Define the schema

Micrograph uses the cohere schema lib. cohere is a tiny lib that compiles attributes and relationships to JSON that is easily transformed into a GraphQL schema, among other things.

The following is an example of a basic schema. Refer to the data modeling section for more information on creating schemas. We're using GraphQL scalar types for attributes.

// schema.js
import Schema, { hasMany, belongsTo } from 'cohere';
import { GraphQLString, GraphQLNonNull, GraphQLList } from 'graphql';
import Blog from './models/Blog';
import User from './models/User';

const schema = new Schema();

schema.defineType('user', {
  attributes: {
    name: GraphQLString,
    email: new GraphQLNonNull(GraphQLString),
  },
  relationships: {
    blogs: hasMany('blog', 'author'),
  },
  model: User,
});

schema.defineType('blog', {
  attributes: {
    title: GraphQLString,    
  },
  relationships: {
    author: belongsTo('user', 'blogs'),
  },
  model: Blog,
});

schema.compile();

export default schema;

cohere offers three relationship types: hasMany, belongsTo, and hasOne. For each relationship, the first argument is the type that the relationship refers to, i.e. the user's blogs relationship points to the blog type. The second argument is the inverse field. For example, the schema above declares that the author relationship of a blog points to the user type and that the user type has a blogs relationship that points back towards the blog type.

The third argument is an optional object.

After defining all your types, you must invoke the compile method, which simply links each relationship to each other. Compiling will let you know if you have any schema errors as well (such as a missing inverse or a relationship that points to an undefined type). This is useful becuase you'll never accidentally deploy a partially broken schema, which is really easy to do once your schema includes >20 types.