ObjFW  Diff

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

#import "unicode.h"

extern const OFChar16 _OFISO8859_2Table[];
extern const size_t _OFISO8859_2TableOffset;
extern const OFChar16 _OFISO8859_3Table[];
extern const size_t _OFISO8859_3TableOffset;
extern const OFChar16 _OFISO8859_15Table[];
extern const size_t _OFISO8859_15TableOffset;
extern const OFChar16 _OFWindows1251Table[];
extern const size_t _OFWindows1251TableOffset;
extern const OFChar16 _OFWindows1252Table[];
extern const size_t _OFWindows1252TableOffset;
extern const OFChar16 _OFCodepage437Table[];
extern const size_t _OFCodepage437TableOffset;
extern const OFChar16 _OFCodepage850Table[];
extern const size_t _OFCodepage850TableOffset;
extern const OFChar16 _OFCodepage858Table[];
extern const size_t _OFCodepage858TableOffset;
extern const OFChar16 _OFMacRomanTable[];
extern const size_t _OFMacRomanTableOffset;
extern const OFChar16 _OFKOI8RTable[];
extern const size_t _OFKOI8RTableOffset;
extern const OFChar16 _OFKOI8UTable[];
extern const size_t _OFKOI8UTableOffset;

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
_OFUTF8StringCheck(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;
}

static size_t
positionToIndex(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
_OFUTF8StringIndexToPosition(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 (_OFUTF8StringCheck(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 (_OFUTF8StringCheck(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 = _OFUTF8StringEncode(
				    (uint8_t)cString[i], buffer);

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

				_s->cStringLength += bytes - 1;

		switch (encoding) {
#define CASE(encoding, var)				\
		case encoding:				\
			table = var;			\
			tableOffset = var##Offset;	\
			break;
#ifdef HAVE_ISO_8859_2
		CASE(OFStringEncodingISO8859_2, _OFISO8859_2Table)
#endif
#ifdef HAVE_ISO_8859_3
		CASE(OFStringEncodingISO8859_3, _OFISO8859_3Table)
#endif
#ifdef HAVE_ISO_8859_15
		CASE(OFStringEncodingISO8859_15, _OFISO8859_15Table)
#endif
#ifdef HAVE_WINDOWS_1251
		CASE(OFStringEncodingWindows1251, _OFWindows1251Table)
#endif
#ifdef HAVE_WINDOWS_1252
		CASE(OFStringEncodingWindows1252, _OFWindows1252Table)
#endif
#ifdef HAVE_CODEPAGE_437
		CASE(OFStringEncodingCodepage437, _OFCodepage437Table)
#endif
#ifdef HAVE_CODEPAGE_850
		CASE(OFStringEncodingCodepage850, _OFCodepage850Table)
#endif
#ifdef HAVE_CODEPAGE_858
		CASE(OFStringEncodingCodepage858, _OFCodepage858Table)
#endif
#ifdef HAVE_MAC_ROMAN
		CASE(OFStringEncodingMacRoman, _OFMacRomanTable)
#endif
#ifdef HAVE_KOI8_R
		CASE(OFStringEncodingKOI8R, _OFKOI8RTable)
#endif
#ifdef HAVE_KOI8_U
		CASE(OFStringEncodingKOI8U, _OFKOI8UTable)
#endif
#undef CASE
		default:
			@throw [OFInvalidArgumentException exception];
		}

		j = 0;

			unichar = table[character - tableOffset];

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

			_s->isUTF8 = true;
			byteLength = _OFUTF8StringEncode(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 (_OFUTF8StringCheck(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 = _OFUTF8StringEncode(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 = _OFUTF8StringEncode(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 = _OFUTF8StringEncode((swap
			    ? OFByteSwap32(characters[i])
			    : characters[i]),
			    buffer);

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

		int cStringLength;

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

		_s = &_storage;

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

		_s->cStringLength = cStringLength;

		@try {
			switch (_OFUTF8StringCheck(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 = _OFUTF8StringDecode(_s->cString + i,
		    _s->cStringLength - i, &c1);
		l2 = _OFUTF8StringDecode(otherCString + j,
		    otherCStringLength - j, &c2);

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

		if (c1 >> 8 < _OFUnicodeCaseFoldingTableSize) {
			OFUnichar tc =
			    _OFUnicodeCaseFoldingTable[c1 >> 8][c1 & 0xFF];

			if (tc)
				c1 = tc;
		}

		if (c2 >> 8 < _OFUnicodeCaseFoldingTableSize) {
			OFUnichar tc =
			    _OFUnicodeCaseFoldingTable[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 = _OFUTF8StringDecode(_s->cString + i,
		    _s->cStringLength - i, &c)) <= 0)
			@throw [OFInvalidEncodingException exception];

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

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

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

	idx = _OFUTF8StringIndexToPosition(_s->cString, idx, _s->cStringLength);

	if (_OFUTF8StringDecode(_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 = _OFUTF8StringIndexToPosition(
		    _s->cString, range.location, _s->cStringLength);
		rangeLength = _OFUTF8StringIndexToPosition(
		    _s->cString + rangeLocation, range.length,
		    _s->cStringLength - rangeLocation);
	} else {
		rangeLocation = range.location;
		rangeLength = range.length;
	}

	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 = _OFUTF8StringIndexToPosition(_s->cString, start,
		    _s->cStringLength);
		end = _OFUTF8StringIndexToPosition(_s->cString, end,
		    _s->cStringLength);
	}

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

	size_t i = 0, j = 0;
	const OFUnichar *ret;

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

		cLen = _OFUTF8StringDecode(_s->cString + i,
		    _s->cStringLength - i, &c);

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

	size_t i = 0, j = 0;
	const OFChar32 *ret;

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

		cLen = _OFUTF8StringDecode(_s->cString + i,
		    _s->cStringLength - i, &c);

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