3 import generic_news_rss_renderer
5 from typing import Dict, List, Optional
9 class cnn_rss_renderer(generic_news_rss_renderer.generic_news_rss_renderer):
12 name_to_timeout_dict: Dict[str, int],
17 super().__init__(name_to_timeout_dict, feed_site, feed_uris, page_title)
19 def get_headlines_page_prefix(self) -> str:
20 return f"cnn-{self.page_title}"
22 def get_details_page_prefix(self) -> str:
23 return f"cnn-details-{self.page_title}"
25 def munge_description(
26 self, description: str, item: xml.etree.ElementTree.Element
28 description = re.sub("[Rr]ead full story for latest details.", "", description)
29 description = re.sub("<[^>]+>", "", description)
32 def find_image(self, item: xml.etree.ElementTree.Element) -> Optional[str]:
33 image = item.findtext("media:thumbnail")
35 image_url = image.get("url")
39 def should_use_https(self) -> bool:
42 def item_is_interesting_for_headlines(
43 self, title: str, description: str, item: xml.etree.ElementTree.Element
45 if self.is_item_older_than_n_days(item, 14, default=True):
47 if re.search(r"[Cc][Nn][Nn][A-Za-z]*\.com", title) is None:
51 def item_is_interesting_for_article(
52 self, title, description, item: xml.etree.ElementTree.Element
54 if self.is_item_older_than_n_days(item, 7):
57 re.search(r"[Cc][Nn][Nn][A-Za-z]*\.com", title) is None
58 and len(description) >= 65
63 # x = cnn_rss_renderer(
70 # "/rss/money_technology.rss",
71 # "/rss/cnn_tech.rss",
72 # "/rss/cnn_topstories.rss",
73 # "/rss/cnn_world.rss",
78 # if x.fetch_news() == 0:
79 # print("Error fetching news, no items fetched.")