Libav
mp3dec.c
Go to the documentation of this file.
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 
35 
36 #define XING_FLAG_FRAMES 0x01
37 #define XING_FLAG_SIZE 0x02
38 #define XING_FLAG_TOC 0x04
39 #define XING_FLAC_QSCALE 0x08
40 
41 #define XING_TOC_COUNT 100
42 
43 typedef struct MP3DecContext {
44  int xing_toc;
45  unsigned frames; /* Total number of frames in file */
46  unsigned size; /* Total number of bytes in the stream */
47  int is_cbr;
49 
50 /* mp3 read */
51 
53 {
54  int max_frames, first_frames = 0;
55  int fsize, frames, sample_rate;
56  uint32_t header;
57  uint8_t *buf, *buf0, *buf2, *end;
58  AVCodecContext avctx;
59 
60  buf0 = p->buf;
61  end = p->buf + p->buf_size - sizeof(uint32_t);
62  while(buf0 < end && !*buf0)
63  buf0++;
64 
65  max_frames = 0;
66  buf = buf0;
67 
68  for(; buf < end; buf= buf2+1) {
69  buf2 = buf;
70 
71  for(frames = 0; buf2 < end; frames++) {
72  header = AV_RB32(buf2);
73  fsize = avpriv_mpa_decode_header(&avctx, header, &sample_rate, &sample_rate, &sample_rate, &sample_rate);
74  if(fsize < 0)
75  break;
76  buf2 += fsize;
77  }
78  max_frames = FFMAX(max_frames, frames);
79  if(buf == buf0)
80  first_frames= frames;
81  }
82  // keep this in sync with ac3 probe, both need to avoid
83  // issues with MPEG-files!
84  if (first_frames >= 10)
85  return AVPROBE_SCORE_EXTENSION + 5;
86  if (first_frames >= 4)
87  return AVPROBE_SCORE_EXTENSION + 1;
88 
89  if (max_frames) {
90  int pes = 0, i;
91  unsigned int code = -1;
92 
93 #define VIDEO_ID 0x000001e0
94 #define AUDIO_ID 0x000001c0
95  /* do a search for mpegps headers to be able to properly bias
96  * towards mpegps if we detect this stream as both. */
97  for (i = 0; i<p->buf_size; i++) {
98  code = (code << 8) + p->buf[i];
99  if ((code & 0xffffff00) == 0x100) {
100  if ((code & 0x1f0) == VIDEO_ID) pes++;
101  else if((code & 0x1e0) == AUDIO_ID) pes++;
102  }
103  }
104 
105  if (pes)
106  max_frames = (max_frames + pes - 1) / pes;
107  }
108  if (max_frames > 500) return AVPROBE_SCORE_EXTENSION;
109  else if (max_frames >= 4) return AVPROBE_SCORE_EXTENSION / 2;
110  else if (max_frames >= 1) return 1;
111  else return 0;
112 //mpegps_mp3_unrecognized_format.mpg has max_frames=3
113 }
114 
115 static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration)
116 {
117  int i;
118  MP3DecContext *mp3 = s->priv_data;
119 
120  if (!filesize &&
121  !(filesize = avio_size(s->pb))) {
122  av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n");
123  return;
124  }
125 
126  for (i = 0; i < XING_TOC_COUNT; i++) {
127  uint8_t b = avio_r8(s->pb);
128 
130  av_rescale(b, filesize, 256),
131  av_rescale(i, duration, XING_TOC_COUNT),
132  0, 0, AVINDEX_KEYFRAME);
133  }
134  mp3->xing_toc = 1;
135 }
136 
138  MPADecodeHeader *c, uint32_t spf)
139 {
140 #define LAST_BITS(k, n) ((k) & ((1 << (n)) - 1))
141 #define MIDDLE_BITS(k, m, n) LAST_BITS((k) >> (m), ((n) - (m)))
142 
143  uint16_t crc;
144  uint32_t v;
145 
146  char version[10];
147 
148  uint32_t peak = 0;
149  int32_t r_gain = INT32_MIN, a_gain = INT32_MIN;
150 
151  MP3DecContext *mp3 = s->priv_data;
152  const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
153 
154  /* Check for Xing / Info tag */
155  avio_skip(s->pb, xing_offtbl[c->lsf == 1][c->nb_channels == 1]);
156  v = avio_rb32(s->pb);
157  mp3->is_cbr = v == MKBETAG('I', 'n', 'f', 'o');
158  if (v != MKBETAG('X', 'i', 'n', 'g') && !mp3->is_cbr)
159  return;
160 
161  v = avio_rb32(s->pb);
162  if (v & XING_FLAG_FRAMES)
163  mp3->frames = avio_rb32(s->pb);
164  if (v & XING_FLAG_SIZE)
165  mp3->size = avio_rb32(s->pb);
166  if (v & XING_FLAG_TOC && mp3->frames)
167  read_xing_toc(s, mp3->size, av_rescale_q(mp3->frames,
168  (AVRational){spf, c->sample_rate},
169  st->time_base));
170 
171  /* VBR quality */
172  if (v & XING_FLAC_QSCALE)
173  avio_rb32(s->pb);
174 
175  /* Encoder short version string */
176  memset(version, 0, sizeof(version));
177  avio_read(s->pb, version, 9);
178 
179  /* Info Tag revision + VBR method */
180  avio_r8(s->pb);
181 
182  /* Lowpass filter value */
183  avio_r8(s->pb);
184 
185  /* ReplayGain peak */
186  v = avio_rb32(s->pb);
187  peak = av_rescale(v, 100000, 1 << 23);
188 
189  /* Radio ReplayGain */
190  v = avio_rb16(s->pb);
191 
192  if (MIDDLE_BITS(v, 13, 15) == 1) {
193  r_gain = MIDDLE_BITS(v, 0, 8) * 10000;
194 
195  if (v & (1 << 9))
196  r_gain *= -1;
197  }
198 
199  /* Audiophile ReplayGain */
200  v = avio_rb16(s->pb);
201 
202  if (MIDDLE_BITS(v, 13, 15) == 2) {
203  a_gain = MIDDLE_BITS(v, 0, 8) * 10000;
204 
205  if (v & (1 << 9))
206  a_gain *= -1;
207  }
208 
209  /* Encoding flags + ATH Type */
210  avio_r8(s->pb);
211 
212  /* if ABR {specified bitrate} else {minimal bitrate} */
213  avio_r8(s->pb);
214 
215  /* Encoder delays */
216  avio_rb24(s->pb);
217 
218  /* Misc */
219  avio_r8(s->pb);
220 
221  /* MP3 gain */
222  avio_r8(s->pb);
223 
224  /* Preset and surround info */
225  avio_rb16(s->pb);
226 
227  /* Music length */
228  avio_rb32(s->pb);
229 
230  /* Music CRC */
231  avio_rb16(s->pb);
232 
233  /* Info Tag CRC */
234  crc = ffio_get_checksum(s->pb);
235  v = avio_rb16(s->pb);
236 
237  if (v == crc) {
238  ff_replaygain_export_raw(st, r_gain, peak, a_gain, 0);
239  av_dict_set(&st->metadata, "encoder", version, 0);
240  }
241 }
242 
243 static void mp3_parse_vbri_tag(AVFormatContext *s, AVStream *st, int64_t base)
244 {
245  uint32_t v;
246  MP3DecContext *mp3 = s->priv_data;
247 
248  /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */
249  avio_seek(s->pb, base + 4 + 32, SEEK_SET);
250  v = avio_rb32(s->pb);
251  if (v == MKBETAG('V', 'B', 'R', 'I')) {
252  /* Check tag version */
253  if (avio_rb16(s->pb) == 1) {
254  /* skip delay and quality */
255  avio_skip(s->pb, 4);
256  mp3->size = avio_rb32(s->pb);
257  mp3->frames = avio_rb32(s->pb);
258  }
259  }
260 }
261 
265 static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base)
266 {
267  uint32_t v, spf;
268  MPADecodeHeader c;
269  int vbrtag_size = 0;
270  MP3DecContext *mp3 = s->priv_data;
271 
273 
274  v = avio_rb32(s->pb);
275  if(ff_mpa_check_header(v) < 0)
276  return -1;
277 
278  if (avpriv_mpegaudio_decode_header(&c, v) == 0)
279  vbrtag_size = c.frame_size;
280  if(c.layer != 3)
281  return -1;
282 
283  spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */
284 
285  mp3->frames = 0;
286  mp3->size = 0;
287 
288  mp3_parse_info_tag(s, st, &c, spf);
289  mp3_parse_vbri_tag(s, st, base);
290 
291  if (!mp3->frames && !mp3->size)
292  return -1;
293 
294  /* Skip the vbr tag frame */
295  avio_seek(s->pb, base + vbrtag_size, SEEK_SET);
296 
297  if (mp3->frames)
298  st->duration = av_rescale_q(mp3->frames, (AVRational){spf, c.sample_rate},
299  st->time_base);
300  if (mp3->size && mp3->frames && !mp3->is_cbr)
301  st->codec->bit_rate = av_rescale(mp3->size, 8 * c.sample_rate, mp3->frames * (int64_t)spf);
302 
303  return 0;
304 }
305 
307 {
308  AVStream *st;
309  int64_t off;
310  int ret;
311 
312  st = avformat_new_stream(s, NULL);
313  if (!st)
314  return AVERROR(ENOMEM);
315 
319  st->start_time = 0;
320 
321  // lcm of all mp3 sample rates
322  avpriv_set_pts_info(st, 64, 1, 14112000);
323 
324  off = avio_tell(s->pb);
325 
327  ff_id3v1_read(s);
328 
329  if (mp3_parse_vbr_tags(s, st, off) < 0)
330  avio_seek(s->pb, off, SEEK_SET);
331 
332  ret = ff_replaygain_export(st, s->metadata);
333  if (ret < 0)
334  return ret;
335 
336  /* the parameters will be extracted from the compressed bitstream */
337  return 0;
338 }
339 
340 #define MP3_PACKET_SIZE 1024
341 
343 {
344  int ret;
345 
346  ret = av_get_packet(s->pb, pkt, MP3_PACKET_SIZE);
347  if (ret < 0)
348  return ret;
349 
350  pkt->stream_index = 0;
351 
352  if (ret > ID3v1_TAG_SIZE &&
353  memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0)
354  ret -= ID3v1_TAG_SIZE;
355 
356  /* note: we need to modify the packet size here to handle the last
357  packet */
358  pkt->size = ret;
359  return ret;
360 }
361 
362 static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp,
363  int flags)
364 {
365  MP3DecContext *mp3 = s->priv_data;
366  AVIndexEntry *ie;
367  AVStream *st = s->streams[0];
368  int64_t ret = av_index_search_timestamp(st, timestamp, flags);
369  uint32_t header = 0;
370 
371  if (!mp3->xing_toc)
372  return AVERROR(ENOSYS);
373 
374  if (ret < 0)
375  return ret;
376 
377  ie = &st->index_entries[ret];
378  ret = avio_seek(s->pb, ie->pos, SEEK_SET);
379  if (ret < 0)
380  return ret;
381 
382  while (!s->pb->eof_reached) {
383  header = (header << 8) + avio_r8(s->pb);
384  if (ff_mpa_check_header(header) >= 0) {
385  ff_update_cur_dts(s, st, ie->timestamp);
386  ret = avio_seek(s->pb, -4, SEEK_CUR);
387  return (ret >= 0) ? 0 : ret;
388  }
389  }
390 
391  return AVERROR_EOF;
392 }
393 
395  .name = "mp3",
396  .long_name = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"),
397  .read_probe = mp3_read_probe,
398  .read_header = mp3_read_header,
399  .read_packet = mp3_read_packet,
400  .read_seek = mp3_seek,
401  .priv_data_size = sizeof(MP3DecContext),
403  .extensions = "mp2,mp3,m2a,mpa", /* XXX: use probe */
404 };
int64_t avio_size(AVIOContext *s)
Get the filesize.
Definition: aviobuf.c:241
int av_add_index_entry(AVStream *st, int64_t pos, int64_t timestamp, int size, int distance, int flags)
Add an index entry into a sorted list.
Definition: utils.c:1181
#define AV_LOG_WARNING
Something somehow does not look correct.
Definition: log.h:129
void avpriv_set_pts_info(AVStream *s, int pts_wrap_bits, unsigned int pts_num, unsigned int pts_den)
Set the time base and wrapping info for a given stream.
Definition: utils.c:2829
int64_t pos
Definition: avformat.h:660
void ff_id3v1_read(AVFormatContext *s)
Read an ID3v1 tag.
Definition: id3v1.c:227
int size
Definition: avcodec.h:974
int64_t avio_seek(AVIOContext *s, int64_t offset, int whence)
fseek() equivalent for AVIOContext.
Definition: aviobuf.c:186
AVIndexEntry * index_entries
Only used if the format does not support seeking natively.
Definition: avformat.h:878
static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration)
Definition: mp3dec.c:115
unsigned int avio_rb16(AVIOContext *s)
Definition: aviobuf.c:580
unsigned frames
Definition: mp3dec.c:45
static int64_t duration
Definition: avplay.c:246
Format I/O context.
Definition: avformat.h:922
void ff_update_cur_dts(AVFormatContext *s, AVStream *ref_st, int64_t timestamp)
Update cur_dts of all streams based on the given timestamp and AVStream.
Definition: utils.c:1104
Public dictionary API.
static int mp3_read_packet(AVFormatContext *s, AVPacket *pkt)
Definition: mp3dec.c:342
uint8_t
#define XING_FLAG_SIZE
Definition: mp3dec.c:37
unsigned int avio_rb32(AVIOContext *s)
Definition: aviobuf.c:595
static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp, int flags)
Definition: mp3dec.c:362
#define AV_RB32
Definition: intreadwrite.h:130
enum AVStreamParseType need_parsing
Definition: avformat.h:867
#define b
Definition: input.c:52
AVStream * avformat_new_stream(AVFormatContext *s, const AVCodec *c)
Add a new stream to a media file.
Definition: utils.c:2521
AVStream ** streams
A list of all streams in the file.
Definition: avformat.h:990
AVDictionaryEntry * av_dict_get(const AVDictionary *m, const char *key, const AVDictionaryEntry *prev, int flags)
Get a dictionary entry with matching key.
Definition: dict.c:38
uint8_t * data
Definition: avcodec.h:973
int avpriv_mpegaudio_decode_header(MPADecodeHeader *s, uint32_t header)
static int flags
Definition: log.c:44
static const uint8_t xing_offtbl[2][2]
Definition: mp3enc.c:109
#define AVERROR_EOF
End of file.
Definition: error.h:51
int av_get_packet(AVIOContext *s, AVPacket *pkt, int size)
Allocate and read the payload of a packet and initialize its fields with default values.
Definition: utils.c:117
static av_always_inline int64_t avio_tell(AVIOContext *s)
ftell() equivalent for AVIOContext.
Definition: avio.h:219
int ff_replaygain_export(AVStream *st, AVDictionary *metadata)
Parse replaygain tags and export them as per-stream side data.
Definition: replaygain.c:110
int avio_read(AVIOContext *s, unsigned char *buf, int size)
Read size bytes from AVIOContext into buf.
Definition: aviobuf.c:452
int64_t av_rescale_q(int64_t a, AVRational bq, AVRational cq)
Rescale a 64-bit integer by 2 rational numbers.
Definition: mathematics.c:129
#define VIDEO_ID
#define AVINDEX_KEYFRAME
Definition: avformat.h:662
AVDictionary * metadata
Metadata that applies to the whole file.
Definition: avformat.h:1130
int av_index_search_timestamp(AVStream *st, int64_t timestamp, int flags)
Get the index for a specific timestamp.
Definition: utils.c:1222
#define AVERROR(e)
Definition: error.h:43
int64_t timestamp
Definition: avformat.h:661
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification. ...
Definition: internal.h:150
static int mp3_read_header(AVFormatContext *s)
Definition: mp3dec.c:306
preferred ID for decoding MPEG audio layer 1, 2 or 3
Definition: avcodec.h:379
void av_log(void *avcl, int level, const char *fmt,...)
Definition: log.c:168
static int ff_mpa_check_header(uint32_t header)
#define FFMAX(a, b)
Definition: common.h:55
int avio_r8(AVIOContext *s)
Definition: aviobuf.c:443
int avpriv_mpa_decode_header(AVCodecContext *avctx, uint32_t head, int *sample_rate, int *channels, int *frame_size, int *bit_rate)
AVCodecContext * codec
Codec context associated with this stream.
Definition: avformat.h:718
int buf_size
Size of buf except extra allocated bytes.
Definition: avformat.h:398
unsigned char * buf
Buffer must have AVPROBE_PADDING_SIZE of extra allocated bytes filled with zero.
Definition: avformat.h:397
void ffio_init_checksum(AVIOContext *s, unsigned long(*update_checksum)(unsigned long c, const uint8_t *p, unsigned int len), unsigned long checksum)
Definition: aviobuf.c:431
int bit_rate
the average bitrate
Definition: avcodec.h:1114
unsigned int avio_rb24(AVIOContext *s)
Definition: aviobuf.c:588
int64_t av_rescale(int64_t a, int64_t b, int64_t c)
Rescale a 64-bit integer with rounding to nearest.
Definition: mathematics.c:116
#define MP3_PACKET_SIZE
Definition: mp3dec.c:340
static av_always_inline int64_t avio_skip(AVIOContext *s, int64_t offset)
Skip given number of bytes forward.
Definition: avio.h:210
static void mp3_parse_info_tag(AVFormatContext *s, AVStream *st, MPADecodeHeader *c, uint32_t spf)
Definition: mp3dec.c:137
int32_t
int xing_toc
Definition: mp3dec.c:44
AVDictionary * metadata
Definition: avformat.h:771
unsigned long ff_crcA001_update(unsigned long checksum, const uint8_t *buf, unsigned int len)
Definition: aviobuf.c:417
static void mp3_parse_vbri_tag(AVFormatContext *s, AVStream *st, int64_t base)
Definition: mp3dec.c:243
if(ac->has_optimized_func)
Stream structure.
Definition: avformat.h:699
static int mp3_read_probe(AVProbeData *p)
Definition: mp3dec.c:52
NULL
Definition: eval.c:55
enum AVMediaType codec_type
Definition: avcodec.h:1058
version
Definition: ffv1enc.c:1080
enum AVCodecID codec_id
Definition: avcodec.h:1067
AVIOContext * pb
I/O context.
Definition: avformat.h:964
main external API structure.
Definition: avcodec.h:1050
#define XING_FLAG_FRAMES
Definition: mp3dec.c:36
int av_dict_set(AVDictionary **pm, const char *key, const char *value, int flags)
Set the given entry in *pm, overwriting an existing entry.
Definition: dict.c:68
#define AVFMT_GENERIC_INDEX
Use generic index building code.
Definition: avformat.h:417
rational number numerator/denominator
Definition: rational.h:43
#define XING_FLAC_QSCALE
Definition: mp3dec.c:39
unsigned long ffio_get_checksum(AVIOContext *s)
Definition: aviobuf.c:423
#define AVPROBE_SCORE_EXTENSION
score for file extension
Definition: avformat.h:402
This structure contains the data a format has to probe a file.
Definition: avformat.h:395
#define AUDIO_ID
int64_t duration
Decoding: duration of the stream, in stream time base.
Definition: avformat.h:756
MPEG Audio header decoder.
int is_cbr
Definition: mp3dec.c:47
full parsing and repack
Definition: avformat.h:653
Main libavformat public API header.
int ff_replaygain_export_raw(AVStream *st, int32_t tg, uint32_t tp, int32_t ag, uint32_t ap)
Export already decoded replaygain values as per-stream side data.
Definition: replaygain.c:69
int64_t start_time
Decoding: pts of the first frame of the stream, in stream time base.
Definition: avformat.h:749
AVInputFormat ff_mp3_demuxer
Definition: mp3dec.c:394
#define MKBETAG(a, b, c, d)
Definition: common.h:239
#define XING_FLAG_TOC
Definition: mp3dec.c:38
#define MIDDLE_BITS(k, m, n)
unsigned size
Definition: mp3dec.c:46
int eof_reached
true if eof reached
Definition: avio.h:96
void * priv_data
Format private data.
Definition: avformat.h:950
#define XING_TOC_COUNT
Definition: mp3dec.c:41
static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base)
Try to find Xing/Info/VBRI tags and compute duration from info therein.
Definition: mp3dec.c:265
#define ID3v1_TAG_SIZE
Definition: id3v1.h:27
const char * name
A comma separated list of short names for the format.
Definition: avformat.h:525
#define AV_DICT_IGNORE_SUFFIX
Definition: dict.h:62
int stream_index
Definition: avcodec.h:975
AVRational time_base
This is the fundamental unit of time (in seconds) in terms of which frame timestamps are represented...
Definition: avformat.h:741
This structure stores compressed data.
Definition: avcodec.h:950