ObjFW  Check-in [ec666ede07]

#endif
extern int of_string_check_utf8(const char*, size_t);
extern size_t of_string_unicode_to_utf8(of_unichar_t, char*);
extern size_t of_string_utf8_to_unicode(const char*, size_t, of_unichar_t*);
extern size_t of_string_position_to_index(const char*, size_t);
extern size_t of_string_index_to_position(const char*, size_t, size_t);
extern size_t of_unicode_string_length(const of_unichar_t*);
extern size_t of_utf16_string_length(const uint16_t*);
#ifdef __cplusplus
}
#endif

@class OFArray;
@class OFURL;

 * \param string The unicode string
 * \param length The length of the unicode string
 * \return A new autoreleased OFString
 */
+ stringWithUnicodeString: (of_unichar_t*)string
		   length: (size_t)length;

/**
 * Creates a new OFString from a UTF-16 encoded string.
 *
 * \param string The UTF-16 string
 * \return A new autoreleased OFString
 */
+ stringWithUTF16String: (uint16_t*)string;

/**
 * Creates a new OFString from a UTF-16 encoded string with the specified
 * length.
 *
 * \param string The UTF-16 string
 * \param length The length of the unicode string
 * \return A new autoreleased OFString
 */
+ stringWithUTF16String: (uint16_t*)string
		 length: (size_t)length;

/**
 * Creates a new OFString from a format string.
 * See printf for the format syntax.
 *
 * \param format A string used as format to initialize the OFString
 * \return A new autoreleased OFString
 */

 * \param string The unicode string
 * \param length The length of the unicode string
 * \return An initialized OFString
 */
- initWithUnicodeString: (of_unichar_t*)string
		 length: (size_t)length;

/**
 * Initializes an already allocated OFString with a UTF-16 string.
 *
 * \param string The UTF-16 string
 * \return An initialized OFString
 */
- initWithUTF16String: (uint16_t*)string;

/**
 * Initializes an already allocated OFString with a UTF-16 string with the
 * specified length.
 *
 * \param string The UTF-16 string
 * \param length The length of the UTF-16 string
 * \return An initialized OFString
 */
- initWithUTF16String: (uint16_t*)string
	       length: (size_t)length;

/**
 * Initializes an already allocated OFString with a format string.
 * See printf for the format syntax.
 *
 * \param format A string used as format to initialize the OFString
 * \return An initialized OFString
 */

	const of_unichar_t *string_ = string;

	while (*string_ != 0)
		string_++;

	return (size_t)(string_ - string);
}

size_t
of_utf16_string_length(const uint16_t *string)
{
	const uint16_t *string_ = string;

	while (*string_ != 0)
		string_++;

	return (size_t)(string_ - string);
}

@implementation OFString
+ string
{
	return [[[self alloc] init] autorelease];
}

+ stringWithUnicodeString: (of_unichar_t*)string
		   length: (size_t)length
{
	return [[[self alloc] initWithUnicodeString: string
					     length: length] autorelease];
}

+ stringWithUTF16String: (uint16_t*)string
{
	return [[[self alloc] initWithUTF16String: string] autorelease];
}

+ stringWithUTF16String: (uint16_t*)string
		 length: (size_t)length
{
	return [[[self alloc] initWithUTF16String: string
					   length: length] autorelease];
}

+ stringWithFormat: (OFString*)format, ...
{
	id ret;
	va_list arguments;

	va_start(arguments, format);

		length = length_;
		string = [self allocMemoryWithSize: (length * 4) + 1];

		for (i = 0; i < length_; i++) {
			size_t characterLen = of_string_unicode_to_utf8(
			    (swap ? of_bswap32(string_[i]) : string_[i]),
			    buffer);

			switch (characterLen) {
			case 1:
				string[j++] = buffer[0];
				break;
			case 2:
				isUTF8 = YES;
				length++;

				memcpy(string + j, buffer, 2);
				j += 2;

				break;
			case 3:
				isUTF8 = YES;
				length += 2;

				memcpy(string + j, buffer, 3);
				j += 3;

				break;
			case 4:
				isUTF8 = YES;
				length += 3;

				memcpy(string + j, buffer, 4);
				j += 4;

				break;
			default:
				@throw [OFInvalidEncodingException
				    newWithClass: isa];
			}
		}

		string[j] = '\0';

		@try {
			string = [self resizeMemory: string
					     toSize: length + 1];
		} @catch (OFOutOfMemoryException *e) {
			/* We don't care, as we only tried to make it smaller */
			[e release];
		}
	} @catch (id e) {
		[self release];
		@throw e;
	}

	return self;
}

- initWithUTF16String: (uint16_t*)string_
{
	return [self initWithUTF16String: string_
				  length: of_utf16_string_length(string_)];
}

- initWithUTF16String: (uint16_t*)string_
	       length: (size_t)length_
{
	self = [super init];

	@try {
		char buffer[4];
		size_t i, j = 0;
		BOOL swap = NO;

		if (*string_ == 0xFEFF) {
			string_++;
			length_--;
		}

		if (*string_ == 0xFFFE) {
			swap = YES;
			string_++;
			length_--;
		}

		length = length_;
		string = [self allocMemoryWithSize: (length * 4) + 1];

		for (i = 0; i < length_; i++) {
			of_unichar_t character =
			    (swap ? of_bswap16(string_[i]) : string_[i]);
			size_t characterLen;

			/* Missed the high surrogate */
			if ((character & 0xFC00) == 0xDC00)
				@throw [OFInvalidEncodingException
				    newWithClass: isa];

			if ((character & 0xFC00) == 0xD800) {
				uint16_t nextCharacter;

				if (length <= i + 1)
					@throw [OFInvalidEncodingException
						newWithClass: isa];

				nextCharacter = (swap
				    ? of_bswap16(string_[i + 1])
				    : string_[i + 1]);
				character = (((character & 0x3FF) << 10) |
				    (nextCharacter & 0x3FF)) + 0x10000;

				i++;
			}

			characterLen = of_string_unicode_to_utf8(
			    character, buffer);

			switch (characterLen) {
			case 1:
				string[j++] = buffer[0];
				break;
			case 2:
				isUTF8 = YES;

static of_unichar_t ucstr[] = {
	0xFEFF, 'f', 0xF6, 0xF6, 'b', 0xE4, 'r', 0x1F03A, 0
};
static of_unichar_t sucstr[] = {
	0xFFFE0000, 0x66000000, 0xF6000000, 0xF6000000, 0x62000000, 0xE4000000,
	0x72000000, 0x3AF00100, 0
};
static uint16_t utf16str[] = {
	0xFEFF, 'f', 0xF6, 0xF6, 'b', 0xE4, 'r', 0xD83C, 0xDC3A, 0
};
static uint16_t sutf16str[] = {
	0xFFFE, 0x6600, 0xF600, 0xF600, 0x6200, 0xE400, 0x7200, 0x3CD8, 0x3ADC,
	0
};

@interface EntityHandler: OFObject <OFStringXMLUnescapingDelegate>
@end

@implementation EntityHandler
-	   (OFString*)string: (OFString*)string
  containsUnknownEntityNamed: (OFString*)entity

	TEST(@"+[stringWithUnicodeString:]",
	    (s[1] = [OFString stringWithUnicodeString: ucstr]) &&
	    [s[1] isEqual: @"fööbär🀺"] &&
	    (s[1] = [OFString stringWithUnicodeString: sucstr]) &&
	    [s[1] isEqual: @"fööbär🀺"])

	TEST(@"+[stringWithUTF16String:]",
	    (s[1] = [OFString stringWithUTF16String: utf16str]) &&
	    [s[1] isEqual: @"fööbär🀺"] &&
	    (s[1] = [OFString stringWithUTF16String: sutf16str]) &&
	    [s[1] isEqual: @"fööbär🀺"])

	TEST(@"+[stringWithContentsOfFile:encoding]", (s[1] = [OFString
	    stringWithContentsOfFile: @"testfile.txt"
			    encoding: OF_STRING_ENCODING_ISO_8859_1]) &&
	    [s[1] isEqual: @"testäöü"])

	TEST(@"+[stringWithContentsOfURL:encoding]", (s[1] = [OFString
	    stringWithContentsOfURL: [OFURL URLWithString: