raw_io module: Add text_input and text_output

recipe_modules/raw_io/example.py

 # Copyright 2014 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.
 
 DEPS = [
   'path',
   'properties',
   'python',
   'raw_io',
   'step',
 ]
 
 
 def RunSteps(api):
   # Read command's stdout and stderr.
   step_result = api.step('echo', ['echo', 'Hello World'],
       stdout=api.raw_io.output(),
       stderr=api.raw_io.output())
   assert step_result.stdout == 'Hello World\n'
   assert step_result.stderr == ''
 
   # Pass stuff to command's stdin, read it from stdout.
   step_result = api.step('cat', ['cat'],
-      stdin=api.raw_io.input(data='hello'),
+      stdin=api.raw_io.input_text(data='hello'),
       stdout=api.raw_io.output('out'))
   assert step_result.stdout == 'hello'
 
+  step_result = api.step('cat', ['cat', api.raw_io.input_text(data='hello')],
+      stdout=api.raw_io.output('out'))
+  assert step_result.stdout == 'hello'
+
+  # \xe2 is not encodable by utf-8 (and has shown up in actual recipe data)
+  # so test that input correctly doesn't try to encode it as utf-8.
+  step_result = api.step('cat', ['cat'],
+      stdin=api.raw_io.input(data='\xe2hello'),
+      stdout=api.raw_io.output())
+  assert step_result.stdout == '\xe2hello'
+
   # Example of auto-mocking stdout. '\n' appended to mock 'echo' behavior.
   step_result = api.step('automock', ['echo', 'huh'],
                  stdout=api.raw_io.output('out'),
                  step_test_data=(
                    lambda: api.raw_io.test_api.stream_output('huh\n')))
   assert step_result.stdout == 'huh\n'
 
   # Example of auto-mocking stdout + stderr.
   step_result = api.step(
     'automock (fail)', ['bash', '-c', 'echo blah && echo fail 1>&2'],
@@ skipping 21 matching lines @@
       leak_to=api.path['tmp_base'].join('out'))])
 
   # Example of overriding default mocked output for a single named output.
   step_result = api.python.inline(
       'override_default_mock',
       """
       import sys
       with open(sys.argv[1], 'w') as f:
         f.write(%r)
       """ % api.properties.get('some_prop', 'good_value'),
-      args=[api.raw_io.output(name='test')],
+      args=[api.raw_io.output_text(name='test')],
       step_test_data=(
-          lambda: api.raw_io.test_api.output('second_bad_value', name='test')))
-  assert step_result.raw_io.outputs['test'] == 'good_value'
-  assert step_result.raw_io.output == 'good_value'
+          lambda: api.raw_io.test_api.output_text(
+              'second_bad_value', name='test')))
+  assert step_result.raw_io.output_texts['test'] == 'good_value'
+  assert step_result.raw_io.output_text == 'good_value'
 
 
 def GenTests(api):
   # This test shows that you can override a specific placeholder, even with
   # default `step_test_data`. However, since this recipe is ACTUALLY run in
   # the presubmit, we need to do a trick with properties:
   #   When run for real, "some_prop" will be "good_value" and pass.
   #   When run for simulation, we override this property to provide a bad value,
   #     AND the default step_test_data in RunSteps above ALSO provides another
   #     bad value, the simulation passes ONLY because of the
   #     'override_default_mock' below.
   yield (api.test('basic') +
       api.properties(some_prop='bad_value') +
       api.step_data('echo',
           stdout=api.raw_io.output('Hello World\n'),
           stderr=api.raw_io.output('')) +
       api.step_data('cat',
           stdout=api.raw_io.output('hello')) +
+      api.step_data('cat (2)',
+          stdout=api.raw_io.output('hello')) +
+      api.step_data('cat (3)',
+          stdout=api.raw_io.output('\xe2hello')) +
       api.step_data('override_default_mock',
-          api.raw_io.output('good_value', name='test'))
+          api.raw_io.output_text('good_value', name='test'))
   )