Initial step on enabling the orcus-based Excel 2003 XML filter.

Still some work remains in the orcus interface implementation code
in sc.

Change-Id: Idc4d95b489e8e55fbe3593557399610abe6c4ae0
Reviewed-on: https://gerrit.libreoffice.org/46655
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Kohei Yoshida <libreoffice@kohei.us>

5 files changed, 54 insertions, 2 deletions

diff --git a/sc/inc/orcusfilters.hxx b/sc/inc/orcusfilters.hxx
index e96facbaa6e3..1f96d6ac1769 100644
--- a/sc/inc/orcusfilters.hxx
+++ b/sc/inc/orcusfilters.hxx
@@ -31,6 +31,8 @@ public:
 
     virtual bool importGnumeric(ScDocument& rDoc, SfxMedium& rMedium) const = 0;
 
+    virtual bool importExcel2003XML(ScDocument& rDoc, SfxMedium& rMedium) const = 0;
+
     virtual bool importXLSX(ScDocument& rDoc, SfxMedium& rMedium) const = 0;
 
     virtual bool importODS(ScDocument& rDoc, SfxMedium& rMedium) const = 0;
diff --git a/sc/source/filter/inc/orcusfiltersimpl.hxx b/sc/source/filter/inc/orcusfiltersimpl.hxx
index ee69df00b552..48011504fc5b 100644
--- a/sc/source/filter/inc/orcusfiltersimpl.hxx
+++ b/sc/source/filter/inc/orcusfiltersimpl.hxx
@@ -21,6 +21,7 @@ public:
 
     virtual bool importCSV(ScDocument& rDoc, SfxMedium& rMedium) const override;
     virtual bool importGnumeric(ScDocument& rDoc, SfxMedium& rMedium) const override;
+    virtual bool importExcel2003XML(ScDocument& rDoc, SfxMedium& rMedium) const override;
     virtual bool importXLSX(ScDocument& rDoc, SfxMedium& rMedium) const override;
     virtual bool importODS(ScDocument& rDoc, SfxMedium& rMedium) const override;
 
diff --git a/sc/source/filter/orcus/filterdetect.cxx b/sc/source/filter/orcus/filterdetect.cxx
index 137585384e48..32bac167063f 100644
--- a/sc/source/filter/orcus/filterdetect.cxx
+++ b/sc/source/filter/orcus/filterdetect.cxx
@@ -80,8 +80,16 @@ OUString OrcusFormatDetect::detect(css::uno::Sequence<css::beans::PropertyValue>
     }
 
     orcus::format_t eFormat = orcus::detect(reinterpret_cast<const unsigned char*>(aContent.getStr()), aContent.getLength());
-    if (eFormat == orcus::format_t::gnumeric)
-        return OUString("Gnumeric XML");
+
+    switch (eFormat)
+    {
+        case orcus::format_t::gnumeric:
+            return OUString("Gnumeric XML");
+        case orcus::format_t::xls_xml:
+            return OUString("calc_MS_Excel_2003_XML");
+        default:
+            ;
+    }
 
     return OUString();
 }
diff --git a/sc/source/filter/orcus/orcusfiltersimpl.cxx b/sc/source/filter/orcus/orcusfiltersimpl.cxx
index 2aa7e8b30dc0..6fc115bdce29 100644
--- a/sc/source/filter/orcus/orcusfiltersimpl.cxx
+++ b/sc/source/filter/orcus/orcusfiltersimpl.cxx
@@ -25,6 +25,7 @@
 #include <orcus/orcus_csv.hpp>
 #include <orcus/orcus_gnumeric.hpp>
 #include <orcus/orcus_xlsx.hpp>
+#include <orcus/orcus_xls_xml.hpp>
 #include <orcus/orcus_ods.hpp>
 #include <orcus/orcus_import_ods.hpp>
 #include <orcus/global.hpp>
@@ -115,6 +116,38 @@ bool ScOrcusFiltersImpl::importGnumeric(ScDocument& rDoc, SfxMedium& rMedium) co
     return true;
 }
 
+bool ScOrcusFiltersImpl::importExcel2003XML(ScDocument& rDoc, SfxMedium& rMedium) const
+{
+    ScOrcusFactory aFactory(rDoc);
+    aFactory.setStatusIndicator(getStatusIndicator(rMedium));
+    SvStream* pStream = rMedium.GetInStream();
+    pStream->Seek(0);
+    static const size_t nReadBuffer = 1024*32;
+    OStringBuffer aBuffer((int(nReadBuffer)));
+    size_t nRead = 0;
+    do
+    {
+        char pData[nReadBuffer];
+        nRead = pStream->ReadBytes(pData, nReadBuffer);
+        aBuffer.append(static_cast<sal_Char*>(pData), nRead);
+    }
+    while (nRead == nReadBuffer);
+
+    try
+    {
+        rDoc.ClearTabs();
+        orcus::orcus_xls_xml filter(&aFactory);
+        filter.read_stream(aBuffer.getStr(), aBuffer.getLength());
+    }
+    catch (const std::exception& e)
+    {
+        SAL_WARN("sc", "Unable to load Excel 2003 XML file! " << e.what());
+        return false;
+    }
+
+    return true;
+}
+
 bool ScOrcusFiltersImpl::importXLSX(ScDocument& rDoc, SfxMedium& rMedium) const
 {
     ScOrcusFactory aFactory(rDoc);
diff --git a/sc/source/ui/docshell/docsh.cxx b/sc/source/ui/docshell/docsh.cxx
index 98e3e5b51eaa..6f1ba210ff4b 100644
--- a/sc/source/ui/docshell/docsh.cxx
+++ b/sc/source/ui/docshell/docsh.cxx
@@ -1169,6 +1169,14 @@ bool ScDocShell::ConvertFrom( SfxMedium& rMedium )
 
             bRet = pOrcus->importGnumeric(aDocument, rMedium);
         }
+        else if (aFltName == "MS Excel 2003 XML Orcus")
+        {
+            ScOrcusFilters* pOrcus = ScFormatFilter::Get().GetOrcusFilters();
+            if (!pOrcus)
+                return false;
+
+            bRet = pOrcus->importExcel2003XML(aDocument, rMedium);
+        }
         else if (aFltName == pFilterAscii)
         {
             SfxItemSet*  pSet = rMedium.GetItemSet();