How do I generate a GraphQL API from existing REST APIs?

You generally do this by building a GraphQL layer (gateway/facade) on top of your existing REST endpoints, rather than “converting” REST into GraphQL automatically.

Common approaches

1. Write a GraphQL server that wraps REST

Create GraphQL types, queries, and mutations, then implement resolvers that call your REST APIs.

Example flow:

• GraphQL query: user(id: "1")
• Resolver calls: GET /users/1
• Resolver maps REST response into GraphQL schema

This is the most flexible and common approach.

---

2. Use API composition / GraphQL gateway

If you have multiple REST services, place a single GraphQL gateway in front of them.

Benefits:

• One GraphQL endpoint for clients
• Aggregates data from multiple REST APIs
• Hides backend complexity
• Allows client-specific shaping of data

---

3. Use code generation / scaffolding tools

Some tools help you bootstrap GraphQL from REST/OpenAPI specs, then you customize the parts that need business logic.

Helpful if your REST APIs already have:

• OpenAPI/Swagger definitions
• Consistent resource models
• Stable endpoints

Examples of useful tools/approaches:

• GraphQL Mesh: can wrap REST/OpenAPI and expose GraphQL quickly
• Apollo Server + custom resolvers
• Hasura / PostGraphile are more for databases, not REST
• Hot Chocolate (C#) / Spring for GraphQL (Java) if you want to build a gateway in those ecosystems

---

Typical architecture

1. Define GraphQL schema

   • Types: User, Order, Product
   • Queries: user, orders, product
   • Mutations: createOrder, updateUser

2. Implement resolvers

   • Each resolver calls one or more REST endpoints
   • Map REST fields to GraphQL fields

3. Handle performance

   • Use batching/caching to avoid N+1 REST calls
   • Consider DataLoader or request-level caching

4. Add auth and error handling

   • Forward tokens to REST services
   • Translate REST errors into GraphQL-friendly errors

5. Expose a single GraphQL endpoint

   • Clients use GraphQL
   • REST stays behind the scenes

---

Simple example

GraphQL schema

type User {
  id: ID!
  name: String!
  email: String
}

type Query {
  user(id: ID!): User
}

Resolver

const resolvers = {
  Query: {
    user: async (_, { id }, { restClient }) => {
      const res = await restClient.get(`/users/${id}`);
      return res.data;
    }
  }
};

---

Things to watch out for

• N+1 problem: GraphQL can trigger many REST calls if nested fields are resolved independently
• Over-fetching/under-fetching: GraphQL helps clients, but your resolvers still need efficient backend calls
• Schema design: Don’t mirror REST 1:1 blindly; model the API around client needs
• Pagination: Standardize how REST pagination maps to GraphQL connections or lists
• Versioning: GraphQL usually evolves by adding fields instead of versioning endpoints

---

If you want the fastest path

If you already have OpenAPI/Swagger specs, try GraphQL Mesh first. It can often generate a GraphQL layer over REST APIs with minimal code.

---

If you want, I can show you:

1. a Node.js/Apollo example,
2. a GraphQL Mesh setup, or
3. how to do this from an OpenAPI spec.