3 * Copyright (c) 2001 Fabrice Bellard.
5 * This file is part of FFmpeg.
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
36 * First version by Francois Revol revol@free.fr
37 * Seek function by Gael Chardon gael.dev@4now.net
39 * Features and limitations:
40 * - reads most of the QT files I have (at least the structure),
41 * the exceptions are .mov with zlib compressed headers ('cmov' section). It shouldn't be hard to implement.
42 * FIXED, Francois Revol, 07/17/2002
43 * - ffmpeg has nearly none of the usual QuickTime codecs,
44 * although I succesfully dumped raw and mp3 audio tracks off .mov files.
45 * Sample QuickTime files with mp3 audio can be found at: http://www.3ivx.com/showcase.html
46 * - .mp4 parsing is still hazardous, although the format really is QuickTime with some minor changes
47 * (to make .mov parser crash maybe ?), despite what they say in the MPEG FAQ at
48 * http://mpeg.telecomitalialab.com/faq.htm
49 * - the code is quite ugly... maybe I won't do it recursive next time :-)
50 * - seek is not supported with files that contain edit list
52 * Funny I didn't know about http://sourceforge.net/projects/qt-ffmpeg/
53 * when coding this :) (it's a writer anyway)
55 * Reference documents:
56 * http://www.geocities.com/xhelmboyx/quicktime/formats/qtm-layout.txt
58 * http://developer.apple.com/documentation/QuickTime/QTFF/
59 * http://developer.apple.com/documentation/QuickTime/QTFF/qtff.pdf
60 * QuickTime is a trademark of Apple (AFAIK :))
63 #include "qtpalette.h"
69 static const CodecTag mov_video_tags[] = {
70 /* { CODEC_ID_, MKTAG('c', 'v', 'i', 'd') }, *//* Cinepak */
71 /* { CODEC_ID_H263, MKTAG('r', 'a', 'w', ' ') }, *//* Uncompressed RGB */
72 /* { CODEC_ID_H263, MKTAG('Y', 'u', 'v', '2') }, *//* Uncompressed YUV422 */
73 /* { CODEC_ID_RAWVIDEO, MKTAG('A', 'V', 'U', 'I') }, *//* YUV with alpha-channel (AVID Uncompressed) */
78 { CODEC_ID_MJPEG, MKTAG('j', 'p', 'e', 'g') }, /* PhotoJPEG */
79 { CODEC_ID_MPEG1VIDEO, MKTAG('m', 'p', 'e', 'g') }, /* MPEG */
80 { CODEC_ID_MJPEG, MKTAG('m', 'j', 'p', 'a') }, /* Motion-JPEG (format A) */
81 { CODEC_ID_MJPEGB, MKTAG('m', 'j', 'p', 'b') }, /* Motion-JPEG (format B) */
82 { CODEC_ID_MJPEG, MKTAG('A', 'V', 'D', 'J') }, /* MJPEG with alpha-channel (AVID JFIF meridien compressed) */
83 /* { CODEC_ID_MJPEG, MKTAG('A', 'V', 'R', 'n') }, *//* MJPEG with alpha-channel (AVID ABVB/Truevision NuVista) */
84 /* { CODEC_ID_GIF, MKTAG('g', 'i', 'f', ' ') }, *//* embedded gif files as frames (usually one "click to play movie" frame) */
86 { CODEC_ID_SVQ1, MKTAG('S', 'V', 'Q', '1') }, /* Sorenson Video v1 */
87 { CODEC_ID_SVQ1, MKTAG('s', 'v', 'q', '1') }, /* Sorenson Video v1 */
88 { CODEC_ID_SVQ1, MKTAG('s', 'v', 'q', 'i') }, /* Sorenson Video v1 (from QT specs)*/
89 { CODEC_ID_SVQ3, MKTAG('S', 'V', 'Q', '3') }, /* Sorenson Video v3 */
90 { CODEC_ID_MPEG4, MKTAG('m', 'p', '4', 'v') },
91 { CODEC_ID_MPEG4, MKTAG('D', 'I', 'V', 'X') }, /* OpenDiVX *//* sample files at http://heroinewarrior.com/xmovie.php3 use this tag */
92 { CODEC_ID_MPEG4, MKTAG('X', 'V', 'I', 'D') },
93 { CODEC_ID_MPEG4, MKTAG('3', 'I', 'V', '2') }, /* experimental: 3IVX files before ivx D4 4.5.1 */
94 /* { CODEC_ID_, MKTAG('I', 'V', '5', '0') }, *//* Indeo 5.0 */
95 { CODEC_ID_H263, MKTAG('h', '2', '6', '3') }, /* H263 */
96 { CODEC_ID_H263, MKTAG('s', '2', '6', '3') }, /* H263 ?? works */
97 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'c', ' ') }, /* DV NTSC */
98 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'c', 'p') }, /* DV PAL */
99 { CODEC_ID_VP3, MKTAG('V', 'P', '3', '1') }, /* On2 VP3 */
100 { CODEC_ID_RPZA, MKTAG('r', 'p', 'z', 'a') }, /* Apple Video (RPZA) */
101 { CODEC_ID_CINEPAK, MKTAG('c', 'v', 'i', 'd') }, /* Cinepak */
102 { CODEC_ID_8BPS, MKTAG('8', 'B', 'P', 'S') }, /* Planar RGB (8BPS) */
103 { CODEC_ID_SMC, MKTAG('s', 'm', 'c', ' ') }, /* Apple Graphics (SMC) */
104 { CODEC_ID_QTRLE, MKTAG('r', 'l', 'e', ' ') }, /* Apple Animation (RLE) */
105 { CODEC_ID_QDRAW, MKTAG('q', 'd', 'r', 'w') }, /* QuickDraw */
106 { CODEC_ID_H264, MKTAG('a', 'v', 'c', '1') }, /* AVC-1/H.264 */
107 { CODEC_ID_MPEG2VIDEO, MKTAG('h', 'd', 'v', '2') }, /* MPEG2 produced by Sony HD camera */
108 { CODEC_ID_MPEG2VIDEO, MKTAG('h', 'd', 'v', '3') }, /* HDV produced by FCP */
109 { CODEC_ID_MPEG2VIDEO, MKTAG('m', 'x', '5', 'n') }, /* MPEG2 IMX NTSC 525/60 50mb/s produced by FCP */
110 { CODEC_ID_MPEG2VIDEO, MKTAG('m', 'x', '5', 'p') }, /* MPEG2 IMX PAL 625/50 50mb/s produced by FCP */
111 { CODEC_ID_MPEG2VIDEO, MKTAG('m', 'x', '3', 'n') }, /* MPEG2 IMX NTSC 525/60 30mb/s produced by FCP */
112 { CODEC_ID_MPEG2VIDEO, MKTAG('m', 'x', '3', 'p') }, /* MPEG2 IMX PAL 625/50 30mb/s produced by FCP */
113 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'p', 'p') }, /* DVCPRO PAL produced by FCP */
114 //{ CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'h', '5') }, /* DVCPRO HD 50i produced by FCP */
115 //{ CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'h', '6') }, /* DVCPRO HD 60i produced by FCP */
116 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', '5', 'p') }, /* DVCPRO50 PAL produced by FCP */
117 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', '5', 'n') }, /* DVCPRO50 NTSC produced by FCP */
118 { CODEC_ID_DVVIDEO, MKTAG('A', 'V', 'd', 'v') }, /* AVID DV */
119 //{ CODEC_ID_JPEG2000, MKTAG('m', 'j', 'p', '2') }, /* JPEG 2000 produced by FCP */
120 { CODEC_ID_RAWVIDEO, MKTAG('2', 'v', 'u', 'y') }, /* UNCOMPRESSED 8BIT 4:2:2 */
121 { CODEC_ID_NONE, 0 },
124 static const CodecTag mov_audio_tags[] = {
125 { CODEC_ID_PCM_S32BE, MKTAG('i', 'n', '3', '2') },
126 { CODEC_ID_PCM_S24BE, MKTAG('i', 'n', '2', '4') },
127 { CODEC_ID_PCM_S16BE, MKTAG('N', 'O', 'N', 'E') }, /* uncompressed */
128 { CODEC_ID_PCM_S16BE, MKTAG('t', 'w', 'o', 's') }, /* 16 bits */
129 { CODEC_ID_PCM_U8, MKTAG('r', 'a', 'w', ' ') }, /* 8 bits unsigned */
130 { CODEC_ID_PCM_S16LE, MKTAG('s', 'o', 'w', 't') }, /* */
131 { CODEC_ID_PCM_MULAW, MKTAG('u', 'l', 'a', 'w') }, /* */
132 { CODEC_ID_PCM_ALAW, MKTAG('a', 'l', 'a', 'w') }, /* */
133 { CODEC_ID_ADPCM_IMA_QT, MKTAG('i', 'm', 'a', '4') }, /* IMA-4 ADPCM */
134 { CODEC_ID_ADPCM_MS, MKTAG('m', 's', 0x00, 0x02) }, /* MS ADPCM */
135 { CODEC_ID_MACE3, MKTAG('M', 'A', 'C', '3') }, /* Macintosh Audio Compression and Expansion 3:1 */
136 { CODEC_ID_MACE6, MKTAG('M', 'A', 'C', '6') }, /* Macintosh Audio Compression and Expansion 6:1 */
138 { CODEC_ID_MP3, MKTAG('.', 'm', 'p', '3') }, /* MPEG layer 3 */ /* sample files at http://www.3ivx.com/showcase.html use this tag */
139 { CODEC_ID_MP2, 0x6D730055 }, /* MPEG layer 3 */
140 { CODEC_ID_MP2, 0x5500736D }, /* MPEG layer 3 *//* XXX: check endianness */
141 /* { CODEC_ID_OGG_VORBIS, MKTAG('O', 'g', 'g', 'S') }, *//* sample files at http://heroinewarrior.com/xmovie.php3 use this tag */
143 { CODEC_ID_AAC, MKTAG('m', 'p', '4', 'a') }, /* MPEG-4 AAC */
144 /* The standard for mpeg4 audio is still not normalised AFAIK anyway */
145 { CODEC_ID_AMR_NB, MKTAG('s', 'a', 'm', 'r') }, /* AMR-NB 3gp */
146 { CODEC_ID_AMR_WB, MKTAG('s', 'a', 'w', 'b') }, /* AMR-WB 3gp */
147 { CODEC_ID_AC3, MKTAG('m', 's', 0x20, 0x00) }, /* Dolby AC-3 */
148 { CODEC_ID_ALAC,MKTAG('a', 'l', 'a', 'c') }, /* Apple Lossless */
149 { CODEC_ID_QDM2,MKTAG('Q', 'D', 'M', '2') }, /* QDM2 */
150 { CODEC_ID_DVAUDIO, MKTAG('v', 'd', 'v', 'a') },
151 { CODEC_ID_DVAUDIO, MKTAG('d', 'v', 'c', 'a') },
152 { CODEC_ID_PCM_S16LE, MKTAG('l', 'p', 'c', 'm') },
153 { CODEC_ID_NONE, 0 },
156 /* the QuickTime file format is quite convoluted...
157 * it has lots of index tables, each indexing something in another one...
158 * Here we just use what is needed to read the chunks
161 typedef struct MOV_sample_to_chunk_tbl {
165 } MOV_sample_to_chunk_tbl;
170 int64_t size; /* total size (excluding the size and type fields) */
180 typedef struct MOV_mdat_atom_s {
187 uint32_t flags; // 24bit
189 /* 0x03 ESDescrTag */
191 #define MP4ODescrTag 0x01
192 #define MP4IODescrTag 0x02
193 #define MP4ESDescrTag 0x03
194 #define MP4DecConfigDescrTag 0x04
195 #define MP4DecSpecificDescrTag 0x05
196 #define MP4SLConfigDescrTag 0x06
197 #define MP4ContentIdDescrTag 0x07
198 #define MP4SupplContentIdDescrTag 0x08
199 #define MP4IPIPtrDescrTag 0x09
200 #define MP4IPMPPtrDescrTag 0x0A
201 #define MP4IPMPDescrTag 0x0B
202 #define MP4RegistrationDescrTag 0x0D
203 #define MP4ESIDIncDescrTag 0x0E
204 #define MP4ESIDRefDescrTag 0x0F
205 #define MP4FileIODescrTag 0x10
206 #define MP4FileODescrTag 0x11
207 #define MP4ExtProfileLevelDescrTag 0x13
208 #define MP4ExtDescrTagsStart 0x80
209 #define MP4ExtDescrTagsEnd 0xFE
210 uint8_t stream_priority;
212 /* 0x04 DecConfigDescrTag */
213 uint8_t object_type_id;
215 /* XXX: really streamType is
216 * only 6bit, followed by:
220 uint32_t buffer_size_db; // 24
221 uint32_t max_bitrate;
222 uint32_t avg_bitrate;
224 /* 0x05 DecSpecificDescrTag */
225 uint8_t decoder_cfg_len;
226 uint8_t *decoder_cfg;
228 /* 0x06 SLConfigDescrTag */
229 uint8_t sl_config_len;
233 struct MOVParseTableEntry;
235 typedef struct MOVStreamContext {
236 int ffindex; /* the ffmpeg stream id */
239 int64_t *chunk_offsets;
241 Time2Sample *stts_data;
243 Time2Sample *ctts_data;
244 int edit_count; /* number of 'edit' (elst atom) */
245 long sample_to_chunk_sz;
246 MOV_sample_to_chunk_tbl *sample_to_chunk;
247 int sample_to_ctime_index;
248 int sample_to_ctime_sample;
258 AVRational sample_size_v1;
259 int dv_audio_container;
262 typedef struct MOVContext {
263 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) */
266 int64_t duration; /* duration of the longest track */
267 int found_moov; /* when both 'moov' and 'mdat' sections has been found */
268 int found_mdat; /* we suppose we have enough data to read the file */
272 /* some streams listed here aren't presented to the ffmpeg API, since they aren't either video nor audio
273 * but we need the info to be able to skip data from those streams in the 'mdat' section
275 MOVStreamContext *streams[MAX_STREAMS];
278 MOV_ctab_t **ctab; /* color tables */
279 const struct MOVParseTableEntry *parse_table; /* could be eventually used to change the table */
280 /* NOTE: for recursion save to/ restore from local variable! */
282 AVPaletteControl palette_control;
283 MOV_mdat_atom_t *mdat_list;
285 DVDemuxContext *dv_demux;
286 AVFormatContext *dv_fctx;
290 /* XXX: it's the first time I make a recursive parser I think... sorry if it's ugly :P */
292 /* those functions parse an atom */
294 1: found what I wanted, exit
295 0: continue to parse next atom
296 -1: error occured, exit
298 typedef int (*mov_parse_function)(MOVContext *ctx, ByteIOContext *pb, MOV_atom_t atom);
300 /* links atom IDs to parse functions */
301 typedef struct MOVParseTableEntry {
303 mov_parse_function func;
304 } MOVParseTableEntry;
306 static int mov_read_leaf(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
309 url_fskip(pb, atom.size);
310 /* url_seek(pb, atom_offset+atom.size, SEEK_SET); */
314 static int mov_read_default(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
316 int64_t total_size = 0;
321 a.offset = atom.offset;
324 atom.size = 0x7fffffffffffffffLL;
325 while(((total_size + 8) < atom.size) && !url_feof(pb) && !err) {
329 a.size = get_be32(pb);
330 a.type = get_le32(pb);
334 dprintf("type: %08x %.4s sz: %"PRIx64" %"PRIx64" %"PRIx64"\n", a.type, (char*)&a.type, a.size, atom.size, total_size);
335 if (a.size == 1) { /* 64 bit extended size */
336 a.size = get_be64(pb) - 8;
341 a.size = atom.size - total_size;
345 for (i = 0; c->parse_table[i].type != 0L
346 && c->parse_table[i].type != a.type; i++)
354 if (c->parse_table[i].type == 0) { /* skip leaf atoms data */
355 url_fskip(pb, a.size);
357 offset_t start_pos = url_ftell(pb);
359 err = (c->parse_table[i].func)(c, pb, a);
360 left = a.size - url_ftell(pb) + start_pos;
361 if (left > 0) /* skip garbage at atom end */
366 total_size += a.size;
369 if (!err && total_size < atom.size && atom.size < 0x7ffff) {
370 url_fskip(pb, atom.size - total_size);
376 static int mov_read_ctab(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
379 url_fskip(pb, atom.size); // for now
381 VERY VERY BROKEN, NEVER execute this, needs rewrite
384 c->ctab = av_realloc(c->ctab, ++c->ctab_size);
385 t = c->ctab[c->ctab_size];
386 t->seed = get_be32(pb);
387 t->flags = get_be16(pb);
388 t->size = get_be16(pb) + 1;
389 len = 2 * t->size * 4;
391 t->clrs = av_malloc(len); // 16bit A R G B
393 get_buffer(pb, t->clrs, len);
400 static int mov_read_hdlr(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
402 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
407 get_byte(pb); /* version */
408 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
411 ctype = get_le32(pb);
412 type = get_le32(pb); /* component subtype */
414 dprintf("ctype= %c%c%c%c (0x%08lx)\n", *((char *)&ctype), ((char *)&ctype)[1], ((char *)&ctype)[2], ((char *)&ctype)[3], (long) ctype);
415 dprintf("stype= %c%c%c%c\n", *((char *)&type), ((char *)&type)[1], ((char *)&type)[2], ((char *)&type)[3]);
416 if(ctype == MKTAG('m', 'h', 'l', 'r')) /* MOV */
420 if(type == MKTAG('v', 'i', 'd', 'e'))
421 st->codec->codec_type = CODEC_TYPE_VIDEO;
422 else if(type == MKTAG('s', 'o', 'u', 'n'))
423 st->codec->codec_type = CODEC_TYPE_AUDIO;
424 get_be32(pb); /* component manufacture */
425 get_be32(pb); /* component flags */
426 get_be32(pb); /* component flags mask */
429 return 0; /* nothing left to read */
430 /* XXX: MP4 uses a C string, not a pascal one */
435 while(get_byte(pb) && (++len < (atom.size - 24)));
437 /* .mov: PASCAL string */
442 url_fskip(pb, atom.size - (url_ftell(pb) - atom.offset));
446 static int mov_mp4_read_descr_len(ByteIOContext *pb)
451 int c = get_byte(pb);
452 len = (len << 7) | (c & 0x7f);
459 static int mov_mp4_read_descr(ByteIOContext *pb, int *tag)
463 len = mov_mp4_read_descr_len(pb);
464 dprintf("MPEG4 description: tag=0x%02x len=%d\n", *tag, len);
468 static int mov_read_esds(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
470 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
471 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
474 /* Well, broken but suffisant for some MP4 streams */
475 get_be32(pb); /* version + flags */
476 len = mov_mp4_read_descr(pb, &tag);
477 if (tag == MP4ESDescrTag) {
478 get_be16(pb); /* ID */
479 get_byte(pb); /* priority */
481 get_be16(pb); /* ID */
483 len = mov_mp4_read_descr(pb, &tag);
484 if (tag == MP4DecConfigDescrTag) {
485 sc->esds.object_type_id = get_byte(pb);
486 sc->esds.stream_type = get_byte(pb);
487 sc->esds.buffer_size_db = get_be24(pb);
488 sc->esds.max_bitrate = get_be32(pb);
489 sc->esds.avg_bitrate = get_be32(pb);
491 st->codec->codec_id= codec_get_id(ff_mov_obj_type, sc->esds.object_type_id);
492 dprintf("esds object type id %d\n", sc->esds.object_type_id);
493 len = mov_mp4_read_descr(pb, &tag);
494 if (tag == MP4DecSpecificDescrTag) {
495 dprintf("Specific MPEG4 header len=%d\n", len);
496 st->codec->extradata = av_mallocz(len + FF_INPUT_BUFFER_PADDING_SIZE);
497 if (st->codec->extradata) {
498 get_buffer(pb, st->codec->extradata, len);
499 st->codec->extradata_size = len;
501 if ((*st->codec->extradata >> 3) == 29) {
502 st->codec->codec_id = CODEC_ID_MP3ON4;
510 /* this atom contains actual media data */
511 static int mov_read_mdat(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
513 if(atom.size == 0) /* wrong one (MP4) */
515 c->mdat_list = av_realloc(c->mdat_list, (c->mdat_count + 1) * sizeof(*c->mdat_list));
516 c->mdat_list[c->mdat_count].offset = atom.offset;
517 c->mdat_list[c->mdat_count].size = atom.size;
520 c->mdat_offset = atom.offset;
521 c->mdat_size = atom.size;
523 return 1; /* found both, just go */
524 url_fskip(pb, atom.size);
525 return 0; /* now go for moov */
528 static int mov_read_ftyp(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
530 uint32_t type = get_le32(pb);
534 case MKTAG('i', 's', 'o', 'm'):
535 case MKTAG('m', 'p', '4', '1'):
536 case MKTAG('m', 'p', '4', '2'):
537 case MKTAG('3', 'g', 'p', '1'):
538 case MKTAG('3', 'g', 'p', '2'):
539 case MKTAG('3', 'g', '2', 'a'):
540 case MKTAG('3', 'g', 'p', '3'):
541 case MKTAG('3', 'g', 'p', '4'):
542 case MKTAG('3', 'g', 'p', '5'):
543 case MKTAG('m', 'm', 'p', '4'): /* Mobile MP4 */
544 case MKTAG('M', '4', 'A', ' '): /* Apple iTunes AAC-LC Audio */
545 case MKTAG('M', '4', 'P', ' '): /* Apple iTunes AAC-LC Protected Audio */
546 case MKTAG('m', 'j', 'p', '2'): /* Motion Jpeg 2000 */
548 case MKTAG('q', 't', ' ', ' '):
550 av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
552 get_be32(pb); /* minor version */
553 url_fskip(pb, atom.size - 8);
557 /* this atom should contain all header atoms */
558 static int mov_read_moov(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
562 err = mov_read_default(c, pb, atom);
563 /* we parsed the 'moov' atom, we can terminate the parsing as soon as we find the 'mdat' */
564 /* so we don't parse the whole file if over a network */
567 return 1; /* found both, just go */
568 return 0; /* now go for mdat */
572 static int mov_read_mdhd(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
574 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
575 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
576 int version = get_byte(pb);
580 return 1; /* unsupported */
582 get_byte(pb); get_byte(pb);
583 get_byte(pb); /* flags */
589 get_be32(pb); /* creation time */
590 get_be32(pb); /* modification time */
593 sc->time_scale = get_be32(pb);
594 st->duration = (version == 1) ? get_be64(pb) : get_be32(pb); /* duration */
596 lang = get_be16(pb); /* language */
597 ff_mov_lang_to_iso639(lang, st->language);
598 get_be16(pb); /* quality */
603 static int mov_read_mvhd(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
605 int version = get_byte(pb); /* version */
606 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
612 get_be32(pb); /* creation time */
613 get_be32(pb); /* modification time */
615 c->time_scale = get_be32(pb); /* time scale */
617 av_log(NULL, AV_LOG_DEBUG, "time scale = %i\n", c->time_scale);
619 c->duration = (version == 1) ? get_be64(pb) : get_be32(pb); /* duration */
620 get_be32(pb); /* preferred scale */
622 get_be16(pb); /* preferred volume */
624 url_fskip(pb, 10); /* reserved */
626 url_fskip(pb, 36); /* display matrix */
628 get_be32(pb); /* preview time */
629 get_be32(pb); /* preview duration */
630 get_be32(pb); /* poster time */
631 get_be32(pb); /* selection time */
632 get_be32(pb); /* selection duration */
633 get_be32(pb); /* current time */
634 get_be32(pb); /* next track ID */
639 static int mov_read_smi(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
641 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
643 if((uint64_t)atom.size > (1<<30))
646 // currently SVQ3 decoder expect full STSD header - so let's fake it
647 // this should be fixed and just SMI header should be passed
648 av_free(st->codec->extradata);
649 st->codec->extradata_size = 0x5a + atom.size;
650 st->codec->extradata = av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
652 if (st->codec->extradata) {
653 strcpy(st->codec->extradata, "SVQ3"); // fake
654 get_buffer(pb, st->codec->extradata + 0x5a, atom.size);
655 dprintf("Reading SMI %"PRId64" %s\n", atom.size, st->codec->extradata + 0x5a);
657 url_fskip(pb, atom.size);
662 static int mov_read_enda(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
664 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
665 int little_endian = get_be16(pb);
668 switch (st->codec->codec_id) {
669 case CODEC_ID_PCM_S24BE:
670 st->codec->codec_id = CODEC_ID_PCM_S24LE;
672 case CODEC_ID_PCM_S32BE:
673 st->codec->codec_id = CODEC_ID_PCM_S32LE;
682 static int mov_read_alac(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
684 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
686 // currently ALAC decoder expect full atom header - so let's fake it
687 // this should be fixed and just ALAC header should be passed
689 av_free(st->codec->extradata);
690 st->codec->extradata_size = 36;
691 st->codec->extradata = av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
693 if (st->codec->extradata) {
694 strcpy(st->codec->extradata + 4, "alac"); // fake
695 get_buffer(pb, st->codec->extradata + 8, 36 - 8);
696 dprintf("Reading alac %d %s\n", st->codec->extradata_size, st->codec->extradata);
698 url_fskip(pb, atom.size);
702 static int mov_read_wave(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
704 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
706 if((uint64_t)atom.size > (1<<30))
709 if (st->codec->codec_id == CODEC_ID_QDM2) {
710 // pass all frma atom to codec, needed at least for QDM2
711 av_free(st->codec->extradata);
712 st->codec->extradata_size = atom.size;
713 st->codec->extradata = av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
715 if (st->codec->extradata) {
716 get_buffer(pb, st->codec->extradata, atom.size);
718 url_fskip(pb, atom.size);
719 } else if (atom.size > 8) { /* to read frma, esds atoms */
720 mov_read_default(c, pb, atom);
722 url_fskip(pb, atom.size);
726 static int mov_read_jp2h(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
728 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
730 if((uint64_t)atom.size > (1<<30))
733 av_free(st->codec->extradata);
735 st->codec->extradata_size = atom.size + 8;
736 st->codec->extradata = av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
738 /* pass all jp2h atom to codec */
739 if (st->codec->extradata) {
740 strcpy(st->codec->extradata + 4, "jp2h");
741 get_buffer(pb, st->codec->extradata + 8, atom.size);
743 url_fskip(pb, atom.size);
747 static int mov_read_avcC(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
749 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
751 if((uint64_t)atom.size > (1<<30))
754 av_free(st->codec->extradata);
756 st->codec->extradata_size = atom.size;
757 st->codec->extradata = av_mallocz(st->codec->extradata_size + FF_INPUT_BUFFER_PADDING_SIZE);
759 if (st->codec->extradata) {
760 get_buffer(pb, st->codec->extradata, atom.size);
762 url_fskip(pb, atom.size);
767 static int mov_read_stco(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
769 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
770 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
771 unsigned int i, entries;
773 get_byte(pb); /* version */
774 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
776 entries = get_be32(pb);
778 if(entries >= UINT_MAX/sizeof(int64_t))
781 sc->chunk_count = entries;
782 sc->chunk_offsets = av_malloc(entries * sizeof(int64_t));
783 if (!sc->chunk_offsets)
785 if (atom.type == MKTAG('s', 't', 'c', 'o')) {
786 for(i=0; i<entries; i++) {
787 sc->chunk_offsets[i] = get_be32(pb);
789 } else if (atom.type == MKTAG('c', 'o', '6', '4')) {
790 for(i=0; i<entries; i++) {
791 sc->chunk_offsets[i] = get_be64(pb);
799 static int mov_read_stsd(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
801 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
802 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
803 int entries, frames_per_sample;
805 uint8_t codec_name[32];
807 /* for palette traversal */
815 unsigned char *color_table;
817 unsigned char r, g, b;
819 get_byte(pb); /* version */
820 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
822 entries = get_be32(pb);
824 while(entries--) { //Parsing Sample description table
826 MOV_atom_t a = { 0, 0, 0 };
827 offset_t start_pos = url_ftell(pb);
828 int size = get_be32(pb); /* size */
829 format = get_le32(pb); /* data format */
831 get_be32(pb); /* reserved */
832 get_be16(pb); /* reserved */
833 get_be16(pb); /* index */
835 if (st->codec->codec_tag) {
836 /* multiple fourcc, just skip for now */
837 url_fskip(pb, size - (url_ftell(pb) - start_pos));
841 st->codec->codec_tag = format;
842 id = codec_get_id(mov_audio_tags, format);
844 st->codec->codec_type = CODEC_TYPE_AUDIO;
845 } else if (format && format != MKTAG('m', 'p', '4', 's')) { /* skip old asf mpeg4 tag */
846 id = codec_get_id(mov_video_tags, format);
848 id = codec_get_id(codec_bmp_tags, format);
850 st->codec->codec_type = CODEC_TYPE_VIDEO;
853 dprintf("size=%d 4CC= %c%c%c%c codec_type=%d\n",
855 (format >> 0) & 0xff, (format >> 8) & 0xff, (format >> 16) & 0xff, (format >> 24) & 0xff,
856 st->codec->codec_type);
858 if(st->codec->codec_type==CODEC_TYPE_VIDEO) {
859 st->codec->codec_id = id;
860 get_be16(pb); /* version */
861 get_be16(pb); /* revision level */
862 get_be32(pb); /* vendor */
863 get_be32(pb); /* temporal quality */
864 get_be32(pb); /* spacial quality */
866 st->codec->width = get_be16(pb); /* width */
867 st->codec->height = get_be16(pb); /* height */
869 get_be32(pb); /* horiz resolution */
870 get_be32(pb); /* vert resolution */
871 get_be32(pb); /* data size, always 0 */
872 frames_per_sample = get_be16(pb); /* frames per samples */
874 av_log(NULL, AV_LOG_DEBUG, "frames/samples = %d\n", frames_per_sample);
876 get_buffer(pb, codec_name, 32); /* codec name, pascal string (FIXME: true for mp4?) */
877 if (codec_name[0] <= 31) {
878 memcpy(st->codec->codec_name, &codec_name[1],codec_name[0]);
879 st->codec->codec_name[codec_name[0]] = 0;
882 st->codec->bits_per_sample = get_be16(pb); /* depth */
883 st->codec->color_table_id = get_be16(pb); /* colortable id */
885 /* figure out the palette situation */
886 color_depth = st->codec->bits_per_sample & 0x1F;
887 color_greyscale = st->codec->bits_per_sample & 0x20;
889 /* if the depth is 2, 4, or 8 bpp, file is palettized */
890 if ((color_depth == 2) || (color_depth == 4) ||
891 (color_depth == 8)) {
893 if (color_greyscale) {
895 /* compute the greyscale palette */
896 color_count = 1 << color_depth;
898 color_dec = 256 / (color_count - 1);
899 for (j = 0; j < color_count; j++) {
900 r = g = b = color_index;
901 c->palette_control.palette[j] =
902 (r << 16) | (g << 8) | (b);
903 color_index -= color_dec;
908 } else if (st->codec->color_table_id & 0x08) {
910 /* if flag bit 3 is set, use the default palette */
911 color_count = 1 << color_depth;
912 if (color_depth == 2)
913 color_table = ff_qt_default_palette_4;
914 else if (color_depth == 4)
915 color_table = ff_qt_default_palette_16;
917 color_table = ff_qt_default_palette_256;
919 for (j = 0; j < color_count; j++) {
920 r = color_table[j * 4 + 0];
921 g = color_table[j * 4 + 1];
922 b = color_table[j * 4 + 2];
923 c->palette_control.palette[j] =
924 (r << 16) | (g << 8) | (b);
929 /* load the palette from the file */
930 color_start = get_be32(pb);
931 color_count = get_be16(pb);
932 color_end = get_be16(pb);
933 for (j = color_start; j <= color_end; j++) {
934 /* each R, G, or B component is 16 bits;
935 * only use the top 8 bits; skip alpha bytes
945 c->palette_control.palette[j] =
946 (r << 16) | (g << 8) | (b);
950 st->codec->palctrl = &c->palette_control;
951 st->codec->palctrl->palette_changed = 1;
953 st->codec->palctrl = NULL;
954 } else if(st->codec->codec_type==CODEC_TYPE_AUDIO) {
956 uint16_t version = get_be16(pb);
958 st->codec->codec_id = id;
959 get_be16(pb); /* revision level */
960 get_be32(pb); /* vendor */
962 st->codec->channels = get_be16(pb); /* channel count */
963 dprintf("audio channels %d\n", st->codec->channels);
964 st->codec->bits_per_sample = get_be16(pb); /* sample size */
965 /* do we need to force to 16 for AMR ? */
967 /* handle specific s8 codec */
968 get_be16(pb); /* compression id = 0*/
969 get_be16(pb); /* packet size = 0 */
971 st->codec->sample_rate = ((get_be32(pb) >> 16));
973 switch (st->codec->codec_id) {
974 case CODEC_ID_PCM_S8:
975 case CODEC_ID_PCM_U8:
976 if (st->codec->bits_per_sample == 16)
977 st->codec->codec_id = CODEC_ID_PCM_S16BE;
979 case CODEC_ID_PCM_S16LE:
980 case CODEC_ID_PCM_S16BE:
981 if (st->codec->bits_per_sample == 8)
982 st->codec->codec_id = CODEC_ID_PCM_S8;
983 else if (st->codec->bits_per_sample == 24)
984 st->codec->codec_id = CODEC_ID_PCM_S24BE;
990 //Read QT version 1 fields. In version 0 theese dont exist
991 dprintf("version =%d mp4=%d\n",version,c->mp4);
993 sc->sample_size_v1.den = get_be32(pb); /* samples per packet */
994 get_be32(pb); /* bytes per packet */
995 sc->sample_size_v1.num = get_be32(pb); /* bytes per frame */
996 get_be32(pb); /* bytes per sample */
997 } else if(version==2) {
998 get_be32(pb); /* sizeof struct only */
999 st->codec->sample_rate = av_int2dbl(get_be64(pb)); /* float 64 */
1000 st->codec->channels = get_be32(pb);
1001 get_be32(pb); /* always 0x7F000000 */
1002 get_be32(pb); /* bits per channel if sound is uncompressed */
1003 get_be32(pb); /* lcpm format specific flag */
1004 get_be32(pb); /* bytes per audio packet if constant */
1005 get_be32(pb); /* lpcm frames per audio packet if constant */
1008 bits_per_sample = av_get_bits_per_sample(st->codec->codec_id);
1009 if (bits_per_sample) {
1010 st->codec->bits_per_sample = bits_per_sample;
1011 sc->sample_size = (bits_per_sample >> 3) * st->codec->channels;
1014 /* other codec type, just skip (rtp, mp4s, tmcd ...) */
1015 url_fskip(pb, size - (url_ftell(pb) - start_pos));
1017 /* this will read extra atoms at the end (wave, alac, damr, avcC, SMI ...) */
1018 a.size = size - (url_ftell(pb) - start_pos);
1020 mov_read_default(c, pb, a);
1021 else if (a.size > 0)
1022 url_fskip(pb, a.size);
1025 if(st->codec->codec_type==CODEC_TYPE_AUDIO && st->codec->sample_rate==0 && sc->time_scale>1) {
1026 st->codec->sample_rate= sc->time_scale;
1029 /* special codec parameters handling */
1030 switch (st->codec->codec_id) {
1031 #ifdef CONFIG_H261_DECODER
1034 #ifdef CONFIG_H263_DECODER
1037 #ifdef CONFIG_MPEG4_DECODER
1038 case CODEC_ID_MPEG4:
1040 st->codec->width= 0; /* let decoder init width/height */
1041 st->codec->height= 0;
1046 #ifdef CONFIG_VORBIS_DECODER
1047 case CODEC_ID_VORBIS:
1049 case CODEC_ID_MP3ON4:
1050 st->codec->sample_rate= 0; /* let decoder init parameters properly */
1052 #ifdef CONFIG_DV_DEMUXER
1053 case CODEC_ID_DVAUDIO:
1054 c->dv_fctx = av_alloc_format_context();
1055 c->dv_demux = dv_init_demux(c->dv_fctx);
1057 av_log(c->fc, AV_LOG_ERROR, "dv demux context init error\n");
1060 sc->dv_audio_container = 1;
1061 st->codec->codec_id = CODEC_ID_PCM_S16LE;
1064 /* no ifdef since parameters are always those */
1065 case CODEC_ID_AMR_WB:
1066 st->codec->sample_rate= 16000;
1067 st->codec->channels= 1; /* really needed */
1069 case CODEC_ID_AMR_NB:
1070 st->codec->sample_rate= 8000;
1071 st->codec->channels= 1; /* really needed */
1074 st->need_parsing = 1;
1083 static int mov_read_stsc(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1085 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1086 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1087 unsigned int i, entries;
1089 get_byte(pb); /* version */
1090 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1092 entries = get_be32(pb);
1094 if(entries >= UINT_MAX / sizeof(MOV_sample_to_chunk_tbl))
1098 av_log(NULL, AV_LOG_DEBUG, "track[%i].stsc.entries = %i\n", c->fc->nb_streams-1, entries);
1100 sc->sample_to_chunk_sz = entries;
1101 sc->sample_to_chunk = av_malloc(entries * sizeof(MOV_sample_to_chunk_tbl));
1102 if (!sc->sample_to_chunk)
1104 for(i=0; i<entries; i++) {
1105 sc->sample_to_chunk[i].first = get_be32(pb);
1106 sc->sample_to_chunk[i].count = get_be32(pb);
1107 sc->sample_to_chunk[i].id = get_be32(pb);
1112 static int mov_read_stss(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1114 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1115 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1116 unsigned int i, entries;
1118 get_byte(pb); /* version */
1119 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1121 entries = get_be32(pb);
1123 if(entries >= UINT_MAX / sizeof(long))
1126 sc->keyframe_count = entries;
1128 av_log(NULL, AV_LOG_DEBUG, "keyframe_count = %ld\n", sc->keyframe_count);
1130 sc->keyframes = av_malloc(entries * sizeof(long));
1133 for(i=0; i<entries; i++) {
1134 sc->keyframes[i] = get_be32(pb);
1136 /* av_log(NULL, AV_LOG_DEBUG, "keyframes[]=%ld\n", sc->keyframes[i]); */
1142 static int mov_read_stsz(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1144 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1145 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1146 unsigned int i, entries, sample_size;
1148 get_byte(pb); /* version */
1149 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1151 sample_size = get_be32(pb);
1152 if (!sc->sample_size) /* do not overwrite value computed in stsd */
1153 sc->sample_size = sample_size;
1154 entries = get_be32(pb);
1155 if(entries >= UINT_MAX / sizeof(long))
1158 sc->sample_count = entries;
1163 av_log(NULL, AV_LOG_DEBUG, "sample_size = %ld sample_count = %ld\n", sc->sample_size, sc->sample_count);
1165 sc->sample_sizes = av_malloc(entries * sizeof(long));
1166 if (!sc->sample_sizes)
1168 for(i=0; i<entries; i++) {
1169 sc->sample_sizes[i] = get_be32(pb);
1171 av_log(NULL, AV_LOG_DEBUG, "sample_sizes[]=%ld\n", sc->sample_sizes[i]);
1177 static int mov_read_stts(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1179 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1180 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1181 unsigned int i, entries;
1183 int64_t total_sample_count=0;
1185 get_byte(pb); /* version */
1186 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1187 entries = get_be32(pb);
1188 if(entries >= UINT_MAX / sizeof(Time2Sample))
1191 sc->stts_count = entries;
1192 sc->stts_data = av_malloc(entries * sizeof(Time2Sample));
1195 av_log(NULL, AV_LOG_DEBUG, "track[%i].stts.entries = %i\n", c->fc->nb_streams-1, entries);
1200 for(i=0; i<entries; i++) {
1201 int sample_duration;
1204 sample_count=get_be32(pb);
1205 sample_duration = get_be32(pb);
1206 sc->stts_data[i].count= sample_count;
1207 sc->stts_data[i].duration= sample_duration;
1209 sc->time_rate= ff_gcd(sc->time_rate, sample_duration);
1211 dprintf("sample_count=%d, sample_duration=%d\n",sample_count,sample_duration);
1213 duration+=(int64_t)sample_duration*sample_count;
1214 total_sample_count+=sample_count;
1217 st->nb_frames= total_sample_count;
1219 st->duration= duration;
1223 static int mov_read_ctts(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1225 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1226 MOVStreamContext *sc = (MOVStreamContext *)st->priv_data;
1227 unsigned int i, entries;
1229 get_byte(pb); /* version */
1230 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1231 entries = get_be32(pb);
1232 if(entries >= UINT_MAX / sizeof(Time2Sample))
1235 sc->ctts_count = entries;
1236 sc->ctts_data = av_malloc(entries * sizeof(Time2Sample));
1238 dprintf("track[%i].ctts.entries = %i\n", c->fc->nb_streams-1, entries);
1240 for(i=0; i<entries; i++) {
1241 int count =get_be32(pb);
1242 int duration =get_be32(pb);
1245 av_log(c->fc, AV_LOG_ERROR, "negative ctts, ignoring\n");
1247 url_fskip(pb, 8 * (entries - i - 1));
1250 sc->ctts_data[i].count = count;
1251 sc->ctts_data[i].duration= duration;
1253 sc->time_rate= ff_gcd(sc->time_rate, duration);
1258 static int mov_read_trak(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1261 MOVStreamContext *sc;
1263 st = av_new_stream(c->fc, c->fc->nb_streams);
1265 sc = av_mallocz(sizeof(MOVStreamContext));
1272 st->codec->codec_type = CODEC_TYPE_DATA;
1273 st->start_time = 0; /* XXX: check */
1274 c->streams[c->fc->nb_streams-1] = sc;
1276 return mov_read_default(c, pb, atom);
1279 static int mov_read_tkhd(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1281 AVStream *st = c->fc->streams[c->fc->nb_streams-1];
1282 int version = get_byte(pb);
1284 get_byte(pb); get_byte(pb);
1285 get_byte(pb); /* flags */
1287 MOV_TRACK_ENABLED 0x0001
1288 MOV_TRACK_IN_MOVIE 0x0002
1289 MOV_TRACK_IN_PREVIEW 0x0004
1290 MOV_TRACK_IN_POSTER 0x0008
1297 get_be32(pb); /* creation time */
1298 get_be32(pb); /* modification time */
1300 st->id = (int)get_be32(pb); /* track id (NOT 0 !)*/
1301 get_be32(pb); /* reserved */
1302 st->start_time = 0; /* check */
1303 (version == 1) ? get_be64(pb) : get_be32(pb); /* highlevel (considering edits) duration in movie timebase */
1304 get_be32(pb); /* reserved */
1305 get_be32(pb); /* reserved */
1307 get_be16(pb); /* layer */
1308 get_be16(pb); /* alternate group */
1309 get_be16(pb); /* volume */
1310 get_be16(pb); /* reserved */
1312 url_fskip(pb, 36); /* display matrix */
1314 /* those are fixed-point */
1315 get_be32(pb); /* track width */
1316 get_be32(pb); /* track height */
1321 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
1322 /* like the files created with Adobe Premiere 5.0, for samples see */
1323 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
1324 static int mov_read_wide(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1329 return 0; /* continue */
1330 if (get_be32(pb) != 0) { /* 0 sized mdat atom... use the 'wide' atom size */
1331 url_fskip(pb, atom.size - 4);
1334 atom.type = get_le32(pb);
1337 if (atom.type != MKTAG('m', 'd', 'a', 't')) {
1338 url_fskip(pb, atom.size);
1341 err = mov_read_mdat(c, pb, atom);
1347 static int null_read_packet(void *opaque, uint8_t *buf, int buf_size)
1352 static int mov_read_cmov(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1356 uint8_t *moov_data; /* uncompressed data */
1357 long cmov_len, moov_len;
1360 get_be32(pb); /* dcom atom */
1361 if (get_le32(pb) != MKTAG( 'd', 'c', 'o', 'm' ))
1363 if (get_le32(pb) != MKTAG( 'z', 'l', 'i', 'b' )) {
1364 av_log(NULL, AV_LOG_ERROR, "unknown compression for cmov atom !");
1367 get_be32(pb); /* cmvd atom */
1368 if (get_le32(pb) != MKTAG( 'c', 'm', 'v', 'd' ))
1370 moov_len = get_be32(pb); /* uncompressed size */
1371 cmov_len = atom.size - 6 * 4;
1373 cmov_data = av_malloc(cmov_len);
1376 moov_data = av_malloc(moov_len);
1381 get_buffer(pb, cmov_data, cmov_len);
1382 if(uncompress (moov_data, (uLongf *) &moov_len, (const Bytef *)cmov_data, cmov_len) != Z_OK)
1384 if(init_put_byte(&ctx, moov_data, moov_len, 0, NULL, null_read_packet, NULL, NULL) != 0)
1386 ctx.buf_end = ctx.buffer + moov_len;
1387 atom.type = MKTAG( 'm', 'o', 'o', 'v' );
1389 atom.size = moov_len;
1391 // { int fd = open("/tmp/uncompheader.mov", O_WRONLY | O_CREAT); write(fd, moov_data, moov_len); close(fd); }
1393 ret = mov_read_default(c, &ctx, atom);
1401 /* edit list atom */
1402 static int mov_read_elst(MOVContext *c, ByteIOContext *pb, MOV_atom_t atom)
1406 get_byte(pb); /* version */
1407 get_byte(pb); get_byte(pb); get_byte(pb); /* flags */
1408 edit_count= c->streams[c->fc->nb_streams-1]->edit_count = get_be32(pb); /* entries */
1410 for(i=0; i<edit_count; i++){
1411 get_be32(pb); /* Track duration */
1412 get_be32(pb); /* Media time */
1413 get_be32(pb); /* Media rate */
1415 dprintf("track[%i].edit_count = %i\n", c->fc->nb_streams-1, c->streams[c->fc->nb_streams-1]->edit_count);
1419 static const MOVParseTableEntry mov_default_parse_table[] = {
1421 { MKTAG( 'c', 'o', '6', '4' ), mov_read_stco },
1422 { MKTAG( 'c', 'p', 'r', 't' ), mov_read_default },
1423 { MKTAG( 'c', 'r', 'h', 'd' ), mov_read_default },
1424 { MKTAG( 'c', 't', 't', 's' ), mov_read_ctts }, /* composition time to sample */
1425 { MKTAG( 'd', 'i', 'n', 'f' ), mov_read_default }, /* data information */
1426 { MKTAG( 'd', 'p', 'n', 'd' ), mov_read_leaf },
1427 { MKTAG( 'd', 'r', 'e', 'f' ), mov_read_leaf },
1428 { MKTAG( 'e', 'd', 't', 's' ), mov_read_default },
1429 { MKTAG( 'e', 'l', 's', 't' ), mov_read_elst },
1430 { MKTAG( 'e', 'n', 'd', 'a' ), mov_read_enda },
1431 { MKTAG( 'f', 'r', 'e', 'e' ), mov_read_leaf },
1432 { MKTAG( 'f', 't', 'y', 'p' ), mov_read_ftyp },
1433 { MKTAG( 'h', 'd', 'l', 'r' ), mov_read_hdlr },
1434 { MKTAG( 'h', 'i', 'n', 't' ), mov_read_leaf },
1435 { MKTAG( 'h', 'm', 'h', 'd' ), mov_read_leaf },
1436 { MKTAG( 'i', 'o', 'd', 's' ), mov_read_leaf },
1437 { MKTAG( 'j', 'p', '2', 'h' ), mov_read_jp2h },
1438 { MKTAG( 'm', 'd', 'a', 't' ), mov_read_mdat },
1439 { MKTAG( 'm', 'd', 'h', 'd' ), mov_read_mdhd },
1440 { MKTAG( 'm', 'd', 'i', 'a' ), mov_read_default },
1441 { MKTAG( 'm', 'i', 'n', 'f' ), mov_read_default },
1442 { MKTAG( 'm', 'o', 'o', 'v' ), mov_read_moov },
1443 { MKTAG( 'm', 'p', '4', 'a' ), mov_read_default },
1444 { MKTAG( 'm', 'p', '4', 's' ), mov_read_default },
1445 { MKTAG( 'm', 'p', '4', 'v' ), mov_read_default },
1446 { MKTAG( 'm', 'p', 'o', 'd' ), mov_read_leaf },
1447 { MKTAG( 'm', 'v', 'h', 'd' ), mov_read_mvhd },
1448 { MKTAG( 'n', 'm', 'h', 'd' ), mov_read_leaf },
1449 { MKTAG( 'o', 'd', 'h', 'd' ), mov_read_default },
1450 { MKTAG( 's', 'd', 'h', 'd' ), mov_read_default },
1451 { MKTAG( 's', 'k', 'i', 'p' ), mov_read_leaf },
1452 { MKTAG( 's', 'm', 'h', 'd' ), mov_read_leaf }, /* sound media info header */
1453 { MKTAG( 'S', 'M', 'I', ' ' ), mov_read_smi }, /* Sorenson extension ??? */
1454 { MKTAG( 'a', 'l', 'a', 'c' ), mov_read_alac }, /* alac specific atom */
1455 { MKTAG( 'a', 'v', 'c', 'C' ), mov_read_avcC },
1456 { MKTAG( 's', 't', 'b', 'l' ), mov_read_default },
1457 { MKTAG( 's', 't', 'c', 'o' ), mov_read_stco },
1458 { MKTAG( 's', 't', 'd', 'p' ), mov_read_default },
1459 { MKTAG( 's', 't', 's', 'c' ), mov_read_stsc },
1460 { MKTAG( 's', 't', 's', 'd' ), mov_read_stsd }, /* sample description */
1461 { MKTAG( 's', 't', 's', 'h' ), mov_read_default },
1462 { MKTAG( 's', 't', 's', 's' ), mov_read_stss }, /* sync sample */
1463 { MKTAG( 's', 't', 's', 'z' ), mov_read_stsz }, /* sample size */
1464 { MKTAG( 's', 't', 't', 's' ), mov_read_stts },
1465 { MKTAG( 't', 'k', 'h', 'd' ), mov_read_tkhd }, /* track header */
1466 { MKTAG( 't', 'r', 'a', 'k' ), mov_read_trak },
1467 { MKTAG( 't', 'r', 'e', 'f' ), mov_read_default }, /* not really */
1468 { MKTAG( 'u', 'd', 't', 'a' ), mov_read_leaf },
1469 { MKTAG( 'u', 'r', 'l', ' ' ), mov_read_leaf },
1470 { MKTAG( 'u', 'r', 'n', ' ' ), mov_read_leaf },
1471 { MKTAG( 'u', 'u', 'i', 'd' ), mov_read_leaf },
1472 { MKTAG( 'v', 'm', 'h', 'd' ), mov_read_leaf }, /* video media info header */
1473 { MKTAG( 'w', 'a', 'v', 'e' ), mov_read_wave },
1475 { MKTAG( 'M', 'D', 'E', 'S' ), mov_read_leaf },
1477 { MKTAG( 'c', 'h', 'a', 'p' ), mov_read_leaf },
1478 { MKTAG( 'c', 'l', 'i', 'p' ), mov_read_default },
1479 { MKTAG( 'c', 'r', 'g', 'n' ), mov_read_leaf },
1480 { MKTAG( 'c', 't', 'a', 'b' ), mov_read_ctab },
1481 { MKTAG( 'e', 's', 'd', 's' ), mov_read_esds },
1482 { MKTAG( 'k', 'm', 'a', 't' ), mov_read_leaf },
1483 { MKTAG( 'm', 'a', 't', 't' ), mov_read_default },
1484 { MKTAG( 'r', 'd', 'r', 'f' ), mov_read_leaf },
1485 { MKTAG( 'r', 'm', 'd', 'a' ), mov_read_default },
1486 { MKTAG( 'r', 'm', 'd', 'r' ), mov_read_leaf },
1487 { MKTAG( 'r', 'm', 'r', 'a' ), mov_read_default },
1488 { MKTAG( 's', 'c', 'p', 't' ), mov_read_leaf },
1489 { MKTAG( 's', 's', 'r', 'c' ), mov_read_leaf },
1490 { MKTAG( 's', 'y', 'n', 'c' ), mov_read_leaf },
1491 { MKTAG( 't', 'c', 'm', 'd' ), mov_read_leaf },
1492 { MKTAG( 'w', 'i', 'd', 'e' ), mov_read_wide }, /* place holder */
1493 //{ MKTAG( 'r', 'm', 'q', 'u' ), mov_read_leaf },
1495 { MKTAG( 'c', 'm', 'o', 'v' ), mov_read_cmov },
1497 { MKTAG( 'c', 'm', 'o', 'v' ), mov_read_leaf },
1499 { 0L, mov_read_leaf }
1502 static void mov_free_stream_context(MOVStreamContext *sc)
1505 av_freep(&sc->ctts_data);
1510 /* XXX: is it sufficient ? */
1511 static int mov_probe(AVProbeData *p)
1513 unsigned int offset;
1517 /* check file header */
1518 if (p->buf_size <= 12)
1522 /* ignore invalid offset */
1523 if ((offset + 8) > (unsigned int)p->buf_size)
1525 tag = LE_32(p->buf + offset + 4);
1527 /* check for obvious tags */
1528 case MKTAG( 'j', 'P', ' ', ' ' ): /* jpeg 2000 signature */
1529 case MKTAG( 'm', 'o', 'o', 'v' ):
1530 case MKTAG( 'm', 'd', 'a', 't' ):
1531 case MKTAG( 'p', 'n', 'o', 't' ): /* detect movs with preview pics like ew.mov and april.mov */
1532 case MKTAG( 'u', 'd', 't', 'a' ): /* Packet Video PVAuthor adds this and a lot of more junk */
1533 return AVPROBE_SCORE_MAX;
1534 /* those are more common words, so rate then a bit less */
1535 case MKTAG( 'w', 'i', 'd', 'e' ):
1536 case MKTAG( 'f', 'r', 'e', 'e' ):
1537 case MKTAG( 'j', 'u', 'n', 'k' ):
1538 case MKTAG( 'p', 'i', 'c', 't' ):
1539 return AVPROBE_SCORE_MAX - 5;
1540 case MKTAG( 'f', 't', 'y', 'p' ):
1541 case MKTAG( 's', 'k', 'i', 'p' ):
1542 case MKTAG( 'u', 'u', 'i', 'd' ):
1543 offset = BE_32(p->buf+offset) + offset;
1544 /* if we only find those cause probedata is too small at least rate them */
1545 score = AVPROBE_SCORE_MAX - 50;
1548 /* unrecognized tag */
1555 static void mov_build_index(MOVContext *mov, AVStream *st)
1557 MOVStreamContext *sc = st->priv_data;
1558 offset_t current_offset;
1559 int64_t current_dts = 0;
1565 if (sc->sample_sizes || st->codec->codec_type == CODEC_TYPE_VIDEO || sc->dv_audio_container) {
1566 int keyframe, sample_size;
1567 int current_sample = 0;
1568 int stts_sample = 0;
1571 st->nb_frames = sc->sample_count;
1572 for (i = 0; i < sc->chunk_count; i++) {
1573 current_offset = sc->chunk_offsets[i];
1574 if (stsc_index + 1 < sc->sample_to_chunk_sz && i + 1 == sc->sample_to_chunk[stsc_index + 1].first)
1576 for (j = 0; j < sc->sample_to_chunk[stsc_index].count; j++) {
1577 keyframe = !sc->keyframe_count || current_sample + 1 == sc->keyframes[stss_index];
1580 if (stss_index + 1 < sc->keyframe_count)
1583 sample_size = sc->sample_size > 0 ? sc->sample_size : sc->sample_sizes[current_sample];
1584 dprintf("AVIndex stream %d, sample %d, offset %llx, dts %lld, size %d, distance %d, keyframe %d\n",
1585 st->index, current_sample, current_offset, current_dts, sample_size, distance, keyframe);
1586 av_add_index_entry(st, current_offset, current_dts, sample_size, distance, keyframe ? AVINDEX_KEYFRAME : 0);
1587 current_offset += sample_size;
1588 assert(sc->stts_data[stts_index].duration % sc->time_rate == 0);
1589 current_dts += sc->stts_data[stts_index].duration / sc->time_rate;
1592 if (current_sample + 1 < sc->sample_count)
1594 if (stts_index + 1 < sc->stts_count && stts_sample == sc->stts_data[stts_index].count) {
1600 } else { /* read whole chunk */
1601 int chunk_samples, chunk_size, chunk_duration;
1603 for (i = 0; i < sc->chunk_count; i++) {
1604 current_offset = sc->chunk_offsets[i];
1605 if (stsc_index + 1 < sc->sample_to_chunk_sz && i + 1 == sc->sample_to_chunk[stsc_index + 1].first)
1607 chunk_samples = sc->sample_to_chunk[stsc_index].count;
1608 /* get chunk size */
1609 if (sc->sample_size > 1 || st->codec->codec_id == CODEC_ID_PCM_U8 || st->codec->codec_id == CODEC_ID_PCM_S8)
1610 chunk_size = chunk_samples * sc->sample_size;
1611 else if (sc->sample_size_v1.den > 0 && (chunk_samples * sc->sample_size_v1.num % sc->sample_size_v1.den == 0))
1612 chunk_size = chunk_samples * sc->sample_size_v1.num / sc->sample_size_v1.den;
1613 else { /* workaround to find nearest next chunk offset */
1614 chunk_size = INT_MAX;
1615 for (j = 0; j < mov->total_streams; j++) {
1616 MOVStreamContext *msc = mov->streams[j];
1618 for (k = msc->next_chunk; k < msc->chunk_count; k++) {
1619 if (msc->chunk_offsets[k] > current_offset && msc->chunk_offsets[k] - current_offset < chunk_size) {
1620 chunk_size = msc->chunk_offsets[k] - current_offset;
1621 msc->next_chunk = k;
1626 /* check for last chunk */
1627 if (chunk_size == INT_MAX)
1628 for (j = 0; j < mov->mdat_count; j++) {
1629 dprintf("mdat %d, offset %llx, size %lld, current offset %llx\n",
1630 j, mov->mdat_list[j].offset, mov->mdat_list[j].size, current_offset);
1631 if (mov->mdat_list[j].offset <= current_offset && mov->mdat_list[j].offset + mov->mdat_list[j].size > current_offset)
1632 chunk_size = mov->mdat_list[j].offset + mov->mdat_list[j].size - current_offset;
1634 assert(chunk_size != INT_MAX);
1635 for (j = 0; j < mov->total_streams; j++) {
1636 mov->streams[j]->next_chunk = 0;
1639 av_add_index_entry(st, current_offset, current_dts, chunk_size, 0, AVINDEX_KEYFRAME);
1640 /* get chunk duration */
1642 while (chunk_samples > 0) {
1643 if (chunk_samples < sc->stts_data[stts_index].count) {
1644 chunk_duration += sc->stts_data[stts_index].duration * chunk_samples;
1645 sc->stts_data[stts_index].count -= chunk_samples;
1648 chunk_duration += sc->stts_data[stts_index].duration * chunk_samples;
1649 chunk_samples -= sc->stts_data[stts_index].count;
1650 if (stts_index + 1 < sc->stts_count) {
1655 dprintf("AVIndex stream %d, chunk %d, offset %llx, dts %lld, size %d, duration %d\n",
1656 st->index, i, current_offset, current_dts, chunk_size, chunk_duration);
1657 assert(chunk_duration % sc->time_rate == 0);
1658 current_dts += chunk_duration / sc->time_rate;
1661 /* adjust sample count to avindex entries */
1662 sc->sample_count = st->nb_index_entries;
1665 static int mov_read_header(AVFormatContext *s, AVFormatParameters *ap)
1667 MOVContext *mov = (MOVContext *) s->priv_data;
1668 ByteIOContext *pb = &s->pb;
1670 MOV_atom_t atom = { 0, 0, 0 };
1673 mov->parse_table = mov_default_parse_table;
1675 if(!url_is_streamed(pb)) /* .mov and .mp4 aren't streamable anyway (only progressive download if moov is before mdat) */
1676 atom.size = url_fsize(pb);
1678 atom.size = 0x7FFFFFFFFFFFFFFFLL;
1680 /* check MOV header */
1681 err = mov_read_default(mov, pb, atom);
1682 if (err<0 || (!mov->found_moov && !mov->found_mdat)) {
1683 av_log(s, AV_LOG_ERROR, "mov: header not found !!! (err:%d, moov:%d, mdat:%d) pos:%"PRId64"\n",
1684 err, mov->found_moov, mov->found_mdat, url_ftell(pb));
1687 dprintf("on_parse_exit_offset=%d\n", (int) url_ftell(pb));
1689 /* some cleanup : make sure we are on the mdat atom */
1690 if(!url_is_streamed(pb) && (url_ftell(pb) != mov->mdat_offset))
1691 url_fseek(pb, mov->mdat_offset, SEEK_SET);
1693 mov->total_streams = s->nb_streams;
1695 for(i=0; i<mov->total_streams; i++) {
1696 MOVStreamContext *sc = mov->streams[i];
1701 sc->time_scale= mov->time_scale;
1702 av_set_pts_info(s->streams[i], 64, sc->time_rate, sc->time_scale);
1704 if(s->streams[i]->duration != AV_NOPTS_VALUE){
1705 assert(s->streams[i]->duration % sc->time_rate == 0);
1706 s->streams[i]->duration /= sc->time_rate;
1709 mov_build_index(mov, s->streams[i]);
1712 for(i=0; i<mov->total_streams; i++) {
1713 /* dont need those anymore */
1714 av_freep(&mov->streams[i]->chunk_offsets);
1715 av_freep(&mov->streams[i]->sample_to_chunk);
1716 av_freep(&mov->streams[i]->sample_sizes);
1717 av_freep(&mov->streams[i]->keyframes);
1718 av_freep(&mov->streams[i]->stts_data);
1720 av_freep(&mov->mdat_list);
1724 static int mov_read_packet(AVFormatContext *s, AVPacket *pkt)
1726 MOVContext *mov = s->priv_data;
1727 MOVStreamContext *sc = 0;
1728 AVIndexEntry *sample = 0;
1729 int64_t best_dts = INT64_MAX;
1732 for (i = 0; i < mov->total_streams; i++) {
1733 MOVStreamContext *msc = mov->streams[i];
1735 if (s->streams[i]->discard != AVDISCARD_ALL && msc->current_sample < msc->sample_count) {
1736 AVIndexEntry *current_sample = &s->streams[i]->index_entries[msc->current_sample];
1737 int64_t dts = av_rescale(current_sample->timestamp * (int64_t)msc->time_rate, AV_TIME_BASE, msc->time_scale);
1739 dprintf("stream %d, sample %ld, dts %lld\n", i, msc->current_sample, dts);
1740 if (dts < best_dts) {
1741 sample = current_sample;
1749 /* must be done just before reading, to avoid infinite loop on sample */
1750 sc->current_sample++;
1751 if (sample->pos >= url_fsize(&s->pb)) {
1752 av_log(mov->fc, AV_LOG_ERROR, "stream %d, offset 0x%llx: partial file\n", sc->ffindex, sample->pos);
1755 #ifdef CONFIG_DV_DEMUXER
1756 if (sc->dv_audio_container) {
1757 dv_get_packet(mov->dv_demux, pkt);
1758 dprintf("dv audio pkt size %d\n", pkt->size);
1761 url_fseek(&s->pb, sample->pos, SEEK_SET);
1762 av_get_packet(&s->pb, pkt, sample->size);
1763 #ifdef CONFIG_DV_DEMUXER
1764 if (mov->dv_demux) {
1765 void *pkt_destruct_func = pkt->destruct;
1766 dv_produce_packet(mov->dv_demux, pkt, pkt->data, pkt->size);
1767 pkt->destruct = pkt_destruct_func;
1771 pkt->stream_index = sc->ffindex;
1772 pkt->dts = sample->timestamp;
1773 if (sc->ctts_data) {
1774 assert(sc->ctts_data[sc->sample_to_ctime_index].duration % sc->time_rate == 0);
1775 pkt->pts = pkt->dts + sc->ctts_data[sc->sample_to_ctime_index].duration / sc->time_rate;
1776 /* update ctts context */
1777 sc->sample_to_ctime_sample++;
1778 if (sc->sample_to_ctime_index < sc->ctts_count && sc->ctts_data[sc->sample_to_ctime_index].count == sc->sample_to_ctime_sample) {
1779 sc->sample_to_ctime_index++;
1780 sc->sample_to_ctime_sample = 0;
1783 pkt->pts = pkt->dts;
1785 pkt->flags |= sample->flags & AVINDEX_KEYFRAME ? PKT_FLAG_KEY : 0;
1786 pkt->pos = sample->pos;
1787 dprintf("stream %d, pts %lld, dts %lld, pos 0x%llx, duration %d\n", pkt->stream_index, pkt->pts, pkt->dts, pkt->pos, pkt->duration);
1791 static int mov_seek_stream(AVStream *st, int64_t timestamp, int flags)
1793 MOVStreamContext *sc = st->priv_data;
1794 int sample, time_sample;
1797 sample = av_index_search_timestamp(st, timestamp, flags);
1798 dprintf("stream %d, timestamp %lld, sample %d\n", st->index, timestamp, sample);
1799 if (sample < 0) /* not sure what to do */
1801 sc->current_sample = sample;
1802 dprintf("stream %d, found sample %ld\n", st->index, sc->current_sample);
1803 /* adjust ctts index */
1804 if (sc->ctts_data) {
1806 for (i = 0; i < sc->ctts_count; i++) {
1807 time_sample += sc->ctts_data[i].count;
1808 if (time_sample >= sc->current_sample) {
1809 sc->sample_to_ctime_index = i;
1810 sc->sample_to_ctime_sample = time_sample - sc->current_sample;
1818 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
1821 int64_t seek_timestamp, timestamp;
1825 if (stream_index >= s->nb_streams)
1828 st = s->streams[stream_index];
1829 sample = mov_seek_stream(st, sample_time, flags);
1833 /* adjust seek timestamp to found sample timestamp */
1834 seek_timestamp = st->index_entries[sample].timestamp;
1836 for (i = 0; i < s->nb_streams; i++) {
1838 if (stream_index == i || st->discard == AVDISCARD_ALL)
1841 timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
1842 mov_seek_stream(st, timestamp, flags);
1847 static int mov_read_close(AVFormatContext *s)
1850 MOVContext *mov = (MOVContext *) s->priv_data;
1851 for(i=0; i<mov->total_streams; i++)
1852 mov_free_stream_context(mov->streams[i]);
1853 /* free color tabs */
1854 for(i=0; i<mov->ctab_size; i++)
1855 av_freep(&mov->ctab[i]);
1857 for(i=0; i<mov->dv_fctx->nb_streams; i++){
1858 av_freep(&mov->dv_fctx->streams[i]->codec);
1859 av_freep(&mov->dv_fctx->streams[i]);
1861 av_freep(&mov->dv_fctx);
1862 av_freep(&mov->dv_demux);
1864 av_freep(&mov->ctab);
1868 AVInputFormat mov_demuxer = {
1869 "mov,mp4,m4a,3gp,3g2,mj2",
1870 "QuickTime/MPEG4/Motion JPEG 2000 format",