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     int is_cbr;
 132
 133     v = avio_rb32(s->pb);
 134     if(ff_mpa_check_header(v) < 0)
 135       return -1;
 136
 137     if (avpriv_mpegaudio_decode_header(&c, v) == 0)
 138         vbrtag_size = c.frame_size;
 139     if(c.layer != 3)
 140         return -1;
 141
 142     spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */
 143
 144     /* Check for Xing / Info tag */
 145     avio_skip(s->pb, xing_offtbl[c.lsf == 1][c.nb_channels == 1]);
 146     v = avio_rb32(s->pb);
 147     is_cbr = v == MKBETAG('I', 'n', 'f', 'o');
 148     if (v == MKBETAG('X', 'i', 'n', 'g') || is_cbr) {
 149         v = avio_rb32(s->pb);
 150         if(v & XING_FLAG_FRAMES)
 151             frames = avio_rb32(s->pb);
 152         if(v & XING_FLAG_SIZE)
 153             size = avio_rb32(s->pb);
 154         if (v & XING_FLAG_TOC && frames)
 155             read_xing_toc(s, size, av_rescale_q(frames, (AVRational){spf, c.sample_rate},
 156                                     st->time_base));
 157     }
 158
 159     /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */
 160     avio_seek(s->pb, base + 4 + 32, SEEK_SET);
 161     v = avio_rb32(s->pb);
 162     if(v == MKBETAG('V', 'B', 'R', 'I')) {
 163         /* Check tag version */
 164         if(avio_rb16(s->pb) == 1) {
 165             /* skip delay and quality */
 166             avio_skip(s->pb, 4);
 167             size = avio_rb32(s->pb);
 168             frames = avio_rb32(s->pb);
 169         }
 170     }
 171
 172     if(!frames && !size)
 173         return -1;
 174
 175     /* Skip the vbr tag frame */
 176     avio_seek(s->pb, base + vbrtag_size, SEEK_SET);
 177
 178     if(frames)
 179         st->duration = av_rescale_q(frames, (AVRational){spf, c.sample_rate},
 180                                     st->time_base);
 181     if (size && frames && !is_cbr)
 182         st->codec->bit_rate = av_rescale(size, 8 * c.sample_rate, frames * (int64_t)spf);
 183
 184     return 0;
 185 }
 186
 187 static int mp3_read_header(AVFormatContext *s)
 188 {
 189     AVStream *st;
 190     int64_t off;
 191
 192     st = avformat_new_stream(s, NULL);
 193     if (!st)
 194         return AVERROR(ENOMEM);
 195
 196     st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
 197     st->codec->codec_id = AV_CODEC_ID_MP3;
 198     st->need_parsing = AVSTREAM_PARSE_FULL;
 199     st->start_time = 0;
 200
 201     // lcm of all mp3 sample rates
 202     avpriv_set_pts_info(st, 64, 1, 14112000);
 203
 204     off = avio_tell(s->pb);
 205
 206     if (!av_dict_get(s->metadata, "", NULL, AV_DICT_IGNORE_SUFFIX))
 207         ff_id3v1_read(s);
 208
 209     if (mp3_parse_vbr_tags(s, st, off) < 0)
 210         avio_seek(s->pb, off, SEEK_SET);
 211
 212     /* the parameters will be extracted from the compressed bitstream */
 213     return 0;
 214 }
 215
 216 #define MP3_PACKET_SIZE 1024
 217
 218 static int mp3_read_packet(AVFormatContext *s, AVPacket *pkt)
 219 {
 220     int ret;
 221
 222     ret = av_get_packet(s->pb, pkt, MP3_PACKET_SIZE);
 223     if (ret < 0)
 224         return ret;
 225
 226     pkt->stream_index = 0;
 227
 228     if (ret > ID3v1_TAG_SIZE &&
 229         memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0)
 230         ret -= ID3v1_TAG_SIZE;
 231
 232     /* note: we need to modify the packet size here to handle the last
 233        packet */
 234     pkt->size = ret;
 235     return ret;
 236 }
 237
 238 static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp,
 239                     int flags)
 240 {
 241     AVIndexEntry *ie;
 242     AVStream *st = s->streams[0];
 243     int64_t ret  = av_index_search_timestamp(st, timestamp, flags);
 244     uint32_t header = 0;
 245
 246     if (ret < 0)
 247         return ret;
 248
 249     ie = &st->index_entries[ret];
 250     ret = avio_seek(s->pb, ie->pos, SEEK_SET);
 251     if (ret < 0)
 252         return ret;
 253
 254     while (!s->pb->eof_reached) {
 255         header = (header << 8) + avio_r8(s->pb);
 256         if (ff_mpa_check_header(header) >= 0) {
 257             ff_update_cur_dts(s, st, ie->timestamp);
 258             ret = avio_seek(s->pb, -4, SEEK_CUR);
 259             return (ret >= 0) ? 0 : ret;
 260         }
 261     }
 262
 263     return AVERROR_EOF;
 264 }
 265
 266 AVInputFormat ff_mp3_demuxer = {
 267     .name           = "mp3",
 268     .long_name      = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"),
 269     .read_probe     = mp3_read_probe,
 270     .read_header    = mp3_read_header,
 271     .read_packet    = mp3_read_packet,
 272     .read_seek      = mp3_seek,
 273     .flags          = AVFMT_GENERIC_INDEX,
 274     .extensions     = "mp2,mp3,m2a", /* XXX: use probe */
 275 };