halacious for HAPI

a better HAL processor for Hapi

Node.js and HAPI Versions

For HAPI v17+ use current release branch (6.x). Note this also requires Node.js v8.10+.

For HAPI v16 use legacy release branch (5.x).

Overview

Halacious is a plugin for the HapiJS web application server that makes HATEOASIFYING your app ridiculously easy. When paired with a well-aged HAL client-side library, you will feel the warmth of loose API coupling and the feeling of moral superiorty as you rid your SPA of hard-coded api links.

Halacious removes the boilerplate standing between you and a Restful application, allowing you to focus on your app's secret sauce. Halacious embraces Hapi's configuration-centric approach to application scaffolding. Most common tasks can be accomplished without writing any code at all.

Features

• Dead-simple namespace/rel registration system that encourages you to document your API as you go
• Automatic api root REST endpoint generation that instantly gives you links to all top-level API endpoints
• Automatic rel documentation site generator so that your fully resolved rel names actually, you know, point somewhere.
• Automatic creation of curie links
• Support for relative and templated link hrefs.
• Auto wrapping of http response entities into HAL representations
• Support for custom object json serialization
• Support for programmatic configuration of HAL entities at the route or entity level
• Bunches of unit tests

Getting Started

Start by npm installing the halacious library into your hapi project folder:

npm install halacious --save

or

yarn add halacious --dev

Register the plugin with the app server

const hapi = require('hapi');
const halacious = require('halacious');

const server = new hapi.Server({ port: 8080 });
server.register(halacious).catch(err => {
  console.error(err);
});

server.route({
    method: 'get',
    path: '/hello/{name}',
    handler: function(req) {
      return({ message: 'Hello, '+req.params.name });
    }
});

server.start().then(() => {
  console.info('Server started at %s', server.info.uri);
}).catch(err => {
  console.error(err);
});

Launch the server:

node ./examples/hello-world

Make a request

curl -H 'Accept: application/hal+json' http://localhost:8080/hello/world

See the response

{
    "_links": {
        "self": {
            "href": "/hello/world"
        }
    },
    "message": "Hello, world"
}

Linking

Links may be declared directly within the route config.

server.route({
    method: 'get',
    path: '/users/{userId}',
    config: {
        handler: function (req, reply) {
            // look up user
            reply({ id: req.params.userId, name: 'User ' + req.params.userId, googlePlusId: '107835557095464780852' });
        },
        plugins: {
            hal: {
                links: {
                    'home': 'http://plus.google.com/{googlePlusId}'
                },
                ignore: 'googlePlusId' // remove the id property from the response
            }
        }
    }
});

curl -H 'Accept: application/hal+json' http://localhost:8080/users/100

will produce:

{
    "_links": {
        "self": {
            "href": "/users/1234"
        },
        "home": {
            "href": "http://plus.google.com/107835557095464780852"
        }
    },
    "id": "100",
    "name": "User 1234"
}

Embedding

HAL allows you to conserve bandwidth by optionally embedding link payloads in the original request. Halacious will automatically convert nested objects into embedded HAL representations (if you ask nicely).

server.route({
    method: 'get',
    path: '/users/{userId}',
    config: {
        handler: function (req, reply) {
            // look up user
            reply({
                id: req.params.userId,
                name: 'User ' + req.params.userId,
                boss: {
                    id: 1234,
                    name: 'Boss Man'
                }
            });
        },
        plugins: {
            hal: {
                embedded: {
                    'boss': {
                        path: 'boss', // the property name of the object to embed
                        href: '../{item.id}'
                    }
                }
            }
        }
    }
});

curl -H 'Accept: application/hal+json' http://localhost:8080/users/100

{
    "_links": {
        "self": {
            "href": "/users/100"
        }
    },
    "id": "100",
    "name": "User 100",
    "_embedded": {
        "boss": {
            "_links": {
                "self": {
                    "href": "/users/1234"
                }
            },
            "id": 1234,
            "name": "Boss Man"
        }
    }
}

Programmatic configuration of HAL representations

You may find the need to take the wheel on occasion and directly configure outbound representions. For example, some links may be conditional on potentially asynchronous criteria. Fortunately, Halacious provides two ways to do this:

1. By providing a prepare() function on the route's hal descriptor (or by assigning the function directly to the hal property)
2. By implementing a toHal() method directly on a wrapped entity.

In either case, the method signature is the same: fn(rep, callback) where

• rep - a representation object with the following properties and functions:
  • factory - a factory reference for creating new representations. The factory object implements one method:
    • create(entity, selfHref) - wraps entity with a new Hal representation, whose self link will point to selfHref
  • request - the originating hapi request
  • self - a shortcut to the representation's self link
  • entity - the original wrapped entity
  • prop(name, value) - manually adds a name/value pair to the representation
  • merge(object) - merges the properties of another object into the representation
  • ignore(...propertyNames) - prevents fields from being included in the response
  • link(relName, href) - adds a new link to the _links collection, returning the new link. Link objects support the following properties (See see http://tools.ietf.org/html/draft-kelly-json-hal-06#section-8.2 for more information):
    • href
    • templated
    • title
    • type
    • deprecation
    • name
    • profile
    • hreflang
  • embed(rel, self, entity) - adds an entity to the representation's _embedded collection with the supplied rel link relation and self href, returning a new representation object for further configuration.
• callback([err], [representation]) - an asynchronous callback function to be called when configuration of the hal entity is complete. Most of the time this function should be called with no arguments. Only pass arguments if there has been an error or if a completely new representation has been created with rep.factory.create().

Example 1: A prepare() function declared in the route descriptor.

server.route({
    method: 'get',
    path: '/users',
    config: {
        handler: function (req, reply) {
            // look up user
            reply({
                start: 0,
                count: 2,
                limit: 2,
                items: [
                    { id: 100, firstName: 'Brad', lastName: 'Leupen', googlePlusId: '107835557095464780852'},
                    { id: 101, firstName: 'Mark', lastName: 'Zuckerberg'}
                ]
            });
        },
        plugins: {
            hal: {
                // you can also assign this function directly to the hal property above as a shortcut
                prepare: function (rep, next) {
                    rep.entity.items.forEach(function (item) {
                        var embed = rep.embed('item', './' + item.id, item);
                        if (item.googlePlusId) {
                            embed.link('home', 'http://plus.google.com/' + item.googlePlusId);
                            embed.ignore('googlePlusId');
                        }
                    });
                    rep.ignore('items');
                    // dont forget to call next!
                    next();
                }
            }
        }
    }
});

curl -H 'Accept: application/hal+json' http://localhost:8080/users

{
    "_links": {
        "self": {
            "href": "/users"
        }
    },
    "start": 0,
    "count": 2,
    "limit": 2,
    "_embedded": {
        "item": [
            {
                "_links": {
                    "self": {
                        "href": "/users/100"
                    },
                    "home": {
                        "href": "http://plus.google.com/107835557095464780852"
                    }
                },
                "id": 100,
                "firstName": "Brad",
                "lastName": "Leupen"
            },
            {
                "_links": {
                    "self": {
                        "href": "/users/101"
                    }
                },
                "id": 101,
                "firstName": "Mark",
                "lastName": "Zuckerberg"
            }
        ]
    }
}

Example 2: Implementing toHal() on a domain entity:

function User(id, firstName, lastName, googlePlusId) {
    this.id = id;
    this.firstName = firstName;
    this.lastName = lastName;
    this.googlePlusId = googlePlusId;
}

User.prototype.toHal = function(rep, next) {
    if (this.googlePlusId) {
        rep.link('home', 'http://plus.google.com/' + this.googlePlusId);
        rep.ignore('googlePlusId');
    }
    next();
};

server.route({
    method: 'get',
    path: '/users',
    config: {
        handler: function (req, reply) {
            // look up user
            reply({
                start: 0,
                count: 2,
                limit: 2,
                items: [
                    new User(100, 'Brad', 'Leupen', '107835557095464780852'),
                    new User(101, 'Mark', 'Zuckerberg')
                ]
            });
        },
        plugins: {
            hal: {
                embedded: {
                    item: {
                        path: 'items',
                        href: './{item.id}'
                    }
                }
            }
        }
    }
});

The HAL route configuration object

The config.plugins.hal route configuration object takes the following format:

• A function fn(rep, next) - for purely programmatic control over the representation or
• An object with the following properties:
  • api - an optional top level api rel name to assign to this route. Setting a value will cause this route to be included in the root api resource's _links collection.
  • prepare(rep, next) - an optional prepare function
  • ignore - A String or array of strings containing the names of properties to remove from the output. Can be used to remove reduntant information from the response
  • query - An RFC 6570 compatible query string that should be communicated to your clients. See: http://tools.ietf.org/html/rfc6570. Example: {?q*,start,limit}. These parameters will be included in top level api links. They will also be included in self links if supplied in the request. Query parameters that are not included in the template, such as runtime tokens, will be excluded from the self href.
  • links - An object whose keys are rel names and whose values are href strings or link objects that contain at least an href property. Hrefs may be absolute or relative to the representation's self link. Hrefs may also contain {expression} template expressions, which are resolved against the wrapped entity.
  • embedded An object whose keys are rel names and whose values are configuration objects with:
    • path - a path expression to evaluate against the wrapped entity to derive the object to embed.
    • href - a Function, String or link object that will be used to define the entity's self relation. Like links, embedded href's may also be templated. Unlike links, embedded href templates have access to two state variables:
      • self - the parent entity
      • item - the child entity
    • links
    • embedded (recursively evaluated)
    • prepare(rep, next)
  • absolute - a boolean true/false. if true, hrefs for this representation will include protocal, server, and port. Default: false

Namespaces and Rels

So far, we have not done a real good job in our examples defining our link relations. Unless registered with the IANA, link relations should really be unique URLs that resolve to documentation regarding their semantics. Halacious will happily let you be lazy but its much better if we do things the Right Way.

Manually creating a namespace

Halacious exposes its api to your Hapi server so that you may configure it at runtime like so:

var server = new hapi.Server();
server.connection({ port: 8080 });
var halacious = require('halacious');
server.register(halacious, function(err){
    if (err) return console.log(err);
    var ns = server.plugins.halacious.namespaces.add({ name: 'mycompany', description: 'My Companys namespace', prefix: 'mco'});
    ns.rel({ name: 'users', description: 'a collection of users' });
    ns.rel({ name: 'user', description: 'a single user' });
    ns.rel({ name: 'boss', description: 'a users boss' });
});

server.route({
    method: 'get',
    path: '/users/{userId}',
    config: {
        handler: function (req, reply) {
            // look up user
            reply({ id: req.params.userId, name: 'User ' + req.params.userId, bossId: 200 });
        },
        plugins: {
            hal: {
                links: {
                    'mco:boss': '../{bossId}'
                },
                ignore: 'bossId'
            }
        }
    }
});

Now, when we access the server we see a new type of link in the _links collection, curies. The curies link provides a mechanism to use shorthand rel names while preserving their uniqueness. Without the curie, the 'mco:boss' rel key would be expanded to read /rels/mycompany/boss

curl -H 'Accept: application/hal+json' http://localhost:8080/users/100

{
    "_links": {
        "self": {
            "href": "/users/100"
        },
        "curies": [
            {
                "name": "mco",
                "href": "/rels/mycompany/{rel}",
                "templated": true
            }
        ],
        "mco:boss": {
            "href": "/users/200"
        }
    },
    "id": "100",
    "name": "User 100"
}

Creating a namespace from a folder of documentated rels

In our examples folder, we have created a folder rels/mycompany containing markdown documents for all of the rels in our company's namespace. We can suck all these into the system in one fell swoop:

var server = new hapi.Server();
server.connection({ port: 8080 });
var halacious = require('halacious');
server.register(halacious, function(err){
    if (err) return console.log(err);
    server.plugins.halacious.namespaces.add({ dir: __dirname + '/rels/mycompany', prefix: 'mco' });
});

Ideally these documents should provide your api consumer enough semantic information to navigate your api.

Rels documentation

Halacious includes an (extremely) barebones namespace / rel navigator for users to browse your documentation. The server binds to the /rels path on your server by default.

_Note: Hapi 9 / 10 users must install and configure the vision views plugin to enable this feature.

Automatic /api root

Discoverability is a key tenant of any hypermedia system. HAL requires that only the root API url be known to clients of your application, from which all other urls may be derived via rel names. If you want, Halacious will create this root api route for you automatically. All you need to do is to identify which resources to include by using the api route configuration option. For example:

server.register(halacious, function(err){
    if (err) return console.log(err);
    var ns = server.plugins.halacious.namespaces.add({ name: 'mycompany', description: 'My Companys namespace', prefix: 'mco'});
    ns.rel({ name: 'users', description: 'a collection of users' });
    ns.rel({ name: 'user', description: 'a single user' });
});

server.route({
    method: 'get',
    path: '/users',
    config: {
        handler: function (req, reply) {
            // look up user
            reply({});
        },
        plugins: {
            hal: {
                api: 'mco:users'
            }
        }
    }
});

server.route({
    method: 'get',
    path: '/users/{userId}',
    config: {
        handler: function (req, reply) {
            // look up user
            reply({});
        },
        plugins: {
            hal: {
                api: 'mco:user'
            }
        }
    }
});

will auto-create the following api root:

curl -H 'Accept: application/hal+json' http://localhost:8080/api/

{
    "_links": {
        "self": {
            "href": "/api/"
        },
        "curies": [
            {
                "name": "mco",
                "href": "/rels/mycompany/{rel}",
                "templated": true
            }
        ],
        "mco:users": {
            "href": "/users"
        },
        "mco:user": {
            "href": "/users/{userId}",
            "templated": true
        }
    }
}

Plugin Options

• strict - setting this to true will cause an exception to be thrown when referencing unregistered local rel. Setting this to true will help catch typos during development. Default: false
• relsPath - the route path to the rels documentation root. Default: /rels
• relsAuth - the hapi authentication setting to use for the documentation routes. Default: false
• relsTemplate - a boolean true/false. if true, rels documentation uses the default template. Default: true
• autoApi - setting this to true will automatically create a root api handler to seed your client application. Default: true
• apiPath - the route path to the api root. Default: /api
• apiAuth - the hapi authentication setting to use for the api route. Default: false
• apiServerLabel - when set, Halacious will select for a specific server to route the api root.
• mediaTypes - an array of media types that will trigger the hal processor to modify the response (e.g. ['application/json', 'application/hal+json']). the media types are checked in order. if any match the accept header parameters, then the response will be halified and the media type of the response will be set to the first winner. Default: ['application/hal+json']
• absolute - a boolean true/false. if true, all hrefs will include the protocol, server, and port. Default: false
• host - a hostname+port string for all absolute link urls
• hostname - a string hostname for all absolute link urls (only used if host is blank)
• port - an integer port for all absolute link urls
• protocol - a string protocol for all absolute link urls
• marked - a object of options to customize marked. Default: {}