FFmpeg
 All Data Structures Namespaces Files Functions Variables Typedefs Enumerations Enumerator Macros Groups Pages
transcoding.c
Go to the documentation of this file.
1 /*
2  * Copyright (c) 2010 Nicolas George
3  * Copyright (c) 2011 Stefano Sabatini
4  * Copyright (c) 2014 Andrey Utkin
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a copy
7  * of this software and associated documentation files (the "Software"), to deal
8  * in the Software without restriction, including without limitation the rights
9  * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
10  * copies of the Software, and to permit persons to whom the Software is
11  * furnished to do so, subject to the following conditions:
12  *
13  * The above copyright notice and this permission notice shall be included in
14  * all copies or substantial portions of the Software.
15  *
16  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
19  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
22  * THE SOFTWARE.
23  */
24 
25 /**
26  * @file
27  * API example for demuxing, decoding, filtering, encoding and muxing
28  * @example transcoding.c
29  */
30 
31 #include <libavcodec/avcodec.h>
32 #include <libavformat/avformat.h>
34 #include <libavfilter/avcodec.h>
35 #include <libavfilter/buffersink.h>
36 #include <libavfilter/buffersrc.h>
37 #include <libavutil/opt.h>
38 #include <libavutil/pixdesc.h>
39 
42 typedef struct FilteringContext {
48 
49 static int open_input_file(const char *filename)
50 {
51  int ret;
52  unsigned int i;
53 
54  ifmt_ctx = NULL;
55  if ((ret = avformat_open_input(&ifmt_ctx, filename, NULL, NULL)) < 0) {
56  av_log(NULL, AV_LOG_ERROR, "Cannot open input file\n");
57  return ret;
58  }
59 
60  if ((ret = avformat_find_stream_info(ifmt_ctx, NULL)) < 0) {
61  av_log(NULL, AV_LOG_ERROR, "Cannot find stream information\n");
62  return ret;
63  }
64 
65  for (i = 0; i < ifmt_ctx->nb_streams; i++) {
66  AVStream *stream;
67  AVCodecContext *codec_ctx;
68  stream = ifmt_ctx->streams[i];
69  codec_ctx = stream->codec;
70  /* Reencode video & audio and remux subtitles etc. */
71  if (codec_ctx->codec_type == AVMEDIA_TYPE_VIDEO
72  || codec_ctx->codec_type == AVMEDIA_TYPE_AUDIO) {
73  /* Open decoder */
74  ret = avcodec_open2(codec_ctx,
75  avcodec_find_decoder(codec_ctx->codec_id), NULL);
76  if (ret < 0) {
77  av_log(NULL, AV_LOG_ERROR, "Failed to open decoder for stream #%u\n", i);
78  return ret;
79  }
80  }
81  }
82 
83  av_dump_format(ifmt_ctx, 0, filename, 0);
84  return 0;
85 }
86 
87 static int open_output_file(const char *filename)
88 {
89  AVStream *out_stream;
90  AVStream *in_stream;
91  AVCodecContext *dec_ctx, *enc_ctx;
92  AVCodec *encoder;
93  int ret;
94  unsigned int i;
95 
96  ofmt_ctx = NULL;
97  avformat_alloc_output_context2(&ofmt_ctx, NULL, NULL, filename);
98  if (!ofmt_ctx) {
99  av_log(NULL, AV_LOG_ERROR, "Could not create output context\n");
100  return AVERROR_UNKNOWN;
101  }
102 
103 
104  for (i = 0; i < ifmt_ctx->nb_streams; i++) {
105  out_stream = avformat_new_stream(ofmt_ctx, NULL);
106  if (!out_stream) {
107  av_log(NULL, AV_LOG_ERROR, "Failed allocating output stream\n");
108  return AVERROR_UNKNOWN;
109  }
110 
111  in_stream = ifmt_ctx->streams[i];
112  dec_ctx = in_stream->codec;
113  enc_ctx = out_stream->codec;
114 
115  if (dec_ctx->codec_type == AVMEDIA_TYPE_VIDEO
116  || dec_ctx->codec_type == AVMEDIA_TYPE_AUDIO) {
117  /* in this example, we choose transcoding to same codec */
118  encoder = avcodec_find_encoder(dec_ctx->codec_id);
119 
120  /* In this example, we transcode to same properties (picture size,
121  * sample rate etc.). These properties can be changed for output
122  * streams easily using filters */
123  if (dec_ctx->codec_type == AVMEDIA_TYPE_VIDEO) {
124  enc_ctx->height = dec_ctx->height;
125  enc_ctx->width = dec_ctx->width;
126  enc_ctx->sample_aspect_ratio = dec_ctx->sample_aspect_ratio;
127  /* take first format from list of supported formats */
128  enc_ctx->pix_fmt = encoder->pix_fmts[0];
129  /* video time_base can be set to whatever is handy and supported by encoder */
130  enc_ctx->time_base = dec_ctx->time_base;
131  } else {
132  enc_ctx->sample_rate = dec_ctx->sample_rate;
133  enc_ctx->channel_layout = dec_ctx->channel_layout;
135  /* take first format from list of supported formats */
136  enc_ctx->sample_fmt = encoder->sample_fmts[0];
137  enc_ctx->time_base = (AVRational){1, enc_ctx->sample_rate};
138  }
139 
140  /* Third parameter can be used to pass settings to encoder */
141  ret = avcodec_open2(enc_ctx, encoder, NULL);
142  if (ret < 0) {
143  av_log(NULL, AV_LOG_ERROR, "Cannot open video encoder for stream #%u\n", i);
144  return ret;
145  }
146  } else if (dec_ctx->codec_type == AVMEDIA_TYPE_UNKNOWN) {
147  av_log(NULL, AV_LOG_FATAL, "Elementary stream #%d is of unknown type, cannot proceed\n", i);
148  return AVERROR_INVALIDDATA;
149  } else {
150  /* if this stream must be remuxed */
151  ret = avcodec_copy_context(ofmt_ctx->streams[i]->codec,
152  ifmt_ctx->streams[i]->codec);
153  if (ret < 0) {
154  av_log(NULL, AV_LOG_ERROR, "Copying stream context failed\n");
155  return ret;
156  }
157  }
158 
159  if (ofmt_ctx->oformat->flags & AVFMT_GLOBALHEADER)
160  enc_ctx->flags |= CODEC_FLAG_GLOBAL_HEADER;
161 
162  }
163  av_dump_format(ofmt_ctx, 0, filename, 1);
164 
165  if (!(ofmt_ctx->oformat->flags & AVFMT_NOFILE)) {
166  ret = avio_open(&ofmt_ctx->pb, filename, AVIO_FLAG_WRITE);
167  if (ret < 0) {
168  av_log(NULL, AV_LOG_ERROR, "Could not open output file '%s'", filename);
169  return ret;
170  }
171  }
172 
173  /* init muxer, write output file header */
174  ret = avformat_write_header(ofmt_ctx, NULL);
175  if (ret < 0) {
176  av_log(NULL, AV_LOG_ERROR, "Error occurred when opening output file\n");
177  return ret;
178  }
179 
180  return 0;
181 }
182 
184  AVCodecContext *enc_ctx, const char *filter_spec)
185 {
186  char args[512];
187  int ret = 0;
188  AVFilter *buffersrc = NULL;
189  AVFilter *buffersink = NULL;
195 
196  if (!outputs || !inputs || !filter_graph) {
197  ret = AVERROR(ENOMEM);
198  goto end;
199  }
200 
201  if (dec_ctx->codec_type == AVMEDIA_TYPE_VIDEO) {
202  buffersrc = avfilter_get_by_name("buffer");
203  buffersink = avfilter_get_by_name("buffersink");
204  if (!buffersrc || !buffersink) {
205  av_log(NULL, AV_LOG_ERROR, "filtering source or sink element not found\n");
206  ret = AVERROR_UNKNOWN;
207  goto end;
208  }
209 
210  snprintf(args, sizeof(args),
211  "video_size=%dx%d:pix_fmt=%d:time_base=%d/%d:pixel_aspect=%d/%d",
212  dec_ctx->width, dec_ctx->height, dec_ctx->pix_fmt,
213  dec_ctx->time_base.num, dec_ctx->time_base.den,
214  dec_ctx->sample_aspect_ratio.num,
215  dec_ctx->sample_aspect_ratio.den);
216 
217  ret = avfilter_graph_create_filter(&buffersrc_ctx, buffersrc, "in",
218  args, NULL, filter_graph);
219  if (ret < 0) {
220  av_log(NULL, AV_LOG_ERROR, "Cannot create buffer source\n");
221  goto end;
222  }
223 
224  ret = avfilter_graph_create_filter(&buffersink_ctx, buffersink, "out",
225  NULL, NULL, filter_graph);
226  if (ret < 0) {
227  av_log(NULL, AV_LOG_ERROR, "Cannot create buffer sink\n");
228  goto end;
229  }
230 
231  ret = av_opt_set_bin(buffersink_ctx, "pix_fmts",
232  (uint8_t*)&enc_ctx->pix_fmt, sizeof(enc_ctx->pix_fmt),
234  if (ret < 0) {
235  av_log(NULL, AV_LOG_ERROR, "Cannot set output pixel format\n");
236  goto end;
237  }
238  } else if (dec_ctx->codec_type == AVMEDIA_TYPE_AUDIO) {
239  buffersrc = avfilter_get_by_name("abuffer");
240  buffersink = avfilter_get_by_name("abuffersink");
241  if (!buffersrc || !buffersink) {
242  av_log(NULL, AV_LOG_ERROR, "filtering source or sink element not found\n");
243  ret = AVERROR_UNKNOWN;
244  goto end;
245  }
246 
247  if (!dec_ctx->channel_layout)
248  dec_ctx->channel_layout =
250  snprintf(args, sizeof(args),
251  "time_base=%d/%d:sample_rate=%d:sample_fmt=%s:channel_layout=0x%"PRIx64,
252  dec_ctx->time_base.num, dec_ctx->time_base.den, dec_ctx->sample_rate,
254  dec_ctx->channel_layout);
255  ret = avfilter_graph_create_filter(&buffersrc_ctx, buffersrc, "in",
256  args, NULL, filter_graph);
257  if (ret < 0) {
258  av_log(NULL, AV_LOG_ERROR, "Cannot create audio buffer source\n");
259  goto end;
260  }
261 
262  ret = avfilter_graph_create_filter(&buffersink_ctx, buffersink, "out",
263  NULL, NULL, filter_graph);
264  if (ret < 0) {
265  av_log(NULL, AV_LOG_ERROR, "Cannot create audio buffer sink\n");
266  goto end;
267  }
268 
269  ret = av_opt_set_bin(buffersink_ctx, "sample_fmts",
270  (uint8_t*)&enc_ctx->sample_fmt, sizeof(enc_ctx->sample_fmt),
272  if (ret < 0) {
273  av_log(NULL, AV_LOG_ERROR, "Cannot set output sample format\n");
274  goto end;
275  }
276 
277  ret = av_opt_set_bin(buffersink_ctx, "channel_layouts",
278  (uint8_t*)&enc_ctx->channel_layout,
279  sizeof(enc_ctx->channel_layout), AV_OPT_SEARCH_CHILDREN);
280  if (ret < 0) {
281  av_log(NULL, AV_LOG_ERROR, "Cannot set output channel layout\n");
282  goto end;
283  }
284 
285  ret = av_opt_set_bin(buffersink_ctx, "sample_rates",
286  (uint8_t*)&enc_ctx->sample_rate, sizeof(enc_ctx->sample_rate),
288  if (ret < 0) {
289  av_log(NULL, AV_LOG_ERROR, "Cannot set output sample rate\n");
290  goto end;
291  }
292  } else {
293  ret = AVERROR_UNKNOWN;
294  goto end;
295  }
296 
297  /* Endpoints for the filter graph. */
298  outputs->name = av_strdup("in");
299  outputs->filter_ctx = buffersrc_ctx;
300  outputs->pad_idx = 0;
301  outputs->next = NULL;
302 
303  inputs->name = av_strdup("out");
304  inputs->filter_ctx = buffersink_ctx;
305  inputs->pad_idx = 0;
306  inputs->next = NULL;
307 
308  if (!outputs->name || !inputs->name) {
309  ret = AVERROR(ENOMEM);
310  goto end;
311  }
312 
313  if ((ret = avfilter_graph_parse_ptr(filter_graph, filter_spec,
314  &inputs, &outputs, NULL)) < 0)
315  goto end;
316 
317  if ((ret = avfilter_graph_config(filter_graph, NULL)) < 0)
318  goto end;
319 
320  /* Fill FilteringContext */
323  fctx->filter_graph = filter_graph;
324 
325 end:
326  avfilter_inout_free(&inputs);
327  avfilter_inout_free(&outputs);
328 
329  return ret;
330 }
331 
332 static int init_filters(void)
333 {
334  const char *filter_spec;
335  unsigned int i;
336  int ret;
337  filter_ctx = av_malloc_array(ifmt_ctx->nb_streams, sizeof(*filter_ctx));
338  if (!filter_ctx)
339  return AVERROR(ENOMEM);
340 
341  for (i = 0; i < ifmt_ctx->nb_streams; i++) {
342  filter_ctx[i].buffersrc_ctx = NULL;
343  filter_ctx[i].buffersink_ctx = NULL;
344  filter_ctx[i].filter_graph = NULL;
345  if (!(ifmt_ctx->streams[i]->codec->codec_type == AVMEDIA_TYPE_AUDIO
346  || ifmt_ctx->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO))
347  continue;
348 
349 
350  if (ifmt_ctx->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO)
351  filter_spec = "null"; /* passthrough (dummy) filter for video */
352  else
353  filter_spec = "anull"; /* passthrough (dummy) filter for audio */
354  ret = init_filter(&filter_ctx[i], ifmt_ctx->streams[i]->codec,
355  ofmt_ctx->streams[i]->codec, filter_spec);
356  if (ret)
357  return ret;
358  }
359  return 0;
360 }
361 
362 static int encode_write_frame(AVFrame *filt_frame, unsigned int stream_index, int *got_frame) {
363  int ret;
364  int got_frame_local;
365  AVPacket enc_pkt;
366  int (*enc_func)(AVCodecContext *, AVPacket *, const AVFrame *, int *) =
367  (ifmt_ctx->streams[stream_index]->codec->codec_type ==
369 
370  if (!got_frame)
371  got_frame = &got_frame_local;
372 
373  av_log(NULL, AV_LOG_INFO, "Encoding frame\n");
374  /* encode filtered frame */
375  enc_pkt.data = NULL;
376  enc_pkt.size = 0;
377  av_init_packet(&enc_pkt);
378  ret = enc_func(ofmt_ctx->streams[stream_index]->codec, &enc_pkt,
379  filt_frame, got_frame);
380  av_frame_free(&filt_frame);
381  if (ret < 0)
382  return ret;
383  if (!(*got_frame))
384  return 0;
385 
386  /* prepare packet for muxing */
387  enc_pkt.stream_index = stream_index;
388  av_packet_rescale_ts(&enc_pkt,
389  ofmt_ctx->streams[stream_index]->codec->time_base,
390  ofmt_ctx->streams[stream_index]->time_base);
391 
392  av_log(NULL, AV_LOG_DEBUG, "Muxing frame\n");
393  /* mux encoded frame */
394  ret = av_interleaved_write_frame(ofmt_ctx, &enc_pkt);
395  return ret;
396 }
397 
398 static int filter_encode_write_frame(AVFrame *frame, unsigned int stream_index)
399 {
400  int ret;
401  AVFrame *filt_frame;
402 
403  av_log(NULL, AV_LOG_INFO, "Pushing decoded frame to filters\n");
404  /* push the decoded frame into the filtergraph */
405  ret = av_buffersrc_add_frame_flags(filter_ctx[stream_index].buffersrc_ctx,
406  frame, 0);
407  if (ret < 0) {
408  av_log(NULL, AV_LOG_ERROR, "Error while feeding the filtergraph\n");
409  return ret;
410  }
411 
412  /* pull filtered frames from the filtergraph */
413  while (1) {
414  filt_frame = av_frame_alloc();
415  if (!filt_frame) {
416  ret = AVERROR(ENOMEM);
417  break;
418  }
419  av_log(NULL, AV_LOG_INFO, "Pulling filtered frame from filters\n");
420  ret = av_buffersink_get_frame(filter_ctx[stream_index].buffersink_ctx,
421  filt_frame);
422  if (ret < 0) {
423  /* if no more frames for output - returns AVERROR(EAGAIN)
424  * if flushed and no more frames for output - returns AVERROR_EOF
425  * rewrite retcode to 0 to show it as normal procedure completion
426  */
427  if (ret == AVERROR(EAGAIN) || ret == AVERROR_EOF)
428  ret = 0;
429  av_frame_free(&filt_frame);
430  break;
431  }
432 
433  filt_frame->pict_type = AV_PICTURE_TYPE_NONE;
434  ret = encode_write_frame(filt_frame, stream_index, NULL);
435  if (ret < 0)
436  break;
437  }
438 
439  return ret;
440 }
441 
442 static int flush_encoder(unsigned int stream_index)
443 {
444  int ret;
445  int got_frame;
446 
447  if (!(ofmt_ctx->streams[stream_index]->codec->codec->capabilities &
449  return 0;
450 
451  while (1) {
452  av_log(NULL, AV_LOG_INFO, "Flushing stream #%u encoder\n", stream_index);
453  ret = encode_write_frame(NULL, stream_index, &got_frame);
454  if (ret < 0)
455  break;
456  if (!got_frame)
457  return 0;
458  }
459  return ret;
460 }
461 
462 int main(int argc, char **argv)
463 {
464  int ret;
465  AVPacket packet = { .data = NULL, .size = 0 };
466  AVFrame *frame = NULL;
467  enum AVMediaType type;
468  unsigned int stream_index;
469  unsigned int i;
470  int got_frame;
471  int (*dec_func)(AVCodecContext *, AVFrame *, int *, const AVPacket *);
472 
473  if (argc != 3) {
474  av_log(NULL, AV_LOG_ERROR, "Usage: %s <input file> <output file>\n", argv[0]);
475  return 1;
476  }
477 
478  av_register_all();
480 
481  if ((ret = open_input_file(argv[1])) < 0)
482  goto end;
483  if ((ret = open_output_file(argv[2])) < 0)
484  goto end;
485  if ((ret = init_filters()) < 0)
486  goto end;
487 
488  /* read all packets */
489  while (1) {
490  if ((ret = av_read_frame(ifmt_ctx, &packet)) < 0)
491  break;
492  stream_index = packet.stream_index;
493  type = ifmt_ctx->streams[packet.stream_index]->codec->codec_type;
494  av_log(NULL, AV_LOG_DEBUG, "Demuxer gave frame of stream_index %u\n",
495  stream_index);
496 
497  if (filter_ctx[stream_index].filter_graph) {
498  av_log(NULL, AV_LOG_DEBUG, "Going to reencode&filter the frame\n");
499  frame = av_frame_alloc();
500  if (!frame) {
501  ret = AVERROR(ENOMEM);
502  break;
503  }
504  av_packet_rescale_ts(&packet,
505  ifmt_ctx->streams[stream_index]->time_base,
506  ifmt_ctx->streams[stream_index]->codec->time_base);
507  dec_func = (type == AVMEDIA_TYPE_VIDEO) ? avcodec_decode_video2 :
509  ret = dec_func(ifmt_ctx->streams[stream_index]->codec, frame,
510  &got_frame, &packet);
511  if (ret < 0) {
512  av_frame_free(&frame);
513  av_log(NULL, AV_LOG_ERROR, "Decoding failed\n");
514  break;
515  }
516 
517  if (got_frame) {
518  frame->pts = av_frame_get_best_effort_timestamp(frame);
519  ret = filter_encode_write_frame(frame, stream_index);
520  av_frame_free(&frame);
521  if (ret < 0)
522  goto end;
523  } else {
524  av_frame_free(&frame);
525  }
526  } else {
527  /* remux this frame without reencoding */
528  av_packet_rescale_ts(&packet,
529  ifmt_ctx->streams[stream_index]->time_base,
530  ofmt_ctx->streams[stream_index]->time_base);
531 
532  ret = av_interleaved_write_frame(ofmt_ctx, &packet);
533  if (ret < 0)
534  goto end;
535  }
536  av_free_packet(&packet);
537  }
538 
539  /* flush filters and encoders */
540  for (i = 0; i < ifmt_ctx->nb_streams; i++) {
541  /* flush filter */
542  if (!filter_ctx[i].filter_graph)
543  continue;
544  ret = filter_encode_write_frame(NULL, i);
545  if (ret < 0) {
546  av_log(NULL, AV_LOG_ERROR, "Flushing filter failed\n");
547  goto end;
548  }
549 
550  /* flush encoder */
551  ret = flush_encoder(i);
552  if (ret < 0) {
553  av_log(NULL, AV_LOG_ERROR, "Flushing encoder failed\n");
554  goto end;
555  }
556  }
557 
558  av_write_trailer(ofmt_ctx);
559 end:
560  av_free_packet(&packet);
561  av_frame_free(&frame);
562  for (i = 0; i < ifmt_ctx->nb_streams; i++) {
563  avcodec_close(ifmt_ctx->streams[i]->codec);
564  if (ofmt_ctx && ofmt_ctx->nb_streams > i && ofmt_ctx->streams[i] && ofmt_ctx->streams[i]->codec)
565  avcodec_close(ofmt_ctx->streams[i]->codec);
566  if (filter_ctx && filter_ctx[i].filter_graph)
567  avfilter_graph_free(&filter_ctx[i].filter_graph);
568  }
569  av_free(filter_ctx);
570  avformat_close_input(&ifmt_ctx);
571  if (ofmt_ctx && !(ofmt_ctx->oformat->flags & AVFMT_NOFILE))
572  avio_close(ofmt_ctx->pb);
573  avformat_free_context(ofmt_ctx);
574 
575  if (ret < 0)
576  av_log(NULL, AV_LOG_ERROR, "Error occurred: %s\n", av_err2str(ret));
577 
578  return ret ? 1 : 0;
579 }