Render inband text tracks in the media pipeline

content/renderer/media/webmediaplayer_impl.cc

Index: content/renderer/media/webmediaplayer_impl.cc
diff --git a/content/renderer/media/webmediaplayer_impl.cc b/content/renderer/media/webmediaplayer_impl.cc
index a99c3aef5920e14e1db9ac4e0afb103eb5867f8b..31015e4f4019f02edb2e8f2d83f1f55ff98bbe1a 100644
--- a/content/renderer/media/webmediaplayer_impl.cc
+++ b/content/renderer/media/webmediaplayer_impl.cc
@@ -15,6 +15,7 @@
 #include "base/debug/crash_logging.h"
 #include "base/message_loop/message_loop_proxy.h"
 #include "base/metrics/histogram.h"
+#include "base/stl_util.h"
 #include "base/strings/string_number_conversions.h"
 #include "base/synchronization/waitable_event.h"
 #include "cc/layers/video_layer.h"
@@ -37,6 +38,8 @@
 #include "media/base/media_log.h"
 #include "media/base/media_switches.h"
 #include "media/base/pipeline.h"
+#include "media/base/text_cue.h"
+#include "media/base/text_renderer.h"
 #include "media/base/video_frame.h"
 #include "media/filters/audio_renderer_impl.h"
 #include "media/filters/chunk_demuxer.h"
@@ -46,6 +49,7 @@
 #include "media/filters/gpu_video_accelerator_factories.h"
 #include "media/filters/gpu_video_decoder.h"
 #include "media/filters/opus_audio_decoder.h"
+#include "media/filters/text_decoder_impl.h"
 #include "media/filters/video_renderer_base.h"
 #include "media/filters/vpx_video_decoder.h"
 #include "third_party/WebKit/public/platform/WebMediaSource.h"
@@ -960,6 +964,15 @@ WebMediaPlayerImpl::OnTextTrack(media::TextKind kind,
                                                         text_track));
 }
 
+void WebMediaPlayerImpl::AddTextStream(media::DemuxerStream* text_stream,
+                                       media::TextKind kind,
+                                       const std::string& label,
+                                       const std::string& language) {
+  DCHECK(main_loop_->BelongsToCurrentThread());
+  scoped_ptr<media::TextTrack> text_track = OnTextTrack(kind, label, language);
+  text_track_map_[text_stream] = text_track.release();

[acolwell GONE FROM CHROMIUM, 2013/09/25 23:23:10]

I don't think you need the map here. You can hide the thread hopping business in
TextTrackImpl and let the renderer own the TextTrack instance.

I think if you use the following signature instead you'd avoid the need for the
map here and avoid map lookups on each cue. It would also allow the TextRenderer
to control the text track lifetime instead of having to coordinate with this map
here.
typedef base::Callback<void(scoped_ptr<media::TextTrack>)>
AddTextTrack(media::TextKind kind,
             const std::string& label,
             const std::string& language,
             const AddTextTrackDoneCB& done_cb)

[Matthew Heaney (Chromium), 2013/09/29 03:31:23]

Done.

+}
+
 void WebMediaPlayerImpl::OnKeyError(const std::string& session_id,
                                     media::MediaKeys::KeyError error_code,
                                     int system_code) {
@@ -1028,6 +1041,9 @@ void WebMediaPlayerImpl::StartPipeline() {
   UMA_HISTOGRAM_BOOLEAN("Media.MSE.Playback",
                         (load_type_ == LoadTypeMediaSource));
 
+  const bool enable_inband_text_tracks =
+      cmd_line->HasSwitch(switches::kEnableInbandTextTracks);
+
   // Figure out which demuxer to use.
   if (load_type_ != LoadTypeMediaSource) {
     DCHECK(!chunk_demuxer_);
@@ -1036,14 +1052,17 @@ void WebMediaPlayerImpl::StartPipeline() {
     demuxer_.reset(new media::FFmpegDemuxer(
         media_loop_, data_source_.get(),
         BIND_TO_RENDER_LOOP_1(&WebMediaPlayerImpl::OnNeedKey, ""),
+        enable_inband_text_tracks,
         media_log_));
   } else {
     DCHECK(!chunk_demuxer_);
     DCHECK(!data_source_);
 
+    // TODO(matthewjheaney): Once we have resolved the threading issues,
+    // change the chunk demuxer to user the same mechanism as for ffmpeg.
     media::AddTextTrackCB add_text_track_cb;
 
-    if (cmd_line->HasSwitch(switches::kEnableInbandTextTracks)) {
+    if (enable_inband_text_tracks) {
       add_text_track_cb =
           base::Bind(&WebMediaPlayerImpl::OnTextTrack, base::Unretained(this));
     }
@@ -1121,6 +1140,20 @@ void WebMediaPlayerImpl::StartPipeline() {
           true));
   filter_collection->SetVideoRenderer(video_renderer.Pass());
 
+  if (load_type_ != LoadTypeMediaSource && enable_inband_text_tracks) {
+    scoped_ptr<media::TextDecoder> text_decoder(
+        new media::TextDecoderImpl(media_loop_));
+
+    scoped_ptr<media::TextRenderer> text_renderer(
+        new media::TextRenderer(
+            media_loop_,
+            text_decoder.Pass(),
+            BIND_TO_RENDER_LOOP(&WebMediaPlayerImpl::AddTextStream),
+            BIND_TO_RENDER_LOOP(&WebMediaPlayerImpl::CueReady)));
+
+    filter_collection->SetTextRenderer(text_renderer.Pass());
+  }
+
   // ... and we're ready to go!
   starting_ = true;
   pipeline_->Start(
@@ -1177,6 +1210,8 @@ void WebMediaPlayerImpl::Destroy() {
       &base::WaitableEvent::Signal, base::Unretained(&waiter)));
   waiter.Wait();
 
+  STLDeleteValues(&text_track_map_);
+
   // Let V8 know we are not using extra resources anymore.
   if (incremented_externally_allocated_memory_) {
     v8::V8::AdjustAmountOfExternalAllocatedMemory(-kPlayerExtraMemory);
@@ -1243,4 +1278,26 @@ void WebMediaPlayerImpl::FrameReady(
       &WebMediaPlayerImpl::Repaint, AsWeakPtr()));
 }
 
+void WebMediaPlayerImpl::CueReady(

[acolwell GONE FROM CHROMIUM, 2013/09/25 23:23:10]

You shouldn't need this. You can hide the thread hopping in TextTrackImpl and
the TextRenderer should just be able to interact with TextTrack objects.

[Matthew Heaney (Chromium), 2013/09/29 03:31:23]

Done.

+    media::DemuxerStream* text_stream,
+    const scoped_refptr<media::TextCue>& text_cue) {
+  DCHECK(main_loop_->BelongsToCurrentThread());
+  DCHECK_NE(text_stream, static_cast<media::DemuxerStream*>(NULL));
+  DCHECK(text_cue != NULL && !text_cue->text().empty());
+
+  TextTrackMap::iterator it = text_track_map_.find(text_stream);
+  DCHECK(it != text_track_map_.end());
+
+  media::TextTrack* text_track = it->second;
+  DCHECK_NE(text_track, static_cast<media::TextTrack*>(NULL));
+
+  base::TimeDelta start = text_cue->timestamp();
+  base::TimeDelta end = start + text_cue->duration();
+
+  text_track->addWebVTTCue(start, end,
+                           text_cue->id(),
+                           text_cue->text(),
+                           text_cue->settings());
+}
+
 }  // namespace content