monolayer / schema / json

Function: json()

json<T>(): PgJson<T, T>

Column that stores JSON data.

Type Parameters

Type Parameter	Default type
T extends JsonValue	JsonValue

Returns

PgJson<T, T>

Remarks

Data stored is an exact copy of the input text and processing functions must reparse on each execution.

Does not support indexing.

Kysely database schema type definition

type JsonArray = JsonValue[];
type JsonValue = boolean | number | string | Record<string, any> | JsonArray;
{
  readonly __select__: JsonValue | null;
  readonly __insert__: JsonValue | null | undefined;
  readonly __update__: JsonValue | null;
};

Nullability and optionality will change according to the column's constraints, generated values, and default data values.

You can customize the data type of the column by providing a type argument to the json function.

Warning: the Zod schema for a json column only validates that data can be conforms to the JsonValue type. When using a custom data type you shoud adapt it. See examples.

Zod Schema

Types:

// type JsonArray = JsonValue[];
// type JsonValue = boolean | number | string | Record<string, any> | JsonArray;
{
  input?: JsonValue | null | undefined;
  output?: JsonValue | null | undefined;
}

Nullability and optionality will change according to the column's constraints, generated values, and default data values.

Validations:

• Explicit undefined values are rejected.
• Input values must be JsonValue or null.
• String values must be valid JSON.
• Record values must be convertible to a JSON string.

Example

Default Data Type

import { json, schema, table } from "monolayer/pg";
import { zodSchema } from "monolayer/zod";

const dbSchema = schema({
  tables: {
    example: table({
      columns: {
        document: json(),
      },
    }),
  },
});

// Kysely database schema type
type DB = typeof dbSchema.infer;
// Zod Schema
const schema = zodSchema(database.tables.example);

Custom Data Example

import { json, schema, table } from "monolayer/pg";
import { zodSchema } from "monolayer/zod";

type Data = { count: number; name: string  };

const dbSchema = schema({
  tables: {
    example: table({
      columns: {
        info: json<Data>(),
      },
    }),
  },
});

// Kysely database schema type
type DB = typeof dbSchema.infer;

// Zod Schema
const schemaShape = zodSchema(database.tables.example).shape;
const schema = z.object({
  ...schemaShape,
  info: schemaShape.id.superRefine((data, ctx) => {
    const objectKeys = Object.keys(data).sort();
    if (
      objectKeys.length !== 2 ||
      objectKeys[0] !== "count" ||
      typeof objectKeys[0] !== "number" ||
      objectKeys[1] !== "name"
      typeof objectKeys[1] !== "string" ||
    ) {
      ctx.addIssue({
        code: z.ZodIssueCode.custom,
        message: "Invalid data",
      });
    }
    return z.NEVER;
  }),
});

See

PostgreSQL Docs: json