mirror of
git://github.com/kovidgoyal/calibre.git
synced 2026-04-22 11:43:28 +02:00
Some minor maintenance on the nytimes
This commit is contained in:
parent
2e20f1aa83
commit
c8ddf2858f
2 changed files with 8 additions and 0 deletions
|
|
@ -219,10 +219,13 @@ def decode_url_date(self,url):
|
|||
re.compile('^leaderboard'),
|
||||
re.compile('^module'),
|
||||
re.compile('commentCount'),
|
||||
'lede-container',
|
||||
'credit'
|
||||
]}),
|
||||
dict(attrs={'class':lambda x: x and 'related-coverage-marginalia' in x.split()}),
|
||||
dict(attrs={'class':lambda x: x and 'interactive' in x.split()}),
|
||||
dict(attrs={'class':lambda x: x and 'sharetools' in x.split()}),
|
||||
dict(attrs={'class':lambda x: x and 'ad' in x.split()}),
|
||||
dict(name='div', attrs={'class':re.compile('toolsList')}), # bits
|
||||
dict(name='div', attrs={'class':re.compile('postNavigation')}), # bits
|
||||
dict(name='div', attrs={'class':'tweet'}),
|
||||
|
|
@ -235,6 +238,7 @@ def decode_url_date(self,url):
|
|||
dict(name='div', attrs={'class':re.compile('entry-tags')}), # pogue
|
||||
dict(id=[
|
||||
'adxLeaderboard',
|
||||
'pagelinks',
|
||||
'adxSponLink',
|
||||
'anchoredAd_module',
|
||||
'anchoredAd_spot',
|
||||
|
|
|
|||
|
|
@ -219,10 +219,13 @@ def decode_url_date(self,url):
|
|||
re.compile('^leaderboard'),
|
||||
re.compile('^module'),
|
||||
re.compile('commentCount'),
|
||||
'lede-container',
|
||||
'credit'
|
||||
]}),
|
||||
dict(attrs={'class':lambda x: x and 'related-coverage-marginalia' in x.split()}),
|
||||
dict(attrs={'class':lambda x: x and 'interactive' in x.split()}),
|
||||
dict(attrs={'class':lambda x: x and 'sharetools' in x.split()}),
|
||||
dict(attrs={'class':lambda x: x and 'ad' in x.split()}),
|
||||
dict(name='div', attrs={'class':re.compile('toolsList')}), # bits
|
||||
dict(name='div', attrs={'class':re.compile('postNavigation')}), # bits
|
||||
dict(name='div', attrs={'class':'tweet'}),
|
||||
|
|
@ -235,6 +238,7 @@ def decode_url_date(self,url):
|
|||
dict(name='div', attrs={'class':re.compile('entry-tags')}), # pogue
|
||||
dict(id=[
|
||||
'adxLeaderboard',
|
||||
'pagelinks',
|
||||
'adxSponLink',
|
||||
'anchoredAd_module',
|
||||
'anchoredAd_spot',
|
||||
|
|
|
|||
Loading…
Reference in a new issue