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);
78 static int mov_metadata_track_or_disc_number(MOVContext *c, AVIOContext *pb,
79 unsigned len, const char *key)
83 short current, total = 0;
84 avio_rb16(pb); // unknown
85 current = avio_rb16(pb);
87 total = avio_rb16(pb);
89 snprintf(buf, sizeof(buf), "%d", current);
91 snprintf(buf, sizeof(buf), "%d/%d", current, total);
92 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
93 av_dict_set(&c->fc->metadata, key, buf, 0);
98 static int mov_metadata_int8_bypass_padding(MOVContext *c, AVIOContext *pb,
99 unsigned len, const char *key)
101 /* bypass padding bytes */
106 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
107 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
112 static int mov_metadata_int8_no_padding(MOVContext *c, AVIOContext *pb,
113 unsigned len, const char *key)
115 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
116 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
121 static int mov_metadata_gnre(MOVContext *c, AVIOContext *pb,
122 unsigned len, const char *key)
126 avio_r8(pb); // unknown
129 if (genre < 1 || genre > ID3v1_GENRE_MAX)
131 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
132 av_dict_set(&c->fc->metadata, key, ff_id3v1_genre_str[genre-1], 0);
137 static const uint32_t mac_to_unicode[128] = {
138 0x00C4,0x00C5,0x00C7,0x00C9,0x00D1,0x00D6,0x00DC,0x00E1,
139 0x00E0,0x00E2,0x00E4,0x00E3,0x00E5,0x00E7,0x00E9,0x00E8,
140 0x00EA,0x00EB,0x00ED,0x00EC,0x00EE,0x00EF,0x00F1,0x00F3,
141 0x00F2,0x00F4,0x00F6,0x00F5,0x00FA,0x00F9,0x00FB,0x00FC,
142 0x2020,0x00B0,0x00A2,0x00A3,0x00A7,0x2022,0x00B6,0x00DF,
143 0x00AE,0x00A9,0x2122,0x00B4,0x00A8,0x2260,0x00C6,0x00D8,
144 0x221E,0x00B1,0x2264,0x2265,0x00A5,0x00B5,0x2202,0x2211,
145 0x220F,0x03C0,0x222B,0x00AA,0x00BA,0x03A9,0x00E6,0x00F8,
146 0x00BF,0x00A1,0x00AC,0x221A,0x0192,0x2248,0x2206,0x00AB,
147 0x00BB,0x2026,0x00A0,0x00C0,0x00C3,0x00D5,0x0152,0x0153,
148 0x2013,0x2014,0x201C,0x201D,0x2018,0x2019,0x00F7,0x25CA,
149 0x00FF,0x0178,0x2044,0x20AC,0x2039,0x203A,0xFB01,0xFB02,
150 0x2021,0x00B7,0x201A,0x201E,0x2030,0x00C2,0x00CA,0x00C1,
151 0x00CB,0x00C8,0x00CD,0x00CE,0x00CF,0x00CC,0x00D3,0x00D4,
152 0xF8FF,0x00D2,0x00DA,0x00DB,0x00D9,0x0131,0x02C6,0x02DC,
153 0x00AF,0x02D8,0x02D9,0x02DA,0x00B8,0x02DD,0x02DB,0x02C7,
156 static int mov_read_mac_string(MOVContext *c, AVIOContext *pb, int len,
157 char *dst, int dstlen)
160 char *end = dst+dstlen-1;
163 for (i = 0; i < len; i++) {
164 uint8_t t, c = avio_r8(pb);
172 PUT_UTF8(mac_to_unicode[c-0x80], t, if (p < end) *p++ = t;);
178 static int mov_read_covr(MOVContext *c, AVIOContext *pb, int type, int len)
182 MOVStreamContext *sc;
187 case 0xd: id = AV_CODEC_ID_MJPEG; break;
188 case 0xe: id = AV_CODEC_ID_PNG; break;
189 case 0x1b: id = AV_CODEC_ID_BMP; break;
191 av_log(c->fc, AV_LOG_WARNING, "Unknown cover type: 0x%x.\n", type);
196 st = avformat_new_stream(c->fc, NULL);
198 return AVERROR(ENOMEM);
199 sc = av_mallocz(sizeof(*sc));
201 return AVERROR(ENOMEM);
204 ret = av_get_packet(pb, &pkt, len);
208 if (pkt.size >= 8 && id != AV_CODEC_ID_BMP) {
209 if (AV_RB64(pkt.data) == 0x89504e470d0a1a0a) {
210 id = AV_CODEC_ID_PNG;
212 id = AV_CODEC_ID_MJPEG;
216 st->disposition |= AV_DISPOSITION_ATTACHED_PIC;
218 st->attached_pic = pkt;
219 st->attached_pic.stream_index = st->index;
220 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
222 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
223 st->codecpar->codec_id = id;
229 static int mov_metadata_loci(MOVContext *c, AVIOContext *pb, unsigned len)
231 char language[4] = { 0 };
232 char buf[200], place[100];
233 uint16_t langcode = 0;
234 double longitude, latitude, altitude;
235 const char *key = "location";
237 if (len < 4 + 2 + 1 + 1 + 4 + 4 + 4) {
238 av_log(c->fc, AV_LOG_ERROR, "loci too short\n");
239 return AVERROR_INVALIDDATA;
242 avio_skip(pb, 4); // version+flags
243 langcode = avio_rb16(pb);
244 ff_mov_lang_to_iso639(langcode, language);
247 len -= avio_get_str(pb, len, place, sizeof(place));
249 av_log(c->fc, AV_LOG_ERROR, "place name too long\n");
250 return AVERROR_INVALIDDATA;
252 avio_skip(pb, 1); // role
256 av_log(c->fc, AV_LOG_ERROR,
257 "loci too short (%u bytes left, need at least %d)\n", len, 12);
258 return AVERROR_INVALIDDATA;
260 longitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
261 latitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
262 altitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
264 // Try to output in the same format as the ?xyz field
265 snprintf(buf, sizeof(buf), "%+08.4f%+09.4f", latitude, longitude);
267 av_strlcatf(buf, sizeof(buf), "%+f", altitude);
268 av_strlcatf(buf, sizeof(buf), "/%s", place);
270 if (*language && strcmp(language, "und")) {
272 snprintf(key2, sizeof(key2), "%s-%s", key, language);
273 av_dict_set(&c->fc->metadata, key2, buf, 0);
275 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
276 return av_dict_set(&c->fc->metadata, key, buf, 0);
279 static int mov_metadata_hmmt(MOVContext *c, AVIOContext *pb, unsigned len)
285 if (c->ignore_chapters)
288 n_hmmt = avio_rb32(pb);
289 for (i = 0; i < n_hmmt && !pb->eof_reached; i++) {
290 int moment_time = avio_rb32(pb);
291 avpriv_new_chapter(c->fc, i, av_make_q(1, 1000), moment_time, AV_NOPTS_VALUE, NULL);
296 static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom)
299 char key2[32], language[4] = {0};
301 const char *key = NULL;
302 uint16_t langcode = 0;
303 uint32_t data_type = 0, str_size, str_size_alloc;
304 int (*parse)(MOVContext*, AVIOContext*, unsigned, const char*) = NULL;
309 case MKTAG( '@','P','R','M'): key = "premiere_version"; raw = 1; break;
310 case MKTAG( '@','P','R','Q'): key = "quicktime_version"; raw = 1; break;
311 case MKTAG( 'X','M','P','_'):
312 if (c->export_xmp) { key = "xmp"; raw = 1; } break;
313 case MKTAG( 'a','A','R','T'): key = "album_artist"; break;
314 case MKTAG( 'a','k','I','D'): key = "account_type";
315 parse = mov_metadata_int8_no_padding; break;
316 case MKTAG( 'a','p','I','D'): key = "account_id"; break;
317 case MKTAG( 'c','a','t','g'): key = "category"; break;
318 case MKTAG( 'c','p','i','l'): key = "compilation";
319 parse = mov_metadata_int8_no_padding; break;
320 case MKTAG( 'c','p','r','t'): key = "copyright"; break;
321 case MKTAG( 'd','e','s','c'): key = "description"; break;
322 case MKTAG( 'd','i','s','k'): key = "disc";
323 parse = mov_metadata_track_or_disc_number; break;
324 case MKTAG( 'e','g','i','d'): key = "episode_uid";
325 parse = mov_metadata_int8_no_padding; break;
326 case MKTAG( 'F','I','R','M'): key = "firmware"; raw = 1; break;
327 case MKTAG( 'g','n','r','e'): key = "genre";
328 parse = mov_metadata_gnre; break;
329 case MKTAG( 'h','d','v','d'): key = "hd_video";
330 parse = mov_metadata_int8_no_padding; break;
331 case MKTAG( 'H','M','M','T'):
332 return mov_metadata_hmmt(c, pb, atom.size);
333 case MKTAG( 'k','e','y','w'): key = "keywords"; break;
334 case MKTAG( 'l','d','e','s'): key = "synopsis"; break;
335 case MKTAG( 'l','o','c','i'):
336 return mov_metadata_loci(c, pb, atom.size);
337 case MKTAG( 'p','c','s','t'): key = "podcast";
338 parse = mov_metadata_int8_no_padding; break;
339 case MKTAG( 'p','g','a','p'): key = "gapless_playback";
340 parse = mov_metadata_int8_no_padding; break;
341 case MKTAG( 'p','u','r','d'): key = "purchase_date"; break;
342 case MKTAG( 'r','t','n','g'): key = "rating";
343 parse = mov_metadata_int8_no_padding; break;
344 case MKTAG( 's','o','a','a'): key = "sort_album_artist"; break;
345 case MKTAG( 's','o','a','l'): key = "sort_album"; break;
346 case MKTAG( 's','o','a','r'): key = "sort_artist"; break;
347 case MKTAG( 's','o','c','o'): key = "sort_composer"; break;
348 case MKTAG( 's','o','n','m'): key = "sort_name"; break;
349 case MKTAG( 's','o','s','n'): key = "sort_show"; break;
350 case MKTAG( 's','t','i','k'): key = "media_type";
351 parse = mov_metadata_int8_no_padding; break;
352 case MKTAG( 't','r','k','n'): key = "track";
353 parse = mov_metadata_track_or_disc_number; break;
354 case MKTAG( 't','v','e','n'): key = "episode_id"; break;
355 case MKTAG( 't','v','e','s'): key = "episode_sort";
356 parse = mov_metadata_int8_bypass_padding; break;
357 case MKTAG( 't','v','n','n'): key = "network"; break;
358 case MKTAG( 't','v','s','h'): key = "show"; break;
359 case MKTAG( 't','v','s','n'): key = "season_number";
360 parse = mov_metadata_int8_bypass_padding; break;
361 case MKTAG(0xa9,'A','R','T'): key = "artist"; break;
362 case MKTAG(0xa9,'P','R','D'): key = "producer"; break;
363 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
364 case MKTAG(0xa9,'a','u','t'): key = "artist"; break;
365 case MKTAG(0xa9,'c','h','p'): key = "chapter"; break;
366 case MKTAG(0xa9,'c','m','t'): key = "comment"; break;
367 case MKTAG(0xa9,'c','o','m'): key = "composer"; break;
368 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
369 case MKTAG(0xa9,'d','a','y'): key = "date"; break;
370 case MKTAG(0xa9,'d','i','r'): key = "director"; break;
371 case MKTAG(0xa9,'d','i','s'): key = "disclaimer"; break;
372 case MKTAG(0xa9,'e','d','1'): key = "edit_date"; break;
373 case MKTAG(0xa9,'e','n','c'): key = "encoder"; break;
374 case MKTAG(0xa9,'f','m','t'): key = "original_format"; break;
375 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
376 case MKTAG(0xa9,'g','r','p'): key = "grouping"; break;
377 case MKTAG(0xa9,'h','s','t'): key = "host_computer"; break;
378 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
379 case MKTAG(0xa9,'l','y','r'): key = "lyrics"; break;
380 case MKTAG(0xa9,'m','a','k'): key = "make"; break;
381 case MKTAG(0xa9,'m','o','d'): key = "model"; break;
382 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
383 case MKTAG(0xa9,'o','p','e'): key = "original_artist"; break;
384 case MKTAG(0xa9,'p','r','d'): key = "producer"; break;
385 case MKTAG(0xa9,'p','r','f'): key = "performers"; break;
386 case MKTAG(0xa9,'r','e','q'): key = "playback_requirements"; break;
387 case MKTAG(0xa9,'s','r','c'): key = "original_source"; break;
388 case MKTAG(0xa9,'s','t','3'): key = "subtitle"; break;
389 case MKTAG(0xa9,'s','w','r'): key = "encoder"; break;
390 case MKTAG(0xa9,'t','o','o'): key = "encoder"; break;
391 case MKTAG(0xa9,'t','r','k'): key = "track"; break;
392 case MKTAG(0xa9,'u','r','l'): key = "URL"; break;
393 case MKTAG(0xa9,'w','r','n'): key = "warning"; break;
394 case MKTAG(0xa9,'w','r','t'): key = "composer"; break;
395 case MKTAG(0xa9,'x','y','z'): key = "location"; break;
398 if (c->itunes_metadata && atom.size > 8) {
399 int data_size = avio_rb32(pb);
400 int tag = avio_rl32(pb);
401 if (tag == MKTAG('d','a','t','a') && data_size <= atom.size) {
402 data_type = avio_rb32(pb); // type
403 avio_rb32(pb); // unknown
404 str_size = data_size - 16;
407 if (atom.type == MKTAG('c', 'o', 'v', 'r')) {
408 int ret = mov_read_covr(c, pb, data_type, str_size);
410 av_log(c->fc, AV_LOG_ERROR, "Error parsing cover art.\n");
413 } else if (!key && c->found_hdlr_mdta && c->meta_keys) {
414 uint32_t index = AV_RB32(&atom.type);
415 if (index < c->meta_keys_count && index > 0) {
416 key = c->meta_keys[index];
418 av_log(c->fc, AV_LOG_WARNING,
419 "The index of 'data' is out of range: %"PRId32" < 1 or >= %d.\n",
420 index, c->meta_keys_count);
424 } else if (atom.size > 4 && key && !c->itunes_metadata && !raw) {
425 str_size = avio_rb16(pb); // string length
426 if (str_size > atom.size) {
428 avio_seek(pb, -2, SEEK_CUR);
429 av_log(c->fc, AV_LOG_WARNING, "UDTA parsing failed retrying raw\n");
432 langcode = avio_rb16(pb);
433 ff_mov_lang_to_iso639(langcode, language);
436 str_size = atom.size;
438 if (c->export_all && !key) {
439 snprintf(tmp_key, 5, "%.4s", (char*)&atom.type);
445 if (atom.size < 0 || str_size >= INT_MAX/2)
446 return AVERROR_INVALIDDATA;
448 // Allocates enough space if data_type is a int32 or float32 number, otherwise
449 // worst-case requirement for output string in case of utf8 coded input
450 num = (data_type >= 21 && data_type <= 23);
451 str_size_alloc = (num ? 512 : (raw ? str_size : str_size * 2)) + 1;
452 str = av_mallocz(str_size_alloc);
454 return AVERROR(ENOMEM);
457 parse(c, pb, str_size, key);
459 if (!raw && (data_type == 3 || (data_type == 0 && (langcode < 0x400 || langcode == 0x7fff)))) { // MAC Encoded
460 mov_read_mac_string(c, pb, str_size, str, str_size_alloc);
461 } else if (data_type == 21) { // BE signed integer, variable size
464 val = (int8_t)avio_r8(pb);
465 else if (str_size == 2)
466 val = (int16_t)avio_rb16(pb);
467 else if (str_size == 3)
468 val = ((int32_t)(avio_rb24(pb)<<8))>>8;
469 else if (str_size == 4)
470 val = (int32_t)avio_rb32(pb);
471 if (snprintf(str, str_size_alloc, "%d", val) >= str_size_alloc) {
472 av_log(c->fc, AV_LOG_ERROR,
473 "Failed to store the number (%d) in string.\n", val);
475 return AVERROR_INVALIDDATA;
477 } else if (data_type == 22) { // BE unsigned integer, variable size
478 unsigned int val = 0;
481 else if (str_size == 2)
483 else if (str_size == 3)
485 else if (str_size == 4)
487 if (snprintf(str, str_size_alloc, "%u", val) >= str_size_alloc) {
488 av_log(c->fc, AV_LOG_ERROR,
489 "Failed to store the number (%u) in string.\n", val);
491 return AVERROR_INVALIDDATA;
493 } else if (data_type == 23 && str_size >= 4) { // BE float32
494 float val = av_int2float(avio_rb32(pb));
495 if (snprintf(str, str_size_alloc, "%f", val) >= str_size_alloc) {
496 av_log(c->fc, AV_LOG_ERROR,
497 "Failed to store the float32 number (%f) in string.\n", val);
499 return AVERROR_INVALIDDATA;
502 int ret = ffio_read_size(pb, str, str_size);
509 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
510 av_dict_set(&c->fc->metadata, key, str, 0);
511 if (*language && strcmp(language, "und")) {
512 snprintf(key2, sizeof(key2), "%s-%s", key, language);
513 av_dict_set(&c->fc->metadata, key2, str, 0);
515 if (!strcmp(key, "encoder")) {
516 int major, minor, micro;
517 if (sscanf(str, "HandBrake %d.%d.%d", &major, &minor, µ) == 3) {
518 c->handbrake_version = 1000000*major + 1000*minor + micro;
527 static int mov_read_chpl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
530 int i, nb_chapters, str_len, version;
534 if (c->ignore_chapters)
537 if ((atom.size -= 5) < 0)
540 version = avio_r8(pb);
543 avio_rb32(pb); // ???
544 nb_chapters = avio_r8(pb);
546 for (i = 0; i < nb_chapters; i++) {
550 start = avio_rb64(pb);
551 str_len = avio_r8(pb);
553 if ((atom.size -= 9+str_len) < 0)
556 ret = ffio_read_size(pb, str, str_len);
560 avpriv_new_chapter(c->fc, i, (AVRational){1,10000000}, start, AV_NOPTS_VALUE, str);
565 #define MIN_DATA_ENTRY_BOX_SIZE 12
566 static int mov_read_dref(MOVContext *c, AVIOContext *pb, MOVAtom atom)
569 MOVStreamContext *sc;
572 if (c->fc->nb_streams < 1)
574 st = c->fc->streams[c->fc->nb_streams-1];
577 avio_rb32(pb); // version + flags
578 entries = avio_rb32(pb);
580 entries > (atom.size - 1) / MIN_DATA_ENTRY_BOX_SIZE + 1 ||
581 entries >= UINT_MAX / sizeof(*sc->drefs))
582 return AVERROR_INVALIDDATA;
586 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
588 return AVERROR(ENOMEM);
589 sc->drefs_count = entries;
591 for (i = 0; i < entries; i++) {
592 MOVDref *dref = &sc->drefs[i];
593 uint32_t size = avio_rb32(pb);
594 int64_t next = avio_tell(pb) + size - 4;
597 return AVERROR_INVALIDDATA;
599 dref->type = avio_rl32(pb);
600 avio_rb32(pb); // version + flags
602 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
603 /* macintosh alias record */
604 uint16_t volume_len, len;
610 volume_len = avio_r8(pb);
611 volume_len = FFMIN(volume_len, 27);
612 ret = ffio_read_size(pb, dref->volume, 27);
615 dref->volume[volume_len] = 0;
616 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", dref->volume, volume_len);
621 len = FFMIN(len, 63);
622 ret = ffio_read_size(pb, dref->filename, 63);
625 dref->filename[len] = 0;
626 av_log(c->fc, AV_LOG_DEBUG, "filename %s, len %d\n", dref->filename, len);
630 /* read next level up_from_alias/down_to_target */
631 dref->nlvl_from = avio_rb16(pb);
632 dref->nlvl_to = avio_rb16(pb);
633 av_log(c->fc, AV_LOG_DEBUG, "nlvl from %d, nlvl to %d\n",
634 dref->nlvl_from, dref->nlvl_to);
638 for (type = 0; type != -1 && avio_tell(pb) < next; ) {
641 type = avio_rb16(pb);
643 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
646 if (type == 2) { // absolute path
648 dref->path = av_mallocz(len+1);
650 return AVERROR(ENOMEM);
652 ret = ffio_read_size(pb, dref->path, len);
654 av_freep(&dref->path);
657 if (len > volume_len && !strncmp(dref->path, dref->volume, volume_len)) {
659 memmove(dref->path, dref->path+volume_len, len);
662 // trim string of any ending zeros
663 for (j = len - 1; j >= 0; j--) {
664 if (dref->path[j] == 0)
669 for (j = 0; j < len; j++)
670 if (dref->path[j] == ':' || dref->path[j] == 0)
672 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
673 } else if (type == 0) { // directory name
675 dref->dir = av_malloc(len+1);
677 return AVERROR(ENOMEM);
679 ret = ffio_read_size(pb, dref->dir, len);
681 av_freep(&dref->dir);
685 for (j = 0; j < len; j++)
686 if (dref->dir[j] == ':')
688 av_log(c->fc, AV_LOG_DEBUG, "dir %s\n", dref->dir);
693 av_log(c->fc, AV_LOG_DEBUG, "Unknown dref type 0x%08"PRIx32" size %"PRIu32"\n",
698 avio_seek(pb, next, SEEK_SET);
703 static int mov_read_hdlr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
712 avio_r8(pb); /* version */
713 avio_rb24(pb); /* flags */
716 ctype = avio_rl32(pb);
717 type = avio_rl32(pb); /* component subtype */
719 av_log(c->fc, AV_LOG_TRACE, "ctype=%s\n", av_fourcc2str(ctype));
720 av_log(c->fc, AV_LOG_TRACE, "stype=%s\n", av_fourcc2str(type));
722 if (c->trak_index < 0) { // meta not inside a trak
723 if (type == MKTAG('m','d','t','a')) {
724 c->found_hdlr_mdta = 1;
729 st = c->fc->streams[c->fc->nb_streams-1];
731 if (type == MKTAG('v','i','d','e'))
732 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
733 else if (type == MKTAG('s','o','u','n'))
734 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
735 else if (type == MKTAG('m','1','a',' '))
736 st->codecpar->codec_id = AV_CODEC_ID_MP2;
737 else if ((type == MKTAG('s','u','b','p')) || (type == MKTAG('c','l','c','p')))
738 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
740 avio_rb32(pb); /* component manufacture */
741 avio_rb32(pb); /* component flags */
742 avio_rb32(pb); /* component flags mask */
744 title_size = atom.size - 24;
745 if (title_size > 0) {
746 if (title_size > FFMIN(INT_MAX, SIZE_MAX-1))
747 return AVERROR_INVALIDDATA;
748 title_str = av_malloc(title_size + 1); /* Add null terminator */
750 return AVERROR(ENOMEM);
752 ret = ffio_read_size(pb, title_str, title_size);
754 av_freep(&title_str);
757 title_str[title_size] = 0;
759 int off = (!c->isom && title_str[0] == title_size - 1);
760 av_dict_set(&st->metadata, "handler_name", title_str + off, 0);
762 av_freep(&title_str);
768 int ff_mov_read_esds(AVFormatContext *fc, AVIOContext *pb)
773 if (fc->nb_streams < 1)
775 st = fc->streams[fc->nb_streams-1];
777 avio_rb32(pb); /* version + flags */
778 ff_mp4_read_descr(fc, pb, &tag);
779 if (tag == MP4ESDescrTag) {
780 ff_mp4_parse_es_descr(pb, NULL);
782 avio_rb16(pb); /* ID */
784 ff_mp4_read_descr(fc, pb, &tag);
785 if (tag == MP4DecConfigDescrTag)
786 ff_mp4_read_dec_config_descr(fc, st, pb);
790 static int mov_read_esds(MOVContext *c, AVIOContext *pb, MOVAtom atom)
792 return ff_mov_read_esds(c->fc, pb);
795 static int mov_read_dac3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
798 enum AVAudioServiceType *ast;
799 int ac3info, acmod, lfeon, bsmod;
801 if (c->fc->nb_streams < 1)
803 st = c->fc->streams[c->fc->nb_streams-1];
805 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
808 return AVERROR(ENOMEM);
810 ac3info = avio_rb24(pb);
811 bsmod = (ac3info >> 14) & 0x7;
812 acmod = (ac3info >> 11) & 0x7;
813 lfeon = (ac3info >> 10) & 0x1;
814 st->codecpar->channels = ((int[]){2,1,2,3,3,4,4,5})[acmod] + lfeon;
815 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
817 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
819 if (st->codecpar->channels > 1 && bsmod == 0x7)
820 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
822 #if FF_API_LAVF_AVCTX
823 FF_DISABLE_DEPRECATION_WARNINGS
824 st->codec->audio_service_type = *ast;
825 FF_ENABLE_DEPRECATION_WARNINGS
831 static int mov_read_dec3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
834 enum AVAudioServiceType *ast;
835 int eac3info, acmod, lfeon, bsmod;
837 if (c->fc->nb_streams < 1)
839 st = c->fc->streams[c->fc->nb_streams-1];
841 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
844 return AVERROR(ENOMEM);
846 /* No need to parse fields for additional independent substreams and its
847 * associated dependent substreams since libavcodec's E-AC-3 decoder
848 * does not support them yet. */
849 avio_rb16(pb); /* data_rate and num_ind_sub */
850 eac3info = avio_rb24(pb);
851 bsmod = (eac3info >> 12) & 0x1f;
852 acmod = (eac3info >> 9) & 0x7;
853 lfeon = (eac3info >> 8) & 0x1;
854 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
856 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
857 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
859 if (st->codecpar->channels > 1 && bsmod == 0x7)
860 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
862 #if FF_API_LAVF_AVCTX
863 FF_DISABLE_DEPRECATION_WARNINGS
864 st->codec->audio_service_type = *ast;
865 FF_ENABLE_DEPRECATION_WARNINGS
871 static int mov_read_ddts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
873 const uint32_t ddts_size = 20;
876 uint32_t frame_duration_code = 0;
877 uint32_t channel_layout_code = 0;
880 buf = av_malloc(ddts_size + AV_INPUT_BUFFER_PADDING_SIZE);
882 return AVERROR(ENOMEM);
884 if (avio_read(pb, buf, ddts_size) < ddts_size) {
886 return AVERROR_INVALIDDATA;
889 init_get_bits(&gb, buf, 8*ddts_size);
891 if (c->fc->nb_streams < 1) {
895 st = c->fc->streams[c->fc->nb_streams-1];
897 st->codecpar->sample_rate = get_bits_long(&gb, 32);
898 if (st->codecpar->sample_rate <= 0) {
899 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
901 return AVERROR_INVALIDDATA;
903 skip_bits_long(&gb, 32); /* max bitrate */
904 st->codecpar->bit_rate = get_bits_long(&gb, 32);
905 st->codecpar->bits_per_coded_sample = get_bits(&gb, 8);
906 frame_duration_code = get_bits(&gb, 2);
907 skip_bits(&gb, 30); /* various fields */
908 channel_layout_code = get_bits(&gb, 16);
910 st->codecpar->frame_size =
911 (frame_duration_code == 0) ? 512 :
912 (frame_duration_code == 1) ? 1024 :
913 (frame_duration_code == 2) ? 2048 :
914 (frame_duration_code == 3) ? 4096 : 0;
916 if (channel_layout_code > 0xff) {
917 av_log(c->fc, AV_LOG_WARNING, "Unsupported DTS audio channel layout");
919 st->codecpar->channel_layout =
920 ((channel_layout_code & 0x1) ? AV_CH_FRONT_CENTER : 0) |
921 ((channel_layout_code & 0x2) ? AV_CH_FRONT_LEFT : 0) |
922 ((channel_layout_code & 0x2) ? AV_CH_FRONT_RIGHT : 0) |
923 ((channel_layout_code & 0x4) ? AV_CH_SIDE_LEFT : 0) |
924 ((channel_layout_code & 0x4) ? AV_CH_SIDE_RIGHT : 0) |
925 ((channel_layout_code & 0x8) ? AV_CH_LOW_FREQUENCY : 0);
927 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
933 static int mov_read_chan(MOVContext *c, AVIOContext *pb, MOVAtom atom)
937 if (c->fc->nb_streams < 1)
939 st = c->fc->streams[c->fc->nb_streams-1];
944 /* skip version and flags */
947 ff_mov_read_chan(c->fc, pb, st, atom.size - 4);
952 static int mov_read_wfex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
957 if (c->fc->nb_streams < 1)
959 st = c->fc->streams[c->fc->nb_streams-1];
961 if ((ret = ff_get_wav_header(c->fc, pb, st->codecpar, atom.size, 0)) < 0)
962 av_log(c->fc, AV_LOG_WARNING, "get_wav_header failed\n");
967 static int mov_read_pasp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
969 const int num = avio_rb32(pb);
970 const int den = avio_rb32(pb);
973 if (c->fc->nb_streams < 1)
975 st = c->fc->streams[c->fc->nb_streams-1];
977 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
978 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num)) {
979 av_log(c->fc, AV_LOG_WARNING,
980 "sample aspect ratio already set to %d:%d, ignoring 'pasp' atom (%d:%d)\n",
981 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den,
983 } else if (den != 0) {
984 av_reduce(&st->sample_aspect_ratio.num, &st->sample_aspect_ratio.den,
990 /* this atom contains actual media data */
991 static int mov_read_mdat(MOVContext *c, AVIOContext *pb, MOVAtom atom)
993 if (atom.size == 0) /* wrong one (MP4) */
996 return 0; /* now go for moov */
999 #define DRM_BLOB_SIZE 56
1001 static int mov_read_adrm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1003 uint8_t intermediate_key[20];
1004 uint8_t intermediate_iv[20];
1007 uint8_t file_checksum[20];
1008 uint8_t calculated_checksum[20];
1012 uint8_t *activation_bytes = c->activation_bytes;
1013 uint8_t *fixed_key = c->audible_fixed_key;
1017 sha = av_sha_alloc();
1019 return AVERROR(ENOMEM);
1020 c->aes_decrypt = av_aes_alloc();
1021 if (!c->aes_decrypt) {
1022 ret = AVERROR(ENOMEM);
1026 /* drm blob processing */
1027 avio_read(pb, output, 8); // go to offset 8, absolute position 0x251
1028 avio_read(pb, input, DRM_BLOB_SIZE);
1029 avio_read(pb, output, 4); // go to offset 4, absolute position 0x28d
1030 avio_read(pb, file_checksum, 20);
1032 av_log(c->fc, AV_LOG_INFO, "[aax] file checksum == "); // required by external tools
1033 for (i = 0; i < 20; i++)
1034 av_log(c->fc, AV_LOG_INFO, "%02x", file_checksum[i]);
1035 av_log(c->fc, AV_LOG_INFO, "\n");
1037 /* verify activation data */
1038 if (!activation_bytes) {
1039 av_log(c->fc, AV_LOG_WARNING, "[aax] activation_bytes option is missing!\n");
1040 ret = 0; /* allow ffprobe to continue working on .aax files */
1043 if (c->activation_bytes_size != 4) {
1044 av_log(c->fc, AV_LOG_FATAL, "[aax] activation_bytes value needs to be 4 bytes!\n");
1045 ret = AVERROR(EINVAL);
1049 /* verify fixed key */
1050 if (c->audible_fixed_key_size != 16) {
1051 av_log(c->fc, AV_LOG_FATAL, "[aax] audible_fixed_key value needs to be 16 bytes!\n");
1052 ret = AVERROR(EINVAL);
1056 /* AAX (and AAX+) key derivation */
1057 av_sha_init(sha, 160);
1058 av_sha_update(sha, fixed_key, 16);
1059 av_sha_update(sha, activation_bytes, 4);
1060 av_sha_final(sha, intermediate_key);
1061 av_sha_init(sha, 160);
1062 av_sha_update(sha, fixed_key, 16);
1063 av_sha_update(sha, intermediate_key, 20);
1064 av_sha_update(sha, activation_bytes, 4);
1065 av_sha_final(sha, intermediate_iv);
1066 av_sha_init(sha, 160);
1067 av_sha_update(sha, intermediate_key, 16);
1068 av_sha_update(sha, intermediate_iv, 16);
1069 av_sha_final(sha, calculated_checksum);
1070 if (memcmp(calculated_checksum, file_checksum, 20)) { // critical error
1071 av_log(c->fc, AV_LOG_ERROR, "[aax] mismatch in checksums!\n");
1072 ret = AVERROR_INVALIDDATA;
1075 av_aes_init(c->aes_decrypt, intermediate_key, 128, 1);
1076 av_aes_crypt(c->aes_decrypt, output, input, DRM_BLOB_SIZE >> 4, intermediate_iv, 1);
1077 for (i = 0; i < 4; i++) {
1078 // file data (in output) is stored in big-endian mode
1079 if (activation_bytes[i] != output[3 - i]) { // critical error
1080 av_log(c->fc, AV_LOG_ERROR, "[aax] error in drm blob decryption!\n");
1081 ret = AVERROR_INVALIDDATA;
1085 memcpy(c->file_key, output + 8, 16);
1086 memcpy(input, output + 26, 16);
1087 av_sha_init(sha, 160);
1088 av_sha_update(sha, input, 16);
1089 av_sha_update(sha, c->file_key, 16);
1090 av_sha_update(sha, fixed_key, 16);
1091 av_sha_final(sha, c->file_iv);
1099 // Audible AAX (and AAX+) bytestream decryption
1100 static int aax_filter(uint8_t *input, int size, MOVContext *c)
1103 unsigned char iv[16];
1105 memcpy(iv, c->file_iv, 16); // iv is overwritten
1106 blocks = size >> 4; // trailing bytes are not encrypted!
1107 av_aes_init(c->aes_decrypt, c->file_key, 128, 1);
1108 av_aes_crypt(c->aes_decrypt, input, input, blocks, iv, 1);
1113 /* read major brand, minor version and compatible brands and store them as metadata */
1114 static int mov_read_ftyp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1117 int comp_brand_size;
1118 char* comp_brands_str;
1119 uint8_t type[5] = {0};
1120 int ret = ffio_read_size(pb, type, 4);
1124 if (strcmp(type, "qt "))
1126 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
1127 av_dict_set(&c->fc->metadata, "major_brand", type, 0);
1128 minor_ver = avio_rb32(pb); /* minor version */
1129 av_dict_set_int(&c->fc->metadata, "minor_version", minor_ver, 0);
1131 comp_brand_size = atom.size - 8;
1132 if (comp_brand_size < 0)
1133 return AVERROR_INVALIDDATA;
1134 comp_brands_str = av_malloc(comp_brand_size + 1); /* Add null terminator */
1135 if (!comp_brands_str)
1136 return AVERROR(ENOMEM);
1138 ret = ffio_read_size(pb, comp_brands_str, comp_brand_size);
1140 av_freep(&comp_brands_str);
1143 comp_brands_str[comp_brand_size] = 0;
1144 av_dict_set(&c->fc->metadata, "compatible_brands", comp_brands_str, 0);
1145 av_freep(&comp_brands_str);
1150 /* this atom should contain all header atoms */
1151 static int mov_read_moov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1155 if (c->found_moov) {
1156 av_log(c->fc, AV_LOG_WARNING, "Found duplicated MOOV Atom. Skipped it\n");
1157 avio_skip(pb, atom.size);
1161 if ((ret = mov_read_default(c, pb, atom)) < 0)
1163 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
1164 /* so we don't parse the whole file if over a network */
1166 return 0; /* now go for mdat */
1169 static int mov_read_moof(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1171 if (!c->has_looked_for_mfra && c->use_mfra_for > 0) {
1172 c->has_looked_for_mfra = 1;
1173 if (pb->seekable & AVIO_SEEKABLE_NORMAL) {
1175 av_log(c->fc, AV_LOG_VERBOSE, "stream has moof boxes, will look "
1177 if ((ret = mov_read_mfra(c, pb)) < 0) {
1178 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but failed to "
1179 "read the mfra (may be a live ismv)\n");
1182 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but stream is not "
1183 "seekable, can not look for mfra\n");
1186 c->fragment.moof_offset = c->fragment.implicit_offset = avio_tell(pb) - 8;
1187 av_log(c->fc, AV_LOG_TRACE, "moof offset %"PRIx64"\n", c->fragment.moof_offset);
1188 return mov_read_default(c, pb, atom);
1191 static void mov_metadata_creation_time(AVDictionary **metadata, int64_t time)
1194 if(time >= 2082844800)
1195 time -= 2082844800; /* seconds between 1904-01-01 and Epoch */
1197 if ((int64_t)(time * 1000000ULL) / 1000000 != time) {
1198 av_log(NULL, AV_LOG_DEBUG, "creation_time is not representable\n");
1202 avpriv_dict_set_timestamp(metadata, "creation_time", time * 1000000);
1206 static int mov_read_mdhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1209 MOVStreamContext *sc;
1211 char language[4] = {0};
1213 int64_t creation_time;
1215 if (c->fc->nb_streams < 1)
1217 st = c->fc->streams[c->fc->nb_streams-1];
1220 if (sc->time_scale) {
1221 av_log(c->fc, AV_LOG_ERROR, "Multiple mdhd?\n");
1222 return AVERROR_INVALIDDATA;
1225 version = avio_r8(pb);
1227 avpriv_request_sample(c->fc, "Version %d", version);
1228 return AVERROR_PATCHWELCOME;
1230 avio_rb24(pb); /* flags */
1232 creation_time = avio_rb64(pb);
1235 creation_time = avio_rb32(pb);
1236 avio_rb32(pb); /* modification time */
1238 mov_metadata_creation_time(&st->metadata, creation_time);
1240 sc->time_scale = avio_rb32(pb);
1241 if (sc->time_scale <= 0) {
1242 av_log(c->fc, AV_LOG_ERROR, "Invalid mdhd time scale %d, defaulting to 1\n", sc->time_scale);
1245 st->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1247 lang = avio_rb16(pb); /* language */
1248 if (ff_mov_lang_to_iso639(lang, language))
1249 av_dict_set(&st->metadata, "language", language, 0);
1250 avio_rb16(pb); /* quality */
1255 static int mov_read_mvhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1258 int64_t creation_time;
1259 int version = avio_r8(pb); /* version */
1260 avio_rb24(pb); /* flags */
1263 creation_time = avio_rb64(pb);
1266 creation_time = avio_rb32(pb);
1267 avio_rb32(pb); /* modification time */
1269 mov_metadata_creation_time(&c->fc->metadata, creation_time);
1270 c->time_scale = avio_rb32(pb); /* time scale */
1271 if (c->time_scale <= 0) {
1272 av_log(c->fc, AV_LOG_ERROR, "Invalid mvhd time scale %d, defaulting to 1\n", c->time_scale);
1275 av_log(c->fc, AV_LOG_TRACE, "time scale = %i\n", c->time_scale);
1277 c->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1278 // set the AVCodecContext duration because the duration of individual tracks
1279 // may be inaccurate
1280 if (c->time_scale > 0 && !c->trex_data)
1281 c->fc->duration = av_rescale(c->duration, AV_TIME_BASE, c->time_scale);
1282 avio_rb32(pb); /* preferred scale */
1284 avio_rb16(pb); /* preferred volume */
1286 avio_skip(pb, 10); /* reserved */
1288 /* movie display matrix, store it in main context and use it later on */
1289 for (i = 0; i < 3; i++) {
1290 c->movie_display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
1291 c->movie_display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
1292 c->movie_display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
1295 avio_rb32(pb); /* preview time */
1296 avio_rb32(pb); /* preview duration */
1297 avio_rb32(pb); /* poster time */
1298 avio_rb32(pb); /* selection time */
1299 avio_rb32(pb); /* selection duration */
1300 avio_rb32(pb); /* current time */
1301 avio_rb32(pb); /* next track ID */
1306 static int mov_read_enda(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1311 if (c->fc->nb_streams < 1)
1313 st = c->fc->streams[c->fc->nb_streams-1];
1315 little_endian = avio_rb16(pb) & 0xFF;
1316 av_log(c->fc, AV_LOG_TRACE, "enda %d\n", little_endian);
1317 if (little_endian == 1) {
1318 switch (st->codecpar->codec_id) {
1319 case AV_CODEC_ID_PCM_S24BE:
1320 st->codecpar->codec_id = AV_CODEC_ID_PCM_S24LE;
1322 case AV_CODEC_ID_PCM_S32BE:
1323 st->codecpar->codec_id = AV_CODEC_ID_PCM_S32LE;
1325 case AV_CODEC_ID_PCM_F32BE:
1326 st->codecpar->codec_id = AV_CODEC_ID_PCM_F32LE;
1328 case AV_CODEC_ID_PCM_F64BE:
1329 st->codecpar->codec_id = AV_CODEC_ID_PCM_F64LE;
1338 static int mov_read_colr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1341 char color_parameter_type[5] = { 0 };
1342 uint16_t color_primaries, color_trc, color_matrix;
1345 if (c->fc->nb_streams < 1)
1347 st = c->fc->streams[c->fc->nb_streams - 1];
1349 ret = ffio_read_size(pb, color_parameter_type, 4);
1352 if (strncmp(color_parameter_type, "nclx", 4) &&
1353 strncmp(color_parameter_type, "nclc", 4)) {
1354 av_log(c->fc, AV_LOG_WARNING, "unsupported color_parameter_type %s\n",
1355 color_parameter_type);
1359 color_primaries = avio_rb16(pb);
1360 color_trc = avio_rb16(pb);
1361 color_matrix = avio_rb16(pb);
1363 av_log(c->fc, AV_LOG_TRACE,
1364 "%s: pri %d trc %d matrix %d",
1365 color_parameter_type, color_primaries, color_trc, color_matrix);
1367 if (!strncmp(color_parameter_type, "nclx", 4)) {
1368 uint8_t color_range = avio_r8(pb) >> 7;
1369 av_log(c->fc, AV_LOG_TRACE, " full %"PRIu8"", color_range);
1371 st->codecpar->color_range = AVCOL_RANGE_JPEG;
1373 st->codecpar->color_range = AVCOL_RANGE_MPEG;
1376 if (!av_color_primaries_name(color_primaries))
1377 color_primaries = AVCOL_PRI_UNSPECIFIED;
1378 if (!av_color_transfer_name(color_trc))
1379 color_trc = AVCOL_TRC_UNSPECIFIED;
1380 if (!av_color_space_name(color_matrix))
1381 color_matrix = AVCOL_SPC_UNSPECIFIED;
1383 st->codecpar->color_primaries = color_primaries;
1384 st->codecpar->color_trc = color_trc;
1385 st->codecpar->color_space = color_matrix;
1386 av_log(c->fc, AV_LOG_TRACE, "\n");
1391 static int mov_read_fiel(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1394 unsigned mov_field_order;
1395 enum AVFieldOrder decoded_field_order = AV_FIELD_UNKNOWN;
1397 if (c->fc->nb_streams < 1) // will happen with jp2 files
1399 st = c->fc->streams[c->fc->nb_streams-1];
1401 return AVERROR_INVALIDDATA;
1402 mov_field_order = avio_rb16(pb);
1403 if ((mov_field_order & 0xFF00) == 0x0100)
1404 decoded_field_order = AV_FIELD_PROGRESSIVE;
1405 else if ((mov_field_order & 0xFF00) == 0x0200) {
1406 switch (mov_field_order & 0xFF) {
1407 case 0x01: decoded_field_order = AV_FIELD_TT;
1409 case 0x06: decoded_field_order = AV_FIELD_BB;
1411 case 0x09: decoded_field_order = AV_FIELD_TB;
1413 case 0x0E: decoded_field_order = AV_FIELD_BT;
1417 if (decoded_field_order == AV_FIELD_UNKNOWN && mov_field_order) {
1418 av_log(NULL, AV_LOG_ERROR, "Unknown MOV field order 0x%04x\n", mov_field_order);
1420 st->codecpar->field_order = decoded_field_order;
1425 static int mov_realloc_extradata(AVCodecParameters *par, MOVAtom atom)
1428 uint64_t size = (uint64_t)par->extradata_size + atom.size + 8 + AV_INPUT_BUFFER_PADDING_SIZE;
1429 if (size > INT_MAX || (uint64_t)atom.size > INT_MAX)
1430 return AVERROR_INVALIDDATA;
1431 if ((err = av_reallocp(&par->extradata, size)) < 0) {
1432 par->extradata_size = 0;
1435 par->extradata_size = size - AV_INPUT_BUFFER_PADDING_SIZE;
1439 /* Read a whole atom into the extradata return the size of the atom read, possibly truncated if != atom.size */
1440 static int64_t mov_read_atom_into_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1441 AVCodecParameters *par, uint8_t *buf)
1443 int64_t result = atom.size;
1446 AV_WB32(buf , atom.size + 8);
1447 AV_WL32(buf + 4, atom.type);
1448 err = ffio_read_size(pb, buf + 8, atom.size);
1450 par->extradata_size -= atom.size;
1452 } else if (err < atom.size) {
1453 av_log(c->fc, AV_LOG_WARNING, "truncated extradata\n");
1454 par->extradata_size -= atom.size - err;
1457 memset(buf + 8 + err, 0, AV_INPUT_BUFFER_PADDING_SIZE);
1461 /* FIXME modify QDM2/SVQ3/H.264 decoders to take full atom as extradata */
1462 static int mov_read_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1463 enum AVCodecID codec_id)
1466 uint64_t original_size;
1469 if (c->fc->nb_streams < 1) // will happen with jp2 files
1471 st = c->fc->streams[c->fc->nb_streams-1];
1473 if (st->codecpar->codec_id != codec_id)
1474 return 0; /* unexpected codec_id - don't mess with extradata */
1476 original_size = st->codecpar->extradata_size;
1477 err = mov_realloc_extradata(st->codecpar, atom);
1481 err = mov_read_atom_into_extradata(c, pb, atom, st->codecpar, st->codecpar->extradata + original_size);
1484 return 0; // Note: this is the original behavior to ignore truncation.
1487 /* wrapper functions for reading ALAC/AVS/MJPEG/MJPEG2000 extradata atoms only for those codecs */
1488 static int mov_read_alac(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1490 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_ALAC);
1493 static int mov_read_avss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1495 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVS);
1498 static int mov_read_jp2h(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1500 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_JPEG2000);
1503 static int mov_read_dpxe(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1505 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_R10K);
1508 static int mov_read_avid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1510 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVUI);
1512 ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_DNXHD);
1516 static int mov_read_targa_y216(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1518 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_TARGA_Y216);
1520 if (!ret && c->fc->nb_streams >= 1) {
1521 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1522 if (par->extradata_size >= 40) {
1523 par->height = AV_RB16(&par->extradata[36]);
1524 par->width = AV_RB16(&par->extradata[38]);
1530 static int mov_read_ares(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1532 if (c->fc->nb_streams >= 1) {
1533 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1534 if (par->codec_tag == MKTAG('A', 'V', 'i', 'n') &&
1535 par->codec_id == AV_CODEC_ID_H264 &&
1539 cid = avio_rb16(pb);
1540 /* For AVID AVCI50, force width of 1440 to be able to select the correct SPS and PPS */
1541 if (cid == 0xd4d || cid == 0xd4e)
1544 } else if ((par->codec_tag == MKTAG('A', 'V', 'd', '1') ||
1545 par->codec_tag == MKTAG('A', 'V', 'd', 'n')) &&
1549 num = avio_rb32(pb);
1550 den = avio_rb32(pb);
1551 if (num <= 0 || den <= 0)
1553 switch (avio_rb32(pb)) {
1555 if (den >= INT_MAX / 2)
1559 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.num = num;
1560 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.den = den;
1567 return mov_read_avid(c, pb, atom);
1570 static int mov_read_aclr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1574 uint64_t original_size;
1575 if (c->fc->nb_streams >= 1) {
1576 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1577 if (par->codec_id == AV_CODEC_ID_H264)
1579 if (atom.size == 16) {
1580 original_size = par->extradata_size;
1581 ret = mov_realloc_extradata(par, atom);
1583 length = mov_read_atom_into_extradata(c, pb, atom, par, par->extradata + original_size);
1584 if (length == atom.size) {
1585 const uint8_t range_value = par->extradata[original_size + 19];
1586 switch (range_value) {
1588 par->color_range = AVCOL_RANGE_MPEG;
1591 par->color_range = AVCOL_RANGE_JPEG;
1594 av_log(c, AV_LOG_WARNING, "ignored unknown aclr value (%d)\n", range_value);
1597 ff_dlog(c, "color_range: %d\n", par->color_range);
1599 /* For some reason the whole atom was not added to the extradata */
1600 av_log(c, AV_LOG_ERROR, "aclr not decoded - incomplete atom\n");
1603 av_log(c, AV_LOG_ERROR, "aclr not decoded - unable to add atom to extradata\n");
1606 av_log(c, AV_LOG_WARNING, "aclr not decoded - unexpected size %"PRId64"\n", atom.size);
1613 static int mov_read_svq3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1615 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_SVQ3);
1618 static int mov_read_wave(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1623 if (c->fc->nb_streams < 1)
1625 st = c->fc->streams[c->fc->nb_streams-1];
1627 if ((uint64_t)atom.size > (1<<30))
1628 return AVERROR_INVALIDDATA;
1630 if (st->codecpar->codec_id == AV_CODEC_ID_QDM2 ||
1631 st->codecpar->codec_id == AV_CODEC_ID_QDMC ||
1632 st->codecpar->codec_id == AV_CODEC_ID_SPEEX) {
1633 // pass all frma atom to codec, needed at least for QDMC and QDM2
1634 av_freep(&st->codecpar->extradata);
1635 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1638 } else if (atom.size > 8) { /* to read frma, esds atoms */
1639 if (st->codecpar->codec_id == AV_CODEC_ID_ALAC && atom.size >= 24) {
1641 ret = ffio_ensure_seekback(pb, 8);
1644 buffer = avio_rb64(pb);
1646 if ( (buffer & 0xFFFFFFFF) == MKBETAG('f','r','m','a')
1647 && buffer >> 32 <= atom.size
1648 && buffer >> 32 >= 8) {
1651 } else if (!st->codecpar->extradata_size) {
1652 #define ALAC_EXTRADATA_SIZE 36
1653 st->codecpar->extradata = av_mallocz(ALAC_EXTRADATA_SIZE + AV_INPUT_BUFFER_PADDING_SIZE);
1654 if (!st->codecpar->extradata)
1655 return AVERROR(ENOMEM);
1656 st->codecpar->extradata_size = ALAC_EXTRADATA_SIZE;
1657 AV_WB32(st->codecpar->extradata , ALAC_EXTRADATA_SIZE);
1658 AV_WB32(st->codecpar->extradata + 4, MKTAG('a','l','a','c'));
1659 AV_WB64(st->codecpar->extradata + 12, buffer);
1660 avio_read(pb, st->codecpar->extradata + 20, 16);
1661 avio_skip(pb, atom.size - 24);
1665 if ((ret = mov_read_default(c, pb, atom)) < 0)
1668 avio_skip(pb, atom.size);
1673 * This function reads atom content and puts data in extradata without tag
1674 * nor size unlike mov_read_extradata.
1676 static int mov_read_glbl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1681 if (c->fc->nb_streams < 1)
1683 st = c->fc->streams[c->fc->nb_streams-1];
1685 if ((uint64_t)atom.size > (1<<30))
1686 return AVERROR_INVALIDDATA;
1688 if (atom.size >= 10) {
1689 // Broken files created by legacy versions of libavformat will
1690 // wrap a whole fiel atom inside of a glbl atom.
1691 unsigned size = avio_rb32(pb);
1692 unsigned type = avio_rl32(pb);
1693 avio_seek(pb, -8, SEEK_CUR);
1694 if (type == MKTAG('f','i','e','l') && size == atom.size)
1695 return mov_read_default(c, pb, atom);
1697 if (st->codecpar->extradata_size > 1 && st->codecpar->extradata) {
1698 av_log(c, AV_LOG_WARNING, "ignoring multiple glbl\n");
1701 av_freep(&st->codecpar->extradata);
1702 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1709 static int mov_read_dvc1(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1712 uint8_t profile_level;
1715 if (c->fc->nb_streams < 1)
1717 st = c->fc->streams[c->fc->nb_streams-1];
1719 if (atom.size >= (1<<28) || atom.size < 7)
1720 return AVERROR_INVALIDDATA;
1722 profile_level = avio_r8(pb);
1723 if ((profile_level & 0xf0) != 0xc0)
1726 avio_seek(pb, 6, SEEK_CUR);
1727 av_freep(&st->codecpar->extradata);
1728 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 7);
1736 * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
1737 * but can have extradata appended at the end after the 40 bytes belonging
1740 static int mov_read_strf(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1745 if (c->fc->nb_streams < 1)
1747 if (atom.size <= 40)
1749 st = c->fc->streams[c->fc->nb_streams-1];
1751 if ((uint64_t)atom.size > (1<<30))
1752 return AVERROR_INVALIDDATA;
1755 av_freep(&st->codecpar->extradata);
1756 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 40);
1763 static int mov_read_stco(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1766 MOVStreamContext *sc;
1767 unsigned int i, entries;
1769 if (c->fc->nb_streams < 1)
1771 st = c->fc->streams[c->fc->nb_streams-1];
1774 avio_r8(pb); /* version */
1775 avio_rb24(pb); /* flags */
1777 entries = avio_rb32(pb);
1782 if (sc->chunk_offsets)
1783 av_log(c->fc, AV_LOG_WARNING, "Duplicated STCO atom\n");
1784 av_free(sc->chunk_offsets);
1785 sc->chunk_count = 0;
1786 sc->chunk_offsets = av_malloc_array(entries, sizeof(*sc->chunk_offsets));
1787 if (!sc->chunk_offsets)
1788 return AVERROR(ENOMEM);
1789 sc->chunk_count = entries;
1791 if (atom.type == MKTAG('s','t','c','o'))
1792 for (i = 0; i < entries && !pb->eof_reached; i++)
1793 sc->chunk_offsets[i] = avio_rb32(pb);
1794 else if (atom.type == MKTAG('c','o','6','4'))
1795 for (i = 0; i < entries && !pb->eof_reached; i++)
1796 sc->chunk_offsets[i] = avio_rb64(pb);
1798 return AVERROR_INVALIDDATA;
1800 sc->chunk_count = i;
1802 if (pb->eof_reached)
1809 * Compute codec id for 'lpcm' tag.
1810 * See CoreAudioTypes and AudioStreamBasicDescription at Apple.
1812 enum AVCodecID ff_mov_get_lpcm_codec_id(int bps, int flags)
1819 return ff_get_pcm_codec_id(bps, flags & 1, flags & 2, flags & 4 ? -1 : 0);
1822 static int mov_codec_id(AVStream *st, uint32_t format)
1824 int id = ff_codec_get_id(ff_codec_movaudio_tags, format);
1827 ((format & 0xFFFF) == 'm' + ('s' << 8) ||
1828 (format & 0xFFFF) == 'T' + ('S' << 8)))
1829 id = ff_codec_get_id(ff_codec_wav_tags, av_bswap32(format) & 0xFFFF);
1831 if (st->codecpar->codec_type != AVMEDIA_TYPE_VIDEO && id > 0) {
1832 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
1833 } else if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO &&
1834 /* skip old ASF MPEG-4 tag */
1835 format && format != MKTAG('m','p','4','s')) {
1836 id = ff_codec_get_id(ff_codec_movvideo_tags, format);
1838 id = ff_codec_get_id(ff_codec_bmp_tags, format);
1840 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
1841 else if (st->codecpar->codec_type == AVMEDIA_TYPE_DATA ||
1842 (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE &&
1843 st->codecpar->codec_id == AV_CODEC_ID_NONE)) {
1844 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
1846 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
1850 st->codecpar->codec_tag = format;
1855 static void mov_parse_stsd_video(MOVContext *c, AVIOContext *pb,
1856 AVStream *st, MOVStreamContext *sc)
1858 uint8_t codec_name[32] = { 0 };
1862 /* The first 16 bytes of the video sample description are already
1863 * read in ff_mov_read_stsd_entries() */
1864 stsd_start = avio_tell(pb) - 16;
1866 avio_rb16(pb); /* version */
1867 avio_rb16(pb); /* revision level */
1868 avio_rb32(pb); /* vendor */
1869 avio_rb32(pb); /* temporal quality */
1870 avio_rb32(pb); /* spatial quality */
1872 st->codecpar->width = avio_rb16(pb); /* width */
1873 st->codecpar->height = avio_rb16(pb); /* height */
1875 avio_rb32(pb); /* horiz resolution */
1876 avio_rb32(pb); /* vert resolution */
1877 avio_rb32(pb); /* data size, always 0 */
1878 avio_rb16(pb); /* frames per samples */
1880 len = avio_r8(pb); /* codec name, pascal string */
1883 mov_read_mac_string(c, pb, len, codec_name, sizeof(codec_name));
1885 avio_skip(pb, 31 - len);
1888 av_dict_set(&st->metadata, "encoder", codec_name, 0);
1890 /* codec_tag YV12 triggers an UV swap in rawdec.c */
1891 if (!memcmp(codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25)) {
1892 st->codecpar->codec_tag = MKTAG('I', '4', '2', '0');
1893 st->codecpar->width &= ~1;
1894 st->codecpar->height &= ~1;
1896 /* Flash Media Server uses tag H.263 with Sorenson Spark */
1897 if (st->codecpar->codec_tag == MKTAG('H','2','6','3') &&
1898 !memcmp(codec_name, "Sorenson H263", 13))
1899 st->codecpar->codec_id = AV_CODEC_ID_FLV1;
1901 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* depth */
1903 avio_seek(pb, stsd_start, SEEK_SET);
1905 if (ff_get_qtpalette(st->codecpar->codec_id, pb, sc->palette)) {
1906 st->codecpar->bits_per_coded_sample &= 0x1F;
1907 sc->has_palette = 1;
1911 static void mov_parse_stsd_audio(MOVContext *c, AVIOContext *pb,
1912 AVStream *st, MOVStreamContext *sc)
1914 int bits_per_sample, flags;
1915 uint16_t version = avio_rb16(pb);
1916 AVDictionaryEntry *compatible_brands = av_dict_get(c->fc->metadata, "compatible_brands", NULL, AV_DICT_MATCH_CASE);
1918 avio_rb16(pb); /* revision level */
1919 avio_rb32(pb); /* vendor */
1921 st->codecpar->channels = avio_rb16(pb); /* channel count */
1922 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* sample size */
1923 av_log(c->fc, AV_LOG_TRACE, "audio channels %d\n", st->codecpar->channels);
1925 sc->audio_cid = avio_rb16(pb);
1926 avio_rb16(pb); /* packet size = 0 */
1928 st->codecpar->sample_rate = ((avio_rb32(pb) >> 16));
1930 // Read QT version 1 fields. In version 0 these do not exist.
1931 av_log(c->fc, AV_LOG_TRACE, "version =%d, isom =%d\n", version, c->isom);
1933 (compatible_brands && strstr(compatible_brands->value, "qt "))) {
1936 sc->samples_per_frame = avio_rb32(pb);
1937 avio_rb32(pb); /* bytes per packet */
1938 sc->bytes_per_frame = avio_rb32(pb);
1939 avio_rb32(pb); /* bytes per sample */
1940 } else if (version == 2) {
1941 avio_rb32(pb); /* sizeof struct only */
1942 st->codecpar->sample_rate = av_int2double(avio_rb64(pb));
1943 st->codecpar->channels = avio_rb32(pb);
1944 avio_rb32(pb); /* always 0x7F000000 */
1945 st->codecpar->bits_per_coded_sample = avio_rb32(pb);
1947 flags = avio_rb32(pb); /* lpcm format specific flag */
1948 sc->bytes_per_frame = avio_rb32(pb);
1949 sc->samples_per_frame = avio_rb32(pb);
1950 if (st->codecpar->codec_tag == MKTAG('l','p','c','m'))
1951 st->codecpar->codec_id =
1952 ff_mov_get_lpcm_codec_id(st->codecpar->bits_per_coded_sample,
1955 if (version == 0 || (version == 1 && sc->audio_cid != -2)) {
1956 /* can't correctly handle variable sized packet as audio unit */
1957 switch (st->codecpar->codec_id) {
1958 case AV_CODEC_ID_MP2:
1959 case AV_CODEC_ID_MP3:
1960 st->need_parsing = AVSTREAM_PARSE_FULL;
1966 if (sc->format == 0) {
1967 if (st->codecpar->bits_per_coded_sample == 8)
1968 st->codecpar->codec_id = mov_codec_id(st, MKTAG('r','a','w',' '));
1969 else if (st->codecpar->bits_per_coded_sample == 16)
1970 st->codecpar->codec_id = mov_codec_id(st, MKTAG('t','w','o','s'));
1973 switch (st->codecpar->codec_id) {
1974 case AV_CODEC_ID_PCM_S8:
1975 case AV_CODEC_ID_PCM_U8:
1976 if (st->codecpar->bits_per_coded_sample == 16)
1977 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16BE;
1979 case AV_CODEC_ID_PCM_S16LE:
1980 case AV_CODEC_ID_PCM_S16BE:
1981 if (st->codecpar->bits_per_coded_sample == 8)
1982 st->codecpar->codec_id = AV_CODEC_ID_PCM_S8;
1983 else if (st->codecpar->bits_per_coded_sample == 24)
1984 st->codecpar->codec_id =
1985 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
1986 AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
1987 else if (st->codecpar->bits_per_coded_sample == 32)
1988 st->codecpar->codec_id =
1989 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
1990 AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
1992 /* set values for old format before stsd version 1 appeared */
1993 case AV_CODEC_ID_MACE3:
1994 sc->samples_per_frame = 6;
1995 sc->bytes_per_frame = 2 * st->codecpar->channels;
1997 case AV_CODEC_ID_MACE6:
1998 sc->samples_per_frame = 6;
1999 sc->bytes_per_frame = 1 * st->codecpar->channels;
2001 case AV_CODEC_ID_ADPCM_IMA_QT:
2002 sc->samples_per_frame = 64;
2003 sc->bytes_per_frame = 34 * st->codecpar->channels;
2005 case AV_CODEC_ID_GSM:
2006 sc->samples_per_frame = 160;
2007 sc->bytes_per_frame = 33;
2013 bits_per_sample = av_get_bits_per_sample(st->codecpar->codec_id);
2014 if (bits_per_sample) {
2015 st->codecpar->bits_per_coded_sample = bits_per_sample;
2016 sc->sample_size = (bits_per_sample >> 3) * st->codecpar->channels;
2020 static void mov_parse_stsd_subtitle(MOVContext *c, AVIOContext *pb,
2021 AVStream *st, MOVStreamContext *sc,
2024 // ttxt stsd contains display flags, justification, background
2025 // color, fonts, and default styles, so fake an atom to read it
2026 MOVAtom fake_atom = { .size = size };
2027 // mp4s contains a regular esds atom
2028 if (st->codecpar->codec_tag != AV_RL32("mp4s"))
2029 mov_read_glbl(c, pb, fake_atom);
2030 st->codecpar->width = sc->width;
2031 st->codecpar->height = sc->height;
2034 static uint32_t yuv_to_rgba(uint32_t ycbcr)
2039 y = (ycbcr >> 16) & 0xFF;
2040 cr = (ycbcr >> 8) & 0xFF;
2043 b = av_clip_uint8((1164 * (y - 16) + 2018 * (cb - 128)) / 1000);
2044 g = av_clip_uint8((1164 * (y - 16) - 813 * (cr - 128) - 391 * (cb - 128)) / 1000);
2045 r = av_clip_uint8((1164 * (y - 16) + 1596 * (cr - 128) ) / 1000);
2047 return (r << 16) | (g << 8) | b;
2050 static int mov_rewrite_dvd_sub_extradata(AVStream *st)
2052 char buf[256] = {0};
2053 uint8_t *src = st->codecpar->extradata;
2056 if (st->codecpar->extradata_size != 64)
2059 if (st->codecpar->width > 0 && st->codecpar->height > 0)
2060 snprintf(buf, sizeof(buf), "size: %dx%d\n",
2061 st->codecpar->width, st->codecpar->height);
2062 av_strlcat(buf, "palette: ", sizeof(buf));
2064 for (i = 0; i < 16; i++) {
2065 uint32_t yuv = AV_RB32(src + i * 4);
2066 uint32_t rgba = yuv_to_rgba(yuv);
2068 av_strlcatf(buf, sizeof(buf), "%06"PRIx32"%s", rgba, i != 15 ? ", " : "");
2071 if (av_strlcat(buf, "\n", sizeof(buf)) >= sizeof(buf))
2074 av_freep(&st->codecpar->extradata);
2075 st->codecpar->extradata_size = 0;
2076 st->codecpar->extradata = av_mallocz(strlen(buf) + AV_INPUT_BUFFER_PADDING_SIZE);
2077 if (!st->codecpar->extradata)
2078 return AVERROR(ENOMEM);
2079 st->codecpar->extradata_size = strlen(buf);
2080 memcpy(st->codecpar->extradata, buf, st->codecpar->extradata_size);
2085 static int mov_parse_stsd_data(MOVContext *c, AVIOContext *pb,
2086 AVStream *st, MOVStreamContext *sc,
2091 if (st->codecpar->codec_tag == MKTAG('t','m','c','d')) {
2092 if ((int)size != size)
2093 return AVERROR(ENOMEM);
2095 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
2099 MOVStreamContext *tmcd_ctx = st->priv_data;
2101 val = AV_RB32(st->codecpar->extradata + 4);
2102 tmcd_ctx->tmcd_flags = val;
2103 st->avg_frame_rate.num = st->codecpar->extradata[16]; /* number of frame */
2104 st->avg_frame_rate.den = 1;
2105 #if FF_API_LAVF_AVCTX
2106 FF_DISABLE_DEPRECATION_WARNINGS
2107 st->codec->time_base = av_inv_q(st->avg_frame_rate);
2108 FF_ENABLE_DEPRECATION_WARNINGS
2110 /* adjust for per frame dur in counter mode */
2111 if (tmcd_ctx->tmcd_flags & 0x0008) {
2112 int timescale = AV_RB32(st->codecpar->extradata + 8);
2113 int framedur = AV_RB32(st->codecpar->extradata + 12);
2114 st->avg_frame_rate.num *= timescale;
2115 st->avg_frame_rate.den *= framedur;
2116 #if FF_API_LAVF_AVCTX
2117 FF_DISABLE_DEPRECATION_WARNINGS
2118 st->codec->time_base.den *= timescale;
2119 st->codec->time_base.num *= framedur;
2120 FF_ENABLE_DEPRECATION_WARNINGS
2124 uint32_t len = AV_RB32(st->codecpar->extradata + 18); /* name atom length */
2125 uint32_t format = AV_RB32(st->codecpar->extradata + 22);
2126 if (format == AV_RB32("name") && (int64_t)size >= (int64_t)len + 18) {
2127 uint16_t str_size = AV_RB16(st->codecpar->extradata + 26); /* string length */
2128 if (str_size > 0 && size >= (int)str_size + 26) {
2129 char *reel_name = av_malloc(str_size + 1);
2131 return AVERROR(ENOMEM);
2132 memcpy(reel_name, st->codecpar->extradata + 30, str_size);
2133 reel_name[str_size] = 0; /* Add null terminator */
2134 /* don't add reel_name if emtpy string */
2135 if (*reel_name == 0) {
2138 av_dict_set(&st->metadata, "reel_name", reel_name, AV_DICT_DONT_STRDUP_VAL);
2145 /* other codec type, just skip (rtp, mp4s ...) */
2146 avio_skip(pb, size);
2151 static int mov_finalize_stsd_codec(MOVContext *c, AVIOContext *pb,
2152 AVStream *st, MOVStreamContext *sc)
2154 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
2155 !st->codecpar->sample_rate && sc->time_scale > 1)
2156 st->codecpar->sample_rate = sc->time_scale;
2158 /* special codec parameters handling */
2159 switch (st->codecpar->codec_id) {
2160 #if CONFIG_DV_DEMUXER
2161 case AV_CODEC_ID_DVAUDIO:
2162 c->dv_fctx = avformat_alloc_context();
2164 av_log(c->fc, AV_LOG_ERROR, "dv demux context alloc error\n");
2165 return AVERROR(ENOMEM);
2167 c->dv_demux = avpriv_dv_init_demux(c->dv_fctx);
2169 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
2170 return AVERROR(ENOMEM);
2172 sc->dv_audio_container = 1;
2173 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16LE;
2176 /* no ifdef since parameters are always those */
2177 case AV_CODEC_ID_QCELP:
2178 st->codecpar->channels = 1;
2179 // force sample rate for qcelp when not stored in mov
2180 if (st->codecpar->codec_tag != MKTAG('Q','c','l','p'))
2181 st->codecpar->sample_rate = 8000;
2182 // FIXME: Why is the following needed for some files?
2183 sc->samples_per_frame = 160;
2184 if (!sc->bytes_per_frame)
2185 sc->bytes_per_frame = 35;
2187 case AV_CODEC_ID_AMR_NB:
2188 st->codecpar->channels = 1;
2189 /* force sample rate for amr, stsd in 3gp does not store sample rate */
2190 st->codecpar->sample_rate = 8000;
2192 case AV_CODEC_ID_AMR_WB:
2193 st->codecpar->channels = 1;
2194 st->codecpar->sample_rate = 16000;
2196 case AV_CODEC_ID_MP2:
2197 case AV_CODEC_ID_MP3:
2198 /* force type after stsd for m1a hdlr */
2199 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2201 case AV_CODEC_ID_GSM:
2202 case AV_CODEC_ID_ADPCM_MS:
2203 case AV_CODEC_ID_ADPCM_IMA_WAV:
2204 case AV_CODEC_ID_ILBC:
2205 case AV_CODEC_ID_MACE3:
2206 case AV_CODEC_ID_MACE6:
2207 case AV_CODEC_ID_QDM2:
2208 st->codecpar->block_align = sc->bytes_per_frame;
2210 case AV_CODEC_ID_ALAC:
2211 if (st->codecpar->extradata_size == 36) {
2212 st->codecpar->channels = AV_RB8 (st->codecpar->extradata + 21);
2213 st->codecpar->sample_rate = AV_RB32(st->codecpar->extradata + 32);
2216 case AV_CODEC_ID_AC3:
2217 case AV_CODEC_ID_EAC3:
2218 case AV_CODEC_ID_MPEG1VIDEO:
2219 case AV_CODEC_ID_VC1:
2220 case AV_CODEC_ID_VP9:
2221 st->need_parsing = AVSTREAM_PARSE_FULL;
2229 static int mov_skip_multiple_stsd(MOVContext *c, AVIOContext *pb,
2230 int codec_tag, int format,
2233 int video_codec_id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2236 (codec_tag != format &&
2237 // AVID 1:1 samples with differing data format and codec tag exist
2238 (codec_tag != AV_RL32("AV1x") || format != AV_RL32("AVup")) &&
2239 // prores is allowed to have differing data format and codec tag
2240 codec_tag != AV_RL32("apcn") && codec_tag != AV_RL32("apch") &&
2242 codec_tag != AV_RL32("dvpp") && codec_tag != AV_RL32("dvcp") &&
2243 (c->fc->video_codec_id ? video_codec_id != c->fc->video_codec_id
2244 : codec_tag != MKTAG('j','p','e','g')))) {
2245 /* Multiple fourcc, we skip JPEG. This is not correct, we should
2246 * export it as a separate AVStream but this needs a few changes
2247 * in the MOV demuxer, patch welcome. */
2249 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
2250 avio_skip(pb, size);
2257 int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries)
2260 MOVStreamContext *sc;
2261 int pseudo_stream_id;
2263 if (c->fc->nb_streams < 1)
2265 st = c->fc->streams[c->fc->nb_streams-1];
2268 for (pseudo_stream_id = 0;
2269 pseudo_stream_id < entries && !pb->eof_reached;
2270 pseudo_stream_id++) {
2271 //Parsing Sample description table
2273 int ret, dref_id = 1;
2274 MOVAtom a = { AV_RL32("stsd") };
2275 int64_t start_pos = avio_tell(pb);
2276 int64_t size = avio_rb32(pb); /* size */
2277 uint32_t format = avio_rl32(pb); /* data format */
2280 avio_rb32(pb); /* reserved */
2281 avio_rb16(pb); /* reserved */
2282 dref_id = avio_rb16(pb);
2283 } else if (size <= 7) {
2284 av_log(c->fc, AV_LOG_ERROR,
2285 "invalid size %"PRId64" in stsd\n", size);
2286 return AVERROR_INVALIDDATA;
2289 if (mov_skip_multiple_stsd(c, pb, st->codecpar->codec_tag, format,
2290 size - (avio_tell(pb) - start_pos)))
2293 sc->pseudo_stream_id = st->codecpar->codec_tag ? -1 : pseudo_stream_id;
2294 sc->dref_id= dref_id;
2295 sc->format = format;
2297 id = mov_codec_id(st, format);
2299 av_log(c->fc, AV_LOG_TRACE,
2300 "size=%"PRId64" 4CC=%s codec_type=%d\n", size,
2301 av_fourcc2str(format), st->codecpar->codec_type);
2303 if (st->codecpar->codec_type==AVMEDIA_TYPE_VIDEO) {
2304 st->codecpar->codec_id = id;
2305 mov_parse_stsd_video(c, pb, st, sc);
2306 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_AUDIO) {
2307 st->codecpar->codec_id = id;
2308 mov_parse_stsd_audio(c, pb, st, sc);
2309 if (st->codecpar->sample_rate < 0) {
2310 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
2311 return AVERROR_INVALIDDATA;
2313 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_SUBTITLE){
2314 st->codecpar->codec_id = id;
2315 mov_parse_stsd_subtitle(c, pb, st, sc,
2316 size - (avio_tell(pb) - start_pos));
2318 ret = mov_parse_stsd_data(c, pb, st, sc,
2319 size - (avio_tell(pb) - start_pos));
2323 /* this will read extra atoms at the end (wave, alac, damr, avcC, hvcC, SMI ...) */
2324 a.size = size - (avio_tell(pb) - start_pos);
2326 if ((ret = mov_read_default(c, pb, a)) < 0)
2328 } else if (a.size > 0)
2329 avio_skip(pb, a.size);
2331 if (sc->extradata && st->codecpar->extradata) {
2332 int extra_size = st->codecpar->extradata_size;
2334 /* Move the current stream extradata to the stream context one. */
2335 sc->extradata_size[pseudo_stream_id] = extra_size;
2336 sc->extradata[pseudo_stream_id] = av_malloc(extra_size + AV_INPUT_BUFFER_PADDING_SIZE);
2337 if (!sc->extradata[pseudo_stream_id])
2338 return AVERROR(ENOMEM);
2339 memcpy(sc->extradata[pseudo_stream_id], st->codecpar->extradata, extra_size);
2340 av_freep(&st->codecpar->extradata);
2341 st->codecpar->extradata_size = 0;
2345 if (pb->eof_reached)
2351 static int mov_read_stsd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2354 MOVStreamContext *sc;
2357 if (c->fc->nb_streams < 1)
2359 st = c->fc->streams[c->fc->nb_streams - 1];
2362 avio_r8(pb); /* version */
2363 avio_rb24(pb); /* flags */
2364 entries = avio_rb32(pb);
2367 av_log(c->fc, AV_LOG_ERROR, "invalid STSD entries %d\n", entries);
2368 return AVERROR_INVALIDDATA;
2371 if (sc->extradata) {
2372 av_log(c->fc, AV_LOG_ERROR, "Duplicate STSD\n");
2373 return AVERROR_INVALIDDATA;
2375 /* Prepare space for hosting multiple extradata. */
2376 sc->extradata = av_mallocz_array(entries, sizeof(*sc->extradata));
2377 sc->extradata_size = av_mallocz_array(entries, sizeof(*sc->extradata_size));
2378 if (!sc->extradata_size || !sc->extradata) {
2379 ret = AVERROR(ENOMEM);
2383 ret = ff_mov_read_stsd_entries(c, pb, entries);
2387 sc->stsd_count = entries;
2389 /* Restore back the primary extradata. */
2390 av_freep(&st->codecpar->extradata);
2391 st->codecpar->extradata_size = sc->extradata_size[0];
2392 if (sc->extradata_size[0]) {
2393 st->codecpar->extradata = av_mallocz(sc->extradata_size[0] + AV_INPUT_BUFFER_PADDING_SIZE);
2394 if (!st->codecpar->extradata)
2395 return AVERROR(ENOMEM);
2396 memcpy(st->codecpar->extradata, sc->extradata[0], sc->extradata_size[0]);
2399 return mov_finalize_stsd_codec(c, pb, st, sc);
2401 av_freep(&sc->extradata);
2402 av_freep(&sc->extradata_size);
2406 static int mov_read_stsc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2409 MOVStreamContext *sc;
2410 unsigned int i, entries;
2412 if (c->fc->nb_streams < 1)
2414 st = c->fc->streams[c->fc->nb_streams-1];
2417 avio_r8(pb); /* version */
2418 avio_rb24(pb); /* flags */
2420 entries = avio_rb32(pb);
2422 av_log(c->fc, AV_LOG_TRACE, "track[%u].stsc.entries = %u\n", c->fc->nb_streams - 1, entries);
2427 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSC atom\n");
2428 av_free(sc->stsc_data);
2430 sc->stsc_data = av_malloc_array(entries, sizeof(*sc->stsc_data));
2432 return AVERROR(ENOMEM);
2434 for (i = 0; i < entries && !pb->eof_reached; i++) {
2435 sc->stsc_data[i].first = avio_rb32(pb);
2436 sc->stsc_data[i].count = avio_rb32(pb);
2437 sc->stsc_data[i].id = avio_rb32(pb);
2442 if (pb->eof_reached)
2448 #define mov_stsc_index_valid(index, count) ((index) < (count) - 1)
2450 /* Compute the samples value for the stsc entry at the given index. */
2451 static inline int mov_get_stsc_samples(MOVStreamContext *sc, int index)
2455 if (mov_stsc_index_valid(index, sc->stsc_count))
2456 chunk_count = sc->stsc_data[index + 1].first - sc->stsc_data[index].first;
2458 chunk_count = sc->chunk_count - (sc->stsc_data[index].first - 1);
2460 return sc->stsc_data[index].count * chunk_count;
2463 static int mov_read_stps(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2466 MOVStreamContext *sc;
2467 unsigned i, entries;
2469 if (c->fc->nb_streams < 1)
2471 st = c->fc->streams[c->fc->nb_streams-1];
2474 avio_rb32(pb); // version + flags
2476 entries = avio_rb32(pb);
2478 av_log(c->fc, AV_LOG_WARNING, "Duplicated STPS atom\n");
2479 av_free(sc->stps_data);
2481 sc->stps_data = av_malloc_array(entries, sizeof(*sc->stps_data));
2483 return AVERROR(ENOMEM);
2485 for (i = 0; i < entries && !pb->eof_reached; i++) {
2486 sc->stps_data[i] = avio_rb32(pb);
2491 if (pb->eof_reached)
2497 static int mov_read_stss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2500 MOVStreamContext *sc;
2501 unsigned int i, entries;
2503 if (c->fc->nb_streams < 1)
2505 st = c->fc->streams[c->fc->nb_streams-1];
2508 avio_r8(pb); /* version */
2509 avio_rb24(pb); /* flags */
2511 entries = avio_rb32(pb);
2513 av_log(c->fc, AV_LOG_TRACE, "keyframe_count = %u\n", entries);
2517 sc->keyframe_absent = 1;
2518 if (!st->need_parsing && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
2519 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2523 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSS atom\n");
2524 if (entries >= UINT_MAX / sizeof(int))
2525 return AVERROR_INVALIDDATA;
2526 av_freep(&sc->keyframes);
2527 sc->keyframe_count = 0;
2528 sc->keyframes = av_malloc_array(entries, sizeof(*sc->keyframes));
2530 return AVERROR(ENOMEM);
2532 for (i = 0; i < entries && !pb->eof_reached; i++) {
2533 sc->keyframes[i] = avio_rb32(pb);
2536 sc->keyframe_count = i;
2538 if (pb->eof_reached)
2544 static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2547 MOVStreamContext *sc;
2548 unsigned int i, entries, sample_size, field_size, num_bytes;
2553 if (c->fc->nb_streams < 1)
2555 st = c->fc->streams[c->fc->nb_streams-1];
2558 avio_r8(pb); /* version */
2559 avio_rb24(pb); /* flags */
2561 if (atom.type == MKTAG('s','t','s','z')) {
2562 sample_size = avio_rb32(pb);
2563 if (!sc->sample_size) /* do not overwrite value computed in stsd */
2564 sc->sample_size = sample_size;
2565 sc->stsz_sample_size = sample_size;
2569 avio_rb24(pb); /* reserved */
2570 field_size = avio_r8(pb);
2572 entries = avio_rb32(pb);
2574 av_log(c->fc, AV_LOG_TRACE, "sample_size = %u sample_count = %u\n", sc->sample_size, entries);
2576 sc->sample_count = entries;
2580 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
2581 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %u\n", field_size);
2582 return AVERROR_INVALIDDATA;
2587 if (entries >= (UINT_MAX - 4) / field_size)
2588 return AVERROR_INVALIDDATA;
2589 if (sc->sample_sizes)
2590 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSZ atom\n");
2591 av_free(sc->sample_sizes);
2592 sc->sample_count = 0;
2593 sc->sample_sizes = av_malloc_array(entries, sizeof(*sc->sample_sizes));
2594 if (!sc->sample_sizes)
2595 return AVERROR(ENOMEM);
2597 num_bytes = (entries*field_size+4)>>3;
2599 buf = av_malloc(num_bytes+AV_INPUT_BUFFER_PADDING_SIZE);
2601 av_freep(&sc->sample_sizes);
2602 return AVERROR(ENOMEM);
2605 ret = ffio_read_size(pb, buf, num_bytes);
2607 av_freep(&sc->sample_sizes);
2612 init_get_bits(&gb, buf, 8*num_bytes);
2614 for (i = 0; i < entries && !pb->eof_reached; i++) {
2615 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
2616 sc->data_size += sc->sample_sizes[i];
2619 sc->sample_count = i;
2623 if (pb->eof_reached)
2629 static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2632 MOVStreamContext *sc;
2633 unsigned int i, entries;
2635 int64_t total_sample_count=0;
2637 if (c->fc->nb_streams < 1)
2639 st = c->fc->streams[c->fc->nb_streams-1];
2642 avio_r8(pb); /* version */
2643 avio_rb24(pb); /* flags */
2644 entries = avio_rb32(pb);
2646 av_log(c->fc, AV_LOG_TRACE, "track[%u].stts.entries = %u\n",
2647 c->fc->nb_streams-1, entries);
2650 av_log(c->fc, AV_LOG_WARNING, "Duplicated STTS atom\n");
2651 av_free(sc->stts_data);
2653 sc->stts_data = av_malloc_array(entries, sizeof(*sc->stts_data));
2655 return AVERROR(ENOMEM);
2657 for (i = 0; i < entries && !pb->eof_reached; i++) {
2658 int sample_duration;
2661 sample_count=avio_rb32(pb);
2662 sample_duration = avio_rb32(pb);
2664 if (sample_count < 0) {
2665 av_log(c->fc, AV_LOG_ERROR, "Invalid sample_count=%d\n", sample_count);
2666 return AVERROR_INVALIDDATA;
2668 sc->stts_data[i].count= sample_count;
2669 sc->stts_data[i].duration= sample_duration;
2671 av_log(c->fc, AV_LOG_TRACE, "sample_count=%d, sample_duration=%d\n",
2672 sample_count, sample_duration);
2676 && sample_count == 1
2677 && total_sample_count > 100
2678 && sample_duration/10 > duration / total_sample_count)
2679 sample_duration = duration / total_sample_count;
2680 duration+=(int64_t)sample_duration*sample_count;
2681 total_sample_count+=sample_count;
2686 sc->duration_for_fps += duration;
2687 sc->nb_frames_for_fps += total_sample_count;
2689 if (pb->eof_reached)
2692 st->nb_frames= total_sample_count;
2694 st->duration= duration;
2695 sc->track_end = duration;
2699 static void mov_update_dts_shift(MOVStreamContext *sc, int duration)
2702 if (duration == INT_MIN) {
2703 av_log(NULL, AV_LOG_WARNING, "mov_update_dts_shift(): dts_shift set to %d\n", INT_MAX);
2706 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
2710 static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2713 MOVStreamContext *sc;
2714 unsigned int i, entries, ctts_count = 0;
2716 if (c->fc->nb_streams < 1)
2718 st = c->fc->streams[c->fc->nb_streams-1];
2721 avio_r8(pb); /* version */
2722 avio_rb24(pb); /* flags */
2723 entries = avio_rb32(pb);
2725 av_log(c->fc, AV_LOG_TRACE, "track[%u].ctts.entries = %u\n", c->fc->nb_streams - 1, entries);
2729 if (entries >= UINT_MAX / sizeof(*sc->ctts_data))
2730 return AVERROR_INVALIDDATA;
2731 av_freep(&sc->ctts_data);
2732 sc->ctts_data = av_realloc(NULL, entries * sizeof(*sc->ctts_data));
2734 return AVERROR(ENOMEM);
2736 for (i = 0; i < entries && !pb->eof_reached; i++) {
2737 int count =avio_rb32(pb);
2738 int duration =avio_rb32(pb);
2741 av_log(c->fc, AV_LOG_TRACE,
2742 "ignoring CTTS entry with count=%d duration=%d\n",
2747 sc->ctts_data[ctts_count].count = count;
2748 sc->ctts_data[ctts_count].duration = duration;
2751 av_log(c->fc, AV_LOG_TRACE, "count=%d, duration=%d\n",
2754 if (FFNABS(duration) < -(1<<28) && i+2<entries) {
2755 av_log(c->fc, AV_LOG_WARNING, "CTTS invalid\n");
2756 av_freep(&sc->ctts_data);
2762 mov_update_dts_shift(sc, duration);
2765 sc->ctts_count = ctts_count;
2767 if (pb->eof_reached)
2770 av_log(c->fc, AV_LOG_TRACE, "dts shift %d\n", sc->dts_shift);
2775 static int mov_read_sbgp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2778 MOVStreamContext *sc;
2779 unsigned int i, entries;
2781 uint32_t grouping_type;
2783 if (c->fc->nb_streams < 1)
2785 st = c->fc->streams[c->fc->nb_streams-1];
2788 version = avio_r8(pb); /* version */
2789 avio_rb24(pb); /* flags */
2790 grouping_type = avio_rl32(pb);
2791 if (grouping_type != MKTAG( 'r','a','p',' '))
2792 return 0; /* only support 'rap ' grouping */
2794 avio_rb32(pb); /* grouping_type_parameter */
2796 entries = avio_rb32(pb);
2800 av_log(c->fc, AV_LOG_WARNING, "Duplicated SBGP atom\n");
2801 av_free(sc->rap_group);
2802 sc->rap_group_count = 0;
2803 sc->rap_group = av_malloc_array(entries, sizeof(*sc->rap_group));
2805 return AVERROR(ENOMEM);
2807 for (i = 0; i < entries && !pb->eof_reached; i++) {
2808 sc->rap_group[i].count = avio_rb32(pb); /* sample_count */
2809 sc->rap_group[i].index = avio_rb32(pb); /* group_description_index */
2812 sc->rap_group_count = i;
2814 return pb->eof_reached ? AVERROR_EOF : 0;
2818 * Get ith edit list entry (media time, duration).
2820 static int get_edit_list_entry(MOVContext *mov,
2821 const MOVStreamContext *msc,
2822 unsigned int edit_list_index,
2823 int64_t *edit_list_media_time,
2824 int64_t *edit_list_duration,
2825 int64_t global_timescale)
2827 if (edit_list_index == msc->elst_count) {
2830 *edit_list_media_time = msc->elst_data[edit_list_index].time;
2831 *edit_list_duration = msc->elst_data[edit_list_index].duration;
2833 /* duration is in global timescale units;convert to msc timescale */
2834 if (global_timescale == 0) {
2835 avpriv_request_sample(mov->fc, "Support for mvhd.timescale = 0 with editlists");
2838 *edit_list_duration = av_rescale(*edit_list_duration, msc->time_scale,
2844 * Find the closest previous frame to the timestamp, in e_old index
2845 * entries. Searching for just any frame / just key frames can be controlled by
2846 * last argument 'flag'.
2847 * Returns the index of the entry in st->index_entries if successful,
2850 static int64_t find_prev_closest_index(AVStream *st,
2851 AVIndexEntry *e_old,
2856 AVIndexEntry *e_keep = st->index_entries;
2857 int nb_keep = st->nb_index_entries;
2861 st->index_entries = e_old;
2862 st->nb_index_entries = nb_old;
2863 found = av_index_search_timestamp(st, timestamp, flag | AVSEEK_FLAG_BACKWARD);
2865 // Keep going backwards in the index entries until the timestamp is the same.
2867 for (i = found; i > 0 && e_old[i].timestamp == e_old[i - 1].timestamp;
2869 if ((flag & AVSEEK_FLAG_ANY) ||
2870 (e_old[i - 1].flags & AVINDEX_KEYFRAME)) {
2876 /* restore AVStream state*/
2877 st->index_entries = e_keep;
2878 st->nb_index_entries = nb_keep;
2883 * Add index entry with the given values, to the end of st->index_entries.
2884 * Returns the new size st->index_entries if successful, else returns -1.
2886 * This function is similar to ff_add_index_entry in libavformat/utils.c
2887 * except that here we are always unconditionally adding an index entry to
2888 * the end, instead of searching the entries list and skipping the add if
2889 * there is an existing entry with the same timestamp.
2890 * This is needed because the mov_fix_index calls this func with the same
2891 * unincremented timestamp for successive discarded frames.
2893 static int64_t add_index_entry(AVStream *st, int64_t pos, int64_t timestamp,
2894 int size, int distance, int flags)
2896 AVIndexEntry *entries, *ie;
2898 const size_t min_size_needed = (st->nb_index_entries + 1) * sizeof(AVIndexEntry);
2900 // Double the allocation each time, to lower memory fragmentation.
2901 // Another difference from ff_add_index_entry function.
2902 const size_t requested_size =
2903 min_size_needed > st->index_entries_allocated_size ?
2904 FFMAX(min_size_needed, 2 * st->index_entries_allocated_size) :
2907 if((unsigned)st->nb_index_entries + 1 >= UINT_MAX / sizeof(AVIndexEntry))
2910 entries = av_fast_realloc(st->index_entries,
2911 &st->index_entries_allocated_size,
2916 st->index_entries= entries;
2918 index= st->nb_index_entries++;
2919 ie= &entries[index];
2922 ie->timestamp = timestamp;
2923 ie->min_distance= distance;
2930 * Rewrite timestamps of index entries in the range [end_index - frame_duration_buffer_size, end_index)
2931 * by subtracting end_ts successively by the amounts given in frame_duration_buffer.
2933 static void fix_index_entry_timestamps(AVStream* st, int end_index, int64_t end_ts,
2934 int64_t* frame_duration_buffer,
2935 int frame_duration_buffer_size) {
2937 av_assert0(end_index >= 0 && end_index <= st->nb_index_entries);
2938 for (i = 0; i < frame_duration_buffer_size; i++) {
2939 end_ts -= frame_duration_buffer[frame_duration_buffer_size - 1 - i];
2940 st->index_entries[end_index - 1 - i].timestamp = end_ts;
2945 * Append a new ctts entry to ctts_data.
2946 * Returns the new ctts_count if successful, else returns -1.
2948 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
2949 int count, int duration)
2951 MOVStts *ctts_buf_new;
2952 const size_t min_size_needed = (*ctts_count + 1) * sizeof(MOVStts);
2953 const size_t requested_size =
2954 min_size_needed > *allocated_size ?
2955 FFMAX(min_size_needed, 2 * (*allocated_size)) :
2958 if((unsigned)(*ctts_count) + 1 >= UINT_MAX / sizeof(MOVStts))
2961 ctts_buf_new = av_fast_realloc(*ctts_data, allocated_size, requested_size);
2966 *ctts_data = ctts_buf_new;
2968 ctts_buf_new[*ctts_count].count = count;
2969 ctts_buf_new[*ctts_count].duration = duration;
2971 *ctts_count = (*ctts_count) + 1;
2975 static void mov_current_sample_inc(MOVStreamContext *sc)
2977 sc->current_sample++;
2978 sc->current_index++;
2979 if (sc->index_ranges &&
2980 sc->current_index >= sc->current_index_range->end &&
2981 sc->current_index_range->end) {
2982 sc->current_index_range++;
2983 sc->current_index = sc->current_index_range->start;
2987 static void mov_current_sample_dec(MOVStreamContext *sc)
2989 sc->current_sample--;
2990 sc->current_index--;
2991 if (sc->index_ranges &&
2992 sc->current_index < sc->current_index_range->start &&
2993 sc->current_index_range > sc->index_ranges) {
2994 sc->current_index_range--;
2995 sc->current_index = sc->current_index_range->end - 1;
2999 static void mov_current_sample_set(MOVStreamContext *sc, int current_sample)
3003 sc->current_sample = current_sample;
3004 sc->current_index = current_sample;
3005 if (!sc->index_ranges) {
3009 for (sc->current_index_range = sc->index_ranges;
3010 sc->current_index_range->end;
3011 sc->current_index_range++) {
3012 range_size = sc->current_index_range->end - sc->current_index_range->start;
3013 if (range_size > current_sample) {
3014 sc->current_index = sc->current_index_range->start + current_sample;
3017 current_sample -= range_size;
3022 * Fix st->index_entries, so that it contains only the entries (and the entries
3023 * which are needed to decode them) that fall in the edit list time ranges.
3024 * Also fixes the timestamps of the index entries to match the timeline
3025 * specified the edit lists.
3027 static void mov_fix_index(MOVContext *mov, AVStream *st)
3029 MOVStreamContext *msc = st->priv_data;
3030 AVIndexEntry *e_old = st->index_entries;
3031 int nb_old = st->nb_index_entries;
3032 const AVIndexEntry *e_old_end = e_old + nb_old;
3033 const AVIndexEntry *current = NULL;
3034 MOVStts *ctts_data_old = msc->ctts_data;
3035 int64_t ctts_index_old = 0;
3036 int64_t ctts_sample_old = 0;
3037 int64_t ctts_count_old = msc->ctts_count;
3038 int64_t edit_list_media_time = 0;
3039 int64_t edit_list_duration = 0;
3040 int64_t frame_duration = 0;
3041 int64_t edit_list_dts_counter = 0;
3042 int64_t edit_list_dts_entry_end = 0;
3043 int64_t edit_list_start_ctts_sample = 0;
3045 int64_t curr_ctts = 0;
3046 int64_t min_corrected_pts = -1;
3047 int64_t empty_edits_sum_duration = 0;
3048 int64_t edit_list_index = 0;
3050 int64_t index_ctts_count;
3052 unsigned int ctts_allocated_size = 0;
3053 int64_t start_dts = 0;
3054 int64_t edit_list_media_time_dts = 0;
3055 int64_t edit_list_start_encountered = 0;
3056 int64_t search_timestamp = 0;
3057 int64_t* frame_duration_buffer = NULL;
3058 int num_discarded_begin = 0;
3059 int first_non_zero_audio_edit = -1;
3060 int packet_skip_samples = 0;
3061 MOVIndexRange *current_index_range;
3064 if (!msc->elst_data || msc->elst_count <= 0 || nb_old <= 0) {
3068 // allocate the index ranges array
3069 msc->index_ranges = av_malloc((msc->elst_count + 1) * sizeof(msc->index_ranges[0]));
3070 if (!msc->index_ranges) {
3071 av_log(mov->fc, AV_LOG_ERROR, "Cannot allocate index ranges buffer\n");
3074 msc->current_index_range = msc->index_ranges;
3075 current_index_range = msc->index_ranges - 1;
3077 // Clean AVStream from traces of old index
3078 st->index_entries = NULL;
3079 st->index_entries_allocated_size = 0;
3080 st->nb_index_entries = 0;
3082 // Clean ctts fields of MOVStreamContext
3083 msc->ctts_data = NULL;
3084 msc->ctts_count = 0;
3085 msc->ctts_index = 0;
3086 msc->ctts_sample = 0;
3088 // If the dts_shift is positive (in case of negative ctts values in mov),
3089 // then negate the DTS by dts_shift
3090 if (msc->dts_shift > 0) {
3091 edit_list_dts_entry_end -= msc->dts_shift;
3092 av_log(mov->fc, AV_LOG_DEBUG, "Shifting DTS by %d because of negative CTTS.\n", msc->dts_shift);
3095 start_dts = edit_list_dts_entry_end;
3097 while (get_edit_list_entry(mov, msc, edit_list_index, &edit_list_media_time,
3098 &edit_list_duration, mov->time_scale)) {
3099 av_log(mov->fc, AV_LOG_DEBUG, "Processing st: %d, edit list %"PRId64" - media time: %"PRId64", duration: %"PRId64"\n",
3100 st->index, edit_list_index, edit_list_media_time, edit_list_duration);
3102 edit_list_dts_counter = edit_list_dts_entry_end;
3103 edit_list_dts_entry_end += edit_list_duration;
3104 num_discarded_begin = 0;
3105 if (edit_list_media_time == -1) {
3106 empty_edits_sum_duration += edit_list_duration;
3110 // If we encounter a non-negative edit list reset the skip_samples/start_pad fields and set them
3111 // according to the edit list below.
3112 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3113 if (first_non_zero_audio_edit < 0) {
3114 first_non_zero_audio_edit = 1;
3116 first_non_zero_audio_edit = 0;
3119 if (first_non_zero_audio_edit > 0)
3120 st->skip_samples = msc->start_pad = 0;
3123 //find closest previous key frame
3124 edit_list_media_time_dts = edit_list_media_time;
3125 if (msc->dts_shift > 0) {
3126 edit_list_media_time_dts -= msc->dts_shift;
3129 // While reordering frame index according to edit list we must handle properly
3130 // the scenario when edit list entry starts from none key frame.
3131 // We find closest previous key frame and preserve it and consequent frames in index.
3132 // All frames which are outside edit list entry time boundaries will be dropped after decoding.
3133 search_timestamp = edit_list_media_time_dts;
3134 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3135 // Audio decoders like AAC need need a decoder delay samples previous to the current sample,
3136 // to correctly decode this frame. Hence for audio we seek to a frame 1 sec. before the
3137 // edit_list_media_time to cover the decoder delay.
3138 search_timestamp = FFMAX(search_timestamp - msc->time_scale, e_old[0].timestamp);
3141 index = find_prev_closest_index(st, e_old, nb_old, search_timestamp, 0);
3143 av_log(mov->fc, AV_LOG_WARNING,
3144 "st: %d edit list: %"PRId64" Missing key frame while searching for timestamp: %"PRId64"\n",
3145 st->index, edit_list_index, search_timestamp);
3146 index = find_prev_closest_index(st, e_old, nb_old, search_timestamp, AVSEEK_FLAG_ANY);
3149 av_log(mov->fc, AV_LOG_WARNING,
3150 "st: %d edit list %"PRId64" Cannot find an index entry before timestamp: %"PRId64".\n"
3151 "Rounding edit list media time to zero.\n",
3152 st->index, edit_list_index, search_timestamp);
3154 edit_list_media_time = 0;
3157 current = e_old + index;
3160 ctts_sample_old = 0;
3162 // set ctts_index properly for the found key frame
3163 for (index_ctts_count = 0; index_ctts_count < index; index_ctts_count++) {
3164 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3166 if (ctts_data_old[ctts_index_old].count == ctts_sample_old) {
3168 ctts_sample_old = 0;
3173 edit_list_start_ctts_sample = ctts_sample_old;
3175 // Iterate over index and arrange it according to edit list
3176 edit_list_start_encountered = 0;
3177 for (; current < e_old_end; current++, index++) {
3178 // check if frame outside edit list mark it for discard
3179 frame_duration = (current + 1 < e_old_end) ?
3180 ((current + 1)->timestamp - current->timestamp) : edit_list_duration;
3182 flags = current->flags;
3184 // frames (pts) before or after edit list
3185 curr_cts = current->timestamp + msc->dts_shift;
3188 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3189 curr_ctts = ctts_data_old[ctts_index_old].duration;
3190 av_log(mov->fc, AV_LOG_DEBUG, "stts: %"PRId64" ctts: %"PRId64", ctts_index: %"PRId64", ctts_count: %"PRId64"\n",
3191 curr_cts, curr_ctts, ctts_index_old, ctts_count_old);
3192 curr_cts += curr_ctts;
3194 if (ctts_sample_old == ctts_data_old[ctts_index_old].count) {
3195 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3196 &ctts_allocated_size,
3197 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3198 ctts_data_old[ctts_index_old].duration) == -1) {
3199 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3201 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3202 ctts_data_old[ctts_index_old].duration);
3206 ctts_sample_old = 0;
3207 edit_list_start_ctts_sample = 0;
3211 if (curr_cts < edit_list_media_time || curr_cts >= (edit_list_duration + edit_list_media_time)) {
3212 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id != AV_CODEC_ID_VORBIS &&
3213 curr_cts < edit_list_media_time && curr_cts + frame_duration > edit_list_media_time &&
3214 first_non_zero_audio_edit > 0) {
3215 packet_skip_samples = edit_list_media_time - curr_cts;
3216 st->skip_samples += packet_skip_samples;
3218 // Shift the index entry timestamp by packet_skip_samples to be correct.
3219 edit_list_dts_counter -= packet_skip_samples;
3220 if (edit_list_start_encountered == 0) {
3221 edit_list_start_encountered = 1;
3222 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3223 // discarded packets.
3224 if (frame_duration_buffer) {
3225 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3226 frame_duration_buffer, num_discarded_begin);
3227 av_freep(&frame_duration_buffer);
3231 av_log(mov->fc, AV_LOG_DEBUG, "skip %d audio samples from curr_cts: %"PRId64"\n", packet_skip_samples, curr_cts);
3233 flags |= AVINDEX_DISCARD_FRAME;
3234 av_log(mov->fc, AV_LOG_DEBUG, "drop a frame at curr_cts: %"PRId64" @ %"PRId64"\n", curr_cts, index);
3236 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && edit_list_start_encountered == 0) {
3237 num_discarded_begin++;
3238 frame_duration_buffer = av_realloc(frame_duration_buffer,
3239 num_discarded_begin * sizeof(int64_t));
3240 if (!frame_duration_buffer) {
3241 av_log(mov->fc, AV_LOG_ERROR, "Cannot reallocate frame duration buffer\n");
3244 frame_duration_buffer[num_discarded_begin - 1] = frame_duration;
3246 // Increment skip_samples for the first non-zero audio edit list
3247 if (first_non_zero_audio_edit > 0 && st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
3248 st->skip_samples += frame_duration;
3249 msc->start_pad = st->skip_samples;
3254 if (min_corrected_pts < 0) {
3255 min_corrected_pts = edit_list_dts_counter + curr_ctts + msc->dts_shift;
3257 min_corrected_pts = FFMIN(min_corrected_pts, edit_list_dts_counter + curr_ctts + msc->dts_shift);
3259 if (edit_list_start_encountered == 0) {
3260 edit_list_start_encountered = 1;
3261 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3262 // discarded packets.
3263 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && frame_duration_buffer) {
3264 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3265 frame_duration_buffer, num_discarded_begin);
3266 av_freep(&frame_duration_buffer);
3271 if (add_index_entry(st, current->pos, edit_list_dts_counter, current->size,
3272 current->min_distance, flags) == -1) {
3273 av_log(mov->fc, AV_LOG_ERROR, "Cannot add index entry\n");
3277 // Update the index ranges array
3278 if (current_index_range < msc->index_ranges || index != current_index_range->end) {
3279 current_index_range++;
3280 current_index_range->start = index;
3282 current_index_range->end = index + 1;
3284 // Only start incrementing DTS in frame_duration amounts, when we encounter a frame in edit list.
3285 if (edit_list_start_encountered > 0) {
3286 edit_list_dts_counter = edit_list_dts_counter + frame_duration;
3289 // Break when found first key frame after edit entry completion
3290 if (((curr_cts + frame_duration) >= (edit_list_duration + edit_list_media_time)) &&
3291 ((flags & AVINDEX_KEYFRAME) || ((st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)))) {
3293 if (ctts_data_old && ctts_sample_old != 0) {
3294 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3295 &ctts_allocated_size,
3296 ctts_sample_old - edit_list_start_ctts_sample,
3297 ctts_data_old[ctts_index_old].duration) == -1) {
3298 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3299 ctts_index_old, ctts_sample_old - edit_list_start_ctts_sample,
3300 ctts_data_old[ctts_index_old].duration);
3308 // If there are empty edits, then min_corrected_pts might be positive intentionally. So we subtract the
3309 // sum duration of emtpy edits here.
3310 min_corrected_pts -= empty_edits_sum_duration;
3312 // If the minimum pts turns out to be greater than zero after fixing the index, then we subtract the
3313 // dts by that amount to make the first pts zero.
3314 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && min_corrected_pts > 0) {
3315 av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by %"PRId64" to make first pts zero.\n", min_corrected_pts);
3316 for (i = 0; i < st->nb_index_entries; ++i) {
3317 st->index_entries[i].timestamp -= min_corrected_pts;
3321 // Update av stream length
3322 st->duration = edit_list_dts_entry_end - start_dts;
3324 // Free the old index and the old CTTS structures
3326 av_free(ctts_data_old);
3328 // Null terminate the index ranges array
3329 current_index_range++;
3330 current_index_range->start = 0;
3331 current_index_range->end = 0;
3332 msc->current_index = msc->index_ranges[0].start;
3335 static void mov_build_index(MOVContext *mov, AVStream *st)
3337 MOVStreamContext *sc = st->priv_data;
3338 int64_t current_offset;
3339 int64_t current_dts = 0;
3340 unsigned int stts_index = 0;
3341 unsigned int stsc_index = 0;
3342 unsigned int stss_index = 0;
3343 unsigned int stps_index = 0;
3345 uint64_t stream_size = 0;
3347 if (sc->elst_count) {
3348 int i, edit_start_index = 0, multiple_edits = 0;
3349 int64_t empty_duration = 0; // empty duration of the first edit list entry
3350 int64_t start_time = 0; // start time of the media
3352 for (i = 0; i < sc->elst_count; i++) {
3353 const MOVElst *e = &sc->elst_data[i];
3354 if (i == 0 && e->time == -1) {
3355 /* if empty, the first entry is the start time of the stream
3356 * relative to the presentation itself */
3357 empty_duration = e->duration;
3358 edit_start_index = 1;
3359 } else if (i == edit_start_index && e->time >= 0) {
3360 start_time = e->time;
3366 if (multiple_edits && !mov->advanced_editlist)
3367 av_log(mov->fc, AV_LOG_WARNING, "multiple edit list entries, "
3368 "Use -advanced_editlist to correctly decode otherwise "
3369 "a/v desync might occur\n");
3371 /* adjust first dts according to edit list */
3372 if ((empty_duration || start_time) && mov->time_scale > 0) {
3374 empty_duration = av_rescale(empty_duration, sc->time_scale, mov->time_scale);
3375 sc->time_offset = start_time - empty_duration;
3376 if (!mov->advanced_editlist)
3377 current_dts = -sc->time_offset;
3380 if (!multiple_edits && !mov->advanced_editlist &&
3381 st->codecpar->codec_id == AV_CODEC_ID_AAC && start_time > 0)
3382 sc->start_pad = start_time;
3385 /* only use old uncompressed audio chunk demuxing when stts specifies it */
3386 if (!(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3387 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
3388 unsigned int current_sample = 0;
3389 unsigned int stts_sample = 0;
3390 unsigned int sample_size;
3391 unsigned int distance = 0;
3392 unsigned int rap_group_index = 0;
3393 unsigned int rap_group_sample = 0;
3394 int64_t last_dts = 0;
3395 int64_t dts_correction = 0;
3396 int rap_group_present = sc->rap_group_count && sc->rap_group;
3397 int key_off = (sc->keyframe_count && sc->keyframes[0] > 0) || (sc->stps_count && sc->stps_data[0] > 0);
3399 current_dts -= sc->dts_shift;
3400 last_dts = current_dts;
3402 if (!sc->sample_count || st->nb_index_entries)
3404 if (sc->sample_count >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3406 if (av_reallocp_array(&st->index_entries,
3407 st->nb_index_entries + sc->sample_count,
3408 sizeof(*st->index_entries)) < 0) {
3409 st->nb_index_entries = 0;
3412 st->index_entries_allocated_size = (st->nb_index_entries + sc->sample_count) * sizeof(*st->index_entries);
3414 for (i = 0; i < sc->chunk_count; i++) {
3415 int64_t next_offset = i+1 < sc->chunk_count ? sc->chunk_offsets[i+1] : INT64_MAX;
3416 current_offset = sc->chunk_offsets[i];
3417 while (mov_stsc_index_valid(stsc_index, sc->stsc_count) &&
3418 i + 1 == sc->stsc_data[stsc_index + 1].first)
3421 if (next_offset > current_offset && sc->sample_size>0 && sc->sample_size < sc->stsz_sample_size &&
3422 sc->stsc_data[stsc_index].count * (int64_t)sc->stsz_sample_size > next_offset - current_offset) {
3423 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too large), ignoring\n", sc->stsz_sample_size);
3424 sc->stsz_sample_size = sc->sample_size;
3426 if (sc->stsz_sample_size>0 && sc->stsz_sample_size < sc->sample_size) {
3427 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too small), ignoring\n", sc->stsz_sample_size);
3428 sc->stsz_sample_size = sc->sample_size;
3431 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
3433 if (current_sample >= sc->sample_count) {
3434 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
3438 if (!sc->keyframe_absent && (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index])) {
3440 if (stss_index + 1 < sc->keyframe_count)
3442 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
3444 if (stps_index + 1 < sc->stps_count)
3447 if (rap_group_present && rap_group_index < sc->rap_group_count) {
3448 if (sc->rap_group[rap_group_index].index > 0)
3450 if (++rap_group_sample == sc->rap_group[rap_group_index].count) {
3451 rap_group_sample = 0;
3455 if (sc->keyframe_absent
3457 && !rap_group_present
3458 && (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO || (i==0 && j==0)))
3462 sample_size = sc->stsz_sample_size > 0 ? sc->stsz_sample_size : sc->sample_sizes[current_sample];
3463 if (sc->pseudo_stream_id == -1 ||
3464 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
3466 if (sample_size > 0x3FFFFFFF) {
3467 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", sample_size);
3470 e = &st->index_entries[st->nb_index_entries++];
3471 e->pos = current_offset;
3472 e->timestamp = current_dts;
3473 e->size = sample_size;
3474 e->min_distance = distance;
3475 e->flags = keyframe ? AVINDEX_KEYFRAME : 0;
3476 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %u, offset %"PRIx64", dts %"PRId64", "
3477 "size %u, distance %u, keyframe %d\n", st->index, current_sample,
3478 current_offset, current_dts, sample_size, distance, keyframe);
3479 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries < 100)
3480 ff_rfps_add_frame(mov->fc, st, current_dts);
3483 current_offset += sample_size;
3484 stream_size += sample_size;
3486 /* A negative sample duration is invalid based on the spec,
3487 * but some samples need it to correct the DTS. */
3488 if (sc->stts_data[stts_index].duration < 0) {
3489 av_log(mov->fc, AV_LOG_WARNING,
3490 "Invalid SampleDelta %d in STTS, at %d st:%d\n",
3491 sc->stts_data[stts_index].duration, stts_index,
3493 dts_correction += sc->stts_data[stts_index].duration - 1;
3494 sc->stts_data[stts_index].duration = 1;
3496 current_dts += sc->stts_data[stts_index].duration;
3497 if (!dts_correction || current_dts + dts_correction > last_dts) {
3498 current_dts += dts_correction;
3501 /* Avoid creating non-monotonous DTS */
3502 dts_correction += current_dts - last_dts - 1;
3503 current_dts = last_dts + 1;
3505 last_dts = current_dts;
3509 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
3515 if (st->duration > 0)
3516 st->codecpar->bit_rate = stream_size*8*sc->time_scale/st->duration;
3518 unsigned chunk_samples, total = 0;
3520 // compute total chunk count
3521 for (i = 0; i < sc->stsc_count; i++) {
3522 unsigned count, chunk_count;
3524 chunk_samples = sc->stsc_data[i].count;
3525 if (i != sc->stsc_count - 1 &&
3526 sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
3527 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
3531 if (sc->samples_per_frame >= 160) { // gsm
3532 count = chunk_samples / sc->samples_per_frame;
3533 } else if (sc->samples_per_frame > 1) {
3534 unsigned samples = (1024/sc->samples_per_frame)*sc->samples_per_frame;
3535 count = (chunk_samples+samples-1) / samples;
3537 count = (chunk_samples+1023) / 1024;
3540 if (mov_stsc_index_valid(i, sc->stsc_count))
3541 chunk_count = sc->stsc_data[i+1].first - sc->stsc_data[i].first;
3543 chunk_count = sc->chunk_count - (sc->stsc_data[i].first - 1);
3544 total += chunk_count * count;
3547 av_log(mov->fc, AV_LOG_TRACE, "chunk count %u\n", total);
3548 if (total >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3550 if (av_reallocp_array(&st->index_entries,
3551 st->nb_index_entries + total,
3552 sizeof(*st->index_entries)) < 0) {
3553 st->nb_index_entries = 0;
3556 st->index_entries_allocated_size = (st->nb_index_entries + total) * sizeof(*st->index_entries);
3559 for (i = 0; i < sc->chunk_count; i++) {
3560 current_offset = sc->chunk_offsets[i];
3561 if (mov_stsc_index_valid(stsc_index, sc->stsc_count) &&
3562 i + 1 == sc->stsc_data[stsc_index + 1].first)
3564 chunk_samples = sc->stsc_data[stsc_index].count;
3566 while (chunk_samples > 0) {
3568 unsigned size, samples;
3570 if (sc->samples_per_frame > 1 && !sc->bytes_per_frame) {
3571 avpriv_request_sample(mov->fc,
3572 "Zero bytes per frame, but %d samples per frame",
3573 sc->samples_per_frame);
3577 if (sc->samples_per_frame >= 160) { // gsm
3578 samples = sc->samples_per_frame;
3579 size = sc->bytes_per_frame;
3581 if (sc->samples_per_frame > 1) {
3582 samples = FFMIN((1024 / sc->samples_per_frame)*
3583 sc->samples_per_frame, chunk_samples);
3584 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
3586 samples = FFMIN(1024, chunk_samples);
3587 size = samples * sc->sample_size;
3591 if (st->nb_index_entries >= total) {
3592 av_log(mov->fc, AV_LOG_ERROR, "wrong chunk count %u\n", total);
3595 if (size > 0x3FFFFFFF) {
3596 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", size);
3599 e = &st->index_entries[st->nb_index_entries++];
3600 e->pos = current_offset;
3601 e->timestamp = current_dts;
3603 e->min_distance = 0;
3604 e->flags = AVINDEX_KEYFRAME;
3605 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, chunk %u, offset %"PRIx64", dts %"PRId64", "
3606 "size %u, duration %u\n", st->index, i, current_offset, current_dts,
3609 current_offset += size;
3610 current_dts += samples;
3611 chunk_samples -= samples;
3616 if (!mov->ignore_editlist && mov->advanced_editlist) {
3617 // Fix index according to edit lists.
3618 mov_fix_index(mov, st);
3622 static int test_same_origin(const char *src, const char *ref) {
3632 av_url_split(src_proto, sizeof(src_proto), src_auth, sizeof(src_auth), src_host, sizeof(src_host), &src_port, NULL, 0, src);
3633 av_url_split(ref_proto, sizeof(ref_proto), ref_auth, sizeof(ref_auth), ref_host, sizeof(ref_host), &ref_port, NULL, 0, ref);
3635 if (strlen(src) == 0) {
3637 } else if (strlen(src_auth) + 1 >= sizeof(src_auth) ||
3638 strlen(ref_auth) + 1 >= sizeof(ref_auth) ||
3639 strlen(src_host) + 1 >= sizeof(src_host) ||
3640 strlen(ref_host) + 1 >= sizeof(ref_host)) {
3642 } else if (strcmp(src_proto, ref_proto) ||
3643 strcmp(src_auth, ref_auth) ||
3644 strcmp(src_host, ref_host) ||
3645 src_port != ref_port) {
3651 static int mov_open_dref(MOVContext *c, AVIOContext **pb, const char *src, MOVDref *ref)
3653 /* try relative path, we do not try the absolute because it can leak information about our
3654 system to an attacker */
3655 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
3656 char filename[1025];
3657 const char *src_path;
3660 /* find a source dir */
3661 src_path = strrchr(src, '/');
3667 /* find a next level down to target */
3668 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
3669 if (ref->path[l] == '/') {
3670 if (i == ref->nlvl_to - 1)
3676 /* compose filename if next level down to target was found */
3677 if (i == ref->nlvl_to - 1 && src_path - src < sizeof(filename)) {
3678 memcpy(filename, src, src_path - src);
3679 filename[src_path - src] = 0;
3681 for (i = 1; i < ref->nlvl_from; i++)
3682 av_strlcat(filename, "../", sizeof(filename));
3684 av_strlcat(filename, ref->path + l + 1, sizeof(filename));
3685 if (!c->use_absolute_path) {
3686 int same_origin = test_same_origin(src, filename);
3689 av_log(c->fc, AV_LOG_ERROR,
3690 "Reference with mismatching origin, %s not tried for security reasons, "
3691 "set demuxer option use_absolute_path to allow it anyway\n",
3693 return AVERROR(ENOENT);
3696 if(strstr(ref->path + l + 1, "..") ||
3697 strstr(ref->path + l + 1, ":") ||
3698 (ref->nlvl_from > 1 && same_origin < 0) ||
3699 (filename[0] == '/' && src_path == src))
3700 return AVERROR(ENOENT);
3703 if (strlen(filename) + 1 == sizeof(filename))
3704 return AVERROR(ENOENT);
3705 if (!c->fc->io_open(c->fc, pb, filename, AVIO_FLAG_READ, NULL))
3708 } else if (c->use_absolute_path) {
3709 av_log(c->fc, AV_LOG_WARNING, "Using absolute path on user request, "
3710 "this is a possible security issue\n");
3711 if (!c->fc->io_open(c->fc, pb, ref->path, AVIO_FLAG_READ, NULL))
3714 av_log(c->fc, AV_LOG_ERROR,
3715 "Absolute path %s not tried for security reasons, "
3716 "set demuxer option use_absolute_path to allow absolute paths\n",
3720 return AVERROR(ENOENT);
3723 static void fix_timescale(MOVContext *c, MOVStreamContext *sc)
3725 if (sc->time_scale <= 0) {
3726 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", sc->ffindex);
3727 sc->time_scale = c->time_scale;
3728 if (sc->time_scale <= 0)
3733 static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3736 MOVStreamContext *sc;
3739 st = avformat_new_stream(c->fc, NULL);
3740 if (!st) return AVERROR(ENOMEM);
3741 st->id = c->fc->nb_streams;
3742 sc = av_mallocz(sizeof(MOVStreamContext));
3743 if (!sc) return AVERROR(ENOMEM);
3746 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
3747 sc->ffindex = st->index;
3748 c->trak_index = st->index;
3750 if ((ret = mov_read_default(c, pb, atom)) < 0)
3756 if (sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
3757 (!sc->sample_size && !sc->sample_count))) {
3758 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
3763 fix_timescale(c, sc);
3765 avpriv_set_pts_info(st, 64, 1, sc->time_scale);
3767 mov_build_index(c, st);
3769 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
3770 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
3771 if (c->enable_drefs) {
3772 if (mov_open_dref(c, &sc->pb, c->fc->filename, dref) < 0)
3773 av_log(c->fc, AV_LOG_ERROR,
3774 "stream %d, error opening alias: path='%s', dir='%s', "
3775 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
3776 st->index, dref->path, dref->dir, dref->filename,
3777 dref->volume, dref->nlvl_from, dref->nlvl_to);
3779 av_log(c->fc, AV_LOG_WARNING,
3780 "Skipped opening external track: "
3781 "stream %d, alias: path='%s', dir='%s', "
3782 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d."
3783 "Set enable_drefs to allow this.\n",
3784 st->index, dref->path, dref->dir, dref->filename,
3785 dref->volume, dref->nlvl_from, dref->nlvl_to);
3789 sc->pb_is_copied = 1;
3792 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
3793 if (!st->sample_aspect_ratio.num && st->codecpar->width && st->codecpar->height &&
3794 sc->height && sc->width &&
3795 (st->codecpar->width != sc->width || st->codecpar->height != sc->height)) {
3796 st->sample_aspect_ratio = av_d2q(((double)st->codecpar->height * sc->width) /
3797 ((double)st->codecpar->width * sc->height), INT_MAX);
3800 #if FF_API_R_FRAME_RATE
3801 if (sc->stts_count == 1 || (sc->stts_count == 2 && sc->stts_data[1].count == 1))
3802 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
3803 sc->time_scale, sc->stts_data[0].duration, INT_MAX);
3807 // done for ai5q, ai52, ai55, ai1q, ai12 and ai15.
3808 if (!st->codecpar->extradata_size && st->codecpar->codec_id == AV_CODEC_ID_H264 &&
3809 TAG_IS_AVCI(st->codecpar->codec_tag)) {
3810 ret = ff_generate_avci_extradata(st);
3815 switch (st->codecpar->codec_id) {
3816 #if CONFIG_H261_DECODER
3817 case AV_CODEC_ID_H261:
3819 #if CONFIG_H263_DECODER
3820 case AV_CODEC_ID_H263:
3822 #if CONFIG_MPEG4_DECODER
3823 case AV_CODEC_ID_MPEG4:
3825 st->codecpar->width = 0; /* let decoder init width/height */
3826 st->codecpar->height= 0;
3830 // If the duration of the mp3 packets is not constant, then they could need a parser
3831 if (st->codecpar->codec_id == AV_CODEC_ID_MP3
3832 && sc->stts_count > 3
3833 && sc->stts_count*10 > st->nb_frames
3834 && sc->time_scale == st->codecpar->sample_rate) {
3835 st->need_parsing = AVSTREAM_PARSE_FULL;
3837 /* Do not need those anymore. */
3838 av_freep(&sc->chunk_offsets);
3839 av_freep(&sc->sample_sizes);
3840 av_freep(&sc->keyframes);
3841 av_freep(&sc->stts_data);
3842 av_freep(&sc->stps_data);
3843 av_freep(&sc->elst_data);
3844 av_freep(&sc->rap_group);
3849 static int mov_read_ilst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3852 c->itunes_metadata = 1;
3853 ret = mov_read_default(c, pb, atom);
3854 c->itunes_metadata = 0;
3858 static int mov_read_keys(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3867 count = avio_rb32(pb);
3868 if (count > UINT_MAX / sizeof(*c->meta_keys) - 1) {
3869 av_log(c->fc, AV_LOG_ERROR,
3870 "The 'keys' atom with the invalid key count: %"PRIu32"\n", count);
3871 return AVERROR_INVALIDDATA;
3874 c->meta_keys_count = count + 1;
3875 c->meta_keys = av_mallocz(c->meta_keys_count * sizeof(*c->meta_keys));
3877 return AVERROR(ENOMEM);
3879 for (i = 1; i <= count; ++i) {
3880 uint32_t key_size = avio_rb32(pb);
3881 uint32_t type = avio_rl32(pb);
3883 av_log(c->fc, AV_LOG_ERROR,
3884 "The key# %"PRIu32" in meta has invalid size:"
3885 "%"PRIu32"\n", i, key_size);
3886 return AVERROR_INVALIDDATA;
3889 if (type != MKTAG('m','d','t','a')) {
3890 avio_skip(pb, key_size);
3892 c->meta_keys[i] = av_mallocz(key_size + 1);
3893 if (!c->meta_keys[i])
3894 return AVERROR(ENOMEM);
3895 avio_read(pb, c->meta_keys[i], key_size);
3901 static int mov_read_custom(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3903 int64_t end = avio_tell(pb) + atom.size;
3904 uint8_t *key = NULL, *val = NULL, *mean = NULL;
3908 MOVStreamContext *sc;
3910 if (c->fc->nb_streams < 1)
3912 st = c->fc->streams[c->fc->nb_streams-1];
3915 for (i = 0; i < 3; i++) {
3919 if (end - avio_tell(pb) <= 12)
3922 len = avio_rb32(pb);
3923 tag = avio_rl32(pb);
3924 avio_skip(pb, 4); // flags
3926 if (len < 12 || len - 12 > end - avio_tell(pb))
3930 if (tag == MKTAG('m', 'e', 'a', 'n'))
3932 else if (tag == MKTAG('n', 'a', 'm', 'e'))
3934 else if (tag == MKTAG('d', 'a', 't', 'a') && len > 4) {
3941 *p = av_malloc(len + 1);
3944 ret = ffio_read_size(pb, *p, len);
3952 if (mean && key && val) {
3953 if (strcmp(key, "iTunSMPB") == 0) {
3954 int priming, remainder, samples;
3955 if(sscanf(val, "%*X %X %X %X", &priming, &remainder, &samples) == 3){
3956 if(priming>0 && priming<16384)
3957 sc->start_pad = priming;
3960 if (strcmp(key, "cdec") != 0) {
3961 av_dict_set(&c->fc->metadata, key, val,
3962 AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL);
3966 av_log(c->fc, AV_LOG_VERBOSE,
3967 "Unhandled or malformed custom metadata of size %"PRId64"\n", atom.size);
3970 avio_seek(pb, end, SEEK_SET);
3977 static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3979 while (atom.size > 8) {
3980 uint32_t tag = avio_rl32(pb);
3982 if (tag == MKTAG('h','d','l','r')) {
3983 avio_seek(pb, -8, SEEK_CUR);
3985 return mov_read_default(c, pb, atom);
3991 // return 1 when matrix is identity, 0 otherwise
3992 #define IS_MATRIX_IDENT(matrix) \
3993 ( (matrix)[0][0] == (1 << 16) && \
3994 (matrix)[1][1] == (1 << 16) && \
3995 (matrix)[2][2] == (1 << 30) && \
3996 !(matrix)[0][1] && !(matrix)[0][2] && \
3997 !(matrix)[1][0] && !(matrix)[1][2] && \
3998 !(matrix)[2][0] && !(matrix)[2][1])
4000 static int mov_read_tkhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4005 int display_matrix[3][3];
4006 int res_display_matrix[3][3] = { { 0 } };
4008 MOVStreamContext *sc;
4012 if (c->fc->nb_streams < 1)
4014 st = c->fc->streams[c->fc->nb_streams-1];
4017 version = avio_r8(pb);
4018 flags = avio_rb24(pb);
4019 st->disposition |= (flags & MOV_TKHD_FLAG_ENABLED) ? AV_DISPOSITION_DEFAULT : 0;
4025 avio_rb32(pb); /* creation time */
4026 avio_rb32(pb); /* modification time */
4028 st->id = (int)avio_rb32(pb); /* track id (NOT 0 !)*/
4029 avio_rb32(pb); /* reserved */
4031 /* highlevel (considering edits) duration in movie timebase */
4032 (version == 1) ? avio_rb64(pb) : avio_rb32(pb);
4033 avio_rb32(pb); /* reserved */
4034 avio_rb32(pb); /* reserved */
4036 avio_rb16(pb); /* layer */
4037 avio_rb16(pb); /* alternate group */
4038 avio_rb16(pb); /* volume */
4039 avio_rb16(pb); /* reserved */
4041 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
4042 // they're kept in fixed point format through all calculations
4043 // save u,v,z to store the whole matrix in the AV_PKT_DATA_DISPLAYMATRIX
4044 // side data, but the scale factor is not needed to calculate aspect ratio
4045 for (i = 0; i < 3; i++) {
4046 display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
4047 display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
4048 display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
4051 width = avio_rb32(pb); // 16.16 fixed point track width
4052 height = avio_rb32(pb); // 16.16 fixed point track height
4053 sc->width = width >> 16;
4054 sc->height = height >> 16;
4056 // apply the moov display matrix (after the tkhd one)
4057 for (i = 0; i < 3; i++) {
4058 const int sh[3] = { 16, 16, 30 };
4059 for (j = 0; j < 3; j++) {
4060 for (e = 0; e < 3; e++) {
4061 res_display_matrix[i][j] +=
4062 ((int64_t) display_matrix[i][e] *
4063 c->movie_display_matrix[e][j]) >> sh[e];
4068 // save the matrix when it is not the default identity
4069 if (!IS_MATRIX_IDENT(res_display_matrix)) {
4072 av_freep(&sc->display_matrix);
4073 sc->display_matrix = av_malloc(sizeof(int32_t) * 9);
4074 if (!sc->display_matrix)
4075 return AVERROR(ENOMEM);
4077 for (i = 0; i < 3; i++)
4078 for (j = 0; j < 3; j++)
4079 sc->display_matrix[i * 3 + j] = res_display_matrix[i][j];
4081 #if FF_API_OLD_ROTATE_API
4082 rotate = av_display_rotation_get(sc->display_matrix);
4083 if (!isnan(rotate)) {
4084 char rotate_buf[64];
4086 if (rotate < 0) // for backward compatibility
4088 snprintf(rotate_buf, sizeof(rotate_buf), "%g", rotate);
4089 av_dict_set(&st->metadata, "rotate", rotate_buf, 0);
4094 // transform the display width/height according to the matrix
4095 // to keep the same scale, use [width height 1<<16]
4096 if (width && height && sc->display_matrix) {
4097 double disp_transform[2];
4099 for (i = 0; i < 2; i++)
4100 disp_transform[i] = hypot(sc->display_matrix[0 + i],
4101 sc->display_matrix[3 + i]);
4103 if (disp_transform[0] > 0 && disp_transform[1] > 0 &&
4104 disp_transform[0] < (1<<24) && disp_transform[1] < (1<<24) &&
4105 fabs((disp_transform[0] / disp_transform[1]) - 1.0) > 0.01)
4106 st->sample_aspect_ratio = av_d2q(
4107 disp_transform[0] / disp_transform[1],
4113 static int mov_read_tfhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4115 MOVFragment *frag = &c->fragment;
4116 MOVTrackExt *trex = NULL;
4117 MOVFragmentIndex* index = NULL;
4118 int flags, track_id, i, found = 0;
4120 avio_r8(pb); /* version */
4121 flags = avio_rb24(pb);
4123 track_id = avio_rb32(pb);
4125 return AVERROR_INVALIDDATA;
4126 frag->track_id = track_id;
4127 for (i = 0; i < c->trex_count; i++)
4128 if (c->trex_data[i].track_id == frag->track_id) {
4129 trex = &c->trex_data[i];
4133 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
4134 return AVERROR_INVALIDDATA;
4137 frag->base_data_offset = flags & MOV_TFHD_BASE_DATA_OFFSET ?
4138 avio_rb64(pb) : flags & MOV_TFHD_DEFAULT_BASE_IS_MOOF ?
4139 frag->moof_offset : frag->implicit_offset;
4140 frag->stsd_id = flags & MOV_TFHD_STSD_ID ? avio_rb32(pb) : trex->stsd_id;
4142 frag->duration = flags & MOV_TFHD_DEFAULT_DURATION ?
4143 avio_rb32(pb) : trex->duration;
4144 frag->size = flags & MOV_TFHD_DEFAULT_SIZE ?
4145 avio_rb32(pb) : trex->size;
4146 frag->flags = flags & MOV_TFHD_DEFAULT_FLAGS ?
4147 avio_rb32(pb) : trex->flags;
4148 frag->time = AV_NOPTS_VALUE;
4149 for (i = 0; i < c->fragment_index_count; i++) {
4151 MOVFragmentIndex* candidate = c->fragment_index_data[i];
4152 if (candidate->track_id == frag->track_id) {
4153 av_log(c->fc, AV_LOG_DEBUG,
4154 "found fragment index for track %u\n", frag->track_id);
4156 for (j = index->current_item; j < index->item_count; j++) {
4157 if (frag->implicit_offset == index->items[j].moof_offset) {
4158 av_log(c->fc, AV_LOG_DEBUG, "found fragment index entry "
4159 "for track %u and moof_offset %"PRId64"\n",
4160 frag->track_id, index->items[j].moof_offset);
4161 frag->time = index->items[j].time;
4162 index->current_item = j + 1;
4171 if (index && !found) {
4172 av_log(c->fc, AV_LOG_DEBUG, "track %u has a fragment index but "
4173 "it doesn't have an (in-order) entry for moof_offset "
4174 "%"PRId64"\n", frag->track_id, frag->implicit_offset);
4176 av_log(c->fc, AV_LOG_TRACE, "frag flags 0x%x\n", frag->flags);
4180 static int mov_read_chap(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4185 num = atom.size / 4;
4186 if (!(new_tracks = av_malloc_array(num, sizeof(int))))
4187 return AVERROR(ENOMEM);
4189 av_free(c->chapter_tracks);
4190 c->chapter_tracks = new_tracks;
4191 c->nb_chapter_tracks = num;
4193 for (i = 0; i < num && !pb->eof_reached; i++)
4194 c->chapter_tracks[i] = avio_rb32(pb);
4199 static int mov_read_trex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4204 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
4205 return AVERROR_INVALIDDATA;
4206 if ((err = av_reallocp_array(&c->trex_data, c->trex_count + 1,
4207 sizeof(*c->trex_data))) < 0) {
4212 c->fc->duration = AV_NOPTS_VALUE; // the duration from mvhd is not representing the whole file when fragments are used.
4214 trex = &c->trex_data[c->trex_count++];
4215 avio_r8(pb); /* version */
4216 avio_rb24(pb); /* flags */
4217 trex->track_id = avio_rb32(pb);
4218 trex->stsd_id = avio_rb32(pb);
4219 trex->duration = avio_rb32(pb);
4220 trex->size = avio_rb32(pb);
4221 trex->flags = avio_rb32(pb);
4225 static int mov_read_tfdt(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4227 MOVFragment *frag = &c->fragment;
4228 AVStream *st = NULL;
4229 MOVStreamContext *sc;
4232 for (i = 0; i < c->fc->nb_streams; i++) {
4233 if (c->fc->streams[i]->id == frag->track_id) {
4234 st = c->fc->streams[i];
4239 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %u\n", frag->track_id);
4240 return AVERROR_INVALIDDATA;
4243 if (sc->pseudo_stream_id + 1 != frag->stsd_id)
4245 version = avio_r8(pb);
4246 avio_rb24(pb); /* flags */
4248 sc->track_end = avio_rb64(pb);
4250 sc->track_end = avio_rb32(pb);
4255 static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4257 MOVFragment *frag = &c->fragment;
4258 AVStream *st = NULL;
4259 MOVStreamContext *sc;
4263 int data_offset = 0;
4264 unsigned entries, first_sample_flags = frag->flags;
4265 int flags, distance, i, err, old_nb_index_entries;
4267 for (i = 0; i < c->fc->nb_streams; i++) {
4268 if (c->fc->streams[i]->id == frag->track_id) {
4269 st = c->fc->streams[i];
4274 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %u\n", frag->track_id);
4275 return AVERROR_INVALIDDATA;
4278 if (sc->pseudo_stream_id+1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4280 avio_r8(pb); /* version */
4281 flags = avio_rb24(pb);
4282 entries = avio_rb32(pb);
4283 av_log(c->fc, AV_LOG_TRACE, "flags 0x%x entries %u\n", flags, entries);
4285 /* Always assume the presence of composition time offsets.
4286 * Without this assumption, for instance, we cannot deal with a track in fragmented movies that meet the following.
4287 * 1) in the initial movie, there are no samples.
4288 * 2) in the first movie fragment, there is only one sample without composition time offset.
4289 * 3) in the subsequent movie fragments, there are samples with composition time offset. */
4290 if (!sc->ctts_count && sc->sample_count)
4292 /* Complement ctts table if moov atom doesn't have ctts atom. */
4293 ctts_data = av_realloc(NULL, sizeof(*sc->ctts_data));
4295 return AVERROR(ENOMEM);
4296 sc->ctts_data = ctts_data;
4297 sc->ctts_data[sc->ctts_count].count = sc->sample_count;
4298 sc->ctts_data[sc->ctts_count].duration = 0;
4301 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
4302 return AVERROR_INVALIDDATA;
4303 if ((err = av_reallocp_array(&sc->ctts_data, entries + sc->ctts_count,
4304 sizeof(*sc->ctts_data))) < 0) {
4308 if (flags & MOV_TRUN_DATA_OFFSET) data_offset = avio_rb32(pb);
4309 if (flags & MOV_TRUN_FIRST_SAMPLE_FLAGS) first_sample_flags = avio_rb32(pb);
4310 dts = sc->track_end - sc->time_offset;
4311 offset = frag->base_data_offset + data_offset;
4313 av_log(c->fc, AV_LOG_TRACE, "first sample flags 0x%x\n", first_sample_flags);
4314 for (i = 0; i < entries && !pb->eof_reached; i++) {
4315 unsigned sample_size = frag->size;
4316 int sample_flags = i ? frag->flags : first_sample_flags;
4317 unsigned sample_duration = frag->duration;
4320 if (flags & MOV_TRUN_SAMPLE_DURATION) sample_duration = avio_rb32(pb);
4321 if (flags & MOV_TRUN_SAMPLE_SIZE) sample_size = avio_rb32(pb);
4322 if (flags & MOV_TRUN_SAMPLE_FLAGS) sample_flags = avio_rb32(pb);
4323 sc->ctts_data[sc->ctts_count].count = 1;
4324 sc->ctts_data[sc->ctts_count].duration = (flags & MOV_TRUN_SAMPLE_CTS) ?
4326 mov_update_dts_shift(sc, sc->ctts_data[sc->ctts_count].duration);
4327 if (frag->time != AV_NOPTS_VALUE) {
4328 if (c->use_mfra_for == FF_MOV_FLAG_MFRA_PTS) {
4329 int64_t pts = frag->time;
4330 av_log(c->fc, AV_LOG_DEBUG, "found frag time %"PRId64
4331 " sc->dts_shift %d ctts.duration %d"
4332 " sc->time_offset %"PRId64" flags & MOV_TRUN_SAMPLE_CTS %d\n", pts,
4333 sc->dts_shift, sc->ctts_data[sc->ctts_count].duration,
4334 sc->time_offset, flags & MOV_TRUN_SAMPLE_CTS);
4335 dts = pts - sc->dts_shift;
4336 if (flags & MOV_TRUN_SAMPLE_CTS) {
4337 dts -= sc->ctts_data[sc->ctts_count].duration;
4339 dts -= sc->time_offset;
4341 av_log(c->fc, AV_LOG_DEBUG, "calculated into dts %"PRId64"\n", dts);
4343 dts = frag->time - sc->time_offset;
4344 av_log(c->fc, AV_LOG_DEBUG, "found frag time %"PRId64
4345 ", using it for dts\n", dts);
4347 frag->time = AV_NOPTS_VALUE;
4350 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
4354 !(sample_flags & (MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC |
4355 MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES));
4358 old_nb_index_entries = st->nb_index_entries;
4359 err = av_add_index_entry(st, offset, dts, sample_size, distance,
4360 keyframe ? AVINDEX_KEYFRAME : 0);
4362 av_log(c->fc, AV_LOG_ERROR, "Failed to add index entry\n");
4363 } else if (err <= sc->current_sample && err + 1 != st->nb_index_entries &&
4364 st->nb_index_entries != old_nb_index_entries) {
4365 // if we inserted a new item before the current sample, move the
4366 // counter ahead so it is still pointing to the same sample.
4367 sc->current_sample++;
4369 av_log(c->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
4370 "size %u, distance %d, keyframe %d\n", st->index, err,
4371 offset, dts, sample_size, distance, keyframe);
4373 dts += sample_duration;
4374 offset += sample_size;
4375 sc->data_size += sample_size;
4376 sc->duration_for_fps += sample_duration;
4377 sc->nb_frames_for_fps ++;
4380 if (pb->eof_reached)
4383 frag->implicit_offset = offset;
4385 sc->track_end = dts + sc->time_offset;
4386 if (st->duration < sc->track_end)
4387 st->duration = sc->track_end;
4392 static int mov_read_sidx(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4394 int64_t offset = avio_tell(pb) + atom.size, pts;
4396 unsigned i, track_id;
4397 AVStream *st = NULL;
4398 AVStream *ref_st = NULL;
4399 MOVStreamContext *sc, *ref_sc = NULL;
4400 MOVFragmentIndex *index = NULL;
4401 MOVFragmentIndex **tmp;
4402 AVRational timescale;
4404 version = avio_r8(pb);
4406 avpriv_request_sample(c->fc, "sidx version %u", version);
4410 avio_rb24(pb); // flags
4412 track_id = avio_rb32(pb); // Reference ID
4413 for (i = 0; i < c->fc->nb_streams; i++) {
4414 if (c->fc->streams[i]->id == track_id) {
4415 st = c->fc->streams[i];
4420 av_log(c->fc, AV_LOG_WARNING, "could not find corresponding track id %d\n", track_id);
4426 timescale = av_make_q(1, avio_rb32(pb));
4428 if (timescale.den <= 0) {
4429 av_log(c->fc, AV_LOG_ERROR, "Invalid sidx timescale 1/%d\n", timescale.den);
4430 return AVERROR_INVALIDDATA;
4434 pts = avio_rb32(pb);
4435 offset += avio_rb32(pb);
4437 pts = avio_rb64(pb);
4438 offset += avio_rb64(pb);
4441 avio_rb16(pb); // reserved
4443 index = av_mallocz(sizeof(MOVFragmentIndex));
4445 return AVERROR(ENOMEM);
4447 index->track_id = track_id;
4449 index->item_count = avio_rb16(pb);
4450 index->items = av_mallocz_array(index->item_count, sizeof(MOVFragmentIndexItem));
4452 if (!index->items) {
4454 return AVERROR(ENOMEM);
4457 for (i = 0; i < index->item_count; i++) {
4458 uint32_t size = avio_rb32(pb);
4459 uint32_t duration = avio_rb32(pb);
4460 if (size & 0x80000000) {
4461 avpriv_request_sample(c->fc, "sidx reference_type 1");
4462 av_freep(&index->items);
4464 return AVERROR_PATCHWELCOME;
4466 avio_rb32(pb); // sap_flags
4467 index->items[i].moof_offset = offset;
4468 index->items[i].time = av_rescale_q(pts, st->time_base, timescale);
4473 st->duration = sc->track_end = pts;
4475 tmp = av_realloc_array(c->fragment_index_data,
4476 c->fragment_index_count + 1,
4477 sizeof(MOVFragmentIndex*));
4479 av_freep(&index->items);
4481 return AVERROR(ENOMEM);
4484 c->fragment_index_data = tmp;
4485 c->fragment_index_data[c->fragment_index_count++] = index;
4488 if (offset == avio_size(pb)) {
4489 for (i = 0; i < c->fc->nb_streams; i++) {
4490 if (c->fc->streams[i]->id == c->fragment_index_data[0]->track_id) {
4491 ref_st = c->fc->streams[i];
4492 ref_sc = ref_st->priv_data;
4496 for (i = 0; i < c->fc->nb_streams; i++) {
4497 st = c->fc->streams[i];
4499 if (!sc->has_sidx) {
4500 st->duration = sc->track_end = av_rescale(ref_st->duration, sc->time_scale, ref_sc->time_scale);
4504 c->fragment_index_complete = 1;
4510 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
4511 /* like the files created with Adobe Premiere 5.0, for samples see */
4512 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
4513 static int mov_read_wide(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4518 return 0; /* continue */
4519 if (avio_rb32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
4520 avio_skip(pb, atom.size - 4);
4523 atom.type = avio_rl32(pb);
4525 if (atom.type != MKTAG('m','d','a','t')) {
4526 avio_skip(pb, atom.size);
4529 err = mov_read_mdat(c, pb, atom);
4533 static int mov_read_cmov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4538 uint8_t *moov_data; /* uncompressed data */
4539 long cmov_len, moov_len;
4542 avio_rb32(pb); /* dcom atom */
4543 if (avio_rl32(pb) != MKTAG('d','c','o','m'))
4544 return AVERROR_INVALIDDATA;
4545 if (avio_rl32(pb) != MKTAG('z','l','i','b')) {
4546 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !\n");
4547 return AVERROR_INVALIDDATA;
4549 avio_rb32(pb); /* cmvd atom */
4550 if (avio_rl32(pb) != MKTAG('c','m','v','d'))
4551 return AVERROR_INVALIDDATA;
4552 moov_len = avio_rb32(pb); /* uncompressed size */
4553 cmov_len = atom.size - 6 * 4;
4555 cmov_data = av_malloc(cmov_len);
4557 return AVERROR(ENOMEM);
4558 moov_data = av_malloc(moov_len);
4561 return AVERROR(ENOMEM);
4563 ret = ffio_read_size(pb, cmov_data, cmov_len);
4565 goto free_and_return;
4567 if (uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
4568 goto free_and_return;
4569 if (ffio_init_context(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
4570 goto free_and_return;
4571 ctx.seekable = AVIO_SEEKABLE_NORMAL;
4572 atom.type = MKTAG('m','o','o','v');
4573 atom.size = moov_len;
4574 ret = mov_read_default(c, &ctx, atom);
4580 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
4581 return AVERROR(ENOSYS);
4585 /* edit list atom */
4586 static int mov_read_elst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4588 MOVStreamContext *sc;
4589 int i, edit_count, version;
4591 if (c->fc->nb_streams < 1 || c->ignore_editlist)
4593 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
4595 version = avio_r8(pb); /* version */
4596 avio_rb24(pb); /* flags */
4597 edit_count = avio_rb32(pb); /* entries */
4602 av_log(c->fc, AV_LOG_WARNING, "Duplicated ELST atom\n");
4603 av_free(sc->elst_data);
4605 sc->elst_data = av_malloc_array(edit_count, sizeof(*sc->elst_data));
4607 return AVERROR(ENOMEM);
4609 av_log(c->fc, AV_LOG_TRACE, "track[%u].edit_count = %i\n", c->fc->nb_streams - 1, edit_count);
4610 for (i = 0; i < edit_count && !pb->eof_reached; i++) {
4611 MOVElst *e = &sc->elst_data[i];
4614 e->duration = avio_rb64(pb);
4615 e->time = avio_rb64(pb);
4617 e->duration = avio_rb32(pb); /* segment duration */
4618 e->time = (int32_t)avio_rb32(pb); /* media time */
4620 e->rate = avio_rb32(pb) / 65536.0;
4621 av_log(c->fc, AV_LOG_TRACE, "duration=%"PRId64" time=%"PRId64" rate=%f\n",
4622 e->duration, e->time, e->rate);
4624 if (e->time < 0 && e->time != -1 &&
4625 c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
4626 av_log(c->fc, AV_LOG_ERROR, "Track %d, edit %d: Invalid edit list media time=%"PRId64"\n",
4627 c->fc->nb_streams-1, i, e->time);
4628 return AVERROR_INVALIDDATA;
4636 static int mov_read_tmcd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4638 MOVStreamContext *sc;
4640 if (c->fc->nb_streams < 1)
4641 return AVERROR_INVALIDDATA;
4642 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
4643 sc->timecode_track = avio_rb32(pb);
4647 static int mov_read_vpcc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4650 int version, color_range, color_primaries, color_trc, color_space;
4652 if (c->fc->nb_streams < 1)
4654 st = c->fc->streams[c->fc->nb_streams - 1];
4656 if (atom.size < 5) {
4657 av_log(c->fc, AV_LOG_ERROR, "Empty VP Codec Configuration box\n");
4658 return AVERROR_INVALIDDATA;
4661 version = avio_r8(pb);
4663 av_log(c->fc, AV_LOG_WARNING, "Unsupported VP Codec Configuration box version %d\n", version);
4666 avio_skip(pb, 3); /* flags */
4668 avio_skip(pb, 2); /* profile + level */
4669 color_range = avio_r8(pb); /* bitDepth, chromaSubsampling, videoFullRangeFlag */
4670 color_primaries = avio_r8(pb);
4671 color_trc = avio_r8(pb);
4672 color_space = avio_r8(pb);
4673 if (avio_rb16(pb)) /* codecIntializationDataSize */
4674 return AVERROR_INVALIDDATA;
4676 if (!av_color_primaries_name(color_primaries))
4677 color_primaries = AVCOL_PRI_UNSPECIFIED;
4678 if (!av_color_transfer_name(color_trc))
4679 color_trc = AVCOL_TRC_UNSPECIFIED;
4680 if (!av_color_space_name(color_space))
4681 color_space = AVCOL_SPC_UNSPECIFIED;
4683 st->codecpar->color_range = (color_range & 1) ? AVCOL_RANGE_JPEG : AVCOL_RANGE_MPEG;
4684 st->codecpar->color_primaries = color_primaries;
4685 st->codecpar->color_trc = color_trc;
4686 st->codecpar->color_space = color_space;
4691 static int mov_read_smdm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4693 MOVStreamContext *sc;
4694 const int chroma_den = 50000;
4695 const int luma_den = 10000;
4698 if (c->fc->nb_streams < 1)
4699 return AVERROR_INVALIDDATA;
4701 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
4703 if (atom.size < 5) {
4704 av_log(c->fc, AV_LOG_ERROR, "Empty Mastering Display Metadata box\n");
4705 return AVERROR_INVALIDDATA;
4708 version = avio_r8(pb);
4710 av_log(c->fc, AV_LOG_WARNING, "Unsupported Mastering Display Metadata box version %d\n", version);
4713 avio_skip(pb, 3); /* flags */
4715 sc->mastering = av_mastering_display_metadata_alloc();
4717 return AVERROR(ENOMEM);
4719 for (i = 0; i < 3; i++)
4720 for (j = 0; j < 2; j++)
4721 sc->mastering->display_primaries[i][j] =
4722 av_make_q(lrint(((double)avio_rb16(pb) / (1 << 16)) * chroma_den), chroma_den);
4723 for (i = 0; i < 2; i++)
4724 sc->mastering->white_point[i] =
4725 av_make_q(lrint(((double)avio_rb16(pb) / (1 << 16)) * chroma_den), chroma_den);
4726 sc->mastering->max_luminance =
4727 av_make_q(lrint(((double)avio_rb32(pb) / (1 << 8)) * luma_den), luma_den);
4728 sc->mastering->min_luminance =
4729 av_make_q(lrint(((double)avio_rb32(pb) / (1 << 14)) * luma_den), luma_den);
4731 sc->mastering->has_primaries = 1;
4732 sc->mastering->has_luminance = 1;
4737 static int mov_read_coll(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4739 MOVStreamContext *sc;
4742 if (c->fc->nb_streams < 1)
4743 return AVERROR_INVALIDDATA;
4745 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
4747 if (atom.size < 5) {
4748 av_log(c->fc, AV_LOG_ERROR, "Empty Content Light Level box\n");
4749 return AVERROR_INVALIDDATA;
4752 version = avio_r8(pb);
4754 av_log(c->fc, AV_LOG_WARNING, "Unsupported Content Light Level box version %d\n", version);
4757 avio_skip(pb, 3); /* flags */
4759 sc->coll = av_content_light_metadata_alloc(&sc->coll_size);
4761 return AVERROR(ENOMEM);
4763 sc->coll->MaxCLL = avio_rb16(pb);
4764 sc->coll->MaxFALL = avio_rb16(pb);
4769 static int mov_read_st3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4772 MOVStreamContext *sc;
4773 enum AVStereo3DType type;
4776 if (c->fc->nb_streams < 1)
4779 st = c->fc->streams[c->fc->nb_streams - 1];
4782 if (atom.size < 5) {
4783 av_log(c->fc, AV_LOG_ERROR, "Empty stereoscopic video box\n");
4784 return AVERROR_INVALIDDATA;
4786 avio_skip(pb, 4); /* version + flags */
4791 type = AV_STEREO3D_2D;
4794 type = AV_STEREO3D_TOPBOTTOM;
4797 type = AV_STEREO3D_SIDEBYSIDE;
4800 av_log(c->fc, AV_LOG_WARNING, "Unknown st3d mode value %d\n", mode);
4804 sc->stereo3d = av_stereo3d_alloc();
4806 return AVERROR(ENOMEM);
4808 sc->stereo3d->type = type;
4812 static int mov_read_sv3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4815 MOVStreamContext *sc;
4817 int32_t yaw, pitch, roll;
4818 uint32_t l = 0, t = 0, r = 0, b = 0;
4819 uint32_t tag, padding = 0;
4820 enum AVSphericalProjection projection;
4822 if (c->fc->nb_streams < 1)
4825 st = c->fc->streams[c->fc->nb_streams - 1];
4828 if (atom.size < 8) {
4829 av_log(c->fc, AV_LOG_ERROR, "Empty spherical video box\n");
4830 return AVERROR_INVALIDDATA;
4833 size = avio_rb32(pb);
4834 if (size <= 12 || size > atom.size)
4835 return AVERROR_INVALIDDATA;
4837 tag = avio_rl32(pb);
4838 if (tag != MKTAG('s','v','h','d')) {
4839 av_log(c->fc, AV_LOG_ERROR, "Missing spherical video header\n");
4842 avio_skip(pb, 4); /* version + flags */
4843 avio_skip(pb, size - 12); /* metadata_source */
4845 size = avio_rb32(pb);
4846 if (size > atom.size)
4847 return AVERROR_INVALIDDATA;
4849 tag = avio_rl32(pb);
4850 if (tag != MKTAG('p','r','o','j')) {
4851 av_log(c->fc, AV_LOG_ERROR, "Missing projection box\n");
4855 size = avio_rb32(pb);
4856 if (size > atom.size)
4857 return AVERROR_INVALIDDATA;
4859 tag = avio_rl32(pb);
4860 if (tag != MKTAG('p','r','h','d')) {
4861 av_log(c->fc, AV_LOG_ERROR, "Missing projection header box\n");
4864 avio_skip(pb, 4); /* version + flags */
4866 /* 16.16 fixed point */
4867 yaw = avio_rb32(pb);
4868 pitch = avio_rb32(pb);
4869 roll = avio_rb32(pb);
4871 size = avio_rb32(pb);
4872 if (size > atom.size)
4873 return AVERROR_INVALIDDATA;
4875 tag = avio_rl32(pb);
4876 avio_skip(pb, 4); /* version + flags */
4878 case MKTAG('c','b','m','p'):
4879 layout = avio_rb32(pb);
4881 av_log(c->fc, AV_LOG_WARNING,
4882 "Unsupported cubemap layout %d\n", layout);
4885 projection = AV_SPHERICAL_CUBEMAP;
4886 padding = avio_rb32(pb);
4888 case MKTAG('e','q','u','i'):
4894 if (b >= UINT_MAX - t || r >= UINT_MAX - l) {
4895 av_log(c->fc, AV_LOG_ERROR,
4896 "Invalid bounding rectangle coordinates "
4897 "%"PRIu32",%"PRIu32",%"PRIu32",%"PRIu32"\n", l, t, r, b);
4898 return AVERROR_INVALIDDATA;
4901 if (l || t || r || b)
4902 projection = AV_SPHERICAL_EQUIRECTANGULAR_TILE;
4904 projection = AV_SPHERICAL_EQUIRECTANGULAR;
4907 av_log(c->fc, AV_LOG_ERROR, "Unknown projection type\n");
4911 sc->spherical = av_spherical_alloc(&sc->spherical_size);
4913 return AVERROR(ENOMEM);
4915 sc->spherical->projection = projection;
4917 sc->spherical->yaw = yaw;
4918 sc->spherical->pitch = pitch;
4919 sc->spherical->roll = roll;
4921 sc->spherical->padding = padding;
4923 sc->spherical->bound_left = l;
4924 sc->spherical->bound_top = t;
4925 sc->spherical->bound_right = r;
4926 sc->spherical->bound_bottom = b;
4931 static int mov_parse_uuid_spherical(MOVStreamContext *sc, AVIOContext *pb, size_t len)
4934 uint8_t *buffer = av_malloc(len + 1);
4938 return AVERROR(ENOMEM);
4941 ret = ffio_read_size(pb, buffer, len);
4945 /* Check for mandatory keys and values, try to support XML as best-effort */
4946 if (av_stristr(buffer, "<GSpherical:StitchingSoftware>") &&
4947 (val = av_stristr(buffer, "<GSpherical:Spherical>")) &&
4948 av_stristr(val, "true") &&
4949 (val = av_stristr(buffer, "<GSpherical:Stitched>")) &&
4950 av_stristr(val, "true") &&
4951 (val = av_stristr(buffer, "<GSpherical:ProjectionType>")) &&
4952 av_stristr(val, "equirectangular")) {
4953 sc->spherical = av_spherical_alloc(&sc->spherical_size);
4957 sc->spherical->projection = AV_SPHERICAL_EQUIRECTANGULAR;
4959 if (av_stristr(buffer, "<GSpherical:StereoMode>")) {
4960 enum AVStereo3DType mode;
4962 if (av_stristr(buffer, "left-right"))
4963 mode = AV_STEREO3D_SIDEBYSIDE;
4964 else if (av_stristr(buffer, "top-bottom"))
4965 mode = AV_STEREO3D_TOPBOTTOM;
4967 mode = AV_STEREO3D_2D;
4969 sc->stereo3d = av_stereo3d_alloc();
4973 sc->stereo3d->type = mode;
4977 val = av_stristr(buffer, "<GSpherical:InitialViewHeadingDegrees>");
4979 sc->spherical->yaw = strtol(val, NULL, 10) * (1 << 16);
4980 val = av_stristr(buffer, "<GSpherical:InitialViewPitchDegrees>");
4982 sc->spherical->pitch = strtol(val, NULL, 10) * (1 << 16);
4983 val = av_stristr(buffer, "<GSpherical:InitialViewRollDegrees>");
4985 sc->spherical->roll = strtol(val, NULL, 10) * (1 << 16);
4993 static int mov_read_uuid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4996 MOVStreamContext *sc;
4999 static const uint8_t uuid_isml_manifest[] = {
5000 0xa5, 0xd4, 0x0b, 0x30, 0xe8, 0x14, 0x11, 0xdd,
5001 0xba, 0x2f, 0x08, 0x00, 0x20, 0x0c, 0x9a, 0x66
5003 static const uint8_t uuid_xmp[] = {
5004 0xbe, 0x7a, 0xcf, 0xcb, 0x97, 0xa9, 0x42, 0xe8,
5005 0x9c, 0x71, 0x99, 0x94, 0x91, 0xe3, 0xaf, 0xac
5007 static const uint8_t uuid_spherical[] = {
5008 0xff, 0xcc, 0x82, 0x63, 0xf8, 0x55, 0x4a, 0x93,
5009 0x88, 0x14, 0x58, 0x7a, 0x02, 0x52, 0x1f, 0xdd,
5012 if (atom.size < sizeof(uuid) || atom.size >= FFMIN(INT_MAX, SIZE_MAX))
5013 return AVERROR_INVALIDDATA;
5015 if (c->fc->nb_streams < 1)
5017 st = c->fc->streams[c->fc->nb_streams - 1];
5020 ret = avio_read(pb, uuid, sizeof(uuid));
5023 } else if (ret != sizeof(uuid)) {
5024 return AVERROR_INVALIDDATA;
5026 if (!memcmp(uuid, uuid_isml_manifest, sizeof(uuid))) {
5027 uint8_t *buffer, *ptr;
5029 size_t len = atom.size - sizeof(uuid);
5032 return AVERROR_INVALIDDATA;
5034 ret = avio_skip(pb, 4); // zeroes
5037 buffer = av_mallocz(len + 1);
5039 return AVERROR(ENOMEM);
5041 ret = avio_read(pb, buffer, len);
5045 } else if (ret != len) {
5047 return AVERROR_INVALIDDATA;
5051 while ((ptr = av_stristr(ptr, "systemBitrate=\""))) {
5052 ptr += sizeof("systemBitrate=\"") - 1;
5053 c->bitrates_count++;
5054 c->bitrates = av_realloc_f(c->bitrates, c->bitrates_count, sizeof(*c->bitrates));
5056 c->bitrates_count = 0;
5058 return AVERROR(ENOMEM);
5061 ret = strtol(ptr, &endptr, 10);
5062 if (ret < 0 || errno || *endptr != '"') {
5063 c->bitrates[c->bitrates_count - 1] = 0;
5065 c->bitrates[c->bitrates_count - 1] = ret;
5070 } else if (!memcmp(uuid, uuid_xmp, sizeof(uuid))) {
5072 size_t len = atom.size - sizeof(uuid);
5073 if (c->export_xmp) {
5074 buffer = av_mallocz(len + 1);
5076 return AVERROR(ENOMEM);
5078 ret = avio_read(pb, buffer, len);
5082 } else if (ret != len) {
5084 return AVERROR_INVALIDDATA;
5087 av_dict_set(&c->fc->metadata, "xmp", buffer, 0);
5090 // skip all uuid atom, which makes it fast for long uuid-xmp file
5091 ret = avio_skip(pb, len);
5095 } else if (!memcmp(uuid, uuid_spherical, sizeof(uuid))) {
5096 size_t len = atom.size - sizeof(uuid);
5097 ret = mov_parse_uuid_spherical(sc, pb, len);
5101 av_log(c->fc, AV_LOG_WARNING, "Invalid spherical metadata found\n"); }
5106 static int mov_read_free(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5109 uint8_t content[16];
5114 ret = avio_read(pb, content, FFMIN(sizeof(content), atom.size));
5120 && !memcmp(content, "Anevia\x1A\x1A", 8)
5121 && c->use_mfra_for == FF_MOV_FLAG_MFRA_AUTO) {
5122 c->use_mfra_for = FF_MOV_FLAG_MFRA_PTS;
5128 static int mov_read_frma(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5130 uint32_t format = avio_rl32(pb);
5131 MOVStreamContext *sc;
5135 if (c->fc->nb_streams < 1)
5137 st = c->fc->streams[c->fc->nb_streams - 1];
5142 case MKTAG('e','n','c','v'): // encrypted video
5143 case MKTAG('e','n','c','a'): // encrypted audio
5144 id = mov_codec_id(st, format);
5145 if (st->codecpar->codec_id != AV_CODEC_ID_NONE &&
5146 st->codecpar->codec_id != id) {
5147 av_log(c->fc, AV_LOG_WARNING,
5148 "ignoring 'frma' atom of '%.4s', stream has codec id %d\n",
5149 (char*)&format, st->codecpar->codec_id);
5153 st->codecpar->codec_id = id;
5154 sc->format = format;
5158 if (format != sc->format) {
5159 av_log(c->fc, AV_LOG_WARNING,
5160 "ignoring 'frma' atom of '%.4s', stream format is '%.4s'\n",
5161 (char*)&format, (char*)&sc->format);
5169 static int mov_read_senc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5172 MOVStreamContext *sc;
5173 size_t auxiliary_info_size;
5175 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
5178 st = c->fc->streams[c->fc->nb_streams - 1];
5181 if (sc->cenc.aes_ctr) {
5182 av_log(c->fc, AV_LOG_ERROR, "duplicate senc atom\n");
5183 return AVERROR_INVALIDDATA;
5186 avio_r8(pb); /* version */
5187 sc->cenc.use_subsamples = avio_rb24(pb) & 0x02; /* flags */
5189 avio_rb32(pb); /* entries */
5191 if (atom.size < 8 || atom.size > FFMIN(INT_MAX, SIZE_MAX)) {
5192 av_log(c->fc, AV_LOG_ERROR, "senc atom size %"PRId64" invalid\n", atom.size);
5193 return AVERROR_INVALIDDATA;
5196 /* save the auxiliary info as is */
5197 auxiliary_info_size = atom.size - 8;
5199 sc->cenc.auxiliary_info = av_malloc(auxiliary_info_size);
5200 if (!sc->cenc.auxiliary_info) {
5201 return AVERROR(ENOMEM);
5204 sc->cenc.auxiliary_info_end = sc->cenc.auxiliary_info + auxiliary_info_size;
5205 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info;
5206 sc->cenc.auxiliary_info_index = 0;
5208 if (avio_read(pb, sc->cenc.auxiliary_info, auxiliary_info_size) != auxiliary_info_size) {
5209 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info");
5210 return AVERROR_INVALIDDATA;
5213 /* initialize the cipher */
5214 sc->cenc.aes_ctr = av_aes_ctr_alloc();
5215 if (!sc->cenc.aes_ctr) {
5216 return AVERROR(ENOMEM);
5219 return av_aes_ctr_init(sc->cenc.aes_ctr, c->decryption_key);
5222 static int mov_read_saiz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5225 MOVStreamContext *sc;
5227 int atom_header_size;
5230 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
5233 st = c->fc->streams[c->fc->nb_streams - 1];
5236 if (sc->cenc.auxiliary_info_sizes || sc->cenc.auxiliary_info_default_size) {
5237 av_log(c->fc, AV_LOG_ERROR, "duplicate saiz atom\n");
5238 return AVERROR_INVALIDDATA;
5241 atom_header_size = 9;
5243 avio_r8(pb); /* version */
5244 flags = avio_rb24(pb);
5246 if ((flags & 0x01) != 0) {
5247 atom_header_size += 8;
5249 avio_rb32(pb); /* info type */
5250 avio_rb32(pb); /* info type param */
5253 sc->cenc.auxiliary_info_default_size = avio_r8(pb);
5254 avio_rb32(pb); /* entries */
5256 if (atom.size <= atom_header_size) {
5260 if (atom.size > FFMIN(INT_MAX, SIZE_MAX)) {
5261 av_log(c->fc, AV_LOG_ERROR, "saiz atom auxiliary_info_sizes size %"PRId64" invalid\n", atom.size);
5262 return AVERROR_INVALIDDATA;
5265 /* save the auxiliary info sizes as is */
5266 data_size = atom.size - atom_header_size;
5268 sc->cenc.auxiliary_info_sizes = av_malloc(data_size);
5269 if (!sc->cenc.auxiliary_info_sizes) {
5270 return AVERROR(ENOMEM);
5273 sc->cenc.auxiliary_info_sizes_count = data_size;
5275 if (avio_read(pb, sc->cenc.auxiliary_info_sizes, data_size) != data_size) {
5276 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info sizes");
5277 return AVERROR_INVALIDDATA;
5283 static int mov_read_dfla(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5286 int last, type, size, ret;
5289 if (c->fc->nb_streams < 1)
5291 st = c->fc->streams[c->fc->nb_streams-1];
5293 if ((uint64_t)atom.size > (1<<30) || atom.size < 42)
5294 return AVERROR_INVALIDDATA;
5296 /* Check FlacSpecificBox version. */
5297 if (avio_r8(pb) != 0)
5298 return AVERROR_INVALIDDATA;
5300 avio_rb24(pb); /* Flags */
5302 avio_read(pb, buf, sizeof(buf));
5303 flac_parse_block_header(buf, &last, &type, &size);
5305 if (type != FLAC_METADATA_TYPE_STREAMINFO || size != FLAC_STREAMINFO_SIZE) {
5306 av_log(c->fc, AV_LOG_ERROR, "STREAMINFO must be first FLACMetadataBlock\n");
5307 return AVERROR_INVALIDDATA;
5310 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
5315 av_log(c->fc, AV_LOG_WARNING, "non-STREAMINFO FLACMetadataBlock(s) ignored\n");
5320 static int mov_seek_auxiliary_info(MOVContext *c, MOVStreamContext *sc, int64_t index)
5322 size_t auxiliary_info_seek_offset = 0;
5325 if (sc->cenc.auxiliary_info_default_size) {
5326 auxiliary_info_seek_offset = (size_t)sc->cenc.auxiliary_info_default_size * index;
5327 } else if (sc->cenc.auxiliary_info_sizes) {
5328 if (index > sc->cenc.auxiliary_info_sizes_count) {
5329 av_log(c, AV_LOG_ERROR, "current sample %"PRId64" greater than the number of auxiliary info sample sizes %"SIZE_SPECIFIER"\n",
5330 index, sc->cenc.auxiliary_info_sizes_count);
5331 return AVERROR_INVALIDDATA;
5334 for (i = 0; i < index; i++) {
5335 auxiliary_info_seek_offset += sc->cenc.auxiliary_info_sizes[i];
5339 if (auxiliary_info_seek_offset > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info) {
5340 av_log(c, AV_LOG_ERROR, "auxiliary info offset %"SIZE_SPECIFIER" greater than auxiliary info size %"SIZE_SPECIFIER"\n",
5341 auxiliary_info_seek_offset, (size_t)(sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info));
5342 return AVERROR_INVALIDDATA;
5345 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info + auxiliary_info_seek_offset;
5346 sc->cenc.auxiliary_info_index = index;
5350 static int cenc_filter(MOVContext *c, MOVStreamContext *sc, int64_t index, uint8_t *input, int size)
5352 uint32_t encrypted_bytes;
5353 uint16_t subsample_count;
5354 uint16_t clear_bytes;
5355 uint8_t* input_end = input + size;
5358 if (index != sc->cenc.auxiliary_info_index) {
5359 ret = mov_seek_auxiliary_info(c, sc, index);
5366 if (AES_CTR_IV_SIZE > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5367 av_log(c->fc, AV_LOG_ERROR, "failed to read iv from the auxiliary info\n");
5368 return AVERROR_INVALIDDATA;
5371 av_aes_ctr_set_iv(sc->cenc.aes_ctr, sc->cenc.auxiliary_info_pos);
5372 sc->cenc.auxiliary_info_pos += AES_CTR_IV_SIZE;
5374 if (!sc->cenc.use_subsamples)
5376 /* decrypt the whole packet */
5377 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, size);
5381 /* read the subsample count */
5382 if (sizeof(uint16_t) > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5383 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample count from the auxiliary info\n");
5384 return AVERROR_INVALIDDATA;
5387 subsample_count = AV_RB16(sc->cenc.auxiliary_info_pos);
5388 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
5390 for (; subsample_count > 0; subsample_count--)
5392 if (6 > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5393 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample from the auxiliary info\n");
5394 return AVERROR_INVALIDDATA;
5397 /* read the number of clear / encrypted bytes */
5398 clear_bytes = AV_RB16(sc->cenc.auxiliary_info_pos);
5399 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
5400 encrypted_bytes = AV_RB32(sc->cenc.auxiliary_info_pos);
5401 sc->cenc.auxiliary_info_pos += sizeof(uint32_t);
5403 if ((uint64_t)clear_bytes + encrypted_bytes > input_end - input) {
5404 av_log(c->fc, AV_LOG_ERROR, "subsample size exceeds the packet size left\n");
5405 return AVERROR_INVALIDDATA;
5408 /* skip the clear bytes */
5409 input += clear_bytes;
5411 /* decrypt the encrypted bytes */
5412 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, encrypted_bytes);
5413 input += encrypted_bytes;
5416 if (input < input_end) {
5417 av_log(c->fc, AV_LOG_ERROR, "leftover packet bytes after subsample processing\n");
5418 return AVERROR_INVALIDDATA;
5421 sc->cenc.auxiliary_info_index++;
5425 static int mov_read_dops(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5427 const int OPUS_SEEK_PREROLL_MS = 80;
5432 if (c->fc->nb_streams < 1)
5434 st = c->fc->streams[c->fc->nb_streams-1];
5436 if ((uint64_t)atom.size > (1<<30) || atom.size < 11)
5437 return AVERROR_INVALIDDATA;
5439 /* Check OpusSpecificBox version. */
5440 if (avio_r8(pb) != 0) {
5441 av_log(c->fc, AV_LOG_ERROR, "unsupported OpusSpecificBox version\n");
5442 return AVERROR_INVALIDDATA;
5445 /* OpusSpecificBox size plus magic for Ogg OpusHead header. */
5446 size = atom.size + 8;
5448 if (ff_alloc_extradata(st->codecpar, size))
5449 return AVERROR(ENOMEM);
5451 AV_WL32(st->codecpar->extradata, MKTAG('O','p','u','s'));
5452 AV_WL32(st->codecpar->extradata + 4, MKTAG('H','e','a','d'));
5453 AV_WB8(st->codecpar->extradata + 8, 1); /* OpusHead version */
5454 avio_read(pb, st->codecpar->extradata + 9, size - 9);
5456 /* OpusSpecificBox is stored in big-endian, but OpusHead is
5457 little-endian; aside from the preceeding magic and version they're
5458 otherwise currently identical. Data after output gain at offset 16
5459 doesn't need to be bytewapped. */
5460 pre_skip = AV_RB16(st->codecpar->extradata + 10);
5461 AV_WL16(st->codecpar->extradata + 10, pre_skip);
5462 AV_WL32(st->codecpar->extradata + 12, AV_RB32(st->codecpar->extradata + 12));
5463 AV_WL16(st->codecpar->extradata + 16, AV_RB16(st->codecpar->extradata + 16));
5465 st->codecpar->initial_padding = pre_skip;
5466 st->codecpar->seek_preroll = av_rescale_q(OPUS_SEEK_PREROLL_MS,
5467 (AVRational){1, 1000},
5468 (AVRational){1, 48000});
5473 static const MOVParseTableEntry mov_default_parse_table[] = {
5474 { MKTAG('A','C','L','R'), mov_read_aclr },
5475 { MKTAG('A','P','R','G'), mov_read_avid },
5476 { MKTAG('A','A','L','P'), mov_read_avid },
5477 { MKTAG('A','R','E','S'), mov_read_ares },
5478 { MKTAG('a','v','s','s'), mov_read_avss },
5479 { MKTAG('c','h','p','l'), mov_read_chpl },
5480 { MKTAG('c','o','6','4'), mov_read_stco },
5481 { MKTAG('c','o','l','r'), mov_read_colr },
5482 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
5483 { MKTAG('d','i','n','f'), mov_read_default },
5484 { MKTAG('D','p','x','E'), mov_read_dpxe },
5485 { MKTAG('d','r','e','f'), mov_read_dref },
5486 { MKTAG('e','d','t','s'), mov_read_default },
5487 { MKTAG('e','l','s','t'), mov_read_elst },
5488 { MKTAG('e','n','d','a'), mov_read_enda },
5489 { MKTAG('f','i','e','l'), mov_read_fiel },
5490 { MKTAG('a','d','r','m'), mov_read_adrm },
5491 { MKTAG('f','t','y','p'), mov_read_ftyp },
5492 { MKTAG('g','l','b','l'), mov_read_glbl },
5493 { MKTAG('h','d','l','r'), mov_read_hdlr },
5494 { MKTAG('i','l','s','t'), mov_read_ilst },
5495 { MKTAG('j','p','2','h'), mov_read_jp2h },
5496 { MKTAG('m','d','a','t'), mov_read_mdat },
5497 { MKTAG('m','d','h','d'), mov_read_mdhd },
5498 { MKTAG('m','d','i','a'), mov_read_default },
5499 { MKTAG('m','e','t','a'), mov_read_meta },
5500 { MKTAG('m','i','n','f'), mov_read_default },
5501 { MKTAG('m','o','o','f'), mov_read_moof },
5502 { MKTAG('m','o','o','v'), mov_read_moov },
5503 { MKTAG('m','v','e','x'), mov_read_default },
5504 { MKTAG('m','v','h','d'), mov_read_mvhd },
5505 { MKTAG('S','M','I',' '), mov_read_svq3 },
5506 { MKTAG('a','l','a','c'), mov_read_alac }, /* alac specific atom */
5507 { MKTAG('a','v','c','C'), mov_read_glbl },
5508 { MKTAG('p','a','s','p'), mov_read_pasp },
5509 { MKTAG('s','i','d','x'), mov_read_sidx },
5510 { MKTAG('s','t','b','l'), mov_read_default },
5511 { MKTAG('s','t','c','o'), mov_read_stco },
5512 { MKTAG('s','t','p','s'), mov_read_stps },
5513 { MKTAG('s','t','r','f'), mov_read_strf },
5514 { MKTAG('s','t','s','c'), mov_read_stsc },
5515 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
5516 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
5517 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
5518 { MKTAG('s','t','t','s'), mov_read_stts },
5519 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
5520 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
5521 { MKTAG('t','f','d','t'), mov_read_tfdt },
5522 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
5523 { MKTAG('t','r','a','k'), mov_read_trak },
5524 { MKTAG('t','r','a','f'), mov_read_default },
5525 { MKTAG('t','r','e','f'), mov_read_default },
5526 { MKTAG('t','m','c','d'), mov_read_tmcd },
5527 { MKTAG('c','h','a','p'), mov_read_chap },
5528 { MKTAG('t','r','e','x'), mov_read_trex },
5529 { MKTAG('t','r','u','n'), mov_read_trun },
5530 { MKTAG('u','d','t','a'), mov_read_default },
5531 { MKTAG('w','a','v','e'), mov_read_wave },
5532 { MKTAG('e','s','d','s'), mov_read_esds },
5533 { MKTAG('d','a','c','3'), mov_read_dac3 }, /* AC-3 info */
5534 { MKTAG('d','e','c','3'), mov_read_dec3 }, /* EAC-3 info */
5535 { MKTAG('d','d','t','s'), mov_read_ddts }, /* DTS audio descriptor */
5536 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
5537 { MKTAG('w','f','e','x'), mov_read_wfex },
5538 { MKTAG('c','m','o','v'), mov_read_cmov },
5539 { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */
5540 { MKTAG('d','v','c','1'), mov_read_dvc1 },
5541 { MKTAG('s','b','g','p'), mov_read_sbgp },
5542 { MKTAG('h','v','c','C'), mov_read_glbl },
5543 { MKTAG('u','u','i','d'), mov_read_uuid },
5544 { MKTAG('C','i','n', 0x8e), mov_read_targa_y216 },
5545 { MKTAG('f','r','e','e'), mov_read_free },
5546 { MKTAG('-','-','-','-'), mov_read_custom },
5547 { MKTAG('s','i','n','f'), mov_read_default },
5548 { MKTAG('f','r','m','a'), mov_read_frma },
5549 { MKTAG('s','e','n','c'), mov_read_senc },
5550 { MKTAG('s','a','i','z'), mov_read_saiz },
5551 { MKTAG('d','f','L','a'), mov_read_dfla },
5552 { MKTAG('s','t','3','d'), mov_read_st3d }, /* stereoscopic 3D video box */
5553 { MKTAG('s','v','3','d'), mov_read_sv3d }, /* spherical video box */
5554 { MKTAG('d','O','p','s'), mov_read_dops },
5555 { MKTAG('S','m','D','m'), mov_read_smdm },
5556 { MKTAG('C','o','L','L'), mov_read_coll },
5557 { MKTAG('v','p','c','C'), mov_read_vpcc },
5561 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5563 int64_t total_size = 0;
5567 if (c->atom_depth > 10) {
5568 av_log(c->fc, AV_LOG_ERROR, "Atoms too deeply nested\n");
5569 return AVERROR_INVALIDDATA;
5574 atom.size = INT64_MAX;
5575 while (total_size + 8 <= atom.size && !avio_feof(pb)) {
5576 int (*parse)(MOVContext*, AVIOContext*, MOVAtom) = NULL;
5579 if (atom.size >= 8) {
5580 a.size = avio_rb32(pb);
5581 a.type = avio_rl32(pb);
5582 if (a.type == MKTAG('f','r','e','e') &&
5584 c->fc->strict_std_compliance < FF_COMPLIANCE_STRICT &&
5587 uint32_t *type = (uint32_t *)buf + 1;
5588 if (avio_read(pb, buf, 8) != 8)
5589 return AVERROR_INVALIDDATA;
5590 avio_seek(pb, -8, SEEK_CUR);
5591 if (*type == MKTAG('m','v','h','d') ||
5592 *type == MKTAG('c','m','o','v')) {
5593 av_log(c->fc, AV_LOG_ERROR, "Detected moov in a free atom.\n");
5594 a.type = MKTAG('m','o','o','v');
5597 if (atom.type != MKTAG('r','o','o','t') &&
5598 atom.type != MKTAG('m','o','o','v'))
5600 if (a.type == MKTAG('t','r','a','k') || a.type == MKTAG('m','d','a','t'))
5602 av_log(c->fc, AV_LOG_ERROR, "Broken file, trak/mdat not at top-level\n");
5609 if (a.size == 1 && total_size + 8 <= atom.size) { /* 64 bit extended size */
5610 a.size = avio_rb64(pb) - 8;
5614 av_log(c->fc, AV_LOG_TRACE, "type:'%s' parent:'%s' sz: %"PRId64" %"PRId64" %"PRId64"\n",
5615 av_fourcc2str(a.type), av_fourcc2str(atom.type), a.size, total_size, atom.size);
5617 a.size = atom.size - total_size + 8;
5622 a.size = FFMIN(a.size, atom.size - total_size);
5624 for (i = 0; mov_default_parse_table[i].type; i++)
5625 if (mov_default_parse_table[i].type == a.type) {
5626 parse = mov_default_parse_table[i].parse;
5630 // container is user data
5631 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
5632 atom.type == MKTAG('i','l','s','t')))
5633 parse = mov_read_udta_string;
5635 // Supports parsing the QuickTime Metadata Keys.
5636 // https://developer.apple.com/library/mac/documentation/QuickTime/QTFF/Metadata/Metadata.html
5637 if (!parse && c->found_hdlr_mdta &&
5638 atom.type == MKTAG('m','e','t','a') &&
5639 a.type == MKTAG('k','e','y','s')) {
5640 parse = mov_read_keys;
5643 if (!parse) { /* skip leaf atoms data */
5644 avio_skip(pb, a.size);
5646 int64_t start_pos = avio_tell(pb);
5648 int err = parse(c, pb, a);
5653 if (c->found_moov && c->found_mdat &&
5654 ((!(pb->seekable & AVIO_SEEKABLE_NORMAL) || c->fc->flags & AVFMT_FLAG_IGNIDX || c->fragment_index_complete) ||
5655 start_pos + a.size == avio_size(pb))) {
5656 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) || c->fc->flags & AVFMT_FLAG_IGNIDX || c->fragment_index_complete)
5657 c->next_root_atom = start_pos + a.size;
5661 left = a.size - avio_tell(pb) + start_pos;
5662 if (left > 0) /* skip garbage at atom end */
5663 avio_skip(pb, left);
5664 else if (left < 0) {
5665 av_log(c->fc, AV_LOG_WARNING,
5666 "overread end of atom '%.4s' by %"PRId64" bytes\n",
5667 (char*)&a.type, -left);
5668 avio_seek(pb, left, SEEK_CUR);
5672 total_size += a.size;
5675 if (total_size < atom.size && atom.size < 0x7ffff)
5676 avio_skip(pb, atom.size - total_size);
5682 static int mov_probe(AVProbeData *p)
5687 int moov_offset = -1;
5689 /* check file header */
5692 /* ignore invalid offset */
5693 if ((offset + 8) > (unsigned int)p->buf_size)
5695 tag = AV_RL32(p->buf + offset + 4);
5697 /* check for obvious tags */
5698 case MKTAG('m','o','o','v'):
5699 moov_offset = offset + 4;
5700 case MKTAG('m','d','a','t'):
5701 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
5702 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
5703 case MKTAG('f','t','y','p'):
5704 if (AV_RB32(p->buf+offset) < 8 &&
5705 (AV_RB32(p->buf+offset) != 1 ||
5706 offset + 12 > (unsigned int)p->buf_size ||
5707 AV_RB64(p->buf+offset + 8) == 0)) {
5708 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
5709 } else if (tag == MKTAG('f','t','y','p') &&
5710 ( AV_RL32(p->buf + offset + 8) == MKTAG('j','p','2',' ')
5711 || AV_RL32(p->buf + offset + 8) == MKTAG('j','p','x',' ')
5713 score = FFMAX(score, 5);
5715 score = AVPROBE_SCORE_MAX;
5717 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5719 /* those are more common words, so rate then a bit less */
5720 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
5721 case MKTAG('w','i','d','e'):
5722 case MKTAG('f','r','e','e'):
5723 case MKTAG('j','u','n','k'):
5724 case MKTAG('p','i','c','t'):
5725 score = FFMAX(score, AVPROBE_SCORE_MAX - 5);
5726 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5728 case MKTAG(0x82,0x82,0x7f,0x7d):
5729 case MKTAG('s','k','i','p'):
5730 case MKTAG('u','u','i','d'):
5731 case MKTAG('p','r','f','l'):
5732 /* if we only find those cause probedata is too small at least rate them */
5733 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
5734 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5737 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5740 if(score > AVPROBE_SCORE_MAX - 50 && moov_offset != -1) {
5741 /* moov atom in the header - we should make sure that this is not a
5742 * MOV-packed MPEG-PS */
5743 offset = moov_offset;
5745 while(offset < (p->buf_size - 16)){ /* Sufficient space */
5746 /* We found an actual hdlr atom */
5747 if(AV_RL32(p->buf + offset ) == MKTAG('h','d','l','r') &&
5748 AV_RL32(p->buf + offset + 8) == MKTAG('m','h','l','r') &&
5749 AV_RL32(p->buf + offset + 12) == MKTAG('M','P','E','G')){
5750 av_log(NULL, AV_LOG_WARNING, "Found media data tag MPEG indicating this is a MOV-packed MPEG-PS.\n");
5751 /* We found a media handler reference atom describing an
5752 * MPEG-PS-in-MOV, return a
5753 * low score to force expanding the probe window until
5754 * mpegps_probe finds what it needs */
5765 // must be done after parsing all trak because there's no order requirement
5766 static void mov_read_chapters(AVFormatContext *s)
5768 MOVContext *mov = s->priv_data;
5770 MOVStreamContext *sc;
5775 for (j = 0; j < mov->nb_chapter_tracks; j++) {
5776 chapter_track = mov->chapter_tracks[j];
5778 for (i = 0; i < s->nb_streams; i++)
5779 if (s->streams[i]->id == chapter_track) {
5784 av_log(s, AV_LOG_ERROR, "Referenced QT chapter track not found\n");
5789 cur_pos = avio_tell(sc->pb);
5791 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
5792 st->disposition |= AV_DISPOSITION_ATTACHED_PIC | AV_DISPOSITION_TIMED_THUMBNAILS;
5793 if (st->nb_index_entries) {
5794 // Retrieve the first frame, if possible
5796 AVIndexEntry *sample = &st->index_entries[0];
5797 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
5798 av_log(s, AV_LOG_ERROR, "Failed to retrieve first frame\n");
5802 if (av_get_packet(sc->pb, &pkt, sample->size) < 0)
5805 st->attached_pic = pkt;
5806 st->attached_pic.stream_index = st->index;
5807 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
5810 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
5811 st->codecpar->codec_id = AV_CODEC_ID_BIN_DATA;
5812 st->discard = AVDISCARD_ALL;
5813 for (i = 0; i < st->nb_index_entries; i++) {
5814 AVIndexEntry *sample = &st->index_entries[i];
5815 int64_t end = i+1 < st->nb_index_entries ? st->index_entries[i+1].timestamp : st->duration;
5820 if (end < sample->timestamp) {
5821 av_log(s, AV_LOG_WARNING, "ignoring stream duration which is shorter than chapters\n");
5822 end = AV_NOPTS_VALUE;
5825 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
5826 av_log(s, AV_LOG_ERROR, "Chapter %d not found in file\n", i);
5830 // the first two bytes are the length of the title
5831 len = avio_rb16(sc->pb);
5832 if (len > sample->size-2)
5834 title_len = 2*len + 1;
5835 if (!(title = av_mallocz(title_len)))
5838 // The samples could theoretically be in any encoding if there's an encd
5839 // atom following, but in practice are only utf-8 or utf-16, distinguished
5840 // instead by the presence of a BOM
5844 ch = avio_rb16(sc->pb);
5846 avio_get_str16be(sc->pb, len, title, title_len);
5847 else if (ch == 0xfffe)
5848 avio_get_str16le(sc->pb, len, title, title_len);
5851 if (len == 1 || len == 2)
5854 avio_get_str(sc->pb, INT_MAX, title + 2, len - 1);
5858 avpriv_new_chapter(s, i, st->time_base, sample->timestamp, end, title);
5863 avio_seek(sc->pb, cur_pos, SEEK_SET);
5867 static int parse_timecode_in_framenum_format(AVFormatContext *s, AVStream *st,
5868 uint32_t value, int flags)
5871 char buf[AV_TIMECODE_STR_SIZE];
5872 AVRational rate = st->avg_frame_rate;
5873 int ret = av_timecode_init(&tc, rate, flags, 0, s);
5876 av_dict_set(&st->metadata, "timecode",
5877 av_timecode_make_string(&tc, buf, value), 0);
5881 static int mov_read_rtmd_track(AVFormatContext *s, AVStream *st)
5883 MOVStreamContext *sc = st->priv_data;
5884 char buf[AV_TIMECODE_STR_SIZE];
5885 int64_t cur_pos = avio_tell(sc->pb);
5886 int hh, mm, ss, ff, drop;
5888 if (!st->nb_index_entries)
5891 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
5892 avio_skip(s->pb, 13);
5893 hh = avio_r8(s->pb);
5894 mm = avio_r8(s->pb);
5895 ss = avio_r8(s->pb);
5896 drop = avio_r8(s->pb);
5897 ff = avio_r8(s->pb);
5898 snprintf(buf, AV_TIMECODE_STR_SIZE, "%02d:%02d:%02d%c%02d",
5899 hh, mm, ss, drop ? ';' : ':', ff);
5900 av_dict_set(&st->metadata, "timecode", buf, 0);
5902 avio_seek(sc->pb, cur_pos, SEEK_SET);
5906 static int mov_read_timecode_track(AVFormatContext *s, AVStream *st)
5908 MOVStreamContext *sc = st->priv_data;
5910 int64_t cur_pos = avio_tell(sc->pb);
5913 if (!st->nb_index_entries)
5916 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
5917 value = avio_rb32(s->pb);
5919 if (sc->tmcd_flags & 0x0001) flags |= AV_TIMECODE_FLAG_DROPFRAME;
5920 if (sc->tmcd_flags & 0x0002) flags |= AV_TIMECODE_FLAG_24HOURSMAX;
5921 if (sc->tmcd_flags & 0x0004) flags |= AV_TIMECODE_FLAG_ALLOWNEGATIVE;
5923 /* Assume Counter flag is set to 1 in tmcd track (even though it is likely
5924 * not the case) and thus assume "frame number format" instead of QT one.
5925 * No sample with tmcd track can be found with a QT timecode at the moment,
5926 * despite what the tmcd track "suggests" (Counter flag set to 0 means QT
5928 parse_timecode_in_framenum_format(s, st, value, flags);
5930 avio_seek(sc->pb, cur_pos, SEEK_SET);
5934 static int mov_read_close(AVFormatContext *s)
5936 MOVContext *mov = s->priv_data;
5939 for (i = 0; i < s->nb_streams; i++) {
5940 AVStream *st = s->streams[i];
5941 MOVStreamContext *sc = st->priv_data;
5946 av_freep(&sc->ctts_data);
5947 for (j = 0; j < sc->drefs_count; j++) {
5948 av_freep(&sc->drefs[j].path);
5949 av_freep(&sc->drefs[j].dir);
5951 av_freep(&sc->drefs);
5953 sc->drefs_count = 0;
5955 if (!sc->pb_is_copied)
5956 ff_format_io_close(s, &sc->pb);
5959 av_freep(&sc->chunk_offsets);
5960 av_freep(&sc->stsc_data);
5961 av_freep(&sc->sample_sizes);
5962 av_freep(&sc->keyframes);
5963 av_freep(&sc->stts_data);
5964 av_freep(&sc->stps_data);
5965 av_freep(&sc->elst_data);
5966 av_freep(&sc->rap_group);
5967 av_freep(&sc->display_matrix);
5968 av_freep(&sc->index_ranges);
5971 for (j = 0; j < sc->stsd_count; j++)
5972 av_free(sc->extradata[j]);
5973 av_freep(&sc->extradata);
5974 av_freep(&sc->extradata_size);
5976 av_freep(&sc->cenc.auxiliary_info);
5977 av_freep(&sc->cenc.auxiliary_info_sizes);
5978 av_aes_ctr_free(sc->cenc.aes_ctr);
5980 av_freep(&sc->stereo3d);
5981 av_freep(&sc->spherical);
5982 av_freep(&sc->mastering);
5983 av_freep(&sc->coll);
5986 if (mov->dv_demux) {
5987 avformat_free_context(mov->dv_fctx);
5988 mov->dv_fctx = NULL;
5991 if (mov->meta_keys) {
5992 for (i = 1; i < mov->meta_keys_count; i++) {
5993 av_freep(&mov->meta_keys[i]);
5995 av_freep(&mov->meta_keys);
5998 av_freep(&mov->trex_data);
5999 av_freep(&mov->bitrates);
6001 for (i = 0; i < mov->fragment_index_count; i++) {
6002 MOVFragmentIndex* index = mov->fragment_index_data[i];
6003 av_freep(&index->items);
6004 av_freep(&mov->fragment_index_data[i]);
6006 av_freep(&mov->fragment_index_data);
6008 av_freep(&mov->aes_decrypt);
6009 av_freep(&mov->chapter_tracks);
6014 static int tmcd_is_referenced(AVFormatContext *s, int tmcd_id)
6018 for (i = 0; i < s->nb_streams; i++) {
6019 AVStream *st = s->streams[i];
6020 MOVStreamContext *sc = st->priv_data;
6022 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO &&
6023 sc->timecode_track == tmcd_id)
6029 /* look for a tmcd track not referenced by any video track, and export it globally */
6030 static void export_orphan_timecode(AVFormatContext *s)
6034 for (i = 0; i < s->nb_streams; i++) {
6035 AVStream *st = s->streams[i];
6037 if (st->codecpar->codec_tag == MKTAG('t','m','c','d') &&
6038 !tmcd_is_referenced(s, i + 1)) {
6039 AVDictionaryEntry *tcr = av_dict_get(st->metadata, "timecode", NULL, 0);
6041 av_dict_set(&s->metadata, "timecode", tcr->value, 0);
6048 static int read_tfra(MOVContext *mov, AVIOContext *f)
6050 MOVFragmentIndex* index = NULL;
6051 int version, fieldlength, i, j;
6052 int64_t pos = avio_tell(f);
6053 uint32_t size = avio_rb32(f);
6056 if (avio_rb32(f) != MKBETAG('t', 'f', 'r', 'a')) {
6059 av_log(mov->fc, AV_LOG_VERBOSE, "found tfra\n");
6060 index = av_mallocz(sizeof(MOVFragmentIndex));
6062 return AVERROR(ENOMEM);
6065 tmp = av_realloc_array(mov->fragment_index_data,
6066 mov->fragment_index_count + 1,
6067 sizeof(MOVFragmentIndex*));
6070 return AVERROR(ENOMEM);
6072 mov->fragment_index_data = tmp;
6073 mov->fragment_index_data[mov->fragment_index_count++] = index;
6075 version = avio_r8(f);
6077 index->track_id = avio_rb32(f);
6078 fieldlength = avio_rb32(f);
6079 index->item_count = avio_rb32(f);
6080 index->items = av_mallocz_array(
6081 index->item_count, sizeof(MOVFragmentIndexItem));
6082 if (!index->items) {
6083 index->item_count = 0;
6084 return AVERROR(ENOMEM);
6086 for (i = 0; i < index->item_count; i++) {
6087 int64_t time, offset;
6089 time = avio_rb64(f);
6090 offset = avio_rb64(f);
6092 time = avio_rb32(f);
6093 offset = avio_rb32(f);
6095 index->items[i].time = time;
6096 index->items[i].moof_offset = offset;
6097 for (j = 0; j < ((fieldlength >> 4) & 3) + 1; j++)
6099 for (j = 0; j < ((fieldlength >> 2) & 3) + 1; j++)
6101 for (j = 0; j < ((fieldlength >> 0) & 3) + 1; j++)
6105 avio_seek(f, pos + size, SEEK_SET);
6109 static int mov_read_mfra(MOVContext *c, AVIOContext *f)
6111 int64_t stream_size = avio_size(f);
6112 int64_t original_pos = avio_tell(f);
6116 if ((seek_ret = avio_seek(f, stream_size - 4, SEEK_SET)) < 0) {
6120 mfra_size = avio_rb32(f);
6121 if (mfra_size < 0 || mfra_size > stream_size) {
6122 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (unreasonable size)\n");
6125 if ((seek_ret = avio_seek(f, -mfra_size, SEEK_CUR)) < 0) {
6129 if (avio_rb32(f) != mfra_size) {
6130 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (size mismatch)\n");
6133 if (avio_rb32(f) != MKBETAG('m', 'f', 'r', 'a')) {
6134 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (tag mismatch)\n");
6137 av_log(c->fc, AV_LOG_VERBOSE, "stream has mfra\n");
6139 ret = read_tfra(c, f);
6145 seek_ret = avio_seek(f, original_pos, SEEK_SET);
6147 av_log(c->fc, AV_LOG_ERROR,
6148 "failed to seek back after looking for mfra\n");
6154 static int mov_read_header(AVFormatContext *s)
6156 MOVContext *mov = s->priv_data;
6157 AVIOContext *pb = s->pb;
6159 MOVAtom atom = { AV_RL32("root") };
6162 if (mov->decryption_key_len != 0 && mov->decryption_key_len != AES_CTR_KEY_SIZE) {
6163 av_log(s, AV_LOG_ERROR, "Invalid decryption key len %d expected %d\n",
6164 mov->decryption_key_len, AES_CTR_KEY_SIZE);
6165 return AVERROR(EINVAL);
6169 mov->trak_index = -1;
6170 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
6171 if (pb->seekable & AVIO_SEEKABLE_NORMAL)
6172 atom.size = avio_size(pb);
6174 atom.size = INT64_MAX;
6176 /* check MOV header */
6178 if (mov->moov_retry)
6179 avio_seek(pb, 0, SEEK_SET);
6180 if ((err = mov_read_default(mov, pb, atom)) < 0) {
6181 av_log(s, AV_LOG_ERROR, "error reading header\n");
6185 } while ((pb->seekable & AVIO_SEEKABLE_NORMAL) && !mov->found_moov && !mov->moov_retry++);
6186 if (!mov->found_moov) {
6187 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
6189 return AVERROR_INVALIDDATA;
6191 av_log(mov->fc, AV_LOG_TRACE, "on_parse_exit_offset=%"PRId64"\n", avio_tell(pb));
6193 if (pb->seekable & AVIO_SEEKABLE_NORMAL) {
6194 if (mov->nb_chapter_tracks > 0 && !mov->ignore_chapters)
6195 mov_read_chapters(s);
6196 for (i = 0; i < s->nb_streams; i++)
6197 if (s->streams[i]->codecpar->codec_tag == AV_RL32("tmcd")) {
6198 mov_read_timecode_track(s, s->streams[i]);
6199 } else if (s->streams[i]->codecpar->codec_tag == AV_RL32("rtmd")) {
6200 mov_read_rtmd_track(s, s->streams[i]);
6204 /* copy timecode metadata from tmcd tracks to the related video streams */
6205 for (i = 0; i < s->nb_streams; i++) {
6206 AVStream *st = s->streams[i];
6207 MOVStreamContext *sc = st->priv_data;
6208 if (sc->timecode_track > 0) {
6209 AVDictionaryEntry *tcr;
6210 int tmcd_st_id = -1;
6212 for (j = 0; j < s->nb_streams; j++)
6213 if (s->streams[j]->id == sc->timecode_track)
6216 if (tmcd_st_id < 0 || tmcd_st_id == i)
6218 tcr = av_dict_get(s->streams[tmcd_st_id]->metadata, "timecode", NULL, 0);
6220 av_dict_set(&st->metadata, "timecode", tcr->value, 0);
6223 export_orphan_timecode(s);
6225 for (i = 0; i < s->nb_streams; i++) {
6226 AVStream *st = s->streams[i];
6227 MOVStreamContext *sc = st->priv_data;
6228 fix_timescale(mov, sc);
6229 if(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id == AV_CODEC_ID_AAC) {
6230 st->skip_samples = sc->start_pad;
6232 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && sc->nb_frames_for_fps > 0 && sc->duration_for_fps > 0)
6233 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
6234 sc->time_scale*(int64_t)sc->nb_frames_for_fps, sc->duration_for_fps, INT_MAX);
6235 if (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE) {
6236 if (st->codecpar->width <= 0 || st->codecpar->height <= 0) {
6237 st->codecpar->width = sc->width;
6238 st->codecpar->height = sc->height;
6240 if (st->codecpar->codec_id == AV_CODEC_ID_DVD_SUBTITLE) {
6241 if ((err = mov_rewrite_dvd_sub_extradata(st)) < 0)
6245 if (mov->handbrake_version &&
6246 mov->handbrake_version <= 1000000*0 + 1000*10 + 2 && // 0.10.2
6247 st->codecpar->codec_id == AV_CODEC_ID_MP3
6249 av_log(s, AV_LOG_VERBOSE, "Forcing full parsing for mp3 stream\n");
6250 st->need_parsing = AVSTREAM_PARSE_FULL;
6254 if (mov->trex_data) {
6255 for (i = 0; i < s->nb_streams; i++) {
6256 AVStream *st = s->streams[i];
6257 MOVStreamContext *sc = st->priv_data;
6258 if (st->duration > 0) {
6259 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
6260 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
6261 sc->data_size, sc->time_scale);
6263 return AVERROR_INVALIDDATA;
6265 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale / st->duration;
6270 if (mov->use_mfra_for > 0) {
6271 for (i = 0; i < s->nb_streams; i++) {
6272 AVStream *st = s->streams[i];
6273 MOVStreamContext *sc = st->priv_data;
6274 if (sc->duration_for_fps > 0) {
6275 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
6276 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
6277 sc->data_size, sc->time_scale);
6279 return AVERROR_INVALIDDATA;
6281 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale /
6282 sc->duration_for_fps;
6287 for (i = 0; i < mov->bitrates_count && i < s->nb_streams; i++) {
6288 if (mov->bitrates[i]) {
6289 s->streams[i]->codecpar->bit_rate = mov->bitrates[i];
6293 ff_rfps_calculate(s);
6295 for (i = 0; i < s->nb_streams; i++) {
6296 AVStream *st = s->streams[i];
6297 MOVStreamContext *sc = st->priv_data;
6299 switch (st->codecpar->codec_type) {
6300 case AVMEDIA_TYPE_AUDIO:
6301 err = ff_replaygain_export(st, s->metadata);
6307 case AVMEDIA_TYPE_VIDEO:
6308 if (sc->display_matrix) {
6309 err = av_stream_add_side_data(st, AV_PKT_DATA_DISPLAYMATRIX, (uint8_t*)sc->display_matrix,
6310 sizeof(int32_t) * 9);
6314 sc->display_matrix = NULL;
6317 err = av_stream_add_side_data(st, AV_PKT_DATA_STEREO3D,
6318 (uint8_t *)sc->stereo3d,
6319 sizeof(*sc->stereo3d));
6323 sc->stereo3d = NULL;
6325 if (sc->spherical) {
6326 err = av_stream_add_side_data(st, AV_PKT_DATA_SPHERICAL,
6327 (uint8_t *)sc->spherical,
6328 sc->spherical_size);
6332 sc->spherical = NULL;
6334 if (sc->mastering) {
6335 err = av_stream_add_side_data(st, AV_PKT_DATA_MASTERING_DISPLAY_METADATA,
6336 (uint8_t *)sc->mastering,
6337 sizeof(*sc->mastering));
6341 sc->mastering = NULL;
6344 err = av_stream_add_side_data(st, AV_PKT_DATA_CONTENT_LIGHT_LEVEL,
6345 (uint8_t *)sc->coll,
6355 ff_configure_buffers_for_index(s, AV_TIME_BASE);
6360 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
6362 AVIndexEntry *sample = NULL;
6363 int64_t best_dts = INT64_MAX;
6365 for (i = 0; i < s->nb_streams; i++) {
6366 AVStream *avst = s->streams[i];
6367 MOVStreamContext *msc = avst->priv_data;
6368 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
6369 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
6370 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
6371 av_log(s, AV_LOG_TRACE, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
6372 if (!sample || (!(s->pb->seekable & AVIO_SEEKABLE_NORMAL) && current_sample->pos < sample->pos) ||
6373 ((s->pb->seekable & AVIO_SEEKABLE_NORMAL) &&
6374 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
6375 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
6376 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
6377 sample = current_sample;
6386 static int should_retry(AVIOContext *pb, int error_code) {
6387 if (error_code == AVERROR_EOF || avio_feof(pb))
6393 static int mov_switch_root(AVFormatContext *s, int64_t target)
6395 MOVContext *mov = s->priv_data;
6397 int already_read = 0;
6399 if (avio_seek(s->pb, target, SEEK_SET) != target) {
6400 av_log(mov->fc, AV_LOG_ERROR, "root atom offset 0x%"PRIx64": partial file\n", target);
6401 return AVERROR_INVALIDDATA;
6404 mov->next_root_atom = 0;
6406 for (i = 0; i < mov->fragment_index_count; i++) {
6407 MOVFragmentIndex *index = mov->fragment_index_data[i];
6409 for (j = 0; j < index->item_count; j++) {
6410 MOVFragmentIndexItem *item = &index->items[j];
6412 mov->next_root_atom = item->moof_offset;
6413 break; // Advance to next index in outer loop
6414 } else if (item->moof_offset == target) {
6415 index->current_item = FFMIN(j, index->current_item);
6416 if (item->headers_read)
6418 item->headers_read = 1;
6423 index->current_item = 0;
6429 mov->found_mdat = 0;
6431 if (mov_read_default(mov, s->pb, (MOVAtom){ AV_RL32("root"), INT64_MAX }) < 0 ||
6434 av_log(s, AV_LOG_TRACE, "read fragments, offset 0x%"PRIx64"\n", avio_tell(s->pb));
6439 static int mov_change_extradata(MOVStreamContext *sc, AVPacket *pkt)
6441 uint8_t *side, *extradata;
6444 /* Save the current index. */
6445 sc->last_stsd_index = sc->stsc_data[sc->stsc_index].id - 1;
6447 /* Notify the decoder that extradata changed. */
6448 extradata_size = sc->extradata_size[sc->last_stsd_index];
6449 extradata = sc->extradata[sc->last_stsd_index];
6450 if (extradata_size > 0 && extradata) {
6451 side = av_packet_new_side_data(pkt,
6452 AV_PKT_DATA_NEW_EXTRADATA,
6455 return AVERROR(ENOMEM);
6456 memcpy(side, extradata, extradata_size);
6462 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
6464 MOVContext *mov = s->priv_data;
6465 MOVStreamContext *sc;
6466 AVIndexEntry *sample;
6467 AVStream *st = NULL;
6468 int64_t current_index;
6472 sample = mov_find_next_sample(s, &st);
6473 if (!sample || (mov->next_root_atom && sample->pos > mov->next_root_atom)) {
6474 if (!mov->next_root_atom)
6476 if ((ret = mov_switch_root(s, mov->next_root_atom)) < 0)
6481 /* must be done just before reading, to avoid infinite loop on sample */
6482 current_index = sc->current_index;
6483 mov_current_sample_inc(sc);
6485 if (mov->next_root_atom) {
6486 sample->pos = FFMIN(sample->pos, mov->next_root_atom);
6487 sample->size = FFMIN(sample->size, (mov->next_root_atom - sample->pos));
6490 if (st->discard != AVDISCARD_ALL) {
6491 int64_t ret64 = avio_seek(sc->pb, sample->pos, SEEK_SET);
6492 if (ret64 != sample->pos) {
6493 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
6494 sc->ffindex, sample->pos);
6495 if (should_retry(sc->pb, ret64)) {
6496 mov_current_sample_dec(sc);
6498 return AVERROR_INVALIDDATA;
6501 if( st->discard == AVDISCARD_NONKEY && 0==(sample->flags & AVINDEX_KEYFRAME) ) {
6502 av_log(mov->fc, AV_LOG_DEBUG, "Nonkey frame from stream %d discarded due to AVDISCARD_NONKEY\n", sc->ffindex);
6506 ret = av_get_packet(sc->pb, pkt, sample->size);
6508 if (should_retry(sc->pb, ret)) {
6509 mov_current_sample_dec(sc);
6513 if (sc->has_palette) {
6516 pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, AVPALETTE_SIZE);
6518 av_log(mov->fc, AV_LOG_ERROR, "Cannot append palette to packet\n");
6520 memcpy(pal, sc->palette, AVPALETTE_SIZE);
6521 sc->has_palette = 0;
6524 #if CONFIG_DV_DEMUXER
6525 if (mov->dv_demux && sc->dv_audio_container) {
6526 avpriv_dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size, pkt->pos);
6527 av_freep(&pkt->data);
6529 ret = avpriv_dv_get_packet(mov->dv_demux, pkt);
6534 if (st->codecpar->codec_id == AV_CODEC_ID_MP3 && !st->need_parsing && pkt->size > 4) {
6535 if (ff_mpa_check_header(AV_RB32(pkt->data)) < 0)
6536 st->need_parsing = AVSTREAM_PARSE_FULL;
6540 pkt->stream_index = sc->ffindex;
6541 pkt->dts = sample->timestamp;
6542 if (sample->flags & AVINDEX_DISCARD_FRAME) {
6543 pkt->flags |= AV_PKT_FLAG_DISCARD;
6545 if (sc->ctts_data && sc->ctts_index < sc->ctts_count) {
6546 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
6547 /* update ctts context */
6549 if (sc->ctts_index < sc->ctts_count &&
6550 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
6552 sc->ctts_sample = 0;
6555 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
6556 st->index_entries[sc->current_sample].timestamp : st->duration;
6557 pkt->duration = next_dts - pkt->dts;
6558 pkt->pts = pkt->dts;
6560 if (st->discard == AVDISCARD_ALL)
6562 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? AV_PKT_FLAG_KEY : 0;
6563 pkt->pos = sample->pos;
6565 /* Multiple stsd handling. */
6566 if (sc->stsc_data) {
6567 /* Keep track of the stsc index for the given sample, then check
6568 * if the stsd index is different from the last used one. */
6570 if (mov_stsc_index_valid(sc->stsc_index, sc->stsc_count) &&
6571 mov_get_stsc_samples(sc, sc->stsc_index) == sc->stsc_sample) {
6573 sc->stsc_sample = 0;
6574 /* Do not check indexes after a switch. */
6575 } else if (sc->stsc_data[sc->stsc_index].id > 0 &&
6576 sc->stsc_data[sc->stsc_index].id - 1 < sc->stsd_count &&
6577 sc->stsc_data[sc->stsc_index].id - 1 != sc->last_stsd_index) {
6578 ret = mov_change_extradata(sc, pkt);
6585 aax_filter(pkt->data, pkt->size, mov);
6587 if (sc->cenc.aes_ctr) {
6588 ret = cenc_filter(mov, sc, current_index, pkt->data, pkt->size);
6597 static int mov_seek_fragment(AVFormatContext *s, AVStream *st, int64_t timestamp)
6599 MOVContext *mov = s->priv_data;
6600 MOVStreamContext *sc = st->priv_data;
6603 if (!mov->fragment_index_complete)
6606 for (i = 0; i < mov->fragment_index_count; i++) {
6607 if (mov->fragment_index_data[i]->track_id == st->id || !sc->has_sidx) {
6608 MOVFragmentIndex *index = mov->fragment_index_data[i];
6609 for (j = index->item_count - 1; j >= 0; j--) {
6610 if (index->items[j].time <= timestamp) {
6611 if (index->items[j].headers_read)
6614 return mov_switch_root(s, index->items[j].moof_offset);
6623 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
6625 MOVStreamContext *sc = st->priv_data;
6626 int sample, time_sample;
6629 int ret = mov_seek_fragment(s, st, timestamp);
6633 sample = av_index_search_timestamp(st, timestamp, flags);
6634 av_log(s, AV_LOG_TRACE, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
6635 if (sample < 0 && st->nb_index_entries && timestamp < st->index_entries[0].timestamp)
6637 if (sample < 0) /* not sure what to do */
6638 return AVERROR_INVALIDDATA;
6639 mov_current_sample_set(sc, sample);
6640 av_log(s, AV_LOG_TRACE, "stream %d, found sample %d\n", st->index, sc->current_sample);
6641 /* adjust ctts index */
6642 if (sc->ctts_data) {
6644 for (i = 0; i < sc->ctts_count; i++) {
6645 int next = time_sample + sc->ctts_data[i].count;
6646 if (next > sc->current_sample) {
6648 sc->ctts_sample = sc->current_sample - time_sample;
6655 /* adjust stsd index */
6657 for (i = 0; i < sc->stsc_count; i++) {
6658 int next = time_sample + mov_get_stsc_samples(sc, i);
6659 if (next > sc->current_sample) {
6661 sc->stsc_sample = sc->current_sample - time_sample;
6670 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
6672 MOVContext *mc = s->priv_data;
6677 if (stream_index >= s->nb_streams)
6678 return AVERROR_INVALIDDATA;
6680 st = s->streams[stream_index];
6681 sample = mov_seek_stream(s, st, sample_time, flags);
6685 if (mc->seek_individually) {
6686 /* adjust seek timestamp to found sample timestamp */
6687 int64_t seek_timestamp = st->index_entries[sample].timestamp;
6689 for (i = 0; i < s->nb_streams; i++) {
6691 MOVStreamContext *sc = s->streams[i]->priv_data;
6693 st->skip_samples = (sample_time <= 0) ? sc->start_pad : 0;
6695 if (stream_index == i)
6698 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
6699 mov_seek_stream(s, st, timestamp, flags);
6702 for (i = 0; i < s->nb_streams; i++) {
6703 MOVStreamContext *sc;
6706 mov_current_sample_set(sc, 0);
6709 MOVStreamContext *sc;
6710 AVIndexEntry *entry = mov_find_next_sample(s, &st);
6712 return AVERROR_INVALIDDATA;
6714 if (sc->ffindex == stream_index && sc->current_sample == sample)
6716 mov_current_sample_inc(sc);
6722 #define OFFSET(x) offsetof(MOVContext, x)
6723 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
6724 static const AVOption mov_options[] = {
6725 {"use_absolute_path",
6726 "allow using absolute path when opening alias, this is a possible security issue",
6727 OFFSET(use_absolute_path), AV_OPT_TYPE_BOOL, {.i64 = 0},
6729 {"seek_streams_individually",
6730 "Seek each stream individually to the to the closest point",
6731 OFFSET(seek_individually), AV_OPT_TYPE_BOOL, { .i64 = 1 },
6733 {"ignore_editlist", "Ignore the edit list atom.", OFFSET(ignore_editlist), AV_OPT_TYPE_BOOL, {.i64 = 0},
6735 {"advanced_editlist",
6736 "Modify the AVIndex according to the editlists. Use this option to decode in the order specified by the edits.",
6737 OFFSET(advanced_editlist), AV_OPT_TYPE_BOOL, {.i64 = 1},
6739 {"ignore_chapters", "", OFFSET(ignore_chapters), AV_OPT_TYPE_BOOL, {.i64 = 0},
6742 "use mfra for fragment timestamps",
6743 OFFSET(use_mfra_for), AV_OPT_TYPE_INT, {.i64 = FF_MOV_FLAG_MFRA_AUTO},
6744 -1, FF_MOV_FLAG_MFRA_PTS, FLAGS,
6746 {"auto", "auto", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_AUTO}, 0, 0,
6747 FLAGS, "use_mfra_for" },
6748 {"dts", "dts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_DTS}, 0, 0,
6749 FLAGS, "use_mfra_for" },
6750 {"pts", "pts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_PTS}, 0, 0,
6751 FLAGS, "use_mfra_for" },
6752 { "export_all", "Export unrecognized metadata entries", OFFSET(export_all),
6753 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
6754 { "export_xmp", "Export full XMP metadata", OFFSET(export_xmp),
6755 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
6756 { "activation_bytes", "Secret bytes for Audible AAX files", OFFSET(activation_bytes),
6757 AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
6758 { "audible_fixed_key", // extracted from libAAX_SDK.so and AAXSDKWin.dll files!
6759 "Fixed key used for handling Audible AAX files", OFFSET(audible_fixed_key),
6760 AV_OPT_TYPE_BINARY, {.str="77214d4b196a87cd520045fd20a51d67"},
6761 .flags = AV_OPT_FLAG_DECODING_PARAM },
6762 { "decryption_key", "The media decryption key (hex)", OFFSET(decryption_key), AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
6763 { "enable_drefs", "Enable external track support.", OFFSET(enable_drefs), AV_OPT_TYPE_BOOL,
6764 {.i64 = 0}, 0, 1, FLAGS },
6769 static const AVClass mov_class = {
6770 .class_name = "mov,mp4,m4a,3gp,3g2,mj2",
6771 .item_name = av_default_item_name,
6772 .option = mov_options,
6773 .version = LIBAVUTIL_VERSION_INT,
6776 AVInputFormat ff_mov_demuxer = {
6777 .name = "mov,mp4,m4a,3gp,3g2,mj2",
6778 .long_name = NULL_IF_CONFIG_SMALL("QuickTime / MOV"),
6779 .priv_class = &mov_class,
6780 .priv_data_size = sizeof(MOVContext),
6781 .extensions = "mov,mp4,m4a,3gp,3g2,mj2",
6782 .read_probe = mov_probe,
6783 .read_header = mov_read_header,
6784 .read_packet = mov_read_packet,
6785 .read_close = mov_read_close,
6786 .read_seek = mov_read_seek,
6787 .flags = AVFMT_NO_BYTE_SEEK,