Commit | Line | Data |
---|---|---|
2ba45a60 DM |
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/crc.h" | |
26 | #include "libavutil/dict.h" | |
27 | #include "libavutil/mathematics.h" | |
28 | #include "avformat.h" | |
29 | #include "internal.h" | |
30 | #include "avio_internal.h" | |
31 | #include "id3v2.h" | |
32 | #include "id3v1.h" | |
33 | #include "replaygain.h" | |
34 | ||
35 | #include "libavcodec/mpegaudiodecheader.h" | |
36 | ||
37 | #define XING_FLAG_FRAMES 0x01 | |
38 | #define XING_FLAG_SIZE 0x02 | |
39 | #define XING_FLAG_TOC 0x04 | |
40 | ||
41 | #define XING_TOC_COUNT 100 | |
42 | ||
43 | typedef struct { | |
44 | AVClass *class; | |
45 | int64_t filesize; | |
46 | int xing_toc; | |
47 | int start_pad; | |
48 | int end_pad; | |
49 | int usetoc; | |
50 | unsigned frames; /* Total number of frames in file */ | |
51 | unsigned header_filesize; /* Total number of bytes in the stream */ | |
52 | int is_cbr; | |
53 | } MP3DecContext; | |
54 | ||
55 | /* mp3 read */ | |
56 | ||
57 | static int mp3_read_probe(AVProbeData *p) | |
58 | { | |
59 | int max_frames, first_frames = 0; | |
60 | int fsize, frames, sample_rate; | |
61 | uint32_t header; | |
62 | const uint8_t *buf, *buf0, *buf2, *end; | |
63 | AVCodecContext avctx; | |
64 | ||
65 | buf0 = p->buf; | |
66 | end = p->buf + p->buf_size - sizeof(uint32_t); | |
67 | while(buf0 < end && !*buf0) | |
68 | buf0++; | |
69 | ||
70 | max_frames = 0; | |
71 | buf = buf0; | |
72 | ||
73 | for(; buf < end; buf= buf2+1) { | |
74 | buf2 = buf; | |
75 | if(ff_mpa_check_header(AV_RB32(buf2))) | |
76 | continue; | |
77 | ||
78 | for(frames = 0; buf2 < end; frames++) { | |
79 | header = AV_RB32(buf2); | |
80 | fsize = avpriv_mpa_decode_header(&avctx, header, &sample_rate, &sample_rate, &sample_rate, &sample_rate); | |
81 | if(fsize < 0) | |
82 | break; | |
83 | buf2 += fsize; | |
84 | } | |
85 | max_frames = FFMAX(max_frames, frames); | |
86 | if(buf == buf0) | |
87 | first_frames= frames; | |
88 | } | |
89 | // keep this in sync with ac3 probe, both need to avoid | |
90 | // issues with MPEG-files! | |
91 | if (first_frames>=4) return AVPROBE_SCORE_EXTENSION + 1; | |
92 | else if(max_frames>200)return AVPROBE_SCORE_EXTENSION; | |
93 | else if(max_frames>=4 && max_frames >= p->buf_size/10000) return AVPROBE_SCORE_EXTENSION / 2; | |
94 | else if(ff_id3v2_match(buf0, ID3v2_DEFAULT_MAGIC) && 2*ff_id3v2_tag_len(buf0) >= p->buf_size) | |
95 | return p->buf_size < PROBE_BUF_MAX ? AVPROBE_SCORE_EXTENSION / 4 : AVPROBE_SCORE_EXTENSION - 2; | |
96 | else if(max_frames>=1 && max_frames >= p->buf_size/10000) return 1; | |
97 | else return 0; | |
98 | //mpegps_mp3_unrecognized_format.mpg has max_frames=3 | |
99 | } | |
100 | ||
101 | static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration) | |
102 | { | |
103 | int i; | |
104 | MP3DecContext *mp3 = s->priv_data; | |
105 | int fill_index = mp3->usetoc && duration > 0; | |
106 | ||
107 | if (!filesize && | |
108 | !(filesize = avio_size(s->pb))) { | |
109 | av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n"); | |
110 | fill_index = 0; | |
111 | } | |
112 | ||
113 | for (i = 0; i < XING_TOC_COUNT; i++) { | |
114 | uint8_t b = avio_r8(s->pb); | |
115 | if (fill_index) | |
116 | av_add_index_entry(s->streams[0], | |
117 | av_rescale(b, filesize, 256), | |
118 | av_rescale(i, duration, XING_TOC_COUNT), | |
119 | 0, 0, AVINDEX_KEYFRAME); | |
120 | } | |
121 | if (fill_index) | |
122 | mp3->xing_toc = 1; | |
123 | } | |
124 | ||
125 | static void mp3_parse_info_tag(AVFormatContext *s, AVStream *st, | |
126 | MPADecodeHeader *c, uint32_t spf) | |
127 | { | |
128 | #define LAST_BITS(k, n) ((k) & ((1 << (n)) - 1)) | |
129 | #define MIDDLE_BITS(k, m, n) LAST_BITS((k) >> (m), ((n) - (m))) | |
130 | ||
131 | uint16_t crc; | |
132 | uint32_t v; | |
133 | ||
134 | char version[10]; | |
135 | ||
136 | uint32_t peak = 0; | |
137 | int32_t r_gain = INT32_MIN, a_gain = INT32_MIN; | |
138 | ||
139 | MP3DecContext *mp3 = s->priv_data; | |
140 | static const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}}; | |
141 | uint64_t fsize = avio_size(s->pb); | |
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 | mp3->is_cbr = v == MKBETAG('I', 'n', 'f', 'o'); | |
147 | if (v != MKBETAG('X', 'i', 'n', 'g') && !mp3->is_cbr) | |
148 | return; | |
149 | ||
150 | v = avio_rb32(s->pb); | |
151 | if (v & XING_FLAG_FRAMES) | |
152 | mp3->frames = avio_rb32(s->pb); | |
153 | if (v & XING_FLAG_SIZE) | |
154 | mp3->header_filesize = avio_rb32(s->pb); | |
155 | if (fsize && mp3->header_filesize) { | |
156 | uint64_t min, delta; | |
157 | min = FFMIN(fsize, mp3->header_filesize); | |
158 | delta = FFMAX(fsize, mp3->header_filesize) - min; | |
159 | if (fsize > mp3->header_filesize && delta > min >> 4) { | |
160 | mp3->frames = 0; | |
161 | } else if (delta > min >> 4) { | |
162 | av_log(s, AV_LOG_WARNING, | |
163 | "filesize and duration do not match (growing file?)\n"); | |
164 | } | |
165 | } | |
166 | if (v & XING_FLAG_TOC) | |
167 | read_xing_toc(s, mp3->header_filesize, av_rescale_q(mp3->frames, | |
168 | (AVRational){spf, c->sample_rate}, | |
169 | st->time_base)); | |
170 | /* VBR quality */ | |
171 | if(v & 8) | |
172 | avio_skip(s->pb, 4); | |
173 | ||
174 | /* Encoder short version string */ | |
175 | memset(version, 0, sizeof(version)); | |
176 | avio_read(s->pb, version, 9); | |
177 | ||
178 | /* Info Tag revision + VBR method */ | |
179 | avio_r8(s->pb); | |
180 | ||
181 | /* Lowpass filter value */ | |
182 | avio_r8(s->pb); | |
183 | ||
184 | /* ReplayGain peak */ | |
185 | v = avio_rb32(s->pb); | |
186 | peak = av_rescale(v, 100000, 1 << 23); | |
187 | ||
188 | /* Radio ReplayGain */ | |
189 | v = avio_rb16(s->pb); | |
190 | ||
191 | if (MIDDLE_BITS(v, 13, 15) == 1) { | |
192 | r_gain = MIDDLE_BITS(v, 0, 8) * 10000; | |
193 | ||
194 | if (v & (1 << 9)) | |
195 | r_gain *= -1; | |
196 | } | |
197 | ||
198 | /* Audiophile ReplayGain */ | |
199 | v = avio_rb16(s->pb); | |
200 | ||
201 | if (MIDDLE_BITS(v, 13, 15) == 2) { | |
202 | a_gain = MIDDLE_BITS(v, 0, 8) * 10000; | |
203 | ||
204 | if (v & (1 << 9)) | |
205 | a_gain *= -1; | |
206 | } | |
207 | ||
208 | /* Encoding flags + ATH Type */ | |
209 | avio_r8(s->pb); | |
210 | ||
211 | /* if ABR {specified bitrate} else {minimal bitrate} */ | |
212 | avio_r8(s->pb); | |
213 | ||
214 | /* Encoder delays */ | |
215 | v= avio_rb24(s->pb); | |
216 | if(AV_RB32(version) == MKBETAG('L', 'A', 'M', 'E') | |
217 | || AV_RB32(version) == MKBETAG('L', 'a', 'v', 'f')) { | |
218 | ||
219 | mp3->start_pad = v>>12; | |
220 | mp3-> end_pad = v&4095; | |
221 | st->skip_samples = mp3->start_pad + 528 + 1; | |
222 | if (!st->start_time) | |
223 | st->start_time = av_rescale_q(st->skip_samples, | |
224 | (AVRational){1, c->sample_rate}, | |
225 | st->time_base); | |
226 | av_log(s, AV_LOG_DEBUG, "pad %d %d\n", mp3->start_pad, mp3-> end_pad); | |
227 | } | |
228 | ||
229 | /* Misc */ | |
230 | avio_r8(s->pb); | |
231 | ||
232 | /* MP3 gain */ | |
233 | avio_r8(s->pb); | |
234 | ||
235 | /* Preset and surround info */ | |
236 | avio_rb16(s->pb); | |
237 | ||
238 | /* Music length */ | |
239 | avio_rb32(s->pb); | |
240 | ||
241 | /* Music CRC */ | |
242 | avio_rb16(s->pb); | |
243 | ||
244 | /* Info Tag CRC */ | |
245 | crc = ffio_get_checksum(s->pb); | |
246 | v = avio_rb16(s->pb); | |
247 | ||
248 | if (v == crc) { | |
249 | ff_replaygain_export_raw(st, r_gain, peak, a_gain, 0); | |
250 | av_dict_set(&st->metadata, "encoder", version, 0); | |
251 | } | |
252 | } | |
253 | ||
254 | static void mp3_parse_vbri_tag(AVFormatContext *s, AVStream *st, int64_t base) | |
255 | { | |
256 | uint32_t v; | |
257 | MP3DecContext *mp3 = s->priv_data; | |
258 | ||
259 | /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */ | |
260 | avio_seek(s->pb, base + 4 + 32, SEEK_SET); | |
261 | v = avio_rb32(s->pb); | |
262 | if (v == MKBETAG('V', 'B', 'R', 'I')) { | |
263 | /* Check tag version */ | |
264 | if (avio_rb16(s->pb) == 1) { | |
265 | /* skip delay and quality */ | |
266 | avio_skip(s->pb, 4); | |
267 | mp3->header_filesize = avio_rb32(s->pb); | |
268 | mp3->frames = avio_rb32(s->pb); | |
269 | } | |
270 | } | |
271 | } | |
272 | ||
273 | /** | |
274 | * Try to find Xing/Info/VBRI tags and compute duration from info therein | |
275 | */ | |
276 | static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base) | |
277 | { | |
278 | uint32_t v, spf; | |
279 | MPADecodeHeader c; | |
280 | int vbrtag_size = 0; | |
281 | MP3DecContext *mp3 = s->priv_data; | |
282 | ||
283 | ffio_init_checksum(s->pb, ff_crcA001_update, 0); | |
284 | ||
285 | v = avio_rb32(s->pb); | |
286 | if(ff_mpa_check_header(v) < 0) | |
287 | return -1; | |
288 | ||
289 | if (avpriv_mpegaudio_decode_header(&c, v) == 0) | |
290 | vbrtag_size = c.frame_size; | |
291 | if(c.layer != 3) | |
292 | return -1; | |
293 | ||
294 | spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */ | |
295 | ||
296 | mp3->frames = 0; | |
297 | mp3->header_filesize = 0; | |
298 | ||
299 | mp3_parse_info_tag(s, st, &c, spf); | |
300 | mp3_parse_vbri_tag(s, st, base); | |
301 | ||
302 | if (!mp3->frames && !mp3->header_filesize) | |
303 | return -1; | |
304 | ||
305 | /* Skip the vbr tag frame */ | |
306 | avio_seek(s->pb, base + vbrtag_size, SEEK_SET); | |
307 | ||
308 | if (mp3->frames) | |
309 | st->duration = av_rescale_q(mp3->frames, (AVRational){spf, c.sample_rate}, | |
310 | st->time_base); | |
311 | if (mp3->header_filesize && mp3->frames && !mp3->is_cbr) | |
312 | st->codec->bit_rate = av_rescale(mp3->header_filesize, 8 * c.sample_rate, mp3->frames * (int64_t)spf); | |
313 | ||
314 | return 0; | |
315 | } | |
316 | ||
317 | static int mp3_read_header(AVFormatContext *s) | |
318 | { | |
319 | MP3DecContext *mp3 = s->priv_data; | |
320 | AVStream *st; | |
321 | int64_t off; | |
322 | int ret; | |
323 | ||
324 | st = avformat_new_stream(s, NULL); | |
325 | if (!st) | |
326 | return AVERROR(ENOMEM); | |
327 | ||
328 | st->codec->codec_type = AVMEDIA_TYPE_AUDIO; | |
329 | st->codec->codec_id = AV_CODEC_ID_MP3; | |
330 | st->need_parsing = AVSTREAM_PARSE_FULL_RAW; | |
331 | st->start_time = 0; | |
332 | ||
333 | // lcm of all mp3 sample rates | |
334 | avpriv_set_pts_info(st, 64, 1, 14112000); | |
335 | ||
336 | s->pb->maxsize = -1; | |
337 | off = avio_tell(s->pb); | |
338 | ||
339 | if (!av_dict_get(s->metadata, "", NULL, AV_DICT_IGNORE_SUFFIX)) | |
340 | ff_id3v1_read(s); | |
341 | ||
342 | if(s->pb->seekable) | |
343 | mp3->filesize = avio_size(s->pb); | |
344 | ||
345 | if (mp3_parse_vbr_tags(s, st, off) < 0) | |
346 | avio_seek(s->pb, off, SEEK_SET); | |
347 | ||
348 | ret = ff_replaygain_export(st, s->metadata); | |
349 | if (ret < 0) | |
350 | return ret; | |
351 | ||
352 | /* the parameters will be extracted from the compressed bitstream */ | |
353 | return 0; | |
354 | } | |
355 | ||
356 | #define MP3_PACKET_SIZE 1024 | |
357 | ||
358 | static int mp3_read_packet(AVFormatContext *s, AVPacket *pkt) | |
359 | { | |
360 | MP3DecContext *mp3 = s->priv_data; | |
361 | int ret, size; | |
362 | int64_t pos; | |
363 | ||
364 | size= MP3_PACKET_SIZE; | |
365 | pos = avio_tell(s->pb); | |
366 | if(mp3->filesize > ID3v1_TAG_SIZE && pos < mp3->filesize) | |
367 | size= FFMIN(size, mp3->filesize - pos); | |
368 | ||
369 | ret= av_get_packet(s->pb, pkt, size); | |
370 | if (ret <= 0) { | |
371 | if(ret<0) | |
372 | return ret; | |
373 | return AVERROR_EOF; | |
374 | } | |
375 | ||
376 | pkt->flags &= ~AV_PKT_FLAG_CORRUPT; | |
377 | pkt->stream_index = 0; | |
378 | ||
379 | if (ret >= ID3v1_TAG_SIZE && | |
380 | memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0) | |
381 | ret -= ID3v1_TAG_SIZE; | |
382 | ||
383 | /* note: we need to modify the packet size here to handle the last | |
384 | packet */ | |
385 | pkt->size = ret; | |
386 | return ret; | |
387 | } | |
388 | ||
389 | static int check(AVFormatContext *s, int64_t pos) | |
390 | { | |
391 | int64_t ret = avio_seek(s->pb, pos, SEEK_SET); | |
392 | unsigned header; | |
393 | MPADecodeHeader sd; | |
394 | if (ret < 0) | |
395 | return ret; | |
396 | header = avio_rb32(s->pb); | |
397 | if (ff_mpa_check_header(header) < 0) | |
398 | return -1; | |
399 | if (avpriv_mpegaudio_decode_header(&sd, header) == 1) | |
400 | return -1; | |
401 | return sd.frame_size; | |
402 | } | |
403 | ||
404 | static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp, | |
405 | int flags) | |
406 | { | |
407 | MP3DecContext *mp3 = s->priv_data; | |
408 | AVIndexEntry *ie, ie1; | |
409 | AVStream *st = s->streams[0]; | |
410 | int64_t ret = av_index_search_timestamp(st, timestamp, flags); | |
411 | int i, j; | |
412 | int dir = (flags&AVSEEK_FLAG_BACKWARD) ? -1 : 1; | |
413 | int64_t best_pos; | |
414 | int best_score; | |
415 | ||
416 | if ( mp3->is_cbr | |
417 | && st->duration > 0 | |
418 | && mp3->header_filesize > s->data_offset | |
419 | && mp3->frames) { | |
420 | int64_t filesize = avio_size(s->pb); | |
421 | int64_t duration; | |
422 | if (filesize <= s->data_offset) | |
423 | filesize = mp3->header_filesize; | |
424 | filesize -= s->data_offset; | |
425 | duration = av_rescale(st->duration, filesize, mp3->header_filesize - s->data_offset); | |
426 | ie = &ie1; | |
427 | timestamp = av_clip64(timestamp, 0, duration); | |
428 | ie->timestamp = timestamp; | |
429 | ie->pos = av_rescale(timestamp, filesize, duration) + s->data_offset; | |
430 | } else if (mp3->xing_toc) { | |
431 | if (ret < 0) | |
432 | return ret; | |
433 | ||
434 | ie = &st->index_entries[ret]; | |
435 | } else { | |
436 | st->skip_samples = timestamp <= 0 ? mp3->start_pad + 528 + 1 : 0; | |
437 | ||
438 | return -1; | |
439 | } | |
440 | ||
441 | avio_seek(s->pb, FFMAX(ie->pos - 4096, 0), SEEK_SET); | |
442 | ret = avio_seek(s->pb, ie->pos, SEEK_SET); | |
443 | if (ret < 0) | |
444 | return ret; | |
445 | ||
446 | #define MIN_VALID 3 | |
447 | best_pos = ie->pos; | |
448 | best_score = 999; | |
449 | for(i=0; i<4096; i++) { | |
450 | int64_t pos = ie->pos + (dir > 0 ? i - 1024 : -i); | |
451 | int64_t candidate = -1; | |
452 | int score = 999; | |
453 | for(j=0; j<MIN_VALID; j++) { | |
454 | ret = check(s, pos); | |
455 | if(ret < 0) | |
456 | break; | |
457 | if ((ie->pos - pos)*dir <= 0 && abs(MIN_VALID/2-j) < score) { | |
458 | candidate = pos; | |
459 | score = abs(MIN_VALID/2-j); | |
460 | } | |
461 | pos += ret; | |
462 | } | |
463 | if (best_score > score && j == MIN_VALID) { | |
464 | best_pos = candidate; | |
465 | best_score = score; | |
466 | if(score == 0) | |
467 | break; | |
468 | } | |
469 | } | |
470 | ||
471 | ret = avio_seek(s->pb, best_pos, SEEK_SET); | |
472 | if (ret < 0) | |
473 | return ret; | |
474 | ff_update_cur_dts(s, st, ie->timestamp); | |
475 | st->skip_samples = ie->timestamp <= 0 ? mp3->start_pad + 528 + 1 : 0; | |
476 | return 0; | |
477 | } | |
478 | ||
479 | static const AVOption options[] = { | |
480 | { "usetoc", "use table of contents", offsetof(MP3DecContext, usetoc), AV_OPT_TYPE_INT, {.i64 = -1}, -1, 1, AV_OPT_FLAG_DECODING_PARAM}, | |
481 | { NULL }, | |
482 | }; | |
483 | ||
484 | static const AVClass demuxer_class = { | |
485 | .class_name = "mp3", | |
486 | .item_name = av_default_item_name, | |
487 | .option = options, | |
488 | .version = LIBAVUTIL_VERSION_INT, | |
489 | .category = AV_CLASS_CATEGORY_DEMUXER, | |
490 | }; | |
491 | ||
492 | AVInputFormat ff_mp3_demuxer = { | |
493 | .name = "mp3", | |
494 | .long_name = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"), | |
495 | .read_probe = mp3_read_probe, | |
496 | .read_header = mp3_read_header, | |
497 | .read_packet = mp3_read_packet, | |
498 | .read_seek = mp3_seek, | |
499 | .priv_data_size = sizeof(MP3DecContext), | |
500 | .flags = AVFMT_GENERIC_INDEX, | |
501 | .extensions = "mp2,mp3,m2a,mpa", /* XXX: use probe */ | |
502 | .priv_class = &demuxer_class, | |
503 | }; |