Tweak naming.
[dcpomatic.git] / src / lib / decoder.cc
index c9235f8e1d595ae0d60b537a88f696222c3b2980..6568ad08915e108e194a966957c7d689ad10da71 100644 (file)
 #include <iostream>
 #include <stdint.h>
 #include <boost/lexical_cast.hpp>
-extern "C" {
-#include <libavfilter/avfiltergraph.h>
-#include <libavfilter/buffersrc.h>
-#if (LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR >= 53 && LIBAVFILTER_VERSION_MINOR <= 77) || LIBAVFILTER_VERSION_MAJOR == 3
-#include <libavfilter/avcodec.h>
-#include <libavfilter/buffersink.h>
-#elif LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR == 15
-#include <libavfilter/vsrc_buffer.h>
-#endif
-#include <libavformat/avio.h>
-}
 #include "film.h"
 #include "format.h"
 #include "job.h"
-#include "film_state.h"
 #include "options.h"
 #include "exceptions.h"
 #include "image.h"
 #include "util.h"
 #include "log.h"
 #include "decoder.h"
-#include "filter.h"
 #include "delay_line.h"
-#include "ffmpeg_compatibility.h"
 #include "subtitle.h"
+#include "filter_graph.h"
 
-using namespace std;
-using namespace boost;
+using std::string;
+using std::stringstream;
+using std::min;
+using std::list;
+using boost::shared_ptr;
 
-/** @param s FilmState of the Film.
+/** @param f Film.
  *  @param o Options.
  *  @param j Job that we are running within, or 0
- *  @param l Log to use.
  *  @param minimal true to do the bare minimum of work; just run through the content.  Useful for acquiring
  *  accurate frame counts as quickly as possible.  This generates no video or audio output.
  *  @param ignore_length Ignore the content's claimed length when computing progress.
  */
-Decoder::Decoder (boost::shared_ptr<const FilmState> s, boost::shared_ptr<const Options> o, Job* j, Log* l, bool minimal, bool ignore_length)
-       : _fs (s)
+Decoder::Decoder (boost::shared_ptr<Film> f, boost::shared_ptr<const Options> o, Job* j, bool minimal, bool ignore_length)
+       : _film (f)
        , _opt (o)
        , _job (j)
-       , _log (l)
        , _minimal (minimal)
        , _ignore_length (ignore_length)
        , _video_frame (0)
-       , _buffer_src_context (0)
-       , _buffer_sink_context (0)
-       , _have_setup_video_filters (false)
        , _delay_line (0)
        , _delay_in_bytes (0)
        , _audio_frames_processed (0)
 {
-       if (_opt->decode_video_frequency != 0 && _fs->length() == 0) {
-               throw DecodeError ("cannot do a partial decode if length == 0");
+       if (_opt->decode_video_frequency != 0 && !_film->length()) {
+               throw DecodeError ("cannot do a partial decode if length is unknown");
        }
 }
 
@@ -90,12 +75,10 @@ Decoder::~Decoder ()
 void
 Decoder::process_begin ()
 {
-       _delay_in_bytes = _fs->total_audio_delay() * _fs->audio_sample_rate() * _fs->audio_channels() * bytes_per_audio_sample() / 1000;
+       _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);
 
-       _log->log (String::compose ("Decoding audio with total delay of %1", _fs->total_audio_delay()));
-
        _audio_frames_processed = 0;
 }
 
@@ -106,7 +89,7 @@ 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 / (_fs->audio_channels() * bytes_per_audio_sample());
+               _audio_frames_processed += _delay_in_bytes / (audio_channels() * bytes_per_audio_sample());
                emit_audio (remainder, -_delay_in_bytes);
        }
 
@@ -114,24 +97,28 @@ Decoder::process_end ()
           in to get it to the right length.
        */
 
-       int64_t const video_length_in_audio_frames = ((int64_t) _fs->dcp_length() * _fs->target_sample_rate() / _fs->frames_per_second());
+       int64_t const video_length_in_audio_frames = ((int64_t) video_frame() * audio_sample_rate() / frames_per_second());
        int64_t const audio_short_by_frames = video_length_in_audio_frames - _audio_frames_processed;
 
-       _log->log (
+       _film->log()->log (
                String::compose ("DCP length is %1 (%2 audio frames); %3 frames of audio processed.",
-                                _fs->dcp_length(),
+                                video_frame(),
                                 video_length_in_audio_frames,
                                 _audio_frames_processed)
                );
        
        if (audio_short_by_frames >= 0 && _opt->decode_audio) {
 
-               _log->log (String::compose ("DCP length is %1; %2 frames of audio processed.", _fs->dcp_length(), _audio_frames_processed));
-               _log->log (String::compose ("Adding %1 frames of silence to the end.", audio_short_by_frames));
+               _film->log()->log (String::compose ("DCP length is %1; %2 frames of audio processed.", video_frame(), _audio_frames_processed));
+               _film->log()->log (String::compose ("Adding %1 frames of silence to the end.", audio_short_by_frames));
 
-               int64_t bytes = audio_short_by_frames * _fs->audio_channels() * bytes_per_audio_sample();
+               /* 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 * _fs->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);
                
@@ -149,13 +136,13 @@ Decoder::go ()
 {
        process_begin ();
 
-       if (_job && _ignore_length) {
+       if (_job && !_film->dcp_length()) {
                _job->set_progress_unknown ();
        }
 
        while (pass () == false) {
-               if (_job && !_ignore_length) {
-                       _job->set_progress (float (_video_frame) / _fs->dcp_length());
+               if (_job && _film->dcp_length()) {
+                       _job->set_progress (float (_video_frame) / _film->dcp_length().get());
                }
        }
 
@@ -169,12 +156,7 @@ Decoder::go ()
 bool
 Decoder::pass ()
 {
-       if (!_have_setup_video_filters) {
-               setup_video_filters ();
-               _have_setup_video_filters = true;
-       }
-       
-       if (!_ignore_length && _video_frame >= _fs->dcp_length()) {
+       if (!_ignore_length && _video_frame >= _film->dcp_length()) {
                return true;
        }
 
@@ -182,7 +164,7 @@ Decoder::pass ()
 }
 
 /** Called by subclasses to tell the world that some audio data is ready
- *  @param data Audio data, in FilmState::audio_sample_format.
+ *  @param data Audio data, in Film::audio_sample_format.
  *  @param size Number of bytes of data.
  */
 void
@@ -199,11 +181,11 @@ Decoder::emit_audio (uint8_t* data, int size)
 {
        /* Deinterleave and convert to float */
 
-       assert ((size % (bytes_per_audio_sample() * _fs->audio_channels())) == 0);
+       assert ((size % (bytes_per_audio_sample() * audio_channels())) == 0);
 
        int const total_samples = size / bytes_per_audio_sample();
-       int const frames = total_samples / _fs->audio_channels();
-       shared_ptr<AudioBuffers> audio (new AudioBuffers (_fs->audio_channels(), frames));
+       int const frames = total_samples / _film->audio_channels();
+       shared_ptr<AudioBuffers> audio (new AudioBuffers (audio_channels(), frames));
 
        switch (audio_sample_format()) {
        case AV_SAMPLE_FMT_S16:
@@ -215,7 +197,7 @@ Decoder::emit_audio (uint8_t* data, int size)
                        audio->data(channel)[sample] = float(*p++) / (1 << 15);
 
                        ++channel;
-                       if (channel == _fs->audio_channels()) {
+                       if (channel == _film->audio_channels()) {
                                channel = 0;
                                ++sample;
                        }
@@ -232,7 +214,7 @@ Decoder::emit_audio (uint8_t* data, int size)
                        audio->data(channel)[sample] = float(*p++) / (1 << 31);
 
                        ++channel;
-                       if (channel == _fs->audio_channels()) {
+                       if (channel == _film->audio_channels()) {
                                channel = 0;
                                ++sample;
                        }
@@ -242,7 +224,7 @@ Decoder::emit_audio (uint8_t* data, int size)
        case AV_SAMPLE_FMT_FLTP:
        {
                float* p = reinterpret_cast<float*> (data);
-               for (int i = 0; i < _fs->audio_channels(); ++i) {
+               for (int i = 0; i < _film->audio_channels(); ++i) {
                        memcpy (audio->data(i), p, frames * sizeof(float));
                        p += frames;
                }
@@ -254,9 +236,9 @@ Decoder::emit_audio (uint8_t* data, int size)
        }
 
        /* Maybe apply gain */
-       if (_fs->audio_gain() != 0) {
-               float const linear_gain = pow (10, _fs->audio_gain() / 20);
-               for (int i = 0; i < _fs->audio_channels(); ++i) {
+       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;
                        }
@@ -264,7 +246,7 @@ Decoder::emit_audio (uint8_t* data, int size)
        }
 
        /* Update the number of audio frames we've pushed to the encoder */
-       _audio_frames_processed += frames;
+       _audio_frames_processed += audio->frames ();
 
        Audio (audio);
 }
@@ -281,11 +263,11 @@ Decoder::process_video (AVFrame* frame)
                return;
        }
 
-       /* Use FilmState::length here as our one may be wrong */
+       /* Use Film::length here as our one may be wrong */
 
        int gap = 0;
        if (_opt->decode_video_frequency != 0) {
-               gap = _fs->length() / _opt->decode_video_frequency;
+               gap = _film->length().get() / _opt->decode_video_frequency;
        }
 
        if (_opt->decode_video_frequency != 0 && gap != 0 && (_video_frame % gap) != 0) {
@@ -293,167 +275,37 @@ Decoder::process_video (AVFrame* frame)
                return;
        }
 
-#if LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR >= 53 && LIBAVFILTER_VERSION_MINOR <= 61
+       shared_ptr<FilterGraph> graph;
 
-       if (av_vsrc_buffer_add_frame (_buffer_src_context, frame, 0) < 0) {
-               throw DecodeError ("could not push buffer into filter chain.");
+       list<shared_ptr<FilterGraph> >::iterator i = _filter_graphs.begin();
+       while (i != _filter_graphs.end() && !(*i)->can_process (Size (frame->width, frame->height), (AVPixelFormat) frame->format)) {
+               ++i;
        }
 
-#elif LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR == 15
-
-       AVRational par;
-       par.num = sample_aspect_ratio_numerator ();
-       par.den = sample_aspect_ratio_denominator ();
-
-       if (av_vsrc_buffer_add_frame (_buffer_src_context, frame, 0, par) < 0) {
-               throw DecodeError ("could not push buffer into filter chain.");
-       }
-
-#else
-
-       if (av_buffersrc_write_frame (_buffer_src_context, frame) < 0) {
-               throw DecodeError ("could not push buffer into filter chain.");
+       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;
        }
 
-#endif 
-       
-#if LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR >= 15 && LIBAVFILTER_VERSION_MINOR <= 61       
-       while (avfilter_poll_frame (_buffer_sink_context->inputs[0])) {
-#else
-       while (av_buffersink_read (_buffer_sink_context, 0)) {
-#endif         
+       list<shared_ptr<Image> > images = graph->process (frame);
 
-#if LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR >= 15
-               
-               int r = avfilter_request_frame (_buffer_sink_context->inputs[0]);
-               if (r < 0) {
-                       throw DecodeError ("could not request filtered frame");
+       for (list<shared_ptr<Image> >::iterator i = images.begin(); i != images.end(); ++i) {
+               if (_opt->black_after > 0 && _video_frame > _opt->black_after) {
+                       (*i)->make_black ();
                }
                
-               AVFilterBufferRef* filter_buffer = _buffer_sink_context->inputs[0]->cur_buf;
-               
-#else
-
-               AVFilterBufferRef* filter_buffer;
-               if (av_buffersink_get_buffer_ref (_buffer_sink_context, &filter_buffer, 0) < 0) {
-                       filter_buffer = 0;
+               shared_ptr<Subtitle> sub;
+               if (_timed_subtitle && _timed_subtitle->displayed_at (double (video_frame()) / _film->frames_per_second())) {
+                       sub = _timed_subtitle->subtitle ();
                }
-
-#endif         
                
-               if (filter_buffer) {
-                       /* This takes ownership of filter_buffer */
-                       shared_ptr<Image> image (new FilterBufferImage ((PixelFormat) frame->format, filter_buffer));
-
-                       if (_opt->black_after > 0 && _video_frame > _opt->black_after) {
-                               image->make_black ();
-                       }
-
-                       shared_ptr<Subtitle> sub;
-                       if (_timed_subtitle && _timed_subtitle->displayed_at (double (last_video_frame()) / rint (_fs->frames_per_second()))) {
-                               sub = _timed_subtitle->subtitle ();
-                       }
-
-                       TIMING ("Decoder emits %1", _video_frame);
-                       Video (image, _video_frame, sub);
-                       ++_video_frame;
-               }
-       }
-}
-
-
-/** Set up a video filtering chain to include cropping and any filters that are specified
- *  by the Film.
- */
-void
-Decoder::setup_video_filters ()
-{
-       stringstream fs;
-       Size size_after_crop;
-       
-       if (_opt->apply_crop) {
-               size_after_crop = _fs->cropped_size (native_size ());
-               fs << crop_string (Position (_fs->crop().left, _fs->crop().top), size_after_crop);
-       } else {
-               size_after_crop = native_size ();
-               fs << crop_string (Position (0, 0), size_after_crop);
-       }
-
-       string filters = Filter::ffmpeg_strings (_fs->filters()).first;
-       if (!filters.empty ()) {
-               filters += ",";
-       }
-
-       filters += fs.str ();
-
-       avfilter_register_all ();
-       
-       AVFilterGraph* graph = avfilter_graph_alloc();
-       if (graph == 0) {
-               throw DecodeError ("Could not create filter graph.");
-       }
-
-       AVFilter* buffer_src = avfilter_get_by_name("buffer");
-       if (buffer_src == 0) {
-               throw DecodeError ("Could not find buffer src filter");
-       }
-
-       AVFilter* buffer_sink = get_sink ();
-
-       stringstream a;
-       a << native_size().width << ":"
-         << native_size().height << ":"
-         << pixel_format() << ":"
-         << time_base_numerator() << ":"
-         << time_base_denominator() << ":"
-         << sample_aspect_ratio_numerator() << ":"
-         << sample_aspect_ratio_denominator();
-
-       int r;
-
-       if ((r = avfilter_graph_create_filter (&_buffer_src_context, buffer_src, "in", a.str().c_str(), 0, graph)) < 0) {
-               throw DecodeError ("could not create buffer source");
-       }
-
-       AVBufferSinkParams* sink_params = av_buffersink_params_alloc ();
-       PixelFormat* pixel_fmts = new PixelFormat[2];
-       pixel_fmts[0] = pixel_format ();
-       pixel_fmts[1] = PIX_FMT_NONE;
-       sink_params->pixel_fmts = pixel_fmts;
-       
-       if (avfilter_graph_create_filter (&_buffer_sink_context, buffer_sink, "out", 0, sink_params, graph) < 0) {
-               throw DecodeError ("could not create buffer sink.");
-       }
-
-       AVFilterInOut* outputs = avfilter_inout_alloc ();
-       outputs->name = av_strdup("in");
-       outputs->filter_ctx = _buffer_src_context;
-       outputs->pad_idx = 0;
-       outputs->next = 0;
-
-       AVFilterInOut* inputs = avfilter_inout_alloc ();
-       inputs->name = av_strdup("out");
-       inputs->filter_ctx = _buffer_sink_context;
-       inputs->pad_idx = 0;
-       inputs->next = 0;
-
-       _log->log ("Using filter chain `" + filters + "'");
-
-#if LIBAVFILTER_VERSION_MAJOR == 2 && LIBAVFILTER_VERSION_MINOR == 15
-       if (avfilter_graph_parse (graph, filters.c_str(), inputs, outputs, 0) < 0) {
-               throw DecodeError ("could not set up filter graph.");
-       }
-#else  
-       if (avfilter_graph_parse (graph, filters.c_str(), &inputs, &outputs, 0) < 0) {
-               throw DecodeError ("could not set up filter graph.");
-       }
-#endif 
-       
-       if (avfilter_graph_config (graph, 0) < 0) {
-               throw DecodeError ("could not configure filter graph.");
+               TIMING ("Decoder emits %1", _video_frame);
+               Video ((*i), _video_frame, sub);
+               ++_video_frame;
        }
-
-       /* XXX: leaking `inputs' / `outputs' ? */
 }
 
 void
@@ -461,9 +313,9 @@ Decoder::process_subtitle (shared_ptr<TimedSubtitle> s)
 {
        _timed_subtitle = s;
        
-       if (_opt->apply_crop) {
+       if (_timed_subtitle && _opt->apply_crop) {
                Position const p = _timed_subtitle->subtitle()->position ();
-               _timed_subtitle->subtitle()->set_position (Position (p.x - _fs->crop().left, p.y - _fs->crop().top));
+               _timed_subtitle->subtitle()->set_position (Position (p.x - _film->crop().left, p.y - _film->crop().top));
        }
 }