Enable filter push-down on nested field

apache · Nov 14, 2024 · 7ea5c2c · 7ea5c2c
1 parent 59d5248
commit 7ea5c2c
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 37 deletions.
diff --git a/cpp/velox/operators/functions/SparkExprToSubfieldFilterParser.cc b/cpp/velox/operators/functions/SparkExprToSubfieldFilterParser.cc
@@ -20,30 +20,6 @@ namespace gluten {
 
 using namespace facebook::velox;
 
-bool SparkExprToSubfieldFilterParser::toSparkSubfield(const core::ITypedExpr* field, common::Subfield& subfield) {
-  std::vector<std::unique_ptr<common::Subfield::PathElement>> path;
-  for (auto* current = field;;) {
-    if (auto* fieldAccess = dynamic_cast<const core::FieldAccessTypedExpr*>(current)) {
-      path.push_back(std::make_unique<common::Subfield::NestedField>(fieldAccess->name()));
-    } else if (dynamic_cast<const core::DereferenceTypedExpr*>(current)) {
-      return false;
-    } else if (dynamic_cast<const core::InputTypedExpr*>(current) == nullptr) {
-      return false;
-    } else {
-      break;
-    }
-
-    if (!current->inputs().empty()) {
-      return false;
-    } else {
-      break;
-    }
-  }
-  std::reverse(path.begin(), path.end());
-  subfield = common::Subfield(std::move(path));
-  return true;
-}
-
 std::unique_ptr<common::Filter> SparkExprToSubfieldFilterParser::leafCallToSubfieldFilter(
     const core::CallTypedExpr& call,
     common::Subfield& subfield,
@@ -56,42 +32,42 @@ std::unique_ptr<common::Filter> SparkExprToSubfieldFilterParser::leafCallToSubfi
   const auto* leftSide = call.inputs()[0].get();
 
   if (call.name() == "equalto") {
-    if (toSparkSubfield(leftSide, subfield)) {
+    if (toSubfield(leftSide, subfield)) {
       return negated ? makeNotEqualFilter(call.inputs()[1], evaluator) : makeEqualFilter(call.inputs()[1], evaluator);
     }
   } else if (call.name() == "lessthanorequal") {
-    if (toSparkSubfield(leftSide, subfield)) {
+    if (toSubfield(leftSide, subfield)) {
       return negated ? makeGreaterThanFilter(call.inputs()[1], evaluator)
                      : makeLessThanOrEqualFilter(call.inputs()[1], evaluator);
     }
   } else if (call.name() == "lessthan") {
-    if (toSparkSubfield(leftSide, subfield)) {
+    if (toSubfield(leftSide, subfield)) {
       return negated ? makeGreaterThanOrEqualFilter(call.inputs()[1], evaluator)
                      : makeLessThanFilter(call.inputs()[1], evaluator);
     }
   } else if (call.name() == "greaterthanorequal") {
-    if (toSparkSubfield(leftSide, subfield)) {
+    if (toSubfield(leftSide, subfield)) {
       return negated ? makeLessThanFilter(call.inputs()[1], evaluator)
                      : makeGreaterThanOrEqualFilter(call.inputs()[1], evaluator);
     }
   } else if (call.name() == "greaterthan") {
-    if (toSparkSubfield(leftSide, subfield)) {
+    if (toSubfield(leftSide, subfield)) {
       return negated ? makeLessThanOrEqualFilter(call.inputs()[1], evaluator)
                      : makeGreaterThanFilter(call.inputs()[1], evaluator);
     }
   } else if (call.name() == "in") {
-    if (toSparkSubfield(leftSide, subfield)) {
+    if (toSubfield(leftSide, subfield)) {
       return makeInFilter(call.inputs()[1], evaluator, negated);
     }
   } else if (call.name() == "isnull") {
-    if (toSparkSubfield(leftSide, subfield)) {
+    if (toSubfield(leftSide, subfield)) {
       if (negated) {
         return exec::isNotNull();
       }
       return exec::isNull();
     }
   } else if (call.name() == "isnotnull") {
-    if (toSparkSubfield(leftSide, subfield)) {
+    if (toSubfield(leftSide, subfield)) {
       if (negated) {
         return exec::isNull();
       }

diff --git a/cpp/velox/operators/functions/SparkExprToSubfieldFilterParser.h b/cpp/velox/operators/functions/SparkExprToSubfieldFilterParser.h
@@ -21,17 +21,12 @@ namespace gluten {
 /// Parses Spark expression into subfield filter. Differences from Presto's parser include:
 /// 1) Some Spark functions are registered under different names.
 /// 2) The supported functions vary.
-/// 3) Filter push-down on nested fields is disabled.
 class SparkExprToSubfieldFilterParser : public facebook::velox::exec::ExprToSubfieldFilterParser {
  public:
   std::unique_ptr<facebook::velox::common::Filter> leafCallToSubfieldFilter(
       const facebook::velox::core::CallTypedExpr& call,
       facebook::velox::common::Subfield& subfield,
       facebook::velox::core::ExpressionEvaluator* evaluator,
       bool negated) override;
-
- private:
-  // Compared to the upstream 'toSubfield', the push-down of filter on nested field is disabled.
-  bool toSparkSubfield(const facebook::velox::core::ITypedExpr* field, facebook::velox::common::Subfield& subfield);
 };
 } // namespace gluten