jackwener
|
fbec2f6f5d
|
feat(snapshot): filter contentinfo subtrees, bilibili ad URLs, boilerplate buttons
- Add contentinfo to subtree-level noise filtering (biggest single win)
- Reuters: 51% → 62%, Google: 57% → 70%, Netflix: 48% → 60%
- Add cm.bilibili.com/cm/api/fees/ ad URL pattern
- Add 广告 keyword to ad detection
- Add back-to-top / 回到顶部 boilerplate button filtering
- Unify ad/boilerplate/contentinfo into single subtree-skip mechanism
- Add vitest config and comprehensive test suite (33 tests)
- Fixture tests skip gracefully when snapshot files are absent
Bump to v0.7.1
|
2026-03-17 01:26:49 +08:00 |
|
jackwener
|
0fd9adf0f5
|
feat: add 8 new sites — bbc, weibo, boss, yahoo-finance, reuters, smzdm, ctrip, youtube
New adapters ported from bb-sites:
- bbc/news: BBC News RSS headlines (public API)
- weibo/hot: 微博热搜 (browser)
- boss/search: BOSS直聘搜索职位 (browser)
- yahoo-finance/quote: 股票行情 (browser)
- reuters/search: 路透社新闻搜索 (browser)
- smzdm/search: 什么值得买搜索好价 (browser)
- ctrip/search: 携程旅行搜索 (browser)
- youtube/search: YouTube视频搜索 (browser)
Also: build script adds clean-yaml step, docs updated to 28 commands / 16 sites
|
2026-03-15 08:46:13 +08:00 |
|