AAC in mov/mp4 patch by (Gildas Bazin <gbazin at altern dot org>)
[libav.git] / libavformat / movenc.c
1 /*
2 * MOV, 3GP, MP4 encoder.
3 * Copyright (c) 2003 Thomas Raivio.
4 * Copyright (c) 2004 Gildas Bazin <gbazin at videolan dot org>.
5 *
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
10 *
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19 */
20 #include "avformat.h"
21 #include "avi.h"
22 #include "avio.h"
23
24 #undef NDEBUG
25 #include <assert.h>
26
27 #define MOV_INDEX_CLUSTER_SIZE 16384
28 #define globalTimescale 1000
29
30 #define MODE_MP4 0
31 #define MODE_MOV 1
32 #define MODE_3GP 2
33
34 typedef struct MOVIentry {
35 unsigned int flags, pos, size;
36 unsigned int samplesInChunk;
37 char key_frame;
38 unsigned int entries;
39 } MOVIentry;
40
41 typedef struct MOVIndex {
42 int mode;
43 int entry;
44 int mdat_size;
45 int ents_allocated;
46 long timescale;
47 long time;
48 long trackDuration;
49 long sampleCount;
50 long sampleDuration;
51 int hasKeyframes;
52 int trackID;
53 AVCodecContext *enc;
54
55 int vosLen;
56 uint8_t *vosData;
57 MOVIentry** cluster;
58 } MOVTrack;
59
60 typedef struct {
61 int mode;
62 long time;
63 int nb_streams;
64 int mdat_written;
65 offset_t mdat_pos;
66 long timescale;
67 MOVTrack tracks[MAX_STREAMS];
68 } MOVContext;
69
70 static int mov_write_esds_tag(ByteIOContext *pb, MOVTrack* track);
71
72 //FIXME supprt 64bit varaint with wide placeholders
73 static int updateSize (ByteIOContext *pb, int pos)
74 {
75 long curpos = url_ftell(pb);
76 url_fseek(pb, pos, SEEK_SET);
77 put_be32(pb, curpos - pos); /* rewrite size */
78 url_fseek(pb, curpos, SEEK_SET);
79
80 return curpos - pos;
81 }
82
83 /* Chunk offset atom */
84 static int mov_write_stco_tag(ByteIOContext *pb, MOVTrack* track)
85 {
86 int i;
87 int pos = url_ftell(pb);
88 put_be32(pb, 0); /* size */
89 put_tag(pb, "stco");
90 put_be32(pb, 0); /* version & flags */
91 put_be32(pb, track->entry); /* entry count */
92 for (i=0; i<track->entry; i++) {
93 int cl = i / MOV_INDEX_CLUSTER_SIZE;
94 int id = i % MOV_INDEX_CLUSTER_SIZE;
95 put_be32(pb, track->cluster[cl][id].pos);
96 }
97 return updateSize (pb, pos);
98 }
99
100 /* Sample size atom */
101 static int mov_write_stsz_tag(ByteIOContext *pb, MOVTrack* track)
102 {
103 int equalChunks = 1;
104 int i, j, entries = 0, tst = -1, oldtst = -1;
105
106 int pos = url_ftell(pb);
107 put_be32(pb, 0); /* size */
108 put_tag(pb, "stsz");
109 put_be32(pb, 0); /* version & flags */
110
111 for (i=0; i<track->entry; i++) {
112 int cl = i / MOV_INDEX_CLUSTER_SIZE;
113 int id = i % MOV_INDEX_CLUSTER_SIZE;
114 tst = track->cluster[cl][id].size/track->cluster[cl][id].entries;
115 if(oldtst != -1 && tst != oldtst) {
116 equalChunks = 0;
117 }
118 oldtst = tst;
119 entries += track->cluster[cl][id].entries;
120 }
121 if (equalChunks) {
122 int sSize = track->cluster[0][0].size/track->cluster[0][0].entries;
123 put_be32(pb, sSize); // sample size
124 put_be32(pb, entries); // sample count
125 }
126 else {
127 put_be32(pb, 0); // sample size
128 put_be32(pb, entries); // sample count
129 for (i=0; i<track->entry; i++) {
130 int cl = i / MOV_INDEX_CLUSTER_SIZE;
131 int id = i % MOV_INDEX_CLUSTER_SIZE;
132 for ( j=0; j<track->cluster[cl][id].entries; j++) {
133 put_be32(pb, track->cluster[cl][id].size /
134 track->cluster[cl][id].entries);
135 }
136 }
137 }
138 return updateSize (pb, pos);
139 }
140
141 /* Sample to chunk atom */
142 static int mov_write_stsc_tag(ByteIOContext *pb, MOVTrack* track)
143 {
144 int index = 0, oldval = -1, i, entryPos, curpos;
145
146 int pos = url_ftell(pb);
147 put_be32(pb, 0); /* size */
148 put_tag(pb, "stsc");
149 put_be32(pb, 0); // version & flags
150 entryPos = url_ftell(pb);
151 put_be32(pb, track->entry); // entry count
152 for (i=0; i<track->entry; i++) {
153 int cl = i / MOV_INDEX_CLUSTER_SIZE;
154 int id = i % MOV_INDEX_CLUSTER_SIZE;
155 if(oldval != track->cluster[cl][id].samplesInChunk)
156 {
157 put_be32(pb, i+1); // first chunk
158 put_be32(pb, track->cluster[cl][id].samplesInChunk); // samples per chunk
159 put_be32(pb, 0x1); // sample description index
160 oldval = track->cluster[cl][id].samplesInChunk;
161 index++;
162 }
163 }
164 curpos = url_ftell(pb);
165 url_fseek(pb, entryPos, SEEK_SET);
166 put_be32(pb, index); // rewrite size
167 url_fseek(pb, curpos, SEEK_SET);
168
169 return updateSize (pb, pos);
170 }
171
172 /* Sync sample atom */
173 static int mov_write_stss_tag(ByteIOContext *pb, MOVTrack* track)
174 {
175 long curpos;
176 int i, index = 0, entryPos;
177 int pos = url_ftell(pb);
178 put_be32(pb, 0); // size
179 put_tag(pb, "stss");
180 put_be32(pb, 0); // version & flags
181 entryPos = url_ftell(pb);
182 put_be32(pb, track->entry); // entry count
183 for (i=0; i<track->entry; i++) {
184 int cl = i / MOV_INDEX_CLUSTER_SIZE;
185 int id = i % MOV_INDEX_CLUSTER_SIZE;
186 if(track->cluster[cl][id].key_frame == 1) {
187 put_be32(pb, i+1);
188 index++;
189 }
190 }
191 curpos = url_ftell(pb);
192 url_fseek(pb, entryPos, SEEK_SET);
193 put_be32(pb, index); // rewrite size
194 url_fseek(pb, curpos, SEEK_SET);
195 return updateSize (pb, pos);
196 }
197
198 static int mov_write_damr_tag(ByteIOContext *pb)
199 {
200 put_be32(pb, 0x11); /* size */
201 put_tag(pb, "damr");
202 put_tag(pb, "FFMP");
203 put_byte(pb, 0);
204
205 put_be16(pb, 0x80); /* Mode set (all modes for AMR_NB) */
206 put_be16(pb, 0xa); /* Mode change period (no restriction) */
207 //put_be16(pb, 0x81ff); /* Mode set (all modes for AMR_NB) */
208 //put_be16(pb, 1); /* Mode change period (no restriction) */
209 return 0x11;
210 }
211
212 static int mov_write_wave_tag(ByteIOContext *pb, MOVTrack* track)
213 {
214 int pos = url_ftell(pb);
215
216 put_be32(pb, 0); /* size */
217 put_tag(pb, "wave");
218
219 put_be32(pb, 12); /* size */
220 put_tag(pb, "frma");
221 put_tag(pb, "mp4a");
222
223 put_be32(pb, 12); /* size */
224 put_tag(pb, "mp4a");
225 put_be32(pb, 0);
226
227 mov_write_esds_tag(pb, track);
228
229 put_be32(pb, 12); /* size */
230 put_tag(pb, "srcq");
231 put_be32(pb, 0x40);
232
233 put_be32(pb, 8); /* size */
234 put_be32(pb, 0); /* null tag */
235
236 return updateSize (pb, pos);
237 }
238
239 const CodecTag codec_movaudio_tags[] = {
240 { CODEC_ID_PCM_MULAW, MKTAG('u', 'l', 'a', 'w') },
241 { CODEC_ID_PCM_ALAW, MKTAG('a', 'l', 'a', 'w') },
242 { CODEC_ID_ADPCM_IMA_QT, MKTAG('i', 'm', 'a', '4') },
243 { CODEC_ID_MACE3, MKTAG('M', 'A', 'C', '3') },
244 { CODEC_ID_MACE6, MKTAG('M', 'A', 'C', '6') },
245 { CODEC_ID_AAC, MKTAG('m', 'p', '4', 'a') },
246 { CODEC_ID_AMR_NB, MKTAG('s', 'a', 'm', 'r') },
247 { CODEC_ID_PCM_S16BE, MKTAG('t', 'w', 'o', 's') },
248 { CODEC_ID_PCM_S16LE, MKTAG('s', 'o', 'w', 't') },
249 { CODEC_ID_MP3, MKTAG('.', 'm', 'p', '3') },
250 { 0, 0 },
251 };
252
253 static int mov_write_audio_tag(ByteIOContext *pb, MOVTrack* track)
254 {
255 int pos = url_ftell(pb);
256 int tag;
257
258 put_be32(pb, 0); /* size */
259
260 tag = codec_get_tag(codec_movaudio_tags, track->enc->codec_id);
261 // if no mac fcc found, try with Microsoft tags
262 if (!tag)
263 {
264 int tmp = codec_get_tag(codec_wav_tags, track->enc->codec_id);
265 tag = MKTAG('m', 's', ((tmp >> 8) & 0xff), (tmp & 0xff));
266 }
267 put_le32(pb, tag); // store it byteswapped
268
269 put_be32(pb, 0); /* Reserved */
270 put_be16(pb, 0); /* Reserved */
271 put_be16(pb, 1); /* Data-reference index, XXX == 1 */
272
273 /* SoundDescription */
274 if(track->mode == MODE_MOV && track->enc->codec_id == CODEC_ID_AAC)
275 put_be16(pb, 1); /* Version 1 */
276 else
277 put_be16(pb, 0); /* Version 0 */
278 put_be16(pb, 0); /* Revision level */
279 put_be32(pb, 0); /* Reserved */
280
281 put_be16(pb, track->enc->channels); /* Number of channels */
282 /* TODO: Currently hard-coded to 16-bit, there doesn't seem
283 to be a good way to get number of bits of audio */
284 put_be16(pb, 0x10); /* Reserved */
285
286 if(track->enc->codec_id == CODEC_ID_AAC ||
287 track->enc->codec_id == CODEC_ID_MP3)
288 {
289 put_be16(pb, 0xfffe); /* compression ID (vbr)*/
290 }
291 else
292 {
293 put_be16(pb, 0); /* compression ID (= 0) */
294 }
295 put_be16(pb, 0); /* packet size (= 0) */
296 put_be16(pb, track->timescale); /* Time scale */
297 put_be16(pb, 0); /* Reserved */
298
299 if(track->mode == MODE_MOV && track->enc->codec_id == CODEC_ID_AAC)
300 {
301 /* SoundDescription V1 extended info */
302 put_be32(pb, track->enc->frame_size); /* Samples per packet */
303 put_be32(pb, 1536); /* Bytes per packet */
304 put_be32(pb, 2); /* Bytes per frame */
305 put_be32(pb, 2); /* Bytes per sample */
306 }
307
308 if(track->enc->codec_id == CODEC_ID_AAC) {
309 if( track->mode == MODE_MOV ) mov_write_wave_tag(pb, track);
310 else mov_write_esds_tag(pb, track);
311 }
312 if(track->enc->codec_id == CODEC_ID_AMR_NB)
313 mov_write_damr_tag(pb);
314 return updateSize (pb, pos);
315 }
316
317 static int mov_write_d263_tag(ByteIOContext *pb)
318 {
319 put_be32(pb, 0xf); /* size */
320 put_tag(pb, "d263");
321 put_tag(pb, "FFMP");
322 put_be16(pb, 0x0a);
323 put_byte(pb, 0);
324 return 0xf;
325 }
326
327 /* TODO: No idea about these values */
328 static int mov_write_svq3_tag(ByteIOContext *pb)
329 {
330 put_be32(pb, 0x15);
331 put_tag(pb, "SMI ");
332 put_tag(pb, "SEQH");
333 put_be32(pb, 0x5);
334 put_be32(pb, 0xe2c0211d);
335 put_be32(pb, 0xc0000000);
336 put_byte(pb, 0);
337 return 0x15;
338 }
339
340 static unsigned int descrLength(unsigned int len)
341 {
342 if (len < 0x00000080)
343 return 2 + len;
344 else if (len < 0x00004000)
345 return 3 + len;
346 else if(len < 0x00200000)
347 return 4 + len;
348 else
349 return 5 + len;
350 }
351
352 static void putDescr(ByteIOContext *pb, int tag, int size)
353 {
354 uint32_t len;
355 uint8_t vals[4];
356
357 len = size;
358 vals[3] = (uint8_t)(len & 0x7f);
359 len >>= 7;
360 vals[2] = (uint8_t)((len & 0x7f) | 0x80);
361 len >>= 7;
362 vals[1] = (uint8_t)((len & 0x7f) | 0x80);
363 len >>= 7;
364 vals[0] = (uint8_t)((len & 0x7f) | 0x80);
365
366 put_byte(pb, tag); // DescriptorTag
367
368 if (size < 0x00000080)
369 {
370 put_byte(pb, vals[3]);
371 }
372 else if (size < 0x00004000)
373 {
374 put_byte(pb, vals[2]);
375 put_byte(pb, vals[3]);
376 }
377 else if (size < 0x00200000)
378 {
379 put_byte(pb, vals[1]);
380 put_byte(pb, vals[2]);
381 put_byte(pb, vals[3]);
382 }
383 else if (size < 0x10000000)
384 {
385 put_byte(pb, vals[0]);
386 put_byte(pb, vals[1]);
387 put_byte(pb, vals[2]);
388 put_byte(pb, vals[3]);
389 }
390 }
391
392 static int mov_write_esds_tag(ByteIOContext *pb, MOVTrack* track) // Basic
393 {
394 int decoderSpecificInfoLen = track->vosLen ? descrLength(track->vosLen):0;
395 int pos = url_ftell(pb);
396
397 put_be32(pb, 0); // size
398 put_tag(pb, "esds");
399 put_be32(pb, 0); // Version
400
401 // ES descriptor
402 putDescr(pb, 0x03, 3 + descrLength(13 + decoderSpecificInfoLen) +
403 descrLength(1));
404 put_be16(pb, 0x0001); // ID (= 1)
405 put_byte(pb, 0x00); // flags (= no flags)
406
407 // DecoderConfig descriptor
408 putDescr(pb, 0x04, 13 + decoderSpecificInfoLen);
409
410 if(track->enc->codec_id == CODEC_ID_AAC)
411 put_byte(pb, 0x40); // Object type indication
412 else if(track->enc->codec_id == CODEC_ID_MPEG4)
413 put_byte(pb, 0x20); // Object type indication (Visual 14496-2)
414
415 if(track->enc->codec_type == CODEC_TYPE_AUDIO)
416 put_byte(pb, 0x15); // flags (= Audiostream)
417 else
418 put_byte(pb, 0x11); // flags (= Visualstream)
419
420 put_byte(pb, 0x0); // Buffersize DB (24 bits)
421 put_be16(pb, 0x0dd2); // Buffersize DB
422
423 // TODO: find real values for these
424 put_be32(pb, 0x0002e918); // maxbitrate
425 put_be32(pb, 0x00017e6b); // avg bitrate
426
427 if (track->vosLen)
428 {
429 // DecoderSpecific info descriptor
430 putDescr(pb, 0x05, track->vosLen);
431 put_buffer(pb, track->vosData, track->vosLen);
432 }
433
434 // SL descriptor
435 putDescr(pb, 0x06, 1);
436 put_byte(pb, 0x02);
437 return updateSize (pb, pos);
438 }
439
440 const CodecTag codec_movvideo_tags[] = {
441 { CODEC_ID_SVQ1, MKTAG('S', 'V', 'Q', '1') },
442 { CODEC_ID_SVQ3, MKTAG('S', 'V', 'Q', '3') },
443 { CODEC_ID_MPEG4, MKTAG('m', 'p', '4', 'v') },
444 { CODEC_ID_H263, MKTAG('s', '2', '6', '3') },
445 { CODEC_ID_DVVIDEO, MKTAG('d', 'v', 'c', ' ') },
446 { 0, 0 },
447 };
448
449 static int mov_write_video_tag(ByteIOContext *pb, MOVTrack* track)
450 {
451 int pos = url_ftell(pb);
452 int tag;
453
454 put_be32(pb, 0); /* size */
455
456 tag = codec_get_tag(codec_movvideo_tags, track->enc->codec_id);
457 // if no mac fcc found, try with Microsoft tags
458 if (!tag)
459 tag = codec_get_tag(codec_bmp_tags, track->enc->codec_id);
460 put_le32(pb, tag); // store it byteswapped
461
462 put_be32(pb, 0); /* Reserved */
463 put_be16(pb, 0); /* Reserved */
464 put_be16(pb, 1); /* Data-reference index */
465
466 put_be32(pb, 0); /* Reserved (= 02000c) */
467 put_be32(pb, 0); /* Reserved ("SVis")*/
468 put_be32(pb, 0); /* Reserved */
469 put_be32(pb, 0); /* Reserved (400)*/
470 put_be16(pb, track->enc->width); /* Video width */
471 put_be16(pb, track->enc->height); /* Video height */
472 put_be32(pb, 0x00480000); /* Reserved */
473 put_be32(pb, 0x00480000); /* Reserved */
474 put_be32(pb, 0); /* Data size (= 0) */
475 put_be16(pb, 1); /* Frame count (= 1) */
476
477 put_be32(pb, 0); /* Reserved */
478 put_be32(pb, 0); /* Reserved */
479 put_be32(pb, 0); /* Reserved */
480 put_be32(pb, 0); /* Reserved */
481 put_be32(pb, 0); /* Reserved */
482 put_be32(pb, 0); /* Reserved */
483 put_be32(pb, 0); /* Reserved */
484 put_be32(pb, 0); /* Reserved */
485
486 put_be16(pb, 0x18); /* Reserved */
487 put_be16(pb, 0xffff); /* Reserved */
488 if(track->enc->codec_id == CODEC_ID_MPEG4)
489 mov_write_esds_tag(pb, track);
490 else if(track->enc->codec_id == CODEC_ID_H263)
491 mov_write_d263_tag(pb);
492 else if(track->enc->codec_id == CODEC_ID_SVQ3)
493 mov_write_svq3_tag(pb);
494
495 return updateSize (pb, pos);
496 }
497
498 static int mov_write_stsd_tag(ByteIOContext *pb, MOVTrack* track)
499 {
500 int pos = url_ftell(pb);
501 put_be32(pb, 0); /* size */
502 put_tag(pb, "stsd");
503 put_be32(pb, 0); /* version & flags */
504 put_be32(pb, 1); /* entry count */
505 if (track->enc->codec_type == CODEC_TYPE_VIDEO)
506 mov_write_video_tag(pb, track);
507 else if (track->enc->codec_type == CODEC_TYPE_AUDIO)
508 mov_write_audio_tag(pb, track);
509 return updateSize(pb, pos);
510 }
511
512 /* TODO?: Currently all samples/frames seem to have same duration */
513 /* Time to sample atom */
514 static int mov_write_stts_tag(ByteIOContext *pb, MOVTrack* track)
515 {
516 put_be32(pb, 0x18); /* size */
517 put_tag(pb, "stts");
518 put_be32(pb, 0); /* version & flags */
519 put_be32(pb, 1); /* entry count */
520
521 put_be32(pb, track->sampleCount); /* sample count */
522 put_be32(pb, track->sampleDuration); /* sample duration */
523 return 0x18;
524 }
525
526 static int mov_write_dref_tag(ByteIOContext *pb)
527 {
528 put_be32(pb, 28); /* size */
529 put_tag(pb, "dref");
530 put_be32(pb, 0); /* version & flags */
531 put_be32(pb, 1); /* entry count */
532
533 put_be32(pb, 0xc); /* size */
534 put_tag(pb, "url ");
535 put_be32(pb, 1); /* version & flags */
536
537 return 28;
538 }
539
540 static int mov_write_stbl_tag(ByteIOContext *pb, MOVTrack* track)
541 {
542 int pos = url_ftell(pb);
543 put_be32(pb, 0); /* size */
544 put_tag(pb, "stbl");
545 mov_write_stsd_tag(pb, track);
546 mov_write_stts_tag(pb, track);
547 if (track->enc->codec_type == CODEC_TYPE_VIDEO &&
548 track->hasKeyframes)
549 mov_write_stss_tag(pb, track);
550 mov_write_stsc_tag(pb, track);
551 mov_write_stsz_tag(pb, track);
552 mov_write_stco_tag(pb, track);
553 return updateSize(pb, pos);
554 }
555
556 static int mov_write_dinf_tag(ByteIOContext *pb)
557 {
558 int pos = url_ftell(pb);
559 put_be32(pb, 0); /* size */
560 put_tag(pb, "dinf");
561 mov_write_dref_tag(pb);
562 return updateSize(pb, pos);
563 }
564
565 static int mov_write_smhd_tag(ByteIOContext *pb)
566 {
567 put_be32(pb, 16); /* size */
568 put_tag(pb, "smhd");
569 put_be32(pb, 0); /* version & flags */
570 put_be16(pb, 0); /* reserved (balance, normally = 0) */
571 put_be16(pb, 0); /* reserved */
572 return 16;
573 }
574
575 static int mov_write_vmhd_tag(ByteIOContext *pb)
576 {
577 put_be32(pb, 0x14); /* size (always 0x14) */
578 put_tag(pb, "vmhd");
579 put_be32(pb, 0x01); /* version & flags */
580 put_be64(pb, 0); /* reserved (graphics mode = copy) */
581 return 0x14;
582 }
583
584 static int mov_write_hdlr_tag(ByteIOContext *pb, MOVTrack* track)
585 {
586 char *descr, *hdlr, *hdlr_type;
587 int pos = url_ftell(pb);
588
589 if (!track) { /* no media --> data handler */
590 hdlr = "dhlr";
591 hdlr_type = "url ";
592 descr = "DataHandler";
593 } else {
594 hdlr = (track->mode == MODE_MOV) ? "mhlr" : "\0\0\0\0";
595 if (track->enc->codec_type == CODEC_TYPE_VIDEO) {
596 hdlr_type = "vide";
597 descr = "VideoHandler";
598 } else {
599 hdlr_type = "soun";
600 descr = "SoundHandler";
601 }
602 }
603
604 put_be32(pb, 0); /* size */
605 put_tag(pb, "hdlr");
606 put_be32(pb, 0); /* Version & flags */
607 put_buffer(pb, hdlr, 4); /* handler */
608 put_tag(pb, hdlr_type); /* handler type */
609 put_be32(pb ,0); /* reserved */
610 put_be32(pb ,0); /* reserved */
611 put_be32(pb ,0); /* reserved */
612 put_byte(pb, strlen(descr)); /* string counter */
613 put_buffer(pb, descr, strlen(descr)); /* handler description */
614 return updateSize(pb, pos);
615 }
616
617 static int mov_write_minf_tag(ByteIOContext *pb, MOVTrack* track)
618 {
619 int pos = url_ftell(pb);
620 put_be32(pb, 0); /* size */
621 put_tag(pb, "minf");
622 if(track->enc->codec_type == CODEC_TYPE_VIDEO)
623 mov_write_vmhd_tag(pb);
624 else
625 mov_write_smhd_tag(pb);
626 if (track->mode == MODE_MOV) /* FIXME: Why do it for MODE_MOV only ? */
627 mov_write_hdlr_tag(pb, NULL);
628 mov_write_dinf_tag(pb);
629 mov_write_stbl_tag(pb, track);
630 return updateSize(pb, pos);
631 }
632
633 static int mov_write_mdhd_tag(ByteIOContext *pb, MOVTrack* track)
634 {
635 put_be32(pb, 32); /* size */
636 put_tag(pb, "mdhd");
637 put_be32(pb, 0); /* Version & flags */
638 put_be32(pb, track->time); /* creation time */
639 put_be32(pb, track->time); /* modification time */
640 put_be32(pb, track->timescale); /* time scale (sample rate for audio) */
641 put_be32(pb, track->trackDuration); /* duration */
642 put_be16(pb, 0); /* language, 0 = english */
643 put_be16(pb, 0); /* reserved (quality) */
644 return 32;
645 }
646
647 static int mov_write_mdia_tag(ByteIOContext *pb, MOVTrack* track)
648 {
649 int pos = url_ftell(pb);
650 put_be32(pb, 0); /* size */
651 put_tag(pb, "mdia");
652 mov_write_mdhd_tag(pb, track);
653 mov_write_hdlr_tag(pb, track);
654 mov_write_minf_tag(pb, track);
655 return updateSize(pb, pos);
656 }
657
658 static int mov_write_tkhd_tag(ByteIOContext *pb, MOVTrack* track)
659 {
660 int64_t maxTrackLenTemp;
661 put_be32(pb, 0x5c); /* size (always 0x5c) */
662 put_tag(pb, "tkhd");
663 put_be32(pb, 0xf); /* version & flags (track enabled) */
664 put_be32(pb, track->time); /* creation time */
665 put_be32(pb, track->time); /* modification time */
666 put_be32(pb, track->trackID); /* track-id */
667 put_be32(pb, 0); /* reserved */
668 maxTrackLenTemp = ((int64_t)globalTimescale*(int64_t)track->trackDuration)/(int64_t)track->timescale;
669 put_be32(pb, (long)maxTrackLenTemp); /* duration */
670
671 put_be32(pb, 0); /* reserved */
672 put_be32(pb, 0); /* reserved */
673 put_be32(pb, 0x0); /* reserved (Layer & Alternate group) */
674 /* Volume, only for audio */
675 if(track->enc->codec_type == CODEC_TYPE_AUDIO)
676 put_be16(pb, 0x0100);
677 else
678 put_be16(pb, 0);
679 put_be16(pb, 0); /* reserved */
680
681 /* Matrix structure */
682 put_be32(pb, 0x00010000); /* reserved */
683 put_be32(pb, 0x0); /* reserved */
684 put_be32(pb, 0x0); /* reserved */
685 put_be32(pb, 0x0); /* reserved */
686 put_be32(pb, 0x00010000); /* reserved */
687 put_be32(pb, 0x0); /* reserved */
688 put_be32(pb, 0x0); /* reserved */
689 put_be32(pb, 0x0); /* reserved */
690 put_be32(pb, 0x40000000); /* reserved */
691
692 /* Track width and height, for visual only */
693 if(track->enc->codec_type == CODEC_TYPE_VIDEO) {
694 double sample_aspect_ratio = av_q2d(track->enc->sample_aspect_ratio);
695 if( !sample_aspect_ratio ) sample_aspect_ratio = 1;
696 put_be32(pb, sample_aspect_ratio * track->enc->width*0x10000);
697 put_be32(pb, track->enc->height*0x10000);
698 }
699 else {
700 put_be32(pb, 0);
701 put_be32(pb, 0);
702 }
703 return 0x5c;
704 }
705
706 static int mov_write_trak_tag(ByteIOContext *pb, MOVTrack* track)
707 {
708 int pos = url_ftell(pb);
709 put_be32(pb, 0); /* size */
710 put_tag(pb, "trak");
711 mov_write_tkhd_tag(pb, track);
712 mov_write_mdia_tag(pb, track);
713 return updateSize(pb, pos);
714 }
715
716 /* TODO: Not sorted out, but not necessary either */
717 static int mov_write_iods_tag(ByteIOContext *pb, MOVContext *mov)
718 {
719 put_be32(pb, 0x15); /* size */
720 put_tag(pb, "iods");
721 put_be32(pb, 0); /* version & flags */
722 put_be16(pb, 0x1007);
723 put_byte(pb, 0);
724 put_be16(pb, 0x4fff);
725 put_be16(pb, 0xfffe);
726 put_be16(pb, 0x01ff);
727 return 0x15;
728 }
729
730 static int mov_write_mvhd_tag(ByteIOContext *pb, MOVContext *mov)
731 {
732 int maxTrackID = 1, maxTrackLen = 0, i;
733 int64_t maxTrackLenTemp;
734
735 put_be32(pb, 0x6c); /* size (always 0x6c) */
736 put_tag(pb, "mvhd");
737 put_be32(pb, 0); /* version & flags */
738 put_be32(pb, mov->time); /* creation time */
739 put_be32(pb, mov->time); /* modification time */
740 put_be32(pb, mov->timescale); /* timescale */
741 for (i=0; i<MAX_STREAMS; i++) {
742 if(mov->tracks[i].entry > 0) {
743 maxTrackLenTemp = ((int64_t)globalTimescale*(int64_t)mov->tracks[i].trackDuration)/(int64_t)mov->tracks[i].timescale;
744 if(maxTrackLen < maxTrackLenTemp)
745 maxTrackLen = maxTrackLenTemp;
746 if(maxTrackID < mov->tracks[i].trackID)
747 maxTrackID = mov->tracks[i].trackID;
748 }
749 }
750 put_be32(pb, maxTrackLen); /* duration of longest track */
751
752 put_be32(pb, 0x00010000); /* reserved (preferred rate) 1.0 = normal */
753 put_be16(pb, 0x0100); /* reserved (preferred volume) 1.0 = normal */
754 put_be16(pb, 0); /* reserved */
755 put_be32(pb, 0); /* reserved */
756 put_be32(pb, 0); /* reserved */
757
758 /* Matrix structure */
759 put_be32(pb, 0x00010000); /* reserved */
760 put_be32(pb, 0x0); /* reserved */
761 put_be32(pb, 0x0); /* reserved */
762 put_be32(pb, 0x0); /* reserved */
763 put_be32(pb, 0x00010000); /* reserved */
764 put_be32(pb, 0x0); /* reserved */
765 put_be32(pb, 0x0); /* reserved */
766 put_be32(pb, 0x0); /* reserved */
767 put_be32(pb, 0x40000000); /* reserved */
768
769 put_be32(pb, 0); /* reserved (preview time) */
770 put_be32(pb, 0); /* reserved (preview duration) */
771 put_be32(pb, 0); /* reserved (poster time) */
772 put_be32(pb, 0); /* reserved (selection time) */
773 put_be32(pb, 0); /* reserved (selection duration) */
774 put_be32(pb, 0); /* reserved (current time) */
775 put_be32(pb, maxTrackID+1); /* Next track id */
776 return 0x6c;
777 }
778
779 static int mov_write_udta_tag(ByteIOContext *pb, MOVContext* mov,
780 AVFormatContext *s)
781 {
782 int pos = url_ftell(pb);
783 int i;
784
785 put_be32(pb, 0); /* size */
786 put_tag(pb, "udta");
787
788 /* Requirements */
789 for (i=0; i<MAX_STREAMS; i++) {
790 if(mov->tracks[i].entry <= 0) continue;
791 if (mov->tracks[i].enc->codec_id == CODEC_ID_AAC ||
792 mov->tracks[i].enc->codec_id == CODEC_ID_MPEG4) {
793 int pos = url_ftell(pb);
794 put_be32(pb, 0); /* size */
795 put_tag(pb, "\251req");
796 put_be16(pb, sizeof("QuickTime 6.0 or greater") - 1);
797 put_be16(pb, 0);
798 put_buffer(pb, "QuickTime 6.0 or greater",
799 sizeof("QuickTime 6.0 or greater") - 1);
800 updateSize(pb, pos);
801 break;
802 }
803 }
804
805 /* Encoder */
806 if(!(mov->tracks[0].enc->flags & CODEC_FLAG_BITEXACT))
807 {
808 int pos = url_ftell(pb);
809 put_be32(pb, 0); /* size */
810 put_tag(pb, "\251enc");
811 put_be16(pb, sizeof(LIBAVFORMAT_IDENT) - 1); /* string length */
812 put_be16(pb, 0);
813 put_buffer(pb, LIBAVFORMAT_IDENT, sizeof(LIBAVFORMAT_IDENT) - 1);
814 updateSize(pb, pos);
815 }
816
817 if( s->title[0] )
818 {
819 int pos = url_ftell(pb);
820 put_be32(pb, 0); /* size */
821 put_tag(pb, "\251nam");
822 put_be16(pb, strlen(s->title)); /* string length */
823 put_be16(pb, 0);
824 put_buffer(pb, s->title, strlen(s->title));
825 updateSize(pb, pos);
826 }
827
828 if( s->author[0] )
829 {
830 int pos = url_ftell(pb);
831 put_be32(pb, 0); /* size */
832 put_tag(pb, /*"\251aut"*/ "\251day" );
833 put_be16(pb, strlen(s->author)); /* string length */
834 put_be16(pb, 0);
835 put_buffer(pb, s->author, strlen(s->author));
836 updateSize(pb, pos);
837 }
838
839 if( s->comment[0] )
840 {
841 int pos = url_ftell(pb);
842 put_be32(pb, 0); /* size */
843 put_tag(pb, "\251des");
844 put_be16(pb, strlen(s->comment)); /* string length */
845 put_be16(pb, 0);
846 put_buffer(pb, s->comment, strlen(s->comment));
847 updateSize(pb, pos);
848 }
849
850 return updateSize(pb, pos);
851 }
852
853 static int mov_write_moov_tag(ByteIOContext *pb, MOVContext *mov,
854 AVFormatContext *s)
855 {
856 int pos, i;
857 pos = url_ftell(pb);
858 put_be32(pb, 0); /* size placeholder*/
859 put_tag(pb, "moov");
860 mov->timescale = globalTimescale;
861
862 for (i=0; i<MAX_STREAMS; i++) {
863 if(mov->tracks[i].entry <= 0) continue;
864
865 if(mov->tracks[i].enc->codec_type == CODEC_TYPE_VIDEO) {
866 mov->tracks[i].timescale = mov->tracks[i].enc->frame_rate;
867 mov->tracks[i].sampleDuration = mov->tracks[i].enc->frame_rate_base;
868 }
869 else if(mov->tracks[i].enc->codec_type == CODEC_TYPE_AUDIO) {
870 /* If AMR, track timescale = 8000, AMR_WB = 16000 */
871 if(mov->tracks[i].enc->codec_id == CODEC_ID_AMR_NB) {
872 mov->tracks[i].sampleDuration = 160; // Bytes per chunk
873 mov->tracks[i].timescale = 8000;
874 }
875 else {
876 mov->tracks[i].timescale = mov->tracks[i].enc->sample_rate;
877 mov->tracks[i].sampleDuration = mov->tracks[i].enc->frame_size;
878 }
879 }
880
881 mov->tracks[i].trackDuration =
882 mov->tracks[i].sampleCount * mov->tracks[i].sampleDuration;
883 mov->tracks[i].time = mov->time;
884 mov->tracks[i].trackID = i+1;
885 }
886
887 mov_write_mvhd_tag(pb, mov);
888 //mov_write_iods_tag(pb, mov);
889 for (i=0; i<MAX_STREAMS; i++) {
890 if(mov->tracks[i].entry > 0) {
891 mov_write_trak_tag(pb, &(mov->tracks[i]));
892 }
893 }
894
895 mov_write_udta_tag(pb, mov, s);
896
897 return updateSize(pb, pos);
898 }
899
900 int mov_write_mdat_tag(ByteIOContext *pb, MOVContext* mov)
901 {
902 mov->mdat_pos = url_ftell(pb);
903 put_be32(pb, 0); /* size placeholder*/
904 put_tag(pb, "mdat");
905 return 0;
906 }
907
908 /* TODO: This needs to be more general */
909 int mov_write_ftyp_tag(ByteIOContext *pb, AVFormatContext *s)
910 {
911 MOVContext *mov = s->priv_data;
912
913 put_be32(pb, 0x14 ); /* size */
914 put_tag(pb, "ftyp");
915
916 if ( mov->mode == MODE_3GP )
917 put_tag(pb, "3gp4");
918 else
919 put_tag(pb, "isom");
920
921 put_be32(pb, 0x200 );
922
923 if ( mov->mode == MODE_3GP )
924 put_tag(pb, "3gp4");
925 else
926 put_tag(pb, "mp41");
927
928 return 0x14;
929 }
930
931 static int mov_write_header(AVFormatContext *s)
932 {
933 ByteIOContext *pb = &s->pb;
934 MOVContext *mov = s->priv_data;
935 int i;
936
937 for(i=0; i<s->nb_streams; i++){
938 AVCodecContext *c= &s->streams[i]->codec;
939
940 if (c->codec_type == CODEC_TYPE_VIDEO){
941 if (!codec_get_tag(codec_movvideo_tags, c->codec_id)){
942 if(!codec_get_tag(codec_bmp_tags, c->codec_id))
943 return -1;
944 else
945 av_log(s, AV_LOG_INFO, "Warning, using MS style video codec tag, the file may be unplayable!\n");
946 }
947 }else if(c->codec_type == CODEC_TYPE_AUDIO){
948 if (!codec_get_tag(codec_movaudio_tags, c->codec_id)){
949 if(!codec_get_tag(codec_wav_tags, c->codec_id))
950 return -1;
951 else
952 av_log(s, AV_LOG_INFO, "Warning, using MS style audio codec tag, the file may be unplayable!\n");
953 }
954 }
955 }
956
957 /* Default mode == MP4 */
958 mov->mode = MODE_MP4;
959
960 if (s->oformat != NULL) {
961 if (!strcmp("3gp", s->oformat->name)) mov->mode = MODE_3GP;
962 else if (!strcmp("mov", s->oformat->name)) mov->mode = MODE_MOV;
963
964 if ( mov->mode == MODE_3GP || mov->mode == MODE_MP4 )
965 mov_write_ftyp_tag(pb,s);
966 }
967
968 for (i=0; i<MAX_STREAMS; i++) {
969 mov->tracks[i].mode = mov->mode;
970 }
971
972 put_flush_packet(pb);
973
974 return 0;
975 }
976
977 static int mov_write_packet(AVFormatContext *s, AVPacket *pkt)
978 {
979 MOVContext *mov = s->priv_data;
980 ByteIOContext *pb = &s->pb;
981 AVCodecContext *enc = &s->streams[pkt->stream_index]->codec;
982 MOVTrack* trk = &mov->tracks[pkt->stream_index];
983 int cl, id;
984 unsigned int samplesInChunk = 0;
985 int size= pkt->size;
986
987 if (url_is_streamed(&s->pb)) return 0; /* Can't handle that */
988 if (!size) return 0; /* Discard 0 sized packets */
989
990 if (enc->codec_type == CODEC_TYPE_VIDEO ) {
991 samplesInChunk = 1;
992 }
993 else if (enc->codec_type == CODEC_TYPE_AUDIO ) {
994 if( enc->codec_id == CODEC_ID_AMR_NB) {
995 /* We must find out how many AMR blocks there are in one packet */
996 static uint16_t packed_size[16] =
997 {13, 14, 16, 18, 20, 21, 27, 32, 6, 0, 0, 0, 0, 0, 0, 0};
998 int len = 0;
999
1000 while (len < size && samplesInChunk < 100) {
1001 len += packed_size[(pkt->data[len] >> 3) & 0x0F];
1002 samplesInChunk++;
1003 }
1004 }
1005 else if(enc->codec_id == CODEC_ID_PCM_ALAW) {
1006 samplesInChunk = size/enc->channels;
1007 }
1008 else if(enc->codec_id == CODEC_ID_PCM_S16BE || enc->codec_id == CODEC_ID_PCM_S16LE) {
1009 samplesInChunk = size/(2*enc->channels);
1010 }
1011 else {
1012 samplesInChunk = 1;
1013 }
1014 }
1015
1016 if ((enc->codec_id == CODEC_ID_MPEG4 || enc->codec_id == CODEC_ID_AAC)
1017 && trk->vosLen == 0) {
1018 // assert(enc->extradata_size);
1019
1020 trk->vosLen = enc->extradata_size;
1021 trk->vosData = av_malloc(trk->vosLen);
1022 memcpy(trk->vosData, enc->extradata, trk->vosLen);
1023 }
1024
1025 cl = trk->entry / MOV_INDEX_CLUSTER_SIZE;
1026 id = trk->entry % MOV_INDEX_CLUSTER_SIZE;
1027
1028 if (trk->ents_allocated <= trk->entry) {
1029 trk->cluster = av_realloc(trk->cluster, (cl+1)*sizeof(void*));
1030 if (!trk->cluster)
1031 return -1;
1032 trk->cluster[cl] = av_malloc(MOV_INDEX_CLUSTER_SIZE*sizeof(MOVIentry));
1033 if (!trk->cluster[cl])
1034 return -1;
1035 trk->ents_allocated += MOV_INDEX_CLUSTER_SIZE;
1036 }
1037 if (mov->mdat_written == 0) {
1038 mov_write_mdat_tag(pb, mov);
1039 mov->mdat_written = 1;
1040 mov->time = s->timestamp;
1041 }
1042
1043 trk->cluster[cl][id].pos = url_ftell(pb);
1044 trk->cluster[cl][id].samplesInChunk = samplesInChunk;
1045 trk->cluster[cl][id].size = size;
1046 trk->cluster[cl][id].entries = samplesInChunk;
1047 if(enc->codec_type == CODEC_TYPE_VIDEO) {
1048 trk->cluster[cl][id].key_frame = !!(pkt->flags & PKT_FLAG_KEY);
1049 if(trk->cluster[cl][id].key_frame)
1050 trk->hasKeyframes = 1;
1051 }
1052 trk->enc = enc;
1053 trk->entry++;
1054 trk->sampleCount += samplesInChunk;
1055 trk->mdat_size += size;
1056
1057 put_buffer(pb, pkt->data, size);
1058
1059 put_flush_packet(pb);
1060 return 0;
1061 }
1062
1063 static int mov_write_trailer(AVFormatContext *s)
1064 {
1065 MOVContext *mov = s->priv_data;
1066 ByteIOContext *pb = &s->pb;
1067 int res = 0;
1068 int i, j;
1069
1070 offset_t moov_pos = url_ftell(pb);
1071
1072 /* Write size of mdat tag */
1073 for (i=0, j=0; i<MAX_STREAMS; i++) {
1074 if(mov->tracks[i].ents_allocated > 0) {
1075 j += mov->tracks[i].mdat_size;
1076 }
1077 }
1078 url_fseek(pb, mov->mdat_pos, SEEK_SET);
1079 put_be32(pb, j+8);
1080 url_fseek(pb, moov_pos, SEEK_SET);
1081
1082 mov_write_moov_tag(pb, mov, s);
1083
1084 for (i=0; i<MAX_STREAMS; i++) {
1085 for (j=0; j<mov->tracks[i].ents_allocated/MOV_INDEX_CLUSTER_SIZE; j++) {
1086 av_free(mov->tracks[i].cluster[j]);
1087 }
1088 av_free(mov->tracks[i].cluster);
1089 if( mov->tracks[i].vosLen ) av_free( mov->tracks[i].vosData );
1090
1091 mov->tracks[i].cluster = NULL;
1092 mov->tracks[i].ents_allocated = mov->tracks[i].entry = 0;
1093 }
1094
1095 put_flush_packet(pb);
1096
1097 return res;
1098 }
1099
1100 static AVOutputFormat mov_oformat = {
1101 "mov",
1102 "mov format",
1103 NULL,
1104 "mov",
1105 sizeof(MOVContext),
1106 CODEC_ID_AAC,
1107 CODEC_ID_MPEG4,
1108 mov_write_header,
1109 mov_write_packet,
1110 mov_write_trailer,
1111 };
1112
1113 static AVOutputFormat _3gp_oformat = {
1114 "3gp",
1115 "3gp format",
1116 NULL,
1117 "3gp",
1118 sizeof(MOVContext),
1119 CODEC_ID_AMR_NB,
1120 CODEC_ID_H263,
1121 mov_write_header,
1122 mov_write_packet,
1123 mov_write_trailer,
1124 };
1125
1126 static AVOutputFormat mp4_oformat = {
1127 "mp4",
1128 "mp4 format",
1129 "application/mp4",
1130 "mp4,m4a",
1131 sizeof(MOVContext),
1132 CODEC_ID_AAC,
1133 CODEC_ID_MPEG4,
1134 mov_write_header,
1135 mov_write_packet,
1136 mov_write_trailer,
1137 };
1138
1139 int movenc_init(void)
1140 {
1141 av_register_output_format(&mov_oformat);
1142 av_register_output_format(&_3gp_oformat);
1143 av_register_output_format(&mp4_oformat);
1144 return 0;
1145 }