Added support for improvement_direction to relevant values, which is propogated to chartjson.

tools/telemetry/telemetry/results/csv_pivot_table_output_formatter_unittest.py

 # 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.
 import os
 import StringIO
 import unittest
 
 from telemetry import page as page_module
 from telemetry.page import page_set
 from telemetry.results import csv_pivot_table_output_formatter
 from telemetry.results import page_test_results
+from telemetry.value import improvement_direction
 from telemetry.value import scalar
 
 
 def _MakePageSet():
   ps = page_set.PageSet(file_path=os.path.dirname(__file__))
   ps.AddUserStory(page_module.Page('http://www.foo.com/', ps, ps.base_dir))
   ps.AddUserStory(page_module.Page('http://www.bar.com/', ps, ps.base_dir))
   return ps
 
 
@@ skipping 28 matching lines @@
         self._results.AddValue(v)
       self._results.DidRunPage(page)
 
   def Format(self):
     self._formatter.Format(self._results)
     return self._output.getvalue()
 
   def testSimple(self):
     # Test a simple benchmark with only one value:
     self.SimulateBenchmarkRun({
-        self._page_set[0]: [scalar.ScalarValue(None, 'foo', 'seconds', 3)]})
+        self._page_set[0]: [scalar.ScalarValue(
+            None, 'foo', 'seconds', 3,
+            improvement_direction=improvement_direction.DOWN)]})
     expected = self._LINE_SEPARATOR.join([
         'page_set,page,name,value,units,run_index',
         'page_set,http://www.foo.com/,foo,3,seconds,0',
         ''])
 
     self.assertEqual(expected, self.Format())
 
   def testMultiplePagesAndValues(self):
     self.SimulateBenchmarkRun({
-        self._page_set[0]: [scalar.ScalarValue(None, 'foo', 'seconds', 4)],
-        self._page_set[1]: [scalar.ScalarValue(None, 'foo', 'seconds', 3.4),
-                            scalar.ScalarValue(None, 'bar', 'km', 10),
-                            scalar.ScalarValue(None, 'baz', 'count', 5)]})
+        self._page_set[0]: [
+            scalar.ScalarValue(
+              None, 'foo', 'seconds', 4,
+              improvement_direction=improvement_direction.DOWN)],
+        self._page_set[1]: [
+            scalar.ScalarValue(
+                None, 'foo', 'seconds', 3.4,
+                improvement_direction=improvement_direction.DOWN),
+            scalar.ScalarValue(
+                None, 'bar', 'km', 10,
+                improvement_direction=improvement_direction.DOWN),
+            scalar.ScalarValue(
+                None, 'baz', 'count', 5,
+                improvement_direction=improvement_direction.DOWN)]})
 
     # Parse CSV output into list of lists.
     csv_string = self.Format()
     lines = csv_string.split(self._LINE_SEPARATOR)
     values = [s.split(',') for s in lines[1:-1]]
 
     self.assertEquals(len(values), 4)  # We expect 4 value in total.
     self.assertEquals(len(set((v[1] for v in values))), 2)  # 2 pages.
     self.assertEquals(len(set((v[2] for v in values))), 3)  # 3 value names.
 
   def testTraceTag(self):
     self.MakeFormatter(trace_tag='date,option')
     self.SimulateBenchmarkRun({
-        self._page_set[0]: [scalar.ScalarValue(None, 'foo', 'seconds', 3),
-                            scalar.ScalarValue(None, 'bar', 'tons', 5)]})
+        self._page_set[0]: [
+            scalar.ScalarValue(
+                None, 'foo', 'seconds', 3,
+                improvement_direction=improvement_direction.DOWN),
+            scalar.ScalarValue(
+                None, 'bar', 'tons', 5,
+                improvement_direction=improvement_direction.DOWN)]})
     output = self.Format().split(self._LINE_SEPARATOR)
 
     self.assertTrue(output[0].endswith(',trace_tag_0,trace_tag_1'))
     for line in output[1:-1]:
       self.assertTrue(line.endswith(',date,option'))