DIYgod/RSSHub Open Pull Requests

chore: integrate knip

hyoban — Sat, 17 Oct 2026 01:46:22 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/cline/blog
/infzm/1

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
- [ ] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] Date and time / 日期和时间
- [ ] Parsed / 可以解析
- [ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Integrates knip to find unused files, exports, dependencies and circular imports. It runs as pnpm knip and as a step in the lint workflow (with the github-actions reporter for inline annotations).

Config (`knip.ts`)

Route modules are loaded by directory import, which knip cannot trace. Files exporting route, namespace or apiRoute are listed as entries, and includeEntryExports is on so other exports in route files are still checked. scripts/knip-preprocessor.ts drops only the registry exports from the report.
Script entries come from knip's package.json / GitHub Actions plugins; only files they can't see are listed (the actions/github-script imports and the Dockerfile minifier).
ignoreIssues is limited to modules whose exports are consumed outside the graph: worker shims and *.worker.* replacements, the npm package entry, Cloudflare entries, and the disabled Twitter login flows (kept on purpose, see the commented-out imports in twitter/api).
Circular imports are reported as errors, and configuration hints fail the run.
Production patterns carry the ! suffix, so knip --production can be run on demand (not in CI).

Cleanup from the findings

Fix: the cline namespace used a default export, so its metadata was never registered.
Removed unreachable files: art-template leftovers from #20777, old-style sjtu/seiee and scnu routes that have not been registered since router.ts was removed, puppeteer-utils.ts (missed in #21976), and never-imported type files.
Removed the needless export from 251 route handlers (the new:route template doesn't export it) and from unused types.
Broke circular imports in infzm, sdo/ff14risingstones and youtube by moving shared code to the module both sides use. This only moves code; behaviour is unchanged.
Dropped unused dependencies: rfc4648, yaml-eslint-parser, eslint-nibble, @types/eslint, @types/module-alias. Added @types/mdast for a JSDoc type import that relied on a transitive dependency.

Verified locally: pnpm knip, pnpm typecheck, oxlint, oxfmt, pnpm build:routes, and the registry / pkg tests.

🤖 Generated with Claude Code

feat(route/cloudflarestatus): add full-text blog feed

ljh12138164 — Fri, 16 Oct 2026 01:33:50 GMT

Involved Issue / 该 PR 相关 Issue

Follow-up to #23543, which was closed after the Cloudflare Status feed was fixed upstream in 9e677874d94237eaf9bc9fbcc5d8c707d033ebdd. This PR focuses on the Blog feed.

Example for the Proposed Route(s) / 路由地址示例

/cloudflarestatus/blog

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有，代码是否有对应措施？
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Adds a full-text Cloudflare Blog feed at /cloudflarestatus/blog.

The official RSS feed provides the article list, authors, publication dates, categories, GUIDs, and cover-image enclosures. The route fetches article pages for the full content, including inline images, code blocks, and copyable prompts that may be missing from the RSS feed.

Article content is cached, relative URLs are resolved, and copyable prompts are converted into code blocks. Article requests run with a concurrency limit of three.

This PR adds one route file under the existing cloudflarestatus namespace. It does not modify the Status route or require additional packages, authentication, or browser automation.

feat(panasonic): add camera and lens firmware feed

fdkevin0 — Thu, 15 Oct 2026 17:23:26 GMT

Involved Issue / 该 PR 相关 Issue

None

Example for the Proposed Route(s) / 路由地址示例

/panasonic/firmware

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [ ] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] Date and time / 日期和时间
- [ ] Parsed / 可以解析
- [ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Adds /panasonic/firmware for the latest firmware listed on Panasonic's global support pages, covering full-frame, Micro Four Thirds, and compact cameras, plus interchangeable lenses including LEICA DG.

Items include the model, firmware version, release date, category, and download-page link. Radar rules are included. Model filtering uses RSSHub's common parameters.

Only the latest listed version per model is included. Historical releases and camera software updates are excluded; intermediate releases may be missed between feed refreshes.

Validation: focused type-aware lint and formatting checks passed. Live route behavior, source date/time-zone semantics, and anti-bot behavior have not yet been verified. No dependencies or browser automation added.

feat(route): add nixpkgs-update package result feeds

DaVinci42 — Sun, 11 Oct 2026 08:54:19 GMT

Involved Issue / 该 PR 相关 Issue

None.

Example for the Proposed Route(s) / 路由地址示例

/nix-community/update-logs/crush
/nix-community/update-logs/changedetection-io

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有，是否有对应措施？
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Summarizes 20 daily logs with update/build outcomes and source links. PR evidence is log-only, not merge status.
UTC dates; stable GUIDs may not re-notify. Concurrency 3; 600s cache. JSX, bilingual docs, Radar.

Test logs (excerpts)

$ pnpm typecheck
> tsc --noEmit
$ pnpm exec oxlint --type-aware lib/routes/nix-community
$ pnpm exec oxfmt --check lib/routes/nix-community
All matched files use the correct format.
$ FULL_ROUTES_TEST=true pnpm vitest run lib/app.test.ts -t '/nix-community/update-logs/:pkg'
Test Files  1 passed (1)
Tests  1 passed | 3888 skipped (3889)

Local server:

python3 - <<'PY'
import urllib.request as u
import xml.etree.ElementTree as E
for pkg in ('crush', 'changedetection-io'):
    for fmt in ('rss', 'atom'):
        with u.urlopen(f'http://127.0.0.1:1200/nix-community/update-logs/{pkg}?format={fmt}', timeout=60) as r:
            root = E.fromstring(r.read())
            path = './channel/item' if fmt == 'rss' else '{http://www.w3.org/2005/Atom}entry'
            count = len(root.findall(path))
            assert r.status == 200 and count == 20
            print(pkg, fmt, r.status, count)
PY

crush rss 200 20
crush atom 200 20
changedetection-io rss 200 20
changedetection-io atom 200 20

58 temporary checks passed; no route tests added. Full suite not run; two extra cache tests failed without Redis. AI-assisted; submitter tested manually.

💘 Generated with Crush

fix(telegram): group album messages into single feed items in mtproto

petrouv — Fri, 09 Oct 2026 14:57:45 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/telegram/channel/awesomeRSSHub

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
- [ ] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] Date and time / 日期和时间
- [ ] Parsed / 可以解析
- [ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

In Telegram MTProto, albums (media groups) are delivered as multiple contiguous Message objects sharing the same groupedId (64-bit integer), where the text caption and forward/story info are typically attached to only one of the messages.

Previously, tglib/channel.ts attempted to buffer media using an ad-hoc text === '' heuristic:

Broken on forwarded albums: when a message is forwarded, message.fwdFrom prepends Forwarded From: ...: to every message, making text non-empty for all messages. Consequently, each photo in the album was emitted as a separate feed item with a fallback UTC date title (e.g. Fri, 09 Oct 2026 07:09:24 GMT).
Broken caption placement: if the caption was attached to the first message in the album, the first message was emitted immediately with only 1 photo, and remaining photos were orphaned or merged into the subsequent unrelated post.
Reversed order: messages from Telegram MTProto arrive in reverse chronological order (descending ID). Without sorting, media in the album was displayed backwards.

This PR addresses the issue according to Telegram's native protocol:

Contiguous messages sharing message.groupedId are grouped together. Standalone messages (groupedId == null) are treated as single items.
Album messages are sorted ascending by id so photos and media appear in their original intended sequence.
All media items across the album are aggregated into a single description.
Text caption, forward attribution (fromName / fwdFrom), and story peer are resolved and rendered once per album.
Canonical link and guid are set to https://t.me/${username}/${primaryMsg.id} (matching the first post ID of the album, consistent with Telegram web preview format).

feat(route): add Oaktree Capital insights

argsno — Fri, 09 Oct 2026 02:59:03 GMT

Involved Issue / 该 PR 相关 Issue

No related issue.

Example for the Proposed Route(s) / 路由地址示例

/oaktree/insights

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Adds /oaktree/insights for Oaktree Capital Insights - Howard Marks memos, market commentary, education and press mentions.

The full insights list is embedded in the page as JSON (no pagination), so the route parses it directly; the latest 30 entries are returned by default and the common limit parameter can be used for more.
Full text is fetched from each internal insight detail page and cached per link with cache.tryGet; press mentions link to external websites.
Items are categorised by content type (memos, press, Market Commentary, Education) and media type (article, audio, video), so both can be filtered with common parameters, e.g. ?filter_category=audio or ?filterout_category=audio|video.

fix(telegram): handle unresolvable forwarded channels and story peers

petrouv — Thu, 08 Oct 2026 13:33:38 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/telegram/channel/awesomeRSSHub

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
- [ ] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] Date and time / 日期和时间
- [ ] Parsed / 可以解析
- [ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

When a Telegram message forwards from a private channel or supergroup that the current session has not joined, client.getEntity(message.fwdFrom.fromId) throws ChannelPrivateError: You haven't joined this channel/supergroup. (caused by channels.GetChannels). Similarly, if a story peer cannot be resolved, client.getEntity(message.media.peer) can throw an error.

Uncaught exceptions in these lookups cause the entire channel route handler to crash with HTTP 503 ("Internal server error"), failing to render the feed even though the main channel messages were successfully retrieved.

This PR wraps both entity resolutions in try ... catch blocks so that unresolvable forward sources and stories are gracefully skipped while the rest of the feed and messages render normally.

fix(route/guancha): fetch full text only for paginated articles and update list APIs

flareskyqq — Wed, 07 Oct 2026 03:07:05 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/guancha
/guancha/all
/guancha/review
/guancha/story
/guancha/fengwen
/guancha/redian
/guancha/gundong
/guancha/home
/guancha/others
/guancha/headline

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
- [ ] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] Date and time / 日期和时间
- [ ] Parsed / 可以解析
- [ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Problem

guancha.cn only serves the _s.shtml variant (the article body on a single page) for multi-page articles. Single-page articles have no such variant, and requesting it 302-redirects to the homepage.

Both routes rewrote every link to _s.shtml unconditionally, so for single-page articles the detail request landed on the homepage:

headline.ts: item.description! += content('.all-txt').html()! appended the literal string null to the description.
index.ts: the description came out null/empty.
The item <link> of those articles also pointed at a URL that redirects to the homepage.

Measured on one /guancha/headline response (20 items):

article	items	`_s.shtml`	body
paginated (`.module-page` present)	5	200	available
single page (`.module-page` absent)	15	302 → `www.guancha.cn`	missing / literal `null`

The correlation is 100%.

Why not just drop the `_s.shtml` rewrite

Because that would be a regression. For paginated articles _s.shtml really is the full text, while the plain URL returns only page 1 — e.g. for one article the plain page yields 1232 chars of text vs 7401 chars for _s.shtml, and the plain page body is a prefix of the _s one. Removing the rewrite would silently truncate every multi-page article to its first page.

What this PR does

Keeps the original link, which also fixes <link> for every article.
Reads the body from the current page first, and only when the article is actually paginated fetches the _s.shtml full text and prefers it. The pagination container .module-page is a reliable signal — measured 1 occurrence for paginated articles and 0 for single-page ones.
Only assigns the body when it was really retrieved, so the literal null is gone.

Second fix: the two list endpoints changed upstream

While verifying the routes above I found five of them were already broken on master, independently of this PR:

/api/redian.htm still responds, but the JSON keys became lowercase (title/httpUrl). Reading TITLE/HTTP_URL produced https://www.guancha.cnundefined for every item link.
/api/gundong.htm now returns 404. The rolling-news list moved to /api/gundong, with the same lowercase keys.

That broke /guancha, /guancha/all, /guancha/redian, /guancha/gundong and /guancha/others — i.e. every route that reads the two lists.

The second commit extracts a single fetchApiList() helper shared by both lists, points the rolling-news request at /api/gundong, and reads item.title ?? item.TITLE / item.httpUrl ?? item.HTTP_URL, so the routes survive the keys being renamed back.

Verified against the live site: both endpoints return 200 and all 18 items across them resolve to article pages that respond 200 with .all-txt present.

feat(route/hotukdeals): add tag route

zhgqthomas — Tue, 06 Oct 2026 03:42:28 GMT

Involved Issue / 该 PR 相关 Issue

N/A

Example for the Proposed Route(s) / 路由地址示例

/hotukdeals/tag/flight
/hotukdeals/tag/flight/new
/hotukdeals/tag/flight/temp
/hotukdeals/tag/flight/highest_price
/hotukdeals/tag/flight/lowest_price
/hotukdeals/tag/flight/discussion

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Adds a generic tag feed for hotukdeals, e.g. https://www.hotukdeals.com/tag/flight.

The tag listing is fetched with a plain HTTP request. Each deal card has its thread data (title, slug, type, publishedAt, author, merchant, image) embedded as JSON in its data-vue3 attribute, so no browser is needed.
sort maps to the site's sortBy query and defaults to new (most recent), which suits a feed better than the site's default "hottest" order.
The listing only contains a truncated snippet, so the full description (threadDetails.preparedHtmlDescription) and categories (groups) are read from each deal page's __INITIAL_STATE__ via evaluateScriptData, with results cached through cache.tryGet.
pubDate is parsed from the Unix timestamp publishedAt, so no time zone conversion is needed.
Unknown tags redirect to /tag on the site; the route detects this and throws a clear error instead of returning an empty feed.
antiCrawler: true is kept for consistency with the existing hotukdeals routes. No blocking was observed during testing.

feat(konek): add Deep Web Konek newest posts

TungHaHuy — Mon, 05 Oct 2026 04:21:46 GMT

Involved Issue / 该 PR 相关 Issue

None.

Example for the Proposed Route(s) / 路由地址示例

/konek/newest

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Adds /konek/newest for Deep Web Konek, an independent cybersecurity news and research publication. Fetches up to 100 published posts from the public HN+ /contents endpoint used by the site's Vue frontend, including full article text, authors, UTC publication dates, categories and tags. Includes a Radar rule and stable GUIDs based on post IDs. No authentication or new dependencies are required.

Validation:

Formatting check and type-aware oxlint passed for both route files.
Executed the final handler against the live API: 100 unique posts with full descriptions and valid publication dates.
Local RSSHub endpoint returned HTTP 200 and valid XML with 100 items. The common parameter limit=3 returned 3 items.

feat(route): add Game Wisdom full-text articles

mcdp-adk — Sun, 04 Oct 2026 10:45:52 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/game-wisdom
/game-wisdom/analysis
/game-wisdom/critical
/game-wisdom/general
/game-wisdom/guest
/game-wisdom/podcast
/game-wisdom/series
/game-wisdom/spotlight
/game-wisdom/videos

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[x] Anti-bot or rate limit / 反爬/频率限制
- [x] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Game Wisdom's official feeds only include excerpts. This route reads the first page of the official site or category feed and adds the full article body from each article page. The site has no public posts API (only oEmbed is exposed under /wp-json/).

The site's robots.txt sets Crawl-delay: 30. The route fetches article pages with a concurrency of 2 and caches each processed article. If an article page cannot be fetched or parsed, that item falls back to the official excerpt without caching it.

feat(route): add 台湾小说网 (xsw)

xbpk3t — Tue, 29 Sep 2026 09:40:24 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/xsw/17301350

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[x] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 无
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确 (站点为东八区, 使用 timezone(..., 8))
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

站点 HTTPS 证书链不完整, 使用 http 直连。
页面 meta 谎报 big5, 实际为 UTF-8, 使用 ofetch 默认解码。
全书无逐章时间戳, pubDate 统一取书页「更新:」时间 (东八区)。
书页内嵌最新 10 条目录, 完整目录需翻页; 本路由提供嵌入式预览条目。

feat(route/cumst): add posts and DMs feeds

gekangen — Sat, 26 Sep 2026 14:31:11 GMT

Involved Issue / 该 PR 相关 Issue

N/A

Example for the Proposed Route(s) / 路由地址示例

/onlyheaven/posts
/onlyheaven/dms
/onlyheaven/onlyfans/16786030
/onlyheaven/onlyfans/16786030/dms
/onlyheaven/fansly/365078231763660800
/onlyheaven/fansly/365078231763660800/dms
/onlyheaven/patreon/4407214

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Adds OnlyHeaven (cum.st) feeds for recent posts and DMs across creators and for individual OnlyFans, Fansly, and Patreon creators. Entries link to their source pages and include available images from unlocked attachments.

Route generation, lint, formatting, typecheck, and 23 offline route and registry tests passed.

fix(douban): use configured cookie for group route

tianalei — Sun, 13 Sep 2026 03:34:37 GMT

Involved Issue / 该 PR 相关 Issue

N/A

Example for the Proposed Route(s) / 路由地址示例

NOROUTE

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
- [ ] Follows Script Standard / 跟随路由规范
[x] Anti-bot or rate limit / 反爬/频率限制
- [x] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] Date and time / 日期和时间
- [ ] Parsed / 可以解析
- [ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

This fixes the existing Douban group route; it does not add a new route, package, or Puppeteer dependency.

The automated route runner cannot exercise this configuration-dependent fix because it does not have a DOUBAN_COOKIE; Douban returns HTTP 403 to the runner without one. The route was therefore marked NOROUTE for the online route check and verified separately with DOUBAN_COOKIE configured.

Summary

Send the configured DOUBAN_COOKIE when fetching a Douban group listing.
Send the same cookie when fetching individual group topic pages.

Without the cookie, Douban can redirect these requests to its security verification page even when the instance operator has configured DOUBAN_COOKIE.

Verification

Built the Chromium-bundled Docker image successfully on Linux/ARM64.
Verified /douban/group/705363 returns HTTP 200 and a valid RSS feed with DOUBAN_COOKIE configured.
Verified the existing Bilibili user-video route still returns HTTP 200.
git diff --check passes.

feat(route): add iandaily daily digest

DeyunMa — Thu, 10 Sep 2026 07:30:22 GMT

Involved Issue / 该 PR 相关 Issue

N/A

Example for the Proposed Route(s) / 路由地址示例

/iandaily

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Dynamically discovers the site's hashed application bundle instead of hardcoding an asset URL.
Extracts the static JSON.parse payload with Acorn without executing remote JavaScript.
Emits one item per daily edition, including its overview, highlights, columns, and original links.
The source exposes calendar dates only, so the route parses them without inventing a publication time.
Verified with formatting, type checking, route/Worker builds, a mocked five-edition route test, the complete workerd suite, and a live request to /iandaily.

feat(route/hackernews): filter stories by minimum comment count

PeterDaveHello — Sun, 06 Sep 2026 16:51:36 GMT

Involved Issue / 该 PR 相关 Issue

Replaces #23216.

Example for the Proposed Route(s) / 路由地址示例

/hackernews
/hackernews/index/sources/10
/hackernews/best/sources/10
/hackernews/newest
/hackernews/index/comments/10?limit=1
/hackernews/over/sources/100/10
/hackernews/submitted/sources/dang/1
/hackernews/item/sources/8863
/hackernews/threads/comments_list/dang
/hackernews/jobs

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
- [ ] Follows [Script Standard](https://docs.rsshub.app/joinus/advanced/script-standard) / 跟随 [路由规范](https://docs.rsshub.app/zh/joinus/advanced/script-standard)
[x] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] [Date and time](https://docs.rsshub.app/joinus/advanced/pub-date) / [日期和时间](https://docs.rsshub.app/zh/joinus/advanced/pub-date)
- [ ] Parsed / 可以解析
- [ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Add minimum-comment filtering before the common limit, preserving existing ID parameters, GUIDs and empty-feed checks. Normalize comment counts independently of GUID construction and separate caches by output format. Move tests outside route sources to fix the Docker build failure in #23216, with type-checking and CI coverage.

Before rebasing, 427 tests (including 24 Hacker News tests), 14 workerd tests, type checking, targeted lint/format checks, Docker build and startup checks passed. Local code review completed without unresolved actionable findings. Git range-diff confirms the patch is unchanged after rebasing; the new HEAD has not been revalidated.

HN may return HTTP 429; this change adds no new anti-bot or rate-limit handling. High thresholds can yield empty feeds. These 10 examples sample distinct parameter and output behaviors to reduce live HN requests; full-comment output uses limit=1.

feat(route): add CPP（无差别同人站） route

marcosteam — Sun, 06 Sep 2026 01:55:00 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/cpp/event
/cpp/event/310100

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

feat(route/gq/tw): add GQ Taiwan category and subcategory feeds

johan456789 — Wed, 02 Sep 2026 15:39:17 GMT

Involved Issue / 该 PR 相关 Issue

This adds a feed that supports category filtering, which the official feed doesn't support.

Continues from #20183. The last requested changes should be fixed.

Example for the Proposed Route(s) / 路由地址示例

/gq/tw/entertainment
/gq/tw/life/outdoor

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

fix(router/twitter): lock tokens temporarily on 401/403 instead of permanent removal

cloneselfish-wq — Mon, 31 Aug 2026 14:57:23 GMT

Involved Issue / 该 PR 相关 Issue

Close #<ISSUE_NUMBER>

Example for the Proposed Route(s) / 路由地址示例

NOROUTE

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
[ ] Follows [Script Standard](https://docs.rsshub.app/joinus/advanced/script-standard) / 跟随 [路由规范](https://docs.rsshub.app/zh/joinus/advanced/script-standard)
[ ] Anti-bot or rate limit / 反爬/频率限制
[ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] [Date and time](https://docs.rsshub.app/joinus/advanced/pub-date) / [日期和时间](https://docs.rsshub.app/zh/joinus/advanced/pub-date)
[ ] Parsed / 可以解析
[ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Behavior hardening for the 401/403 path in twitterGot() (lib/routes/twitter/api/web-api/utils.ts), building on #<PR1_NUMBER>.

Problem

On 401/403 the current code does

config.twitter.authToken?.splice(tokenIndex, 1);

which removes the token from the in-memory array irreversibly (until process restart) — even for a transient error. With a single-token deployment, one transient 401/403 permanently disables authenticated access. This is also the first half of the silent guest-degradation chain fixed in #<PR1_NUMBER>.

Changes

splice() → existing time-based lock (3600s) on 401/403, consistent with the 429/rate-limit path. Transient auth errors lock the token temporarily instead of deleting it permanently.
Defensive check in twitterGot: auth && !auth.token now throws ConfigNotFoundError, guaranteeing a guest cookie jar can never be cached under the literal key twitter:cookie:undefined, even if getAuth changes in the future.
Removed the dead commented-out re-login fallback code in this block (the username/password fields it references are already commented out in getAuth).

Design notes (per discussion with @dosu in #<ISSUE_NUMBER>)

Failure still surfaces: once every token is locked, getAuth(30) exhausts its retries (~30s worst case) and returns undefined, so twitterGot throws ConfigNotFoundError instead of looping forever — the failure is no longer silent.
Recovery: a locked token is retried after the 3600s lock expires. If maintainers prefer a different duration now that the lock is the sole recovery mechanism (vs a backstop before permanent removal), happy to adjust — e.g. escalate the TTL on repeated 401/403 for the same token.
Multi-token pools keep rotating through unlocked tokens while one is locked.

Issue: #<ISSUE_NUMBER>
Depends on #<PR1_NUMBER>


---

## 给 @dosu 的 issue 回复（确认拆分方案）

```markdown
Thanks @dosu — agreed on splitting, that's cleaner for review and bisecting. Plan:

**PR 1** (minimal, obviously correct): the `!config.twitter.authToken?.length` guard
alone.

**PR 2** (behavior change): `splice()` → temporary lock on 401/403 + the
`auth && !auth.token` defensive check.

On your design question: keeping the 3600s lock for now, and I believe the
"eventually surface the failure" requirement is already satisfied without extra code —
once all tokens are locked, `getAuth(30)` exhausts its retries (each retry sleeps
500–1000ms and moves to the next index) and returns `undefined`, so `twitterGot`
throws `ConfigNotFoundError` after ~30s worst case rather than looping forever. The
failure stops being silent either way.

The remaining judgment call is the lock TTL: with `splice()` gone, 3600s becomes the
sole recovery mechanism, so a permanently revoked token would retry hourly forever
(though each retry only costs one failed request, and the failure is now logged and
surfaced). If maintainers prefer, I can escalate the TTL on repeated 401/403 for the
same token instead of a fixed 3600s — will leave that open for review.

Opening the PRs now.

feat(route/nytimes): add Wirecutter

IvanWng97 — Sun, 30 Aug 2026 05:57:49 GMT

Involved Issue / 该 PR 相关 Issue

N/A

Example for the Proposed Route(s) / 路由地址示例

/nytimes/wirecutter

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[x] Anti-bot or rate limit / 反爬/频率限制
- [x] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Why. Wirecutter has an official feed at nytimes.com/wirecutter/feed/, but it only carries the opening few paragraphs and stops before any of the actual picks.

How. The feed supplies the item list. Each post is then rendered from props.pageProps.post in the page's __NEXT_DATA__:

the hero image, then the opening paragraphs (structuredLede), then each chapter under its title as a heading
node trees come in two formats, a DOM-shaped one (type: 'tag' with a name) and a document one (the element is the type, with marks for formatting), and both go through a tag/attribute allowlist that drops layout classes and data-gtm-* tracking attributes
shortcode-callout / product_callout (the picks) render as the ribbon, product name, photo, verdict and description; shortcode-gallery, shortcode-caption and scoop_image_block become figures; shortcode-pullquote becomes a blockquote
adslot, shortcode-recirc and shortcode-scoop_form_callout are dropped as ads, in-house promos and a form, and script / style are dropped
an unknown node keeps its content without its element, so a new block type never fails the item
images drop the resizing query string so the original is served
category holds the section and subsection names, e.g. Electronics and Laptops

Anti-bot. The site sits behind DataDome. Articles are fetched two at a time with p-map and 403s are not retried. When an article can't be fetched or parsed, the item falls back to the opening paragraphs from the feed's content:encoded. That fallback happens outside cache.tryGet, so it isn't cached and the article is tried again on the next run.

🤖 Generated with Claude Code

https://claude.ai/code/session_017gmEyEJ8ufxFDtbwMCXNiL

fix(route/xueqiu): avoid caching transient detail failures

sirius60111 — Sat, 29 Aug 2026 11:36:04 GMT

Involved Issue / 该 PR 相关 Issue

NOROUTE

Example for the Proposed Route(s) / 路由地址示例

/xueqiu/user/1505944393

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
[x] Anti-bot or rate limit / 反爬/频率限制
- [x] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Follow-up fix for the existing /xueqiu/user/:id/:type? route after #22708.

The route fetches each status detail through statuses/show.json inside cache.tryGet. The current implementation treats every response containing an error_code as a permanent failure and returns the timeline fallback from inside the cache callback. Authentication or other transient business errors are therefore cached as incomplete items. HTTP 200 responses with an empty detail body are also cached. A single transient upstream failure can consequently leave only some items in the same feed without their details until the content cache expires.

修复已有 /xueqiu/user 路由在 #22708 后仍存在的详情缓存污染问题。当前实现将任意带 error_code 的响应都视为永久失败，并在 cache.tryGet 回调内返回列表摘要，因此鉴权等瞬态业务错误也会被缓存；HTTP 200 但详情正文为空时同样会缓存空详情。这会让同一 feed 中部分 item 长期缺少详情。

Changes / 改动

Cache the timeline fallback only for the explicitly known permanent error 20210; authentication, WAF, rate-limit, network, unknown business errors, and empty HTTP 200 details fall back for the current feed without entering the detail cache. / 仅明确的永久错误 20210 缓存摘要；鉴权、WAF、限流、网络、未知业务错误和 HTTP 200 空详情仅在当次 feed 回退，不写详情缓存。
Validate the assembled detail after including text, images, and repost content, and use detail.description when detail.text is empty. / 组合文字、图片和转发内容后再校验详情，并在 detail.text 为空时使用 detail.description。
Use item.text || item.description for the timeline fallback and keep the item even when both are empty. / 摘要回退优先使用 item.text，其次使用 item.description；二者为空时仍保留 item。
Disable automatic retries for show.json so an HTTP 400 is not amplified into three requests, while retaining the existing per-handler concurrency of 3. / 对 show.json 设置 retry: 0，避免一次 HTTP 400 放大为三次请求，同时保留现有单 handler 并发 3。
Add debug diagnostics with the item id, target, HTTP status, business error code, and failure classification, without logging cookies or content. / 增加 debug 诊断日志，只记录 item id、target、HTTP 状态、业务错误码和失败分类，不记录 Cookie 或正文。

The existing cache key is intentionally unchanged, so this PR prevents new transient failures from being cached but does not migrate previously cached incomplete items.

Local cold-cache validation for /xueqiu/user/1505944393 completed twice: both requests returned HTTP 200 with 20 items and 0 empty descriptions in 6.97 s and 6.36 s respectively. Type checking, linting, formatting, and route generation pass.

fix(route/anthropic): locate research publications anywhere in flight data

naplesblue — Thu, 27 Aug 2026 11:25:11 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Same symptom as the previously fixed #21581 (Error: this route is empty), but a different cause — the page layout changed again.

Example for the Proposed Route(s) / 路由地址示例

/anthropic/research

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

/anthropic/research currently fails with Error: this route is empty.

The route reads the publication list out of the Next.js flight payload at the fixed path item.page.sections. www.anthropic.com/research now nests that section deeper inside the React element tree, so .flatMap((item) => item?.page?.sections ?? []) resolves to an empty array, no Publications section is found, and the route returns no items.

This PR walks the parsed flight payload and picks up the Publications section wherever it sits, instead of relying on one fixed path. Posts are deduplicated by slug in case the section is embedded more than once (server payload + client hydration).

Verified against the live page on pnpm dev:

/anthropic/research → 20 items (the route's default limit), previously 0
/anthropic/research?limit=3 → 3 items, limit still honored (as added in #23075)
titles, links, pubDate and full article bodies all populate correctly

Not switching to DOM parsing like the sibling /anthropic/news route: the research page only server-renders ~10 publications before the "See more" button, which would silently cut the route's default of 20.

feat(route): add zjnu

gushen610140 — Sat, 15 Aug 2026 18:27:03 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/zjnu/news
/zjnu/academic
/zjnu/media
/zjnu/general

New RSS Route Checklist / 新 RSS 路由检查表

[x] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[x] Date and time / 日期和时间
- [x] Parsed / 可以解析
- [x] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Adds a route for zjnu.edu.cn

feat(utils): add parse date range to extract start&end date

magazian — Sat, 01 Aug 2026 05:53:41 GMT

Involved Issue / 该 PR 相关 Issue

Close #

Example for the Proposed Route(s) / 路由地址示例

/shanghaimuseum/display/offline-exhibit
/chnmuseum/zl
/dpm/exhibitions
/chnmus/information/exhibition/special
/hebeimuseum/list/special
/sdmuseum/exhibitions
/sxhm/temporary
/njmuseum/exhibitionIndex/
/hnmuseum/current-exhibitions
/capitalmuseum/exhibition
/gdmuseum/exhibition
/3gmuseum/tempexhibition
/szmuseum/temporary
/tjbwg/exhibition
/chinasilkmuseum/zz
/grandcanalmuseum/linzhantezhan
/ahm/exhibition/xztj
/minhangmuseum/interim
/canalmuseum/lszl
/namoc/zhanlan
/wzbwg/zhanlan/specialexhibition

New RSS Route Checklist / 新 RSS 路由检查表

[ ] New Route / 新的路由
- [x] Follows Script Standard / 跟随路由规范
[ ] Anti-bot or rate limit / 反爬/频率限制
- [ ] If yes, do your code reflect this sign? / 如果有, 是否有对应的措施?
[ ] Date and time / 日期和时间
- [ ] Parsed / 可以解析
- [ ] Correct time zone / 时区正确
[ ] New package added / 添加了新的包
[ ] Puppeteer

Note / 说明

Adds a new utility lib/utils/parse-date-range.ts that extracts a start date and end date from free-form date range strings, returning both in YYYY-MM-DD format, can support both CN&EN date range format. This utility was originally written in the chnmuseum route. It is now promoted to a standalone, well-tested helper so that other routes (e.g. museum, gallery, and exhibition feeds) can reuse it directly instead of reimplementing the same logic.

DIYgod/RSSHub Open Pull Requests

chore: integrate knip

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

Config (knip.ts)

Cleanup from the findings

feat(route/cloudflarestatus): add full-text blog feed

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

feat(panasonic): add camera and lens firmware feed

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

feat(route): add nixpkgs-update package result feeds

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

Test logs (excerpts)

fix(telegram): group album messages into single feed items in mtproto

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

feat(route): add Oaktree Capital insights

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

fix(telegram): handle unresolvable forwarded channels and story peers

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

fix(route/guancha): fetch full text only for paginated articles and update list APIs

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

Problem

Why not just drop the _s.shtml rewrite

What this PR does

Second fix: the two list endpoints changed upstream

feat(route/hotukdeals): add tag route

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

feat(konek): add Deep Web Konek newest posts

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

feat(route): add Game Wisdom full-text articles

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

feat(route): add 台湾小说网 (xsw)

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

feat(route/cumst): add posts and DMs feeds

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

fix(douban): use configured cookie for group route

Involved Issue / 该 PR 相关 Issue

Example for the Proposed Route(s) / 路由地址示例

New RSS Route Checklist / 新 RSS 路由检查表

Note / 说明

Summary

Verification

Config (`knip.ts`)

Why not just drop the `_s.shtml` rewrite