tdf#82254 - Don't remove UTF-8 BOM from CSV when saving file

Don't remove the byte-order-mark in the resulting CSV file
when it was present in the CSV source file.

Change-Id: Id26abad2686917f320f2ace85441621bcf57ea9e
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/145879
Tested-by: Jenkins
Reviewed-by: Eike Rathke <erack@redhat.com>

2 files changed, 20 insertions, 4 deletions

diff --git a/sc/source/ui/docshell/docsh.cxx b/sc/source/ui/docshell/docsh.cxx
index f37b969bd344..4bf9343734a7 100644
--- a/sc/source/ui/docshell/docsh.cxx
+++ b/sc/source/ui/docshell/docsh.cxx
@@ -1301,6 +1301,15 @@ bool ScDocShell::ConvertFrom( SfxMedium& rMedium )
                     sc::SetFormulaDirtyContext aCxt;
                     m_pDocument->SetAllFormulasDirty(aCxt);
 
+                    // tdf#82254 - check whether to include a byte-order-mark in the output
+                    if (const bool bIncludeBOM = aImpEx.GetIncludeBOM())
+                    {
+                        aOptions.SetIncludeBOM(bIncludeBOM);
+                        if (rMedium.GetItemSet() != nullptr)
+                            rMedium.GetItemSet()->Put(
+                                SfxStringItem(SID_FILE_FILTEROPTIONS, aOptions.WriteToString()));
+                    }
+
                     // for mobile case, we use a copy of the original document and give it a temporary name before editing
                     // Therefore, the sheet name becomes ugly, long and nonsensical.
 #if !(defined ANDROID)
@@ -1939,6 +1948,7 @@ void ScDocShell::AsciiSave( SvStream& rStream, const ScImportOptions& rAsciiOpt,
     bool bSaveNumberAsSuch = rAsciiOpt.bSaveNumberAsSuch;
     bool bSaveAsShown     = rAsciiOpt.bSaveAsShown;
     bool bShowFormulas    = rAsciiOpt.bSaveFormulas;
+    bool bIncludeBOM      = rAsciiOpt.bIncludeBOM;
 
     rtl_TextEncoding eOldCharSet = rStream.GetStreamCharSet();
     rStream.SetStreamCharSet( eCharSet );
@@ -1955,6 +1965,9 @@ void ScDocShell::AsciiSave( SvStream& rStream, const ScImportOptions& rAsciiOpt,
     }
     else
     {
+        // tdf#82254 - check whether to include a byte-order-mark in the output
+        if (bIncludeBOM && eCharSet == RTL_TEXTENCODING_UTF8)
+            rStream.WriteUChar(0xEF).WriteUChar(0xBB).WriteUChar(0xBF);
         aStrDelimEncoded = OString(&cStrDelim, 1, eCharSet);
         aDelimEncoded = OString(&cDelim, 1, eCharSet);
         rtl_TextEncodingInfo aInfo;
diff --git a/sc/source/ui/docshell/impex.cxx b/sc/source/ui/docshell/impex.cxx
index 1b8d715890b9..87735c2087a3 100644
--- a/sc/source/ui/docshell/impex.cxx
+++ b/sc/source/ui/docshell/impex.cxx
@@ -114,7 +114,7 @@ ScImportExport::ScImportExport( ScDocument& r )
       bFormulas( false ), bIncludeFiltered( true ),
       bAll( true ), bSingle( true ), bUndo( false ),
       bOverflowRow( false ), bOverflowCol( false ), bOverflowCell( false ),
-      mbApi( true ), mbImportBroadcast(false), mbOverwriting( false )
+      mbApi( true ), mbImportBroadcast(false), mbOverwriting( false ), mbIncludeBOM(false)
 {
     pUndoDoc = nullptr;
     pExtOptions = nullptr;
@@ -129,7 +129,7 @@ ScImportExport::ScImportExport( ScDocument& r, const ScAddress& rPt )
       bFormulas( false ), bIncludeFiltered( true ),
       bAll( false ), bSingle( true ), bUndo( pDocSh != nullptr ),
       bOverflowRow( false ), bOverflowCol( false ), bOverflowCell( false ),
-      mbApi( true ), mbImportBroadcast(false), mbOverwriting( false )
+      mbApi( true ), mbImportBroadcast(false), mbOverwriting( false ), mbIncludeBOM(false)
 {
     pUndoDoc = nullptr;
     pExtOptions = nullptr;
@@ -145,7 +145,7 @@ ScImportExport::ScImportExport( ScDocument& r, const ScRange& rRange )
       bFormulas( false ), bIncludeFiltered( true ),
       bAll( false ), bSingle( false ), bUndo( pDocSh != nullptr ),
       bOverflowRow( false ), bOverflowCol( false ), bOverflowCell( false ),
-      mbApi( true ), mbImportBroadcast(false), mbOverwriting( false )
+      mbApi( true ), mbImportBroadcast(false), mbOverwriting( false ), mbIncludeBOM(false)
 {
     pUndoDoc = nullptr;
     pExtOptions = nullptr;
@@ -162,7 +162,7 @@ ScImportExport::ScImportExport( ScDocument& r, const OUString& rPos )
       bFormulas( false ), bIncludeFiltered( true ),
       bAll( false ), bSingle( true ), bUndo( pDocSh != nullptr ),
       bOverflowRow( false ), bOverflowCol( false ), bOverflowCell( false ),
-      mbApi( true ), mbImportBroadcast(false), mbOverwriting( false )
+      mbApi( true ), mbImportBroadcast(false), mbOverwriting( false ), mbIncludeBOM(false)
 {
     pUndoDoc = nullptr;
     pExtOptions = nullptr;
@@ -1577,6 +1577,9 @@ bool ScImportExport::ExtText2Doc( SvStream& rStrm )
     std::unique_ptr<ScProgress> xProgress( new ScProgress( pDocSh,
             ScResId( STR_LOAD_DOC ), nRemaining, true ));
     rStrm.StartReadingUnicodeText( rStrm.GetStreamCharSet() );
+    // tdf#82254 - check whether to include a byte-order-mark in the output
+    if (nOldPos != rStrm.Tell())
+        mbIncludeBOM = true;
 
     SCCOL nStartCol = aRange.aStart.Col();
     SCCOL nEndCol = aRange.aEnd.Col();