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.
26 #include "libavutil/avstring.h"
27 #include "libavutil/bswap.h"
28 #include "libavutil/dict.h"
29 #include "libavcodec/get_bits.h"
30 #include "libavcodec/bytestream.h"
31 #include "libavcodec/vorbis_parser.h"
35 #include "vorbiscomment.h"
37 static int ogm_chapter(AVFormatContext *as, uint8_t *key, uint8_t *val)
39 int i, cnum, h, m, s, ms, keylen = strlen(key);
40 AVChapter *chapter = NULL;
42 if (keylen < 9 || sscanf(key, "CHAPTER%02d", &cnum) != 1)
46 if (sscanf(val, "%02d:%02d:%02d.%03d", &h, &m, &s, &ms) < 4)
49 avpriv_new_chapter(as, cnum, (AVRational){1,1000},
50 ms + 1000*(s + 60*(m + 60*h)),
51 AV_NOPTS_VALUE, NULL);
53 } else if (!strcmp(key+9, "NAME")) {
54 for(i = 0; i < as->nb_chapters; i++)
55 if (as->chapters[i]->id == cnum) {
56 chapter = as->chapters[i];
62 av_dict_set(&chapter->metadata, "title", val,
63 AV_DICT_DONT_STRDUP_VAL);
72 ff_vorbis_comment(AVFormatContext * as, AVDictionary **m, const uint8_t *buf, int size)
74 const uint8_t *p = buf;
75 const uint8_t *end = buf + size;
79 if (size < 8) /* must have vendor_length and user_comment_list_length */
82 s = bytestream_get_le32(&p);
84 if (end - p - 4 < s || s < 0)
89 n = bytestream_get_le32(&p);
91 while (end - p >= 4 && n > 0) {
95 s = bytestream_get_le32(&p);
97 if (end - p < s || s < 0)
104 v = memchr(t, '=', s);
115 tt = av_malloc(tl + 1);
116 ct = av_malloc(vl + 1);
120 av_log(as, AV_LOG_WARNING, "out-of-memory error. skipping VorbisComment tag.\n");
124 for (j = 0; j < tl; j++)
125 tt[j] = toupper(t[j]);
131 if (!ogm_chapter(as, tt, ct))
132 av_dict_set(m, tt, ct,
133 AV_DICT_DONT_STRDUP_KEY |
134 AV_DICT_DONT_STRDUP_VAL);
139 av_log(as, AV_LOG_INFO, "%ti bytes of comment header remain\n", end-p);
141 av_log(as, AV_LOG_INFO,
142 "truncated comment header, %i comments not found\n", n);
144 ff_metadata_conv(m, NULL, ff_vorbiscomment_metadata_conv);
150 /** Parse the vorbis header
151 * Vorbis Identification header from Vorbis_I_spec.html#vorbis-spec-codec
152 * [vorbis_version] = read 32 bits as unsigned integer | Not used
153 * [audio_channels] = read 8 bit integer as unsigned | Used
154 * [audio_sample_rate] = read 32 bits as unsigned integer | Used
155 * [bitrate_maximum] = read 32 bits as signed integer | Not used yet
156 * [bitrate_nominal] = read 32 bits as signed integer | Not used yet
157 * [bitrate_minimum] = read 32 bits as signed integer | Used as bitrate
158 * [blocksize_0] = read 4 bits as unsigned integer | Not Used
159 * [blocksize_1] = read 4 bits as unsigned integer | Not Used
160 * [framing_flag] = read one bit | Not Used
163 struct oggvorbis_private {
165 unsigned char *packet[3];
166 VorbisParseContext vp;
173 fixup_vorbis_headers(AVFormatContext * as, struct oggvorbis_private *priv,
179 len = priv->len[0] + priv->len[1] + priv->len[2];
180 ptr = *buf = av_mallocz(len + len/255 + 64);
184 offset += av_xiphlacing(&ptr[offset], priv->len[0]);
185 offset += av_xiphlacing(&ptr[offset], priv->len[1]);
186 for (i = 0; i < 3; i++) {
187 memcpy(&ptr[offset], priv->packet[i], priv->len[i]);
188 offset += priv->len[i];
189 av_freep(&priv->packet[i]);
191 *buf = av_realloc(*buf, offset + FF_INPUT_BUFFER_PADDING_SIZE);
195 static void vorbis_cleanup(AVFormatContext *s, int idx)
197 struct ogg *ogg = s->priv_data;
198 struct ogg_stream *os = ogg->streams + idx;
199 struct oggvorbis_private *priv = os->private;
202 for (i = 0; i < 3; i++)
203 av_freep(&priv->packet[i]);
207 vorbis_header (AVFormatContext * s, int idx)
209 struct ogg *ogg = s->priv_data;
210 struct ogg_stream *os = ogg->streams + idx;
211 AVStream *st = s->streams[idx];
212 struct oggvorbis_private *priv;
213 int pkt_type = os->buf[os->pstart];
216 return os->private ? 0 : -1;
219 os->private = av_mallocz(sizeof(struct oggvorbis_private));
224 if (os->psize < 1 || pkt_type > 5)
229 if (priv->packet[pkt_type>>1])
231 if (pkt_type > 1 && !priv->packet[0] || pkt_type > 3 && !priv->packet[1])
234 priv->len[pkt_type >> 1] = os->psize;
235 priv->packet[pkt_type >> 1] = av_mallocz(os->psize);
236 if (!priv->packet[pkt_type >> 1])
237 return AVERROR(ENOMEM);
238 memcpy(priv->packet[pkt_type >> 1], os->buf + os->pstart, os->psize);
239 if (os->buf[os->pstart] == 1) {
240 const uint8_t *p = os->buf + os->pstart + 7; /* skip "\001vorbis" tag */
241 unsigned blocksize, bs0, bs1;
248 if (bytestream_get_le32(&p) != 0) /* vorbis_version */
251 channels= bytestream_get_byte(&p);
252 if (st->codec->channels && channels != st->codec->channels) {
253 av_log(s, AV_LOG_ERROR, "Channel change is not supported\n");
254 return AVERROR_PATCHWELCOME;
256 st->codec->channels = channels;
257 srate = bytestream_get_le32(&p);
258 p += 4; // skip maximum bitrate
259 st->codec->bit_rate = bytestream_get_le32(&p); // nominal bitrate
260 p += 4; // skip minimum bitrate
262 blocksize = bytestream_get_byte(&p);
263 bs0 = blocksize & 15;
264 bs1 = blocksize >> 4;
268 if (bs0 < 6 || bs1 > 13)
271 if (bytestream_get_byte(&p) != 1) /* framing_flag */
274 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
275 st->codec->codec_id = AV_CODEC_ID_VORBIS;
278 st->codec->sample_rate = srate;
279 avpriv_set_pts_info(st, 64, 1, srate);
281 } else if (os->buf[os->pstart] == 3) {
283 ff_vorbis_comment(s, &st->metadata, os->buf + os->pstart + 7, os->psize - 8) >= 0) {
284 // drop all metadata we parsed and which is not required by libvorbis
285 unsigned new_len = 7 + 4 + AV_RL32(priv->packet[1] + 7) + 4 + 1;
286 if (new_len >= 16 && new_len < os->psize) {
287 AV_WL32(priv->packet[1] + new_len - 5, 0);
288 priv->packet[1][new_len - 1] = 1;
289 priv->len[1] = new_len;
294 st->codec->extradata_size =
295 fixup_vorbis_headers(s, priv, &st->codec->extradata);
296 if ((ret = avpriv_vorbis_parse_extradata(st->codec, &priv->vp))) {
297 av_freep(&st->codec->extradata);
298 st->codec->extradata_size = 0;
306 static int vorbis_packet(AVFormatContext *s, int idx)
308 struct ogg *ogg = s->priv_data;
309 struct ogg_stream *os = ogg->streams + idx;
310 struct oggvorbis_private *priv = os->private;
313 /* first packet handling
314 here we parse the duration of each packet in the first page and compare
315 the total duration to the page granule to find the encoder delay and
316 set the first timestamp */
317 if ((!os->lastpts || os->lastpts == AV_NOPTS_VALUE) && !(os->flags & OGG_FLAG_EOS)) {
319 uint8_t *last_pkt = os->buf + os->pstart;
320 uint8_t *next_pkt = last_pkt;
322 avpriv_vorbis_parse_reset(&priv->vp);
325 d = avpriv_vorbis_parse_frame(&priv->vp, last_pkt, 1);
327 os->pflags |= AV_PKT_FLAG_CORRUPT;
331 last_pkt = next_pkt = next_pkt + os->psize;
332 for (; seg < os->nsegs; seg++) {
333 if (os->segments[seg] < 255) {
334 int d = avpriv_vorbis_parse_frame(&priv->vp, last_pkt, 1);
336 duration = os->granule;
340 last_pkt = next_pkt + os->segments[seg];
342 next_pkt += os->segments[seg];
344 os->lastpts = os->lastdts = os->granule - duration;
345 if(s->streams[idx]->start_time == AV_NOPTS_VALUE) {
346 s->streams[idx]->start_time = FFMAX(os->lastpts, 0);
347 if (s->streams[idx]->duration)
348 s->streams[idx]->duration -= s->streams[idx]->start_time;
350 priv->final_pts = AV_NOPTS_VALUE;
351 avpriv_vorbis_parse_reset(&priv->vp);
354 /* parse packet duration */
356 duration = avpriv_vorbis_parse_frame(&priv->vp, os->buf + os->pstart, 1);
358 os->pflags |= AV_PKT_FLAG_CORRUPT;
361 os->pduration = duration;
364 /* final packet handling
365 here we save the pts of the first packet in the final page, sum up all
366 packet durations in the final page except for the last one, and compare
367 to the page granule to find the duration of the final packet */
368 if (os->flags & OGG_FLAG_EOS) {
369 if (os->lastpts != AV_NOPTS_VALUE) {
370 priv->final_pts = os->lastpts;
371 priv->final_duration = 0;
373 if (os->segp == os->nsegs)
374 os->pduration = os->granule - priv->final_pts - priv->final_duration;
375 priv->final_duration += os->pduration;
381 const struct ogg_codec ff_vorbis_codec = {
382 .magic = "\001vorbis",
384 .header = vorbis_header,
385 .packet = vorbis_packet,
386 .cleanup= vorbis_cleanup,