ObjFW  Diff

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

#import "unicode.h"

extern const OFChar16 of_iso_8859_2_table[];
extern const size_t of_iso_8859_2_table_offset;
extern const OFChar16 of_iso_8859_3_table[];
extern const size_t of_iso_8859_3_table_offset;
extern const OFChar16 of_iso_8859_15_table[];
extern const size_t of_iso_8859_15_table_offset;
extern const OFChar16 of_windows_1251_table[];
extern const size_t of_windows_1251_table_offset;
extern const OFChar16 of_windows_1252_table[];
extern const size_t of_windows_1252_table_offset;
extern const OFChar16 of_codepage_437_table[];
extern const size_t of_codepage_437_table_offset;
extern const OFChar16 of_codepage_850_table[];
extern const size_t of_codepage_850_table_offset;
extern const OFChar16 of_codepage_858_table[];
extern const size_t of_codepage_858_table_offset;
extern const OFChar16 of_mac_roman_table[];
extern const size_t of_mac_roman_table_offset;
extern const OFChar16 of_koi8_r_table[];
extern const size_t of_koi8_r_table_offset;
extern const OFChar16 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];

			return OFOrderedAscending;
	}

	return OFOrderedSame;
}

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

	for (size_t i = 0; i < UTF8Length; i++) {
		/* No sign of UTF-8 here */
		if OF_LIKELY (!(UTF8String[i] & 0x80))

	if (length != NULL)
		*length = tmpLength;

	return isUTF8;
}

size_t
of_string_utf8_get_index(const char *string, size_t position)
{
	size_t idx = position;

	for (size_t i = 0; i < position; i++)
		if OF_UNLIKELY ((string[i] & 0xC0) == 0x80)
			idx--;

	return idx;
}

size_t
of_string_utf8_get_position(const char *string, size_t idx, size_t length)
{
	for (size_t i = 0; i <= idx; i++)
		if OF_UNLIKELY ((string[i] & 0xC0) == 0x80)
			if (++idx > length)
				@throw [OFInvalidFormatException exception];

	return idx;

		}

		_s = &_storage;

		_s->cString = storage;
		_s->cStringLength = UTF8StringLength;

		switch (of_string_utf8_check(UTF8String, UTF8StringLength,
		    &_s->length)) {
		case 1:
			_s->isUTF8 = true;
			break;
		case -1:
			@throw [OFInvalidEncodingException exception];
		}

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

		if (encoding == OFStringEncodingUTF8 ||
		    encoding == OFStringEncodingASCII) {
			switch (of_string_utf8_check(cString, cStringLength,
			    &_s->length)) {
			case 1:
				if (encoding == OFStringEncodingASCII)
					@throw [OFInvalidEncodingException
					    exception];

				_s->isUTF8 = true;

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

				_s->isUTF8 = true;
				bytes = of_string_utf8_encode(
				    (uint8_t)cString[i], buffer);

				if (bytes == 0)
					@throw [OFInvalidEncodingException
					    exception];

				_s->cStringLength += bytes - 1;

			return self;
		}

		switch (encoding) {
#define CASE(encoding, var)				\
		case encoding:				\
			table = var;			\
			tableOffset = var##_offset;	\
			break;
#ifdef HAVE_ISO_8859_2
		CASE(OFStringEncodingISO8859_2, of_iso_8859_2_table)
#endif
#ifdef HAVE_ISO_8859_3
		CASE(OFStringEncodingISO8859_3, of_iso_8859_3_table)
#endif
#ifdef HAVE_ISO_8859_15
		CASE(OFStringEncodingISO8859_15, of_iso_8859_15_table)
#endif
#ifdef HAVE_WINDOWS_1251
		CASE(OFStringEncodingWindows1251, of_windows_1251_table)
#endif
#ifdef HAVE_WINDOWS_1252
		CASE(OFStringEncodingWindows1252, of_windows_1252_table)
#endif
#ifdef HAVE_CODEPAGE_437
		CASE(OFStringEncodingCodepage437, of_codepage_437_table)
#endif
#ifdef HAVE_CODEPAGE_850
		CASE(OFStringEncodingCodepage850, of_codepage_850_table)
#endif
#ifdef HAVE_CODEPAGE_858
		CASE(OFStringEncodingCodepage858, of_codepage_858_table)
#endif
#ifdef HAVE_MAC_ROMAN
		CASE(OFStringEncodingMacRoman, of_mac_roman_table)
#endif
#ifdef HAVE_KOI8_R
		CASE(OFStringEncodingKOI8R, of_koi8_r_table)
#endif
#ifdef HAVE_KOI8_U
		CASE(OFStringEncodingKOI8U, of_koi8_u_table)
#endif
#undef CASE
		default:
			@throw [OFInvalidEncodingException exception];
		}

		j = 0;

			unichar = table[character - tableOffset];

			if (unichar == 0xFFFF)
				@throw [OFInvalidEncodingException exception];

			_s->isUTF8 = true;
			byteLength = of_string_utf8_encode(unichar, buffer);

			if (byteLength == 0)
				@throw [OFInvalidEncodingException exception];

			_s->cStringLength += byteLength - 1;
			_s->cString = OFResizeMemory(_s->cString,
			    _s->cStringLength + 1, 1);

		if (UTF8StringLength >= 3 &&
		    memcmp(UTF8String, "\xEF\xBB\xBF", 3) == 0) {
			UTF8String += 3;
			UTF8StringLength -= 3;
		}

		switch (of_string_utf8_check(UTF8String, UTF8StringLength,
		    &_s->length)) {
		case 1:
			_s->isUTF8 = true;
			break;
		case -1:
			@throw [OFInvalidEncodingException exception];
		}

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

		j = 0;
		for (size_t i = 0; i < length; i++) {
			size_t len = of_string_utf8_encode(characters[i],
			    _s->cString + j);

			if (len == 0)
				@throw [OFInvalidEncodingException exception];

			if (len > 1)
				_s->isUTF8 = true;

				character = (((character & 0x3FF) << 10) |
				    (nextCharacter & 0x3FF)) + 0x10000;

				i++;
				_s->length--;
			}

			len = of_string_utf8_encode(character, _s->cString + j);

			if (len == 0)
				@throw [OFInvalidEncodingException exception];

			if (len > 1)
				_s->isUTF8 = true;

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

		j = 0;
		for (size_t i = 0; i < length; i++) {
			char buffer[4];
			size_t len = of_string_utf8_encode((swap
			    ? OFByteSwap32(characters[i])
			    : characters[i]),
			    buffer);

			switch (len) {
			case 1:
				_s->cString[j++] = buffer[0];

		if ((cStringLength = OFVASPrintF(&tmp, format.UTF8String,
		    arguments)) == -1)
			@throw [OFInvalidFormatException exception];

		_s->cStringLength = cStringLength;

		@try {
			switch (of_string_utf8_check(tmp, cStringLength,
			    &_s->length)) {
			case 1:
				_s->isUTF8 = true;
				break;
			case -1:
				@throw [OFInvalidEncodingException exception];
			}

	i = j = 0;

	while (i < _s->cStringLength && j < otherCStringLength) {
		OFUnichar 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) {
			OFUnichar tc =
			    of_unicode_casefolding_table[c1 >> 8][c1 & 0xFF];

			if (tc)
				c1 = tc;
		}

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

			if (tc)
				c2 = tc;
		}

		if (c1 > c2)
			return OFOrderedDescending;

	OFHashInit(&hash);

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

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

		OFHashAdd(&hash, (c & 0xFF0000) >> 16);
		OFHashAdd(&hash, (c & 0x00FF00) >> 8);
		OFHashAdd(&hash, c & 0x0000FF);

	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: (OFUnichar *)buffer inRange: (OFRange)range
{

	size_t rangeLocation, rangeLength;

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

	if (_s->isUTF8) {
		rangeLocation = of_string_utf8_get_position(
		    _s->cString, range.location, _s->cStringLength);
		rangeLength = of_string_utf8_get_position(
		    _s->cString + rangeLocation, range.length,
		    _s->cStringLength - rangeLocation);
	} else {
		rangeLocation = range.location;
		rangeLength = range.length;
	}

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

	if (cStringLength > rangeLength)
		return OFRangeMake(OFNotFound, 0);

	if (options & OFStringSearchBackwards) {
		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 OFRangeMake(OFNotFound, 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;
			}
		}
	}

	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];

	if (_s->isUTF8) {
		start = of_string_utf8_get_position(_s->cString, start,
		    _s->cStringLength);
		end = of_string_utf8_get_position(_s->cString, end,
		    _s->cStringLength);
	}

	return [OFString stringWithUTF8String: _s->cString + start
				       length: end - start];
}

	OFUnichar *buffer = OFAllocMemory(_s->length, sizeof(OFUnichar));
	size_t i = 0, j = 0;

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

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

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

	OFChar32 *buffer = OFAllocMemory(_s->length + 1, sizeof(OFChar32));
	size_t i = 0, j = 0;

	while (i < _s->cStringLength) {
		OFChar32 c;
		ssize_t cLen;

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

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