duckdb 0.7.1-dev37.0 → 0.7.1-dev415.0

package/src/duckdb/src/parser/statement/delete_statement.cpp

@@ -13,6 +13,9 @@ DeleteStatement::DeleteStatement(const DeleteStatement &other) : SQLStatement(ot
 	for (const auto &using_clause : other.using_clauses) {
 		using_clauses.push_back(using_clause->Copy());
 	}
+	for (auto &expr : other.returning_list) {
+		returning_list.emplace_back(expr->Copy());
+	}
 	cte_map = other.cte_map.Copy();
 }
 

package/src/duckdb/src/parser/statement/detach_statement.cpp

@@ -0,0 +1,15 @@
+#include "duckdb/parser/statement/detach_statement.hpp"
+
+namespace duckdb {
+
+DetachStatement::DetachStatement() : SQLStatement(StatementType::DETACH_STATEMENT) {
+}
+
+DetachStatement::DetachStatement(const DetachStatement &other) : SQLStatement(other), info(other.info->Copy()) {
+}
+
+unique_ptr<SQLStatement> DetachStatement::Copy() const {
+	return unique_ptr<DetachStatement>(new DetachStatement(*this));
+}
+
+} // namespace duckdb

package/src/duckdb/src/parser/statement/insert_statement.cpp

@@ -13,6 +13,9 @@ OnConflictInfo::OnConflictInfo(const OnConflictInfo &other)
 	if (other.set_info) {
 		set_info = other.set_info->Copy();
 	}
+	if (other.condition) {
+		condition = other.condition->Copy();
+	}
 }
 
 unique_ptr<OnConflictInfo> OnConflictInfo::Copy() const {
@@ -28,6 +31,12 @@ InsertStatement::InsertStatement(const InsertStatement &other)
       select_statement(unique_ptr_cast<SQLStatement, SelectStatement>(other.select_statement->Copy())),
       columns(other.columns), table(other.table), schema(other.schema), catalog(other.catalog) {
 	cte_map = other.cte_map.Copy();
+	for (auto &expr : other.returning_list) {
+		returning_list.emplace_back(expr->Copy());
+	}
+	if (other.table_ref) {
+		table_ref = other.table_ref->Copy();
+	}
 	if (other.on_conflict_info) {
 		on_conflict_info = other.on_conflict_info->Copy();
 	}

package/src/duckdb/src/parser/statement/update_statement.cpp

@@ -27,6 +27,9 @@ UpdateStatement::UpdateStatement(const UpdateStatement &other)
 	if (other.from_table) {
 		from_table = other.from_table->Copy();
 	}
+	for (auto &expr : other.returning_list) {
+		returning_list.emplace_back(expr->Copy());
+	}
 	cte_map = other.cte_map.Copy();
 }
 

package/src/duckdb/src/parser/transform/expression/transform_case.cpp

@@ -9,16 +9,16 @@ unique_ptr<ParsedExpression> Transformer::TransformCase(duckdb_libpgquery::PGCas
 	D_ASSERT(root);
 
 	auto case_node = make_unique<CaseExpression>();
+	auto root_arg = TransformExpression(reinterpret_cast<duckdb_libpgquery::PGNode *>(root->arg));
 	for (auto cell = root->args->head; cell != nullptr; cell = cell->next) {
 		CaseCheck case_check;
 
 		auto w = reinterpret_cast<duckdb_libpgquery::PGCaseWhen *>(cell->data.ptr_value);
 		auto test_raw = TransformExpression(reinterpret_cast<duckdb_libpgquery::PGNode *>(w->expr));
 		unique_ptr<ParsedExpression> test;
-		auto arg = TransformExpression(reinterpret_cast<duckdb_libpgquery::PGNode *>(root->arg));
-		if (arg) {
+		if (root_arg) {
 			case_check.when_expr =
-			    make_unique<ComparisonExpression>(ExpressionType::COMPARE_EQUAL, std::move(arg), std::move(test_raw));
+			    make_unique<ComparisonExpression>(ExpressionType::COMPARE_EQUAL, root_arg->Copy(), std::move(test_raw));
 		} else {
 			case_check.when_expr = std::move(test_raw);
 		}

package/src/duckdb/src/parser/transform/statement/transform_create_database.cpp

@@ -11,7 +11,6 @@ unique_ptr<CreateStatement> Transformer::TransformCreateDatabase(duckdb_libpgque
 	auto result = make_unique<CreateStatement>();
 	auto info = make_unique<CreateDatabaseInfo>();
 
-	info->extension_name = stmt->extension ? stmt->extension : string();
 	info->path = stmt->path ? stmt->path : string();
 
 	auto qualified_name = TransformQualifiedName(stmt->name);

package/src/duckdb/src/parser/transform/statement/transform_detach.cpp

@@ -0,0 +1,19 @@
+#include "duckdb/parser/transformer.hpp"
+#include "duckdb/parser/statement/detach_statement.hpp"
+#include "duckdb/parser/expression/constant_expression.hpp"
+#include "duckdb/common/string_util.hpp"
+
+namespace duckdb {
+
+unique_ptr<DetachStatement> Transformer::TransformDetach(duckdb_libpgquery::PGNode *node) {
+	auto stmt = reinterpret_cast<duckdb_libpgquery::PGDetachStmt *>(node);
+	auto result = make_unique<DetachStatement>();
+	auto info = make_unique<DetachInfo>();
+	info->name = stmt->db_name;
+	info->if_exists = stmt->missing_ok;
+
+	result->info = std::move(info);
+	return result;
+}
+
+} // namespace duckdb

package/src/duckdb/src/parser/transformer.cpp

@@ -145,6 +145,8 @@ unique_ptr<SQLStatement> Transformer::TransformStatementInternal(duckdb_libpgque
 		return TransformAlterSequence(stmt);
 	case duckdb_libpgquery::T_PGAttachStmt:
 		return TransformAttach(stmt);
+	case duckdb_libpgquery::T_PGDetachStmt:
+		return TransformDetach(stmt);
 	case duckdb_libpgquery::T_PGUseStmt:
 		return TransformUse(stmt);
 	case duckdb_libpgquery::T_PGCreateDatabaseStmt:

package/src/duckdb/src/planner/bind_context.cpp

@@ -44,7 +44,7 @@ vector<string> BindContext::GetSimilarBindings(const string &column_name) {
 	for (auto &kv : bindings) {
 		auto binding = kv.second.get();
 		for (auto &name : binding->names) {
-			idx_t distance = StringUtil::LevenshteinDistance(name, column_name);
+			idx_t distance = StringUtil::SimilarityScore(name, column_name);
 			scores.emplace_back(binding->alias + "." + name, distance);
 		}
 	}

package/src/duckdb/src/planner/binder/expression/bind_aggregate_expression.cpp

@@ -72,6 +72,9 @@ static void NegatePercentileFractions(ClientContext &context, unique_ptr<ParsedE
 		for (const auto &element_val : ListValue::GetChildren(value)) {
 			values.push_back(NegatePercentileValue(element_val, desc));
 		}
+		if (values.empty()) {
+			throw BinderException("Empty list in percentile not allowed");
+		}
 		bound.expr = make_unique<BoundConstantExpression>(Value::LIST(values));
 	} else {
 		bound.expr = make_unique<BoundConstantExpression>(NegatePercentileValue(value, desc));

package/src/duckdb/src/planner/binder/statement/bind_copy.cpp

@@ -23,16 +23,6 @@
 
 namespace duckdb {
 
-static vector<idx_t> ColumnListToIndices(const vector<bool> &vec) {
-	vector<idx_t> ret;
-	for (idx_t i = 0; i < vec.size(); i++) {
-		if (vec[i]) {
-			ret.push_back(i);
-		}
-	}
-	return ret;
-}
-
 vector<string> GetUniqueNames(const vector<string> &original_names) {
 	unordered_set<string> name_set;
 	vector<string> unique_names;
@@ -97,22 +87,25 @@ BoundStatement Binder::BindCopyTo(CopyStatement &stmt) {
 	for (auto &option : original_options) {
 		auto loption = StringUtil::Lower(option.first);
 		if (loption == "use_tmp_file") {
-			use_tmp_file = option.second[0].CastAs(context, LogicalType::BOOLEAN).GetValue<bool>();
+			use_tmp_file =
+			    option.second.empty() || option.second[0].CastAs(context, LogicalType::BOOLEAN).GetValue<bool>();
 			user_set_use_tmp_file = true;
 			continue;
 		}
 		if (loption == "allow_overwrite") {
-			allow_overwrite = option.second[0].CastAs(context, LogicalType::BOOLEAN).GetValue<bool>();
+			allow_overwrite =
+			    option.second.empty() || option.second[0].CastAs(context, LogicalType::BOOLEAN).GetValue<bool>();
 			continue;
 		}
 
 		if (loption == "per_thread_output") {
-			per_thread_output = option.second[0].CastAs(context, LogicalType::BOOLEAN).GetValue<bool>();
+			per_thread_output =
+			    option.second.empty() || option.second[0].CastAs(context, LogicalType::BOOLEAN).GetValue<bool>();
 			continue;
 		}
 		if (loption == "partition_by") {
 			auto converted = ConvertVectorToValue(std::move(option.second));
-			partition_cols = ColumnListToIndices(ParseColumnList(converted, select_node.names, loption));
+			partition_cols = ParseColumnsOrdered(converted, select_node.names, loption);
 			continue;
 		}
 		stmt.info->options[option.first] = option.second;

package/src/duckdb/src/planner/binder/statement/bind_create.cpp

@@ -12,13 +12,11 @@
 #include "duckdb/parser/parsed_data/create_macro_info.hpp"
 #include "duckdb/parser/parsed_data/create_view_info.hpp"
 #include "duckdb/parser/parsed_data/create_database_info.hpp"
-#include "duckdb/function/create_database_extension.hpp"
 #include "duckdb/parser/tableref/table_function_ref.hpp"
 #include "duckdb/parser/parsed_expression_iterator.hpp"
 #include "duckdb/parser/statement/create_statement.hpp"
 #include "duckdb/planner/binder.hpp"
 #include "duckdb/planner/bound_query_node.hpp"
-#include "duckdb/planner/expression_binder/aggregate_binder.hpp"
 #include "duckdb/planner/expression_binder/index_binder.hpp"
 #include "duckdb/planner/expression_binder/select_binder.hpp"
 #include "duckdb/planner/operator/logical_create.hpp"
@@ -26,13 +24,13 @@
 #include "duckdb/planner/operator/logical_create_table.hpp"
 #include "duckdb/planner/operator/logical_get.hpp"
 #include "duckdb/planner/operator/logical_distinct.hpp"
-#include "duckdb/planner/parsed_data/bound_create_function_info.hpp"
 #include "duckdb/planner/parsed_data/bound_create_table_info.hpp"
 #include "duckdb/planner/query_node/bound_select_node.hpp"
 #include "duckdb/planner/tableref/bound_basetableref.hpp"
 #include "duckdb/parser/constraints/foreign_key_constraint.hpp"
 #include "duckdb/function/scalar_macro_function.hpp"
 #include "duckdb/storage/data_table.hpp"
+#include "duckdb/storage/storage_extension.hpp"
 #include "duckdb/main/client_data.hpp"
 #include "duckdb/parser/constraints/unique_constraint.hpp"
 #include "duckdb/parser/constraints/list.hpp"
@@ -651,23 +649,27 @@ BoundStatement Binder::Bind(CreateStatement &stmt) {
 	case CatalogType::DATABASE_ENTRY: {
 		// not supported in DuckDB yet but allow extensions to intercept and implement this functionality
 		auto &base = (CreateDatabaseInfo &)*stmt.info;
-		string extension_name = base.extension_name;
 		string database_name = base.name;
 		string source_path = base.path;
 
 		auto &config = DBConfig::GetConfig(context);
-		for (auto &extension : config.create_database_extensions) {
-			auto create_database_function_ref =
-			    extension.function(context, extension_name, database_name, source_path, extension.data.get());
-			if (create_database_function_ref) {
-				auto bound_create_database_func = Bind(*create_database_function_ref);
-				result.plan = CreatePlan(*bound_create_database_func);
-				break;
-			}
-		}
-		if (!result.plan) {
+
+		if (config.storage_extensions.empty()) {
 			throw NotImplementedException("CREATE DATABASE not supported in DuckDB yet");
 		}
+		// for now assume only one storage extension provides the custom create_database impl
+		for (auto &extension_entry : config.storage_extensions) {
+			if (extension_entry.second->create_database != nullptr) {
+				auto &storage_extension = extension_entry.second;
+				auto create_database_function_ref = storage_extension->create_database(
+				    storage_extension->storage_info.get(), context, database_name, source_path);
+				if (create_database_function_ref) {
+					auto bound_create_database_func = Bind(*create_database_function_ref);
+					result.plan = CreatePlan(*bound_create_database_func);
+					break;
+				}
+			}
+		}
 		break;
 	}
 	default:

package/src/duckdb/src/planner/binder/statement/bind_create_table.cpp

@@ -16,6 +16,8 @@
 #include "duckdb/parser/expression/list.hpp"
 #include "duckdb/common/index_map.hpp"
 #include "duckdb/planner/expression_iterator.hpp"
+#include "duckdb/planner/expression_binder/index_binder.hpp"
+#include "duckdb/parser/parsed_data/create_index_info.hpp"
 
 #include <algorithm>
 
@@ -300,4 +302,15 @@ unique_ptr<BoundCreateTableInfo> Binder::BindCreateTableInfo(unique_ptr<CreateIn
 	return BindCreateTableInfo(std::move(info), schema);
 }
 
+vector<unique_ptr<Expression>> Binder::BindCreateIndexExpressions(TableCatalogEntry *table, CreateIndexInfo *info) {
+	vector<unique_ptr<Expression>> expressions;
+
+	auto index_binder = IndexBinder(*this, this->context, table, info);
+	for (auto &expr : info->expressions) {
+		expressions.push_back(index_binder.Bind(expr));
+	}
+
+	return expressions;
+}
+
 } // namespace duckdb

package/src/duckdb/src/planner/binder/statement/bind_detach.cpp

@@ -0,0 +1,19 @@
+#include "duckdb/parser/statement/detach_statement.hpp"
+#include "duckdb/planner/binder.hpp"
+#include "duckdb/planner/operator/logical_simple.hpp"
+#include "duckdb/main/config.hpp"
+
+namespace duckdb {
+
+BoundStatement Binder::Bind(DetachStatement &stmt) {
+	BoundStatement result;
+
+	result.plan = make_unique<LogicalSimple>(LogicalOperatorType::LOGICAL_DETACH, std::move(stmt.info));
+	result.names = {"Success"};
+	result.types = {LogicalType::BOOLEAN};
+	properties.allow_stream_result = false;
+	properties.return_type = StatementReturnType::NOTHING;
+	return result;
+}
+
+} // namespace duckdb

package/src/duckdb/src/planner/binder/statement/bind_drop.cpp

@@ -1,9 +1,13 @@
 #include "duckdb/parser/statement/drop_statement.hpp"
 #include "duckdb/planner/binder.hpp"
+#include "duckdb/planner/bound_tableref.hpp"
 #include "duckdb/planner/operator/logical_simple.hpp"
 #include "duckdb/catalog/catalog.hpp"
 #include "duckdb/catalog/standard_entry.hpp"
 #include "duckdb/catalog/catalog_entry/schema_catalog_entry.hpp"
+#include "duckdb/parser/parsed_data/drop_info.hpp"
+#include "duckdb/main/config.hpp"
+#include "duckdb/storage/storage_extension.hpp"
 
 namespace duckdb {
 
@@ -43,10 +47,31 @@ BoundStatement Binder::Bind(DropStatement &stmt) {
 		stmt.info->schema = entry->schema->name;
 		break;
 	}
-	case CatalogType::DATABASE_ENTRY:
-		// attaching and detaching is read-only
-		stmt.info->catalog = SYSTEM_CATALOG;
-		break;
+	case CatalogType::DATABASE_ENTRY: {
+		auto &base = (DropInfo &)*stmt.info;
+		string database_name = base.name;
+
+		auto &config = DBConfig::GetConfig(context);
+		// for now assume only one storage extension provides the custom drop_database impl
+		for (auto &extension_entry : config.storage_extensions) {
+			if (extension_entry.second->drop_database == nullptr) {
+				continue;
+			}
+			auto &storage_extension = extension_entry.second;
+			auto drop_database_function_ref =
+			    storage_extension->drop_database(storage_extension->storage_info.get(), context, database_name);
+			if (drop_database_function_ref) {
+				auto bound_drop_database_func = Bind(*drop_database_function_ref);
+				result.plan = CreatePlan(*bound_drop_database_func);
+				result.names = {"Success"};
+				result.types = {LogicalType::BIGINT};
+				properties.allow_stream_result = false;
+				properties.return_type = StatementReturnType::NOTHING;
+				return result;
+			}
+		}
+		throw BinderException("Drop is not supported for this database!");
+	}
 	default:
 		throw BinderException("Unknown catalog type for drop statement!");
 	}

package/src/duckdb/src/planner/binder/statement/bind_insert.cpp

@@ -301,7 +301,28 @@ void Binder::BindOnConflictClause(LogicalInsert &insert, TableCatalogEntry &tabl
 		insert.on_conflict_condition = std::move(condition);
 	}
 
-	auto projection_index = insert.children[0]->GetTableIndex()[0];
+	auto bindings = insert.children[0]->GetColumnBindings();
+	idx_t projection_index = DConstants::INVALID_INDEX;
+	std::vector<unique_ptr<LogicalOperator>> *insert_child_operators;
+	insert_child_operators = &insert.children;
+	while (projection_index == DConstants::INVALID_INDEX) {
+		if (insert_child_operators->empty()) {
+			// No further children to visit
+			break;
+		}
+		D_ASSERT(insert_child_operators->size() >= 1);
+		auto &current_child = (*insert_child_operators)[0];
+		auto table_indices = current_child->GetTableIndex();
+		if (table_indices.empty()) {
+			// This operator does not have a table index to refer to, we have to visit its children
+			insert_child_operators = &current_child->children;
+			continue;
+		}
+		projection_index = table_indices[0];
+	}
+	if (projection_index == DConstants::INVALID_INDEX) {
+		throw InternalException("Could not locate a table_index from the children of the insert");
+	}
 
 	string unused;
 	auto original_binding = bind_context.GetBinding(table_alias, unused);

package/src/duckdb/src/planner/binder.cpp

@@ -90,6 +90,8 @@ BoundStatement Binder::Bind(SQLStatement &statement) {
 		return Bind((LogicalPlanStatement &)statement);
 	case StatementType::ATTACH_STATEMENT:
 		return Bind((AttachStatement &)statement);
+	case StatementType::DETACH_STATEMENT:
+		return Bind((DetachStatement &)statement);
 	default: // LCOV_EXCL_START
 		throw NotImplementedException("Unimplemented statement type \"%s\" for Bind",
 		                              StatementTypeToString(statement.type));

package/src/duckdb/src/planner/expression_binder/index_binder.cpp

@@ -1,8 +1,14 @@
 #include "duckdb/planner/expression_binder/index_binder.hpp"
 
+#include "duckdb/parser/parsed_data/create_index_info.hpp"
+#include "duckdb/parser/expression/columnref_expression.hpp"
+#include "duckdb/planner/expression/bound_columnref_expression.hpp"
+#include "duckdb/planner/column_binding.hpp"
+
 namespace duckdb {
 
-IndexBinder::IndexBinder(Binder &binder, ClientContext &context) : ExpressionBinder(binder, context) {
+IndexBinder::IndexBinder(Binder &binder, ClientContext &context, TableCatalogEntry *table, CreateIndexInfo *info)
+    : ExpressionBinder(binder, context), table(table), info(info) {
 }
 
 BindResult IndexBinder::BindExpression(unique_ptr<ParsedExpression> *expr_ptr, idx_t depth, bool root_expression) {
@@ -12,6 +18,31 @@ BindResult IndexBinder::BindExpression(unique_ptr<ParsedExpression> *expr_ptr, i
 		return BindResult("window functions are not allowed in index expressions");
 	case ExpressionClass::SUBQUERY:
 		return BindResult("cannot use subquery in index expressions");
+	case ExpressionClass::COLUMN_REF: {
+		if (table) {
+			// WAL replay
+			// we assume that the parsed expressions have qualified column names
+			// and that the columns exist in the table
+			auto &col_ref = (ColumnRefExpression &)expr;
+			auto col_idx = table->GetColumnIndex(col_ref.column_names.back());
+			auto col_type = table->GetColumn(col_idx).GetType();
+
+			// find the col_idx in the index.column_ids
+			auto col_id_idx = DConstants::INVALID_INDEX;
+			for (idx_t i = 0; i < info->column_ids.size(); i++) {
+				if (col_idx.index == info->column_ids[i]) {
+					col_id_idx = i;
+				}
+			}
+
+			if (col_id_idx == DConstants::INVALID_INDEX) {
+				throw InternalException("failed to replay CREATE INDEX statement - column id not found");
+			}
+			return BindResult(
+			    make_unique<BoundColumnRefExpression>(col_ref.alias, col_type, ColumnBinding(0, col_id_idx)));
+		}
+		return ExpressionBinder::BindExpression(expr_ptr, depth);
+	}
 	default:
 		return ExpressionBinder::BindExpression(expr_ptr, depth);
 	}

package/src/duckdb/src/planner/logical_operator.cpp

@@ -57,6 +57,7 @@ void LogicalOperator::ResolveOperatorTypes() {
 
 vector<ColumnBinding> LogicalOperator::GenerateColumnBindings(idx_t table_idx, idx_t column_count) {
 	vector<ColumnBinding> result;
+	result.reserve(column_count);
 	for (idx_t i = 0; i < column_count; i++) {
 		result.emplace_back(table_idx, i);
 	}
@@ -84,6 +85,7 @@ vector<ColumnBinding> LogicalOperator::MapBindings(const vector<ColumnBinding> &
 		vector<ColumnBinding> result_bindings;
 		result_bindings.reserve(projection_map.size());
 		for (auto index : projection_map) {
+			D_ASSERT(index < bindings.size());
 			result_bindings.push_back(bindings[index]);
 		}
 		return result_bindings;
@@ -165,7 +167,8 @@ idx_t LogicalOperator::EstimateCardinality(ClientContext &context) {
 		max_cardinality = MaxValue(child->EstimateCardinality(context), max_cardinality);
 	}
 	has_estimated_cardinality = true;
-	return max_cardinality;
+	estimated_cardinality = max_cardinality;
+	return estimated_cardinality;
 }
 
 void LogicalOperator::Print() {
@@ -338,6 +341,8 @@ unique_ptr<LogicalOperator> LogicalOperator::Deserialize(Deserializer &deseriali
 	case LogicalOperatorType::LOGICAL_DROP:
 		result = LogicalSimple::Deserialize(state, reader);
 		break;
+	case LogicalOperatorType::LOGICAL_DETACH:
+		throw SerializationException("Logical Detach does not support serialization");
 	case LogicalOperatorType::LOGICAL_EXTENSION_OPERATOR:
 		result = LogicalExtensionOperator::Deserialize(state, reader);
 		break;

package/src/duckdb/src/planner/planner.cpp

@@ -133,6 +133,7 @@ void Planner::CreatePlan(unique_ptr<SQLStatement> statement) {
 	case StatementType::EXECUTE_STATEMENT:
 	case StatementType::LOGICAL_PLAN_STATEMENT:
 	case StatementType::ATTACH_STATEMENT:
+	case StatementType::DETACH_STATEMENT:
 		CreatePlan(*statement);
 		break;
 	default: