tclstringobj.c

来自「tcl是工具命令语言」· C语言 代码 · 共 1,878 行 · 第 1/4 页

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

void
Tcl_SetUnicodeObj(objPtr, unicode, numChars)
    Tcl_Obj *objPtr;		/* The object to set the string of. */
    CONST Tcl_UniChar *unicode;	/* The unicode string used to initialize
				 * the object. */
    int numChars;		/* Number of characters in the unicode
				 * string. */
{
    Tcl_ObjType *typePtr;
    String *stringPtr;
    size_t uallocated;

    if (numChars < 0) {
	numChars = 0;
	if (unicode) {
	    while (unicode[numChars] != 0) { numChars++; }
	}
    }
    uallocated = STRING_UALLOC(numChars);

    /*
     * Free the internal rep if one exists, and invalidate the string rep.
     */

    typePtr = objPtr->typePtr;
    if ((typePtr != NULL) && (typePtr->freeIntRepProc) != NULL) {
	(*typePtr->freeIntRepProc)(objPtr);
    }
    objPtr->typePtr = &tclStringType;

    /*
     * Allocate enough space for the String structure + Unicode string.
     */
	
    stringPtr = (String *) ckalloc(STRING_SIZE(uallocated));
    stringPtr->numChars = numChars;
    stringPtr->uallocated = uallocated;
    stringPtr->hasUnicode = (numChars > 0);
    stringPtr->allocated = 0;
    memcpy((VOID *) stringPtr->unicode, (VOID *) unicode, uallocated);
    stringPtr->unicode[numChars] = 0;
    SET_STRING(objPtr, stringPtr);
    Tcl_InvalidateStringRep(objPtr);
    return;
}

/*
 *----------------------------------------------------------------------
 *
 * Tcl_AppendToObj --
 *
 *	This procedure appends a sequence of bytes to an object.
 *
 * Results:
 *	None.
 *
 * Side effects:
 *	The bytes at *bytes are appended to the string representation
 *	of objPtr.
 *
 *----------------------------------------------------------------------
 */

void
Tcl_AppendToObj(objPtr, bytes, length)
    register Tcl_Obj *objPtr;	/* Points to the object to append to. */
    CONST char *bytes;		/* Points to the bytes to append to the
				 * object. */
    register int length;	/* The number of bytes to append from
				 * "bytes". If < 0, then append all bytes
				 * up to NULL byte. */
{
    String *stringPtr;

    if (Tcl_IsShared(objPtr)) {
	panic("Tcl_AppendToObj called with shared object");
    }
    
    SetStringFromAny(NULL, objPtr);

    if (length < 0) {
	length = (bytes ? strlen(bytes) : 0);
    }
    if (length == 0) {
	return;
    }

    /*
     * If objPtr has a valid Unicode rep, then append the Unicode
     * conversion of "bytes" to the objPtr's Unicode rep, otherwise
     * append "bytes" to objPtr's string rep.
     */

    stringPtr = GET_STRING(objPtr);
    if (stringPtr->hasUnicode != 0) {
	AppendUtfToUnicodeRep(objPtr, bytes, length);

	stringPtr = GET_STRING(objPtr);
    } else {
	AppendUtfToUtfRep(objPtr, bytes, length);
    }
}

/*
 *----------------------------------------------------------------------
 *
 * Tcl_AppendUnicodeToObj --
 *
 *	This procedure appends a Unicode string to an object in the
 *	most efficient manner possible.  Length must be >= 0.
 *
 * Results:
 *	None.
 *
 * Side effects:
 *	Invalidates the string rep and creates a new Unicode string.
 *
 *----------------------------------------------------------------------
 */

void
Tcl_AppendUnicodeToObj(objPtr, unicode, length)
    register Tcl_Obj *objPtr;	/* Points to the object to append to. */
    CONST Tcl_UniChar *unicode;	/* The unicode string to append to the
			         * object. */
    int length;			/* Number of chars in "unicode". */
{
    String *stringPtr;

    if (Tcl_IsShared(objPtr)) {
	panic("Tcl_AppendUnicodeToObj called with shared object");
    }

    if (length == 0) {
	return;
    }

    SetStringFromAny(NULL, objPtr);
    stringPtr = GET_STRING(objPtr);

    /*
     * If objPtr has a valid Unicode rep, then append the "unicode"
     * to the objPtr's Unicode rep, otherwise the UTF conversion of
     * "unicode" to objPtr's string rep.
     */

    if (stringPtr->hasUnicode != 0) {
	AppendUnicodeToUnicodeRep(objPtr, unicode, length);
    } else {
	AppendUnicodeToUtfRep(objPtr, unicode, length);
    }
}

/*
 *----------------------------------------------------------------------
 *
 * Tcl_AppendObjToObj --
 *
 *	This procedure appends the string rep of one object to another.
 *	"objPtr" cannot be a shared object.
 *
 * Results:
 *	None.
 *
 * Side effects:
 *	The string rep of appendObjPtr is appended to the string 
 *	representation of objPtr.
 *
 *----------------------------------------------------------------------
 */

void
Tcl_AppendObjToObj(objPtr, appendObjPtr)
    Tcl_Obj *objPtr;		/* Points to the object to append to. */
    Tcl_Obj *appendObjPtr;	/* Object to append. */
{
    String *stringPtr;
    int length, numChars, allOneByteChars;
    char *bytes;

    SetStringFromAny(NULL, objPtr);

    /*
     * If objPtr has a valid Unicode rep, then get a Unicode string
     * from appendObjPtr and append it.
     */

    stringPtr = GET_STRING(objPtr);
    if (stringPtr->hasUnicode != 0) {
	
	/*
	 * If appendObjPtr is not of the "String" type, don't convert it.
	 */

	if (appendObjPtr->typePtr == &tclStringType) {
	    stringPtr = GET_STRING(appendObjPtr);
	    if ((stringPtr->numChars == -1)
		    || (stringPtr->hasUnicode == 0)) {
		
		/*
		 * If appendObjPtr is a string obj with no valid Unicode
		 * rep, then fill its unicode rep.
		 */

		FillUnicodeRep(appendObjPtr);
		stringPtr = GET_STRING(appendObjPtr);
	    }
	    AppendUnicodeToUnicodeRep(objPtr, stringPtr->unicode,
		    stringPtr->numChars);
	} else {
	    bytes = Tcl_GetStringFromObj(appendObjPtr, &length);
	    AppendUtfToUnicodeRep(objPtr, bytes, length);
	}
	return;
    }

    /*
     * Append to objPtr's UTF string rep.  If we know the number of
     * characters in both objects before appending, then set the combined
     * number of characters in the final (appended-to) object.
     */

    bytes = Tcl_GetStringFromObj(appendObjPtr, &length);

    allOneByteChars = 0;
    numChars = stringPtr->numChars;
    if ((numChars >= 0) && (appendObjPtr->typePtr == &tclStringType)) {
	stringPtr = GET_STRING(appendObjPtr);
	if ((stringPtr->numChars >= 0) && (stringPtr->numChars == length)) {
	    numChars += stringPtr->numChars;
	    allOneByteChars = 1;
	}
    }

    AppendUtfToUtfRep(objPtr, bytes, length);

    if (allOneByteChars) {
	stringPtr = GET_STRING(objPtr);
	stringPtr->numChars = numChars;
    }
}

/*
 *----------------------------------------------------------------------
 *
 * AppendUnicodeToUnicodeRep --
 *
 *	This procedure appends the contents of "unicode" to the Unicode
 *	rep of "objPtr".  objPtr must already have a valid Unicode rep.
 *
 * Results:
 *	None.
 *
 * Side effects:
 *	objPtr's internal rep is reallocated.
 *
 *----------------------------------------------------------------------
 */

static void
AppendUnicodeToUnicodeRep(objPtr, unicode, appendNumChars)
    Tcl_Obj *objPtr;	        /* Points to the object to append to. */
    CONST Tcl_UniChar *unicode; /* String to append. */
    int appendNumChars;	        /* Number of chars of "unicode" to append. */
{
    String *stringPtr, *tmpString;
    size_t numChars;

    if (appendNumChars < 0) {
	appendNumChars = 0;
	if (unicode) {
	    while (unicode[appendNumChars] != 0) { appendNumChars++; }
	}
    }
    if (appendNumChars == 0) {
	return;
    }

    SetStringFromAny(NULL, objPtr);
    stringPtr = GET_STRING(objPtr);

    /*
     * If not enough space has been allocated for the unicode rep,
     * reallocate the internal rep object with additional space.  First
     * try to double the required allocation; if that fails, try a more
     * modest increase.  See the "TCL STRING GROWTH ALGORITHM" comment at
     * the top of this file for an explanation of this growth algorithm.
     */

    numChars = stringPtr->numChars + appendNumChars;

    if (STRING_UALLOC(numChars) >= stringPtr->uallocated) {
     	stringPtr->uallocated = STRING_UALLOC(2 * numChars);
	tmpString = (String *) attemptckrealloc((char *)stringPtr,
		STRING_SIZE(stringPtr->uallocated));
	if (tmpString == NULL) {
	    stringPtr->uallocated =
	        STRING_UALLOC(numChars + appendNumChars) 
		+ TCL_GROWTH_MIN_ALLOC;
	    tmpString = (String *) ckrealloc((char *)stringPtr,
		    STRING_SIZE(stringPtr->uallocated));
	}
	stringPtr = tmpString;
	SET_STRING(objPtr, stringPtr);
    }

    /*
     * Copy the new string onto the end of the old string, then add the
     * trailing null.
     */

    memcpy((VOID*) (stringPtr->unicode + stringPtr->numChars), unicode,
	    appendNumChars * sizeof(Tcl_UniChar));
    stringPtr->unicode[numChars] = 0;
    stringPtr->numChars = numChars;

    Tcl_InvalidateStringRep(objPtr);
}

/*
 *----------------------------------------------------------------------
 *
 * AppendUnicodeToUtfRep --
 *
 *	This procedure converts the contents of "unicode" to UTF and
 *	appends the UTF to the string rep of "objPtr".
 *
 * Results:
 *	None.
 *
 * Side effects:
 *	objPtr's internal rep is reallocated.
 *
 *----------------------------------------------------------------------
 */

static void
AppendUnicodeToUtfRep(objPtr, unicode, numChars)
    Tcl_Obj *objPtr;	        /* Points to the object to append to. */
    CONST Tcl_UniChar *unicode; /* String to convert to UTF. */
    int numChars;	        /* Number of chars of "unicode" to convert. */
{
    Tcl_DString dsPtr;
    CONST char *bytes;
    
    if (numChars < 0) {
	numChars = 0;
	if (unicode) {
	    while (unicode[numChars] != 0) { numChars++; }
	}
    }
    if (numChars == 0) {
	return;
    }

    Tcl_DStringInit(&dsPtr);
    bytes = Tcl_UniCharToUtfDString(unicode, numChars, &dsPtr);
    AppendUtfToUtfRep(objPtr, bytes, Tcl_DStringLength(&dsPtr));
    Tcl_DStringFree(&dsPtr);
}

/*
 *----------------------------------------------------------------------
 *
 * AppendUtfToUnicodeRep --
 *
 *	This procedure converts the contents of "bytes" to Unicode and
 *	appends the Unicode to the Unicode rep of "objPtr".  objPtr must
 *	already have a valid Unicode rep.
 *
 * Results:
 *	None.
 *
 * Side effects:
 *	objPtr's internal rep is reallocated.
 *
 *----------------------------------------------------------------------
 */

static void
AppendUtfToUnicodeRep(objPtr, bytes, numBytes)
    Tcl_Obj *objPtr;	/* Points to the object to append to. */
    CONST char *bytes;	/* String to convert to Unicode. */
    int numBytes;	/* Number of bytes of "bytes" to convert. */
{
    Tcl_DString dsPtr;
    int numChars;
    Tcl_UniChar *unicode;

    if (numBytes < 0) {
	numBytes = (bytes ? strlen(bytes) : 0);
    }
    if (numBytes == 0) {
	return;
    }
    
    Tcl_DStringInit(&dsPtr);
    numChars = Tcl_NumUtfChars(bytes, numBytes);
    unicode = (Tcl_UniChar *)Tcl_UtfToUniCharDString(bytes, numBytes, &dsPtr);
    AppendUnicodeToUnicodeRep(objPtr, unicode, numChars);
    Tcl_DStringFree(&dsPtr);
}

/*
 *----------------------------------------------------------------------
 *
 * AppendUtfToUtfRep --
 *
 *	This procedure appends "numBytes" bytes of "bytes" to the UTF string
 *	rep of "objPtr".  objPtr must already have a valid String rep.
 *
 * Results:
 *	None.
 *
 * Side effects:
 *	objPtr's internal rep is reallocated.
 *
 *----------------------------------------------------------------------
 */

static void
AppendUtfToUtfRep(objPtr, bytes, numBytes)
    Tcl_Obj *objPtr;	/* Points to the object to append to. */
    CONST char *bytes;	/* String to append. */
    int numBytes;	/* Number of bytes of "bytes" to append. */
{
    String *stringPtr;
    int newLength, oldLength;

    if (numBytes < 0) {
	numBytes = (bytes ? strlen(bytes) : 0);
    }
    if (numBytes == 0) {
	return;
    }

    /*
     * Copy the new string onto the end of the old string, then add the
     * trailing null.
     */

    oldLength = objPtr->length;
    newLength = numBytes + oldLength;

    stringPtr = GET_STRING(objPtr);
    if (newLength > (int) stringPtr->allocated) {

	/*
	 * There isn't currently enough space in the string representation
	 * so allocate additional space.  First, try to double the length
	 * required.  If that fails, try a more modest allocation.  See the
	 * "TCL STRING GROWTH ALGORITHM" comment at the top of this file for an
	 * explanation of this growth algorithm.
	 */

	if (Tcl_AttemptSetObjLength(objPtr, 2 * newLength) == 0) {
	    Tcl_SetObjLength(objPtr,
		    newLength + numBytes + TCL_GROWTH_MIN_ALLOC);
	}
    }

    /*
     * Invalidate the unicode data.
     */
    
    stringPtr->numChars = -1;