Chop off platform name from step name

appengine/findit/waterfall/try_job_util.py

 # Copyright 2015 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 datetime import datetime
 from datetime import timedelta
 import logging
 
 from google.appengine.ext import ndb
 
@@ skipping 130 matching lines @@
   if not failed_steps:
     return failed_steps_and_tests
 
   for step_name, step in failed_steps.iteritems():
     for test_name in step.get('tests', [None]):
       failed_steps_and_tests.append([step_name, test_name])
 
   return sorted(failed_steps_and_tests)
 
 
+def _RemovePlatformFromStepName(step_name):
+  """Returns step name without platform.
+
+  Args:
+    step_name: Raw step name. Example: 'net_unittests on Windows-10'.
+
+  Returns:
+    Step name without platform or the string ' on '. Example: 'net_unittests'.
+  """
+  separator = ' on '
+  return step_name.split(separator)[0]

[lijeffrey, 2016/08/10 23:24:25]

I believe step_name should always be 1 word, so you should be able just to use
step_name.split()[0] as we do in handlers/handler_util.py.

Chan can you please confirm this is the behavior that is to be expected?

If so this method is unneeded and you can use the 1-liner. Else we should
refactor that into a shared module that does the .split()[0] that this and
handlers_util.py can both use but that's probably outside the scope of this CL.
If we choose to go that route, put a TODO(lijeffrey): refactor this method to
share with handlers_util.py.

[chanli, 2016/08/11 20:47:06]

The changes here and in handler_util are all temporary change: they should work
for all cases we've seen so far but not 100% guaranteed. I'm thinking of keeping
track of the real step name without platform in the analysis results.

So I think this change is fine for me right now and feel free to add TODO for
later change.

+
+
 def GetSuspectedCLsWithFailures(heuristic_result):
   """Generates a list of suspected CLs with failures.
 
   Args:
     heuristic_result: the heuristic_result from which to generate the list of
     suspected CLs with failures.
 
   Returns:
     A list of suspected CLs with failures that each could look like:
 
         [step_name, revision, test_name]
 
     or could look like:
 
         [step_name, revision, None]
   """
   suspected_cls_with_failures = []
 
   if not heuristic_result:
     return suspected_cls_with_failures
 
   # Iterates through the failures, tests, and suspected_cls, appending suspected
   # CLs and failures to the list.
   for failure in heuristic_result['failures']:
     if failure.get('tests'):
       for test in failure['tests']:
         for suspected_cl in test.get('suspected_cls', []):
           suspected_cls_with_failures.append([
-              failure['step_name'],
+              _RemovePlatformFromStepName(failure['step_name']),
               suspected_cl['revision'],
               test['test_name']])
     else:
       for suspected_cl in failure['suspected_cls']:
         suspected_cls_with_failures.append([
-            failure['step_name'],
+            _RemovePlatformFromStepName(failure['step_name']),
             suspected_cl['revision'],
             None])
 
   return suspected_cls_with_failures
 
 
 def _LinkAnalysisToBuildFailureGroup(
     master_name, builder_name, build_number, failure_group_key):
   analysis = WfAnalysis.Get(master_name, builder_name, build_number)
   analysis.failure_group_key = failure_group_key
@@ skipping 251 matching lines @@
               pipeline.pipeline_status_path, try_job_type)
     else:  # pragma: no cover
       logging_str = (
           'Try job was scheduled for build %s, %s, %s: %s because of %s '
           'failure.') % (
               master_name, builder_name, build_number,
               pipeline.pipeline_status_path, try_job_type)
     logging.info(logging_str)
 
   return failure_result_map