Update FAZ.net

This commit is contained in:
Kovid Goyal 2015-02-04 08:24:44 +05:30
parent ff1a8f9eec
commit 2a4e027b0b

View File

@ -20,16 +20,18 @@ class FazNet(BasicNewsRecipe):
encoding = 'utf-8'
remove_javascript = True
keep_only_tags = [{'class':['FAZArtikelEinleitung']},
keep_only_tags = [
{'class':['FAZArtikelEinleitung']},
dict(name='div', attrs={'class':'FAZSlimHeader'}),
{'id':'ArtikelTabContent_0'}
]
remove_tags_after = [dict(name='div', attrs={'class':['ArtikelFooter']})]
remove_tags = [dict(name='div', attrs={'class':['ArtikelFooter','clear']}),
dict(name='a', attrs={'title':['Vergrößern']}), #AGe 2014-10-22
dict(name='img', attrs={'class':['VideoCtrlIcon']}), #AGe 2014-10-22
dict(name='span', attrs={'class':['shareAutor']}) #AGe 2014-10-22
remove_tags = [
dict(name='div', attrs={'class':['ArtikelFooter','clear']}),
dict(name='a', attrs={'title':['Vergrößern']}), # AGe 2014-10-22
dict(name='img', attrs={'class':['VideoCtrlIcon']}), # AGe 2014-10-22
dict(name='span', attrs={'class':['shareAutor']}) # AGe 2014-10-22
]
feeds = [
@ -56,13 +58,17 @@ class FazNet(BasicNewsRecipe):
nexturl = self.INDEX + pager['href']
soup2 = self.index_to_soup(nexturl)
texttag = soup2.find('div', attrs={'class':'FAZArtikelContent'})
for cls in ('ArtikelFooter', 'ArtikelAbbinder', 'ArtikelKommentieren Artikelfuss GETS;tk;boxen.top-lesermeinungen;tp;content', 'Anzeige GoogleAdsBuehne',
'ThemenLinks', 'rechtehinweis', 'stageModule Ressortmodul Rubrikenkopf clearfix', 'VideoCtrlIcon', 'ArtikelAbbinder clearfix',
'stageModule clearfix GETS;tk;artikel.empfehlungen.weitere-artikel;tp;content'): #AGe 2014-10-22
for cls in (
'ArtikelFooter', 'ArtikelAbbinder',
'ArtikelKommentieren Artikelfuss GETS;tk;boxen.top-lesermeinungen;tp;content',
'Anzeige GoogleAdsBuehne', 'ThemenLinks', 'rechtehinweis',
'stageModule Ressortmodul Rubrikenkopf clearfix', 'VideoCtrlIcon',
'ArtikelAbbinder clearfix',
'stageModule clearfix GETS;tk;artikel.empfehlungen.weitere-artikel;tp;content'): # AGe 2014-10-22
div = texttag.find(attrs={'class':cls})
if div is not None:
div.extract()
div = texttag.find(attrs={'title':'Vergrößern'}) #AGe 2014-10-22
div = texttag.find(attrs={'title':'Vergrößern'}) # AGe 2014-10-22
if div is not None:
div.extract()
@ -74,6 +80,8 @@ class FazNet(BasicNewsRecipe):
def preprocess_html(self, soup):
self.append_page(soup, soup.body, 3)
for img in soup.findAll('img', attrs={'data-src':True}):
img['src'] = img['data-src']
return self.adeify_images(soup)
def postprocess_html(self, soup, first_fetch):