2 * WAV muxer and demuxer
3 * Copyright (c) 2001, 2002 Fabrice Bellard
7 * Copyright (c) 2009 Daniel Verkamp
9 * This file is part of FFmpeg.
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.
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.
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
26 #include "libavutil/avassert.h"
27 #include "libavutil/dict.h"
28 #include "libavutil/log.h"
29 #include "libavutil/mathematics.h"
30 #include "libavutil/opt.h"
32 #include "avio_internal.h"
36 #include "avio_internal.h"
50 int smv_frames_per_jpeg;
58 static inline void bwf_write_bext_string(AVFormatContext *s, const char *key, int maxlen)
60 AVDictionaryEntry *tag;
63 if (tag = av_dict_get(s->metadata, key, NULL, 0)) {
64 len = strlen(tag->value);
65 len = FFMIN(len, maxlen);
66 avio_write(s->pb, tag->value, len);
69 ffio_fill(s->pb, 0, maxlen - len);
72 static void bwf_write_bext_chunk(AVFormatContext *s)
74 AVDictionaryEntry *tmp_tag;
75 uint64_t time_reference = 0;
76 int64_t bext = ff_start_tag(s->pb, "bext");
78 bwf_write_bext_string(s, "description", 256);
79 bwf_write_bext_string(s, "originator", 32);
80 bwf_write_bext_string(s, "originator_reference", 32);
81 bwf_write_bext_string(s, "origination_date", 10);
82 bwf_write_bext_string(s, "origination_time", 8);
84 if (tmp_tag = av_dict_get(s->metadata, "time_reference", NULL, 0))
85 time_reference = strtoll(tmp_tag->value, NULL, 10);
86 avio_wl64(s->pb, time_reference);
87 avio_wl16(s->pb, 1); // set version to 1
89 if (tmp_tag = av_dict_get(s->metadata, "umid", NULL, 0)) {
90 unsigned char umidpart_str[17] = {0};
93 int len = strlen(tmp_tag->value+2);
95 for (i = 0; i < len/16; i++) {
96 memcpy(umidpart_str, tmp_tag->value + 2 + (i*16), 16);
97 umidpart = strtoll(umidpart_str, NULL, 16);
98 avio_wb64(s->pb, umidpart);
100 ffio_fill(s->pb, 0, 64 - i*8);
102 ffio_fill(s->pb, 0, 64); // zero UMID
104 ffio_fill(s->pb, 0, 190); // Reserved
106 if (tmp_tag = av_dict_get(s->metadata, "coding_history", NULL, 0))
107 avio_put_str(s->pb, tmp_tag->value);
109 ff_end_tag(s->pb, bext);
112 static int wav_write_header(AVFormatContext *s)
114 WAVContext *wav = s->priv_data;
115 AVIOContext *pb = s->pb;
118 ffio_wfourcc(pb, "RIFF");
119 avio_wl32(pb, 0); /* file length */
120 ffio_wfourcc(pb, "WAVE");
123 fmt = ff_start_tag(pb, "fmt ");
124 if (ff_put_wav_header(pb, s->streams[0]->codec) < 0) {
125 av_log(s, AV_LOG_ERROR, "%s codec not supported in WAVE format\n",
126 s->streams[0]->codec->codec ? s->streams[0]->codec->codec->name : "NONE");
131 if (s->streams[0]->codec->codec_tag != 0x01 /* hence for all other than PCM */
132 && s->pb->seekable) {
133 fact = ff_start_tag(pb, "fact");
135 ff_end_tag(pb, fact);
139 bwf_write_bext_chunk(s);
141 av_set_pts_info(s->streams[0], 64, 1, s->streams[0]->codec->sample_rate);
142 wav->maxpts = wav->last_duration = 0;
143 wav->minpts = INT64_MAX;
146 wav->data = ff_start_tag(pb, "data");
153 static int wav_write_packet(AVFormatContext *s, AVPacket *pkt)
155 AVIOContext *pb = s->pb;
156 WAVContext *wav = s->priv_data;
157 avio_write(pb, pkt->data, pkt->size);
158 if(pkt->pts != AV_NOPTS_VALUE) {
159 wav->minpts = FFMIN(wav->minpts, pkt->pts);
160 wav->maxpts = FFMAX(wav->maxpts, pkt->pts);
161 wav->last_duration = pkt->duration;
163 av_log(s, AV_LOG_ERROR, "wav_write_packet: NOPTS\n");
167 static int wav_write_trailer(AVFormatContext *s)
169 AVIOContext *pb = s->pb;
170 WAVContext *wav = s->priv_data;
175 if (s->pb->seekable) {
176 ff_end_tag(pb, wav->data);
178 /* update file size */
179 file_size = avio_tell(pb);
180 avio_seek(pb, 4, SEEK_SET);
181 avio_wl32(pb, (uint32_t)(file_size - 8));
182 avio_seek(pb, file_size, SEEK_SET);
186 if(s->streams[0]->codec->codec_tag != 0x01) {
187 /* Update num_samps in fact chunk */
188 int number_of_samples;
189 number_of_samples = av_rescale(wav->maxpts - wav->minpts + wav->last_duration,
190 s->streams[0]->codec->sample_rate * (int64_t)s->streams[0]->time_base.num,
191 s->streams[0]->time_base.den);
192 avio_seek(pb, wav->data-12, SEEK_SET);
193 avio_wl32(pb, number_of_samples);
194 avio_seek(pb, file_size, SEEK_SET);
201 #define OFFSET(x) offsetof(WAVContext, x)
202 #define ENC AV_OPT_FLAG_ENCODING_PARAM
203 static const AVOption options[] = {
204 { "write_bext", "Write BEXT chunk.", OFFSET(write_bext), AV_OPT_TYPE_INT, { 0 }, 0, 1, ENC },
208 static const AVClass wav_muxer_class = {
209 .class_name = "WAV muxer",
210 .item_name = av_default_item_name,
212 .version = LIBAVUTIL_VERSION_INT,
215 AVOutputFormat ff_wav_muxer = {
217 .long_name = NULL_IF_CONFIG_SMALL("WAV format"),
218 .mime_type = "audio/x-wav",
220 .priv_data_size = sizeof(WAVContext),
221 .audio_codec = CODEC_ID_PCM_S16LE,
222 .video_codec = CODEC_ID_NONE,
223 .write_header = wav_write_header,
224 .write_packet = wav_write_packet,
225 .write_trailer = wav_write_trailer,
226 .codec_tag= (const AVCodecTag* const []){ff_codec_wav_tags, 0},
227 .priv_class = &wav_muxer_class,
229 #endif /* CONFIG_WAV_MUXER */
232 #if CONFIG_WAV_DEMUXER
234 static int64_t next_tag(AVIOContext *pb, unsigned int *tag)
236 *tag = avio_rl32(pb);
237 return avio_rl32(pb);
240 /* return the size of the found tag */
241 static int64_t find_tag(AVIOContext *pb, uint32_t tag1)
249 size = next_tag(pb, &tag);
257 static int wav_probe(AVProbeData *p)
259 /* check file header */
260 if (p->buf_size <= 32)
262 if (!memcmp(p->buf + 8, "WAVE", 4)) {
263 if (!memcmp(p->buf, "RIFF", 4))
265 Since ACT demuxer has standard WAV header at top of it's own,
266 returning score is decreased to avoid probe conflict
269 return AVPROBE_SCORE_MAX - 1;
270 else if (!memcmp(p->buf, "RF64", 4) &&
271 !memcmp(p->buf + 12, "ds64", 4))
272 return AVPROBE_SCORE_MAX;
277 static int wav_parse_fmt_tag(AVFormatContext *s, int64_t size, AVStream **st)
279 AVIOContext *pb = s->pb;
282 /* parse fmt header */
283 *st = avformat_new_stream(s, NULL);
285 return AVERROR(ENOMEM);
287 ret = ff_get_wav_header(pb, (*st)->codec, size);
290 (*st)->need_parsing = AVSTREAM_PARSE_FULL;
292 av_set_pts_info(*st, 64, 1, (*st)->codec->sample_rate);
297 static inline int wav_parse_bext_string(AVFormatContext *s, const char *key,
303 av_assert0(length <= sizeof(temp));
304 if ((ret = avio_read(s->pb, temp, length)) < 0)
310 return av_dict_set(&s->metadata, key, temp, 0);
315 static int wav_parse_bext_tag(AVFormatContext *s, int64_t size)
317 char temp[131], *coding_history;
319 uint64_t time_reference;
320 int64_t umid_parts[8], umid_mask = 0;
322 if ((ret = wav_parse_bext_string(s, "description", 256)) < 0 ||
323 (ret = wav_parse_bext_string(s, "originator", 32)) < 0 ||
324 (ret = wav_parse_bext_string(s, "originator_reference", 32)) < 0 ||
325 (ret = wav_parse_bext_string(s, "origination_date", 10)) < 0 ||
326 (ret = wav_parse_bext_string(s, "origination_time", 8)) < 0)
329 time_reference = avio_rl64(s->pb);
330 snprintf(temp, sizeof(temp), "%"PRIu64, time_reference);
331 if ((ret = av_dict_set(&s->metadata, "time_reference", temp, 0)) < 0)
334 /* check if version is >= 1, in which case an UMID may be present */
335 if (avio_rl16(s->pb) >= 1) {
336 for (x = 0; x < 8; x++)
337 umid_mask |= umid_parts[x] = avio_rb64(s->pb);
340 /* the string formatting below is per SMPTE 330M-2004 Annex C */
341 if (umid_parts[4] == 0 && umid_parts[5] == 0 && umid_parts[6] == 0 && umid_parts[7] == 0) {
343 snprintf(temp, sizeof(temp), "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
344 umid_parts[0], umid_parts[1], umid_parts[2], umid_parts[3]);
347 snprintf(temp, sizeof(temp), "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64
348 "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
349 umid_parts[0], umid_parts[1], umid_parts[2], umid_parts[3],
350 umid_parts[4], umid_parts[5], umid_parts[6], umid_parts[7]);
353 if ((ret = av_dict_set(&s->metadata, "umid", temp, 0)) < 0)
357 avio_skip(s->pb, 190);
359 avio_skip(s->pb, 254);
362 /* CodingHistory present */
365 if (!(coding_history = av_malloc(size+1)))
366 return AVERROR(ENOMEM);
368 if ((ret = avio_read(s->pb, coding_history, size)) < 0)
371 coding_history[size] = 0;
372 if ((ret = av_dict_set(&s->metadata, "coding_history", coding_history,
373 AV_DICT_DONT_STRDUP_VAL)) < 0)
380 static const AVMetadataConv wav_metadata_conv[] = {
381 {"description", "comment" },
382 {"originator", "encoded_by" },
383 {"origination_date", "date" },
384 {"origination_time", "creation_time"},
389 static int wav_read_header(AVFormatContext *s,
390 AVFormatParameters *ap)
392 int64_t size, av_uninit(data_size);
393 int64_t sample_count=0;
396 AVIOContext *pb = s->pb;
398 WAVContext *wav = s->priv_data;
399 int ret, got_fmt = 0;
400 int64_t next_tag_ofs, data_ofs = -1;
402 wav->smv_data_ofs = -1;
404 /* check RIFF header */
407 rf64 = tag == MKTAG('R', 'F', '6', '4');
408 if (!rf64 && tag != MKTAG('R', 'I', 'F', 'F'))
410 avio_rl32(pb); /* file size */
412 if (tag != MKTAG('W', 'A', 'V', 'E'))
416 if (avio_rl32(pb) != MKTAG('d', 's', '6', '4'))
418 size = avio_rl32(pb);
421 avio_rl64(pb); /* RIFF size */
422 data_size = avio_rl64(pb);
423 sample_count = avio_rl64(pb);
424 if (data_size < 0 || sample_count < 0) {
425 av_log(s, AV_LOG_ERROR, "negative data_size and/or sample_count in "
426 "ds64: data_size = %"PRId64", sample_count = %"PRId64"\n",
427 data_size, sample_count);
428 return AVERROR_INVALIDDATA;
430 avio_skip(pb, size - 24); /* skip rest of ds64 chunk */
436 size = next_tag(pb, &tag);
437 next_tag_ofs = avio_tell(pb) + size;
443 case MKTAG('f', 'm', 't', ' '):
444 /* only parse the first 'fmt ' tag found */
445 if (!got_fmt && (ret = wav_parse_fmt_tag(s, size, &st)) < 0) {
448 av_log(s, AV_LOG_WARNING, "found more than one 'fmt ' tag\n");
452 case MKTAG('d', 'a', 't', 'a'):
454 av_log(s, AV_LOG_ERROR, "found no 'fmt ' tag before the 'data' tag\n");
455 return AVERROR_INVALIDDATA;
459 next_tag_ofs = wav->data_end = avio_tell(pb) + data_size;
462 next_tag_ofs = wav->data_end = size ? next_tag_ofs : INT64_MAX;
465 data_ofs = avio_tell(pb);
467 /* don't look for footer metadata if we can't seek or if we don't
468 * know where the data tag ends
470 if (!pb->seekable || (!rf64 && !size))
473 case MKTAG('f','a','c','t'):
475 sample_count = avio_rl32(pb);
477 case MKTAG('b','e','x','t'):
478 if ((ret = wav_parse_bext_tag(s, size)) < 0)
481 case MKTAG('S','M','V','0'):
482 // SMV file, a wav file with video appended.
483 if (size != MKTAG('0','2','0','0')) {
484 av_log(s, AV_LOG_ERROR, "Unknown SMV version found\n");
487 av_log(s, AV_LOG_DEBUG, "Found SMV data\n");
488 vst = avformat_new_stream(s, NULL);
490 return AVERROR(ENOMEM);
493 vst->codec->codec_type = AVMEDIA_TYPE_VIDEO;
494 vst->codec->codec_id = CODEC_ID_MJPEG;
495 vst->codec->width = avio_rl24(pb);
496 vst->codec->height = avio_rl24(pb);
497 size = avio_rl24(pb);
498 wav->smv_data_ofs = avio_tell(pb) + (size - 5) * 3;
500 wav->smv_block_size = avio_rl24(pb);
501 av_set_pts_info(vst, 32, 1, avio_rl24(pb));
502 vst->duration = avio_rl24(pb);
505 wav->smv_frames_per_jpeg = avio_rl24(pb);
509 /* seek to next tag unless we know that we'll run into EOF */
510 if ((avio_size(pb) > 0 && next_tag_ofs >= avio_size(pb)) ||
511 avio_seek(pb, next_tag_ofs, SEEK_SET) < 0) {
517 av_log(s, AV_LOG_ERROR, "no 'data' tag found\n");
518 return AVERROR_INVALIDDATA;
521 avio_seek(pb, data_ofs, SEEK_SET);
523 if (!sample_count && st->codec->channels && av_get_bits_per_sample(st->codec->codec_id))
524 sample_count = (data_size<<3) / (st->codec->channels * (uint64_t)av_get_bits_per_sample(st->codec->codec_id));
526 st->duration = sample_count;
528 ff_metadata_conv_ctx(s, NULL, wav_metadata_conv);
533 /** Find chunk with w64 GUID by skipping over other chunks
534 * @return the size of the found chunk
536 static int64_t find_guid(AVIOContext *pb, const uint8_t guid1[16])
541 while (!url_feof(pb)) {
542 avio_read(pb, guid, 16);
543 size = avio_rl64(pb);
546 if (!memcmp(guid, guid1, 16))
548 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 24);
553 static const uint8_t guid_data[16] = { 'd', 'a', 't', 'a',
554 0xF3, 0xAC, 0xD3, 0x11, 0x8C, 0xD1, 0x00, 0xC0, 0x4F, 0x8E, 0xDB, 0x8A };
556 #define MAX_SIZE 4096
558 static int wav_read_packet(AVFormatContext *s,
564 WAVContext *wav = s->priv_data;
566 if (wav->smv_data_ofs > 0) {
567 int64_t audio_dts, video_dts;
569 audio_dts = s->streams[0]->cur_dts;
570 video_dts = s->streams[1]->cur_dts;
571 if (audio_dts != AV_NOPTS_VALUE && video_dts != AV_NOPTS_VALUE) {
572 audio_dts = av_rescale_q(audio_dts, s->streams[0]->time_base, AV_TIME_BASE_Q);
573 video_dts = av_rescale_q(video_dts, s->streams[1]->time_base, AV_TIME_BASE_Q);
574 wav->smv_last_stream = video_dts >= audio_dts;
576 wav->smv_last_stream = !wav->smv_last_stream;
577 wav->smv_last_stream |= wav->audio_eof;
578 wav->smv_last_stream &= !wav->smv_eof;
579 if (wav->smv_last_stream) {
580 uint64_t old_pos = avio_tell(s->pb);
581 uint64_t new_pos = wav->smv_data_ofs +
582 wav->smv_block * wav->smv_block_size;
583 if (avio_seek(s->pb, new_pos, SEEK_SET) < 0) {
587 size = avio_rl24(s->pb);
588 ret = av_get_packet(s->pb, pkt, size);
592 pkt->pts = wav->smv_block * wav->smv_frames_per_jpeg;
594 pkt->stream_index = 1;
596 avio_seek(s->pb, old_pos, SEEK_SET);
597 if (ret == AVERROR_EOF) {
607 left = wav->data_end - avio_tell(s->pb);
609 if (CONFIG_W64_DEMUXER && wav->w64)
610 left = find_guid(s->pb, guid_data) - 24;
612 left = find_tag(s->pb, MKTAG('d', 'a', 't', 'a'));
615 if (wav->smv_data_ofs > 0 && !wav->smv_eof)
619 wav->data_end= avio_tell(s->pb) + left;
623 if (st->codec->block_align > 1) {
624 if (size < st->codec->block_align)
625 size = st->codec->block_align;
626 size = (size / st->codec->block_align) * st->codec->block_align;
628 size = FFMIN(size, left);
629 ret = av_get_packet(s->pb, pkt, size);
632 pkt->stream_index = 0;
637 static int wav_read_seek(AVFormatContext *s,
638 int stream_index, int64_t timestamp, int flags)
640 WAVContext *wav = s->priv_data;
644 if (wav->smv_data_ofs > 0) {
645 int64_t smv_timestamp = timestamp;
646 if (stream_index == 0)
647 smv_timestamp = av_rescale_q(timestamp, s->streams[0]->time_base, s->streams[1]->time_base);
649 timestamp = av_rescale_q(smv_timestamp, s->streams[1]->time_base, s->streams[0]->time_base);
650 wav->smv_block = smv_timestamp / wav->smv_frames_per_jpeg;
654 switch (st->codec->codec_id) {
659 /* use generic seeking with dynamically generated indexes */
664 return pcm_read_seek(s, stream_index, timestamp, flags);
667 AVInputFormat ff_wav_demuxer = {
669 .long_name = NULL_IF_CONFIG_SMALL("WAV format"),
670 .priv_data_size = sizeof(WAVContext),
671 .read_probe = wav_probe,
672 .read_header = wav_read_header,
673 .read_packet = wav_read_packet,
674 .read_seek = wav_read_seek,
675 .flags= AVFMT_GENERIC_INDEX,
676 .codec_tag= (const AVCodecTag* const []){ff_codec_wav_tags, 0},
678 #endif /* CONFIG_WAV_DEMUXER */
681 #if CONFIG_W64_DEMUXER
682 static const uint8_t guid_riff[16] = { 'r', 'i', 'f', 'f',
683 0x2E, 0x91, 0xCF, 0x11, 0xA5, 0xD6, 0x28, 0xDB, 0x04, 0xC1, 0x00, 0x00 };
685 static const uint8_t guid_wave[16] = { 'w', 'a', 'v', 'e',
686 0xF3, 0xAC, 0xD3, 0x11, 0x8C, 0xD1, 0x00, 0xC0, 0x4F, 0x8E, 0xDB, 0x8A };
688 static const uint8_t guid_fmt [16] = { 'f', 'm', 't', ' ',
689 0xF3, 0xAC, 0xD3, 0x11, 0x8C, 0xD1, 0x00, 0xC0, 0x4F, 0x8E, 0xDB, 0x8A };
691 static int w64_probe(AVProbeData *p)
693 if (p->buf_size <= 40)
695 if (!memcmp(p->buf, guid_riff, 16) &&
696 !memcmp(p->buf + 24, guid_wave, 16))
697 return AVPROBE_SCORE_MAX;
702 static int w64_read_header(AVFormatContext *s, AVFormatParameters *ap)
705 AVIOContext *pb = s->pb;
706 WAVContext *wav = s->priv_data;
711 avio_read(pb, guid, 16);
712 if (memcmp(guid, guid_riff, 16))
715 if (avio_rl64(pb) < 16 + 8 + 16 + 8 + 16 + 8) /* riff + wave + fmt + sizes */
718 avio_read(pb, guid, 16);
719 if (memcmp(guid, guid_wave, 16)) {
720 av_log(s, AV_LOG_ERROR, "could not find wave guid\n");
724 size = find_guid(pb, guid_fmt);
726 av_log(s, AV_LOG_ERROR, "could not find fmt guid\n");
730 st = avformat_new_stream(s, NULL);
732 return AVERROR(ENOMEM);
734 /* subtract chunk header size - normal wav file doesn't count it */
735 ret = ff_get_wav_header(pb, st->codec, size - 24);
738 avio_skip(pb, FFALIGN(size, INT64_C(8)) - size);
740 st->need_parsing = AVSTREAM_PARSE_FULL;
742 av_set_pts_info(st, 64, 1, st->codec->sample_rate);
744 size = find_guid(pb, guid_data);
746 av_log(s, AV_LOG_ERROR, "could not find data guid\n");
749 wav->data_end = avio_tell(pb) + size - 24;
755 AVInputFormat ff_w64_demuxer = {
757 .long_name = NULL_IF_CONFIG_SMALL("Sony Wave64 format"),
758 .priv_data_size = sizeof(WAVContext),
759 .read_probe = w64_probe,
760 .read_header = w64_read_header,
761 .read_packet = wav_read_packet,
762 .read_seek = wav_read_seek,
763 .flags = AVFMT_GENERIC_INDEX,
764 .codec_tag = (const AVCodecTag* const []){ff_codec_wav_tags, 0},
766 #endif /* CONFIG_W64_DEMUXER */