14d8254f4c15463c4e63b557c0f68928f3b88593
[libav.git] / libavformat / mp3dec.c
1 /*
2 * MP3 demuxer
3 * Copyright (c) 2003 Fabrice Bellard
4 *
5 * This file is part of Libav.
6 *
7 * Libav is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * Libav is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22 #include "libavutil/avstring.h"
23 #include "libavutil/intreadwrite.h"
24 #include "libavutil/crc.h"
25 #include "libavutil/dict.h"
26 #include "libavutil/mathematics.h"
27 #include "avformat.h"
28 #include "internal.h"
29 #include "avio_internal.h"
30 #include "id3v2.h"
31 #include "id3v1.h"
32 #include "replaygain.h"
33
34 #include "libavcodec/mpegaudiodecheader.h"
35
36 #define XING_FLAG_FRAMES 0x01
37 #define XING_FLAG_SIZE 0x02
38 #define XING_FLAG_TOC 0x04
39
40 #define XING_TOC_COUNT 100
41
42 typedef struct MP3DecContext {
43 int xing_toc;
44 unsigned frames; /* Total number of frames in file */
45 unsigned size; /* Total number of bytes in the stream */
46 int is_cbr;
47 } MP3DecContext;
48
49 /* mp3 read */
50
51 static int mp3_read_probe(AVProbeData *p)
52 {
53 int max_frames, first_frames = 0;
54 int fsize, frames, sample_rate;
55 uint32_t header;
56 uint8_t *buf, *buf0, *buf2, *end;
57 AVCodecContext avctx;
58
59 buf0 = p->buf;
60 end = p->buf + p->buf_size - sizeof(uint32_t);
61 while(buf0 < end && !*buf0)
62 buf0++;
63
64 max_frames = 0;
65 buf = buf0;
66
67 for(; buf < end; buf= buf2+1) {
68 buf2 = buf;
69
70 for(frames = 0; buf2 < end; frames++) {
71 header = AV_RB32(buf2);
72 fsize = avpriv_mpa_decode_header(&avctx, header, &sample_rate, &sample_rate, &sample_rate, &sample_rate);
73 if(fsize < 0)
74 break;
75 buf2 += fsize;
76 }
77 max_frames = FFMAX(max_frames, frames);
78 if(buf == buf0)
79 first_frames= frames;
80 }
81 // keep this in sync with ac3 probe, both need to avoid
82 // issues with MPEG-files!
83 if (first_frames >= 4) return AVPROBE_SCORE_EXTENSION + 1;
84
85 if (max_frames) {
86 int pes = 0, i;
87 unsigned int code = -1;
88
89 #define VIDEO_ID 0x000001e0
90 #define AUDIO_ID 0x000001c0
91 /* do a search for mpegps headers to be able to properly bias
92 * towards mpegps if we detect this stream as both. */
93 for (i = 0; i<p->buf_size; i++) {
94 code = (code << 8) + p->buf[i];
95 if ((code & 0xffffff00) == 0x100) {
96 if ((code & 0x1f0) == VIDEO_ID) pes++;
97 else if((code & 0x1e0) == AUDIO_ID) pes++;
98 }
99 }
100
101 if (pes)
102 max_frames = (max_frames + pes - 1) / pes;
103 }
104 if (max_frames > 500) return AVPROBE_SCORE_EXTENSION;
105 else if (max_frames >= 4) return AVPROBE_SCORE_EXTENSION / 2;
106 else if (max_frames >= 1) return 1;
107 else return 0;
108 //mpegps_mp3_unrecognized_format.mpg has max_frames=3
109 }
110
111 static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration)
112 {
113 int i;
114 MP3DecContext *mp3 = s->priv_data;
115
116 if (!filesize &&
117 !(filesize = avio_size(s->pb))) {
118 av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n");
119 return;
120 }
121
122 for (i = 0; i < XING_TOC_COUNT; i++) {
123 uint8_t b = avio_r8(s->pb);
124
125 av_add_index_entry(s->streams[0],
126 av_rescale(b, filesize, 256),
127 av_rescale(i, duration, XING_TOC_COUNT),
128 0, 0, AVINDEX_KEYFRAME);
129 }
130 mp3->xing_toc = 1;
131 }
132
133 static void mp3_parse_info_tag(AVFormatContext *s, AVStream *st,
134 MPADecodeHeader *c, uint32_t spf)
135 {
136 #define LAST_BITS(k, n) ((k) & ((1 << (n)) - 1))
137 #define MIDDLE_BITS(k, m, n) LAST_BITS((k) >> (m), ((n) - (m)))
138
139 uint16_t crc;
140 uint32_t v;
141
142 char version[10];
143
144 uint32_t peak = 0;
145 int32_t r_gain = INT32_MIN, a_gain = INT32_MIN;
146
147 MP3DecContext *mp3 = s->priv_data;
148 const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
149
150 /* Check for Xing / Info tag */
151 avio_skip(s->pb, xing_offtbl[c->lsf == 1][c->nb_channels == 1]);
152 v = avio_rb32(s->pb);
153 mp3->is_cbr = v == MKBETAG('I', 'n', 'f', 'o');
154 if (v != MKBETAG('X', 'i', 'n', 'g') && !mp3->is_cbr)
155 return;
156
157 v = avio_rb32(s->pb);
158 if (v & XING_FLAG_FRAMES)
159 mp3->frames = avio_rb32(s->pb);
160 if (v & XING_FLAG_SIZE)
161 mp3->size = avio_rb32(s->pb);
162 if (v & XING_FLAG_TOC && mp3->frames)
163 read_xing_toc(s, mp3->size, av_rescale_q(mp3->frames,
164 (AVRational){spf, c->sample_rate},
165 st->time_base));
166
167 /* VBR quality */
168 avio_rb32(s->pb);
169
170 /* Encoder short version string */
171 memset(version, 0, sizeof(version));
172 avio_read(s->pb, version, 9);
173
174 /* Info Tag revision + VBR method */
175 avio_r8(s->pb);
176
177 /* Lowpass filter value */
178 avio_r8(s->pb);
179
180 /* ReplayGain peak */
181 v = avio_rb32(s->pb);
182 peak = av_rescale(v, 100000, 1 << 23);
183
184 /* Radio ReplayGain */
185 v = avio_rb16(s->pb);
186
187 if (MIDDLE_BITS(v, 13, 15) == 1) {
188 r_gain = MIDDLE_BITS(v, 0, 8) * 10000;
189
190 if (v & (1 << 9))
191 r_gain *= -1;
192 }
193
194 /* Audiophile ReplayGain */
195 v = avio_rb16(s->pb);
196
197 if (MIDDLE_BITS(v, 13, 15) == 2) {
198 a_gain = MIDDLE_BITS(v, 0, 8) * 10000;
199
200 if (v & (1 << 9))
201 a_gain *= -1;
202 }
203
204 /* Encoding flags + ATH Type */
205 avio_r8(s->pb);
206
207 /* if ABR {specified bitrate} else {minimal bitrate} */
208 avio_r8(s->pb);
209
210 /* Encoder delays */
211 avio_rb24(s->pb);
212
213 /* Misc */
214 avio_r8(s->pb);
215
216 /* MP3 gain */
217 avio_r8(s->pb);
218
219 /* Preset and surround info */
220 avio_rb16(s->pb);
221
222 /* Music length */
223 avio_rb32(s->pb);
224
225 /* Music CRC */
226 avio_rb16(s->pb);
227
228 /* Info Tag CRC */
229 crc = ffio_get_checksum(s->pb);
230 v = avio_rb16(s->pb);
231
232 if (v == crc) {
233 ff_replaygain_export_raw(st, r_gain, peak, a_gain, 0);
234 av_dict_set(&st->metadata, "encoder", version, 0);
235 }
236 }
237
238 static void mp3_parse_vbri_tag(AVFormatContext *s, AVStream *st, int64_t base)
239 {
240 uint32_t v;
241 MP3DecContext *mp3 = s->priv_data;
242
243 /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */
244 avio_seek(s->pb, base + 4 + 32, SEEK_SET);
245 v = avio_rb32(s->pb);
246 if (v == MKBETAG('V', 'B', 'R', 'I')) {
247 /* Check tag version */
248 if (avio_rb16(s->pb) == 1) {
249 /* skip delay and quality */
250 avio_skip(s->pb, 4);
251 mp3->size = avio_rb32(s->pb);
252 mp3->frames = avio_rb32(s->pb);
253 }
254 }
255 }
256
257 /**
258 * Try to find Xing/Info/VBRI tags and compute duration from info therein
259 */
260 static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base)
261 {
262 uint32_t v, spf;
263 MPADecodeHeader c;
264 int vbrtag_size = 0;
265 MP3DecContext *mp3 = s->priv_data;
266
267 ffio_init_checksum(s->pb, ff_crcA001_update, 0);
268
269 v = avio_rb32(s->pb);
270 if(ff_mpa_check_header(v) < 0)
271 return -1;
272
273 if (avpriv_mpegaudio_decode_header(&c, v) == 0)
274 vbrtag_size = c.frame_size;
275 if(c.layer != 3)
276 return -1;
277
278 spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */
279
280 mp3->frames = 0;
281 mp3->size = 0;
282
283 mp3_parse_info_tag(s, st, &c, spf);
284 mp3_parse_vbri_tag(s, st, base);
285
286 if (!mp3->frames && !mp3->size)
287 return -1;
288
289 /* Skip the vbr tag frame */
290 avio_seek(s->pb, base + vbrtag_size, SEEK_SET);
291
292 if (mp3->frames)
293 st->duration = av_rescale_q(mp3->frames, (AVRational){spf, c.sample_rate},
294 st->time_base);
295 if (mp3->size && mp3->frames && !mp3->is_cbr)
296 st->codec->bit_rate = av_rescale(mp3->size, 8 * c.sample_rate, mp3->frames * (int64_t)spf);
297
298 return 0;
299 }
300
301 static int mp3_read_header(AVFormatContext *s)
302 {
303 AVStream *st;
304 int64_t off;
305 int ret;
306
307 st = avformat_new_stream(s, NULL);
308 if (!st)
309 return AVERROR(ENOMEM);
310
311 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
312 st->codec->codec_id = AV_CODEC_ID_MP3;
313 st->need_parsing = AVSTREAM_PARSE_FULL;
314 st->start_time = 0;
315
316 // lcm of all mp3 sample rates
317 avpriv_set_pts_info(st, 64, 1, 14112000);
318
319 off = avio_tell(s->pb);
320
321 if (!av_dict_get(s->metadata, "", NULL, AV_DICT_IGNORE_SUFFIX))
322 ff_id3v1_read(s);
323
324 if (mp3_parse_vbr_tags(s, st, off) < 0)
325 avio_seek(s->pb, off, SEEK_SET);
326
327 ret = ff_replaygain_export(st, s->metadata);
328 if (ret < 0)
329 return ret;
330
331 /* the parameters will be extracted from the compressed bitstream */
332 return 0;
333 }
334
335 #define MP3_PACKET_SIZE 1024
336
337 static int mp3_read_packet(AVFormatContext *s, AVPacket *pkt)
338 {
339 int ret;
340
341 ret = av_get_packet(s->pb, pkt, MP3_PACKET_SIZE);
342 if (ret < 0)
343 return ret;
344
345 pkt->stream_index = 0;
346
347 if (ret > ID3v1_TAG_SIZE &&
348 memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0)
349 ret -= ID3v1_TAG_SIZE;
350
351 /* note: we need to modify the packet size here to handle the last
352 packet */
353 pkt->size = ret;
354 return ret;
355 }
356
357 static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp,
358 int flags)
359 {
360 MP3DecContext *mp3 = s->priv_data;
361 AVIndexEntry *ie;
362 AVStream *st = s->streams[0];
363 int64_t ret = av_index_search_timestamp(st, timestamp, flags);
364 uint32_t header = 0;
365
366 if (!mp3->xing_toc)
367 return AVERROR(ENOSYS);
368
369 if (ret < 0)
370 return ret;
371
372 ie = &st->index_entries[ret];
373 ret = avio_seek(s->pb, ie->pos, SEEK_SET);
374 if (ret < 0)
375 return ret;
376
377 while (!s->pb->eof_reached) {
378 header = (header << 8) + avio_r8(s->pb);
379 if (ff_mpa_check_header(header) >= 0) {
380 ff_update_cur_dts(s, st, ie->timestamp);
381 ret = avio_seek(s->pb, -4, SEEK_CUR);
382 return (ret >= 0) ? 0 : ret;
383 }
384 }
385
386 return AVERROR_EOF;
387 }
388
389 AVInputFormat ff_mp3_demuxer = {
390 .name = "mp3",
391 .long_name = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"),
392 .read_probe = mp3_read_probe,
393 .read_header = mp3_read_header,
394 .read_packet = mp3_read_packet,
395 .read_seek = mp3_seek,
396 .priv_data_size = sizeof(MP3DecContext),
397 .flags = AVFMT_GENERIC_INDEX,
398 .extensions = "mp2,mp3,m2a,mpa", /* XXX: use probe */
399 };