tdf#150582 linguistic: fix always rejected words with U+2019 apostrophe

During spell checking, words added to the user dictionaries still
weren't recognized by Calc, Impress and Draw because of the unnecessary
conversion their typographic apostrophes to ASCII apostrophes.

Note: Writer has no such problem because of its (obsolete) apostrophe
conversion for the user dictionary.

First skip the obsolete apostrophe conversion in isValid(), but keep
it as a fallback for Writer users.

Change-Id: I09870a35d7a91017281ba4b228d338336e7bd9c1
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/168858
Tested-by: Jenkins
Reviewed-by: László Németh <nemeth@numbertext.org>

linguistic/source/spelldsp.cxx

@@ -199,10 +199,19 @@ sal_Bool SAL_CALL
             const css::uno::Sequence< ::css::beans::PropertyValue >& rProperties )
 {
     MutexGuard  aGuard( GetLinguMutex() );
-    return isValid_Impl( rWord, LinguLocaleToLanguage( rLocale ), rProperties );
+    // for historical reasons, the word can be only with ASCII apostrophe in the dictionaries,
+    // so as a fallback, convert typographical apostrophes to avoid annoying users, if they
+    // have old (user) dictionaries only with the obsolete ASCII apostrophe.
+    bool bConvert = false;
+    bool bRet = isValid_Impl( rWord, LinguLocaleToLanguage( rLocale ), rProperties, bConvert );
+    if (!bRet && bConvert)
+    {
+        // fallback: convert the apostrophes
+        bRet = isValid_Impl( rWord, LinguLocaleToLanguage( rLocale ), rProperties, bConvert );
+    }
+    return bRet;
 }
 
-
 Reference< XSpellAlternatives > SAL_CALL
     SpellCheckerDispatcher::spell( const OUString& rWord, const Locale& rLocale,
             const css::uno::Sequence< ::css::beans::PropertyValue >& rProperties )
@@ -249,7 +258,8 @@ static Reference< XDictionaryEntry > lcl_GetRulingDictionaryEntry(
 bool SpellCheckerDispatcher::isValid_Impl(
             const OUString& rWord,
             LanguageType nLanguage,
-            const PropertyValues& rProperties)
+            const PropertyValues& rProperties,
+            bool& rConvertApostrophe)
 {
     MutexGuard  aGuard( GetLinguMutex() );
 
@@ -267,11 +277,21 @@ bool SpellCheckerDispatcher::isValid_Impl(
         OUString aChkWord( rWord );
         Locale aLocale( LanguageTag::convertToLocale( nLanguage ) );
 
-        // replace typographical apostroph by ascii apostroph
+        // replace typographical apostrophe by ASCII apostrophe only as a fallback
+        // for old user dictionaries before the time of the default typographical apostrophe
+        // (Note: otherwise also no problem with non-Unicode Hunspell dictionaries, because
+        // the character conversion converts also the typographical apostrophe to the ASCII one)
         OUString aSingleQuote( GetLocaleDataWrapper( nLanguage ).getQuotationMarkEnd() );
         DBG_ASSERT( 1 == aSingleQuote.getLength(), "unexpected length of quotation mark" );
-        if (!aSingleQuote.isEmpty())
+        if (!aSingleQuote.isEmpty() && aChkWord.indexOf(aSingleQuote[0]) > -1)
-            aChkWord = aChkWord.replace( aSingleQuote[0], '\'' );
+        {
+            // tdf#150582 first check with the original typographical apostrophe,
+            // and convert it only on the second try
+            if (rConvertApostrophe)
+                aChkWord = aChkWord.replace( aSingleQuote[0], '\'' );
+            else
+                rConvertApostrophe = true;
+        }
 
         RemoveHyphens( aChkWord );
         if (IsIgnoreControlChars( rProperties, GetPropSet() ))

linguistic/source/spelldsp.hxx

@@ -69,7 +69,7 @@ class SpellCheckerDispatcher :
     /// @throws css::uno::RuntimeException
     /// @throws css::lang::IllegalArgumentException
     bool    isValid_Impl(const OUString& aWord, LanguageType nLanguage,
-                    const css::beans::PropertyValues& aProperties);
+                    const css::beans::PropertyValues& aProperties, bool& rConvertApostrophe);
 
     /// @throws css::uno::RuntimeException
     /// @throws css::lang::IllegalArgumentException