libavformat/aiffdec.c

   1 /*
   2  * AIFF/AIFF-C demuxer
   3  * Copyright (c) 2006  Patrick Guimond
   4  *
   5  * This file is part of FFmpeg.
   6  *
   7  * FFmpeg 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  * FFmpeg 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 FFmpeg; 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/intreadwrite.h"
  23 #include "libavutil/dict.h"
  24 #include "libavutil/mem.h"
  25 #include "avformat.h"
  26 #include "demux.h"
  27 #include "internal.h"
  28 #include "pcm.h"
  29 #include "aiff.h"
  30 #include "id3v2.h"
  31 #include "mov_chan.h"
  32 #include "replaygain.h"
  33
  34 #define AIFF                    0
  35 #define AIFF_C_VERSION1         0xA2805140
  36
  37 typedef struct AIFFInputContext {
  38     int64_t data_end;
  39     int block_duration;
  40 } AIFFInputContext;
  41
  42 static enum AVCodecID aiff_codec_get_id(int bps)
  43 {
  44     if (bps <= 8)
  45         return AV_CODEC_ID_PCM_S8;
  46     if (bps <= 16)
  47         return AV_CODEC_ID_PCM_S16BE;
  48     if (bps <= 24)
  49         return AV_CODEC_ID_PCM_S24BE;
  50     if (bps <= 32)
  51         return AV_CODEC_ID_PCM_S32BE;
  52
  53     /* bigger than 32 isn't allowed  */
  54     return AV_CODEC_ID_NONE;
  55 }
  56
  57 /* returns the size of the found tag */
  58 static int64_t get_tag(AVIOContext *pb, uint32_t * tag)
  59 {
  60     int64_t size;
  61
  62     if (avio_feof(pb))
  63         return AVERROR_INVALIDDATA;
  64
  65     *tag = avio_rl32(pb);
  66     size = avio_rb32(pb);
  67
  68     return size;
  69 }
  70
  71 /* Metadata string read */
  72 static void get_meta(AVFormatContext *s, const char *key, int64_t size)
  73 {
  74     uint8_t *str = NULL;
  75
  76     if (size < SIZE_MAX)
  77         str = av_malloc(size+1);
  78
  79     if (str) {
  80         int res = avio_read(s->pb, str, size);
  81         if (res < 0){
  82             av_free(str);
  83             return;
  84         }
  85         size -= res;
  86         str[res] = 0;
  87         av_dict_set(&s->metadata, key, str, AV_DICT_DONT_STRDUP_VAL);
  88     }
  89
  90     avio_skip(s->pb, size);
  91 }
  92
  93 /* Returns the number of sound data frames or negative on error */
  94 static int get_aiff_header(AVFormatContext *s, int64_t size,
  95                                     unsigned version)
  96 {
  97     AVIOContext *pb        = s->pb;
  98     AVCodecParameters *par = s->streams[0]->codecpar;
  99     AIFFInputContext *aiff = s->priv_data;
 100     int exp;
 101     uint64_t val;
 102     int sample_rate;
 103     unsigned int num_frames;
 104     int channels;
 105
 106     if (size & 1)
 107         size++;
 108     par->codec_type = AVMEDIA_TYPE_AUDIO;
 109     channels = avio_rb16(pb);
 110     if (par->ch_layout.nb_channels && par->ch_layout.nb_channels != channels)
 111         return AVERROR_INVALIDDATA;
 112     par->ch_layout.nb_channels = channels;
 113     num_frames = avio_rb32(pb);
 114     par->bits_per_coded_sample = avio_rb16(pb);
 115
 116     exp = avio_rb16(pb) - 16383 - 63;
 117     val = avio_rb64(pb);
 118     if (exp <-63 || exp >63) {
 119         av_log(s, AV_LOG_ERROR, "exp %d is out of range\n", exp);
 120         return AVERROR_INVALIDDATA;
 121     }
 122     if (exp >= 0)
 123         sample_rate = val << exp;
 124     else
 125         sample_rate = (val + (1ULL<<(-exp-1))) >> -exp;
 126     if (sample_rate <= 0)
 127         return AVERROR_INVALIDDATA;
 128
 129     par->sample_rate = sample_rate;
 130     if (size < 18)
 131         return AVERROR_INVALIDDATA;
 132     size -= 18;
 133
 134     /* get codec id for AIFF-C */
 135     if (size < 4) {
 136         version = AIFF;
 137     } else if (version == AIFF_C_VERSION1) {
 138         par->codec_tag = avio_rl32(pb);
 139         par->codec_id  = ff_codec_get_id(ff_codec_aiff_tags, par->codec_tag);
 140         if (par->codec_id == AV_CODEC_ID_NONE)
 141             avpriv_request_sample(s, "unknown or unsupported codec tag: %s",
 142                                   av_fourcc2str(par->codec_tag));
 143         size -= 4;
 144     }
 145
 146     if (version != AIFF_C_VERSION1 || par->codec_id == AV_CODEC_ID_PCM_S16BE) {
 147         par->codec_id = aiff_codec_get_id(par->bits_per_coded_sample);
 148         par->bits_per_coded_sample = av_get_bits_per_sample(par->codec_id);
 149         aiff->block_duration = 1;
 150     } else {
 151         switch (par->codec_id) {
 152         case AV_CODEC_ID_PCM_F32BE:
 153         case AV_CODEC_ID_PCM_F64BE:
 154         case AV_CODEC_ID_PCM_S16LE:
 155         case AV_CODEC_ID_PCM_ALAW:
 156         case AV_CODEC_ID_PCM_MULAW:
 157             aiff->block_duration = 1;
 158             break;
 159         case AV_CODEC_ID_ADPCM_IMA_QT:
 160             par->block_align = 34 * channels;
 161             break;
 162         case AV_CODEC_ID_MACE3:
 163             par->block_align = 2 * channels;
 164             break;
 165         case AV_CODEC_ID_ADPCM_G726LE:
 166             par->bits_per_coded_sample = 5;
 167         case AV_CODEC_ID_ADPCM_IMA_WS:
 168         case AV_CODEC_ID_ADPCM_G722:
 169         case AV_CODEC_ID_MACE6:
 170         case AV_CODEC_ID_CBD2_DPCM:
 171         case AV_CODEC_ID_SDX2_DPCM:
 172             par->block_align = 1 * channels;
 173             break;
 174         case AV_CODEC_ID_GSM:
 175             par->block_align = 33;
 176             break;
 177         case AV_CODEC_ID_G728:
 178             par->block_align = 5;
 179             break;
 180         case AV_CODEC_ID_ADPCM_N64:
 181             par->block_align = 9;
 182             break;
 183         default:
 184             aiff->block_duration = 1;
 185             break;
 186         }
 187         if (par->block_align > 0)
 188             aiff->block_duration = av_get_audio_frame_duration2(par,
 189                                                                 par->block_align);
 190     }
 191
 192     /* Block align needs to be computed in all cases, as the definition
 193      * is specific to applications -> here we use the WAVE format definition */
 194     if (!par->block_align)
 195         par->block_align = (av_get_bits_per_sample(par->codec_id) * channels) >> 3;
 196
 197     if (aiff->block_duration) {
 198         par->bit_rate = av_rescale(par->sample_rate, par->block_align * 8LL,
 199                                    aiff->block_duration);
 200         if (par->bit_rate < 0)
 201             par->bit_rate = 0;
 202     }
 203
 204     /* Chunk is over */
 205     if (size)
 206         avio_skip(pb, size);
 207
 208     return num_frames;
 209 }
 210
 211 static int aiff_probe(const AVProbeData *p)
 212 {
 213     /* check file header */
 214     if (AV_RL32(p->buf) == MKTAG('F', 'O', 'R', 'M') &&
 215         AV_RB32(p->buf + 4) >= 4 &&
 216         p->buf[8] == 'A' && p->buf[9] == 'I' &&
 217         p->buf[10] == 'F' && (p->buf[11] == 'F' || p->buf[11] == 'C'))
 218         return AVPROBE_SCORE_MAX;
 219     else
 220         return 0;
 221 }
 222
 223 /* aiff input */
 224 static int aiff_read_header(AVFormatContext *s)
 225 {
 226     int ret;
 227     int64_t filesize, size;
 228     int64_t offset = 0, position;
 229     uint32_t tag;
 230     unsigned version = AIFF_C_VERSION1;
 231     AVIOContext *pb = s->pb;
 232     AVStream * st;
 233     AIFFInputContext *aiff = s->priv_data;
 234     ID3v2ExtraMeta *id3v2_extra_meta;
 235
 236     /* check FORM header */
 237     filesize = get_tag(pb, &tag);
 238     if (filesize < 4 || tag != MKTAG('F', 'O', 'R', 'M'))
 239         return AVERROR_INVALIDDATA;
 240
 241     /* AIFF data type */
 242     tag = avio_rl32(pb);
 243     if (tag == MKTAG('A', 'I', 'F', 'F'))       /* Got an AIFF file */
 244         version = AIFF;
 245     else if (tag != MKTAG('A', 'I', 'F', 'C'))  /* An AIFF-C file then */
 246         return AVERROR_INVALIDDATA;
 247
 248     filesize -= 4;
 249
 250     st = avformat_new_stream(s, NULL);
 251     if (!st)
 252         return AVERROR(ENOMEM);
 253
 254     while (filesize > 0) {
 255         /* parse different chunks */
 256         size = get_tag(pb, &tag);
 257
 258         if (size == AVERROR_EOF && offset > 0 && st->codecpar->block_align) {
 259             av_log(s, AV_LOG_WARNING, "header parser hit EOF\n");
 260             goto got_sound;
 261         }
 262         if (size < 0)
 263             return size;
 264
 265         filesize -= size + 8;
 266
 267         switch (tag) {
 268         case MKTAG('C', 'O', 'M', 'M'):     /* Common chunk */
 269             /* Then for the complete header info */
 270             st->nb_frames = get_aiff_header(s, size, version);
 271             if (st->nb_frames < 0)
 272                 return st->nb_frames;
 273             if (offset > 0) // COMM is after SSND
 274                 goto got_sound;
 275             break;
 276         case MKTAG('I', 'D', '3', ' '):
 277             position = avio_tell(pb);
 278             ff_id3v2_read(s, ID3v2_DEFAULT_MAGIC, &id3v2_extra_meta, size);
 279             if (id3v2_extra_meta)
 280                 if ((ret = ff_id3v2_parse_apic(s, id3v2_extra_meta)) < 0 ||
 281                     (ret = ff_id3v2_parse_chapters(s, id3v2_extra_meta)) < 0) {
 282                     ff_id3v2_free_extra_meta(&id3v2_extra_meta);
 283                     return ret;
 284                 }
 285             ff_id3v2_free_extra_meta(&id3v2_extra_meta);
 286             if (position + size > avio_tell(pb))
 287                 avio_skip(pb, position + size - avio_tell(pb));
 288             break;
 289         case MKTAG('F', 'V', 'E', 'R'):     /* Version chunk */
 290             version = avio_rb32(pb);
 291             break;
 292         case MKTAG('N', 'A', 'M', 'E'):     /* Sample name chunk */
 293             get_meta(s, "title"    , size);
 294             break;
 295         case MKTAG('A', 'U', 'T', 'H'):     /* Author chunk */
 296             get_meta(s, "author"   , size);
 297             break;
 298         case MKTAG('(', 'c', ')', ' '):     /* Copyright chunk */
 299             get_meta(s, "copyright", size);
 300             break;
 301         case MKTAG('A', 'N', 'N', 'O'):     /* Annotation chunk */
 302             get_meta(s, "comment"  , size);
 303             break;
 304         case MKTAG('S', 'S', 'N', 'D'):     /* Sampled sound chunk */
 305             if (size < 8)
 306                 return AVERROR_INVALIDDATA;
 307             aiff->data_end = avio_tell(pb) + size;
 308             offset = avio_rb32(pb);      /* Offset of sound data */
 309             avio_rb32(pb);               /* BlockSize... don't care */
 310             offset += avio_tell(pb);    /* Compute absolute data offset */
 311             if (st->codecpar->block_align && !(pb->seekable & AVIO_SEEKABLE_NORMAL))    /* Assume COMM already parsed */
 312                 goto got_sound;
 313             if (!(pb->seekable & AVIO_SEEKABLE_NORMAL)) {
 314                 av_log(s, AV_LOG_ERROR, "file is not seekable\n");
 315                 return -1;
 316             }
 317             avio_skip(pb, size - 8);
 318             break;
 319         case MKTAG('w', 'a', 'v', 'e'):
 320             if ((uint64_t)size > (1<<30))
 321                 return AVERROR_INVALIDDATA;
 322             if ((ret = ff_get_extradata(s, st->codecpar, pb, size)) < 0)
 323                 return ret;
 324             if (   (st->codecpar->codec_id == AV_CODEC_ID_QDMC || st->codecpar->codec_id == AV_CODEC_ID_QDM2)
 325                 && size>=12*4 && !st->codecpar->block_align) {
 326                 st->codecpar->block_align = AV_RB32(st->codecpar->extradata+11*4);
 327                 aiff->block_duration = AV_RB32(st->codecpar->extradata+9*4);
 328             } else if (st->codecpar->codec_id == AV_CODEC_ID_QCELP) {
 329                 char rate = 0;
 330                 if (size >= 25)
 331                     rate = st->codecpar->extradata[24];
 332                 switch (rate) {
 333                 case 'H': // RATE_HALF
 334                     st->codecpar->block_align = 17;
 335                     break;
 336                 case 'F': // RATE_FULL
 337                 default:
 338                     st->codecpar->block_align = 35;
 339                 }
 340                 aiff->block_duration = 160;
 341                 st->codecpar->bit_rate = (int64_t)st->codecpar->sample_rate * (st->codecpar->block_align << 3) /
 342                                          aiff->block_duration;
 343             }
 344             break;
 345         case MKTAG('C','H','A','N'):
 346             if ((ret = ff_mov_read_chan(s, pb, st, size)) < 0)
 347                 return ret;
 348             break;
 349         case MKTAG('A','P','C','M'): /* XA ADPCM compressed sound chunk */
 350             st->codecpar->codec_id = AV_CODEC_ID_ADPCM_XA;
 351             aiff->data_end = avio_tell(pb) + size;
 352             offset = avio_tell(pb) + 8;
 353             /* This field is unknown and its data seems to be irrelevant */
 354             avio_rb32(pb);
 355             st->codecpar->block_align = avio_rb32(pb);
 356
 357             goto got_sound;
 358             break;
 359         case MKTAG('A','P','P','L'):
 360             if (size > 4) {
 361                 uint32_t chunk = avio_rl32(pb);
 362
 363                 size -= 4;
 364                 if (chunk == MKTAG('s','t','o','c')) {
 365                     int len = avio_r8(pb);
 366
 367                     size--;
 368                     if (len == 11 && size > 11) {
 369                         uint8_t chunk[11];
 370
 371                         ret = avio_read(pb, chunk, 11);
 372                         if (ret > 0)
 373                             size -= ret;
 374                         if (!memcmp(chunk, "VADPCMCODES", sizeof(chunk))) {
 375                             if ((ret = ff_get_extradata(s, st->codecpar, pb, size)) < 0)
 376                                 return ret;
 377                             size -= ret;
 378                         }
 379                     }
 380                 }
 381             }
 382             avio_skip(pb, size);
 383             break;
 384         case 0:
 385             if (offset > 0 && st->codecpar->block_align) // COMM && SSND
 386                 goto got_sound;
 387         default: /* Jump */
 388             avio_skip(pb, size);
 389         }
 390
 391         /* Skip required padding byte for odd-sized chunks. */
 392         if (size & 1) {
 393             filesize--;
 394             avio_skip(pb, 1);
 395         }
 396     }
 397
 398     ret = ff_replaygain_export(st, s->metadata);
 399     if (ret < 0)
 400         return ret;
 401
 402 got_sound:
 403     if (!st->codecpar->block_align && st->codecpar->codec_id == AV_CODEC_ID_QCELP) {
 404         av_log(s, AV_LOG_WARNING, "qcelp without wave chunk, assuming full rate\n");
 405         st->codecpar->block_align = 35;
 406     } else if (st->codecpar->block_align <= 0) {
 407         av_log(s, AV_LOG_ERROR, "could not find COMM tag or invalid block_align value\n");
 408         return AVERROR_INVALIDDATA;
 409     }
 410     if (aiff->block_duration < 0)
 411         return AVERROR_INVALIDDATA;
 412
 413     /* Now positioned, get the sound data start and end */
 414     avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
 415     st->start_time = 0;
 416     st->duration = st->nb_frames * aiff->block_duration;
 417
 418     /* Position the stream at the first block */
 419     avio_seek(pb, offset, SEEK_SET);
 420
 421     return 0;
 422 }
 423
 424 #define MAX_SIZE 4096
 425
 426 static int aiff_read_packet(AVFormatContext *s,
 427                             AVPacket *pkt)
 428 {
 429     AVStream *st = s->streams[0];
 430     AIFFInputContext *aiff = s->priv_data;
 431     int64_t max_size;
 432     int res, size;
 433
 434     /* calculate size of remaining data */
 435     max_size = aiff->data_end - avio_tell(s->pb);
 436     if (max_size <= 0)
 437         return AVERROR_EOF;
 438
 439     if (!st->codecpar->block_align) {
 440         av_log(s, AV_LOG_ERROR, "block_align not set\n");
 441         return AVERROR_INVALIDDATA;
 442     }
 443
 444     /* Now for that packet */
 445     switch (st->codecpar->codec_id) {
 446     case AV_CODEC_ID_ADPCM_IMA_QT:
 447     case AV_CODEC_ID_GSM:
 448     case AV_CODEC_ID_QDM2:
 449     case AV_CODEC_ID_QCELP:
 450         size = st->codecpar->block_align;
 451         break;
 452     default:
 453         size = st->codecpar->block_align ? (MAX_SIZE / st->codecpar->block_align) * st->codecpar->block_align : MAX_SIZE;
 454         if (!size)
 455             return AVERROR_INVALIDDATA;
 456     }
 457     size = FFMIN(max_size, size);
 458     res = av_get_packet(s->pb, pkt, size);
 459     if (res < 0)
 460         return res;
 461
 462     if (size >= st->codecpar->block_align)
 463         pkt->flags &= ~AV_PKT_FLAG_CORRUPT;
 464     /* Only one stream in an AIFF file */
 465     pkt->stream_index = 0;
 466     pkt->duration     = (res / st->codecpar->block_align) * (int64_t) aiff->block_duration;
 467     return 0;
 468 }
 469
 470 const FFInputFormat ff_aiff_demuxer = {
 471     .p.name         = "aiff",
 472     .p.long_name    = NULL_IF_CONFIG_SMALL("Audio IFF"),
 473     .p.codec_tag    = ff_aiff_codec_tags_list,
 474     .priv_data_size = sizeof(AIFFInputContext),
 475     .read_probe     = aiff_probe,
 476     .read_header    = aiff_read_header,
 477     .read_packet    = aiff_read_packet,
 478     .read_seek      = ff_pcm_read_seek,
 479 };