fix: is_select check for lowercase select with "WITH" clauses (#22370)

superset/sql_parse.py

@@ -228,7 +228,7 @@ def is_select(self) -> bool:
  # for `UNKNOWN`, check all DDL/DML explicitly: only `SELECT` DML is allowed,
  # and no DDL is allowed
  if any(token.ttype == DDL for token in parsed[0]) or any(
- token.ttype == DML and token.value != "SELECT" for token in parsed[0]
+ token.ttype == DML and token.normalized != "SELECT" for token in parsed[0]
  ):
  return False
 
@@ -237,7 +237,7 @@ def is_select(self) -> bool:
  return False
 
  return any(
- token.ttype == DML and token.value == "SELECT" for token in parsed[0]
+ token.ttype == DML and token.normalized == "SELECT" for token in parsed[0]
  )
 
  def is_valid_ctas(self) -> bool:

tests/unit_tests/sql_parse_tests.py

@@ -1008,6 +1008,28 @@ def test_cte_is_select() -> None:
  assert sql.is_select()
 
 
+def test_cte_is_select_lowercase() -> None:
+ """
+ Some CTEs with lowercase select are not correctly identified as SELECTS.
+ """
+ sql = ParsedQuery(
+ """WITH foo AS(
+select
+ FLOOR(__time TO WEEK) AS "week",
+ name,
+ COUNT(DISTINCT user_id) AS "unique_users"
+FROM "druid"."my_table"
+GROUP BY 1,2
+)
+select
+ f.week,
+ f.name,
+ f.unique_users
+FROM foo f"""
+ )
+ assert sql.is_select()
+
+
 def test_unknown_select() -> None:
  """
  Test that `is_select` works when sqlparse fails to identify the type.