9da9aa8b1628876bc0b53400f2d8d50bb82e8303
[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/dict.h"
25 #include "libavutil/mathematics.h"
26 #include "avformat.h"
27 #include "internal.h"
28 #include "id3v2.h"
29 #include "id3v1.h"
30 #include "libavcodec/mpegaudiodecheader.h"
31
32 #define XING_FLAG_FRAMES 0x01
33 #define XING_FLAG_SIZE 0x02
34 #define XING_FLAG_TOC 0x04
35
36 #define XING_TOC_COUNT 100
37
38 typedef struct MP3DecContext {
39 int xing_toc;
40 } MP3DecContext;
41
42 /* mp3 read */
43
44 static int mp3_read_probe(AVProbeData *p)
45 {
46 int max_frames, first_frames = 0;
47 int fsize, frames, sample_rate;
48 uint32_t header;
49 uint8_t *buf, *buf0, *buf2, *end;
50 AVCodecContext avctx;
51
52 buf0 = p->buf;
53 end = p->buf + p->buf_size - sizeof(uint32_t);
54 while(buf0 < end && !*buf0)
55 buf0++;
56
57 max_frames = 0;
58 buf = buf0;
59
60 for(; buf < end; buf= buf2+1) {
61 buf2 = buf;
62
63 for(frames = 0; buf2 < end; frames++) {
64 header = AV_RB32(buf2);
65 fsize = avpriv_mpa_decode_header(&avctx, header, &sample_rate, &sample_rate, &sample_rate, &sample_rate);
66 if(fsize < 0)
67 break;
68 buf2 += fsize;
69 }
70 max_frames = FFMAX(max_frames, frames);
71 if(buf == buf0)
72 first_frames= frames;
73 }
74 // keep this in sync with ac3 probe, both need to avoid
75 // issues with MPEG-files!
76 if (first_frames >= 4) return AVPROBE_SCORE_MAX / 2 + 1;
77
78 if (max_frames) {
79 int pes = 0, i;
80 unsigned int code = -1;
81
82 #define VIDEO_ID 0x000001e0
83 #define AUDIO_ID 0x000001c0
84 /* do a search for mpegps headers to be able to properly bias
85 * towards mpegps if we detect this stream as both. */
86 for (i = 0; i<p->buf_size; i++) {
87 code = (code << 8) + p->buf[i];
88 if ((code & 0xffffff00) == 0x100) {
89 if ((code & 0x1f0) == VIDEO_ID) pes++;
90 else if((code & 0x1e0) == AUDIO_ID) pes++;
91 }
92 }
93
94 if (pes)
95 max_frames = (max_frames + pes - 1) / pes;
96 }
97 if (max_frames > 500) return AVPROBE_SCORE_MAX / 2;
98 else if (max_frames >= 4) return AVPROBE_SCORE_MAX / 4;
99 else if (max_frames >= 1) return 1;
100 else return 0;
101 //mpegps_mp3_unrecognized_format.mpg has max_frames=3
102 }
103
104 static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration)
105 {
106 int i;
107 MP3DecContext *mp3 = s->priv_data;
108
109 if (!filesize &&
110 !(filesize = avio_size(s->pb))) {
111 av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n");
112 return;
113 }
114
115 for (i = 0; i < XING_TOC_COUNT; i++) {
116 uint8_t b = avio_r8(s->pb);
117
118 av_add_index_entry(s->streams[0],
119 av_rescale(b, filesize, 256),
120 av_rescale(i, duration, XING_TOC_COUNT),
121 0, 0, AVINDEX_KEYFRAME);
122 }
123 mp3->xing_toc = 1;
124 }
125
126 /**
127 * Try to find Xing/Info/VBRI tags and compute duration from info therein
128 */
129 static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base)
130 {
131 uint32_t v, spf;
132 unsigned frames = 0; /* Total number of frames in file */
133 unsigned size = 0; /* Total number of bytes in the stream */
134 const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
135 MPADecodeHeader c;
136 int vbrtag_size = 0;
137 int is_cbr;
138
139 v = avio_rb32(s->pb);
140 if(ff_mpa_check_header(v) < 0)
141 return -1;
142
143 if (avpriv_mpegaudio_decode_header(&c, v) == 0)
144 vbrtag_size = c.frame_size;
145 if(c.layer != 3)
146 return -1;
147
148 spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */
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 is_cbr = v == MKBETAG('I', 'n', 'f', 'o');
154 if (v == MKBETAG('X', 'i', 'n', 'g') || is_cbr) {
155 v = avio_rb32(s->pb);
156 if(v & XING_FLAG_FRAMES)
157 frames = avio_rb32(s->pb);
158 if(v & XING_FLAG_SIZE)
159 size = avio_rb32(s->pb);
160 if (v & XING_FLAG_TOC && frames)
161 read_xing_toc(s, size, av_rescale_q(frames, (AVRational){spf, c.sample_rate},
162 st->time_base));
163 }
164
165 /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */
166 avio_seek(s->pb, base + 4 + 32, SEEK_SET);
167 v = avio_rb32(s->pb);
168 if(v == MKBETAG('V', 'B', 'R', 'I')) {
169 /* Check tag version */
170 if(avio_rb16(s->pb) == 1) {
171 /* skip delay and quality */
172 avio_skip(s->pb, 4);
173 size = avio_rb32(s->pb);
174 frames = avio_rb32(s->pb);
175 }
176 }
177
178 if(!frames && !size)
179 return -1;
180
181 /* Skip the vbr tag frame */
182 avio_seek(s->pb, base + vbrtag_size, SEEK_SET);
183
184 if(frames)
185 st->duration = av_rescale_q(frames, (AVRational){spf, c.sample_rate},
186 st->time_base);
187 if (size && frames && !is_cbr)
188 st->codec->bit_rate = av_rescale(size, 8 * c.sample_rate, frames * (int64_t)spf);
189
190 return 0;
191 }
192
193 static int mp3_read_header(AVFormatContext *s)
194 {
195 AVStream *st;
196 int64_t off;
197
198 st = avformat_new_stream(s, NULL);
199 if (!st)
200 return AVERROR(ENOMEM);
201
202 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
203 st->codec->codec_id = AV_CODEC_ID_MP3;
204 st->need_parsing = AVSTREAM_PARSE_FULL;
205 st->start_time = 0;
206
207 // lcm of all mp3 sample rates
208 avpriv_set_pts_info(st, 64, 1, 14112000);
209
210 off = avio_tell(s->pb);
211
212 if (!av_dict_get(s->metadata, "", NULL, AV_DICT_IGNORE_SUFFIX))
213 ff_id3v1_read(s);
214
215 if (mp3_parse_vbr_tags(s, st, off) < 0)
216 avio_seek(s->pb, off, SEEK_SET);
217
218 /* the parameters will be extracted from the compressed bitstream */
219 return 0;
220 }
221
222 #define MP3_PACKET_SIZE 1024
223
224 static int mp3_read_packet(AVFormatContext *s, AVPacket *pkt)
225 {
226 int ret;
227
228 ret = av_get_packet(s->pb, pkt, MP3_PACKET_SIZE);
229 if (ret < 0)
230 return ret;
231
232 pkt->stream_index = 0;
233
234 if (ret > ID3v1_TAG_SIZE &&
235 memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0)
236 ret -= ID3v1_TAG_SIZE;
237
238 /* note: we need to modify the packet size here to handle the last
239 packet */
240 pkt->size = ret;
241 return ret;
242 }
243
244 static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp,
245 int flags)
246 {
247 MP3DecContext *mp3 = s->priv_data;
248 AVIndexEntry *ie;
249 AVStream *st = s->streams[0];
250 int64_t ret = av_index_search_timestamp(st, timestamp, flags);
251 uint32_t header = 0;
252
253 if (!mp3->xing_toc)
254 return AVERROR(ENOSYS);
255
256 if (ret < 0)
257 return ret;
258
259 ie = &st->index_entries[ret];
260 ret = avio_seek(s->pb, ie->pos, SEEK_SET);
261 if (ret < 0)
262 return ret;
263
264 while (!s->pb->eof_reached) {
265 header = (header << 8) + avio_r8(s->pb);
266 if (ff_mpa_check_header(header) >= 0) {
267 ff_update_cur_dts(s, st, ie->timestamp);
268 ret = avio_seek(s->pb, -4, SEEK_CUR);
269 return (ret >= 0) ? 0 : ret;
270 }
271 }
272
273 return AVERROR_EOF;
274 }
275
276 AVInputFormat ff_mp3_demuxer = {
277 .name = "mp3",
278 .long_name = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"),
279 .read_probe = mp3_read_probe,
280 .read_header = mp3_read_header,
281 .read_packet = mp3_read_packet,
282 .read_seek = mp3_seek,
283 .priv_data_size = sizeof(MP3DecContext),
284 .flags = AVFMT_GENERIC_INDEX,
285 .extensions = "mp2,mp3,m2a", /* XXX: use probe */
286 };