3 * Copyright (c) 2001 Fabrice Bellard.
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2 of the License, or (at your option) any later version.
10 * This library is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
34 * First version by Francois Revol revol@free.fr
35 * Seek function by Gael Chardon gael.dev@4now.net
37 * Features and limitations:
38 * - reads most of the QT files I have (at least the structure),
39 * the exceptions are .mov with zlib compressed headers ('cmov' section). It shouldn't be hard to implement.
40 * FIXED, Francois Revol, 07/17/2002
41 * - ffmpeg has nearly none of the usual QuickTime codecs,
42 * although I succesfully dumped raw and mp3 audio tracks off .mov files.
43 * Sample QuickTime files with mp3 audio can be found at: http://www.3ivx.com/showcase.html
44 * - .mp4 parsing is still hazardous, although the format really is QuickTime with some minor changes
45 * (to make .mov parser crash maybe ?), despite what they say in the MPEG FAQ at
46 * http://mpeg.telecomitalialab.com/faq.htm
47 * - the code is quite ugly... maybe I won't do it recursive next time :-)
48 * - seek is not supported with files that contain edit list
50 * Funny I didn't know about http://sourceforge.net/projects/qt-ffmpeg/
51 * when coding this :) (it's a writer anyway)
53 * Reference documents:
54 * http://www.geocities.com/xhelmboyx/quicktime/formats/qtm-layout.txt
56 * http://developer.apple.com/documentation/QuickTime/QTFF/
57 * http://developer.apple.com/documentation/QuickTime/QTFF/qtff.pdf
58 * QuickTime is a trademark of Apple (AFAIK :))
61 #include "qtpalette.h"
67 static const CodecTag mov_video_tags[] = {
68 /* { CODEC_ID_, MKTAG('c', 'v', 'i', 'd') }, *//* Cinepak */
69 /* { CODEC_ID_H263, MKTAG('r', 'a', 'w', ' ') }, *//* Uncompressed RGB */
70 /* { CODEC_ID_H263, MKTAG('Y', 'u', 'v', '2') }, *//* Uncompressed YUV422 */
71 /* { CODEC_ID_RAWVIDEO, MKTAG('A', 'V', 'U', 'I') }, *//* YUV with alpha-channel (AVID Uncompressed) */
76 { CODEC_ID_MJPEG, MKTAG('j', 'p', 'e', 'g') }, /* PhotoJPEG */
77 { CODEC_ID_MPEG1VIDEO, MKTAG('m', 'p', 'e', 'g') }, /* MPEG */
78 { CODEC_ID_MJPEG, MKTAG('m', 'j', 'p', 'a') }, /* Motion-JPEG (format A) */
79 { CODEC_ID_MJPEGB, MKTAG('m', 'j', 'p', 'b') }, /* Motion-JPEG (format B) */
80 { CODEC_ID_MJPEG, MKTAG('A', 'V', 'D', 'J') }, /* MJPEG with alpha-channel (AVID JFIF meridien compressed) */
81 /* { CODEC_ID_MJPEG, MKTAG('A', 'V', 'R', 'n') }, *//* MJPEG with alpha-channel (AVID ABVB/Truevision NuVista) */
82 /* { CODEC_ID_GIF, MKTAG('g', 'i', 'f', ' ') }, *//* embedded gif files as frames (usually one "click to play movie" frame) */
84 { CODEC_ID_SVQ1, MKTAG('S', 'V', 'Q', '1') }, /* Sorenson Video v1 */
85 { CODEC_ID_SVQ1, MKTAG('s', 'v', 'q', '1') }, /* Sorenson Video v1 */
86 { CODEC_ID_SVQ1, MKTAG('s', 'v', 'q', 'i') }, /* Sorenson Video v1 (from QT specs)*/
87 { CODEC_ID_SVQ3, MKTAG('S', 'V', 'Q', '3') }, /* Sorenson Video v3 */
88 { CODEC_ID_MPEG4, MKTAG('m', 'p', '4', 'v') },
89 { CODEC_ID_MPEG4, MKTAG('D', 'I', 'V', 'X') }, /* OpenDiVX *//* sample files at http://heroinewarrior.com/xmovie.php3 use this tag */
90 { CODEC_ID_MPEG4, MKTAG('X', 'V', 'I', 'D') },
91 { CODEC_ID_MPEG4, MKTAG('3', 'I', 'V', '2') }, /* experimental: 3IVX files before ivx D4 4.5.1 */
92 /* { CODEC_ID_, MKTAG('I', 'V', '5', '0') }, *//* Indeo 5.0 */
93 { CODEC_ID_H263, MKTAG('h', '2', '6', '3') }, /* H263 */
94 { CODEC_ID_H263, MKTAG('s', '2', '6', '3') }, /* H263 ?? works */
95 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'c', ' ') }, /* DV NTSC */
96 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'c', 'p') }, /* DV PAL */
97 { CODEC_ID_VP3, MKTAG('V', 'P', '3', '1') }, /* On2 VP3 */
98 { CODEC_ID_RPZA, MKTAG('r', 'p', 'z', 'a') }, /* Apple Video (RPZA) */
99 { CODEC_ID_CINEPAK, MKTAG('c', 'v', 'i', 'd') }, /* Cinepak */
100 { CODEC_ID_8BPS, MKTAG('8', 'B', 'P', 'S') }, /* Planar RGB (8BPS) */
101 { CODEC_ID_SMC, MKTAG('s', 'm', 'c', ' ') }, /* Apple Graphics (SMC) */
102 { CODEC_ID_QTRLE, MKTAG('r', 'l', 'e', ' ') }, /* Apple Animation (RLE) */
103 { CODEC_ID_QDRAW, MKTAG('q', 'd', 'r', 'w') }, /* QuickDraw */
104 { CODEC_ID_H264, MKTAG('a', 'v', 'c', '1') }, /* AVC-1/H.264 */
105 { CODEC_ID_MPEG2VIDEO, MKTAG('h', 'd', 'v', '2') }, /* MPEG2 produced by Sony HD camera */
106 { CODEC_ID_MPEG2VIDEO, MKTAG('h', 'd', 'v', '3') }, /* HDV produced by FCP */
107 { CODEC_ID_MPEG2VIDEO, MKTAG('m', 'x', '5', 'n') }, /* MPEG2 IMX NTSC 525/60 50mb/s produced by FCP */
108 { CODEC_ID_MPEG2VIDEO, MKTAG('m', 'x', '5', 'p') }, /* MPEG2 IMX PAL 625/50 50mb/s produced by FCP */
109 { CODEC_ID_MPEG2VIDEO, MKTAG('m', 'x', '3', 'n') }, /* MPEG2 IMX NTSC 525/60 30mb/s produced by FCP */
110 { CODEC_ID_MPEG2VIDEO, MKTAG('m', 'x', '3', 'p') }, /* MPEG2 IMX PAL 625/50 30mb/s produced by FCP */
111 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'p', 'p') }, /* DVCPRO PAL produced by FCP */
112 //{ CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'h', '5') }, /* DVCPRO HD 50i produced by FCP */
113 //{ CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'h', '6') }, /* DVCPRO HD 60i produced by FCP */
114 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', '5', 'p') }, /* DVCPRO50 PAL produced by FCP */
115 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', '5', 'n') }, /* DVCPRO50 NTSC produced by FCP */
116 { CODEC_ID_DVVIDEO, MKTAG('A', 'V', 'd', 'v') }, /* AVID DV */
117 //{ CODEC_ID_JPEG2000, MKTAG('m', 'j', 'p', '2') }, /* JPEG 2000 produced by FCP */
118 { CODEC_ID_RAWVIDEO, MKTAG('2', 'v', 'u', 'y') }, /* UNCOMPRESSED 8BIT 4:2:2 */
119 { CODEC_ID_NONE, 0 },
122 static const CodecTag mov_audio_tags[] = {
123 { CODEC_ID_PCM_S32BE, MKTAG('i', 'n', '3', '2') },
124 { CODEC_ID_PCM_S24BE, MKTAG('i', 'n', '2', '4') },
125 { CODEC_ID_PCM_S16BE, MKTAG('N', 'O', 'N', 'E') }, /* uncompressed */
126 { CODEC_ID_PCM_S16BE, MKTAG('t', 'w', 'o', 's') }, /* 16 bits */
127 { CODEC_ID_PCM_U8, MKTAG('r', 'a', 'w', ' ') }, /* 8 bits unsigned */
128 { CODEC_ID_PCM_S16LE, MKTAG('s', 'o', 'w', 't') }, /* */
129 { CODEC_ID_PCM_MULAW, MKTAG('u', 'l', 'a', 'w') }, /* */
130 { CODEC_ID_PCM_ALAW, MKTAG('a', 'l', 'a', 'w') }, /* */
131 { CODEC_ID_ADPCM_IMA_QT, MKTAG('i', 'm', 'a', '4') }, /* IMA-4 ADPCM */
132 { CODEC_ID_ADPCM_MS, MKTAG('m', 's', 0x00, 0x02) }, /* MS ADPCM */
133 { CODEC_ID_MACE3, MKTAG('M', 'A', 'C', '3') }, /* Macintosh Audio Compression and Expansion 3:1 */
134 { CODEC_ID_MACE6, MKTAG('M', 'A', 'C', '6') }, /* Macintosh Audio Compression and Expansion 6:1 */
136 { CODEC_ID_MP3, MKTAG('.', 'm', 'p', '3') }, /* MPEG layer 3 */ /* sample files at http://www.3ivx.com/showcase.html use this tag */
137 { CODEC_ID_MP2, 0x6D730055 }, /* MPEG layer 3 */
138 { CODEC_ID_MP2, 0x5500736D }, /* MPEG layer 3 *//* XXX: check endianness */
139 /* { CODEC_ID_OGG_VORBIS, MKTAG('O', 'g', 'g', 'S') }, *//* sample files at http://heroinewarrior.com/xmovie.php3 use this tag */
141 { CODEC_ID_AAC, MKTAG('m', 'p', '4', 'a') }, /* MPEG-4 AAC */
142 /* The standard for mpeg4 audio is still not normalised AFAIK anyway */
143 { CODEC_ID_AMR_NB, MKTAG('s', 'a', 'm', 'r') }, /* AMR-NB 3gp */
144 { CODEC_ID_AMR_WB, MKTAG('s', 'a', 'w', 'b') }, /* AMR-WB 3gp */
145 { CODEC_ID_AC3, MKTAG('m', 's', 0x20, 0x00) }, /* Dolby AC-3 */
146 { CODEC_ID_ALAC,MKTAG('a', 'l', 'a', 'c') }, /* Apple Lossless */
147 { CODEC_ID_QDM2,MKTAG('Q', 'D', 'M', '2') }, /* QDM2 */
148 { CODEC_ID_DVAUDIO, MKTAG('v', 'd', 'v', 'a') },
149 { CODEC_ID_DVAUDIO, MKTAG('d', 'v', 'c', 'a') },
150 { CODEC_ID_PCM_S16LE, MKTAG('l', 'p', 'c', 'm') },
151 { CODEC_ID_NONE, 0 },
154 /* the QuickTime file format is quite convoluted...
155 * it has lots of index tables, each indexing something in another one...
156 * Here we just use what is needed to read the chunks
159 typedef struct MOV_sample_to_chunk_tbl {
163 } MOV_sample_to_chunk_tbl;
168 int64_t size; /* total size (excluding the size and type fields) */
178 typedef struct MOV_mdat_atom_s {
185 uint32_t flags; // 24bit
187 /* 0x03 ESDescrTag */
189 #define MP4ODescrTag 0x01
190 #define MP4IODescrTag 0x02
191 #define MP4ESDescrTag 0x03
192 #define MP4DecConfigDescrTag 0x04
193 #define MP4DecSpecificDescrTag 0x05
194 #define MP4SLConfigDescrTag 0x06
195 #define MP4ContentIdDescrTag 0x07
196 #define MP4SupplContentIdDescrTag 0x08
197 #define MP4IPIPtrDescrTag 0x09
198 #define MP4IPMPPtrDescrTag 0x0A
199 #define MP4IPMPDescrTag 0x0B
200 #define MP4RegistrationDescrTag 0x0D
201 #define MP4ESIDIncDescrTag 0x0E
202 #define MP4ESIDRefDescrTag 0x0F
203 #define MP4FileIODescrTag 0x10
204 #define MP4FileODescrTag 0x11
205 #define MP4ExtProfileLevelDescrTag 0x13
206 #define MP4ExtDescrTagsStart 0x80
207 #define MP4ExtDescrTagsEnd 0xFE
208 uint8_t stream_priority;
210 /* 0x04 DecConfigDescrTag */
211 uint8_t object_type_id;
213 /* XXX: really streamType is
214 * only 6bit, followed by:
218 uint32_t buffer_size_db; // 24
219 uint32_t max_bitrate;
220 uint32_t avg_bitrate;
222 /* 0x05 DecSpecificDescrTag */
223 uint8_t decoder_cfg_len;
224 uint8_t *decoder_cfg;
226 /* 0x06 SLConfigDescrTag */
227 uint8_t sl_config_len;
231 struct MOVParseTableEntry;
233 typedef struct MOVStreamContext {
234 int ffindex; /* the ffmpeg stream id */
237 int64_t *chunk_offsets;
239 Time2Sample *stts_data;
241 Time2Sample *ctts_data;
242 int edit_count; /* number of 'edit' (elst atom) */
243 long sample_to_chunk_sz;
244 MOV_sample_to_chunk_tbl *sample_to_chunk;
245 int sample_to_ctime_index;
246 int sample_to_ctime_sample;
256 AVRational sample_size_v1;
257 int dv_audio_container;
260 typedef struct MOVContext {
261 int mp4; /* set to 1 as soon as we are sure that the file is an .mp4 file (even some header parsing depends on this) */
264 int64_t duration; /* duration of the longest track */
265 int found_moov; /* when both 'moov' and 'mdat' sections has been found */
266 int found_mdat; /* we suppose we have enough data to read the file */
270 /* some streams listed here aren't presented to the ffmpeg API, since they aren't either video nor audio
271 * but we need the info to be able to skip data from those streams in the 'mdat' section
273 MOVStreamContext *streams[MAX_STREAMS];
276 MOV_ctab_t **ctab; /* color tables */
277 const struct MOVParseTableEntry *parse_table; /* could be eventually used to change the table */
278 /* NOTE: for recursion save to/ restore from local variable! */
280 AVPaletteControl palette_control;
281 MOV_mdat_atom_t *mdat_list;
283 DVDemuxContext *dv_demux;
284 AVFormatContext *dv_fctx;
288 /* XXX: it's the first time I make a recursive parser I think... sorry if it's ugly :P */
290 /* those functions parse an atom */
292 1: found what I wanted, exit
293 0: continue to parse next atom
294 -1: error occured, exit
296 typedef int (*mov_parse_function)(MOVContext *ctx, ByteIOContext *pb, MOV_atom_t atom);
298 /* links atom IDs to parse functions */
299 typedef struct MOVParseTableEntry {
301 mov_parse_function func;
302 } MOVParseTableEntry;
304 static int mov_read_leaf(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
307 url_fskip(pb, atom.size);
308 /* url_seek(pb, atom_offset+atom.size, SEEK_SET); */
312 static int mov_read_default(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
314 int64_t total_size = 0;
319 a.offset = atom.offset;
322 atom.size = 0x7fffffffffffffffLL;
323 while(((total_size + 8) < atom.size) && !url_feof(pb) && !err) {
327 a.size = get_be32(pb);
328 a.type = get_le32(pb);
332 dprintf("type: %08x %.4s sz: %"PRIx64" %"PRIx64" %"PRIx64"\n", a.type, (char*)&a.type, a.size, atom.size, total_size);
333 if (a.size == 1) { /* 64 bit extended size */
334 a.size = get_be64(pb) - 8;
339 a.size = atom.size - total_size;
343 for (i = 0; c->parse_table[i].type != 0L
344 && c->parse_table[i].type != a.type; i++)
352 if (c->parse_table[i].type == 0) { /* skip leaf atoms data */
353 url_fskip(pb, a.size);
355 offset_t start_pos = url_ftell(pb);
357 err = (c->parse_table[i].func)(c, pb, a);
358 left = a.size - url_ftell(pb) + start_pos;
359 if (left > 0) /* skip garbage at atom end */
364 total_size += a.size;
367 if (!err && total_size < atom.size && atom.size < 0x7ffff) {
368 url_fskip(pb, atom.size - total_size);
374 static int mov_read_ctab(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
377 url_fskip(pb, atom.size); // for now
379 VERY VERY BROKEN, NEVER execute this, needs rewrite
382 c->ctab = av_realloc(c->ctab, ++c->ctab_size);
383 t = c->ctab[c->ctab_size];
384 t->seed = get_be32(pb);
385 t->flags = get_be16(pb);
386 t->size = get_be16(pb) + 1;
387 len = 2 * t->size * 4;
389 t->clrs = av_malloc(len); // 16bit A R G B
391 get_buffer(pb, t->clrs, len);
398 static int mov_read_hdlr(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
400 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
405 get_byte(pb); /* version */
406 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
409 ctype = get_le32(pb);
410 type = get_le32(pb); /* component subtype */
412 dprintf("ctype= %c%c%c%c (0x%08lx)\n", *((char *)&ctype), ((char *)&ctype)[1], ((char *)&ctype)[2], ((char *)&ctype)[3], (long) ctype);
413 dprintf("stype= %c%c%c%c\n", *((char *)&type), ((char *)&type)[1], ((char *)&type)[2], ((char *)&type)[3]);
414 if(ctype == MKTAG('m', 'h', 'l', 'r')) /* MOV */
418 if(type == MKTAG('v', 'i', 'd', 'e'))
419 st->codec->codec_type = CODEC_TYPE_VIDEO;
420 else if(type == MKTAG('s', 'o', 'u', 'n'))
421 st->codec->codec_type = CODEC_TYPE_AUDIO;
422 get_be32(pb); /* component manufacture */
423 get_be32(pb); /* component flags */
424 get_be32(pb); /* component flags mask */
427 return 0; /* nothing left to read */
428 /* XXX: MP4 uses a C string, not a pascal one */
433 while(get_byte(pb) && (++len < (atom.size - 24)));
435 /* .mov: PASCAL string */
440 url_fskip(pb, atom.size - (url_ftell(pb) - atom.offset));
444 static int mov_mp4_read_descr_len(ByteIOContext *pb)
449 int c = get_byte(pb);
450 len = (len << 7) | (c & 0x7f);
457 static int mov_mp4_read_descr(ByteIOContext *pb, int *tag)
461 len = mov_mp4_read_descr_len(pb);
462 dprintf("MPEG4 description: tag=0x%02x len=%d\n", *tag, len);
466 static int mov_read_esds(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
468 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
469 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
472 /* Well, broken but suffisant for some MP4 streams */
473 get_be32(pb); /* version + flags */
474 len = mov_mp4_read_descr(pb, &tag);
475 if (tag == MP4ESDescrTag) {
476 get_be16(pb); /* ID */
477 get_byte(pb); /* priority */
479 get_be16(pb); /* ID */
481 len = mov_mp4_read_descr(pb, &tag);
482 if (tag == MP4DecConfigDescrTag) {
483 sc->esds.object_type_id = get_byte(pb);
484 sc->esds.stream_type = get_byte(pb);
485 sc->esds.buffer_size_db = get_be24(pb);
486 sc->esds.max_bitrate = get_be32(pb);
487 sc->esds.avg_bitrate = get_be32(pb);
489 st->codec->codec_id= codec_get_id(ff_mov_obj_type, sc->esds.object_type_id);
490 dprintf("esds object type id %d\n", sc->esds.object_type_id);
491 len = mov_mp4_read_descr(pb, &tag);
492 if (tag == MP4DecSpecificDescrTag) {
493 dprintf("Specific MPEG4 header len=%d\n", len);
494 st->codec->extradata = (uint8_t*) av_mallocz(len + FF_INPUT_BUFFER_PADDING_SIZE);
495 if (st->codec->extradata) {
496 get_buffer(pb, st->codec->extradata, len);
497 st->codec->extradata_size = len;
499 if ((*(uint8_t *)st->codec->extradata >> 3) == 29) {
500 st->codec->codec_id = CODEC_ID_MP3ON4;
508 /* this atom contains actual media data */
509 static int mov_read_mdat(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
511 if(atom.size == 0) /* wrong one (MP4) */
513 c->mdat_list = av_realloc(c->mdat_list, (c->mdat_count + 1) * sizeof(*c->mdat_list));
514 c->mdat_list[c->mdat_count].offset = atom.offset;
515 c->mdat_list[c->mdat_count].size = atom.size;
518 c->mdat_offset = atom.offset;
519 c->mdat_size = atom.size;
521 return 1; /* found both, just go */
522 url_fskip(pb, atom.size);
523 return 0; /* now go for moov */
526 static int mov_read_ftyp(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
528 uint32_t type = get_le32(pb);
532 case MKTAG('i', 's', 'o', 'm'):
533 case MKTAG('m', 'p', '4', '1'):
534 case MKTAG('m', 'p', '4', '2'):
535 case MKTAG('3', 'g', 'p', '1'):
536 case MKTAG('3', 'g', 'p', '2'):
537 case MKTAG('3', 'g', '2', 'a'):
538 case MKTAG('3', 'g', 'p', '3'):
539 case MKTAG('3', 'g', 'p', '4'):
540 case MKTAG('3', 'g', 'p', '5'):
541 case MKTAG('m', 'm', 'p', '4'): /* Mobile MP4 */
542 case MKTAG('M', '4', 'A', ' '): /* Apple iTunes AAC-LC Audio */
543 case MKTAG('M', '4', 'P', ' '): /* Apple iTunes AAC-LC Protected Audio */
544 case MKTAG('m', 'j', 'p', '2'): /* Motion Jpeg 2000 */
546 case MKTAG('q', 't', ' ', ' '):
548 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
550 get_be32(pb); /* minor version */
551 url_fskip(pb, atom.size - 8);
555 /* this atom should contain all header atoms */
556 static int mov_read_moov(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
560 err = mov_read_default(c, pb, atom);
561 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
562 /* so we don't parse the whole file if over a network */
565 return 1; /* found both, just go */
566 return 0; /* now go for mdat */
570 static int mov_read_mdhd(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
572 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
573 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
574 int version = get_byte(pb);
578 return 1; /* unsupported */
580 get_byte(pb); get_byte(pb);
581 get_byte(pb); /* flags */
587 get_be32(pb); /* creation time */
588 get_be32(pb); /* modification time */
591 sc->time_scale = get_be32(pb);
592 st->duration = (version == 1) ? get_be64(pb) : get_be32(pb); /* duration */
594 lang = get_be16(pb); /* language */
595 ff_mov_lang_to_iso639(lang, st->language);
596 get_be16(pb); /* quality */
601 static int mov_read_mvhd(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
603 int version = get_byte(pb); /* version */
604 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
610 get_be32(pb); /* creation time */
611 get_be32(pb); /* modification time */
613 c->time_scale = get_be32(pb); /* time scale */
615 av_log(NULL, AV_LOG_DEBUG, "time scale = %i\n", c->time_scale);
617 c->duration = (version == 1) ? get_be64(pb) : get_be32(pb); /* duration */
618 get_be32(pb); /* preferred scale */
620 get_be16(pb); /* preferred volume */
622 url_fskip(pb, 10); /* reserved */
624 url_fskip(pb, 36); /* display matrix */
626 get_be32(pb); /* preview time */
627 get_be32(pb); /* preview duration */
628 get_be32(pb); /* poster time */
629 get_be32(pb); /* selection time */
630 get_be32(pb); /* selection duration */
631 get_be32(pb); /* current time */
632 get_be32(pb); /* next track ID */
637 static int mov_read_smi(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
639 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
641 if((uint64_t)atom.size > (1<<30))
644 // currently SVQ3 decoder expect full STSD header - so let's fake it
645 // this should be fixed and just SMI header should be passed
646 av_free(st->codec->extradata);
647 st->codec->extradata_size = 0x5a + atom.size;
648 st->codec->extradata = (uint8_t*) av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
650 if (st->codec->extradata) {
651 strcpy(st->codec->extradata, "SVQ3"); // fake
652 get_buffer(pb, st->codec->extradata + 0x5a, atom.size);
653 dprintf("Reading SMI %"PRId64" %s\n", atom.size, (char*)st->codec->extradata + 0x5a);
655 url_fskip(pb, atom.size);
660 static int mov_read_enda(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
662 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
663 int little_endian = get_be16(pb);
666 switch (st->codec->codec_id) {
667 case CODEC_ID_PCM_S24BE:
668 st->codec->codec_id = CODEC_ID_PCM_S24LE;
670 case CODEC_ID_PCM_S32BE:
671 st->codec->codec_id = CODEC_ID_PCM_S32LE;
680 static int mov_read_alac(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
682 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
684 // currently ALAC decoder expect full atom header - so let's fake it
685 // this should be fixed and just ALAC header should be passed
687 av_free(st->codec->extradata);
688 st->codec->extradata_size = 36;
689 st->codec->extradata = (uint8_t*) av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
691 if (st->codec->extradata) {
692 strcpy(st->codec->extradata + 4, "alac"); // fake
693 get_buffer(pb, st->codec->extradata + 8, 36 - 8);
694 dprintf("Reading alac %d %s\n", st->codec->extradata_size, (char*)st->codec->extradata);
696 url_fskip(pb, atom.size);
700 static int mov_read_wave(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
702 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
704 if((uint64_t)atom.size > (1<<30))
707 if (st->codec->codec_id == CODEC_ID_QDM2) {
708 // pass all frma atom to codec, needed at least for QDM2
709 av_free(st->codec->extradata);
710 st->codec->extradata_size = atom.size;
711 st->codec->extradata = (uint8_t*) av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
713 if (st->codec->extradata) {
714 get_buffer(pb, st->codec->extradata, atom.size);
716 url_fskip(pb, atom.size);
717 } else if (atom.size > 8) { /* to read frma, esds atoms */
718 mov_read_default(c, pb, atom);
720 url_fskip(pb, atom.size);
724 static int mov_read_jp2h(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
726 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
728 if((uint64_t)atom.size > (1<<30))
731 av_free(st->codec->extradata);
733 st->codec->extradata_size = atom.size + 8;
734 st->codec->extradata = (uint8_t*) av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
736 /* pass all jp2h atom to codec */
737 if (st->codec->extradata) {
738 strcpy(st->codec->extradata + 4, "jp2h");
739 get_buffer(pb, st->codec->extradata + 8, atom.size);
741 url_fskip(pb, atom.size);
745 static int mov_read_avcC(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
747 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
749 if((uint64_t)atom.size > (1<<30))
752 av_free(st->codec->extradata);
754 st->codec->extradata_size = atom.size;
755 st->codec->extradata = (uint8_t*) av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
757 if (st->codec->extradata) {
758 get_buffer(pb, st->codec->extradata, atom.size);
760 url_fskip(pb, atom.size);
765 static int mov_read_stco(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
767 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
768 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
769 unsigned int i, entries;
771 get_byte(pb); /* version */
772 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
774 entries = get_be32(pb);
776 if(entries >= UINT_MAX/sizeof(int64_t))
779 sc->chunk_count = entries;
780 sc->chunk_offsets = (int64_t*) av_malloc(entries * sizeof(int64_t));
781 if (!sc->chunk_offsets)
783 if (atom.type == MKTAG('s', 't', 'c', 'o')) {
784 for(i=0; i<entries; i++) {
785 sc->chunk_offsets[i] = get_be32(pb);
787 } else if (atom.type == MKTAG('c', 'o', '6', '4')) {
788 for(i=0; i<entries; i++) {
789 sc->chunk_offsets[i] = get_be64(pb);
797 static int mov_read_stsd(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
799 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
800 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
801 int entries, frames_per_sample;
803 uint8_t codec_name[32];
805 /* for palette traversal */
813 unsigned char *color_table;
815 unsigned char r, g, b;
817 get_byte(pb); /* version */
818 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
820 entries = get_be32(pb);
822 while(entries--) { //Parsing Sample description table
824 MOV_atom_t a = { 0, 0, 0 };
825 offset_t start_pos = url_ftell(pb);
826 int size = get_be32(pb); /* size */
827 format = get_le32(pb); /* data format */
829 get_be32(pb); /* reserved */
830 get_be16(pb); /* reserved */
831 get_be16(pb); /* index */
833 if (st->codec->codec_tag) {
834 /* multiple fourcc, just skip for now */
835 url_fskip(pb, size - (url_ftell(pb) - start_pos));
839 st->codec->codec_tag = format;
840 id = codec_get_id(mov_audio_tags, format);
842 st->codec->codec_type = CODEC_TYPE_AUDIO;
843 } else if (format && format != MKTAG('m', 'p', '4', 's')) { /* skip old asf mpeg4 tag */
844 id = codec_get_id(mov_video_tags, format);
846 id = codec_get_id(codec_bmp_tags, format);
848 st->codec->codec_type = CODEC_TYPE_VIDEO;
851 dprintf("size=%d 4CC= %c%c%c%c codec_type=%d\n",
853 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff, (format >> 24) & 0xff,
854 st->codec->codec_type);
856 if(st->codec->codec_type==CODEC_TYPE_VIDEO) {
857 st->codec->codec_id = id;
858 get_be16(pb); /* version */
859 get_be16(pb); /* revision level */
860 get_be32(pb); /* vendor */
861 get_be32(pb); /* temporal quality */
862 get_be32(pb); /* spacial quality */
864 st->codec->width = get_be16(pb); /* width */
865 st->codec->height = get_be16(pb); /* height */
867 get_be32(pb); /* horiz resolution */
868 get_be32(pb); /* vert resolution */
869 get_be32(pb); /* data size, always 0 */
870 frames_per_sample = get_be16(pb); /* frames per samples */
872 av_log(NULL, AV_LOG_DEBUG, "frames/samples = %d\n", frames_per_sample);
874 get_buffer(pb, codec_name, 32); /* codec name, pascal string (FIXME: true for mp4?) */
875 if (codec_name[0] <= 31) {
876 memcpy(st->codec->codec_name, &codec_name[1],codec_name[0]);
877 st->codec->codec_name[codec_name[0]] = 0;
880 st->codec->bits_per_sample = get_be16(pb); /* depth */
881 st->codec->color_table_id = get_be16(pb); /* colortable id */
883 /* figure out the palette situation */
884 color_depth = st->codec->bits_per_sample & 0x1F;
885 color_greyscale = st->codec->bits_per_sample & 0x20;
887 /* if the depth is 2, 4, or 8 bpp, file is palettized */
888 if ((color_depth == 2) || (color_depth == 4) ||
889 (color_depth == 8)) {
891 if (color_greyscale) {
893 /* compute the greyscale palette */
894 color_count = 1 << color_depth;
896 color_dec = 256 / (color_count - 1);
897 for (j = 0; j < color_count; j++) {
898 r = g = b = color_index;
899 c->palette_control.palette[j] =
900 (r << 16) | (g << 8) | (b);
901 color_index -= color_dec;
906 } else if (st->codec->color_table_id & 0x08) {
908 /* if flag bit 3 is set, use the default palette */
909 color_count = 1 << color_depth;
910 if (color_depth == 2)
911 color_table = ff_qt_default_palette_4;
912 else if (color_depth == 4)
913 color_table = ff_qt_default_palette_16;
915 color_table = ff_qt_default_palette_256;
917 for (j = 0; j < color_count; j++) {
918 r = color_table[j * 4 + 0];
919 g = color_table[j * 4 + 1];
920 b = color_table[j * 4 + 2];
921 c->palette_control.palette[j] =
922 (r << 16) | (g << 8) | (b);
927 /* load the palette from the file */
928 color_start = get_be32(pb);
929 color_count = get_be16(pb);
930 color_end = get_be16(pb);
931 for (j = color_start; j <= color_end; j++) {
932 /* each R, G, or B component is 16 bits;
933 * only use the top 8 bits; skip alpha bytes
943 c->palette_control.palette[j] =
944 (r << 16) | (g << 8) | (b);
948 st->codec->palctrl = &c->palette_control;
949 st->codec->palctrl->palette_changed = 1;
951 st->codec->palctrl = NULL;
952 } else if(st->codec->codec_type==CODEC_TYPE_AUDIO) {
954 uint16_t version = get_be16(pb);
956 st->codec->codec_id = id;
957 get_be16(pb); /* revision level */
958 get_be32(pb); /* vendor */
960 st->codec->channels = get_be16(pb); /* channel count */
961 dprintf("audio channels %d\n", st->codec->channels);
962 st->codec->bits_per_sample = get_be16(pb); /* sample size */
963 /* do we need to force to 16 for AMR ? */
965 /* handle specific s8 codec */
966 get_be16(pb); /* compression id = 0*/
967 get_be16(pb); /* packet size = 0 */
969 st->codec->sample_rate = ((get_be32(pb) >> 16));
971 switch (st->codec->codec_id) {
972 case CODEC_ID_PCM_S8:
973 case CODEC_ID_PCM_U8:
974 if (st->codec->bits_per_sample == 16)
975 st->codec->codec_id = CODEC_ID_PCM_S16BE;
977 case CODEC_ID_PCM_S16LE:
978 case CODEC_ID_PCM_S16BE:
979 if (st->codec->bits_per_sample == 8)
980 st->codec->codec_id = CODEC_ID_PCM_S8;
981 else if (st->codec->bits_per_sample == 24)
982 st->codec->codec_id = CODEC_ID_PCM_S24BE;
988 //Read QT version 1 fields. In version 0 theese dont exist
989 dprintf("version =%d mp4=%d\n",version,c->mp4);
991 sc->sample_size_v1.den = get_be32(pb); /* samples per packet */
992 get_be32(pb); /* bytes per packet */
993 sc->sample_size_v1.num = get_be32(pb); /* bytes per frame */
994 get_be32(pb); /* bytes per sample */
995 } else if(version==2) {
996 get_be32(pb); /* sizeof struct only */
997 st->codec->sample_rate = av_int2dbl(get_be64(pb)); /* float 64 */
998 st->codec->channels = get_be32(pb);
999 get_be32(pb); /* always 0x7F000000 */
1000 get_be32(pb); /* bits per channel if sound is uncompressed */
1001 get_be32(pb); /* lcpm format specific flag */
1002 get_be32(pb); /* bytes per audio packet if constant */
1003 get_be32(pb); /* lpcm frames per audio packet if constant */
1006 bits_per_sample = av_get_bits_per_sample(st->codec->codec_id);
1007 if (bits_per_sample) {
1008 st->codec->bits_per_sample = bits_per_sample;
1009 sc->sample_size = (bits_per_sample >> 3) * st->codec->channels;
1012 /* other codec type, just skip (rtp, mp4s, tmcd ...) */
1013 url_fskip(pb, size - (url_ftell(pb) - start_pos));
1015 /* this will read extra atoms at the end (wave, alac, damr, avcC, SMI ...) */
1016 a.size = size - (url_ftell(pb) - start_pos);
1018 mov_read_default(c, pb, a);
1019 else if (a.size > 0)
1020 url_fskip(pb, a.size);
1023 if(st->codec->codec_type==CODEC_TYPE_AUDIO && st->codec->sample_rate==0 && sc->time_scale>1) {
1024 st->codec->sample_rate= sc->time_scale;
1027 /* special codec parameters handling */
1028 switch (st->codec->codec_id) {
1029 #ifdef CONFIG_H261_DECODER
1032 #ifdef CONFIG_H263_DECODER
1035 #ifdef CONFIG_MPEG4_DECODER
1036 case CODEC_ID_MPEG4:
1038 st->codec->width= 0; /* let decoder init width/height */
1039 st->codec->height= 0;
1044 #ifdef CONFIG_VORBIS_DECODER
1045 case CODEC_ID_VORBIS:
1047 case CODEC_ID_MP3ON4:
1048 st->codec->sample_rate= 0; /* let decoder init parameters properly */
1050 #ifdef CONFIG_DV_DEMUXER
1051 case CODEC_ID_DVAUDIO:
1052 c->dv_fctx = av_alloc_format_context();
1053 c->dv_demux = dv_init_demux(c->dv_fctx);
1055 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
1058 sc->dv_audio_container = 1;
1059 st->codec->codec_id = CODEC_ID_PCM_S16LE;
1062 /* no ifdef since parameters are always those */
1063 case CODEC_ID_AMR_WB:
1064 st->codec->sample_rate= 16000;
1065 st->codec->channels= 1; /* really needed */
1067 case CODEC_ID_AMR_NB:
1068 st->codec->sample_rate= 8000;
1069 st->codec->channels= 1; /* really needed */
1072 st->need_parsing = 1;
1081 static int mov_read_stsc(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1083 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1084 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1085 unsigned int i, entries;
1087 get_byte(pb); /* version */
1088 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1090 entries = get_be32(pb);
1092 if(entries >= UINT_MAX / sizeof(MOV_sample_to_chunk_tbl))
1096 av_log(NULL, AV_LOG_DEBUG, "track[%i].stsc.entries = %i\n", c->fc->nb_streams-1, entries);
1098 sc->sample_to_chunk_sz = entries;
1099 sc->sample_to_chunk = (MOV_sample_to_chunk_tbl*) av_malloc(entries * sizeof(MOV_sample_to_chunk_tbl));
1100 if (!sc->sample_to_chunk)
1102 for(i=0; i<entries; i++) {
1103 sc->sample_to_chunk[i].first = get_be32(pb);
1104 sc->sample_to_chunk[i].count = get_be32(pb);
1105 sc->sample_to_chunk[i].id = get_be32(pb);
1110 static int mov_read_stss(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1112 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1113 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1114 unsigned int i, entries;
1116 get_byte(pb); /* version */
1117 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1119 entries = get_be32(pb);
1121 if(entries >= UINT_MAX / sizeof(long))
1124 sc->keyframe_count = entries;
1126 av_log(NULL, AV_LOG_DEBUG, "keyframe_count = %ld\n", sc->keyframe_count);
1128 sc->keyframes = (long*) av_malloc(entries * sizeof(long));
1131 for(i=0; i<entries; i++) {
1132 sc->keyframes[i] = get_be32(pb);
1134 /* av_log(NULL, AV_LOG_DEBUG, "keyframes[]=%ld\n", sc->keyframes[i]); */
1140 static int mov_read_stsz(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1142 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1143 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1144 unsigned int i, entries, sample_size;
1146 get_byte(pb); /* version */
1147 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1149 sample_size = get_be32(pb);
1150 if (!sc->sample_size) /* do not overwrite value computed in stsd */
1151 sc->sample_size = sample_size;
1152 entries = get_be32(pb);
1153 if(entries >= UINT_MAX / sizeof(long))
1156 sc->sample_count = entries;
1161 av_log(NULL, AV_LOG_DEBUG, "sample_size = %ld sample_count = %ld\n", sc->sample_size, sc->sample_count);
1163 sc->sample_sizes = (long*) av_malloc(entries * sizeof(long));
1164 if (!sc->sample_sizes)
1166 for(i=0; i<entries; i++) {
1167 sc->sample_sizes[i] = get_be32(pb);
1169 av_log(NULL, AV_LOG_DEBUG, "sample_sizes[]=%ld\n", sc->sample_sizes[i]);
1175 static int mov_read_stts(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1177 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1178 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1179 unsigned int i, entries;
1181 int64_t total_sample_count=0;
1183 get_byte(pb); /* version */
1184 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1185 entries = get_be32(pb);
1186 if(entries >= UINT_MAX / sizeof(Time2Sample))
1189 sc->stts_count = entries;
1190 sc->stts_data = av_malloc(entries * sizeof(Time2Sample));
1193 av_log(NULL, AV_LOG_DEBUG, "track[%i].stts.entries = %i\n", c->fc->nb_streams-1, entries);
1198 for(i=0; i<entries; i++) {
1199 int sample_duration;
1202 sample_count=get_be32(pb);
1203 sample_duration = get_be32(pb);
1204 sc->stts_data[i].count= sample_count;
1205 sc->stts_data[i].duration= sample_duration;
1207 sc->time_rate= ff_gcd(sc->time_rate, sample_duration);
1209 dprintf("sample_count=%d, sample_duration=%d\n",sample_count,sample_duration);
1211 duration+=(int64_t)sample_duration*sample_count;
1212 total_sample_count+=sample_count;
1215 st->nb_frames= total_sample_count;
1217 st->duration= duration;
1221 static int mov_read_ctts(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1223 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1224 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1225 unsigned int i, entries;
1227 get_byte(pb); /* version */
1228 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1229 entries = get_be32(pb);
1230 if(entries >= UINT_MAX / sizeof(Time2Sample))
1233 sc->ctts_count = entries;
1234 sc->ctts_data = av_malloc(entries * sizeof(Time2Sample));
1236 dprintf("track[%i].ctts.entries = %i\n", c->fc->nb_streams-1, entries);
1238 for(i=0; i<entries; i++) {
1239 int count =get_be32(pb);
1240 int duration =get_be32(pb);
1243 av_log(c->fc, AV_LOG_ERROR, "negative ctts, ignoring\n");
1245 url_fskip(pb, 8 * (entries - i - 1));
1248 sc->ctts_data[i].count = count;
1249 sc->ctts_data[i].duration= duration;
1251 sc->time_rate= ff_gcd(sc->time_rate, duration);
1256 static int mov_read_trak(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1259 MOVStreamContext *sc;
1261 st = av_new_stream(c->fc, c->fc->nb_streams);
1263 sc = (MOVStreamContext*) av_mallocz(sizeof(MOVStreamContext));
1270 st->codec->codec_type = CODEC_TYPE_DATA;
1271 st->start_time = 0; /* XXX: check */
1272 c->streams[c->fc->nb_streams-1] = sc;
1274 return mov_read_default(c, pb, atom);
1277 static int mov_read_tkhd(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1279 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1280 int version = get_byte(pb);
1282 get_byte(pb); get_byte(pb);
1283 get_byte(pb); /* flags */
1285 MOV_TRACK_ENABLED 0x0001
1286 MOV_TRACK_IN_MOVIE 0x0002
1287 MOV_TRACK_IN_PREVIEW 0x0004
1288 MOV_TRACK_IN_POSTER 0x0008
1295 get_be32(pb); /* creation time */
1296 get_be32(pb); /* modification time */
1298 st->id = (int)get_be32(pb); /* track id (NOT 0 !)*/
1299 get_be32(pb); /* reserved */
1300 st->start_time = 0; /* check */
1301 (version == 1) ? get_be64(pb) : get_be32(pb); /* highlevel (considering edits) duration in movie timebase */
1302 get_be32(pb); /* reserved */
1303 get_be32(pb); /* reserved */
1305 get_be16(pb); /* layer */
1306 get_be16(pb); /* alternate group */
1307 get_be16(pb); /* volume */
1308 get_be16(pb); /* reserved */
1310 url_fskip(pb, 36); /* display matrix */
1312 /* those are fixed-point */
1313 get_be32(pb); /* track width */
1314 get_be32(pb); /* track height */
1319 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
1320 /* like the files created with Adobe Premiere 5.0, for samples see */
1321 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
1322 static int mov_read_wide(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1327 return 0; /* continue */
1328 if (get_be32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
1329 url_fskip(pb, atom.size - 4);
1332 atom.type = get_le32(pb);
1335 if (atom.type != MKTAG('m', 'd', 'a', 't')) {
1336 url_fskip(pb, atom.size);
1339 err = mov_read_mdat(c, pb, atom);
1345 static int null_read_packet(void *opaque, uint8_t *buf, int buf_size)
1350 static int mov_read_cmov(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1354 uint8_t *moov_data; /* uncompressed data */
1355 long cmov_len, moov_len;
1358 get_be32(pb); /* dcom atom */
1359 if (get_le32(pb) != MKTAG( 'd', 'c', 'o', 'm' ))
1361 if (get_le32(pb) != MKTAG( 'z', 'l', 'i', 'b' )) {
1362 av_log(NULL, AV_LOG_ERROR, "unknown compression for cmov atom !");
1365 get_be32(pb); /* cmvd atom */
1366 if (get_le32(pb) != MKTAG( 'c', 'm', 'v', 'd' ))
1368 moov_len = get_be32(pb); /* uncompressed size */
1369 cmov_len = atom.size - 6 * 4;
1371 cmov_data = (uint8_t *) av_malloc(cmov_len);
1374 moov_data = (uint8_t *) av_malloc(moov_len);
1379 get_buffer(pb, cmov_data, cmov_len);
1380 if(uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
1382 if(init_put_byte(&ctx, moov_data, moov_len, 0, NULL, null_read_packet, NULL, NULL) != 0)
1384 ctx.buf_end = ctx.buffer + moov_len;
1385 atom.type = MKTAG( 'm', 'o', 'o', 'v' );
1387 atom.size = moov_len;
1389 // { int fd = open("/tmp/uncompheader.mov", O_WRONLY | O_CREAT); write(fd, moov_data, moov_len); close(fd); }
1391 ret = mov_read_default(c, &ctx, atom);
1399 /* edit list atom */
1400 static int mov_read_elst(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1404 get_byte(pb); /* version */
1405 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1406 edit_count= c->streams[c->fc->nb_streams-1]->edit_count = get_be32(pb); /* entries */
1408 for(i=0; i<edit_count; i++){
1409 get_be32(pb); /* Track duration */
1410 get_be32(pb); /* Media time */
1411 get_be32(pb); /* Media rate */
1413 dprintf("track[%i].edit_count = %i\n", c->fc->nb_streams-1, c->streams[c->fc->nb_streams-1]->edit_count);
1417 static const MOVParseTableEntry mov_default_parse_table[] = {
1419 { MKTAG( 'c', 'o', '6', '4' ), mov_read_stco },
1420 { MKTAG( 'c', 'p', 'r', 't' ), mov_read_default },
1421 { MKTAG( 'c', 'r', 'h', 'd' ), mov_read_default },
1422 { MKTAG( 'c', 't', 't', 's' ), mov_read_ctts }, /* composition time to sample */
1423 { MKTAG( 'd', 'i', 'n', 'f' ), mov_read_default }, /* data information */
1424 { MKTAG( 'd', 'p', 'n', 'd' ), mov_read_leaf },
1425 { MKTAG( 'd', 'r', 'e', 'f' ), mov_read_leaf },
1426 { MKTAG( 'e', 'd', 't', 's' ), mov_read_default },
1427 { MKTAG( 'e', 'l', 's', 't' ), mov_read_elst },
1428 { MKTAG( 'e', 'n', 'd', 'a' ), mov_read_enda },
1429 { MKTAG( 'f', 'r', 'e', 'e' ), mov_read_leaf },
1430 { MKTAG( 'f', 't', 'y', 'p' ), mov_read_ftyp },
1431 { MKTAG( 'h', 'd', 'l', 'r' ), mov_read_hdlr },
1432 { MKTAG( 'h', 'i', 'n', 't' ), mov_read_leaf },
1433 { MKTAG( 'h', 'm', 'h', 'd' ), mov_read_leaf },
1434 { MKTAG( 'i', 'o', 'd', 's' ), mov_read_leaf },
1435 { MKTAG( 'j', 'p', '2', 'h' ), mov_read_jp2h },
1436 { MKTAG( 'm', 'd', 'a', 't' ), mov_read_mdat },
1437 { MKTAG( 'm', 'd', 'h', 'd' ), mov_read_mdhd },
1438 { MKTAG( 'm', 'd', 'i', 'a' ), mov_read_default },
1439 { MKTAG( 'm', 'i', 'n', 'f' ), mov_read_default },
1440 { MKTAG( 'm', 'o', 'o', 'v' ), mov_read_moov },
1441 { MKTAG( 'm', 'p', '4', 'a' ), mov_read_default },
1442 { MKTAG( 'm', 'p', '4', 's' ), mov_read_default },
1443 { MKTAG( 'm', 'p', '4', 'v' ), mov_read_default },
1444 { MKTAG( 'm', 'p', 'o', 'd' ), mov_read_leaf },
1445 { MKTAG( 'm', 'v', 'h', 'd' ), mov_read_mvhd },
1446 { MKTAG( 'n', 'm', 'h', 'd' ), mov_read_leaf },
1447 { MKTAG( 'o', 'd', 'h', 'd' ), mov_read_default },
1448 { MKTAG( 's', 'd', 'h', 'd' ), mov_read_default },
1449 { MKTAG( 's', 'k', 'i', 'p' ), mov_read_leaf },
1450 { MKTAG( 's', 'm', 'h', 'd' ), mov_read_leaf }, /* sound media info header */
1451 { MKTAG( 'S', 'M', 'I', ' ' ), mov_read_smi }, /* Sorenson extension ??? */
1452 { MKTAG( 'a', 'l', 'a', 'c' ), mov_read_alac }, /* alac specific atom */
1453 { MKTAG( 'a', 'v', 'c', 'C' ), mov_read_avcC },
1454 { MKTAG( 's', 't', 'b', 'l' ), mov_read_default },
1455 { MKTAG( 's', 't', 'c', 'o' ), mov_read_stco },
1456 { MKTAG( 's', 't', 'd', 'p' ), mov_read_default },
1457 { MKTAG( 's', 't', 's', 'c' ), mov_read_stsc },
1458 { MKTAG( 's', 't', 's', 'd' ), mov_read_stsd }, /* sample description */
1459 { MKTAG( 's', 't', 's', 'h' ), mov_read_default },
1460 { MKTAG( 's', 't', 's', 's' ), mov_read_stss }, /* sync sample */
1461 { MKTAG( 's', 't', 's', 'z' ), mov_read_stsz }, /* sample size */
1462 { MKTAG( 's', 't', 't', 's' ), mov_read_stts },
1463 { MKTAG( 't', 'k', 'h', 'd' ), mov_read_tkhd }, /* track header */
1464 { MKTAG( 't', 'r', 'a', 'k' ), mov_read_trak },
1465 { MKTAG( 't', 'r', 'e', 'f' ), mov_read_default }, /* not really */
1466 { MKTAG( 'u', 'd', 't', 'a' ), mov_read_leaf },
1467 { MKTAG( 'u', 'r', 'l', ' ' ), mov_read_leaf },
1468 { MKTAG( 'u', 'r', 'n', ' ' ), mov_read_leaf },
1469 { MKTAG( 'u', 'u', 'i', 'd' ), mov_read_leaf },
1470 { MKTAG( 'v', 'm', 'h', 'd' ), mov_read_leaf }, /* video media info header */
1471 { MKTAG( 'w', 'a', 'v', 'e' ), mov_read_wave },
1473 { MKTAG( 'M', 'D', 'E', 'S' ), mov_read_leaf },
1475 { MKTAG( 'c', 'h', 'a', 'p' ), mov_read_leaf },
1476 { MKTAG( 'c', 'l', 'i', 'p' ), mov_read_default },
1477 { MKTAG( 'c', 'r', 'g', 'n' ), mov_read_leaf },
1478 { MKTAG( 'c', 't', 'a', 'b' ), mov_read_ctab },
1479 { MKTAG( 'e', 's', 'd', 's' ), mov_read_esds },
1480 { MKTAG( 'k', 'm', 'a', 't' ), mov_read_leaf },
1481 { MKTAG( 'm', 'a', 't', 't' ), mov_read_default },
1482 { MKTAG( 'r', 'd', 'r', 'f' ), mov_read_leaf },
1483 { MKTAG( 'r', 'm', 'd', 'a' ), mov_read_default },
1484 { MKTAG( 'r', 'm', 'd', 'r' ), mov_read_leaf },
1485 { MKTAG( 'r', 'm', 'r', 'a' ), mov_read_default },
1486 { MKTAG( 's', 'c', 'p', 't' ), mov_read_leaf },
1487 { MKTAG( 's', 's', 'r', 'c' ), mov_read_leaf },
1488 { MKTAG( 's', 'y', 'n', 'c' ), mov_read_leaf },
1489 { MKTAG( 't', 'c', 'm', 'd' ), mov_read_leaf },
1490 { MKTAG( 'w', 'i', 'd', 'e' ), mov_read_wide }, /* place holder */
1491 //{ MKTAG( 'r', 'm', 'q', 'u' ), mov_read_leaf },
1493 { MKTAG( 'c', 'm', 'o', 'v' ), mov_read_cmov },
1495 { MKTAG( 'c', 'm', 'o', 'v' ), mov_read_leaf },
1497 { 0L, mov_read_leaf }
1500 static void mov_free_stream_context(MOVStreamContext *sc)
1503 av_freep(&sc->ctts_data);
1508 /* XXX: is it sufficient ? */
1509 static int mov_probe(AVProbeData *p)
1511 unsigned int offset;
1515 /* check file header */
1516 if (p->buf_size <= 12)
1520 /* ignore invalid offset */
1521 if ((offset + 8) > (unsigned int)p->buf_size)
1523 tag = LE_32(p->buf + offset + 4);
1525 /* check for obvious tags */
1526 case MKTAG( 'j', 'P', ' ', ' ' ): /* jpeg 2000 signature */
1527 case MKTAG( 'm', 'o', 'o', 'v' ):
1528 case MKTAG( 'm', 'd', 'a', 't' ):
1529 case MKTAG( 'p', 'n', 'o', 't' ): /* detect movs with preview pics like ew.mov and april.mov */
1530 case MKTAG( 'u', 'd', 't', 'a' ): /* Packet Video PVAuthor adds this and a lot of more junk */
1531 return AVPROBE_SCORE_MAX;
1532 /* those are more common words, so rate then a bit less */
1533 case MKTAG( 'w', 'i', 'd', 'e' ):
1534 case MKTAG( 'f', 'r', 'e', 'e' ):
1535 case MKTAG( 'j', 'u', 'n', 'k' ):
1536 case MKTAG( 'p', 'i', 'c', 't' ):
1537 return AVPROBE_SCORE_MAX - 5;
1538 case MKTAG( 'f', 't', 'y', 'p' ):
1539 case MKTAG( 's', 'k', 'i', 'p' ):
1540 case MKTAG( 'u', 'u', 'i', 'd' ):
1541 offset = BE_32(p->buf+offset) + offset;
1542 /* if we only find those cause probedata is too small at least rate them */
1543 score = AVPROBE_SCORE_MAX - 50;
1546 /* unrecognized tag */
1553 static void mov_build_index(MOVContext *mov, AVStream *st)
1555 MOVStreamContext *sc = st->priv_data;
1556 offset_t current_offset;
1557 int64_t current_dts = 0;
1563 if (sc->sample_sizes || st->codec->codec_type == CODEC_TYPE_VIDEO || sc->dv_audio_container) {
1564 int keyframe, sample_size;
1565 int current_sample = 0;
1566 int stts_sample = 0;
1569 st->nb_frames = sc->sample_count;
1570 for (i = 0; i < sc->chunk_count; i++) {
1571 current_offset = sc->chunk_offsets[i];
1572 if (stsc_index + 1 < sc->sample_to_chunk_sz && i + 1 == sc->sample_to_chunk[stsc_index + 1].first)
1574 for (j = 0; j < sc->sample_to_chunk[stsc_index].count; j++) {
1575 keyframe = !sc->keyframe_count || current_sample + 1 == sc->keyframes[stss_index];
1578 if (stss_index + 1 < sc->keyframe_count)
1581 sample_size = sc->sample_size > 0 ? sc->sample_size : sc->sample_sizes[current_sample];
1582 dprintf("AVIndex stream %d, sample %d, offset %llx, dts %lld, size %d, distance %d, keyframe %d\n",
1583 st->index, current_sample, current_offset, current_dts, sample_size, distance, keyframe);
1584 av_add_index_entry(st, current_offset, current_dts, sample_size, distance, keyframe ? AVINDEX_KEYFRAME : 0);
1585 current_offset += sample_size;
1586 assert(sc->stts_data[stts_index].duration % sc->time_rate == 0);
1587 current_dts += sc->stts_data[stts_index].duration / sc->time_rate;
1590 if (current_sample + 1 < sc->sample_count)
1592 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
1598 } else { /* read whole chunk */
1599 int chunk_samples, chunk_size, chunk_duration;
1601 for (i = 0; i < sc->chunk_count; i++) {
1602 current_offset = sc->chunk_offsets[i];
1603 if (stsc_index + 1 < sc->sample_to_chunk_sz && i + 1 == sc->sample_to_chunk[stsc_index + 1].first)
1605 chunk_samples = sc->sample_to_chunk[stsc_index].count;
1606 /* get chunk size */
1607 if (sc->sample_size > 1 || st->codec->codec_id == CODEC_ID_PCM_U8 || st->codec->codec_id == CODEC_ID_PCM_S8)
1608 chunk_size = chunk_samples * sc->sample_size;
1609 else if (sc->sample_size_v1.den > 0 && (chunk_samples * sc->sample_size_v1.num % sc->sample_size_v1.den == 0))
1610 chunk_size = chunk_samples * sc->sample_size_v1.num / sc->sample_size_v1.den;
1611 else { /* workaround to find nearest next chunk offset */
1612 chunk_size = INT_MAX;
1613 for (j = 0; j < mov->total_streams; j++) {
1614 MOVStreamContext *msc = mov->streams[j];
1616 for (k = msc->next_chunk; k < msc->chunk_count; k++) {
1617 if (msc->chunk_offsets[k] > current_offset && msc->chunk_offsets[k] - current_offset < chunk_size) {
1618 chunk_size = msc->chunk_offsets[k] - current_offset;
1619 msc->next_chunk = k;
1624 /* check for last chunk */
1625 if (chunk_size == INT_MAX)
1626 for (j = 0; j < mov->mdat_count; j++) {
1627 dprintf("mdat %d, offset %llx, size %lld, current offset %llx\n",
1628 j, mov->mdat_list[j].offset, mov->mdat_list[j].size, current_offset);
1629 if (mov->mdat_list[j].offset <= current_offset && mov->mdat_list[j].offset + mov->mdat_list[j].size > current_offset)
1630 chunk_size = mov->mdat_list[j].offset + mov->mdat_list[j].size - current_offset;
1632 assert(chunk_size != INT_MAX);
1633 for (j = 0; j < mov->total_streams; j++) {
1634 mov->streams[j]->next_chunk = 0;
1637 av_add_index_entry(st, current_offset, current_dts, chunk_size, 0, AVINDEX_KEYFRAME);
1638 /* get chunk duration */
1640 while (chunk_samples > 0) {
1641 if (chunk_samples < sc->stts_data[stts_index].count) {
1642 chunk_duration += sc->stts_data[stts_index].duration * chunk_samples;
1643 sc->stts_data[stts_index].count -= chunk_samples;
1646 chunk_duration += sc->stts_data[stts_index].duration * chunk_samples;
1647 chunk_samples -= sc->stts_data[stts_index].count;
1648 if (stts_index + 1 < sc->stts_count) {
1653 dprintf("AVIndex stream %d, chunk %d, offset %llx, dts %lld, size %d, duration %d\n",
1654 st->index, i, current_offset, current_dts, chunk_size, chunk_duration);
1655 assert(chunk_duration % sc->time_rate == 0);
1656 current_dts += chunk_duration / sc->time_rate;
1659 /* adjust sample count to avindex entries */
1660 sc->sample_count = st->nb_index_entries;
1663 static int mov_read_header(AVFormatContext *s, AVFormatParameters *ap)
1665 MOVContext *mov = (MOVContext *) s->priv_data;
1666 ByteIOContext *pb = &s->pb;
1668 MOV_atom_t atom = { 0, 0, 0 };
1671 mov->parse_table = mov_default_parse_table;
1673 if(!url_is_streamed(pb)) /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
1674 atom.size = url_fsize(pb);
1676 atom.size = 0x7FFFFFFFFFFFFFFFLL;
1678 /* check MOV header */
1679 err = mov_read_default(mov, pb, atom);
1680 if (err<0 || (!mov->found_moov && !mov->found_mdat)) {
1681 av_log(s, AV_LOG_ERROR, "mov: header not found !!! (err:%d, moov:%d, mdat:%d) pos:%"PRId64"\n",
1682 err, mov->found_moov, mov->found_mdat, url_ftell(pb));
1685 dprintf("on_parse_exit_offset=%d\n", (int) url_ftell(pb));
1687 /* some cleanup : make sure we are on the mdat atom */
1688 if(!url_is_streamed(pb) && (url_ftell(pb) != mov->mdat_offset))
1689 url_fseek(pb, mov->mdat_offset, SEEK_SET);
1691 mov->total_streams = s->nb_streams;
1693 for(i=0; i<mov->total_streams; i++) {
1694 MOVStreamContext *sc = mov->streams[i];
1699 sc->time_scale= mov->time_scale;
1700 av_set_pts_info(s->streams[i], 64, sc->time_rate, sc->time_scale);
1702 if(s->streams[i]->duration != AV_NOPTS_VALUE){
1703 assert(s->streams[i]->duration % sc->time_rate == 0);
1704 s->streams[i]->duration /= sc->time_rate;
1707 mov_build_index(mov, s->streams[i]);
1710 for(i=0; i<mov->total_streams; i++) {
1711 /* dont need those anymore */
1712 av_freep(&mov->streams[i]->chunk_offsets);
1713 av_freep(&mov->streams[i]->sample_to_chunk);
1714 av_freep(&mov->streams[i]->sample_sizes);
1715 av_freep(&mov->streams[i]->keyframes);
1716 av_freep(&mov->streams[i]->stts_data);
1718 av_freep(&mov->mdat_list);
1722 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
1724 MOVContext *mov = s->priv_data;
1725 MOVStreamContext *sc = 0;
1726 AVIndexEntry *sample = 0;
1727 int64_t best_dts = INT64_MAX;
1730 for (i = 0; i < mov->total_streams; i++) {
1731 MOVStreamContext *msc = mov->streams[i];
1733 if (s->streams[i]->discard != AVDISCARD_ALL && msc->current_sample < msc->sample_count) {
1734 AVIndexEntry *current_sample = &s->streams[i]->index_entries[msc->current_sample];
1735 int64_t dts = av_rescale(current_sample->timestamp * (int64_t)msc->time_rate, AV_TIME_BASE, msc->time_scale);
1737 dprintf("stream %d, sample %ld, dts %lld\n", i, msc->current_sample, dts);
1738 if (dts < best_dts) {
1739 sample = current_sample;
1747 /* must be done just before reading, to avoid infinite loop on sample */
1748 sc->current_sample++;
1749 if (sample->pos >= url_fsize(&s->pb)) {
1750 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%llx: partial file\n", sc->ffindex, sample->pos);
1753 #ifdef CONFIG_DV_DEMUXER
1754 if (sc->dv_audio_container) {
1755 dv_get_packet(mov->dv_demux, pkt);
1756 dprintf("dv audio pkt size %d\n", pkt->size);
1759 url_fseek(&s->pb, sample->pos, SEEK_SET);
1760 av_get_packet(&s->pb, pkt, sample->size);
1761 #ifdef CONFIG_DV_DEMUXER
1762 if (mov->dv_demux) {
1763 void *pkt_destruct_func = pkt->destruct;
1764 dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size);
1765 pkt->destruct = pkt_destruct_func;
1769 pkt->stream_index = sc->ffindex;
1770 pkt->dts = sample->timestamp;
1771 if (sc->ctts_data) {
1772 assert(sc->ctts_data[sc->sample_to_ctime_index].duration % sc->time_rate == 0);
1773 pkt->pts = pkt->dts + sc->ctts_data[sc->sample_to_ctime_index].duration / sc->time_rate;
1774 /* update ctts context */
1775 sc->sample_to_ctime_sample++;
1776 if (sc->sample_to_ctime_index < sc->ctts_count && sc->ctts_data[sc->sample_to_ctime_index].count == sc->sample_to_ctime_sample) {
1777 sc->sample_to_ctime_index++;
1778 sc->sample_to_ctime_sample = 0;
1781 pkt->pts = pkt->dts;
1783 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? PKT_FLAG_KEY : 0;
1784 pkt->pos = sample->pos;
1785 dprintf("stream %d, pts %lld, dts %lld, pos 0x%llx, duration %d\n", pkt->stream_index, pkt->pts, pkt->dts, pkt->pos, pkt->duration);
1789 static int mov_seek_stream(AVStream *st, int64_t timestamp, int flags)
1791 MOVStreamContext *sc = st->priv_data;
1792 int sample, time_sample;
1795 sample = av_index_search_timestamp(st, timestamp, flags);
1796 dprintf("stream %d, timestamp %lld, sample %d\n", st->index, timestamp, sample);
1797 if (sample < 0) /* not sure what to do */
1799 sc->current_sample = sample;
1800 dprintf("stream %d, found sample %ld\n", st->index, sc->current_sample);
1801 /* adjust ctts index */
1802 if (sc->ctts_data) {
1804 for (i = 0; i < sc->ctts_count; i++) {
1805 time_sample += sc->ctts_data[i].count;
1806 if (time_sample >= sc->current_sample) {
1807 sc->sample_to_ctime_index = i;
1808 sc->sample_to_ctime_sample = time_sample - sc->current_sample;
1816 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
1819 int64_t seek_timestamp, timestamp;
1823 if (stream_index >= s->nb_streams)
1826 st = s->streams[stream_index];
1827 sample = mov_seek_stream(st, sample_time, flags);
1831 /* adjust seek timestamp to found sample timestamp */
1832 seek_timestamp = st->index_entries[sample].timestamp;
1834 for (i = 0; i < s->nb_streams; i++) {
1836 if (stream_index == i || st->discard == AVDISCARD_ALL)
1839 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
1840 mov_seek_stream(st, timestamp, flags);
1845 static int mov_read_close(AVFormatContext *s)
1848 MOVContext *mov = (MOVContext *) s->priv_data;
1849 for(i=0; i<mov->total_streams; i++)
1850 mov_free_stream_context(mov->streams[i]);
1851 /* free color tabs */
1852 for(i=0; i<mov->ctab_size; i++)
1853 av_freep(&mov->ctab[i]);
1855 for(i=0; i<mov->dv_fctx->nb_streams; i++){
1856 av_freep(&mov->dv_fctx->streams[i]->codec);
1857 av_freep(&mov->dv_fctx->streams[i]);
1859 av_freep(&mov->dv_fctx);
1860 av_freep(&mov->dv_demux);
1862 av_freep(&mov->ctab);
1866 AVInputFormat mov_demuxer = {
1867 "mov,mp4,m4a,3gp,3g2,mj2",
1868 "QuickTime/MPEG4/Motion JPEG 2000 format",