Update Tagesanzeiger

recipes/tagesan.recipe

@@ -1,45 +1,51 @@
 from calibre.web.feeds.news import BasicNewsRecipe
 
-class AdvancedUserRecipe1284927619(BasicNewsRecipe):
-    title = u'Tagesanzeiger'
-    publisher = u'Tamedia AG'
-    oldest_article = 2
-    __author__ = 'noxxx'
-    max_articles_per_feed = 100
-    description = 'tagesanzeiger.ch: Nichts verpassen'
-    category = 'News, Politik, Nachrichten, Schweiz, Zuerich'
-    language = 'de'
 
-    conversion_options = {
+class AdvancedUserRecipe1450864843(BasicNewsRecipe):
-                             'comments'  : description
+    title = u'Tagesanzeiger'
-                            ,'tags'      : category
+    __author__ = 'noxxx'
-                            ,'language'  : language
+    language = 'de'
-                            ,'publisher' : publisher
+    oldest_article = 1
+    max_articles_per_feed = 8
+    no_stylesheets = True
+    auto_cleanup = True
+    scale_news_images_to_device = True
+    compress_news_images = True
+    remove_javascript = True
+    encoding = 'utf-8'
+    remove_empty_feeds = True
+    ignore_duplicate_articles = {'title', 'url'}
+
+    feeds = [
+        (u'Front', u'http://www.tagesanzeiger.ch/rss.html'),
+        (u'Zurich', u'http://www.tagesanzeiger.ch/zuerich/rss.html'),
+        (u'Schweiz', u'http://www.tagesanzeiger.ch/schweiz/rss.html'),
+        (u'International', u'http://www.tagesanzeiger.ch/ausland/rss.html'),
+        (u'Wirtschaft', u'http://www.tagesanzeiger.ch/wirtschaft/rss.html'),
+        (u'Sport', u'http://www.tagesanzeiger.ch/sport/rss.html'),
+        (u'Kultur', u'http://www.tagesanzeiger.ch/kultur/rss.html'),
+        (u'Leben', u'http://www.tagesanzeiger.ch/leben/rss.html'),
+        (u'Panorama', u'http://www.tagesanzeiger.ch/panorama/rss.html'),
+        (u'Digital', u'http://www.tagesanzeiger.ch/digital/rss.html')
+    ]
+
+    keep_only_tags = [dict(name='h1'),
+                      dict(name='h3'),
+                      dict(name='div', attrs={'id': 'topElement'}),
+                      dict(name='div', attrs={'id': 'mainContent'})
+                      ]
+
+    auto_cleanup_keep = {'//div[@id="article-image"]',
+                         '//*[@class="standard"]'
                          }
 
-    remove_tags = [
+    conversion_options = {'base_font_size': 12}
-     dict(name='img')
-                    ,dict(name='div',attrs={'class':['swissquote ad','boxNews','centerAD','contentTabs2','sbsLabel']})
-                    ,dict(name='div',attrs={'id':['colRightAd','singleRight','singleSmallRight','MailInfo','metaLine','sidebarSky','contentFooter','commentInfo','commentInfo2','commentInfo3','footerBottom','clear','boxExclusiv','singleLogo','navSearch','headerLogin','headerBottomRight','horizontalNavigation','subnavigation','googleAdSense','footerAd','contentbox','articleGalleryNav']})
-    ,dict(name='form',attrs={'id':['articleMailForm','commentform']})
-    ,dict(name='div',attrs={'style':['position:absolute']})
-    ,dict(name='script',attrs={'type':['text/javascript']})
-    ,dict(name='p',attrs={'class':['schreiben','smallPrint','charCounter','caption']})
-     ]
-    feeds = [
-     (u'Front', u'http://www.tagesanzeiger.ch/rss.html')
-    ,(u'Zürich', u'http://www.tagesanzeiger.ch/zuerich/rss.html')
-    ,(u'Schweiz', u'http://www.tagesanzeiger.ch/schweiz/rss.html')
-    ,(u'Ausland', u'http://www.tagesanzeiger.ch/ausland/rss.html')
-    ,(u'Digital', u'http://www.tagesanzeiger.ch/digital/rss.html')
-    ,(u'Wissen', u'http://www.tagesanzeiger.ch/wissen/rss.html')
-    ,(u'Panorama', u'http://www.tagesanzeiger.ch/panorama/rss.html')
-    ,(u'Wirtschaft', u'http://www.tagesanzeiger.ch/wirtschaft/rss.html')
-    ,(u'Sport', u'http://www.tagesanzeiger.ch/sport/rss.html')
-    ,(u'Kultur', u'http://www.tagesanzeiger.ch/kultur/rss.html')
-    ,(u'Leben', u'http://www.tagesanzeiger.ch/leben/rss.html')
-    ,(u'Auto', u'http://www.tagesanzeiger.ch/auto/rss.html')]
 
-    def print_version(self, url):
+    extra_css = 'h1 { font-size: 150%; font-weight: bold; } \
-        return url + '/print.html'
+                 h2 { font-size: 120%; font-weight: bold; } \
+                 h3 { font-size: 100%; font-style: italic; } \
+                 h4, h5, h6 { font-size: 100%; font-style: italic; }'
 
+    def get_cover_url(self):
+        self.cover_url = 'http://newsmail.cleartext.ch/referenz_tagesanzeiger.jpg'
+        return getattr(self, 'cover_url', None)