fix: Fallback to Spark for LIKE with custom escape character (apache#478

) * Fallback to Spark for LIKE with custom escape character Currently, LIKE with custom escape character produces incorrect results. * For custom escape character, provide user with specific info message * Test case for default escape char with checkSparkAnswerAndOperator (cherry picked from commit fd90a28)
himadripal · May 31, 2024 · b17abe7 · b17abe7
1 parent ba16f40
commit b17abe7
Show file tree

Hide file tree

Showing 2 changed files with 37 additions and 14 deletions.
diff --git a/spark/src/main/scala/org/apache/comet/serde/QueryPlanSerde.scala b/spark/src/main/scala/org/apache/comet/serde/QueryPlanSerde.scala
@@ -1074,23 +1074,28 @@ object QueryPlanSerde extends Logging with ShimQueryPlanSerde with CometExprShim
             None
           }
 
-        case Like(left, right, _) =>
-          // TODO escapeChar
-          val leftExpr = exprToProtoInternal(left, inputs)
-          val rightExpr = exprToProtoInternal(right, inputs)
+        case Like(left, right, escapeChar) =>
+          if (escapeChar == '\\') {
+            val leftExpr = exprToProtoInternal(left, inputs)
+            val rightExpr = exprToProtoInternal(right, inputs)
 
-          if (leftExpr.isDefined && rightExpr.isDefined) {
-            val builder = ExprOuterClass.Like.newBuilder()
-            builder.setLeft(leftExpr.get)
-            builder.setRight(rightExpr.get)
+            if (leftExpr.isDefined && rightExpr.isDefined) {
+              val builder = ExprOuterClass.Like.newBuilder()
+              builder.setLeft(leftExpr.get)
+              builder.setRight(rightExpr.get)
 
-            Some(
-              ExprOuterClass.Expr
-                .newBuilder()
-                .setLike(builder)
-                .build())
+              Some(
+                ExprOuterClass.Expr
+                  .newBuilder()
+                  .setLike(builder)
+                  .build())
+            } else {
+              withInfo(expr, left, right)
+              None
+            }
           } else {
-            withInfo(expr, left, right)
+            // TODO custom escape char
+            withInfo(expr, s"custom escape character $escapeChar not supported in LIKE")
             None
           }
 

diff --git a/spark/src/test/scala/org/apache/comet/CometExpressionSuite.scala b/spark/src/test/scala/org/apache/comet/CometExpressionSuite.scala
@@ -594,6 +594,24 @@ class CometExpressionSuite extends CometTestBase with AdaptiveSparkPlanHelper {
     }
   }
 
+  test("like with custom escape") {
+    val table = "names"
+    withTable(table) {
+      sql(s"create table $table(id int, name varchar(20)) using parquet")
+      sql(s"insert into $table values(1,'James Smith')")
+      sql(s"insert into $table values(2,'Michael_Rose')")
+      sql(s"insert into $table values(3,'Robert_R_Williams')")
+
+      // Filter column having values that include underscores
+      val queryDefaultEscape = sql("select id from names where name like '%\\_%'")
+      checkSparkAnswerAndOperator(queryDefaultEscape)
+
+      val queryCustomEscape = sql("select id from names where name like '%$_%' escape '$'")
+      checkAnswer(queryCustomEscape, Row(2) :: Row(3) :: Nil)
+
+    }
+  }
+
   test("contains") {
     assume(!isSpark32)