[sqlalchemy-spanner]: _designate_type() raises KeyError on unrecognized Spanner column types instead of falling back to NullType #18839

Determine this is the right repository

• I determined this is the correct repository in which to report this bug.

Summary of the issue

SpannerDialect._designate_type() performs an unguarded _type_map[str_repr] lookup and raises KeyError whenever a reflected table (or any table in the same schema during get_multi_columns()) contains a Spanner column type not in _type_map (such as INTERVAL, PROTO<...>, ENUM<...>, or STRUCT<...>, as well as UUID and TOKENLIST prior to 1.20.0).

Environment details

• OS: Linux
• Python version: 3.13
• sqlalchemy-spanner version: 1.17.2 and main / 1.20.1 (eb8a76628adf10093959600071a18f50285c2ecf)
• sqlalchemy version: 2.0.52

Problem description

In SpannerDialect._designate_type:

    def _designate_type(self, str_repr):
        if str_repr.startswith("STRING"):
            end = str_repr.index(")")
            size = int_from_size(str_repr[7:end])
            return _type_map["STRING"](length=size)
        elif str_repr.startswith("BYTES"):
            end = str_repr.index(")")
            size = int_from_size(str_repr[6:end])
            return _type_map["BYTES"](length=size)
        elif str_repr.startswith("ARRAY"):
            inner_type_str = str_repr[6:-1]
            inner_type = self._designate_type(inner_type_str)
            return _type_map["ARRAY"](inner_type)
        else:
            return _type_map[str_repr]

While UUID (#17913) and TOKENLIST (#17911) were added to _type_map in 1.20.0, _designate_type() still raises an unhandled KeyError for any other valid Spanner column type not in _type_map (e.g. INTERVAL, PROTO<...>, ENUM<...>, STRUCT<...>). Because get_multi_columns() reflects all tables in a schema when filter_names=None (or when MetaData.reflect() reflects a schema), a single table with an unrecognized column type crashes reflection for the entire schema.

In SQLAlchemy's built-in dialects (e.g., PostgreSQL, MySQL, SQLite, Oracle), unrecognized column types during reflection emit sa.util.warn(f"Did not recognize type '{str_repr}' ...") and return sa.types.NullType(). (See also #16622.)

Reproduction steps: code

from google.cloud.sqlalchemy_spanner.sqlalchemy_spanner import SpannerDialect

dialect = SpannerDialect()
for spanner_type in ("INTERVAL", "PROTO<my.pkg.Message>", "ENUM<my.pkg.Status>"):
    try:
        dialect._designate_type(spanner_type)
    except KeyError as e:
        print(f"{spanner_type} -> KeyError: {e}")

Actual output

INTERVAL -> KeyError: 'INTERVAL'
PROTO<my.pkg.Message> -> KeyError: 'PROTO<my.pkg.Message>'
ENUM<my.pkg.Status> -> KeyError: 'ENUM<my.pkg.Status>'

Expected output

_designate_type() maps supported Spanner types and falls back to sa.types.NullType() with sa.util.warn(...) for unrecognized column types instead of raising KeyError.