diff options
| -rw-r--r-- | unbiased/sources/fox.py | 18 |
1 files changed, 10 insertions, 8 deletions
diff --git a/unbiased/sources/fox.py b/unbiased/sources/fox.py index 63f853e..55de92b 100644 --- a/unbiased/sources/fox.py +++ b/unbiased/sources/fox.py @@ -20,15 +20,17 @@ class Fox(NewsSource): """ soup = cls._fetch_content(cls.url) - primary = soup.find('div', class_='main-primary')\ - .find('div', class_='collection-spotlight')\ - .find_all('article', class_='article') - h1s = (primary[0].find('header', class_='info-header').find('a')['href'],) - h2s = tuple(x.find('header', class_='info-header').find('a')['href'] for x in primary[1:]) + hero = soup.select('div.collection-spotlight.has-hero .title a') + related = soup.select('div.collection-spotlight.has-hero .related a') + h1s = [x['href'] for x in hero + related if 'live-news' not in x['href']][:1] + + h2s = soup.select('div.collection-spotlight .title a') + h2s = [x['href'] for x in h2s if 'live-news' not in x['href']] + h2s = set(h2s) - set(h1s) - h3s = soup.find('div', class_='main-primary')\ - .find('div', class_='collection-article-list')\ + h3s = soup.find('div', class_='collection-article-list')\ .find_all('article', class_='article') - h3s = tuple(x.find('header', class_='info-header').find('h2', class_='title').find('a')['href'] for x in h3s) + h3s = tuple(x.find('header', class_='info-header')\ + .find('h2', class_='title').find('a')['href'] for x in h3s) return h1s, h2s, h3s |