Type inference (#12980)

* adding type hinting, changed tests to have INT64 as default for integers, not INT32

* added test for entity type, need to verify this works in python2

* fixed some silly linting issues

* using six to check for string and binary types

* adding six.integer_types as well

* modifying entity inference

* prefixing string with `u` to ensure string type matching

* unicoded strings in tests

* linting fixes

* addressing izzys nits

sdk/tables/azure-data-tables/azure/data/tables/_deserialize.py

@@ -61,11 +61,11 @@ def _deserialize_table_creation(response, _, headers):
 
 
 def _from_entity_binary(value):
-    return EntityProperty(EdmType.BINARY, _decode_base64_to_bytes(value))
+    return EntityProperty(_decode_base64_to_bytes(value))
 
 
 def _from_entity_int32(value):
-    return EntityProperty(EdmType.INT32, int(value))
+    return EntityProperty(int(value))
 
 
 zero = datetime.timedelta(0)  # same as 00:00

sdk/tables/azure-data-tables/azure/data/tables/_entity.py

@@ -4,6 +4,9 @@
 # license information.
 # --------------------------------------------------------------------------
 from enum import Enum
+from datetime import datetime
+from uuid import UUID
+import six
 
 from ._error import _ERROR_ATTRIBUTE_MISSING
 
@@ -75,17 +78,41 @@ class EntityProperty(object):
     """
 
     def __init__(self,
-                 type=None,  # type: Union[str,EdmType] # pylint:disable=W0622
-                 value=None  # type: Any
-                 ):
+        value=None, # type: Any
+        type=None,  # type: Union[str,EdmType] # pylint:disable=W0622
+    ):
         """
         Represents an Azure Table. Returned by list_tables.
 
         :param Union[str, EdmType] type: The type of the property.
         :param Any value: The value of the property.
         """
-        self.type = type
         self.value = value
+        if type is not None:
+            self.type = type
+        elif isinstance(value, six.text_type):
+            try:
+                self.value = UUID(value)
+                self.type = EdmType.GUID
+            except ValueError:
+                self.type = EdmType.STRING
+        elif isinstance(value, six.binary_type):
+            self.type = EdmType.BINARY
+        elif isinstance(value, bool):
+            self.type = EdmType.BOOLEAN
+        elif isinstance(value, six.integer_types):
+            self.type = EdmType.INT64
+        elif isinstance(value, datetime):
+            self.type = EdmType.DATETIME
+        elif isinstance(value, float):
+            self.type = EdmType.DOUBLE
+        else:
+            raise ValueError(
+                """Type of {} could not be inferred. Acceptable types are bytes, int, uuid.UUID,
+                datetime, string, int32, int64, float, and boolean. Refer to
+                azure.data.tables.EdmType for more information.
+                """.format(value)
+                )
 
 
 class EdmType(str, Enum):

sdk/tables/azure-data-tables/tests/test_table_batch.py

@@ -16,6 +16,7 @@
 from azure.core import MatchConditions
 from azure.core.exceptions import (
     ResourceExistsError)
+from azure.data.tables import EdmType, TableEntity, EntityProperty
 
 from _shared.testcase import GlobalStorageAccountPreparer, TableTestCase, LogCaptured
 
@@ -78,7 +79,7 @@ def _create_random_entity_dict(self, pk=None, rk=None):
             'Birthday': datetime(1973, 10, 4, tzinfo=tzutc()),
             'birthday': datetime(1970, 10, 4, tzinfo=tzutc()),
             'binary': b'binary',
-            'other': EntityProperty(EdmType.INT32, 20),
+            'other': EntityProperty(20, EdmType.INT32),
             'clsid': uuid.UUID('c9da6455-213d-42c9-9a79-3e9149a57833')
         }
         return Entity(**properties)
@@ -148,6 +149,33 @@ def _assert_updated_entity(self, entity):
         self.assertIsInstance(entity.timestamp, datetime)
 
     #--Test cases for batch ---------------------------------------------
+
+    def test_inferred_types(self):
+        # Arrange
+        # Act
+        entity = TableEntity()
+        entity.PartitionKey = '003'
+        entity.RowKey = 'batch_all_operations_together-1'
+        entity.test = EntityProperty(True)
+        entity.test2 = EntityProperty(b'abcdef')
+        entity.test3 = EntityProperty(u'c9da6455-213d-42c9-9a79-3e9149a57833')
+        entity.test4 = EntityProperty(datetime(1973, 10, 4, tzinfo=tzutc()))
+        entity.test5 = EntityProperty(u"stringystring")
+        entity.test6 = EntityProperty(3.14159)
+        entity.test7 = EntityProperty(100)
+        entity.test8 = EntityProperty(10, EdmType.INT32)
+
+        # Assert
+        self.assertEqual(entity.test.type, EdmType.BOOLEAN)
+        self.assertEqual(entity.test2.type, EdmType.BINARY)
+        self.assertEqual(entity.test3.type, EdmType.GUID)
+        self.assertEqual(entity.test4.type, EdmType.DATETIME)
+        self.assertEqual(entity.test5.type, EdmType.STRING)
+        self.assertEqual(entity.test6.type, EdmType.DOUBLE)
+        self.assertEqual(entity.test7.type, EdmType.INT64)
+        self.assertEqual(entity.test8.type, EdmType.INT32)
+
+
     @pytest.mark.skip("pending")
     @GlobalStorageAccountPreparer()
     def test_batch_insert(self, resource_group, location, storage_account, storage_account_key):
@@ -158,10 +186,10 @@ def test_batch_insert(self, resource_group, location, storage_account, storage_a
             entity = Entity()
             entity.PartitionKey = '001'
             entity.RowKey = 'batch_insert'
-            entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+            entity.test = EntityProperty(True)
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
 
             batch = self.table.create_batch()
@@ -185,10 +213,10 @@ def test_batch_update(self, resource_group, location, storage_account, storage_a
             entity = Entity()
             entity.PartitionKey = '001'
             entity.RowKey = 'batch_update'
-            entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+            entity.test = EntityProperty(True)
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
             self.table.create_item(entity)
 
@@ -218,10 +246,10 @@ def test_batch_merge(self, resource_group, location, storage_account, storage_ac
             entity = Entity()
             entity.PartitionKey = '001'
             entity.RowKey = 'batch_merge'
-            entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+            entity.test = EntityProperty(True)
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
             self.table.create_item(entity)
 
@@ -313,7 +341,7 @@ def test_batch_insert_replace(self, resource_group, location, storage_account, s
             entity.test = True
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
 
             batch = self.table.create_batch()
@@ -343,7 +371,7 @@ def test_batch_insert_merge(self, resource_group, location, storage_account, sto
             entity.test = True
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
 
             batch = self.table.create_batch()
@@ -370,10 +398,10 @@ def test_batch_delete(self, resource_group, location, storage_account, storage_a
             entity = Entity()
             entity.PartitionKey = '001'
             entity.RowKey = 'batch_delete'
-            entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+            entity.test = EntityProperty(True)
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
             self.table.create_item(entity)
 
@@ -399,10 +427,10 @@ def test_batch_inserts(self, resource_group, location, storage_account, storage_
             # Act
             entity = Entity()
             entity.PartitionKey = 'batch_inserts'
-            entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+            entity.test = EntityProperty(True)
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
 
             batch = self.table.create_batch()
             for i in range(100):
@@ -428,10 +456,10 @@ def test_batch_all_operations_together(self, resource_group, location, storage_a
             entity = Entity()
             entity.PartitionKey = '003'
             entity.RowKey = 'batch_all_operations_together-1'
-            entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+            entity.test = EntityProperty(True)
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
             self.table.create_item(entity)
             entity.RowKey = 'batch_all_operations_together-2'
@@ -476,10 +504,10 @@ def test_batch_all_operations_together_context_manager(self, resource_group, loc
             entity = Entity()
             entity.PartitionKey = '003'
             entity.RowKey = 'batch_all_operations_together-1'
-            entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+            entity.test = EntityProperty(True)
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
             self.table.create_item(entity)
             entity.RowKey = 'batch_all_operations_together-2'
@@ -525,10 +553,10 @@ def test_batch_reuse(self, resource_group, location, storage_account, storage_ac
             entity = Entity()
             entity.PartitionKey = '003'
             entity.RowKey = 'batch_all_operations_together-1'
-            entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+            entity.test = EntityProperty(True)
             entity.test2 = 'value'
             entity.test3 = 3
-            entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+            entity.test4 = EntityProperty(1234567890)
             entity.test5 = datetime.utcnow()
 
             batch = TableBatchClient()

sdk/tables/azure-data-tables/tests/test_table_entity.py

@@ -115,7 +115,7 @@ def _create_random_entity_dict(self, pk=None, rk=None):
             'Birthday': datetime(1973, 10, 4, tzinfo=tzutc()),
             'birthday': datetime(1970, 10, 4, tzinfo=tzutc()),
             'binary': b'binary',
-            'other': EntityProperty(type=EdmType.INT32, value=20),
+            'other': EntityProperty(value=20, type=EdmType.INT32),
             'clsid': uuid.UUID('c9da6455-213d-42c9-9a79-3e9149a57833')
         }
         return TableEntity(**properties)
@@ -160,7 +160,7 @@ def _assert_default_entity(self, entity, headers=None):
         self.assertEqual(entity['birthday'], datetime(1970, 10, 4, tzinfo=tzutc()))
         self.assertEqual(entity['binary'].value, b'binary')
         self.assertIsInstance(entity['other'], EntityProperty)
-        self.assertEqual(entity['other'].type, EdmType.INT32)
+        self.assertEqual(entity['other'].type, EdmType.INT64)
         self.assertEqual(entity['other'].value, 20)
         self.assertEqual(entity['clsid'], uuid.UUID('c9da6455-213d-42c9-9a79-3e9149a57833'))
         # self.assertTrue('metadata' in entity.odata)
@@ -188,7 +188,7 @@ def _assert_default_entity_json_full_metadata(self, entity, headers=None):
         self.assertEqual(entity['birthday'], datetime(1970, 10, 4, tzinfo=tzutc()))
         self.assertEqual(entity['binary'].value, b'binary')
         self.assertIsInstance(entity['other'], EntityProperty)
-        self.assertEqual(entity['other'].type, EdmType.INT32)
+        self.assertEqual(entity['other'].type, EdmType.INT64)
         self.assertEqual(entity['other'].value, 20)
         self.assertEqual(entity['clsid'], uuid.UUID('c9da6455-213d-42c9-9a79-3e9149a57833'))
         # self.assertTrue('metadata' in entity.odata)
@@ -222,7 +222,7 @@ def _assert_default_entity_json_no_metadata(self, entity, headers=None):
         self.assertTrue(entity['birthday'].endswith('00Z'))
         self.assertEqual(entity['binary'], b64encode(b'binary').decode('utf-8'))
         self.assertIsInstance(entity['other'], EntityProperty)
-        self.assertEqual(entity['other'].type, EdmType.INT32)
+        self.assertEqual(entity['other'].type, EdmType.INT64)
         self.assertEqual(entity['other'].value, 20)
         self.assertEqual(entity['clsid'], 'c9da6455-213d-42c9-9a79-3e9149a57833')
         # self.assertIsNone(entity.odata)
@@ -273,7 +273,7 @@ def _assert_merged_entity(self, entity):
         self.assertEqual(entity.Birthday, datetime(1973, 10, 4, tzinfo=tzutc()))
         self.assertEqual(entity.birthday, datetime(1991, 10, 4, tzinfo=tzutc()))
         self.assertIsInstance(entity.other, EntityProperty)
-        self.assertEqual(entity.other.type, EdmType.INT32)
+        self.assertEqual(entity.other.type, EdmType.INT64)
         self.assertEqual(entity.other.value, 20)
         self.assertIsInstance(entity.clsid, uuid.UUID)
         self.assertEqual(str(entity.clsid), 'c9da6455-213d-42c9-9a79-3e9149a57833')
@@ -417,13 +417,13 @@ def test_insert_entity_with_large_int32_value_throws(self, resource_group, locat
         try:
             # Act
             dict32 = self._create_random_base_entity_dict()
-            dict32['large'] = EntityProperty(EdmType.INT32, 2 ** 31)
+            dict32['large'] = EntityProperty(2 ** 31, EdmType.INT32)
 
             # Assert
             with self.assertRaises(TypeError):
                 self.table.create_entity(entity=dict32)
 
-            dict32['large'] = EntityProperty(EdmType.INT32, -(2 ** 31 + 1))
+            dict32['large'] = EntityProperty(-(2 ** 31 + 1), EdmType.INT32)
             with self.assertRaises(TypeError):
                 self.table.create_entity(entity=dict32)
         finally:
@@ -438,13 +438,13 @@ def test_insert_entity_with_large_int64_value_throws(self, resource_group, locat
         try:
             # Act
             dict64 = self._create_random_base_entity_dict()
-            dict64['large'] = EntityProperty(EdmType.INT64, 2 ** 63)
+            dict64['large'] = EntityProperty(2 ** 63)
 
             # Assert
             with self.assertRaises(TypeError):
                 self.table.create_entity(entity=dict64)
 
-            dict64['large'] = EntityProperty(EdmType.INT64, -(2 ** 63 + 1))
+            dict64['large'] = EntityProperty(-(2 ** 63 + 1))
             with self.assertRaises(TypeError):
                 self.table.create_entity(entity=dict64)
         finally:
@@ -1304,10 +1304,10 @@ def test_query_entities_large(self, resource_group, location, storage_account, s
                 entity = Entity()
                 entity.PartitionKey = 'large'
                 entity.RowKey = 'batch{0}-item{1}'.format(j, i)
-                entity.test = EntityProperty(EdmType.BOOLEAN, 'true')
+                entity.test = EntityProperty(True)
                 entity.test2 = 'hello world;' * 100
                 entity.test3 = 3
-                entity.test4 = EntityProperty(EdmType.INT64, '1234567890')
+                entity.test4 = EntityProperty(1234567890)
                 entity.test5 = datetime(2016, 12, 31, 11, 59, 59, 0)
                 batch.create_entity(entity)
             self.ts.commit_batch(table_name, batch)