check: Make item stream resync a bit more robust

attic/archive.py

@@ -431,6 +431,10 @@ class RobustUnpacker():
             while self._resync:
                 if not data:
                     raise StopIteration
+                # Abort early if the data does not look like a serialized item
+                if len(data) < 2 or ((data[0] & 0xf0) != 0x80) or ((data[1] & 0xe0) != 0xa0) or not b'\xa4path' in data:
+                    data = data[1:]
+                    continue
                 self._unpacker = msgpack.Unpacker(object_hook=StableDict)
                 self._unpacker.feed(data)
                 try:

attic/testsuite/archive.py

@@ -35,10 +35,10 @@ class ChunkBufferTestCase(AtticTestCase):
 class RobustUnpackerTestCase(AtticTestCase):
 
     def make_chunks(self, items):
-        return b''.join(msgpack.packb(item) for item in items)
+        return b''.join(msgpack.packb({'path': item}) for item in items)
 
     def _validator(self, value):
-        return value in (b'foo', b'bar', b'boo', b'baz')
+        return isinstance(value, dict) and value.get(b'path') in (b'foo', b'bar', b'boo', b'baz')
 
     def process(self, input):
         unpacker = RobustUnpacker(validator=self._validator)
@@ -56,7 +56,11 @@ class RobustUnpackerTestCase(AtticTestCase):
         chunks = [(False, [self.make_chunks([b'foo', b'bar'])]),
                   (False, [b'garbage'] + [self.make_chunks([b'boo', b'baz'])])]
         result = self.process(chunks)
-        self.assert_equal(result, [b'foo', b'bar', 103, 97, 114, 98, 97, 103, 101, b'boo', b'baz'])
+        self.assert_equal(result, [
+            {b'path': b'foo'}, {b'path': b'bar'},
+            103, 97, 114, 98, 97, 103, 101,
+            {b'path': b'boo'},
+            {b'path': b'baz'}])
 
     def split(self, left, length):
         parts = []
@@ -69,16 +73,16 @@ class RobustUnpackerTestCase(AtticTestCase):
         chunks = self.split(self.make_chunks([b'foo', b'bar', b'boo', b'baz']), 2)
         input = [(False, chunks)]
         result = self.process(input)
-        self.assert_equal(result, [b'foo', b'bar', b'boo', b'baz'])
+        self.assert_equal(result, [{b'path': b'foo'}, {b'path': b'bar'}, {b'path': b'boo'}, {b'path': b'baz'}])
 
     def test_missing_chunk(self):
-        chunks = self.split(self.make_chunks([b'foo', b'bar', b'boo', b'baz']), 2)
+        chunks = self.split(self.make_chunks([b'foo', b'bar', b'boo', b'baz']), 4)
         input = [(False, chunks[:3]), (True, chunks[4:])]
         result = self.process(input)
-        self.assert_equal(result, [b'foo', b'boo', b'baz'])
+        self.assert_equal(result, [{b'path': b'foo'}, {b'path': b'boo'}, {b'path': b'baz'}])
 
     def test_corrupt_chunk(self):
-        chunks = self.split(self.make_chunks([b'foo', b'bar', b'boo', b'baz']), 2)
+        chunks = self.split(self.make_chunks([b'foo', b'bar', b'boo', b'baz']), 4)
         input = [(False, chunks[:3]), (True, [b'gar', b'bage'] + chunks[3:])]
         result = self.process(input)
-        self.assert_equal(result, [b'foo', b'boo', b'baz'])
+        self.assert_equal(result, [{b'path': b'foo'}, {b'path': b'boo'}, {b'path': b'baz'}])