def test_parse_url_invalid_req_cols(input_df):
expected_error = ValueError(
"Given req_cols must be subset of %s" %
(["hostname", "subdomain", "domain", "suffix"]))
with pytest.raises(ValueError) as actual_error:
dns.parse_url(input_df["url"], req_cols={"test"})
assert actual_error == expected_error
def test_parse_url(input_df):
expected_output_df = DataFrame({
"domain": [
"google",
"gmail",
"github",
"pydata",
"worldbank",
"waiterrant",
"cnn",
"cnn",
"cnn",
"news",
"news",
"news",
"sbcglobal",
"akamaitechnologies",
],
"suffix": [
"com",
"com",
"com",
"org",
"org.kg",
"blogspot.com",
"com.ac",
"ac",
"com",
"uk",
"co.uk",
"co.uk",
"net",
"com",
],
})
output_df = dns.parse_url(input_df["url"], req_cols={"domain", "suffix"})
for col in expected_output_df.columns:
assert expected_output_df[col].equals(output_df[col])
def test2_parse_url(input_df):
expected_output_df = DataFrame({
"hostname": [
"www.google.com",
"gmail.com",
"github.com",
"pandas.pydata.org",
"www.worldbank.org.kg",
"waiterrant.blogspot.com",
"forums.news.cnn.com.ac",
"forums.news.cnn.ac",
"b.cnn.com",
"a.news.uk",
"a.news.co.uk",
"a.news.co.uk",
"107-193-100-2.lightspeed.cicril.sbcglobal.net",
"a23-44-13-2.deploy.static.akamaitechnologies.com",
],
"subdomain": [
"www",
"",
"",
"pandas",
"www",
"",
"forums.news",
"forums.news",
"b",
"a",
"a",
"a",
"107-193-100-2.lightspeed.cicril",
"a23-44-13-2.deploy.static",
],
"domain": [
"google",
"gmail",
"github",
"pydata",
"worldbank",
"waiterrant",
"cnn",
"cnn",
"cnn",
"news",
"news",
"news",
"sbcglobal",
"akamaitechnologies",
],
"suffix": [
"com",
"com",
"com",
"org",
"org.kg",
"blogspot.com",
"com.ac",
"ac",
"com",
"uk",
"co.uk",
"co.uk",
"net",
"com",
],
})
output_df = dns.parse_url(input_df["url"])
assert expected_output_df.equals(output_df)