public override void Execute(WorkItem workItem)
{
try
{
switch (workItem.Extension)
{
case "doc":
ExecuteWithWord(workItem);
break;
case "ppt":
ExecuteWithPowerpoint(workItem);
break;
case "xls":
ExecuteWithExcel(workItem);
break;
default:
throw new Exception("File extension not expected : " + workItem.Extension);
}
}
catch (Exception e)
{
workItem.FailureMode = FailureMode.DocumentCorruption;
throw e;
}
}
public override void Execute(WorkItem workItem)
{
try
{
m_workItem = workItem;
DocumentText dtCheck;
switch (workItem.Extension)
{
case "doc":
WordDocumentReader wrdRdr = new WordDocumentReader(workItem.BaseFileName);
try
{
dtCheck = wrdRdr.Read();
}
catch (Exception)
{
return; // failed to discover source document
}
CompareDiscoveryForDoc(workItem.BaseFileName, workItem.DiscoveredText, dtCheck);
break;
case "ppt":
PptDocumentReader pptRdr = new PptDocumentReader(workItem.BaseFileName);
try
{
dtCheck = pptRdr.Read();
}
catch (Exception)
{
return; // failed to discover source document
}
CompareDiscoveryForPpt(workItem.BaseFileName, workItem.DiscoveredText, dtCheck);
break;
case "xls":
ExcelDocumentReader xlsRdr = new ExcelDocumentReader(workItem.BaseFileName);
try
{
dtCheck = xlsRdr.Read();
}
catch (Exception)
{
return; // failed to discover source document
}
CompareDiscoveryForXls(workItem.BaseFileName, workItem.DiscoveredText, dtCheck);
break;
default:
throw new Exception("File extension not expected : " + workItem.Extension);
}
}
catch (Exception e)
{
workItem.FailureMode = FailureMode.DiscoveryMismatch;
throw e;
}
}
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();
}
}
private void ExecuteWithExcel(WorkItem workItem)
{
try
{
ExcelUtils.CheckFileValid(workItem.CleanedFileName);
}
catch (Exception e)
{
if (e.Message.StartsWith("The remote procedure call failed"))
{
ExcelUtils.DiscardExcelApp();
}
if (e.Message.StartsWith("The RPC server is unavailable"))
{
ExcelUtils.DiscardExcelApp();
}
throw e;
}
}
public abstract void Execute(WorkItem workItem);
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;
}
}
