Apply new uniquing rules for future ORM selects

Fixed issue where usage of the :meth:`_result.Result.unique` method with an
ORM result that included column expressions with unhashable types, such as
``JSON`` or ``ARRAY`` using non-tuples would silently fall back to using
the ``id()`` function, rather than raising an error. This now raises an
error when the :meth:`_result.Result.unique` method is used in a 2.0 style
ORM query. Additionally, hashability is assumed to be True for result
values of unknown type, such as often happens when using SQL functions of
unknown return type; if values are truly not hashable then the ``hash()``
itself will raise.

For legacy ORM queries, since the legacy :class:`_orm.Query` object
uniquifies in all cases, the old rules remain in place, which is to use
``id()`` for result values of unknown type as this legacy uniquing is
mostly for the purpose of uniquing ORM entities and not column values.

Fixes: #6769
Change-Id: I5747f706f1e97c78867b5cf28c73360497273808

4 files changed, 45 insertions, 10 deletions

diff --git a/lib/sqlalchemy/orm/context.py b/lib/sqlalchemy/orm/context.py
index 0af3fd6af..c4b695687 100644
--- a/lib/sqlalchemy/orm/context.py
+++ b/lib/sqlalchemy/orm/context.py
@@ -81,6 +81,7 @@ class QueryContext(object):
         _yield_per = None
         _refresh_state = None
         _lazy_loaded_from = None
+        _legacy_uniquing = False
 
     def __init__(
         self,
@@ -2257,6 +2258,10 @@ class _QueryEntity(object):
 
     __slots__ = ()
 
+    _non_hashable_value = False
+    _null_column_type = False
+    use_id_for_hash = False
+
     @classmethod
     def to_compile_state(cls, compile_state, entities, entities_collection):
 
@@ -2387,6 +2392,7 @@ class _MapperEntity(_QueryEntity):
 
     supports_single_entity = True
 
+    _non_hashable_value = True
     use_id_for_hash = True
 
     @property
@@ -2483,7 +2489,6 @@ class _MapperEntity(_QueryEntity):
 
 
 class _BundleEntity(_QueryEntity):
-    use_id_for_hash = False
 
     _extra_entities = ()
 
@@ -2663,9 +2668,13 @@ class _ColumnEntity(_QueryEntity):
         return self.column.type
 
     @property
-    def use_id_for_hash(self):
+    def _non_hashable_value(self):
         return not self.column.type.hashable
 
+    @property
+    def _null_column_type(self):
+        return self.column.type._isnull
+
     def row_processor(self, context, result):
         compile_state = context.compile_state
 
diff --git a/lib/sqlalchemy/orm/loading.py b/lib/sqlalchemy/orm/loading.py
index 948f33ad5..abc8780ed 100644
--- a/lib/sqlalchemy/orm/loading.py
+++ b/lib/sqlalchemy/orm/loading.py
@@ -92,17 +92,43 @@ def instances(cursor, context):
             "Can't use the ORM yield_per feature in conjunction with unique()"
         )
 
-    row_metadata = SimpleResultMetaData(
-        labels,
-        extra,
-        _unique_filters=[
+    def _not_hashable(datatype):
+        def go(obj):
+            raise sa_exc.InvalidRequestError(
+                "Can't apply uniqueness to row tuple containing value of "
+                "type %r; this datatype produces non-hashable values"
+                % datatype
+            )
+
+        return go
+
+    if context.load_options._legacy_uniquing:
+        unique_filters = [
+            _no_unique
+            if context.yield_per
+            else id
+            if (
+                ent.use_id_for_hash
+                or ent._non_hashable_value
+                or ent._null_column_type
+            )
+            else None
+            for ent in context.compile_state._entities
+        ]
+    else:
+        unique_filters = [
             _no_unique
             if context.yield_per
+            else _not_hashable(ent.column.type)
+            if (not ent.use_id_for_hash and ent._non_hashable_value)
             else id
             if ent.use_id_for_hash
             else None
             for ent in context.compile_state._entities
-        ],
+        ]
+
+    row_metadata = SimpleResultMetaData(
+        labels, extra, _unique_filters=unique_filters
     )
 
     def chunks(size):
diff --git a/lib/sqlalchemy/orm/query.py b/lib/sqlalchemy/orm/query.py
index 6ad7f3020..9a97d37b0 100644
--- a/lib/sqlalchemy/orm/query.py
+++ b/lib/sqlalchemy/orm/query.py
@@ -130,7 +130,9 @@ class Query(
 
     _compile_options = ORMCompileState.default_compile_options
 
-    load_options = QueryContext.default_load_options
+    load_options = QueryContext.default_load_options + {
+        "_legacy_uniquing": True
+    }
 
     _params = util.EMPTY_DICT
 
diff --git a/lib/sqlalchemy/sql/sqltypes.py b/lib/sqlalchemy/sql/sqltypes.py
index 1b05465c9..44431d38f 100644
--- a/lib/sqlalchemy/sql/sqltypes.py
+++ b/lib/sqlalchemy/sql/sqltypes.py
@@ -3184,8 +3184,6 @@ class NullType(TypeEngine):
 
     _isnull = True
 
-    hashable = False
-
     def literal_processor(self, dialect):
         def process(value):
             raise exc.CompileError(