2 * Copyright (C) 2005 Michael Ahlberg, Måns Rullgård
4 * Permission is hereby granted, free of charge, to any person
5 * obtaining a copy of this software and associated documentation
6 * files (the "Software"), to deal in the Software without
7 * restriction, including without limitation the rights to use, copy,
8 * modify, merge, publish, distribute, sublicense, and/or sell copies
9 * of the Software, and to permit persons to whom the Software is
10 * furnished to do so, subject to the following conditions:
12 * The above copyright notice and this permission notice shall be
13 * included in all copies or substantial portions of the Software.
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
16 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
17 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
18 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
19 * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
20 * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
22 * DEALINGS IN THE SOFTWARE.
27 #include "libavutil/avstring.h"
28 #include "libavutil/base64.h"
29 #include "libavutil/bswap.h"
30 #include "libavutil/dict.h"
31 #include "libavcodec/bytestream.h"
32 #include "libavcodec/get_bits.h"
33 #include "libavcodec/vorbis_parser.h"
35 #include "flac_picture.h"
38 #include "vorbiscomment.h"
39 #include "replaygain.h"
41 static int ogm_chapter(AVFormatContext *as, uint8_t *key, uint8_t *val)
43 int i, cnum, h, m, s, ms, keylen = strlen(key);
44 AVChapter *chapter = NULL;
46 if (keylen < 9 || sscanf(key, "CHAPTER%03d", &cnum) != 1)
50 if (sscanf(val, "%02d:%02d:%02d.%03d", &h, &m, &s, &ms) < 4)
53 avpriv_new_chapter(as, cnum, (AVRational) { 1, 1000 },
54 ms + 1000 * (s + 60 * (m + 60 * h)),
55 AV_NOPTS_VALUE, NULL);
57 } else if (!strcmp(key + keylen - 4, "NAME")) {
58 for (i = 0; i < as->nb_chapters; i++)
59 if (as->chapters[i]->id == cnum) {
60 chapter = as->chapters[i];
66 av_dict_set(&chapter->metadata, "title", val, AV_DICT_DONT_STRDUP_VAL);
74 int ff_vorbis_stream_comment(AVFormatContext *as, AVStream *st,
75 const uint8_t *buf, int size)
77 int updates = ff_vorbis_comment(as, &st->metadata, buf, size, 1);
80 st->event_flags |= AVSTREAM_EVENT_FLAG_METADATA_UPDATED;
86 int ff_vorbis_comment(AVFormatContext *as, AVDictionary **m,
87 const uint8_t *buf, int size,
90 const uint8_t *p = buf;
91 const uint8_t *end = buf + size;
96 /* must have vendor_length and user_comment_list_length */
98 return AVERROR_INVALIDDATA;
100 s = bytestream_get_le32(&p);
102 if (end - p - 4 < s || s < 0)
103 return AVERROR_INVALIDDATA;
107 n = bytestream_get_le32(&p);
109 while (end - p >= 4 && n > 0) {
113 s = bytestream_get_le32(&p);
115 if (end - p < s || s < 0)
122 v = memchr(t, '=', s);
133 tt = av_malloc(tl + 1);
134 ct = av_malloc(vl + 1);
138 return AVERROR(ENOMEM);
141 for (j = 0; j < tl; j++)
142 tt[j] = av_toupper(t[j]);
148 /* The format in which the pictures are stored is the FLAC format.
149 * Xiph says: "The binary FLAC picture structure is base64 encoded
150 * and placed within a VorbisComment with the tag name
151 * 'METADATA_BLOCK_PICTURE'. This is the preferred and
152 * recommended way of embedding cover art within VorbisComments."
154 if (!strcmp(tt, "METADATA_BLOCK_PICTURE") && parse_picture) {
156 char *pict = av_malloc(vl);
159 av_log(as, AV_LOG_WARNING, "out-of-memory error. Skipping cover art block.\n");
164 if ((ret = av_base64_decode(pict, ct, vl)) > 0)
165 ret = ff_flac_parse_picture(as, pict, ret);
170 av_log(as, AV_LOG_WARNING, "Failed to parse cover art block.\n");
173 } else if (!ogm_chapter(as, tt, ct)) {
175 if (av_dict_get(*m, tt, NULL, 0)) {
176 av_dict_set(m, tt, ";", AV_DICT_APPEND);
178 av_dict_set(m, tt, ct,
179 AV_DICT_DONT_STRDUP_KEY |
187 av_log(as, AV_LOG_INFO,
188 "%"PTRDIFF_SPECIFIER" bytes of comment header remain\n", end - p);
190 av_log(as, AV_LOG_INFO,
191 "truncated comment header, %i comments not found\n", n);
193 ff_metadata_conv(m, NULL, ff_vorbiscomment_metadata_conv);
199 * Parse the vorbis header
201 * Vorbis Identification header from Vorbis_I_spec.html#vorbis-spec-codec
202 * [vorbis_version] = read 32 bits as unsigned integer | Not used
203 * [audio_channels] = read 8 bit integer as unsigned | Used
204 * [audio_sample_rate] = read 32 bits as unsigned integer | Used
205 * [bitrate_maximum] = read 32 bits as signed integer | Not used yet
206 * [bitrate_nominal] = read 32 bits as signed integer | Not used yet
207 * [bitrate_minimum] = read 32 bits as signed integer | Used as bitrate
208 * [blocksize_0] = read 4 bits as unsigned integer | Not Used
209 * [blocksize_1] = read 4 bits as unsigned integer | Not Used
210 * [framing_flag] = read one bit | Not Used
213 struct oggvorbis_private {
215 unsigned char *packet[3];
216 AVVorbisParseContext *vp;
221 static int fixup_vorbis_headers(AVFormatContext *as,
222 struct oggvorbis_private *priv,
225 int i, offset, len, err;
229 len = priv->len[0] + priv->len[1] + priv->len[2];
230 buf_len = len + len / 255 + 64;
231 ptr = *buf = av_realloc(NULL, buf_len);
233 return AVERROR(ENOMEM);
234 memset(*buf, '\0', buf_len);
238 offset += av_xiphlacing(&ptr[offset], priv->len[0]);
239 offset += av_xiphlacing(&ptr[offset], priv->len[1]);
240 for (i = 0; i < 3; i++) {
241 memcpy(&ptr[offset], priv->packet[i], priv->len[i]);
242 offset += priv->len[i];
243 av_freep(&priv->packet[i]);
245 if ((err = av_reallocp(buf, offset + AV_INPUT_BUFFER_PADDING_SIZE)) < 0)
250 static void vorbis_cleanup(AVFormatContext *s, int idx)
252 struct ogg *ogg = s->priv_data;
253 struct ogg_stream *os = ogg->streams + idx;
254 struct oggvorbis_private *priv = os->private;
257 av_vorbis_parse_free(&priv->vp);
258 for (i = 0; i < 3; i++)
259 av_freep(&priv->packet[i]);
263 static int vorbis_update_metadata(AVFormatContext *s, int idx)
265 struct ogg *ogg = s->priv_data;
266 struct ogg_stream *os = ogg->streams + idx;
267 AVStream *st = s->streams[idx];
273 /* New metadata packet; release old data. */
274 av_dict_free(&st->metadata);
275 ret = ff_vorbis_stream_comment(s, st, os->buf + os->pstart + 7,
280 /* Update the metadata if possible. */
281 av_freep(&os->new_metadata);
283 os->new_metadata = av_packet_pack_dictionary(st->metadata, &os->new_metadata_size);
284 /* Send an empty dictionary to indicate that metadata has been cleared. */
286 os->new_metadata = av_malloc(1);
287 os->new_metadata_size = 0;
293 static int vorbis_header(AVFormatContext *s, int idx)
295 struct ogg *ogg = s->priv_data;
296 AVStream *st = s->streams[idx];
297 struct ogg_stream *os = ogg->streams + idx;
298 struct oggvorbis_private *priv;
299 int pkt_type = os->buf[os->pstart];
302 os->private = av_mallocz(sizeof(struct oggvorbis_private));
304 return AVERROR(ENOMEM);
310 return priv->vp ? 0 : AVERROR_INVALIDDATA;
312 if (os->psize < 1 || pkt_type > 5)
313 return AVERROR_INVALIDDATA;
315 if (priv->packet[pkt_type >> 1])
316 return AVERROR_INVALIDDATA;
317 if (pkt_type > 1 && !priv->packet[0] || pkt_type > 3 && !priv->packet[1])
318 return AVERROR_INVALIDDATA;
320 priv->len[pkt_type >> 1] = os->psize;
321 priv->packet[pkt_type >> 1] = av_mallocz(os->psize);
322 if (!priv->packet[pkt_type >> 1])
323 return AVERROR(ENOMEM);
324 memcpy(priv->packet[pkt_type >> 1], os->buf + os->pstart, os->psize);
325 if (os->buf[os->pstart] == 1) {
326 const uint8_t *p = os->buf + os->pstart + 7; /* skip "\001vorbis" tag */
327 unsigned blocksize, bs0, bs1;
332 return AVERROR_INVALIDDATA;
334 if (bytestream_get_le32(&p) != 0) /* vorbis_version */
335 return AVERROR_INVALIDDATA;
337 channels = bytestream_get_byte(&p);
338 if (st->codecpar->channels && channels != st->codecpar->channels) {
339 av_log(s, AV_LOG_ERROR, "Channel change is not supported\n");
340 return AVERROR_PATCHWELCOME;
342 st->codecpar->channels = channels;
343 srate = bytestream_get_le32(&p);
344 p += 4; // skip maximum bitrate
345 st->codecpar->bit_rate = bytestream_get_le32(&p); // nominal bitrate
346 p += 4; // skip minimum bitrate
348 blocksize = bytestream_get_byte(&p);
349 bs0 = blocksize & 15;
350 bs1 = blocksize >> 4;
353 return AVERROR_INVALIDDATA;
354 if (bs0 < 6 || bs1 > 13)
355 return AVERROR_INVALIDDATA;
357 if (bytestream_get_byte(&p) != 1) /* framing_flag */
358 return AVERROR_INVALIDDATA;
360 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
361 st->codecpar->codec_id = AV_CODEC_ID_VORBIS;
364 st->codecpar->sample_rate = srate;
365 avpriv_set_pts_info(st, 64, 1, srate);
367 } else if (os->buf[os->pstart] == 3) {
368 if (vorbis_update_metadata(s, idx) >= 0 && priv->len[1] > 10) {
371 int ret = ff_replaygain_export(st, st->metadata);
375 // drop all metadata we parsed and which is not required by libvorbis
376 new_len = 7 + 4 + AV_RL32(priv->packet[1] + 7) + 4 + 1;
377 if (new_len >= 16 && new_len < os->psize) {
378 AV_WL32(priv->packet[1] + new_len - 5, 0);
379 priv->packet[1][new_len - 1] = 1;
380 priv->len[1] = new_len;
384 int ret = fixup_vorbis_headers(s, priv, &st->codecpar->extradata);
386 st->codecpar->extradata_size = 0;
389 st->codecpar->extradata_size = ret;
391 priv->vp = av_vorbis_parse_init(st->codecpar->extradata, st->codecpar->extradata_size);
393 av_freep(&st->codecpar->extradata);
394 st->codecpar->extradata_size = 0;
395 return AVERROR_UNKNOWN;
402 static int vorbis_packet(AVFormatContext *s, int idx)
404 struct ogg *ogg = s->priv_data;
405 struct ogg_stream *os = ogg->streams + idx;
406 struct oggvorbis_private *priv = os->private;
407 int duration, flags = 0;
410 return AVERROR_INVALIDDATA;
412 /* first packet handling
413 * here we parse the duration of each packet in the first page and compare
414 * the total duration to the page granule to find the encoder delay and
415 * set the first timestamp */
416 if ((!os->lastpts || os->lastpts == AV_NOPTS_VALUE) && !(os->flags & OGG_FLAG_EOS) && (int64_t)os->granule>=0) {
418 uint8_t *last_pkt = os->buf + os->pstart;
419 uint8_t *next_pkt = last_pkt;
421 av_vorbis_parse_reset(priv->vp);
424 d = av_vorbis_parse_frame_flags(priv->vp, last_pkt, 1, &flags);
426 os->pflags |= AV_PKT_FLAG_CORRUPT;
428 } else if (flags & VORBIS_FLAG_COMMENT) {
429 vorbis_update_metadata(s, idx);
433 last_pkt = next_pkt = next_pkt + os->psize;
434 for (; seg < os->nsegs; seg++) {
435 if (os->segments[seg] < 255) {
436 int d = av_vorbis_parse_frame_flags(priv->vp, last_pkt, 1, &flags);
438 duration = os->granule;
440 } else if (flags & VORBIS_FLAG_COMMENT) {
441 vorbis_update_metadata(s, idx);
445 last_pkt = next_pkt + os->segments[seg];
447 next_pkt += os->segments[seg];
450 os->lastdts = os->granule - duration;
452 if (!os->granule && duration) //hack to deal with broken files (Ticket3710)
453 os->lastpts = os->lastdts = AV_NOPTS_VALUE;
455 if (s->streams[idx]->start_time == AV_NOPTS_VALUE) {
456 s->streams[idx]->start_time = FFMAX(os->lastpts, 0);
457 if (s->streams[idx]->duration != AV_NOPTS_VALUE)
458 s->streams[idx]->duration -= s->streams[idx]->start_time;
460 priv->final_pts = AV_NOPTS_VALUE;
461 av_vorbis_parse_reset(priv->vp);
464 /* parse packet duration */
466 duration = av_vorbis_parse_frame_flags(priv->vp, os->buf + os->pstart, 1, &flags);
468 os->pflags |= AV_PKT_FLAG_CORRUPT;
470 } else if (flags & VORBIS_FLAG_COMMENT) {
471 vorbis_update_metadata(s, idx);
474 os->pduration = duration;
477 /* final packet handling
478 * here we save the pts of the first packet in the final page, sum up all
479 * packet durations in the final page except for the last one, and compare
480 * to the page granule to find the duration of the final packet */
481 if (os->flags & OGG_FLAG_EOS) {
482 if (os->lastpts != AV_NOPTS_VALUE) {
483 priv->final_pts = os->lastpts;
484 priv->final_duration = 0;
486 if (os->segp == os->nsegs)
487 os->pduration = os->granule - priv->final_pts - priv->final_duration;
488 priv->final_duration += os->pduration;
494 const struct ogg_codec ff_vorbis_codec = {
495 .magic = "\001vorbis",
497 .header = vorbis_header,
498 .packet = vorbis_packet,
499 .cleanup = vorbis_cleanup,