feat(optimize): rewrite LIKE '%' (e.g., '%%', '%%%') as IS NOT NULL (#18288)

* optimize like followed by '%' or '%....' literal string

Author: chagelo

src/query/sql/src/planner/semantic/type_check.rs

@@ -4434,7 +4434,15 @@ impl<'a> TypeChecker<'a> {
         } else {
             Cow::Borrowed(like_str)
         };
-        if check_const(&new_like_str) {
+        if check_percent(&new_like_str) {
+            // Convert to `a is not null`
+            let is_not_null = Expr::IsNull {
+                span: None,
+                expr: Box::new(left.clone()),
+                not: true,
+            };
+            self.resolve(&is_not_null)
+        } else if check_const(&new_like_str) {
             // Convert to equal comparison
             self.resolve_binary_op(span, &BinaryOperator::Eq, left, right)
         } else if check_prefix(&new_like_str) {
@@ -5762,6 +5770,11 @@ pub fn validate_function_arg(
     }
 }
 
+// optimize special cases for like expression
+fn check_percent(like_str: &str) -> bool {
+    !like_str.is_empty() && like_str.chars().all(|c| c == '%')
+}
+
 // Some check functions for like expression
 fn check_const(like_str: &str) -> bool {
     for char in like_str.chars() {

tests/sqllogictests/suites/mode/standalone/explain/explain_like.test

@@ -19,7 +19,7 @@ explain select * from t1 where s like '%';
 ----
 Filter
 ├── output columns: [t1.s (#0)]
-├── filters: [is_true(like(t1.s (#0), '%'))]
+├── filters: [is_not_null(t1.s (#0))]
 ├── estimated rows: 5.00
 └── TableScan
     ├── table: default.default.t1
@@ -29,7 +29,7 @@ Filter
     ├── partitions total: 1
     ├── partitions scanned: 1
     ├── pruning stats: [segments: <range pruning: 1 to 1>, blocks: <range pruning: 1 to 1>]
-    ├── push downs: [filters: [is_true(like(t1.s (#0), '%'))], limit: NONE]
+    ├── push downs: [filters: [is_not_null(t1.s (#0))], limit: NONE]
     └── estimated rows: 5.00
 
 query T

tests/sqllogictests/suites/mode/standalone/explain/selectivity/is_not_null.test

@@ -24,3 +24,30 @@ Filter
 
 statement ok
 DROP TABLE twocolumn;
+
+statement ok
+create table if not exists t_user(id int not null, name varchar(10));
+
+statement ok
+insert into t_user values (10, 'alice'), (103, 'bod');
+
+query T
+EXPLAIN SELECT * FROM t_user WHERE name LIKE '%';
+----
+Filter
+├── output columns: [t_user.id (#0), t_user.name (#1)]
+├── filters: [is_not_null(t_user.name (#1))]
+├── estimated rows: 2.00
+└── TableScan
+    ├── table: default.default.t_user
+    ├── output columns: [id (#0), name (#1)]
+    ├── read rows: 2
+    ├── read size: < 1 KiB
+    ├── partitions total: 1
+    ├── partitions scanned: 1
+    ├── pruning stats: [segments: <range pruning: 1 to 1>, blocks: <range pruning: 1 to 1>]
+    ├── push downs: [filters: [is_not_null(t_user.name (#1))], limit: NONE]
+    └── estimated rows: 2.00
+
+statement ok
+drop table t_user;