async def test_github_data_preparer():
matcher = CountryNameMatcher()
data: str = await mock_load_github()
data = GithibDataPreparer.prepare(data, matcher)
assert type(data) == pd.DataFrame
columns = data.to_dict()
check_type(None, columns, Dict[datetime, Dict[str, int]])
exceptions = {'kosovo'}
assert set(data.index) - matcher.keys() == exceptions
assert 'all' in data.index
assert len(data) > 200
async def test_region_handling():
matcher = CountryNameMatcher()
data: str = await mock_load_github()
data = GithibDataPreparer._csv_to_dataframe(data)
data = GithibDataPreparer._colnames_to_datetime(data)
data = GithibDataPreparer._country_and_region_names_to_keys(data, matcher)
unhandled_regions = data.filter(['country', 'region']).dropna()
unhandled_regions = unhandled_regions.loc[~unhandled_regions.region.
isin(matcher.keys())]
exceptions = {'australia', 'canada', 'china'}
unhandled_regions = unhandled_regions.loc[~unhandled_regions.country.
isin(exceptions)]
assert len(unhandled_regions) == 0, str(unhandled_regions)