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/sha.h"
45 #include "libavutil/spherical.h"
46 #include "libavutil/stereo3d.h"
47 #include "libavutil/timecode.h"
48 #include "libavcodec/ac3tab.h"
49 #include "libavcodec/flac.h"
50 #include "libavcodec/mpegaudiodecheader.h"
53 #include "avio_internal.h"
56 #include "libavcodec/get_bits.h"
59 #include "replaygain.h"
65 #include "qtpalette.h"
67 /* those functions parse an atom */
68 /* links atom IDs to parse functions */
69 typedef struct MOVParseTableEntry {
71 int (*parse)(MOVContext *ctx, AVIOContext *pb, MOVAtom atom);
74 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom);
75 static int mov_read_mfra(MOVContext *c, AVIOContext *f);
77 static int mov_metadata_track_or_disc_number(MOVContext *c, AVIOContext *pb,
78 unsigned len, const char *key)
82 short current, total = 0;
83 avio_rb16(pb); // unknown
84 current = avio_rb16(pb);
86 total = avio_rb16(pb);
88 snprintf(buf, sizeof(buf), "%d", current);
90 snprintf(buf, sizeof(buf), "%d/%d", current, total);
91 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
92 av_dict_set(&c->fc->metadata, key, buf, 0);
97 static int mov_metadata_int8_bypass_padding(MOVContext *c, AVIOContext *pb,
98 unsigned len, const char *key)
100 /* bypass padding bytes */
105 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
106 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
111 static int mov_metadata_int8_no_padding(MOVContext *c, AVIOContext *pb,
112 unsigned len, const char *key)
114 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
115 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
120 static int mov_metadata_gnre(MOVContext *c, AVIOContext *pb,
121 unsigned len, const char *key)
125 avio_r8(pb); // unknown
128 if (genre < 1 || genre > ID3v1_GENRE_MAX)
130 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
131 av_dict_set(&c->fc->metadata, key, ff_id3v1_genre_str[genre-1], 0);
136 static const uint32_t mac_to_unicode[128] = {
137 0x00C4,0x00C5,0x00C7,0x00C9,0x00D1,0x00D6,0x00DC,0x00E1,
138 0x00E0,0x00E2,0x00E4,0x00E3,0x00E5,0x00E7,0x00E9,0x00E8,
139 0x00EA,0x00EB,0x00ED,0x00EC,0x00EE,0x00EF,0x00F1,0x00F3,
140 0x00F2,0x00F4,0x00F6,0x00F5,0x00FA,0x00F9,0x00FB,0x00FC,
141 0x2020,0x00B0,0x00A2,0x00A3,0x00A7,0x2022,0x00B6,0x00DF,
142 0x00AE,0x00A9,0x2122,0x00B4,0x00A8,0x2260,0x00C6,0x00D8,
143 0x221E,0x00B1,0x2264,0x2265,0x00A5,0x00B5,0x2202,0x2211,
144 0x220F,0x03C0,0x222B,0x00AA,0x00BA,0x03A9,0x00E6,0x00F8,
145 0x00BF,0x00A1,0x00AC,0x221A,0x0192,0x2248,0x2206,0x00AB,
146 0x00BB,0x2026,0x00A0,0x00C0,0x00C3,0x00D5,0x0152,0x0153,
147 0x2013,0x2014,0x201C,0x201D,0x2018,0x2019,0x00F7,0x25CA,
148 0x00FF,0x0178,0x2044,0x20AC,0x2039,0x203A,0xFB01,0xFB02,
149 0x2021,0x00B7,0x201A,0x201E,0x2030,0x00C2,0x00CA,0x00C1,
150 0x00CB,0x00C8,0x00CD,0x00CE,0x00CF,0x00CC,0x00D3,0x00D4,
151 0xF8FF,0x00D2,0x00DA,0x00DB,0x00D9,0x0131,0x02C6,0x02DC,
152 0x00AF,0x02D8,0x02D9,0x02DA,0x00B8,0x02DD,0x02DB,0x02C7,
155 static int mov_read_mac_string(MOVContext *c, AVIOContext *pb, int len,
156 char *dst, int dstlen)
159 char *end = dst+dstlen-1;
162 for (i = 0; i < len; i++) {
163 uint8_t t, c = avio_r8(pb);
164 if (c < 0x80 && p < end)
167 PUT_UTF8(mac_to_unicode[c-0x80], t, if (p < end) *p++ = t;);
173 static int mov_read_covr(MOVContext *c, AVIOContext *pb, int type, int len)
177 MOVStreamContext *sc;
182 case 0xd: id = AV_CODEC_ID_MJPEG; break;
183 case 0xe: id = AV_CODEC_ID_PNG; break;
184 case 0x1b: id = AV_CODEC_ID_BMP; break;
186 av_log(c->fc, AV_LOG_WARNING, "Unknown cover type: 0x%x.\n", type);
191 st = avformat_new_stream(c->fc, NULL);
193 return AVERROR(ENOMEM);
194 sc = av_mallocz(sizeof(*sc));
196 return AVERROR(ENOMEM);
199 ret = av_get_packet(pb, &pkt, len);
203 if (pkt.size >= 8 && id != AV_CODEC_ID_BMP) {
204 if (AV_RB64(pkt.data) == 0x89504e470d0a1a0a) {
205 id = AV_CODEC_ID_PNG;
207 id = AV_CODEC_ID_MJPEG;
211 st->disposition |= AV_DISPOSITION_ATTACHED_PIC;
213 st->attached_pic = pkt;
214 st->attached_pic.stream_index = st->index;
215 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
217 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
218 st->codecpar->codec_id = id;
224 static int mov_metadata_loci(MOVContext *c, AVIOContext *pb, unsigned len)
226 char language[4] = { 0 };
227 char buf[200], place[100];
228 uint16_t langcode = 0;
229 double longitude, latitude, altitude;
230 const char *key = "location";
232 if (len < 4 + 2 + 1 + 1 + 4 + 4 + 4) {
233 av_log(c->fc, AV_LOG_ERROR, "loci too short\n");
234 return AVERROR_INVALIDDATA;
237 avio_skip(pb, 4); // version+flags
238 langcode = avio_rb16(pb);
239 ff_mov_lang_to_iso639(langcode, language);
242 len -= avio_get_str(pb, len, place, sizeof(place));
244 av_log(c->fc, AV_LOG_ERROR, "place name too long\n");
245 return AVERROR_INVALIDDATA;
247 avio_skip(pb, 1); // role
251 av_log(c->fc, AV_LOG_ERROR,
252 "loci too short (%u bytes left, need at least %d)\n", len, 12);
253 return AVERROR_INVALIDDATA;
255 longitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
256 latitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
257 altitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
259 // Try to output in the same format as the ?xyz field
260 snprintf(buf, sizeof(buf), "%+08.4f%+09.4f", latitude, longitude);
262 av_strlcatf(buf, sizeof(buf), "%+f", altitude);
263 av_strlcatf(buf, sizeof(buf), "/%s", place);
265 if (*language && strcmp(language, "und")) {
267 snprintf(key2, sizeof(key2), "%s-%s", key, language);
268 av_dict_set(&c->fc->metadata, key2, buf, 0);
270 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
271 return av_dict_set(&c->fc->metadata, key, buf, 0);
274 static int mov_metadata_hmmt(MOVContext *c, AVIOContext *pb, unsigned len)
280 if (c->ignore_chapters)
283 n_hmmt = avio_rb32(pb);
284 for (i = 0; i < n_hmmt && !pb->eof_reached; i++) {
285 int moment_time = avio_rb32(pb);
286 avpriv_new_chapter(c->fc, i, av_make_q(1, 1000), moment_time, AV_NOPTS_VALUE, NULL);
291 static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom)
294 char key2[32], language[4] = {0};
296 const char *key = NULL;
297 uint16_t langcode = 0;
298 uint32_t data_type = 0, str_size, str_size_alloc;
299 int (*parse)(MOVContext*, AVIOContext*, unsigned, const char*) = NULL;
304 case MKTAG( '@','P','R','M'): key = "premiere_version"; raw = 1; break;
305 case MKTAG( '@','P','R','Q'): key = "quicktime_version"; raw = 1; break;
306 case MKTAG( 'X','M','P','_'):
307 if (c->export_xmp) { key = "xmp"; raw = 1; } break;
308 case MKTAG( 'a','A','R','T'): key = "album_artist"; break;
309 case MKTAG( 'a','k','I','D'): key = "account_type";
310 parse = mov_metadata_int8_no_padding; break;
311 case MKTAG( 'a','p','I','D'): key = "account_id"; break;
312 case MKTAG( 'c','a','t','g'): key = "category"; break;
313 case MKTAG( 'c','p','i','l'): key = "compilation";
314 parse = mov_metadata_int8_no_padding; break;
315 case MKTAG( 'c','p','r','t'): key = "copyright"; break;
316 case MKTAG( 'd','e','s','c'): key = "description"; break;
317 case MKTAG( 'd','i','s','k'): key = "disc";
318 parse = mov_metadata_track_or_disc_number; break;
319 case MKTAG( 'e','g','i','d'): key = "episode_uid";
320 parse = mov_metadata_int8_no_padding; break;
321 case MKTAG( 'F','I','R','M'): key = "firmware"; raw = 1; break;
322 case MKTAG( 'g','n','r','e'): key = "genre";
323 parse = mov_metadata_gnre; break;
324 case MKTAG( 'h','d','v','d'): key = "hd_video";
325 parse = mov_metadata_int8_no_padding; break;
326 case MKTAG( 'H','M','M','T'):
327 return mov_metadata_hmmt(c, pb, atom.size);
328 case MKTAG( 'k','e','y','w'): key = "keywords"; break;
329 case MKTAG( 'l','d','e','s'): key = "synopsis"; break;
330 case MKTAG( 'l','o','c','i'):
331 return mov_metadata_loci(c, pb, atom.size);
332 case MKTAG( 'p','c','s','t'): key = "podcast";
333 parse = mov_metadata_int8_no_padding; break;
334 case MKTAG( 'p','g','a','p'): key = "gapless_playback";
335 parse = mov_metadata_int8_no_padding; break;
336 case MKTAG( 'p','u','r','d'): key = "purchase_date"; break;
337 case MKTAG( 'r','t','n','g'): key = "rating";
338 parse = mov_metadata_int8_no_padding; break;
339 case MKTAG( 's','o','a','a'): key = "sort_album_artist"; break;
340 case MKTAG( 's','o','a','l'): key = "sort_album"; break;
341 case MKTAG( 's','o','a','r'): key = "sort_artist"; break;
342 case MKTAG( 's','o','c','o'): key = "sort_composer"; break;
343 case MKTAG( 's','o','n','m'): key = "sort_name"; break;
344 case MKTAG( 's','o','s','n'): key = "sort_show"; break;
345 case MKTAG( 's','t','i','k'): key = "media_type";
346 parse = mov_metadata_int8_no_padding; break;
347 case MKTAG( 't','r','k','n'): key = "track";
348 parse = mov_metadata_track_or_disc_number; break;
349 case MKTAG( 't','v','e','n'): key = "episode_id"; break;
350 case MKTAG( 't','v','e','s'): key = "episode_sort";
351 parse = mov_metadata_int8_bypass_padding; break;
352 case MKTAG( 't','v','n','n'): key = "network"; break;
353 case MKTAG( 't','v','s','h'): key = "show"; break;
354 case MKTAG( 't','v','s','n'): key = "season_number";
355 parse = mov_metadata_int8_bypass_padding; break;
356 case MKTAG(0xa9,'A','R','T'): key = "artist"; break;
357 case MKTAG(0xa9,'P','R','D'): key = "producer"; break;
358 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
359 case MKTAG(0xa9,'a','u','t'): key = "artist"; break;
360 case MKTAG(0xa9,'c','h','p'): key = "chapter"; break;
361 case MKTAG(0xa9,'c','m','t'): key = "comment"; break;
362 case MKTAG(0xa9,'c','o','m'): key = "composer"; break;
363 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
364 case MKTAG(0xa9,'d','a','y'): key = "date"; break;
365 case MKTAG(0xa9,'d','i','r'): key = "director"; break;
366 case MKTAG(0xa9,'d','i','s'): key = "disclaimer"; break;
367 case MKTAG(0xa9,'e','d','1'): key = "edit_date"; break;
368 case MKTAG(0xa9,'e','n','c'): key = "encoder"; break;
369 case MKTAG(0xa9,'f','m','t'): key = "original_format"; break;
370 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
371 case MKTAG(0xa9,'g','r','p'): key = "grouping"; break;
372 case MKTAG(0xa9,'h','s','t'): key = "host_computer"; break;
373 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
374 case MKTAG(0xa9,'l','y','r'): key = "lyrics"; break;
375 case MKTAG(0xa9,'m','a','k'): key = "make"; break;
376 case MKTAG(0xa9,'m','o','d'): key = "model"; break;
377 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
378 case MKTAG(0xa9,'o','p','e'): key = "original_artist"; break;
379 case MKTAG(0xa9,'p','r','d'): key = "producer"; break;
380 case MKTAG(0xa9,'p','r','f'): key = "performers"; break;
381 case MKTAG(0xa9,'r','e','q'): key = "playback_requirements"; break;
382 case MKTAG(0xa9,'s','r','c'): key = "original_source"; break;
383 case MKTAG(0xa9,'s','t','3'): key = "subtitle"; break;
384 case MKTAG(0xa9,'s','w','r'): key = "encoder"; break;
385 case MKTAG(0xa9,'t','o','o'): key = "encoder"; break;
386 case MKTAG(0xa9,'t','r','k'): key = "track"; break;
387 case MKTAG(0xa9,'u','r','l'): key = "URL"; break;
388 case MKTAG(0xa9,'w','r','n'): key = "warning"; break;
389 case MKTAG(0xa9,'w','r','t'): key = "composer"; break;
390 case MKTAG(0xa9,'x','y','z'): key = "location"; break;
393 if (c->itunes_metadata && atom.size > 8) {
394 int data_size = avio_rb32(pb);
395 int tag = avio_rl32(pb);
396 if (tag == MKTAG('d','a','t','a') && data_size <= atom.size) {
397 data_type = avio_rb32(pb); // type
398 avio_rb32(pb); // unknown
399 str_size = data_size - 16;
402 if (atom.type == MKTAG('c', 'o', 'v', 'r')) {
403 int ret = mov_read_covr(c, pb, data_type, str_size);
405 av_log(c->fc, AV_LOG_ERROR, "Error parsing cover art.\n");
408 } else if (!key && c->found_hdlr_mdta && c->meta_keys) {
409 uint32_t index = AV_RB32(&atom.type);
410 if (index < c->meta_keys_count) {
411 key = c->meta_keys[index];
413 av_log(c->fc, AV_LOG_WARNING,
414 "The index of 'data' is out of range: %d >= %d.\n",
415 index, c->meta_keys_count);
419 } else if (atom.size > 4 && key && !c->itunes_metadata && !raw) {
420 str_size = avio_rb16(pb); // string length
421 if (str_size > atom.size) {
423 avio_seek(pb, -2, SEEK_CUR);
424 av_log(c->fc, AV_LOG_WARNING, "UDTA parsing failed retrying raw\n");
427 langcode = avio_rb16(pb);
428 ff_mov_lang_to_iso639(langcode, language);
431 str_size = atom.size;
433 if (c->export_all && !key) {
434 snprintf(tmp_key, 5, "%.4s", (char*)&atom.type);
440 if (atom.size < 0 || str_size >= INT_MAX/2)
441 return AVERROR_INVALIDDATA;
443 // Allocates enough space if data_type is a int32 or float32 number, otherwise
444 // worst-case requirement for output string in case of utf8 coded input
445 num = (data_type >= 21 && data_type <= 23);
446 str_size_alloc = (num ? 512 : (raw ? str_size : str_size * 2)) + 1;
447 str = av_mallocz(str_size_alloc);
449 return AVERROR(ENOMEM);
452 parse(c, pb, str_size, key);
454 if (!raw && (data_type == 3 || (data_type == 0 && (langcode < 0x400 || langcode == 0x7fff)))) { // MAC Encoded
455 mov_read_mac_string(c, pb, str_size, str, str_size_alloc);
456 } else if (data_type == 21) { // BE signed integer, variable size
459 val = (int8_t)avio_r8(pb);
460 else if (str_size == 2)
461 val = (int16_t)avio_rb16(pb);
462 else if (str_size == 3)
463 val = ((int32_t)(avio_rb24(pb)<<8))>>8;
464 else if (str_size == 4)
465 val = (int32_t)avio_rb32(pb);
466 if (snprintf(str, str_size_alloc, "%d", val) >= str_size_alloc) {
467 av_log(c->fc, AV_LOG_ERROR,
468 "Failed to store the number (%d) in string.\n", val);
470 return AVERROR_INVALIDDATA;
472 } else if (data_type == 22) { // BE unsigned integer, variable size
473 unsigned int val = 0;
476 else if (str_size == 2)
478 else if (str_size == 3)
480 else if (str_size == 4)
482 if (snprintf(str, str_size_alloc, "%u", val) >= str_size_alloc) {
483 av_log(c->fc, AV_LOG_ERROR,
484 "Failed to store the number (%u) in string.\n", val);
486 return AVERROR_INVALIDDATA;
488 } else if (data_type == 23 && str_size >= 4) { // BE float32
489 float val = av_int2float(avio_rb32(pb));
490 if (snprintf(str, str_size_alloc, "%f", val) >= str_size_alloc) {
491 av_log(c->fc, AV_LOG_ERROR,
492 "Failed to store the float32 number (%f) in string.\n", val);
494 return AVERROR_INVALIDDATA;
497 int ret = ffio_read_size(pb, str, str_size);
504 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
505 av_dict_set(&c->fc->metadata, key, str, 0);
506 if (*language && strcmp(language, "und")) {
507 snprintf(key2, sizeof(key2), "%s-%s", key, language);
508 av_dict_set(&c->fc->metadata, key2, str, 0);
510 if (!strcmp(key, "encoder")) {
511 int major, minor, micro;
512 if (sscanf(str, "HandBrake %d.%d.%d", &major, &minor, µ) == 3) {
513 c->handbrake_version = 1000000*major + 1000*minor + micro;
517 av_log(c->fc, AV_LOG_TRACE, "lang \"%3s\" ", language);
518 av_log(c->fc, AV_LOG_TRACE, "tag \"%s\" value \"%s\" atom \"%.4s\" %d %"PRId64"\n",
519 key, str, (char*)&atom.type, str_size_alloc, atom.size);
525 static int mov_read_chpl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
528 int i, nb_chapters, str_len, version;
532 if (c->ignore_chapters)
535 if ((atom.size -= 5) < 0)
538 version = avio_r8(pb);
541 avio_rb32(pb); // ???
542 nb_chapters = avio_r8(pb);
544 for (i = 0; i < nb_chapters; i++) {
548 start = avio_rb64(pb);
549 str_len = avio_r8(pb);
551 if ((atom.size -= 9+str_len) < 0)
554 ret = ffio_read_size(pb, str, str_len);
558 avpriv_new_chapter(c->fc, i, (AVRational){1,10000000}, start, AV_NOPTS_VALUE, str);
563 #define MIN_DATA_ENTRY_BOX_SIZE 12
564 static int mov_read_dref(MOVContext *c, AVIOContext *pb, MOVAtom atom)
567 MOVStreamContext *sc;
570 if (c->fc->nb_streams < 1)
572 st = c->fc->streams[c->fc->nb_streams-1];
575 avio_rb32(pb); // version + flags
576 entries = avio_rb32(pb);
578 entries > (atom.size - 1) / MIN_DATA_ENTRY_BOX_SIZE + 1 ||
579 entries >= UINT_MAX / sizeof(*sc->drefs))
580 return AVERROR_INVALIDDATA;
584 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
586 return AVERROR(ENOMEM);
587 sc->drefs_count = entries;
589 for (i = 0; i < entries; i++) {
590 MOVDref *dref = &sc->drefs[i];
591 uint32_t size = avio_rb32(pb);
592 int64_t next = avio_tell(pb) + size - 4;
595 return AVERROR_INVALIDDATA;
597 dref->type = avio_rl32(pb);
598 avio_rb32(pb); // version + flags
599 av_log(c->fc, AV_LOG_TRACE, "type %.4s size %d\n", (char*)&dref->type, size);
601 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
602 /* macintosh alias record */
603 uint16_t volume_len, len;
609 volume_len = avio_r8(pb);
610 volume_len = FFMIN(volume_len, 27);
611 ret = ffio_read_size(pb, dref->volume, 27);
614 dref->volume[volume_len] = 0;
615 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", dref->volume, volume_len);
620 len = FFMIN(len, 63);
621 ret = ffio_read_size(pb, dref->filename, 63);
624 dref->filename[len] = 0;
625 av_log(c->fc, AV_LOG_DEBUG, "filename %s, len %d\n", dref->filename, len);
629 /* read next level up_from_alias/down_to_target */
630 dref->nlvl_from = avio_rb16(pb);
631 dref->nlvl_to = avio_rb16(pb);
632 av_log(c->fc, AV_LOG_DEBUG, "nlvl from %d, nlvl to %d\n",
633 dref->nlvl_from, dref->nlvl_to);
637 for (type = 0; type != -1 && avio_tell(pb) < next; ) {
640 type = avio_rb16(pb);
642 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
645 if (type == 2) { // absolute path
647 dref->path = av_mallocz(len+1);
649 return AVERROR(ENOMEM);
651 ret = ffio_read_size(pb, dref->path, len);
653 av_freep(&dref->path);
656 if (len > volume_len && !strncmp(dref->path, dref->volume, volume_len)) {
658 memmove(dref->path, dref->path+volume_len, len);
661 // trim string of any ending zeros
662 for (j = len - 1; j >= 0; j--) {
663 if (dref->path[j] == 0)
668 for (j = 0; j < len; j++)
669 if (dref->path[j] == ':' || dref->path[j] == 0)
671 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
672 } else if (type == 0) { // directory name
674 dref->dir = av_malloc(len+1);
676 return AVERROR(ENOMEM);
678 ret = ffio_read_size(pb, dref->dir, len);
680 av_freep(&dref->dir);
684 for (j = 0; j < len; j++)
685 if (dref->dir[j] == ':')
687 av_log(c->fc, AV_LOG_DEBUG, "dir %s\n", dref->dir);
692 av_log(c->fc, AV_LOG_DEBUG, "Unknown dref type 0x08%x size %d\n",
697 avio_seek(pb, next, SEEK_SET);
702 static int mov_read_hdlr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
706 uint32_t av_unused ctype;
711 avio_r8(pb); /* version */
712 avio_rb24(pb); /* flags */
715 ctype = avio_rl32(pb);
716 type = avio_rl32(pb); /* component subtype */
718 av_log(c->fc, AV_LOG_TRACE, "ctype= %.4s (0x%08x)\n", (char*)&ctype, ctype);
719 av_log(c->fc, AV_LOG_TRACE, "stype= %.4s\n", (char*)&type);
721 if (c->trak_index < 0) { // meta not inside a trak
722 if (type == MKTAG('m','d','t','a')) {
723 c->found_hdlr_mdta = 1;
728 st = c->fc->streams[c->fc->nb_streams-1];
730 if (type == MKTAG('v','i','d','e'))
731 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
732 else if (type == MKTAG('s','o','u','n'))
733 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
734 else if (type == MKTAG('m','1','a',' '))
735 st->codecpar->codec_id = AV_CODEC_ID_MP2;
736 else if ((type == MKTAG('s','u','b','p')) || (type == MKTAG('c','l','c','p')))
737 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
739 avio_rb32(pb); /* component manufacture */
740 avio_rb32(pb); /* component flags */
741 avio_rb32(pb); /* component flags mask */
743 title_size = atom.size - 24;
744 if (title_size > 0) {
745 title_str = av_malloc(title_size + 1); /* Add null terminator */
747 return AVERROR(ENOMEM);
749 ret = ffio_read_size(pb, title_str, title_size);
751 av_freep(&title_str);
754 title_str[title_size] = 0;
756 int off = (!c->isom && title_str[0] == title_size - 1);
757 av_dict_set(&st->metadata, "handler_name", title_str + off, 0);
759 av_freep(&title_str);
765 int ff_mov_read_esds(AVFormatContext *fc, AVIOContext *pb)
770 if (fc->nb_streams < 1)
772 st = fc->streams[fc->nb_streams-1];
774 avio_rb32(pb); /* version + flags */
775 ff_mp4_read_descr(fc, pb, &tag);
776 if (tag == MP4ESDescrTag) {
777 ff_mp4_parse_es_descr(pb, NULL);
779 avio_rb16(pb); /* ID */
781 ff_mp4_read_descr(fc, pb, &tag);
782 if (tag == MP4DecConfigDescrTag)
783 ff_mp4_read_dec_config_descr(fc, st, pb);
787 static int mov_read_esds(MOVContext *c, AVIOContext *pb, MOVAtom atom)
789 return ff_mov_read_esds(c->fc, pb);
792 static int mov_read_dac3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
795 enum AVAudioServiceType *ast;
796 int ac3info, acmod, lfeon, bsmod;
798 if (c->fc->nb_streams < 1)
800 st = c->fc->streams[c->fc->nb_streams-1];
802 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
805 return AVERROR(ENOMEM);
807 ac3info = avio_rb24(pb);
808 bsmod = (ac3info >> 14) & 0x7;
809 acmod = (ac3info >> 11) & 0x7;
810 lfeon = (ac3info >> 10) & 0x1;
811 st->codecpar->channels = ((int[]){2,1,2,3,3,4,4,5})[acmod] + lfeon;
812 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
814 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
816 if (st->codecpar->channels > 1 && bsmod == 0x7)
817 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
819 #if FF_API_LAVF_AVCTX
820 FF_DISABLE_DEPRECATION_WARNINGS
821 st->codec->audio_service_type = *ast;
822 FF_ENABLE_DEPRECATION_WARNINGS
828 static int mov_read_dec3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
831 enum AVAudioServiceType *ast;
832 int eac3info, acmod, lfeon, bsmod;
834 if (c->fc->nb_streams < 1)
836 st = c->fc->streams[c->fc->nb_streams-1];
838 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
841 return AVERROR(ENOMEM);
843 /* No need to parse fields for additional independent substreams and its
844 * associated dependent substreams since libavcodec's E-AC-3 decoder
845 * does not support them yet. */
846 avio_rb16(pb); /* data_rate and num_ind_sub */
847 eac3info = avio_rb24(pb);
848 bsmod = (eac3info >> 12) & 0x1f;
849 acmod = (eac3info >> 9) & 0x7;
850 lfeon = (eac3info >> 8) & 0x1;
851 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
853 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
854 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
856 if (st->codecpar->channels > 1 && bsmod == 0x7)
857 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
859 #if FF_API_LAVF_AVCTX
860 FF_DISABLE_DEPRECATION_WARNINGS
861 st->codec->audio_service_type = *ast;
862 FF_ENABLE_DEPRECATION_WARNINGS
868 static int mov_read_ddts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
870 const uint32_t ddts_size = 20;
873 uint32_t frame_duration_code = 0;
874 uint32_t channel_layout_code = 0;
877 buf = av_malloc(ddts_size + FF_INPUT_BUFFER_PADDING_SIZE);
879 return AVERROR(ENOMEM);
881 if (avio_read(pb, buf, ddts_size) < ddts_size) {
883 return AVERROR_INVALIDDATA;
886 init_get_bits(&gb, buf, 8*ddts_size);
888 if (c->fc->nb_streams < 1) {
891 st = c->fc->streams[c->fc->nb_streams-1];
893 st->codecpar->sample_rate = get_bits_long(&gb, 32);
894 if (st->codecpar->sample_rate <= 0) {
895 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
896 return AVERROR_INVALIDDATA;
898 skip_bits_long(&gb, 32); /* max bitrate */
899 st->codecpar->bit_rate = get_bits_long(&gb, 32);
900 st->codecpar->bits_per_coded_sample = get_bits(&gb, 8);
901 frame_duration_code = get_bits(&gb, 2);
902 skip_bits(&gb, 30); /* various fields */
903 channel_layout_code = get_bits(&gb, 16);
905 st->codecpar->frame_size =
906 (frame_duration_code == 0) ? 512 :
907 (frame_duration_code == 1) ? 1024 :
908 (frame_duration_code == 2) ? 2048 :
909 (frame_duration_code == 3) ? 4096 : 0;
911 if (channel_layout_code > 0xff) {
912 av_log(c->fc, AV_LOG_WARNING, "Unsupported DTS audio channel layout");
914 st->codecpar->channel_layout =
915 ((channel_layout_code & 0x1) ? AV_CH_FRONT_CENTER : 0) |
916 ((channel_layout_code & 0x2) ? AV_CH_FRONT_LEFT : 0) |
917 ((channel_layout_code & 0x2) ? AV_CH_FRONT_RIGHT : 0) |
918 ((channel_layout_code & 0x4) ? AV_CH_SIDE_LEFT : 0) |
919 ((channel_layout_code & 0x4) ? AV_CH_SIDE_RIGHT : 0) |
920 ((channel_layout_code & 0x8) ? AV_CH_LOW_FREQUENCY : 0);
922 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
927 static int mov_read_chan(MOVContext *c, AVIOContext *pb, MOVAtom atom)
931 if (c->fc->nb_streams < 1)
933 st = c->fc->streams[c->fc->nb_streams-1];
938 /* skip version and flags */
941 ff_mov_read_chan(c->fc, pb, st, atom.size - 4);
946 static int mov_read_wfex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
951 if (c->fc->nb_streams < 1)
953 st = c->fc->streams[c->fc->nb_streams-1];
955 if ((ret = ff_get_wav_header(c->fc, pb, st->codecpar, atom.size, 0)) < 0)
956 av_log(c->fc, AV_LOG_WARNING, "get_wav_header failed\n");
961 static int mov_read_pasp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
963 const int num = avio_rb32(pb);
964 const int den = avio_rb32(pb);
967 if (c->fc->nb_streams < 1)
969 st = c->fc->streams[c->fc->nb_streams-1];
971 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
972 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num)) {
973 av_log(c->fc, AV_LOG_WARNING,
974 "sample aspect ratio already set to %d:%d, ignoring 'pasp' atom (%d:%d)\n",
975 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den,
977 } else if (den != 0) {
978 av_reduce(&st->sample_aspect_ratio.num, &st->sample_aspect_ratio.den,
984 /* this atom contains actual media data */
985 static int mov_read_mdat(MOVContext *c, AVIOContext *pb, MOVAtom atom)
987 if (atom.size == 0) /* wrong one (MP4) */
990 return 0; /* now go for moov */
993 #define DRM_BLOB_SIZE 56
995 static int mov_read_adrm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
997 uint8_t intermediate_key[20];
998 uint8_t intermediate_iv[20];
1001 uint8_t file_checksum[20];
1002 uint8_t calculated_checksum[20];
1006 uint8_t *activation_bytes = c->activation_bytes;
1007 uint8_t *fixed_key = c->audible_fixed_key;
1011 sha = av_sha_alloc();
1013 return AVERROR(ENOMEM);
1014 c->aes_decrypt = av_aes_alloc();
1015 if (!c->aes_decrypt) {
1016 ret = AVERROR(ENOMEM);
1020 /* drm blob processing */
1021 avio_read(pb, output, 8); // go to offset 8, absolute position 0x251
1022 avio_read(pb, input, DRM_BLOB_SIZE);
1023 avio_read(pb, output, 4); // go to offset 4, absolute position 0x28d
1024 avio_read(pb, file_checksum, 20);
1026 av_log(c->fc, AV_LOG_INFO, "[aax] file checksum == "); // required by external tools
1027 for (i = 0; i < 20; i++)
1028 av_log(c->fc, AV_LOG_INFO, "%02x", file_checksum[i]);
1029 av_log(c->fc, AV_LOG_INFO, "\n");
1031 /* verify activation data */
1032 if (!activation_bytes) {
1033 av_log(c->fc, AV_LOG_WARNING, "[aax] activation_bytes option is missing!\n");
1034 ret = 0; /* allow ffprobe to continue working on .aax files */
1037 if (c->activation_bytes_size != 4) {
1038 av_log(c->fc, AV_LOG_FATAL, "[aax] activation_bytes value needs to be 4 bytes!\n");
1039 ret = AVERROR(EINVAL);
1043 /* verify fixed key */
1044 if (c->audible_fixed_key_size != 16) {
1045 av_log(c->fc, AV_LOG_FATAL, "[aax] audible_fixed_key value needs to be 16 bytes!\n");
1046 ret = AVERROR(EINVAL);
1050 /* AAX (and AAX+) key derivation */
1051 av_sha_init(sha, 160);
1052 av_sha_update(sha, fixed_key, 16);
1053 av_sha_update(sha, activation_bytes, 4);
1054 av_sha_final(sha, intermediate_key);
1055 av_sha_init(sha, 160);
1056 av_sha_update(sha, fixed_key, 16);
1057 av_sha_update(sha, intermediate_key, 20);
1058 av_sha_update(sha, activation_bytes, 4);
1059 av_sha_final(sha, intermediate_iv);
1060 av_sha_init(sha, 160);
1061 av_sha_update(sha, intermediate_key, 16);
1062 av_sha_update(sha, intermediate_iv, 16);
1063 av_sha_final(sha, calculated_checksum);
1064 if (memcmp(calculated_checksum, file_checksum, 20)) { // critical error
1065 av_log(c->fc, AV_LOG_ERROR, "[aax] mismatch in checksums!\n");
1066 ret = AVERROR_INVALIDDATA;
1069 av_aes_init(c->aes_decrypt, intermediate_key, 128, 1);
1070 av_aes_crypt(c->aes_decrypt, output, input, DRM_BLOB_SIZE >> 4, intermediate_iv, 1);
1071 for (i = 0; i < 4; i++) {
1072 // file data (in output) is stored in big-endian mode
1073 if (activation_bytes[i] != output[3 - i]) { // critical error
1074 av_log(c->fc, AV_LOG_ERROR, "[aax] error in drm blob decryption!\n");
1075 ret = AVERROR_INVALIDDATA;
1079 memcpy(c->file_key, output + 8, 16);
1080 memcpy(input, output + 26, 16);
1081 av_sha_init(sha, 160);
1082 av_sha_update(sha, input, 16);
1083 av_sha_update(sha, c->file_key, 16);
1084 av_sha_update(sha, fixed_key, 16);
1085 av_sha_final(sha, c->file_iv);
1093 // Audible AAX (and AAX+) bytestream decryption
1094 static int aax_filter(uint8_t *input, int size, MOVContext *c)
1097 unsigned char iv[16];
1099 memcpy(iv, c->file_iv, 16); // iv is overwritten
1100 blocks = size >> 4; // trailing bytes are not encrypted!
1101 av_aes_init(c->aes_decrypt, c->file_key, 128, 1);
1102 av_aes_crypt(c->aes_decrypt, input, input, blocks, iv, 1);
1107 /* read major brand, minor version and compatible brands and store them as metadata */
1108 static int mov_read_ftyp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1111 int comp_brand_size;
1112 char* comp_brands_str;
1113 uint8_t type[5] = {0};
1114 int ret = ffio_read_size(pb, type, 4);
1118 if (strcmp(type, "qt "))
1120 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
1121 av_dict_set(&c->fc->metadata, "major_brand", type, 0);
1122 minor_ver = avio_rb32(pb); /* minor version */
1123 av_dict_set_int(&c->fc->metadata, "minor_version", minor_ver, 0);
1125 comp_brand_size = atom.size - 8;
1126 if (comp_brand_size < 0)
1127 return AVERROR_INVALIDDATA;
1128 comp_brands_str = av_malloc(comp_brand_size + 1); /* Add null terminator */
1129 if (!comp_brands_str)
1130 return AVERROR(ENOMEM);
1132 ret = ffio_read_size(pb, comp_brands_str, comp_brand_size);
1134 av_freep(&comp_brands_str);
1137 comp_brands_str[comp_brand_size] = 0;
1138 av_dict_set(&c->fc->metadata, "compatible_brands", comp_brands_str, 0);
1139 av_freep(&comp_brands_str);
1144 /* this atom should contain all header atoms */
1145 static int mov_read_moov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1149 if (c->found_moov) {
1150 av_log(c->fc, AV_LOG_WARNING, "Found duplicated MOOV Atom. Skipped it\n");
1151 avio_skip(pb, atom.size);
1155 if ((ret = mov_read_default(c, pb, atom)) < 0)
1157 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
1158 /* so we don't parse the whole file if over a network */
1160 return 0; /* now go for mdat */
1163 static int mov_read_moof(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1165 if (!c->has_looked_for_mfra && c->use_mfra_for > 0) {
1166 c->has_looked_for_mfra = 1;
1169 av_log(c->fc, AV_LOG_VERBOSE, "stream has moof boxes, will look "
1171 if ((ret = mov_read_mfra(c, pb)) < 0) {
1172 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but failed to "
1173 "read the mfra (may be a live ismv)\n");
1176 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but stream is not "
1177 "seekable, can not look for mfra\n");
1180 c->fragment.moof_offset = c->fragment.implicit_offset = avio_tell(pb) - 8;
1181 av_log(c->fc, AV_LOG_TRACE, "moof offset %"PRIx64"\n", c->fragment.moof_offset);
1182 return mov_read_default(c, pb, atom);
1185 static void mov_metadata_creation_time(AVDictionary **metadata, int64_t time)
1188 if(time >= 2082844800)
1189 time -= 2082844800; /* seconds between 1904-01-01 and Epoch */
1190 avpriv_dict_set_timestamp(metadata, "creation_time", time * 1000000);
1194 static int mov_read_mdhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1197 MOVStreamContext *sc;
1199 char language[4] = {0};
1201 int64_t creation_time;
1203 if (c->fc->nb_streams < 1)
1205 st = c->fc->streams[c->fc->nb_streams-1];
1208 if (sc->time_scale) {
1209 av_log(c->fc, AV_LOG_ERROR, "Multiple mdhd?\n");
1210 return AVERROR_INVALIDDATA;
1213 version = avio_r8(pb);
1215 avpriv_request_sample(c->fc, "Version %d", version);
1216 return AVERROR_PATCHWELCOME;
1218 avio_rb24(pb); /* flags */
1220 creation_time = avio_rb64(pb);
1223 creation_time = avio_rb32(pb);
1224 avio_rb32(pb); /* modification time */
1226 mov_metadata_creation_time(&st->metadata, creation_time);
1228 sc->time_scale = avio_rb32(pb);
1229 if (sc->time_scale <= 0) {
1230 av_log(c->fc, AV_LOG_ERROR, "Invalid mdhd time scale %d\n", sc->time_scale);
1231 return AVERROR_INVALIDDATA;
1233 st->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1235 lang = avio_rb16(pb); /* language */
1236 if (ff_mov_lang_to_iso639(lang, language))
1237 av_dict_set(&st->metadata, "language", language, 0);
1238 avio_rb16(pb); /* quality */
1243 static int mov_read_mvhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1246 int64_t creation_time;
1247 int version = avio_r8(pb); /* version */
1248 avio_rb24(pb); /* flags */
1251 creation_time = avio_rb64(pb);
1254 creation_time = avio_rb32(pb);
1255 avio_rb32(pb); /* modification time */
1257 mov_metadata_creation_time(&c->fc->metadata, creation_time);
1258 c->time_scale = avio_rb32(pb); /* time scale */
1259 if (c->time_scale <= 0) {
1260 av_log(c->fc, AV_LOG_ERROR, "Invalid mvhd time scale %d\n", c->time_scale);
1261 return AVERROR_INVALIDDATA;
1263 av_log(c->fc, AV_LOG_TRACE, "time scale = %i\n", c->time_scale);
1265 c->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1266 // set the AVCodecContext duration because the duration of individual tracks
1267 // may be inaccurate
1268 if (c->time_scale > 0 && !c->trex_data)
1269 c->fc->duration = av_rescale(c->duration, AV_TIME_BASE, c->time_scale);
1270 avio_rb32(pb); /* preferred scale */
1272 avio_rb16(pb); /* preferred volume */
1274 avio_skip(pb, 10); /* reserved */
1276 /* movie display matrix, store it in main context and use it later on */
1277 for (i = 0; i < 3; i++) {
1278 c->movie_display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
1279 c->movie_display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
1280 c->movie_display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
1283 avio_rb32(pb); /* preview time */
1284 avio_rb32(pb); /* preview duration */
1285 avio_rb32(pb); /* poster time */
1286 avio_rb32(pb); /* selection time */
1287 avio_rb32(pb); /* selection duration */
1288 avio_rb32(pb); /* current time */
1289 avio_rb32(pb); /* next track ID */
1294 static int mov_read_enda(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1299 if (c->fc->nb_streams < 1)
1301 st = c->fc->streams[c->fc->nb_streams-1];
1303 little_endian = avio_rb16(pb) & 0xFF;
1304 av_log(c->fc, AV_LOG_TRACE, "enda %d\n", little_endian);
1305 if (little_endian == 1) {
1306 switch (st->codecpar->codec_id) {
1307 case AV_CODEC_ID_PCM_S24BE:
1308 st->codecpar->codec_id = AV_CODEC_ID_PCM_S24LE;
1310 case AV_CODEC_ID_PCM_S32BE:
1311 st->codecpar->codec_id = AV_CODEC_ID_PCM_S32LE;
1313 case AV_CODEC_ID_PCM_F32BE:
1314 st->codecpar->codec_id = AV_CODEC_ID_PCM_F32LE;
1316 case AV_CODEC_ID_PCM_F64BE:
1317 st->codecpar->codec_id = AV_CODEC_ID_PCM_F64LE;
1326 static int mov_read_colr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1329 char color_parameter_type[5] = { 0 };
1330 uint16_t color_primaries, color_trc, color_matrix;
1333 if (c->fc->nb_streams < 1)
1335 st = c->fc->streams[c->fc->nb_streams - 1];
1337 ret = ffio_read_size(pb, color_parameter_type, 4);
1340 if (strncmp(color_parameter_type, "nclx", 4) &&
1341 strncmp(color_parameter_type, "nclc", 4)) {
1342 av_log(c->fc, AV_LOG_WARNING, "unsupported color_parameter_type %s\n",
1343 color_parameter_type);
1347 color_primaries = avio_rb16(pb);
1348 color_trc = avio_rb16(pb);
1349 color_matrix = avio_rb16(pb);
1351 av_log(c->fc, AV_LOG_TRACE,
1352 "%s: pri %d trc %d matrix %d",
1353 color_parameter_type, color_primaries, color_trc, color_matrix);
1355 if (!strncmp(color_parameter_type, "nclx", 4)) {
1356 uint8_t color_range = avio_r8(pb) >> 7;
1357 av_log(c->fc, AV_LOG_TRACE, " full %"PRIu8"", color_range);
1359 st->codecpar->color_range = AVCOL_RANGE_JPEG;
1361 st->codecpar->color_range = AVCOL_RANGE_MPEG;
1363 if (color_primaries >= AVCOL_PRI_NB)
1364 color_primaries = AVCOL_PRI_UNSPECIFIED;
1365 if (color_trc >= AVCOL_TRC_NB)
1366 color_trc = AVCOL_TRC_UNSPECIFIED;
1367 if (color_matrix >= AVCOL_SPC_NB)
1368 color_matrix = AVCOL_SPC_UNSPECIFIED;
1369 st->codecpar->color_primaries = color_primaries;
1370 st->codecpar->color_trc = color_trc;
1371 st->codecpar->color_space = color_matrix;
1372 av_log(c->fc, AV_LOG_TRACE, "\n");
1377 static int mov_read_fiel(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1380 unsigned mov_field_order;
1381 enum AVFieldOrder decoded_field_order = AV_FIELD_UNKNOWN;
1383 if (c->fc->nb_streams < 1) // will happen with jp2 files
1385 st = c->fc->streams[c->fc->nb_streams-1];
1387 return AVERROR_INVALIDDATA;
1388 mov_field_order = avio_rb16(pb);
1389 if ((mov_field_order & 0xFF00) == 0x0100)
1390 decoded_field_order = AV_FIELD_PROGRESSIVE;
1391 else if ((mov_field_order & 0xFF00) == 0x0200) {
1392 switch (mov_field_order & 0xFF) {
1393 case 0x01: decoded_field_order = AV_FIELD_TT;
1395 case 0x06: decoded_field_order = AV_FIELD_BB;
1397 case 0x09: decoded_field_order = AV_FIELD_TB;
1399 case 0x0E: decoded_field_order = AV_FIELD_BT;
1403 if (decoded_field_order == AV_FIELD_UNKNOWN && mov_field_order) {
1404 av_log(NULL, AV_LOG_ERROR, "Unknown MOV field order 0x%04x\n", mov_field_order);
1406 st->codecpar->field_order = decoded_field_order;
1411 static int mov_realloc_extradata(AVCodecParameters *par, MOVAtom atom)
1414 uint64_t size = (uint64_t)par->extradata_size + atom.size + 8 + AV_INPUT_BUFFER_PADDING_SIZE;
1415 if (size > INT_MAX || (uint64_t)atom.size > INT_MAX)
1416 return AVERROR_INVALIDDATA;
1417 if ((err = av_reallocp(&par->extradata, size)) < 0) {
1418 par->extradata_size = 0;
1421 par->extradata_size = size - AV_INPUT_BUFFER_PADDING_SIZE;
1425 /* Read a whole atom into the extradata return the size of the atom read, possibly truncated if != atom.size */
1426 static int64_t mov_read_atom_into_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1427 AVCodecParameters *par, uint8_t *buf)
1429 int64_t result = atom.size;
1432 AV_WB32(buf , atom.size + 8);
1433 AV_WL32(buf + 4, atom.type);
1434 err = ffio_read_size(pb, buf + 8, atom.size);
1436 par->extradata_size -= atom.size;
1438 } else if (err < atom.size) {
1439 av_log(c->fc, AV_LOG_WARNING, "truncated extradata\n");
1440 par->extradata_size -= atom.size - err;
1443 memset(buf + 8 + err, 0, AV_INPUT_BUFFER_PADDING_SIZE);
1447 /* FIXME modify QDM2/SVQ3/H.264 decoders to take full atom as extradata */
1448 static int mov_read_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1449 enum AVCodecID codec_id)
1452 uint64_t original_size;
1455 if (c->fc->nb_streams < 1) // will happen with jp2 files
1457 st = c->fc->streams[c->fc->nb_streams-1];
1459 if (st->codecpar->codec_id != codec_id)
1460 return 0; /* unexpected codec_id - don't mess with extradata */
1462 original_size = st->codecpar->extradata_size;
1463 err = mov_realloc_extradata(st->codecpar, atom);
1467 err = mov_read_atom_into_extradata(c, pb, atom, st->codecpar, st->codecpar->extradata + original_size);
1470 return 0; // Note: this is the original behavior to ignore truncation.
1473 /* wrapper functions for reading ALAC/AVS/MJPEG/MJPEG2000 extradata atoms only for those codecs */
1474 static int mov_read_alac(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1476 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_ALAC);
1479 static int mov_read_avss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1481 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVS);
1484 static int mov_read_jp2h(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1486 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_JPEG2000);
1489 static int mov_read_dpxe(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1491 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_R10K);
1494 static int mov_read_avid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1496 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVUI);
1498 ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_DNXHD);
1502 static int mov_read_targa_y216(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1504 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_TARGA_Y216);
1506 if (!ret && c->fc->nb_streams >= 1) {
1507 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1508 if (par->extradata_size >= 40) {
1509 par->height = AV_RB16(&par->extradata[36]);
1510 par->width = AV_RB16(&par->extradata[38]);
1516 static int mov_read_ares(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1518 if (c->fc->nb_streams >= 1) {
1519 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1520 if (par->codec_tag == MKTAG('A', 'V', 'i', 'n') &&
1521 par->codec_id == AV_CODEC_ID_H264 &&
1525 cid = avio_rb16(pb);
1526 /* For AVID AVCI50, force width of 1440 to be able to select the correct SPS and PPS */
1527 if (cid == 0xd4d || cid == 0xd4e)
1530 } else if ((par->codec_tag == MKTAG('A', 'V', 'd', '1') ||
1531 par->codec_tag == MKTAG('A', 'V', 'd', 'n')) &&
1535 num = avio_rb32(pb);
1536 den = avio_rb32(pb);
1537 if (num <= 0 || den <= 0)
1539 switch (avio_rb32(pb)) {
1541 if (den >= INT_MAX / 2)
1545 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.num = num;
1546 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.den = den;
1553 return mov_read_avid(c, pb, atom);
1556 static int mov_read_aclr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1560 uint64_t original_size;
1561 if (c->fc->nb_streams >= 1) {
1562 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1563 if (par->codec_id == AV_CODEC_ID_H264)
1565 if (atom.size == 16) {
1566 original_size = par->extradata_size;
1567 ret = mov_realloc_extradata(par, atom);
1569 length = mov_read_atom_into_extradata(c, pb, atom, par, par->extradata + original_size);
1570 if (length == atom.size) {
1571 const uint8_t range_value = par->extradata[original_size + 19];
1572 switch (range_value) {
1574 par->color_range = AVCOL_RANGE_MPEG;
1577 par->color_range = AVCOL_RANGE_JPEG;
1580 av_log(c, AV_LOG_WARNING, "ignored unknown aclr value (%d)\n", range_value);
1583 ff_dlog(c, "color_range: %d\n", par->color_range);
1585 /* For some reason the whole atom was not added to the extradata */
1586 av_log(c, AV_LOG_ERROR, "aclr not decoded - incomplete atom\n");
1589 av_log(c, AV_LOG_ERROR, "aclr not decoded - unable to add atom to extradata\n");
1592 av_log(c, AV_LOG_WARNING, "aclr not decoded - unexpected size %"PRId64"\n", atom.size);
1599 static int mov_read_svq3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1601 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_SVQ3);
1604 static int mov_read_wave(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1609 if (c->fc->nb_streams < 1)
1611 st = c->fc->streams[c->fc->nb_streams-1];
1613 if ((uint64_t)atom.size > (1<<30))
1614 return AVERROR_INVALIDDATA;
1616 if (st->codecpar->codec_id == AV_CODEC_ID_QDM2 ||
1617 st->codecpar->codec_id == AV_CODEC_ID_QDMC ||
1618 st->codecpar->codec_id == AV_CODEC_ID_SPEEX) {
1619 // pass all frma atom to codec, needed at least for QDMC and QDM2
1620 av_freep(&st->codecpar->extradata);
1621 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1624 } else if (atom.size > 8) { /* to read frma, esds atoms */
1625 if (st->codecpar->codec_id == AV_CODEC_ID_ALAC && atom.size >= 24) {
1627 ret = ffio_ensure_seekback(pb, 8);
1630 buffer = avio_rb64(pb);
1632 if ( (buffer & 0xFFFFFFFF) == MKBETAG('f','r','m','a')
1633 && buffer >> 32 <= atom.size
1634 && buffer >> 32 >= 8) {
1637 } else if (!st->codecpar->extradata_size) {
1638 #define ALAC_EXTRADATA_SIZE 36
1639 st->codecpar->extradata = av_mallocz(ALAC_EXTRADATA_SIZE + AV_INPUT_BUFFER_PADDING_SIZE);
1640 if (!st->codecpar->extradata)
1641 return AVERROR(ENOMEM);
1642 st->codecpar->extradata_size = ALAC_EXTRADATA_SIZE;
1643 AV_WB32(st->codecpar->extradata , ALAC_EXTRADATA_SIZE);
1644 AV_WB32(st->codecpar->extradata + 4, MKTAG('a','l','a','c'));
1645 AV_WB64(st->codecpar->extradata + 12, buffer);
1646 avio_read(pb, st->codecpar->extradata + 20, 16);
1647 avio_skip(pb, atom.size - 24);
1651 if ((ret = mov_read_default(c, pb, atom)) < 0)
1654 avio_skip(pb, atom.size);
1659 * This function reads atom content and puts data in extradata without tag
1660 * nor size unlike mov_read_extradata.
1662 static int mov_read_glbl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1667 if (c->fc->nb_streams < 1)
1669 st = c->fc->streams[c->fc->nb_streams-1];
1671 if ((uint64_t)atom.size > (1<<30))
1672 return AVERROR_INVALIDDATA;
1674 if (atom.size >= 10) {
1675 // Broken files created by legacy versions of libavformat will
1676 // wrap a whole fiel atom inside of a glbl atom.
1677 unsigned size = avio_rb32(pb);
1678 unsigned type = avio_rl32(pb);
1679 avio_seek(pb, -8, SEEK_CUR);
1680 if (type == MKTAG('f','i','e','l') && size == atom.size)
1681 return mov_read_default(c, pb, atom);
1683 if (st->codecpar->extradata_size > 1 && st->codecpar->extradata) {
1684 av_log(c, AV_LOG_WARNING, "ignoring multiple glbl\n");
1687 av_freep(&st->codecpar->extradata);
1688 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1695 static int mov_read_dvc1(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1698 uint8_t profile_level;
1701 if (c->fc->nb_streams < 1)
1703 st = c->fc->streams[c->fc->nb_streams-1];
1705 if (atom.size >= (1<<28) || atom.size < 7)
1706 return AVERROR_INVALIDDATA;
1708 profile_level = avio_r8(pb);
1709 if ((profile_level & 0xf0) != 0xc0)
1712 avio_seek(pb, 6, SEEK_CUR);
1713 av_freep(&st->codecpar->extradata);
1714 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 7);
1722 * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
1723 * but can have extradata appended at the end after the 40 bytes belonging
1726 static int mov_read_strf(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1731 if (c->fc->nb_streams < 1)
1733 if (atom.size <= 40)
1735 st = c->fc->streams[c->fc->nb_streams-1];
1737 if ((uint64_t)atom.size > (1<<30))
1738 return AVERROR_INVALIDDATA;
1741 av_freep(&st->codecpar->extradata);
1742 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 40);
1749 static int mov_read_stco(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1752 MOVStreamContext *sc;
1753 unsigned int i, entries;
1755 if (c->fc->nb_streams < 1)
1757 st = c->fc->streams[c->fc->nb_streams-1];
1760 avio_r8(pb); /* version */
1761 avio_rb24(pb); /* flags */
1763 entries = avio_rb32(pb);
1768 if (sc->chunk_offsets)
1769 av_log(c->fc, AV_LOG_WARNING, "Duplicated STCO atom\n");
1770 av_free(sc->chunk_offsets);
1771 sc->chunk_count = 0;
1772 sc->chunk_offsets = av_malloc_array(entries, sizeof(*sc->chunk_offsets));
1773 if (!sc->chunk_offsets)
1774 return AVERROR(ENOMEM);
1775 sc->chunk_count = entries;
1777 if (atom.type == MKTAG('s','t','c','o'))
1778 for (i = 0; i < entries && !pb->eof_reached; i++)
1779 sc->chunk_offsets[i] = avio_rb32(pb);
1780 else if (atom.type == MKTAG('c','o','6','4'))
1781 for (i = 0; i < entries && !pb->eof_reached; i++)
1782 sc->chunk_offsets[i] = avio_rb64(pb);
1784 return AVERROR_INVALIDDATA;
1786 sc->chunk_count = i;
1788 if (pb->eof_reached)
1795 * Compute codec id for 'lpcm' tag.
1796 * See CoreAudioTypes and AudioStreamBasicDescription at Apple.
1798 enum AVCodecID ff_mov_get_lpcm_codec_id(int bps, int flags)
1805 return ff_get_pcm_codec_id(bps, flags & 1, flags & 2, flags & 4 ? -1 : 0);
1808 static int mov_codec_id(AVStream *st, uint32_t format)
1810 int id = ff_codec_get_id(ff_codec_movaudio_tags, format);
1813 ((format & 0xFFFF) == 'm' + ('s' << 8) ||
1814 (format & 0xFFFF) == 'T' + ('S' << 8)))
1815 id = ff_codec_get_id(ff_codec_wav_tags, av_bswap32(format) & 0xFFFF);
1817 if (st->codecpar->codec_type != AVMEDIA_TYPE_VIDEO && id > 0) {
1818 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
1819 } else if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO &&
1820 /* skip old ASF MPEG-4 tag */
1821 format && format != MKTAG('m','p','4','s')) {
1822 id = ff_codec_get_id(ff_codec_movvideo_tags, format);
1824 id = ff_codec_get_id(ff_codec_bmp_tags, format);
1826 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
1827 else if (st->codecpar->codec_type == AVMEDIA_TYPE_DATA ||
1828 (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE &&
1829 st->codecpar->codec_id == AV_CODEC_ID_NONE)) {
1830 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
1832 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
1836 st->codecpar->codec_tag = format;
1841 static void mov_parse_stsd_video(MOVContext *c, AVIOContext *pb,
1842 AVStream *st, MOVStreamContext *sc)
1844 uint8_t codec_name[32] = { 0 };
1848 /* The first 16 bytes of the video sample description are already
1849 * read in ff_mov_read_stsd_entries() */
1850 stsd_start = avio_tell(pb) - 16;
1852 avio_rb16(pb); /* version */
1853 avio_rb16(pb); /* revision level */
1854 avio_rb32(pb); /* vendor */
1855 avio_rb32(pb); /* temporal quality */
1856 avio_rb32(pb); /* spatial quality */
1858 st->codecpar->width = avio_rb16(pb); /* width */
1859 st->codecpar->height = avio_rb16(pb); /* height */
1861 avio_rb32(pb); /* horiz resolution */
1862 avio_rb32(pb); /* vert resolution */
1863 avio_rb32(pb); /* data size, always 0 */
1864 avio_rb16(pb); /* frames per samples */
1866 len = avio_r8(pb); /* codec name, pascal string */
1869 mov_read_mac_string(c, pb, len, codec_name, sizeof(codec_name));
1871 avio_skip(pb, 31 - len);
1874 av_dict_set(&st->metadata, "encoder", codec_name, 0);
1876 /* codec_tag YV12 triggers an UV swap in rawdec.c */
1877 if (!memcmp(codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25)) {
1878 st->codecpar->codec_tag = MKTAG('I', '4', '2', '0');
1879 st->codecpar->width &= ~1;
1880 st->codecpar->height &= ~1;
1882 /* Flash Media Server uses tag H.263 with Sorenson Spark */
1883 if (st->codecpar->codec_tag == MKTAG('H','2','6','3') &&
1884 !memcmp(codec_name, "Sorenson H263", 13))
1885 st->codecpar->codec_id = AV_CODEC_ID_FLV1;
1887 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* depth */
1889 avio_seek(pb, stsd_start, SEEK_SET);
1891 if (ff_get_qtpalette(st->codecpar->codec_id, pb, sc->palette)) {
1892 st->codecpar->bits_per_coded_sample &= 0x1F;
1893 sc->has_palette = 1;
1897 static void mov_parse_stsd_audio(MOVContext *c, AVIOContext *pb,
1898 AVStream *st, MOVStreamContext *sc)
1900 int bits_per_sample, flags;
1901 uint16_t version = avio_rb16(pb);
1902 AVDictionaryEntry *compatible_brands = av_dict_get(c->fc->metadata, "compatible_brands", NULL, AV_DICT_MATCH_CASE);
1904 avio_rb16(pb); /* revision level */
1905 avio_rb32(pb); /* vendor */
1907 st->codecpar->channels = avio_rb16(pb); /* channel count */
1908 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* sample size */
1909 av_log(c->fc, AV_LOG_TRACE, "audio channels %d\n", st->codecpar->channels);
1911 sc->audio_cid = avio_rb16(pb);
1912 avio_rb16(pb); /* packet size = 0 */
1914 st->codecpar->sample_rate = ((avio_rb32(pb) >> 16));
1916 // Read QT version 1 fields. In version 0 these do not exist.
1917 av_log(c->fc, AV_LOG_TRACE, "version =%d, isom =%d\n", version, c->isom);
1919 (compatible_brands && strstr(compatible_brands->value, "qt "))) {
1922 sc->samples_per_frame = avio_rb32(pb);
1923 avio_rb32(pb); /* bytes per packet */
1924 sc->bytes_per_frame = avio_rb32(pb);
1925 avio_rb32(pb); /* bytes per sample */
1926 } else if (version == 2) {
1927 avio_rb32(pb); /* sizeof struct only */
1928 st->codecpar->sample_rate = av_int2double(avio_rb64(pb));
1929 st->codecpar->channels = avio_rb32(pb);
1930 avio_rb32(pb); /* always 0x7F000000 */
1931 st->codecpar->bits_per_coded_sample = avio_rb32(pb);
1933 flags = avio_rb32(pb); /* lpcm format specific flag */
1934 sc->bytes_per_frame = avio_rb32(pb);
1935 sc->samples_per_frame = avio_rb32(pb);
1936 if (st->codecpar->codec_tag == MKTAG('l','p','c','m'))
1937 st->codecpar->codec_id =
1938 ff_mov_get_lpcm_codec_id(st->codecpar->bits_per_coded_sample,
1941 if (version == 0 || (version == 1 && sc->audio_cid != -2)) {
1942 /* can't correctly handle variable sized packet as audio unit */
1943 switch (st->codecpar->codec_id) {
1944 case AV_CODEC_ID_MP2:
1945 case AV_CODEC_ID_MP3:
1946 st->need_parsing = AVSTREAM_PARSE_FULL;
1952 if (sc->format == 0) {
1953 if (st->codecpar->bits_per_coded_sample == 8)
1954 st->codecpar->codec_id = mov_codec_id(st, MKTAG('r','a','w',' '));
1955 else if (st->codecpar->bits_per_coded_sample == 16)
1956 st->codecpar->codec_id = mov_codec_id(st, MKTAG('t','w','o','s'));
1959 switch (st->codecpar->codec_id) {
1960 case AV_CODEC_ID_PCM_S8:
1961 case AV_CODEC_ID_PCM_U8:
1962 if (st->codecpar->bits_per_coded_sample == 16)
1963 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16BE;
1965 case AV_CODEC_ID_PCM_S16LE:
1966 case AV_CODEC_ID_PCM_S16BE:
1967 if (st->codecpar->bits_per_coded_sample == 8)
1968 st->codecpar->codec_id = AV_CODEC_ID_PCM_S8;
1969 else if (st->codecpar->bits_per_coded_sample == 24)
1970 st->codecpar->codec_id =
1971 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
1972 AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
1973 else if (st->codecpar->bits_per_coded_sample == 32)
1974 st->codecpar->codec_id =
1975 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
1976 AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
1978 /* set values for old format before stsd version 1 appeared */
1979 case AV_CODEC_ID_MACE3:
1980 sc->samples_per_frame = 6;
1981 sc->bytes_per_frame = 2 * st->codecpar->channels;
1983 case AV_CODEC_ID_MACE6:
1984 sc->samples_per_frame = 6;
1985 sc->bytes_per_frame = 1 * st->codecpar->channels;
1987 case AV_CODEC_ID_ADPCM_IMA_QT:
1988 sc->samples_per_frame = 64;
1989 sc->bytes_per_frame = 34 * st->codecpar->channels;
1991 case AV_CODEC_ID_GSM:
1992 sc->samples_per_frame = 160;
1993 sc->bytes_per_frame = 33;
1999 bits_per_sample = av_get_bits_per_sample(st->codecpar->codec_id);
2000 if (bits_per_sample) {
2001 st->codecpar->bits_per_coded_sample = bits_per_sample;
2002 sc->sample_size = (bits_per_sample >> 3) * st->codecpar->channels;
2006 static void mov_parse_stsd_subtitle(MOVContext *c, AVIOContext *pb,
2007 AVStream *st, MOVStreamContext *sc,
2010 // ttxt stsd contains display flags, justification, background
2011 // color, fonts, and default styles, so fake an atom to read it
2012 MOVAtom fake_atom = { .size = size };
2013 // mp4s contains a regular esds atom
2014 if (st->codecpar->codec_tag != AV_RL32("mp4s"))
2015 mov_read_glbl(c, pb, fake_atom);
2016 st->codecpar->width = sc->width;
2017 st->codecpar->height = sc->height;
2020 static uint32_t yuv_to_rgba(uint32_t ycbcr)
2025 y = (ycbcr >> 16) & 0xFF;
2026 cr = (ycbcr >> 8) & 0xFF;
2029 b = av_clip_uint8((1164 * (y - 16) + 2018 * (cb - 128)) / 1000);
2030 g = av_clip_uint8((1164 * (y - 16) - 813 * (cr - 128) - 391 * (cb - 128)) / 1000);
2031 r = av_clip_uint8((1164 * (y - 16) + 1596 * (cr - 128) ) / 1000);
2033 return (r << 16) | (g << 8) | b;
2036 static int mov_rewrite_dvd_sub_extradata(AVStream *st)
2038 char buf[256] = {0};
2039 uint8_t *src = st->codecpar->extradata;
2042 if (st->codecpar->extradata_size != 64)
2045 if (st->codecpar->width > 0 && st->codecpar->height > 0)
2046 snprintf(buf, sizeof(buf), "size: %dx%d\n",
2047 st->codecpar->width, st->codecpar->height);
2048 av_strlcat(buf, "palette: ", sizeof(buf));
2050 for (i = 0; i < 16; i++) {
2051 uint32_t yuv = AV_RB32(src + i * 4);
2052 uint32_t rgba = yuv_to_rgba(yuv);
2054 av_strlcatf(buf, sizeof(buf), "%06"PRIx32"%s", rgba, i != 15 ? ", " : "");
2057 if (av_strlcat(buf, "\n", sizeof(buf)) >= sizeof(buf))
2060 av_freep(&st->codecpar->extradata);
2061 st->codecpar->extradata_size = 0;
2062 st->codecpar->extradata = av_mallocz(strlen(buf) + AV_INPUT_BUFFER_PADDING_SIZE);
2063 if (!st->codecpar->extradata)
2064 return AVERROR(ENOMEM);
2065 st->codecpar->extradata_size = strlen(buf);
2066 memcpy(st->codecpar->extradata, buf, st->codecpar->extradata_size);
2071 static int mov_parse_stsd_data(MOVContext *c, AVIOContext *pb,
2072 AVStream *st, MOVStreamContext *sc,
2077 if (st->codecpar->codec_tag == MKTAG('t','m','c','d')) {
2078 if ((int)size != size)
2079 return AVERROR(ENOMEM);
2081 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
2085 MOVStreamContext *tmcd_ctx = st->priv_data;
2087 val = AV_RB32(st->codecpar->extradata + 4);
2088 tmcd_ctx->tmcd_flags = val;
2089 st->avg_frame_rate.num = st->codecpar->extradata[16]; /* number of frame */
2090 st->avg_frame_rate.den = 1;
2091 #if FF_API_LAVF_AVCTX
2092 FF_DISABLE_DEPRECATION_WARNINGS
2093 st->codec->time_base = av_inv_q(st->avg_frame_rate);
2094 FF_ENABLE_DEPRECATION_WARNINGS
2096 /* adjust for per frame dur in counter mode */
2097 if (tmcd_ctx->tmcd_flags & 0x0008) {
2098 int timescale = AV_RB32(st->codecpar->extradata + 8);
2099 int framedur = AV_RB32(st->codecpar->extradata + 12);
2100 st->avg_frame_rate.num *= timescale;
2101 st->avg_frame_rate.den *= framedur;
2102 #if FF_API_LAVF_AVCTX
2103 FF_DISABLE_DEPRECATION_WARNINGS
2104 st->codec->time_base.den *= timescale;
2105 st->codec->time_base.num *= framedur;
2106 FF_ENABLE_DEPRECATION_WARNINGS
2110 uint32_t len = AV_RB32(st->codecpar->extradata + 18); /* name atom length */
2111 uint32_t format = AV_RB32(st->codecpar->extradata + 22);
2112 if (format == AV_RB32("name") && (int64_t)size >= (int64_t)len + 18) {
2113 uint16_t str_size = AV_RB16(st->codecpar->extradata + 26); /* string length */
2114 if (str_size > 0 && size >= (int)str_size + 26) {
2115 char *reel_name = av_malloc(str_size + 1);
2117 return AVERROR(ENOMEM);
2118 memcpy(reel_name, st->codecpar->extradata + 30, str_size);
2119 reel_name[str_size] = 0; /* Add null terminator */
2120 /* don't add reel_name if emtpy string */
2121 if (*reel_name == 0) {
2124 av_dict_set(&st->metadata, "reel_name", reel_name, AV_DICT_DONT_STRDUP_VAL);
2131 /* other codec type, just skip (rtp, mp4s ...) */
2132 avio_skip(pb, size);
2137 static int mov_finalize_stsd_codec(MOVContext *c, AVIOContext *pb,
2138 AVStream *st, MOVStreamContext *sc)
2140 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
2141 !st->codecpar->sample_rate && sc->time_scale > 1)
2142 st->codecpar->sample_rate = sc->time_scale;
2144 /* special codec parameters handling */
2145 switch (st->codecpar->codec_id) {
2146 #if CONFIG_DV_DEMUXER
2147 case AV_CODEC_ID_DVAUDIO:
2148 c->dv_fctx = avformat_alloc_context();
2150 av_log(c->fc, AV_LOG_ERROR, "dv demux context alloc error\n");
2151 return AVERROR(ENOMEM);
2153 c->dv_demux = avpriv_dv_init_demux(c->dv_fctx);
2155 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
2156 return AVERROR(ENOMEM);
2158 sc->dv_audio_container = 1;
2159 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16LE;
2162 /* no ifdef since parameters are always those */
2163 case AV_CODEC_ID_QCELP:
2164 st->codecpar->channels = 1;
2165 // force sample rate for qcelp when not stored in mov
2166 if (st->codecpar->codec_tag != MKTAG('Q','c','l','p'))
2167 st->codecpar->sample_rate = 8000;
2168 // FIXME: Why is the following needed for some files?
2169 sc->samples_per_frame = 160;
2170 if (!sc->bytes_per_frame)
2171 sc->bytes_per_frame = 35;
2173 case AV_CODEC_ID_AMR_NB:
2174 st->codecpar->channels = 1;
2175 /* force sample rate for amr, stsd in 3gp does not store sample rate */
2176 st->codecpar->sample_rate = 8000;
2178 case AV_CODEC_ID_AMR_WB:
2179 st->codecpar->channels = 1;
2180 st->codecpar->sample_rate = 16000;
2182 case AV_CODEC_ID_MP2:
2183 case AV_CODEC_ID_MP3:
2184 /* force type after stsd for m1a hdlr */
2185 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2187 case AV_CODEC_ID_GSM:
2188 case AV_CODEC_ID_ADPCM_MS:
2189 case AV_CODEC_ID_ADPCM_IMA_WAV:
2190 case AV_CODEC_ID_ILBC:
2191 case AV_CODEC_ID_MACE3:
2192 case AV_CODEC_ID_MACE6:
2193 case AV_CODEC_ID_QDM2:
2194 st->codecpar->block_align = sc->bytes_per_frame;
2196 case AV_CODEC_ID_ALAC:
2197 if (st->codecpar->extradata_size == 36) {
2198 st->codecpar->channels = AV_RB8 (st->codecpar->extradata + 21);
2199 st->codecpar->sample_rate = AV_RB32(st->codecpar->extradata + 32);
2202 case AV_CODEC_ID_AC3:
2203 case AV_CODEC_ID_EAC3:
2204 case AV_CODEC_ID_MPEG1VIDEO:
2205 case AV_CODEC_ID_VC1:
2206 case AV_CODEC_ID_VP9:
2207 st->need_parsing = AVSTREAM_PARSE_FULL;
2215 static int mov_skip_multiple_stsd(MOVContext *c, AVIOContext *pb,
2216 int codec_tag, int format,
2219 int video_codec_id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2222 (codec_tag != format &&
2223 // AVID 1:1 samples with differing data format and codec tag exist
2224 (codec_tag != AV_RL32("AV1x") || format != AV_RL32("AVup")) &&
2225 // prores is allowed to have differing data format and codec tag
2226 codec_tag != AV_RL32("apcn") && codec_tag != AV_RL32("apch") &&
2228 codec_tag != AV_RL32("dvpp") && codec_tag != AV_RL32("dvcp") &&
2229 (c->fc->video_codec_id ? video_codec_id != c->fc->video_codec_id
2230 : codec_tag != MKTAG('j','p','e','g')))) {
2231 /* Multiple fourcc, we skip JPEG. This is not correct, we should
2232 * export it as a separate AVStream but this needs a few changes
2233 * in the MOV demuxer, patch welcome. */
2235 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
2236 avio_skip(pb, size);
2243 int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries)
2246 MOVStreamContext *sc;
2247 int pseudo_stream_id;
2249 if (c->fc->nb_streams < 1)
2251 st = c->fc->streams[c->fc->nb_streams-1];
2254 for (pseudo_stream_id = 0;
2255 pseudo_stream_id < entries && !pb->eof_reached;
2256 pseudo_stream_id++) {
2257 //Parsing Sample description table
2259 int ret, dref_id = 1;
2260 MOVAtom a = { AV_RL32("stsd") };
2261 int64_t start_pos = avio_tell(pb);
2262 int64_t size = avio_rb32(pb); /* size */
2263 uint32_t format = avio_rl32(pb); /* data format */
2266 avio_rb32(pb); /* reserved */
2267 avio_rb16(pb); /* reserved */
2268 dref_id = avio_rb16(pb);
2269 } else if (size <= 7) {
2270 av_log(c->fc, AV_LOG_ERROR,
2271 "invalid size %"PRId64" in stsd\n", size);
2272 return AVERROR_INVALIDDATA;
2275 if (mov_skip_multiple_stsd(c, pb, st->codecpar->codec_tag, format,
2276 size - (avio_tell(pb) - start_pos)))
2279 sc->pseudo_stream_id = st->codecpar->codec_tag ? -1 : pseudo_stream_id;
2280 sc->dref_id= dref_id;
2281 sc->format = format;
2283 id = mov_codec_id(st, format);
2285 av_log(c->fc, AV_LOG_TRACE,
2286 "size=%"PRId64" 4CC= %c%c%c%c/0x%08x codec_type=%d\n", size,
2287 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff,
2288 (format >> 24) & 0xff, format, st->codecpar->codec_type);
2290 if (st->codecpar->codec_type==AVMEDIA_TYPE_VIDEO) {
2291 st->codecpar->codec_id = id;
2292 mov_parse_stsd_video(c, pb, st, sc);
2293 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_AUDIO) {
2294 st->codecpar->codec_id = id;
2295 mov_parse_stsd_audio(c, pb, st, sc);
2296 if (st->codecpar->sample_rate < 0) {
2297 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
2298 return AVERROR_INVALIDDATA;
2300 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_SUBTITLE){
2301 st->codecpar->codec_id = id;
2302 mov_parse_stsd_subtitle(c, pb, st, sc,
2303 size - (avio_tell(pb) - start_pos));
2305 ret = mov_parse_stsd_data(c, pb, st, sc,
2306 size - (avio_tell(pb) - start_pos));
2310 /* this will read extra atoms at the end (wave, alac, damr, avcC, hvcC, SMI ...) */
2311 a.size = size - (avio_tell(pb) - start_pos);
2313 if ((ret = mov_read_default(c, pb, a)) < 0)
2315 } else if (a.size > 0)
2316 avio_skip(pb, a.size);
2318 if (sc->extradata) {
2319 int extra_size = st->codecpar->extradata_size;
2321 /* Move the current stream extradata to the stream context one. */
2322 sc->extradata_size[pseudo_stream_id] = extra_size;
2323 sc->extradata[pseudo_stream_id] = av_malloc(extra_size + AV_INPUT_BUFFER_PADDING_SIZE);
2324 if (!sc->extradata[pseudo_stream_id])
2325 return AVERROR(ENOMEM);
2326 memcpy(sc->extradata[pseudo_stream_id], st->codecpar->extradata, extra_size);
2327 av_freep(&st->codecpar->extradata);
2328 st->codecpar->extradata_size = 0;
2332 if (pb->eof_reached)
2338 static int mov_read_stsd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2341 MOVStreamContext *sc;
2344 if (c->fc->nb_streams < 1)
2346 st = c->fc->streams[c->fc->nb_streams - 1];
2349 avio_r8(pb); /* version */
2350 avio_rb24(pb); /* flags */
2351 entries = avio_rb32(pb);
2354 av_log(c->fc, AV_LOG_ERROR, "invalid STSD entries %d\n", entries);
2355 return AVERROR_INVALIDDATA;
2358 if (sc->extradata) {
2359 av_log(c->fc, AV_LOG_ERROR, "Duplicate STSD\n");
2360 return AVERROR_INVALIDDATA;
2362 /* Prepare space for hosting multiple extradata. */
2363 sc->extradata = av_mallocz_array(entries, sizeof(*sc->extradata));
2364 sc->extradata_size = av_mallocz_array(entries, sizeof(*sc->extradata_size));
2365 if (!sc->extradata_size || !sc->extradata) {
2366 ret = AVERROR(ENOMEM);
2370 ret = ff_mov_read_stsd_entries(c, pb, entries);
2374 sc->stsd_count = entries;
2376 /* Restore back the primary extradata. */
2377 av_freep(&st->codecpar->extradata);
2378 st->codecpar->extradata_size = sc->extradata_size[0];
2379 if (sc->extradata_size[0]) {
2380 st->codecpar->extradata = av_mallocz(sc->extradata_size[0] + AV_INPUT_BUFFER_PADDING_SIZE);
2381 if (!st->codecpar->extradata)
2382 return AVERROR(ENOMEM);
2383 memcpy(st->codecpar->extradata, sc->extradata[0], sc->extradata_size[0]);
2386 return mov_finalize_stsd_codec(c, pb, st, sc);
2388 av_freep(&sc->extradata);
2389 av_freep(&sc->extradata_size);
2393 static int mov_read_stsc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2396 MOVStreamContext *sc;
2397 unsigned int i, entries;
2399 if (c->fc->nb_streams < 1)
2401 st = c->fc->streams[c->fc->nb_streams-1];
2404 avio_r8(pb); /* version */
2405 avio_rb24(pb); /* flags */
2407 entries = avio_rb32(pb);
2409 av_log(c->fc, AV_LOG_TRACE, "track[%i].stsc.entries = %i\n", c->fc->nb_streams-1, entries);
2414 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSC atom\n");
2415 av_free(sc->stsc_data);
2417 sc->stsc_data = av_malloc_array(entries, sizeof(*sc->stsc_data));
2419 return AVERROR(ENOMEM);
2421 for (i = 0; i < entries && !pb->eof_reached; i++) {
2422 sc->stsc_data[i].first = avio_rb32(pb);
2423 sc->stsc_data[i].count = avio_rb32(pb);
2424 sc->stsc_data[i].id = avio_rb32(pb);
2429 if (pb->eof_reached)
2435 /* Compute the samples value for the stsc entry at the given index. */
2436 static inline int mov_get_stsc_samples(MOVStreamContext *sc, int index)
2440 if (index < sc->stsc_count - 1)
2441 chunk_count = sc->stsc_data[index + 1].first - sc->stsc_data[index].first;
2443 chunk_count = sc->chunk_count - (sc->stsc_data[index].first - 1);
2445 return sc->stsc_data[index].count * chunk_count;
2448 static int mov_read_stps(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2451 MOVStreamContext *sc;
2452 unsigned i, entries;
2454 if (c->fc->nb_streams < 1)
2456 st = c->fc->streams[c->fc->nb_streams-1];
2459 avio_rb32(pb); // version + flags
2461 entries = avio_rb32(pb);
2463 av_log(c->fc, AV_LOG_WARNING, "Duplicated STPS atom\n");
2464 av_free(sc->stps_data);
2466 sc->stps_data = av_malloc_array(entries, sizeof(*sc->stps_data));
2468 return AVERROR(ENOMEM);
2470 for (i = 0; i < entries && !pb->eof_reached; i++) {
2471 sc->stps_data[i] = avio_rb32(pb);
2472 //av_log(c->fc, AV_LOG_TRACE, "stps %d\n", sc->stps_data[i]);
2477 if (pb->eof_reached)
2483 static int mov_read_stss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2486 MOVStreamContext *sc;
2487 unsigned int i, entries;
2489 if (c->fc->nb_streams < 1)
2491 st = c->fc->streams[c->fc->nb_streams-1];
2494 avio_r8(pb); /* version */
2495 avio_rb24(pb); /* flags */
2497 entries = avio_rb32(pb);
2499 av_log(c->fc, AV_LOG_TRACE, "keyframe_count = %d\n", entries);
2503 sc->keyframe_absent = 1;
2504 if (!st->need_parsing && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
2505 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2509 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSS atom\n");
2510 if (entries >= UINT_MAX / sizeof(int))
2511 return AVERROR_INVALIDDATA;
2512 av_freep(&sc->keyframes);
2513 sc->keyframe_count = 0;
2514 sc->keyframes = av_malloc_array(entries, sizeof(*sc->keyframes));
2516 return AVERROR(ENOMEM);
2518 for (i = 0; i < entries && !pb->eof_reached; i++) {
2519 sc->keyframes[i] = avio_rb32(pb);
2520 //av_log(c->fc, AV_LOG_TRACE, "keyframes[]=%d\n", sc->keyframes[i]);
2523 sc->keyframe_count = i;
2525 if (pb->eof_reached)
2531 static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2534 MOVStreamContext *sc;
2535 unsigned int i, entries, sample_size, field_size, num_bytes;
2540 if (c->fc->nb_streams < 1)
2542 st = c->fc->streams[c->fc->nb_streams-1];
2545 avio_r8(pb); /* version */
2546 avio_rb24(pb); /* flags */
2548 if (atom.type == MKTAG('s','t','s','z')) {
2549 sample_size = avio_rb32(pb);
2550 if (!sc->sample_size) /* do not overwrite value computed in stsd */
2551 sc->sample_size = sample_size;
2552 sc->stsz_sample_size = sample_size;
2556 avio_rb24(pb); /* reserved */
2557 field_size = avio_r8(pb);
2559 entries = avio_rb32(pb);
2561 av_log(c->fc, AV_LOG_TRACE, "sample_size = %d sample_count = %d\n", sc->sample_size, entries);
2563 sc->sample_count = entries;
2567 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
2568 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %d\n", field_size);
2569 return AVERROR_INVALIDDATA;
2574 if (entries >= (UINT_MAX - 4) / field_size)
2575 return AVERROR_INVALIDDATA;
2576 if (sc->sample_sizes)
2577 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSZ atom\n");
2578 av_free(sc->sample_sizes);
2579 sc->sample_count = 0;
2580 sc->sample_sizes = av_malloc_array(entries, sizeof(*sc->sample_sizes));
2581 if (!sc->sample_sizes)
2582 return AVERROR(ENOMEM);
2584 num_bytes = (entries*field_size+4)>>3;
2586 buf = av_malloc(num_bytes+AV_INPUT_BUFFER_PADDING_SIZE);
2588 av_freep(&sc->sample_sizes);
2589 return AVERROR(ENOMEM);
2592 ret = ffio_read_size(pb, buf, num_bytes);
2594 av_freep(&sc->sample_sizes);
2599 init_get_bits(&gb, buf, 8*num_bytes);
2601 for (i = 0; i < entries && !pb->eof_reached; i++) {
2602 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
2603 sc->data_size += sc->sample_sizes[i];
2606 sc->sample_count = i;
2610 if (pb->eof_reached)
2616 static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2619 MOVStreamContext *sc;
2620 unsigned int i, entries;
2622 int64_t total_sample_count=0;
2624 if (c->fc->nb_streams < 1)
2626 st = c->fc->streams[c->fc->nb_streams-1];
2629 avio_r8(pb); /* version */
2630 avio_rb24(pb); /* flags */
2631 entries = avio_rb32(pb);
2633 av_log(c->fc, AV_LOG_TRACE, "track[%i].stts.entries = %i\n",
2634 c->fc->nb_streams-1, entries);
2637 av_log(c->fc, AV_LOG_WARNING, "Duplicated STTS atom\n");
2638 av_free(sc->stts_data);
2640 sc->stts_data = av_malloc_array(entries, sizeof(*sc->stts_data));
2642 return AVERROR(ENOMEM);
2644 for (i = 0; i < entries && !pb->eof_reached; i++) {
2645 int sample_duration;
2648 sample_count=avio_rb32(pb);
2649 sample_duration = avio_rb32(pb);
2651 if (sample_count < 0) {
2652 av_log(c->fc, AV_LOG_ERROR, "Invalid sample_count=%d\n", sample_count);
2653 return AVERROR_INVALIDDATA;
2655 sc->stts_data[i].count= sample_count;
2656 sc->stts_data[i].duration= sample_duration;
2658 av_log(c->fc, AV_LOG_TRACE, "sample_count=%d, sample_duration=%d\n",
2659 sample_count, sample_duration);
2663 && sample_count == 1
2664 && total_sample_count > 100
2665 && sample_duration/10 > duration / total_sample_count)
2666 sample_duration = duration / total_sample_count;
2667 duration+=(int64_t)sample_duration*sample_count;
2668 total_sample_count+=sample_count;
2673 sc->duration_for_fps += duration;
2674 sc->nb_frames_for_fps += total_sample_count;
2676 if (pb->eof_reached)
2679 st->nb_frames= total_sample_count;
2681 st->duration= duration;
2682 sc->track_end = duration;
2686 static void mov_update_dts_shift(MOVStreamContext *sc, int duration)
2689 if (duration == INT_MIN) {
2690 av_log(NULL, AV_LOG_WARNING, "mov_update_dts_shift(): dts_shift set to %d\n", INT_MAX);
2693 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
2697 static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2700 MOVStreamContext *sc;
2701 unsigned int i, entries, ctts_count = 0;
2703 if (c->fc->nb_streams < 1)
2705 st = c->fc->streams[c->fc->nb_streams-1];
2708 avio_r8(pb); /* version */
2709 avio_rb24(pb); /* flags */
2710 entries = avio_rb32(pb);
2712 av_log(c->fc, AV_LOG_TRACE, "track[%i].ctts.entries = %i\n", c->fc->nb_streams-1, entries);
2716 if (entries >= UINT_MAX / sizeof(*sc->ctts_data))
2717 return AVERROR_INVALIDDATA;
2718 av_freep(&sc->ctts_data);
2719 sc->ctts_data = av_realloc(NULL, entries * sizeof(*sc->ctts_data));
2721 return AVERROR(ENOMEM);
2723 for (i = 0; i < entries && !pb->eof_reached; i++) {
2724 int count =avio_rb32(pb);
2725 int duration =avio_rb32(pb);
2728 av_log(c->fc, AV_LOG_TRACE,
2729 "ignoring CTTS entry with count=%d duration=%d\n",
2734 sc->ctts_data[ctts_count].count = count;
2735 sc->ctts_data[ctts_count].duration = duration;
2738 av_log(c->fc, AV_LOG_TRACE, "count=%d, duration=%d\n",
2741 if (FFNABS(duration) < -(1<<28) && i+2<entries) {
2742 av_log(c->fc, AV_LOG_WARNING, "CTTS invalid\n");
2743 av_freep(&sc->ctts_data);
2749 mov_update_dts_shift(sc, duration);
2752 sc->ctts_count = ctts_count;
2754 if (pb->eof_reached)
2757 av_log(c->fc, AV_LOG_TRACE, "dts shift %d\n", sc->dts_shift);
2762 static int mov_read_sbgp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2765 MOVStreamContext *sc;
2766 unsigned int i, entries;
2768 uint32_t grouping_type;
2770 if (c->fc->nb_streams < 1)
2772 st = c->fc->streams[c->fc->nb_streams-1];
2775 version = avio_r8(pb); /* version */
2776 avio_rb24(pb); /* flags */
2777 grouping_type = avio_rl32(pb);
2778 if (grouping_type != MKTAG( 'r','a','p',' '))
2779 return 0; /* only support 'rap ' grouping */
2781 avio_rb32(pb); /* grouping_type_parameter */
2783 entries = avio_rb32(pb);
2787 av_log(c->fc, AV_LOG_WARNING, "Duplicated SBGP atom\n");
2788 av_free(sc->rap_group);
2789 sc->rap_group_count = 0;
2790 sc->rap_group = av_malloc_array(entries, sizeof(*sc->rap_group));
2792 return AVERROR(ENOMEM);
2794 for (i = 0; i < entries && !pb->eof_reached; i++) {
2795 sc->rap_group[i].count = avio_rb32(pb); /* sample_count */
2796 sc->rap_group[i].index = avio_rb32(pb); /* group_description_index */
2799 sc->rap_group_count = i;
2801 return pb->eof_reached ? AVERROR_EOF : 0;
2805 * Get ith edit list entry (media time, duration).
2807 static int get_edit_list_entry(MOVContext *mov,
2808 const MOVStreamContext *msc,
2809 unsigned int edit_list_index,
2810 int64_t *edit_list_media_time,
2811 int64_t *edit_list_duration,
2812 int64_t global_timescale)
2814 if (edit_list_index == msc->elst_count) {
2817 *edit_list_media_time = msc->elst_data[edit_list_index].time;
2818 *edit_list_duration = msc->elst_data[edit_list_index].duration;
2820 /* duration is in global timescale units;convert to msc timescale */
2821 if (global_timescale == 0) {
2822 avpriv_request_sample(mov->fc, "Support for mvhd.timescale = 0 with editlists");
2825 *edit_list_duration = av_rescale(*edit_list_duration, msc->time_scale,
2831 * Find the closest previous frame to the timestamp, in e_old index
2832 * entries. Searching for just any frame / just key frames can be controlled by
2833 * last argument 'flag'.
2834 * Returns the index of the entry in st->index_entries if successful,
2837 static int64_t find_prev_closest_index(AVStream *st,
2838 AVIndexEntry *e_old,
2843 AVIndexEntry *e_keep = st->index_entries;
2844 int nb_keep = st->nb_index_entries;
2847 st->index_entries = e_old;
2848 st->nb_index_entries = nb_old;
2849 found = av_index_search_timestamp(st, timestamp, flag | AVSEEK_FLAG_BACKWARD);
2851 /* restore AVStream state*/
2852 st->index_entries = e_keep;
2853 st->nb_index_entries = nb_keep;
2858 * Add index entry with the given values, to the end of st->index_entries.
2859 * Returns the new size st->index_entries if successful, else returns -1.
2861 * This function is similar to ff_add_index_entry in libavformat/utils.c
2862 * except that here we are always unconditionally adding an index entry to
2863 * the end, instead of searching the entries list and skipping the add if
2864 * there is an existing entry with the same timestamp.
2865 * This is needed because the mov_fix_index calls this func with the same
2866 * unincremented timestamp for successive discarded frames.
2868 static int64_t add_index_entry(AVStream *st, int64_t pos, int64_t timestamp,
2869 int size, int distance, int flags)
2871 AVIndexEntry *entries, *ie;
2873 const size_t min_size_needed = (st->nb_index_entries + 1) * sizeof(AVIndexEntry);
2875 // Double the allocation each time, to lower memory fragmentation.
2876 // Another difference from ff_add_index_entry function.
2877 const size_t requested_size =
2878 min_size_needed > st->index_entries_allocated_size ?
2879 FFMAX(min_size_needed, 2 * st->index_entries_allocated_size) :
2882 if((unsigned)st->nb_index_entries + 1 >= UINT_MAX / sizeof(AVIndexEntry))
2885 entries = av_fast_realloc(st->index_entries,
2886 &st->index_entries_allocated_size,
2891 st->index_entries= entries;
2893 index= st->nb_index_entries++;
2894 ie= &entries[index];
2897 ie->timestamp = timestamp;
2898 ie->min_distance= distance;
2905 * Rewrite timestamps of index entries in the range [end_index - frame_duration_buffer_size, end_index)
2906 * by subtracting end_ts successively by the amounts given in frame_duration_buffer.
2908 static void fix_index_entry_timestamps(AVStream* st, int end_index, int64_t end_ts,
2909 int64_t* frame_duration_buffer,
2910 int frame_duration_buffer_size) {
2912 av_assert0(end_index >= 0 && end_index <= st->nb_index_entries);
2913 for (i = 0; i < frame_duration_buffer_size; i++) {
2914 end_ts -= frame_duration_buffer[frame_duration_buffer_size - 1 - i];
2915 st->index_entries[end_index - 1 - i].timestamp = end_ts;
2920 * Append a new ctts entry to ctts_data.
2921 * Returns the new ctts_count if successful, else returns -1.
2923 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
2924 int count, int duration)
2926 MOVStts *ctts_buf_new;
2927 const size_t min_size_needed = (*ctts_count + 1) * sizeof(MOVStts);
2928 const size_t requested_size =
2929 min_size_needed > *allocated_size ?
2930 FFMAX(min_size_needed, 2 * (*allocated_size)) :
2933 if((unsigned)(*ctts_count) + 1 >= UINT_MAX / sizeof(MOVStts))
2936 ctts_buf_new = av_fast_realloc(*ctts_data, allocated_size, requested_size);
2941 *ctts_data = ctts_buf_new;
2943 ctts_buf_new[*ctts_count].count = count;
2944 ctts_buf_new[*ctts_count].duration = duration;
2946 *ctts_count = (*ctts_count) + 1;
2951 * Fix st->index_entries, so that it contains only the entries (and the entries
2952 * which are needed to decode them) that fall in the edit list time ranges.
2953 * Also fixes the timestamps of the index entries to match the timeline
2954 * specified the edit lists.
2956 static void mov_fix_index(MOVContext *mov, AVStream *st)
2958 MOVStreamContext *msc = st->priv_data;
2959 AVIndexEntry *e_old = st->index_entries;
2960 int nb_old = st->nb_index_entries;
2961 const AVIndexEntry *e_old_end = e_old + nb_old;
2962 const AVIndexEntry *current = NULL;
2963 MOVStts *ctts_data_old = msc->ctts_data;
2964 int64_t ctts_index_old = 0;
2965 int64_t ctts_sample_old = 0;
2966 int64_t ctts_count_old = msc->ctts_count;
2967 int64_t edit_list_media_time = 0;
2968 int64_t edit_list_duration = 0;
2969 int64_t frame_duration = 0;
2970 int64_t edit_list_dts_counter = 0;
2971 int64_t edit_list_dts_entry_end = 0;
2972 int64_t edit_list_start_ctts_sample = 0;
2974 int64_t edit_list_index = 0;
2976 int64_t index_ctts_count;
2978 unsigned int ctts_allocated_size = 0;
2979 int64_t start_dts = 0;
2980 int64_t edit_list_media_time_dts = 0;
2981 int64_t edit_list_start_encountered = 0;
2982 int64_t search_timestamp = 0;
2983 int64_t* frame_duration_buffer = NULL;
2984 int num_discarded_begin = 0;
2985 int first_non_zero_audio_edit = -1;
2986 int packet_skip_samples = 0;
2988 if (!msc->elst_data || msc->elst_count <= 0 || nb_old <= 0) {
2991 // Clean AVStream from traces of old index
2992 st->index_entries = NULL;
2993 st->index_entries_allocated_size = 0;
2994 st->nb_index_entries = 0;
2996 // Clean ctts fields of MOVStreamContext
2997 msc->ctts_data = NULL;
2998 msc->ctts_count = 0;
2999 msc->ctts_index = 0;
3000 msc->ctts_sample = 0;
3002 // If the dts_shift is positive (in case of negative ctts values in mov),
3003 // then negate the DTS by dts_shift
3004 if (msc->dts_shift > 0)
3005 edit_list_dts_entry_end -= msc->dts_shift;
3007 // Offset the DTS by ctts[0] to make the PTS of the first frame 0
3008 if (ctts_data_old && ctts_count_old > 0) {
3009 edit_list_dts_entry_end -= ctts_data_old[0].duration;
3010 av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by ctts[%d].duration: %d\n", 0, ctts_data_old[0].duration);
3013 start_dts = edit_list_dts_entry_end;
3015 while (get_edit_list_entry(mov, msc, edit_list_index, &edit_list_media_time,
3016 &edit_list_duration, mov->time_scale)) {
3017 av_log(mov->fc, AV_LOG_DEBUG, "Processing st: %d, edit list %"PRId64" - media time: %"PRId64", duration: %"PRId64"\n",
3018 st->index, edit_list_index, edit_list_media_time, edit_list_duration);
3020 edit_list_dts_counter = edit_list_dts_entry_end;
3021 edit_list_dts_entry_end += edit_list_duration;
3022 num_discarded_begin = 0;
3023 if (edit_list_media_time == -1) {
3027 // If we encounter a non-negative edit list reset the skip_samples/start_pad fields and set them
3028 // according to the edit list below.
3029 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3030 if (first_non_zero_audio_edit < 0) {
3031 first_non_zero_audio_edit = 1;
3033 first_non_zero_audio_edit = 0;
3036 if (first_non_zero_audio_edit > 0)
3037 st->skip_samples = msc->start_pad = 0;
3040 //find closest previous key frame
3041 edit_list_media_time_dts = edit_list_media_time;
3042 if (msc->dts_shift > 0) {
3043 edit_list_media_time_dts -= msc->dts_shift;
3046 // While reordering frame index according to edit list we must handle properly
3047 // the scenario when edit list entry starts from none key frame.
3048 // We find closest previous key frame and preserve it and consequent frames in index.
3049 // All frames which are outside edit list entry time boundaries will be dropped after decoding.
3050 search_timestamp = edit_list_media_time_dts;
3051 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3052 // Audio decoders like AAC need need a decoder delay samples previous to the current sample,
3053 // to correctly decode this frame. Hence for audio we seek to a frame 1 sec. before the
3054 // edit_list_media_time to cover the decoder delay.
3055 search_timestamp = FFMAX(search_timestamp - msc->time_scale, e_old[0].timestamp);
3058 index = find_prev_closest_index(st, e_old, nb_old, search_timestamp, 0);
3060 av_log(mov->fc, AV_LOG_WARNING,
3061 "st: %d edit list: %"PRId64" Missing key frame while searching for timestamp: %"PRId64"\n",
3062 st->index, edit_list_index, search_timestamp);
3063 index = find_prev_closest_index(st, e_old, nb_old, search_timestamp, AVSEEK_FLAG_ANY);
3066 av_log(mov->fc, AV_LOG_WARNING,
3067 "st: %d edit list %"PRId64" Cannot find an index entry before timestamp: %"PRId64".\n"
3068 "Rounding edit list media time to zero.\n",
3069 st->index, edit_list_index, search_timestamp);
3071 edit_list_media_time = 0;
3074 current = e_old + index;
3077 ctts_sample_old = 0;
3079 // set ctts_index properly for the found key frame
3080 for (index_ctts_count = 0; index_ctts_count < index; index_ctts_count++) {
3081 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3083 if (ctts_data_old[ctts_index_old].count == ctts_sample_old) {
3085 ctts_sample_old = 0;
3090 edit_list_start_ctts_sample = ctts_sample_old;
3092 // Iterate over index and arrange it according to edit list
3093 edit_list_start_encountered = 0;
3094 for (; current < e_old_end; current++, index++) {
3095 // check if frame outside edit list mark it for discard
3096 frame_duration = (current + 1 < e_old_end) ?
3097 ((current + 1)->timestamp - current->timestamp) : edit_list_duration;
3099 flags = current->flags;
3101 // frames (pts) before or after edit list
3102 curr_cts = current->timestamp + msc->dts_shift;
3104 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3105 av_log(mov->fc, AV_LOG_DEBUG, "shifted frame pts, curr_cts: %"PRId64" @ %"PRId64", ctts: %d, ctts_count: %"PRId64"\n",
3106 curr_cts, ctts_index_old, ctts_data_old[ctts_index_old].duration, ctts_count_old);
3107 curr_cts += ctts_data_old[ctts_index_old].duration;
3109 if (ctts_sample_old == ctts_data_old[ctts_index_old].count) {
3110 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3111 &ctts_allocated_size,
3112 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3113 ctts_data_old[ctts_index_old].duration) == -1) {
3114 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3116 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3117 ctts_data_old[ctts_index_old].duration);
3121 ctts_sample_old = 0;
3122 edit_list_start_ctts_sample = 0;
3126 if (curr_cts < edit_list_media_time || curr_cts >= (edit_list_duration + edit_list_media_time)) {
3127 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id != AV_CODEC_ID_VORBIS &&
3128 curr_cts < edit_list_media_time && curr_cts + frame_duration > edit_list_media_time &&
3129 first_non_zero_audio_edit > 0) {
3130 packet_skip_samples = edit_list_media_time - curr_cts;
3131 st->skip_samples += packet_skip_samples;
3133 // Shift the index entry timestamp by packet_skip_samples to be correct.
3134 edit_list_dts_counter -= packet_skip_samples;
3135 if (edit_list_start_encountered == 0) {
3136 edit_list_start_encountered = 1;
3137 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3138 // discarded packets.
3139 if (frame_duration_buffer) {
3140 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3141 frame_duration_buffer, num_discarded_begin);
3142 av_freep(&frame_duration_buffer);
3146 av_log(mov->fc, AV_LOG_DEBUG, "skip %d audio samples from curr_cts: %"PRId64"\n", packet_skip_samples, curr_cts);
3148 flags |= AVINDEX_DISCARD_FRAME;
3149 av_log(mov->fc, AV_LOG_DEBUG, "drop a frame at curr_cts: %"PRId64" @ %"PRId64"\n", curr_cts, index);
3151 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && edit_list_start_encountered == 0) {
3152 num_discarded_begin++;
3153 frame_duration_buffer = av_realloc(frame_duration_buffer,
3154 num_discarded_begin * sizeof(int64_t));
3155 if (!frame_duration_buffer) {
3156 av_log(mov->fc, AV_LOG_ERROR, "Cannot reallocate frame duration buffer\n");
3159 frame_duration_buffer[num_discarded_begin - 1] = frame_duration;
3161 // Increment skip_samples for the first non-zero audio edit list
3162 if (first_non_zero_audio_edit > 0 && st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
3163 st->skip_samples += frame_duration;
3164 msc->start_pad = st->skip_samples;
3168 } else if (edit_list_start_encountered == 0) {
3169 edit_list_start_encountered = 1;
3170 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3171 // discarded packets.
3172 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && frame_duration_buffer) {
3173 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3174 frame_duration_buffer, num_discarded_begin);
3175 av_freep(&frame_duration_buffer);
3179 if (add_index_entry(st, current->pos, edit_list_dts_counter, current->size,
3180 current->min_distance, flags) == -1) {
3181 av_log(mov->fc, AV_LOG_ERROR, "Cannot add index entry\n");
3185 // Only start incrementing DTS in frame_duration amounts, when we encounter a frame in edit list.
3186 if (edit_list_start_encountered > 0) {
3187 edit_list_dts_counter = edit_list_dts_counter + frame_duration;
3190 // Break when found first key frame after edit entry completion
3191 if (((curr_cts + frame_duration) >= (edit_list_duration + edit_list_media_time)) &&
3192 ((flags & AVINDEX_KEYFRAME) || ((st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)))) {
3194 if (ctts_data_old && ctts_sample_old != 0) {
3195 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3196 &ctts_allocated_size,
3197 ctts_sample_old - 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",
3200 ctts_index_old, ctts_sample_old - edit_list_start_ctts_sample,
3201 ctts_data_old[ctts_index_old].duration);
3209 // Update av stream length
3210 st->duration = edit_list_dts_entry_end - start_dts;
3212 // Free the old index and the old CTTS structures
3214 av_free(ctts_data_old);
3217 static void mov_build_index(MOVContext *mov, AVStream *st)
3219 MOVStreamContext *sc = st->priv_data;
3220 int64_t current_offset;
3221 int64_t current_dts = 0;
3222 unsigned int stts_index = 0;
3223 unsigned int stsc_index = 0;
3224 unsigned int stss_index = 0;
3225 unsigned int stps_index = 0;
3227 uint64_t stream_size = 0;
3229 if (sc->elst_count) {
3230 int i, edit_start_index = 0;
3231 int64_t empty_duration = 0; // empty duration of the first edit list entry
3232 int64_t start_time = 0; // start time of the media
3234 for (i = 0; i < sc->elst_count; i++) {
3235 const MOVElst *e = &sc->elst_data[i];
3236 if (i == 0 && e->time == -1) {
3237 /* if empty, the first entry is the start time of the stream
3238 * relative to the presentation itself */
3239 empty_duration = e->duration;
3240 edit_start_index = 1;
3241 } else if (i == edit_start_index && e->time >= 0) {
3242 start_time = e->time;
3246 /* adjust first dts according to edit list */
3247 if ((empty_duration || start_time) && mov->time_scale > 0) {
3249 empty_duration = av_rescale(empty_duration, sc->time_scale, mov->time_scale);
3250 sc->time_offset = start_time - empty_duration;
3254 /* only use old uncompressed audio chunk demuxing when stts specifies it */
3255 if (!(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3256 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
3257 unsigned int current_sample = 0;
3258 unsigned int stts_sample = 0;
3259 unsigned int sample_size;
3260 unsigned int distance = 0;
3261 unsigned int rap_group_index = 0;
3262 unsigned int rap_group_sample = 0;
3263 int64_t last_dts = 0;
3264 int64_t dts_correction = 0;
3265 int rap_group_present = sc->rap_group_count && sc->rap_group;
3266 int key_off = (sc->keyframe_count && sc->keyframes[0] > 0) || (sc->stps_count && sc->stps_data[0] > 0);
3268 current_dts -= sc->dts_shift;
3269 last_dts = current_dts;
3271 if (!sc->sample_count || st->nb_index_entries)
3273 if (sc->sample_count >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3275 if (av_reallocp_array(&st->index_entries,
3276 st->nb_index_entries + sc->sample_count,
3277 sizeof(*st->index_entries)) < 0) {
3278 st->nb_index_entries = 0;
3281 st->index_entries_allocated_size = (st->nb_index_entries + sc->sample_count) * sizeof(*st->index_entries);
3283 for (i = 0; i < sc->chunk_count; i++) {
3284 int64_t next_offset = i+1 < sc->chunk_count ? sc->chunk_offsets[i+1] : INT64_MAX;
3285 current_offset = sc->chunk_offsets[i];
3286 while (stsc_index + 1 < sc->stsc_count &&
3287 i + 1 == sc->stsc_data[stsc_index + 1].first)
3290 if (next_offset > current_offset && sc->sample_size>0 && sc->sample_size < sc->stsz_sample_size &&
3291 sc->stsc_data[stsc_index].count * (int64_t)sc->stsz_sample_size > next_offset - current_offset) {
3292 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too large), ignoring\n", sc->stsz_sample_size);
3293 sc->stsz_sample_size = sc->sample_size;
3295 if (sc->stsz_sample_size>0 && sc->stsz_sample_size < sc->sample_size) {
3296 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too small), ignoring\n", sc->stsz_sample_size);
3297 sc->stsz_sample_size = sc->sample_size;
3300 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
3302 if (current_sample >= sc->sample_count) {
3303 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
3307 if (!sc->keyframe_absent && (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index])) {
3309 if (stss_index + 1 < sc->keyframe_count)
3311 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
3313 if (stps_index + 1 < sc->stps_count)
3316 if (rap_group_present && rap_group_index < sc->rap_group_count) {
3317 if (sc->rap_group[rap_group_index].index > 0)
3319 if (++rap_group_sample == sc->rap_group[rap_group_index].count) {
3320 rap_group_sample = 0;
3324 if (sc->keyframe_absent
3326 && !rap_group_present
3327 && (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO || (i==0 && j==0)))
3331 sample_size = sc->stsz_sample_size > 0 ? sc->stsz_sample_size : sc->sample_sizes[current_sample];
3332 if (sc->pseudo_stream_id == -1 ||
3333 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
3335 if (sample_size > 0x3FFFFFFF) {
3336 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", sample_size);
3339 e = &st->index_entries[st->nb_index_entries++];
3340 e->pos = current_offset;
3341 e->timestamp = current_dts;
3342 e->size = sample_size;
3343 e->min_distance = distance;
3344 e->flags = keyframe ? AVINDEX_KEYFRAME : 0;
3345 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
3346 "size %d, distance %d, keyframe %d\n", st->index, current_sample,
3347 current_offset, current_dts, sample_size, distance, keyframe);
3348 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries < 100)
3349 ff_rfps_add_frame(mov->fc, st, current_dts);
3352 current_offset += sample_size;
3353 stream_size += sample_size;
3355 /* A negative sample duration is invalid based on the spec,
3356 * but some samples need it to correct the DTS. */
3357 if (sc->stts_data[stts_index].duration < 0) {
3358 av_log(mov->fc, AV_LOG_WARNING,
3359 "Invalid SampleDelta %d in STTS, at %d st:%d\n",
3360 sc->stts_data[stts_index].duration, stts_index,
3362 dts_correction += sc->stts_data[stts_index].duration - 1;
3363 sc->stts_data[stts_index].duration = 1;
3365 current_dts += sc->stts_data[stts_index].duration;
3366 if (!dts_correction || current_dts + dts_correction > last_dts) {
3367 current_dts += dts_correction;
3370 /* Avoid creating non-monotonous DTS */
3371 dts_correction += current_dts - last_dts - 1;
3372 current_dts = last_dts + 1;
3374 last_dts = current_dts;
3378 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
3384 if (st->duration > 0)
3385 st->codecpar->bit_rate = stream_size*8*sc->time_scale/st->duration;
3387 unsigned chunk_samples, total = 0;
3389 // compute total chunk count
3390 for (i = 0; i < sc->stsc_count; i++) {
3391 unsigned count, chunk_count;
3393 chunk_samples = sc->stsc_data[i].count;
3394 if (i != sc->stsc_count - 1 &&
3395 sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
3396 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
3400 if (sc->samples_per_frame >= 160) { // gsm
3401 count = chunk_samples / sc->samples_per_frame;
3402 } else if (sc->samples_per_frame > 1) {
3403 unsigned samples = (1024/sc->samples_per_frame)*sc->samples_per_frame;
3404 count = (chunk_samples+samples-1) / samples;
3406 count = (chunk_samples+1023) / 1024;
3409 if (i < sc->stsc_count - 1)
3410 chunk_count = sc->stsc_data[i+1].first - sc->stsc_data[i].first;
3412 chunk_count = sc->chunk_count - (sc->stsc_data[i].first - 1);
3413 total += chunk_count * count;
3416 av_log(mov->fc, AV_LOG_TRACE, "chunk count %d\n", total);
3417 if (total >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3419 if (av_reallocp_array(&st->index_entries,
3420 st->nb_index_entries + total,
3421 sizeof(*st->index_entries)) < 0) {
3422 st->nb_index_entries = 0;
3425 st->index_entries_allocated_size = (st->nb_index_entries + total) * sizeof(*st->index_entries);
3428 for (i = 0; i < sc->chunk_count; i++) {
3429 current_offset = sc->chunk_offsets[i];
3430 if (stsc_index + 1 < sc->stsc_count &&
3431 i + 1 == sc->stsc_data[stsc_index + 1].first)
3433 chunk_samples = sc->stsc_data[stsc_index].count;
3435 while (chunk_samples > 0) {
3437 unsigned size, samples;
3439 if (sc->samples_per_frame > 1 && !sc->bytes_per_frame) {
3440 avpriv_request_sample(mov->fc,
3441 "Zero bytes per frame, but %d samples per frame",
3442 sc->samples_per_frame);
3446 if (sc->samples_per_frame >= 160) { // gsm
3447 samples = sc->samples_per_frame;
3448 size = sc->bytes_per_frame;
3450 if (sc->samples_per_frame > 1) {
3451 samples = FFMIN((1024 / sc->samples_per_frame)*
3452 sc->samples_per_frame, chunk_samples);
3453 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
3455 samples = FFMIN(1024, chunk_samples);
3456 size = samples * sc->sample_size;
3460 if (st->nb_index_entries >= total) {
3461 av_log(mov->fc, AV_LOG_ERROR, "wrong chunk count %d\n", total);
3464 if (size > 0x3FFFFFFF) {
3465 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", size);
3468 e = &st->index_entries[st->nb_index_entries++];
3469 e->pos = current_offset;
3470 e->timestamp = current_dts;
3472 e->min_distance = 0;
3473 e->flags = AVINDEX_KEYFRAME;
3474 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, chunk %d, offset %"PRIx64", dts %"PRId64", "
3475 "size %d, duration %d\n", st->index, i, current_offset, current_dts,
3478 current_offset += size;
3479 current_dts += samples;
3480 chunk_samples -= samples;
3485 // Fix index according to edit lists.
3486 mov_fix_index(mov, st);
3489 static int test_same_origin(const char *src, const char *ref) {
3499 av_url_split(src_proto, sizeof(src_proto), src_auth, sizeof(src_auth), src_host, sizeof(src_host), &src_port, NULL, 0, src);
3500 av_url_split(ref_proto, sizeof(ref_proto), ref_auth, sizeof(ref_auth), ref_host, sizeof(ref_host), &ref_port, NULL, 0, ref);
3502 if (strlen(src) == 0) {
3504 } else if (strlen(src_auth) + 1 >= sizeof(src_auth) ||
3505 strlen(ref_auth) + 1 >= sizeof(ref_auth) ||
3506 strlen(src_host) + 1 >= sizeof(src_host) ||
3507 strlen(ref_host) + 1 >= sizeof(ref_host)) {
3509 } else if (strcmp(src_proto, ref_proto) ||
3510 strcmp(src_auth, ref_auth) ||
3511 strcmp(src_host, ref_host) ||
3512 src_port != ref_port) {
3518 static int mov_open_dref(MOVContext *c, AVIOContext **pb, const char *src, MOVDref *ref)
3520 /* try relative path, we do not try the absolute because it can leak information about our
3521 system to an attacker */
3522 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
3523 char filename[1025];
3524 const char *src_path;
3527 /* find a source dir */
3528 src_path = strrchr(src, '/');
3534 /* find a next level down to target */
3535 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
3536 if (ref->path[l] == '/') {
3537 if (i == ref->nlvl_to - 1)
3543 /* compose filename if next level down to target was found */
3544 if (i == ref->nlvl_to - 1 && src_path - src < sizeof(filename)) {
3545 memcpy(filename, src, src_path - src);
3546 filename[src_path - src] = 0;
3548 for (i = 1; i < ref->nlvl_from; i++)
3549 av_strlcat(filename, "../", sizeof(filename));
3551 av_strlcat(filename, ref->path + l + 1, sizeof(filename));
3552 if (!c->use_absolute_path) {
3553 int same_origin = test_same_origin(src, filename);
3556 av_log(c->fc, AV_LOG_ERROR,
3557 "Reference with mismatching origin, %s not tried for security reasons, "
3558 "set demuxer option use_absolute_path to allow it anyway\n",
3560 return AVERROR(ENOENT);
3563 if(strstr(ref->path + l + 1, "..") ||
3564 strstr(ref->path + l + 1, ":") ||
3565 (ref->nlvl_from > 1 && same_origin < 0) ||
3566 (filename[0] == '/' && src_path == src))
3567 return AVERROR(ENOENT);
3570 if (strlen(filename) + 1 == sizeof(filename))
3571 return AVERROR(ENOENT);
3572 if (!c->fc->io_open(c->fc, pb, filename, AVIO_FLAG_READ, NULL))
3575 } else if (c->use_absolute_path) {
3576 av_log(c->fc, AV_LOG_WARNING, "Using absolute path on user request, "
3577 "this is a possible security issue\n");
3578 if (!c->fc->io_open(c->fc, pb, ref->path, AVIO_FLAG_READ, NULL))
3581 av_log(c->fc, AV_LOG_ERROR,
3582 "Absolute path %s not tried for security reasons, "
3583 "set demuxer option use_absolute_path to allow absolute paths\n",
3587 return AVERROR(ENOENT);
3590 static void fix_timescale(MOVContext *c, MOVStreamContext *sc)
3592 if (sc->time_scale <= 0) {
3593 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", sc->ffindex);
3594 sc->time_scale = c->time_scale;
3595 if (sc->time_scale <= 0)
3600 static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3603 MOVStreamContext *sc;
3606 st = avformat_new_stream(c->fc, NULL);
3607 if (!st) return AVERROR(ENOMEM);
3608 st->id = c->fc->nb_streams;
3609 sc = av_mallocz(sizeof(MOVStreamContext));
3610 if (!sc) return AVERROR(ENOMEM);
3613 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
3614 sc->ffindex = st->index;
3615 c->trak_index = st->index;
3617 if ((ret = mov_read_default(c, pb, atom)) < 0)
3623 if (sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
3624 (!sc->sample_size && !sc->sample_count))) {
3625 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
3630 fix_timescale(c, sc);
3632 avpriv_set_pts_info(st, 64, 1, sc->time_scale);
3634 mov_build_index(c, st);
3636 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
3637 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
3638 if (c->enable_drefs) {
3639 if (mov_open_dref(c, &sc->pb, c->fc->filename, dref) < 0)
3640 av_log(c->fc, AV_LOG_ERROR,
3641 "stream %d, error opening alias: path='%s', dir='%s', "
3642 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
3643 st->index, dref->path, dref->dir, dref->filename,
3644 dref->volume, dref->nlvl_from, dref->nlvl_to);
3646 av_log(c->fc, AV_LOG_WARNING,
3647 "Skipped opening external track: "
3648 "stream %d, alias: path='%s', dir='%s', "
3649 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d."
3650 "Set enable_drefs to allow this.\n",
3651 st->index, dref->path, dref->dir, dref->filename,
3652 dref->volume, dref->nlvl_from, dref->nlvl_to);
3656 sc->pb_is_copied = 1;
3659 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
3660 if (!st->sample_aspect_ratio.num && st->codecpar->width && st->codecpar->height &&
3661 sc->height && sc->width &&
3662 (st->codecpar->width != sc->width || st->codecpar->height != sc->height)) {
3663 st->sample_aspect_ratio = av_d2q(((double)st->codecpar->height * sc->width) /
3664 ((double)st->codecpar->width * sc->height), INT_MAX);
3667 #if FF_API_R_FRAME_RATE
3668 if (sc->stts_count == 1 || (sc->stts_count == 2 && sc->stts_data[1].count == 1))
3669 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
3670 sc->time_scale, sc->stts_data[0].duration, INT_MAX);
3674 // done for ai5q, ai52, ai55, ai1q, ai12 and ai15.
3675 if (!st->codecpar->extradata_size && st->codecpar->codec_id == AV_CODEC_ID_H264 &&
3676 TAG_IS_AVCI(st->codecpar->codec_tag)) {
3677 ret = ff_generate_avci_extradata(st);
3682 switch (st->codecpar->codec_id) {
3683 #if CONFIG_H261_DECODER
3684 case AV_CODEC_ID_H261:
3686 #if CONFIG_H263_DECODER
3687 case AV_CODEC_ID_H263:
3689 #if CONFIG_MPEG4_DECODER
3690 case AV_CODEC_ID_MPEG4:
3692 st->codecpar->width = 0; /* let decoder init width/height */
3693 st->codecpar->height= 0;
3697 // If the duration of the mp3 packets is not constant, then they could need a parser
3698 if (st->codecpar->codec_id == AV_CODEC_ID_MP3
3699 && sc->stts_count > 3
3700 && sc->stts_count*10 > st->nb_frames
3701 && sc->time_scale == st->codecpar->sample_rate) {
3702 st->need_parsing = AVSTREAM_PARSE_FULL;
3704 /* Do not need those anymore. */
3705 av_freep(&sc->chunk_offsets);
3706 av_freep(&sc->sample_sizes);
3707 av_freep(&sc->keyframes);
3708 av_freep(&sc->stts_data);
3709 av_freep(&sc->stps_data);
3710 av_freep(&sc->elst_data);
3711 av_freep(&sc->rap_group);
3716 static int mov_read_ilst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3719 c->itunes_metadata = 1;
3720 ret = mov_read_default(c, pb, atom);
3721 c->itunes_metadata = 0;
3725 static int mov_read_keys(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3734 count = avio_rb32(pb);
3735 if (count > UINT_MAX / sizeof(*c->meta_keys) - 1) {
3736 av_log(c->fc, AV_LOG_ERROR,
3737 "The 'keys' atom with the invalid key count: %d\n", count);
3738 return AVERROR_INVALIDDATA;
3741 c->meta_keys_count = count + 1;
3742 c->meta_keys = av_mallocz(c->meta_keys_count * sizeof(*c->meta_keys));
3744 return AVERROR(ENOMEM);
3746 for (i = 1; i <= count; ++i) {
3747 uint32_t key_size = avio_rb32(pb);
3748 uint32_t type = avio_rl32(pb);
3750 av_log(c->fc, AV_LOG_ERROR,
3751 "The key# %d in meta has invalid size: %d\n", i, key_size);
3752 return AVERROR_INVALIDDATA;
3755 if (type != MKTAG('m','d','t','a')) {
3756 avio_skip(pb, key_size);
3758 c->meta_keys[i] = av_mallocz(key_size + 1);
3759 if (!c->meta_keys[i])
3760 return AVERROR(ENOMEM);
3761 avio_read(pb, c->meta_keys[i], key_size);
3767 static int mov_read_custom(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3769 int64_t end = avio_tell(pb) + atom.size;
3770 uint8_t *key = NULL, *val = NULL, *mean = NULL;
3774 MOVStreamContext *sc;
3776 if (c->fc->nb_streams < 1)
3778 st = c->fc->streams[c->fc->nb_streams-1];
3781 for (i = 0; i < 3; i++) {
3785 if (end - avio_tell(pb) <= 12)
3788 len = avio_rb32(pb);
3789 tag = avio_rl32(pb);
3790 avio_skip(pb, 4); // flags
3792 if (len < 12 || len - 12 > end - avio_tell(pb))
3796 if (tag == MKTAG('m', 'e', 'a', 'n'))
3798 else if (tag == MKTAG('n', 'a', 'm', 'e'))
3800 else if (tag == MKTAG('d', 'a', 't', 'a') && len > 4) {
3807 *p = av_malloc(len + 1);
3810 ret = ffio_read_size(pb, *p, len);
3818 if (mean && key && val) {
3819 if (strcmp(key, "iTunSMPB") == 0) {
3820 int priming, remainder, samples;
3821 if(sscanf(val, "%*X %X %X %X", &priming, &remainder, &samples) == 3){
3822 if(priming>0 && priming<16384)
3823 sc->start_pad = priming;
3826 if (strcmp(key, "cdec") != 0) {
3827 av_dict_set(&c->fc->metadata, key, val,
3828 AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL);
3832 av_log(c->fc, AV_LOG_VERBOSE,
3833 "Unhandled or malformed custom metadata of size %"PRId64"\n", atom.size);
3836 avio_seek(pb, end, SEEK_SET);
3843 static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3845 while (atom.size > 8) {
3846 uint32_t tag = avio_rl32(pb);
3848 if (tag == MKTAG('h','d','l','r')) {
3849 avio_seek(pb, -8, SEEK_CUR);
3851 return mov_read_default(c, pb, atom);
3857 // return 1 when matrix is identity, 0 otherwise
3858 #define IS_MATRIX_IDENT(matrix) \
3859 ( (matrix)[0][0] == (1 << 16) && \
3860 (matrix)[1][1] == (1 << 16) && \
3861 (matrix)[2][2] == (1 << 30) && \
3862 !(matrix)[0][1] && !(matrix)[0][2] && \
3863 !(matrix)[1][0] && !(matrix)[1][2] && \
3864 !(matrix)[2][0] && !(matrix)[2][1])
3866 static int mov_read_tkhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3871 int display_matrix[3][3];
3872 int res_display_matrix[3][3] = { { 0 } };
3874 MOVStreamContext *sc;
3878 if (c->fc->nb_streams < 1)
3880 st = c->fc->streams[c->fc->nb_streams-1];
3883 version = avio_r8(pb);
3884 flags = avio_rb24(pb);
3885 st->disposition |= (flags & MOV_TKHD_FLAG_ENABLED) ? AV_DISPOSITION_DEFAULT : 0;
3891 avio_rb32(pb); /* creation time */
3892 avio_rb32(pb); /* modification time */
3894 st->id = (int)avio_rb32(pb); /* track id (NOT 0 !)*/
3895 avio_rb32(pb); /* reserved */
3897 /* highlevel (considering edits) duration in movie timebase */
3898 (version == 1) ? avio_rb64(pb) : avio_rb32(pb);
3899 avio_rb32(pb); /* reserved */
3900 avio_rb32(pb); /* reserved */
3902 avio_rb16(pb); /* layer */
3903 avio_rb16(pb); /* alternate group */
3904 avio_rb16(pb); /* volume */
3905 avio_rb16(pb); /* reserved */
3907 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
3908 // they're kept in fixed point format through all calculations
3909 // save u,v,z to store the whole matrix in the AV_PKT_DATA_DISPLAYMATRIX
3910 // side data, but the scale factor is not needed to calculate aspect ratio
3911 for (i = 0; i < 3; i++) {
3912 display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
3913 display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
3914 display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
3917 width = avio_rb32(pb); // 16.16 fixed point track width
3918 height = avio_rb32(pb); // 16.16 fixed point track height
3919 sc->width = width >> 16;
3920 sc->height = height >> 16;
3922 // apply the moov display matrix (after the tkhd one)
3923 for (i = 0; i < 3; i++) {
3924 const int sh[3] = { 16, 16, 30 };
3925 for (j = 0; j < 3; j++) {
3926 for (e = 0; e < 3; e++) {
3927 res_display_matrix[i][j] +=
3928 ((int64_t) display_matrix[i][e] *
3929 c->movie_display_matrix[e][j]) >> sh[e];
3934 // save the matrix when it is not the default identity
3935 if (!IS_MATRIX_IDENT(res_display_matrix)) {
3938 av_freep(&sc->display_matrix);
3939 sc->display_matrix = av_malloc(sizeof(int32_t) * 9);
3940 if (!sc->display_matrix)
3941 return AVERROR(ENOMEM);
3943 for (i = 0; i < 3; i++)
3944 for (j = 0; j < 3; j++)
3945 sc->display_matrix[i * 3 + j] = res_display_matrix[i][j];
3947 rotate = av_display_rotation_get(sc->display_matrix);
3948 if (!isnan(rotate)) {
3949 char rotate_buf[64];
3951 if (rotate < 0) // for backward compatibility
3953 snprintf(rotate_buf, sizeof(rotate_buf), "%g", rotate);
3954 av_dict_set(&st->metadata, "rotate", rotate_buf, 0);
3958 // transform the display width/height according to the matrix
3959 // to keep the same scale, use [width height 1<<16]
3960 if (width && height && sc->display_matrix) {
3961 double disp_transform[2];
3963 for (i = 0; i < 2; i++)
3964 disp_transform[i] = hypot(sc->display_matrix[0 + i],
3965 sc->display_matrix[3 + i]);
3967 if (disp_transform[0] > 0 && disp_transform[1] > 0 &&
3968 disp_transform[0] < (1<<24) && disp_transform[1] < (1<<24) &&
3969 fabs((disp_transform[0] / disp_transform[1]) - 1.0) > 0.01)
3970 st->sample_aspect_ratio = av_d2q(
3971 disp_transform[0] / disp_transform[1],
3977 static int mov_read_tfhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3979 MOVFragment *frag = &c->fragment;
3980 MOVTrackExt *trex = NULL;
3981 MOVFragmentIndex* index = NULL;
3982 int flags, track_id, i, found = 0;
3984 avio_r8(pb); /* version */
3985 flags = avio_rb24(pb);
3987 track_id = avio_rb32(pb);
3989 return AVERROR_INVALIDDATA;
3990 frag->track_id = track_id;
3991 for (i = 0; i < c->trex_count; i++)
3992 if (c->trex_data[i].track_id == frag->track_id) {
3993 trex = &c->trex_data[i];
3997 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
3998 return AVERROR_INVALIDDATA;
4001 frag->base_data_offset = flags & MOV_TFHD_BASE_DATA_OFFSET ?
4002 avio_rb64(pb) : flags & MOV_TFHD_DEFAULT_BASE_IS_MOOF ?
4003 frag->moof_offset : frag->implicit_offset;
4004 frag->stsd_id = flags & MOV_TFHD_STSD_ID ? avio_rb32(pb) : trex->stsd_id;
4006 frag->duration = flags & MOV_TFHD_DEFAULT_DURATION ?
4007 avio_rb32(pb) : trex->duration;
4008 frag->size = flags & MOV_TFHD_DEFAULT_SIZE ?
4009 avio_rb32(pb) : trex->size;
4010 frag->flags = flags & MOV_TFHD_DEFAULT_FLAGS ?
4011 avio_rb32(pb) : trex->flags;
4012 frag->time = AV_NOPTS_VALUE;
4013 for (i = 0; i < c->fragment_index_count; i++) {
4015 MOVFragmentIndex* candidate = c->fragment_index_data[i];
4016 if (candidate->track_id == frag->track_id) {
4017 av_log(c->fc, AV_LOG_DEBUG,
4018 "found fragment index for track %u\n", frag->track_id);
4020 for (j = index->current_item; j < index->item_count; j++) {
4021 if (frag->implicit_offset == index->items[j].moof_offset) {
4022 av_log(c->fc, AV_LOG_DEBUG, "found fragment index entry "
4023 "for track %u and moof_offset %"PRId64"\n",
4024 frag->track_id, index->items[j].moof_offset);
4025 frag->time = index->items[j].time;
4026 index->current_item = j + 1;
4035 if (index && !found) {
4036 av_log(c->fc, AV_LOG_DEBUG, "track %u has a fragment index but "
4037 "it doesn't have an (in-order) entry for moof_offset "
4038 "%"PRId64"\n", frag->track_id, frag->implicit_offset);
4040 av_log(c->fc, AV_LOG_TRACE, "frag flags 0x%x\n", frag->flags);
4044 static int mov_read_chap(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4049 num = atom.size / 4;
4050 if (!(new_tracks = av_malloc_array(num, sizeof(int))))
4051 return AVERROR(ENOMEM);
4053 av_free(c->chapter_tracks);
4054 c->chapter_tracks = new_tracks;
4055 c->nb_chapter_tracks = num;
4057 for (i = 0; i < num && !pb->eof_reached; i++)
4058 c->chapter_tracks[i] = avio_rb32(pb);
4063 static int mov_read_trex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4068 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
4069 return AVERROR_INVALIDDATA;
4070 if ((err = av_reallocp_array(&c->trex_data, c->trex_count + 1,
4071 sizeof(*c->trex_data))) < 0) {
4076 c->fc->duration = AV_NOPTS_VALUE; // the duration from mvhd is not representing the whole file when fragments are used.
4078 trex = &c->trex_data[c->trex_count++];
4079 avio_r8(pb); /* version */
4080 avio_rb24(pb); /* flags */
4081 trex->track_id = avio_rb32(pb);
4082 trex->stsd_id = avio_rb32(pb);
4083 trex->duration = avio_rb32(pb);
4084 trex->size = avio_rb32(pb);
4085 trex->flags = avio_rb32(pb);
4089 static int mov_read_tfdt(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4091 MOVFragment *frag = &c->fragment;
4092 AVStream *st = NULL;
4093 MOVStreamContext *sc;
4096 for (i = 0; i < c->fc->nb_streams; i++) {
4097 if (c->fc->streams[i]->id == frag->track_id) {
4098 st = c->fc->streams[i];
4103 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
4104 return AVERROR_INVALIDDATA;
4107 if (sc->pseudo_stream_id + 1 != frag->stsd_id)
4109 version = avio_r8(pb);
4110 avio_rb24(pb); /* flags */
4112 sc->track_end = avio_rb64(pb);
4114 sc->track_end = avio_rb32(pb);
4119 static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4121 MOVFragment *frag = &c->fragment;
4122 AVStream *st = NULL;
4123 MOVStreamContext *sc;
4127 int data_offset = 0;
4128 unsigned entries, first_sample_flags = frag->flags;
4129 int flags, distance, i, err;
4131 for (i = 0; i < c->fc->nb_streams; i++) {
4132 if (c->fc->streams[i]->id == frag->track_id) {
4133 st = c->fc->streams[i];
4138 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
4139 return AVERROR_INVALIDDATA;
4142 if (sc->pseudo_stream_id+1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4144 avio_r8(pb); /* version */
4145 flags = avio_rb24(pb);
4146 entries = avio_rb32(pb);
4147 av_log(c->fc, AV_LOG_TRACE, "flags 0x%x entries %d\n", flags, entries);
4149 /* Always assume the presence of composition time offsets.
4150 * Without this assumption, for instance, we cannot deal with a track in fragmented movies that meet the following.
4151 * 1) in the initial movie, there are no samples.
4152 * 2) in the first movie fragment, there is only one sample without composition time offset.
4153 * 3) in the subsequent movie fragments, there are samples with composition time offset. */
4154 if (!sc->ctts_count && sc->sample_count)
4156 /* Complement ctts table if moov atom doesn't have ctts atom. */
4157 ctts_data = av_realloc(NULL, sizeof(*sc->ctts_data));
4159 return AVERROR(ENOMEM);
4160 sc->ctts_data = ctts_data;
4161 sc->ctts_data[sc->ctts_count].count = sc->sample_count;
4162 sc->ctts_data[sc->ctts_count].duration = 0;
4165 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
4166 return AVERROR_INVALIDDATA;
4167 if ((err = av_reallocp_array(&sc->ctts_data, entries + sc->ctts_count,
4168 sizeof(*sc->ctts_data))) < 0) {
4172 if (flags & MOV_TRUN_DATA_OFFSET) data_offset = avio_rb32(pb);
4173 if (flags & MOV_TRUN_FIRST_SAMPLE_FLAGS) first_sample_flags = avio_rb32(pb);
4174 dts = sc->track_end - sc->time_offset;
4175 offset = frag->base_data_offset + data_offset;
4177 av_log(c->fc, AV_LOG_TRACE, "first sample flags 0x%x\n", first_sample_flags);
4178 for (i = 0; i < entries && !pb->eof_reached; i++) {
4179 unsigned sample_size = frag->size;
4180 int sample_flags = i ? frag->flags : first_sample_flags;
4181 unsigned sample_duration = frag->duration;
4184 if (flags & MOV_TRUN_SAMPLE_DURATION) sample_duration = avio_rb32(pb);
4185 if (flags & MOV_TRUN_SAMPLE_SIZE) sample_size = avio_rb32(pb);
4186 if (flags & MOV_TRUN_SAMPLE_FLAGS) sample_flags = avio_rb32(pb);
4187 sc->ctts_data[sc->ctts_count].count = 1;
4188 sc->ctts_data[sc->ctts_count].duration = (flags & MOV_TRUN_SAMPLE_CTS) ?
4190 mov_update_dts_shift(sc, sc->ctts_data[sc->ctts_count].duration);
4191 if (frag->time != AV_NOPTS_VALUE) {
4192 if (c->use_mfra_for == FF_MOV_FLAG_MFRA_PTS) {
4193 int64_t pts = frag->time;
4194 av_log(c->fc, AV_LOG_DEBUG, "found frag time %"PRId64
4195 " sc->dts_shift %d ctts.duration %d"
4196 " sc->time_offset %"PRId64" flags & MOV_TRUN_SAMPLE_CTS %d\n", pts,
4197 sc->dts_shift, sc->ctts_data[sc->ctts_count].duration,
4198 sc->time_offset, flags & MOV_TRUN_SAMPLE_CTS);
4199 dts = pts - sc->dts_shift;
4200 if (flags & MOV_TRUN_SAMPLE_CTS) {
4201 dts -= sc->ctts_data[sc->ctts_count].duration;
4203 dts -= sc->time_offset;
4205 av_log(c->fc, AV_LOG_DEBUG, "calculated into dts %"PRId64"\n", dts);
4207 dts = frag->time - sc->time_offset;
4208 av_log(c->fc, AV_LOG_DEBUG, "found frag time %"PRId64
4209 ", using it for dts\n", dts);
4211 frag->time = AV_NOPTS_VALUE;
4214 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
4218 !(sample_flags & (MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC |
4219 MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES));
4222 err = av_add_index_entry(st, offset, dts, sample_size, distance,
4223 keyframe ? AVINDEX_KEYFRAME : 0);
4225 av_log(c->fc, AV_LOG_ERROR, "Failed to add index entry\n");
4227 av_log(c->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
4228 "size %d, distance %d, keyframe %d\n", st->index, sc->sample_count+i,
4229 offset, dts, sample_size, distance, keyframe);
4231 dts += sample_duration;
4232 offset += sample_size;
4233 sc->data_size += sample_size;
4234 sc->duration_for_fps += sample_duration;
4235 sc->nb_frames_for_fps ++;
4238 if (pb->eof_reached)
4241 frag->implicit_offset = offset;
4243 sc->track_end = dts + sc->time_offset;
4244 if (st->duration < sc->track_end)
4245 st->duration = sc->track_end;
4250 static int mov_read_sidx(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4252 int64_t offset = avio_tell(pb) + atom.size, pts;
4254 unsigned i, track_id;
4255 AVStream *st = NULL;
4257 MOVStreamContext *sc, *ref_sc;
4258 MOVFragmentIndex *index = NULL;
4259 MOVFragmentIndex **tmp;
4260 AVRational timescale;
4262 version = avio_r8(pb);
4264 avpriv_request_sample(c->fc, "sidx version %u", version);
4268 avio_rb24(pb); // flags
4270 track_id = avio_rb32(pb); // Reference ID
4271 for (i = 0; i < c->fc->nb_streams; i++) {
4272 if (c->fc->streams[i]->id == track_id) {
4273 st = c->fc->streams[i];
4278 av_log(c->fc, AV_LOG_WARNING, "could not find corresponding track id %d\n", track_id);
4284 timescale = av_make_q(1, avio_rb32(pb));
4286 if (timescale.den <= 0) {
4287 av_log(c->fc, AV_LOG_ERROR, "Invalid sidx timescale 1/%d\n", timescale.den);
4288 return AVERROR_INVALIDDATA;
4292 pts = avio_rb32(pb);
4293 offset += avio_rb32(pb);
4295 pts = avio_rb64(pb);
4296 offset += avio_rb64(pb);
4299 avio_rb16(pb); // reserved
4301 index = av_mallocz(sizeof(MOVFragmentIndex));
4303 return AVERROR(ENOMEM);
4305 index->track_id = track_id;
4307 index->item_count = avio_rb16(pb);
4308 index->items = av_mallocz_array(index->item_count, sizeof(MOVFragmentIndexItem));
4310 if (!index->items) {
4312 return AVERROR(ENOMEM);
4315 for (i = 0; i < index->item_count; i++) {
4316 uint32_t size = avio_rb32(pb);
4317 uint32_t duration = avio_rb32(pb);
4318 if (size & 0x80000000) {
4319 avpriv_request_sample(c->fc, "sidx reference_type 1");
4320 av_freep(&index->items);
4322 return AVERROR_PATCHWELCOME;
4324 avio_rb32(pb); // sap_flags
4325 index->items[i].moof_offset = offset;
4326 index->items[i].time = av_rescale_q(pts, st->time_base, timescale);
4331 st->duration = sc->track_end = pts;
4333 tmp = av_realloc_array(c->fragment_index_data,
4334 c->fragment_index_count + 1,
4335 sizeof(MOVFragmentIndex*));
4337 av_freep(&index->items);
4339 return AVERROR(ENOMEM);
4342 c->fragment_index_data = tmp;
4343 c->fragment_index_data[c->fragment_index_count++] = index;
4346 if (offset == avio_size(pb)) {
4347 for (i = 0; i < c->fc->nb_streams; i++) {
4348 if (c->fc->streams[i]->id == c->fragment_index_data[0]->track_id) {
4349 ref_st = c->fc->streams[i];
4350 ref_sc = ref_st->priv_data;
4354 for (i = 0; i < c->fc->nb_streams; i++) {
4355 st = c->fc->streams[i];
4357 if (!sc->has_sidx) {
4358 st->duration = sc->track_end = av_rescale(ref_st->duration, sc->time_scale, ref_sc->time_scale);
4362 c->fragment_index_complete = 1;
4368 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
4369 /* like the files created with Adobe Premiere 5.0, for samples see */
4370 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
4371 static int mov_read_wide(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4376 return 0; /* continue */
4377 if (avio_rb32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
4378 avio_skip(pb, atom.size - 4);
4381 atom.type = avio_rl32(pb);
4383 if (atom.type != MKTAG('m','d','a','t')) {
4384 avio_skip(pb, atom.size);
4387 err = mov_read_mdat(c, pb, atom);
4391 static int mov_read_cmov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4396 uint8_t *moov_data; /* uncompressed data */
4397 long cmov_len, moov_len;
4400 avio_rb32(pb); /* dcom atom */
4401 if (avio_rl32(pb) != MKTAG('d','c','o','m'))
4402 return AVERROR_INVALIDDATA;
4403 if (avio_rl32(pb) != MKTAG('z','l','i','b')) {
4404 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !\n");
4405 return AVERROR_INVALIDDATA;
4407 avio_rb32(pb); /* cmvd atom */
4408 if (avio_rl32(pb) != MKTAG('c','m','v','d'))
4409 return AVERROR_INVALIDDATA;
4410 moov_len = avio_rb32(pb); /* uncompressed size */
4411 cmov_len = atom.size - 6 * 4;
4413 cmov_data = av_malloc(cmov_len);
4415 return AVERROR(ENOMEM);
4416 moov_data = av_malloc(moov_len);
4419 return AVERROR(ENOMEM);
4421 ret = ffio_read_size(pb, cmov_data, cmov_len);
4423 goto free_and_return;
4425 if (uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
4426 goto free_and_return;
4427 if (ffio_init_context(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
4428 goto free_and_return;
4429 ctx.seekable = AVIO_SEEKABLE_NORMAL;
4430 atom.type = MKTAG('m','o','o','v');
4431 atom.size = moov_len;
4432 ret = mov_read_default(c, &ctx, atom);
4438 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
4439 return AVERROR(ENOSYS);
4443 /* edit list atom */
4444 static int mov_read_elst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4446 MOVStreamContext *sc;
4447 int i, edit_count, version;
4449 if (c->fc->nb_streams < 1 || c->ignore_editlist)
4451 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
4453 version = avio_r8(pb); /* version */
4454 avio_rb24(pb); /* flags */
4455 edit_count = avio_rb32(pb); /* entries */
4460 av_log(c->fc, AV_LOG_WARNING, "Duplicated ELST atom\n");
4461 av_free(sc->elst_data);
4463 sc->elst_data = av_malloc_array(edit_count, sizeof(*sc->elst_data));
4465 return AVERROR(ENOMEM);
4467 av_log(c->fc, AV_LOG_TRACE, "track[%i].edit_count = %i\n", c->fc->nb_streams-1, edit_count);
4468 for (i = 0; i < edit_count && !pb->eof_reached; i++) {
4469 MOVElst *e = &sc->elst_data[i];
4472 e->duration = avio_rb64(pb);
4473 e->time = avio_rb64(pb);
4475 e->duration = avio_rb32(pb); /* segment duration */
4476 e->time = (int32_t)avio_rb32(pb); /* media time */
4478 e->rate = avio_rb32(pb) / 65536.0;
4479 av_log(c->fc, AV_LOG_TRACE, "duration=%"PRId64" time=%"PRId64" rate=%f\n",
4480 e->duration, e->time, e->rate);
4482 if (e->time < 0 && e->time != -1 &&
4483 c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
4484 av_log(c->fc, AV_LOG_ERROR, "Track %d, edit %d: Invalid edit list media time=%"PRId64"\n",
4485 c->fc->nb_streams-1, i, e->time);
4486 return AVERROR_INVALIDDATA;
4494 static int mov_read_tmcd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4496 MOVStreamContext *sc;
4498 if (c->fc->nb_streams < 1)
4499 return AVERROR_INVALIDDATA;
4500 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
4501 sc->timecode_track = avio_rb32(pb);
4505 static int mov_read_st3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4508 MOVStreamContext *sc;
4509 enum AVStereo3DType type;
4512 if (c->fc->nb_streams < 1)
4515 st = c->fc->streams[c->fc->nb_streams - 1];
4518 if (atom.size < 5) {
4519 av_log(c->fc, AV_LOG_ERROR, "Empty stereoscopic video box\n");
4520 return AVERROR_INVALIDDATA;
4522 avio_skip(pb, 4); /* version + flags */
4527 type = AV_STEREO3D_2D;
4530 type = AV_STEREO3D_TOPBOTTOM;
4533 type = AV_STEREO3D_SIDEBYSIDE;
4536 av_log(c->fc, AV_LOG_WARNING, "Unknown st3d mode value %d\n", mode);
4540 sc->stereo3d = av_stereo3d_alloc();
4542 return AVERROR(ENOMEM);
4544 sc->stereo3d->type = type;
4548 static int mov_read_sv3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4551 MOVStreamContext *sc;
4553 int32_t yaw, pitch, roll;
4555 enum AVSphericalProjection projection;
4557 if (c->fc->nb_streams < 1)
4560 st = c->fc->streams[c->fc->nb_streams - 1];
4563 if (atom.size < 8) {
4564 av_log(c->fc, AV_LOG_ERROR, "Empty spherical video box\n");
4565 return AVERROR_INVALIDDATA;
4568 size = avio_rb32(pb);
4569 if (size > atom.size)
4570 return AVERROR_INVALIDDATA;
4572 tag = avio_rl32(pb);
4573 if (tag != MKTAG('s','v','h','d')) {
4574 av_log(c->fc, AV_LOG_ERROR, "Missing spherical video header\n");
4577 avio_skip(pb, 4); /* version + flags */
4578 avio_skip(pb, avio_r8(pb)); /* metadata_source */
4580 size = avio_rb32(pb);
4581 if (size > atom.size)
4582 return AVERROR_INVALIDDATA;
4584 tag = avio_rl32(pb);
4585 if (tag != MKTAG('p','r','o','j')) {
4586 av_log(c->fc, AV_LOG_ERROR, "Missing projection box\n");
4590 size = avio_rb32(pb);
4591 if (size > atom.size)
4592 return AVERROR_INVALIDDATA;
4594 tag = avio_rl32(pb);
4595 if (tag != MKTAG('p','r','h','d')) {
4596 av_log(c->fc, AV_LOG_ERROR, "Missing projection header box\n");
4599 avio_skip(pb, 4); /* version + flags */
4601 /* 16.16 fixed point */
4602 yaw = avio_rb32(pb);
4603 pitch = avio_rb32(pb);
4604 roll = avio_rb32(pb);
4606 size = avio_rb32(pb);
4607 if (size > atom.size)
4608 return AVERROR_INVALIDDATA;
4610 tag = avio_rl32(pb);
4611 avio_skip(pb, 4); /* version + flags */
4613 case MKTAG('c','b','m','p'):
4614 projection = AV_SPHERICAL_CUBEMAP;
4616 case MKTAG('e','q','u','i'):
4617 projection = AV_SPHERICAL_EQUIRECTANGULAR;
4620 av_log(c->fc, AV_LOG_ERROR, "Unknown projection type\n");
4624 sc->spherical = av_spherical_alloc(&sc->spherical_size);
4626 return AVERROR(ENOMEM);
4628 sc->spherical->projection = projection;
4630 sc->spherical->yaw = yaw;
4631 sc->spherical->pitch = pitch;
4632 sc->spherical->roll = roll;
4637 static int mov_parse_uuid_spherical(MOVStreamContext *sc, AVIOContext *pb, size_t len)
4640 uint8_t *buffer = av_malloc(len + 1);
4644 return AVERROR(ENOMEM);
4647 ret = ffio_read_size(pb, buffer, len);
4651 /* Check for mandatory keys and values, try to support XML as best-effort */
4652 if (av_stristr(buffer, "<GSpherical:StitchingSoftware>") &&
4653 (val = av_stristr(buffer, "<GSpherical:Spherical>")) &&
4654 av_stristr(val, "true") &&
4655 (val = av_stristr(buffer, "<GSpherical:Stitched>")) &&
4656 av_stristr(val, "true") &&
4657 (val = av_stristr(buffer, "<GSpherical:ProjectionType>")) &&
4658 av_stristr(val, "equirectangular")) {
4659 sc->spherical = av_spherical_alloc(&sc->spherical_size);
4663 sc->spherical->projection = AV_SPHERICAL_EQUIRECTANGULAR;
4665 if (av_stristr(buffer, "<GSpherical:StereoMode>")) {
4666 enum AVStereo3DType mode;
4668 if (av_stristr(buffer, "left-right"))
4669 mode = AV_STEREO3D_SIDEBYSIDE;
4670 else if (av_stristr(buffer, "top-bottom"))
4671 mode = AV_STEREO3D_TOPBOTTOM;
4673 mode = AV_STEREO3D_2D;
4675 sc->stereo3d = av_stereo3d_alloc();
4679 sc->stereo3d->type = mode;
4683 val = av_stristr(buffer, "<GSpherical:InitialViewHeadingDegrees>");
4685 sc->spherical->yaw = strtol(val, NULL, 10) * (1 << 16);
4686 val = av_stristr(buffer, "<GSpherical:InitialViewPitchDegrees>");
4688 sc->spherical->pitch = strtol(val, NULL, 10) * (1 << 16);
4689 val = av_stristr(buffer, "<GSpherical:InitialViewRollDegrees>");
4691 sc->spherical->roll = strtol(val, NULL, 10) * (1 << 16);
4699 static int mov_read_uuid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4702 MOVStreamContext *sc;
4705 static const uint8_t uuid_isml_manifest[] = {
4706 0xa5, 0xd4, 0x0b, 0x30, 0xe8, 0x14, 0x11, 0xdd,
4707 0xba, 0x2f, 0x08, 0x00, 0x20, 0x0c, 0x9a, 0x66
4709 static const uint8_t uuid_xmp[] = {
4710 0xbe, 0x7a, 0xcf, 0xcb, 0x97, 0xa9, 0x42, 0xe8,
4711 0x9c, 0x71, 0x99, 0x94, 0x91, 0xe3, 0xaf, 0xac
4713 static const uint8_t uuid_spherical[] = {
4714 0xff, 0xcc, 0x82, 0x63, 0xf8, 0x55, 0x4a, 0x93,
4715 0x88, 0x14, 0x58, 0x7a, 0x02, 0x52, 0x1f, 0xdd,
4718 if (atom.size < sizeof(uuid) || atom.size == INT64_MAX)
4719 return AVERROR_INVALIDDATA;
4721 if (c->fc->nb_streams < 1)
4723 st = c->fc->streams[c->fc->nb_streams - 1];
4726 ret = avio_read(pb, uuid, sizeof(uuid));
4729 } else if (ret != sizeof(uuid)) {
4730 return AVERROR_INVALIDDATA;
4732 if (!memcmp(uuid, uuid_isml_manifest, sizeof(uuid))) {
4733 uint8_t *buffer, *ptr;
4735 size_t len = atom.size - sizeof(uuid);
4738 return AVERROR_INVALIDDATA;
4740 ret = avio_skip(pb, 4); // zeroes
4743 buffer = av_mallocz(len + 1);
4745 return AVERROR(ENOMEM);
4747 ret = avio_read(pb, buffer, len);
4751 } else if (ret != len) {
4753 return AVERROR_INVALIDDATA;
4757 while ((ptr = av_stristr(ptr, "systemBitrate=\""))) {
4758 ptr += sizeof("systemBitrate=\"") - 1;
4759 c->bitrates_count++;
4760 c->bitrates = av_realloc_f(c->bitrates, c->bitrates_count, sizeof(*c->bitrates));
4762 c->bitrates_count = 0;
4764 return AVERROR(ENOMEM);
4767 ret = strtol(ptr, &endptr, 10);
4768 if (ret < 0 || errno || *endptr != '"') {
4769 c->bitrates[c->bitrates_count - 1] = 0;
4771 c->bitrates[c->bitrates_count - 1] = ret;
4776 } else if (!memcmp(uuid, uuid_xmp, sizeof(uuid))) {
4778 size_t len = atom.size - sizeof(uuid);
4780 buffer = av_mallocz(len + 1);
4782 return AVERROR(ENOMEM);
4784 ret = avio_read(pb, buffer, len);
4788 } else if (ret != len) {
4790 return AVERROR_INVALIDDATA;
4792 if (c->export_xmp) {
4794 av_dict_set(&c->fc->metadata, "xmp", buffer, 0);
4797 } else if (!memcmp(uuid, uuid_spherical, sizeof(uuid))) {
4798 size_t len = atom.size - sizeof(uuid);
4799 ret = mov_parse_uuid_spherical(sc, pb, len);
4803 av_log(c->fc, AV_LOG_WARNING, "Invalid spherical metadata found\n"); }
4808 static int mov_read_free(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4811 uint8_t content[16];
4816 ret = avio_read(pb, content, FFMIN(sizeof(content), atom.size));
4822 && !memcmp(content, "Anevia\x1A\x1A", 8)
4823 && c->use_mfra_for == FF_MOV_FLAG_MFRA_AUTO) {
4824 c->use_mfra_for = FF_MOV_FLAG_MFRA_PTS;
4830 static int mov_read_frma(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4832 uint32_t format = avio_rl32(pb);
4833 MOVStreamContext *sc;
4837 if (c->fc->nb_streams < 1)
4839 st = c->fc->streams[c->fc->nb_streams - 1];
4844 case MKTAG('e','n','c','v'): // encrypted video
4845 case MKTAG('e','n','c','a'): // encrypted audio
4846 id = mov_codec_id(st, format);
4847 if (st->codecpar->codec_id != AV_CODEC_ID_NONE &&
4848 st->codecpar->codec_id != id) {
4849 av_log(c->fc, AV_LOG_WARNING,
4850 "ignoring 'frma' atom of '%.4s', stream has codec id %d\n",
4851 (char*)&format, st->codecpar->codec_id);
4855 st->codecpar->codec_id = id;
4856 sc->format = format;
4860 if (format != sc->format) {
4861 av_log(c->fc, AV_LOG_WARNING,
4862 "ignoring 'frma' atom of '%.4s', stream format is '%.4s'\n",
4863 (char*)&format, (char*)&sc->format);
4871 static int mov_read_senc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4874 MOVStreamContext *sc;
4875 size_t auxiliary_info_size;
4877 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
4880 st = c->fc->streams[c->fc->nb_streams - 1];
4883 if (sc->cenc.aes_ctr) {
4884 av_log(c->fc, AV_LOG_ERROR, "duplicate senc atom\n");
4885 return AVERROR_INVALIDDATA;
4888 avio_r8(pb); /* version */
4889 sc->cenc.use_subsamples = avio_rb24(pb) & 0x02; /* flags */
4891 avio_rb32(pb); /* entries */
4893 if (atom.size < 8) {
4894 av_log(c->fc, AV_LOG_ERROR, "senc atom size %"PRId64" too small\n", atom.size);
4895 return AVERROR_INVALIDDATA;
4898 /* save the auxiliary info as is */
4899 auxiliary_info_size = atom.size - 8;
4901 sc->cenc.auxiliary_info = av_malloc(auxiliary_info_size);
4902 if (!sc->cenc.auxiliary_info) {
4903 return AVERROR(ENOMEM);
4906 sc->cenc.auxiliary_info_end = sc->cenc.auxiliary_info + auxiliary_info_size;
4908 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info;
4910 if (avio_read(pb, sc->cenc.auxiliary_info, auxiliary_info_size) != auxiliary_info_size) {
4911 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info");
4912 return AVERROR_INVALIDDATA;
4915 /* initialize the cipher */
4916 sc->cenc.aes_ctr = av_aes_ctr_alloc();
4917 if (!sc->cenc.aes_ctr) {
4918 return AVERROR(ENOMEM);
4921 return av_aes_ctr_init(sc->cenc.aes_ctr, c->decryption_key);
4924 static int mov_read_saiz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4927 MOVStreamContext *sc;
4929 int atom_header_size;
4932 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
4935 st = c->fc->streams[c->fc->nb_streams - 1];
4938 if (sc->cenc.auxiliary_info_sizes || sc->cenc.auxiliary_info_default_size) {
4939 av_log(c->fc, AV_LOG_ERROR, "duplicate saiz atom\n");
4940 return AVERROR_INVALIDDATA;
4943 atom_header_size = 9;
4945 avio_r8(pb); /* version */
4946 flags = avio_rb24(pb);
4948 if ((flags & 0x01) != 0) {
4949 atom_header_size += 8;
4951 avio_rb32(pb); /* info type */
4952 avio_rb32(pb); /* info type param */
4955 sc->cenc.auxiliary_info_default_size = avio_r8(pb);
4956 avio_rb32(pb); /* entries */
4958 if (atom.size <= atom_header_size) {
4962 /* save the auxiliary info sizes as is */
4963 data_size = atom.size - atom_header_size;
4965 sc->cenc.auxiliary_info_sizes = av_malloc(data_size);
4966 if (!sc->cenc.auxiliary_info_sizes) {
4967 return AVERROR(ENOMEM);
4970 sc->cenc.auxiliary_info_sizes_count = data_size;
4972 if (avio_read(pb, sc->cenc.auxiliary_info_sizes, data_size) != data_size) {
4973 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info sizes");
4974 return AVERROR_INVALIDDATA;
4980 static int mov_read_dfla(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4983 int last, type, size, ret;
4986 if (c->fc->nb_streams < 1)
4988 st = c->fc->streams[c->fc->nb_streams-1];
4990 if ((uint64_t)atom.size > (1<<30) || atom.size < 42)
4991 return AVERROR_INVALIDDATA;
4993 /* Check FlacSpecificBox version. */
4994 if (avio_r8(pb) != 0)
4995 return AVERROR_INVALIDDATA;
4997 avio_rb24(pb); /* Flags */
4999 avio_read(pb, buf, sizeof(buf));
5000 flac_parse_block_header(buf, &last, &type, &size);
5002 if (type != FLAC_METADATA_TYPE_STREAMINFO || size != FLAC_STREAMINFO_SIZE) {
5003 av_log(c->fc, AV_LOG_ERROR, "STREAMINFO must be first FLACMetadataBlock\n");
5004 return AVERROR_INVALIDDATA;
5007 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
5012 av_log(c->fc, AV_LOG_WARNING, "non-STREAMINFO FLACMetadataBlock(s) ignored\n");
5017 static int cenc_filter(MOVContext *c, MOVStreamContext *sc, uint8_t *input, int size)
5019 uint32_t encrypted_bytes;
5020 uint16_t subsample_count;
5021 uint16_t clear_bytes;
5022 uint8_t* input_end = input + size;
5025 if (AES_CTR_IV_SIZE > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5026 av_log(c->fc, AV_LOG_ERROR, "failed to read iv from the auxiliary info\n");
5027 return AVERROR_INVALIDDATA;
5030 av_aes_ctr_set_iv(sc->cenc.aes_ctr, sc->cenc.auxiliary_info_pos);
5031 sc->cenc.auxiliary_info_pos += AES_CTR_IV_SIZE;
5033 if (!sc->cenc.use_subsamples)
5035 /* decrypt the whole packet */
5036 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, size);
5040 /* read the subsample count */
5041 if (sizeof(uint16_t) > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5042 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample count from the auxiliary info\n");
5043 return AVERROR_INVALIDDATA;
5046 subsample_count = AV_RB16(sc->cenc.auxiliary_info_pos);
5047 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
5049 for (; subsample_count > 0; subsample_count--)
5051 if (6 > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
5052 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample from the auxiliary info\n");
5053 return AVERROR_INVALIDDATA;
5056 /* read the number of clear / encrypted bytes */
5057 clear_bytes = AV_RB16(sc->cenc.auxiliary_info_pos);
5058 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
5059 encrypted_bytes = AV_RB32(sc->cenc.auxiliary_info_pos);
5060 sc->cenc.auxiliary_info_pos += sizeof(uint32_t);
5062 if ((uint64_t)clear_bytes + encrypted_bytes > input_end - input) {
5063 av_log(c->fc, AV_LOG_ERROR, "subsample size exceeds the packet size left\n");
5064 return AVERROR_INVALIDDATA;
5067 /* skip the clear bytes */
5068 input += clear_bytes;
5070 /* decrypt the encrypted bytes */
5071 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, encrypted_bytes);
5072 input += encrypted_bytes;
5075 if (input < input_end) {
5076 av_log(c->fc, AV_LOG_ERROR, "leftover packet bytes after subsample processing\n");
5077 return AVERROR_INVALIDDATA;
5083 static int mov_seek_auxiliary_info(AVFormatContext *s, MOVStreamContext *sc)
5085 size_t auxiliary_info_seek_offset = 0;
5088 if (sc->cenc.auxiliary_info_default_size) {
5089 auxiliary_info_seek_offset = (size_t)sc->cenc.auxiliary_info_default_size * sc->current_sample;
5090 } else if (sc->cenc.auxiliary_info_sizes) {
5091 if (sc->current_sample > sc->cenc.auxiliary_info_sizes_count) {
5092 av_log(s, AV_LOG_ERROR, "current sample %d greater than the number of auxiliary info sample sizes %"SIZE_SPECIFIER"\n",
5093 sc->current_sample, sc->cenc.auxiliary_info_sizes_count);
5094 return AVERROR_INVALIDDATA;
5097 for (i = 0; i < sc->current_sample; i++) {
5098 auxiliary_info_seek_offset += sc->cenc.auxiliary_info_sizes[i];
5102 if (auxiliary_info_seek_offset > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info) {
5103 av_log(s, AV_LOG_ERROR, "auxiliary info offset %"SIZE_SPECIFIER" greater than auxiliary info size %"SIZE_SPECIFIER"\n",
5104 auxiliary_info_seek_offset, (size_t)(sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info));
5105 return AVERROR_INVALIDDATA;
5108 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info + auxiliary_info_seek_offset;
5113 static const MOVParseTableEntry mov_default_parse_table[] = {
5114 { MKTAG('A','C','L','R'), mov_read_aclr },
5115 { MKTAG('A','P','R','G'), mov_read_avid },
5116 { MKTAG('A','A','L','P'), mov_read_avid },
5117 { MKTAG('A','R','E','S'), mov_read_ares },
5118 { MKTAG('a','v','s','s'), mov_read_avss },
5119 { MKTAG('c','h','p','l'), mov_read_chpl },
5120 { MKTAG('c','o','6','4'), mov_read_stco },
5121 { MKTAG('c','o','l','r'), mov_read_colr },
5122 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
5123 { MKTAG('d','i','n','f'), mov_read_default },
5124 { MKTAG('D','p','x','E'), mov_read_dpxe },
5125 { MKTAG('d','r','e','f'), mov_read_dref },
5126 { MKTAG('e','d','t','s'), mov_read_default },
5127 { MKTAG('e','l','s','t'), mov_read_elst },
5128 { MKTAG('e','n','d','a'), mov_read_enda },
5129 { MKTAG('f','i','e','l'), mov_read_fiel },
5130 { MKTAG('a','d','r','m'), mov_read_adrm },
5131 { MKTAG('f','t','y','p'), mov_read_ftyp },
5132 { MKTAG('g','l','b','l'), mov_read_glbl },
5133 { MKTAG('h','d','l','r'), mov_read_hdlr },
5134 { MKTAG('i','l','s','t'), mov_read_ilst },
5135 { MKTAG('j','p','2','h'), mov_read_jp2h },
5136 { MKTAG('m','d','a','t'), mov_read_mdat },
5137 { MKTAG('m','d','h','d'), mov_read_mdhd },
5138 { MKTAG('m','d','i','a'), mov_read_default },
5139 { MKTAG('m','e','t','a'), mov_read_meta },
5140 { MKTAG('m','i','n','f'), mov_read_default },
5141 { MKTAG('m','o','o','f'), mov_read_moof },
5142 { MKTAG('m','o','o','v'), mov_read_moov },
5143 { MKTAG('m','v','e','x'), mov_read_default },
5144 { MKTAG('m','v','h','d'), mov_read_mvhd },
5145 { MKTAG('S','M','I',' '), mov_read_svq3 },
5146 { MKTAG('a','l','a','c'), mov_read_alac }, /* alac specific atom */
5147 { MKTAG('a','v','c','C'), mov_read_glbl },
5148 { MKTAG('p','a','s','p'), mov_read_pasp },
5149 { MKTAG('s','i','d','x'), mov_read_sidx },
5150 { MKTAG('s','t','b','l'), mov_read_default },
5151 { MKTAG('s','t','c','o'), mov_read_stco },
5152 { MKTAG('s','t','p','s'), mov_read_stps },
5153 { MKTAG('s','t','r','f'), mov_read_strf },
5154 { MKTAG('s','t','s','c'), mov_read_stsc },
5155 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
5156 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
5157 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
5158 { MKTAG('s','t','t','s'), mov_read_stts },
5159 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
5160 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
5161 { MKTAG('t','f','d','t'), mov_read_tfdt },
5162 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
5163 { MKTAG('t','r','a','k'), mov_read_trak },
5164 { MKTAG('t','r','a','f'), mov_read_default },
5165 { MKTAG('t','r','e','f'), mov_read_default },
5166 { MKTAG('t','m','c','d'), mov_read_tmcd },
5167 { MKTAG('c','h','a','p'), mov_read_chap },
5168 { MKTAG('t','r','e','x'), mov_read_trex },
5169 { MKTAG('t','r','u','n'), mov_read_trun },
5170 { MKTAG('u','d','t','a'), mov_read_default },
5171 { MKTAG('w','a','v','e'), mov_read_wave },
5172 { MKTAG('e','s','d','s'), mov_read_esds },
5173 { MKTAG('d','a','c','3'), mov_read_dac3 }, /* AC-3 info */
5174 { MKTAG('d','e','c','3'), mov_read_dec3 }, /* EAC-3 info */
5175 { MKTAG('d','d','t','s'), mov_read_ddts }, /* DTS audio descriptor */
5176 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
5177 { MKTAG('w','f','e','x'), mov_read_wfex },
5178 { MKTAG('c','m','o','v'), mov_read_cmov },
5179 { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */
5180 { MKTAG('d','v','c','1'), mov_read_dvc1 },
5181 { MKTAG('s','b','g','p'), mov_read_sbgp },
5182 { MKTAG('h','v','c','C'), mov_read_glbl },
5183 { MKTAG('u','u','i','d'), mov_read_uuid },
5184 { MKTAG('C','i','n', 0x8e), mov_read_targa_y216 },
5185 { MKTAG('f','r','e','e'), mov_read_free },
5186 { MKTAG('-','-','-','-'), mov_read_custom },
5187 { MKTAG('s','i','n','f'), mov_read_default },
5188 { MKTAG('f','r','m','a'), mov_read_frma },
5189 { MKTAG('s','e','n','c'), mov_read_senc },
5190 { MKTAG('s','a','i','z'), mov_read_saiz },
5191 { MKTAG('d','f','L','a'), mov_read_dfla },
5192 { MKTAG('s','t','3','d'), mov_read_st3d }, /* stereoscopic 3D video box */
5193 { MKTAG('s','v','3','d'), mov_read_sv3d }, /* spherical video box */
5197 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5199 int64_t total_size = 0;
5203 if (c->atom_depth > 10) {
5204 av_log(c->fc, AV_LOG_ERROR, "Atoms too deeply nested\n");
5205 return AVERROR_INVALIDDATA;
5210 atom.size = INT64_MAX;
5211 while (total_size + 8 <= atom.size && !avio_feof(pb)) {
5212 int (*parse)(MOVContext*, AVIOContext*, MOVAtom) = NULL;
5215 if (atom.size >= 8) {
5216 a.size = avio_rb32(pb);
5217 a.type = avio_rl32(pb);
5218 if (a.type == MKTAG('f','r','e','e') &&
5220 c->fc->strict_std_compliance < FF_COMPLIANCE_STRICT &&
5223 uint32_t *type = (uint32_t *)buf + 1;
5224 if (avio_read(pb, buf, 8) != 8)
5225 return AVERROR_INVALIDDATA;
5226 avio_seek(pb, -8, SEEK_CUR);
5227 if (*type == MKTAG('m','v','h','d') ||
5228 *type == MKTAG('c','m','o','v')) {
5229 av_log(c->fc, AV_LOG_ERROR, "Detected moov in a free atom.\n");
5230 a.type = MKTAG('m','o','o','v');
5233 if (atom.type != MKTAG('r','o','o','t') &&
5234 atom.type != MKTAG('m','o','o','v'))
5236 if (a.type == MKTAG('t','r','a','k') || a.type == MKTAG('m','d','a','t'))
5238 av_log(c->fc, AV_LOG_ERROR, "Broken file, trak/mdat not at top-level\n");
5245 if (a.size == 1 && total_size + 8 <= atom.size) { /* 64 bit extended size */
5246 a.size = avio_rb64(pb) - 8;
5250 av_log(c->fc, AV_LOG_TRACE, "type: %08x '%.4s' parent:'%.4s' sz: %"PRId64" %"PRId64" %"PRId64"\n",
5251 a.type, (char*)&a.type, (char*)&atom.type, a.size, total_size, atom.size);
5253 a.size = atom.size - total_size + 8;
5258 a.size = FFMIN(a.size, atom.size - total_size);
5260 for (i = 0; mov_default_parse_table[i].type; i++)
5261 if (mov_default_parse_table[i].type == a.type) {
5262 parse = mov_default_parse_table[i].parse;
5266 // container is user data
5267 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
5268 atom.type == MKTAG('i','l','s','t')))
5269 parse = mov_read_udta_string;
5271 // Supports parsing the QuickTime Metadata Keys.
5272 // https://developer.apple.com/library/mac/documentation/QuickTime/QTFF/Metadata/Metadata.html
5273 if (!parse && c->found_hdlr_mdta &&
5274 atom.type == MKTAG('m','e','t','a') &&
5275 a.type == MKTAG('k','e','y','s')) {
5276 parse = mov_read_keys;
5279 if (!parse) { /* skip leaf atoms data */
5280 avio_skip(pb, a.size);
5282 int64_t start_pos = avio_tell(pb);
5284 int err = parse(c, pb, a);
5289 if (c->found_moov && c->found_mdat &&
5290 ((!pb->seekable || c->fc->flags & AVFMT_FLAG_IGNIDX || c->fragment_index_complete) ||
5291 start_pos + a.size == avio_size(pb))) {
5292 if (!pb->seekable || c->fc->flags & AVFMT_FLAG_IGNIDX || c->fragment_index_complete)
5293 c->next_root_atom = start_pos + a.size;
5297 left = a.size - avio_tell(pb) + start_pos;
5298 if (left > 0) /* skip garbage at atom end */
5299 avio_skip(pb, left);
5300 else if (left < 0) {
5301 av_log(c->fc, AV_LOG_WARNING,
5302 "overread end of atom '%.4s' by %"PRId64" bytes\n",
5303 (char*)&a.type, -left);
5304 avio_seek(pb, left, SEEK_CUR);
5308 total_size += a.size;
5311 if (total_size < atom.size && atom.size < 0x7ffff)
5312 avio_skip(pb, atom.size - total_size);
5318 static int mov_probe(AVProbeData *p)
5323 int moov_offset = -1;
5325 /* check file header */
5328 /* ignore invalid offset */
5329 if ((offset + 8) > (unsigned int)p->buf_size)
5331 tag = AV_RL32(p->buf + offset + 4);
5333 /* check for obvious tags */
5334 case MKTAG('m','o','o','v'):
5335 moov_offset = offset + 4;
5336 case MKTAG('m','d','a','t'):
5337 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
5338 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
5339 case MKTAG('f','t','y','p'):
5340 if (AV_RB32(p->buf+offset) < 8 &&
5341 (AV_RB32(p->buf+offset) != 1 ||
5342 offset + 12 > (unsigned int)p->buf_size ||
5343 AV_RB64(p->buf+offset + 8) == 0)) {
5344 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
5345 } else if (tag == MKTAG('f','t','y','p') &&
5346 ( AV_RL32(p->buf + offset + 8) == MKTAG('j','p','2',' ')
5347 || AV_RL32(p->buf + offset + 8) == MKTAG('j','p','x',' ')
5349 score = FFMAX(score, 5);
5351 score = AVPROBE_SCORE_MAX;
5353 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5355 /* those are more common words, so rate then a bit less */
5356 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
5357 case MKTAG('w','i','d','e'):
5358 case MKTAG('f','r','e','e'):
5359 case MKTAG('j','u','n','k'):
5360 case MKTAG('p','i','c','t'):
5361 score = FFMAX(score, AVPROBE_SCORE_MAX - 5);
5362 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5364 case MKTAG(0x82,0x82,0x7f,0x7d):
5365 case MKTAG('s','k','i','p'):
5366 case MKTAG('u','u','i','d'):
5367 case MKTAG('p','r','f','l'):
5368 /* if we only find those cause probedata is too small at least rate them */
5369 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
5370 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5373 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5376 if(score > AVPROBE_SCORE_MAX - 50 && moov_offset != -1) {
5377 /* moov atom in the header - we should make sure that this is not a
5378 * MOV-packed MPEG-PS */
5379 offset = moov_offset;
5381 while(offset < (p->buf_size - 16)){ /* Sufficient space */
5382 /* We found an actual hdlr atom */
5383 if(AV_RL32(p->buf + offset ) == MKTAG('h','d','l','r') &&
5384 AV_RL32(p->buf + offset + 8) == MKTAG('m','h','l','r') &&
5385 AV_RL32(p->buf + offset + 12) == MKTAG('M','P','E','G')){
5386 av_log(NULL, AV_LOG_WARNING, "Found media data tag MPEG indicating this is a MOV-packed MPEG-PS.\n");
5387 /* We found a media handler reference atom describing an
5388 * MPEG-PS-in-MOV, return a
5389 * low score to force expanding the probe window until
5390 * mpegps_probe finds what it needs */
5401 // must be done after parsing all trak because there's no order requirement
5402 static void mov_read_chapters(AVFormatContext *s)
5404 MOVContext *mov = s->priv_data;
5406 MOVStreamContext *sc;
5411 for (j = 0; j < mov->nb_chapter_tracks; j++) {
5412 chapter_track = mov->chapter_tracks[j];
5414 for (i = 0; i < s->nb_streams; i++)
5415 if (s->streams[i]->id == chapter_track) {
5420 av_log(s, AV_LOG_ERROR, "Referenced QT chapter track not found\n");
5425 cur_pos = avio_tell(sc->pb);
5427 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
5428 st->disposition |= AV_DISPOSITION_ATTACHED_PIC | AV_DISPOSITION_TIMED_THUMBNAILS;
5429 if (st->nb_index_entries) {
5430 // Retrieve the first frame, if possible
5432 AVIndexEntry *sample = &st->index_entries[0];
5433 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
5434 av_log(s, AV_LOG_ERROR, "Failed to retrieve first frame\n");
5438 if (av_get_packet(sc->pb, &pkt, sample->size) < 0)
5441 st->attached_pic = pkt;
5442 st->attached_pic.stream_index = st->index;
5443 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
5446 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
5447 st->codecpar->codec_id = AV_CODEC_ID_BIN_DATA;
5448 st->discard = AVDISCARD_ALL;
5449 for (i = 0; i < st->nb_index_entries; i++) {
5450 AVIndexEntry *sample = &st->index_entries[i];
5451 int64_t end = i+1 < st->nb_index_entries ? st->index_entries[i+1].timestamp : st->duration;
5456 if (end < sample->timestamp) {
5457 av_log(s, AV_LOG_WARNING, "ignoring stream duration which is shorter than chapters\n");
5458 end = AV_NOPTS_VALUE;
5461 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
5462 av_log(s, AV_LOG_ERROR, "Chapter %d not found in file\n", i);
5466 // the first two bytes are the length of the title
5467 len = avio_rb16(sc->pb);
5468 if (len > sample->size-2)
5470 title_len = 2*len + 1;
5471 if (!(title = av_mallocz(title_len)))
5474 // The samples could theoretically be in any encoding if there's an encd
5475 // atom following, but in practice are only utf-8 or utf-16, distinguished
5476 // instead by the presence of a BOM
5480 ch = avio_rb16(sc->pb);
5482 avio_get_str16be(sc->pb, len, title, title_len);
5483 else if (ch == 0xfffe)
5484 avio_get_str16le(sc->pb, len, title, title_len);
5487 if (len == 1 || len == 2)
5490 avio_get_str(sc->pb, INT_MAX, title + 2, len - 1);
5494 avpriv_new_chapter(s, i, st->time_base, sample->timestamp, end, title);
5499 avio_seek(sc->pb, cur_pos, SEEK_SET);
5503 static int parse_timecode_in_framenum_format(AVFormatContext *s, AVStream *st,
5504 uint32_t value, int flags)
5507 char buf[AV_TIMECODE_STR_SIZE];
5508 AVRational rate = st->avg_frame_rate;
5509 int ret = av_timecode_init(&tc, rate, flags, 0, s);
5512 av_dict_set(&st->metadata, "timecode",
5513 av_timecode_make_string(&tc, buf, value), 0);
5517 static int mov_read_rtmd_track(AVFormatContext *s, AVStream *st)
5519 MOVStreamContext *sc = st->priv_data;
5520 char buf[AV_TIMECODE_STR_SIZE];
5521 int64_t cur_pos = avio_tell(sc->pb);
5522 int hh, mm, ss, ff, drop;
5524 if (!st->nb_index_entries)
5527 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
5528 avio_skip(s->pb, 13);
5529 hh = avio_r8(s->pb);
5530 mm = avio_r8(s->pb);
5531 ss = avio_r8(s->pb);
5532 drop = avio_r8(s->pb);
5533 ff = avio_r8(s->pb);
5534 snprintf(buf, AV_TIMECODE_STR_SIZE, "%02d:%02d:%02d%c%02d",
5535 hh, mm, ss, drop ? ';' : ':', ff);
5536 av_dict_set(&st->metadata, "timecode", buf, 0);
5538 avio_seek(sc->pb, cur_pos, SEEK_SET);
5542 static int mov_read_timecode_track(AVFormatContext *s, AVStream *st)
5544 MOVStreamContext *sc = st->priv_data;
5546 int64_t cur_pos = avio_tell(sc->pb);
5549 if (!st->nb_index_entries)
5552 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
5553 value = avio_rb32(s->pb);
5555 if (sc->tmcd_flags & 0x0001) flags |= AV_TIMECODE_FLAG_DROPFRAME;
5556 if (sc->tmcd_flags & 0x0002) flags |= AV_TIMECODE_FLAG_24HOURSMAX;
5557 if (sc->tmcd_flags & 0x0004) flags |= AV_TIMECODE_FLAG_ALLOWNEGATIVE;
5559 /* Assume Counter flag is set to 1 in tmcd track (even though it is likely
5560 * not the case) and thus assume "frame number format" instead of QT one.
5561 * No sample with tmcd track can be found with a QT timecode at the moment,
5562 * despite what the tmcd track "suggests" (Counter flag set to 0 means QT
5564 parse_timecode_in_framenum_format(s, st, value, flags);
5566 avio_seek(sc->pb, cur_pos, SEEK_SET);
5570 static int mov_read_close(AVFormatContext *s)
5572 MOVContext *mov = s->priv_data;
5575 for (i = 0; i < s->nb_streams; i++) {
5576 AVStream *st = s->streams[i];
5577 MOVStreamContext *sc = st->priv_data;
5582 av_freep(&sc->ctts_data);
5583 for (j = 0; j < sc->drefs_count; j++) {
5584 av_freep(&sc->drefs[j].path);
5585 av_freep(&sc->drefs[j].dir);
5587 av_freep(&sc->drefs);
5589 sc->drefs_count = 0;
5591 if (!sc->pb_is_copied)
5592 ff_format_io_close(s, &sc->pb);
5595 av_freep(&sc->chunk_offsets);
5596 av_freep(&sc->stsc_data);
5597 av_freep(&sc->sample_sizes);
5598 av_freep(&sc->keyframes);
5599 av_freep(&sc->stts_data);
5600 av_freep(&sc->stps_data);
5601 av_freep(&sc->elst_data);
5602 av_freep(&sc->rap_group);
5603 av_freep(&sc->display_matrix);
5606 for (j = 0; j < sc->stsd_count; j++)
5607 av_free(sc->extradata[j]);
5608 av_freep(&sc->extradata);
5609 av_freep(&sc->extradata_size);
5611 av_freep(&sc->cenc.auxiliary_info);
5612 av_freep(&sc->cenc.auxiliary_info_sizes);
5613 av_aes_ctr_free(sc->cenc.aes_ctr);
5615 av_freep(&sc->stereo3d);
5616 av_freep(&sc->spherical);
5619 if (mov->dv_demux) {
5620 avformat_free_context(mov->dv_fctx);
5621 mov->dv_fctx = NULL;
5624 if (mov->meta_keys) {
5625 for (i = 1; i < mov->meta_keys_count; i++) {
5626 av_freep(&mov->meta_keys[i]);
5628 av_freep(&mov->meta_keys);
5631 av_freep(&mov->trex_data);
5632 av_freep(&mov->bitrates);
5634 for (i = 0; i < mov->fragment_index_count; i++) {
5635 MOVFragmentIndex* index = mov->fragment_index_data[i];
5636 av_freep(&index->items);
5637 av_freep(&mov->fragment_index_data[i]);
5639 av_freep(&mov->fragment_index_data);
5641 av_freep(&mov->aes_decrypt);
5646 static int tmcd_is_referenced(AVFormatContext *s, int tmcd_id)
5650 for (i = 0; i < s->nb_streams; i++) {
5651 AVStream *st = s->streams[i];
5652 MOVStreamContext *sc = st->priv_data;
5654 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO &&
5655 sc->timecode_track == tmcd_id)
5661 /* look for a tmcd track not referenced by any video track, and export it globally */
5662 static void export_orphan_timecode(AVFormatContext *s)
5666 for (i = 0; i < s->nb_streams; i++) {
5667 AVStream *st = s->streams[i];
5669 if (st->codecpar->codec_tag == MKTAG('t','m','c','d') &&
5670 !tmcd_is_referenced(s, i + 1)) {
5671 AVDictionaryEntry *tcr = av_dict_get(st->metadata, "timecode", NULL, 0);
5673 av_dict_set(&s->metadata, "timecode", tcr->value, 0);
5680 static int read_tfra(MOVContext *mov, AVIOContext *f)
5682 MOVFragmentIndex* index = NULL;
5683 int version, fieldlength, i, j;
5684 int64_t pos = avio_tell(f);
5685 uint32_t size = avio_rb32(f);
5688 if (avio_rb32(f) != MKBETAG('t', 'f', 'r', 'a')) {
5691 av_log(mov->fc, AV_LOG_VERBOSE, "found tfra\n");
5692 index = av_mallocz(sizeof(MOVFragmentIndex));
5694 return AVERROR(ENOMEM);
5697 tmp = av_realloc_array(mov->fragment_index_data,
5698 mov->fragment_index_count + 1,
5699 sizeof(MOVFragmentIndex*));
5702 return AVERROR(ENOMEM);
5704 mov->fragment_index_data = tmp;
5705 mov->fragment_index_data[mov->fragment_index_count++] = index;
5707 version = avio_r8(f);
5709 index->track_id = avio_rb32(f);
5710 fieldlength = avio_rb32(f);
5711 index->item_count = avio_rb32(f);
5712 index->items = av_mallocz_array(
5713 index->item_count, sizeof(MOVFragmentIndexItem));
5714 if (!index->items) {
5715 index->item_count = 0;
5716 return AVERROR(ENOMEM);
5718 for (i = 0; i < index->item_count; i++) {
5719 int64_t time, offset;
5721 time = avio_rb64(f);
5722 offset = avio_rb64(f);
5724 time = avio_rb32(f);
5725 offset = avio_rb32(f);
5727 index->items[i].time = time;
5728 index->items[i].moof_offset = offset;
5729 for (j = 0; j < ((fieldlength >> 4) & 3) + 1; j++)
5731 for (j = 0; j < ((fieldlength >> 2) & 3) + 1; j++)
5733 for (j = 0; j < ((fieldlength >> 0) & 3) + 1; j++)
5737 avio_seek(f, pos + size, SEEK_SET);
5741 static int mov_read_mfra(MOVContext *c, AVIOContext *f)
5743 int64_t stream_size = avio_size(f);
5744 int64_t original_pos = avio_tell(f);
5748 if ((seek_ret = avio_seek(f, stream_size - 4, SEEK_SET)) < 0) {
5752 mfra_size = avio_rb32(f);
5753 if (mfra_size < 0 || mfra_size > stream_size) {
5754 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (unreasonable size)\n");
5757 if ((seek_ret = avio_seek(f, -mfra_size, SEEK_CUR)) < 0) {
5761 if (avio_rb32(f) != mfra_size) {
5762 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (size mismatch)\n");
5765 if (avio_rb32(f) != MKBETAG('m', 'f', 'r', 'a')) {
5766 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (tag mismatch)\n");
5769 av_log(c->fc, AV_LOG_VERBOSE, "stream has mfra\n");
5771 ret = read_tfra(c, f);
5777 seek_ret = avio_seek(f, original_pos, SEEK_SET);
5779 av_log(c->fc, AV_LOG_ERROR,
5780 "failed to seek back after looking for mfra\n");
5786 static int mov_read_header(AVFormatContext *s)
5788 MOVContext *mov = s->priv_data;
5789 AVIOContext *pb = s->pb;
5791 MOVAtom atom = { AV_RL32("root") };
5794 if (mov->decryption_key_len != 0 && mov->decryption_key_len != AES_CTR_KEY_SIZE) {
5795 av_log(s, AV_LOG_ERROR, "Invalid decryption key len %d expected %d\n",
5796 mov->decryption_key_len, AES_CTR_KEY_SIZE);
5797 return AVERROR(EINVAL);
5801 mov->trak_index = -1;
5802 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
5804 atom.size = avio_size(pb);
5806 atom.size = INT64_MAX;
5808 /* check MOV header */
5810 if (mov->moov_retry)
5811 avio_seek(pb, 0, SEEK_SET);
5812 if ((err = mov_read_default(mov, pb, atom)) < 0) {
5813 av_log(s, AV_LOG_ERROR, "error reading header\n");
5817 } while (pb->seekable && !mov->found_moov && !mov->moov_retry++);
5818 if (!mov->found_moov) {
5819 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
5821 return AVERROR_INVALIDDATA;
5823 av_log(mov->fc, AV_LOG_TRACE, "on_parse_exit_offset=%"PRId64"\n", avio_tell(pb));
5826 if (mov->nb_chapter_tracks > 0 && !mov->ignore_chapters)
5827 mov_read_chapters(s);
5828 for (i = 0; i < s->nb_streams; i++)
5829 if (s->streams[i]->codecpar->codec_tag == AV_RL32("tmcd")) {
5830 mov_read_timecode_track(s, s->streams[i]);
5831 } else if (s->streams[i]->codecpar->codec_tag == AV_RL32("rtmd")) {
5832 mov_read_rtmd_track(s, s->streams[i]);
5836 /* copy timecode metadata from tmcd tracks to the related video streams */
5837 for (i = 0; i < s->nb_streams; i++) {
5838 AVStream *st = s->streams[i];
5839 MOVStreamContext *sc = st->priv_data;
5840 if (sc->timecode_track > 0) {
5841 AVDictionaryEntry *tcr;
5842 int tmcd_st_id = -1;
5844 for (j = 0; j < s->nb_streams; j++)
5845 if (s->streams[j]->id == sc->timecode_track)
5848 if (tmcd_st_id < 0 || tmcd_st_id == i)
5850 tcr = av_dict_get(s->streams[tmcd_st_id]->metadata, "timecode", NULL, 0);
5852 av_dict_set(&st->metadata, "timecode", tcr->value, 0);
5855 export_orphan_timecode(s);
5857 for (i = 0; i < s->nb_streams; i++) {
5858 AVStream *st = s->streams[i];
5859 MOVStreamContext *sc = st->priv_data;
5860 fix_timescale(mov, sc);
5861 if(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id == AV_CODEC_ID_AAC) {
5862 st->skip_samples = sc->start_pad;
5864 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && sc->nb_frames_for_fps > 0 && sc->duration_for_fps > 0)
5865 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
5866 sc->time_scale*(int64_t)sc->nb_frames_for_fps, sc->duration_for_fps, INT_MAX);
5867 if (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE) {
5868 if (st->codecpar->width <= 0 || st->codecpar->height <= 0) {
5869 st->codecpar->width = sc->width;
5870 st->codecpar->height = sc->height;
5872 if (st->codecpar->codec_id == AV_CODEC_ID_DVD_SUBTITLE) {
5873 if ((err = mov_rewrite_dvd_sub_extradata(st)) < 0)
5877 if (mov->handbrake_version &&
5878 mov->handbrake_version <= 1000000*0 + 1000*10 + 2 && // 0.10.2
5879 st->codecpar->codec_id == AV_CODEC_ID_MP3
5881 av_log(s, AV_LOG_VERBOSE, "Forcing full parsing for mp3 stream\n");
5882 st->need_parsing = AVSTREAM_PARSE_FULL;
5886 if (mov->trex_data) {
5887 for (i = 0; i < s->nb_streams; i++) {
5888 AVStream *st = s->streams[i];
5889 MOVStreamContext *sc = st->priv_data;
5890 if (st->duration > 0) {
5891 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
5892 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
5893 sc->data_size, sc->time_scale);
5895 return AVERROR_INVALIDDATA;
5897 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale / st->duration;
5902 if (mov->use_mfra_for > 0) {
5903 for (i = 0; i < s->nb_streams; i++) {
5904 AVStream *st = s->streams[i];
5905 MOVStreamContext *sc = st->priv_data;
5906 if (sc->duration_for_fps > 0) {
5907 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
5908 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
5909 sc->data_size, sc->time_scale);
5911 return AVERROR_INVALIDDATA;
5913 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale /
5914 sc->duration_for_fps;
5919 for (i = 0; i < mov->bitrates_count && i < s->nb_streams; i++) {
5920 if (mov->bitrates[i]) {
5921 s->streams[i]->codecpar->bit_rate = mov->bitrates[i];
5925 ff_rfps_calculate(s);
5927 for (i = 0; i < s->nb_streams; i++) {
5928 AVStream *st = s->streams[i];
5929 MOVStreamContext *sc = st->priv_data;
5931 switch (st->codecpar->codec_type) {
5932 case AVMEDIA_TYPE_AUDIO:
5933 err = ff_replaygain_export(st, s->metadata);
5939 case AVMEDIA_TYPE_VIDEO:
5940 if (sc->display_matrix) {
5941 err = av_stream_add_side_data(st, AV_PKT_DATA_DISPLAYMATRIX, (uint8_t*)sc->display_matrix,
5942 sizeof(int32_t) * 9);
5946 sc->display_matrix = NULL;
5949 err = av_stream_add_side_data(st, AV_PKT_DATA_STEREO3D,
5950 (uint8_t *)sc->stereo3d,
5951 sizeof(*sc->stereo3d));
5955 sc->stereo3d = NULL;
5957 if (sc->spherical) {
5958 err = av_stream_add_side_data(st, AV_PKT_DATA_SPHERICAL,
5959 (uint8_t *)sc->spherical,
5960 sc->spherical_size);
5964 sc->spherical = NULL;
5969 ff_configure_buffers_for_index(s, AV_TIME_BASE);
5974 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
5976 AVIndexEntry *sample = NULL;
5977 int64_t best_dts = INT64_MAX;
5979 for (i = 0; i < s->nb_streams; i++) {
5980 AVStream *avst = s->streams[i];
5981 MOVStreamContext *msc = avst->priv_data;
5982 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
5983 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
5984 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
5985 av_log(s, AV_LOG_TRACE, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
5986 if (!sample || (!s->pb->seekable && current_sample->pos < sample->pos) ||
5988 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
5989 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
5990 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
5991 sample = current_sample;
6000 static int should_retry(AVIOContext *pb, int error_code) {
6001 if (error_code == AVERROR_EOF || avio_feof(pb))
6007 static int mov_switch_root(AVFormatContext *s, int64_t target)
6009 MOVContext *mov = s->priv_data;
6011 int already_read = 0;
6013 if (avio_seek(s->pb, target, SEEK_SET) != target) {
6014 av_log(mov->fc, AV_LOG_ERROR, "root atom offset 0x%"PRIx64": partial file\n", target);
6015 return AVERROR_INVALIDDATA;
6018 mov->next_root_atom = 0;
6020 for (i = 0; i < mov->fragment_index_count; i++) {
6021 MOVFragmentIndex *index = mov->fragment_index_data[i];
6023 for (j = 0; j < index->item_count; j++) {
6024 MOVFragmentIndexItem *item = &index->items[j];
6026 mov->next_root_atom = item->moof_offset;
6027 break; // Advance to next index in outer loop
6028 } else if (item->moof_offset == target) {
6029 index->current_item = FFMIN(j, index->current_item);
6030 if (item->headers_read)
6032 item->headers_read = 1;
6037 index->current_item = 0;
6043 mov->found_mdat = 0;
6045 if (mov_read_default(mov, s->pb, (MOVAtom){ AV_RL32("root"), INT64_MAX }) < 0 ||
6048 av_log(s, AV_LOG_TRACE, "read fragments, offset 0x%"PRIx64"\n", avio_tell(s->pb));
6053 static int mov_change_extradata(MOVStreamContext *sc, AVPacket *pkt)
6055 uint8_t *side, *extradata;
6058 /* Save the current index. */
6059 sc->last_stsd_index = sc->stsc_data[sc->stsc_index].id - 1;
6061 /* Notify the decoder that extradata changed. */
6062 extradata_size = sc->extradata_size[sc->last_stsd_index];
6063 extradata = sc->extradata[sc->last_stsd_index];
6064 if (extradata_size > 0 && extradata) {
6065 side = av_packet_new_side_data(pkt,
6066 AV_PKT_DATA_NEW_EXTRADATA,
6069 return AVERROR(ENOMEM);
6070 memcpy(side, extradata, extradata_size);
6076 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
6078 MOVContext *mov = s->priv_data;
6079 MOVStreamContext *sc;
6080 AVIndexEntry *sample;
6081 AVStream *st = NULL;
6085 sample = mov_find_next_sample(s, &st);
6086 if (!sample || (mov->next_root_atom && sample->pos > mov->next_root_atom)) {
6087 if (!mov->next_root_atom)
6089 if ((ret = mov_switch_root(s, mov->next_root_atom)) < 0)
6094 /* must be done just before reading, to avoid infinite loop on sample */
6095 sc->current_sample++;
6097 if (mov->next_root_atom) {
6098 sample->pos = FFMIN(sample->pos, mov->next_root_atom);
6099 sample->size = FFMIN(sample->size, (mov->next_root_atom - sample->pos));
6102 if (st->discard != AVDISCARD_ALL) {
6103 int64_t ret64 = avio_seek(sc->pb, sample->pos, SEEK_SET);
6104 if (ret64 != sample->pos) {
6105 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
6106 sc->ffindex, sample->pos);
6107 sc->current_sample -= should_retry(sc->pb, ret64);
6108 return AVERROR_INVALIDDATA;
6111 if( st->discard == AVDISCARD_NONKEY && 0==(sample->flags & AVINDEX_KEYFRAME) ) {
6112 av_log(mov->fc, AV_LOG_DEBUG, "Nonkey frame from stream %d discarded due to AVDISCARD_NONKEY\n", sc->ffindex);
6116 ret = av_get_packet(sc->pb, pkt, sample->size);
6118 sc->current_sample -= should_retry(sc->pb, ret);
6121 if (sc->has_palette) {
6124 pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, AVPALETTE_SIZE);
6126 av_log(mov->fc, AV_LOG_ERROR, "Cannot append palette to packet\n");
6128 memcpy(pal, sc->palette, AVPALETTE_SIZE);
6129 sc->has_palette = 0;
6132 #if CONFIG_DV_DEMUXER
6133 if (mov->dv_demux && sc->dv_audio_container) {
6134 avpriv_dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size, pkt->pos);
6135 av_freep(&pkt->data);
6137 ret = avpriv_dv_get_packet(mov->dv_demux, pkt);
6142 if (st->codecpar->codec_id == AV_CODEC_ID_MP3 && !st->need_parsing && pkt->size > 4) {
6143 if (ff_mpa_check_header(AV_RB32(pkt->data)) < 0)
6144 st->need_parsing = AVSTREAM_PARSE_FULL;
6148 pkt->stream_index = sc->ffindex;
6149 pkt->dts = sample->timestamp;
6150 if (sample->flags & AVINDEX_DISCARD_FRAME) {
6151 pkt->flags |= AV_PKT_FLAG_DISCARD;
6153 if (sc->ctts_data && sc->ctts_index < sc->ctts_count) {
6154 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
6155 /* update ctts context */
6157 if (sc->ctts_index < sc->ctts_count &&
6158 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
6160 sc->ctts_sample = 0;
6163 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
6164 st->index_entries[sc->current_sample].timestamp : st->duration;
6165 pkt->duration = next_dts - pkt->dts;
6166 pkt->pts = pkt->dts;
6168 if (st->discard == AVDISCARD_ALL)
6170 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? AV_PKT_FLAG_KEY : 0;
6171 pkt->pos = sample->pos;
6173 /* Multiple stsd handling. */
6174 if (sc->stsc_data) {
6175 /* Keep track of the stsc index for the given sample, then check
6176 * if the stsd index is different from the last used one. */
6178 if (sc->stsc_index < sc->stsc_count - 1 &&
6179 mov_get_stsc_samples(sc, sc->stsc_index) == sc->stsc_sample) {
6181 sc->stsc_sample = 0;
6182 /* Do not check indexes after a switch. */
6183 } else if (sc->stsc_data[sc->stsc_index].id > 0 &&
6184 sc->stsc_data[sc->stsc_index].id - 1 < sc->stsd_count &&
6185 sc->stsc_data[sc->stsc_index].id - 1 != sc->last_stsd_index) {
6186 ret = mov_change_extradata(sc, pkt);
6193 aax_filter(pkt->data, pkt->size, mov);
6195 if (sc->cenc.aes_ctr) {
6196 ret = cenc_filter(mov, sc, pkt->data, pkt->size);
6205 static int mov_seek_fragment(AVFormatContext *s, AVStream *st, int64_t timestamp)
6207 MOVContext *mov = s->priv_data;
6208 MOVStreamContext *sc = st->priv_data;
6211 if (!mov->fragment_index_complete)
6214 for (i = 0; i < mov->fragment_index_count; i++) {
6215 if (mov->fragment_index_data[i]->track_id == st->id || !sc->has_sidx) {
6216 MOVFragmentIndex *index = mov->fragment_index_data[i];
6217 for (j = index->item_count - 1; j >= 0; j--) {
6218 if (index->items[j].time <= timestamp) {
6219 if (index->items[j].headers_read)
6222 return mov_switch_root(s, index->items[j].moof_offset);
6231 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
6233 MOVStreamContext *sc = st->priv_data;
6234 int sample, time_sample;
6237 int ret = mov_seek_fragment(s, st, timestamp);
6241 sample = av_index_search_timestamp(st, timestamp, flags);
6242 av_log(s, AV_LOG_TRACE, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
6243 if (sample < 0 && st->nb_index_entries && timestamp < st->index_entries[0].timestamp)
6245 if (sample < 0) /* not sure what to do */
6246 return AVERROR_INVALIDDATA;
6247 sc->current_sample = sample;
6248 av_log(s, AV_LOG_TRACE, "stream %d, found sample %d\n", st->index, sc->current_sample);
6249 /* adjust ctts index */
6250 if (sc->ctts_data) {
6252 for (i = 0; i < sc->ctts_count; i++) {
6253 int next = time_sample + sc->ctts_data[i].count;
6254 if (next > sc->current_sample) {
6256 sc->ctts_sample = sc->current_sample - time_sample;
6263 /* adjust stsd index */
6265 for (i = 0; i < sc->stsc_count; i++) {
6266 int next = time_sample + mov_get_stsc_samples(sc, i);
6267 if (next > sc->current_sample) {
6269 sc->stsc_sample = sc->current_sample - time_sample;
6275 ret = mov_seek_auxiliary_info(s, sc);
6283 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
6285 MOVContext *mc = s->priv_data;
6290 if (stream_index >= s->nb_streams)
6291 return AVERROR_INVALIDDATA;
6293 st = s->streams[stream_index];
6294 sample = mov_seek_stream(s, st, sample_time, flags);
6298 if (mc->seek_individually) {
6299 /* adjust seek timestamp to found sample timestamp */
6300 int64_t seek_timestamp = st->index_entries[sample].timestamp;
6302 for (i = 0; i < s->nb_streams; i++) {
6304 MOVStreamContext *sc = s->streams[i]->priv_data;
6306 st->skip_samples = (sample_time <= 0) ? sc->start_pad : 0;
6308 if (stream_index == i)
6311 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
6312 mov_seek_stream(s, st, timestamp, flags);
6315 for (i = 0; i < s->nb_streams; i++) {
6316 MOVStreamContext *sc;
6319 sc->current_sample = 0;
6322 MOVStreamContext *sc;
6323 AVIndexEntry *entry = mov_find_next_sample(s, &st);
6325 return AVERROR_INVALIDDATA;
6327 if (sc->ffindex == stream_index && sc->current_sample == sample)
6329 sc->current_sample++;
6335 #define OFFSET(x) offsetof(MOVContext, x)
6336 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
6337 static const AVOption mov_options[] = {
6338 {"use_absolute_path",
6339 "allow using absolute path when opening alias, this is a possible security issue",
6340 OFFSET(use_absolute_path), AV_OPT_TYPE_BOOL, {.i64 = 0},
6342 {"seek_streams_individually",
6343 "Seek each stream individually to the to the closest point",
6344 OFFSET(seek_individually), AV_OPT_TYPE_BOOL, { .i64 = 1 },
6346 {"ignore_editlist", "", OFFSET(ignore_editlist), AV_OPT_TYPE_BOOL, {.i64 = 0},
6348 {"ignore_chapters", "", OFFSET(ignore_chapters), AV_OPT_TYPE_BOOL, {.i64 = 0},
6351 "use mfra for fragment timestamps",
6352 OFFSET(use_mfra_for), AV_OPT_TYPE_INT, {.i64 = FF_MOV_FLAG_MFRA_AUTO},
6353 -1, FF_MOV_FLAG_MFRA_PTS, FLAGS,
6355 {"auto", "auto", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_AUTO}, 0, 0,
6356 FLAGS, "use_mfra_for" },
6357 {"dts", "dts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_DTS}, 0, 0,
6358 FLAGS, "use_mfra_for" },
6359 {"pts", "pts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_PTS}, 0, 0,
6360 FLAGS, "use_mfra_for" },
6361 { "export_all", "Export unrecognized metadata entries", OFFSET(export_all),
6362 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
6363 { "export_xmp", "Export full XMP metadata", OFFSET(export_xmp),
6364 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
6365 { "activation_bytes", "Secret bytes for Audible AAX files", OFFSET(activation_bytes),
6366 AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
6367 { "audible_fixed_key", // extracted from libAAX_SDK.so and AAXSDKWin.dll files!
6368 "Fixed key used for handling Audible AAX files", OFFSET(audible_fixed_key),
6369 AV_OPT_TYPE_BINARY, {.str="77214d4b196a87cd520045fd20a51d67"},
6370 .flags = AV_OPT_FLAG_DECODING_PARAM },
6371 { "decryption_key", "The media decryption key (hex)", OFFSET(decryption_key), AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
6372 { "enable_drefs", "Enable external track support.", OFFSET(enable_drefs), AV_OPT_TYPE_BOOL,
6373 {.i64 = 0}, 0, 1, FLAGS },
6378 static const AVClass mov_class = {
6379 .class_name = "mov,mp4,m4a,3gp,3g2,mj2",
6380 .item_name = av_default_item_name,
6381 .option = mov_options,
6382 .version = LIBAVUTIL_VERSION_INT,
6385 AVInputFormat ff_mov_demuxer = {
6386 .name = "mov,mp4,m4a,3gp,3g2,mj2",
6387 .long_name = NULL_IF_CONFIG_SMALL("QuickTime / MOV"),
6388 .priv_class = &mov_class,
6389 .priv_data_size = sizeof(MOVContext),
6390 .extensions = "mov,mp4,m4a,3gp,3g2,mj2",
6391 .read_probe = mov_probe,
6392 .read_header = mov_read_header,
6393 .read_packet = mov_read_packet,
6394 .read_close = mov_read_close,
6395 .read_seek = mov_read_seek,
6396 .flags = AVFMT_NO_BYTE_SEEK,