calibre/recipes/kathemerini.recipe

from calibre.web.feeds.recipes import BasicNewsRecipe


class Kathimerini(BasicNewsRecipe):
    title = 'Kathimerini'
    __author__ = 'jenniepet'
    description = 'News from Greece'
    max_articles_per_feed = 100
    oldest_article = 1
    publisher = 'Kathimerini'
    category = 'news, GR'
    language = 'el'
    encoding = 'utf-8'
    conversion_options = {'linearize_tables': True}
    no_stylesheets = True
    remove_tags_before = dict(id='site-body')
    remove_tags_after = [dict(id='social')]
    remove_tags = [
        dict(attrs={'class': ['post-tools', 'edition edition_PRINT']})]
# to remove images comment the line above and uncomment the line below
#    remove_tags            = [dict(attrs={'class':['post-tools', 'edition edition_PRINT', 'clearing-featured-img']})]

# Categories In order of Appearance: Politics-1-2-3, Greece-1-2, World-1-2, People-Specials
# Greek Economy-1-2, Business, International Economy, Real Estate
    feeds = [(
        u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03BA\u03AE', 'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=2&type=&edition=PRINT&author=0&fromDate=&toDate='),  # noqa
        (u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03BA\u03AE 2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=2&type=&edition=PRINT&author=0&fromDate=&toDate=&page=1'),
        (u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03BA\u03AE 3',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=2&type=&edition=PRINT&author=0&fromDate=&toDate=&page=2'),
        (u'\u0395\u03BB\u03BB\u03AC\u03B4\u03B1',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=4&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u0395\u03BB\u03BB\u03AC\u03B4\u03B1 2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=4&type=&edition=PRINT&author=0&fromDate=&toDate=&page=1'),
        (u'\u039A\u03CC\u03C3\u03BC\u03BF\u03C2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=5&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u039A\u03CC\u03C3\u03BC\u03BF\u03C2 2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=5&type=&edition=PRINT&author=0&fromDate=&toDate=&page=1'),
        (u'\u03A0\u03C1\u03CC\u03C3\u03C9\u03C0\u03B1',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=41&cat=42&cat=43&cat=24&cat=25&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u0395\u03BB\u03BB\u03B7\u03BD\u03B9\u03BA\u03AE \u039F\u03B9\u03BA\u03BF\u03BD\u03BF\u03BC\u03AF\u03B1',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=17&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u0395\u03BB\u03BB\u03B7\u03BD\u03B9\u03BA\u03AE \u039F\u03B9\u03BA\u03BF\u03BD\u03BF\u03BC\u03AF\u03B1 2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=17&type=&edition=PRINT&author=0&fromDate=&toDate=&page=1'),
        (u'\u0395\u03C0\u03B9\u03C7\u03B5\u03B9\u03C1\u03AE\u03C3\u03B5\u03B9\u03C2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=18&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u0394\u03B9\u03B5\u03B8\u03BD\u03AE\u03C2 \u039F\u03B9\u03BA\u03BF\u03BD\u03BF\u03BC\u03AF\u03B1',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=19&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'Real Estate', 'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=21&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u03A0\u03B5\u03C1\u03B9\u03B2\u03AC\u03BB\u03BB\u03BF\u03BD',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=6&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u0395\u03C0\u03B9\u03C3\u03C4\u03AE\u03BC\u03B7',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=7&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u03A4\u03B5\u03C7\u03BD\u03BF\u03BB\u03BF\u03B3\u03AF\u03B1',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=47&cat=48&cat=49&cat=50&cat=51&type=&edition=PRINT&author=0&fromDate=&toDate='),
        (u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=31&cat=32&cat=33&cat=34&cat=35&cat=36&cat=37&cat=38&cat=39&type=&edition=PRINT&author=0&fromDate=&toDate='),  # noqa
        (u'\u03A0\u03BF\u03BB\u03B9\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2 2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=31&cat=32&cat=33&cat=34&cat=35&cat=36&cat=37&cat=38&cat=39&type=&edition=PRINT&author=0&fromDate=&toDate=&page=1'),  # noqa
        (u'\u03A4\u03B1\u03BE\u03AF\u03B4\u03B9\u03B1',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=11&cat=10&cat=12&cat=14&cat=15&cat=13&type=&edition=PRINT&author=0&fromDate=&toDate='),  # noqa
        (u'\u0391\u03B8\u03BB\u03B7\u03C4\u03B9\u03C3\u03BC\u03CC\u03C2',
         'http://www.kathimerini.gr/rss?i=news.el.search&q=&t=0&w=&c=&s=p&cat=3&type=&edition=PRINT&author=0&fromDate=&toDate=')
    ]

    def get_cover_url(self):
        import time
        return 'http://s.kathimerini.gr/resources/issue-cover/%s.jpg' % time.strftime('%d-%m-%Y')