Handle invalid tar layers and add tests

test/test_streamlayerformat.py

@@ -3,6 +3,7 @@ import tarfile
 
 from StringIO import StringIO
 from util.streamlayerformat import StreamLayerMerger, AUFS_WHITEOUT
+from util.tarlayerformat import TarLayerReadException
 
 class TestStreamLayerMerger(unittest.TestCase):
   def create_layer(self, **kwargs):
@@ -30,6 +31,9 @@ class TestStreamLayerMerger(unittest.TestCase):
 
     return output.getvalue()
 
+  def create_empty_layer(self):
+    return ''
+
   def squash_layers(self, layers):
     def get_layers():
       return [StringIO(layer) for layer in layers]
@@ -337,6 +341,7 @@ class TestStreamLayerMerger(unittest.TestCase):
     self.assertHasFile(squashed, 'foobar/baz/some_file', 'foo')
     self.assertDoesNotHaveFile(squashed, 'foo/another_file')
 
+
   def test_delete_root_directory(self):
     third_layer = self.create_layer(
       foo = 'build/first_file',
@@ -350,5 +355,45 @@ class TestStreamLayerMerger(unittest.TestCase):
     self.assertDoesNotHaveFile(squashed, 'build/first_file')
     self.assertDoesNotHaveFile(squashed, 'build/second_file')
 
+
+  def test_tar_empty_layer(self):
+    third_layer = self.create_layer(
+      foo = 'build/first_file',
+      bar = 'build/second_file')
+
+    empty_layer = self.create_layer()
+
+    squashed = self.squash_layers([empty_layer, third_layer])
+
+    self.assertHasFile(squashed, 'build/first_file', 'foo')
+    self.assertHasFile(squashed, 'build/second_file', 'bar')
+
+
+  def test_data_empty_layer(self):
+    third_layer = self.create_layer(
+      foo = 'build/first_file',
+      bar = 'build/second_file')
+
+    empty_layer = self.create_empty_layer()
+
+    squashed = self.squash_layers([empty_layer, third_layer])
+
+    self.assertHasFile(squashed, 'build/first_file', 'foo')
+    self.assertHasFile(squashed, 'build/second_file', 'bar')
+
+
+  def test_broken_layer(self):
+    third_layer = self.create_layer(
+      foo = 'build/first_file',
+      bar = 'build/second_file')
+
+    broken_layer = 'not valid data'
+
+    try:
+      self.squash_layers([broken_layer, third_layer])
+      self.fail('Expected exception')
+    except TarLayerReadException as ex:
+      self.assertEquals('Could not read layer', ex.message)
+
 if __name__ == '__main__':
   unittest.main()

util/tarlayerformat.py

@@ -1,6 +1,11 @@
 import os
 import tarfile
 
+class TarLayerReadException(Exception):
+  """ Exception raised when reading a layer has failed. """
+  pass
+  
+
 class TarLayerFormat(object):
   """ Class which creates a generator of the combined TAR data. """
   def __init__(self, tar_iterator):
@@ -10,9 +15,14 @@ class TarLayerFormat(object):
     for current_tar in self.tar_iterator():
       # Read the current TAR. If it is empty, we just continue
       # to the next one.
+      tar_file = None
       try:
         tar_file = tarfile.open(mode='r|*', fileobj=current_tar)
       except tarfile.ReadError as re:
+        if re.message != 'empty file':
+          raise TarLayerReadException('Could not read layer')
+
+      if not tar_file:
         continue
 
       # For each of the tar entries, yield them IF and ONLY IF we have not