Worker RSS Bridge

Single-file Cloudflare Worker RSS bridge implemented in worker.js.

Bridge plugins live in bridges/. Add a new site by creating a new plugin file that exports a standard bridge object, then register it in worker.js.

See PLUGIN_DEVELOPMENT.md for the full plugin interface, caching behavior, development checklist, and validation commands.

Routes

• / - HTML usage page
• /health - JSON health check
• /feed?bridge=nikkeiAsiaChina
• /feed/nikkeiAsiaChina?category=chinaTech&limit=15&fulltext=true
• /feed?bridge=bbcWorld
• /feed/bbcWorld?limit=20&fulltext=false
• /feed?bridge=apNews
• /feed/apNews?category=world&limit=20&fulltext=true
• /feed?bridge=abcNewsJustIn
• /feed/abcNewsJustIn?category=world&limit=20&fulltext=true
• /feed/linuxDo?limit=20&fulltext=true
• /feed/jin10Topics?category=middleEast&limit=30
• /feed/samsungMobileW24?limit=20
• /feed/innomad?limit=20&fulltext=true
• /feed/nessLabs?limit=20&fulltext=true
• /feed/foreverBlog?limit=30
• /feed/bokequan?limit=40
• /feed/hiKebaHost?limit=30&fulltext=true
• /feed/reutersMobile?category=world&limit=20
• /feed/pbcNews?limit=30
• /feed/ftChinese?limit=20&fulltext=false
• /feed/nytimesChinese?category=world&limit=20&fulltext=true
• /feed/politicoMobile?category=latestNews&limit=20&fulltext=true
• /feed/theAtlantic?limit=20&fulltext=true
• /feed/flyertDigest?limit=30
• /feed/bloombergMobile?category=latest&limit=20&fulltext=true
• /feed/scmpFullText?limit=20&fulltext=true
• /feed/morganStanleyGlobal?limit=20
• /feed/morganStanleyChina?limit=20
• /image?url=... - cached image proxy
• /telegram?chat_id=-1001234567890&limit=20 - RSS from Telegram Bot API updates
• /telegram?channel=public_channel_name&limit=20 - same feed, after the bot has received a post from the channel
• /telegram/webhook - Telegram Bot API webhook receiver

BBC World

BBC World News bridge:

/feed?bridge=bbcWorld
/feed/bbcWorld?limit=15&fulltext=true

Parameters:

• category: world
• limit: 1-50, default 15
• fulltext: default true; use 0 or false for summary mode

AP News

AP News bridge:

/feed?bridge=apNews
/feed/apNews?category=world&limit=15&fulltext=true

Parameters:

• category: home, factCheck, us, world, investigations
• limit: 1-50, default 15
• fulltext: default true; use 0 or false for summary mode

ABC News

ABC News bridge:

/feed?bridge=abcNewsJustIn
/feed/abcNewsJustIn?limit=15&fulltext=true
/feed/abcNewsJustIn?category=world&limit=15&fulltext=true
/feed/abcNewsJustIn?category=world&limit=10&fulltext=false

Parameters:

• category: justIn, world
• limit: 1-50, default 15
• fulltext: default true; use 0 or false for summary mode

Nikkei Asia secret

Full logged-in article content can use a Worker Secret:

wrangler secret put NIKKEI_ASIA_COOKIE

Migrated PHP sources

• linuxDo from quanwenrss.com_tjfZJ/linux.do/linux.do.php
• jin10Topics from quanwenrss.com_tjfZJ/jin10/jin10.php
• samsungMobileW24 from quanwenrss.com_tjfZJ/blog/samsungmobile/samsungmobile.php
• innomad from quanwenrss.com_tjfZJ/blog/innomad.io/generate_rss.php
• nessLabs from quanwenrss.com_tjfZJ/blog/nesslabs/nesslabs_rss.php
• foreverBlog from quanwenrss.com_tjfZJ/blog/foreverblog/fetch_rss.php
• bokequan from quanwenrss.com_tjfZJ/blog/bokequan/bokequan_rss.php
• hiKebaHost from quanwenrss.com_tjfZJ/blog/keba.host/generate_hi_keba_rss.php
• reutersMobile from quanwenrss.com_tjfZJ/reuters/reuters_rss.php
• pbcNews from quanwenrss.com_tjfZJ/pbc/zhengcejiedu.php
• ftChinese from quanwenrss.com_tjfZJ/ftchinese/ft_rss_gen.php
• nytimesChinese from quanwenrss.com_tjfZJ/nytimes/nytimes_rss.php
• politicoMobile from quanwenrss.com_tjfZJ/politico/rss.php and its category config in quanwenrss.com_tjfZJ/politico/config.php
• theAtlantic from quanwenrss.com_tjfZJ/Atlantic/rss.php
• flyertDigest from quanwenrss.com_tjfZJ/flyert/flyert-digest-rss.php
• bloombergMobile from quanwenrss.com_tjfZJ/bloomberg/bloomberg.php
• scmpFullText from quanwenrss.com_tjfZJ/scmp/scmp.php
• morganStanleyGlobal from quanwenrss.com_tjfZJ/morganstanley/morganstanley_global_rss.php
• morganStanleyChina from quanwenrss.com_tjfZJ/morganstanley/morganstanley_rss.php

Excluded source artifacts

These old PHP files are intentionally not separate Worker plugins:

• quanwenrss.com_tjfZJ/bloomberg/feeds/bilingual_rss.php - DeepL translation post-processor for generated RSS files
• quanwenrss.com_tjfZJ/morganstanley/generate_all_rss.php - batch wrapper around the two Morgan Stanley generators
• quanwenrss.com_tjfZJ/reuters/reuters_api_scraper.php - older Reuters pf/api/v3 generator; the endpoint returned 401 during migration, so reutersMobile uses the working mobile API instead
• quanwenrss.com_tjfZJ/reuters/reuters_api_scraper_debug.php - diagnostic helper for the old Reuters API
• quanwenrss.com_tjfZJ/scmp/fetch_articles.php - database population job; scmpFullText fetches RSS and article pages directly
• quanwenrss.com_tjfZJ/scmp/generate_atom.php - database-to-Atom generator; scmpFullText renders RSS in the Worker

Generated XML, caches, databases, translation post-processors, .env files, and vendor/ files are source artifacts, not Worker plugins.

SCMP secret

SCMP full text can use a Worker Secret for article cookies:

wrangler secret put SCMP_COOKIES

Scheduled feed pre-generation

Slow full-text feeds can be pre-generated by Cron and served from KV. This keeps reader requests fast after the scheduled job has warmed the XML cache.

Create a KV namespace and bind it as FEED_CACHE:

npx wrangler kv namespace create FEED_CACHE

Add the returned namespace ID to wrangler.toml:

[[kv_namespaces]]
binding = "FEED_CACHE"
id = "your-kv-namespace-id"

Configure the public Worker URL and the feeds to pre-generate:

[vars]
FEED_BASE_URL = "https://your-worker-domain.example"
FEED_PREGENERATE_URLS = """
/feed/apNews?category=world&limit=15&fulltext=true
/feed/abcNewsJustIn?category=world&limit=15&fulltext=true
"""
FEED_CACHE_FRESH_SECONDS = "1800"
FEED_CACHE_STALE_SECONDS = "86400"

FEED_PREGENERATE_URLS accepts newline-separated paths, comma-separated paths, or a JSON array. Absolute feed URLs also work and do not need FEED_BASE_URL.

Telegram Bot API feeds

The Telegram route is API-based and never scrapes t.me/s. It also supports private and restricted channels, provided the bot can receive their posts.

Create a bot with BotFather, add it to each source channel as an administrator, and bind a KV namespace as TELEGRAM_FEEDS (it may use the same namespace as FEED_CACHE):

[[kv_namespaces]]
binding = "TELEGRAM_FEEDS"
id = "your-kv-namespace-id"

Set these Worker secrets:

wrangler secret put TELEGRAM_BOT_TOKEN
wrangler secret put TELEGRAM_WEBHOOK_SECRET

Register the webhook, replacing the placeholders with the deployed Worker URL, bot token, and the exact webhook secret:

curl "https://api.telegram.org/bot<BOT_TOKEN>/setWebhook?url=https%3A%2F%2Fyour-worker.example%2Ftelegram%2Fwebhook&secret_token=<WEBHOOK_SECRET>&allowed_updates=%5B%22channel_post%22%2C%22edited_channel_post%22%5D"

Telegram only delivers posts created after the bot is made an administrator and the webhook is registered; its Bot API cannot fetch a channel's earlier history. Once the first post is received, use the numeric chat id shown by the update or the channel username:

/telegram?chat_id=-1001234567890&limit=20
/telegram?channel=public_channel_name&limit=20

Channel photos are served through /telegram/media so the bot token remains a Worker secret.

Deploy

npx wrangler deploy