duckdb 0.5.2-dev1092.0 → 0.5.2-dev1104.0

package/package.json

@@ -2,7 +2,7 @@
   "name": "duckdb",
   "main": "./lib/duckdb.js",
   "types": "./lib/duckdb.d.ts",
-  "version": "0.5.2-dev1092.0",
+  "version": "0.5.2-dev1104.0",
   "description": "DuckDB node.js API",
   "gypfile": true,
   "dependencies": {

package/src/duckdb.cpp

@@ -58640,19 +58640,6 @@ void Construct(vector<Key> &keys, row_t *row_ids, Node *&node, KeySection &key_s
 	}
 }
 
-void FindFirstNotNullKey(vector<Key> &keys, bool &skipped_all_nulls, idx_t &start_idx) {
-
-	if (!skipped_all_nulls) {
-		for (idx_t i = 0; i < keys.size(); i++) {
-			if (!keys[i].Empty()) {
-				start_idx = i;
-				skipped_all_nulls = true;
-				return;
-			}
-		}
-	}
-}
-
 void ART::ConstructAndMerge(IndexLock &lock, PayloadScanner &scanner, Allocator &allocator) {
 
 	auto payload_types = logical_types;
@@ -58661,7 +58648,6 @@ void ART::ConstructAndMerge(IndexLock &lock, PayloadScanner &scanner, Allocator
 	ArenaAllocator arena_allocator(allocator);
 	vector<Key> keys(STANDARD_VECTOR_SIZE);
 
-	auto skipped_all_nulls = false;
 	auto temp_art = make_unique<ART>(this->column_ids, this->table_io_manager, this->unbound_expressions,
 	                                 this->constraint_type, this->db);
 
@@ -58686,22 +58672,6 @@ void ART::ConstructAndMerge(IndexLock &lock, PayloadScanner &scanner, Allocator
 		arena_allocator.Reset();
 		GenerateKeys(arena_allocator, ordered_chunk, keys);
 
-		// we order NULLS FIRST, so we might have to skip nulls at the start of our sorted data
-		idx_t start_idx = 0;
-		FindFirstNotNullKey(keys, skipped_all_nulls, start_idx);
-
-		if (start_idx != 0 && IsPrimary()) {
-			throw ConstraintException("NULLs in new data violate the primary key constraint of the index");
-		}
-
-		if (!skipped_all_nulls) {
-			if (IsPrimary()) {
-				// chunk consists only of NULLs
-				throw ConstraintException("NULLs in new data violate the primary key constraint of the index");
-			}
-			continue;
-		}
-
 		// prepare the row_identifiers
 		row_identifiers.Flatten(ordered_chunk.size());
 		auto row_ids = FlatVector::GetData<row_t>(row_identifiers);
@@ -58709,7 +58679,7 @@ void ART::ConstructAndMerge(IndexLock &lock, PayloadScanner &scanner, Allocator
 		// construct the ART of this chunk
 		auto art = make_unique<ART>(this->column_ids, this->table_io_manager, this->unbound_expressions,
 		                            this->constraint_type, this->db);
-		auto key_section = KeySection(start_idx, ordered_chunk.size() - 1, 0, 0);
+		auto key_section = KeySection(0, ordered_chunk.size() - 1, 0, 0);
 		auto has_constraint = IsUnique();
 		Construct(keys, row_ids, art->tree, key_section, has_constraint);
 
@@ -82706,17 +82676,22 @@ unique_ptr<PhysicalOperator> PhysicalPlanGenerator::CreatePlan(LogicalCreate &op
 
 
 
+
+
+
 namespace duckdb {
 
 unique_ptr<PhysicalOperator> PhysicalPlanGenerator::CreatePlan(LogicalCreateIndex &op) {
 
 	D_ASSERT(op.children.empty());
 
+	// table scan operator for index key columns and row IDs
 	unique_ptr<TableFilterSet> table_filters;
 	op.info->column_ids.emplace_back(COLUMN_IDENTIFIER_ROW_ID);
 
 	auto &bind_data = (TableScanBindData &)*op.bind_data;
 	bind_data.is_create_index = true;
+
 	auto table_scan =
 	    make_unique<PhysicalTableScan>(op.info->scan_types, op.function, move(op.bind_data), op.info->column_ids,
 	                                   op.info->names, move(table_filters), op.estimated_cardinality);
@@ -82724,10 +82699,32 @@ unique_ptr<PhysicalOperator> PhysicalPlanGenerator::CreatePlan(LogicalCreateInde
 	dependencies.insert(&op.table);
 	op.info->column_ids.pop_back();
 
+	D_ASSERT(op.info->scan_types.size() - 1 <= op.info->names.size());
+	D_ASSERT(op.info->scan_types.size() - 1 <= op.info->column_ids.size());
+
+	// filter operator for IS_NOT_NULL on each key column
+	vector<LogicalType> filter_types;
+	vector<unique_ptr<Expression>> filter_select_list;
+
+	for (idx_t i = 0; i < op.info->scan_types.size() - 1; i++) {
+		filter_types.push_back(op.info->scan_types[i]);
+		auto is_not_null_expr =
+		    make_unique<BoundOperatorExpression>(ExpressionType::OPERATOR_IS_NOT_NULL, LogicalType::BOOLEAN);
+		auto bound_ref =
+		    make_unique<BoundReferenceExpression>(op.info->names[op.info->column_ids[i]], op.info->scan_types[i], i);
+		is_not_null_expr->children.push_back(move(bound_ref));
+		filter_select_list.push_back(move(is_not_null_expr));
+	}
+
+	auto null_filter = make_unique<PhysicalFilter>(move(filter_types), move(filter_select_list), STANDARD_VECTOR_SIZE);
+	null_filter->types.emplace_back(LogicalType::ROW_TYPE);
+	null_filter->children.push_back(move(table_scan));
+
+	// actual physical create index operator
 	auto physical_create_index =
 	    make_unique<PhysicalCreateIndex>(op, op.table, op.info->column_ids, move(op.expressions), move(op.info),
 	                                     move(op.unbound_expressions), op.estimated_cardinality);
-	physical_create_index->children.push_back(move(table_scan));
+	physical_create_index->children.push_back(move(null_filter));
 	return move(physical_create_index);
 }
 
@@ -114079,12 +114076,11 @@ static idx_t LevenshteinDistance(const string_t &txt, const string_t &tgt) {
 	auto txt_len = txt.GetSize();
 	auto tgt_len = tgt.GetSize();
 
-	if (txt_len < 1) {
-		throw InvalidInputException("Levenshtein Function: 1st argument too short");
-	}
-
-	if (tgt_len < 1) {
-		throw InvalidInputException("Levenshtein Function: 2nd argument too short");
+	// If one string is empty, the distance equals the length of the other string
+	if (txt_len == 0) {
+		return tgt_len;
+	} else if (tgt_len == 0) {
+		return txt_len;
 	}
 
 	auto txt_str = txt.GetDataUnsafe();

package/src/duckdb.hpp

@@ -11,8 +11,8 @@ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLI
 #pragma once
 #define DUCKDB_AMALGAMATION 1
 #define DUCKDB_AMALGAMATION_EXTENDED 1
-#define DUCKDB_SOURCE_ID "0060a0f5a"
-#define DUCKDB_VERSION "v0.5.2-dev1092"
+#define DUCKDB_SOURCE_ID "b8ed45a8d"
+#define DUCKDB_VERSION "v0.5.2-dev1104"
 //===----------------------------------------------------------------------===//
 //                         DuckDB
 //