do {
snprintf( key, sizeof(key), "%d.consumer", index++ );
nested = mlt_properties_get_data( properties, key, NULL );
- if ( nested ) mlt_consumer_start( nested );
+ if ( nested )
+ {
+ mlt_properties nested_props = MLT_CONSUMER_PROPERTIES(nested);
+ mlt_properties_set_position( nested_props, "_multi_position", 0 );
+ mlt_properties_set_data( nested_props, "_multi_audio", NULL, 0, NULL, NULL );
+ mlt_properties_set_int( nested_props, "_multi_samples", 0 );
+ mlt_consumer_start( nested );
+ }
} while ( nested );
}
do {
snprintf( key, sizeof(key), "%d.consumer", index++ );
nested = mlt_properties_get_data( properties, key, NULL );
- if ( nested ) mlt_consumer_put_frame( nested, mlt_frame_clone( frame, 0 ) );
+ if ( nested )
+ {
+ mlt_properties nested_props = MLT_CONSUMER_PROPERTIES(nested);
+ double self_fps = mlt_properties_get_double( properties, "fps" );
+ double nested_fps = mlt_properties_get_double( nested_props, "fps" );
+ mlt_position nested_pos = mlt_properties_get_position( nested_props, "_multi_position" );
+ mlt_position self_pos = mlt_frame_get_position( frame );
+ double self_time = self_pos / self_fps;
+ double nested_time = nested_pos / nested_fps;
+
+ // get the audio for the current frame
+ uint8_t *buffer = NULL;
+ mlt_audio_format format = mlt_audio_s16;
+ int channels = mlt_properties_get_int( properties, "channels" );
+ int frequency = mlt_properties_get_int( properties, "frequency" );
+ int current_samples = mlt_sample_calculator( self_fps, frequency, self_pos );
+ mlt_frame_get_audio( frame, (void**) &buffer, &format, &frequency, &channels, ¤t_samples );
+ int current_size = mlt_audio_format_size( format, current_samples, channels );
+
+ // get any leftover audio
+ int prev_size = 0;
+ uint8_t *prev_buffer = mlt_properties_get_data( nested_props, "_multi_audio", &prev_size );
+ uint8_t *new_buffer = NULL;
+ if ( prev_size > 0 )
+ {
+ new_buffer = mlt_pool_alloc( prev_size + current_size );
+ memcpy( new_buffer, prev_buffer, prev_size );
+ memcpy( new_buffer + prev_size, buffer, current_size );
+ buffer = new_buffer;
+ }
+ current_size += prev_size;
+ current_samples += mlt_properties_get_int( nested_props, "_multi_samples" );
+
+ while ( nested_time <= self_time )
+ {
+ // put ideal number of samples into cloned frame
+ mlt_frame clone_frame = mlt_frame_clone( frame, 0 );
+ int nested_samples = mlt_sample_calculator( nested_fps, frequency, nested_pos );
+ // -10 is an optimization to avoid tiny amounts of leftover samples
+ nested_samples = nested_samples > current_samples - 10 ? current_samples : nested_samples;
+ int nested_size = mlt_audio_format_size( format, nested_samples, channels );
+ if ( nested_size > 0 )
+ {
+ prev_buffer = mlt_pool_alloc( nested_size );
+ memcpy( prev_buffer, buffer, nested_size );
+ }
+ else
+ {
+ prev_buffer = NULL;
+ nested_size = 0;
+ }
+ mlt_frame_set_audio( clone_frame, prev_buffer, format, nested_size, mlt_pool_release );
+ mlt_properties_set_int( MLT_FRAME_PROPERTIES(clone_frame), "audio_samples", nested_samples );
+ mlt_properties_set_int( MLT_FRAME_PROPERTIES(clone_frame), "audio_frequency", frequency );
+ mlt_properties_set_int( MLT_FRAME_PROPERTIES(clone_frame), "audio_channels", channels );
+
+ // chomp the audio
+ current_samples -= nested_samples;
+ current_size -= nested_size;
+ buffer += nested_size;
+
+ // send frame to nested consumer
+ mlt_consumer_put_frame( nested, clone_frame );
+ mlt_properties_set_position( nested_props, "_multi_position", ++nested_pos );
+ nested_time = nested_pos / nested_fps;
+ }
+
+ // save any remaining audio
+ if ( current_size > 0 )
+ {
+ prev_buffer = mlt_pool_alloc( current_size );
+ memcpy( prev_buffer, buffer, current_size );
+ }
+ else
+ {
+ prev_buffer = NULL;
+ current_size = 0;
+ }
+ mlt_pool_release( new_buffer );
+ mlt_properties_set_data( nested_props, "_multi_audio", prev_buffer, current_size, mlt_pool_release, NULL );
+ mlt_properties_set_int( nested_props, "_multi_samples", current_samples );
+ }
} while ( nested );
}
// Check that we have a frame to work with
if ( frame && !terminated && !is_stopped( consumer ) )
{
- if ( !mlt_properties_get_int( MLT_FRAME_PROPERTIES(frame), "rendered" ) )
+ if ( mlt_properties_get_int( MLT_FRAME_PROPERTIES(frame), "rendered" ) )
+ {
+ if ( mlt_properties_get_int( MLT_FRAME_PROPERTIES(frame), "_speed" ) == 0 )
+ foreach_consumer_refresh( consumer );
+ foreach_consumer_put( consumer, frame );
+ mlt_events_fire( properties, "consumer-frame-show", frame, NULL );
+ }
+ else
{
int dropped = mlt_properties_get_int( properties, "_dropped" );
- mlt_frame_close( frame );
mlt_log_info( MLT_CONSUMER_SERVICE(consumer), "dropped frame %d\n", ++dropped );
mlt_properties_set_int( properties, "_dropped", dropped );
- continue;
}
- if ( mlt_properties_get_int( MLT_FRAME_PROPERTIES(frame), "_speed" ) == 0 )
- foreach_consumer_refresh( consumer );
- foreach_consumer_put( consumer, frame );
- mlt_events_fire( properties, "consumer-frame-show", frame, NULL );
mlt_frame_close( frame );
}
else