Merge pull request #10 from duckdb/updatevendoredsourcesv0.10.3

update vendored sources to 0.10.3

Author: hannes

Diff for: CMakeLists.txt

@@ -62,8 +62,35 @@ struct IcuBindData : public FunctionData {
 		auto &other = other_p.Cast<IcuBindData>();
 		return language == other.language && country == other.country;
 	}
+
+	static void Serialize(Serializer &serializer, const optional_ptr<FunctionData> bind_data_p,
+	                      const ScalarFunction &function) {
+		auto &bind_data = bind_data_p->Cast<IcuBindData>();
+		serializer.WriteProperty(100, "language", bind_data.language);
+		serializer.WriteProperty(101, "country", bind_data.country);
+	}
+
+	static unique_ptr<FunctionData> Deserialize(Deserializer &deserializer, ScalarFunction &function) {
+		string language;
+		string country;
+		deserializer.ReadProperty(100, "language", language);
+		deserializer.ReadProperty(101, "country", country);
+
+		return make_uniq<IcuBindData>(language, country);
+	}
+
+	static const string FUNCTION_PREFIX;
+
+	static string EncodeFunctionName(const string &collation) {
+		return FUNCTION_PREFIX + collation;
+	}
+	static string DecodeFunctionName(const string &fname) {
+		return fname.substr(FUNCTION_PREFIX.size());
+	}
 };
 
+const string IcuBindData::FUNCTION_PREFIX = "icu_collate_";
+
 static int32_t ICUGetSortKey(icu::Collator &collator, string_t input, duckdb::unique_ptr<char[]> &buffer,
                              int32_t &buffer_size) {
 	icu::UnicodeString unicode_string =
@@ -108,7 +135,9 @@ static void ICUCollateFunction(DataChunk &args, ExpressionState &state, Vector &
 
 static duckdb::unique_ptr<FunctionData> ICUCollateBind(ClientContext &context, ScalarFunction &bound_function,
                                                        vector<duckdb::unique_ptr<Expression>> &arguments) {
-	auto splits = StringUtil::Split(bound_function.name, "_");
+
+	const auto collation = IcuBindData::DecodeFunctionName(bound_function.name);
+	auto splits = StringUtil::Split(collation, "_");
 	if (splits.size() == 1) {
 		return make_uniq<IcuBindData>(splits[0], "");
 	} else if (splits.size() == 2) {
@@ -137,19 +166,11 @@ static duckdb::unique_ptr<FunctionData> ICUSortKeyBind(ClientContext &context, S
 	}
 }
 
-static void ICUCollateSerialize(Serializer &serializer, const optional_ptr<FunctionData> bind_data,
-                                const ScalarFunction &function) {
-	throw NotImplementedException("FIXME: serialize icu-collate");
-}
-
-static duckdb::unique_ptr<FunctionData> ICUCollateDeserialize(Deserializer &deserializer, ScalarFunction &function) {
-	throw NotImplementedException("FIXME: serialize icu-collate");
-}
-
-static ScalarFunction GetICUFunction(const string &collation) {
-	ScalarFunction result(collation, {LogicalType::VARCHAR}, LogicalType::VARCHAR, ICUCollateFunction, ICUCollateBind);
-	result.serialize = ICUCollateSerialize;
-	result.deserialize = ICUCollateDeserialize;
+static ScalarFunction GetICUCollateFunction(const string &collation) {
+	string fname = IcuBindData::EncodeFunctionName(collation);
+	ScalarFunction result(fname, {LogicalType::VARCHAR}, LogicalType::VARCHAR, ICUCollateFunction, ICUCollateBind);
+	result.serialize = IcuBindData::Serialize;
+	result.deserialize = IcuBindData::Deserialize;
 	return result;
 }
 
@@ -238,7 +259,7 @@ static void LoadInternal(DuckDB &ddb) {
 		}
 		collation = StringUtil::Lower(collation);
 
-		CreateCollationInfo info(collation, GetICUFunction(collation), false, false);
+		CreateCollationInfo info(collation, GetICUCollateFunction(collation), false, false);
 		ExtensionUtil::RegisterCollation(db, info);
 	}
 	ScalarFunction sort_key("icu_sort_key", {LogicalType::VARCHAR, LogicalType::VARCHAR}, LogicalType::VARCHAR,
@@ -283,6 +304,14 @@ std::string IcuExtension::Name() {
 	return "icu";
 }
 
+std::string IcuExtension::Version() const {
+#ifdef EXT_VERSION_ICU
+	return EXT_VERSION_ICU;
+#else
+	return "";
+#endif
+}
+
 } // namespace duckdb
 
 extern "C" {

Diff for: src/duckdb/extension/icu/icu_extension.cpp

@@ -16,6 +16,7 @@ class IcuExtension : public Extension {
 public:
 	void Load(DuckDB &db) override;
 	std::string Name() override;
+	std::string Version() const override;
 };
 
 } // namespace duckdb

Diff for: src/duckdb/extension/icu/include/icu_extension.hpp

@@ -63,7 +63,7 @@ BasicTimeZone::hasEquivalentTransitions(const BasicTimeZone& tz, UDate start, UD
     } else {
         if (raw1 != raw2 || dst1 != dst2) {
             return FALSE;
-        }            
+        }
     }
     // Check transitions in the range
     UDate time = start;
@@ -159,7 +159,7 @@ BasicTimeZone::getSimpleRulesNear(UDate date, InitialTimeZoneRule*& initial,
         if (((tr.getFrom()->getDSTSavings() == 0 && tr.getTo()->getDSTSavings() != 0)
               || (tr.getFrom()->getDSTSavings() != 0 && tr.getTo()->getDSTSavings() == 0))
             && (date + MILLIS_PER_YEAR > nextTransitionTime)) {
- 
+
             int32_t year, month, dom, dow, doy, mid;
             UDate d;
 
@@ -375,13 +375,13 @@ BasicTimeZone::getTimeZoneRulesAfter(UDate start, InitialTimeZoneRule*& initial,
         UDate updatedTime = tzt.getTime();
         if (updatedTime == time) {
             // Can get here if rules for start & end of daylight time have exactly
-            // the same time.  
+            // the same time.
             // TODO:  fix getNextTransition() to prevent it?
             status = U_INVALID_STATE_ERROR;
             goto error;
         }
         time = updatedTime;
- 
+
         const TimeZoneRule *toRule = tzt.getTo();
         for (i = 0; i < ruleCount; i++) {
             r = (TimeZoneRule*)orgRules->elementAt(i);
@@ -408,7 +408,7 @@ BasicTimeZone::getTimeZoneRulesAfter(UDate start, InitialTimeZoneRule*& initial,
                 if (!avail) {
                     break;
                 }
-                if (*(tzt0.getTo()) == *tar) {
+                if (*tar == *(tzt0.getTo())) {
                     break;
                 }
                 t = tzt0.getTime();

Diff for: src/duckdb/extension/icu/third_party/icu/i18n/basictz.cpp

@@ -14,6 +14,8 @@
 #include "duckdb/planner/expression/bound_function_expression.hpp"
 #include "yyjson.hpp"
 
+using namespace duckdb_yyjson; // NOLINT
+
 namespace duckdb {
 
 //! JSON allocator is a custom allocator for yyjson that prevents many tiny allocations
@@ -108,6 +110,7 @@ struct JSONCommon {
 		switch (yyjson_get_tag(val)) {
 		case YYJSON_TYPE_NULL | YYJSON_SUBTYPE_NONE:
 			return JSONCommon::TYPE_STRING_NULL;
+		case YYJSON_TYPE_STR | YYJSON_SUBTYPE_NOESC:
 		case YYJSON_TYPE_STR | YYJSON_SUBTYPE_NONE:
 			return JSONCommon::TYPE_STRING_VARCHAR;
 		case YYJSON_TYPE_ARR | YYJSON_SUBTYPE_NONE:
@@ -136,6 +139,7 @@ struct JSONCommon {
 		switch (yyjson_get_tag(val)) {
 		case YYJSON_TYPE_NULL | YYJSON_SUBTYPE_NONE:
 			return LogicalTypeId::SQLNULL;
+		case YYJSON_TYPE_STR | YYJSON_SUBTYPE_NOESC:
 		case YYJSON_TYPE_STR | YYJSON_SUBTYPE_NONE:
 			return LogicalTypeId::VARCHAR;
 		case YYJSON_TYPE_ARR | YYJSON_SUBTYPE_NONE:
@@ -286,7 +290,8 @@ struct JSONCommon {
 private:
 	//! Get JSON pointer (/field/index/... syntax)
 	static inline yyjson_val *GetPointer(yyjson_val *val, const char *ptr, const idx_t &len) {
-		return len == 1 ? val : unsafe_yyjson_get_pointer(val, ptr, len);
+		yyjson_ptr_err err;
+		return len == 1 ? val : unsafe_yyjson_ptr_getx(val, ptr, len, &err);
 	}
 	//! Get JSON path ($.field[index]... syntax)
 	static yyjson_val *GetPath(yyjson_val *val, const char *ptr, const idx_t &len);

Diff for: src/duckdb/extension/json/include/json_common.hpp

@@ -30,7 +30,7 @@ struct JSONExecutors {
 	}
 
 	//! Two-argument JSON read function (with path query), i.e. json_type('[1, 2, 3]', '$[0]')
-	template <class T>
+	template <class T, bool NULL_IF_NULL = true>
 	static void BinaryExecute(DataChunk &args, ExpressionState &state, Vector &result,
 	                          std::function<T(yyjson_val *, yyjson_alc *, Vector &)> fun) {
 		auto &func_expr = state.expr.Cast<BoundFunctionExpression>();
@@ -48,7 +48,7 @@ struct JSONExecutors {
 					    auto doc =
 					        JSONCommon::ReadDocument(input, JSONCommon::READ_FLAG, lstate.json_allocator.GetYYAlc());
 					    auto val = JSONCommon::GetUnsafe(doc->root, ptr, len);
-					    if (!val || unsafe_yyjson_is_null(val)) {
+					    if (!val || (NULL_IF_NULL && unsafe_yyjson_is_null(val))) {
 						    mask.SetInvalid(idx);
 						    return T {};
 					    } else {
@@ -76,7 +76,7 @@ struct JSONExecutors {
 					for (idx_t i = 0; i < vals.size(); i++) {
 						auto &val = vals[i];
 						D_ASSERT(val != nullptr); // Wildcard extract shouldn't give back nullptrs
-						if (unsafe_yyjson_is_null(val)) {
+						if (NULL_IF_NULL && unsafe_yyjson_is_null(val)) {
 							child_validity.SetInvalid(current_size + i);
 						} else {
 							child_vals[current_size + i] = fun(val, alc, result);
@@ -109,7 +109,7 @@ struct JSONExecutors {
 	}
 
 	//! JSON read function with list of path queries, i.e. json_type('[1, 2, 3]', ['$[0]', '$[1]'])
-	template <class T>
+	template <class T, bool NULL_IF_NULL = true>
 	static void ExecuteMany(DataChunk &args, ExpressionState &state, Vector &result,
 	                        std::function<T(yyjson_val *, yyjson_alc *, Vector &)> fun) {
 		auto &func_expr = state.expr.Cast<BoundFunctionExpression>();
@@ -148,7 +148,7 @@ struct JSONExecutors {
 			for (idx_t path_i = 0; path_i < num_paths; path_i++) {
 				auto child_idx = offset + path_i;
 				val = JSONCommon::GetUnsafe(doc->root, info.ptrs[path_i], info.lens[path_i]);
-				if (!val || unsafe_yyjson_is_null(val)) {
+				if (!val || (NULL_IF_NULL && unsafe_yyjson_is_null(val))) {
 					child_validity.SetInvalid(child_idx);
 				} else {
 					child_data[child_idx] = fun(val, alc, child);

Diff for: src/duckdb/extension/json/include/json_executors.hpp

@@ -16,6 +16,7 @@ class JsonExtension : public Extension {
 public:
 	void Load(DuckDB &db) override;
 	std::string Name() override;
+	std::string Version() const override;
 };
 
 } // namespace duckdb

Diff for: src/duckdb/extension/json/include/json_extension.hpp

@@ -70,8 +70,8 @@ class JSONFunctions {
 	static vector<ScalarFunctionSet> GetScalarFunctions();
 	static vector<PragmaFunctionSet> GetPragmaFunctions();
 	static vector<TableFunctionSet> GetTableFunctions();
-	static unique_ptr<TableRef> ReadJSONReplacement(ClientContext &context, const string &table_name,
-	                                                ReplacementScanData *data);
+	static unique_ptr<TableRef> ReadJSONReplacement(ClientContext &context, ReplacementScanInput &input,
+	                                                optional_ptr<ReplacementScanData> data);
 	static TableFunction GetReadJSONTableFunction(shared_ptr<JSONScanInfo> function_info);
 	static CopyFunction GetJSONCopyFunction();
 	static void RegisterSimpleCastFunctions(CastFunctionSet &casts);

Diff for: src/duckdb/extension/json/include/json_functions.hpp

@@ -27,8 +27,8 @@ struct JsonSerializer : Serializer {
 public:
 	explicit JsonSerializer(yyjson_mut_doc *doc, bool skip_if_null, bool skip_if_empty, bool skip_if_default)
 	    : doc(doc), stack({yyjson_mut_obj(doc)}), skip_if_null(skip_if_null), skip_if_empty(skip_if_empty) {
-		serialize_enum_as_string = true;
-		serialize_default_values = !skip_if_default;
+		options.serialize_enum_as_string = true;
+		options.serialize_default_values = !skip_if_default;
 	}
 
 	template <class T>