ObjFW  Diff

#import "OFInvalidFormatException.h"
#import "OFOutOfMemoryException.h"
#import "OFOutOfRangeException.h"

#import "of_asprintf.h"
#import "unicode.h"

extern const of_char16_t of_iso_8859_2_table[];
extern const size_t of_iso_8859_2_table_offset;
extern const of_char16_t of_iso_8859_3_table[];
extern const size_t of_iso_8859_3_table_offset;
extern const of_char16_t of_iso_8859_15_table[];
extern const size_t of_iso_8859_15_table_offset;
extern const of_char16_t of_windows_1251_table[];
extern const size_t of_windows_1251_table_offset;
extern const of_char16_t of_windows_1252_table[];
extern const size_t of_windows_1252_table_offset;
extern const of_char16_t of_codepage_437_table[];
extern const size_t of_codepage_437_table_offset;
extern const of_char16_t of_codepage_850_table[];
extern const size_t of_codepage_850_table_offset;
extern const of_char16_t of_codepage_858_table[];
extern const size_t of_codepage_858_table_offset;
extern const of_char16_t of_mac_roman_table[];
extern const size_t of_mac_roman_table_offset;
extern const of_char16_t of_koi8_r_table[];
extern const size_t of_koi8_r_table_offset;
extern const of_char16_t of_koi8_u_table[];
extern const size_t of_koi8_u_table_offset;

static inline int
memcasecmp(const char *first, const char *second, size_t length)
{
	for (size_t i = 0; i < length; i++) {
		unsigned char f = first[i];
		unsigned char s = second[i];

		f = of_ascii_toupper(f);
		s = of_ascii_toupper(s);

		if (f > s)
			return OF_ORDERED_DESCENDING;
		if (f < s)
			return OF_ORDERED_ASCENDING;
	}

	return OF_ORDERED_SAME;
}

int
of_string_utf8_check(const char *UTF8String, size_t UTF8Length, size_t *length)
{
	size_t tmpLength = UTF8Length;
	int isUTF8 = 0;

		@throw e;
	}

	return self;
}

- (instancetype)initWithCString: (const char *)cString
		       encoding: (of_string_encoding_t)encoding
			 length: (size_t)cStringLength
{
	self = [super init];

	@try {
		const of_char16_t *table;
		size_t tableOffset, j;

		if (encoding == OF_STRING_ENCODING_UTF_8 &&
		    cStringLength >= 3 &&
		    memcmp(cString, "\xEF\xBB\xBF", 3) == 0) {
			cString += 3;
			cStringLength -= 3;
		}

		_s = &_storage;

		_s->cString = of_alloc(cStringLength + 1, 1);
		_s->cStringLength = cStringLength;
		_s->freeWhenDone = true;

		if (encoding == OF_STRING_ENCODING_UTF_8 ||
		    encoding == OF_STRING_ENCODING_ASCII) {
			switch (of_string_utf8_check(cString, cStringLength,
			    &_s->length)) {
			case 1:
				if (encoding == OF_STRING_ENCODING_ASCII)
					@throw [OFInvalidEncodingException
					    exception];

				_s->isUTF8 = true;
				break;
			case -1:
				@throw [OFInvalidEncodingException exception];
			}

			memcpy(_s->cString, cString, cStringLength);
			_s->cString[cStringLength] = 0;

			return self;
		}

		/* All other encodings we support are single byte encodings */
		_s->length = cStringLength;

		if (encoding == OF_STRING_ENCODING_ISO_8859_1) {
			j = 0;
			for (size_t i = 0; i < cStringLength; i++) {
				char buffer[4];
				size_t bytes;

				if (!(cString[i] & 0x80)) {
					_s->cString[j++] = cString[i];

		switch (encoding) {
#define CASE(encoding, var)				\
		case encoding:				\
			table = var;			\
			tableOffset = var##_offset;	\
			break;
#ifdef HAVE_ISO_8859_2
		CASE(OF_STRING_ENCODING_ISO_8859_2, of_iso_8859_2_table)
#endif
#ifdef HAVE_ISO_8859_3
		CASE(OF_STRING_ENCODING_ISO_8859_3, of_iso_8859_3_table)
#endif
#ifdef HAVE_ISO_8859_15
		CASE(OF_STRING_ENCODING_ISO_8859_15, of_iso_8859_15_table)
#endif
#ifdef HAVE_WINDOWS_1251
		CASE(OF_STRING_ENCODING_WINDOWS_1251, of_windows_1251_table)
#endif
#ifdef HAVE_WINDOWS_1252
		CASE(OF_STRING_ENCODING_WINDOWS_1252, of_windows_1252_table)
#endif
#ifdef HAVE_CODEPAGE_437
		CASE(OF_STRING_ENCODING_CODEPAGE_437, of_codepage_437_table)
#endif
#ifdef HAVE_CODEPAGE_850
		CASE(OF_STRING_ENCODING_CODEPAGE_850, of_codepage_850_table)
#endif
#ifdef HAVE_CODEPAGE_858
		CASE(OF_STRING_ENCODING_CODEPAGE_858, of_codepage_858_table)
#endif
#ifdef HAVE_MAC_ROMAN
		CASE(OF_STRING_ENCODING_MAC_ROMAN, of_mac_roman_table)
#endif
#ifdef HAVE_KOI8_R
		CASE(OF_STRING_ENCODING_KOI8_R, of_koi8_r_table)
#endif
#ifdef HAVE_KOI8_U
		CASE(OF_STRING_ENCODING_KOI8_U, of_koi8_u_table)
#endif
#undef CASE
		default:
			@throw [OFInvalidEncodingException exception];
		}

		j = 0;
		for (size_t i = 0; i < cStringLength; i++) {
			unsigned char character = (unsigned char)cString[i];
			of_unichar_t unichar;
			char buffer[4];
			size_t byteLength;

			if (character < tableOffset) {
				_s->cString[j++] = cString[i];
				continue;
			}

		[self release];
		@throw e;
	}

	return self;
}

- (instancetype)initWithCharacters: (const of_unichar_t *)characters
			    length: (size_t)length
{
	self = [super init];

	@try {
		size_t j;

		[self release];
		@throw e;
	}

	return self;
}

- (instancetype)initWithUTF16String: (const of_char16_t *)string
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)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)
			swap = true;

		_s = &_storage;

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

		j = 0;
		for (size_t i = 0; i < length; i++) {
			of_unichar_t character =
			    (swap ? OF_BSWAP16(string[i]) : string[i]);
			size_t len;

			/* Missing high surrogate */
			if ((character & 0xFC00) == 0xDC00)
				@throw [OFInvalidEncodingException exception];

			if ((character & 0xFC00) == 0xD800) {
				of_char16_t nextCharacter;

				if (length <= i + 1)
					@throw [OFInvalidEncodingException
					    exception];

				nextCharacter = (swap
				    ? OF_BSWAP16(string[i + 1])

		[self release];
		@throw e;
	}

	return self;
}

- (instancetype)initWithUTF32String: (const of_char32_t *)characters
			     length: (size_t)length
			  byteOrder: (of_byte_order_t)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)
			swap = true;

		_s = &_storage;

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

		free(_s->cString);

	[super dealloc];
}

- (size_t)getCString: (char *)cString
	   maxLength: (size_t)maxLength
	    encoding: (of_string_encoding_t)encoding
{
	switch (encoding) {
	case OF_STRING_ENCODING_ASCII:
		if (_s->isUTF8)
			@throw [OFInvalidEncodingException exception];
		/* intentional fall-through */
	case OF_STRING_ENCODING_UTF_8:
		if (_s->cStringLength + 1 > maxLength)
			@throw [OFOutOfRangeException exception];

		memcpy(cString, _s->cString, _s->cStringLength + 1);

		return _s->cStringLength;
	default:
		return [super getCString: cString
			       maxLength: maxLength
				encoding: encoding];
	}
}

- (const char *)cStringWithEncoding: (of_string_encoding_t)encoding
{
	switch (encoding) {
	case OF_STRING_ENCODING_ASCII:
		if (_s->isUTF8)
			@throw [OFInvalidEncodingException exception];
		/* intentional fall-through */
	case OF_STRING_ENCODING_UTF_8:
		return _s->cString;
	default:
		return [super cStringWithEncoding: encoding];
	}
}

- (const char *)UTF8String
{
	return _s->cString;
}

- (size_t)length
{
	return _s->length;
}

- (size_t)cStringLengthWithEncoding: (of_string_encoding_t)encoding
{
	switch (encoding) {
	case OF_STRING_ENCODING_UTF_8:
	case OF_STRING_ENCODING_ASCII:
		return _s->cStringLength;
	default:
		return [super cStringLengthWithEncoding: encoding];
	}
}

- (size_t)UTF8StringLength

	if (strcmp(_s->cString, string.UTF8String) != 0)
		return false;

	return true;
}

- (of_comparison_result_t)compare: (OFString *)string
{
	size_t otherCStringLength, minimumCStringLength;
	int compare;

	if (string == self)
		return OF_ORDERED_SAME;

	if (![string isKindOfClass: [OFString class]])
		@throw [OFInvalidArgumentException exception];

	otherCStringLength = string.UTF8StringLength;
	minimumCStringLength = (_s->cStringLength > otherCStringLength
	    ? otherCStringLength : _s->cStringLength);

	if ((compare = memcmp(_s->cString, string.UTF8String,
	    minimumCStringLength)) == 0) {
		if (_s->cStringLength > otherCStringLength)
			return OF_ORDERED_DESCENDING;
		if (_s->cStringLength < otherCStringLength)
			return OF_ORDERED_ASCENDING;
		return OF_ORDERED_SAME;
	}

	if (compare > 0)
		return OF_ORDERED_DESCENDING;
	else
		return OF_ORDERED_ASCENDING;
}

- (of_comparison_result_t)caseInsensitiveCompare: (OFString *)string
{
	const char *otherCString;
	size_t otherCStringLength, minimumCStringLength;
#ifdef OF_HAVE_UNICODE_TABLES
	size_t i, j;
#endif
	int compare;

	if (string == self)
		return OF_ORDERED_SAME;

	otherCString = string.UTF8String;
	otherCStringLength = string.UTF8StringLength;

#ifdef OF_HAVE_UNICODE_TABLES
	if (!_s->isUTF8) {
#endif
		minimumCStringLength = (_s->cStringLength > otherCStringLength
		    ? otherCStringLength : _s->cStringLength);

		if ((compare = memcasecmp(_s->cString, otherCString,
		    minimumCStringLength)) == 0) {
			if (_s->cStringLength > otherCStringLength)
				return OF_ORDERED_DESCENDING;
			if (_s->cStringLength < otherCStringLength)
				return OF_ORDERED_ASCENDING;
			return OF_ORDERED_SAME;
		}

		if (compare > 0)
			return OF_ORDERED_DESCENDING;
		else
			return OF_ORDERED_ASCENDING;
#ifdef OF_HAVE_UNICODE_TABLES
	}

	i = j = 0;

	while (i < _s->cStringLength && j < otherCStringLength) {
		of_unichar_t c1, c2;
		ssize_t l1, l2;

		l1 = of_string_utf8_decode(_s->cString + i,
		    _s->cStringLength - i, &c1);
		l2 = of_string_utf8_decode(otherCString + j,
		    otherCStringLength - j, &c2);

		if (l1 <= 0 || l2 <= 0 || c1 > 0x10FFFF || c2 > 0x10FFFF)
			@throw [OFInvalidEncodingException exception];

		if (c1 >> 8 < OF_UNICODE_CASEFOLDING_TABLE_SIZE) {
			of_unichar_t tc =
			    of_unicode_casefolding_table[c1 >> 8][c1 & 0xFF];

			if (tc)
				c1 = tc;
		}

		if (c2 >> 8 < OF_UNICODE_CASEFOLDING_TABLE_SIZE) {
			of_unichar_t tc =
			    of_unicode_casefolding_table[c2 >> 8][c2 & 0xFF];

			if (tc)
				c2 = tc;
		}

		if (c1 > c2)
			return OF_ORDERED_DESCENDING;
		if (c1 < c2)
			return OF_ORDERED_ASCENDING;

		i += l1;
		j += l2;
	}

	if (_s->cStringLength - i > otherCStringLength - j)
		return OF_ORDERED_DESCENDING;
	else if (_s->cStringLength - i < otherCStringLength - j)
		return OF_ORDERED_ASCENDING;
#endif

	return OF_ORDERED_SAME;
}

- (unsigned long)hash
{
	uint32_t hash;

	if (_s->hashed)
		return _s->hash;

	OF_HASH_INIT(hash);

	for (size_t i = 0; i < _s->cStringLength; i++) {
		of_unichar_t c;
		ssize_t length;

		if ((length = of_string_utf8_decode(_s->cString + i,
		    _s->cStringLength - i, &c)) <= 0)
			@throw [OFInvalidEncodingException exception];

		OF_HASH_ADD(hash, (c & 0xFF0000) >> 16);

	_s->hash = hash;
	_s->hashed = true;

	return hash;
}

- (of_unichar_t)characterAtIndex: (size_t)idx
{
	of_unichar_t character;

	if (idx >= _s->length)
		@throw [OFOutOfRangeException exception];

	if (!_s->isUTF8)
		return _s->cString[idx];

	idx = of_string_utf8_get_position(_s->cString, idx, _s->cStringLength);

	if (of_string_utf8_decode(_s->cString + idx,
	    _s->cStringLength - idx, &character) <= 0)
		@throw [OFInvalidEncodingException exception];

	return character;
}

- (void)getCharacters: (of_unichar_t *)buffer inRange: (of_range_t)range
{
	/* TODO: Could be slightly optimized */
	void *pool = objc_autoreleasePoolPush();
	const of_unichar_t *characters = self.characters;

	if (range.length > SIZE_MAX - range.location ||
	    range.location + range.length > _s->length)
		@throw [OFOutOfRangeException exception];

	memcpy(buffer, characters + range.location,
	    range.length * sizeof(of_unichar_t));

	objc_autoreleasePoolPop(pool);
}

- (of_range_t)rangeOfString: (OFString *)string
		    options: (int)options
		      range: (of_range_t)range
{
	const char *cString = string.UTF8String;
	size_t cStringLength = string.UTF8StringLength;
	size_t rangeLocation, rangeLength;

	if (range.length > SIZE_MAX - range.location ||
	    range.location + range.length > _s->length)

		    _s->cStringLength - rangeLocation);
	} else {
		rangeLocation = range.location;
		rangeLength = range.length;
	}

	if (cStringLength == 0)
		return of_range(0, 0);

	if (cStringLength > rangeLength)
		return of_range(OF_NOT_FOUND, 0);

	if (options & OF_STRING_SEARCH_BACKWARDS) {
		for (size_t i = rangeLength - cStringLength;; i--) {
			if (memcmp(_s->cString + rangeLocation + i, cString,
			    cStringLength) == 0) {
				range.location += of_string_utf8_get_index(
				    _s->cString + rangeLocation, i);
				range.length = string.length;

				return range;
			}

			/* Did not match and we're at the last char */
			if (i == 0)
				return of_range(OF_NOT_FOUND, 0);
		}
	} else {
		for (size_t i = 0; i <= rangeLength - cStringLength; i++) {
			if (memcmp(_s->cString + rangeLocation + i, cString,
			    cStringLength) == 0) {
				range.location += of_string_utf8_get_index(
				    _s->cString + rangeLocation, i);
				range.length = string.length;

				return range;
			}
		}
	}

	return of_range(OF_NOT_FOUND, 0);
}

- (bool)containsString: (OFString *)string
{
	const char *cString = string.UTF8String;
	size_t cStringLength = string.UTF8StringLength;

	if (cStringLength == 0)
		return true;

	if (cStringLength > _s->cStringLength)
		return false;

	for (size_t i = 0; i <= _s->cStringLength - cStringLength; i++)
		if (memcmp(_s->cString + i, cString, cStringLength) == 0)
			return true;

	return false;
}

- (OFString *)substringWithRange: (of_range_t)range
{
	size_t start = range.location;
	size_t end = range.location + range.length;

	if (range.length > SIZE_MAX - range.location || end > _s->length)
		@throw [OFOutOfRangeException exception];

		return false;

	return (memcmp(_s->cString + (_s->cStringLength - cStringLength),
	    suffix.UTF8String, cStringLength) == 0);
}

- (OFArray *)componentsSeparatedByString: (OFString *)delimiter
				 options: (int)options
{
	void *pool;
	OFMutableArray *array;
	const char *cString;
	size_t cStringLength;
	bool skipEmpty = (options & OF_STRING_SKIP_EMPTY);
	size_t last;
	OFString *component;

	if (delimiter == nil)
		@throw [OFInvalidArgumentException exception];

	if (delimiter.length == 0)

	[array makeImmutable];

	objc_autoreleasePoolPop(pool);

	return array;
}

- (const of_unichar_t *)characters
{
	of_unichar_t *buffer = of_alloc(_s->length, sizeof(of_unichar_t));
	size_t i = 0, j = 0;

	while (i < _s->cStringLength) {
		of_unichar_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];
		}

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

	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
{
	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)
			buffer[j++] = OF_BSWAP32(c);
		else
			buffer[j++] = c;

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

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

#ifdef OF_HAVE_BLOCKS
- (void)enumerateLinesUsingBlock: (of_string_line_enumeration_block_t)block
{
	void *pool;