public void TestGetSimpleText()
{
// a very simple file
XSSFExcelExtractor extractor = GetExtractor("sample.xlsx");
String text = extractor.Text;
Assert.IsTrue(text.Length > 0);
// Check sheet names
Assert.IsTrue(text.StartsWith("Sheet1"));
Assert.IsTrue(text.EndsWith("Sheet3\n"));
// Now without, will have text
extractor.SetIncludeSheetNames(false);
text = extractor.Text;
String CHUNK1 =
"Lorem\t111\n" +
"ipsum\t222\n" +
"dolor\t333\n" +
"sit\t444\n" +
"amet\t555\n" +
"consectetuer\t666\n" +
"adipiscing\t777\n" +
"elit\t888\n" +
"Nunc\t999\n";
String CHUNK2 =
"The quick brown fox jumps over the lazy dog\n" +
"hello, xssf hello, xssf\n"+
"hello, xssf hello, xssf\n"+
"hello, xssf hello, xssf\n"+
"hello, xssf hello, xssf\n";
Assert.AreEqual(
CHUNK1 +
"at\t4995\n" +
CHUNK2
, text);
// Now Get formulas not their values
extractor.SetFormulasNotResults(true);
text = extractor.Text;
Assert.AreEqual(
CHUNK1 +
"at\tSUM(B1:B9)\n" +
CHUNK2, text);
// With sheet names too
extractor.SetIncludeSheetNames(true);
text = extractor.Text;
Assert.AreEqual(
"Sheet1\n" +
CHUNK1 +
"at\tSUM(B1:B9)\n" +
"rich test\n" +
CHUNK2 +
"Sheet3\n"
, text);
extractor.Close();
}
public override string Parse()
{
if (!File.Exists(Context.Path))
throw new FileNotFoundException("File " + Context.Path + " is not found");
IWorkbook workbook = WorkbookFactory.Create(Context.Path);
bool extractHeaderFooter = false;
if (Context.Properties.ContainsKey("IncludeHeaderFooter"))
{
extractHeaderFooter = Utility.IsTrue(Context.Properties["IncludeHeaderFooter"]);
}
bool showCalculatedResult = false;
if (Context.Properties.ContainsKey("ShowCalculatedResult"))
{
showCalculatedResult = Utility.IsTrue(Context.Properties["ShowCalculatedResult"]);
}
bool includeSheetNames = true;
if (Context.Properties.ContainsKey("IncludeSheetNames"))
{
includeSheetNames = Utility.IsTrue(Context.Properties["IncludeSheetNames"]);
}
bool includeComment = true;
if (Context.Properties.ContainsKey("IncludeComments"))
{
includeComment = Utility.IsTrue(Context.Properties["IncludeComments"]);
}
if (workbook is XSSFWorkbook)
{
XSSFExcelExtractor extractor = new XSSFExcelExtractor((XSSFWorkbook)workbook);
extractor.SetIncludeHeadersFooters(extractHeaderFooter);
extractor.SetIncludeCellComments(includeComment);
extractor.SetIncludeSheetNames(includeSheetNames);
extractor.SetFormulasNotResults(!showCalculatedResult);
return extractor.Text;
}
else //if (workbook is HSSFWorkbook)
{
ExcelExtractor extractor = new ExcelExtractor((HSSFWorkbook)workbook);
extractor.IncludeHeaderFooter = extractHeaderFooter;
extractor.IncludeCellComments= includeComment;
extractor.IncludeSheetNames = includeSheetNames;
extractor.FormulasNotResults = !showCalculatedResult;
return extractor.Text;
}
}
public void TestInlineStrings()
{
XSSFExcelExtractor extractor = GetExtractor("InlineStrings.xlsx");
extractor.SetFormulasNotResults(true);
String text = extractor.Text;
// Numbers
Assert.IsTrue(text.Contains("43"), "Unable to find expected word in text\n" + text);
Assert.IsTrue(text.Contains("22"), "Unable to find expected word in text\n" + text);
// Strings
Assert.IsTrue(text.Contains("ABCDE"), "Unable to find expected word in text\n" + text);
Assert.IsTrue(text.Contains("Long Text"), "Unable to find expected word in text\n" + text);
// Inline Strings
Assert.IsTrue(text.Contains("1st Inline String"), "Unable to find expected word in text\n" + text);
Assert.IsTrue(text.Contains("And More"), "Unable to find expected word in text\n" + text);
// Formulas
Assert.IsTrue(text.Contains("A2"), "Unable to find expected word in text\n" + text);
Assert.IsTrue(text.Contains("A5-A$2"), "Unable to find expected word in text\n" + text);
}