2 Copyright (C) 2012 Carl Hetherington <cth@carlh.net>
4 This program is free software; you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation; either version 2 of the License, or
7 (at your option) any later version.
9 This program 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
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program; if not, write to the Free Software
16 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
20 /** @file src/decoder.cc
21 * @brief Parent class for decoders of content.
26 #include <boost/lexical_cast.hpp>
28 #include <libavfilter/avfiltergraph.h>
29 #include <libavfilter/buffersrc.h>
30 #if (LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR >= 53 && LIBAVFILTER_VERSION_MINOR <= 77) || LIBAVFILTER_VERSION_MAJOR == 3
31 #include <libavfilter/avcodec.h>
32 #include <libavfilter/buffersink.h>
33 #elif LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR == 15
34 #include <libavfilter/vsrc_buffer.h>
36 #include <libavformat/avio.h>
41 #include "film_state.h"
43 #include "exceptions.h"
49 #include "delay_line.h"
50 #include "ffmpeg_compatibility.h"
54 using namespace boost;
56 /** @param s FilmState of the Film.
58 * @param j Job that we are running within, or 0
59 * @param l Log to use.
60 * @param minimal true to do the bare minimum of work; just run through the content. Useful for acquiring
61 * accurate frame counts as quickly as possible. This generates no video or audio output.
62 * @param ignore_length Ignore the content's claimed length when computing progress.
64 Decoder::Decoder (boost::shared_ptr<const FilmState> s, boost::shared_ptr<const Options> o, Job* j, Log* l, bool minimal, bool ignore_length)
70 , _ignore_length (ignore_length)
72 , _buffer_src_context (0)
73 , _buffer_sink_context (0)
74 , _have_setup_video_filters (false)
77 , _audio_frames_processed (0)
79 if (_opt->decode_video_frequency != 0 && _fs->length == 0) {
80 throw DecodeError ("cannot do a partial decode if length == 0");
89 /** Start off a decode processing run */
91 Decoder::process_begin ()
93 _delay_in_bytes = _fs->audio_delay * _fs->audio_sample_rate * _fs->audio_channels * _fs->bytes_per_sample() / 1000;
95 _delay_line = new DelayLine (_delay_in_bytes);
97 _audio_frames_processed = 0;
100 /** Finish off a decode processing run */
102 Decoder::process_end ()
104 if (_delay_in_bytes < 0) {
105 uint8_t remainder[-_delay_in_bytes];
106 _delay_line->get_remaining (remainder);
107 _audio_frames_processed += _delay_in_bytes / (_fs->audio_channels * _fs->bytes_per_sample());
108 Audio (remainder, _delay_in_bytes);
111 /* If we cut the decode off, the audio may be short; push some silence
112 in to get it to the right length.
115 int64_t const audio_short_by_frames =
116 ((int64_t) _fs->dcp_length() * _fs->target_sample_rate() / _fs->frames_per_second)
117 - _audio_frames_processed;
119 if (audio_short_by_frames >= 0) {
122 s << "Adding " << audio_short_by_frames << " frames of silence to the end.";
123 _log->log (s.str ());
125 int64_t bytes = audio_short_by_frames * _fs->audio_channels * _fs->bytes_per_sample();
127 int64_t const silence_size = 64 * 1024;
128 uint8_t silence[silence_size];
129 memset (silence, 0, silence_size);
132 int64_t const t = min (bytes, silence_size);
139 /** Start decoding */
145 if (_job && _ignore_length) {
146 _job->set_progress_unknown ();
149 while (pass () == false) {
150 if (_job && !_ignore_length) {
151 _job->set_progress (float (_video_frame) / _fs->dcp_length());
158 /** Run one pass. This may or may not generate any actual video / audio data;
159 * some decoders may require several passes to generate a single frame.
160 * @return true if we have finished processing all data; otherwise false.
165 if (!_have_setup_video_filters) {
166 setup_video_filters ();
167 _have_setup_video_filters = true;
170 if (_video_frame >= _fs->dcp_length()) {
177 /** Called by subclasses to tell the world that some audio data is ready
178 * @param data Interleaved audio data, in FilmState::audio_sample_format.
179 * @param size Number of bytes of data.
182 Decoder::process_audio (uint8_t* data, int size)
184 /* Samples per channel */
185 int const samples = size / _fs->bytes_per_sample();
187 /* Maybe apply gain */
188 if (_fs->audio_gain != 0) {
189 float const linear_gain = pow (10, _fs->audio_gain / 20);
191 switch (_fs->audio_sample_format) {
192 case AV_SAMPLE_FMT_S16:
193 for (int i = 0; i < samples; ++i) {
194 /* XXX: assumes little-endian; also we should probably be dithering here */
196 /* unsigned sample */
197 int const ou = p[0] | (p[1] << 8);
200 int const os = ou >= 0x8000 ? (- 0x10000 + ou) : ou;
202 /* signed sample with altered gain */
203 int const gs = int (os * linear_gain);
205 /* unsigned sample with altered gain */
206 int const gu = gs > 0 ? gs : (0x10000 + gs);
210 p[1] = (gu & 0xff00) >> 8;
219 /* Update the number of audio frames we've pushed to the encoder */
220 _audio_frames_processed += size / (_fs->audio_channels * _fs->bytes_per_sample ());
222 /* Push into the delay line and then tell the world what we've got */
223 int available = _delay_line->feed (data, size);
224 Audio (data, available);
227 /** Called by subclasses to tell the world that some video data is ready.
228 * We do some post-processing / filtering then emit it for listeners.
229 * @param frame to decode; caller manages memory.
232 Decoder::process_video (AVFrame* frame)
239 /* Use FilmState::length here as our one may be wrong */
242 if (_opt->decode_video_frequency != 0) {
243 gap = _fs->length / _opt->decode_video_frequency;
246 if (_opt->decode_video_frequency != 0 && gap != 0 && (_video_frame % gap) != 0) {
251 #if LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR >= 53 && LIBAVFILTER_VERSION_MINOR <= 61
253 if (av_vsrc_buffer_add_frame (_buffer_src_context, frame, 0) < 0) {
254 throw DecodeError ("could not push buffer into filter chain.");
257 #elif LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR == 15
260 par.num = sample_aspect_ratio_numerator ();
261 par.den = sample_aspect_ratio_denominator ();
263 if (av_vsrc_buffer_add_frame (_buffer_src_context, frame, 0, par) < 0) {
264 throw DecodeError ("could not push buffer into filter chain.");
269 if (av_buffersrc_write_frame (_buffer_src_context, frame) < 0) {
270 throw DecodeError ("could not push buffer into filter chain.");
275 #if LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR >= 15 && LIBAVFILTER_VERSION_MINOR <= 61
276 while (avfilter_poll_frame (_buffer_sink_context->inputs[0])) {
278 while (av_buffersink_read (_buffer_sink_context, 0)) {
281 #if LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR >= 15
283 int r = avfilter_request_frame (_buffer_sink_context->inputs[0]);
285 throw DecodeError ("could not request filtered frame");
288 AVFilterBufferRef* filter_buffer = _buffer_sink_context->inputs[0]->cur_buf;
292 AVFilterBufferRef* filter_buffer;
293 if (av_buffersink_get_buffer_ref (_buffer_sink_context, &filter_buffer, 0) < 0) {
300 /* This takes ownership of filter_buffer */
301 shared_ptr<Image> image (new FilterBufferImage ((PixelFormat) frame->format, filter_buffer));
303 if (_opt->black_after > 0 && _video_frame > _opt->black_after) {
304 image->make_black ();
307 shared_ptr<Subtitle> sub;
308 if (_subtitle && _subtitle->displayed_at (double (last_video_frame()) / rint (_fs->frames_per_second))) {
312 TIMING ("Decoder emits %1", _video_frame);
313 Video (image, _video_frame, sub);
320 /** Set up a video filtering chain to include cropping and any filters that are specified
324 Decoder::setup_video_filters ()
327 Size size_after_crop;
329 if (_opt->apply_crop) {
330 size_after_crop = _fs->cropped_size (native_size ());
331 fs << crop_string (Position (_fs->crop.left, _fs->crop.top), size_after_crop);
333 size_after_crop = native_size ();
334 fs << crop_string (Position (0, 0), size_after_crop);
337 string filters = Filter::ffmpeg_strings (_fs->filters).first;
338 if (!filters.empty ()) {
342 filters += fs.str ();
344 avfilter_register_all ();
346 AVFilterGraph* graph = avfilter_graph_alloc();
348 throw DecodeError ("Could not create filter graph.");
351 AVFilter* buffer_src = avfilter_get_by_name("buffer");
352 if (buffer_src == 0) {
353 throw DecodeError ("Could not find buffer src filter");
356 AVFilter* buffer_sink = get_sink ();
359 a << native_size().width << ":"
360 << native_size().height << ":"
361 << pixel_format() << ":"
362 << time_base_numerator() << ":"
363 << time_base_denominator() << ":"
364 << sample_aspect_ratio_numerator() << ":"
365 << sample_aspect_ratio_denominator();
369 if ((r = avfilter_graph_create_filter (&_buffer_src_context, buffer_src, "in", a.str().c_str(), 0, graph)) < 0) {
370 throw DecodeError ("could not create buffer source");
373 AVBufferSinkParams* sink_params = av_buffersink_params_alloc ();
374 PixelFormat* pixel_fmts = new PixelFormat[2];
375 pixel_fmts[0] = pixel_format ();
376 pixel_fmts[1] = PIX_FMT_NONE;
377 sink_params->pixel_fmts = pixel_fmts;
379 if (avfilter_graph_create_filter (&_buffer_sink_context, buffer_sink, "out", 0, sink_params, graph) < 0) {
380 throw DecodeError ("could not create buffer sink.");
383 AVFilterInOut* outputs = avfilter_inout_alloc ();
384 outputs->name = av_strdup("in");
385 outputs->filter_ctx = _buffer_src_context;
386 outputs->pad_idx = 0;
389 AVFilterInOut* inputs = avfilter_inout_alloc ();
390 inputs->name = av_strdup("out");
391 inputs->filter_ctx = _buffer_sink_context;
395 _log->log ("Using filter chain `" + filters + "'");
397 #if LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR == 15
398 if (avfilter_graph_parse (graph, filters.c_str(), inputs, outputs, 0) < 0) {
399 throw DecodeError ("could not set up filter graph.");
402 if (avfilter_graph_parse (graph, filters.c_str(), &inputs, &outputs, 0) < 0) {
403 throw DecodeError ("could not set up filter graph.");
407 if (avfilter_graph_config (graph, 0) < 0) {
408 throw DecodeError ("could not configure filter graph.");
411 /* XXX: leaking `inputs' / `outputs' ? */
415 Decoder::process_subtitle (shared_ptr<Subtitle> s)
419 if (_opt->apply_crop) {
420 list<shared_ptr<SubtitleImage> > im = _subtitle->images ();
421 for (list<shared_ptr<SubtitleImage> >::iterator i = im.begin(); i != im.end(); ++i) {
422 Position const p = (*i)->position ();
423 (*i)->set_position (Position (p.x - _fs->crop.left, p.y - _fs->crop.top));