Add and update unit tests

Signed-off-by: Tai Le Manh <[email protected]>

Author: tlm365

datafusion/optimizer/src/unwrap_cast_in_comparison.rs

@@ -159,20 +159,26 @@ impl TreeNodeRewriter for UnwrapCastExprRewriter {
                             expr: right_expr, ..
                         }),
                     ) => {
-                        // if the left_lit_value can be casted to the type of expr
+                        // if the left_lit_value can be cast to the type of expr
                         // we need to unwrap the cast for cast/try_cast expr, and add cast to the literal
                         let Ok(expr_type) = right_expr.get_type(&self.schema) else {
                             return Ok(Transformed::no(expr));
                         };
-                        let Some(value) =
-                            try_cast_literal_to_type(left_lit_value, &expr_type)
-                        else {
-                            return Ok(Transformed::no(expr));
-                        };
-                        **left = lit(value);
-                        // unwrap the cast/try_cast for the right expr
-                        **right = mem::take(right_expr);
-                        Ok(Transformed::yes(expr))
+                        match expr_type {
+                            // https://github.com/apache/datafusion/issues/12180
+                            DataType::Utf8View => Ok(Transformed::no(expr)),
+                            _ => {
+                                let Some(value) =
+                                    try_cast_literal_to_type(left_lit_value, &expr_type)
+                                else {
+                                    return Ok(Transformed::no(expr));
+                                };
+                                **left = lit(value);
+                                // unwrap the cast/try_cast for the right expr
+                                **right = mem::take(right_expr);
+                                Ok(Transformed::yes(expr))
+                            }
+                        }
                     }
                     (
                         Expr::TryCast(TryCast {
@@ -183,20 +189,26 @@ impl TreeNodeRewriter for UnwrapCastExprRewriter {
                         }),
                         Expr::Literal(right_lit_value),
                     ) => {
-                        // if the right_lit_value can be casted to the type of expr
+                        // if the right_lit_value can be cast to the type of expr
                         // we need to unwrap the cast for cast/try_cast expr, and add cast to the literal
                         let Ok(expr_type) = left_expr.get_type(&self.schema) else {
                             return Ok(Transformed::no(expr));
                         };
-                        let Some(value) =
-                            try_cast_literal_to_type(right_lit_value, &expr_type)
-                        else {
-                            return Ok(Transformed::no(expr));
-                        };
-                        // unwrap the cast/try_cast for the left expr
-                        **left = mem::take(left_expr);
-                        **right = lit(value);
-                        Ok(Transformed::yes(expr))
+                        match expr_type {
+                            // https://github.com/apache/datafusion/issues/12180
+                            DataType::Utf8View => Ok(Transformed::no(expr)),
+                            _ => {
+                                let Some(value) =
+                                    try_cast_literal_to_type(right_lit_value, &expr_type)
+                                else {
+                                    return Ok(Transformed::no(expr));
+                                };
+                                // unwrap the cast/try_cast for the left expr
+                                **left = mem::take(left_expr);
+                                **right = lit(value);
+                                Ok(Transformed::yes(expr))
+                            }
+                        }
                     }
                     _ => Ok(Transformed::no(expr)),
                 }

datafusion/sqllogictest/test_files/string_view.slt

@@ -1226,43 +1226,83 @@ NULL NULL NULL
 # `~` operator (regex match)
 query TT
 EXPLAIN SELECT
-  column1_utf8view ~ 'foo' AS c1
+  column1_utf8view ~ 'an' AS c1
 FROM test;
 ----
 logical_plan
-01)Projection: CAST(test.column1_utf8view AS Utf8) LIKE Utf8("%foo%") AS c1
+01)Projection: CAST(test.column1_utf8view AS Utf8) LIKE Utf8("%an%") AS c1
 02)--TableScan: test projection=[column1_utf8view]
 
+query B
+SELECT
+  column1_utf8view ~ 'an' AS c1
+FROM test;
+----
+false
+true
+false
+NULL
+
 # `~*` operator (regex match case-insensitive)
 query TT
 EXPLAIN SELECT
-  column1_utf8view ~* 'foo' AS c1
+  column1_utf8view ~* '^a.{3}e' AS c1
 FROM test;
 ----
 logical_plan
-01)Projection: CAST(test.column1_utf8view AS Utf8) ILIKE Utf8("%foo%") AS c1
+01)Projection: CAST(test.column1_utf8view AS Utf8) ~* Utf8("^a.{3}e") AS c1
 02)--TableScan: test projection=[column1_utf8view]
 
+query B
+SELECT
+  column1_utf8view ~* '^a.{3}e' AS c1
+FROM test;
+----
+true
+false
+false
+NULL
+
 # `!~~` operator (not like match)
 query TT
 EXPLAIN SELECT
-  column1_utf8view !~~ 'an' AS c1
+  column1_utf8view !~~ 'xia_g%g' AS c1
 FROM test;
 ----
 logical_plan
-01)Projection: CAST(test.column1_utf8view AS Utf8) !~~ Utf8("an") AS c1
+01)Projection: CAST(test.column1_utf8view AS Utf8) !~~ Utf8("xia_g%g") AS c1
 02)--TableScan: test projection=[column1_utf8view]
 
+query B
+SELECT
+  column1_utf8view !~~ 'xia_g%g' AS c1
+FROM test;
+----
+true
+true
+true
+NULL
+
 # `!~~*` operator (not like match case-insensitive)
 query TT
 EXPLAIN SELECT
-  column1_utf8view !~~* 'an' AS c1
+  column1_utf8view !~~* 'xia_g%g' AS c1
 FROM test;
 ----
 logical_plan
-01)Projection: CAST(test.column1_utf8view AS Utf8) !~~* Utf8("an") AS c1
+01)Projection: CAST(test.column1_utf8view AS Utf8) !~~* Utf8("xia_g%g") AS c1
 02)--TableScan: test projection=[column1_utf8view]
 
+query B
+SELECT
+  column1_utf8view !~~* 'xia_g%g' AS c1
+FROM test;
+----
+true
+false
+true
+NULL
+
 statement ok
 drop table test;