Qpoint

Intelligence at the edge - an edge endpoint framework

Compose powerful edge capabilities to analyze, transform, reject, or proxy traffic as it passes through the edge to your apps.

Designed to run within worker runtimes, a qpoint endpoint can be deployed trivially to edge networks like Cloudflare Workers and Deno Deploy, or with the help of Qpoint, deploy to any platform including your own servers.

Example

import Endpoint from '@qpoint/endpoint'
import proxy from '@qpoint/proxy'
import maskUrls from '@qpoint/mask-urls'
import replaceContent from '@qpoint/replace-content'
import rewriteHtml from '@qpoint/rewrite-html'

// initialize and export the endpoint
export default new Endpoint()

  // proxy request to app
  .use(proxy({ appUrl:"https://qdemo.io" }))

  // mask urls in html response
  .use(maskUrls())

  // replace occurrences of qdemo with qpoint
  .use(replaceContent({ rules: [{ from: 'qdemo', to: 'qpoint' }] }))

  // rewrite html (trigger htmlrewriter rules)
  .use(rewriteHtml())

Composable adapters

Adapters are middleware functions to be executed in a chain, each potentially modifying the request/response until finally returning the response.

Example: Reject the request (at the edge) if no auth is provided

endpoint.use((ctx: Context, next: Function) => {
  
  // check for the Authorization header
  if (!ctx.request.headers.has("Authorization")) {
    // set the response to unauthorized
    ctx.response = new Response(null, { status: 401 });

    // return without calling next() to terminate the chain
    return
  }

  // continue the chain to the next
  return next();
})

Context, Request and Response

Each adapter receives a Qpoint Context object that wraps an incoming request and the corresponding response. ctx is often used as the parameter name for the context object.

endpoint.use(async (ctx: Context, next: Function) => { await next(); });

After each of the adapters have run, the response as set on the context will be returned.

Proxies and Load Balancers

A very common case for Qpoint is building intelligent proxies and load balancers, and since the original Request object cannot be modified, the proxy is a copy of the original request that can be fetched by a proxy or load-balancer adapter.

In such a scenario, adapters that need to modify the request before a proxy fetch occurs will sequencially modify or replace the proxy instance as the chain progresses.

Releasing New Versions

To release a new version create a New Release in GitHub with an incremented tag in the format of v#.#.#. GitHub will automatically generate the changelog since the last release version. After creation, a GitHub action will be kicked off to build, tests, set the npm version, and publish.