Update subprocess2.check_output() to behave like subprocess.check_output().

presubmit_canned_checks.py

 # Copyright (c) 2011 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.
 
 """Generic presubmit checks that can be reused by other presubmit checks."""
 
 
 ### Description checks
 
 def CheckChangeHasTestField(input_api, output_api):
@@ skipping 497 matching lines @@
       cmd = [input_api.python_executable]
     cmd.append(unit_test)
     if input_api.verbose:
       print('Running %s' % unit_test)
       cmd.append('--verbose')
     try:
       if input_api.verbose:
         input_api.subprocess.check_call(cmd, cwd=input_api.PresubmitLocalPath())
       else:
         input_api.subprocess.check_output(
-            cmd, cwd=input_api.PresubmitLocalPath())
+            cmd,
+            stderr=input_api.subprocess.STDOUT,
+            cwd=input_api.PresubmitLocalPath())
     except (OSError, input_api.subprocess.CalledProcessError), e:
       results.append(message_type('%s failed!\n%s' % (unit_test, e)))
   return results
 
 
 def RunPythonUnitTests(input_api, output_api, unit_tests):
   """Run the unit tests out of process, capture the output and use the result
   code to determine success.
 
   DEPRECATED.
@@ skipping 22 matching lines @@
       env = input_api.environ.copy()
       # At least on Windows, it seems '.' must explicitly be in PYTHONPATH
       backpath = [
           '.', input_api.os_path.pathsep.join(['..'] * (cwd.count('/') + 1))
         ]
       if env.get('PYTHONPATH'):
         backpath.append(env.get('PYTHONPATH'))
       env['PYTHONPATH'] = input_api.os_path.pathsep.join((backpath))
     cmd = [input_api.python_executable, '-m', '%s' % unit_test]
     try:
-      input_api.subprocess.check_output(cmd, cwd=cwd, env=env)
+      input_api.subprocess.check_output(
+          cmd, stderr=input_api.subprocess.STDOUT, cwd=cwd, env=env)
     except (OSError, input_api.subprocess.CalledProcessError), e:
       results.append(message_type('%s failed!\n%s' % (unit_test_name, e)))
   return results
 
 
 def _FetchAllFiles(input_api, white_list, black_list):
   """Hack to fetch all files."""
   # We cannot use AffectedFiles here because we want to test every python
   # file on each single python change. It's because a change in a python file
   # can break another unmodified file.
@@ skipping 367 matching lines @@
     results.extend(input_api.canned_checks.CheckSvnForCommonMimeTypes(
         input_api, output_api))
     snapshot("checking license")
     results.extend(input_api.canned_checks.CheckLicense(
         input_api, output_api, license_header, source_file_filter=sources))
     snapshot("checking was uploaded")
     results.extend(input_api.canned_checks.CheckChangeWasUploaded(
         input_api, output_api))
   snapshot("done")
   return results