BUG: Map pandas integer to optimal SQLAlchemy integer type (GH35076)

Author: avinashpancham

pandas/io/sql.py

@@ -1097,6 +1097,7 @@ def _sqlalchemy_type(self, col):
             DateTime,
             Float,
             Integer,
+            SmallInteger,
             Text,
             Time,
         )
@@ -1127,8 +1128,13 @@ def _sqlalchemy_type(self, col):
             else:
                 return Float(precision=53)
         elif col_type == "integer":
-            if col.dtype == "int32":
+            # GH35076 Map pandas integer to optimal SQLAlchemy integer type
+            if col.dtype.name.lower() in ("int8", "uint8", "int16"):
+                return SmallInteger
+            elif col.dtype.name.lower() in ("uint16", "int32"):
                 return Integer
+            elif col.dtype.name.lower() == "uint64":
+                raise ValueError("Unsigned 64 bit integer datatype is not supported")
             else:
                 return BigInteger
         elif col_type == "boolean":

pandas/tests/io/test_sql.py

@@ -1139,6 +1139,45 @@ def test_sqlalchemy_type_mapping(self):
         # GH 9086: TIMESTAMP is the suggested type for datetimes with timezones
         assert isinstance(table.table.c["time"].type, sqltypes.TIMESTAMP)
 
+    @pytest.mark.parametrize(
+        "integer, expected",
+        [
+            ("int8", "SMALLINT"),
+            ("Int8", "SMALLINT"),
+            ("uint8", "SMALLINT"),
+            ("UInt8", "SMALLINT"),
+            ("int16", "SMALLINT"),
+            ("Int16", "SMALLINT"),
+            ("uint16", "INTEGER"),
+            ("UInt16", "INTEGER"),
+            ("int32", "INTEGER"),
+            ("Int32", "INTEGER"),
+            ("uint32", "BIGINT"),
+            ("UInt32", "BIGINT"),
+            (int, "BIGINT"),
+            ("int64", "BIGINT"),
+            ("Int64", "BIGINT"),
+        ],
+    )
+    def test_sqlalchemy_integer_mapping(self, integer, expected):
+        # GH35076 Map pandas integer to optimal SQLAlchemy integer type
+        df = DataFrame([0, 1], columns=["a"], dtype=integer)
+        db = sql.SQLDatabase(self.conn)
+        table = sql.SQLTable("test_type", db, frame=df)
+
+        result = str(table.table.c["a"].type)
+        assert result == expected
+
+    @pytest.mark.parametrize("integer", ["uint64", "UInt64"])
+    def test_sqlalchemy_integer_overload_mapping(self, integer):
+        # GH35076 Map pandas integer to optimal SQLAlchemy integer type
+        df = DataFrame([0, 1], columns=["a"], dtype=integer)
+        db = sql.SQLDatabase(self.conn)
+        with pytest.raises(
+            ValueError, match="Unsigned 64 bit integer datatype is not supported"
+        ):
+            sql.SQLTable("test_type", db, frame=df)
+
     def test_database_uri_string(self):
 
         # Test read_sql and .to_sql method with a database URI (GH10654)