Fix Dockerfile parsing for unicode and add testing

Fixes #423

test/test_dockerfileparse.py

@@ -0,0 +1,66 @@
+# -*- coding: utf-8 -*-
+
+import unittest
+
+from util.dockerfileparse import parse_dockerfile
+
+class TestParsedDockerfile(unittest.TestCase):
+  def test_basic_parse(self):
+    parsed = parse_dockerfile("""
+      FROM someimage:latest
+      RUN dosomething
+    """)
+
+    self.assertEquals(("someimage", "latest"), parsed.get_image_and_tag())
+    self.assertEquals("someimage", parsed.get_base_image())
+
+  def test_basic_parse_notag(self):
+    parsed = parse_dockerfile("""
+      FROM someimage
+      RUN dosomething
+    """)
+
+    self.assertEquals(("someimage", "latest"), parsed.get_image_and_tag())
+    self.assertEquals("someimage", parsed.get_base_image())
+
+  def test_two_from_lines(self):
+    parsed = parse_dockerfile("""
+      FROM someimage:latest
+      FROM secondimage:second
+    """)
+
+    self.assertEquals(("secondimage", "second"), parsed.get_image_and_tag())
+    self.assertEquals("secondimage", parsed.get_base_image())
+
+  def test_parse_comments(self):
+    parsed = parse_dockerfile("""
+      # FROM someimage:latest
+      FROM anotherimage:foobar # This is a comment
+      RUN dosomething
+    """)
+
+    self.assertEquals(("anotherimage", "foobar"), parsed.get_image_and_tag())
+    self.assertEquals("anotherimage", parsed.get_base_image())
+
+  def test_unicode_parse_as_ascii(self):
+    parsed = parse_dockerfile("""
+      FROM someimage:latest
+      MAINTAINER José Schorr <jschorr@whatever.com>
+    """)
+
+    self.assertEquals(("someimage", "latest"), parsed.get_image_and_tag())
+    self.assertEquals("someimage", parsed.get_base_image())
+
+
+  def test_unicode_parse_as_unicode(self):
+    parsed = parse_dockerfile("""
+      FROM someimage:latest
+      MAINTAINER José Schorr <jschorr@whatever.com>
+    """.decode('utf-8'))
+
+    self.assertEquals(("someimage", "latest"), parsed.get_image_and_tag())
+    self.assertEquals("someimage", parsed.get_base_image())
+
+if __name__ == '__main__':
+  unittest.main()
+

tools/reparsedockerfile.py

@@ -1,26 +0,0 @@
-from util.dockerfileparse import parse_dockerfile, ParsedDockerfile, serialize_dockerfile
-
-with open('Dockerfile.test', 'r') as dockerfileobj:
-  parsed_dockerfile = parse_dockerfile(dockerfileobj.read())
-
-quay_reponame = 'something'
-env_command = {
-  'command': 'ENV',
-  'parameters': 'QUAY_REPOSITORY %s' % quay_reponame
-}
-
-for index, command in reversed(list(enumerate(parsed_dockerfile.commands))):
-  if command['command'] == 'FROM':
-    new_command_index = index + 1
-    parsed_dockerfile.commands.insert(new_command_index, env_command)
-    break
-
-image_and_tag_tuple = parsed_dockerfile.get_image_and_tag()
-print image_and_tag_tuple
-if image_and_tag_tuple is None or image_and_tag_tuple[0] is None:
-  raise Exception('Missing FROM command in Dockerfile')
-
-print serialize_dockerfile(parsed_dockerfile)
-
-with open('Dockerfile.test.out', 'w') as dockerfileobj:
-  dockerfileobj.write(serialize_dockerfile(parsed_dockerfile))

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()
+      command = match_command.group(1).upper()
-      parameters = match_command.group(2).decode('utf-8')
+      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])