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()))
{
XlsxDocumentReader reader = new XlsxDocumentReader(bData);
reader.Read();
}
return true;
}
catch (System.Exception ex)
{
Logger.LogError("ExcelX Verification failed");
Logger.LogError(ex);
}
return false;
}
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;
}
}