username & email extractor for VCS & RhodeCode

gistfile1.py

def author_email(author):
    """
    returns email address of given author.
    If any of <,> sign are found, it fallbacks to regex findall()
    and returns first found result or empty string
    
    Regex taken from http://www.regular-expressions.info/email.html
    """
    import re
    r = author.find('>')
    l = author.find('<')


    if l == -1 or r == -1:
        # fallback to regex match of email out of a string
        email_re = re.compile(r"""[a-z0-9!#$%&'*+/=?^_`{|}~-]+(?:\.[a-z0-9!"""
                              r"""#$%&'*+/=?^_`{|}~-]+)*@(?:[a-z0-9](?:[a-z"""
                              r"""0-9-]*[a-z0-9])?\.)+[a-z0-9](?:[a-z0-9-]"""
                              r"""*[a-z0-9])?""", re.IGNORECASE)
        m = re.findall(email_re, author)
        return m[0] if m else ''

    return author[l + 1:r].strip()


def author_name(author):
    """
    get name of author, or else username.
    It'll try to find an email in the author string and just cut it off
    to get the username
    """

    if not '@' in author:
        return author
    else:
        return author.replace(author_email(author), '').replace('<', '')\
            .replace('>', '').strip()



#==============================================================================
# TESTS
#==============================================================================

test_usernames = [('Marcin Kuzminski <[email protected]>',
                    ('Marcin Kuzminski', '[email protected]')),
                  ('Marcin Kuzminski Spaces < [email protected] >',
                    ('Marcin Kuzminski Spaces', '[email protected]')),
                  ('Marcin Kuzminski <[email protected]>',
                    ('Marcin Kuzminski', '[email protected]')),
                  ('mrf RFC_SPEC <[email protected]>',
                    ('mrf RFC_SPEC', '[email protected]')),
                  ('username <[email protected]>',
                    ('username', '[email protected]')),
                  ('username <[email protected]',
                   ('username', '[email protected]')),
                  ('broken [email protected]',
                   ('broken', '[email protected]')),
                  ('<[email protected]>',
                   ('', '[email protected]')),
                  ('justname',
                   ('justname', '')),
                  ('Mr Double Name [email protected] ',
                   ('Mr Double Name', '[email protected]')),
                  ]

def test_author_email():

    for test_str, result in test_usernames:
        assert result[1] == author_email(test_str)


def test_author_name():

    for test_str, result in test_usernames:
        assert result[0] == author_name(test_str)


if __name__ == '__main__':

    test_author_email()
    test_author_name()