/gqtx

Code-first type-safe GraphQL Server without codegen or metaprogramming

Primary LanguageTypeScript

Getting Started

yarn add gqtx

Type-safety without manual work

gqtx is a thin layer on top of graphql-js for writing a type-safe GraphQL server in TypeScript. It provides you with a set of helper functions to create an intermediate representation of a GraphQL schema, and then converts that schema to a raw graphql-js schema. So you get to use everything from the reference implementation of GraphQL, but with way more type safety.

If a schema compiles, the following holds:

  • The type of a field agrees with the return type of the resolver.
  • The arguments of a field agrees with the accepted arguments of the resolver.
  • The source of a field agrees with the type of the object to which it belongs.
  • The return type of the resolver will not be input types (InputObject)
  • The arguments of a field will not be abstract types (Interface, Union)
  • The context argument for all resolver functions in a schema agree.

Most importantly, we achieve all this without having to:

  • Set up code generation tools
  • Write SDL and having your schema partially defined in code and in a DSL file
  • Require special compiler magic such as reflect-metadata and decorators

What does it look like?

import { createTypesFactory, buildGraphQLSchema } from 'gqtx';

enum Role {
  Admin,
  User,
}

type User = {
  id: string;
  role: Role;
  name: string;
};

const users: User[] = [
  { id: "1", role: Role.Admin, name: 'Sikan' },
  { id: "2", role: Role.User, name: 'Nicole' },
];

type AppContext = {
  viewerId: 1,
  users: User[]
}

// We can set the app context type once, and it will
// be automatically inferred for all our resolvers! :)
const t = createTypesFactory<AppContext>();

const RoleEnum = t.enumType({
  name: 'Role',
  description: 'A user role',
  values: [{ name: 'Admin', value: Role.Admin }, { name: 'User', value: Role.User }],
});

const UserType = t.objectType<User>({
  name: 'User',
  description: 'A User',
  fields: () => [
    t.defaultField('id', t.NonNull(t.ID)),
    t.defaultField('role', t.NonNull(RoleEnum)),
    // `defaultField` is the safe version of a default resolver
    // field. In this case, field 'name' must exist on `User`
    // and its type must be `string`
    t.defaultField('name', t.NonNull(t.String)),
  ],
});

const Query = t.queryType({
  fields: [
    t.field('userById', {
      type: UserType,
      args: {
        id: t.arg(t.NonNullInput(t.ID)),
      },
      resolve: (_, args, ctx) => {
        // `args` is automatically inferred as { id: string }
        // `ctx` is also automatically inferred as AppContext
        //  All with no extra work!
        const user = ctx.users.find(u => u.id === args.id);
        // Also ensures we return an `User | null` type :)
        return user || null;
      },
    })
  ],
});

const schema = buildGraphQLSchema({
  query: Query,
});

Use your favorite server option to serve the schema!

import express from 'express';
import graphqlHTTP from 'express-graphql';

const app = express();

app.use(
  '/graphql',
  graphqlHTTP({
    schema,
    graphiql: true,
  })
);

app.listen(4000);

To Recap

  • We created an intermediate representation of a GraphQL schema via the helper functions exported by this library.
  • Then, we converted the schema to a real graphql-js schema by calling buildGraphQLSchema at server startup time.
  • Used existing express middleware express-graphql to server our schema with graphiql explorer
  • That's it! We get a fully typesafe server with almost zero type annotation needed