Add CSV export, suitable for use with "pivot tables" in spreadsheets.

tools/telemetry/telemetry/results/csv_pivot_table_output_formatter.py

Index: tools/telemetry/telemetry/results/csv_pivot_table_output_formatter.py
diff --git a/tools/telemetry/telemetry/results/csv_pivot_table_output_formatter.py b/tools/telemetry/telemetry/results/csv_pivot_table_output_formatter.py
new file mode 100644
index 0000000000000000000000000000000000000000..d6cd103199c7111e47e3ee1409125b1ac20fbe39
--- /dev/null
+++ b/tools/telemetry/telemetry/results/csv_pivot_table_output_formatter.py
@@ -0,0 +1,58 @@
+# 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 csv
+
+from telemetry.results import output_formatter
+from telemetry.value import scalar
+
+
+class CsvPivotTableOutputFormatter(output_formatter.OutputFormatter):
+  """Output the results as CSV suitable for reading into a spreadsheet.
+
+  This will write a header row, and one row for each value. Each value row
+  contains the value and unit, identifies the value (page_set, page, name), and
+  (optionally) data from --output-trace-tag. This format matches what
+  spreadsheet programs expect as input for a "pivot table".
+
+  A trace tag (--output-trace-tag) can be used to tag each value, to allow
+  easy combination of the resulting CSVs from several runs.
+  If the trace_tag contains a comma, it will be written as several
+  comma-separated values.
+
+  This class only processes scalar values.
+  """
+
+  FIELDS = ['page_set', 'page', 'name', 'value', 'units', 'run_index']
+
+  def __init__(self, output_stream, trace_tag=''):
+    super(CsvPivotTableOutputFormatter, self).__init__(output_stream)
+    self._trace_tag = trace_tag
+
+  def Format(self, page_test_results):
+    csv_writer = csv.writer(self.output_stream)
+
+    # Observe trace_tag. Use comma to split up the trace tag.
+    tag_values = self._trace_tag.split(',')
+    tag_values = filter(lambda x: x, tag_values) # filter empty list entries
+    tag_headers = ['trace_tag_%d' % i for i in range(len(tag_values))]
+
+    # Write header.
+    csv_writer.writerow(self.FIELDS + tag_headers)
+
+    # Write all values. Each row contains a value + page-level metadata.
+    for run in page_test_results.all_page_runs:
+      run_index = page_test_results.all_page_runs.index(run)
+      page_dict = {
+          'page': run.page.display_name,
+          'page_set': run.page.page_set.Name(),
+          'run_index': run_index,
+      }
+      for value in run.values:
+        if isinstance(value, scalar.ScalarValue):
+          value_dict = value.AsDict()

[nednguyen, 2014/10/09 17:05:14]

I don't like the usage of value dict and this dynamic lookup since it creates
dependency between your header field names and the value dict. 

Please use direct fields: 
page_set --> page_set.Name()
page --> page.display_name
name --> value.name
units --> value.units

[vogelheim, 2014/10/10 11:27:43]

Done.

+          value_dict.update(page_dict.items())
+          csv_writer.writerow(
+              [value_dict[field] for field in self.FIELDS] + tag_values)
+

[nednguyen, 2014/10/09 17:05:14]

Nits: extra blank line here and in .._unittest file

[vogelheim, 2014/10/10 11:27:43]

Done.