FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
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 FFmpeg.
6  *
7  * FFmpeg 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  * FFmpeg 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 FFmpeg; 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/opt.h"
23 #include "libavutil/avstring.h"
24 #include "libavutil/intreadwrite.h"
25 #include "libavutil/dict.h"
26 #include "libavutil/mathematics.h"
27 #include "avformat.h"
28 #include "internal.h"
29 #include "id3v2.h"
30 #include "id3v1.h"
32 
33 #define XING_FLAG_FRAMES 0x01
34 #define XING_FLAG_SIZE 0x02
35 #define XING_FLAG_TOC 0x04
36 
37 #define XING_TOC_COUNT 100
38 
39 typedef struct {
40  AVClass *class;
41  int64_t filesize;
42  int64_t header_filesize;
43  int xing_toc;
44  int start_pad;
45  int end_pad;
46  int usetoc;
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  const 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  if(ff_mpa_check_header(AV_RB32(buf2)))
71  continue;
72 
73  for(frames = 0; buf2 < end; frames++) {
74  header = AV_RB32(buf2);
75  fsize = avpriv_mpa_decode_header(&avctx, header, &sample_rate, &sample_rate, &sample_rate, &sample_rate);
76  if(fsize < 0)
77  break;
78  buf2 += fsize;
79  }
80  max_frames = FFMAX(max_frames, frames);
81  if(buf == buf0)
82  first_frames= frames;
83  }
84  // keep this in sync with ac3 probe, both need to avoid
85  // issues with MPEG-files!
86  if (first_frames>=4) return AVPROBE_SCORE_EXTENSION + 1;
87  else if(max_frames>200)return AVPROBE_SCORE_EXTENSION;
88  else if(max_frames>=4) return AVPROBE_SCORE_EXTENSION / 2;
89  else if(ff_id3v2_match(buf0, ID3v2_DEFAULT_MAGIC) && 2*ff_id3v2_tag_len(buf0) >= p->buf_size)
91  else if(max_frames>=1) return 1;
92  else return 0;
93 //mpegps_mp3_unrecognized_format.mpg has max_frames=3
94 }
95 
96 static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration)
97 {
98  int i;
99  MP3DecContext *mp3 = s->priv_data;
100  int fill_index = mp3->usetoc && duration > 0;
101 
102  if (!filesize &&
103  !(filesize = avio_size(s->pb))) {
104  av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n");
105  fill_index = 0;
106  }
107 
108  for (i = 0; i < XING_TOC_COUNT; i++) {
109  uint8_t b = avio_r8(s->pb);
110  if (fill_index)
112  av_rescale(b, filesize, 256),
113  av_rescale(i, duration, XING_TOC_COUNT),
114  0, 0, AVINDEX_KEYFRAME);
115  }
116  if (fill_index)
117  mp3->xing_toc = 1;
118 }
119 
120 /**
121  * Try to find Xing/Info/VBRI tags and compute duration from info therein
122  */
123 static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base)
124 {
125  MP3DecContext *mp3 = s->priv_data;
126  uint32_t v, spf;
127  unsigned frames = 0; /* Total number of frames in file */
128  unsigned size = 0; /* Total number of bytes in the stream */
129  static const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
131  int vbrtag_size = 0;
132  int is_cbr;
133 
134  v = avio_rb32(s->pb);
135  if(ff_mpa_check_header(v) < 0)
136  return -1;
137 
138  if (avpriv_mpegaudio_decode_header(&c, v) == 0)
139  vbrtag_size = c.frame_size;
140  if(c.layer != 3)
141  return -1;
142 
143  spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */
144 
145  /* Check for Xing / Info tag */
146  avio_skip(s->pb, xing_offtbl[c.lsf == 1][c.nb_channels == 1]);
147  v = avio_rb32(s->pb);
148  is_cbr = v == MKBETAG('I', 'n', 'f', 'o');
149  if (v == MKBETAG('X', 'i', 'n', 'g') || is_cbr) {
150  v = avio_rb32(s->pb);
151  if(v & XING_FLAG_FRAMES)
152  frames = avio_rb32(s->pb);
153  if(v & XING_FLAG_SIZE)
154  size = avio_rb32(s->pb);
155  if (v & XING_FLAG_TOC)
156  read_xing_toc(s, size, av_rescale_q(frames, (AVRational){spf, c.sample_rate},
157  st->time_base));
158  if(v & 8)
159  avio_skip(s->pb, 4);
160 
161  v = avio_rb32(s->pb);
162  if(v == MKBETAG('L', 'A', 'M', 'E') || v == MKBETAG('L', 'a', 'v', 'f')) {
163  avio_skip(s->pb, 21-4);
164  v= avio_rb24(s->pb);
165  mp3->start_pad = v>>12;
166  mp3-> end_pad = v&4095;
167  st->skip_samples = mp3->start_pad + 528 + 1;
168  if (!st->start_time)
170  (AVRational){1, c.sample_rate},
171  st->time_base);
172  av_log(s, AV_LOG_DEBUG, "pad %d %d\n", mp3->start_pad, mp3-> end_pad);
173  }
174  }
175 
176  /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */
177  avio_seek(s->pb, base + 4 + 32, SEEK_SET);
178  v = avio_rb32(s->pb);
179  if(v == MKBETAG('V', 'B', 'R', 'I')) {
180  /* Check tag version */
181  if(avio_rb16(s->pb) == 1) {
182  /* skip delay and quality */
183  avio_skip(s->pb, 4);
184  size = avio_rb32(s->pb);
185  frames = avio_rb32(s->pb);
186  }
187  }
188 
189  if(!frames && !size)
190  return -1;
191 
192  /* Skip the vbr tag frame */
193  avio_seek(s->pb, base + vbrtag_size, SEEK_SET);
194 
195  if(frames)
196  st->duration = av_rescale_q(frames, (AVRational){spf, c.sample_rate},
197  st->time_base);
198  if (size && frames && !is_cbr)
199  st->codec->bit_rate = av_rescale(size, 8 * c.sample_rate, frames * (int64_t)spf);
200 
201  mp3->is_cbr = is_cbr;
202  mp3->header_filesize = size;
203 
204  return 0;
205 }
206 
208 {
209  MP3DecContext *mp3 = s->priv_data;
210  AVStream *st;
211  int64_t off;
212 
213  st = avformat_new_stream(s, NULL);
214  if (!st)
215  return AVERROR(ENOMEM);
216 
220  st->start_time = 0;
221 
222  // lcm of all mp3 sample rates
223  avpriv_set_pts_info(st, 64, 1, 14112000);
224 
225  s->pb->maxsize = -1;
226  off = avio_tell(s->pb);
227 
228  if (!av_dict_get(s->metadata, "", NULL, AV_DICT_IGNORE_SUFFIX))
229  ff_id3v1_read(s);
230 
231  if(s->pb->seekable)
232  mp3->filesize = avio_size(s->pb);
233 
234  if (mp3_parse_vbr_tags(s, st, off) < 0)
235  avio_seek(s->pb, off, SEEK_SET);
236 
237  /* the parameters will be extracted from the compressed bitstream */
238  return 0;
239 }
240 
241 #define MP3_PACKET_SIZE 1024
242 
244 {
245  MP3DecContext *mp3 = s->priv_data;
246  int ret, size;
247  int64_t pos;
248 
249  size= MP3_PACKET_SIZE;
250  pos = avio_tell(s->pb);
251  if(mp3->filesize > ID3v1_TAG_SIZE && pos < mp3->filesize)
252  size= FFMIN(size, mp3->filesize - pos);
253 
254  ret= av_get_packet(s->pb, pkt, size);
255  if (ret <= 0) {
256  if(ret<0)
257  return ret;
258  return AVERROR_EOF;
259  }
260 
261  pkt->flags &= ~AV_PKT_FLAG_CORRUPT;
262  pkt->stream_index = 0;
263 
264  if (ret >= ID3v1_TAG_SIZE &&
265  memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0)
266  ret -= ID3v1_TAG_SIZE;
267 
268  /* note: we need to modify the packet size here to handle the last
269  packet */
270  pkt->size = ret;
271  return ret;
272 }
273 
274 static int check(AVFormatContext *s, int64_t pos)
275 {
276  int64_t ret = avio_seek(s->pb, pos, SEEK_SET);
277  unsigned header;
278  MPADecodeHeader sd;
279  if (ret < 0)
280  return ret;
281  header = avio_rb32(s->pb);
282  if (ff_mpa_check_header(header) < 0)
283  return -1;
284  if (avpriv_mpegaudio_decode_header(&sd, header) == 1)
285  return -1;
286  return sd.frame_size;
287 }
288 
289 static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp,
290  int flags)
291 {
292  MP3DecContext *mp3 = s->priv_data;
293  AVIndexEntry *ie, ie1;
294  AVStream *st = s->streams[0];
295  int64_t ret = av_index_search_timestamp(st, timestamp, flags);
296  int i, j;
297  int dir = (flags&AVSEEK_FLAG_BACKWARD) ? -1 : 1;
298 
299  if (mp3->is_cbr && st->duration > 0 && mp3->header_filesize > s->data_offset) {
300  int64_t filesize = avio_size(s->pb);
301  int64_t duration;
302  if (filesize <= s->data_offset)
303  filesize = mp3->header_filesize;
304  filesize -= s->data_offset;
305  duration = av_rescale(st->duration, filesize, mp3->header_filesize - s->data_offset);
306  ie = &ie1;
307  timestamp = av_clip64(timestamp, 0, duration);
308  ie->timestamp = timestamp;
309  ie->pos = av_rescale(timestamp, filesize, duration) + s->data_offset;
310  } else if (mp3->xing_toc) {
311  if (ret < 0)
312  return ret;
313 
314  ie = &st->index_entries[ret];
315  } else {
316  st->skip_samples = timestamp <= 0 ? mp3->start_pad + 528 + 1 : 0;
317 
318  return -1;
319  }
320 
321  if (dir < 0)
322  avio_seek(s->pb, FFMAX(ie->pos - 4096, 0), SEEK_SET);
323  ret = avio_seek(s->pb, ie->pos, SEEK_SET);
324  if (ret < 0)
325  return ret;
326 
327 #define MIN_VALID 3
328  for(i=0; i<4096; i++) {
329  int64_t pos = ie->pos + i*dir;
330  for(j=0; j<MIN_VALID; j++) {
331  ret = check(s, pos);
332  if(ret < 0)
333  break;
334  pos += ret;
335  }
336  if(j==MIN_VALID)
337  break;
338  }
339  if(j!=MIN_VALID)
340  i=0;
341 
342  ret = avio_seek(s->pb, ie->pos + i*dir, SEEK_SET);
343  if (ret < 0)
344  return ret;
345  ff_update_cur_dts(s, st, ie->timestamp);
346  st->skip_samples = ie->timestamp <= 0 ? mp3->start_pad + 528 + 1 : 0;
347  return 0;
348 }
349 
350 static const AVOption options[] = {
351  { "usetoc", "use table of contents", offsetof(MP3DecContext, usetoc), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 1, AV_OPT_FLAG_DECODING_PARAM},
352  { NULL },
353 };
354 
355 static const AVClass demuxer_class = {
356  .class_name = "mp3",
357  .item_name = av_default_item_name,
358  .option = options,
359  .version = LIBAVUTIL_VERSION_INT,
360  .category = AV_CLASS_CATEGORY_DEMUXER,
361 };
362 
364  .name = "mp3",
365  .long_name = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"),
366  .read_probe = mp3_read_probe,
367  .read_header = mp3_read_header,
368  .read_packet = mp3_read_packet,
369  .read_seek = mp3_seek,
370  .priv_data_size = sizeof(MP3DecContext),
372  .extensions = "mp2,mp3,m2a,mpa", /* XXX: use probe */
373  .priv_class = &demuxer_class,
374 };