mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-07-09 03:04:10 -04:00
Update El Correo
This commit is contained in:
Kovid Goyal
parent
a585b65f3b
commit
25e60aa62a
@ -1,10 +1,10 @@
|
|||||||
#!/usr/bin/env python2
|
#!/usr/bin/env python2
|
||||||
__license__ = 'GPL v3'
|
__license__ = 'GPL v3'
|
||||||
__copyright__ = '08 Januery 2011, desUBIKado'
|
__copyright__ = '08 Januery 2011, desUBIKado'
|
||||||
__author__ = 'desUBIKado'
|
__author__ = 'desUBIKado'
|
||||||
__description__ = 'Daily newspaper from Biscay'
|
__description__ = 'Daily newspaper from Biscay'
|
||||||
__version__ = 'v0.11'
|
__version__ = 'v0.13'
|
||||||
__date__ = '26, July 2014'
|
__date__ = '28, July 2016'
|
||||||
'''
|
'''
|
||||||
http://www.elcorreo.com/
|
http://www.elcorreo.com/
|
||||||
'''
|
'''
|
||||||
@ -32,8 +32,8 @@ class elcorreo(BasicNewsRecipe):
|
|||||||
remove_javascript = True
|
remove_javascript = True
|
||||||
|
|
||||||
feeds = [
|
feeds = [
|
||||||
(u'Portada', u'http://www.elcorreo.com/bizkaia/rss/atom/portada'),
|
(u'Portada', u'http://www.elcorreo.com/bizkaia/rss/atom/portada'),
|
||||||
(u'Mundo', u'http://www.elcorreo.com/bizkaia/rss/atom?seccion=internacional'),
|
(u'Mundo', u'http://www.elcorreo.com/bizkaia/rss/atom?seccion=internacional'),
|
||||||
(u'Bizkaia', u'http://www.elcorreo.com/bizkaia/rss/atom?seccion=bizkaia'),
|
(u'Bizkaia', u'http://www.elcorreo.com/bizkaia/rss/atom?seccion=bizkaia'),
|
||||||
(u'Guipuzkoa', u'http://www.elcorreo.com/bizkaia/rss/atom?seccion=gipuzkoa'),
|
(u'Guipuzkoa', u'http://www.elcorreo.com/bizkaia/rss/atom?seccion=gipuzkoa'),
|
||||||
(u'Araba', u'http://www.elcorreo.com/bizkaia/rss/atom?seccion=araba'),
|
(u'Araba', u'http://www.elcorreo.com/bizkaia/rss/atom?seccion=araba'),
|
||||||
@ -58,93 +58,100 @@ class elcorreo(BasicNewsRecipe):
|
|||||||
]
|
]
|
||||||
|
|
||||||
keep_only_tags = [
|
keep_only_tags = [
|
||||||
dict(name='ul', attrs={'class':['media-list']})
|
dict(name='article', attrs={'class':['story media-list ']})
|
||||||
]
|
]
|
||||||
|
|
||||||
remove_tags = [
|
remove_tags = [
|
||||||
dict(name='span', attrs={'class':['no-comments']})
|
dict(name='span', attrs={'class':['no-comments']}),
|
||||||
|
dict(name='div', attrs={'class':['compApoyosText compNoticiasR']})
|
||||||
]
|
]
|
||||||
|
|
||||||
remove_tags_before = dict(name='ul' , attrs={'class':'media-list'})
|
remove_tags_before = dict(name='article' , attrs={'class':'story media-list '})
|
||||||
remove_tags_after = dict(name='ul' , attrs={'class':'media-list'})
|
remove_tags_after = dict(name='article' , attrs={'class':'story media-list '})
|
||||||
|
|
||||||
|
|
||||||
# Usamos la versión para móviles
|
# Usamos la versión para móviles
|
||||||
|
|
||||||
def print_version(self, url):
|
def print_version(self, url):
|
||||||
return url.replace('http://www.', 'http://m.')
|
return url.replace('http://www.', 'http://m.')
|
||||||
|
|
||||||
|
|
||||||
|
|
||||||
_processed_links = []
|
_processed_links = []
|
||||||
|
|
||||||
def get_article_url(self, article):
|
def get_article_url(self, article):
|
||||||
|
|
||||||
link = article.get('link', None)
|
link = article.get('link', None)
|
||||||
|
|
||||||
if link is None:
|
if link is None:
|
||||||
return article
|
return article
|
||||||
|
|
||||||
# modificamos la url de las noticias de los equipos deportivos para que funcionen, por ejemplo:
|
# modificamos la url de las noticias de los equipos deportivos para que funcionen, por ejemplo:
|
||||||
# http://athletic.elcorreo.com/noticias/201407/27/muniain-estrella-athletic-para-20140727093046.html
|
# http://athletic.elcorreo.com/noticias/201407/27/muniain-estrella-athletic-para-20140727093046.html
|
||||||
# http://m.elcorreo.com/noticias/201407/27/muniain-estrella-athletic-para-20140727093046.html?external=deportes/athletic
|
# http://m.elcorreo.com/noticias/201407/27/muniain-estrella-athletic-para-20140727093046.html?external=deportes/athletic
|
||||||
|
|
||||||
parte = link.split('/')
|
parte = link.split('/')
|
||||||
|
|
||||||
if parte[2] == 'athletic.elcorreo.com':
|
if parte[2] == 'athletic.elcorreo.com':
|
||||||
link = 'http://m.elcorreo.com/' + parte[3] + '/' + parte[4] + '/' + parte[5] + '/' + parte[6] + '?external=deportes/athletic'
|
link = 'http://m.elcorreo.com/' + parte[3] + '/' + parte[4] + '/' + parte[5] + '/' + parte[6] + '?external=deportes/athletic'
|
||||||
else:
|
else:
|
||||||
if parte[2] == 'baskonia.elcorreo.com':
|
if parte[2] == 'baskonia.elcorreo.com':
|
||||||
link = 'http://m.elcorreo.com/' + parte[3] + '/' + parte[4] + '/' + parte[5] + '/' + parte[6] + '?external=deportes/baskonia'
|
link = 'http://m.elcorreo.com/' + parte[3] + '/' + parte[4] + '/' + parte[5] + '/' + parte[6] + '?external=deportes/baskonia'
|
||||||
else:
|
else:
|
||||||
if parte[2] == 'bilbaobasket.elcorreo.com':
|
if parte[2] == 'bilbaobasket.elcorreo.com':
|
||||||
link = 'http://m.elcorreo.com/' + parte[3] + '/' + parte[4] + '/' + parte[5] + '/' + parte[6] + '?external=deportes/bilbaobasket'
|
link = 'http://m.elcorreo.com/' + parte[3] + '/' + parte[4] + '/' + parte[5] + '/' + parte[6] + '?external=deportes/bilbaobasket'
|
||||||
else:
|
else:
|
||||||
if parte[2] == 'alaves.elcorreo.com':
|
if parte[2] == 'alaves.elcorreo.com':
|
||||||
link = 'http://m.elcorreo.com/' + parte[3] + '/' + parte[4] + '/' + parte[5] + '/' + parte[6] + '?external=deportes/alaves'
|
link = 'http://m.elcorreo.com/' + parte[3] + '/' + parte[4] + '/' + parte[5] + '/' + parte[6] + '?external=deportes/alaves'
|
||||||
|
|
||||||
# A veces el mismo articulo aparece en la versión de Alava y en la de Bizkaia. Por ejemplo:
|
|
||||||
# http://www.elcorreo.com/alava/deportes/motor/formula-1/201407/27/ecclestone-quiere-briatore-ayude-20140727140820-rc.html
|
|
||||||
# http://www.elcorreo.com/bizkaia/deportes/motor/formula-1/201407/27/ecclestone-quiere-briatore-ayude-20140727140820-rc.html
|
|
||||||
# para controlar los duplicados, unificamos las url para que sean siempre de bizkaia (excepto para la sección "araba")
|
|
||||||
|
|
||||||
if ((parte[3] == 'alava') and (parte[4] != 'araba')):
|
# A veces el mismo articulo aparece en la versión de Alava y en la de Bizkaia. Por ejemplo:
|
||||||
link = link.replace('elcorreo.com/alava', 'elcorreo.com/bizkaia')
|
# http://www.elcorreo.com/alava/deportes/motor/formula-1/201407/27/ecclestone-quiere-briatore-ayude-20140727140820-rc.html
|
||||||
|
# http://www.elcorreo.com/bizkaia/deportes/motor/formula-1/201407/27/ecclestone-quiere-briatore-ayude-20140727140820-rc.html
|
||||||
|
# para controlar los duplicados, unificamos las url para que sean siempre de bizkaia (excepto para la sección "araba")
|
||||||
|
|
||||||
# Controlamos si el artículo ha sido incluido en otro feed para eliminarlo
|
if ((parte[3] == 'alava') and (parte[4] != 'araba')):
|
||||||
|
link = link.replace('elcorreo.com/alava', 'elcorreo.com/bizkaia')
|
||||||
|
|
||||||
if not (link in self._processed_links):
|
|
||||||
|
# Controlamos si el artículo ha sido incluido en otro feed para eliminarlo
|
||||||
|
|
||||||
|
if not (link in self._processed_links):
|
||||||
self._processed_links.append(link)
|
self._processed_links.append(link)
|
||||||
else:
|
else:
|
||||||
link = None
|
link = None
|
||||||
|
|
||||||
return link
|
|
||||||
|
return link
|
||||||
|
|
||||||
# Recuperamos la portada de papel (la imagen format=1 tiene mayor resolucion)
|
# Recuperamos la portada de papel (la imagen format=1 tiene mayor resolucion)
|
||||||
|
|
||||||
def get_cover_url(self):
|
def get_cover_url(self):
|
||||||
cover = None
|
cover = None
|
||||||
st = time.localtime()
|
st = time.localtime()
|
||||||
year = str(st.tm_year)
|
year = str(st.tm_year)
|
||||||
month = "%.2d" % st.tm_mon
|
month = "%.2d" % st.tm_mon
|
||||||
day = "%.2d" % st.tm_mday
|
day = "%.2d" % st.tm_mday
|
||||||
# http://info.elcorreo.com/pdf/07082013-viz.pdf
|
#http://info.elcorreo.com/pdf/07082013-viz.pdf
|
||||||
cover='http://info.elcorreo.com/pdf/'+ day + month + year +'-viz.pdf'
|
cover='http://info.elcorreo.com/pdf/'+ day + month + year +'-viz.pdf'
|
||||||
br = BasicNewsRecipe.get_browser(self)
|
br = BasicNewsRecipe.get_browser(self)
|
||||||
try:
|
try:
|
||||||
br.open(cover)
|
br.open(cover)
|
||||||
except:
|
except:
|
||||||
self.log("\nPortada no disponible")
|
self.log("\nPortada no disponible")
|
||||||
cover ='http://www.elcorreo.com/vizcaya/noticias/201002/02/Media/logo-elcorreo-nuevo.png'
|
cover ='http://www.elcorreo.com/vizcaya/noticias/201002/02/Media/logo-elcorreo-nuevo.png'
|
||||||
return cover
|
return cover
|
||||||
|
|
||||||
# Para cambiar el estilo del texto
|
# Para cambiar el estilo del texto
|
||||||
|
|
||||||
extra_css = '''
|
extra_css = '''
|
||||||
h4 {font-family:Arial,Helvetica,sans-serif; font-weight:bold;font-size:28px;}
|
h4 {font-family:Arial,Helvetica,sans-serif; font-weight:bold;font-size:28px;}
|
||||||
.place {font-family:Arial,Helvetica,sans-serif; font-weight:normal;font-size:12px;}
|
.place {font-family:Arial,Helvetica,sans-serif; font-weight:normal;font-size:12px;}
|
||||||
.name {font-family:Arial,Helvetica,sans-serif; font-weight:bold;font-size:12px;}
|
.name {font-family:Arial,Helvetica,sans-serif; font-weight:bold;font-size:12px;}
|
||||||
.overhead, .compLadillo, description {font-family:georgia,serif; font-weight:bold;font-size:18px;}
|
.overhead, .compLadillo, description {font-family:georgia,serif; font-weight:bold;font-size:18px;}
|
||||||
.compSumario, .detalle-titular {font-family:georgia,serif; font-style:italic; font-weight:normal;font-size:22px;color:#4D4D4D;}
|
.compSumario, .detalle-titular {font-family:georgia,serif; font-style:italic; font-weight:normal;font-size:22px;color:#4D4D4D;}
|
||||||
img{margin-bottom: 0.4em}
|
img{margin-bottom: 0.4em}
|
||||||
'''
|
'''
|
||||||
|
|
||||||
preprocess_regexps = [
|
preprocess_regexps = [
|
||||||
|
|
||||||
|
|||||||
Loading…
x
Reference in New Issue
Block a user