fate-run: make avconv() properly deal with multiple inputs.
[libav.git] / libavfilter / af_asyncts.c
CommitLineData
9f26421b
AK
1/*
2 * This file is part of Libav.
3 *
4 * Libav is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public
6 * License as published by the Free Software Foundation; either
7 * version 2.1 of the License, or (at your option) any later version.
8 *
9 * Libav is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
13 *
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with Libav; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17 */
18
19#include "libavresample/avresample.h"
20#include "libavutil/audio_fifo.h"
21#include "libavutil/mathematics.h"
22#include "libavutil/opt.h"
23#include "libavutil/samplefmt.h"
24
25#include "audio.h"
26#include "avfilter.h"
803391f7 27#include "internal.h"
9f26421b
AK
28
29typedef struct ASyncContext {
30 const AVClass *class;
31
32 AVAudioResampleContext *avr;
33 int64_t pts; ///< timestamp in samples of the first sample in fifo
34 int min_delta; ///< pad/trim min threshold in samples
35
36 /* options */
37 int resample;
38 float min_delta_sec;
39 int max_comp;
6f834293
AK
40
41 /* set by filter_samples() to signal an output frame to request_frame() */
42 int got_output;
9f26421b
AK
43} ASyncContext;
44
45#define OFFSET(x) offsetof(ASyncContext, x)
46#define A AV_OPT_FLAG_AUDIO_PARAM
47static const AVOption options[] = {
48 { "compensate", "Stretch/squeeze the data to make it match the timestamps", OFFSET(resample), AV_OPT_TYPE_INT, { 0 }, 0, 1, A },
49 { "min_delta", "Minimum difference between timestamps and audio data "
50 "(in seconds) to trigger padding/trimmin the data.", OFFSET(min_delta_sec), AV_OPT_TYPE_FLOAT, { 0.1 }, 0, INT_MAX, A },
51 { "max_comp", "Maximum compensation in samples per second.", OFFSET(max_comp), AV_OPT_TYPE_INT, { 500 }, 0, INT_MAX, A },
52 { NULL },
53};
54
55static const AVClass async_class = {
56 .class_name = "asyncts filter",
57 .item_name = av_default_item_name,
58 .option = options,
59 .version = LIBAVUTIL_VERSION_INT,
60};
61
a5e8c41c 62static int init(AVFilterContext *ctx, const char *args)
9f26421b
AK
63{
64 ASyncContext *s = ctx->priv;
65 int ret;
66
67 s->class = &async_class;
68 av_opt_set_defaults(s);
69
70 if ((ret = av_set_options_string(s, args, "=", ":")) < 0) {
71 av_log(ctx, AV_LOG_ERROR, "Error parsing options string '%s'.\n", args);
72 return ret;
73 }
74 av_opt_free(s);
75
76 s->pts = AV_NOPTS_VALUE;
77
78 return 0;
79}
80
81static void uninit(AVFilterContext *ctx)
82{
83 ASyncContext *s = ctx->priv;
84
85 if (s->avr) {
86 avresample_close(s->avr);
87 avresample_free(&s->avr);
88 }
89}
90
91static int config_props(AVFilterLink *link)
92{
93 ASyncContext *s = link->src->priv;
94 int ret;
95
96 s->min_delta = s->min_delta_sec * link->sample_rate;
97 link->time_base = (AVRational){1, link->sample_rate};
98
99 s->avr = avresample_alloc_context();
100 if (!s->avr)
101 return AVERROR(ENOMEM);
102
103 av_opt_set_int(s->avr, "in_channel_layout", link->channel_layout, 0);
104 av_opt_set_int(s->avr, "out_channel_layout", link->channel_layout, 0);
105 av_opt_set_int(s->avr, "in_sample_fmt", link->format, 0);
106 av_opt_set_int(s->avr, "out_sample_fmt", link->format, 0);
107 av_opt_set_int(s->avr, "in_sample_rate", link->sample_rate, 0);
108 av_opt_set_int(s->avr, "out_sample_rate", link->sample_rate, 0);
109
110 if (s->resample)
111 av_opt_set_int(s->avr, "force_resampling", 1, 0);
112
113 if ((ret = avresample_open(s->avr)) < 0)
114 return ret;
115
116 return 0;
117}
118
119static int request_frame(AVFilterLink *link)
120{
121 AVFilterContext *ctx = link->src;
122 ASyncContext *s = ctx->priv;
6f834293 123 int ret = 0;
9f26421b
AK
124 int nb_samples;
125
6f834293
AK
126 s->got_output = 0;
127 while (ret >= 0 && !s->got_output)
128 ret = ff_request_frame(ctx->inputs[0]);
129
9f26421b
AK
130 /* flush the fifo */
131 if (ret == AVERROR_EOF && (nb_samples = avresample_get_delay(s->avr))) {
132 AVFilterBufferRef *buf = ff_get_audio_buffer(link, AV_PERM_WRITE,
133 nb_samples);
134 if (!buf)
135 return AVERROR(ENOMEM);
136 avresample_convert(s->avr, (void**)buf->extended_data, buf->linesize[0],
137 nb_samples, NULL, 0, 0);
138 buf->pts = s->pts;
139 ff_filter_samples(link, buf);
140 return 0;
141 }
142
143 return ret;
144}
145
146static void write_to_fifo(ASyncContext *s, AVFilterBufferRef *buf)
147{
148 avresample_convert(s->avr, NULL, 0, 0, (void**)buf->extended_data,
149 buf->linesize[0], buf->audio->nb_samples);
150 avfilter_unref_buffer(buf);
151}
152
153/* get amount of data currently buffered, in samples */
154static int64_t get_delay(ASyncContext *s)
155{
156 return avresample_available(s->avr) + avresample_get_delay(s->avr);
157}
158
159static void filter_samples(AVFilterLink *inlink, AVFilterBufferRef *buf)
160{
161 AVFilterContext *ctx = inlink->dst;
162 ASyncContext *s = ctx->priv;
163 AVFilterLink *outlink = ctx->outputs[0];
164 int nb_channels = av_get_channel_layout_nb_channels(buf->audio->channel_layout);
165 int64_t pts = (buf->pts == AV_NOPTS_VALUE) ? buf->pts :
166 av_rescale_q(buf->pts, inlink->time_base, outlink->time_base);
167 int out_size;
168 int64_t delta;
169
170 /* buffer data until we get the first timestamp */
171 if (s->pts == AV_NOPTS_VALUE) {
172 if (pts != AV_NOPTS_VALUE) {
173 s->pts = pts - get_delay(s);
174 }
175 write_to_fifo(s, buf);
176 return;
177 }
178
179 /* now wait for the next timestamp */
180 if (pts == AV_NOPTS_VALUE) {
181 write_to_fifo(s, buf);
182 return;
183 }
184
185 /* when we have two timestamps, compute how many samples would we have
186 * to add/remove to get proper sync between data and timestamps */
187 delta = pts - s->pts - get_delay(s);
188 out_size = avresample_available(s->avr);
189
190 if (labs(delta) > s->min_delta) {
191 av_log(ctx, AV_LOG_VERBOSE, "Discontinuity - %"PRId64" samples.\n", delta);
192 out_size += delta;
f297dd38
AK
193 } else {
194 if (s->resample) {
195 int comp = av_clip(delta, -s->max_comp, s->max_comp);
196 av_log(ctx, AV_LOG_VERBOSE, "Compensating %d samples per second.\n", comp);
197 avresample_set_compensation(s->avr, delta, inlink->sample_rate);
198 }
199 delta = 0;
9f26421b
AK
200 }
201
202 if (out_size > 0) {
203 AVFilterBufferRef *buf_out = ff_get_audio_buffer(outlink, AV_PERM_WRITE,
204 out_size);
205 if (!buf_out)
206 return;
207
208 avresample_read(s->avr, (void**)buf_out->extended_data, out_size);
209 buf_out->pts = s->pts;
210
211 if (delta > 0) {
212 av_samples_set_silence(buf_out->extended_data, out_size - delta,
213 delta, nb_channels, buf->format);
214 }
215 ff_filter_samples(outlink, buf_out);
6f834293 216 s->got_output = 1;
9f26421b
AK
217 } else {
218 av_log(ctx, AV_LOG_WARNING, "Non-monotonous timestamps, dropping "
219 "whole buffer.\n");
220 }
221
222 /* drain any remaining buffered data */
223 avresample_read(s->avr, NULL, avresample_available(s->avr));
224
225 s->pts = pts - avresample_get_delay(s->avr);
226 avresample_convert(s->avr, NULL, 0, 0, (void**)buf->extended_data,
227 buf->linesize[0], buf->audio->nb_samples);
228 avfilter_unref_buffer(buf);
229}
230
231AVFilter avfilter_af_asyncts = {
232 .name = "asyncts",
233 .description = NULL_IF_CONFIG_SMALL("Sync audio data to timestamps"),
234
235 .init = init,
236 .uninit = uninit,
237
238 .priv_size = sizeof(ASyncContext),
239
240 .inputs = (const AVFilterPad[]) {{ .name = "default",
241 .type = AVMEDIA_TYPE_AUDIO,
242 .filter_samples = filter_samples },
243 { NULL }},
244 .outputs = (const AVFilterPad[]) {{ .name = "default",
245 .type = AVMEDIA_TYPE_AUDIO,
246 .config_props = config_props,
247 .request_frame = request_frame },
248 { NULL }},
249};