def test_read_from_file_should_deliver_text(): """ Test, if reading a file returns a string that is not empty. """ e = Event() e.read_file("test/test_articles/01_wikipedia_syrian_civil_war.txt") assert_not_equal(e.text, "")
def test_correct_amount_of_events_should_be_extracted(): """ Test, if all events of the test-file marked_sentences.txt are extracted """ e = Event() e.read_file("test/test_articles/marked_sentences.txt") assert_equal(len(e.start_extraction()), 19)
def test_remove_references(): e = Event() e.read_file("test/test_articles/01_wikipedia_syrian_civil_war.txt") text = remove_references(e.text) if re.search(str('\[\d{1,3}]'), text): print("CHECK") else: print("NO")
def test_references_should_be_removed(): """ Check, if references are removed from text. Maybe to be refined later on, because, it is the same regex command that removes the references from text - therefore it is most likely that this test is fine. hmm... """ e = Event() e.read_file("test/test_articles/01_wikipedia_syrian_civil_war.txt") # re searches for references in bracktes, like "[12]" and should not find any. assert_false(re.search(str('\[\d{1,3}]'), remove_references(e.text)))
def test_read_from_file(): e = Event() e.read_file("test/test_articles/random_test.txt") print(e.text)
def test_start_extraction(): e = Event() e.read_file("test/test_articles/random_test.txt") e.text = remove_references(e.text) e.start_extraction()