libavfilter/af_resample.c
d371e7b9
 /*
d814a839
  * This file is part of FFmpeg.
d371e7b9
  *
d814a839
  * FFmpeg is free software; you can redistribute it and/or
d371e7b9
  * 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.
  *
d814a839
  * FFmpeg is distributed in the hope that it will be useful,
d371e7b9
  * 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
d814a839
  * License along with FFmpeg; if not, write to the Free Software
d371e7b9
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
 /**
  * @file
  * sample format and channel layout conversion audio filter
  */
 
 #include "libavutil/avassert.h"
 #include "libavutil/avstring.h"
1d9c2dc8
 #include "libavutil/common.h"
9f122356
 #include "libavutil/dict.h"
d371e7b9
 #include "libavutil/mathematics.h"
 #include "libavutil/opt.h"
 
 #include "libavresample/avresample.h"
 
 #include "audio.h"
 #include "avfilter.h"
ff1f51a8
 #include "formats.h"
d371e7b9
 #include "internal.h"
 
 typedef struct ResampleContext {
4fa1f52e
     const AVClass *class;
d371e7b9
     AVAudioResampleContext *avr;
9f122356
     AVDictionary *options;
d371e7b9
 
6b15874f
     int resampling;
d371e7b9
     int64_t next_pts;
6cbbf059
     int64_t next_in_pts;
1ffb6456
 
565e4993
     /* set by filter_frame() to signal an output frame to request_frame() */
1ffb6456
     int got_output;
d371e7b9
 } ResampleContext;
 
4fa1f52e
 static av_cold int init(AVFilterContext *ctx, AVDictionary **opts)
9f122356
 {
     ResampleContext *s = ctx->priv;
4fa1f52e
     const AVClass *avr_class = avresample_get_class();
     AVDictionaryEntry *e = NULL;
9f122356
 
4fa1f52e
     while ((e = av_dict_get(*opts, "", e, AV_DICT_IGNORE_SUFFIX))) {
         if (av_opt_find(&avr_class, e->key, NULL, 0,
                         AV_OPT_SEARCH_FAKE_OBJ | AV_OPT_SEARCH_CHILDREN))
             av_dict_set(&s->options, e->key, e->value, 0);
9f122356
     }
 
4fa1f52e
     e = NULL;
     while ((e = av_dict_get(s->options, "", e, AV_DICT_IGNORE_SUFFIX)))
         av_dict_set(opts, e->key, NULL, 0);
 
     /* do not allow the user to override basic format options */
     av_dict_set(&s->options,  "in_channel_layout", NULL, 0);
     av_dict_set(&s->options, "out_channel_layout", NULL, 0);
     av_dict_set(&s->options,  "in_sample_fmt",     NULL, 0);
     av_dict_set(&s->options, "out_sample_fmt",     NULL, 0);
     av_dict_set(&s->options,  "in_sample_rate",    NULL, 0);
     av_dict_set(&s->options, "out_sample_rate",    NULL, 0);
 
9f122356
     return 0;
 }
 
d371e7b9
 static av_cold void uninit(AVFilterContext *ctx)
 {
     ResampleContext *s = ctx->priv;
 
     if (s->avr) {
         avresample_close(s->avr);
         avresample_free(&s->avr);
     }
9f122356
     av_dict_free(&s->options);
d371e7b9
 }
 
 static int query_formats(AVFilterContext *ctx)
 {
     AVFilterLink *inlink  = ctx->inputs[0];
     AVFilterLink *outlink = ctx->outputs[0];
6aaac24d
     AVFilterFormats *in_formats, *out_formats, *in_samplerates, *out_samplerates;
     AVFilterChannelLayouts *in_layouts, *out_layouts;
     int ret;
d371e7b9
 
6aaac24d
     if (!(in_formats      = ff_all_formats         (AVMEDIA_TYPE_AUDIO)) ||
         !(out_formats     = ff_all_formats         (AVMEDIA_TYPE_AUDIO)) ||
         !(in_samplerates  = ff_all_samplerates     (                  )) ||
         !(out_samplerates = ff_all_samplerates     (                  )) ||
         !(in_layouts      = ff_all_channel_layouts (                  )) ||
         !(out_layouts     = ff_all_channel_layouts (                  )))
         return AVERROR(ENOMEM);
ff1f51a8
 
6aaac24d
     if ((ret = ff_formats_ref         (in_formats,      &inlink->out_formats        )) < 0 ||
         (ret = ff_formats_ref         (out_formats,     &outlink->in_formats        )) < 0 ||
         (ret = ff_formats_ref         (in_samplerates,  &inlink->out_samplerates    )) < 0 ||
         (ret = ff_formats_ref         (out_samplerates, &outlink->in_samplerates    )) < 0 ||
         (ret = ff_channel_layouts_ref (in_layouts,      &inlink->out_channel_layouts)) < 0 ||
         (ret = ff_channel_layouts_ref (out_layouts,     &outlink->in_channel_layouts)) < 0)
         return ret;
ff1f51a8
 
d371e7b9
     return 0;
 }
 
 static int config_output(AVFilterLink *outlink)
 {
     AVFilterContext *ctx = outlink->src;
     AVFilterLink *inlink = ctx->inputs[0];
     ResampleContext   *s = ctx->priv;
     char buf1[64], buf2[64];
     int ret;
 
6b15874f
     int64_t resampling_forced;
 
d371e7b9
     if (s->avr) {
         avresample_close(s->avr);
         avresample_free(&s->avr);
     }
 
     if (inlink->channel_layout == outlink->channel_layout &&
         inlink->sample_rate    == outlink->sample_rate    &&
7b556be6
         (inlink->format        == outlink->format ||
         (av_get_channel_layout_nb_channels(inlink->channel_layout)  == 1 &&
          av_get_channel_layout_nb_channels(outlink->channel_layout) == 1 &&
          av_get_planar_sample_fmt(inlink->format) ==
          av_get_planar_sample_fmt(outlink->format))))
d371e7b9
         return 0;
 
     if (!(s->avr = avresample_alloc_context()))
         return AVERROR(ENOMEM);
 
9f122356
     if (s->options) {
0d989dbf
         int ret;
9f122356
         AVDictionaryEntry *e = NULL;
         while ((e = av_dict_get(s->options, "", e, AV_DICT_IGNORE_SUFFIX)))
             av_log(ctx, AV_LOG_VERBOSE, "lavr option: %s=%s\n", e->key, e->value);
 
0d989dbf
         ret = av_opt_set_dict(s->avr, &s->options);
         if (ret < 0)
             return ret;
9f122356
     }
 
d371e7b9
     av_opt_set_int(s->avr,  "in_channel_layout", inlink ->channel_layout, 0);
     av_opt_set_int(s->avr, "out_channel_layout", outlink->channel_layout, 0);
     av_opt_set_int(s->avr,  "in_sample_fmt",     inlink ->format,         0);
     av_opt_set_int(s->avr, "out_sample_fmt",     outlink->format,         0);
     av_opt_set_int(s->avr,  "in_sample_rate",    inlink ->sample_rate,    0);
     av_opt_set_int(s->avr, "out_sample_rate",    outlink->sample_rate,    0);
 
     if ((ret = avresample_open(s->avr)) < 0)
         return ret;
 
6b15874f
     av_opt_get_int(s->avr, "force_resampling", 0, &resampling_forced);
     s->resampling = resampling_forced || (inlink->sample_rate != outlink->sample_rate);
 
     if (s->resampling) {
         outlink->time_base = (AVRational){ 1, outlink->sample_rate };
         s->next_pts        = AV_NOPTS_VALUE;
         s->next_in_pts     = AV_NOPTS_VALUE;
     } else
         outlink->time_base = inlink->time_base;
d371e7b9
 
     av_get_channel_layout_string(buf1, sizeof(buf1),
                                  -1, inlink ->channel_layout);
     av_get_channel_layout_string(buf2, sizeof(buf2),
                                  -1, outlink->channel_layout);
     av_log(ctx, AV_LOG_VERBOSE,
e0d8427d
            "fmt:%s srate:%d cl:%s -> fmt:%s srate:%d cl:%s\n",
d371e7b9
            av_get_sample_fmt_name(inlink ->format), inlink ->sample_rate, buf1,
            av_get_sample_fmt_name(outlink->format), outlink->sample_rate, buf2);
 
     return 0;
 }
 
 static int request_frame(AVFilterLink *outlink)
 {
     AVFilterContext *ctx = outlink->src;
     ResampleContext   *s = ctx->priv;
1ffb6456
     int ret = 0;
 
     s->got_output = 0;
     while (ret >= 0 && !s->got_output)
         ret = ff_request_frame(ctx->inputs[0]);
d371e7b9
 
     /* flush the lavr delay buffer */
     if (ret == AVERROR_EOF && s->avr) {
7e350379
         AVFrame *frame;
ad0fe2f4
         int nb_samples = avresample_get_out_samples(s->avr, 0);
d371e7b9
 
         if (!nb_samples)
             return ret;
 
7e350379
         frame = ff_get_audio_buffer(outlink, nb_samples);
         if (!frame)
d371e7b9
             return AVERROR(ENOMEM);
 
7e350379
         ret = avresample_convert(s->avr, frame->extended_data,
                                  frame->linesize[0], nb_samples,
d371e7b9
                                  NULL, 0, 0);
         if (ret <= 0) {
7e350379
             av_frame_free(&frame);
d371e7b9
             return (ret == 0) ? AVERROR_EOF : ret;
         }
 
c060d046
         frame->nb_samples = ret;
7e350379
         frame->pts = s->next_pts;
         return ff_filter_frame(outlink, frame);
d371e7b9
     }
     return ret;
 }
 
7e350379
 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
d371e7b9
 {
     AVFilterContext  *ctx = inlink->dst;
     ResampleContext    *s = ctx->priv;
     AVFilterLink *outlink = ctx->outputs[0];
cd991462
     int ret;
d371e7b9
 
     if (s->avr) {
7e350379
         AVFrame *out;
cd991462
         int delay, nb_samples;
d371e7b9
 
         /* maximum possible samples lavr can output */
         delay      = avresample_get_delay(s->avr);
ad0fe2f4
         nb_samples = avresample_get_out_samples(s->avr, in->nb_samples);
d371e7b9
 
7e350379
         out = ff_get_audio_buffer(outlink, nb_samples);
         if (!out) {
cd991462
             ret = AVERROR(ENOMEM);
             goto fail;
         }
 
7e350379
         ret = avresample_convert(s->avr, out->extended_data, out->linesize[0],
                                  nb_samples, in->extended_data, in->linesize[0],
                                  in->nb_samples);
ac9a8956
         if (ret <= 0) {
7e350379
             av_frame_free(&out);
ac9a8956
             if (ret < 0)
                 goto fail;
cd991462
         }
d371e7b9
 
         av_assert0(!avresample_available(s->avr));
 
6b15874f
         if (s->resampling && s->next_pts == AV_NOPTS_VALUE) {
7e350379
             if (in->pts == AV_NOPTS_VALUE) {
d371e7b9
                 av_log(ctx, AV_LOG_WARNING, "First timestamp is missing, "
                        "assuming 0.\n");
                 s->next_pts = 0;
             } else
7e350379
                 s->next_pts = av_rescale_q(in->pts, inlink->time_base,
d371e7b9
                                            outlink->time_base);
         }
 
         if (ret > 0) {
7e350379
             out->nb_samples = ret;
dcc7e4bf
 
             ret = av_frame_copy_props(out, in);
             if (ret < 0) {
                 av_frame_free(&out);
                 goto fail;
             }
 
6b15874f
             if (s->resampling) {
                 out->sample_rate = outlink->sample_rate;
                 /* Only convert in->pts if there is a discontinuous jump.
                    This ensures that out->pts tracks the number of samples actually
                    output by the resampler in the absence of such a jump.
                    Otherwise, the rounding in av_rescale_q() and av_rescale()
                    causes off-by-1 errors. */
                 if (in->pts != AV_NOPTS_VALUE && in->pts != s->next_in_pts) {
                     out->pts = av_rescale_q(in->pts, inlink->time_base,
                                                 outlink->time_base) -
                                    av_rescale(delay, outlink->sample_rate,
                                               inlink->sample_rate);
                 } else
                     out->pts = s->next_pts;
 
                 s->next_pts = out->pts + out->nb_samples;
                 s->next_in_pts = in->pts + in->nb_samples;
d371e7b9
             } else
6b15874f
                 out->pts = in->pts;
d371e7b9
 
7e350379
             ret = ff_filter_frame(outlink, out);
1ffb6456
             s->got_output = 1;
d371e7b9
         }
cd991462
 
 fail:
7e350379
         av_frame_free(&in);
1ffb6456
     } else {
7e350379
         in->format = outlink->format;
         ret = ff_filter_frame(outlink, in);
1ffb6456
         s->got_output = 1;
     }
cd991462
 
     return ret;
d371e7b9
 }
 
4fa1f52e
 static const AVClass *resample_child_class_next(const AVClass *prev)
 {
     return prev ? NULL : avresample_get_class();
 }
 
 static void *resample_child_next(void *obj, void *prev)
 {
     ResampleContext *s = obj;
     return prev ? NULL : s->avr;
 }
 
 static const AVClass resample_class = {
     .class_name       = "resample",
     .item_name        = av_default_item_name,
     .version          = LIBAVUTIL_VERSION_INT,
     .child_class_next = resample_child_class_next,
     .child_next       = resample_child_next,
 };
 
568c70e7
 static const AVFilterPad avfilter_af_resample_inputs[] = {
     {
b211607b
         .name          = "default",
         .type          = AVMEDIA_TYPE_AUDIO,
         .filter_frame  = filter_frame,
568c70e7
     },
     { NULL }
 };
 
 static const AVFilterPad avfilter_af_resample_outputs[] = {
     {
         .name          = "default",
         .type          = AVMEDIA_TYPE_AUDIO,
         .config_props  = config_output,
         .request_frame = request_frame
     },
     { NULL }
 };
 
cd43ca04
 AVFilter ff_af_resample = {
d371e7b9
     .name          = "resample",
     .description   = NULL_IF_CONFIG_SMALL("Audio resampling and conversion."),
     .priv_size     = sizeof(ResampleContext),
4fa1f52e
     .priv_class    = &resample_class,
b211607b
     .init_dict     = init,
     .uninit        = uninit,
     .query_formats = query_formats,
     .inputs        = avfilter_af_resample_inputs,
     .outputs       = avfilter_af_resample_outputs,
d371e7b9
 };