libavformat/4xm.c

   1 /*
   2  * 4X Technologies .4xm File Demuxer (no muxer)
   3  * Copyright (c) 2003  The FFmpeg project
   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 /**
  23  * @file
  24  * 4X Technologies file demuxer
  25  * by Mike Melanson (melanson@pcisys.net)
  26  * for more information on the .4xm file format, visit:
  27  *   http://www.pcisys.net/~melanson/codecs/
  28  */
  29
  30 #include "libavutil/intreadwrite.h"
  31 #include "libavutil/intfloat.h"
  32 #include "libavcodec/internal.h"
  33 #include "avformat.h"
  34 #include "internal.h"
  35
  36 #define     RIFF_TAG MKTAG('R', 'I', 'F', 'F')
  37 #define  FOURXMV_TAG MKTAG('4', 'X', 'M', 'V')
  38 #define     LIST_TAG MKTAG('L', 'I', 'S', 'T')
  39 #define     HEAD_TAG MKTAG('H', 'E', 'A', 'D')
  40 #define     TRK__TAG MKTAG('T', 'R', 'K', '_')
  41 #define     MOVI_TAG MKTAG('M', 'O', 'V', 'I')
  42 #define     VTRK_TAG MKTAG('V', 'T', 'R', 'K')
  43 #define     STRK_TAG MKTAG('S', 'T', 'R', 'K')
  44 #define     std__TAG MKTAG('s', 't', 'd', '_')
  45 #define     name_TAG MKTAG('n', 'a', 'm', 'e')
  46 #define     vtrk_TAG MKTAG('v', 't', 'r', 'k')
  47 #define     strk_TAG MKTAG('s', 't', 'r', 'k')
  48 #define     ifrm_TAG MKTAG('i', 'f', 'r', 'm')
  49 #define     pfrm_TAG MKTAG('p', 'f', 'r', 'm')
  50 #define     cfrm_TAG MKTAG('c', 'f', 'r', 'm')
  51 #define     ifr2_TAG MKTAG('i', 'f', 'r', '2')
  52 #define     pfr2_TAG MKTAG('p', 'f', 'r', '2')
  53 #define     cfr2_TAG MKTAG('c', 'f', 'r', '2')
  54 #define     snd__TAG MKTAG('s', 'n', 'd', '_')
  55
  56 #define vtrk_SIZE 0x44
  57 #define strk_SIZE 0x28
  58
  59 #define GET_LIST_HEADER() \
  60     fourcc_tag = avio_rl32(pb); \
  61     size       = avio_rl32(pb); \
  62     if (fourcc_tag != LIST_TAG) { \
  63         ret = AVERROR_INVALIDDATA; \
  64         goto fail; \
  65     } \
  66     fourcc_tag = avio_rl32(pb);
  67
  68 typedef struct AudioTrack {
  69     int sample_rate;
  70     int bits;
  71     int channels;
  72     int stream_index;
  73     int adpcm;
  74     int64_t audio_pts;
  75 } AudioTrack;
  76
  77 typedef struct FourxmDemuxContext {
  78     int video_stream_index;
  79     int track_count;
  80     AudioTrack *tracks;
  81
  82     int64_t video_pts;
  83     AVRational fps;
  84 } FourxmDemuxContext;
  85
  86 static int fourxm_probe(const AVProbeData *p)
  87 {
  88     if ((AV_RL32(&p->buf[0]) != RIFF_TAG) ||
  89         (AV_RL32(&p->buf[8]) != FOURXMV_TAG))
  90         return 0;
  91
  92     return AVPROBE_SCORE_MAX;
  93 }
  94
  95 static int parse_vtrk(AVFormatContext *s,
  96                       FourxmDemuxContext *fourxm, uint8_t *buf, int size,
  97                       int left)
  98 {
  99     AVStream *st;
 100     /* check that there is enough data */
 101     if (size != vtrk_SIZE || left < size + 8) {
 102         return AVERROR_INVALIDDATA;
 103     }
 104
 105     /* allocate a new AVStream */
 106     st = avformat_new_stream(s, NULL);
 107     if (!st)
 108         return AVERROR(ENOMEM);
 109
 110     avpriv_set_pts_info(st, 60, fourxm->fps.den, fourxm->fps.num);
 111
 112     fourxm->video_stream_index = st->index;
 113
 114     st->codecpar->codec_type     = AVMEDIA_TYPE_VIDEO;
 115     st->codecpar->codec_id       = AV_CODEC_ID_4XM;
 116
 117     st->codecpar->extradata      = av_mallocz(4 + AV_INPUT_BUFFER_PADDING_SIZE);
 118     if (!st->codecpar->extradata)
 119         return AVERROR(ENOMEM);
 120     st->codecpar->extradata_size = 4;
 121     AV_WL32(st->codecpar->extradata, AV_RL32(buf + 16));
 122     st->codecpar->width  = AV_RL32(buf + 36);
 123     st->codecpar->height = AV_RL32(buf + 40);
 124
 125     return 0;
 126 }
 127
 128
 129 static int parse_strk(AVFormatContext *s,
 130                       FourxmDemuxContext *fourxm, uint8_t *buf, int size,
 131                       int left)
 132 {
 133     AVStream *st;
 134     int track;
 135     /* check that there is enough data */
 136     if (size != strk_SIZE || left < size + 8)
 137         return AVERROR_INVALIDDATA;
 138
 139     track = AV_RL32(buf + 8);
 140     if ((unsigned)track >= UINT_MAX / sizeof(AudioTrack) - 1 ||
 141         track >= s->max_streams) {
 142         av_log(s, AV_LOG_ERROR, "current_track too large\n");
 143         return AVERROR_INVALIDDATA;
 144     }
 145
 146     if (track + 1 > fourxm->track_count) {
 147         if (av_reallocp_array(&fourxm->tracks, track + 1, sizeof(AudioTrack)))
 148             return AVERROR(ENOMEM);
 149         memset(&fourxm->tracks[fourxm->track_count], 0,
 150                sizeof(AudioTrack) * (track + 1 - fourxm->track_count));
 151         fourxm->track_count = track + 1;
 152     } else {
 153         if (fourxm->tracks[track].bits)
 154             return AVERROR_INVALIDDATA;
 155     }
 156     fourxm->tracks[track].adpcm       = AV_RL32(buf + 12);
 157     fourxm->tracks[track].channels    = AV_RL32(buf + 36);
 158     fourxm->tracks[track].sample_rate = AV_RL32(buf + 40);
 159     fourxm->tracks[track].bits        = AV_RL32(buf + 44);
 160     fourxm->tracks[track].audio_pts   = 0;
 161
 162     if (fourxm->tracks[track].channels    <= 0 ||
 163         fourxm->tracks[track].channels     > FF_SANE_NB_CHANNELS ||
 164         fourxm->tracks[track].sample_rate <= 0 ||
 165         fourxm->tracks[track].bits        <= 0 ||
 166         fourxm->tracks[track].bits         > INT_MAX / FF_SANE_NB_CHANNELS) {
 167         av_log(s, AV_LOG_ERROR, "audio header invalid\n");
 168         return AVERROR_INVALIDDATA;
 169     }
 170     if (!fourxm->tracks[track].adpcm && fourxm->tracks[track].bits<8) {
 171         av_log(s, AV_LOG_ERROR, "bits unspecified for non ADPCM\n");
 172         return AVERROR_INVALIDDATA;
 173     }
 174
 175     if (fourxm->tracks[track].sample_rate > INT64_MAX / fourxm->tracks[track].bits / fourxm->tracks[track].channels) {
 176         av_log(s, AV_LOG_ERROR, "Overflow during bit rate calculation %d * %d * %d\n",
 177                fourxm->tracks[track].sample_rate, fourxm->tracks[track].bits, fourxm->tracks[track].channels);
 178         return AVERROR_INVALIDDATA;
 179     }
 180
 181     /* allocate a new AVStream */
 182     st = avformat_new_stream(s, NULL);
 183     if (!st)
 184         return AVERROR(ENOMEM);
 185
 186     st->id = track;
 187     avpriv_set_pts_info(st, 60, 1, fourxm->tracks[track].sample_rate);
 188
 189     fourxm->tracks[track].stream_index = st->index;
 190
 191     st->codecpar->codec_type            = AVMEDIA_TYPE_AUDIO;
 192     st->codecpar->codec_tag             = 0;
 193     st->codecpar->channels              = fourxm->tracks[track].channels;
 194     st->codecpar->sample_rate           = fourxm->tracks[track].sample_rate;
 195     st->codecpar->bits_per_coded_sample = fourxm->tracks[track].bits;
 196     st->codecpar->bit_rate              = (int64_t)st->codecpar->channels *
 197                                           st->codecpar->sample_rate *
 198                                           st->codecpar->bits_per_coded_sample;
 199     st->codecpar->block_align           = st->codecpar->channels *
 200                                           st->codecpar->bits_per_coded_sample;
 201
 202     if (fourxm->tracks[track].adpcm){
 203         st->codecpar->codec_id = AV_CODEC_ID_ADPCM_4XM;
 204     } else if (st->codecpar->bits_per_coded_sample == 8) {
 205         st->codecpar->codec_id = AV_CODEC_ID_PCM_U8;
 206     } else
 207         st->codecpar->codec_id = AV_CODEC_ID_PCM_S16LE;
 208
 209     return 0;
 210 }
 211
 212 static int fourxm_read_header(AVFormatContext *s)
 213 {
 214     AVIOContext *pb = s->pb;
 215     unsigned int fourcc_tag;
 216     unsigned int size;
 217     int header_size;
 218     FourxmDemuxContext *fourxm = s->priv_data;
 219     unsigned char *header = NULL;
 220     int i, ret;
 221
 222     fourxm->track_count = 0;
 223     fourxm->tracks      = NULL;
 224     fourxm->fps         = (AVRational){1,1};
 225     fourxm->video_stream_index = -1;
 226
 227     /* skip the first 3 32-bit numbers */
 228     avio_skip(pb, 12);
 229
 230     /* check for LIST-HEAD */
 231     GET_LIST_HEADER();
 232     header_size = size - 4;
 233     if (fourcc_tag != HEAD_TAG || header_size < 0)
 234         return AVERROR_INVALIDDATA;
 235
 236     /* allocate space for the header and load the whole thing */
 237     header = av_malloc(header_size);
 238     if (!header)
 239         return AVERROR(ENOMEM);
 240     if (avio_read(pb, header, header_size) != header_size) {
 241         av_free(header);
 242         return AVERROR(EIO);
 243     }
 244
 245     /* take the lazy approach and search for any and all vtrk and strk chunks */
 246     for (i = 0; i < header_size - 8; i++) {
 247         fourcc_tag = AV_RL32(&header[i]);
 248         size       = AV_RL32(&header[i + 4]);
 249         if (size > header_size - i - 8 && (fourcc_tag == vtrk_TAG || fourcc_tag == strk_TAG)) {
 250             av_log(s, AV_LOG_ERROR, "chunk larger than array %d>%d\n", size, header_size - i - 8);
 251             ret = AVERROR_INVALIDDATA;
 252             goto fail;
 253         }
 254
 255         if (fourcc_tag == std__TAG) {
 256             if (header_size - i < 16) {
 257                 av_log(s, AV_LOG_ERROR, "std TAG truncated\n");
 258                 ret = AVERROR_INVALIDDATA;
 259                 goto fail;
 260             }
 261             fourxm->fps = av_d2q(av_int2float(AV_RL32(&header[i + 12])), 10000);
 262         } else if (fourcc_tag == vtrk_TAG) {
 263             if ((ret = parse_vtrk(s, fourxm, header + i, size,
 264                                   header_size - i)) < 0)
 265                 goto fail;
 266
 267             i += 8 + size;
 268         } else if (fourcc_tag == strk_TAG) {
 269             if ((ret = parse_strk(s, fourxm, header + i, size,
 270                                   header_size - i)) < 0)
 271                 goto fail;
 272
 273             i += 8 + size;
 274         }
 275     }
 276
 277     /* skip over the LIST-MOVI chunk (which is where the stream should be */
 278     GET_LIST_HEADER();
 279     if (fourcc_tag != MOVI_TAG) {
 280         ret = AVERROR_INVALIDDATA;
 281         goto fail;
 282     }
 283
 284     av_free(header);
 285     /* initialize context members */
 286     fourxm->video_pts = -1;  /* first frame will push to 0 */
 287
 288     return 0;
 289 fail:
 290     av_freep(&fourxm->tracks);
 291     av_free(header);
 292     return ret;
 293 }
 294
 295 static int fourxm_read_packet(AVFormatContext *s,
 296                               AVPacket *pkt)
 297 {
 298     FourxmDemuxContext *fourxm = s->priv_data;
 299     AVIOContext *pb            = s->pb;
 300     unsigned int fourcc_tag;
 301     unsigned int size;
 302     int ret = 0;
 303     unsigned int track_number;
 304     int packet_read = 0;
 305     unsigned char header[8];
 306     int64_t audio_frame_count;
 307
 308     while (!packet_read) {
 309         if ((ret = avio_read(s->pb, header, 8)) < 0)
 310             return ret;
 311         fourcc_tag = AV_RL32(&header[0]);
 312         size       = AV_RL32(&header[4]);
 313         if (avio_feof(pb))
 314             return AVERROR(EIO);
 315         switch (fourcc_tag) {
 316         case LIST_TAG:
 317             /* this is a good time to bump the video pts */
 318             fourxm->video_pts++;
 319
 320             /* skip the LIST-* tag and move on to the next fourcc */
 321             avio_rl32(pb);
 322             break;
 323
 324         case ifrm_TAG:
 325         case pfrm_TAG:
 326         case cfrm_TAG:
 327         case ifr2_TAG:
 328         case pfr2_TAG:
 329         case cfr2_TAG:
 330             /* allocate 8 more bytes than 'size' to account for fourcc
 331              * and size */
 332             if (size > INT_MAX - AV_INPUT_BUFFER_PADDING_SIZE - 8)
 333                 return AVERROR_INVALIDDATA;
 334             if (fourxm->video_stream_index < 0)
 335                 return AVERROR_INVALIDDATA;
 336             if ((ret = av_new_packet(pkt, size + 8)) < 0)
 337                 return ret;
 338             pkt->stream_index = fourxm->video_stream_index;
 339             pkt->pts          = fourxm->video_pts;
 340             pkt->pos          = avio_tell(s->pb);
 341             memcpy(pkt->data, header, 8);
 342             ret = avio_read(s->pb, &pkt->data[8], size);
 343
 344             if (ret < 0) {
 345                 av_packet_unref(pkt);
 346             } else {
 347                 packet_read = 1;
 348                 av_shrink_packet(pkt, ret + 8);
 349             }
 350             break;
 351
 352         case snd__TAG:
 353             track_number = avio_rl32(pb);
 354             avio_skip(pb, 4);
 355             size -= 8;
 356
 357             if (track_number < fourxm->track_count &&
 358                 fourxm->tracks[track_number].channels > 0) {
 359                 ret = av_get_packet(s->pb, pkt, size);
 360                 if (ret < 0)
 361                     return ret;
 362                 pkt->stream_index =
 363                     fourxm->tracks[track_number].stream_index;
 364                 pkt->pts    = fourxm->tracks[track_number].audio_pts;
 365                 packet_read = 1;
 366
 367                 /* pts accounting */
 368                 audio_frame_count = size;
 369                 if (fourxm->tracks[track_number].adpcm)
 370                     audio_frame_count -= 2 * (fourxm->tracks[track_number].channels);
 371                 audio_frame_count /= fourxm->tracks[track_number].channels;
 372                 if (fourxm->tracks[track_number].adpcm) {
 373                     audio_frame_count *= 2;
 374                 } else
 375                     audio_frame_count /=
 376                         (fourxm->tracks[track_number].bits / 8);
 377                 fourxm->tracks[track_number].audio_pts += audio_frame_count;
 378             } else {
 379                 avio_skip(pb, size);
 380             }
 381             break;
 382
 383         default:
 384             avio_skip(pb, size);
 385             break;
 386         }
 387     }
 388     return ret;
 389 }
 390
 391 static int fourxm_read_close(AVFormatContext *s)
 392 {
 393     FourxmDemuxContext *fourxm = s->priv_data;
 394
 395     av_freep(&fourxm->tracks);
 396
 397     return 0;
 398 }
 399
 400 AVInputFormat ff_fourxm_demuxer = {
 401     .name           = "4xm",
 402     .long_name      = NULL_IF_CONFIG_SMALL("4X Technologies"),
 403     .priv_data_size = sizeof(FourxmDemuxContext),
 404     .read_probe     = fourxm_probe,
 405     .read_header    = fourxm_read_header,
 406     .read_packet    = fourxm_read_packet,
 407     .read_close     = fourxm_read_close,
 408 };