1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-02 03:06:28 +02:00
FFmpeg/libavformat/westwood_vqa.c
Michael Niedermayer e37f161e66 Merge remote-tracking branch 'qatar/master'
* qatar/master: (71 commits)
  movenc: Allow writing to a non-seekable output if using empty moov
  movenc: Support adding isml (smooth streaming live) metadata
  libavcodec: Don't crash in avcodec_encode_audio if time_base isn't set
  sunrast: Document the different Sun Raster file format types.
  sunrast: Add a check for experimental type.
  libspeexenc: use AVSampleFormat instead of deprecated/removed SampleFormat
  lavf: remove disabled FF_API_SET_PTS_INFO cruft
  lavf: remove disabled FF_API_OLD_INTERRUPT_CB cruft
  lavf: remove disabled FF_API_REORDER_PRIVATE cruft
  lavf: remove disabled FF_API_SEEK_PUBLIC cruft
  lavf: remove disabled FF_API_STREAM_COPY cruft
  lavf: remove disabled FF_API_PRELOAD cruft
  lavf: remove disabled FF_API_NEW_STREAM cruft
  lavf: remove disabled FF_API_RTSP_URL_OPTIONS cruft
  lavf: remove disabled FF_API_MUXRATE cruft
  lavf: remove disabled FF_API_FILESIZE cruft
  lavf: remove disabled FF_API_TIMESTAMP cruft
  lavf: remove disabled FF_API_LOOP_OUTPUT cruft
  lavf: remove disabled FF_API_LOOP_INPUT cruft
  lavf: remove disabled FF_API_AVSTREAM_QUALITY cruft
  ...

Conflicts:
	doc/APIchanges
	libavcodec/8bps.c
	libavcodec/avcodec.h
	libavcodec/libx264.c
	libavcodec/mjpegbdec.c
	libavcodec/options.c
	libavcodec/sunrast.c
	libavcodec/utils.c
	libavcodec/version.h
	libavcodec/x86/h264_deblock.asm
	libavdevice/libdc1394.c
	libavdevice/v4l2.c
	libavformat/avformat.h
	libavformat/avio.c
	libavformat/avio.h
	libavformat/aviobuf.c
	libavformat/dv.c
	libavformat/mov.c
	libavformat/utils.c
	libavformat/version.h
	libavformat/wtv.c
	libavutil/Makefile
	libavutil/file.c
	libswscale/x86/input.asm
	libswscale/x86/swscale_mmx.c
	libswscale/x86/swscale_template.c
	tests/ref/lavf/ffm

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-01-28 07:53:34 +01:00

267 lines
8.7 KiB
C

/*
* Westwood Studios VQA Format Demuxer
* Copyright (c) 2003 The ffmpeg Project
*
* This file is part of FFmpeg.
*
* FFmpeg is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* FFmpeg is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
/**
* @file
* Westwood Studios VQA file demuxer
* by Mike Melanson (melanson@pcisys.net)
* for more information on the Westwood file formats, visit:
* http://www.pcisys.net/~melanson/codecs/
* http://www.geocities.com/SiliconValley/8682/aud3.txt
*/
#include "libavutil/intreadwrite.h"
#include "avformat.h"
#include "internal.h"
#define FORM_TAG MKBETAG('F', 'O', 'R', 'M')
#define WVQA_TAG MKBETAG('W', 'V', 'Q', 'A')
#define VQHD_TAG MKBETAG('V', 'Q', 'H', 'D')
#define FINF_TAG MKBETAG('F', 'I', 'N', 'F')
#define SND0_TAG MKBETAG('S', 'N', 'D', '0')
#define SND1_TAG MKBETAG('S', 'N', 'D', '1')
#define SND2_TAG MKBETAG('S', 'N', 'D', '2')
#define VQFR_TAG MKBETAG('V', 'Q', 'F', 'R')
/* don't know what these tags are for, but acknowledge their existence */
#define CINF_TAG MKBETAG('C', 'I', 'N', 'F')
#define CINH_TAG MKBETAG('C', 'I', 'N', 'H')
#define CIND_TAG MKBETAG('C', 'I', 'N', 'D')
#define PINF_TAG MKBETAG('P', 'I', 'N', 'F')
#define PINH_TAG MKBETAG('P', 'I', 'N', 'H')
#define PIND_TAG MKBETAG('P', 'I', 'N', 'D')
#define CMDS_TAG MKBETAG('C', 'M', 'D', 'S')
#define VQA_HEADER_SIZE 0x2A
#define VQA_PREAMBLE_SIZE 8
typedef struct WsVqaDemuxContext {
int audio_channels;
int audio_stream_index;
int video_stream_index;
} WsVqaDemuxContext;
static int wsvqa_probe(AVProbeData *p)
{
/* need 12 bytes to qualify */
if (p->buf_size < 12)
return 0;
/* check for the VQA signatures */
if ((AV_RB32(&p->buf[0]) != FORM_TAG) ||
(AV_RB32(&p->buf[8]) != WVQA_TAG))
return 0;
return AVPROBE_SCORE_MAX;
}
static int wsvqa_read_header(AVFormatContext *s)
{
WsVqaDemuxContext *wsvqa = s->priv_data;
AVIOContext *pb = s->pb;
AVStream *st;
unsigned char *header;
unsigned char scratch[VQA_PREAMBLE_SIZE];
unsigned int chunk_tag;
unsigned int chunk_size;
int fps, version, flags, sample_rate, channels;
/* initialize the video decoder stream */
st = avformat_new_stream(s, NULL);
if (!st)
return AVERROR(ENOMEM);
st->start_time = 0;
wsvqa->video_stream_index = st->index;
st->codec->codec_type = AVMEDIA_TYPE_VIDEO;
st->codec->codec_id = CODEC_ID_WS_VQA;
st->codec->codec_tag = 0; /* no fourcc */
/* skip to the start of the VQA header */
avio_seek(pb, 20, SEEK_SET);
/* the VQA header needs to go to the decoder */
st->codec->extradata_size = VQA_HEADER_SIZE;
st->codec->extradata = av_mallocz(VQA_HEADER_SIZE + FF_INPUT_BUFFER_PADDING_SIZE);
header = (unsigned char *)st->codec->extradata;
if (avio_read(pb, st->codec->extradata, VQA_HEADER_SIZE) !=
VQA_HEADER_SIZE) {
av_free(st->codec->extradata);
return AVERROR(EIO);
}
st->codec->width = AV_RL16(&header[6]);
st->codec->height = AV_RL16(&header[8]);
fps = header[12];
if (fps < 1 || fps > 30) {
av_log(s, AV_LOG_ERROR, "invalid fps: %d\n", fps);
return AVERROR_INVALIDDATA;
}
avpriv_set_pts_info(st, 64, 1, fps);
/* initialize the audio decoder stream for VQA v1 or nonzero samplerate */
version = AV_RL16(&header[ 0]);
flags = AV_RL16(&header[ 2]);
sample_rate = AV_RL16(&header[24]);
channels = header[26];
if (sample_rate || (version == 1 && flags == 1)) {
st = avformat_new_stream(s, NULL);
if (!st)
return AVERROR(ENOMEM);
st->start_time = 0;
st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
st->codec->extradata_size = VQA_HEADER_SIZE;
st->codec->extradata = av_mallocz(VQA_HEADER_SIZE + FF_INPUT_BUFFER_PADDING_SIZE);
if (!st->codec->extradata)
return AVERROR(ENOMEM);
memcpy(st->codec->extradata, header, VQA_HEADER_SIZE);
if (!sample_rate)
sample_rate = 22050;
st->codec->sample_rate = sample_rate;
avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate);
if (!channels)
channels = 1;
st->codec->channels = channels;
switch (version) {
case 1:
st->codec->codec_id = CODEC_ID_WESTWOOD_SND1;
break;
case 2:
case 3:
st->codec->codec_id = CODEC_ID_ADPCM_IMA_WS;
st->codec->bits_per_coded_sample = 4;
st->codec->bit_rate = channels * sample_rate * 4;
break;
default:
/* NOTE: version 0 is supposedly raw pcm_u8 or pcm_s16le, but we do
not have any samples to validate this */
av_log_ask_for_sample(s, "VQA version %d audio\n", version);
return AVERROR_PATCHWELCOME;
}
wsvqa->audio_stream_index = st->index;
wsvqa->audio_channels = st->codec->channels;
}
/* there are 0 or more chunks before the FINF chunk; iterate until
* FINF has been skipped and the file will be ready to be demuxed */
do {
if (avio_read(pb, scratch, VQA_PREAMBLE_SIZE) != VQA_PREAMBLE_SIZE)
return AVERROR(EIO);
chunk_tag = AV_RB32(&scratch[0]);
chunk_size = AV_RB32(&scratch[4]);
/* catch any unknown header tags, for curiousity */
switch (chunk_tag) {
case CINF_TAG:
case CINH_TAG:
case CIND_TAG:
case PINF_TAG:
case PINH_TAG:
case PIND_TAG:
case FINF_TAG:
case CMDS_TAG:
break;
default:
av_log (s, AV_LOG_ERROR, " note: unknown chunk seen (%c%c%c%c)\n",
scratch[0], scratch[1],
scratch[2], scratch[3]);
break;
}
avio_skip(pb, chunk_size);
} while (chunk_tag != FINF_TAG);
return 0;
}
static int wsvqa_read_packet(AVFormatContext *s,
AVPacket *pkt)
{
WsVqaDemuxContext *wsvqa = s->priv_data;
AVIOContext *pb = s->pb;
int ret = -1;
unsigned char preamble[VQA_PREAMBLE_SIZE];
unsigned int chunk_type;
unsigned int chunk_size;
int skip_byte;
while (avio_read(pb, preamble, VQA_PREAMBLE_SIZE) == VQA_PREAMBLE_SIZE) {
chunk_type = AV_RB32(&preamble[0]);
chunk_size = AV_RB32(&preamble[4]);
skip_byte = chunk_size & 0x01;
if ((chunk_type == SND2_TAG || chunk_type == SND1_TAG) && wsvqa->audio_channels == 0) {
av_log(s, AV_LOG_ERROR, "audio chunk without any audio header information found\n");
return AVERROR_INVALIDDATA;
}
if ((chunk_type == SND1_TAG) || (chunk_type == SND2_TAG) || (chunk_type == VQFR_TAG)) {
ret= av_get_packet(pb, pkt, chunk_size);
if (ret<0)
return AVERROR(EIO);
if (chunk_type == SND2_TAG) {
pkt->stream_index = wsvqa->audio_stream_index;
/* 2 samples/byte, 1 or 2 samples per frame depending on stereo */
pkt->duration = (chunk_size * 2) / wsvqa->audio_channels;
} else if(chunk_type == SND1_TAG) {
pkt->stream_index = wsvqa->audio_stream_index;
/* unpacked size is stored in header */
pkt->duration = AV_RL16(pkt->data) / wsvqa->audio_channels;
} else {
pkt->stream_index = wsvqa->video_stream_index;
pkt->duration = 1;
}
/* stay on 16-bit alignment */
if (skip_byte)
avio_skip(pb, 1);
return ret;
} else {
switch(chunk_type){
case CMDS_TAG:
case SND0_TAG:
break;
default:
av_log(s, AV_LOG_INFO, "Skipping unknown chunk 0x%08X\n", chunk_type);
}
avio_skip(pb, chunk_size + skip_byte);
}
}
return ret;
}
AVInputFormat ff_wsvqa_demuxer = {
.name = "wsvqa",
.long_name = NULL_IF_CONFIG_SMALL("Westwood Studios VQA format"),
.priv_data_size = sizeof(WsVqaDemuxContext),
.read_probe = wsvqa_probe,
.read_header = wsvqa_read_header,
.read_packet = wsvqa_read_packet,
};