3 * Copyright (c) 2001 Fabrice Bellard
4 * Copyright (c) 2009 Baptiste Coudurier <baptiste dot coudurier at gmail dot com>
6 * first version by Francois Revol <revol@free.fr>
7 * seek function by Gael Chardon <gael.dev@4now.net>
9 * This file is part of FFmpeg.
11 * FFmpeg is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public
13 * License as published by the Free Software Foundation; either
14 * version 2.1 of the License, or (at your option) any later version.
16 * FFmpeg is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with FFmpeg; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
30 #include "libavutil/attributes.h"
31 #include "libavutil/channel_layout.h"
32 #include "libavutil/internal.h"
33 #include "libavutil/intreadwrite.h"
34 #include "libavutil/intfloat.h"
35 #include "libavutil/mathematics.h"
36 #include "libavutil/time_internal.h"
37 #include "libavutil/avassert.h"
38 #include "libavutil/avstring.h"
39 #include "libavutil/dict.h"
40 #include "libavutil/display.h"
41 #include "libavutil/opt.h"
42 #include "libavutil/aes.h"
43 #include "libavutil/aes_ctr.h"
44 #include "libavutil/sha.h"
45 #include "libavutil/timecode.h"
46 #include "libavcodec/ac3tab.h"
47 #include "libavcodec/mpegaudiodecheader.h"
50 #include "avio_internal.h"
53 #include "libavcodec/get_bits.h"
56 #include "replaygain.h"
62 #include "qtpalette.h"
64 /* those functions parse an atom */
65 /* links atom IDs to parse functions */
66 typedef struct MOVParseTableEntry {
68 int (*parse)(MOVContext *ctx, AVIOContext *pb, MOVAtom atom);
71 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom);
72 static int mov_read_mfra(MOVContext *c, AVIOContext *f);
74 static int mov_metadata_track_or_disc_number(MOVContext *c, AVIOContext *pb,
75 unsigned len, const char *key)
79 short current, total = 0;
80 avio_rb16(pb); // unknown
81 current = avio_rb16(pb);
83 total = avio_rb16(pb);
85 snprintf(buf, sizeof(buf), "%d", current);
87 snprintf(buf, sizeof(buf), "%d/%d", current, total);
88 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
89 av_dict_set(&c->fc->metadata, key, buf, 0);
94 static int mov_metadata_int8_bypass_padding(MOVContext *c, AVIOContext *pb,
95 unsigned len, const char *key)
97 /* bypass padding bytes */
102 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
103 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
108 static int mov_metadata_int8_no_padding(MOVContext *c, AVIOContext *pb,
109 unsigned len, const char *key)
111 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
112 av_dict_set_int(&c->fc->metadata, key, avio_r8(pb), 0);
117 static int mov_metadata_gnre(MOVContext *c, AVIOContext *pb,
118 unsigned len, const char *key)
122 avio_r8(pb); // unknown
125 if (genre < 1 || genre > ID3v1_GENRE_MAX)
127 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
128 av_dict_set(&c->fc->metadata, key, ff_id3v1_genre_str[genre-1], 0);
133 static const uint32_t mac_to_unicode[128] = {
134 0x00C4,0x00C5,0x00C7,0x00C9,0x00D1,0x00D6,0x00DC,0x00E1,
135 0x00E0,0x00E2,0x00E4,0x00E3,0x00E5,0x00E7,0x00E9,0x00E8,
136 0x00EA,0x00EB,0x00ED,0x00EC,0x00EE,0x00EF,0x00F1,0x00F3,
137 0x00F2,0x00F4,0x00F6,0x00F5,0x00FA,0x00F9,0x00FB,0x00FC,
138 0x2020,0x00B0,0x00A2,0x00A3,0x00A7,0x2022,0x00B6,0x00DF,
139 0x00AE,0x00A9,0x2122,0x00B4,0x00A8,0x2260,0x00C6,0x00D8,
140 0x221E,0x00B1,0x2264,0x2265,0x00A5,0x00B5,0x2202,0x2211,
141 0x220F,0x03C0,0x222B,0x00AA,0x00BA,0x03A9,0x00E6,0x00F8,
142 0x00BF,0x00A1,0x00AC,0x221A,0x0192,0x2248,0x2206,0x00AB,
143 0x00BB,0x2026,0x00A0,0x00C0,0x00C3,0x00D5,0x0152,0x0153,
144 0x2013,0x2014,0x201C,0x201D,0x2018,0x2019,0x00F7,0x25CA,
145 0x00FF,0x0178,0x2044,0x20AC,0x2039,0x203A,0xFB01,0xFB02,
146 0x2021,0x00B7,0x201A,0x201E,0x2030,0x00C2,0x00CA,0x00C1,
147 0x00CB,0x00C8,0x00CD,0x00CE,0x00CF,0x00CC,0x00D3,0x00D4,
148 0xF8FF,0x00D2,0x00DA,0x00DB,0x00D9,0x0131,0x02C6,0x02DC,
149 0x00AF,0x02D8,0x02D9,0x02DA,0x00B8,0x02DD,0x02DB,0x02C7,
152 static int mov_read_mac_string(MOVContext *c, AVIOContext *pb, int len,
153 char *dst, int dstlen)
156 char *end = dst+dstlen-1;
159 for (i = 0; i < len; i++) {
160 uint8_t t, c = avio_r8(pb);
161 if (c < 0x80 && p < end)
164 PUT_UTF8(mac_to_unicode[c-0x80], t, if (p < end) *p++ = t;);
170 static int mov_read_covr(MOVContext *c, AVIOContext *pb, int type, int len)
174 MOVStreamContext *sc;
179 case 0xd: id = AV_CODEC_ID_MJPEG; break;
180 case 0xe: id = AV_CODEC_ID_PNG; break;
181 case 0x1b: id = AV_CODEC_ID_BMP; break;
183 av_log(c->fc, AV_LOG_WARNING, "Unknown cover type: 0x%x.\n", type);
188 st = avformat_new_stream(c->fc, NULL);
190 return AVERROR(ENOMEM);
191 sc = av_mallocz(sizeof(*sc));
193 return AVERROR(ENOMEM);
196 ret = av_get_packet(pb, &pkt, len);
200 if (pkt.size >= 8 && id != AV_CODEC_ID_BMP) {
201 if (AV_RB64(pkt.data) == 0x89504e470d0a1a0a) {
202 id = AV_CODEC_ID_PNG;
204 id = AV_CODEC_ID_MJPEG;
208 st->disposition |= AV_DISPOSITION_ATTACHED_PIC;
210 st->attached_pic = pkt;
211 st->attached_pic.stream_index = st->index;
212 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
214 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
215 st->codecpar->codec_id = id;
221 static int mov_metadata_loci(MOVContext *c, AVIOContext *pb, unsigned len)
223 char language[4] = { 0 };
224 char buf[200], place[100];
225 uint16_t langcode = 0;
226 double longitude, latitude, altitude;
227 const char *key = "location";
229 if (len < 4 + 2 + 1 + 1 + 4 + 4 + 4) {
230 av_log(c->fc, AV_LOG_ERROR, "loci too short\n");
231 return AVERROR_INVALIDDATA;
234 avio_skip(pb, 4); // version+flags
235 langcode = avio_rb16(pb);
236 ff_mov_lang_to_iso639(langcode, language);
239 len -= avio_get_str(pb, len, place, sizeof(place));
241 av_log(c->fc, AV_LOG_ERROR, "place name too long\n");
242 return AVERROR_INVALIDDATA;
244 avio_skip(pb, 1); // role
248 av_log(c->fc, AV_LOG_ERROR,
249 "loci too short (%u bytes left, need at least %d)\n", len, 12);
250 return AVERROR_INVALIDDATA;
252 longitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
253 latitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
254 altitude = ((int32_t) avio_rb32(pb)) / (float) (1 << 16);
256 // Try to output in the same format as the ?xyz field
257 snprintf(buf, sizeof(buf), "%+08.4f%+09.4f", latitude, longitude);
259 av_strlcatf(buf, sizeof(buf), "%+f", altitude);
260 av_strlcatf(buf, sizeof(buf), "/%s", place);
262 if (*language && strcmp(language, "und")) {
264 snprintf(key2, sizeof(key2), "%s-%s", key, language);
265 av_dict_set(&c->fc->metadata, key2, buf, 0);
267 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
268 return av_dict_set(&c->fc->metadata, key, buf, 0);
271 static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom)
274 char key2[32], language[4] = {0};
276 const char *key = NULL;
277 uint16_t langcode = 0;
278 uint32_t data_type = 0, str_size, str_size_alloc;
279 int (*parse)(MOVContext*, AVIOContext*, unsigned, const char*) = NULL;
284 case MKTAG( '@','P','R','M'): key = "premiere_version"; raw = 1; break;
285 case MKTAG( '@','P','R','Q'): key = "quicktime_version"; raw = 1; break;
286 case MKTAG( 'X','M','P','_'):
287 if (c->export_xmp) { key = "xmp"; raw = 1; } break;
288 case MKTAG( 'a','A','R','T'): key = "album_artist"; break;
289 case MKTAG( 'a','k','I','D'): key = "account_type";
290 parse = mov_metadata_int8_no_padding; break;
291 case MKTAG( 'a','p','I','D'): key = "account_id"; break;
292 case MKTAG( 'c','a','t','g'): key = "category"; break;
293 case MKTAG( 'c','p','i','l'): key = "compilation";
294 parse = mov_metadata_int8_no_padding; break;
295 case MKTAG( 'c','p','r','t'): key = "copyright"; break;
296 case MKTAG( 'd','e','s','c'): key = "description"; break;
297 case MKTAG( 'd','i','s','k'): key = "disc";
298 parse = mov_metadata_track_or_disc_number; break;
299 case MKTAG( 'e','g','i','d'): key = "episode_uid";
300 parse = mov_metadata_int8_no_padding; break;
301 case MKTAG( 'g','n','r','e'): key = "genre";
302 parse = mov_metadata_gnre; break;
303 case MKTAG( 'h','d','v','d'): key = "hd_video";
304 parse = mov_metadata_int8_no_padding; break;
305 case MKTAG( 'k','e','y','w'): key = "keywords"; break;
306 case MKTAG( 'l','d','e','s'): key = "synopsis"; break;
307 case MKTAG( 'l','o','c','i'):
308 return mov_metadata_loci(c, pb, atom.size);
309 case MKTAG( 'p','c','s','t'): key = "podcast";
310 parse = mov_metadata_int8_no_padding; break;
311 case MKTAG( 'p','g','a','p'): key = "gapless_playback";
312 parse = mov_metadata_int8_no_padding; break;
313 case MKTAG( 'p','u','r','d'): key = "purchase_date"; break;
314 case MKTAG( 'r','t','n','g'): key = "rating";
315 parse = mov_metadata_int8_no_padding; break;
316 case MKTAG( 's','o','a','a'): key = "sort_album_artist"; break;
317 case MKTAG( 's','o','a','l'): key = "sort_album"; break;
318 case MKTAG( 's','o','a','r'): key = "sort_artist"; break;
319 case MKTAG( 's','o','c','o'): key = "sort_composer"; break;
320 case MKTAG( 's','o','n','m'): key = "sort_name"; break;
321 case MKTAG( 's','o','s','n'): key = "sort_show"; break;
322 case MKTAG( 's','t','i','k'): key = "media_type";
323 parse = mov_metadata_int8_no_padding; break;
324 case MKTAG( 't','r','k','n'): key = "track";
325 parse = mov_metadata_track_or_disc_number; break;
326 case MKTAG( 't','v','e','n'): key = "episode_id"; break;
327 case MKTAG( 't','v','e','s'): key = "episode_sort";
328 parse = mov_metadata_int8_bypass_padding; break;
329 case MKTAG( 't','v','n','n'): key = "network"; break;
330 case MKTAG( 't','v','s','h'): key = "show"; break;
331 case MKTAG( 't','v','s','n'): key = "season_number";
332 parse = mov_metadata_int8_bypass_padding; break;
333 case MKTAG(0xa9,'A','R','T'): key = "artist"; break;
334 case MKTAG(0xa9,'P','R','D'): key = "producer"; break;
335 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
336 case MKTAG(0xa9,'a','u','t'): key = "artist"; break;
337 case MKTAG(0xa9,'c','h','p'): key = "chapter"; break;
338 case MKTAG(0xa9,'c','m','t'): key = "comment"; break;
339 case MKTAG(0xa9,'c','o','m'): key = "composer"; break;
340 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
341 case MKTAG(0xa9,'d','a','y'): key = "date"; break;
342 case MKTAG(0xa9,'d','i','r'): key = "director"; break;
343 case MKTAG(0xa9,'d','i','s'): key = "disclaimer"; break;
344 case MKTAG(0xa9,'e','d','1'): key = "edit_date"; break;
345 case MKTAG(0xa9,'e','n','c'): key = "encoder"; break;
346 case MKTAG(0xa9,'f','m','t'): key = "original_format"; break;
347 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
348 case MKTAG(0xa9,'g','r','p'): key = "grouping"; break;
349 case MKTAG(0xa9,'h','s','t'): key = "host_computer"; break;
350 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
351 case MKTAG(0xa9,'l','y','r'): key = "lyrics"; break;
352 case MKTAG(0xa9,'m','a','k'): key = "make"; break;
353 case MKTAG(0xa9,'m','o','d'): key = "model"; break;
354 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
355 case MKTAG(0xa9,'o','p','e'): key = "original_artist"; break;
356 case MKTAG(0xa9,'p','r','d'): key = "producer"; break;
357 case MKTAG(0xa9,'p','r','f'): key = "performers"; break;
358 case MKTAG(0xa9,'r','e','q'): key = "playback_requirements"; break;
359 case MKTAG(0xa9,'s','r','c'): key = "original_source"; break;
360 case MKTAG(0xa9,'s','t','3'): key = "subtitle"; break;
361 case MKTAG(0xa9,'s','w','r'): key = "encoder"; break;
362 case MKTAG(0xa9,'t','o','o'): key = "encoder"; break;
363 case MKTAG(0xa9,'t','r','k'): key = "track"; break;
364 case MKTAG(0xa9,'u','r','l'): key = "URL"; break;
365 case MKTAG(0xa9,'w','r','n'): key = "warning"; break;
366 case MKTAG(0xa9,'w','r','t'): key = "composer"; break;
367 case MKTAG(0xa9,'x','y','z'): key = "location"; break;
370 if (c->itunes_metadata && atom.size > 8) {
371 int data_size = avio_rb32(pb);
372 int tag = avio_rl32(pb);
373 if (tag == MKTAG('d','a','t','a') && data_size <= atom.size) {
374 data_type = avio_rb32(pb); // type
375 avio_rb32(pb); // unknown
376 str_size = data_size - 16;
379 if (atom.type == MKTAG('c', 'o', 'v', 'r')) {
380 int ret = mov_read_covr(c, pb, data_type, str_size);
382 av_log(c->fc, AV_LOG_ERROR, "Error parsing cover art.\n");
385 } else if (!key && c->found_hdlr_mdta && c->meta_keys) {
386 uint32_t index = AV_RB32(&atom.type);
387 if (index < c->meta_keys_count) {
388 key = c->meta_keys[index];
390 av_log(c->fc, AV_LOG_WARNING,
391 "The index of 'data' is out of range: %d >= %d.\n",
392 index, c->meta_keys_count);
396 } else if (atom.size > 4 && key && !c->itunes_metadata && !raw) {
397 str_size = avio_rb16(pb); // string length
398 if (str_size > atom.size) {
400 avio_seek(pb, -2, SEEK_CUR);
401 av_log(c->fc, AV_LOG_WARNING, "UDTA parsing failed retrying raw\n");
404 langcode = avio_rb16(pb);
405 ff_mov_lang_to_iso639(langcode, language);
408 str_size = atom.size;
410 if (c->export_all && !key) {
411 snprintf(tmp_key, 5, "%.4s", (char*)&atom.type);
417 if (atom.size < 0 || str_size >= INT_MAX/2)
418 return AVERROR_INVALIDDATA;
420 // Allocates enough space if data_type is a int32 or float32 number, otherwise
421 // worst-case requirement for output string in case of utf8 coded input
422 num = (data_type >= 21 && data_type <= 23);
423 str_size_alloc = (num ? 512 : (raw ? str_size : str_size * 2)) + 1;
424 str = av_mallocz(str_size_alloc);
426 return AVERROR(ENOMEM);
429 parse(c, pb, str_size, key);
431 if (!raw && (data_type == 3 || (data_type == 0 && (langcode < 0x400 || langcode == 0x7fff)))) { // MAC Encoded
432 mov_read_mac_string(c, pb, str_size, str, str_size_alloc);
433 } else if (data_type == 21) { // BE signed integer, variable size
436 val = (int8_t)avio_r8(pb);
437 else if (str_size == 2)
438 val = (int16_t)avio_rb16(pb);
439 else if (str_size == 3)
440 val = ((int32_t)(avio_rb24(pb)<<8))>>8;
441 else if (str_size == 4)
442 val = (int32_t)avio_rb32(pb);
443 if (snprintf(str, str_size_alloc, "%d", val) >= str_size_alloc) {
444 av_log(c->fc, AV_LOG_ERROR,
445 "Failed to store the number (%d) in string.\n", val);
447 return AVERROR_INVALIDDATA;
449 } else if (data_type == 22) { // BE unsigned integer, variable size
450 unsigned int val = 0;
453 else if (str_size == 2)
455 else if (str_size == 3)
457 else if (str_size == 4)
459 if (snprintf(str, str_size_alloc, "%u", val) >= str_size_alloc) {
460 av_log(c->fc, AV_LOG_ERROR,
461 "Failed to store the number (%u) in string.\n", val);
463 return AVERROR_INVALIDDATA;
465 } else if (data_type == 23 && str_size >= 4) { // BE float32
466 float val = av_int2float(avio_rb32(pb));
467 if (snprintf(str, str_size_alloc, "%f", val) >= str_size_alloc) {
468 av_log(c->fc, AV_LOG_ERROR,
469 "Failed to store the float32 number (%f) in string.\n", val);
471 return AVERROR_INVALIDDATA;
474 int ret = ffio_read_size(pb, str, str_size);
481 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
482 av_dict_set(&c->fc->metadata, key, str, 0);
483 if (*language && strcmp(language, "und")) {
484 snprintf(key2, sizeof(key2), "%s-%s", key, language);
485 av_dict_set(&c->fc->metadata, key2, str, 0);
487 if (!strcmp(key, "encoder")) {
488 int major, minor, micro;
489 if (sscanf(str, "HandBrake %d.%d.%d", &major, &minor, µ) == 3) {
490 c->handbrake_version = 1000000*major + 1000*minor + micro;
494 av_log(c->fc, AV_LOG_TRACE, "lang \"%3s\" ", language);
495 av_log(c->fc, AV_LOG_TRACE, "tag \"%s\" value \"%s\" atom \"%.4s\" %d %"PRId64"\n",
496 key, str, (char*)&atom.type, str_size_alloc, atom.size);
502 static int mov_read_chpl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
505 int i, nb_chapters, str_len, version;
509 if (c->ignore_chapters)
512 if ((atom.size -= 5) < 0)
515 version = avio_r8(pb);
518 avio_rb32(pb); // ???
519 nb_chapters = avio_r8(pb);
521 for (i = 0; i < nb_chapters; i++) {
525 start = avio_rb64(pb);
526 str_len = avio_r8(pb);
528 if ((atom.size -= 9+str_len) < 0)
531 ret = ffio_read_size(pb, str, str_len);
535 avpriv_new_chapter(c->fc, i, (AVRational){1,10000000}, start, AV_NOPTS_VALUE, str);
540 #define MIN_DATA_ENTRY_BOX_SIZE 12
541 static int mov_read_dref(MOVContext *c, AVIOContext *pb, MOVAtom atom)
544 MOVStreamContext *sc;
547 if (c->fc->nb_streams < 1)
549 st = c->fc->streams[c->fc->nb_streams-1];
552 avio_rb32(pb); // version + flags
553 entries = avio_rb32(pb);
555 entries > (atom.size - 1) / MIN_DATA_ENTRY_BOX_SIZE + 1 ||
556 entries >= UINT_MAX / sizeof(*sc->drefs))
557 return AVERROR_INVALIDDATA;
561 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
563 return AVERROR(ENOMEM);
564 sc->drefs_count = entries;
566 for (i = 0; i < entries; i++) {
567 MOVDref *dref = &sc->drefs[i];
568 uint32_t size = avio_rb32(pb);
569 int64_t next = avio_tell(pb) + size - 4;
572 return AVERROR_INVALIDDATA;
574 dref->type = avio_rl32(pb);
575 avio_rb32(pb); // version + flags
576 av_log(c->fc, AV_LOG_TRACE, "type %.4s size %d\n", (char*)&dref->type, size);
578 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
579 /* macintosh alias record */
580 uint16_t volume_len, len;
586 volume_len = avio_r8(pb);
587 volume_len = FFMIN(volume_len, 27);
588 ret = ffio_read_size(pb, dref->volume, 27);
591 dref->volume[volume_len] = 0;
592 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", dref->volume, volume_len);
597 len = FFMIN(len, 63);
598 ret = ffio_read_size(pb, dref->filename, 63);
601 dref->filename[len] = 0;
602 av_log(c->fc, AV_LOG_DEBUG, "filename %s, len %d\n", dref->filename, len);
606 /* read next level up_from_alias/down_to_target */
607 dref->nlvl_from = avio_rb16(pb);
608 dref->nlvl_to = avio_rb16(pb);
609 av_log(c->fc, AV_LOG_DEBUG, "nlvl from %d, nlvl to %d\n",
610 dref->nlvl_from, dref->nlvl_to);
614 for (type = 0; type != -1 && avio_tell(pb) < next; ) {
617 type = avio_rb16(pb);
619 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
622 if (type == 2) { // absolute path
624 dref->path = av_mallocz(len+1);
626 return AVERROR(ENOMEM);
628 ret = ffio_read_size(pb, dref->path, len);
630 av_freep(&dref->path);
633 if (len > volume_len && !strncmp(dref->path, dref->volume, volume_len)) {
635 memmove(dref->path, dref->path+volume_len, len);
638 // trim string of any ending zeros
639 for (j = len - 1; j >= 0; j--) {
640 if (dref->path[j] == 0)
645 for (j = 0; j < len; j++)
646 if (dref->path[j] == ':' || dref->path[j] == 0)
648 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
649 } else if (type == 0) { // directory name
651 dref->dir = av_malloc(len+1);
653 return AVERROR(ENOMEM);
655 ret = ffio_read_size(pb, dref->dir, len);
657 av_freep(&dref->dir);
661 for (j = 0; j < len; j++)
662 if (dref->dir[j] == ':')
664 av_log(c->fc, AV_LOG_DEBUG, "dir %s\n", dref->dir);
669 av_log(c->fc, AV_LOG_DEBUG, "Unknown dref type 0x08%x size %d\n",
674 avio_seek(pb, next, SEEK_SET);
679 static int mov_read_hdlr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
683 uint32_t av_unused ctype;
688 avio_r8(pb); /* version */
689 avio_rb24(pb); /* flags */
692 ctype = avio_rl32(pb);
693 type = avio_rl32(pb); /* component subtype */
695 av_log(c->fc, AV_LOG_TRACE, "ctype= %.4s (0x%08x)\n", (char*)&ctype, ctype);
696 av_log(c->fc, AV_LOG_TRACE, "stype= %.4s\n", (char*)&type);
698 if (c->trak_index < 0) { // meta not inside a trak
699 if (type == MKTAG('m','d','t','a')) {
700 c->found_hdlr_mdta = 1;
705 st = c->fc->streams[c->fc->nb_streams-1];
707 if (type == MKTAG('v','i','d','e'))
708 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
709 else if (type == MKTAG('s','o','u','n'))
710 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
711 else if (type == MKTAG('m','1','a',' '))
712 st->codecpar->codec_id = AV_CODEC_ID_MP2;
713 else if ((type == MKTAG('s','u','b','p')) || (type == MKTAG('c','l','c','p')))
714 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
716 avio_rb32(pb); /* component manufacture */
717 avio_rb32(pb); /* component flags */
718 avio_rb32(pb); /* component flags mask */
720 title_size = atom.size - 24;
721 if (title_size > 0) {
722 title_str = av_malloc(title_size + 1); /* Add null terminator */
724 return AVERROR(ENOMEM);
726 ret = ffio_read_size(pb, title_str, title_size);
728 av_freep(&title_str);
731 title_str[title_size] = 0;
733 int off = (!c->isom && title_str[0] == title_size - 1);
734 av_dict_set(&st->metadata, "handler_name", title_str + off, 0);
736 av_freep(&title_str);
742 int ff_mov_read_esds(AVFormatContext *fc, AVIOContext *pb)
747 if (fc->nb_streams < 1)
749 st = fc->streams[fc->nb_streams-1];
751 avio_rb32(pb); /* version + flags */
752 ff_mp4_read_descr(fc, pb, &tag);
753 if (tag == MP4ESDescrTag) {
754 ff_mp4_parse_es_descr(pb, NULL);
756 avio_rb16(pb); /* ID */
758 ff_mp4_read_descr(fc, pb, &tag);
759 if (tag == MP4DecConfigDescrTag)
760 ff_mp4_read_dec_config_descr(fc, st, pb);
764 static int mov_read_esds(MOVContext *c, AVIOContext *pb, MOVAtom atom)
766 return ff_mov_read_esds(c->fc, pb);
769 static int mov_read_dac3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
772 enum AVAudioServiceType *ast;
773 int ac3info, acmod, lfeon, bsmod;
775 if (c->fc->nb_streams < 1)
777 st = c->fc->streams[c->fc->nb_streams-1];
779 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
782 return AVERROR(ENOMEM);
784 ac3info = avio_rb24(pb);
785 bsmod = (ac3info >> 14) & 0x7;
786 acmod = (ac3info >> 11) & 0x7;
787 lfeon = (ac3info >> 10) & 0x1;
788 st->codecpar->channels = ((int[]){2,1,2,3,3,4,4,5})[acmod] + lfeon;
789 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
791 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
793 if (st->codecpar->channels > 1 && bsmod == 0x7)
794 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
796 #if FF_API_LAVF_AVCTX
797 FF_DISABLE_DEPRECATION_WARNINGS
798 st->codec->audio_service_type = *ast;
799 FF_ENABLE_DEPRECATION_WARNINGS
805 static int mov_read_dec3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
808 enum AVAudioServiceType *ast;
809 int eac3info, acmod, lfeon, bsmod;
811 if (c->fc->nb_streams < 1)
813 st = c->fc->streams[c->fc->nb_streams-1];
815 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
818 return AVERROR(ENOMEM);
820 /* No need to parse fields for additional independent substreams and its
821 * associated dependent substreams since libavcodec's E-AC-3 decoder
822 * does not support them yet. */
823 avio_rb16(pb); /* data_rate and num_ind_sub */
824 eac3info = avio_rb24(pb);
825 bsmod = (eac3info >> 12) & 0x1f;
826 acmod = (eac3info >> 9) & 0x7;
827 lfeon = (eac3info >> 8) & 0x1;
828 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
830 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
831 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
833 if (st->codecpar->channels > 1 && bsmod == 0x7)
834 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
836 #if FF_API_LAVF_AVCTX
837 FF_DISABLE_DEPRECATION_WARNINGS
838 st->codec->audio_service_type = *ast;
839 FF_ENABLE_DEPRECATION_WARNINGS
845 static int mov_read_ddts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
847 const uint32_t ddts_size = 20;
850 uint32_t frame_duration_code = 0;
851 uint32_t channel_layout_code = 0;
854 buf = av_malloc(ddts_size + FF_INPUT_BUFFER_PADDING_SIZE);
856 return AVERROR(ENOMEM);
858 if (avio_read(pb, buf, ddts_size) < ddts_size) {
860 return AVERROR_INVALIDDATA;
863 init_get_bits(&gb, buf, 8*ddts_size);
865 if (c->fc->nb_streams < 1) {
868 st = c->fc->streams[c->fc->nb_streams-1];
870 st->codecpar->sample_rate = get_bits_long(&gb, 32);
871 skip_bits_long(&gb, 32); /* max bitrate */
872 st->codecpar->bit_rate = get_bits_long(&gb, 32);
873 st->codecpar->bits_per_coded_sample = get_bits(&gb, 8);
874 frame_duration_code = get_bits(&gb, 2);
875 skip_bits(&gb, 30); /* various fields */
876 channel_layout_code = get_bits(&gb, 16);
878 st->codecpar->frame_size =
879 (frame_duration_code == 0) ? 512 :
880 (frame_duration_code == 1) ? 1024 :
881 (frame_duration_code == 2) ? 2048 :
882 (frame_duration_code == 3) ? 4096 : 0;
884 if (channel_layout_code > 0xff) {
885 av_log(c->fc, AV_LOG_WARNING, "Unsupported DTS audio channel layout");
887 st->codecpar->channel_layout =
888 ((channel_layout_code & 0x1) ? AV_CH_FRONT_CENTER : 0) |
889 ((channel_layout_code & 0x2) ? AV_CH_FRONT_LEFT : 0) |
890 ((channel_layout_code & 0x2) ? AV_CH_FRONT_RIGHT : 0) |
891 ((channel_layout_code & 0x4) ? AV_CH_SIDE_LEFT : 0) |
892 ((channel_layout_code & 0x4) ? AV_CH_SIDE_RIGHT : 0) |
893 ((channel_layout_code & 0x8) ? AV_CH_LOW_FREQUENCY : 0);
895 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
900 static int mov_read_chan(MOVContext *c, AVIOContext *pb, MOVAtom atom)
904 if (c->fc->nb_streams < 1)
906 st = c->fc->streams[c->fc->nb_streams-1];
911 /* skip version and flags */
914 ff_mov_read_chan(c->fc, pb, st, atom.size - 4);
919 static int mov_read_wfex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
924 if (c->fc->nb_streams < 1)
926 st = c->fc->streams[c->fc->nb_streams-1];
928 if ((ret = ff_get_wav_header(c->fc, pb, st->codecpar, atom.size, 0)) < 0)
929 av_log(c->fc, AV_LOG_WARNING, "get_wav_header failed\n");
934 static int mov_read_pasp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
936 const int num = avio_rb32(pb);
937 const int den = avio_rb32(pb);
940 if (c->fc->nb_streams < 1)
942 st = c->fc->streams[c->fc->nb_streams-1];
944 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
945 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num)) {
946 av_log(c->fc, AV_LOG_WARNING,
947 "sample aspect ratio already set to %d:%d, ignoring 'pasp' atom (%d:%d)\n",
948 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den,
950 } else if (den != 0) {
951 av_reduce(&st->sample_aspect_ratio.num, &st->sample_aspect_ratio.den,
957 /* this atom contains actual media data */
958 static int mov_read_mdat(MOVContext *c, AVIOContext *pb, MOVAtom atom)
960 if (atom.size == 0) /* wrong one (MP4) */
963 return 0; /* now go for moov */
966 #define DRM_BLOB_SIZE 56
968 static int mov_read_adrm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
970 uint8_t intermediate_key[20];
971 uint8_t intermediate_iv[20];
974 uint8_t file_checksum[20];
975 uint8_t calculated_checksum[20];
979 uint8_t *activation_bytes = c->activation_bytes;
980 uint8_t *fixed_key = c->audible_fixed_key;
984 sha = av_sha_alloc();
986 return AVERROR(ENOMEM);
987 c->aes_decrypt = av_aes_alloc();
988 if (!c->aes_decrypt) {
989 ret = AVERROR(ENOMEM);
993 /* drm blob processing */
994 avio_read(pb, output, 8); // go to offset 8, absolute position 0x251
995 avio_read(pb, input, DRM_BLOB_SIZE);
996 avio_read(pb, output, 4); // go to offset 4, absolute position 0x28d
997 avio_read(pb, file_checksum, 20);
999 av_log(c->fc, AV_LOG_INFO, "[aax] file checksum == "); // required by external tools
1000 for (i = 0; i < 20; i++)
1001 av_log(c->fc, AV_LOG_INFO, "%02x", file_checksum[i]);
1002 av_log(c->fc, AV_LOG_INFO, "\n");
1004 /* verify activation data */
1005 if (!activation_bytes) {
1006 av_log(c->fc, AV_LOG_WARNING, "[aax] activation_bytes option is missing!\n");
1007 ret = 0; /* allow ffprobe to continue working on .aax files */
1010 if (c->activation_bytes_size != 4) {
1011 av_log(c->fc, AV_LOG_FATAL, "[aax] activation_bytes value needs to be 4 bytes!\n");
1012 ret = AVERROR(EINVAL);
1016 /* verify fixed key */
1017 if (c->audible_fixed_key_size != 16) {
1018 av_log(c->fc, AV_LOG_FATAL, "[aax] audible_fixed_key value needs to be 16 bytes!\n");
1019 ret = AVERROR(EINVAL);
1023 /* AAX (and AAX+) key derivation */
1024 av_sha_init(sha, 160);
1025 av_sha_update(sha, fixed_key, 16);
1026 av_sha_update(sha, activation_bytes, 4);
1027 av_sha_final(sha, intermediate_key);
1028 av_sha_init(sha, 160);
1029 av_sha_update(sha, fixed_key, 16);
1030 av_sha_update(sha, intermediate_key, 20);
1031 av_sha_update(sha, activation_bytes, 4);
1032 av_sha_final(sha, intermediate_iv);
1033 av_sha_init(sha, 160);
1034 av_sha_update(sha, intermediate_key, 16);
1035 av_sha_update(sha, intermediate_iv, 16);
1036 av_sha_final(sha, calculated_checksum);
1037 if (memcmp(calculated_checksum, file_checksum, 20)) { // critical error
1038 av_log(c->fc, AV_LOG_ERROR, "[aax] mismatch in checksums!\n");
1039 ret = AVERROR_INVALIDDATA;
1042 av_aes_init(c->aes_decrypt, intermediate_key, 128, 1);
1043 av_aes_crypt(c->aes_decrypt, output, input, DRM_BLOB_SIZE >> 4, intermediate_iv, 1);
1044 for (i = 0; i < 4; i++) {
1045 // file data (in output) is stored in big-endian mode
1046 if (activation_bytes[i] != output[3 - i]) { // critical error
1047 av_log(c->fc, AV_LOG_ERROR, "[aax] error in drm blob decryption!\n");
1048 ret = AVERROR_INVALIDDATA;
1052 memcpy(c->file_key, output + 8, 16);
1053 memcpy(input, output + 26, 16);
1054 av_sha_init(sha, 160);
1055 av_sha_update(sha, input, 16);
1056 av_sha_update(sha, c->file_key, 16);
1057 av_sha_update(sha, fixed_key, 16);
1058 av_sha_final(sha, c->file_iv);
1066 // Audible AAX (and AAX+) bytestream decryption
1067 static int aax_filter(uint8_t *input, int size, MOVContext *c)
1070 unsigned char iv[16];
1072 memcpy(iv, c->file_iv, 16); // iv is overwritten
1073 blocks = size >> 4; // trailing bytes are not encrypted!
1074 av_aes_init(c->aes_decrypt, c->file_key, 128, 1);
1075 av_aes_crypt(c->aes_decrypt, input, input, blocks, iv, 1);
1080 /* read major brand, minor version and compatible brands and store them as metadata */
1081 static int mov_read_ftyp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1084 int comp_brand_size;
1085 char* comp_brands_str;
1086 uint8_t type[5] = {0};
1087 int ret = ffio_read_size(pb, type, 4);
1091 if (strcmp(type, "qt "))
1093 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
1094 av_dict_set(&c->fc->metadata, "major_brand", type, 0);
1095 minor_ver = avio_rb32(pb); /* minor version */
1096 av_dict_set_int(&c->fc->metadata, "minor_version", minor_ver, 0);
1098 comp_brand_size = atom.size - 8;
1099 if (comp_brand_size < 0)
1100 return AVERROR_INVALIDDATA;
1101 comp_brands_str = av_malloc(comp_brand_size + 1); /* Add null terminator */
1102 if (!comp_brands_str)
1103 return AVERROR(ENOMEM);
1105 ret = ffio_read_size(pb, comp_brands_str, comp_brand_size);
1107 av_freep(&comp_brands_str);
1110 comp_brands_str[comp_brand_size] = 0;
1111 av_dict_set(&c->fc->metadata, "compatible_brands", comp_brands_str, 0);
1112 av_freep(&comp_brands_str);
1117 /* this atom should contain all header atoms */
1118 static int mov_read_moov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1122 if (c->found_moov) {
1123 av_log(c->fc, AV_LOG_WARNING, "Found duplicated MOOV Atom. Skipped it\n");
1124 avio_skip(pb, atom.size);
1128 if ((ret = mov_read_default(c, pb, atom)) < 0)
1130 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
1131 /* so we don't parse the whole file if over a network */
1133 return 0; /* now go for mdat */
1136 static int mov_read_moof(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1138 if (!c->has_looked_for_mfra && c->use_mfra_for > 0) {
1139 c->has_looked_for_mfra = 1;
1142 av_log(c->fc, AV_LOG_VERBOSE, "stream has moof boxes, will look "
1144 if ((ret = mov_read_mfra(c, pb)) < 0) {
1145 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but failed to "
1146 "read the mfra (may be a live ismv)\n");
1149 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but stream is not "
1150 "seekable, can not look for mfra\n");
1153 c->fragment.moof_offset = c->fragment.implicit_offset = avio_tell(pb) - 8;
1154 av_log(c->fc, AV_LOG_TRACE, "moof offset %"PRIx64"\n", c->fragment.moof_offset);
1155 return mov_read_default(c, pb, atom);
1158 static void mov_metadata_creation_time(AVDictionary **metadata, int64_t time)
1161 if(time >= 2082844800)
1162 time -= 2082844800; /* seconds between 1904-01-01 and Epoch */
1163 avpriv_dict_set_timestamp(metadata, "creation_time", time * 1000000);
1167 static int mov_read_mdhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1170 MOVStreamContext *sc;
1172 char language[4] = {0};
1174 int64_t creation_time;
1176 if (c->fc->nb_streams < 1)
1178 st = c->fc->streams[c->fc->nb_streams-1];
1181 if (sc->time_scale) {
1182 av_log(c->fc, AV_LOG_ERROR, "Multiple mdhd?\n");
1183 return AVERROR_INVALIDDATA;
1186 version = avio_r8(pb);
1188 avpriv_request_sample(c->fc, "Version %d", version);
1189 return AVERROR_PATCHWELCOME;
1191 avio_rb24(pb); /* flags */
1193 creation_time = avio_rb64(pb);
1196 creation_time = avio_rb32(pb);
1197 avio_rb32(pb); /* modification time */
1199 mov_metadata_creation_time(&st->metadata, creation_time);
1201 sc->time_scale = avio_rb32(pb);
1202 st->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1204 lang = avio_rb16(pb); /* language */
1205 if (ff_mov_lang_to_iso639(lang, language))
1206 av_dict_set(&st->metadata, "language", language, 0);
1207 avio_rb16(pb); /* quality */
1212 static int mov_read_mvhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1214 int64_t creation_time;
1215 int version = avio_r8(pb); /* version */
1216 avio_rb24(pb); /* flags */
1219 creation_time = avio_rb64(pb);
1222 creation_time = avio_rb32(pb);
1223 avio_rb32(pb); /* modification time */
1225 mov_metadata_creation_time(&c->fc->metadata, creation_time);
1226 c->time_scale = avio_rb32(pb); /* time scale */
1228 av_log(c->fc, AV_LOG_TRACE, "time scale = %i\n", c->time_scale);
1230 c->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1231 // set the AVCodecContext duration because the duration of individual tracks
1232 // may be inaccurate
1233 if (c->time_scale > 0 && !c->trex_data)
1234 c->fc->duration = av_rescale(c->duration, AV_TIME_BASE, c->time_scale);
1235 avio_rb32(pb); /* preferred scale */
1237 avio_rb16(pb); /* preferred volume */
1239 avio_skip(pb, 10); /* reserved */
1241 avio_skip(pb, 36); /* display matrix */
1243 avio_rb32(pb); /* preview time */
1244 avio_rb32(pb); /* preview duration */
1245 avio_rb32(pb); /* poster time */
1246 avio_rb32(pb); /* selection time */
1247 avio_rb32(pb); /* selection duration */
1248 avio_rb32(pb); /* current time */
1249 avio_rb32(pb); /* next track ID */
1254 static int mov_read_enda(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1259 if (c->fc->nb_streams < 1)
1261 st = c->fc->streams[c->fc->nb_streams-1];
1263 little_endian = avio_rb16(pb) & 0xFF;
1264 av_log(c->fc, AV_LOG_TRACE, "enda %d\n", little_endian);
1265 if (little_endian == 1) {
1266 switch (st->codecpar->codec_id) {
1267 case AV_CODEC_ID_PCM_S24BE:
1268 st->codecpar->codec_id = AV_CODEC_ID_PCM_S24LE;
1270 case AV_CODEC_ID_PCM_S32BE:
1271 st->codecpar->codec_id = AV_CODEC_ID_PCM_S32LE;
1273 case AV_CODEC_ID_PCM_F32BE:
1274 st->codecpar->codec_id = AV_CODEC_ID_PCM_F32LE;
1276 case AV_CODEC_ID_PCM_F64BE:
1277 st->codecpar->codec_id = AV_CODEC_ID_PCM_F64LE;
1286 static int mov_read_colr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1289 char color_parameter_type[5] = { 0 };
1290 uint16_t color_primaries, color_trc, color_matrix;
1293 if (c->fc->nb_streams < 1)
1295 st = c->fc->streams[c->fc->nb_streams - 1];
1297 ret = ffio_read_size(pb, color_parameter_type, 4);
1300 if (strncmp(color_parameter_type, "nclx", 4) &&
1301 strncmp(color_parameter_type, "nclc", 4)) {
1302 av_log(c->fc, AV_LOG_WARNING, "unsupported color_parameter_type %s\n",
1303 color_parameter_type);
1307 color_primaries = avio_rb16(pb);
1308 color_trc = avio_rb16(pb);
1309 color_matrix = avio_rb16(pb);
1311 av_log(c->fc, AV_LOG_TRACE,
1312 "%s: pri %d trc %d matrix %d",
1313 color_parameter_type, color_primaries, color_trc, color_matrix);
1315 if (!strncmp(color_parameter_type, "nclx", 4)) {
1316 uint8_t color_range = avio_r8(pb) >> 7;
1317 av_log(c->fc, AV_LOG_TRACE, " full %"PRIu8"", color_range);
1319 st->codecpar->color_range = AVCOL_RANGE_JPEG;
1321 st->codecpar->color_range = AVCOL_RANGE_MPEG;
1323 if (color_primaries >= AVCOL_PRI_NB)
1324 color_primaries = AVCOL_PRI_UNSPECIFIED;
1325 if (color_trc >= AVCOL_TRC_NB)
1326 color_trc = AVCOL_TRC_UNSPECIFIED;
1327 if (color_matrix >= AVCOL_SPC_NB)
1328 color_matrix = AVCOL_SPC_UNSPECIFIED;
1329 st->codecpar->color_primaries = color_primaries;
1330 st->codecpar->color_trc = color_trc;
1331 st->codecpar->color_space = color_matrix;
1332 av_log(c->fc, AV_LOG_TRACE, "\n");
1337 static int mov_read_fiel(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1340 unsigned mov_field_order;
1341 enum AVFieldOrder decoded_field_order = AV_FIELD_UNKNOWN;
1343 if (c->fc->nb_streams < 1) // will happen with jp2 files
1345 st = c->fc->streams[c->fc->nb_streams-1];
1347 return AVERROR_INVALIDDATA;
1348 mov_field_order = avio_rb16(pb);
1349 if ((mov_field_order & 0xFF00) == 0x0100)
1350 decoded_field_order = AV_FIELD_PROGRESSIVE;
1351 else if ((mov_field_order & 0xFF00) == 0x0200) {
1352 switch (mov_field_order & 0xFF) {
1353 case 0x01: decoded_field_order = AV_FIELD_TT;
1355 case 0x06: decoded_field_order = AV_FIELD_BB;
1357 case 0x09: decoded_field_order = AV_FIELD_TB;
1359 case 0x0E: decoded_field_order = AV_FIELD_BT;
1363 if (decoded_field_order == AV_FIELD_UNKNOWN && mov_field_order) {
1364 av_log(NULL, AV_LOG_ERROR, "Unknown MOV field order 0x%04x\n", mov_field_order);
1366 st->codecpar->field_order = decoded_field_order;
1371 static int mov_realloc_extradata(AVCodecParameters *par, MOVAtom atom)
1374 uint64_t size = (uint64_t)par->extradata_size + atom.size + 8 + AV_INPUT_BUFFER_PADDING_SIZE;
1375 if (size > INT_MAX || (uint64_t)atom.size > INT_MAX)
1376 return AVERROR_INVALIDDATA;
1377 if ((err = av_reallocp(&par->extradata, size)) < 0) {
1378 par->extradata_size = 0;
1381 par->extradata_size = size - AV_INPUT_BUFFER_PADDING_SIZE;
1385 /* Read a whole atom into the extradata return the size of the atom read, possibly truncated if != atom.size */
1386 static int64_t mov_read_atom_into_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1387 AVCodecParameters *par, uint8_t *buf)
1389 int64_t result = atom.size;
1392 AV_WB32(buf , atom.size + 8);
1393 AV_WL32(buf + 4, atom.type);
1394 err = ffio_read_size(pb, buf + 8, atom.size);
1396 par->extradata_size -= atom.size;
1398 } else if (err < atom.size) {
1399 av_log(c->fc, AV_LOG_WARNING, "truncated extradata\n");
1400 par->extradata_size -= atom.size - err;
1403 memset(buf + 8 + err, 0, AV_INPUT_BUFFER_PADDING_SIZE);
1407 /* FIXME modify QDM2/SVQ3/H.264 decoders to take full atom as extradata */
1408 static int mov_read_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1409 enum AVCodecID codec_id)
1412 uint64_t original_size;
1415 if (c->fc->nb_streams < 1) // will happen with jp2 files
1417 st = c->fc->streams[c->fc->nb_streams-1];
1419 if (st->codecpar->codec_id != codec_id)
1420 return 0; /* unexpected codec_id - don't mess with extradata */
1422 original_size = st->codecpar->extradata_size;
1423 err = mov_realloc_extradata(st->codecpar, atom);
1427 err = mov_read_atom_into_extradata(c, pb, atom, st->codecpar, st->codecpar->extradata + original_size);
1430 return 0; // Note: this is the original behavior to ignore truncation.
1433 /* wrapper functions for reading ALAC/AVS/MJPEG/MJPEG2000 extradata atoms only for those codecs */
1434 static int mov_read_alac(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1436 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_ALAC);
1439 static int mov_read_avss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1441 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVS);
1444 static int mov_read_jp2h(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1446 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_JPEG2000);
1449 static int mov_read_dpxe(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1451 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_R10K);
1454 static int mov_read_avid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1456 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVUI);
1458 ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_DNXHD);
1462 static int mov_read_targa_y216(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1464 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_TARGA_Y216);
1466 if (!ret && c->fc->nb_streams >= 1) {
1467 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1468 if (par->extradata_size >= 40) {
1469 par->height = AV_RB16(&par->extradata[36]);
1470 par->width = AV_RB16(&par->extradata[38]);
1476 static int mov_read_ares(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1478 if (c->fc->nb_streams >= 1) {
1479 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1480 if (par->codec_tag == MKTAG('A', 'V', 'i', 'n') &&
1481 par->codec_id == AV_CODEC_ID_H264 &&
1485 cid = avio_rb16(pb);
1486 /* For AVID AVCI50, force width of 1440 to be able to select the correct SPS and PPS */
1487 if (cid == 0xd4d || cid == 0xd4e)
1490 } else if ((par->codec_tag == MKTAG('A', 'V', 'd', '1') ||
1491 par->codec_tag == MKTAG('A', 'V', 'd', 'n')) &&
1495 num = avio_rb32(pb);
1496 den = avio_rb32(pb);
1497 if (num <= 0 || den <= 0)
1499 switch (avio_rb32(pb)) {
1501 if (den >= INT_MAX / 2)
1505 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.num = num;
1506 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.den = den;
1513 return mov_read_avid(c, pb, atom);
1516 static int mov_read_aclr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1520 uint64_t original_size;
1521 if (c->fc->nb_streams >= 1) {
1522 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1523 if (par->codec_id == AV_CODEC_ID_H264)
1525 if (atom.size == 16) {
1526 original_size = par->extradata_size;
1527 ret = mov_realloc_extradata(par, atom);
1529 length = mov_read_atom_into_extradata(c, pb, atom, par, par->extradata + original_size);
1530 if (length == atom.size) {
1531 const uint8_t range_value = par->extradata[original_size + 19];
1532 switch (range_value) {
1534 par->color_range = AVCOL_RANGE_MPEG;
1537 par->color_range = AVCOL_RANGE_JPEG;
1540 av_log(c, AV_LOG_WARNING, "ignored unknown aclr value (%d)\n", range_value);
1543 ff_dlog(c, "color_range: %d\n", par->color_range);
1545 /* For some reason the whole atom was not added to the extradata */
1546 av_log(c, AV_LOG_ERROR, "aclr not decoded - incomplete atom\n");
1549 av_log(c, AV_LOG_ERROR, "aclr not decoded - unable to add atom to extradata\n");
1552 av_log(c, AV_LOG_WARNING, "aclr not decoded - unexpected size %"PRId64"\n", atom.size);
1559 static int mov_read_svq3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1561 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_SVQ3);
1564 static int mov_read_wave(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1569 if (c->fc->nb_streams < 1)
1571 st = c->fc->streams[c->fc->nb_streams-1];
1573 if ((uint64_t)atom.size > (1<<30))
1574 return AVERROR_INVALIDDATA;
1576 if (st->codecpar->codec_id == AV_CODEC_ID_QDM2 ||
1577 st->codecpar->codec_id == AV_CODEC_ID_QDMC ||
1578 st->codecpar->codec_id == AV_CODEC_ID_SPEEX) {
1579 // pass all frma atom to codec, needed at least for QDMC and QDM2
1580 av_freep(&st->codecpar->extradata);
1581 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1584 } else if (atom.size > 8) { /* to read frma, esds atoms */
1585 if (st->codecpar->codec_id == AV_CODEC_ID_ALAC && atom.size >= 24) {
1587 ret = ffio_ensure_seekback(pb, 8);
1590 buffer = avio_rb64(pb);
1592 if ( (buffer & 0xFFFFFFFF) == MKBETAG('f','r','m','a')
1593 && buffer >> 32 <= atom.size
1594 && buffer >> 32 >= 8) {
1597 } else if (!st->codecpar->extradata_size) {
1598 #define ALAC_EXTRADATA_SIZE 36
1599 st->codecpar->extradata = av_mallocz(ALAC_EXTRADATA_SIZE + AV_INPUT_BUFFER_PADDING_SIZE);
1600 if (!st->codecpar->extradata)
1601 return AVERROR(ENOMEM);
1602 st->codecpar->extradata_size = ALAC_EXTRADATA_SIZE;
1603 AV_WB32(st->codecpar->extradata , ALAC_EXTRADATA_SIZE);
1604 AV_WB32(st->codecpar->extradata + 4, MKTAG('a','l','a','c'));
1605 AV_WB64(st->codecpar->extradata + 12, buffer);
1606 avio_read(pb, st->codecpar->extradata + 20, 16);
1607 avio_skip(pb, atom.size - 24);
1611 if ((ret = mov_read_default(c, pb, atom)) < 0)
1614 avio_skip(pb, atom.size);
1619 * This function reads atom content and puts data in extradata without tag
1620 * nor size unlike mov_read_extradata.
1622 static int mov_read_glbl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1627 if (c->fc->nb_streams < 1)
1629 st = c->fc->streams[c->fc->nb_streams-1];
1631 if ((uint64_t)atom.size > (1<<30))
1632 return AVERROR_INVALIDDATA;
1634 if (atom.size >= 10) {
1635 // Broken files created by legacy versions of libavformat will
1636 // wrap a whole fiel atom inside of a glbl atom.
1637 unsigned size = avio_rb32(pb);
1638 unsigned type = avio_rl32(pb);
1639 avio_seek(pb, -8, SEEK_CUR);
1640 if (type == MKTAG('f','i','e','l') && size == atom.size)
1641 return mov_read_default(c, pb, atom);
1643 if (st->codecpar->extradata_size > 1 && st->codecpar->extradata) {
1644 av_log(c, AV_LOG_WARNING, "ignoring multiple glbl\n");
1647 av_freep(&st->codecpar->extradata);
1648 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1655 static int mov_read_dvc1(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1658 uint8_t profile_level;
1661 if (c->fc->nb_streams < 1)
1663 st = c->fc->streams[c->fc->nb_streams-1];
1665 if (atom.size >= (1<<28) || atom.size < 7)
1666 return AVERROR_INVALIDDATA;
1668 profile_level = avio_r8(pb);
1669 if ((profile_level & 0xf0) != 0xc0)
1672 avio_seek(pb, 6, SEEK_CUR);
1673 av_freep(&st->codecpar->extradata);
1674 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 7);
1682 * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
1683 * but can have extradata appended at the end after the 40 bytes belonging
1686 static int mov_read_strf(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1691 if (c->fc->nb_streams < 1)
1693 if (atom.size <= 40)
1695 st = c->fc->streams[c->fc->nb_streams-1];
1697 if ((uint64_t)atom.size > (1<<30))
1698 return AVERROR_INVALIDDATA;
1701 av_freep(&st->codecpar->extradata);
1702 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 40);
1709 static int mov_read_stco(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1712 MOVStreamContext *sc;
1713 unsigned int i, entries;
1715 if (c->fc->nb_streams < 1)
1717 st = c->fc->streams[c->fc->nb_streams-1];
1720 avio_r8(pb); /* version */
1721 avio_rb24(pb); /* flags */
1723 entries = avio_rb32(pb);
1728 if (sc->chunk_offsets)
1729 av_log(c->fc, AV_LOG_WARNING, "Duplicated STCO atom\n");
1730 av_free(sc->chunk_offsets);
1731 sc->chunk_count = 0;
1732 sc->chunk_offsets = av_malloc_array(entries, sizeof(*sc->chunk_offsets));
1733 if (!sc->chunk_offsets)
1734 return AVERROR(ENOMEM);
1735 sc->chunk_count = entries;
1737 if (atom.type == MKTAG('s','t','c','o'))
1738 for (i = 0; i < entries && !pb->eof_reached; i++)
1739 sc->chunk_offsets[i] = avio_rb32(pb);
1740 else if (atom.type == MKTAG('c','o','6','4'))
1741 for (i = 0; i < entries && !pb->eof_reached; i++)
1742 sc->chunk_offsets[i] = avio_rb64(pb);
1744 return AVERROR_INVALIDDATA;
1746 sc->chunk_count = i;
1748 if (pb->eof_reached)
1755 * Compute codec id for 'lpcm' tag.
1756 * See CoreAudioTypes and AudioStreamBasicDescription at Apple.
1758 enum AVCodecID ff_mov_get_lpcm_codec_id(int bps, int flags)
1765 return ff_get_pcm_codec_id(bps, flags & 1, flags & 2, flags & 4 ? -1 : 0);
1768 static int mov_codec_id(AVStream *st, uint32_t format)
1770 int id = ff_codec_get_id(ff_codec_movaudio_tags, format);
1773 ((format & 0xFFFF) == 'm' + ('s' << 8) ||
1774 (format & 0xFFFF) == 'T' + ('S' << 8)))
1775 id = ff_codec_get_id(ff_codec_wav_tags, av_bswap32(format) & 0xFFFF);
1777 if (st->codecpar->codec_type != AVMEDIA_TYPE_VIDEO && id > 0) {
1778 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
1779 } else if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO &&
1780 /* skip old ASF MPEG-4 tag */
1781 format && format != MKTAG('m','p','4','s')) {
1782 id = ff_codec_get_id(ff_codec_movvideo_tags, format);
1784 id = ff_codec_get_id(ff_codec_bmp_tags, format);
1786 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
1787 else if (st->codecpar->codec_type == AVMEDIA_TYPE_DATA ||
1788 (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE &&
1789 st->codecpar->codec_id == AV_CODEC_ID_NONE)) {
1790 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
1792 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
1796 st->codecpar->codec_tag = format;
1801 static void mov_parse_stsd_video(MOVContext *c, AVIOContext *pb,
1802 AVStream *st, MOVStreamContext *sc)
1804 uint8_t codec_name[32];
1808 /* The first 16 bytes of the video sample description are already
1809 * read in ff_mov_read_stsd_entries() */
1810 stsd_start = avio_tell(pb) - 16;
1812 avio_rb16(pb); /* version */
1813 avio_rb16(pb); /* revision level */
1814 avio_rb32(pb); /* vendor */
1815 avio_rb32(pb); /* temporal quality */
1816 avio_rb32(pb); /* spatial quality */
1818 st->codecpar->width = avio_rb16(pb); /* width */
1819 st->codecpar->height = avio_rb16(pb); /* height */
1821 avio_rb32(pb); /* horiz resolution */
1822 avio_rb32(pb); /* vert resolution */
1823 avio_rb32(pb); /* data size, always 0 */
1824 avio_rb16(pb); /* frames per samples */
1826 len = avio_r8(pb); /* codec name, pascal string */
1829 mov_read_mac_string(c, pb, len, codec_name, sizeof(codec_name));
1831 avio_skip(pb, 31 - len);
1834 av_dict_set(&st->metadata, "encoder", codec_name, 0);
1836 /* codec_tag YV12 triggers an UV swap in rawdec.c */
1837 if (!memcmp(codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25)) {
1838 st->codecpar->codec_tag = MKTAG('I', '4', '2', '0');
1839 st->codecpar->width &= ~1;
1840 st->codecpar->height &= ~1;
1842 /* Flash Media Server uses tag H.263 with Sorenson Spark */
1843 if (st->codecpar->codec_tag == MKTAG('H','2','6','3') &&
1844 !memcmp(codec_name, "Sorenson H263", 13))
1845 st->codecpar->codec_id = AV_CODEC_ID_FLV1;
1847 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* depth */
1849 avio_seek(pb, stsd_start, SEEK_SET);
1851 if (ff_get_qtpalette(st->codecpar->codec_id, pb, sc->palette)) {
1852 st->codecpar->bits_per_coded_sample &= 0x1F;
1853 sc->has_palette = 1;
1857 static void mov_parse_stsd_audio(MOVContext *c, AVIOContext *pb,
1858 AVStream *st, MOVStreamContext *sc)
1860 int bits_per_sample, flags;
1861 uint16_t version = avio_rb16(pb);
1862 AVDictionaryEntry *compatible_brands = av_dict_get(c->fc->metadata, "compatible_brands", NULL, AV_DICT_MATCH_CASE);
1864 avio_rb16(pb); /* revision level */
1865 avio_rb32(pb); /* vendor */
1867 st->codecpar->channels = avio_rb16(pb); /* channel count */
1868 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* sample size */
1869 av_log(c->fc, AV_LOG_TRACE, "audio channels %d\n", st->codecpar->channels);
1871 sc->audio_cid = avio_rb16(pb);
1872 avio_rb16(pb); /* packet size = 0 */
1874 st->codecpar->sample_rate = ((avio_rb32(pb) >> 16));
1876 // Read QT version 1 fields. In version 0 these do not exist.
1877 av_log(c->fc, AV_LOG_TRACE, "version =%d, isom =%d\n", version, c->isom);
1879 (compatible_brands && strstr(compatible_brands->value, "qt "))) {
1882 sc->samples_per_frame = avio_rb32(pb);
1883 avio_rb32(pb); /* bytes per packet */
1884 sc->bytes_per_frame = avio_rb32(pb);
1885 avio_rb32(pb); /* bytes per sample */
1886 } else if (version == 2) {
1887 avio_rb32(pb); /* sizeof struct only */
1888 st->codecpar->sample_rate = av_int2double(avio_rb64(pb));
1889 st->codecpar->channels = avio_rb32(pb);
1890 avio_rb32(pb); /* always 0x7F000000 */
1891 st->codecpar->bits_per_coded_sample = avio_rb32(pb);
1893 flags = avio_rb32(pb); /* lpcm format specific flag */
1894 sc->bytes_per_frame = avio_rb32(pb);
1895 sc->samples_per_frame = avio_rb32(pb);
1896 if (st->codecpar->codec_tag == MKTAG('l','p','c','m'))
1897 st->codecpar->codec_id =
1898 ff_mov_get_lpcm_codec_id(st->codecpar->bits_per_coded_sample,
1901 if (version == 0 || (version == 1 && sc->audio_cid != -2)) {
1902 /* can't correctly handle variable sized packet as audio unit */
1903 switch (st->codecpar->codec_id) {
1904 case AV_CODEC_ID_MP2:
1905 case AV_CODEC_ID_MP3:
1906 st->need_parsing = AVSTREAM_PARSE_FULL;
1912 if (sc->format == 0) {
1913 if (st->codecpar->bits_per_coded_sample == 8)
1914 st->codecpar->codec_id = mov_codec_id(st, MKTAG('r','a','w',' '));
1915 else if (st->codecpar->bits_per_coded_sample == 16)
1916 st->codecpar->codec_id = mov_codec_id(st, MKTAG('t','w','o','s'));
1919 switch (st->codecpar->codec_id) {
1920 case AV_CODEC_ID_PCM_S8:
1921 case AV_CODEC_ID_PCM_U8:
1922 if (st->codecpar->bits_per_coded_sample == 16)
1923 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16BE;
1925 case AV_CODEC_ID_PCM_S16LE:
1926 case AV_CODEC_ID_PCM_S16BE:
1927 if (st->codecpar->bits_per_coded_sample == 8)
1928 st->codecpar->codec_id = AV_CODEC_ID_PCM_S8;
1929 else if (st->codecpar->bits_per_coded_sample == 24)
1930 st->codecpar->codec_id =
1931 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
1932 AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
1933 else if (st->codecpar->bits_per_coded_sample == 32)
1934 st->codecpar->codec_id =
1935 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
1936 AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
1938 /* set values for old format before stsd version 1 appeared */
1939 case AV_CODEC_ID_MACE3:
1940 sc->samples_per_frame = 6;
1941 sc->bytes_per_frame = 2 * st->codecpar->channels;
1943 case AV_CODEC_ID_MACE6:
1944 sc->samples_per_frame = 6;
1945 sc->bytes_per_frame = 1 * st->codecpar->channels;
1947 case AV_CODEC_ID_ADPCM_IMA_QT:
1948 sc->samples_per_frame = 64;
1949 sc->bytes_per_frame = 34 * st->codecpar->channels;
1951 case AV_CODEC_ID_GSM:
1952 sc->samples_per_frame = 160;
1953 sc->bytes_per_frame = 33;
1959 bits_per_sample = av_get_bits_per_sample(st->codecpar->codec_id);
1960 if (bits_per_sample) {
1961 st->codecpar->bits_per_coded_sample = bits_per_sample;
1962 sc->sample_size = (bits_per_sample >> 3) * st->codecpar->channels;
1966 static void mov_parse_stsd_subtitle(MOVContext *c, AVIOContext *pb,
1967 AVStream *st, MOVStreamContext *sc,
1970 // ttxt stsd contains display flags, justification, background
1971 // color, fonts, and default styles, so fake an atom to read it
1972 MOVAtom fake_atom = { .size = size };
1973 // mp4s contains a regular esds atom
1974 if (st->codecpar->codec_tag != AV_RL32("mp4s"))
1975 mov_read_glbl(c, pb, fake_atom);
1976 st->codecpar->width = sc->width;
1977 st->codecpar->height = sc->height;
1980 static uint32_t yuv_to_rgba(uint32_t ycbcr)
1985 y = (ycbcr >> 16) & 0xFF;
1986 cr = (ycbcr >> 8) & 0xFF;
1989 b = av_clip_uint8((1164 * (y - 16) + 2018 * (cb - 128)) / 1000);
1990 g = av_clip_uint8((1164 * (y - 16) - 813 * (cr - 128) - 391 * (cb - 128)) / 1000);
1991 r = av_clip_uint8((1164 * (y - 16) + 1596 * (cr - 128) ) / 1000);
1993 return (r << 16) | (g << 8) | b;
1996 static int mov_rewrite_dvd_sub_extradata(AVStream *st)
1998 char buf[256] = {0};
1999 uint8_t *src = st->codecpar->extradata;
2002 if (st->codecpar->extradata_size != 64)
2005 if (st->codecpar->width > 0 && st->codecpar->height > 0)
2006 snprintf(buf, sizeof(buf), "size: %dx%d\n",
2007 st->codecpar->width, st->codecpar->height);
2008 av_strlcat(buf, "palette: ", sizeof(buf));
2010 for (i = 0; i < 16; i++) {
2011 uint32_t yuv = AV_RB32(src + i * 4);
2012 uint32_t rgba = yuv_to_rgba(yuv);
2014 av_strlcatf(buf, sizeof(buf), "%06"PRIx32"%s", rgba, i != 15 ? ", " : "");
2017 if (av_strlcat(buf, "\n", sizeof(buf)) >= sizeof(buf))
2020 av_freep(&st->codecpar->extradata);
2021 st->codecpar->extradata_size = 0;
2022 st->codecpar->extradata = av_mallocz(strlen(buf) + AV_INPUT_BUFFER_PADDING_SIZE);
2023 if (!st->codecpar->extradata)
2024 return AVERROR(ENOMEM);
2025 st->codecpar->extradata_size = strlen(buf);
2026 memcpy(st->codecpar->extradata, buf, st->codecpar->extradata_size);
2031 static int mov_parse_stsd_data(MOVContext *c, AVIOContext *pb,
2032 AVStream *st, MOVStreamContext *sc,
2037 if (st->codecpar->codec_tag == MKTAG('t','m','c','d')) {
2038 if ((int)size != size)
2039 return AVERROR(ENOMEM);
2041 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
2045 MOVStreamContext *tmcd_ctx = st->priv_data;
2047 val = AV_RB32(st->codecpar->extradata + 4);
2048 tmcd_ctx->tmcd_flags = val;
2049 st->avg_frame_rate.num = st->codecpar->extradata[16]; /* number of frame */
2050 st->avg_frame_rate.den = 1;
2051 #if FF_API_LAVF_AVCTX
2052 FF_DISABLE_DEPRECATION_WARNINGS
2053 st->codec->time_base = av_inv_q(st->avg_frame_rate);
2054 FF_ENABLE_DEPRECATION_WARNINGS
2056 /* adjust for per frame dur in counter mode */
2057 if (tmcd_ctx->tmcd_flags & 0x0008) {
2058 int timescale = AV_RB32(st->codecpar->extradata + 8);
2059 int framedur = AV_RB32(st->codecpar->extradata + 12);
2060 st->avg_frame_rate.num *= timescale;
2061 st->avg_frame_rate.den *= framedur;
2062 #if FF_API_LAVF_AVCTX
2063 FF_DISABLE_DEPRECATION_WARNINGS
2064 st->codec->time_base.den *= timescale;
2065 st->codec->time_base.num *= framedur;
2066 FF_ENABLE_DEPRECATION_WARNINGS
2070 uint32_t len = AV_RB32(st->codecpar->extradata + 18); /* name atom length */
2071 uint32_t format = AV_RB32(st->codecpar->extradata + 22);
2072 if (format == AV_RB32("name") && (int64_t)size >= (int64_t)len + 18) {
2073 uint16_t str_size = AV_RB16(st->codecpar->extradata + 26); /* string length */
2074 if (str_size > 0 && size >= (int)str_size + 26) {
2075 char *reel_name = av_malloc(str_size + 1);
2077 return AVERROR(ENOMEM);
2078 memcpy(reel_name, st->codecpar->extradata + 30, str_size);
2079 reel_name[str_size] = 0; /* Add null terminator */
2080 /* don't add reel_name if emtpy string */
2081 if (*reel_name == 0) {
2084 av_dict_set(&st->metadata, "reel_name", reel_name, AV_DICT_DONT_STRDUP_VAL);
2091 /* other codec type, just skip (rtp, mp4s ...) */
2092 avio_skip(pb, size);
2097 static int mov_finalize_stsd_codec(MOVContext *c, AVIOContext *pb,
2098 AVStream *st, MOVStreamContext *sc)
2100 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
2101 !st->codecpar->sample_rate && sc->time_scale > 1)
2102 st->codecpar->sample_rate = sc->time_scale;
2104 /* special codec parameters handling */
2105 switch (st->codecpar->codec_id) {
2106 #if CONFIG_DV_DEMUXER
2107 case AV_CODEC_ID_DVAUDIO:
2108 c->dv_fctx = avformat_alloc_context();
2110 av_log(c->fc, AV_LOG_ERROR, "dv demux context alloc error\n");
2111 return AVERROR(ENOMEM);
2113 c->dv_demux = avpriv_dv_init_demux(c->dv_fctx);
2115 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
2116 return AVERROR(ENOMEM);
2118 sc->dv_audio_container = 1;
2119 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16LE;
2122 /* no ifdef since parameters are always those */
2123 case AV_CODEC_ID_QCELP:
2124 st->codecpar->channels = 1;
2125 // force sample rate for qcelp when not stored in mov
2126 if (st->codecpar->codec_tag != MKTAG('Q','c','l','p'))
2127 st->codecpar->sample_rate = 8000;
2128 // FIXME: Why is the following needed for some files?
2129 sc->samples_per_frame = 160;
2130 if (!sc->bytes_per_frame)
2131 sc->bytes_per_frame = 35;
2133 case AV_CODEC_ID_AMR_NB:
2134 st->codecpar->channels = 1;
2135 /* force sample rate for amr, stsd in 3gp does not store sample rate */
2136 st->codecpar->sample_rate = 8000;
2138 case AV_CODEC_ID_AMR_WB:
2139 st->codecpar->channels = 1;
2140 st->codecpar->sample_rate = 16000;
2142 case AV_CODEC_ID_MP2:
2143 case AV_CODEC_ID_MP3:
2144 /* force type after stsd for m1a hdlr */
2145 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2147 case AV_CODEC_ID_GSM:
2148 case AV_CODEC_ID_ADPCM_MS:
2149 case AV_CODEC_ID_ADPCM_IMA_WAV:
2150 case AV_CODEC_ID_ILBC:
2151 case AV_CODEC_ID_MACE3:
2152 case AV_CODEC_ID_MACE6:
2153 case AV_CODEC_ID_QDM2:
2154 st->codecpar->block_align = sc->bytes_per_frame;
2156 case AV_CODEC_ID_ALAC:
2157 if (st->codecpar->extradata_size == 36) {
2158 st->codecpar->channels = AV_RB8 (st->codecpar->extradata + 21);
2159 st->codecpar->sample_rate = AV_RB32(st->codecpar->extradata + 32);
2162 case AV_CODEC_ID_AC3:
2163 case AV_CODEC_ID_EAC3:
2164 case AV_CODEC_ID_MPEG1VIDEO:
2165 case AV_CODEC_ID_VC1:
2166 case AV_CODEC_ID_VP9:
2167 st->need_parsing = AVSTREAM_PARSE_FULL;
2175 static int mov_skip_multiple_stsd(MOVContext *c, AVIOContext *pb,
2176 int codec_tag, int format,
2179 int video_codec_id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2182 (codec_tag != format &&
2183 // prores is allowed to have differing data format and codec tag
2184 codec_tag != AV_RL32("apcn") && codec_tag != AV_RL32("apch") &&
2185 (c->fc->video_codec_id ? video_codec_id != c->fc->video_codec_id
2186 : codec_tag != MKTAG('j','p','e','g')))) {
2187 /* Multiple fourcc, we skip JPEG. This is not correct, we should
2188 * export it as a separate AVStream but this needs a few changes
2189 * in the MOV demuxer, patch welcome. */
2191 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
2192 avio_skip(pb, size);
2195 if ( codec_tag == AV_RL32("hvc1") ||
2196 codec_tag == AV_RL32("hev1")
2198 av_log(c->fc, AV_LOG_WARNING, "Concatenated H.264 or H.265 might not play correctly.\n");
2203 int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries)
2206 MOVStreamContext *sc;
2207 int pseudo_stream_id;
2209 if (c->fc->nb_streams < 1)
2211 st = c->fc->streams[c->fc->nb_streams-1];
2214 for (pseudo_stream_id = 0;
2215 pseudo_stream_id < entries && !pb->eof_reached;
2216 pseudo_stream_id++) {
2217 //Parsing Sample description table
2219 int ret, dref_id = 1;
2220 MOVAtom a = { AV_RL32("stsd") };
2221 int64_t start_pos = avio_tell(pb);
2222 int64_t size = avio_rb32(pb); /* size */
2223 uint32_t format = avio_rl32(pb); /* data format */
2226 avio_rb32(pb); /* reserved */
2227 avio_rb16(pb); /* reserved */
2228 dref_id = avio_rb16(pb);
2229 } else if (size <= 7) {
2230 av_log(c->fc, AV_LOG_ERROR,
2231 "invalid size %"PRId64" in stsd\n", size);
2232 return AVERROR_INVALIDDATA;
2235 if (mov_skip_multiple_stsd(c, pb, st->codecpar->codec_tag, format,
2236 size - (avio_tell(pb) - start_pos)))
2239 sc->pseudo_stream_id = st->codecpar->codec_tag ? -1 : pseudo_stream_id;
2240 sc->dref_id= dref_id;
2241 sc->format = format;
2243 id = mov_codec_id(st, format);
2245 av_log(c->fc, AV_LOG_TRACE,
2246 "size=%"PRId64" 4CC= %c%c%c%c/0x%08x codec_type=%d\n", size,
2247 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff,
2248 (format >> 24) & 0xff, format, st->codecpar->codec_type);
2250 if (st->codecpar->codec_type==AVMEDIA_TYPE_VIDEO) {
2251 st->codecpar->codec_id = id;
2252 mov_parse_stsd_video(c, pb, st, sc);
2253 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_AUDIO) {
2254 st->codecpar->codec_id = id;
2255 mov_parse_stsd_audio(c, pb, st, sc);
2256 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_SUBTITLE){
2257 st->codecpar->codec_id = id;
2258 mov_parse_stsd_subtitle(c, pb, st, sc,
2259 size - (avio_tell(pb) - start_pos));
2261 ret = mov_parse_stsd_data(c, pb, st, sc,
2262 size - (avio_tell(pb) - start_pos));
2266 /* this will read extra atoms at the end (wave, alac, damr, avcC, hvcC, SMI ...) */
2267 a.size = size - (avio_tell(pb) - start_pos);
2269 if ((ret = mov_read_default(c, pb, a)) < 0)
2271 } else if (a.size > 0)
2272 avio_skip(pb, a.size);
2274 if (sc->extradata) {
2275 int extra_size = st->codecpar->extradata_size;
2277 /* Move the current stream extradata to the stream context one. */
2278 sc->extradata_size[pseudo_stream_id] = extra_size;
2279 sc->extradata[pseudo_stream_id] = av_malloc(extra_size + AV_INPUT_BUFFER_PADDING_SIZE);
2280 if (!sc->extradata[pseudo_stream_id])
2281 return AVERROR(ENOMEM);
2282 memcpy(sc->extradata[pseudo_stream_id], st->codecpar->extradata, extra_size);
2283 av_freep(&st->codecpar->extradata);
2284 st->codecpar->extradata_size = 0;
2288 if (pb->eof_reached)
2291 return mov_finalize_stsd_codec(c, pb, st, sc);
2294 static int mov_read_stsd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2297 MOVStreamContext *sc;
2301 if (c->fc->nb_streams < 1)
2303 st = c->fc->streams[c->fc->nb_streams - 1];
2306 avio_r8(pb); /* version */
2307 avio_rb24(pb); /* flags */
2308 entries = avio_rb32(pb); /* entries */
2311 av_log(c->fc, AV_LOG_ERROR, "invalid STSD entries %d\n", entries);
2312 return AVERROR_INVALIDDATA;
2315 if (sc->extradata) {
2316 av_log(c->fc, AV_LOG_ERROR, "Duplicate STSD\n");
2317 return AVERROR_INVALIDDATA;
2319 /* Prepare space for hosting multiple extradata. */
2320 sc->extradata = av_mallocz_array(entries, sizeof(*sc->extradata));
2321 sc->extradata_size = av_mallocz_array(entries, sizeof(*sc->extradata_size));
2322 if (!sc->extradata_size || !sc->extradata) {
2323 ret = AVERROR(ENOMEM);
2327 ret = ff_mov_read_stsd_entries(c, pb, entries);
2331 sc->stsd_count = entries;
2333 /* Restore back the primary extradata. */
2334 av_freep(&st->codecpar->extradata);
2335 st->codecpar->extradata_size = sc->extradata_size[0];
2336 if (sc->extradata_size[0]) {
2337 st->codecpar->extradata = av_mallocz(sc->extradata_size[0] + AV_INPUT_BUFFER_PADDING_SIZE);
2338 if (!st->codecpar->extradata)
2339 return AVERROR(ENOMEM);
2340 memcpy(st->codecpar->extradata, sc->extradata[0], sc->extradata_size[0]);
2345 av_freep(&sc->extradata);
2346 av_freep(&sc->extradata_size);
2350 static int mov_read_stsc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2353 MOVStreamContext *sc;
2354 unsigned int i, entries;
2356 if (c->fc->nb_streams < 1)
2358 st = c->fc->streams[c->fc->nb_streams-1];
2361 avio_r8(pb); /* version */
2362 avio_rb24(pb); /* flags */
2364 entries = avio_rb32(pb);
2366 av_log(c->fc, AV_LOG_TRACE, "track[%i].stsc.entries = %i\n", c->fc->nb_streams-1, entries);
2371 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSC atom\n");
2372 av_free(sc->stsc_data);
2374 sc->stsc_data = av_malloc_array(entries, sizeof(*sc->stsc_data));
2376 return AVERROR(ENOMEM);
2378 for (i = 0; i < entries && !pb->eof_reached; i++) {
2379 sc->stsc_data[i].first = avio_rb32(pb);
2380 sc->stsc_data[i].count = avio_rb32(pb);
2381 sc->stsc_data[i].id = avio_rb32(pb);
2386 if (pb->eof_reached)
2392 /* Compute the samples value for the stsc entry at the given index. */
2393 static inline int mov_get_stsc_samples(MOVStreamContext *sc, int index)
2397 if (index < sc->stsc_count - 1)
2398 chunk_count = sc->stsc_data[index + 1].first - sc->stsc_data[index].first;
2400 chunk_count = sc->chunk_count - (sc->stsc_data[index].first - 1);
2402 return sc->stsc_data[index].count * chunk_count;
2405 static int mov_read_stps(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2408 MOVStreamContext *sc;
2409 unsigned i, entries;
2411 if (c->fc->nb_streams < 1)
2413 st = c->fc->streams[c->fc->nb_streams-1];
2416 avio_rb32(pb); // version + flags
2418 entries = avio_rb32(pb);
2420 av_log(c->fc, AV_LOG_WARNING, "Duplicated STPS atom\n");
2421 av_free(sc->stps_data);
2423 sc->stps_data = av_malloc_array(entries, sizeof(*sc->stps_data));
2425 return AVERROR(ENOMEM);
2427 for (i = 0; i < entries && !pb->eof_reached; i++) {
2428 sc->stps_data[i] = avio_rb32(pb);
2429 //av_log(c->fc, AV_LOG_TRACE, "stps %d\n", sc->stps_data[i]);
2434 if (pb->eof_reached)
2440 static int mov_read_stss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2443 MOVStreamContext *sc;
2444 unsigned int i, entries;
2446 if (c->fc->nb_streams < 1)
2448 st = c->fc->streams[c->fc->nb_streams-1];
2451 avio_r8(pb); /* version */
2452 avio_rb24(pb); /* flags */
2454 entries = avio_rb32(pb);
2456 av_log(c->fc, AV_LOG_TRACE, "keyframe_count = %d\n", entries);
2460 sc->keyframe_absent = 1;
2461 if (!st->need_parsing && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
2462 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2466 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSS atom\n");
2467 if (entries >= UINT_MAX / sizeof(int))
2468 return AVERROR_INVALIDDATA;
2469 av_freep(&sc->keyframes);
2470 sc->keyframe_count = 0;
2471 sc->keyframes = av_malloc_array(entries, sizeof(*sc->keyframes));
2473 return AVERROR(ENOMEM);
2475 for (i = 0; i < entries && !pb->eof_reached; i++) {
2476 sc->keyframes[i] = avio_rb32(pb);
2477 //av_log(c->fc, AV_LOG_TRACE, "keyframes[]=%d\n", sc->keyframes[i]);
2480 sc->keyframe_count = i;
2482 if (pb->eof_reached)
2488 static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2491 MOVStreamContext *sc;
2492 unsigned int i, entries, sample_size, field_size, num_bytes;
2497 if (c->fc->nb_streams < 1)
2499 st = c->fc->streams[c->fc->nb_streams-1];
2502 avio_r8(pb); /* version */
2503 avio_rb24(pb); /* flags */
2505 if (atom.type == MKTAG('s','t','s','z')) {
2506 sample_size = avio_rb32(pb);
2507 if (!sc->sample_size) /* do not overwrite value computed in stsd */
2508 sc->sample_size = sample_size;
2509 sc->stsz_sample_size = sample_size;
2513 avio_rb24(pb); /* reserved */
2514 field_size = avio_r8(pb);
2516 entries = avio_rb32(pb);
2518 av_log(c->fc, AV_LOG_TRACE, "sample_size = %d sample_count = %d\n", sc->sample_size, entries);
2520 sc->sample_count = entries;
2524 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
2525 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %d\n", field_size);
2526 return AVERROR_INVALIDDATA;
2531 if (entries >= (UINT_MAX - 4) / field_size)
2532 return AVERROR_INVALIDDATA;
2533 if (sc->sample_sizes)
2534 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSZ atom\n");
2535 av_free(sc->sample_sizes);
2536 sc->sample_count = 0;
2537 sc->sample_sizes = av_malloc_array(entries, sizeof(*sc->sample_sizes));
2538 if (!sc->sample_sizes)
2539 return AVERROR(ENOMEM);
2541 num_bytes = (entries*field_size+4)>>3;
2543 buf = av_malloc(num_bytes+AV_INPUT_BUFFER_PADDING_SIZE);
2545 av_freep(&sc->sample_sizes);
2546 return AVERROR(ENOMEM);
2549 ret = ffio_read_size(pb, buf, num_bytes);
2551 av_freep(&sc->sample_sizes);
2556 init_get_bits(&gb, buf, 8*num_bytes);
2558 for (i = 0; i < entries && !pb->eof_reached; i++) {
2559 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
2560 sc->data_size += sc->sample_sizes[i];
2563 sc->sample_count = i;
2567 if (pb->eof_reached)
2573 static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2576 MOVStreamContext *sc;
2577 unsigned int i, entries;
2579 int64_t total_sample_count=0;
2581 if (c->fc->nb_streams < 1)
2583 st = c->fc->streams[c->fc->nb_streams-1];
2586 avio_r8(pb); /* version */
2587 avio_rb24(pb); /* flags */
2588 entries = avio_rb32(pb);
2590 av_log(c->fc, AV_LOG_TRACE, "track[%i].stts.entries = %i\n",
2591 c->fc->nb_streams-1, entries);
2594 av_log(c->fc, AV_LOG_WARNING, "Duplicated STTS atom\n");
2595 av_free(sc->stts_data);
2597 sc->stts_data = av_malloc_array(entries, sizeof(*sc->stts_data));
2599 return AVERROR(ENOMEM);
2601 for (i = 0; i < entries && !pb->eof_reached; i++) {
2602 int sample_duration;
2605 sample_count=avio_rb32(pb);
2606 sample_duration = avio_rb32(pb);
2608 if (sample_count < 0) {
2609 av_log(c->fc, AV_LOG_ERROR, "Invalid sample_count=%d\n", sample_count);
2610 return AVERROR_INVALIDDATA;
2612 sc->stts_data[i].count= sample_count;
2613 sc->stts_data[i].duration= sample_duration;
2615 av_log(c->fc, AV_LOG_TRACE, "sample_count=%d, sample_duration=%d\n",
2616 sample_count, sample_duration);
2620 && sample_count == 1
2621 && total_sample_count > 100
2622 && sample_duration/10 > duration / total_sample_count)
2623 sample_duration = duration / total_sample_count;
2624 duration+=(int64_t)sample_duration*sample_count;
2625 total_sample_count+=sample_count;
2630 sc->duration_for_fps += duration;
2631 sc->nb_frames_for_fps += total_sample_count;
2633 if (pb->eof_reached)
2636 st->nb_frames= total_sample_count;
2638 st->duration= duration;
2639 sc->track_end = duration;
2643 static void mov_update_dts_shift(MOVStreamContext *sc, int duration)
2646 if (duration == INT_MIN) {
2647 av_log(NULL, AV_LOG_WARNING, "mov_update_dts_shift(): dts_shift set to %d\n", INT_MAX);
2650 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
2654 static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2657 MOVStreamContext *sc;
2658 unsigned int i, entries, ctts_count = 0;
2660 if (c->fc->nb_streams < 1)
2662 st = c->fc->streams[c->fc->nb_streams-1];
2665 avio_r8(pb); /* version */
2666 avio_rb24(pb); /* flags */
2667 entries = avio_rb32(pb);
2669 av_log(c->fc, AV_LOG_TRACE, "track[%i].ctts.entries = %i\n", c->fc->nb_streams-1, entries);
2673 if (entries >= UINT_MAX / sizeof(*sc->ctts_data))
2674 return AVERROR_INVALIDDATA;
2675 av_freep(&sc->ctts_data);
2676 sc->ctts_data = av_realloc(NULL, entries * sizeof(*sc->ctts_data));
2678 return AVERROR(ENOMEM);
2680 for (i = 0; i < entries && !pb->eof_reached; i++) {
2681 int count =avio_rb32(pb);
2682 int duration =avio_rb32(pb);
2685 av_log(c->fc, AV_LOG_TRACE,
2686 "ignoring CTTS entry with count=%d duration=%d\n",
2691 sc->ctts_data[ctts_count].count = count;
2692 sc->ctts_data[ctts_count].duration = duration;
2695 av_log(c->fc, AV_LOG_TRACE, "count=%d, duration=%d\n",
2698 if (FFNABS(duration) < -(1<<28) && i+2<entries) {
2699 av_log(c->fc, AV_LOG_WARNING, "CTTS invalid\n");
2700 av_freep(&sc->ctts_data);
2706 mov_update_dts_shift(sc, duration);
2709 sc->ctts_count = ctts_count;
2711 if (pb->eof_reached)
2714 av_log(c->fc, AV_LOG_TRACE, "dts shift %d\n", sc->dts_shift);
2719 static int mov_read_sbgp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2722 MOVStreamContext *sc;
2723 unsigned int i, entries;
2725 uint32_t grouping_type;
2727 if (c->fc->nb_streams < 1)
2729 st = c->fc->streams[c->fc->nb_streams-1];
2732 version = avio_r8(pb); /* version */
2733 avio_rb24(pb); /* flags */
2734 grouping_type = avio_rl32(pb);
2735 if (grouping_type != MKTAG( 'r','a','p',' '))
2736 return 0; /* only support 'rap ' grouping */
2738 avio_rb32(pb); /* grouping_type_parameter */
2740 entries = avio_rb32(pb);
2744 av_log(c->fc, AV_LOG_WARNING, "Duplicated SBGP atom\n");
2745 av_free(sc->rap_group);
2746 sc->rap_group_count = 0;
2747 sc->rap_group = av_malloc_array(entries, sizeof(*sc->rap_group));
2749 return AVERROR(ENOMEM);
2751 for (i = 0; i < entries && !pb->eof_reached; i++) {
2752 sc->rap_group[i].count = avio_rb32(pb); /* sample_count */
2753 sc->rap_group[i].index = avio_rb32(pb); /* group_description_index */
2756 sc->rap_group_count = i;
2758 return pb->eof_reached ? AVERROR_EOF : 0;
2762 * Get ith edit list entry (media time, duration).
2764 static int get_edit_list_entry(const MOVStreamContext *msc,
2765 unsigned int edit_list_index,
2766 int64_t *edit_list_media_time,
2767 int64_t *edit_list_duration,
2768 int64_t global_timescale)
2770 if (edit_list_index == msc->elst_count) {
2773 *edit_list_media_time = msc->elst_data[edit_list_index].time;
2774 *edit_list_duration = msc->elst_data[edit_list_index].duration;
2776 /* duration is in global timescale units;convert to msc timescale */
2777 if (global_timescale == 0) {
2778 avpriv_request_sample(msc, "Support for mvhd.timescale = 0 with editlists");
2781 *edit_list_duration = av_rescale(*edit_list_duration, msc->time_scale,
2787 * Find the closest previous keyframe to the timestamp, in e_old index
2789 * Returns the index of the entry in st->index_entries if successful,
2792 static int64_t find_prev_closest_keyframe_index(AVStream *st,
2793 AVIndexEntry *e_old,
2798 AVIndexEntry *e_keep = st->index_entries;
2799 int nb_keep = st->nb_index_entries;
2802 st->index_entries = e_old;
2803 st->nb_index_entries = nb_old;
2804 found = av_index_search_timestamp(st, timestamp, flag | AVSEEK_FLAG_BACKWARD);
2806 /* restore AVStream state*/
2807 st->index_entries = e_keep;
2808 st->nb_index_entries = nb_keep;
2813 * Add index entry with the given values, to the end of st->index_entries.
2814 * Returns the new size st->index_entries if successful, else returns -1.
2816 * This function is similar to ff_add_index_entry in libavformat/utils.c
2817 * except that here we are always unconditionally adding an index entry to
2818 * the end, instead of searching the entries list and skipping the add if
2819 * there is an existing entry with the same timestamp.
2820 * This is needed because the mov_fix_index calls this func with the same
2821 * unincremented timestamp for successive discarded frames.
2823 static int64_t add_index_entry(AVStream *st, int64_t pos, int64_t timestamp,
2824 int size, int distance, int flags)
2826 AVIndexEntry *entries, *ie;
2828 const size_t min_size_needed = (st->nb_index_entries + 1) * sizeof(AVIndexEntry);
2830 // Double the allocation each time, to lower memory fragmentation.
2831 // Another difference from ff_add_index_entry function.
2832 const size_t requested_size =
2833 min_size_needed > st->index_entries_allocated_size ?
2834 FFMAX(min_size_needed, 2 * st->index_entries_allocated_size) :
2837 if((unsigned)st->nb_index_entries + 1 >= UINT_MAX / sizeof(AVIndexEntry))
2840 entries = av_fast_realloc(st->index_entries,
2841 &st->index_entries_allocated_size,
2846 st->index_entries= entries;
2848 index= st->nb_index_entries++;
2849 ie= &entries[index];
2852 ie->timestamp = timestamp;
2853 ie->min_distance= distance;
2860 * Rewrite timestamps of index entries in the range [end_index - frame_duration_buffer_size, end_index)
2861 * by subtracting end_ts successively by the amounts given in frame_duration_buffer.
2863 static void fix_index_entry_timestamps(AVStream* st, int end_index, int64_t end_ts,
2864 int64_t* frame_duration_buffer,
2865 int frame_duration_buffer_size) {
2867 av_assert0(end_index >= 0 && end_index <= st->nb_index_entries);
2868 for (i = 0; i < frame_duration_buffer_size; i++) {
2869 end_ts -= frame_duration_buffer[frame_duration_buffer_size - 1 - i];
2870 st->index_entries[end_index - 1 - i].timestamp = end_ts;
2875 * Append a new ctts entry to ctts_data.
2876 * Returns the new ctts_count if successful, else returns -1.
2878 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
2879 int count, int duration)
2881 MOVStts *ctts_buf_new;
2882 const size_t min_size_needed = (*ctts_count + 1) * sizeof(MOVStts);
2883 const size_t requested_size =
2884 min_size_needed > *allocated_size ?
2885 FFMAX(min_size_needed, 2 * (*allocated_size)) :
2888 if((unsigned)(*ctts_count) + 1 >= UINT_MAX / sizeof(MOVStts))
2891 ctts_buf_new = av_fast_realloc(*ctts_data, allocated_size, requested_size);
2896 *ctts_data = ctts_buf_new;
2898 ctts_buf_new[*ctts_count].count = count;
2899 ctts_buf_new[*ctts_count].duration = duration;
2901 *ctts_count = (*ctts_count) + 1;
2906 * Fix st->index_entries, so that it contains only the entries (and the entries
2907 * which are needed to decode them) that fall in the edit list time ranges.
2908 * Also fixes the timestamps of the index entries to match the timeline
2909 * specified the edit lists.
2911 static void mov_fix_index(MOVContext *mov, AVStream *st)
2913 MOVStreamContext *msc = st->priv_data;
2914 AVIndexEntry *e_old = st->index_entries;
2915 int nb_old = st->nb_index_entries;
2916 const AVIndexEntry *e_old_end = e_old + nb_old;
2917 const AVIndexEntry *current = NULL;
2918 MOVStts *ctts_data_old = msc->ctts_data;
2919 int64_t ctts_index_old = 0;
2920 int64_t ctts_sample_old = 0;
2921 int64_t ctts_count_old = msc->ctts_count;
2922 int64_t edit_list_media_time = 0;
2923 int64_t edit_list_duration = 0;
2924 int64_t frame_duration = 0;
2925 int64_t edit_list_dts_counter = 0;
2926 int64_t edit_list_dts_entry_end = 0;
2927 int64_t edit_list_start_ctts_sample = 0;
2929 int64_t edit_list_index = 0;
2931 int64_t index_ctts_count;
2933 unsigned int ctts_allocated_size = 0;
2934 int64_t start_dts = 0;
2935 int64_t edit_list_media_time_dts = 0;
2936 int64_t edit_list_start_encountered = 0;
2937 int64_t search_timestamp = 0;
2938 int64_t* frame_duration_buffer = NULL;
2939 int num_discarded_begin = 0;
2940 int first_non_zero_audio_edit = -1;
2941 int packet_skip_samples = 0;
2943 if (!msc->elst_data || msc->elst_count <= 0) {
2946 // Clean AVStream from traces of old index
2947 st->index_entries = NULL;
2948 st->index_entries_allocated_size = 0;
2949 st->nb_index_entries = 0;
2951 // Clean ctts fields of MOVStreamContext
2952 msc->ctts_data = NULL;
2953 msc->ctts_count = 0;
2954 msc->ctts_index = 0;
2955 msc->ctts_sample = 0;
2957 // If the dts_shift is positive (in case of negative ctts values in mov),
2958 // then negate the DTS by dts_shift
2959 if (msc->dts_shift > 0)
2960 edit_list_dts_entry_end -= msc->dts_shift;
2962 // Offset the DTS by ctts[0] to make the PTS of the first frame 0
2963 if (ctts_data_old && ctts_count_old > 0) {
2964 edit_list_dts_entry_end -= ctts_data_old[0].duration;
2965 av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by ctts[%d].duration: %d\n", 0, ctts_data_old[0].duration);
2968 start_dts = edit_list_dts_entry_end;
2970 while (get_edit_list_entry(msc, edit_list_index, &edit_list_media_time,
2971 &edit_list_duration, mov->time_scale)) {
2972 av_log(mov->fc, AV_LOG_DEBUG, "Processing st: %d, edit list %"PRId64" - media time: %"PRId64", duration: %"PRId64"\n",
2973 st->index, edit_list_index, edit_list_media_time, edit_list_duration);
2975 edit_list_dts_counter = edit_list_dts_entry_end;
2976 edit_list_dts_entry_end += edit_list_duration;
2977 num_discarded_begin = 0;
2978 if (edit_list_media_time == -1) {
2982 // If we encounter a non-negative edit list reset the skip_samples/start_pad fields and set them
2983 // according to the edit list below.
2984 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
2985 if (first_non_zero_audio_edit < 0) {
2986 first_non_zero_audio_edit = 1;
2988 first_non_zero_audio_edit = 0;
2991 if (first_non_zero_audio_edit > 0)
2992 st->skip_samples = msc->start_pad = 0;
2995 //find closest previous key frame
2996 edit_list_media_time_dts = edit_list_media_time;
2997 if (msc->dts_shift > 0) {
2998 edit_list_media_time_dts -= msc->dts_shift;
3001 // While reordering frame index according to edit list we must handle properly
3002 // the scenario when edit list entry starts from none key frame.
3003 // We find closest previous key frame and preserve it and consequent frames in index.
3004 // All frames which are outside edit list entry time boundaries will be dropped after decoding.
3005 search_timestamp = edit_list_media_time_dts;
3006 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3007 // Audio decoders like AAC need need a decoder delay samples previous to the current sample,
3008 // to correctly decode this frame. Hence for audio we seek to a frame 1 sec. before the
3009 // edit_list_media_time to cover the decoder delay.
3010 search_timestamp = FFMAX(search_timestamp - mov->time_scale, e_old[0].timestamp);
3013 index = find_prev_closest_keyframe_index(st, e_old, nb_old, search_timestamp, 0);
3015 av_log(mov->fc, AV_LOG_ERROR, "Missing key frame while reordering index according to edit list\n");
3018 current = e_old + index;
3021 ctts_sample_old = 0;
3023 // set ctts_index properly for the found key frame
3024 for (index_ctts_count = 0; index_ctts_count < index; index_ctts_count++) {
3025 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3027 if (ctts_data_old[ctts_index_old].count == ctts_sample_old) {
3029 ctts_sample_old = 0;
3034 edit_list_start_ctts_sample = ctts_sample_old;
3036 // Iterate over index and arrange it according to edit list
3037 edit_list_start_encountered = 0;
3038 for (; current < e_old_end; current++, index++) {
3039 // check if frame outside edit list mark it for discard
3040 frame_duration = (current + 1 < e_old_end) ?
3041 ((current + 1)->timestamp - current->timestamp) : edit_list_duration;
3043 flags = current->flags;
3045 // frames (pts) before or after edit list
3046 curr_cts = current->timestamp + msc->dts_shift;
3048 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3049 av_log(mov->fc, AV_LOG_DEBUG, "shifted frame pts, curr_cts: %"PRId64" @ %"PRId64", ctts: %d, ctts_count: %"PRId64"\n",
3050 curr_cts, ctts_index_old, ctts_data_old[ctts_index_old].duration, ctts_count_old);
3051 curr_cts += ctts_data_old[ctts_index_old].duration;
3053 if (ctts_sample_old == ctts_data_old[ctts_index_old].count) {
3054 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3055 &ctts_allocated_size,
3056 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3057 ctts_data_old[ctts_index_old].duration) == -1) {
3058 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3060 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3061 ctts_data_old[ctts_index_old].duration);
3065 ctts_sample_old = 0;
3066 edit_list_start_ctts_sample = 0;
3070 if (curr_cts < edit_list_media_time || curr_cts >= (edit_list_duration + edit_list_media_time)) {
3071 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id != AV_CODEC_ID_VORBIS &&
3072 curr_cts < edit_list_media_time && curr_cts + frame_duration > edit_list_media_time &&
3073 first_non_zero_audio_edit > 0) {
3074 packet_skip_samples = edit_list_media_time - curr_cts;
3075 st->skip_samples += packet_skip_samples;
3077 // Shift the index entry timestamp by packet_skip_samples to be correct.
3078 edit_list_dts_counter -= packet_skip_samples;
3079 if (edit_list_start_encountered == 0) {
3080 edit_list_start_encountered = 1;
3081 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3082 // discarded packets.
3083 if (frame_duration_buffer) {
3084 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3085 frame_duration_buffer, num_discarded_begin);
3086 av_freep(&frame_duration_buffer);
3090 av_log(mov->fc, AV_LOG_DEBUG, "skip %d audio samples from curr_cts: %"PRId64"\n", packet_skip_samples, curr_cts);
3092 flags |= AVINDEX_DISCARD_FRAME;
3093 av_log(mov->fc, AV_LOG_DEBUG, "drop a frame at curr_cts: %"PRId64" @ %"PRId64"\n", curr_cts, index);
3095 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && edit_list_start_encountered == 0) {
3096 num_discarded_begin++;
3097 frame_duration_buffer = av_realloc(frame_duration_buffer,
3098 num_discarded_begin * sizeof(int64_t));
3099 if (!frame_duration_buffer) {
3100 av_log(mov->fc, AV_LOG_ERROR, "Cannot reallocate frame duration buffer\n");
3103 frame_duration_buffer[num_discarded_begin - 1] = frame_duration;
3105 // Increment skip_samples for the first non-zero audio edit list
3106 if (first_non_zero_audio_edit > 0 && st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
3107 st->skip_samples += frame_duration;
3108 msc->start_pad = st->skip_samples;
3112 } else if (edit_list_start_encountered == 0) {
3113 edit_list_start_encountered = 1;
3114 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3115 // discarded packets.
3116 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && frame_duration_buffer) {
3117 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3118 frame_duration_buffer, num_discarded_begin);
3119 av_freep(&frame_duration_buffer);
3123 if (add_index_entry(st, current->pos, edit_list_dts_counter, current->size,
3124 current->min_distance, flags) == -1) {
3125 av_log(mov->fc, AV_LOG_ERROR, "Cannot add index entry\n");
3129 // Only start incrementing DTS in frame_duration amounts, when we encounter a frame in edit list.
3130 if (edit_list_start_encountered > 0) {
3131 edit_list_dts_counter = edit_list_dts_counter + frame_duration;
3134 // Break when found first key frame after edit entry completion
3135 if (((curr_cts + frame_duration) >= (edit_list_duration + edit_list_media_time)) &&
3136 ((flags & AVINDEX_KEYFRAME) || ((st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)))) {
3138 if (ctts_data_old && ctts_sample_old != 0) {
3139 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3140 &ctts_allocated_size,
3141 ctts_sample_old - edit_list_start_ctts_sample,
3142 ctts_data_old[ctts_index_old].duration) == -1) {
3143 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3144 ctts_index_old, ctts_sample_old - edit_list_start_ctts_sample,
3145 ctts_data_old[ctts_index_old].duration);
3153 // Update av stream length
3154 st->duration = edit_list_dts_entry_end - start_dts;
3156 // Free the old index and the old CTTS structures
3158 av_free(ctts_data_old);
3161 static void mov_build_index(MOVContext *mov, AVStream *st)
3163 MOVStreamContext *sc = st->priv_data;
3164 int64_t current_offset;
3165 int64_t current_dts = 0;
3166 unsigned int stts_index = 0;
3167 unsigned int stsc_index = 0;
3168 unsigned int stss_index = 0;
3169 unsigned int stps_index = 0;
3171 uint64_t stream_size = 0;
3173 if (sc->elst_count) {
3174 int i, edit_start_index = 0;
3175 int64_t empty_duration = 0; // empty duration of the first edit list entry
3176 int64_t start_time = 0; // start time of the media
3178 for (i = 0; i < sc->elst_count; i++) {
3179 const MOVElst *e = &sc->elst_data[i];
3180 if (i == 0 && e->time == -1) {
3181 /* if empty, the first entry is the start time of the stream
3182 * relative to the presentation itself */
3183 empty_duration = e->duration;
3184 edit_start_index = 1;
3185 } else if (i == edit_start_index && e->time >= 0) {
3186 start_time = e->time;
3190 /* adjust first dts according to edit list */
3191 if ((empty_duration || start_time) && mov->time_scale > 0) {
3193 empty_duration = av_rescale(empty_duration, sc->time_scale, mov->time_scale);
3194 sc->time_offset = start_time - empty_duration;
3198 /* only use old uncompressed audio chunk demuxing when stts specifies it */
3199 if (!(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3200 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
3201 unsigned int current_sample = 0;
3202 unsigned int stts_sample = 0;
3203 unsigned int sample_size;
3204 unsigned int distance = 0;
3205 unsigned int rap_group_index = 0;
3206 unsigned int rap_group_sample = 0;
3207 int64_t last_dts = 0;
3208 int64_t dts_correction = 0;
3209 int rap_group_present = sc->rap_group_count && sc->rap_group;
3210 int key_off = (sc->keyframe_count && sc->keyframes[0] > 0) || (sc->stps_count && sc->stps_data[0] > 0);
3212 current_dts -= sc->dts_shift;
3213 last_dts = current_dts;
3215 if (!sc->sample_count || st->nb_index_entries)
3217 if (sc->sample_count >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3219 if (av_reallocp_array(&st->index_entries,
3220 st->nb_index_entries + sc->sample_count,
3221 sizeof(*st->index_entries)) < 0) {
3222 st->nb_index_entries = 0;
3225 st->index_entries_allocated_size = (st->nb_index_entries + sc->sample_count) * sizeof(*st->index_entries);
3227 for (i = 0; i < sc->chunk_count; i++) {
3228 int64_t next_offset = i+1 < sc->chunk_count ? sc->chunk_offsets[i+1] : INT64_MAX;
3229 current_offset = sc->chunk_offsets[i];
3230 while (stsc_index + 1 < sc->stsc_count &&
3231 i + 1 == sc->stsc_data[stsc_index + 1].first)
3234 if (next_offset > current_offset && sc->sample_size>0 && sc->sample_size < sc->stsz_sample_size &&
3235 sc->stsc_data[stsc_index].count * (int64_t)sc->stsz_sample_size > next_offset - current_offset) {
3236 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too large), ignoring\n", sc->stsz_sample_size);
3237 sc->stsz_sample_size = sc->sample_size;
3239 if (sc->stsz_sample_size>0 && sc->stsz_sample_size < sc->sample_size) {
3240 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too small), ignoring\n", sc->stsz_sample_size);
3241 sc->stsz_sample_size = sc->sample_size;
3244 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
3246 if (current_sample >= sc->sample_count) {
3247 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
3251 if (!sc->keyframe_absent && (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index])) {
3253 if (stss_index + 1 < sc->keyframe_count)
3255 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
3257 if (stps_index + 1 < sc->stps_count)
3260 if (rap_group_present && rap_group_index < sc->rap_group_count) {
3261 if (sc->rap_group[rap_group_index].index > 0)
3263 if (++rap_group_sample == sc->rap_group[rap_group_index].count) {
3264 rap_group_sample = 0;
3268 if (sc->keyframe_absent
3270 && !rap_group_present
3271 && (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO || (i==0 && j==0)))
3275 sample_size = sc->stsz_sample_size > 0 ? sc->stsz_sample_size : sc->sample_sizes[current_sample];
3276 if (sc->pseudo_stream_id == -1 ||
3277 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
3279 if (sample_size > 0x3FFFFFFF) {
3280 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", sample_size);
3283 e = &st->index_entries[st->nb_index_entries++];
3284 e->pos = current_offset;
3285 e->timestamp = current_dts;
3286 e->size = sample_size;
3287 e->min_distance = distance;
3288 e->flags = keyframe ? AVINDEX_KEYFRAME : 0;
3289 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
3290 "size %d, distance %d, keyframe %d\n", st->index, current_sample,
3291 current_offset, current_dts, sample_size, distance, keyframe);
3292 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries < 100)
3293 ff_rfps_add_frame(mov->fc, st, current_dts);
3296 current_offset += sample_size;
3297 stream_size += sample_size;
3299 /* A negative sample duration is invalid based on the spec,
3300 * but some samples need it to correct the DTS. */
3301 if (sc->stts_data[stts_index].duration < 0) {
3302 av_log(mov->fc, AV_LOG_WARNING,
3303 "Invalid SampleDelta %d in STTS, at %d st:%d\n",
3304 sc->stts_data[stts_index].duration, stts_index,
3306 dts_correction += sc->stts_data[stts_index].duration - 1;
3307 sc->stts_data[stts_index].duration = 1;
3309 current_dts += sc->stts_data[stts_index].duration;
3310 if (!dts_correction || current_dts + dts_correction > last_dts) {
3311 current_dts += dts_correction;
3314 /* Avoid creating non-monotonous DTS */
3315 dts_correction += current_dts - last_dts - 1;
3316 current_dts = last_dts + 1;
3318 last_dts = current_dts;
3322 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
3328 if (st->duration > 0)
3329 st->codecpar->bit_rate = stream_size*8*sc->time_scale/st->duration;
3331 unsigned chunk_samples, total = 0;
3333 // compute total chunk count
3334 for (i = 0; i < sc->stsc_count; i++) {
3335 unsigned count, chunk_count;
3337 chunk_samples = sc->stsc_data[i].count;
3338 if (i != sc->stsc_count - 1 &&
3339 sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
3340 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
3344 if (sc->samples_per_frame >= 160) { // gsm
3345 count = chunk_samples / sc->samples_per_frame;
3346 } else if (sc->samples_per_frame > 1) {
3347 unsigned samples = (1024/sc->samples_per_frame)*sc->samples_per_frame;
3348 count = (chunk_samples+samples-1) / samples;
3350 count = (chunk_samples+1023) / 1024;
3353 if (i < sc->stsc_count - 1)
3354 chunk_count = sc->stsc_data[i+1].first - sc->stsc_data[i].first;
3356 chunk_count = sc->chunk_count - (sc->stsc_data[i].first - 1);
3357 total += chunk_count * count;
3360 av_log(mov->fc, AV_LOG_TRACE, "chunk count %d\n", total);
3361 if (total >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3363 if (av_reallocp_array(&st->index_entries,
3364 st->nb_index_entries + total,
3365 sizeof(*st->index_entries)) < 0) {
3366 st->nb_index_entries = 0;
3369 st->index_entries_allocated_size = (st->nb_index_entries + total) * sizeof(*st->index_entries);
3372 for (i = 0; i < sc->chunk_count; i++) {
3373 current_offset = sc->chunk_offsets[i];
3374 if (stsc_index + 1 < sc->stsc_count &&
3375 i + 1 == sc->stsc_data[stsc_index + 1].first)
3377 chunk_samples = sc->stsc_data[stsc_index].count;
3379 while (chunk_samples > 0) {
3381 unsigned size, samples;
3383 if (sc->samples_per_frame > 1 && !sc->bytes_per_frame) {
3384 avpriv_request_sample(mov->fc,
3385 "Zero bytes per frame, but %d samples per frame",
3386 sc->samples_per_frame);
3390 if (sc->samples_per_frame >= 160) { // gsm
3391 samples = sc->samples_per_frame;
3392 size = sc->bytes_per_frame;
3394 if (sc->samples_per_frame > 1) {
3395 samples = FFMIN((1024 / sc->samples_per_frame)*
3396 sc->samples_per_frame, chunk_samples);
3397 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
3399 samples = FFMIN(1024, chunk_samples);
3400 size = samples * sc->sample_size;
3404 if (st->nb_index_entries >= total) {
3405 av_log(mov->fc, AV_LOG_ERROR, "wrong chunk count %d\n", total);
3408 if (size > 0x3FFFFFFF) {
3409 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", size);
3412 e = &st->index_entries[st->nb_index_entries++];
3413 e->pos = current_offset;
3414 e->timestamp = current_dts;
3416 e->min_distance = 0;
3417 e->flags = AVINDEX_KEYFRAME;
3418 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, chunk %d, offset %"PRIx64", dts %"PRId64", "
3419 "size %d, duration %d\n", st->index, i, current_offset, current_dts,
3422 current_offset += size;
3423 current_dts += samples;
3424 chunk_samples -= samples;
3429 // Fix index according to edit lists.
3430 mov_fix_index(mov, st);
3433 static int test_same_origin(const char *src, const char *ref) {
3443 av_url_split(src_proto, sizeof(src_proto), src_auth, sizeof(src_auth), src_host, sizeof(src_host), &src_port, NULL, 0, src);
3444 av_url_split(ref_proto, sizeof(ref_proto), ref_auth, sizeof(ref_auth), ref_host, sizeof(ref_host), &ref_port, NULL, 0, ref);
3446 if (strlen(src) == 0) {
3448 } else if (strlen(src_auth) + 1 >= sizeof(src_auth) ||
3449 strlen(ref_auth) + 1 >= sizeof(ref_auth) ||
3450 strlen(src_host) + 1 >= sizeof(src_host) ||
3451 strlen(ref_host) + 1 >= sizeof(ref_host)) {
3453 } else if (strcmp(src_proto, ref_proto) ||
3454 strcmp(src_auth, ref_auth) ||
3455 strcmp(src_host, ref_host) ||
3456 src_port != ref_port) {
3462 static int mov_open_dref(MOVContext *c, AVIOContext **pb, const char *src, MOVDref *ref)
3464 /* try relative path, we do not try the absolute because it can leak information about our
3465 system to an attacker */
3466 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
3467 char filename[1025];
3468 const char *src_path;
3471 /* find a source dir */
3472 src_path = strrchr(src, '/');
3478 /* find a next level down to target */
3479 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
3480 if (ref->path[l] == '/') {
3481 if (i == ref->nlvl_to - 1)
3487 /* compose filename if next level down to target was found */
3488 if (i == ref->nlvl_to - 1 && src_path - src < sizeof(filename)) {
3489 memcpy(filename, src, src_path - src);
3490 filename[src_path - src] = 0;
3492 for (i = 1; i < ref->nlvl_from; i++)
3493 av_strlcat(filename, "../", sizeof(filename));
3495 av_strlcat(filename, ref->path + l + 1, sizeof(filename));
3496 if (!c->use_absolute_path) {
3497 int same_origin = test_same_origin(src, filename);
3500 av_log(c->fc, AV_LOG_ERROR,
3501 "Reference with mismatching origin, %s not tried for security reasons, "
3502 "set demuxer option use_absolute_path to allow it anyway\n",
3504 return AVERROR(ENOENT);
3507 if(strstr(ref->path + l + 1, "..") ||
3508 strstr(ref->path + l + 1, ":") ||
3509 (ref->nlvl_from > 1 && same_origin < 0) ||
3510 (filename[0] == '/' && src_path == src))
3511 return AVERROR(ENOENT);
3514 if (strlen(filename) + 1 == sizeof(filename))
3515 return AVERROR(ENOENT);
3516 if (!c->fc->io_open(c->fc, pb, filename, AVIO_FLAG_READ, NULL))
3519 } else if (c->use_absolute_path) {
3520 av_log(c->fc, AV_LOG_WARNING, "Using absolute path on user request, "
3521 "this is a possible security issue\n");
3522 if (!c->fc->io_open(c->fc, pb, ref->path, AVIO_FLAG_READ, NULL))
3525 av_log(c->fc, AV_LOG_ERROR,
3526 "Absolute path %s not tried for security reasons, "
3527 "set demuxer option use_absolute_path to allow absolute paths\n",
3531 return AVERROR(ENOENT);
3534 static void fix_timescale(MOVContext *c, MOVStreamContext *sc)
3536 if (sc->time_scale <= 0) {
3537 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", sc->ffindex);
3538 sc->time_scale = c->time_scale;
3539 if (sc->time_scale <= 0)
3544 static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3547 MOVStreamContext *sc;
3550 st = avformat_new_stream(c->fc, NULL);
3551 if (!st) return AVERROR(ENOMEM);
3552 st->id = c->fc->nb_streams;
3553 sc = av_mallocz(sizeof(MOVStreamContext));
3554 if (!sc) return AVERROR(ENOMEM);
3557 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
3558 sc->ffindex = st->index;
3559 c->trak_index = st->index;
3561 if ((ret = mov_read_default(c, pb, atom)) < 0)
3567 if (sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
3568 (!sc->sample_size && !sc->sample_count))) {
3569 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
3574 fix_timescale(c, sc);
3576 avpriv_set_pts_info(st, 64, 1, sc->time_scale);
3578 mov_build_index(c, st);
3580 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
3581 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
3582 if (c->enable_drefs) {
3583 if (mov_open_dref(c, &sc->pb, c->fc->filename, dref) < 0)
3584 av_log(c->fc, AV_LOG_ERROR,
3585 "stream %d, error opening alias: path='%s', dir='%s', "
3586 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
3587 st->index, dref->path, dref->dir, dref->filename,
3588 dref->volume, dref->nlvl_from, dref->nlvl_to);
3590 av_log(c->fc, AV_LOG_WARNING,
3591 "Skipped opening external track: "
3592 "stream %d, alias: path='%s', dir='%s', "
3593 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d."
3594 "Set enable_drefs to allow this.\n",
3595 st->index, dref->path, dref->dir, dref->filename,
3596 dref->volume, dref->nlvl_from, dref->nlvl_to);
3600 sc->pb_is_copied = 1;
3603 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
3604 if (!st->sample_aspect_ratio.num && st->codecpar->width && st->codecpar->height &&
3605 sc->height && sc->width &&
3606 (st->codecpar->width != sc->width || st->codecpar->height != sc->height)) {
3607 st->sample_aspect_ratio = av_d2q(((double)st->codecpar->height * sc->width) /
3608 ((double)st->codecpar->width * sc->height), INT_MAX);
3611 #if FF_API_R_FRAME_RATE
3612 if (sc->stts_count == 1 || (sc->stts_count == 2 && sc->stts_data[1].count == 1))
3613 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
3614 sc->time_scale, sc->stts_data[0].duration, INT_MAX);
3618 // done for ai5q, ai52, ai55, ai1q, ai12 and ai15.
3619 if (!st->codecpar->extradata_size && st->codecpar->codec_id == AV_CODEC_ID_H264 &&
3620 TAG_IS_AVCI(st->codecpar->codec_tag)) {
3621 ret = ff_generate_avci_extradata(st);
3626 switch (st->codecpar->codec_id) {
3627 #if CONFIG_H261_DECODER
3628 case AV_CODEC_ID_H261:
3630 #if CONFIG_H263_DECODER
3631 case AV_CODEC_ID_H263:
3633 #if CONFIG_MPEG4_DECODER
3634 case AV_CODEC_ID_MPEG4:
3636 st->codecpar->width = 0; /* let decoder init width/height */
3637 st->codecpar->height= 0;
3641 // If the duration of the mp3 packets is not constant, then they could need a parser
3642 if (st->codecpar->codec_id == AV_CODEC_ID_MP3
3643 && sc->stts_count > 3
3644 && sc->stts_count*10 > st->nb_frames
3645 && sc->time_scale == st->codecpar->sample_rate) {
3646 st->need_parsing = AVSTREAM_PARSE_FULL;
3648 /* Do not need those anymore. */
3649 av_freep(&sc->chunk_offsets);
3650 av_freep(&sc->sample_sizes);
3651 av_freep(&sc->keyframes);
3652 av_freep(&sc->stts_data);
3653 av_freep(&sc->stps_data);
3654 av_freep(&sc->elst_data);
3655 av_freep(&sc->rap_group);
3660 static int mov_read_ilst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3663 c->itunes_metadata = 1;
3664 ret = mov_read_default(c, pb, atom);
3665 c->itunes_metadata = 0;
3669 static int mov_read_keys(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3678 count = avio_rb32(pb);
3679 if (count > UINT_MAX / sizeof(*c->meta_keys) - 1) {
3680 av_log(c->fc, AV_LOG_ERROR,
3681 "The 'keys' atom with the invalid key count: %d\n", count);
3682 return AVERROR_INVALIDDATA;
3685 c->meta_keys_count = count + 1;
3686 c->meta_keys = av_mallocz(c->meta_keys_count * sizeof(*c->meta_keys));
3688 return AVERROR(ENOMEM);
3690 for (i = 1; i <= count; ++i) {
3691 uint32_t key_size = avio_rb32(pb);
3692 uint32_t type = avio_rl32(pb);
3694 av_log(c->fc, AV_LOG_ERROR,
3695 "The key# %d in meta has invalid size: %d\n", i, key_size);
3696 return AVERROR_INVALIDDATA;
3699 if (type != MKTAG('m','d','t','a')) {
3700 avio_skip(pb, key_size);
3702 c->meta_keys[i] = av_mallocz(key_size + 1);
3703 if (!c->meta_keys[i])
3704 return AVERROR(ENOMEM);
3705 avio_read(pb, c->meta_keys[i], key_size);
3711 static int mov_read_custom(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3713 int64_t end = avio_tell(pb) + atom.size;
3714 uint8_t *key = NULL, *val = NULL, *mean = NULL;
3718 MOVStreamContext *sc;
3720 if (c->fc->nb_streams < 1)
3722 st = c->fc->streams[c->fc->nb_streams-1];
3725 for (i = 0; i < 3; i++) {
3729 if (end - avio_tell(pb) <= 12)
3732 len = avio_rb32(pb);
3733 tag = avio_rl32(pb);
3734 avio_skip(pb, 4); // flags
3736 if (len < 12 || len - 12 > end - avio_tell(pb))
3740 if (tag == MKTAG('m', 'e', 'a', 'n'))
3742 else if (tag == MKTAG('n', 'a', 'm', 'e'))
3744 else if (tag == MKTAG('d', 'a', 't', 'a') && len > 4) {
3751 *p = av_malloc(len + 1);
3754 ret = ffio_read_size(pb, *p, len);
3762 if (mean && key && val) {
3763 if (strcmp(key, "iTunSMPB") == 0) {
3764 int priming, remainder, samples;
3765 if(sscanf(val, "%*X %X %X %X", &priming, &remainder, &samples) == 3){
3766 if(priming>0 && priming<16384)
3767 sc->start_pad = priming;
3770 if (strcmp(key, "cdec") != 0) {
3771 av_dict_set(&c->fc->metadata, key, val,
3772 AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL);
3776 av_log(c->fc, AV_LOG_VERBOSE,
3777 "Unhandled or malformed custom metadata of size %"PRId64"\n", atom.size);
3780 avio_seek(pb, end, SEEK_SET);
3787 static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3789 while (atom.size > 8) {
3790 uint32_t tag = avio_rl32(pb);
3792 if (tag == MKTAG('h','d','l','r')) {
3793 avio_seek(pb, -8, SEEK_CUR);
3795 return mov_read_default(c, pb, atom);
3801 static int mov_read_tkhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3806 int display_matrix[3][3];
3808 MOVStreamContext *sc;
3812 if (c->fc->nb_streams < 1)
3814 st = c->fc->streams[c->fc->nb_streams-1];
3817 version = avio_r8(pb);
3818 flags = avio_rb24(pb);
3819 st->disposition |= (flags & MOV_TKHD_FLAG_ENABLED) ? AV_DISPOSITION_DEFAULT : 0;
3825 avio_rb32(pb); /* creation time */
3826 avio_rb32(pb); /* modification time */
3828 st->id = (int)avio_rb32(pb); /* track id (NOT 0 !)*/
3829 avio_rb32(pb); /* reserved */
3831 /* highlevel (considering edits) duration in movie timebase */
3832 (version == 1) ? avio_rb64(pb) : avio_rb32(pb);
3833 avio_rb32(pb); /* reserved */
3834 avio_rb32(pb); /* reserved */
3836 avio_rb16(pb); /* layer */
3837 avio_rb16(pb); /* alternate group */
3838 avio_rb16(pb); /* volume */
3839 avio_rb16(pb); /* reserved */
3841 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
3842 // they're kept in fixed point format through all calculations
3843 // save u,v,z to store the whole matrix in the AV_PKT_DATA_DISPLAYMATRIX
3844 // side data, but the scale factor is not needed to calculate aspect ratio
3845 for (i = 0; i < 3; i++) {
3846 display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
3847 display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
3848 display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
3851 width = avio_rb32(pb); // 16.16 fixed point track width
3852 height = avio_rb32(pb); // 16.16 fixed point track height
3853 sc->width = width >> 16;
3854 sc->height = height >> 16;
3856 // save the matrix and add rotate metadata when it is not the default
3858 if (display_matrix[0][0] != (1 << 16) ||
3859 display_matrix[1][1] != (1 << 16) ||
3860 display_matrix[2][2] != (1 << 30) ||
3861 display_matrix[0][1] || display_matrix[0][2] ||
3862 display_matrix[1][0] || display_matrix[1][2] ||
3863 display_matrix[2][0] || display_matrix[2][1]) {
3867 av_freep(&sc->display_matrix);
3868 sc->display_matrix = av_malloc(sizeof(int32_t) * 9);
3869 if (!sc->display_matrix)
3870 return AVERROR(ENOMEM);
3872 for (i = 0; i < 3; i++)
3873 for (j = 0; j < 3; j++)
3874 sc->display_matrix[i * 3 + j] = display_matrix[i][j];
3876 rotate = av_display_rotation_get(sc->display_matrix);
3877 if (!isnan(rotate)) {
3878 char rotate_buf[64];
3880 if (rotate < 0) // for backward compatibility
3882 snprintf(rotate_buf, sizeof(rotate_buf), "%g", rotate);
3883 av_dict_set(&st->metadata, "rotate", rotate_buf, 0);
3887 // transform the display width/height according to the matrix
3888 // to keep the same scale, use [width height 1<<16]
3889 if (width && height && sc->display_matrix) {
3890 double disp_transform[2];
3892 for (i = 0; i < 2; i++)
3893 disp_transform[i] = hypot(display_matrix[i][0], display_matrix[i][1]);
3895 if (disp_transform[0] > 0 && disp_transform[1] > 0 &&
3896 disp_transform[0] < (1<<24) && disp_transform[1] < (1<<24) &&
3897 fabs((disp_transform[0] / disp_transform[1]) - 1.0) > 0.01)
3898 st->sample_aspect_ratio = av_d2q(
3899 disp_transform[0] / disp_transform[1],
3905 static int mov_read_tfhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3907 MOVFragment *frag = &c->fragment;
3908 MOVTrackExt *trex = NULL;
3909 MOVFragmentIndex* index = NULL;
3910 int flags, track_id, i, found = 0;
3912 avio_r8(pb); /* version */
3913 flags = avio_rb24(pb);
3915 track_id = avio_rb32(pb);
3917 return AVERROR_INVALIDDATA;
3918 frag->track_id = track_id;
3919 for (i = 0; i < c->trex_count; i++)
3920 if (c->trex_data[i].track_id == frag->track_id) {
3921 trex = &c->trex_data[i];
3925 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
3926 return AVERROR_INVALIDDATA;
3929 frag->base_data_offset = flags & MOV_TFHD_BASE_DATA_OFFSET ?
3930 avio_rb64(pb) : flags & MOV_TFHD_DEFAULT_BASE_IS_MOOF ?
3931 frag->moof_offset : frag->implicit_offset;
3932 frag->stsd_id = flags & MOV_TFHD_STSD_ID ? avio_rb32(pb) : trex->stsd_id;
3934 frag->duration = flags & MOV_TFHD_DEFAULT_DURATION ?
3935 avio_rb32(pb) : trex->duration;
3936 frag->size = flags & MOV_TFHD_DEFAULT_SIZE ?
3937 avio_rb32(pb) : trex->size;
3938 frag->flags = flags & MOV_TFHD_DEFAULT_FLAGS ?
3939 avio_rb32(pb) : trex->flags;
3940 frag->time = AV_NOPTS_VALUE;
3941 for (i = 0; i < c->fragment_index_count; i++) {
3943 MOVFragmentIndex* candidate = c->fragment_index_data[i];
3944 if (candidate->track_id == frag->track_id) {
3945 av_log(c->fc, AV_LOG_DEBUG,
3946 "found fragment index for track %u\n", frag->track_id);
3948 for (j = index->current_item; j < index->item_count; j++) {
3949 if (frag->implicit_offset == index->items[j].moof_offset) {
3950 av_log(c->fc, AV_LOG_DEBUG, "found fragment index entry "
3951 "for track %u and moof_offset %"PRId64"\n",
3952 frag->track_id, index->items[j].moof_offset);
3953 frag->time = index->items[j].time;
3954 index->current_item = j + 1;
3963 if (index && !found) {
3964 av_log(c->fc, AV_LOG_DEBUG, "track %u has a fragment index but "
3965 "it doesn't have an (in-order) entry for moof_offset "
3966 "%"PRId64"\n", frag->track_id, frag->implicit_offset);
3968 av_log(c->fc, AV_LOG_TRACE, "frag flags 0x%x\n", frag->flags);
3972 static int mov_read_chap(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3974 c->chapter_track = avio_rb32(pb);
3978 static int mov_read_trex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3983 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
3984 return AVERROR_INVALIDDATA;
3985 if ((err = av_reallocp_array(&c->trex_data, c->trex_count + 1,
3986 sizeof(*c->trex_data))) < 0) {
3991 c->fc->duration = AV_NOPTS_VALUE; // the duration from mvhd is not representing the whole file when fragments are used.
3993 trex = &c->trex_data[c->trex_count++];
3994 avio_r8(pb); /* version */
3995 avio_rb24(pb); /* flags */
3996 trex->track_id = avio_rb32(pb);
3997 trex->stsd_id = avio_rb32(pb);
3998 trex->duration = avio_rb32(pb);
3999 trex->size = avio_rb32(pb);
4000 trex->flags = avio_rb32(pb);
4004 static int mov_read_tfdt(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4006 MOVFragment *frag = &c->fragment;
4007 AVStream *st = NULL;
4008 MOVStreamContext *sc;
4011 for (i = 0; i < c->fc->nb_streams; i++) {
4012 if (c->fc->streams[i]->id == frag->track_id) {
4013 st = c->fc->streams[i];
4018 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
4019 return AVERROR_INVALIDDATA;
4022 if (sc->pseudo_stream_id + 1 != frag->stsd_id)
4024 version = avio_r8(pb);
4025 avio_rb24(pb); /* flags */
4027 sc->track_end = avio_rb64(pb);
4029 sc->track_end = avio_rb32(pb);
4034 static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4036 MOVFragment *frag = &c->fragment;
4037 AVStream *st = NULL;
4038 MOVStreamContext *sc;
4042 int data_offset = 0;
4043 unsigned entries, first_sample_flags = frag->flags;
4044 int flags, distance, i, err;
4046 for (i = 0; i < c->fc->nb_streams; i++) {
4047 if (c->fc->streams[i]->id == frag->track_id) {
4048 st = c->fc->streams[i];
4053 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
4054 return AVERROR_INVALIDDATA;
4057 if (sc->pseudo_stream_id+1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4059 avio_r8(pb); /* version */
4060 flags = avio_rb24(pb);
4061 entries = avio_rb32(pb);
4062 av_log(c->fc, AV_LOG_TRACE, "flags 0x%x entries %d\n", flags, entries);
4064 /* Always assume the presence of composition time offsets.
4065 * Without this assumption, for instance, we cannot deal with a track in fragmented movies that meet the following.
4066 * 1) in the initial movie, there are no samples.
4067 * 2) in the first movie fragment, there is only one sample without composition time offset.
4068 * 3) in the subsequent movie fragments, there are samples with composition time offset. */
4069 if (!sc->ctts_count && sc->sample_count)
4071 /* Complement ctts table if moov atom doesn't have ctts atom. */
4072 ctts_data = av_realloc(NULL, sizeof(*sc->ctts_data));
4074 return AVERROR(ENOMEM);
4075 sc->ctts_data = ctts_data;
4076 sc->ctts_data[sc->ctts_count].count = sc->sample_count;
4077 sc->ctts_data[sc->ctts_count].duration = 0;
4080 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
4081 return AVERROR_INVALIDDATA;
4082 if ((err = av_reallocp_array(&sc->ctts_data, entries + sc->ctts_count,
4083 sizeof(*sc->ctts_data))) < 0) {
4087 if (flags & MOV_TRUN_DATA_OFFSET) data_offset = avio_rb32(pb);
4088 if (flags & MOV_TRUN_FIRST_SAMPLE_FLAGS) first_sample_flags = avio_rb32(pb);
4089 dts = sc->track_end - sc->time_offset;
4090 offset = frag->base_data_offset + data_offset;
4092 av_log(c->fc, AV_LOG_TRACE, "first sample flags 0x%x\n", first_sample_flags);
4093 for (i = 0; i < entries && !pb->eof_reached; i++) {
4094 unsigned sample_size = frag->size;
4095 int sample_flags = i ? frag->flags : first_sample_flags;
4096 unsigned sample_duration = frag->duration;
4099 if (flags & MOV_TRUN_SAMPLE_DURATION) sample_duration = avio_rb32(pb);
4100 if (flags & MOV_TRUN_SAMPLE_SIZE) sample_size = avio_rb32(pb);
4101 if (flags & MOV_TRUN_SAMPLE_FLAGS) sample_flags = avio_rb32(pb);
4102 sc->ctts_data[sc->ctts_count].count = 1;
4103 sc->ctts_data[sc->ctts_count].duration = (flags & MOV_TRUN_SAMPLE_CTS) ?
4105 mov_update_dts_shift(sc, sc->ctts_data[sc->ctts_count].duration);
4106 if (frag->time != AV_NOPTS_VALUE) {
4107 if (c->use_mfra_for == FF_MOV_FLAG_MFRA_PTS) {
4108 int64_t pts = frag->time;
4109 av_log(c->fc, AV_LOG_DEBUG, "found frag time %"PRId64
4110 " sc->dts_shift %d ctts.duration %d"
4111 " sc->time_offset %"PRId64" flags & MOV_TRUN_SAMPLE_CTS %d\n", pts,
4112 sc->dts_shift, sc->ctts_data[sc->ctts_count].duration,
4113 sc->time_offset, flags & MOV_TRUN_SAMPLE_CTS);
4114 dts = pts - sc->dts_shift;
4115 if (flags & MOV_TRUN_SAMPLE_CTS) {
4116 dts -= sc->ctts_data[sc->ctts_count].duration;
4118 dts -= sc->time_offset;
4120 av_log(c->fc, AV_LOG_DEBUG, "calculated into dts %"PRId64"\n", dts);
4122 dts = frag->time - sc->time_offset;
4123 av_log(c->fc, AV_LOG_DEBUG, "found frag time %"PRId64
4124 ", using it for dts\n", dts);
4126 frag->time = AV_NOPTS_VALUE;
4129 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
4133 !(sample_flags & (MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC |
4134 MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES));
4137 err = av_add_index_entry(st, offset, dts, sample_size, distance,
4138 keyframe ? AVINDEX_KEYFRAME : 0);
4140 av_log(c->fc, AV_LOG_ERROR, "Failed to add index entry\n");
4142 av_log(c->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
4143 "size %d, distance %d, keyframe %d\n", st->index, sc->sample_count+i,
4144 offset, dts, sample_size, distance, keyframe);
4146 dts += sample_duration;
4147 offset += sample_size;
4148 sc->data_size += sample_size;
4149 sc->duration_for_fps += sample_duration;
4150 sc->nb_frames_for_fps ++;
4153 if (pb->eof_reached)
4156 frag->implicit_offset = offset;
4158 sc->track_end = dts + sc->time_offset;
4159 if (st->duration < sc->track_end)
4160 st->duration = sc->track_end;
4165 static int mov_read_sidx(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4167 int64_t offset = avio_tell(pb) + atom.size, pts;
4169 unsigned i, track_id;
4170 AVStream *st = NULL;
4171 MOVStreamContext *sc;
4172 MOVFragmentIndex *index = NULL;
4173 MOVFragmentIndex **tmp;
4174 AVRational timescale;
4176 version = avio_r8(pb);
4178 avpriv_request_sample(c->fc, "sidx version %u", version);
4182 avio_rb24(pb); // flags
4184 track_id = avio_rb32(pb); // Reference ID
4185 for (i = 0; i < c->fc->nb_streams; i++) {
4186 if (c->fc->streams[i]->id == track_id) {
4187 st = c->fc->streams[i];
4192 av_log(c->fc, AV_LOG_WARNING, "could not find corresponding track id %d\n", track_id);
4198 timescale = av_make_q(1, avio_rb32(pb));
4201 pts = avio_rb32(pb);
4202 offset += avio_rb32(pb);
4204 pts = avio_rb64(pb);
4205 offset += avio_rb64(pb);
4208 avio_rb16(pb); // reserved
4210 index = av_mallocz(sizeof(MOVFragmentIndex));
4212 return AVERROR(ENOMEM);
4214 index->track_id = track_id;
4216 index->item_count = avio_rb16(pb);
4217 index->items = av_mallocz_array(index->item_count, sizeof(MOVFragmentIndexItem));
4219 if (!index->items) {
4221 return AVERROR(ENOMEM);
4224 for (i = 0; i < index->item_count; i++) {
4225 uint32_t size = avio_rb32(pb);
4226 uint32_t duration = avio_rb32(pb);
4227 if (size & 0x80000000) {
4228 avpriv_request_sample(c->fc, "sidx reference_type 1");
4229 av_freep(&index->items);
4231 return AVERROR_PATCHWELCOME;
4233 avio_rb32(pb); // sap_flags
4234 index->items[i].moof_offset = offset;
4235 index->items[i].time = av_rescale_q(pts, st->time_base, timescale);
4240 st->duration = sc->track_end = pts;
4242 tmp = av_realloc_array(c->fragment_index_data,
4243 c->fragment_index_count + 1,
4244 sizeof(MOVFragmentIndex*));
4246 av_freep(&index->items);
4248 return AVERROR(ENOMEM);
4251 c->fragment_index_data = tmp;
4252 c->fragment_index_data[c->fragment_index_count++] = index;
4254 if (offset == avio_size(pb))
4255 c->fragment_index_complete = 1;
4260 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
4261 /* like the files created with Adobe Premiere 5.0, for samples see */
4262 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
4263 static int mov_read_wide(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4268 return 0; /* continue */
4269 if (avio_rb32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
4270 avio_skip(pb, atom.size - 4);
4273 atom.type = avio_rl32(pb);
4275 if (atom.type != MKTAG('m','d','a','t')) {
4276 avio_skip(pb, atom.size);
4279 err = mov_read_mdat(c, pb, atom);
4283 static int mov_read_cmov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4288 uint8_t *moov_data; /* uncompressed data */
4289 long cmov_len, moov_len;
4292 avio_rb32(pb); /* dcom atom */
4293 if (avio_rl32(pb) != MKTAG('d','c','o','m'))
4294 return AVERROR_INVALIDDATA;
4295 if (avio_rl32(pb) != MKTAG('z','l','i','b')) {
4296 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !\n");
4297 return AVERROR_INVALIDDATA;
4299 avio_rb32(pb); /* cmvd atom */
4300 if (avio_rl32(pb) != MKTAG('c','m','v','d'))
4301 return AVERROR_INVALIDDATA;
4302 moov_len = avio_rb32(pb); /* uncompressed size */
4303 cmov_len = atom.size - 6 * 4;
4305 cmov_data = av_malloc(cmov_len);
4307 return AVERROR(ENOMEM);
4308 moov_data = av_malloc(moov_len);
4311 return AVERROR(ENOMEM);
4313 ret = ffio_read_size(pb, cmov_data, cmov_len);
4315 goto free_and_return;
4317 if (uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
4318 goto free_and_return;
4319 if (ffio_init_context(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
4320 goto free_and_return;
4321 ctx.seekable = AVIO_SEEKABLE_NORMAL;
4322 atom.type = MKTAG('m','o','o','v');
4323 atom.size = moov_len;
4324 ret = mov_read_default(c, &ctx, atom);
4330 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
4331 return AVERROR(ENOSYS);
4335 /* edit list atom */
4336 static int mov_read_elst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4338 MOVStreamContext *sc;
4339 int i, edit_count, version;
4341 if (c->fc->nb_streams < 1 || c->ignore_editlist)
4343 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
4345 version = avio_r8(pb); /* version */
4346 avio_rb24(pb); /* flags */
4347 edit_count = avio_rb32(pb); /* entries */
4352 av_log(c->fc, AV_LOG_WARNING, "Duplicated ELST atom\n");
4353 av_free(sc->elst_data);
4355 sc->elst_data = av_malloc_array(edit_count, sizeof(*sc->elst_data));
4357 return AVERROR(ENOMEM);
4359 av_log(c->fc, AV_LOG_TRACE, "track[%i].edit_count = %i\n", c->fc->nb_streams-1, edit_count);
4360 for (i = 0; i < edit_count && !pb->eof_reached; i++) {
4361 MOVElst *e = &sc->elst_data[i];
4364 e->duration = avio_rb64(pb);
4365 e->time = avio_rb64(pb);
4367 e->duration = avio_rb32(pb); /* segment duration */
4368 e->time = (int32_t)avio_rb32(pb); /* media time */
4370 e->rate = avio_rb32(pb) / 65536.0;
4371 av_log(c->fc, AV_LOG_TRACE, "duration=%"PRId64" time=%"PRId64" rate=%f\n",
4372 e->duration, e->time, e->rate);
4379 static int mov_read_tmcd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4381 MOVStreamContext *sc;
4383 if (c->fc->nb_streams < 1)
4384 return AVERROR_INVALIDDATA;
4385 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
4386 sc->timecode_track = avio_rb32(pb);
4390 static int mov_read_uuid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4394 static const uint8_t uuid_isml_manifest[] = {
4395 0xa5, 0xd4, 0x0b, 0x30, 0xe8, 0x14, 0x11, 0xdd,
4396 0xba, 0x2f, 0x08, 0x00, 0x20, 0x0c, 0x9a, 0x66
4398 static const uint8_t uuid_xmp[] = {
4399 0xbe, 0x7a, 0xcf, 0xcb, 0x97, 0xa9, 0x42, 0xe8,
4400 0x9c, 0x71, 0x99, 0x94, 0x91, 0xe3, 0xaf, 0xac
4403 if (atom.size < sizeof(uuid) || atom.size == INT64_MAX)
4404 return AVERROR_INVALIDDATA;
4406 ret = avio_read(pb, uuid, sizeof(uuid));
4409 } else if (ret != sizeof(uuid)) {
4410 return AVERROR_INVALIDDATA;
4412 if (!memcmp(uuid, uuid_isml_manifest, sizeof(uuid))) {
4413 uint8_t *buffer, *ptr;
4415 size_t len = atom.size - sizeof(uuid);
4418 return AVERROR_INVALIDDATA;
4420 ret = avio_skip(pb, 4); // zeroes
4423 buffer = av_mallocz(len + 1);
4425 return AVERROR(ENOMEM);
4427 ret = avio_read(pb, buffer, len);
4431 } else if (ret != len) {
4433 return AVERROR_INVALIDDATA;
4437 while ((ptr = av_stristr(ptr, "systemBitrate=\""))) {
4438 ptr += sizeof("systemBitrate=\"") - 1;
4439 c->bitrates_count++;
4440 c->bitrates = av_realloc_f(c->bitrates, c->bitrates_count, sizeof(*c->bitrates));
4442 c->bitrates_count = 0;
4444 return AVERROR(ENOMEM);
4447 ret = strtol(ptr, &endptr, 10);
4448 if (ret < 0 || errno || *endptr != '"') {
4449 c->bitrates[c->bitrates_count - 1] = 0;
4451 c->bitrates[c->bitrates_count - 1] = ret;
4456 } else if (!memcmp(uuid, uuid_xmp, sizeof(uuid))) {
4458 size_t len = atom.size - sizeof(uuid);
4460 buffer = av_mallocz(len + 1);
4462 return AVERROR(ENOMEM);
4464 ret = avio_read(pb, buffer, len);
4468 } else if (ret != len) {
4470 return AVERROR_INVALIDDATA;
4472 if (c->export_xmp) {
4474 av_dict_set(&c->fc->metadata, "xmp", buffer, 0);
4481 static int mov_read_free(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4484 uint8_t content[16];
4489 ret = avio_read(pb, content, FFMIN(sizeof(content), atom.size));
4495 && !memcmp(content, "Anevia\x1A\x1A", 8)
4496 && c->use_mfra_for == FF_MOV_FLAG_MFRA_AUTO) {
4497 c->use_mfra_for = FF_MOV_FLAG_MFRA_PTS;
4503 static int mov_read_frma(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4505 uint32_t format = avio_rl32(pb);
4506 MOVStreamContext *sc;
4510 if (c->fc->nb_streams < 1)
4512 st = c->fc->streams[c->fc->nb_streams - 1];
4517 case MKTAG('e','n','c','v'): // encrypted video
4518 case MKTAG('e','n','c','a'): // encrypted audio
4519 id = mov_codec_id(st, format);
4520 if (st->codecpar->codec_id != AV_CODEC_ID_NONE &&
4521 st->codecpar->codec_id != id) {
4522 av_log(c->fc, AV_LOG_WARNING,
4523 "ignoring 'frma' atom of '%.4s', stream has codec id %d\n",
4524 (char*)&format, st->codecpar->codec_id);
4528 st->codecpar->codec_id = id;
4529 sc->format = format;
4533 if (format != sc->format) {
4534 av_log(c->fc, AV_LOG_WARNING,
4535 "ignoring 'frma' atom of '%.4s', stream format is '%.4s'\n",
4536 (char*)&format, (char*)&sc->format);
4544 static int mov_read_senc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4547 MOVStreamContext *sc;
4548 size_t auxiliary_info_size;
4550 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
4553 st = c->fc->streams[c->fc->nb_streams - 1];
4556 if (sc->cenc.aes_ctr) {
4557 av_log(c->fc, AV_LOG_ERROR, "duplicate senc atom\n");
4558 return AVERROR_INVALIDDATA;
4561 avio_r8(pb); /* version */
4562 sc->cenc.use_subsamples = avio_rb24(pb) & 0x02; /* flags */
4564 avio_rb32(pb); /* entries */
4566 if (atom.size < 8) {
4567 av_log(c->fc, AV_LOG_ERROR, "senc atom size %"PRId64" too small\n", atom.size);
4568 return AVERROR_INVALIDDATA;
4571 /* save the auxiliary info as is */
4572 auxiliary_info_size = atom.size - 8;
4574 sc->cenc.auxiliary_info = av_malloc(auxiliary_info_size);
4575 if (!sc->cenc.auxiliary_info) {
4576 return AVERROR(ENOMEM);
4579 sc->cenc.auxiliary_info_end = sc->cenc.auxiliary_info + auxiliary_info_size;
4581 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info;
4583 if (avio_read(pb, sc->cenc.auxiliary_info, auxiliary_info_size) != auxiliary_info_size) {
4584 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info");
4585 return AVERROR_INVALIDDATA;
4588 /* initialize the cipher */
4589 sc->cenc.aes_ctr = av_aes_ctr_alloc();
4590 if (!sc->cenc.aes_ctr) {
4591 return AVERROR(ENOMEM);
4594 return av_aes_ctr_init(sc->cenc.aes_ctr, c->decryption_key);
4597 static int mov_read_saiz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4600 MOVStreamContext *sc;
4602 int atom_header_size;
4605 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
4608 st = c->fc->streams[c->fc->nb_streams - 1];
4611 if (sc->cenc.auxiliary_info_sizes || sc->cenc.auxiliary_info_default_size) {
4612 av_log(c->fc, AV_LOG_ERROR, "duplicate saiz atom\n");
4613 return AVERROR_INVALIDDATA;
4616 atom_header_size = 9;
4618 avio_r8(pb); /* version */
4619 flags = avio_rb24(pb);
4621 if ((flags & 0x01) != 0) {
4622 atom_header_size += 8;
4624 avio_rb32(pb); /* info type */
4625 avio_rb32(pb); /* info type param */
4628 sc->cenc.auxiliary_info_default_size = avio_r8(pb);
4629 avio_rb32(pb); /* entries */
4631 if (atom.size <= atom_header_size) {
4635 /* save the auxiliary info sizes as is */
4636 data_size = atom.size - atom_header_size;
4638 sc->cenc.auxiliary_info_sizes = av_malloc(data_size);
4639 if (!sc->cenc.auxiliary_info_sizes) {
4640 return AVERROR(ENOMEM);
4643 sc->cenc.auxiliary_info_sizes_count = data_size;
4645 if (avio_read(pb, sc->cenc.auxiliary_info_sizes, data_size) != data_size) {
4646 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info sizes");
4647 return AVERROR_INVALIDDATA;
4653 static int cenc_filter(MOVContext *c, MOVStreamContext *sc, uint8_t *input, int size)
4655 uint32_t encrypted_bytes;
4656 uint16_t subsample_count;
4657 uint16_t clear_bytes;
4658 uint8_t* input_end = input + size;
4661 if (AES_CTR_IV_SIZE > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
4662 av_log(c->fc, AV_LOG_ERROR, "failed to read iv from the auxiliary info\n");
4663 return AVERROR_INVALIDDATA;
4666 av_aes_ctr_set_iv(sc->cenc.aes_ctr, sc->cenc.auxiliary_info_pos);
4667 sc->cenc.auxiliary_info_pos += AES_CTR_IV_SIZE;
4669 if (!sc->cenc.use_subsamples)
4671 /* decrypt the whole packet */
4672 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, size);
4676 /* read the subsample count */
4677 if (sizeof(uint16_t) > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
4678 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample count from the auxiliary info\n");
4679 return AVERROR_INVALIDDATA;
4682 subsample_count = AV_RB16(sc->cenc.auxiliary_info_pos);
4683 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
4685 for (; subsample_count > 0; subsample_count--)
4687 if (6 > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
4688 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample from the auxiliary info\n");
4689 return AVERROR_INVALIDDATA;
4692 /* read the number of clear / encrypted bytes */
4693 clear_bytes = AV_RB16(sc->cenc.auxiliary_info_pos);
4694 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
4695 encrypted_bytes = AV_RB32(sc->cenc.auxiliary_info_pos);
4696 sc->cenc.auxiliary_info_pos += sizeof(uint32_t);
4698 if ((uint64_t)clear_bytes + encrypted_bytes > input_end - input) {
4699 av_log(c->fc, AV_LOG_ERROR, "subsample size exceeds the packet size left\n");
4700 return AVERROR_INVALIDDATA;
4703 /* skip the clear bytes */
4704 input += clear_bytes;
4706 /* decrypt the encrypted bytes */
4707 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, encrypted_bytes);
4708 input += encrypted_bytes;
4711 if (input < input_end) {
4712 av_log(c->fc, AV_LOG_ERROR, "leftover packet bytes after subsample processing\n");
4713 return AVERROR_INVALIDDATA;
4719 static int mov_seek_auxiliary_info(AVFormatContext *s, MOVStreamContext *sc)
4721 size_t auxiliary_info_seek_offset = 0;
4724 if (sc->cenc.auxiliary_info_default_size) {
4725 auxiliary_info_seek_offset = (size_t)sc->cenc.auxiliary_info_default_size * sc->current_sample;
4726 } else if (sc->cenc.auxiliary_info_sizes) {
4727 if (sc->current_sample > sc->cenc.auxiliary_info_sizes_count) {
4728 av_log(s, AV_LOG_ERROR, "current sample %d greater than the number of auxiliary info sample sizes %"SIZE_SPECIFIER"\n",
4729 sc->current_sample, sc->cenc.auxiliary_info_sizes_count);
4730 return AVERROR_INVALIDDATA;
4733 for (i = 0; i < sc->current_sample; i++) {
4734 auxiliary_info_seek_offset += sc->cenc.auxiliary_info_sizes[i];
4738 if (auxiliary_info_seek_offset > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info) {
4739 av_log(s, AV_LOG_ERROR, "auxiliary info offset %"SIZE_SPECIFIER" greater than auxiliary info size %"SIZE_SPECIFIER"\n",
4740 auxiliary_info_seek_offset, (size_t)(sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info));
4741 return AVERROR_INVALIDDATA;
4744 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info + auxiliary_info_seek_offset;
4749 static const MOVParseTableEntry mov_default_parse_table[] = {
4750 { MKTAG('A','C','L','R'), mov_read_aclr },
4751 { MKTAG('A','P','R','G'), mov_read_avid },
4752 { MKTAG('A','A','L','P'), mov_read_avid },
4753 { MKTAG('A','R','E','S'), mov_read_ares },
4754 { MKTAG('a','v','s','s'), mov_read_avss },
4755 { MKTAG('c','h','p','l'), mov_read_chpl },
4756 { MKTAG('c','o','6','4'), mov_read_stco },
4757 { MKTAG('c','o','l','r'), mov_read_colr },
4758 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
4759 { MKTAG('d','i','n','f'), mov_read_default },
4760 { MKTAG('D','p','x','E'), mov_read_dpxe },
4761 { MKTAG('d','r','e','f'), mov_read_dref },
4762 { MKTAG('e','d','t','s'), mov_read_default },
4763 { MKTAG('e','l','s','t'), mov_read_elst },
4764 { MKTAG('e','n','d','a'), mov_read_enda },
4765 { MKTAG('f','i','e','l'), mov_read_fiel },
4766 { MKTAG('a','d','r','m'), mov_read_adrm },
4767 { MKTAG('f','t','y','p'), mov_read_ftyp },
4768 { MKTAG('g','l','b','l'), mov_read_glbl },
4769 { MKTAG('h','d','l','r'), mov_read_hdlr },
4770 { MKTAG('i','l','s','t'), mov_read_ilst },
4771 { MKTAG('j','p','2','h'), mov_read_jp2h },
4772 { MKTAG('m','d','a','t'), mov_read_mdat },
4773 { MKTAG('m','d','h','d'), mov_read_mdhd },
4774 { MKTAG('m','d','i','a'), mov_read_default },
4775 { MKTAG('m','e','t','a'), mov_read_meta },
4776 { MKTAG('m','i','n','f'), mov_read_default },
4777 { MKTAG('m','o','o','f'), mov_read_moof },
4778 { MKTAG('m','o','o','v'), mov_read_moov },
4779 { MKTAG('m','v','e','x'), mov_read_default },
4780 { MKTAG('m','v','h','d'), mov_read_mvhd },
4781 { MKTAG('S','M','I',' '), mov_read_svq3 },
4782 { MKTAG('a','l','a','c'), mov_read_alac }, /* alac specific atom */
4783 { MKTAG('a','v','c','C'), mov_read_glbl },
4784 { MKTAG('p','a','s','p'), mov_read_pasp },
4785 { MKTAG('s','i','d','x'), mov_read_sidx },
4786 { MKTAG('s','t','b','l'), mov_read_default },
4787 { MKTAG('s','t','c','o'), mov_read_stco },
4788 { MKTAG('s','t','p','s'), mov_read_stps },
4789 { MKTAG('s','t','r','f'), mov_read_strf },
4790 { MKTAG('s','t','s','c'), mov_read_stsc },
4791 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
4792 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
4793 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
4794 { MKTAG('s','t','t','s'), mov_read_stts },
4795 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
4796 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
4797 { MKTAG('t','f','d','t'), mov_read_tfdt },
4798 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
4799 { MKTAG('t','r','a','k'), mov_read_trak },
4800 { MKTAG('t','r','a','f'), mov_read_default },
4801 { MKTAG('t','r','e','f'), mov_read_default },
4802 { MKTAG('t','m','c','d'), mov_read_tmcd },
4803 { MKTAG('c','h','a','p'), mov_read_chap },
4804 { MKTAG('t','r','e','x'), mov_read_trex },
4805 { MKTAG('t','r','u','n'), mov_read_trun },
4806 { MKTAG('u','d','t','a'), mov_read_default },
4807 { MKTAG('w','a','v','e'), mov_read_wave },
4808 { MKTAG('e','s','d','s'), mov_read_esds },
4809 { MKTAG('d','a','c','3'), mov_read_dac3 }, /* AC-3 info */
4810 { MKTAG('d','e','c','3'), mov_read_dec3 }, /* EAC-3 info */
4811 { MKTAG('d','d','t','s'), mov_read_ddts }, /* DTS audio descriptor */
4812 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
4813 { MKTAG('w','f','e','x'), mov_read_wfex },
4814 { MKTAG('c','m','o','v'), mov_read_cmov },
4815 { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */
4816 { MKTAG('d','v','c','1'), mov_read_dvc1 },
4817 { MKTAG('s','b','g','p'), mov_read_sbgp },
4818 { MKTAG('h','v','c','C'), mov_read_glbl },
4819 { MKTAG('u','u','i','d'), mov_read_uuid },
4820 { MKTAG('C','i','n', 0x8e), mov_read_targa_y216 },
4821 { MKTAG('f','r','e','e'), mov_read_free },
4822 { MKTAG('-','-','-','-'), mov_read_custom },
4823 { MKTAG('s','i','n','f'), mov_read_default },
4824 { MKTAG('f','r','m','a'), mov_read_frma },
4825 { MKTAG('s','e','n','c'), mov_read_senc },
4826 { MKTAG('s','a','i','z'), mov_read_saiz },
4830 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4832 int64_t total_size = 0;
4836 if (c->atom_depth > 10) {
4837 av_log(c->fc, AV_LOG_ERROR, "Atoms too deeply nested\n");
4838 return AVERROR_INVALIDDATA;
4843 atom.size = INT64_MAX;
4844 while (total_size + 8 <= atom.size && !avio_feof(pb)) {
4845 int (*parse)(MOVContext*, AVIOContext*, MOVAtom) = NULL;
4848 if (atom.size >= 8) {
4849 a.size = avio_rb32(pb);
4850 a.type = avio_rl32(pb);
4851 if (a.type == MKTAG('f','r','e','e') &&
4855 uint32_t *type = (uint32_t *)buf + 1;
4856 if (avio_read(pb, buf, 8) != 8)
4857 return AVERROR_INVALIDDATA;
4858 avio_seek(pb, -8, SEEK_CUR);
4859 if (*type == MKTAG('m','v','h','d') ||
4860 *type == MKTAG('c','m','o','v')) {
4861 av_log(c->fc, AV_LOG_ERROR, "Detected moov in a free atom.\n");
4862 a.type = MKTAG('m','o','o','v');
4865 if (atom.type != MKTAG('r','o','o','t') &&
4866 atom.type != MKTAG('m','o','o','v'))
4868 if (a.type == MKTAG('t','r','a','k') || a.type == MKTAG('m','d','a','t'))
4870 av_log(c->fc, AV_LOG_ERROR, "Broken file, trak/mdat not at top-level\n");
4877 if (a.size == 1 && total_size + 8 <= atom.size) { /* 64 bit extended size */
4878 a.size = avio_rb64(pb) - 8;
4882 av_log(c->fc, AV_LOG_TRACE, "type: %08x '%.4s' parent:'%.4s' sz: %"PRId64" %"PRId64" %"PRId64"\n",
4883 a.type, (char*)&a.type, (char*)&atom.type, a.size, total_size, atom.size);
4885 a.size = atom.size - total_size + 8;
4890 a.size = FFMIN(a.size, atom.size - total_size);
4892 for (i = 0; mov_default_parse_table[i].type; i++)
4893 if (mov_default_parse_table[i].type == a.type) {
4894 parse = mov_default_parse_table[i].parse;
4898 // container is user data
4899 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
4900 atom.type == MKTAG('i','l','s','t')))
4901 parse = mov_read_udta_string;
4903 // Supports parsing the QuickTime Metadata Keys.
4904 // https://developer.apple.com/library/mac/documentation/QuickTime/QTFF/Metadata/Metadata.html
4905 if (!parse && c->found_hdlr_mdta &&
4906 atom.type == MKTAG('m','e','t','a') &&
4907 a.type == MKTAG('k','e','y','s')) {
4908 parse = mov_read_keys;
4911 if (!parse) { /* skip leaf atoms data */
4912 avio_skip(pb, a.size);
4914 int64_t start_pos = avio_tell(pb);
4916 int err = parse(c, pb, a);
4921 if (c->found_moov && c->found_mdat &&
4922 ((!pb->seekable || c->fc->flags & AVFMT_FLAG_IGNIDX || c->fragment_index_complete) ||
4923 start_pos + a.size == avio_size(pb))) {
4924 if (!pb->seekable || c->fc->flags & AVFMT_FLAG_IGNIDX || c->fragment_index_complete)
4925 c->next_root_atom = start_pos + a.size;
4929 left = a.size - avio_tell(pb) + start_pos;
4930 if (left > 0) /* skip garbage at atom end */
4931 avio_skip(pb, left);
4932 else if (left < 0) {
4933 av_log(c->fc, AV_LOG_WARNING,
4934 "overread end of atom '%.4s' by %"PRId64" bytes\n",
4935 (char*)&a.type, -left);
4936 avio_seek(pb, left, SEEK_CUR);
4940 total_size += a.size;
4943 if (total_size < atom.size && atom.size < 0x7ffff)
4944 avio_skip(pb, atom.size - total_size);
4950 static int mov_probe(AVProbeData *p)
4955 int moov_offset = -1;
4957 /* check file header */
4960 /* ignore invalid offset */
4961 if ((offset + 8) > (unsigned int)p->buf_size)
4963 tag = AV_RL32(p->buf + offset + 4);
4965 /* check for obvious tags */
4966 case MKTAG('m','o','o','v'):
4967 moov_offset = offset + 4;
4968 case MKTAG('m','d','a','t'):
4969 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
4970 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
4971 case MKTAG('f','t','y','p'):
4972 if (AV_RB32(p->buf+offset) < 8 &&
4973 (AV_RB32(p->buf+offset) != 1 ||
4974 offset + 12 > (unsigned int)p->buf_size ||
4975 AV_RB64(p->buf+offset + 8) == 0)) {
4976 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
4977 } else if (tag == MKTAG('f','t','y','p') &&
4978 ( AV_RL32(p->buf + offset + 8) == MKTAG('j','p','2',' ')
4979 || AV_RL32(p->buf + offset + 8) == MKTAG('j','p','x',' ')
4981 score = FFMAX(score, 5);
4983 score = AVPROBE_SCORE_MAX;
4985 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
4987 /* those are more common words, so rate then a bit less */
4988 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
4989 case MKTAG('w','i','d','e'):
4990 case MKTAG('f','r','e','e'):
4991 case MKTAG('j','u','n','k'):
4992 case MKTAG('p','i','c','t'):
4993 score = FFMAX(score, AVPROBE_SCORE_MAX - 5);
4994 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
4996 case MKTAG(0x82,0x82,0x7f,0x7d):
4997 case MKTAG('s','k','i','p'):
4998 case MKTAG('u','u','i','d'):
4999 case MKTAG('p','r','f','l'):
5000 /* if we only find those cause probedata is too small at least rate them */
5001 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
5002 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5005 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5008 if(score > AVPROBE_SCORE_MAX - 50 && moov_offset != -1) {
5009 /* moov atom in the header - we should make sure that this is not a
5010 * MOV-packed MPEG-PS */
5011 offset = moov_offset;
5013 while(offset < (p->buf_size - 16)){ /* Sufficient space */
5014 /* We found an actual hdlr atom */
5015 if(AV_RL32(p->buf + offset ) == MKTAG('h','d','l','r') &&
5016 AV_RL32(p->buf + offset + 8) == MKTAG('m','h','l','r') &&
5017 AV_RL32(p->buf + offset + 12) == MKTAG('M','P','E','G')){
5018 av_log(NULL, AV_LOG_WARNING, "Found media data tag MPEG indicating this is a MOV-packed MPEG-PS.\n");
5019 /* We found a media handler reference atom describing an
5020 * MPEG-PS-in-MOV, return a
5021 * low score to force expanding the probe window until
5022 * mpegps_probe finds what it needs */
5033 // must be done after parsing all trak because there's no order requirement
5034 static void mov_read_chapters(AVFormatContext *s)
5036 MOVContext *mov = s->priv_data;
5037 AVStream *st = NULL;
5038 MOVStreamContext *sc;
5042 for (i = 0; i < s->nb_streams; i++)
5043 if (s->streams[i]->id == mov->chapter_track) {
5048 av_log(s, AV_LOG_ERROR, "Referenced QT chapter track not found\n");
5052 st->discard = AVDISCARD_ALL;
5054 cur_pos = avio_tell(sc->pb);
5056 for (i = 0; i < st->nb_index_entries; i++) {
5057 AVIndexEntry *sample = &st->index_entries[i];
5058 int64_t end = i+1 < st->nb_index_entries ? st->index_entries[i+1].timestamp : st->duration;
5063 if (end < sample->timestamp) {
5064 av_log(s, AV_LOG_WARNING, "ignoring stream duration which is shorter than chapters\n");
5065 end = AV_NOPTS_VALUE;
5068 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
5069 av_log(s, AV_LOG_ERROR, "Chapter %d not found in file\n", i);
5073 // the first two bytes are the length of the title
5074 len = avio_rb16(sc->pb);
5075 if (len > sample->size-2)
5077 title_len = 2*len + 1;
5078 if (!(title = av_mallocz(title_len)))
5081 // The samples could theoretically be in any encoding if there's an encd
5082 // atom following, but in practice are only utf-8 or utf-16, distinguished
5083 // instead by the presence of a BOM
5087 ch = avio_rb16(sc->pb);
5089 avio_get_str16be(sc->pb, len, title, title_len);
5090 else if (ch == 0xfffe)
5091 avio_get_str16le(sc->pb, len, title, title_len);
5094 if (len == 1 || len == 2)
5097 avio_get_str(sc->pb, INT_MAX, title + 2, len - 1);
5101 avpriv_new_chapter(s, i, st->time_base, sample->timestamp, end, title);
5105 avio_seek(sc->pb, cur_pos, SEEK_SET);
5108 static int parse_timecode_in_framenum_format(AVFormatContext *s, AVStream *st,
5109 uint32_t value, int flags)
5112 char buf[AV_TIMECODE_STR_SIZE];
5113 AVRational rate = st->avg_frame_rate;
5114 int ret = av_timecode_init(&tc, rate, flags, 0, s);
5117 av_dict_set(&st->metadata, "timecode",
5118 av_timecode_make_string(&tc, buf, value), 0);
5122 static int mov_read_rtmd_track(AVFormatContext *s, AVStream *st)
5124 MOVStreamContext *sc = st->priv_data;
5125 char buf[AV_TIMECODE_STR_SIZE];
5126 int64_t cur_pos = avio_tell(sc->pb);
5127 int hh, mm, ss, ff, drop;
5129 if (!st->nb_index_entries)
5132 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
5133 avio_skip(s->pb, 13);
5134 hh = avio_r8(s->pb);
5135 mm = avio_r8(s->pb);
5136 ss = avio_r8(s->pb);
5137 drop = avio_r8(s->pb);
5138 ff = avio_r8(s->pb);
5139 snprintf(buf, AV_TIMECODE_STR_SIZE, "%02d:%02d:%02d%c%02d",
5140 hh, mm, ss, drop ? ';' : ':', ff);
5141 av_dict_set(&st->metadata, "timecode", buf, 0);
5143 avio_seek(sc->pb, cur_pos, SEEK_SET);
5147 static int mov_read_timecode_track(AVFormatContext *s, AVStream *st)
5149 MOVStreamContext *sc = st->priv_data;
5151 int64_t cur_pos = avio_tell(sc->pb);
5154 if (!st->nb_index_entries)
5157 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
5158 value = avio_rb32(s->pb);
5160 if (sc->tmcd_flags & 0x0001) flags |= AV_TIMECODE_FLAG_DROPFRAME;
5161 if (sc->tmcd_flags & 0x0002) flags |= AV_TIMECODE_FLAG_24HOURSMAX;
5162 if (sc->tmcd_flags & 0x0004) flags |= AV_TIMECODE_FLAG_ALLOWNEGATIVE;
5164 /* Assume Counter flag is set to 1 in tmcd track (even though it is likely
5165 * not the case) and thus assume "frame number format" instead of QT one.
5166 * No sample with tmcd track can be found with a QT timecode at the moment,
5167 * despite what the tmcd track "suggests" (Counter flag set to 0 means QT
5169 parse_timecode_in_framenum_format(s, st, value, flags);
5171 avio_seek(sc->pb, cur_pos, SEEK_SET);
5175 static int mov_read_close(AVFormatContext *s)
5177 MOVContext *mov = s->priv_data;
5180 for (i = 0; i < s->nb_streams; i++) {
5181 AVStream *st = s->streams[i];
5182 MOVStreamContext *sc = st->priv_data;
5187 av_freep(&sc->ctts_data);
5188 for (j = 0; j < sc->drefs_count; j++) {
5189 av_freep(&sc->drefs[j].path);
5190 av_freep(&sc->drefs[j].dir);
5192 av_freep(&sc->drefs);
5194 sc->drefs_count = 0;
5196 if (!sc->pb_is_copied)
5197 ff_format_io_close(s, &sc->pb);
5200 av_freep(&sc->chunk_offsets);
5201 av_freep(&sc->stsc_data);
5202 av_freep(&sc->sample_sizes);
5203 av_freep(&sc->keyframes);
5204 av_freep(&sc->stts_data);
5205 av_freep(&sc->stps_data);
5206 av_freep(&sc->elst_data);
5207 av_freep(&sc->rap_group);
5208 av_freep(&sc->display_matrix);
5211 for (j = 0; j < sc->stsd_count; j++)
5212 av_free(sc->extradata[j]);
5213 av_freep(&sc->extradata);
5214 av_freep(&sc->extradata_size);
5216 av_freep(&sc->cenc.auxiliary_info);
5217 av_freep(&sc->cenc.auxiliary_info_sizes);
5218 av_aes_ctr_free(sc->cenc.aes_ctr);
5221 if (mov->dv_demux) {
5222 avformat_free_context(mov->dv_fctx);
5223 mov->dv_fctx = NULL;
5226 if (mov->meta_keys) {
5227 for (i = 1; i < mov->meta_keys_count; i++) {
5228 av_freep(&mov->meta_keys[i]);
5230 av_freep(&mov->meta_keys);
5233 av_freep(&mov->trex_data);
5234 av_freep(&mov->bitrates);
5236 for (i = 0; i < mov->fragment_index_count; i++) {
5237 MOVFragmentIndex* index = mov->fragment_index_data[i];
5238 av_freep(&index->items);
5239 av_freep(&mov->fragment_index_data[i]);
5241 av_freep(&mov->fragment_index_data);
5243 av_freep(&mov->aes_decrypt);
5248 static int tmcd_is_referenced(AVFormatContext *s, int tmcd_id)
5252 for (i = 0; i < s->nb_streams; i++) {
5253 AVStream *st = s->streams[i];
5254 MOVStreamContext *sc = st->priv_data;
5256 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO &&
5257 sc->timecode_track == tmcd_id)
5263 /* look for a tmcd track not referenced by any video track, and export it globally */
5264 static void export_orphan_timecode(AVFormatContext *s)
5268 for (i = 0; i < s->nb_streams; i++) {
5269 AVStream *st = s->streams[i];
5271 if (st->codecpar->codec_tag == MKTAG('t','m','c','d') &&
5272 !tmcd_is_referenced(s, i + 1)) {
5273 AVDictionaryEntry *tcr = av_dict_get(st->metadata, "timecode", NULL, 0);
5275 av_dict_set(&s->metadata, "timecode", tcr->value, 0);
5282 static int read_tfra(MOVContext *mov, AVIOContext *f)
5284 MOVFragmentIndex* index = NULL;
5285 int version, fieldlength, i, j;
5286 int64_t pos = avio_tell(f);
5287 uint32_t size = avio_rb32(f);
5290 if (avio_rb32(f) != MKBETAG('t', 'f', 'r', 'a')) {
5293 av_log(mov->fc, AV_LOG_VERBOSE, "found tfra\n");
5294 index = av_mallocz(sizeof(MOVFragmentIndex));
5296 return AVERROR(ENOMEM);
5299 tmp = av_realloc_array(mov->fragment_index_data,
5300 mov->fragment_index_count + 1,
5301 sizeof(MOVFragmentIndex*));
5304 return AVERROR(ENOMEM);
5306 mov->fragment_index_data = tmp;
5307 mov->fragment_index_data[mov->fragment_index_count++] = index;
5309 version = avio_r8(f);
5311 index->track_id = avio_rb32(f);
5312 fieldlength = avio_rb32(f);
5313 index->item_count = avio_rb32(f);
5314 index->items = av_mallocz_array(
5315 index->item_count, sizeof(MOVFragmentIndexItem));
5316 if (!index->items) {
5317 index->item_count = 0;
5318 return AVERROR(ENOMEM);
5320 for (i = 0; i < index->item_count; i++) {
5321 int64_t time, offset;
5323 time = avio_rb64(f);
5324 offset = avio_rb64(f);
5326 time = avio_rb32(f);
5327 offset = avio_rb32(f);
5329 index->items[i].time = time;
5330 index->items[i].moof_offset = offset;
5331 for (j = 0; j < ((fieldlength >> 4) & 3) + 1; j++)
5333 for (j = 0; j < ((fieldlength >> 2) & 3) + 1; j++)
5335 for (j = 0; j < ((fieldlength >> 0) & 3) + 1; j++)
5339 avio_seek(f, pos + size, SEEK_SET);
5343 static int mov_read_mfra(MOVContext *c, AVIOContext *f)
5345 int64_t stream_size = avio_size(f);
5346 int64_t original_pos = avio_tell(f);
5350 if ((seek_ret = avio_seek(f, stream_size - 4, SEEK_SET)) < 0) {
5354 mfra_size = avio_rb32(f);
5355 if (mfra_size < 0 || mfra_size > stream_size) {
5356 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (unreasonable size)\n");
5359 if ((seek_ret = avio_seek(f, -mfra_size, SEEK_CUR)) < 0) {
5363 if (avio_rb32(f) != mfra_size) {
5364 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (size mismatch)\n");
5367 if (avio_rb32(f) != MKBETAG('m', 'f', 'r', 'a')) {
5368 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (tag mismatch)\n");
5371 av_log(c->fc, AV_LOG_VERBOSE, "stream has mfra\n");
5373 ret = read_tfra(c, f);
5379 seek_ret = avio_seek(f, original_pos, SEEK_SET);
5381 av_log(c->fc, AV_LOG_ERROR,
5382 "failed to seek back after looking for mfra\n");
5388 static int mov_read_header(AVFormatContext *s)
5390 MOVContext *mov = s->priv_data;
5391 AVIOContext *pb = s->pb;
5393 MOVAtom atom = { AV_RL32("root") };
5396 if (mov->decryption_key_len != 0 && mov->decryption_key_len != AES_CTR_KEY_SIZE) {
5397 av_log(s, AV_LOG_ERROR, "Invalid decryption key len %d expected %d\n",
5398 mov->decryption_key_len, AES_CTR_KEY_SIZE);
5399 return AVERROR(EINVAL);
5403 mov->trak_index = -1;
5404 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
5406 atom.size = avio_size(pb);
5408 atom.size = INT64_MAX;
5410 /* check MOV header */
5412 if (mov->moov_retry)
5413 avio_seek(pb, 0, SEEK_SET);
5414 if ((err = mov_read_default(mov, pb, atom)) < 0) {
5415 av_log(s, AV_LOG_ERROR, "error reading header\n");
5419 } while (pb->seekable && !mov->found_moov && !mov->moov_retry++);
5420 if (!mov->found_moov) {
5421 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
5423 return AVERROR_INVALIDDATA;
5425 av_log(mov->fc, AV_LOG_TRACE, "on_parse_exit_offset=%"PRId64"\n", avio_tell(pb));
5428 if (mov->chapter_track > 0 && !mov->ignore_chapters)
5429 mov_read_chapters(s);
5430 for (i = 0; i < s->nb_streams; i++)
5431 if (s->streams[i]->codecpar->codec_tag == AV_RL32("tmcd")) {
5432 mov_read_timecode_track(s, s->streams[i]);
5433 } else if (s->streams[i]->codecpar->codec_tag == AV_RL32("rtmd")) {
5434 mov_read_rtmd_track(s, s->streams[i]);
5438 /* copy timecode metadata from tmcd tracks to the related video streams */
5439 for (i = 0; i < s->nb_streams; i++) {
5440 AVStream *st = s->streams[i];
5441 MOVStreamContext *sc = st->priv_data;
5442 if (sc->timecode_track > 0) {
5443 AVDictionaryEntry *tcr;
5444 int tmcd_st_id = -1;
5446 for (j = 0; j < s->nb_streams; j++)
5447 if (s->streams[j]->id == sc->timecode_track)
5450 if (tmcd_st_id < 0 || tmcd_st_id == i)
5452 tcr = av_dict_get(s->streams[tmcd_st_id]->metadata, "timecode", NULL, 0);
5454 av_dict_set(&st->metadata, "timecode", tcr->value, 0);
5457 export_orphan_timecode(s);
5459 for (i = 0; i < s->nb_streams; i++) {
5460 AVStream *st = s->streams[i];
5461 MOVStreamContext *sc = st->priv_data;
5462 fix_timescale(mov, sc);
5463 if(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id == AV_CODEC_ID_AAC) {
5464 st->skip_samples = sc->start_pad;
5466 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && sc->nb_frames_for_fps > 0 && sc->duration_for_fps > 0)
5467 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
5468 sc->time_scale*(int64_t)sc->nb_frames_for_fps, sc->duration_for_fps, INT_MAX);
5469 if (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE) {
5470 if (st->codecpar->width <= 0 || st->codecpar->height <= 0) {
5471 st->codecpar->width = sc->width;
5472 st->codecpar->height = sc->height;
5474 if (st->codecpar->codec_id == AV_CODEC_ID_DVD_SUBTITLE) {
5475 if ((err = mov_rewrite_dvd_sub_extradata(st)) < 0)
5479 if (mov->handbrake_version &&
5480 mov->handbrake_version <= 1000000*0 + 1000*10 + 2 && // 0.10.2
5481 st->codecpar->codec_id == AV_CODEC_ID_MP3
5483 av_log(s, AV_LOG_VERBOSE, "Forcing full parsing for mp3 stream\n");
5484 st->need_parsing = AVSTREAM_PARSE_FULL;
5488 if (mov->trex_data) {
5489 for (i = 0; i < s->nb_streams; i++) {
5490 AVStream *st = s->streams[i];
5491 MOVStreamContext *sc = st->priv_data;
5492 if (st->duration > 0)
5493 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale / st->duration;
5497 if (mov->use_mfra_for > 0) {
5498 for (i = 0; i < s->nb_streams; i++) {
5499 AVStream *st = s->streams[i];
5500 MOVStreamContext *sc = st->priv_data;
5501 if (sc->duration_for_fps > 0) {
5502 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale /
5503 sc->duration_for_fps;
5508 for (i = 0; i < mov->bitrates_count && i < s->nb_streams; i++) {
5509 if (mov->bitrates[i]) {
5510 s->streams[i]->codecpar->bit_rate = mov->bitrates[i];
5514 ff_rfps_calculate(s);
5516 for (i = 0; i < s->nb_streams; i++) {
5517 AVStream *st = s->streams[i];
5518 MOVStreamContext *sc = st->priv_data;
5520 switch (st->codecpar->codec_type) {
5521 case AVMEDIA_TYPE_AUDIO:
5522 err = ff_replaygain_export(st, s->metadata);
5528 case AVMEDIA_TYPE_VIDEO:
5529 if (sc->display_matrix) {
5530 AVPacketSideData *sd, *tmp;
5532 tmp = av_realloc_array(st->side_data,
5533 st->nb_side_data + 1, sizeof(*tmp));
5535 return AVERROR(ENOMEM);
5537 st->side_data = tmp;
5540 sd = &st->side_data[st->nb_side_data - 1];
5541 sd->type = AV_PKT_DATA_DISPLAYMATRIX;
5542 sd->size = sizeof(int32_t) * 9;
5543 sd->data = (uint8_t*)sc->display_matrix;
5544 sc->display_matrix = NULL;
5549 ff_configure_buffers_for_index(s, AV_TIME_BASE);
5554 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
5556 AVIndexEntry *sample = NULL;
5557 int64_t best_dts = INT64_MAX;
5559 for (i = 0; i < s->nb_streams; i++) {
5560 AVStream *avst = s->streams[i];
5561 MOVStreamContext *msc = avst->priv_data;
5562 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
5563 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
5564 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
5565 av_log(s, AV_LOG_TRACE, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
5566 if (!sample || (!s->pb->seekable && current_sample->pos < sample->pos) ||
5568 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
5569 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
5570 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
5571 sample = current_sample;
5580 static int should_retry(AVIOContext *pb, int error_code) {
5581 if (error_code == AVERROR_EOF || avio_feof(pb))
5587 static int mov_switch_root(AVFormatContext *s, int64_t target)
5589 MOVContext *mov = s->priv_data;
5591 int already_read = 0;
5593 if (avio_seek(s->pb, target, SEEK_SET) != target) {
5594 av_log(mov->fc, AV_LOG_ERROR, "root atom offset 0x%"PRIx64": partial file\n", target);
5595 return AVERROR_INVALIDDATA;
5598 mov->next_root_atom = 0;
5600 for (i = 0; i < mov->fragment_index_count; i++) {
5601 MOVFragmentIndex *index = mov->fragment_index_data[i];
5603 for (j = 0; j < index->item_count; j++) {
5604 MOVFragmentIndexItem *item = &index->items[j];
5606 mov->next_root_atom = item->moof_offset;
5607 break; // Advance to next index in outer loop
5608 } else if (item->moof_offset == target) {
5609 index->current_item = FFMIN(j, index->current_item);
5610 if (item->headers_read)
5612 item->headers_read = 1;
5617 index->current_item = 0;
5623 mov->found_mdat = 0;
5625 if (mov_read_default(mov, s->pb, (MOVAtom){ AV_RL32("root"), INT64_MAX }) < 0 ||
5628 av_log(s, AV_LOG_TRACE, "read fragments, offset 0x%"PRIx64"\n", avio_tell(s->pb));
5633 static int mov_change_extradata(MOVStreamContext *sc, AVPacket *pkt)
5635 uint8_t *side, *extradata;
5638 /* Save the current index. */
5639 sc->last_stsd_index = sc->stsc_data[sc->stsc_index].id - 1;
5641 /* Notify the decoder that extradata changed. */
5642 extradata_size = sc->extradata_size[sc->last_stsd_index];
5643 extradata = sc->extradata[sc->last_stsd_index];
5644 if (extradata_size > 0 && extradata) {
5645 side = av_packet_new_side_data(pkt,
5646 AV_PKT_DATA_NEW_EXTRADATA,
5649 return AVERROR(ENOMEM);
5650 memcpy(side, extradata, extradata_size);
5656 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
5658 MOVContext *mov = s->priv_data;
5659 MOVStreamContext *sc;
5660 AVIndexEntry *sample;
5661 AVStream *st = NULL;
5665 sample = mov_find_next_sample(s, &st);
5666 if (!sample || (mov->next_root_atom && sample->pos > mov->next_root_atom)) {
5667 if (!mov->next_root_atom)
5669 if ((ret = mov_switch_root(s, mov->next_root_atom)) < 0)
5674 /* must be done just before reading, to avoid infinite loop on sample */
5675 sc->current_sample++;
5677 if (mov->next_root_atom) {
5678 sample->pos = FFMIN(sample->pos, mov->next_root_atom);
5679 sample->size = FFMIN(sample->size, (mov->next_root_atom - sample->pos));
5682 if (st->discard != AVDISCARD_ALL) {
5683 int64_t ret64 = avio_seek(sc->pb, sample->pos, SEEK_SET);
5684 if (ret64 != sample->pos) {
5685 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
5686 sc->ffindex, sample->pos);
5687 sc->current_sample -= should_retry(sc->pb, ret64);
5688 return AVERROR_INVALIDDATA;
5691 if( st->discard == AVDISCARD_NONKEY && 0==(sample->flags & AVINDEX_KEYFRAME) ) {
5692 av_log(mov->fc, AV_LOG_DEBUG, "Nonkey frame from stream %d discarded due to AVDISCARD_NONKEY\n", sc->ffindex);
5696 ret = av_get_packet(sc->pb, pkt, sample->size);
5698 sc->current_sample -= should_retry(sc->pb, ret);
5701 if (sc->has_palette) {
5704 pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, AVPALETTE_SIZE);
5706 av_log(mov->fc, AV_LOG_ERROR, "Cannot append palette to packet\n");
5708 memcpy(pal, sc->palette, AVPALETTE_SIZE);
5709 sc->has_palette = 0;
5712 #if CONFIG_DV_DEMUXER
5713 if (mov->dv_demux && sc->dv_audio_container) {
5714 avpriv_dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size, pkt->pos);
5715 av_freep(&pkt->data);
5717 ret = avpriv_dv_get_packet(mov->dv_demux, pkt);
5722 if (st->codecpar->codec_id == AV_CODEC_ID_MP3 && !st->need_parsing && pkt->size > 4) {
5723 if (ff_mpa_check_header(AV_RB32(pkt->data)) < 0)
5724 st->need_parsing = AVSTREAM_PARSE_FULL;
5728 pkt->stream_index = sc->ffindex;
5729 pkt->dts = sample->timestamp;
5730 if (sample->flags & AVINDEX_DISCARD_FRAME) {
5731 pkt->flags |= AV_PKT_FLAG_DISCARD;
5733 if (sc->ctts_data && sc->ctts_index < sc->ctts_count) {
5734 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
5735 /* update ctts context */
5737 if (sc->ctts_index < sc->ctts_count &&
5738 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
5740 sc->ctts_sample = 0;
5743 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
5744 st->index_entries[sc->current_sample].timestamp : st->duration;
5745 pkt->duration = next_dts - pkt->dts;
5746 pkt->pts = pkt->dts;
5748 if (st->discard == AVDISCARD_ALL)
5750 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? AV_PKT_FLAG_KEY : 0;
5751 pkt->pos = sample->pos;
5753 /* Multiple stsd handling. */
5754 if (sc->stsc_data) {
5755 /* Keep track of the stsc index for the given sample, then check
5756 * if the stsd index is different from the last used one. */
5758 if (sc->stsc_index < sc->stsc_count - 1 &&
5759 mov_get_stsc_samples(sc, sc->stsc_index) == sc->stsc_sample) {
5761 sc->stsc_sample = 0;
5762 /* Do not check indexes after a switch. */
5763 } else if (sc->stsc_data[sc->stsc_index].id > 0 &&
5764 sc->stsc_data[sc->stsc_index].id - 1 < sc->stsd_count &&
5765 sc->stsc_data[sc->stsc_index].id - 1 != sc->last_stsd_index) {
5766 ret = mov_change_extradata(sc, pkt);
5773 aax_filter(pkt->data, pkt->size, mov);
5775 if (sc->cenc.aes_ctr) {
5776 ret = cenc_filter(mov, sc, pkt->data, pkt->size);
5785 static int mov_seek_fragment(AVFormatContext *s, AVStream *st, int64_t timestamp)
5787 MOVContext *mov = s->priv_data;
5790 if (!mov->fragment_index_complete)
5793 for (i = 0; i < mov->fragment_index_count; i++) {
5794 if (mov->fragment_index_data[i]->track_id == st->id) {
5795 MOVFragmentIndex *index = mov->fragment_index_data[i];
5796 for (j = index->item_count - 1; j >= 0; j--) {
5797 if (index->items[j].time <= timestamp) {
5798 if (index->items[j].headers_read)
5801 return mov_switch_root(s, index->items[j].moof_offset);
5810 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
5812 MOVStreamContext *sc = st->priv_data;
5813 int sample, time_sample;
5816 int ret = mov_seek_fragment(s, st, timestamp);
5820 sample = av_index_search_timestamp(st, timestamp, flags);
5821 av_log(s, AV_LOG_TRACE, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
5822 if (sample < 0 && st->nb_index_entries && timestamp < st->index_entries[0].timestamp)
5824 if (sample < 0) /* not sure what to do */
5825 return AVERROR_INVALIDDATA;
5826 sc->current_sample = sample;
5827 av_log(s, AV_LOG_TRACE, "stream %d, found sample %d\n", st->index, sc->current_sample);
5828 /* adjust ctts index */
5829 if (sc->ctts_data) {
5831 for (i = 0; i < sc->ctts_count; i++) {
5832 int next = time_sample + sc->ctts_data[i].count;
5833 if (next > sc->current_sample) {
5835 sc->ctts_sample = sc->current_sample - time_sample;
5842 /* adjust stsd index */
5844 for (i = 0; i < sc->stsc_count; i++) {
5845 int next = time_sample + mov_get_stsc_samples(sc, i);
5846 if (next > sc->current_sample) {
5848 sc->stsc_sample = sc->current_sample - time_sample;
5854 ret = mov_seek_auxiliary_info(s, sc);
5862 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
5864 MOVContext *mc = s->priv_data;
5869 if (stream_index >= s->nb_streams)
5870 return AVERROR_INVALIDDATA;
5872 st = s->streams[stream_index];
5873 sample = mov_seek_stream(s, st, sample_time, flags);
5877 if (mc->seek_individually) {
5878 /* adjust seek timestamp to found sample timestamp */
5879 int64_t seek_timestamp = st->index_entries[sample].timestamp;
5881 for (i = 0; i < s->nb_streams; i++) {
5883 MOVStreamContext *sc = s->streams[i]->priv_data;
5885 st->skip_samples = (sample_time <= 0) ? sc->start_pad : 0;
5887 if (stream_index == i)
5890 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
5891 mov_seek_stream(s, st, timestamp, flags);
5894 for (i = 0; i < s->nb_streams; i++) {
5895 MOVStreamContext *sc;
5898 sc->current_sample = 0;
5901 MOVStreamContext *sc;
5902 AVIndexEntry *entry = mov_find_next_sample(s, &st);
5904 return AVERROR_INVALIDDATA;
5906 if (sc->ffindex == stream_index && sc->current_sample == sample)
5908 sc->current_sample++;
5914 #define OFFSET(x) offsetof(MOVContext, x)
5915 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
5916 static const AVOption mov_options[] = {
5917 {"use_absolute_path",
5918 "allow using absolute path when opening alias, this is a possible security issue",
5919 OFFSET(use_absolute_path), AV_OPT_TYPE_BOOL, {.i64 = 0},
5921 {"seek_streams_individually",
5922 "Seek each stream individually to the to the closest point",
5923 OFFSET(seek_individually), AV_OPT_TYPE_BOOL, { .i64 = 1 },
5925 {"ignore_editlist", "", OFFSET(ignore_editlist), AV_OPT_TYPE_BOOL, {.i64 = 0},
5927 {"ignore_chapters", "", OFFSET(ignore_chapters), AV_OPT_TYPE_BOOL, {.i64 = 0},
5930 "use mfra for fragment timestamps",
5931 OFFSET(use_mfra_for), AV_OPT_TYPE_INT, {.i64 = FF_MOV_FLAG_MFRA_AUTO},
5932 -1, FF_MOV_FLAG_MFRA_PTS, FLAGS,
5934 {"auto", "auto", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_AUTO}, 0, 0,
5935 FLAGS, "use_mfra_for" },
5936 {"dts", "dts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_DTS}, 0, 0,
5937 FLAGS, "use_mfra_for" },
5938 {"pts", "pts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_PTS}, 0, 0,
5939 FLAGS, "use_mfra_for" },
5940 { "export_all", "Export unrecognized metadata entries", OFFSET(export_all),
5941 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
5942 { "export_xmp", "Export full XMP metadata", OFFSET(export_xmp),
5943 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
5944 { "activation_bytes", "Secret bytes for Audible AAX files", OFFSET(activation_bytes),
5945 AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
5946 { "audible_fixed_key", // extracted from libAAX_SDK.so and AAXSDKWin.dll files!
5947 "Fixed key used for handling Audible AAX files", OFFSET(audible_fixed_key),
5948 AV_OPT_TYPE_BINARY, {.str="77214d4b196a87cd520045fd20a51d67"},
5949 .flags = AV_OPT_FLAG_DECODING_PARAM },
5950 { "decryption_key", "The media decryption key (hex)", OFFSET(decryption_key), AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
5951 { "enable_drefs", "Enable external track support.", OFFSET(enable_drefs), AV_OPT_TYPE_BOOL,
5952 {.i64 = 0}, 0, 1, FLAGS },
5957 static const AVClass mov_class = {
5958 .class_name = "mov,mp4,m4a,3gp,3g2,mj2",
5959 .item_name = av_default_item_name,
5960 .option = mov_options,
5961 .version = LIBAVUTIL_VERSION_INT,
5964 AVInputFormat ff_mov_demuxer = {
5965 .name = "mov,mp4,m4a,3gp,3g2,mj2",
5966 .long_name = NULL_IF_CONFIG_SMALL("QuickTime / MOV"),
5967 .priv_class = &mov_class,
5968 .priv_data_size = sizeof(MOVContext),
5969 .extensions = "mov,mp4,m4a,3gp,3g2,mj2",
5970 .read_probe = mov_probe,
5971 .read_header = mov_read_header,
5972 .read_packet = mov_read_packet,
5973 .read_close = mov_read_close,
5974 .read_seek = mov_read_seek,
5975 .flags = AVFMT_NO_BYTE_SEEK,