protected void TestCleanTextTypeFromDocument(string fileName, ContentType contentType, string[] textThatShouldntBeCleaned)
{
fileName = TESTFILE_DIR + fileName;
string outFileName = fileName + "_cleaned.doc";
File.Copy(fileName, outFileName, true);
using (Workshare.FCS.Lite.WordDocumentReader docReader = new WordDocumentReader(outFileName, true))
{
docReader.Clean(BuildContentTypeList(contentType));
}
using (WordDocumentReader docReader2 = new WordDocumentReader(outFileName, false))
{
DocumentText dt = docReader2.Read();
IAbstractTextType tt = dt.GetUniqueTextType(contentType);
IAbstractTextType ttHid = dt.GetUniqueTextType(ContentType.HiddenText);
if (tt == null && ttHid == null)
{
CheckHaventCleanedTooMuch(dt.GetUniqueTextType(ContentType.Paragraph), textThatShouldntBeCleaned);
return;
}
if (tt != null)
Assert.AreEqual(0, tt.Nodes.Count, "Expected all the items of type " + contentType.ToString() + " to have been cleaned");
if (ttHid != null)
Assert.AreEqual(0, ttHid.Nodes.Count, "Expected no significant hidden text to show up");
}
}
private void TestCleanTCsFromDocument(string fileName, bool checkHiddenText)
{
string outFileName = fileName + "_cleaned.doc";
File.Copy(fileName, outFileName, true);
using (Workshare.FCS.Lite.WordDocumentReader docReader = new WordDocumentReader(outFileName, true))
{
docReader.Clean(BuildContentTypeList(ContentType.TrackChange));
}
using (WordDocumentReader docReader2 = new WordDocumentReader(outFileName, false))
{
DocumentText dt = docReader2.Read();
IAbstractTextType tt = dt.GetUniqueTextType(ContentType.TrackChange);
IAbstractTextType ttHid = dt.GetUniqueTextType(ContentType.HiddenText);
if (tt == null && ttHid == null)
return;
if (tt != null)
Assert.AreEqual(0, tt.Nodes.Count, "Expected all the track changes to have been cleaned");
if (checkHiddenText)
{
if (ttHid != null)
Assert.AreEqual(0, ttHid.Nodes.Count, "Expected no significant hidden text to show up");
}
}
}
private void CleanFile(string fileName, List<ContentType> listContentTypes, List<Exclusion> listExclusion)
{
try
{
using (WordDocumentReader wrdReader = new WordDocumentReader(fileName, true))
{
wrdReader.Clean(listContentTypes, listExclusion);
}
}
catch (System.Exception ex)
{
Logger.LogError("Word Binary cleaning failed");
Logger.LogError(ex);
throw;
}
}
/// <summary>
/// Original incarnation of lightspeed clean method. Left for comparison. There is at least one document that gets corrupted by the
/// lightspeed clean user action as opposed to the method below. see Rally DE8912 in Professional & Deltaview project.
/// </summary>
/// <param name="listContentTypes"></param>
private void DoLightSpeedClean(List<ContentType> listContentTypes)
{
System.Diagnostics.Stopwatch watch = new System.Diagnostics.Stopwatch();
watch.Start();
switch (m_ft)
{
case Workshare.Policy.FileType.WordDocument:
using (WordDocumentReader Word2003Reader1 = new WordDocumentReader(m_sFileForBinClean, true))
{
Word2003Reader1.Clean(listContentTypes);
}
break;
case Workshare.Policy.FileType.ExcelSheet:
using (ExcelDocumentReader Excel2003Reader1 = new ExcelDocumentReader(m_sFileForBinClean, true))
{
Excel2003Reader1.Clean(listContentTypes);
}
break;
case Workshare.Policy.FileType.PowerPoint:
using (Workshare.FCS.Lite.PptDocumentReader Ppt2003Reader1 = new PptDocumentReader(m_sFileForBinClean, true))
{
Ppt2003Reader1.Clean(listContentTypes);
}
break;
case Workshare.Policy.FileType.WordDocumentX:
case Workshare.Policy.FileType.WordDocumentMacroX:
case Workshare.Policy.FileType.WordDocumentTemplateX:
case Workshare.Policy.FileType.WordDocumentMacroTemplateX:
{
string outFileName = System.IO.Path.GetTempFileName();
using (DocxDocumentReader Word2007Reader1 = new DocxDocumentReader(m_sFileForBinClean))
{
using (Stream outStr = File.Open(outFileName, FileMode.Create))
{
Word2007Reader1.CleanTo(outStr, listContentTypes);
}
}
File.Copy(outFileName, m_sFileForBinClean, true);
File.Delete(outFileName);
break;
}
case Workshare.Policy.FileType.ExcelSheetX:
case Workshare.Policy.FileType.ExcelSheetMacroX:
case Workshare.Policy.FileType.ExcelSheetTemplateX:
case Workshare.Policy.FileType.ExcelSheetMacroTemplateX:
{
string outFileName = System.IO.Path.GetTempFileName();
using (XlsxDocumentReader Excel2007Reader1 = new XlsxDocumentReader(m_sFileForBinClean))
{
using (Stream outStr = File.Open(outFileName, FileMode.Create))
{
Excel2007Reader1.CleanTo(outStr, listContentTypes);
}
}
File.Copy(outFileName, m_sFileForBinClean, true);
File.Delete(outFileName);
break;
}
case Workshare.Policy.FileType.PowerPointX:
case Workshare.Policy.FileType.PowerPointMacroX:
case Workshare.Policy.FileType.PowerPointTemplateX:
case Workshare.Policy.FileType.PowerPointMacroTemplateX:
case Workshare.Policy.FileType.PowerPointShowX:
case Workshare.Policy.FileType.PowerPointMacroShowX:
{
string outFileName = System.IO.Path.GetTempFileName();
using (PptxDocumentReader Ppt2007Reader1 = new PptxDocumentReader(m_sFileForBinClean))
{
using (Stream outStr = File.Open(outFileName, FileMode.Create))
{
Ppt2007Reader1.CleanTo(outStr, listContentTypes);
}
}
File.Copy(outFileName, m_sFileForBinClean, true);
File.Delete(outFileName);
break;
}
default:
break;
}
watch.Stop();
m_binCleanTime = watch.Elapsed.TotalSeconds;
}
public void DoBinaryClean(List<ContentType> listContentTypes)
{
switch (m_ft)
{
case Workshare.Policy.FileType.WordDocument:
using (WordDocumentReader Word2003Reader1 = new WordDocumentReader(m_filenameIn, true))
{
Word2003Reader1.Clean(listContentTypes);
}
break;
case Workshare.Policy.FileType.ExcelSheet:
using (ExcelDocumentReader Excel2003Reader1 = new ExcelDocumentReader(m_filenameIn, true))
{
Excel2003Reader1.Clean(listContentTypes);
}
break;
case Workshare.Policy.FileType.PowerPoint:
using (Workshare.FCS.Lite.PptDocumentReader Ppt2003Reader1 = new PptDocumentReader(m_filenameIn, true))
{
Ppt2003Reader1.Clean(listContentTypes);
}
break;
default:
break;
}
}
