Attachment #57916 for issue #51772

View | Details | Raw Unified | Return to issue 51772
Collapse All | Expand All

Lines 120-126 Link Here

(-)i18npool/inc/breakiteratorImpl.hxx (-1 / +2 lines)
120	virtual com::sun::star::uno::Sequence< rtl::OUString > SAL_CALL getSupportedServiceNames()	120	virtual com::sun::star::uno::Sequence< rtl::OUString > SAL_CALL getSupportedServiceNames()
121	throw( com::sun::star::uno::RuntimeException );	121	throw( com::sun::star::uno::RuntimeException );
122		122
123	static sal_Int16 SAL_CALL getScriptClass(sal_uInt32 currentChar);	123	static sal_Int16 SAL_CALL getScriptClass(sal_uInt32 currentChar, sal_uInt32 lastChar = 0);
		124	static sal_Bool SAL_CALL changeToWeak(sal_uInt32 ch, sal_uInt32 prev_ch);
124	protected:	125	protected:
125	Boundary result; // for word break iterator	126	Boundary result; // for word break iterator
126		127





        sal_Int32 strLen = Text.getLength();
		sal_uInt32 ch=0;
		sal_uInt32 prev_ch=Text.getStr()[0];
        while(iterateCodePoints(Text, nStartPos, 1, ch) < strLen ) {
            sal_Int16 currentCharScriptType = getScriptClass(ch, prev_ch);
            if(ScriptType != currentCharScriptType && currentCharScriptType != ScriptType::WEAK)
                break;
	    prev_ch = ch;
        }
        return  nStartPos;
}


#define scriptListCount sizeof (scriptList) / sizeof (UBlock2Script)

sal_Bool BreakIteratorImpl::changeToWeak(sal_uInt32 ch, sal_uInt32 prev_ch)
{
	switch (ch) {
		case 1 :
		case 2 :
		case 0x20 :
		case 0xA0 :
			return true;
		// Geresh or Gershayim or quotes following Hebrew
		case 0x22 : // quotation mark 
		case 0x05F4 : // Gershayim
		case 0x27 : // apostrophe
		case 0x05F3 : // Geresh
			if ( prev_ch >= 0x05D0 && prev_ch <= 0x05EA)
				return true;
		default :
			return false;
	}
}

sal_Int16  BreakIteratorImpl::getScriptClass(sal_uInt32 currentChar, sal_uInt32 lastChar)
{
        static sal_uInt32 lastChar = 0;
        static sal_Int16 nRet = 0;

        if (currentChar != lastChar) {
            lastChar = currentChar;

            //JP 21.9.2001: handle specific characters - always as weak
            //                  definition of 1 - this breaks a word
            //                  2 - this can be inside a word
            //                  0x20 & 0xA0 - Bug 102975, declare western space and non-break space as WEAK char.
            if ( changeToWeak(currentChar, lastChar))
                nRet = ScriptType::WEAK;
            // workaround for Coptic
            else if ( 0x2C80 <= currentChar && 0x2CE3 >= currentChar)

                }
                nRet=(i < scriptListCount && block >= scriptList[i].from) ? scriptList[i].script : ScriptType::WEAK;
            }
            lastChar = currentChar;
        }
        return nRet;
}

Return to issue 51772

Lines 300-309 Link Here

(-)i18npool/source/breakiterator/breakiteratorImpl.cxx (-6 / +26 lines)
300		300
301	sal_Int32 strLen = Text.getLength();	301	sal_Int32 strLen = Text.getLength();
302	sal_uInt32 ch=0;	302	sal_uInt32 ch=0;
		303	sal_uInt32 prev_ch=Text.getStr()[0];
303	while(iterateCodePoints(Text, nStartPos, 1, ch) < strLen ) {	304	while(iterateCodePoints(Text, nStartPos, 1, ch) < strLen ) {
304	sal_Int16 currentCharScriptType = getScriptClass(ch);	305	sal_Int16 currentCharScriptType = getScriptClass(ch, prev_ch);
305	if(ScriptType != currentCharScriptType && currentCharScriptType != ScriptType::WEAK)	306	if(ScriptType != currentCharScriptType && currentCharScriptType != ScriptType::WEAK)
306	break;	307	break;
		308	prev_ch = ch;
307	}	309	}
308	return nStartPos;	310	return nStartPos;
309	}	311	}
Lines 456-474 Link Here
456		458
457	#define scriptListCount sizeof (scriptList) / sizeof (UBlock2Script)	459	#define scriptListCount sizeof (scriptList) / sizeof (UBlock2Script)
458		460
459	sal_Int16 BreakIteratorImpl::getScriptClass(sal_uInt32 currentChar)	461	sal_Bool BreakIteratorImpl::changeToWeak(sal_uInt32 ch, sal_uInt32 prev_ch)
		462	{
		463	switch (ch) {
		464	case 1 :
		465	case 2 :
		466	case 0x20 :
		467	case 0xA0 :
		468	return true;
		469	// Geresh or Gershayim or quotes following Hebrew
		470	case 0x22 : // quotation mark
		471	case 0x05F4 : // Gershayim
		472	case 0x27 : // apostrophe
		473	case 0x05F3 : // Geresh
		474	if ( prev_ch >= 0x05D0 && prev_ch <= 0x05EA)
		475	return true;
		476	default :
		477	return false;
		478	}
		479	}
		480
		481	sal_Int16 BreakIteratorImpl::getScriptClass(sal_uInt32 currentChar, sal_uInt32 lastChar)
460	{	482	{
461	static sal_uInt32 lastChar = 0;
462	static sal_Int16 nRet = 0;	483	static sal_Int16 nRet = 0;
463		484
464	if (currentChar != lastChar) {	485	if (currentChar != lastChar) {
465	lastChar = currentChar;
466
467	//JP 21.9.2001: handle specific characters - always as weak	486	//JP 21.9.2001: handle specific characters - always as weak
468	// definition of 1 - this breaks a word	487	// definition of 1 - this breaks a word
469	// 2 - this can be inside a word	488	// 2 - this can be inside a word
470	// 0x20 & 0xA0 - Bug 102975, declare western space and non-break space as WEAK char.	489	// 0x20 & 0xA0 - Bug 102975, declare western space and non-break space as WEAK char.
471	if( 1 == currentChar \|\| 2 == currentChar \|\| 0x20 == currentChar \|\| 0xA0 == currentChar)	490	if ( changeToWeak(currentChar, lastChar))
472	nRet = ScriptType::WEAK;	491	nRet = ScriptType::WEAK;
473	// workaround for Coptic	492	// workaround for Coptic
474	else if ( 0x2C80 <= currentChar && 0x2CE3 >= currentChar)	493	else if ( 0x2C80 <= currentChar && 0x2CE3 >= currentChar)
Lines 481-486 Link Here
481	}	500	}
482	nRet=(i < scriptListCount && block >= scriptList[i].from) ? scriptList[i].script : ScriptType::WEAK;	501	nRet=(i < scriptListCount && block >= scriptList[i].from) ? scriptList[i].script : ScriptType::WEAK;
483	}	502	}
		503	lastChar = currentChar;
484	}	504	}
485	return nRet;	505	return nRet;
486	}	506	}