Refactor the util directory to use subpackages.

util/registry/streamlayerformat.py

@@ -0,0 +1,53 @@
+import marisa_trie
+import os
+from util.registry.aufs import is_aufs_metadata, get_deleted_prefix
+from util.registry.tarlayerformat import TarLayerFormat
+
+AUFS_METADATA = u'.wh..wh.'
+
+AUFS_WHITEOUT = u'.wh.'
+AUFS_WHITEOUT_PREFIX_LENGTH = len(AUFS_WHITEOUT)
+
+class StreamLayerMerger(TarLayerFormat):
+  """ Class which creates a generator of the combined TAR data for a set of Docker layers. """
+  def __init__(self, layer_iterator, path_prefix=None):
+    super(StreamLayerMerger, self).__init__(layer_iterator, path_prefix)
+
+    self.path_trie = marisa_trie.Trie()
+    self.path_encountered = []
+
+    self.prefix_trie = marisa_trie.Trie()
+    self.prefix_encountered = []
+
+  def after_tar_layer(self, current_layer):
+      # Update the tries.
+      self.path_trie = marisa_trie.Trie(self.path_encountered)
+      self.prefix_trie = marisa_trie.Trie(self.prefix_encountered)
+
+  def check_tar_info(self, tar_info):
+    absolute = os.path.relpath(tar_info.name.decode('utf-8'), './')
+
+    # Skip metadata.
+    if is_aufs_metadata(absolute):
+      return False
+
+    # Add any prefix of deleted paths to the prefix list.
+    deleted_prefix = get_deleted_prefix(absolute)
+    if deleted_prefix is not None:
+      self.prefix_encountered.append(deleted_prefix)
+      return False
+
+    # Check if this file has already been encountered somewhere. If so,
+    # skip it.
+    ubsolute = unicode(absolute)
+    if ubsolute in self.path_trie:
+      return False
+
+    # Check if this file is under a deleted path.
+    for prefix in self.prefix_trie.iter_prefixes(ubsolute):
+      if not os.path.relpath(ubsolute, prefix).startswith('..'):
+        return False
+
+    # Otherwise, add the path to the encountered list and return it.
+    self.path_encountered.append(absolute)
+    return True