static IAdapter GetAdapter(string inputFile)
{
string extension = Path.GetExtension(inputFile).ToLower();
switch (extension)
{
case ".htm":
case ".html":
if (HtmAdapter.CanProcess(inputFile))
{
return(new HtmAdapter(inputFile));
}
else
{
return(new AngleHtmlAdapter(inputFile, MaxRowsToProcess));
}
case ".pdf":
case ".xhtml":
case ".doc":
case ".rtf":
case ".toloka_json":
case ".docx":
return(OpenXmlWordAdapter.CreateAdapter(inputFile, MaxRowsToProcess));
case ".xls":
case ".xlsx":
if (AdapterFamily == "aspose" || AdapterFamily == "prod")
{
if (!AsposeLicense.Licensed && extension == ".xls")
{
throw new Exception("xls file format is not supported");
}
if (AsposeLicense.Licensed)
{
return(AsposeExcelAdapter.CreateAdapter(inputFile, MaxRowsToProcess));
}
}
else if (AdapterFamily == "npoi")
{
return(NpoiExcelAdapter.CreateAdapter(inputFile, MaxRowsToProcess));
}
else
{
return(null);
}
break;
default:
Logger.Error("Unknown file extension " + extension);
return(null);
}
Logger.Error("Cannot find adapter for " + inputFile);
return(null);
}
public void RealEstateColumnDetector()
{
string docxFile = Path.Combine(TestUtil.GetTestDataPath(), "glav_44_2010.doc");
IAdapter adapter = OpenXmlWordAdapter.CreateAdapter(docxFile, -1);
ColumnOrdering ordering = ColumnDetector.ExamineTableBeginning(adapter);
Assert.AreEqual(ordering.ColumnOrder.Count, 9);
}
public void SpendingsWrongColumnTest()
{
string docxFile = Path.Combine(TestUtil.GetTestDataPath(), "82442.doc");
IAdapter adapter = OpenXmlWordAdapter.CreateAdapter(docxFile, -1);
ColumnOrdering ordering = ColumnDetector.ExamineTableBeginning(adapter);
Assert.AreEqual(ordering.ColumnOrder[DeclarationField.DeclaredYearlyIncome].BeginColumn, 1);
}
public void TwoRowHeaderEmptyTopCellTest()
{
string docxFile = Path.Combine(TestUtil.GetTestDataPath(), "57715.doc");
IAdapter adapter = OpenXmlWordAdapter.CreateAdapter(docxFile, -1);
ColumnOrdering ordering = ColumnDetector.ExamineTableBeginning(adapter);
Assert.AreEqual(ordering.ColumnOrder.Count, 13);
Assert.AreEqual(ordering.ColumnOrder[DeclarationField.Vehicle].BeginColumn, 10);
Assert.AreEqual(ordering.ColumnOrder[DeclarationField.DeclaredYearlyIncome].BeginColumn, 11);
}
public void RedundantColumnDetector()
{
string docxFile = Path.Combine(TestUtil.GetTestDataPath(), "18664.docx");
IAdapter adapter = OpenXmlWordAdapter.CreateAdapter(docxFile, -1);
ColumnOrdering ordering = ColumnDetector.ExamineTableBeginning(adapter);
Assert.AreEqual(ordering.ColumnOrder.Count, 13);
Assert.AreEqual(ordering.ColumnOrder[DeclarationField.AcquiredProperty].BeginColumn, 11);
Assert.AreEqual(ordering.ColumnOrder[DeclarationField.MoneySources].BeginColumn, 12);
}
private static IAdapter GetCommonAdapter(string inputFile)
{
if (AdapterFamily != "aspose")
{
if (AdapterFamily == "prod")
{
return(OpenXmlWordAdapter.CreateAdapter(inputFile, MaxRowsToProcess));
}
}
else if (!AsposeLicense.Licensed)
{
throw new Exception("doc and docx file format is not supported");
}
return(AsposeDocAdapter.CreateAdapter(inputFile));
}
