Handle English vs locale's General keyword, tdf#33689 follow-up, tdf#107867

Caller of ImpSvNumberformatScan::GetKeyWord() needs to know if
English or locale's keyword was found and act on it.

Pointed out by an assert when attempting to copy "STANDARD" length
characters (8) from "General" (7), which doesn't harm because when
not in debug it copied the terminating 0-character. However, the
keyword name also needs to be replaced for the locale's resulting
format code.

Change-Id: Ia5211b307ea44c1ba9c2b86786368355c570d4a2
Reviewed-on: https://gerrit.libreoffice.org/55293
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit bfe6e9ecb18e8a3d9da368a1eb48e807ab9041cf)
Reviewed-on: https://gerrit.libreoffice.org/55297

2 files changed, 41 insertions, 12 deletions

diff --git a/svl/source/numbers/zforscan.cxx b/svl/source/numbers/zforscan.cxx
index d9e1fba36a8f..40b522d074a9 100644
--- a/svl/source/numbers/zforscan.cxx
+++ b/svl/source/numbers/zforscan.cxx
@@ -667,17 +667,21 @@ Color* ImpSvNumberformatScan::GetColor(OUString& sStr)
     return pResult;
 }
 
-short ImpSvNumberformatScan::GetKeyWord( const OUString& sSymbol, sal_Int32 nPos ) const
+short ImpSvNumberformatScan::GetKeyWord( const OUString& sSymbol, sal_Int32 nPos, bool& rbFoundEnglish ) const
 {
     OUString sString = pFormatter->GetCharClass()->uppercase( sSymbol, nPos, sSymbol.getLength() - nPos );
     const NfKeywordTable & rKeyword = GetKeywords();
     // #77026# for the Xcl perverts: the GENERAL keyword is recognized anywhere
-    if ( sString.startsWith( rKeyword[NF_KEY_GENERAL] ) ||
-            ((meKeywordLocalization == KeywordLocalization::AllowEnglish) &&
-             sString.startsWith( sEnglishKeyword[NF_KEY_GENERAL])))
+    if (sString.startsWith( rKeyword[NF_KEY_GENERAL] ))
     {
         return NF_KEY_GENERAL;
     }
+    if ((meKeywordLocalization == KeywordLocalization::AllowEnglish) &&
+            sString.startsWith( sEnglishKeyword[NF_KEY_GENERAL]))
+    {
+        rbFoundEnglish = true;
+        return NF_KEY_GENERAL;
+    }
     //! MUST be a reverse search to find longer strings first
     short i = NF_KEYWORD_ENTRIES_COUNT-1;
     bool bFound = false;
@@ -733,6 +737,7 @@ short ImpSvNumberformatScan::GetKeyWord( const OUString& sSymbol, sal_Int32 nPos
                 }
                 if ( j && sEnglishKeyword[j].getLength() > sEnglishKeyword[i].getLength() )
                 {
+                    rbFoundEnglish = true;
                     return j;
                 }
             }
@@ -898,7 +903,8 @@ short ImpSvNumberformatScan::Next_Symbol( const OUString& rStr,
                 }
                 else if ( pChrCls->isLetter( rStr, nPos-1 ) )
                 {
-                    short nTmpType = GetKeyWord( rStr, nPos-1 );
+                    bool bFoundEnglish = false;
+                    short nTmpType = GetKeyWord( rStr, nPos-1, bFoundEnglish);
                     if ( nTmpType )
                     {
                         bool bCurrency = false;
@@ -906,7 +912,7 @@ short ImpSvNumberformatScan::Next_Symbol( const OUString& rStr,
                         // like "R" (Rand) and 'R' (era)
                         if ( nCurrPos >= 0 &&
                              nPos-1 + sCurString.getLength() <= rStr.getLength() &&
-                             sCurString.startsWith( sKeyword[nTmpType] ) )
+                             sCurString.startsWith( bFoundEnglish ? sEnglishKeyword[nTmpType] : sKeyword[nTmpType]))
                         {
                             OUString aTest = pChrCls->uppercase( rStr.copy( nPos-1, sCurString.getLength() ) );
                             if ( aTest == sCurString )
@@ -922,8 +928,22 @@ short ImpSvNumberformatScan::Next_Symbol( const OUString& rStr,
                         else
                         {
                             eType = nTmpType;
-                            sal_Int32 nLen = sKeyword[eType].getLength();
-                            sSymbol = rStr.copy( nPos-1, nLen );
+                            // The code to be advanced is the detected keyword,
+                            // not necessarily the locale's keyword, but the
+                            // symbol is to be the locale's keyword.
+                            sal_Int32 nLen;
+                            if (bFoundEnglish)
+                            {
+                                nLen = sEnglishKeyword[eType].getLength();
+                                // Use the locale's General keyword name, not uppercase.
+                                sSymbol = (eType == NF_KEY_GENERAL ? sNameStandardFormat : sKeyword[eType]);
+                            }
+                            else
+                            {
+                                nLen = sKeyword[eType].getLength();
+                                // Preserve a locale's keyword's case as entered.
+                                sSymbol = rStr.copy( nPos-1, nLen);
+                            }
                             if ((eType == NF_KEY_E || IsAmbiguousE(eType)) && nPos < rStr.getLength())
                             {
                                 sal_Unicode cNext = rStr[nPos];
@@ -975,7 +995,8 @@ short ImpSvNumberformatScan::Next_Symbol( const OUString& rStr,
         case SsGetWord:
             if ( pChrCls->isLetter( rStr, nPos-1 ) )
             {
-                short nTmpType = GetKeyWord( rStr, nPos-1 );
+                bool bFoundEnglish = false;
+                short nTmpType = GetKeyWord( rStr, nPos-1, bFoundEnglish);
                 if ( nTmpType )
                 {
                     // beginning of keyword, stop scan and put back
@@ -3070,7 +3091,8 @@ sal_Int32 ImpSvNumberformatScan::FinalScan( OUString& rString )
                 sal_Int32 nLen = rStr.getLength();
                 for ( sal_Int32 j = 0; j < nLen; j++ )
                 {
-                    if ( (j == 0 || rStr[j - 1] != '\\') && GetKeyWord( rStr, j ) )
+                    bool bFoundEnglish = false;
+                    if ( (j == 0 || rStr[j - 1] != '\\') && GetKeyWord( rStr, j, bFoundEnglish) )
                     {
                         rStr = "\"" + rStr + "\"";
                         break; // for
diff --git a/svl/source/numbers/zforscan.hxx b/svl/source/numbers/zforscan.hxx
index e2246f574dd5..b7ee25d2ce52 100644
--- a/svl/source/numbers/zforscan.hxx
+++ b/svl/source/numbers/zforscan.hxx
@@ -247,8 +247,15 @@ private: // Private section
     short PreviousType( sal_uInt16 i ) const;   // Returns type before position skips EMPTY
     bool IsLastBlankBeforeFrac(sal_uInt16 i) const; // True <=> there won't be a ' ' until the '/'
     void Reset();                               // Reset all variables before starting the analysis
-    short GetKeyWord( const OUString& sSymbol,  // Determine keyword at nPos
-                      sal_Int32 nPos ) const;   // Return 0 <=> not found
+
+    /** Determine keyword at nPos.
+        @param  rbFoundEnglish set if English instead of locale's keyword
+                found, never cleared, thus init with false.
+        @return 0 if not found, else keyword enumeration.
+     */
+    short GetKeyWord( const OUString& sSymbol,
+                      sal_Int32 nPos,
+                      bool& rbFoundEnglish ) const;
 
     bool IsAmbiguousE( short nKey ) const  // whether nKey is ambiguous E of NF_KEY_E/NF_KEY_EC
         {