SearchFlags::WILD_MATCH_SELECTION, SearchOptions2::WildcardEscapeCharacter

At least '\' (search in Word) and '~' (search in Excel) should be
supported as escape character.

Being able to restrict a match to entire selection instead of substring
speeds up the Calc match whole cell scenario.

Change-Id: Ice242b9cd59009f172b724e03c2cc08feda4cd3c

i18npool/qa/cppunit/test_textsearch.cxx

@@ -145,6 +145,9 @@ void TestTextSearch::testWildcardSearch()
     util::SearchResult aRes;
 
     aOptions.AlgorithmType2 = util::SearchAlgorithms2::WILDCARD ;
+    aOptions.WildcardEscapeCharacter = '~';
+    // aOptions.searchFlag = ::css::util::SearchFlags::WILD_MATCH_SELECTION;
+    // is not set, so substring match is allowed.
     aOptions.transliterateFlags = ::css::i18n::TransliterationModules::TransliterationModules_IGNORE_CASE;
     aText = "abAca";
 

i18npool/source/search/textsearch.cxx

@@ -113,8 +113,6 @@ TextSearch::TextSearch(const Reference < XComponentContext > & rxContext)
         , pJumpTable2( nullptr )
         , pRegexMatcher( nullptr )
         , pWLD( nullptr )
-        , mcWildcardEscapeChar('~')         /* TODO: make this option available through API */
-        , mbWildcardAllowSubstring(true)    /* TODO: make this option available through API */
 {
     SearchOptions2 aOpt;
     aOpt.AlgorithmType2 = SearchAlgorithms2::ABSOLUTE;
@@ -245,6 +243,8 @@ void TextSearch::setOptions2( const SearchOptions2& rOptions ) throw( RuntimeExc
             break;
 
         case SearchAlgorithms2::WILDCARD:
+            mcWildcardEscapeChar = static_cast<sal_uInt32>(aSrchPara.WildcardEscapeCharacter);
+            mbWildcardAllowSubstring = ((aSrchPara.searchFlag & SearchFlags::WILD_MATCH_SELECTION) == 0);
             fnForward = &TextSearch::WildcardSrchFrwrd;
             fnBackward = &TextSearch::WildcardSrchBkwrd;
             break;
@@ -289,7 +289,8 @@ void TextSearch::setOptions( const SearchOptions& rOptions ) throw( RuntimeExcep
             rOptions.deletedChars,
             rOptions.insertedChars,
             rOptions.transliterateFlags,
-            nAlgorithmType2
+            nAlgorithmType2,
+            0   // no wildcard search, no escape character..
             );
     setOptions2( aOptions2);
 }

include/unotools/textsearch.hxx

@@ -96,9 +96,12 @@ private:
 
     SearchType m_eSrchType;       // search normal/regular/LevDist
 
+    sal_uInt32 m_cWildEscChar;      // wildcard escape character
+
     bool m_bWordOnly   : 1;        // used by normal search
     bool m_bSrchInSel  : 1;        // search only in the selection
     bool m_bCaseSense  : 1;
+    bool m_bWildMatchSel : 1;       // wildcard pattern must match entire selection
 
     // values for the "weight Levenshtein-Distance"
     bool bLEV_Relaxed : 1;
@@ -114,7 +117,9 @@ public:
                     SearchType eSrchType = SearchParam::SRCH_NORMAL,
                     bool bCaseSensitive = true,
                     bool bWordOnly = false,
-                    bool bSearchInSelection = false );
+                    bool bSearchInSelection = false,
+                    sal_uInt32 cWildEscChar = '\\',
+                    bool bWildMatchSel = false );
 
     SearchParam( const SearchParam& );
 
@@ -127,6 +132,10 @@ public:
     bool            IsCaseSensitive() const     { return m_bCaseSense; }
     bool            IsSrchInSelection() const   { return m_bSrchInSel; }
     bool            IsSrchWordOnly() const      { return m_bWordOnly; }
+    bool            IsWildMatchSel() const      { return m_bWildMatchSel; }
+
+    // signed return for API use
+    sal_Int32       GetWildEscChar() const      { return static_cast<sal_Int32>(m_cWildEscChar); }
 
     bool            IsSrchRelaxed() const       { return bLEV_Relaxed; }
     int             GetLEVOther() const         { return nLEV_OtherX; }

offapi/com/sun/star/util/SearchFlags.idl

@@ -121,6 +121,20 @@ published constants SearchFlags
         positives, but meets user expectation better. </p>
     */
     const long  LEV_RELAXED     = 0x00010000;
+
+    /** Flag for wildcards search if entire selection must match the
+        pattern.
+
+        <p> If com::sun::star::util::SearchOptions2::AlgorithmType2 is
+        com::sun::star::util::SearchAlgorithms2::WILDCARD specifies
+        whether a wildcard pattern must match the entire selected range
+        of the string from start position to end position or a substring
+        match is allowed. </p>
+
+        <p> If set, the entire selection must match. If not set, a
+        substring match is allowed. </p>
+     */
+    const long  WILD_MATCH_SELECTION = 0x00100000;
 };
 
 }; }; }; };

offapi/com/sun/star/util/SearchOptions2.idl

@@ -29,6 +29,20 @@ published struct SearchOptions2 : com::sun::star::util::SearchOptions {
         SearchAlgorithms SearchOptions::algorithmType enum field.
      */
     short   AlgorithmType2;
+
+    /** The escape character to be used with a
+        com::sun::star::util::SearchAlgorithms2::WILDCARD search.
+
+        <p> A Unicode character, if not 0 escapes the special meaning of
+        a question mark, asterisk or escape character that follows
+        immediately after the escape character. If 0 defines no escape
+        character is used. </p>
+
+        <p> Common values are '\' (U+005C REVERSE SOLIDUS) aka backslash
+        in text processing context, or '~' (U+007E TILDE) in spreadsheet
+        processing context. </p>
+     */
+    long    WildcardEscapeCharacter;
 };
 
 }; }; }; };

sc/inc/queryentry.hxx

@@ -59,7 +59,8 @@ struct SC_DLLPUBLIC ScQueryEntry
     ~ScQueryEntry();
 
     /// creates pSearchParam and pSearchText if necessary
-    utl::TextSearch* GetSearchTextPtr( utl::SearchParam::SearchType eSearchType, bool bCaseSens ) const;
+    utl::TextSearch* GetSearchTextPtr( utl::SearchParam::SearchType eSearchType, bool bCaseSens,
+            bool bWildMatchSel ) const;
 
     QueryItemsType& GetQueryItems() { return maQueryItems;}
     const QueryItemsType& GetQueryItems() const { return maQueryItems;}

sc/source/core/data/dpcache.cxx

@@ -537,7 +537,7 @@ bool ScDPCache::ValidQuery( SCROW nRow, const ScQueryParam &rParam) const
                 sal_Int32 nStart = 0;
                 sal_Int32 nEnd   = aCellStr.getLength();
 
-                bool bMatch = (bool) rEntry.GetSearchTextPtr( rParam.eSearchType, rParam.bCaseSens )
+                bool bMatch = (bool) rEntry.GetSearchTextPtr( rParam.eSearchType, rParam.bCaseSens, bMatchWholeCell )
                               ->SearchForward( aCellStr, &nStart, &nEnd );
                 // from 614 on, nEnd is behind the found text
                 if (bMatch && bMatchWholeCell

sc/source/core/data/table3.cxx

@@ -2423,12 +2423,12 @@ public:
             {
                 nEnd = 0;
                 nStart = aCellStr.getLength();
-                bMatch = rEntry.GetSearchTextPtr( mrParam.eSearchType, mrParam.bCaseSens )
+                bMatch = rEntry.GetSearchTextPtr( mrParam.eSearchType, mrParam.bCaseSens, bMatchWholeCell )
                     ->SearchBackward(aCellStr.getString(), &nStart, &nEnd);
             }
             else
             {
-                bMatch = rEntry.GetSearchTextPtr( mrParam.eSearchType, mrParam.bCaseSens )
+                bMatch = rEntry.GetSearchTextPtr( mrParam.eSearchType, mrParam.bCaseSens, bMatchWholeCell )
                     ->SearchForward(aCellStr.getString(), &nStart, &nEnd);
             }
             if ( bMatch && bMatchWholeCell

sc/source/core/tool/compare.cxx

@@ -140,9 +140,8 @@ double CompareFunc( const Compare& rComp, CompareOptions* pOptions )
             {
                 sal_Int32 nStart = 0;
                 sal_Int32 nStop  = rCell1.maStr.getLength();
-                bool bMatch = rEntry.GetSearchTextPtr( pOptions->eSearchType,
-                        !rComp.mbIgnoreCase)->SearchForward(
-                            rCell1.maStr.getString(), &nStart, &nStop);
+                bool bMatch = rEntry.GetSearchTextPtr( pOptions->eSearchType, !rComp.mbIgnoreCase,
+                        pOptions->bMatchWholeCell)->SearchForward( rCell1.maStr.getString(), &nStart, &nStop);
                 if (bMatch && pOptions->bMatchWholeCell && (nStart != 0 || nStop != rCell1.maStr.getLength()))
                     bMatch = false;     // RegEx must match entire string.
                 fRes = (bMatch ? 0 : 1);

sc/source/core/tool/interpr1.cxx

@@ -8148,7 +8148,7 @@ void ScInterpreter::ScSearch()
         else
         {
             utl::SearchParam::SearchType eSearchType = DetectSearchType( SearchStr, pDok );
-            utl::SearchParam sPar(SearchStr, eSearchType, false, false, false);
+            utl::SearchParam sPar(SearchStr, eSearchType, false, false, false, '~', false);
             utl::TextSearch sT( sPar, *ScGlobal::pCharClass );
             bool bBool = sT.SearchForward(sStr, &nPos, &nEndPos);
             if (!bBool)

sc/source/core/tool/queryentry.cxx

@@ -164,13 +164,14 @@ bool ScQueryEntry::operator==( const ScQueryEntry& r ) const
     // do not compare pSearchParam and pSearchText!
 }
 
-utl::TextSearch* ScQueryEntry::GetSearchTextPtr( utl::SearchParam::SearchType eSearchType, bool bCaseSens ) const
+utl::TextSearch* ScQueryEntry::GetSearchTextPtr( utl::SearchParam::SearchType eSearchType, bool bCaseSens,
+        bool bWildMatchSel ) const
 {
     if ( !pSearchParam )
     {
         OUString aStr = maQueryItems[0].maString.getString();
         pSearchParam = new utl::SearchParam(
-            aStr, eSearchType, bCaseSens, false, false);
+            aStr, eSearchType, bCaseSens, false, false, '~', bWildMatchSel);
         pSearchText = new utl::TextSearch( *pSearchParam, *ScGlobal::pCharClass );
     }
     return pSearchText;

unotools/source/i18n/textsearch.cxx

@@ -41,14 +41,19 @@ SearchParam::SearchParam( const OUString &rText,
                                 SearchType eType,
                                 bool bCaseSensitive,
                                 bool bWrdOnly,
-                                bool bSearchInSel )
+                                bool bSearchInSel,
+                                sal_uInt32 cWildEscChar,
+                                bool bWildMatchSel )
 {
     sSrchStr        = rText;
     m_eSrchType     = eType;
 
+    m_cWildEscChar  = cWildEscChar;
+
     m_bWordOnly     = bWrdOnly;
     m_bSrchInSel    = bSearchInSel;
     m_bCaseSense    = bCaseSensitive;
+    m_bWildMatchSel = bWildMatchSel;
 
     nTransliterationFlags = 0;
 
@@ -65,9 +70,12 @@ SearchParam::SearchParam( const SearchParam& rParam )
     sReplaceStr     = rParam.sReplaceStr;
     m_eSrchType     = rParam.m_eSrchType;
 
+    m_cWildEscChar  = rParam.m_cWildEscChar;
+
     m_bWordOnly     = rParam.m_bWordOnly;
     m_bSrchInSel    = rParam.m_bSrchInSel;
     m_bCaseSense    = rParam.m_bCaseSense;
+    m_bWildMatchSel = rParam.m_bWildMatchSel;
 
     bLEV_Relaxed    = rParam.bLEV_Relaxed;
     nLEV_OtherX     = rParam.nLEV_OtherX;
@@ -83,6 +91,7 @@ static bool lcl_Equals( const SearchOptions2& rSO1, const SearchOptions2& rSO2 )
 {
     return
         rSO1.AlgorithmType2 == rSO2.AlgorithmType2 &&
+        rSO1.WildcardEscapeCharacter == rSO2.WildcardEscapeCharacter &&
         rSO1.algorithmType == rSO2.algorithmType &&
         rSO1.searchFlag == rSO2.searchFlag &&
         rSO1.searchString.equals(rSO2.searchString) &&
@@ -175,7 +184,8 @@ css::util::SearchOptions2 TextSearch::UpgradeToSearchOptions2( const css::util::
             rOptions.deletedChars,
             rOptions.insertedChars,
             rOptions.transliterateFlags,
-            nAlgorithmType2
+            nAlgorithmType2,
+            0       // no wildcard search, no escape character..
             );
     return aOptions2;
 }
@@ -190,6 +200,9 @@ void TextSearch::Init( const SearchParam & rParam,
     {
     case SearchParam::SRCH_WILDCARD:
         aSOpt.AlgorithmType2 = SearchAlgorithms2::WILDCARD;
+        aSOpt.WildcardEscapeCharacter = rParam.GetWildEscChar();
+        if (rParam.IsWildMatchSel())
+            aSOpt.searchFlag |= SearchFlags::WILD_MATCH_SELECTION;
         aSOpt.algorithmType = static_cast<SearchAlgorithms>(-1);    // no old enum for that
         break;
 
@@ -237,7 +250,6 @@ void TextSearch::Init( const SearchParam & rParam,
 void TextSearch::SetLocale( const css::util::SearchOptions2& rOptions,
                             const css::lang::Locale& rLocale )
 {
-    // convert SearchParam to the UNO SearchOptions2
     SearchOptions2 aSOpt( rOptions );
     aSOpt.Locale = rLocale;