Leveraging Transform to Create Specialized Types

Sometimes you want a transform: say you accept a string but store it as a Date. Zod’s .transform() method has your back. You still validate the input first, then morph it into whatever form you need:

const dateSchema = z
	.string()
	.refine((val) => !isNaN(new Date(val).valueOf()), {
		message: 'Invalid date string',
	})
	.transform((val) => new Date(val));

// Now your run-of-the-mill string is validated and transformed into a Date object
const date: Date = dateSchema.parse('2025-03-20');

This approach is great for bridging the gap between what your API or form input looks like versus what your code actually needs.

Pipelining Transforms

Zod allows chaining transforms. You can do .transform() multiple times if each transform’s output type matches the next transform’s input type.

Example: Combining Validation + Chained Transforms

import { z } from 'zod';

// Start with a string
const pipeline = z
	.string()
	.transform((val) => val.trim()) // first transform
	.transform((trimmed) => trimmed.toUpperCase()); // second transform

// " hello " -> "hello" -> "HELLO"
pipeline.parse('  hello  ');

You can also do advanced checks between transforms. For multi-step transformations, consider if .superRefine() might be simpler if you need to add multiple error issues.