@saws/api
v1.0.11
Published
<div align='center'>
Downloads
2
Readme
API Services
Services for building APIs on top of APIGateway and AWS Lambda.
Table of Contents
Installation
From the command line run:
npm install @saws/api
Then add one of the included services (RestAPIService
or GraphQLApiService
) to your saws.js
file.
Development
In development, API services will build your API from the entrypoint file and register it with the local Lambda server. It will then start up an HTTP server that will behave similarly to API Gateway and route calls to your API server. Anytime any files change in your API service, it will rebuild the code automatically.
In development, the HTTP server that is started for your GraphQLApiService
will also expose an endpoint to access a Graphiql IDE to explore your GraphQL server. You can access it at http://localhost:PORT/graphiql
If your API service depends on the @saws/cognito
CognitoService
. It will also authenticate all HTTP requests against the CognitoService
's user pool.
The first time you run npx saws dev
after adding a new API service to your saws.js
file, it will install any missing dependencies and create a folder with a hello world lambda function entrypoint.
Deployment
When you deploy an API service, a number of resources will be stood up and configured for you:
- S3 bucket for holding your Lambda function code
- Lambda function
- API gateway configured to trigger your lambda function
- If your API service depends on a
CognitoService
it will also configure API Gateway with an authorizer requiring aBearer
token in theAuthorization
header. That token should be the access token from an authenticated session with your cognito user pool. For more information on how to use theCognitoService
and it's libraries to obtain an access token, see@saws/cognito
README
Services
There are two flavors of API services. RestAPIService
and GraphQLApiService
. At the moment, @saws/api
does not provide preconfigured Axios
or Apollo
clients to interact with your rest and graphql apis. Though this could be added in the future.
RestAPIService
You can require the RestAPIService
and use it in your saws.js
file like so:
const { RestAPIService } = require('@saws/api/rest-api-service')
const restApi = new RestAPIService({
name: 'my-rest-api'
})
module.exports = restApi
The RestAPIService
constructor accepts the following options:
name: string
The name of your service. This should be unique across all of your services.
dependencies: ServiceDefinition[]
An array of all of the other services this service depends on. This will ensure that permissions, environment variables, and execution order are all set up.
handler: string
Allows you to provide the path to your entrypoint if it is not index.ts
.
externalPackages: string[]
Allows you to provide a list of modules or files to not include in your code bundle.
port: number
The port to run the HTTP server on. If it's already in use, a different port will be automatically selected for you.
include: string[]
A list of additional files you would like to include in your ZIP archive uploaded to S3 and used by Lambda. This is useful for files that you might need to read from disk. The file path in the zip archive will match the path from the root of your API service.
For example, given the following directory structure:
- my-rest-api
- index.ts
- other-folder
- include-this-file.txt
- other-module.ts
The ZIP archive structure will look like:
- my-rest-api
- index.js
- other-folder
- include-this-file.txt
So make sure your path handling logic will still work when your code is bundled.
GraphQLAPIService
You can require the GraphQLAPIService
and use it in your saws.js
like so:
const { GraphQLAPIService } = require('@saws/api/rest-api-service')
const graphqlApi = new GraphQLAPIService({
name: 'my-graphql-api'
})
module.exports = graphqlApi
The GraphQLAPIService
constructor accepts the following options:
name: string
The name of your service. This should be unique across all of your services.
dependencies: ServiceDefinition[]
An array of all of the other services this service depends on. This will ensure that permissions, environment variables, and execution order are all set up.
handler: string
Allows you to provide the path to your entrypoint if it is not index.ts
.
externalPackages: string[]
Allows you to provide a list of modules or files to not include in your code bundle.
port: number
The port to run the HTTP server on. If it's already in use, a different port will be automatically selected for you.
include: string[]
A list of additional files you would like to include in your ZIP archive uploaded to S3 and used by Lambda. This is useful for files that you might need to read from disk. The file path in the zip archive will match the path from the root of your API service.
For example, given the following directory structure:
- my-rest-api
- index.ts
- other-folder
- include-this-file.txt
- other-module.ts
The ZIP archive structure will look like:
- my-rest-api
- index.js
- other-folder
- include-this-file.txt
So make sure your path handling logic will still work when your code is bundled.
When used as a dependency
When an API service is used as a dependency to other services, it will automatically attach (where applicable) the following environment variables into the dependent services:
SERVICE_NAME_API_URL: string
- the URL of the api
Libraries
@saws/api
includes 2 libraries to help establish your API entrypoints. RestApi
for RestAPIService
s and GraphQLAPI
for GraphQLAPIService
s.
RestAPI
The RestAPI
library is a class that accepts an express application object and makes it easy to create your Lambda function entrypoint for a rest api.
Example usage:
import { RestAPI } from '@saws/api/rest-api'
import express from 'express'
const app = express()
// do whatever you need to do with your express app
app.use(...)
const restApi = new RestApi(app)
export const handler = restApi.createLambdaHandler()
When your RestAPIService
depends on a CognitoService
, the information from the provided access token will be attached to req.user
.
GraphQLAPI
The GraphQLAPI
library is a class that accepts typedefs
, resolvers
and an onError
callback. With these options it will create an ApolloServer
, configured to run in a Lambda function.
Example usage:
import { GraphQLAPI } from '@saws/api/graphql-api';
import { ApolloContext } from 'apollo-server-lambda';
import { mergeTypeDefs, mergeResolvers } from "@graphql-tools/merge";
import {
types as myTypeDefs,
resolvers as myResolvers,
} from "./hello-world";
import { IExecutableSchemaDefinition } from "@graphql-tools/schema";
const api = new GraphQLAPI({
typeDefs: mergeTypeDefs([
myTypeDefs
]),
resolvers: mergeResolvers([
myResolvers
]) as IExecutableSchemaDefinition<ApolloContext>["resolvers"],
});
export const handler = api.createLambdaHandler();
When your GraphQLAPIService
depends on a CognitoService
, the information from the provided access token will be attached to ctx.user
and ctx.accessToken
inside of your resolver functions.