`TypeError [ERR_INVALID_URL]: Invalid URL: [object Object]` for requests not made with an `URL` object

Environment

Name	Version
msw	0.28.1
node	15.14.0
OS	Debian

Request handlers

import { setupServer } from 'msw/node'
import { rest } from 'msw'

const server = setupServer(
  rest.get('https://github.com', (req, res, ctx) => {
        return res(ctx.status(200));
    })
)

server.listen()

Actual request

import Wreck from '@hapi/wreck';

const request = async () => {
    return await Wreck.get('https://github.com');
}

Current behavior

Requests made in application code that don’t use the URL object can’t be mocked, and cause an Invalid URL error:

    TypeError [ERR_INVALID_URL]: Invalid URL: [object Object]
        at new NodeError (node:internal/errors:329:5)
        at onParseError (node:internal/url:537:9)
        at new URL (node:internal/url:613:5)
        at CookieStore.get (/srv/node_modules/msw/node/lib/index.js:7404:28)
        at setRequestCookies (/srv/node_modules/msw/node/lib/index.js:7550:109)
        at /srv/node_modules/msw/node/lib/index.js:7609:21
        at Generator.next (<anonymous>)
        at /srv/node_modules/msw/node/lib/index.js:53:71
        at new Promise (<anonymous>)
        at __awaiter (/srv/node_modules/msw/node/lib/index.js:49:12) {
      input: '[object Object]',
      code: 'ERR_INVALID_URL',
      trace: [ { method: 'GET', url: 'https://github.com' } ]
    }

Expected behavior

Requests somewhat comforming to the structure of an URL object can be mocked as well.

Details

In my example, I’m using the Wreck module provided by Hapi to make HTTP requests. This causes the follow chain of events:

• Wreck uses a simple object to construct the url of the request (instead of the native URL)
• since msw calls .toString when setting request cookies and reading response cookies, this causes the above mentioned error

This could be fixed by using request.url.href instead of request.url.toString(), which also seems to be the behaviour of calling toString on a standard URL object.

I’ll be happy to provide a PR for this, but there might be a better way to support this (or it might not be desired to do so).

[kettanaito]

No worries! We can bump the “interceptors” version in a new pull request here and add it to the next release.

[lauraseidler]

I think what happens is that this condition is true:

https://github.com/mswjs/interceptors/blob/16fd35622efe6328497ade15e21a84f6afda0a93/src/interceptors/ClientRequest/utils/normalizeHttpRequestParams.ts#L70

also supported by debug logs:

http normalizeHttpRequestParams given a URL: {
  setHost: false,
  host: 'github.com',
  origin: 'https://github.com',
  searchParams: URLSearchParams {},
  protocol: 'https:',
  hostname: 'github.com',
  hash: '',
  search: '',
  pathname: '/',
  path: '/',
  href: 'https://github.com/',
  method: 'GET',
  headers: { host: 'github.com' },
  agent: Agent {
    _events: [Object: null prototype] {
      free: [Function (anonymous)],
      newListener: [Function: maybeEnableKeylog]
    },
    _eventsCount: 2,
    _maxListeners: undefined,
    defaultPort: 443,
    protocol: 'https:',
    options: { maxSockets: Infinity, path: null },
    requests: {},
    sockets: {},
    freeSockets: {},
    keepAliveMsecs: 1000,
    keepAlive: false,
    maxSockets: Infinity,
    maxFreeSockets: 256,
    scheduling: 'lifo',
    maxTotalSockets: Infinity,
    totalSocketCount: 0,
    maxCachedSessions: 100,
    _sessionCache: { map: {}, list: [] },
    [Symbol(kCapture)]: false
  }
} +0ms

So the interceptor assumes if origin is present, it’s a URL, and doesn’t process it further?