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 "replaygain.h"
  31
  32 #include "libavcodec/mpegaudiodecheader.h"
  33
  34 #define XING_FLAG_FRAMES 0x01
  35 #define XING_FLAG_SIZE   0x02
  36 #define XING_FLAG_TOC    0x04
  37
  38 #define XING_TOC_COUNT 100
  39
  40 typedef struct MP3DecContext {
  41     int xing_toc;
  42 } MP3DecContext;
  43
  44 /* mp3 read */
  45
  46 static int mp3_read_probe(AVProbeData *p)
  47 {
  48     int max_frames, first_frames = 0;
  49     int fsize, frames, sample_rate;
  50     uint32_t header;
  51     uint8_t *buf, *buf0, *buf2, *end;
  52     AVCodecContext avctx;
  53
  54     buf0 = p->buf;
  55     end = p->buf + p->buf_size - sizeof(uint32_t);
  56     while(buf0 < end && !*buf0)
  57         buf0++;
  58
  59     max_frames = 0;
  60     buf = buf0;
  61
  62     for(; buf < end; buf= buf2+1) {
  63         buf2 = buf;
  64
  65         for(frames = 0; buf2 < end; frames++) {
  66             header = AV_RB32(buf2);
  67             fsize = avpriv_mpa_decode_header(&avctx, header, &sample_rate, &sample_rate, &sample_rate, &sample_rate);
  68             if(fsize < 0)
  69                 break;
  70             buf2 += fsize;
  71         }
  72         max_frames = FFMAX(max_frames, frames);
  73         if(buf == buf0)
  74             first_frames= frames;
  75     }
  76     // keep this in sync with ac3 probe, both need to avoid
  77     // issues with MPEG-files!
  78     if (first_frames >= 4) return AVPROBE_SCORE_EXTENSION + 1;
  79
  80     if (max_frames) {
  81         int pes = 0, i;
  82         unsigned int code = -1;
  83
  84 #define VIDEO_ID 0x000001e0
  85 #define AUDIO_ID 0x000001c0
  86         /* do a search for mpegps headers to be able to properly bias
  87          * towards mpegps if we detect this stream as both. */
  88         for (i = 0; i<p->buf_size; i++) {
  89             code = (code << 8) + p->buf[i];
  90             if ((code & 0xffffff00) == 0x100) {
  91                 if     ((code & 0x1f0) == VIDEO_ID) pes++;
  92                 else if((code & 0x1e0) == AUDIO_ID) pes++;
  93             }
  94         }
  95
  96         if (pes)
  97             max_frames = (max_frames + pes - 1) / pes;
  98     }
  99     if      (max_frames >  500) return AVPROBE_SCORE_EXTENSION;
 100     else if (max_frames >= 4)   return AVPROBE_SCORE_EXTENSION / 2;
 101     else if (max_frames >= 1)   return 1;
 102     else                        return 0;
 103 //mpegps_mp3_unrecognized_format.mpg has max_frames=3
 104 }
 105
 106 static void read_xing_toc(AVFormatContext *s, int64_t filesize, int64_t duration)
 107 {
 108     int i;
 109     MP3DecContext *mp3 = s->priv_data;
 110
 111     if (!filesize &&
 112         !(filesize = avio_size(s->pb))) {
 113         av_log(s, AV_LOG_WARNING, "Cannot determine file size, skipping TOC table.\n");
 114         return;
 115     }
 116
 117     for (i = 0; i < XING_TOC_COUNT; i++) {
 118         uint8_t b = avio_r8(s->pb);
 119
 120         av_add_index_entry(s->streams[0],
 121                            av_rescale(b, filesize, 256),
 122                            av_rescale(i, duration, XING_TOC_COUNT),
 123                            0, 0, AVINDEX_KEYFRAME);
 124     }
 125     mp3->xing_toc = 1;
 126 }
 127
 128 /**
 129  * Try to find Xing/Info/VBRI tags and compute duration from info therein
 130  */
 131 static int mp3_parse_vbr_tags(AVFormatContext *s, AVStream *st, int64_t base)
 132 {
 133     uint32_t v, spf;
 134     unsigned frames = 0; /* Total number of frames in file */
 135     unsigned size = 0; /* Total number of bytes in the stream */
 136     const int64_t xing_offtbl[2][2] = {{32, 17}, {17,9}};
 137     MPADecodeHeader c;
 138     int vbrtag_size = 0;
 139     int is_cbr;
 140
 141     v = avio_rb32(s->pb);
 142     if(ff_mpa_check_header(v) < 0)
 143       return -1;
 144
 145     if (avpriv_mpegaudio_decode_header(&c, v) == 0)
 146         vbrtag_size = c.frame_size;
 147     if(c.layer != 3)
 148         return -1;
 149
 150     spf = c.lsf ? 576 : 1152; /* Samples per frame, layer 3 */
 151
 152     /* Check for Xing / Info tag */
 153     avio_skip(s->pb, xing_offtbl[c.lsf == 1][c.nb_channels == 1]);
 154     v = avio_rb32(s->pb);
 155     is_cbr = v == MKBETAG('I', 'n', 'f', 'o');
 156     if (v == MKBETAG('X', 'i', 'n', 'g') || is_cbr) {
 157         v = avio_rb32(s->pb);
 158         if(v & XING_FLAG_FRAMES)
 159             frames = avio_rb32(s->pb);
 160         if(v & XING_FLAG_SIZE)
 161             size = avio_rb32(s->pb);
 162         if (v & XING_FLAG_TOC && frames)
 163             read_xing_toc(s, size, av_rescale_q(frames, (AVRational){spf, c.sample_rate},
 164                                     st->time_base));
 165     }
 166
 167     /* Check for VBRI tag (always 32 bytes after end of mpegaudio header) */
 168     avio_seek(s->pb, base + 4 + 32, SEEK_SET);
 169     v = avio_rb32(s->pb);
 170     if(v == MKBETAG('V', 'B', 'R', 'I')) {
 171         /* Check tag version */
 172         if(avio_rb16(s->pb) == 1) {
 173             /* skip delay and quality */
 174             avio_skip(s->pb, 4);
 175             size = avio_rb32(s->pb);
 176             frames = avio_rb32(s->pb);
 177         }
 178     }
 179
 180     if(!frames && !size)
 181         return -1;
 182
 183     /* Skip the vbr tag frame */
 184     avio_seek(s->pb, base + vbrtag_size, SEEK_SET);
 185
 186     if(frames)
 187         st->duration = av_rescale_q(frames, (AVRational){spf, c.sample_rate},
 188                                     st->time_base);
 189     if (size && frames && !is_cbr)
 190         st->codec->bit_rate = av_rescale(size, 8 * c.sample_rate, frames * (int64_t)spf);
 191
 192     return 0;
 193 }
 194
 195 static int mp3_read_header(AVFormatContext *s)
 196 {
 197     AVStream *st;
 198     int64_t off;
 199     int ret;
 200
 201     st = avformat_new_stream(s, NULL);
 202     if (!st)
 203         return AVERROR(ENOMEM);
 204
 205     st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
 206     st->codec->codec_id = AV_CODEC_ID_MP3;
 207     st->need_parsing = AVSTREAM_PARSE_FULL;
 208     st->start_time = 0;
 209
 210     // lcm of all mp3 sample rates
 211     avpriv_set_pts_info(st, 64, 1, 14112000);
 212
 213     off = avio_tell(s->pb);
 214
 215     if (!av_dict_get(s->metadata, "", NULL, AV_DICT_IGNORE_SUFFIX))
 216         ff_id3v1_read(s);
 217
 218     if (mp3_parse_vbr_tags(s, st, off) < 0)
 219         avio_seek(s->pb, off, SEEK_SET);
 220
 221     ret = ff_replaygain_export(st, s->metadata);
 222     if (ret < 0)
 223         return ret;
 224
 225     /* the parameters will be extracted from the compressed bitstream */
 226     return 0;
 227 }
 228
 229 #define MP3_PACKET_SIZE 1024
 230
 231 static int mp3_read_packet(AVFormatContext *s, AVPacket *pkt)
 232 {
 233     int ret;
 234
 235     ret = av_get_packet(s->pb, pkt, MP3_PACKET_SIZE);
 236     if (ret < 0)
 237         return ret;
 238
 239     pkt->stream_index = 0;
 240
 241     if (ret > ID3v1_TAG_SIZE &&
 242         memcmp(&pkt->data[ret - ID3v1_TAG_SIZE], "TAG", 3) == 0)
 243         ret -= ID3v1_TAG_SIZE;
 244
 245     /* note: we need to modify the packet size here to handle the last
 246        packet */
 247     pkt->size = ret;
 248     return ret;
 249 }
 250
 251 static int mp3_seek(AVFormatContext *s, int stream_index, int64_t timestamp,
 252                     int flags)
 253 {
 254     MP3DecContext *mp3 = s->priv_data;
 255     AVIndexEntry *ie;
 256     AVStream *st = s->streams[0];
 257     int64_t ret  = av_index_search_timestamp(st, timestamp, flags);
 258     uint32_t header = 0;
 259
 260     if (!mp3->xing_toc)
 261         return AVERROR(ENOSYS);
 262
 263     if (ret < 0)
 264         return ret;
 265
 266     ie = &st->index_entries[ret];
 267     ret = avio_seek(s->pb, ie->pos, SEEK_SET);
 268     if (ret < 0)
 269         return ret;
 270
 271     while (!s->pb->eof_reached) {
 272         header = (header << 8) + avio_r8(s->pb);
 273         if (ff_mpa_check_header(header) >= 0) {
 274             ff_update_cur_dts(s, st, ie->timestamp);
 275             ret = avio_seek(s->pb, -4, SEEK_CUR);
 276             return (ret >= 0) ? 0 : ret;
 277         }
 278     }
 279
 280     return AVERROR_EOF;
 281 }
 282
 283 AVInputFormat ff_mp3_demuxer = {
 284     .name           = "mp3",
 285     .long_name      = NULL_IF_CONFIG_SMALL("MP2/3 (MPEG audio layer 2/3)"),
 286     .read_probe     = mp3_read_probe,
 287     .read_header    = mp3_read_header,
 288     .read_packet    = mp3_read_packet,
 289     .read_seek      = mp3_seek,
 290     .priv_data_size = sizeof(MP3DecContext),
 291     .flags          = AVFMT_GENERIC_INDEX,
 292     .extensions     = "mp2,mp3,m2a,mpa", /* XXX: use probe */
 293 };