Accessor for ClosedCaptionsDialog.
[dcpomatic.git] / src / lib / ffmpeg_file_encoder.cc
index eb7036429fb44bbeb05893bffbfa6f2ab89ee331..e9809008098536ee6591904e2047b378683d22d1 100644 (file)
@@ -38,6 +38,7 @@ using std::pair;
 using boost::shared_ptr;
 using boost::bind;
 using boost::weak_ptr;
+using boost::optional;
 using namespace dcpomatic;
 
 int FFmpegFileEncoder::_video_stream_index = 0;
@@ -51,6 +52,10 @@ FFmpegFileEncoder::FFmpegFileEncoder (
        ExportFormat format,
        int x264_crf,
        boost::filesystem::path output
+#ifdef DCPOMATIC_VARIANT_SWAROOP
+       , optional<dcp::Key> key
+       , optional<string> id
+#endif
        )
        : _video_options (0)
        , _audio_channels (channels)
@@ -58,6 +63,7 @@ FFmpegFileEncoder::FFmpegFileEncoder (
        , _video_frame_size (video_frame_size)
        , _video_frame_rate (video_frame_rate)
        , _audio_frame_rate (audio_frame_rate)
+       , _audio_frames (0)
 {
        _pixel_format = pixel_format (format);
 
@@ -69,18 +75,30 @@ FFmpegFileEncoder::FFmpegFileEncoder (
                av_dict_set (&_video_options, "profile", "3", 0);
                av_dict_set (&_video_options, "threads", "auto", 0);
                break;
-       case EXPORT_FORMAT_H264:
+       case EXPORT_FORMAT_H264_AAC:
                _sample_format = AV_SAMPLE_FMT_FLTP;
                _video_codec_name = "libx264";
                _audio_codec_name = "aac";
                av_dict_set_int (&_video_options, "crf", x264_crf, 0);
                break;
+       case EXPORT_FORMAT_H264_PCM:
+               _sample_format = AV_SAMPLE_FMT_S32;
+               _video_codec_name = "libx264";
+               _audio_codec_name = "pcm_s24le";
+               av_dict_set_int (&_video_options, "crf", x264_crf, 0);
+               break;
+       default:
+               DCPOMATIC_ASSERT (false);
        }
 
        setup_video ();
        setup_audio ();
 
+#ifdef DCPOMATIC_VARIANT_SWAROOP
+       int r = avformat_alloc_output_context2 (&_format_context, av_guess_format("mov", 0, 0), 0, 0);
+#else
        int r = avformat_alloc_output_context2 (&_format_context, 0, 0, _output.string().c_str());
+#endif
        if (!_format_context) {
                throw runtime_error (String::compose("could not allocate FFmpeg format context (%1)", r));
        }
@@ -116,7 +134,24 @@ FFmpegFileEncoder::FFmpegFileEncoder (
                throw runtime_error ("could not open FFmpeg output file");
        }
 
-       if (avformat_write_header (_format_context, 0) < 0) {
+       AVDictionary* options = 0;
+
+#ifdef DCPOMATIC_VARIANT_SWAROOP
+       if (key) {
+               av_dict_set (&options, "encryption_key", key->hex().c_str(), 0);
+               /* XXX: is this OK? */
+               av_dict_set (&options, "encryption_kid", "00000000000000000000000000000000", 0);
+               av_dict_set (&options, "encryption_scheme", "cenc-aes-ctr", 0);
+       }
+
+       if (id) {
+               if (av_dict_set(&_format_context->metadata, SWAROOP_ID_TAG, id->c_str(), 0) < 0) {
+                       throw runtime_error ("Could not write ID to output");
+               }
+       }
+#endif
+
+       if (avformat_write_header (_format_context, &options) < 0) {
                throw runtime_error ("could not write header to FFmpeg output file");
        }
 
@@ -129,7 +164,8 @@ FFmpegFileEncoder::pixel_format (ExportFormat format)
        switch (format) {
        case EXPORT_FORMAT_PRORES:
                return AV_PIX_FMT_YUV422P10;
-       case EXPORT_FORMAT_H264:
+       case EXPORT_FORMAT_H264_AAC:
+       case EXPORT_FORMAT_H264_PCM:
                return AV_PIX_FMT_YUV420P;
        default:
                DCPOMATIC_ASSERT (false);
@@ -178,7 +214,7 @@ FFmpegFileEncoder::setup_audio ()
        avcodec_get_context_defaults3 (_audio_codec_context, _audio_codec);
 
        /* XXX: configurable */
-       _audio_codec_context->bit_rate = 256 * 1024;
+       _audio_codec_context->bit_rate = _audio_channels * 128 * 1024;
        _audio_codec_context->sample_fmt = _sample_format;
        _audio_codec_context->sample_rate = _audio_frame_rate;
        _audio_codec_context->channel_layout = av_get_default_channel_layout (_audio_channels);
@@ -245,7 +281,11 @@ FFmpegFileEncoder::video (shared_ptr<PlayerVideo> video, DCPTime time)
        AVFrame* frame = av_frame_alloc ();
        DCPOMATIC_ASSERT (frame);
 
-       _pending_images[image->data()[0]] = image;
+       {
+               boost::mutex::scoped_lock lm (_pending_images_mutex);
+               _pending_images[image->data()[0]] = image;
+       }
+
        for (int i = 0; i < 3; ++i) {
                AVBufferRef* buffer = av_buffer_create(image->data()[i], image->stride()[i] * image->size().height, &buffer_free, this, 0);
                frame->buf[i] = av_buffer_ref (buffer);
@@ -257,7 +297,8 @@ FFmpegFileEncoder::video (shared_ptr<PlayerVideo> video, DCPTime time)
        frame->width = image->size().width;
        frame->height = image->size().height;
        frame->format = _pixel_format;
-       frame->pts = time.seconds() / av_q2d (_video_stream->time_base);
+       DCPOMATIC_ASSERT (_video_stream->time_base.num == 1);
+       frame->pts = time.get() * _video_stream->time_base.den / DCPTime::HZ;
 
        AVPacket packet;
        av_init_packet (&packet);
@@ -329,6 +370,16 @@ FFmpegFileEncoder::audio_frame (int size)
                }
                break;
        }
+       case AV_SAMPLE_FMT_S32:
+       {
+               int32_t* q = reinterpret_cast<int32_t*> (samples);
+               for (int i = 0; i < size; ++i) {
+                       for (int j = 0; j < channels; ++j) {
+                               *q++ = p[j][i] * 2147483647;
+                       }
+               }
+               break;
+       }
        case AV_SAMPLE_FMT_FLTP:
        {
                float* q = reinterpret_cast<float*> (samples);
@@ -342,6 +393,9 @@ FFmpegFileEncoder::audio_frame (int size)
                DCPOMATIC_ASSERT (false);
        }
 
+       DCPOMATIC_ASSERT (_audio_stream->time_base.num == 1);
+       frame->pts = _audio_frames * _audio_stream->time_base.den / _audio_frame_rate;
+
        AVPacket packet;
        av_init_packet (&packet);
        packet.data = 0;
@@ -362,6 +416,7 @@ FFmpegFileEncoder::audio_frame (int size)
        av_frame_free (&frame);
 
        _pending_audio->trim_start (size);
+       _audio_frames += size;
 }
 
 void
@@ -379,5 +434,8 @@ FFmpegFileEncoder::buffer_free (void* opaque, uint8_t* data)
 void
 FFmpegFileEncoder::buffer_free2 (uint8_t* data)
 {
-       _pending_images.erase (data);
+       boost::mutex::scoped_lock lm (_pending_images_mutex);
+       if (_pending_images.find(data) != _pending_images.end()) {
+               _pending_images.erase (data);
+       }
 }