mp3dec: read Xing frame TOC index
[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 /* mp3 read */
39
40 static int mp3_read_probe(AVProbeData *p)
41 {
42 int max_frames, first_frames = 0;
43 int fsize, frames, sample_rate;
44 uint32_t header;
45 uint8_t *buf, *buf0, *buf2, *end;
46 AVCodecContext avctx;
47
48 buf0 = p->buf;
49 end = p->buf + p->buf_size - sizeof(uint32_t);
50 while(buf0 < end && !*buf0)
51 buf0++;
52
53 max_frames = 0;
54 buf = buf0;
55
56 for(; buf < end; buf= buf2+1) {
57 buf2 = buf;
58
59 for(frames = 0; buf2 < end; frames++) {
60 header = AV_RB32(buf2);
61 fsize = avpriv_mpa_decode_header(&avctx, header, &sample_rate, &sample_rate, &sample_rate, &sample_rate);
62 if(fsize < 0)
63 break;
64 buf2 += fsize;
65 }
66 max_frames = FFMAX(max_frames, frames);
67 if(buf == buf0)
68 first_frames= frames;
69 }
70 // keep this in sync with ac3 probe, both need to avoid
71 // issues with MPEG-files!
72 if (first_frames >= 4) return AVPROBE_SCORE_MAX / 2 + 1;
73
74 if (max_frames) {
75 int pes = 0, i;
76 unsigned int code = -1;
77
78 #define VIDEO_ID 0x000001e0
79 #define AUDIO_ID 0x000001c0
80 /* do a search for mpegps headers to be able to properly bias
81 * towards mpegps if we detect this stream as both. */
82 for (i = 0; i<p->buf_size; i++) {
83 code = (code << 8) + p->buf[i];
84 if ((code & 0xffffff00) == 0x100) {
85 if ((code & 0x1f0) == VIDEO_ID) pes++;
86 else if((code & 0x1e0) == AUDIO_ID) pes++;
87 }
88 }
89
90 if (pes)
91 max_frames = (max_frames + pes - 1) / pes;
92 }
93 if (max_frames > 500) return AVPROBE_SCORE_MAX / 2;
94 else if (max_frames >= 4) return AVPROBE_SCORE_MAX / 4;
95 else if (max_frames >= 1) return 1;
96 else return 0;
97 //mpegps_mp3_unrecognized_format.mpg has max_frames=3
98 }
99
100 static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration)
101 {
102 int i;
103
104 if (!filesize &&
105 !(filesize = avio_size(s->pb))) {
106 av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n");
107 return;
108 }
109
110 for (i = 0; i < XING_TOC_COUNT; i++) {
111 uint8_t b = avio_r8(s->pb);
112
113 av_add_index_entry(s->streams[0],
114 av_rescale(b, filesize, 256),
115 av_rescale(i, duration, XING_TOC_COUNT),
116 0, 0, AVINDEX_KEYFRAME);
117 }
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 uint32_t v, spf;
126 unsigned frames = 0; /* Total number of frames in file */
127 unsigned size = 0; /* Total number of bytes in the stream */
128 const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
129 MPADecodeHeader c;
130 int vbrtag_size = 0;
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 if(v == MKBETAG('X', 'i', 'n', 'g') || v == MKBETAG('I', 'n', 'f', 'o')) {
147 v = avio_rb32(s->pb);
148 if(v & XING_FLAG_FRAMES)
149 frames = avio_rb32(s->pb);
150 if(v & XING_FLAG_SIZE)
151 size = avio_rb32(s->pb);
152 if (v & XING_FLAG_TOC && frames)
153 read_xing_toc(s, size, av_rescale_q(frames, (AVRational){spf, c.sample_rate},
154 st->time_base));
155 }
156
157 /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */
158 avio_seek(s->pb, base + 4 + 32, SEEK_SET);
159 v = avio_rb32(s->pb);
160 if(v == MKBETAG('V', 'B', 'R', 'I')) {
161 /* Check tag version */
162 if(avio_rb16(s->pb) == 1) {
163 /* skip delay and quality */
164 avio_skip(s->pb, 4);
165 size = avio_rb32(s->pb);
166 frames = avio_rb32(s->pb);
167 }
168 }
169
170 if(!frames && !size)
171 return -1;
172
173 /* Skip the vbr tag frame */
174 avio_seek(s->pb, base + vbrtag_size, SEEK_SET);
175
176 if(frames)
177 st->duration = av_rescale_q(frames, (AVRational){spf, c.sample_rate},
178 st->time_base);
179 if(size && frames)
180 st->codec->bit_rate = av_rescale(size, 8 * c.sample_rate, frames * (int64_t)spf);
181
182 return 0;
183 }
184
185 static int mp3_read_header(AVFormatContext *s)
186 {
187 AVStream *st;
188 int64_t off;
189
190 st = avformat_new_stream(s, NULL);
191 if (!st)
192 return AVERROR(ENOMEM);
193
194 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
195 st->codec->codec_id = AV_CODEC_ID_MP3;
196 st->need_parsing = AVSTREAM_PARSE_FULL;
197 st->start_time = 0;
198
199 // lcm of all mp3 sample rates
200 avpriv_set_pts_info(st, 64, 1, 14112000);
201
202 off = avio_tell(s->pb);
203
204 if (!av_dict_get(s->metadata, "", NULL, AV_DICT_IGNORE_SUFFIX))
205 ff_id3v1_read(s);
206
207 if (mp3_parse_vbr_tags(s, st, off) < 0)
208 avio_seek(s->pb, off, SEEK_SET);
209
210 /* the parameters will be extracted from the compressed bitstream */
211 return 0;
212 }
213
214 #define MP3_PACKET_SIZE 1024
215
216 static int mp3_read_packet(AVFormatContext *s, AVPacket *pkt)
217 {
218 int ret;
219
220 ret = av_get_packet(s->pb, pkt, MP3_PACKET_SIZE);
221 if (ret < 0)
222 return ret;
223
224 pkt->stream_index = 0;
225
226 if (ret > ID3v1_TAG_SIZE &&
227 memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0)
228 ret -= ID3v1_TAG_SIZE;
229
230 /* note: we need to modify the packet size here to handle the last
231 packet */
232 pkt->size = ret;
233 return ret;
234 }
235
236 static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp,
237 int flags)
238 {
239 AVIndexEntry *ie;
240 AVStream *st = s->streams[0];
241 int64_t ret = av_index_search_timestamp(st, timestamp, flags);
242 uint32_t header = 0;
243
244 if (ret < 0)
245 return ret;
246
247 ie = &st->index_entries[ret];
248 ret = avio_seek(s->pb, ie->pos, SEEK_SET);
249 if (ret < 0)
250 return ret;
251
252 while (!s->pb->eof_reached) {
253 header = (header << 8) + avio_r8(s->pb);
254 if (ff_mpa_check_header(header) >= 0) {
255 ff_update_cur_dts(s, st, ie->timestamp);
256 ret = avio_seek(s->pb, -4, SEEK_CUR);
257 return (ret >= 0) ? 0 : ret;
258 }
259 }
260
261 return AVERROR_EOF;
262 }
263
264 AVInputFormat ff_mp3_demuxer = {
265 .name = "mp3",
266 .long_name = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"),
267 .read_probe = mp3_read_probe,
268 .read_header = mp3_read_header,
269 .read_packet = mp3_read_packet,
270 .read_seek = mp3_seek,
271 .flags = AVFMT_GENERIC_INDEX,
272 .extensions = "mp2,mp3,m2a", /* XXX: use probe */
273 };