fabo serverless framework

Build scalable serverless APIs in minutes, not days.

Easy backend serverless yaml based REST API generator, which aims to replace a lot of the boilerplate which goes along with creating jamstack mobile/web apps.

• This project is in ALPHA, documentation is a work in progress, caveat emptor

Introduction

fabo is a small serverless framework for auto generating rest api's on top of a yaml configurations. It was made out of necessity of reducing boilerplate and being able to provide sensible defaults for scaffolding serverless + mongodb projects.

feedback @publicbuilds(which is built with fabo): https://www.publicbuilds.com/builds/fpaboim/FABOjs

Features

• Serverless
• Easy to eject: it's just serverless + express + mongodb
• Automatic schema generation based on yaml interface
• Automatic API generation based on yaml interface
• Easy query building and validation
• Easy client side and server-side schema validation
• Commandline client for scaffolding

Instructions

• Install the latest version of mongodb 4.4+ if not already installed
• Install serverless and login for deploying the serverless backend npm install -g serverless or with pnpm pnpm add -g serverless
• Install fabo npm install -g fabo or with pnpm pnpm add -g fabo
• To create a project in the current directory fabo init or to make a new directory with your project files fabo init newdirectory

Guide

Structure

Your schema definition and methods relating to the schema reside in the ./models folder. Creating a folder in the models directory automatically creates a route with the lowercased folder name and corresponding routes. The folder can have the following files which auto-generate a schema and API:

• schema.yaml
• schemaHooks.yaml
• api.yaml
• methods.js

Examples:

Schema (/models/modelroute/schema.yaml)

Example schema:

imports:
  uniqid: uniqid
schema:
  username:
    type     : String
    unique   : false
    required: [true, Username is required.]
    validations:
      - validator: isLength
        arguments: [3, 50]
        message: Name should be between {ARGS[0]} and {ARGS[1]} characters
      - validator: isAlphanumeric
        passIfEmpty: true
        message: Name should contain alpha-numeric characters only
  email   :
    type    : String
    unique  : true
    lowercase: true
    trim    : true
    required: [true, 'Email is required.']
    validations:
      - validator: isEmail
        message: Please enter a valid email
      - validator: isLength
        only: server # one of ['client', 'server']
        arguments: [4, 100]
        message: Email should be between {ARGS[0]} and {ARGS[1]} characters
  password:
    type    : String
    trim    : true
    required: [true, 'Password is required.']
    validations:
      - validator: isLength
        arguments: [8, 40]
        message: Password should be between {ARGS[0]} and {ARGS[1]} characters
  imagepath:
    type: String
    default: uniqid()
  joined:
    type: Date
    default: Date.now
  verified  :
    type: Boolean
    default: false
  roles  :
    type: [String]
    default: [C.ROLES.USER]
  liked:
    type: [Id]
    default: "[]"
    ref: Post
  messages:
    type: [Id]
    default: "[]"
    ref: Message

Schema Hooks (/models/modelroute/schemaHooks.js)

Example pre save mongoose hook for salting and hashing a password before saving:

import bcrypt   from 'bcryptjs'

const hooks = {
  pre: {
    save: function (next) {
      if (this.isModified('imagepath')) {
        return next()
      }

      // only run this if we're messing with the password field, or else bcrypt
      // will on all saves!
      if (!this.isModified('password')) {
        return next()
      }

      bcrypt.genSalt(10, (err, salt) => {
        if (err) {
          console.log('ERR:', err)
          return next(err)
        }
        bcrypt.hash(this.password, salt, (err, hash) => {
          if (err) {
            console.log('BCRYPT ERR:', err)
            return next(err)
          }
          this.password = hash
          // console.log('newpass', this.password)
          next()
        })
      })
    }
  }
}

export default hooks

API (/models/modelroute/api.yaml)

Example API:

endpoints:
  count: true
  delete:
    auth:
      - role: C.ROLES.ADMIN
  find:
    query:
      filter:
        allow: [build]
        required: [build]
  findone: true
  create:
    login: true
    pre:
      allowFields: [body, build]
      setFields:
        - author: user.username
        - edited: Date.now()
        - created: Date.now()
  updateone:
    login: true
    auth:
      - role: C.ROLES.USER
      - role: C.ROLES.EDITOR
    query:
      filter:
        setFields:
          - field: author
            value: user.username
    pre:
      allowFields: [body]
      setFields:
        - edited: Date.now()
query:
  filter:
    required: [build]
  skip: true
  limit:
    default: 25
    max: 25
    min: 1
  sort:
    default: -created
  fields: false
  populate: false

Example custom API:

endpoints:
  auth/login:
    method: signinUser
    middlewares: [apiLimiter]
  auth/signup:
    method: signupUser
    middlewares: [apiLimiter]
  auth/refreshtoken:
    method: refreshToken
    middlewares: [apiLimiter]
  auth/verify:
    method: verifyEmail
    middlewares: [apiLimiter]
  profile:
    method: getCurrentUser
    auth:
      - role: C.ROLES.USER
  auth/get:
    method: getUser
    middlewares: [apiLimiter]

Extending API Methods (/models/modelroute/methods.yaml)

The canonical User model uses custom methods which are mapped to routes in the example above.

import User from "./";
import jwt from 'jsonwebtoken'
import bcrypt from 'bcryptjs'

const createToken = (user, secret, expiresIn='2d') => {
  return jwt.sign({ email: user.email, _id: user._id }, secret, { expiresIn })
}


const methods = {
  getCurrentUser: async (req, res, next) => {
    const user = req.user
    if (!user) {
      return null
    } else {
      // const user = await User.findOne({email: user.email}, {password: false, favorites: false})
      return res.send({user})
    }
  },

  verifyEmail: async (req, res, next) => {
    const user = await User.findById(req.user.id, '-password').lean()
    if (!user) {
      return res.status(401).send({errors: {email: {message: 'Error refreshing token.'}}})
    }
  },


  refreshToken: async (req, res, next) => {
    const user = await User.findById(req.user.id, '-password').lean()
    if (!user) {
      return res.status(401).send({errors: {email: {message: 'Error refreshing token.'}}})
    }

    return res.status(200).send({ token: createToken(user, process.env.SECRET) })
  },

  signinUser: async (req, res, next) => {
    const {email, password} = req.body
    // console.log('signing in', email)
    let user = await User.findOne({email}).lean()
    // console.log('signing in user:', user)
    if (!user) {
      return res.status(401).send({errors: {email: {message: 'Email not found.'}}})
    }
    const isValidPassword = await bcrypt.compare(password, user.password)
    if (!isValidPassword) {
      return res.status(401).send({errors: {password: {message: 'Invalid password.'}}})
    }

    delete user.password
    // console.log('token for:', user)

    const token = createToken(user, process.env.SECRET)

    return res.status(200).send({...user, token})
  },

  signupUser: async (req, res, next) => {
    try {
      console.log('signup')
      // console.log('USER:', email, password)
      const {username, email, password} = req.body

      const user = await User.findOne({ email }).lean()
      const user2 = await User.findOne({ username }).lean()

      if (user) {
        return res.status(400).send({errors: {email: {message: 'Email already registered.'}}})
      }

      if (user2) {
        return res.status(400).send({errors: {username: {message: 'Username already exists.'}}})
      }

      let newUser = await new User({
        username,
        email,
        password
      }).save()
      newUser=newUser.toObject()
      const token = createToken(newUser, process.env.SECRET)

      return res.status(200).send({...newUser, token})
    } catch(err) {
      console.log('err:', err)
      return res.status(400).json(err)
    }
  }
};

export default methods

Querying the API

fabo uses the awesome api-query-params to parse query strings and assist in build your mongodb queries with ease.

Example query:

const res = await post('/build/find?status=sent&timestamp>2016-01-01&author.firstName=/john/i&limit=100&skip=50&sort=-timestamp&populate=logs&fields=id,logs.ip')

Supported query features by api-query-params

Filtering operators

MongoDB	URI	Example	Result
$eq	key=val	type=public	{filter: {type: 'public'}}
$gt	key>val	count>5	{filter: {count: {$gt: 5}}}
$gte	key>=val	rating>=9.5	{filter: {rating: {$gte: 9.5}}}
$lt	key<val	createdAt<2016-01-01	{filter: {createdAt: {$lt: Fri Jan 01 2016 01:00:00 GMT+0100 (CET)}}}
$lte	key<=val	score<=-5	{filter: {score: {$lte: -5}}}
$ne	key!=val	status!=success	{filter: {status: {$ne: 'success'}}}
$in	key=val1,val2	country=GB,US	{filter: {country: {$in: ['GB', 'US']}}}
$nin	key!=val1,val2	lang!=fr,en	{filter: {lang: {$nin: ['fr', 'en']}}}
$exists	key	phone	{filter: {phone: {$exists: true}}}
$exists	!key	!email	{filter: {email: {$exists: false}}}
$regex	key=/value/<opts>	email=/@gmail\.com$/i	{filter: {email: /@gmail.com$/i}}
$regex	key!=/value/<opts>	phone!=/^06/	{filter: {phone: { $not: /^06/}}}

For more advanced usage ($or, $type, $elemMatch, etc.), pass any MongoDB query filter object as JSON string in the filter query parameter, ie:

'filter={"$or":[{"key1":"value1"},{"key2":"value2"}]}'
//  {
//    filter: {
//      $or: [
//        { key1: 'value1' },
//        { key2: 'value2' }
//      ]
//    },
//  }

Skip / Limit operators

• Useful to limit the number of records returned.
• Default operator keys are skip and limit.

'skip=5&limit=10'
//  {
//    skip: 5,
//    limit: 10
//  }

Projection operator

• Useful to limit fields to return in each records.
• Default operator key is fields.
• It accepts a comma-separated list of fields. Default behavior is to specify fields to return. Use - prefixes to return all fields except some specific fields.
• Due to a MongoDB limitation, you cannot combine inclusion and exclusion semantics in a single projection with the exception of the _id field.
• It also accepts JSON string to use more powerful projection operators ($, $elemMatch or $slice)

'fields=id,url'
//  {
//    projection: { id: 1, url: 1}
//  }

'fields=-_id,-email'
//  {
//    projection: { _id: 0, email: 0 }
//  }

'fields={"comments":{"$slice":[20,10]}}'
//  {
//    projection: { comments: { $slice: [ 20, 10 ] } }
//  }

Sort operator

• Useful to sort returned records.
• Default operator key is sort.
• It accepts a comma-separated list of fields. Default behavior is to sort in ascending order. Use - prefixes to sort in descending order.

'sort=-points,createdAt'
//  {
//    sort: { points: -1, createdAt: 1 }
//  }

Population operator

• Useful to populate (reference documents in other collections) returned records. This is a mongoose-only feature.
• Default operator key is populate.
• It accepts a comma-separated list of fields.
• It extracts projection on populated documents from the projection object.

'populate=a,b&fields=foo,bar,a.baz'
// {
//    population: [ { path: 'a', select: { baz: 1 } } ],
//    projection: { foo: 1, bar: 1 },
//  }

Keys with multiple values

Any operators which process a list of fields ($in, $nin, sort and projection) can accept a comma-separated string or multiple pairs of key/value:

• country=GB,US is equivalent to country=GB&country=US
• sort=-createdAt,lastName is equivalent to sort=-createdAt&sort=lastName

Embedded documents using . notation

Any operators can be applied on deep properties using . notation:

'followers[0].id=123&sort=-metadata.created_at'
//  {
//    filter: {
//      'followers[0].id': 123,
//    },
//    sort: { 'metadata.created_at': -1 }
//  }

Automatic type casting

The following types are automatically casted: Number, RegExp, Date and Boolean. null string is also casted:

'date=2016-01-01&boolean=true&integer=10&regexp=/foobar/i&null=null'
// {
//   filter: {
//     date: Fri Jan 01 2016 01:00:00 GMT+0100 (CET),
//     boolean: true,
//     integer: 10,
//     regexp: /foobar/i,
//     null: null
//   }
// }

If you need to disable or force type casting, you can wrap the values with string(), date() built-in casters or by specifying your own custom functions (See below):

'key1=string(10)&key2=date(2016)&key3=string(null)'
// {
//   filter: {
//     key1: '10',
//     key2: Fri Jan 01 2016 01:00:00 GMT+0100 (CET),
//     key3: 'null'
//   }
// }

Available options (opts)

Customize operator keys

The following options are useful to change the operator default keys:

• skipKey: custom skip operator key (default is skip)
• limitKey: custom limit operator key (default is limit)
• projectionKey: custom projection operator key (default is fields)
• sortKey: custom sort operator key (default is sort)
• filterKey: custom filter operator key (default is filter)
• populationKey: custom populate operator key (default is populate)

'organizationId=123&offset=10&max=125',
  limitKey: 'max',
  skipKey: 'offset',
});
// {
//   filter: {
//     organizationId: 123,
//   },
//   skip: 10,
//   limit: 125
// }

Blacklist / Whitelist

The following options are useful to specify which keys to use in the filter object. (ie, avoid that authentication parameter like apiKey ends up in a mongoDB query). All operator keys are (sort, limit, etc.) already ignored.

• blacklist: filter on all keys except the ones specified
• whitelist: filter only on the keys specified

'id=e9117e5c-c405-489b-9c12-d9f398c7a112&apiKey=foobar',
  blacklist: ['apiKey'],
});
// {
//   filter: {
//     id: 'e9117e5c-c405-489b-9c12-d9f398c7a112',
//   }
// }

Add custom casting functions

You can specify you own casting functions to apply to query parameter values, either by explicitly wrapping the value in URL with your custom function name (See example below) or by implictly mapping a key to a function (See Specify casting per param keys below). Note that you can also override built-in casting functions: boolean, date ,null ,number ,regex and string.

• casters: object to specify custom casters, key is the caster name, and value is a function which is passed the query parameter value as parameter.

'key1=lowercase(VALUE)&key2=int(10.5)&key3=true',
  casters: {
    lowercase: val => val.toLowerCase(),
    int: val => parseInt(val, 10),
    boolean: val => (val === 'true' ? '1' : '0'),
  },
});
// {
//   filter: {
//     key1: 'value',
//     key2: 10,
//     key3: '1',
//   }
// }

Specify casting per param keys

You can specify how query parameter values are casted by passing an object.

• castParams: object which map keys to casters (built-in or custom ones using the casters option).

'key1=VALUE&key2=10.5&key3=20&key4=foo',
  casters: {
    lowercase: val => val.toLowerCase(),
    int: val => parseInt(val, 10),
  },
  castParams: {
    key1: 'lowercase',
    key2: 'int',
    key3: 'string',
    key4: 'unknown',
  },
});
// {
//   filter: {
//     key1: 'value',
//     key2: 10,
//     key3: '20',
//     key4: 'foo',
//   }
// }

Config for S3 Uploads

• Install aws cli client here
• Create S3 bucket for storing files:

  aws s3api create-bucket --bucket my-bucket --region us-west-2 --create-bucket-configuration LocationConstraint=us-west-2
  

• Configure bucket:

  aws s3api put-bucket-cors --bucket my-bucket --cors-configuration '{"CORSRules": [{
  "AllowedOrigins": ["*"],
  "AllowedHeaders": ["Authorization"],
  "AllowedMethods": ["GET", "POST"],
  "MaxAgeSeconds": 3000
  }]}'
  

• pnpm i
• pnpm run dev

Configuring Cache w/ S3 and cloudfront

https://aws.amazon.com/blogs/networking-and-content-delivery/amazon-s3-amazon-cloudfront-a-match-made-in-the-cloud/

Constants (models/constants.js)

Defines constants which your API and schema have access to, like roles and enums.

Client

fabo currently was built with svelte kit in mind, but a general schema to integrate with any client side framework in on the roadmap.

Securing your server

• Per ip rate limiting with AWS WAF: https://docs.aws.amazon.com/waf/latest/developerguide/how-aws-waf-works.html

WIP

Admin

• View db fields
• Delete db entries
• Update db fields

Todo

planned

• Admin
• Register KPIs on admin
• Easier image uploads
• fabo config file for configuring different client frameworks
• Page for the docs
• Validate query parameters
• Alias endpoints with different config
• Global query limits config
• Tests

maybe?

• Svelte autoforms
• Make schema requirements specific to server or client
• Upload generic media