- Make the layer going over the estimated size raise an exception

- Add a heuristic for estimating the layer size if it is 0
- Add a method where we can add a custom defined map of image -> size

util/dockerloadformat.py

@@ -6,6 +6,14 @@ import copy
 import json
 import tarfile
 
+class FileEstimationException(Exception):
+  """ Exception raised by build_docker_load_stream if the estimated size of the layer TAR
+      was lower than the actual size. This means the sent TAR header is wrong, and we have
+      to fail.
+  """
+  pass
+
+
 def build_docker_load_stream(namespace, repository, tag, synthetic_image_id,
                              layer_json, get_image_iterator, get_layer_iterator):
   """ Builds and streams a synthetic .tar.gz that represents a squashed version
@@ -50,7 +58,9 @@ def _import_format_generator(namespace, repository, tag, synthetic_image_id,
   # Yield the merged layer data's header.
   estimated_file_size = 0
   for image in get_image_iterator():
-    estimated_file_size += image.storage.uncompressed_size or 0
+    estimated_file_size += (image.storage.uncompressed_size or 
+                            _get_mapped_size(image) or 
+                            _estimate_size(image))
 
   yield _tar_file_header(synthetic_image_id + '/layer.tar', estimated_file_size)
 
@@ -60,6 +70,11 @@ def _import_format_generator(namespace, repository, tag, synthetic_image_id,
     yield entry
     yielded_size += len(entry)
 
+  # If the yielded size is more than the estimated size (which is unlikely but possible), then
+  # raise an exception since the tar header will be wrong.
+  if yielded_size > estimated_file_size:
+    raise FileEstimationException()
+
   # If the yielded size is less than the estimated size (which is likely), fill the rest with
   # zeros.
   if yielded_size < estimated_file_size:
@@ -113,3 +128,13 @@ def _tar_folder(name):
   info = tarfile.TarInfo(name=name)
   info.type = tarfile.DIRTYPE
   return info.tobuf()
+
+def _get_mapped_size(image):
+  """ Returns a predefined image size for the given image or None if not found. """
+  return None
+
+def _estimate_size(image):
+  """ Estimates a file size based on a heuristic. """
+  # More than 1 SD away from the size difference in the DB, as of 9/29/2014
+  return image.storage.image_size * 12
+