How to validate an object with keys that follow specific pattern (regex)?

[akrynski-1]

I have object with a shape similar to one below:

{
  foo: 123,
  bar: "abc",
  field_x: 1,
  field_y: 2,
  field_z: 3,
  field_something_else: 4,
  random_trash_i_dont_need: null
}

How can I validate it using valibot?
I have known and static fields: foo and bar
I have dynamic fields following specific key pattern: field_*, they all have the same type.
There can be fields I don't care about at all, they won't be in the schema, I don't want them to result in failed validation, they should also work with looseObject because maybe they will be useful somewhere else.

I tried playing with record and intersect but this will fail because keys like email and password don't match the regex here:

import * as v from 'valibot';

const Schema = v.object({
  email: v.pipe(v.string(), v.email()),
  password: v.pipe(v.string(), v.minLength(8)),
});

const PatternSchema = v.record(
  v.pipe(v.string(), v.regex(/^field_/)),
  v.number()
);

const result = v.safeParse(v.intersect([Schema, PatternSchema]), {
  email: 'jane@example.com',
  password: '12345678',
});

console.log(result);

It feels like it should be possible without hacks.
Thank you in advance!

[fabian-hiller]

Unfortunately, this harder then I wish because we do not support custom key schemas with objectWithRest. It is still possible to implement it, but it requires a bit more code. Here is an example. Test it out in our playground.

import * as v from 'valibot';

const BaseSchema = v.object({
  email: v.pipe(v.string(), v.email()),
  password: v.pipe(v.string(), v.minLength(8)),
});

const ExtendedSchema = v.pipe(
  v.objectWithRest(BaseSchema.entries, v.number()),
  v.check((input) => {
    for (const key in input) {
      if (!(key in BaseSchema.entries) && !key.startsWith('field_')) {
        return false;
      }
    }
    return true;
  }, 'Additional keys must be prefixed with "field_"'),
);

[akrynski-1]

Hmm, this wouldn't work for other random fields like random_trash_i_dont_need from my example object.
I didn't think about using check and running custom validations for the subset of the fields. I think it should be possible to extend this solution to handle unknown keys correctly. I don't know what would merging issues from other schemas look like.

I decided to keep it simple and validate these field_* keys on demand when needed (same as with the other, unknown fields).