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
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"
36 #include "avio_internal.h"
44 typedef struct WAVDemuxContext {
50 int smv_frames_per_jpeg;
59 int unaligned; // e.g. if an odd number of bytes ID3 tag was prepended
60 int rifx; // RIFX: integer byte order for parameters is big endian
63 #if CONFIG_WAV_DEMUXER
65 static int64_t next_tag(AVIOContext *pb, uint32_t *tag, int big_endian)
75 /* RIFF chunks are always at even offsets relative to where they start. */
76 static int64_t wav_seek_tag(WAVDemuxContext * wav, AVIOContext *s, int64_t offset, int whence)
78 offset += offset < INT64_MAX && offset + wav->unaligned & 1;
80 return avio_seek(s, offset, whence);
83 /* return the size of the found tag */
84 static int64_t find_tag(WAVDemuxContext * wav, AVIOContext *pb, uint32_t tag1)
92 size = next_tag(pb, &tag, wav->rifx);
95 wav_seek_tag(wav, pb, size, SEEK_CUR);
100 static int wav_probe(AVProbeData *p)
102 /* check file header */
103 if (p->buf_size <= 32)
105 if (!memcmp(p->buf + 8, "WAVE", 4)) {
106 if (!memcmp(p->buf, "RIFF", 4) || !memcmp(p->buf, "RIFX", 4))
107 /* Since the ACT demuxer has a standard WAV header at the top of
108 * its own, the returned score is decreased to avoid a probe
109 * conflict between ACT and WAV. */
110 return AVPROBE_SCORE_MAX - 1;
111 else if (!memcmp(p->buf, "RF64", 4) &&
112 !memcmp(p->buf + 12, "ds64", 4))
113 return AVPROBE_SCORE_MAX;
118 static void handle_stream_probing(AVStream *st)
120 if (st->codec->codec_id == AV_CODEC_ID_PCM_S16LE) {
121 st->request_probe = AVPROBE_SCORE_EXTENSION;
122 st->probe_packets = FFMIN(st->probe_packets, 32);
126 static int wav_parse_fmt_tag(AVFormatContext *s, int64_t size, AVStream **st)
128 AVIOContext *pb = s->pb;
129 WAVDemuxContext *wav = s->priv_data;
132 /* parse fmt header */
133 *st = avformat_new_stream(s, NULL);
135 return AVERROR(ENOMEM);
137 ret = ff_get_wav_header(s, pb, (*st)->codec, size, wav->rifx);
140 handle_stream_probing(*st);
142 (*st)->need_parsing = AVSTREAM_PARSE_FULL_RAW;
144 avpriv_set_pts_info(*st, 64, 1, (*st)->codec->sample_rate);
149 static inline int wav_parse_bext_string(AVFormatContext *s, const char *key,
155 av_assert0(length <= sizeof(temp));
156 if ((ret = avio_read(s->pb, temp, length)) < 0)
162 return av_dict_set(&s->metadata, key, temp, 0);
167 static int wav_parse_bext_tag(AVFormatContext *s, int64_t size)
169 char temp[131], *coding_history;
171 uint64_t time_reference;
172 int64_t umid_parts[8], umid_mask = 0;
174 if ((ret = wav_parse_bext_string(s, "description", 256)) < 0 ||
175 (ret = wav_parse_bext_string(s, "originator", 32)) < 0 ||
176 (ret = wav_parse_bext_string(s, "originator_reference", 32)) < 0 ||
177 (ret = wav_parse_bext_string(s, "origination_date", 10)) < 0 ||
178 (ret = wav_parse_bext_string(s, "origination_time", 8)) < 0)
181 time_reference = avio_rl64(s->pb);
182 snprintf(temp, sizeof(temp), "%"PRIu64, time_reference);
183 if ((ret = av_dict_set(&s->metadata, "time_reference", temp, 0)) < 0)
186 /* check if version is >= 1, in which case an UMID may be present */
187 if (avio_rl16(s->pb) >= 1) {
188 for (x = 0; x < 8; x++)
189 umid_mask |= umid_parts[x] = avio_rb64(s->pb);
192 /* the string formatting below is per SMPTE 330M-2004 Annex C */
193 if (umid_parts[4] == 0 && umid_parts[5] == 0 &&
194 umid_parts[6] == 0 && umid_parts[7] == 0) {
196 snprintf(temp, sizeof(temp),
197 "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
198 umid_parts[0], umid_parts[1],
199 umid_parts[2], umid_parts[3]);
202 snprintf(temp, sizeof(temp),
203 "0x%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64
204 "%016"PRIX64"%016"PRIX64"%016"PRIX64"%016"PRIX64,
205 umid_parts[0], umid_parts[1],
206 umid_parts[2], umid_parts[3],
207 umid_parts[4], umid_parts[5],
208 umid_parts[6], umid_parts[7]);
211 if ((ret = av_dict_set(&s->metadata, "umid", temp, 0)) < 0)
215 avio_skip(s->pb, 190);
217 avio_skip(s->pb, 254);
220 /* CodingHistory present */
223 if (!(coding_history = av_malloc(size + 1)))
224 return AVERROR(ENOMEM);
226 if ((ret = avio_read(s->pb, coding_history, size)) < 0)
229 coding_history[size] = 0;
230 if ((ret = av_dict_set(&s->metadata, "coding_history", coding_history,
231 AV_DICT_DONT_STRDUP_VAL)) < 0)
238 static const AVMetadataConv wav_metadata_conv[] = {
239 { "description", "comment" },
240 { "originator", "encoded_by" },
241 { "origination_date", "date" },
242 { "origination_time", "creation_time" },
247 static int wav_read_header(AVFormatContext *s)
249 int64_t size, av_uninit(data_size);
250 int64_t sample_count = 0;
254 AVIOContext *pb = s->pb;
256 WAVDemuxContext *wav = s->priv_data;
257 int ret, got_fmt = 0;
258 int64_t next_tag_ofs, data_ofs = -1;
260 wav->unaligned = avio_tell(s->pb) & 1;
262 wav->smv_data_ofs = -1;
267 case MKTAG('R', 'I', 'F', 'F'):
269 case MKTAG('R', 'I', 'F', 'X'):
272 case MKTAG('R', 'F', '6', '4'):
276 av_get_codec_tag_string(start_code, sizeof(start_code), tag);
277 av_log(s, AV_LOG_ERROR, "invalid start code %s in RIFF header\n", start_code);
278 return AVERROR_INVALIDDATA;
281 /* read chunk size */
285 if (avio_rl32(pb) != MKTAG('W', 'A', 'V', 'E')) {
286 av_log(s, AV_LOG_ERROR, "invalid format in RIFF header\n");
287 return AVERROR_INVALIDDATA;
291 if (avio_rl32(pb) != MKTAG('d', 's', '6', '4'))
292 return AVERROR_INVALIDDATA;
293 size = avio_rl32(pb);
295 return AVERROR_INVALIDDATA;
296 avio_rl64(pb); /* RIFF size */
298 data_size = avio_rl64(pb);
299 sample_count = avio_rl64(pb);
301 if (data_size < 0 || sample_count < 0) {
302 av_log(s, AV_LOG_ERROR, "negative data_size and/or sample_count in "
303 "ds64: data_size = %"PRId64", sample_count = %"PRId64"\n",
304 data_size, sample_count);
305 return AVERROR_INVALIDDATA;
307 avio_skip(pb, size - 24); /* skip rest of ds64 chunk */
313 size = next_tag(pb, &tag, wav->rifx);
314 next_tag_ofs = avio_tell(pb) + size;
320 case MKTAG('f', 'm', 't', ' '):
321 /* only parse the first 'fmt ' tag found */
322 if (!got_fmt && (ret = wav_parse_fmt_tag(s, size, &st)) < 0) {
325 av_log(s, AV_LOG_WARNING, "found more than one 'fmt ' tag\n");
329 case MKTAG('d', 'a', 't', 'a'):
331 av_log(s, AV_LOG_ERROR,
332 "found no 'fmt ' tag before the 'data' tag\n");
333 return AVERROR_INVALIDDATA;
337 next_tag_ofs = wav->data_end = avio_tell(pb) + data_size;
338 } else if (size != 0xFFFFFFFF) {
340 next_tag_ofs = wav->data_end = size ? next_tag_ofs : INT64_MAX;
342 av_log(s, AV_LOG_WARNING, "Ignoring maximum wav data size, "
343 "file may be invalid\n");
345 next_tag_ofs = wav->data_end = INT64_MAX;
348 data_ofs = avio_tell(pb);
350 /* don't look for footer metadata if we can't seek or if we don't
351 * know where the data tag ends
353 if (!pb->seekable || (!rf64 && !size))
356 case MKTAG('f', 'a', 'c', 't'):
358 sample_count = (!wav->rifx ? avio_rl32(pb) : avio_rb32(pb));
360 case MKTAG('b', 'e', 'x', 't'):
361 if ((ret = wav_parse_bext_tag(s, size)) < 0)
364 case MKTAG('S','M','V','0'):
366 av_log(s, AV_LOG_ERROR, "found no 'fmt ' tag before the 'SMV0' tag\n");
367 return AVERROR_INVALIDDATA;
369 // SMV file, a wav file with video appended.
370 if (size != MKTAG('0','2','0','0')) {
371 av_log(s, AV_LOG_ERROR, "Unknown SMV version found\n");
374 av_log(s, AV_LOG_DEBUG, "Found SMV data\n");
375 wav->smv_given_first = 0;
376 vst = avformat_new_stream(s, NULL);
378 return AVERROR(ENOMEM);
381 vst->codec->codec_type = AVMEDIA_TYPE_VIDEO;
382 vst->codec->codec_id = AV_CODEC_ID_SMVJPEG;
383 vst->codec->width = avio_rl24(pb);
384 vst->codec->height = avio_rl24(pb);
385 if (ff_alloc_extradata(vst->codec, 4)) {
386 av_log(s, AV_LOG_ERROR, "Could not allocate extradata.\n");
387 return AVERROR(ENOMEM);
389 size = avio_rl24(pb);
390 wav->smv_data_ofs = avio_tell(pb) + (size - 5) * 3;
392 wav->smv_block_size = avio_rl24(pb);
393 avpriv_set_pts_info(vst, 32, 1, avio_rl24(pb));
394 vst->duration = avio_rl24(pb);
397 wav->smv_frames_per_jpeg = avio_rl24(pb);
398 if (wav->smv_frames_per_jpeg > 65536) {
399 av_log(s, AV_LOG_ERROR, "too many frames per jpeg\n");
400 return AVERROR_INVALIDDATA;
402 AV_WL32(vst->codec->extradata, wav->smv_frames_per_jpeg);
405 case MKTAG('L', 'I', 'S', 'T'):
407 av_log(s, AV_LOG_ERROR, "too short LIST tag\n");
408 return AVERROR_INVALIDDATA;
410 switch (avio_rl32(pb)) {
411 case MKTAG('I', 'N', 'F', 'O'):
412 ff_read_riff_info(s, size - 4);
417 /* seek to next tag unless we know that we'll run into EOF */
418 if ((avio_size(pb) > 0 && next_tag_ofs >= avio_size(pb)) ||
419 wav_seek_tag(wav, pb, next_tag_ofs, SEEK_SET) < 0) {
426 av_log(s, AV_LOG_ERROR, "no 'data' tag found\n");
427 return AVERROR_INVALIDDATA;
430 avio_seek(pb, data_ofs, SEEK_SET);
432 if (data_size > (INT64_MAX>>3)) {
433 av_log(s, AV_LOG_WARNING, "Data size %"PRId64" is too large\n", data_size);
437 if ( data_size > 0 && sample_count && st->codec->channels
438 && (data_size << 3) / sample_count / st->codec->channels > st->codec->bits_per_coded_sample) {
439 av_log(s, AV_LOG_WARNING, "ignoring wrong sample_count %"PRId64"\n", sample_count);
443 /* G.729 hack (for Ticket4577)
444 * FIXME: Come up with cleaner, more general solution */
445 if (st->codec->codec_id == AV_CODEC_ID_G729 && sample_count && (data_size << 3) > sample_count) {
446 av_log(s, AV_LOG_WARNING, "ignoring wrong sample_count %"PRId64"\n", sample_count);
450 if (!sample_count || av_get_exact_bits_per_sample(st->codec->codec_id) > 0)
451 if ( st->codec->channels
453 && av_get_bits_per_sample(st->codec->codec_id)
454 && wav->data_end <= avio_size(pb))
455 sample_count = (data_size << 3)
457 (st->codec->channels * (uint64_t)av_get_bits_per_sample(st->codec->codec_id));
460 st->duration = sample_count;
462 ff_metadata_conv_ctx(s, NULL, wav_metadata_conv);
463 ff_metadata_conv_ctx(s, NULL, ff_riff_info_conv);
469 * Find chunk with w64 GUID by skipping over other chunks.
470 * @return the size of the found chunk
472 static int64_t find_guid(AVIOContext *pb, const uint8_t guid1[16])
477 while (!avio_feof(pb)) {
478 avio_read(pb, guid, 16);
479 size = avio_rl64(pb);
481 return AVERROR_INVALIDDATA;
482 if (!memcmp(guid, guid1, 16))
484 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 24);
489 #define MAX_SIZE 4096
491 static int wav_read_packet(AVFormatContext *s, AVPacket *pkt)
496 WAVDemuxContext *wav = s->priv_data;
498 if (CONFIG_SPDIF_DEMUXER && wav->spdif == 0 &&
499 s->streams[0]->codec->codec_tag == 1) {
500 enum AVCodecID codec;
501 ret = ff_spdif_probe(s->pb->buffer, s->pb->buf_end - s->pb->buffer,
503 if (ret > AVPROBE_SCORE_EXTENSION) {
504 s->streams[0]->codec->codec_id = codec;
510 if (CONFIG_SPDIF_DEMUXER && wav->spdif == 1)
511 return ff_spdif_read_packet(s, pkt);
513 if (wav->smv_data_ofs > 0) {
514 int64_t audio_dts, video_dts;
516 audio_dts = (int32_t)s->streams[0]->cur_dts;
517 video_dts = (int32_t)s->streams[1]->cur_dts;
519 if (audio_dts != AV_NOPTS_VALUE && video_dts != AV_NOPTS_VALUE) {
520 /*We always return a video frame first to get the pixel format first*/
521 wav->smv_last_stream = wav->smv_given_first ?
522 av_compare_ts(video_dts, s->streams[1]->time_base,
523 audio_dts, s->streams[0]->time_base) > 0 : 0;
524 wav->smv_given_first = 1;
526 wav->smv_last_stream = !wav->smv_last_stream;
527 wav->smv_last_stream |= wav->audio_eof;
528 wav->smv_last_stream &= !wav->smv_eof;
529 if (wav->smv_last_stream) {
530 uint64_t old_pos = avio_tell(s->pb);
531 uint64_t new_pos = wav->smv_data_ofs +
532 wav->smv_block * wav->smv_block_size;
533 if (avio_seek(s->pb, new_pos, SEEK_SET) < 0) {
537 size = avio_rl24(s->pb);
538 ret = av_get_packet(s->pb, pkt, size);
542 pkt->pts = wav->smv_block * wav->smv_frames_per_jpeg + wav->smv_cur_pt;
544 if (wav->smv_frames_per_jpeg > 0)
545 wav->smv_cur_pt %= wav->smv_frames_per_jpeg;
546 if (!wav->smv_cur_pt)
549 pkt->stream_index = 1;
551 avio_seek(s->pb, old_pos, SEEK_SET);
552 if (ret == AVERROR_EOF) {
562 left = wav->data_end - avio_tell(s->pb);
563 if (wav->ignore_length)
566 if (CONFIG_W64_DEMUXER && wav->w64)
567 left = find_guid(s->pb, ff_w64_guid_data) - 24;
569 left = find_tag(wav, s->pb, MKTAG('d', 'a', 't', 'a'));
572 if (wav->smv_data_ofs > 0 && !wav->smv_eof)
576 wav->data_end = avio_tell(s->pb) + left;
580 if (st->codec->block_align > 1) {
581 if (size < st->codec->block_align)
582 size = st->codec->block_align;
583 size = (size / st->codec->block_align) * st->codec->block_align;
585 size = FFMIN(size, left);
586 ret = av_get_packet(s->pb, pkt, size);
589 pkt->stream_index = 0;
594 static int wav_read_seek(AVFormatContext *s,
595 int stream_index, int64_t timestamp, int flags)
597 WAVDemuxContext *wav = s->priv_data;
601 if (wav->smv_data_ofs > 0) {
602 int64_t smv_timestamp = timestamp;
603 if (stream_index == 0)
604 smv_timestamp = av_rescale_q(timestamp, s->streams[0]->time_base, s->streams[1]->time_base);
606 timestamp = av_rescale_q(smv_timestamp, s->streams[1]->time_base, s->streams[0]->time_base);
607 if (wav->smv_frames_per_jpeg > 0) {
608 wav->smv_block = smv_timestamp / wav->smv_frames_per_jpeg;
609 wav->smv_cur_pt = smv_timestamp % wav->smv_frames_per_jpeg;
614 switch (st->codec->codec_id) {
615 case AV_CODEC_ID_MP2:
616 case AV_CODEC_ID_MP3:
617 case AV_CODEC_ID_AC3:
618 case AV_CODEC_ID_DTS:
619 /* use generic seeking with dynamically generated indexes */
624 return ff_pcm_read_seek(s, stream_index, timestamp, flags);
627 #define OFFSET(x) offsetof(WAVDemuxContext, x)
628 #define DEC AV_OPT_FLAG_DECODING_PARAM
629 static const AVOption demux_options[] = {
630 { "ignore_length", "Ignore length", OFFSET(ignore_length), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, DEC },
634 static const AVClass wav_demuxer_class = {
635 .class_name = "WAV demuxer",
636 .item_name = av_default_item_name,
637 .option = demux_options,
638 .version = LIBAVUTIL_VERSION_INT,
640 AVInputFormat ff_wav_demuxer = {
642 .long_name = NULL_IF_CONFIG_SMALL("WAV / WAVE (Waveform Audio)"),
643 .priv_data_size = sizeof(WAVDemuxContext),
644 .read_probe = wav_probe,
645 .read_header = wav_read_header,
646 .read_packet = wav_read_packet,
647 .read_seek = wav_read_seek,
648 .flags = AVFMT_GENERIC_INDEX,
649 .codec_tag = (const AVCodecTag * const []) { ff_codec_wav_tags, 0 },
650 .priv_class = &wav_demuxer_class,
652 #endif /* CONFIG_WAV_DEMUXER */
654 #if CONFIG_W64_DEMUXER
655 static int w64_probe(AVProbeData *p)
657 if (p->buf_size <= 40)
659 if (!memcmp(p->buf, ff_w64_guid_riff, 16) &&
660 !memcmp(p->buf + 24, ff_w64_guid_wave, 16))
661 return AVPROBE_SCORE_MAX;
666 static int w64_read_header(AVFormatContext *s)
668 int64_t size, data_ofs = 0;
669 AVIOContext *pb = s->pb;
670 WAVDemuxContext *wav = s->priv_data;
675 avio_read(pb, guid, 16);
676 if (memcmp(guid, ff_w64_guid_riff, 16))
677 return AVERROR_INVALIDDATA;
679 /* riff + wave + fmt + sizes */
680 if (avio_rl64(pb) < 16 + 8 + 16 + 8 + 16 + 8)
681 return AVERROR_INVALIDDATA;
683 avio_read(pb, guid, 16);
684 if (memcmp(guid, ff_w64_guid_wave, 16)) {
685 av_log(s, AV_LOG_ERROR, "could not find wave guid\n");
686 return AVERROR_INVALIDDATA;
691 st = avformat_new_stream(s, NULL);
693 return AVERROR(ENOMEM);
695 while (!avio_feof(pb)) {
696 if (avio_read(pb, guid, 16) != 16)
698 size = avio_rl64(pb);
699 if (size <= 24 || INT64_MAX - size < avio_tell(pb))
700 return AVERROR_INVALIDDATA;
702 if (!memcmp(guid, ff_w64_guid_fmt, 16)) {
703 /* subtract chunk header size - normal wav file doesn't count it */
704 ret = ff_get_wav_header(s, pb, st->codec, size - 24, 0);
707 avio_skip(pb, FFALIGN(size, INT64_C(8)) - size);
709 avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate);
710 } else if (!memcmp(guid, ff_w64_guid_fact, 16)) {
713 samples = avio_rl64(pb);
715 st->duration = samples;
716 } else if (!memcmp(guid, ff_w64_guid_data, 16)) {
717 wav->data_end = avio_tell(pb) + size - 24;
719 data_ofs = avio_tell(pb);
723 avio_skip(pb, size - 24);
724 } else if (!memcmp(guid, ff_w64_guid_summarylist, 16)) {
725 int64_t start, end, cur;
726 uint32_t count, chunk_size, i;
728 start = avio_tell(pb);
729 end = start + FFALIGN(size, INT64_C(8)) - 24;
730 count = avio_rl32(pb);
732 for (i = 0; i < count; i++) {
733 char chunk_key[5], *value;
735 if (avio_feof(pb) || (cur = avio_tell(pb)) < 0 || cur > end - 8 /* = tag + size */)
739 avio_read(pb, chunk_key, 4);
740 chunk_size = avio_rl32(pb);
742 value = av_mallocz(chunk_size + 1);
744 return AVERROR(ENOMEM);
746 ret = avio_get_str16le(pb, chunk_size, value, chunk_size);
747 avio_skip(pb, chunk_size - ret);
749 av_dict_set(&s->metadata, chunk_key, value, AV_DICT_DONT_STRDUP_VAL);
752 avio_skip(pb, end - avio_tell(pb));
754 av_log(s, AV_LOG_DEBUG, "unknown guid: "FF_PRI_GUID"\n", FF_ARG_GUID(guid));
755 avio_skip(pb, FFALIGN(size, INT64_C(8)) - 24);
762 ff_metadata_conv_ctx(s, NULL, wav_metadata_conv);
763 ff_metadata_conv_ctx(s, NULL, ff_riff_info_conv);
765 handle_stream_probing(st);
766 st->need_parsing = AVSTREAM_PARSE_FULL_RAW;
768 avio_seek(pb, data_ofs, SEEK_SET);
773 AVInputFormat ff_w64_demuxer = {
775 .long_name = NULL_IF_CONFIG_SMALL("Sony Wave64"),
776 .priv_data_size = sizeof(WAVDemuxContext),
777 .read_probe = w64_probe,
778 .read_header = w64_read_header,
779 .read_packet = wav_read_packet,
780 .read_seek = wav_read_seek,
781 .flags = AVFMT_GENERIC_INDEX,
782 .codec_tag = (const AVCodecTag * const []) { ff_codec_wav_tags, 0 },
784 #endif /* CONFIG_W64_DEMUXER */