duckdb 0.6.2-dev490.0 → 0.6.2-dev503.0

package/package.json

@@ -2,7 +2,7 @@
   "name": "duckdb",
   "main": "./lib/duckdb.js",
   "types": "./lib/duckdb.d.ts",
-  "version": "0.6.2-dev490.0",
+  "version": "0.6.2-dev503.0",
   "description": "DuckDB node.js API",
   "gypfile": true,
   "dependencies": {

package/src/duckdb.cpp

@@ -80940,7 +80940,23 @@ vector<LogicalType> BufferedCSVReader::SniffCSV(const vector<LogicalType> &reque
 	// #######
 	options.num_cols = best_num_cols;
 	DetectHeader(best_sql_types_candidates, best_header_row);
-
+	auto sql_types_per_column = options.sql_types_per_column;
+	for (idx_t i = 0; i < col_names.size(); i++) {
+		auto it = sql_types_per_column.find(col_names[i]);
+		if (it != sql_types_per_column.end()) {
+			best_sql_types_candidates[i] = {it->second};
+			sql_types_per_column.erase(col_names[i]);
+		}
+	}
+	if (!sql_types_per_column.empty()) {
+		string exception = "COLUMN_TYPES error: Columns with names: ";
+		for (auto &col : sql_types_per_column) {
+			exception += "\"" + col.first + "\",";
+		}
+		exception.pop_back();
+		exception += " do not exist in the CSV File";
+		throw BinderException(exception);
+	}
 	// #######
 	// ### type detection (refining)
 	// #######
@@ -127123,6 +127139,26 @@ static unique_ptr<FunctionData> ReadCSVBind(ClientContext &context, TableFunctio
 			if (names.empty()) {
 				throw BinderException("read_csv requires at least a single column as input!");
 			}
+		} else if (loption == "column_types") {
+			auto &child_type = kv.second.type();
+			if (child_type.id() != LogicalTypeId::STRUCT) {
+				throw BinderException("read_csv_auto column_types requires a struct as input");
+			}
+			auto &struct_children = StructValue::GetChildren(kv.second);
+			D_ASSERT(StructType::GetChildCount(child_type) == struct_children.size());
+			for (idx_t i = 0; i < struct_children.size(); i++) {
+				auto &name = StructType::GetChildName(child_type, i);
+				auto &val = struct_children[i];
+				if (val.type().id() != LogicalTypeId::VARCHAR) {
+					throw BinderException("read_csv_auto requires a type specification as string");
+				}
+				auto def_type = TransformStringToLogicalType(StringValue::Get(val));
+				if (def_type.id() == LogicalTypeId::USER) {
+					throw BinderException("Unrecognized type for read_csv_auto column_types definition");
+				}
+				options.sql_types_per_column[name] = def_type;
+			}
+
 		} else if (loption == "all_varchar") {
 			options.all_varchar = BooleanValue::Get(kv.second);
 		} else if (loption == "normalize_names") {
@@ -127761,6 +127797,7 @@ TableFunction ReadCSVTableFunction::GetAutoFunction(bool list_parameter) {
 	read_csv_auto.get_batch_index = CSVReaderGetBatchIndex;
 	read_csv_auto.cardinality = CSVReaderCardinality;
 	ReadCSVAddNamedParameters(read_csv_auto);
+	read_csv_auto.named_parameters["column_types"] = LogicalType::ANY;
 	return read_csv_auto;
 }
 
@@ -156976,6 +157013,7 @@ private:
 	void UpdateDPTree(JoinNode *new_plan);
 
 	void UpdateJoinNodesInFullPlan(JoinNode *node);
+	bool NodeInFullPlan(JoinNode *node);
 
 	std::pair<JoinRelationSet *, unique_ptr<LogicalOperator>>
 	GenerateJoins(vector<unique_ptr<LogicalOperator>> &extracted_relations, JoinNode *node);
@@ -157706,7 +157744,6 @@ string JoinNode::ToString() {
 
 
 
-
 #include <algorithm>
 
 namespace std {
@@ -157975,6 +158012,10 @@ unique_ptr<JoinNode> JoinOrderOptimizer::CreateJoinTree(JoinRelationSet *set,
 	return result;
 }
 
+bool JoinOrderOptimizer::NodeInFullPlan(JoinNode *node) {
+	return join_nodes_in_full_plan.find(node->set->ToString()) != join_nodes_in_full_plan.end();
+}
+
 void JoinOrderOptimizer::UpdateJoinNodesInFullPlan(JoinNode *node) {
 	if (!node) {
 		return;
@@ -158205,7 +158246,7 @@ static vector<unordered_set<idx_t>> AddSuperSets(vector<unordered_set<idx_t>> cu
 }
 
 // works by first creating all sets with cardinality 1
-// then iterates over each previously create group of subsets and will only add a neighbor if the neighbor
+// then iterates over each previously created group of subsets and will only add a neighbor if the neighbor
 // is greater than all relations in the set.
 static vector<unordered_set<idx_t>> GetAllNeighborSets(JoinRelationSet *new_set, unordered_set<idx_t> &exclusion_set,
                                                        vector<idx_t> neighbors) {
@@ -158241,6 +158282,11 @@ static vector<unordered_set<idx_t>> GetAllNeighborSets(JoinRelationSet *new_set,
 }
 
 void JoinOrderOptimizer::UpdateDPTree(JoinNode *new_plan) {
+	if (!NodeInFullPlan(new_plan)) {
+		// if the new node is not in the full plan, feel free to return
+		// because you won't be updating the full plan.
+		return;
+	}
 	auto new_set = new_plan->set;
 	// now update every plan that uses this plan
 	unordered_set<idx_t> exclusion_set;
@@ -158302,6 +158348,7 @@ void JoinOrderOptimizer::SolveJoinOrderApproximately() {
 					// update the DP tree in case a plan created by the DP algorithm uses the node
 					// that was potentially just updated by EmitPair. You will get a use-after-free
 					// error if future plans rely on the old node that was just replaced.
+					// if node in FullPath, then updateDP tree.
 					UpdateDPTree(node);
 
 					if (!best_connection || node->GetCost() < best_connection->GetCost()) {