twochmatome.py

#!/usr/bin/env python
# -*- coding:utf-8 -*-

import urllib2

from pyquery import PyQuery as pq

MATOME_PORTALS = {
    u'blomaga': u'http://ch.nicovideo.jp/portal/blomaga/matomemedia',
    u'livedoor': u'http://blog.livedoor.com/category/243/',
}


def get_matome_dom(portal_name, encoding='utf-8'):
    res = urllib2.urlopen(MATOME_PORTALS[portal_name])
    html = res.read().decode(encoding)
    dom = pq(html)
    return dom


def get_matome_info_from_livedoor():
    dom = get_matome_dom('livedoor')
    matome_info_doms = dom('ul.list[data-id="243"] h3.ttl>a')
    matome_info = set()
    for matome_info_dom in matome_info_doms:
        matome_info.add(
            (matome_info_dom.text, matome_info_dom.get('href'))
        )
    return matome_info


def get_matome_info_from_blomaga():
    dom = get_matome_dom('blomaga')
    title_doms = dom('span.blomaga_name.matome>a>span')
    url_doms = dom('span.blomaga_name.matome>a')
    matome_info_doms = zip(title_doms, url_doms)
    matome_info = set()
    for (title_dom, url_dom) in matome_info_doms:
        matome_info.add(
            (title_dom.text, url_dom.get('href'))
        )
    return matome_info


def get_matome_info(portal_name):
    if portal_name == u'livedoor' or portal_name == u'ld':
        matome_info = get_matome_info_from_livedoor()
    elif portal_name == u'blomaga' or portal_name == u'bm':
        matome_info = get_matome_info_from_blomaga()
    return matome_info


def main():
    blomaga_matome = get_matome_info('livedoor')
    livedoor_matome = get_matome_info('blomaga')
    for matome in [blomaga_matome, livedoor_matome]:
        for title, url in matome:
            print title
            print url + '\n'

if __name__ == '__main__':
    main()