def test_fix_term_in_the_end(self):
checker = chks.SentCorrectnessChecker(['term_in_the_end'])
chunk = Chunk("", "Without term ", "", "", 1)
checker.fix(chunk)
self.assertEqual("Without term.", chunk.get_mod_text())
chunk = Chunk("", ["Without term ", "wo term"], "", "", 1)
checker.fix(chunk)
self.assertEqual("Without term. wo term.", chunk.get_mod_text())
chunk = Chunk("", ["Boring sent.", "text"], "", "", 1)
checker.fix(chunk)
self.assertEqual("Boring sent. text.", chunk.get_mod_text())
def test_fix_title_case(self):
checker = chks.SentCorrectnessChecker(['title_case'])
chunk = Chunk("", "маленькая буква.", "", "", 1)
checker.fix(chunk)
self.assertEqual("Маленькая буква.", chunk.get_mod_text())
chunk = Chunk(
"", ". В основном токсин из организма выводится через почки.", "",
"", 1)
checker.fix(chunk)
self.assertEqual(1, len(chunk.get_mod_sents()))
self.assertEqual(
"В основном токсин из организма выводится через почки.",
chunk.get_mod_text())
def create_checkers(opts, sources_dir,
spell_checker_whitelist = None):
return [
chks.OrigSentChecker(opts),
chks.SourceDocsChecker(opts, sources_dir),
chks.PRChecker(opts, fluctuation_delta = 5),
chks.AddChecker(opts, fluctuation_delta = 5),
chks.DelChecker(opts, fluctuation_delta = 5),
chks.CPYChecker(opts, fluctuation_delta = 5),
chks.CctChecker(opts, fluctuation_delta = 5),
chks.SspChecker(opts, fluctuation_delta = 5),
chks.ORIGModTypeChecker(),
chks.SentCorrectnessChecker(),
chks.CyrillicAlphabetChecker(opts),
chks.SpellChecker(whitelist = spell_checker_whitelist)
]
def test_term_in_the_end(self):
checker = chks.SentCorrectnessChecker(['term_in_the_end'])
chunk = Chunk("", "Correct sent with trailing spaces!!!!! ", "", "",
1)
checker(chunk, None)
self.assertEqual(1, len(chunk.get_mod_sents()))
self.assertEqual(0, len(checker.get_errors()))
chunk = Chunk("", "Boring sent.", "", "", 1)
checker(chunk, None)
self.assertEqual(0, len(checker.get_errors()))
chunk = Chunk("", "Question mark?", "", "", 1)
checker(chunk, None)
self.assertEqual(0, len(checker.get_errors()))
chunk = Chunk("", "Without term ", "", "", 1)
checker(chunk, None)
self.assertEqual(1, len(checker.get_errors()))
def test_title_case(self):
checker = chks.SentCorrectnessChecker(['title_case'])
chunk = Chunk("", "Корректное предложение!", "", "", 1)
checker(chunk, None)
self.assertEqual(0, len(checker.get_errors()))
chunk = Chunk("", "Из-за дефиса не работает str.istitle.", "", "", 1)
checker(chunk, None)
self.assertEqual(0, len(checker.get_errors()))
chunk = Chunk("", '"Цитата: текст"', "", "", 1)
checker(chunk, None)
self.assertEqual(0, len(checker.get_errors()))
chunk = Chunk("", '2009 number is ok.', "", "", 1)
checker(chunk, None)
self.assertEqual(0, len(checker.get_errors()))
chunk = Chunk("", "маленькая буква.", "", "", 1)
checker(chunk, None)
self.assertEqual(1, len(checker.get_errors()))
def create_checkers(opts, sources_dir, spell_checker_whitelist=None):
return [
chks.OriginalityChecker(opts),
chks.OrigSentChecker(opts),
chks.ModSentChecker(opts),
chks.SourceDocsChecker(opts, sources_dir),
chks.PRChecker(opts),
# chks.AddChecker(opts),
# chks.DelChecker(opts),
# chks.CPYChecker(opts),
chks.CctChecker(opts),
# chks.SspChecker(opts),
chks.SHFChecker(opts),
# chks.SYNChecker(opts),
chks.LexicalSimChecker(opts),
trans_chks.ORIGModTypeChecker(),
chks.SentCorrectnessChecker(),
trans_chks.TranslationChecker(opts),
trans_chks.ManualTranslationChecker(opts),
chks.CyrillicAlphabetChecker(opts),
chks.SpellChecker(whitelist=spell_checker_whitelist)
]