rusackas commented on code in PR #41804:
URL: https://github.com/apache/superset/pull/41804#discussion_r3540253248


##########
superset/models/helpers.py:
##########
@@ -3746,7 +3758,11 @@ def get_sqla_query(  # pylint: 
disable=too-many-arguments,too-many-locals,too-ma
                         utils.FilterOperator.ILIKE,
                         utils.FilterOperator.LIKE,
                     }:
-                        if target_generic_type != GenericDataType.STRING:
+                        # Native UUID columns report GenericDataType.STRING but
+                        # reject LIKE/ILIKE without a cast (see issue #41795)
+                        if target_generic_type != GenericDataType.STRING or (
+                            is_uuid_native_type(col_type)
+                        ):
                             sqla_col = sa.cast(sqla_col, sa.String)

Review Comment:
   Merged the LIKE and NOT LIKE branches into one `elif` so the cast condition 
(now `needs_string_cast_for_like`) is computed once and reused for all four ops.



##########
superset/models/helpers.py:
##########
@@ -250,6 +250,18 @@ def json_to_dict(json_str: str) -> dict[Any, Any]:
     return {}
 
 
+def is_uuid_native_type(native_type: Optional[str]) -> bool:
+    """
+    Return True if a native column type represents a UUID.
+
+    Engines such as PostgreSQL and ClickHouse expose native UUID column types
+    (e.g. ``UUID``, ``Nullable(UUID)``) that map to ``GenericDataType.STRING``
+    yet reject LIKE/ILIKE against the raw column, so these columns need an
+    explicit cast to string before pattern matching.
+    """
+    return native_type is not None and "uuid" in native_type.lower()

Review Comment:
   Switched to a `\buuid\b` word-boundary regex (case-insensitive, with strip) 
so an unrelated type that happens to contain "uuid" as part of a longer word 
wont get misclassified.



##########
tests/unit_tests/models/helpers_test.py:
##########
@@ -3410,3 +3410,71 @@ def fake_engine(*args, **kwargs):
     # ```forecasts.original`.`total_cost``` (the regression), so this negative
     # assertion catches the actual failure mode, not just an exact-string 
match.
     assert "`forecasts.original`" not in sql
+
+
[email protected](
+    "native_type",
+    ["UUID", "uuid", "Nullable(UUID)"],
+)
[email protected](
+    "op",
+    ["LIKE", "ILIKE", "NOT LIKE", "NOT ILIKE"],
+)
+def test_like_filter_on_uuid_column_casts_to_string(
+    database: Database, native_type: str, op: str
+) -> None:
+    """
+    LIKE-family filters on native UUID columns must cast the column to string.
+
+    UUID columns map to ``GenericDataType.STRING``, so the generic-type guard
+    alone skips the string cast — but engines such as PostgreSQL and ClickHouse
+    reject LIKE/ILIKE against a raw UUID column (issue #41795: table chart
+    server-pagination search fails with e.g. "Illegal type UUID of argument of
+    function ilike"). The native column type must force the cast.
+    """
+    from superset.connectors.sqla.models import SqlaTable, TableColumn
+
+    table = SqlaTable(
+        database=database,
+        schema=None,
+        table_name="t",
+        columns=[TableColumn(column_name="event_id", type=native_type)],
+    )
+
+    result = table.get_sqla_query(
+        columns=["event_id"],
+        metrics=[],
+        extras={},
+        filter=[{"col": "event_id", "op": op, "val": "abc%"}],
+        granularity=None,
+        is_timeseries=False,
+        orderby=[],
+    )
+    sql = str(result.sqla_query)
+    assert "CAST" in sql.upper(), f"Expected string cast in SQL: {sql}"

Review Comment:
   Fair point, swapped both assertions to walk the whereclause with 
`sqlalchemy.sql.visitors.iterate` and check for a `Cast` node instead of 
grepping the compiled SQL string.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to