libavfilter/af_resample.c

   1 /*
   2  *
   3  * This file is part of Libav.
   4  *
   5  * Libav is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU Lesser General Public
   7  * License as published by the Free Software Foundation; either
   8  * version 2.1 of the License, or (at your option) any later version.
   9  *
  10  * Libav is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  13  * Lesser General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Lesser General Public
  16  * License along with Libav; if not, write to the Free Software
  17  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  18  */
  19
  20 /**
  21  * @file
  22  * sample format and channel layout conversion audio filter
  23  */
  24
  25 #include "libavutil/avassert.h"
  26 #include "libavutil/avstring.h"
  27 #include "libavutil/common.h"
  28 #include "libavutil/dict.h"
  29 #include "libavutil/mathematics.h"
  30 #include "libavutil/opt.h"
  31
  32 #include "libavresample/avresample.h"
  33
  34 #include "audio.h"
  35 #include "avfilter.h"
  36 #include "formats.h"
  37 #include "internal.h"
  38
  39 typedef struct ResampleContext {
  40     AVAudioResampleContext *avr;
  41     AVDictionary *options;
  42
  43     int64_t next_pts;
  44
  45     /* set by filter_frame() to signal an output frame to request_frame() */
  46     int got_output;
  47 } ResampleContext;
  48
  49 static av_cold int init(AVFilterContext *ctx, const char *args)
  50 {
  51     ResampleContext *s = ctx->priv;
  52
  53     if (args) {
  54         int ret = av_dict_parse_string(&s->options, args, "=", ":", 0);
  55         if (ret < 0) {
  56             av_log(ctx, AV_LOG_ERROR, "error setting option string: %s\n", args);
  57             return ret;
  58         }
  59
  60         /* do not allow the user to override basic format options */
  61         av_dict_set(&s->options,  "in_channel_layout", NULL, 0);
  62         av_dict_set(&s->options, "out_channel_layout", NULL, 0);
  63         av_dict_set(&s->options,  "in_sample_fmt",     NULL, 0);
  64         av_dict_set(&s->options, "out_sample_fmt",     NULL, 0);
  65         av_dict_set(&s->options,  "in_sample_rate",    NULL, 0);
  66         av_dict_set(&s->options, "out_sample_rate",    NULL, 0);
  67     }
  68
  69     return 0;
  70 }
  71
  72 static av_cold void uninit(AVFilterContext *ctx)
  73 {
  74     ResampleContext *s = ctx->priv;
  75
  76     if (s->avr) {
  77         avresample_close(s->avr);
  78         avresample_free(&s->avr);
  79     }
  80     av_dict_free(&s->options);
  81 }
  82
  83 static int query_formats(AVFilterContext *ctx)
  84 {
  85     AVFilterLink *inlink  = ctx->inputs[0];
  86     AVFilterLink *outlink = ctx->outputs[0];
  87
  88     AVFilterFormats        *in_formats      = ff_all_formats(AVMEDIA_TYPE_AUDIO);
  89     AVFilterFormats        *out_formats     = ff_all_formats(AVMEDIA_TYPE_AUDIO);
  90     AVFilterFormats        *in_samplerates  = ff_all_samplerates();
  91     AVFilterFormats        *out_samplerates = ff_all_samplerates();
  92     AVFilterChannelLayouts *in_layouts      = ff_all_channel_layouts();
  93     AVFilterChannelLayouts *out_layouts     = ff_all_channel_layouts();
  94
  95     ff_formats_ref(in_formats,  &inlink->out_formats);
  96     ff_formats_ref(out_formats, &outlink->in_formats);
  97
  98     ff_formats_ref(in_samplerates,  &inlink->out_samplerates);
  99     ff_formats_ref(out_samplerates, &outlink->in_samplerates);
 100
 101     ff_channel_layouts_ref(in_layouts,  &inlink->out_channel_layouts);
 102     ff_channel_layouts_ref(out_layouts, &outlink->in_channel_layouts);
 103
 104     return 0;
 105 }
 106
 107 static int config_output(AVFilterLink *outlink)
 108 {
 109     AVFilterContext *ctx = outlink->src;
 110     AVFilterLink *inlink = ctx->inputs[0];
 111     ResampleContext   *s = ctx->priv;
 112     char buf1[64], buf2[64];
 113     int ret;
 114
 115     if (s->avr) {
 116         avresample_close(s->avr);
 117         avresample_free(&s->avr);
 118     }
 119
 120     if (inlink->channel_layout == outlink->channel_layout &&
 121         inlink->sample_rate    == outlink->sample_rate    &&
 122         (inlink->format        == outlink->format ||
 123         (av_get_channel_layout_nb_channels(inlink->channel_layout)  == 1 &&
 124          av_get_channel_layout_nb_channels(outlink->channel_layout) == 1 &&
 125          av_get_planar_sample_fmt(inlink->format) ==
 126          av_get_planar_sample_fmt(outlink->format))))
 127         return 0;
 128
 129     if (!(s->avr = avresample_alloc_context()))
 130         return AVERROR(ENOMEM);
 131
 132     if (s->options) {
 133         AVDictionaryEntry *e = NULL;
 134         while ((e = av_dict_get(s->options, "", e, AV_DICT_IGNORE_SUFFIX)))
 135             av_log(ctx, AV_LOG_VERBOSE, "lavr option: %s=%s\n", e->key, e->value);
 136
 137         av_opt_set_dict(s->avr, &s->options);
 138     }
 139
 140     av_opt_set_int(s->avr,  "in_channel_layout", inlink ->channel_layout, 0);
 141     av_opt_set_int(s->avr, "out_channel_layout", outlink->channel_layout, 0);
 142     av_opt_set_int(s->avr,  "in_sample_fmt",     inlink ->format,         0);
 143     av_opt_set_int(s->avr, "out_sample_fmt",     outlink->format,         0);
 144     av_opt_set_int(s->avr,  "in_sample_rate",    inlink ->sample_rate,    0);
 145     av_opt_set_int(s->avr, "out_sample_rate",    outlink->sample_rate,    0);
 146
 147     if ((ret = avresample_open(s->avr)) < 0)
 148         return ret;
 149
 150     outlink->time_base = (AVRational){ 1, outlink->sample_rate };
 151     s->next_pts        = AV_NOPTS_VALUE;
 152
 153     av_get_channel_layout_string(buf1, sizeof(buf1),
 154                                  -1, inlink ->channel_layout);
 155     av_get_channel_layout_string(buf2, sizeof(buf2),
 156                                  -1, outlink->channel_layout);
 157     av_log(ctx, AV_LOG_VERBOSE,
 158            "fmt:%s srate:%d cl:%s -> fmt:%s srate:%d cl:%s\n",
 159            av_get_sample_fmt_name(inlink ->format), inlink ->sample_rate, buf1,
 160            av_get_sample_fmt_name(outlink->format), outlink->sample_rate, buf2);
 161
 162     return 0;
 163 }
 164
 165 static int request_frame(AVFilterLink *outlink)
 166 {
 167     AVFilterContext *ctx = outlink->src;
 168     ResampleContext   *s = ctx->priv;
 169     int ret = 0;
 170
 171     s->got_output = 0;
 172     while (ret >= 0 && !s->got_output)
 173         ret = ff_request_frame(ctx->inputs[0]);
 174
 175     /* flush the lavr delay buffer */
 176     if (ret == AVERROR_EOF && s->avr) {
 177         AVFrame *frame;
 178         int nb_samples = av_rescale_rnd(avresample_get_delay(s->avr),
 179                                         outlink->sample_rate,
 180                                         ctx->inputs[0]->sample_rate,
 181                                         AV_ROUND_UP);
 182
 183         if (!nb_samples)
 184             return ret;
 185
 186         frame = ff_get_audio_buffer(outlink, nb_samples);
 187         if (!frame)
 188             return AVERROR(ENOMEM);
 189
 190         ret = avresample_convert(s->avr, frame->extended_data,
 191                                  frame->linesize[0], nb_samples,
 192                                  NULL, 0, 0);
 193         if (ret <= 0) {
 194             av_frame_free(&frame);
 195             return (ret == 0) ? AVERROR_EOF : ret;
 196         }
 197
 198         frame->pts = s->next_pts;
 199         return ff_filter_frame(outlink, frame);
 200     }
 201     return ret;
 202 }
 203
 204 static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 205 {
 206     AVFilterContext  *ctx = inlink->dst;
 207     ResampleContext    *s = ctx->priv;
 208     AVFilterLink *outlink = ctx->outputs[0];
 209     int ret;
 210
 211     if (s->avr) {
 212         AVFrame *out;
 213         int delay, nb_samples;
 214
 215         /* maximum possible samples lavr can output */
 216         delay      = avresample_get_delay(s->avr);
 217         nb_samples = av_rescale_rnd(in->nb_samples + delay,
 218                                     outlink->sample_rate, inlink->sample_rate,
 219                                     AV_ROUND_UP);
 220
 221         out = ff_get_audio_buffer(outlink, nb_samples);
 222         if (!out) {
 223             ret = AVERROR(ENOMEM);
 224             goto fail;
 225         }
 226
 227         ret = avresample_convert(s->avr, out->extended_data, out->linesize[0],
 228                                  nb_samples, in->extended_data, in->linesize[0],
 229                                  in->nb_samples);
 230         if (ret <= 0) {
 231             av_frame_free(&out);
 232             if (ret < 0)
 233                 goto fail;
 234         }
 235
 236         av_assert0(!avresample_available(s->avr));
 237
 238         if (s->next_pts == AV_NOPTS_VALUE) {
 239             if (in->pts == AV_NOPTS_VALUE) {
 240                 av_log(ctx, AV_LOG_WARNING, "First timestamp is missing, "
 241                        "assuming 0.\n");
 242                 s->next_pts = 0;
 243             } else
 244                 s->next_pts = av_rescale_q(in->pts, inlink->time_base,
 245                                            outlink->time_base);
 246         }
 247
 248         if (ret > 0) {
 249             out->nb_samples = ret;
 250             if (in->pts != AV_NOPTS_VALUE) {
 251                 out->pts = av_rescale_q(in->pts, inlink->time_base,
 252                                             outlink->time_base) -
 253                                av_rescale(delay, outlink->sample_rate,
 254                                           inlink->sample_rate);
 255             } else
 256                 out->pts = s->next_pts;
 257
 258             s->next_pts = out->pts + out->nb_samples;
 259
 260             ret = ff_filter_frame(outlink, out);
 261             s->got_output = 1;
 262         }
 263
 264 fail:
 265         av_frame_free(&in);
 266     } else {
 267         in->format = outlink->format;
 268         ret = ff_filter_frame(outlink, in);
 269         s->got_output = 1;
 270     }
 271
 272     return ret;
 273 }
 274
 275 static const AVFilterPad avfilter_af_resample_inputs[] = {
 276     {
 277         .name           = "default",
 278         .type           = AVMEDIA_TYPE_AUDIO,
 279         .filter_frame   = filter_frame,
 280     },
 281     { NULL }
 282 };
 283
 284 static const AVFilterPad avfilter_af_resample_outputs[] = {
 285     {
 286         .name          = "default",
 287         .type          = AVMEDIA_TYPE_AUDIO,
 288         .config_props  = config_output,
 289         .request_frame = request_frame
 290     },
 291     { NULL }
 292 };
 293
 294 AVFilter avfilter_af_resample = {
 295     .name          = "resample",
 296     .description   = NULL_IF_CONFIG_SMALL("Audio resampling and conversion."),
 297     .priv_size     = sizeof(ResampleContext),
 298
 299     .init           = init,
 300     .uninit         = uninit,
 301     .query_formats  = query_formats,
 302
 303     .inputs    = avfilter_af_resample_inputs,
 304     .outputs   = avfilter_af_resample_outputs,
 305 };