tdf#35208 Allow choosing encoding for old Excel files

Lots of BIFF2-BIFF5 files out there don't have CODEPAGE
record at all, or have one with a wrong value, and/or
FONT records with wrong charset value.

To solve that, this patch adds a new "Choose Encoding"
entry to the file picker, so that users could specify the
encoding to use for a given file, instead of relying on
the information that might be present (or not) inside the
file. It can be also used in headless mode,
e.g. --infilter="MS Excel (encoded)":61

This doesn't affect BIFF8 import, nor the default behavior
when not explicitly using the "Choose Encoding" entry.

Change-Id: I89d850d7679b81bd399044478fac7a02e8b7680e

5 files changed, 36 insertions, 3 deletions

diff --git a/sc/source/ui/docshell/docsh.cxx b/sc/source/ui/docshell/docsh.cxx
index de05c84cc7cc..599fce43d954 100644
--- a/sc/source/ui/docshell/docsh.cxx
+++ b/sc/source/ui/docshell/docsh.cxx
@@ -153,6 +153,7 @@ static const sal_Char pFilterExcel95[]  = "MS Excel 95";
 static const sal_Char pFilterEx95Temp[] = "MS Excel 95 Vorlage/Template";
 static const sal_Char pFilterExcel97[]  = "MS Excel 97";
 static const sal_Char pFilterEx97Temp[] = "MS Excel 97 Vorlage/Template";
+static const sal_Char pFilterExcelEnc[] = "MS Excel (encoded)";
 static const sal_Char pFilterDBase[]        = "dBase";
 static const sal_Char pFilterDif[]      = "DIF";
 static const sal_Char pFilterSylk[]     = "SYLK";
@@ -1110,7 +1111,8 @@ bool ScDocShell::ConvertFrom( SfxMedium& rMedium )
             bSetColWidths = true;
             bSetRowHeights = true;
         }
-        else if ( aFltName == pFilterExcel4 || aFltName == pFilterExcel5 ||
+        else if (  aFltName == pFilterExcelEnc ||
+                   aFltName == pFilterExcel4 || aFltName == pFilterExcel5 ||
                    aFltName == pFilterExcel95 || aFltName == pFilterExcel97 ||
                    aFltName == pFilterEx4Temp || aFltName == pFilterEx5Temp ||
                    aFltName == pFilterEx95Temp || aFltName == pFilterEx97Temp )
@@ -1124,9 +1126,24 @@ bool ScDocShell::ConvertFrom( SfxMedium& rMedium )
             else if ( aFltName == pFilterExcel97 || aFltName == pFilterEx97Temp )
                 eFormat = EIF_BIFF8;
 
+            rtl_TextEncoding eTextEnc = RTL_TEXTENCODING_DONTKNOW;
+            if ( eFormat != EIF_BIFF8 )
+            {
+                OUString sItStr;
+                SfxItemSet* pSet = rMedium.GetItemSet();
+                const SfxPoolItem* pItem;
+                if ( pSet && SfxItemState::SET ==
+                     pSet->GetItemState( SID_FILE_FILTEROPTIONS, true, &pItem ) )
+                {
+                    sItStr = static_cast<const SfxStringItem*>(pItem)->GetValue();
+                    if ( !sItStr.isEmpty() )
+                        eTextEnc = ScGlobal::GetCharsetValue( sItStr );
+                }
+            }
+
             MakeDrawLayer(); //! In the filter
             CalcOutputFactor(); // prepare update of row height
-            FltError eError = ScFormatFilter::Get().ScImportExcel( rMedium, &aDocument, eFormat );
+            FltError eError = ScFormatFilter::Get().ScImportExcel( rMedium, &aDocument, eFormat, eTextEnc );
             aDocument.UpdateFontCharSet();
             if ( aDocument.IsChartListenerCollectionNeedsUpdate() )
                 aDocument.UpdateChartListenerCollection(); //! For all imports?
@@ -2598,6 +2615,11 @@ OUString ScDocShell::GetDifFilterName()
     return OUString(pFilterDif);
 }
 
+OUString ScDocShell::GetExcelEncodingFilterName()
+{
+    return OUString(pFilterExcelEnc);
+}
+
 bool ScDocShell::HasAutomaticTableName( const OUString& rFilter )
 {
     //  sal_True for those filters that keep the default table name
diff --git a/sc/source/ui/docshell/impex.cxx b/sc/source/ui/docshell/impex.cxx
index 47a796414ce8..9b23831e1c99 100644
--- a/sc/source/ui/docshell/impex.cxx
+++ b/sc/source/ui/docshell/impex.cxx
@@ -2217,7 +2217,7 @@ class ScFormatFilterMissing : public ScFormatFilterPlugin {
     virtual ~ScFormatFilterMissing() {}
     virtual FltError ScImportLotus123( SfxMedium&, ScDocument*, rtl_TextEncoding ) override { return eERR_INTERN; }
     virtual FltError ScImportQuattroPro( SfxMedium &, ScDocument * ) override { return eERR_INTERN; }
-    virtual FltError ScImportExcel( SfxMedium&, ScDocument*, const EXCIMPFORMAT ) override { return eERR_INTERN; }
+    virtual FltError ScImportExcel( SfxMedium&, ScDocument*, const EXCIMPFORMAT, rtl_TextEncoding ) override { return eERR_INTERN; }
     virtual FltError ScImportStarCalc10( SvStream&, ScDocument* ) override { return eERR_INTERN; }
     virtual FltError ScImportDif( SvStream&, ScDocument*, const ScAddress&,
                  const rtl_TextEncoding, sal_uInt32 ) override { return eERR_INTERN; }
diff --git a/sc/source/ui/inc/docsh.hxx b/sc/source/ui/inc/docsh.hxx
index 079b542e68dc..914f369d6ba1 100644
--- a/sc/source/ui/inc/docsh.hxx
+++ b/sc/source/ui/inc/docsh.hxx
@@ -398,6 +398,7 @@ public:
     static OUString   GetLotusFilterName();
     static OUString   GetDBaseFilterName();
     static OUString   GetDifFilterName();
+    static OUString   GetExcelEncodingFilterName();
     static bool       HasAutomaticTableName( const OUString& rFilter );
 
     DECL_LINK_TYPED( RefreshDBDataHdl, Timer*, void );
diff --git a/sc/source/ui/src/globstr.src b/sc/source/ui/src/globstr.src
index c74373f662b3..f026a13e5640 100644
--- a/sc/source/ui/src/globstr.src
+++ b/sc/source/ui/src/globstr.src
@@ -2109,6 +2109,10 @@ Resource RID_GLOBSTR
     {
         Text [ en-US ] = "%1 and %2 more";
     };
+    String STR_IMPORT_EXCEL_ENCODING
+    {
+        Text [ en-US ] = "Import Excel file";
+    };
 };
 
 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */
diff --git a/sc/source/ui/unoobj/filtuno.cxx b/sc/source/ui/unoobj/filtuno.cxx
index 96f74ce7b949..eac4e805da4c 100644
--- a/sc/source/ui/unoobj/filtuno.cxx
+++ b/sc/source/ui/unoobj/filtuno.cxx
@@ -274,6 +274,12 @@ sal_Int16 SAL_CALL ScFilterOptionsObj::execute() throw(uno::RuntimeException, st
             // common for DIF import/export
             eEncoding = RTL_TEXTENCODING_MS_1252;
         }
+        else if ( aFilterString == ScDocShell::GetExcelEncodingFilterName() )
+        {
+            SAL_WARN_IF( bExport, "sc.ui", "Filter Options for Excel Export is not implemented" );
+            aTitle = ScGlobal::GetRscString( STR_IMPORT_EXCEL_ENCODING );
+            eEncoding = RTL_TEXTENCODING_MS_1252;
+        }
 
         ScImportOptions aOptions( cAsciiDel, cStrDel, eEncoding);