X-Git-Url: https://git.carlh.net/gitweb/?a=blobdiff_plain;f=src%2Flib%2Fdecoder.cc;h=5d191512874916c74a82eed7ce049b9e2e33803d;hb=194c2ac76b4bb4fc9fae5bc21d4e221ed194d444;hp=34c04b260502ab75c424dc9635c2aee3e1515956;hpb=ac712b45bad948fc8f38cf4fa48a529f2024c7f5;p=dcpomatic.git diff --git a/src/lib/decoder.cc b/src/lib/decoder.cc index 34c04b260..5d1915128 100644 --- a/src/lib/decoder.cc +++ b/src/lib/decoder.cc @@ -1,321 +1,101 @@ /* - Copyright (C) 2012 Carl Hetherington + Copyright (C) 2012-2021 Carl Hetherington - This program is free software; you can redistribute it and/or modify + This file is part of DCP-o-matic. + + DCP-o-matic is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. - This program is distributed in the hope that it will be useful, + DCP-o-matic is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License - along with this program; if not, write to the Free Software - Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + along with DCP-o-matic. If not, see . */ -/** @file src/decoder.cc - * @brief Parent class for decoders of content. - */ -#include -#include -#include -#include "film.h" -#include "format.h" -#include "job.h" -#include "options.h" -#include "exceptions.h" -#include "image.h" -#include "util.h" -#include "log.h" #include "decoder.h" -#include "delay_line.h" -#include "subtitle.h" -#include "filter_graph.h" +#include "video_decoder.h" +#include "audio_decoder.h" +#include "text_decoder.h" +#include +#include -using std::string; -using std::stringstream; -using std::min; -using std::list; -using boost::shared_ptr; -/** @param f Film. - * @param o Options. - * @param j Job that we are running within, or 0 - */ -Decoder::Decoder (boost::shared_ptr f, boost::shared_ptr o, Job* j) - : _film (f) - , _opt (o) - , _job (j) - , _video_frame_index (0) - , _delay_line (0) - , _delay_in_bytes (0) - , _audio_frames_processed (0) -{ - -} +using std::cout; +using boost::optional; +using std::shared_ptr; +using std::weak_ptr; +using namespace dcpomatic; -Decoder::~Decoder () -{ - delete _delay_line; -} -/** Start off a decode processing run */ -void -Decoder::process_begin () +Decoder::Decoder (weak_ptr film) + : WeakConstFilm (film) { - _delay_in_bytes = _film->audio_delay() * audio_sample_rate() * audio_channels() * bytes_per_audio_sample() / 1000; - delete _delay_line; - _delay_line = new DelayLine (_delay_in_bytes); - _audio_frames_processed = 0; } -/** Finish off a decode processing run */ -void -Decoder::process_end () -{ - if (_delay_in_bytes < 0) { - uint8_t remainder[-_delay_in_bytes]; - _delay_line->get_remaining (remainder); - _audio_frames_processed += _delay_in_bytes / (audio_channels() * bytes_per_audio_sample()); - emit_audio (remainder, -_delay_in_bytes); - } - - /* If we cut the decode off, the audio may be short; push some silence - in to get it to the right length. - */ - - int64_t const video_length_in_audio_frames = ((int64_t) video_frame_index() * audio_sample_rate() / frames_per_second()); - int64_t const audio_short_by_frames = video_length_in_audio_frames - _audio_frames_processed; - - _film->log()->log ( - String::compose ("Source length is %1 (%2 audio frames); %3 frames of audio processed.", - video_frame_index(), - video_length_in_audio_frames, - _audio_frames_processed) - ); - - if (audio_short_by_frames >= 0 && _opt->decode_audio) { - _film->log()->log (String::compose ("Source length is %1; %2 frames of audio processed.", video_frame_index(), _audio_frames_processed)); - _film->log()->log (String::compose ("Adding %1 frames of silence to the end.", audio_short_by_frames)); - - /* XXX: this is slightly questionable; does memset () give silence with all - sample formats? - */ - - int64_t bytes = audio_short_by_frames * _film->audio_channels() * bytes_per_audio_sample(); - - int64_t const silence_size = 16 * 1024 * _film->audio_channels() * bytes_per_audio_sample(); - uint8_t silence[silence_size]; - memset (silence, 0, silence_size); - - while (bytes) { - int64_t const t = min (bytes, silence_size); - emit_audio (silence, t); - bytes -= t; - } - } -} - -/** Start decoding */ -void -Decoder::go () +/** @return Earliest time of content that the next pass() will emit */ +ContentTime +Decoder::position () const { - process_begin (); + optional pos; + auto f = film(); - if (_job && !_film->dcp_length()) { - _job->set_progress_unknown (); + if (video && !video->ignore() && (!pos || video->position(f).get_value_or(ContentTime()) < *pos)) { + pos = video->position(f); } - while (pass () == false) { - if (_job && _film->dcp_length()) { - SourceFrame const p = _video_frame_index - _film->dcp_trim_start(); - _job->set_progress (float (p) / _film->dcp_length().get()); - } - } - - process_end (); -} - -/** Called by subclasses to tell the world that some audio data is ready - * @param data Audio data, in Film::audio_sample_format. - * @param size Number of bytes of data. - */ -void -Decoder::process_audio (uint8_t* data, int size) -{ - /* Push into the delay line */ - size = _delay_line->feed (data, size); - - emit_audio (data, size); -} - -void -Decoder::emit_audio (uint8_t* data, int size) -{ - if (size == 0) { - return; + if (audio && !audio->ignore() && (!pos || audio->position(f).get_value_or(ContentTime()) < *pos)) { + pos = audio->position(f); } - - assert (_film->audio_channels()); - assert (bytes_per_audio_sample()); - - /* Deinterleave and convert to float */ - - assert ((size % (bytes_per_audio_sample() * audio_channels())) == 0); - - int const total_samples = size / bytes_per_audio_sample(); - int const frames = total_samples / _film->audio_channels(); - shared_ptr audio (new AudioBuffers (audio_channels(), frames)); - - switch (audio_sample_format()) { - case AV_SAMPLE_FMT_S16: - { - int16_t* p = (int16_t *) data; - int sample = 0; - int channel = 0; - for (int i = 0; i < total_samples; ++i) { - audio->data(channel)[sample] = float(*p++) / (1 << 15); - ++channel; - if (channel == _film->audio_channels()) { - channel = 0; - ++sample; - } - } - } - break; - - case AV_SAMPLE_FMT_S32: - { - int32_t* p = (int32_t *) data; - int sample = 0; - int channel = 0; - for (int i = 0; i < total_samples; ++i) { - audio->data(channel)[sample] = float(*p++) / (1 << 31); - - ++channel; - if (channel == _film->audio_channels()) { - channel = 0; - ++sample; - } - } - } - - case AV_SAMPLE_FMT_FLTP: - { - float* p = reinterpret_cast (data); - for (int i = 0; i < _film->audio_channels(); ++i) { - memcpy (audio->data(i), p, frames * sizeof(float)); - p += frames; - } - } - break; - - default: - assert (false); - } - - /* Maybe apply gain */ - if (_film->audio_gain() != 0) { - float const linear_gain = pow (10, _film->audio_gain() / 20); - for (int i = 0; i < _film->audio_channels(); ++i) { - for (int j = 0; j < frames; ++j) { - audio->data(i)[j] *= linear_gain; + /* Only decide position based on subtitle sources if there is nothing else + to go on. Otherwise we can have problems with muxed sources which have + (for example) video, audio and a subtitle. If the subtitle data runs out + before the video/audio the position() call will return the position of the + end of the subs. This causes this file to be pass()ed in favour of others, + which can cause bugs like #1581. + */ + if (!pos) { + for (auto i: text) { + if (!i->ignore() && (!pos || i->position(f) < *pos)) { + pos = i->position(f); } } } - /* Update the number of audio frames we've pushed to the encoder */ - _audio_frames_processed += audio->frames (); - - Audio (audio); + return pos.get_value_or(ContentTime()); } -/** Called by subclasses to tell the world that some video data is ready. - * We do some post-processing / filtering then emit it for listeners. - * @param frame to decode; caller manages memory. - */ + void -Decoder::process_video (AVFrame* frame) +Decoder::seek (ContentTime, bool) { - assert (_film->length()); - - /* Use Film::length here as our one may be wrong */ - - if (_opt->decode_video_skip != 0 && (_video_frame_index % _opt->decode_video_skip) != 0) { - ++_video_frame_index; - return; + if (video) { + video->seek (); } - - if (_film->dcp_trim_start() > _video_frame_index || (_film->length().get() - _film->dcp_trim_end()) < _video_frame_index) { - ++_video_frame_index; - return; + if (audio) { + audio->seek (); } - - shared_ptr graph; - - list >::iterator i = _filter_graphs.begin(); - while (i != _filter_graphs.end() && !(*i)->can_process (Size (frame->width, frame->height), (AVPixelFormat) frame->format)) { - ++i; - } - - if (i == _filter_graphs.end ()) { - graph.reset (new FilterGraph (_film, this, _opt->apply_crop, Size (frame->width, frame->height), (AVPixelFormat) frame->format)); - _filter_graphs.push_back (graph); - _film->log()->log (String::compose ("New graph for %1x%2, pixel format %3", frame->width, frame->height, frame->format)); - } else { - graph = *i; - } - - list > images = graph->process (frame); - - for (list >::iterator i = images.begin(); i != images.end(); ++i) { - shared_ptr sub; - if (_timed_subtitle && _timed_subtitle->displayed_at (double (video_frame_index()) / _film->frames_per_second())) { - sub = _timed_subtitle->subtitle (); - } - - TIMING ("Decoder emits %1", _video_frame_index); - Video (*i, _video_frame_index, sub); - ++_video_frame_index; - _last_image = *i; - _last_subtitle = sub; + for (auto i: text) { + i->seek (); } } -void -Decoder::repeat_last_video () -{ - if (!_last_image) { - _last_image.reset (new CompactImage (pixel_format(), native_size())); - _last_image->make_black (); - } - - Video (_last_image, _video_frame_index, _last_subtitle); - ++_video_frame_index; -} -void -Decoder::process_subtitle (shared_ptr s) +shared_ptr +Decoder::only_text () const { - _timed_subtitle = s; - - if (_timed_subtitle && _opt->apply_crop) { - Position const p = _timed_subtitle->subtitle()->position (); - _timed_subtitle->subtitle()->set_position (Position (p.x - _film->crop().left, p.y - _film->crop().top)); + DCPOMATIC_ASSERT (text.size() < 2); + if (text.empty()) { + return {}; } -} - - -int -Decoder::bytes_per_audio_sample () const -{ - return av_get_bytes_per_sample (audio_sample_format ()); + return text.front(); }