Added test that SurfaceFlinger metrics are calculated on android

tools/telemetry/telemetry/timeline/tab_id_importer.py

Index: tools/telemetry/telemetry/timeline/tab_id_importer.py
diff --git a/tools/telemetry/telemetry/timeline/tab_id_importer.py b/tools/telemetry/telemetry/timeline/tab_id_importer.py
new file mode 100644
index 0000000000000000000000000000000000000000..f5929d5bc6a67d40b6a53cdfaa08d6acae9c8430
--- /dev/null
+++ b/tools/telemetry/telemetry/timeline/tab_id_importer.py
@@ -0,0 +1,54 @@
+# Copyright 2014 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.
+
+from telemetry.timeline import importer
+from telemetry.timeline import trace_data as trace_data_module
+
+class TraceBufferOverflowException(Exception):
+  pass
+
+
+class TabIdImporter(importer.TimelineImporter):
+  def __init__(self, model, trace_data):
+    # Needs to run after all other importers so overflow events have been
+    # created on the model.
+    super(TabIdImporter, self).__init__(
+        model,
+        trace_data,
+        import_order=999)
+    self._trace_data = trace_data
+
+  @staticmethod
+  def GetSupportedPart():
+    return trace_data_module.TAB_ID_PART
+
+  def ImportEvents(self):
+    pass
+
+  def FinalizeImport(self):
+    self._CheckTraceBufferOverflow()
+    self._CreateTabIdsToThreadsMap()
+
+  def _CheckTraceBufferOverflow(self):
+    # Since _CreateTabIdsToThreadsMap() relies on markers output on timeline
+    # tracing data, it may not work in case we have trace events dropped due to
+    # trace buffer overflow.
+    for process in self._model.GetAllProcesses():
+      if process.trace_buffer_did_overflow:
+        raise TraceBufferOverflowException(
+            'Trace buffer of process with pid=%d overflowed at timestamp %d. '
+            'Raw trace data:\n%s' %
+            (process.pid, process.trace_buffer_overflow_event.start,
+             repr(self._trace_data)))
+
+  def _CreateTabIdsToThreadsMap(self):
+    tab_id_events = self._trace_data.GetEventsFor(
+        trace_data_module.TAB_ID_PART)
+
+    for tab_id in tab_id_events:
+      timeline_markers = self._model.FindTimelineMarkers(tab_id)
+      assert(len(timeline_markers) == 1)
+      assert(timeline_markers[0].start_thread == timeline_markers[0].end_thread)
+      self._model.AddMappingFromTabIdToRendererThread(
+          tab_id, timeline_markers[0].start_thread)