from calibre.web.feeds.news import BasicNewsRecipe
import re
# History:
# 1: Base Version
# 2: Added rules for wdr.de, ndr.de, br-online.de
# 3: Added rules for rbb-online.de, boerse.ard.de, sportschau.de
# 4: New design of tagesschau.de implemented. Simplyfied.
# 5: Taken out the pictures.


class Tagesschau(BasicNewsRecipe):
    title = 'Tagesschau'
    description = 'Nachrichten der ARD'
    publisher = 'ARD'
    language = 'de'
    version = 5

    __author__ = 'Florian Andreas Pfaff, a.peter'
    oldest_article = 7
    max_articles_per_feed = 100
    no_stylesheets = True
    remove_javascript = True

    feeds = [('Tagesschau', 'http://www.tagesschau.de/xml/rss2')]

    remove_tags = [
        dict(attrs={'class': ['socialMedia', 'mediaLink', 'mediaInfo',
                              'linklist teaserImTeaser', 'modCon modConComments']}),
    ]

    keep_only_tags = [
        dict(name='div', attrs={
             'class': ['section sectionZ', 'section sectionZ sectionArticle']}),
        dict(name='div', attrs={'class': re.compile(r'.*containerArticle.*')})
    ]

    def get_article_url(self, article):
        ans = BasicNewsRecipe.get_article_url(self, article)
        if '/multimedia/' in ans:
            ans = None
        return ans