Some endianness related changes which were never pushed

src/common/parsers/json.cpp

@@ -40,11 +40,27 @@ object json_parser::deserialize_bytes(bytes_view_type in_data, text_encoding in_
 		deserialize_json<wchar_t, true>(result, data_view);
 	}
 	else if (in_write_encoding == text_encoding::multibyte) {
-		// TODO: support without copying... somehow
+		// TODO: support without copying
 		auto u8_data = mbstring_to_ustring<char8_t>(jessilib::string_view_cast<char>(in_data));
 		std::u8string_view data_view = u8_data.second;
 		deserialize_json<char8_t, true>(result, data_view);
 	}
+	else if (in_write_encoding == text_encoding::utf_16_foreign) {
+		// TODO: support without copying
+		std::u16string u16_data{ jessilib::string_view_cast<char16_t>(in_data) };
+		array_byteswap(u16_data.data(), u16_data.data() + u16_data.size());
+		std::u16string_view data_view = u16_data;
+
+		deserialize_json<char16_t, true>(result, data_view);
+	}
+	else if (in_write_encoding == text_encoding::utf_32_foreign) {
+		// TODO: support without copying
+		std::u32string u32_data{ jessilib::string_view_cast<char32_t>(in_data) };
+		array_byteswap(u32_data.data(), u32_data.data() + u32_data.size());
+		std::u32string_view data_view = u32_data;
+
+		deserialize_json<char32_t, true>(result, data_view);
+	}
 
 	return result;
 }
@@ -56,9 +72,23 @@ std::string json_parser::serialize_bytes(const object& in_object, text_encoding
 		case text_encoding::utf_16:
 			return serialize_impl<char16_t, char>(in_object);
 		case text_encoding::utf_32:
-			return serialize_impl<char16_t, char>(in_object);
+			return serialize_impl<char32_t, char>(in_object);
 		case text_encoding::wchar:
-			return serialize_impl<char16_t, char>(in_object);
+			return serialize_impl<wchar_t, char>(in_object);
+		case text_encoding::multibyte:
+			return ustring_to_mbstring(serialize_impl<char8_t>(in_object)).second;
+
+			// Other-endianness
+		case text_encoding::utf_16_foreign: {
+			std::string result = serialize_impl<char16_t, char>(in_object);
+			string_byteswap<char16_t>(result);
+			return result;
+		}
+		case text_encoding::utf_32_foreign: {
+			std::string result = serialize_impl<char32_t, char>(in_object);
+			string_byteswap<char32_t>(result);
+			return result;
+		}
 		default:
 			break;
 	}

src/include/jessilib/parser.hpp

@@ -28,7 +28,7 @@ namespace jessilib {
 class parser {
 public:
 	virtual ~parser() = default;
-	using byte_type = uint8_t;
+	using byte_type = char;
 	using bytes_view_type = std::basic_string_view<byte_type>;
 
 	/** Interface methods */

src/include/jessilib/util.hpp

@@ -20,14 +20,25 @@
 
 #include <cstddef>
 #include <charconv>
+#include <string>
 
 /** Macros */
 
-#define JESSILIB_FILENAME \
-	(::jessilib::impl::filename_from_string(__FILE__))
+#define JESSILIB_IMPL_TYPE_LITERAL(IN, ...) __VA_ARGS__ ## IN
+#define JESSILIB_TYPE_LITERAL(IN, ...) JESSILIB_IMPL_TYPE_LITERAL(IN, __VA_ARGS__)
+
+#define JESSILIB_FILENAME(...) \
+	(::jessilib::impl::filename_from_string(JESSILIB_TYPE_LITERAL(__FILE__, __VA_ARGS__)))
+
+#define JESSILIB_FILENAME_VIEW(...) (::jessilib::make_string_view(JESSILIB_FILENAME(__VA_ARGS__)))
 
 namespace jessilib {
 
+template<typename CharT>
+std::basic_string_view<CharT> make_string_view(const CharT* in_ntstring) {
+	return in_ntstring;
+}
+
 template<typename CharT, typename NumberT>
 const CharT* parse_decimal_part(const CharT* in_str, const CharT* in_str_end, NumberT& out_value) {
 	int denominator = 10;
@@ -129,19 +140,71 @@ constexpr T square(T in_value) {
 	return in_value * in_value;
 }
 
-/** Implementation details */
+template<typename IntegerT>
+constexpr IntegerT byteswap(IntegerT in_integer) { // TODO: Remove w/ C++23
+	static_assert(sizeof(IntegerT) > 1, "byteswap on single byte does nothing");
 
+	if constexpr (sizeof(IntegerT) == 2) {
+		return (in_integer << 8) | (in_integer >> 8);
+	}
+	else if constexpr (sizeof(IntegerT) == 4) {
+		return (in_integer << 24)
+			| ((in_integer & 0xFF00) << 8)
+			| ((in_integer & 0xFF0000) >> 8)
+			| (in_integer >> 24);
+	}
+	else if constexpr (sizeof(IntegerT) == 8) {
+		return (in_integer << 56)
+			| ((in_integer & 0xFF00) << 40)
+			| ((in_integer & 0xFF0000) << 24)
+			| ((in_integer & 0xFF000000) << 8)
+			| ((in_integer & 0xFF00000000) >> 8)
+			| ((in_integer & 0xFF0000000000) >> 24)
+			| ((in_integer & 0xFF000000000000) >> 40)
+			| (in_integer >> 56);
+	}
+	else {
+		// Fallback to std::reverse for exotic integer types
+		std::reverse(reinterpret_cast<unsigned char*>(&in_integer),
+			reinterpret_cast<unsigned char*>(&in_integer) + sizeof(IntegerT));
+		return in_integer;
+	}
+}
+
+template<typename CharT>
+void array_byteswap(CharT* begin, CharT* end) {
+	while (begin != end) {
+		*begin = byteswap(*begin);
+		++begin;
+	}
+}
+
+template<typename CharT>
+void string_byteswap(std::string& in_string) {
+	// Ensure correct number of bytes for underlying representation type
+	if (in_string.size() % sizeof(CharT) != 0) {
+		return;
+	}
+
+	for (auto itr = in_string.begin(); itr != in_string.end();) {
+		auto unit_end = itr + sizeof(CharT);
+		std::reverse(itr, unit_end);
+		itr = unit_end;
+	}
+}
+
+/** Implementation details */
 
 namespace impl {
 
-template<size_t in_filename_length>
-constexpr const char* filename_from_string(const char (&in_filename)[in_filename_length]) {
-	const char* filename_itr = in_filename;
-	const char* filename_end = filename_itr + in_filename_length;
-	const char* result = filename_itr;
+template<typename CharT, size_t in_filename_length>
+constexpr const CharT* filename_from_string(const CharT (&in_filename)[in_filename_length]) {
+	const CharT* filename_itr = in_filename;
+	const CharT* filename_end = filename_itr + in_filename_length;
+	const CharT* result = filename_itr;
 
 	while (filename_itr != filename_end) {
-		if (*filename_itr == '/' || *filename_itr == '\\') {
+		if (*filename_itr == U'/' || *filename_itr == U'\\') {
 			++filename_itr;
 			result = filename_itr;
 		}
@@ -153,5 +216,28 @@ constexpr const char* filename_from_string(const char (&in_filename)[in_filename
 	return result;
 }
 
+template<typename CharT, size_t in_filename_length>
+constexpr const CharT* file_extension_from_string(const CharT (&in_filename)[in_filename_length]) {
+	const CharT* filename_itr = in_filename;
+	const CharT* filename_end = filename_itr + in_filename_length;
+	const CharT* result = filename_end;
+
+	while (filename_itr != filename_end) {
+		if (*filename_itr == U'/' || *filename_itr == U'\\') {
+			++filename_itr;
+			result = filename_end;
+		}
+		else {
+			if (*filename_itr == '.') {
+				result = filename_itr;
+			}
+
+			++filename_itr;
+		}
+	}
+
+	return result;
+}
+
 } // namespace impl
 } // namespace jessilib

src/test/parsers/json.cpp

@@ -137,12 +137,12 @@ TEST(JsonParser, deserialize_string) {
 	std::u16string_view u16text = uR"json("text")json"sv;
 	EXPECT_TRUE(deserialize_json(obj, u16text));
 	EXPECT_EQ(obj, u8"text"sv);
-	EXPECT_TRUE(u8text.empty());
+	EXPECT_TRUE(u16text.empty());
 
 	std::u32string_view u32text = UR"json("text")json"sv;
 	EXPECT_TRUE(deserialize_json(obj, u32text));
 	EXPECT_EQ(obj, u8"text"sv);
-	EXPECT_TRUE(u8text.empty());
+	EXPECT_TRUE(u32text.empty());
 }
 
 TEST(JsonParser, deserialize_array) {
@@ -273,3 +273,50 @@ TEST(JsonParser, deserialize_map_nested) {
 	EXPECT_EQ(obj[u8"some_other_object"][u8"beans"][u8"magical"], true);
 	EXPECT_EQ(obj[u8"some_other_object"][u8"beans"][u8"makes toot"], true);
 }
+
+/** Some basic foreign-encoding tests */
+
+template<typename CharT>
+std::string make_foreign_string(std::basic_string_view<CharT> in_string) {
+	std::string result{ reinterpret_cast<const char*>(in_string.data()),
+		reinterpret_cast<const char*>(in_string.data() + in_string.size()) };
+	jessilib::string_byteswap<CharT>(result);
+
+	return result;
+}
+
+TEST(JsonParser, deserialize_fu16_string) {
+	json_parser parser;
+	std::string fu16text = make_foreign_string(uR"json("text")json"sv);
+
+	// Deserialize foreign utf-16 text
+	object obj = parser.deserialize_bytes(fu16text, text_encoding::utf_16_foreign);
+	EXPECT_EQ(obj, u8"text"sv);
+}
+
+TEST(JsonParser, deserialize_fu32_string) {
+	json_parser parser;
+	std::string fu32text = make_foreign_string(UR"json("text")json"sv);
+
+	// Deserialize foreign utf-16 text
+	object obj = parser.deserialize_bytes(fu32text, text_encoding::utf_32_foreign);
+	EXPECT_EQ(obj, u8"text"sv);
+}
+
+TEST(JsonParser, serialize_fu16_string) {
+	json_parser parser;
+	std::string fu16text = make_foreign_string(uR"json("\"text\"")json"sv);
+
+	// Deserialize foreign utf-16 text
+	std::string serialized_bytes = parser.serialize_bytes(u8R"json("text")json"sv, text_encoding::utf_16_foreign);
+	EXPECT_EQ(serialized_bytes, fu16text);
+}
+
+TEST(JsonParser, serialize_fu32_string) {
+	json_parser parser;
+	std::string fu32text = make_foreign_string(UR"json("\"text\"")json"sv);
+
+	// Deserialize foreign utf-16 text
+	std::string serialized_bytes = parser.serialize_bytes(u8R"json("text")json"sv, text_encoding::utf_32_foreign);
+	EXPECT_EQ(serialized_bytes, fu32text);
+}

src/test/util.cpp

@@ -23,6 +23,60 @@ using namespace jessilib;
 using namespace std::literals;
 
 TEST(UtilTest, filename) {
-	constexpr const char* filename = JESSILIB_FILENAME;
-	EXPECT_STREQ(filename, "util.cpp");
+	EXPECT_EQ(JESSILIB_FILENAME_VIEW(), "util.cpp");
+	EXPECT_EQ(JESSILIB_FILENAME_VIEW(u8), u8"util.cpp"sv);
+	EXPECT_EQ(JESSILIB_FILENAME_VIEW(u), u"util.cpp"sv);
+	EXPECT_EQ(JESSILIB_FILENAME_VIEW(U), U"util.cpp"sv);
+	EXPECT_EQ(JESSILIB_FILENAME_VIEW(L), L"util.cpp"sv);
+	//EXPECT_EQ(JESSILIB_FILENAME(u8), u8".cpp"sv);
+}
+
+TEST(UtilTest, byteswap) {
+	EXPECT_EQ(byteswap(uint16_t{0x1234}), uint16_t{0x3412})
+		<< "hex: " << std::hex << byteswap(uint16_t{0x1234});
+	EXPECT_EQ(byteswap(uint32_t{0x12345678UL}), uint32_t{0x78563412UL})
+		<< "hex: " << std::hex << byteswap(uint32_t{0x12345678UL});
+	EXPECT_EQ(byteswap(uint64_t{0x123456789ABCDEF0ULL}), uint64_t{0xF0DEBC9A78563412ULL})
+		<< "hex: " << std::hex << byteswap(uint64_t{0x123456789ABCDEF0ULL});
+}
+
+TEST(UtilTest, array_byteswap) {
+	uint16_t numbers[]{
+		0x1234,
+		0x5678,
+		0x9ABC,
+		0xDEF0
+	};
+
+	uint16_t byteswapped_numbers[]{
+		0x3412,
+		0x7856,
+		0xBC9A,
+		0xF0DE
+	};
+
+	array_byteswap(std::begin(numbers), std::end(numbers));
+	EXPECT_EQ(numbers[0], byteswapped_numbers[0]);
+	EXPECT_EQ(numbers[1], byteswapped_numbers[1]);
+	EXPECT_EQ(numbers[2], byteswapped_numbers[2]);
+	EXPECT_EQ(numbers[3], byteswapped_numbers[3]);
+}
+
+TEST(UtilTest, string_byteswap) {
+	std::string numbers{
+		0x12, 0x34,
+		0x56, 0x78,
+		static_cast<char>(0x9A), static_cast<char>(0xBC),
+		static_cast<char>(0xDE), static_cast<char>(0xF0)
+	};
+
+	std::string u16byteswapped_numbers{
+		0x34, 0x12,
+		0x78, 0x56,
+		static_cast<char>(0xBC), static_cast<char>(0x9A),
+		static_cast<char>(0xF0), static_cast<char>(0xDE)
+	};
+
+	string_byteswap<char16_t>(numbers);
+	EXPECT_EQ(numbers, u16byteswapped_numbers);
 }