Logging improvements to allow prettier displays in the server GUI.
[dcpomatic.git] / src / lib / ffmpeg_decoder.cc
index d40b798baafb77058f019e2bc6cf71817b5bd2f8..83bce18308ae3b42de919ad6ffc26fe356ee3d61 100644 (file)
@@ -1,5 +1,5 @@
 /*
-    Copyright (C) 2012-2014 Carl Hetherington <cth@carlh.net>
+    Copyright (C) 2012-2015 Carl Hetherington <cth@carlh.net>
 
     This program is free software; you can redistribute it and/or modify
     it under the terms of the GNU General Public License as published by
  *  @brief A decoder using FFmpeg to decode content.
  */
 
-#include <stdexcept>
-#include <vector>
-#include <iomanip>
-#include <iostream>
-#include <stdint.h>
-#include <sndfile.h>
-extern "C" {
-#include <libavcodec/avcodec.h>
-#include <libavformat/avformat.h>
-}
-#include "film.h"
 #include "filter.h"
 #include "exceptions.h"
 #include "image.h"
 #include "util.h"
 #include "log.h"
 #include "ffmpeg_decoder.h"
+#include "ffmpeg_audio_stream.h"
+#include "ffmpeg_subtitle_stream.h"
 #include "filter_graph.h"
 #include "audio_buffers.h"
 #include "ffmpeg_content.h"
-#include "image_proxy.h"
+#include "raw_image_proxy.h"
+#include "film.h"
+#include "compose.hpp"
+extern "C" {
+#include <libavcodec/avcodec.h>
+#include <libavformat/avformat.h>
+}
+#include <boost/foreach.hpp>
+#include <vector>
+#include <iomanip>
+#include <iostream>
+#include <stdint.h>
 
 #include "i18n.h"
 
-#define LOG_GENERAL(...) film->log()->log (String::compose (__VA_ARGS__), Log::TYPE_GENERAL);
-#define LOG_ERROR(...) film->log()->log (String::compose (__VA_ARGS__), Log::TYPE_ERROR);
-#define LOG_WARNING(...) film->log()->log (__VA_ARGS__, Log::TYPE_WARNING);
+#define LOG_GENERAL(...) _log->log (String::compose (__VA_ARGS__), LogEntry::TYPE_GENERAL);
+#define LOG_ERROR(...) _log->log (String::compose (__VA_ARGS__), LogEntry::TYPE_ERROR);
+#define LOG_WARNING_NC(...) _log->log (__VA_ARGS__, LogEntry::TYPE_WARNING);
+#define LOG_WARNING(...) _log->log (String::compose (__VA_ARGS__), LogEntry::TYPE_WARNING);
 
 using std::cout;
-using std::string;
 using std::vector;
 using std::list;
 using std::min;
 using std::pair;
+using std::max;
 using boost::shared_ptr;
-using boost::optional;
-using boost::dynamic_pointer_cast;
-using libdcp::Size;
-
-FFmpegDecoder::FFmpegDecoder (shared_ptr<const Film> f, shared_ptr<const FFmpegContent> c, bool video, bool audio)
-       : Decoder (f)
-       , VideoDecoder (f, c)
-       , AudioDecoder (f, c)
-       , SubtitleDecoder (f)
-       , FFmpeg (c)
-       , _subtitle_codec_context (0)
-       , _subtitle_codec (0)
-       , _decode_video (video)
-       , _decode_audio (audio)
-       , _pts_offset (0)
-       , _just_sought (false)
-{
-       setup_subtitle ();
-
-       /* Audio and video frame PTS values may not start with 0.  We want
-          to fiddle them so that:
-
-          1.  One of them starts at time 0.
-          2.  The first video PTS value ends up on a frame boundary.
-
-          Then we remove big initial gaps in PTS and we allow our
-          insertion of black frames to work.
-
-          We will do:
-            audio_pts_to_use = audio_pts_from_ffmpeg + pts_offset;
-            video_pts_to_use = video_pts_from_ffmpeg + pts_offset;
-       */
-
-       bool const have_video = video && c->first_video();
-       bool const have_audio = audio && c->audio_stream() && c->audio_stream()->first_audio;
-
-       /* First, make one of them start at 0 */
-
-       if (have_audio && have_video) {
-               _pts_offset = - min (c->first_video().get(), c->audio_stream()->first_audio.get());
-       } else if (have_video) {
-               _pts_offset = - c->first_video().get();
-       } else if (have_audio) {
-               _pts_offset = - c->audio_stream()->first_audio.get();
-       }
-
-       /* Now adjust both so that the video pts starts on a frame */
-       if (have_video && have_audio) {
-               double first_video = c->first_video().get() + _pts_offset;
-               double const old_first_video = first_video;
-               
-               /* Round the first video up to a frame boundary */
-               if (fabs (rint (first_video * c->video_frame_rate()) - first_video * c->video_frame_rate()) > 1e-6) {
-                       first_video = ceil (first_video * c->video_frame_rate()) / c->video_frame_rate ();
-               }
-
-               _pts_offset += first_video - old_first_video;
-       }
-}
+using dcp::Size;
 
-FFmpegDecoder::~FFmpegDecoder ()
+FFmpegDecoder::FFmpegDecoder (shared_ptr<const FFmpegContent> c, shared_ptr<Log> log, bool fast)
+       : VideoDecoder (c)
+       , AudioDecoder (c, fast)
+       , SubtitleDecoder (c)
+       , FFmpeg (c)
+       , _log (log)
+       , _pts_offset (pts_offset (c->ffmpeg_audio_streams(), c->first_video(), c->video_frame_rate()))
 {
-       boost::mutex::scoped_lock lm (_mutex);
 
-       if (_subtitle_codec_context) {
-               avcodec_close (_subtitle_codec_context);
-       }
 }
 
 void
 FFmpegDecoder::flush ()
 {
        /* Get any remaining frames */
-       
+
        _packet.data = 0;
        _packet.size = 0;
-       
+
        /* XXX: should we reset _packet.data and size after each *_decode_* call? */
-       
-       if (_decode_video) {
-               while (decode_video_packet ()) {}
-       }
-       
-       if (_ffmpeg_content->audio_stream() && _decode_audio) {
-               decode_audio_packet ();
-       }
 
-       /* Stop us being asked for any more data */
-       _video_position = _ffmpeg_content->video_length_after_3d_combine ();
-       _audio_position = _ffmpeg_content->audio_length ();
+       while (decode_video_packet ()) {}
+
+       decode_audio_packet ();
+       AudioDecoder::flush ();
 }
 
-void
+bool
 FFmpegDecoder::pass ()
 {
        int r = av_read_frame (_format_context, &_packet);
 
-       if (r < 0) {
+       /* AVERROR_INVALIDDATA can apparently be returned sometimes even when av_read_frame
+          has pretty-much succeeded (and hence generated data which should be processed).
+          Hence it makes sense to continue here in that case.
+       */
+       if (r < 0 && r != AVERROR_INVALIDDATA) {
                if (r != AVERROR_EOF) {
                        /* Maybe we should fail here, but for now we'll just finish off instead */
                        char buf[256];
                        av_strerror (r, buf, sizeof(buf));
-                       shared_ptr<const Film> film = _film.lock ();
-                       assert (film);
                        LOG_ERROR (N_("error on av_read_frame (%1) (%2)"), buf, r);
                }
 
                flush ();
-               return;
+               return true;
        }
 
-       shared_ptr<const Film> film = _film.lock ();
-       assert (film);
-
        int const si = _packet.stream_index;
-       
-       if (si == _video_stream && _decode_video) {
+       shared_ptr<const FFmpegContent> fc = _ffmpeg_content;
+
+       if (si == _video_stream && !_ignore_video) {
                decode_video_packet ();
-       } else if (_ffmpeg_content->audio_stream() && _ffmpeg_content->audio_stream()->uses_index (_format_context, si) && _decode_audio) {
-               decode_audio_packet ();
-       } else if (_ffmpeg_content->subtitle_stream() && _ffmpeg_content->subtitle_stream()->uses_index (_format_context, si) && film->with_subtitles ()) {
+       } else if (fc->subtitle_stream() && fc->subtitle_stream()->uses_index (_format_context, si)) {
                decode_subtitle_packet ();
+       } else {
+               decode_audio_packet ();
        }
 
        av_free_packet (&_packet);
+       return false;
 }
 
 /** @param data pointer to array of pointers to buffers.
  *  Only the first buffer will be used for non-planar data, otherwise there will be one per channel.
  */
 shared_ptr<AudioBuffers>
-FFmpegDecoder::deinterleave_audio (uint8_t** data, int size)
+FFmpegDecoder::deinterleave_audio (shared_ptr<FFmpegAudioStream> stream, uint8_t** data, int size)
 {
-       assert (_ffmpeg_content->audio_channels());
-       assert (bytes_per_audio_sample());
+       DCPOMATIC_ASSERT (bytes_per_audio_sample (stream));
 
        /* Deinterleave and convert to float */
 
-       assert ((size % (bytes_per_audio_sample() * _ffmpeg_content->audio_channels())) == 0);
-
-       int const total_samples = size / bytes_per_audio_sample();
-       int const frames = total_samples / _ffmpeg_content->audio_channels();
-       shared_ptr<AudioBuffers> audio (new AudioBuffers (_ffmpeg_content->audio_channels(), frames));
+       /* total_samples and frames will be rounded down here, so if there are stray samples at the end
+          of the block that do not form a complete sample or frame they will be dropped.
+       */
+       int const total_samples = size / bytes_per_audio_sample (stream);
+       int const frames = total_samples / stream->channels();
+       shared_ptr<AudioBuffers> audio (new AudioBuffers (stream->channels(), frames));
 
-       switch (audio_sample_format()) {
+       switch (audio_sample_format (stream)) {
        case AV_SAMPLE_FMT_U8:
        {
                uint8_t* p = reinterpret_cast<uint8_t *> (data[0]);
@@ -210,14 +151,14 @@ FFmpegDecoder::deinterleave_audio (uint8_t** data, int size)
                        audio->data(channel)[sample] = float(*p++) / (1 << 23);
 
                        ++channel;
-                       if (channel == _ffmpeg_content->audio_channels()) {
+                       if (channel == stream->channels()) {
                                channel = 0;
                                ++sample;
                        }
                }
        }
        break;
-       
+
        case AV_SAMPLE_FMT_S16:
        {
                int16_t* p = reinterpret_cast<int16_t *> (data[0]);
@@ -227,7 +168,7 @@ FFmpegDecoder::deinterleave_audio (uint8_t** data, int size)
                        audio->data(channel)[sample] = float(*p++) / (1 << 15);
 
                        ++channel;
-                       if (channel == _ffmpeg_content->audio_channels()) {
+                       if (channel == stream->channels()) {
                                channel = 0;
                                ++sample;
                        }
@@ -238,14 +179,14 @@ FFmpegDecoder::deinterleave_audio (uint8_t** data, int size)
        case AV_SAMPLE_FMT_S16P:
        {
                int16_t** p = reinterpret_cast<int16_t **> (data);
-               for (int i = 0; i < _ffmpeg_content->audio_channels(); ++i) {
+               for (int i = 0; i < stream->channels(); ++i) {
                        for (int j = 0; j < frames; ++j) {
                                audio->data(i)[j] = static_cast<float>(p[i][j]) / (1 << 15);
                        }
                }
        }
        break;
-       
+
        case AV_SAMPLE_FMT_S32:
        {
                int32_t* p = reinterpret_cast<int32_t *> (data[0]);
@@ -255,7 +196,7 @@ FFmpegDecoder::deinterleave_audio (uint8_t** data, int size)
                        audio->data(channel)[sample] = static_cast<float>(*p++) / (1 << 31);
 
                        ++channel;
-                       if (channel == _ffmpeg_content->audio_channels()) {
+                       if (channel == stream->channels()) {
                                channel = 0;
                                ++sample;
                        }
@@ -272,123 +213,73 @@ FFmpegDecoder::deinterleave_audio (uint8_t** data, int size)
                        audio->data(channel)[sample] = *p++;
 
                        ++channel;
-                       if (channel == _ffmpeg_content->audio_channels()) {
+                       if (channel == stream->channels()) {
                                channel = 0;
                                ++sample;
                        }
                }
        }
        break;
-               
+
        case AV_SAMPLE_FMT_FLTP:
        {
                float** p = reinterpret_cast<float**> (data);
-               for (int i = 0; i < _ffmpeg_content->audio_channels(); ++i) {
+               for (int i = 0; i < stream->channels(); ++i) {
                        memcpy (audio->data(i), p[i], frames * sizeof(float));
                }
        }
        break;
 
        default:
-               throw DecodeError (String::compose (_("Unrecognised audio sample format (%1)"), static_cast<int> (audio_sample_format())));
+               throw DecodeError (String::compose (_("Unrecognised audio sample format (%1)"), static_cast<int> (audio_sample_format (stream))));
        }
 
        return audio;
 }
 
 AVSampleFormat
-FFmpegDecoder::audio_sample_format () const
+FFmpegDecoder::audio_sample_format (shared_ptr<FFmpegAudioStream> stream) const
 {
-       if (!_ffmpeg_content->audio_stream()) {
-               return (AVSampleFormat) 0;
-       }
-       
-       return audio_codec_context()->sample_fmt;
+       return stream->stream (_format_context)->codec->sample_fmt;
 }
 
 int
-FFmpegDecoder::bytes_per_audio_sample () const
+FFmpegDecoder::bytes_per_audio_sample (shared_ptr<FFmpegAudioStream> stream) const
 {
-       return av_get_bytes_per_sample (audio_sample_format ());
+       return av_get_bytes_per_sample (audio_sample_format (stream));
 }
 
 void
-FFmpegDecoder::seek (VideoContent::Frame frame, bool accurate)
+FFmpegDecoder::seek (ContentTime time, bool accurate)
 {
-       double const time_base = av_q2d (_format_context->streams[_video_stream]->time_base);
+       VideoDecoder::seek (time, accurate);
+       AudioDecoder::seek (time, accurate);
+       SubtitleDecoder::seek (time, accurate);
 
-       /* If we are doing an accurate seek, our initial shot will be 5 frames (5 being
-          a number plucked from the air) earlier than we want to end up.  The loop below
-          will hopefully then step through to where we want to be.
+       /* If we are doing an `accurate' seek, we need to use pre-roll, as
+          we don't really know what the seek will give us.
        */
-       int initial = frame;
-
-       if (accurate) {
-               initial -= 5;
-       }
 
-       if (initial < 0) {
-               initial = 0;
-       }
+       ContentTime pre_roll = accurate ? ContentTime::from_seconds (2) : ContentTime (0);
+       time -= pre_roll;
 
-       /* Initial seek time in the stream's timebase */
-       int64_t const initial_vt = ((initial / _ffmpeg_content->original_video_frame_rate()) - _pts_offset) / time_base;
-
-       av_seek_frame (_format_context, _video_stream, initial_vt, AVSEEK_FLAG_BACKWARD);
-
-       avcodec_flush_buffers (video_codec_context());
-       if (_subtitle_codec_context) {
-               avcodec_flush_buffers (_subtitle_codec_context);
-       }
-
-       /* This !accurate is piling hack upon hack; setting _just_sought to true
-          even with accurate == true defeats our attempt to align the start
-          of the video and audio.  Here we disable that defeat when accurate == true
-          i.e. when we are making a DCP rather than just previewing one.
-          Ewww.  This should be gone in 2.0.
+       /* XXX: it seems debatable whether PTS should be used here...
+          http://www.mjbshaw.com/2012/04/seeking-in-ffmpeg-know-your-timestamp.html
        */
-       if (!accurate) {
-               _just_sought = true;
-       }
-       
-       _video_position = frame;
-       
-       if (frame == 0 || !accurate) {
-               /* We're already there, or we're as close as we need to be */
-               return;
+
+       ContentTime u = time - _pts_offset;
+       if (u < ContentTime ()) {
+               u = ContentTime ();
        }
+       av_seek_frame (_format_context, _video_stream, u.seconds() / av_q2d (_format_context->streams[_video_stream]->time_base), AVSEEK_FLAG_BACKWARD);
 
-       while (true) {
-               int r = av_read_frame (_format_context, &_packet);
-               if (r < 0) {
-                       return;
-               }
+       avcodec_flush_buffers (video_codec_context());
 
-               if (_packet.stream_index != _video_stream) {
-                       av_free_packet (&_packet);
-                       continue;
-               }
-               
-               int finished = 0;
-               r = avcodec_decode_video2 (video_codec_context(), _frame, &finished, &_packet);
-               if (r >= 0 && finished) {
-                       _video_position = rint (
-                               (av_frame_get_best_effort_timestamp (_frame) * time_base + _pts_offset) * _ffmpeg_content->original_video_frame_rate()
-                               );
+       /* XXX: should be flushing audio buffers? */
 
-                       if (_video_position >= (frame - 1)) {
-                               av_free_packet (&_packet);
-                               break;
-                       }
-               }
-               
-               av_free_packet (&_packet);
+       if (subtitle_codec_context ()) {
+               avcodec_flush_buffers (subtitle_codec_context ());
        }
-
-       /* _video_position should be the next thing to be emitted, which will the one after the thing
-          we just saw.
-       */
-       _video_position++;
 }
 
 void
@@ -397,51 +288,66 @@ FFmpegDecoder::decode_audio_packet ()
        /* Audio packets can contain multiple frames, so we may have to call avcodec_decode_audio4
           several times.
        */
-       
+
        AVPacket copy_packet = _packet;
-       
+
+       /* XXX: inefficient */
+       vector<shared_ptr<FFmpegAudioStream> > streams = ffmpeg_content()->ffmpeg_audio_streams ();
+       vector<shared_ptr<FFmpegAudioStream> >::const_iterator stream = streams.begin ();
+       while (stream != streams.end () && !(*stream)->uses_index (_format_context, copy_packet.stream_index)) {
+               ++stream;
+       }
+
+       if (stream == streams.end ()) {
+               /* The packet's stream may not be an audio one; just ignore it in this method if so */
+               return;
+       }
+
        while (copy_packet.size > 0) {
 
                int frame_finished;
-               int const decode_result = avcodec_decode_audio4 (audio_codec_context(), _frame, &frame_finished, &copy_packet);
+               int decode_result = avcodec_decode_audio4 ((*stream)->stream (_format_context)->codec, _frame, &frame_finished, &copy_packet);
                if (decode_result < 0) {
-                       shared_ptr<const Film> film = _film.lock ();
-                       assert (film);
-                       LOG_ERROR ("avcodec_decode_audio4 failed (%1)", decode_result);
-                       return;
+                       /* avcodec_decode_audio4 can sometimes return an error even though it has decoded
+                          some valid data; for example dca_subframe_footer can return AVERROR_INVALIDDATA
+                          if it overreads the auxiliary data.  ffplay carries on if frame_finished is true,
+                          even in the face of such an error, so I think we should too.
+
+                          Returning from the method here caused mantis #352.
+                       */
+                       LOG_WARNING ("avcodec_decode_audio4 failed (%1)", decode_result);
+
+                       /* Fudge decode_result so that we come out of the while loop when
+                          we've processed this data.
+                       */
+                       decode_result = copy_packet.size;
                }
 
                if (frame_finished) {
-                       
-                       if (_audio_position == 0) {
-                               /* Where we are in the source, in seconds */
-                               double const pts = av_q2d (_format_context->streams[copy_packet.stream_index]->time_base)
-                                       * av_frame_get_best_effort_timestamp(_frame) + _pts_offset;
-
-                               if (pts > 0) {
-                                       /* Emit some silence */
-                                       int64_t frames = pts * _ffmpeg_content->content_audio_frame_rate ();
-                                       while (frames > 0) {
-                                               int64_t const this_time = min (frames, (int64_t) _ffmpeg_content->content_audio_frame_rate() / 2);
-                                               
-                                               shared_ptr<AudioBuffers> silence (
-                                                       new AudioBuffers (_ffmpeg_content->audio_channels(), this_time)
-                                                       );
-                                       
-                                               silence->make_silent ();
-                                               audio (silence, _audio_position);
-                                               frames -= this_time;
-                                       }
-                               }
-                       }
-                       
+                       ContentTime ct = ContentTime::from_seconds (
+                               av_frame_get_best_effort_timestamp (_frame) *
+                               av_q2d ((*stream)->stream (_format_context)->time_base))
+                               + _pts_offset;
+
                        int const data_size = av_samples_get_buffer_size (
-                               0, audio_codec_context()->channels, _frame->nb_samples, audio_sample_format (), 1
+                               0, (*stream)->stream(_format_context)->codec->channels, _frame->nb_samples, audio_sample_format (*stream), 1
                                );
-                       
-                       audio (deinterleave_audio (_frame->data, data_size), _audio_position);
+
+                       shared_ptr<AudioBuffers> data = deinterleave_audio (*stream, _frame->data, data_size);
+
+                       if (ct < ContentTime ()) {
+                               /* Discard audio data that comes before time 0 */
+                               Frame const remove = min (int64_t (data->frames()), (-ct).frames_ceil(double((*stream)->frame_rate ())));
+                               data->move (remove, 0, data->frames() - remove);
+                               data->set_frames (data->frames() - remove);
+                               ct += ContentTime::from_frames (remove, (*stream)->frame_rate ());
+                       }
+
+                       if (data->frames() > 0) {
+                               audio (*stream, data, ct);
+                       }
                }
-                       
+
                copy_packet.data += decode_result;
                copy_packet.size -= decode_result;
        }
@@ -458,19 +364,15 @@ FFmpegDecoder::decode_video_packet ()
        boost::mutex::scoped_lock lm (_filter_graphs_mutex);
 
        shared_ptr<FilterGraph> graph;
-       
+
        list<shared_ptr<FilterGraph> >::iterator i = _filter_graphs.begin();
-       while (i != _filter_graphs.end() && !(*i)->can_process (libdcp::Size (_frame->width, _frame->height), (AVPixelFormat) _frame->format)) {
+       while (i != _filter_graphs.end() && !(*i)->can_process (dcp::Size (_frame->width, _frame->height), (AVPixelFormat) _frame->format)) {
                ++i;
        }
 
        if (i == _filter_graphs.end ()) {
-               shared_ptr<const Film> film = _film.lock ();
-               assert (film);
-
-               graph.reset (new FilterGraph (_ffmpeg_content, libdcp::Size (_frame->width, _frame->height), (AVPixelFormat) _frame->format));
+               graph.reset (new FilterGraph (_ffmpeg_content, dcp::Size (_frame->width, _frame->height), (AVPixelFormat) _frame->format));
                _filter_graphs.push_back (graph);
-
                LOG_GENERAL (N_("New graph for %1x%2, pixel format %3"), _frame->width, _frame->height, _frame->format);
        } else {
                graph = *i;
@@ -478,136 +380,94 @@ FFmpegDecoder::decode_video_packet ()
 
        list<pair<shared_ptr<Image>, int64_t> > images = graph->process (_frame);
 
-       shared_ptr<const Film> film = _film.lock ();
-       assert (film);
-
        for (list<pair<shared_ptr<Image>, int64_t> >::iterator i = images.begin(); i != images.end(); ++i) {
 
                shared_ptr<Image> image = i->first;
-               
-               if (i->second != AV_NOPTS_VALUE) {
-
-                       double const pts = i->second * av_q2d (_format_context->streams[_video_stream]->time_base) + _pts_offset;
-
-                       if (_just_sought) {
-                               /* We just did a seek, so disable any attempts to correct for where we
-                                  are / should be.
-                               */
-                               _video_position = rint (pts * _ffmpeg_content->original_video_frame_rate ());
-                               _just_sought = false;
-                       }
-
-                       double const next = _video_position / _ffmpeg_content->original_video_frame_rate();
-                       double const one_frame = 1 / _ffmpeg_content->original_video_frame_rate ();
-                       double delta = pts - next;
-
-                       while (delta > one_frame) {
-                               /* This PTS is more than one frame forward in time of where we think we should be; emit
-                                  a black frame.
-                               */
-
-                               /* XXX: I think this should be a copy of the last frame... */
-                               boost::shared_ptr<Image> black (
-                                       new Image (
-                                               static_cast<AVPixelFormat> (_frame->format),
-                                               libdcp::Size (video_codec_context()->width, video_codec_context()->height),
-                                               true
-                                               )
-                                       );
-                               
-                               shared_ptr<const Film> film = _film.lock ();
-                               assert (film);
-
-                               black->make_black ();
-                               video (shared_ptr<ImageProxy> (new RawImageProxy (image, film->log())), false, _video_position);
-                               delta -= one_frame;
-                       }
 
-                       if (delta > -one_frame) {
-                               /* This PTS is within a frame of being right; emit this (otherwise it will be dropped) */
-                               video (shared_ptr<ImageProxy> (new RawImageProxy (image, film->log())), false, _video_position);
-                       }
-                               
+               if (i->second != AV_NOPTS_VALUE) {
+                       double const pts = i->second * av_q2d (_format_context->streams[_video_stream]->time_base) + _pts_offset.seconds ();
+                       video (
+                               shared_ptr<ImageProxy> (new RawImageProxy (image)),
+                               llrint (pts * _ffmpeg_content->video_frame_rate ())
+                               );
                } else {
-                       LOG_WARNING ("Dropping frame without PTS");
+                       LOG_WARNING_NC ("Dropping frame without PTS");
                }
        }
 
        return true;
 }
 
-       
-void
-FFmpegDecoder::setup_subtitle ()
-{
-       boost::mutex::scoped_lock lm (_mutex);
-       
-       if (!_ffmpeg_content->subtitle_stream()) {
-               return;
-       }
-
-       _subtitle_codec_context = _ffmpeg_content->subtitle_stream()->stream(_format_context)->codec;
-       if (_subtitle_codec_context == 0) {
-               throw DecodeError (N_("could not find subtitle stream"));
-       }
-
-       _subtitle_codec = avcodec_find_decoder (_subtitle_codec_context->codec_id);
-
-       if (_subtitle_codec == 0) {
-               throw DecodeError (N_("could not find subtitle decoder"));
-       }
-       
-       if (avcodec_open2 (_subtitle_codec_context, _subtitle_codec, 0) < 0) {
-               throw DecodeError (N_("could not open subtitle decoder"));
-       }
-}
-
-bool
-FFmpegDecoder::done () const
-{
-       bool const vd = !_decode_video || (_video_position >= _ffmpeg_content->video_length());
-       bool const ad = !_decode_audio || !_ffmpeg_content->audio_stream() || (_audio_position >= _ffmpeg_content->audio_length());
-       return vd && ad;
-}
-       
 void
 FFmpegDecoder::decode_subtitle_packet ()
 {
        int got_subtitle;
        AVSubtitle sub;
-       if (avcodec_decode_subtitle2 (_subtitle_codec_context, &sub, &got_subtitle, &_packet) < 0 || !got_subtitle) {
+       if (avcodec_decode_subtitle2 (subtitle_codec_context(), &sub, &got_subtitle, &_packet) < 0 || !got_subtitle) {
                return;
        }
 
-       /* Sometimes we get an empty AVSubtitle, which is used by some codecs to
-          indicate that the previous subtitle should stop.
-       */
        if (sub.num_rects <= 0) {
-               subtitle (shared_ptr<Image> (), dcpomatic::Rect<double> (), 0, 0);
+               /* Sometimes we get an empty AVSubtitle, which is used by some codecs to
+                  indicate that the previous subtitle should stop.  We can ignore it here.
+               */
                return;
-       } else if (sub.num_rects > 1) {
-               throw DecodeError (_("multi-part subtitles not yet supported"));
        }
-               
-       /* Subtitle PTS in seconds (within the source, not taking into account any of the
-          source that we may have chopped off for the DCP)
+
+       /* Subtitle PTS (within the source, not taking into account any of the
+          source that we may have chopped off for the DCP).
        */
-       double const packet_time = (static_cast<double> (sub.pts ) / AV_TIME_BASE) + _pts_offset;
+       FFmpegSubtitlePeriod sub_period = subtitle_period (sub);
+       ContentTimePeriod period;
+       period.from = sub_period.from + _pts_offset;
+       if (sub_period.to) {
+               /* We already know the subtitle period `to' time */
+               period.to = sub_period.to.get() + _pts_offset;
+       } else {
+               /* We have to look up the `to' time in the stream's records */
+               period.to = ffmpeg_content()->subtitle_stream()->find_subtitle_to (period.from);
+       }
+
+       for (unsigned int i = 0; i < sub.num_rects; ++i) {
+               AVSubtitleRect const * rect = sub.rects[i];
+
+               switch (rect->type) {
+               case SUBTITLE_NONE:
+                       break;
+               case SUBTITLE_BITMAP:
+                       decode_bitmap_subtitle (rect, period);
+                       break;
+               case SUBTITLE_TEXT:
+                       cout << "XXX: SUBTITLE_TEXT " << rect->text << "\n";
+                       break;
+               case SUBTITLE_ASS:
+                       cout << "XXX: SUBTITLE_ASS " << rect->ass << "\n";
+                       break;
+               }
+       }
 
-       /* hence start time for this sub */
-       Time const from = (packet_time + (double (sub.start_display_time) / 1e3)) * TIME_HZ;
-       Time const to = (packet_time + (double (sub.end_display_time) / 1e3)) * TIME_HZ;
+       avsubtitle_free (&sub);
+}
 
-       AVSubtitleRect const * rect = sub.rects[0];
+list<ContentTimePeriod>
+FFmpegDecoder::image_subtitles_during (ContentTimePeriod p, bool starting) const
+{
+       return _ffmpeg_content->subtitles_during (p, starting);
+}
 
-       if (rect->type != SUBTITLE_BITMAP) {
-               throw DecodeError (_("non-bitmap subtitles not yet supported"));
-       }
+list<ContentTimePeriod>
+FFmpegDecoder::text_subtitles_during (ContentTimePeriod, bool) const
+{
+       return list<ContentTimePeriod> ();
+}
 
+void
+FFmpegDecoder::decode_bitmap_subtitle (AVSubtitleRect const * rect, ContentTimePeriod period)
+{
        /* Note RGBA is expressed little-endian, so the first byte in the word is R, second
           G, third B, fourth A.
        */
-       shared_ptr<Image> image (new Image (PIX_FMT_RGBA, libdcp::Size (rect->w, rect->h), true));
+       shared_ptr<Image> image (new Image (PIX_FMT_RGBA, dcp::Size (rect->w, rect->h), true));
 
        /* Start of the first line in the subtitle */
        uint8_t* sub_p = rect->pict.data[0];
@@ -629,20 +489,13 @@ FFmpegDecoder::decode_subtitle_packet ()
                out_p += image->stride()[0] / sizeof (uint32_t);
        }
 
-       libdcp::Size const vs = _ffmpeg_content->video_size ();
-
-       subtitle (
-               image,
-               dcpomatic::Rect<double> (
-                       static_cast<double> (rect->x) / vs.width,
-                       static_cast<double> (rect->y) / vs.height,
-                       static_cast<double> (rect->w) / vs.width,
-                       static_cast<double> (rect->h) / vs.height
-                       ),
-               from,
-               to
+       dcp::Size const vs = _ffmpeg_content->video_size ();
+       dcpomatic::Rect<double> const scaled_rect (
+               static_cast<double> (rect->x) / vs.width,
+               static_cast<double> (rect->y) / vs.height,
+               static_cast<double> (rect->w) / vs.width,
+               static_cast<double> (rect->h) / vs.height
                );
-                         
-       
-       avsubtitle_free (&sub);
+
+       image_subtitle (period, image, scaled_rect);
 }