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"
33 #include "avio_internal.h"
50 int smv_frames_per_jpeg;
59 static inline void bwf_write_bext_string(AVFormatContext *s, const char *key, int maxlen)
61 AVDictionaryEntry *tag;
64 if (tag = av_dict_get(s->metadata, key, NULL, 0)) {
65 len = strlen(tag->value);
66 len = FFMIN(len, maxlen);
67 avio_write(s->pb, tag->value, len);
70 ffio_fill(s->pb, 0, maxlen - len);
73 static void bwf_write_bext_chunk(AVFormatContext *s)
75 AVDictionaryEntry *tmp_tag;
76 uint64_t time_reference = 0;
77 int64_t bext = ff_start_tag(s->pb, "bext");
79 bwf_write_bext_string(s, "description", 256);
80 bwf_write_bext_string(s, "originator", 32);
81 bwf_write_bext_string(s, "originator_reference", 32);
82 bwf_write_bext_string(s, "origination_date", 10);
83 bwf_write_bext_string(s, "origination_time", 8);
85 if (tmp_tag = av_dict_get(s->metadata, "time_reference", NULL, 0))
86 time_reference = strtoll(tmp_tag->value, NULL, 10);
87 avio_wl64(s->pb, time_reference);
88 avio_wl16(s->pb, 1); // set version to 1
90 if (tmp_tag = av_dict_get(s->metadata, "umid", NULL, 0)) {
91 unsigned char umidpart_str[17] = {0};
94 int len = strlen(tmp_tag->value+2);
96 for (i = 0; i < len/16; i++) {
97 memcpy(umidpart_str, tmp_tag->value + 2 + (i*16), 16);
98 umidpart = strtoll(umidpart_str, NULL, 16);
99 avio_wb64(s->pb, umidpart);
101 ffio_fill(s->pb, 0, 64 - i*8);
103 ffio_fill(s->pb, 0, 64); // zero UMID
105 ffio_fill(s->pb, 0, 190); // Reserved
107 if (tmp_tag = av_dict_get(s->metadata, "coding_history", NULL, 0))
108 avio_put_str(s->pb, tmp_tag->value);
110 ff_end_tag(s->pb, bext);
113 static int wav_write_header(AVFormatContext *s)
115 WAVContext *wav = s->priv_data;
116 AVIOContext *pb = s->pb;
119 ffio_wfourcc(pb, "RIFF");
120 avio_wl32(pb, 0); /* file length */
121 ffio_wfourcc(pb, "WAVE");
124 fmt = ff_start_tag(pb, "fmt ");
125 if (ff_put_wav_header(pb, s->streams[0]->codec) < 0) {
126 av_log(s, AV_LOG_ERROR, "%s codec not supported in WAVE format\n",
127 s->streams[0]->codec->codec ? s->streams[0]->codec->codec->name : "NONE");
132 if (s->streams[0]->codec->codec_tag != 0x01 /* hence for all other than PCM */
133 && s->pb->seekable) {
134 fact = ff_start_tag(pb, "fact");
136 ff_end_tag(pb, fact);
140 bwf_write_bext_chunk(s);
142 avpriv_set_pts_info(s->streams[0], 64, 1, s->streams[0]->codec->sample_rate);
143 wav->maxpts = wav->last_duration = 0;
144 wav->minpts = INT64_MAX;
147 wav->data = ff_start_tag(pb, "data");
154 static int wav_write_packet(AVFormatContext *s, AVPacket *pkt)
156 AVIOContext *pb = s->pb;
157 WAVContext *wav = s->priv_data;
158 avio_write(pb, pkt->data, pkt->size);
159 if(pkt->pts != AV_NOPTS_VALUE) {
160 wav->minpts = FFMIN(wav->minpts, pkt->pts);
161 wav->maxpts = FFMAX(wav->maxpts, pkt->pts);
162 wav->last_duration = pkt->duration;
164 av_log(s, AV_LOG_ERROR, "wav_write_packet: NOPTS\n");
168 static int wav_write_trailer(AVFormatContext *s)
170 AVIOContext *pb = s->pb;
171 WAVContext *wav = s->priv_data;
176 if (s->pb->seekable) {
177 ff_end_tag(pb, wav->data);
179 /* update file size */
180 file_size = avio_tell(pb);
181 avio_seek(pb, 4, SEEK_SET);
182 avio_wl32(pb, (uint32_t)(file_size - 8));
183 avio_seek(pb, file_size, SEEK_SET);
187 if(s->streams[0]->codec->codec_tag != 0x01) {
188 /* Update num_samps in fact chunk */
189 int number_of_samples;
190 number_of_samples = av_rescale(wav->maxpts - wav->minpts + wav->last_duration,
191 s->streams[0]->codec->sample_rate * (int64_t)s->streams[0]->time_base.num,
192 s->streams[0]->time_base.den);
193 avio_seek(pb, wav->data-12, SEEK_SET);
194 avio_wl32(pb, number_of_samples);
195 avio_seek(pb, file_size, SEEK_SET);
202 #define OFFSET(x) offsetof(WAVContext, x)
203 #define ENC AV_OPT_FLAG_ENCODING_PARAM
204 static const AVOption options[] = {
205 { "write_bext", "Write BEXT chunk.", OFFSET(write_bext), AV_OPT_TYPE_INT, { 0 }, 0, 1, ENC },
209 static const AVClass wav_muxer_class = {
210 .class_name = "WAV muxer",
211 .item_name = av_default_item_name,
213 .version = LIBAVUTIL_VERSION_INT,
216 AVOutputFormat ff_wav_muxer = {
218 .long_name = NULL_IF_CONFIG_SMALL("WAV format"),
219 .mime_type = "audio/x-wav",
221 .priv_data_size = sizeof(WAVContext),
222 .audio_codec = CODEC_ID_PCM_S16LE,
223 .video_codec = CODEC_ID_NONE,
224 .write_header = wav_write_header,
225 .write_packet = wav_write_packet,
226 .write_trailer = wav_write_trailer,
227 .flags = AVFMT_TS_NONSTRICT,
228 .codec_tag = (const AVCodecTag* const []){ ff_codec_wav_tags, 0 },
229 .priv_class = &wav_muxer_class,
231 #endif /* CONFIG_WAV_MUXER */
234 #if CONFIG_WAV_DEMUXER
236 static int64_t next_tag(AVIOContext *pb, uint32_t *tag)
238 *tag = avio_rl32(pb);
239 return avio_rl32(pb);
242 /* return the size of the found tag */
243 static int64_t find_tag(AVIOContext *pb, uint32_t tag1)
251 size = next_tag(pb, &tag);
259 static int wav_probe(AVProbeData *p)
261 /* check file header */
262 if (p->buf_size <= 32)
264 if (!memcmp(p->buf + 8, "WAVE", 4)) {
265 if (!memcmp(p->buf, "RIFF", 4))
267 Since ACT demuxer has standard WAV header at top of it's own,
268 returning score is decreased to avoid probe conflict
271 return AVPROBE_SCORE_MAX - 1;
272 else if (!memcmp(p->buf, "RF64", 4) &&
273 !memcmp(p->buf + 12, "ds64", 4))
274 return AVPROBE_SCORE_MAX;
279 static int wav_parse_fmt_tag(AVFormatContext *s, int64_t size, AVStream **st)
281 AVIOContext *pb = s->pb;
284 /* parse fmt header */
285 *st = avformat_new_stream(s, NULL);
287 return AVERROR(ENOMEM);
289 ret = ff_get_wav_header(pb, (*st)->codec, size);
292 (*st)->need_parsing = AVSTREAM_PARSE_FULL;
294 avpriv_set_pts_info(*st, 64, 1, (*st)->codec->sample_rate);
299 static inline int wav_parse_bext_string(AVFormatContext *s, const char *key,
305 av_assert0(length <= sizeof(temp));
306 if ((ret = avio_read(s->pb, temp, length)) < 0)
312 return av_dict_set(&s->metadata, key, temp, 0);
317 static int wav_parse_bext_tag(AVFormatContext *s, int64_t size)
319 char temp[131], *coding_history;
321 uint64_t time_reference;
322 int64_t umid_parts[8], umid_mask = 0;
324 if ((ret = wav_parse_bext_string(s, "description", 256)) < 0 ||
325 (ret = wav_parse_bext_string(s, "originator", 32)) < 0 ||
326 (ret = wav_parse_bext_string(s, "originator_reference", 32)) < 0 ||
327 (ret = wav_parse_bext_string(s, "origination_date", 10)) < 0 ||
328 (ret = wav_parse_bext_string(s, "origination_time", 8)) < 0)
331 time_reference = avio_rl64(s->pb);
332 snprintf(temp, sizeof(temp), "%"PRIu64, time_reference);
333 if ((ret = av_dict_set(&s->metadata, "time_reference", temp, 0)) < 0)
336 /* check if version is >= 1, in which case an UMID may be present */
337 if (avio_rl16(s->pb) >= 1) {
338 for (x = 0; x < 8; x++)
339 umid_mask |= umid_parts[x] = avio_rb64(s->pb);
342 /* the string formatting below is per SMPTE 330M-2004 Annex C */
343 if (umid_parts[4] == 0 && umid_parts[5] == 0 && umid_parts[6] == 0 && umid_parts[7] == 0) {
345 snprintf(temp, sizeof(temp), "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
346 umid_parts[0], umid_parts[1], umid_parts[2], umid_parts[3]);
349 snprintf(temp, sizeof(temp), "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64
350 "%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
351 umid_parts[0], umid_parts[1], umid_parts[2], umid_parts[3],
352 umid_parts[4], umid_parts[5], umid_parts[6], umid_parts[7]);
355 if ((ret = av_dict_set(&s->metadata, "umid", temp, 0)) < 0)
359 avio_skip(s->pb, 190);
361 avio_skip(s->pb, 254);
364 /* CodingHistory present */
367 if (!(coding_history = av_malloc(size+1)))
368 return AVERROR(ENOMEM);
370 if ((ret = avio_read(s->pb, coding_history, size)) < 0)
373 coding_history[size] = 0;
374 if ((ret = av_dict_set(&s->metadata, "coding_history", coding_history,
375 AV_DICT_DONT_STRDUP_VAL)) < 0)
382 static const AVMetadataConv wav_metadata_conv[] = {
383 {"description", "comment" },
384 {"originator", "encoded_by" },
385 {"origination_date", "date" },
386 {"origination_time", "creation_time"},
391 static int wav_read_header(AVFormatContext *s)
393 int64_t size, av_uninit(data_size);
394 int64_t sample_count=0;
396 uint32_t tag, list_type;
397 AVIOContext *pb = s->pb;
399 WAVContext *wav = s->priv_data;
400 int ret, got_fmt = 0;
401 int64_t next_tag_ofs, data_ofs = -1;
403 wav->smv_data_ofs = -1;
405 /* check RIFF header */
408 rf64 = tag == MKTAG('R', 'F', '6', '4');
409 if (!rf64 && tag != MKTAG('R', 'I', 'F', 'F'))
411 avio_rl32(pb); /* file size */
413 if (tag != MKTAG('W', 'A', 'V', 'E'))
417 if (avio_rl32(pb) != MKTAG('d', 's', '6', '4'))
419 size = avio_rl32(pb);
422 avio_rl64(pb); /* RIFF size */
423 data_size = avio_rl64(pb);
424 sample_count = avio_rl64(pb);
425 if (data_size < 0 || sample_count < 0) {
426 av_log(s, AV_LOG_ERROR, "negative data_size and/or sample_count in "
427 "ds64: data_size = %"PRId64", sample_count = %"PRId64"\n",
428 data_size, sample_count);
429 return AVERROR_INVALIDDATA;
431 avio_skip(pb, size - 24); /* skip rest of ds64 chunk */
437 size = next_tag(pb, &tag);
438 next_tag_ofs = avio_tell(pb) + size;
444 case MKTAG('f', 'm', 't', ' '):
445 /* only parse the first 'fmt ' tag found */
446 if (!got_fmt && (ret = wav_parse_fmt_tag(s, size, &st)) < 0) {
449 av_log(s, AV_LOG_WARNING, "found more than one 'fmt ' tag\n");
453 case MKTAG('d', 'a', 't', 'a'):
455 av_log(s, AV_LOG_ERROR, "found no 'fmt ' tag before the 'data' tag\n");
456 return AVERROR_INVALIDDATA;
460 next_tag_ofs = wav->data_end = avio_tell(pb) + data_size;
463 next_tag_ofs = wav->data_end = size ? next_tag_ofs : INT64_MAX;
466 data_ofs = avio_tell(pb);
468 /* don't look for footer metadata if we can't seek or if we don't
469 * know where the data tag ends
471 if (!pb->seekable || (!rf64 && !size))
474 case MKTAG('f','a','c','t'):
476 sample_count = avio_rl32(pb);
478 case MKTAG('b','e','x','t'):
479 if ((ret = wav_parse_bext_tag(s, size)) < 0)
482 case MKTAG('S','M','V','0'):
483 // SMV file, a wav file with video appended.
484 if (size != MKTAG('0','2','0','0')) {
485 av_log(s, AV_LOG_ERROR, "Unknown SMV version found\n");
488 av_log(s, AV_LOG_DEBUG, "Found SMV data\n");
489 vst = avformat_new_stream(s, NULL);
491 return AVERROR(ENOMEM);
494 vst->codec->codec_type = AVMEDIA_TYPE_VIDEO;
495 vst->codec->codec_id = CODEC_ID_MJPEG;
496 vst->codec->width = avio_rl24(pb);
497 vst->codec->height = avio_rl24(pb);
498 size = avio_rl24(pb);
499 wav->smv_data_ofs = avio_tell(pb) + (size - 5) * 3;
501 wav->smv_block_size = avio_rl24(pb);
502 avpriv_set_pts_info(vst, 32, 1, avio_rl24(pb));
503 vst->duration = avio_rl24(pb);
506 wav->smv_frames_per_jpeg = avio_rl24(pb);
508 case MKTAG('L', 'I', 'S', 'T'):
509 list_type = avio_rl32(pb);
511 av_log(s, AV_LOG_ERROR, "too short LIST tag\n");
512 return AVERROR_INVALIDDATA;
515 case MKTAG('I', 'N', 'F', 'O'):
516 if ((ret = ff_read_riff_info(s, size - 4)) < 0)
522 /* seek to next tag unless we know that we'll run into EOF */
523 if ((avio_size(pb) > 0 && next_tag_ofs >= avio_size(pb)) ||
524 avio_seek(pb, next_tag_ofs, SEEK_SET) < 0) {
530 av_log(s, AV_LOG_ERROR, "no 'data' tag found\n");
531 return AVERROR_INVALIDDATA;
534 avio_seek(pb, data_ofs, SEEK_SET);
536 if (!sample_count && st->codec->channels && av_get_bits_per_sample(st->codec->codec_id))
537 sample_count = (data_size<<3) / (st->codec->channels * (uint64_t)av_get_bits_per_sample(st->codec->codec_id));
539 st->duration = sample_count;
541 ff_metadata_conv_ctx(s, NULL, wav_metadata_conv);
542 ff_metadata_conv_ctx(s, NULL, ff_riff_info_conv);
547 /** Find chunk with w64 GUID by skipping over other chunks
548 * @return the size of the found chunk
550 static int64_t find_guid(AVIOContext *pb, const uint8_t guid1[16])
555 while (!url_feof(pb)) {
556 avio_read(pb, guid, 16);
557 size = avio_rl64(pb);
560 if (!memcmp(guid, guid1, 16))
562 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 24);
567 static const uint8_t guid_data[16] = { 'd', 'a', 't', 'a',
568 0xF3, 0xAC, 0xD3, 0x11, 0x8C, 0xD1, 0x00, 0xC0, 0x4F, 0x8E, 0xDB, 0x8A };
570 #define MAX_SIZE 4096
572 static int wav_read_packet(AVFormatContext *s,
578 WAVContext *wav = s->priv_data;
580 if (wav->smv_data_ofs > 0) {
581 int64_t audio_dts, video_dts;
583 audio_dts = s->streams[0]->cur_dts;
584 video_dts = s->streams[1]->cur_dts;
585 if (audio_dts != AV_NOPTS_VALUE && video_dts != AV_NOPTS_VALUE) {
586 audio_dts = av_rescale_q(audio_dts, s->streams[0]->time_base, AV_TIME_BASE_Q);
587 video_dts = av_rescale_q(video_dts, s->streams[1]->time_base, AV_TIME_BASE_Q);
588 wav->smv_last_stream = video_dts >= audio_dts;
590 wav->smv_last_stream = !wav->smv_last_stream;
591 wav->smv_last_stream |= wav->audio_eof;
592 wav->smv_last_stream &= !wav->smv_eof;
593 if (wav->smv_last_stream) {
594 uint64_t old_pos = avio_tell(s->pb);
595 uint64_t new_pos = wav->smv_data_ofs +
596 wav->smv_block * wav->smv_block_size;
597 if (avio_seek(s->pb, new_pos, SEEK_SET) < 0) {
601 size = avio_rl24(s->pb);
602 ret = av_get_packet(s->pb, pkt, size);
606 pkt->pts = wav->smv_block * wav->smv_frames_per_jpeg;
608 pkt->stream_index = 1;
610 avio_seek(s->pb, old_pos, SEEK_SET);
611 if (ret == AVERROR_EOF) {
621 left = wav->data_end - avio_tell(s->pb);
622 if (wav->ignore_length)
625 if (CONFIG_W64_DEMUXER && wav->w64)
626 left = find_guid(s->pb, guid_data) - 24;
628 left = find_tag(s->pb, MKTAG('d', 'a', 't', 'a'));
631 if (wav->smv_data_ofs > 0 && !wav->smv_eof)
635 wav->data_end= avio_tell(s->pb) + left;
639 if (st->codec->block_align > 1) {
640 if (size < st->codec->block_align)
641 size = st->codec->block_align;
642 size = (size / st->codec->block_align) * st->codec->block_align;
644 size = FFMIN(size, left);
645 ret = av_get_packet(s->pb, pkt, size);
648 pkt->stream_index = 0;
653 static int wav_read_seek(AVFormatContext *s,
654 int stream_index, int64_t timestamp, int flags)
656 WAVContext *wav = s->priv_data;
660 if (wav->smv_data_ofs > 0) {
661 int64_t smv_timestamp = timestamp;
662 if (stream_index == 0)
663 smv_timestamp = av_rescale_q(timestamp, s->streams[0]->time_base, s->streams[1]->time_base);
665 timestamp = av_rescale_q(smv_timestamp, s->streams[1]->time_base, s->streams[0]->time_base);
666 wav->smv_block = smv_timestamp / wav->smv_frames_per_jpeg;
670 switch (st->codec->codec_id) {
675 /* use generic seeking with dynamically generated indexes */
680 return ff_pcm_read_seek(s, stream_index, timestamp, flags);
683 #define OFFSET(x) offsetof(WAVContext, x)
684 #define DEC AV_OPT_FLAG_DECODING_PARAM
685 static const AVOption demux_options[] = {
686 { "ignore_length", "Ignore length", OFFSET(ignore_length), AV_OPT_TYPE_INT, { 0 }, 0, 1, DEC },
690 static const AVClass wav_demuxer_class = {
691 .class_name = "WAV demuxer",
692 .item_name = av_default_item_name,
693 .option = demux_options,
694 .version = LIBAVUTIL_VERSION_INT,
696 AVInputFormat ff_wav_demuxer = {
698 .long_name = NULL_IF_CONFIG_SMALL("WAV format"),
699 .priv_data_size = sizeof(WAVContext),
700 .read_probe = wav_probe,
701 .read_header = wav_read_header,
702 .read_packet = wav_read_packet,
703 .read_seek = wav_read_seek,
704 .flags = AVFMT_GENERIC_INDEX,
705 .codec_tag = (const AVCodecTag* const []){ ff_codec_wav_tags, 0 },
706 .priv_class = &wav_demuxer_class,
708 #endif /* CONFIG_WAV_DEMUXER */
711 #if CONFIG_W64_DEMUXER
712 static const uint8_t guid_riff[16] = { 'r', 'i', 'f', 'f',
713 0x2E, 0x91, 0xCF, 0x11, 0xA5, 0xD6, 0x28, 0xDB, 0x04, 0xC1, 0x00, 0x00 };
715 static const uint8_t guid_wave[16] = { 'w', 'a', 'v', 'e',
716 0xF3, 0xAC, 0xD3, 0x11, 0x8C, 0xD1, 0x00, 0xC0, 0x4F, 0x8E, 0xDB, 0x8A };
718 static const uint8_t guid_fmt [16] = { 'f', 'm', 't', ' ',
719 0xF3, 0xAC, 0xD3, 0x11, 0x8C, 0xD1, 0x00, 0xC0, 0x4F, 0x8E, 0xDB, 0x8A };
721 static int w64_probe(AVProbeData *p)
723 if (p->buf_size <= 40)
725 if (!memcmp(p->buf, guid_riff, 16) &&
726 !memcmp(p->buf + 24, guid_wave, 16))
727 return AVPROBE_SCORE_MAX;
732 static int w64_read_header(AVFormatContext *s)
735 AVIOContext *pb = s->pb;
736 WAVContext *wav = s->priv_data;
741 avio_read(pb, guid, 16);
742 if (memcmp(guid, guid_riff, 16))
745 if (avio_rl64(pb) < 16 + 8 + 16 + 8 + 16 + 8) /* riff + wave + fmt + sizes */
748 avio_read(pb, guid, 16);
749 if (memcmp(guid, guid_wave, 16)) {
750 av_log(s, AV_LOG_ERROR, "could not find wave guid\n");
754 size = find_guid(pb, guid_fmt);
756 av_log(s, AV_LOG_ERROR, "could not find fmt guid\n");
760 st = avformat_new_stream(s, NULL);
762 return AVERROR(ENOMEM);
764 /* subtract chunk header size - normal wav file doesn't count it */
765 ret = ff_get_wav_header(pb, st->codec, size - 24);
768 avio_skip(pb, FFALIGN(size, INT64_C(8)) - size);
770 st->need_parsing = AVSTREAM_PARSE_FULL;
772 avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate);
774 size = find_guid(pb, guid_data);
776 av_log(s, AV_LOG_ERROR, "could not find data guid\n");
779 wav->data_end = avio_tell(pb) + size - 24;
785 AVInputFormat ff_w64_demuxer = {
787 .long_name = NULL_IF_CONFIG_SMALL("Sony Wave64 format"),
788 .priv_data_size = sizeof(WAVContext),
789 .read_probe = w64_probe,
790 .read_header = w64_read_header,
791 .read_packet = wav_read_packet,
792 .read_seek = wav_read_seek,
793 .flags = AVFMT_GENERIC_INDEX,
794 .codec_tag = (const AVCodecTag* const []){ ff_codec_wav_tags, 0 },
796 #endif /* CONFIG_W64_DEMUXER */