Attachment 55467 Details for Bug 40504 – A patch by tkos

[patch] A patch by tkos

skipDiacritics2.patch (text/plain), 2.88 KB, created by Lior Kaplan on 2012-01-11 13:37:30 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Lior Kaplan

Created: 2012-01-11 13:37:30 UTC

Size: 2.88 KB

patch

obsolete

>Index: linguistic/source/spelldsp.cxx
>===================================================================
>--- linguistic/source/spelldsp.cxx	(revision 271264)
>+++ linguistic/source/spelldsp.cxx	(working copy)
>@@ -279,6 +279,40 @@
> }
> 
> 
>+// Hebrew is usually written without diacritics.
>+// However, sometimes the diacritics are written as special marks located within, above, or below consonants.
>+// The diacritics are represented internally as separate Unicode characters.
>+// Hebrew dictionaries check for words without diacritics and will continue to do so for the forseeable future
>+// This function filters the diacritics out of a word, before spellchecking it
>+// (Using breakiterator is not appropriate, since we don't want word-breaking at the diacritics)
>+::rtl::OUString SpellCheckerDispatcher::removeDiacritics(const ::rtl::OUString& rWord, LanguageType nLanguage)
>+{
>+	OUString tmpWord;
>+	OUString diacritics;
>+	// Build a string containing diacritics to be skipped
>+	switch (nLanguage)
>+	{
>+		case LANGUAGE_HEBREW :
>+			int beginDiacritics;
>+			int endDiacritics;
>+			beginDiacritics = 0x591; // first of Hebrew cantillation marks
>+			endDiacritics = 0x5C5; // last of Hebrew niqqud
>+			for (int i = beginDiacritics; i <= endDiacritics; i++)
>+				diacritics += OUString((sal_Unicode)i);
>+			break;
>+		default:
>+			return rWord;
>+	}
>+	// Filter the diacritics from the word
>+	for (int i = 0; i < rWord.getLength(); i ++)
>+	{
>+		sal_Unicode c = rWord.getStr()[i];
>+		if (diacritics.indexOf(c) < 0)
>+			tmpWord += rWord.copy(i, 1);
>+	}
>+	return tmpWord;
>+}
>+
> BOOL SpellCheckerDispatcher::isValid_Impl(
> 			const OUString& rWord, 
>             LanguageType nLanguage, 
>@@ -304,7 +338,9 @@
> 	}
> 	else
> 	{
>-		OUString aChkWord( rWord );
>+	bool bRemoveDiacritics = 
>+		nLanguage == LANGUAGE_HEBREW ;
>+	OUString aChkWord( bRemoveDiacritics ? removeDiacritics(rWord, nLanguage) : rWord ); 
>         Locale aLocale( CreateLocale( nLanguage ) );
> 
>         // replace typographical apostroph by ascii apostroph
>@@ -480,7 +516,9 @@
> 	}
> 	else
> 	{
>-		OUString aChkWord( rWord );
>+	bool bRemoveDiacritics = 
>+		nLanguage == LANGUAGE_HEBREW ;
>+	OUString aChkWord( bRemoveDiacritics ? removeDiacritics(rWord, nLanguage) : rWord ); 
>         Locale aLocale( CreateLocale( nLanguage ) );
> 
>         // replace typographical apostroph by ascii apostroph
>Index: linguistic/source/spelldsp.hxx
>===================================================================
>--- linguistic/source/spelldsp.hxx	(revision 271264)
>+++ linguistic/source/spelldsp.hxx	(working copy)
>@@ -93,6 +93,8 @@
> 
> 	void	ClearSvcList();
> 
>+	::rtl::OUString removeDiacritics(const ::rtl::OUString& rWord, LanguageType nLanguage);
>+    
>     BOOL    isValid_Impl(const ::rtl::OUString& aWord, LanguageType nLanguage,
> 					const ::com::sun::star::beans::PropertyValues& aProperties,
> 					BOOL bCheckDics)

Actions: View

Attachments on bug 40504: 55467