Schemas

Summary

We use schemas for the purpose of validating data. But instead of "just" validating data we adopt the approach of parsing instead of validating. This way all data enters your code through typed boundaries and nothing gets exiled and every part is taken care of.

Cook

import { z, schemaForType } from "@dendronhq/common-all";

export const aSchema = schemaForType<ASchemaType>()(
  z
    .object({
        aKey: 'aValue'
    })
)

The schemaForType function allows us the define the type outside of the library (zod in our case). The reason is that most types currently exist outside of the library and to keep types seperate from dependencies seems like a good thing. Note that when

parsing

To parse a schema use the parse function which will return a `Result` type.

import { parse, z } from "@dendronhq/common-all";

const schema = z.object({ /* some schema */ })
const input = /* some unknown (from typescript POV) data */
const outputResult = parse(schema, input, "Optional error message")
if (outputResult.isOk()) {
    outputResult.value // do something with the parsed/valid input
}

optional vs default

const schema = z.object({
  y: z.string().optional(),
  x: z.string().default("str"),
});

type InputType = z.input<typeof schema> // y optional, x optional
type OutputType = z.infer<typeof schema> // y optional, x non-optional

This will allow the input for x to be undefined but the output to be non-optional. So it will not throw but return the field (after being parsed) with the default value. See: https://github.com/colinhacks/zod/issues/1456#issuecomment-1271874390

Reference

Backlinks