Tcl Source Code

StringRplcCmd(
    ClientData dummy,		/* Not used. */
    Tcl_Interp *interp,		/* Current interpreter. */
    int objc,			/* Number of arguments. */
    Tcl_Obj *const objv[])	/* Argument objects. */
{
    Tcl_UniChar *ustring;
    int first, last, length, end;

    if (objc < 4 || objc > 5) {
	Tcl_WrongNumArgs(interp, 1, objv, "string first last ?string?");
	return TCL_ERROR;
    }

    ustring = Tcl_GetUnicodeFromObj(objv[1], &length);
    end = length - 1;

    if (TclGetIntForIndexM(interp, objv[2], end, &first) != TCL_OK ||
	    TclGetIntForIndexM(interp, objv[3], end, &last) != TCL_OK){
	return TCL_ERROR;
    }

    /*
     * The following test screens out most empty substrings as
     * candidates for replacement. When they are detected, no
     * replacement is done, and the result is the original string,
     */
    if ((last < 0) ||		/* Range ends before start of string */
	    (first > end) ||	/* Range begins after end of string */
	    (last < first)) {	/* Range begins after it starts */

	/*
	 * BUT!!! when (end < 0) -- an empty original string -- we can
	 * have (first <= end < 0 <= last) and an empty string is permitted
	 * to be replaced.
	 */
	Tcl_SetObjResult(interp, objv[1]);
    } else {
	Tcl_Obj *resultPtr;

	/*
	 * We are re-fetching in case the string argument is same value as 
	 * an index argument, and shimmering cost us our ustring.
	 */

	ustring = Tcl_GetUnicodeFromObj(objv[1], &length);
	end = length-1;

	if (first < 0) {
	    first = 0;
	}

	resultPtr = Tcl_NewUnicodeObj(ustring, first);
	if (objc == 5) {
	    Tcl_AppendObjToObj(resultPtr, objv[4]);
	}
	if (last < end) {
	    Tcl_AppendUnicodeToObj(resultPtr, ustring + last + 1,
		    end - last);
	}
	Tcl_SetObjResult(interp, resultPtr);
    }
    return TCL_OK;
}

/*

	length2 = strlen(tclDefaultTrimSet);
    } else {
	Tcl_WrongNumArgs(interp, 1, objv, "string ?chars?");
	return TCL_ERROR;
    }
    string1 = TclGetStringFromObj(objv[1], &length1);

    triml = TclTrim(string1, length1, string2, length2, &trimr);


    Tcl_SetObjResult(interp,
	    Tcl_NewStringObj(string1 + triml, length1 - triml - trimr));
    return TCL_OK;
}

/*

    Tcl_Interp *interp,		/* Tcl interpreter for context. */
    Tcl_Parse *parsePtr,	/* Points to a parse structure for the
				 * command. */
    Command *cmdPtr,		/* Points to defintion of command being
				 * compiled. */
    CompileEnv *envPtr)		/* Holds the resulting instructions. */
{
    Tcl_Token *tokenPtr, *valueTokenPtr;
    DefineLineInformation;	/* TIP #280 */
    int first, last;

    if (parsePtr->numWords < 4 || parsePtr->numWords > 5) {
	return TCL_ERROR;
    }
 
    /* Bytecode to compute/push string argument being replaced */
    valueTokenPtr = TokenAfter(parsePtr->tokenPtr);

    CompileWord(envPtr, valueTokenPtr, interp, 1);



    /*
     * Check for first index known and useful at compile time. 
     */
    tokenPtr = TokenAfter(valueTokenPtr);
    if (TclGetIndexFromToken(tokenPtr, TCL_INDEX_BEFORE, TCL_INDEX_AFTER,
	    &first) != TCL_OK) {
	goto genericReplace;
    }

    /*

     * Check for last index known and useful at compile time. 
     */

    tokenPtr = TokenAfter(tokenPtr);
    if (TclGetIndexFromToken(tokenPtr, TCL_INDEX_BEFORE, TCL_INDEX_AFTER,
	    &last) != TCL_OK) {
	goto genericReplace;
    }

    /* 
     * [string replace] is an odd bird.  For many arguments it is
     * a conventional substring replacer.  However it also goes out
     * of its way to become a no-op for many cases where it would be
     * replacing an empty substring.  Precisely, it is a no-op when
     *
     *		(last < first)		OR
     *		(last < 0)		OR
     *		(end < first)
     *
     * For some compile-time values we can detect these cases, and
     * compile direct to bytecode implementing the no-op.
     */

    if ((last == TCL_INDEX_BEFORE)		/* Know (last < 0) */
	    || (first == TCL_INDEX_AFTER)	/* Know (first > end) */

	/*
	 * Tricky to determine when runtime (last < first) can be
	 * certainly known based on the encoded values. Consider the
	 * cases...
	 *
	 * (first <= TCL_INDEX_END) &&
	 *	(last == TCL_INDEX_AFTER) => cannot tell REJECT
	 *	(last <= TCL_INDEX END) && (last < first) => ACCEPT
	 *	else => cannot tell REJECT
	 */
	    || ((first <= TCL_INDEX_END) && (last <= TCL_INDEX_END)
		&& (last < first))		/* Know (last < first) */
	/*
	 * (first == TCL_INDEX_BEFORE) &&
	 *	(last == TCL_INDEX_AFTER) => (first < last) REJECT
	 *	(last <= TCL_INDEX_END) => cannot tell REJECT
	 *	else		=> (first < last) REJECT
	 *







	 * else [[first >= TCL_INDEX_START]] &&
	 *	(last == TCL_INDEX_AFTER) => cannot tell REJECT
	 *	(last <= TCL_INDEX_END) => cannot tell REJECT
	 *	else [[last >= TCL_INDEX START]] && (last < first) => ACCEPT
	 */
	    || ((first >= TCL_INDEX_START) && (last >= TCL_INDEX_START)


		&& (last < first))) {		/* Know (last < first) */
	if (parsePtr->numWords == 5) {
	    tokenPtr = TokenAfter(tokenPtr);
	    CompileWord(envPtr, tokenPtr, interp, 4);
	    OP(		POP);		/* Pop newString */


	}

	/* Original string argument now on TOS as result */


	return TCL_OK;
    }



    if (parsePtr->numWords == 5) {
    /*

     * When we have a string replacement, we have to take care about
     * not replacing empty substrings that [string replace] promises
     * not to replace

     *
     * The remaining index values might be suitable for conventional
     * string replacement, but only if they cannot possibly meet the
     * conditions described above at runtime. If there's a chance they
     * might, we would have to emit bytecode to check and at that point
     * we're paying more in bytecode execution time than would make
     * things worthwhile. Trouble is we are very limited in
     * how much we can detect that at compile time. After decoding,
     * we need, first:
     *





     *		(first <= end)



     *
     * The encoded indices (first <= TCL_INDEX END) and
     * (first == TCL_INDEX_BEFORE) always meets this condition, but

     * any other encoded first index has some list for which it fails.
     *
     * We also need, second:

     *
     *		(last >= 0)
     *
     * The encoded indices (last >= TCL_INDEX_START) and
     * (last == TCL_INDEX_AFTER) always meet this condition but any
     * other encoded last index has some list for which it fails.
     *
     * Finally we need, third:
     *

     *		(first <= last)

     * 
     * Considered in combination with the constraints we already have,
     * we see that we can proceed when (first == TCL_INDEX_BEFORE)
     * or (last == TCL_INDEX_AFTER). These also permit simplification
     * of the prefix|replace|suffix construction. The other constraints,
     * though, interfere with getting a guarantee that first <= last. 
     */

    if ((first == TCL_INDEX_BEFORE) && (last >= TCL_INDEX_START)) {
	/* empty prefix */
	tokenPtr = TokenAfter(tokenPtr);
	CompileWord(envPtr, tokenPtr, interp, 4);
	OP4(		REVERSE, 2);
	if (last == TCL_INDEX_AFTER) {
	    OP(		POP);		/* Pop  original */
	} else {
	    OP44(	STR_RANGE_IMM, last + 1, TCL_INDEX_END);
	    OP1(	STR_CONCAT1, 2);
	}
	return TCL_OK;
    }




    if ((last == TCL_INDEX_AFTER) && (first <= TCL_INDEX_END)) {









	OP44(		STR_RANGE_IMM, 0, first-1);
	tokenPtr = TokenAfter(tokenPtr);
	CompileWord(envPtr, tokenPtr, interp, 4);
	OP1(		STR_CONCAT1, 2);

	return TCL_OK;
    }

	/* FLOW THROUGH TO genericReplace */

    } else {
	/* 


	 * When we have no replacement string to worry about, we may

	 * have more luck, because the forbidden empty string replacements
	 * are harmless when they are replaced by another empty string.
	 */

	if ((first == TCL_INDEX_BEFORE) || (first == TCL_INDEX_START)) {
	    /* empty prefix - build suffix only */

	    if ((last == TCL_INDEX_END) || (last == TCL_INDEX_AFTER)) {
		/* empty suffix too => empty result */
		OP(	POP);		/* Pop  original */
		PUSH	(	"");
		return TCL_OK;
	    }
	    OP44(	STR_RANGE_IMM, last + 1, TCL_INDEX_END);
	    return TCL_OK;
	} else {
	    if ((last == TCL_INDEX_END) || (last == TCL_INDEX_AFTER)) {
		/* empty suffix - build prefix only */
		OP44(	STR_RANGE_IMM, 0, first-1);
		return TCL_OK;
	    }
	    OP(		DUP);
	    OP44(	STR_RANGE_IMM, 0, first-1);
	    OP4(	REVERSE, 2);
	    OP44(	STR_RANGE_IMM, last + 1, TCL_INDEX_END);
	    OP1(	STR_CONCAT1, 2);
	    return TCL_OK;
	}
    }

    genericReplace:

	tokenPtr = TokenAfter(valueTokenPtr);
	CompileWord(envPtr, tokenPtr, interp, 2);
	tokenPtr = TokenAfter(tokenPtr);
	CompileWord(envPtr, tokenPtr, interp, 3);
	if (parsePtr->numWords == 5) {
	    tokenPtr = TokenAfter(tokenPtr);
	    CompileWord(envPtr, tokenPtr, interp, 4);
	} else {
	    PUSH(	"");
	}
	OP(		STR_REPLACE);
	return TCL_OK;

}

int
TclCompileStringTrimLCmd(
    Tcl_Interp *interp,		/* Used for error reporting. */
    Tcl_Parse *parsePtr,	/* Points to a parse structure for the command
				 * created by Tcl_ParseCommand. */

	    TclNewObj(objResultPtr);
	}
	TRACE_APPEND(("%.30s\n", O2S(objResultPtr)));
	NEXT_INST_F(9, 1, 1);

    {
	Tcl_UniChar *ustring1, *ustring2, *ustring3, *end, *p;
	int length3, endIdx;
	Tcl_Obj *value3Ptr;

    case INST_STR_REPLACE:
	value3Ptr = POP_OBJECT();
	valuePtr = OBJ_AT_DEPTH(2);
	endIdx = Tcl_GetCharLength(valuePtr) - 1;
	TRACE(("\"%.20s\" %s %s \"%.20s\" => ", O2S(valuePtr),
		O2S(OBJ_UNDER_TOS), O2S(OBJ_AT_TOS), O2S(value3Ptr)));
	if (TclGetIntForIndexM(interp, OBJ_UNDER_TOS, endIdx,
		    &fromIdx) != TCL_OK
	    || TclGetIntForIndexM(interp, OBJ_AT_TOS, endIdx,
		    &toIdx) != TCL_OK) {
	    TclDecrRefCount(value3Ptr);
	    TRACE_ERROR(interp);
	    goto gotError;
	}
	TclDecrRefCount(OBJ_AT_TOS);
	(void) POP_OBJECT();
	TclDecrRefCount(OBJ_AT_TOS);
	(void) POP_OBJECT();



	if ((toIdx < 0) ||
		(fromIdx > endIdx) ||
		(toIdx < fromIdx)) {
	    TRACE_APPEND(("\"%.30s\"\n", O2S(valuePtr)));
	    TclDecrRefCount(value3Ptr);
	    NEXT_INST_F(1, 0, 0);
	}

	if (fromIdx < 0) {
	    fromIdx = 0;
	}

	if (toIdx > endIdx) {
	    toIdx = endIdx;
	}

	if (fromIdx == 0 && toIdx == endIdx) {
	    TclDecrRefCount(OBJ_AT_TOS);
	    OBJ_AT_TOS = value3Ptr;
	    TRACE_APPEND(("\"%.30s\"\n", O2S(value3Ptr)));
	    NEXT_INST_F(1, 0, 0);
	}

	length3 = Tcl_GetCharLength(value3Ptr);

	trim1 = 0;
	goto createTrimmedString;
    case INST_STR_TRIM:
	valuePtr = OBJ_UNDER_TOS;	/* String */
	value2Ptr = OBJ_AT_TOS;		/* TrimSet */
	string2 = TclGetStringFromObj(value2Ptr, &length2);
	string1 = TclGetStringFromObj(valuePtr, &length);
	trim1 = TclTrim(string1, length, string2, length2, &trim2);





    createTrimmedString:
	/*
	 * Careful here; trim set often contains non-ASCII characters so we
	 * take care when printing. [Bug 971cb4f1db]
	 */

#ifdef TCL_COMPILE_DEBUG

			    Tcl_Obj *const opts[], int *flagPtr);
MODULE_SCOPE void	TclSubstParse(Tcl_Interp *interp, const char *bytes,
			    int numBytes, int flags, Tcl_Parse *parsePtr,
			    Tcl_InterpState *statePtr);
MODULE_SCOPE int	TclSubstTokens(Tcl_Interp *interp, Tcl_Token *tokenPtr,
			    int count, int *tokensLeftPtr, int line,
			    int *clNextOuter, const char *outerScript);
MODULE_SCOPE int	TclTrim(const char *bytes, int numBytes,
			    const char *trim, int numTrim, int *trimRight);
MODULE_SCOPE int	TclTrimLeft(const char *bytes, int numBytes,
			    const char *trim, int numTrim);
MODULE_SCOPE int	TclTrimRight(const char *bytes, int numBytes,
			    const char *trim, int numTrim);
MODULE_SCOPE int	TclUtfCasecmp(const char *cs, const char *ct);
MODULE_SCOPE Tcl_Obj *	TclpNativeToNormalized(ClientData clientData);
MODULE_SCOPE Tcl_Obj *	TclpFilesystemPathType(Tcl_Obj *pathPtr);

    TclUtfToUniChar(buf, &ch);
    return (char) ch;
}

/*
 *----------------------------------------------------------------------
 *
 * UtfWellFormedEnd --
 *	Checks the end of utf string is malformed, if yes - wraps bytes
 *	to the given buffer (as well-formed NTS string).  The buffer
 *	argument should be initialized by the caller and ready to use.
 *
 * Results:
 *	The bytes with well-formed end of the string.
 *
 * Side effects:
 *	Buffer (DString) may be allocated, so must be released.
 *
 *----------------------------------------------------------------------
 */

static inline const char*
UtfWellFormedEnd(
    Tcl_DString *buffer,	/* Buffer used to hold well-formed string. */
    const char *bytes,		/* Pointer to the beginning of the string. */
    int length)			/* Length of the string. */
{
    const char *l = bytes + length;
    const char *p = Tcl_UtfPrev(l, bytes);

    if (Tcl_UtfCharComplete(p, l - p)) {
	return bytes;
    }
    /* 
     * Malformed utf-8 end, be sure we've NTS to safe compare of end-character,
     * avoid segfault by access violation out of range.
     */
    Tcl_DStringAppend(buffer, bytes, length);
    return Tcl_DStringValue(buffer);
}
/*
 *----------------------------------------------------------------------
 *
 * TclTrimRight --
 *	Takes two counted strings in the Tcl encoding.  Conceptually

 *	finds the sub string (offset) to trim from the right side of the
 *	first string all characters found in the second string.
 *
 * Results:
 *	The number of bytes to be removed from the end of the string.
 *
 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

static inline int
TrimRight(
    const char *bytes,		/* String to be trimmed... */
    int numBytes,		/* ...and its length in bytes */
    const char *trim,		/* String of trim characters... */
    int numTrim)		/* ...and its length in bytes */
{
    const char *p = bytes + numBytes;
    int pInc;













    /*
     * Outer loop: iterate over string to be trimmed.
     */

    do {
	Tcl_UniChar ch1;
	const char *q = trim;

	    p += pInc;
	    break;
	}
    } while (p > bytes);

    return numBytes - (p - bytes);
}

int
TclTrimRight(
    const char *bytes,	/* String to be trimmed... */
    int numBytes,	/* ...and its length in bytes */
    const char *trim,	/* String of trim characters... */
    int numTrim)	/* ...and its length in bytes */
{
    int res;
    Tcl_DString bytesBuf, trimBuf;

    /* Empty strings -> nothing to do */
    if ((numBytes == 0) || (numTrim == 0)) {
	return 0;
    }

    Tcl_DStringInit(&bytesBuf);
    Tcl_DStringInit(&trimBuf);
    bytes = UtfWellFormedEnd(&bytesBuf, bytes, numBytes);
    trim = UtfWellFormedEnd(&trimBuf, trim, numTrim);

    res = TrimRight(bytes, numBytes, trim, numTrim);
    if (res > numBytes) {
	res = numBytes;
    }

    Tcl_DStringFree(&bytesBuf);
    Tcl_DStringFree(&trimBuf);

    return res;
}

/*
 *----------------------------------------------------------------------
 *
 * TclTrimLeft --
 *
 *	Takes two counted strings in the Tcl encoding.  Conceptually
 *	finds the sub string (offset) to trim from the left side of the
 *	first string all characters found in the second string.
 *
 * Results:
 *	The number of bytes to be removed from the start of the string.
 *
 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

static inline int
TrimLeft(
    const char *bytes,		/* String to be trimmed... */
    int numBytes,		/* ...and its length in bytes */
    const char *trim,		/* String of trim characters... */
    int numTrim)		/* ...and its length in bytes */
{
    const char *p = bytes;













    /*
     * Outer loop: iterate over string to be trimmed.
     */

    do {
	Tcl_UniChar ch1;
	int pInc = TclUtfToUniChar(p, &ch1);

	     */

	    break;
	}

	p += pInc;
	numBytes -= pInc;
    } while (numBytes > 0);

    return p - bytes;
}

int
TclTrimLeft(
    const char *bytes,	/* String to be trimmed... */
    int numBytes,	/* ...and its length in bytes */
    const char *trim,	/* String of trim characters... */
    int numTrim)	/* ...and its length in bytes */
{
    int res;
    Tcl_DString bytesBuf, trimBuf;

    /* Empty strings -> nothing to do */
    if ((numBytes == 0) || (numTrim == 0)) {
	return 0;
    }

    Tcl_DStringInit(&bytesBuf);
    Tcl_DStringInit(&trimBuf);
    bytes = UtfWellFormedEnd(&bytesBuf, bytes, numBytes);
    trim = UtfWellFormedEnd(&trimBuf, trim, numTrim);

    res = TrimLeft(bytes, numBytes, trim, numTrim);
    if (res > numBytes) {
	res = numBytes;
    }

    Tcl_DStringFree(&bytesBuf);
    Tcl_DStringFree(&trimBuf);

    return res;
}

/*
 *----------------------------------------------------------------------
 *
 * TclTrim --
 *	Finds the sub string (offset) to trim from both sides of the
 *	first string all characters found in the second string.
 *
 * Results:
 *	The number of bytes to be removed from the start of the string
 *
 * Side effects:
 *	None.
 *
 *----------------------------------------------------------------------
 */

int
TclTrim(
    const char *bytes,	/* String to be trimmed... */
    int numBytes,	/* ...and its length in bytes */
    const char *trim,	/* String of trim characters... */
    int numTrim,	/* ...and its length in bytes */
    int *trimRight)		/* Offset from the end of the string. */
{
    int trimLeft;
    Tcl_DString bytesBuf, trimBuf;

    *trimRight = 0;
    /* Empty strings -> nothing to do */
    if ((numBytes == 0) || (numTrim == 0)) {
	return 0;
    }

    Tcl_DStringInit(&bytesBuf);
    Tcl_DStringInit(&trimBuf);
    bytes = UtfWellFormedEnd(&bytesBuf, bytes, numBytes);
    trim = UtfWellFormedEnd(&trimBuf, trim, numTrim);

    trimLeft = TrimLeft(bytes, numBytes, trim, numTrim);
    if (trimLeft > numBytes) {
	trimLeft = numBytes;
    }
    numBytes -= trimLeft;
    /* have to trim yet (first char was already verified within TrimLeft) */
    if (numBytes > 1) {
	bytes += trimLeft;
	*trimRight = TrimRight(bytes, numBytes, trim, numTrim);
	if (*trimRight > numBytes) {
	    *trimRight = numBytes;
	}
    }

    Tcl_DStringFree(&bytesBuf);
    Tcl_DStringFree(&trimBuf);

    return trimLeft;
}

/*
 *----------------------------------------------------------------------
 *
 * Tcl_Concat --
 *
 *	Concatenate a set of strings into a single large string.

    /*
     * All element bytes + (argc - 1) spaces + 1 terminating NULL.
     */

    result = ckalloc((unsigned) (bytesNeeded + argc));

    for (p = result, i = 0;  i < argc;  i++) {
	int triml, trimr, elemLength;
	const char *element;

	element = argv[i];
	elemLength = strlen(argv[i]);


	/* Trim away the leading/trailing whitespace. */


	triml = TclTrim(element, elemLength, CONCAT_TRIM_SET,
		CONCAT_WS_SIZE, &trimr);
	element += triml;
	elemLength -= triml + trimr;



	/* Do not permit trimming to expose a final backslash character. */




	elemLength += trimr && (element[elemLength - 1] == '\\');


	/*
	 * If we're left with empty element after trimming, do nothing.
	 */

	if (elemLength == 0) {
	    continue;

     */

    TclNewObj(resPtr);
    (void) Tcl_AttemptSetObjLength(resPtr, bytesNeeded + objc - 1);
    Tcl_SetObjLength(resPtr, 0);

    for (i = 0;  i < objc;  i++) {
	int triml, trimr;

	element = TclGetStringFromObj(objv[i], &elemLength);


	/* Trim away the leading/trailing whitespace. */


	triml = TclTrim(element, elemLength, CONCAT_TRIM_SET,
		CONCAT_WS_SIZE, &trimr);
	element += triml;
	elemLength -= triml + trimr;



	/* Do not permit trimming to expose a final backslash character. */




	elemLength += trimr && (element[elemLength - 1] == '\\');


	/*
	 * If we're left with empty element after trimming, do nothing.
	 */

	if (elemLength == 0) {
	    continue;

} {abcdeNEWop}
test string-14.16 {string replace} {
    string replace abcdefghijklmnop 0 end foo
} {foo}
test string-14.17 {string replace} {
    string replace abcdefghijklmnop end end-1
} {abcdefghijklmnop}
test string-14.18 {string replace} {
    string replace abcdefghijklmnop 10 9 XXX
} {abcdefghijklmnop}
test string-14.19 {string replace} {
    string replace {} -1 0 A
} A

test string-15.1 {string tolower too few args} {
    list [catch {string tolower} msg] $msg
} {1 {wrong # args: should be "string tolower string ?first? ?last?"}}
test string-15.2 {string tolower bad args} {
    list [catch {string tolower a b} msg] $msg
} {1 {bad index "b": must be integer?[+-]integer? or end?[+-]integer?}}