fix: use trafilatura.extract() for text + bare_extraction(with_metadata=True) for date/image

fix: use BeautifulSoup for OG image extraction (handles name=, twitter:image, article:image, link[rel=image_src])

fix: walk ancestor chain, handle lazy-load attrs & relative URLs for images

fix: restore image extraction in HTML scraper

fix: extract dates from article pages via trafilatura instead of homepage HTML

feat: extract dates from HTML-scraped articles

a try

a try

a try

a try

Reformat SPONSOR_INFO, add owner_wikis, update templates/webapp

Replace World Health feeds: STAT, ScienceDaily, Science News, Medscape, NIH, Harvard Gazette

Fix sourcing penalty: detect Article A citing Article B from another outlet, not generic sourcing

Add sourcing penalty: negative points for articles that source from other reporting

Add sponsor modal per article with Wikipedia link, remove standalone sponsors page

Fix political leaning: expanded keywords, lower thresholds

Add MBFC bias/factuality badges

Rich sponsor ownership: shareholders, types, ownership chains

Add sponsor recognition for all 42 news sources

Deploy NewsApp