Filter out null map key entries by default

Summary: When reading from file, Presto Java filter out null maps by default, so we need to do the same in Velox. Differential Revision: D56522977
Yuhta · Apr 24, 2024 · ad63e90 · ad63e90
1 parent 8ac7e39
commit ad63e90
Show file tree

Hide file tree

Showing 4 changed files with 58 additions and 2 deletions.
diff --git a/velox/connectors/hive/HiveConnectorUtil.cpp b/velox/connectors/hive/HiveConnectorUtil.cpp
@@ -329,7 +329,8 @@ std::shared_ptr<common::ScanSpec> makeScanSpec(
         partitionKeys,
     const std::unordered_map<std::string, std::shared_ptr<HiveColumnHandle>>&
         infoColumns,
-    memory::MemoryPool* pool) {
+    memory::MemoryPool* pool,
+    bool mapNullKeysEnabled) {
   auto spec = std::make_shared<common::ScanSpec>("root");
   folly::F14FastMap<std::string, std::vector<const common::Subfield*>>
       filterSubfields;
@@ -397,6 +398,15 @@ std::shared_ptr<common::ScanSpec> makeScanSpec(
     fieldSpec->addFilter(*pair.second);
   }
 
+  if (!mapNullKeysEnabled) {
+    spec->visit(*rowType, [](const Type& type, common::ScanSpec& node) {
+      if (type.isMap()) {
+        node.childByName(common::ScanSpec::kMapKeysFieldName)
+            ->addFilter(common::IsNotNull());
+      }
+    });
+  }
+
   return spec;
 }
 

diff --git a/velox/connectors/hive/HiveConnectorUtil.h b/velox/connectors/hive/HiveConnectorUtil.h
@@ -57,7 +57,8 @@ std::shared_ptr<common::ScanSpec> makeScanSpec(
         partitionKeys,
     const std::unordered_map<std::string, std::shared_ptr<HiveColumnHandle>>&
         infoColumns,
-    memory::MemoryPool* pool);
+    memory::MemoryPool* pool,
+    bool mapNullKeysEnabled = false);
 
 void configureReaderOptions(
     dwio::common::ReaderOptions& readerOptions,

diff --git a/velox/dwio/common/ScanSpec.h b/velox/dwio/common/ScanSpec.h
@@ -322,6 +322,10 @@ class ScanSpec {
     flatMapFeatureSelection_ = std::move(features);
   }
 
+  /// Invoke the function provided on each node of the ScanSpec tree.
+  template <typename F>
+  void visit(const Type& type, F&& f);
+
  private:
   void reorder();
 
@@ -403,6 +407,30 @@ class ScanSpec {
   std::vector<std::string> flatMapFeatureSelection_;
 };
 
+template <typename F>
+void ScanSpec::visit(const Type& type, F&& f) {
+  f(type, *this);
+  switch (type.kind()) {
+    case TypeKind::ROW:
+      for (auto& child : children_) {
+        child->visit(*type.childAt(child->channel()), std::forward<F>(f));
+      }
+      break;
+    case TypeKind::MAP:
+      childByName(kMapKeysFieldName)
+          ->visit(*type.childAt(0), std::forward<F>(f));
+      childByName(kMapValuesFieldName)
+          ->visit(*type.childAt(1), std::forward<F>(f));
+      break;
+    case TypeKind::ARRAY:
+      childByName(kArrayElementsFieldName)
+          ->visit(*type.childAt(0), std::forward<F>(f));
+      break;
+    default:
+      break;
+  }
+}
+
 // Returns false if no value from a range defined by stats can pass the
 // filter. True, otherwise.
 bool testFilter(

diff --git a/velox/exec/tests/TableScanTest.cpp b/velox/exec/tests/TableScanTest.cpp
@@ -975,6 +975,23 @@ TEST_F(TableScanTest, subfieldPruningArrayType) {
   }
 }
 
+TEST_F(TableScanTest, skipNullMapKeys) {
+  auto vector = makeRowVector({makeMapVector(
+      {0, 2},
+      makeNullableFlatVector<int64_t>({std::nullopt, 2}),
+      makeFlatVector<int64_t>({1, 2}))});
+  auto rowType = asRowType(vector->type());
+  auto filePath = TempFilePath::create();
+  writeToFile(filePath->getPath(), {vector});
+  auto plan = PlanBuilder().tableScan(rowType).planNode();
+  auto split = makeHiveConnectorSplit(filePath->getPath());
+  auto expected = makeRowVector({makeMapVector(
+      {0, 1},
+      makeNullableFlatVector(std::vector<std::optional<int64_t>>(1, 2)),
+      makeFlatVector(std::vector<int64_t>(1, 2)))});
+  AssertQueryBuilder(plan).split(split).assertResults(expected);
+}
+
 // Test reading files written before schema change, e.g. missing newly added
 // columns.
 TEST_F(TableScanTest, missingColumns) {