distrowatch to CSV extractor

distros.py

#!/usr/bin/env python

import logging

import pulldistros

log = logging.getLogger(__name__)

def main():
    distros = pulldistros.distros()
    for distro in distros:
        log.debug("Getting Distro %s", distro)
        dinfo = pulldistros.distinfo(distro)
        log.debug("%s: %r", distro, dinfo)
        releases = dinfo['releases']
        date_name = sorted(((value['Release Date'], key) for key, value in releases.items()))
        first_release_date, first_release_name = map(lambda s: s.strip(), date_name[0])
        last_release_date, last_release_name = map(lambda s: s.strip(), date_name[-1])
        number_of_releases = len(date_name)

        log.info('%s Home Page: %r', distro, dinfo['Home Page'])
        url = dinfo['Home Page'].split()[0].replace('https://', 'http://').strip()
        name = distro.strip()
        popularity = dinfo['Popularity'].strip()
        status = dinfo['Status'].strip()
        origin = dinfo['Origin'].strip()
        dwurl = "http://distrowatch.com/" + distro


        s = ("%s ; %s; %s; %s; %s"
            ";%s; %s"
            ";%s"
            ";%s;%s"
            ";%s")
        s %= (url, name, popularity, status, origin,
            first_release_name, first_release_date,
            number_of_releases,
            last_release_date, last_release_name,
            dwurl)
        log.info("Printing now... %s", s)

        assert '\n' not in url
        assert '\n' not in s
        print ("%s" % s)

if __name__ == "__main__":
    logging.basicConfig(level=logging.DEBUG)
    print (main())