mpv/demux/demux_raw.c

330 lines
9.8 KiB
C
Raw Normal View History

/*
* This file is part of mpv.
*
* mpv 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.
*
* mpv 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 mpv. If not, see <http://www.gnu.org/licenses/>.
*/
#include "config.h"
#include <stdlib.h>
#include <stdio.h>
#include <unistd.h>
#include <string.h>
#include <libavcodec/avcodec.h>
#include <libavutil/common.h>
#include "common/av_common.h"
#include "options/m_config.h"
#include "options/m_option.h"
#include "stream/stream.h"
#include "demux.h"
#include "stheader.h"
#include "codec_tags.h"
#include "video/fmt-conversion.h"
#include "video/img_format.h"
#include "osdep/endian.h"
struct demux_rawaudio_opts {
struct m_channels channels;
int samplerate;
int aformat;
};
// Ad-hoc schema to systematically encode the format as int
#define PCM(sign, is_float, bits, is_be) \
((sign) | ((is_float) << 1) | ((is_be) << 2) | ((bits) << 3))
#define NE (BYTE_ORDER == BIG_ENDIAN)
#define OPT_BASE_STRUCT struct demux_rawaudio_opts
const struct m_sub_options demux_rawaudio_conf = {
.opts = (const m_option_t[]) {
OPT_CHANNELS("channels", channels, 0, .min = 1),
OPT_INTRANGE("rate", samplerate, 0, 1000, 8 * 48000),
OPT_CHOICE("format", aformat, 0,
({"u8", PCM(0, 0, 8, 0)},
{"s8", PCM(1, 0, 8, 0)},
{"u16le", PCM(0, 0, 16, 0)}, {"u16be", PCM(0, 0, 16, 1)},
{"s16le", PCM(1, 0, 16, 0)}, {"s16be", PCM(1, 0, 16, 1)},
{"u24le", PCM(0, 0, 24, 0)}, {"u24be", PCM(0, 0, 24, 1)},
{"s24le", PCM(1, 0, 24, 0)}, {"s24be", PCM(1, 0, 24, 1)},
{"u32le", PCM(0, 0, 32, 0)}, {"u32be", PCM(0, 0, 32, 1)},
{"s32le", PCM(1, 0, 32, 0)}, {"s32be", PCM(1, 0, 32, 1)},
{"floatle", PCM(0, 1, 32, 0)}, {"floatbe", PCM(0, 1, 32, 1)},
{"doublele",PCM(0, 1, 64, 0)}, {"doublebe", PCM(0, 1, 64, 1)},
{"u16", PCM(0, 0, 16, NE)},
{"s16", PCM(1, 0, 16, NE)},
{"u24", PCM(0, 0, 24, NE)},
{"s24", PCM(1, 0, 24, NE)},
{"u32", PCM(0, 0, 32, NE)},
{"s32", PCM(1, 0, 32, NE)},
{"float", PCM(0, 1, 32, NE)},
{"double", PCM(0, 1, 64, NE)})),
{0}
},
.size = sizeof(struct demux_rawaudio_opts),
.defaults = &(const struct demux_rawaudio_opts){
// Note that currently, stream_cdda expects exactly these parameters!
.channels = {
.set = 1,
.chmaps = (struct mp_chmap[]){ MP_CHMAP_INIT_STEREO, },
.num_chmaps = 1,
},
.samplerate = 44100,
.aformat = PCM(1, 0, 16, 0), // s16le
},
};
#undef PCM
#undef NE
struct demux_rawvideo_opts {
int vformat;
int mp_format;
char *codec;
int width;
int height;
float fps;
int imgsize;
};
#undef OPT_BASE_STRUCT
#define OPT_BASE_STRUCT struct demux_rawvideo_opts
const struct m_sub_options demux_rawvideo_conf = {
.opts = (const m_option_t[]) {
OPT_INTRANGE("w", width, 0, 1, 8192),
OPT_INTRANGE("h", height, 0, 1, 8192),
OPT_GENERAL(int, "format", vformat, 0, .type = &m_option_type_fourcc),
OPT_IMAGEFORMAT("mp-format", mp_format, 0),
OPT_STRING("codec", codec, 0),
OPT_FLOATRANGE("fps", fps, 0, 0.001, 1000),
OPT_INTRANGE("size", imgsize, 0, 1, 8192 * 8192 * 4),
{0}
},
.size = sizeof(struct demux_rawvideo_opts),
.defaults = &(const struct demux_rawvideo_opts){
.vformat = MKTAG('I', '4', '2', '0'),
.width = 1280,
.height = 720,
.fps = 25,
},
};
struct priv {
struct sh_stream *sh;
int frame_size;
int read_frames;
double frame_rate;
};
static int generic_open(struct demuxer *demuxer)
{
struct stream *s = demuxer->stream;
struct priv *p = demuxer->priv;
int64_t end = stream_get_size(s);
if (end >= 0)
demuxer->duration = (end / p->frame_size) / p->frame_rate;
return 0;
}
static int demux_rawaudio_open(demuxer_t *demuxer, enum demux_check check)
{
struct demux_rawaudio_opts *opts =
mp_get_config_group(demuxer, demuxer->global, &demux_rawaudio_conf);
if (check != DEMUX_CHECK_REQUEST && check != DEMUX_CHECK_FORCE)
return -1;
if (opts->channels.num_chmaps != 1) {
MP_ERR(demuxer, "Invalid channels option given.\n");
return -1;
}
struct sh_stream *sh = demux_alloc_sh_stream(STREAM_AUDIO);
struct mp_codec_params *c = sh->codec;
c->channels = opts->channels.chmaps[0];
c->force_channels = true;
c->samplerate = opts->samplerate;
c->native_tb_num = 1;
c->native_tb_den = c->samplerate;
int f = opts->aformat;
// See PCM(): sign float bits endian
mp_set_pcm_codec(sh->codec, f & 1, f & 2, f >> 3, f & 4);
int samplesize = ((f >> 3) + 7) / 8;
demux_add_sh_stream(demuxer, sh);
struct priv *p = talloc_ptrtype(demuxer, p);
demuxer->priv = p;
*p = (struct priv) {
.sh = sh,
.frame_size = samplesize * c->channels.num,
.frame_rate = c->samplerate,
.read_frames = c->samplerate / 8,
};
return generic_open(demuxer);
}
static int demux_rawvideo_open(demuxer_t *demuxer, enum demux_check check)
{
struct demux_rawvideo_opts *opts =
mp_get_config_group(demuxer, demuxer->global, &demux_rawvideo_conf);
if (check != DEMUX_CHECK_REQUEST && check != DEMUX_CHECK_FORCE)
return -1;
int width = opts->width;
int height = opts->height;
if (!width || !height) {
MP_ERR(demuxer, "rawvideo: width or height not specified!\n");
return -1;
video: decouple internal pixel formats from FourCCs mplayer's video chain traditionally used FourCCs for pixel formats. For example, it used IMGFMT_YV12 for 4:2:0 YUV, which was defined to the string 'YV12' interpreted as unsigned int. Additionally, it used to encode information into the numeric values of some formats. The RGB formats had their bit depth and endian encoded into the least significant byte. Extended planar formats (420P10 etc.) had chroma shift, endian, and component bit depth encoded. (This has been removed in recent commits.) Replace the FourCC mess with a simple enum. Remove all the redundant formats like YV12/I420/IYUV. Replace some image format names by something more intuitive, most importantly IMGFMT_YV12 -> IMGFMT_420P. Add img_fourcc.h, which contains the old IDs for code that actually uses FourCCs. Change the way demuxers, that output raw video, identify the video format: they set either MP_FOURCC_RAWVIDEO or MP_FOURCC_IMGFMT to request the rawvideo decoder, and sh_video->imgfmt specifies the pixel format. Like the previous hack, this is supposed to avoid the need for a complete codecs.cfg entry per format, or other lookup tables. (Note that the RGB raw video FourCCs mostly rely on ffmpeg's mappings for NUT raw video, but this is still considered better than adding a raw video decoder - even if trivial, it would be full of annoying lookup tables.) The TV code has not been tested. Some corrective changes regarding endian and other image format flags creep in.
2012-12-23 19:03:30 +00:00
}
const char *decoder = "rawvideo";
int imgfmt = opts->vformat;
int imgsize = opts->imgsize;
int mp_imgfmt = 0;
if (opts->mp_format && !IMGFMT_IS_HWACCEL(opts->mp_format)) {
mp_imgfmt = opts->mp_format;
if (!imgsize) {
struct mp_imgfmt_desc desc = mp_imgfmt_get_desc(opts->mp_format);
for (int p = 0; p < desc.num_planes; p++) {
imgsize += ((width >> desc.xs[p]) * (height >> desc.ys[p]) *
desc.bpp[p] + 7) / 8;
}
}
} else if (opts->codec && opts->codec[0])
decoder = talloc_strdup(demuxer, opts->codec);
if (!imgsize) {
int bpp = 0;
switch (imgfmt) {
case MKTAG('Y', 'V', '1', '2'):
case MKTAG('I', '4', '2', '0'):
case MKTAG('I', 'Y', 'U', 'V'):
bpp = 12;
break;
case MKTAG('U', 'Y', 'V', 'Y'):
case MKTAG('Y', 'U', 'Y', '2'):
bpp = 16;
break;
}
if (!bpp) {
MP_ERR(demuxer, "rawvideo: img size not specified and unknown format!\n");
return -1;
}
imgsize = width * height * bpp / 8;
video: decouple internal pixel formats from FourCCs mplayer's video chain traditionally used FourCCs for pixel formats. For example, it used IMGFMT_YV12 for 4:2:0 YUV, which was defined to the string 'YV12' interpreted as unsigned int. Additionally, it used to encode information into the numeric values of some formats. The RGB formats had their bit depth and endian encoded into the least significant byte. Extended planar formats (420P10 etc.) had chroma shift, endian, and component bit depth encoded. (This has been removed in recent commits.) Replace the FourCC mess with a simple enum. Remove all the redundant formats like YV12/I420/IYUV. Replace some image format names by something more intuitive, most importantly IMGFMT_YV12 -> IMGFMT_420P. Add img_fourcc.h, which contains the old IDs for code that actually uses FourCCs. Change the way demuxers, that output raw video, identify the video format: they set either MP_FOURCC_RAWVIDEO or MP_FOURCC_IMGFMT to request the rawvideo decoder, and sh_video->imgfmt specifies the pixel format. Like the previous hack, this is supposed to avoid the need for a complete codecs.cfg entry per format, or other lookup tables. (Note that the RGB raw video FourCCs mostly rely on ffmpeg's mappings for NUT raw video, but this is still considered better than adding a raw video decoder - even if trivial, it would be full of annoying lookup tables.) The TV code has not been tested. Some corrective changes regarding endian and other image format flags creep in.
2012-12-23 19:03:30 +00:00
}
struct sh_stream *sh = demux_alloc_sh_stream(STREAM_VIDEO);
struct mp_codec_params *c = sh->codec;
c->codec = decoder;
c->codec_tag = imgfmt;
c->fps = opts->fps;
c->reliable_fps = true;
c->disp_w = width;
c->disp_h = height;
if (mp_imgfmt) {
c->lav_codecpar = avcodec_parameters_alloc();
if (!c->lav_codecpar)
abort();
c->lav_codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
c->lav_codecpar->codec_id = mp_codec_to_av_codec_id(decoder);
c->lav_codecpar->format = imgfmt2pixfmt(mp_imgfmt);
c->lav_codecpar->width = width;
c->lav_codecpar->height = height;
}
demux_add_sh_stream(demuxer, sh);
struct priv *p = talloc_ptrtype(demuxer, p);
demuxer->priv = p;
*p = (struct priv) {
.sh = sh,
.frame_size = imgsize,
.frame_rate = c->fps,
.read_frames = 1,
};
return generic_open(demuxer);
}
static bool raw_read_packet(struct demuxer *demuxer, struct demux_packet **pkt)
{
struct priv *p = demuxer->priv;
2013-07-08 22:03:14 +00:00
if (demuxer->stream->eof)
return false;
2013-07-08 22:03:14 +00:00
struct demux_packet *dp = new_demux_packet(p->frame_size * p->read_frames);
if (!dp) {
MP_ERR(demuxer, "Can't read packet.\n");
return true;
}
dp->keyframe = true;
dp->pos = stream_tell(demuxer->stream);
dp->pts = (dp->pos / p->frame_size) / p->frame_rate;
2013-07-08 22:03:14 +00:00
int len = stream_read(demuxer->stream, dp->buffer, dp->len);
demux_packet_shorten(dp, len);
2013-07-08 22:03:14 +00:00
dp->stream = p->sh->index;
*pkt = dp;
return true;
}
static void raw_seek(demuxer_t *demuxer, double seek_pts, int flags)
{
struct priv *p = demuxer->priv;
stream_t *s = demuxer->stream;
int64_t end = stream_get_size(s);
demux_raw: fix operation with demuxer cache and backward playback Raw audio formats can be accessed sample-wise, and logically audio packets demuxed from it would contain only 1 sample. This is inefficient, so raw audio demuxers typically "bundle" multiple samples in one packet. The problem for the demuxer cache and backward playback is that they need properly aligned packets to make seeking "deterministic". The requirement is that if you read some packets, and then seek back, you eventually see the same packets again. demux_raw basically allowed to seek into the middle of a previously returned packet, which makes it impossible to make the transition seamless. (Unless you'd be aware of the packet data format and cut them to make it seamless, which is too complex for such a use case.) Solve this by always aligning seeks to packet boundaries. This reduces the seek accuracy to the arbitrarily chosen packet size. But you can use hr-seek to fix this. The gain from not making raw audio an awful special case pays in exchange for this "stupid" suggestion to use hr-seek. It appears this also fixes that it could and did seek into the middle of the frame (not sure if this code was ever tested - it goes back to removing the code duplication between the former demux_rawaudio.c and demux_rawvideo.c). If you really cared, you could introduce a seek flag that controls whether the seek is aligned or not. Then code which requires "deterministic" demuxing could set it. But this isn't really useful for us, and we'd always set the flag anyway, unless maybe the caching were forced disabled. libavformat's wav demuxer exhibits the same issue. We can't fix it (it would require the unpleasant experience of contributing to FFmpeg), so document this in otions.rst. In theory, this also affects seek range joining, but the only bad effect should be that cached data is discarded.
2019-05-20 00:18:59 +00:00
int64_t frame_nr = seek_pts * p->frame_rate;
frame_nr = frame_nr - (frame_nr % p->read_frames);
int64_t pos = frame_nr * p->frame_size;
if (flags & SEEK_FACTOR)
pos = end * seek_pts;
if (pos < 0)
pos = 0;
if (end > 0 && pos > end)
pos = end;
stream_seek(s, (pos / p->frame_size) * p->frame_size);
}
const demuxer_desc_t demuxer_desc_rawaudio = {
.name = "rawaudio",
.desc = "Uncompressed audio",
.open = demux_rawaudio_open,
.read_packet = raw_read_packet,
.seek = raw_seek,
};
const demuxer_desc_t demuxer_desc_rawvideo = {
.name = "rawvideo",
.desc = "Uncompressed video",
.open = demux_rawvideo_open,
.read_packet = raw_read_packet,
.seek = raw_seek,
};