def __init__(self, file_name: str, name: str):
self.__file_name = file_name
self.__name = Utils.replace_redundant_ws(name).lower()
self.__cleaned_name = cleanco(self.__name).clean_name()
self.__tokens = set(self.__cleaned_name.split(" "))
if len(self.tokens) == 0:
raise AssertionError("length of list is zero")
def create_mapper_from_dataframe(dataframe: DataFrame):
dataframe.columns = Index(
map(Utils.normalize_string, dataframe.columns))
if not set(CompanyMapper.COLUMN_NAMES).issubset(set(
dataframe.columns)):
raise AssertionError("necessary columns do not exist")
dataframe = dataframe[CompanyMapper.COLUMN_NAMES]
name_to_group: Dict[CompanyNameWithFileName, int] = {}
for _, row in dataframe.iterrows():
name = CompanyNameWithFileName(
row[CompanyMapper.COLUMN_NAMES[0]],
Utils.normalize_string(row[CompanyMapper.COLUMN_NAMES[1]]))
name_to_group[name] = row[CompanyMapper.COLUMN_NAMES[2]]
return CompanyMapper(name_to_group)
def _standartize_columns_names(dataframe: pd.DataFrame):
dataframe.columns = pd.Index(data=[
Utils.replace_redundant_ws(column_name).lower()
for column_name in dataframe.columns
])
def __init__(self, key_name: str):
self.__original_key_name = key_name
self.__key_name = Utils.replace_redundant_ws(key_name).lower()
def test_contract(self):
try:
Utils.replace_redundant_ws(1)
assert 1 == 0, "contract doesn't work"
except ContractNotRespected:
pass
def test_newline_replace(self):
arg = "\n\n abc\r\n\r\n\tabc"
expected = "abc abc"
actual = Utils.replace_redundant_ws(arg)
assert actual == expected
def test_ws_remove_2(self):
arg = "\tabc \t abc\t\tabc"
expected = "abc abc abc"
actual = Utils.replace_redundant_ws(arg)
assert expected == actual