Fixes Caterpillar crashing while generating report.

src/end_to_end_test.py

Index: src/end_to_end_test.py
diff --git a/src/end_to_end_test.py b/src/end_to_end_test.py
index 8c3582154c9e9bb3cc253fe54c07e95a1a37b6bd..16d2099af8adc3f008f834ea381b684f005b1c37 100644
--- a/src/end_to_end_test.py
+++ b/src/end_to_end_test.py
@@ -68,42 +68,53 @@ class TestEndToEndConvert(caterpillar_test.TestCaseWithTempDir):
 
     if not os.path.exists(self.config_path):
       raise RuntimeError('Configuration file generation failed.')
+    if process.wait():
+      raise subprocess.CalledProcessError()
 
-    process = subprocess.Popen(
+    if not os.path.exists(self.config_path):
+      raise RuntimeError('Configuration file generation failed.')
+
+    output = subprocess.check_output(
         [CATERPILLAR_PATH, 'convert', '-c', self.config_path, self.input_dir,
-         self.output_dir],
-        stdin=subprocess.PIPE,
-        stdout=subprocess.PIPE,
-        stderr=subprocess.STDOUT)
-    process.wait()
+         self.output_dir])
 
-  def test_output_matches_reference(self):
-    """Tests that the output matches the reference output."""
-    expected_files = set()
-    for dirname, _, filenames in os.walk(TTS_REFERENCE_PATH):
-      for filename in filenames:
-        relpath = os.path.relpath(
-            os.path.join(dirname, filename), TTS_REFERENCE_PATH)
-        expected_files.add(relpath)
+  def get_relative_filepaths(self, directory, ignore_folders=None):
+    """Returns all relative filepaths in a directory and children.
 
-    for dirname, _, filenames in os.walk(self.output_dir):
-      for filename in filenames:
-        relpath = os.path.relpath(
-            os.path.join(dirname, filename), self.output_dir)
-        self.assertIn(relpath, expected_files)
+    Args:
+      directory: Path to directory.
+      ignore_folders: List of folder names to ignore.

[Matt Giuca, 2016/02/04 06:36:34]

directory / folder are interchangeable, so don't mix them. (Consistently use
"directory".)

Also make it a set.

[Matthew Alger, 2016/02/04 22:47:06]

Done.

 
-    expected_files = set()
-    for dirname, _, filenames in os.walk(self.output_dir):
-      for filename in filenames:
-        relpath = os.path.relpath(
-            os.path.join(dirname, filename), self.output_dir)
-        expected_files.add(relpath)
+    Returns:
+      set of relative filepaths.
+    """
+    if not ignore_folders:
+      ignore_folders = []
+
+    filepaths = set()
+    for dirname, _, filenames in os.walk(directory):
+      skip = False
+      for folder in ignore_folders:
+        if folder in dirname:

[Matt Giuca, 2016/02/04 06:36:34]

This is not precise enough. If ignore_folders = ['bower_components'], this will
match any path with "bower_components" anywhere in the string, including if it
is part of a directory name.

Is |ignore_folders| supposed to be any folder called that, or only folders whose
full path (relative to |directory|) matches?

Have a look at os.walk with topdown=True. The documentation there explicitly
mentions pruning the search. If |ignore_folders| is supposed to be any directory
name, you can just delete any item in |dirnames| (in the result from walk) that
is in ignore_folders, and you're done.

[Matthew Alger, 2016/02/04 22:47:06]

Oh, that's much cleaner. Done.

+          skip = True
+          break
+
+      if skip:
+        continue
 
-    for dirname, _, filenames in os.walk(TTS_REFERENCE_PATH):
       for filename in filenames:
-        relpath = os.path.relpath(
-            os.path.join(dirname, filename), TTS_REFERENCE_PATH)
-        self.assertIn(relpath, expected_files)
+        relpath = os.path.relpath(os.path.join(dirname, filename), directory)
+        filepaths.add(relpath)
+
+    return filepaths
+
+  def test_output_matches_reference(self):
+    """Tests that the output matches the reference output."""
+    expected_files = self.get_relative_filepaths(TTS_REFERENCE_PATH,
+                                                 ['bower_components'])
+    actual_files = self.get_relative_filepaths(self.output_dir,
+                                               ['bower_components'])

[Matt Giuca, 2016/02/04 06:36:34]

{'bower_components'}

[Matthew Alger, 2016/02/04 22:47:06]

Done.

+    self.assertEqual(expected_files, actual_files)
 
   def test_all_correct_contents(self):
     """Tests that the content of all non-static output files is expected."""