return "cnn-details-%s" % (self.page_title)
def munge_description(self, description):
- description = re.sub('[Rr]ead full story for latest details.', '', description)
+ description = re.sub('[Rr]ead full story for latest details.',
+ '',
+ description)
description = re.sub('<[^>]+>', '', description)
return description
+ def find_image(self, item):
+ image = item.findtext('media:thumbnail')
+ if image is not None:
+ image_url = image.get('url')
+ return image_url
+ return None
+
def should_use_https(self):
return False
def item_is_interesting_for_headlines(self, title, description, item):
+ if self.is_item_older_than_n_days(item, 14):
+ self.debug_print("%s: is too old!" % title)
+ return False
return re.search(r'[Cc][Nn][Nn][A-Za-z]*\.com', title) is None
def item_is_interesting_for_article(self, title, description, item):
+ if self.is_item_older_than_n_days(item, 7):
+ self.debug_print("%s: is too old!" % title)
+ return False
return (re.search(r'[Cc][Nn][Nn][A-Za-z]*\.com', title) is None and
len(description) >= 65)
# ],
# "Test" )
#if x.fetch_news() == 0:
-# print "Error fetching news, no items fetched."
+# print("Error fetching news, no items fetched.")
#x.shuffle_news()