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 
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>=4) return AVPROBE_SCORE_EXTENSION + 1;
85  else if(max_frames>200)return AVPROBE_SCORE_EXTENSION;
86  else if(max_frames>=4) return AVPROBE_SCORE_EXTENSION / 2;
87  else if(ff_id3v2_match(buf0, ID3v2_DEFAULT_MAGIC) && 2*ff_id3v2_tag_len(buf0) >= p->buf_size)
89  else if(max_frames>=1) return 1;
90  else return 0;
91 //mpegps_mp3_unrecognized_format.mpg has max_frames=3
92 }
93 
94 static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration)
95 {
96  int i;
97  MP3DecContext *mp3 = s->priv_data;
98  int fill_index = mp3->usetoc && duration > 0;
99 
100  if (!filesize &&
101  !(filesize = avio_size(s->pb))) {
102  av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n");
103  fill_index = 0;
104  }
105 
106  for (i = 0; i < XING_TOC_COUNT; i++) {
107  uint8_t b = avio_r8(s->pb);
108  if (fill_index)
110  av_rescale(b, filesize, 256),
111  av_rescale(i, duration, XING_TOC_COUNT),
112  0, 0, AVINDEX_KEYFRAME);
113  }
114  if (fill_index)
115  mp3->xing_toc = 1;
116 }
117 
118 /**
119  * Try to find Xing/Info/VBRI tags and compute duration from info therein
120  */
121 static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base)
122 {
123  MP3DecContext *mp3 = s->priv_data;
124  uint32_t v, spf;
125  unsigned frames = 0; /* Total number of frames in file */
126  unsigned size = 0; /* Total number of bytes in the stream */
127  static const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
129  int vbrtag_size = 0;
130  int is_cbr;
131 
132  v = avio_rb32(s->pb);
133  if(ff_mpa_check_header(v) < 0)
134  return -1;
135 
136  if (avpriv_mpegaudio_decode_header(&c, v) == 0)
137  vbrtag_size = c.frame_size;
138  if(c.layer != 3)
139  return -1;
140 
141  spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */
142 
143  /* Check for Xing / Info tag */
144  avio_skip(s->pb, xing_offtbl[c.lsf == 1][c.nb_channels == 1]);
145  v = avio_rb32(s->pb);
146  is_cbr = v == MKBETAG('I', 'n', 'f', 'o');
147  if (v == MKBETAG('X', 'i', 'n', 'g') || is_cbr) {
148  v = avio_rb32(s->pb);
149  if(v & XING_FLAG_FRAMES)
150  frames = avio_rb32(s->pb);
151  if(v & XING_FLAG_SIZE)
152  size = avio_rb32(s->pb);
153  if (v & XING_FLAG_TOC)
154  read_xing_toc(s, size, av_rescale_q(frames, (AVRational){spf, c.sample_rate},
155  st->time_base));
156  if(v & 8)
157  avio_skip(s->pb, 4);
158 
159  v = avio_rb32(s->pb);
160  if(v == MKBETAG('L', 'A', 'M', 'E') || v == MKBETAG('L', 'a', 'v', 'f')) {
161  avio_skip(s->pb, 21-4);
162  v= avio_rb24(s->pb);
163  mp3->start_pad = v>>12;
164  mp3-> end_pad = v&4095;
165  st->skip_samples = mp3->start_pad + 528 + 1;
166  av_log(s, AV_LOG_DEBUG, "pad %d %d\n", mp3->start_pad, mp3-> end_pad);
167  }
168  }
169 
170  /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */
171  avio_seek(s->pb, base + 4 + 32, SEEK_SET);
172  v = avio_rb32(s->pb);
173  if(v == MKBETAG('V', 'B', 'R', 'I')) {
174  /* Check tag version */
175  if(avio_rb16(s->pb) == 1) {
176  /* skip delay and quality */
177  avio_skip(s->pb, 4);
178  size = avio_rb32(s->pb);
179  frames = avio_rb32(s->pb);
180  }
181  }
182 
183  if(!frames && !size)
184  return -1;
185 
186  /* Skip the vbr tag frame */
187  avio_seek(s->pb, base + vbrtag_size, SEEK_SET);
188 
189  if(frames)
190  st->duration = av_rescale_q(frames, (AVRational){spf, c.sample_rate},
191  st->time_base);
192  if (size && frames && !is_cbr)
193  st->codec->bit_rate = av_rescale(size, 8 * c.sample_rate, frames * (int64_t)spf);
194 
195  mp3->is_cbr = is_cbr;
196  mp3->header_filesize = size;
197 
198  return 0;
199 }
200 
202 {
203  MP3DecContext *mp3 = s->priv_data;
204  AVStream *st;
205  int64_t off;
206 
207  st = avformat_new_stream(s, NULL);
208  if (!st)
209  return AVERROR(ENOMEM);
210 
214  st->start_time = 0;
215 
216  // lcm of all mp3 sample rates
217  avpriv_set_pts_info(st, 64, 1, 14112000);
218 
219  s->pb->maxsize = -1;
220  off = avio_tell(s->pb);
221 
222  if (!av_dict_get(s->metadata, "", NULL, AV_DICT_IGNORE_SUFFIX))
223  ff_id3v1_read(s);
224 
225  if(s->pb->seekable)
226  mp3->filesize = avio_size(s->pb);
227 
228  if (mp3_parse_vbr_tags(s, st, off) < 0)
229  avio_seek(s->pb, off, SEEK_SET);
230 
231  /* the parameters will be extracted from the compressed bitstream */
232  return 0;
233 }
234 
235 #define MP3_PACKET_SIZE 1024
236 
238 {
239  MP3DecContext *mp3 = s->priv_data;
240  int ret, size;
241  int64_t pos;
242 
243  size= MP3_PACKET_SIZE;
244  pos = avio_tell(s->pb);
245  if(mp3->filesize > ID3v1_TAG_SIZE && pos < mp3->filesize)
246  size= FFMIN(size, mp3->filesize - pos);
247 
248  ret= av_get_packet(s->pb, pkt, size);
249  if (ret <= 0) {
250  if(ret<0)
251  return ret;
252  return AVERROR_EOF;
253  }
254 
255  pkt->flags &= ~AV_PKT_FLAG_CORRUPT;
256  pkt->stream_index = 0;
257 
258  if (ret >= ID3v1_TAG_SIZE &&
259  memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0)
260  ret -= ID3v1_TAG_SIZE;
261 
262  /* note: we need to modify the packet size here to handle the last
263  packet */
264  pkt->size = ret;
265  return ret;
266 }
267 
268 static int check(AVFormatContext *s, int64_t pos)
269 {
270  int64_t ret = avio_seek(s->pb, pos, SEEK_SET);
271  unsigned header;
272  MPADecodeHeader sd;
273  if (ret < 0)
274  return ret;
275  header = avio_rb32(s->pb);
276  if (ff_mpa_check_header(header) < 0)
277  return -1;
278  if (avpriv_mpegaudio_decode_header(&sd, header) == 1)
279  return -1;
280  return sd.frame_size;
281 }
282 
283 static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp,
284  int flags)
285 {
286  MP3DecContext *mp3 = s->priv_data;
287  AVIndexEntry *ie, ie1;
288  AVStream *st = s->streams[0];
289  int64_t ret = av_index_search_timestamp(st, timestamp, flags);
290  int i, j;
291  int dir = (flags&AVSEEK_FLAG_BACKWARD) ? -1 : 1;
292 
293  if (mp3->is_cbr && st->duration > 0 && mp3->header_filesize > s->data_offset) {
294  int64_t filesize = avio_size(s->pb);
295  int64_t duration;
296  if (filesize <= s->data_offset)
297  filesize = mp3->header_filesize;
298  filesize -= s->data_offset;
299  duration = av_rescale(st->duration, filesize, mp3->header_filesize - s->data_offset);
300  ie = &ie1;
301  timestamp = av_clip64(timestamp, 0, duration);
302  ie->timestamp = timestamp;
303  ie->pos = av_rescale(timestamp, filesize, duration) + s->data_offset;
304  } else if (mp3->xing_toc) {
305  if (ret < 0)
306  return ret;
307 
308  ie = &st->index_entries[ret];
309  } else {
310  st->skip_samples = timestamp <= 0 ? mp3->start_pad + 528 + 1 : 0;
311 
312  return -1;
313  }
314 
315  ret = avio_seek(s->pb, ie->pos, SEEK_SET);
316  if (ret < 0)
317  return ret;
318 
319 #define MIN_VALID 3
320  for(i=0; i<4096; i++) {
321  int64_t pos = ie->pos + i*dir;
322  for(j=0; j<MIN_VALID; j++) {
323  ret = check(s, pos);
324  if(ret < 0)
325  break;
326  pos += ret;
327  }
328  if(j==MIN_VALID)
329  break;
330  }
331  if(j!=MIN_VALID)
332  i=0;
333 
334  ret = avio_seek(s->pb, ie->pos + i*dir, SEEK_SET);
335  if (ret < 0)
336  return ret;
337  ff_update_cur_dts(s, st, ie->timestamp);
338  st->skip_samples = ie->timestamp <= 0 ? mp3->start_pad + 528 + 1 : 0;
339  return 0;
340 }
341 
342 static const AVOption options[] = {
343  { "usetoc", "use table of contents", offsetof(MP3DecContext, usetoc), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 1, AV_OPT_FLAG_DECODING_PARAM},
344  { NULL },
345 };
346 
347 static const AVClass demuxer_class = {
348  .class_name = "mp3",
349  .item_name = av_default_item_name,
350  .option = options,
351  .version = LIBAVUTIL_VERSION_INT,
352  .category = AV_CLASS_CATEGORY_DEMUXER,
353 };
354 
356  .name = "mp3",
357  .long_name = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"),
358  .read_probe = mp3_read_probe,
359  .read_header = mp3_read_header,
360  .read_packet = mp3_read_packet,
361  .read_seek = mp3_seek,
362  .priv_data_size = sizeof(MP3DecContext),
364  .extensions = "mp2,mp3,m2a,mpa", /* XXX: use probe */
365  .priv_class = &demuxer_class,
366 };