def test_missing(self):
validator = StaticValidator(self.params)
report = ValidationReport([])
validator.params.missing_sample_size_allowed = True
df = pd.DataFrame([[np.nan, np.nan, np.nan]],
columns=["val", "se", "sample_size"])
validator.check_bad_sample_size(df, "name", report)
assert len(report.raised_errors) == 0
validator.params.missing_sample_size_allowed = False
validator.check_bad_sample_size(df, "name", report)
assert len(report.raised_errors) == 1
assert "check_n_missing" in [
err.check_data_id[0] for err in report.raised_errors
]
def test_same_day(self):
params = {
"data_source": "",
"span_length": 0,
"end_date": "2020-09-01",
"expected_lag": {}
}
validator = StaticValidator(params)
report = ValidationReport([])
filenames = [("20200901_county_signal_signal.csv", "match_obj")]
validator.check_missing_date_files(filenames, report)
assert len(report.raised_errors) == 0
assert "check_missing_date_files" not in [
err.check_data_id[0] for err in report.raised_errors
]
def test_invalid_geo_id_hrr(self):
validator = StaticValidator(self.params)
report = ValidationReport([])
df = pd.DataFrame(["1", "11", "111", "8", "88", "888"],
columns=["geo_id"])
validator.check_bad_geo_id_value(df, "name", "hrr", report)
assert len(report.raised_errors) == 1
assert "check_bad_geo_id_value" in report.raised_errors[
0].check_data_id
assert len(report.raised_errors[0].expression) == 3
assert "1" not in report.raised_errors[0].expression
assert "11" not in report.raised_errors[0].expression
assert "111" not in report.raised_errors[0].expression
assert "8" in report.raised_errors[0].expression
assert "88" in report.raised_errors[0].expression
assert "888" in report.raised_errors[0].expression
def test_empty_filelist(self):
params = {
"data_source": "",
"span_length": 8,
"end_date": "2020-09-09",
"expected_lag": {}
}
validator = StaticValidator(params)
report = ValidationReport([])
report = ValidationReport([])
filenames = list()
validator.check_missing_date_files(filenames, report)
assert len(report.raised_errors) == 1
assert "check_missing_date_files" in [
err.check_data_id[0] for err in report.raised_errors
]
assert len(report.raised_errors[0].expression) == 9
def test_empty_df(self):
validator = StaticValidator(self.params)
report = ValidationReport([])
empty_df = pd.DataFrame(columns=["geo_id"], dtype=str)
validator.check_bad_geo_id_value(empty_df, "name", "county", report)
assert len(report.raised_errors) == 0
def test_single_column_duplicates_but_not_row(self):
validator = StaticValidator(self.params)
report = ValidationReport([])
df = pd.DataFrame([["a", "1"], ["a", "2"], ["b", "2"]])
validator.check_duplicate_rows(df, "file", report)
assert len(report.raised_warnings) == 0