[Telemetry] Add capability for values to reference external files.

tools/telemetry/telemetry/results/json_output_formatter.py

Index: tools/telemetry/telemetry/results/json_output_formatter.py
diff --git a/tools/telemetry/telemetry/results/json_output_formatter.py b/tools/telemetry/telemetry/results/json_output_formatter.py
index bf23ede9b8452d039c828b6e99720b65c74370b9..6517c46b6a82c2e6be41d7bd0a26044239e3d881 100644
--- a/tools/telemetry/telemetry/results/json_output_formatter.py
+++ b/tools/telemetry/telemetry/results/json_output_formatter.py
@@ -3,6 +3,7 @@
 # found in the LICENSE file.
 
 import json
+import os
 
 from telemetry.results import output_formatter
 
@@ -15,18 +16,32 @@ def ResultsAsDict(page_test_results, benchmark_metadata):
   name).
 
   Args:
-    page_test_results: a PageTestResults object
-    benchmark_metadata: a benchmark.BenchmarkMetadata object
+    page_test_results: A PageTestResults object.
+    benchmark_metadata: A benchmark.BenchmarkMetadata object.
   """
   result_dict = {
     'format_version': '0.2',
     'benchmark_name': benchmark_metadata.name,
-    'summary_values': [v.AsDict() for v in
-                       page_test_results.all_summary_values],
-    'per_page_values': [v.AsDict() for v in
-                        page_test_results.all_page_specific_values],
-    'pages': {p.id: p.AsDict() for p in _GetAllPages(page_test_results)}
+    'pages': {p.id: p.AsDict() for p in _GetAllPages(page_test_results)},
   }
+  paths = page_test_results.all_paths
+  if paths:
+    paths_to_ids = {path: path_id for path_id, path in
+                    enumerate(page_test_results.all_paths)}
+    ids_to_paths = {path_id: path for path, path_id
+                    in paths_to_ids.iteritems()}
+    result_dict['paths'] = ids_to_paths
+    result_dict['summary_values'] = (
+        [_UpdatePaths(v.AsDict(), paths_to_ids)
+         for v in page_test_results.all_summary_values])
+    result_dict['per_page_values'] = (
+        [_UpdatePaths(v.AsDict(), paths_to_ids)
+         for v in page_test_results.all_page_specific_values])
+  else:

[nednguyen, 2014/09/08 18:45:24]

You can just keep the initialization of results_dict as the current states, and
then swap 'paths' : [....] with 'path_ids': [....] for pages' values,
summary_values, and per_page_values. It seems to make the code flow easier to
follow, e.g:
..
if paths:
  _ReplacePathsWithPathIds(results_dict['summary_values'], paths_to_ids)
   _ReplacePathsWithPathIds(results_dict['summary_values'], paths_to_ids)

+    result_dict['summary_values'] = [v.AsDict() for v in
+                                     page_test_results.all_summary_values]
+    result_dict['per_page_values'] = (
+        [v.AsDict() for v in page_test_results.all_page_specific_values])
 
   return result_dict
 
@@ -35,6 +50,31 @@ def _GetAllPages(page_test_results):
               page_test_results.all_page_runs)
   return pages
 
+def _UpdatePaths(value_dict, paths_to_ids):
+  """Takes a value dict and converts its paths to IDs.
+
+  Values hold an optional list of file paths, as strings, until they need to be
+  serialized. At serialization time, we figure out a canonical mapping of IDs
+  to absolute paths, and correct the list for each value here.
+
+  Args:
+    value_dict: A value dict having an optional paths field.
+    paths_to_ids: A dict mapping path strings to IDs.
+
+  Returns:
+    A value dict with a list of path IDs in the paths field.
+  """
+
+  paths = value_dict.get('paths', None)
+  if paths:
+    new_dict = value_dict.copy()
+    new_paths = [paths_to_ids[os.path.normcase(os.path.realpath(path))]
+                 for path in paths]
+    new_dict['paths'] = new_paths

[nednguyen, 2014/09/08 18:45:24]

I think 'path_ids' is the right term. You can just remove the 'paths' and
replace with 'path_ids'.

+    return new_dict
+  else:
+    return value_dict
+
 class JsonOutputFormatter(output_formatter.OutputFormatter):
   def __init__(self, output_stream, benchmark_metadata):
     super(JsonOutputFormatter, self).__init__(output_stream)