bug: mssql backend fails to resolve schema for user-defined types

[kolatat]

What happened?

The MSSQL backend fails when the query result contains a user-defined type (UDT). When I use .sql() on the mssql backend:

ibis.mssql.connect(...).sql('SELECT someFieldWithUDTResult FROM someTable')

exception 'NoneType' object has no attribute 'lower' is thrown. This occurs because .sql() tries to resolve the schema using mssql._get_schema_using_query() using this statement:

ibis/ibis/backends/mssql/__init__.py

Lines 336 to 347 in d9402b4

	query = f"""
	SELECT
	name,
	is_nullable,
	system_type_name,
	precision,
	scale,
	error_number,
	error_message
	FROM sys.dm_exec_describe_first_result_set(N{tsql}, NULL, 0)
	ORDER BY column_ordinal
	""" # noqa: S608

relying on the system_type_name column from sys.dm_exec_describe_first_result_set. According to Microsoft documentation, system_type_name returns NULL for user-defined types. When this None value is passed to the SQL compiler TypeMapper, it results in the exception when

ibis/ibis/backends/sql/datatypes.py

Lines 182 to 184 in d9402b4

	def from_string(cls, text: str, nullable: bool | None = None) -> dt.DataType:
	if dtype := cls.unknown_type_strings.get(text.lower()):
	return dtype

calls text.lower() but text is None.

What version of ibis are you using?

10.8.0

What backend(s) are you using, if any?

MSSQL

Relevant log output

Code of Conduct

• I agree to follow this project's Code of Conduct

[kolatat]

Maybe the SQL

SELECT
	name,
	is_nullable,
	system_type_name,
	precision,
	scale,
	error_number,
	error_message
FROM sys.dm_exec_describe_first_result_set(N{tsql}, NULL, 0)
ORDER BY column_ordinal

can be replaced with

SELECT
	r.name,
	r.is_nullable,
	COALESCE(r.system_type_name, st.name),
	r.precision,
	r.scale,
	r.error_number,
	r.error_message
FROM sys.dm_exec_describe_first_result_set(N{tsql}, NULL, 0) r
LEFT JOIN sys.types st
ON r.system_type_id=st.user_type_id AND st.is_user_defined=0
ORDER BY column_ordinal

to always get a valid underlying system type name.