Fix Dockerfile parsing for unicode and add testing

Fixes #423

util/dockerfileparse.py

@@ -63,7 +63,16 @@ class ParsedDockerfile(object):
 
 
 def strip_comments(contents):
-  lines = [line for line in contents.split('\n') if not line.startswith(COMMENT_CHARACTER)]
+  lines = []
+  for line in contents.split('\n'):
+    index = line.find(COMMENT_CHARACTER)
+    if index < 0:
+      lines.append(line)
+      continue
+
+    line = line[:index]
+    lines.append(line)
+
   return '\n'.join(lines)
 
 
@@ -72,15 +81,22 @@ def join_continued_lines(contents):
 
 
 def parse_dockerfile(contents):
+  # If we receive ASCII, translate into unicode.
+  try:
+    contents = contents.decode('utf-8')
+  except ValueError:
+    # Already unicode or unable to convert.
+    pass
+
   contents = join_continued_lines(strip_comments(contents))
-  lines = [line for line in contents.split('\n') if len(line) > 0]
+  lines = [line.strip() for line in contents.split('\n') if len(line) > 0]
 
   commands = []
   for line in lines:
     match_command = COMMAND_REGEX.match(line)
     if match_command:
-      command = match_command.group(1).decode('utf-8').upper()
-      parameters = match_command.group(2).decode('utf-8')
+      command = match_command.group(1).upper()
+      parameters = match_command.group(2)
 
       commands.append({
         'command': command,
@@ -88,9 +104,3 @@ def parse_dockerfile(contents):
       })
 
   return ParsedDockerfile(commands)
-
-
-def serialize_dockerfile(parsed_dockerfile):
-  return '\n'.join([' '.join([command['command'].encode('utf-8'),
-                              command['parameters'].encode('utf-8')])
-                    for command in parsed_dockerfile.commands])