*----------------------------------------------------------------------
 */

static int
DictionaryCompare(
    const char *left, const char *right)	/* The strings to compare. */
{
    int uniLeft, uniRight, uniLeftLower, uniRightLower;
    int diff, zeros;
    int secondaryDiff = 0;

    while (1) {
	if (isdigit(UCHAR(*right))		/* INTL: digit */
		&& isdigit(UCHAR(*left))) {	/* INTL: digit */
	    /*

	/*
	 * Convert character to Unicode for comparison purposes. If either
	 * string is at the terminating null, do a byte-wise comparison and
	 * bail out immediately.
	 */

	if ((*left != '\0') && (*right != '\0')) {
	    Tcl_UniChar lch = 0, rch = 0;

	    left += TclUtfToUniChar(left, &lch);
	    right += TclUtfToUniChar(right, &rch);

	    uniLeft = lch;
	    uniRight = rch;
#if TCL_UTF_MAX == 3
	    if (*left && ((lch & 0xFC00) == 0xD800)) {
		int len2 = TclUtfToUniChar(left, &lch);

		if ((lch & 0xFC00) == 0xDC00) {
		    uniLeft = (((uniLeft&0x3FF)<<10) | (lch&0x3FF)) + 0x10000;
		    left += len2;
		}
	    }
	    if (*right && ((rch & 0xFC00) == 0xD800)) {
		int len2 = TclUtfToUniChar(right, &rch);

		if ((rch & 0xFC00) == 0xDC00) {
		    uniRight = (((uniRight&0x3FF)<<10) | (rch&0x3FF)) + 0x10000;
		    right += len2;
		}
	    }
#endif
	    /*
	     * Convert both chars to lower for the comparison, because
	     * dictionary sorts are case insensitve. Covert to lower, not
	     * upper, so chars between Z and a will sort before A (where most
	     * other interesting punctuations occur).
	     */

	    uniLeftLower = TclUCS4ToLower(uniLeft);
	    uniRightLower = TclUCS4ToLower(uniRight);
	} else {
	    diff = UCHAR(*left) - UCHAR(*right);
	    break;
	}

	diff = uniLeftLower - uniRightLower;
	if (diff) {

static Tcl_NRPostProc	SwitchPostProc;
static Tcl_NRPostProc	TryPostBody;
static Tcl_NRPostProc	TryPostFinal;
static Tcl_NRPostProc	TryPostHandler;
static int		UniCharIsAscii(int character);
static int		UniCharIsHexDigit(int character);

#if TCL_UTF_MAX == 3
static int		NumCodePointsUtf(const char *src, int length);
static int		NumCodePointsUnicode(const Tcl_UniChar *src,
			    int length);
static int		UniCharNcmp(const Tcl_UniChar *ucs,
			    const Tcl_UniChar *uct, unsigned long numCp);
static int		UtfNcasecmp(const char *cs, const char *ct,
			    unsigned long numCp);
static int		UtfNcmp(const char *cs, const char *ct,
			    unsigned long numCp);
#endif

/*
 * Default set of characters to trim in [string trim] and friends. This is a
 * UTF-8 literal string containing all Unicode space characters [TIP #413]
 */

const char tclDefaultTrimSet[] =
	"\x09\x0a\x0b\x0c\x0d " /* ASCII */

	"\xe2\x80\xa9" /* paragraph separator (U+2029) */
	"\xe2\x80\xaf" /* narrow no-break space (U+202f) */
	"\xe2\x81\x9f" /* medium mathematical space (U+205f) */
	"\xe2\x81\xa0" /* word joiner (U+2060) */
	"\xe3\x80\x80" /* ideographic space (U+3000) */
	"\xef\xbb\xbf" /* zero width no-break space (U+feff) */
;

#if TCL_UTF_MAX == 3
/*
 *---------------------------------------------------------------------------
 *
 * NumCodePointsUtf --
 *
 *	Like Tcl_NumUtfChars() but returns the number of code points.
 *	Problem: single high surrogates (0xD800..0xDBFF) at the very
 *	end of the string are not counted. If they were, the functions
 *	UtfNcmp() and UtfNcasecmp() would read beyond the buffer.
 *
 * Results:
 *	As above.
 *
 * Side effects:
 *	None.
 *
 *---------------------------------------------------------------------------
 */

static int
NumCodePointsUtf(
    const char *src,		/* The UTF-8 string to measure. */
    int length)			/* The length of the string in bytes. */
{
    Tcl_UniChar ch = 0;
    int i = 0;

    const char *endPtr = src + length - TCL_UTF_MAX;

    while (src < endPtr) {
	src += TclUtfToUniChar(src, &ch);
	if ((ch & 0xFC00) == 0xD800) {
	    if ((src < endPtr) && Tcl_UtfCharComplete(src, endPtr - src)) {
		int len = TclUtfToUniChar(src, &ch);

		if ((ch & 0xFC00) == 0xDC00) {
		    --i;
		    src += len;
		}
	    }
	}
	i++;
    }
    endPtr += TCL_UTF_MAX;
    while ((src < endPtr) && Tcl_UtfCharComplete(src, endPtr - src)) {
	src += TclUtfToUniChar(src, &ch);
	if ((ch & 0xFC00) == 0xD800) {
	    if ((src < endPtr) && Tcl_UtfCharComplete(src, endPtr - src)) {
		int len = TclUtfToUniChar(src, &ch);

		if ((ch & 0xFC00) == 0xDC00) {
		    --i;
		    src += len;
		}
	    }
	}
	i++;
    }
    if (src < endPtr) {
	i += endPtr - src;
    } else if (i && ((ch & 0xFC00) == 0xD800)) {
	--i;
    }
    return i;
}
#endif

#if TCL_UTF_MAX == 3
/*
 *----------------------------------------------------------------------
 *
 * UtfNcmp --
 *
 *	Like Tcl_UtfNcmp() but the limit is guaranteed and specified in
 *	code points.
 *
 * Results:
 *	Return <0 if cs < ct, 0 if cs == ct, or >0 if cs > ct.
 *
 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

static int
UtfNcmp(
    const char *cs,		/* UTF string to compare to ct. */
    const char *ct,		/* UTF string cs is compared to. */
    unsigned long numCp)	/* Number of code points to compare. */
{
    Tcl_UniChar ch1 = 0, ch2 = 0;
    int uch1, uch2;

    while (numCp-- > 0) {

	cs += TclUtfToUniChar(cs, &ch1);
	ct += TclUtfToUniChar(ct, &ch2);

	uch1 = ch1;
	uch2 = ch2;

	if ((ch1 & 0xFC00) == 0xD800) {
	    int len = TclUtfToUniChar(cs, &ch1);

	    if ((ch1 & 0xFC00) == 0xDC00) {
		uch1 = (((uch1&0x3FF)<<10) | (ch1&0x3FF)) + 0x10000;
		cs += len;
	    }
	}
	if ((ch2 & 0xFC00) == 0xD800) {
	    int len = TclUtfToUniChar(ct, &ch2);

	    if ((ch2 & 0xFC00) == 0xDC00) {
		uch2 = (((uch2&0x3FF)<<10) | (ch2&0x3FF)) + 0x10000;
		ct += len;
	    }
	}

	if (uch1 != uch2) {
	    return (uch1 - uch2);
	}
    }
    return 0;
}
#endif

#if TCL_UTF_MAX == 3
/*
 *----------------------------------------------------------------------
 *
 * UtfNcasecmp --
 *
 *	Like Tcl_UtfNcasecmp() but the limit is guaranteed and specified in
 *	code points.
 *
 * Results:
 *	Return <0 if cs < ct, 0 if cs == ct, or >0 if cs > ct.
 *
 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

static int
UtfNcasecmp(
    const char *cs,		/* UTF string to compare to ct. */
    const char *ct,		/* UTF string cs is compared to. */
    unsigned long numCp)	/* Number of code points to compare. */
{
    Tcl_UniChar ch1 = 0, ch2 = 0;
    int uch1, uch2;

    while (numCp-- > 0) {

	cs += TclUtfToUniChar(cs, &ch1);
	ct += TclUtfToUniChar(ct, &ch2);

	uch1 = ch1;
	uch2 = ch2;

	if ((ch1 & 0xFC00) == 0xD800) {
	    int len = TclUtfToUniChar(cs, &ch1);

	    if ((ch1 & 0xFC00) == 0xDC00) {
		uch1 = (((uch1&0x3FF)<<10) | (ch1&0x3FF)) + 0x10000;
		cs += len;
	    }
	}
	if ((ch2 & 0xFC00) == 0xD800) {
	    int len = TclUtfToUniChar(ct, &ch2);

	    if ((ch2 & 0xFC00) == 0xDC00) {
		uch2 = (((uch2&0x3FF)<<10) | (ch2&0x3FF)) + 0x10000;
		ct += len;
	    }
	}

	if (uch1 != uch2) {
	    uch1 = TclUCS4ToLower(uch1);
	    uch2 = TclUCS4ToLower(uch2);
	    if (uch1 != uch2) {
		return (uch1 - uch2);
	    }
	}
    }
    return 0;
}
#endif

#if TCL_UTF_MAX == 3
/*
 *---------------------------------------------------------------------------
 *
 * NumCodePointsUnicode --
 *
 *	Returns the number of code points of a Tcl_UniChar array.
 *	Problem: single high surrogates (0xD800..0xDBFF) at the
 *	very end of the array are not counted. If they were, the
 *	functions UniCharNcmp() and UniCharNcasecmp() would read
 *	beyond the buffer.
 *
 * Results:
 *	As above.
 *
 * Side effects:
 *	None.
 *
 *---------------------------------------------------------------------------
 */

static int
NumCodePointsUnicode(
    const Tcl_UniChar *src,	/* The array to measure. */
    int length)			/* The length of the array in elements. */
{
    int i, n = 0;

    for (i = 0; i < length; i++, n++) {
	if ((src[i] & 0xFC00) == 0xD800) {
	    if (i + 1 >= length) {
		n--;
	    }
	    if ((i + 1 < length) && ((src[i+1] & 0xFC00) == 0xDC00)) {
		i++;
	    }
	}
    }
    return n;
}
#endif

#if TCL_UTF_MAX == 3
/*
 *----------------------------------------------------------------------
 *
 * UniCharNcmp --
 *
 *	Like Tcl_UniCharNcmp() but the limit is guaranteed and specified in
 *	code points.
 *
 * Results:
 *	Return <0 if ucs < uct, 0 if ucs == uct, or >0 if ucs > uct.
 *
 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

static int
UniCharNcmp(
    const Tcl_UniChar *ucs,	/* Unicode string to compare to uct. */
    const Tcl_UniChar *uct,	/* Unicode string ucs is compared to. */
    unsigned long numCp)	/* Number of code points to compare. */
{
    int lcs, lct;

    for ( ; numCp != 0; numCp--, ucs++, uct++) {
	lcs = *ucs;
	lct = *uct;
	if ((lcs & 0xFC00) == 0xD800) {
	    if ((ucs[1] & 0xFC00) == 0xDC00) {
		lcs = (((lcs&0x3FF)<<10) | (ucs[1]&0x3FF)) + 0x10000;
		ucs++;
	    }
	}
	if ((lct & 0xFC00) == 0xD800) {
	    if ((uct[1] & 0xFC00) == 0xDC00) {
		lct = (((lct&0x3FF)<<10) | (uct[1]&0x3FF)) + 0x10000;
		uct++;
	    }
	}
	if (lcs != lct) {
	    return (lcs - lct);
	}
    }
    return 0;
}
#endif

#if TCL_UTF_MAX == 3
/*
 *----------------------------------------------------------------------
 *
 * UniCharNcasecmp --
 *
 *	Like Tcl_UniCharNcasecmp() but the limit is guaranteed and specified in
 *	code points.
 *
 * Results:
 *	Return <0 if ucs < uct, 0 if ucs == uct, or >0 if ucs > uct.
 *
 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

static int
UniCharNcasecmp(
    const Tcl_UniChar *ucs,	/* Unicode string to compare to uct. */
    const Tcl_UniChar *uct,	/* Unicode string ucs is compared to. */
    unsigned long numCp)	/* Number of code points to compare. */
{
    int lcs, lct;

    for ( ; numCp != 0; numCp--, ucs++, uct++) {
	lcs = *ucs;
	lct = *uct;
	if ((lcs & 0xFC00) == 0xD800) {
	    if ((ucs[1] & 0xFC00) == 0xDC00) {
		lcs = (((lcs&0x3FF)<<10) | (ucs[1]&0x3FF)) + 0x10000;
		ucs++;
	    }
	}
	if ((lct & 0xFC00) == 0xD800) {
	    if ((uct[1] & 0xFC00) == 0xDC00) {
		lct = (((lct&0x3FF)<<10) | (uct[1]&0x3FF)) + 0x10000;
		uct++;
	    }
	}
	if (lcs != lct) {
	    lcs = TclUCS4ToLower(lcs);
	    lct = TclUCS4ToLower(lct);
	    if (lcs != lct) {
		return (lcs - lct);
	    }
	}
    }
    return 0;
}
#endif

/*
 *----------------------------------------------------------------------
 *
 * Tcl_PwdObjCmd --
 *
 *	This procedure is invoked to process the "pwd" Tcl command. See the

	 * benchmark testing this proved the most efficient check between the
	 * unicode and string comparison operations.
	 */

	if (nocase) {
	    s1 = (char *) Tcl_GetUnicodeFromObj(value1Ptr, &s1len);
	    s2 = (char *) Tcl_GetUnicodeFromObj(value2Ptr, &s2len);
#if TCL_UTF_MAX == 3
	    s1len = NumCodePointsUnicode((Tcl_UniChar *) s1, s1len);
	    s2len = NumCodePointsUnicode((Tcl_UniChar *) s2, s2len);
	    memCmpFn = (memCmpFn_t) UniCharNcasecmp;
#else
	    memCmpFn = (memCmpFn_t)Tcl_UniCharNcasecmp;
#endif
	} else {
	    s1len = Tcl_GetCharLength(value1Ptr);
	    s2len = Tcl_GetCharLength(value2Ptr);
	    if ((s1len == value1Ptr->length)
		    && (value1Ptr->bytes != NULL)
		    && (s2len == value2Ptr->length)
		    && (value2Ptr->bytes != NULL)) {

			checkEq
#endif /* WORDS_BIGENDIAN */
		        ) {
		    memCmpFn = memcmp;
		    s1len *= sizeof(Tcl_UniChar);
		    s2len *= sizeof(Tcl_UniChar);
		} else {
#if TCL_UTF_MAX == 3
		    s1len = NumCodePointsUnicode((Tcl_UniChar *) s1, s1len);
		    s2len = NumCodePointsUnicode((Tcl_UniChar *) s2, s2len);
		    memCmpFn = (memCmpFn_t) UniCharNcmp;
#else
		    memCmpFn = (memCmpFn_t) Tcl_UniCharNcmp;
#endif
		}
	    }
	}
    } else {
	/*
	 * Get the string representations, being careful in case we have
	 * special empty string objects about.

	} else {
	    /*
	     * As a catch-all we will work with UTF-8. We cannot use memcmp()
	     * as that is unsafe with any string containing NUL (\xC0\x80 in
	     * Tcl's utf rep). We can use the more efficient TclpUtfNcmp2 if
	     * we are case-sensitive and no specific length was requested.
	     */
#if TCL_UTF_MAX == 3
	    s1len = NumCodePointsUtf(s1, s1len);
	    s2len = NumCodePointsUtf(s2, s2len);
	    memCmpFn = (memCmpFn_t)
		    (nocase ? UtfNcasecmp : UtfNcmp);
#else
	    if ((reqlength < 0) && !nocase) {
		memCmpFn = (memCmpFn_t) TclpUtfNcmp2;
	    } else {
		s1len = Tcl_NumUtfChars(s1, s1len);
		s2len = Tcl_NumUtfChars(s2, s2len);
		memCmpFn = (memCmpFn_t)
			(nocase ? Tcl_UtfNcasecmp : Tcl_UtfNcmp);
	    }
#endif
	}
    }

    length = (s1len < s2len) ? s1len : s2len;
    if (reqlength > 0 && reqlength < length) {
	length = reqlength;
    } else if (reqlength < 0) {

MODULE_SCOPE void *	TclpThreadGetMasterTSD(void *tsdKeyPtr);

MODULE_SCOPE void	TclErrorStackResetIf(Tcl_Interp *interp, const char *msg, int length);

#if TCL_UTF_MAX > 3
MODULE_SCOPE int	TclCollapseSurrogatePair(Tcl_Token *tokenPtr,
			    int *numReadPtr, char *buffer);
#define	TclUCS4ToUpper(ch) Tcl_UniCharToUpper((ch))
#define	TclUCS4ToLower(ch) Tcl_UniCharToLower((ch))
#define	TclUCS4ToTitle(ch) Tcl_UniCharToTitle((ch))
#else
MODULE_SCOPE int	TclUniCharToUtfExt(int ch, char *buf);
MODULE_SCOPE int	TclUtfToUniCharExt(const char *src, int *chPtr);
MODULE_SCOPE int	TclUtfCharCompleteExt(const char *src, int len);
MODULE_SCOPE int	TclUCS4ToUpper(int ch);
MODULE_SCOPE int	TclUCS4ToLower(int ch);
MODULE_SCOPE int	TclUCS4ToTitle(int ch);
#endif

/*
 * Many parsing tasks need a common definition of whitespace.
 * Use this routine and macro to achieve that and place
 * optimization (fragile on changes) in one place.
 */

int
TclCollapseSurrogatePair(
    Tcl_Token *tokenPtr,	/* Pointer to token to be checked. */
    int *numReadPtr,		/* Pointer to number of consumed input chars. */
    char *buffer)		/* Buffer holding UTF data of previous token. */
{
    int count, numRead;
    Tcl_UniChar ch = 0, ch2 = 0;
    char buffer2[TCL_UTF_MAX];

    TclUtfToUniChar(buffer, &ch);
    if ((ch <= 0xFFFF) && ((ch & 0xFC00) == 0xD800)) {
	if (tokenPtr->type == TCL_TOKEN_BS) {
	    count = TclParseBackslash(tokenPtr->start, tokenPtr->size,
			    &numRead, buffer2);
	    if (count <= 0) {
		return 0;
	    }
	    TclUtfToUniChar(buffer2, &ch2);
	    if ((ch2 <= 0xFFFF) && ((ch2 & 0xFC00) == 0xDC00)) {
		unsigned int uch = ((ch & 0x3FF) << 10) + (ch2 & 0x3FF);

		uch += 0x10000;
		if (numReadPtr != NULL) {
		    *numReadPtr = numRead;
		}

/*
 * Functions used only in this module.
 */

static int		UtfCount(int ch);
static int		Invalid(unsigned char *src);






/*
 *---------------------------------------------------------------------------
 *
 * UtfCount --
 *
 *	Find the number of bytes in the Utf character "ch".

	    int len2 = TclUtfToUniChar(src + len, &ch);

	    if ((ch & 0xFC00) == 0xDC00) {
		len += len2;
		upChar = (((upChar&0x3FF)<<10) | (ch&0x3FF)) + 0x10000;
	    }
	}
	upChar = TclUCS4ToUpper(upChar);
#else
	upChar = Tcl_UniCharToUpper(ch);
#endif
	/*
	 * To keep badly formed Utf strings from getting inflated by the
	 * conversion (thereby causing a segfault), only copy the upper case
	 * char to dst if its size is <= the original char.

	    int len2 = TclUtfToUniChar(src + len, &ch);

	    if ((ch & 0xFC00) == 0xDC00) {
		len += len2;
		lowChar = (((lowChar&0x3FF)<<10) | (ch&0x3FF)) + 0x10000;
	    }
	}
	lowChar = TclUCS4ToLower(lowChar);
#else
	lowChar = Tcl_UniCharToLower(ch);
#endif

	/*
	 * To keep badly formed Utf strings from getting inflated by the
	 * conversion (thereby causing a segfault), only copy the lower case

	    int len2 = TclUtfToUniChar(src + len, &ch);

	    if ((ch & 0xFC00) == 0xDC00) {
		len += len2;
		titleChar = (((titleChar&0x3FF)<<10) | (ch&0x3FF)) + 0x10000;
	    }
	}
	titleChar = TclUCS4ToTitle(titleChar);
#else
	titleChar = Tcl_UniCharToTitle(ch);
#endif

	if (len < UtfCount(titleChar)) {
	    memmove(dst, src, len);
	    dst += len;

	    if ((ch & 0xFC00) == 0xDC00) {
		len += len2;
		lowChar = (((lowChar&0x3FF)<<10) | (ch&0x3FF)) + 0x10000;
	    }
	}
	/* Special exception for Georgian Asomtavruli chars, no titlecase. */
	if ((unsigned)(lowChar - 0x1C90) >= 0x30) {
	    lowChar = TclUCS4ToLower(lowChar);
	}
#else
	/* Special exception for Georgian Asomtavruli chars, no titlecase. */
	if ((unsigned)(lowChar - 0x1C90) >= 0x30) {
	    lowChar = Tcl_UniCharToLower(lowChar);
	}
#endif

int
Tcl_UtfNcmp(
    const char *cs,		/* UTF string to compare to ct. */
    const char *ct,		/* UTF string cs is compared to. */
    unsigned long numChars)	/* Number of UTF chars to compare. */
{
    Tcl_UniChar ch1 = 0, ch2 = 0;
    int uch1, uch2;
#if TCL_UTF_MAX == 3
    int num1 = numChars, num2 = numChars;
#endif

    /*
     * Cannot use 'memcmp(cs, ct, n);' as byte representation of \u0000 (the
     * pair of bytes 0xC0,0x80) is larger than byte representation of \u0001
     * (the byte 0x01.)
     */

    while (
#if TCL_UTF_MAX == 3
	(num1-- > 0) && (num2 > 0)
#else
	numChars-- > 0
#endif
    ) {
	/*
	 * n must be interpreted as chars, not bytes. This should be called
	 * only when both strings are of at least n chars long (no need for \0
	 * check)
	 */

	cs += TclUtfToUniChar(cs, &ch1);
	ct += TclUtfToUniChar(ct, &ch2);

	uch1 = ch1;
	uch2 = ch2;

#if TCL_UTF_MAX == 3
	if ((num1 > 0) && ((ch1 & 0xFC00) == 0xD800)) {
	    int len = TclUtfToUniChar(cs, &ch1);

	    if ((ch1 & 0xFC00) == 0xDC00) {
		uch1 = (((uch1&0x3FF)<<10) | (ch1&0x3FF)) + 0x10000;
		cs += len;
		num1--;
	    }
	}
	if ((num2 > 0) && ((ch2 & 0xFC00) == 0xD800)) {

	    int len = TclUtfToUniChar(ct, &ch2);

	    if ((ch2 & 0xFC00) == 0xDC00) {
		uch2 = (((uch2&0x3FF)<<10) | (ch2&0x3FF)) + 0x10000;
		ct += len;
		num2--;
	    }
	}
#endif

	if (uch1 != uch2) {
	    return (uch1 - uch2);
	}
    }
    return 0;
}

/*
 *----------------------------------------------------------------------

int
Tcl_UtfNcasecmp(
    const char *cs,		/* UTF string to compare to ct. */
    const char *ct,		/* UTF string cs is compared to. */
    unsigned long numChars)	/* Number of UTF chars to compare. */
{
    Tcl_UniChar ch1 = 0, ch2 = 0;
    int uch1, uch2;
#if TCL_UTF_MAX == 3
    int num1 = numChars, num2 = numChars;
#endif

    while (
#if TCL_UTF_MAX == 3
	(num1-- > 0) && (num2-- > 0)
#else
	numChars-- > 0
#endif
    ) {
	
	/*
	 * n must be interpreted as chars, not bytes.
	 * This should be called only when both strings are of
	 * at least n chars long (no need for \0 check)
	 */
	cs += TclUtfToUniChar(cs, &ch1);
	ct += TclUtfToUniChar(ct, &ch2);

	uch1 = ch1;
	uch2 = ch2;

#if TCL_UTF_MAX == 3
	if ((num1 > 0) && ((ch1 & 0xFC00) == 0xD800)) {
	    int len = TclUtfToUniChar(cs, &ch1);

	    if ((ch1 & 0xFC00) == 0xDC00) {
		uch1 = (((uch1&0x3FF)<<10) | (ch1&0x3FF)) + 0x10000;
		cs += len;
		num1--;
	    }
	}
	if ((num2 > 0) && ((ch2 & 0xFC00) == 0xD800)) {

	    int len = TclUtfToUniChar(ct, &ch2);

	    if ((ch2 & 0xFC00) == 0xDC00) {
		uch2 = (((uch2&0x3FF)<<10) | (ch2&0x3FF)) + 0x10000;
		ct += len;
		num2--;
	    }
	}
#endif

	if (uch1 != uch2) {
	    uch1 = TclUCS4ToLower(uch1);
	    uch2 = TclUCS4ToLower(uch2);
	    if (uch1 != uch2) {
		return (uch1 - uch2);
	    }
	}
    }
    return 0;
}

/*

int
TclUtfCasecmp(
    const char *cs,		/* UTF string to compare to ct. */
    const char *ct)		/* UTF string cs is compared to. */
{
    Tcl_UniChar ch1 = 0, ch2 = 0;
    int uch1, uch2;

    while (*cs && *ct) {
	cs += TclUtfToUniChar(cs, &ch1);
	ct += TclUtfToUniChar(ct, &ch2);

	uch1 = ch1;
	uch2 = ch2;

#if TCL_UTF_MAX == 3

	if (*cs && ((ch1 & 0xFC00) == 0xD800)) {
	    int len = TclUtfToUniChar(cs, &ch1);

	    if ((ch1 & 0xFC00) == 0xDC00) {
		uch1 = (((uch1&0x3FF)<<10) | (ch1&0x3FF)) + 0x10000;
		cs += len;
	    }
	}
	if (*ct && ((ch2 & 0xFC00) == 0xD800)) {
	    int len = TclUtfToUniChar(ct, &ch2);

	    if ((ch2 & 0xFC00) == 0xDC00) {
		uch2 = (((uch2&0x3FF)<<10) | (ch2&0x3FF)) + 0x10000;
		ct += len;
	    }
	}
#endif

	if (uch1 != uch2) {
#if TCL_UTF_MAX == 3
	    uch1 = TclUCS4ToLower(uch1);
	    uch2 = TclUCS4ToLower(uch2);
#else
	    uch1 = Tcl_UniCharToLower(uch1);
	    uch2 = Tcl_UniCharToLower(uch2);
#endif
	    if (uch1 != uch2) {
		return uch1 - uch2;
	    }
	}
    }
    return UCHAR(*cs) - UCHAR(*ct);
}



 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

#if TCL_UTF_MAX == 3
int
TclUCS4ToUpper(
    int ch)			/* Unicode character to convert. */
{
    if (!UNICODE_OUT_OF_RANGE(ch)) {
	int info = GetUniCharInfo(ch);

	if (GetCaseType(info) & 0x04) {
	    ch -= GetDelta(info);

 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

#if TCL_UTF_MAX == 3
int
TclUCS4ToLower(
    int ch)			/* Unicode character to convert. */
{
    if (!UNICODE_OUT_OF_RANGE(ch)) {
	int info = GetUniCharInfo(ch);
	int mode = GetCaseType(info);

	if ((mode & 0x02) && (mode != 0x7)) {

 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

#if TCL_UTF_MAX == 3
int
TclUCS4ToTitle(
    int ch)			/* Unicode character to convert. */
{
    if (!UNICODE_OUT_OF_RANGE(ch)) {
	int info = GetUniCharInfo(ch);
	int mode = GetCaseType(info);

	if (mode & 0x1) {

int
Tcl_UniCharNcmp(
    const Tcl_UniChar *ucs,	/* Unicode string to compare to uct. */
    const Tcl_UniChar *uct,	/* Unicode string ucs is compared to. */
    unsigned long numChars)	/* Number of unichars to compare. */
{
#if TCL_UTF_MAX == 3
    int lcs, lct, nums = numChars, numt = numChars;

    for ( ; nums != 0 && numt != 0; nums--, numt--, ucs++, uct++) {
	lcs = *ucs;
	lct = *uct;
	if ((nums > 1) && ((lcs & 0xFC00) == 0xD800)) {
	    if ((ucs[1] & 0xFC00) == 0xDC00) {
		lcs = (((lcs&0x3FF)<<10) | (ucs[1]&0x3FF)) + 0x10000;
		ucs++;
		nums--;
	    }
	}
	if ((numt > 1) && ((lct & 0xFC00) == 0xD800)) {
	    if ((uct[1] & 0xFC00) == 0xDC00) {
		lct = (((lct&0x3FF)<<10) | (uct[1]&0x3FF)) + 0x10000;
		uct++;
		numt--;
	    }
	}
	if (lcs != lct) {
	    return (lcs - lct);
	}
    }
    return 0;
#else
#ifdef WORDS_BIGENDIAN
    /*
     * We are definitely on a big-endian machine; memcmp() is safe
     */

    return memcmp(ucs, uct, numChars*sizeof(Tcl_UniChar));

#else /* !WORDS_BIGENDIAN */
    /*
     * We can't simply call memcmp() because that is not lexically correct.
     */

    for ( ; numChars != 0; ucs++, uct++, numChars--) {
	if (*ucs != *uct) {
	    return (*ucs - *uct);
	}
    }
    return 0;
#endif /* WORDS_BIGENDIAN */
#endif /* TCL_UTF_MAX == 3 */
}

/*
 *----------------------------------------------------------------------
 *
 * Tcl_UniCharNcasecmp --
 *

int
Tcl_UniCharNcasecmp(
    const Tcl_UniChar *ucs,	/* Unicode string to compare to uct. */
    const Tcl_UniChar *uct,	/* Unicode string ucs is compared to. */
    unsigned long numChars)	/* Number of unichars to compare. */
{
#if TCL_UTF_MAX == 3
    int lcs, lct, nums = numChars, numt = numChars;

    for ( ; nums != 0 && numt != 0; nums--, numt--, ucs++, uct++) {
	lcs = *ucs;
	lct = *uct;
	if ((nums > 1) && ((lcs & 0xFC00) == 0xD800)) {
	    if ((ucs[1] & 0xFC00) == 0xDC00) {
		lcs = (((lcs&0x3FF)<<10) | (ucs[1]&0x3FF)) + 0x10000;
		ucs++;
		nums--;
	    }
	}
	if ((numt > 1) && ((lct & 0xFC00) == 0xD800)) {
	    if ((uct[1] & 0xFC00) == 0xDC00) {
		lct = (((lct&0x3FF)<<10) | (uct[1]&0x3FF)) + 0x10000;
		uct++;
		numt--;
	    }
	}
	if (lcs != lct) {
	    lcs = TclUCS4ToLower(lcs);
	    lct = TclUCS4ToLower(lct);
	    if (lcs != lct) {
		return (lcs - lct);
	    }
	}
    }
#else
    for ( ; numChars != 0; numChars--, ucs++, uct++) {
	if (*ucs != *uct) {
	    Tcl_UniChar lcs = Tcl_UniCharToLower(*ucs);
	    Tcl_UniChar lct = Tcl_UniCharToLower(*uct);

	    if (lcs != lct) {
		return (lcs - lct);
	    }
	}
    }
#endif
    return 0;
}

/*
 *----------------------------------------------------------------------
 *
 * Tcl_UniCharIsAlnum --

			    if ((string + 1 < stringEnd) &&
				    ((string[1] & 0xFC00) == 0xDC00)) {
				q = (((q&0x3FF)<<10) | (string[1]&0x3FF))
					+ 0x10000;
			    }
			}
			if ((p == q) || (nocase &&
				(p == TclUCS4ToLower(q)))) {
			    break;
			}
			if (q > 0xFFFF) {
			    string++;
			}
			string++;
		    }

		if ((string + 1 < stringEnd) &&
			((string[1] & 0xFC00) == 0xDC00)) {
		    ch1 = (((ch1&0x3FF)<<10) | (string[1]&0x3FF)) + 0x10000;
		    string++;
		}
	    }
	    if (nocase) {
		ch1 = TclUCS4ToLower(ch1);
	    }
#else
	    ch1 = (nocase ? Tcl_UniCharToLower(*string) : *string);
#endif
	    string++;
	    while (1) {
		if ((*pattern == ']') || (pattern == patternEnd)) {
		    return 0;
		}
#if TCL_UTF_MAX == 3
		startChar = *pattern;
		if ((startChar & 0xFC00) == 0xD800) {
		    if ((pattern + 1 < patternEnd) &&
			    ((pattern[1] & 0xFC00) == 0xDC00)) {
			startChar = (((startChar&0x3FF)<<10) |
					(pattern[1]&0x3FF)) + 0x10000;
			pattern++;
		    }
		}
	 	if (nocase) {
		    startChar = TclUCS4ToLower(startChar);
		}
#else
		startChar = (nocase ? Tcl_UniCharToLower(*pattern) : *pattern);
#endif
		pattern++;
		if (*pattern == '-') {
		    pattern++;

				((pattern[1] & 0xFC00) == 0xDC00)) {
			    endChar = (((endChar&0x3FF)<<10) |
					    (pattern[1]&0x3FF)) + 0x10000;
			    pattern++;
			}
		    }
		    if (nocase) {
			endChar = TclUCS4ToLower(endChar);
		    }
#else
		    endChar = (nocase ? Tcl_UniCharToLower(*pattern)
			    : *pattern);
#endif
		    pattern++;
		    if (((startChar <= ch1) && (ch1 <= endChar))

	    if ((string + 1 < stringEnd) &&
		    ((string[1] & 0xFC00) == 0xDC00)) {
		q = (((q&0x3FF)<<10) | (string[1]&0x3FF)) + 0x10000;
		string++;
	    }
	}
	if (nocase) {
	    if (TclUCS4ToLower(q) != TclUCS4ToLower(p)) {
		return 0;
	    }
	} else if (q != p) {
	    return 0;
	}
#else
	if (nocase) {

	    if (UCHAR(*pattern) < 0x80) {
		ch2 = (int)
			(nocase ? tolower(UCHAR(*pattern)) : UCHAR(*pattern));
	    } else {
		UtfToUniChar(pattern, &ch2);
		if (nocase) {
		    ch2 = TclUCS4ToLower(ch2);
		}
	    }

	    while (1) {
		/*
		 * Optimization for matching - cruise through the string
		 * quickly if the next char in the pattern isn't a special
		 * character
		 */

		if ((p != '[') && (p != '?') && (p != '\\')) {
		    if (nocase) {
			while (*str) {
			    charLen = UtfToUniChar(str, &ch1);
			    if (ch2==ch1 || ch2==TclUCS4ToLower(ch1)) {
				break;
			    }
			    str += charLen;
			}
		    } else {
			/*
			 * There's no point in trying to make this code

	    if (UCHAR(*str) < 0x80) {
		ch1 = (int)
			(nocase ? tolower(UCHAR(*str)) : UCHAR(*str));
		str++;
	    } else {
		str += UtfToUniChar(str, &ch1);
		if (nocase) {
		    ch1 = TclUCS4ToLower(ch1);
		}
	    }
	    while (1) {
		if ((*pattern == ']') || (*pattern == '\0')) {
		    return 0;
		}
		if (UCHAR(*pattern) < 0x80) {
		    startChar = (int) (nocase
			    ? tolower(UCHAR(*pattern)) : UCHAR(*pattern));
		    pattern++;
		} else {
		    pattern += UtfToUniChar(pattern, &startChar);
		    if (nocase) {
			startChar = TclUCS4ToLower(startChar);
		    }
		}
		if (*pattern == '-') {
		    pattern++;
		    if (*pattern == '\0') {
			return 0;
		    }
		    if (UCHAR(*pattern) < 0x80) {
			endChar = (int) (nocase
				? tolower(UCHAR(*pattern)) : UCHAR(*pattern));
			pattern++;
		    } else {
			pattern += UtfToUniChar(pattern, &endChar);
			if (nocase) {
			    endChar = TclUCS4ToLower(endChar);
			}
		    }
		    if (((startChar <= ch1) && (ch1 <= endChar))
			    || ((endChar <= ch1) && (ch1 <= startChar))) {
			/*
			 * Matches ranges of form [a-z] or [z-a].
			 */

	 * There's no special character. Just make sure that the next bytes of
	 * each string match.
	 */

	str += UtfToUniChar(str, &ch1);
	pattern += UtfToUniChar(pattern, &ch2);
	if (nocase) {
	    if (TclUCS4ToLower(ch1) != TclUCS4ToLower(ch2)) {
		return 0;
	    }
	} else if (ch1 != ch2) {
	    return 0;
	}
    }
}

    test_lsort 0
} -result 0 -cleanup {
    rename test_lsort ""
}
test cmdIL-5.6 {lsort with multiple list-style index options} {
    lsort -index {1 2 3} -index 0 {{a b} {c d} {b e}}
} {{a b} {b e} {c d}}
test cmdIL-5.7 {lsort unicode beyond U+FFFF} {
    lsort {\uD83D\uDE03 \uD83D\uDE02 \uD83D\uDE04}
} "\uD83D\uDE02 \uD83D\uDE03 \uD83D\udE04"
test cmdIL-5.7 {lsort unicode beyond U+FFFF} {
    lsort -decreasing {\uD83D\uDE03 \uD83D\uDE02 \uD83D\uDE04}
} "\uD83D\uDE04 \uD83D\uDE03 \uD83D\udE02"
test cmdIL-5.8 {lsort unicode beyond U+FFFF} {
    lsort -nocase {\U0001F603 \U0001F602 \U0001F604}
} "\U0001F602 \U0001F603 \U0001F604"
test cmdIL-5.9 {lsort unicode beyond U+FFFF} {
    lsort -dictionary {\U0001F603x1 \U0001F602y1 \U0001F602y \U0001F603xx}
} "\U0001F602y \U0001F602y1 \U0001F603x1 \U0001F603xx"
test cmdIL-5.9 {lsort unicode beyond U+FFFF} {
    lsort -dictionary {b\U0001F60320 c\U0001F60230 c\U0001F6023x b\U0001F6032}
} "b\U0001F6032 b\U0001F60320 c\U0001F6023x c\U0001F60230"
test cmdIL-5.10 {lsort unicode beyond U+FFFF} {
    lsort -nocase {b\U00010428a B\U00010400C}
} "b\U00010428a B\U00010400C"
test cmdIL-5.11 {lsort unicode beyond U+FFFF} {
    lsort -dictionary -nocase {b\U00010428a B\U00010400C}
} "b\U00010428a B\U00010400C"

# Compiled version
test cmdIL-6.1 {lassign command syntax} -returnCodes error -body {
    apply {{} { lassign }}
} -result {wrong # args: should be "lassign list ?varName ...?"}
test cmdIL-6.2 {lassign command syntax} {
    apply {{} { lassign x }}

::tcltest::loadTestedCommands
catch [list package require -exact Tcltest [info patchlevel]]

# Some tests require the testobj command

testConstraint testobj [expr {[info commands testobj] != {}}]
testConstraint testindexobj [expr {[info commands testindexobj] != {}}]
testConstraint testbytestring [expr {[info commands testbytestring] != {}}]
testConstraint wtf8 [expr {[string length \U00010000] != 1}]

# Used for constraining memory leak tests
testConstraint memory [llength [info commands memory]]

test string-1.1 {error conditions} {
    list [catch {string gorp a b} msg] $msg
} {1 {unknown or ambiguous subcommand "gorp": must be bytelength, cat, compare, equal, first, index, is, last, length, map, match, range, repeat, replace, reverse, tolower, totitle, toupper, trim, trimleft, trimright, wordend, or wordstart}}

} {000341 000341 0341}
test string-12.22 {string range, shimmering binary/index} {
    set s 0000000001
    binary scan $s a* x
    string range $s $s end
} 000000001
test string-12.23 {string range, surrogates, bug [11ae2be95dac9417]} wtf8 {
    list [string range a\U00100000b 1 1] \
	[string range a\U00100000b 2 2] \
	[string range a\U00100000b 3 3]
} [list \U00100000 {} b]
test string-12.23 {string range, surrogates, bug [11ae2be95dac9417]} !wtf8 {
    list [string range a\U00100000b 1 1] \
	[string range a\U00100000b 2 2] \
	[string range a\U00100000b 3 3]
} [list \U00100000 b {}]

test string-13.1 {string repeat} {
    list [catch {string repeat} msg] $msg
} {1 {wrong # args: should be "string repeat string count"}}
test string-13.2 {string repeat} {
    list [catch {string repeat abc 10 oops} msg] $msg
} {1 {wrong # args: should be "string repeat string count"}}

    } 0 {}
    {binary neq} {
	string compare [binary format a100a 0 1] [binary format a100a 0 0]
    } 1 {}
    {binary neq inequal length} {
	string compare [binary format a20a 0 1] [binary format a100a 0 0]
    } 1 {}
    {unicode corner cases} {
	# high surrogate at end is like empty string for TCL_UTF_MAX==3
	set ret [string compare \uD7FF \uD800]
	if {[string length \U00010000] > 1} {
	    set ret [expr {- $ret}]
	}
	set ret
    } -1 {}
    {unicode corner cases} {
	string compare \uDBFF \uDC00
    } -1 {}
    {unicode corner cases} {
	string compare \uD83D \uDE00
    } -1 {}
    {unicode corner cases} {
	string compare \uE000 \uDFFF
    } 1 {}
    {unicode beyond U+FFFF} {
	string compare \uFFFF \U00010000
    } -1 {}
    {unicode beyond U+FFFF} {
	string compare \U00010000 \U0000FFFF
    } 1 {}
    {unicode beyond U+FFFF} {
	string compare ab \U0001F600\U0001F601\U0001F602\U0001F603
    } -1 {}
    {unicode beyond U+FFFF} {
	string compare \U0001F600\U0001F601\U0001F602\U0001F603 ab
    } 1 {}
    {unicode beyond U+FFFF} {
	string compare \U0001F601\U0001F602 \U0001F600\U0001F601
    } 1 {}
    {unicode beyond U+FFFF} {
	string compare \uD83D\uDE00\uD83D\uDE01 \U0001F600\U0001F601
    } 0 {}
    {unicode beyond U+FFFF} {
	string compare \uD83D\uDE00 \uD83D\uDE01\U0001F600\U0001F601
    } -1 {}
} {
    if {$tname eq ""} { continue }
    if {$tcode eq ""} { set tcode ok }
    test stringComp-2.[incr i] "string compare, $tname" \
	-body [list eval $tbody] \
	-returnCodes $tcode -result $tresult
    test stringComp-2.[incr i] "string compare bc, $tname" \

    incr count
}
variable count 1
UniCharCaseCmpTest < a b
UniCharCaseCmpTest > b a
UniCharCaseCmpTest > B a
UniCharCaseCmpTest > aBcB abca
UniCharCaseCmpTest < \uFFFF [format %c 0x10000]
UniCharCaseCmpTest < \uFFFF \U10000
UniCharCaseCmpTest > [format %c 0x10000] \uFFFF
UniCharCaseCmpTest > \U10000 \uFFFF





unset count
rename UniCharCaseCmpTest {}

	dcb.XonChar = argv[0][0];
	dcb.XoffChar = argv[1][0];
	if (argv[0][0] & 0x80 || argv[1][0] & 0x80) {
	    Tcl_UniChar character;
	    int charLen;

	    charLen = TclUtfToUniChar(argv[0], &character);
	    if ((character & ~0xFF) || argv[0][charLen]) {
		goto badXchar;
	    }
	    dcb.XonChar = (char) character;
	    charLen = TclUtfToUniChar(argv[1], &character);
	    if ((character & ~0xFF) || argv[1][charLen]) {
		goto badXchar;
	    }
	    dcb.XoffChar = (char) character;
	}
	ckfree(argv);