Running yt-dlp, youtubei.js, and googlevideo in a dedicated fastify service.

Requirements:

• Node.js > 22 (Top level)
• Python 3 (ytdlp-server)
• Protobuf

This is a rapidly evolving solution, so its doesn't make much sense to document thoroughly.

This service starts two processes: A fastify server which listens on 5000 and a python server (loopback only) on 5001. The fastify server makes requests to the flask server.

This is orchestrated in a Dockerfile, and can also boot from an npm start.

Use package imports aliases for cross-area imports that would otherwise require deep ../../ paths, and keep relative imports for nearby sibling files. Examples:

import { build } from '#test/helper.js'
/** @import { OnesieFormatResults } from '#lib/onesie/index.js' */
/** @import { ExtractKnownResponseType } from '#types/fastify-utils.js' */

For TypeScript source files behind an alias, import the runtime .js specifier and let package.json#imports map it to the .ts source for type checking.

Python Notes