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);
1905 if (atom.type == MKTAG('h','v','c','C') && st->codecpar->codec_tag == MKTAG('d','v','h','1'))
1906 st->codecpar->codec_id = AV_CODEC_ID_HEVC;
1911 static int mov_read_dvc1(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1914 uint8_t profile_level;
1917 if (c->fc->nb_streams < 1)
1919 st = c->fc->streams[c->fc->nb_streams-1];
1921 if (atom.size >= (1<<28) || atom.size < 7)
1922 return AVERROR_INVALIDDATA;
1924 profile_level = avio_r8(pb);
1925 if ((profile_level & 0xf0) != 0xc0)
1928 avio_seek(pb, 6, SEEK_CUR);
1929 av_freep(&st->codecpar->extradata);
1930 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 7);
1938 * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
1939 * but can have extradata appended at the end after the 40 bytes belonging
1942 static int mov_read_strf(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1947 if (c->fc->nb_streams < 1)
1949 if (atom.size <= 40)
1951 st = c->fc->streams[c->fc->nb_streams-1];
1953 if ((uint64_t)atom.size > (1<<30))
1954 return AVERROR_INVALIDDATA;
1957 av_freep(&st->codecpar->extradata);
1958 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 40);
1965 static int mov_read_stco(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1968 MOVStreamContext *sc;
1969 unsigned int i, entries;
1971 if (c->fc->nb_streams < 1)
1973 st = c->fc->streams[c->fc->nb_streams-1];
1976 avio_r8(pb); /* version */
1977 avio_rb24(pb); /* flags */
1979 entries = avio_rb32(pb);
1984 if (sc->chunk_offsets)
1985 av_log(c->fc, AV_LOG_WARNING, "Duplicated STCO atom\n");
1986 av_free(sc->chunk_offsets);
1987 sc->chunk_count = 0;
1988 sc->chunk_offsets = av_malloc_array(entries, sizeof(*sc->chunk_offsets));
1989 if (!sc->chunk_offsets)
1990 return AVERROR(ENOMEM);
1991 sc->chunk_count = entries;
1993 if (atom.type == MKTAG('s','t','c','o'))
1994 for (i = 0; i < entries && !pb->eof_reached; i++)
1995 sc->chunk_offsets[i] = avio_rb32(pb);
1996 else if (atom.type == MKTAG('c','o','6','4'))
1997 for (i = 0; i < entries && !pb->eof_reached; i++)
1998 sc->chunk_offsets[i] = avio_rb64(pb);
2000 return AVERROR_INVALIDDATA;
2002 sc->chunk_count = i;
2004 if (pb->eof_reached) {
2005 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STCO atom\n");
2012 static int mov_codec_id(AVStream *st, uint32_t format)
2014 int id = ff_codec_get_id(ff_codec_movaudio_tags, format);
2017 ((format & 0xFFFF) == 'm' + ('s' << 8) ||
2018 (format & 0xFFFF) == 'T' + ('S' << 8)))
2019 id = ff_codec_get_id(ff_codec_wav_tags, av_bswap32(format) & 0xFFFF);
2021 if (st->codecpar->codec_type != AVMEDIA_TYPE_VIDEO && id > 0) {
2022 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2023 } else if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO &&
2024 /* skip old ASF MPEG-4 tag */
2025 format && format != MKTAG('m','p','4','s')) {
2026 id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2028 id = ff_codec_get_id(ff_codec_bmp_tags, format);
2030 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
2031 else if (st->codecpar->codec_type == AVMEDIA_TYPE_DATA ||
2032 (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE &&
2033 st->codecpar->codec_id == AV_CODEC_ID_NONE)) {
2034 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
2036 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
2038 id = ff_codec_get_id(ff_codec_movdata_tags, format);
2042 st->codecpar->codec_tag = format;
2047 static void mov_parse_stsd_video(MOVContext *c, AVIOContext *pb,
2048 AVStream *st, MOVStreamContext *sc)
2050 uint8_t codec_name[32] = { 0 };
2054 /* The first 16 bytes of the video sample description are already
2055 * read in ff_mov_read_stsd_entries() */
2056 stsd_start = avio_tell(pb) - 16;
2058 avio_rb16(pb); /* version */
2059 avio_rb16(pb); /* revision level */
2060 avio_rb32(pb); /* vendor */
2061 avio_rb32(pb); /* temporal quality */
2062 avio_rb32(pb); /* spatial quality */
2064 st->codecpar->width = avio_rb16(pb); /* width */
2065 st->codecpar->height = avio_rb16(pb); /* height */
2067 avio_rb32(pb); /* horiz resolution */
2068 avio_rb32(pb); /* vert resolution */
2069 avio_rb32(pb); /* data size, always 0 */
2070 avio_rb16(pb); /* frames per samples */
2072 len = avio_r8(pb); /* codec name, pascal string */
2075 mov_read_mac_string(c, pb, len, codec_name, sizeof(codec_name));
2077 avio_skip(pb, 31 - len);
2080 av_dict_set(&st->metadata, "encoder", codec_name, 0);
2082 /* codec_tag YV12 triggers an UV swap in rawdec.c */
2083 if (!strncmp(codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25)) {
2084 st->codecpar->codec_tag = MKTAG('I', '4', '2', '0');
2085 st->codecpar->width &= ~1;
2086 st->codecpar->height &= ~1;
2088 /* Flash Media Server uses tag H.263 with Sorenson Spark */
2089 if (st->codecpar->codec_tag == MKTAG('H','2','6','3') &&
2090 !strncmp(codec_name, "Sorenson H263", 13))
2091 st->codecpar->codec_id = AV_CODEC_ID_FLV1;
2093 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* depth */
2095 avio_seek(pb, stsd_start, SEEK_SET);
2097 if (ff_get_qtpalette(st->codecpar->codec_id, pb, sc->palette)) {
2098 st->codecpar->bits_per_coded_sample &= 0x1F;
2099 sc->has_palette = 1;
2103 static void mov_parse_stsd_audio(MOVContext *c, AVIOContext *pb,
2104 AVStream *st, MOVStreamContext *sc)
2106 int bits_per_sample, flags;
2107 uint16_t version = avio_rb16(pb);
2108 AVDictionaryEntry *compatible_brands = av_dict_get(c->fc->metadata, "compatible_brands", NULL, AV_DICT_MATCH_CASE);
2110 avio_rb16(pb); /* revision level */
2111 avio_rb32(pb); /* vendor */
2113 st->codecpar->channels = avio_rb16(pb); /* channel count */
2114 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* sample size */
2115 av_log(c->fc, AV_LOG_TRACE, "audio channels %d\n", st->codecpar->channels);
2117 sc->audio_cid = avio_rb16(pb);
2118 avio_rb16(pb); /* packet size = 0 */
2120 st->codecpar->sample_rate = ((avio_rb32(pb) >> 16));
2122 // Read QT version 1 fields. In version 0 these do not exist.
2123 av_log(c->fc, AV_LOG_TRACE, "version =%d, isom =%d\n", version, c->isom);
2125 (compatible_brands && strstr(compatible_brands->value, "qt ")) ||
2126 (sc->stsd_version == 0 && version > 0)) {
2128 sc->samples_per_frame = avio_rb32(pb);
2129 avio_rb32(pb); /* bytes per packet */
2130 sc->bytes_per_frame = avio_rb32(pb);
2131 avio_rb32(pb); /* bytes per sample */
2132 } else if (version == 2) {
2133 avio_rb32(pb); /* sizeof struct only */
2134 st->codecpar->sample_rate = av_int2double(avio_rb64(pb));
2135 st->codecpar->channels = avio_rb32(pb);
2136 avio_rb32(pb); /* always 0x7F000000 */
2137 st->codecpar->bits_per_coded_sample = avio_rb32(pb);
2139 flags = avio_rb32(pb); /* lpcm format specific flag */
2140 sc->bytes_per_frame = avio_rb32(pb);
2141 sc->samples_per_frame = avio_rb32(pb);
2142 if (st->codecpar->codec_tag == MKTAG('l','p','c','m'))
2143 st->codecpar->codec_id =
2144 ff_mov_get_lpcm_codec_id(st->codecpar->bits_per_coded_sample,
2147 if (version == 0 || (version == 1 && sc->audio_cid != -2)) {
2148 /* can't correctly handle variable sized packet as audio unit */
2149 switch (st->codecpar->codec_id) {
2150 case AV_CODEC_ID_MP2:
2151 case AV_CODEC_ID_MP3:
2152 st->need_parsing = AVSTREAM_PARSE_FULL;
2158 if (sc->format == 0) {
2159 if (st->codecpar->bits_per_coded_sample == 8)
2160 st->codecpar->codec_id = mov_codec_id(st, MKTAG('r','a','w',' '));
2161 else if (st->codecpar->bits_per_coded_sample == 16)
2162 st->codecpar->codec_id = mov_codec_id(st, MKTAG('t','w','o','s'));
2165 switch (st->codecpar->codec_id) {
2166 case AV_CODEC_ID_PCM_S8:
2167 case AV_CODEC_ID_PCM_U8:
2168 if (st->codecpar->bits_per_coded_sample == 16)
2169 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16BE;
2171 case AV_CODEC_ID_PCM_S16LE:
2172 case AV_CODEC_ID_PCM_S16BE:
2173 if (st->codecpar->bits_per_coded_sample == 8)
2174 st->codecpar->codec_id = AV_CODEC_ID_PCM_S8;
2175 else if (st->codecpar->bits_per_coded_sample == 24)
2176 st->codecpar->codec_id =
2177 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
2178 AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
2179 else if (st->codecpar->bits_per_coded_sample == 32)
2180 st->codecpar->codec_id =
2181 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
2182 AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
2184 /* set values for old format before stsd version 1 appeared */
2185 case AV_CODEC_ID_MACE3:
2186 sc->samples_per_frame = 6;
2187 sc->bytes_per_frame = 2 * st->codecpar->channels;
2189 case AV_CODEC_ID_MACE6:
2190 sc->samples_per_frame = 6;
2191 sc->bytes_per_frame = 1 * st->codecpar->channels;
2193 case AV_CODEC_ID_ADPCM_IMA_QT:
2194 sc->samples_per_frame = 64;
2195 sc->bytes_per_frame = 34 * st->codecpar->channels;
2197 case AV_CODEC_ID_GSM:
2198 sc->samples_per_frame = 160;
2199 sc->bytes_per_frame = 33;
2205 bits_per_sample = av_get_bits_per_sample(st->codecpar->codec_id);
2206 if (bits_per_sample) {
2207 st->codecpar->bits_per_coded_sample = bits_per_sample;
2208 sc->sample_size = (bits_per_sample >> 3) * st->codecpar->channels;
2212 static void mov_parse_stsd_subtitle(MOVContext *c, AVIOContext *pb,
2213 AVStream *st, MOVStreamContext *sc,
2216 // ttxt stsd contains display flags, justification, background
2217 // color, fonts, and default styles, so fake an atom to read it
2218 MOVAtom fake_atom = { .size = size };
2219 // mp4s contains a regular esds atom
2220 if (st->codecpar->codec_tag != AV_RL32("mp4s"))
2221 mov_read_glbl(c, pb, fake_atom);
2222 st->codecpar->width = sc->width;
2223 st->codecpar->height = sc->height;
2226 static uint32_t yuv_to_rgba(uint32_t ycbcr)
2231 y = (ycbcr >> 16) & 0xFF;
2232 cr = (ycbcr >> 8) & 0xFF;
2235 b = av_clip_uint8((1164 * (y - 16) + 2018 * (cb - 128)) / 1000);
2236 g = av_clip_uint8((1164 * (y - 16) - 813 * (cr - 128) - 391 * (cb - 128)) / 1000);
2237 r = av_clip_uint8((1164 * (y - 16) + 1596 * (cr - 128) ) / 1000);
2239 return (r << 16) | (g << 8) | b;
2242 static int mov_rewrite_dvd_sub_extradata(AVStream *st)
2244 char buf[256] = {0};
2245 uint8_t *src = st->codecpar->extradata;
2248 if (st->codecpar->extradata_size != 64)
2251 if (st->codecpar->width > 0 && st->codecpar->height > 0)
2252 snprintf(buf, sizeof(buf), "size: %dx%d\n",
2253 st->codecpar->width, st->codecpar->height);
2254 av_strlcat(buf, "palette: ", sizeof(buf));
2256 for (i = 0; i < 16; i++) {
2257 uint32_t yuv = AV_RB32(src + i * 4);
2258 uint32_t rgba = yuv_to_rgba(yuv);
2260 av_strlcatf(buf, sizeof(buf), "%06"PRIx32"%s", rgba, i != 15 ? ", " : "");
2263 if (av_strlcat(buf, "\n", sizeof(buf)) >= sizeof(buf))
2266 av_freep(&st->codecpar->extradata);
2267 st->codecpar->extradata_size = 0;
2268 st->codecpar->extradata = av_mallocz(strlen(buf) + AV_INPUT_BUFFER_PADDING_SIZE);
2269 if (!st->codecpar->extradata)
2270 return AVERROR(ENOMEM);
2271 st->codecpar->extradata_size = strlen(buf);
2272 memcpy(st->codecpar->extradata, buf, st->codecpar->extradata_size);
2277 static int mov_parse_stsd_data(MOVContext *c, AVIOContext *pb,
2278 AVStream *st, MOVStreamContext *sc,
2283 if (st->codecpar->codec_tag == MKTAG('t','m','c','d')) {
2284 if ((int)size != size)
2285 return AVERROR(ENOMEM);
2287 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
2291 MOVStreamContext *tmcd_ctx = st->priv_data;
2293 val = AV_RB32(st->codecpar->extradata + 4);
2294 tmcd_ctx->tmcd_flags = val;
2295 st->avg_frame_rate.num = st->codecpar->extradata[16]; /* number of frame */
2296 st->avg_frame_rate.den = 1;
2297 #if FF_API_LAVF_AVCTX
2298 FF_DISABLE_DEPRECATION_WARNINGS
2299 st->codec->time_base = av_inv_q(st->avg_frame_rate);
2300 FF_ENABLE_DEPRECATION_WARNINGS
2302 /* adjust for per frame dur in counter mode */
2303 if (tmcd_ctx->tmcd_flags & 0x0008) {
2304 int timescale = AV_RB32(st->codecpar->extradata + 8);
2305 int framedur = AV_RB32(st->codecpar->extradata + 12);
2306 st->avg_frame_rate.num *= timescale;
2307 st->avg_frame_rate.den *= framedur;
2308 #if FF_API_LAVF_AVCTX
2309 FF_DISABLE_DEPRECATION_WARNINGS
2310 st->codec->time_base.den *= timescale;
2311 st->codec->time_base.num *= framedur;
2312 FF_ENABLE_DEPRECATION_WARNINGS
2316 uint32_t len = AV_RB32(st->codecpar->extradata + 18); /* name atom length */
2317 uint32_t format = AV_RB32(st->codecpar->extradata + 22);
2318 if (format == AV_RB32("name") && (int64_t)size >= (int64_t)len + 18) {
2319 uint16_t str_size = AV_RB16(st->codecpar->extradata + 26); /* string length */
2320 if (str_size > 0 && size >= (int)str_size + 26) {
2321 char *reel_name = av_malloc(str_size + 1);
2323 return AVERROR(ENOMEM);
2324 memcpy(reel_name, st->codecpar->extradata + 30, str_size);
2325 reel_name[str_size] = 0; /* Add null terminator */
2326 /* don't add reel_name if emtpy string */
2327 if (*reel_name == 0) {
2330 av_dict_set(&st->metadata, "reel_name", reel_name, AV_DICT_DONT_STRDUP_VAL);
2337 /* other codec type, just skip (rtp, mp4s ...) */
2338 avio_skip(pb, size);
2343 static int mov_finalize_stsd_codec(MOVContext *c, AVIOContext *pb,
2344 AVStream *st, MOVStreamContext *sc)
2346 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
2347 !st->codecpar->sample_rate && sc->time_scale > 1)
2348 st->codecpar->sample_rate = sc->time_scale;
2350 /* special codec parameters handling */
2351 switch (st->codecpar->codec_id) {
2352 #if CONFIG_DV_DEMUXER
2353 case AV_CODEC_ID_DVAUDIO:
2354 c->dv_fctx = avformat_alloc_context();
2356 av_log(c->fc, AV_LOG_ERROR, "dv demux context alloc error\n");
2357 return AVERROR(ENOMEM);
2359 c->dv_demux = avpriv_dv_init_demux(c->dv_fctx);
2361 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
2362 return AVERROR(ENOMEM);
2364 sc->dv_audio_container = 1;
2365 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16LE;
2368 /* no ifdef since parameters are always those */
2369 case AV_CODEC_ID_QCELP:
2370 st->codecpar->channels = 1;
2371 // force sample rate for qcelp when not stored in mov
2372 if (st->codecpar->codec_tag != MKTAG('Q','c','l','p'))
2373 st->codecpar->sample_rate = 8000;
2374 // FIXME: Why is the following needed for some files?
2375 sc->samples_per_frame = 160;
2376 if (!sc->bytes_per_frame)
2377 sc->bytes_per_frame = 35;
2379 case AV_CODEC_ID_AMR_NB:
2380 st->codecpar->channels = 1;
2381 /* force sample rate for amr, stsd in 3gp does not store sample rate */
2382 st->codecpar->sample_rate = 8000;
2384 case AV_CODEC_ID_AMR_WB:
2385 st->codecpar->channels = 1;
2386 st->codecpar->sample_rate = 16000;
2388 case AV_CODEC_ID_MP2:
2389 case AV_CODEC_ID_MP3:
2390 /* force type after stsd for m1a hdlr */
2391 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2393 case AV_CODEC_ID_GSM:
2394 case AV_CODEC_ID_ADPCM_MS:
2395 case AV_CODEC_ID_ADPCM_IMA_WAV:
2396 case AV_CODEC_ID_ILBC:
2397 case AV_CODEC_ID_MACE3:
2398 case AV_CODEC_ID_MACE6:
2399 case AV_CODEC_ID_QDM2:
2400 st->codecpar->block_align = sc->bytes_per_frame;
2402 case AV_CODEC_ID_ALAC:
2403 if (st->codecpar->extradata_size == 36) {
2404 st->codecpar->channels = AV_RB8 (st->codecpar->extradata + 21);
2405 st->codecpar->sample_rate = AV_RB32(st->codecpar->extradata + 32);
2408 case AV_CODEC_ID_AC3:
2409 case AV_CODEC_ID_EAC3:
2410 case AV_CODEC_ID_MPEG1VIDEO:
2411 case AV_CODEC_ID_VC1:
2412 case AV_CODEC_ID_VP8:
2413 case AV_CODEC_ID_VP9:
2414 st->need_parsing = AVSTREAM_PARSE_FULL;
2422 static int mov_skip_multiple_stsd(MOVContext *c, AVIOContext *pb,
2423 int codec_tag, int format,
2426 int video_codec_id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2429 (codec_tag != format &&
2430 // AVID 1:1 samples with differing data format and codec tag exist
2431 (codec_tag != AV_RL32("AV1x") || format != AV_RL32("AVup")) &&
2432 // prores is allowed to have differing data format and codec tag
2433 codec_tag != AV_RL32("apcn") && codec_tag != AV_RL32("apch") &&
2435 codec_tag != AV_RL32("dvpp") && codec_tag != AV_RL32("dvcp") &&
2436 (c->fc->video_codec_id ? video_codec_id != c->fc->video_codec_id
2437 : codec_tag != MKTAG('j','p','e','g')))) {
2438 /* Multiple fourcc, we skip JPEG. This is not correct, we should
2439 * export it as a separate AVStream but this needs a few changes
2440 * in the MOV demuxer, patch welcome. */
2442 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
2443 avio_skip(pb, size);
2450 int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries)
2453 MOVStreamContext *sc;
2454 int pseudo_stream_id;
2456 av_assert0 (c->fc->nb_streams >= 1);
2457 st = c->fc->streams[c->fc->nb_streams-1];
2460 for (pseudo_stream_id = 0;
2461 pseudo_stream_id < entries && !pb->eof_reached;
2462 pseudo_stream_id++) {
2463 //Parsing Sample description table
2465 int ret, dref_id = 1;
2466 MOVAtom a = { AV_RL32("stsd") };
2467 int64_t start_pos = avio_tell(pb);
2468 int64_t size = avio_rb32(pb); /* size */
2469 uint32_t format = avio_rl32(pb); /* data format */
2472 avio_rb32(pb); /* reserved */
2473 avio_rb16(pb); /* reserved */
2474 dref_id = avio_rb16(pb);
2475 } else if (size <= 7) {
2476 av_log(c->fc, AV_LOG_ERROR,
2477 "invalid size %"PRId64" in stsd\n", size);
2478 return AVERROR_INVALIDDATA;
2481 if (mov_skip_multiple_stsd(c, pb, st->codecpar->codec_tag, format,
2482 size - (avio_tell(pb) - start_pos))) {
2487 sc->pseudo_stream_id = st->codecpar->codec_tag ? -1 : pseudo_stream_id;
2488 sc->dref_id= dref_id;
2489 sc->format = format;
2491 id = mov_codec_id(st, format);
2493 av_log(c->fc, AV_LOG_TRACE,
2494 "size=%"PRId64" 4CC=%s codec_type=%d\n", size,
2495 av_fourcc2str(format), st->codecpar->codec_type);
2497 st->codecpar->codec_id = id;
2498 if (st->codecpar->codec_type==AVMEDIA_TYPE_VIDEO) {
2499 mov_parse_stsd_video(c, pb, st, sc);
2500 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_AUDIO) {
2501 mov_parse_stsd_audio(c, pb, st, sc);
2502 if (st->codecpar->sample_rate < 0) {
2503 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
2504 return AVERROR_INVALIDDATA;
2506 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_SUBTITLE){
2507 mov_parse_stsd_subtitle(c, pb, st, sc,
2508 size - (avio_tell(pb) - start_pos));
2510 ret = mov_parse_stsd_data(c, pb, st, sc,
2511 size - (avio_tell(pb) - start_pos));
2515 /* this will read extra atoms at the end (wave, alac, damr, avcC, hvcC, SMI ...) */
2516 a.size = size - (avio_tell(pb) - start_pos);
2518 if ((ret = mov_read_default(c, pb, a)) < 0)
2520 } else if (a.size > 0)
2521 avio_skip(pb, a.size);
2523 if (sc->extradata && st->codecpar->extradata) {
2524 int extra_size = st->codecpar->extradata_size;
2526 /* Move the current stream extradata to the stream context one. */
2527 sc->extradata_size[pseudo_stream_id] = extra_size;
2528 sc->extradata[pseudo_stream_id] = av_malloc(extra_size + AV_INPUT_BUFFER_PADDING_SIZE);
2529 if (!sc->extradata[pseudo_stream_id])
2530 return AVERROR(ENOMEM);
2531 memcpy(sc->extradata[pseudo_stream_id], st->codecpar->extradata, extra_size);
2532 av_freep(&st->codecpar->extradata);
2533 st->codecpar->extradata_size = 0;
2538 if (pb->eof_reached) {
2539 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSD atom\n");
2546 static int mov_read_stsd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2549 MOVStreamContext *sc;
2552 if (c->fc->nb_streams < 1)
2554 st = c->fc->streams[c->fc->nb_streams - 1];
2557 sc->stsd_version = avio_r8(pb);
2558 avio_rb24(pb); /* flags */
2559 entries = avio_rb32(pb);
2561 /* Each entry contains a size (4 bytes) and format (4 bytes). */
2562 if (entries <= 0 || entries > atom.size / 8) {
2563 av_log(c->fc, AV_LOG_ERROR, "invalid STSD entries %d\n", entries);
2564 return AVERROR_INVALIDDATA;
2567 if (sc->extradata) {
2568 av_log(c->fc, AV_LOG_ERROR,
2569 "Duplicate stsd found in this track.\n");
2570 return AVERROR_INVALIDDATA;
2573 /* Prepare space for hosting multiple extradata. */
2574 sc->extradata = av_mallocz_array(entries, sizeof(*sc->extradata));
2576 return AVERROR(ENOMEM);
2578 sc->extradata_size = av_mallocz_array(entries, sizeof(*sc->extradata_size));
2579 if (!sc->extradata_size) {
2580 ret = AVERROR(ENOMEM);
2584 ret = ff_mov_read_stsd_entries(c, pb, entries);
2588 /* Restore back the primary extradata. */
2589 av_freep(&st->codecpar->extradata);
2590 st->codecpar->extradata_size = sc->extradata_size[0];
2591 if (sc->extradata_size[0]) {
2592 st->codecpar->extradata = av_mallocz(sc->extradata_size[0] + AV_INPUT_BUFFER_PADDING_SIZE);
2593 if (!st->codecpar->extradata)
2594 return AVERROR(ENOMEM);
2595 memcpy(st->codecpar->extradata, sc->extradata[0], sc->extradata_size[0]);
2598 return mov_finalize_stsd_codec(c, pb, st, sc);
2600 if (sc->extradata) {
2602 for (j = 0; j < sc->stsd_count; j++)
2603 av_freep(&sc->extradata[j]);
2606 av_freep(&sc->extradata);
2607 av_freep(&sc->extradata_size);
2611 static int mov_read_stsc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2614 MOVStreamContext *sc;
2615 unsigned int i, entries;
2617 if (c->fc->nb_streams < 1)
2619 st = c->fc->streams[c->fc->nb_streams-1];
2622 avio_r8(pb); /* version */
2623 avio_rb24(pb); /* flags */
2625 entries = avio_rb32(pb);
2626 if ((uint64_t)entries * 12 + 4 > atom.size)
2627 return AVERROR_INVALIDDATA;
2629 av_log(c->fc, AV_LOG_TRACE, "track[%u].stsc.entries = %u\n", c->fc->nb_streams - 1, entries);
2634 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSC atom\n");
2635 av_free(sc->stsc_data);
2637 sc->stsc_data = av_malloc_array(entries, sizeof(*sc->stsc_data));
2639 return AVERROR(ENOMEM);
2641 for (i = 0; i < entries && !pb->eof_reached; i++) {
2642 sc->stsc_data[i].first = avio_rb32(pb);
2643 sc->stsc_data[i].count = avio_rb32(pb);
2644 sc->stsc_data[i].id = avio_rb32(pb);
2648 for (i = sc->stsc_count - 1; i < UINT_MAX; i--) {
2649 int64_t first_min = i + 1;
2650 if ((i+1 < sc->stsc_count && sc->stsc_data[i].first >= sc->stsc_data[i+1].first) ||
2651 (i > 0 && sc->stsc_data[i].first <= sc->stsc_data[i-1].first) ||
2652 sc->stsc_data[i].first < first_min ||
2653 sc->stsc_data[i].count < 1 ||
2654 sc->stsc_data[i].id < 1) {
2655 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);
2656 if (i+1 >= sc->stsc_count) {
2657 sc->stsc_data[i].first = FFMAX(sc->stsc_data[i].first, first_min);
2658 if (i > 0 && sc->stsc_data[i].first <= sc->stsc_data[i-1].first)
2659 sc->stsc_data[i].first = FFMIN(sc->stsc_data[i-1].first + 1LL, INT_MAX);
2660 sc->stsc_data[i].count = FFMAX(sc->stsc_data[i].count, 1);
2661 sc->stsc_data[i].id = FFMAX(sc->stsc_data[i].id, 1);
2664 av_assert0(sc->stsc_data[i+1].first >= 2);
2665 // We replace this entry by the next valid
2666 sc->stsc_data[i].first = sc->stsc_data[i+1].first - 1;
2667 sc->stsc_data[i].count = sc->stsc_data[i+1].count;
2668 sc->stsc_data[i].id = sc->stsc_data[i+1].id;
2672 if (pb->eof_reached) {
2673 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSC atom\n");
2680 static inline int mov_stsc_index_valid(unsigned int index, unsigned int count)
2682 return index < count - 1;
2685 /* Compute the samples value for the stsc entry at the given index. */
2686 static inline int64_t mov_get_stsc_samples(MOVStreamContext *sc, unsigned int index)
2690 if (mov_stsc_index_valid(index, sc->stsc_count))
2691 chunk_count = sc->stsc_data[index + 1].first - sc->stsc_data[index].first;
2693 chunk_count = sc->chunk_count - (sc->stsc_data[index].first - 1);
2695 return sc->stsc_data[index].count * (int64_t)chunk_count;
2698 static int mov_read_stps(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2701 MOVStreamContext *sc;
2702 unsigned i, entries;
2704 if (c->fc->nb_streams < 1)
2706 st = c->fc->streams[c->fc->nb_streams-1];
2709 avio_rb32(pb); // version + flags
2711 entries = avio_rb32(pb);
2713 av_log(c->fc, AV_LOG_WARNING, "Duplicated STPS atom\n");
2714 av_free(sc->stps_data);
2716 sc->stps_data = av_malloc_array(entries, sizeof(*sc->stps_data));
2718 return AVERROR(ENOMEM);
2720 for (i = 0; i < entries && !pb->eof_reached; i++) {
2721 sc->stps_data[i] = avio_rb32(pb);
2726 if (pb->eof_reached) {
2727 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STPS atom\n");
2734 static int mov_read_stss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2737 MOVStreamContext *sc;
2738 unsigned int i, entries;
2740 if (c->fc->nb_streams < 1)
2742 st = c->fc->streams[c->fc->nb_streams-1];
2745 avio_r8(pb); /* version */
2746 avio_rb24(pb); /* flags */
2748 entries = avio_rb32(pb);
2750 av_log(c->fc, AV_LOG_TRACE, "keyframe_count = %u\n", entries);
2754 sc->keyframe_absent = 1;
2755 if (!st->need_parsing && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
2756 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2760 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSS atom\n");
2761 if (entries >= UINT_MAX / sizeof(int))
2762 return AVERROR_INVALIDDATA;
2763 av_freep(&sc->keyframes);
2764 sc->keyframe_count = 0;
2765 sc->keyframes = av_malloc_array(entries, sizeof(*sc->keyframes));
2767 return AVERROR(ENOMEM);
2769 for (i = 0; i < entries && !pb->eof_reached; i++) {
2770 sc->keyframes[i] = avio_rb32(pb);
2773 sc->keyframe_count = i;
2775 if (pb->eof_reached) {
2776 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSS atom\n");
2783 static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2786 MOVStreamContext *sc;
2787 unsigned int i, entries, sample_size, field_size, num_bytes;
2792 if (c->fc->nb_streams < 1)
2794 st = c->fc->streams[c->fc->nb_streams-1];
2797 avio_r8(pb); /* version */
2798 avio_rb24(pb); /* flags */
2800 if (atom.type == MKTAG('s','t','s','z')) {
2801 sample_size = avio_rb32(pb);
2802 if (!sc->sample_size) /* do not overwrite value computed in stsd */
2803 sc->sample_size = sample_size;
2804 sc->stsz_sample_size = sample_size;
2808 avio_rb24(pb); /* reserved */
2809 field_size = avio_r8(pb);
2811 entries = avio_rb32(pb);
2813 av_log(c->fc, AV_LOG_TRACE, "sample_size = %u sample_count = %u\n", sc->sample_size, entries);
2815 sc->sample_count = entries;
2819 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
2820 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %u\n", field_size);
2821 return AVERROR_INVALIDDATA;
2826 if (entries >= (UINT_MAX - 4) / field_size)
2827 return AVERROR_INVALIDDATA;
2828 if (sc->sample_sizes)
2829 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSZ atom\n");
2830 av_free(sc->sample_sizes);
2831 sc->sample_count = 0;
2832 sc->sample_sizes = av_malloc_array(entries, sizeof(*sc->sample_sizes));
2833 if (!sc->sample_sizes)
2834 return AVERROR(ENOMEM);
2836 num_bytes = (entries*field_size+4)>>3;
2838 buf = av_malloc(num_bytes+AV_INPUT_BUFFER_PADDING_SIZE);
2840 av_freep(&sc->sample_sizes);
2841 return AVERROR(ENOMEM);
2844 ret = ffio_read_size(pb, buf, num_bytes);
2846 av_freep(&sc->sample_sizes);
2848 av_log(c->fc, AV_LOG_WARNING, "STSZ atom truncated\n");
2852 init_get_bits(&gb, buf, 8*num_bytes);
2854 for (i = 0; i < entries && !pb->eof_reached; i++) {
2855 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
2856 sc->data_size += sc->sample_sizes[i];
2859 sc->sample_count = i;
2863 if (pb->eof_reached) {
2864 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STSZ atom\n");
2871 static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2874 MOVStreamContext *sc;
2875 unsigned int i, entries, alloc_size = 0;
2877 int64_t total_sample_count=0;
2879 if (c->fc->nb_streams < 1)
2881 st = c->fc->streams[c->fc->nb_streams-1];
2884 avio_r8(pb); /* version */
2885 avio_rb24(pb); /* flags */
2886 entries = avio_rb32(pb);
2888 av_log(c->fc, AV_LOG_TRACE, "track[%u].stts.entries = %u\n",
2889 c->fc->nb_streams-1, entries);
2892 av_log(c->fc, AV_LOG_WARNING, "Duplicated STTS atom\n");
2893 av_freep(&sc->stts_data);
2895 if (entries >= INT_MAX / sizeof(*sc->stts_data))
2896 return AVERROR(ENOMEM);
2898 for (i = 0; i < entries && !pb->eof_reached; i++) {
2899 int sample_duration;
2900 unsigned int sample_count;
2901 unsigned int min_entries = FFMIN(FFMAX(i + 1, 1024 * 1024), entries);
2902 MOVStts *stts_data = av_fast_realloc(sc->stts_data, &alloc_size,
2903 min_entries * sizeof(*sc->stts_data));
2905 av_freep(&sc->stts_data);
2907 return AVERROR(ENOMEM);
2909 sc->stts_count = min_entries;
2910 sc->stts_data = stts_data;
2912 sample_count=avio_rb32(pb);
2913 sample_duration = avio_rb32(pb);
2915 sc->stts_data[i].count= sample_count;
2916 sc->stts_data[i].duration= sample_duration;
2918 av_log(c->fc, AV_LOG_TRACE, "sample_count=%d, sample_duration=%d\n",
2919 sample_count, sample_duration);
2923 && sample_count == 1
2924 && total_sample_count > 100
2925 && sample_duration/10 > duration / total_sample_count)
2926 sample_duration = duration / total_sample_count;
2927 duration+=(int64_t)sample_duration*(uint64_t)sample_count;
2928 total_sample_count+=sample_count;
2934 duration <= INT64_MAX - sc->duration_for_fps &&
2935 total_sample_count <= INT64_MAX - sc->nb_frames_for_fps
2937 sc->duration_for_fps += duration;
2938 sc->nb_frames_for_fps += total_sample_count;
2941 if (pb->eof_reached) {
2942 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted STTS atom\n");
2946 st->nb_frames= total_sample_count;
2948 st->duration= FFMIN(st->duration, duration);
2949 sc->track_end = duration;
2953 static void mov_update_dts_shift(MOVStreamContext *sc, int duration)
2956 if (duration == INT_MIN) {
2957 av_log(NULL, AV_LOG_WARNING, "mov_update_dts_shift(): dts_shift set to %d\n", INT_MAX);
2960 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
2964 static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2967 MOVStreamContext *sc;
2968 unsigned int i, entries, ctts_count = 0;
2970 if (c->fc->nb_streams < 1)
2972 st = c->fc->streams[c->fc->nb_streams-1];
2975 avio_r8(pb); /* version */
2976 avio_rb24(pb); /* flags */
2977 entries = avio_rb32(pb);
2979 av_log(c->fc, AV_LOG_TRACE, "track[%u].ctts.entries = %u\n", c->fc->nb_streams - 1, entries);
2983 if (entries >= UINT_MAX / sizeof(*sc->ctts_data))
2984 return AVERROR_INVALIDDATA;
2985 av_freep(&sc->ctts_data);
2986 sc->ctts_data = av_fast_realloc(NULL, &sc->ctts_allocated_size, entries * sizeof(*sc->ctts_data));
2988 return AVERROR(ENOMEM);
2990 for (i = 0; i < entries && !pb->eof_reached; i++) {
2991 int count =avio_rb32(pb);
2992 int duration =avio_rb32(pb);
2995 av_log(c->fc, AV_LOG_TRACE,
2996 "ignoring CTTS entry with count=%d duration=%d\n",
3001 add_ctts_entry(&sc->ctts_data, &ctts_count, &sc->ctts_allocated_size,
3004 av_log(c->fc, AV_LOG_TRACE, "count=%d, duration=%d\n",
3007 if (FFNABS(duration) < -(1<<28) && i+2<entries) {
3008 av_log(c->fc, AV_LOG_WARNING, "CTTS invalid\n");
3009 av_freep(&sc->ctts_data);
3015 mov_update_dts_shift(sc, duration);
3018 sc->ctts_count = ctts_count;
3020 if (pb->eof_reached) {
3021 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted CTTS atom\n");
3025 av_log(c->fc, AV_LOG_TRACE, "dts shift %d\n", sc->dts_shift);
3030 static int mov_read_sbgp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3033 MOVStreamContext *sc;
3034 unsigned int i, entries;
3036 uint32_t grouping_type;
3038 if (c->fc->nb_streams < 1)
3040 st = c->fc->streams[c->fc->nb_streams-1];
3043 version = avio_r8(pb); /* version */
3044 avio_rb24(pb); /* flags */
3045 grouping_type = avio_rl32(pb);
3046 if (grouping_type != MKTAG( 'r','a','p',' '))
3047 return 0; /* only support 'rap ' grouping */
3049 avio_rb32(pb); /* grouping_type_parameter */
3051 entries = avio_rb32(pb);
3055 av_log(c->fc, AV_LOG_WARNING, "Duplicated SBGP atom\n");
3056 av_free(sc->rap_group);
3057 sc->rap_group_count = 0;
3058 sc->rap_group = av_malloc_array(entries, sizeof(*sc->rap_group));
3060 return AVERROR(ENOMEM);
3062 for (i = 0; i < entries && !pb->eof_reached; i++) {
3063 sc->rap_group[i].count = avio_rb32(pb); /* sample_count */
3064 sc->rap_group[i].index = avio_rb32(pb); /* group_description_index */
3067 sc->rap_group_count = i;
3069 if (pb->eof_reached) {
3070 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted SBGP atom\n");
3078 * Get ith edit list entry (media time, duration).
3080 static int get_edit_list_entry(MOVContext *mov,
3081 const MOVStreamContext *msc,
3082 unsigned int edit_list_index,
3083 int64_t *edit_list_media_time,
3084 int64_t *edit_list_duration,
3085 int64_t global_timescale)
3087 if (edit_list_index == msc->elst_count) {
3090 *edit_list_media_time = msc->elst_data[edit_list_index].time;
3091 *edit_list_duration = msc->elst_data[edit_list_index].duration;
3093 /* duration is in global timescale units;convert to msc timescale */
3094 if (global_timescale == 0) {
3095 avpriv_request_sample(mov->fc, "Support for mvhd.timescale = 0 with editlists");
3098 *edit_list_duration = av_rescale(*edit_list_duration, msc->time_scale,
3104 * Find the closest previous frame to the timestamp_pts, in e_old index
3105 * entries. Searching for just any frame / just key frames can be controlled by
3106 * last argument 'flag'.
3107 * Note that if ctts_data is not NULL, we will always search for a key frame
3108 * irrespective of the value of 'flag'. If we don't find any keyframe, we will
3109 * return the first frame of the video.
3111 * Here the timestamp_pts is considered to be a presentation timestamp and
3112 * the timestamp of index entries are considered to be decoding timestamps.
3114 * Returns 0 if successful in finding a frame, else returns -1.
3115 * Places the found index corresponding output arg.
3117 * If ctts_old is not NULL, then refines the searched entry by searching
3118 * backwards from the found timestamp, to find the frame with correct PTS.
3120 * Places the found ctts_index and ctts_sample in corresponding output args.
3122 static int find_prev_closest_index(AVStream *st,
3123 AVIndexEntry *e_old,
3127 int64_t timestamp_pts,
3130 int64_t* ctts_index,
3131 int64_t* ctts_sample)
3133 MOVStreamContext *msc = st->priv_data;
3134 AVIndexEntry *e_keep = st->index_entries;
3135 int nb_keep = st->nb_index_entries;
3137 int64_t index_ctts_count;
3141 // If dts_shift > 0, then all the index timestamps will have to be offset by
3142 // at least dts_shift amount to obtain PTS.
3143 // Hence we decrement the searched timestamp_pts by dts_shift to find the closest index element.
3144 if (msc->dts_shift > 0) {
3145 timestamp_pts -= msc->dts_shift;
3148 st->index_entries = e_old;
3149 st->nb_index_entries = nb_old;
3150 *index = av_index_search_timestamp(st, timestamp_pts, flag | AVSEEK_FLAG_BACKWARD);
3152 // Keep going backwards in the index entries until the timestamp is the same.
3154 for (i = *index; i > 0 && e_old[i].timestamp == e_old[i - 1].timestamp;
3156 if ((flag & AVSEEK_FLAG_ANY) ||
3157 (e_old[i - 1].flags & AVINDEX_KEYFRAME)) {
3163 // If we have CTTS then refine the search, by searching backwards over PTS
3164 // computed by adding corresponding CTTS durations to index timestamps.
3165 if (ctts_data && *index >= 0) {
3166 av_assert0(ctts_index);
3167 av_assert0(ctts_sample);
3168 // Find out the ctts_index for the found frame.
3171 for (index_ctts_count = 0; index_ctts_count < *index; index_ctts_count++) {
3172 if (*ctts_index < ctts_count) {
3174 if (ctts_data[*ctts_index].count == *ctts_sample) {
3181 while (*index >= 0 && (*ctts_index) >= 0 && (*ctts_index) < ctts_count) {
3182 // Find a "key frame" with PTS <= timestamp_pts (So that we can decode B-frames correctly).
3183 // No need to add dts_shift to the timestamp here becase timestamp_pts has already been
3184 // compensated by dts_shift above.
3185 if ((e_old[*index].timestamp + ctts_data[*ctts_index].duration) <= timestamp_pts &&
3186 (e_old[*index].flags & AVINDEX_KEYFRAME)) {
3191 if (*ctts_sample == 0) {
3193 if (*ctts_index >= 0)
3194 *ctts_sample = ctts_data[*ctts_index].count - 1;
3201 /* restore AVStream state*/
3202 st->index_entries = e_keep;
3203 st->nb_index_entries = nb_keep;
3204 return *index >= 0 ? 0 : -1;
3208 * Add index entry with the given values, to the end of st->index_entries.
3209 * Returns the new size st->index_entries if successful, else returns -1.
3211 * This function is similar to ff_add_index_entry in libavformat/utils.c
3212 * except that here we are always unconditionally adding an index entry to
3213 * the end, instead of searching the entries list and skipping the add if
3214 * there is an existing entry with the same timestamp.
3215 * This is needed because the mov_fix_index calls this func with the same
3216 * unincremented timestamp for successive discarded frames.
3218 static int64_t add_index_entry(AVStream *st, int64_t pos, int64_t timestamp,
3219 int size, int distance, int flags)
3221 AVIndexEntry *entries, *ie;
3223 const size_t min_size_needed = (st->nb_index_entries + 1) * sizeof(AVIndexEntry);
3225 // Double the allocation each time, to lower memory fragmentation.
3226 // Another difference from ff_add_index_entry function.
3227 const size_t requested_size =
3228 min_size_needed > st->index_entries_allocated_size ?
3229 FFMAX(min_size_needed, 2 * st->index_entries_allocated_size) :
3232 if((unsigned)st->nb_index_entries + 1 >= UINT_MAX / sizeof(AVIndexEntry))
3235 entries = av_fast_realloc(st->index_entries,
3236 &st->index_entries_allocated_size,
3241 st->index_entries= entries;
3243 index= st->nb_index_entries++;
3244 ie= &entries[index];
3247 ie->timestamp = timestamp;
3248 ie->min_distance= distance;
3255 * Rewrite timestamps of index entries in the range [end_index - frame_duration_buffer_size, end_index)
3256 * by subtracting end_ts successively by the amounts given in frame_duration_buffer.
3258 static void fix_index_entry_timestamps(AVStream* st, int end_index, int64_t end_ts,
3259 int64_t* frame_duration_buffer,
3260 int frame_duration_buffer_size) {
3262 av_assert0(end_index >= 0 && end_index <= st->nb_index_entries);
3263 for (i = 0; i < frame_duration_buffer_size; i++) {
3264 end_ts -= frame_duration_buffer[frame_duration_buffer_size - 1 - i];
3265 st->index_entries[end_index - 1 - i].timestamp = end_ts;
3270 * Append a new ctts entry to ctts_data.
3271 * Returns the new ctts_count if successful, else returns -1.
3273 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
3274 int count, int duration)
3276 MOVStts *ctts_buf_new;
3277 const size_t min_size_needed = (*ctts_count + 1) * sizeof(MOVStts);
3278 const size_t requested_size =
3279 min_size_needed > *allocated_size ?
3280 FFMAX(min_size_needed, 2 * (*allocated_size)) :
3283 if((unsigned)(*ctts_count) >= UINT_MAX / sizeof(MOVStts) - 1)
3286 ctts_buf_new = av_fast_realloc(*ctts_data, allocated_size, requested_size);
3291 *ctts_data = ctts_buf_new;
3293 ctts_buf_new[*ctts_count].count = count;
3294 ctts_buf_new[*ctts_count].duration = duration;
3296 *ctts_count = (*ctts_count) + 1;
3300 #define MAX_REORDER_DELAY 16
3301 static void mov_estimate_video_delay(MOVContext *c, AVStream* st) {
3302 MOVStreamContext *msc = st->priv_data;
3305 int ctts_sample = 0;
3306 int64_t pts_buf[MAX_REORDER_DELAY + 1]; // Circular buffer to sort pts.
3308 int j, r, num_swaps;
3310 for (j = 0; j < MAX_REORDER_DELAY + 1; j++)
3311 pts_buf[j] = INT64_MIN;
3313 if (st->codecpar->video_delay <= 0 && msc->ctts_data &&
3314 st->codecpar->codec_id == AV_CODEC_ID_H264) {
3315 st->codecpar->video_delay = 0;
3316 for(ind = 0; ind < st->nb_index_entries && ctts_ind < msc->ctts_count; ++ind) {
3317 // Point j to the last elem of the buffer and insert the current pts there.
3319 buf_start = (buf_start + 1);
3320 if (buf_start == MAX_REORDER_DELAY + 1)
3323 pts_buf[j] = st->index_entries[ind].timestamp + msc->ctts_data[ctts_ind].duration;
3325 // The timestamps that are already in the sorted buffer, and are greater than the
3326 // current pts, are exactly the timestamps that need to be buffered to output PTS
3327 // in correct sorted order.
3328 // Hence the video delay (which is the buffer size used to sort DTS and output PTS),
3329 // can be computed as the maximum no. of swaps any particular timestamp needs to
3330 // go through, to keep this buffer in sorted order.
3332 while (j != buf_start) {
3334 if (r < 0) r = MAX_REORDER_DELAY;
3335 if (pts_buf[j] < pts_buf[r]) {
3336 FFSWAP(int64_t, pts_buf[j], pts_buf[r]);
3343 st->codecpar->video_delay = FFMAX(st->codecpar->video_delay, num_swaps);
3346 if (ctts_sample == msc->ctts_data[ctts_ind].count) {
3351 av_log(c->fc, AV_LOG_DEBUG, "Setting codecpar->delay to %d for stream st: %d\n",
3352 st->codecpar->video_delay, st->index);
3356 static void mov_current_sample_inc(MOVStreamContext *sc)
3358 sc->current_sample++;
3359 sc->current_index++;
3360 if (sc->index_ranges &&
3361 sc->current_index >= sc->current_index_range->end &&
3362 sc->current_index_range->end) {
3363 sc->current_index_range++;
3364 sc->current_index = sc->current_index_range->start;
3368 static void mov_current_sample_dec(MOVStreamContext *sc)
3370 sc->current_sample--;
3371 sc->current_index--;
3372 if (sc->index_ranges &&
3373 sc->current_index < sc->current_index_range->start &&
3374 sc->current_index_range > sc->index_ranges) {
3375 sc->current_index_range--;
3376 sc->current_index = sc->current_index_range->end - 1;
3380 static void mov_current_sample_set(MOVStreamContext *sc, int current_sample)
3384 sc->current_sample = current_sample;
3385 sc->current_index = current_sample;
3386 if (!sc->index_ranges) {
3390 for (sc->current_index_range = sc->index_ranges;
3391 sc->current_index_range->end;
3392 sc->current_index_range++) {
3393 range_size = sc->current_index_range->end - sc->current_index_range->start;
3394 if (range_size > current_sample) {
3395 sc->current_index = sc->current_index_range->start + current_sample;
3398 current_sample -= range_size;
3403 * Fix st->index_entries, so that it contains only the entries (and the entries
3404 * which are needed to decode them) that fall in the edit list time ranges.
3405 * Also fixes the timestamps of the index entries to match the timeline
3406 * specified the edit lists.
3408 static void mov_fix_index(MOVContext *mov, AVStream *st)
3410 MOVStreamContext *msc = st->priv_data;
3411 AVIndexEntry *e_old = st->index_entries;
3412 int nb_old = st->nb_index_entries;
3413 const AVIndexEntry *e_old_end = e_old + nb_old;
3414 const AVIndexEntry *current = NULL;
3415 MOVStts *ctts_data_old = msc->ctts_data;
3416 int64_t ctts_index_old = 0;
3417 int64_t ctts_sample_old = 0;
3418 int64_t ctts_count_old = msc->ctts_count;
3419 int64_t edit_list_media_time = 0;
3420 int64_t edit_list_duration = 0;
3421 int64_t frame_duration = 0;
3422 int64_t edit_list_dts_counter = 0;
3423 int64_t edit_list_dts_entry_end = 0;
3424 int64_t edit_list_start_ctts_sample = 0;
3426 int64_t curr_ctts = 0;
3427 int64_t empty_edits_sum_duration = 0;
3428 int64_t edit_list_index = 0;
3431 int64_t start_dts = 0;
3432 int64_t edit_list_start_encountered = 0;
3433 int64_t search_timestamp = 0;
3434 int64_t* frame_duration_buffer = NULL;
3435 int num_discarded_begin = 0;
3436 int first_non_zero_audio_edit = -1;
3437 int packet_skip_samples = 0;
3438 MOVIndexRange *current_index_range;
3440 int found_keyframe_after_edit = 0;
3442 if (!msc->elst_data || msc->elst_count <= 0 || nb_old <= 0) {
3446 // allocate the index ranges array
3447 msc->index_ranges = av_malloc((msc->elst_count + 1) * sizeof(msc->index_ranges[0]));
3448 if (!msc->index_ranges) {
3449 av_log(mov->fc, AV_LOG_ERROR, "Cannot allocate index ranges buffer\n");
3452 msc->current_index_range = msc->index_ranges;
3453 current_index_range = msc->index_ranges - 1;
3455 // Clean AVStream from traces of old index
3456 st->index_entries = NULL;
3457 st->index_entries_allocated_size = 0;
3458 st->nb_index_entries = 0;
3460 // Clean ctts fields of MOVStreamContext
3461 msc->ctts_data = NULL;
3462 msc->ctts_count = 0;
3463 msc->ctts_index = 0;
3464 msc->ctts_sample = 0;
3465 msc->ctts_allocated_size = 0;
3467 // Reinitialize min_corrected_pts so that it can be computed again.
3468 msc->min_corrected_pts = -1;
3470 // If the dts_shift is positive (in case of negative ctts values in mov),
3471 // then negate the DTS by dts_shift
3472 if (msc->dts_shift > 0) {
3473 edit_list_dts_entry_end -= msc->dts_shift;
3474 av_log(mov->fc, AV_LOG_DEBUG, "Shifting DTS by %d because of negative CTTS.\n", msc->dts_shift);
3477 start_dts = edit_list_dts_entry_end;
3479 while (get_edit_list_entry(mov, msc, edit_list_index, &edit_list_media_time,
3480 &edit_list_duration, mov->time_scale)) {
3481 av_log(mov->fc, AV_LOG_DEBUG, "Processing st: %d, edit list %"PRId64" - media time: %"PRId64", duration: %"PRId64"\n",
3482 st->index, edit_list_index, edit_list_media_time, edit_list_duration);
3484 edit_list_dts_counter = edit_list_dts_entry_end;
3485 edit_list_dts_entry_end += edit_list_duration;
3486 num_discarded_begin = 0;
3487 if (edit_list_media_time == -1) {
3488 empty_edits_sum_duration += edit_list_duration;
3492 // If we encounter a non-negative edit list reset the skip_samples/start_pad fields and set them
3493 // according to the edit list below.
3494 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3495 if (first_non_zero_audio_edit < 0) {
3496 first_non_zero_audio_edit = 1;
3498 first_non_zero_audio_edit = 0;
3501 if (first_non_zero_audio_edit > 0)
3502 st->skip_samples = msc->start_pad = 0;
3505 // While reordering frame index according to edit list we must handle properly
3506 // the scenario when edit list entry starts from none key frame.
3507 // We find closest previous key frame and preserve it and consequent frames in index.
3508 // All frames which are outside edit list entry time boundaries will be dropped after decoding.
3509 search_timestamp = edit_list_media_time;
3510 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3511 // Audio decoders like AAC need need a decoder delay samples previous to the current sample,
3512 // to correctly decode this frame. Hence for audio we seek to a frame 1 sec. before the
3513 // edit_list_media_time to cover the decoder delay.
3514 search_timestamp = FFMAX(search_timestamp - msc->time_scale, e_old[0].timestamp);
3517 if (find_prev_closest_index(st, e_old, nb_old, ctts_data_old, ctts_count_old, search_timestamp, 0,
3518 &index, &ctts_index_old, &ctts_sample_old) < 0) {
3519 av_log(mov->fc, AV_LOG_WARNING,
3520 "st: %d edit list: %"PRId64" Missing key frame while searching for timestamp: %"PRId64"\n",
3521 st->index, edit_list_index, search_timestamp);
3522 if (find_prev_closest_index(st, e_old, nb_old, ctts_data_old, ctts_count_old, search_timestamp, AVSEEK_FLAG_ANY,
3523 &index, &ctts_index_old, &ctts_sample_old) < 0) {
3524 av_log(mov->fc, AV_LOG_WARNING,
3525 "st: %d edit list %"PRId64" Cannot find an index entry before timestamp: %"PRId64".\n",
3526 st->index, edit_list_index, search_timestamp);
3529 ctts_sample_old = 0;
3532 current = e_old + index;
3533 edit_list_start_ctts_sample = ctts_sample_old;
3535 // Iterate over index and arrange it according to edit list
3536 edit_list_start_encountered = 0;
3537 found_keyframe_after_edit = 0;
3538 for (; current < e_old_end; current++, index++) {
3539 // check if frame outside edit list mark it for discard
3540 frame_duration = (current + 1 < e_old_end) ?
3541 ((current + 1)->timestamp - current->timestamp) : edit_list_duration;
3543 flags = current->flags;
3545 // frames (pts) before or after edit list
3546 curr_cts = current->timestamp + msc->dts_shift;
3549 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3550 curr_ctts = ctts_data_old[ctts_index_old].duration;
3551 av_log(mov->fc, AV_LOG_DEBUG, "stts: %"PRId64" ctts: %"PRId64", ctts_index: %"PRId64", ctts_count: %"PRId64"\n",
3552 curr_cts, curr_ctts, ctts_index_old, ctts_count_old);
3553 curr_cts += curr_ctts;
3555 if (ctts_sample_old == ctts_data_old[ctts_index_old].count) {
3556 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3557 &msc->ctts_allocated_size,
3558 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3559 ctts_data_old[ctts_index_old].duration) == -1) {
3560 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3562 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3563 ctts_data_old[ctts_index_old].duration);
3567 ctts_sample_old = 0;
3568 edit_list_start_ctts_sample = 0;
3572 if (curr_cts < edit_list_media_time || curr_cts >= (edit_list_duration + edit_list_media_time)) {
3573 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id != AV_CODEC_ID_VORBIS &&
3574 curr_cts < edit_list_media_time && curr_cts + frame_duration > edit_list_media_time &&
3575 first_non_zero_audio_edit > 0) {
3576 packet_skip_samples = edit_list_media_time - curr_cts;
3577 st->skip_samples += packet_skip_samples;
3579 // Shift the index entry timestamp by packet_skip_samples to be correct.
3580 edit_list_dts_counter -= packet_skip_samples;
3581 if (edit_list_start_encountered == 0) {
3582 edit_list_start_encountered = 1;
3583 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3584 // discarded packets.
3585 if (frame_duration_buffer) {
3586 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3587 frame_duration_buffer, num_discarded_begin);
3588 av_freep(&frame_duration_buffer);
3592 av_log(mov->fc, AV_LOG_DEBUG, "skip %d audio samples from curr_cts: %"PRId64"\n", packet_skip_samples, curr_cts);
3594 flags |= AVINDEX_DISCARD_FRAME;
3595 av_log(mov->fc, AV_LOG_DEBUG, "drop a frame at curr_cts: %"PRId64" @ %"PRId64"\n", curr_cts, index);
3597 if (edit_list_start_encountered == 0) {
3598 num_discarded_begin++;
3599 frame_duration_buffer = av_realloc(frame_duration_buffer,
3600 num_discarded_begin * sizeof(int64_t));
3601 if (!frame_duration_buffer) {
3602 av_log(mov->fc, AV_LOG_ERROR, "Cannot reallocate frame duration buffer\n");
3605 frame_duration_buffer[num_discarded_begin - 1] = frame_duration;
3607 // Increment skip_samples for the first non-zero audio edit list
3608 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3609 first_non_zero_audio_edit > 0 && st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
3610 st->skip_samples += frame_duration;
3615 if (msc->min_corrected_pts < 0) {
3616 msc->min_corrected_pts = edit_list_dts_counter + curr_ctts + msc->dts_shift;
3618 msc->min_corrected_pts = FFMIN(msc->min_corrected_pts, edit_list_dts_counter + curr_ctts + msc->dts_shift);
3620 if (edit_list_start_encountered == 0) {
3621 edit_list_start_encountered = 1;
3622 // Make timestamps strictly monotonically increasing by rewriting timestamps for
3623 // discarded packets.
3624 if (frame_duration_buffer) {
3625 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3626 frame_duration_buffer, num_discarded_begin);
3627 av_freep(&frame_duration_buffer);
3632 if (add_index_entry(st, current->pos, edit_list_dts_counter, current->size,
3633 current->min_distance, flags) == -1) {
3634 av_log(mov->fc, AV_LOG_ERROR, "Cannot add index entry\n");
3638 // Update the index ranges array
3639 if (current_index_range < msc->index_ranges || index != current_index_range->end) {
3640 current_index_range++;
3641 current_index_range->start = index;
3643 current_index_range->end = index + 1;
3645 // Only start incrementing DTS in frame_duration amounts, when we encounter a frame in edit list.
3646 if (edit_list_start_encountered > 0) {
3647 edit_list_dts_counter = edit_list_dts_counter + frame_duration;
3650 // Break when found first key frame after edit entry completion
3651 if ((curr_cts + frame_duration >= (edit_list_duration + edit_list_media_time)) &&
3652 ((flags & AVINDEX_KEYFRAME) || ((st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)))) {
3653 if (ctts_data_old) {
3654 // If we have CTTS and this is the first keyframe after edit elist,
3655 // wait for one more, because there might be trailing B-frames after this I-frame
3656 // that do belong to the edit.
3657 if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO && found_keyframe_after_edit == 0) {
3658 found_keyframe_after_edit = 1;
3661 if (ctts_sample_old != 0) {
3662 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3663 &msc->ctts_allocated_size,
3664 ctts_sample_old - edit_list_start_ctts_sample,
3665 ctts_data_old[ctts_index_old].duration) == -1) {
3666 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3667 ctts_index_old, ctts_sample_old - edit_list_start_ctts_sample,
3668 ctts_data_old[ctts_index_old].duration);
3677 // If there are empty edits, then msc->min_corrected_pts might be positive
3678 // intentionally. So we subtract the sum duration of emtpy edits here.
3679 msc->min_corrected_pts -= empty_edits_sum_duration;
3681 // If the minimum pts turns out to be greater than zero after fixing the index, then we subtract the
3682 // dts by that amount to make the first pts zero.
3683 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
3684 if (msc->min_corrected_pts > 0) {
3685 av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by %"PRId64" to make first pts zero.\n", msc->min_corrected_pts);
3686 for (i = 0; i < st->nb_index_entries; ++i) {
3687 st->index_entries[i].timestamp -= msc->min_corrected_pts;
3691 // Start time should be equal to zero or the duration of any empty edits.
3692 st->start_time = empty_edits_sum_duration;
3694 // Update av stream length, if it ends up shorter than the track's media duration
3695 st->duration = FFMIN(st->duration, edit_list_dts_entry_end - start_dts);
3696 msc->start_pad = st->skip_samples;
3698 // Free the old index and the old CTTS structures
3700 av_free(ctts_data_old);
3701 av_freep(&frame_duration_buffer);
3703 // Null terminate the index ranges array
3704 current_index_range++;
3705 current_index_range->start = 0;
3706 current_index_range->end = 0;
3707 msc->current_index = msc->index_ranges[0].start;
3710 static void mov_build_index(MOVContext *mov, AVStream *st)
3712 MOVStreamContext *sc = st->priv_data;
3713 int64_t current_offset;
3714 int64_t current_dts = 0;
3715 unsigned int stts_index = 0;
3716 unsigned int stsc_index = 0;
3717 unsigned int stss_index = 0;
3718 unsigned int stps_index = 0;
3720 uint64_t stream_size = 0;
3721 MOVStts *ctts_data_old = sc->ctts_data;
3722 unsigned int ctts_count_old = sc->ctts_count;
3724 if (sc->elst_count) {
3725 int i, edit_start_index = 0, multiple_edits = 0;
3726 int64_t empty_duration = 0; // empty duration of the first edit list entry
3727 int64_t start_time = 0; // start time of the media
3729 for (i = 0; i < sc->elst_count; i++) {
3730 const MOVElst *e = &sc->elst_data[i];
3731 if (i == 0 && e->time == -1) {
3732 /* if empty, the first entry is the start time of the stream
3733 * relative to the presentation itself */
3734 empty_duration = e->duration;
3735 edit_start_index = 1;
3736 } else if (i == edit_start_index && e->time >= 0) {
3737 start_time = e->time;
3743 if (multiple_edits && !mov->advanced_editlist)
3744 av_log(mov->fc, AV_LOG_WARNING, "multiple edit list entries, "
3745 "Use -advanced_editlist to correctly decode otherwise "
3746 "a/v desync might occur\n");
3748 /* adjust first dts according to edit list */
3749 if ((empty_duration || start_time) && mov->time_scale > 0) {
3751 empty_duration = av_rescale(empty_duration, sc->time_scale, mov->time_scale);
3752 sc->time_offset = start_time - empty_duration;
3753 sc->min_corrected_pts = start_time;
3754 if (!mov->advanced_editlist)
3755 current_dts = -sc->time_offset;
3758 if (!multiple_edits && !mov->advanced_editlist &&
3759 st->codecpar->codec_id == AV_CODEC_ID_AAC && start_time > 0)
3760 sc->start_pad = start_time;
3763 /* only use old uncompressed audio chunk demuxing when stts specifies it */
3764 if (!(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3765 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
3766 unsigned int current_sample = 0;
3767 unsigned int stts_sample = 0;
3768 unsigned int sample_size;
3769 unsigned int distance = 0;
3770 unsigned int rap_group_index = 0;
3771 unsigned int rap_group_sample = 0;
3772 int64_t last_dts = 0;
3773 int64_t dts_correction = 0;
3774 int rap_group_present = sc->rap_group_count && sc->rap_group;
3775 int key_off = (sc->keyframe_count && sc->keyframes[0] > 0) || (sc->stps_count && sc->stps_data[0] > 0);
3777 current_dts -= sc->dts_shift;
3778 last_dts = current_dts;
3780 if (!sc->sample_count || st->nb_index_entries)
3782 if (sc->sample_count >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3784 if (av_reallocp_array(&st->index_entries,
3785 st->nb_index_entries + sc->sample_count,
3786 sizeof(*st->index_entries)) < 0) {
3787 st->nb_index_entries = 0;
3790 st->index_entries_allocated_size = (st->nb_index_entries + sc->sample_count) * sizeof(*st->index_entries);
3792 if (ctts_data_old) {
3793 // Expand ctts entries such that we have a 1-1 mapping with samples
3794 if (sc->sample_count >= UINT_MAX / sizeof(*sc->ctts_data))
3797 sc->ctts_allocated_size = 0;
3798 sc->ctts_data = av_fast_realloc(NULL, &sc->ctts_allocated_size,
3799 sc->sample_count * sizeof(*sc->ctts_data));
3800 if (!sc->ctts_data) {
3801 av_free(ctts_data_old);
3805 memset((uint8_t*)(sc->ctts_data), 0, sc->ctts_allocated_size);
3807 for (i = 0; i < ctts_count_old &&
3808 sc->ctts_count < sc->sample_count; i++)
3809 for (j = 0; j < ctts_data_old[i].count &&
3810 sc->ctts_count < sc->sample_count; j++)
3811 add_ctts_entry(&sc->ctts_data, &sc->ctts_count,
3812 &sc->ctts_allocated_size, 1,
3813 ctts_data_old[i].duration);
3814 av_free(ctts_data_old);
3817 for (i = 0; i < sc->chunk_count; i++) {
3818 int64_t next_offset = i+1 < sc->chunk_count ? sc->chunk_offsets[i+1] : INT64_MAX;
3819 current_offset = sc->chunk_offsets[i];
3820 while (mov_stsc_index_valid(stsc_index, sc->stsc_count) &&
3821 i + 1 == sc->stsc_data[stsc_index + 1].first)
3824 if (next_offset > current_offset && sc->sample_size>0 && sc->sample_size < sc->stsz_sample_size &&
3825 sc->stsc_data[stsc_index].count * (int64_t)sc->stsz_sample_size > next_offset - current_offset) {
3826 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too large), ignoring\n", sc->stsz_sample_size);
3827 sc->stsz_sample_size = sc->sample_size;
3829 if (sc->stsz_sample_size>0 && sc->stsz_sample_size < sc->sample_size) {
3830 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too small), ignoring\n", sc->stsz_sample_size);
3831 sc->stsz_sample_size = sc->sample_size;
3834 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
3836 if (current_sample >= sc->sample_count) {
3837 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
3841 if (!sc->keyframe_absent && (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index])) {
3843 if (stss_index + 1 < sc->keyframe_count)
3845 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
3847 if (stps_index + 1 < sc->stps_count)
3850 if (rap_group_present && rap_group_index < sc->rap_group_count) {
3851 if (sc->rap_group[rap_group_index].index > 0)
3853 if (++rap_group_sample == sc->rap_group[rap_group_index].count) {
3854 rap_group_sample = 0;
3858 if (sc->keyframe_absent
3860 && !rap_group_present
3861 && (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO || (i==0 && j==0)))
3865 sample_size = sc->stsz_sample_size > 0 ? sc->stsz_sample_size : sc->sample_sizes[current_sample];
3866 if (sc->pseudo_stream_id == -1 ||
3867 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
3869 if (sample_size > 0x3FFFFFFF) {
3870 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", sample_size);
3873 e = &st->index_entries[st->nb_index_entries++];
3874 e->pos = current_offset;
3875 e->timestamp = current_dts;
3876 e->size = sample_size;
3877 e->min_distance = distance;
3878 e->flags = keyframe ? AVINDEX_KEYFRAME : 0;
3879 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %u, offset %"PRIx64", dts %"PRId64", "
3880 "size %u, distance %u, keyframe %d\n", st->index, current_sample,
3881 current_offset, current_dts, sample_size, distance, keyframe);
3882 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries < 100)
3883 ff_rfps_add_frame(mov->fc, st, current_dts);
3886 current_offset += sample_size;
3887 stream_size += sample_size;
3889 /* A negative sample duration is invalid based on the spec,
3890 * but some samples need it to correct the DTS. */
3891 if (sc->stts_data[stts_index].duration < 0) {
3892 av_log(mov->fc, AV_LOG_WARNING,
3893 "Invalid SampleDelta %d in STTS, at %d st:%d\n",
3894 sc->stts_data[stts_index].duration, stts_index,
3896 dts_correction += sc->stts_data[stts_index].duration - 1;
3897 sc->stts_data[stts_index].duration = 1;
3899 current_dts += sc->stts_data[stts_index].duration;
3900 if (!dts_correction || current_dts + dts_correction > last_dts) {
3901 current_dts += dts_correction;
3904 /* Avoid creating non-monotonous DTS */
3905 dts_correction += current_dts - last_dts - 1;
3906 current_dts = last_dts + 1;
3908 last_dts = current_dts;
3912 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
3918 if (st->duration > 0)
3919 st->codecpar->bit_rate = stream_size*8*sc->time_scale/st->duration;
3921 unsigned chunk_samples, total = 0;
3923 if (!sc->chunk_count)
3926 // compute total chunk count
3927 for (i = 0; i < sc->stsc_count; i++) {
3928 unsigned count, chunk_count;
3930 chunk_samples = sc->stsc_data[i].count;
3931 if (i != sc->stsc_count - 1 &&
3932 sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
3933 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
3937 if (sc->samples_per_frame >= 160) { // gsm
3938 count = chunk_samples / sc->samples_per_frame;
3939 } else if (sc->samples_per_frame > 1) {
3940 unsigned samples = (1024/sc->samples_per_frame)*sc->samples_per_frame;
3941 count = (chunk_samples+samples-1) / samples;
3943 count = (chunk_samples+1023) / 1024;
3946 if (mov_stsc_index_valid(i, sc->stsc_count))
3947 chunk_count = sc->stsc_data[i+1].first - sc->stsc_data[i].first;
3949 chunk_count = sc->chunk_count - (sc->stsc_data[i].first - 1);
3950 total += chunk_count * count;
3953 av_log(mov->fc, AV_LOG_TRACE, "chunk count %u\n", total);
3954 if (total >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3956 if (av_reallocp_array(&st->index_entries,
3957 st->nb_index_entries + total,
3958 sizeof(*st->index_entries)) < 0) {
3959 st->nb_index_entries = 0;
3962 st->index_entries_allocated_size = (st->nb_index_entries + total) * sizeof(*st->index_entries);
3965 for (i = 0; i < sc->chunk_count; i++) {
3966 current_offset = sc->chunk_offsets[i];
3967 if (mov_stsc_index_valid(stsc_index, sc->stsc_count) &&
3968 i + 1 == sc->stsc_data[stsc_index + 1].first)
3970 chunk_samples = sc->stsc_data[stsc_index].count;
3972 while (chunk_samples > 0) {
3974 unsigned size, samples;
3976 if (sc->samples_per_frame > 1 && !sc->bytes_per_frame) {
3977 avpriv_request_sample(mov->fc,
3978 "Zero bytes per frame, but %d samples per frame",
3979 sc->samples_per_frame);
3983 if (sc->samples_per_frame >= 160) { // gsm
3984 samples = sc->samples_per_frame;
3985 size = sc->bytes_per_frame;
3987 if (sc->samples_per_frame > 1) {
3988 samples = FFMIN((1024 / sc->samples_per_frame)*
3989 sc->samples_per_frame, chunk_samples);
3990 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
3992 samples = FFMIN(1024, chunk_samples);
3993 size = samples * sc->sample_size;
3997 if (st->nb_index_entries >= total) {
3998 av_log(mov->fc, AV_LOG_ERROR, "wrong chunk count %u\n", total);
4001 if (size > 0x3FFFFFFF) {
4002 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", size);
4005 e = &st->index_entries[st->nb_index_entries++];
4006 e->pos = current_offset;
4007 e->timestamp = current_dts;
4009 e->min_distance = 0;
4010 e->flags = AVINDEX_KEYFRAME;
4011 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, chunk %u, offset %"PRIx64", dts %"PRId64", "
4012 "size %u, duration %u\n", st->index, i, current_offset, current_dts,
4015 current_offset += size;
4016 current_dts += samples;
4017 chunk_samples -= samples;
4022 if (!mov->ignore_editlist && mov->advanced_editlist) {
4023 // Fix index according to edit lists.
4024 mov_fix_index(mov, st);
4027 // Update start time of the stream.
4028 if (st->start_time == AV_NOPTS_VALUE && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries > 0) {
4029 st->start_time = st->index_entries[0].timestamp + sc->dts_shift;
4030 if (sc->ctts_data) {
4031 st->start_time += sc->ctts_data[0].duration;
4035 mov_estimate_video_delay(mov, st);
4038 static int test_same_origin(const char *src, const char *ref) {
4048 av_url_split(src_proto, sizeof(src_proto), src_auth, sizeof(src_auth), src_host, sizeof(src_host), &src_port, NULL, 0, src);
4049 av_url_split(ref_proto, sizeof(ref_proto), ref_auth, sizeof(ref_auth), ref_host, sizeof(ref_host), &ref_port, NULL, 0, ref);
4051 if (strlen(src) == 0) {
4053 } else if (strlen(src_auth) + 1 >= sizeof(src_auth) ||
4054 strlen(ref_auth) + 1 >= sizeof(ref_auth) ||
4055 strlen(src_host) + 1 >= sizeof(src_host) ||
4056 strlen(ref_host) + 1 >= sizeof(ref_host)) {
4058 } else if (strcmp(src_proto, ref_proto) ||
4059 strcmp(src_auth, ref_auth) ||
4060 strcmp(src_host, ref_host) ||
4061 src_port != ref_port) {
4067 static int mov_open_dref(MOVContext *c, AVIOContext **pb, const char *src, MOVDref *ref)
4069 /* try relative path, we do not try the absolute because it can leak information about our
4070 system to an attacker */
4071 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
4072 char filename[1025];
4073 const char *src_path;
4076 /* find a source dir */
4077 src_path = strrchr(src, '/');
4083 /* find a next level down to target */
4084 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
4085 if (ref->path[l] == '/') {
4086 if (i == ref->nlvl_to - 1)
4092 /* compose filename if next level down to target was found */
4093 if (i == ref->nlvl_to - 1 && src_path - src < sizeof(filename)) {
4094 memcpy(filename, src, src_path - src);
4095 filename[src_path - src] = 0;
4097 for (i = 1; i < ref->nlvl_from; i++)
4098 av_strlcat(filename, "../", sizeof(filename));
4100 av_strlcat(filename, ref->path + l + 1, sizeof(filename));
4101 if (!c->use_absolute_path) {
4102 int same_origin = test_same_origin(src, filename);
4105 av_log(c->fc, AV_LOG_ERROR,
4106 "Reference with mismatching origin, %s not tried for security reasons, "
4107 "set demuxer option use_absolute_path to allow it anyway\n",
4109 return AVERROR(ENOENT);
4112 if(strstr(ref->path + l + 1, "..") ||
4113 strstr(ref->path + l + 1, ":") ||
4114 (ref->nlvl_from > 1 && same_origin < 0) ||
4115 (filename[0] == '/' && src_path == src))
4116 return AVERROR(ENOENT);
4119 if (strlen(filename) + 1 == sizeof(filename))
4120 return AVERROR(ENOENT);
4121 if (!c->fc->io_open(c->fc, pb, filename, AVIO_FLAG_READ, NULL))
4124 } else if (c->use_absolute_path) {
4125 av_log(c->fc, AV_LOG_WARNING, "Using absolute path on user request, "
4126 "this is a possible security issue\n");
4127 if (!c->fc->io_open(c->fc, pb, ref->path, AVIO_FLAG_READ, NULL))
4130 av_log(c->fc, AV_LOG_ERROR,
4131 "Absolute path %s not tried for security reasons, "
4132 "set demuxer option use_absolute_path to allow absolute paths\n",
4136 return AVERROR(ENOENT);
4139 static void fix_timescale(MOVContext *c, MOVStreamContext *sc)
4141 if (sc->time_scale <= 0) {
4142 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", sc->ffindex);
4143 sc->time_scale = c->time_scale;
4144 if (sc->time_scale <= 0)
4149 static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4152 MOVStreamContext *sc;
4155 st = avformat_new_stream(c->fc, NULL);
4156 if (!st) return AVERROR(ENOMEM);
4157 st->id = c->fc->nb_streams;
4158 sc = av_mallocz(sizeof(MOVStreamContext));
4159 if (!sc) return AVERROR(ENOMEM);
4162 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
4163 sc->ffindex = st->index;
4164 c->trak_index = st->index;
4166 if ((ret = mov_read_default(c, pb, atom)) < 0)
4172 if ((sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
4173 (!sc->sample_size && !sc->sample_count))) ||
4174 (!sc->chunk_count && sc->sample_count)) {
4175 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
4179 if (sc->chunk_count && sc->stsc_count && sc->stsc_data[ sc->stsc_count - 1 ].first > sc->chunk_count) {
4180 av_log(c->fc, AV_LOG_ERROR, "stream %d, contradictionary STSC and STCO\n",
4182 return AVERROR_INVALIDDATA;
4185 fix_timescale(c, sc);
4187 avpriv_set_pts_info(st, 64, 1, sc->time_scale);
4189 mov_build_index(c, st);
4191 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
4192 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
4193 if (c->enable_drefs) {
4194 if (mov_open_dref(c, &sc->pb, c->fc->url, dref) < 0)
4195 av_log(c->fc, AV_LOG_ERROR,
4196 "stream %d, error opening alias: path='%s', dir='%s', "
4197 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
4198 st->index, dref->path, dref->dir, dref->filename,
4199 dref->volume, dref->nlvl_from, dref->nlvl_to);
4201 av_log(c->fc, AV_LOG_WARNING,
4202 "Skipped opening external track: "
4203 "stream %d, alias: path='%s', dir='%s', "
4204 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d."
4205 "Set enable_drefs to allow this.\n",
4206 st->index, dref->path, dref->dir, dref->filename,
4207 dref->volume, dref->nlvl_from, dref->nlvl_to);
4211 sc->pb_is_copied = 1;
4214 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
4215 if (!st->sample_aspect_ratio.num && st->codecpar->width && st->codecpar->height &&
4216 sc->height && sc->width &&
4217 (st->codecpar->width != sc->width || st->codecpar->height != sc->height)) {
4218 st->sample_aspect_ratio = av_d2q(((double)st->codecpar->height * sc->width) /
4219 ((double)st->codecpar->width * sc->height), INT_MAX);
4222 #if FF_API_R_FRAME_RATE
4223 if (sc->stts_count == 1 || (sc->stts_count == 2 && sc->stts_data[1].count == 1))
4224 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
4225 sc->time_scale, sc->stts_data[0].duration, INT_MAX);
4229 // done for ai5q, ai52, ai55, ai1q, ai12 and ai15.
4230 if (!st->codecpar->extradata_size && st->codecpar->codec_id == AV_CODEC_ID_H264 &&
4231 TAG_IS_AVCI(st->codecpar->codec_tag)) {
4232 ret = ff_generate_avci_extradata(st);
4237 switch (st->codecpar->codec_id) {
4238 #if CONFIG_H261_DECODER
4239 case AV_CODEC_ID_H261:
4241 #if CONFIG_H263_DECODER
4242 case AV_CODEC_ID_H263:
4244 #if CONFIG_MPEG4_DECODER
4245 case AV_CODEC_ID_MPEG4:
4247 st->codecpar->width = 0; /* let decoder init width/height */
4248 st->codecpar->height= 0;
4252 // If the duration of the mp3 packets is not constant, then they could need a parser
4253 if (st->codecpar->codec_id == AV_CODEC_ID_MP3
4254 && sc->stts_count > 3
4255 && sc->stts_count*10 > st->nb_frames
4256 && sc->time_scale == st->codecpar->sample_rate) {
4257 st->need_parsing = AVSTREAM_PARSE_FULL;
4259 /* Do not need those anymore. */
4260 av_freep(&sc->chunk_offsets);
4261 av_freep(&sc->sample_sizes);
4262 av_freep(&sc->keyframes);
4263 av_freep(&sc->stts_data);
4264 av_freep(&sc->stps_data);
4265 av_freep(&sc->elst_data);
4266 av_freep(&sc->rap_group);
4271 static int mov_read_ilst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4274 c->itunes_metadata = 1;
4275 ret = mov_read_default(c, pb, atom);
4276 c->itunes_metadata = 0;
4280 static int mov_read_keys(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4289 count = avio_rb32(pb);
4290 if (count > UINT_MAX / sizeof(*c->meta_keys) - 1) {
4291 av_log(c->fc, AV_LOG_ERROR,
4292 "The 'keys' atom with the invalid key count: %"PRIu32"\n", count);
4293 return AVERROR_INVALIDDATA;
4296 c->meta_keys_count = count + 1;
4297 c->meta_keys = av_mallocz(c->meta_keys_count * sizeof(*c->meta_keys));
4299 return AVERROR(ENOMEM);
4301 for (i = 1; i <= count; ++i) {
4302 uint32_t key_size = avio_rb32(pb);
4303 uint32_t type = avio_rl32(pb);
4305 av_log(c->fc, AV_LOG_ERROR,
4306 "The key# %"PRIu32" in meta has invalid size:"
4307 "%"PRIu32"\n", i, key_size);
4308 return AVERROR_INVALIDDATA;
4311 if (type != MKTAG('m','d','t','a')) {
4312 avio_skip(pb, key_size);
4314 c->meta_keys[i] = av_mallocz(key_size + 1);
4315 if (!c->meta_keys[i])
4316 return AVERROR(ENOMEM);
4317 avio_read(pb, c->meta_keys[i], key_size);
4323 static int mov_read_custom(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4325 int64_t end = avio_tell(pb) + atom.size;
4326 uint8_t *key = NULL, *val = NULL, *mean = NULL;
4330 MOVStreamContext *sc;
4332 if (c->fc->nb_streams < 1)
4334 st = c->fc->streams[c->fc->nb_streams-1];
4337 for (i = 0; i < 3; i++) {
4341 if (end - avio_tell(pb) <= 12)
4344 len = avio_rb32(pb);
4345 tag = avio_rl32(pb);
4346 avio_skip(pb, 4); // flags
4348 if (len < 12 || len - 12 > end - avio_tell(pb))
4352 if (tag == MKTAG('m', 'e', 'a', 'n'))
4354 else if (tag == MKTAG('n', 'a', 'm', 'e'))
4356 else if (tag == MKTAG('d', 'a', 't', 'a') && len > 4) {
4363 *p = av_malloc(len + 1);
4365 ret = AVERROR(ENOMEM);
4368 ret = ffio_read_size(pb, *p, len);
4376 if (mean && key && val) {
4377 if (strcmp(key, "iTunSMPB") == 0) {
4378 int priming, remainder, samples;
4379 if(sscanf(val, "%*X %X %X %X", &priming, &remainder, &samples) == 3){
4380 if(priming>0 && priming<16384)
4381 sc->start_pad = priming;
4384 if (strcmp(key, "cdec") != 0) {
4385 av_dict_set(&c->fc->metadata, key, val,
4386 AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL);
4390 av_log(c->fc, AV_LOG_VERBOSE,
4391 "Unhandled or malformed custom metadata of size %"PRId64"\n", atom.size);
4394 avio_seek(pb, end, SEEK_SET);
4401 static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4403 while (atom.size > 8) {
4404 uint32_t tag = avio_rl32(pb);
4406 if (tag == MKTAG('h','d','l','r')) {
4407 avio_seek(pb, -8, SEEK_CUR);
4409 return mov_read_default(c, pb, atom);
4415 // return 1 when matrix is identity, 0 otherwise
4416 #define IS_MATRIX_IDENT(matrix) \
4417 ( (matrix)[0][0] == (1 << 16) && \
4418 (matrix)[1][1] == (1 << 16) && \
4419 (matrix)[2][2] == (1 << 30) && \
4420 !(matrix)[0][1] && !(matrix)[0][2] && \
4421 !(matrix)[1][0] && !(matrix)[1][2] && \
4422 !(matrix)[2][0] && !(matrix)[2][1])
4424 static int mov_read_tkhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4429 int display_matrix[3][3];
4430 int res_display_matrix[3][3] = { { 0 } };
4432 MOVStreamContext *sc;
4436 if (c->fc->nb_streams < 1)
4438 st = c->fc->streams[c->fc->nb_streams-1];
4441 version = avio_r8(pb);
4442 flags = avio_rb24(pb);
4443 st->disposition |= (flags & MOV_TKHD_FLAG_ENABLED) ? AV_DISPOSITION_DEFAULT : 0;
4449 avio_rb32(pb); /* creation time */
4450 avio_rb32(pb); /* modification time */
4452 st->id = (int)avio_rb32(pb); /* track id (NOT 0 !)*/
4453 avio_rb32(pb); /* reserved */
4455 /* highlevel (considering edits) duration in movie timebase */
4456 (version == 1) ? avio_rb64(pb) : avio_rb32(pb);
4457 avio_rb32(pb); /* reserved */
4458 avio_rb32(pb); /* reserved */
4460 avio_rb16(pb); /* layer */
4461 avio_rb16(pb); /* alternate group */
4462 avio_rb16(pb); /* volume */
4463 avio_rb16(pb); /* reserved */
4465 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
4466 // they're kept in fixed point format through all calculations
4467 // save u,v,z to store the whole matrix in the AV_PKT_DATA_DISPLAYMATRIX
4468 // side data, but the scale factor is not needed to calculate aspect ratio
4469 for (i = 0; i < 3; i++) {
4470 display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
4471 display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
4472 display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
4475 width = avio_rb32(pb); // 16.16 fixed point track width
4476 height = avio_rb32(pb); // 16.16 fixed point track height
4477 sc->width = width >> 16;
4478 sc->height = height >> 16;
4480 // apply the moov display matrix (after the tkhd one)
4481 for (i = 0; i < 3; i++) {
4482 const int sh[3] = { 16, 16, 30 };
4483 for (j = 0; j < 3; j++) {
4484 for (e = 0; e < 3; e++) {
4485 res_display_matrix[i][j] +=
4486 ((int64_t) display_matrix[i][e] *
4487 c->movie_display_matrix[e][j]) >> sh[e];
4492 // save the matrix when it is not the default identity
4493 if (!IS_MATRIX_IDENT(res_display_matrix)) {
4496 av_freep(&sc->display_matrix);
4497 sc->display_matrix = av_malloc(sizeof(int32_t) * 9);
4498 if (!sc->display_matrix)
4499 return AVERROR(ENOMEM);
4501 for (i = 0; i < 3; i++)
4502 for (j = 0; j < 3; j++)
4503 sc->display_matrix[i * 3 + j] = res_display_matrix[i][j];
4505 #if FF_API_OLD_ROTATE_API
4506 rotate = av_display_rotation_get(sc->display_matrix);
4507 if (!isnan(rotate)) {
4508 char rotate_buf[64];
4510 if (rotate < 0) // for backward compatibility
4512 snprintf(rotate_buf, sizeof(rotate_buf), "%g", rotate);
4513 av_dict_set(&st->metadata, "rotate", rotate_buf, 0);
4518 // transform the display width/height according to the matrix
4519 // to keep the same scale, use [width height 1<<16]
4520 if (width && height && sc->display_matrix) {
4521 double disp_transform[2];
4523 for (i = 0; i < 2; i++)
4524 disp_transform[i] = hypot(sc->display_matrix[0 + i],
4525 sc->display_matrix[3 + i]);
4527 if (disp_transform[0] > 0 && disp_transform[1] > 0 &&
4528 disp_transform[0] < (1<<24) && disp_transform[1] < (1<<24) &&
4529 fabs((disp_transform[0] / disp_transform[1]) - 1.0) > 0.01)
4530 st->sample_aspect_ratio = av_d2q(
4531 disp_transform[0] / disp_transform[1],
4537 static int mov_read_tfhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4539 MOVFragment *frag = &c->fragment;
4540 MOVTrackExt *trex = NULL;
4541 int flags, track_id, i;
4543 avio_r8(pb); /* version */
4544 flags = avio_rb24(pb);
4546 track_id = avio_rb32(pb);
4548 return AVERROR_INVALIDDATA;
4549 frag->track_id = track_id;
4550 set_frag_stream(&c->frag_index, track_id);
4551 for (i = 0; i < c->trex_count; i++)
4552 if (c->trex_data[i].track_id == frag->track_id) {
4553 trex = &c->trex_data[i];
4557 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
4558 return AVERROR_INVALIDDATA;
4561 frag->base_data_offset = flags & MOV_TFHD_BASE_DATA_OFFSET ?
4562 avio_rb64(pb) : flags & MOV_TFHD_DEFAULT_BASE_IS_MOOF ?
4563 frag->moof_offset : frag->implicit_offset;
4564 frag->stsd_id = flags & MOV_TFHD_STSD_ID ? avio_rb32(pb) : trex->stsd_id;
4566 frag->duration = flags & MOV_TFHD_DEFAULT_DURATION ?
4567 avio_rb32(pb) : trex->duration;
4568 frag->size = flags & MOV_TFHD_DEFAULT_SIZE ?
4569 avio_rb32(pb) : trex->size;
4570 frag->flags = flags & MOV_TFHD_DEFAULT_FLAGS ?
4571 avio_rb32(pb) : trex->flags;
4572 av_log(c->fc, AV_LOG_TRACE, "frag flags 0x%x\n", frag->flags);
4577 static int mov_read_chap(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4582 num = atom.size / 4;
4583 if (!(new_tracks = av_malloc_array(num, sizeof(int))))
4584 return AVERROR(ENOMEM);
4586 av_free(c->chapter_tracks);
4587 c->chapter_tracks = new_tracks;
4588 c->nb_chapter_tracks = num;
4590 for (i = 0; i < num && !pb->eof_reached; i++)
4591 c->chapter_tracks[i] = avio_rb32(pb);
4596 static int mov_read_trex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4601 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
4602 return AVERROR_INVALIDDATA;
4603 if ((err = av_reallocp_array(&c->trex_data, c->trex_count + 1,
4604 sizeof(*c->trex_data))) < 0) {
4609 c->fc->duration = AV_NOPTS_VALUE; // the duration from mvhd is not representing the whole file when fragments are used.
4611 trex = &c->trex_data[c->trex_count++];
4612 avio_r8(pb); /* version */
4613 avio_rb24(pb); /* flags */
4614 trex->track_id = avio_rb32(pb);
4615 trex->stsd_id = avio_rb32(pb);
4616 trex->duration = avio_rb32(pb);
4617 trex->size = avio_rb32(pb);
4618 trex->flags = avio_rb32(pb);
4622 static int mov_read_tfdt(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4624 MOVFragment *frag = &c->fragment;
4625 AVStream *st = NULL;
4626 MOVStreamContext *sc;
4628 MOVFragmentStreamInfo * frag_stream_info;
4629 int64_t base_media_decode_time;
4631 for (i = 0; i < c->fc->nb_streams; i++) {
4632 if (c->fc->streams[i]->id == frag->track_id) {
4633 st = c->fc->streams[i];
4638 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %u\n", frag->track_id);
4639 return AVERROR_INVALIDDATA;
4642 if (sc->pseudo_stream_id + 1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4644 version = avio_r8(pb);
4645 avio_rb24(pb); /* flags */
4647 base_media_decode_time = avio_rb64(pb);
4649 base_media_decode_time = avio_rb32(pb);
4652 frag_stream_info = get_current_frag_stream_info(&c->frag_index);
4653 if (frag_stream_info)
4654 frag_stream_info->tfdt_dts = base_media_decode_time;
4655 sc->track_end = base_media_decode_time;
4660 static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4662 MOVFragment *frag = &c->fragment;
4663 AVStream *st = NULL;
4664 MOVStreamContext *sc;
4667 int64_t dts, pts = AV_NOPTS_VALUE;
4668 int data_offset = 0;
4669 unsigned entries, first_sample_flags = frag->flags;
4670 int flags, distance, i;
4671 int64_t prev_dts = AV_NOPTS_VALUE;
4672 int next_frag_index = -1, index_entry_pos;
4673 size_t requested_size;
4674 size_t old_ctts_allocated_size;
4675 AVIndexEntry *new_entries;
4676 MOVFragmentStreamInfo * frag_stream_info;
4678 for (i = 0; i < c->fc->nb_streams; i++) {
4679 if (c->fc->streams[i]->id == frag->track_id) {
4680 st = c->fc->streams[i];
4685 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %u\n", frag->track_id);
4686 return AVERROR_INVALIDDATA;
4689 if (sc->pseudo_stream_id+1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4692 // Find the next frag_index index that has a valid index_entry for
4693 // the current track_id.
4695 // A valid index_entry means the trun for the fragment was read
4696 // and it's samples are in index_entries at the given position.
4697 // New index entries will be inserted before the index_entry found.
4698 index_entry_pos = st->nb_index_entries;
4699 for (i = c->frag_index.current + 1; i < c->frag_index.nb_items; i++) {
4700 frag_stream_info = get_frag_stream_info(&c->frag_index, i, frag->track_id);
4701 if (frag_stream_info && frag_stream_info->index_entry >= 0) {
4702 next_frag_index = i;
4703 index_entry_pos = frag_stream_info->index_entry;
4708 avio_r8(pb); /* version */
4709 flags = avio_rb24(pb);
4710 entries = avio_rb32(pb);
4711 av_log(c->fc, AV_LOG_TRACE, "flags 0x%x entries %u\n", flags, entries);
4713 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
4714 return AVERROR_INVALIDDATA;
4715 if (flags & MOV_TRUN_DATA_OFFSET) data_offset = avio_rb32(pb);
4716 if (flags & MOV_TRUN_FIRST_SAMPLE_FLAGS) first_sample_flags = avio_rb32(pb);
4718 frag_stream_info = get_current_frag_stream_info(&c->frag_index);
4719 if (frag_stream_info)
4721 if (frag_stream_info->first_tfra_pts != AV_NOPTS_VALUE &&
4722 c->use_mfra_for == FF_MOV_FLAG_MFRA_PTS) {
4723 pts = frag_stream_info->first_tfra_pts;
4724 av_log(c->fc, AV_LOG_DEBUG, "found mfra time %"PRId64
4725 ", using it for pts\n", pts);
4726 } else if (frag_stream_info->sidx_pts != AV_NOPTS_VALUE) {
4727 // FIXME: sidx earliest_presentation_time is *PTS*, s.b.
4728 // pts = frag_stream_info->sidx_pts;
4729 dts = frag_stream_info->sidx_pts - sc->time_offset;
4730 av_log(c->fc, AV_LOG_DEBUG, "found sidx time %"PRId64
4731 ", using it for pts\n", pts);
4732 } else if (frag_stream_info->tfdt_dts != AV_NOPTS_VALUE) {
4733 dts = frag_stream_info->tfdt_dts - sc->time_offset;
4734 av_log(c->fc, AV_LOG_DEBUG, "found tfdt time %"PRId64
4735 ", using it for dts\n", dts);
4737 dts = sc->track_end - sc->time_offset;
4738 av_log(c->fc, AV_LOG_DEBUG, "found track end time %"PRId64
4739 ", using it for dts\n", dts);
4742 dts = sc->track_end - sc->time_offset;
4743 av_log(c->fc, AV_LOG_DEBUG, "found track end time %"PRId64
4744 ", using it for dts\n", dts);
4746 offset = frag->base_data_offset + data_offset;
4748 av_log(c->fc, AV_LOG_TRACE, "first sample flags 0x%x\n", first_sample_flags);
4750 // realloc space for new index entries
4751 if((unsigned)st->nb_index_entries + entries >= UINT_MAX / sizeof(AVIndexEntry)) {
4752 entries = UINT_MAX / sizeof(AVIndexEntry) - st->nb_index_entries;
4753 av_log(c->fc, AV_LOG_ERROR, "Failed to add index entry\n");
4758 requested_size = (st->nb_index_entries + entries) * sizeof(AVIndexEntry);
4759 new_entries = av_fast_realloc(st->index_entries,
4760 &st->index_entries_allocated_size,
4763 return AVERROR(ENOMEM);
4764 st->index_entries= new_entries;
4766 requested_size = (st->nb_index_entries + entries) * sizeof(*sc->ctts_data);
4767 old_ctts_allocated_size = sc->ctts_allocated_size;
4768 ctts_data = av_fast_realloc(sc->ctts_data, &sc->ctts_allocated_size,
4771 return AVERROR(ENOMEM);
4772 sc->ctts_data = ctts_data;
4774 // In case there were samples without ctts entries, ensure they get
4775 // zero valued entries. This ensures clips which mix boxes with and
4776 // without ctts entries don't pickup uninitialized data.
4777 memset((uint8_t*)(sc->ctts_data) + old_ctts_allocated_size, 0,
4778 sc->ctts_allocated_size - old_ctts_allocated_size);
4780 if (index_entry_pos < st->nb_index_entries) {
4781 // Make hole in index_entries and ctts_data for new samples
4782 memmove(st->index_entries + index_entry_pos + entries,
4783 st->index_entries + index_entry_pos,
4784 sizeof(*st->index_entries) *
4785 (st->nb_index_entries - index_entry_pos));
4786 memmove(sc->ctts_data + index_entry_pos + entries,
4787 sc->ctts_data + index_entry_pos,
4788 sizeof(*sc->ctts_data) * (sc->ctts_count - index_entry_pos));
4789 if (index_entry_pos < sc->current_sample) {
4790 sc->current_sample += entries;
4794 st->nb_index_entries += entries;
4795 sc->ctts_count = st->nb_index_entries;
4797 // Record the index_entry position in frag_index of this fragment
4798 if (frag_stream_info)
4799 frag_stream_info->index_entry = index_entry_pos;
4801 if (index_entry_pos > 0)
4802 prev_dts = st->index_entries[index_entry_pos-1].timestamp;
4804 for (i = 0; i < entries && !pb->eof_reached; i++) {
4805 unsigned sample_size = frag->size;
4806 int sample_flags = i ? frag->flags : first_sample_flags;
4807 unsigned sample_duration = frag->duration;
4808 unsigned ctts_duration = 0;
4810 int index_entry_flags = 0;
4812 if (flags & MOV_TRUN_SAMPLE_DURATION) sample_duration = avio_rb32(pb);
4813 if (flags & MOV_TRUN_SAMPLE_SIZE) sample_size = avio_rb32(pb);
4814 if (flags & MOV_TRUN_SAMPLE_FLAGS) sample_flags = avio_rb32(pb);
4815 if (flags & MOV_TRUN_SAMPLE_CTS) ctts_duration = avio_rb32(pb);
4817 mov_update_dts_shift(sc, ctts_duration);
4818 if (pts != AV_NOPTS_VALUE) {
4819 dts = pts - sc->dts_shift;
4820 if (flags & MOV_TRUN_SAMPLE_CTS) {
4821 dts -= ctts_duration;
4823 dts -= sc->time_offset;
4825 av_log(c->fc, AV_LOG_DEBUG,
4826 "pts %"PRId64" calculated dts %"PRId64
4827 " sc->dts_shift %d ctts.duration %d"
4828 " sc->time_offset %"PRId64
4829 " flags & MOV_TRUN_SAMPLE_CTS %d\n",
4831 sc->dts_shift, ctts_duration,
4832 sc->time_offset, flags & MOV_TRUN_SAMPLE_CTS);
4833 pts = AV_NOPTS_VALUE;
4836 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
4840 !(sample_flags & (MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC |
4841 MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES));
4844 index_entry_flags |= AVINDEX_KEYFRAME;
4846 // Fragments can overlap in time. Discard overlapping frames after
4848 if (prev_dts >= dts)
4849 index_entry_flags |= AVINDEX_DISCARD_FRAME;
4851 st->index_entries[index_entry_pos].pos = offset;
4852 st->index_entries[index_entry_pos].timestamp = dts;
4853 st->index_entries[index_entry_pos].size= sample_size;
4854 st->index_entries[index_entry_pos].min_distance= distance;
4855 st->index_entries[index_entry_pos].flags = index_entry_flags;
4857 sc->ctts_data[index_entry_pos].count = 1;
4858 sc->ctts_data[index_entry_pos].duration = ctts_duration;
4861 av_log(c->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
4862 "size %u, distance %d, keyframe %d\n", st->index,
4863 index_entry_pos, offset, dts, sample_size, distance, keyframe);
4865 dts += sample_duration;
4866 offset += sample_size;
4867 sc->data_size += sample_size;
4869 if (sample_duration <= INT64_MAX - sc->duration_for_fps &&
4870 1 <= INT64_MAX - sc->nb_frames_for_fps
4872 sc->duration_for_fps += sample_duration;
4873 sc->nb_frames_for_fps ++;
4877 // EOF found before reading all entries. Fix the hole this would
4878 // leave in index_entries and ctts_data
4879 int gap = entries - i;
4880 memmove(st->index_entries + index_entry_pos,
4881 st->index_entries + index_entry_pos + gap,
4882 sizeof(*st->index_entries) *
4883 (st->nb_index_entries - (index_entry_pos + gap)));
4884 memmove(sc->ctts_data + index_entry_pos,
4885 sc->ctts_data + index_entry_pos + gap,
4886 sizeof(*sc->ctts_data) *
4887 (sc->ctts_count - (index_entry_pos + gap)));
4889 st->nb_index_entries -= gap;
4890 sc->ctts_count -= gap;
4891 if (index_entry_pos < sc->current_sample) {
4892 sc->current_sample -= gap;
4897 // The end of this new fragment may overlap in time with the start
4898 // of the next fragment in index_entries. Mark the samples in the next
4899 // fragment that overlap with AVINDEX_DISCARD_FRAME
4900 prev_dts = AV_NOPTS_VALUE;
4901 if (index_entry_pos > 0)
4902 prev_dts = st->index_entries[index_entry_pos-1].timestamp;
4903 for (i = index_entry_pos; i < st->nb_index_entries; i++) {
4904 if (prev_dts < st->index_entries[i].timestamp)
4906 st->index_entries[i].flags |= AVINDEX_DISCARD_FRAME;
4909 // If a hole was created to insert the new index_entries into,
4910 // the index_entry recorded for all subsequent moof must
4911 // be incremented by the number of entries inserted.
4912 fix_frag_index_entries(&c->frag_index, next_frag_index,
4913 frag->track_id, entries);
4915 if (pb->eof_reached) {
4916 av_log(c->fc, AV_LOG_WARNING, "reached eof, corrupted TRUN atom\n");
4920 frag->implicit_offset = offset;
4922 sc->track_end = dts + sc->time_offset;
4923 if (st->duration < sc->track_end)
4924 st->duration = sc->track_end;
4929 static int mov_read_sidx(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4931 int64_t offset = avio_tell(pb) + atom.size, pts, timestamp;
4933 unsigned i, j, track_id, item_count;
4934 AVStream *st = NULL;
4935 AVStream *ref_st = NULL;
4936 MOVStreamContext *sc, *ref_sc = NULL;
4937 AVRational timescale;
4939 version = avio_r8(pb);
4941 avpriv_request_sample(c->fc, "sidx version %u", version);
4945 avio_rb24(pb); // flags
4947 track_id = avio_rb32(pb); // Reference ID
4948 for (i = 0; i < c->fc->nb_streams; i++) {
4949 if (c->fc->streams[i]->id == track_id) {
4950 st = c->fc->streams[i];
4955 av_log(c->fc, AV_LOG_WARNING, "could not find corresponding track id %d\n", track_id);
4961 timescale = av_make_q(1, avio_rb32(pb));
4963 if (timescale.den <= 0) {
4964 av_log(c->fc, AV_LOG_ERROR, "Invalid sidx timescale 1/%d\n", timescale.den);
4965 return AVERROR_INVALIDDATA;
4969 pts = avio_rb32(pb);
4970 offset += avio_rb32(pb);
4972 pts = avio_rb64(pb);
4973 offset += avio_rb64(pb);
4976 avio_rb16(pb); // reserved
4978 item_count = avio_rb16(pb);
4980 for (i = 0; i < item_count; i++) {
4982 MOVFragmentStreamInfo * frag_stream_info;
4983 uint32_t size = avio_rb32(pb);
4984 uint32_t duration = avio_rb32(pb);
4985 if (size & 0x80000000) {
4986 avpriv_request_sample(c->fc, "sidx reference_type 1");
4987 return AVERROR_PATCHWELCOME;
4989 avio_rb32(pb); // sap_flags
4990 timestamp = av_rescale_q(pts, st->time_base, timescale);
4992 index = update_frag_index(c, offset);
4993 frag_stream_info = get_frag_stream_info(&c->frag_index, index, track_id);
4994 if (frag_stream_info)
4995 frag_stream_info->sidx_pts = timestamp;
5001 st->duration = sc->track_end = pts;
5005 if (offset == avio_size(pb)) {
5006 // Find first entry in fragment index that came from an sidx.
5007 // This will pretty much always be the first entry.
5008 for (i = 0; i < c->frag_index.nb_items; i++) {
5009 MOVFragmentIndexItem * item = &c->frag_index.item[i];
5010 for (j = 0; ref_st == NULL && j < item->nb_stream_info; j++) {
5011 MOVFragmentStreamInfo * si;
5012 si = &item->stream_info[j];
5013 if (si->sidx_pts != AV_NOPTS_VALUE) {
5014 ref_st = c->fc->streams[j];
5015 ref_sc = ref_st->priv_data;
5020 for (i = 0; i < c->fc->nb_streams; i++) {
5021 st = c->fc->streams[i];
5023 if (!sc->has_sidx) {
5024 st->duration = sc->track_end = av_rescale(ref_st->duration, sc->time_scale, ref_sc->time_scale);
5028 c->frag_index.complete = 1;
5034 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
5035 /* like the files created with Adobe Premiere 5.0, for samples see */
5036 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
5037 static int mov_read_wide(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5042 return 0; /* continue */
5043 if (avio_rb32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
5044 avio_skip(pb, atom.size - 4);
5047 atom.type = avio_rl32(pb);
5049 if (atom.type != MKTAG('m','d','a','t')) {
5050 avio_skip(pb, atom.size);
5053 err = mov_read_mdat(c, pb, atom);
5057 static int mov_read_cmov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5062 uint8_t *moov_data; /* uncompressed data */
5063 long cmov_len, moov_len;
5066 avio_rb32(pb); /* dcom atom */
5067 if (avio_rl32(pb) != MKTAG('d','c','o','m'))
5068 return AVERROR_INVALIDDATA;
5069 if (avio_rl32(pb) != MKTAG('z','l','i','b')) {
5070 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !\n");
5071 return AVERROR_INVALIDDATA;
5073 avio_rb32(pb); /* cmvd atom */
5074 if (avio_rl32(pb) != MKTAG('c','m','v','d'))
5075 return AVERROR_INVALIDDATA;
5076 moov_len = avio_rb32(pb); /* uncompressed size */
5077 cmov_len = atom.size - 6 * 4;
5079 cmov_data = av_malloc(cmov_len);
5081 return AVERROR(ENOMEM);
5082 moov_data = av_malloc(moov_len);
5085 return AVERROR(ENOMEM);
5087 ret = ffio_read_size(pb, cmov_data, cmov_len);
5089 goto free_and_return;
5091 ret = AVERROR_INVALIDDATA;
5092 if (uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
5093 goto free_and_return;
5094 if (ffio_init_context(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
5095 goto free_and_return;
5096 ctx.seekable = AVIO_SEEKABLE_NORMAL;
5097 atom.type = MKTAG('m','o','o','v');
5098 atom.size = moov_len;
5099 ret = mov_read_default(c, &ctx, atom);
5105 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
5106 return AVERROR(ENOSYS);
5110 /* edit list atom */
5111 static int mov_read_elst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5113 MOVStreamContext *sc;
5114 int i, edit_count, version;
5115 int64_t elst_entry_size;
5117 if (c->fc->nb_streams < 1 || c->ignore_editlist)
5119 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
5121 version = avio_r8(pb); /* version */
5122 avio_rb24(pb); /* flags */
5123 edit_count = avio_rb32(pb); /* entries */
5126 elst_entry_size = version == 1 ? 20 : 12;
5127 if (atom.size != edit_count * elst_entry_size) {
5128 if (c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
5129 av_log(c->fc, AV_LOG_ERROR, "Invalid edit list entry_count: %d for elst atom of size: %"PRId64" bytes.\n",
5130 edit_count, atom.size + 8);
5131 return AVERROR_INVALIDDATA;
5133 edit_count = atom.size / elst_entry_size;
5134 if (edit_count * elst_entry_size != atom.size) {
5135 av_log(c->fc, AV_LOG_WARNING, "ELST atom of %"PRId64" bytes, bigger than %d entries.", atom.size, edit_count);
5143 av_log(c->fc, AV_LOG_WARNING, "Duplicated ELST atom\n");
5144 av_free(sc->elst_data);
5146 sc->elst_data = av_malloc_array(edit_count, sizeof(*sc->elst_data));
5148 return AVERROR(ENOMEM);
5150 av_log(c->fc, AV_LOG_TRACE, "track[%u].edit_count = %i\n", c->fc->nb_streams - 1, edit_count);
5151 for (i = 0; i < edit_count && atom.size > 0 && !pb->eof_reached; i++) {
5152 MOVElst *e = &sc->elst_data[i];
5155 e->duration = avio_rb64(pb);
5156 e->time = avio_rb64(pb);
5159 e->duration = avio_rb32(pb); /* segment duration */
5160 e->time = (int32_t)avio_rb32(pb); /* media time */
5163 e->rate = avio_rb32(pb) / 65536.0;
5165 av_log(c->fc, AV_LOG_TRACE, "duration=%"PRId64" time=%"PRId64" rate=%f\n",
5166 e->duration, e->time, e->rate);
5168 if (e->time < 0 && e->time != -1 &&
5169 c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
5170 av_log(c->fc, AV_LOG_ERROR, "Track %d, edit %d: Invalid edit list media time=%"PRId64"\n",
5171 c->fc->nb_streams-1, i, e->time);
5172 return AVERROR_INVALIDDATA;
5180 static int mov_read_tmcd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5182 MOVStreamContext *sc;
5184 if (c->fc->nb_streams < 1)
5185 return AVERROR_INVALIDDATA;
5186 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5187 sc->timecode_track = avio_rb32(pb);
5191 static int mov_read_av1c(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5196 if (c->fc->nb_streams < 1)
5198 st = c->fc->streams[c->fc->nb_streams - 1];
5200 if (atom.size < 4) {
5201 av_log(c->fc, AV_LOG_ERROR, "Empty AV1 Codec Configuration Box\n");
5202 return AVERROR_INVALIDDATA;
5205 /* For now, propagate only the OBUs, if any. Once libavcodec is
5206 updated to handle isobmff style extradata this can be removed. */
5212 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 4);
5219 static int mov_read_vpcc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5222 int version, color_range, color_primaries, color_trc, color_space;
5224 if (c->fc->nb_streams < 1)
5226 st = c->fc->streams[c->fc->nb_streams - 1];
5228 if (atom.size < 5) {
5229 av_log(c->fc, AV_LOG_ERROR, "Empty VP Codec Configuration box\n");
5230 return AVERROR_INVALIDDATA;
5233 version = avio_r8(pb);
5235 av_log(c->fc, AV_LOG_WARNING, "Unsupported VP Codec Configuration box version %d\n", version);
5238 avio_skip(pb, 3); /* flags */
5240 avio_skip(pb, 2); /* profile + level */
5241 color_range = avio_r8(pb); /* bitDepth, chromaSubsampling, videoFullRangeFlag */
5242 color_primaries = avio_r8(pb);
5243 color_trc = avio_r8(pb);
5244 color_space = avio_r8(pb);
5245 if (avio_rb16(pb)) /* codecIntializationDataSize */
5246 return AVERROR_INVALIDDATA;
5248 if (!av_color_primaries_name(color_primaries))
5249 color_primaries = AVCOL_PRI_UNSPECIFIED;
5250 if (!av_color_transfer_name(color_trc))
5251 color_trc = AVCOL_TRC_UNSPECIFIED;
5252 if (!av_color_space_name(color_space))
5253 color_space = AVCOL_SPC_UNSPECIFIED;
5255 st->codecpar->color_range = (color_range & 1) ? AVCOL_RANGE_JPEG : AVCOL_RANGE_MPEG;
5256 st->codecpar->color_primaries = color_primaries;
5257 st->codecpar->color_trc = color_trc;
5258 st->codecpar->color_space = color_space;
5263 static int mov_read_smdm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5265 MOVStreamContext *sc;
5266 const int chroma_den = 50000;
5267 const int luma_den = 10000;
5270 if (c->fc->nb_streams < 1)
5271 return AVERROR_INVALIDDATA;
5273 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5275 if (atom.size < 5) {
5276 av_log(c->fc, AV_LOG_ERROR, "Empty Mastering Display Metadata box\n");
5277 return AVERROR_INVALIDDATA;
5280 version = avio_r8(pb);
5282 av_log(c->fc, AV_LOG_WARNING, "Unsupported Mastering Display Metadata box version %d\n", version);
5285 avio_skip(pb, 3); /* flags */
5287 sc->mastering = av_mastering_display_metadata_alloc();
5289 return AVERROR(ENOMEM);
5291 for (i = 0; i < 3; i++)
5292 for (j = 0; j < 2; j++)
5293 sc->mastering->display_primaries[i][j] =
5294 av_make_q(lrint(((double)avio_rb16(pb) / (1 << 16)) * chroma_den), chroma_den);
5295 for (i = 0; i < 2; i++)
5296 sc->mastering->white_point[i] =
5297 av_make_q(lrint(((double)avio_rb16(pb) / (1 << 16)) * chroma_den), chroma_den);
5298 sc->mastering->max_luminance =
5299 av_make_q(lrint(((double)avio_rb32(pb) / (1 << 8)) * luma_den), luma_den);
5300 sc->mastering->min_luminance =
5301 av_make_q(lrint(((double)avio_rb32(pb) / (1 << 14)) * luma_den), luma_den);
5303 sc->mastering->has_primaries = 1;
5304 sc->mastering->has_luminance = 1;
5309 static int mov_read_mdcv(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5311 MOVStreamContext *sc;
5312 const int mapping[3] = {1, 2, 0};
5313 const int chroma_den = 50000;
5314 const int luma_den = 10000;
5317 if (c->fc->nb_streams < 1)
5318 return AVERROR_INVALIDDATA;
5320 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5322 if (atom.size < 24) {
5323 av_log(c->fc, AV_LOG_ERROR, "Invalid Mastering Display Color Volume box\n");
5324 return AVERROR_INVALIDDATA;
5327 sc->mastering = av_mastering_display_metadata_alloc();
5329 return AVERROR(ENOMEM);
5331 for (i = 0; i < 3; i++) {
5332 const int j = mapping[i];
5333 sc->mastering->display_primaries[j][0] = av_make_q(avio_rb16(pb), chroma_den);
5334 sc->mastering->display_primaries[j][1] = av_make_q(avio_rb16(pb), chroma_den);
5336 sc->mastering->white_point[0] = av_make_q(avio_rb16(pb), chroma_den);
5337 sc->mastering->white_point[1] = av_make_q(avio_rb16(pb), chroma_den);
5339 sc->mastering->max_luminance = av_make_q(avio_rb32(pb), luma_den);
5340 sc->mastering->min_luminance = av_make_q(avio_rb32(pb), luma_den);
5342 sc->mastering->has_luminance = 1;
5343 sc->mastering->has_primaries = 1;
5348 static int mov_read_coll(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5350 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 < 5) {
5359 av_log(c->fc, AV_LOG_ERROR, "Empty Content Light Level box\n");
5360 return AVERROR_INVALIDDATA;
5363 version = avio_r8(pb);
5365 av_log(c->fc, AV_LOG_WARNING, "Unsupported Content Light Level box version %d\n", version);
5368 avio_skip(pb, 3); /* flags */
5370 sc->coll = av_content_light_metadata_alloc(&sc->coll_size);
5372 return AVERROR(ENOMEM);
5374 sc->coll->MaxCLL = avio_rb16(pb);
5375 sc->coll->MaxFALL = avio_rb16(pb);
5380 static int mov_read_clli(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5382 MOVStreamContext *sc;
5384 if (c->fc->nb_streams < 1)
5385 return AVERROR_INVALIDDATA;
5387 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
5389 if (atom.size < 4) {
5390 av_log(c->fc, AV_LOG_ERROR, "Empty Content Light Level Info box\n");
5391 return AVERROR_INVALIDDATA;
5394 sc->coll = av_content_light_metadata_alloc(&sc->coll_size);
5396 return AVERROR(ENOMEM);
5398 sc->coll->MaxCLL = avio_rb16(pb);
5399 sc->coll->MaxFALL = avio_rb16(pb);
5404 static int mov_read_st3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5407 MOVStreamContext *sc;
5408 enum AVStereo3DType type;
5411 if (c->fc->nb_streams < 1)
5414 st = c->fc->streams[c->fc->nb_streams - 1];
5417 if (atom.size < 5) {
5418 av_log(c->fc, AV_LOG_ERROR, "Empty stereoscopic video box\n");
5419 return AVERROR_INVALIDDATA;
5421 avio_skip(pb, 4); /* version + flags */
5426 type = AV_STEREO3D_2D;
5429 type = AV_STEREO3D_TOPBOTTOM;
5432 type = AV_STEREO3D_SIDEBYSIDE;
5435 av_log(c->fc, AV_LOG_WARNING, "Unknown st3d mode value %d\n", mode);
5439 sc->stereo3d = av_stereo3d_alloc();
5441 return AVERROR(ENOMEM);
5443 sc->stereo3d->type = type;
5447 static int mov_read_sv3d(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5450 MOVStreamContext *sc;
5451 int size, version, layout;
5452 int32_t yaw, pitch, roll;
5453 uint32_t l = 0, t = 0, r = 0, b = 0;
5454 uint32_t tag, padding = 0;
5455 enum AVSphericalProjection projection;
5457 if (c->fc->nb_streams < 1)
5460 st = c->fc->streams[c->fc->nb_streams - 1];
5463 if (atom.size < 8) {
5464 av_log(c->fc, AV_LOG_ERROR, "Empty spherical video box\n");
5465 return AVERROR_INVALIDDATA;
5468 size = avio_rb32(pb);
5469 if (size <= 12 || size > atom.size)
5470 return AVERROR_INVALIDDATA;
5472 tag = avio_rl32(pb);
5473 if (tag != MKTAG('s','v','h','d')) {
5474 av_log(c->fc, AV_LOG_ERROR, "Missing spherical video header\n");
5477 version = avio_r8(pb);
5479 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5483 avio_skip(pb, 3); /* flags */
5484 avio_skip(pb, size - 12); /* metadata_source */
5486 size = avio_rb32(pb);
5487 if (size > atom.size)
5488 return AVERROR_INVALIDDATA;
5490 tag = avio_rl32(pb);
5491 if (tag != MKTAG('p','r','o','j')) {
5492 av_log(c->fc, AV_LOG_ERROR, "Missing projection box\n");
5496 size = avio_rb32(pb);
5497 if (size > atom.size)
5498 return AVERROR_INVALIDDATA;
5500 tag = avio_rl32(pb);
5501 if (tag != MKTAG('p','r','h','d')) {
5502 av_log(c->fc, AV_LOG_ERROR, "Missing projection header box\n");
5505 version = avio_r8(pb);
5507 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5511 avio_skip(pb, 3); /* flags */
5513 /* 16.16 fixed point */
5514 yaw = avio_rb32(pb);
5515 pitch = avio_rb32(pb);
5516 roll = avio_rb32(pb);
5518 size = avio_rb32(pb);
5519 if (size > atom.size)
5520 return AVERROR_INVALIDDATA;
5522 tag = avio_rl32(pb);
5523 version = avio_r8(pb);
5525 av_log(c->fc, AV_LOG_WARNING, "Unknown spherical version %d\n",
5529 avio_skip(pb, 3); /* flags */
5531 case MKTAG('c','b','m','p'):
5532 layout = avio_rb32(pb);
5534 av_log(c->fc, AV_LOG_WARNING,
5535 "Unsupported cubemap layout %d\n", layout);
5538 projection = AV_SPHERICAL_CUBEMAP;
5539 padding = avio_rb32(pb);
5541 case MKTAG('e','q','u','i'):
5547 if (b >= UINT_MAX - t || r >= UINT_MAX - l) {
5548 av_log(c->fc, AV_LOG_ERROR,
5549 "Invalid bounding rectangle coordinates "
5550 "%"PRIu32",%"PRIu32",%"PRIu32",%"PRIu32"\n", l, t, r, b);
5551 return AVERROR_INVALIDDATA;
5554 if (l || t || r || b)
5555 projection = AV_SPHERICAL_EQUIRECTANGULAR_TILE;
5557 projection = AV_SPHERICAL_EQUIRECTANGULAR;
5560 av_log(c->fc, AV_LOG_ERROR, "Unknown projection type: %s\n", av_fourcc2str(tag));
5564 sc->spherical = av_spherical_alloc(&sc->spherical_size);
5566 return AVERROR(ENOMEM);
5568 sc->spherical->projection = projection;
5570 sc->spherical->yaw = yaw;
5571 sc->spherical->pitch = pitch;
5572 sc->spherical->roll = roll;
5574 sc->spherical->padding = padding;
5576 sc->spherical->bound_left = l;
5577 sc->spherical->bound_top = t;
5578 sc->spherical->bound_right = r;
5579 sc->spherical->bound_bottom = b;
5584 static int mov_parse_uuid_spherical(MOVStreamContext *sc, AVIOContext *pb, size_t len)
5587 uint8_t *buffer = av_malloc(len + 1);
5591 return AVERROR(ENOMEM);
5594 ret = ffio_read_size(pb, buffer, len);
5598 /* Check for mandatory keys and values, try to support XML as best-effort */
5599 if (!sc->spherical &&
5600 av_stristr(buffer, "<GSpherical:StitchingSoftware>") &&
5601 (val = av_stristr(buffer, "<GSpherical:Spherical>")) &&
5602 av_stristr(val, "true") &&
5603 (val = av_stristr(buffer, "<GSpherical:Stitched>")) &&
5604 av_stristr(val, "true") &&
5605 (val = av_stristr(buffer, "<GSpherical:ProjectionType>")) &&
5606 av_stristr(val, "equirectangular")) {
5607 sc->spherical = av_spherical_alloc(&sc->spherical_size);
5611 sc->spherical->projection = AV_SPHERICAL_EQUIRECTANGULAR;
5613 if (av_stristr(buffer, "<GSpherical:StereoMode>") && !sc->stereo3d) {
5614 enum AVStereo3DType mode;
5616 if (av_stristr(buffer, "left-right"))
5617 mode = AV_STEREO3D_SIDEBYSIDE;
5618 else if (av_stristr(buffer, "top-bottom"))
5619 mode = AV_STEREO3D_TOPBOTTOM;
5621 mode = AV_STEREO3D_2D;
5623 sc->stereo3d = av_stereo3d_alloc();
5627 sc->stereo3d->type = mode;
5631 val = av_stristr(buffer, "<GSpherical:InitialViewHeadingDegrees>");
5633 sc->spherical->yaw = strtol(val, NULL, 10) * (1 << 16);
5634 val = av_stristr(buffer, "<GSpherical:InitialViewPitchDegrees>");
5636 sc->spherical->pitch = strtol(val, NULL, 10) * (1 << 16);
5637 val = av_stristr(buffer, "<GSpherical:InitialViewRollDegrees>");
5639 sc->spherical->roll = strtol(val, NULL, 10) * (1 << 16);
5647 static int mov_read_uuid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5650 MOVStreamContext *sc;
5653 static const uint8_t uuid_isml_manifest[] = {
5654 0xa5, 0xd4, 0x0b, 0x30, 0xe8, 0x14, 0x11, 0xdd,
5655 0xba, 0x2f, 0x08, 0x00, 0x20, 0x0c, 0x9a, 0x66
5657 static const uint8_t uuid_xmp[] = {
5658 0xbe, 0x7a, 0xcf, 0xcb, 0x97, 0xa9, 0x42, 0xe8,
5659 0x9c, 0x71, 0x99, 0x94, 0x91, 0xe3, 0xaf, 0xac
5661 static const uint8_t uuid_spherical[] = {
5662 0xff, 0xcc, 0x82, 0x63, 0xf8, 0x55, 0x4a, 0x93,
5663 0x88, 0x14, 0x58, 0x7a, 0x02, 0x52, 0x1f, 0xdd,
5666 if (atom.size < sizeof(uuid) || atom.size >= FFMIN(INT_MAX, SIZE_MAX))
5667 return AVERROR_INVALIDDATA;
5669 if (c->fc->nb_streams < 1)
5671 st = c->fc->streams[c->fc->nb_streams - 1];
5674 ret = avio_read(pb, uuid, sizeof(uuid));
5677 } else if (ret != sizeof(uuid)) {
5678 return AVERROR_INVALIDDATA;
5680 if (!memcmp(uuid, uuid_isml_manifest, sizeof(uuid))) {
5681 uint8_t *buffer, *ptr;
5683 size_t len = atom.size - sizeof(uuid);
5686 return AVERROR_INVALIDDATA;
5688 ret = avio_skip(pb, 4); // zeroes
5691 buffer = av_mallocz(len + 1);
5693 return AVERROR(ENOMEM);
5695 ret = avio_read(pb, buffer, len);
5699 } else if (ret != len) {
5701 return AVERROR_INVALIDDATA;
5705 while ((ptr = av_stristr(ptr, "systemBitrate=\""))) {
5706 ptr += sizeof("systemBitrate=\"") - 1;
5707 c->bitrates_count++;
5708 c->bitrates = av_realloc_f(c->bitrates, c->bitrates_count, sizeof(*c->bitrates));
5710 c->bitrates_count = 0;
5712 return AVERROR(ENOMEM);
5715 ret = strtol(ptr, &endptr, 10);
5716 if (ret < 0 || errno || *endptr != '"') {
5717 c->bitrates[c->bitrates_count - 1] = 0;
5719 c->bitrates[c->bitrates_count - 1] = ret;
5724 } else if (!memcmp(uuid, uuid_xmp, sizeof(uuid))) {
5726 size_t len = atom.size - sizeof(uuid);
5727 if (c->export_xmp) {
5728 buffer = av_mallocz(len + 1);
5730 return AVERROR(ENOMEM);
5732 ret = avio_read(pb, buffer, len);
5736 } else if (ret != len) {
5738 return AVERROR_INVALIDDATA;
5741 av_dict_set(&c->fc->metadata, "xmp", buffer, 0);
5744 // skip all uuid atom, which makes it fast for long uuid-xmp file
5745 ret = avio_skip(pb, len);
5749 } else if (!memcmp(uuid, uuid_spherical, sizeof(uuid))) {
5750 size_t len = atom.size - sizeof(uuid);
5751 ret = mov_parse_uuid_spherical(sc, pb, len);
5755 av_log(c->fc, AV_LOG_WARNING, "Invalid spherical metadata found\n");
5761 static int mov_read_free(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5764 uint8_t content[16];
5769 ret = avio_read(pb, content, FFMIN(sizeof(content), atom.size));
5775 && !memcmp(content, "Anevia\x1A\x1A", 8)
5776 && c->use_mfra_for == FF_MOV_FLAG_MFRA_AUTO) {
5777 c->use_mfra_for = FF_MOV_FLAG_MFRA_PTS;
5783 static int mov_read_frma(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5785 uint32_t format = avio_rl32(pb);
5786 MOVStreamContext *sc;
5790 if (c->fc->nb_streams < 1)
5792 st = c->fc->streams[c->fc->nb_streams - 1];
5797 case MKTAG('e','n','c','v'): // encrypted video
5798 case MKTAG('e','n','c','a'): // encrypted audio
5799 id = mov_codec_id(st, format);
5800 if (st->codecpar->codec_id != AV_CODEC_ID_NONE &&
5801 st->codecpar->codec_id != id) {
5802 av_log(c->fc, AV_LOG_WARNING,
5803 "ignoring 'frma' atom of '%.4s', stream has codec id %d\n",
5804 (char*)&format, st->codecpar->codec_id);
5808 st->codecpar->codec_id = id;
5809 sc->format = format;
5813 if (format != sc->format) {
5814 av_log(c->fc, AV_LOG_WARNING,
5815 "ignoring 'frma' atom of '%.4s', stream format is '%.4s'\n",
5816 (char*)&format, (char*)&sc->format);
5825 * Gets the current encryption info and associated current stream context. If
5826 * we are parsing a track fragment, this will return the specific encryption
5827 * info for this fragment; otherwise this will return the global encryption
5828 * info for the current stream.
5830 static int get_current_encryption_info(MOVContext *c, MOVEncryptionIndex **encryption_index, MOVStreamContext **sc)
5832 MOVFragmentStreamInfo *frag_stream_info;
5836 frag_stream_info = get_current_frag_stream_info(&c->frag_index);
5837 if (frag_stream_info) {
5838 for (i = 0; i < c->fc->nb_streams; i++) {
5839 if (c->fc->streams[i]->id == frag_stream_info->id) {
5840 st = c->fc->streams[i];
5844 if (i == c->fc->nb_streams)
5846 *sc = st->priv_data;
5848 if (!frag_stream_info->encryption_index) {
5849 // If this stream isn't encrypted, don't create the index.
5850 if (!(*sc)->cenc.default_encrypted_sample)
5852 frag_stream_info->encryption_index = av_mallocz(sizeof(*frag_stream_info->encryption_index));
5853 if (!frag_stream_info->encryption_index)
5854 return AVERROR(ENOMEM);
5856 *encryption_index = frag_stream_info->encryption_index;
5859 // No current track fragment, using stream level encryption info.
5861 if (c->fc->nb_streams < 1)
5863 st = c->fc->streams[c->fc->nb_streams - 1];
5864 *sc = st->priv_data;
5866 if (!(*sc)->cenc.encryption_index) {
5867 // If this stream isn't encrypted, don't create the index.
5868 if (!(*sc)->cenc.default_encrypted_sample)
5870 (*sc)->cenc.encryption_index = av_mallocz(sizeof(*frag_stream_info->encryption_index));
5871 if (!(*sc)->cenc.encryption_index)
5872 return AVERROR(ENOMEM);
5875 *encryption_index = (*sc)->cenc.encryption_index;
5880 static int mov_read_sample_encryption_info(MOVContext *c, AVIOContext *pb, MOVStreamContext *sc, AVEncryptionInfo **sample, int use_subsamples)
5883 unsigned int subsample_count;
5884 AVSubsampleEncryptionInfo *subsamples;
5886 if (!sc->cenc.default_encrypted_sample) {
5887 av_log(c->fc, AV_LOG_ERROR, "Missing schm or tenc\n");
5888 return AVERROR_INVALIDDATA;
5891 *sample = av_encryption_info_clone(sc->cenc.default_encrypted_sample);
5893 return AVERROR(ENOMEM);
5895 if (sc->cenc.per_sample_iv_size != 0) {
5896 if (avio_read(pb, (*sample)->iv, sc->cenc.per_sample_iv_size) != sc->cenc.per_sample_iv_size) {
5897 av_log(c->fc, AV_LOG_ERROR, "failed to read the initialization vector\n");
5898 av_encryption_info_free(*sample);
5900 return AVERROR_INVALIDDATA;
5904 if (use_subsamples) {
5905 subsample_count = avio_rb16(pb);
5906 av_free((*sample)->subsamples);
5907 (*sample)->subsamples = av_mallocz_array(subsample_count, sizeof(*subsamples));
5908 if (!(*sample)->subsamples) {
5909 av_encryption_info_free(*sample);
5911 return AVERROR(ENOMEM);
5914 for (i = 0; i < subsample_count && !pb->eof_reached; i++) {
5915 (*sample)->subsamples[i].bytes_of_clear_data = avio_rb16(pb);
5916 (*sample)->subsamples[i].bytes_of_protected_data = avio_rb32(pb);
5919 if (pb->eof_reached) {
5920 av_log(c->fc, AV_LOG_ERROR, "hit EOF while reading sub-sample encryption info\n");
5921 av_encryption_info_free(*sample);
5923 return AVERROR_INVALIDDATA;
5925 (*sample)->subsample_count = subsample_count;
5931 static int mov_read_senc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
5933 AVEncryptionInfo **encrypted_samples;
5934 MOVEncryptionIndex *encryption_index;
5935 MOVStreamContext *sc;
5936 int use_subsamples, ret;
5937 unsigned int sample_count, i, alloc_size = 0;
5939 ret = get_current_encryption_info(c, &encryption_index, &sc);
5943 if (encryption_index->nb_encrypted_samples) {
5944 // This can happen if we have both saio/saiz and senc atoms.
5945 av_log(c->fc, AV_LOG_DEBUG, "Ignoring duplicate encryption info in senc\n");
5949 avio_r8(pb); /* version */
5950 use_subsamples = avio_rb24(pb) & 0x02; /* flags */
5952 sample_count = avio_rb32(pb);
5953 if (sample_count >= INT_MAX / sizeof(*encrypted_samples))
5954 return AVERROR(ENOMEM);
5956 for (i = 0; i < sample_count; i++) {
5957 unsigned int min_samples = FFMIN(FFMAX(i + 1, 1024 * 1024), sample_count);
5958 encrypted_samples = av_fast_realloc(encryption_index->encrypted_samples, &alloc_size,
5959 min_samples * sizeof(*encrypted_samples));
5960 if (encrypted_samples) {
5961 encryption_index->encrypted_samples = encrypted_samples;
5963 ret = mov_read_sample_encryption_info(
5964 c, pb, sc, &encryption_index->encrypted_samples[i], use_subsamples);
5966 ret = AVERROR(ENOMEM);
5968 if (pb->eof_reached) {
5969 av_log(c->fc, AV_LOG_ERROR, "Hit EOF while reading senc\n");
5970 ret = AVERROR_INVALIDDATA;
5975 av_encryption_info_free(encryption_index->encrypted_samples[i - 1]);
5976 av_freep(&encryption_index->encrypted_samples);
5980 encryption_index->nb_encrypted_samples = sample_count;
5985 static int mov_parse_auxiliary_info(MOVContext *c, MOVStreamContext *sc, AVIOContext *pb, MOVEncryptionIndex *encryption_index)
5987 AVEncryptionInfo **sample, **encrypted_samples;
5989 size_t sample_count, sample_info_size, i;
5991 unsigned int alloc_size = 0;
5993 if (encryption_index->nb_encrypted_samples)
5995 sample_count = encryption_index->auxiliary_info_sample_count;
5996 if (encryption_index->auxiliary_offsets_count != 1) {
5997 av_log(c->fc, AV_LOG_ERROR, "Multiple auxiliary info chunks are not supported\n");
5998 return AVERROR_PATCHWELCOME;
6000 if (sample_count >= INT_MAX / sizeof(*encrypted_samples))
6001 return AVERROR(ENOMEM);
6003 prev_pos = avio_tell(pb);
6004 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) ||
6005 avio_seek(pb, encryption_index->auxiliary_offsets[0], SEEK_SET) != encryption_index->auxiliary_offsets[0]) {
6006 av_log(c->fc, AV_LOG_INFO, "Failed to seek for auxiliary info, will only parse senc atoms for encryption info\n");
6010 for (i = 0; i < sample_count && !pb->eof_reached; i++) {
6011 unsigned int min_samples = FFMIN(FFMAX(i + 1, 1024 * 1024), sample_count);
6012 encrypted_samples = av_fast_realloc(encryption_index->encrypted_samples, &alloc_size,
6013 min_samples * sizeof(*encrypted_samples));
6014 if (!encrypted_samples) {
6015 ret = AVERROR(ENOMEM);
6018 encryption_index->encrypted_samples = encrypted_samples;
6020 sample = &encryption_index->encrypted_samples[i];
6021 sample_info_size = encryption_index->auxiliary_info_default_size
6022 ? encryption_index->auxiliary_info_default_size
6023 : encryption_index->auxiliary_info_sizes[i];
6025 ret = mov_read_sample_encryption_info(c, pb, sc, sample, sample_info_size > sc->cenc.per_sample_iv_size);
6029 if (pb->eof_reached) {
6030 av_log(c->fc, AV_LOG_ERROR, "Hit EOF while reading auxiliary info\n");
6031 ret = AVERROR_INVALIDDATA;
6033 encryption_index->nb_encrypted_samples = sample_count;
6037 avio_seek(pb, prev_pos, SEEK_SET);
6039 for (; i > 0; i--) {
6040 av_encryption_info_free(encryption_index->encrypted_samples[i - 1]);
6042 av_freep(&encryption_index->encrypted_samples);
6048 * Tries to read the given number of bytes from the stream and puts it in a
6049 * newly allocated buffer. This reads in small chunks to avoid allocating large
6050 * memory if the file contains an invalid/malicious size value.
6052 static int mov_try_read_block(AVIOContext *pb, size_t size, uint8_t **data)
6054 const unsigned int block_size = 1024 * 1024;
6055 uint8_t *buffer = NULL;
6056 unsigned int alloc_size = 0, offset = 0;
6057 while (offset < size) {
6058 unsigned int new_size =
6059 alloc_size >= INT_MAX - block_size ? INT_MAX : alloc_size + block_size;
6060 uint8_t *new_buffer = av_fast_realloc(buffer, &alloc_size, new_size);
6061 unsigned int to_read = FFMIN(size, alloc_size) - offset;
6064 return AVERROR(ENOMEM);
6066 buffer = new_buffer;
6068 if (avio_read(pb, buffer + offset, to_read) != to_read) {
6070 return AVERROR_INVALIDDATA;
6079 static int mov_read_saiz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6081 MOVEncryptionIndex *encryption_index;
6082 MOVStreamContext *sc;
6084 unsigned int sample_count, aux_info_type, aux_info_param;
6086 ret = get_current_encryption_info(c, &encryption_index, &sc);
6090 if (encryption_index->nb_encrypted_samples) {
6091 // This can happen if we have both saio/saiz and senc atoms.
6092 av_log(c->fc, AV_LOG_DEBUG, "Ignoring duplicate encryption info in saiz\n");
6096 if (encryption_index->auxiliary_info_sample_count) {
6097 av_log(c->fc, AV_LOG_ERROR, "Duplicate saiz atom\n");
6098 return AVERROR_INVALIDDATA;
6101 avio_r8(pb); /* version */
6102 if (avio_rb24(pb) & 0x01) { /* flags */
6103 aux_info_type = avio_rb32(pb);
6104 aux_info_param = avio_rb32(pb);
6105 if (sc->cenc.default_encrypted_sample) {
6106 if (aux_info_type != sc->cenc.default_encrypted_sample->scheme) {
6107 av_log(c->fc, AV_LOG_DEBUG, "Ignoring saiz box with non-zero aux_info_type\n");
6110 if (aux_info_param != 0) {
6111 av_log(c->fc, AV_LOG_DEBUG, "Ignoring saiz box with non-zero aux_info_type_parameter\n");
6115 // Didn't see 'schm' or 'tenc', so this isn't encrypted.
6116 if ((aux_info_type == MKBETAG('c','e','n','c') ||
6117 aux_info_type == MKBETAG('c','e','n','s') ||
6118 aux_info_type == MKBETAG('c','b','c','1') ||
6119 aux_info_type == MKBETAG('c','b','c','s')) &&
6120 aux_info_param == 0) {
6121 av_log(c->fc, AV_LOG_ERROR, "Saw encrypted saiz without schm/tenc\n");
6122 return AVERROR_INVALIDDATA;
6127 } else if (!sc->cenc.default_encrypted_sample) {
6128 // Didn't see 'schm' or 'tenc', so this isn't encrypted.
6132 encryption_index->auxiliary_info_default_size = avio_r8(pb);
6133 sample_count = avio_rb32(pb);
6134 encryption_index->auxiliary_info_sample_count = sample_count;
6136 if (encryption_index->auxiliary_info_default_size == 0) {
6137 ret = mov_try_read_block(pb, sample_count, &encryption_index->auxiliary_info_sizes);
6139 av_log(c->fc, AV_LOG_ERROR, "Failed to read the auxiliary info\n");
6144 if (encryption_index->auxiliary_offsets_count) {
6145 return mov_parse_auxiliary_info(c, sc, pb, encryption_index);
6151 static int mov_read_saio(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6153 uint64_t *auxiliary_offsets;
6154 MOVEncryptionIndex *encryption_index;
6155 MOVStreamContext *sc;
6157 unsigned int version, entry_count, aux_info_type, aux_info_param;
6158 unsigned int alloc_size = 0;
6160 ret = get_current_encryption_info(c, &encryption_index, &sc);
6164 if (encryption_index->nb_encrypted_samples) {
6165 // This can happen if we have both saio/saiz and senc atoms.
6166 av_log(c->fc, AV_LOG_DEBUG, "Ignoring duplicate encryption info in saio\n");
6170 if (encryption_index->auxiliary_offsets_count) {
6171 av_log(c->fc, AV_LOG_ERROR, "Duplicate saio atom\n");
6172 return AVERROR_INVALIDDATA;
6175 version = avio_r8(pb); /* version */
6176 if (avio_rb24(pb) & 0x01) { /* flags */
6177 aux_info_type = avio_rb32(pb);
6178 aux_info_param = avio_rb32(pb);
6179 if (sc->cenc.default_encrypted_sample) {
6180 if (aux_info_type != sc->cenc.default_encrypted_sample->scheme) {
6181 av_log(c->fc, AV_LOG_DEBUG, "Ignoring saio box with non-zero aux_info_type\n");
6184 if (aux_info_param != 0) {
6185 av_log(c->fc, AV_LOG_DEBUG, "Ignoring saio box with non-zero aux_info_type_parameter\n");
6189 // Didn't see 'schm' or 'tenc', so this isn't encrypted.
6190 if ((aux_info_type == MKBETAG('c','e','n','c') ||
6191 aux_info_type == MKBETAG('c','e','n','s') ||
6192 aux_info_type == MKBETAG('c','b','c','1') ||
6193 aux_info_type == MKBETAG('c','b','c','s')) &&
6194 aux_info_param == 0) {
6195 av_log(c->fc, AV_LOG_ERROR, "Saw encrypted saio without schm/tenc\n");
6196 return AVERROR_INVALIDDATA;
6201 } else if (!sc->cenc.default_encrypted_sample) {
6202 // Didn't see 'schm' or 'tenc', so this isn't encrypted.
6206 entry_count = avio_rb32(pb);
6207 if (entry_count >= INT_MAX / sizeof(*auxiliary_offsets))
6208 return AVERROR(ENOMEM);
6210 for (i = 0; i < entry_count && !pb->eof_reached; i++) {
6211 unsigned int min_offsets = FFMIN(FFMAX(i + 1, 1024), entry_count);
6212 auxiliary_offsets = av_fast_realloc(
6213 encryption_index->auxiliary_offsets, &alloc_size,
6214 min_offsets * sizeof(*auxiliary_offsets));
6215 if (!auxiliary_offsets) {
6216 av_freep(&encryption_index->auxiliary_offsets);
6217 return AVERROR(ENOMEM);
6219 encryption_index->auxiliary_offsets = auxiliary_offsets;
6222 encryption_index->auxiliary_offsets[i] = avio_rb32(pb);
6224 encryption_index->auxiliary_offsets[i] = avio_rb64(pb);
6226 if (c->frag_index.current >= 0) {
6227 encryption_index->auxiliary_offsets[i] += c->fragment.base_data_offset;
6231 if (pb->eof_reached) {
6232 av_log(c->fc, AV_LOG_ERROR, "Hit EOF while reading saio\n");
6233 av_freep(&encryption_index->auxiliary_offsets);
6234 return AVERROR_INVALIDDATA;
6237 encryption_index->auxiliary_offsets_count = entry_count;
6239 if (encryption_index->auxiliary_info_sample_count) {
6240 return mov_parse_auxiliary_info(c, sc, pb, encryption_index);
6246 static int mov_read_pssh(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6248 AVEncryptionInitInfo *info, *old_init_info;
6251 uint8_t *side_data, *extra_data, *old_side_data;
6252 size_t side_data_size;
6253 int ret = 0, old_side_data_size;
6254 unsigned int version, kid_count, extra_data_size, alloc_size = 0;
6256 if (c->fc->nb_streams < 1)
6258 st = c->fc->streams[c->fc->nb_streams-1];
6260 version = avio_r8(pb); /* version */
6261 avio_rb24(pb); /* flags */
6263 info = av_encryption_init_info_alloc(/* system_id_size */ 16, /* num_key_ids */ 0,
6264 /* key_id_size */ 16, /* data_size */ 0);
6266 return AVERROR(ENOMEM);
6268 if (avio_read(pb, info->system_id, 16) != 16) {
6269 av_log(c->fc, AV_LOG_ERROR, "Failed to read the system id\n");
6270 ret = AVERROR_INVALIDDATA;
6275 kid_count = avio_rb32(pb);
6276 if (kid_count >= INT_MAX / sizeof(*key_ids))
6277 return AVERROR(ENOMEM);
6279 for (unsigned int i = 0; i < kid_count && !pb->eof_reached; i++) {
6280 unsigned int min_kid_count = FFMIN(FFMAX(i + 1, 1024), kid_count);
6281 key_ids = av_fast_realloc(info->key_ids, &alloc_size,
6282 min_kid_count * sizeof(*key_ids));
6284 ret = AVERROR(ENOMEM);
6287 info->key_ids = key_ids;
6289 info->key_ids[i] = av_mallocz(16);
6290 if (!info->key_ids[i]) {
6291 ret = AVERROR(ENOMEM);
6294 info->num_key_ids = i + 1;
6296 if (avio_read(pb, info->key_ids[i], 16) != 16) {
6297 av_log(c->fc, AV_LOG_ERROR, "Failed to read the key id\n");
6298 ret = AVERROR_INVALIDDATA;
6303 if (pb->eof_reached) {
6304 av_log(c->fc, AV_LOG_ERROR, "Hit EOF while reading pssh\n");
6305 ret = AVERROR_INVALIDDATA;
6310 extra_data_size = avio_rb32(pb);
6311 ret = mov_try_read_block(pb, extra_data_size, &extra_data);
6315 av_freep(&info->data); // malloc(0) may still allocate something.
6316 info->data = extra_data;
6317 info->data_size = extra_data_size;
6319 // If there is existing initialization data, append to the list.
6320 old_side_data = av_stream_get_side_data(st, AV_PKT_DATA_ENCRYPTION_INIT_INFO, &old_side_data_size);
6321 if (old_side_data) {
6322 old_init_info = av_encryption_init_info_get_side_data(old_side_data, old_side_data_size);
6323 if (old_init_info) {
6324 // Append to the end of the list.
6325 for (AVEncryptionInitInfo *cur = old_init_info;; cur = cur->next) {
6331 info = old_init_info;
6333 // Assume existing side-data will be valid, so the only error we could get is OOM.
6334 ret = AVERROR(ENOMEM);
6339 side_data = av_encryption_init_info_add_side_data(info, &side_data_size);
6341 ret = AVERROR(ENOMEM);
6344 ret = av_stream_add_side_data(st, AV_PKT_DATA_ENCRYPTION_INIT_INFO,
6345 side_data, side_data_size);
6350 av_encryption_init_info_free(info);
6354 static int mov_read_schm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6357 MOVStreamContext *sc;
6359 if (c->fc->nb_streams < 1)
6361 st = c->fc->streams[c->fc->nb_streams-1];
6364 if (sc->pseudo_stream_id != 0) {
6365 av_log(c->fc, AV_LOG_ERROR, "schm boxes are only supported in first sample descriptor\n");
6366 return AVERROR_PATCHWELCOME;
6370 return AVERROR_INVALIDDATA;
6372 avio_rb32(pb); /* version and flags */
6374 if (!sc->cenc.default_encrypted_sample) {
6375 sc->cenc.default_encrypted_sample = av_encryption_info_alloc(0, 16, 16);
6376 if (!sc->cenc.default_encrypted_sample) {
6377 return AVERROR(ENOMEM);
6381 sc->cenc.default_encrypted_sample->scheme = avio_rb32(pb);
6385 static int mov_read_tenc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6388 MOVStreamContext *sc;
6389 unsigned int version, pattern, is_protected, iv_size;
6391 if (c->fc->nb_streams < 1)
6393 st = c->fc->streams[c->fc->nb_streams-1];
6396 if (sc->pseudo_stream_id != 0) {
6397 av_log(c->fc, AV_LOG_ERROR, "tenc atom are only supported in first sample descriptor\n");
6398 return AVERROR_PATCHWELCOME;
6401 if (!sc->cenc.default_encrypted_sample) {
6402 sc->cenc.default_encrypted_sample = av_encryption_info_alloc(0, 16, 16);
6403 if (!sc->cenc.default_encrypted_sample) {
6404 return AVERROR(ENOMEM);
6409 return AVERROR_INVALIDDATA;
6411 version = avio_r8(pb); /* version */
6412 avio_rb24(pb); /* flags */
6414 avio_r8(pb); /* reserved */
6415 pattern = avio_r8(pb);
6418 sc->cenc.default_encrypted_sample->crypt_byte_block = pattern >> 4;
6419 sc->cenc.default_encrypted_sample->skip_byte_block = pattern & 0xf;
6422 is_protected = avio_r8(pb);
6423 if (is_protected && !sc->cenc.encryption_index) {
6424 // The whole stream should be by-default encrypted.
6425 sc->cenc.encryption_index = av_mallocz(sizeof(MOVEncryptionIndex));
6426 if (!sc->cenc.encryption_index)
6427 return AVERROR(ENOMEM);
6429 sc->cenc.per_sample_iv_size = avio_r8(pb);
6430 if (sc->cenc.per_sample_iv_size != 0 && sc->cenc.per_sample_iv_size != 8 &&
6431 sc->cenc.per_sample_iv_size != 16) {
6432 av_log(c->fc, AV_LOG_ERROR, "invalid per-sample IV size value\n");
6433 return AVERROR_INVALIDDATA;
6435 if (avio_read(pb, sc->cenc.default_encrypted_sample->key_id, 16) != 16) {
6436 av_log(c->fc, AV_LOG_ERROR, "failed to read the default key ID\n");
6437 return AVERROR_INVALIDDATA;
6440 if (is_protected && !sc->cenc.per_sample_iv_size) {
6441 iv_size = avio_r8(pb);
6442 if (iv_size != 8 && iv_size != 16) {
6443 av_log(c->fc, AV_LOG_ERROR, "invalid default_constant_IV_size in tenc atom\n");
6444 return AVERROR_INVALIDDATA;
6447 if (avio_read(pb, sc->cenc.default_encrypted_sample->iv, iv_size) != iv_size) {
6448 av_log(c->fc, AV_LOG_ERROR, "failed to read the default IV\n");
6449 return AVERROR_INVALIDDATA;
6456 static int mov_read_dfla(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6459 int last, type, size, ret;
6462 if (c->fc->nb_streams < 1)
6464 st = c->fc->streams[c->fc->nb_streams-1];
6466 if ((uint64_t)atom.size > (1<<30) || atom.size < 42)
6467 return AVERROR_INVALIDDATA;
6469 /* Check FlacSpecificBox version. */
6470 if (avio_r8(pb) != 0)
6471 return AVERROR_INVALIDDATA;
6473 avio_rb24(pb); /* Flags */
6475 avio_read(pb, buf, sizeof(buf));
6476 flac_parse_block_header(buf, &last, &type, &size);
6478 if (type != FLAC_METADATA_TYPE_STREAMINFO || size != FLAC_STREAMINFO_SIZE) {
6479 av_log(c->fc, AV_LOG_ERROR, "STREAMINFO must be first FLACMetadataBlock\n");
6480 return AVERROR_INVALIDDATA;
6483 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
6488 av_log(c->fc, AV_LOG_WARNING, "non-STREAMINFO FLACMetadataBlock(s) ignored\n");
6493 static int cenc_decrypt(MOVContext *c, MOVStreamContext *sc, AVEncryptionInfo *sample, uint8_t *input, int size)
6497 if (sample->scheme != MKBETAG('c','e','n','c') || sample->crypt_byte_block != 0 || sample->skip_byte_block != 0) {
6498 av_log(c->fc, AV_LOG_ERROR, "Only the 'cenc' encryption scheme is supported\n");
6499 return AVERROR_PATCHWELCOME;
6502 if (!sc->cenc.aes_ctr) {
6503 /* initialize the cipher */
6504 sc->cenc.aes_ctr = av_aes_ctr_alloc();
6505 if (!sc->cenc.aes_ctr) {
6506 return AVERROR(ENOMEM);
6509 ret = av_aes_ctr_init(sc->cenc.aes_ctr, c->decryption_key);
6515 av_aes_ctr_set_full_iv(sc->cenc.aes_ctr, sample->iv);
6517 if (!sample->subsample_count)
6519 /* decrypt the whole packet */
6520 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, size);
6524 for (i = 0; i < sample->subsample_count; i++)
6526 if (sample->subsamples[i].bytes_of_clear_data + sample->subsamples[i].bytes_of_protected_data > size) {
6527 av_log(c->fc, AV_LOG_ERROR, "subsample size exceeds the packet size left\n");
6528 return AVERROR_INVALIDDATA;
6531 /* skip the clear bytes */
6532 input += sample->subsamples[i].bytes_of_clear_data;
6533 size -= sample->subsamples[i].bytes_of_clear_data;
6535 /* decrypt the encrypted bytes */
6536 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, sample->subsamples[i].bytes_of_protected_data);
6537 input += sample->subsamples[i].bytes_of_protected_data;
6538 size -= sample->subsamples[i].bytes_of_protected_data;
6542 av_log(c->fc, AV_LOG_ERROR, "leftover packet bytes after subsample processing\n");
6543 return AVERROR_INVALIDDATA;
6549 static int cenc_filter(MOVContext *mov, MOVStreamContext *sc, AVPacket *pkt, int current_index)
6551 MOVFragmentStreamInfo *frag_stream_info;
6552 MOVEncryptionIndex *encryption_index;
6553 AVEncryptionInfo *encrypted_sample;
6554 int encrypted_index, ret;
6556 frag_stream_info = get_current_frag_stream_info(&mov->frag_index);
6557 encrypted_index = current_index;
6558 encryption_index = NULL;
6559 if (frag_stream_info) {
6560 // Note this only supports encryption info in the first sample descriptor.
6561 if (mov->fragment.stsd_id == 1) {
6562 if (frag_stream_info->encryption_index) {
6563 encrypted_index = current_index - frag_stream_info->index_entry;
6564 encryption_index = frag_stream_info->encryption_index;
6566 encryption_index = sc->cenc.encryption_index;
6570 encryption_index = sc->cenc.encryption_index;
6573 if (encryption_index) {
6574 if (encryption_index->auxiliary_info_sample_count &&
6575 !encryption_index->nb_encrypted_samples) {
6576 av_log(mov->fc, AV_LOG_ERROR, "saiz atom found without saio\n");
6577 return AVERROR_INVALIDDATA;
6579 if (encryption_index->auxiliary_offsets_count &&
6580 !encryption_index->nb_encrypted_samples) {
6581 av_log(mov->fc, AV_LOG_ERROR, "saio atom found without saiz\n");
6582 return AVERROR_INVALIDDATA;
6585 if (!encryption_index->nb_encrypted_samples) {
6586 // Full-sample encryption with default settings.
6587 encrypted_sample = sc->cenc.default_encrypted_sample;
6588 } else if (encrypted_index >= 0 && encrypted_index < encryption_index->nb_encrypted_samples) {
6589 // Per-sample setting override.
6590 encrypted_sample = encryption_index->encrypted_samples[encrypted_index];
6592 av_log(mov->fc, AV_LOG_ERROR, "Incorrect number of samples in encryption info\n");
6593 return AVERROR_INVALIDDATA;
6596 if (mov->decryption_key) {
6597 return cenc_decrypt(mov, sc, encrypted_sample, pkt->data, pkt->size);
6600 uint8_t *side_data = av_encryption_info_add_side_data(encrypted_sample, &size);
6602 return AVERROR(ENOMEM);
6603 ret = av_packet_add_side_data(pkt, AV_PKT_DATA_ENCRYPTION_INFO, side_data, size);
6613 static int mov_read_dops(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6615 const int OPUS_SEEK_PREROLL_MS = 80;
6620 if (c->fc->nb_streams < 1)
6622 st = c->fc->streams[c->fc->nb_streams-1];
6624 if ((uint64_t)atom.size > (1<<30) || atom.size < 11)
6625 return AVERROR_INVALIDDATA;
6627 /* Check OpusSpecificBox version. */
6628 if (avio_r8(pb) != 0) {
6629 av_log(c->fc, AV_LOG_ERROR, "unsupported OpusSpecificBox version\n");
6630 return AVERROR_INVALIDDATA;
6633 /* OpusSpecificBox size plus magic for Ogg OpusHead header. */
6634 size = atom.size + 8;
6636 if (ff_alloc_extradata(st->codecpar, size))
6637 return AVERROR(ENOMEM);
6639 AV_WL32(st->codecpar->extradata, MKTAG('O','p','u','s'));
6640 AV_WL32(st->codecpar->extradata + 4, MKTAG('H','e','a','d'));
6641 AV_WB8(st->codecpar->extradata + 8, 1); /* OpusHead version */
6642 avio_read(pb, st->codecpar->extradata + 9, size - 9);
6644 /* OpusSpecificBox is stored in big-endian, but OpusHead is
6645 little-endian; aside from the preceeding magic and version they're
6646 otherwise currently identical. Data after output gain at offset 16
6647 doesn't need to be bytewapped. */
6648 pre_skip = AV_RB16(st->codecpar->extradata + 10);
6649 AV_WL16(st->codecpar->extradata + 10, pre_skip);
6650 AV_WL32(st->codecpar->extradata + 12, AV_RB32(st->codecpar->extradata + 12));
6651 AV_WL16(st->codecpar->extradata + 16, AV_RB16(st->codecpar->extradata + 16));
6653 st->codecpar->initial_padding = pre_skip;
6654 st->codecpar->seek_preroll = av_rescale_q(OPUS_SEEK_PREROLL_MS,
6655 (AVRational){1, 1000},
6656 (AVRational){1, 48000});
6661 static const MOVParseTableEntry mov_default_parse_table[] = {
6662 { MKTAG('A','C','L','R'), mov_read_aclr },
6663 { MKTAG('A','P','R','G'), mov_read_avid },
6664 { MKTAG('A','A','L','P'), mov_read_avid },
6665 { MKTAG('A','R','E','S'), mov_read_ares },
6666 { MKTAG('a','v','s','s'), mov_read_avss },
6667 { MKTAG('a','v','1','C'), mov_read_av1c },
6668 { MKTAG('c','h','p','l'), mov_read_chpl },
6669 { MKTAG('c','o','6','4'), mov_read_stco },
6670 { MKTAG('c','o','l','r'), mov_read_colr },
6671 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
6672 { MKTAG('d','i','n','f'), mov_read_default },
6673 { MKTAG('D','p','x','E'), mov_read_dpxe },
6674 { MKTAG('d','r','e','f'), mov_read_dref },
6675 { MKTAG('e','d','t','s'), mov_read_default },
6676 { MKTAG('e','l','s','t'), mov_read_elst },
6677 { MKTAG('e','n','d','a'), mov_read_enda },
6678 { MKTAG('f','i','e','l'), mov_read_fiel },
6679 { MKTAG('a','d','r','m'), mov_read_adrm },
6680 { MKTAG('f','t','y','p'), mov_read_ftyp },
6681 { MKTAG('g','l','b','l'), mov_read_glbl },
6682 { MKTAG('h','d','l','r'), mov_read_hdlr },
6683 { MKTAG('i','l','s','t'), mov_read_ilst },
6684 { MKTAG('j','p','2','h'), mov_read_jp2h },
6685 { MKTAG('m','d','a','t'), mov_read_mdat },
6686 { MKTAG('m','d','h','d'), mov_read_mdhd },
6687 { MKTAG('m','d','i','a'), mov_read_default },
6688 { MKTAG('m','e','t','a'), mov_read_meta },
6689 { MKTAG('m','i','n','f'), mov_read_default },
6690 { MKTAG('m','o','o','f'), mov_read_moof },
6691 { MKTAG('m','o','o','v'), mov_read_moov },
6692 { MKTAG('m','v','e','x'), mov_read_default },
6693 { MKTAG('m','v','h','d'), mov_read_mvhd },
6694 { MKTAG('S','M','I',' '), mov_read_svq3 },
6695 { MKTAG('a','l','a','c'), mov_read_alac }, /* alac specific atom */
6696 { MKTAG('a','v','c','C'), mov_read_glbl },
6697 { MKTAG('p','a','s','p'), mov_read_pasp },
6698 { MKTAG('s','i','d','x'), mov_read_sidx },
6699 { MKTAG('s','t','b','l'), mov_read_default },
6700 { MKTAG('s','t','c','o'), mov_read_stco },
6701 { MKTAG('s','t','p','s'), mov_read_stps },
6702 { MKTAG('s','t','r','f'), mov_read_strf },
6703 { MKTAG('s','t','s','c'), mov_read_stsc },
6704 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
6705 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
6706 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
6707 { MKTAG('s','t','t','s'), mov_read_stts },
6708 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
6709 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
6710 { MKTAG('t','f','d','t'), mov_read_tfdt },
6711 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
6712 { MKTAG('t','r','a','k'), mov_read_trak },
6713 { MKTAG('t','r','a','f'), mov_read_default },
6714 { MKTAG('t','r','e','f'), mov_read_default },
6715 { MKTAG('t','m','c','d'), mov_read_tmcd },
6716 { MKTAG('c','h','a','p'), mov_read_chap },
6717 { MKTAG('t','r','e','x'), mov_read_trex },
6718 { MKTAG('t','r','u','n'), mov_read_trun },
6719 { MKTAG('u','d','t','a'), mov_read_default },
6720 { MKTAG('w','a','v','e'), mov_read_wave },
6721 { MKTAG('e','s','d','s'), mov_read_esds },
6722 { MKTAG('d','a','c','3'), mov_read_dac3 }, /* AC-3 info */
6723 { MKTAG('d','e','c','3'), mov_read_dec3 }, /* EAC-3 info */
6724 { MKTAG('d','d','t','s'), mov_read_ddts }, /* DTS audio descriptor */
6725 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
6726 { MKTAG('w','f','e','x'), mov_read_wfex },
6727 { MKTAG('c','m','o','v'), mov_read_cmov },
6728 { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */
6729 { MKTAG('d','v','c','1'), mov_read_dvc1 },
6730 { MKTAG('s','b','g','p'), mov_read_sbgp },
6731 { MKTAG('h','v','c','C'), mov_read_glbl },
6732 { MKTAG('u','u','i','d'), mov_read_uuid },
6733 { MKTAG('C','i','n', 0x8e), mov_read_targa_y216 },
6734 { MKTAG('f','r','e','e'), mov_read_free },
6735 { MKTAG('-','-','-','-'), mov_read_custom },
6736 { MKTAG('s','i','n','f'), mov_read_default },
6737 { MKTAG('f','r','m','a'), mov_read_frma },
6738 { MKTAG('s','e','n','c'), mov_read_senc },
6739 { MKTAG('s','a','i','z'), mov_read_saiz },
6740 { MKTAG('s','a','i','o'), mov_read_saio },
6741 { MKTAG('p','s','s','h'), mov_read_pssh },
6742 { MKTAG('s','c','h','m'), mov_read_schm },
6743 { MKTAG('s','c','h','i'), mov_read_default },
6744 { MKTAG('t','e','n','c'), mov_read_tenc },
6745 { MKTAG('d','f','L','a'), mov_read_dfla },
6746 { MKTAG('s','t','3','d'), mov_read_st3d }, /* stereoscopic 3D video box */
6747 { MKTAG('s','v','3','d'), mov_read_sv3d }, /* spherical video box */
6748 { MKTAG('d','O','p','s'), mov_read_dops },
6749 { MKTAG('S','m','D','m'), mov_read_smdm },
6750 { MKTAG('C','o','L','L'), mov_read_coll },
6751 { MKTAG('v','p','c','C'), mov_read_vpcc },
6752 { MKTAG('m','d','c','v'), mov_read_mdcv },
6753 { MKTAG('c','l','l','i'), mov_read_clli },
6757 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom)
6759 int64_t total_size = 0;
6763 if (c->atom_depth > 10) {
6764 av_log(c->fc, AV_LOG_ERROR, "Atoms too deeply nested\n");
6765 return AVERROR_INVALIDDATA;
6770 atom.size = INT64_MAX;
6771 while (total_size <= atom.size - 8 && !avio_feof(pb)) {
6772 int (*parse)(MOVContext*, AVIOContext*, MOVAtom) = NULL;
6775 if (atom.size >= 8) {
6776 a.size = avio_rb32(pb);
6777 a.type = avio_rl32(pb);
6778 if (a.type == MKTAG('f','r','e','e') &&
6780 c->fc->strict_std_compliance < FF_COMPLIANCE_STRICT &&
6783 uint32_t *type = (uint32_t *)buf + 1;
6784 if (avio_read(pb, buf, 8) != 8)
6785 return AVERROR_INVALIDDATA;
6786 avio_seek(pb, -8, SEEK_CUR);
6787 if (*type == MKTAG('m','v','h','d') ||
6788 *type == MKTAG('c','m','o','v')) {
6789 av_log(c->fc, AV_LOG_ERROR, "Detected moov in a free atom.\n");
6790 a.type = MKTAG('m','o','o','v');
6793 if (atom.type != MKTAG('r','o','o','t') &&
6794 atom.type != MKTAG('m','o','o','v'))
6796 if (a.type == MKTAG('t','r','a','k') || a.type == MKTAG('m','d','a','t'))
6798 av_log(c->fc, AV_LOG_ERROR, "Broken file, trak/mdat not at top-level\n");
6805 if (a.size == 1 && total_size + 8 <= atom.size) { /* 64 bit extended size */
6806 a.size = avio_rb64(pb) - 8;
6810 av_log(c->fc, AV_LOG_TRACE, "type:'%s' parent:'%s' sz: %"PRId64" %"PRId64" %"PRId64"\n",
6811 av_fourcc2str(a.type), av_fourcc2str(atom.type), a.size, total_size, atom.size);
6813 a.size = atom.size - total_size + 8;
6818 a.size = FFMIN(a.size, atom.size - total_size);
6820 for (i = 0; mov_default_parse_table[i].type; i++)
6821 if (mov_default_parse_table[i].type == a.type) {
6822 parse = mov_default_parse_table[i].parse;
6826 // container is user data
6827 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
6828 atom.type == MKTAG('i','l','s','t')))
6829 parse = mov_read_udta_string;
6831 // Supports parsing the QuickTime Metadata Keys.
6832 // https://developer.apple.com/library/mac/documentation/QuickTime/QTFF/Metadata/Metadata.html
6833 if (!parse && c->found_hdlr_mdta &&
6834 atom.type == MKTAG('m','e','t','a') &&
6835 a.type == MKTAG('k','e','y','s')) {
6836 parse = mov_read_keys;
6839 if (!parse) { /* skip leaf atoms data */
6840 avio_skip(pb, a.size);
6842 int64_t start_pos = avio_tell(pb);
6844 int err = parse(c, pb, a);
6849 if (c->found_moov && c->found_mdat &&
6850 ((!(pb->seekable & AVIO_SEEKABLE_NORMAL) || c->fc->flags & AVFMT_FLAG_IGNIDX || c->frag_index.complete) ||
6851 start_pos + a.size == avio_size(pb))) {
6852 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) || c->fc->flags & AVFMT_FLAG_IGNIDX || c->frag_index.complete)
6853 c->next_root_atom = start_pos + a.size;
6857 left = a.size - avio_tell(pb) + start_pos;
6858 if (left > 0) /* skip garbage at atom end */
6859 avio_skip(pb, left);
6860 else if (left < 0) {
6861 av_log(c->fc, AV_LOG_WARNING,
6862 "overread end of atom '%.4s' by %"PRId64" bytes\n",
6863 (char*)&a.type, -left);
6864 avio_seek(pb, left, SEEK_CUR);
6868 total_size += a.size;
6871 if (total_size < atom.size && atom.size < 0x7ffff)
6872 avio_skip(pb, atom.size - total_size);
6878 static int mov_probe(AVProbeData *p)
6883 int moov_offset = -1;
6885 /* check file header */
6888 /* ignore invalid offset */
6889 if ((offset + 8) > (unsigned int)p->buf_size)
6891 tag = AV_RL32(p->buf + offset + 4);
6893 /* check for obvious tags */
6894 case MKTAG('m','o','o','v'):
6895 moov_offset = offset + 4;
6896 case MKTAG('m','d','a','t'):
6897 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
6898 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
6899 case MKTAG('f','t','y','p'):
6900 if (AV_RB32(p->buf+offset) < 8 &&
6901 (AV_RB32(p->buf+offset) != 1 ||
6902 offset + 12 > (unsigned int)p->buf_size ||
6903 AV_RB64(p->buf+offset + 8) == 0)) {
6904 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
6905 } else if (tag == MKTAG('f','t','y','p') &&
6906 ( AV_RL32(p->buf + offset + 8) == MKTAG('j','p','2',' ')
6907 || AV_RL32(p->buf + offset + 8) == MKTAG('j','p','x',' ')
6909 score = FFMAX(score, 5);
6911 score = AVPROBE_SCORE_MAX;
6913 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6915 /* those are more common words, so rate then a bit less */
6916 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
6917 case MKTAG('w','i','d','e'):
6918 case MKTAG('f','r','e','e'):
6919 case MKTAG('j','u','n','k'):
6920 case MKTAG('p','i','c','t'):
6921 score = FFMAX(score, AVPROBE_SCORE_MAX - 5);
6922 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6924 case MKTAG(0x82,0x82,0x7f,0x7d):
6925 case MKTAG('s','k','i','p'):
6926 case MKTAG('u','u','i','d'):
6927 case MKTAG('p','r','f','l'):
6928 /* if we only find those cause probedata is too small at least rate them */
6929 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
6930 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6933 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
6936 if(score > AVPROBE_SCORE_MAX - 50 && moov_offset != -1) {
6937 /* moov atom in the header - we should make sure that this is not a
6938 * MOV-packed MPEG-PS */
6939 offset = moov_offset;
6941 while(offset < (p->buf_size - 16)){ /* Sufficient space */
6942 /* We found an actual hdlr atom */
6943 if(AV_RL32(p->buf + offset ) == MKTAG('h','d','l','r') &&
6944 AV_RL32(p->buf + offset + 8) == MKTAG('m','h','l','r') &&
6945 AV_RL32(p->buf + offset + 12) == MKTAG('M','P','E','G')){
6946 av_log(NULL, AV_LOG_WARNING, "Found media data tag MPEG indicating this is a MOV-packed MPEG-PS.\n");
6947 /* We found a media handler reference atom describing an
6948 * MPEG-PS-in-MOV, return a
6949 * low score to force expanding the probe window until
6950 * mpegps_probe finds what it needs */
6961 // must be done after parsing all trak because there's no order requirement
6962 static void mov_read_chapters(AVFormatContext *s)
6964 MOVContext *mov = s->priv_data;
6966 MOVStreamContext *sc;
6971 for (j = 0; j < mov->nb_chapter_tracks; j++) {
6972 chapter_track = mov->chapter_tracks[j];
6974 for (i = 0; i < s->nb_streams; i++)
6975 if (s->streams[i]->id == chapter_track) {
6980 av_log(s, AV_LOG_ERROR, "Referenced QT chapter track not found\n");
6985 cur_pos = avio_tell(sc->pb);
6987 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
6988 st->disposition |= AV_DISPOSITION_ATTACHED_PIC | AV_DISPOSITION_TIMED_THUMBNAILS;
6989 if (st->nb_index_entries) {
6990 // Retrieve the first frame, if possible
6992 AVIndexEntry *sample = &st->index_entries[0];
6993 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
6994 av_log(s, AV_LOG_ERROR, "Failed to retrieve first frame\n");
6998 if (av_get_packet(sc->pb, &pkt, sample->size) < 0)
7001 st->attached_pic = pkt;
7002 st->attached_pic.stream_index = st->index;
7003 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
7006 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
7007 st->codecpar->codec_id = AV_CODEC_ID_BIN_DATA;
7008 st->discard = AVDISCARD_ALL;
7009 for (i = 0; i < st->nb_index_entries; i++) {
7010 AVIndexEntry *sample = &st->index_entries[i];
7011 int64_t end = i+1 < st->nb_index_entries ? st->index_entries[i+1].timestamp : st->duration;
7016 if (end < sample->timestamp) {
7017 av_log(s, AV_LOG_WARNING, "ignoring stream duration which is shorter than chapters\n");
7018 end = AV_NOPTS_VALUE;
7021 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
7022 av_log(s, AV_LOG_ERROR, "Chapter %d not found in file\n", i);
7026 // the first two bytes are the length of the title
7027 len = avio_rb16(sc->pb);
7028 if (len > sample->size-2)
7030 title_len = 2*len + 1;
7031 if (!(title = av_mallocz(title_len)))
7034 // The samples could theoretically be in any encoding if there's an encd
7035 // atom following, but in practice are only utf-8 or utf-16, distinguished
7036 // instead by the presence of a BOM
7040 ch = avio_rb16(sc->pb);
7042 avio_get_str16be(sc->pb, len, title, title_len);
7043 else if (ch == 0xfffe)
7044 avio_get_str16le(sc->pb, len, title, title_len);
7047 if (len == 1 || len == 2)
7050 avio_get_str(sc->pb, INT_MAX, title + 2, len - 1);
7054 avpriv_new_chapter(s, i, st->time_base, sample->timestamp, end, title);
7059 avio_seek(sc->pb, cur_pos, SEEK_SET);
7063 static int parse_timecode_in_framenum_format(AVFormatContext *s, AVStream *st,
7064 uint32_t value, int flags)
7067 char buf[AV_TIMECODE_STR_SIZE];
7068 AVRational rate = st->avg_frame_rate;
7069 int ret = av_timecode_init(&tc, rate, flags, 0, s);
7072 av_dict_set(&st->metadata, "timecode",
7073 av_timecode_make_string(&tc, buf, value), 0);
7077 static int mov_read_rtmd_track(AVFormatContext *s, AVStream *st)
7079 MOVStreamContext *sc = st->priv_data;
7080 char buf[AV_TIMECODE_STR_SIZE];
7081 int64_t cur_pos = avio_tell(sc->pb);
7082 int hh, mm, ss, ff, drop;
7084 if (!st->nb_index_entries)
7087 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
7088 avio_skip(s->pb, 13);
7089 hh = avio_r8(s->pb);
7090 mm = avio_r8(s->pb);
7091 ss = avio_r8(s->pb);
7092 drop = avio_r8(s->pb);
7093 ff = avio_r8(s->pb);
7094 snprintf(buf, AV_TIMECODE_STR_SIZE, "%02d:%02d:%02d%c%02d",
7095 hh, mm, ss, drop ? ';' : ':', ff);
7096 av_dict_set(&st->metadata, "timecode", buf, 0);
7098 avio_seek(sc->pb, cur_pos, SEEK_SET);
7102 static int mov_read_timecode_track(AVFormatContext *s, AVStream *st)
7104 MOVStreamContext *sc = st->priv_data;
7106 int64_t cur_pos = avio_tell(sc->pb);
7109 if (!st->nb_index_entries)
7112 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
7113 value = avio_rb32(s->pb);
7115 if (sc->tmcd_flags & 0x0001) flags |= AV_TIMECODE_FLAG_DROPFRAME;
7116 if (sc->tmcd_flags & 0x0002) flags |= AV_TIMECODE_FLAG_24HOURSMAX;
7117 if (sc->tmcd_flags & 0x0004) flags |= AV_TIMECODE_FLAG_ALLOWNEGATIVE;
7119 /* Assume Counter flag is set to 1 in tmcd track (even though it is likely
7120 * not the case) and thus assume "frame number format" instead of QT one.
7121 * No sample with tmcd track can be found with a QT timecode at the moment,
7122 * despite what the tmcd track "suggests" (Counter flag set to 0 means QT
7124 parse_timecode_in_framenum_format(s, st, value, flags);
7126 avio_seek(sc->pb, cur_pos, SEEK_SET);
7130 static void mov_free_encryption_index(MOVEncryptionIndex **index) {
7132 if (!index || !*index) return;
7133 for (i = 0; i < (*index)->nb_encrypted_samples; i++) {
7134 av_encryption_info_free((*index)->encrypted_samples[i]);
7136 av_freep(&(*index)->encrypted_samples);
7137 av_freep(&(*index)->auxiliary_info_sizes);
7138 av_freep(&(*index)->auxiliary_offsets);
7142 static int mov_read_close(AVFormatContext *s)
7144 MOVContext *mov = s->priv_data;
7147 for (i = 0; i < s->nb_streams; i++) {
7148 AVStream *st = s->streams[i];
7149 MOVStreamContext *sc = st->priv_data;
7154 av_freep(&sc->ctts_data);
7155 for (j = 0; j < sc->drefs_count; j++) {
7156 av_freep(&sc->drefs[j].path);
7157 av_freep(&sc->drefs[j].dir);
7159 av_freep(&sc->drefs);
7161 sc->drefs_count = 0;
7163 if (!sc->pb_is_copied)
7164 ff_format_io_close(s, &sc->pb);
7167 av_freep(&sc->chunk_offsets);
7168 av_freep(&sc->stsc_data);
7169 av_freep(&sc->sample_sizes);
7170 av_freep(&sc->keyframes);
7171 av_freep(&sc->stts_data);
7172 av_freep(&sc->stps_data);
7173 av_freep(&sc->elst_data);
7174 av_freep(&sc->rap_group);
7175 av_freep(&sc->display_matrix);
7176 av_freep(&sc->index_ranges);
7179 for (j = 0; j < sc->stsd_count; j++)
7180 av_free(sc->extradata[j]);
7181 av_freep(&sc->extradata);
7182 av_freep(&sc->extradata_size);
7184 mov_free_encryption_index(&sc->cenc.encryption_index);
7185 av_encryption_info_free(sc->cenc.default_encrypted_sample);
7186 av_aes_ctr_free(sc->cenc.aes_ctr);
7188 av_freep(&sc->stereo3d);
7189 av_freep(&sc->spherical);
7190 av_freep(&sc->mastering);
7191 av_freep(&sc->coll);
7194 if (mov->dv_demux) {
7195 avformat_free_context(mov->dv_fctx);
7196 mov->dv_fctx = NULL;
7199 if (mov->meta_keys) {
7200 for (i = 1; i < mov->meta_keys_count; i++) {
7201 av_freep(&mov->meta_keys[i]);
7203 av_freep(&mov->meta_keys);
7206 av_freep(&mov->trex_data);
7207 av_freep(&mov->bitrates);
7209 for (i = 0; i < mov->frag_index.nb_items; i++) {
7210 MOVFragmentStreamInfo *frag = mov->frag_index.item[i].stream_info;
7211 for (j = 0; j < mov->frag_index.item[i].nb_stream_info; j++) {
7212 mov_free_encryption_index(&frag[j].encryption_index);
7214 av_freep(&mov->frag_index.item[i].stream_info);
7216 av_freep(&mov->frag_index.item);
7218 av_freep(&mov->aes_decrypt);
7219 av_freep(&mov->chapter_tracks);
7224 static int tmcd_is_referenced(AVFormatContext *s, int tmcd_id)
7228 for (i = 0; i < s->nb_streams; i++) {
7229 AVStream *st = s->streams[i];
7230 MOVStreamContext *sc = st->priv_data;
7232 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO &&
7233 sc->timecode_track == tmcd_id)
7239 /* look for a tmcd track not referenced by any video track, and export it globally */
7240 static void export_orphan_timecode(AVFormatContext *s)
7244 for (i = 0; i < s->nb_streams; i++) {
7245 AVStream *st = s->streams[i];
7247 if (st->codecpar->codec_tag == MKTAG('t','m','c','d') &&
7248 !tmcd_is_referenced(s, i + 1)) {
7249 AVDictionaryEntry *tcr = av_dict_get(st->metadata, "timecode", NULL, 0);
7251 av_dict_set(&s->metadata, "timecode", tcr->value, 0);
7258 static int read_tfra(MOVContext *mov, AVIOContext *f)
7260 int version, fieldlength, i, j;
7261 int64_t pos = avio_tell(f);
7262 uint32_t size = avio_rb32(f);
7263 unsigned track_id, item_count;
7265 if (avio_rb32(f) != MKBETAG('t', 'f', 'r', 'a')) {
7268 av_log(mov->fc, AV_LOG_VERBOSE, "found tfra\n");
7270 version = avio_r8(f);
7272 track_id = avio_rb32(f);
7273 fieldlength = avio_rb32(f);
7274 item_count = avio_rb32(f);
7275 for (i = 0; i < item_count; i++) {
7276 int64_t time, offset;
7278 MOVFragmentStreamInfo * frag_stream_info;
7281 return AVERROR_INVALIDDATA;
7285 time = avio_rb64(f);
7286 offset = avio_rb64(f);
7288 time = avio_rb32(f);
7289 offset = avio_rb32(f);
7292 // The first sample of each stream in a fragment is always a random
7293 // access sample. So it's entry in the tfra can be used as the
7294 // initial PTS of the fragment.
7295 index = update_frag_index(mov, offset);
7296 frag_stream_info = get_frag_stream_info(&mov->frag_index, index, track_id);
7297 if (frag_stream_info &&
7298 frag_stream_info->first_tfra_pts == AV_NOPTS_VALUE)
7299 frag_stream_info->first_tfra_pts = time;
7301 for (j = 0; j < ((fieldlength >> 4) & 3) + 1; j++)
7303 for (j = 0; j < ((fieldlength >> 2) & 3) + 1; j++)
7305 for (j = 0; j < ((fieldlength >> 0) & 3) + 1; j++)
7309 avio_seek(f, pos + size, SEEK_SET);
7313 static int mov_read_mfra(MOVContext *c, AVIOContext *f)
7315 int64_t stream_size = avio_size(f);
7316 int64_t original_pos = avio_tell(f);
7320 if ((seek_ret = avio_seek(f, stream_size - 4, SEEK_SET)) < 0) {
7324 mfra_size = avio_rb32(f);
7325 if (mfra_size < 0 || mfra_size > stream_size) {
7326 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (unreasonable size)\n");
7329 if ((seek_ret = avio_seek(f, -mfra_size, SEEK_CUR)) < 0) {
7333 if (avio_rb32(f) != mfra_size) {
7334 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (size mismatch)\n");
7337 if (avio_rb32(f) != MKBETAG('m', 'f', 'r', 'a')) {
7338 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (tag mismatch)\n");
7341 av_log(c->fc, AV_LOG_VERBOSE, "stream has mfra\n");
7343 ret = read_tfra(c, f);
7349 seek_ret = avio_seek(f, original_pos, SEEK_SET);
7351 av_log(c->fc, AV_LOG_ERROR,
7352 "failed to seek back after looking for mfra\n");
7358 static int mov_read_header(AVFormatContext *s)
7360 MOVContext *mov = s->priv_data;
7361 AVIOContext *pb = s->pb;
7363 MOVAtom atom = { AV_RL32("root") };
7366 if (mov->decryption_key_len != 0 && mov->decryption_key_len != AES_CTR_KEY_SIZE) {
7367 av_log(s, AV_LOG_ERROR, "Invalid decryption key len %d expected %d\n",
7368 mov->decryption_key_len, AES_CTR_KEY_SIZE);
7369 return AVERROR(EINVAL);
7373 mov->trak_index = -1;
7374 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
7375 if (pb->seekable & AVIO_SEEKABLE_NORMAL)
7376 atom.size = avio_size(pb);
7378 atom.size = INT64_MAX;
7380 /* check MOV header */
7382 if (mov->moov_retry)
7383 avio_seek(pb, 0, SEEK_SET);
7384 if ((err = mov_read_default(mov, pb, atom)) < 0) {
7385 av_log(s, AV_LOG_ERROR, "error reading header\n");
7389 } while ((pb->seekable & AVIO_SEEKABLE_NORMAL) && !mov->found_moov && !mov->moov_retry++);
7390 if (!mov->found_moov) {
7391 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
7393 return AVERROR_INVALIDDATA;
7395 av_log(mov->fc, AV_LOG_TRACE, "on_parse_exit_offset=%"PRId64"\n", avio_tell(pb));
7397 if (pb->seekable & AVIO_SEEKABLE_NORMAL) {
7398 if (mov->nb_chapter_tracks > 0 && !mov->ignore_chapters)
7399 mov_read_chapters(s);
7400 for (i = 0; i < s->nb_streams; i++)
7401 if (s->streams[i]->codecpar->codec_tag == AV_RL32("tmcd")) {
7402 mov_read_timecode_track(s, s->streams[i]);
7403 } else if (s->streams[i]->codecpar->codec_tag == AV_RL32("rtmd")) {
7404 mov_read_rtmd_track(s, s->streams[i]);
7408 /* copy timecode metadata from tmcd tracks to the related video streams */
7409 for (i = 0; i < s->nb_streams; i++) {
7410 AVStream *st = s->streams[i];
7411 MOVStreamContext *sc = st->priv_data;
7412 if (sc->timecode_track > 0) {
7413 AVDictionaryEntry *tcr;
7414 int tmcd_st_id = -1;
7416 for (j = 0; j < s->nb_streams; j++)
7417 if (s->streams[j]->id == sc->timecode_track)
7420 if (tmcd_st_id < 0 || tmcd_st_id == i)
7422 tcr = av_dict_get(s->streams[tmcd_st_id]->metadata, "timecode", NULL, 0);
7424 av_dict_set(&st->metadata, "timecode", tcr->value, 0);
7427 export_orphan_timecode(s);
7429 for (i = 0; i < s->nb_streams; i++) {
7430 AVStream *st = s->streams[i];
7431 MOVStreamContext *sc = st->priv_data;
7432 fix_timescale(mov, sc);
7433 if(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id == AV_CODEC_ID_AAC) {
7434 st->skip_samples = sc->start_pad;
7436 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && sc->nb_frames_for_fps > 0 && sc->duration_for_fps > 0)
7437 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
7438 sc->time_scale*(int64_t)sc->nb_frames_for_fps, sc->duration_for_fps, INT_MAX);
7439 if (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE) {
7440 if (st->codecpar->width <= 0 || st->codecpar->height <= 0) {
7441 st->codecpar->width = sc->width;
7442 st->codecpar->height = sc->height;
7444 if (st->codecpar->codec_id == AV_CODEC_ID_DVD_SUBTITLE) {
7445 if ((err = mov_rewrite_dvd_sub_extradata(st)) < 0)
7449 if (mov->handbrake_version &&
7450 mov->handbrake_version <= 1000000*0 + 1000*10 + 2 && // 0.10.2
7451 st->codecpar->codec_id == AV_CODEC_ID_MP3
7453 av_log(s, AV_LOG_VERBOSE, "Forcing full parsing for mp3 stream\n");
7454 st->need_parsing = AVSTREAM_PARSE_FULL;
7458 if (mov->trex_data) {
7459 for (i = 0; i < s->nb_streams; i++) {
7460 AVStream *st = s->streams[i];
7461 MOVStreamContext *sc = st->priv_data;
7462 if (st->duration > 0) {
7463 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
7464 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
7465 sc->data_size, sc->time_scale);
7467 return AVERROR_INVALIDDATA;
7469 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale / st->duration;
7474 if (mov->use_mfra_for > 0) {
7475 for (i = 0; i < s->nb_streams; i++) {
7476 AVStream *st = s->streams[i];
7477 MOVStreamContext *sc = st->priv_data;
7478 if (sc->duration_for_fps > 0) {
7479 if (sc->data_size > INT64_MAX / sc->time_scale / 8) {
7480 av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %"PRId64" * 8 * %d\n",
7481 sc->data_size, sc->time_scale);
7483 return AVERROR_INVALIDDATA;
7485 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale /
7486 sc->duration_for_fps;
7491 for (i = 0; i < mov->bitrates_count && i < s->nb_streams; i++) {
7492 if (mov->bitrates[i]) {
7493 s->streams[i]->codecpar->bit_rate = mov->bitrates[i];
7497 ff_rfps_calculate(s);
7499 for (i = 0; i < s->nb_streams; i++) {
7500 AVStream *st = s->streams[i];
7501 MOVStreamContext *sc = st->priv_data;
7503 switch (st->codecpar->codec_type) {
7504 case AVMEDIA_TYPE_AUDIO:
7505 err = ff_replaygain_export(st, s->metadata);
7511 case AVMEDIA_TYPE_VIDEO:
7512 if (sc->display_matrix) {
7513 err = av_stream_add_side_data(st, AV_PKT_DATA_DISPLAYMATRIX, (uint8_t*)sc->display_matrix,
7514 sizeof(int32_t) * 9);
7518 sc->display_matrix = NULL;
7521 err = av_stream_add_side_data(st, AV_PKT_DATA_STEREO3D,
7522 (uint8_t *)sc->stereo3d,
7523 sizeof(*sc->stereo3d));
7527 sc->stereo3d = NULL;
7529 if (sc->spherical) {
7530 err = av_stream_add_side_data(st, AV_PKT_DATA_SPHERICAL,
7531 (uint8_t *)sc->spherical,
7532 sc->spherical_size);
7536 sc->spherical = NULL;
7538 if (sc->mastering) {
7539 err = av_stream_add_side_data(st, AV_PKT_DATA_MASTERING_DISPLAY_METADATA,
7540 (uint8_t *)sc->mastering,
7541 sizeof(*sc->mastering));
7545 sc->mastering = NULL;
7548 err = av_stream_add_side_data(st, AV_PKT_DATA_CONTENT_LIGHT_LEVEL,
7549 (uint8_t *)sc->coll,
7559 ff_configure_buffers_for_index(s, AV_TIME_BASE);
7561 for (i = 0; i < mov->frag_index.nb_items; i++)
7562 if (mov->frag_index.item[i].moof_offset <= mov->fragment.moof_offset)
7563 mov->frag_index.item[i].headers_read = 1;
7568 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
7570 AVIndexEntry *sample = NULL;
7571 int64_t best_dts = INT64_MAX;
7573 for (i = 0; i < s->nb_streams; i++) {
7574 AVStream *avst = s->streams[i];
7575 MOVStreamContext *msc = avst->priv_data;
7576 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
7577 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
7578 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
7579 av_log(s, AV_LOG_TRACE, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
7580 if (!sample || (!(s->pb->seekable & AVIO_SEEKABLE_NORMAL) && current_sample->pos < sample->pos) ||
7581 ((s->pb->seekable & AVIO_SEEKABLE_NORMAL) &&
7582 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
7583 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
7584 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
7585 sample = current_sample;
7594 static int should_retry(AVIOContext *pb, int error_code) {
7595 if (error_code == AVERROR_EOF || avio_feof(pb))
7601 static int mov_switch_root(AVFormatContext *s, int64_t target, int index)
7604 MOVContext *mov = s->priv_data;
7606 if (index >= 0 && index < mov->frag_index.nb_items)
7607 target = mov->frag_index.item[index].moof_offset;
7608 if (avio_seek(s->pb, target, SEEK_SET) != target) {
7609 av_log(mov->fc, AV_LOG_ERROR, "root atom offset 0x%"PRIx64": partial file\n", target);
7610 return AVERROR_INVALIDDATA;
7613 mov->next_root_atom = 0;
7614 if (index < 0 || index >= mov->frag_index.nb_items)
7615 index = search_frag_moof_offset(&mov->frag_index, target);
7616 if (index < mov->frag_index.nb_items) {
7617 if (index + 1 < mov->frag_index.nb_items)
7618 mov->next_root_atom = mov->frag_index.item[index + 1].moof_offset;
7619 if (mov->frag_index.item[index].headers_read)
7621 mov->frag_index.item[index].headers_read = 1;
7624 mov->found_mdat = 0;
7626 ret = mov_read_default(mov, s->pb, (MOVAtom){ AV_RL32("root"), INT64_MAX });
7629 if (avio_feof(s->pb))
7631 av_log(s, AV_LOG_TRACE, "read fragments, offset 0x%"PRIx64"\n", avio_tell(s->pb));
7636 static int mov_change_extradata(MOVStreamContext *sc, AVPacket *pkt)
7638 uint8_t *side, *extradata;
7641 /* Save the current index. */
7642 sc->last_stsd_index = sc->stsc_data[sc->stsc_index].id - 1;
7644 /* Notify the decoder that extradata changed. */
7645 extradata_size = sc->extradata_size[sc->last_stsd_index];
7646 extradata = sc->extradata[sc->last_stsd_index];
7647 if (extradata_size > 0 && extradata) {
7648 side = av_packet_new_side_data(pkt,
7649 AV_PKT_DATA_NEW_EXTRADATA,
7652 return AVERROR(ENOMEM);
7653 memcpy(side, extradata, extradata_size);
7659 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
7661 MOVContext *mov = s->priv_data;
7662 MOVStreamContext *sc;
7663 AVIndexEntry *sample;
7664 AVStream *st = NULL;
7665 int64_t current_index;
7669 sample = mov_find_next_sample(s, &st);
7670 if (!sample || (mov->next_root_atom && sample->pos > mov->next_root_atom)) {
7671 if (!mov->next_root_atom)
7673 if ((ret = mov_switch_root(s, mov->next_root_atom, -1)) < 0)
7678 /* must be done just before reading, to avoid infinite loop on sample */
7679 current_index = sc->current_index;
7680 mov_current_sample_inc(sc);
7682 if (mov->next_root_atom) {
7683 sample->pos = FFMIN(sample->pos, mov->next_root_atom);
7684 sample->size = FFMIN(sample->size, (mov->next_root_atom - sample->pos));
7687 if (st->discard != AVDISCARD_ALL) {
7688 int64_t ret64 = avio_seek(sc->pb, sample->pos, SEEK_SET);
7689 if (ret64 != sample->pos) {
7690 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
7691 sc->ffindex, sample->pos);
7692 if (should_retry(sc->pb, ret64)) {
7693 mov_current_sample_dec(sc);
7695 return AVERROR_INVALIDDATA;
7698 if( st->discard == AVDISCARD_NONKEY && 0==(sample->flags & AVINDEX_KEYFRAME) ) {
7699 av_log(mov->fc, AV_LOG_DEBUG, "Nonkey frame from stream %d discarded due to AVDISCARD_NONKEY\n", sc->ffindex);
7703 ret = av_get_packet(sc->pb, pkt, sample->size);
7705 if (should_retry(sc->pb, ret)) {
7706 mov_current_sample_dec(sc);
7710 if (sc->has_palette) {
7713 pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, AVPALETTE_SIZE);
7715 av_log(mov->fc, AV_LOG_ERROR, "Cannot append palette to packet\n");
7717 memcpy(pal, sc->palette, AVPALETTE_SIZE);
7718 sc->has_palette = 0;
7721 #if CONFIG_DV_DEMUXER
7722 if (mov->dv_demux && sc->dv_audio_container) {
7723 avpriv_dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size, pkt->pos);
7724 av_freep(&pkt->data);
7726 ret = avpriv_dv_get_packet(mov->dv_demux, pkt);
7731 if (st->codecpar->codec_id == AV_CODEC_ID_MP3 && !st->need_parsing && pkt->size > 4) {
7732 if (ff_mpa_check_header(AV_RB32(pkt->data)) < 0)
7733 st->need_parsing = AVSTREAM_PARSE_FULL;
7737 pkt->stream_index = sc->ffindex;
7738 pkt->dts = sample->timestamp;
7739 if (sample->flags & AVINDEX_DISCARD_FRAME) {
7740 pkt->flags |= AV_PKT_FLAG_DISCARD;
7742 if (sc->ctts_data && sc->ctts_index < sc->ctts_count) {
7743 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
7744 /* update ctts context */
7746 if (sc->ctts_index < sc->ctts_count &&
7747 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
7749 sc->ctts_sample = 0;
7752 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
7753 st->index_entries[sc->current_sample].timestamp : st->duration;
7755 if (next_dts >= pkt->dts)
7756 pkt->duration = next_dts - pkt->dts;
7757 pkt->pts = pkt->dts;
7759 if (st->discard == AVDISCARD_ALL)
7761 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? AV_PKT_FLAG_KEY : 0;
7762 pkt->pos = sample->pos;
7764 /* Multiple stsd handling. */
7765 if (sc->stsc_data) {
7766 /* Keep track of the stsc index for the given sample, then check
7767 * if the stsd index is different from the last used one. */
7769 if (mov_stsc_index_valid(sc->stsc_index, sc->stsc_count) &&
7770 mov_get_stsc_samples(sc, sc->stsc_index) == sc->stsc_sample) {
7772 sc->stsc_sample = 0;
7773 /* Do not check indexes after a switch. */
7774 } else if (sc->stsc_data[sc->stsc_index].id > 0 &&
7775 sc->stsc_data[sc->stsc_index].id - 1 < sc->stsd_count &&
7776 sc->stsc_data[sc->stsc_index].id - 1 != sc->last_stsd_index) {
7777 ret = mov_change_extradata(sc, pkt);
7784 aax_filter(pkt->data, pkt->size, mov);
7786 ret = cenc_filter(mov, sc, pkt, current_index);
7793 static int mov_seek_fragment(AVFormatContext *s, AVStream *st, int64_t timestamp)
7795 MOVContext *mov = s->priv_data;
7798 if (!mov->frag_index.complete)
7801 index = search_frag_timestamp(&mov->frag_index, st, timestamp);
7804 if (!mov->frag_index.item[index].headers_read)
7805 return mov_switch_root(s, -1, index);
7806 if (index + 1 < mov->frag_index.nb_items)
7807 mov->next_root_atom = mov->frag_index.item[index + 1].moof_offset;
7812 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
7814 MOVStreamContext *sc = st->priv_data;
7815 int sample, time_sample, ret;
7818 // Here we consider timestamp to be PTS, hence try to offset it so that we
7819 // can search over the DTS timeline.
7820 timestamp -= (sc->min_corrected_pts + sc->dts_shift);
7822 ret = mov_seek_fragment(s, st, timestamp);
7826 sample = av_index_search_timestamp(st, timestamp, flags);
7827 av_log(s, AV_LOG_TRACE, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
7828 if (sample < 0 && st->nb_index_entries && timestamp < st->index_entries[0].timestamp)
7830 if (sample < 0) /* not sure what to do */
7831 return AVERROR_INVALIDDATA;
7832 mov_current_sample_set(sc, sample);
7833 av_log(s, AV_LOG_TRACE, "stream %d, found sample %d\n", st->index, sc->current_sample);
7834 /* adjust ctts index */
7835 if (sc->ctts_data) {
7837 for (i = 0; i < sc->ctts_count; i++) {
7838 int next = time_sample + sc->ctts_data[i].count;
7839 if (next > sc->current_sample) {
7841 sc->ctts_sample = sc->current_sample - time_sample;
7848 /* adjust stsd index */
7850 for (i = 0; i < sc->stsc_count; i++) {
7851 int64_t next = time_sample + mov_get_stsc_samples(sc, i);
7852 if (next > sc->current_sample) {
7854 sc->stsc_sample = sc->current_sample - time_sample;
7857 av_assert0(next == (int)next);
7864 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
7866 MOVContext *mc = s->priv_data;
7871 if (stream_index >= s->nb_streams)
7872 return AVERROR_INVALIDDATA;
7874 st = s->streams[stream_index];
7875 sample = mov_seek_stream(s, st, sample_time, flags);
7879 if (mc->seek_individually) {
7880 /* adjust seek timestamp to found sample timestamp */
7881 int64_t seek_timestamp = st->index_entries[sample].timestamp;
7883 for (i = 0; i < s->nb_streams; i++) {
7885 MOVStreamContext *sc = s->streams[i]->priv_data;
7887 st->skip_samples = (sample_time <= 0) ? sc->start_pad : 0;
7889 if (stream_index == i)
7892 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
7893 mov_seek_stream(s, st, timestamp, flags);
7896 for (i = 0; i < s->nb_streams; i++) {
7897 MOVStreamContext *sc;
7900 mov_current_sample_set(sc, 0);
7903 MOVStreamContext *sc;
7904 AVIndexEntry *entry = mov_find_next_sample(s, &st);
7906 return AVERROR_INVALIDDATA;
7908 if (sc->ffindex == stream_index && sc->current_sample == sample)
7910 mov_current_sample_inc(sc);
7916 #define OFFSET(x) offsetof(MOVContext, x)
7917 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
7918 static const AVOption mov_options[] = {
7919 {"use_absolute_path",
7920 "allow using absolute path when opening alias, this is a possible security issue",
7921 OFFSET(use_absolute_path), AV_OPT_TYPE_BOOL, {.i64 = 0},
7923 {"seek_streams_individually",
7924 "Seek each stream individually to the to the closest point",
7925 OFFSET(seek_individually), AV_OPT_TYPE_BOOL, { .i64 = 1 },
7927 {"ignore_editlist", "Ignore the edit list atom.", OFFSET(ignore_editlist), AV_OPT_TYPE_BOOL, {.i64 = 0},
7929 {"advanced_editlist",
7930 "Modify the AVIndex according to the editlists. Use this option to decode in the order specified by the edits.",
7931 OFFSET(advanced_editlist), AV_OPT_TYPE_BOOL, {.i64 = 1},
7933 {"ignore_chapters", "", OFFSET(ignore_chapters), AV_OPT_TYPE_BOOL, {.i64 = 0},
7936 "use mfra for fragment timestamps",
7937 OFFSET(use_mfra_for), AV_OPT_TYPE_INT, {.i64 = FF_MOV_FLAG_MFRA_AUTO},
7938 -1, FF_MOV_FLAG_MFRA_PTS, FLAGS,
7940 {"auto", "auto", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_AUTO}, 0, 0,
7941 FLAGS, "use_mfra_for" },
7942 {"dts", "dts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_DTS}, 0, 0,
7943 FLAGS, "use_mfra_for" },
7944 {"pts", "pts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_PTS}, 0, 0,
7945 FLAGS, "use_mfra_for" },
7946 { "export_all", "Export unrecognized metadata entries", OFFSET(export_all),
7947 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
7948 { "export_xmp", "Export full XMP metadata", OFFSET(export_xmp),
7949 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
7950 { "activation_bytes", "Secret bytes for Audible AAX files", OFFSET(activation_bytes),
7951 AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
7952 { "audible_fixed_key", // extracted from libAAX_SDK.so and AAXSDKWin.dll files!
7953 "Fixed key used for handling Audible AAX files", OFFSET(audible_fixed_key),
7954 AV_OPT_TYPE_BINARY, {.str="77214d4b196a87cd520045fd20a51d67"},
7955 .flags = AV_OPT_FLAG_DECODING_PARAM },
7956 { "decryption_key", "The media decryption key (hex)", OFFSET(decryption_key), AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
7957 { "enable_drefs", "Enable external track support.", OFFSET(enable_drefs), AV_OPT_TYPE_BOOL,
7958 {.i64 = 0}, 0, 1, FLAGS },
7963 static const AVClass mov_class = {
7964 .class_name = "mov,mp4,m4a,3gp,3g2,mj2",
7965 .item_name = av_default_item_name,
7966 .option = mov_options,
7967 .version = LIBAVUTIL_VERSION_INT,
7970 AVInputFormat ff_mov_demuxer = {
7971 .name = "mov,mp4,m4a,3gp,3g2,mj2",
7972 .long_name = NULL_IF_CONFIG_SMALL("QuickTime / MOV"),
7973 .priv_class = &mov_class,
7974 .priv_data_size = sizeof(MOVContext),
7975 .extensions = "mov,mp4,m4a,3gp,3g2,mj2",
7976 .read_probe = mov_probe,
7977 .read_header = mov_read_header,
7978 .read_packet = mov_read_packet,
7979 .read_close = mov_read_close,
7980 .read_seek = mov_read_seek,
7981 .flags = AVFMT_NO_BYTE_SEEK,