[GLUTEN-3861][CH] Fix parse exception when join postJoinFilter contai…

…ns singularOrList
apache · Nov 29, 2023 · 80e2229 · 80e2229
1 parent e0f197f
commit 80e2229
Show file tree

Hide file tree

Showing 3 changed files with 30 additions and 3 deletions.
diff --git a/...ickhouse/src/test/scala/io/glutenproject/execution/GlutenClickHouseTPCHParquetSuite.scala b/...ickhouse/src/test/scala/io/glutenproject/execution/GlutenClickHouseTPCHParquetSuite.scala
@@ -2197,6 +2197,28 @@ class GlutenClickHouseTPCHParquetSuite extends GlutenClickHouseTPCHAbstractSuite
     }
   }
 
+  test("GLUTEN-3861: Fix parse exception when join postJoinFilter contains singularOrList") {
+    withSQLConf(("spark.sql.autoBroadcastJoinThreshold", "-1")) {
+      val sql =
+        """
+          |select t1.l_orderkey, t2.o_orderkey, t1.l_year, t2.o_year
+          |from (
+          |  select l_orderkey, extract(year from l_shipdate) as l_year, count(1) as l_cnt
+          |  from lineitem
+          |  group by l_orderkey, l_shipdate) t1
+          |left join (
+          |  select o_orderkey, extract(year from o_orderdate) as o_year, count(1) as o_cnt
+          |  from orders
+          |  group by o_orderkey, o_orderdate) t2
+          |on t1.l_orderkey = t2.o_orderkey
+          | and l_year in (1997, 1995, 1993)
+          |order by t1.l_orderkey, t2.o_orderkey, t1.l_year, t2.o_year
+          |limit 100
+          |""".stripMargin
+      compareResultsAgainstVanillaSpark(sql, true, { _ => })
+    }
+  }
+
   test("GLUTEN-3467: Fix 'Names of tuple elements must be unique' error for ch backend") {
     val sql =
       """

diff --git a/cpp-ch/local-engine/Parser/JoinRelParser.cpp b/cpp-ch/local-engine/Parser/JoinRelParser.cpp
@@ -397,7 +397,7 @@ bool JoinRelParser::tryAddPushDownFilter(
             }
         }
     }
-    // if ch not support the join type or join conditions, it will throw an exception like 'not support'.
+    // if ch does not support the join type or join conditions, it will throw an exception like 'not support'.
     catch (Poco::Exception & e)
     {
         // CH not support join condition has 'or' and has different table in each side.

diff --git a/cpp-ch/local-engine/Parser/SerializedPlanParser.cpp b/cpp-ch/local-engine/Parser/SerializedPlanParser.cpp
@@ -1916,6 +1916,8 @@ ActionsDAGPtr ASTParser::convertToActions(const NamesAndTypesList & name_and_typ
 ASTPtr ASTParser::parseToAST(const Names & names, const substrait::Expression & rel)
 {
     LOG_DEBUG(&Poco::Logger::get("ASTParser"), "substrait plan:\n{}", rel.DebugString());
+    if (rel.has_singular_or_list())
+        return parseArgumentToAST(names, rel);
     if (!rel.has_scalar_function())
         throw Exception(ErrorCodes::BAD_ARGUMENTS, "the root of expression should be a scalar function:\n {}", rel.DebugString());
 
@@ -2021,7 +2023,8 @@ ASTPtr ASTParser::parseArgumentToAST(const Names & names, const substrait::Expre
 
             bool nullable = false;
             size_t options_len = options.size();
-            args.reserve(options_len);
+            ASTs in_args;
+            in_args.reserve(options_len);
 
             for (int i = 0; i < static_cast<int>(options_len); ++i)
             {
@@ -2044,8 +2047,10 @@ ASTPtr ASTParser::parseArgumentToAST(const Names & names, const substrait::Expre
                         elem_type->getName(),
                         option_type->getName());
 
-                args.emplace_back(std::make_shared<ASTLiteral>(type_and_field.second));
+                in_args.emplace_back(std::make_shared<ASTLiteral>(type_and_field.second));
             }
+            auto array_ast = makeASTFunction("array", in_args);
+            args.emplace_back(array_ast);
 
             auto ast = makeASTFunction("in", args);
             if (nullable)