Express routing matching extra URL tokens

Question

I would like to get URLs of the form (strictly) /forumId/slug/ with Express' router.get(). Additional tokens should result in a 404.

I came up with this path route: /:forumId([^?\/]{0,}):parameters1?/:slug([^?\/]{0,})?:parameters2? which according to express-route-tester behaves as expected.

However Express seems to also be catching paths with extra tokens after the slug. Directly passing a Javascript regular expression instead of a string has the same effect.

Here is the regex I used: /^\/(\d+)(?:\?[^\?\/]{0,})?(?:$|\/([^\?\/]{0,})(?:\?[^\?\/]{0,})?\/?$)/

Examples of strings that should match (those work):

/7721
/7721/
/7721?page=2
/7721/ForumTitle/
/7721/AnotherForumTitle?test/
/7721/YetAnotherForumTitle?page=2
/7721?page=3/ForumTitle?page=2
/7721?page=3/ForumTitle?page=2/

Example of strings that shouldn't match:

/7721?page=3/ForumTitle?page=2/threadId

Express uses path-to-regexp to parse the string and according to the docs (https://www.npmjs.com/package/path-to-regexp#user-content-usage), the 'end' option is set to true by default which seems to corroborate that the match should stop right after the slug.

What am I missing?

The version of Express is 4.16.4.

UPDATE: removing the part about the query parameters (?page=2) in the regex solved the problem. Why is that so? Was my regex flawed (I tried spotting possible greedy matches but couldn't find one) or is this expected from path-to-regexp?

Try this https://regexr.com/44hu7 and also remember about `^` and `$` operators for limit line. — Ihor Voronin, Dec 07 '18 at 15:42
@IhorVoronin 'ForumTitle' is just an example of title but could be any series of characters. I edited the question to include the regex I used that didn't work either. — Kathandrax, Dec 07 '18 at 17:13

Kathandrax · Accepted Answer · 2018-12-08T03:09:44.797

As explained in Express routing docs,

query strings are not part of the route path

(https://expressjs.com/en/guide/routing.html#route-paths).

The part trying to catch the query parameters in /^\/(\d+)(?:\?[^\?\/]{0,})?(?:$|\/([^\?\/]{0,})(?:\?[^\?\/]{0,})?\/?$)/ was therefore unnecessary, resulting in the simplified /^\/(\d{1,})(?:$|\/([^\/]{0,})\/?$)/:

^\/: beginning of the endpoint followed by a slash '/'
(\d{1,}): the forum identifier consisting of digits
(?:$|\/([^\/]{0,})\/?$)/: either ends there or captures an optional slug

This regex matched every provided example but also captured /7721/ForumTitle?page=2/test which should result in a 404. Upon further investigation I noticed '/test' was being captured as part of the query parameters in req.query ({"page": "2/test"}).

This is odd, since the slash '/' is supposed to be a reserved character for delimitation (Characters allowed in GET parameter). Additional information would be helpful.

EDIT: to quote Douglas Wilson (from there), Express.js uses Node.js core to parse the paths according to the url spec. Node.js's URL module follows the WHATWG URL Standard.

Express routing matching extra URL tokens

1 Answers1