recipes: remove some garbage from alejakomiksu

This commit is contained in:
Tomasz Długosz 2016-10-07 21:45:22 +02:00
parent f00b58c766
commit d41b8fc707

View File

@ -10,8 +10,6 @@ class AlejaKomiksu(BasicNewsRecipe):
category = 'comics'
language = 'pl'
extra_css = 'ul {list-style-type: none;} .gfx_news {float: right;}'
preprocess_regexps = [(re.compile(ur'((<li class="no_img_b">(Do poczytania)|(Nowości):</li>)|(<p class="head2">Komentarze</p>)).*</body>',
re.DOTALL | re.IGNORECASE), lambda match: '</body>')]
cover_url = 'http://www.alejakomiksu.com/gfx/build/logo.png'
masthead_url = 'http://www.alejakomiksu.com/gfx/build/logo.png'
use_embedded_content = False
@ -23,7 +21,7 @@ class AlejaKomiksu(BasicNewsRecipe):
remove_attributes = ['style', 'font']
ignore_duplicate_articles = {'title', 'url'}
keep_only_tags = [dict(attrs={'class': 'cont_tresc'})]
keep_only_tags = dict(attrs={'class': ['akNews__header','akNews__body']})
feeds = [(u'Wiadomości', 'http://www.alejakomiksu.com/rss.php5')]