Update cgm.pl, historia.pl and tablety.pl

This commit is contained in:
Kovid Goyal 2011-11-14 09:47:00 +05:30
parent 4409c8c413
commit 798ad00029
3 changed files with 14 additions and 4 deletions

View File

@ -27,7 +27,7 @@ class CGM(BasicNewsRecipe):
del item['style'] del item['style']
ad=soup.findAll('a') ad=soup.findAll('a')
for r in ad: for r in ad:
if 'http://www.hustla.pl' in r['href']: if 'http://www.hustla.pl' in r['href'] or 'http://www.ebilet.pl' in r['href']:
r.extract() r.extract()
gallery=soup.find('div', attrs={'class':'galleryFlash'}) gallery=soup.find('div', attrs={'class':'galleryFlash'})
if gallery: if gallery:

View File

@ -8,6 +8,15 @@ class Historia_org_pl(BasicNewsRecipe):
category = 'history' category = 'history'
language = 'pl' language = 'pl'
oldest_article = 8 oldest_article = 8
remove_empty_feeds=True
max_articles_per_feed = 100 max_articles_per_feed = 100
feeds = [(u'Artykuły', u'http://www.historia.org.pl/index.php?format=feed&type=rss')] feeds = [(u'Wszystkie', u'http://www.historia.org.pl/index.php?format=feed&type=rss'),
(u'Wiadomości', u'http://www.historia.org.pl/index.php/wiadomosci.feed?type=rss'),
(u'Publikacje', u'http://www.historia.org.pl/index.php/publikacje.feed?type=rss'),
(u'Publicystyka', u'http://www.historia.org.pl/index.php/publicystyka.feed?type=rss'),
(u'Recenzje', u'http://historia.org.pl/index.php/recenzje.feed?type=rss'),
(u'Kultura i sztuka', u'http://www.historia.org.pl/index.php/kultura-i-sztuka.feed?type=rss'),
(u'Rekonstykcje', u'http://www.historia.org.pl/index.php/rekonstrukcje.feed?type=rss'),
(u'Projekty', u'http://www.historia.org.pl/index.php/projekty.feed?type=rss'),
(u'Konkursy'), (u'http://www.historia.org.pl/index.php/konkursy.feed?type=rss')]

View File

@ -9,6 +9,7 @@ class Tablety_pl(BasicNewsRecipe):
language = 'pl' language = 'pl'
oldest_article = 8 oldest_article = 8
max_articles_per_feed = 100 max_articles_per_feed = 100
keep_only_tags=[dict(name='header', attrs={'class':'entry-header'}), dict(name='div', attrs={'class':'entry-content clearfix'})] remove_tags_before=dict(name="h1", attrs={'class':'entry-title'})
remove_tags=[dict(name='div', attrs={'class':'snap_nopreview sharing robots-nocontent'}), dict(name='span', attrs={'class':'dsq-postid'})] remove_tags_after=dict(name="div", attrs={'class':'snap_nopreview sharing robots-nocontent'})
remove_tags=[dict(name='div', attrs={'class':'snap_nopreview sharing robots-nocontent'})]
feeds = [(u'Najnowsze posty', u'http://www.tablety.pl/feed/')] feeds = [(u'Najnowsze posty', u'http://www.tablety.pl/feed/')]