1#!/usr/bin/env python 2# -*- coding: utf-8 -*- 3# Copyright (C) 2012-2014 Bastian Kleineidam 4# Copyright (C) 2015-2016 Tobias Gruetzmacher 5""" 6Script to get a list of creators.com comics and save the info in a JSON file 7for further processing. 8""" 9from __future__ import absolute_import, division, print_function 10 11from scriptutil import ComicListUpdater 12 13 14class CreatorsUpdater(ComicListUpdater): 15 dup_templates = ('GoComics/%s',) 16 17 # names of comics to exclude 18 excluded_comics = ( 19 # no images 20 'Doodles', 21 ) 22 23 def handle_url(self, url): 24 """Parse one listing page.""" 25 data = self.get_url(url) 26 27 for comicdiv in data.cssselect('ul.all-test li'): 28 comiclink = comicdiv.cssselect('a')[0] 29 comicurl = comiclink.attrib['href'] 30 name = comicdiv.cssselect('p strong')[0].text 31 32 self.add_comic(name, comicurl.rsplit('/', 1)[1]) 33 34 def collect_results(self): 35 """Parse all search result pages.""" 36 self.handle_url('https://www.creators.com/categories/comics/all') 37 self.handle_url('https://www.creators.com/categories/cartoons/all') 38 39 def get_entry(self, name, path): 40 langopt = ", 'es'" if name.lower().endswith('spanish') else '' 41 return u"cls('%s', '%s'%s)," % (name, path, langopt) 42 43 44if __name__ == '__main__': 45 CreatorsUpdater(__file__).run() 46