[Findit] Show build analysis references in UI for Findit Cross-platform auto-triage

appengine/findit/handlers/triage_analysis.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.
 
 """This module is to handle manual triage of analysis result.
 
 This handler will flag the analysis result as correct or incorrect.
 TODO: work on an automatic or semi-automatic way to triage analysis result.
 """
 
@@ skipping 72 matching lines @@
 
   # Both analyses must have non-empty potential culprit lists.
   if not potential_culprit_tuple_list_1 or not potential_culprit_tuple_list_2:
     return False
 
   # Both analyses must have matching potential culprit lists.
   return (sorted(potential_culprit_tuple_list_1) ==
           sorted(potential_culprit_tuple_list_2))
 
 
-def _AppendTriageHistoryRecord(analysis, is_correct, user_name):
+def _AppendTriageHistoryRecord(
+    analysis, is_correct, user_name, is_duplicate=False):
   """Appends a triage history record to the given analysis.
 
   Args:
     analysis: The analysis to which to append the history record.
     is_correct: True if the history record should indicate a correct judgement,
         otherwise False.
     user_name: The user_name of the person to include in the triage record.
+    is_duplicate: True if the history record should indicate that this triage is
+        a duplicate, otherwise False.

[chanli, 2016/06/30 23:52:25]

After the discussion, is_duplicate is not determined by history record to me.
It's actually determined by whether someone triages this analysis result or it's
triaged as a duplicate.

[josiahk, 2016/07/01 21:45:27]

Thanks! I have changed the docstring.

   """
   if is_correct:
     if analysis.suspected_cls:
-      analysis.result_status = result_status.FOUND_CORRECT
+      if is_duplicate:
+        analysis.result_status = result_status.FOUND_CORRECT_DUPLICATE
+      else:
+        analysis.result_status = result_status.FOUND_CORRECT
       analysis.culprit_cls = analysis.suspected_cls
     else:
       analysis.result_status = result_status.NOT_FOUND_CORRECT
       analysis.culprit_cls = None
   else:
     analysis.culprit_cls = None
     if analysis.suspected_cls:
-      analysis.result_status = result_status.FOUND_INCORRECT
+      if is_duplicate:
+        analysis.result_status = result_status.FOUND_INCORRECT_DUPLICATE
+      else:
+        analysis.result_status = result_status.FOUND_INCORRECT
     else:
       analysis.result_status = result_status.NOT_FOUND_INCORRECT
 
+  if not is_duplicate:
+    # Resets the reference to the "duplicator" triage analysis.

[chanli, 2016/06/30 23:46:37]

This comment is a little confusing...

How about adding reasoning about why we need to reset here?

[josiahk, 2016/07/01 21:45:27]

Done.

+    analysis.triage_reference_analysis_master_name = None
+    analysis.triage_reference_analysis_builder_name = None
+    analysis.triage_reference_analysis_build_number = None
+
   triage_record = {
       'triage_timestamp': calendar.timegm(datetime.utcnow().timetuple()),
       'user_name': user_name,
       'result_status': analysis.result_status,
       'version': analysis.version,
   }
   if not analysis.triage_history:
     analysis.triage_history = []
   analysis.triage_history.append(triage_record)
 
   analysis.put()
 
 
 @ndb.transactional
 def _UpdateAnalysisResultStatus(
     master_name, builder_name, build_number, is_correct, user_name=None):
   analysis = WfAnalysis.Get(master_name, builder_name, build_number)
   if not analysis or not analysis.completed:
     return False, None
 
-  _AppendTriageHistoryRecord(analysis, is_correct, user_name)
+  _AppendTriageHistoryRecord(analysis, is_correct, user_name,
+                             is_duplicate=False)
 
   return True, analysis
 
 
 def _GetDuplicateAnalyses(original_analysis):
   start_time = (original_analysis.build_start_time -
                 timedelta(hours=MATCHING_ANALYSIS_HOURS_AGO_START))
   end_time = (original_analysis.build_start_time +
               timedelta(hours=MATCHING_ANALYSIS_HOURS_AGO_END))
 
   # Don't count any analyses from today (except for exactly at midnight local
   # time).
   # Get current time (UTC).
   current_time_as_utc = pytz.utc.localize(datetime.utcnow())
 
   # Convert to local time.
   current_time_as_local = current_time_as_utc.astimezone(
       pytz.timezone(MATCHING_ANALYSIS_END_BOUND_TIME_ZONE))
 
   # Set hours and minutes to 0 to get midnight.
   local_midnight_as_local = current_time_as_local.replace(
       hour=0, minute=0, second=0, microsecond=0)
 
   # Convert back to UTC time.
   local_midnight_as_utc = local_midnight_as_local.astimezone(pytz.utc)
 
   # Strip timezone.
   local_midnight = local_midnight_as_utc.replace(tzinfo=None)
 
-  if end_time > local_midnight:  # pragma: no branch
+  if end_time > local_midnight:
     end_time = local_midnight
 
   # Retrieve potential duplicate build analyses.
   analysis_results = WfAnalysis.query(ndb.AND(
       WfAnalysis.build_start_time >= start_time,
       WfAnalysis.build_start_time <= end_time,
       WfAnalysis.result_status == result_status.FOUND_UNTRIAGED
       )).fetch()
 
   # Further filter potential duplicates and return them.
   return [analysis for analysis in analysis_results if
+          analysis.completed and
+          analysis.result and
           _DoAnalysesMatch(original_analysis, analysis) and
-          original_analysis.key is not analysis.key and
-          analysis.completed]
+          original_analysis.key is not analysis.key]
 
 
-def _TriageDuplicateResults(original_analysis, is_correct, user_name=None):
+def _TriageAndCountDuplicateResults(original_analysis, is_correct,
+                                    user_name=None):
   matching_analyses = _GetDuplicateAnalyses(original_analysis)
 
   for analysis in matching_analyses:
-    _AppendTriageHistoryRecord(analysis, is_correct, user_name)
+    analysis.triage_reference_analysis_master_name = (
+        original_analysis.master_name)
+    analysis.triage_reference_analysis_builder_name = (
+        original_analysis.builder_name)
+    analysis.triage_reference_analysis_build_number = (
+        original_analysis.build_number)
+    _AppendTriageHistoryRecord(analysis, is_correct, user_name,
+                               is_duplicate=True)
+
+  return len(matching_analyses)
 
 
 class TriageAnalysis(BaseHandler):
   PERMISSION_LEVEL = Permission.CORP_USER
 
   def HandleGet(self):  # pragma: no cover
     return self.HandlePost()
 
   def HandlePost(self):
     """Sets the manual triage result for the analysis.
 
     Mark the analysis result as correct/wrong/etc.
     TODO: make it possible to set the real culprit CLs.
     """
     url = self.request.get('url').strip()
     build_info = buildbot.ParseBuildUrl(url)
     if not build_info:
       return {'data': {'success': False}}
     master_name, builder_name, build_number = build_info
 
     is_correct = self.request.get('correct').lower() == 'true'
     # As the permission level is CORP_USER, we could assume the current user
     # already logged in.
     user_name = users.get_current_user().email().split('@')[0]
     success, original_analysis = _UpdateAnalysisResultStatus(
         master_name, builder_name, build_number, is_correct, user_name)
+    num_duplicate_analyses = 0
     if success:
-      _TriageDuplicateResults(original_analysis, is_correct, user_name)
-    return {'data': {'success': success}}
+      num_duplicate_analyses = _TriageAndCountDuplicateResults(
+          original_analysis, is_correct, user_name)
+    return {'data': {'success': success,
+                     'num_duplicate_analyses': num_duplicate_analyses}}