def setUp(self):
self.__simpleNormalization = SimpleNormalization()
self.__dirPath = os.path.abspath(os.curdir)
self.__normalizeWords = [
"thi", "test", "file", "encod", "проверк", "определен", "кодировк",
"фа"
]
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, "")
def setUp(self):
self.__mongoUtils = MongoSaveUtils(HOST, PORT, USR, PWD, DB, FC_N,
FC_DN, MDN)
self.__simpleNormal = SimpleNormalization()
self.__fileSourceCustom = FileSourceCustom()
self.__fileBlockSource = FileBlockSource()
self.__calcMongo = CalcMongo()
self.__mongoStatistics = MongoStatistic(self.__mongoUtils)
self.__dirPath = os.path.abspath(os.curdir)
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,
"")
def setUp(self):
self.__dirPath = os.path.abspath(os.curdir)
firstPath = os.path.join(self.__dirPath, "resources/first")
secondPath = os.path.join(self.__dirPath, "resources/second")
self.__mongoSaveUtils = MongoSaveRetrievalUtils(
HOST, PORT, USR, PWD, DB, FC_N, FC_DN, MDN)
self.__mongoReadUtils = MongoReadUtils(HOST, PORT, USR, PWD, DB, FC_N,
FC_DN)
self.__simN = SimpleNormalization()
self.__simNamesN = FileNameNormalization()
self.__ms = StatisticFactory().createStatistic(MONGO_TYPE,
self.__mongoSaveUtils)
self.__scc = FSSourceCustomCallback()
self.__rnFS = ReaderNameFS([firstPath, secondPath])
def setUp(self):
self.__dirPath = os.path.abspath(os.curdir)
firstPath = os.path.join(self.__dirPath, "resources/first")
secondPath = os.path.join(self.__dirPath, "resources/second")
self.__mongoUtils = MongoSaveRetrievalUtils(HOST, PORT, USR, PWD, DB,
FC_N, FC_DN, MDN)
self.__mongoUtilsTypeError = MongoSaveUtils(HOST, PORT, USR, PWD, DB,
FC_N, FC_DN, MDN)
self.__mongoReadUtils = MongoReadUtils(HOST, PORT, USR, PWD, DB, FC_N,
FC_DN)
self.__smN = SimpleNormalization()
self.__scc = FSSourceCustomCallback()
self.__rFS = ReaderNameFS([firstPath, secondPath])
self.__fsWorker = DataSourceWorkerFS()
def testParseQ(self):
smN = SimpleNormalization()
parseList = self.spq.parseQ("testing q q q q testing", None, smN)
self.assertEqual(parseList[0], TYPE_Q_FULL)
self.assertListEqual(parseList[1], ["test", "q"])
def setUp(self):
self.__simpleNormalization = SimpleNormalization()
self.__dirPath = os.path.abspath(os.curdir)
self.__normalizeWords = ["thi", "test", "file", "encod", "проверк", "определен", "кодировк", "фа"]
