Fix CND recipe

This commit is contained in:
Kovid Goyal 2011-12-14 09:08:12 +05:30
parent 4e13502caa
commit 225dc41252

View File

@ -23,7 +23,9 @@ class TheCND(BasicNewsRecipe):
remove_tags = [dict(name='table', attrs={'align':'right'}), dict(name='img', attrs={'src':'http://my.cnd.org/images/logo.gif'}), dict(name='hr', attrs={}), dict(name='small', attrs={})] remove_tags = [dict(name='table', attrs={'align':'right'}), dict(name='img', attrs={'src':'http://my.cnd.org/images/logo.gif'}), dict(name='hr', attrs={}), dict(name='small', attrs={})]
no_stylesheets = True no_stylesheets = True
preprocess_regexps = [(re.compile(r'<!--.*?-->', re.DOTALL), lambda m: '')] preprocess_regexps = [ (re.compile(r'<!--.*?-->', re.DOTALL), lambda m: ''),
(re.compile('<table width.*?</table>', re.DOTALL), lambda m: ''),
]
def print_version(self, url): def print_version(self, url):
if url.find('news/article.php') >= 0: if url.find('news/article.php') >= 0:
@ -46,16 +48,18 @@ class TheCND(BasicNewsRecipe):
title = self.tag_to_string(a) title = self.tag_to_string(a)
self.log('\tFound article: ', title, 'at', url) self.log('\tFound article: ', title, 'at', url)
date = a.nextSibling date = a.nextSibling
if re.search('cm', date):
continue
if (date is not None) and len(date)>2: if (date is not None) and len(date)>2:
if not articles.has_key(date): if not articles.has_key(date):
articles[date] = [] articles[date] = []
articles[date].append({'title':title, 'url':url, 'description': '', 'date':''}) articles[date].append({'title':title, 'url':url, 'description': '', 'date':''})
self.log('\t\tAppend to : ', date) self.log('\t\tAppend to : ', date)
self.log('log articles', articles) #self.log('log articles', articles)
mostCurrent = sorted(articles).pop() mostCurrent = sorted(articles).pop()
self.title = 'CND ' + mostCurrent self.title = 'CND ' + mostCurrent
feeds.append((self.title, articles[mostCurrent])) feeds.append((self.title, articles[mostCurrent]))
return feeds return feeds