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 "libavutil/mem.h"
  26 #include "avformat.h"
  27 #include "demux.h"
  28 #include "internal.h"
  29 #include "riff.h"
  30
  31 /*
  32  * Demuxer for xWMA, a Microsoft audio container used by XAudio 2.
  33  */
  34
  35 typedef struct XWMAContext {
  36     int64_t data_end;
  37 } XWMAContext;
  38
  39 static int xwma_probe(const AVProbeData *p)
  40 {
  41     if (!memcmp(p->buf, "RIFF", 4) && !memcmp(p->buf + 8, "XWMA", 4))
  42         return AVPROBE_SCORE_MAX;
  43     return 0;
  44 }
  45
  46 static int xwma_read_header(AVFormatContext *s)
  47 {
  48     int64_t size;
  49     int ret = 0;
  50     uint32_t dpds_table_size = 0;
  51     uint32_t *dpds_table = NULL;
  52     unsigned int tag;
  53     AVIOContext *pb = s->pb;
  54     AVStream *st;
  55     XWMAContext *xwma = s->priv_data;
  56     int i;
  57
  58     /* The following code is mostly copied from wav.c, with some
  59      * minor alterations.
  60      */
  61
  62     /* check RIFF header */
  63     tag = avio_rl32(pb);
  64     if (tag != MKTAG('R', 'I', 'F', 'F'))
  65         return AVERROR_INVALIDDATA;
  66     avio_rl32(pb); /* file size */
  67     tag = avio_rl32(pb);
  68     if (tag != MKTAG('X', 'W', 'M', 'A'))
  69         return AVERROR_INVALIDDATA;
  70
  71     /* parse fmt header */
  72     tag = avio_rl32(pb);
  73     if (tag != MKTAG('f', 'm', 't', ' '))
  74         return AVERROR_INVALIDDATA;
  75     size = avio_rl32(pb);
  76     st = avformat_new_stream(s, NULL);
  77     if (!st)
  78         return AVERROR(ENOMEM);
  79
  80     ret = ff_get_wav_header(s, pb, st->codecpar, size, 0);
  81     if (ret < 0)
  82         return ret;
  83     ffstream(st)->need_parsing = AVSTREAM_PARSE_NONE;
  84
  85     /* XWMA encoder only allows a few channel/sample rate/bitrate combinations,
  86      * but some create identical files with fake bitrate (1ch 22050hz at
  87      * 20/48/192kbps are all 20kbps, with the exact same codec data).
  88      * Decoder needs correct bitrate to work, so it's normalized here. */
  89     if (st->codecpar->codec_id == AV_CODEC_ID_WMAV2) {
  90         int ch = st->codecpar->ch_layout.nb_channels;
  91         int sr = st->codecpar->sample_rate;
  92         int br = st->codecpar->bit_rate;
  93
  94         if (ch == 1) {
  95             if (sr == 22050 && (br==48000 || br==192000))
  96                 br = 20000;
  97             else if (sr == 32000 && (br==48000 || br==192000))
  98                 br = 20000;
  99             else if (sr == 44100 && (br==96000 || br==192000))
 100                 br = 48000;
 101         }
 102         else if (ch == 2) {
 103             if (sr == 22050 && (br==48000 || br==192000))
 104                 br = 32000;
 105             else if (sr == 32000 && (br==192000))
 106                 br = 48000;
 107         }
 108
 109         st->codecpar->bit_rate = br;
 110     }
 111
 112     /* Normally xWMA can only contain WMAv2 with 1/2 channels,
 113      * and WMAPRO with 6 channels. */
 114     if (st->codecpar->codec_id != AV_CODEC_ID_WMAV2 &&
 115         st->codecpar->codec_id != AV_CODEC_ID_WMAPRO) {
 116         avpriv_request_sample(s, "Unexpected codec (tag %s; id %d)",
 117                               av_fourcc2str(st->codecpar->codec_tag),
 118                               st->codecpar->codec_id);
 119     } else {
 120         /* xWMA shouldn't have extradata. But the WMA codecs require it,
 121          * so we provide our own fake extradata.
 122          *
 123          * First, check that there really was no extradata in the header. If
 124          * there was, then try to use it, after asking the user to provide a
 125          * sample of this unusual file.
 126          */
 127         if (st->codecpar->extradata_size != 0) {
 128             /* Surprise, surprise: We *did* get some extradata. No idea
 129              * if it will work, but just go on and try it, after asking
 130              * the user for a sample.
 131              */
 132             avpriv_request_sample(s, "Unexpected extradata (%d bytes)",
 133                                   st->codecpar->extradata_size);
 134         } else if (st->codecpar->codec_id == AV_CODEC_ID_WMAPRO) {
 135             if ((ret = ff_alloc_extradata(st->codecpar, 18)) < 0)
 136                 return ret;
 137
 138             memset(st->codecpar->extradata, 0, st->codecpar->extradata_size);
 139             st->codecpar->extradata[ 0] = st->codecpar->bits_per_coded_sample;
 140             st->codecpar->extradata[14] = 224;
 141         } else {
 142             if ((ret = ff_alloc_extradata(st->codecpar, 6)) < 0)
 143                 return ret;
 144
 145             memset(st->codecpar->extradata, 0, st->codecpar->extradata_size);
 146             /* setup extradata with our experimentally obtained value */
 147             st->codecpar->extradata[4] = 31;
 148         }
 149     }
 150
 151     if (!av_channel_layout_check(&st->codecpar->ch_layout)) {
 152         av_log(s, AV_LOG_WARNING, "Invalid channel count: %d\n",
 153                st->codecpar->ch_layout.nb_channels);
 154         return AVERROR_INVALIDDATA;
 155     }
 156     if (!st->codecpar->bits_per_coded_sample || st->codecpar->bits_per_coded_sample > 64) {
 157         av_log(s, AV_LOG_WARNING, "Invalid bits_per_coded_sample: %d\n",
 158                st->codecpar->bits_per_coded_sample);
 159         return AVERROR_INVALIDDATA;
 160     }
 161
 162     /* set the sample rate */
 163     avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
 164
 165     /* parse the remaining RIFF chunks */
 166     for (;;) {
 167         if (pb->eof_reached) {
 168             ret = AVERROR_EOF;
 169             goto fail;
 170         }
 171         /* read next chunk tag */
 172         tag = avio_rl32(pb);
 173         size = avio_rl32(pb);
 174         if (tag == MKTAG('d', 'a', 't', 'a')) {
 175             /* We assume that the data chunk comes last. */
 176             break;
 177         } else if (tag == MKTAG('d','p','d','s')) {
 178             /* Quoting the MSDN xWMA docs on the dpds chunk: "Contains the
 179              * decoded packet cumulative data size array, each element is the
 180              * number of bytes accumulated after the corresponding xWMA packet
 181              * is decoded in order."
 182              *
 183              * Each packet has size equal to st->codecpar->block_align, which in
 184              * all cases I saw so far was always 2230. Thus, we can use the
 185              * dpds data to compute a seeking index.
 186              */
 187
 188             /* Error out if there is more than one dpds chunk. */
 189             if (dpds_table) {
 190                 av_log(s, AV_LOG_ERROR, "two dpds chunks present\n");
 191                 ret = AVERROR_INVALIDDATA;
 192                 goto fail;
 193             }
 194
 195             /* Compute the number of entries in the dpds chunk. */
 196             if (size & 3) {  /* Size should be divisible by four */
 197                 av_log(s, AV_LOG_WARNING,
 198                        "dpds chunk size %"PRId64" not divisible by 4\n", size);
 199             }
 200             dpds_table_size = size / 4;
 201             if (dpds_table_size == 0 || dpds_table_size >= INT_MAX / 4) {
 202                 av_log(s, AV_LOG_ERROR,
 203                        "dpds chunk size %"PRId64" invalid\n", size);
 204                 return AVERROR_INVALIDDATA;
 205             }
 206
 207             /* Allocate some temporary storage to keep the dpds data around.
 208              * for processing later on.
 209              */
 210             dpds_table = av_malloc_array(dpds_table_size, sizeof(uint32_t));
 211             if (!dpds_table) {
 212                 return AVERROR(ENOMEM);
 213             }
 214
 215             for (i = 0; i < dpds_table_size; ++i) {
 216                 if (avio_feof(pb)) {
 217                     ret = AVERROR_INVALIDDATA;
 218                     goto fail;
 219                 }
 220                 dpds_table[i] = avio_rl32(pb);
 221                 size -= 4;
 222             }
 223         }
 224         avio_skip(pb, size);
 225     }
 226
 227     /* Determine overall data length */
 228     if (size < 0) {
 229         ret = AVERROR_INVALIDDATA;
 230         goto fail;
 231     }
 232     if (!size) {
 233         xwma->data_end = INT64_MAX;
 234     } else
 235         xwma->data_end = avio_tell(pb) + size;
 236
 237
 238     if (dpds_table && dpds_table_size) {
 239         int64_t cur_pos;
 240         const uint32_t bytes_per_sample
 241                 = (st->codecpar->ch_layout.nb_channels * st->codecpar->bits_per_coded_sample) >> 3;
 242
 243         /* Estimate the duration from the total number of output bytes. */
 244         const uint64_t total_decoded_bytes = dpds_table[dpds_table_size - 1];
 245
 246         if (!bytes_per_sample) {
 247             av_log(s, AV_LOG_ERROR,
 248                    "Invalid bits_per_coded_sample %d for %d channels\n",
 249                    st->codecpar->bits_per_coded_sample, st->codecpar->ch_layout.nb_channels);
 250             ret = AVERROR_INVALIDDATA;
 251             goto fail;
 252         }
 253
 254         st->duration = total_decoded_bytes / bytes_per_sample;
 255
 256         /* Use the dpds data to build a seek table.  We can only do this after
 257          * we know the offset to the data chunk, as we need that to determine
 258          * the actual offset to each input block.
 259          * Note: If we allowed ourselves to assume that the data chunk always
 260          * follows immediately after the dpds block, we could of course guess
 261          * the data block's start offset already while reading the dpds chunk.
 262          * I decided against that, just in case other chunks ever are
 263          * discovered.
 264          */
 265         cur_pos = avio_tell(pb);
 266         for (i = 0; i < dpds_table_size; ++i) {
 267             /* From the number of output bytes that would accumulate in the
 268              * output buffer after decoding the first (i+1) packets, we compute
 269              * an offset / timestamp pair.
 270              */
 271             av_add_index_entry(st,
 272                                cur_pos + (i+1) * st->codecpar->block_align, /* pos */
 273                                dpds_table[i] / bytes_per_sample,            /* timestamp */
 274                                st->codecpar->block_align,                   /* size */
 275                                0,                                           /* duration */
 276                                AVINDEX_KEYFRAME);
 277         }
 278     } else if (st->codecpar->bit_rate) {
 279         /* No dpds chunk was present (or only an empty one), so estimate
 280          * the total duration using the average bits per sample and the
 281          * total data length.
 282          */
 283         st->duration = av_rescale((size<<3), st->codecpar->sample_rate, st->codecpar->bit_rate);
 284     }
 285
 286 fail:
 287     av_free(dpds_table);
 288
 289     return ret;
 290 }
 291
 292 static int xwma_read_packet(AVFormatContext *s, AVPacket *pkt)
 293 {
 294     int ret, size;
 295     int64_t left;
 296     AVStream *st;
 297     XWMAContext *xwma = s->priv_data;
 298
 299     st = s->streams[0];
 300
 301     left = xwma->data_end - avio_tell(s->pb);
 302     if (left <= 0) {
 303         return AVERROR_EOF;
 304     }
 305
 306     /* read a single block; the default block size is 2230. */
 307     size = (st->codecpar->block_align > 1) ? st->codecpar->block_align : 2230;
 308     size = FFMIN(size, left);
 309
 310     ret  = av_get_packet(s->pb, pkt, size);
 311     if (ret < 0)
 312         return ret;
 313
 314     pkt->stream_index = 0;
 315     return ret;
 316 }
 317
 318 const FFInputFormat ff_xwma_demuxer = {
 319     .p.name         = "xwma",
 320     .p.long_name    = NULL_IF_CONFIG_SMALL("Microsoft xWMA"),
 321     .priv_data_size = sizeof(XWMAContext),
 322     .read_probe     = xwma_probe,
 323     .read_header    = xwma_read_header,
 324     .read_packet    = xwma_read_packet,
 325 };