tdf#105657 Treat "Precision as shown" for fractions

For Option "Precision as shown",
 fraction must specificly be treated
ImpGetFractionElements retrieves values of each part
of fraction (integer, numerator, denominator)
independently from its exact representation
Update: avoid include of zformat.hxx in document4.cxx

Change-Id: Ia3ea2322f3d311c04ef71f3260730c7154c3dc15
Reviewed-on: https://gerrit.libreoffice.org/34331
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>

5 files changed, 119 insertions, 63 deletions

diff --git a/include/svl/zforlist.hxx b/include/svl/zforlist.hxx
index 37c2b26c91cd..fdc00627d051 100644
--- a/include/svl/zforlist.hxx
+++ b/include/svl/zforlist.hxx
@@ -530,6 +530,9 @@ public:
                                      sal_uInt16& nPrecision, sal_uInt16& nAnzLeading,
                                      LanguageType eLnge = LANGUAGE_DONTKNOW );
 
+    /// Get round value with fraction representation
+    double GetRoundFractionValue( sal_uInt32 nFormat, double fValue ) const;
+
     /// Check if format code string may be deleted by user
     bool IsUserDefined( const OUString& sStr, LanguageType eLnge = LANGUAGE_DONTKNOW );
 
diff --git a/include/svl/zformat.hxx b/include/svl/zformat.hxx
index ce632ae04920..d68a1007cc24 100644
--- a/include/svl/zformat.hxx
+++ b/include/svl/zformat.hxx
@@ -212,6 +212,9 @@ public:
                               sal_uInt16& nPrecision,
                               sal_uInt16& nAnzLeading) const;
 
+    /// Get index of subformat (0..3) according to conditions and fNumber value
+    sal_uInt16 GetSubformatIndex( double fNumber ) const;
+
     /// Count of decimal precision
     sal_uInt16 GetFormatPrecision() const   { return NumFor[0].Info().nCntPost; }
 
@@ -242,6 +245,9 @@ public:
     OUString GetDenominatorString( sal_uInt16 nNumFor ) const;
     OUString GetNumeratorString( sal_uInt16 nNumFor ) const;
     OUString GetIntegerFractionDelimiterString( sal_uInt16 nNumFor ) const;
+    /// Round fNumber to its fraction representation
+    double GetRoundFractionValue ( double fNumber ) const;
+
     /** If the count of string elements (substrings, ignoring [modifiers] and
         so on) in a subformat code nNumFor (0..3) is equal to the given number.
         Used by ImpSvNumberInputScan::IsNumberFormatMain() to detect a matched
@@ -579,6 +585,19 @@ private:
                                  sal_uInt16 nIx,
                                  bool bInteger );
 
+    /** Calculate each element of fraction:
+     * integer part, numerator part, denominator part
+     * @param fNumber value to be represented as fraction. Will contain absolute fractional part
+     * @param nIx subformat number 0..3
+     * @param fIntPart integral part of fraction
+     * @param nFrac numerator of fraction
+     * @param nDic denominator of fraction
+     */
+    SVL_DLLPRIVATE void ImpGetFractionElements( double& fNumber,
+                                                sal_uInt16 nIx,
+                                                double& fIntPart,
+                                                sal_uInt64& nFrac,
+                                                sal_uInt64& nDiv ) const;
     SVL_DLLPRIVATE bool ImpGetFractionOutput(double fNumber,
                                              sal_uInt16 nIx,
                                              OUStringBuffer& OutString);
diff --git a/sc/source/core/data/documen4.cxx b/sc/source/core/data/documen4.cxx
index c503b2fa6137..09b7c543e7e8 100644
--- a/sc/source/core/data/documen4.cxx
+++ b/sc/source/core/data/documen4.cxx
@@ -672,6 +672,10 @@ double ScDocument::RoundValueAsShown( double fVal, sal_uInt32 nFormat ) const
                         nPrecision = sal::static_int_cast<short>( nPrecision - (short)floor( log10( -fVal ) ) );
                     break;
                 }
+                case css::util::NumberFormat::FRACTION:     // get value of fraction representation
+                {
+                    return GetFormatTable()->GetRoundFractionValue( nFormat, fVal );
+                }
             }
         }
         else
diff --git a/svl/source/numbers/zforlist.cxx b/svl/source/numbers/zforlist.cxx
index 964abdce3a0c..71a071825a0a 100644
--- a/svl/source/numbers/zforlist.cxx
+++ b/svl/source/numbers/zforlist.cxx
@@ -1963,6 +1963,15 @@ void SvNumberFormatter::GetFormatSpecialInfo(sal_uInt32 nFormat,
     }
 }
 
+double SvNumberFormatter::GetRoundFractionValue( sal_uInt32 nFormat, double fValue ) const
+{
+    const SvNumberformat* pFormat = GetFormatEntry( nFormat );
+    if ( pFormat )
+        return pFormat->GetRoundFractionValue( fValue );
+    else
+        return fValue;
+}
+
 sal_uInt16 SvNumberFormatter::GetFormatPrecision( sal_uInt32 nFormat ) const
 {
     const SvNumberformat* pFormat = GetFormatEntry( nFormat );
diff --git a/svl/source/numbers/zformat.cxx b/svl/source/numbers/zformat.cxx
index ac0c6080ee4c..5d5349ee8f70 100644
--- a/svl/source/numbers/zformat.cxx
+++ b/svl/source/numbers/zformat.cxx
@@ -2347,6 +2347,31 @@ bool SvNumberformat::GetOutputString(double fNumber, sal_uInt16 nCharCount, OUSt
     return true;
 }
 
+sal_uInt16 SvNumberformat::GetSubformatIndex (double fNumber ) const
+{
+    sal_uInt16 nIx; // Index of the partial format
+    double fLimit_1 = fLimit1;
+    short nCheck = ImpCheckCondition(fNumber, fLimit_1, eOp1);
+    if (nCheck == -1 || nCheck == 1) // Only 1 String or True
+    {
+        nIx = 0;
+    }
+    else
+    {
+        double fLimit_2 = fLimit2;
+        nCheck = ImpCheckCondition(fNumber, fLimit_2, eOp2);
+        if (nCheck == -1 || nCheck == 1)
+        {
+            nIx = 1;
+        }
+        else
+        {
+            nIx = 2;
+        }
+    }
+    return nIx;
+}
+
 bool SvNumberformat::GetOutputString(double fNumber,
                                      OUString& OutString,
                                      Color** ppColor)
@@ -2437,24 +2462,7 @@ bool SvNumberformat::GetOutputString(double fNumber,
     }
     if ( !bHadStandard )
     {
-        sal_uInt16 nIx; // Index of the partial format
-        short nCheck = ImpCheckCondition(fNumber, fLimit1, eOp1);
-        if (nCheck == -1 || nCheck == 1) // Only 1 String or True
-        {
-            nIx = 0;
-        }
-        else
-        {
-            nCheck = ImpCheckCondition(fNumber, fLimit2, eOp2);
-            if (nCheck == -1 || nCheck == 1)
-            {
-                nIx = 1;
-            }
-            else
-            {
-                nIx = 2;
-            }
-        }
+        sal_uInt16 nIx = GetSubformatIndex ( fNumber ); // Index of the partial format
         if (fNumber < 0.0 &&
                 ((nIx == 0 && IsFirstSubformatRealNegative()) || // 1st, usually positive subformat
                  (nIx == 1 && IsSecondSubformatRealNegative()))) // 2nd, usually negative subformat
@@ -2677,45 +2685,31 @@ bool SvNumberformat::ImpGetScientificOutput(double fNumber,
     return bRes;
 }
 
-bool SvNumberformat::ImpGetFractionOutput(double fNumber,
-                                          sal_uInt16 nIx,
-                                          OUStringBuffer& sBuff)
+double SvNumberformat::GetRoundFractionValue ( double fNumber ) const
 {
-    bool bRes = false;
-    const ImpSvNumberformatInfo& rInfo = NumFor[nIx].Info();
-    const sal_uInt16 nAnz = NumFor[nIx].GetCount();
-    OUStringBuffer sStr, sFrac, sDiv; // Strings, value for
-    sal_uInt64 nFrac=0, nDiv=1;       // Integral part
-    bool bSign = false;               // Numerator and denominator
-    const OUString sIntegerFormat = lcl_GetFractionIntegerString(rInfo, nAnz);
-    const OUString sNumeratorFormat = lcl_GetNumeratorString(rInfo, nAnz);
-    const OUString sDenominatorFormat = lcl_GetDenominatorString(rInfo, nAnz);
+    sal_uInt16 nIx = GetSubformatIndex ( fNumber );
+    double fIntPart = 0.0;           // integer part of fraction
+    sal_uInt64 nFrac = 0, nDiv = 1;  // numerator and denominator
+    double fSign = (fNumber < 0.0) ? -1.0 : 1.0;
+    // fNumber is modified in ImpGetFractionElements to absolute fractional part
+    ImpGetFractionElements ( fNumber, nIx, fIntPart, nFrac, nDiv );
+    if ( nDiv > 0 )
+        return fSign * ( fIntPart + (double)nFrac / (double)nDiv );
+    else
+        return fSign * fIntPart;
+}
 
-    if (fNumber < 0)
-    {
-        if (nIx == 0) // Not in the ones at the end
-            bSign = true; // Formats
+void SvNumberformat::ImpGetFractionElements ( double& fNumber, sal_uInt16 nIx,
+                                              double& fIntPart, sal_uInt64& nFrac, sal_uInt64& nDiv ) const
+{
+    if ( fNumber < 0.0 )
         fNumber = -fNumber;
-    }
-
-    double fNum = floor(fNumber); // Integral part
-
-    fNumber -= fNum; // Fractional part
-    if (fNum > D_MAX_U_INT32 || rInfo.nCntExp > 9) // Too large
-    {
-        sBuff = rScan.GetErrorString();
-        return false;
-    }
-    if (rInfo.nCntExp == 0)
-    {
-        SAL_WARN( "svl.numbers", "SvNumberformat:: Fraction, nCntExp == 0");
-        sBuff.truncate();
-        return false;
-    }
-
-    if( sal_Int32 nForcedDiv = sDenominatorFormat.toInt32() )
+    fIntPart = floor(fNumber); // Integral part
+    fNumber -= fIntPart;         // Fractional part
+    const ImpSvNumberformatInfo& rInfo = NumFor[nIx].Info();
+    nDiv = lcl_GetDenominatorString( rInfo, NumFor[nIx].GetCount() ).toInt32();
+    if( nDiv > 0 )
     {   // Forced Denominator
-        nDiv = (sal_uInt64) nForcedDiv;
         nFrac = (sal_uInt64)floor ( fNumber * nDiv );
         double fFracNew = (double)nFrac / (double)nDiv;
         double fFracNew1 = (double)(nFrac + 1) / (double)nDiv;
@@ -2724,14 +2718,10 @@ bool SvNumberformat::ImpGetFractionOutput(double fNumber,
         {
             nFrac++;
         }
-        if( nFrac >= nDiv )
-        {
-            nFrac = nDiv = 0;
-            fNum = fNum + 1.0;
-        }
     }
     else // Calculated Denominator
     {
+        nDiv = 1;
         sal_uInt64 nBasis = ((sal_uInt64)floor( pow(10.0,rInfo.nCntExp))) - 1; // 9, 99, 999 ,...
         sal_uInt64 nFracPrev = 1L, nDivPrev = 0, nFracNext, nDivNext, nPartialDenom;
         double fRemainder = fNumber;
@@ -2769,12 +2759,43 @@ bool SvNumberformat::ImpGetFractionOutput(double fNumber,
                 fRemainder = 0.0; // exit while loop
             }
         }
-        if (nFrac == nDiv)
-        {
-            ++fNum;
-            nFrac = 0;
-        }
     }
+    if (nFrac >= nDiv)
+    {
+        ++fIntPart;
+        nFrac = nDiv = 0;
+    }
+}
+
+bool SvNumberformat::ImpGetFractionOutput(double fNumber,
+                                          sal_uInt16 nIx,
+                                          OUStringBuffer& sBuff)
+{
+    bool bRes = false;
+    const ImpSvNumberformatInfo& rInfo = NumFor[nIx].Info();
+    const sal_uInt16 nAnz = NumFor[nIx].GetCount();
+    OUStringBuffer sStr, sFrac, sDiv; // Strings, value for Integral part Numerator and denominator
+    bool bSign = ( (fNumber < 0) && (nIx == 0) ); // sign Not in the ones at the end
+    const OUString sIntegerFormat = lcl_GetFractionIntegerString(rInfo, nAnz);
+    const OUString sNumeratorFormat = lcl_GetNumeratorString(rInfo, nAnz);
+    const OUString sDenominatorFormat = lcl_GetDenominatorString(rInfo, nAnz);
+
+    sal_uInt64 nFrac = 0, nDiv = 1;
+    double fNum = floor(fNumber); // Integral part
+
+    if (fNum > D_MAX_U_INT32 || rInfo.nCntExp > 9) // Too large
+    {
+        sBuff = rScan.GetErrorString();
+        return false;
+    }
+    if (rInfo.nCntExp == 0)
+    {
+        SAL_WARN( "svl.numbers", "SvNumberformat:: Fraction, nCntExp == 0");
+        sBuff.truncate();
+        return false;
+    }
+
+    ImpGetFractionElements( fNumber, nIx, fNum, nFrac, nDiv);
 
     if (rInfo.nCntPre == 0) // Improper fraction
     {