electrodb-zod-example.ts

import { z } from 'zod';

const VehicleSchema = z.object({
  id: z.string(),
  itemType: z.literal('VEHICLE'),
  licencePlateNumber: z.string(),
});

const HouseSchema = z.object({
  id: z.string(),
  itemType: z.literal('HOUSE'),
  address: z.string(),
  city: z.string(),
  country: z.string(),
});


const ItemModelZod = z.discriminatedUnion('itemType', [VehicleSchema, HouseSchema]);


const ItemEntity = new Entity({
  model: {
    entity: 'item',
    service: 'auth',
    version: '01',
  },
  attributes: {
    id: {
      type: 'string',
    },
    itemType: {
      type: 'string'
    },
    /* 
        also allow any other attributes to be inserted besides these two i.e. in this case 
        - licencePlateNumber
        - address
        - city
        - country
        without having to explicitly define them here
        
        With for example large/complex discriminated unions keeping this up-to-date could be quite cumbersome
        and force us to basically list all the possible attributes that an item in the union might have
        
        One interesting option could of course be if these could be automatically created from the zod schema...
    */
  },
  indexes: {
    byItemType: {
      pk: {
        field: 'pk',
        composite: ['itemType']
      },
      sk: {
        field: 'sk',
        composite: ['id']
      }
    }
  }
}, { table, client });


const item = ItemModelZod.parse({
  id: '123456',
  itemType: 'VEHICLE',
  licencePlateNumber: 'abcde',
});
await ItemEntity.put(item).go();

In it case helps anyone, here is link to discussion with type inference / assertion solution that has been working quite well tywalch/electrodb#216 (comment)