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..3318a38b82b5f076a54a4df2e8c12b833c413dff 100644
--- a/src/end_to_end_test.py
+++ b/src/end_to_end_test.py
@@ -68,42 +68,50 @@ 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_directories=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_directories: Set of directory names to ignore. Note that all
+        children will be ignored too.
 
-    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_directories:
+      ignore_directories = set()
+
+    filepaths = set()
+    for dirname, dirnames, filenames in os.walk(directory, topdown=True):
+      basename = os.path.basename(os.path.normpath(dirname))

[Matt Giuca, 2016/02/05 00:23:49]

This is descending into the ignored directory, and then deleting all of its
subdirectories.

Instead, you should just delete the ignored directory when it shows up in its
parent (then you can just directory compare directory names without having to
mess around with normpath and basename).

# Do not descend into ignored directories.
for i in range(len(dirnames) - 1, -1, -1):
  if dirnames[i] in ignore_directories:
    del dirnames[i]

I think this is the intended usage of os.walk(topdown=true).

+      if basename in ignore_directories:
+        dirnames[:] = []
+        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'})
+    self.assertEqual(expected_files, actual_files)
 
   def test_all_correct_contents(self):
     """Tests that the content of all non-static output files is expected."""