ObjFW  Check-in [1046d10a20]

- (const of_char16_t *)UTF16String
{
	[self finishInitialization];
	return self.UTF16String;
}

- (const of_char16_t *)UTF16StringWithByteOrder: (of_byte_order_t)byteOrder
- (const of_char16_t *)UTF16StringWithByteOrder: (OFByteOrder)byteOrder
{
	[self finishInitialization];
	return [self UTF16StringWithByteOrder: byteOrder];
}

- (size_t)UTF16StringLength
{
	[self finishInitialization];
	return self.UTF16StringLength;
}

- (const of_char32_t *)UTF32String
{
	[self finishInitialization];
	return self.UTF32String;
}

- (const of_char32_t *)UTF32StringWithByteOrder: (of_byte_order_t)byteOrder
- (const of_char32_t *)UTF32StringWithByteOrder: (OFByteOrder)byteOrder
{
	[self finishInitialization];
	return [self UTF32StringWithByteOrder: byteOrder];
}

- (OFData *)dataWithEncoding: (of_string_encoding_t)encoding
{

			     length: (size_t)length
{
	return (id)[[OFMutableUTF8String alloc] initWithUTF16String: string
							      length: length];
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return (id)[[OFMutableUTF8String alloc] initWithUTF16String: string
							  byteOrder: byteOrder];
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return (id)[[OFMutableUTF8String alloc] initWithUTF16String: string
							     length: length
							  byteOrder: byteOrder];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
{
	return (id)[[OFMutableUTF8String alloc] initWithUTF32String: string];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			     length: (size_t)length
{
	return (id)[[OFMutableUTF8String alloc] initWithUTF32String: string
							     length: length];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return (id)[[OFMutableUTF8String alloc] initWithUTF32String: string
							  byteOrder: byteOrder];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return (id)[[OFMutableUTF8String alloc] initWithUTF32String: string
							     length: length
							  byteOrder: byteOrder];
}

- (instancetype)initWithFormat: (OFConstantString *)format, ...

#endif

/**
 * @brief An enum for storing endianess.
 */
typedef enum {
	/** Most significant byte first (big endian) */
	OF_BYTE_ORDER_BIG_ENDIAN,
	OFByteOrderBigEndian,
	/** Least significant byte first (little endian) */
	OFByteOrderLittleEndian,
	/** Native byte order of the system */
	OF_BYTE_ORDER_LITTLE_ENDIAN
} of_byte_order_t;
#ifdef OF_BIG_ENDIAN
	OFByteOrderNative = OFByteOrderBigEndian
#else
	OFByteOrderNative = OFByteOrderLittleEndian
#endif
} OFByteOrder;

/**
 * @struct of_range_t OFObject.h ObjFW/OFObject.h
 *
 * @brief A range.
 */
struct OF_BOXABLE of_range_t {

 *	  specified byte order if no byte order mark is found.
 *
 * @param string The UTF-16 string
 * @param byteOrder The byte order to assume if there is no byte order mark
 * @return A new autoreleased OFString
 */
+ (instancetype)stringWithUTF16String: (const of_char16_t *)string
			    byteOrder: (of_byte_order_t)byteOrder;
			    byteOrder: (OFByteOrder)byteOrder;

/**
 * @brief Creates a new OFString from a UTF-16 encoded string with the
 *	  specified length, assuming the specified byte order if no byte order
 *	  mark is found.
 *
 * @param string The UTF-16 string
 * @param length The length of the UTF-16 string
 * @param byteOrder The byte order to assume if there is no byte order mark
 * @return A new autoreleased OFString
 */
+ (instancetype)stringWithUTF16String: (const of_char16_t *)string
			       length: (size_t)length
			    byteOrder: (of_byte_order_t)byteOrder;
			    byteOrder: (OFByteOrder)byteOrder;

/**
 * @brief Creates a new OFString from a UTF-32 encoded string.
 *
 * @param string The UTF-32 string
 * @return A new autoreleased OFString
 */

 *	  specified byte order if no byte order mark is found.
 *
 * @param string The UTF-32 string
 * @param byteOrder The byte order to assume if there is no byte order mark
 * @return A new autoreleased OFString
 */
+ (instancetype)stringWithUTF32String: (const of_char32_t *)string
			    byteOrder: (of_byte_order_t)byteOrder;
			    byteOrder: (OFByteOrder)byteOrder;

/**
 * @brief Creates a new OFString from a UTF-32 encoded string with the
 *	  specified length, assuming the specified byte order if no byte order
 *	  mark is found.
 *
 * @param string The UTF-32 string
 * @param length The length of the UTF-32 string
 * @param byteOrder The byte order to assume if there is no byte order mark
 * @return A new autoreleased OFString
 */
+ (instancetype)stringWithUTF32String: (const of_char32_t *)string
			       length: (size_t)length
			    byteOrder: (of_byte_order_t)byteOrder;
			    byteOrder: (OFByteOrder)byteOrder;

/**
 * @brief Creates a new OFString from a format string.
 *
 * See printf for the format syntax. As an addition, `%@` is available as
 * format specifier for objects, `%C` for `of_unichar_t` and `%S` for
 * `const of_unichar_t *`.

 *	  assuming the specified byte order if no byte order mark is found.
 *
 * @param string The UTF-16 string
 * @param byteOrder The byte order to assume if there is no byte order mark
 * @return An initialized OFString
 */
- (instancetype)initWithUTF16String: (const of_char16_t *)string
			  byteOrder: (of_byte_order_t)byteOrder;
			  byteOrder: (OFByteOrder)byteOrder;

/**
 * @brief Initializes an already allocated OFString with a UTF-16 string with
 *	  the specified length, assuming the specified byte order if no byte
 *	  order mark is found.
 *
 * @param string The UTF-16 string
 * @param length The length of the UTF-16 string
 * @param byteOrder The byte order to assume if there is no byte order mark
 * @return An initialized OFString
 */
- (instancetype)initWithUTF16String: (const of_char16_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder;
			  byteOrder: (OFByteOrder)byteOrder;

/**
 * @brief Initializes an already allocated OFString with a UTF-32 string.
 *
 * @param string The UTF-32 string
 * @return An initialized OFString
 */

 *	  assuming the specified byte order if no byte order mark is found.
 *
 * @param string The UTF-32 string
 * @param byteOrder The byte order to assume if there is no byte order mark
 * @return An initialized OFString
 */
- (instancetype)initWithUTF32String: (const of_char32_t *)string
			  byteOrder: (of_byte_order_t)byteOrder;
			  byteOrder: (OFByteOrder)byteOrder;

/**
 * @brief Initializes an already allocated OFString with a UTF-32 string with
 *	  the specified length, assuming the specified byte order if no byte
 *	  order mark is found.
 *
 * @param string The UTF-32 string
 * @param length The length of the UTF-32 string
 * @param byteOrder The byte order to assume if there is no byte order mark
 * @return An initialized OFString
 */
- (instancetype)initWithUTF32String: (const of_char32_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder;
			  byteOrder: (OFByteOrder)byteOrder;

/**
 * @brief Initializes an already allocated OFString with a format string.
 *
 * See printf for the format syntax. As an addition, `%@` is available as
 * format specifier for objects, `%C` for `of_unichar_t` and `%S` for
 * `const of_unichar_t *`.

 * The result is valid until the autorelease pool is released. If you want to
 * use the result outside the scope of the current autorelease pool, you have to
 * copy it.
 *
 * @param byteOrder The byte order for the UTF-16 encoding
 * @return The string in UTF-16 encoding with the specified byte order
 */
- (const of_char16_t *)UTF16StringWithByteOrder: (of_byte_order_t)byteOrder
- (const of_char16_t *)UTF16StringWithByteOrder: (OFByteOrder)byteOrder
    OF_RETURNS_INNER_POINTER;

/**
 * @brief Returns the string in UTF-32 encoding with the specified byte order.
 *
 * The result is valid until the autorelease pool is released. If you want to
 * use the result outside the scope of the current autorelease pool, you have to
 * copy it.
 *
 * @param byteOrder The byte order for the UTF-32 encoding
 * @return The string in UTF-32 encoding with the specified byte order
 */
- (const of_char32_t *)UTF32StringWithByteOrder: (of_byte_order_t)byteOrder
- (const of_char32_t *)UTF32StringWithByteOrder: (OFByteOrder)byteOrder
    OF_RETURNS_INNER_POINTER;

/**
 * @brief Returns the string as OFData with the specified encoding.
 *
 * @param encoding The encoding to use for the returned OFData
 * @return The string as OFData with the specified encoding

			     length: (size_t)length
{
	return (id)[[OFUTF8String alloc] initWithUTF16String: string
						      length: length];
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return (id)[[OFUTF8String alloc] initWithUTF16String: string
						   byteOrder: byteOrder];
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return (id)[[OFUTF8String alloc] initWithUTF16String: string
						      length: length
						   byteOrder: byteOrder];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
{
	return (id)[[OFUTF8String alloc] initWithUTF32String: string];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			     length: (size_t)length
{
	return (id)[[OFUTF8String alloc] initWithUTF32String: string
						      length: length];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return (id)[[OFUTF8String alloc] initWithUTF32String: string
						   byteOrder: byteOrder];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return (id)[[OFUTF8String alloc] initWithUTF32String: string
						      length: length
						   byteOrder: byteOrder];
}

- (instancetype)initWithFormat: (OFConstantString *)format, ...

			       length: (size_t)length
{
	return [[[self alloc] initWithUTF16String: string
					   length: length] autorelease];
}

+ (instancetype)stringWithUTF16String: (const of_char16_t *)string
			    byteOrder: (of_byte_order_t)byteOrder
			    byteOrder: (OFByteOrder)byteOrder
{
	return [[[self alloc] initWithUTF16String: string
					byteOrder: byteOrder] autorelease];
}

+ (instancetype)stringWithUTF16String: (const of_char16_t *)string
			       length: (size_t)length
			    byteOrder: (of_byte_order_t)byteOrder
			    byteOrder: (OFByteOrder)byteOrder
{
	return [[[self alloc] initWithUTF16String: string
					   length: length
					byteOrder: byteOrder] autorelease];
}

+ (instancetype)stringWithUTF32String: (const of_char32_t *)string
{
	return [[[self alloc] initWithUTF32String: string] autorelease];
}

+ (instancetype)stringWithUTF32String: (const of_char32_t *)string
			       length: (size_t)length
{
	return [[[self alloc] initWithUTF32String: string
					   length: length] autorelease];
}

+ (instancetype)stringWithUTF32String: (const of_char32_t *)string
			    byteOrder: (of_byte_order_t)byteOrder
			    byteOrder: (OFByteOrder)byteOrder
{
	return [[[self alloc] initWithUTF32String: string
					byteOrder: byteOrder] autorelease];
}

+ (instancetype)stringWithUTF32String: (const of_char32_t *)string
			       length: (size_t)length
			    byteOrder: (of_byte_order_t)byteOrder
			    byteOrder: (OFByteOrder)byteOrder
{
	return [[[self alloc] initWithUTF32String: string
					   length: length
					byteOrder: byteOrder] autorelease];
}

+ (instancetype)stringWithFormat: (OFConstantString *)format, ...

	OF_INVALID_INIT_METHOD
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
{
	return [self initWithUTF16String: string
				  length: of_string_utf16_length(string)
			       byteOrder: OF_BYTE_ORDER_NATIVE];
			       byteOrder: OFByteOrderNative];
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			     length: (size_t)length
{
	return [self initWithUTF16String: string
				  length: length
			       byteOrder: OF_BYTE_ORDER_NATIVE];
			       byteOrder: OFByteOrderNative];
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return [self initWithUTF16String: string
				  length: of_string_utf16_length(string)
			       byteOrder: byteOrder];
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	OF_INVALID_INIT_METHOD
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
{
	return [self initWithUTF32String: string
				  length: of_string_utf32_length(string)
			       byteOrder: OF_BYTE_ORDER_NATIVE];
			       byteOrder: OFByteOrderNative];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			     length: (size_t)length
{
	return [self initWithUTF32String: string
				  length: length
			       byteOrder: OF_BYTE_ORDER_NATIVE];
			       byteOrder: OFByteOrderNative];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	return [self initWithUTF32String: string
				  length: of_string_utf32_length(string)
			       byteOrder: byteOrder];
}

- (instancetype)initWithUTF32String: (const of_char32_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	OF_INVALID_INIT_METHOD
}

- (instancetype)initWithFormat: (OFConstantString *)format, ...
{
	id ret;

		free(buffer);
		@throw e;
	}
}

- (const of_char16_t *)UTF16String
{
	return [self UTF16StringWithByteOrder: OF_BYTE_ORDER_NATIVE];
	return [self UTF16StringWithByteOrder: OFByteOrderNative];
}

- (const of_char16_t *)UTF16StringWithByteOrder: (of_byte_order_t)byteOrder
- (const of_char16_t *)UTF16StringWithByteOrder: (OFByteOrder)byteOrder
{
	void *pool = objc_autoreleasePoolPush();
	const of_unichar_t *characters = self.characters;
	size_t length = self.length;
	of_char16_t *buffer;
	size_t j;
	bool swap = (byteOrder != OF_BYTE_ORDER_NATIVE);
	bool swap = (byteOrder != OFByteOrderNative);

	/* Allocate memory for the worst case */
	buffer = of_alloc((length + 1) * 2, sizeof(of_char16_t));

	j = 0;
	for (size_t i = 0; i < length; i++) {
		of_unichar_t c = characters[i];

			UTF16StringLength++;

	return UTF16StringLength;
}

- (const of_char32_t *)UTF32String
{
	return [self UTF32StringWithByteOrder: OF_BYTE_ORDER_NATIVE];
	return [self UTF32StringWithByteOrder: OFByteOrderNative];
}

- (const of_char32_t *)UTF32StringWithByteOrder: (of_byte_order_t)byteOrder
- (const of_char32_t *)UTF32StringWithByteOrder: (OFByteOrder)byteOrder
{
	size_t length = self.length;
	of_char32_t *buffer;

	buffer = of_alloc(length + 1, sizeof(of_char32_t));
	@try {
		[self getCharacters: buffer inRange: of_range(0, length)];
		buffer[length] = 0;

		if (byteOrder != OF_BYTE_ORDER_NATIVE)
		if (byteOrder != OFByteOrderNative)
			for (size_t i = 0; i < length; i++)
				buffer[i] = OF_BSWAP32(buffer[i]);

		return [[OFData dataWithItemsNoCopy: buffer
					      count: length + 1
					   itemSize: sizeof(of_char32_t)
				       freeWhenDone: true] items];

	}

	return self;
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	self = [super init];

	@try {
		size_t j;
		bool swap = false;

		if (length > 0 && *string == 0xFEFF) {
			string++;
			length--;
		} else if (length > 0 && *string == 0xFFFE) {
			swap = true;
			string++;
			length--;
		} else if (byteOrder != OF_BYTE_ORDER_NATIVE)
		} else if (byteOrder != OFByteOrderNative)
			swap = true;

		_s = &_storage;

		_s->cString = of_alloc((length * 4) + 1, 1);
		_s->length = length;
		_s->freeWhenDone = true;

	}

	return self;
}

- (instancetype)initWithUTF32String: (const of_char32_t *)characters
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	self = [super init];

	@try {
		size_t j;
		bool swap = false;

		if (length > 0 && *characters == 0xFEFF) {
			characters++;
			length--;
		} else if (length > 0 && *characters == 0xFFFE0000) {
			swap = true;
			characters++;
			length--;
		} else if (byteOrder != OF_BYTE_ORDER_NATIVE)
		} else if (byteOrder != OFByteOrderNative)
			swap = true;

		_s = &_storage;

		_s->cString = of_alloc((length * 4) + 1, 1);
		_s->length = length;
		_s->freeWhenDone = true;

	return [[OFData dataWithItemsNoCopy: buffer
				      count: _s->length
				   itemSize: sizeof(of_unichar_t)
			       freeWhenDone: true] items];
}

- (const of_char32_t *)UTF32StringWithByteOrder: (of_byte_order_t)byteOrder
- (const of_char32_t *)UTF32StringWithByteOrder: (OFByteOrder)byteOrder
{
	of_char32_t *buffer = of_alloc(_s->length + 1, sizeof(of_char32_t));
	size_t i = 0, j = 0;

	while (i < _s->cStringLength) {
		of_char32_t c;
		ssize_t cLen;

		cLen = of_string_utf8_decode(_s->cString + i,
		    _s->cStringLength - i, &c);

		if (cLen <= 0 || c > 0x10FFFF) {
			free(buffer);
			@throw [OFInvalidEncodingException exception];
		}

		if (byteOrder != OF_BYTE_ORDER_NATIVE)
		if (byteOrder != OFByteOrderNative)
			buffer[j++] = OF_BSWAP32(c);
		else
			buffer[j++] = c;

		i += cLen;
	}
	buffer[j] = 0;

# define OF_LIKELY(cond) (cond)
# define OF_UNLIKELY(cond) (cond)
# define OF_CONST_FUNC
# define OF_NO_RETURN_FUNC
# define OF_WEAK_REF(sym)
#endif

#ifdef OF_BIG_ENDIAN
# define OF_BYTE_ORDER_NATIVE OF_BYTE_ORDER_BIG_ENDIAN
#else
# define OF_BYTE_ORDER_NATIVE OF_BYTE_ORDER_LITTLE_ENDIAN
#endif

#if __STDC_VERSION__ >= 201112L
# define OF_ALIGNOF(type) _Alignof(type)
# define OF_ALIGNAS(type) _Alignas(type)
#else
# define OF_ALIGNOF(type) __alignof__(type)
# define OF_ALIGNAS(type) __attribute__((__aligned__(__alignof__(type))))
#endif

	}

	return self;
}

- (instancetype)initWithUTF16String: (const of_char16_t *)UTF16String
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	self = [super init];

	@try {
		_string = [[OFMutableString alloc]
		    initWithUTF16String: UTF16String
				 length: length
			      byteOrder: byteOrder];
	} @catch (id e) {
		[self release];
		@throw e;
	}

	return self;
}

- (instancetype)initWithUTF32String: (const of_char32_t *)UTF32String
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)byteOrder
			  byteOrder: (OFByteOrder)byteOrder
{
	self = [super init];

	@try {
		_string = [[OFMutableString alloc]
		    initWithUTF32String: UTF32String
				 length: length

	       @"0123456789ABCDEF0123456789ABCDEF0123456789ABCDEF")
	    unsignedLongLongValueWithBase: 16])

	TEST(@"-[characters]", (ua = C(@"fööbär🀺").characters) &&
	    !memcmp(ua, ucstr + 1, sizeof(ucstr) - 8))

#ifdef OF_BIG_ENDIAN
# define SWAPPED_BYTE_ORDER OF_BYTE_ORDER_LITTLE_ENDIAN
# define SWAPPED_BYTE_ORDER OFByteOrderLittleEndian
#else
# define SWAPPED_BYTE_ORDER OF_BYTE_ORDER_BIG_ENDIAN
# define SWAPPED_BYTE_ORDER OFByteOrderBigEndian
#endif
	TEST(@"-[UTF16String]", (u16a = C(@"fööbär🀺").UTF16String) &&
	    !memcmp(u16a, utf16str + 1, of_string_utf16_length(utf16str) * 2) &&
	    (u16a = [C(@"fööbär🀺")
	    UTF16StringWithByteOrder: SWAPPED_BYTE_ORDER]) &&
	    !memcmp(u16a, sutf16str + 1, of_string_utf16_length(sutf16str) * 2))