shapearabic.c

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

    default :
        return(0);
        break;
    }
}

/*
 *Name     : specialChar
 *Function : Special Arabic characters need special handling in the shapeUnicode
 *           function, this function returns 1 or 2 for these special characters
 */
static int32_t
specialChar(UChar ch) {

    if( (ch>0x0621 && ch<0x0626)||(ch==0x0627)||(ch>0x062e && ch<0x0633)||
        (ch>0x0647 && ch<0x064a)||(ch==0x0629) ) {
        return (1);
    }
    else
    if( ch>=0x064B && ch<= 0x0652 )
        return (2);
    else
    if( (ch>=0x0653 && ch<= 0x0655) || ch == 0x0670 ||
        (ch>=0xFE70 && ch<= 0xFE7F) )
        return (3);
    else
        return (0);
}

/*
 *Name     : getLink
 *Function : Resolves the link between the characters as
 *           Arabic characters have four forms :
 *           Isolated, Initial, Middle and Final Form
 */
static UChar
getLink(UChar ch) {

    if(ch >= 0x0622 && ch <= 0x06D3) {
        return(araLink[ch-0x0622]);
    } else if(ch == 0x200D) {
        return(3);
    } else if(ch >= 0x206D && ch <= 0x206F) {
        return(4);
    } else if(ch >= 0xFE70 && ch <= 0xFEFC) {
        return(presLink[ch-0xFE70]);
    } else {
        return(0);
    }
}

/*
 *Name     : isTashkeelChar
 *Function : Returns 1 for Tashkeel characters else return 0
 */
static int32_t
isTashkeelChar(UChar ch) {

    if( ch>=0x064B && ch<= 0x0652 )
        return (1);
    else
        return (0);
}

/*
 *Name     : shapeUnicode
 *Function : Converts an Arabic Unicode buffer in 06xx Range into a shaped
 *           arabic Unicode buffer in FExx Range
 */
static int32_t
shapeUnicode(UChar *dest, int32_t sourceLength,
             int32_t destSize,
             int tashkeelFlag) {

    int32_t          i, iend;
    int32_t          prevPos, lastPos,Nx, Nw;
    unsigned int     Shape;
    int32_t          flag;
    int32_t          lamalef_found = 0;
    UChar            prevLink = 0, lastLink = 0, currLink, nextLink = 0;
    UChar            wLamalef;

    /*
     * Converts the input buffer from FExx Range into 06xx Range
     * to make sure that all characters are in the 06xx range
     * even the lamalef is converted to the special region in
     * the 06xx range
     */
    for (i = 0; i < sourceLength; i++) {
        UChar inputChar = dest[i];
        if ( (inputChar >= 0xFE70) && (inputChar <= 0xFEFC)) {
            dest[i] = convertFEto06 [ (inputChar - 0xFE70) ] ;
        }
    }

    /* sets the index to the end of the buffer, together with the step point to -1 */
    i = 0;
    iend = sourceLength;

    /*
     * This function resolves the link between the characters .
     * Arabic characters have four forms :
     * Isolated Form, Initial Form, Middle Form and Final Form
     */
    currLink = getLink(dest[i]);

    prevPos = i;
    lastPos = i;
    Nx = sourceLength + 2, Nw = 0;

    while (i != iend) {
        /* If high byte of currLink > 0 then more than one shape */
        if ((currLink & 0xFF00) > 0 || isTashkeelChar(dest[i])) {
            Nw = i + 1;
            while (Nx >= sourceLength) {         /* we need to know about next char */
                if(Nw == iend) {
                    nextLink = 0;
                    Nx = -1;
                } else {
                    nextLink = getLink(dest[Nw]);
                    if((nextLink & IRRELEVANT) == 0) {
                        Nx = Nw;
                    } else {
                        Nw = Nw + 1;
                    }
                }
            }

            if ( ((currLink & ALEFTYPE) > 0)  &&  ((lastLink & LAMTYPE) > 0) ) {
                lamalef_found = 1;
                wLamalef = changeLamAlef(dest[i]); /*get from 0x065C-0x065f */
                if ( wLamalef != 0) {
                    dest[i] = ' ';               /* The default case is to drop the Alef and replace */
                    dest[lastPos] =wLamalef;     /* it by a space.                                   */
                    i=lastPos;
                }
                lastLink = prevLink;
                currLink = getLink(wLamalef);
            }
            /*
             * get the proper shape according to link ability of neighbors
             * and of character; depends on the order of the shapes
             * (isolated, initial, middle, final) in the compatibility area
             */
             flag  = specialChar(dest[i]);

             Shape = shapeTable[nextLink & (LINKR + LINKL)]
                               [lastLink & (LINKR + LINKL)]
                               [currLink & (LINKR + LINKL)];

             if (flag == 1) {
                 Shape = (Shape == 1 || Shape == 3) ? 1 : 0;
             }
             else
             if(flag == 2) {
                 if( (lastLink & LINKL) && (nextLink & LINKR) && (tashkeelFlag == 1) &&
                      dest[i] != 0x064C && dest[i] != 0x064D ) {
                     Shape = 1;
                     if( (nextLink&ALEFTYPE) == ALEFTYPE && (lastLink&LAMTYPE) == LAMTYPE )
                         Shape = 0;
                 }
                 else {
                     Shape = 0;
                 }
             }

             if(flag == 2) {
                 dest[i] =  0xFE70 + IrrelevantPos[(dest[i] - 0x064B)] + Shape;
             }
             else
                 dest[i] = (UChar)((dest[i] < 0x0670 ? 0xFE70 : 0xFB50) + (currLink >> 8) + Shape);
        }

        /* move one notch forward */
        if ((currLink & IRRELEVANT) == 0) {
              prevLink = lastLink;
              lastLink = currLink;
              prevPos = lastPos;
              lastPos = i;
        }

        i++;
        if (i == Nx) {
            currLink = nextLink;
            Nx = sourceLength + 2;
        }
        else if(i != iend) {
            currLink = getLink(dest[i]);
        }
    }

    destSize = sourceLength;

    return destSize;
}

int32_t shapeArabic(const UChar *source, int32_t sourceLength, UChar *dest, int32_t destCapacity, uint32_t options, UErrorCode *pErrorCode) {
    int32_t destLength;

    /* usual error checking */
    if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
        return 0;
    }

    /* make sure that no reserved options values are used; allow dest==NULL only for preflighting */
    if( source==NULL || sourceLength<-1 ||
        (dest==NULL && destCapacity!=0) || destCapacity<0 ||
        options>=U_SHAPE_DIGIT_TYPE_RESERVED ||
        (options&U_SHAPE_DIGITS_MASK)>=U_SHAPE_DIGITS_RESERVED
    ) {
        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
        return 0;
    }

    /* determine the source length */
    if(sourceLength==-1) {
        sourceLength=u_strlen(source);
    }
    if(sourceLength==0) {
        return 0;
    }

    /* check that source and destination do not overlap */
    if( dest!=NULL &&
        ((source<=dest && dest<source+sourceLength) ||
         (dest<=source && source<dest+destCapacity))
    ) {
        *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
        return 0;
    }

    if((options&U_SHAPE_LETTERS_MASK)!=U_SHAPE_LETTERS_NOOP) {
        int32_t outputSize = sourceLength;

        /* calculate destination size */
        /* TODO: do we ever need to do this pure preflighting? */
        ASSERT((options&U_SHAPE_LENGTH_MASK) != U_SHAPE_LENGTH_GROW_SHRINK);

        if(outputSize>destCapacity) {
            *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
            return outputSize;
        }

        /* Start of Arabic letter shaping part */
        memcpy(dest, source, sourceLength*U_SIZEOF_UCHAR);

        ASSERT((options&U_SHAPE_TEXT_DIRECTION_MASK) == U_SHAPE_TEXT_DIRECTION_LOGICAL);

        switch(options&U_SHAPE_LETTERS_MASK) {
        case U_SHAPE_LETTERS_SHAPE :
            /* Call the shaping function with tashkeel flag == 1 */
            destLength = shapeUnicode(dest,sourceLength,destCapacity,1);
            break;
        case U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED :
            /* Call the shaping function with tashkeel flag == 0 */
            destLength = shapeUnicode(dest,sourceLength,destCapacity,0);
            break;
        case U_SHAPE_LETTERS_UNSHAPE :
            ASSERT_NOT_REACHED();
            break;
        default :
            /* will never occur because of validity checks above */
            destLength = 0;
            break;
        }

        /* End of Arabic letter shaping part */
    } else
        ASSERT_NOT_REACHED();

    ASSERT((options & U_SHAPE_DIGITS_MASK) == U_SHAPE_DIGITS_NOOP); 

    return sourceLength;
}

#endif // USE(ATSUI)