[Findit] Strip platform from step_name before triggering try job.

appengine/findit/waterfall/process_swarming_task_result_pipeline.py

Index: appengine/findit/waterfall/process_swarming_task_result_pipeline.py
diff --git a/appengine/findit/waterfall/process_swarming_task_result_pipeline.py b/appengine/findit/waterfall/process_swarming_task_result_pipeline.py
index d33dc8bd77a13fa408852fccb7a2b8a32cc34192..7d18d1d6ec8e5cbc4339fe63aa199f4b2bf12f85 100644
--- a/appengine/findit/waterfall/process_swarming_task_result_pipeline.py
+++ b/appengine/findit/waterfall/process_swarming_task_result_pipeline.py
@@ -85,12 +85,15 @@ class ProcessSwarmingTaskResultPipeline(BasePipeline):
     task_started = False
     task_completed = False
     tests_statuses = {}
+    step_name_no_platform = step_name

[stgao, 2016/03/10 23:50:28]

Should not default to step name for steps running on Swarming.

[chanli, 2016/03/11 00:49:41]

I think we need to fall back to step name just in case.

[stgao, 2016/03/11 01:03:20]

In which scenario do you think we need to fall back to step name?

[chanli, 2016/03/11 03:47:00]

Just for precaution. 

[stgao, 2016/03/11 06:42:58]

But the step names always includes the platform name now. That won't work.

 
     while not task_completed:
       # Keeps monitoring the swarming task, waits for it to complete.
       data = swarming_util.GetSwarmingTaskResultById(
           task_id, self.HTTP_CLIENT)
       task_state = data['state']
+      step_name_no_platform = swarming_util.GetTagValue(
+          data.get('tags', {}), 'ref_name') or step_name
       if task_state not in swarming_util.STATES_RUNNING:
         task_completed = True
         task = WfSwarmingTask.Get(
@@ -140,4 +143,4 @@ class ProcessSwarmingTaskResultPipeline(BasePipeline):
     task.completed_time = _ConvertDateTime(data.get('completed_ts'))
     task.put()
 
-    return step_name, task.classified_tests
+    return step_name, (step_name_no_platform, task.classified_tests)