Express Toolkit
Tiny little utilities for reducing expressjs boilerplate code when building simple, mongodb backed, http apis
TL;DR
With express-toolkit you can easily create a basic REST resource and mount it into an express application. The app will provide basic CRUD methods:
const express = require('express')
const Resource = require('../../src/resource')
const mongoose = require('mongoose')
// Let's create our Model with Mongoose
const schema = new mongoose.Schema({
name: String
})
const PetsResource = new Resource({
name: 'pets',
id: 'uuid',
model: mongoose.model('pets', schema, 'pets')
})
PetsResource.registerHook('pre:find', (req, res, next) => {
console.log('Looking for Pets')
next()
})
// Remember to extend the router AFTER adding hooks,
// otherwise the router will be overwritten without this route
PetsResource.router.get('/actions/eat',(req,res) => {
return res.send('Om nom nom')
})
// Now the Express related stuff
const app = express()
const port = 3000
app.get('/', (req, res) => res.send('Hello World!'))
// Here we mount the Pets resource under the /pets path
PetsResource.mount('/pets', app)
// After mongoose is ready, we start listening on the TCP port
mongoose.connect('mongodb://localhost:27017/pets', {})
.then(() => {
console.log('Connection to Mongodb Established')
app.listen(port, () => console.log(`Example app listening on port ${port}!`))
})
.catch(error => {
console.log('Unable to establish connection to Mongodb', error)
})Under the hood, the resource object uses three other objects: Model, Controller and Router.
- Routers are plain express routers to which the library mounts some default REST routes
- Controllers implement CRUD methods
- Models define a mongoose model
When you create a resource object, the library will create a model a controller and a router for you, if you need to add custom logic to those components you can retrieve them as properties of the resource:
// Let's create our Model with Mongoose
const schema = new mongoose.Schema({
name: String
})
const PetsResource = new Resource({
name: 'pets',
id: 'uuid',
model: mongoose.model('pets', schema, 'pets')
})
// Let's add a hook to the controller
const ctrl = PetsResource.controller
// more on this method later in this document
ctrl.registerHook('post:create',(req,res,next) => {
console.log("Hello I created a resource")
next()
})
// Let's add a custom route to the router
const router = PetsResource.router
router.get('/hello/world',(req,res,next) => {
res.send("Hello")
})Models, Controllers and Routers
Suppose we need to build an http microservice for handling dinosaurs (tired of cats).
First of all we will need a model file, powered by mongoose
const mongoose = require('mongoose')
const Schema = mongoose.Schema
// Dinosaurs are simple
const DinosaurSchema = new Schema({
name: {
type: String,
required:true
},
type: {
type: String,
required: true
}
})
const DinosaurModel = mongoose.model('Dinosaur', DinosaurSchema, 'dinosaurs')
module.exports = {DinosaurSchema, DinosaurModel}Then the controller file
const { Controller } = require('express-toolkit')
const { DinosaurModel } = require('./path/to/dinosaur.model.js')
const myController = new Controller({
model: DinosaurModel,
name: 'dinosaurs'
})
module.exports = myControllerFinally the router file
const { buildRouter } = require('express-toolkit')
const DinosaurController = require('./dinosaur.controller.js')
module.exports = buildRouter({
controller: DinosaurController,
options: {} // See expressJS router options
})Then, somewehere in your express app, where you mount routes:
const express = require('express')
const app = express()
const dinosaursResource = require('./path/to/dinosaur.router.js')
//...
app.use('/dinosaurs',dinosaursResource)
// ...
app.listen(1337)Default endpoints
In the following table, every path showed in the Path column is meant to be appended to the resource base path which simply is /<resourcename>. Following the dinosaurs examples, would be /dinosaurs
| Name | Http verb | Path | Description |
|---|---|---|---|
| Create | POST | / | Creates a new resource and returns it. |
| List | GET | / | Get a paginated and filtered list of resources of the given type |
| GetById | GET | /{uuid} | Get a resource by id |
| UpdateById | PUT | /{uuid} | Updates a resource |
| UpdateByQuery | PUT | / | Updates resources that matches query parameters |
| PatchById | PATCH | /{uuid} | Updates a resource by id using PATCH semantics |
| ReplaceById | PUT | /{uuid}/replace | Replaces a resource by id. Primary id field (and _id if not the same) are not replaced |
| DeleteById | DELETE | /{uuid} | Deletes a resource |
| DeleteByQuery | DELETE | / | Deletes resources matching filters in the querystring |
| Count | GET | / | Count resources in collection matching filters in the querysting |
Disable endpoints
By default, all endpoints are enabled, to control which endpoints should be disabled, you can use the endpoints router parameter
// Default behaviour, endpoints is an optional parameter
const router = buildRouter({
controller: require('./mycontroller.js'),
endpoints: {
find: true,
findById: true,
create: true,
updateById: true,
updateByQuery: true,
deleteById: true,
deleteByQuery: true,
count: true,
patchById: true,
replaceById: true
}
})
// Default resource deletion
const router = buildRouter({
controller: require('./mycontroller.js'),
endpoints: {
deleteById: true,
deleteByQuery: true
}
})Sorting
GET endpoints support result sorting thanks to two query string parameters:
-
sortorder <String>that can only have two values:DESCfor descending sorting andASCfor ascending order. -
sortby <String>can be used to select the sorting parameter.
For example, the following request would sort dinosaurs by age, oldest to youngest:
GET /dinosaurs?sortby=age&sortorder=DESCPagination
GET endpoints support result pagination through skip and limit parameters:
-
skip <Number>tells the endpoint how many results to skip -
limit <Numbertells the endpoint how many results to include in the response
To implement a pagination scheme, you can leverage these two parameters in the following way: Suppose you want to return R results per page and you want to return page number P, you just need to set limit to R and skip to (P-1)*R
Projection
Sometimes you don't need the whole resource object but just some of its attributes, in these cases you can use the fields query string parameter.
Suppose the dinosaur resource has name, type and age attributes, but we just want names and age:
GET /dinosaurs?fields=name,ageOr just names
GET /dinosaurs?fields=nameOr every field but the age and the name
GET /dinosaurs?fields=-age,-nameIf you don't specify a fields parameter, every attribute will be returned.
Custom primary key
By defaults, resources are handled as if their primary key is the _id field, which is automatically added by mongodb. Sometimes you might want to provide your own key such as an uuid field added to the model. For such cases you can provide the id attribute to the controller's config:
const myController = new Controller({
model: MyModel,
name: 'dinosaurs',
id: 'uuid'
})Hooks
Every resource endpoint can have multiple pre and post hooks. These hooks will be run by the router before and after the related controller method.
Typically, in pre hooks you will want to manually edit requests or do some kind of prior validation on the request, while on post hooks you would fetch/add more data or generate other actions such as logging business logic events.
List of hooks
- pre:count
- post:count
- pre:find
- post:find
- pre:findById
- post:findById
- pre:create
- post:create
- pre:updateById
- post:updateById
- pre:updateByQuery
- post:updateByQuery
- pre:deleteById
- post:deleteById
- pre:deleteByQuery
- post:deleteByQuery
- pre:patchById
- post:patchById
- pre:replaceById
- post:replaceById
- pre:*
- post:*
- pre:finalize
pre:finalize is called on every endpoint, just before sending the response payload to the client.
Here you can hijack req.toSend and update it as you need.
pre:* if defined, is called on every endpoint of that resource before any other "pre" hook, in the same way post:* is called after any other post hook. For every endpoint the order is:
pre:*pre:<methodName>middlewarepost:<methodname>post:*pre:finalizefinalize
For example, you might want to check the Accept HTTP header and convert the response from JSON to YAML, or XML.
Examples
const { Controller } = require('express-toolkit')
const { DinosaurModel } = require('./path/to/dinosaur.model.js')
const myController = new Controller({
model: DinosaurModel,
name: 'dinosaurs'
})
// Check authorization on all dinosaurs routes:
myController.registerHook('pre:*', (req,res,next) => {
//This is just an example, a bad auth example.
if (req.headers.authorization !== "supersecret") {
return res.sendStatus(401)
}
next()
})
// Force all find queries to look for velociraptor type
myController.registerHook('pre:find', (req,res,next) => {
req.query.type = 'velociraptor'
next()
})
// Before returning dinosaurs to the client we convert timestamps to date strings
myController.registerHook('post:find', (req,res,next) => {
req.toSend = req.toSend.map(dinosaur => {
let dino = Object.assign({},dinosaur)
dino.createdAt = String(new Date(dino.createdAt))
return dino
})
next()
})
module.exports = myController
![dependabot[bot]](https://avatars.githubusercontent.com/u/49699333?size=120)
