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>
31 #include "exceptions.h"
36 #include "delay_line.h"
38 #include "filter_graph.h"
41 using std::stringstream;
44 using boost::shared_ptr;
48 * @param j Job that we are running within, or 0
49 * @param minimal true to do the bare minimum of work; just run through the content. Useful for acquiring
50 * accurate frame counts as quickly as possible. This generates no video or audio output.
52 Decoder::Decoder (boost::shared_ptr<Film> f, boost::shared_ptr<const Options> o, Job* j, bool minimal)
57 , _video_frames_in (0)
58 , _video_frames_out (0)
59 , _audio_frames_in (0)
60 , _audio_frames_out (0)
72 /** Start off a decode processing run. This should only be called once on
73 * a given Decoder object.
76 Decoder::process_begin ()
78 _delay_in_bytes = _film->audio_delay() * audio_sample_rate() * audio_channels() * bytes_per_audio_sample() / 1000;
79 _delay_line = new DelayLine (_delay_in_bytes);
82 /** Finish off a decode processing run */
84 Decoder::process_end ()
86 if (_delay_in_bytes < 0) {
87 /* Empty the delay line */
88 uint8_t remainder[-_delay_in_bytes];
89 _delay_line->get_remaining (remainder);
90 emit_audio (remainder, -_delay_in_bytes);
93 if (_opt->decode_audio) {
95 /* Ensure that our video and audio emissions are the same length */
97 int64_t video_frames_out_in_audio_frames = ((int64_t) _video_frames_out * audio_sample_rate() / frames_per_second());
98 int64_t audio_short_by_frames = video_frames_out_in_audio_frames - _audio_frames_out;
101 String::compose ("Decoder has emitted %1 video frames (which equals %2 audio frames) and %3 audio frames",
103 video_frames_out_in_audio_frames,
107 if (audio_short_by_frames < 0) {
109 _film->log()->log (String::compose ("Emitted %1 too many audio frames", -audio_short_by_frames));
111 /* We have emitted more audio than video. Emit enough black video frames so that we reverse this */
112 int const black_video_frames = ceil (-audio_short_by_frames * frames_per_second() / audio_sample_rate());
114 _film->log()->log (String::compose ("Emitting %1 frames of black video", black_video_frames));
116 shared_ptr<Image> black (new CompactImage (pixel_format(), native_size()));
117 black->make_black ();
118 for (int i = 0; i < black_video_frames; ++i) {
119 emit_video (black, shared_ptr<Subtitle> ());
122 /* Now recompute our check values */
123 video_frames_out_in_audio_frames = ((int64_t) _video_frames_out * audio_sample_rate() / frames_per_second());
124 audio_short_by_frames = video_frames_out_in_audio_frames - _audio_frames_out;
127 if (audio_short_by_frames > 0) {
129 _film->log()->log (String::compose ("Emitted %1 too few audio frames", audio_short_by_frames));
131 /* XXX: this is slightly questionable; does memset () give silence with all
135 int64_t bytes = audio_short_by_frames * _film->audio_channels() * bytes_per_audio_sample();
137 int64_t const silence_size = 16 * 1024 * _film->audio_channels() * bytes_per_audio_sample();
138 uint8_t silence[silence_size];
139 memset (silence, 0, silence_size);
142 int64_t const t = min (bytes, silence_size);
143 emit_audio (silence, t);
150 /** Start decoding */
156 if (_job && !_film->dcp_length()) {
157 _job->set_progress_unknown ();
160 while (pass () == false) {
161 if (_job && _film->dcp_length()) {
162 _job->set_progress (float (_video_frames_out) / _film->dcp_length().get());
169 /** Called by subclasses to tell the world that some audio data is ready
170 * @param data Audio data, in Film::audio_sample_format.
171 * @param size Number of bytes of data.
174 Decoder::process_audio (uint8_t* data, int size)
176 int const audio_frames_in_in_video_frames = _audio_frames_in * frames_per_second() / audio_sample_rate();
177 if (!within_range (audio_frames_in_in_video_frames)) {
181 /* Push into the delay line */
182 size = _delay_line->feed (data, size);
184 emit_audio (data, size);
188 Decoder::emit_audio (uint8_t* data, int size)
194 assert (_film->audio_channels());
195 assert (bytes_per_audio_sample());
197 /* Deinterleave and convert to float */
199 assert ((size % (bytes_per_audio_sample() * audio_channels())) == 0);
201 int const total_samples = size / bytes_per_audio_sample();
202 int const frames = total_samples / _film->audio_channels();
203 shared_ptr<AudioBuffers> audio (new AudioBuffers (audio_channels(), frames));
205 switch (audio_sample_format()) {
206 case AV_SAMPLE_FMT_S16:
208 int16_t* p = (int16_t *) data;
211 for (int i = 0; i < total_samples; ++i) {
212 audio->data(channel)[sample] = float(*p++) / (1 << 15);
215 if (channel == _film->audio_channels()) {
223 case AV_SAMPLE_FMT_S32:
225 int32_t* p = (int32_t *) data;
228 for (int i = 0; i < total_samples; ++i) {
229 audio->data(channel)[sample] = float(*p++) / (1 << 31);
232 if (channel == _film->audio_channels()) {
239 case AV_SAMPLE_FMT_FLTP:
241 float* p = reinterpret_cast<float*> (data);
242 for (int i = 0; i < _film->audio_channels(); ++i) {
243 memcpy (audio->data(i), p, frames * sizeof(float));
253 /* Maybe apply gain */
254 if (_film->audio_gain() != 0) {
255 float const linear_gain = pow (10, _film->audio_gain() / 20);
256 for (int i = 0; i < _film->audio_channels(); ++i) {
257 for (int j = 0; j < frames; ++j) {
258 audio->data(i)[j] *= linear_gain;
263 /* Update the number of audio frames we've pushed to the encoder */
264 _audio_frames_out += audio->frames ();
269 /** Called by subclasses to tell the world that some video data is ready.
270 * We do some post-processing / filtering then emit it for listeners.
271 * @param frame to decode; caller manages memory.
274 Decoder::process_video (AVFrame* frame)
276 assert (_film->length());
283 /* Use Film::length here as our one may be wrong */
285 if (_opt->decode_video_skip != 0 && (_video_frames_in % _opt->decode_video_skip) != 0) {
290 if (!within_range (_video_frames_in)) {
295 shared_ptr<FilterGraph> graph;
297 list<shared_ptr<FilterGraph> >::iterator i = _filter_graphs.begin();
298 while (i != _filter_graphs.end() && !(*i)->can_process (Size (frame->width, frame->height), (AVPixelFormat) frame->format)) {
302 if (i == _filter_graphs.end ()) {
303 graph.reset (new FilterGraph (_film, this, _opt->apply_crop, Size (frame->width, frame->height), (AVPixelFormat) frame->format));
304 _filter_graphs.push_back (graph);
305 _film->log()->log (String::compose ("New graph for %1x%2, pixel format %3", frame->width, frame->height, frame->format));
310 list<shared_ptr<Image> > images = graph->process (frame);
312 for (list<shared_ptr<Image> >::iterator i = images.begin(); i != images.end(); ++i) {
313 shared_ptr<Subtitle> sub;
314 if (_timed_subtitle && _timed_subtitle->displayed_at (double (video_frames_in()) / _film->frames_per_second())) {
315 sub = _timed_subtitle->subtitle ();
318 emit_video (*i, sub);
323 Decoder::repeat_last_video ()
326 _last_image.reset (new CompactImage (pixel_format(), native_size()));
327 _last_image->make_black ();
330 emit_video (_last_image, _last_subtitle);
334 Decoder::emit_video (shared_ptr<Image> image, shared_ptr<Subtitle> sub)
336 TIMING ("Decoder emits %1", _video_frames_out);
337 Video (image, _video_frames_out, sub);
340 _last_subtitle = sub;
344 Decoder::process_subtitle (shared_ptr<TimedSubtitle> s)
348 if (_timed_subtitle && _opt->apply_crop) {
349 Position const p = _timed_subtitle->subtitle()->position ();
350 _timed_subtitle->subtitle()->set_position (Position (p.x - _film->crop().left, p.y - _film->crop().top));
356 Decoder::bytes_per_audio_sample () const
358 return av_get_bytes_per_sample (audio_sample_format ());
361 /** @param s A video frame index within the source */
363 Decoder::within_range (SourceFrames s) const
365 return (s >= _film->dcp_trim_start() && s < (_film->length().get() + _film->dcp_trim_start()));