def test_invalid_geo_id_value_nation(self):
validator = StaticValidator(self.params)
report = ValidationReport([])
df = pd.DataFrame(["us", "zz"], columns=["geo_id"])
validator.check_bad_geo_id_value(df, FILENAME, "nation", report)
assert len(report.raised_errors) == 1
assert report.raised_errors[0].check_name == "check_bad_geo_id_value"
def test_uppercase_geo_id(self):
validator = StaticValidator(self.params)
report = ValidationReport([])
df = pd.DataFrame(["ak", "AK"], columns=["geo_id"])
validator.check_bad_geo_id_value(df, FILENAME, "state", report)
assert len(report.raised_errors) == 0
assert len(report.raised_warnings) == 1
assert report.raised_warnings[0].check_name == "check_geo_id_lowercase"
def test_state_level_fips(self):
validator = StaticValidator(self.params)
report = ValidationReport([])
df = pd.DataFrame(["37183", "56000", "04000", "60000", "78000"], columns=["geo_id"])
validator.check_bad_geo_id_value(df, FILENAME, "county", report)
assert len(report.raised_errors) == 0
df = pd.DataFrame(["37183", "56000", "04000", "60000", "78000", "99000"], columns=["geo_id"])
validator.check_bad_geo_id_value(df, FILENAME, "county", report)
assert len(report.raised_errors) == 1
assert report.raised_errors[0].check_name == "check_bad_geo_id_value"
def test_additional_valid_geo_ids(self):
params = self.params.copy()
params["static"] = {
"additional_valid_geo_values": {
"state": ["state1"],
"county": ["county1", "county2"]
}
}
validator = StaticValidator(params)
report = ValidationReport([])
df = pd.DataFrame(["05109", "06019", "county2"], columns=["geo_id"])
validator.check_bad_geo_id_value(df, FILENAME, "county", report)
assert len(report.raised_errors) == 0
df = pd.DataFrame(["ma", "state1", "mi"], columns=["geo_id"])
validator.check_bad_geo_id_value(df, FILENAME, "state", report)
assert len(report.raised_errors) == 0
df = pd.DataFrame(["county2", "02"], columns=["geo_id"])
validator.check_bad_geo_id_value(df, FILENAME, "hhs", report)
assert len(report.raised_errors) == 1
assert report.raised_errors[0].check_name == "check_bad_geo_id_value"
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, FILENAME, "county", report)
assert len(report.raised_errors) == 0