public override void RemoveMetadata(TempFileForActions tempFile, CleanActionPropertySet cleanProperties)
{
try
{
m_elementsToClean = GetRationalizedElementsToClean(GetListOfEnabledElementsToClean(cleanProperties), CleanActionPropertySet.TargetApplication.Word);
List<Exclusion> listExclusion = GetListOfExcludedElements(cleanProperties);
using (BinaryData bData = new BinaryData(tempFile.GetMemoryStream()))
{
using (DocxDocumentReader reader = new DocxDocumentReader(bData))
{
using (Stream str = GetOutputStream())
{
reader.CleanTo(str, m_elementsToClean, listExclusion);
File.Copy(InterimTempFileName, tempFile.TempFile, true);
}
}
}
}
catch (System.Exception ex)
{
Logger.LogError("WordX cleaning failed");
Logger.LogError(ex);
throw;
}
finally
{
CleanUp();
}
}
private void ExecuteUsingOfficeOpenXml(TempFileForActions tempFile, CleanActionPropertySet elementsToClean)
{
try
{
m_elementsToClean = GetElementsNotRemovedByDomClean(GetListOfEnabledElementsToClean(elementsToClean));
using (BinaryData bData = new BinaryData(tempFile.GetMemoryStream()))
{
using (DocxDocumentReader reader = new DocxDocumentReader(bData))
{
using (Stream str = GetOutputStream())
{
List<Exclusion> excludedElements = GetListOfExcludedElements(elementsToClean);
reader.CleanTo(str, m_elementsToClean, excludedElements);
File.Copy(InterimTempFileName, tempFile.TempFile, true);
}
}
}
}
catch (System.Exception ex)
{
Logger.LogError(ex);
throw;
}
finally
{
CleanUp();
}
}
public override void Execute(WorkItem workItem)
{
try
{
switch (workItem.Extension)
{
case "doc":
using (DocxDocumentReader reader = new DocxDocumentReader(workItem.ConvertedFileName))
{
workItem.discoveryTimer.Start();
DocumentText results = reader.Read();
workItem.DiscoveredText = results;
workItem.FileSize += new FileInfo(workItem.ConvertedFileName).Length;
}
break;
case "ppt":
using (PptxDocumentReader reader = new PptxDocumentReader(workItem.ConvertedFileName))
{
workItem.discoveryTimer.Start();
DocumentText results = reader.Read();
workItem.DiscoveredText = results;
workItem.FileSize += new FileInfo(workItem.ConvertedFileName).Length;
}
break;
case "xls":
using (XlsxDocumentReader reader = new XlsxDocumentReader(workItem.ConvertedFileName))
{
workItem.discoveryTimer.Start();
DocumentText results = reader.Read();
workItem.DiscoveredText = results;
workItem.FileSize += new FileInfo(workItem.ConvertedFileName).Length;
}
break;
default:
throw new Exception("File extension not expected : " + workItem.Extension);
}
}
catch (Exception e)
{
workItem.FailureMode = FailureMode.DiscoveryFailed;
workItem.ShouldAbort = true;
throw e;
}
finally
{
workItem.discoveryTimer.Stop();
}
}
public override bool VerifyFile(TempFileForActions tempFile)
{
try
{
using (BinaryData bData = new BinaryData(tempFile.GetMemoryStream()))
{
using (DocxDocumentReader reader = new DocxDocumentReader(bData))
{
reader.Read();
}
}
return true;
}
catch (System.Exception ex)
{
Logger.LogError("WordX Verification failed");
Logger.LogError(ex);
}
return false;
}
/// <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;
}
DocumentText DiscoverDocument(string sFilename)
{
switch (m_ft)
{
case Workshare.Policy.FileType.WordDocument:
using (WordDocumentReader Word2003Reader1 = new WordDocumentReader(sFilename, false))
{
return Word2003Reader1.Read();
}
case Workshare.Policy.FileType.ExcelSheet:
using (ExcelDocumentReader Excel2003Reader1 = new ExcelDocumentReader(sFilename, false))
{
return Excel2003Reader1.Read();
}
case Workshare.Policy.FileType.PowerPoint:
using (Workshare.FCS.Lite.PptDocumentReader Ppt2003Reader1 = new PptDocumentReader(sFilename, false))
{
return Ppt2003Reader1.Read();
}
case Workshare.Policy.FileType.WordDocumentX:
case Workshare.Policy.FileType.WordDocumentMacroX:
case Workshare.Policy.FileType.WordDocumentTemplateX:
case Workshare.Policy.FileType.WordDocumentMacroTemplateX:
using (DocxDocumentReader Word2007Reader1 = new DocxDocumentReader(sFilename))
{
return Word2007Reader1.Read();
}
case Workshare.Policy.FileType.ExcelSheetX:
case Workshare.Policy.FileType.ExcelSheetMacroX:
case Workshare.Policy.FileType.ExcelSheetTemplateX:
case Workshare.Policy.FileType.ExcelSheetMacroTemplateX:
using (XlsxDocumentReader Excel2007Reader1 = new XlsxDocumentReader(sFilename))
{
return Excel2007Reader1.Read();
}
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:
using (PptxDocumentReader Ppt2007Reader1 = new PptxDocumentReader(sFilename))
{
return Ppt2007Reader1.Read();
}
default:
break;
}
return null;
}
public override void Execute(WorkItem workItem)
{
try
{
switch (workItem.Extension)
{
case "doc":
using (DocxDocumentReader reader = new DocxDocumentReader(workItem.CleanedFileName))
{
DocumentText results = reader.Read();
//workItem.FileSize += new FileInfo(workItem.ConvertedFileName).Length;
foreach (IAbstractTextType tt in results.GetTextTypes())
{
if (tt.GetContentType() == ContentType.WorkshareProperty)
continue; // We explicitly never clean these
if (tt.GetChildCount() > 0 && (tt.GetContentType() != ContentType.Paragraph))
{
workItem.Info = DumpTextType(tt);
workItem.ContentType = tt.GetContentType();
throw new Exception("Unexpected content type found in cleaned doc");
}
}
}
break;
case "ppt":
using (PptxDocumentReader reader = new PptxDocumentReader(workItem.CleanedFileName))
{
DocumentText results = reader.Read();
//workItem.FileSize += new FileInfo(workItem.ConvertedFileName).Length;
foreach (IAbstractTextType tt in results.GetTextTypes())
{
if (tt.GetContentType() == ContentType.TextBox)
continue;
if (tt.GetContentType() == ContentType.Paragraph)
continue;
if (tt.GetChildCount() > 0 )
{
workItem.Info = DumpTextType(tt);
workItem.ContentType = tt.GetContentType();
throw new Exception("Unexpected content type found in cleaned ppt");
}
}
}
break;
case "xls":
using (XlsxDocumentReader reader = new XlsxDocumentReader(workItem.CleanedFileName))
{
DocumentText results = reader.Read();
//workItem.FileSize += new FileInfo(workItem.ConvertedFileName).Length;
foreach (IAbstractTextType tt in results.GetTextTypes())
{
if (tt.GetContentType() == ContentType.RoutingSlip)
continue;
if (tt.GetChildCount() > 0 && (tt.GetContentType() != ContentType.CellText))
{
workItem.Info = DumpTextType(tt);
workItem.ContentType = tt.GetContentType();
throw new Exception("Unexpected content type found in cleaned xls");
}
}
}
break;
default:
throw new Exception("File extension not expected : " + workItem.Extension);
}
}
catch (Exception e)
{
workItem.FailureMode = FailureMode.ThingsNotCleaned;
throw e;
}
}
/// <summary>
/// </summary>
/// <returns>A FCS.Lite.DocumentText object</returns>
public FCS.Lite.Base.DocumentText AdvancedRead()
{
if (m_docText == null)
{
FCS.Lite.Base.DocumentReader docReader;
switch (m_fileData.FileType)
{
case FileType.WordDocument:
docReader = new WordDocumentReader(m_fileData.BinaryFileData);
break;
case FileType.PowerPoint:
docReader = new PptDocumentReader(m_fileData.BinaryFileData);
break;
case FileType.ExcelSheet:
docReader = new ExcelDocumentReader(m_fileData.BinaryFileData);
break;
case FileType.WordDocumentX:
case FileType.WordDocumentMacroX:
case FileType.WordDocumentTemplateX:
case FileType.WordDocumentMacroTemplateX:
docReader = new DocxDocumentReader(m_fileData.BinaryFileData);
break;
case FileType.PowerPointX:
case FileType.PowerPointMacroX:
case FileType.PowerPointTemplateX:
case FileType.PowerPointMacroTemplateX:
case FileType.PowerPointShowX:
case FileType.PowerPointMacroShowX:
docReader = new PptxDocumentReader(m_fileData.BinaryFileData);
break;
case FileType.ExcelSheetX:
case FileType.ExcelSheetMacroX:
case FileType.ExcelSheetTemplateX:
case FileType.ExcelSheetMacroTemplateX:
docReader = new XlsxDocumentReader(m_fileData.BinaryFileData);
break;
case FileType.PDFDocument:
docReader = new PdfDocumentReader(m_fileData.BinaryFileData);
break;
default:
return m_docText;
}
using (docReader)
{
m_docText = docReader.Read();
}
}
return m_docText;
}
