From 4fa6bb4c64e90eb5c3c11074cf83747f01bd7fd7 Mon Sep 17 00:00:00 2001 From: Matt Singleton Date: Sat, 14 Oct 2017 19:05:39 -0400 Subject: BBC News closes #3 --- unbiased/sources/bbc.py | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+) create mode 100644 unbiased/sources/bbc.py diff --git a/unbiased/sources/bbc.py b/unbiased/sources/bbc.py new file mode 100644 index 0000000..0dd0f80 --- /dev/null +++ b/unbiased/sources/bbc.py @@ -0,0 +1,26 @@ +from unbiased.sources.base import NewsSource + +class BBC(NewsSource): + + name = 'BBC News' + shortname = 'bbc' + url = 'http://www.bbc.com/news/world/us_and_canada' + + bad_images = ['bbc_news_logo.png'] + + @classmethod + def _fetch_urls(cls): + soup = cls._fetch_content(cls.url) + + h1s = soup.find('div', class_='buzzard-item')\ + .find('a', class_='title-link') + h1s = (h1s['href'],) + + h2s = soup.find_all('div', attrs={'class': 'pigeon__column', 'data-entityid': True}) + h2s = tuple(x.find('a', class_='title-link')['href'] for x in h2s) + + # get tertiary headlines + h3s = soup.find_all('div', attrs={'class': 'macaw-item', 'data-entityid': True}) + h3s = tuple(x.find('a', class_='title-link')['href'] for x in h3s) + + return h1s, h2s, h3s -- cgit v1.2.3