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_metadata_hmmt(MOVContext *c, AVIOContext *pb, unsigned len)
277 if (c->ignore_chapters)
280 n_hmmt = avio_rb32(pb);
281 for (i = 0; i < n_hmmt && !pb->eof_reached; i++) {
282 int moment_time = avio_rb32(pb);
283 avpriv_new_chapter(c->fc, i, av_make_q(1, 1000), moment_time, AV_NOPTS_VALUE, NULL);
288 static int mov_read_udta_string(MOVContext *c, AVIOContext *pb, MOVAtom atom)
291 char key2[32], language[4] = {0};
293 const char *key = NULL;
294 uint16_t langcode = 0;
295 uint32_t data_type = 0, str_size, str_size_alloc;
296 int (*parse)(MOVContext*, AVIOContext*, unsigned, const char*) = NULL;
301 case MKTAG( '@','P','R','M'): key = "premiere_version"; raw = 1; break;
302 case MKTAG( '@','P','R','Q'): key = "quicktime_version"; raw = 1; break;
303 case MKTAG( 'X','M','P','_'):
304 if (c->export_xmp) { key = "xmp"; raw = 1; } break;
305 case MKTAG( 'a','A','R','T'): key = "album_artist"; break;
306 case MKTAG( 'a','k','I','D'): key = "account_type";
307 parse = mov_metadata_int8_no_padding; break;
308 case MKTAG( 'a','p','I','D'): key = "account_id"; break;
309 case MKTAG( 'c','a','t','g'): key = "category"; break;
310 case MKTAG( 'c','p','i','l'): key = "compilation";
311 parse = mov_metadata_int8_no_padding; break;
312 case MKTAG( 'c','p','r','t'): key = "copyright"; break;
313 case MKTAG( 'd','e','s','c'): key = "description"; break;
314 case MKTAG( 'd','i','s','k'): key = "disc";
315 parse = mov_metadata_track_or_disc_number; break;
316 case MKTAG( 'e','g','i','d'): key = "episode_uid";
317 parse = mov_metadata_int8_no_padding; break;
318 case MKTAG( 'F','I','R','M'): key = "firmware"; raw = 1; break;
319 case MKTAG( 'g','n','r','e'): key = "genre";
320 parse = mov_metadata_gnre; break;
321 case MKTAG( 'h','d','v','d'): key = "hd_video";
322 parse = mov_metadata_int8_no_padding; break;
323 case MKTAG( 'H','M','M','T'):
324 return mov_metadata_hmmt(c, pb, atom.size);
325 case MKTAG( 'k','e','y','w'): key = "keywords"; break;
326 case MKTAG( 'l','d','e','s'): key = "synopsis"; break;
327 case MKTAG( 'l','o','c','i'):
328 return mov_metadata_loci(c, pb, atom.size);
329 case MKTAG( 'p','c','s','t'): key = "podcast";
330 parse = mov_metadata_int8_no_padding; break;
331 case MKTAG( 'p','g','a','p'): key = "gapless_playback";
332 parse = mov_metadata_int8_no_padding; break;
333 case MKTAG( 'p','u','r','d'): key = "purchase_date"; break;
334 case MKTAG( 'r','t','n','g'): key = "rating";
335 parse = mov_metadata_int8_no_padding; break;
336 case MKTAG( 's','o','a','a'): key = "sort_album_artist"; break;
337 case MKTAG( 's','o','a','l'): key = "sort_album"; break;
338 case MKTAG( 's','o','a','r'): key = "sort_artist"; break;
339 case MKTAG( 's','o','c','o'): key = "sort_composer"; break;
340 case MKTAG( 's','o','n','m'): key = "sort_name"; break;
341 case MKTAG( 's','o','s','n'): key = "sort_show"; break;
342 case MKTAG( 's','t','i','k'): key = "media_type";
343 parse = mov_metadata_int8_no_padding; break;
344 case MKTAG( 't','r','k','n'): key = "track";
345 parse = mov_metadata_track_or_disc_number; break;
346 case MKTAG( 't','v','e','n'): key = "episode_id"; break;
347 case MKTAG( 't','v','e','s'): key = "episode_sort";
348 parse = mov_metadata_int8_bypass_padding; break;
349 case MKTAG( 't','v','n','n'): key = "network"; break;
350 case MKTAG( 't','v','s','h'): key = "show"; break;
351 case MKTAG( 't','v','s','n'): key = "season_number";
352 parse = mov_metadata_int8_bypass_padding; break;
353 case MKTAG(0xa9,'A','R','T'): key = "artist"; break;
354 case MKTAG(0xa9,'P','R','D'): key = "producer"; break;
355 case MKTAG(0xa9,'a','l','b'): key = "album"; break;
356 case MKTAG(0xa9,'a','u','t'): key = "artist"; break;
357 case MKTAG(0xa9,'c','h','p'): key = "chapter"; break;
358 case MKTAG(0xa9,'c','m','t'): key = "comment"; break;
359 case MKTAG(0xa9,'c','o','m'): key = "composer"; break;
360 case MKTAG(0xa9,'c','p','y'): key = "copyright"; break;
361 case MKTAG(0xa9,'d','a','y'): key = "date"; break;
362 case MKTAG(0xa9,'d','i','r'): key = "director"; break;
363 case MKTAG(0xa9,'d','i','s'): key = "disclaimer"; break;
364 case MKTAG(0xa9,'e','d','1'): key = "edit_date"; break;
365 case MKTAG(0xa9,'e','n','c'): key = "encoder"; break;
366 case MKTAG(0xa9,'f','m','t'): key = "original_format"; break;
367 case MKTAG(0xa9,'g','e','n'): key = "genre"; break;
368 case MKTAG(0xa9,'g','r','p'): key = "grouping"; break;
369 case MKTAG(0xa9,'h','s','t'): key = "host_computer"; break;
370 case MKTAG(0xa9,'i','n','f'): key = "comment"; break;
371 case MKTAG(0xa9,'l','y','r'): key = "lyrics"; break;
372 case MKTAG(0xa9,'m','a','k'): key = "make"; break;
373 case MKTAG(0xa9,'m','o','d'): key = "model"; break;
374 case MKTAG(0xa9,'n','a','m'): key = "title"; break;
375 case MKTAG(0xa9,'o','p','e'): key = "original_artist"; break;
376 case MKTAG(0xa9,'p','r','d'): key = "producer"; break;
377 case MKTAG(0xa9,'p','r','f'): key = "performers"; break;
378 case MKTAG(0xa9,'r','e','q'): key = "playback_requirements"; break;
379 case MKTAG(0xa9,'s','r','c'): key = "original_source"; break;
380 case MKTAG(0xa9,'s','t','3'): key = "subtitle"; break;
381 case MKTAG(0xa9,'s','w','r'): key = "encoder"; break;
382 case MKTAG(0xa9,'t','o','o'): key = "encoder"; break;
383 case MKTAG(0xa9,'t','r','k'): key = "track"; break;
384 case MKTAG(0xa9,'u','r','l'): key = "URL"; break;
385 case MKTAG(0xa9,'w','r','n'): key = "warning"; break;
386 case MKTAG(0xa9,'w','r','t'): key = "composer"; break;
387 case MKTAG(0xa9,'x','y','z'): key = "location"; break;
390 if (c->itunes_metadata && atom.size > 8) {
391 int data_size = avio_rb32(pb);
392 int tag = avio_rl32(pb);
393 if (tag == MKTAG('d','a','t','a') && data_size <= atom.size) {
394 data_type = avio_rb32(pb); // type
395 avio_rb32(pb); // unknown
396 str_size = data_size - 16;
399 if (atom.type == MKTAG('c', 'o', 'v', 'r')) {
400 int ret = mov_read_covr(c, pb, data_type, str_size);
402 av_log(c->fc, AV_LOG_ERROR, "Error parsing cover art.\n");
405 } else if (!key && c->found_hdlr_mdta && c->meta_keys) {
406 uint32_t index = AV_RB32(&atom.type);
407 if (index < c->meta_keys_count) {
408 key = c->meta_keys[index];
410 av_log(c->fc, AV_LOG_WARNING,
411 "The index of 'data' is out of range: %d >= %d.\n",
412 index, c->meta_keys_count);
416 } else if (atom.size > 4 && key && !c->itunes_metadata && !raw) {
417 str_size = avio_rb16(pb); // string length
418 if (str_size > atom.size) {
420 avio_seek(pb, -2, SEEK_CUR);
421 av_log(c->fc, AV_LOG_WARNING, "UDTA parsing failed retrying raw\n");
424 langcode = avio_rb16(pb);
425 ff_mov_lang_to_iso639(langcode, language);
428 str_size = atom.size;
430 if (c->export_all && !key) {
431 snprintf(tmp_key, 5, "%.4s", (char*)&atom.type);
437 if (atom.size < 0 || str_size >= INT_MAX/2)
438 return AVERROR_INVALIDDATA;
440 // Allocates enough space if data_type is a int32 or float32 number, otherwise
441 // worst-case requirement for output string in case of utf8 coded input
442 num = (data_type >= 21 && data_type <= 23);
443 str_size_alloc = (num ? 512 : (raw ? str_size : str_size * 2)) + 1;
444 str = av_mallocz(str_size_alloc);
446 return AVERROR(ENOMEM);
449 parse(c, pb, str_size, key);
451 if (!raw && (data_type == 3 || (data_type == 0 && (langcode < 0x400 || langcode == 0x7fff)))) { // MAC Encoded
452 mov_read_mac_string(c, pb, str_size, str, str_size_alloc);
453 } else if (data_type == 21) { // BE signed integer, variable size
456 val = (int8_t)avio_r8(pb);
457 else if (str_size == 2)
458 val = (int16_t)avio_rb16(pb);
459 else if (str_size == 3)
460 val = ((int32_t)(avio_rb24(pb)<<8))>>8;
461 else if (str_size == 4)
462 val = (int32_t)avio_rb32(pb);
463 if (snprintf(str, str_size_alloc, "%d", val) >= str_size_alloc) {
464 av_log(c->fc, AV_LOG_ERROR,
465 "Failed to store the number (%d) in string.\n", val);
467 return AVERROR_INVALIDDATA;
469 } else if (data_type == 22) { // BE unsigned integer, variable size
470 unsigned int val = 0;
473 else if (str_size == 2)
475 else if (str_size == 3)
477 else if (str_size == 4)
479 if (snprintf(str, str_size_alloc, "%u", val) >= str_size_alloc) {
480 av_log(c->fc, AV_LOG_ERROR,
481 "Failed to store the number (%u) in string.\n", val);
483 return AVERROR_INVALIDDATA;
485 } else if (data_type == 23 && str_size >= 4) { // BE float32
486 float val = av_int2float(avio_rb32(pb));
487 if (snprintf(str, str_size_alloc, "%f", val) >= str_size_alloc) {
488 av_log(c->fc, AV_LOG_ERROR,
489 "Failed to store the float32 number (%f) in string.\n", val);
491 return AVERROR_INVALIDDATA;
494 int ret = ffio_read_size(pb, str, str_size);
501 c->fc->event_flags |= AVFMT_EVENT_FLAG_METADATA_UPDATED;
502 av_dict_set(&c->fc->metadata, key, str, 0);
503 if (*language && strcmp(language, "und")) {
504 snprintf(key2, sizeof(key2), "%s-%s", key, language);
505 av_dict_set(&c->fc->metadata, key2, str, 0);
507 if (!strcmp(key, "encoder")) {
508 int major, minor, micro;
509 if (sscanf(str, "HandBrake %d.%d.%d", &major, &minor, µ) == 3) {
510 c->handbrake_version = 1000000*major + 1000*minor + micro;
514 av_log(c->fc, AV_LOG_TRACE, "lang \"%3s\" ", language);
515 av_log(c->fc, AV_LOG_TRACE, "tag \"%s\" value \"%s\" atom \"%.4s\" %d %"PRId64"\n",
516 key, str, (char*)&atom.type, str_size_alloc, atom.size);
522 static int mov_read_chpl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
525 int i, nb_chapters, str_len, version;
529 if (c->ignore_chapters)
532 if ((atom.size -= 5) < 0)
535 version = avio_r8(pb);
538 avio_rb32(pb); // ???
539 nb_chapters = avio_r8(pb);
541 for (i = 0; i < nb_chapters; i++) {
545 start = avio_rb64(pb);
546 str_len = avio_r8(pb);
548 if ((atom.size -= 9+str_len) < 0)
551 ret = ffio_read_size(pb, str, str_len);
555 avpriv_new_chapter(c->fc, i, (AVRational){1,10000000}, start, AV_NOPTS_VALUE, str);
560 #define MIN_DATA_ENTRY_BOX_SIZE 12
561 static int mov_read_dref(MOVContext *c, AVIOContext *pb, MOVAtom atom)
564 MOVStreamContext *sc;
567 if (c->fc->nb_streams < 1)
569 st = c->fc->streams[c->fc->nb_streams-1];
572 avio_rb32(pb); // version + flags
573 entries = avio_rb32(pb);
575 entries > (atom.size - 1) / MIN_DATA_ENTRY_BOX_SIZE + 1 ||
576 entries >= UINT_MAX / sizeof(*sc->drefs))
577 return AVERROR_INVALIDDATA;
581 sc->drefs = av_mallocz(entries * sizeof(*sc->drefs));
583 return AVERROR(ENOMEM);
584 sc->drefs_count = entries;
586 for (i = 0; i < entries; i++) {
587 MOVDref *dref = &sc->drefs[i];
588 uint32_t size = avio_rb32(pb);
589 int64_t next = avio_tell(pb) + size - 4;
592 return AVERROR_INVALIDDATA;
594 dref->type = avio_rl32(pb);
595 avio_rb32(pb); // version + flags
596 av_log(c->fc, AV_LOG_TRACE, "type %.4s size %d\n", (char*)&dref->type, size);
598 if (dref->type == MKTAG('a','l','i','s') && size > 150) {
599 /* macintosh alias record */
600 uint16_t volume_len, len;
606 volume_len = avio_r8(pb);
607 volume_len = FFMIN(volume_len, 27);
608 ret = ffio_read_size(pb, dref->volume, 27);
611 dref->volume[volume_len] = 0;
612 av_log(c->fc, AV_LOG_DEBUG, "volume %s, len %d\n", dref->volume, volume_len);
617 len = FFMIN(len, 63);
618 ret = ffio_read_size(pb, dref->filename, 63);
621 dref->filename[len] = 0;
622 av_log(c->fc, AV_LOG_DEBUG, "filename %s, len %d\n", dref->filename, len);
626 /* read next level up_from_alias/down_to_target */
627 dref->nlvl_from = avio_rb16(pb);
628 dref->nlvl_to = avio_rb16(pb);
629 av_log(c->fc, AV_LOG_DEBUG, "nlvl from %d, nlvl to %d\n",
630 dref->nlvl_from, dref->nlvl_to);
634 for (type = 0; type != -1 && avio_tell(pb) < next; ) {
637 type = avio_rb16(pb);
639 av_log(c->fc, AV_LOG_DEBUG, "type %d, len %d\n", type, len);
642 if (type == 2) { // absolute path
644 dref->path = av_mallocz(len+1);
646 return AVERROR(ENOMEM);
648 ret = ffio_read_size(pb, dref->path, len);
650 av_freep(&dref->path);
653 if (len > volume_len && !strncmp(dref->path, dref->volume, volume_len)) {
655 memmove(dref->path, dref->path+volume_len, len);
658 // trim string of any ending zeros
659 for (j = len - 1; j >= 0; j--) {
660 if (dref->path[j] == 0)
665 for (j = 0; j < len; j++)
666 if (dref->path[j] == ':' || dref->path[j] == 0)
668 av_log(c->fc, AV_LOG_DEBUG, "path %s\n", dref->path);
669 } else if (type == 0) { // directory name
671 dref->dir = av_malloc(len+1);
673 return AVERROR(ENOMEM);
675 ret = ffio_read_size(pb, dref->dir, len);
677 av_freep(&dref->dir);
681 for (j = 0; j < len; j++)
682 if (dref->dir[j] == ':')
684 av_log(c->fc, AV_LOG_DEBUG, "dir %s\n", dref->dir);
689 av_log(c->fc, AV_LOG_DEBUG, "Unknown dref type 0x08%x size %d\n",
694 avio_seek(pb, next, SEEK_SET);
699 static int mov_read_hdlr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
703 uint32_t av_unused ctype;
708 avio_r8(pb); /* version */
709 avio_rb24(pb); /* flags */
712 ctype = avio_rl32(pb);
713 type = avio_rl32(pb); /* component subtype */
715 av_log(c->fc, AV_LOG_TRACE, "ctype= %.4s (0x%08x)\n", (char*)&ctype, ctype);
716 av_log(c->fc, AV_LOG_TRACE, "stype= %.4s\n", (char*)&type);
718 if (c->trak_index < 0) { // meta not inside a trak
719 if (type == MKTAG('m','d','t','a')) {
720 c->found_hdlr_mdta = 1;
725 st = c->fc->streams[c->fc->nb_streams-1];
727 if (type == MKTAG('v','i','d','e'))
728 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
729 else if (type == MKTAG('s','o','u','n'))
730 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
731 else if (type == MKTAG('m','1','a',' '))
732 st->codecpar->codec_id = AV_CODEC_ID_MP2;
733 else if ((type == MKTAG('s','u','b','p')) || (type == MKTAG('c','l','c','p')))
734 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
736 avio_rb32(pb); /* component manufacture */
737 avio_rb32(pb); /* component flags */
738 avio_rb32(pb); /* component flags mask */
740 title_size = atom.size - 24;
741 if (title_size > 0) {
742 title_str = av_malloc(title_size + 1); /* Add null terminator */
744 return AVERROR(ENOMEM);
746 ret = ffio_read_size(pb, title_str, title_size);
748 av_freep(&title_str);
751 title_str[title_size] = 0;
753 int off = (!c->isom && title_str[0] == title_size - 1);
754 av_dict_set(&st->metadata, "handler_name", title_str + off, 0);
756 av_freep(&title_str);
762 int ff_mov_read_esds(AVFormatContext *fc, AVIOContext *pb)
767 if (fc->nb_streams < 1)
769 st = fc->streams[fc->nb_streams-1];
771 avio_rb32(pb); /* version + flags */
772 ff_mp4_read_descr(fc, pb, &tag);
773 if (tag == MP4ESDescrTag) {
774 ff_mp4_parse_es_descr(pb, NULL);
776 avio_rb16(pb); /* ID */
778 ff_mp4_read_descr(fc, pb, &tag);
779 if (tag == MP4DecConfigDescrTag)
780 ff_mp4_read_dec_config_descr(fc, st, pb);
784 static int mov_read_esds(MOVContext *c, AVIOContext *pb, MOVAtom atom)
786 return ff_mov_read_esds(c->fc, pb);
789 static int mov_read_dac3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
792 enum AVAudioServiceType *ast;
793 int ac3info, acmod, lfeon, bsmod;
795 if (c->fc->nb_streams < 1)
797 st = c->fc->streams[c->fc->nb_streams-1];
799 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
802 return AVERROR(ENOMEM);
804 ac3info = avio_rb24(pb);
805 bsmod = (ac3info >> 14) & 0x7;
806 acmod = (ac3info >> 11) & 0x7;
807 lfeon = (ac3info >> 10) & 0x1;
808 st->codecpar->channels = ((int[]){2,1,2,3,3,4,4,5})[acmod] + lfeon;
809 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
811 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
813 if (st->codecpar->channels > 1 && bsmod == 0x7)
814 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
816 #if FF_API_LAVF_AVCTX
817 FF_DISABLE_DEPRECATION_WARNINGS
818 st->codec->audio_service_type = *ast;
819 FF_ENABLE_DEPRECATION_WARNINGS
825 static int mov_read_dec3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
828 enum AVAudioServiceType *ast;
829 int eac3info, acmod, lfeon, bsmod;
831 if (c->fc->nb_streams < 1)
833 st = c->fc->streams[c->fc->nb_streams-1];
835 ast = (enum AVAudioServiceType*)av_stream_new_side_data(st, AV_PKT_DATA_AUDIO_SERVICE_TYPE,
838 return AVERROR(ENOMEM);
840 /* No need to parse fields for additional independent substreams and its
841 * associated dependent substreams since libavcodec's E-AC-3 decoder
842 * does not support them yet. */
843 avio_rb16(pb); /* data_rate and num_ind_sub */
844 eac3info = avio_rb24(pb);
845 bsmod = (eac3info >> 12) & 0x1f;
846 acmod = (eac3info >> 9) & 0x7;
847 lfeon = (eac3info >> 8) & 0x1;
848 st->codecpar->channel_layout = avpriv_ac3_channel_layout_tab[acmod];
850 st->codecpar->channel_layout |= AV_CH_LOW_FREQUENCY;
851 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
853 if (st->codecpar->channels > 1 && bsmod == 0x7)
854 *ast = AV_AUDIO_SERVICE_TYPE_KARAOKE;
856 #if FF_API_LAVF_AVCTX
857 FF_DISABLE_DEPRECATION_WARNINGS
858 st->codec->audio_service_type = *ast;
859 FF_ENABLE_DEPRECATION_WARNINGS
865 static int mov_read_ddts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
867 const uint32_t ddts_size = 20;
870 uint32_t frame_duration_code = 0;
871 uint32_t channel_layout_code = 0;
874 buf = av_malloc(ddts_size + FF_INPUT_BUFFER_PADDING_SIZE);
876 return AVERROR(ENOMEM);
878 if (avio_read(pb, buf, ddts_size) < ddts_size) {
880 return AVERROR_INVALIDDATA;
883 init_get_bits(&gb, buf, 8*ddts_size);
885 if (c->fc->nb_streams < 1) {
888 st = c->fc->streams[c->fc->nb_streams-1];
890 st->codecpar->sample_rate = get_bits_long(&gb, 32);
891 if (st->codecpar->sample_rate <= 0) {
892 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
893 return AVERROR_INVALIDDATA;
895 skip_bits_long(&gb, 32); /* max bitrate */
896 st->codecpar->bit_rate = get_bits_long(&gb, 32);
897 st->codecpar->bits_per_coded_sample = get_bits(&gb, 8);
898 frame_duration_code = get_bits(&gb, 2);
899 skip_bits(&gb, 30); /* various fields */
900 channel_layout_code = get_bits(&gb, 16);
902 st->codecpar->frame_size =
903 (frame_duration_code == 0) ? 512 :
904 (frame_duration_code == 1) ? 1024 :
905 (frame_duration_code == 2) ? 2048 :
906 (frame_duration_code == 3) ? 4096 : 0;
908 if (channel_layout_code > 0xff) {
909 av_log(c->fc, AV_LOG_WARNING, "Unsupported DTS audio channel layout");
911 st->codecpar->channel_layout =
912 ((channel_layout_code & 0x1) ? AV_CH_FRONT_CENTER : 0) |
913 ((channel_layout_code & 0x2) ? AV_CH_FRONT_LEFT : 0) |
914 ((channel_layout_code & 0x2) ? AV_CH_FRONT_RIGHT : 0) |
915 ((channel_layout_code & 0x4) ? AV_CH_SIDE_LEFT : 0) |
916 ((channel_layout_code & 0x4) ? AV_CH_SIDE_RIGHT : 0) |
917 ((channel_layout_code & 0x8) ? AV_CH_LOW_FREQUENCY : 0);
919 st->codecpar->channels = av_get_channel_layout_nb_channels(st->codecpar->channel_layout);
924 static int mov_read_chan(MOVContext *c, AVIOContext *pb, MOVAtom atom)
928 if (c->fc->nb_streams < 1)
930 st = c->fc->streams[c->fc->nb_streams-1];
935 /* skip version and flags */
938 ff_mov_read_chan(c->fc, pb, st, atom.size - 4);
943 static int mov_read_wfex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
948 if (c->fc->nb_streams < 1)
950 st = c->fc->streams[c->fc->nb_streams-1];
952 if ((ret = ff_get_wav_header(c->fc, pb, st->codecpar, atom.size, 0)) < 0)
953 av_log(c->fc, AV_LOG_WARNING, "get_wav_header failed\n");
958 static int mov_read_pasp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
960 const int num = avio_rb32(pb);
961 const int den = avio_rb32(pb);
964 if (c->fc->nb_streams < 1)
966 st = c->fc->streams[c->fc->nb_streams-1];
968 if ((st->sample_aspect_ratio.den != 1 || st->sample_aspect_ratio.num) && // default
969 (den != st->sample_aspect_ratio.den || num != st->sample_aspect_ratio.num)) {
970 av_log(c->fc, AV_LOG_WARNING,
971 "sample aspect ratio already set to %d:%d, ignoring 'pasp' atom (%d:%d)\n",
972 st->sample_aspect_ratio.num, st->sample_aspect_ratio.den,
974 } else if (den != 0) {
975 av_reduce(&st->sample_aspect_ratio.num, &st->sample_aspect_ratio.den,
981 /* this atom contains actual media data */
982 static int mov_read_mdat(MOVContext *c, AVIOContext *pb, MOVAtom atom)
984 if (atom.size == 0) /* wrong one (MP4) */
987 return 0; /* now go for moov */
990 #define DRM_BLOB_SIZE 56
992 static int mov_read_adrm(MOVContext *c, AVIOContext *pb, MOVAtom atom)
994 uint8_t intermediate_key[20];
995 uint8_t intermediate_iv[20];
998 uint8_t file_checksum[20];
999 uint8_t calculated_checksum[20];
1003 uint8_t *activation_bytes = c->activation_bytes;
1004 uint8_t *fixed_key = c->audible_fixed_key;
1008 sha = av_sha_alloc();
1010 return AVERROR(ENOMEM);
1011 c->aes_decrypt = av_aes_alloc();
1012 if (!c->aes_decrypt) {
1013 ret = AVERROR(ENOMEM);
1017 /* drm blob processing */
1018 avio_read(pb, output, 8); // go to offset 8, absolute position 0x251
1019 avio_read(pb, input, DRM_BLOB_SIZE);
1020 avio_read(pb, output, 4); // go to offset 4, absolute position 0x28d
1021 avio_read(pb, file_checksum, 20);
1023 av_log(c->fc, AV_LOG_INFO, "[aax] file checksum == "); // required by external tools
1024 for (i = 0; i < 20; i++)
1025 av_log(c->fc, AV_LOG_INFO, "%02x", file_checksum[i]);
1026 av_log(c->fc, AV_LOG_INFO, "\n");
1028 /* verify activation data */
1029 if (!activation_bytes) {
1030 av_log(c->fc, AV_LOG_WARNING, "[aax] activation_bytes option is missing!\n");
1031 ret = 0; /* allow ffprobe to continue working on .aax files */
1034 if (c->activation_bytes_size != 4) {
1035 av_log(c->fc, AV_LOG_FATAL, "[aax] activation_bytes value needs to be 4 bytes!\n");
1036 ret = AVERROR(EINVAL);
1040 /* verify fixed key */
1041 if (c->audible_fixed_key_size != 16) {
1042 av_log(c->fc, AV_LOG_FATAL, "[aax] audible_fixed_key value needs to be 16 bytes!\n");
1043 ret = AVERROR(EINVAL);
1047 /* AAX (and AAX+) key derivation */
1048 av_sha_init(sha, 160);
1049 av_sha_update(sha, fixed_key, 16);
1050 av_sha_update(sha, activation_bytes, 4);
1051 av_sha_final(sha, intermediate_key);
1052 av_sha_init(sha, 160);
1053 av_sha_update(sha, fixed_key, 16);
1054 av_sha_update(sha, intermediate_key, 20);
1055 av_sha_update(sha, activation_bytes, 4);
1056 av_sha_final(sha, intermediate_iv);
1057 av_sha_init(sha, 160);
1058 av_sha_update(sha, intermediate_key, 16);
1059 av_sha_update(sha, intermediate_iv, 16);
1060 av_sha_final(sha, calculated_checksum);
1061 if (memcmp(calculated_checksum, file_checksum, 20)) { // critical error
1062 av_log(c->fc, AV_LOG_ERROR, "[aax] mismatch in checksums!\n");
1063 ret = AVERROR_INVALIDDATA;
1066 av_aes_init(c->aes_decrypt, intermediate_key, 128, 1);
1067 av_aes_crypt(c->aes_decrypt, output, input, DRM_BLOB_SIZE >> 4, intermediate_iv, 1);
1068 for (i = 0; i < 4; i++) {
1069 // file data (in output) is stored in big-endian mode
1070 if (activation_bytes[i] != output[3 - i]) { // critical error
1071 av_log(c->fc, AV_LOG_ERROR, "[aax] error in drm blob decryption!\n");
1072 ret = AVERROR_INVALIDDATA;
1076 memcpy(c->file_key, output + 8, 16);
1077 memcpy(input, output + 26, 16);
1078 av_sha_init(sha, 160);
1079 av_sha_update(sha, input, 16);
1080 av_sha_update(sha, c->file_key, 16);
1081 av_sha_update(sha, fixed_key, 16);
1082 av_sha_final(sha, c->file_iv);
1090 // Audible AAX (and AAX+) bytestream decryption
1091 static int aax_filter(uint8_t *input, int size, MOVContext *c)
1094 unsigned char iv[16];
1096 memcpy(iv, c->file_iv, 16); // iv is overwritten
1097 blocks = size >> 4; // trailing bytes are not encrypted!
1098 av_aes_init(c->aes_decrypt, c->file_key, 128, 1);
1099 av_aes_crypt(c->aes_decrypt, input, input, blocks, iv, 1);
1104 /* read major brand, minor version and compatible brands and store them as metadata */
1105 static int mov_read_ftyp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1108 int comp_brand_size;
1109 char* comp_brands_str;
1110 uint8_t type[5] = {0};
1111 int ret = ffio_read_size(pb, type, 4);
1115 if (strcmp(type, "qt "))
1117 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
1118 av_dict_set(&c->fc->metadata, "major_brand", type, 0);
1119 minor_ver = avio_rb32(pb); /* minor version */
1120 av_dict_set_int(&c->fc->metadata, "minor_version", minor_ver, 0);
1122 comp_brand_size = atom.size - 8;
1123 if (comp_brand_size < 0)
1124 return AVERROR_INVALIDDATA;
1125 comp_brands_str = av_malloc(comp_brand_size + 1); /* Add null terminator */
1126 if (!comp_brands_str)
1127 return AVERROR(ENOMEM);
1129 ret = ffio_read_size(pb, comp_brands_str, comp_brand_size);
1131 av_freep(&comp_brands_str);
1134 comp_brands_str[comp_brand_size] = 0;
1135 av_dict_set(&c->fc->metadata, "compatible_brands", comp_brands_str, 0);
1136 av_freep(&comp_brands_str);
1141 /* this atom should contain all header atoms */
1142 static int mov_read_moov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1146 if (c->found_moov) {
1147 av_log(c->fc, AV_LOG_WARNING, "Found duplicated MOOV Atom. Skipped it\n");
1148 avio_skip(pb, atom.size);
1152 if ((ret = mov_read_default(c, pb, atom)) < 0)
1154 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
1155 /* so we don't parse the whole file if over a network */
1157 return 0; /* now go for mdat */
1160 static int mov_read_moof(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1162 if (!c->has_looked_for_mfra && c->use_mfra_for > 0) {
1163 c->has_looked_for_mfra = 1;
1166 av_log(c->fc, AV_LOG_VERBOSE, "stream has moof boxes, will look "
1168 if ((ret = mov_read_mfra(c, pb)) < 0) {
1169 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but failed to "
1170 "read the mfra (may be a live ismv)\n");
1173 av_log(c->fc, AV_LOG_VERBOSE, "found a moof box but stream is not "
1174 "seekable, can not look for mfra\n");
1177 c->fragment.moof_offset = c->fragment.implicit_offset = avio_tell(pb) - 8;
1178 av_log(c->fc, AV_LOG_TRACE, "moof offset %"PRIx64"\n", c->fragment.moof_offset);
1179 return mov_read_default(c, pb, atom);
1182 static void mov_metadata_creation_time(AVDictionary **metadata, int64_t time)
1185 if(time >= 2082844800)
1186 time -= 2082844800; /* seconds between 1904-01-01 and Epoch */
1187 avpriv_dict_set_timestamp(metadata, "creation_time", time * 1000000);
1191 static int mov_read_mdhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1194 MOVStreamContext *sc;
1196 char language[4] = {0};
1198 int64_t creation_time;
1200 if (c->fc->nb_streams < 1)
1202 st = c->fc->streams[c->fc->nb_streams-1];
1205 if (sc->time_scale) {
1206 av_log(c->fc, AV_LOG_ERROR, "Multiple mdhd?\n");
1207 return AVERROR_INVALIDDATA;
1210 version = avio_r8(pb);
1212 avpriv_request_sample(c->fc, "Version %d", version);
1213 return AVERROR_PATCHWELCOME;
1215 avio_rb24(pb); /* flags */
1217 creation_time = avio_rb64(pb);
1220 creation_time = avio_rb32(pb);
1221 avio_rb32(pb); /* modification time */
1223 mov_metadata_creation_time(&st->metadata, creation_time);
1225 sc->time_scale = avio_rb32(pb);
1226 if (sc->time_scale <= 0) {
1227 av_log(c->fc, AV_LOG_ERROR, "Invalid mdhd time scale %d\n", sc->time_scale);
1228 return AVERROR_INVALIDDATA;
1230 st->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1232 lang = avio_rb16(pb); /* language */
1233 if (ff_mov_lang_to_iso639(lang, language))
1234 av_dict_set(&st->metadata, "language", language, 0);
1235 avio_rb16(pb); /* quality */
1240 static int mov_read_mvhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1242 int64_t creation_time;
1243 int version = avio_r8(pb); /* version */
1244 avio_rb24(pb); /* flags */
1247 creation_time = avio_rb64(pb);
1250 creation_time = avio_rb32(pb);
1251 avio_rb32(pb); /* modification time */
1253 mov_metadata_creation_time(&c->fc->metadata, creation_time);
1254 c->time_scale = avio_rb32(pb); /* time scale */
1255 if (c->time_scale <= 0) {
1256 av_log(c->fc, AV_LOG_ERROR, "Invalid mvhd time scale %d\n", c->time_scale);
1257 return AVERROR_INVALIDDATA;
1259 av_log(c->fc, AV_LOG_TRACE, "time scale = %i\n", c->time_scale);
1261 c->duration = (version == 1) ? avio_rb64(pb) : avio_rb32(pb); /* duration */
1262 // set the AVCodecContext duration because the duration of individual tracks
1263 // may be inaccurate
1264 if (c->time_scale > 0 && !c->trex_data)
1265 c->fc->duration = av_rescale(c->duration, AV_TIME_BASE, c->time_scale);
1266 avio_rb32(pb); /* preferred scale */
1268 avio_rb16(pb); /* preferred volume */
1270 avio_skip(pb, 10); /* reserved */
1272 avio_skip(pb, 36); /* display matrix */
1274 avio_rb32(pb); /* preview time */
1275 avio_rb32(pb); /* preview duration */
1276 avio_rb32(pb); /* poster time */
1277 avio_rb32(pb); /* selection time */
1278 avio_rb32(pb); /* selection duration */
1279 avio_rb32(pb); /* current time */
1280 avio_rb32(pb); /* next track ID */
1285 static int mov_read_enda(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1290 if (c->fc->nb_streams < 1)
1292 st = c->fc->streams[c->fc->nb_streams-1];
1294 little_endian = avio_rb16(pb) & 0xFF;
1295 av_log(c->fc, AV_LOG_TRACE, "enda %d\n", little_endian);
1296 if (little_endian == 1) {
1297 switch (st->codecpar->codec_id) {
1298 case AV_CODEC_ID_PCM_S24BE:
1299 st->codecpar->codec_id = AV_CODEC_ID_PCM_S24LE;
1301 case AV_CODEC_ID_PCM_S32BE:
1302 st->codecpar->codec_id = AV_CODEC_ID_PCM_S32LE;
1304 case AV_CODEC_ID_PCM_F32BE:
1305 st->codecpar->codec_id = AV_CODEC_ID_PCM_F32LE;
1307 case AV_CODEC_ID_PCM_F64BE:
1308 st->codecpar->codec_id = AV_CODEC_ID_PCM_F64LE;
1317 static int mov_read_colr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1320 char color_parameter_type[5] = { 0 };
1321 uint16_t color_primaries, color_trc, color_matrix;
1324 if (c->fc->nb_streams < 1)
1326 st = c->fc->streams[c->fc->nb_streams - 1];
1328 ret = ffio_read_size(pb, color_parameter_type, 4);
1331 if (strncmp(color_parameter_type, "nclx", 4) &&
1332 strncmp(color_parameter_type, "nclc", 4)) {
1333 av_log(c->fc, AV_LOG_WARNING, "unsupported color_parameter_type %s\n",
1334 color_parameter_type);
1338 color_primaries = avio_rb16(pb);
1339 color_trc = avio_rb16(pb);
1340 color_matrix = avio_rb16(pb);
1342 av_log(c->fc, AV_LOG_TRACE,
1343 "%s: pri %d trc %d matrix %d",
1344 color_parameter_type, color_primaries, color_trc, color_matrix);
1346 if (!strncmp(color_parameter_type, "nclx", 4)) {
1347 uint8_t color_range = avio_r8(pb) >> 7;
1348 av_log(c->fc, AV_LOG_TRACE, " full %"PRIu8"", color_range);
1350 st->codecpar->color_range = AVCOL_RANGE_JPEG;
1352 st->codecpar->color_range = AVCOL_RANGE_MPEG;
1354 if (color_primaries >= AVCOL_PRI_NB)
1355 color_primaries = AVCOL_PRI_UNSPECIFIED;
1356 if (color_trc >= AVCOL_TRC_NB)
1357 color_trc = AVCOL_TRC_UNSPECIFIED;
1358 if (color_matrix >= AVCOL_SPC_NB)
1359 color_matrix = AVCOL_SPC_UNSPECIFIED;
1360 st->codecpar->color_primaries = color_primaries;
1361 st->codecpar->color_trc = color_trc;
1362 st->codecpar->color_space = color_matrix;
1363 av_log(c->fc, AV_LOG_TRACE, "\n");
1368 static int mov_read_fiel(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1371 unsigned mov_field_order;
1372 enum AVFieldOrder decoded_field_order = AV_FIELD_UNKNOWN;
1374 if (c->fc->nb_streams < 1) // will happen with jp2 files
1376 st = c->fc->streams[c->fc->nb_streams-1];
1378 return AVERROR_INVALIDDATA;
1379 mov_field_order = avio_rb16(pb);
1380 if ((mov_field_order & 0xFF00) == 0x0100)
1381 decoded_field_order = AV_FIELD_PROGRESSIVE;
1382 else if ((mov_field_order & 0xFF00) == 0x0200) {
1383 switch (mov_field_order & 0xFF) {
1384 case 0x01: decoded_field_order = AV_FIELD_TT;
1386 case 0x06: decoded_field_order = AV_FIELD_BB;
1388 case 0x09: decoded_field_order = AV_FIELD_TB;
1390 case 0x0E: decoded_field_order = AV_FIELD_BT;
1394 if (decoded_field_order == AV_FIELD_UNKNOWN && mov_field_order) {
1395 av_log(NULL, AV_LOG_ERROR, "Unknown MOV field order 0x%04x\n", mov_field_order);
1397 st->codecpar->field_order = decoded_field_order;
1402 static int mov_realloc_extradata(AVCodecParameters *par, MOVAtom atom)
1405 uint64_t size = (uint64_t)par->extradata_size + atom.size + 8 + AV_INPUT_BUFFER_PADDING_SIZE;
1406 if (size > INT_MAX || (uint64_t)atom.size > INT_MAX)
1407 return AVERROR_INVALIDDATA;
1408 if ((err = av_reallocp(&par->extradata, size)) < 0) {
1409 par->extradata_size = 0;
1412 par->extradata_size = size - AV_INPUT_BUFFER_PADDING_SIZE;
1416 /* Read a whole atom into the extradata return the size of the atom read, possibly truncated if != atom.size */
1417 static int64_t mov_read_atom_into_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1418 AVCodecParameters *par, uint8_t *buf)
1420 int64_t result = atom.size;
1423 AV_WB32(buf , atom.size + 8);
1424 AV_WL32(buf + 4, atom.type);
1425 err = ffio_read_size(pb, buf + 8, atom.size);
1427 par->extradata_size -= atom.size;
1429 } else if (err < atom.size) {
1430 av_log(c->fc, AV_LOG_WARNING, "truncated extradata\n");
1431 par->extradata_size -= atom.size - err;
1434 memset(buf + 8 + err, 0, AV_INPUT_BUFFER_PADDING_SIZE);
1438 /* FIXME modify QDM2/SVQ3/H.264 decoders to take full atom as extradata */
1439 static int mov_read_extradata(MOVContext *c, AVIOContext *pb, MOVAtom atom,
1440 enum AVCodecID codec_id)
1443 uint64_t original_size;
1446 if (c->fc->nb_streams < 1) // will happen with jp2 files
1448 st = c->fc->streams[c->fc->nb_streams-1];
1450 if (st->codecpar->codec_id != codec_id)
1451 return 0; /* unexpected codec_id - don't mess with extradata */
1453 original_size = st->codecpar->extradata_size;
1454 err = mov_realloc_extradata(st->codecpar, atom);
1458 err = mov_read_atom_into_extradata(c, pb, atom, st->codecpar, st->codecpar->extradata + original_size);
1461 return 0; // Note: this is the original behavior to ignore truncation.
1464 /* wrapper functions for reading ALAC/AVS/MJPEG/MJPEG2000 extradata atoms only for those codecs */
1465 static int mov_read_alac(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1467 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_ALAC);
1470 static int mov_read_avss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1472 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVS);
1475 static int mov_read_jp2h(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1477 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_JPEG2000);
1480 static int mov_read_dpxe(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1482 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_R10K);
1485 static int mov_read_avid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1487 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_AVUI);
1489 ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_DNXHD);
1493 static int mov_read_targa_y216(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1495 int ret = mov_read_extradata(c, pb, atom, AV_CODEC_ID_TARGA_Y216);
1497 if (!ret && c->fc->nb_streams >= 1) {
1498 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1499 if (par->extradata_size >= 40) {
1500 par->height = AV_RB16(&par->extradata[36]);
1501 par->width = AV_RB16(&par->extradata[38]);
1507 static int mov_read_ares(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1509 if (c->fc->nb_streams >= 1) {
1510 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1511 if (par->codec_tag == MKTAG('A', 'V', 'i', 'n') &&
1512 par->codec_id == AV_CODEC_ID_H264 &&
1516 cid = avio_rb16(pb);
1517 /* For AVID AVCI50, force width of 1440 to be able to select the correct SPS and PPS */
1518 if (cid == 0xd4d || cid == 0xd4e)
1521 } else if ((par->codec_tag == MKTAG('A', 'V', 'd', '1') ||
1522 par->codec_tag == MKTAG('A', 'V', 'd', 'n')) &&
1526 num = avio_rb32(pb);
1527 den = avio_rb32(pb);
1528 if (num <= 0 || den <= 0)
1530 switch (avio_rb32(pb)) {
1532 if (den >= INT_MAX / 2)
1536 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.num = num;
1537 c->fc->streams[c->fc->nb_streams-1]->display_aspect_ratio.den = den;
1544 return mov_read_avid(c, pb, atom);
1547 static int mov_read_aclr(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1551 uint64_t original_size;
1552 if (c->fc->nb_streams >= 1) {
1553 AVCodecParameters *par = c->fc->streams[c->fc->nb_streams-1]->codecpar;
1554 if (par->codec_id == AV_CODEC_ID_H264)
1556 if (atom.size == 16) {
1557 original_size = par->extradata_size;
1558 ret = mov_realloc_extradata(par, atom);
1560 length = mov_read_atom_into_extradata(c, pb, atom, par, par->extradata + original_size);
1561 if (length == atom.size) {
1562 const uint8_t range_value = par->extradata[original_size + 19];
1563 switch (range_value) {
1565 par->color_range = AVCOL_RANGE_MPEG;
1568 par->color_range = AVCOL_RANGE_JPEG;
1571 av_log(c, AV_LOG_WARNING, "ignored unknown aclr value (%d)\n", range_value);
1574 ff_dlog(c, "color_range: %d\n", par->color_range);
1576 /* For some reason the whole atom was not added to the extradata */
1577 av_log(c, AV_LOG_ERROR, "aclr not decoded - incomplete atom\n");
1580 av_log(c, AV_LOG_ERROR, "aclr not decoded - unable to add atom to extradata\n");
1583 av_log(c, AV_LOG_WARNING, "aclr not decoded - unexpected size %"PRId64"\n", atom.size);
1590 static int mov_read_svq3(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1592 return mov_read_extradata(c, pb, atom, AV_CODEC_ID_SVQ3);
1595 static int mov_read_wave(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1600 if (c->fc->nb_streams < 1)
1602 st = c->fc->streams[c->fc->nb_streams-1];
1604 if ((uint64_t)atom.size > (1<<30))
1605 return AVERROR_INVALIDDATA;
1607 if (st->codecpar->codec_id == AV_CODEC_ID_QDM2 ||
1608 st->codecpar->codec_id == AV_CODEC_ID_QDMC ||
1609 st->codecpar->codec_id == AV_CODEC_ID_SPEEX) {
1610 // pass all frma atom to codec, needed at least for QDMC and QDM2
1611 av_freep(&st->codecpar->extradata);
1612 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1615 } else if (atom.size > 8) { /* to read frma, esds atoms */
1616 if (st->codecpar->codec_id == AV_CODEC_ID_ALAC && atom.size >= 24) {
1618 ret = ffio_ensure_seekback(pb, 8);
1621 buffer = avio_rb64(pb);
1623 if ( (buffer & 0xFFFFFFFF) == MKBETAG('f','r','m','a')
1624 && buffer >> 32 <= atom.size
1625 && buffer >> 32 >= 8) {
1628 } else if (!st->codecpar->extradata_size) {
1629 #define ALAC_EXTRADATA_SIZE 36
1630 st->codecpar->extradata = av_mallocz(ALAC_EXTRADATA_SIZE + AV_INPUT_BUFFER_PADDING_SIZE);
1631 if (!st->codecpar->extradata)
1632 return AVERROR(ENOMEM);
1633 st->codecpar->extradata_size = ALAC_EXTRADATA_SIZE;
1634 AV_WB32(st->codecpar->extradata , ALAC_EXTRADATA_SIZE);
1635 AV_WB32(st->codecpar->extradata + 4, MKTAG('a','l','a','c'));
1636 AV_WB64(st->codecpar->extradata + 12, buffer);
1637 avio_read(pb, st->codecpar->extradata + 20, 16);
1638 avio_skip(pb, atom.size - 24);
1642 if ((ret = mov_read_default(c, pb, atom)) < 0)
1645 avio_skip(pb, atom.size);
1650 * This function reads atom content and puts data in extradata without tag
1651 * nor size unlike mov_read_extradata.
1653 static int mov_read_glbl(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1658 if (c->fc->nb_streams < 1)
1660 st = c->fc->streams[c->fc->nb_streams-1];
1662 if ((uint64_t)atom.size > (1<<30))
1663 return AVERROR_INVALIDDATA;
1665 if (atom.size >= 10) {
1666 // Broken files created by legacy versions of libavformat will
1667 // wrap a whole fiel atom inside of a glbl atom.
1668 unsigned size = avio_rb32(pb);
1669 unsigned type = avio_rl32(pb);
1670 avio_seek(pb, -8, SEEK_CUR);
1671 if (type == MKTAG('f','i','e','l') && size == atom.size)
1672 return mov_read_default(c, pb, atom);
1674 if (st->codecpar->extradata_size > 1 && st->codecpar->extradata) {
1675 av_log(c, AV_LOG_WARNING, "ignoring multiple glbl\n");
1678 av_freep(&st->codecpar->extradata);
1679 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size);
1686 static int mov_read_dvc1(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1689 uint8_t profile_level;
1692 if (c->fc->nb_streams < 1)
1694 st = c->fc->streams[c->fc->nb_streams-1];
1696 if (atom.size >= (1<<28) || atom.size < 7)
1697 return AVERROR_INVALIDDATA;
1699 profile_level = avio_r8(pb);
1700 if ((profile_level & 0xf0) != 0xc0)
1703 avio_seek(pb, 6, SEEK_CUR);
1704 av_freep(&st->codecpar->extradata);
1705 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 7);
1713 * An strf atom is a BITMAPINFOHEADER struct. This struct is 40 bytes itself,
1714 * but can have extradata appended at the end after the 40 bytes belonging
1717 static int mov_read_strf(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1722 if (c->fc->nb_streams < 1)
1724 if (atom.size <= 40)
1726 st = c->fc->streams[c->fc->nb_streams-1];
1728 if ((uint64_t)atom.size > (1<<30))
1729 return AVERROR_INVALIDDATA;
1732 av_freep(&st->codecpar->extradata);
1733 ret = ff_get_extradata(c->fc, st->codecpar, pb, atom.size - 40);
1740 static int mov_read_stco(MOVContext *c, AVIOContext *pb, MOVAtom atom)
1743 MOVStreamContext *sc;
1744 unsigned int i, entries;
1746 if (c->fc->nb_streams < 1)
1748 st = c->fc->streams[c->fc->nb_streams-1];
1751 avio_r8(pb); /* version */
1752 avio_rb24(pb); /* flags */
1754 entries = avio_rb32(pb);
1759 if (sc->chunk_offsets)
1760 av_log(c->fc, AV_LOG_WARNING, "Duplicated STCO atom\n");
1761 av_free(sc->chunk_offsets);
1762 sc->chunk_count = 0;
1763 sc->chunk_offsets = av_malloc_array(entries, sizeof(*sc->chunk_offsets));
1764 if (!sc->chunk_offsets)
1765 return AVERROR(ENOMEM);
1766 sc->chunk_count = entries;
1768 if (atom.type == MKTAG('s','t','c','o'))
1769 for (i = 0; i < entries && !pb->eof_reached; i++)
1770 sc->chunk_offsets[i] = avio_rb32(pb);
1771 else if (atom.type == MKTAG('c','o','6','4'))
1772 for (i = 0; i < entries && !pb->eof_reached; i++)
1773 sc->chunk_offsets[i] = avio_rb64(pb);
1775 return AVERROR_INVALIDDATA;
1777 sc->chunk_count = i;
1779 if (pb->eof_reached)
1786 * Compute codec id for 'lpcm' tag.
1787 * See CoreAudioTypes and AudioStreamBasicDescription at Apple.
1789 enum AVCodecID ff_mov_get_lpcm_codec_id(int bps, int flags)
1796 return ff_get_pcm_codec_id(bps, flags & 1, flags & 2, flags & 4 ? -1 : 0);
1799 static int mov_codec_id(AVStream *st, uint32_t format)
1801 int id = ff_codec_get_id(ff_codec_movaudio_tags, format);
1804 ((format & 0xFFFF) == 'm' + ('s' << 8) ||
1805 (format & 0xFFFF) == 'T' + ('S' << 8)))
1806 id = ff_codec_get_id(ff_codec_wav_tags, av_bswap32(format) & 0xFFFF);
1808 if (st->codecpar->codec_type != AVMEDIA_TYPE_VIDEO && id > 0) {
1809 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
1810 } else if (st->codecpar->codec_type != AVMEDIA_TYPE_AUDIO &&
1811 /* skip old ASF MPEG-4 tag */
1812 format && format != MKTAG('m','p','4','s')) {
1813 id = ff_codec_get_id(ff_codec_movvideo_tags, format);
1815 id = ff_codec_get_id(ff_codec_bmp_tags, format);
1817 st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
1818 else if (st->codecpar->codec_type == AVMEDIA_TYPE_DATA ||
1819 (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE &&
1820 st->codecpar->codec_id == AV_CODEC_ID_NONE)) {
1821 id = ff_codec_get_id(ff_codec_movsubtitle_tags, format);
1823 st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
1827 st->codecpar->codec_tag = format;
1832 static void mov_parse_stsd_video(MOVContext *c, AVIOContext *pb,
1833 AVStream *st, MOVStreamContext *sc)
1835 uint8_t codec_name[32] = { 0 };
1839 /* The first 16 bytes of the video sample description are already
1840 * read in ff_mov_read_stsd_entries() */
1841 stsd_start = avio_tell(pb) - 16;
1843 avio_rb16(pb); /* version */
1844 avio_rb16(pb); /* revision level */
1845 avio_rb32(pb); /* vendor */
1846 avio_rb32(pb); /* temporal quality */
1847 avio_rb32(pb); /* spatial quality */
1849 st->codecpar->width = avio_rb16(pb); /* width */
1850 st->codecpar->height = avio_rb16(pb); /* height */
1852 avio_rb32(pb); /* horiz resolution */
1853 avio_rb32(pb); /* vert resolution */
1854 avio_rb32(pb); /* data size, always 0 */
1855 avio_rb16(pb); /* frames per samples */
1857 len = avio_r8(pb); /* codec name, pascal string */
1860 mov_read_mac_string(c, pb, len, codec_name, sizeof(codec_name));
1862 avio_skip(pb, 31 - len);
1865 av_dict_set(&st->metadata, "encoder", codec_name, 0);
1867 /* codec_tag YV12 triggers an UV swap in rawdec.c */
1868 if (!memcmp(codec_name, "Planar Y'CbCr 8-bit 4:2:0", 25)) {
1869 st->codecpar->codec_tag = MKTAG('I', '4', '2', '0');
1870 st->codecpar->width &= ~1;
1871 st->codecpar->height &= ~1;
1873 /* Flash Media Server uses tag H.263 with Sorenson Spark */
1874 if (st->codecpar->codec_tag == MKTAG('H','2','6','3') &&
1875 !memcmp(codec_name, "Sorenson H263", 13))
1876 st->codecpar->codec_id = AV_CODEC_ID_FLV1;
1878 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* depth */
1880 avio_seek(pb, stsd_start, SEEK_SET);
1882 if (ff_get_qtpalette(st->codecpar->codec_id, pb, sc->palette)) {
1883 st->codecpar->bits_per_coded_sample &= 0x1F;
1884 sc->has_palette = 1;
1888 static void mov_parse_stsd_audio(MOVContext *c, AVIOContext *pb,
1889 AVStream *st, MOVStreamContext *sc)
1891 int bits_per_sample, flags;
1892 uint16_t version = avio_rb16(pb);
1893 AVDictionaryEntry *compatible_brands = av_dict_get(c->fc->metadata, "compatible_brands", NULL, AV_DICT_MATCH_CASE);
1895 avio_rb16(pb); /* revision level */
1896 avio_rb32(pb); /* vendor */
1898 st->codecpar->channels = avio_rb16(pb); /* channel count */
1899 st->codecpar->bits_per_coded_sample = avio_rb16(pb); /* sample size */
1900 av_log(c->fc, AV_LOG_TRACE, "audio channels %d\n", st->codecpar->channels);
1902 sc->audio_cid = avio_rb16(pb);
1903 avio_rb16(pb); /* packet size = 0 */
1905 st->codecpar->sample_rate = ((avio_rb32(pb) >> 16));
1907 // Read QT version 1 fields. In version 0 these do not exist.
1908 av_log(c->fc, AV_LOG_TRACE, "version =%d, isom =%d\n", version, c->isom);
1910 (compatible_brands && strstr(compatible_brands->value, "qt "))) {
1913 sc->samples_per_frame = avio_rb32(pb);
1914 avio_rb32(pb); /* bytes per packet */
1915 sc->bytes_per_frame = avio_rb32(pb);
1916 avio_rb32(pb); /* bytes per sample */
1917 } else if (version == 2) {
1918 avio_rb32(pb); /* sizeof struct only */
1919 st->codecpar->sample_rate = av_int2double(avio_rb64(pb));
1920 st->codecpar->channels = avio_rb32(pb);
1921 avio_rb32(pb); /* always 0x7F000000 */
1922 st->codecpar->bits_per_coded_sample = avio_rb32(pb);
1924 flags = avio_rb32(pb); /* lpcm format specific flag */
1925 sc->bytes_per_frame = avio_rb32(pb);
1926 sc->samples_per_frame = avio_rb32(pb);
1927 if (st->codecpar->codec_tag == MKTAG('l','p','c','m'))
1928 st->codecpar->codec_id =
1929 ff_mov_get_lpcm_codec_id(st->codecpar->bits_per_coded_sample,
1932 if (version == 0 || (version == 1 && sc->audio_cid != -2)) {
1933 /* can't correctly handle variable sized packet as audio unit */
1934 switch (st->codecpar->codec_id) {
1935 case AV_CODEC_ID_MP2:
1936 case AV_CODEC_ID_MP3:
1937 st->need_parsing = AVSTREAM_PARSE_FULL;
1943 if (sc->format == 0) {
1944 if (st->codecpar->bits_per_coded_sample == 8)
1945 st->codecpar->codec_id = mov_codec_id(st, MKTAG('r','a','w',' '));
1946 else if (st->codecpar->bits_per_coded_sample == 16)
1947 st->codecpar->codec_id = mov_codec_id(st, MKTAG('t','w','o','s'));
1950 switch (st->codecpar->codec_id) {
1951 case AV_CODEC_ID_PCM_S8:
1952 case AV_CODEC_ID_PCM_U8:
1953 if (st->codecpar->bits_per_coded_sample == 16)
1954 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16BE;
1956 case AV_CODEC_ID_PCM_S16LE:
1957 case AV_CODEC_ID_PCM_S16BE:
1958 if (st->codecpar->bits_per_coded_sample == 8)
1959 st->codecpar->codec_id = AV_CODEC_ID_PCM_S8;
1960 else if (st->codecpar->bits_per_coded_sample == 24)
1961 st->codecpar->codec_id =
1962 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
1963 AV_CODEC_ID_PCM_S24BE : AV_CODEC_ID_PCM_S24LE;
1964 else if (st->codecpar->bits_per_coded_sample == 32)
1965 st->codecpar->codec_id =
1966 st->codecpar->codec_id == AV_CODEC_ID_PCM_S16BE ?
1967 AV_CODEC_ID_PCM_S32BE : AV_CODEC_ID_PCM_S32LE;
1969 /* set values for old format before stsd version 1 appeared */
1970 case AV_CODEC_ID_MACE3:
1971 sc->samples_per_frame = 6;
1972 sc->bytes_per_frame = 2 * st->codecpar->channels;
1974 case AV_CODEC_ID_MACE6:
1975 sc->samples_per_frame = 6;
1976 sc->bytes_per_frame = 1 * st->codecpar->channels;
1978 case AV_CODEC_ID_ADPCM_IMA_QT:
1979 sc->samples_per_frame = 64;
1980 sc->bytes_per_frame = 34 * st->codecpar->channels;
1982 case AV_CODEC_ID_GSM:
1983 sc->samples_per_frame = 160;
1984 sc->bytes_per_frame = 33;
1990 bits_per_sample = av_get_bits_per_sample(st->codecpar->codec_id);
1991 if (bits_per_sample) {
1992 st->codecpar->bits_per_coded_sample = bits_per_sample;
1993 sc->sample_size = (bits_per_sample >> 3) * st->codecpar->channels;
1997 static void mov_parse_stsd_subtitle(MOVContext *c, AVIOContext *pb,
1998 AVStream *st, MOVStreamContext *sc,
2001 // ttxt stsd contains display flags, justification, background
2002 // color, fonts, and default styles, so fake an atom to read it
2003 MOVAtom fake_atom = { .size = size };
2004 // mp4s contains a regular esds atom
2005 if (st->codecpar->codec_tag != AV_RL32("mp4s"))
2006 mov_read_glbl(c, pb, fake_atom);
2007 st->codecpar->width = sc->width;
2008 st->codecpar->height = sc->height;
2011 static uint32_t yuv_to_rgba(uint32_t ycbcr)
2016 y = (ycbcr >> 16) & 0xFF;
2017 cr = (ycbcr >> 8) & 0xFF;
2020 b = av_clip_uint8((1164 * (y - 16) + 2018 * (cb - 128)) / 1000);
2021 g = av_clip_uint8((1164 * (y - 16) - 813 * (cr - 128) - 391 * (cb - 128)) / 1000);
2022 r = av_clip_uint8((1164 * (y - 16) + 1596 * (cr - 128) ) / 1000);
2024 return (r << 16) | (g << 8) | b;
2027 static int mov_rewrite_dvd_sub_extradata(AVStream *st)
2029 char buf[256] = {0};
2030 uint8_t *src = st->codecpar->extradata;
2033 if (st->codecpar->extradata_size != 64)
2036 if (st->codecpar->width > 0 && st->codecpar->height > 0)
2037 snprintf(buf, sizeof(buf), "size: %dx%d\n",
2038 st->codecpar->width, st->codecpar->height);
2039 av_strlcat(buf, "palette: ", sizeof(buf));
2041 for (i = 0; i < 16; i++) {
2042 uint32_t yuv = AV_RB32(src + i * 4);
2043 uint32_t rgba = yuv_to_rgba(yuv);
2045 av_strlcatf(buf, sizeof(buf), "%06"PRIx32"%s", rgba, i != 15 ? ", " : "");
2048 if (av_strlcat(buf, "\n", sizeof(buf)) >= sizeof(buf))
2051 av_freep(&st->codecpar->extradata);
2052 st->codecpar->extradata_size = 0;
2053 st->codecpar->extradata = av_mallocz(strlen(buf) + AV_INPUT_BUFFER_PADDING_SIZE);
2054 if (!st->codecpar->extradata)
2055 return AVERROR(ENOMEM);
2056 st->codecpar->extradata_size = strlen(buf);
2057 memcpy(st->codecpar->extradata, buf, st->codecpar->extradata_size);
2062 static int mov_parse_stsd_data(MOVContext *c, AVIOContext *pb,
2063 AVStream *st, MOVStreamContext *sc,
2068 if (st->codecpar->codec_tag == MKTAG('t','m','c','d')) {
2069 if ((int)size != size)
2070 return AVERROR(ENOMEM);
2072 ret = ff_get_extradata(c->fc, st->codecpar, pb, size);
2076 MOVStreamContext *tmcd_ctx = st->priv_data;
2078 val = AV_RB32(st->codecpar->extradata + 4);
2079 tmcd_ctx->tmcd_flags = val;
2080 st->avg_frame_rate.num = st->codecpar->extradata[16]; /* number of frame */
2081 st->avg_frame_rate.den = 1;
2082 #if FF_API_LAVF_AVCTX
2083 FF_DISABLE_DEPRECATION_WARNINGS
2084 st->codec->time_base = av_inv_q(st->avg_frame_rate);
2085 FF_ENABLE_DEPRECATION_WARNINGS
2087 /* adjust for per frame dur in counter mode */
2088 if (tmcd_ctx->tmcd_flags & 0x0008) {
2089 int timescale = AV_RB32(st->codecpar->extradata + 8);
2090 int framedur = AV_RB32(st->codecpar->extradata + 12);
2091 st->avg_frame_rate.num *= timescale;
2092 st->avg_frame_rate.den *= framedur;
2093 #if FF_API_LAVF_AVCTX
2094 FF_DISABLE_DEPRECATION_WARNINGS
2095 st->codec->time_base.den *= timescale;
2096 st->codec->time_base.num *= framedur;
2097 FF_ENABLE_DEPRECATION_WARNINGS
2101 uint32_t len = AV_RB32(st->codecpar->extradata + 18); /* name atom length */
2102 uint32_t format = AV_RB32(st->codecpar->extradata + 22);
2103 if (format == AV_RB32("name") && (int64_t)size >= (int64_t)len + 18) {
2104 uint16_t str_size = AV_RB16(st->codecpar->extradata + 26); /* string length */
2105 if (str_size > 0 && size >= (int)str_size + 26) {
2106 char *reel_name = av_malloc(str_size + 1);
2108 return AVERROR(ENOMEM);
2109 memcpy(reel_name, st->codecpar->extradata + 30, str_size);
2110 reel_name[str_size] = 0; /* Add null terminator */
2111 /* don't add reel_name if emtpy string */
2112 if (*reel_name == 0) {
2115 av_dict_set(&st->metadata, "reel_name", reel_name, AV_DICT_DONT_STRDUP_VAL);
2122 /* other codec type, just skip (rtp, mp4s ...) */
2123 avio_skip(pb, size);
2128 static int mov_finalize_stsd_codec(MOVContext *c, AVIOContext *pb,
2129 AVStream *st, MOVStreamContext *sc)
2131 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
2132 !st->codecpar->sample_rate && sc->time_scale > 1)
2133 st->codecpar->sample_rate = sc->time_scale;
2135 /* special codec parameters handling */
2136 switch (st->codecpar->codec_id) {
2137 #if CONFIG_DV_DEMUXER
2138 case AV_CODEC_ID_DVAUDIO:
2139 c->dv_fctx = avformat_alloc_context();
2141 av_log(c->fc, AV_LOG_ERROR, "dv demux context alloc error\n");
2142 return AVERROR(ENOMEM);
2144 c->dv_demux = avpriv_dv_init_demux(c->dv_fctx);
2146 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
2147 return AVERROR(ENOMEM);
2149 sc->dv_audio_container = 1;
2150 st->codecpar->codec_id = AV_CODEC_ID_PCM_S16LE;
2153 /* no ifdef since parameters are always those */
2154 case AV_CODEC_ID_QCELP:
2155 st->codecpar->channels = 1;
2156 // force sample rate for qcelp when not stored in mov
2157 if (st->codecpar->codec_tag != MKTAG('Q','c','l','p'))
2158 st->codecpar->sample_rate = 8000;
2159 // FIXME: Why is the following needed for some files?
2160 sc->samples_per_frame = 160;
2161 if (!sc->bytes_per_frame)
2162 sc->bytes_per_frame = 35;
2164 case AV_CODEC_ID_AMR_NB:
2165 st->codecpar->channels = 1;
2166 /* force sample rate for amr, stsd in 3gp does not store sample rate */
2167 st->codecpar->sample_rate = 8000;
2169 case AV_CODEC_ID_AMR_WB:
2170 st->codecpar->channels = 1;
2171 st->codecpar->sample_rate = 16000;
2173 case AV_CODEC_ID_MP2:
2174 case AV_CODEC_ID_MP3:
2175 /* force type after stsd for m1a hdlr */
2176 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
2178 case AV_CODEC_ID_GSM:
2179 case AV_CODEC_ID_ADPCM_MS:
2180 case AV_CODEC_ID_ADPCM_IMA_WAV:
2181 case AV_CODEC_ID_ILBC:
2182 case AV_CODEC_ID_MACE3:
2183 case AV_CODEC_ID_MACE6:
2184 case AV_CODEC_ID_QDM2:
2185 st->codecpar->block_align = sc->bytes_per_frame;
2187 case AV_CODEC_ID_ALAC:
2188 if (st->codecpar->extradata_size == 36) {
2189 st->codecpar->channels = AV_RB8 (st->codecpar->extradata + 21);
2190 st->codecpar->sample_rate = AV_RB32(st->codecpar->extradata + 32);
2193 case AV_CODEC_ID_AC3:
2194 case AV_CODEC_ID_EAC3:
2195 case AV_CODEC_ID_MPEG1VIDEO:
2196 case AV_CODEC_ID_VC1:
2197 case AV_CODEC_ID_VP9:
2198 st->need_parsing = AVSTREAM_PARSE_FULL;
2206 static int mov_skip_multiple_stsd(MOVContext *c, AVIOContext *pb,
2207 int codec_tag, int format,
2210 int video_codec_id = ff_codec_get_id(ff_codec_movvideo_tags, format);
2213 (codec_tag != format &&
2214 // prores is allowed to have differing data format and codec tag
2215 codec_tag != AV_RL32("apcn") && codec_tag != AV_RL32("apch") &&
2217 codec_tag != AV_RL32("dvpp") && codec_tag != AV_RL32("dvcp") &&
2218 (c->fc->video_codec_id ? video_codec_id != c->fc->video_codec_id
2219 : codec_tag != MKTAG('j','p','e','g')))) {
2220 /* Multiple fourcc, we skip JPEG. This is not correct, we should
2221 * export it as a separate AVStream but this needs a few changes
2222 * in the MOV demuxer, patch welcome. */
2224 av_log(c->fc, AV_LOG_WARNING, "multiple fourcc not supported\n");
2225 avio_skip(pb, size);
2228 if ( codec_tag == AV_RL32("hvc1") ||
2229 codec_tag == AV_RL32("hev1")
2231 av_log(c->fc, AV_LOG_WARNING, "Concatenated H.264 or H.265 might not play correctly.\n");
2236 int ff_mov_read_stsd_entries(MOVContext *c, AVIOContext *pb, int entries)
2239 MOVStreamContext *sc;
2240 int pseudo_stream_id;
2242 if (c->fc->nb_streams < 1)
2244 st = c->fc->streams[c->fc->nb_streams-1];
2247 for (pseudo_stream_id = 0;
2248 pseudo_stream_id < entries && !pb->eof_reached;
2249 pseudo_stream_id++) {
2250 //Parsing Sample description table
2252 int ret, dref_id = 1;
2253 MOVAtom a = { AV_RL32("stsd") };
2254 int64_t start_pos = avio_tell(pb);
2255 int64_t size = avio_rb32(pb); /* size */
2256 uint32_t format = avio_rl32(pb); /* data format */
2259 avio_rb32(pb); /* reserved */
2260 avio_rb16(pb); /* reserved */
2261 dref_id = avio_rb16(pb);
2262 } else if (size <= 7) {
2263 av_log(c->fc, AV_LOG_ERROR,
2264 "invalid size %"PRId64" in stsd\n", size);
2265 return AVERROR_INVALIDDATA;
2268 if (mov_skip_multiple_stsd(c, pb, st->codecpar->codec_tag, format,
2269 size - (avio_tell(pb) - start_pos)))
2272 sc->pseudo_stream_id = st->codecpar->codec_tag ? -1 : pseudo_stream_id;
2273 sc->dref_id= dref_id;
2274 sc->format = format;
2276 id = mov_codec_id(st, format);
2278 av_log(c->fc, AV_LOG_TRACE,
2279 "size=%"PRId64" 4CC= %c%c%c%c/0x%08x codec_type=%d\n", size,
2280 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff,
2281 (format >> 24) & 0xff, format, st->codecpar->codec_type);
2283 if (st->codecpar->codec_type==AVMEDIA_TYPE_VIDEO) {
2284 st->codecpar->codec_id = id;
2285 mov_parse_stsd_video(c, pb, st, sc);
2286 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_AUDIO) {
2287 st->codecpar->codec_id = id;
2288 mov_parse_stsd_audio(c, pb, st, sc);
2289 if (st->codecpar->sample_rate < 0) {
2290 av_log(c->fc, AV_LOG_ERROR, "Invalid sample rate %d\n", st->codecpar->sample_rate);
2291 return AVERROR_INVALIDDATA;
2293 } else if (st->codecpar->codec_type==AVMEDIA_TYPE_SUBTITLE){
2294 st->codecpar->codec_id = id;
2295 mov_parse_stsd_subtitle(c, pb, st, sc,
2296 size - (avio_tell(pb) - start_pos));
2298 ret = mov_parse_stsd_data(c, pb, st, sc,
2299 size - (avio_tell(pb) - start_pos));
2303 /* this will read extra atoms at the end (wave, alac, damr, avcC, hvcC, SMI ...) */
2304 a.size = size - (avio_tell(pb) - start_pos);
2306 if ((ret = mov_read_default(c, pb, a)) < 0)
2308 } else if (a.size > 0)
2309 avio_skip(pb, a.size);
2311 if (sc->extradata) {
2312 int extra_size = st->codecpar->extradata_size;
2314 /* Move the current stream extradata to the stream context one. */
2315 sc->extradata_size[pseudo_stream_id] = extra_size;
2316 sc->extradata[pseudo_stream_id] = av_malloc(extra_size + AV_INPUT_BUFFER_PADDING_SIZE);
2317 if (!sc->extradata[pseudo_stream_id])
2318 return AVERROR(ENOMEM);
2319 memcpy(sc->extradata[pseudo_stream_id], st->codecpar->extradata, extra_size);
2320 av_freep(&st->codecpar->extradata);
2321 st->codecpar->extradata_size = 0;
2325 if (pb->eof_reached)
2331 static int mov_read_stsd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2334 MOVStreamContext *sc;
2337 if (c->fc->nb_streams < 1)
2339 st = c->fc->streams[c->fc->nb_streams - 1];
2342 avio_r8(pb); /* version */
2343 avio_rb24(pb); /* flags */
2344 entries = avio_rb32(pb);
2347 av_log(c->fc, AV_LOG_ERROR, "invalid STSD entries %d\n", entries);
2348 return AVERROR_INVALIDDATA;
2351 if (sc->extradata) {
2352 av_log(c->fc, AV_LOG_ERROR, "Duplicate STSD\n");
2353 return AVERROR_INVALIDDATA;
2355 /* Prepare space for hosting multiple extradata. */
2356 sc->extradata = av_mallocz_array(entries, sizeof(*sc->extradata));
2357 sc->extradata_size = av_mallocz_array(entries, sizeof(*sc->extradata_size));
2358 if (!sc->extradata_size || !sc->extradata) {
2359 ret = AVERROR(ENOMEM);
2363 ret = ff_mov_read_stsd_entries(c, pb, entries);
2367 sc->stsd_count = entries;
2369 /* Restore back the primary extradata. */
2370 av_freep(&st->codecpar->extradata);
2371 st->codecpar->extradata_size = sc->extradata_size[0];
2372 if (sc->extradata_size[0]) {
2373 st->codecpar->extradata = av_mallocz(sc->extradata_size[0] + AV_INPUT_BUFFER_PADDING_SIZE);
2374 if (!st->codecpar->extradata)
2375 return AVERROR(ENOMEM);
2376 memcpy(st->codecpar->extradata, sc->extradata[0], sc->extradata_size[0]);
2379 return mov_finalize_stsd_codec(c, pb, st, sc);
2381 av_freep(&sc->extradata);
2382 av_freep(&sc->extradata_size);
2386 static int mov_read_stsc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2389 MOVStreamContext *sc;
2390 unsigned int i, entries;
2392 if (c->fc->nb_streams < 1)
2394 st = c->fc->streams[c->fc->nb_streams-1];
2397 avio_r8(pb); /* version */
2398 avio_rb24(pb); /* flags */
2400 entries = avio_rb32(pb);
2402 av_log(c->fc, AV_LOG_TRACE, "track[%i].stsc.entries = %i\n", c->fc->nb_streams-1, entries);
2407 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSC atom\n");
2408 av_free(sc->stsc_data);
2410 sc->stsc_data = av_malloc_array(entries, sizeof(*sc->stsc_data));
2412 return AVERROR(ENOMEM);
2414 for (i = 0; i < entries && !pb->eof_reached; i++) {
2415 sc->stsc_data[i].first = avio_rb32(pb);
2416 sc->stsc_data[i].count = avio_rb32(pb);
2417 sc->stsc_data[i].id = avio_rb32(pb);
2422 if (pb->eof_reached)
2428 /* Compute the samples value for the stsc entry at the given index. */
2429 static inline int mov_get_stsc_samples(MOVStreamContext *sc, int index)
2433 if (index < sc->stsc_count - 1)
2434 chunk_count = sc->stsc_data[index + 1].first - sc->stsc_data[index].first;
2436 chunk_count = sc->chunk_count - (sc->stsc_data[index].first - 1);
2438 return sc->stsc_data[index].count * chunk_count;
2441 static int mov_read_stps(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2444 MOVStreamContext *sc;
2445 unsigned i, entries;
2447 if (c->fc->nb_streams < 1)
2449 st = c->fc->streams[c->fc->nb_streams-1];
2452 avio_rb32(pb); // version + flags
2454 entries = avio_rb32(pb);
2456 av_log(c->fc, AV_LOG_WARNING, "Duplicated STPS atom\n");
2457 av_free(sc->stps_data);
2459 sc->stps_data = av_malloc_array(entries, sizeof(*sc->stps_data));
2461 return AVERROR(ENOMEM);
2463 for (i = 0; i < entries && !pb->eof_reached; i++) {
2464 sc->stps_data[i] = avio_rb32(pb);
2465 //av_log(c->fc, AV_LOG_TRACE, "stps %d\n", sc->stps_data[i]);
2470 if (pb->eof_reached)
2476 static int mov_read_stss(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2479 MOVStreamContext *sc;
2480 unsigned int i, entries;
2482 if (c->fc->nb_streams < 1)
2484 st = c->fc->streams[c->fc->nb_streams-1];
2487 avio_r8(pb); /* version */
2488 avio_rb24(pb); /* flags */
2490 entries = avio_rb32(pb);
2492 av_log(c->fc, AV_LOG_TRACE, "keyframe_count = %d\n", entries);
2496 sc->keyframe_absent = 1;
2497 if (!st->need_parsing && st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO)
2498 st->need_parsing = AVSTREAM_PARSE_HEADERS;
2502 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSS atom\n");
2503 if (entries >= UINT_MAX / sizeof(int))
2504 return AVERROR_INVALIDDATA;
2505 av_freep(&sc->keyframes);
2506 sc->keyframe_count = 0;
2507 sc->keyframes = av_malloc_array(entries, sizeof(*sc->keyframes));
2509 return AVERROR(ENOMEM);
2511 for (i = 0; i < entries && !pb->eof_reached; i++) {
2512 sc->keyframes[i] = avio_rb32(pb);
2513 //av_log(c->fc, AV_LOG_TRACE, "keyframes[]=%d\n", sc->keyframes[i]);
2516 sc->keyframe_count = i;
2518 if (pb->eof_reached)
2524 static int mov_read_stsz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2527 MOVStreamContext *sc;
2528 unsigned int i, entries, sample_size, field_size, num_bytes;
2533 if (c->fc->nb_streams < 1)
2535 st = c->fc->streams[c->fc->nb_streams-1];
2538 avio_r8(pb); /* version */
2539 avio_rb24(pb); /* flags */
2541 if (atom.type == MKTAG('s','t','s','z')) {
2542 sample_size = avio_rb32(pb);
2543 if (!sc->sample_size) /* do not overwrite value computed in stsd */
2544 sc->sample_size = sample_size;
2545 sc->stsz_sample_size = sample_size;
2549 avio_rb24(pb); /* reserved */
2550 field_size = avio_r8(pb);
2552 entries = avio_rb32(pb);
2554 av_log(c->fc, AV_LOG_TRACE, "sample_size = %d sample_count = %d\n", sc->sample_size, entries);
2556 sc->sample_count = entries;
2560 if (field_size != 4 && field_size != 8 && field_size != 16 && field_size != 32) {
2561 av_log(c->fc, AV_LOG_ERROR, "Invalid sample field size %d\n", field_size);
2562 return AVERROR_INVALIDDATA;
2567 if (entries >= (UINT_MAX - 4) / field_size)
2568 return AVERROR_INVALIDDATA;
2569 if (sc->sample_sizes)
2570 av_log(c->fc, AV_LOG_WARNING, "Duplicated STSZ atom\n");
2571 av_free(sc->sample_sizes);
2572 sc->sample_count = 0;
2573 sc->sample_sizes = av_malloc_array(entries, sizeof(*sc->sample_sizes));
2574 if (!sc->sample_sizes)
2575 return AVERROR(ENOMEM);
2577 num_bytes = (entries*field_size+4)>>3;
2579 buf = av_malloc(num_bytes+AV_INPUT_BUFFER_PADDING_SIZE);
2581 av_freep(&sc->sample_sizes);
2582 return AVERROR(ENOMEM);
2585 ret = ffio_read_size(pb, buf, num_bytes);
2587 av_freep(&sc->sample_sizes);
2592 init_get_bits(&gb, buf, 8*num_bytes);
2594 for (i = 0; i < entries && !pb->eof_reached; i++) {
2595 sc->sample_sizes[i] = get_bits_long(&gb, field_size);
2596 sc->data_size += sc->sample_sizes[i];
2599 sc->sample_count = i;
2603 if (pb->eof_reached)
2609 static int mov_read_stts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2612 MOVStreamContext *sc;
2613 unsigned int i, entries;
2615 int64_t total_sample_count=0;
2617 if (c->fc->nb_streams < 1)
2619 st = c->fc->streams[c->fc->nb_streams-1];
2622 avio_r8(pb); /* version */
2623 avio_rb24(pb); /* flags */
2624 entries = avio_rb32(pb);
2626 av_log(c->fc, AV_LOG_TRACE, "track[%i].stts.entries = %i\n",
2627 c->fc->nb_streams-1, entries);
2630 av_log(c->fc, AV_LOG_WARNING, "Duplicated STTS atom\n");
2631 av_free(sc->stts_data);
2633 sc->stts_data = av_malloc_array(entries, sizeof(*sc->stts_data));
2635 return AVERROR(ENOMEM);
2637 for (i = 0; i < entries && !pb->eof_reached; i++) {
2638 int sample_duration;
2641 sample_count=avio_rb32(pb);
2642 sample_duration = avio_rb32(pb);
2644 if (sample_count < 0) {
2645 av_log(c->fc, AV_LOG_ERROR, "Invalid sample_count=%d\n", sample_count);
2646 return AVERROR_INVALIDDATA;
2648 sc->stts_data[i].count= sample_count;
2649 sc->stts_data[i].duration= sample_duration;
2651 av_log(c->fc, AV_LOG_TRACE, "sample_count=%d, sample_duration=%d\n",
2652 sample_count, sample_duration);
2656 && sample_count == 1
2657 && total_sample_count > 100
2658 && sample_duration/10 > duration / total_sample_count)
2659 sample_duration = duration / total_sample_count;
2660 duration+=(int64_t)sample_duration*sample_count;
2661 total_sample_count+=sample_count;
2666 sc->duration_for_fps += duration;
2667 sc->nb_frames_for_fps += total_sample_count;
2669 if (pb->eof_reached)
2672 st->nb_frames= total_sample_count;
2674 st->duration= duration;
2675 sc->track_end = duration;
2679 static void mov_update_dts_shift(MOVStreamContext *sc, int duration)
2682 if (duration == INT_MIN) {
2683 av_log(NULL, AV_LOG_WARNING, "mov_update_dts_shift(): dts_shift set to %d\n", INT_MAX);
2686 sc->dts_shift = FFMAX(sc->dts_shift, -duration);
2690 static int mov_read_ctts(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2693 MOVStreamContext *sc;
2694 unsigned int i, entries, ctts_count = 0;
2696 if (c->fc->nb_streams < 1)
2698 st = c->fc->streams[c->fc->nb_streams-1];
2701 avio_r8(pb); /* version */
2702 avio_rb24(pb); /* flags */
2703 entries = avio_rb32(pb);
2705 av_log(c->fc, AV_LOG_TRACE, "track[%i].ctts.entries = %i\n", c->fc->nb_streams-1, entries);
2709 if (entries >= UINT_MAX / sizeof(*sc->ctts_data))
2710 return AVERROR_INVALIDDATA;
2711 av_freep(&sc->ctts_data);
2712 sc->ctts_data = av_realloc(NULL, entries * sizeof(*sc->ctts_data));
2714 return AVERROR(ENOMEM);
2716 for (i = 0; i < entries && !pb->eof_reached; i++) {
2717 int count =avio_rb32(pb);
2718 int duration =avio_rb32(pb);
2721 av_log(c->fc, AV_LOG_TRACE,
2722 "ignoring CTTS entry with count=%d duration=%d\n",
2727 sc->ctts_data[ctts_count].count = count;
2728 sc->ctts_data[ctts_count].duration = duration;
2731 av_log(c->fc, AV_LOG_TRACE, "count=%d, duration=%d\n",
2734 if (FFNABS(duration) < -(1<<28) && i+2<entries) {
2735 av_log(c->fc, AV_LOG_WARNING, "CTTS invalid\n");
2736 av_freep(&sc->ctts_data);
2742 mov_update_dts_shift(sc, duration);
2745 sc->ctts_count = ctts_count;
2747 if (pb->eof_reached)
2750 av_log(c->fc, AV_LOG_TRACE, "dts shift %d\n", sc->dts_shift);
2755 static int mov_read_sbgp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
2758 MOVStreamContext *sc;
2759 unsigned int i, entries;
2761 uint32_t grouping_type;
2763 if (c->fc->nb_streams < 1)
2765 st = c->fc->streams[c->fc->nb_streams-1];
2768 version = avio_r8(pb); /* version */
2769 avio_rb24(pb); /* flags */
2770 grouping_type = avio_rl32(pb);
2771 if (grouping_type != MKTAG( 'r','a','p',' '))
2772 return 0; /* only support 'rap ' grouping */
2774 avio_rb32(pb); /* grouping_type_parameter */
2776 entries = avio_rb32(pb);
2780 av_log(c->fc, AV_LOG_WARNING, "Duplicated SBGP atom\n");
2781 av_free(sc->rap_group);
2782 sc->rap_group_count = 0;
2783 sc->rap_group = av_malloc_array(entries, sizeof(*sc->rap_group));
2785 return AVERROR(ENOMEM);
2787 for (i = 0; i < entries && !pb->eof_reached; i++) {
2788 sc->rap_group[i].count = avio_rb32(pb); /* sample_count */
2789 sc->rap_group[i].index = avio_rb32(pb); /* group_description_index */
2792 sc->rap_group_count = i;
2794 return pb->eof_reached ? AVERROR_EOF : 0;
2798 * Get ith edit list entry (media time, duration).
2800 static int get_edit_list_entry(MOVContext *mov,
2801 const MOVStreamContext *msc,
2802 unsigned int edit_list_index,
2803 int64_t *edit_list_media_time,
2804 int64_t *edit_list_duration,
2805 int64_t global_timescale)
2807 if (edit_list_index == msc->elst_count) {
2810 *edit_list_media_time = msc->elst_data[edit_list_index].time;
2811 *edit_list_duration = msc->elst_data[edit_list_index].duration;
2813 /* duration is in global timescale units;convert to msc timescale */
2814 if (global_timescale == 0) {
2815 avpriv_request_sample(mov->fc, "Support for mvhd.timescale = 0 with editlists");
2818 *edit_list_duration = av_rescale(*edit_list_duration, msc->time_scale,
2824 * Find the closest previous frame to the timestamp, in e_old index
2825 * entries. Searching for just any frame / just key frames can be controlled by
2826 * last argument 'flag'.
2827 * Returns the index of the entry in st->index_entries if successful,
2830 static int64_t find_prev_closest_index(AVStream *st,
2831 AVIndexEntry *e_old,
2836 AVIndexEntry *e_keep = st->index_entries;
2837 int nb_keep = st->nb_index_entries;
2840 st->index_entries = e_old;
2841 st->nb_index_entries = nb_old;
2842 found = av_index_search_timestamp(st, timestamp, flag | AVSEEK_FLAG_BACKWARD);
2844 /* restore AVStream state*/
2845 st->index_entries = e_keep;
2846 st->nb_index_entries = nb_keep;
2851 * Add index entry with the given values, to the end of st->index_entries.
2852 * Returns the new size st->index_entries if successful, else returns -1.
2854 * This function is similar to ff_add_index_entry in libavformat/utils.c
2855 * except that here we are always unconditionally adding an index entry to
2856 * the end, instead of searching the entries list and skipping the add if
2857 * there is an existing entry with the same timestamp.
2858 * This is needed because the mov_fix_index calls this func with the same
2859 * unincremented timestamp for successive discarded frames.
2861 static int64_t add_index_entry(AVStream *st, int64_t pos, int64_t timestamp,
2862 int size, int distance, int flags)
2864 AVIndexEntry *entries, *ie;
2866 const size_t min_size_needed = (st->nb_index_entries + 1) * sizeof(AVIndexEntry);
2868 // Double the allocation each time, to lower memory fragmentation.
2869 // Another difference from ff_add_index_entry function.
2870 const size_t requested_size =
2871 min_size_needed > st->index_entries_allocated_size ?
2872 FFMAX(min_size_needed, 2 * st->index_entries_allocated_size) :
2875 if((unsigned)st->nb_index_entries + 1 >= UINT_MAX / sizeof(AVIndexEntry))
2878 entries = av_fast_realloc(st->index_entries,
2879 &st->index_entries_allocated_size,
2884 st->index_entries= entries;
2886 index= st->nb_index_entries++;
2887 ie= &entries[index];
2890 ie->timestamp = timestamp;
2891 ie->min_distance= distance;
2898 * Rewrite timestamps of index entries in the range [end_index - frame_duration_buffer_size, end_index)
2899 * by subtracting end_ts successively by the amounts given in frame_duration_buffer.
2901 static void fix_index_entry_timestamps(AVStream* st, int end_index, int64_t end_ts,
2902 int64_t* frame_duration_buffer,
2903 int frame_duration_buffer_size) {
2905 av_assert0(end_index >= 0 && end_index <= st->nb_index_entries);
2906 for (i = 0; i < frame_duration_buffer_size; i++) {
2907 end_ts -= frame_duration_buffer[frame_duration_buffer_size - 1 - i];
2908 st->index_entries[end_index - 1 - i].timestamp = end_ts;
2913 * Append a new ctts entry to ctts_data.
2914 * Returns the new ctts_count if successful, else returns -1.
2916 static int64_t add_ctts_entry(MOVStts** ctts_data, unsigned int* ctts_count, unsigned int* allocated_size,
2917 int count, int duration)
2919 MOVStts *ctts_buf_new;
2920 const size_t min_size_needed = (*ctts_count + 1) * sizeof(MOVStts);
2921 const size_t requested_size =
2922 min_size_needed > *allocated_size ?
2923 FFMAX(min_size_needed, 2 * (*allocated_size)) :
2926 if((unsigned)(*ctts_count) + 1 >= UINT_MAX / sizeof(MOVStts))
2929 ctts_buf_new = av_fast_realloc(*ctts_data, allocated_size, requested_size);
2934 *ctts_data = ctts_buf_new;
2936 ctts_buf_new[*ctts_count].count = count;
2937 ctts_buf_new[*ctts_count].duration = duration;
2939 *ctts_count = (*ctts_count) + 1;
2944 * Fix st->index_entries, so that it contains only the entries (and the entries
2945 * which are needed to decode them) that fall in the edit list time ranges.
2946 * Also fixes the timestamps of the index entries to match the timeline
2947 * specified the edit lists.
2949 static void mov_fix_index(MOVContext *mov, AVStream *st)
2951 MOVStreamContext *msc = st->priv_data;
2952 AVIndexEntry *e_old = st->index_entries;
2953 int nb_old = st->nb_index_entries;
2954 const AVIndexEntry *e_old_end = e_old + nb_old;
2955 const AVIndexEntry *current = NULL;
2956 MOVStts *ctts_data_old = msc->ctts_data;
2957 int64_t ctts_index_old = 0;
2958 int64_t ctts_sample_old = 0;
2959 int64_t ctts_count_old = msc->ctts_count;
2960 int64_t edit_list_media_time = 0;
2961 int64_t edit_list_duration = 0;
2962 int64_t frame_duration = 0;
2963 int64_t edit_list_dts_counter = 0;
2964 int64_t edit_list_dts_entry_end = 0;
2965 int64_t edit_list_start_ctts_sample = 0;
2967 int64_t edit_list_index = 0;
2969 int64_t index_ctts_count;
2971 unsigned int ctts_allocated_size = 0;
2972 int64_t start_dts = 0;
2973 int64_t edit_list_media_time_dts = 0;
2974 int64_t edit_list_start_encountered = 0;
2975 int64_t search_timestamp = 0;
2976 int64_t* frame_duration_buffer = NULL;
2977 int num_discarded_begin = 0;
2978 int first_non_zero_audio_edit = -1;
2979 int packet_skip_samples = 0;
2981 if (!msc->elst_data || msc->elst_count <= 0 || nb_old <= 0) {
2984 // Clean AVStream from traces of old index
2985 st->index_entries = NULL;
2986 st->index_entries_allocated_size = 0;
2987 st->nb_index_entries = 0;
2989 // Clean ctts fields of MOVStreamContext
2990 msc->ctts_data = NULL;
2991 msc->ctts_count = 0;
2992 msc->ctts_index = 0;
2993 msc->ctts_sample = 0;
2995 // If the dts_shift is positive (in case of negative ctts values in mov),
2996 // then negate the DTS by dts_shift
2997 if (msc->dts_shift > 0)
2998 edit_list_dts_entry_end -= msc->dts_shift;
3000 // Offset the DTS by ctts[0] to make the PTS of the first frame 0
3001 if (ctts_data_old && ctts_count_old > 0) {
3002 edit_list_dts_entry_end -= ctts_data_old[0].duration;
3003 av_log(mov->fc, AV_LOG_DEBUG, "Offset DTS by ctts[%d].duration: %d\n", 0, ctts_data_old[0].duration);
3006 start_dts = edit_list_dts_entry_end;
3008 while (get_edit_list_entry(mov, msc, edit_list_index, &edit_list_media_time,
3009 &edit_list_duration, mov->time_scale)) {
3010 av_log(mov->fc, AV_LOG_DEBUG, "Processing st: %d, edit list %"PRId64" - media time: %"PRId64", duration: %"PRId64"\n",
3011 st->index, edit_list_index, edit_list_media_time, edit_list_duration);
3013 edit_list_dts_counter = edit_list_dts_entry_end;
3014 edit_list_dts_entry_end += edit_list_duration;
3015 num_discarded_begin = 0;
3016 if (edit_list_media_time == -1) {
3020 // If we encounter a non-negative edit list reset the skip_samples/start_pad fields and set them
3021 // according to the edit list below.
3022 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3023 if (first_non_zero_audio_edit < 0) {
3024 first_non_zero_audio_edit = 1;
3026 first_non_zero_audio_edit = 0;
3029 if (first_non_zero_audio_edit > 0)
3030 st->skip_samples = msc->start_pad = 0;
3033 //find closest previous key frame
3034 edit_list_media_time_dts = edit_list_media_time;
3035 if (msc->dts_shift > 0) {
3036 edit_list_media_time_dts -= msc->dts_shift;
3039 // While reordering frame index according to edit list we must handle properly
3040 // the scenario when edit list entry starts from none key frame.
3041 // We find closest previous key frame and preserve it and consequent frames in index.
3042 // All frames which are outside edit list entry time boundaries will be dropped after decoding.
3043 search_timestamp = edit_list_media_time_dts;
3044 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
3045 // Audio decoders like AAC need need a decoder delay samples previous to the current sample,
3046 // to correctly decode this frame. Hence for audio we seek to a frame 1 sec. before the
3047 // edit_list_media_time to cover the decoder delay.
3048 search_timestamp = FFMAX(search_timestamp - msc->time_scale, e_old[0].timestamp);
3051 index = find_prev_closest_index(st, e_old, nb_old, search_timestamp, 0);
3053 av_log(mov->fc, AV_LOG_WARNING,
3054 "st: %d edit list: %"PRId64" Missing key frame while searching for timestamp: %"PRId64"\n",
3055 st->index, edit_list_index, search_timestamp);
3056 index = find_prev_closest_index(st, e_old, nb_old, search_timestamp, AVSEEK_FLAG_ANY);
3059 av_log(mov->fc, AV_LOG_WARNING,
3060 "st: %d edit list %"PRId64" Cannot find an index entry before timestamp: %"PRId64".\n"
3061 "Rounding edit list media time to zero.\n",
3062 st->index, edit_list_index, search_timestamp);
3064 edit_list_media_time = 0;
3067 current = e_old + index;
3070 ctts_sample_old = 0;
3072 // set ctts_index properly for the found key frame
3073 for (index_ctts_count = 0; index_ctts_count < index; index_ctts_count++) {
3074 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3076 if (ctts_data_old[ctts_index_old].count == ctts_sample_old) {
3078 ctts_sample_old = 0;
3083 edit_list_start_ctts_sample = ctts_sample_old;
3085 // Iterate over index and arrange it according to edit list
3086 edit_list_start_encountered = 0;
3087 for (; current < e_old_end; current++, index++) {
3088 // check if frame outside edit list mark it for discard
3089 frame_duration = (current + 1 < e_old_end) ?
3090 ((current + 1)->timestamp - current->timestamp) : edit_list_duration;
3092 flags = current->flags;
3094 // frames (pts) before or after edit list
3095 curr_cts = current->timestamp + msc->dts_shift;
3097 if (ctts_data_old && ctts_index_old < ctts_count_old) {
3098 av_log(mov->fc, AV_LOG_DEBUG, "shifted frame pts, curr_cts: %"PRId64" @ %"PRId64", ctts: %d, ctts_count: %"PRId64"\n",
3099 curr_cts, ctts_index_old, ctts_data_old[ctts_index_old].duration, ctts_count_old);
3100 curr_cts += ctts_data_old[ctts_index_old].duration;
3102 if (ctts_sample_old == ctts_data_old[ctts_index_old].count) {
3103 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3104 &ctts_allocated_size,
3105 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3106 ctts_data_old[ctts_index_old].duration) == -1) {
3107 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3109 ctts_data_old[ctts_index_old].count - edit_list_start_ctts_sample,
3110 ctts_data_old[ctts_index_old].duration);
3114 ctts_sample_old = 0;
3115 edit_list_start_ctts_sample = 0;
3119 if (curr_cts < edit_list_media_time || curr_cts >= (edit_list_duration + edit_list_media_time)) {
3120 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id != AV_CODEC_ID_VORBIS &&
3121 curr_cts < edit_list_media_time && curr_cts + frame_duration > edit_list_media_time &&
3122 first_non_zero_audio_edit > 0) {
3123 packet_skip_samples = edit_list_media_time - curr_cts;
3124 st->skip_samples += packet_skip_samples;
3126 // Shift the index entry timestamp by packet_skip_samples to be correct.
3127 edit_list_dts_counter -= packet_skip_samples;
3128 if (edit_list_start_encountered == 0) {
3129 edit_list_start_encountered = 1;
3130 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3131 // discarded packets.
3132 if (frame_duration_buffer) {
3133 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3134 frame_duration_buffer, num_discarded_begin);
3135 av_freep(&frame_duration_buffer);
3139 av_log(mov->fc, AV_LOG_DEBUG, "skip %d audio samples from curr_cts: %"PRId64"\n", packet_skip_samples, curr_cts);
3141 flags |= AVINDEX_DISCARD_FRAME;
3142 av_log(mov->fc, AV_LOG_DEBUG, "drop a frame at curr_cts: %"PRId64" @ %"PRId64"\n", curr_cts, index);
3144 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && edit_list_start_encountered == 0) {
3145 num_discarded_begin++;
3146 frame_duration_buffer = av_realloc(frame_duration_buffer,
3147 num_discarded_begin * sizeof(int64_t));
3148 if (!frame_duration_buffer) {
3149 av_log(mov->fc, AV_LOG_ERROR, "Cannot reallocate frame duration buffer\n");
3152 frame_duration_buffer[num_discarded_begin - 1] = frame_duration;
3154 // Increment skip_samples for the first non-zero audio edit list
3155 if (first_non_zero_audio_edit > 0 && st->codecpar->codec_id != AV_CODEC_ID_VORBIS) {
3156 st->skip_samples += frame_duration;
3157 msc->start_pad = st->skip_samples;
3161 } else if (edit_list_start_encountered == 0) {
3162 edit_list_start_encountered = 1;
3163 // Make timestamps strictly monotonically increasing for audio, by rewriting timestamps for
3164 // discarded packets.
3165 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && frame_duration_buffer) {
3166 fix_index_entry_timestamps(st, st->nb_index_entries, edit_list_dts_counter,
3167 frame_duration_buffer, num_discarded_begin);
3168 av_freep(&frame_duration_buffer);
3172 if (add_index_entry(st, current->pos, edit_list_dts_counter, current->size,
3173 current->min_distance, flags) == -1) {
3174 av_log(mov->fc, AV_LOG_ERROR, "Cannot add index entry\n");
3178 // Only start incrementing DTS in frame_duration amounts, when we encounter a frame in edit list.
3179 if (edit_list_start_encountered > 0) {
3180 edit_list_dts_counter = edit_list_dts_counter + frame_duration;
3183 // Break when found first key frame after edit entry completion
3184 if (((curr_cts + frame_duration) >= (edit_list_duration + edit_list_media_time)) &&
3185 ((flags & AVINDEX_KEYFRAME) || ((st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)))) {
3187 if (ctts_data_old && ctts_sample_old != 0) {
3188 if (add_ctts_entry(&msc->ctts_data, &msc->ctts_count,
3189 &ctts_allocated_size,
3190 ctts_sample_old - edit_list_start_ctts_sample,
3191 ctts_data_old[ctts_index_old].duration) == -1) {
3192 av_log(mov->fc, AV_LOG_ERROR, "Cannot add CTTS entry %"PRId64" - {%"PRId64", %d}\n",
3193 ctts_index_old, ctts_sample_old - edit_list_start_ctts_sample,
3194 ctts_data_old[ctts_index_old].duration);
3202 // Update av stream length
3203 st->duration = edit_list_dts_entry_end - start_dts;
3205 // Free the old index and the old CTTS structures
3207 av_free(ctts_data_old);
3210 static void mov_build_index(MOVContext *mov, AVStream *st)
3212 MOVStreamContext *sc = st->priv_data;
3213 int64_t current_offset;
3214 int64_t current_dts = 0;
3215 unsigned int stts_index = 0;
3216 unsigned int stsc_index = 0;
3217 unsigned int stss_index = 0;
3218 unsigned int stps_index = 0;
3220 uint64_t stream_size = 0;
3222 if (sc->elst_count) {
3223 int i, edit_start_index = 0;
3224 int64_t empty_duration = 0; // empty duration of the first edit list entry
3225 int64_t start_time = 0; // start time of the media
3227 for (i = 0; i < sc->elst_count; i++) {
3228 const MOVElst *e = &sc->elst_data[i];
3229 if (i == 0 && e->time == -1) {
3230 /* if empty, the first entry is the start time of the stream
3231 * relative to the presentation itself */
3232 empty_duration = e->duration;
3233 edit_start_index = 1;
3234 } else if (i == edit_start_index && e->time >= 0) {
3235 start_time = e->time;
3239 /* adjust first dts according to edit list */
3240 if ((empty_duration || start_time) && mov->time_scale > 0) {
3242 empty_duration = av_rescale(empty_duration, sc->time_scale, mov->time_scale);
3243 sc->time_offset = start_time - empty_duration;
3247 /* only use old uncompressed audio chunk demuxing when stts specifies it */
3248 if (!(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO &&
3249 sc->stts_count == 1 && sc->stts_data[0].duration == 1)) {
3250 unsigned int current_sample = 0;
3251 unsigned int stts_sample = 0;
3252 unsigned int sample_size;
3253 unsigned int distance = 0;
3254 unsigned int rap_group_index = 0;
3255 unsigned int rap_group_sample = 0;
3256 int64_t last_dts = 0;
3257 int64_t dts_correction = 0;
3258 int rap_group_present = sc->rap_group_count && sc->rap_group;
3259 int key_off = (sc->keyframe_count && sc->keyframes[0] > 0) || (sc->stps_count && sc->stps_data[0] > 0);
3261 current_dts -= sc->dts_shift;
3262 last_dts = current_dts;
3264 if (!sc->sample_count || st->nb_index_entries)
3266 if (sc->sample_count >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3268 if (av_reallocp_array(&st->index_entries,
3269 st->nb_index_entries + sc->sample_count,
3270 sizeof(*st->index_entries)) < 0) {
3271 st->nb_index_entries = 0;
3274 st->index_entries_allocated_size = (st->nb_index_entries + sc->sample_count) * sizeof(*st->index_entries);
3276 for (i = 0; i < sc->chunk_count; i++) {
3277 int64_t next_offset = i+1 < sc->chunk_count ? sc->chunk_offsets[i+1] : INT64_MAX;
3278 current_offset = sc->chunk_offsets[i];
3279 while (stsc_index + 1 < sc->stsc_count &&
3280 i + 1 == sc->stsc_data[stsc_index + 1].first)
3283 if (next_offset > current_offset && sc->sample_size>0 && sc->sample_size < sc->stsz_sample_size &&
3284 sc->stsc_data[stsc_index].count * (int64_t)sc->stsz_sample_size > next_offset - current_offset) {
3285 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too large), ignoring\n", sc->stsz_sample_size);
3286 sc->stsz_sample_size = sc->sample_size;
3288 if (sc->stsz_sample_size>0 && sc->stsz_sample_size < sc->sample_size) {
3289 av_log(mov->fc, AV_LOG_WARNING, "STSZ sample size %d invalid (too small), ignoring\n", sc->stsz_sample_size);
3290 sc->stsz_sample_size = sc->sample_size;
3293 for (j = 0; j < sc->stsc_data[stsc_index].count; j++) {
3295 if (current_sample >= sc->sample_count) {
3296 av_log(mov->fc, AV_LOG_ERROR, "wrong sample count\n");
3300 if (!sc->keyframe_absent && (!sc->keyframe_count || current_sample+key_off == sc->keyframes[stss_index])) {
3302 if (stss_index + 1 < sc->keyframe_count)
3304 } else if (sc->stps_count && current_sample+key_off == sc->stps_data[stps_index]) {
3306 if (stps_index + 1 < sc->stps_count)
3309 if (rap_group_present && rap_group_index < sc->rap_group_count) {
3310 if (sc->rap_group[rap_group_index].index > 0)
3312 if (++rap_group_sample == sc->rap_group[rap_group_index].count) {
3313 rap_group_sample = 0;
3317 if (sc->keyframe_absent
3319 && !rap_group_present
3320 && (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO || (i==0 && j==0)))
3324 sample_size = sc->stsz_sample_size > 0 ? sc->stsz_sample_size : sc->sample_sizes[current_sample];
3325 if (sc->pseudo_stream_id == -1 ||
3326 sc->stsc_data[stsc_index].id - 1 == sc->pseudo_stream_id) {
3328 if (sample_size > 0x3FFFFFFF) {
3329 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", sample_size);
3332 e = &st->index_entries[st->nb_index_entries++];
3333 e->pos = current_offset;
3334 e->timestamp = current_dts;
3335 e->size = sample_size;
3336 e->min_distance = distance;
3337 e->flags = keyframe ? AVINDEX_KEYFRAME : 0;
3338 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
3339 "size %d, distance %d, keyframe %d\n", st->index, current_sample,
3340 current_offset, current_dts, sample_size, distance, keyframe);
3341 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && st->nb_index_entries < 100)
3342 ff_rfps_add_frame(mov->fc, st, current_dts);
3345 current_offset += sample_size;
3346 stream_size += sample_size;
3348 /* A negative sample duration is invalid based on the spec,
3349 * but some samples need it to correct the DTS. */
3350 if (sc->stts_data[stts_index].duration < 0) {
3351 av_log(mov->fc, AV_LOG_WARNING,
3352 "Invalid SampleDelta %d in STTS, at %d st:%d\n",
3353 sc->stts_data[stts_index].duration, stts_index,
3355 dts_correction += sc->stts_data[stts_index].duration - 1;
3356 sc->stts_data[stts_index].duration = 1;
3358 current_dts += sc->stts_data[stts_index].duration;
3359 if (!dts_correction || current_dts + dts_correction > last_dts) {
3360 current_dts += dts_correction;
3363 /* Avoid creating non-monotonous DTS */
3364 dts_correction += current_dts - last_dts - 1;
3365 current_dts = last_dts + 1;
3367 last_dts = current_dts;
3371 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
3377 if (st->duration > 0)
3378 st->codecpar->bit_rate = stream_size*8*sc->time_scale/st->duration;
3380 unsigned chunk_samples, total = 0;
3382 // compute total chunk count
3383 for (i = 0; i < sc->stsc_count; i++) {
3384 unsigned count, chunk_count;
3386 chunk_samples = sc->stsc_data[i].count;
3387 if (i != sc->stsc_count - 1 &&
3388 sc->samples_per_frame && chunk_samples % sc->samples_per_frame) {
3389 av_log(mov->fc, AV_LOG_ERROR, "error unaligned chunk\n");
3393 if (sc->samples_per_frame >= 160) { // gsm
3394 count = chunk_samples / sc->samples_per_frame;
3395 } else if (sc->samples_per_frame > 1) {
3396 unsigned samples = (1024/sc->samples_per_frame)*sc->samples_per_frame;
3397 count = (chunk_samples+samples-1) / samples;
3399 count = (chunk_samples+1023) / 1024;
3402 if (i < sc->stsc_count - 1)
3403 chunk_count = sc->stsc_data[i+1].first - sc->stsc_data[i].first;
3405 chunk_count = sc->chunk_count - (sc->stsc_data[i].first - 1);
3406 total += chunk_count * count;
3409 av_log(mov->fc, AV_LOG_TRACE, "chunk count %d\n", total);
3410 if (total >= UINT_MAX / sizeof(*st->index_entries) - st->nb_index_entries)
3412 if (av_reallocp_array(&st->index_entries,
3413 st->nb_index_entries + total,
3414 sizeof(*st->index_entries)) < 0) {
3415 st->nb_index_entries = 0;
3418 st->index_entries_allocated_size = (st->nb_index_entries + total) * sizeof(*st->index_entries);
3421 for (i = 0; i < sc->chunk_count; i++) {
3422 current_offset = sc->chunk_offsets[i];
3423 if (stsc_index + 1 < sc->stsc_count &&
3424 i + 1 == sc->stsc_data[stsc_index + 1].first)
3426 chunk_samples = sc->stsc_data[stsc_index].count;
3428 while (chunk_samples > 0) {
3430 unsigned size, samples;
3432 if (sc->samples_per_frame > 1 && !sc->bytes_per_frame) {
3433 avpriv_request_sample(mov->fc,
3434 "Zero bytes per frame, but %d samples per frame",
3435 sc->samples_per_frame);
3439 if (sc->samples_per_frame >= 160) { // gsm
3440 samples = sc->samples_per_frame;
3441 size = sc->bytes_per_frame;
3443 if (sc->samples_per_frame > 1) {
3444 samples = FFMIN((1024 / sc->samples_per_frame)*
3445 sc->samples_per_frame, chunk_samples);
3446 size = (samples / sc->samples_per_frame) * sc->bytes_per_frame;
3448 samples = FFMIN(1024, chunk_samples);
3449 size = samples * sc->sample_size;
3453 if (st->nb_index_entries >= total) {
3454 av_log(mov->fc, AV_LOG_ERROR, "wrong chunk count %d\n", total);
3457 if (size > 0x3FFFFFFF) {
3458 av_log(mov->fc, AV_LOG_ERROR, "Sample size %u is too large\n", size);
3461 e = &st->index_entries[st->nb_index_entries++];
3462 e->pos = current_offset;
3463 e->timestamp = current_dts;
3465 e->min_distance = 0;
3466 e->flags = AVINDEX_KEYFRAME;
3467 av_log(mov->fc, AV_LOG_TRACE, "AVIndex stream %d, chunk %d, offset %"PRIx64", dts %"PRId64", "
3468 "size %d, duration %d\n", st->index, i, current_offset, current_dts,
3471 current_offset += size;
3472 current_dts += samples;
3473 chunk_samples -= samples;
3478 // Fix index according to edit lists.
3479 mov_fix_index(mov, st);
3482 static int test_same_origin(const char *src, const char *ref) {
3492 av_url_split(src_proto, sizeof(src_proto), src_auth, sizeof(src_auth), src_host, sizeof(src_host), &src_port, NULL, 0, src);
3493 av_url_split(ref_proto, sizeof(ref_proto), ref_auth, sizeof(ref_auth), ref_host, sizeof(ref_host), &ref_port, NULL, 0, ref);
3495 if (strlen(src) == 0) {
3497 } else if (strlen(src_auth) + 1 >= sizeof(src_auth) ||
3498 strlen(ref_auth) + 1 >= sizeof(ref_auth) ||
3499 strlen(src_host) + 1 >= sizeof(src_host) ||
3500 strlen(ref_host) + 1 >= sizeof(ref_host)) {
3502 } else if (strcmp(src_proto, ref_proto) ||
3503 strcmp(src_auth, ref_auth) ||
3504 strcmp(src_host, ref_host) ||
3505 src_port != ref_port) {
3511 static int mov_open_dref(MOVContext *c, AVIOContext **pb, const char *src, MOVDref *ref)
3513 /* try relative path, we do not try the absolute because it can leak information about our
3514 system to an attacker */
3515 if (ref->nlvl_to > 0 && ref->nlvl_from > 0) {
3516 char filename[1025];
3517 const char *src_path;
3520 /* find a source dir */
3521 src_path = strrchr(src, '/');
3527 /* find a next level down to target */
3528 for (i = 0, l = strlen(ref->path) - 1; l >= 0; l--)
3529 if (ref->path[l] == '/') {
3530 if (i == ref->nlvl_to - 1)
3536 /* compose filename if next level down to target was found */
3537 if (i == ref->nlvl_to - 1 && src_path - src < sizeof(filename)) {
3538 memcpy(filename, src, src_path - src);
3539 filename[src_path - src] = 0;
3541 for (i = 1; i < ref->nlvl_from; i++)
3542 av_strlcat(filename, "../", sizeof(filename));
3544 av_strlcat(filename, ref->path + l + 1, sizeof(filename));
3545 if (!c->use_absolute_path) {
3546 int same_origin = test_same_origin(src, filename);
3549 av_log(c->fc, AV_LOG_ERROR,
3550 "Reference with mismatching origin, %s not tried for security reasons, "
3551 "set demuxer option use_absolute_path to allow it anyway\n",
3553 return AVERROR(ENOENT);
3556 if(strstr(ref->path + l + 1, "..") ||
3557 strstr(ref->path + l + 1, ":") ||
3558 (ref->nlvl_from > 1 && same_origin < 0) ||
3559 (filename[0] == '/' && src_path == src))
3560 return AVERROR(ENOENT);
3563 if (strlen(filename) + 1 == sizeof(filename))
3564 return AVERROR(ENOENT);
3565 if (!c->fc->io_open(c->fc, pb, filename, AVIO_FLAG_READ, NULL))
3568 } else if (c->use_absolute_path) {
3569 av_log(c->fc, AV_LOG_WARNING, "Using absolute path on user request, "
3570 "this is a possible security issue\n");
3571 if (!c->fc->io_open(c->fc, pb, ref->path, AVIO_FLAG_READ, NULL))
3574 av_log(c->fc, AV_LOG_ERROR,
3575 "Absolute path %s not tried for security reasons, "
3576 "set demuxer option use_absolute_path to allow absolute paths\n",
3580 return AVERROR(ENOENT);
3583 static void fix_timescale(MOVContext *c, MOVStreamContext *sc)
3585 if (sc->time_scale <= 0) {
3586 av_log(c->fc, AV_LOG_WARNING, "stream %d, timescale not set\n", sc->ffindex);
3587 sc->time_scale = c->time_scale;
3588 if (sc->time_scale <= 0)
3593 static int mov_read_trak(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3596 MOVStreamContext *sc;
3599 st = avformat_new_stream(c->fc, NULL);
3600 if (!st) return AVERROR(ENOMEM);
3601 st->id = c->fc->nb_streams;
3602 sc = av_mallocz(sizeof(MOVStreamContext));
3603 if (!sc) return AVERROR(ENOMEM);
3606 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
3607 sc->ffindex = st->index;
3608 c->trak_index = st->index;
3610 if ((ret = mov_read_default(c, pb, atom)) < 0)
3616 if (sc->chunk_count && (!sc->stts_count || !sc->stsc_count ||
3617 (!sc->sample_size && !sc->sample_count))) {
3618 av_log(c->fc, AV_LOG_ERROR, "stream %d, missing mandatory atoms, broken header\n",
3623 fix_timescale(c, sc);
3625 avpriv_set_pts_info(st, 64, 1, sc->time_scale);
3627 mov_build_index(c, st);
3629 if (sc->dref_id-1 < sc->drefs_count && sc->drefs[sc->dref_id-1].path) {
3630 MOVDref *dref = &sc->drefs[sc->dref_id - 1];
3631 if (c->enable_drefs) {
3632 if (mov_open_dref(c, &sc->pb, c->fc->filename, dref) < 0)
3633 av_log(c->fc, AV_LOG_ERROR,
3634 "stream %d, error opening alias: path='%s', dir='%s', "
3635 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d\n",
3636 st->index, dref->path, dref->dir, dref->filename,
3637 dref->volume, dref->nlvl_from, dref->nlvl_to);
3639 av_log(c->fc, AV_LOG_WARNING,
3640 "Skipped opening external track: "
3641 "stream %d, alias: path='%s', dir='%s', "
3642 "filename='%s', volume='%s', nlvl_from=%d, nlvl_to=%d."
3643 "Set enable_drefs to allow this.\n",
3644 st->index, dref->path, dref->dir, dref->filename,
3645 dref->volume, dref->nlvl_from, dref->nlvl_to);
3649 sc->pb_is_copied = 1;
3652 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
3653 if (!st->sample_aspect_ratio.num && st->codecpar->width && st->codecpar->height &&
3654 sc->height && sc->width &&
3655 (st->codecpar->width != sc->width || st->codecpar->height != sc->height)) {
3656 st->sample_aspect_ratio = av_d2q(((double)st->codecpar->height * sc->width) /
3657 ((double)st->codecpar->width * sc->height), INT_MAX);
3660 #if FF_API_R_FRAME_RATE
3661 if (sc->stts_count == 1 || (sc->stts_count == 2 && sc->stts_data[1].count == 1))
3662 av_reduce(&st->r_frame_rate.num, &st->r_frame_rate.den,
3663 sc->time_scale, sc->stts_data[0].duration, INT_MAX);
3667 // done for ai5q, ai52, ai55, ai1q, ai12 and ai15.
3668 if (!st->codecpar->extradata_size && st->codecpar->codec_id == AV_CODEC_ID_H264 &&
3669 TAG_IS_AVCI(st->codecpar->codec_tag)) {
3670 ret = ff_generate_avci_extradata(st);
3675 switch (st->codecpar->codec_id) {
3676 #if CONFIG_H261_DECODER
3677 case AV_CODEC_ID_H261:
3679 #if CONFIG_H263_DECODER
3680 case AV_CODEC_ID_H263:
3682 #if CONFIG_MPEG4_DECODER
3683 case AV_CODEC_ID_MPEG4:
3685 st->codecpar->width = 0; /* let decoder init width/height */
3686 st->codecpar->height= 0;
3690 // If the duration of the mp3 packets is not constant, then they could need a parser
3691 if (st->codecpar->codec_id == AV_CODEC_ID_MP3
3692 && sc->stts_count > 3
3693 && sc->stts_count*10 > st->nb_frames
3694 && sc->time_scale == st->codecpar->sample_rate) {
3695 st->need_parsing = AVSTREAM_PARSE_FULL;
3697 /* Do not need those anymore. */
3698 av_freep(&sc->chunk_offsets);
3699 av_freep(&sc->sample_sizes);
3700 av_freep(&sc->keyframes);
3701 av_freep(&sc->stts_data);
3702 av_freep(&sc->stps_data);
3703 av_freep(&sc->elst_data);
3704 av_freep(&sc->rap_group);
3709 static int mov_read_ilst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3712 c->itunes_metadata = 1;
3713 ret = mov_read_default(c, pb, atom);
3714 c->itunes_metadata = 0;
3718 static int mov_read_keys(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3727 count = avio_rb32(pb);
3728 if (count > UINT_MAX / sizeof(*c->meta_keys) - 1) {
3729 av_log(c->fc, AV_LOG_ERROR,
3730 "The 'keys' atom with the invalid key count: %d\n", count);
3731 return AVERROR_INVALIDDATA;
3734 c->meta_keys_count = count + 1;
3735 c->meta_keys = av_mallocz(c->meta_keys_count * sizeof(*c->meta_keys));
3737 return AVERROR(ENOMEM);
3739 for (i = 1; i <= count; ++i) {
3740 uint32_t key_size = avio_rb32(pb);
3741 uint32_t type = avio_rl32(pb);
3743 av_log(c->fc, AV_LOG_ERROR,
3744 "The key# %d in meta has invalid size: %d\n", i, key_size);
3745 return AVERROR_INVALIDDATA;
3748 if (type != MKTAG('m','d','t','a')) {
3749 avio_skip(pb, key_size);
3751 c->meta_keys[i] = av_mallocz(key_size + 1);
3752 if (!c->meta_keys[i])
3753 return AVERROR(ENOMEM);
3754 avio_read(pb, c->meta_keys[i], key_size);
3760 static int mov_read_custom(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3762 int64_t end = avio_tell(pb) + atom.size;
3763 uint8_t *key = NULL, *val = NULL, *mean = NULL;
3767 MOVStreamContext *sc;
3769 if (c->fc->nb_streams < 1)
3771 st = c->fc->streams[c->fc->nb_streams-1];
3774 for (i = 0; i < 3; i++) {
3778 if (end - avio_tell(pb) <= 12)
3781 len = avio_rb32(pb);
3782 tag = avio_rl32(pb);
3783 avio_skip(pb, 4); // flags
3785 if (len < 12 || len - 12 > end - avio_tell(pb))
3789 if (tag == MKTAG('m', 'e', 'a', 'n'))
3791 else if (tag == MKTAG('n', 'a', 'm', 'e'))
3793 else if (tag == MKTAG('d', 'a', 't', 'a') && len > 4) {
3800 *p = av_malloc(len + 1);
3803 ret = ffio_read_size(pb, *p, len);
3811 if (mean && key && val) {
3812 if (strcmp(key, "iTunSMPB") == 0) {
3813 int priming, remainder, samples;
3814 if(sscanf(val, "%*X %X %X %X", &priming, &remainder, &samples) == 3){
3815 if(priming>0 && priming<16384)
3816 sc->start_pad = priming;
3819 if (strcmp(key, "cdec") != 0) {
3820 av_dict_set(&c->fc->metadata, key, val,
3821 AV_DICT_DONT_STRDUP_KEY | AV_DICT_DONT_STRDUP_VAL);
3825 av_log(c->fc, AV_LOG_VERBOSE,
3826 "Unhandled or malformed custom metadata of size %"PRId64"\n", atom.size);
3829 avio_seek(pb, end, SEEK_SET);
3836 static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3838 while (atom.size > 8) {
3839 uint32_t tag = avio_rl32(pb);
3841 if (tag == MKTAG('h','d','l','r')) {
3842 avio_seek(pb, -8, SEEK_CUR);
3844 return mov_read_default(c, pb, atom);
3850 static int mov_read_tkhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3855 int display_matrix[3][3];
3857 MOVStreamContext *sc;
3861 if (c->fc->nb_streams < 1)
3863 st = c->fc->streams[c->fc->nb_streams-1];
3866 version = avio_r8(pb);
3867 flags = avio_rb24(pb);
3868 st->disposition |= (flags & MOV_TKHD_FLAG_ENABLED) ? AV_DISPOSITION_DEFAULT : 0;
3874 avio_rb32(pb); /* creation time */
3875 avio_rb32(pb); /* modification time */
3877 st->id = (int)avio_rb32(pb); /* track id (NOT 0 !)*/
3878 avio_rb32(pb); /* reserved */
3880 /* highlevel (considering edits) duration in movie timebase */
3881 (version == 1) ? avio_rb64(pb) : avio_rb32(pb);
3882 avio_rb32(pb); /* reserved */
3883 avio_rb32(pb); /* reserved */
3885 avio_rb16(pb); /* layer */
3886 avio_rb16(pb); /* alternate group */
3887 avio_rb16(pb); /* volume */
3888 avio_rb16(pb); /* reserved */
3890 //read in the display matrix (outlined in ISO 14496-12, Section 6.2.2)
3891 // they're kept in fixed point format through all calculations
3892 // save u,v,z to store the whole matrix in the AV_PKT_DATA_DISPLAYMATRIX
3893 // side data, but the scale factor is not needed to calculate aspect ratio
3894 for (i = 0; i < 3; i++) {
3895 display_matrix[i][0] = avio_rb32(pb); // 16.16 fixed point
3896 display_matrix[i][1] = avio_rb32(pb); // 16.16 fixed point
3897 display_matrix[i][2] = avio_rb32(pb); // 2.30 fixed point
3900 width = avio_rb32(pb); // 16.16 fixed point track width
3901 height = avio_rb32(pb); // 16.16 fixed point track height
3902 sc->width = width >> 16;
3903 sc->height = height >> 16;
3905 // save the matrix and add rotate metadata when it is not the default
3907 if (display_matrix[0][0] != (1 << 16) ||
3908 display_matrix[1][1] != (1 << 16) ||
3909 display_matrix[2][2] != (1 << 30) ||
3910 display_matrix[0][1] || display_matrix[0][2] ||
3911 display_matrix[1][0] || display_matrix[1][2] ||
3912 display_matrix[2][0] || display_matrix[2][1]) {
3916 av_freep(&sc->display_matrix);
3917 sc->display_matrix = av_malloc(sizeof(int32_t) * 9);
3918 if (!sc->display_matrix)
3919 return AVERROR(ENOMEM);
3921 for (i = 0; i < 3; i++)
3922 for (j = 0; j < 3; j++)
3923 sc->display_matrix[i * 3 + j] = display_matrix[i][j];
3925 rotate = av_display_rotation_get(sc->display_matrix);
3926 if (!isnan(rotate)) {
3927 char rotate_buf[64];
3929 if (rotate < 0) // for backward compatibility
3931 snprintf(rotate_buf, sizeof(rotate_buf), "%g", rotate);
3932 av_dict_set(&st->metadata, "rotate", rotate_buf, 0);
3936 // transform the display width/height according to the matrix
3937 // to keep the same scale, use [width height 1<<16]
3938 if (width && height && sc->display_matrix) {
3939 double disp_transform[2];
3941 for (i = 0; i < 2; i++)
3942 disp_transform[i] = hypot(display_matrix[i][0], display_matrix[i][1]);
3944 if (disp_transform[0] > 0 && disp_transform[1] > 0 &&
3945 disp_transform[0] < (1<<24) && disp_transform[1] < (1<<24) &&
3946 fabs((disp_transform[0] / disp_transform[1]) - 1.0) > 0.01)
3947 st->sample_aspect_ratio = av_d2q(
3948 disp_transform[0] / disp_transform[1],
3954 static int mov_read_tfhd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
3956 MOVFragment *frag = &c->fragment;
3957 MOVTrackExt *trex = NULL;
3958 MOVFragmentIndex* index = NULL;
3959 int flags, track_id, i, found = 0;
3961 avio_r8(pb); /* version */
3962 flags = avio_rb24(pb);
3964 track_id = avio_rb32(pb);
3966 return AVERROR_INVALIDDATA;
3967 frag->track_id = track_id;
3968 for (i = 0; i < c->trex_count; i++)
3969 if (c->trex_data[i].track_id == frag->track_id) {
3970 trex = &c->trex_data[i];
3974 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding trex\n");
3975 return AVERROR_INVALIDDATA;
3978 frag->base_data_offset = flags & MOV_TFHD_BASE_DATA_OFFSET ?
3979 avio_rb64(pb) : flags & MOV_TFHD_DEFAULT_BASE_IS_MOOF ?
3980 frag->moof_offset : frag->implicit_offset;
3981 frag->stsd_id = flags & MOV_TFHD_STSD_ID ? avio_rb32(pb) : trex->stsd_id;
3983 frag->duration = flags & MOV_TFHD_DEFAULT_DURATION ?
3984 avio_rb32(pb) : trex->duration;
3985 frag->size = flags & MOV_TFHD_DEFAULT_SIZE ?
3986 avio_rb32(pb) : trex->size;
3987 frag->flags = flags & MOV_TFHD_DEFAULT_FLAGS ?
3988 avio_rb32(pb) : trex->flags;
3989 frag->time = AV_NOPTS_VALUE;
3990 for (i = 0; i < c->fragment_index_count; i++) {
3992 MOVFragmentIndex* candidate = c->fragment_index_data[i];
3993 if (candidate->track_id == frag->track_id) {
3994 av_log(c->fc, AV_LOG_DEBUG,
3995 "found fragment index for track %u\n", frag->track_id);
3997 for (j = index->current_item; j < index->item_count; j++) {
3998 if (frag->implicit_offset == index->items[j].moof_offset) {
3999 av_log(c->fc, AV_LOG_DEBUG, "found fragment index entry "
4000 "for track %u and moof_offset %"PRId64"\n",
4001 frag->track_id, index->items[j].moof_offset);
4002 frag->time = index->items[j].time;
4003 index->current_item = j + 1;
4012 if (index && !found) {
4013 av_log(c->fc, AV_LOG_DEBUG, "track %u has a fragment index but "
4014 "it doesn't have an (in-order) entry for moof_offset "
4015 "%"PRId64"\n", frag->track_id, frag->implicit_offset);
4017 av_log(c->fc, AV_LOG_TRACE, "frag flags 0x%x\n", frag->flags);
4021 static int mov_read_chap(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4026 num = atom.size / 4;
4027 if (!(new_tracks = av_malloc_array(num, sizeof(int))))
4028 return AVERROR(ENOMEM);
4030 av_free(c->chapter_tracks);
4031 c->chapter_tracks = new_tracks;
4032 c->nb_chapter_tracks = num;
4034 for (i = 0; i < num && !pb->eof_reached; i++)
4035 c->chapter_tracks[i] = avio_rb32(pb);
4040 static int mov_read_trex(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4045 if ((uint64_t)c->trex_count+1 >= UINT_MAX / sizeof(*c->trex_data))
4046 return AVERROR_INVALIDDATA;
4047 if ((err = av_reallocp_array(&c->trex_data, c->trex_count + 1,
4048 sizeof(*c->trex_data))) < 0) {
4053 c->fc->duration = AV_NOPTS_VALUE; // the duration from mvhd is not representing the whole file when fragments are used.
4055 trex = &c->trex_data[c->trex_count++];
4056 avio_r8(pb); /* version */
4057 avio_rb24(pb); /* flags */
4058 trex->track_id = avio_rb32(pb);
4059 trex->stsd_id = avio_rb32(pb);
4060 trex->duration = avio_rb32(pb);
4061 trex->size = avio_rb32(pb);
4062 trex->flags = avio_rb32(pb);
4066 static int mov_read_tfdt(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4068 MOVFragment *frag = &c->fragment;
4069 AVStream *st = NULL;
4070 MOVStreamContext *sc;
4073 for (i = 0; i < c->fc->nb_streams; i++) {
4074 if (c->fc->streams[i]->id == frag->track_id) {
4075 st = c->fc->streams[i];
4080 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
4081 return AVERROR_INVALIDDATA;
4084 if (sc->pseudo_stream_id + 1 != frag->stsd_id)
4086 version = avio_r8(pb);
4087 avio_rb24(pb); /* flags */
4089 sc->track_end = avio_rb64(pb);
4091 sc->track_end = avio_rb32(pb);
4096 static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4098 MOVFragment *frag = &c->fragment;
4099 AVStream *st = NULL;
4100 MOVStreamContext *sc;
4104 int data_offset = 0;
4105 unsigned entries, first_sample_flags = frag->flags;
4106 int flags, distance, i, err;
4108 for (i = 0; i < c->fc->nb_streams; i++) {
4109 if (c->fc->streams[i]->id == frag->track_id) {
4110 st = c->fc->streams[i];
4115 av_log(c->fc, AV_LOG_ERROR, "could not find corresponding track id %d\n", frag->track_id);
4116 return AVERROR_INVALIDDATA;
4119 if (sc->pseudo_stream_id+1 != frag->stsd_id && sc->pseudo_stream_id != -1)
4121 avio_r8(pb); /* version */
4122 flags = avio_rb24(pb);
4123 entries = avio_rb32(pb);
4124 av_log(c->fc, AV_LOG_TRACE, "flags 0x%x entries %d\n", flags, entries);
4126 /* Always assume the presence of composition time offsets.
4127 * Without this assumption, for instance, we cannot deal with a track in fragmented movies that meet the following.
4128 * 1) in the initial movie, there are no samples.
4129 * 2) in the first movie fragment, there is only one sample without composition time offset.
4130 * 3) in the subsequent movie fragments, there are samples with composition time offset. */
4131 if (!sc->ctts_count && sc->sample_count)
4133 /* Complement ctts table if moov atom doesn't have ctts atom. */
4134 ctts_data = av_realloc(NULL, sizeof(*sc->ctts_data));
4136 return AVERROR(ENOMEM);
4137 sc->ctts_data = ctts_data;
4138 sc->ctts_data[sc->ctts_count].count = sc->sample_count;
4139 sc->ctts_data[sc->ctts_count].duration = 0;
4142 if ((uint64_t)entries+sc->ctts_count >= UINT_MAX/sizeof(*sc->ctts_data))
4143 return AVERROR_INVALIDDATA;
4144 if ((err = av_reallocp_array(&sc->ctts_data, entries + sc->ctts_count,
4145 sizeof(*sc->ctts_data))) < 0) {
4149 if (flags & MOV_TRUN_DATA_OFFSET) data_offset = avio_rb32(pb);
4150 if (flags & MOV_TRUN_FIRST_SAMPLE_FLAGS) first_sample_flags = avio_rb32(pb);
4151 dts = sc->track_end - sc->time_offset;
4152 offset = frag->base_data_offset + data_offset;
4154 av_log(c->fc, AV_LOG_TRACE, "first sample flags 0x%x\n", first_sample_flags);
4155 for (i = 0; i < entries && !pb->eof_reached; i++) {
4156 unsigned sample_size = frag->size;
4157 int sample_flags = i ? frag->flags : first_sample_flags;
4158 unsigned sample_duration = frag->duration;
4161 if (flags & MOV_TRUN_SAMPLE_DURATION) sample_duration = avio_rb32(pb);
4162 if (flags & MOV_TRUN_SAMPLE_SIZE) sample_size = avio_rb32(pb);
4163 if (flags & MOV_TRUN_SAMPLE_FLAGS) sample_flags = avio_rb32(pb);
4164 sc->ctts_data[sc->ctts_count].count = 1;
4165 sc->ctts_data[sc->ctts_count].duration = (flags & MOV_TRUN_SAMPLE_CTS) ?
4167 mov_update_dts_shift(sc, sc->ctts_data[sc->ctts_count].duration);
4168 if (frag->time != AV_NOPTS_VALUE) {
4169 if (c->use_mfra_for == FF_MOV_FLAG_MFRA_PTS) {
4170 int64_t pts = frag->time;
4171 av_log(c->fc, AV_LOG_DEBUG, "found frag time %"PRId64
4172 " sc->dts_shift %d ctts.duration %d"
4173 " sc->time_offset %"PRId64" flags & MOV_TRUN_SAMPLE_CTS %d\n", pts,
4174 sc->dts_shift, sc->ctts_data[sc->ctts_count].duration,
4175 sc->time_offset, flags & MOV_TRUN_SAMPLE_CTS);
4176 dts = pts - sc->dts_shift;
4177 if (flags & MOV_TRUN_SAMPLE_CTS) {
4178 dts -= sc->ctts_data[sc->ctts_count].duration;
4180 dts -= sc->time_offset;
4182 av_log(c->fc, AV_LOG_DEBUG, "calculated into dts %"PRId64"\n", dts);
4184 dts = frag->time - sc->time_offset;
4185 av_log(c->fc, AV_LOG_DEBUG, "found frag time %"PRId64
4186 ", using it for dts\n", dts);
4188 frag->time = AV_NOPTS_VALUE;
4191 if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO)
4195 !(sample_flags & (MOV_FRAG_SAMPLE_FLAG_IS_NON_SYNC |
4196 MOV_FRAG_SAMPLE_FLAG_DEPENDS_YES));
4199 err = av_add_index_entry(st, offset, dts, sample_size, distance,
4200 keyframe ? AVINDEX_KEYFRAME : 0);
4202 av_log(c->fc, AV_LOG_ERROR, "Failed to add index entry\n");
4204 av_log(c->fc, AV_LOG_TRACE, "AVIndex stream %d, sample %d, offset %"PRIx64", dts %"PRId64", "
4205 "size %d, distance %d, keyframe %d\n", st->index, sc->sample_count+i,
4206 offset, dts, sample_size, distance, keyframe);
4208 dts += sample_duration;
4209 offset += sample_size;
4210 sc->data_size += sample_size;
4211 sc->duration_for_fps += sample_duration;
4212 sc->nb_frames_for_fps ++;
4215 if (pb->eof_reached)
4218 frag->implicit_offset = offset;
4220 sc->track_end = dts + sc->time_offset;
4221 if (st->duration < sc->track_end)
4222 st->duration = sc->track_end;
4227 static int mov_read_sidx(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4229 int64_t offset = avio_tell(pb) + atom.size, pts;
4231 unsigned i, track_id;
4232 AVStream *st = NULL;
4234 MOVStreamContext *sc, *ref_sc;
4235 MOVFragmentIndex *index = NULL;
4236 MOVFragmentIndex **tmp;
4237 AVRational timescale;
4239 version = avio_r8(pb);
4241 avpriv_request_sample(c->fc, "sidx version %u", version);
4245 avio_rb24(pb); // flags
4247 track_id = avio_rb32(pb); // Reference ID
4248 for (i = 0; i < c->fc->nb_streams; i++) {
4249 if (c->fc->streams[i]->id == track_id) {
4250 st = c->fc->streams[i];
4255 av_log(c->fc, AV_LOG_WARNING, "could not find corresponding track id %d\n", track_id);
4261 timescale = av_make_q(1, avio_rb32(pb));
4263 if (timescale.den <= 0) {
4264 av_log(c->fc, AV_LOG_ERROR, "Invalid sidx timescale 1/%d\n", timescale.den);
4265 return AVERROR_INVALIDDATA;
4269 pts = avio_rb32(pb);
4270 offset += avio_rb32(pb);
4272 pts = avio_rb64(pb);
4273 offset += avio_rb64(pb);
4276 avio_rb16(pb); // reserved
4278 index = av_mallocz(sizeof(MOVFragmentIndex));
4280 return AVERROR(ENOMEM);
4282 index->track_id = track_id;
4284 index->item_count = avio_rb16(pb);
4285 index->items = av_mallocz_array(index->item_count, sizeof(MOVFragmentIndexItem));
4287 if (!index->items) {
4289 return AVERROR(ENOMEM);
4292 for (i = 0; i < index->item_count; i++) {
4293 uint32_t size = avio_rb32(pb);
4294 uint32_t duration = avio_rb32(pb);
4295 if (size & 0x80000000) {
4296 avpriv_request_sample(c->fc, "sidx reference_type 1");
4297 av_freep(&index->items);
4299 return AVERROR_PATCHWELCOME;
4301 avio_rb32(pb); // sap_flags
4302 index->items[i].moof_offset = offset;
4303 index->items[i].time = av_rescale_q(pts, st->time_base, timescale);
4308 st->duration = sc->track_end = pts;
4310 tmp = av_realloc_array(c->fragment_index_data,
4311 c->fragment_index_count + 1,
4312 sizeof(MOVFragmentIndex*));
4314 av_freep(&index->items);
4316 return AVERROR(ENOMEM);
4319 c->fragment_index_data = tmp;
4320 c->fragment_index_data[c->fragment_index_count++] = index;
4323 if (offset == avio_size(pb)) {
4324 for (i = 0; i < c->fc->nb_streams; i++) {
4325 if (c->fc->streams[i]->id == c->fragment_index_data[0]->track_id) {
4326 ref_st = c->fc->streams[i];
4327 ref_sc = ref_st->priv_data;
4331 for (i = 0; i < c->fc->nb_streams; i++) {
4332 st = c->fc->streams[i];
4334 if (!sc->has_sidx) {
4335 st->duration = sc->track_end = av_rescale(ref_st->duration, sc->time_scale, ref_sc->time_scale);
4339 c->fragment_index_complete = 1;
4345 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
4346 /* like the files created with Adobe Premiere 5.0, for samples see */
4347 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
4348 static int mov_read_wide(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4353 return 0; /* continue */
4354 if (avio_rb32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
4355 avio_skip(pb, atom.size - 4);
4358 atom.type = avio_rl32(pb);
4360 if (atom.type != MKTAG('m','d','a','t')) {
4361 avio_skip(pb, atom.size);
4364 err = mov_read_mdat(c, pb, atom);
4368 static int mov_read_cmov(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4373 uint8_t *moov_data; /* uncompressed data */
4374 long cmov_len, moov_len;
4377 avio_rb32(pb); /* dcom atom */
4378 if (avio_rl32(pb) != MKTAG('d','c','o','m'))
4379 return AVERROR_INVALIDDATA;
4380 if (avio_rl32(pb) != MKTAG('z','l','i','b')) {
4381 av_log(c->fc, AV_LOG_ERROR, "unknown compression for cmov atom !\n");
4382 return AVERROR_INVALIDDATA;
4384 avio_rb32(pb); /* cmvd atom */
4385 if (avio_rl32(pb) != MKTAG('c','m','v','d'))
4386 return AVERROR_INVALIDDATA;
4387 moov_len = avio_rb32(pb); /* uncompressed size */
4388 cmov_len = atom.size - 6 * 4;
4390 cmov_data = av_malloc(cmov_len);
4392 return AVERROR(ENOMEM);
4393 moov_data = av_malloc(moov_len);
4396 return AVERROR(ENOMEM);
4398 ret = ffio_read_size(pb, cmov_data, cmov_len);
4400 goto free_and_return;
4402 if (uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
4403 goto free_and_return;
4404 if (ffio_init_context(&ctx, moov_data, moov_len, 0, NULL, NULL, NULL, NULL) != 0)
4405 goto free_and_return;
4406 ctx.seekable = AVIO_SEEKABLE_NORMAL;
4407 atom.type = MKTAG('m','o','o','v');
4408 atom.size = moov_len;
4409 ret = mov_read_default(c, &ctx, atom);
4415 av_log(c->fc, AV_LOG_ERROR, "this file requires zlib support compiled in\n");
4416 return AVERROR(ENOSYS);
4420 /* edit list atom */
4421 static int mov_read_elst(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4423 MOVStreamContext *sc;
4424 int i, edit_count, version;
4426 if (c->fc->nb_streams < 1 || c->ignore_editlist)
4428 sc = c->fc->streams[c->fc->nb_streams-1]->priv_data;
4430 version = avio_r8(pb); /* version */
4431 avio_rb24(pb); /* flags */
4432 edit_count = avio_rb32(pb); /* entries */
4437 av_log(c->fc, AV_LOG_WARNING, "Duplicated ELST atom\n");
4438 av_free(sc->elst_data);
4440 sc->elst_data = av_malloc_array(edit_count, sizeof(*sc->elst_data));
4442 return AVERROR(ENOMEM);
4444 av_log(c->fc, AV_LOG_TRACE, "track[%i].edit_count = %i\n", c->fc->nb_streams-1, edit_count);
4445 for (i = 0; i < edit_count && !pb->eof_reached; i++) {
4446 MOVElst *e = &sc->elst_data[i];
4449 e->duration = avio_rb64(pb);
4450 e->time = avio_rb64(pb);
4452 e->duration = avio_rb32(pb); /* segment duration */
4453 e->time = (int32_t)avio_rb32(pb); /* media time */
4455 e->rate = avio_rb32(pb) / 65536.0;
4456 av_log(c->fc, AV_LOG_TRACE, "duration=%"PRId64" time=%"PRId64" rate=%f\n",
4457 e->duration, e->time, e->rate);
4459 if (e->time < 0 && e->time != -1 &&
4460 c->fc->strict_std_compliance >= FF_COMPLIANCE_STRICT) {
4461 av_log(c->fc, AV_LOG_ERROR, "Track %d, edit %d: Invalid edit list media time=%"PRId64"\n",
4462 c->fc->nb_streams-1, i, e->time);
4463 return AVERROR_INVALIDDATA;
4471 static int mov_read_tmcd(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4473 MOVStreamContext *sc;
4475 if (c->fc->nb_streams < 1)
4476 return AVERROR_INVALIDDATA;
4477 sc = c->fc->streams[c->fc->nb_streams - 1]->priv_data;
4478 sc->timecode_track = avio_rb32(pb);
4482 static int mov_read_uuid(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4486 static const uint8_t uuid_isml_manifest[] = {
4487 0xa5, 0xd4, 0x0b, 0x30, 0xe8, 0x14, 0x11, 0xdd,
4488 0xba, 0x2f, 0x08, 0x00, 0x20, 0x0c, 0x9a, 0x66
4490 static const uint8_t uuid_xmp[] = {
4491 0xbe, 0x7a, 0xcf, 0xcb, 0x97, 0xa9, 0x42, 0xe8,
4492 0x9c, 0x71, 0x99, 0x94, 0x91, 0xe3, 0xaf, 0xac
4495 if (atom.size < sizeof(uuid) || atom.size == INT64_MAX)
4496 return AVERROR_INVALIDDATA;
4498 ret = avio_read(pb, uuid, sizeof(uuid));
4501 } else if (ret != sizeof(uuid)) {
4502 return AVERROR_INVALIDDATA;
4504 if (!memcmp(uuid, uuid_isml_manifest, sizeof(uuid))) {
4505 uint8_t *buffer, *ptr;
4507 size_t len = atom.size - sizeof(uuid);
4510 return AVERROR_INVALIDDATA;
4512 ret = avio_skip(pb, 4); // zeroes
4515 buffer = av_mallocz(len + 1);
4517 return AVERROR(ENOMEM);
4519 ret = avio_read(pb, buffer, len);
4523 } else if (ret != len) {
4525 return AVERROR_INVALIDDATA;
4529 while ((ptr = av_stristr(ptr, "systemBitrate=\""))) {
4530 ptr += sizeof("systemBitrate=\"") - 1;
4531 c->bitrates_count++;
4532 c->bitrates = av_realloc_f(c->bitrates, c->bitrates_count, sizeof(*c->bitrates));
4534 c->bitrates_count = 0;
4536 return AVERROR(ENOMEM);
4539 ret = strtol(ptr, &endptr, 10);
4540 if (ret < 0 || errno || *endptr != '"') {
4541 c->bitrates[c->bitrates_count - 1] = 0;
4543 c->bitrates[c->bitrates_count - 1] = ret;
4548 } else if (!memcmp(uuid, uuid_xmp, sizeof(uuid))) {
4550 size_t len = atom.size - sizeof(uuid);
4552 buffer = av_mallocz(len + 1);
4554 return AVERROR(ENOMEM);
4556 ret = avio_read(pb, buffer, len);
4560 } else if (ret != len) {
4562 return AVERROR_INVALIDDATA;
4564 if (c->export_xmp) {
4566 av_dict_set(&c->fc->metadata, "xmp", buffer, 0);
4573 static int mov_read_free(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4576 uint8_t content[16];
4581 ret = avio_read(pb, content, FFMIN(sizeof(content), atom.size));
4587 && !memcmp(content, "Anevia\x1A\x1A", 8)
4588 && c->use_mfra_for == FF_MOV_FLAG_MFRA_AUTO) {
4589 c->use_mfra_for = FF_MOV_FLAG_MFRA_PTS;
4595 static int mov_read_frma(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4597 uint32_t format = avio_rl32(pb);
4598 MOVStreamContext *sc;
4602 if (c->fc->nb_streams < 1)
4604 st = c->fc->streams[c->fc->nb_streams - 1];
4609 case MKTAG('e','n','c','v'): // encrypted video
4610 case MKTAG('e','n','c','a'): // encrypted audio
4611 id = mov_codec_id(st, format);
4612 if (st->codecpar->codec_id != AV_CODEC_ID_NONE &&
4613 st->codecpar->codec_id != id) {
4614 av_log(c->fc, AV_LOG_WARNING,
4615 "ignoring 'frma' atom of '%.4s', stream has codec id %d\n",
4616 (char*)&format, st->codecpar->codec_id);
4620 st->codecpar->codec_id = id;
4621 sc->format = format;
4625 if (format != sc->format) {
4626 av_log(c->fc, AV_LOG_WARNING,
4627 "ignoring 'frma' atom of '%.4s', stream format is '%.4s'\n",
4628 (char*)&format, (char*)&sc->format);
4636 static int mov_read_senc(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4639 MOVStreamContext *sc;
4640 size_t auxiliary_info_size;
4642 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
4645 st = c->fc->streams[c->fc->nb_streams - 1];
4648 if (sc->cenc.aes_ctr) {
4649 av_log(c->fc, AV_LOG_ERROR, "duplicate senc atom\n");
4650 return AVERROR_INVALIDDATA;
4653 avio_r8(pb); /* version */
4654 sc->cenc.use_subsamples = avio_rb24(pb) & 0x02; /* flags */
4656 avio_rb32(pb); /* entries */
4658 if (atom.size < 8) {
4659 av_log(c->fc, AV_LOG_ERROR, "senc atom size %"PRId64" too small\n", atom.size);
4660 return AVERROR_INVALIDDATA;
4663 /* save the auxiliary info as is */
4664 auxiliary_info_size = atom.size - 8;
4666 sc->cenc.auxiliary_info = av_malloc(auxiliary_info_size);
4667 if (!sc->cenc.auxiliary_info) {
4668 return AVERROR(ENOMEM);
4671 sc->cenc.auxiliary_info_end = sc->cenc.auxiliary_info + auxiliary_info_size;
4673 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info;
4675 if (avio_read(pb, sc->cenc.auxiliary_info, auxiliary_info_size) != auxiliary_info_size) {
4676 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info");
4677 return AVERROR_INVALIDDATA;
4680 /* initialize the cipher */
4681 sc->cenc.aes_ctr = av_aes_ctr_alloc();
4682 if (!sc->cenc.aes_ctr) {
4683 return AVERROR(ENOMEM);
4686 return av_aes_ctr_init(sc->cenc.aes_ctr, c->decryption_key);
4689 static int mov_read_saiz(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4692 MOVStreamContext *sc;
4694 int atom_header_size;
4697 if (c->decryption_key_len == 0 || c->fc->nb_streams < 1)
4700 st = c->fc->streams[c->fc->nb_streams - 1];
4703 if (sc->cenc.auxiliary_info_sizes || sc->cenc.auxiliary_info_default_size) {
4704 av_log(c->fc, AV_LOG_ERROR, "duplicate saiz atom\n");
4705 return AVERROR_INVALIDDATA;
4708 atom_header_size = 9;
4710 avio_r8(pb); /* version */
4711 flags = avio_rb24(pb);
4713 if ((flags & 0x01) != 0) {
4714 atom_header_size += 8;
4716 avio_rb32(pb); /* info type */
4717 avio_rb32(pb); /* info type param */
4720 sc->cenc.auxiliary_info_default_size = avio_r8(pb);
4721 avio_rb32(pb); /* entries */
4723 if (atom.size <= atom_header_size) {
4727 /* save the auxiliary info sizes as is */
4728 data_size = atom.size - atom_header_size;
4730 sc->cenc.auxiliary_info_sizes = av_malloc(data_size);
4731 if (!sc->cenc.auxiliary_info_sizes) {
4732 return AVERROR(ENOMEM);
4735 sc->cenc.auxiliary_info_sizes_count = data_size;
4737 if (avio_read(pb, sc->cenc.auxiliary_info_sizes, data_size) != data_size) {
4738 av_log(c->fc, AV_LOG_ERROR, "failed to read the auxiliary info sizes");
4739 return AVERROR_INVALIDDATA;
4745 static int cenc_filter(MOVContext *c, MOVStreamContext *sc, uint8_t *input, int size)
4747 uint32_t encrypted_bytes;
4748 uint16_t subsample_count;
4749 uint16_t clear_bytes;
4750 uint8_t* input_end = input + size;
4753 if (AES_CTR_IV_SIZE > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
4754 av_log(c->fc, AV_LOG_ERROR, "failed to read iv from the auxiliary info\n");
4755 return AVERROR_INVALIDDATA;
4758 av_aes_ctr_set_iv(sc->cenc.aes_ctr, sc->cenc.auxiliary_info_pos);
4759 sc->cenc.auxiliary_info_pos += AES_CTR_IV_SIZE;
4761 if (!sc->cenc.use_subsamples)
4763 /* decrypt the whole packet */
4764 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, size);
4768 /* read the subsample count */
4769 if (sizeof(uint16_t) > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
4770 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample count from the auxiliary info\n");
4771 return AVERROR_INVALIDDATA;
4774 subsample_count = AV_RB16(sc->cenc.auxiliary_info_pos);
4775 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
4777 for (; subsample_count > 0; subsample_count--)
4779 if (6 > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info_pos) {
4780 av_log(c->fc, AV_LOG_ERROR, "failed to read subsample from the auxiliary info\n");
4781 return AVERROR_INVALIDDATA;
4784 /* read the number of clear / encrypted bytes */
4785 clear_bytes = AV_RB16(sc->cenc.auxiliary_info_pos);
4786 sc->cenc.auxiliary_info_pos += sizeof(uint16_t);
4787 encrypted_bytes = AV_RB32(sc->cenc.auxiliary_info_pos);
4788 sc->cenc.auxiliary_info_pos += sizeof(uint32_t);
4790 if ((uint64_t)clear_bytes + encrypted_bytes > input_end - input) {
4791 av_log(c->fc, AV_LOG_ERROR, "subsample size exceeds the packet size left\n");
4792 return AVERROR_INVALIDDATA;
4795 /* skip the clear bytes */
4796 input += clear_bytes;
4798 /* decrypt the encrypted bytes */
4799 av_aes_ctr_crypt(sc->cenc.aes_ctr, input, input, encrypted_bytes);
4800 input += encrypted_bytes;
4803 if (input < input_end) {
4804 av_log(c->fc, AV_LOG_ERROR, "leftover packet bytes after subsample processing\n");
4805 return AVERROR_INVALIDDATA;
4811 static int mov_seek_auxiliary_info(AVFormatContext *s, MOVStreamContext *sc)
4813 size_t auxiliary_info_seek_offset = 0;
4816 if (sc->cenc.auxiliary_info_default_size) {
4817 auxiliary_info_seek_offset = (size_t)sc->cenc.auxiliary_info_default_size * sc->current_sample;
4818 } else if (sc->cenc.auxiliary_info_sizes) {
4819 if (sc->current_sample > sc->cenc.auxiliary_info_sizes_count) {
4820 av_log(s, AV_LOG_ERROR, "current sample %d greater than the number of auxiliary info sample sizes %"SIZE_SPECIFIER"\n",
4821 sc->current_sample, sc->cenc.auxiliary_info_sizes_count);
4822 return AVERROR_INVALIDDATA;
4825 for (i = 0; i < sc->current_sample; i++) {
4826 auxiliary_info_seek_offset += sc->cenc.auxiliary_info_sizes[i];
4830 if (auxiliary_info_seek_offset > sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info) {
4831 av_log(s, AV_LOG_ERROR, "auxiliary info offset %"SIZE_SPECIFIER" greater than auxiliary info size %"SIZE_SPECIFIER"\n",
4832 auxiliary_info_seek_offset, (size_t)(sc->cenc.auxiliary_info_end - sc->cenc.auxiliary_info));
4833 return AVERROR_INVALIDDATA;
4836 sc->cenc.auxiliary_info_pos = sc->cenc.auxiliary_info + auxiliary_info_seek_offset;
4841 static const MOVParseTableEntry mov_default_parse_table[] = {
4842 { MKTAG('A','C','L','R'), mov_read_aclr },
4843 { MKTAG('A','P','R','G'), mov_read_avid },
4844 { MKTAG('A','A','L','P'), mov_read_avid },
4845 { MKTAG('A','R','E','S'), mov_read_ares },
4846 { MKTAG('a','v','s','s'), mov_read_avss },
4847 { MKTAG('c','h','p','l'), mov_read_chpl },
4848 { MKTAG('c','o','6','4'), mov_read_stco },
4849 { MKTAG('c','o','l','r'), mov_read_colr },
4850 { MKTAG('c','t','t','s'), mov_read_ctts }, /* composition time to sample */
4851 { MKTAG('d','i','n','f'), mov_read_default },
4852 { MKTAG('D','p','x','E'), mov_read_dpxe },
4853 { MKTAG('d','r','e','f'), mov_read_dref },
4854 { MKTAG('e','d','t','s'), mov_read_default },
4855 { MKTAG('e','l','s','t'), mov_read_elst },
4856 { MKTAG('e','n','d','a'), mov_read_enda },
4857 { MKTAG('f','i','e','l'), mov_read_fiel },
4858 { MKTAG('a','d','r','m'), mov_read_adrm },
4859 { MKTAG('f','t','y','p'), mov_read_ftyp },
4860 { MKTAG('g','l','b','l'), mov_read_glbl },
4861 { MKTAG('h','d','l','r'), mov_read_hdlr },
4862 { MKTAG('i','l','s','t'), mov_read_ilst },
4863 { MKTAG('j','p','2','h'), mov_read_jp2h },
4864 { MKTAG('m','d','a','t'), mov_read_mdat },
4865 { MKTAG('m','d','h','d'), mov_read_mdhd },
4866 { MKTAG('m','d','i','a'), mov_read_default },
4867 { MKTAG('m','e','t','a'), mov_read_meta },
4868 { MKTAG('m','i','n','f'), mov_read_default },
4869 { MKTAG('m','o','o','f'), mov_read_moof },
4870 { MKTAG('m','o','o','v'), mov_read_moov },
4871 { MKTAG('m','v','e','x'), mov_read_default },
4872 { MKTAG('m','v','h','d'), mov_read_mvhd },
4873 { MKTAG('S','M','I',' '), mov_read_svq3 },
4874 { MKTAG('a','l','a','c'), mov_read_alac }, /* alac specific atom */
4875 { MKTAG('a','v','c','C'), mov_read_glbl },
4876 { MKTAG('p','a','s','p'), mov_read_pasp },
4877 { MKTAG('s','i','d','x'), mov_read_sidx },
4878 { MKTAG('s','t','b','l'), mov_read_default },
4879 { MKTAG('s','t','c','o'), mov_read_stco },
4880 { MKTAG('s','t','p','s'), mov_read_stps },
4881 { MKTAG('s','t','r','f'), mov_read_strf },
4882 { MKTAG('s','t','s','c'), mov_read_stsc },
4883 { MKTAG('s','t','s','d'), mov_read_stsd }, /* sample description */
4884 { MKTAG('s','t','s','s'), mov_read_stss }, /* sync sample */
4885 { MKTAG('s','t','s','z'), mov_read_stsz }, /* sample size */
4886 { MKTAG('s','t','t','s'), mov_read_stts },
4887 { MKTAG('s','t','z','2'), mov_read_stsz }, /* compact sample size */
4888 { MKTAG('t','k','h','d'), mov_read_tkhd }, /* track header */
4889 { MKTAG('t','f','d','t'), mov_read_tfdt },
4890 { MKTAG('t','f','h','d'), mov_read_tfhd }, /* track fragment header */
4891 { MKTAG('t','r','a','k'), mov_read_trak },
4892 { MKTAG('t','r','a','f'), mov_read_default },
4893 { MKTAG('t','r','e','f'), mov_read_default },
4894 { MKTAG('t','m','c','d'), mov_read_tmcd },
4895 { MKTAG('c','h','a','p'), mov_read_chap },
4896 { MKTAG('t','r','e','x'), mov_read_trex },
4897 { MKTAG('t','r','u','n'), mov_read_trun },
4898 { MKTAG('u','d','t','a'), mov_read_default },
4899 { MKTAG('w','a','v','e'), mov_read_wave },
4900 { MKTAG('e','s','d','s'), mov_read_esds },
4901 { MKTAG('d','a','c','3'), mov_read_dac3 }, /* AC-3 info */
4902 { MKTAG('d','e','c','3'), mov_read_dec3 }, /* EAC-3 info */
4903 { MKTAG('d','d','t','s'), mov_read_ddts }, /* DTS audio descriptor */
4904 { MKTAG('w','i','d','e'), mov_read_wide }, /* place holder */
4905 { MKTAG('w','f','e','x'), mov_read_wfex },
4906 { MKTAG('c','m','o','v'), mov_read_cmov },
4907 { MKTAG('c','h','a','n'), mov_read_chan }, /* channel layout */
4908 { MKTAG('d','v','c','1'), mov_read_dvc1 },
4909 { MKTAG('s','b','g','p'), mov_read_sbgp },
4910 { MKTAG('h','v','c','C'), mov_read_glbl },
4911 { MKTAG('u','u','i','d'), mov_read_uuid },
4912 { MKTAG('C','i','n', 0x8e), mov_read_targa_y216 },
4913 { MKTAG('f','r','e','e'), mov_read_free },
4914 { MKTAG('-','-','-','-'), mov_read_custom },
4915 { MKTAG('s','i','n','f'), mov_read_default },
4916 { MKTAG('f','r','m','a'), mov_read_frma },
4917 { MKTAG('s','e','n','c'), mov_read_senc },
4918 { MKTAG('s','a','i','z'), mov_read_saiz },
4922 static int mov_read_default(MOVContext *c, AVIOContext *pb, MOVAtom atom)
4924 int64_t total_size = 0;
4928 if (c->atom_depth > 10) {
4929 av_log(c->fc, AV_LOG_ERROR, "Atoms too deeply nested\n");
4930 return AVERROR_INVALIDDATA;
4935 atom.size = INT64_MAX;
4936 while (total_size + 8 <= atom.size && !avio_feof(pb)) {
4937 int (*parse)(MOVContext*, AVIOContext*, MOVAtom) = NULL;
4940 if (atom.size >= 8) {
4941 a.size = avio_rb32(pb);
4942 a.type = avio_rl32(pb);
4943 if (a.type == MKTAG('f','r','e','e') &&
4945 c->fc->strict_std_compliance < FF_COMPLIANCE_STRICT &&
4948 uint32_t *type = (uint32_t *)buf + 1;
4949 if (avio_read(pb, buf, 8) != 8)
4950 return AVERROR_INVALIDDATA;
4951 avio_seek(pb, -8, SEEK_CUR);
4952 if (*type == MKTAG('m','v','h','d') ||
4953 *type == MKTAG('c','m','o','v')) {
4954 av_log(c->fc, AV_LOG_ERROR, "Detected moov in a free atom.\n");
4955 a.type = MKTAG('m','o','o','v');
4958 if (atom.type != MKTAG('r','o','o','t') &&
4959 atom.type != MKTAG('m','o','o','v'))
4961 if (a.type == MKTAG('t','r','a','k') || a.type == MKTAG('m','d','a','t'))
4963 av_log(c->fc, AV_LOG_ERROR, "Broken file, trak/mdat not at top-level\n");
4970 if (a.size == 1 && total_size + 8 <= atom.size) { /* 64 bit extended size */
4971 a.size = avio_rb64(pb) - 8;
4975 av_log(c->fc, AV_LOG_TRACE, "type: %08x '%.4s' parent:'%.4s' sz: %"PRId64" %"PRId64" %"PRId64"\n",
4976 a.type, (char*)&a.type, (char*)&atom.type, a.size, total_size, atom.size);
4978 a.size = atom.size - total_size + 8;
4983 a.size = FFMIN(a.size, atom.size - total_size);
4985 for (i = 0; mov_default_parse_table[i].type; i++)
4986 if (mov_default_parse_table[i].type == a.type) {
4987 parse = mov_default_parse_table[i].parse;
4991 // container is user data
4992 if (!parse && (atom.type == MKTAG('u','d','t','a') ||
4993 atom.type == MKTAG('i','l','s','t')))
4994 parse = mov_read_udta_string;
4996 // Supports parsing the QuickTime Metadata Keys.
4997 // https://developer.apple.com/library/mac/documentation/QuickTime/QTFF/Metadata/Metadata.html
4998 if (!parse && c->found_hdlr_mdta &&
4999 atom.type == MKTAG('m','e','t','a') &&
5000 a.type == MKTAG('k','e','y','s')) {
5001 parse = mov_read_keys;
5004 if (!parse) { /* skip leaf atoms data */
5005 avio_skip(pb, a.size);
5007 int64_t start_pos = avio_tell(pb);
5009 int err = parse(c, pb, a);
5014 if (c->found_moov && c->found_mdat &&
5015 ((!pb->seekable || c->fc->flags & AVFMT_FLAG_IGNIDX || c->fragment_index_complete) ||
5016 start_pos + a.size == avio_size(pb))) {
5017 if (!pb->seekable || c->fc->flags & AVFMT_FLAG_IGNIDX || c->fragment_index_complete)
5018 c->next_root_atom = start_pos + a.size;
5022 left = a.size - avio_tell(pb) + start_pos;
5023 if (left > 0) /* skip garbage at atom end */
5024 avio_skip(pb, left);
5025 else if (left < 0) {
5026 av_log(c->fc, AV_LOG_WARNING,
5027 "overread end of atom '%.4s' by %"PRId64" bytes\n",
5028 (char*)&a.type, -left);
5029 avio_seek(pb, left, SEEK_CUR);
5033 total_size += a.size;
5036 if (total_size < atom.size && atom.size < 0x7ffff)
5037 avio_skip(pb, atom.size - total_size);
5043 static int mov_probe(AVProbeData *p)
5048 int moov_offset = -1;
5050 /* check file header */
5053 /* ignore invalid offset */
5054 if ((offset + 8) > (unsigned int)p->buf_size)
5056 tag = AV_RL32(p->buf + offset + 4);
5058 /* check for obvious tags */
5059 case MKTAG('m','o','o','v'):
5060 moov_offset = offset + 4;
5061 case MKTAG('m','d','a','t'):
5062 case MKTAG('p','n','o','t'): /* detect movs with preview pics like ew.mov and april.mov */
5063 case MKTAG('u','d','t','a'): /* Packet Video PVAuthor adds this and a lot of more junk */
5064 case MKTAG('f','t','y','p'):
5065 if (AV_RB32(p->buf+offset) < 8 &&
5066 (AV_RB32(p->buf+offset) != 1 ||
5067 offset + 12 > (unsigned int)p->buf_size ||
5068 AV_RB64(p->buf+offset + 8) == 0)) {
5069 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
5070 } else if (tag == MKTAG('f','t','y','p') &&
5071 ( AV_RL32(p->buf + offset + 8) == MKTAG('j','p','2',' ')
5072 || AV_RL32(p->buf + offset + 8) == MKTAG('j','p','x',' ')
5074 score = FFMAX(score, 5);
5076 score = AVPROBE_SCORE_MAX;
5078 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5080 /* those are more common words, so rate then a bit less */
5081 case MKTAG('e','d','i','w'): /* xdcam files have reverted first tags */
5082 case MKTAG('w','i','d','e'):
5083 case MKTAG('f','r','e','e'):
5084 case MKTAG('j','u','n','k'):
5085 case MKTAG('p','i','c','t'):
5086 score = FFMAX(score, AVPROBE_SCORE_MAX - 5);
5087 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5089 case MKTAG(0x82,0x82,0x7f,0x7d):
5090 case MKTAG('s','k','i','p'):
5091 case MKTAG('u','u','i','d'):
5092 case MKTAG('p','r','f','l'):
5093 /* if we only find those cause probedata is too small at least rate them */
5094 score = FFMAX(score, AVPROBE_SCORE_EXTENSION);
5095 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5098 offset = FFMAX(4, AV_RB32(p->buf+offset)) + offset;
5101 if(score > AVPROBE_SCORE_MAX - 50 && moov_offset != -1) {
5102 /* moov atom in the header - we should make sure that this is not a
5103 * MOV-packed MPEG-PS */
5104 offset = moov_offset;
5106 while(offset < (p->buf_size - 16)){ /* Sufficient space */
5107 /* We found an actual hdlr atom */
5108 if(AV_RL32(p->buf + offset ) == MKTAG('h','d','l','r') &&
5109 AV_RL32(p->buf + offset + 8) == MKTAG('m','h','l','r') &&
5110 AV_RL32(p->buf + offset + 12) == MKTAG('M','P','E','G')){
5111 av_log(NULL, AV_LOG_WARNING, "Found media data tag MPEG indicating this is a MOV-packed MPEG-PS.\n");
5112 /* We found a media handler reference atom describing an
5113 * MPEG-PS-in-MOV, return a
5114 * low score to force expanding the probe window until
5115 * mpegps_probe finds what it needs */
5126 // must be done after parsing all trak because there's no order requirement
5127 static void mov_read_chapters(AVFormatContext *s)
5129 MOVContext *mov = s->priv_data;
5131 MOVStreamContext *sc;
5136 for (j = 0; j < mov->nb_chapter_tracks; j++) {
5137 chapter_track = mov->chapter_tracks[j];
5139 for (i = 0; i < s->nb_streams; i++)
5140 if (s->streams[i]->id == chapter_track) {
5145 av_log(s, AV_LOG_ERROR, "Referenced QT chapter track not found\n");
5150 cur_pos = avio_tell(sc->pb);
5152 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
5153 st->disposition |= AV_DISPOSITION_ATTACHED_PIC | AV_DISPOSITION_TIMED_THUMBNAILS;
5154 if (st->nb_index_entries) {
5155 // Retrieve the first frame, if possible
5157 AVIndexEntry *sample = &st->index_entries[0];
5158 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
5159 av_log(s, AV_LOG_ERROR, "Failed to retrieve first frame\n");
5163 if (av_get_packet(sc->pb, &pkt, sample->size) < 0)
5166 st->attached_pic = pkt;
5167 st->attached_pic.stream_index = st->index;
5168 st->attached_pic.flags |= AV_PKT_FLAG_KEY;
5171 st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
5172 st->codecpar->codec_id = AV_CODEC_ID_BIN_DATA;
5173 st->discard = AVDISCARD_ALL;
5174 for (i = 0; i < st->nb_index_entries; i++) {
5175 AVIndexEntry *sample = &st->index_entries[i];
5176 int64_t end = i+1 < st->nb_index_entries ? st->index_entries[i+1].timestamp : st->duration;
5181 if (end < sample->timestamp) {
5182 av_log(s, AV_LOG_WARNING, "ignoring stream duration which is shorter than chapters\n");
5183 end = AV_NOPTS_VALUE;
5186 if (avio_seek(sc->pb, sample->pos, SEEK_SET) != sample->pos) {
5187 av_log(s, AV_LOG_ERROR, "Chapter %d not found in file\n", i);
5191 // the first two bytes are the length of the title
5192 len = avio_rb16(sc->pb);
5193 if (len > sample->size-2)
5195 title_len = 2*len + 1;
5196 if (!(title = av_mallocz(title_len)))
5199 // The samples could theoretically be in any encoding if there's an encd
5200 // atom following, but in practice are only utf-8 or utf-16, distinguished
5201 // instead by the presence of a BOM
5205 ch = avio_rb16(sc->pb);
5207 avio_get_str16be(sc->pb, len, title, title_len);
5208 else if (ch == 0xfffe)
5209 avio_get_str16le(sc->pb, len, title, title_len);
5212 if (len == 1 || len == 2)
5215 avio_get_str(sc->pb, INT_MAX, title + 2, len - 1);
5219 avpriv_new_chapter(s, i, st->time_base, sample->timestamp, end, title);
5224 avio_seek(sc->pb, cur_pos, SEEK_SET);
5228 static int parse_timecode_in_framenum_format(AVFormatContext *s, AVStream *st,
5229 uint32_t value, int flags)
5232 char buf[AV_TIMECODE_STR_SIZE];
5233 AVRational rate = st->avg_frame_rate;
5234 int ret = av_timecode_init(&tc, rate, flags, 0, s);
5237 av_dict_set(&st->metadata, "timecode",
5238 av_timecode_make_string(&tc, buf, value), 0);
5242 static int mov_read_rtmd_track(AVFormatContext *s, AVStream *st)
5244 MOVStreamContext *sc = st->priv_data;
5245 char buf[AV_TIMECODE_STR_SIZE];
5246 int64_t cur_pos = avio_tell(sc->pb);
5247 int hh, mm, ss, ff, drop;
5249 if (!st->nb_index_entries)
5252 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
5253 avio_skip(s->pb, 13);
5254 hh = avio_r8(s->pb);
5255 mm = avio_r8(s->pb);
5256 ss = avio_r8(s->pb);
5257 drop = avio_r8(s->pb);
5258 ff = avio_r8(s->pb);
5259 snprintf(buf, AV_TIMECODE_STR_SIZE, "%02d:%02d:%02d%c%02d",
5260 hh, mm, ss, drop ? ';' : ':', ff);
5261 av_dict_set(&st->metadata, "timecode", buf, 0);
5263 avio_seek(sc->pb, cur_pos, SEEK_SET);
5267 static int mov_read_timecode_track(AVFormatContext *s, AVStream *st)
5269 MOVStreamContext *sc = st->priv_data;
5271 int64_t cur_pos = avio_tell(sc->pb);
5274 if (!st->nb_index_entries)
5277 avio_seek(sc->pb, st->index_entries->pos, SEEK_SET);
5278 value = avio_rb32(s->pb);
5280 if (sc->tmcd_flags & 0x0001) flags |= AV_TIMECODE_FLAG_DROPFRAME;
5281 if (sc->tmcd_flags & 0x0002) flags |= AV_TIMECODE_FLAG_24HOURSMAX;
5282 if (sc->tmcd_flags & 0x0004) flags |= AV_TIMECODE_FLAG_ALLOWNEGATIVE;
5284 /* Assume Counter flag is set to 1 in tmcd track (even though it is likely
5285 * not the case) and thus assume "frame number format" instead of QT one.
5286 * No sample with tmcd track can be found with a QT timecode at the moment,
5287 * despite what the tmcd track "suggests" (Counter flag set to 0 means QT
5289 parse_timecode_in_framenum_format(s, st, value, flags);
5291 avio_seek(sc->pb, cur_pos, SEEK_SET);
5295 static int mov_read_close(AVFormatContext *s)
5297 MOVContext *mov = s->priv_data;
5300 for (i = 0; i < s->nb_streams; i++) {
5301 AVStream *st = s->streams[i];
5302 MOVStreamContext *sc = st->priv_data;
5307 av_freep(&sc->ctts_data);
5308 for (j = 0; j < sc->drefs_count; j++) {
5309 av_freep(&sc->drefs[j].path);
5310 av_freep(&sc->drefs[j].dir);
5312 av_freep(&sc->drefs);
5314 sc->drefs_count = 0;
5316 if (!sc->pb_is_copied)
5317 ff_format_io_close(s, &sc->pb);
5320 av_freep(&sc->chunk_offsets);
5321 av_freep(&sc->stsc_data);
5322 av_freep(&sc->sample_sizes);
5323 av_freep(&sc->keyframes);
5324 av_freep(&sc->stts_data);
5325 av_freep(&sc->stps_data);
5326 av_freep(&sc->elst_data);
5327 av_freep(&sc->rap_group);
5328 av_freep(&sc->display_matrix);
5331 for (j = 0; j < sc->stsd_count; j++)
5332 av_free(sc->extradata[j]);
5333 av_freep(&sc->extradata);
5334 av_freep(&sc->extradata_size);
5336 av_freep(&sc->cenc.auxiliary_info);
5337 av_freep(&sc->cenc.auxiliary_info_sizes);
5338 av_aes_ctr_free(sc->cenc.aes_ctr);
5341 if (mov->dv_demux) {
5342 avformat_free_context(mov->dv_fctx);
5343 mov->dv_fctx = NULL;
5346 if (mov->meta_keys) {
5347 for (i = 1; i < mov->meta_keys_count; i++) {
5348 av_freep(&mov->meta_keys[i]);
5350 av_freep(&mov->meta_keys);
5353 av_freep(&mov->trex_data);
5354 av_freep(&mov->bitrates);
5356 for (i = 0; i < mov->fragment_index_count; i++) {
5357 MOVFragmentIndex* index = mov->fragment_index_data[i];
5358 av_freep(&index->items);
5359 av_freep(&mov->fragment_index_data[i]);
5361 av_freep(&mov->fragment_index_data);
5363 av_freep(&mov->aes_decrypt);
5368 static int tmcd_is_referenced(AVFormatContext *s, int tmcd_id)
5372 for (i = 0; i < s->nb_streams; i++) {
5373 AVStream *st = s->streams[i];
5374 MOVStreamContext *sc = st->priv_data;
5376 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO &&
5377 sc->timecode_track == tmcd_id)
5383 /* look for a tmcd track not referenced by any video track, and export it globally */
5384 static void export_orphan_timecode(AVFormatContext *s)
5388 for (i = 0; i < s->nb_streams; i++) {
5389 AVStream *st = s->streams[i];
5391 if (st->codecpar->codec_tag == MKTAG('t','m','c','d') &&
5392 !tmcd_is_referenced(s, i + 1)) {
5393 AVDictionaryEntry *tcr = av_dict_get(st->metadata, "timecode", NULL, 0);
5395 av_dict_set(&s->metadata, "timecode", tcr->value, 0);
5402 static int read_tfra(MOVContext *mov, AVIOContext *f)
5404 MOVFragmentIndex* index = NULL;
5405 int version, fieldlength, i, j;
5406 int64_t pos = avio_tell(f);
5407 uint32_t size = avio_rb32(f);
5410 if (avio_rb32(f) != MKBETAG('t', 'f', 'r', 'a')) {
5413 av_log(mov->fc, AV_LOG_VERBOSE, "found tfra\n");
5414 index = av_mallocz(sizeof(MOVFragmentIndex));
5416 return AVERROR(ENOMEM);
5419 tmp = av_realloc_array(mov->fragment_index_data,
5420 mov->fragment_index_count + 1,
5421 sizeof(MOVFragmentIndex*));
5424 return AVERROR(ENOMEM);
5426 mov->fragment_index_data = tmp;
5427 mov->fragment_index_data[mov->fragment_index_count++] = index;
5429 version = avio_r8(f);
5431 index->track_id = avio_rb32(f);
5432 fieldlength = avio_rb32(f);
5433 index->item_count = avio_rb32(f);
5434 index->items = av_mallocz_array(
5435 index->item_count, sizeof(MOVFragmentIndexItem));
5436 if (!index->items) {
5437 index->item_count = 0;
5438 return AVERROR(ENOMEM);
5440 for (i = 0; i < index->item_count; i++) {
5441 int64_t time, offset;
5443 time = avio_rb64(f);
5444 offset = avio_rb64(f);
5446 time = avio_rb32(f);
5447 offset = avio_rb32(f);
5449 index->items[i].time = time;
5450 index->items[i].moof_offset = offset;
5451 for (j = 0; j < ((fieldlength >> 4) & 3) + 1; j++)
5453 for (j = 0; j < ((fieldlength >> 2) & 3) + 1; j++)
5455 for (j = 0; j < ((fieldlength >> 0) & 3) + 1; j++)
5459 avio_seek(f, pos + size, SEEK_SET);
5463 static int mov_read_mfra(MOVContext *c, AVIOContext *f)
5465 int64_t stream_size = avio_size(f);
5466 int64_t original_pos = avio_tell(f);
5470 if ((seek_ret = avio_seek(f, stream_size - 4, SEEK_SET)) < 0) {
5474 mfra_size = avio_rb32(f);
5475 if (mfra_size < 0 || mfra_size > stream_size) {
5476 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (unreasonable size)\n");
5479 if ((seek_ret = avio_seek(f, -mfra_size, SEEK_CUR)) < 0) {
5483 if (avio_rb32(f) != mfra_size) {
5484 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (size mismatch)\n");
5487 if (avio_rb32(f) != MKBETAG('m', 'f', 'r', 'a')) {
5488 av_log(c->fc, AV_LOG_DEBUG, "doesn't look like mfra (tag mismatch)\n");
5491 av_log(c->fc, AV_LOG_VERBOSE, "stream has mfra\n");
5493 ret = read_tfra(c, f);
5499 seek_ret = avio_seek(f, original_pos, SEEK_SET);
5501 av_log(c->fc, AV_LOG_ERROR,
5502 "failed to seek back after looking for mfra\n");
5508 static int mov_read_header(AVFormatContext *s)
5510 MOVContext *mov = s->priv_data;
5511 AVIOContext *pb = s->pb;
5513 MOVAtom atom = { AV_RL32("root") };
5516 if (mov->decryption_key_len != 0 && mov->decryption_key_len != AES_CTR_KEY_SIZE) {
5517 av_log(s, AV_LOG_ERROR, "Invalid decryption key len %d expected %d\n",
5518 mov->decryption_key_len, AES_CTR_KEY_SIZE);
5519 return AVERROR(EINVAL);
5523 mov->trak_index = -1;
5524 /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
5526 atom.size = avio_size(pb);
5528 atom.size = INT64_MAX;
5530 /* check MOV header */
5532 if (mov->moov_retry)
5533 avio_seek(pb, 0, SEEK_SET);
5534 if ((err = mov_read_default(mov, pb, atom)) < 0) {
5535 av_log(s, AV_LOG_ERROR, "error reading header\n");
5539 } while (pb->seekable && !mov->found_moov && !mov->moov_retry++);
5540 if (!mov->found_moov) {
5541 av_log(s, AV_LOG_ERROR, "moov atom not found\n");
5543 return AVERROR_INVALIDDATA;
5545 av_log(mov->fc, AV_LOG_TRACE, "on_parse_exit_offset=%"PRId64"\n", avio_tell(pb));
5548 if (mov->nb_chapter_tracks > 0 && !mov->ignore_chapters)
5549 mov_read_chapters(s);
5550 for (i = 0; i < s->nb_streams; i++)
5551 if (s->streams[i]->codecpar->codec_tag == AV_RL32("tmcd")) {
5552 mov_read_timecode_track(s, s->streams[i]);
5553 } else if (s->streams[i]->codecpar->codec_tag == AV_RL32("rtmd")) {
5554 mov_read_rtmd_track(s, s->streams[i]);
5558 /* copy timecode metadata from tmcd tracks to the related video streams */
5559 for (i = 0; i < s->nb_streams; i++) {
5560 AVStream *st = s->streams[i];
5561 MOVStreamContext *sc = st->priv_data;
5562 if (sc->timecode_track > 0) {
5563 AVDictionaryEntry *tcr;
5564 int tmcd_st_id = -1;
5566 for (j = 0; j < s->nb_streams; j++)
5567 if (s->streams[j]->id == sc->timecode_track)
5570 if (tmcd_st_id < 0 || tmcd_st_id == i)
5572 tcr = av_dict_get(s->streams[tmcd_st_id]->metadata, "timecode", NULL, 0);
5574 av_dict_set(&st->metadata, "timecode", tcr->value, 0);
5577 export_orphan_timecode(s);
5579 for (i = 0; i < s->nb_streams; i++) {
5580 AVStream *st = s->streams[i];
5581 MOVStreamContext *sc = st->priv_data;
5582 fix_timescale(mov, sc);
5583 if(st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO && st->codecpar->codec_id == AV_CODEC_ID_AAC) {
5584 st->skip_samples = sc->start_pad;
5586 if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO && sc->nb_frames_for_fps > 0 && sc->duration_for_fps > 0)
5587 av_reduce(&st->avg_frame_rate.num, &st->avg_frame_rate.den,
5588 sc->time_scale*(int64_t)sc->nb_frames_for_fps, sc->duration_for_fps, INT_MAX);
5589 if (st->codecpar->codec_type == AVMEDIA_TYPE_SUBTITLE) {
5590 if (st->codecpar->width <= 0 || st->codecpar->height <= 0) {
5591 st->codecpar->width = sc->width;
5592 st->codecpar->height = sc->height;
5594 if (st->codecpar->codec_id == AV_CODEC_ID_DVD_SUBTITLE) {
5595 if ((err = mov_rewrite_dvd_sub_extradata(st)) < 0)
5599 if (mov->handbrake_version &&
5600 mov->handbrake_version <= 1000000*0 + 1000*10 + 2 && // 0.10.2
5601 st->codecpar->codec_id == AV_CODEC_ID_MP3
5603 av_log(s, AV_LOG_VERBOSE, "Forcing full parsing for mp3 stream\n");
5604 st->need_parsing = AVSTREAM_PARSE_FULL;
5608 if (mov->trex_data) {
5609 for (i = 0; i < s->nb_streams; i++) {
5610 AVStream *st = s->streams[i];
5611 MOVStreamContext *sc = st->priv_data;
5612 if (st->duration > 0)
5613 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale / st->duration;
5617 if (mov->use_mfra_for > 0) {
5618 for (i = 0; i < s->nb_streams; i++) {
5619 AVStream *st = s->streams[i];
5620 MOVStreamContext *sc = st->priv_data;
5621 if (sc->duration_for_fps > 0) {
5622 st->codecpar->bit_rate = sc->data_size * 8 * sc->time_scale /
5623 sc->duration_for_fps;
5628 for (i = 0; i < mov->bitrates_count && i < s->nb_streams; i++) {
5629 if (mov->bitrates[i]) {
5630 s->streams[i]->codecpar->bit_rate = mov->bitrates[i];
5634 ff_rfps_calculate(s);
5636 for (i = 0; i < s->nb_streams; i++) {
5637 AVStream *st = s->streams[i];
5638 MOVStreamContext *sc = st->priv_data;
5640 switch (st->codecpar->codec_type) {
5641 case AVMEDIA_TYPE_AUDIO:
5642 err = ff_replaygain_export(st, s->metadata);
5648 case AVMEDIA_TYPE_VIDEO:
5649 if (sc->display_matrix) {
5650 AVPacketSideData *sd, *tmp;
5652 tmp = av_realloc_array(st->side_data,
5653 st->nb_side_data + 1, sizeof(*tmp));
5655 return AVERROR(ENOMEM);
5657 st->side_data = tmp;
5660 sd = &st->side_data[st->nb_side_data - 1];
5661 sd->type = AV_PKT_DATA_DISPLAYMATRIX;
5662 sd->size = sizeof(int32_t) * 9;
5663 sd->data = (uint8_t*)sc->display_matrix;
5664 sc->display_matrix = NULL;
5669 ff_configure_buffers_for_index(s, AV_TIME_BASE);
5674 static AVIndexEntry *mov_find_next_sample(AVFormatContext *s, AVStream **st)
5676 AVIndexEntry *sample = NULL;
5677 int64_t best_dts = INT64_MAX;
5679 for (i = 0; i < s->nb_streams; i++) {
5680 AVStream *avst = s->streams[i];
5681 MOVStreamContext *msc = avst->priv_data;
5682 if (msc->pb && msc->current_sample < avst->nb_index_entries) {
5683 AVIndexEntry *current_sample = &avst->index_entries[msc->current_sample];
5684 int64_t dts = av_rescale(current_sample->timestamp, AV_TIME_BASE, msc->time_scale);
5685 av_log(s, AV_LOG_TRACE, "stream %d, sample %d, dts %"PRId64"\n", i, msc->current_sample, dts);
5686 if (!sample || (!s->pb->seekable && current_sample->pos < sample->pos) ||
5688 ((msc->pb != s->pb && dts < best_dts) || (msc->pb == s->pb &&
5689 ((FFABS(best_dts - dts) <= AV_TIME_BASE && current_sample->pos < sample->pos) ||
5690 (FFABS(best_dts - dts) > AV_TIME_BASE && dts < best_dts)))))) {
5691 sample = current_sample;
5700 static int should_retry(AVIOContext *pb, int error_code) {
5701 if (error_code == AVERROR_EOF || avio_feof(pb))
5707 static int mov_switch_root(AVFormatContext *s, int64_t target)
5709 MOVContext *mov = s->priv_data;
5711 int already_read = 0;
5713 if (avio_seek(s->pb, target, SEEK_SET) != target) {
5714 av_log(mov->fc, AV_LOG_ERROR, "root atom offset 0x%"PRIx64": partial file\n", target);
5715 return AVERROR_INVALIDDATA;
5718 mov->next_root_atom = 0;
5720 for (i = 0; i < mov->fragment_index_count; i++) {
5721 MOVFragmentIndex *index = mov->fragment_index_data[i];
5723 for (j = 0; j < index->item_count; j++) {
5724 MOVFragmentIndexItem *item = &index->items[j];
5726 mov->next_root_atom = item->moof_offset;
5727 break; // Advance to next index in outer loop
5728 } else if (item->moof_offset == target) {
5729 index->current_item = FFMIN(j, index->current_item);
5730 if (item->headers_read)
5732 item->headers_read = 1;
5737 index->current_item = 0;
5743 mov->found_mdat = 0;
5745 if (mov_read_default(mov, s->pb, (MOVAtom){ AV_RL32("root"), INT64_MAX }) < 0 ||
5748 av_log(s, AV_LOG_TRACE, "read fragments, offset 0x%"PRIx64"\n", avio_tell(s->pb));
5753 static int mov_change_extradata(MOVStreamContext *sc, AVPacket *pkt)
5755 uint8_t *side, *extradata;
5758 /* Save the current index. */
5759 sc->last_stsd_index = sc->stsc_data[sc->stsc_index].id - 1;
5761 /* Notify the decoder that extradata changed. */
5762 extradata_size = sc->extradata_size[sc->last_stsd_index];
5763 extradata = sc->extradata[sc->last_stsd_index];
5764 if (extradata_size > 0 && extradata) {
5765 side = av_packet_new_side_data(pkt,
5766 AV_PKT_DATA_NEW_EXTRADATA,
5769 return AVERROR(ENOMEM);
5770 memcpy(side, extradata, extradata_size);
5776 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
5778 MOVContext *mov = s->priv_data;
5779 MOVStreamContext *sc;
5780 AVIndexEntry *sample;
5781 AVStream *st = NULL;
5785 sample = mov_find_next_sample(s, &st);
5786 if (!sample || (mov->next_root_atom && sample->pos > mov->next_root_atom)) {
5787 if (!mov->next_root_atom)
5789 if ((ret = mov_switch_root(s, mov->next_root_atom)) < 0)
5794 /* must be done just before reading, to avoid infinite loop on sample */
5795 sc->current_sample++;
5797 if (mov->next_root_atom) {
5798 sample->pos = FFMIN(sample->pos, mov->next_root_atom);
5799 sample->size = FFMIN(sample->size, (mov->next_root_atom - sample->pos));
5802 if (st->discard != AVDISCARD_ALL) {
5803 int64_t ret64 = avio_seek(sc->pb, sample->pos, SEEK_SET);
5804 if (ret64 != sample->pos) {
5805 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%"PRIx64": partial file\n",
5806 sc->ffindex, sample->pos);
5807 sc->current_sample -= should_retry(sc->pb, ret64);
5808 return AVERROR_INVALIDDATA;
5811 if( st->discard == AVDISCARD_NONKEY && 0==(sample->flags & AVINDEX_KEYFRAME) ) {
5812 av_log(mov->fc, AV_LOG_DEBUG, "Nonkey frame from stream %d discarded due to AVDISCARD_NONKEY\n", sc->ffindex);
5816 ret = av_get_packet(sc->pb, pkt, sample->size);
5818 sc->current_sample -= should_retry(sc->pb, ret);
5821 if (sc->has_palette) {
5824 pal = av_packet_new_side_data(pkt, AV_PKT_DATA_PALETTE, AVPALETTE_SIZE);
5826 av_log(mov->fc, AV_LOG_ERROR, "Cannot append palette to packet\n");
5828 memcpy(pal, sc->palette, AVPALETTE_SIZE);
5829 sc->has_palette = 0;
5832 #if CONFIG_DV_DEMUXER
5833 if (mov->dv_demux && sc->dv_audio_container) {
5834 avpriv_dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size, pkt->pos);
5835 av_freep(&pkt->data);
5837 ret = avpriv_dv_get_packet(mov->dv_demux, pkt);
5842 if (st->codecpar->codec_id == AV_CODEC_ID_MP3 && !st->need_parsing && pkt->size > 4) {
5843 if (ff_mpa_check_header(AV_RB32(pkt->data)) < 0)
5844 st->need_parsing = AVSTREAM_PARSE_FULL;
5848 pkt->stream_index = sc->ffindex;
5849 pkt->dts = sample->timestamp;
5850 if (sample->flags & AVINDEX_DISCARD_FRAME) {
5851 pkt->flags |= AV_PKT_FLAG_DISCARD;
5853 if (sc->ctts_data && sc->ctts_index < sc->ctts_count) {
5854 pkt->pts = pkt->dts + sc->dts_shift + sc->ctts_data[sc->ctts_index].duration;
5855 /* update ctts context */
5857 if (sc->ctts_index < sc->ctts_count &&
5858 sc->ctts_data[sc->ctts_index].count == sc->ctts_sample) {
5860 sc->ctts_sample = 0;
5863 int64_t next_dts = (sc->current_sample < st->nb_index_entries) ?
5864 st->index_entries[sc->current_sample].timestamp : st->duration;
5865 pkt->duration = next_dts - pkt->dts;
5866 pkt->pts = pkt->dts;
5868 if (st->discard == AVDISCARD_ALL)
5870 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? AV_PKT_FLAG_KEY : 0;
5871 pkt->pos = sample->pos;
5873 /* Multiple stsd handling. */
5874 if (sc->stsc_data) {
5875 /* Keep track of the stsc index for the given sample, then check
5876 * if the stsd index is different from the last used one. */
5878 if (sc->stsc_index < sc->stsc_count - 1 &&
5879 mov_get_stsc_samples(sc, sc->stsc_index) == sc->stsc_sample) {
5881 sc->stsc_sample = 0;
5882 /* Do not check indexes after a switch. */
5883 } else if (sc->stsc_data[sc->stsc_index].id > 0 &&
5884 sc->stsc_data[sc->stsc_index].id - 1 < sc->stsd_count &&
5885 sc->stsc_data[sc->stsc_index].id - 1 != sc->last_stsd_index) {
5886 ret = mov_change_extradata(sc, pkt);
5893 aax_filter(pkt->data, pkt->size, mov);
5895 if (sc->cenc.aes_ctr) {
5896 ret = cenc_filter(mov, sc, pkt->data, pkt->size);
5905 static int mov_seek_fragment(AVFormatContext *s, AVStream *st, int64_t timestamp)
5907 MOVContext *mov = s->priv_data;
5908 MOVStreamContext *sc = st->priv_data;
5911 if (!mov->fragment_index_complete)
5914 for (i = 0; i < mov->fragment_index_count; i++) {
5915 if (mov->fragment_index_data[i]->track_id == st->id || !sc->has_sidx) {
5916 MOVFragmentIndex *index = mov->fragment_index_data[i];
5917 for (j = index->item_count - 1; j >= 0; j--) {
5918 if (index->items[j].time <= timestamp) {
5919 if (index->items[j].headers_read)
5922 return mov_switch_root(s, index->items[j].moof_offset);
5931 static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp, int flags)
5933 MOVStreamContext *sc = st->priv_data;
5934 int sample, time_sample;
5937 int ret = mov_seek_fragment(s, st, timestamp);
5941 sample = av_index_search_timestamp(st, timestamp, flags);
5942 av_log(s, AV_LOG_TRACE, "stream %d, timestamp %"PRId64", sample %d\n", st->index, timestamp, sample);
5943 if (sample < 0 && st->nb_index_entries && timestamp < st->index_entries[0].timestamp)
5945 if (sample < 0) /* not sure what to do */
5946 return AVERROR_INVALIDDATA;
5947 sc->current_sample = sample;
5948 av_log(s, AV_LOG_TRACE, "stream %d, found sample %d\n", st->index, sc->current_sample);
5949 /* adjust ctts index */
5950 if (sc->ctts_data) {
5952 for (i = 0; i < sc->ctts_count; i++) {
5953 int next = time_sample + sc->ctts_data[i].count;
5954 if (next > sc->current_sample) {
5956 sc->ctts_sample = sc->current_sample - time_sample;
5963 /* adjust stsd index */
5965 for (i = 0; i < sc->stsc_count; i++) {
5966 int next = time_sample + mov_get_stsc_samples(sc, i);
5967 if (next > sc->current_sample) {
5969 sc->stsc_sample = sc->current_sample - time_sample;
5975 ret = mov_seek_auxiliary_info(s, sc);
5983 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
5985 MOVContext *mc = s->priv_data;
5990 if (stream_index >= s->nb_streams)
5991 return AVERROR_INVALIDDATA;
5993 st = s->streams[stream_index];
5994 sample = mov_seek_stream(s, st, sample_time, flags);
5998 if (mc->seek_individually) {
5999 /* adjust seek timestamp to found sample timestamp */
6000 int64_t seek_timestamp = st->index_entries[sample].timestamp;
6002 for (i = 0; i < s->nb_streams; i++) {
6004 MOVStreamContext *sc = s->streams[i]->priv_data;
6006 st->skip_samples = (sample_time <= 0) ? sc->start_pad : 0;
6008 if (stream_index == i)
6011 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
6012 mov_seek_stream(s, st, timestamp, flags);
6015 for (i = 0; i < s->nb_streams; i++) {
6016 MOVStreamContext *sc;
6019 sc->current_sample = 0;
6022 MOVStreamContext *sc;
6023 AVIndexEntry *entry = mov_find_next_sample(s, &st);
6025 return AVERROR_INVALIDDATA;
6027 if (sc->ffindex == stream_index && sc->current_sample == sample)
6029 sc->current_sample++;
6035 #define OFFSET(x) offsetof(MOVContext, x)
6036 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
6037 static const AVOption mov_options[] = {
6038 {"use_absolute_path",
6039 "allow using absolute path when opening alias, this is a possible security issue",
6040 OFFSET(use_absolute_path), AV_OPT_TYPE_BOOL, {.i64 = 0},
6042 {"seek_streams_individually",
6043 "Seek each stream individually to the to the closest point",
6044 OFFSET(seek_individually), AV_OPT_TYPE_BOOL, { .i64 = 1 },
6046 {"ignore_editlist", "", OFFSET(ignore_editlist), AV_OPT_TYPE_BOOL, {.i64 = 0},
6048 {"ignore_chapters", "", OFFSET(ignore_chapters), AV_OPT_TYPE_BOOL, {.i64 = 0},
6051 "use mfra for fragment timestamps",
6052 OFFSET(use_mfra_for), AV_OPT_TYPE_INT, {.i64 = FF_MOV_FLAG_MFRA_AUTO},
6053 -1, FF_MOV_FLAG_MFRA_PTS, FLAGS,
6055 {"auto", "auto", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_AUTO}, 0, 0,
6056 FLAGS, "use_mfra_for" },
6057 {"dts", "dts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_DTS}, 0, 0,
6058 FLAGS, "use_mfra_for" },
6059 {"pts", "pts", 0, AV_OPT_TYPE_CONST, {.i64 = FF_MOV_FLAG_MFRA_PTS}, 0, 0,
6060 FLAGS, "use_mfra_for" },
6061 { "export_all", "Export unrecognized metadata entries", OFFSET(export_all),
6062 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
6063 { "export_xmp", "Export full XMP metadata", OFFSET(export_xmp),
6064 AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, .flags = FLAGS },
6065 { "activation_bytes", "Secret bytes for Audible AAX files", OFFSET(activation_bytes),
6066 AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
6067 { "audible_fixed_key", // extracted from libAAX_SDK.so and AAXSDKWin.dll files!
6068 "Fixed key used for handling Audible AAX files", OFFSET(audible_fixed_key),
6069 AV_OPT_TYPE_BINARY, {.str="77214d4b196a87cd520045fd20a51d67"},
6070 .flags = AV_OPT_FLAG_DECODING_PARAM },
6071 { "decryption_key", "The media decryption key (hex)", OFFSET(decryption_key), AV_OPT_TYPE_BINARY, .flags = AV_OPT_FLAG_DECODING_PARAM },
6072 { "enable_drefs", "Enable external track support.", OFFSET(enable_drefs), AV_OPT_TYPE_BOOL,
6073 {.i64 = 0}, 0, 1, FLAGS },
6078 static const AVClass mov_class = {
6079 .class_name = "mov,mp4,m4a,3gp,3g2,mj2",
6080 .item_name = av_default_item_name,
6081 .option = mov_options,
6082 .version = LIBAVUTIL_VERSION_INT,
6085 AVInputFormat ff_mov_demuxer = {
6086 .name = "mov,mp4,m4a,3gp,3g2,mj2",
6087 .long_name = NULL_IF_CONFIG_SMALL("QuickTime / MOV"),
6088 .priv_class = &mov_class,
6089 .priv_data_size = sizeof(MOVContext),
6090 .extensions = "mov,mp4,m4a,3gp,3g2,mj2",
6091 .read_probe = mov_probe,
6092 .read_header = mov_read_header,
6093 .read_packet = mov_read_packet,
6094 .read_close = mov_read_close,
6095 .read_seek = mov_read_seek,
6096 .flags = AVFMT_NO_BYTE_SEEK,