support legacy compressed properties back and forth (googleapis#183)

* support legacy compressed properties back and forth

google/cloud/ndb/model.py

@@ -324,7 +324,13 @@ class Person(Model):
 
 
 _MEANING_PREDEFINED_ENTITY_USER = 20
-_MEANING_URI_COMPRESSED = "ZLIB"
+_MEANING_COMPRESSED = 22
+
+# As produced by zlib. Indicates compressed byte sequence using DEFLATE at
+# default compression level, with a 32K window size.
+# From https://github.com/madler/zlib/blob/master/doc/rfc1950.txt
+_ZLIB_COMPRESSION_MARKER = b"x\x9c"
+
 _MAX_STRING_LENGTH = 1500
 Key = key_module.Key
 BlobKey = _datastore_types.BlobKey
@@ -627,6 +633,8 @@ def new_entity(key):
             else:
                 value = _BaseValue(value)
 
+        value = prop._from_datastore(ds_entity, value)
+
         prop._store_value(entity, value)
 
     return entity
@@ -721,6 +729,13 @@ def _entity_to_ds_entity(entity, set_key=True):
         ds_entity = ds_entity_module.Entity(
             exclude_from_indexes=exclude_from_indexes
         )
+
+    # Some properties may need to set meanings for backwards compatibility,
+    # so we look for them. They are set using the _to_datastore calls above.
+    meanings = data.pop("_meanings", None)
+    if meanings is not None:
+        ds_entity._meanings = meanings
+
     ds_entity.update(data)
 
     return ds_entity
@@ -2034,6 +2049,25 @@ def _to_datastore(self, entity, data, prefix="", repeated=False):
 
         return (key,)
 
+    def _from_datastore(self, ds_entity, value):
+        """Helper to convert property value from Datastore serializable data.
+
+        Called to modify the value of a property during deserialization from
+        storage. Subclasses (like BlobProperty) may need to override the
+        default behavior, which is simply to return the received value without
+        modification.
+
+        Args:
+            ds_entity (~google.cloud.datastore.Entity): The Datastore entity to
+                convert.
+            value (_BaseValue): The stored value of this property for the
+                entity being deserialized.
+
+        Return:
+            value [Any]: The transformed value.
+        """
+        return value
+
 
 def _validate_key(value, entity=None):
     """Validate a key.
@@ -2414,11 +2448,48 @@ def _from_base_type(self, value):
             decompressed.
         """
         if self._compressed and not isinstance(value, _CompressedValue):
+            if not value.startswith(_ZLIB_COMPRESSION_MARKER):
+                value = zlib.compress(value)
             value = _CompressedValue(value)
 
         if isinstance(value, _CompressedValue):
             return zlib.decompress(value.z_val)
 
+    def _to_datastore(self, entity, data, prefix="", repeated=False):
+        """Override of :method:`Property._to_datastore`.
+
+        If this is a compressed property, we need to set the backwards-
+        compatible `_meanings` field, so that it can be properly read later.
+        """
+        keys = super(BlobProperty, self)._to_datastore(
+            entity, data, prefix=prefix, repeated=repeated
+        )
+        if self._compressed:
+            value = data[self._name]
+            if isinstance(value, _CompressedValue):
+                value = value.z_val
+                data[self._name] = value
+            if not value.startswith(_ZLIB_COMPRESSION_MARKER):
+                value = zlib.compress(value)
+                data[self._name] = value
+            data.setdefault("_meanings", {})[self._name] = (
+                _MEANING_COMPRESSED,
+                value,
+            )
+        return keys
+
+    def _from_datastore(self, ds_entity, value):
+        """Override of :method:`Property._from_datastore`.
+
+        Need to check the ds_entity for a compressed meaning that would
+        indicate we are getting a compressed value.
+        """
+        if self._name in ds_entity._meanings:
+            meaning = ds_entity._meanings[self._name][0]
+            if meaning == _MEANING_COMPRESSED and not self._compressed:
+                value.b_val = zlib.decompress(value.b_val)
+        return value
+
     def _db_set_compressed_meaning(self, p):
         """Helper for :meth:`_db_set_value`.
 

tests/system/conftest.py

@@ -91,6 +91,23 @@ def make_entity(*key_args, **entity_kwargs):
     yield make_entity
 
 
+@pytest.fixture
+def ds_entity_with_meanings(with_ds_client, dispose_of):
+    def make_entity(*key_args, **entity_kwargs):
+        meanings = key_args[0]
+        key = with_ds_client.key(*key_args[1:])
+        assert with_ds_client.get(key) is None
+        entity = datastore.Entity(key=key, exclude_from_indexes=("blob",))
+        entity._meanings = meanings
+        entity.update(entity_kwargs)
+        with_ds_client.put(entity)
+        dispose_of(key)
+
+        return entity
+
+    yield make_entity
+
+
 @pytest.fixture
 def dispose_of(with_ds_client, to_delete):
     def delete_entity(ds_key):

tests/system/test_crud.py

@@ -20,6 +20,7 @@
 import operator
 import os
 import threading
+import zlib
 
 from unittest import mock
 
@@ -315,6 +316,28 @@ class SomeKind(ndb.Model):
     dispose_of(key._key)
 
 
+@pytest.mark.usefixtures("client_context")
+def test_retrieve_entity_with_legacy_compressed_property(
+    ds_entity_with_meanings
+):
+    class SomeKind(ndb.Model):
+        blob = ndb.BlobProperty()
+
+    value = b"abc" * 1000
+    compressed_value = zlib.compress(value)
+    entity_id = test_utils.system.unique_resource_id()
+    ds_entity_with_meanings(
+        {"blob": (22, compressed_value)},
+        KIND,
+        entity_id,
+        **{"blob": compressed_value}
+    )
+
+    key = ndb.Key(KIND, entity_id)
+    retrieved = key.get()
+    assert retrieved.blob == value
+
+
 @pytest.mark.usefixtures("client_context")
 def test_large_pickle_property(dispose_of, ds_client):
     class SomeKind(ndb.Model):

tests/unit/test_model.py

@@ -1722,12 +1722,19 @@ def test__from_base_type():
         assert converted == original
 
     @staticmethod
-    def test__from_base_type_no_compressed_value():
-        prop = model.BlobProperty(name="blob")
+    def test__from_base_type_no_compressed_value_uncompressed():
+        prop = model.BlobProperty(name="blob", compressed=True)
         original = b"abc" * 10
-        value = zlib.compress(original)
-        prop._compressed = True
-        converted = prop._from_base_type(value)
+        converted = prop._from_base_type(original)
+
+        assert converted == original
+
+    @staticmethod
+    def test__from_base_type_no_compressed_value_compressed():
+        prop = model.BlobProperty(name="blob", compressed=True)
+        original = b"abc" * 10
+        z_val = zlib.compress(original)
+        converted = prop._from_base_type(z_val)
 
         assert converted == original
 
@@ -1761,6 +1768,100 @@ def test__db_get_value():
         with pytest.raises(NotImplementedError):
             prop._db_get_value(None, None)
 
+    @staticmethod
+    @pytest.mark.usefixtures("in_context")
+    def test__to_datastore_compressed():
+        class ThisKind(model.Model):
+            foo = model.BlobProperty(compressed=True)
+
+        uncompressed_value = b"abc" * 1000
+        compressed_value = zlib.compress(uncompressed_value)
+        entity = ThisKind(foo=uncompressed_value)
+        ds_entity = model._entity_to_ds_entity(entity)
+        assert "foo" in ds_entity._meanings
+        assert ds_entity._meanings["foo"][0] == model._MEANING_COMPRESSED
+        assert ds_entity._meanings["foo"][1] == compressed_value
+
+    @staticmethod
+    @pytest.mark.usefixtures("in_context")
+    def test__to_datastore_uncompressed():
+        class ThisKind(model.Model):
+            foo = model.BlobProperty(compressed=False)
+
+        uncompressed_value = b"abc"
+        entity = ThisKind(foo=uncompressed_value)
+        ds_entity = model._entity_to_ds_entity(entity)
+        assert "foo" not in ds_entity._meanings
+
+    @staticmethod
+    @pytest.mark.usefixtures("in_context")
+    def test__from_datastore_compressed_to_uncompressed():
+        class ThisKind(model.Model):
+            foo = model.BlobProperty(compressed=False)
+
+        key = datastore.Key("ThisKind", 123, project="testing")
+        datastore_entity = datastore.Entity(key=key)
+        uncompressed_value = b"abc" * 1000
+        compressed_value = zlib.compress(uncompressed_value)
+        datastore_entity.update({"foo": compressed_value})
+        meanings = {"foo": (model._MEANING_COMPRESSED, compressed_value)}
+        datastore_entity._meanings = meanings
+        protobuf = helpers.entity_to_protobuf(datastore_entity)
+        entity = model._entity_from_protobuf(protobuf)
+        assert entity.foo == uncompressed_value
+        ds_entity = model._entity_to_ds_entity(entity)
+        assert ds_entity["foo"] == uncompressed_value
+
+    @staticmethod
+    @pytest.mark.usefixtures("in_context")
+    def test__from_datastore_compressed_to_compressed():
+        class ThisKind(model.Model):
+            foo = model.BlobProperty(compressed=True)
+
+        key = datastore.Key("ThisKind", 123, project="testing")
+        datastore_entity = datastore.Entity(key=key)
+        uncompressed_value = b"abc" * 1000
+        compressed_value = zlib.compress(uncompressed_value)
+        datastore_entity.update({"foo": compressed_value})
+        meanings = {"foo": (model._MEANING_COMPRESSED, compressed_value)}
+        datastore_entity._meanings = meanings
+        protobuf = helpers.entity_to_protobuf(datastore_entity)
+        entity = model._entity_from_protobuf(protobuf)
+        ds_entity = model._entity_to_ds_entity(entity)
+        assert ds_entity["foo"] == compressed_value
+
+    @staticmethod
+    @pytest.mark.usefixtures("in_context")
+    def test__from_datastore_uncompressed_to_uncompressed():
+        class ThisKind(model.Model):
+            foo = model.BlobProperty(compressed=False)
+
+        key = datastore.Key("ThisKind", 123, project="testing")
+        datastore_entity = datastore.Entity(key=key)
+        uncompressed_value = b"abc" * 1000
+        datastore_entity.update({"foo": uncompressed_value})
+        protobuf = helpers.entity_to_protobuf(datastore_entity)
+        entity = model._entity_from_protobuf(protobuf)
+        assert entity.foo == uncompressed_value
+        ds_entity = model._entity_to_ds_entity(entity)
+        assert ds_entity["foo"] == uncompressed_value
+
+    @staticmethod
+    @pytest.mark.usefixtures("in_context")
+    def test__from_datastore_uncompressed_to_compressed():
+        class ThisKind(model.Model):
+            foo = model.BlobProperty(compressed=True)
+
+        key = datastore.Key("ThisKind", 123, project="testing")
+        datastore_entity = datastore.Entity(key=key)
+        uncompressed_value = b"abc" * 1000
+        compressed_value = zlib.compress(uncompressed_value)
+        datastore_entity.update({"foo": uncompressed_value})
+        protobuf = helpers.entity_to_protobuf(datastore_entity)
+        entity = model._entity_from_protobuf(protobuf)
+        ds_entity = model._entity_to_ds_entity(entity)
+        assert ds_entity["foo"] == compressed_value
+
 
 class TestTextProperty:
     @staticmethod