duckdb 0.7.2-dev3515.0 → 0.7.2-dev3666.0

package/src/duckdb/extension/json/json_functions/json_create.cpp

@@ -494,9 +494,9 @@ static void CreateValues(const StructNames &names, yyjson_mut_doc *doc, yyjson_m
 
 static void ObjectFunction(DataChunk &args, ExpressionState &state, Vector &result) {
 	auto &func_expr = state.expr.Cast<BoundFunctionExpression>();
-	const auto &info = (JSONCreateFunctionData &)*func_expr.bind_info;
+	const auto &info = func_expr.bind_info->Cast<JSONCreateFunctionData>();
 	auto &lstate = JSONFunctionLocalState::ResetAndGet(state);
-	auto alc = lstate.json_allocator.GetYYJSONAllocator();
+	auto alc = lstate.json_allocator.GetYYAlc();
 
 	// Initialize values
 	const idx_t count = args.size();
@@ -526,9 +526,9 @@ static void ObjectFunction(DataChunk &args, ExpressionState &state, Vector &resu
 
 static void ArrayFunction(DataChunk &args, ExpressionState &state, Vector &result) {
 	auto &func_expr = state.expr.Cast<BoundFunctionExpression>();
-	const auto &info = (JSONCreateFunctionData &)*func_expr.bind_info;
+	const auto &info = func_expr.bind_info->Cast<JSONCreateFunctionData>();
 	auto &lstate = JSONFunctionLocalState::ResetAndGet(state);
-	auto alc = lstate.json_allocator.GetYYJSONAllocator();
+	auto alc = lstate.json_allocator.GetYYAlc();
 
 	// Initialize arrays
 	const idx_t count = args.size();
@@ -585,9 +585,9 @@ static void ToJSONFunctionInternal(const StructNames &names, Vector &input, cons
 
 static void ToJSONFunction(DataChunk &args, ExpressionState &state, Vector &result) {
 	auto &func_expr = state.expr.Cast<BoundFunctionExpression>();
-	const auto &info = (JSONCreateFunctionData &)*func_expr.bind_info;
+	const auto &info = func_expr.bind_info->Cast<JSONCreateFunctionData>();
 	auto &lstate = JSONFunctionLocalState::ResetAndGet(state);
-	auto alc = lstate.json_allocator.GetYYJSONAllocator();
+	auto alc = lstate.json_allocator.GetYYAlc();
 
 	ToJSONFunctionInternal(info.const_struct_names, args.data[0], args.size(), result, alc);
 }
@@ -649,7 +649,7 @@ public:
 static bool AnyToJSONCast(Vector &source, Vector &result, idx_t count, CastParameters &parameters) {
 	auto &lstate = parameters.local_state->Cast<JSONFunctionLocalState>();
 	lstate.json_allocator.Reset();
-	auto alc = lstate.json_allocator.GetYYJSONAllocator();
+	auto alc = lstate.json_allocator.GetYYAlc();
 	const auto &names = parameters.cast_data->Cast<NestedToJSONCastData>().const_struct_names;
 
 	ToJSONFunctionInternal(names, source, count, result, alc);

package/src/duckdb/extension/json/json_functions/json_merge_patch.cpp

@@ -3,8 +3,8 @@
 
 namespace duckdb {
 
-static duckdb::unique_ptr<FunctionData> JSONMergePatchBind(ClientContext &context, ScalarFunction &bound_function,
-                                                           vector<duckdb::unique_ptr<Expression>> &arguments) {
+static unique_ptr<FunctionData> JSONMergePatchBind(ClientContext &context, ScalarFunction &bound_function,
+                                                   vector<unique_ptr<Expression>> &arguments) {
 	if (arguments.size() < 2) {
 		throw InvalidInputException("json_merge_patch requires at least two parameters");
 	}
@@ -53,9 +53,9 @@ static inline void ReadObjects(yyjson_mut_doc *doc, Vector &input, yyjson_mut_va
 //! Follows MySQL behaviour
 static void MergePatchFunction(DataChunk &args, ExpressionState &state, Vector &result) {
 	auto &lstate = JSONFunctionLocalState::ResetAndGet(state);
-	auto alc = lstate.json_allocator.GetYYJSONAllocator();
+	auto alc = lstate.json_allocator.GetYYAlc();
 
-	auto doc = JSONCommon::CreateDocument(lstate.json_allocator.GetYYJSONAllocator());
+	auto doc = JSONCommon::CreateDocument(lstate.json_allocator.GetYYAlc());
 	const auto count = args.size();
 
 	// Read the first json arg

package/src/duckdb/extension/json/json_functions/json_serialize_sql.cpp

@@ -76,7 +76,7 @@ static unique_ptr<FunctionData> JsonSerializeBind(ClientContext &context, Scalar
 
 static void JsonSerializeFunction(DataChunk &args, ExpressionState &state, Vector &result) {
 	auto &local_state = JSONFunctionLocalState::ResetAndGet(state);
-	auto alc = local_state.json_allocator.GetYYJSONAllocator();
+	auto alc = local_state.json_allocator.GetYYAlc();
 	auto &inputs = args.data[0];
 
 	auto &func_expr = state.expr.Cast<BoundFunctionExpression>();
@@ -193,7 +193,7 @@ static unique_ptr<SelectStatement> DeserializeSelectStatement(string_t input, yy
 static void JsonDeserializeFunction(DataChunk &args, ExpressionState &state, Vector &result) {
 
 	auto &local_state = JSONFunctionLocalState::ResetAndGet(state);
-	auto alc = local_state.json_allocator.GetYYJSONAllocator();
+	auto alc = local_state.json_allocator.GetYYAlc();
 	auto &inputs = args.data[0];
 
 	UnaryExecutor::Execute<string_t, string_t>(inputs, result, args.size(), [&](string_t input) {
@@ -214,7 +214,7 @@ ScalarFunctionSet JSONFunctions::GetDeserializeSqlFunction() {
 //----------------------------------------------------------------------
 static string ExecuteJsonSerializedSqlPragmaFunction(ClientContext &context, const FunctionParameters &parameters) {
 	JSONFunctionLocalState local_state(context);
-	auto alc = local_state.json_allocator.GetYYJSONAllocator();
+	auto alc = local_state.json_allocator.GetYYAlc();
 
 	auto input = parameters.values[0].GetValueUnsafe<string_t>();
 	auto stmt = DeserializeSelectStatement(input, alc);
@@ -239,7 +239,7 @@ struct ExecuteSqlTableFunction {
 	static unique_ptr<FunctionData> Bind(ClientContext &context, TableFunctionBindInput &input,
 	                                     vector<LogicalType> &return_types, vector<string> &names) {
 		JSONFunctionLocalState local_state(context);
-		auto alc = local_state.json_allocator.GetYYJSONAllocator();
+		auto alc = local_state.json_allocator.GetYYAlc();
 
 		auto result = make_uniq<BindData>();
 

package/src/duckdb/extension/json/json_functions/json_structure.cpp

@@ -1,9 +1,9 @@
 #include "json_structure.hpp"
 
+#include "duckdb/common/enum_util.hpp"
 #include "json_executors.hpp"
 #include "json_scan.hpp"
 #include "json_transform.hpp"
-#include "duckdb/common/enum_util.hpp"
 
 namespace duckdb {
 
@@ -93,7 +93,7 @@ bool JSONStructureNode::ContainsVarchar() const {
 }
 
 void JSONStructureNode::InitializeCandidateTypes(const idx_t max_depth, idx_t depth) {
-	if (depth > max_depth) {
+	if (depth >= max_depth) {
 		return;
 	}
 	if (descriptions.size() != 1) {
@@ -521,11 +521,11 @@ static LogicalType StructureToTypeString(const JSONStructureNode &node) {
 
 LogicalType JSONStructure::StructureToType(ClientContext &context, const JSONStructureNode &node, const idx_t max_depth,
                                            idx_t depth) {
-	if (depth > max_depth) {
+	if (depth >= max_depth) {
 		return JSONCommon::JSONType();
 	}
 	if (node.descriptions.empty()) {
-		return LogicalTypeId::SQLNULL;
+		return JSONCommon::JSONType();
 	}
 	if (node.descriptions.size() != 1) { // Inconsistent types, so we resort to JSON
 		return JSONCommon::JSONType();
@@ -540,7 +540,9 @@ LogicalType JSONStructure::StructureToType(ClientContext &context, const JSONStr
 	case LogicalTypeId::VARCHAR:
 		return StructureToTypeString(node);
 	case LogicalTypeId::SQLNULL:
-		return LogicalTypeId::INTEGER;
+		return JSONCommon::JSONType();
+	case LogicalTypeId::UBIGINT:
+		return LogicalTypeId::BIGINT; // We prefer not to return UBIGINT in our type auto-detection
 	default:
 		return desc.type;
 	}

package/src/duckdb/extension/json/json_functions/json_transform.cpp

@@ -20,7 +20,7 @@ JSONTransformOptions::JSONTransformOptions(bool strict_cast_p, bool error_duplic
       error_unknown_key(error_unkown_key_p) {
 }
 
-void JSONTransformOptions::Serialize(FieldWriter &writer) {
+void JSONTransformOptions::Serialize(FieldWriter &writer) const {
 	writer.WriteField(strict_cast);
 	writer.WriteField(error_duplicate_key);
 	writer.WriteField(error_missing_key);
@@ -98,8 +98,8 @@ static unique_ptr<FunctionData> JSONTransformBind(ClientContext &context, Scalar
 		auto structure_string = structure_val.GetValueUnsafe<string_t>();
 		JSONAllocator json_allocator(Allocator::DefaultAllocator());
 		yyjson_read_err err;
-		auto doc = JSONCommon::ReadDocumentUnsafe(structure_string, JSONCommon::READ_FLAG,
-		                                          json_allocator.GetYYJSONAllocator(), &err);
+		auto doc =
+		    JSONCommon::ReadDocumentUnsafe(structure_string, JSONCommon::READ_FLAG, json_allocator.GetYYAlc(), &err);
 		if (err.code != YYJSON_READ_SUCCESS) {
 			JSONCommon::ThrowParseError(structure_string.GetData(), structure_string.GetSize(), err);
 		}
@@ -615,7 +615,7 @@ static bool TransformObjectToMap(yyjson_val *objects[], yyjson_alc *alc, Vector
 	ListVector::SetListSize(result, list_size);
 
 	auto list_entries = FlatVector::GetData<list_entry_t>(result);
-	auto list_validity = FlatVector::Validity(result);
+	auto &list_validity = FlatVector::Validity(result);
 
 	auto keys = (yyjson_val **)alc->malloc(alc->ctx, sizeof(yyjson_val *) * list_size);
 	auto vals = (yyjson_val **)alc->malloc(alc->ctx, sizeof(yyjson_val *) * list_size);
@@ -654,6 +654,7 @@ static bool TransformObjectToMap(yyjson_val *objects[], yyjson_alc *alc, Vector
 			list_offset++;
 		}
 	}
+	D_ASSERT(list_offset == list_size);
 
 	// Transform keys
 	if (!JSONTransform::Transform(keys, alc, MapVector::GetKeys(result), list_size, options)) {
@@ -678,7 +679,7 @@ bool TransformToJSON(yyjson_val *vals[], yyjson_alc *alc, Vector &result, const
 	auto &validity = FlatVector::Validity(result);
 	for (idx_t i = 0; i < count; i++) {
 		const auto &val = vals[i];
-		if (!val) {
+		if (!val || unsafe_yyjson_is_null(val)) {
 			validity.SetInvalid(i);
 		} else {
 			data[i] = JSONCommon::WriteVal(val, alc);
@@ -691,7 +692,8 @@ bool TransformToJSON(yyjson_val *vals[], yyjson_alc *alc, Vector &result, const
 bool JSONTransform::Transform(yyjson_val *vals[], yyjson_alc *alc, Vector &result, const idx_t count,
                               JSONTransformOptions &options) {
 	auto result_type = result.GetType();
-	if ((result_type == LogicalTypeId::TIMESTAMP || result_type == LogicalTypeId::DATE) && options.date_format_map) {
+	if ((result_type == LogicalTypeId::TIMESTAMP || result_type == LogicalTypeId::DATE) && options.date_format_map &&
+	    options.date_format_map->HasFormats(result_type.id())) {
 		// Auto-detected date/timestamp format during sampling
 		return TransformFromStringWithFormat(vals, result, count, options);
 	}
@@ -803,7 +805,7 @@ static bool TransformFunctionInternal(Vector &input, const idx_t count, Vector &
 template <bool strict>
 static void TransformFunction(DataChunk &args, ExpressionState &state, Vector &result) {
 	auto &lstate = JSONFunctionLocalState::ResetAndGet(state);
-	auto alc = lstate.json_allocator.GetYYJSONAllocator();
+	auto alc = lstate.json_allocator.GetYYAlc();
 
 	JSONTransformOptions options(strict, strict, strict, false);
 	if (!TransformFunctionInternal(args.data[0], args.size(), result, alc, options)) {
@@ -838,7 +840,7 @@ ScalarFunctionSet JSONFunctions::GetTransformStrictFunction() {
 static bool JSONToAnyCast(Vector &source, Vector &result, idx_t count, CastParameters &parameters) {
 	auto &lstate = parameters.local_state->Cast<JSONFunctionLocalState>();
 	lstate.json_allocator.Reset();
-	auto alc = lstate.json_allocator.GetYYJSONAllocator();
+	auto alc = lstate.json_allocator.GetYYAlc();
 
 	JSONTransformOptions options(true, true, true, true);
 	options.delay_error = true;

package/src/duckdb/extension/json/json_functions/json_valid.cpp

@@ -4,7 +4,7 @@ namespace duckdb {
 
 static void ValidFunction(DataChunk &args, ExpressionState &state, Vector &result) {
 	auto &lstate = JSONFunctionLocalState::ResetAndGet(state);
-	auto alc = lstate.json_allocator.GetYYJSONAllocator();
+	auto alc = lstate.json_allocator.GetYYAlc();
 	auto &inputs = args.data[0];
 	UnaryExecutor::Execute<string_t, bool>(inputs, result, args.size(), [&](string_t input) {
 		return JSONCommon::ReadDocumentUnsafe(input, JSONCommon::READ_FLAG, alc);