libavformat/xwma.c

   1 /*
   2  * xWMA demuxer
   3  * Copyright (c) 2011 Max Horn
   4  *
   5  * This file is part of Libav.
   6  *
   7  * Libav 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  * Libav 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 Libav; 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
  24 #include "avformat.h"
  25 #include "internal.h"
  26 #include "riff.h"
  27
  28 /*
  29  * Demuxer for xWMA, a Microsoft audio container used by XAudio 2.
  30  */
  31
  32 typedef struct {
  33     int64_t data_end;
  34 } XWMAContext;
  35
  36 static int xwma_probe(AVProbeData *p)
  37 {
  38     if (!memcmp(p->buf, "RIFF", 4) && !memcmp(p->buf + 8, "XWMA", 4))
  39         return AVPROBE_SCORE_MAX;
  40     return 0;
  41 }
  42
  43 static int xwma_read_header(AVFormatContext *s)
  44 {
  45     int64_t size;
  46     int ret;
  47     uint32_t dpds_table_size = 0;
  48     uint32_t *dpds_table = 0;
  49     unsigned int tag;
  50     AVIOContext *pb = s->pb;
  51     AVStream *st;
  52     XWMAContext *xwma = s->priv_data;
  53     int i;
  54
  55     /* The following code is mostly copied from wav.c, with some
  56      * minor alterations.
  57      */
  58
  59     /* check RIFF header */
  60     tag = avio_rl32(pb);
  61     if (tag != MKTAG('R', 'I', 'F', 'F'))
  62         return -1;
  63     avio_rl32(pb); /* file size */
  64     tag = avio_rl32(pb);
  65     if (tag != MKTAG('X', 'W', 'M', 'A'))
  66         return -1;
  67
  68     /* parse fmt header */
  69     tag = avio_rl32(pb);
  70     if (tag != MKTAG('f', 'm', 't', ' '))
  71         return -1;
  72     size = avio_rl32(pb);
  73     st = avformat_new_stream(s, NULL);
  74     if (!st)
  75         return AVERROR(ENOMEM);
  76
  77     ret = ff_get_wav_header(pb, st->codec, size);
  78     if (ret < 0)
  79         return ret;
  80     st->need_parsing = AVSTREAM_PARSE_NONE;
  81
  82     /* All xWMA files I have seen contained WMAv2 data. If there are files
  83      * using WMA Pro or some other codec, then we need to figure out the right
  84      * extradata for that. Thus, ask the user for feedback, but try to go on
  85      * anyway.
  86      */
  87     if (st->codec->codec_id != AV_CODEC_ID_WMAV2) {
  88         av_log(s, AV_LOG_WARNING, "unexpected codec (tag 0x04%x; id %d)\n",
  89                               st->codec->codec_tag, st->codec->codec_id);
  90         av_log_ask_for_sample(s, NULL);
  91     } else {
  92         /* In all xWMA files I have seen, there is no extradata. But the WMA
  93          * codecs require extradata, so we provide our own fake extradata.
  94          *
  95          * First, check that there really was no extradata in the header. If
  96          * there was, then try to use it, after asking the user to provide a
  97          * sample of this unusual file.
  98          */
  99         if (st->codec->extradata_size != 0) {
 100             /* Surprise, surprise: We *did* get some extradata. No idea
 101              * if it will work, but just go on and try it, after asking
 102              * the user for a sample.
 103              */
 104             av_log(s, AV_LOG_WARNING, "unexpected extradata (%d bytes)\n",
 105                                   st->codec->extradata_size);
 106             av_log_ask_for_sample(s, NULL);
 107         } else {
 108             st->codec->extradata_size = 6;
 109             st->codec->extradata      = av_mallocz(6 + FF_INPUT_BUFFER_PADDING_SIZE);
 110             if (!st->codec->extradata)
 111                 return AVERROR(ENOMEM);
 112
 113             /* setup extradata with our experimentally obtained value */
 114             st->codec->extradata[4] = 31;
 115         }
 116     }
 117
 118     if (!st->codec->channels) {
 119         av_log(s, AV_LOG_WARNING, "Invalid channel count: %d\n",
 120                st->codec->channels);
 121         return AVERROR_INVALIDDATA;
 122     }
 123     if (!st->codec->bits_per_coded_sample) {
 124         av_log(s, AV_LOG_WARNING, "Invalid bits_per_coded_sample: %d\n",
 125                st->codec->bits_per_coded_sample);
 126         return AVERROR_INVALIDDATA;
 127     }
 128
 129     /* set the sample rate */
 130     avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate);
 131
 132     /* parse the remaining RIFF chunks */
 133     for (;;) {
 134         if (pb->eof_reached)
 135             return -1;
 136         /* read next chunk tag */
 137         tag = avio_rl32(pb);
 138         size = avio_rl32(pb);
 139         if (tag == MKTAG('d', 'a', 't', 'a')) {
 140             /* We assume that the data chunk comes last. */
 141             break;
 142         } else if (tag == MKTAG('d','p','d','s')) {
 143             /* Quoting the MSDN xWMA docs on the dpds chunk: "Contains the
 144              * decoded packet cumulative data size array, each element is the
 145              * number of bytes accumulated after the corresponding xWMA packet
 146              * is decoded in order."
 147              *
 148              * Each packet has size equal to st->codec->block_align, which in
 149              * all cases I saw so far was always 2230. Thus, we can use the
 150              * dpds data to compute a seeking index.
 151              */
 152
 153             /* Error out if there is more than one dpds chunk. */
 154             if (dpds_table) {
 155                 av_log(s, AV_LOG_ERROR, "two dpds chunks present\n");
 156                 return -1;
 157             }
 158
 159             /* Compute the number of entries in the dpds chunk. */
 160             if (size & 3) {  /* Size should be divisible by four */
 161                 av_log(s, AV_LOG_WARNING,
 162                        "dpds chunk size %"PRId64" not divisible by 4\n", size);
 163             }
 164             dpds_table_size = size / 4;
 165             if (dpds_table_size == 0 || dpds_table_size >= INT_MAX / 4) {
 166                 av_log(s, AV_LOG_ERROR,
 167                        "dpds chunk size %"PRId64" invalid\n", size);
 168                 return -1;
 169             }
 170
 171             /* Allocate some temporary storage to keep the dpds data around.
 172              * for processing later on.
 173              */
 174             dpds_table = av_malloc(dpds_table_size * sizeof(uint32_t));
 175             if (!dpds_table) {
 176                 return AVERROR(ENOMEM);
 177             }
 178
 179             for (i = 0; i < dpds_table_size; ++i) {
 180                 dpds_table[i] = avio_rl32(pb);
 181                 size -= 4;
 182             }
 183         }
 184         avio_skip(pb, size);
 185     }
 186
 187     /* Determine overall data length */
 188     if (size < 0)
 189         return -1;
 190     if (!size) {
 191         xwma->data_end = INT64_MAX;
 192     } else
 193         xwma->data_end = avio_tell(pb) + size;
 194
 195
 196     if (dpds_table && dpds_table_size) {
 197         int64_t cur_pos;
 198         const uint32_t bytes_per_sample
 199                 = (st->codec->channels * st->codec->bits_per_coded_sample) >> 3;
 200
 201         /* Estimate the duration from the total number of output bytes. */
 202         const uint64_t total_decoded_bytes = dpds_table[dpds_table_size - 1];
 203         st->duration = total_decoded_bytes / bytes_per_sample;
 204
 205         /* Use the dpds data to build a seek table.  We can only do this after
 206          * we know the offset to the data chunk, as we need that to determine
 207          * the actual offset to each input block.
 208          * Note: If we allowed ourselves to assume that the data chunk always
 209          * follows immediately after the dpds block, we could of course guess
 210          * the data block's start offset already while reading the dpds chunk.
 211          * I decided against that, just in case other chunks ever are
 212          * discovered.
 213          */
 214         cur_pos = avio_tell(pb);
 215         for (i = 0; i < dpds_table_size; ++i) {
 216             /* From the number of output bytes that would accumulate in the
 217              * output buffer after decoding the first (i+1) packets, we compute
 218              * an offset / timestamp pair.
 219              */
 220             av_add_index_entry(st,
 221                                cur_pos + (i+1) * st->codec->block_align, /* pos */
 222                                dpds_table[i] / bytes_per_sample,         /* timestamp */
 223                                st->codec->block_align,                   /* size */
 224                                0,                                        /* duration */
 225                                AVINDEX_KEYFRAME);
 226         }
 227     } else if (st->codec->bit_rate) {
 228         /* No dpds chunk was present (or only an empty one), so estimate
 229          * the total duration using the average bits per sample and the
 230          * total data length.
 231          */
 232         st->duration = (size<<3) * st->codec->sample_rate / st->codec->bit_rate;
 233     }
 234
 235     av_free(dpds_table);
 236
 237     return 0;
 238 }
 239
 240 static int xwma_read_packet(AVFormatContext *s, AVPacket *pkt)
 241 {
 242     int ret, size;
 243     int64_t left;
 244     AVStream *st;
 245     XWMAContext *xwma = s->priv_data;
 246
 247     st = s->streams[0];
 248
 249     left = xwma->data_end - avio_tell(s->pb);
 250     if (left <= 0) {
 251         return AVERROR_EOF;
 252     }
 253
 254     /* read a single block; the default block size is 2230. */
 255     size = (st->codec->block_align > 1) ? st->codec->block_align : 2230;
 256     size = FFMIN(size, left);
 257
 258     ret  = av_get_packet(s->pb, pkt, size);
 259     if (ret < 0)
 260         return ret;
 261
 262     pkt->stream_index = 0;
 263     return ret;
 264 }
 265
 266 AVInputFormat ff_xwma_demuxer = {
 267     .name           = "xwma",
 268     .long_name      = NULL_IF_CONFIG_SMALL("Microsoft xWMA"),
 269     .priv_data_size = sizeof(XWMAContext),
 270     .read_probe     = xwma_probe,
 271     .read_header    = xwma_read_header,
 272     .read_packet    = xwma_read_packet,
 273 };