diff --git a/sql/src/sparksql_upgrade/rules.py b/sql/src/sparksql_upgrade/rules.py
index 28f32b2..bd3eeb5 100644
--- a/sql/src/sparksql_upgrade/rules.py
+++ b/sql/src/sparksql_upgrade/rules.py
@@ -447,10 +447,11 @@ def _eval(self, context: RuleContext) -> Optional[LintResult]:
         bracketed_segments = children.first(sp.is_type("bracketed"))
 
         if function_name == "APPROX_PERCENTILE" or function_name == "PERCENTILE_APPROX":
-            print("Found approx function!")
+            print(f"Found approx function! {function_name}")
 
             expression_count = 0
             expression_segment = None
+            # Find "middle" of the approx_percentile(bloop) (e.g. bloop)
             for segment in bracketed_segments.children().iterate_segments(
                 sp.is_type("expression")
             ):
@@ -460,7 +461,12 @@ def _eval(self, context: RuleContext) -> Optional[LintResult]:
 
             if expression_segment is not None:
                 expression_child = expression_segment.children().first()
-                if expression_child[0].type == "function":
+                # cast can either be a keyword or a function depending on if were iterating on
+                # parsed on updated code.
+                if expression_child[0].type == "keyword":
+                    if expression_child.child[0].raw == "cast":
+                        return None
+                elif expression_child[0].type == "function":
                     function_name_id_seg = (
                         expression_child.children()
                         .first(sp.is_type("function_name"))