FFmpeg
westwood_vqa.c
Go to the documentation of this file.
1 /*
2  * Westwood Studios VQA Format Demuxer
3  * Copyright (c) 2003 Mike Melanson <melanson@pcisys.net>
4  * Copyright (c) 2021 Pekka Väänänen <pekka.vaananen@iki.fi>
5  *
6  * This file is part of FFmpeg.
7  *
8  * FFmpeg is free software; you can redistribute it and/or
9  * modify it under the terms of the GNU Lesser General Public
10  * License as published by the Free Software Foundation; either
11  * version 2.1 of the License, or (at your option) any later version.
12  *
13  * FFmpeg is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16  * Lesser General Public License for more details.
17  *
18  * You should have received a copy of the GNU Lesser General Public
19  * License along with FFmpeg; if not, write to the Free Software
20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21  */
22 
23 /**
24  * @file
25  * Westwood Studios VQA file demuxer
26  * by Mike Melanson (melanson@pcisys.net)
27  * for more information on the Westwood file formats, visit:
28  * http://www.pcisys.net/~melanson/codecs/
29  * http://www.geocities.com/SiliconValley/8682/aud3.txt
30  */
31 
32 #include "libavutil/intreadwrite.h"
33 #include "avformat.h"
34 #include "avio_internal.h"
35 #include "demux.h"
36 #include "internal.h"
37 
38 #define FORM_TAG MKBETAG('F', 'O', 'R', 'M')
39 #define WVQA_TAG MKBETAG('W', 'V', 'Q', 'A')
40 #define VQHD_TAG MKBETAG('V', 'Q', 'H', 'D')
41 #define FINF_TAG MKBETAG('F', 'I', 'N', 'F')
42 #define SND0_TAG MKBETAG('S', 'N', 'D', '0')
43 #define SND1_TAG MKBETAG('S', 'N', 'D', '1')
44 #define SND2_TAG MKBETAG('S', 'N', 'D', '2')
45 #define VQFR_TAG MKBETAG('V', 'Q', 'F', 'R')
46 #define VQFL_TAG MKBETAG('V', 'Q', 'F', 'L')
47 
48 /* don't know what these tags are for, but acknowledge their existence */
49 #define CINF_TAG MKBETAG('C', 'I', 'N', 'F')
50 #define CINH_TAG MKBETAG('C', 'I', 'N', 'H')
51 #define CIND_TAG MKBETAG('C', 'I', 'N', 'D')
52 #define LINF_TAG MKBETAG('L', 'I', 'N', 'F')
53 #define PINF_TAG MKBETAG('P', 'I', 'N', 'F')
54 #define PINH_TAG MKBETAG('P', 'I', 'N', 'H')
55 #define PIND_TAG MKBETAG('P', 'I', 'N', 'D')
56 #define CMDS_TAG MKBETAG('C', 'M', 'D', 'S')
57 #define SN2J_TAG MKBETAG('S', 'N', '2', 'J')
58 #define VIEW_TAG MKBETAG('V', 'I', 'E', 'W')
59 #define ZBUF_TAG MKBETAG('Z', 'B', 'U', 'F')
60 
61 #define VQA_HEADER_SIZE 0x2A
62 #define VQA_PREAMBLE_SIZE 8
63 
64 typedef struct WsVqaDemuxContext {
65  int version;
66  int bps;
67  int channels;
71  int64_t vqfl_chunk_pos;
74 
75 static int wsvqa_probe(const AVProbeData *p)
76 {
77  /* need 12 bytes to qualify */
78  if (p->buf_size < 12)
79  return 0;
80 
81  /* check for the VQA signatures */
82  if ((AV_RB32(&p->buf[0]) != FORM_TAG) ||
83  (AV_RB32(&p->buf[8]) != WVQA_TAG))
84  return 0;
85 
86  return AVPROBE_SCORE_MAX;
87 }
88 
90 {
91  WsVqaDemuxContext *wsvqa = s->priv_data;
92  AVIOContext *pb = s->pb;
93  AVStream *st;
94  uint8_t *header;
95  uint8_t scratch[VQA_PREAMBLE_SIZE];
96  uint32_t chunk_tag;
97  uint32_t chunk_size;
98  int fps, ret;
99 
100  /* initialize the video decoder stream */
101  st = avformat_new_stream(s, NULL);
102  if (!st)
103  return AVERROR(ENOMEM);
104  st->start_time = 0;
105  wsvqa->video_stream_index = st->index;
108  st->codecpar->codec_tag = 0; /* no fourcc */
109 
110  /* skip to the start of the VQA header */
111  avio_seek(pb, 20, SEEK_SET);
112 
113  /* the VQA header needs to go to the decoder */
114  if ((ret = ff_get_extradata(s, st->codecpar, pb, VQA_HEADER_SIZE)) < 0)
115  return ret;
116  header = st->codecpar->extradata;
117  st->codecpar->width = AV_RL16(&header[6]);
118  st->codecpar->height = AV_RL16(&header[8]);
119  fps = header[12];
120  st->nb_frames =
121  st->duration = AV_RL16(&header[4]);
122  if (fps < 1 || fps > 30) {
123  av_log(s, AV_LOG_ERROR, "invalid fps: %d\n", fps);
124  return AVERROR_INVALIDDATA;
125  }
126  avpriv_set_pts_info(st, 64, 1, fps);
127 
128  wsvqa->version = AV_RL16(&header[ 0]);
129  wsvqa->sample_rate = AV_RL16(&header[24]);
130  wsvqa->channels = header[26];
131  wsvqa->bps = header[27];
132  wsvqa->audio_stream_index = -1;
133  wsvqa->vqfl_chunk_pos = 0;
134  wsvqa->vqfl_chunk_size = 0;
135 
136  s->ctx_flags |= AVFMTCTX_NOHEADER;
137 
138  /* there are 0 or more chunks before the FINF chunk; iterate until
139  * FINF has been skipped and the file will be ready to be demuxed */
140  do {
141  if (avio_read(pb, scratch, VQA_PREAMBLE_SIZE) != VQA_PREAMBLE_SIZE)
142  return AVERROR(EIO);
143  chunk_tag = AV_RB32(&scratch[0]);
144  chunk_size = AV_RB32(&scratch[4]);
145 
146  /* catch any unknown header tags, for curiosity */
147  switch (chunk_tag) {
148  case CINF_TAG:
149  case CINH_TAG:
150  case CIND_TAG:
151  case LINF_TAG:
152  case PINF_TAG:
153  case PINH_TAG:
154  case PIND_TAG:
155  case FINF_TAG:
156  case CMDS_TAG:
157  case VIEW_TAG:
158  case ZBUF_TAG:
159  break;
160 
161  default:
162  av_log(s, AV_LOG_ERROR, " note: unknown chunk seen (%s)\n",
163  av_fourcc2str(chunk_tag));
164  break;
165  }
166 
167  avio_skip(pb, chunk_size);
168  } while (chunk_tag != FINF_TAG);
169 
170  return 0;
171 }
172 
174  AVPacket *pkt)
175 {
176  WsVqaDemuxContext *wsvqa = s->priv_data;
177  AVIOContext *pb = s->pb;
178  int ret = -1;
179  uint8_t preamble[VQA_PREAMBLE_SIZE];
180  uint32_t chunk_type;
181  int chunk_size;
182  unsigned skip_byte;
183 
184  while (avio_read(pb, preamble, VQA_PREAMBLE_SIZE) == VQA_PREAMBLE_SIZE) {
185  chunk_type = AV_RB32(&preamble[0]);
186  chunk_size = AV_RB32(&preamble[4]);
187 
188  if (chunk_size < 0)
189  return AVERROR_INVALIDDATA;
190  skip_byte = chunk_size & 0x01;
191 
192  if (chunk_type == VQFL_TAG) {
193  /* Each VQFL chunk carries only a codebook update inside which must be applied
194  * before the next VQFR is rendered. That's why we stash the VQFL offset here
195  * so it can be combined with the next VQFR packet. This way each packet
196  * includes a whole frame as expected. */
197  wsvqa->vqfl_chunk_pos = avio_tell(pb);
198  if (chunk_size > 3 * (1 << 20))
199  return AVERROR_INVALIDDATA;
200  wsvqa->vqfl_chunk_size = chunk_size;
201  /* We need a big seekback buffer because there can be SNxx, VIEW and ZBUF
202  * chunks (<512 KiB total) in the stream before we read VQFR (<256 KiB) and
203  * seek back here. */
204  ffio_ensure_seekback(pb, wsvqa->vqfl_chunk_size + (512 + 256) * 1024);
205  avio_skip(pb, chunk_size + skip_byte);
206  continue;
207  } else if ((chunk_type == SND0_TAG) || (chunk_type == SND1_TAG) ||
208  (chunk_type == SND2_TAG) || (chunk_type == VQFR_TAG)) {
209 
210  ret= av_get_packet(pb, pkt, chunk_size);
211  if (ret<0)
212  return AVERROR(EIO);
213 
214  switch (chunk_type) {
215  case SND0_TAG:
216  case SND1_TAG:
217  case SND2_TAG:
218  if (wsvqa->audio_stream_index == -1) {
220  if (!st)
221  return AVERROR(ENOMEM);
222 
223  wsvqa->audio_stream_index = st->index;
224  if (!wsvqa->sample_rate)
225  wsvqa->sample_rate = 22050;
226  if (!wsvqa->channels)
227  wsvqa->channels = 1;
228  if (!wsvqa->bps)
229  wsvqa->bps = 8;
230  st->codecpar->sample_rate = wsvqa->sample_rate;
231  st->codecpar->bits_per_coded_sample = wsvqa->bps;
234 
235  avpriv_set_pts_info(st, 64, 1, st->codecpar->sample_rate);
236 
237  switch (chunk_type) {
238  case SND0_TAG:
239  if (wsvqa->bps == 16)
241  else
243  break;
244  case SND1_TAG:
246  break;
247  case SND2_TAG:
249  if ((ret = ff_alloc_extradata(st->codecpar, 2)) < 0)
250  return ret;
251  AV_WL16(st->codecpar->extradata, wsvqa->version);
252  break;
253  }
254  }
255 
257  switch (chunk_type) {
258  case SND1_TAG:
259  /* unpacked size is stored in header */
260  if(pkt->data)
261  pkt->duration = AV_RL16(pkt->data) / wsvqa->channels;
262  break;
263  case SND2_TAG:
264  /* 2 samples/byte, 1 or 2 samples per frame depending on stereo */
265  pkt->duration = (chunk_size * 2) / wsvqa->channels;
266  break;
267  }
268  break;
269  case VQFR_TAG:
270  /* if a new codebook is available inside an earlier a VQFL chunk then
271  * append it to 'pkt' */
272  if (wsvqa->vqfl_chunk_size > 0) {
273  int64_t current_pos = pkt->pos;
274 
275  if (avio_seek(pb, wsvqa->vqfl_chunk_pos, SEEK_SET) < 0)
276  return AVERROR(EIO);
277 
278  /* the decoder expects chunks to be 16-bit aligned */
279  if (wsvqa->vqfl_chunk_size % 2 == 1)
280  wsvqa->vqfl_chunk_size++;
281 
282  if (av_append_packet(pb, pkt, wsvqa->vqfl_chunk_size) < 0)
283  return AVERROR(EIO);
284 
285  if (avio_seek(pb, current_pos, SEEK_SET) < 0)
286  return AVERROR(EIO);
287 
288  wsvqa->vqfl_chunk_pos = 0;
289  wsvqa->vqfl_chunk_size = 0;
290  }
291 
293  pkt->duration = 1;
294  break;
295  }
296 
297  /* stay on 16-bit alignment */
298  if (skip_byte)
299  avio_skip(pb, 1);
300 
301  return ret;
302  } else {
303  switch(chunk_type){
304  case CMDS_TAG:
305  case SN2J_TAG:
306  case VIEW_TAG:
307  case ZBUF_TAG:
308  break;
309  default:
310  av_log(s, AV_LOG_INFO, "Skipping unknown chunk %s\n",
311  av_fourcc2str(av_bswap32(chunk_type)));
312  }
313  avio_skip(pb, chunk_size + skip_byte);
314  }
315  }
316 
317  return ret;
318 }
319 
321  .name = "wsvqa",
322  .long_name = NULL_IF_CONFIG_SMALL("Westwood Studios VQA"),
323  .priv_data_size = sizeof(WsVqaDemuxContext),
327 };
AV_CODEC_ID_PCM_S16LE
@ AV_CODEC_ID_PCM_S16LE
Definition: codec_id.h:331
AVCodecParameters::extradata
uint8_t * extradata
Extra binary data needed for initializing the decoder, codec-dependent.
Definition: codec_par.h:69
AVERROR
Filter the word “frame” indicates either a video frame or a group of audio as stored in an AVFrame structure Format for each input and each output the list of supported formats For video that means pixel format For audio that means channel sample they are references to shared objects When the negotiation mechanism computes the intersection of the formats supported at each end of a all references to both lists are replaced with a reference to the intersection And when a single format is eventually chosen for a link amongst the remaining all references to the list are updated That means that if a filter requires that its input and output have the same format amongst a supported all it has to do is use a reference to the same list of formats query_formats can leave some formats unset and return AVERROR(EAGAIN) to cause the negotiation mechanism toagain later. That can be used by filters with complex requirements to use the format negotiated on one link to set the formats supported on another. Frame references ownership and permissions
AVCodecParameters::codec_type
enum AVMediaType codec_type
General type of the encoded data.
Definition: codec_par.h:51
PIND_TAG
#define PIND_TAG
Definition: westwood_vqa.c:55
avformat_new_stream
AVStream * avformat_new_stream(AVFormatContext *s, const struct AVCodec *c)
Add a new stream to a media file.
wsvqa_read_packet
static int wsvqa_read_packet(AVFormatContext *s, AVPacket *pkt)
Definition: westwood_vqa.c:173
AVPacket::data
uint8_t * data
Definition: packet.h:522
ZBUF_TAG
#define ZBUF_TAG
Definition: westwood_vqa.c:59
WsVqaDemuxContext::version
int version
Definition: westwood_vqa.c:65
AVPacket::duration
int64_t duration
Duration of this packet in AVStream->time_base units, 0 if unknown.
Definition: packet.h:540
AVCodecParameters::codec_tag
uint32_t codec_tag
Additional information about the codec (corresponds to the AVI FOURCC).
Definition: codec_par.h:59
AVProbeData::buf_size
int buf_size
Size of buf except extra allocated bytes.
Definition: avformat.h:455
WsVqaDemuxContext::vqfl_chunk_pos
int64_t vqfl_chunk_pos
Definition: westwood_vqa.c:71
WsVqaDemuxContext::channels
int channels
Definition: westwood_vqa.c:67
ff_get_extradata
int ff_get_extradata(void *logctx, AVCodecParameters *par, AVIOContext *pb, int size)
Allocate extradata with additional AV_INPUT_BUFFER_PADDING_SIZE at end which is always set to 0 and f...
Definition: demux_utils.c:355
AVPROBE_SCORE_MAX
#define AVPROBE_SCORE_MAX
maximum score
Definition: avformat.h:464
WsVqaDemuxContext::vqfl_chunk_size
int vqfl_chunk_size
Definition: westwood_vqa.c:72
WsVqaDemuxContext::bps
int bps
Definition: westwood_vqa.c:66
avpriv_set_pts_info
void avpriv_set_pts_info(AVStream *st, int pts_wrap_bits, unsigned int pts_num, unsigned int pts_den)
Set the time base and wrapping info for a given stream.
Definition: avformat.c:843
SN2J_TAG
#define SN2J_TAG
Definition: westwood_vqa.c:57
CMDS_TAG
#define CMDS_TAG
Definition: westwood_vqa.c:56
WsVqaDemuxContext::video_stream_index
int video_stream_index
Definition: westwood_vqa.c:70
avio_tell
static av_always_inline int64_t avio_tell(AVIOContext *s)
ftell() equivalent for AVIOContext.
Definition: avio.h:513
VQFR_TAG
#define VQFR_TAG
Definition: westwood_vqa.c:45
CINH_TAG
#define CINH_TAG
Definition: westwood_vqa.c:50
AVStream::duration
int64_t duration
Decoding: duration of the stream, in stream time base.
Definition: avformat.h:900
FORM_TAG
#define FORM_TAG
Definition: westwood_vqa.c:38
VQFL_TAG
#define VQFL_TAG
Definition: westwood_vqa.c:46
pkt
AVPacket * pkt
Definition: movenc.c:59
AV_LOG_ERROR
#define AV_LOG_ERROR
Something went wrong and cannot losslessly be recovered.
Definition: log.h:180
AVInputFormat
Definition: avformat.h:549
LINF_TAG
#define LINF_TAG
Definition: westwood_vqa.c:52
read_packet
static int read_packet(void *opaque, uint8_t *buf, int buf_size)
Definition: avio_read_callback.c:41
intreadwrite.h
s
#define s(width, name)
Definition: cbs_vp9.c:198
AVInputFormat::name
const char * name
A comma separated list of short names for the format.
Definition: avformat.h:554
AVProbeData::buf
unsigned char * buf
Buffer must have AVPROBE_PADDING_SIZE of extra allocated bytes filled with zero.
Definition: avformat.h:454
AVMEDIA_TYPE_AUDIO
@ AVMEDIA_TYPE_AUDIO
Definition: avutil.h:202
AVCodecParameters::width
int width
Video only.
Definition: codec_par.h:121
AV_CODEC_ID_WS_VQA
@ AV_CODEC_ID_WS_VQA
Definition: codec_id.h:96
AV_RL16
uint64_t_TMPL AV_WL64 unsigned int_TMPL AV_WL32 unsigned int_TMPL AV_WL24 unsigned int_TMPL AV_RL16
Definition: bytestream.h:94
PINH_TAG
#define PINH_TAG
Definition: westwood_vqa.c:54
AVFormatContext
Format I/O context.
Definition: avformat.h:1363
internal.h
AVStream::codecpar
AVCodecParameters * codecpar
Codec parameters associated with this stream.
Definition: avformat.h:864
read_header
static int read_header(FFV1Context *f)
Definition: ffv1dec.c:550
NULL
#define NULL
Definition: coverity.c:32
SND0_TAG
#define SND0_TAG
Definition: westwood_vqa.c:42
AVFMTCTX_NOHEADER
#define AVFMTCTX_NOHEADER
signal that no header is present (streams are added dynamically)
Definition: avformat.h:1314
AV_CODEC_ID_ADPCM_IMA_WS
@ AV_CODEC_ID_ADPCM_IMA_WS
Definition: codec_id.h:374
AVProbeData
This structure contains the data a format has to probe a file.
Definition: avformat.h:452
ff_wsvqa_demuxer
const AVInputFormat ff_wsvqa_demuxer
Definition: westwood_vqa.c:320
AVCodecParameters::ch_layout
AVChannelLayout ch_layout
Audio only.
Definition: codec_par.h:206
AVCodecParameters::sample_rate
int sample_rate
Audio only.
Definition: codec_par.h:171
AVStream::nb_frames
int64_t nb_frames
number of frames in this stream if known or 0
Definition: avformat.h:902
WVQA_TAG
#define WVQA_TAG
Definition: westwood_vqa.c:39
AVIOContext
Bytestream IO Context.
Definition: avio.h:166
NULL_IF_CONFIG_SMALL
#define NULL_IF_CONFIG_SMALL(x)
Return NULL if CONFIG_SMALL is true, otherwise the argument without modification.
Definition: internal.h:106
VQA_PREAMBLE_SIZE
#define VQA_PREAMBLE_SIZE
Definition: westwood_vqa.c:62
av_bswap32
#define av_bswap32
Definition: bswap.h:28
AV_RB32
uint64_t_TMPL AV_WL64 unsigned int_TMPL AV_WL32 unsigned int_TMPL AV_WL24 unsigned int_TMPL AV_WL16 uint64_t_TMPL AV_WB64 unsigned int_TMPL AV_RB32
Definition: bytestream.h:96
SND2_TAG
#define SND2_TAG
Definition: westwood_vqa.c:44
header
static const uint8_t header[24]
Definition: sdr2.c:67
AV_WL16
#define AV_WL16(p, v)
Definition: intreadwrite.h:410
ffio_ensure_seekback
int ffio_ensure_seekback(AVIOContext *s, int64_t buf_size)
Ensures that the requested seekback buffer size will be available.
Definition: aviobuf.c:1135
FINF_TAG
#define FINF_TAG
Definition: westwood_vqa.c:41
AV_LOG_INFO
#define AV_LOG_INFO
Standard information.
Definition: log.h:191
av_channel_layout_default
void av_channel_layout_default(AVChannelLayout *ch_layout, int nb_channels)
Get the default channel layout for a given number of channels.
Definition: channel_layout.c:1033
CINF_TAG
#define CINF_TAG
Definition: westwood_vqa.c:49
VQA_HEADER_SIZE
#define VQA_HEADER_SIZE
Definition: westwood_vqa.c:61
avio_internal.h
AVCodecParameters::height
int height
Definition: codec_par.h:122
PINF_TAG
#define PINF_TAG
Definition: westwood_vqa.c:53
wsvqa_read_header
static int wsvqa_read_header(AVFormatContext *s)
Definition: westwood_vqa.c:89
demux.h
av_get_packet
int av_get_packet(AVIOContext *s, AVPacket *pkt, int size)
Allocate and read the payload of a packet and initialize its fields with default values.
Definition: utils.c:103
ret
ret
Definition: filter_design.txt:187
AVStream
Stream structure.
Definition: avformat.h:841
avio_seek
int64_t avio_seek(AVIOContext *s, int64_t offset, int whence)
fseek() equivalent for AVIOContext.
Definition: aviobuf.c:278
av_append_packet
int av_append_packet(AVIOContext *s, AVPacket *pkt, int size)
Read data and append it to the current content of the AVPacket.
Definition: utils.c:119
avformat.h
AV_CODEC_ID_WESTWOOD_SND1
@ AV_CODEC_ID_WESTWOOD_SND1
Definition: codec_id.h:460
VIEW_TAG
#define VIEW_TAG
Definition: westwood_vqa.c:58
AVStream::index
int index
stream index in AVFormatContext
Definition: avformat.h:847
wsvqa_probe
static int wsvqa_probe(const AVProbeData *p)
Definition: westwood_vqa.c:75
WsVqaDemuxContext
Definition: westwood_vqa.c:64
avio_read
int avio_read(AVIOContext *s, unsigned char *buf, int size)
Read size bytes from AVIOContext into buf.
Definition: aviobuf.c:659
AVPacket::stream_index
int stream_index
Definition: packet.h:524
avio_skip
int64_t avio_skip(AVIOContext *s, int64_t offset)
Skip given number of bytes forward.
Definition: aviobuf.c:365
AVMEDIA_TYPE_VIDEO
@ AVMEDIA_TYPE_VIDEO
Definition: avutil.h:201
SND1_TAG
#define SND1_TAG
Definition: westwood_vqa.c:43
read_probe
static int read_probe(const AVProbeData *p)
Definition: cdg.c:29
AVCodecParameters::bits_per_coded_sample
int bits_per_coded_sample
The number of bits per sample in the codedwords.
Definition: codec_par.h:97
CIND_TAG
#define CIND_TAG
Definition: westwood_vqa.c:51
AV_CODEC_ID_PCM_U8
@ AV_CODEC_ID_PCM_U8
Definition: codec_id.h:336
WsVqaDemuxContext::sample_rate
int sample_rate
Definition: westwood_vqa.c:68
AVCodecParameters::codec_id
enum AVCodecID codec_id
Specific type of the encoded data (the codec used).
Definition: codec_par.h:55
AVPacket
This structure stores compressed data.
Definition: packet.h:499
AVPacket::pos
int64_t pos
byte position in stream, -1 if unknown
Definition: packet.h:542
av_log
#define av_log(a,...)
Definition: tableprint_vlc.h:27
AVERROR_INVALIDDATA
#define AVERROR_INVALIDDATA
Invalid data found when processing input.
Definition: error.h:61
AVStream::start_time
int64_t start_time
Decoding: pts of the first frame of the stream in presentation order, in stream time base.
Definition: avformat.h:890
WsVqaDemuxContext::audio_stream_index
int audio_stream_index
Definition: westwood_vqa.c:69
av_fourcc2str
#define av_fourcc2str(fourcc)
Definition: avutil.h:358
ff_alloc_extradata
int ff_alloc_extradata(AVCodecParameters *par, int size)
Allocate extradata with additional AV_INPUT_BUFFER_PADDING_SIZE at end which is always set to 0.
Definition: utils.c:239