Zod Schemas

You can generate Zod schemas for tables declared with @monolayer/pg with the zodSchema function before inserting or updating data to the database.

The Zod schemas will take into account:

The data type of the column.
The column nullability and its default value.
Whether the column is a primary key or generated.
Specific column data type validations that match validations at database level. For example, the range of an integer column is: -2147483648 to +2147483647.

Validating Schemas

import {
  enumType,
  enumerated,
  integer,
  primaryKey,
  table,
  text,
  timestampWithTimeZone
} from "@monolayer/pg/schema"
import { zodSchema } from "@monolayer/pg/zod"
import { sql } from "kysely";
import z from "zod";

// `users` table
const userRole = enumType("user_role", ["admin", "user"]);
const users = table({
  columns: {
    id: integer().generatedAlwaysAsIdentity(),
    name: text(),
    email: text().notNull(),
    role: enumerated(userRole).notNull(),
    createdAt: timestampWithTimeZone().default(sql`now()`).notNull(),
  },
  constraints: {
    primaryKey: primaryKey(["id"]),
  },
});

// Zod Schema for `users`
const schema = zodSchema(users);

// Usage
const user = schema.parse({
  name: "John Smith",
  email: 'john@smith.com',
  role: 'admin',
});

type InputType = z.input<typeof schema>;
//{
//   id: never;
//   name?: string | null | undefined;
//   email: string;
//   role: "user" | "admin";
//   createdAt?: Date | string | undefined;
//}

type OutputType = z.output<typeof schema>;
//{
//   id: never;
//   name?: string | null | undefined;
//   email: string;
//   role: "user" | "admin";
//   createdAt?: Date | undefined;
// }

Extending Schemas

You can use the Zod API to customize schemas.

Example

Adding an email validation to the email column in the users table from the previous example:

const schemaWithEmailValidation = zodSchema(users).extend({
  email: schema.shape.email.pipe(z.string().email())
})

type InputTypeEmail = z.input<typeof schemaWithEmailValidation>;
//{
//   id: never;
//   name: string;
//   email: string;
//   role: "user" | "admin";
//   createdAt?: Date | string | undefined;
//}
type OutputTypeEmail = z.output<typeof schemaWithEmailValidation>;
//{
//   id: never;
//   name: string;
//   email: string;
//   role: "user" | "admin";
//   createdAt?: Date | undefined;
// }

See: Zod Objects

Default input and output types

Each column data type has input and output types in the generated schema:

Boolish and JsonValue types

type Boolish = "true" | "false" | "yes" | "no" |
               1 | 0 | "1" | "0" | "on" | "off";

type JsonArray = JsonValue[];
type JsonValue = boolean | number | string |
                 Record<string, unknown> | JsonArray;

Nullability and optionality

Depending on a column constraints (NOT NULL, primary key), the default data value, and whether it's generated, input and output types in the schema will change:

Column	Input and output types allow `null`	Input and output types allow `undefined`
with default data value	N/A	yes
with `NOT NULL` constraint	no	no
with `NOT NULL` constraint and default data value	no	yes
primary key	no	no
generated by default as identity	no	yes
serial	no	yes
bigserial	no	yes

Generated always by identity columns

Generated always by identity columns have never as input and output type.