morss-tg

this is a cloudflare worker script that basically takes a list of feed urls, and based on the configured cron frequency decides which feed url it has to fetch at the time it's running.

my approach to saving cpu time and computation (since i'm using the free plan and i don't intend on paying :p) was to create a build.js script which compiles the list of feeds.txt in a few steps:

• read feeds.txt and clear the empty lines or the lines that start with #
• if there's a WORKER_CRON environment variable is provided, parse it
• if there's no WORKER_CRON environment variable is provided, calculate the needed time frequency.
• then the build script creates the wrangler.toml needed by wrangler to publish the worker. (i had to make this part a compile-time step since i wanted to use sensitive info like kv storage id or worker name hidden)
• the build script also produces a src/data.json which is imported by src/index.ts (see the DataJSON interface in src/util.ts )
• now the ESBuild bundles the files in src and outputs a single dist/worker.js that is the ACTUAL worker entry point used by wrangler

secrets

to let people fork this repo and publish their changes publicly without worrying about api keys and sensitive info, the ci workflows uses Github Secrets. (github secrets get redacted in the build logs or outputs):

• CLOUDFLARE_ACCOUNT_ID
• CLOUDFLARE_API_TOKEN
• DST_CHAT
• TG_API_TOKEN
• WORKER_KV_ID
• WORKER_NAME

these secret variable names are pretty self-explanatory except DST_CHAT! fill this variable value with the @username or numeric chat_id of a Telegram chat. (NOTE that these chat ids must be valid to Telegram's bot api)

environment variables

there are three plain text environment variables that are not sensitive:

• POST_MARKUP, Telegram bot api's parse_mode field (e.g. HTML)
• WORKER_CRON, an optional cron string (e.g. */5 * * * *). if not provided the build script generates one based on the feeds count.
• POST_FMT, the format string containing {TITLE}, {URL}, {DESC} and/or {TIME}