public void issue_81_file_1()
{
var filePath = _filePathParent + "sad-files/EI-73-1018-2_5632837.doc";
Action act = () => _cut.Parse(filePath);
act.Should().Throw <TextExtractionException>();
//act.ShouldThrow<TextExtractionException>();
}
public void ParallelTestWithLokalParser()
{
var plist = new List <KeyValuePair <string, string> >();
var list = new List <KeyValuePair <string, string> >();
list.Add(new KeyValuePair <string, string>(_filePathParent + "files/Tika.pptx", "Tika Test Presentation"));
list.Add(new KeyValuePair <string, string>(_filePathParent + "files/Tika.docx", "formatted in interesting ways"));
list.Add(new KeyValuePair <string, string>(_filePathParent + "files/Tika.xlsx", "Use the force duke"));
for (int i = 0; i < 1000; i++)
{
plist.AddRange(list);
}
Parallel.ForEach(plist, (test) =>
{
var tika2 = new Tika();
var result = tika2.Parse(test.Key);
result.Text.Should().Contain(test.Value);
var result2 = tika.ParseToString(test.Key);
result2.Should().Contain(test.Value);
});
}
public void StreamParsing()
{
string filePath = _filePathParent + "files/Tika.rtf";
using (FileStream fileStream = new FileStream(filePath, FileMode.Open, FileAccess.Read))
{
var result = tika.Parse(fileStream);
result.Text.Should().Contain("pack of pickled almonds");
}
}
public void should_extract_author_list_from_pdf()
{
var textExtractionResult = _cut.Parse(_filePathParent + "files/file_author.pdf", CreateCustomResult);
textExtractionResult.Metadata["meta:author"].Should().ContainInOrder("Bernal, M. A.", "deAlmeida, C. E.", "Incerti, S.", "Champion, C.", "Ivanchenko, V.", "Francis, Z.");
}
public void non_existing_files_should_fail_with_exception()
{
string fileName = _filePathParent + "files/doesnotexist.mp3";
Action act = () => _cut.Parse(fileName);
act.Should().Throw <TextExtractionException>().Which.Message.Should().Contain(fileName);
//act.ShouldThrow<TextExtractionException>()
// .Which.Message.Should().Contain(fileName);
}