Move audio sample format stuff out of Decoder base class.
[dcpomatic.git] / src / lib / decoder.cc
index 673d571c40078d1b7adc2e785f7d5f525bc29968..01c4edf604b9864e3f011d7a45afee81ffd47217 100644 (file)
@@ -40,6 +40,7 @@
 using std::string;
 using std::stringstream;
 using std::min;
+using std::pair;
 using std::list;
 using boost::shared_ptr;
 
@@ -49,17 +50,14 @@ using boost::shared_ptr;
  *  @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.
  */
-Decoder::Decoder (boost::shared_ptr<Film> f, boost::shared_ptr<const Options> o, Job* j, bool minimal)
+Decoder::Decoder (boost::shared_ptr<Film> f, boost::shared_ptr<const Options> o, Job* j)
        : _film (f)
        , _opt (o)
        , _job (j)
-       , _minimal (minimal)
-       , _video_frames_in (0)
-       , _video_frames_out (0)
-       , _audio_frames_in (0)
-       , _audio_frames_out (0)
+       , _video_frame (0)
+       , _audio_frame (0)
        , _delay_line (0)
-       , _delay_in_bytes (0)
+       , _delay_in_frames (0)
 {
        
 }
@@ -75,33 +73,33 @@ Decoder::~Decoder ()
 void
 Decoder::process_begin ()
 {
-       _delay_in_bytes = _film->audio_delay() * audio_sample_rate() * audio_channels() * bytes_per_audio_sample() / 1000;
-       _delay_line = new DelayLine (_delay_in_bytes);
+       _delay_in_frames = _film->audio_delay() * audio_sample_rate() / 1000;
+       _delay_line = new DelayLine (audio_channels(), _delay_in_frames);
 }
 
 /** Finish off a decode processing run */
 void
 Decoder::process_end ()
 {
-       if (_delay_in_bytes < 0) {
-               /* Empty the delay line */
-               uint8_t remainder[-_delay_in_bytes];
-               _delay_line->get_remaining (remainder);
-               emit_audio (remainder, -_delay_in_bytes);
+       if (_delay_in_frames < 0 && _opt->decode_audio && audio_channels()) {
+               shared_ptr<AudioBuffers> b (new AudioBuffers (audio_channels(), -_delay_in_frames));
+               b->make_silent ();
+               emit_audio (b);
        }
 
-       if (_opt->decode_audio) {
+       if (_opt->decode_audio && audio_channels()) {
 
                /* Ensure that our video and audio emissions are the same length */
 
-               int64_t video_frames_out_in_audio_frames = ((int64_t) _video_frames_out * audio_sample_rate() / frames_per_second());
-               int64_t audio_short_by_frames = video_frames_out_in_audio_frames - _audio_frames_out;
+               int64_t audio_short_by_frames = video_frames_to_audio_frames (_video_frame, audio_sample_rate(), frames_per_second()) - _audio_frame;
 
                _film->log()->log (
-                       String::compose ("Decoder has emitted %1 video frames (which equals %2 audio frames) and %3 audio frames",
-                                        _video_frames_out,
-                                        video_frames_out_in_audio_frames,
-                                        _audio_frames_out)
+                       String::compose (
+                               "Decoder has emitted %1 video frames (which equals %2 audio frames) and %3 audio frames",
+                               _video_frame,
+                               video_frames_to_audio_frames (_video_frame, audio_sample_rate(), frames_per_second()),
+                               _audio_frame
+                               )
                        );
 
                if (audio_short_by_frames < 0) {
@@ -119,30 +117,15 @@ Decoder::process_end ()
                                emit_video (black, shared_ptr<Subtitle> ());
                        }
 
-                       /* Now recompute our check values */
-                       video_frames_out_in_audio_frames = ((int64_t) _video_frames_out * audio_sample_rate() / frames_per_second());
-                       audio_short_by_frames = video_frames_out_in_audio_frames - _audio_frames_out;
+                       /* Now recompute our check value */
+                       audio_short_by_frames = video_frames_to_audio_frames (_video_frame, audio_sample_rate(), frames_per_second()) - _audio_frame;
                }
        
                if (audio_short_by_frames > 0) {
-
                        _film->log()->log (String::compose ("Emitted %1 too few audio frames", 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;
-                       }
+                       shared_ptr<AudioBuffers> b (new AudioBuffers (audio_channels(), audio_short_by_frames));
+                       b->make_silent ();
+                       emit_audio (b);
                }
        }
 }
@@ -159,7 +142,7 @@ Decoder::go ()
 
        while (pass () == false) {
                if (_job && _film->dcp_length()) {
-                       _job->set_progress (float (_video_frames_out) / _film->dcp_length().get());
+                       _job->set_progress (float (_video_frame) / _film->length().get());
                }
        }
 
@@ -171,94 +154,20 @@ Decoder::go ()
  *  @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)
+Decoder::process_audio (shared_ptr<AudioBuffers> audio)
 {
-       if (size == 0) {
-               return;
-       }
-       
-       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<AudioBuffers> 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<float*> (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) {
+               for (int i = 0; i < audio->channels(); ++i) {
+                       for (int j = 0; j < audio->frames(); ++j) {
                                audio->data(i)[j] *= linear_gain;
                        }
                }
        }
 
-       /* Update the number of audio frames we've pushed to the encoder */
-       _audio_frames_out += audio->frames ();
-
-       Audio (audio);
+       _delay_line->feed (audio);
+       emit_audio (audio);
 }
 
 /** Called by subclasses to tell the world that some video data is ready.
@@ -268,25 +177,6 @@ Decoder::emit_audio (uint8_t* data, int size)
 void
 Decoder::process_video (AVFrame* frame)
 {
-       assert (_film->length());
-
-       if (_minimal) {
-               ++_video_frames_in;
-               return;
-       }
-
-       /* Use Film::length here as our one may be wrong */
-
-       if (_opt->decode_video_skip != 0 && (_video_frames_in % _opt->decode_video_skip) != 0) {
-               ++_video_frames_in;
-               return;
-       }
-
-       if (_film->dcp_trim_start() > _video_frames_in || (_film->length().get() + _film->dcp_trim_start()) < _video_frames_in) {
-               ++_video_frames_in;
-               return;
-       }
-
        shared_ptr<FilterGraph> graph;
 
        list<shared_ptr<FilterGraph> >::iterator i = _filter_graphs.begin();
@@ -306,7 +196,7 @@ Decoder::process_video (AVFrame* frame)
 
        for (list<shared_ptr<Image> >::iterator i = images.begin(); i != images.end(); ++i) {
                shared_ptr<Subtitle> sub;
-               if (_timed_subtitle && _timed_subtitle->displayed_at (double (video_frames_in()) / _film->frames_per_second())) {
+               if (_timed_subtitle && _timed_subtitle->displayed_at (double (video_frame()) / _film->frames_per_second())) {
                        sub = _timed_subtitle->subtitle ();
                }
 
@@ -328,13 +218,21 @@ Decoder::repeat_last_video ()
 void
 Decoder::emit_video (shared_ptr<Image> image, shared_ptr<Subtitle> sub)
 {
-       TIMING ("Decoder emits %1", _video_frames_out);
-       Video (image, _video_frames_out, sub);
-       ++_video_frames_out;
+       TIMING ("Decoder emits %1", _video_frame);
+       Video (image, _video_frame, sub);
+       ++_video_frame;
+
        _last_image = image;
        _last_subtitle = sub;
 }
 
+void
+Decoder::emit_audio (shared_ptr<AudioBuffers> audio)
+{
+       Audio (audio, _audio_frame);
+       _audio_frame += audio->frames ();
+}
+
 void
 Decoder::process_subtitle (shared_ptr<TimedSubtitle> s)
 {
@@ -345,10 +243,3 @@ Decoder::process_subtitle (shared_ptr<TimedSubtitle> s)
                _timed_subtitle->subtitle()->set_position (Position (p.x - _film->crop().left, p.y - _film->crop().top));
        }
 }
-
-
-int
-Decoder::bytes_per_audio_sample () const
-{
-       return av_get_bytes_per_sample (audio_sample_format ());
-}