3 * Copyright (c) 2001 Fabrice Bellard
4 * Copyright (c) 2009 Baptiste Coudurier <baptiste dot coudurier at gmail dot com>
6 * first version by Francois Revol <revol@free.fr>
7 * seek function by Gael Chardon <gael.dev@4now.net>
9 * This file is part of FFmpeg.
11 * FFmpeg is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public
13 * License as published by the Free Software Foundation; either
14 * version 2.1 of the License, or (at your option) any later version.
16 * FFmpeg is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with FFmpeg; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
30 #include "libavutil/attributes.h"
31 #include "libavutil/channel_layout.h"
32 #include "libavutil/internal.h"
33 #include "libavutil/intreadwrite.h"
34 #include "libavutil/intfloat.h"
35 #include "libavutil/mathematics.h"
36 #include "libavutil/time_internal.h"
37 #include "libavutil/avassert.h"
38 #include "libavutil/avstring.h"
39 #include "libavutil/dict.h"
40 #include "libavutil/display.h"
41 #include "libavutil/opt.h"
42 #include "libavutil/aes.h"
43 #include "libavutil/aes_ctr.h"
44 #include "libavutil/pixdesc.h"
45 #include "libavutil/sha.h"
46 #include "libavutil/spherical.h"
47 #include "libavutil/stereo3d.h"
48 #include "libavutil/timecode.h"
49 #include "libavcodec/ac3tab.h"
50 #include "libavcodec/flac.h"
51 #include "libavcodec/mpegaudiodecheader.h"
54 #include "avio_internal.h"
57 #include "libavcodec/get_bits.h"
60 #include "replaygain.h"
66 #include "qtpalette.h"
68 /* those functions parse an atom */
69 /* links atom IDs to parse functions */
70 typedef struct MOVParseTableEntry {
72 int (*parse)(MOVContext *ctx, AVIOContext *pb, MOVAtom atom);
75 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom);
76 static int mov_read_mfra(MOVContext *c, AVIOContext *f);
77 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
78 int count, int duration);
80 static int mov_metadata_track_or_disc_number(MOVContext *c, AVIOContext *pb,
81 unsigned len, const char *key)
85 short current, total = 0;
86 avio_rb16(pb); // unknown
87 current = avio_rb16(pb);
89 total = avio_rb16(pb);
91 snprintf(buf, sizeof(buf), "%d", current);
93 snprintf(buf, sizeof(buf), "%d/%d", current, total);
94 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
95 av_dict_set(&c->fc->metadata, key, buf, 0);
100 static int mov_metadata_int8_bypass_padding(MOVContext *c, AVIOContext *pb,
101 unsigned len, const char *key)
103 /* bypass padding bytes */
108 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
109 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
114 static int mov_metadata_int8_no_padding(MOVContext *c, AVIOContext *pb,
115 unsigned len, const char *key)
117 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
118 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
123 static int mov_metadata_gnre(MOVContext *c, AVIOContext *pb,
124 unsigned len, const char *key)
128 avio_r8(pb); // unknown
131 if (genre < 1 || genre > ID3v1_GENRE_MAX)
133 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
134 av_dict_set(&c->fc->metadata, key, ff_id3v1_genre_str[genre-1], 0);
139 static const uint32_t mac_to_unicode[128] = {
140 0x00C4,0x00C5,0x00C7,0x00C9,0x00D1,0x00D6,0x00DC,0x00E1,
141 0x00E0,0x00E2,0x00E4,0x00E3,0x00E5,0x00E7,0x00E9,0x00E8,
142 0x00EA,0x00EB,0x00ED,0x00EC,0x00EE,0x00EF,0x00F1,0x00F3,
143 0x00F2,0x00F4,0x00F6,0x00F5,0x00FA,0x00F9,0x00FB,0x00FC,
144 0x2020,0x00B0,0x00A2,0x00A3,0x00A7,0x2022,0x00B6,0x00DF,
145 0x00AE,0x00A9,0x2122,0x00B4,0x00A8,0x2260,0x00C6,0x00D8,
146 0x221E,0x00B1,0x2264,0x2265,0x00A5,0x00B5,0x2202,0x2211,
147 0x220F,0x03C0,0x222B,0x00AA,0x00BA,0x03A9,0x00E6,0x00F8,
148 0x00BF,0x00A1,0x00AC,0x221A,0x0192,0x2248,0x2206,0x00AB,
149 0x00BB,0x2026,0x00A0,0x00C0,0x00C3,0x00D5,0x0152,0x0153,
150 0x2013,0x2014,0x201C,0x201D,0x2018,0x2019,0x00F7,0x25CA,
151 0x00FF,0x0178,0x2044,0x20AC,0x2039,0x203A,0xFB01,0xFB02,
152 0x2021,0x00B7,0x201A,0x201E,0x2030,0x00C2,0x00CA,0x00C1,
153 0x00CB,0x00C8,0x00CD,0x00CE,0x00CF,0x00CC,0x00D3,0x00D4,
154 0xF8FF,0x00D2,0x00DA,0x00DB,0x00D9,0x0131,0x02C6,0x02DC,
155 0x00AF,0x02D8,0x02D9,0x02DA,0x00B8,0x02DD,0x02DB,0x02C7,
158 static int mov_read_mac_string(MOVContext *c, AVIOContext *pb, int len,
159 char *dst, int dstlen)
162 char *end = dst+dstlen-1;
165 for (i = 0; i < len; i++) {
166 uint8_t t, c = avio_r8(pb);
174 PUT_UTF8(mac_to_unicode[c-0x80], t, if (p < end) *p++ = t;);
180 static int mov_read_covr(MOVContext *c, AVIOContext *pb, int type, int len)
184 MOVStreamContext *sc;
189 case 0xd: id = AV_CODEC_ID_MJPEG; break;
190 case 0xe: id = AV_CODEC_ID_PNG; break;
191 case 0x1b: id = AV_CODEC_ID_BMP; break;
193 av_log(c->fc, AV_LOG_WARNING, "Unknown cover type: 0x%x.\n", type);
198 st = avformat_new_stream(c->fc, NULL);
200 return AVERROR(ENOMEM);
201 sc = av_mallocz(sizeof(*sc));
203 return AVERROR(ENOMEM);
206 ret = av_get_packet(pb, &pkt, len);
210 if (pkt.size >= 8 && id != AV_CODEC_ID_BMP) {
211 if (AV_RB64(pkt.data) == 0x89504e470d0a1a0a) {
212 id = AV_CODEC_ID_PNG;
214 id = AV_CODEC_ID_MJPEG;
218 st->disposition |= AV_DISPOSITION_ATTACHED_PIC;
220 st->attached_pic = pkt;
221 st->attached_pic.stream_index = st->index;
222 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
224 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
225 st->codecpar->codec_id = id;
231 static int mov_metadata_loci(MOVContext *c, AVIOContext *pb, unsigned len)
233 char language[4] = { 0 };
234 char buf[200], place[100];
235 uint16_t langcode = 0;
236 double longitude, latitude, altitude;
237 const char *key = "location";
239 if (len < 4 + 2 + 1 + 1 + 4 + 4 + 4) {
240 av_log(c->fc, AV_LOG_ERROR, "loci too short\n");
241 return AVERROR_INVALIDDATA;
244 avio_skip(pb, 4); // version+flags
245 langcode = avio_rb16(pb);
246 ff_mov_lang_to_iso639(langcode, language);
249 len -= avio_get_str(pb, len, place, sizeof(place));
251 av_log(c->fc, AV_LOG_ERROR, "place name too long\n");
252 return AVERROR_INVALIDDATA;
254 avio_skip(pb, 1); // role
258 av_log(c->fc, AV_LOG_ERROR,
259 "loci too short (%u bytes left, need at least %d)\n", len, 12);
260 return AVERROR_INVALIDDATA;
262 longitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
263 latitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
264 altitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
266 // Try to output in the same format as the ?xyz field
267 snprintf(buf, sizeof(buf), "%+08.4f%+09.4f", latitude, longitude);
269 av_strlcatf(buf, sizeof(buf), "%+f", altitude);
270 av_strlcatf(buf, sizeof(buf), "/%s", place);
272 if (*language && strcmp(language, "und")) {
274 snprintf(key2, sizeof(key2), "%s-%s", key, language);
275 av_dict_set(&c->fc->metadata, key2, buf, 0);
277 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
278 return av_dict_set(&c->fc->metadata, key, buf, 0);
281 static int mov_metadata_hmmt(MOVContext *c, AVIOContext *pb, unsigned len)
287 if (c->ignore_chapters)
290 n_hmmt = avio_rb32(pb);
291 for (i = 0; i < n_hmmt && !pb->eof_reached; i++) {
292 int moment_time = avio_rb32(pb);
293 avpriv_new_chapter(c->fc, i, av_make_q(1, 1000), moment_time, AV_NOPTS_VALUE, NULL);
298 static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom)
301 char key2[32], language[4] = {0};
303 const char *key = NULL;
304 uint16_t langcode = 0;
305 uint32_t data_type = 0, str_size, str_size_alloc;
306 int (*parse)(MOVContext*, AVIOContext*, unsigned, const char*) = NULL;
311 case MKTAG( '@','P','R','M'): key = "premiere_version"; raw = 1; break;
312 case MKTAG( '@','P','R','Q'): key = "quicktime_version"; raw = 1; break;
313 case MKTAG( 'X','M','P','_'):
314 if (c->export_xmp) { key = "xmp"; raw = 1; } break;
315 case MKTAG( 'a','A','R','T'): key = "album_artist"; break;
316 case MKTAG( 'a','k','I','D'): key = "account_type";
317 parse = mov_metadata_int8_no_padding; break;
318 case MKTAG( 'a','p','I','D'): key = "account_id"; break;
319 case MKTAG( 'c','a','t','g'): key = "category"; break;
320 case MKTAG( 'c','p','i','l'): key = "compilation";
321 parse = mov_metadata_int8_no_padding; break;
322 case MKTAG( 'c','p','r','t'): key = "copyright"; break;
323 case MKTAG( 'd','e','s','c'): key = "description"; break;
324 case MKTAG( 'd','i','s','k'): key = "disc";
325 parse = mov_metadata_track_or_disc_number; break;
326 case MKTAG( 'e','g','i','d'): key = "episode_uid";
327 parse = mov_metadata_int8_no_padding; break;
328 case MKTAG( 'F','I','R','M'): key = "firmware"; raw = 1; break;
329 case MKTAG( 'g','n','r','e'): key = "genre";
330 parse = mov_metadata_gnre; break;
331 case MKTAG( 'h','d','v','d'): key = "hd_video";
332 parse = mov_metadata_int8_no_padding; break;
333 case MKTAG( 'H','M','M','T'):
334 return mov_metadata_hmmt(c, pb, atom.size);
335 case MKTAG( 'k','e','y','w'): key = "keywords"; break;
336 case MKTAG( 'l','d','e','s'): key = "synopsis"; break;
337 case MKTAG( 'l','o','c','i'):
338 return mov_metadata_loci(c, pb, atom.size);
339 case MKTAG( 'm','a','n','u'): key = "make"; break;
340 case MKTAG( 'm','o','d','l'): key = "model"; break;
341 case MKTAG( 'p','c','s','t'): key = "podcast";
342 parse = mov_metadata_int8_no_padding; break;
343 case MKTAG( 'p','g','a','p'): key = "gapless_playback";
344 parse = mov_metadata_int8_no_padding; break;
345 case MKTAG( 'p','u','r','d'): key = "purchase_date"; break;
346 case MKTAG( 'r','t','n','g'): key = "rating";
347 parse = mov_metadata_int8_no_padding; break;
348 case MKTAG( 's','o','a','a'): key = "sort_album_artist"; break;
349 case MKTAG( 's','o','a','l'): key = "sort_album"; break;
350 case MKTAG( 's','o','a','r'): key = "sort_artist"; break;
351 case MKTAG( 's','o','c','o'): key = "sort_composer"; break;
352 case MKTAG( 's','o','n','m'): key = "sort_name"; break;
353 case MKTAG( 's','o','s','n'): key = "sort_show"; break;
354 case MKTAG( 's','t','i','k'): key = "media_type";
355 parse = mov_metadata_int8_no_padding; break;
356 case MKTAG( 't','r','k','n'): key = "track";
357 parse = mov_metadata_track_or_disc_number; break;
358 case MKTAG( 't','v','e','n'): key = "episode_id"; break;
359 case MKTAG( 't','v','e','s'): key = "episode_sort";
360 parse = mov_metadata_int8_bypass_padding; break;
361 case MKTAG( 't','v','n','n'): key = "network"; break;
362 case MKTAG( 't','v','s','h'): key = "show"; break;
363 case MKTAG( 't','v','s','n'): key = "season_number";
364 parse = mov_metadata_int8_bypass_padding; break;
365 case MKTAG(0xa9,'A','R','T'): key = "artist"; break;
366 case MKTAG(0xa9,'P','R','D'): key = "producer"; break;
367 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
368 case MKTAG(0xa9,'a','u','t'): key = "artist"; break;
369 case MKTAG(0xa9,'c','h','p'): key = "chapter"; break;
370 case MKTAG(0xa9,'c','m','t'): key = "comment"; break;
371 case MKTAG(0xa9,'c','o','m'): key = "composer"; break;
372 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
373 case MKTAG(0xa9,'d','a','y'): key = "date"; break;
374 case MKTAG(0xa9,'d','i','r'): key = "director"; break;
375 case MKTAG(0xa9,'d','i','s'): key = "disclaimer"; break;
376 case MKTAG(0xa9,'e','d','1'): key = "edit_date"; break;
377 case MKTAG(0xa9,'e','n','c'): key = "encoder"; break;
378 case MKTAG(0xa9,'f','m','t'): key = "original_format"; break;
379 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
380 case MKTAG(0xa9,'g','r','p'): key = "grouping"; break;
381 case MKTAG(0xa9,'h','s','t'): key = "host_computer"; break;
382 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
383 case MKTAG(0xa9,'l','y','r'): key = "lyrics"; break;
384 case MKTAG(0xa9,'m','a','k'): key = "make"; break;
385 case MKTAG(0xa9,'m','o','d'): key = "model"; break;
386 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
387 case MKTAG(0xa9,'o','p','e'): key = "original_artist"; break;
388 case MKTAG(0xa9,'p','r','d'): key = "producer"; break;
389 case MKTAG(0xa9,'p','r','f'): key = "performers"; break;
390 case MKTAG(0xa9,'r','e','q'): key = "playback_requirements"; break;
391 case MKTAG(0xa9,'s','r','c'): key = "original_source"; break;
392 case MKTAG(0xa9,'s','t','3'): key = "subtitle"; break;
393 case MKTAG(0xa9,'s','w','r'): key = "encoder"; break;
394 case MKTAG(0xa9,'t','o','o'): key = "encoder"; break;
395 case MKTAG(0xa9,'t','r','k'): key = "track"; break;
396 case MKTAG(0xa9,'u','r','l'): key = "URL"; break;
397 case MKTAG(0xa9,'w','r','n'): key = "warning"; break;
398 case MKTAG(0xa9,'w','r','t'): key = "composer"; break;
399 case MKTAG(0xa9,'x','y','z'): key = "location"; break;
402 if (c->itunes_metadata && atom.size > 8) {
403 int data_size = avio_rb32(pb);
404 int tag = avio_rl32(pb);
405 if (tag == MKTAG('d','a','t','a') && data_size <= atom.size) {
406 data_type = avio_rb32(pb); // type
407 avio_rb32(pb); // unknown
408 str_size = data_size - 16;
411 if (atom.type == MKTAG('c', 'o', 'v', 'r')) {
412 int ret = mov_read_covr(c, pb, data_type, str_size);
414 av_log(c->fc, AV_LOG_ERROR, "Error parsing cover art.\n");
417 atom.size -= str_size;
421 } else if (!key && c->found_hdlr_mdta && c->meta_keys) {
422 uint32_t index = AV_RB32(&atom.type);
423 if (index < c->meta_keys_count && index > 0) {
424 key = c->meta_keys[index];
426 av_log(c->fc, AV_LOG_WARNING,
427 "The index of 'data' is out of range: %"PRId32" < 1 or >= %d.\n",
428 index, c->meta_keys_count);
432 } else if (atom.size > 4 && key && !c->itunes_metadata && !raw) {
433 str_size = avio_rb16(pb); // string length
434 if (str_size > atom.size) {
436 avio_seek(pb, -2, SEEK_CUR);
437 av_log(c->fc, AV_LOG_WARNING, "UDTA parsing failed retrying raw\n");
440 langcode = avio_rb16(pb);
441 ff_mov_lang_to_iso639(langcode, language);
444 str_size = atom.size;
446 if (c->export_all && !key) {
447 snprintf(tmp_key, 5, "%.4s", (char*)&atom.type);
453 if (atom.size < 0 || str_size >= INT_MAX/2)
454 return AVERROR_INVALIDDATA;
456 // Allocates enough space if data_type is a int32 or float32 number, otherwise
457 // worst-case requirement for output string in case of utf8 coded input
458 num = (data_type >= 21 && data_type <= 23);
459 str_size_alloc = (num ? 512 : (raw ? str_size : str_size * 2)) + 1;
460 str = av_mallocz(str_size_alloc);
462 return AVERROR(ENOMEM);
465 parse(c, pb, str_size, key);
467 if (!raw && (data_type == 3 || (data_type == 0 && (langcode < 0x400 || langcode == 0x7fff)))) { // MAC Encoded
468 mov_read_mac_string(c, pb, str_size, str, str_size_alloc);
469 } else if (data_type == 21) { // BE signed integer, variable size
472 val = (int8_t)avio_r8(pb);
473 else if (str_size == 2)
474 val = (int16_t)avio_rb16(pb);
475 else if (str_size == 3)
476 val = ((int32_t)(avio_rb24(pb)<<8))>>8;
477 else if (str_size == 4)
478 val = (int32_t)avio_rb32(pb);
479 if (snprintf(str, str_size_alloc, "%d", val) >= str_size_alloc) {
480 av_log(c->fc, AV_LOG_ERROR,
481 "Failed to store the number (%d) in string.\n", val);
483 return AVERROR_INVALIDDATA;
485 } else if (data_type == 22) { // BE unsigned integer, variable size
486 unsigned int val = 0;
489 else if (str_size == 2)
491 else if (str_size == 3)
493 else if (str_size == 4)
495 if (snprintf(str, str_size_alloc, "%u", val) >= str_size_alloc) {
496 av_log(c->fc, AV_LOG_ERROR,
497 "Failed to store the number (%u) in string.\n", val);
499 return AVERROR_INVALIDDATA;
501 } else if (data_type == 23 && str_size >= 4) { // BE float32
502 float val = av_int2float(avio_rb32(pb));
503 if (snprintf(str, str_size_alloc, "%f", val) >= str_size_alloc) {
504 av_log(c->fc, AV_LOG_ERROR,
505 "Failed to store the float32 number (%f) in string.\n", val);
507 return AVERROR_INVALIDDATA;
510 int ret = ffio_read_size(pb, str, str_size);
517 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
518 av_dict_set(&c->fc->metadata, key, str, 0);
519 if (*language && strcmp(language, "und")) {
520 snprintf(key2, sizeof(key2), "%s-%s", key, language);
521 av_dict_set(&c->fc->metadata, key2, str, 0);
523 if (!strcmp(key, "encoder")) {
524 int major, minor, micro;
525 if (sscanf(str, "HandBrake %d.%d.%d", &major, &minor, µ) == 3) {
526 c->handbrake_version = 1000000*major + 1000*minor + micro;
535 static int mov_read_chpl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
538 int i, nb_chapters, str_len, version;
542 if (c->ignore_chapters)
545 if ((atom.size -= 5) < 0)
548 version = avio_r8(pb);
551 avio_rb32(pb); // ???
552 nb_chapters = avio_r8(pb);
554 for (i = 0; i < nb_chapters; i++) {
558 start = avio_rb64(pb);
559 str_len = avio_r8(pb);
561 if ((atom.size -= 9+str_len) < 0)
564 ret = ffio_read_size(pb, str, str_len);
568 avpriv_new_chapter(c->fc, i, (AVRational){1,10000000}, start, AV_NOPTS_VALUE, str);
573 #define MIN_DATA_ENTRY_BOX_SIZE 12
574 static int mov_read_dref(MOVContext *c, AVIOContext *pb, MOVAtom atom)
577 MOVStreamContext *sc;
580 if (c->fc->nb_streams < 1)
582 st = c->fc->streams[c->fc->nb_streams-1];
585 avio_rb32(pb); // version + flags
586 entries = avio_rb32(pb);
588 entries > (atom.size - 1) / MIN_DATA_ENTRY_BOX_SIZE + 1 ||
589 entries >= UINT_MAX / sizeof(*sc->drefs))
590 return AVERROR_INVALIDDATA;
594 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
596 return AVERROR(ENOMEM);
597 sc->drefs_count = entries;
599 for (i = 0; i < entries; i++) {
600 MOVDref *dref = &sc->drefs[i];
601 uint32_t size = avio_rb32(pb);
602 int64_t next = avio_tell(pb) + size - 4;
605 return AVERROR_INVALIDDATA;
607 dref->type = avio_rl32(pb);
608 avio_rb32(pb); // version + flags
610 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
611 /* macintosh alias record */
612 uint16_t volume_len, len;
618 volume_len = avio_r8(pb);
619 volume_len = FFMIN(volume_len, 27);
620 ret = ffio_read_size(pb, dref->volume, 27);
623 dref->volume[volume_len] = 0;
624 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", dref->volume, volume_len);
629 len = FFMIN(len, 63);
630 ret = ffio_read_size(pb, dref->filename, 63);
633 dref->filename[len] = 0;
634 av_log(c->fc, AV_LOG_DEBUG, "filename %s, len %d\n", dref->filename, len);
638 /* read next level up_from_alias/down_to_target */
639 dref->nlvl_from = avio_rb16(pb);
640 dref->nlvl_to = avio_rb16(pb);
641 av_log(c->fc, AV_LOG_DEBUG, "nlvl from %d, nlvl to %d\n",
642 dref->nlvl_from, dref->nlvl_to);
646 for (type = 0; type != -1 && avio_tell(pb) < next; ) {
649 type = avio_rb16(pb);
651 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
654 if (type == 2) { // absolute path
656 dref->path = av_mallocz(len+1);
658 return AVERROR(ENOMEM);
660 ret = ffio_read_size(pb, dref->path, len);
662 av_freep(&dref->path);
665 if (len > volume_len && !strncmp(dref->path, dref->volume, volume_len)) {
667 memmove(dref->path, dref->path+volume_len, len);
670 // trim string of any ending zeros
671 for (j = len - 1; j >= 0; j--) {
672 if (dref->path[j] == 0)
677 for (j = 0; j < len; j++)
678 if (dref->path[j] == ':' || dref->path[j] == 0)
680 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
681 } else if (type == 0) { // directory name
683 dref->dir = av_malloc(len+1);
685 return AVERROR(ENOMEM);
687 ret = ffio_read_size(pb, dref->dir, len);
689 av_freep(&dref->dir);
693 for (j = 0; j < len; j++)
694 if (dref->dir[j] == ':')
696 av_log(c->fc, AV_LOG_DEBUG, "dir %s\n", dref->dir);
701 av_log(c->fc, AV_LOG_DEBUG, "Unknown dref type 0x%08"PRIx32" size %"PRIu32"\n",
706 avio_seek(pb, next, SEEK_SET);
711 static int mov_read_hdlr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
720 avio_r8(pb); /* version */
721 avio_rb24(pb); /* flags */
724 ctype = avio_rl32(pb);
725 type = avio_rl32(pb); /* component subtype */
727 av_log(c->fc, AV_LOG_TRACE, "ctype=%s\n", av_fourcc2str(ctype));
728 av_log(c->fc, AV_LOG_TRACE, "stype=%s\n", av_fourcc2str(type));
730 if (c->trak_index < 0) { // meta not inside a trak
731 if (type == MKTAG('m','d','t','a')) {
732 c->found_hdlr_mdta = 1;
737 st = c->fc->streams[c->fc->nb_streams-1];
739 if (type == MKTAG('v','i','d','e'))
740 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
741 else if (type == MKTAG('s','o','u','n'))
742 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
743 else if (type == MKTAG('m','1','a',' '))
744 st->codecpar->codec_id = AV_CODEC_ID_MP2;
745 else if ((type == MKTAG('s','u','b','p')) || (type == MKTAG('c','l','c','p')))
746 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
748 avio_rb32(pb); /* component manufacture */
749 avio_rb32(pb); /* component flags */
750 avio_rb32(pb); /* component flags mask */
752 title_size = atom.size - 24;
753 if (title_size > 0) {
754 if (title_size > FFMIN(INT_MAX, SIZE_MAX-1))
755 return AVERROR_INVALIDDATA;
756 title_str = av_malloc(title_size + 1); /* Add null terminator */
758 return AVERROR(ENOMEM);
760 ret = ffio_read_size(pb, title_str, title_size);
762 av_freep(&title_str);
765 title_str[title_size] = 0;
767 int off = (!c->isom && title_str[0] == title_size - 1);
768 // flag added so as to not set stream handler name if already set from mdia->hdlr
769 av_dict_set(&st->metadata, "handler_name", title_str + off, AV_DICT_DONT_OVERWRITE);
771 av_freep(&title_str);
777 static int mov_read_esds(MOVContext *c, AVIOContext *pb, MOVAtom atom)
779 return ff_mov_read_esds(c->fc, pb);
782 static int mov_read_dac3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
785 enum AVAudioServiceType *ast;
786 int ac3info, acmod, lfeon, bsmod;
788 if (c->fc->nb_streams < 1)
790 st = c->fc->streams[c->fc->nb_streams-1];
792 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
795 return AVERROR(ENOMEM);
797 ac3info = avio_rb24(pb);
798 bsmod = (ac3info >> 14) & 0x7;
799 acmod = (ac3info >> 11) & 0x7;
800 lfeon = (ac3info >> 10) & 0x1;
801 st->codecpar->channels = ((int[]){2,1,2,3,3,4,4,5})[acmod] + lfeon;
802 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
804 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
806 if (st->codecpar->channels > 1 && bsmod == 0x7)
807 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
809 #if FF_API_LAVF_AVCTX
810 FF_DISABLE_DEPRECATION_WARNINGS
811 st->codec->audio_service_type = *ast;
812 FF_ENABLE_DEPRECATION_WARNINGS
818 static int mov_read_dec3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
821 enum AVAudioServiceType *ast;
822 int eac3info, acmod, lfeon, bsmod;
824 if (c->fc->nb_streams < 1)
826 st = c->fc->streams[c->fc->nb_streams-1];
828 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
831 return AVERROR(ENOMEM);
833 /* No need to parse fields for additional independent substreams and its
834 * associated dependent substreams since libavcodec's E-AC-3 decoder
835 * does not support them yet. */
836 avio_rb16(pb); /* data_rate and num_ind_sub */
837 eac3info = avio_rb24(pb);
838 bsmod = (eac3info >> 12) & 0x1f;
839 acmod = (eac3info >> 9) & 0x7;
840 lfeon = (eac3info >> 8) & 0x1;
841 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
843 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
844 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
846 if (st->codecpar->channels > 1 && bsmod == 0x7)
847 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
849 #if FF_API_LAVF_AVCTX
850 FF_DISABLE_DEPRECATION_WARNINGS
851 st->codec->audio_service_type = *ast;
852 FF_ENABLE_DEPRECATION_WARNINGS
858 static int mov_read_ddts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
860 const uint32_t ddts_size = 20;
863 uint32_t frame_duration_code = 0;
864 uint32_t channel_layout_code = 0;
867 buf = av_malloc(ddts_size + AV_INPUT_BUFFER_PADDING_SIZE);
869 return AVERROR(ENOMEM);
871 if (avio_read(pb, buf, ddts_size) < ddts_size) {
873 return AVERROR_INVALIDDATA;
876 init_get_bits(&gb, buf, 8*ddts_size);
878 if (c->fc->nb_streams < 1) {
882 st = c->fc->streams[c->fc->nb_streams-1];
884 st->codecpar->sample_rate = get_bits_long(&gb, 32);
885 if (st->codecpar->sample_rate <= 0) {
886 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
888 return AVERROR_INVALIDDATA;
890 skip_bits_long(&gb, 32); /* max bitrate */
891 st->codecpar->bit_rate = get_bits_long(&gb, 32);
892 st->codecpar->bits_per_coded_sample = get_bits(&gb, 8);
893 frame_duration_code = get_bits(&gb, 2);
894 skip_bits(&gb, 30); /* various fields */
895 channel_layout_code = get_bits(&gb, 16);
897 st->codecpar->frame_size =
898 (frame_duration_code == 0) ? 512 :
899 (frame_duration_code == 1) ? 1024 :
900 (frame_duration_code == 2) ? 2048 :
901 (frame_duration_code == 3) ? 4096 : 0;
903 if (channel_layout_code > 0xff) {
904 av_log(c->fc, AV_LOG_WARNING, "Unsupported DTS audio channel layout");
906 st->codecpar->channel_layout =
907 ((channel_layout_code & 0x1) ? AV_CH_FRONT_CENTER : 0) |
908 ((channel_layout_code & 0x2) ? AV_CH_FRONT_LEFT : 0) |
909 ((channel_layout_code & 0x2) ? AV_CH_FRONT_RIGHT : 0) |
910 ((channel_layout_code & 0x4) ? AV_CH_SIDE_LEFT : 0) |
911 ((channel_layout_code & 0x4) ? AV_CH_SIDE_RIGHT : 0) |
912 ((channel_layout_code & 0x8) ? AV_CH_LOW_FREQUENCY : 0);
914 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
920 static int mov_read_chan(MOVContext *c, AVIOContext *pb, MOVAtom atom)
924 if (c->fc->nb_streams < 1)
926 st = c->fc->streams[c->fc->nb_streams-1];
931 /* skip version and flags */
934 ff_mov_read_chan(c->fc, pb, st, atom.size - 4);
939 static int mov_read_wfex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
944 if (c->fc->nb_streams < 1)
946 st = c->fc->streams[c->fc->nb_streams-1];
948 if ((ret = ff_get_wav_header(c->fc, pb, st->codecpar, atom.size, 0)) < 0)
949 av_log(c->fc, AV_LOG_WARNING, "get_wav_header failed\n");
954 static int mov_read_pasp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
956 const int num = avio_rb32(pb);
957 const int den = avio_rb32(pb);
960 if (c->fc->nb_streams < 1)
962 st = c->fc->streams[c->fc->nb_streams-1];
964 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
965 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num)) {
966 av_log(c->fc, AV_LOG_WARNING,
967 "sample aspect ratio already set to %d:%d, ignoring 'pasp' atom (%d:%d)\n",
968 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den,
970 } else if (den != 0) {
971 av_reduce(&st->sample_aspect_ratio.num, &st->sample_aspect_ratio.den,
977 /* this atom contains actual media data */
978 static int mov_read_mdat(MOVContext *c, AVIOContext *pb, MOVAtom atom)
980 if (atom.size == 0) /* wrong one (MP4) */
983 return 0; /* now go for moov */
986 #define DRM_BLOB_SIZE 56
988 static int mov_read_adrm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
990 uint8_t intermediate_key[20];
991 uint8_t intermediate_iv[20];
994 uint8_t file_checksum[20];
995 uint8_t calculated_checksum[20];
999 uint8_t *activation_bytes = c->activation_bytes;
1000 uint8_t *fixed_key = c->audible_fixed_key;
1004 sha = av_sha_alloc();
1006 return AVERROR(ENOMEM);
1007 c->aes_decrypt = av_aes_alloc();
1008 if (!c->aes_decrypt) {
1009 ret = AVERROR(ENOMEM);
1013 /* drm blob processing */
1014 avio_read(pb, output, 8); // go to offset 8, absolute position 0x251
1015 avio_read(pb, input, DRM_BLOB_SIZE);
1016 avio_read(pb, output, 4); // go to offset 4, absolute position 0x28d
1017 avio_read(pb, file_checksum, 20);
1019 av_log(c->fc, AV_LOG_INFO, "[aax] file checksum == "); // required by external tools
1020 for (i = 0; i < 20; i++)
1021 av_log(c->fc, AV_LOG_INFO, "%02x", file_checksum[i]);
1022 av_log(c->fc, AV_LOG_INFO, "\n");
1024 /* verify activation data */
1025 if (!activation_bytes) {
1026 av_log(c->fc, AV_LOG_WARNING, "[aax] activation_bytes option is missing!\n");
1027 ret = 0; /* allow ffprobe to continue working on .aax files */
1030 if (c->activation_bytes_size != 4) {
1031 av_log(c->fc, AV_LOG_FATAL, "[aax] activation_bytes value needs to be 4 bytes!\n");
1032 ret = AVERROR(EINVAL);
1036 /* verify fixed key */
1037 if (c->audible_fixed_key_size != 16) {
1038 av_log(c->fc, AV_LOG_FATAL, "[aax] audible_fixed_key value needs to be 16 bytes!\n");
1039 ret = AVERROR(EINVAL);
1043 /* AAX (and AAX+) key derivation */
1044 av_sha_init(sha, 160);
1045 av_sha_update(sha, fixed_key, 16);
1046 av_sha_update(sha, activation_bytes, 4);
1047 av_sha_final(sha, intermediate_key);
1048 av_sha_init(sha, 160);
1049 av_sha_update(sha, fixed_key, 16);
1050 av_sha_update(sha, intermediate_key, 20);
1051 av_sha_update(sha, activation_bytes, 4);
1052 av_sha_final(sha, intermediate_iv);
1053 av_sha_init(sha, 160);
1054 av_sha_update(sha, intermediate_key, 16);
1055 av_sha_update(sha, intermediate_iv, 16);
1056 av_sha_final(sha, calculated_checksum);
1057 if (memcmp(calculated_checksum, file_checksum, 20)) { // critical error
1058 av_log(c->fc, AV_LOG_ERROR, "[aax] mismatch in checksums!\n");
1059 ret = AVERROR_INVALIDDATA;
1062 av_aes_init(c->aes_decrypt, intermediate_key, 128, 1);
1063 av_aes_crypt(c->aes_decrypt, output, input, DRM_BLOB_SIZE >> 4, intermediate_iv, 1);
1064 for (i = 0; i < 4; i++) {
1065 // file data (in output) is stored in big-endian mode
1066 if (activation_bytes[i] != output[3 - i]) { // critical error
1067 av_log(c->fc, AV_LOG_ERROR, "[aax] error in drm blob decryption!\n");
1068 ret = AVERROR_INVALIDDATA;
1072 memcpy(c->file_key, output + 8, 16);
1073 memcpy(input, output + 26, 16);
1074 av_sha_init(sha, 160);
1075 av_sha_update(sha, input, 16);
1076 av_sha_update(sha, c->file_key, 16);
1077 av_sha_update(sha, fixed_key, 16);
1078 av_sha_final(sha, c->file_iv);
1086 // Audible AAX (and AAX+) bytestream decryption
1087 static int aax_filter(uint8_t *input, int size, MOVContext *c)
1090 unsigned char iv[16];
1092 memcpy(iv, c->file_iv, 16); // iv is overwritten
1093 blocks = size >> 4; // trailing bytes are not encrypted!
1094 av_aes_init(c->aes_decrypt, c->file_key, 128, 1);
1095 av_aes_crypt(c->aes_decrypt, input, input, blocks, iv, 1);
1100 /* read major brand, minor version and compatible brands and store them as metadata */
1101 static int mov_read_ftyp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1104 int comp_brand_size;
1105 char* comp_brands_str;
1106 uint8_t type[5] = {0};
1107 int ret = ffio_read_size(pb, type, 4);
1111 if (strcmp(type, "qt "))
1113 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
1114 av_dict_set(&c->fc->metadata, "major_brand", type, 0);
1115 minor_ver = avio_rb32(pb); /* minor version */
1116 av_dict_set_int(&c->fc->metadata, "minor_version", minor_ver, 0);
1118 comp_brand_size = atom.size - 8;
1119 if (comp_brand_size < 0)
1120 return AVERROR_INVALIDDATA;
1121 comp_brands_str = av_malloc(comp_brand_size + 1); /* Add null terminator */
1122 if (!comp_brands_str)
1123 return AVERROR(ENOMEM);
1125 ret = ffio_read_size(pb, comp_brands_str, comp_brand_size);
1127 av_freep(&comp_brands_str);
1130 comp_brands_str[comp_brand_size] = 0;
1131 av_dict_set(&c->fc->metadata, "compatible_brands", comp_brands_str, 0);
1132 av_freep(&comp_brands_str);
1137 /* this atom should contain all header atoms */
1138 static int mov_read_moov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1142 if (c->found_moov) {
1143 av_log(c->fc, AV_LOG_WARNING, "Found duplicated MOOV Atom. Skipped it\n");
1144 avio_skip(pb, atom.size);
1148 if ((ret = mov_read_default(c, pb, atom)) < 0)
1150 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
1151 /* so we don't parse the whole file if over a network */
1153 return 0; /* now go for mdat */
1156 static MOVFragmentStreamInfo * get_frag_stream_info(
1157 MOVFragmentIndex *frag_index,
1162 MOVFragmentIndexItem * item;
1164 if (index < 0 || index >= frag_index->nb_items)
1166 item = &frag_index->item[index];
1167 for (i = 0; i < item->nb_stream_info; i++)
1168 if (item->stream_info[i].id == id)
1169 return &item->stream_info[i];
1171 // This shouldn't happen
1175 static void set_frag_stream(MOVFragmentIndex *frag_index, int id)
1178 MOVFragmentIndexItem * item;
1180 if (frag_index->current < 0 ||
1181 frag_index->current >= frag_index->nb_items)
1184 item = &frag_index->item[frag_index->current];
1185 for (i = 0; i < item->nb_stream_info; i++)
1186 if (item->stream_info[i].id == id) {
1191 // id not found. This shouldn't happen.
1195 static MOVFragmentStreamInfo * get_current_frag_stream_info(
1196 MOVFragmentIndex *frag_index)
1198 MOVFragmentIndexItem *item;
1199 if (frag_index->current < 0 ||
1200 frag_index->current >= frag_index->nb_items)
1203 item = &frag_index->item[frag_index->current];
1204 if (item->current >= 0 && item->current < item->nb_stream_info)
1205 return &item->stream_info[item->current];
1207 // This shouldn't happen
1211 static int search_frag_moof_offset(MOVFragmentIndex *frag_index, int64_t offset)
1214 int64_t moof_offset;
1216 // Optimize for appending new entries
1217 if (!frag_index->nb_items ||
1218 frag_index->item[frag_index->nb_items - 1].moof_offset < offset)
1219 return frag_index->nb_items;
1222 b = frag_index->nb_items;
1226 moof_offset = frag_index->item[m].moof_offset;
1227 if (moof_offset >= offset)
1229 if (moof_offset <= offset)
1235 static int64_t get_stream_info_time(MOVFragmentStreamInfo * frag_stream_info)
1238 if (frag_stream_info) {
1239 if (frag_stream_info->sidx_pts != AV_NOPTS_VALUE)
1240 return frag_stream_info->sidx_pts;
1241 if (frag_stream_info->first_tfra_pts != AV_NOPTS_VALUE)
1242 return frag_stream_info->first_tfra_pts;
1243 if (frag_stream_info->tfdt_dts != AV_NOPTS_VALUE)
1244 return frag_stream_info->tfdt_dts;
1246 return AV_NOPTS_VALUE;
1249 static int64_t get_frag_time(MOVFragmentIndex *frag_index,
1250 int index, int track_id)
1252 MOVFragmentStreamInfo * frag_stream_info;
1256 if (track_id >= 0) {
1257 frag_stream_info = get_frag_stream_info(frag_index, index, track_id);
1258 return frag_stream_info->sidx_pts;
1261 for (i = 0; i < frag_index->item[index].nb_stream_info; i++) {
1262 frag_stream_info = &frag_index->item[index].stream_info[i];
1263 timestamp = get_stream_info_time(frag_stream_info);
1264 if (timestamp != AV_NOPTS_VALUE)
1267 return AV_NOPTS_VALUE;
1270 static int search_frag_timestamp(MOVFragmentIndex *frag_index,
1271 AVStream *st, int64_t timestamp)
1278 // If the stream is referenced by any sidx, limit the search
1279 // to fragments that referenced this stream in the sidx
1280 MOVStreamContext *sc = st->priv_data;
1286 b = frag_index->nb_items;
1290 frag_time = get_frag_time(frag_index, m, id);
1291 if (frag_time != AV_NOPTS_VALUE) {
1292 if (frag_time >= timestamp)
1294 if (frag_time <= timestamp)
1301 static int update_frag_index(MOVContext *c, int64_t offset)
1304 MOVFragmentIndexItem * item;
1305 MOVFragmentStreamInfo * frag_stream_info;
1307 // If moof_offset already exists in frag_index, return index to it
1308 index = search_frag_moof_offset(&c->frag_index, offset);
1309 if (index < c->frag_index.nb_items &&
1310 c->frag_index.item[index].moof_offset == offset)
1313 // offset is not yet in frag index.
1314 // Insert new item at index (sorted by moof offset)
1315 item = av_fast_realloc(c->frag_index.item,
1316 &c->frag_index.allocated_size,
1317 (c->frag_index.nb_items + 1) *
1318 sizeof(*c->frag_index.item));
1321 c->frag_index.item = item;
1323 frag_stream_info = av_realloc_array(NULL, c->fc->nb_streams,
1324 sizeof(*item->stream_info));
1325 if (!frag_stream_info)
1328 for (i = 0; i < c->fc->nb_streams; i++) {
1329 frag_stream_info[i].id = c->fc->streams[i]->id;
1330 frag_stream_info[i].sidx_pts = AV_NOPTS_VALUE;
1331 frag_stream_info[i].tfdt_dts = AV_NOPTS_VALUE;
1332 frag_stream_info[i].first_tfra_pts = AV_NOPTS_VALUE;
1333 frag_stream_info[i].index_entry = -1;
1334 frag_stream_info[i].encryption_index = NULL;
1337 if (index < c->frag_index.nb_items)
1338 memmove(c->frag_index.item + index + 1, c->frag_index.item + index,
1339 (c->frag_index.nb_items - index) * sizeof(*c->frag_index.item));
1341 item = &c->frag_index.item[index];
1342 item->headers_read = 0;
1344 item->nb_stream_info = c->fc->nb_streams;
1345 item->moof_offset = offset;
1346 item->stream_info = frag_stream_info;
1347 c->frag_index.nb_items++;
1352 static void fix_frag_index_entries(MOVFragmentIndex *frag_index, int index,
1353 int id, int entries)
1356 MOVFragmentStreamInfo * frag_stream_info;
1360 for (i = index; i < frag_index->nb_items; i++) {
1361 frag_stream_info = get_frag_stream_info(frag_index, i, id);
1362 if (frag_stream_info && frag_stream_info->index_entry >= 0)
1363 frag_stream_info->index_entry += entries;
1367 static int mov_read_moof(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1369 if (!c->has_looked_for_mfra && c->use_mfra_for > 0) {
1370 c->has_looked_for_mfra = 1;
1371 if (pb->seekable & AVIO_SEEKABLE_NORMAL) {
1373 av_log(c->fc, AV_LOG_VERBOSE, "stream has moof boxes, will look "
1375 if ((ret = mov_read_mfra(c, pb)) < 0) {
1376 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but failed to "
1377 "read the mfra (may be a live ismv)\n");
1380 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but stream is not "
1381 "seekable, can not look for mfra\n");
1384 c->fragment.moof_offset = c->fragment.implicit_offset = avio_tell(pb) - 8;
1385 av_log(c->fc, AV_LOG_TRACE, "moof offset %"PRIx64"\n", c->fragment.moof_offset);
1386 c->frag_index.current = update_frag_index(c, c->fragment.moof_offset);
1387 return mov_read_default(c, pb, atom);
1390 static void mov_metadata_creation_time(AVDictionary **metadata, int64_t time)
1393 if(time >= 2082844800)
1394 time -= 2082844800; /* seconds between 1904-01-01 and Epoch */
1396 if ((int64_t)(time * 1000000ULL) / 1000000 != time) {
1397 av_log(NULL, AV_LOG_DEBUG, "creation_time is not representable\n");
1401 avpriv_dict_set_timestamp(metadata, "creation_time", time * 1000000);
1405 static int mov_read_mdhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1408 MOVStreamContext *sc;
1410 char language[4] = {0};
1412 int64_t creation_time;
1414 if (c->fc->nb_streams < 1)
1416 st = c->fc->streams[c->fc->nb_streams-1];
1419 if (sc->time_scale) {
1420 av_log(c->fc, AV_LOG_ERROR, "Multiple mdhd?\n");
1421 return AVERROR_INVALIDDATA;
1424 version = avio_r8(pb);
1426 avpriv_request_sample(c->fc, "Version %d", version);
1427 return AVERROR_PATCHWELCOME;
1429 avio_rb24(pb); /* flags */
1431 creation_time = avio_rb64(pb);
1434 creation_time = avio_rb32(pb);
1435 avio_rb32(pb); /* modification time */
1437 mov_metadata_creation_time(&st->metadata, creation_time);
1439 sc->time_scale = avio_rb32(pb);
1440 if (sc->time_scale <= 0) {
1441 av_log(c->fc, AV_LOG_ERROR, "Invalid mdhd time scale %d, defaulting to 1\n", sc->time_scale);
1444 st->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1446 lang = avio_rb16(pb); /* language */
1447 if (ff_mov_lang_to_iso639(lang, language))
1448 av_dict_set(&st->metadata, "language", language, 0);
1449 avio_rb16(pb); /* quality */
1454 static int mov_read_mvhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1457 int64_t creation_time;
1458 int version = avio_r8(pb); /* version */
1459 avio_rb24(pb); /* flags */
1462 creation_time = avio_rb64(pb);
1465 creation_time = avio_rb32(pb);
1466 avio_rb32(pb); /* modification time */
1468 mov_metadata_creation_time(&c->fc->metadata, creation_time);
1469 c->time_scale = avio_rb32(pb); /* time scale */
1470 if (c->time_scale <= 0) {
1471 av_log(c->fc, AV_LOG_ERROR, "Invalid mvhd time scale %d, defaulting to 1\n", c->time_scale);
1474 av_log(c->fc, AV_LOG_TRACE, "time scale = %i\n", c->time_scale);
1476 c->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1477 // set the AVCodecContext duration because the duration of individual tracks
1478 // may be inaccurate
1479 if (c->time_scale > 0 && !c->trex_data)
1480 c->fc->duration = av_rescale(c->duration, AV_TIME_BASE, c->time_scale);
1481 avio_rb32(pb); /* preferred scale */
1483 avio_rb16(pb); /* preferred volume */
1485 avio_skip(pb, 10); /* reserved */
1487 /* movie display matrix, store it in main context and use it later on */
1488 for (i = 0; i < 3; i++) {
1489 c->movie_display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
1490 c->movie_display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
1491 c->movie_display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
1494 avio_rb32(pb); /* preview time */
1495 avio_rb32(pb); /* preview duration */
1496 avio_rb32(pb); /* poster time */
1497 avio_rb32(pb); /* selection time */
1498 avio_rb32(pb); /* selection duration */
1499 avio_rb32(pb); /* current time */
1500 avio_rb32(pb); /* next track ID */
1505 static int mov_read_enda(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1510 if (c->fc->nb_streams < 1)
1512 st = c->fc->streams[c->fc->nb_streams-1];
1514 little_endian = avio_rb16(pb) & 0xFF;
1515 av_log(c->fc, AV_LOG_TRACE, "enda %d\n", little_endian);
1516 if (little_endian == 1) {
1517 switch (st->codecpar->codec_id) {
1518 case AV_CODEC_ID_PCM_S24BE:
1519 st->codecpar->codec_id = AV_CODEC_ID_PCM_S24LE;
1521 case AV_CODEC_ID_PCM_S32BE:
1522 st->codecpar->codec_id = AV_CODEC_ID_PCM_S32LE;
1524 case AV_CODEC_ID_PCM_F32BE:
1525 st->codecpar->codec_id = AV_CODEC_ID_PCM_F32LE;
1527 case AV_CODEC_ID_PCM_F64BE:
1528 st->codecpar->codec_id = AV_CODEC_ID_PCM_F64LE;
1537 static int mov_read_colr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1540 char color_parameter_type[5] = { 0 };
1541 uint16_t color_primaries, color_trc, color_matrix;
1544 if (c->fc->nb_streams < 1)
1546 st = c->fc->streams[c->fc->nb_streams - 1];
1548 ret = ffio_read_size(pb, color_parameter_type, 4);
1551 if (strncmp(color_parameter_type, "nclx", 4) &&
1552 strncmp(color_parameter_type, "nclc", 4)) {
1553 av_log(c->fc, AV_LOG_WARNING, "unsupported color_parameter_type %s\n",
1554 color_parameter_type);
1558 color_primaries = avio_rb16(pb);
1559 color_trc = avio_rb16(pb);
1560 color_matrix = avio_rb16(pb);
1562 av_log(c->fc, AV_LOG_TRACE,
1563 "%s: pri %d trc %d matrix %d",
1564 color_parameter_type, color_primaries, color_trc, color_matrix);
1566 if (!strncmp(color_parameter_type, "nclx", 4)) {
1567 uint8_t color_range = avio_r8(pb) >> 7;
1568 av_log(c->fc, AV_LOG_TRACE, " full %"PRIu8"", color_range);
1570 st->codecpar->color_range = AVCOL_RANGE_JPEG;
1572 st->codecpar->color_range = AVCOL_RANGE_MPEG;
1575 if (!av_color_primaries_name(color_primaries))
1576 color_primaries = AVCOL_PRI_UNSPECIFIED;
1577 if (!av_color_transfer_name(color_trc))
1578 color_trc = AVCOL_TRC_UNSPECIFIED;
1579 if (!av_color_space_name(color_matrix))
1580 color_matrix = AVCOL_SPC_UNSPECIFIED;
1582 st->codecpar->color_primaries = color_primaries;
1583 st->codecpar->color_trc = color_trc;
1584 st->codecpar->color_space = color_matrix;
1585 av_log(c->fc, AV_LOG_TRACE, "\n");
1590 static int mov_read_fiel(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1593 unsigned mov_field_order;
1594 enum AVFieldOrder decoded_field_order = AV_FIELD_UNKNOWN;
1596 if (c->fc->nb_streams < 1) // will happen with jp2 files
1598 st = c->fc->streams[c->fc->nb_streams-1];
1600 return AVERROR_INVALIDDATA;
1601 mov_field_order = avio_rb16(pb);
1602 if ((mov_field_order & 0xFF00) == 0x0100)
1603 decoded_field_order = AV_FIELD_PROGRESSIVE;
1604 else if ((mov_field_order & 0xFF00) == 0x0200) {
1605 switch (mov_field_order & 0xFF) {
1606 case 0x01: decoded_field_order = AV_FIELD_TT;
1608 case 0x06: decoded_field_order = AV_FIELD_BB;
1610 case 0x09: decoded_field_order = AV_FIELD_TB;
1612 case 0x0E: decoded_field_order = AV_FIELD_BT;
1616 if (decoded_field_order == AV_FIELD_UNKNOWN && mov_field_order) {
1617 av_log(NULL, AV_LOG_ERROR, "Unknown MOV field order 0x%04x\n", mov_field_order);
1619 st->codecpar->field_order = decoded_field_order;
1624 static int mov_realloc_extradata(AVCodecParameters *par, MOVAtom atom)
1627 uint64_t size = (uint64_t)par->extradata_size + atom.size + 8 + AV_INPUT_BUFFER_PADDING_SIZE;
1628 if (size > INT_MAX || (uint64_t)atom.size > INT_MAX)
1629 return AVERROR_INVALIDDATA;
1630 if ((err = av_reallocp(&par->extradata, size)) < 0) {
1631 par->extradata_size = 0;
1634 par->extradata_size = size - AV_INPUT_BUFFER_PADDING_SIZE;
1638 /* Read a whole atom into the extradata return the size of the atom read, possibly truncated if != atom.size */
1639 static int64_t mov_read_atom_into_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1640 AVCodecParameters *par, uint8_t *buf)
1642 int64_t result = atom.size;
1645 AV_WB32(buf , atom.size + 8);
1646 AV_WL32(buf + 4, atom.type);
1647 err = ffio_read_size(pb, buf + 8, atom.size);
1649 par->extradata_size -= atom.size;
1651 } else if (err < atom.size) {
1652 av_log(c->fc, AV_LOG_WARNING, "truncated extradata\n");
1653 par->extradata_size -= atom.size - err;
1656 memset(buf + 8 + err, 0, AV_INPUT_BUFFER_PADDING_SIZE);
1660 /* FIXME modify QDM2/SVQ3/H.264 decoders to take full atom as extradata */
1661 static int mov_read_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1662 enum AVCodecID codec_id)
1665 uint64_t original_size;
1668 if (c->fc->nb_streams < 1) // will happen with jp2 files
1670 st = c->fc->streams[c->fc->nb_streams-1];
1672 if (st->codecpar->codec_id != codec_id)
1673 return 0; /* unexpected codec_id - don't mess with extradata */
1675 original_size = st->codecpar->extradata_size;
1676 err = mov_realloc_extradata(st->codecpar, atom);
1680 err = mov_read_atom_into_extradata(c, pb, atom, st->codecpar, st->codecpar->extradata + original_size);
1683 return 0; // Note: this is the original behavior to ignore truncation.
1686 /* wrapper functions for reading ALAC/AVS/MJPEG/MJPEG2000 extradata atoms only for those codecs */
1687 static int mov_read_alac(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1689 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_ALAC);
1692 static int mov_read_avss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1694 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVS);
1697 static int mov_read_jp2h(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1699 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_JPEG2000);
1702 static int mov_read_dpxe(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1704 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_R10K);
1707 static int mov_read_avid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1709 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVUI);
1711 ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_DNXHD);
1715 static int mov_read_targa_y216(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1717 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_TARGA_Y216);
1719 if (!ret && c->fc->nb_streams >= 1) {
1720 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1721 if (par->extradata_size >= 40) {
1722 par->height = AV_RB16(&par->extradata[36]);
1723 par->width = AV_RB16(&par->extradata[38]);
1729 static int mov_read_ares(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1731 if (c->fc->nb_streams >= 1) {
1732 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1733 if (par->codec_tag == MKTAG('A', 'V', 'i', 'n') &&
1734 par->codec_id == AV_CODEC_ID_H264 &&
1738 cid = avio_rb16(pb);
1739 /* For AVID AVCI50, force width of 1440 to be able to select the correct SPS and PPS */
1740 if (cid == 0xd4d || cid == 0xd4e)
1743 } else if ((par->codec_tag == MKTAG('A', 'V', 'd', '1') ||
1744 par->codec_tag == MKTAG('A', 'V', 'j', '2') ||
1745 par->codec_tag == MKTAG('A', 'V', 'd', 'n')) &&
1749 num = avio_rb32(pb);
1750 den = avio_rb32(pb);
1751 if (num <= 0 || den <= 0)
1753 switch (avio_rb32(pb)) {
1755 if (den >= INT_MAX / 2)
1759 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.num = num;
1760 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.den = den;
1767 return mov_read_avid(c, pb, atom);
1770 static int mov_read_aclr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1774 uint64_t original_size;
1775 if (c->fc->nb_streams >= 1) {
1776 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1777 if (par->codec_id == AV_CODEC_ID_H264)
1779 if (atom.size == 16) {
1780 original_size = par->extradata_size;
1781 ret = mov_realloc_extradata(par, atom);
1783 length = mov_read_atom_into_extradata(c, pb, atom, par, par->extradata + original_size);
1784 if (length == atom.size) {
1785 const uint8_t range_value = par->extradata[original_size + 19];
1786 switch (range_value) {
1788 par->color_range = AVCOL_RANGE_MPEG;
1791 par->color_range = AVCOL_RANGE_JPEG;
1794 av_log(c, AV_LOG_WARNING, "ignored unknown aclr value (%d)\n", range_value);
1797 ff_dlog(c, "color_range: %d\n", par->color_range);
1799 /* For some reason the whole atom was not added to the extradata */
1800 av_log(c, AV_LOG_ERROR, "aclr not decoded - incomplete atom\n");
1803 av_log(c, AV_LOG_ERROR, "aclr not decoded - unable to add atom to extradata\n");
1806 av_log(c, AV_LOG_WARNING, "aclr not decoded - unexpected size %"PRId64"\n", atom.size);
1813 static int mov_read_svq3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1815 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_SVQ3);
1818 static int mov_read_wave(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1823 if (c->fc->nb_streams < 1)
1825 st = c->fc->streams[c->fc->nb_streams-1];
1827 if ((uint64_t)atom.size > (1<<30))
1828 return AVERROR_INVALIDDATA;
1830 if (st->codecpar->codec_id == AV_CODEC_ID_QDM2 ||
1831 st->codecpar->codec_id == AV_CODEC_ID_QDMC ||
1832 st->codecpar->codec_id == AV_CODEC_ID_SPEEX) {
1833 // pass all frma atom to codec, needed at least for QDMC and QDM2
1834 av_freep(&st->codecpar->extradata);
1835 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1838 } else if (atom.size > 8) { /* to read frma, esds atoms */
1839 if (st->codecpar->codec_id == AV_CODEC_ID_ALAC && atom.size >= 24) {
1841 ret = ffio_ensure_seekback(pb, 8);
1844 buffer = avio_rb64(pb);
1846 if ( (buffer & 0xFFFFFFFF) == MKBETAG('f','r','m','a')
1847 && buffer >> 32 <= atom.size
1848 && buffer >> 32 >= 8) {
1851 } else if (!st->codecpar->extradata_size) {
1852 #define ALAC_EXTRADATA_SIZE 36
1853 st->codecpar->extradata = av_mallocz(ALAC_EXTRADATA_SIZE + AV_INPUT_BUFFER_PADDING_SIZE);
1854 if (!st->codecpar->extradata)
1855 return AVERROR(ENOMEM);
1856 st->codecpar->extradata_size = ALAC_EXTRADATA_SIZE;
1857 AV_WB32(st->codecpar->extradata , ALAC_EXTRADATA_SIZE);
1858 AV_WB32(st->codecpar->extradata + 4, MKTAG('a','l','a','c'));
1859 AV_WB64(st->codecpar->extradata + 12, buffer);
1860 avio_read(pb, st->codecpar->extradata + 20, 16);
1861 avio_skip(pb, atom.size - 24);
1865 if ((ret = mov_read_default(c, pb, atom)) < 0)
1868 avio_skip(pb, atom.size);
1873 * This function reads atom content and puts data in extradata without tag
1874 * nor size unlike mov_read_extradata.
1876 static int mov_read_glbl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1881 if (c->fc->nb_streams < 1)
1883 st = c->fc->streams[c->fc->nb_streams-1];
1885 if ((uint64_t)atom.size > (1<<30))
1886 return AVERROR_INVALIDDATA;
1888 if (atom.size >= 10) {
1889 // Broken files created by legacy versions of libavformat will
1890 // wrap a whole fiel atom inside of a glbl atom.
1891 unsigned size = avio_rb32(pb);
1892 unsigned type = avio_rl32(pb);
1893 avio_seek(pb, -8, SEEK_CUR);
1894 if (type == MKTAG('f','i','e','l') && size == atom.size)
1895 return mov_read_default(c, pb, atom);
1897 if (st->codecpar->extradata_size > 1 && st->codecpar->extradata) {
1898 av_log(c, AV_LOG_WARNING, "ignoring multiple glbl\n");
1901 av_freep(&st->codecpar->extradata);
1902 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1909 static int mov_read_dvc1(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1912 uint8_t profile_level;
1915 if (c->fc->nb_streams < 1)
1917 st = c->fc->streams[c->fc->nb_streams-1];
1919 if (atom.size >= (1<<28) || atom.size < 7)
1920 return AVERROR_INVALIDDATA;
1922 profile_level = avio_r8(pb);
1923 if ((profile_level & 0xf0) != 0xc0)
1926 avio_seek(pb, 6, SEEK_CUR);
1927 av_freep(&st->codecpar->extradata);
1928 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 7);
1936 * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
1937 * but can have extradata appended at the end after the 40 bytes belonging
1940 static int mov_read_strf(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1945 if (c->fc->nb_streams < 1)
1947 if (atom.size <= 40)
1949 st = c->fc->streams[c->fc->nb_streams-1];
1951 if ((uint64_t)atom.size > (1<<30))
1952 return AVERROR_INVALIDDATA;
1955 av_freep(&st->codecpar->extradata);
1956 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 40);
1963 static int mov_read_stco(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1966 MOVStreamContext *sc;
1967 unsigned int i, entries;
1969 if (c->fc->nb_streams < 1)
1971 st = c->fc->streams[c->fc->nb_streams-1];
1974 avio_r8(pb); /* version */
1975 avio_rb24(pb); /* flags */
1977 entries = avio_rb32(pb);
1982 if (sc->chunk_offsets)
1983 av_log(c->fc, AV_LOG_WARNING, "Duplicated STCO atom\n");
1984 av_free(sc->chunk_offsets);
1985 sc->chunk_count = 0;
1986 sc->chunk_offsets = av_malloc_array(entries, sizeof(*sc->chunk_offsets));
1987 if (!sc->chunk_offsets)
1988 return AVERROR(ENOMEM);
1989 sc->chunk_count = entries;
1991 if (atom.type == MKTAG('s','t','c','o'))
1992 for (i = 0; i < entries && !pb->eof_reached; i++)
1993 sc->chunk_offsets[i] = avio_rb32(pb);
1994 else if (atom.type == MKTAG('c','o','6','4'))
1995 for (i = 0; i < entries && !pb->eof_reached; i++)
1996 sc->chunk_offsets[i] = avio_rb64(pb);
1998 return AVERROR_INVALIDDATA;
2000 sc->chunk_count = i;
2002 if (pb->eof_reached) {
2003 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STCO atom\n");
2010 static int mov_codec_id(AVStream *st, uint32_t format)
2012 int id = ff_codec_get_id(ff_codec_movaudio_tags, format);
2015 ((format & 0xFFFF) == 'm' + ('s' << 8) ||
2016 (format & 0xFFFF) == 'T' + ('S' << 8)))
2017 id = ff_codec_get_id(ff_codec_wav_tags, av_bswap32(format) & 0xFFFF);
2019 if (st->codecpar->codec_type != AVMEDIA_TYPE_VIDEO && id > 0) {
2020 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2021 } else if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO &&
2022 /* skip old ASF MPEG-4 tag */
2023 format && format != MKTAG('m','p','4','s')) {
2024 id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2026 id = ff_codec_get_id(ff_codec_bmp_tags, format);
2028 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
2029 else if (st->codecpar->codec_type == AVMEDIA_TYPE_DATA ||
2030 (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE &&
2031 st->codecpar->codec_id == AV_CODEC_ID_NONE)) {
2032 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
2034 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
2038 st->codecpar->codec_tag = format;
2043 static void mov_parse_stsd_video(MOVContext *c, AVIOContext *pb,
2044 AVStream *st, MOVStreamContext *sc)
2046 uint8_t codec_name[32] = { 0 };
2050 /* The first 16 bytes of the video sample description are already
2051 * read in ff_mov_read_stsd_entries() */
2052 stsd_start = avio_tell(pb) - 16;
2054 avio_rb16(pb); /* version */
2055 avio_rb16(pb); /* revision level */
2056 avio_rb32(pb); /* vendor */
2057 avio_rb32(pb); /* temporal quality */
2058 avio_rb32(pb); /* spatial quality */
2060 st->codecpar->width = avio_rb16(pb); /* width */
2061 st->codecpar->height = avio_rb16(pb); /* height */
2063 avio_rb32(pb); /* horiz resolution */
2064 avio_rb32(pb); /* vert resolution */
2065 avio_rb32(pb); /* data size, always 0 */
2066 avio_rb16(pb); /* frames per samples */
2068 len = avio_r8(pb); /* codec name, pascal string */
2071 mov_read_mac_string(c, pb, len, codec_name, sizeof(codec_name));
2073 avio_skip(pb, 31 - len);
2076 av_dict_set(&st->metadata, "encoder", codec_name, 0);
2078 /* codec_tag YV12 triggers an UV swap in rawdec.c */
2079 if (!strncmp(codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25)) {
2080 st->codecpar->codec_tag = MKTAG('I', '4', '2', '0');
2081 st->codecpar->width &= ~1;
2082 st->codecpar->height &= ~1;
2084 /* Flash Media Server uses tag H.263 with Sorenson Spark */
2085 if (st->codecpar->codec_tag == MKTAG('H','2','6','3') &&
2086 !strncmp(codec_name, "Sorenson H263", 13))
2087 st->codecpar->codec_id = AV_CODEC_ID_FLV1;
2089 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* depth */
2091 avio_seek(pb, stsd_start, SEEK_SET);
2093 if (ff_get_qtpalette(st->codecpar->codec_id, pb, sc->palette)) {
2094 st->codecpar->bits_per_coded_sample &= 0x1F;
2095 sc->has_palette = 1;
2099 static void mov_parse_stsd_audio(MOVContext *c, AVIOContext *pb,
2100 AVStream *st, MOVStreamContext *sc)
2102 int bits_per_sample, flags;
2103 uint16_t version = avio_rb16(pb);
2104 AVDictionaryEntry *compatible_brands = av_dict_get(c->fc->metadata, "compatible_brands", NULL, AV_DICT_MATCH_CASE);
2106 avio_rb16(pb); /* revision level */
2107 avio_rb32(pb); /* vendor */
2109 st->codecpar->channels = avio_rb16(pb); /* channel count */
2110 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* sample size */
2111 av_log(c->fc, AV_LOG_TRACE, "audio channels %d\n", st->codecpar->channels);
2113 sc->audio_cid = avio_rb16(pb);
2114 avio_rb16(pb); /* packet size = 0 */
2116 st->codecpar->sample_rate = ((avio_rb32(pb) >> 16));
2118 // Read QT version 1 fields. In version 0 these do not exist.
2119 av_log(c->fc, AV_LOG_TRACE, "version =%d, isom =%d\n", version, c->isom);
2121 (compatible_brands && strstr(compatible_brands->value, "qt "))) {
2124 sc->samples_per_frame = avio_rb32(pb);
2125 avio_rb32(pb); /* bytes per packet */
2126 sc->bytes_per_frame = avio_rb32(pb);
2127 avio_rb32(pb); /* bytes per sample */
2128 } else if (version == 2) {
2129 avio_rb32(pb); /* sizeof struct only */
2130 st->codecpar->sample_rate = av_int2double(avio_rb64(pb));
2131 st->codecpar->channels = avio_rb32(pb);
2132 avio_rb32(pb); /* always 0x7F000000 */
2133 st->codecpar->bits_per_coded_sample = avio_rb32(pb);
2135 flags = avio_rb32(pb); /* lpcm format specific flag */
2136 sc->bytes_per_frame = avio_rb32(pb);
2137 sc->samples_per_frame = avio_rb32(pb);
2138 if (st->codecpar->codec_tag == MKTAG('l','p','c','m'))
2139 st->codecpar->codec_id =
2140 ff_mov_get_lpcm_codec_id(st->codecpar->bits_per_coded_sample,
2143 if (version == 0 || (version == 1 && sc->audio_cid != -2)) {
2144 /* can't correctly handle variable sized packet as audio unit */
2145 switch (st->codecpar->codec_id) {
2146 case AV_CODEC_ID_MP2:
2147 case AV_CODEC_ID_MP3:
2148 st->need_parsing = AVSTREAM_PARSE_FULL;
2154 if (sc->format == 0) {
2155 if (st->codecpar->bits_per_coded_sample == 8)
2156 st->codecpar->codec_id = mov_codec_id(st, MKTAG('r','a','w',' '));
2157 else if (st->codecpar->bits_per_coded_sample == 16)
2158 st->codecpar->codec_id = mov_codec_id(st, MKTAG('t','w','o','s'));
2161 switch (st->codecpar->codec_id) {
2162 case AV_CODEC_ID_PCM_S8:
2163 case AV_CODEC_ID_PCM_U8:
2164 if (st->codecpar->bits_per_coded_sample == 16)
2165 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16BE;
2167 case AV_CODEC_ID_PCM_S16LE:
2168 case AV_CODEC_ID_PCM_S16BE:
2169 if (st->codecpar->bits_per_coded_sample == 8)
2170 st->codecpar->codec_id = AV_CODEC_ID_PCM_S8;
2171 else if (st->codecpar->bits_per_coded_sample == 24)
2172 st->codecpar->codec_id =
2173 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
2174 AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
2175 else if (st->codecpar->bits_per_coded_sample == 32)
2176 st->codecpar->codec_id =
2177 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
2178 AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
2180 /* set values for old format before stsd version 1 appeared */
2181 case AV_CODEC_ID_MACE3:
2182 sc->samples_per_frame = 6;
2183 sc->bytes_per_frame = 2 * st->codecpar->channels;
2185 case AV_CODEC_ID_MACE6:
2186 sc->samples_per_frame = 6;
2187 sc->bytes_per_frame = 1 * st->codecpar->channels;
2189 case AV_CODEC_ID_ADPCM_IMA_QT:
2190 sc->samples_per_frame = 64;
2191 sc->bytes_per_frame = 34 * st->codecpar->channels;
2193 case AV_CODEC_ID_GSM:
2194 sc->samples_per_frame = 160;
2195 sc->bytes_per_frame = 33;
2201 bits_per_sample = av_get_bits_per_sample(st->codecpar->codec_id);
2202 if (bits_per_sample) {
2203 st->codecpar->bits_per_coded_sample = bits_per_sample;
2204 sc->sample_size = (bits_per_sample >> 3) * st->codecpar->channels;
2208 static void mov_parse_stsd_subtitle(MOVContext *c, AVIOContext *pb,
2209 AVStream *st, MOVStreamContext *sc,
2212 // ttxt stsd contains display flags, justification, background
2213 // color, fonts, and default styles, so fake an atom to read it
2214 MOVAtom fake_atom = { .size = size };
2215 // mp4s contains a regular esds atom
2216 if (st->codecpar->codec_tag != AV_RL32("mp4s"))
2217 mov_read_glbl(c, pb, fake_atom);
2218 st->codecpar->width = sc->width;
2219 st->codecpar->height = sc->height;
2222 static uint32_t yuv_to_rgba(uint32_t ycbcr)
2227 y = (ycbcr >> 16) & 0xFF;
2228 cr = (ycbcr >> 8) & 0xFF;
2231 b = av_clip_uint8((1164 * (y - 16) + 2018 * (cb - 128)) / 1000);
2232 g = av_clip_uint8((1164 * (y - 16) - 813 * (cr - 128) - 391 * (cb - 128)) / 1000);
2233 r = av_clip_uint8((1164 * (y - 16) + 1596 * (cr - 128) ) / 1000);
2235 return (r << 16) | (g << 8) | b;
2238 static int mov_rewrite_dvd_sub_extradata(AVStream *st)
2240 char buf[256] = {0};
2241 uint8_t *src = st->codecpar->extradata;
2244 if (st->codecpar->extradata_size != 64)
2247 if (st->codecpar->width > 0 && st->codecpar->height > 0)
2248 snprintf(buf, sizeof(buf), "size: %dx%d\n",
2249 st->codecpar->width, st->codecpar->height);
2250 av_strlcat(buf, "palette: ", sizeof(buf));
2252 for (i = 0; i < 16; i++) {
2253 uint32_t yuv = AV_RB32(src + i * 4);
2254 uint32_t rgba = yuv_to_rgba(yuv);
2256 av_strlcatf(buf, sizeof(buf), "%06"PRIx32"%s", rgba, i != 15 ? ", " : "");
2259 if (av_strlcat(buf, "\n", sizeof(buf)) >= sizeof(buf))
2262 av_freep(&st->codecpar->extradata);
2263 st->codecpar->extradata_size = 0;
2264 st->codecpar->extradata = av_mallocz(strlen(buf) + AV_INPUT_BUFFER_PADDING_SIZE);
2265 if (!st->codecpar->extradata)
2266 return AVERROR(ENOMEM);
2267 st->codecpar->extradata_size = strlen(buf);
2268 memcpy(st->codecpar->extradata, buf, st->codecpar->extradata_size);
2273 static int mov_parse_stsd_data(MOVContext *c, AVIOContext *pb,
2274 AVStream *st, MOVStreamContext *sc,
2279 if (st->codecpar->codec_tag == MKTAG('t','m','c','d')) {
2280 if ((int)size != size)
2281 return AVERROR(ENOMEM);
2283 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
2287 MOVStreamContext *tmcd_ctx = st->priv_data;
2289 val = AV_RB32(st->codecpar->extradata + 4);
2290 tmcd_ctx->tmcd_flags = val;
2291 st->avg_frame_rate.num = st->codecpar->extradata[16]; /* number of frame */
2292 st->avg_frame_rate.den = 1;
2293 #if FF_API_LAVF_AVCTX
2294 FF_DISABLE_DEPRECATION_WARNINGS
2295 st->codec->time_base = av_inv_q(st->avg_frame_rate);
2296 FF_ENABLE_DEPRECATION_WARNINGS
2298 /* adjust for per frame dur in counter mode */
2299 if (tmcd_ctx->tmcd_flags & 0x0008) {
2300 int timescale = AV_RB32(st->codecpar->extradata + 8);
2301 int framedur = AV_RB32(st->codecpar->extradata + 12);
2302 st->avg_frame_rate.num *= timescale;
2303 st->avg_frame_rate.den *= framedur;
2304 #if FF_API_LAVF_AVCTX
2305 FF_DISABLE_DEPRECATION_WARNINGS
2306 st->codec->time_base.den *= timescale;
2307 st->codec->time_base.num *= framedur;
2308 FF_ENABLE_DEPRECATION_WARNINGS
2312 uint32_t len = AV_RB32(st->codecpar->extradata + 18); /* name atom length */
2313 uint32_t format = AV_RB32(st->codecpar->extradata + 22);
2314 if (format == AV_RB32("name") && (int64_t)size >= (int64_t)len + 18) {
2315 uint16_t str_size = AV_RB16(st->codecpar->extradata + 26); /* string length */
2316 if (str_size > 0 && size >= (int)str_size + 26) {
2317 char *reel_name = av_malloc(str_size + 1);
2319 return AVERROR(ENOMEM);
2320 memcpy(reel_name, st->codecpar->extradata + 30, str_size);
2321 reel_name[str_size] = 0; /* Add null terminator */
2322 /* don't add reel_name if emtpy string */
2323 if (*reel_name == 0) {
2326 av_dict_set(&st->metadata, "reel_name", reel_name, AV_DICT_DONT_STRDUP_VAL);
2333 /* other codec type, just skip (rtp, mp4s ...) */
2334 avio_skip(pb, size);
2339 static int mov_finalize_stsd_codec(MOVContext *c, AVIOContext *pb,
2340 AVStream *st, MOVStreamContext *sc)
2342 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
2343 !st->codecpar->sample_rate && sc->time_scale > 1)
2344 st->codecpar->sample_rate = sc->time_scale;
2346 /* special codec parameters handling */
2347 switch (st->codecpar->codec_id) {
2348 #if CONFIG_DV_DEMUXER
2349 case AV_CODEC_ID_DVAUDIO:
2350 c->dv_fctx = avformat_alloc_context();
2352 av_log(c->fc, AV_LOG_ERROR, "dv demux context alloc error\n");
2353 return AVERROR(ENOMEM);
2355 c->dv_demux = avpriv_dv_init_demux(c->dv_fctx);
2357 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
2358 return AVERROR(ENOMEM);
2360 sc->dv_audio_container = 1;
2361 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16LE;
2364 /* no ifdef since parameters are always those */
2365 case AV_CODEC_ID_QCELP:
2366 st->codecpar->channels = 1;
2367 // force sample rate for qcelp when not stored in mov
2368 if (st->codecpar->codec_tag != MKTAG('Q','c','l','p'))
2369 st->codecpar->sample_rate = 8000;
2370 // FIXME: Why is the following needed for some files?
2371 sc->samples_per_frame = 160;
2372 if (!sc->bytes_per_frame)
2373 sc->bytes_per_frame = 35;
2375 case AV_CODEC_ID_AMR_NB:
2376 st->codecpar->channels = 1;
2377 /* force sample rate for amr, stsd in 3gp does not store sample rate */
2378 st->codecpar->sample_rate = 8000;
2380 case AV_CODEC_ID_AMR_WB:
2381 st->codecpar->channels = 1;
2382 st->codecpar->sample_rate = 16000;
2384 case AV_CODEC_ID_MP2:
2385 case AV_CODEC_ID_MP3:
2386 /* force type after stsd for m1a hdlr */
2387 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2389 case AV_CODEC_ID_GSM:
2390 case AV_CODEC_ID_ADPCM_MS:
2391 case AV_CODEC_ID_ADPCM_IMA_WAV:
2392 case AV_CODEC_ID_ILBC:
2393 case AV_CODEC_ID_MACE3:
2394 case AV_CODEC_ID_MACE6:
2395 case AV_CODEC_ID_QDM2:
2396 st->codecpar->block_align = sc->bytes_per_frame;
2398 case AV_CODEC_ID_ALAC:
2399 if (st->codecpar->extradata_size == 36) {
2400 st->codecpar->channels = AV_RB8 (st->codecpar->extradata + 21);
2401 st->codecpar->sample_rate = AV_RB32(st->codecpar->extradata + 32);
2404 case AV_CODEC_ID_AC3:
2405 case AV_CODEC_ID_EAC3:
2406 case AV_CODEC_ID_MPEG1VIDEO:
2407 case AV_CODEC_ID_VC1:
2408 case AV_CODEC_ID_VP8:
2409 case AV_CODEC_ID_VP9:
2410 st->need_parsing = AVSTREAM_PARSE_FULL;
2418 static int mov_skip_multiple_stsd(MOVContext *c, AVIOContext *pb,
2419 int codec_tag, int format,
2422 int video_codec_id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2425 (codec_tag != format &&
2426 // AVID 1:1 samples with differing data format and codec tag exist
2427 (codec_tag != AV_RL32("AV1x") || format != AV_RL32("AVup")) &&
2428 // prores is allowed to have differing data format and codec tag
2429 codec_tag != AV_RL32("apcn") && codec_tag != AV_RL32("apch") &&
2431 codec_tag != AV_RL32("dvpp") && codec_tag != AV_RL32("dvcp") &&
2432 (c->fc->video_codec_id ? video_codec_id != c->fc->video_codec_id
2433 : codec_tag != MKTAG('j','p','e','g')))) {
2434 /* Multiple fourcc, we skip JPEG. This is not correct, we should
2435 * export it as a separate AVStream but this needs a few changes
2436 * in the MOV demuxer, patch welcome. */
2438 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
2439 avio_skip(pb, size);
2446 int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries)
2449 MOVStreamContext *sc;
2450 int pseudo_stream_id;
2452 av_assert0 (c->fc->nb_streams >= 1);
2453 st = c->fc->streams[c->fc->nb_streams-1];
2456 for (pseudo_stream_id = 0;
2457 pseudo_stream_id < entries && !pb->eof_reached;
2458 pseudo_stream_id++) {
2459 //Parsing Sample description table
2461 int ret, dref_id = 1;
2462 MOVAtom a = { AV_RL32("stsd") };
2463 int64_t start_pos = avio_tell(pb);
2464 int64_t size = avio_rb32(pb); /* size */
2465 uint32_t format = avio_rl32(pb); /* data format */
2468 avio_rb32(pb); /* reserved */
2469 avio_rb16(pb); /* reserved */
2470 dref_id = avio_rb16(pb);
2471 } else if (size <= 7) {
2472 av_log(c->fc, AV_LOG_ERROR,
2473 "invalid size %"PRId64" in stsd\n", size);
2474 return AVERROR_INVALIDDATA;
2477 if (mov_skip_multiple_stsd(c, pb, st->codecpar->codec_tag, format,
2478 size - (avio_tell(pb) - start_pos))) {
2483 sc->pseudo_stream_id = st->codecpar->codec_tag ? -1 : pseudo_stream_id;
2484 sc->dref_id= dref_id;
2485 sc->format = format;
2487 id = mov_codec_id(st, format);
2489 av_log(c->fc, AV_LOG_TRACE,
2490 "size=%"PRId64" 4CC=%s codec_type=%d\n", size,
2491 av_fourcc2str(format), st->codecpar->codec_type);
2493 if (st->codecpar->codec_type==AVMEDIA_TYPE_VIDEO) {
2494 st->codecpar->codec_id = id;
2495 mov_parse_stsd_video(c, pb, st, sc);
2496 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_AUDIO) {
2497 st->codecpar->codec_id = id;
2498 mov_parse_stsd_audio(c, pb, st, sc);
2499 if (st->codecpar->sample_rate < 0) {
2500 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
2501 return AVERROR_INVALIDDATA;
2503 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_SUBTITLE){
2504 st->codecpar->codec_id = id;
2505 mov_parse_stsd_subtitle(c, pb, st, sc,
2506 size - (avio_tell(pb) - start_pos));
2508 ret = mov_parse_stsd_data(c, pb, st, sc,
2509 size - (avio_tell(pb) - start_pos));
2513 /* this will read extra atoms at the end (wave, alac, damr, avcC, hvcC, SMI ...) */
2514 a.size = size - (avio_tell(pb) - start_pos);
2516 if ((ret = mov_read_default(c, pb, a)) < 0)
2518 } else if (a.size > 0)
2519 avio_skip(pb, a.size);
2521 if (sc->extradata && st->codecpar->extradata) {
2522 int extra_size = st->codecpar->extradata_size;
2524 /* Move the current stream extradata to the stream context one. */
2525 sc->extradata_size[pseudo_stream_id] = extra_size;
2526 sc->extradata[pseudo_stream_id] = av_malloc(extra_size + AV_INPUT_BUFFER_PADDING_SIZE);
2527 if (!sc->extradata[pseudo_stream_id])
2528 return AVERROR(ENOMEM);
2529 memcpy(sc->extradata[pseudo_stream_id], st->codecpar->extradata, extra_size);
2530 av_freep(&st->codecpar->extradata);
2531 st->codecpar->extradata_size = 0;
2536 if (pb->eof_reached) {
2537 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSD atom\n");
2544 static int mov_read_stsd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2547 MOVStreamContext *sc;
2550 if (c->fc->nb_streams < 1)
2552 st = c->fc->streams[c->fc->nb_streams - 1];
2555 avio_r8(pb); /* version */
2556 avio_rb24(pb); /* flags */
2557 entries = avio_rb32(pb);
2560 av_log(c->fc, AV_LOG_ERROR, "invalid STSD entries %d\n", entries);
2561 return AVERROR_INVALIDDATA;
2564 if (sc->extradata) {
2565 av_log(c->fc, AV_LOG_ERROR,
2566 "Duplicate stsd found in this track.\n");
2567 return AVERROR_INVALIDDATA;
2570 /* Prepare space for hosting multiple extradata. */
2571 sc->extradata = av_mallocz_array(entries, sizeof(*sc->extradata));
2573 return AVERROR(ENOMEM);
2575 sc->extradata_size = av_mallocz_array(entries, sizeof(*sc->extradata_size));
2576 if (!sc->extradata_size) {
2577 ret = AVERROR(ENOMEM);
2581 ret = ff_mov_read_stsd_entries(c, pb, entries);
2585 /* Restore back the primary extradata. */
2586 av_freep(&st->codecpar->extradata);
2587 st->codecpar->extradata_size = sc->extradata_size[0];
2588 if (sc->extradata_size[0]) {
2589 st->codecpar->extradata = av_mallocz(sc->extradata_size[0] + AV_INPUT_BUFFER_PADDING_SIZE);
2590 if (!st->codecpar->extradata)
2591 return AVERROR(ENOMEM);
2592 memcpy(st->codecpar->extradata, sc->extradata[0], sc->extradata_size[0]);
2595 return mov_finalize_stsd_codec(c, pb, st, sc);
2597 if (sc->extradata) {
2599 for (j = 0; j < sc->stsd_count; j++)
2600 av_freep(&sc->extradata[j]);
2603 av_freep(&sc->extradata);
2604 av_freep(&sc->extradata_size);
2608 static int mov_read_stsc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2611 MOVStreamContext *sc;
2612 unsigned int i, entries;
2614 if (c->fc->nb_streams < 1)
2616 st = c->fc->streams[c->fc->nb_streams-1];
2619 avio_r8(pb); /* version */
2620 avio_rb24(pb); /* flags */
2622 entries = avio_rb32(pb);
2623 if ((uint64_t)entries * 12 + 4 > atom.size)
2624 return AVERROR_INVALIDDATA;
2626 av_log(c->fc, AV_LOG_TRACE, "track[%u].stsc.entries = %u\n", c->fc->nb_streams - 1, entries);
2631 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSC atom\n");
2632 av_free(sc->stsc_data);
2634 sc->stsc_data = av_malloc_array(entries, sizeof(*sc->stsc_data));
2636 return AVERROR(ENOMEM);
2638 for (i = 0; i < entries && !pb->eof_reached; i++) {
2639 sc->stsc_data[i].first = avio_rb32(pb);
2640 sc->stsc_data[i].count = avio_rb32(pb);
2641 sc->stsc_data[i].id = avio_rb32(pb);
2645 for (i = sc->stsc_count - 1; i < UINT_MAX; i--) {
2646 int64_t first_min = i + 1;
2647 if ((i+1 < sc->stsc_count && sc->stsc_data[i].first >= sc->stsc_data[i+1].first) ||
2648 (i > 0 && sc->stsc_data[i].first <= sc->stsc_data[i-1].first) ||
2649 sc->stsc_data[i].first < first_min ||
2650 sc->stsc_data[i].count < 1 ||
2651 sc->stsc_data[i].id < 1) {
2652 av_log(c->fc, AV_LOG_WARNING, "STSC entry %d is invalid (first=%d count=%d id=%d)\n", i, sc->stsc_data[i].first, sc->stsc_data[i].count, sc->stsc_data[i].id);
2653 if (i+1 >= sc->stsc_count) {
2654 sc->stsc_data[i].first = FFMAX(sc->stsc_data[i].first, first_min);
2655 if (i > 0 && sc->stsc_data[i].first <= sc->stsc_data[i-1].first)
2656 sc->stsc_data[i].first = FFMIN(sc->stsc_data[i-1].first + 1LL, INT_MAX);
2657 sc->stsc_data[i].count = FFMAX(sc->stsc_data[i].count, 1);
2658 sc->stsc_data[i].id = FFMAX(sc->stsc_data[i].id, 1);
2661 av_assert0(sc->stsc_data[i+1].first >= 2);
2662 // We replace this entry by the next valid
2663 sc->stsc_data[i].first = sc->stsc_data[i+1].first - 1;
2664 sc->stsc_data[i].count = sc->stsc_data[i+1].count;
2665 sc->stsc_data[i].id = sc->stsc_data[i+1].id;
2669 if (pb->eof_reached) {
2670 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSC atom\n");
2677 static inline int mov_stsc_index_valid(unsigned int index, unsigned int count)
2679 return index < count - 1;
2682 /* Compute the samples value for the stsc entry at the given index. */
2683 static inline int64_t mov_get_stsc_samples(MOVStreamContext *sc, unsigned int index)
2687 if (mov_stsc_index_valid(index, sc->stsc_count))
2688 chunk_count = sc->stsc_data[index + 1].first - sc->stsc_data[index].first;
2690 chunk_count = sc->chunk_count - (sc->stsc_data[index].first - 1);
2692 return sc->stsc_data[index].count * (int64_t)chunk_count;
2695 static int mov_read_stps(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2698 MOVStreamContext *sc;
2699 unsigned i, entries;
2701 if (c->fc->nb_streams < 1)
2703 st = c->fc->streams[c->fc->nb_streams-1];
2706 avio_rb32(pb); // version + flags
2708 entries = avio_rb32(pb);
2710 av_log(c->fc, AV_LOG_WARNING, "Duplicated STPS atom\n");
2711 av_free(sc->stps_data);
2713 sc->stps_data = av_malloc_array(entries, sizeof(*sc->stps_data));
2715 return AVERROR(ENOMEM);
2717 for (i = 0; i < entries && !pb->eof_reached; i++) {
2718 sc->stps_data[i] = avio_rb32(pb);
2723 if (pb->eof_reached) {
2724 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STPS atom\n");
2731 static int mov_read_stss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2734 MOVStreamContext *sc;
2735 unsigned int i, entries;
2737 if (c->fc->nb_streams < 1)
2739 st = c->fc->streams[c->fc->nb_streams-1];
2742 avio_r8(pb); /* version */
2743 avio_rb24(pb); /* flags */
2745 entries = avio_rb32(pb);
2747 av_log(c->fc, AV_LOG_TRACE, "keyframe_count = %u\n", entries);
2751 sc->keyframe_absent = 1;
2752 if (!st->need_parsing && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
2753 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2757 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSS atom\n");
2758 if (entries >= UINT_MAX / sizeof(int))
2759 return AVERROR_INVALIDDATA;
2760 av_freep(&sc->keyframes);
2761 sc->keyframe_count = 0;
2762 sc->keyframes = av_malloc_array(entries, sizeof(*sc->keyframes));
2764 return AVERROR(ENOMEM);
2766 for (i = 0; i < entries && !pb->eof_reached; i++) {
2767 sc->keyframes[i] = avio_rb32(pb);
2770 sc->keyframe_count = i;
2772 if (pb->eof_reached) {
2773 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSS atom\n");
2780 static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2783 MOVStreamContext *sc;
2784 unsigned int i, entries, sample_size, field_size, num_bytes;
2789 if (c->fc->nb_streams < 1)
2791 st = c->fc->streams[c->fc->nb_streams-1];
2794 avio_r8(pb); /* version */
2795 avio_rb24(pb); /* flags */
2797 if (atom.type == MKTAG('s','t','s','z')) {
2798 sample_size = avio_rb32(pb);
2799 if (!sc->sample_size) /* do not overwrite value computed in stsd */
2800 sc->sample_size = sample_size;
2801 sc->stsz_sample_size = sample_size;
2805 avio_rb24(pb); /* reserved */
2806 field_size = avio_r8(pb);
2808 entries = avio_rb32(pb);
2810 av_log(c->fc, AV_LOG_TRACE, "sample_size = %u sample_count = %u\n", sc->sample_size, entries);
2812 sc->sample_count = entries;
2816 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
2817 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %u\n", field_size);
2818 return AVERROR_INVALIDDATA;
2823 if (entries >= (UINT_MAX - 4) / field_size)
2824 return AVERROR_INVALIDDATA;
2825 if (sc->sample_sizes)
2826 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSZ atom\n");
2827 av_free(sc->sample_sizes);
2828 sc->sample_count = 0;
2829 sc->sample_sizes = av_malloc_array(entries, sizeof(*sc->sample_sizes));
2830 if (!sc->sample_sizes)
2831 return AVERROR(ENOMEM);
2833 num_bytes = (entries*field_size+4)>>3;
2835 buf = av_malloc(num_bytes+AV_INPUT_BUFFER_PADDING_SIZE);
2837 av_freep(&sc->sample_sizes);
2838 return AVERROR(ENOMEM);
2841 ret = ffio_read_size(pb, buf, num_bytes);
2843 av_freep(&sc->sample_sizes);
2845 av_log(c->fc, AV_LOG_WARNING, "STSZ atom truncated\n");
2849 init_get_bits(&gb, buf, 8*num_bytes);
2851 for (i = 0; i < entries && !pb->eof_reached; i++) {
2852 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
2853 sc->data_size += sc->sample_sizes[i];
2856 sc->sample_count = i;
2860 if (pb->eof_reached) {
2861 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSZ atom\n");
2868 static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2871 MOVStreamContext *sc;
2872 unsigned int i, entries, alloc_size = 0;
2874 int64_t total_sample_count=0;
2876 if (c->fc->nb_streams < 1)
2878 st = c->fc->streams[c->fc->nb_streams-1];
2881 avio_r8(pb); /* version */
2882 avio_rb24(pb); /* flags */
2883 entries = avio_rb32(pb);
2885 av_log(c->fc, AV_LOG_TRACE, "track[%u].stts.entries = %u\n",
2886 c->fc->nb_streams-1, entries);
2889 av_log(c->fc, AV_LOG_WARNING, "Duplicated STTS atom\n");
2890 av_freep(&sc->stts_data);
2892 if (entries >= INT_MAX / sizeof(*sc->stts_data))
2893 return AVERROR(ENOMEM);
2895 for (i = 0; i < entries && !pb->eof_reached; i++) {
2896 int sample_duration;
2897 unsigned int sample_count;
2898 unsigned int min_entries = FFMIN(FFMAX(i + 1, 1024 * 1024), entries);
2899 MOVStts *stts_data = av_fast_realloc(sc->stts_data, &alloc_size,
2900 min_entries * sizeof(*sc->stts_data));
2902 av_freep(&sc->stts_data);
2904 return AVERROR(ENOMEM);
2906 sc->stts_count = min_entries;
2907 sc->stts_data = stts_data;
2909 sample_count=avio_rb32(pb);
2910 sample_duration = avio_rb32(pb);
2912 sc->stts_data[i].count= sample_count;
2913 sc->stts_data[i].duration= sample_duration;
2915 av_log(c->fc, AV_LOG_TRACE, "sample_count=%d, sample_duration=%d\n",
2916 sample_count, sample_duration);
2920 && sample_count == 1
2921 && total_sample_count > 100
2922 && sample_duration/10 > duration / total_sample_count)
2923 sample_duration = duration / total_sample_count;
2924 duration+=(int64_t)sample_duration*(uint64_t)sample_count;
2925 total_sample_count+=sample_count;
2931 duration <= INT64_MAX - sc->duration_for_fps &&
2932 total_sample_count <= INT64_MAX - sc->nb_frames_for_fps
2934 sc->duration_for_fps += duration;
2935 sc->nb_frames_for_fps += total_sample_count;
2938 if (pb->eof_reached) {
2939 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STTS atom\n");
2943 st->nb_frames= total_sample_count;
2945 st->duration= FFMIN(st->duration, duration);
2946 sc->track_end = duration;
2950 static void mov_update_dts_shift(MOVStreamContext *sc, int duration)
2953 if (duration == INT_MIN) {
2954 av_log(NULL, AV_LOG_WARNING, "mov_update_dts_shift(): dts_shift set to %d\n", INT_MAX);
2957 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
2961 static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2964 MOVStreamContext *sc;
2965 unsigned int i, entries, ctts_count = 0;
2967 if (c->fc->nb_streams < 1)
2969 st = c->fc->streams[c->fc->nb_streams-1];
2972 avio_r8(pb); /* version */
2973 avio_rb24(pb); /* flags */
2974 entries = avio_rb32(pb);
2976 av_log(c->fc, AV_LOG_TRACE, "track[%u].ctts.entries = %u\n", c->fc->nb_streams - 1, entries);
2980 if (entries >= UINT_MAX / sizeof(*sc->ctts_data))
2981 return AVERROR_INVALIDDATA;
2982 av_freep(&sc->ctts_data);
2983 sc->ctts_data = av_fast_realloc(NULL, &sc->ctts_allocated_size, entries * sizeof(*sc->ctts_data));
2985 return AVERROR(ENOMEM);
2987 for (i = 0; i < entries && !pb->eof_reached; i++) {
2988 int count =avio_rb32(pb);
2989 int duration =avio_rb32(pb);
2992 av_log(c->fc, AV_LOG_TRACE,
2993 "ignoring CTTS entry with count=%d duration=%d\n",
2998 add_ctts_entry(&sc->ctts_data, &ctts_count, &sc->ctts_allocated_size,
3001 av_log(c->fc, AV_LOG_TRACE, "count=%d, duration=%d\n",
3004 if (FFNABS(duration) < -(1<<28) && i+2<entries) {
3005 av_log(c->fc, AV_LOG_WARNING, "CTTS invalid\n");
3006 av_freep(&sc->ctts_data);
3012 mov_update_dts_shift(sc, duration);
3015 sc->ctts_count = ctts_count;
3017 if (pb->eof_reached) {
3018 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted CTTS atom\n");
3022 av_log(c->fc, AV_LOG_TRACE, "dts shift %d\n", sc->dts_shift);
3027 static int mov_read_sbgp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3030 MOVStreamContext *sc;
3031 unsigned int i, entries;
3033 uint32_t grouping_type;
3035 if (c->fc->nb_streams < 1)
3037 st = c->fc->streams[c->fc->nb_streams-1];
3040 version = avio_r8(pb); /* version */
3041 avio_rb24(pb); /* flags */
3042 grouping_type = avio_rl32(pb);
3043 if (grouping_type != MKTAG( 'r','a','p',' '))
3044 return 0; /* only support 'rap ' grouping */
3046 avio_rb32(pb); /* grouping_type_parameter */
3048 entries = avio_rb32(pb);
3052 av_log(c->fc, AV_LOG_WARNING, "Duplicated SBGP atom\n");
3053 av_free(sc->rap_group);
3054 sc->rap_group_count = 0;
3055 sc->rap_group = av_malloc_array(entries, sizeof(*sc->rap_group));
3057 return AVERROR(ENOMEM);
3059 for (i = 0; i < entries && !pb->eof_reached; i++) {
3060 sc->rap_group[i].count = avio_rb32(pb); /* sample_count */
3061 sc->rap_group[i].index = avio_rb32(pb); /* group_description_index */
3064 sc->rap_group_count = i;
3066 if (pb->eof_reached) {
3067 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted SBGP atom\n");
3075 * Get ith edit list entry (media time, duration).
3077 static int get_edit_list_entry(MOVContext *mov,
3078 const MOVStreamContext *msc,
3079 unsigned int edit_list_index,
3080 int64_t *edit_list_media_time,
3081 int64_t *edit_list_duration,
3082 int64_t global_timescale)
3084 if (edit_list_index == msc->elst_count) {
3087 *edit_list_media_time = msc->elst_data[edit_list_index].time;
3088 *edit_list_duration = msc->elst_data[edit_list_index].duration;
3090 /* duration is in global timescale units;convert to msc timescale */
3091 if (global_timescale == 0) {
3092 avpriv_request_sample(mov->fc, "Support for mvhd.timescale = 0 with editlists");
3095 *edit_list_duration = av_rescale(*edit_list_duration, msc->time_scale,
3101 * Find the closest previous frame to the timestamp_pts, in e_old index
3102 * entries. Searching for just any frame / just key frames can be controlled by
3103 * last argument 'flag'.
3104 * Note that if ctts_data is not NULL, we will always search for a key frame
3105 * irrespective of the value of 'flag'. If we don't find any keyframe, we will
3106 * return the first frame of the video.
3108 * Here the timestamp_pts is considered to be a presentation timestamp and
3109 * the timestamp of index entries are considered to be decoding timestamps.
3111 * Returns 0 if successful in finding a frame, else returns -1.
3112 * Places the found index corresponding output arg.
3114 * If ctts_old is not NULL, then refines the searched entry by searching
3115 * backwards from the found timestamp, to find the frame with correct PTS.
3117 * Places the found ctts_index and ctts_sample in corresponding output args.
3119 static int find_prev_closest_index(AVStream *st,
3120 AVIndexEntry *e_old,
3124 int64_t timestamp_pts,
3127 int64_t* ctts_index,
3128 int64_t* ctts_sample)
3130 MOVStreamContext *msc = st->priv_data;
3131 AVIndexEntry *e_keep = st->index_entries;
3132 int nb_keep = st->nb_index_entries;
3134 int64_t index_ctts_count;
3138 // If dts_shift > 0, then all the index timestamps will have to be offset by
3139 // at least dts_shift amount to obtain PTS.
3140 // Hence we decrement the searched timestamp_pts by dts_shift to find the closest index element.
3141 if (msc->dts_shift > 0) {
3142 timestamp_pts -= msc->dts_shift;
3145 st->index_entries = e_old;
3146 st->nb_index_entries = nb_old;
3147 *index = av_index_search_timestamp(st, timestamp_pts, flag | AVSEEK_FLAG_BACKWARD);
3149 // Keep going backwards in the index entries until the timestamp is the same.
3151 for (i = *index; i > 0 && e_old[i].timestamp == e_old[i - 1].timestamp;
3153 if ((flag & AVSEEK_FLAG_ANY) ||
3154 (e_old[i - 1].flags & AVINDEX_KEYFRAME)) {
3160 // If we have CTTS then refine the search, by searching backwards over PTS
3161 // computed by adding corresponding CTTS durations to index timestamps.
3162 if (ctts_data && *index >= 0) {
3163 av_assert0(ctts_index);
3164 av_assert0(ctts_sample);
3165 // Find out the ctts_index for the found frame.
3168 for (index_ctts_count = 0; index_ctts_count < *index; index_ctts_count++) {
3169 if (*ctts_index < ctts_count) {
3171 if (ctts_data[*ctts_index].count == *ctts_sample) {
3178 while (*index >= 0 && (*ctts_index) >= 0 && (*ctts_index) < ctts_count) {
3179 // Find a "key frame" with PTS <= timestamp_pts (So that we can decode B-frames correctly).
3180 // No need to add dts_shift to the timestamp here becase timestamp_pts has already been
3181 // compensated by dts_shift above.
3182 if ((e_old[*index].timestamp + ctts_data[*ctts_index].duration) <= timestamp_pts &&
3183 (e_old[*index].flags & AVINDEX_KEYFRAME)) {
3188 if (*ctts_sample == 0) {
3190 if (*ctts_index >= 0)
3191 *ctts_sample = ctts_data[*ctts_index].count - 1;
3198 /* restore AVStream state*/
3199 st->index_entries = e_keep;
3200 st->nb_index_entries = nb_keep;
3201 return *index >= 0 ? 0 : -1;
3205 * Add index entry with the given values, to the end of st->index_entries.
3206 * Returns the new size st->index_entries if successful, else returns -1.
3208 * This function is similar to ff_add_index_entry in libavformat/utils.c
3209 * except that here we are always unconditionally adding an index entry to
3210 * the end, instead of searching the entries list and skipping the add if
3211 * there is an existing entry with the same timestamp.
3212 * This is needed because the mov_fix_index calls this func with the same
3213 * unincremented timestamp for successive discarded frames.
3215 static int64_t add_index_entry(AVStream *st, int64_t pos, int64_t timestamp,
3216 int size, int distance, int flags)
3218 AVIndexEntry *entries, *ie;
3220 const size_t min_size_needed = (st->nb_index_entries + 1) * sizeof(AVIndexEntry);
3222 // Double the allocation each time, to lower memory fragmentation.
3223 // Another difference from ff_add_index_entry function.
3224 const size_t requested_size =
3225 min_size_needed > st->index_entries_allocated_size ?
3226 FFMAX(min_size_needed, 2 * st->index_entries_allocated_size) :
3229 if((unsigned)st->nb_index_entries + 1 >= UINT_MAX / sizeof(AVIndexEntry))
3232 entries = av_fast_realloc(st->index_entries,
3233 &st->index_entries_allocated_size,
3238 st->index_entries= entries;
3240 index= st->nb_index_entries++;
3241 ie= &entries[index];
3244 ie->timestamp = timestamp;
3245 ie->min_distance= distance;
3252 * Rewrite timestamps of index entries in the range [end_index - frame_duration_buffer_size, end_index)
3253 * by subtracting end_ts successively by the amounts given in frame_duration_buffer.
3255 static void fix_index_entry_timestamps(AVStream* st, int end_index, int64_t end_ts,
3256 int64_t* frame_duration_buffer,
3257 int frame_duration_buffer_size) {
3259 av_assert0(end_index >= 0 && end_index <= st->nb_index_entries);
3260 for (i = 0; i < frame_duration_buffer_size; i++) {
3261 end_ts -= frame_duration_buffer[frame_duration_buffer_size - 1 - i];
3262 st->index_entries[end_index - 1 - i].timestamp = end_ts;
3267 * Append a new ctts entry to ctts_data.
3268 * Returns the new ctts_count if successful, else returns -1.
3270 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
3271 int count, int duration)
3273 MOVStts *ctts_buf_new;
3274 const size_t min_size_needed = (*ctts_count + 1) * sizeof(MOVStts);
3275 const size_t requested_size =
3276 min_size_needed > *allocated_size ?
3277 FFMAX(min_size_needed, 2 * (*allocated_size)) :
3280 if((unsigned)(*ctts_count) >= UINT_MAX / sizeof(MOVStts) - 1)
3283 ctts_buf_new = av_fast_realloc(*ctts_data, allocated_size, requested_size);
3288 *ctts_data = ctts_buf_new;
3290 ctts_buf_new[*ctts_count].count = count;
3291 ctts_buf_new[*ctts_count].duration = duration;
3293 *ctts_count = (*ctts_count) + 1;
3297 #define MAX_REORDER_DELAY 16
3298 static void mov_estimate_video_delay(MOVContext *c, AVStream* st) {
3299 MOVStreamContext *msc = st->priv_data;
3302 int ctts_sample = 0;
3303 int64_t pts_buf[MAX_REORDER_DELAY + 1]; // Circular buffer to sort pts.
3305 int j, r, num_swaps;
3307 for (j = 0; j < MAX_REORDER_DELAY + 1; j++)
3308 pts_buf[j] = INT64_MIN;
3310 if (st->codecpar->video_delay <= 0 && msc->ctts_data &&
3311 st->codecpar->codec_id == AV_CODEC_ID_H264) {
3312 st->codecpar->video_delay = 0;
3313 for(ind = 0; ind < st->nb_index_entries && ctts_ind < msc->ctts_count; ++ind) {
3314 // Point j to the last elem of the buffer and insert the current pts there.
3316 buf_start = (buf_start + 1);
3317 if (buf_start == MAX_REORDER_DELAY + 1)
3320 pts_buf[j] = st->index_entries[ind].timestamp + msc->ctts_data[ctts_ind].duration;
3322 // The timestamps that are already in the sorted buffer, and are greater than the
3323 // current pts, are exactly the timestamps that need to be buffered to output PTS
3324 // in correct sorted order.
3325 // Hence the video delay (which is the buffer size used to sort DTS and output PTS),
3326 // can be computed as the maximum no. of swaps any particular timestamp needs to
3327 // go through, to keep this buffer in sorted order.
3329 while (j != buf_start) {
3331 if (r < 0) r = MAX_REORDER_DELAY;
3332 if (pts_buf[j] < pts_buf[r]) {
3333 FFSWAP(int64_t, pts_buf[j], pts_buf[r]);
3340 st->codecpar->video_delay = FFMAX(st->codecpar->video_delay, num_swaps);
3343 if (ctts_sample == msc->ctts_data[ctts_ind].count) {
3348 av_log(c->fc, AV_LOG_DEBUG, "Setting codecpar->delay to %d for stream st: %d\n",
3349 st->codecpar->video_delay, st->index);
3353 static void mov_current_sample_inc(MOVStreamContext *sc)
3355 sc->current_sample++;
3356 sc->current_index++;
3357 if (sc->index_ranges &&
3358 sc->current_index >= sc->current_index_range->end &&
3359 sc->current_index_range->end) {
3360 sc->current_index_range++;
3361 sc->current_index = sc->current_index_range->start;
3365 static void mov_current_sample_dec(MOVStreamContext *sc)
3367 sc->current_sample--;
3368 sc->current_index--;
3369 if (sc->index_ranges &&
3370 sc->current_index < sc->current_index_range->start &&
3371 sc->current_index_range > sc->index_ranges) {
3372 sc->current_index_range--;
3373 sc->current_index = sc->current_index_range->end - 1;
3377 static void mov_current_sample_set(MOVStreamContext *sc, int current_sample)
3381 sc->current_sample = current_sample;
3382 sc->current_index = current_sample;
3383 if (!sc->index_ranges) {
3387 for (sc->current_index_range = sc->index_ranges;
3388 sc->current_index_range->end;
3389 sc->current_index_range++) {
3390 range_size = sc->current_index_range->end - sc->current_index_range->start;
3391 if (range_size > current_sample) {
3392 sc->current_index = sc->current_index_range->start + current_sample;
3395 current_sample -= range_size;
3400 * Fix st->index_entries, so that it contains only the entries (and the entries
3401 * which are needed to decode them) that fall in the edit list time ranges.
3402 * Also fixes the timestamps of the index entries to match the timeline
3403 * specified the edit lists.
3405 static void mov_fix_index(MOVContext *mov, AVStream *st)
3407 MOVStreamContext *msc = st->priv_data;
3408 AVIndexEntry *e_old = st->index_entries;
3409 int nb_old = st->nb_index_entries;
3410 const AVIndexEntry *e_old_end = e_old + nb_old;
3411 const AVIndexEntry *current = NULL;
3412 MOVStts *ctts_data_old = msc->ctts_data;
3413 int64_t ctts_index_old = 0;
3414 int64_t ctts_sample_old = 0;
3415 int64_t ctts_count_old = msc->ctts_count;
3416 int64_t edit_list_media_time = 0;
3417 int64_t edit_list_duration = 0;
3418 int64_t frame_duration = 0;
3419 int64_t edit_list_dts_counter = 0;
3420 int64_t edit_list_dts_entry_end = 0;
3421 int64_t edit_list_start_ctts_sample = 0;
3423 int64_t curr_ctts = 0;
3424 int64_t empty_edits_sum_duration = 0;
3425 int64_t edit_list_index = 0;
3428 int64_t start_dts = 0;
3429 int64_t edit_list_start_encountered = 0;
3430 int64_t search_timestamp = 0;
3431 int64_t* frame_duration_buffer = NULL;
3432 int num_discarded_begin = 0;
3433 int first_non_zero_audio_edit = -1;
3434 int packet_skip_samples = 0;
3435 MOVIndexRange *current_index_range;
3437 int found_keyframe_after_edit = 0;
3439 if (!msc->elst_data || msc->elst_count <= 0 || nb_old <= 0) {
3443 // allocate the index ranges array
3444 msc->index_ranges = av_malloc((msc->elst_count + 1) * sizeof(msc->index_ranges[0]));
3445 if (!msc->index_ranges) {
3446 av_log(mov->fc, AV_LOG_ERROR, "Cannot allocate index ranges buffer\n");
3449 msc->current_index_range = msc->index_ranges;
3450 current_index_range = msc->index_ranges - 1;
3452 // Clean AVStream from traces of old index
3453 st->index_entries = NULL;
3454 st->index_entries_allocated_size = 0;
3455 st->nb_index_entries = 0;
3457 // Clean ctts fields of MOVStreamContext
3458 msc->ctts_data = NULL;
3459 msc->ctts_count = 0;
3460 msc->ctts_index = 0;
3461 msc->ctts_sample = 0;
3462 msc->ctts_allocated_size = 0;
3464 // Reinitialize min_corrected_pts so that it can be computed again.
3465 msc->min_corrected_pts = -1;
3467 // If the dts_shift is positive (in case of negative ctts values in mov),
3468 // then negate the DTS by dts_shift
3469 if (msc->dts_shift > 0) {
3470 edit_list_dts_entry_end -= msc->dts_shift;
3471 av_log(mov->fc, AV_LOG_DEBUG, "Shifting DTS by %d because of negative CTTS.\n", msc->dts_shift);
3474 start_dts = edit_list_dts_entry_end;
3476 while (get_edit_list_entry(mov, msc, edit_list_index, &edit_list_media_time,
3477 &edit_list_duration, mov->time_scale)) {
3478 av_log(mov->fc, AV_LOG_DEBUG, "Processing st: %d, edit list %"PRId64" - media time: %"PRId64", duration: %"PRId64"\n",
3479 st->index, edit_list_index, edit_list_media_time, edit_list_duration);
3481 edit_list_dts_counter = edit_list_dts_entry_end;
3482 edit_list_dts_entry_end += edit_list_duration;
3483 num_discarded_begin = 0;
3484 if (edit_list_media_time == -1) {
3485 empty_edits_sum_duration += edit_list_duration;
3489 // If we encounter a non-negative edit list reset the skip_samples/start_pad fields and set them
3490 // according to the edit list below.
3491 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3492 if (first_non_zero_audio_edit < 0) {
3493 first_non_zero_audio_edit = 1;
3495 first_non_zero_audio_edit = 0;
3498 if (first_non_zero_audio_edit > 0)
3499 st->skip_samples = msc->start_pad = 0;
3502 // While reordering frame index according to edit list we must handle properly
3503 // the scenario when edit list entry starts from none key frame.
3504 // We find closest previous key frame and preserve it and consequent frames in index.
3505 // All frames which are outside edit list entry time boundaries will be dropped after decoding.
3506 search_timestamp = edit_list_media_time;
3507 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3508 // Audio decoders like AAC need need a decoder delay samples previous to the current sample,
3509 // to correctly decode this frame. Hence for audio we seek to a frame 1 sec. before the
3510 // edit_list_media_time to cover the decoder delay.
3511 search_timestamp = FFMAX(search_timestamp - msc->time_scale, e_old[0].timestamp);
3514 if (find_prev_closest_index(st, e_old, nb_old, ctts_data_old, ctts_count_old, search_timestamp, 0,
3515 &index, &ctts_index_old, &ctts_sample_old) < 0) {
3516 av_log(mov->fc, AV_LOG_WARNING,
3517 "st: %d edit list: %"PRId64" Missing key frame while searching for timestamp: %"PRId64"\n",
3518 st->index, edit_list_index, search_timestamp);
3519 if (find_prev_closest_index(st, e_old, nb_old, ctts_data_old, ctts_count_old, search_timestamp, AVSEEK_FLAG_ANY,
3520 &index, &ctts_index_old, &ctts_sample_old) < 0) {
3521 av_log(mov->fc, AV_LOG_WARNING,
3522 "st: %d edit list %"PRId64" Cannot find an index entry before timestamp: %"PRId64".\n",
3523 st->index, edit_list_index, search_timestamp);
3526 ctts_sample_old = 0;
3529 current = e_old + index;
3530 edit_list_start_ctts_sample = ctts_sample_old;
3532 // Iterate over index and arrange it according to edit list
3533 edit_list_start_encountered = 0;
3534 found_keyframe_after_edit = 0;
3535 for (; current < e_old_end; current++, index++) {
3536 // check if frame outside edit list mark it for discard
3537 frame_duration = (current + 1 < e_old_end) ?
3538 ((current + 1)->timestamp - current->timestamp) : edit_list_duration;
3540 flags = current->flags;
3542 // frames (pts) before or after edit list
3543 curr_cts = current->timestamp + msc->dts_shift;
3546 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3547 curr_ctts = ctts_data_old[ctts_index_old].duration;
3548 av_log(mov->fc, AV_LOG_DEBUG, "stts: %"PRId64" ctts: %"PRId64", ctts_index: %"PRId64", ctts_count: %"PRId64"\n",
3549 curr_cts, curr_ctts, ctts_index_old, ctts_count_old);
3550 curr_cts += curr_ctts;
3552 if (ctts_sample_old == ctts_data_old[ctts_index_old].count) {
3553 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3554 &msc->ctts_allocated_size,
3555 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3556 ctts_data_old[ctts_index_old].duration) == -1) {
3557 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3559 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3560 ctts_data_old[ctts_index_old].duration);
3564 ctts_sample_old = 0;
3565 edit_list_start_ctts_sample = 0;
3569 if (curr_cts < edit_list_media_time || curr_cts >= (edit_list_duration + edit_list_media_time)) {
3570 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id != AV_CODEC_ID_VORBIS &&
3571 curr_cts < edit_list_media_time && curr_cts + frame_duration > edit_list_media_time &&
3572 first_non_zero_audio_edit > 0) {
3573 packet_skip_samples = edit_list_media_time - curr_cts;
3574 st->skip_samples += packet_skip_samples;
3576 // Shift the index entry timestamp by packet_skip_samples to be correct.
3577 edit_list_dts_counter -= packet_skip_samples;
3578 if (edit_list_start_encountered == 0) {
3579 edit_list_start_encountered = 1;
3580 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3581 // discarded packets.
3582 if (frame_duration_buffer) {
3583 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3584 frame_duration_buffer, num_discarded_begin);
3585 av_freep(&frame_duration_buffer);
3589 av_log(mov->fc, AV_LOG_DEBUG, "skip %d audio samples from curr_cts: %"PRId64"\n", packet_skip_samples, curr_cts);
3591 flags |= AVINDEX_DISCARD_FRAME;
3592 av_log(mov->fc, AV_LOG_DEBUG, "drop a frame at curr_cts: %"PRId64" @ %"PRId64"\n", curr_cts, index);
3594 if (edit_list_start_encountered == 0) {
3595 num_discarded_begin++;
3596 frame_duration_buffer = av_realloc(frame_duration_buffer,
3597 num_discarded_begin * sizeof(int64_t));
3598 if (!frame_duration_buffer) {
3599 av_log(mov->fc, AV_LOG_ERROR, "Cannot reallocate frame duration buffer\n");
3602 frame_duration_buffer[num_discarded_begin - 1] = frame_duration;
3604 // Increment skip_samples for the first non-zero audio edit list
3605 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3606 first_non_zero_audio_edit > 0 && st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
3607 st->skip_samples += frame_duration;
3612 if (msc->min_corrected_pts < 0) {
3613 msc->min_corrected_pts = edit_list_dts_counter + curr_ctts + msc->dts_shift;
3615 msc->min_corrected_pts = FFMIN(msc->min_corrected_pts, edit_list_dts_counter + curr_ctts + msc->dts_shift);
3617 if (edit_list_start_encountered == 0) {
3618 edit_list_start_encountered = 1;
3619 // Make timestamps strictly monotonically increasing by rewriting timestamps for
3620 // discarded packets.
3621 if (frame_duration_buffer) {
3622 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3623 frame_duration_buffer, num_discarded_begin);
3624 av_freep(&frame_duration_buffer);
3629 if (add_index_entry(st, current->pos, edit_list_dts_counter, current->size,
3630 current->min_distance, flags) == -1) {
3631 av_log(mov->fc, AV_LOG_ERROR, "Cannot add index entry\n");
3635 // Update the index ranges array
3636 if (current_index_range < msc->index_ranges || index != current_index_range->end) {
3637 current_index_range++;
3638 current_index_range->start = index;
3640 current_index_range->end = index + 1;
3642 // Only start incrementing DTS in frame_duration amounts, when we encounter a frame in edit list.
3643 if (edit_list_start_encountered > 0) {
3644 edit_list_dts_counter = edit_list_dts_counter + frame_duration;
3647 // Break when found first key frame after edit entry completion
3648 if ((curr_cts + frame_duration >= (edit_list_duration + edit_list_media_time)) &&
3649 ((flags & AVINDEX_KEYFRAME) || ((st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)))) {
3650 if (ctts_data_old) {
3651 // If we have CTTS and this is the first keyframe after edit elist,
3652 // wait for one more, because there might be trailing B-frames after this I-frame
3653 // that do belong to the edit.
3654 if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO && found_keyframe_after_edit == 0) {
3655 found_keyframe_after_edit = 1;
3658 if (ctts_sample_old != 0) {
3659 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3660 &msc->ctts_allocated_size,
3661 ctts_sample_old - edit_list_start_ctts_sample,
3662 ctts_data_old[ctts_index_old].duration) == -1) {
3663 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3664 ctts_index_old, ctts_sample_old - edit_list_start_ctts_sample,
3665 ctts_data_old[ctts_index_old].duration);
3674 // If there are empty edits, then msc->min_corrected_pts might be positive
3675 // intentionally. So we subtract the sum duration of emtpy edits here.
3676 msc->min_corrected_pts -= empty_edits_sum_duration;
3678 // If the minimum pts turns out to be greater than zero after fixing the index, then we subtract the
3679 // dts by that amount to make the first pts zero.
3680 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
3681 if (msc->min_corrected_pts > 0) {
3682 av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by %"PRId64" to make first pts zero.\n", msc->min_corrected_pts);
3683 for (i = 0; i < st->nb_index_entries; ++i) {
3684 st->index_entries[i].timestamp -= msc->min_corrected_pts;
3687 // Start time should be equal to zero or the duration of any empty edits.
3688 st->start_time = empty_edits_sum_duration;
3691 // Update av stream length, if it ends up shorter than the track's media duration
3692 st->duration = FFMIN(st->duration, edit_list_dts_entry_end - start_dts);
3693 msc->start_pad = st->skip_samples;
3695 // Free the old index and the old CTTS structures
3697 av_free(ctts_data_old);
3698 av_freep(&frame_duration_buffer);
3700 // Null terminate the index ranges array
3701 current_index_range++;
3702 current_index_range->start = 0;
3703 current_index_range->end = 0;
3704 msc->current_index = msc->index_ranges[0].start;
3707 static void mov_build_index(MOVContext *mov, AVStream *st)
3709 MOVStreamContext *sc = st->priv_data;
3710 int64_t current_offset;
3711 int64_t current_dts = 0;
3712 unsigned int stts_index = 0;
3713 unsigned int stsc_index = 0;
3714 unsigned int stss_index = 0;
3715 unsigned int stps_index = 0;
3717 uint64_t stream_size = 0;
3718 MOVStts *ctts_data_old = sc->ctts_data;
3719 unsigned int ctts_count_old = sc->ctts_count;
3721 if (sc->elst_count) {
3722 int i, edit_start_index = 0, multiple_edits = 0;
3723 int64_t empty_duration = 0; // empty duration of the first edit list entry
3724 int64_t start_time = 0; // start time of the media
3726 for (i = 0; i < sc->elst_count; i++) {
3727 const MOVElst *e = &sc->elst_data[i];
3728 if (i == 0 && e->time == -1) {
3729 /* if empty, the first entry is the start time of the stream
3730 * relative to the presentation itself */
3731 empty_duration = e->duration;
3732 edit_start_index = 1;
3733 } else if (i == edit_start_index && e->time >= 0) {
3734 start_time = e->time;
3740 if (multiple_edits && !mov->advanced_editlist)
3741 av_log(mov->fc, AV_LOG_WARNING, "multiple edit list entries, "
3742 "Use -advanced_editlist to correctly decode otherwise "
3743 "a/v desync might occur\n");
3745 /* adjust first dts according to edit list */
3746 if ((empty_duration || start_time) && mov->time_scale > 0) {
3748 empty_duration = av_rescale(empty_duration, sc->time_scale, mov->time_scale);
3749 sc->time_offset = start_time - empty_duration;
3750 sc->min_corrected_pts = start_time;
3751 if (!mov->advanced_editlist)
3752 current_dts = -sc->time_offset;
3755 if (!multiple_edits && !mov->advanced_editlist &&
3756 st->codecpar->codec_id == AV_CODEC_ID_AAC && start_time > 0)
3757 sc->start_pad = start_time;
3760 /* only use old uncompressed audio chunk demuxing when stts specifies it */
3761 if (!(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3762 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
3763 unsigned int current_sample = 0;
3764 unsigned int stts_sample = 0;
3765 unsigned int sample_size;
3766 unsigned int distance = 0;
3767 unsigned int rap_group_index = 0;
3768 unsigned int rap_group_sample = 0;
3769 int64_t last_dts = 0;
3770 int64_t dts_correction = 0;
3771 int rap_group_present = sc->rap_group_count && sc->rap_group;
3772 int key_off = (sc->keyframe_count && sc->keyframes[0] > 0) || (sc->stps_count && sc->stps_data[0] > 0);
3774 current_dts -= sc->dts_shift;
3775 last_dts = current_dts;
3777 if (!sc->sample_count || st->nb_index_entries)
3779 if (sc->sample_count >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3781 if (av_reallocp_array(&st->index_entries,
3782 st->nb_index_entries + sc->sample_count,
3783 sizeof(*st->index_entries)) < 0) {
3784 st->nb_index_entries = 0;
3787 st->index_entries_allocated_size = (st->nb_index_entries + sc->sample_count) * sizeof(*st->index_entries);
3789 if (ctts_data_old) {
3790 // Expand ctts entries such that we have a 1-1 mapping with samples
3791 if (sc->sample_count >= UINT_MAX / sizeof(*sc->ctts_data))
3794 sc->ctts_allocated_size = 0;
3795 sc->ctts_data = av_fast_realloc(NULL, &sc->ctts_allocated_size,
3796 sc->sample_count * sizeof(*sc->ctts_data));
3797 if (!sc->ctts_data) {
3798 av_free(ctts_data_old);
3802 memset((uint8_t*)(sc->ctts_data), 0, sc->ctts_allocated_size);
3804 for (i = 0; i < ctts_count_old &&
3805 sc->ctts_count < sc->sample_count; i++)
3806 for (j = 0; j < ctts_data_old[i].count &&
3807 sc->ctts_count < sc->sample_count; j++)
3808 add_ctts_entry(&sc->ctts_data, &sc->ctts_count,
3809 &sc->ctts_allocated_size, 1,
3810 ctts_data_old[i].duration);
3811 av_free(ctts_data_old);
3814 for (i = 0; i < sc->chunk_count; i++) {
3815 int64_t next_offset = i+1 < sc->chunk_count ? sc->chunk_offsets[i+1] : INT64_MAX;
3816 current_offset = sc->chunk_offsets[i];
3817 while (mov_stsc_index_valid(stsc_index, sc->stsc_count) &&
3818 i + 1 == sc->stsc_data[stsc_index + 1].first)
3821 if (next_offset > current_offset && sc->sample_size>0 && sc->sample_size < sc->stsz_sample_size &&
3822 sc->stsc_data[stsc_index].count * (int64_t)sc->stsz_sample_size > next_offset - current_offset) {
3823 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too large), ignoring\n", sc->stsz_sample_size);
3824 sc->stsz_sample_size = sc->sample_size;
3826 if (sc->stsz_sample_size>0 && sc->stsz_sample_size < sc->sample_size) {
3827 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too small), ignoring\n", sc->stsz_sample_size);
3828 sc->stsz_sample_size = sc->sample_size;
3831 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
3833 if (current_sample >= sc->sample_count) {
3834 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
3838 if (!sc->keyframe_absent && (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index])) {
3840 if (stss_index + 1 < sc->keyframe_count)
3842 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
3844 if (stps_index + 1 < sc->stps_count)
3847 if (rap_group_present && rap_group_index < sc->rap_group_count) {
3848 if (sc->rap_group[rap_group_index].index > 0)
3850 if (++rap_group_sample == sc->rap_group[rap_group_index].count) {
3851 rap_group_sample = 0;
3855 if (sc->keyframe_absent
3857 && !rap_group_present
3858 && (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO || (i==0 && j==0)))
3862 sample_size = sc->stsz_sample_size > 0 ? sc->stsz_sample_size : sc->sample_sizes[current_sample];
3863 if (sc->pseudo_stream_id == -1 ||
3864 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
3866 if (sample_size > 0x3FFFFFFF) {
3867 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", sample_size);
3870 e = &st->index_entries[st->nb_index_entries++];
3871 e->pos = current_offset;
3872 e->timestamp = current_dts;
3873 e->size = sample_size;
3874 e->min_distance = distance;
3875 e->flags = keyframe ? AVINDEX_KEYFRAME : 0;
3876 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %u, offset %"PRIx64", dts %"PRId64", "
3877 "size %u, distance %u, keyframe %d\n", st->index, current_sample,
3878 current_offset, current_dts, sample_size, distance, keyframe);
3879 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries < 100)
3880 ff_rfps_add_frame(mov->fc, st, current_dts);
3883 current_offset += sample_size;
3884 stream_size += sample_size;
3886 /* A negative sample duration is invalid based on the spec,
3887 * but some samples need it to correct the DTS. */
3888 if (sc->stts_data[stts_index].duration < 0) {
3889 av_log(mov->fc, AV_LOG_WARNING,
3890 "Invalid SampleDelta %d in STTS, at %d st:%d\n",
3891 sc->stts_data[stts_index].duration, stts_index,
3893 dts_correction += sc->stts_data[stts_index].duration - 1;
3894 sc->stts_data[stts_index].duration = 1;
3896 current_dts += sc->stts_data[stts_index].duration;
3897 if (!dts_correction || current_dts + dts_correction > last_dts) {
3898 current_dts += dts_correction;
3901 /* Avoid creating non-monotonous DTS */
3902 dts_correction += current_dts - last_dts - 1;
3903 current_dts = last_dts + 1;
3905 last_dts = current_dts;
3909 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
3915 if (st->duration > 0)
3916 st->codecpar->bit_rate = stream_size*8*sc->time_scale/st->duration;
3918 unsigned chunk_samples, total = 0;
3920 if (!sc->chunk_count)
3923 // compute total chunk count
3924 for (i = 0; i < sc->stsc_count; i++) {
3925 unsigned count, chunk_count;
3927 chunk_samples = sc->stsc_data[i].count;
3928 if (i != sc->stsc_count - 1 &&
3929 sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
3930 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
3934 if (sc->samples_per_frame >= 160) { // gsm
3935 count = chunk_samples / sc->samples_per_frame;
3936 } else if (sc->samples_per_frame > 1) {
3937 unsigned samples = (1024/sc->samples_per_frame)*sc->samples_per_frame;
3938 count = (chunk_samples+samples-1) / samples;
3940 count = (chunk_samples+1023) / 1024;
3943 if (mov_stsc_index_valid(i, sc->stsc_count))
3944 chunk_count = sc->stsc_data[i+1].first - sc->stsc_data[i].first;
3946 chunk_count = sc->chunk_count - (sc->stsc_data[i].first - 1);
3947 total += chunk_count * count;
3950 av_log(mov->fc, AV_LOG_TRACE, "chunk count %u\n", total);
3951 if (total >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3953 if (av_reallocp_array(&st->index_entries,
3954 st->nb_index_entries + total,
3955 sizeof(*st->index_entries)) < 0) {
3956 st->nb_index_entries = 0;
3959 st->index_entries_allocated_size = (st->nb_index_entries + total) * sizeof(*st->index_entries);
3962 for (i = 0; i < sc->chunk_count; i++) {
3963 current_offset = sc->chunk_offsets[i];
3964 if (mov_stsc_index_valid(stsc_index, sc->stsc_count) &&
3965 i + 1 == sc->stsc_data[stsc_index + 1].first)
3967 chunk_samples = sc->stsc_data[stsc_index].count;
3969 while (chunk_samples > 0) {
3971 unsigned size, samples;
3973 if (sc->samples_per_frame > 1 && !sc->bytes_per_frame) {
3974 avpriv_request_sample(mov->fc,
3975 "Zero bytes per frame, but %d samples per frame",
3976 sc->samples_per_frame);
3980 if (sc->samples_per_frame >= 160) { // gsm
3981 samples = sc->samples_per_frame;
3982 size = sc->bytes_per_frame;
3984 if (sc->samples_per_frame > 1) {
3985 samples = FFMIN((1024 / sc->samples_per_frame)*
3986 sc->samples_per_frame, chunk_samples);
3987 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
3989 samples = FFMIN(1024, chunk_samples);
3990 size = samples * sc->sample_size;
3994 if (st->nb_index_entries >= total) {
3995 av_log(mov->fc, AV_LOG_ERROR, "wrong chunk count %u\n", total);
3998 if (size > 0x3FFFFFFF) {
3999 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", size);
4002 e = &st->index_entries[st->nb_index_entries++];
4003 e->pos = current_offset;
4004 e->timestamp = current_dts;
4006 e->min_distance = 0;
4007 e->flags = AVINDEX_KEYFRAME;
4008 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, chunk %u, offset %"PRIx64", dts %"PRId64", "
4009 "size %u, duration %u\n", st->index, i, current_offset, current_dts,
4012 current_offset += size;
4013 current_dts += samples;
4014 chunk_samples -= samples;
4019 if (!mov->ignore_editlist && mov->advanced_editlist) {
4020 // Fix index according to edit lists.
4021 mov_fix_index(mov, st);
4024 // Update start time of the stream.
4025 if (st->start_time == AV_NOPTS_VALUE && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries > 0) {
4026 st->start_time = st->index_entries[0].timestamp + sc->dts_shift;
4027 if (sc->ctts_data) {
4028 st->start_time += sc->ctts_data[0].duration;
4032 mov_estimate_video_delay(mov, st);
4035 static int test_same_origin(const char *src, const char *ref) {
4045 av_url_split(src_proto, sizeof(src_proto), src_auth, sizeof(src_auth), src_host, sizeof(src_host), &src_port, NULL, 0, src);
4046 av_url_split(ref_proto, sizeof(ref_proto), ref_auth, sizeof(ref_auth), ref_host, sizeof(ref_host), &ref_port, NULL, 0, ref);
4048 if (strlen(src) == 0) {
4050 } else if (strlen(src_auth) + 1 >= sizeof(src_auth) ||
4051 strlen(ref_auth) + 1 >= sizeof(ref_auth) ||
4052 strlen(src_host) + 1 >= sizeof(src_host) ||
4053 strlen(ref_host) + 1 >= sizeof(ref_host)) {
4055 } else if (strcmp(src_proto, ref_proto) ||
4056 strcmp(src_auth, ref_auth) ||
4057 strcmp(src_host, ref_host) ||
4058 src_port != ref_port) {
4064 static int mov_open_dref(MOVContext *c, AVIOContext **pb, const char *src, MOVDref *ref)
4066 /* try relative path, we do not try the absolute because it can leak information about our
4067 system to an attacker */
4068 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
4069 char filename[1025];
4070 const char *src_path;
4073 /* find a source dir */
4074 src_path = strrchr(src, '/');
4080 /* find a next level down to target */
4081 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
4082 if (ref->path[l] == '/') {
4083 if (i == ref->nlvl_to - 1)
4089 /* compose filename if next level down to target was found */
4090 if (i == ref->nlvl_to - 1 && src_path - src < sizeof(filename)) {
4091 memcpy(filename, src, src_path - src);
4092 filename[src_path - src] = 0;
4094 for (i = 1; i < ref->nlvl_from; i++)
4095 av_strlcat(filename, "../", sizeof(filename));
4097 av_strlcat(filename, ref->path + l + 1, sizeof(filename));
4098 if (!c->use_absolute_path) {
4099 int same_origin = test_same_origin(src, filename);
4102 av_log(c->fc, AV_LOG_ERROR,
4103 "Reference with mismatching origin, %s not tried for security reasons, "
4104 "set demuxer option use_absolute_path to allow it anyway\n",
4106 return AVERROR(ENOENT);
4109 if(strstr(ref->path + l + 1, "..") ||
4110 strstr(ref->path + l + 1, ":") ||
4111 (ref->nlvl_from > 1 && same_origin < 0) ||
4112 (filename[0] == '/' && src_path == src))
4113 return AVERROR(ENOENT);
4116 if (strlen(filename) + 1 == sizeof(filename))
4117 return AVERROR(ENOENT);
4118 if (!c->fc->io_open(c->fc, pb, filename, AVIO_FLAG_READ, NULL))
4121 } else if (c->use_absolute_path) {
4122 av_log(c->fc, AV_LOG_WARNING, "Using absolute path on user request, "
4123 "this is a possible security issue\n");
4124 if (!c->fc->io_open(c->fc, pb, ref->path, AVIO_FLAG_READ, NULL))
4127 av_log(c->fc, AV_LOG_ERROR,
4128 "Absolute path %s not tried for security reasons, "
4129 "set demuxer option use_absolute_path to allow absolute paths\n",
4133 return AVERROR(ENOENT);
4136 static void fix_timescale(MOVContext *c, MOVStreamContext *sc)
4138 if (sc->time_scale <= 0) {
4139 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", sc->ffindex);
4140 sc->time_scale = c->time_scale;
4141 if (sc->time_scale <= 0)
4146 static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4149 MOVStreamContext *sc;
4152 st = avformat_new_stream(c->fc, NULL);
4153 if (!st) return AVERROR(ENOMEM);
4154 st->id = c->fc->nb_streams;
4155 sc = av_mallocz(sizeof(MOVStreamContext));
4156 if (!sc) return AVERROR(ENOMEM);
4159 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
4160 sc->ffindex = st->index;
4161 c->trak_index = st->index;
4163 if ((ret = mov_read_default(c, pb, atom)) < 0)
4169 if ((sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
4170 (!sc->sample_size && !sc->sample_count))) ||
4171 (!sc->chunk_count && sc->sample_count)) {
4172 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
4176 if (sc->chunk_count && sc->stsc_count && sc->stsc_data[ sc->stsc_count - 1 ].first > sc->chunk_count) {
4177 av_log(c->fc, AV_LOG_ERROR, "stream %d, contradictionary STSC and STCO\n",
4179 return AVERROR_INVALIDDATA;
4182 fix_timescale(c, sc);
4184 avpriv_set_pts_info(st, 64, 1, sc->time_scale);
4186 mov_build_index(c, st);
4188 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
4189 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
4190 if (c->enable_drefs) {
4191 if (mov_open_dref(c, &sc->pb, c->fc->url, dref) < 0)
4192 av_log(c->fc, AV_LOG_ERROR,
4193 "stream %d, error opening alias: path='%s', dir='%s', "
4194 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
4195 st->index, dref->path, dref->dir, dref->filename,
4196 dref->volume, dref->nlvl_from, dref->nlvl_to);
4198 av_log(c->fc, AV_LOG_WARNING,
4199 "Skipped opening external track: "
4200 "stream %d, alias: path='%s', dir='%s', "
4201 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d."
4202 "Set enable_drefs to allow this.\n",
4203 st->index, dref->path, dref->dir, dref->filename,
4204 dref->volume, dref->nlvl_from, dref->nlvl_to);
4208 sc->pb_is_copied = 1;
4211 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
4212 if (!st->sample_aspect_ratio.num && st->codecpar->width && st->codecpar->height &&
4213 sc->height && sc->width &&
4214 (st->codecpar->width != sc->width || st->codecpar->height != sc->height)) {
4215 st->sample_aspect_ratio = av_d2q(((double)st->codecpar->height * sc->width) /
4216 ((double)st->codecpar->width * sc->height), INT_MAX);
4219 #if FF_API_R_FRAME_RATE
4220 if (sc->stts_count == 1 || (sc->stts_count == 2 && sc->stts_data[1].count == 1))
4221 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
4222 sc->time_scale, sc->stts_data[0].duration, INT_MAX);
4226 // done for ai5q, ai52, ai55, ai1q, ai12 and ai15.
4227 if (!st->codecpar->extradata_size && st->codecpar->codec_id == AV_CODEC_ID_H264 &&
4228 TAG_IS_AVCI(st->codecpar->codec_tag)) {
4229 ret = ff_generate_avci_extradata(st);
4234 switch (st->codecpar->codec_id) {
4235 #if CONFIG_H261_DECODER
4236 case AV_CODEC_ID_H261:
4238 #if CONFIG_H263_DECODER
4239 case AV_CODEC_ID_H263:
4241 #if CONFIG_MPEG4_DECODER
4242 case AV_CODEC_ID_MPEG4:
4244 st->codecpar->width = 0; /* let decoder init width/height */
4245 st->codecpar->height= 0;
4249 // If the duration of the mp3 packets is not constant, then they could need a parser
4250 if (st->codecpar->codec_id == AV_CODEC_ID_MP3
4251 && sc->stts_count > 3
4252 && sc->stts_count*10 > st->nb_frames
4253 && sc->time_scale == st->codecpar->sample_rate) {
4254 st->need_parsing = AVSTREAM_PARSE_FULL;
4256 /* Do not need those anymore. */
4257 av_freep(&sc->chunk_offsets);
4258 av_freep(&sc->sample_sizes);
4259 av_freep(&sc->keyframes);
4260 av_freep(&sc->stts_data);
4261 av_freep(&sc->stps_data);
4262 av_freep(&sc->elst_data);
4263 av_freep(&sc->rap_group);
4268 static int mov_read_ilst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4271 c->itunes_metadata = 1;
4272 ret = mov_read_default(c, pb, atom);
4273 c->itunes_metadata = 0;
4277 static int mov_read_keys(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4286 count = avio_rb32(pb);
4287 if (count > UINT_MAX / sizeof(*c->meta_keys) - 1) {
4288 av_log(c->fc, AV_LOG_ERROR,
4289 "The 'keys' atom with the invalid key count: %"PRIu32"\n", count);
4290 return AVERROR_INVALIDDATA;
4293 c->meta_keys_count = count + 1;
4294 c->meta_keys = av_mallocz(c->meta_keys_count * sizeof(*c->meta_keys));
4296 return AVERROR(ENOMEM);
4298 for (i = 1; i <= count; ++i) {
4299 uint32_t key_size = avio_rb32(pb);
4300 uint32_t type = avio_rl32(pb);
4302 av_log(c->fc, AV_LOG_ERROR,
4303 "The key# %"PRIu32" in meta has invalid size:"
4304 "%"PRIu32"\n", i, key_size);
4305 return AVERROR_INVALIDDATA;
4308 if (type != MKTAG('m','d','t','a')) {
4309 avio_skip(pb, key_size);
4311 c->meta_keys[i] = av_mallocz(key_size + 1);
4312 if (!c->meta_keys[i])
4313 return AVERROR(ENOMEM);
4314 avio_read(pb, c->meta_keys[i], key_size);
4320 static int mov_read_custom(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4322 int64_t end = avio_tell(pb) + atom.size;
4323 uint8_t *key = NULL, *val = NULL, *mean = NULL;
4327 MOVStreamContext *sc;
4329 if (c->fc->nb_streams < 1)
4331 st = c->fc->streams[c->fc->nb_streams-1];
4334 for (i = 0; i < 3; i++) {
4338 if (end - avio_tell(pb) <= 12)
4341 len = avio_rb32(pb);
4342 tag = avio_rl32(pb);
4343 avio_skip(pb, 4); // flags
4345 if (len < 12 || len - 12 > end - avio_tell(pb))
4349 if (tag == MKTAG('m', 'e', 'a', 'n'))
4351 else if (tag == MKTAG('n', 'a', 'm', 'e'))
4353 else if (tag == MKTAG('d', 'a', 't', 'a') && len > 4) {
4360 *p = av_malloc(len + 1);
4362 ret = AVERROR(ENOMEM);
4365 ret = ffio_read_size(pb, *p, len);
4373 if (mean && key && val) {
4374 if (strcmp(key, "iTunSMPB") == 0) {
4375 int priming, remainder, samples;
4376 if(sscanf(val, "%*X %X %X %X", &priming, &remainder, &samples) == 3){
4377 if(priming>0 && priming<16384)
4378 sc->start_pad = priming;
4381 if (strcmp(key, "cdec") != 0) {
4382 av_dict_set(&c->fc->metadata, key, val,
4383 AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL);
4387 av_log(c->fc, AV_LOG_VERBOSE,
4388 "Unhandled or malformed custom metadata of size %"PRId64"\n", atom.size);
4391 avio_seek(pb, end, SEEK_SET);
4398 static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4400 while (atom.size > 8) {
4401 uint32_t tag = avio_rl32(pb);
4403 if (tag == MKTAG('h','d','l','r')) {
4404 avio_seek(pb, -8, SEEK_CUR);
4406 return mov_read_default(c, pb, atom);
4412 // return 1 when matrix is identity, 0 otherwise
4413 #define IS_MATRIX_IDENT(matrix) \
4414 ( (matrix)[0][0] == (1 << 16) && \
4415 (matrix)[1][1] == (1 << 16) && \
4416 (matrix)[2][2] == (1 << 30) && \
4417 !(matrix)[0][1] && !(matrix)[0][2] && \
4418 !(matrix)[1][0] && !(matrix)[1][2] && \
4419 !(matrix)[2][0] && !(matrix)[2][1])
4421 static int mov_read_tkhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4426 int display_matrix[3][3];
4427 int res_display_matrix[3][3] = { { 0 } };
4429 MOVStreamContext *sc;
4433 if (c->fc->nb_streams < 1)
4435 st = c->fc->streams[c->fc->nb_streams-1];
4438 version = avio_r8(pb);
4439 flags = avio_rb24(pb);
4440 st->disposition |= (flags & MOV_TKHD_FLAG_ENABLED) ? AV_DISPOSITION_DEFAULT : 0;
4446 avio_rb32(pb); /* creation time */
4447 avio_rb32(pb); /* modification time */
4449 st->id = (int)avio_rb32(pb); /* track id (NOT 0 !)*/
4450 avio_rb32(pb); /* reserved */
4452 /* highlevel (considering edits) duration in movie timebase */
4453 (version == 1) ? avio_rb64(pb) : avio_rb32(pb);
4454 avio_rb32(pb); /* reserved */
4455 avio_rb32(pb); /* reserved */
4457 avio_rb16(pb); /* layer */
4458 avio_rb16(pb); /* alternate group */
4459 avio_rb16(pb); /* volume */
4460 avio_rb16(pb); /* reserved */
4462 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
4463 // they're kept in fixed point format through all calculations
4464 // save u,v,z to store the whole matrix in the AV_PKT_DATA_DISPLAYMATRIX
4465 // side data, but the scale factor is not needed to calculate aspect ratio
4466 for (i = 0; i < 3; i++) {
4467 display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
4468 display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
4469 display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
4472 width = avio_rb32(pb); // 16.16 fixed point track width
4473 height = avio_rb32(pb); // 16.16 fixed point track height
4474 sc->width = width >> 16;
4475 sc->height = height >> 16;
4477 // apply the moov display matrix (after the tkhd one)
4478 for (i = 0; i < 3; i++) {
4479 const int sh[3] = { 16, 16, 30 };
4480 for (j = 0; j < 3; j++) {
4481 for (e = 0; e < 3; e++) {
4482 res_display_matrix[i][j] +=
4483 ((int64_t) display_matrix[i][e] *
4484 c->movie_display_matrix[e][j]) >> sh[e];
4489 // save the matrix when it is not the default identity
4490 if (!IS_MATRIX_IDENT(res_display_matrix)) {
4493 av_freep(&sc->display_matrix);
4494 sc->display_matrix = av_malloc(sizeof(int32_t) * 9);
4495 if (!sc->display_matrix)
4496 return AVERROR(ENOMEM);
4498 for (i = 0; i < 3; i++)
4499 for (j = 0; j < 3; j++)
4500 sc->display_matrix[i * 3 + j] = res_display_matrix[i][j];
4502 #if FF_API_OLD_ROTATE_API
4503 rotate = av_display_rotation_get(sc->display_matrix);
4504 if (!isnan(rotate)) {
4505 char rotate_buf[64];
4507 if (rotate < 0) // for backward compatibility
4509 snprintf(rotate_buf, sizeof(rotate_buf), "%g", rotate);
4510 av_dict_set(&st->metadata, "rotate", rotate_buf, 0);
4515 // transform the display width/height according to the matrix
4516 // to keep the same scale, use [width height 1<<16]
4517 if (width && height && sc->display_matrix) {
4518 double disp_transform[2];
4520 for (i = 0; i < 2; i++)
4521 disp_transform[i] = hypot(sc->display_matrix[0 + i],
4522 sc->display_matrix[3 + i]);
4524 if (disp_transform[0] > 0 && disp_transform[1] > 0 &&
4525 disp_transform[0] < (1<<24) && disp_transform[1] < (1<<24) &&
4526 fabs((disp_transform[0] / disp_transform[1]) - 1.0) > 0.01)
4527 st->sample_aspect_ratio = av_d2q(
4528 disp_transform[0] / disp_transform[1],
4534 static int mov_read_tfhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4536 MOVFragment *frag = &c->fragment;
4537 MOVTrackExt *trex = NULL;
4538 int flags, track_id, i;
4540 avio_r8(pb); /* version */
4541 flags = avio_rb24(pb);
4543 track_id = avio_rb32(pb);
4545 return AVERROR_INVALIDDATA;
4546 frag->track_id = track_id;
4547 set_frag_stream(&c->frag_index, track_id);
4548 for (i = 0; i < c->trex_count; i++)
4549 if (c->trex_data[i].track_id == frag->track_id) {
4550 trex = &c->trex_data[i];
4554 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
4555 return AVERROR_INVALIDDATA;
4558 frag->base_data_offset = flags & MOV_TFHD_BASE_DATA_OFFSET ?
4559 avio_rb64(pb) : flags & MOV_TFHD_DEFAULT_BASE_IS_MOOF ?
4560 frag->moof_offset : frag->implicit_offset;
4561 frag->stsd_id = flags & MOV_TFHD_STSD_ID ? avio_rb32(pb) : trex->stsd_id;
4563 frag->duration = flags & MOV_TFHD_DEFAULT_DURATION ?
4564 avio_rb32(pb) : trex->duration;
4565 frag->size = flags & MOV_TFHD_DEFAULT_SIZE ?
4566 avio_rb32(pb) : trex->size;
4567 frag->flags = flags & MOV_TFHD_DEFAULT_FLAGS ?
4568 avio_rb32(pb) : trex->flags;
4569 av_log(c->fc, AV_LOG_TRACE, "frag flags 0x%x\n", frag->flags);
4574 static int mov_read_chap(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4579 num = atom.size / 4;
4580 if (!(new_tracks = av_malloc_array(num, sizeof(int))))
4581 return AVERROR(ENOMEM);
4583 av_free(c->chapter_tracks);
4584 c->chapter_tracks = new_tracks;
4585 c->nb_chapter_tracks = num;
4587 for (i = 0; i < num && !pb->eof_reached; i++)
4588 c->chapter_tracks[i] = avio_rb32(pb);
4593 static int mov_read_trex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4598 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
4599 return AVERROR_INVALIDDATA;
4600 if ((err = av_reallocp_array(&c->trex_data, c->trex_count + 1,
4601 sizeof(*c->trex_data))) < 0) {
4606 c->fc->duration = AV_NOPTS_VALUE; // the duration from mvhd is not representing the whole file when fragments are used.
4608 trex = &c->trex_data[c->trex_count++];
4609 avio_r8(pb); /* version */
4610 avio_rb24(pb); /* flags */
4611 trex->track_id = avio_rb32(pb);
4612 trex->stsd_id = avio_rb32(pb);
4613 trex->duration = avio_rb32(pb);
4614 trex->size = avio_rb32(pb);
4615 trex->flags = avio_rb32(pb);
4619 static int mov_read_tfdt(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4621 MOVFragment *frag = &c->fragment;
4622 AVStream *st = NULL;
4623 MOVStreamContext *sc;
4625 MOVFragmentStreamInfo * frag_stream_info;
4626 int64_t base_media_decode_time;
4628 for (i = 0; i < c->fc->nb_streams; i++) {
4629 if (c->fc->streams[i]->id == frag->track_id) {
4630 st = c->fc->streams[i];
4635 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %u\n", frag->track_id);
4636 return AVERROR_INVALIDDATA;
4639 if (sc->pseudo_stream_id + 1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4641 version = avio_r8(pb);
4642 avio_rb24(pb); /* flags */
4644 base_media_decode_time = avio_rb64(pb);
4646 base_media_decode_time = avio_rb32(pb);
4649 frag_stream_info = get_current_frag_stream_info(&c->frag_index);
4650 if (frag_stream_info)
4651 frag_stream_info->tfdt_dts = base_media_decode_time;
4652 sc->track_end = base_media_decode_time;
4657 static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4659 MOVFragment *frag = &c->fragment;
4660 AVStream *st = NULL;
4661 MOVStreamContext *sc;
4664 int64_t dts, pts = AV_NOPTS_VALUE;
4665 int data_offset = 0;
4666 unsigned entries, first_sample_flags = frag->flags;
4667 int flags, distance, i;
4668 int64_t prev_dts = AV_NOPTS_VALUE;
4669 int next_frag_index = -1, index_entry_pos;
4670 size_t requested_size;
4671 size_t old_ctts_allocated_size;
4672 AVIndexEntry *new_entries;
4673 MOVFragmentStreamInfo * frag_stream_info;
4675 for (i = 0; i < c->fc->nb_streams; i++) {
4676 if (c->fc->streams[i]->id == frag->track_id) {
4677 st = c->fc->streams[i];
4682 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %u\n", frag->track_id);
4683 return AVERROR_INVALIDDATA;
4686 if (sc->pseudo_stream_id+1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4689 // Find the next frag_index index that has a valid index_entry for
4690 // the current track_id.
4692 // A valid index_entry means the trun for the fragment was read
4693 // and it's samples are in index_entries at the given position.
4694 // New index entries will be inserted before the index_entry found.
4695 index_entry_pos = st->nb_index_entries;
4696 for (i = c->frag_index.current + 1; i < c->frag_index.nb_items; i++) {
4697 frag_stream_info = get_frag_stream_info(&c->frag_index, i, frag->track_id);
4698 if (frag_stream_info && frag_stream_info->index_entry >= 0) {
4699 next_frag_index = i;
4700 index_entry_pos = frag_stream_info->index_entry;
4705 avio_r8(pb); /* version */
4706 flags = avio_rb24(pb);
4707 entries = avio_rb32(pb);
4708 av_log(c->fc, AV_LOG_TRACE, "flags 0x%x entries %u\n", flags, entries);
4710 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
4711 return AVERROR_INVALIDDATA;
4712 if (flags & MOV_TRUN_DATA_OFFSET) data_offset = avio_rb32(pb);
4713 if (flags & MOV_TRUN_FIRST_SAMPLE_FLAGS) first_sample_flags = avio_rb32(pb);
4715 frag_stream_info = get_current_frag_stream_info(&c->frag_index);
4716 if (frag_stream_info)
4718 if (frag_stream_info->first_tfra_pts != AV_NOPTS_VALUE &&
4719 c->use_mfra_for == FF_MOV_FLAG_MFRA_PTS) {
4720 pts = frag_stream_info->first_tfra_pts;
4721 av_log(c->fc, AV_LOG_DEBUG, "found mfra time %"PRId64
4722 ", using it for pts\n", pts);
4723 } else if (frag_stream_info->sidx_pts != AV_NOPTS_VALUE) {
4724 // FIXME: sidx earliest_presentation_time is *PTS*, s.b.
4725 // pts = frag_stream_info->sidx_pts;
4726 dts = frag_stream_info->sidx_pts - sc->time_offset;
4727 av_log(c->fc, AV_LOG_DEBUG, "found sidx time %"PRId64
4728 ", using it for pts\n", pts);
4729 } else if (frag_stream_info->tfdt_dts != AV_NOPTS_VALUE) {
4730 dts = frag_stream_info->tfdt_dts - sc->time_offset;
4731 av_log(c->fc, AV_LOG_DEBUG, "found tfdt time %"PRId64
4732 ", using it for dts\n", dts);
4734 dts = sc->track_end - sc->time_offset;
4735 av_log(c->fc, AV_LOG_DEBUG, "found track end time %"PRId64
4736 ", using it for dts\n", dts);
4739 dts = sc->track_end - sc->time_offset;
4740 av_log(c->fc, AV_LOG_DEBUG, "found track end time %"PRId64
4741 ", using it for dts\n", dts);
4743 offset = frag->base_data_offset + data_offset;
4745 av_log(c->fc, AV_LOG_TRACE, "first sample flags 0x%x\n", first_sample_flags);
4747 // realloc space for new index entries
4748 if((unsigned)st->nb_index_entries + entries >= UINT_MAX / sizeof(AVIndexEntry)) {
4749 entries = UINT_MAX / sizeof(AVIndexEntry) - st->nb_index_entries;
4750 av_log(c->fc, AV_LOG_ERROR, "Failed to add index entry\n");
4755 requested_size = (st->nb_index_entries + entries) * sizeof(AVIndexEntry);
4756 new_entries = av_fast_realloc(st->index_entries,
4757 &st->index_entries_allocated_size,
4760 return AVERROR(ENOMEM);
4761 st->index_entries= new_entries;
4763 requested_size = (st->nb_index_entries + entries) * sizeof(*sc->ctts_data);
4764 old_ctts_allocated_size = sc->ctts_allocated_size;
4765 ctts_data = av_fast_realloc(sc->ctts_data, &sc->ctts_allocated_size,
4768 return AVERROR(ENOMEM);
4769 sc->ctts_data = ctts_data;
4771 // In case there were samples without ctts entries, ensure they get
4772 // zero valued entries. This ensures clips which mix boxes with and
4773 // without ctts entries don't pickup uninitialized data.
4774 memset((uint8_t*)(sc->ctts_data) + old_ctts_allocated_size, 0,
4775 sc->ctts_allocated_size - old_ctts_allocated_size);
4777 if (index_entry_pos < st->nb_index_entries) {
4778 // Make hole in index_entries and ctts_data for new samples
4779 memmove(st->index_entries + index_entry_pos + entries,
4780 st->index_entries + index_entry_pos,
4781 sizeof(*st->index_entries) *
4782 (st->nb_index_entries - index_entry_pos));
4783 memmove(sc->ctts_data + index_entry_pos + entries,
4784 sc->ctts_data + index_entry_pos,
4785 sizeof(*sc->ctts_data) * (sc->ctts_count - index_entry_pos));
4786 if (index_entry_pos < sc->current_sample) {
4787 sc->current_sample += entries;
4791 st->nb_index_entries += entries;
4792 sc->ctts_count = st->nb_index_entries;
4794 // Record the index_entry position in frag_index of this fragment
4795 if (frag_stream_info)
4796 frag_stream_info->index_entry = index_entry_pos;
4798 if (index_entry_pos > 0)
4799 prev_dts = st->index_entries[index_entry_pos-1].timestamp;
4801 for (i = 0; i < entries && !pb->eof_reached; i++) {
4802 unsigned sample_size = frag->size;
4803 int sample_flags = i ? frag->flags : first_sample_flags;
4804 unsigned sample_duration = frag->duration;
4805 unsigned ctts_duration = 0;
4807 int index_entry_flags = 0;
4809 if (flags & MOV_TRUN_SAMPLE_DURATION) sample_duration = avio_rb32(pb);
4810 if (flags & MOV_TRUN_SAMPLE_SIZE) sample_size = avio_rb32(pb);
4811 if (flags & MOV_TRUN_SAMPLE_FLAGS) sample_flags = avio_rb32(pb);
4812 if (flags & MOV_TRUN_SAMPLE_CTS) ctts_duration = avio_rb32(pb);
4814 mov_update_dts_shift(sc, ctts_duration);
4815 if (pts != AV_NOPTS_VALUE) {
4816 dts = pts - sc->dts_shift;
4817 if (flags & MOV_TRUN_SAMPLE_CTS) {
4818 dts -= ctts_duration;
4820 dts -= sc->time_offset;
4822 av_log(c->fc, AV_LOG_DEBUG,
4823 "pts %"PRId64" calculated dts %"PRId64
4824 " sc->dts_shift %d ctts.duration %d"
4825 " sc->time_offset %"PRId64
4826 " flags & MOV_TRUN_SAMPLE_CTS %d\n",
4828 sc->dts_shift, ctts_duration,
4829 sc->time_offset, flags & MOV_TRUN_SAMPLE_CTS);
4830 pts = AV_NOPTS_VALUE;
4833 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
4837 !(sample_flags & (MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC |
4838 MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES));
4841 index_entry_flags |= AVINDEX_KEYFRAME;
4843 // Fragments can overlap in time. Discard overlapping frames after
4845 if (prev_dts >= dts)
4846 index_entry_flags |= AVINDEX_DISCARD_FRAME;
4848 st->index_entries[index_entry_pos].pos = offset;
4849 st->index_entries[index_entry_pos].timestamp = dts;
4850 st->index_entries[index_entry_pos].size= sample_size;
4851 st->index_entries[index_entry_pos].min_distance= distance;
4852 st->index_entries[index_entry_pos].flags = index_entry_flags;
4854 sc->ctts_data[index_entry_pos].count = 1;
4855 sc->ctts_data[index_entry_pos].duration = ctts_duration;
4858 av_log(c->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
4859 "size %u, distance %d, keyframe %d\n", st->index,
4860 index_entry_pos, offset, dts, sample_size, distance, keyframe);
4862 dts += sample_duration;
4863 offset += sample_size;
4864 sc->data_size += sample_size;
4866 if (sample_duration <= INT64_MAX - sc->duration_for_fps &&
4867 1 <= INT64_MAX - sc->nb_frames_for_fps
4869 sc->duration_for_fps += sample_duration;
4870 sc->nb_frames_for_fps ++;
4874 // EOF found before reading all entries. Fix the hole this would
4875 // leave in index_entries and ctts_data
4876 int gap = entries - i;
4877 memmove(st->index_entries + index_entry_pos,
4878 st->index_entries + index_entry_pos + gap,
4879 sizeof(*st->index_entries) *
4880 (st->nb_index_entries - (index_entry_pos + gap)));
4881 memmove(sc->ctts_data + index_entry_pos,
4882 sc->ctts_data + index_entry_pos + gap,
4883 sizeof(*sc->ctts_data) *
4884 (sc->ctts_count - (index_entry_pos + gap)));
4886 st->nb_index_entries -= gap;
4887 sc->ctts_count -= gap;
4888 if (index_entry_pos < sc->current_sample) {
4889 sc->current_sample -= gap;
4894 // The end of this new fragment may overlap in time with the start
4895 // of the next fragment in index_entries. Mark the samples in the next
4896 // fragment that overlap with AVINDEX_DISCARD_FRAME
4897 prev_dts = AV_NOPTS_VALUE;
4898 if (index_entry_pos > 0)
4899 prev_dts = st->index_entries[index_entry_pos-1].timestamp;
4900 for (i = index_entry_pos; i < st->nb_index_entries; i++) {
4901 if (prev_dts < st->index_entries[i].timestamp)
4903 st->index_entries[i].flags |= AVINDEX_DISCARD_FRAME;
4906 // If a hole was created to insert the new index_entries into,
4907 // the index_entry recorded for all subsequent moof must
4908 // be incremented by the number of entries inserted.
4909 fix_frag_index_entries(&c->frag_index, next_frag_index,
4910 frag->track_id, entries);
4912 if (pb->eof_reached) {
4913 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted TRUN atom\n");
4917 frag->implicit_offset = offset;
4919 sc->track_end = dts + sc->time_offset;
4920 if (st->duration < sc->track_end)
4921 st->duration = sc->track_end;
4926 static int mov_read_sidx(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4928 int64_t offset = avio_tell(pb) + atom.size, pts, timestamp;
4930 unsigned i, j, track_id, item_count;
4931 AVStream *st = NULL;
4932 AVStream *ref_st = NULL;
4933 MOVStreamContext *sc, *ref_sc = NULL;
4934 AVRational timescale;
4936 version = avio_r8(pb);
4938 avpriv_request_sample(c->fc, "sidx version %u", version);
4942 avio_rb24(pb); // flags
4944 track_id = avio_rb32(pb); // Reference ID
4945 for (i = 0; i < c->fc->nb_streams; i++) {
4946 if (c->fc->streams[i]->id == track_id) {
4947 st = c->fc->streams[i];
4952 av_log(c->fc, AV_LOG_WARNING, "could not find corresponding track id %d\n", track_id);
4958 timescale = av_make_q(1, avio_rb32(pb));
4960 if (timescale.den <= 0) {
4961 av_log(c->fc, AV_LOG_ERROR, "Invalid sidx timescale 1/%d\n", timescale.den);
4962 return AVERROR_INVALIDDATA;
4966 pts = avio_rb32(pb);
4967 offset += avio_rb32(pb);
4969 pts = avio_rb64(pb);
4970 offset += avio_rb64(pb);
4973 avio_rb16(pb); // reserved
4975 item_count = avio_rb16(pb);
4977 for (i = 0; i < item_count; i++) {
4979 MOVFragmentStreamInfo * frag_stream_info;
4980 uint32_t size = avio_rb32(pb);
4981 uint32_t duration = avio_rb32(pb);
4982 if (size & 0x80000000) {
4983 avpriv_request_sample(c->fc, "sidx reference_type 1");
4984 return AVERROR_PATCHWELCOME;
4986 avio_rb32(pb); // sap_flags
4987 timestamp = av_rescale_q(pts, st->time_base, timescale);
4989 index = update_frag_index(c, offset);
4990 frag_stream_info = get_frag_stream_info(&c->frag_index, index, track_id);
4991 if (frag_stream_info)
4992 frag_stream_info->sidx_pts = timestamp;
4998 st->duration = sc->track_end = pts;
5002 if (offset == avio_size(pb)) {
5003 // Find first entry in fragment index that came from an sidx.
5004 // This will pretty much always be the first entry.
5005 for (i = 0; i < c->frag_index.nb_items; i++) {
5006 MOVFragmentIndexItem * item = &c->frag_index.item[i];
5007 for (j = 0; ref_st == NULL && j < item->nb_stream_info; j++) {
5008 MOVFragmentStreamInfo * si;
5009 si = &item->stream_info[j];
5010 if (si->sidx_pts != AV_NOPTS_VALUE) {
5011 ref_st = c->fc->streams[j];
5012 ref_sc = ref_st->priv_data;
5017 for (i = 0; i < c->fc->nb_streams; i++) {
5018 st = c->fc->streams[i];
5020 if (!sc->has_sidx) {
5021 st->duration = sc->track_end = av_rescale(ref_st->duration, sc->time_scale, ref_sc->time_scale);
5025 c->frag_index.complete = 1;
5031 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
5032 /* like the files created with Adobe Premiere 5.0, for samples see */
5033 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
5034 static int mov_read_wide(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5039 return 0; /* continue */
5040 if (avio_rb32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
5041 avio_skip(pb, atom.size - 4);
5044 atom.type = avio_rl32(pb);
5046 if (atom.type != MKTAG('m','d','a','t')) {
5047 avio_skip(pb, atom.size);
5050 err = mov_read_mdat(c, pb, atom);
5054 static int mov_read_cmov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5059 uint8_t *moov_data; /* uncompressed data */
5060 long cmov_len, moov_len;
5063 avio_rb32(pb); /* dcom atom */
5064 if (avio_rl32(pb) != MKTAG('d','c','o','m'))
5065 return AVERROR_INVALIDDATA;
5066 if (avio_rl32(pb) != MKTAG('z','l','i','b')) {
5067 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !\n");
5068 return AVERROR_INVALIDDATA;
5070 avio_rb32(pb); /* cmvd atom */
5071 if (avio_rl32(pb) != MKTAG('c','m','v','d'))
5072 return AVERROR_INVALIDDATA;
5073 moov_len = avio_rb32(pb); /* uncompressed size */
5074 cmov_len = atom.size - 6 * 4;
5076 cmov_data = av_malloc(cmov_len);
5078 return AVERROR(ENOMEM);
5079 moov_data = av_malloc(moov_len);
5082 return AVERROR(ENOMEM);
5084 ret = ffio_read_size(pb, cmov_data, cmov_len);
5086 goto free_and_return;
5088 ret = AVERROR_INVALIDDATA;
5089 if (uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
5090 goto free_and_return;
5091 if (ffio_init_context(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
5092 goto free_and_return;
5093 ctx.seekable = AVIO_SEEKABLE_NORMAL;
5094 atom.type = MKTAG('m','o','o','v');
5095 atom.size = moov_len;
5096 ret = mov_read_default(c, &ctx, atom);
5102 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
5103 return AVERROR(ENOSYS);
5107 /* edit list atom */
5108 static int mov_read_elst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5110 MOVStreamContext *sc;
5111 int i, edit_count, version;
5112 int64_t elst_entry_size;
5114 if (c->fc->nb_streams < 1 || c->ignore_editlist)
5116 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
5118 version = avio_r8(pb); /* version */
5119 avio_rb24(pb); /* flags */
5120 edit_count = avio_rb32(pb); /* entries */
5123 elst_entry_size = version == 1 ? 20 : 12;
5124 if (atom.size != edit_count * elst_entry_size) {
5125 if (c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
5126 av_log(c->fc, AV_LOG_ERROR, "Invalid edit list entry_count: %d for elst atom of size: %"PRId64" bytes.\n",
5127 edit_count, atom.size + 8);
5128 return AVERROR_INVALIDDATA;
5130 edit_count = atom.size / elst_entry_size;
5131 if (edit_count * elst_entry_size != atom.size) {
5132 av_log(c->fc, AV_LOG_WARNING, "ELST atom of %"PRId64" bytes, bigger than %d entries.", atom.size, edit_count);
5140 av_log(c->fc, AV_LOG_WARNING, "Duplicated ELST atom\n");
5141 av_free(sc->elst_data);
5143 sc->elst_data = av_malloc_array(edit_count, sizeof(*sc->elst_data));
5145 return AVERROR(ENOMEM);
5147 av_log(c->fc, AV_LOG_TRACE, "track[%u].edit_count = %i\n", c->fc->nb_streams - 1, edit_count);
5148 for (i = 0; i < edit_count && atom.size > 0 && !pb->eof_reached; i++) {
5149 MOVElst *e = &sc->elst_data[i];
5152 e->duration = avio_rb64(pb);
5153 e->time = avio_rb64(pb);
5156 e->duration = avio_rb32(pb); /* segment duration */
5157 e->time = (int32_t)avio_rb32(pb); /* media time */
5160 e->rate = avio_rb32(pb) / 65536.0;
5162 av_log(c->fc, AV_LOG_TRACE, "duration=%"PRId64" time=%"PRId64" rate=%f\n",
5163 e->duration, e->time, e->rate);
5165 if (e->time < 0 && e->time != -1 &&
5166 c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
5167 av_log(c->fc, AV_LOG_ERROR, "Track %d, edit %d: Invalid edit list media time=%"PRId64"\n",
5168 c->fc->nb_streams-1, i, e->time);
5169 return AVERROR_INVALIDDATA;
5177 static int mov_read_tmcd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5179 MOVStreamContext *sc;
5181 if (c->fc->nb_streams < 1)
5182 return AVERROR_INVALIDDATA;
5183 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5184 sc->timecode_track = avio_rb32(pb);
5188 static int mov_read_vpcc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5191 int version, color_range, color_primaries, color_trc, color_space;
5193 if (c->fc->nb_streams < 1)
5195 st = c->fc->streams[c->fc->nb_streams - 1];
5197 if (atom.size < 5) {
5198 av_log(c->fc, AV_LOG_ERROR, "Empty VP Codec Configuration box\n");
5199 return AVERROR_INVALIDDATA;
5202 version = avio_r8(pb);
5204 av_log(c->fc, AV_LOG_WARNING, "Unsupported VP Codec Configuration box version %d\n", version);
5207 avio_skip(pb, 3); /* flags */
5209 avio_skip(pb, 2); /* profile + level */
5210 color_range = avio_r8(pb); /* bitDepth, chromaSubsampling, videoFullRangeFlag */
5211 color_primaries = avio_r8(pb);
5212 color_trc = avio_r8(pb);
5213 color_space = avio_r8(pb);
5214 if (avio_rb16(pb)) /* codecIntializationDataSize */
5215 return AVERROR_INVALIDDATA;
5217 if (!av_color_primaries_name(color_primaries))
5218 color_primaries = AVCOL_PRI_UNSPECIFIED;
5219 if (!av_color_transfer_name(color_trc))
5220 color_trc = AVCOL_TRC_UNSPECIFIED;
5221 if (!av_color_space_name(color_space))
5222 color_space = AVCOL_SPC_UNSPECIFIED;
5224 st->codecpar->color_range = (color_range & 1) ? AVCOL_RANGE_JPEG : AVCOL_RANGE_MPEG;
5225 st->codecpar->color_primaries = color_primaries;
5226 st->codecpar->color_trc = color_trc;
5227 st->codecpar->color_space = color_space;
5232 static int mov_read_smdm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5234 MOVStreamContext *sc;
5235 const int chroma_den = 50000;
5236 const int luma_den = 10000;
5239 if (c->fc->nb_streams < 1)
5240 return AVERROR_INVALIDDATA;
5242 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5244 if (atom.size < 5) {
5245 av_log(c->fc, AV_LOG_ERROR, "Empty Mastering Display Metadata box\n");
5246 return AVERROR_INVALIDDATA;
5249 version = avio_r8(pb);
5251 av_log(c->fc, AV_LOG_WARNING, "Unsupported Mastering Display Metadata box version %d\n", version);
5254 avio_skip(pb, 3); /* flags */
5256 sc->mastering = av_mastering_display_metadata_alloc();
5258 return AVERROR(ENOMEM);
5260 for (i = 0; i < 3; i++)
5261 for (j = 0; j < 2; j++)
5262 sc->mastering->display_primaries[i][j] =
5263 av_make_q(lrint(((double)avio_rb16(pb) / (1 << 16)) * chroma_den), chroma_den);
5264 for (i = 0; i < 2; i++)
5265 sc->mastering->white_point[i] =
5266 av_make_q(lrint(((double)avio_rb16(pb) / (1 << 16)) * chroma_den), chroma_den);
5267 sc->mastering->max_luminance =
5268 av_make_q(lrint(((double)avio_rb32(pb) / (1 << 8)) * luma_den), luma_den);
5269 sc->mastering->min_luminance =
5270 av_make_q(lrint(((double)avio_rb32(pb) / (1 << 14)) * luma_den), luma_den);
5272 sc->mastering->has_primaries = 1;
5273 sc->mastering->has_luminance = 1;
5278 static int mov_read_mdcv(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5280 MOVStreamContext *sc;
5281 const int mapping[3] = {1, 2, 0};
5282 const int chroma_den = 50000;
5283 const int luma_den = 10000;
5286 if (c->fc->nb_streams < 1)
5287 return AVERROR_INVALIDDATA;
5289 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5291 if (atom.size < 24) {
5292 av_log(c->fc, AV_LOG_ERROR, "Invalid Mastering Display Color Volume box\n");
5293 return AVERROR_INVALIDDATA;
5296 sc->mastering = av_mastering_display_metadata_alloc();
5298 return AVERROR(ENOMEM);
5300 for (i = 0; i < 3; i++) {
5301 const int j = mapping[i];
5302 sc->mastering->display_primaries[j][0] = av_make_q(avio_rb16(pb), chroma_den);
5303 sc->mastering->display_primaries[j][1] = av_make_q(avio_rb16(pb), chroma_den);
5305 sc->mastering->white_point[0] = av_make_q(avio_rb16(pb), chroma_den);
5306 sc->mastering->white_point[1] = av_make_q(avio_rb16(pb), chroma_den);
5308 sc->mastering->max_luminance = av_make_q(avio_rb32(pb), luma_den);
5309 sc->mastering->min_luminance = av_make_q(avio_rb32(pb), luma_den);
5311 sc->mastering->has_luminance = 1;
5312 sc->mastering->has_primaries = 1;
5317 static int mov_read_coll(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5319 MOVStreamContext *sc;
5322 if (c->fc->nb_streams < 1)
5323 return AVERROR_INVALIDDATA;
5325 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5327 if (atom.size < 5) {
5328 av_log(c->fc, AV_LOG_ERROR, "Empty Content Light Level box\n");
5329 return AVERROR_INVALIDDATA;
5332 version = avio_r8(pb);
5334 av_log(c->fc, AV_LOG_WARNING, "Unsupported Content Light Level box version %d\n", version);
5337 avio_skip(pb, 3); /* flags */
5339 sc->coll = av_content_light_metadata_alloc(&sc->coll_size);
5341 return AVERROR(ENOMEM);
5343 sc->coll->MaxCLL = avio_rb16(pb);
5344 sc->coll->MaxFALL = avio_rb16(pb);
5349 static int mov_read_clli(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5351 MOVStreamContext *sc;
5353 if (c->fc->nb_streams < 1)
5354 return AVERROR_INVALIDDATA;
5356 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5358 if (atom.size < 4) {
5359 av_log(c->fc, AV_LOG_ERROR, "Empty Content Light Level Info box\n");
5360 return AVERROR_INVALIDDATA;
5363 sc->coll = av_content_light_metadata_alloc(&sc->coll_size);
5365 return AVERROR(ENOMEM);
5367 sc->coll->MaxCLL = avio_rb16(pb);
5368 sc->coll->MaxFALL = avio_rb16(pb);
5373 static int mov_read_st3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5376 MOVStreamContext *sc;
5377 enum AVStereo3DType type;
5380 if (c->fc->nb_streams < 1)
5383 st = c->fc->streams[c->fc->nb_streams - 1];
5386 if (atom.size < 5) {
5387 av_log(c->fc, AV_LOG_ERROR, "Empty stereoscopic video box\n");
5388 return AVERROR_INVALIDDATA;
5390 avio_skip(pb, 4); /* version + flags */
5395 type = AV_STEREO3D_2D;
5398 type = AV_STEREO3D_TOPBOTTOM;
5401 type = AV_STEREO3D_SIDEBYSIDE;
5404 av_log(c->fc, AV_LOG_WARNING, "Unknown st3d mode value %d\n", mode);
5408 sc->stereo3d = av_stereo3d_alloc();
5410 return AVERROR(ENOMEM);
5412 sc->stereo3d->type = type;
5416 static int mov_read_sv3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5419 MOVStreamContext *sc;
5420 int size, version, layout;
5421 int32_t yaw, pitch, roll;
5422 uint32_t l = 0, t = 0, r = 0, b = 0;
5423 uint32_t tag, padding = 0;
5424 enum AVSphericalProjection projection;
5426 if (c->fc->nb_streams < 1)
5429 st = c->fc->streams[c->fc->nb_streams - 1];
5432 if (atom.size < 8) {
5433 av_log(c->fc, AV_LOG_ERROR, "Empty spherical video box\n");
5434 return AVERROR_INVALIDDATA;
5437 size = avio_rb32(pb);
5438 if (size <= 12 || size > atom.size)
5439 return AVERROR_INVALIDDATA;
5441 tag = avio_rl32(pb);
5442 if (tag != MKTAG('s','v','h','d')) {
5443 av_log(c->fc, AV_LOG_ERROR, "Missing spherical video header\n");
5446 version = avio_r8(pb);
5448 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5452 avio_skip(pb, 3); /* flags */
5453 avio_skip(pb, size - 12); /* metadata_source */
5455 size = avio_rb32(pb);
5456 if (size > atom.size)
5457 return AVERROR_INVALIDDATA;
5459 tag = avio_rl32(pb);
5460 if (tag != MKTAG('p','r','o','j')) {
5461 av_log(c->fc, AV_LOG_ERROR, "Missing projection box\n");
5465 size = avio_rb32(pb);
5466 if (size > atom.size)
5467 return AVERROR_INVALIDDATA;
5469 tag = avio_rl32(pb);
5470 if (tag != MKTAG('p','r','h','d')) {
5471 av_log(c->fc, AV_LOG_ERROR, "Missing projection header box\n");
5474 version = avio_r8(pb);
5476 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5480 avio_skip(pb, 3); /* flags */
5482 /* 16.16 fixed point */
5483 yaw = avio_rb32(pb);
5484 pitch = avio_rb32(pb);
5485 roll = avio_rb32(pb);
5487 size = avio_rb32(pb);
5488 if (size > atom.size)
5489 return AVERROR_INVALIDDATA;
5491 tag = avio_rl32(pb);
5492 version = avio_r8(pb);
5494 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5498 avio_skip(pb, 3); /* flags */
5500 case MKTAG('c','b','m','p'):
5501 layout = avio_rb32(pb);
5503 av_log(c->fc, AV_LOG_WARNING,
5504 "Unsupported cubemap layout %d\n", layout);
5507 projection = AV_SPHERICAL_CUBEMAP;
5508 padding = avio_rb32(pb);
5510 case MKTAG('e','q','u','i'):
5516 if (b >= UINT_MAX - t || r >= UINT_MAX - l) {
5517 av_log(c->fc, AV_LOG_ERROR,
5518 "Invalid bounding rectangle coordinates "
5519 "%"PRIu32",%"PRIu32",%"PRIu32",%"PRIu32"\n", l, t, r, b);
5520 return AVERROR_INVALIDDATA;
5523 if (l || t || r || b)
5524 projection = AV_SPHERICAL_EQUIRECTANGULAR_TILE;
5526 projection = AV_SPHERICAL_EQUIRECTANGULAR;
5529 av_log(c->fc, AV_LOG_ERROR, "Unknown projection type: %s\n", av_fourcc2str(tag));
5533 sc->spherical = av_spherical_alloc(&sc->spherical_size);
5535 return AVERROR(ENOMEM);
5537 sc->spherical->projection = projection;
5539 sc->spherical->yaw = yaw;
5540 sc->spherical->pitch = pitch;
5541 sc->spherical->roll = roll;
5543 sc->spherical->padding = padding;
5545 sc->spherical->bound_left = l;
5546 sc->spherical->bound_top = t;
5547 sc->spherical->bound_right = r;
5548 sc->spherical->bound_bottom = b;
5553 static int mov_parse_uuid_spherical(MOVStreamContext *sc, AVIOContext *pb, size_t len)
5556 uint8_t *buffer = av_malloc(len + 1);
5560 return AVERROR(ENOMEM);
5563 ret = ffio_read_size(pb, buffer, len);
5567 /* Check for mandatory keys and values, try to support XML as best-effort */
5568 if (!sc->spherical &&
5569 av_stristr(buffer, "<GSpherical:StitchingSoftware>") &&
5570 (val = av_stristr(buffer, "<GSpherical:Spherical>")) &&
5571 av_stristr(val, "true") &&
5572 (val = av_stristr(buffer, "<GSpherical:Stitched>")) &&
5573 av_stristr(val, "true") &&
5574 (val = av_stristr(buffer, "<GSpherical:ProjectionType>")) &&
5575 av_stristr(val, "equirectangular")) {
5576 sc->spherical = av_spherical_alloc(&sc->spherical_size);
5580 sc->spherical->projection = AV_SPHERICAL_EQUIRECTANGULAR;
5582 if (av_stristr(buffer, "<GSpherical:StereoMode>") && !sc->stereo3d) {
5583 enum AVStereo3DType mode;
5585 if (av_stristr(buffer, "left-right"))
5586 mode = AV_STEREO3D_SIDEBYSIDE;
5587 else if (av_stristr(buffer, "top-bottom"))
5588 mode = AV_STEREO3D_TOPBOTTOM;
5590 mode = AV_STEREO3D_2D;
5592 sc->stereo3d = av_stereo3d_alloc();
5596 sc->stereo3d->type = mode;
5600 val = av_stristr(buffer, "<GSpherical:InitialViewHeadingDegrees>");
5602 sc->spherical->yaw = strtol(val, NULL, 10) * (1 << 16);
5603 val = av_stristr(buffer, "<GSpherical:InitialViewPitchDegrees>");
5605 sc->spherical->pitch = strtol(val, NULL, 10) * (1 << 16);
5606 val = av_stristr(buffer, "<GSpherical:InitialViewRollDegrees>");
5608 sc->spherical->roll = strtol(val, NULL, 10) * (1 << 16);
5616 static int mov_read_uuid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5619 MOVStreamContext *sc;
5622 static const uint8_t uuid_isml_manifest[] = {
5623 0xa5, 0xd4, 0x0b, 0x30, 0xe8, 0x14, 0x11, 0xdd,
5624 0xba, 0x2f, 0x08, 0x00, 0x20, 0x0c, 0x9a, 0x66
5626 static const uint8_t uuid_xmp[] = {
5627 0xbe, 0x7a, 0xcf, 0xcb, 0x97, 0xa9, 0x42, 0xe8,
5628 0x9c, 0x71, 0x99, 0x94, 0x91, 0xe3, 0xaf, 0xac
5630 static const uint8_t uuid_spherical[] = {
5631 0xff, 0xcc, 0x82, 0x63, 0xf8, 0x55, 0x4a, 0x93,
5632 0x88, 0x14, 0x58, 0x7a, 0x02, 0x52, 0x1f, 0xdd,
5635 if (atom.size < sizeof(uuid) || atom.size >= FFMIN(INT_MAX, SIZE_MAX))
5636 return AVERROR_INVALIDDATA;
5638 if (c->fc->nb_streams < 1)
5640 st = c->fc->streams[c->fc->nb_streams - 1];
5643 ret = avio_read(pb, uuid, sizeof(uuid));
5646 } else if (ret != sizeof(uuid)) {
5647 return AVERROR_INVALIDDATA;
5649 if (!memcmp(uuid, uuid_isml_manifest, sizeof(uuid))) {
5650 uint8_t *buffer, *ptr;
5652 size_t len = atom.size - sizeof(uuid);
5655 return AVERROR_INVALIDDATA;
5657 ret = avio_skip(pb, 4); // zeroes
5660 buffer = av_mallocz(len + 1);
5662 return AVERROR(ENOMEM);
5664 ret = avio_read(pb, buffer, len);
5668 } else if (ret != len) {
5670 return AVERROR_INVALIDDATA;
5674 while ((ptr = av_stristr(ptr, "systemBitrate=\""))) {
5675 ptr += sizeof("systemBitrate=\"") - 1;
5676 c->bitrates_count++;
5677 c->bitrates = av_realloc_f(c->bitrates, c->bitrates_count, sizeof(*c->bitrates));
5679 c->bitrates_count = 0;
5681 return AVERROR(ENOMEM);
5684 ret = strtol(ptr, &endptr, 10);
5685 if (ret < 0 || errno || *endptr != '"') {
5686 c->bitrates[c->bitrates_count - 1] = 0;
5688 c->bitrates[c->bitrates_count - 1] = ret;
5693 } else if (!memcmp(uuid, uuid_xmp, sizeof(uuid))) {
5695 size_t len = atom.size - sizeof(uuid);
5696 if (c->export_xmp) {
5697 buffer = av_mallocz(len + 1);
5699 return AVERROR(ENOMEM);
5701 ret = avio_read(pb, buffer, len);
5705 } else if (ret != len) {
5707 return AVERROR_INVALIDDATA;
5710 av_dict_set(&c->fc->metadata, "xmp", buffer, 0);
5713 // skip all uuid atom, which makes it fast for long uuid-xmp file
5714 ret = avio_skip(pb, len);
5718 } else if (!memcmp(uuid, uuid_spherical, sizeof(uuid))) {
5719 size_t len = atom.size - sizeof(uuid);
5720 ret = mov_parse_uuid_spherical(sc, pb, len);
5724 av_log(c->fc, AV_LOG_WARNING, "Invalid spherical metadata found\n");
5730 static int mov_read_free(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5733 uint8_t content[16];
5738 ret = avio_read(pb, content, FFMIN(sizeof(content), atom.size));
5744 && !memcmp(content, "Anevia\x1A\x1A", 8)
5745 && c->use_mfra_for == FF_MOV_FLAG_MFRA_AUTO) {
5746 c->use_mfra_for = FF_MOV_FLAG_MFRA_PTS;
5752 static int mov_read_frma(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5754 uint32_t format = avio_rl32(pb);
5755 MOVStreamContext *sc;
5759 if (c->fc->nb_streams < 1)
5761 st = c->fc->streams[c->fc->nb_streams - 1];
5766 case MKTAG('e','n','c','v'): // encrypted video
5767 case MKTAG('e','n','c','a'): // encrypted audio
5768 id = mov_codec_id(st, format);
5769 if (st->codecpar->codec_id != AV_CODEC_ID_NONE &&
5770 st->codecpar->codec_id != id) {
5771 av_log(c->fc, AV_LOG_WARNING,
5772 "ignoring 'frma' atom of '%.4s', stream has codec id %d\n",
5773 (char*)&format, st->codecpar->codec_id);
5777 st->codecpar->codec_id = id;
5778 sc->format = format;
5782 if (format != sc->format) {
5783 av_log(c->fc, AV_LOG_WARNING,
5784 "ignoring 'frma' atom of '%.4s', stream format is '%.4s'\n",
5785 (char*)&format, (char*)&sc->format);
5794 * Gets the current encryption info and associated current stream context. If
5795 * we are parsing a track fragment, this will return the specific encryption
5796 * info for this fragment; otherwise this will return the global encryption
5797 * info for the current stream.
5799 static int get_current_encryption_info(MOVContext *c, MOVEncryptionIndex **encryption_index, MOVStreamContext **sc)
5801 MOVFragmentStreamInfo *frag_stream_info;
5805 frag_stream_info = get_current_frag_stream_info(&c->frag_index);
5806 if (frag_stream_info) {
5807 for (i = 0; i < c->fc->nb_streams; i++) {
5808 if (c->fc->streams[i]->id == frag_stream_info->id) {
5809 st = c->fc->streams[i];
5813 if (i == c->fc->nb_streams)
5815 *sc = st->priv_data;
5817 if (!frag_stream_info->encryption_index) {
5818 frag_stream_info->encryption_index = av_mallocz(sizeof(*frag_stream_info->encryption_index));
5819 if (!frag_stream_info->encryption_index)
5820 return AVERROR(ENOMEM);
5822 *encryption_index = frag_stream_info->encryption_index;
5825 // No current track fragment, using stream level encryption info.
5827 if (c->fc->nb_streams < 1)
5829 st = c->fc->streams[c->fc->nb_streams - 1];
5830 *sc = st->priv_data;
5832 if (!(*sc)->cenc.encryption_index) {
5833 (*sc)->cenc.encryption_index = av_mallocz(sizeof(*frag_stream_info->encryption_index));
5834 if (!(*sc)->cenc.encryption_index)
5835 return AVERROR(ENOMEM);
5838 *encryption_index = (*sc)->cenc.encryption_index;
5843 static int mov_read_sample_encryption_info(MOVContext *c, AVIOContext *pb, MOVStreamContext *sc, AVEncryptionInfo **sample, int use_subsamples)
5846 unsigned int subsample_count;
5847 AVSubsampleEncryptionInfo *subsamples;
5849 *sample = av_encryption_info_clone(sc->cenc.default_encrypted_sample);
5851 return AVERROR(ENOMEM);
5853 if (sc->cenc.per_sample_iv_size != 0) {
5854 if (avio_read(pb, (*sample)->iv, sc->cenc.per_sample_iv_size) != sc->cenc.per_sample_iv_size) {
5855 av_log(c->fc, AV_LOG_ERROR, "failed to read the initialization vector\n");
5856 av_encryption_info_free(*sample);
5858 return AVERROR_INVALIDDATA;
5862 if (use_subsamples) {
5863 subsample_count = avio_rb16(pb);
5864 av_free((*sample)->subsamples);
5865 (*sample)->subsamples = av_mallocz_array(subsample_count, sizeof(*subsamples));
5866 if (!(*sample)->subsamples) {
5867 av_encryption_info_free(*sample);
5869 return AVERROR(ENOMEM);
5872 for (i = 0; i < subsample_count && !pb->eof_reached; i++) {
5873 (*sample)->subsamples[i].bytes_of_clear_data = avio_rb16(pb);
5874 (*sample)->subsamples[i].bytes_of_protected_data = avio_rb32(pb);
5877 if (pb->eof_reached) {
5878 av_log(c->fc, AV_LOG_ERROR, "hit EOF while reading sub-sample encryption info\n");
5879 av_encryption_info_free(*sample);
5881 return AVERROR_INVALIDDATA;
5883 (*sample)->subsample_count = subsample_count;
5889 static int mov_read_senc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5891 AVEncryptionInfo **encrypted_samples;
5892 MOVEncryptionIndex *encryption_index;
5893 MOVStreamContext *sc;
5894 int use_subsamples, ret;
5895 unsigned int sample_count, i, alloc_size = 0;
5897 ret = get_current_encryption_info(c, &encryption_index, &sc);
5901 if (encryption_index->nb_encrypted_samples) {
5902 // This can happen if we have both saio/saiz and senc atoms.
5903 av_log(c->fc, AV_LOG_DEBUG, "Ignoring duplicate encryption info in senc\n");
5907 avio_r8(pb); /* version */
5908 use_subsamples = avio_rb24(pb) & 0x02; /* flags */
5910 sample_count = avio_rb32(pb);
5911 if (sample_count >= INT_MAX / sizeof(*encrypted_samples))
5912 return AVERROR(ENOMEM);
5914 for (i = 0; i < sample_count; i++) {
5915 unsigned int min_samples = FFMIN(FFMAX(i + 1, 1024 * 1024), sample_count);
5916 encrypted_samples = av_fast_realloc(encryption_index->encrypted_samples, &alloc_size,
5917 min_samples * sizeof(*encrypted_samples));
5918 if (encrypted_samples) {
5919 encryption_index->encrypted_samples = encrypted_samples;
5921 ret = mov_read_sample_encryption_info(
5922 c, pb, sc, &encryption_index->encrypted_samples[i], use_subsamples);
5924 ret = AVERROR(ENOMEM);
5926 if (pb->eof_reached) {
5927 av_log(c->fc, AV_LOG_ERROR, "Hit EOF while reading senc\n");
5928 ret = AVERROR_INVALIDDATA;
5933 av_encryption_info_free(encryption_index->encrypted_samples[i - 1]);
5934 av_freep(&encryption_index->encrypted_samples);
5938 encryption_index->nb_encrypted_samples = sample_count;
5943 static int mov_parse_auxiliary_info(MOVContext *c, MOVStreamContext *sc, AVIOContext *pb, MOVEncryptionIndex *encryption_index)
5945 AVEncryptionInfo **sample, **encrypted_samples;
5947 size_t sample_count, sample_info_size, i;
5949 unsigned int alloc_size = 0;
5951 if (encryption_index->nb_encrypted_samples)
5953 sample_count = encryption_index->auxiliary_info_sample_count;
5954 if (encryption_index->auxiliary_offsets_count != 1) {
5955 av_log(c->fc, AV_LOG_ERROR, "Multiple auxiliary info chunks are not supported\n");
5956 return AVERROR_PATCHWELCOME;
5958 if (sample_count >= INT_MAX / sizeof(*encrypted_samples))
5959 return AVERROR(ENOMEM);
5961 prev_pos = avio_tell(pb);
5962 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) ||
5963 avio_seek(pb, encryption_index->auxiliary_offsets[0], SEEK_SET) != encryption_index->auxiliary_offsets[0]) {
5964 av_log(c->fc, AV_LOG_INFO, "Failed to seek for auxiliary info, will only parse senc atoms for encryption info\n");
5968 for (i = 0; i < sample_count && !pb->eof_reached; i++) {
5969 unsigned int min_samples = FFMIN(FFMAX(i + 1, 1024 * 1024), sample_count);
5970 encrypted_samples = av_fast_realloc(encryption_index->encrypted_samples, &alloc_size,
5971 min_samples * sizeof(*encrypted_samples));
5972 if (!encrypted_samples) {
5973 ret = AVERROR(ENOMEM);
5976 encryption_index->encrypted_samples = encrypted_samples;
5978 sample = &encryption_index->encrypted_samples[i];
5979 sample_info_size = encryption_index->auxiliary_info_default_size
5980 ? encryption_index->auxiliary_info_default_size
5981 : encryption_index->auxiliary_info_sizes[i];
5983 ret = mov_read_sample_encryption_info(c, pb, sc, sample, sample_info_size > sc->cenc.per_sample_iv_size);
5987 if (pb->eof_reached) {
5988 av_log(c->fc, AV_LOG_ERROR, "Hit EOF while reading auxiliary info\n");
5989 ret = AVERROR_INVALIDDATA;
5991 encryption_index->nb_encrypted_samples = sample_count;
5995 avio_seek(pb, prev_pos, SEEK_SET);
5997 for (; i > 0; i--) {
5998 av_encryption_info_free(encryption_index->encrypted_samples[i - 1]);
6000 av_freep(&encryption_index->encrypted_samples);
6006 * Tries to read the given number of bytes from the stream and puts it in a
6007 * newly allocated buffer. This reads in small chunks to avoid allocating large
6008 * memory if the file contains an invalid/malicious size value.
6010 static int mov_try_read_block(AVIOContext *pb, size_t size, uint8_t **data)
6012 const unsigned int block_size = 1024 * 1024;
6013 uint8_t *buffer = NULL;
6014 unsigned int alloc_size = 0, offset = 0;
6015 while (offset < size) {
6016 unsigned int new_size =
6017 alloc_size >= INT_MAX - block_size ? INT_MAX : alloc_size + block_size;
6018 uint8_t *new_buffer = av_fast_realloc(buffer, &alloc_size, new_size);
6019 unsigned int to_read = FFMIN(size, alloc_size) - offset;
6022 return AVERROR(ENOMEM);
6024 buffer = new_buffer;
6026 if (avio_read(pb, buffer + offset, to_read) != to_read) {
6028 return AVERROR_INVALIDDATA;
6037 static int mov_read_saiz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6039 MOVEncryptionIndex *encryption_index;
6040 MOVStreamContext *sc;
6042 unsigned int sample_count, aux_info_type, aux_info_param;
6044 ret = get_current_encryption_info(c, &encryption_index, &sc);
6048 if (encryption_index->nb_encrypted_samples) {
6049 // This can happen if we have both saio/saiz and senc atoms.
6050 av_log(c->fc, AV_LOG_DEBUG, "Ignoring duplicate encryption info in saiz\n");
6054 if (encryption_index->auxiliary_info_sample_count) {
6055 av_log(c->fc, AV_LOG_ERROR, "Duplicate saiz atom\n");
6056 return AVERROR_INVALIDDATA;
6059 avio_r8(pb); /* version */
6060 if (avio_rb24(pb) & 0x01) { /* flags */
6061 aux_info_type = avio_rb32(pb);
6062 aux_info_param = avio_rb32(pb);
6063 if (sc->cenc.default_encrypted_sample) {
6064 if (aux_info_type != sc->cenc.default_encrypted_sample->scheme) {
6065 av_log(c->fc, AV_LOG_DEBUG, "Ignoring saiz box with non-zero aux_info_type\n");
6068 if (aux_info_param != 0) {
6069 av_log(c->fc, AV_LOG_DEBUG, "Ignoring saiz box with non-zero aux_info_type_parameter\n");
6073 // Didn't see 'schm' or 'tenc', so this isn't encrypted.
6074 if ((aux_info_type == MKBETAG('c','e','n','c') ||
6075 aux_info_type == MKBETAG('c','e','n','s') ||
6076 aux_info_type == MKBETAG('c','b','c','1') ||
6077 aux_info_type == MKBETAG('c','b','c','s')) &&
6078 aux_info_param == 0) {
6079 av_log(c->fc, AV_LOG_ERROR, "Saw encrypted saiz without schm/tenc\n");
6080 return AVERROR_INVALIDDATA;
6085 } else if (!sc->cenc.default_encrypted_sample) {
6086 // Didn't see 'schm' or 'tenc', so this isn't encrypted.
6090 encryption_index->auxiliary_info_default_size = avio_r8(pb);
6091 sample_count = avio_rb32(pb);
6092 encryption_index->auxiliary_info_sample_count = sample_count;
6094 if (encryption_index->auxiliary_info_default_size == 0) {
6095 ret = mov_try_read_block(pb, sample_count, &encryption_index->auxiliary_info_sizes);
6097 av_log(c->fc, AV_LOG_ERROR, "Failed to read the auxiliary info\n");
6102 if (encryption_index->auxiliary_offsets_count) {
6103 return mov_parse_auxiliary_info(c, sc, pb, encryption_index);
6109 static int mov_read_saio(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6111 uint64_t *auxiliary_offsets;
6112 MOVEncryptionIndex *encryption_index;
6113 MOVStreamContext *sc;
6115 unsigned int version, entry_count, aux_info_type, aux_info_param;
6116 unsigned int alloc_size = 0;
6118 ret = get_current_encryption_info(c, &encryption_index, &sc);
6122 if (encryption_index->nb_encrypted_samples) {
6123 // This can happen if we have both saio/saiz and senc atoms.
6124 av_log(c->fc, AV_LOG_DEBUG, "Ignoring duplicate encryption info in saio\n");
6128 if (encryption_index->auxiliary_offsets_count) {
6129 av_log(c->fc, AV_LOG_ERROR, "Duplicate saio atom\n");
6130 return AVERROR_INVALIDDATA;
6133 version = avio_r8(pb); /* version */
6134 if (avio_rb24(pb) & 0x01) { /* flags */
6135 aux_info_type = avio_rb32(pb);
6136 aux_info_param = avio_rb32(pb);
6137 if (sc->cenc.default_encrypted_sample) {
6138 if (aux_info_type != sc->cenc.default_encrypted_sample->scheme) {
6139 av_log(c->fc, AV_LOG_DEBUG, "Ignoring saio box with non-zero aux_info_type\n");
6142 if (aux_info_param != 0) {
6143 av_log(c->fc, AV_LOG_DEBUG, "Ignoring saio box with non-zero aux_info_type_parameter\n");
6147 // Didn't see 'schm' or 'tenc', so this isn't encrypted.
6148 if ((aux_info_type == MKBETAG('c','e','n','c') ||
6149 aux_info_type == MKBETAG('c','e','n','s') ||
6150 aux_info_type == MKBETAG('c','b','c','1') ||
6151 aux_info_type == MKBETAG('c','b','c','s')) &&
6152 aux_info_param == 0) {
6153 av_log(c->fc, AV_LOG_ERROR, "Saw encrypted saio without schm/tenc\n");
6154 return AVERROR_INVALIDDATA;
6159 } else if (!sc->cenc.default_encrypted_sample) {
6160 // Didn't see 'schm' or 'tenc', so this isn't encrypted.
6164 entry_count = avio_rb32(pb);
6165 if (entry_count >= INT_MAX / sizeof(*auxiliary_offsets))
6166 return AVERROR(ENOMEM);
6168 for (i = 0; i < entry_count && !pb->eof_reached; i++) {
6169 unsigned int min_offsets = FFMIN(FFMAX(i + 1, 1024), entry_count);
6170 auxiliary_offsets = av_fast_realloc(
6171 encryption_index->auxiliary_offsets, &alloc_size,
6172 min_offsets * sizeof(*auxiliary_offsets));
6173 if (!auxiliary_offsets) {
6174 av_freep(&encryption_index->auxiliary_offsets);
6175 return AVERROR(ENOMEM);
6177 encryption_index->auxiliary_offsets = auxiliary_offsets;
6180 encryption_index->auxiliary_offsets[i] = avio_rb32(pb);
6182 encryption_index->auxiliary_offsets[i] = avio_rb64(pb);
6184 if (c->frag_index.current >= 0) {
6185 encryption_index->auxiliary_offsets[i] += c->fragment.base_data_offset;
6189 if (pb->eof_reached) {
6190 av_log(c->fc, AV_LOG_ERROR, "Hit EOF while reading saio\n");
6191 av_freep(&encryption_index->auxiliary_offsets);
6192 return AVERROR_INVALIDDATA;
6195 encryption_index->auxiliary_offsets_count = entry_count;
6197 if (encryption_index->auxiliary_info_sample_count) {
6198 return mov_parse_auxiliary_info(c, sc, pb, encryption_index);
6204 static int mov_read_pssh(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6206 AVEncryptionInitInfo *info, *old_init_info;
6209 uint8_t *side_data, *extra_data, *old_side_data;
6210 size_t side_data_size;
6211 int ret = 0, old_side_data_size;
6212 unsigned int version, kid_count, extra_data_size, alloc_size = 0;
6214 if (c->fc->nb_streams < 1)
6216 st = c->fc->streams[c->fc->nb_streams-1];
6218 version = avio_r8(pb); /* version */
6219 avio_rb24(pb); /* flags */
6221 info = av_encryption_init_info_alloc(/* system_id_size */ 16, /* num_key_ids */ 0,
6222 /* key_id_size */ 16, /* data_size */ 0);
6224 return AVERROR(ENOMEM);
6226 if (avio_read(pb, info->system_id, 16) != 16) {
6227 av_log(c->fc, AV_LOG_ERROR, "Failed to read the system id\n");
6228 ret = AVERROR_INVALIDDATA;
6233 kid_count = avio_rb32(pb);
6234 if (kid_count >= INT_MAX / sizeof(*key_ids))
6235 return AVERROR(ENOMEM);
6237 for (unsigned int i = 0; i < kid_count && !pb->eof_reached; i++) {
6238 unsigned int min_kid_count = FFMIN(FFMAX(i + 1, 1024), kid_count);
6239 key_ids = av_fast_realloc(info->key_ids, &alloc_size,
6240 min_kid_count * sizeof(*key_ids));
6242 ret = AVERROR(ENOMEM);
6245 info->key_ids = key_ids;
6247 info->key_ids[i] = av_mallocz(16);
6248 if (!info->key_ids[i]) {
6249 ret = AVERROR(ENOMEM);
6252 info->num_key_ids = i + 1;
6254 if (avio_read(pb, info->key_ids[i], 16) != 16) {
6255 av_log(c->fc, AV_LOG_ERROR, "Failed to read the key id\n");
6256 ret = AVERROR_INVALIDDATA;
6261 if (pb->eof_reached) {
6262 av_log(c->fc, AV_LOG_ERROR, "Hit EOF while reading pssh\n");
6263 ret = AVERROR_INVALIDDATA;
6268 extra_data_size = avio_rb32(pb);
6269 ret = mov_try_read_block(pb, extra_data_size, &extra_data);
6273 av_freep(&info->data); // malloc(0) may still allocate something.
6274 info->data = extra_data;
6275 info->data_size = extra_data_size;
6277 // If there is existing initialization data, append to the list.
6278 old_side_data = av_stream_get_side_data(st, AV_PKT_DATA_ENCRYPTION_INIT_INFO, &old_side_data_size);
6279 if (old_side_data) {
6280 old_init_info = av_encryption_init_info_get_side_data(old_side_data, old_side_data_size);
6281 if (old_init_info) {
6282 // Append to the end of the list.
6283 for (AVEncryptionInitInfo *cur = old_init_info;; cur = cur->next) {
6289 info = old_init_info;
6291 // Assume existing side-data will be valid, so the only error we could get is OOM.
6292 ret = AVERROR(ENOMEM);
6297 side_data = av_encryption_init_info_add_side_data(info, &side_data_size);
6299 ret = AVERROR(ENOMEM);
6302 ret = av_stream_add_side_data(st, AV_PKT_DATA_ENCRYPTION_INIT_INFO,
6303 side_data, side_data_size);
6308 av_encryption_init_info_free(info);
6312 static int mov_read_schm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6315 MOVStreamContext *sc;
6317 if (c->fc->nb_streams < 1)
6319 st = c->fc->streams[c->fc->nb_streams-1];
6322 if (sc->pseudo_stream_id != 0) {
6323 av_log(c->fc, AV_LOG_ERROR, "schm boxes are only supported in first sample descriptor\n");
6324 return AVERROR_PATCHWELCOME;
6328 return AVERROR_INVALIDDATA;
6330 avio_rb32(pb); /* version and flags */
6332 if (!sc->cenc.default_encrypted_sample) {
6333 sc->cenc.default_encrypted_sample = av_encryption_info_alloc(0, 16, 16);
6334 if (!sc->cenc.default_encrypted_sample) {
6335 return AVERROR(ENOMEM);
6339 sc->cenc.default_encrypted_sample->scheme = avio_rb32(pb);
6343 static int mov_read_tenc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6346 MOVStreamContext *sc;
6347 unsigned int version, pattern, is_protected, iv_size;
6349 if (c->fc->nb_streams < 1)
6351 st = c->fc->streams[c->fc->nb_streams-1];
6354 if (sc->pseudo_stream_id != 0) {
6355 av_log(c->fc, AV_LOG_ERROR, "tenc atom are only supported in first sample descriptor\n");
6356 return AVERROR_PATCHWELCOME;
6359 if (!sc->cenc.default_encrypted_sample) {
6360 sc->cenc.default_encrypted_sample = av_encryption_info_alloc(0, 16, 16);
6361 if (!sc->cenc.default_encrypted_sample) {
6362 return AVERROR(ENOMEM);
6367 return AVERROR_INVALIDDATA;
6369 version = avio_r8(pb); /* version */
6370 avio_rb24(pb); /* flags */
6372 avio_r8(pb); /* reserved */
6373 pattern = avio_r8(pb);
6376 sc->cenc.default_encrypted_sample->crypt_byte_block = pattern >> 4;
6377 sc->cenc.default_encrypted_sample->skip_byte_block = pattern & 0xf;
6380 is_protected = avio_r8(pb);
6381 if (is_protected && !sc->cenc.encryption_index) {
6382 // The whole stream should be by-default encrypted.
6383 sc->cenc.encryption_index = av_mallocz(sizeof(MOVEncryptionIndex));
6384 if (!sc->cenc.encryption_index)
6385 return AVERROR(ENOMEM);
6387 sc->cenc.per_sample_iv_size = avio_r8(pb);
6388 if (sc->cenc.per_sample_iv_size != 0 && sc->cenc.per_sample_iv_size != 8 &&
6389 sc->cenc.per_sample_iv_size != 16) {
6390 av_log(c->fc, AV_LOG_ERROR, "invalid per-sample IV size value\n");
6391 return AVERROR_INVALIDDATA;
6393 if (avio_read(pb, sc->cenc.default_encrypted_sample->key_id, 16) != 16) {
6394 av_log(c->fc, AV_LOG_ERROR, "failed to read the default key ID\n");
6395 return AVERROR_INVALIDDATA;
6398 if (is_protected && !sc->cenc.per_sample_iv_size) {
6399 iv_size = avio_r8(pb);
6400 if (iv_size != 8 && iv_size != 16) {
6401 av_log(c->fc, AV_LOG_ERROR, "invalid default_constant_IV_size in tenc atom\n");
6402 return AVERROR_INVALIDDATA;
6405 if (avio_read(pb, sc->cenc.default_encrypted_sample->iv, iv_size) != iv_size) {
6406 av_log(c->fc, AV_LOG_ERROR, "failed to read the default IV\n");
6407 return AVERROR_INVALIDDATA;
6414 static int mov_read_dfla(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6417 int last, type, size, ret;
6420 if (c->fc->nb_streams < 1)
6422 st = c->fc->streams[c->fc->nb_streams-1];
6424 if ((uint64_t)atom.size > (1<<30) || atom.size < 42)
6425 return AVERROR_INVALIDDATA;
6427 /* Check FlacSpecificBox version. */
6428 if (avio_r8(pb) != 0)
6429 return AVERROR_INVALIDDATA;
6431 avio_rb24(pb); /* Flags */
6433 avio_read(pb, buf, sizeof(buf));
6434 flac_parse_block_header(buf, &last, &type, &size);
6436 if (type != FLAC_METADATA_TYPE_STREAMINFO || size != FLAC_STREAMINFO_SIZE) {
6437 av_log(c->fc, AV_LOG_ERROR, "STREAMINFO must be first FLACMetadataBlock\n");
6438 return AVERROR_INVALIDDATA;
6441 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
6446 av_log(c->fc, AV_LOG_WARNING, "non-STREAMINFO FLACMetadataBlock(s) ignored\n");
6451 static int cenc_decrypt(MOVContext *c, MOVStreamContext *sc, AVEncryptionInfo *sample, uint8_t *input, int size)
6455 if (sample->scheme != MKBETAG('c','e','n','c') || sample->crypt_byte_block != 0 || sample->skip_byte_block != 0) {
6456 av_log(c->fc, AV_LOG_ERROR, "Only the 'cenc' encryption scheme is supported\n");
6457 return AVERROR_PATCHWELCOME;
6460 if (!sc->cenc.aes_ctr) {
6461 /* initialize the cipher */
6462 sc->cenc.aes_ctr = av_aes_ctr_alloc();
6463 if (!sc->cenc.aes_ctr) {
6464 return AVERROR(ENOMEM);
6467 ret = av_aes_ctr_init(sc->cenc.aes_ctr, c->decryption_key);
6473 av_aes_ctr_set_full_iv(sc->cenc.aes_ctr, sample->iv);
6475 if (!sample->subsample_count)
6477 /* decrypt the whole packet */
6478 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, size);
6482 for (i = 0; i < sample->subsample_count; i++)
6484 if (sample->subsamples[i].bytes_of_clear_data + sample->subsamples[i].bytes_of_protected_data > size) {
6485 av_log(c->fc, AV_LOG_ERROR, "subsample size exceeds the packet size left\n");
6486 return AVERROR_INVALIDDATA;
6489 /* skip the clear bytes */
6490 input += sample->subsamples[i].bytes_of_clear_data;
6491 size -= sample->subsamples[i].bytes_of_clear_data;
6493 /* decrypt the encrypted bytes */
6494 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, sample->subsamples[i].bytes_of_protected_data);
6495 input += sample->subsamples[i].bytes_of_protected_data;
6496 size -= sample->subsamples[i].bytes_of_protected_data;
6500 av_log(c->fc, AV_LOG_ERROR, "leftover packet bytes after subsample processing\n");
6501 return AVERROR_INVALIDDATA;
6507 static int cenc_filter(MOVContext *mov, MOVStreamContext *sc, AVPacket *pkt, int current_index)
6509 MOVFragmentStreamInfo *frag_stream_info;
6510 MOVEncryptionIndex *encryption_index;
6511 AVEncryptionInfo *encrypted_sample;
6512 int encrypted_index, ret;
6514 frag_stream_info = get_current_frag_stream_info(&mov->frag_index);
6515 encrypted_index = current_index;
6516 encryption_index = NULL;
6517 if (frag_stream_info) {
6518 // Note this only supports encryption info in the first sample descriptor.
6519 if (mov->fragment.stsd_id == 1) {
6520 if (frag_stream_info->encryption_index) {
6521 encrypted_index = current_index - frag_stream_info->index_entry;
6522 encryption_index = frag_stream_info->encryption_index;
6524 encryption_index = sc->cenc.encryption_index;
6528 encryption_index = sc->cenc.encryption_index;
6531 if (encryption_index) {
6532 if (encryption_index->auxiliary_info_sample_count &&
6533 !encryption_index->nb_encrypted_samples) {
6534 av_log(mov->fc, AV_LOG_ERROR, "saiz atom found without saio\n");
6535 return AVERROR_INVALIDDATA;
6537 if (encryption_index->auxiliary_offsets_count &&
6538 !encryption_index->nb_encrypted_samples) {
6539 av_log(mov->fc, AV_LOG_ERROR, "saio atom found without saiz\n");
6540 return AVERROR_INVALIDDATA;
6543 if (!encryption_index->nb_encrypted_samples) {
6544 // Full-sample encryption with default settings.
6545 encrypted_sample = sc->cenc.default_encrypted_sample;
6546 } else if (encrypted_index >= 0 && encrypted_index < encryption_index->nb_encrypted_samples) {
6547 // Per-sample setting override.
6548 encrypted_sample = encryption_index->encrypted_samples[encrypted_index];
6550 av_log(mov->fc, AV_LOG_ERROR, "Incorrect number of samples in encryption info\n");
6551 return AVERROR_INVALIDDATA;
6554 if (mov->decryption_key) {
6555 return cenc_decrypt(mov, sc, encrypted_sample, pkt->data, pkt->size);
6558 uint8_t *side_data = av_encryption_info_add_side_data(encrypted_sample, &size);
6560 return AVERROR(ENOMEM);
6561 ret = av_packet_add_side_data(pkt, AV_PKT_DATA_ENCRYPTION_INFO, side_data, size);
6571 static int mov_read_dops(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6573 const int OPUS_SEEK_PREROLL_MS = 80;
6578 if (c->fc->nb_streams < 1)
6580 st = c->fc->streams[c->fc->nb_streams-1];
6582 if ((uint64_t)atom.size > (1<<30) || atom.size < 11)
6583 return AVERROR_INVALIDDATA;
6585 /* Check OpusSpecificBox version. */
6586 if (avio_r8(pb) != 0) {
6587 av_log(c->fc, AV_LOG_ERROR, "unsupported OpusSpecificBox version\n");
6588 return AVERROR_INVALIDDATA;
6591 /* OpusSpecificBox size plus magic for Ogg OpusHead header. */
6592 size = atom.size + 8;
6594 if (ff_alloc_extradata(st->codecpar, size))
6595 return AVERROR(ENOMEM);
6597 AV_WL32(st->codecpar->extradata, MKTAG('O','p','u','s'));
6598 AV_WL32(st->codecpar->extradata + 4, MKTAG('H','e','a','d'));
6599 AV_WB8(st->codecpar->extradata + 8, 1); /* OpusHead version */
6600 avio_read(pb, st->codecpar->extradata + 9, size - 9);
6602 /* OpusSpecificBox is stored in big-endian, but OpusHead is
6603 little-endian; aside from the preceeding magic and version they're
6604 otherwise currently identical. Data after output gain at offset 16
6605 doesn't need to be bytewapped. */
6606 pre_skip = AV_RB16(st->codecpar->extradata + 10);
6607 AV_WL16(st->codecpar->extradata + 10, pre_skip);
6608 AV_WL32(st->codecpar->extradata + 12, AV_RB32(st->codecpar->extradata + 12));
6609 AV_WL16(st->codecpar->extradata + 16, AV_RB16(st->codecpar->extradata + 16));
6611 st->codecpar->initial_padding = pre_skip;
6612 st->codecpar->seek_preroll = av_rescale_q(OPUS_SEEK_PREROLL_MS,
6613 (AVRational){1, 1000},
6614 (AVRational){1, 48000});
6619 static const MOVParseTableEntry mov_default_parse_table[] = {
6620 { MKTAG('A','C','L','R'), mov_read_aclr },
6621 { MKTAG('A','P','R','G'), mov_read_avid },
6622 { MKTAG('A','A','L','P'), mov_read_avid },
6623 { MKTAG('A','R','E','S'), mov_read_ares },
6624 { MKTAG('a','v','s','s'), mov_read_avss },
6625 { MKTAG('c','h','p','l'), mov_read_chpl },
6626 { MKTAG('c','o','6','4'), mov_read_stco },
6627 { MKTAG('c','o','l','r'), mov_read_colr },
6628 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
6629 { MKTAG('d','i','n','f'), mov_read_default },
6630 { MKTAG('D','p','x','E'), mov_read_dpxe },
6631 { MKTAG('d','r','e','f'), mov_read_dref },
6632 { MKTAG('e','d','t','s'), mov_read_default },
6633 { MKTAG('e','l','s','t'), mov_read_elst },
6634 { MKTAG('e','n','d','a'), mov_read_enda },
6635 { MKTAG('f','i','e','l'), mov_read_fiel },
6636 { MKTAG('a','d','r','m'), mov_read_adrm },
6637 { MKTAG('f','t','y','p'), mov_read_ftyp },
6638 { MKTAG('g','l','b','l'), mov_read_glbl },
6639 { MKTAG('h','d','l','r'), mov_read_hdlr },
6640 { MKTAG('i','l','s','t'), mov_read_ilst },
6641 { MKTAG('j','p','2','h'), mov_read_jp2h },
6642 { MKTAG('m','d','a','t'), mov_read_mdat },
6643 { MKTAG('m','d','h','d'), mov_read_mdhd },
6644 { MKTAG('m','d','i','a'), mov_read_default },
6645 { MKTAG('m','e','t','a'), mov_read_meta },
6646 { MKTAG('m','i','n','f'), mov_read_default },
6647 { MKTAG('m','o','o','f'), mov_read_moof },
6648 { MKTAG('m','o','o','v'), mov_read_moov },
6649 { MKTAG('m','v','e','x'), mov_read_default },
6650 { MKTAG('m','v','h','d'), mov_read_mvhd },
6651 { MKTAG('S','M','I',' '), mov_read_svq3 },
6652 { MKTAG('a','l','a','c'), mov_read_alac }, /* alac specific atom */
6653 { MKTAG('a','v','c','C'), mov_read_glbl },
6654 { MKTAG('p','a','s','p'), mov_read_pasp },
6655 { MKTAG('s','i','d','x'), mov_read_sidx },
6656 { MKTAG('s','t','b','l'), mov_read_default },
6657 { MKTAG('s','t','c','o'), mov_read_stco },
6658 { MKTAG('s','t','p','s'), mov_read_stps },
6659 { MKTAG('s','t','r','f'), mov_read_strf },
6660 { MKTAG('s','t','s','c'), mov_read_stsc },
6661 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
6662 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
6663 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
6664 { MKTAG('s','t','t','s'), mov_read_stts },
6665 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
6666 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
6667 { MKTAG('t','f','d','t'), mov_read_tfdt },
6668 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
6669 { MKTAG('t','r','a','k'), mov_read_trak },
6670 { MKTAG('t','r','a','f'), mov_read_default },
6671 { MKTAG('t','r','e','f'), mov_read_default },
6672 { MKTAG('t','m','c','d'), mov_read_tmcd },
6673 { MKTAG('c','h','a','p'), mov_read_chap },
6674 { MKTAG('t','r','e','x'), mov_read_trex },
6675 { MKTAG('t','r','u','n'), mov_read_trun },
6676 { MKTAG('u','d','t','a'), mov_read_default },
6677 { MKTAG('w','a','v','e'), mov_read_wave },
6678 { MKTAG('e','s','d','s'), mov_read_esds },
6679 { MKTAG('d','a','c','3'), mov_read_dac3 }, /* AC-3 info */
6680 { MKTAG('d','e','c','3'), mov_read_dec3 }, /* EAC-3 info */
6681 { MKTAG('d','d','t','s'), mov_read_ddts }, /* DTS audio descriptor */
6682 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
6683 { MKTAG('w','f','e','x'), mov_read_wfex },
6684 { MKTAG('c','m','o','v'), mov_read_cmov },
6685 { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */
6686 { MKTAG('d','v','c','1'), mov_read_dvc1 },
6687 { MKTAG('s','b','g','p'), mov_read_sbgp },
6688 { MKTAG('h','v','c','C'), mov_read_glbl },
6689 { MKTAG('u','u','i','d'), mov_read_uuid },
6690 { MKTAG('C','i','n', 0x8e), mov_read_targa_y216 },
6691 { MKTAG('f','r','e','e'), mov_read_free },
6692 { MKTAG('-','-','-','-'), mov_read_custom },
6693 { MKTAG('s','i','n','f'), mov_read_default },
6694 { MKTAG('f','r','m','a'), mov_read_frma },
6695 { MKTAG('s','e','n','c'), mov_read_senc },
6696 { MKTAG('s','a','i','z'), mov_read_saiz },
6697 { MKTAG('s','a','i','o'), mov_read_saio },
6698 { MKTAG('p','s','s','h'), mov_read_pssh },
6699 { MKTAG('s','c','h','m'), mov_read_schm },
6700 { MKTAG('s','c','h','i'), mov_read_default },
6701 { MKTAG('t','e','n','c'), mov_read_tenc },
6702 { MKTAG('d','f','L','a'), mov_read_dfla },
6703 { MKTAG('s','t','3','d'), mov_read_st3d }, /* stereoscopic 3D video box */
6704 { MKTAG('s','v','3','d'), mov_read_sv3d }, /* spherical video box */
6705 { MKTAG('d','O','p','s'), mov_read_dops },
6706 { MKTAG('S','m','D','m'), mov_read_smdm },
6707 { MKTAG('C','o','L','L'), mov_read_coll },
6708 { MKTAG('v','p','c','C'), mov_read_vpcc },
6709 { MKTAG('m','d','c','v'), mov_read_mdcv },
6710 { MKTAG('c','l','l','i'), mov_read_clli },
6714 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6716 int64_t total_size = 0;
6720 if (c->atom_depth > 10) {
6721 av_log(c->fc, AV_LOG_ERROR, "Atoms too deeply nested\n");
6722 return AVERROR_INVALIDDATA;
6727 atom.size = INT64_MAX;
6728 while (total_size <= atom.size - 8 && !avio_feof(pb)) {
6729 int (*parse)(MOVContext*, AVIOContext*, MOVAtom) = NULL;
6732 if (atom.size >= 8) {
6733 a.size = avio_rb32(pb);
6734 a.type = avio_rl32(pb);
6735 if (a.type == MKTAG('f','r','e','e') &&
6737 c->fc->strict_std_compliance < FF_COMPLIANCE_STRICT &&
6740 uint32_t *type = (uint32_t *)buf + 1;
6741 if (avio_read(pb, buf, 8) != 8)
6742 return AVERROR_INVALIDDATA;
6743 avio_seek(pb, -8, SEEK_CUR);
6744 if (*type == MKTAG('m','v','h','d') ||
6745 *type == MKTAG('c','m','o','v')) {
6746 av_log(c->fc, AV_LOG_ERROR, "Detected moov in a free atom.\n");
6747 a.type = MKTAG('m','o','o','v');
6750 if (atom.type != MKTAG('r','o','o','t') &&
6751 atom.type != MKTAG('m','o','o','v'))
6753 if (a.type == MKTAG('t','r','a','k') || a.type == MKTAG('m','d','a','t'))
6755 av_log(c->fc, AV_LOG_ERROR, "Broken file, trak/mdat not at top-level\n");
6762 if (a.size == 1 && total_size + 8 <= atom.size) { /* 64 bit extended size */
6763 a.size = avio_rb64(pb) - 8;
6767 av_log(c->fc, AV_LOG_TRACE, "type:'%s' parent:'%s' sz: %"PRId64" %"PRId64" %"PRId64"\n",
6768 av_fourcc2str(a.type), av_fourcc2str(atom.type), a.size, total_size, atom.size);
6770 a.size = atom.size - total_size + 8;
6775 a.size = FFMIN(a.size, atom.size - total_size);
6777 for (i = 0; mov_default_parse_table[i].type; i++)
6778 if (mov_default_parse_table[i].type == a.type) {
6779 parse = mov_default_parse_table[i].parse;
6783 // container is user data
6784 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
6785 atom.type == MKTAG('i','l','s','t')))
6786 parse = mov_read_udta_string;
6788 // Supports parsing the QuickTime Metadata Keys.
6789 // https://developer.apple.com/library/mac/documentation/QuickTime/QTFF/Metadata/Metadata.html
6790 if (!parse && c->found_hdlr_mdta &&
6791 atom.type == MKTAG('m','e','t','a') &&
6792 a.type == MKTAG('k','e','y','s')) {
6793 parse = mov_read_keys;
6796 if (!parse) { /* skip leaf atoms data */
6797 avio_skip(pb, a.size);
6799 int64_t start_pos = avio_tell(pb);
6801 int err = parse(c, pb, a);
6806 if (c->found_moov && c->found_mdat &&
6807 ((!(pb->seekable & AVIO_SEEKABLE_NORMAL) || c->fc->flags & AVFMT_FLAG_IGNIDX || c->frag_index.complete) ||
6808 start_pos + a.size == avio_size(pb))) {
6809 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) || c->fc->flags & AVFMT_FLAG_IGNIDX || c->frag_index.complete)
6810 c->next_root_atom = start_pos + a.size;
6814 left = a.size - avio_tell(pb) + start_pos;
6815 if (left > 0) /* skip garbage at atom end */
6816 avio_skip(pb, left);
6817 else if (left < 0) {
6818 av_log(c->fc, AV_LOG_WARNING,
6819 "overread end of atom '%.4s' by %"PRId64" bytes\n",
6820 (char*)&a.type, -left);
6821 avio_seek(pb, left, SEEK_CUR);
6825 total_size += a.size;
6828 if (total_size < atom.size && atom.size < 0x7ffff)
6829 avio_skip(pb, atom.size - total_size);
6835 static int mov_probe(AVProbeData *p)
6840 int moov_offset = -1;
6842 /* check file header */
6845 /* ignore invalid offset */
6846 if ((offset + 8) > (unsigned int)p->buf_size)
6848 tag = AV_RL32(p->buf + offset + 4);
6850 /* check for obvious tags */
6851 case MKTAG('m','o','o','v'):
6852 moov_offset = offset + 4;
6853 case MKTAG('m','d','a','t'):
6854 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
6855 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
6856 case MKTAG('f','t','y','p'):
6857 if (AV_RB32(p->buf+offset) < 8 &&
6858 (AV_RB32(p->buf+offset) != 1 ||
6859 offset + 12 > (unsigned int)p->buf_size ||
6860 AV_RB64(p->buf+offset + 8) == 0)) {
6861 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
6862 } else if (tag == MKTAG('f','t','y','p') &&
6863 ( AV_RL32(p->buf + offset + 8) == MKTAG('j','p','2',' ')
6864 || AV_RL32(p->buf + offset + 8) == MKTAG('j','p','x',' ')
6866 score = FFMAX(score, 5);
6868 score = AVPROBE_SCORE_MAX;
6870 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6872 /* those are more common words, so rate then a bit less */
6873 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
6874 case MKTAG('w','i','d','e'):
6875 case MKTAG('f','r','e','e'):
6876 case MKTAG('j','u','n','k'):
6877 case MKTAG('p','i','c','t'):
6878 score = FFMAX(score, AVPROBE_SCORE_MAX - 5);
6879 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6881 case MKTAG(0x82,0x82,0x7f,0x7d):
6882 case MKTAG('s','k','i','p'):
6883 case MKTAG('u','u','i','d'):
6884 case MKTAG('p','r','f','l'):
6885 /* if we only find those cause probedata is too small at least rate them */
6886 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
6887 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6890 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6893 if(score > AVPROBE_SCORE_MAX - 50 && moov_offset != -1) {
6894 /* moov atom in the header - we should make sure that this is not a
6895 * MOV-packed MPEG-PS */
6896 offset = moov_offset;
6898 while(offset < (p->buf_size - 16)){ /* Sufficient space */
6899 /* We found an actual hdlr atom */
6900 if(AV_RL32(p->buf + offset ) == MKTAG('h','d','l','r') &&
6901 AV_RL32(p->buf + offset + 8) == MKTAG('m','h','l','r') &&
6902 AV_RL32(p->buf + offset + 12) == MKTAG('M','P','E','G')){
6903 av_log(NULL, AV_LOG_WARNING, "Found media data tag MPEG indicating this is a MOV-packed MPEG-PS.\n");
6904 /* We found a media handler reference atom describing an
6905 * MPEG-PS-in-MOV, return a
6906 * low score to force expanding the probe window until
6907 * mpegps_probe finds what it needs */
6918 // must be done after parsing all trak because there's no order requirement
6919 static void mov_read_chapters(AVFormatContext *s)
6921 MOVContext *mov = s->priv_data;
6923 MOVStreamContext *sc;
6928 for (j = 0; j < mov->nb_chapter_tracks; j++) {
6929 chapter_track = mov->chapter_tracks[j];
6931 for (i = 0; i < s->nb_streams; i++)
6932 if (s->streams[i]->id == chapter_track) {
6937 av_log(s, AV_LOG_ERROR, "Referenced QT chapter track not found\n");
6942 cur_pos = avio_tell(sc->pb);
6944 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
6945 st->disposition |= AV_DISPOSITION_ATTACHED_PIC | AV_DISPOSITION_TIMED_THUMBNAILS;
6946 if (st->nb_index_entries) {
6947 // Retrieve the first frame, if possible
6949 AVIndexEntry *sample = &st->index_entries[0];
6950 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
6951 av_log(s, AV_LOG_ERROR, "Failed to retrieve first frame\n");
6955 if (av_get_packet(sc->pb, &pkt, sample->size) < 0)
6958 st->attached_pic = pkt;
6959 st->attached_pic.stream_index = st->index;
6960 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
6963 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
6964 st->codecpar->codec_id = AV_CODEC_ID_BIN_DATA;
6965 st->discard = AVDISCARD_ALL;
6966 for (i = 0; i < st->nb_index_entries; i++) {
6967 AVIndexEntry *sample = &st->index_entries[i];
6968 int64_t end = i+1 < st->nb_index_entries ? st->index_entries[i+1].timestamp : st->duration;
6973 if (end < sample->timestamp) {
6974 av_log(s, AV_LOG_WARNING, "ignoring stream duration which is shorter than chapters\n");
6975 end = AV_NOPTS_VALUE;
6978 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
6979 av_log(s, AV_LOG_ERROR, "Chapter %d not found in file\n", i);
6983 // the first two bytes are the length of the title
6984 len = avio_rb16(sc->pb);
6985 if (len > sample->size-2)
6987 title_len = 2*len + 1;
6988 if (!(title = av_mallocz(title_len)))
6991 // The samples could theoretically be in any encoding if there's an encd
6992 // atom following, but in practice are only utf-8 or utf-16, distinguished
6993 // instead by the presence of a BOM
6997 ch = avio_rb16(sc->pb);
6999 avio_get_str16be(sc->pb, len, title, title_len);
7000 else if (ch == 0xfffe)
7001 avio_get_str16le(sc->pb, len, title, title_len);
7004 if (len == 1 || len == 2)
7007 avio_get_str(sc->pb, INT_MAX, title + 2, len - 1);
7011 avpriv_new_chapter(s, i, st->time_base, sample->timestamp, end, title);
7016 avio_seek(sc->pb, cur_pos, SEEK_SET);
7020 static int parse_timecode_in_framenum_format(AVFormatContext *s, AVStream *st,
7021 uint32_t value, int flags)
7024 char buf[AV_TIMECODE_STR_SIZE];
7025 AVRational rate = st->avg_frame_rate;
7026 int ret = av_timecode_init(&tc, rate, flags, 0, s);
7029 av_dict_set(&st->metadata, "timecode",
7030 av_timecode_make_string(&tc, buf, value), 0);
7034 static int mov_read_rtmd_track(AVFormatContext *s, AVStream *st)
7036 MOVStreamContext *sc = st->priv_data;
7037 char buf[AV_TIMECODE_STR_SIZE];
7038 int64_t cur_pos = avio_tell(sc->pb);
7039 int hh, mm, ss, ff, drop;
7041 if (!st->nb_index_entries)
7044 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
7045 avio_skip(s->pb, 13);
7046 hh = avio_r8(s->pb);
7047 mm = avio_r8(s->pb);
7048 ss = avio_r8(s->pb);
7049 drop = avio_r8(s->pb);
7050 ff = avio_r8(s->pb);
7051 snprintf(buf, AV_TIMECODE_STR_SIZE, "%02d:%02d:%02d%c%02d",
7052 hh, mm, ss, drop ? ';' : ':', ff);
7053 av_dict_set(&st->metadata, "timecode", buf, 0);
7055 avio_seek(sc->pb, cur_pos, SEEK_SET);
7059 static int mov_read_timecode_track(AVFormatContext *s, AVStream *st)
7061 MOVStreamContext *sc = st->priv_data;
7063 int64_t cur_pos = avio_tell(sc->pb);
7066 if (!st->nb_index_entries)
7069 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
7070 value = avio_rb32(s->pb);
7072 if (sc->tmcd_flags & 0x0001) flags |= AV_TIMECODE_FLAG_DROPFRAME;
7073 if (sc->tmcd_flags & 0x0002) flags |= AV_TIMECODE_FLAG_24HOURSMAX;
7074 if (sc->tmcd_flags & 0x0004) flags |= AV_TIMECODE_FLAG_ALLOWNEGATIVE;
7076 /* Assume Counter flag is set to 1 in tmcd track (even though it is likely
7077 * not the case) and thus assume "frame number format" instead of QT one.
7078 * No sample with tmcd track can be found with a QT timecode at the moment,
7079 * despite what the tmcd track "suggests" (Counter flag set to 0 means QT
7081 parse_timecode_in_framenum_format(s, st, value, flags);
7083 avio_seek(sc->pb, cur_pos, SEEK_SET);
7087 static void mov_free_encryption_index(MOVEncryptionIndex **index) {
7089 if (!index || !*index) return;
7090 for (i = 0; i < (*index)->nb_encrypted_samples; i++) {
7091 av_encryption_info_free((*index)->encrypted_samples[i]);
7093 av_freep(&(*index)->encrypted_samples);
7094 av_freep(&(*index)->auxiliary_info_sizes);
7095 av_freep(&(*index)->auxiliary_offsets);
7099 static int mov_read_close(AVFormatContext *s)
7101 MOVContext *mov = s->priv_data;
7104 for (i = 0; i < s->nb_streams; i++) {
7105 AVStream *st = s->streams[i];
7106 MOVStreamContext *sc = st->priv_data;
7111 av_freep(&sc->ctts_data);
7112 for (j = 0; j < sc->drefs_count; j++) {
7113 av_freep(&sc->drefs[j].path);
7114 av_freep(&sc->drefs[j].dir);
7116 av_freep(&sc->drefs);
7118 sc->drefs_count = 0;
7120 if (!sc->pb_is_copied)
7121 ff_format_io_close(s, &sc->pb);
7124 av_freep(&sc->chunk_offsets);
7125 av_freep(&sc->stsc_data);
7126 av_freep(&sc->sample_sizes);
7127 av_freep(&sc->keyframes);
7128 av_freep(&sc->stts_data);
7129 av_freep(&sc->stps_data);
7130 av_freep(&sc->elst_data);
7131 av_freep(&sc->rap_group);
7132 av_freep(&sc->display_matrix);
7133 av_freep(&sc->index_ranges);
7136 for (j = 0; j < sc->stsd_count; j++)
7137 av_free(sc->extradata[j]);
7138 av_freep(&sc->extradata);
7139 av_freep(&sc->extradata_size);
7141 mov_free_encryption_index(&sc->cenc.encryption_index);
7142 av_encryption_info_free(sc->cenc.default_encrypted_sample);
7143 av_aes_ctr_free(sc->cenc.aes_ctr);
7145 av_freep(&sc->stereo3d);
7146 av_freep(&sc->spherical);
7147 av_freep(&sc->mastering);
7148 av_freep(&sc->coll);
7151 if (mov->dv_demux) {
7152 avformat_free_context(mov->dv_fctx);
7153 mov->dv_fctx = NULL;
7156 if (mov->meta_keys) {
7157 for (i = 1; i < mov->meta_keys_count; i++) {
7158 av_freep(&mov->meta_keys[i]);
7160 av_freep(&mov->meta_keys);
7163 av_freep(&mov->trex_data);
7164 av_freep(&mov->bitrates);
7166 for (i = 0; i < mov->frag_index.nb_items; i++) {
7167 MOVFragmentStreamInfo *frag = mov->frag_index.item[i].stream_info;
7168 for (j = 0; j < mov->frag_index.item[i].nb_stream_info; j++) {
7169 mov_free_encryption_index(&frag[j].encryption_index);
7171 av_freep(&mov->frag_index.item[i].stream_info);
7173 av_freep(&mov->frag_index.item);
7175 av_freep(&mov->aes_decrypt);
7176 av_freep(&mov->chapter_tracks);
7181 static int tmcd_is_referenced(AVFormatContext *s, int tmcd_id)
7185 for (i = 0; i < s->nb_streams; i++) {
7186 AVStream *st = s->streams[i];
7187 MOVStreamContext *sc = st->priv_data;
7189 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO &&
7190 sc->timecode_track == tmcd_id)
7196 /* look for a tmcd track not referenced by any video track, and export it globally */
7197 static void export_orphan_timecode(AVFormatContext *s)
7201 for (i = 0; i < s->nb_streams; i++) {
7202 AVStream *st = s->streams[i];
7204 if (st->codecpar->codec_tag == MKTAG('t','m','c','d') &&
7205 !tmcd_is_referenced(s, i + 1)) {
7206 AVDictionaryEntry *tcr = av_dict_get(st->metadata, "timecode", NULL, 0);
7208 av_dict_set(&s->metadata, "timecode", tcr->value, 0);
7215 static int read_tfra(MOVContext *mov, AVIOContext *f)
7217 int version, fieldlength, i, j;
7218 int64_t pos = avio_tell(f);
7219 uint32_t size = avio_rb32(f);
7220 unsigned track_id, item_count;
7222 if (avio_rb32(f) != MKBETAG('t', 'f', 'r', 'a')) {
7225 av_log(mov->fc, AV_LOG_VERBOSE, "found tfra\n");
7227 version = avio_r8(f);
7229 track_id = avio_rb32(f);
7230 fieldlength = avio_rb32(f);
7231 item_count = avio_rb32(f);
7232 for (i = 0; i < item_count; i++) {
7233 int64_t time, offset;
7235 MOVFragmentStreamInfo * frag_stream_info;
7238 return AVERROR_INVALIDDATA;
7242 time = avio_rb64(f);
7243 offset = avio_rb64(f);
7245 time = avio_rb32(f);
7246 offset = avio_rb32(f);
7249 // The first sample of each stream in a fragment is always a random
7250 // access sample. So it's entry in the tfra can be used as the
7251 // initial PTS of the fragment.
7252 index = update_frag_index(mov, offset);
7253 frag_stream_info = get_frag_stream_info(&mov->frag_index, index, track_id);
7254 if (frag_stream_info &&
7255 frag_stream_info->first_tfra_pts == AV_NOPTS_VALUE)
7256 frag_stream_info->first_tfra_pts = time;
7258 for (j = 0; j < ((fieldlength >> 4) & 3) + 1; j++)
7260 for (j = 0; j < ((fieldlength >> 2) & 3) + 1; j++)
7262 for (j = 0; j < ((fieldlength >> 0) & 3) + 1; j++)
7266 avio_seek(f, pos + size, SEEK_SET);
7270 static int mov_read_mfra(MOVContext *c, AVIOContext *f)
7272 int64_t stream_size = avio_size(f);
7273 int64_t original_pos = avio_tell(f);
7277 if ((seek_ret = avio_seek(f, stream_size - 4, SEEK_SET)) < 0) {
7281 mfra_size = avio_rb32(f);
7282 if (mfra_size < 0 || mfra_size > stream_size) {
7283 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (unreasonable size)\n");
7286 if ((seek_ret = avio_seek(f, -mfra_size, SEEK_CUR)) < 0) {
7290 if (avio_rb32(f) != mfra_size) {
7291 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (size mismatch)\n");
7294 if (avio_rb32(f) != MKBETAG('m', 'f', 'r', 'a')) {
7295 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (tag mismatch)\n");
7298 av_log(c->fc, AV_LOG_VERBOSE, "stream has mfra\n");
7300 ret = read_tfra(c, f);
7306 seek_ret = avio_seek(f, original_pos, SEEK_SET);
7308 av_log(c->fc, AV_LOG_ERROR,
7309 "failed to seek back after looking for mfra\n");
7315 static int mov_read_header(AVFormatContext *s)
7317 MOVContext *mov = s->priv_data;
7318 AVIOContext *pb = s->pb;
7320 MOVAtom atom = { AV_RL32("root") };
7323 if (mov->decryption_key_len != 0 && mov->decryption_key_len != AES_CTR_KEY_SIZE) {
7324 av_log(s, AV_LOG_ERROR, "Invalid decryption key len %d expected %d\n",
7325 mov->decryption_key_len, AES_CTR_KEY_SIZE);
7326 return AVERROR(EINVAL);
7330 mov->trak_index = -1;
7331 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
7332 if (pb->seekable & AVIO_SEEKABLE_NORMAL)
7333 atom.size = avio_size(pb);
7335 atom.size = INT64_MAX;
7337 /* check MOV header */
7339 if (mov->moov_retry)
7340 avio_seek(pb, 0, SEEK_SET);
7341 if ((err = mov_read_default(mov, pb, atom)) < 0) {
7342 av_log(s, AV_LOG_ERROR, "error reading header\n");
7346 } while ((pb->seekable & AVIO_SEEKABLE_NORMAL) && !mov->found_moov && !mov->moov_retry++);
7347 if (!mov->found_moov) {
7348 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
7350 return AVERROR_INVALIDDATA;
7352 av_log(mov->fc, AV_LOG_TRACE, "on_parse_exit_offset=%"PRId64"\n", avio_tell(pb));
7354 if (pb->seekable & AVIO_SEEKABLE_NORMAL) {
7355 if (mov->nb_chapter_tracks > 0 && !mov->ignore_chapters)
7356 mov_read_chapters(s);
7357 for (i = 0; i < s->nb_streams; i++)
7358 if (s->streams[i]->codecpar->codec_tag == AV_RL32("tmcd")) {
7359 mov_read_timecode_track(s, s->streams[i]);
7360 } else if (s->streams[i]->codecpar->codec_tag == AV_RL32("rtmd")) {
7361 mov_read_rtmd_track(s, s->streams[i]);
7365 /* copy timecode metadata from tmcd tracks to the related video streams */
7366 for (i = 0; i < s->nb_streams; i++) {
7367 AVStream *st = s->streams[i];
7368 MOVStreamContext *sc = st->priv_data;
7369 if (sc->timecode_track > 0) {
7370 AVDictionaryEntry *tcr;
7371 int tmcd_st_id = -1;
7373 for (j = 0; j < s->nb_streams; j++)
7374 if (s->streams[j]->id == sc->timecode_track)
7377 if (tmcd_st_id < 0 || tmcd_st_id == i)
7379 tcr = av_dict_get(s->streams[tmcd_st_id]->metadata, "timecode", NULL, 0);
7381 av_dict_set(&st->metadata, "timecode", tcr->value, 0);
7384 export_orphan_timecode(s);
7386 for (i = 0; i < s->nb_streams; i++) {
7387 AVStream *st = s->streams[i];
7388 MOVStreamContext *sc = st->priv_data;
7389 fix_timescale(mov, sc);
7390 if(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id == AV_CODEC_ID_AAC) {
7391 st->skip_samples = sc->start_pad;
7393 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && sc->nb_frames_for_fps > 0 && sc->duration_for_fps > 0)
7394 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
7395 sc->time_scale*(int64_t)sc->nb_frames_for_fps, sc->duration_for_fps, INT_MAX);
7396 if (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE) {
7397 if (st->codecpar->width <= 0 || st->codecpar->height <= 0) {
7398 st->codecpar->width = sc->width;
7399 st->codecpar->height = sc->height;
7401 if (st->codecpar->codec_id == AV_CODEC_ID_DVD_SUBTITLE) {
7402 if ((err = mov_rewrite_dvd_sub_extradata(st)) < 0)
7406 if (mov->handbrake_version &&
7407 mov->handbrake_version <= 1000000*0 + 1000*10 + 2 && // 0.10.2
7408 st->codecpar->codec_id == AV_CODEC_ID_MP3
7410 av_log(s, AV_LOG_VERBOSE, "Forcing full parsing for mp3 stream\n");
7411 st->need_parsing = AVSTREAM_PARSE_FULL;
7415 if (mov->trex_data) {
7416 for (i = 0; i < s->nb_streams; i++) {
7417 AVStream *st = s->streams[i];
7418 MOVStreamContext *sc = st->priv_data;
7419 if (st->duration > 0) {
7420 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
7421 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
7422 sc->data_size, sc->time_scale);
7424 return AVERROR_INVALIDDATA;
7426 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale / st->duration;
7431 if (mov->use_mfra_for > 0) {
7432 for (i = 0; i < s->nb_streams; i++) {
7433 AVStream *st = s->streams[i];
7434 MOVStreamContext *sc = st->priv_data;
7435 if (sc->duration_for_fps > 0) {
7436 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
7437 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
7438 sc->data_size, sc->time_scale);
7440 return AVERROR_INVALIDDATA;
7442 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale /
7443 sc->duration_for_fps;
7448 for (i = 0; i < mov->bitrates_count && i < s->nb_streams; i++) {
7449 if (mov->bitrates[i]) {
7450 s->streams[i]->codecpar->bit_rate = mov->bitrates[i];
7454 ff_rfps_calculate(s);
7456 for (i = 0; i < s->nb_streams; i++) {
7457 AVStream *st = s->streams[i];
7458 MOVStreamContext *sc = st->priv_data;
7460 switch (st->codecpar->codec_type) {
7461 case AVMEDIA_TYPE_AUDIO:
7462 err = ff_replaygain_export(st, s->metadata);
7468 case AVMEDIA_TYPE_VIDEO:
7469 if (sc->display_matrix) {
7470 err = av_stream_add_side_data(st, AV_PKT_DATA_DISPLAYMATRIX, (uint8_t*)sc->display_matrix,
7471 sizeof(int32_t) * 9);
7475 sc->display_matrix = NULL;
7478 err = av_stream_add_side_data(st, AV_PKT_DATA_STEREO3D,
7479 (uint8_t *)sc->stereo3d,
7480 sizeof(*sc->stereo3d));
7484 sc->stereo3d = NULL;
7486 if (sc->spherical) {
7487 err = av_stream_add_side_data(st, AV_PKT_DATA_SPHERICAL,
7488 (uint8_t *)sc->spherical,
7489 sc->spherical_size);
7493 sc->spherical = NULL;
7495 if (sc->mastering) {
7496 err = av_stream_add_side_data(st, AV_PKT_DATA_MASTERING_DISPLAY_METADATA,
7497 (uint8_t *)sc->mastering,
7498 sizeof(*sc->mastering));
7502 sc->mastering = NULL;
7505 err = av_stream_add_side_data(st, AV_PKT_DATA_CONTENT_LIGHT_LEVEL,
7506 (uint8_t *)sc->coll,
7516 ff_configure_buffers_for_index(s, AV_TIME_BASE);
7518 for (i = 0; i < mov->frag_index.nb_items; i++)
7519 if (mov->frag_index.item[i].moof_offset <= mov->fragment.moof_offset)
7520 mov->frag_index.item[i].headers_read = 1;
7525 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
7527 AVIndexEntry *sample = NULL;
7528 int64_t best_dts = INT64_MAX;
7530 for (i = 0; i < s->nb_streams; i++) {
7531 AVStream *avst = s->streams[i];
7532 MOVStreamContext *msc = avst->priv_data;
7533 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
7534 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
7535 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
7536 av_log(s, AV_LOG_TRACE, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
7537 if (!sample || (!(s->pb->seekable & AVIO_SEEKABLE_NORMAL) && current_sample->pos < sample->pos) ||
7538 ((s->pb->seekable & AVIO_SEEKABLE_NORMAL) &&
7539 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
7540 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
7541 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
7542 sample = current_sample;
7551 static int should_retry(AVIOContext *pb, int error_code) {
7552 if (error_code == AVERROR_EOF || avio_feof(pb))
7558 static int mov_switch_root(AVFormatContext *s, int64_t target, int index)
7561 MOVContext *mov = s->priv_data;
7563 if (index >= 0 && index < mov->frag_index.nb_items)
7564 target = mov->frag_index.item[index].moof_offset;
7565 if (avio_seek(s->pb, target, SEEK_SET) != target) {
7566 av_log(mov->fc, AV_LOG_ERROR, "root atom offset 0x%"PRIx64": partial file\n", target);
7567 return AVERROR_INVALIDDATA;
7570 mov->next_root_atom = 0;
7571 if (index < 0 || index >= mov->frag_index.nb_items)
7572 index = search_frag_moof_offset(&mov->frag_index, target);
7573 if (index < mov->frag_index.nb_items) {
7574 if (index + 1 < mov->frag_index.nb_items)
7575 mov->next_root_atom = mov->frag_index.item[index + 1].moof_offset;
7576 if (mov->frag_index.item[index].headers_read)
7578 mov->frag_index.item[index].headers_read = 1;
7581 mov->found_mdat = 0;
7583 ret = mov_read_default(mov, s->pb, (MOVAtom){ AV_RL32("root"), INT64_MAX });
7586 if (avio_feof(s->pb))
7588 av_log(s, AV_LOG_TRACE, "read fragments, offset 0x%"PRIx64"\n", avio_tell(s->pb));
7593 static int mov_change_extradata(MOVStreamContext *sc, AVPacket *pkt)
7595 uint8_t *side, *extradata;
7598 /* Save the current index. */
7599 sc->last_stsd_index = sc->stsc_data[sc->stsc_index].id - 1;
7601 /* Notify the decoder that extradata changed. */
7602 extradata_size = sc->extradata_size[sc->last_stsd_index];
7603 extradata = sc->extradata[sc->last_stsd_index];
7604 if (extradata_size > 0 && extradata) {
7605 side = av_packet_new_side_data(pkt,
7606 AV_PKT_DATA_NEW_EXTRADATA,
7609 return AVERROR(ENOMEM);
7610 memcpy(side, extradata, extradata_size);
7616 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
7618 MOVContext *mov = s->priv_data;
7619 MOVStreamContext *sc;
7620 AVIndexEntry *sample;
7621 AVStream *st = NULL;
7622 int64_t current_index;
7626 sample = mov_find_next_sample(s, &st);
7627 if (!sample || (mov->next_root_atom && sample->pos > mov->next_root_atom)) {
7628 if (!mov->next_root_atom)
7630 if ((ret = mov_switch_root(s, mov->next_root_atom, -1)) < 0)
7635 /* must be done just before reading, to avoid infinite loop on sample */
7636 current_index = sc->current_index;
7637 mov_current_sample_inc(sc);
7639 if (mov->next_root_atom) {
7640 sample->pos = FFMIN(sample->pos, mov->next_root_atom);
7641 sample->size = FFMIN(sample->size, (mov->next_root_atom - sample->pos));
7644 if (st->discard != AVDISCARD_ALL) {
7645 int64_t ret64 = avio_seek(sc->pb, sample->pos, SEEK_SET);
7646 if (ret64 != sample->pos) {
7647 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
7648 sc->ffindex, sample->pos);
7649 if (should_retry(sc->pb, ret64)) {
7650 mov_current_sample_dec(sc);
7652 return AVERROR_INVALIDDATA;
7655 if( st->discard == AVDISCARD_NONKEY && 0==(sample->flags & AVINDEX_KEYFRAME) ) {
7656 av_log(mov->fc, AV_LOG_DEBUG, "Nonkey frame from stream %d discarded due to AVDISCARD_NONKEY\n", sc->ffindex);
7660 ret = av_get_packet(sc->pb, pkt, sample->size);
7662 if (should_retry(sc->pb, ret)) {
7663 mov_current_sample_dec(sc);
7667 if (sc->has_palette) {
7670 pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, AVPALETTE_SIZE);
7672 av_log(mov->fc, AV_LOG_ERROR, "Cannot append palette to packet\n");
7674 memcpy(pal, sc->palette, AVPALETTE_SIZE);
7675 sc->has_palette = 0;
7678 #if CONFIG_DV_DEMUXER
7679 if (mov->dv_demux && sc->dv_audio_container) {
7680 avpriv_dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size, pkt->pos);
7681 av_freep(&pkt->data);
7683 ret = avpriv_dv_get_packet(mov->dv_demux, pkt);
7688 if (st->codecpar->codec_id == AV_CODEC_ID_MP3 && !st->need_parsing && pkt->size > 4) {
7689 if (ff_mpa_check_header(AV_RB32(pkt->data)) < 0)
7690 st->need_parsing = AVSTREAM_PARSE_FULL;
7694 pkt->stream_index = sc->ffindex;
7695 pkt->dts = sample->timestamp;
7696 if (sample->flags & AVINDEX_DISCARD_FRAME) {
7697 pkt->flags |= AV_PKT_FLAG_DISCARD;
7699 if (sc->ctts_data && sc->ctts_index < sc->ctts_count) {
7700 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
7701 /* update ctts context */
7703 if (sc->ctts_index < sc->ctts_count &&
7704 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
7706 sc->ctts_sample = 0;
7709 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
7710 st->index_entries[sc->current_sample].timestamp : st->duration;
7712 if (next_dts >= pkt->dts)
7713 pkt->duration = next_dts - pkt->dts;
7714 pkt->pts = pkt->dts;
7716 if (st->discard == AVDISCARD_ALL)
7718 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? AV_PKT_FLAG_KEY : 0;
7719 pkt->pos = sample->pos;
7721 /* Multiple stsd handling. */
7722 if (sc->stsc_data) {
7723 /* Keep track of the stsc index for the given sample, then check
7724 * if the stsd index is different from the last used one. */
7726 if (mov_stsc_index_valid(sc->stsc_index, sc->stsc_count) &&
7727 mov_get_stsc_samples(sc, sc->stsc_index) == sc->stsc_sample) {
7729 sc->stsc_sample = 0;
7730 /* Do not check indexes after a switch. */
7731 } else if (sc->stsc_data[sc->stsc_index].id > 0 &&
7732 sc->stsc_data[sc->stsc_index].id - 1 < sc->stsd_count &&
7733 sc->stsc_data[sc->stsc_index].id - 1 != sc->last_stsd_index) {
7734 ret = mov_change_extradata(sc, pkt);
7741 aax_filter(pkt->data, pkt->size, mov);
7743 ret = cenc_filter(mov, sc, pkt, current_index);
7750 static int mov_seek_fragment(AVFormatContext *s, AVStream *st, int64_t timestamp)
7752 MOVContext *mov = s->priv_data;
7755 if (!mov->frag_index.complete)
7758 index = search_frag_timestamp(&mov->frag_index, st, timestamp);
7761 if (!mov->frag_index.item[index].headers_read)
7762 return mov_switch_root(s, -1, index);
7763 if (index + 1 < mov->frag_index.nb_items)
7764 mov->next_root_atom = mov->frag_index.item[index + 1].moof_offset;
7769 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
7771 MOVStreamContext *sc = st->priv_data;
7772 int sample, time_sample, ret;
7775 // Here we consider timestamp to be PTS, hence try to offset it so that we
7776 // can search over the DTS timeline.
7777 timestamp -= (sc->min_corrected_pts + sc->dts_shift);
7779 ret = mov_seek_fragment(s, st, timestamp);
7783 sample = av_index_search_timestamp(st, timestamp, flags);
7784 av_log(s, AV_LOG_TRACE, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
7785 if (sample < 0 && st->nb_index_entries && timestamp < st->index_entries[0].timestamp)
7787 if (sample < 0) /* not sure what to do */
7788 return AVERROR_INVALIDDATA;
7789 mov_current_sample_set(sc, sample);
7790 av_log(s, AV_LOG_TRACE, "stream %d, found sample %d\n", st->index, sc->current_sample);
7791 /* adjust ctts index */
7792 if (sc->ctts_data) {
7794 for (i = 0; i < sc->ctts_count; i++) {
7795 int next = time_sample + sc->ctts_data[i].count;
7796 if (next > sc->current_sample) {
7798 sc->ctts_sample = sc->current_sample - time_sample;
7805 /* adjust stsd index */
7807 for (i = 0; i < sc->stsc_count; i++) {
7808 int64_t next = time_sample + mov_get_stsc_samples(sc, i);
7809 if (next > sc->current_sample) {
7811 sc->stsc_sample = sc->current_sample - time_sample;
7814 av_assert0(next == (int)next);
7821 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
7823 MOVContext *mc = s->priv_data;
7828 if (stream_index >= s->nb_streams)
7829 return AVERROR_INVALIDDATA;
7831 st = s->streams[stream_index];
7832 sample = mov_seek_stream(s, st, sample_time, flags);
7836 if (mc->seek_individually) {
7837 /* adjust seek timestamp to found sample timestamp */
7838 int64_t seek_timestamp = st->index_entries[sample].timestamp;
7840 for (i = 0; i < s->nb_streams; i++) {
7842 MOVStreamContext *sc = s->streams[i]->priv_data;
7844 st->skip_samples = (sample_time <= 0) ? sc->start_pad : 0;
7846 if (stream_index == i)
7849 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
7850 mov_seek_stream(s, st, timestamp, flags);
7853 for (i = 0; i < s->nb_streams; i++) {
7854 MOVStreamContext *sc;
7857 mov_current_sample_set(sc, 0);
7860 MOVStreamContext *sc;
7861 AVIndexEntry *entry = mov_find_next_sample(s, &st);
7863 return AVERROR_INVALIDDATA;
7865 if (sc->ffindex == stream_index && sc->current_sample == sample)
7867 mov_current_sample_inc(sc);
7873 #define OFFSET(x) offsetof(MOVContext, x)
7874 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
7875 static const AVOption mov_options[] = {
7876 {"use_absolute_path",
7877 "allow using absolute path when opening alias, this is a possible security issue",
7878 OFFSET(use_absolute_path), AV_OPT_TYPE_BOOL, {.i64 = 0},
7880 {"seek_streams_individually",
7881 "Seek each stream individually to the to the closest point",
7882 OFFSET(seek_individually), AV_OPT_TYPE_BOOL, { .i64 = 1 },
7884 {"ignore_editlist", "Ignore the edit list atom.", OFFSET(ignore_editlist), AV_OPT_TYPE_BOOL, {.i64 = 0},
7886 {"advanced_editlist",
7887 "Modify the AVIndex according to the editlists. Use this option to decode in the order specified by the edits.",
7888 OFFSET(advanced_editlist), AV_OPT_TYPE_BOOL, {.i64 = 1},
7890 {"ignore_chapters", "", OFFSET(ignore_chapters), AV_OPT_TYPE_BOOL, {.i64 = 0},
7893 "use mfra for fragment timestamps",
7894 OFFSET(use_mfra_for), AV_OPT_TYPE_INT, {.i64 = FF_MOV_FLAG_MFRA_AUTO},
7895 -1, FF_MOV_FLAG_MFRA_PTS, FLAGS,
7897 {"auto", "auto", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_AUTO}, 0, 0,
7898 FLAGS, "use_mfra_for" },
7899 {"dts", "dts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_DTS}, 0, 0,
7900 FLAGS, "use_mfra_for" },
7901 {"pts", "pts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_PTS}, 0, 0,
7902 FLAGS, "use_mfra_for" },
7903 { "export_all", "Export unrecognized metadata entries", OFFSET(export_all),
7904 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
7905 { "export_xmp", "Export full XMP metadata", OFFSET(export_xmp),
7906 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
7907 { "activation_bytes", "Secret bytes for Audible AAX files", OFFSET(activation_bytes),
7908 AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
7909 { "audible_fixed_key", // extracted from libAAX_SDK.so and AAXSDKWin.dll files!
7910 "Fixed key used for handling Audible AAX files", OFFSET(audible_fixed_key),
7911 AV_OPT_TYPE_BINARY, {.str="77214d4b196a87cd520045fd20a51d67"},
7912 .flags = AV_OPT_FLAG_DECODING_PARAM },
7913 { "decryption_key", "The media decryption key (hex)", OFFSET(decryption_key), AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
7914 { "enable_drefs", "Enable external track support.", OFFSET(enable_drefs), AV_OPT_TYPE_BOOL,
7915 {.i64 = 0}, 0, 1, FLAGS },
7920 static const AVClass mov_class = {
7921 .class_name = "mov,mp4,m4a,3gp,3g2,mj2",
7922 .item_name = av_default_item_name,
7923 .option = mov_options,
7924 .version = LIBAVUTIL_VERSION_INT,
7927 AVInputFormat ff_mov_demuxer = {
7928 .name = "mov,mp4,m4a,3gp,3g2,mj2",
7929 .long_name = NULL_IF_CONFIG_SMALL("QuickTime / MOV"),
7930 .priv_class = &mov_class,
7931 .priv_data_size = sizeof(MOVContext),
7932 .extensions = "mov,mp4,m4a,3gp,3g2,mj2",
7933 .read_probe = mov_probe,
7934 .read_header = mov_read_header,
7935 .read_packet = mov_read_packet,
7936 .read_close = mov_read_close,
7937 .read_seek = mov_read_seek,
7938 .flags = AVFMT_NO_BYTE_SEEK,