json: Improve error handling and adding ability to limit log output.

recipe_modules/json/test_api.py

 # Copyright 2015 The LUCI Authors. All rights reserved.
 # Use of this source code is governed under the Apache License, Version 2.0
 # that can be found in the LICENSE file.
 
 import json
 
 from recipe_engine import recipe_test_api
 
 class JsonTestApi(recipe_test_api.RecipeTestApi):
   @recipe_test_api.placeholder_step_data
   @staticmethod
   def output(data, retcode=None, name=None):
     """Supplies placeholder data for a json.output. `data` should be a jsonish
     python object (e.g. dict, list, str, bool, int, etc). It will be dumped out
     with json.dumps and the step will be observed to return that dumped value.
     """
     return json.dumps(data), retcode, name
 
   def invalid(self, raw_data_str, retcode=None, name=None):
     """Can be used to supply data for a json.output, except that it takes a raw
     string rather than a json object."""
     ret = recipe_test_api.StepTestData()
     ret.retcode=retcode
     placeholder_data = recipe_test_api.PlaceholderTestData(
       data=raw_data_str, name=name)
     ret.placeholder_data[(self._module.NAME, 'output', name)] = placeholder_data
     return ret
 
+  def not_found(self, retcode=None, name=None):
+    """Can be used to have the json.output file be missing."""
+    ret = recipe_test_api.StepTestData()
+    ret.retcode=retcode
+    # FIXME: How do I get None into the data?

[Paweł Hajdan Jr., 2017/03/23 12:22:12]

What happens if you just use None below instead of '' ?

[iannucci, 2017/03/23 19:44:13]

That's what I would expect as well... however I would also expect that passing
None to invalid() above will also work?

+    ret.placeholder_data[(self._module.NAME, 'output', name)] = ''
+    return ret
+
   def output_stream(self, data, stream='stdout', retcode=None, name=None):
     assert stream in ('stdout', 'stderr')
     ret = recipe_test_api.StepTestData()
     step_data = self.output(data, retcode=retcode, name=name)
     setattr(ret, stream, step_data.unwrap_placeholder())
     return ret