[native] Add more semi- and anti-join e2e tests

wypb · Nov 11, 2022 · 79dde4b · 79dde4b
1 parent f43daa2
commit 79dde4b
Show file tree

Hide file tree

Showing 2 changed files with 30 additions and 10 deletions.
diff --git a/presto_cpp/main/types/PrestoToVeloxQueryPlan.cpp b/presto_cpp/main/types/PrestoToVeloxQueryPlan.cpp
@@ -1121,12 +1121,6 @@ core::PlanNodePtr VeloxQueryPlanConverter::toVeloxQueryPlan(
       }
     }
 
-    if (!joinType.has_value()) {
-      VELOX_UNSUPPORTED(
-          "Unsupported Filter over SemiJoin: {}",
-          toJsonString(node->predicate));
-    }
-
     std::vector<core::FieldAccessTypedExprPtr> leftKeys = {
         exprConverter_.toVeloxExpr(semiJoin->sourceJoinVariable)};
     std::vector<core::FieldAccessTypedExprPtr> rightKeys = {
@@ -1139,10 +1133,26 @@ core::PlanNodePtr VeloxQueryPlanConverter::toVeloxQueryPlan(
     const auto& leftNames = left->outputType()->names();
     const auto& leftTypes = left->outputType()->children();
 
-    std::vector<std::string> names;
-    names.reserve(leftNames.size() + 1);
-    std::copy(leftNames.begin(), leftNames.end(), std::back_inserter(names));
-    names.emplace_back(semiJoin->semiJoinOutput.name);
+    auto names = leftNames;
+    names.push_back(semiJoin->semiJoinOutput.name);
+
+    if (!joinType.has_value()) {
+      auto types = leftTypes;
+      types.push_back(BOOLEAN());
+
+      return std::make_shared<core::FilterNode>(
+          node->id,
+          exprConverter_.toVeloxExpr(node->predicate),
+          std::make_shared<core::HashJoinNode>(
+              semiJoin->id,
+              core::JoinType::kLeftSemiProject,
+              leftKeys,
+              rightKeys,
+              nullptr, // filter
+              left,
+              right,
+              ROW(std::move(names), std::move(types))));
+    }
 
     std::vector<core::TypedExprPtr> projections;
     projections.reserve(leftNames.size() + 1);

diff --git a/src/test/java/com/facebook/presto/nativeworker/TestHiveJoinQueries.java b/src/test/java/com/facebook/presto/nativeworker/TestHiveJoinQueries.java
@@ -54,6 +54,11 @@ public void testSemiJoin()
 
         assertQuery(partitionedJoin, "SELECT * FROM lineitem WHERE orderkey IN (SELECT orderkey FROM orders WHERE (orderkey + custkey) % 2 = 0)");
         assertQuery(broadcastJoin, "SELECT * FROM lineitem WHERE orderkey IN (SELECT orderkey FROM orders WHERE (orderkey + custkey) % 2 = 0)");
+
+        assertQuery(partitionedJoin, "SELECT * FROM lineitem " +
+                "WHERE linenumber = 3 OR orderkey IN (SELECT orderkey FROM orders WHERE (orderkey + custkey) % 2 = 0)");
+        assertQuery(broadcastJoin, "SELECT linenumber, quantity FROM lineitem " +
+                "WHERE linenumber = 3 OR orderkey IN (SELECT orderkey FROM orders WHERE (orderkey + custkey) % 2 = 0)");
     }
 
     @Test
@@ -64,6 +69,11 @@ public void testAntiJoin()
 
         assertQuery(partitionedJoin, "SELECT * FROM lineitem WHERE orderkey NOT IN (SELECT orderkey FROM orders WHERE (orderkey + custkey) % 2 = 0)");
         assertQuery(broadcastJoin, "SELECT * FROM lineitem WHERE orderkey NOT IN (SELECT orderkey FROM orders WHERE (orderkey + custkey) % 2 = 0)");
+
+        assertQuery(partitionedJoin, "SELECT * FROM lineitem " +
+                "WHERE linenumber = 3 OR orderkey NOT IN (SELECT orderkey FROM orders WHERE (orderkey + custkey) % 2 = 0)");
+        assertQuery(broadcastJoin, "SELECT * FROM lineitem " +
+                "WHERE linenumber = 3 OR orderkey NOT IN (SELECT orderkey FROM orders WHERE (orderkey + custkey) % 2 = 0)");
     }
 
     @Test