f1 ultra

recipes/f1_ultra.recipe

@@ -0,0 +1,35 @@
+from calibre.web.feeds.news import BasicNewsRecipe
+import re
+
+class f1ultra(BasicNewsRecipe):
+    title = u'Formuła 1 - F1 ultra'
+    __license__ = 'GPL v3'
+    __author__ = 'MrStefan <mrstefaan@gmail.com>, Artur Stachecki <artur.stachecki@gmail.com>'
+    language = 'pl'
+    description =u'Formuła 1, Robert Kubica, F3, GP2 oraz inne serie wyścigowe.'
+    masthead_url='http://www.f1ultra.pl/templates/f1ultra/images/logo.gif'
+    remove_empty_feeds= True
+    oldest_article = 1
+    max_articles_per_feed = 100
+    remove_javascript=True
+    no_stylesheets=True
+
+    keep_only_tags =[(dict(name = 'div', attrs = {'id' : 'main'}))]
+    remove_tags_after =[dict(attrs = {'style' : 'margin-top:5px;margin-bottom:5px;display: inline;'})]
+    remove_tags =[(dict(attrs = {'class' : ['buttonheading', 'avPlayerContainer', 'createdate']}))]
+    remove_tags.append(dict(attrs = {'title' : ['PDF', 'Drukuj', 'Email']}))
+    remove_tags.append(dict(name = 'form', attrs = {'method' : 'post'}))
+    remove_tags.append(dict(name = 'hr', attrs = {'size' : '2'}))
+    
+    preprocess_regexps = [(re.compile(r'align="left"'), lambda match: ''),
+		          (re.compile(r'align="right"'), lambda match: ''),
+		          (re.compile(r'width=\"*\"'), lambda match: ''),
+        		  (re.compile(r'\<table .*?\>'), lambda match: '')]
+  
+
+    extra_css = '''.contentheading { font-size: 1.4em; font-weight: bold; }
+	           img { display: block; clear: both;}
+	        '''
+    remove_attributes = ['width','height','position','float','padding-left','padding-right','padding','text-align']
+
+    feeds = [(u'F1 Ultra', u'http://www.f1ultra.pl/index.php?option=com_rd_rss&id=1&Itemid=245')]