Show common leading users for 2 different SE sites.

so_leaders.py

"""Show common leading users for different SE sites.

Pass --grid from the command line to get a grid of common users
between askubuntu.com, unix.stackexchange.com,
superuser.com, and serverfault.com.

For the monthly report, it currently looks like this:
            s   s   a   u
            e   u   s   n
            r   p   k   i
            v   e   u   x
            e   r   b   .
            r   u   u   s
            f   s   n   t
            a   e   t   a

serverfa   36   2   0   4
superuse    2  36   0   3
askubunt    0   0  36   2
unix.sta    4   3   2  36
"""
from lxml import html
import textwrap
import pprint
import argparse

UNIX_SERVER = 'unix.stackexchange.com'
AU_SERVER = 'askubuntu.com'

USER_LINKS = '//div[@class="user-details"]/a/@href'

def users_url(server, filter='week'):
    """Build a top users URL for the server and filter."""
    return 'http://{0}/users?tab=reputation&filter={1}'.format(server, filter)

def username(path):
    """Returns the username from a SE path."""
    return path.split('/')[-1]

def common_users(url1, url2, xpath=USER_LINKS, parse=html.parse):
    """Returns the set of matching elements at `xpath` common between URLS."""
    def users(url):
        return (username(user_path)
                for user_path in parse(url).xpath(xpath))

    sets = [set(users(url)) for url in (url1, url2)]
    return set.intersection(*sets)

def site_top_user_report(url):
    print "Top users at '{0}':".format(url)
    usernames = (username(path)
                 for path in html.parse(url).xpath(USER_LINKS))
    print textwrap.fill(' '.join(usernames))

def render_grid(server_names, filter='week'):

    def abbrev(server):
        return '{0:8s}'.format(server[0:8])
    server_abbrs = [abbrev(server) for server in server_names]

    def header(server_abbrs):
        def line(index):
            return ' ' * 12 + '   '.join(name[index] for name in server_abbrs)
        return '\n'.join(line(i) for i in range(8)) + '\n\n'

    def body(server_names):
        import StringIO
        import urllib

        pages = [StringIO.StringIO(urllib.urlopen(users_url(server, filter)).read())
                 for server in server_names]
        counts = [[len(common_users(p1, p2)) for p1 in pages] for p2 in pages]

        def line(name, counts):
            counts_str = (' '.join('{0:3d}'.format(c) for c in counts))
            return '{0}  {1}'.format(name, counts_str)

        return '\n'.join(line(abbrev(name), row)
                         for name, row in zip(server_names, counts))

    return header(server_abbrs) + body(server_names)

if __name__ == '__main__':
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument('-f', '--filter', default='week')
    parser.add_argument('-1', '--server-1', default=UNIX_SERVER)
    parser.add_argument('-2', '--server-2', default=AU_SERVER)
    parser.add_argument('-g', '--grid', action='store_true')
    options = parser.parse_args()

    if options.grid:
        servers = ['serverfault.com',
                   'superuser.com',
                   'askubuntu.com',
                   'unix.stackexchange.com']
        print render_grid(servers, options.filter)

    else:
        unix_url = users_url(options.server_1, options.filter)
        au_url = users_url(options.server_2, options.filter)

        site_top_user_report(unix_url)
        print
        site_top_user_report(au_url)
        print

        print "Common users:"
        pprint.pprint(common_users(unix_url, au_url))