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.
51 * @param ignore_length Ignore the content's claimed length when computing progress.
53 Decoder::Decoder (boost::shared_ptr<Film> f, boost::shared_ptr<const Options> o, Job* j, bool minimal, bool ignore_length)
58 , _ignore_length (ignore_length)
59 , _video_frame_index (0)
62 , _audio_frames_processed (0)
64 if (_opt->decode_video_frequency != 0 && !_film->length()) {
65 throw DecodeError ("cannot do a partial decode if length is unknown");
74 /** Start off a decode processing run */
76 Decoder::process_begin ()
78 _delay_in_bytes = _film->audio_delay() * audio_sample_rate() * audio_channels() * bytes_per_audio_sample() / 1000;
80 _delay_line = new DelayLine (_delay_in_bytes);
82 _audio_frames_processed = 0;
85 /** Finish off a decode processing run */
87 Decoder::process_end ()
89 if (_delay_in_bytes < 0) {
90 uint8_t remainder[-_delay_in_bytes];
91 _delay_line->get_remaining (remainder);
92 _audio_frames_processed += _delay_in_bytes / (audio_channels() * bytes_per_audio_sample());
93 emit_audio (remainder, -_delay_in_bytes);
96 /* If we cut the decode off, the audio may be short; push some silence
97 in to get it to the right length.
100 int64_t const video_length_in_audio_frames = ((int64_t) video_frame_index() * audio_sample_rate() / frames_per_second());
101 int64_t const audio_short_by_frames = video_length_in_audio_frames - _audio_frames_processed;
104 String::compose ("DCP length is %1 (%2 audio frames); %3 frames of audio processed.",
106 video_length_in_audio_frames,
107 _audio_frames_processed)
110 if (audio_short_by_frames >= 0 && _opt->decode_audio) {
112 _film->log()->log (String::compose ("DCP length is %1; %2 frames of audio processed.", video_frame_index(), _audio_frames_processed));
113 _film->log()->log (String::compose ("Adding %1 frames of silence to the end.", audio_short_by_frames));
115 /* XXX: this is slightly questionable; does memset () give silence with all
119 int64_t bytes = audio_short_by_frames * _film->audio_channels() * bytes_per_audio_sample();
121 int64_t const silence_size = 16 * 1024 * _film->audio_channels() * bytes_per_audio_sample();
122 uint8_t silence[silence_size];
123 memset (silence, 0, silence_size);
126 int64_t const t = min (bytes, silence_size);
127 emit_audio (silence, t);
133 /** Start decoding */
139 if (_job && !_film->dcp_length()) {
140 _job->set_progress_unknown ();
143 while (pass () == false) {
144 if (_job && _film->dcp_length()) {
145 _job->set_progress (float (_video_frame_index) / _film->dcp_length().get());
152 /** Run one pass. This may or may not generate any actual video / audio data;
153 * some decoders may require several passes to generate a single frame.
154 * @return true if we have finished processing all data; otherwise false.
159 if (!_ignore_length && _video_frame_index >= _film->dcp_length()) {
166 /** Called by subclasses to tell the world that some audio data is ready
167 * @param data Audio data, in Film::audio_sample_format.
168 * @param size Number of bytes of data.
171 Decoder::process_audio (uint8_t* data, int size)
173 /* Push into the delay line */
174 size = _delay_line->feed (data, size);
176 emit_audio (data, size);
180 Decoder::emit_audio (uint8_t* data, int size)
186 assert (_film->audio_channels());
187 assert (bytes_per_audio_sample());
189 /* Deinterleave and convert to float */
191 assert ((size % (bytes_per_audio_sample() * audio_channels())) == 0);
193 int const total_samples = size / bytes_per_audio_sample();
194 int const frames = total_samples / _film->audio_channels();
195 shared_ptr<AudioBuffers> audio (new AudioBuffers (audio_channels(), frames));
197 switch (audio_sample_format()) {
198 case AV_SAMPLE_FMT_S16:
200 int16_t* p = (int16_t *) data;
203 for (int i = 0; i < total_samples; ++i) {
204 audio->data(channel)[sample] = float(*p++) / (1 << 15);
207 if (channel == _film->audio_channels()) {
215 case AV_SAMPLE_FMT_S32:
217 int32_t* p = (int32_t *) data;
220 for (int i = 0; i < total_samples; ++i) {
221 audio->data(channel)[sample] = float(*p++) / (1 << 31);
224 if (channel == _film->audio_channels()) {
231 case AV_SAMPLE_FMT_FLTP:
233 float* p = reinterpret_cast<float*> (data);
234 for (int i = 0; i < _film->audio_channels(); ++i) {
235 memcpy (audio->data(i), p, frames * sizeof(float));
245 /* Maybe apply gain */
246 if (_film->audio_gain() != 0) {
247 float const linear_gain = pow (10, _film->audio_gain() / 20);
248 for (int i = 0; i < _film->audio_channels(); ++i) {
249 for (int j = 0; j < frames; ++j) {
250 audio->data(i)[j] *= linear_gain;
255 /* Update the number of audio frames we've pushed to the encoder */
256 _audio_frames_processed += audio->frames ();
261 /** Called by subclasses to tell the world that some video data is ready.
262 * We do some post-processing / filtering then emit it for listeners.
263 * @param frame to decode; caller manages memory.
266 Decoder::process_video (AVFrame* frame)
269 ++_video_frame_index;
273 /* Use Film::length here as our one may be wrong */
276 if (_opt->decode_video_frequency != 0) {
277 gap = _film->length().get() / _opt->decode_video_frequency;
280 if (_opt->decode_video_frequency != 0 && gap != 0 && (_video_frame_index % gap) != 0) {
281 ++_video_frame_index;
285 shared_ptr<FilterGraph> graph;
287 list<shared_ptr<FilterGraph> >::iterator i = _filter_graphs.begin();
288 while (i != _filter_graphs.end() && !(*i)->can_process (Size (frame->width, frame->height), (AVPixelFormat) frame->format)) {
292 if (i == _filter_graphs.end ()) {
293 graph.reset (new FilterGraph (_film, this, _opt->apply_crop, Size (frame->width, frame->height), (AVPixelFormat) frame->format));
294 _filter_graphs.push_back (graph);
295 _film->log()->log (String::compose ("New graph for %1x%2, pixel format %3", frame->width, frame->height, frame->format));
300 list<shared_ptr<Image> > images = graph->process (frame);
302 for (list<shared_ptr<Image> >::iterator i = images.begin(); i != images.end(); ++i) {
303 shared_ptr<Subtitle> sub;
304 if (_timed_subtitle && _timed_subtitle->displayed_at (double (video_frame_index()) / _film->frames_per_second())) {
305 sub = _timed_subtitle->subtitle ();
308 TIMING ("Decoder emits %1", _video_frame_index);
309 Video (*i, _video_frame_index, sub);
310 ++_video_frame_index;
312 _last_subtitle = sub;
317 Decoder::repeat_last_video ()
320 _last_image.reset (new CompactImage (pixel_format(), native_size()));
321 _last_image->make_black ();
324 Video (_last_image, _video_frame_index, _last_subtitle);
325 ++_video_frame_index;
329 Decoder::process_subtitle (shared_ptr<TimedSubtitle> s)
333 if (_timed_subtitle && _opt->apply_crop) {
334 Position const p = _timed_subtitle->subtitle()->position ();
335 _timed_subtitle->subtitle()->set_position (Position (p.x - _film->crop().left, p.y - _film->crop().top));
341 Decoder::bytes_per_audio_sample () const
343 return av_get_bytes_per_sample (audio_sample_format ());