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_comment(AVFormatContext *as, AVDictionary **m,
75 const uint8_t *buf, int size,
78 const uint8_t *p = buf;
79 const uint8_t *end = buf + size;
83 /* must have vendor_length and user_comment_list_length */
85 return AVERROR_INVALIDDATA;
87 s = bytestream_get_le32(&p);
89 if (end - p - 4 < s || s < 0)
90 return AVERROR_INVALIDDATA;
94 n = bytestream_get_le32(&p);
96 while (end - p >= 4 && n > 0) {
100 s = bytestream_get_le32(&p);
102 if (end - p < s || s < 0)
109 v = memchr(t, '=', s);
120 tt = av_malloc(tl + 1);
121 ct = av_malloc(vl + 1);
125 return AVERROR(ENOMEM);
128 for (j = 0; j < tl; j++)
129 tt[j] = av_toupper(t[j]);
135 /* The format in which the pictures are stored is the FLAC format.
136 * Xiph says: "The binary FLAC picture structure is base64 encoded
137 * and placed within a VorbisComment with the tag name
138 * 'METADATA_BLOCK_PICTURE'. This is the preferred and
139 * recommended way of embedding cover art within VorbisComments."
141 if (!strcmp(tt, "METADATA_BLOCK_PICTURE") && parse_picture) {
143 char *pict = av_malloc(vl);
148 return AVERROR(ENOMEM);
150 if ((ret = av_base64_decode(pict, ct, vl)) > 0)
151 ret = ff_flac_parse_picture(as, pict, ret);
156 av_log(as, AV_LOG_WARNING, "Failed to parse cover art block.\n");
159 } else if (!ogm_chapter(as, tt, ct))
160 av_dict_set(m, tt, ct,
161 AV_DICT_DONT_STRDUP_KEY |
162 AV_DICT_DONT_STRDUP_VAL);
167 av_log(as, AV_LOG_INFO,
168 "%ti bytes of comment header remain\n", end - p);
170 av_log(as, AV_LOG_INFO,
171 "truncated comment header, %i comments not found\n", n);
173 ff_metadata_conv(m, NULL, ff_vorbiscomment_metadata_conv);
179 * Parse the vorbis header
181 * Vorbis Identification header from Vorbis_I_spec.html#vorbis-spec-codec
182 * [vorbis_version] = read 32 bits as unsigned integer | Not used
183 * [audio_channels] = read 8 bit integer as unsigned | Used
184 * [audio_sample_rate] = read 32 bits as unsigned integer | Used
185 * [bitrate_maximum] = read 32 bits as signed integer | Not used yet
186 * [bitrate_nominal] = read 32 bits as signed integer | Not used yet
187 * [bitrate_minimum] = read 32 bits as signed integer | Used as bitrate
188 * [blocksize_0] = read 4 bits as unsigned integer | Not Used
189 * [blocksize_1] = read 4 bits as unsigned integer | Not Used
190 * [framing_flag] = read one bit | Not Used
193 struct oggvorbis_private {
195 unsigned char *packet[3];
196 VorbisParseContext vp;
201 static int fixup_vorbis_headers(AVFormatContext *as,
202 struct oggvorbis_private *priv,
205 int i, offset, len, err;
208 len = priv->len[0] + priv->len[1] + priv->len[2];
209 ptr = *buf = av_mallocz(len + len / 255 + 64);
211 return AVERROR(ENOMEM);
215 offset += av_xiphlacing(&ptr[offset], priv->len[0]);
216 offset += av_xiphlacing(&ptr[offset], priv->len[1]);
217 for (i = 0; i < 3; i++) {
218 memcpy(&ptr[offset], priv->packet[i], priv->len[i]);
219 offset += priv->len[i];
220 av_freep(&priv->packet[i]);
222 if ((err = av_reallocp(buf, offset + FF_INPUT_BUFFER_PADDING_SIZE)) < 0)
227 static void vorbis_cleanup(AVFormatContext *s, int idx)
229 struct ogg *ogg = s->priv_data;
230 struct ogg_stream *os = ogg->streams + idx;
231 struct oggvorbis_private *priv = os->private;
234 for (i = 0; i < 3; i++)
235 av_freep(&priv->packet[i]);
238 static int vorbis_header(AVFormatContext *s, int idx)
240 struct ogg *ogg = s->priv_data;
241 AVStream *st = s->streams[idx];
242 struct ogg_stream *os = ogg->streams + idx;
243 struct oggvorbis_private *priv;
244 int pkt_type = os->buf[os->pstart];
247 os->private = av_mallocz(sizeof(struct oggvorbis_private));
249 return AVERROR(ENOMEM);
255 if (os->psize < 1 || pkt_type > 5)
256 return AVERROR_INVALIDDATA;
260 if (priv->packet[pkt_type >> 1])
261 return AVERROR_INVALIDDATA;
262 if (pkt_type > 1 && !priv->packet[0] || pkt_type > 3 && !priv->packet[1])
263 return AVERROR_INVALIDDATA;
265 priv->len[pkt_type >> 1] = os->psize;
266 priv->packet[pkt_type >> 1] = av_mallocz(os->psize);
267 if (!priv->packet[pkt_type >> 1])
268 return AVERROR(ENOMEM);
269 memcpy(priv->packet[pkt_type >> 1], os->buf + os->pstart, os->psize);
270 if (os->buf[os->pstart] == 1) {
271 const uint8_t *p = os->buf + os->pstart + 7; /* skip "\001vorbis" tag */
272 unsigned blocksize, bs0, bs1;
276 return AVERROR_INVALIDDATA;
278 if (bytestream_get_le32(&p) != 0) /* vorbis_version */
279 return AVERROR_INVALIDDATA;
281 st->codec->channels = bytestream_get_byte(&p);
282 srate = bytestream_get_le32(&p);
283 p += 4; // skip maximum bitrate
284 st->codec->bit_rate = bytestream_get_le32(&p); // nominal bitrate
285 p += 4; // skip minimum bitrate
287 blocksize = bytestream_get_byte(&p);
288 bs0 = blocksize & 15;
289 bs1 = blocksize >> 4;
292 return AVERROR_INVALIDDATA;
293 if (bs0 < 6 || bs1 > 13)
294 return AVERROR_INVALIDDATA;
296 if (bytestream_get_byte(&p) != 1) /* framing_flag */
297 return AVERROR_INVALIDDATA;
299 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
300 st->codec->codec_id = AV_CODEC_ID_VORBIS;
303 st->codec->sample_rate = srate;
304 avpriv_set_pts_info(st, 64, 1, srate);
306 } else if (os->buf[os->pstart] == 3) {
308 ff_vorbis_comment(s, &st->metadata, os->buf + os->pstart + 7,
309 os->psize - 8, 1) >= 0) {
312 int ret = ff_replaygain_export(st, st->metadata);
316 // drop all metadata we parsed and which is not required by libvorbis
317 new_len = 7 + 4 + AV_RL32(priv->packet[1] + 7) + 4 + 1;
318 if (new_len >= 16 && new_len < os->psize) {
319 AV_WL32(priv->packet[1] + new_len - 5, 0);
320 priv->packet[1][new_len - 1] = 1;
321 priv->len[1] = new_len;
325 int ret = fixup_vorbis_headers(s, priv, &st->codec->extradata);
327 st->codec->extradata_size = 0;
330 st->codec->extradata_size = ret;
331 if ((ret = avpriv_vorbis_parse_extradata(st->codec, &priv->vp))) {
332 av_freep(&st->codec->extradata);
333 st->codec->extradata_size = 0;
341 static int vorbis_packet(AVFormatContext *s, int idx)
343 struct ogg *ogg = s->priv_data;
344 struct ogg_stream *os = ogg->streams + idx;
345 struct oggvorbis_private *priv = os->private;
348 /* first packet handling
349 * here we parse the duration of each packet in the first page and compare
350 * the total duration to the page granule to find the encoder delay and
351 * set the first timestamp */
354 uint8_t *last_pkt = os->buf + os->pstart;
355 uint8_t *next_pkt = last_pkt;
356 int first_duration = 0;
358 avpriv_vorbis_parse_reset(&priv->vp);
360 for (seg = 0; seg < os->nsegs; seg++) {
361 if (os->segments[seg] < 255) {
362 int d = avpriv_vorbis_parse_frame(&priv->vp, last_pkt, 1);
364 duration = os->granule;
370 last_pkt = next_pkt + os->segments[seg];
372 next_pkt += os->segments[seg];
375 os->lastdts = os->granule - duration;
376 s->streams[idx]->start_time = os->lastpts + first_duration;
377 if (s->streams[idx]->duration)
378 s->streams[idx]->duration -= s->streams[idx]->start_time;
379 s->streams[idx]->cur_dts = AV_NOPTS_VALUE;
380 priv->final_pts = AV_NOPTS_VALUE;
381 avpriv_vorbis_parse_reset(&priv->vp);
384 /* parse packet duration */
386 duration = avpriv_vorbis_parse_frame(&priv->vp, os->buf + os->pstart, 1);
388 os->pflags |= AV_PKT_FLAG_CORRUPT;
391 os->pduration = duration;
394 /* final packet handling
395 * here we save the pts of the first packet in the final page, sum up all
396 * packet durations in the final page except for the last one, and compare
397 * to the page granule to find the duration of the final packet */
398 if (os->flags & OGG_FLAG_EOS) {
399 if (os->lastpts != AV_NOPTS_VALUE) {
400 priv->final_pts = os->lastpts;
401 priv->final_duration = 0;
403 if (os->segp == os->nsegs)
404 os->pduration = os->granule - priv->final_pts - priv->final_duration;
405 priv->final_duration += os->pduration;
411 const struct ogg_codec ff_vorbis_codec = {
412 .magic = "\001vorbis",
414 .header = vorbis_header,
415 .packet = vorbis_packet,
416 .cleanup = vorbis_cleanup,