Adjustment to see if this PoC works properly after some PR Requests

pdet · Sep 3, 2024 · 73b072a · 73b072a
1 parent 359be4e
commit 73b072a
Show file tree

Hide file tree

Showing 5 changed files with 671 additions and 740 deletions.
diff --git a/src/from_substrait.cpp b/src/from_substrait.cpp
@@ -416,7 +416,7 @@ shared_ptr<Relation> SubstraitToDuckDB::TransformJoinOp(const substrait::Rel &so
 	case substrait::JoinRel::JoinType::JoinRel_JoinType_JOIN_TYPE_LEFT_SEMI:
 		djointype = JoinType::SEMI;
 		break;
-	case substrait::JoinRel::JoinType::JoinRel_JoinType_JOIN_TYPE_MARK:
+	case substrait::JoinRel::JoinType::JoinRel_JoinType_JOIN_TYPE_LEFT_MARK:
 		djointype = JoinType::MARK;
 		break;
 	case substrait::JoinRel::JoinType::JoinRel_JoinType_JOIN_TYPE_RIGHT_SEMI:
@@ -439,7 +439,7 @@ shared_ptr<Relation> SubstraitToDuckDB::TransformDelimJoinOp(const substrait::Re
 		duplicate_eliminated_columns.emplace_back(
 		    make_uniq<PositionalReferenceExpression>(col.direct_reference().struct_field().field() + 1));
 	}
-
+	duplicate_eliminated_columns_ptr = &duplicate_eliminated_columns;
 	JoinType djointype;
 	switch (sjoin.type()) {
 	case substrait::DuplicateEliminatedJoinRel_JoinType::DuplicateEliminatedJoinRel_JoinType_JOIN_TYPE_INNER:
@@ -457,7 +457,7 @@ shared_ptr<Relation> SubstraitToDuckDB::TransformDelimJoinOp(const substrait::Re
 	case substrait::DuplicateEliminatedJoinRel_JoinType::DuplicateEliminatedJoinRel_JoinType_JOIN_TYPE_RIGHT_SEMI:
 		djointype = JoinType::RIGHT_SEMI;
 		break;
-	case substrait::DuplicateEliminatedJoinRel_JoinType::DuplicateEliminatedJoinRel_JoinType_JOIN_TYPE_MARK:
+	case substrait::DuplicateEliminatedJoinRel_JoinType::DuplicateEliminatedJoinRel_JoinType_JOIN_TYPE_LEFT_MARK:
 		djointype = JoinType::MARK;
 		break;
 	case substrait::DuplicateEliminatedJoinRel_JoinType::DuplicateEliminatedJoinRel_JoinType_JOIN_TYPE_RIGHT_ANTI:
@@ -490,9 +490,11 @@ shared_ptr<Relation> SubstraitToDuckDB::TransformDelimGetOp(const substrait::Rel
 	auto &client_context = con.context;
 	vector<LogicalType> chunk_types;
 	auto &input_columns = subtree->Columns();
-	for (auto &col_ref : delim_get.column_ids()) {
-		chunk_types.emplace_back(input_columns[col_ref.direct_reference().struct_field().field()].Type());
+	for (auto &col : *duplicate_eliminated_columns_ptr) {
+		auto& col_ref = col->Cast<PositionalReferenceExpression>();
+		chunk_types.emplace_back(input_columns[col_ref.index - 1].Type());
 	}
+	duplicate_eliminated_columns_ptr = nullptr;
 	return make_shared_ptr<DelimGetRelation>(client_context, chunk_types);
 }
 

diff --git a/src/include/from_substrait.hpp b/src/include/from_substrait.hpp
@@ -60,5 +60,7 @@ class SubstraitToDuckDB {
 	//! names
 	static const unordered_map<std::string, std::string> function_names_remap;
 	static const case_insensitive_set_t valid_extract_subfields;
+	//! Pointer to last seen duplicate_eliminated_columns
+	vector<unique_ptr<ParsedExpression>> *duplicate_eliminated_columns_ptr = nullptr;
 };
 } // namespace duckdb
diff --git a/src/to_substrait.cpp b/src/to_substrait.cpp
@@ -929,7 +929,7 @@ substrait::Rel *DuckDBToSubstrait::TransformComparisonJoin(LogicalOperator &dop)
 		sjoin->set_type(substrait::JoinRel::JoinType::JoinRel_JoinType_JOIN_TYPE_LEFT_SEMI);
 		break;
 	case JoinType::MARK:
-		sjoin->set_type(substrait::JoinRel::JoinType::JoinRel_JoinType_JOIN_TYPE_MARK);
+		sjoin->set_type(substrait::JoinRel::JoinType::JoinRel_JoinType_JOIN_TYPE_LEFT_MARK);
 		break;
 	case JoinType::RIGHT_SEMI:
 		sjoin->set_type(substrait::JoinRel::JoinType::JoinRel_JoinType_JOIN_TYPE_RIGHT_SEMI);
@@ -1000,7 +1000,7 @@ substrait::Rel *DuckDBToSubstrait::TransformDelimiterJoin(LogicalOperator &dop)
 		break;
 	case JoinType::MARK:
 		sjoin->set_type(
-		    substrait::DuplicateEliminatedJoinRel_JoinType::DuplicateEliminatedJoinRel_JoinType_JOIN_TYPE_MARK);
+		    substrait::DuplicateEliminatedJoinRel_JoinType::DuplicateEliminatedJoinRel_JoinType_JOIN_TYPE_LEFT_MARK);
 		break;
 	case JoinType::RIGHT_ANTI:
 		sjoin->set_type(
@@ -1397,12 +1397,6 @@ substrait::Rel *DuckDBToSubstrait::TransformIntersect(LogicalOperator &dop) {
 substrait::Rel *DuckDBToSubstrait::TransformDelimGet() {
 	auto rel = new substrait::Rel();
 	auto delim_get = rel->mutable_duplicate_eliminated_get();
-	D_ASSERT(duplicate_eliminated_parent_ptr);
-	for (auto &dup_col : duplicate_eliminated_parent_ptr->duplicate_eliminated_columns) {
-		auto &dref = dup_col->Cast<BoundReferenceExpression>();
-		auto s_dup_col = delim_get->add_column_ids();
-		s_dup_col->mutable_direct_reference()->mutable_struct_field()->set_field(static_cast<int32_t>(dref.index));
-	}
 	auto ref_input = delim_get->mutable_input();
 	ref_input->set_subtree_ordinal(cur_subtree_relation);
 	duplicate_eliminated_parent_ptr = nullptr;