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;
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.
53 Decoder::Decoder (boost::shared_ptr<Film> f, boost::shared_ptr<const Options> o, Job* j, bool minimal)
58 , _video_frames_in (0)
59 , _video_frames_out (0)
60 , _audio_frames_in (0)
61 , _audio_frames_out (0)
63 , _delay_in_frames (0)
73 /** Start off a decode processing run. This should only be called once on
74 * a given Decoder object.
77 Decoder::process_begin ()
79 _delay_in_frames = _film->audio_delay() * audio_sample_rate() / 1000;
80 _delay_line = new DelayLine (audio_channels(), _delay_in_frames);
83 /** Finish off a decode processing run */
85 Decoder::process_end ()
87 if (_delay_in_frames < 0 && _opt->decode_audio && audio_channels()) {
88 shared_ptr<AudioBuffers> b (new AudioBuffers (audio_channels(), -_delay_in_frames));
93 if (_opt->decode_audio && audio_channels()) {
95 /* Ensure that our video and audio emissions are the same length */
97 int64_t audio_short_by_frames = video_frames_to_audio_frames (_video_frames_out) - _audio_frames_out;
100 String::compose ("Decoder has emitted %1 video frames (which equals %2 audio frames) and %3 audio frames",
102 video_frames_to_audio_frames (_video_frames_out),
106 if (audio_short_by_frames < 0) {
108 _film->log()->log (String::compose ("Emitted %1 too many audio frames", -audio_short_by_frames));
110 /* We have emitted more audio than video. Emit enough black video frames so that we reverse this */
111 int const black_video_frames = ceil (-audio_short_by_frames * frames_per_second() / audio_sample_rate());
113 _film->log()->log (String::compose ("Emitting %1 frames of black video", black_video_frames));
115 shared_ptr<Image> black (new CompactImage (pixel_format(), native_size()));
116 black->make_black ();
117 for (int i = 0; i < black_video_frames; ++i) {
118 emit_video (black, shared_ptr<Subtitle> ());
120 /* This is a bit of a hack, but you can sort-of justify it if you squint at it right.
121 It's important because the encoder will probably use this to name its output frame.
126 /* Now recompute our check value */
127 audio_short_by_frames = video_frames_to_audio_frames (_video_frames_out) - _audio_frames_out;
130 if (audio_short_by_frames > 0) {
131 _film->log()->log (String::compose ("Emitted %1 too few audio frames", audio_short_by_frames));
132 shared_ptr<AudioBuffers> b (new AudioBuffers (audio_channels(), audio_short_by_frames));
139 /** Start decoding */
145 if (_job && !_film->dcp_length()) {
146 _job->set_progress_unknown ();
149 while (pass () == false) {
150 if (_job && _film->dcp_length()) {
151 _job->set_progress (float (_video_frames_out) / _film->dcp_length().get());
158 /** Called by subclasses to tell the world that some audio data is ready
159 * @param data Audio data, in Film::audio_sample_format.
160 * @param size Number of bytes of data.
163 Decoder::process_audio (uint8_t* data, int size)
169 assert (_film->audio_channels());
170 assert (bytes_per_audio_sample());
172 /* Deinterleave and convert to float */
174 assert ((size % (bytes_per_audio_sample() * audio_channels())) == 0);
176 int const total_samples = size / bytes_per_audio_sample();
177 int const frames = total_samples / _film->audio_channels();
178 shared_ptr<AudioBuffers> audio (new AudioBuffers (audio_channels(), frames));
180 switch (audio_sample_format()) {
181 case AV_SAMPLE_FMT_S16:
183 int16_t* p = (int16_t *) data;
186 for (int i = 0; i < total_samples; ++i) {
187 audio->data(channel)[sample] = float(*p++) / (1 << 15);
190 if (channel == _film->audio_channels()) {
198 case AV_SAMPLE_FMT_S32:
200 int32_t* p = (int32_t *) data;
203 for (int i = 0; i < total_samples; ++i) {
204 audio->data(channel)[sample] = float(*p++) / (1 << 31);
207 if (channel == _film->audio_channels()) {
214 case AV_SAMPLE_FMT_FLTP:
216 float* p = reinterpret_cast<float*> (data);
217 for (int i = 0; i < _film->audio_channels(); ++i) {
218 memcpy (audio->data(i), p, frames * sizeof(float));
228 /* Maybe apply gain */
229 if (_film->audio_gain() != 0) {
230 float const linear_gain = pow (10, _film->audio_gain() / 20);
231 for (int i = 0; i < _film->audio_channels(); ++i) {
232 for (int j = 0; j < frames; ++j) {
233 audio->data(i)[j] *= linear_gain;
238 _delay_line->feed (audio);
240 int const in_frames = audio->frames ();
242 if (_opt->decode_range) {
243 /* Decode range in audio frames */
244 pair<int64_t, int64_t> required_range (
245 video_frames_to_audio_frames (_opt->decode_range.get().first),
246 video_frames_to_audio_frames (_opt->decode_range.get().second)
249 /* Range of this block of data */
250 pair<int64_t, int64_t> this_range (
252 _audio_frames_in + audio->frames()
255 if (this_range.second < required_range.first || required_range.second < this_range.first) {
256 /* No part of this audio is within the required range */
257 audio->set_frames (0);
258 } else if (required_range.first >= this_range.first && required_range.first < this_range.second) {
260 int64_t const shift = required_range.first - this_range.first;
261 audio->move (shift, 0, audio->frames() - shift);
262 audio->set_frames (audio->frames() - shift);
263 } else if (required_range.second >= this_range.first && required_range.second < this_range.second) {
265 audio->set_frames (required_range.second - this_range.first);
269 if (audio->frames()) {
273 _audio_frames_in += in_frames;
277 Decoder::emit_audio (shared_ptr<AudioBuffers> audio)
280 _audio_frames_out += audio->frames ();
283 /** Called by subclasses to tell the world that some video data is ready.
284 * We do some post-processing / filtering then emit it for listeners.
285 * @param frame to decode; caller manages memory.
288 Decoder::process_video (AVFrame* frame)
295 if (_opt->decode_video_skip != 0 && (_video_frames_in % _opt->decode_video_skip) != 0) {
300 if (_opt->decode_range) {
301 pair<SourceFrame, SourceFrame> r = _opt->decode_range.get();
302 if (_video_frames_in < r.first || _video_frames_in >= r.second) {
308 shared_ptr<FilterGraph> graph;
310 list<shared_ptr<FilterGraph> >::iterator i = _filter_graphs.begin();
311 while (i != _filter_graphs.end() && !(*i)->can_process (Size (frame->width, frame->height), (AVPixelFormat) frame->format)) {
315 if (i == _filter_graphs.end ()) {
316 graph.reset (new FilterGraph (_film, this, _opt->apply_crop, Size (frame->width, frame->height), (AVPixelFormat) frame->format));
317 _filter_graphs.push_back (graph);
318 _film->log()->log (String::compose ("New graph for %1x%2, pixel format %3", frame->width, frame->height, frame->format));
323 list<shared_ptr<Image> > images = graph->process (frame);
325 for (list<shared_ptr<Image> >::iterator i = images.begin(); i != images.end(); ++i) {
326 shared_ptr<Subtitle> sub;
327 if (_timed_subtitle && _timed_subtitle->displayed_at (double (video_frames_in()) / _film->frames_per_second())) {
328 sub = _timed_subtitle->subtitle ();
331 emit_video (*i, sub);
337 Decoder::repeat_last_video ()
340 _last_image.reset (new CompactImage (pixel_format(), native_size()));
341 _last_image->make_black ();
344 emit_video (_last_image, _last_subtitle);
349 Decoder::emit_video (shared_ptr<Image> image, shared_ptr<Subtitle> sub)
351 TIMING ("Decoder emits %1", _video_frames_out);
352 Video (image, _video_frames_in, sub);
355 _last_subtitle = sub;
359 Decoder::process_subtitle (shared_ptr<TimedSubtitle> s)
363 if (_timed_subtitle && _opt->apply_crop) {
364 Position const p = _timed_subtitle->subtitle()->position ();
365 _timed_subtitle->subtitle()->set_position (Position (p.x - _film->crop().left, p.y - _film->crop().top));
371 Decoder::bytes_per_audio_sample () const
373 return av_get_bytes_per_sample (audio_sample_format ());
377 Decoder::video_frames_to_audio_frames (SourceFrame v) const
379 return ((int64_t) v * audio_sample_rate() / frames_per_second());