class TestSimpleNormalization(unittest.TestCase):
def setUp(self):
self.__simpleNormalization = SimpleNormalization()
self.__dirPath = os.path.abspath(os.curdir)
self.__normalizeWords = ["thi", "test", "file", "encod", "проверк", "определен", "кодировк", "фа"]
def testNormalizeText(self):
filePath = os.path.join(self.__dirPath, "resources/test_encode_utf8")
with open(filePath) as utf8File:
words = self.__simpleNormalization.normalizeText(utf8File.read())
for itemWord in words:
self.assertIn(itemWord, self.__normalizeWords, "not normalized test_encode_utf8")
filePath = os.path.join(self.__dirPath, "resources/test_encode_win1251")
with open(filePath) as win1251File:
words = self.__simpleNormalization.normalizeText(win1251File.read())
for itemWord in words:
self.assertIn(itemWord, self.__normalizeWords, "not normalized test_encode_win1251")
filePath = os.path.join(self.__dirPath, "resources/test_encode_win866")
with open(filePath) as win866File:
words = self.__simpleNormalization.normalizeText(win866File.read())
for itemWord in words:
self.assertIn(itemWord, self.__normalizeWords, "not normalized test_encode_win866")
def testNormalizeTextParamError(self):
self.assertRaises(ParamError, self.__simpleNormalization.normalizeText, None)
self.assertRaises(ParamError, self.__simpleNormalization.normalizeText, "")
class TestSimpleNormalization(unittest.TestCase):
def setUp(self):
self.__simpleNormalization = SimpleNormalization()
self.__dirPath = os.path.abspath(os.curdir)
self.__normalizeWords = [
"thi", "test", "file", "encod", "проверк", "определен", "кодировк",
"фа"
]
def testNormalizeText(self):
filePath = os.path.join(self.__dirPath, "resources/test_encode_utf8")
with open(filePath) as utf8File:
words = self.__simpleNormalization.normalizeText(utf8File.read())
for itemWord in words:
self.assertIn(itemWord, self.__normalizeWords,
"not normalized test_encode_utf8")
filePath = os.path.join(self.__dirPath,
"resources/test_encode_win1251")
with open(filePath) as win1251File:
words = self.__simpleNormalization.normalizeText(
win1251File.read())
for itemWord in words:
self.assertIn(itemWord, self.__normalizeWords,
"not normalized test_encode_win1251")
filePath = os.path.join(self.__dirPath, "resources/test_encode_win866")
with open(filePath) as win866File:
words = self.__simpleNormalization.normalizeText(win866File.read())
for itemWord in words:
self.assertIn(itemWord, self.__normalizeWords,
"not normalized test_encode_win866")
def testNormalizeTextParamError(self):
self.assertRaises(ParamError, self.__simpleNormalization.normalizeText,
None)
self.assertRaises(ParamError, self.__simpleNormalization.normalizeText,
"")
