- Notifications
You must be signed in to change notification settings - Fork21
Create GraphQL schema with TypeScript classes.
License
prismake/typegql
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
typegql is set of decorators allowing creating GraphQL APIs quickly and in type-safe way.
- Basic Express example
- Typeorm integration example
- Forward resolution - eg. query only needed db fields
- Nested mutations or queries
- Custom decorators / Higher order decorators
- Serverless eg. AWS Lambda
- Merge schemas
Example below is able to resolve such query
query {hello(name:"Bob") # will resolve to 'Hello, Bob!'}
import{compileSchema,SchemaRoot,Query}from'typegql';@SchemaRoot()classSuperSchema{ @Query()hello(name:string):string{return`Hello,${name}!`;}}constcompiledSchema=compileSchema({roots:[SuperSchema]});
compiledSchema
is regular executable schema compatible withgraphql-js
library.
To use it withexpress
, you'd have to simply:
import*asexpressfrom'express';import*asgraphqlHTTPfrom'express-graphql';constapp=express();app.use('/graphql',graphqlHTTP({schema:compiledSchema,graphiql:true,}),);app.listen(3000,()=>console.log('Graphql API ready on http://localhost:3000/graphql'));
For now, our query field returned scalar (string). Let's return something more complex. Schema will look like:
mutation {createProduct(name:"Chair",price:99.99) {namepriceisExpensive }}
Such query will have a bit more code and here it is:
import{Schema,Query,ObjectType,Field,Mutation,compileSchema}from'typegql';@ObjectType({description:'Simple product object type'})classProduct{ @Field()name:string; @Field()price:number; @Field()isExpensive(){returnthis.price>50;}}@Schema()classSuperSchema{ @Mutation()createProduct(name:string,price:number):Product{constproduct=newProduct();product.name=name;product.price=price;returnproduct;}}constcompiledSchema=compileSchema(SuperSchema);
Until now,typegql
was able to guess type of every field from typescript type definitions.
There are, however, some cases where we'd have to define them explicitly.
- We want to strictly tell if field is nullable or not
- We want to be explicit about if some
number
type isFloat
orInt
(GraphQLFloat
orGraphQLInt
) etc - Function we use returns type of
Promise<SomeType>
while field itself is typed asSomeType
- List (Array) type is used. (For now, typescript
Reflect
api is not able to guess type of single array item. This might change in the future)
Let's modify ourProduct
so it has additionalcategories
field that will return array of strings. For sake of readibility, I'll ommit all fields we've defined previously.
@ObjectType()classProduct{ @Field({type:[String]})// note we can use any native type like GraphQLString!categories():string[]{return['Tables','Furniture'];}}
We've added{ type: [String] }
as@Field
options. Type can be anything that is resolvable toGraphQL
type
- Native JS scalars:
String
,Number
,Boolean
. - Any type that is already compiled to
graphql
eg.GraphQLFloat
or any type from external graphql library etc - Every class decorated with
@ObjectType
- One element array of any of above for list types eg.
[String]
or[GraphQLFloat]
Every field function we write can beasync
and returnPromise
. Let's say, instead of hard-coding our categories, we want to fetch it from some external API:
@ObjectType()classProduct{ @Field({type:[String]})// note we can use any native type like GraphQLString!asynccategories():Promise<string[]>{constcategories=awaitapi.fetchCategories();returncategories.map(cat=>cat.name);}}
Before version1.0.0
consider APIs oftypegql
to be subject to change. We encourage you to try this library out and provide us feedback so we can polish it to be as usable and efficent as possible.