Skip to content

metabolize/1schema

Repository files navigation

1schema

1schema provides friendly developer tooling for runtime checking of TypeScript schemas. It uses the powerful and effective ts-json-schema-generator to generate JSON Schema which can be validated at runtime using 1schema's built-in support for Ajv or any other JSON Schema validator. For example, in our Python projects we use jsonschema.

This is way better than writing and maintaining JSON Schema by hand, and if you use TypeScript you also get the benefit of compile-time checking.

Since the schemas are written in TypeScript it's ideal for TypeScript projects, however it's easy to use in JavaScript projects, and works in non-JavaScript projects too.

I didn't write this blog post, but it explains the idea really well.

We've been using this pattern for a while at Metabolize–Curvewise, both for validating user uploads and validating across interface boundaries. This open-source tooling is new and considered alpha. Developer feedback and contributions welcome!

Why not use ts-json-schema-generator directly?

You can! However, to provide a smooth development experience, 1schema provides a few niceties:

  1. Globbing
  2. Formatting
  3. Checking that the schema are up to date
  4. Pruning JSON schemas when their corresponding source files are removed

How it works

  • In your project, create a schema.ts file:

    export type ContactMethodType = 'mobile' | 'home' | 'work' | 'other'
    
    export interface Address {
      streetAddress: string
      locality: string
      region: string
      postalCode: string
      country: string
    }
    
    export interface Contact {
      familyName: string
      givenName: string
      honorificPrefix?: string
      honorificSuffix?: string
      nickname?: string
      url?: string
      imageUrl?: string
      email: {
        address: string
        type: ContactMethodType
      }[]
      phone: {
        phoneNumber: string
        type: ContactMethodType
      }[]
      address: Array<Address & { type: ContactMethodType }>
      birthdate: Date
      gender?: string
    }
  • Run 1schema update to generate generated/schema.json with all exported types and their dependents. Check in this file.

  • COMING SOON: At runtime, import { validate } from '1schema' and validate(inputData).

  • If you're using TypeScript, cast the validated input to the appropriate type from your schema (e.g. const contact = inputData as Contact) to get compile-time checking.

  • In CI, run 1schema check to verify the generated schema are up to date.

Your schema files are just ordinary TypeScript files so they can import and extend other TypeScript types and schemas, so long as the types are supported by ts-json-schema-validator.

If you have a tsconfig.json it will be used and if not one is provided for you.

You can spread schemas across multiple files: If you create: this.schema.ts, that.schema.ts, the-other/schema.ts. Running 1schema update will generate generated/this.schema.json, generated/that.schema.json and the-other/schema.json.

Related projects

We use 1schema with Werkit, a toolkit for encapsulating Python functions on AWS Lambda.

Acknowledgements

Serious thanks to Dominik Moritz for maintaining the wonderful ts-json-schema-validator tool. And thanks to Jacob Beard who turned me back onto JSON Schema in the first place.