Render inband text tracks in the media pipeline

media/filters/webvtt_util.h

Index: media/filters/webvtt_util.h
diff --git a/media/filters/webvtt_util.h b/media/filters/webvtt_util.h
new file mode 100644
index 0000000000000000000000000000000000000000..391f322d27bce14f3d8f8941ff518631cbc4cd75
--- /dev/null
+++ b/media/filters/webvtt_util.h
@@ -0,0 +1,36 @@
+// Copyright (c) 2013 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef MEDIA_FILTERS_WEBVTT_UTIL_H_
+#define MEDIA_FILTERS_WEBVTT_UTIL_H_
+
+#include <vector>
+
+namespace media {
+
+// Utility function to create side data item for decoder buffer.
+template<typename T>
+static void MakeSideData(T id_begin, T id_end,
+                         T settings_begin, T settings_end,
+                         std::vector<uint8>* side_data);
+
+}  // namespace media
+
+template<typename T>

[acolwell GONE FROM CHROMIUM, 2013/10/29 21:14:19]

nit: Do we really need a template here?

[Matthew Heaney (Chromium), 2013/11/05 04:52:40]

What is the alternative?  In the ffmpeg demuxer case, you have uint8*, and in
the chunk demuxer case, you have string.

+inline void media::MakeSideData(
+    T id_begin, T id_end,
+    T settings_begin, T settings_end,
+    std::vector<uint8>* side_data) {
+  // The DecoderBuffer only supports a single side data item. In the case of
+  // a WebVTT cue, we can have potentially two side data items. In order to
+  // avoid disrupting DecoderBuffer any more than we need to, we copy both
+  // side data items onto a single one, and terminate each with a NUL marker.
+  side_data->clear();
+  side_data->insert(side_data->end(), id_begin, id_end);
+  side_data->push_back(0);
+  side_data->insert(side_data->end(), settings_begin, settings_end);
+  side_data->push_back(0);
+}
+
+#endif  // MEDIA_FILTERS_WEBVTT_UTIL_H_