def test_should_ignore_nan_secondary_email_address(self):
assert extract_distinct_emails_from_data_frame(pd.DataFrame([{
'email': np.nan,
'secondary_email': np.nan
}])) == set()
def test_should_strip_spaces_and_ignore_blank(self):
assert extract_distinct_emails_from_data_frame(pd.DataFrame([{
'email': ' ',
'secondary_email': ' %s ' % EMAIL_2
}])) == {EMAIL_2}
def test_should_ignore_blank_secondary_email_address(self):
assert extract_distinct_emails_from_data_frame(pd.DataFrame([{
'email': EMAIL_1,
'secondary_email': ''
}])) == {EMAIL_1}
def test_should_extract_secondary_email_address_if_column_present(self):
assert extract_distinct_emails_from_data_frame(pd.DataFrame([{
'email': EMAIL_1,
'secondary_email': EMAIL_2
}])) == {EMAIL_1, EMAIL_2}
def test_should_extract_primary_email_address_only_once(self):
assert extract_distinct_emails_from_data_frame(pd.DataFrame([{
'email': EMAIL_1
}, {
'email': EMAIL_1
}])) == {EMAIL_1}
def test_should_extract_multiple_primary_email_addresses(self):
assert extract_distinct_emails_from_data_frame(pd.DataFrame([{
'email': EMAIL_1
}, {
'email': EMAIL_2
}])) == {EMAIL_1, EMAIL_2}