recipe for Express.de - german

[schuster]

that's the complete site with all regions,
if no interest to köln or düsseldoft than kick it

class AdvancedUserRecipe1303841067(BasicNewsRecipe):

    title          = u'Express.de'
    __author__  = 'schuster'
    oldest_article = 2
    max_articles_per_feed = 50
    no_stylesheets         = True
    use_embedded_content   = False
    language               = 'de'
    extra_css = '''
                    h2{font-family:Arial,Helvetica,sans-serif; font-size: x-small;}
                    h1{ font-family:Arial,Helvetica,sans-serif;  font-size:x-large; font-weight:bold;}

                '''
    remove_javascript      = True
    remove_tags_befor = [dict(name='div', attrs={'class':'Datum'})]
    remove_tags_after = [dict(name='div', attrs={'class':'MoreNews'})]

    remove_tags = [dict(id='kalaydo'),
		   dict(id='Header'),
		   dict(id='Searchline'),
		   dict(id='MainNav'),
		   dict(id='Logo'),
		   dict(id='MainLinkSpacer'),
		   dict(id='MainLinks'),
		   dict(title='Diese Seite Bookmarken'),

                                           dict(name='span'),
		   dict(name='div', attrs={'class':'spacer_leftneu'}),
		   dict(name='div', attrs={'class':'button kalaydologo'}),
		   dict(name='div', attrs={'class':'button stellenneu'}),
		   dict(name='div', attrs={'class':'button autoneu'}),
		   dict(name='div', attrs={'class':'button immobilienneu'}),
		   dict(name='div', attrs={'class':'button kleinanzeigen'}),
		   dict(name='div', attrs={'class':'button tiereneu'}),
		   dict(name='div', attrs={'class':'button ferienwohnungen'}),
		   dict(name='div', attrs={'class':'button inserierenneu'}),
		   dict(name='div', attrs={'class':'spacer_rightneu'}),
		   dict(name='div', attrs={'class':'spacer_rightcorner'}),
		   dict(name='div', attrs={'class':'HeaderMetaNav'}),
		   dict(name='div', attrs={'class':'HeaderSearchOption'}),
		   dict(name='div', attrs={'class':'HeaderSearch'}),
		   dict(name='div', attrs={'class':'sbutton'}),
		   dict(name='div', attrs={'class':'active'}),

]


    def preprocess_html(self, soup):
        for alink in soup.findAll('a'):
            if alink.string is not None:
               tstr = alink.string
               alink.replaceWith(tstr)
        return soup

    feeds          = [(u'Top-Themen', u'http://www.express.de/home/-/2126/2126/-/view/asFeed/-/index.xml'),
                          (u'Regional - Köln', u'http://www.express.de/regional/koeln/-/2856/2856/-/view/asFeed/-/index.xml'),
                          (u'Regional - Bonn', u'http://www.express.de/regional/bonn/-/2860/2860/-/view/asFeed/-/index.xml'),
                          (u'Regional - Düsseldorf', u'http://www.express.de/regional/duesseldorf/-/2858/2858/-/view/asFeed/-/index.xml'),
                          (u'Regional - Region', u'http://www.express.de/regional/-/2178/2178/-/view/asFeed/-/index.xml'),
                          (u'Sport-News', u'http://www.express.de/sport/-/2176/2176/-/view/asFeed/-/index.xml'),
                          (u'Fussball-News', u'http://www.express.de/sport/fussball/-/3186/3186/-/view/asFeed/-/index.xml'),
                          (u'1.FC Köln News', u'http://www.express.de/sport/fussball/fc-koeln/-/3192/3192/-/view/asFeed/-/index.xml'),
                          (u'Alemannia Aachen News', u'http://www.express.de/sport/fussball/alemannia/-/3290/3290/-/view/asFeed/-/index.xml'),
                          (u'Borussia M~Gladbach', u'http://www.express.de/sport/fussball/gladbach/-/3286/3286/-/view/asFeed/-/index.xml'),
                          (u'Fortuna D~Dorf', u'http://www.express.de/sport/fussball/fortuna/-/3292/3292/-/view/asFeed/-/index.xml'),
                          (u'Basketball News', u'http://www.express.de/sport/basketball/-/3190/3190/-/view/asFeed/-/index.xml'),
                          (u'Big Brother', u'http://www.express.de/news/promi-show/big-brother/-/2402/2402/-/view/asFeed/-/index.xml'),



]

[schuster]

from calibre.web.feeds.news import BasicNewsRecipe
class AdvancedUserRecipe1303841067(BasicNewsRecipe):

    title          = u'Express.de'
    __author__  = 'schuster'
    oldest_article = 2
    max_articles_per_feed = 50
    no_stylesheets         = True
    use_embedded_content   = False
    language               = 'de'
    extra_css = '''
                    h2{font-family:Arial,Helvetica,sans-serif; font-size: x-small;}
                    h1{ font-family:Arial,Helvetica,sans-serif;  font-size:x-large; font-weight:bold;}
                '''
    remove_javascript      = True
    remove_tags_befor = [dict(name='div', attrs={'class':'Datum'})]
    remove_tags_after = [dict(name='div', attrs={'class':'MoreNews'})]

    remove_tags = [dict(id='kalaydo'),
           dict(id='Header'),
           dict(id='Searchline'),
           dict(id='MainNav'),
           dict(id='Logo'),
           dict(id='MainLinkSpacer'),
           dict(id='MainLinks'),
           dict(id='ContainerPfad'), #neu
           dict(title='Diese Seite Bookmarken'),

                                           dict(name='span'),
           dict(name='div', attrs={'class':'spacer_leftneu'}),
           dict(name='div', attrs={'class':'button kalaydologo'}),
           dict(name='div', attrs={'class':'button stellenneu'}),
           dict(name='div', attrs={'class':'button autoneu'}),
           dict(name='div', attrs={'class':'button immobilienneu'}),
           dict(name='div', attrs={'class':'button kleinanzeigen'}),
           dict(name='div', attrs={'class':'button tiereneu'}),
           dict(name='div', attrs={'class':'button ferienwohnungen'}),
           dict(name='div', attrs={'class':'button inserierenneu'}),
           dict(name='div', attrs={'class':'spacer_rightneu'}),
           dict(name='div', attrs={'class':'spacer_rightcorner'}),
           dict(name='div', attrs={'class':'HeaderMetaNav'}),
           dict(name='div', attrs={'class':'HeaderSearchOption'}),
           dict(name='div', attrs={'class':'HeaderSearch'}),
           dict(name='div', attrs={'class':'sbutton'}),
           dict(name='div', attrs={'class':'active'}),
           dict(name='div', attrs={'class':'MoreNews'}), #neu
           dict(name='div', attrs={'class':'ContentBoxSubline'}) #neu
]


    def preprocess_html(self, soup):
        for alink in soup.findAll('a'):
            if alink.string is not None:
               tstr = alink.string
               alink.replaceWith(tstr)
        return soup

    feeds          = [(u'Top-Themen', u'http://www.express.de/home/-/2126/2126/-/view/asFeed/-/index.xml'),
                          (u'Regional - Köln', u'http://www.express.de/regional/koeln/-/2856/2856/-/view/asFeed/-/index.xml'),
                          (u'Regional - Bonn', u'http://www.express.de/regional/bonn/-/2860/2860/-/view/asFeed/-/index.xml'),
                          (u'Regional - Düsseldorf', u'http://www.express.de/regional/duesseldorf/-/2858/2858/-/view/asFeed/-/index.xml'),
                          (u'Regional - Region', u'http://www.express.de/regional/-/2178/2178/-/view/asFeed/-/index.xml'),
                          (u'Sport-News', u'http://www.express.de/sport/-/2176/2176/-/view/asFeed/-/index.xml'),
                          (u'Fussball-News', u'http://www.express.de/sport/fussball/-/3186/3186/-/view/asFeed/-/index.xml'),
                          (u'1.FC Köln News', u'http://www.express.de/sport/fussball/fc-koeln/-/3192/3192/-/view/asFeed/-/index.xml'),
                          (u'Alemannia Aachen News', u'http://www.express.de/sport/fussball/alemannia/-/3290/3290/-/view/asFeed/-/index.xml'),
                          (u'Borussia M~Gladbach', u'http://www.express.de/sport/fussball/gladbach/-/3286/3286/-/view/asFeed/-/index.xml'),
                          (u'Fortuna D~Dorf', u'http://www.express.de/sport/fussball/fortuna/-/3292/3292/-/view/asFeed/-/index.xml'),
                          (u'Basketball News', u'http://www.express.de/sport/basketball/-/3190/3190/-/view/asFeed/-/index.xml'),
                          (u'Big Brother', u'http://www.express.de/news/promi-show/big-brother/-/2402/2402/-/view/asFeed/-/index.xml'),



]