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 * BW64 demuxer
10 *
11 * This file is part of FFmpeg.
12 *
13 * FFmpeg is free software; you can redistribute it and/or
14 * modify it under the terms of the GNU Lesser General Public
15 * License as published by the Free Software Foundation; either
16 * version 2.1 of the License, or (at your option) any later version.
17 *
18 * FFmpeg is distributed in the hope that it will be useful,
19 * but WITHOUT ANY WARRANTY; without even the implied warranty of
20 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
21 * Lesser General Public License for more details.
22 *
23 * You should have received a copy of the GNU Lesser General Public
24 * License along with FFmpeg; if not, write to the Free Software
25 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
26 */
27
28 #include <stdint.h>
29
30 #include "config.h"
31 #include "libavutil/avassert.h"
32 #include "libavutil/dict.h"
33 #include "libavutil/intreadwrite.h"
34 #include "libavutil/log.h"
35 #include "libavutil/mathematics.h"
36 #include "libavutil/opt.h"
37 #include "avformat.h"
38 #include "avio.h"
39 #include "avio_internal.h"
40 #include "id3v2.h"
41 #include "internal.h"
42 #include "metadata.h"
43 #include "pcm.h"
44 #include "riff.h"
45 #include "w64.h"
46 #include "spdif.h"
47
48 typedef struct WAVDemuxContext {
49 const AVClass *class;
50 int64_t data_end;
51 int w64;
52 int64_t smv_data_ofs;
53 int smv_block_size;
54 int smv_frames_per_jpeg;
55 int smv_block;
56 int smv_last_stream;
57 int smv_eof;
58 int audio_eof;
59 int ignore_length;
60 int max_size;
61 int spdif;
62 int smv_given_first;
63 int unaligned; // e.g. if an odd number of bytes ID3 tag was prepended
64 int rifx; // RIFX: integer byte order for parameters is big endian
65 } WAVDemuxContext;
66
67 #define OFFSET(x) offsetof(WAVDemuxContext, x)
68 #define DEC AV_OPT_FLAG_DECODING_PARAM
69 static const AVOption demux_options[] = {
70 #define W64_DEMUXER_OPTIONS_OFFSET (1 * CONFIG_WAV_DEMUXER)
71 #if CONFIG_WAV_DEMUXER
72 { "ignore_length", "Ignore length", OFFSET(ignore_length), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, DEC },
73 #endif
74 { "max_size", "max size of single packet", OFFSET(max_size), AV_OPT_TYPE_INT, { .i64 = 4096 }, 1024, 1 << 22, DEC },
75 { NULL },
76 };
77
set_spdif(AVFormatContext * s,WAVDemuxContext * wav)78 static void set_spdif(AVFormatContext *s, WAVDemuxContext *wav)
79 {
80 if (CONFIG_SPDIF_DEMUXER && s->streams[0]->codecpar->codec_tag == 1) {
81 enum AVCodecID codec;
82 int len = 1<<16;
83 int ret = ffio_ensure_seekback(s->pb, len);
84
85 if (ret >= 0) {
86 uint8_t *buf = av_malloc(len + AV_INPUT_BUFFER_PADDING_SIZE);
87 if (!buf) {
88 ret = AVERROR(ENOMEM);
89 } else {
90 int64_t pos = avio_tell(s->pb);
91 len = ret = avio_read(s->pb, buf, len);
92 if (len >= 0) {
93 ret = ff_spdif_probe(buf, len, &codec);
94 if (ret > AVPROBE_SCORE_EXTENSION) {
95 s->streams[0]->codecpar->codec_id = codec;
96 wav->spdif = 1;
97 }
98 }
99 avio_seek(s->pb, pos, SEEK_SET);
100 av_free(buf);
101 }
102 }
103
104 if (ret < 0)
105 av_log(s, AV_LOG_WARNING, "Cannot check for SPDIF\n");
106 }
107 }
108
109 #if CONFIG_WAV_DEMUXER
110
next_tag(AVIOContext * pb,uint32_t * tag,int big_endian)111 static int64_t next_tag(AVIOContext *pb, uint32_t *tag, int big_endian)
112 {
113 *tag = avio_rl32(pb);
114 if (!big_endian) {
115 return avio_rl32(pb);
116 } else {
117 return avio_rb32(pb);
118 }
119 }
120
121 /* RIFF chunks are always at even offsets relative to where they start. */
wav_seek_tag(WAVDemuxContext * wav,AVIOContext * s,int64_t offset,int whence)122 static int64_t wav_seek_tag(WAVDemuxContext * wav, AVIOContext *s, int64_t offset, int whence)
123 {
124 offset += offset < INT64_MAX && offset + wav->unaligned & 1;
125
126 return avio_seek(s, offset, whence);
127 }
128
129 /* return the size of the found tag */
find_tag(WAVDemuxContext * wav,AVIOContext * pb,uint32_t tag1)130 static int64_t find_tag(WAVDemuxContext * wav, AVIOContext *pb, uint32_t tag1)
131 {
132 unsigned int tag;
133 int64_t size;
134
135 for (;;) {
136 if (avio_feof(pb))
137 return AVERROR_EOF;
138 size = next_tag(pb, &tag, wav->rifx);
139 if (tag == tag1)
140 break;
141 wav_seek_tag(wav, pb, size, SEEK_CUR);
142 }
143 return size;
144 }
145
wav_probe(const AVProbeData * p)146 static int wav_probe(const AVProbeData *p)
147 {
148 /* check file header */
149 if (p->buf_size <= 32)
150 return 0;
151 if (!memcmp(p->buf + 8, "WAVE", 4)) {
152 if (!memcmp(p->buf, "RIFF", 4) || !memcmp(p->buf, "RIFX", 4))
153 /* Since the ACT demuxer has a standard WAV header at the top of
154 * its own, the returned score is decreased to avoid a probe
155 * conflict between ACT and WAV. */
156 return AVPROBE_SCORE_MAX - 1;
157 else if ((!memcmp(p->buf, "RF64", 4) ||
158 !memcmp(p->buf, "BW64", 4)) &&
159 !memcmp(p->buf + 12, "ds64", 4))
160 return AVPROBE_SCORE_MAX;
161 }
162 return 0;
163 }
164
handle_stream_probing(AVStream * st)165 static void handle_stream_probing(AVStream *st)
166 {
167 if (st->codecpar->codec_id == AV_CODEC_ID_PCM_S16LE) {
168 st->internal->request_probe = AVPROBE_SCORE_EXTENSION;
169 st->probe_packets = FFMIN(st->probe_packets, 32);
170 }
171 }
172
wav_parse_fmt_tag(AVFormatContext * s,int64_t size,AVStream ** st)173 static int wav_parse_fmt_tag(AVFormatContext *s, int64_t size, AVStream **st)
174 {
175 AVIOContext *pb = s->pb;
176 WAVDemuxContext *wav = s->priv_data;
177 int ret;
178
179 /* parse fmt header */
180 *st = avformat_new_stream(s, NULL);
181 if (!*st)
182 return AVERROR(ENOMEM);
183
184 ret = ff_get_wav_header(s, pb, (*st)->codecpar, size, wav->rifx);
185 if (ret < 0)
186 return ret;
187 handle_stream_probing(*st);
188
189 (*st)->need_parsing = AVSTREAM_PARSE_FULL_RAW;
190
191 avpriv_set_pts_info(*st, 64, 1, (*st)->codecpar->sample_rate);
192
193 return 0;
194 }
195
wav_parse_xma2_tag(AVFormatContext * s,int64_t size,AVStream ** st)196 static int wav_parse_xma2_tag(AVFormatContext *s, int64_t size, AVStream **st)
197 {
198 AVIOContext *pb = s->pb;
199 int version, num_streams, i, channels = 0, ret;
200
201 if (size < 36)
202 return AVERROR_INVALIDDATA;
203
204 *st = avformat_new_stream(s, NULL);
205 if (!*st)
206 return AVERROR(ENOMEM);
207
208 (*st)->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
209 (*st)->codecpar->codec_id = AV_CODEC_ID_XMA2;
210 (*st)->need_parsing = AVSTREAM_PARSE_FULL_RAW;
211
212 version = avio_r8(pb);
213 if (version != 3 && version != 4)
214 return AVERROR_INVALIDDATA;
215 num_streams = avio_r8(pb);
216 if (size != (32 + ((version==3)?0:8) + 4*num_streams))
217 return AVERROR_INVALIDDATA;
218 avio_skip(pb, 10);
219 (*st)->codecpar->sample_rate = avio_rb32(pb);
220 if (version == 4)
221 avio_skip(pb, 8);
222 avio_skip(pb, 4);
223 (*st)->duration = avio_rb32(pb);
224 avio_skip(pb, 8);
225
226 for (i = 0; i < num_streams; i++) {
227 channels += avio_r8(pb);
228 avio_skip(pb, 3);
229 }
230 (*st)->codecpar->channels = channels;
231
232 if ((*st)->codecpar->channels <= 0 || (*st)->codecpar->sample_rate <= 0)
233 return AVERROR_INVALIDDATA;
234
235 avpriv_set_pts_info(*st, 64, 1, (*st)->codecpar->sample_rate);
236
237 avio_seek(pb, -size, SEEK_CUR);
238 if ((ret = ff_get_extradata(s, (*st)->codecpar, pb, size)) < 0)
239 return ret;
240
241 return 0;
242 }
243
wav_parse_bext_string(AVFormatContext * s,const char * key,int length)244 static inline int wav_parse_bext_string(AVFormatContext *s, const char *key,
245 int length)
246 {
247 char temp[257];
248 int ret;
249
250 av_assert0(length < sizeof(temp));
251 if ((ret = avio_read(s->pb, temp, length)) != length)
252 return ret < 0 ? ret : AVERROR_INVALIDDATA;
253
254 temp[length] = 0;
255
256 if (strlen(temp))
257 return av_dict_set(&s->metadata, key, temp, 0);
258
259 return 0;
260 }
261
wav_parse_bext_tag(AVFormatContext * s,int64_t size)262 static int wav_parse_bext_tag(AVFormatContext *s, int64_t size)
263 {
264 char temp[131], *coding_history;
265 int ret, x;
266 uint64_t time_reference;
267 int64_t umid_parts[8], umid_mask = 0;
268
269 if ((ret = wav_parse_bext_string(s, "description", 256)) < 0 ||
270 (ret = wav_parse_bext_string(s, "originator", 32)) < 0 ||
271 (ret = wav_parse_bext_string(s, "originator_reference", 32)) < 0 ||
272 (ret = wav_parse_bext_string(s, "origination_date", 10)) < 0 ||
273 (ret = wav_parse_bext_string(s, "origination_time", 8)) < 0)
274 return ret;
275
276 time_reference = avio_rl64(s->pb);
277 snprintf(temp, sizeof(temp), "%"PRIu64, time_reference);
278 if ((ret = av_dict_set(&s->metadata, "time_reference", temp, 0)) < 0)
279 return ret;
280
281 /* check if version is >= 1, in which case an UMID may be present */
282 if (avio_rl16(s->pb) >= 1) {
283 for (x = 0; x < 8; x++)
284 umid_mask |= umid_parts[x] = avio_rb64(s->pb);
285
286 if (umid_mask) {
287 /* the string formatting below is per SMPTE 330M-2004 Annex C */
288 if (umid_parts[4] == 0 && umid_parts[5] == 0 &&
289 umid_parts[6] == 0 && umid_parts[7] == 0) {
290 /* basic UMID */
291 snprintf(temp, sizeof(temp),
292 "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
293 umid_parts[0], umid_parts[1],
294 umid_parts[2], umid_parts[3]);
295 } else {
296 /* extended UMID */
297 snprintf(temp, sizeof(temp),
298 "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64
299 "%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
300 umid_parts[0], umid_parts[1],
301 umid_parts[2], umid_parts[3],
302 umid_parts[4], umid_parts[5],
303 umid_parts[6], umid_parts[7]);
304 }
305
306 if ((ret = av_dict_set(&s->metadata, "umid", temp, 0)) < 0)
307 return ret;
308 }
309
310 avio_skip(s->pb, 190);
311 } else
312 avio_skip(s->pb, 254);
313
314 if (size > 602) {
315 /* CodingHistory present */
316 size -= 602;
317
318 if (!(coding_history = av_malloc(size + 1)))
319 return AVERROR(ENOMEM);
320
321 if ((ret = avio_read(s->pb, coding_history, size)) != size) {
322 av_free(coding_history);
323 return ret < 0 ? ret : AVERROR_INVALIDDATA;
324 }
325
326 coding_history[size] = 0;
327 if ((ret = av_dict_set(&s->metadata, "coding_history", coding_history,
328 AV_DICT_DONT_STRDUP_VAL)) < 0)
329 return ret;
330 }
331
332 return 0;
333 }
334
335 static const AVMetadataConv wav_metadata_conv[] = {
336 { "description", "comment" },
337 { "originator", "encoded_by" },
338 { "origination_date", "date" },
339 { "origination_time", "creation_time" },
340 { 0 },
341 };
342
343 /* wav input */
wav_read_header(AVFormatContext * s)344 static int wav_read_header(AVFormatContext *s)
345 {
346 int64_t size, av_uninit(data_size);
347 int64_t sample_count = 0;
348 int rf64 = 0, bw64 = 0;
349 uint32_t tag;
350 AVIOContext *pb = s->pb;
351 AVStream *st = NULL;
352 WAVDemuxContext *wav = s->priv_data;
353 int ret, got_fmt = 0, got_xma2 = 0;
354 int64_t next_tag_ofs, data_ofs = -1;
355
356 wav->unaligned = avio_tell(s->pb) & 1;
357
358 wav->smv_data_ofs = -1;
359
360 /* read chunk ID */
361 tag = avio_rl32(pb);
362 switch (tag) {
363 case MKTAG('R', 'I', 'F', 'F'):
364 break;
365 case MKTAG('R', 'I', 'F', 'X'):
366 wav->rifx = 1;
367 break;
368 case MKTAG('R', 'F', '6', '4'):
369 rf64 = 1;
370 break;
371 case MKTAG('B', 'W', '6', '4'):
372 bw64 = 1;
373 break;
374 default:
375 av_log(s, AV_LOG_ERROR, "invalid start code %s in RIFF header\n",
376 av_fourcc2str(tag));
377 return AVERROR_INVALIDDATA;
378 }
379
380 /* read chunk size */
381 avio_rl32(pb);
382
383 /* read format */
384 if (avio_rl32(pb) != MKTAG('W', 'A', 'V', 'E')) {
385 av_log(s, AV_LOG_ERROR, "invalid format in RIFF header\n");
386 return AVERROR_INVALIDDATA;
387 }
388
389 if (rf64 || bw64) {
390 if (avio_rl32(pb) != MKTAG('d', 's', '6', '4'))
391 return AVERROR_INVALIDDATA;
392 size = avio_rl32(pb);
393 if (size < 24)
394 return AVERROR_INVALIDDATA;
395 avio_rl64(pb); /* RIFF size */
396
397 data_size = avio_rl64(pb);
398 sample_count = avio_rl64(pb);
399
400 if (data_size < 0 || sample_count < 0) {
401 av_log(s, AV_LOG_ERROR, "negative data_size and/or sample_count in "
402 "ds64: data_size = %"PRId64", sample_count = %"PRId64"\n",
403 data_size, sample_count);
404 return AVERROR_INVALIDDATA;
405 }
406 avio_skip(pb, size - 24); /* skip rest of ds64 chunk */
407
408 }
409
410 for (;;) {
411 AVStream *vst;
412 size = next_tag(pb, &tag, wav->rifx);
413 next_tag_ofs = avio_tell(pb) + size;
414
415 if (avio_feof(pb))
416 break;
417
418 switch (tag) {
419 case MKTAG('f', 'm', 't', ' '):
420 /* only parse the first 'fmt ' tag found */
421 if (!got_xma2 && !got_fmt && (ret = wav_parse_fmt_tag(s, size, &st)) < 0) {
422 return ret;
423 } else if (got_fmt)
424 av_log(s, AV_LOG_WARNING, "found more than one 'fmt ' tag\n");
425
426 got_fmt = 1;
427 break;
428 case MKTAG('X', 'M', 'A', '2'):
429 /* only parse the first 'XMA2' tag found */
430 if (!got_fmt && !got_xma2 && (ret = wav_parse_xma2_tag(s, size, &st)) < 0) {
431 return ret;
432 } else if (got_xma2)
433 av_log(s, AV_LOG_WARNING, "found more than one 'XMA2' tag\n");
434
435 got_xma2 = 1;
436 break;
437 case MKTAG('d', 'a', 't', 'a'):
438 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) && !got_fmt && !got_xma2) {
439 av_log(s, AV_LOG_ERROR,
440 "found no 'fmt ' tag before the 'data' tag\n");
441 return AVERROR_INVALIDDATA;
442 }
443
444 if (rf64 || bw64) {
445 next_tag_ofs = wav->data_end = avio_tell(pb) + data_size;
446 } else if (size != 0xFFFFFFFF) {
447 data_size = size;
448 next_tag_ofs = wav->data_end = size ? next_tag_ofs : INT64_MAX;
449 } else {
450 av_log(s, AV_LOG_WARNING, "Ignoring maximum wav data size, "
451 "file may be invalid\n");
452 data_size = 0;
453 next_tag_ofs = wav->data_end = INT64_MAX;
454 }
455
456 data_ofs = avio_tell(pb);
457
458 /* don't look for footer metadata if we can't seek or if we don't
459 * know where the data tag ends
460 */
461 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL) || (!(rf64 && !bw64) && !size))
462 goto break_loop;
463 break;
464 case MKTAG('f', 'a', 'c', 't'):
465 if (!sample_count)
466 sample_count = (!wav->rifx ? avio_rl32(pb) : avio_rb32(pb));
467 break;
468 case MKTAG('b', 'e', 'x', 't'):
469 if ((ret = wav_parse_bext_tag(s, size)) < 0)
470 return ret;
471 break;
472 case MKTAG('S','M','V','0'):
473 if (!got_fmt) {
474 av_log(s, AV_LOG_ERROR, "found no 'fmt ' tag before the 'SMV0' tag\n");
475 return AVERROR_INVALIDDATA;
476 }
477 // SMV file, a wav file with video appended.
478 if (size != MKTAG('0','2','0','0')) {
479 av_log(s, AV_LOG_ERROR, "Unknown SMV version found\n");
480 goto break_loop;
481 }
482 av_log(s, AV_LOG_DEBUG, "Found SMV data\n");
483 wav->smv_given_first = 0;
484 vst = avformat_new_stream(s, NULL);
485 if (!vst)
486 return AVERROR(ENOMEM);
487 avio_r8(pb);
488 vst->id = 1;
489 vst->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
490 vst->codecpar->codec_id = AV_CODEC_ID_SMVJPEG;
491 vst->codecpar->width = avio_rl24(pb);
492 vst->codecpar->height = avio_rl24(pb);
493 if ((ret = ff_alloc_extradata(vst->codecpar, 4)) < 0) {
494 av_log(s, AV_LOG_ERROR, "Could not allocate extradata.\n");
495 return ret;
496 }
497 size = avio_rl24(pb);
498 wav->smv_data_ofs = avio_tell(pb) + (size - 5) * 3;
499 avio_rl24(pb);
500 wav->smv_block_size = avio_rl24(pb);
501 if (!wav->smv_block_size)
502 return AVERROR_INVALIDDATA;
503 avpriv_set_pts_info(vst, 32, 1, avio_rl24(pb));
504 vst->duration = avio_rl24(pb);
505 avio_rl24(pb);
506 avio_rl24(pb);
507 wav->smv_frames_per_jpeg = avio_rl24(pb);
508 if (wav->smv_frames_per_jpeg > 65536) {
509 av_log(s, AV_LOG_ERROR, "too many frames per jpeg\n");
510 return AVERROR_INVALIDDATA;
511 }
512 AV_WL32(vst->codecpar->extradata, wav->smv_frames_per_jpeg);
513 goto break_loop;
514 case MKTAG('L', 'I', 'S', 'T'):
515 case MKTAG('l', 'i', 's', 't'):
516 if (size < 4) {
517 av_log(s, AV_LOG_ERROR, "too short LIST tag\n");
518 return AVERROR_INVALIDDATA;
519 }
520 switch (avio_rl32(pb)) {
521 case MKTAG('I', 'N', 'F', 'O'):
522 ff_read_riff_info(s, size - 4);
523 break;
524 case MKTAG('a', 'd', 't', 'l'):
525 if (s->nb_chapters > 0) {
526 while (avio_tell(pb) < next_tag_ofs &&
527 !avio_feof(pb)) {
528 char cue_label[512];
529 unsigned id, sub_size;
530
531 if (avio_rl32(pb) != MKTAG('l', 'a', 'b', 'l'))
532 break;
533
534 sub_size = avio_rl32(pb);
535 if (sub_size < 5)
536 break;
537 id = avio_rl32(pb);
538 avio_get_str(pb, sub_size - 4, cue_label, sizeof(cue_label));
539 avio_skip(pb, avio_tell(pb) & 1);
540
541 for (int i = 0; i < s->nb_chapters; i++) {
542 if (s->chapters[i]->id == id) {
543 av_dict_set(&s->chapters[i]->metadata, "title", cue_label, 0);
544 break;
545 }
546 }
547 }
548 }
549 break;
550 }
551 break;
552 case MKTAG('I', 'D', '3', ' '):
553 case MKTAG('i', 'd', '3', ' '): {
554 ID3v2ExtraMeta *id3v2_extra_meta = NULL;
555 ff_id3v2_read_dict(pb, &s->internal->id3v2_meta, ID3v2_DEFAULT_MAGIC, &id3v2_extra_meta);
556 if (id3v2_extra_meta) {
557 ff_id3v2_parse_apic(s, id3v2_extra_meta);
558 ff_id3v2_parse_chapters(s, id3v2_extra_meta);
559 ff_id3v2_parse_priv(s, id3v2_extra_meta);
560 }
561 ff_id3v2_free_extra_meta(&id3v2_extra_meta);
562 }
563 break;
564 case MKTAG('c', 'u', 'e', ' '):
565 if (size >= 4 && got_fmt && st->codecpar->sample_rate > 0) {
566 AVRational tb = {1, st->codecpar->sample_rate};
567 unsigned nb_cues = avio_rl32(pb);
568
569 if (size >= nb_cues * 24LL + 4LL) {
570 for (int i = 0; i < nb_cues; i++) {
571 unsigned offset, id = avio_rl32(pb);
572
573 if (avio_feof(pb))
574 return AVERROR_INVALIDDATA;
575
576 avio_skip(pb, 16);
577 offset = avio_rl32(pb);
578
579 if (!avpriv_new_chapter(s, id, tb, offset, AV_NOPTS_VALUE, NULL))
580 return AVERROR(ENOMEM);
581 }
582 }
583 }
584 break;
585 }
586
587 /* seek to next tag unless we know that we'll run into EOF */
588 if ((avio_size(pb) > 0 && next_tag_ofs >= avio_size(pb)) ||
589 wav_seek_tag(wav, pb, next_tag_ofs, SEEK_SET) < 0) {
590 break;
591 }
592 }
593
594 break_loop:
595 if (!got_fmt && !got_xma2) {
596 av_log(s, AV_LOG_ERROR, "no 'fmt ' or 'XMA2' tag found\n");
597 return AVERROR_INVALIDDATA;
598 }
599
600 if (data_ofs < 0) {
601 av_log(s, AV_LOG_ERROR, "no 'data' tag found\n");
602 return AVERROR_INVALIDDATA;
603 }
604
605 avio_seek(pb, data_ofs, SEEK_SET);
606
607 if (data_size > (INT64_MAX>>3)) {
608 av_log(s, AV_LOG_WARNING, "Data size %"PRId64" is too large\n", data_size);
609 data_size = 0;
610 }
611
612 if ( st->codecpar->bit_rate > 0 && data_size > 0
613 && st->codecpar->sample_rate > 0
614 && sample_count > 0 && st->codecpar->channels > 1
615 && sample_count % st->codecpar->channels == 0) {
616 if (fabs(8.0 * data_size * st->codecpar->channels * st->codecpar->sample_rate /
617 sample_count /st->codecpar->bit_rate - 1.0) < 0.3)
618 sample_count /= st->codecpar->channels;
619 }
620
621 if ( data_size > 0 && sample_count && st->codecpar->channels
622 && (data_size << 3) / sample_count / st->codecpar->channels > st->codecpar->bits_per_coded_sample + 1) {
623 av_log(s, AV_LOG_WARNING, "ignoring wrong sample_count %"PRId64"\n", sample_count);
624 sample_count = 0;
625 }
626
627 /* G.729 hack (for Ticket4577)
628 * FIXME: Come up with cleaner, more general solution */
629 if (st->codecpar->codec_id == AV_CODEC_ID_G729 && sample_count && (data_size << 3) > sample_count) {
630 av_log(s, AV_LOG_WARNING, "ignoring wrong sample_count %"PRId64"\n", sample_count);
631 sample_count = 0;
632 }
633
634 if (!sample_count || av_get_exact_bits_per_sample(st->codecpar->codec_id) > 0)
635 if ( st->codecpar->channels
636 && data_size
637 && av_get_bits_per_sample(st->codecpar->codec_id)
638 && wav->data_end <= avio_size(pb))
639 sample_count = (data_size << 3)
640 /
641 (st->codecpar->channels * (uint64_t)av_get_bits_per_sample(st->codecpar->codec_id));
642
643 if (sample_count)
644 st->duration = sample_count;
645
646 if (st->codecpar->codec_id == AV_CODEC_ID_PCM_S32LE &&
647 st->codecpar->block_align == st->codecpar->channels * 4 &&
648 st->codecpar->bits_per_coded_sample == 32 &&
649 st->codecpar->extradata_size == 2 &&
650 AV_RL16(st->codecpar->extradata) == 1) {
651 st->codecpar->codec_id = AV_CODEC_ID_PCM_F16LE;
652 st->codecpar->bits_per_coded_sample = 16;
653 } else if (st->codecpar->codec_id == AV_CODEC_ID_PCM_S24LE &&
654 st->codecpar->block_align == st->codecpar->channels * 4 &&
655 st->codecpar->bits_per_coded_sample == 24) {
656 st->codecpar->codec_id = AV_CODEC_ID_PCM_F24LE;
657 } else if (st->codecpar->codec_id == AV_CODEC_ID_XMA1 ||
658 st->codecpar->codec_id == AV_CODEC_ID_XMA2) {
659 st->codecpar->block_align = 2048;
660 } else if (st->codecpar->codec_id == AV_CODEC_ID_ADPCM_MS && st->codecpar->channels > 2 &&
661 st->codecpar->block_align < INT_MAX / st->codecpar->channels) {
662 st->codecpar->block_align *= st->codecpar->channels;
663 }
664
665 ff_metadata_conv_ctx(s, NULL, wav_metadata_conv);
666 ff_metadata_conv_ctx(s, NULL, ff_riff_info_conv);
667
668 set_spdif(s, wav);
669
670 return 0;
671 }
672
673 /**
674 * Find chunk with w64 GUID by skipping over other chunks.
675 * @return the size of the found chunk
676 */
find_guid(AVIOContext * pb,const uint8_t guid1[16])677 static int64_t find_guid(AVIOContext *pb, const uint8_t guid1[16])
678 {
679 uint8_t guid[16];
680 int64_t size;
681
682 while (!avio_feof(pb)) {
683 avio_read(pb, guid, 16);
684 size = avio_rl64(pb);
685 if (size <= 24 || size > INT64_MAX - 8)
686 return AVERROR_INVALIDDATA;
687 if (!memcmp(guid, guid1, 16))
688 return size;
689 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 24);
690 }
691 return AVERROR_EOF;
692 }
693
wav_read_packet(AVFormatContext * s,AVPacket * pkt)694 static int wav_read_packet(AVFormatContext *s, AVPacket *pkt)
695 {
696 int ret, size;
697 int64_t left;
698 AVStream *st;
699 WAVDemuxContext *wav = s->priv_data;
700
701 if (CONFIG_SPDIF_DEMUXER && wav->spdif == 1)
702 return ff_spdif_read_packet(s, pkt);
703
704 if (wav->smv_data_ofs > 0) {
705 int64_t audio_dts, video_dts;
706 smv_retry:
707 audio_dts = (int32_t)s->streams[0]->cur_dts;
708 video_dts = (int32_t)s->streams[1]->cur_dts;
709
710 if (audio_dts != AV_NOPTS_VALUE && video_dts != AV_NOPTS_VALUE) {
711 /*We always return a video frame first to get the pixel format first*/
712 wav->smv_last_stream = wav->smv_given_first ?
713 av_compare_ts(video_dts, s->streams[1]->time_base,
714 audio_dts, s->streams[0]->time_base) > 0 : 0;
715 wav->smv_given_first = 1;
716 }
717 wav->smv_last_stream = !wav->smv_last_stream;
718 wav->smv_last_stream |= wav->audio_eof;
719 wav->smv_last_stream &= !wav->smv_eof;
720 if (wav->smv_last_stream) {
721 uint64_t old_pos = avio_tell(s->pb);
722 uint64_t new_pos = wav->smv_data_ofs +
723 wav->smv_block * (int64_t)wav->smv_block_size;
724 if (avio_seek(s->pb, new_pos, SEEK_SET) < 0) {
725 ret = AVERROR_EOF;
726 goto smv_out;
727 }
728 size = avio_rl24(s->pb);
729 ret = av_get_packet(s->pb, pkt, size);
730 if (ret < 0)
731 goto smv_out;
732 pkt->pos -= 3;
733 pkt->pts = wav->smv_block * wav->smv_frames_per_jpeg;
734 pkt->duration = wav->smv_frames_per_jpeg;
735 wav->smv_block++;
736
737 pkt->stream_index = 1;
738 smv_out:
739 avio_seek(s->pb, old_pos, SEEK_SET);
740 if (ret == AVERROR_EOF) {
741 wav->smv_eof = 1;
742 goto smv_retry;
743 }
744 return ret;
745 }
746 }
747
748 st = s->streams[0];
749
750 left = wav->data_end - avio_tell(s->pb);
751 if (wav->ignore_length)
752 left = INT_MAX;
753 if (left <= 0) {
754 if (CONFIG_W64_DEMUXER && wav->w64)
755 left = find_guid(s->pb, ff_w64_guid_data) - 24;
756 else
757 left = find_tag(wav, s->pb, MKTAG('d', 'a', 't', 'a'));
758 if (left < 0) {
759 wav->audio_eof = 1;
760 if (wav->smv_data_ofs > 0 && !wav->smv_eof)
761 goto smv_retry;
762 return AVERROR_EOF;
763 }
764 wav->data_end = avio_tell(s->pb) + left;
765 }
766
767 size = wav->max_size;
768 if (st->codecpar->block_align > 1) {
769 if (size < st->codecpar->block_align)
770 size = st->codecpar->block_align;
771 size = (size / st->codecpar->block_align) * st->codecpar->block_align;
772 }
773 size = FFMIN(size, left);
774 ret = av_get_packet(s->pb, pkt, size);
775 if (ret < 0)
776 return ret;
777 pkt->stream_index = 0;
778
779 return ret;
780 }
781
wav_read_seek(AVFormatContext * s,int stream_index,int64_t timestamp,int flags)782 static int wav_read_seek(AVFormatContext *s,
783 int stream_index, int64_t timestamp, int flags)
784 {
785 WAVDemuxContext *wav = s->priv_data;
786 AVStream *st;
787 wav->smv_eof = 0;
788 wav->audio_eof = 0;
789 if (wav->smv_data_ofs > 0) {
790 int64_t smv_timestamp = timestamp;
791 if (stream_index == 0)
792 smv_timestamp = av_rescale_q(timestamp, s->streams[0]->time_base, s->streams[1]->time_base);
793 else
794 timestamp = av_rescale_q(smv_timestamp, s->streams[1]->time_base, s->streams[0]->time_base);
795 if (wav->smv_frames_per_jpeg > 0) {
796 wav->smv_block = smv_timestamp / wav->smv_frames_per_jpeg;
797 }
798 }
799
800 st = s->streams[0];
801 switch (st->codecpar->codec_id) {
802 case AV_CODEC_ID_MP2:
803 case AV_CODEC_ID_MP3:
804 case AV_CODEC_ID_AC3:
805 case AV_CODEC_ID_DTS:
806 case AV_CODEC_ID_XMA2:
807 /* use generic seeking with dynamically generated indexes */
808 return -1;
809 default:
810 break;
811 }
812 return ff_pcm_read_seek(s, stream_index, timestamp, flags);
813 }
814
815 static const AVClass wav_demuxer_class = {
816 .class_name = "WAV demuxer",
817 .item_name = av_default_item_name,
818 .option = demux_options,
819 .version = LIBAVUTIL_VERSION_INT,
820 };
821 AVInputFormat ff_wav_demuxer = {
822 .name = "wav",
823 .long_name = NULL_IF_CONFIG_SMALL("WAV / WAVE (Waveform Audio)"),
824 .priv_data_size = sizeof(WAVDemuxContext),
825 .read_probe = wav_probe,
826 .read_header = wav_read_header,
827 .read_packet = wav_read_packet,
828 .read_seek = wav_read_seek,
829 .flags = AVFMT_GENERIC_INDEX,
830 .codec_tag = ff_wav_codec_tags_list,
831 .priv_class = &wav_demuxer_class,
832 };
833 #endif /* CONFIG_WAV_DEMUXER */
834
835 #if CONFIG_W64_DEMUXER
w64_probe(const AVProbeData * p)836 static int w64_probe(const AVProbeData *p)
837 {
838 if (p->buf_size <= 40)
839 return 0;
840 if (!memcmp(p->buf, ff_w64_guid_riff, 16) &&
841 !memcmp(p->buf + 24, ff_w64_guid_wave, 16))
842 return AVPROBE_SCORE_MAX;
843 else
844 return 0;
845 }
846
w64_read_header(AVFormatContext * s)847 static int w64_read_header(AVFormatContext *s)
848 {
849 int64_t size, data_ofs = 0;
850 AVIOContext *pb = s->pb;
851 WAVDemuxContext *wav = s->priv_data;
852 AVStream *st;
853 uint8_t guid[16];
854 int ret;
855
856 avio_read(pb, guid, 16);
857 if (memcmp(guid, ff_w64_guid_riff, 16))
858 return AVERROR_INVALIDDATA;
859
860 /* riff + wave + fmt + sizes */
861 if (avio_rl64(pb) < 16 + 8 + 16 + 8 + 16 + 8)
862 return AVERROR_INVALIDDATA;
863
864 avio_read(pb, guid, 16);
865 if (memcmp(guid, ff_w64_guid_wave, 16)) {
866 av_log(s, AV_LOG_ERROR, "could not find wave guid\n");
867 return AVERROR_INVALIDDATA;
868 }
869
870 wav->w64 = 1;
871
872 st = avformat_new_stream(s, NULL);
873 if (!st)
874 return AVERROR(ENOMEM);
875
876 while (!avio_feof(pb)) {
877 if (avio_read(pb, guid, 16) != 16)
878 break;
879 size = avio_rl64(pb);
880 if (size <= 24 || INT64_MAX - size < avio_tell(pb))
881 return AVERROR_INVALIDDATA;
882
883 if (!memcmp(guid, ff_w64_guid_fmt, 16)) {
884 /* subtract chunk header size - normal wav file doesn't count it */
885 ret = ff_get_wav_header(s, pb, st->codecpar, size - 24, 0);
886 if (ret < 0)
887 return ret;
888 avio_skip(pb, FFALIGN(size, INT64_C(8)) - size);
889
890 avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
891 } else if (!memcmp(guid, ff_w64_guid_fact, 16)) {
892 int64_t samples;
893
894 samples = avio_rl64(pb);
895 if (samples > 0)
896 st->duration = samples;
897 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 32);
898 } else if (!memcmp(guid, ff_w64_guid_data, 16)) {
899 wav->data_end = avio_tell(pb) + size - 24;
900
901 data_ofs = avio_tell(pb);
902 if (!(pb->seekable & AVIO_SEEKABLE_NORMAL))
903 break;
904
905 avio_skip(pb, size - 24);
906 } else if (!memcmp(guid, ff_w64_guid_summarylist, 16)) {
907 int64_t start, end, cur;
908 uint32_t count, chunk_size, i;
909 int64_t filesize = avio_size(s->pb);
910
911 start = avio_tell(pb);
912 end = start + FFALIGN(size, INT64_C(8)) - 24;
913 count = avio_rl32(pb);
914
915 for (i = 0; i < count; i++) {
916 char chunk_key[5], *value;
917
918 if (avio_feof(pb) || (cur = avio_tell(pb)) < 0 || cur > end - 8 /* = tag + size */)
919 break;
920
921 chunk_key[4] = 0;
922 avio_read(pb, chunk_key, 4);
923 chunk_size = avio_rl32(pb);
924 if (chunk_size == UINT32_MAX || (filesize >= 0 && chunk_size > filesize))
925 return AVERROR_INVALIDDATA;
926
927 value = av_malloc(chunk_size + 1);
928 if (!value)
929 return AVERROR(ENOMEM);
930
931 ret = avio_get_str16le(pb, chunk_size, value, chunk_size);
932 if (ret < 0) {
933 av_free(value);
934 return ret;
935 }
936 avio_skip(pb, chunk_size - ret);
937
938 av_dict_set(&s->metadata, chunk_key, value, AV_DICT_DONT_STRDUP_VAL);
939 }
940
941 avio_skip(pb, end - avio_tell(pb));
942 } else {
943 av_log(s, AV_LOG_DEBUG, "unknown guid: "FF_PRI_GUID"\n", FF_ARG_GUID(guid));
944 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 24);
945 }
946 }
947
948 if (!data_ofs)
949 return AVERROR_EOF;
950
951 ff_metadata_conv_ctx(s, NULL, wav_metadata_conv);
952 ff_metadata_conv_ctx(s, NULL, ff_riff_info_conv);
953
954 handle_stream_probing(st);
955 st->need_parsing = AVSTREAM_PARSE_FULL_RAW;
956
957 avio_seek(pb, data_ofs, SEEK_SET);
958
959 set_spdif(s, wav);
960
961 return 0;
962 }
963
964 static const AVClass w64_demuxer_class = {
965 .class_name = "W64 demuxer",
966 .item_name = av_default_item_name,
967 .option = &demux_options[W64_DEMUXER_OPTIONS_OFFSET],
968 .version = LIBAVUTIL_VERSION_INT,
969 };
970
971 AVInputFormat ff_w64_demuxer = {
972 .name = "w64",
973 .long_name = NULL_IF_CONFIG_SMALL("Sony Wave64"),
974 .priv_data_size = sizeof(WAVDemuxContext),
975 .read_probe = w64_probe,
976 .read_header = w64_read_header,
977 .read_packet = wav_read_packet,
978 .read_seek = wav_read_seek,
979 .flags = AVFMT_GENERIC_INDEX,
980 .codec_tag = ff_wav_codec_tags_list,
981 .priv_class = &w64_demuxer_class,
982 };
983 #endif /* CONFIG_W64_DEMUXER */
984