Update vendored DuckDB sources to 7ce99bc041

duckdblabs-bot · github-actions[bot] · commit 4e0781008f4b · 2025-11-08T06:44:51.000Z
diff --git a/src/duckdb/src/execution/join_hashtable.cpp b/src/duckdb/src/execution/join_hashtable.cpp
@@ -888,19 +888,25 @@ idx_t ScanStructure::ResolvePredicates(DataChunk &keys, SelectionVector &match_s
 	}
 
 	// If there is a matcher for the probing side because of non-equality predicates, use it
+	idx_t result_count;
 	if (ht.needs_chain_matcher) {
 		idx_t no_match_count = 0;
 		auto &matcher = no_match_sel ? ht.row_matcher_probe_no_match_sel : ht.row_matcher_probe;
 		D_ASSERT(matcher);
 
 		// we need to only use the vectors with the indices of the columns that are used in the probe phase, namely
 		// the non-equality columns
-		return matcher->Match(keys, key_state.vector_data, match_sel, this->count, pointers, no_match_sel,
-		                      no_match_count);
+		result_count =
+		    matcher->Match(keys, key_state.vector_data, match_sel, this->count, pointers, no_match_sel, no_match_count);
 	} else {
 		// no match sel is the opposite of match sel
-		return this->count;
+		result_count = this->count;
 	}
+
+	// Update total probe match count
+	ht.total_probe_matches.fetch_add(result_count, std::memory_order_relaxed);
+
+	return result_count;
 }
 
 idx_t ScanStructure::ScanInnerJoin(DataChunk &keys, SelectionVector &result_vector) {
diff --git a/src/duckdb/src/execution/operator/join/physical_hash_join.cpp b/src/duckdb/src/execution/operator/join/physical_hash_join.cpp
@@ -164,6 +164,11 @@ class HashJoinGlobalSinkState : public GlobalSinkState {
 		}
 	}
 
+	~HashJoinGlobalSinkState() override {
+		DUCKDB_LOG(context, PhysicalOperatorLogType, op, "PhysicalHashJoin", "GetData",
+		           {{"total_probe_matches", to_string(hash_table->total_probe_matches)}});
+	}
+
 	void ScheduleFinalize(Pipeline &pipeline, Event &event);
 	void InitializeProbeSpill();
 
diff --git a/src/duckdb/src/function/table/version/pragma_version.cpp b/src/duckdb/src/function/table/version/pragma_version.cpp
@@ -1,5 +1,5 @@
 #ifndef DUCKDB_PATCH_VERSION
-#define DUCKDB_PATCH_VERSION "2-dev279"
+#define DUCKDB_PATCH_VERSION "2-dev284"
 #endif
 #ifndef DUCKDB_MINOR_VERSION
 #define DUCKDB_MINOR_VERSION 4
@@ -8,10 +8,10 @@
 #define DUCKDB_MAJOR_VERSION 1
 #endif
 #ifndef DUCKDB_VERSION
-#define DUCKDB_VERSION "v1.4.2-dev279"
+#define DUCKDB_VERSION "v1.4.2-dev284"
 #endif
 #ifndef DUCKDB_SOURCE_ID
-#define DUCKDB_SOURCE_ID "783f08ffd8"
+#define DUCKDB_SOURCE_ID "7ce99bc041"
 #endif
 #include "duckdb/function/table/system_functions.hpp"
 #include "duckdb/main/database.hpp"
diff --git a/src/duckdb/src/include/duckdb/execution/join_hashtable.hpp b/src/duckdb/src/include/duckdb/execution/join_hashtable.hpp
@@ -277,6 +277,8 @@ class JoinHashTable {
 	uint64_t bitmask = DConstants::INVALID_INDEX;
 	//! Whether or not we error on multiple rows found per match in a SINGLE join
 	bool single_join_error_on_multiple_rows = true;
+	//! Number of probe matches
+	atomic<idx_t> total_probe_matches {0};
 
 	struct {
 		mutex mj_lock;
diff --git a/src/duckdb/src/include/duckdb/main/relation.hpp b/src/duckdb/src/include/duckdb/main/relation.hpp
@@ -162,19 +162,27 @@ class Relation : public enable_shared_from_this<Relation> {
 
 	//! Insert the data from this relation into a table
 	DUCKDB_API shared_ptr<Relation> InsertRel(const string &schema_name, const string &table_name);
+	DUCKDB_API shared_ptr<Relation> InsertRel(const string &catalog_name, const string &schema_name,
+	                                          const string &table_name);
 	DUCKDB_API void Insert(const string &table_name);
 	DUCKDB_API void Insert(const string &schema_name, const string &table_name);
+	DUCKDB_API void Insert(const string &catalog_name, const string &schema_name, const string &table_name);
 	//! Insert a row (i.e.,list of values) into a table
-	DUCKDB_API void Insert(const vector<vector<Value>> &values);
-	DUCKDB_API void Insert(vector<vector<unique_ptr<ParsedExpression>>> &&expressions);
+	DUCKDB_API virtual void Insert(const vector<vector<Value>> &values);
+	DUCKDB_API virtual void Insert(vector<vector<unique_ptr<ParsedExpression>>> &&expressions);
 	//! Create a table and insert the data from this relation into that table
 	DUCKDB_API shared_ptr<Relation> CreateRel(const string &schema_name, const string &table_name,
 	                                          bool temporary = false,
 	                                          OnCreateConflict on_conflict = OnCreateConflict::ERROR_ON_CONFLICT);
+	DUCKDB_API shared_ptr<Relation> CreateRel(const string &catalog_name, const string &schema_name,
+	                                          const string &table_name, bool temporary = false,
+	                                          OnCreateConflict on_conflict = OnCreateConflict::ERROR_ON_CONFLICT);
 	DUCKDB_API void Create(const string &table_name, bool temporary = false,
 	                       OnCreateConflict on_conflict = OnCreateConflict::ERROR_ON_CONFLICT);
 	DUCKDB_API void Create(const string &schema_name, const string &table_name, bool temporary = false,
 	                       OnCreateConflict on_conflict = OnCreateConflict::ERROR_ON_CONFLICT);
+	DUCKDB_API void Create(const string &catalog_name, const string &schema_name, const string &table_name,
+	                       bool temporary = false, OnCreateConflict on_conflict = OnCreateConflict::ERROR_ON_CONFLICT);
 
 	//! Write a relation to a CSV file
 	DUCKDB_API shared_ptr<Relation>
diff --git a/src/duckdb/src/include/duckdb/main/relation/create_table_relation.hpp b/src/duckdb/src/include/duckdb/main/relation/create_table_relation.hpp
@@ -16,8 +16,11 @@ class CreateTableRelation : public Relation {
 public:
 	CreateTableRelation(shared_ptr<Relation> child, string schema_name, string table_name, bool temporary,
 	                    OnCreateConflict on_conflict);
+	CreateTableRelation(shared_ptr<Relation> child, string catalog_name, string schema_name, string table_name,
+	                    bool temporary, OnCreateConflict on_conflict);
 
 	shared_ptr<Relation> child;
+	string catalog_name;
 	string schema_name;
 	string table_name;
 	vector<ColumnDefinition> columns;
diff --git a/src/duckdb/src/include/duckdb/main/relation/insert_relation.hpp b/src/duckdb/src/include/duckdb/main/relation/insert_relation.hpp
@@ -15,8 +15,10 @@ namespace duckdb {
 class InsertRelation : public Relation {
 public:
 	InsertRelation(shared_ptr<Relation> child, string schema_name, string table_name);
+	InsertRelation(shared_ptr<Relation> child, string catalog_name, string schema_name, string table_name);
 
 	shared_ptr<Relation> child;
+	string catalog_name;
 	string schema_name;
 	string table_name;
 	vector<ColumnDefinition> columns;
diff --git a/src/duckdb/src/include/duckdb/main/relation/table_relation.hpp b/src/duckdb/src/include/duckdb/main/relation/table_relation.hpp
@@ -29,6 +29,8 @@ class TableRelation : public Relation {
 
 	unique_ptr<TableRef> GetTableRef() override;
 
+	void Insert(const vector<vector<Value>> &values) override;
+	void Insert(vector<vector<unique_ptr<ParsedExpression>>> &&expressions) override;
 	void Update(const string &update, const string &condition = string()) override;
 	void Update(vector<string> column_names, vector<unique_ptr<ParsedExpression>> &&update,
 	            unique_ptr<ParsedExpression> condition = nullptr) override;
diff --git a/src/duckdb/src/main/relation.cpp b/src/duckdb/src/main/relation.cpp
@@ -241,15 +241,24 @@ BoundStatement Relation::Bind(Binder &binder) {
 }
 
 shared_ptr<Relation> Relation::InsertRel(const string &schema_name, const string &table_name) {
-	return make_shared_ptr<InsertRelation>(shared_from_this(), schema_name, table_name);
+	return InsertRel(INVALID_CATALOG, schema_name, table_name);
+}
+
+shared_ptr<Relation> Relation::InsertRel(const string &catalog_name, const string &schema_name,
+                                         const string &table_name) {
+	return make_shared_ptr<InsertRelation>(shared_from_this(), catalog_name, schema_name, table_name);
 }
 
 void Relation::Insert(const string &table_name) {
 	Insert(INVALID_SCHEMA, table_name);
 }
 
 void Relation::Insert(const string &schema_name, const string &table_name) {
-	auto insert = InsertRel(schema_name, table_name);
+	Insert(INVALID_CATALOG, schema_name, table_name);
+}
+
+void Relation::Insert(const string &catalog_name, const string &schema_name, const string &table_name) {
+	auto insert = InsertRel(catalog_name, schema_name, table_name);
 	auto res = insert->Execute();
 	if (res->HasError()) {
 		const string prepended_message = "Failed to insert into table '" + table_name + "': ";
@@ -258,30 +267,37 @@ void Relation::Insert(const string &schema_name, const string &table_name) {
 }
 
 void Relation::Insert(const vector<vector<Value>> &values) {
-	vector<string> column_names;
-	auto rel = make_shared_ptr<ValueRelation>(context->GetContext(), values, std::move(column_names), "values");
-	rel->Insert(GetAlias());
+	throw InvalidInputException("INSERT with values can only be used on base tables!");
 }
 
 void Relation::Insert(vector<vector<unique_ptr<ParsedExpression>>> &&expressions) {
-	vector<string> column_names;
-	auto rel = make_shared_ptr<ValueRelation>(context->GetContext(), std::move(expressions), std::move(column_names),
-	                                          "values");
-	rel->Insert(GetAlias());
+	(void)std::move(expressions);
+	throw InvalidInputException("INSERT with expressions can only be used on base tables!");
 }
 
 shared_ptr<Relation> Relation::CreateRel(const string &schema_name, const string &table_name, bool temporary,
                                          OnCreateConflict on_conflict) {
-	return make_shared_ptr<CreateTableRelation>(shared_from_this(), schema_name, table_name, temporary, on_conflict);
+	return CreateRel(INVALID_CATALOG, schema_name, table_name, temporary, on_conflict);
+}
+
+shared_ptr<Relation> Relation::CreateRel(const string &catalog_name, const string &schema_name,
+                                         const string &table_name, bool temporary, OnCreateConflict on_conflict) {
+	return make_shared_ptr<CreateTableRelation>(shared_from_this(), catalog_name, schema_name, table_name, temporary,
+	                                            on_conflict);
 }
 
 void Relation::Create(const string &table_name, bool temporary, OnCreateConflict on_conflict) {
-	Create(INVALID_SCHEMA, table_name, temporary, on_conflict);
+	Create(INVALID_CATALOG, INVALID_SCHEMA, table_name, temporary, on_conflict);
 }
 
 void Relation::Create(const string &schema_name, const string &table_name, bool temporary,
                       OnCreateConflict on_conflict) {
-	auto create = CreateRel(schema_name, table_name, temporary, on_conflict);
+	Create(INVALID_CATALOG, schema_name, table_name, temporary, on_conflict);
+}
+
+void Relation::Create(const string &catalog_name, const string &schema_name, const string &table_name, bool temporary,
+                      OnCreateConflict on_conflict) {
+	auto create = CreateRel(catalog_name, schema_name, table_name, temporary, on_conflict);
 	auto res = create->Execute();
 	if (res->HasError()) {
 		const string prepended_message = "Failed to create table '" + table_name + "': ";
diff --git a/src/duckdb/src/main/relation/create_table_relation.cpp b/src/duckdb/src/main/relation/create_table_relation.cpp
@@ -14,12 +14,21 @@ CreateTableRelation::CreateTableRelation(shared_ptr<Relation> child_p, string sc
 	TryBindRelation(columns);
 }
 
+CreateTableRelation::CreateTableRelation(shared_ptr<Relation> child_p, string catalog_name, string schema_name,
+                                         string table_name, bool temporary_p, OnCreateConflict on_conflict)
+    : Relation(child_p->context, RelationType::CREATE_TABLE_RELATION), child(std::move(child_p)),
+      catalog_name(std::move(catalog_name)), schema_name(std::move(schema_name)), table_name(std::move(table_name)),
+      temporary(temporary_p), on_conflict(on_conflict) {
+	TryBindRelation(columns);
+}
+
 BoundStatement CreateTableRelation::Bind(Binder &binder) {
 	auto select = make_uniq<SelectStatement>();
 	select->node = child->GetQueryNode();
 
 	CreateStatement stmt;
 	auto info = make_uniq<CreateTableInfo>();
+	info->catalog = catalog_name;
 	info->schema = schema_name;
 	info->table = table_name;
 	info->query = std::move(select);
diff --git a/src/duckdb/src/main/relation/insert_relation.cpp b/src/duckdb/src/main/relation/insert_relation.cpp
@@ -13,11 +13,18 @@ InsertRelation::InsertRelation(shared_ptr<Relation> child_p, string schema_name,
 	TryBindRelation(columns);
 }
 
+InsertRelation::InsertRelation(shared_ptr<Relation> child_p, string catalog_name, string schema_name, string table_name)
+    : Relation(child_p->context, RelationType::INSERT_RELATION), child(std::move(child_p)),
+      catalog_name(std::move(catalog_name)), schema_name(std::move(schema_name)), table_name(std::move(table_name)) {
+	TryBindRelation(columns);
+}
+
 BoundStatement InsertRelation::Bind(Binder &binder) {
 	InsertStatement stmt;
 	auto select = make_uniq<SelectStatement>();
 	select->node = child->GetQueryNode();
 
+	stmt.catalog = catalog_name;
 	stmt.schema = schema_name;
 	stmt.table = table_name;
 	stmt.select_statement = std::move(select);
diff --git a/src/duckdb/src/main/relation/table_relation.cpp b/src/duckdb/src/main/relation/table_relation.cpp
@@ -3,6 +3,7 @@
 #include "duckdb/parser/query_node/select_node.hpp"
 #include "duckdb/parser/expression/star_expression.hpp"
 #include "duckdb/main/relation/delete_relation.hpp"
+#include "duckdb/main/relation/value_relation.hpp"
 #include "duckdb/main/relation/update_relation.hpp"
 #include "duckdb/parser/parser.hpp"
 #include "duckdb/main/client_context.hpp"
@@ -87,4 +88,17 @@ void TableRelation::Delete(const string &condition) {
 	del->Execute();
 }
 
+void TableRelation::Insert(const vector<vector<Value>> &values) {
+	vector<string> column_names;
+	auto rel = make_shared_ptr<ValueRelation>(context->GetContext(), values, std::move(column_names), "values");
+	rel->Insert(description->database, description->schema, description->table);
+}
+
+void TableRelation::Insert(vector<vector<unique_ptr<ParsedExpression>>> &&expressions) {
+	vector<string> column_names;
+	auto rel = make_shared_ptr<ValueRelation>(context->GetContext(), std::move(expressions), std::move(column_names),
+	                                          "values");
+	rel->Insert(description->database, description->schema, description->table);
+}
+
 } // namespace duckdb

Original file line number	Diff line number	Diff line change
`@@ -164,6 +164,11 @@ class HashJoinGlobalSinkState : public GlobalSinkState {`
`164`	`164`	`}`
`165`	`165`	`}`
`166`	`166`
	`167`	`+ ~HashJoinGlobalSinkState() override {`
	`168`	`+ DUCKDB_LOG(context, PhysicalOperatorLogType, op, "PhysicalHashJoin", "GetData",`
	`169`	`+ {{"total_probe_matches", to_string(hash_table->total_probe_matches)}});`
	`170`	`+ }`
	`171`	`+`
`167`	`172`	`void ScheduleFinalize(Pipeline &pipeline, Event &event);`
`168`	`173`	`void InitializeProbeSpill();`
`169`	`174`