Use a new caching algorithm which can limit the size for the build nodes. Stop treating public images as special. Add a new phase to the builder for pulling.

util/dockerfileparse.py

@@ -4,23 +4,27 @@ LINE_CONTINUATION_REGEX = re.compile('\s*\\\s*\n')
 COMMAND_REGEX = re.compile('([A-Za-z]+)\s(.*)')
 
 COMMENT_CHARACTER = '#'
+LATEST_TAG = 'latest'
 
 class ParsedDockerfile(object):
   def __init__(self, commands):
     self.commands = commands
 
-  def get_commands_of_kind(self, kind):
+  def _get_commands_of_kind(self, kind):
     return [command for command in self.commands if command['command'] == kind]
 
-  def get_base_image(self):
-    image_and_tag = self.get_base_image_and_tag()
-    if not image_and_tag:
+  def _get_from_image_identifier(self):
+    from_commands = self._get_commands_of_kind('FROM')
+    if not from_commands:
       return None
 
-    return self.base_image_from_repo_identifier(image_and_tag)
+    return from_commands[-1]['parameters']
 
   @staticmethod
-  def base_image_from_repo_identifier(image_and_tag):
+  def parse_image_identifier(image_identifier):
+    """ Parses a docker image identifier, and returns a tuple of image name and tag, where the tag
+        is filled in with "latest" if left unspecified.
+    """
     # Note:
     # Dockerfile images references can be of multiple forms:
     #   server:port/some/path
@@ -28,29 +32,34 @@ class ParsedDockerfile(object):
     #   server/some/path
     #   server/some/path:tag
     #   server:port/some/path:tag
-    parts = image_and_tag.strip().split(':')
+    parts = image_identifier.strip().split(':')
 
     if len(parts) == 1:
       # somepath
-      return parts[0]
+      return (parts[0], LATEST_TAG)
 
     # Otherwise, determine if the last part is a port
     # or a tag.
     if parts[-1].find('/') >= 0:
       # Last part is part of the hostname.
-      return image_and_tag
+      return (image_identifier, LATEST_TAG)
 
     # Remaining cases:
     #   server/some/path:tag
     #   server:port/some/path:tag    
-    return ':'.join(parts[0:-1])
+    return (':'.join(parts[0:-1]), parts[-1])    
 
-  def get_base_image_and_tag(self):
-    from_commands = self.get_commands_of_kind('FROM')
-    if not from_commands:
-      return None
+  def get_base_image(self):
+    """ Return the base image without the tag name. """
+    return self.get_image_and_tag()[0]
 
-    return from_commands[-1]['parameters']
+  def get_image_and_tag(self):
+    """ Returns the image and tag from the FROM line of the dockerfile. """
+    image_identifier = self._get_from_image_identifier()
+    if image_identifier is None:
+      return (None, None)
+
+    return self.parse_image_identifier(image_identifier)
 
 
 def strip_comments(contents):