libavformat/xwma.c

   1 /*
   2  * xWMA demuxer
   3  * Copyright (c) 2011 Max Horn
   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 <inttypes.h>
  23 #include <stdint.h>
  24
  25 #include "avformat.h"
  26 #include "internal.h"
  27 #include "riff.h"
  28
  29 /*
  30  * Demuxer for xWMA, a Microsoft audio container used by XAudio 2.
  31  */
  32
  33 typedef struct XWMAContext {
  34     int64_t data_end;
  35 } XWMAContext;
  36
  37 static int xwma_probe(const AVProbeData *p)
  38 {
  39     if (!memcmp(p->buf, "RIFF", 4) && !memcmp(p->buf + 8, "XWMA", 4))
  40         return AVPROBE_SCORE_MAX;
  41     return 0;
  42 }
  43
  44 static int xwma_read_header(AVFormatContext *s)
  45 {
  46     int64_t size;
  47     int ret = 0;
  48     uint32_t dpds_table_size = 0;
  49     uint32_t *dpds_table = NULL;
  50     unsigned int tag;
  51     AVIOContext *pb = s->pb;
  52     AVStream *st;
  53     XWMAContext *xwma = s->priv_data;
  54     int i;
  55
  56     /* The following code is mostly copied from wav.c, with some
  57      * minor alterations.
  58      */
  59
  60     /* check RIFF header */
  61     tag = avio_rl32(pb);
  62     if (tag != MKTAG('R', 'I', 'F', 'F'))
  63         return AVERROR_INVALIDDATA;
  64     avio_rl32(pb); /* file size */
  65     tag = avio_rl32(pb);
  66     if (tag != MKTAG('X', 'W', 'M', 'A'))
  67         return AVERROR_INVALIDDATA;
  68
  69     /* parse fmt header */
  70     tag = avio_rl32(pb);
  71     if (tag != MKTAG('f', 'm', 't', ' '))
  72         return AVERROR_INVALIDDATA;
  73     size = avio_rl32(pb);
  74     st = avformat_new_stream(s, NULL);
  75     if (!st)
  76         return AVERROR(ENOMEM);
  77
  78     ret = ff_get_wav_header(s, pb, st->codecpar, size, 0);
  79     if (ret < 0)
  80         return ret;
  81     st->need_parsing = AVSTREAM_PARSE_NONE;
  82
  83     /* XWMA encoder only allows a few channel/sample rate/bitrate combinations,
  84      * but some create identical files with fake bitrate (1ch 22050hz at
  85      * 20/48/192kbps are all 20kbps, with the exact same codec data).
  86      * Decoder needs correct bitrate to work, so it's normalized here. */
  87     if (st->codecpar->codec_id == AV_CODEC_ID_WMAV2) {
  88         int ch = st->codecpar->channels;
  89         int sr = st->codecpar->sample_rate;
  90         int br = st->codecpar->bit_rate;
  91
  92         if (ch == 1) {
  93             if (sr == 22050 && (br==48000 || br==192000))
  94                 br = 20000;
  95             else if (sr == 32000 && (br==48000 || br==192000))
  96                 br = 20000;
  97             else if (sr == 44100 && (br==96000 || br==192000))
  98                 br = 48000;
  99         }
 100         else if (ch == 2) {
 101             if (sr == 22050 && (br==48000 || br==192000))
 102                 br = 32000;
 103             else if (sr == 32000 && (br==192000))
 104                 br = 48000;
 105         }
 106
 107         st->codecpar->bit_rate = br;
 108     }
 109
 110     /* Normally xWMA can only contain WMAv2 with 1/2 channels,
 111      * and WMAPRO with 6 channels. */
 112     if (st->codecpar->codec_id != AV_CODEC_ID_WMAV2 &&
 113         st->codecpar->codec_id != AV_CODEC_ID_WMAPRO) {
 114         avpriv_request_sample(s, "Unexpected codec (tag %s; id %d)",
 115                               av_fourcc2str(st->codecpar->codec_tag),
 116                               st->codecpar->codec_id);
 117     } else {
 118         /* xWMA shouldn't have extradata. But the WMA codecs require it,
 119          * so we provide our own fake extradata.
 120          *
 121          * First, check that there really was no extradata in the header. If
 122          * there was, then try to use it, after asking the user to provide a
 123          * sample of this unusual file.
 124          */
 125         if (st->codecpar->extradata_size != 0) {
 126             /* Surprise, surprise: We *did* get some extradata. No idea
 127              * if it will work, but just go on and try it, after asking
 128              * the user for a sample.
 129              */
 130             avpriv_request_sample(s, "Unexpected extradata (%d bytes)",
 131                                   st->codecpar->extradata_size);
 132         } else if (st->codecpar->codec_id == AV_CODEC_ID_WMAPRO) {
 133             if ((ret = ff_alloc_extradata(st->codecpar, 18)) < 0)
 134                 return ret;
 135
 136             memset(st->codecpar->extradata, 0, st->codecpar->extradata_size);
 137             st->codecpar->extradata[ 0] = st->codecpar->bits_per_coded_sample;
 138             st->codecpar->extradata[14] = 224;
 139         } else {
 140             if ((ret = ff_alloc_extradata(st->codecpar, 6)) < 0)
 141                 return ret;
 142
 143             memset(st->codecpar->extradata, 0, st->codecpar->extradata_size);
 144             /* setup extradata with our experimentally obtained value */
 145             st->codecpar->extradata[4] = 31;
 146         }
 147     }
 148
 149     if (!st->codecpar->channels) {
 150         av_log(s, AV_LOG_WARNING, "Invalid channel count: %d\n",
 151                st->codecpar->channels);
 152         return AVERROR_INVALIDDATA;
 153     }
 154     if (!st->codecpar->bits_per_coded_sample || st->codecpar->bits_per_coded_sample > 64) {
 155         av_log(s, AV_LOG_WARNING, "Invalid bits_per_coded_sample: %d\n",
 156                st->codecpar->bits_per_coded_sample);
 157         return AVERROR_INVALIDDATA;
 158     }
 159
 160     /* set the sample rate */
 161     avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
 162
 163     /* parse the remaining RIFF chunks */
 164     for (;;) {
 165         if (pb->eof_reached) {
 166             ret = AVERROR_EOF;
 167             goto fail;
 168         }
 169         /* read next chunk tag */
 170         tag = avio_rl32(pb);
 171         size = avio_rl32(pb);
 172         if (tag == MKTAG('d', 'a', 't', 'a')) {
 173             /* We assume that the data chunk comes last. */
 174             break;
 175         } else if (tag == MKTAG('d','p','d','s')) {
 176             /* Quoting the MSDN xWMA docs on the dpds chunk: "Contains the
 177              * decoded packet cumulative data size array, each element is the
 178              * number of bytes accumulated after the corresponding xWMA packet
 179              * is decoded in order."
 180              *
 181              * Each packet has size equal to st->codecpar->block_align, which in
 182              * all cases I saw so far was always 2230. Thus, we can use the
 183              * dpds data to compute a seeking index.
 184              */
 185
 186             /* Error out if there is more than one dpds chunk. */
 187             if (dpds_table) {
 188                 av_log(s, AV_LOG_ERROR, "two dpds chunks present\n");
 189                 ret = AVERROR_INVALIDDATA;
 190                 goto fail;
 191             }
 192
 193             /* Compute the number of entries in the dpds chunk. */
 194             if (size & 3) {  /* Size should be divisible by four */
 195                 av_log(s, AV_LOG_WARNING,
 196                        "dpds chunk size %"PRId64" not divisible by 4\n", size);
 197             }
 198             dpds_table_size = size / 4;
 199             if (dpds_table_size == 0 || dpds_table_size >= INT_MAX / 4) {
 200                 av_log(s, AV_LOG_ERROR,
 201                        "dpds chunk size %"PRId64" invalid\n", size);
 202                 return AVERROR_INVALIDDATA;
 203             }
 204
 205             /* Allocate some temporary storage to keep the dpds data around.
 206              * for processing later on.
 207              */
 208             dpds_table = av_malloc_array(dpds_table_size, sizeof(uint32_t));
 209             if (!dpds_table) {
 210                 return AVERROR(ENOMEM);
 211             }
 212
 213             for (i = 0; i < dpds_table_size; ++i) {
 214                 if (avio_feof(pb)) {
 215                     ret = AVERROR_INVALIDDATA;
 216                     goto fail;
 217                 }
 218                 dpds_table[i] = avio_rl32(pb);
 219                 size -= 4;
 220             }
 221         }
 222         avio_skip(pb, size);
 223     }
 224
 225     /* Determine overall data length */
 226     if (size < 0) {
 227         ret = AVERROR_INVALIDDATA;
 228         goto fail;
 229     }
 230     if (!size) {
 231         xwma->data_end = INT64_MAX;
 232     } else
 233         xwma->data_end = avio_tell(pb) + size;
 234
 235
 236     if (dpds_table && dpds_table_size) {
 237         int64_t cur_pos;
 238         const uint32_t bytes_per_sample
 239                 = (st->codecpar->channels * st->codecpar->bits_per_coded_sample) >> 3;
 240
 241         /* Estimate the duration from the total number of output bytes. */
 242         const uint64_t total_decoded_bytes = dpds_table[dpds_table_size - 1];
 243
 244         if (!bytes_per_sample) {
 245             av_log(s, AV_LOG_ERROR,
 246                    "Invalid bits_per_coded_sample %d for %d channels\n",
 247                    st->codecpar->bits_per_coded_sample, st->codecpar->channels);
 248             ret = AVERROR_INVALIDDATA;
 249             goto fail;
 250         }
 251
 252         st->duration = total_decoded_bytes / bytes_per_sample;
 253
 254         /* Use the dpds data to build a seek table.  We can only do this after
 255          * we know the offset to the data chunk, as we need that to determine
 256          * the actual offset to each input block.
 257          * Note: If we allowed ourselves to assume that the data chunk always
 258          * follows immediately after the dpds block, we could of course guess
 259          * the data block's start offset already while reading the dpds chunk.
 260          * I decided against that, just in case other chunks ever are
 261          * discovered.
 262          */
 263         cur_pos = avio_tell(pb);
 264         for (i = 0; i < dpds_table_size; ++i) {
 265             /* From the number of output bytes that would accumulate in the
 266              * output buffer after decoding the first (i+1) packets, we compute
 267              * an offset / timestamp pair.
 268              */
 269             av_add_index_entry(st,
 270                                cur_pos + (i+1) * st->codecpar->block_align, /* pos */
 271                                dpds_table[i] / bytes_per_sample,            /* timestamp */
 272                                st->codecpar->block_align,                   /* size */
 273                                0,                                           /* duration */
 274                                AVINDEX_KEYFRAME);
 275         }
 276     } else if (st->codecpar->bit_rate) {
 277         /* No dpds chunk was present (or only an empty one), so estimate
 278          * the total duration using the average bits per sample and the
 279          * total data length.
 280          */
 281         st->duration = av_rescale((size<<3), st->codecpar->sample_rate, st->codecpar->bit_rate);
 282     }
 283
 284 fail:
 285     av_free(dpds_table);
 286
 287     return ret;
 288 }
 289
 290 static int xwma_read_packet(AVFormatContext *s, AVPacket *pkt)
 291 {
 292     int ret, size;
 293     int64_t left;
 294     AVStream *st;
 295     XWMAContext *xwma = s->priv_data;
 296
 297     st = s->streams[0];
 298
 299     left = xwma->data_end - avio_tell(s->pb);
 300     if (left <= 0) {
 301         return AVERROR_EOF;
 302     }
 303
 304     /* read a single block; the default block size is 2230. */
 305     size = (st->codecpar->block_align > 1) ? st->codecpar->block_align : 2230;
 306     size = FFMIN(size, left);
 307
 308     ret  = av_get_packet(s->pb, pkt, size);
 309     if (ret < 0)
 310         return ret;
 311
 312     pkt->stream_index = 0;
 313     return ret;
 314 }
 315
 316 AVInputFormat ff_xwma_demuxer = {
 317     .name           = "xwma",
 318     .long_name      = NULL_IF_CONFIG_SMALL("Microsoft xWMA"),
 319     .priv_data_size = sizeof(XWMAContext),
 320     .read_probe     = xwma_probe,
 321     .read_header    = xwma_read_header,
 322     .read_packet    = xwma_read_packet,
 323 };