From: J. Nick Koston <nick@koston.org>
Date: Tue, 25 Apr 2023 02:32:17 +0000 (-0400)
Subject: disable "bytes" handler for all drivers other than psycopg2
X-Git-Tag: rel_2_0_11~5^2
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=188cb4226ac7b337446689ab3498b4397d0b7d2d;p=thirdparty%2Fsqlalchemy%2Fsqlalchemy.git

disable "bytes" handler for all drivers other than psycopg2

Improved row processing performance for "binary" datatypes by making the
"bytes" handler conditional on a per driver basis.  As a result, the
"bytes" result handler has been disabled for nearly all drivers other than
psycopg2, all of which in modern forms support returning Python "bytes"
directly.  Pull request courtesy J. Nick Koston.

Fixes: #9680
Closes: #9681
Pull-request: https://github.com/sqlalchemy/sqlalchemy/pull/9681
Pull-request-sha: 4f2fd88bd9af54c54438a3b72a2f30384b0f8898

Change-Id: I394bdcbebaab272e63b13cc02f60813b7aa76839
---

diff --git a/doc/build/changelog/unreleased_20/9680.rst b/doc/build/changelog/unreleased_20/9680.rst
new file mode 100644
index 0000000000..e64d649db3
--- /dev/null
+++ b/doc/build/changelog/unreleased_20/9680.rst
@@ -0,0 +1,9 @@
+.. change::
+    :tags: performance, sql
+    :tickets: 9680
+
+    Improved row processing performance for "binary" datatypes by making the
+    "bytes" handler conditional on a per driver basis.  As a result, the
+    "bytes" result handler has been disabled for nearly all drivers other than
+    psycopg2, all of which in modern forms support returning Python "bytes"
+    directly.  Pull request courtesy J. Nick Koston.
diff --git a/lib/sqlalchemy/dialects/mssql/base.py b/lib/sqlalchemy/dialects/mssql/base.py
index 4a7e48ab8a..0afd726fd9 100644
--- a/lib/sqlalchemy/dialects/mssql/base.py
+++ b/lib/sqlalchemy/dialects/mssql/base.py
@@ -1347,7 +1347,8 @@ class TIMESTAMP(sqltypes._Binary):
         if self.convert_int:
 
             def process(value):
-                value = super_(value)
+                if super_:
+                    value = super_(value)
                 if value is not None:
                     # https://stackoverflow.com/a/30403242/34549
                     value = int(codecs.encode(value, "hex"), 16)
@@ -2974,6 +2975,8 @@ class MSDialect(default.DefaultDialect):
     supports_empty_insert = False
     favor_returning_over_lastrowid = True
 
+    returns_native_bytes = True
+
     supports_comments = True
     supports_default_metavalue = False
     """dialect supports INSERT... VALUES (DEFAULT) syntax -
diff --git a/lib/sqlalchemy/dialects/mysql/base.py b/lib/sqlalchemy/dialects/mysql/base.py
index eb9ccc6064..2ed2bbc7a0 100644
--- a/lib/sqlalchemy/dialects/mysql/base.py
+++ b/lib/sqlalchemy/dialects/mysql/base.py
@@ -2398,6 +2398,8 @@ class MySQLDialect(default.DefaultDialect):
 
     supports_native_enum = True
 
+    returns_native_bytes = True
+
     supports_sequences = False  # default for MySQL ...
     # ... may be updated to True for MariaDB 10.3+ in initialize()
 
diff --git a/lib/sqlalchemy/dialects/oracle/base.py b/lib/sqlalchemy/dialects/oracle/base.py
index 08ab35bea1..a3e724cbeb 100644
--- a/lib/sqlalchemy/dialects/oracle/base.py
+++ b/lib/sqlalchemy/dialects/oracle/base.py
@@ -1405,6 +1405,7 @@ class OracleDialect(default.DefaultDialect):
 
     supports_simple_order_by_label = False
     cte_follows_insert = True
+    returns_native_bytes = True
 
     supports_sequences = True
     sequences_optional = False
diff --git a/lib/sqlalchemy/dialects/postgresql/base.py b/lib/sqlalchemy/dialects/postgresql/base.py
index ad5e346b76..8e99942930 100644
--- a/lib/sqlalchemy/dialects/postgresql/base.py
+++ b/lib/sqlalchemy/dialects/postgresql/base.py
@@ -2912,6 +2912,8 @@ class PGDialect(default.DefaultDialect):
     postfetch_lastrowid = False
     use_insertmanyvalues = True
 
+    returns_native_bytes = True
+
     insertmanyvalues_implicit_sentinel = (
         InsertmanyvaluesSentinelOpts.ANY_AUTOINCREMENT
         | InsertmanyvaluesSentinelOpts.USE_INSERT_FROM_SELECT
diff --git a/lib/sqlalchemy/dialects/postgresql/psycopg2.py b/lib/sqlalchemy/dialects/postgresql/psycopg2.py
index e28bd8fdad..5cdd341833 100644
--- a/lib/sqlalchemy/dialects/postgresql/psycopg2.py
+++ b/lib/sqlalchemy/dialects/postgresql/psycopg2.py
@@ -600,6 +600,8 @@ class PGDialect_psycopg2(_PGDialect_common_psycopg):
     psycopg2_version = (0, 0)
     use_insertmanyvalues_wo_returning = True
 
+    returns_native_bytes = False
+
     _has_native_hstore = True
 
     colspecs = util.update_copy(
diff --git a/lib/sqlalchemy/dialects/sqlite/pysqlite.py b/lib/sqlalchemy/dialects/sqlite/pysqlite.py
index a40e3d256f..71da4d0ef5 100644
--- a/lib/sqlalchemy/dialects/sqlite/pysqlite.py
+++ b/lib/sqlalchemy/dialects/sqlite/pysqlite.py
@@ -486,6 +486,7 @@ class _SQLite_pysqliteDate(DATE):
 class SQLiteDialect_pysqlite(SQLiteDialect):
     default_paramstyle = "qmark"
     supports_statement_cache = True
+    returns_native_bytes = True
 
     colspecs = util.update_copy(
         SQLiteDialect.colspecs,
diff --git a/lib/sqlalchemy/engine/default.py b/lib/sqlalchemy/engine/default.py
index 8992334ee6..d604282877 100644
--- a/lib/sqlalchemy/engine/default.py
+++ b/lib/sqlalchemy/engine/default.py
@@ -157,6 +157,8 @@ class DefaultDialect(Dialect):
     supports_native_enum = False
     supports_native_boolean = False
     supports_native_uuid = False
+    returns_native_bytes = False
+
     non_native_boolean_check_constraint = True
 
     supports_simple_order_by_label = True
diff --git a/lib/sqlalchemy/engine/interfaces.py b/lib/sqlalchemy/engine/interfaces.py
index 0216c155d1..e4914551cf 100644
--- a/lib/sqlalchemy/engine/interfaces.py
+++ b/lib/sqlalchemy/engine/interfaces.py
@@ -1055,6 +1055,14 @@ class Dialect(EventTarget):
 
     """
 
+    returns_native_bytes: bool
+    """indicates if Python bytes() objects are returned natively by the
+    driver for SQL "binary" datatypes.
+
+    .. versionadded:: 2.0.11
+
+    """
+
     construct_arguments: Optional[
         List[Tuple[Type[Union[SchemaItem, ClauseElement]], Mapping[str, Any]]]
     ] = None
diff --git a/lib/sqlalchemy/sql/sqltypes.py b/lib/sqlalchemy/sql/sqltypes.py
index 4e7514e38d..ce579cca2a 100644
--- a/lib/sqlalchemy/sql/sqltypes.py
+++ b/lib/sqlalchemy/sql/sqltypes.py
@@ -934,6 +934,9 @@ class _Binary(TypeEngine[bytes]):
     # both sqlite3 and pg8000 seem to return it,
     # psycopg2 as of 2.5 returns 'memoryview'
     def result_processor(self, dialect, coltype):
+        if dialect.returns_native_bytes:
+            return None
+
         def process(value):
             if value is not None:
                 value = bytes(value)
diff --git a/lib/sqlalchemy/testing/suite/test_types.py b/lib/sqlalchemy/testing/suite/test_types.py
index 3ba7f8e897..72f1e8c10f 100644
--- a/lib/sqlalchemy/testing/suite/test_types.py
+++ b/lib/sqlalchemy/testing/suite/test_types.py
@@ -279,7 +279,6 @@ class ArrayTest(_LiteralRoundTripFixture, fixtures.TablesTest):
 
 
 class BinaryTest(_LiteralRoundTripFixture, fixtures.TablesTest):
-    __requires__ = ("binary_literals",)
     __backend__ = True
 
     @classmethod
@@ -294,14 +293,15 @@ class BinaryTest(_LiteralRoundTripFixture, fixtures.TablesTest):
             Column("pickle_data", PickleType),
         )
 
-    def test_binary_roundtrip(self, connection):
+    @testing.combinations(b"this is binary", b"7\xe7\x9f", argnames="data")
+    def test_binary_roundtrip(self, connection, data):
         binary_table = self.tables.binary_table
 
         connection.execute(
-            binary_table.insert(), {"id": 1, "binary_data": b"this is binary"}
+            binary_table.insert(), {"id": 1, "binary_data": data}
         )
         row = connection.execute(select(binary_table.c.binary_data)).first()
-        eq_(row, (b"this is binary",))
+        eq_(row, (data,))
 
     def test_pickle_roundtrip(self, connection):
         binary_table = self.tables.binary_table