如何在我的应用程序中限制对 API 子集的访问？

Question

Let's say I have app.js file which exposes 5 operations (mix of GET & POST): /op1, /op2, /op3, /op4 & /op5. I want to run same app.js file on 2 different servers wherein I want /op1, /op2, /op3 services to be accessible on server running on port# 3000 and /op4 and /op5 services to be accessible from port# 500. Is there anyway I can achieve this without creating separate app.js files for each server with relevant services?

Answer 1

You can create conditional code that registers certain routes only under some conditions.

Assuming the desired port to run the server on is either passed as a command line argument or in an environment variable and you've read that into a local variable port , you can then just conditionally check that port value and decide which routes to register:

if (port === 3000) {
     app.get("/op1", ...)
     app.get("/op2", ...)
     app.get("/op3", ...)
} else {
     app.get("/op4", ...)
     app.get("/op5", ...)
}

It may be cleaner to break the separate groups of routes into their own modules and then conditionally import only the routes files that are appropriate for a given situation.

if (port === 3000) {
     require('routesA')(app);
} else {
     require('routesB')(app);
}

Each of those modules would export a function that gets called here with the app object as an argument so each module can then register it's routes as appropriate.

---

My personal opinion is somewhat in line with some of what was expressed in the comments that you shouldn't really need to do it this way. If you have two separate servers with different routes on different ports, then you should probably have two separate main server files that each establish the configuration of one of the servers.

With appropriate use of modules, you can share as much code as you want between the two servers (as much as 99.99% if you want).

But, it does make some logical sense that the main server file would be separate for the two servers since they are intended to be different servers. It could literally just be a tiny shell that establishes the port and which route modules it wants to load.

Answer 2

A simple solution to your question would be to use an api gateway setup.This api gateway/load balancer has to be run as a separate server and the single endpoint exposed to client apps. U can use nginx to forward your requests to specific servers based on the endpoint . So /op1,/op2,/op3 forward request to server1 /op4,/op5 forward request to server2 .

https://docs.nginx.com/nginx/admin-guide/web-server/reverse-proxy/

But as mentioned in comments, you can get an eye on the microservices architecture which seems to more elegant and efficient way to go about this problem.

https://microservices.io/patterns/microservices.html