1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-12-12 19:18:44 +02:00
FFmpeg/libavfilter/af_asyncts.c
Michael Niedermayer a05a44e205 Merge commit '7e350379f87e7f74420b4813170fe808e2313911'
* commit '7e350379f87e7f74420b4813170fe808e2313911':
  lavfi: switch to AVFrame.

Conflicts:
	doc/filters.texi
	libavfilter/af_ashowinfo.c
	libavfilter/audio.c
	libavfilter/avfilter.c
	libavfilter/avfilter.h
	libavfilter/buffersink.c
	libavfilter/buffersrc.c
	libavfilter/buffersrc.h
	libavfilter/f_select.c
	libavfilter/f_setpts.c
	libavfilter/fifo.c
	libavfilter/split.c
	libavfilter/src_movie.c
	libavfilter/version.h
	libavfilter/vf_aspect.c
	libavfilter/vf_bbox.c
	libavfilter/vf_blackframe.c
	libavfilter/vf_delogo.c
	libavfilter/vf_drawbox.c
	libavfilter/vf_drawtext.c
	libavfilter/vf_fade.c
	libavfilter/vf_fieldorder.c
	libavfilter/vf_fps.c
	libavfilter/vf_frei0r.c
	libavfilter/vf_gradfun.c
	libavfilter/vf_hqdn3d.c
	libavfilter/vf_lut.c
	libavfilter/vf_overlay.c
	libavfilter/vf_pad.c
	libavfilter/vf_scale.c
	libavfilter/vf_showinfo.c
	libavfilter/vf_transpose.c
	libavfilter/vf_vflip.c
	libavfilter/vf_yadif.c
	libavfilter/video.c
	libavfilter/vsrc_testsrc.c
	libavfilter/yadif.h

Following are notes about the merge authorship and various technical details.

Michael Niedermayer:
  * Main merge operation, notably avfilter.c and video.c
  * Switch to AVFrame:
    - afade
    - anullsrc
    - apad
    - aresample
    - blackframe
    - deshake
    - idet
    - il
    - mandelbrot
    - mptestsrc
    - noise
    - setfield
    - smartblur
    - tinterlace
  * various merge changes and fixes in:
    - ashowinfo
    - blackdetect
    - field
    - fps
    - select
    - testsrc
    - yadif

Nicolas George:
  * Switch to AVFrame:
    - make rawdec work with refcounted frames. Adapted from commit
      759001c534 by Anton Khirnov.
      Also, fix the use of || instead of | in a flags check.
    - make buffer sink and src, audio and video work all together

Clément Bœsch:
  * Switch to AVFrame:
    - aevalsrc
    - alphaextract
    - blend
    - cellauto
    - colormatrix
    - concat
    - earwax
    - ebur128
    - edgedetect
    - geq
    - histeq
    - histogram
    - hue
    - kerndeint
    - life
    - movie
    - mp (with the help of Michael)
    - overlay
    - pad
    - pan
    - pp
    - pp
    - removelogo
    - sendcmd
    - showspectrum
    - showwaves
    - silencedetect
    - stereo3d
    - subtitles
    - super2xsai
    - swapuv
    - thumbnail
    - tile

Hendrik Leppkes:
  * Switch to AVFrame:
    - aconvert
    - amerge
    - asetnsamples
    - atempo
    - biquads

Matthieu Bouron:
  * Switch to AVFrame
    - alphamerge
    - decimate
    - volumedetect

Stefano Sabatini:
  * Switch to AVFrame:
    - astreamsync
    - flite
    - framestep

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: Nicolas George <nicolas.george@normalesup.org>
Signed-off-by: Clément Bœsch <ubitux@gmail.com>
Signed-off-by: Hendrik Leppkes <h.leppkes@gmail.com>
Signed-off-by: Matthieu Bouron <matthieu.bouron@gmail.com>
Signed-off-by: Stefano Sabatini <stefasab@gmail.com>

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2013-03-10 01:40:35 +01:00

308 lines
9.7 KiB
C

/*
* This file is part of Libav.
*
* Libav 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.
*
* Libav 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 Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "libavresample/avresample.h"
#include "libavutil/audio_fifo.h"
#include "libavutil/common.h"
#include "libavutil/mathematics.h"
#include "libavutil/opt.h"
#include "libavutil/samplefmt.h"
#include "audio.h"
#include "avfilter.h"
#include "internal.h"
typedef struct ASyncContext {
const AVClass *class;
AVAudioResampleContext *avr;
int64_t pts; ///< timestamp in samples of the first sample in fifo
int min_delta; ///< pad/trim min threshold in samples
int first_frame; ///< 1 until filter_frame() has processed at least 1 frame with a pts != AV_NOPTS_VALUE
int64_t first_pts; ///< user-specified first expected pts, in samples
/* options */
int resample;
float min_delta_sec;
int max_comp;
/* set by filter_frame() to signal an output frame to request_frame() */
int got_output;
} ASyncContext;
#define OFFSET(x) offsetof(ASyncContext, x)
#define A AV_OPT_FLAG_AUDIO_PARAM
#define F AV_OPT_FLAG_FILTERING_PARAM
static const AVOption asyncts_options[] = {
{ "compensate", "Stretch/squeeze the data to make it match the timestamps", OFFSET(resample), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, A|F },
{ "min_delta", "Minimum difference between timestamps and audio data "
"(in seconds) to trigger padding/trimmin the data.", OFFSET(min_delta_sec), AV_OPT_TYPE_FLOAT, { .dbl = 0.1 }, 0, INT_MAX, A|F },
{ "max_comp", "Maximum compensation in samples per second.", OFFSET(max_comp), AV_OPT_TYPE_INT, { .i64 = 500 }, 0, INT_MAX, A|F },
{ "first_pts", "Assume the first pts should be this value.", OFFSET(first_pts), AV_OPT_TYPE_INT64, { .i64 = AV_NOPTS_VALUE }, INT64_MIN, INT64_MAX, A|F },
{ NULL },
};
AVFILTER_DEFINE_CLASS(asyncts);
static int init(AVFilterContext *ctx, const char *args)
{
ASyncContext *s = ctx->priv;
int ret;
s->class = &asyncts_class;
av_opt_set_defaults(s);
if ((ret = av_set_options_string(s, args, "=", ":")) < 0)
return ret;
av_opt_free(s);
s->pts = AV_NOPTS_VALUE;
s->first_frame = 1;
return 0;
}
static void uninit(AVFilterContext *ctx)
{
ASyncContext *s = ctx->priv;
if (s->avr) {
avresample_close(s->avr);
avresample_free(&s->avr);
}
}
static int config_props(AVFilterLink *link)
{
ASyncContext *s = link->src->priv;
int ret;
s->min_delta = s->min_delta_sec * link->sample_rate;
link->time_base = (AVRational){1, link->sample_rate};
s->avr = avresample_alloc_context();
if (!s->avr)
return AVERROR(ENOMEM);
av_opt_set_int(s->avr, "in_channel_layout", link->channel_layout, 0);
av_opt_set_int(s->avr, "out_channel_layout", link->channel_layout, 0);
av_opt_set_int(s->avr, "in_sample_fmt", link->format, 0);
av_opt_set_int(s->avr, "out_sample_fmt", link->format, 0);
av_opt_set_int(s->avr, "in_sample_rate", link->sample_rate, 0);
av_opt_set_int(s->avr, "out_sample_rate", link->sample_rate, 0);
if (s->resample)
av_opt_set_int(s->avr, "force_resampling", 1, 0);
if ((ret = avresample_open(s->avr)) < 0)
return ret;
return 0;
}
/* get amount of data currently buffered, in samples */
static int64_t get_delay(ASyncContext *s)
{
return avresample_available(s->avr) + avresample_get_delay(s->avr);
}
static void handle_trimming(AVFilterContext *ctx)
{
ASyncContext *s = ctx->priv;
if (s->pts < s->first_pts) {
int delta = FFMIN(s->first_pts - s->pts, avresample_available(s->avr));
av_log(ctx, AV_LOG_VERBOSE, "Trimming %d samples from start\n",
delta);
avresample_read(s->avr, NULL, delta);
s->pts += delta;
} else if (s->first_frame)
s->pts = s->first_pts;
}
static int request_frame(AVFilterLink *link)
{
AVFilterContext *ctx = link->src;
ASyncContext *s = ctx->priv;
int ret = 0;
int nb_samples;
s->got_output = 0;
while (ret >= 0 && !s->got_output)
ret = ff_request_frame(ctx->inputs[0]);
/* flush the fifo */
if (ret == AVERROR_EOF) {
if (s->first_pts != AV_NOPTS_VALUE)
handle_trimming(ctx);
if (nb_samples = get_delay(s)) {
AVFrame *buf = ff_get_audio_buffer(link, nb_samples);
if (!buf)
return AVERROR(ENOMEM);
ret = avresample_convert(s->avr, buf->extended_data,
buf->linesize[0], nb_samples, NULL, 0, 0);
if (ret <= 0) {
av_frame_free(&buf);
return (ret < 0) ? ret : AVERROR_EOF;
}
buf->pts = s->pts;
return ff_filter_frame(link, buf);
}
}
return ret;
}
static int write_to_fifo(ASyncContext *s, AVFrame *buf)
{
int ret = avresample_convert(s->avr, NULL, 0, 0, buf->extended_data,
buf->linesize[0], buf->nb_samples);
av_frame_free(&buf);
return ret;
}
static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
{
AVFilterContext *ctx = inlink->dst;
ASyncContext *s = ctx->priv;
AVFilterLink *outlink = ctx->outputs[0];
int nb_channels = av_get_channel_layout_nb_channels(buf->channel_layout);
int64_t pts = (buf->pts == AV_NOPTS_VALUE) ? buf->pts :
av_rescale_q(buf->pts, inlink->time_base, outlink->time_base);
int out_size, ret;
int64_t delta;
/* buffer data until we get the next timestamp */
if (s->pts == AV_NOPTS_VALUE || pts == AV_NOPTS_VALUE) {
if (pts != AV_NOPTS_VALUE) {
s->pts = pts - get_delay(s);
}
return write_to_fifo(s, buf);
}
if (s->first_pts != AV_NOPTS_VALUE) {
handle_trimming(ctx);
if (!avresample_available(s->avr))
return write_to_fifo(s, buf);
}
/* when we have two timestamps, compute how many samples would we have
* to add/remove to get proper sync between data and timestamps */
delta = pts - s->pts - get_delay(s);
out_size = avresample_available(s->avr);
if (labs(delta) > s->min_delta ||
(s->first_frame && delta && s->first_pts != AV_NOPTS_VALUE)) {
av_log(ctx, AV_LOG_VERBOSE, "Discontinuity - %"PRId64" samples.\n", delta);
out_size = av_clipl_int32((int64_t)out_size + delta);
} else {
if (s->resample) {
int comp = av_clip(delta, -s->max_comp, s->max_comp);
av_log(ctx, AV_LOG_VERBOSE, "Compensating %d samples per second.\n", comp);
avresample_set_compensation(s->avr, comp, inlink->sample_rate);
}
delta = 0;
}
if (out_size > 0) {
AVFrame *buf_out = ff_get_audio_buffer(outlink, out_size);
if (!buf_out) {
ret = AVERROR(ENOMEM);
goto fail;
}
if (s->first_frame && delta > 0) {
int ch;
av_samples_set_silence(buf_out->extended_data, 0, delta,
nb_channels, buf->format);
for (ch = 0; ch < nb_channels; ch++)
buf_out->extended_data[ch] += delta;
avresample_read(s->avr, buf_out->extended_data, out_size);
for (ch = 0; ch < nb_channels; ch++)
buf_out->extended_data[ch] -= delta;
} else {
avresample_read(s->avr, buf_out->extended_data, out_size);
if (delta > 0) {
av_samples_set_silence(buf_out->extended_data, out_size - delta,
delta, nb_channels, buf->format);
}
}
buf_out->pts = s->pts;
ret = ff_filter_frame(outlink, buf_out);
if (ret < 0)
goto fail;
s->got_output = 1;
} else if (avresample_available(s->avr)) {
av_log(ctx, AV_LOG_WARNING, "Non-monotonous timestamps, dropping "
"whole buffer.\n");
}
/* drain any remaining buffered data */
avresample_read(s->avr, NULL, avresample_available(s->avr));
s->pts = pts - avresample_get_delay(s->avr);
ret = avresample_convert(s->avr, NULL, 0, 0, buf->extended_data,
buf->linesize[0], buf->nb_samples);
s->first_frame = 0;
fail:
av_frame_free(&buf);
return ret;
}
static const AVFilterPad avfilter_af_asyncts_inputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_AUDIO,
.filter_frame = filter_frame
},
{ NULL }
};
static const AVFilterPad avfilter_af_asyncts_outputs[] = {
{
.name = "default",
.type = AVMEDIA_TYPE_AUDIO,
.config_props = config_props,
.request_frame = request_frame
},
{ NULL }
};
AVFilter avfilter_af_asyncts = {
.name = "asyncts",
.description = NULL_IF_CONFIG_SMALL("Sync audio data to timestamps"),
.init = init,
.uninit = uninit,
.priv_size = sizeof(ASyncContext),
.inputs = avfilter_af_asyncts_inputs,
.outputs = avfilter_af_asyncts_outputs,
.priv_class = &asyncts_class,
};