summaryrefslogtreecommitdiff
path: root/libavformat/xwma.c (plain)
blob: 1c187729289b8c3a2d2c9769516529bf04593cdd
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
33typedef struct XWMAContext {
34 int64_t data_end;
35} XWMAContext;
36
37static int xwma_probe(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
44static 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 -1;
64 avio_rl32(pb); /* file size */
65 tag = avio_rl32(pb);
66 if (tag != MKTAG('X', 'W', 'M', 'A'))
67 return -1;
68
69 /* parse fmt header */
70 tag = avio_rl32(pb);
71 if (tag != MKTAG('f', 'm', 't', ' '))
72 return -1;
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 /* All xWMA files I have seen contained WMAv2 data. If there are files
84 * using WMA Pro or some other codec, then we need to figure out the right
85 * extradata for that. Thus, ask the user for feedback, but try to go on
86 * anyway.
87 */
88 if (st->codecpar->codec_id != AV_CODEC_ID_WMAV2 &&
89 st->codecpar->codec_id != AV_CODEC_ID_WMAPRO) {
90 avpriv_request_sample(s, "Unexpected codec (tag %s; id %d)",
91 av_fourcc2str(st->codecpar->codec_tag),
92 st->codecpar->codec_id);
93 } else {
94 /* In all xWMA files I have seen, there is no extradata. But the WMA
95 * codecs require extradata, so we provide our own fake extradata.
96 *
97 * First, check that there really was no extradata in the header. If
98 * there was, then try to use it, after asking the user to provide a
99 * sample of this unusual file.
100 */
101 if (st->codecpar->extradata_size != 0) {
102 /* Surprise, surprise: We *did* get some extradata. No idea
103 * if it will work, but just go on and try it, after asking
104 * the user for a sample.
105 */
106 avpriv_request_sample(s, "Unexpected extradata (%d bytes)",
107 st->codecpar->extradata_size);
108 } else if (st->codecpar->codec_id == AV_CODEC_ID_WMAPRO) {
109 if (ff_alloc_extradata(st->codecpar, 18))
110 return AVERROR(ENOMEM);
111
112 memset(st->codecpar->extradata, 0, st->codecpar->extradata_size);
113 st->codecpar->extradata[ 0] = st->codecpar->bits_per_coded_sample;
114 st->codecpar->extradata[14] = 224;
115 } else {
116 if (ff_alloc_extradata(st->codecpar, 6))
117 return AVERROR(ENOMEM);
118
119 memset(st->codecpar->extradata, 0, st->codecpar->extradata_size);
120 /* setup extradata with our experimentally obtained value */
121 st->codecpar->extradata[4] = 31;
122 }
123 }
124
125 if (!st->codecpar->channels) {
126 av_log(s, AV_LOG_WARNING, "Invalid channel count: %d\n",
127 st->codecpar->channels);
128 return AVERROR_INVALIDDATA;
129 }
130 if (!st->codecpar->bits_per_coded_sample) {
131 av_log(s, AV_LOG_WARNING, "Invalid bits_per_coded_sample: %d\n",
132 st->codecpar->bits_per_coded_sample);
133 return AVERROR_INVALIDDATA;
134 }
135
136 /* set the sample rate */
137 avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
138
139 /* parse the remaining RIFF chunks */
140 for (;;) {
141 if (pb->eof_reached) {
142 ret = AVERROR_EOF;
143 goto fail;
144 }
145 /* read next chunk tag */
146 tag = avio_rl32(pb);
147 size = avio_rl32(pb);
148 if (tag == MKTAG('d', 'a', 't', 'a')) {
149 /* We assume that the data chunk comes last. */
150 break;
151 } else if (tag == MKTAG('d','p','d','s')) {
152 /* Quoting the MSDN xWMA docs on the dpds chunk: "Contains the
153 * decoded packet cumulative data size array, each element is the
154 * number of bytes accumulated after the corresponding xWMA packet
155 * is decoded in order."
156 *
157 * Each packet has size equal to st->codecpar->block_align, which in
158 * all cases I saw so far was always 2230. Thus, we can use the
159 * dpds data to compute a seeking index.
160 */
161
162 /* Error out if there is more than one dpds chunk. */
163 if (dpds_table) {
164 av_log(s, AV_LOG_ERROR, "two dpds chunks present\n");
165 ret = AVERROR_INVALIDDATA;
166 goto fail;
167 }
168
169 /* Compute the number of entries in the dpds chunk. */
170 if (size & 3) { /* Size should be divisible by four */
171 av_log(s, AV_LOG_WARNING,
172 "dpds chunk size %"PRId64" not divisible by 4\n", size);
173 }
174 dpds_table_size = size / 4;
175 if (dpds_table_size == 0 || dpds_table_size >= INT_MAX / 4) {
176 av_log(s, AV_LOG_ERROR,
177 "dpds chunk size %"PRId64" invalid\n", size);
178 return AVERROR_INVALIDDATA;
179 }
180
181 /* Allocate some temporary storage to keep the dpds data around.
182 * for processing later on.
183 */
184 dpds_table = av_malloc_array(dpds_table_size, sizeof(uint32_t));
185 if (!dpds_table) {
186 return AVERROR(ENOMEM);
187 }
188
189 for (i = 0; i < dpds_table_size; ++i) {
190 dpds_table[i] = avio_rl32(pb);
191 size -= 4;
192 }
193 }
194 avio_skip(pb, size);
195 }
196
197 /* Determine overall data length */
198 if (size < 0) {
199 ret = AVERROR_INVALIDDATA;
200 goto fail;
201 }
202 if (!size) {
203 xwma->data_end = INT64_MAX;
204 } else
205 xwma->data_end = avio_tell(pb) + size;
206
207
208 if (dpds_table && dpds_table_size) {
209 int64_t cur_pos;
210 const uint32_t bytes_per_sample
211 = (st->codecpar->channels * st->codecpar->bits_per_coded_sample) >> 3;
212
213 /* Estimate the duration from the total number of output bytes. */
214 const uint64_t total_decoded_bytes = dpds_table[dpds_table_size - 1];
215
216 if (!bytes_per_sample) {
217 av_log(s, AV_LOG_ERROR,
218 "Invalid bits_per_coded_sample %d for %d channels\n",
219 st->codecpar->bits_per_coded_sample, st->codecpar->channels);
220 ret = AVERROR_INVALIDDATA;
221 goto fail;
222 }
223
224 st->duration = total_decoded_bytes / bytes_per_sample;
225
226 /* Use the dpds data to build a seek table. We can only do this after
227 * we know the offset to the data chunk, as we need that to determine
228 * the actual offset to each input block.
229 * Note: If we allowed ourselves to assume that the data chunk always
230 * follows immediately after the dpds block, we could of course guess
231 * the data block's start offset already while reading the dpds chunk.
232 * I decided against that, just in case other chunks ever are
233 * discovered.
234 */
235 cur_pos = avio_tell(pb);
236 for (i = 0; i < dpds_table_size; ++i) {
237 /* From the number of output bytes that would accumulate in the
238 * output buffer after decoding the first (i+1) packets, we compute
239 * an offset / timestamp pair.
240 */
241 av_add_index_entry(st,
242 cur_pos + (i+1) * st->codecpar->block_align, /* pos */
243 dpds_table[i] / bytes_per_sample, /* timestamp */
244 st->codecpar->block_align, /* size */
245 0, /* duration */
246 AVINDEX_KEYFRAME);
247 }
248 } else if (st->codecpar->bit_rate) {
249 /* No dpds chunk was present (or only an empty one), so estimate
250 * the total duration using the average bits per sample and the
251 * total data length.
252 */
253 st->duration = (size<<3) * st->codecpar->sample_rate / st->codecpar->bit_rate;
254 }
255
256fail:
257 av_free(dpds_table);
258
259 return ret;
260}
261
262static int xwma_read_packet(AVFormatContext *s, AVPacket *pkt)
263{
264 int ret, size;
265 int64_t left;
266 AVStream *st;
267 XWMAContext *xwma = s->priv_data;
268
269 st = s->streams[0];
270
271 left = xwma->data_end - avio_tell(s->pb);
272 if (left <= 0) {
273 return AVERROR_EOF;
274 }
275
276 /* read a single block; the default block size is 2230. */
277 size = (st->codecpar->block_align > 1) ? st->codecpar->block_align : 2230;
278 size = FFMIN(size, left);
279
280 ret = av_get_packet(s->pb, pkt, size);
281 if (ret < 0)
282 return ret;
283
284 pkt->stream_index = 0;
285 return ret;
286}
287
288AVInputFormat ff_xwma_demuxer = {
289 .name = "xwma",
290 .long_name = NULL_IF_CONFIG_SMALL("Microsoft xWMA"),
291 .priv_data_size = sizeof(XWMAContext),
292 .read_probe = xwma_probe,
293 .read_header = xwma_read_header,
294 .read_packet = xwma_read_packet,
295};
296
generated by cgit at 2024-05-17 06:12:03 (GMT)