info;
5555 sc->cenc.auxiliary_info_index = 0;
5557 if (avio_read(pb, sc->cenc.auxiliary_info, auxiliary_info_size) != auxiliary_info_size) {
5558 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info");
5559 return AVERROR_INVALIDDATA;
5562 /* initialize the cipher */
5563 sc->cenc.aes_ctr = av_aes_ctr_alloc();
5564 if (!sc->cenc.aes_ctr) {
5565 return AVERROR(ENOMEM);
5568 return av_aes_ctr_init(sc->cenc.aes_ctr, c->decryption_key);
5571 static int mov_read_saiz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5574 MOVStreamContext *sc;
5576 int atom_header_size;
5579 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
5582 st = c->fc->streams[c->fc->nb_streams - 1];
5585 if (sc->cenc.auxiliary_info_sizes || sc->cenc.auxiliary_info_default_size) {
5586 av_log(c->fc, AV_LOG_ERROR, "duplicate saiz atom\n");
5587 return AVERROR_INVALIDDATA;
5590 atom_header_size = 9;
5592 avio_r8(pb); /* version */
5593 flags = avio_rb24(pb);
5595 if ((flags & 0x01) != 0) {
5596 atom_header_size += 8;
5598 avio_rb32(pb); /* info type */
5599 avio_rb32(pb); /* info type param */
5602 sc->cenc.auxiliary_info_default_size = avio_r8(pb);
5603 avio_rb32(pb); /* entries */
5605 if (atom.size <= atom_header_size) {
5609 if (atom.size > FFMIN(INT_MAX, SIZE_MAX)) {
5610 av_log(c->fc, AV_LOG_ERROR, "saiz atom auxiliary_info_sizes size %"PRId64" invalid\n", atom.size);
5611 return AVERROR_INVALIDDATA;
5614 /* save the auxiliary info sizes as is */
5615 data_size = atom.size - atom_header_size;
5617 sc->cenc.auxiliary_info_sizes = av_malloc(data_size);
5618 if (!sc->cenc.auxiliary_info_sizes) {
5619 return AVERROR(ENOMEM);
5622 sc->cenc.auxiliary_info_sizes_count = data_size;
5624 if (avio_read(pb, sc->cenc.auxiliary_info_sizes, data_size) != data_size) {
5625 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info sizes");
5626 return AVERROR_INVALIDDATA;
5632 static int mov_read_dfla(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5635 int last, type, size, ret;
5638 if (c->fc->nb_streams < 1)
5640 st = c->fc->streams[c->fc->nb_streams-1];
5642 if ((uint64_t)atom.size > (1<<30) || atom.size < 42)
5643 return AVERROR_INVALIDDATA;
5645 /* Check FlacSpecificBox version. */
5646 if (avio_r8(pb) != 0)
5647 return AVERROR_INVALIDDATA;
5649 avio_rb24(pb); /* Flags */
5651 avio_read(pb, buf, sizeof(buf));
5652 flac_parse_block_header(buf, &last, &type, &size);
5654 if (type != FLAC_METADATA_TYPE_STREAMINFO || size != FLAC_STREAMINFO_SIZE) {
5655 av_log(c->fc, AV_LOG_ERROR, "STREAMINFO must be first FLACMetadataBlock\n");
5656 return AVERROR_INVALIDDATA;
5659 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
5664 av_log(c->fc, AV_LOG_WARNING, "non-STREAMINFO FLACMetadataBlock(s) ignored\n");
5669 static int mov_seek_auxiliary_info(MOVContext *c, MOVStreamContext *sc, int64_t index)
5671 size_t auxiliary_info_seek_offset = 0;
5674 if (sc->cenc.auxiliary_info_default_size) {
5675 auxiliary_info_seek_offset = (size_t)sc->cenc.auxiliary_info_default_size * index;
5676 } else if (sc->cenc.auxiliary_info_sizes) {
5677 if (index > sc->cenc.auxiliary_info_sizes_count) {
5678 av_log(c, AV_LOG_ERROR, "current sample %"PRId64" greater than the number of auxiliary info sample sizes %"SIZE_SPECIFIER"\n",
5679 index, sc->cenc.auxiliary_info_sizes_count);
5680 return AVERROR_INVALIDDATA;
5683 for (i = 0; i < index; i++) {
5684 auxiliary_info_seek_offset += sc->cenc.auxiliary_info_sizes[i];
5688 if (auxiliary_info_seek_offset > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info) {
5689 av_log(c, AV_LOG_ERROR, "auxiliary info offset %"SIZE_SPECIFIER" greater than auxiliary info size %"SIZE_SPECIFIER"\n",
5690 auxiliary_info_seek_offset, (size_t)(sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info));
5691 return AVERROR_INVALIDDATA;
5694 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info + auxiliary_info_seek_offset;
5695 sc->cenc.auxiliary_info_index = index;
5699 static int cenc_filter(MOVContext *c, MOVStreamContext *sc, int64_t index, uint8_t *input, int size)
5701 uint32_t encrypted_bytes;
5702 uint16_t subsample_count;
5703 uint16_t clear_bytes;
5704 uint8_t* input_end = input + size;
5707 if (index != sc->cenc.auxiliary_info_index) {
5708 ret = mov_seek_auxiliary_info(c, sc, index);
5715 if (AES_CTR_IV_SIZE > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5716 av_log(c->fc, AV_LOG_ERROR, "failed to read iv from the auxiliary info\n");
5717 return AVERROR_INVALIDDATA;
5720 av_aes_ctr_set_iv(sc->cenc.aes_ctr, sc->cenc.auxiliary_info_pos);
5721 sc->cenc.auxiliary_info_pos += AES_CTR_IV_SIZE;
5723 if (!sc->cenc.use_subsamples)
5725 /* decrypt the whole packet */
5726 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, size);
5730 /* read the subsample count */
5731 if (sizeof(uint16_t) > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5732 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample count from the auxiliary info\n");
5733 return AVERROR_INVALIDDATA;
5736 subsample_count = AV_RB16(sc->cenc.auxiliary_info_pos);
5737 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
5739 for (; subsample_count > 0; subsample_count--)
5741 if (6 > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5742 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample from the auxiliary info\n");
5743 return AVERROR_INVALIDDATA;
5746 /* read the number of clear / encrypted bytes */
5747 clear_bytes = AV_RB16(sc->cenc.auxiliary_info_pos);
5748 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
5749 encrypted_bytes = AV_RB32(sc->cenc.auxiliary_info_pos);
5750 sc->cenc.auxiliary_info_pos += sizeof(uint32_t);
5752 if ((uint64_t)clear_bytes + encrypted_bytes > input_end - input) {
5753 av_log(c->fc, AV_LOG_ERROR, "subsample size exceeds the packet size left\n");
5754 return AVERROR_INVALIDDATA;
5757 /* skip the clear bytes */
5758 input += clear_bytes;
5760 /* decrypt the encrypted bytes */
5761 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, encrypted_bytes);
5762 input += encrypted_bytes;
5765 if (input < input_end) {
5766 av_log(c->fc, AV_LOG_ERROR, "leftover packet bytes after subsample processing\n");
5767 return AVERROR_INVALIDDATA;
5770 sc->cenc.auxiliary_info_index++;
5774 static int mov_read_dops(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5776 const int OPUS_SEEK_PREROLL_MS = 80;
5781 if (c->fc->nb_streams < 1)
5783 st = c->fc->streams[c->fc->nb_streams-1];
5785 if ((uint64_t)atom.size > (1<<30) || atom.size < 11)
5786 return AVERROR_INVALIDDATA;
5788 /* Check OpusSpecificBox version. */
5789 if (avio_r8(pb) != 0) {
5790 av_log(c->fc, AV_LOG_ERROR, "unsupported OpusSpecificBox version\n");
5791 return AVERROR_INVALIDDATA;
5794 /* OpusSpecificBox size plus magic for Ogg OpusHead header. */
5795 size = atom.size + 8;
5797 if (ff_alloc_extradata(st->codecpar, size))
5798 return AVERROR(ENOMEM);
5800 AV_WL32(st->codecpar->extradata, MKTAG('O','p','u','s'));
5801 AV_WL32(st->codecpar->extradata + 4, MKTAG('H','e','a','d'));
5802 AV_WB8(st->codecpar->extradata + 8, 1); /* OpusHead version */
5803 avio_read(pb, st->codecpar->extradata + 9, size - 9);
5805 /* OpusSpecificBox is stored in big-endian, but OpusHead is
5806 little-endian; aside from the preceeding magic and version they're
5807 otherwise currently identical. Data after output gain at offset 16
5808 doesn't need to be bytewapped. */
5809 pre_skip = AV_RB16(st->codecpar->extradata + 10);
5810 AV_WL16(st->codecpar->extradata + 10, pre_skip);
5811 AV_WL32(st->codecpar->extradata + 12, AV_RB32(st->codecpar->extradata + 12));
5812 AV_WL16(st->codecpar->extradata + 16, AV_RB16(st->codecpar->extradata + 16));
5814 st->codecpar->initial_padding = pre_skip;
5815 st->codecpar->seek_preroll = av_rescale_q(OPUS_SEEK_PREROLL_MS,
5816 (AVRational){1, 1000},
5817 (AVRational){1, 48000});
5822 static const MOVParseTableEntry mov_default_parse_table[] = {
5823 { MKTAG('A','C','L','R'), mov_read_aclr },
5824 { MKTAG('A','P','R','G'), mov_read_avid },
5825 { MKTAG('A','A','L','P'), mov_read_avid },
5826 { MKTAG('A','R','E','S'), mov_read_ares },
5827 { MKTAG('a','v','s','s'), mov_read_avss },
5828 { MKTAG('c','h','p','l'), mov_read_chpl },
5829 { MKTAG('c','o','6','4'), mov_read_stco },
5830 { MKTAG('c','o','l','r'), mov_read_colr },
5831 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
5832 { MKTAG('d','i','n','f'), mov_read_default },
5833 { MKTAG('D','p','x','E'), mov_read_dpxe },
5834 { MKTAG('d','r','e','f'), mov_read_dref },
5835 { MKTAG('e','d','t','s'), mov_read_default },
5836 { MKTAG('e','l','s','t'), mov_read_elst },
5837 { MKTAG('e','n','d','a'), mov_read_enda },
5838 { MKTAG('f','i','e','l'), mov_read_fiel },
5839 { MKTAG('a','d','r','m'), mov_read_adrm },
5840 { MKTAG('f','t','y','p'), mov_read_ftyp },
5841 { MKTAG('g','l','b','l'), mov_read_glbl },
5842 { MKTAG('h','d','l','r'), mov_read_hdlr },
5843 { MKTAG('i','l','s','t'), mov_read_ilst },
5844 { MKTAG('j','p','2','h'), mov_read_jp2h },
5845 { MKTAG('m','d','a','t'), mov_read_mdat },
5846 { MKTAG('m','d','h','d'), mov_read_mdhd },
5847 { MKTAG('m','d','i','a'), mov_read_default },
5848 { MKTAG('m','e','t','a'), mov_read_meta },
5849 { MKTAG('m','i','n','f'), mov_read_default },
5850 { MKTAG('m','o','o','f'), mov_read_moof },
5851 { MKTAG('m','o','o','v'), mov_read_moov },
5852 { MKTAG('m','v','e','x'), mov_read_default },
5853 { MKTAG('m','v','h','d'), mov_read_mvhd },
5854 { MKTAG('S','M','I',' '), mov_read_svq3 },
5855 { MKTAG('a','l','a','c'), mov_read_alac }, /* alac specific atom */
5856 { MKTAG('a','v','c','C'), mov_read_glbl },
5857 { MKTAG('p','a','s','p'), mov_read_pasp },
5858 { MKTAG('s','i','d','x'), mov_read_sidx },
5859 { MKTAG('s','t','b','l'), mov_read_default },
5860 { MKTAG('s','t','c','o'), mov_read_stco },
5861 { MKTAG('s','t','p','s'), mov_read_stps },
5862 { MKTAG('s','t','r','f'), mov_read_strf },
5863 { MKTAG('s','t','s','c'), mov_read_stsc },
5864 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
5865 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
5866 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
5867 { MKTAG('s','t','t','s'), mov_read_stts },
5868 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
5869 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
5870 { MKTAG('t','f','d','t'), mov_read_tfdt },
5871 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
5872 { MKTAG('t','r','a','k'), mov_read_trak },
5873 { MKTAG('t','r','a','f'), mov_read_default },
5874 { MKTAG('t','r','e','f'), mov_read_default },
5875 { MKTAG('t','m','c','d'), mov_read_tmcd },
5876 { MKTAG('c','h','a','p'), mov_read_chap },
5877 { MKTAG('t','r','e','x'), mov_read_trex },
5878 { MKTAG('t','r','u','n'), mov_read_trun },
5879 { MKTAG('u','d','t','a'), mov_read_default },
5880 { MKTAG('w','a','v','e'), mov_read_wave },
5881 { MKTAG('e','s','d','s'), mov_read_esds },
5882 { MKTAG('d','a','c','3'), mov_read_dac3 }, /* AC-3 info */
5883 { MKTAG('d','e','c','3'), mov_read_dec3 }, /* EAC-3 info */
5884 { MKTAG('d','d','t','s'), mov_read_ddts }, /* DTS audio descriptor */
5885 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
5886 { MKTAG('w','f','e','x'), mov_read_wfex },
5887 { MKTAG('c','m','o','v'), mov_read_cmov },
5888 { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */
5889 { MKTAG('d','v','c','1'), mov_read_dvc1 },
5890 { MKTAG('s','b','g','p'), mov_read_sbgp },
5891 { MKTAG('h','v','c','C'), mov_read_glbl },
5892 { MKTAG('u','u','i','d'), mov_read_uuid },
5893 { MKTAG('C','i','n', 0x8e), mov_read_targa_y216 },
5894 { MKTAG('f','r','e','e'), mov_read_free },
5895 { MKTAG('-','-','-','-'), mov_read_custom },
5896 { MKTAG('s','i','n','f'), mov_read_default },
5897 { MKTAG('f','r','m','a'), mov_read_frma },
5898 { MKTAG('s','e','n','c'), mov_read_senc },
5899 { MKTAG('s','a','i','z'), mov_read_saiz },
5900 { MKTAG('d','f','L','a'), mov_read_dfla },
5901 { MKTAG('s','t','3','d'), mov_read_st3d }, /* stereoscopic 3D video box */
5902 { MKTAG('s','v','3','d'), mov_read_sv3d }, /* spherical video box */
5903 { MKTAG('d','O','p','s'), mov_read_dops },
5904 { MKTAG('S','m','D','m'), mov_read_smdm },
5905 { MKTAG('C','o','L','L'), mov_read_coll },
5906 { MKTAG('v','p','c','C'), mov_read_vpcc },
5910 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5912 int64_t total_size = 0;
5916 if (c->atom_depth > 10) {
5917 av_log(c->fc, AV_LOG_ERROR, "Atoms too deeply nested\n");
5918 return AVERROR_INVALIDDATA;
5923 atom.size = INT64_MAX;
5924 while (total_size <= atom.size - 8 && !avio_feof(pb)) {
5925 int (*parse)(MOVContext*, AVIOContext*, MOVAtom) = NULL;
5928 if (atom.size >= 8) {
5929 a.size = avio_rb32(pb);
5930 a.type = avio_rl32(pb);
5931 if (a.type == MKTAG('f','r','e','e') &&
5933 c->fc->strict_std_compliance < FF_COMPLIANCE_STRICT &&
5936 uint32_t *type = (uint32_t *)buf + 1;
5937 if (avio_read(pb, buf, 8) != 8)
5938 return AVERROR_INVALIDDATA;
5939 avio_seek(pb, -8, SEEK_CUR);
5940 if (*type == MKTAG('m','v','h','d') ||
5941 *type == MKTAG('c','m','o','v')) {
5942 av_log(c->fc, AV_LOG_ERROR, "Detected moov in a free atom.\n");
5943 a.type = MKTAG('m','o','o','v');
5946 if (atom.type != MKTAG('r','o','o','t') &&
5947 atom.type != MKTAG('m','o','o','v'))
5949 if (a.type == MKTAG('t','r','a','k') || a.type == MKTAG('m','d','a','t'))
5951 av_log(c->fc, AV_LOG_ERROR, "Broken file, trak/mdat not at top-level\n");
5958 if (a.size == 1 && total_size + 8 <= atom.size) { /* 64 bit extended size */
5959 a.size = avio_rb64(pb) - 8;
5963 av_log(c->fc, AV_LOG_TRACE, "type:'%s' parent:'%s' sz: %"PRId64" %"PRId64" %"PRId64"\n",
5964 av_fourcc2str(a.type), av_fourcc2str(atom.type), a.size, total_size, atom.size);
5966 a.size = atom.size - total_size + 8;
5971 a.size = FFMIN(a.size, atom.size - total_size);
5973 for (i = 0; mov_default_parse_table[i].type; i++)
5974 if (mov_default_parse_table[i].type == a.type) {
5975 parse = mov_default_parse_table[i].parse;
5979 // container is user data
5980 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
5981 atom.type == MKTAG('i','l','s','t')))
5982 parse = mov_read_udta_string;
5984 // Supports parsing the QuickTime Metadata Keys.
5985 // https://developer.apple.com/library/mac/documentation/QuickTime/QTFF/Metadata/Metadata.html
5986 if (!parse && c->found_hdlr_mdta &&
5987 atom.type == MKTAG('m','e','t','a') &&
5988 a.type == MKTAG('k','e','y','s')) {
5989 parse = mov_read_keys;
5992 if (!parse) { /* skip leaf atoms data */
5993 avio_skip(pb, a.size);
5995 int64_t start_pos = avio_tell(pb);
5997 int err = parse(c, pb, a);
6002 if (c->found_moov && c->found_mdat &&
6003 ((!(pb->seekable & AVIO_SEEKABLE_NORMAL) || c->fc->flags & AVFMT_FLAG_IGNIDX || c->frag_index.complete) ||
6004 start_pos + a.size == avio_size(pb))) {
6005 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) || c->fc->flags & AVFMT_FLAG_IGNIDX || c->frag_index.complete)
6006 c->next_root_atom = start_pos + a.size;
6010 left = a.size - avio_tell(pb) + start_pos;
6011 if (left > 0) /* skip garbage at atom end */
6012 avio_skip(pb, left);
6013 else if (left < 0) {
6014 av_log(c->fc, AV_LOG_WARNING,
6015 "overread end of atom '%.4s' by %"PRId64" bytes\n",
6016 (char*)&a.type, -left);
6017 avio_seek(pb, left, SEEK_CUR);
6021 total_size += a.size;
6024 if (total_size < atom.size && atom.size < 0x7ffff)
6025 avio_skip(pb, atom.size - total_size);
6031 static int mov_probe(AVProbeData *p)
6036 int moov_offset = -1;
6038 /* check file header */
6041 /* ignore invalid offset */
6042 if ((offset + 8) > (unsigned int)p->buf_size)
6044 tag = AV_RL32(p->buf + offset + 4);
6046 /* check for obvious tags */
6047 case MKTAG('m','o','o','v'):
6048 moov_offset = offset + 4;
6049 case MKTAG('m','d','a','t'):
6050 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
6051 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
6052 case MKTAG('f','t','y','p'):
6053 if (AV_RB32(p->buf+offset) < 8 &&
6054 (AV_RB32(p->buf+offset) != 1 ||
6055 offset + 12 > (unsigned int)p->buf_size ||
6056 AV_RB64(p->buf+offset + 8) == 0)) {
6057 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
6058 } else if (tag == MKTAG('f','t','y','p') &&
6059 ( AV_RL32(p->buf + offset + 8) == MKTAG('j','p','2',' ')
6060 || AV_RL32(p->buf + offset + 8) == MKTAG('j','p','x',' ')
6062 score = FFMAX(score, 5);
6064 score = AVPROBE_SCORE_MAX;
6066 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6068 /* those are more common words, so rate then a bit less */
6069 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
6070 case MKTAG('w','i','d','e'):
6071 case MKTAG('f','r','e','e'):
6072 case MKTAG('j','u','n','k'):
6073 case MKTAG('p','i','c','t'):
6074 score = FFMAX(score, AVPROBE_SCORE_MAX - 5);
6075 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6077 case MKTAG(0x82,0x82,0x7f,0x7d):
6078 case MKTAG('s','k','i','p'):
6079 case MKTAG('u','u','i','d'):
6080 case MKTAG('p','r','f','l'):
6081 /* if we only find those cause probedata is too small at least rate them */
6082 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
6083 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6086 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6089 if(score > AVPROBE_SCORE_MAX - 50 && moov_offset != -1) {
6090 /* moov atom in the header - we should make sure that this is not a
6091 * MOV-packed MPEG-PS */
6092 offset = moov_offset;
6094 while(offset < (p->buf_size - 16)){ /* Sufficient space */
6095 /* We found an actual hdlr atom */
6096 if(AV_RL32(p->buf + offset ) == MKTAG('h','d','l','r') &&
6097 AV_RL32(p->buf + offset + 8) == MKTAG('m','h','l','r') &&
6098 AV_RL32(p->buf + offset + 12) == MKTAG('M','P','E','G')){
6099 av_log(NULL, AV_LOG_WARNING, "Found media data tag MPEG indicating this is a MOV-packed MPEG-PS.\n");
6100 /* We found a media handler reference atom describing an
6101 * MPEG-PS-in-MOV, return a
6102 * low score to force expanding the probe window until
6103 * mpegps_probe finds what it needs */
6114 // must be done after parsing all trak because there's no order requirement
6115 static void mov_read_chapters(AVFormatContext *s)
6117 MOVContext *mov = s->priv_data;
6119 MOVStreamContext *sc;
6124 for (j = 0; j < mov->nb_chapter_tracks; j++) {
6125 chapter_track = mov->chapter_tracks[j];
6127 for (i = 0; i < s->nb_streams; i++)
6128 if (s->streams[i]->id == chapter_track) {
6133 av_log(s, AV_LOG_ERROR, "Referenced QT chapter track not found\n");
6138 cur_pos = avio_tell(sc->pb);
6140 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
6141 st->disposition |= AV_DISPOSITION_ATTACHED_PIC | AV_DISPOSITION_TIMED_THUMBNAILS;
6142 if (st->nb_index_entries) {
6143 // Retrieve the first frame, if possible
6145 AVIndexEntry *sample = &st->index_entries[0];
6146 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
6147 av_log(s, AV_LOG_ERROR, "Failed to retrieve first frame\n");
6151 if (av_get_packet(sc->pb, &pkt, sample->size) < 0)
6154 st->attached_pic = pkt;
6155 st->attached_pic.stream_index = st->index;
6156 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
6159 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
6160 st->codecpar->codec_id = AV_CODEC_ID_BIN_DATA;
6161 st->discard = AVDISCARD_ALL;
6162 for (i = 0; i < st->nb_index_entries; i++) {
6163 AVIndexEntry *sample = &st->index_entries[i];
6164 int64_t end = i+1 < st->nb_index_entries ? st->index_entries[i+1].timestamp : st->duration;
6169 if (end < sample->timestamp) {
6170 av_log(s, AV_LOG_WARNING, "ignoring stream duration which is shorter than chapters\n");
6171 end = AV_NOPTS_VALUE;
6174 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
6175 av_log(s, AV_LOG_ERROR, "Chapter %d not found in file\n", i);
6179 // the first two bytes are the length of the title
6180 len = avio_rb16(sc->pb);
6181 if (len > sample->size-2)
6183 title_len = 2*len + 1;
6184 if (!(title = av_mallocz(title_len)))
6187 // The samples could theoretically be in any encoding if there's an encd
6188 // atom following, but in practice are only utf-8 or utf-16, distinguished
6189 // instead by the presence of a BOM
6193 ch = avio_rb16(sc->pb);
6195 avio_get_str16be(sc->pb, len, title, title_len);
6196 else if (ch == 0xfffe)
6197 avio_get_str16le(sc->pb, len, title, title_len);
6200 if (len == 1 || len == 2)
6203 avio_get_str(sc->pb, INT_MAX, title + 2, len - 1);
6207 avpriv_new_chapter(s, i, st->time_base, sample->timestamp, end, title);
6212 avio_seek(sc->pb, cur_pos, SEEK_SET);
6216 static int parse_timecode_in_framenum_format(AVFormatContext *s, AVStream *st,
6217 uint32_t value, int flags)
6220 char buf[AV_TIMECODE_STR_SIZE];
6221 AVRational rate = st->avg_frame_rate;
6222 int ret = av_timecode_init(&tc, rate, flags, 0, s);
6225 av_dict_set(&st->metadata, "timecode",
6226 av_timecode_make_string(&tc, buf, value), 0);
6230 static int mov_read_rtmd_track(AVFormatContext *s, AVStream *st)
6232 MOVStreamContext *sc = st->priv_data;
6233 char buf[AV_TIMECODE_STR_SIZE];
6234 int64_t cur_pos = avio_tell(sc->pb);
6235 int hh, mm, ss, ff, drop;
6237 if (!st->nb_index_entries)
6240 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
6241 avio_skip(s->pb, 13);
6242 hh = avio_r8(s->pb);
6243 mm = avio_r8(s->pb);
6244 ss = avio_r8(s->pb);
6245 drop = avio_r8(s->pb);
6246 ff = avio_r8(s->pb);
6247 snprintf(buf, AV_TIMECODE_STR_SIZE, "%02d:%02d:%02d%c%02d",
6248 hh, mm, ss, drop ? ';' : ':', ff);
6249 av_dict_set(&st->metadata, "timecode", buf, 0);
6251 avio_seek(sc->pb, cur_pos, SEEK_SET);
6255 static int mov_read_timecode_track(AVFormatContext *s, AVStream *st)
6257 MOVStreamContext *sc = st->priv_data;
6259 int64_t cur_pos = avio_tell(sc->pb);
6262 if (!st->nb_index_entries)
6265 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
6266 value = avio_rb32(s->pb);
6268 if (sc->tmcd_flags & 0x0001) flags |= AV_TIMECODE_FLAG_DROPFRAME;
6269 if (sc->tmcd_flags & 0x0002) flags |= AV_TIMECODE_FLAG_24HOURSMAX;
6270 if (sc->tmcd_flags & 0x0004) flags |= AV_TIMECODE_FLAG_ALLOWNEGATIVE;
6272 /* Assume Counter flag is set to 1 in tmcd track (even though it is likely
6273 * not the case) and thus assume "frame number format" instead of QT one.
6274 * No sample with tmcd track can be found with a QT timecode at the moment,
6275 * despite what the tmcd track "suggests" (Counter flag set to 0 means QT
6277 parse_timecode_in_framenum_format(s, st, value, flags);
6279 avio_seek(sc->pb, cur_pos, SEEK_SET);
6283 static int mov_read_close(AVFormatContext *s)
6285 MOVContext *mov = s->priv_data;
6288 for (i = 0; i < s->nb_streams; i++) {
6289 AVStream *st = s->streams[i];
6290 MOVStreamContext *sc = st->priv_data;
6295 av_freep(&sc->ctts_data);
6296 for (j = 0; j < sc->drefs_count; j++) {
6297 av_freep(&sc->drefs[j].path);
6298 av_freep(&sc->drefs[j].dir);
6300 av_freep(&sc->drefs);
6302 sc->drefs_count = 0;
6304 if (!sc->pb_is_copied)
6305 ff_format_io_close(s, &sc->pb);
6308 av_freep(&sc->chunk_offsets);
6309 av_freep(&sc->stsc_data);
6310 av_freep(&sc->sample_sizes);
6311 av_freep(&sc->keyframes);
6312 av_freep(&sc->stts_data);
6313 av_freep(&sc->stps_data);
6314 av_freep(&sc->elst_data);
6315 av_freep(&sc->rap_group);
6316 av_freep(&sc->display_matrix);
6317 av_freep(&sc->index_ranges);
6320 for (j = 0; j < sc->stsd_count; j++)
6321 av_free(sc->extradata[j]);
6322 av_freep(&sc->extradata);
6323 av_freep(&sc->extradata_size);
6325 av_freep(&sc->cenc.auxiliary_info);
6326 av_freep(&sc->cenc.auxiliary_info_sizes);
6327 av_aes_ctr_free(sc->cenc.aes_ctr);
6329 av_freep(&sc->stereo3d);
6330 av_freep(&sc->spherical);
6331 av_freep(&sc->mastering);
6332 av_freep(&sc->coll);
6335 if (mov->dv_demux) {
6336 avformat_free_context(mov->dv_fctx);
6337 mov->dv_fctx = NULL;
6340 if (mov->meta_keys) {
6341 for (i = 1; i < mov->meta_keys_count; i++) {
6342 av_freep(&mov->meta_keys[i]);
6344 av_freep(&mov->meta_keys);
6347 av_freep(&mov->trex_data);
6348 av_freep(&mov->bitrates);
6350 for (i = 0; i < mov->frag_index.nb_items; i++) {
6351 av_freep(&mov->frag_index.item[i].stream_info);
6353 av_freep(&mov->frag_index.item);
6355 av_freep(&mov->aes_decrypt);
6356 av_freep(&mov->chapter_tracks);
6361 static int tmcd_is_referenced(AVFormatContext *s, int tmcd_id)
6365 for (i = 0; i < s->nb_streams; i++) {
6366 AVStream *st = s->streams[i];
6367 MOVStreamContext *sc = st->priv_data;
6369 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO &&
6370 sc->timecode_track == tmcd_id)
6376 /* look for a tmcd track not referenced by any video track, and export it globally */
6377 static void export_orphan_timecode(AVFormatContext *s)
6381 for (i = 0; i < s->nb_streams; i++) {
6382 AVStream *st = s->streams[i];
6384 if (st->codecpar->codec_tag == MKTAG('t','m','c','d') &&
6385 !tmcd_is_referenced(s, i + 1)) {
6386 AVDictionaryEntry *tcr = av_dict_get(st->metadata, "timecode", NULL, 0);
6388 av_dict_set(&s->metadata, "timecode", tcr->value, 0);
6395 static int read_tfra(MOVContext *mov, AVIOContext *f)
6397 int version, fieldlength, i, j;
6398 int64_t pos = avio_tell(f);
6399 uint32_t size = avio_rb32(f);
6400 unsigned track_id, item_count;
6402 if (avio_rb32(f) != MKBETAG('t', 'f', 'r', 'a')) {
6405 av_log(mov->fc, AV_LOG_VERBOSE, "found tfra\n");
6407 version = avio_r8(f);
6409 track_id = avio_rb32(f);
6410 fieldlength = avio_rb32(f);
6411 item_count = avio_rb32(f);
6412 for (i = 0; i < item_count; i++) {
6413 int64_t time, offset;
6415 MOVFragmentStreamInfo * frag_stream_info;
6418 return AVERROR_INVALIDDATA;
6422 time = avio_rb64(f);
6423 offset = avio_rb64(f);
6425 time = avio_rb32(f);
6426 offset = avio_rb32(f);
6429 // The first sample of each stream in a fragment is always a random
6430 // access sample. So it's entry in the tfra can be used as the
6431 // initial PTS of the fragment.
6432 index = update_frag_index(mov, offset);
6433 frag_stream_info = get_frag_stream_info(&mov->frag_index, index, track_id);
6434 if (frag_stream_info &&
6435 frag_stream_info->first_tfra_pts == AV_NOPTS_VALUE)
6436 frag_stream_info->first_tfra_pts = time;
6438 for (j = 0; j < ((fieldlength >> 4) & 3) + 1; j++)
6440 for (j = 0; j < ((fieldlength >> 2) & 3) + 1; j++)
6442 for (j = 0; j < ((fieldlength >> 0) & 3) + 1; j++)
6446 avio_seek(f, pos + size, SEEK_SET);
6450 static int mov_read_mfra(MOVContext *c, AVIOContext *f)
6452 int64_t stream_size = avio_size(f);
6453 int64_t original_pos = avio_tell(f);
6457 if ((seek_ret = avio_seek(f, stream_size - 4, SEEK_SET)) < 0) {
6461 mfra_size = avio_rb32(f);
6462 if (mfra_size < 0 || mfra_size > stream_size) {
6463 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (unreasonable size)\n");
6466 if ((seek_ret = avio_seek(f, -mfra_size, SEEK_CUR)) < 0) {
6470 if (avio_rb32(f) != mfra_size) {
6471 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (size mismatch)\n");
6474 if (avio_rb32(f) != MKBETAG('m', 'f', 'r', 'a')) {
6475 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (tag mismatch)\n");
6478 av_log(c->fc, AV_LOG_VERBOSE, "stream has mfra\n");
6480 ret = read_tfra(c, f);
6486 seek_ret = avio_seek(f, original_pos, SEEK_SET);
6488 av_log(c->fc, AV_LOG_ERROR,
6489 "failed to seek back after looking for mfra\n");
6495 static int mov_read_header(AVFormatContext *s)
6497 MOVContext *mov = s->priv_data;
6498 AVIOContext *pb = s->pb;
6500 MOVAtom atom = { AV_RL32("root") };
6503 if (mov->decryption_key_len != 0 && mov->decryption_key_len != AES_CTR_KEY_SIZE) {
6504 av_log(s, AV_LOG_ERROR, "Invalid decryption key len %d expected %d\n",
6505 mov->decryption_key_len, AES_CTR_KEY_SIZE);
6506 return AVERROR(EINVAL);
6510 mov->trak_index = -1;
6511 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
6512 if (pb->seekable & AVIO_SEEKABLE_NORMAL)
6513 atom.size = avio_size(pb);
6515 atom.size = INT64_MAX;
6517 /* check MOV header */
6519 if (mov->moov_retry)
6520 avio_seek(pb, 0, SEEK_SET);
6521 if ((err = mov_read_default(mov, pb, atom)) < 0) {
6522 av_log(s, AV_LOG_ERROR, "error reading header\n");
6526 } while ((pb->seekable & AVIO_SEEKABLE_NORMAL) && !mov->found_moov && !mov->moov_retry++);
6527 if (!mov->found_moov) {
6528 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
6530 return AVERROR_INVALIDDATA;
6532 av_log(mov->fc, AV_LOG_TRACE, "on_parse_exit_offset=%"PRId64"\n", avio_tell(pb));
6534 if (pb->seekable & AVIO_SEEKABLE_NORMAL) {
6535 if (mov->nb_chapter_tracks > 0 && !mov->ignore_chapters)
6536 mov_read_chapters(s);
6537 for (i = 0; i < s->nb_streams; i++)
6538 if (s->streams[i]->codecpar->codec_tag == AV_RL32("tmcd")) {
6539 mov_read_timecode_track(s, s->streams[i]);
6540 } else if (s->streams[i]->codecpar->codec_tag == AV_RL32("rtmd")) {
6541 mov_read_rtmd_track(s, s->streams[i]);
6545 /* copy timecode metadata from tmcd tracks to the related video streams */
6546 for (i = 0; i < s->nb_streams; i++) {
6547 AVStream *st = s->streams[i];
6548 MOVStreamContext *sc = st->priv_data;
6549 if (sc->timecode_track > 0) {
6550 AVDictionaryEntry *tcr;
6551 int tmcd_st_id = -1;
6553 for (j = 0; j < s->nb_streams; j++)
6554 if (s->streams[j]->id == sc->timecode_track)
6557 if (tmcd_st_id < 0 || tmcd_st_id == i)
6559 tcr = av_dict_get(s->streams[tmcd_st_id]->metadata, "timecode", NULL, 0);
6561 av_dict_set(&st->metadata, "timecode", tcr->value, 0);
6564 export_orphan_timecode(s);
6566 for (i = 0; i < s->nb_streams; i++) {
6567 AVStream *st = s->streams[i];
6568 MOVStreamContext *sc = st->priv_data;
6569 fix_timescale(mov, sc);
6570 if(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id == AV_CODEC_ID_AAC) {
6571 st->skip_samples = sc->start_pad;
6573 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && sc->nb_frames_for_fps > 0 && sc->duration_for_fps > 0)
6574 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
6575 sc->time_scale*(int64_t)sc->nb_frames_for_fps, sc->duration_for_fps, INT_MAX);
6576 if (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE) {
6577 if (st->codecpar->width <= 0 || st->codecpar->height <= 0) {
6578 st->codecpar->width = sc->width;
6579 st->codecpar->height = sc->height;
6581 if (st->codecpar->codec_id == AV_CODEC_ID_DVD_SUBTITLE) {
6582 if ((err = mov_rewrite_dvd_sub_extradata(st)) < 0)
6586 if (mov->handbrake_version &&
6587 mov->handbrake_version <= 1000000*0 + 1000*10 + 2 && // 0.10.2
6588 st->codecpar->codec_id == AV_CODEC_ID_MP3
6590 av_log(s, AV_LOG_VERBOSE, "Forcing full parsing for mp3 stream\n");
6591 st->need_parsing = AVSTREAM_PARSE_FULL;
6595 if (mov->trex_data) {
6596 for (i = 0; i < s->nb_streams; i++) {
6597 AVStream *st = s->streams[i];
6598 MOVStreamContext *sc = st->priv_data;
6599 if (st->duration > 0) {
6600 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
6601 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
6602 sc->data_size, sc->time_scale);
6604 return AVERROR_INVALIDDATA;
6606 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale / st->duration;
6611 if (mov->use_mfra_for > 0) {
6612 for (i = 0; i < s->nb_streams; i++) {
6613 AVStream *st = s->streams[i];
6614 MOVStreamContext *sc = st->priv_data;
6615 if (sc->duration_for_fps > 0) {
6616 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
6617 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
6618 sc->data_size, sc->time_scale);
6620 return AVERROR_INVALIDDATA;
6622 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale /
6623 sc->duration_for_fps;
6628 for (i = 0; i < mov->bitrates_count && i < s->nb_streams; i++) {
6629 if (mov->bitrates[i]) {
6630 s->streams[i]->codecpar->bit_rate = mov->bitrates[i];
6634 ff_rfps_calculate(s);
6636 for (i = 0; i < s->nb_streams; i++) {
6637 AVStream *st = s->streams[i];
6638 MOVStreamContext *sc = st->priv_data;
6640 switch (st->codecpar->codec_type) {
6641 case AVMEDIA_TYPE_AUDIO:
6642 err = ff_replaygain_export(st, s->metadata);
6648 case AVMEDIA_TYPE_VIDEO:
6649 if (sc->display_matrix) {
6650 err = av_stream_add_side_data(st, AV_PKT_DATA_DISPLAYMATRIX, (uint8_t*)sc->display_matrix,
6651 sizeof(int32_t) * 9);
6655 sc->display_matrix = NULL;
6658 err = av_stream_add_side_data(st, AV_PKT_DATA_STEREO3D,
6659 (uint8_t *)sc->stereo3d,
6660 sizeof(*sc->stereo3d));
6664 sc->stereo3d = NULL;
6666 if (sc->spherical) {
6667 err = av_stream_add_side_data(st, AV_PKT_DATA_SPHERICAL,
6668 (uint8_t *)sc->spherical,
6669 sc->spherical_size);
6673 sc->spherical = NULL;
6675 if (sc->mastering) {
6676 err = av_stream_add_side_data(st, AV_PKT_DATA_MASTERING_DISPLAY_METADATA,
6677 (uint8_t *)sc->mastering,
6678 sizeof(*sc->mastering));
6682 sc->mastering = NULL;
6685 err = av_stream_add_side_data(st, AV_PKT_DATA_CONTENT_LIGHT_LEVEL,
6686 (uint8_t *)sc->coll,
6696 ff_configure_buffers_for_index(s, AV_TIME_BASE);
6698 for (i = 0; i < mov->frag_index.nb_items; i++)
6699 if (mov->frag_index.item[i].moof_offset <= mov->fragment.moof_offset)
6700 mov->frag_index.item[i].headers_read = 1;
6705 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
6707 AVIndexEntry *sample = NULL;
6708 int64_t best_dts = INT64_MAX;
6710 for (i = 0; i < s->nb_streams; i++) {
6711 AVStream *avst = s->streams[i];
6712 MOVStreamContext *msc = avst->priv_data;
6713 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
6714 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
6715 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
6716 av_log(s, AV_LOG_TRACE, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
6717 if (!sample || (!(s->pb->seekable & AVIO_SEEKABLE_NORMAL) && current_sample->pos < sample->pos) ||
6718 ((s->pb->seekable & AVIO_SEEKABLE_NORMAL) &&
6719 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
6720 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
6721 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
6722 sample = current_sample;
6731 static int should_retry(AVIOContext *pb, int error_code) {
6732 if (error_code == AVERROR_EOF || avio_feof(pb))
6738 static int mov_switch_root(AVFormatContext *s, int64_t target, int index)
6740 MOVContext *mov = s->priv_data;
6742 if (index >= 0 && index < mov->frag_index.nb_items)
6743 target = mov->frag_index.item[index].moof_offset;
6744 if (avio_seek(s->pb, target, SEEK_SET) != target) {
6745 av_log(mov->fc, AV_LOG_ERROR, "root atom offset 0x%"PRIx64": partial file\n", target);
6746 return AVERROR_INVALIDDATA;
6749 mov->next_root_atom = 0;
6750 if (index < 0 || index >= mov->frag_index.nb_items)
6751 index = search_frag_moof_offset(&mov->frag_index, target);
6752 if (index < mov->frag_index.nb_items) {
6753 if (index + 1 < mov->frag_index.nb_items)
6754 mov->next_root_atom = mov->frag_index.item[index + 1].moof_offset;
6755 if (mov->frag_index.item[index].headers_read)
6757 mov->frag_index.item[index].headers_read = 1;
6760 mov->found_mdat = 0;
6762 if (mov_read_default(mov, s->pb, (MOVAtom){ AV_RL32("root"), INT64_MAX }) < 0 ||
6765 av_log(s, AV_LOG_TRACE, "read fragments, offset 0x%"PRIx64"\n", avio_tell(s->pb));
6770 static int mov_change_extradata(MOVStreamContext *sc, AVPacket *pkt)
6772 uint8_t *side, *extradata;
6775 /* Save the current index. */
6776 sc->last_stsd_index = sc->stsc_data[sc->stsc_index].id - 1;
6778 /* Notify the decoder that extradata changed. */
6779 extradata_size = sc->extradata_size[sc->last_stsd_index];
6780 extradata = sc->extradata[sc->last_stsd_index];
6781 if (extradata_size > 0 && extradata) {
6782 side = av_packet_new_side_data(pkt,
6783 AV_PKT_DATA_NEW_EXTRADATA,
6786 return AVERROR(ENOMEM);
6787 memcpy(side, extradata, extradata_size);
6793 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
6795 MOVContext *mov = s->priv_data;
6796 MOVStreamContext *sc;
6797 AVIndexEntry *sample;
6798 AVStream *st = NULL;
6799 int64_t current_index;
6803 sample = mov_find_next_sample(s, &st);
6804 if (!sample || (mov->next_root_atom && sample->pos > mov->next_root_atom)) {
6805 if (!mov->next_root_atom)
6807 if ((ret = mov_switch_root(s, mov->next_root_atom, -1)) < 0)
6812 /* must be done just before reading, to avoid infinite loop on sample */
6813 current_index = sc->current_index;
6814 mov_current_sample_inc(sc);
6816 if (mov->next_root_atom) {
6817 sample->pos = FFMIN(sample->pos, mov->next_root_atom);
6818 sample->size = FFMIN(sample->size, (mov->next_root_atom - sample->pos));
6821 if (st->discard != AVDISCARD_ALL) {
6822 int64_t ret64 = avio_seek(sc->pb, sample->pos, SEEK_SET);
6823 if (ret64 != sample->pos) {
6824 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
6825 sc->ffindex, sample->pos);
6826 if (should_retry(sc->pb, ret64)) {
6827 mov_current_sample_dec(sc);
6829 return AVERROR_INVALIDDATA;
6832 if( st->discard == AVDISCARD_NONKEY && 0==(sample->flags & AVINDEX_KEYFRAME) ) {
6833 av_log(mov->fc, AV_LOG_DEBUG, "Nonkey frame from stream %d discarded due to AVDISCARD_NONKEY\n", sc->ffindex);
6837 ret = av_get_packet(sc->pb, pkt, sample->size);
6839 if (should_retry(sc->pb, ret)) {
6840 mov_current_sample_dec(sc);
6844 if (sc->has_palette) {
6847 pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, AVPALETTE_SIZE);
6849 av_log(mov->fc, AV_LOG_ERROR, "Cannot append palette to packet\n");
6851 memcpy(pal, sc->palette, AVPALETTE_SIZE);
6852 sc->has_palette = 0;
6855 #if CONFIG_DV_DEMUXER
6856 if (mov->dv_demux && sc->dv_audio_container) {
6857 avpriv_dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size, pkt->pos);
6858 av_freep(&pkt->data);
6860 ret = avpriv_dv_get_packet(mov->dv_demux, pkt);
6865 if (st->codecpar->codec_id == AV_CODEC_ID_MP3 && !st->need_parsing && pkt->size > 4) {
6866 if (ff_mpa_check_header(AV_RB32(pkt->data)) < 0)
6867 st->need_parsing = AVSTREAM_PARSE_FULL;
6871 pkt->stream_index = sc->ffindex;
6872 pkt->dts = sample->timestamp;
6873 if (sample->flags & AVINDEX_DISCARD_FRAME) {
6874 pkt->flags |= AV_PKT_FLAG_DISCARD;
6876 if (sc->ctts_data && sc->ctts_index < sc->ctts_count) {
6877 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
6878 /* update ctts context */
6880 if (sc->ctts_index < sc->ctts_count &&
6881 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
6883 sc->ctts_sample = 0;
6886 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
6887 st->index_entries[sc->current_sample].timestamp : st->duration;
6888 pkt->duration = next_dts - pkt->dts;
6889 pkt->pts = pkt->dts;
6891 if (st->discard == AVDISCARD_ALL)
6893 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? AV_PKT_FLAG_KEY : 0;
6894 pkt->pos = sample->pos;
6896 /* Multiple stsd handling. */
6897 if (sc->stsc_data) {
6898 /* Keep track of the stsc index for the given sample, then check
6899 * if the stsd index is different from the last used one. */
6901 if (mov_stsc_index_valid(sc->stsc_index, sc->stsc_count) &&
6902 mov_get_stsc_samples(sc, sc->stsc_index) == sc->stsc_sample) {
6904 sc->stsc_sample = 0;
6905 /* Do not check indexes after a switch. */
6906 } else if (sc->stsc_data[sc->stsc_index].id > 0 &&
6907 sc->stsc_data[sc->stsc_index].id - 1 < sc->stsd_count &&
6908 sc->stsc_data[sc->stsc_index].id - 1 != sc->last_stsd_index) {
6909 ret = mov_change_extradata(sc, pkt);
6916 aax_filter(pkt->data, pkt->size, mov);
6918 if (sc->cenc.aes_ctr) {
6919 ret = cenc_filter(mov, sc, current_index, pkt->data, pkt->size);
6928 static int mov_seek_fragment(AVFormatContext *s, AVStream *st, int64_t timestamp)
6930 MOVContext *mov = s->priv_data;
6933 if (!mov->frag_index.complete)
6936 index = search_frag_timestamp(&mov->frag_index, st, timestamp);
6939 if (!mov->frag_index.item[index].headers_read)
6940 return mov_switch_root(s, -1, index);
6941 if (index + 1 < mov->frag_index.nb_items)
6942 mov->next_root_atom = mov->frag_index.item[index + 1].moof_offset;
6947 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
6949 MOVStreamContext *sc = st->priv_data;
6950 int sample, time_sample, ret;
6953 timestamp -= sc->time_offset;
6955 ret = mov_seek_fragment(s, st, timestamp);
6959 sample = av_index_search_timestamp(st, timestamp, flags);
6960 av_log(s, AV_LOG_TRACE, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
6961 if (sample < 0 && st->nb_index_entries && timestamp < st->index_entries[0].timestamp)
6963 if (sample < 0) /* not sure what to do */
6964 return AVERROR_INVALIDDATA;
6965 mov_current_sample_set(sc, sample);
6966 av_log(s, AV_LOG_TRACE, "stream %d, found sample %d\n", st->index, sc->current_sample);
6967 /* adjust ctts index */
6968 if (sc->ctts_data) {
6970 for (i = 0; i < sc->ctts_count; i++) {
6971 int next = time_sample + sc->ctts_data[i].count;
6972 if (next > sc->current_sample) {
6974 sc->ctts_sample = sc->current_sample - time_sample;
6981 /* adjust stsd index */
6983 for (i = 0; i < sc->stsc_count; i++) {
6984 int next = time_sample + mov_get_stsc_samples(sc, i);
6985 if (next > sc->current_sample) {
6987 sc->stsc_sample = sc->current_sample - time_sample;
6996 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
6998 MOVContext *mc = s->priv_data;
7003 if (stream_index >= s->nb_streams)
7004 return AVERROR_INVALIDDATA;
7006 st = s->streams[stream_index];
7007 sample = mov_seek_stream(s, st, sample_time, flags);
7011 if (mc->seek_individually) {
7012 /* adjust seek timestamp to found sample timestamp */
7013 int64_t seek_timestamp = st->index_entries[sample].timestamp;
7015 for (i = 0; i < s->nb_streams; i++) {
7017 MOVStreamContext *sc = s->streams[i]->priv_data;
7019 st->skip_samples = (sample_time <= 0) ? sc->start_pad : 0;
7021 if (stream_index == i)
7024 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
7025 mov_seek_stream(s, st, timestamp, flags);
7028 for (i = 0; i < s->nb_streams; i++) {
7029 MOVStreamContext *sc;
7032 mov_current_sample_set(sc, 0);
7035 MOVStreamContext *sc;
7036 AVIndexEntry *entry = mov_find_next_sample(s, &st);
7038 return AVERROR_INVALIDDATA;
7040 if (sc->ffindex == stream_index && sc->current_sample == sample)
7042 mov_current_sample_inc(sc);
7048 #define OFFSET(x) offsetof(MOVContext, x)
7049 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
7050 static const AVOption mov_options[] = {
7051 {"use_absolute_path",
7052 "allow using absolute path when opening alias, this is a possible security issue",
7053 OFFSET(use_absolute_path), AV_OPT_TYPE_BOOL, {.i64 = 0},
7055 {"seek_streams_individually",
7056 "Seek each stream individually to the to the closest point",
7057 OFFSET(seek_individually), AV_OPT_TYPE_BOOL, { .i64 = 1 },
7059 {"ignore_editlist", "Ignore the edit list atom.", OFFSET(ignore_editlist), AV_OPT_TYPE_BOOL, {.i64 = 0},
7061 {"advanced_editlist",
7062 "Modify the AVIndex according to the editlists. Use this option to decode in the order specified by the edits.",
7063 OFFSET(advanced_editlist), AV_OPT_TYPE_BOOL, {.i64 = 1},
7065 {"ignore_chapters", "", OFFSET(ignore_chapters), AV_OPT_TYPE_BOOL, {.i64 = 0},
7068 "use mfra for fragment timestamps",
7069 OFFSET(use_mfra_for), AV_OPT_TYPE_INT, {.i64 = FF_MOV_FLAG_MFRA_AUTO},
7070 -1, FF_MOV_FLAG_MFRA_PTS, FLAGS,
7072 {"auto", "auto", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_AUTO}, 0, 0,
7073 FLAGS, "use_mfra_for" },
7074 {"dts", "dts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_DTS}, 0, 0,
7075 FLAGS, "use_mfra_for" },
7076 {"pts", "pts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_PTS}, 0, 0,
7077 FLAGS, "use_mfra_for" },
7078 { "export_all", "Export unrecognized metadata entries", OFFSET(export_all),
7079 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
7080 { "export_xmp", "Export full XMP metadata", OFFSET(export_xmp),
7081 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
7082 { "activation_bytes", "Secret bytes for Audible AAX files", OFFSET(activation_bytes),
7083 AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
7084 { "audible_fixed_key", // extracted from libAAX_SDK.so and AAXSDKWin.dll files!
7085 "Fixed key used for handling Audible AAX files", OFFSET(audible_fixed_key),
7086 AV_OPT_TYPE_BINARY, {.str="77214d4b196a87cd520045fd20a51d67"},
7087 .flags = AV_OPT_FLAG_DECODING_PARAM },
7088 { "decryption_key", "The media decryption key (hex)", OFFSET(decryption_key), AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
7089 { "enable_drefs", "Enable external track support.", OFFSET(enable_drefs), AV_OPT_TYPE_BOOL,
7090 {.i64 = 0}, 0, 1, FLAGS },
7095 static const AVClass mov_class = {
7096 .class_name = "mov,mp4,m4a,3gp,3g2,mj2",
7097 .item_name = av_default_item_name,
7098 .option = mov_options,
7099 .version = LIBAVUTIL_VERSION_INT,
7102 AVInputFormat ff_mov_demuxer = {
7103 .name = "mov,mp4,m4a,3gp,3g2,mj2",
7104 .long_name = NULL_IF_CONFIG_SMALL("QuickTime / MOV"),
7105 .priv_class = &mov_class,
7106 .priv_data_size = sizeof(MOVContext),
7107 .extensions = "mov,mp4,m4a,3gp,3g2,mj2",
7108 .read_probe = mov_probe,
7109 .read_header = mov_read_header,
7110 .read_packet = mov_read_packet,
7111 .read_close = mov_read_close,
7112 .read_seek = mov_read_seek,
7113 .flags = AVFMT_NO_BYTE_SEEK,