Tweakers.net by Roedi06

2025-07-09 03:04:10 -04:00 · 2012-01-17 18:12:45 +05:30 · 2012-01-17 18:12:45 +05:30 · 1a361f694f
commit 1a361f694f
parent ffeb865728
1 changed files with 66 additions and 0 deletions
--- a/recipes/tweakers_net.recipe
+++ b/recipes/tweakers_net.recipe
@ -0,0 +1,66 @@
+#!/usr/bin/env python
+# vim:fileencoding=UTF-8:ts=4:sw=4:sta:et:sts=4:ai
+from __future__ import with_statement
+
+__license__   = 'GPL v3'
+__docformat__ = 'restructuredtext en'
+
+import re
+from calibre.web.feeds.news import BasicNewsRecipe
+
+class Tweakers(BasicNewsRecipe):
+     title          = u'Tweakers.net - with Reactions'
+     __author__     = 'Roedi06'
+     language       = 'nl'
+     oldest_article = 7
+     max_articles_per_feed = 100
+     cover_url       = 'http://img51.imageshack.us/img51/7470/tweakersnetebook.gif'
+
+     keep_only_tags = [dict(name='div', attrs={'class':'columnwrapper news'}),
+    {'id':'reacties'},
+      ]
+
+     remove_tags    = [dict(name='div', attrs={'id' : ['utracker']}),
+                        {'id' : ['channelNav']},
+                        {'id' : ['contentArea']},
+                        {'class' : ['breadCrumb']},
+                        {'class' : ['nextPrevious ellipsis']},
+                        {'class' : ['advertorial']},
+                        {'class' : ['sidebar']},
+                        {'class' : ['filterBox']},
+                        {'id' : ['toggleButtonTxt']},
+                        {'id' : ['socialButtons']},
+                        {'class' : ['button']},
+                        {'class' : ['textadTop']},
+                        {'class' : ['commentLink']},
+                        {'title' : ['Reageer op deze reactie']},
+                        {'class' : ['pageIndex']},
+        {'class' : ['reactieHeader collapsed']},
+                      ]
+     no_stylesheets=True
+
+     preprocess_regexps = [
+        (re.compile(r'<hr*?>', re.IGNORECASE | re.DOTALL), lambda match : ''),
+        (re.compile(r'<p>', re.IGNORECASE | re.DOTALL), lambda match : ''),
+        (re.compile(r'</p>', re.IGNORECASE | re.DOTALL), lambda match : ''),
+        (re.compile(r'<a.*?>'), lambda h1: '<b><u>'),
+        (re.compile(r'</a>'), lambda h2: '</u></b>'),
+        (re.compile(r'<span class="new">', re.IGNORECASE | re.DOTALL), lambda match : ''),
+        (re.compile(r'</span>', re.IGNORECASE | re.DOTALL), lambda match : ''),
+        (re.compile(r'<div class="moderation"><img src="http://tweakimg.net/g/if/comments/score_0'), lambda match : ' - moderated 0<div class="moderation"><img src="http://tweakimg.net/g/if/comments/score_0'),
+        (re.compile(r'<div class="moderation"><img src="http://tweakimg.net/g/if/comments/score_1'), lambda match : ' - moderated +1<div class="moderation"><img src="http://tweakimg.net/g/if/comments/score_1'),
+        (re.compile(r'<div class="moderation"><img src="http://tweakimg.net/g/if/comments/score_2'), lambda match : ' - moderated +2<div class="moderation"><img src="http://tweakimg.net/g/if/comments/score_2'),
+        (re.compile(r'<div class="moderation"><img src="http://tweakimg.net/g/if/comments/score_3'), lambda match : ' - moderated +3<div class="moderation"><img src="http://tweakimg.net/g/if/comments/score_3'),
+        (re.compile(r'<div class="moderation">.*?</div>'), lambda h1: ''),
+     ]
+
+     extra_css = '.reactieHeader { color: #333333; font-size: 6px; border-bottom:solid 2px #333333; border-top:solid 1px #333333; } \
+       .reactieContent { font-family:"Times New Roman",Georgia,Serif; color: #000000; font-size: 8px; } \
+       .quote { font-family:"Times New Roman",Georgia,Serif; padding-left:2px; border-left:solid 3px #666666; color: #666666; }'
+
+
+     feeds          = [(u'Tweakers.net', u'http://feeds.feedburner.com/tweakers/nieuws')]
+
+     def print_version(self, url):
+        return url + '?max=200'
+