mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-08-11 09:13:57 -04:00
Update Deutsche Welle (English and Spanish editions)
This commit is contained in:
Kovid Goyal
parent
81182e7ae8
commit
1bac058322
@ -1,66 +1,73 @@
|
|||||||
|
#!/usr/bin/env python2
|
||||||
|
# vim:fileencoding=utf-8
|
||||||
|
from __future__ import unicode_literals, division, absolute_import, print_function
|
||||||
|
|
||||||
__license__ = 'GPL v3'
|
__license__ = 'GPL v3'
|
||||||
__copyright__ = '2010, Darko Miletic <darko.miletic at gmail.com>'
|
__copyright__ = '2010, Darko Miletic <darko.miletic at gmail.com>'
|
||||||
|
|
||||||
'''
|
'''
|
||||||
dw-world.de
|
Deutsche Welle (english) - dw.com/en
|
||||||
'''
|
'''
|
||||||
|
|
||||||
|
import re
|
||||||
from calibre.web.feeds.news import BasicNewsRecipe
|
from calibre.web.feeds.news import BasicNewsRecipe
|
||||||
|
|
||||||
class DeutscheWelle_en(BasicNewsRecipe):
|
class DeutscheWelle_en(BasicNewsRecipe):
|
||||||
title = 'Deutsche Welle'
|
title = 'Deutsche Welle'
|
||||||
__author__ = 'Darko Miletic'
|
__author__ = 'Darko Miletic'
|
||||||
description = 'News from Germany and World'
|
description = 'News from Germany and the world'
|
||||||
publisher = 'Deutsche Welle'
|
publisher = 'Deutsche Welle'
|
||||||
category = 'news, politics, Germany'
|
language = 'en'
|
||||||
oldest_article = 1
|
|
||||||
max_articles_per_feed = 100
|
|
||||||
use_embedded_content = False
|
|
||||||
no_stylesheets = True
|
|
||||||
language = 'en'
|
|
||||||
publication_type = 'newsportal'
|
|
||||||
remove_empty_feeds = True
|
|
||||||
masthead_url = 'http://www.dw-world.de/skins/std/channel1/pics/dw_logo1024.gif'
|
|
||||||
extra_css = """
|
|
||||||
body{font-family: Arial,sans-serif}
|
|
||||||
img{margin-top: 0.5em; margin-bottom: 0.2em; display: block}
|
|
||||||
.caption{font-size: x-small; display: block; margin-bottom: 0.4em}
|
|
||||||
"""
|
|
||||||
|
|
||||||
|
oldest_article = 1
|
||||||
|
max_articles_per_feed = 50
|
||||||
|
no_stylesheets = True
|
||||||
|
remove_javascript = True
|
||||||
|
remove_empty_feeds = True
|
||||||
|
ignore_duplicate_articles = {'title', 'url'}
|
||||||
|
|
||||||
conversion_options = {
|
feeds = [
|
||||||
'comment' : description
|
('Top Stories', 'http://rss.dw-world.de/rdf/rss-en-top'),
|
||||||
, 'tags' : category
|
('World', 'http://rss.dw.de/rdf/rss-en-world'),
|
||||||
, 'publisher': publisher
|
('Germany', 'http://rss.dw.de/rdf/rss-en-ger'),
|
||||||
, 'language' : language
|
('Europe', 'http://rss.dw.de/rdf/rss-en-eu'),
|
||||||
}
|
('Business', 'http://rss.dw.de/rdf/rss-en-bus'),
|
||||||
|
('Culture & Lifestyle', 'http://rss.dw.de/rdf/rss-en-cul'),
|
||||||
|
('Sports', 'http://rss.dw.de/rdf/rss-en-sports'),
|
||||||
|
('Visit Germany', 'http://rss.dw.de/rdf/rss-en-visitgermany'),
|
||||||
|
('Asia', 'http://rss.dw.de/rdf/rss-en-asia')
|
||||||
|
]
|
||||||
|
|
||||||
|
keep_only_tags=[
|
||||||
|
dict(name='div', attrs={'class':'col3'})
|
||||||
|
]
|
||||||
|
|
||||||
|
remove_tags_after = [
|
||||||
|
dict(name='div', attrs={'class':'group'})
|
||||||
|
]
|
||||||
|
|
||||||
remove_tags = [
|
remove_tags = [
|
||||||
dict(name=['iframe','embed','object','form','base','meta','link'])
|
dict(name='div', attrs={'class':'col1'}),
|
||||||
,dict(attrs={'class':'actionFooter'})
|
dict(name='div', attrs={'class':re.compile('gallery')}),
|
||||||
]
|
dict(name='div', attrs={'class':re.compile('audio')}),
|
||||||
keep_only_tags=[dict(attrs={'class':'ArticleDetail detail'})]
|
dict(name='div', attrs={'class':re.compile('video')})
|
||||||
remove_attributes = ['height','width','onclick','border','lang']
|
]
|
||||||
|
|
||||||
feeds = [(u'All news', u'http://rss.dw-world.de/rdf/rss-en-all')]
|
remove_attributes = ['height', 'width', 'onclick', 'border', 'lang', 'link']
|
||||||
|
|
||||||
def print_version(self, url):
|
extra_css = '''
|
||||||
artl = url.rpartition('/')[2]
|
h1 {font-size: 1.6em; margin-top: 0em}
|
||||||
return 'http://www.dw-world.de/popups/popup_printcontent/' + artl
|
.artikel {font-size: 1em; text-transform: uppercase; margin: 0em}
|
||||||
|
'''
|
||||||
|
|
||||||
def preprocess_html(self, soup):
|
def preprocess_html(self, soup):
|
||||||
for item in soup.findAll('a'):
|
# convert local hyperlinks
|
||||||
limg = item.find('img')
|
for a in soup.findAll('a', href=True):
|
||||||
if item.string is not None:
|
if a['href'].startswith('/'):
|
||||||
str = item.string
|
a['href'] = 'http://www.dw.com' + a['href']
|
||||||
item.replaceWith(str)
|
elif a['href'].startswith('#'):
|
||||||
else:
|
del a['href']
|
||||||
if limg:
|
# remove all style attributes with an effect on font size
|
||||||
item.name = 'div'
|
for item in soup.findAll(attrs={'style':re.compile('font-size')}):
|
||||||
del item['href']
|
del item['style']
|
||||||
if item.has_key('target'):
|
|
||||||
del item['target']
|
|
||||||
else:
|
|
||||||
str = self.tag_to_string(item)
|
|
||||||
item.replaceWith(str)
|
|
||||||
return soup
|
return soup
|
||||||
|
|
||||||
|
|||||||
@ -1,66 +1,74 @@
|
|||||||
|
#!/usr/bin/env python2
|
||||||
|
# vim:fileencoding=utf-8
|
||||||
|
from __future__ import unicode_literals, division, absolute_import, print_function
|
||||||
|
|
||||||
__license__ = 'GPL v3'
|
__license__ = 'GPL v3'
|
||||||
__copyright__ = '2010, Darko Miletic <darko.miletic at gmail.com>'
|
__copyright__ = '2010, Darko Miletic <darko.miletic at gmail.com>'
|
||||||
|
|
||||||
'''
|
'''
|
||||||
dw-world.de
|
Deutsche Welle (español) - dw.com/es
|
||||||
'''
|
'''
|
||||||
|
|
||||||
|
import re
|
||||||
from calibre.web.feeds.news import BasicNewsRecipe
|
from calibre.web.feeds.news import BasicNewsRecipe
|
||||||
|
|
||||||
class DeutscheWelle_es(BasicNewsRecipe):
|
class DeutscheWelle_es(BasicNewsRecipe):
|
||||||
title = 'Deutsche Welle'
|
title = 'Deutsche Welle'
|
||||||
__author__ = 'Darko Miletic'
|
__author__ = 'Darko Miletic'
|
||||||
description = 'Noticias desde Alemania y mundo'
|
description = 'Noticias desde Alemania y mundo'
|
||||||
publisher = 'Deutsche Welle'
|
publisher = 'Deutsche Welle'
|
||||||
category = 'news, politics, Germany'
|
language = 'es'
|
||||||
oldest_article = 1
|
|
||||||
max_articles_per_feed = 100
|
|
||||||
use_embedded_content = False
|
|
||||||
no_stylesheets = True
|
|
||||||
language = 'de'
|
|
||||||
publication_type = 'newsportal'
|
|
||||||
remove_empty_feeds = True
|
|
||||||
masthead_url = 'http://www.dw-world.de/skins/std/channel1/pics/dw_logo1024.gif'
|
|
||||||
extra_css = """
|
|
||||||
body{font-family: Arial,sans-serif}
|
|
||||||
img{margin-top: 0.5em; margin-bottom: 0.2em; display: block}
|
|
||||||
.caption{font-size: x-small; display: block; margin-bottom: 0.4em}
|
|
||||||
"""
|
|
||||||
|
|
||||||
|
oldest_article = 2
|
||||||
|
max_articles_per_feed = 50
|
||||||
|
no_stylesheets = True
|
||||||
|
remove_javascript = True
|
||||||
|
remove_empty_feeds = True
|
||||||
|
ignore_duplicate_articles = {'title', 'url'}
|
||||||
|
|
||||||
conversion_options = {
|
feeds = [
|
||||||
'comment' : description
|
('Titulares', 'http://rss.dw-world.de/rdf/rss-sp-top'),
|
||||||
, 'tags' : category
|
('Noticias de Alemania', 'http://rss.dw-world.de/rdf/rss-sp-ale'),
|
||||||
, 'publisher': publisher
|
('Internacionales', 'http://rss.dw-world.de/rdf/rss-sp-inter'),
|
||||||
, 'language' : language
|
('Cultura', 'http://rss.dw-world.de/rdf/rss-sp-cul'),
|
||||||
}
|
('Ciencia y Tecnología', 'http://rss.dw-world.de/rdf/rss-sp-cyt'),
|
||||||
|
('Economía', 'http://rss.dw-world.de/rdf/rss-sp-eco'),
|
||||||
|
('La prensa opina', 'http://rss.dw-world.de/rdf/rss-sp-press'),
|
||||||
|
('Ecología', 'http://rss.dw-world.de/rdf/rss-sp-ecol'),
|
||||||
|
('Futbol alemán', 'http://rss.dw-world.de/rdf/rss-sp-fut'),
|
||||||
|
('Conozca Alemania', 'http://rss.dw-world.de/rdf/rss-sp-con')
|
||||||
|
]
|
||||||
|
|
||||||
|
keep_only_tags=[
|
||||||
|
dict(name='div', attrs={'class':'col3'})
|
||||||
|
]
|
||||||
|
|
||||||
|
remove_tags_after = [
|
||||||
|
dict(name='div', attrs={'class':'group'})
|
||||||
|
]
|
||||||
|
|
||||||
remove_tags = [
|
remove_tags = [
|
||||||
dict(name=['iframe','embed','object','form','base','meta','link'])
|
dict(name='div', attrs={'class':'col1'}),
|
||||||
,dict(attrs={'class':'actionFooter'})
|
dict(name='div', attrs={'class':re.compile('gallery')}),
|
||||||
]
|
dict(name='div', attrs={'class':re.compile('audio')}),
|
||||||
keep_only_tags=[dict(attrs={'class':'ArticleDetail detail'})]
|
dict(name='div', attrs={'class':re.compile('video')})
|
||||||
remove_attributes = ['height','width','onclick','border','lang']
|
]
|
||||||
|
|
||||||
feeds = [(u'Noticias', u'http://rss.dw-world.de/rdf/rss-sp-all')]
|
remove_attributes = ['height', 'width', 'onclick', 'border', 'lang', 'link']
|
||||||
|
|
||||||
def print_version(self, url):
|
extra_css = '''
|
||||||
artl = url.rpartition('/')[2]
|
h1 {font-size: 1.6em; margin-top: 0em}
|
||||||
return 'http://www.dw-world.de/popups/popup_printcontent/' + artl
|
.artikel {font-size: 1em; text-transform: uppercase; margin: 0em}
|
||||||
|
'''
|
||||||
|
|
||||||
def preprocess_html(self, soup):
|
def preprocess_html(self, soup):
|
||||||
for item in soup.findAll('a'):
|
# convert local hyperlinks
|
||||||
limg = item.find('img')
|
for a in soup.findAll('a', href=True):
|
||||||
if item.string is not None:
|
if a['href'].startswith('/'):
|
||||||
str = item.string
|
a['href'] = 'http://www.dw.com' + a['href']
|
||||||
item.replaceWith(str)
|
elif a['href'].startswith('#'):
|
||||||
else:
|
del a['href']
|
||||||
if limg:
|
# remove all style attributes with an effect on font size
|
||||||
item.name = 'div'
|
for item in soup.findAll(attrs={'style':re.compile('font-size')}):
|
||||||
del item['href']
|
del item['style']
|
||||||
if item.has_key('target'):
|
|
||||||
del item['target']
|
|
||||||
else:
|
|
||||||
str = self.tag_to_string(item)
|
|
||||||
item.replaceWith(str)
|
|
||||||
return soup
|
return soup
|
||||||
|
|
||||||
|
|||||||
Loading…
x
Reference in New Issue
Block a user