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"
37 #include "delay_line.h"
38 #include "ffmpeg_compatibility.h"
42 using std::stringstream;
45 using boost::shared_ptr;
49 * @param j Job that we are running within, or 0
50 * @param minimal true to do the bare minimum of work; just run through the content. Useful for acquiring
51 * accurate frame counts as quickly as possible. This generates no video or audio output.
52 * @param ignore_length Ignore the content's claimed length when computing progress.
54 Decoder::Decoder (boost::shared_ptr<Film> f, boost::shared_ptr<const Options> o, Job* j, bool minimal, bool ignore_length)
59 , _ignore_length (ignore_length)
63 , _audio_frames_processed (0)
65 if (_opt->decode_video_frequency != 0 && !_film->length()) {
66 throw DecodeError ("cannot do a partial decode if length is unknown");
75 /** Start off a decode processing run */
77 Decoder::process_begin ()
79 _delay_in_bytes = _film->audio_delay() * audio_sample_rate() * audio_channels() * bytes_per_audio_sample() / 1000;
81 _delay_line = new DelayLine (_delay_in_bytes);
83 _audio_frames_processed = 0;
86 /** Finish off a decode processing run */
88 Decoder::process_end ()
90 if (_delay_in_bytes < 0) {
91 uint8_t remainder[-_delay_in_bytes];
92 _delay_line->get_remaining (remainder);
93 _audio_frames_processed += _delay_in_bytes / (audio_channels() * bytes_per_audio_sample());
94 emit_audio (remainder, -_delay_in_bytes);
97 /* If we cut the decode off, the audio may be short; push some silence
98 in to get it to the right length.
101 int64_t const video_length_in_audio_frames = ((int64_t) last_video_frame() * audio_sample_rate() / frames_per_second());
102 int64_t const audio_short_by_frames = video_length_in_audio_frames - _audio_frames_processed;
105 String::compose ("DCP length is %1 (%2 audio frames); %3 frames of audio processed.",
107 video_length_in_audio_frames,
108 _audio_frames_processed)
111 if (audio_short_by_frames >= 0 && _opt->decode_audio) {
113 _film->log()->log (String::compose ("DCP length is %1; %2 frames of audio processed.", last_video_frame(), _audio_frames_processed));
114 _film->log()->log (String::compose ("Adding %1 frames of silence to the end.", audio_short_by_frames));
116 /* XXX: this is slightly questionable; does memset () give silence with all
120 int64_t bytes = audio_short_by_frames * _film->audio_channels() * bytes_per_audio_sample();
122 int64_t const silence_size = 16 * 1024 * _film->audio_channels() * bytes_per_audio_sample();
123 uint8_t silence[silence_size];
124 memset (silence, 0, silence_size);
127 int64_t const t = min (bytes, silence_size);
128 emit_audio (silence, t);
134 /** Start decoding */
140 if (_job && !_film->dcp_length()) {
141 _job->set_progress_unknown ();
144 while (pass () == false) {
145 if (_job && _film->dcp_length()) {
146 _job->set_progress (float (_video_frame) / _film->dcp_length().get());
153 /** Run one pass. This may or may not generate any actual video / audio data;
154 * some decoders may require several passes to generate a single frame.
155 * @return true if we have finished processing all data; otherwise false.
160 if (!_ignore_length && _video_frame >= _film->dcp_length()) {
167 /** Called by subclasses to tell the world that some audio data is ready
168 * @param data Audio data, in Film::audio_sample_format.
169 * @param size Number of bytes of data.
172 Decoder::process_audio (uint8_t* data, int size)
174 /* Push into the delay line */
175 size = _delay_line->feed (data, size);
177 emit_audio (data, size);
181 Decoder::emit_audio (uint8_t* data, int size)
183 /* Deinterleave and convert to float */
185 assert ((size % (bytes_per_audio_sample() * audio_channels())) == 0);
187 int const total_samples = size / bytes_per_audio_sample();
188 int const frames = total_samples / _film->audio_channels();
189 shared_ptr<AudioBuffers> audio (new AudioBuffers (audio_channels(), frames));
191 switch (audio_sample_format()) {
192 case AV_SAMPLE_FMT_S16:
194 int16_t* p = (int16_t *) data;
197 for (int i = 0; i < total_samples; ++i) {
198 audio->data(channel)[sample] = float(*p++) / (1 << 15);
201 if (channel == _film->audio_channels()) {
209 case AV_SAMPLE_FMT_S32:
211 int32_t* p = (int32_t *) data;
214 for (int i = 0; i < total_samples; ++i) {
215 audio->data(channel)[sample] = float(*p++) / (1 << 31);
218 if (channel == _film->audio_channels()) {
225 case AV_SAMPLE_FMT_FLTP:
227 float* p = reinterpret_cast<float*> (data);
228 for (int i = 0; i < _film->audio_channels(); ++i) {
229 memcpy (audio->data(i), p, frames * sizeof(float));
239 /* Maybe apply gain */
240 if (_film->audio_gain() != 0) {
241 float const linear_gain = pow (10, _film->audio_gain() / 20);
242 for (int i = 0; i < _film->audio_channels(); ++i) {
243 for (int j = 0; j < frames; ++j) {
244 audio->data(i)[j] *= linear_gain;
249 /* Update the number of audio frames we've pushed to the encoder */
250 _audio_frames_processed += audio->frames ();
255 /** Called by subclasses to tell the world that some video data is ready.
256 * We do some post-processing / filtering then emit it for listeners.
257 * @param frame to decode; caller manages memory.
260 Decoder::process_video (AVFrame* frame)
267 /* Use Film::length here as our one may be wrong */
270 if (_opt->decode_video_frequency != 0) {
271 gap = _film->length().get() / _opt->decode_video_frequency;
274 if (_opt->decode_video_frequency != 0 && gap != 0 && (_video_frame % gap) != 0) {
279 shared_ptr<FilterGraph> graph;
281 list<shared_ptr<FilterGraph> >::iterator i = _filter_graphs.begin();
282 while (i != _filter_graphs.end() && !(*i)->can_process (Size (frame->width, frame->height), (AVPixelFormat) frame->format)) {
286 if (i == _filter_graphs.end ()) {
287 graph.reset (new FilterGraph (_film, this, _opt->apply_crop, Size (frame->width, frame->height), (AVPixelFormat) frame->format));
288 _filter_graphs.push_back (graph);
289 _film->log()->log (String::compose ("New graph for %1x%2, pixel format %3", frame->width, frame->height, frame->format));
294 list<shared_ptr<Image> > images = graph->process (frame);
296 for (list<shared_ptr<Image> >::iterator i = images.begin(); i != images.end(); ++i) {
297 if (_opt->black_after > 0 && _video_frame > _opt->black_after) {
301 shared_ptr<Subtitle> sub;
302 if (_timed_subtitle && _timed_subtitle->displayed_at (double (last_video_frame()) / _film->frames_per_second())) {
303 sub = _timed_subtitle->subtitle ();
306 TIMING ("Decoder emits %1", _video_frame);
307 Video ((*i), _video_frame, sub);
313 Decoder::process_subtitle (shared_ptr<TimedSubtitle> s)
317 if (_timed_subtitle && _opt->apply_crop) {
318 Position const p = _timed_subtitle->subtitle()->position ();
319 _timed_subtitle->subtitle()->set_position (Position (p.x - _film->crop().left, p.y - _film->crop().top));
325 Decoder::bytes_per_audio_sample () const
327 return av_get_bytes_per_sample (audio_sample_format ());