Update Heraldo.es

This commit is contained in:
Kovid Goyal 2012-06-30 20:02:50 +05:30
parent 50a13872c8
commit 39e8b5305c

View File

@ -3,8 +3,8 @@ __license__ = 'GPL v3'
__copyright__ = '04 December 2010, desUBIKado' __copyright__ = '04 December 2010, desUBIKado'
__author__ = 'desUBIKado' __author__ = 'desUBIKado'
__description__ = 'Daily newspaper from Aragon' __description__ = 'Daily newspaper from Aragon'
__version__ = 'v0.04' __version__ = 'v0.05'
__date__ = '6, Januery 2011' __date__ = '5, Februery 2012'
''' '''
[url]http://www.heraldo.es/[/url] [url]http://www.heraldo.es/[/url]
''' '''
@ -38,7 +38,7 @@ class heraldo(BasicNewsRecipe):
keep_only_tags = [dict(name='div', attrs={'id':['dts','com']})] keep_only_tags = [dict(name='div', attrs={'id':['dts','com']})]
remove_tags = [dict(name='a', attrs={'class':['com flo-r','enl-if','enl-df']}), remove_tags = [dict(name='a', attrs={'class':['com flo-r','enl-if','enl-df']}),
dict(name='div', attrs={'class':['brb-b-s con marg-btt','cnt-rel con']}), dict(name='div', attrs={'class':['brb-b-s con marg-btt','cnt-rel con','col5-f1']}),
dict(name='form', attrs={'class':'form'}), dict(name='form', attrs={'class':'form'}),
dict(name='ul', attrs={'id':['cont-tags','pag-1']})] dict(name='ul', attrs={'id':['cont-tags','pag-1']})]
@ -72,6 +72,9 @@ class heraldo(BasicNewsRecipe):
preprocess_regexps = [ preprocess_regexps = [
# To separate the comments with a blank line # Para separar los comentarios con una linea en blanco
(re.compile(r'<div id="com"', re.DOTALL|re.IGNORECASE), lambda match: '<br><div id="com"') (re.compile(r'<div id="com"', re.DOTALL|re.IGNORECASE), lambda match: '<br><div id="com"')
] ]