mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-07-09 03:04:10 -04:00
Fix Austin Statesman
This commit is contained in:
parent
cb7814dfd5
commit
ce2f9396b8
@ -10,12 +10,14 @@ class AdvancedUserRecipe1278049615(BasicNewsRecipe):
|
||||
|
||||
max_articles_per_feed = 100
|
||||
|
||||
feeds = [(u'News', u'http://www.statesman.com/section-rss.do?source=news&includeSubSections=true'),
|
||||
(u'Business', u'http://www.statesman.com/section-rss.do?source=business&includeSubSections=true'),
|
||||
(u'Life', u'http://www.statesman.com/section-rss.do?source=life&includesubsection=true'),
|
||||
(u'Editorial', u'http://www.statesman.com/section-rss.do?source=opinion&includesubsections=true'),
|
||||
(u'Sports', u'http://www.statesman.com/section-rss.do?source=sports&includeSubSections=true')
|
||||
]
|
||||
feeds = [(u'News',
|
||||
u'http://www.statesman.com/section-rss.do?source=news&includeSubSections=true'),
|
||||
(u'Local', u'http://www.statesman.com/section-rss.do?source=local&includeSubSections=true'),
|
||||
(u'Business', u'http://www.statesman.com/section-rss.do?source=business&includeSubSections=true'),
|
||||
(u'Life', u'http://www.statesman.com/section-rss.do?source=life&includesubsection=true'),
|
||||
(u'Editorial', u'http://www.statesman.com/section-rss.do?source=opinion&includesubsections=true'),
|
||||
(u'Sports', u'http://www.statesman.com/section-rss.do?source=sports&includeSubSections=true')
|
||||
]
|
||||
masthead_url = "http://www.statesman.com/images/cmg-logo.gif"
|
||||
#temp_files = []
|
||||
#articles_are_obfuscated = True
|
||||
@ -28,8 +30,11 @@ class AdvancedUserRecipe1278049615(BasicNewsRecipe):
|
||||
conversion_options = {'linearize_tables':True}
|
||||
remove_tags = [
|
||||
dict(name='div', attrs={'id':'cxArticleOptions'}),
|
||||
{'class':['perma', 'comments', 'trail', 'share-buttons',
|
||||
'toggle_show_on']},
|
||||
]
|
||||
keep_only_tags = [
|
||||
dict(name='div', attrs={'class':'cxArticleHeader'}),
|
||||
dict(name='div', attrs={'id':'cxArticleBodyText'}),
|
||||
dict(name='div', attrs={'class':'cxArticleHeader'}),
|
||||
dict(name='div', attrs={'id':['cxArticleBodyText',
|
||||
'content']}),
|
||||
]
|
||||
|
Loading…
x
Reference in New Issue
Block a user