calibre/recipes/pilot.recipe

46 lines
1.8 KiB
Plaintext

from calibre.web.feeds.news import BasicNewsRecipe
class PilotOnline(BasicNewsRecipe):
title = u'Pilot Online'
language = 'en'
__author__ = 'Krittika Goyal'
oldest_article = 120 #days
max_articles_per_feed = 25
no_stylesheets = True
remove_tags_before = dict(name='div', attrs={'id':'wrapper'})
#remove_tags_after = dict(name='td', attrs={'class':'asset-bar'})
remove_tags = [
dict(name='iframe'),
dict(name='div', attrs={'class':['breadcrumb', 'asset-bar', 'footerLogos', 'footerLinks', 'lefty', 'righty', 'last', 'subFooter', 'footerLinks-last', 'comment clear-block', 'box', 'related', ]}),
dict(name='div', attrs={'id':['rightColumn']}),
dict(name='span', attrs={'class':'comment_forbidden'}),
dict(name='ul', attrs={'class':'links inline'}),
dict(name='p', attrs={'id':'commentadvisory'}),
dict(name='div', attrs={'style':['width: 300px; margin-right: 2em; float: left;']}),
dict(name='div', style="float:right; width: 300px;"),
dict(name='p', style="clear:both;"),
#dict(name='p', attrs={'name':'&lpos=footer_textlinks'}),
dict(name='span', text=':'),
]
feeds = [
('Hampton Roads',
'http://hamptonroads.com/taxterm/11'),
'http://hamptonroads.com/taxterm/12',
'http://hamptonroads.com/taxterm/222',
'http://hamptonroads.com/taxterm/224',
'http://hamptonroads.com/taxterm/226',
'http://hamptonroads.com/taxterm/3',
'http://hamptonroads.com/taxterm/149',
'http://hamptonroads.com/taxterm/228',
'http://hamptonroads.com/taxterm/4',
'http://hamptonroads.com/taxterm/522',
'http://hamptonroads.com/taxterm/420',
'http://hamptonroads.com/taxterm/227',
'http://hamptonroads.com/taxterm/231',
'http://hamptonroads.com/taxterm/232',
]