Feature Proposal: Preprocessed string types

[prescience-data]

I use Zod heavily for parsing environment variables and in each project need to reimplement transformer definitions to convert the raw env object (either from process.env or dotenv) to their intended types.

What would alleviate a bunch of pain in this use case is having built in <number|boolean|bigint|null>String(), for example:

const isString = (value: unknown) : value is string => typeof value === "string"

const booleanString = z.preprocess(
  (value) => (isString(value) ? value === "true" : value),
  z.boolean()
)

const numberString = z.preprocess(
  (value) => (isString(value) ? parseInt(value) : value),
  z.number()
)

const bigintString = z.preprocess(
  (value) => (isString(value) ? BigInt(value) : value),
  z.bigint()
)

const nullString = z.preprocess(
  (value) => (isString(value) && value === "null" ? null : value),
  z.null()
)

The ideal implementation would be for the z import to provide these like:

import { z } from "zod"

const EnvSchema = z.object({
  FOO_NUMBER: z.numberString().default(100),
  BAR_BOOL: z.booleanString().default(false)
  // etc...
})

Happy to do an MR if this is something that might be accepted?

[FlorianWendelborn]

To me this looks more like something that could/should be implemented by a utility library. You could even include the schema itself in it, by using it like z.preprocess(...castToBoolean), so it’d really not be a lot of added code if this isn’t in zod itself.

[prescience-data]

Maybe a better / more abstract interface could be something like

z.from(z.string()).number()

which would have the additional isomorphic benefit of things like

z.from(z.bigint()).number()

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[ConProgramming]

For any modern googlers - the solution is coercion