apps/metadata/ogg.c

   1 /***************************************************************************
   2  *             __________               __   ___.
   3  *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
   4  *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
   5  *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
   6  *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
   7  *                     \/            \/     \/    \/            \/
   8  * $Id$
   9  *
  10  * Copyright (C) 2005 Dave Chapman
  11  *
  12  * This program is free software; you can redistribute it and/or
  13  * modify it under the terms of the GNU General Public License
  14  * as published by the Free Software Foundation; either version 2
  15  * of the License, or (at your option) any later version.
  16  *
  17  * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
  18  * KIND, either express or implied.
  19  *
  20  ****************************************************************************/
  21 #include <stdio.h>
  22 #include <string.h>
  23 #include <stdlib.h>
  24 #include <ctype.h>
  25 #include <inttypes.h>
  26
  27 #include "system.h"
  28 #include "id3.h"
  29 #include "metadata_common.h"
  30 #include "metadata_parsers.h"
  31 #include "logf.h"
  32
  33 /* A simple parser to read vital metadata from an Ogg Vorbis file.
  34  * Can also handle parsing Ogg Speex files for metadata. Returns
  35  * false if metadata needed by the codec couldn't be read.
  36  */
  37 bool get_ogg_metadata(int fd, struct mp3entry* id3)
  38 {
  39     /* An Ogg File is split into pages, each starting with the string
  40      * "OggS". Each page has a timestamp (in PCM samples) referred to as
  41      * the "granule position".
  42      *
  43      * An Ogg Vorbis has the following structure:
  44      * 1) Identification header (containing samplerate, numchannels, etc)
  45      * 2) Comment header - containing the Vorbis Comments
  46      * 3) Setup header - containing codec setup information
  47      * 4) Many audio packets...
  48      *
  49      * An Ogg Speex has the following structure:
  50      * 1) Identification header (containing samplerate, numchannels, etc)
  51      *    Described in this page: (http://www.speex.org/manual2/node7.html)
  52      * 2) Comment header - containing the Vorbis Comments
  53      * 3) Many audio packets.
  54      */
  55
  56     /* Use the path name of the id3 structure as a temporary buffer. */
  57     unsigned char* buf = (unsigned char *)id3->path;
  58     long comment_size;
  59     long remaining = 0;
  60     long last_serial = 0;
  61     long serial, r;
  62     int segments, header_size;
  63     int i;
  64     bool eof = false;
  65
  66     /* 92 bytes is enough for both Vorbis and Speex headers */
  67     if ((lseek(fd, 0, SEEK_SET) < 0) || (read(fd, buf, 92) < 92))
  68     {
  69         return false;
  70     }
  71
  72     /* All Ogg streams start with OggS */
  73     if (memcmp(buf, "OggS", 4) != 0)
  74     {
  75         return false;
  76     }
  77
  78     /* Check for format magic and then get metadata */
  79     if (memcmp(&buf[29], "vorbis", 6) == 0)
  80     {
  81         id3->codectype = AFMT_OGG_VORBIS;
  82         id3->frequency = get_long_le(&buf[40]);
  83         id3->vbr = true;
  84
  85         /* Comments are in second Ogg page (byte 58 onwards for Vorbis) */
  86         if (lseek(fd, 58, SEEK_SET) < 0)
  87         {
  88             return false;
  89         }
  90     }
  91     else if (memcmp(&buf[28], "Speex   ", 8) == 0)
  92     {
  93         id3->codectype = AFMT_SPEEX;
  94         id3->frequency = get_slong(&buf[64]);
  95         id3->vbr = get_long_le(&buf[88]);
  96
  97         header_size = get_long_le(&buf[60]);
  98
  99         /* Comments are in second Ogg page (byte 108 onwards for Speex) */
 100         if (lseek(fd, 28 + header_size, SEEK_SET) < 0)
 101         {
 102             return false;
 103         }
 104     }
 105     else
 106     {
 107         /* Unsupported format, try to print the marker, catches Ogg/FLAC at least */
 108         DEBUGF("Usupported format in Ogg stream: %16s\n", &buf[28]);
 109         return false;
 110     }
 111
 112     id3->filesize = filesize(fd);
 113
 114     /* We need to ensure the serial number from this page is the same as the
 115      * one from the last page (since we only support a single bitstream).
 116      */
 117     serial = get_long_le(&buf[14]);
 118
 119     /* Minimum header length for Ogg pages is 27. */
 120     if (read(fd, buf, 27) < 27)
 121     {
 122         return false;
 123     }
 124
 125     if (memcmp(buf, "OggS", 4) !=0 )
 126     {
 127         return false;
 128     }
 129
 130     segments = buf[26];
 131
 132     /* read in segment table */
 133     if (read(fd, buf, segments) < segments)
 134     {
 135         return false;
 136     }
 137
 138     /* The second packet in a vorbis stream is the comment packet. It *may*
 139      * extend beyond the second page, but usually does not. Here we find the
 140      * length of the comment packet (or the rest of the page if the comment
 141      * packet extends to the third page).
 142      */
 143     for (i = 0; i < segments; i++)
 144     {
 145         remaining += buf[i];
 146
 147         /* The last segment of a packet is always < 255 bytes */
 148         if (buf[i] < 255)
 149         {
 150               break;
 151         }
 152     }
 153
 154     comment_size = remaining;
 155
 156     if (id3->codectype == AFMT_OGG_VORBIS) {
 157         /* Now read in packet header (type and id string) */
 158         if (read(fd, buf, 7) < 7)
 159         {
 160             return false;
 161         }
 162
 163         remaining -= 7;
 164
 165         /* The first byte of a packet is the packet type; comment packets are
 166          * type 3.
 167          */
 168         if (buf[0] != 3)
 169         {
 170             return false;
 171         }
 172     }
 173
 174     /* Failure to read the tags isn't fatal. */
 175     read_vorbis_tags(fd, id3, remaining);
 176
 177     /* We now need to search for the last page in the file - identified by
 178      * by ('O','g','g','S',0) and retrieve totalsamples.
 179      */
 180
 181     /* A page is always < 64 kB */
 182     if (lseek(fd, -(MIN(64 * 1024, id3->filesize)), SEEK_END) < 0)
 183     {
 184         return false;
 185     }
 186
 187     remaining = 0;
 188
 189     while (!eof)
 190     {
 191         r = read(fd, &buf[remaining], MAX_PATH - remaining);
 192
 193         if (r <= 0)
 194         {
 195             eof = true;
 196         }
 197         else
 198         {
 199             remaining += r;
 200         }
 201
 202         /* Inefficient (but simple) search */
 203         i = 0;
 204
 205         while (i < (remaining - 3))
 206         {
 207             if ((buf[i] == 'O') && (memcmp(&buf[i], "OggS", 4) == 0))
 208             {
 209                 if (i < (remaining - 17))
 210                 {
 211                     /* Note that this only reads the low 32 bits of a
 212                      * 64 bit value.
 213                      */
 214                      id3->samples = get_long_le(&buf[i + 6]);
 215                      last_serial = get_long_le(&buf[i + 14]);
 216
 217                     /* If this page is very small the beginning of the next
 218                      * header could be in buffer. Jump near end of this header
 219                      * and continue */
 220                     i += 27;
 221                 }
 222                 else
 223                 {
 224                     break;
 225                 }
 226             }
 227             else
 228             {
 229                 i++;
 230             }
 231         }
 232
 233         if (i < remaining)
 234         {
 235             /* Move the remaining bytes to start of buffer.
 236              * Reuse var 'segments' as it is no longer needed */
 237             segments = 0;
 238             while (i < remaining)
 239             {
 240                 buf[segments++] = buf[i++];
 241             }
 242             remaining = segments;
 243         }
 244         else
 245         {
 246             /* Discard the rest of the buffer */
 247             remaining = 0;
 248         }
 249     }
 250
 251     /* This file has mutiple vorbis bitstreams (or is corrupt). */
 252     /* FIXME we should display an error here. */
 253     if (serial != last_serial)
 254     {
 255         logf("serialno mismatch");
 256         logf("%ld", serial);
 257         logf("%ld", last_serial);
 258         return false;
 259     }
 260
 261     id3->length = ((int64_t) id3->samples * 1000) / id3->frequency;
 262     if (id3->length <= 0)
 263     {
 264         logf("ogg length invalid!");
 265         return false;
 266     }
 267
 268     id3->bitrate = (((int64_t) id3->filesize - comment_size) * 8) / id3->length;
 269
 270     return true;
 271 }
 272