1 /*
2 * WAV demuxer
3 * Copyright (c) 2001, 2002 Fabrice Bellard
4 *
5 * Sony Wave64 demuxer
6 * RF64 demuxer
7 * Copyright (c) 2009 Daniel Verkamp
8 *
9 * This file is part of FFmpeg.
10 *
11 * FFmpeg is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public
13 * License as published by the Free Software Foundation; either
14 * version 2.1 of the License, or (at your option) any later version.
15 *
16 * FFmpeg is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
20 *
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with FFmpeg; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 */
25
26 #include <stdint.h>
27
28 #include "libavutil/avassert.h"
29 #include "libavutil/dict.h"
30 #include "libavutil/intreadwrite.h"
31 #include "libavutil/log.h"
32 #include "libavutil/mathematics.h"
33 #include "libavutil/opt.h"
34 #include "avformat.h"
35 #include "avio.h"
36 #include "avio_internal.h"
37 #include "id3v2.h"
38 #include "internal.h"
39 #include "metadata.h"
40 #include "pcm.h"
41 #include "riff.h"
42 #include "w64.h"
43 #include "spdif.h"
44
45 typedef struct WAVDemuxContext {
46 const AVClass *class;
47 int64_t data_end;
48 int w64;
49 int64_t smv_data_ofs;
50 int smv_block_size;
51 int smv_frames_per_jpeg;
52 int smv_block;
53 int smv_last_stream;
54 int smv_eof;
55 int audio_eof;
56 int ignore_length;
57 int spdif;
58 int smv_cur_pt;
59 int smv_given_first;
60 int unaligned; // e.g. if an odd number of bytes ID3 tag was prepended
61 int rifx; // RIFX: integer byte order for parameters is big endian
62 } WAVDemuxContext;
63
set_spdif(AVFormatContext * s,WAVDemuxContext * wav)64 static void set_spdif(AVFormatContext *s, WAVDemuxContext *wav)
65 {
66 if (CONFIG_SPDIF_DEMUXER && s->streams[0]->codecpar->codec_tag == 1) {
67 enum AVCodecID codec;
68 int len = 1<<16;
69 int ret = ffio_ensure_seekback(s->pb, len);
70
71 if (ret >= 0) {
72 uint8_t *buf = av_malloc(len + AV_INPUT_BUFFER_PADDING_SIZE);
73 if (!buf) {
74 ret = AVERROR(ENOMEM);
75 } else {
76 int64_t pos = avio_tell(s->pb);
77 len = ret = avio_read(s->pb, buf, len);
78 if (len >= 0) {
79 ret = ff_spdif_probe(buf, len, &codec);
80 if (ret > AVPROBE_SCORE_EXTENSION) {
81 s->streams[0]->codecpar->codec_id = codec;
82 wav->spdif = 1;
83 }
84 }
85 avio_seek(s->pb, pos, SEEK_SET);
86 av_free(buf);
87 }
88 }
89
90 if (ret < 0)
91 av_log(s, AV_LOG_WARNING, "Cannot check for SPDIF\n");
92 }
93 }
94
95 #if CONFIG_WAV_DEMUXER
96
next_tag(AVIOContext * pb,uint32_t * tag,int big_endian)97 static int64_t next_tag(AVIOContext *pb, uint32_t *tag, int big_endian)
98 {
99 *tag = avio_rl32(pb);
100 if (!big_endian) {
101 return avio_rl32(pb);
102 } else {
103 return avio_rb32(pb);
104 }
105 }
106
107 /* RIFF chunks are always at even offsets relative to where they start. */
wav_seek_tag(WAVDemuxContext * wav,AVIOContext * s,int64_t offset,int whence)108 static int64_t wav_seek_tag(WAVDemuxContext * wav, AVIOContext *s, int64_t offset, int whence)
109 {
110 offset += offset < INT64_MAX && offset + wav->unaligned & 1;
111
112 return avio_seek(s, offset, whence);
113 }
114
115 /* return the size of the found tag */
find_tag(WAVDemuxContext * wav,AVIOContext * pb,uint32_t tag1)116 static int64_t find_tag(WAVDemuxContext * wav, AVIOContext *pb, uint32_t tag1)
117 {
118 unsigned int tag;
119 int64_t size;
120
121 for (;;) {
122 if (avio_feof(pb))
123 return AVERROR_EOF;
124 size = next_tag(pb, &tag, wav->rifx);
125 if (tag == tag1)
126 break;
127 wav_seek_tag(wav, pb, size, SEEK_CUR);
128 }
129 return size;
130 }
131
wav_probe(const AVProbeData * p)132 static int wav_probe(const AVProbeData *p)
133 {
134 /* check file header */
135 if (p->buf_size <= 32)
136 return 0;
137 if (!memcmp(p->buf + 8, "WAVE", 4)) {
138 if (!memcmp(p->buf, "RIFF", 4) || !memcmp(p->buf, "RIFX", 4))
139 /* Since the ACT demuxer has a standard WAV header at the top of
140 * its own, the returned score is decreased to avoid a probe
141 * conflict between ACT and WAV. */
142 return AVPROBE_SCORE_MAX - 1;
143 else if (!memcmp(p->buf, "RF64", 4) &&
144 !memcmp(p->buf + 12, "ds64", 4))
145 return AVPROBE_SCORE_MAX;
146 }
147 return 0;
148 }
149
handle_stream_probing(AVStream * st)150 static void handle_stream_probing(AVStream *st)
151 {
152 if (st->codecpar->codec_id == AV_CODEC_ID_PCM_S16LE) {
153 st->request_probe = AVPROBE_SCORE_EXTENSION;
154 st->probe_packets = FFMIN(st->probe_packets, 32);
155 }
156 }
157
wav_parse_fmt_tag(AVFormatContext * s,int64_t size,AVStream ** st)158 static int wav_parse_fmt_tag(AVFormatContext *s, int64_t size, AVStream **st)
159 {
160 AVIOContext *pb = s->pb;
161 WAVDemuxContext *wav = s->priv_data;
162 int ret;
163
164 /* parse fmt header */
165 *st = avformat_new_stream(s, NULL);
166 if (!*st)
167 return AVERROR(ENOMEM);
168
169 ret = ff_get_wav_header(s, pb, (*st)->codecpar, size, wav->rifx);
170 if (ret < 0)
171 return ret;
172 handle_stream_probing(*st);
173
174 (*st)->need_parsing = AVSTREAM_PARSE_FULL_RAW;
175
176 avpriv_set_pts_info(*st, 64, 1, (*st)->codecpar->sample_rate);
177
178 return 0;
179 }
180
wav_parse_xma2_tag(AVFormatContext * s,int64_t size,AVStream ** st)181 static int wav_parse_xma2_tag(AVFormatContext *s, int64_t size, AVStream **st)
182 {
183 AVIOContext *pb = s->pb;
184 int version, num_streams, i, channels = 0, ret;
185
186 if (size < 36)
187 return AVERROR_INVALIDDATA;
188
189 *st = avformat_new_stream(s, NULL);
190 if (!*st)
191 return AVERROR(ENOMEM);
192
193 (*st)->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
194 (*st)->codecpar->codec_id = AV_CODEC_ID_XMA2;
195 (*st)->need_parsing = AVSTREAM_PARSE_FULL_RAW;
196
197 version = avio_r8(pb);
198 if (version != 3 && version != 4)
199 return AVERROR_INVALIDDATA;
200 num_streams = avio_r8(pb);
201 if (size != (32 + ((version==3)?0:8) + 4*num_streams))
202 return AVERROR_INVALIDDATA;
203 avio_skip(pb, 10);
204 (*st)->codecpar->sample_rate = avio_rb32(pb);
205 if (version == 4)
206 avio_skip(pb, 8);
207 avio_skip(pb, 4);
208 (*st)->duration = avio_rb32(pb);
209 avio_skip(pb, 8);
210
211 for (i = 0; i < num_streams; i++) {
212 channels += avio_r8(pb);
213 avio_skip(pb, 3);
214 }
215 (*st)->codecpar->channels = channels;
216
217 if ((*st)->codecpar->channels <= 0 || (*st)->codecpar->sample_rate <= 0)
218 return AVERROR_INVALIDDATA;
219
220 avpriv_set_pts_info(*st, 64, 1, (*st)->codecpar->sample_rate);
221
222 avio_seek(pb, -size, SEEK_CUR);
223 if ((ret = ff_get_extradata(s, (*st)->codecpar, pb, size)) < 0)
224 return ret;
225
226 return 0;
227 }
228
wav_parse_bext_string(AVFormatContext * s,const char * key,int length)229 static inline int wav_parse_bext_string(AVFormatContext *s, const char *key,
230 int length)
231 {
232 char temp[257];
233 int ret;
234
235 av_assert0(length < sizeof(temp));
236 if ((ret = avio_read(s->pb, temp, length)) != length)
237 return ret < 0 ? ret : AVERROR_INVALIDDATA;
238
239 temp[length] = 0;
240
241 if (strlen(temp))
242 return av_dict_set(&s->metadata, key, temp, 0);
243
244 return 0;
245 }
246
wav_parse_bext_tag(AVFormatContext * s,int64_t size)247 static int wav_parse_bext_tag(AVFormatContext *s, int64_t size)
248 {
249 char temp[131], *coding_history;
250 int ret, x;
251 uint64_t time_reference;
252 int64_t umid_parts[8], umid_mask = 0;
253
254 if ((ret = wav_parse_bext_string(s, "description", 256)) < 0 ||
255 (ret = wav_parse_bext_string(s, "originator", 32)) < 0 ||
256 (ret = wav_parse_bext_string(s, "originator_reference", 32)) < 0 ||
257 (ret = wav_parse_bext_string(s, "origination_date", 10)) < 0 ||
258 (ret = wav_parse_bext_string(s, "origination_time", 8)) < 0)
259 return ret;
260
261 time_reference = avio_rl64(s->pb);
262 snprintf(temp, sizeof(temp), "%"PRIu64, time_reference);
263 if ((ret = av_dict_set(&s->metadata, "time_reference", temp, 0)) < 0)
264 return ret;
265
266 /* check if version is >= 1, in which case an UMID may be present */
267 if (avio_rl16(s->pb) >= 1) {
268 for (x = 0; x < 8; x++)
269 umid_mask |= umid_parts[x] = avio_rb64(s->pb);
270
271 if (umid_mask) {
272 /* the string formatting below is per SMPTE 330M-2004 Annex C */
273 if (umid_parts[4] == 0 && umid_parts[5] == 0 &&
274 umid_parts[6] == 0 && umid_parts[7] == 0) {
275 /* basic UMID */
276 snprintf(temp, sizeof(temp),
277 "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
278 umid_parts[0], umid_parts[1],
279 umid_parts[2], umid_parts[3]);
280 } else {
281 /* extended UMID */
282 snprintf(temp, sizeof(temp),
283 "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64
284 "%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
285 umid_parts[0], umid_parts[1],
286 umid_parts[2], umid_parts[3],
287 umid_parts[4], umid_parts[5],
288 umid_parts[6], umid_parts[7]);
289 }
290
291 if ((ret = av_dict_set(&s->metadata, "umid", temp, 0)) < 0)
292 return ret;
293 }
294
295 avio_skip(s->pb, 190);
296 } else
297 avio_skip(s->pb, 254);
298
299 if (size > 602) {
300 /* CodingHistory present */
301 size -= 602;
302
303 if (!(coding_history = av_malloc(size + 1)))
304 return AVERROR(ENOMEM);
305
306 if ((ret = avio_read(s->pb, coding_history, size)) != size) {
307 av_free(coding_history);
308 return ret < 0 ? ret : AVERROR_INVALIDDATA;
309 }
310
311 coding_history[size] = 0;
312 if ((ret = av_dict_set(&s->metadata, "coding_history", coding_history,
313 AV_DICT_DONT_STRDUP_VAL)) < 0)
314 return ret;
315 }
316
317 return 0;
318 }
319
320 static const AVMetadataConv wav_metadata_conv[] = {
321 { "description", "comment" },
322 { "originator", "encoded_by" },
323 { "origination_date", "date" },
324 { "origination_time", "creation_time" },
325 { 0 },
326 };
327
328 /* wav input */
wav_read_header(AVFormatContext * s)329 static int wav_read_header(AVFormatContext *s)
330 {
331 int64_t size, av_uninit(data_size);
332 int64_t sample_count = 0;
333 int rf64 = 0;
334 uint32_t tag;
335 AVIOContext *pb = s->pb;
336 AVStream *st = NULL;
337 WAVDemuxContext *wav = s->priv_data;
338 int ret, got_fmt = 0, got_xma2 = 0;
339 int64_t next_tag_ofs, data_ofs = -1;
340
341 wav->unaligned = avio_tell(s->pb) & 1;
342
343 wav->smv_data_ofs = -1;
344
345 /* read chunk ID */
346 tag = avio_rl32(pb);
347 switch (tag) {
348 case MKTAG('R', 'I', 'F', 'F'):
349 break;
350 case MKTAG('R', 'I', 'F', 'X'):
351 wav->rifx = 1;
352 break;
353 case MKTAG('R', 'F', '6', '4'):
354 rf64 = 1;
355 break;
356 default:
357 av_log(s, AV_LOG_ERROR, "invalid start code %s in RIFF header\n",
358 av_fourcc2str(tag));
359 return AVERROR_INVALIDDATA;
360 }
361
362 /* read chunk size */
363 avio_rl32(pb);
364
365 /* read format */
366 if (avio_rl32(pb) != MKTAG('W', 'A', 'V', 'E')) {
367 av_log(s, AV_LOG_ERROR, "invalid format in RIFF header\n");
368 return AVERROR_INVALIDDATA;
369 }
370
371 if (rf64) {
372 if (avio_rl32(pb) != MKTAG('d', 's', '6', '4'))
373 return AVERROR_INVALIDDATA;
374 size = avio_rl32(pb);
375 if (size < 24)
376 return AVERROR_INVALIDDATA;
377 avio_rl64(pb); /* RIFF size */
378
379 data_size = avio_rl64(pb);
380 sample_count = avio_rl64(pb);
381
382 if (data_size < 0 || sample_count < 0) {
383 av_log(s, AV_LOG_ERROR, "negative data_size and/or sample_count in "
384 "ds64: data_size = %"PRId64", sample_count = %"PRId64"\n",
385 data_size, sample_count);
386 return AVERROR_INVALIDDATA;
387 }
388 avio_skip(pb, size - 24); /* skip rest of ds64 chunk */
389
390 }
391
392 for (;;) {
393 AVStream *vst;
394 size = next_tag(pb, &tag, wav->rifx);
395 next_tag_ofs = avio_tell(pb) + size;
396
397 if (avio_feof(pb))
398 break;
399
400 switch (tag) {
401 case MKTAG('f', 'm', 't', ' '):
402 /* only parse the first 'fmt ' tag found */
403 if (!got_xma2 && !got_fmt && (ret = wav_parse_fmt_tag(s, size, &st)) < 0) {
404 return ret;
405 } else if (got_fmt)
406 av_log(s, AV_LOG_WARNING, "found more than one 'fmt ' tag\n");
407
408 got_fmt = 1;
409 break;
410 case MKTAG('X', 'M', 'A', '2'):
411 /* only parse the first 'XMA2' tag found */
412 if (!got_fmt && !got_xma2 && (ret = wav_parse_xma2_tag(s, size, &st)) < 0) {
413 return ret;
414 } else if (got_xma2)
415 av_log(s, AV_LOG_WARNING, "found more than one 'XMA2' tag\n");
416
417 got_xma2 = 1;
418 break;
419 case MKTAG('d', 'a', 't', 'a'):
420 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) && !got_fmt && !got_xma2) {
421 av_log(s, AV_LOG_ERROR,
422 "found no 'fmt ' tag before the 'data' tag\n");
423 return AVERROR_INVALIDDATA;
424 }
425
426 if (rf64) {
427 next_tag_ofs = wav->data_end = avio_tell(pb) + data_size;
428 } else if (size != 0xFFFFFFFF) {
429 data_size = size;
430 next_tag_ofs = wav->data_end = size ? next_tag_ofs : INT64_MAX;
431 } else {
432 av_log(s, AV_LOG_WARNING, "Ignoring maximum wav data size, "
433 "file may be invalid\n");
434 data_size = 0;
435 next_tag_ofs = wav->data_end = INT64_MAX;
436 }
437
438 data_ofs = avio_tell(pb);
439
440 /* don't look for footer metadata if we can't seek or if we don't
441 * know where the data tag ends
442 */
443 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) || (!rf64 && !size))
444 goto break_loop;
445 break;
446 case MKTAG('f', 'a', 'c', 't'):
447 if (!sample_count)
448 sample_count = (!wav->rifx ? avio_rl32(pb) : avio_rb32(pb));
449 break;
450 case MKTAG('b', 'e', 'x', 't'):
451 if ((ret = wav_parse_bext_tag(s, size)) < 0)
452 return ret;
453 break;
454 case MKTAG('S','M','V','0'):
455 if (!got_fmt) {
456 av_log(s, AV_LOG_ERROR, "found no 'fmt ' tag before the 'SMV0' tag\n");
457 return AVERROR_INVALIDDATA;
458 }
459 // SMV file, a wav file with video appended.
460 if (size != MKTAG('0','2','0','0')) {
461 av_log(s, AV_LOG_ERROR, "Unknown SMV version found\n");
462 goto break_loop;
463 }
464 av_log(s, AV_LOG_DEBUG, "Found SMV data\n");
465 wav->smv_given_first = 0;
466 vst = avformat_new_stream(s, NULL);
467 if (!vst)
468 return AVERROR(ENOMEM);
469 avio_r8(pb);
470 vst->id = 1;
471 vst->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
472 vst->codecpar->codec_id = AV_CODEC_ID_SMVJPEG;
473 vst->codecpar->width = avio_rl24(pb);
474 vst->codecpar->height = avio_rl24(pb);
475 if ((ret = ff_alloc_extradata(vst->codecpar, 4)) < 0) {
476 av_log(s, AV_LOG_ERROR, "Could not allocate extradata.\n");
477 return ret;
478 }
479 size = avio_rl24(pb);
480 wav->smv_data_ofs = avio_tell(pb) + (size - 5) * 3;
481 avio_rl24(pb);
482 wav->smv_block_size = avio_rl24(pb);
483 avpriv_set_pts_info(vst, 32, 1, avio_rl24(pb));
484 vst->duration = avio_rl24(pb);
485 avio_rl24(pb);
486 avio_rl24(pb);
487 wav->smv_frames_per_jpeg = avio_rl24(pb);
488 if (wav->smv_frames_per_jpeg > 65536) {
489 av_log(s, AV_LOG_ERROR, "too many frames per jpeg\n");
490 return AVERROR_INVALIDDATA;
491 }
492 AV_WL32(vst->codecpar->extradata, wav->smv_frames_per_jpeg);
493 wav->smv_cur_pt = 0;
494 goto break_loop;
495 case MKTAG('L', 'I', 'S', 'T'):
496 if (size < 4) {
497 av_log(s, AV_LOG_ERROR, "too short LIST tag\n");
498 return AVERROR_INVALIDDATA;
499 }
500 switch (avio_rl32(pb)) {
501 case MKTAG('I', 'N', 'F', 'O'):
502 ff_read_riff_info(s, size - 4);
503 }
504 break;
505 case MKTAG('I', 'D', '3', ' '):
506 case MKTAG('i', 'd', '3', ' '): {
507 ID3v2ExtraMeta *id3v2_extra_meta = NULL;
508 ff_id3v2_read_dict(pb, &s->internal->id3v2_meta, ID3v2_DEFAULT_MAGIC, &id3v2_extra_meta);
509 if (id3v2_extra_meta) {
510 ff_id3v2_parse_apic(s, id3v2_extra_meta);
511 ff_id3v2_parse_chapters(s, id3v2_extra_meta);
512 ff_id3v2_parse_priv(s, id3v2_extra_meta);
513 }
514 ff_id3v2_free_extra_meta(&id3v2_extra_meta);
515 }
516 break;
517 }
518
519 /* seek to next tag unless we know that we'll run into EOF */
520 if ((avio_size(pb) > 0 && next_tag_ofs >= avio_size(pb)) ||
521 wav_seek_tag(wav, pb, next_tag_ofs, SEEK_SET) < 0) {
522 break;
523 }
524 }
525
526 break_loop:
527 if (!got_fmt && !got_xma2) {
528 av_log(s, AV_LOG_ERROR, "no 'fmt ' or 'XMA2' tag found\n");
529 return AVERROR_INVALIDDATA;
530 }
531
532 if (data_ofs < 0) {
533 av_log(s, AV_LOG_ERROR, "no 'data' tag found\n");
534 return AVERROR_INVALIDDATA;
535 }
536
537 avio_seek(pb, data_ofs, SEEK_SET);
538
539 if (data_size > (INT64_MAX>>3)) {
540 av_log(s, AV_LOG_WARNING, "Data size %"PRId64" is too large\n", data_size);
541 data_size = 0;
542 }
543
544 if ( st->codecpar->bit_rate > 0 && data_size > 0
545 && st->codecpar->sample_rate > 0
546 && sample_count > 0 && st->codecpar->channels > 1
547 && sample_count % st->codecpar->channels == 0) {
548 if (fabs(8.0 * data_size * st->codecpar->channels * st->codecpar->sample_rate /
549 sample_count /st->codecpar->bit_rate - 1.0) < 0.3)
550 sample_count /= st->codecpar->channels;
551 }
552
553 if ( data_size > 0 && sample_count && st->codecpar->channels
554 && (data_size << 3) / sample_count / st->codecpar->channels > st->codecpar->bits_per_coded_sample + 1) {
555 av_log(s, AV_LOG_WARNING, "ignoring wrong sample_count %"PRId64"\n", sample_count);
556 sample_count = 0;
557 }
558
559 /* G.729 hack (for Ticket4577)
560 * FIXME: Come up with cleaner, more general solution */
561 if (st->codecpar->codec_id == AV_CODEC_ID_G729 && sample_count && (data_size << 3) > sample_count) {
562 av_log(s, AV_LOG_WARNING, "ignoring wrong sample_count %"PRId64"\n", sample_count);
563 sample_count = 0;
564 }
565
566 if (!sample_count || av_get_exact_bits_per_sample(st->codecpar->codec_id) > 0)
567 if ( st->codecpar->channels
568 && data_size
569 && av_get_bits_per_sample(st->codecpar->codec_id)
570 && wav->data_end <= avio_size(pb))
571 sample_count = (data_size << 3)
572 /
573 (st->codecpar->channels * (uint64_t)av_get_bits_per_sample(st->codecpar->codec_id));
574
575 if (sample_count)
576 st->duration = sample_count;
577
578 if (st->codecpar->codec_id == AV_CODEC_ID_PCM_S32LE &&
579 st->codecpar->block_align == st->codecpar->channels * 4 &&
580 st->codecpar->bits_per_coded_sample == 32 &&
581 st->codecpar->extradata_size == 2 &&
582 AV_RL16(st->codecpar->extradata) == 1) {
583 st->codecpar->codec_id = AV_CODEC_ID_PCM_F16LE;
584 st->codecpar->bits_per_coded_sample = 16;
585 } else if (st->codecpar->codec_id == AV_CODEC_ID_PCM_S24LE &&
586 st->codecpar->block_align == st->codecpar->channels * 4 &&
587 st->codecpar->bits_per_coded_sample == 24) {
588 st->codecpar->codec_id = AV_CODEC_ID_PCM_F24LE;
589 } else if (st->codecpar->codec_id == AV_CODEC_ID_XMA1 ||
590 st->codecpar->codec_id == AV_CODEC_ID_XMA2) {
591 st->codecpar->block_align = 2048;
592 } else if (st->codecpar->codec_id == AV_CODEC_ID_ADPCM_MS && st->codecpar->channels > 2 &&
593 st->codecpar->block_align < INT_MAX / st->codecpar->channels) {
594 st->codecpar->block_align *= st->codecpar->channels;
595 }
596
597 ff_metadata_conv_ctx(s, NULL, wav_metadata_conv);
598 ff_metadata_conv_ctx(s, NULL, ff_riff_info_conv);
599
600 set_spdif(s, wav);
601
602 return 0;
603 }
604
605 /**
606 * Find chunk with w64 GUID by skipping over other chunks.
607 * @return the size of the found chunk
608 */
find_guid(AVIOContext * pb,const uint8_t guid1[16])609 static int64_t find_guid(AVIOContext *pb, const uint8_t guid1[16])
610 {
611 uint8_t guid[16];
612 int64_t size;
613
614 while (!avio_feof(pb)) {
615 avio_read(pb, guid, 16);
616 size = avio_rl64(pb);
617 if (size <= 24 || size > INT64_MAX - 8)
618 return AVERROR_INVALIDDATA;
619 if (!memcmp(guid, guid1, 16))
620 return size;
621 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 24);
622 }
623 return AVERROR_EOF;
624 }
625
626 #define MAX_SIZE 4096
627
wav_read_packet(AVFormatContext * s,AVPacket * pkt)628 static int wav_read_packet(AVFormatContext *s, AVPacket *pkt)
629 {
630 int ret, size;
631 int64_t left;
632 AVStream *st;
633 WAVDemuxContext *wav = s->priv_data;
634
635 if (CONFIG_SPDIF_DEMUXER && wav->spdif == 1)
636 return ff_spdif_read_packet(s, pkt);
637
638 if (wav->smv_data_ofs > 0) {
639 int64_t audio_dts, video_dts;
640 smv_retry:
641 audio_dts = (int32_t)s->streams[0]->cur_dts;
642 video_dts = (int32_t)s->streams[1]->cur_dts;
643
644 if (audio_dts != AV_NOPTS_VALUE && video_dts != AV_NOPTS_VALUE) {
645 /*We always return a video frame first to get the pixel format first*/
646 wav->smv_last_stream = wav->smv_given_first ?
647 av_compare_ts(video_dts, s->streams[1]->time_base,
648 audio_dts, s->streams[0]->time_base) > 0 : 0;
649 wav->smv_given_first = 1;
650 }
651 wav->smv_last_stream = !wav->smv_last_stream;
652 wav->smv_last_stream |= wav->audio_eof;
653 wav->smv_last_stream &= !wav->smv_eof;
654 if (wav->smv_last_stream) {
655 uint64_t old_pos = avio_tell(s->pb);
656 uint64_t new_pos = wav->smv_data_ofs +
657 wav->smv_block * wav->smv_block_size;
658 if (avio_seek(s->pb, new_pos, SEEK_SET) < 0) {
659 ret = AVERROR_EOF;
660 goto smv_out;
661 }
662 size = avio_rl24(s->pb);
663 ret = av_get_packet(s->pb, pkt, size);
664 if (ret < 0)
665 goto smv_out;
666 pkt->pos -= 3;
667 pkt->pts = wav->smv_block * wav->smv_frames_per_jpeg + wav->smv_cur_pt;
668 wav->smv_cur_pt++;
669 if (wav->smv_frames_per_jpeg > 0)
670 wav->smv_cur_pt %= wav->smv_frames_per_jpeg;
671 if (!wav->smv_cur_pt)
672 wav->smv_block++;
673
674 pkt->stream_index = 1;
675 smv_out:
676 avio_seek(s->pb, old_pos, SEEK_SET);
677 if (ret == AVERROR_EOF) {
678 wav->smv_eof = 1;
679 goto smv_retry;
680 }
681 return ret;
682 }
683 }
684
685 st = s->streams[0];
686
687 left = wav->data_end - avio_tell(s->pb);
688 if (wav->ignore_length)
689 left = INT_MAX;
690 if (left <= 0) {
691 if (CONFIG_W64_DEMUXER && wav->w64)
692 left = find_guid(s->pb, ff_w64_guid_data) - 24;
693 else
694 left = find_tag(wav, s->pb, MKTAG('d', 'a', 't', 'a'));
695 if (left < 0) {
696 wav->audio_eof = 1;
697 if (wav->smv_data_ofs > 0 && !wav->smv_eof)
698 goto smv_retry;
699 return AVERROR_EOF;
700 }
701 wav->data_end = avio_tell(s->pb) + left;
702 }
703
704 size = MAX_SIZE;
705 if (st->codecpar->block_align > 1) {
706 if (size < st->codecpar->block_align)
707 size = st->codecpar->block_align;
708 size = (size / st->codecpar->block_align) * st->codecpar->block_align;
709 }
710 size = FFMIN(size, left);
711 ret = av_get_packet(s->pb, pkt, size);
712 if (ret < 0)
713 return ret;
714 pkt->stream_index = 0;
715
716 return ret;
717 }
718
wav_read_seek(AVFormatContext * s,int stream_index,int64_t timestamp,int flags)719 static int wav_read_seek(AVFormatContext *s,
720 int stream_index, int64_t timestamp, int flags)
721 {
722 WAVDemuxContext *wav = s->priv_data;
723 AVStream *st;
724 wav->smv_eof = 0;
725 wav->audio_eof = 0;
726 if (wav->smv_data_ofs > 0) {
727 int64_t smv_timestamp = timestamp;
728 if (stream_index == 0)
729 smv_timestamp = av_rescale_q(timestamp, s->streams[0]->time_base, s->streams[1]->time_base);
730 else
731 timestamp = av_rescale_q(smv_timestamp, s->streams[1]->time_base, s->streams[0]->time_base);
732 if (wav->smv_frames_per_jpeg > 0) {
733 wav->smv_block = smv_timestamp / wav->smv_frames_per_jpeg;
734 wav->smv_cur_pt = smv_timestamp % wav->smv_frames_per_jpeg;
735 }
736 }
737
738 st = s->streams[0];
739 switch (st->codecpar->codec_id) {
740 case AV_CODEC_ID_MP2:
741 case AV_CODEC_ID_MP3:
742 case AV_CODEC_ID_AC3:
743 case AV_CODEC_ID_DTS:
744 case AV_CODEC_ID_XMA2:
745 /* use generic seeking with dynamically generated indexes */
746 return -1;
747 default:
748 break;
749 }
750 return ff_pcm_read_seek(s, stream_index, timestamp, flags);
751 }
752
753 #define OFFSET(x) offsetof(WAVDemuxContext, x)
754 #define DEC AV_OPT_FLAG_DECODING_PARAM
755 static const AVOption demux_options[] = {
756 { "ignore_length", "Ignore length", OFFSET(ignore_length), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, DEC },
757 { NULL },
758 };
759
760 static const AVClass wav_demuxer_class = {
761 .class_name = "WAV demuxer",
762 .item_name = av_default_item_name,
763 .option = demux_options,
764 .version = LIBAVUTIL_VERSION_INT,
765 };
766 AVInputFormat ff_wav_demuxer = {
767 .name = "wav",
768 .long_name = NULL_IF_CONFIG_SMALL("WAV / WAVE (Waveform Audio)"),
769 .priv_data_size = sizeof(WAVDemuxContext),
770 .read_probe = wav_probe,
771 .read_header = wav_read_header,
772 .read_packet = wav_read_packet,
773 .read_seek = wav_read_seek,
774 .flags = AVFMT_GENERIC_INDEX,
775 .codec_tag = (const AVCodecTag * const []) { ff_codec_wav_tags, 0 },
776 .priv_class = &wav_demuxer_class,
777 };
778 #endif /* CONFIG_WAV_DEMUXER */
779
780 #if CONFIG_W64_DEMUXER
w64_probe(const AVProbeData * p)781 static int w64_probe(const AVProbeData *p)
782 {
783 if (p->buf_size <= 40)
784 return 0;
785 if (!memcmp(p->buf, ff_w64_guid_riff, 16) &&
786 !memcmp(p->buf + 24, ff_w64_guid_wave, 16))
787 return AVPROBE_SCORE_MAX;
788 else
789 return 0;
790 }
791
w64_read_header(AVFormatContext * s)792 static int w64_read_header(AVFormatContext *s)
793 {
794 int64_t size, data_ofs = 0;
795 AVIOContext *pb = s->pb;
796 WAVDemuxContext *wav = s->priv_data;
797 AVStream *st;
798 uint8_t guid[16];
799 int ret;
800
801 avio_read(pb, guid, 16);
802 if (memcmp(guid, ff_w64_guid_riff, 16))
803 return AVERROR_INVALIDDATA;
804
805 /* riff + wave + fmt + sizes */
806 if (avio_rl64(pb) < 16 + 8 + 16 + 8 + 16 + 8)
807 return AVERROR_INVALIDDATA;
808
809 avio_read(pb, guid, 16);
810 if (memcmp(guid, ff_w64_guid_wave, 16)) {
811 av_log(s, AV_LOG_ERROR, "could not find wave guid\n");
812 return AVERROR_INVALIDDATA;
813 }
814
815 wav->w64 = 1;
816
817 st = avformat_new_stream(s, NULL);
818 if (!st)
819 return AVERROR(ENOMEM);
820
821 while (!avio_feof(pb)) {
822 if (avio_read(pb, guid, 16) != 16)
823 break;
824 size = avio_rl64(pb);
825 if (size <= 24 || INT64_MAX - size < avio_tell(pb))
826 return AVERROR_INVALIDDATA;
827
828 if (!memcmp(guid, ff_w64_guid_fmt, 16)) {
829 /* subtract chunk header size - normal wav file doesn't count it */
830 ret = ff_get_wav_header(s, pb, st->codecpar, size - 24, 0);
831 if (ret < 0)
832 return ret;
833 avio_skip(pb, FFALIGN(size, INT64_C(8)) - size);
834
835 avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
836 } else if (!memcmp(guid, ff_w64_guid_fact, 16)) {
837 int64_t samples;
838
839 samples = avio_rl64(pb);
840 if (samples > 0)
841 st->duration = samples;
842 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 32);
843 } else if (!memcmp(guid, ff_w64_guid_data, 16)) {
844 wav->data_end = avio_tell(pb) + size - 24;
845
846 data_ofs = avio_tell(pb);
847 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL))
848 break;
849
850 avio_skip(pb, size - 24);
851 } else if (!memcmp(guid, ff_w64_guid_summarylist, 16)) {
852 int64_t start, end, cur;
853 uint32_t count, chunk_size, i;
854 int64_t filesize = avio_size(s->pb);
855
856 start = avio_tell(pb);
857 end = start + FFALIGN(size, INT64_C(8)) - 24;
858 count = avio_rl32(pb);
859
860 for (i = 0; i < count; i++) {
861 char chunk_key[5], *value;
862
863 if (avio_feof(pb) || (cur = avio_tell(pb)) < 0 || cur > end - 8 /* = tag + size */)
864 break;
865
866 chunk_key[4] = 0;
867 avio_read(pb, chunk_key, 4);
868 chunk_size = avio_rl32(pb);
869 if (chunk_size == UINT32_MAX || (filesize >= 0 && chunk_size > filesize))
870 return AVERROR_INVALIDDATA;
871
872 value = av_mallocz(chunk_size + 1);
873 if (!value)
874 return AVERROR(ENOMEM);
875
876 ret = avio_get_str16le(pb, chunk_size, value, chunk_size);
877 if (ret < 0) {
878 av_free(value);
879 return ret;
880 }
881 avio_skip(pb, chunk_size - ret);
882
883 av_dict_set(&s->metadata, chunk_key, value, AV_DICT_DONT_STRDUP_VAL);
884 }
885
886 avio_skip(pb, end - avio_tell(pb));
887 } else {
888 av_log(s, AV_LOG_DEBUG, "unknown guid: "FF_PRI_GUID"\n", FF_ARG_GUID(guid));
889 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 24);
890 }
891 }
892
893 if (!data_ofs)
894 return AVERROR_EOF;
895
896 ff_metadata_conv_ctx(s, NULL, wav_metadata_conv);
897 ff_metadata_conv_ctx(s, NULL, ff_riff_info_conv);
898
899 handle_stream_probing(st);
900 st->need_parsing = AVSTREAM_PARSE_FULL_RAW;
901
902 avio_seek(pb, data_ofs, SEEK_SET);
903
904 set_spdif(s, wav);
905
906 return 0;
907 }
908
909 AVInputFormat ff_w64_demuxer = {
910 .name = "w64",
911 .long_name = NULL_IF_CONFIG_SMALL("Sony Wave64"),
912 .priv_data_size = sizeof(WAVDemuxContext),
913 .read_probe = w64_probe,
914 .read_header = w64_read_header,
915 .read_packet = wav_read_packet,
916 .read_seek = wav_read_seek,
917 .flags = AVFMT_GENERIC_INDEX,
918 .codec_tag = (const AVCodecTag * const []) { ff_codec_wav_tags, 0 },
919 };
920 #endif /* CONFIG_W64_DEMUXER */
921