This commit is contained in:
Kovid Goyal 2018-05-22 19:28:55 +05:30
commit 029ad0d0d5
No known key found for this signature in database
GPG Key ID: 06BC317B515ACE7C

View File

@ -79,19 +79,19 @@ class NewYorkTimes(BasicNewsRecipe):
no_stylesheets = True
compress_news_images = True
compress_news_images_auto_size = 5
preprocess_regexps = [(re.compile(r'(?s)<!--.*?-->'), lambda m: '')]
remove_tags = [
dict(attrs={'aria-label':'tools'.split()}),
dict(attrs={'data-videoid':True}),
dict(name='button meta link'.split()),
dict(id=lambda x: x and x.startswith('story-ad-')),
dict(name='head'),
dict(name='a', href=lambda x: x and '#story-continues-' in x),
dict(name='a', href=lambda x: x and '#whats-next' in x),
dict(id=lambda x: x and 'sharetools-' in x),
dict(id='newsletter-promo supported-by-ad bottom-wrapper'.split()),
classes('story-print-citation supported-by accessibility-ad-header visually-hidden bottom-of-article ad'),
dict(attrs={'class': lambda x: x and ('SectionBar' in x or 'recirculation' in x or 'ResponsiveAd' in x)}),
dict(attrs={'class': lambda x: x and ('SectionBar' in x or 'recirculation' in x or 'ResponsiveAd' in x or 'accessibility-visuallyHidden' in x or 'RelatedCoverage' in x)}),
]
def preprocess_html(self, soup):