Updated Wall Street Journal recipe

This commit is contained in:
Kovid Goyal 2009-07-07 23:58:18 -06:00
parent e0e6ddb6cf
commit 5faeaba6c2

View File

@ -53,6 +53,10 @@ class WallStreetJournal(BasicNewsRecipe):
def postprocess_html(self, soup, first): def postprocess_html(self, soup, first):
for tag in soup.findAll(name=['table', 'tr', 'td']): for tag in soup.findAll(name=['table', 'tr', 'td']):
tag.name = 'div' tag.name = 'div'
for tag in soup.findAll('div', dict(id=["articleImage_1", "articleImage_2", "articleImage_3", "articleImage_4", "articleImage_5", "articleImage_6", "articleImage_7"])):
tag.extract()
return soup return soup
def get_article_url(self, article): def get_article_url(self, article):