import factory from '@hippiemedia/agent';

agent = factory();
let resource = await agent.follow('http://haltalk.herokuapp.com/');

An hypermedia agent capable of discovering and navigating a web of linked resources.

npm install @hippiemedia/agent

npm install
npm run build

Then you can create an agent. It's like your browser, but it understands more hypermedia formats:

import factory from '@hippiemedia/agent';

agent = factory();

• html [ ]
• ld+json (+hydra) [ ]
• siren+json [ ]
• hal+json [x]
• hal-forms [x]
• vnd.api+json [ ]
• collection+json [ ]
• uber+json [ ]
• your own adapter...

This agent is capable of fetching a resource, parsing its content based on its content type,
and populating a Resource object with operations, links and queries.

It's important to let the agent discover links instead of hardcoding them.
You can do that by introspecting resources:

let menu = resource.links.map(link => render(`<a href="${link.href}">${link.rel}</a>`));

let actions = resource.operations.map(o => render(`<form method="${o.method}" action="${o.href}">
    ${o.fields.map(field => render(`<input type=${field.type} />`))}
</form>`));

You can apply some operations on a resource, like POSTing a new resource to a collection, or DELETE a resource.

let resource = await agent.follow('http://some-hypermedia-api.com/resource1');
let newResource = await resource.follow('comment').then(r => {
    r.operation('default').fill({
        author: 'me',
        text: 'fuck yeah'
    }).submit()
});

You can follow links as you follow them in your browser.

let resource = await agent.follow('http://some-hypermedia-api.com/resource1');
console.log(resource.links);
let comments = await resource.followAll('comments');

You can follow templated links as you follow links:

let resource = await agent.follow('http://some-hypermedia-api.com/resource1');
let comments = await resource.follow('user', {id: '1'});

The current state of http APIs sucks. Everybody makes its own shit in its own corner
without embracing what made the web so successful (hateoas, links, auto-discovery, ...)

json-ld + hydra is the future, yet everybody tries to replace poor REST with graphql,
thinking that graphql replaces REST (while it's only complementary at best),
thinking that client-side-hardcoded-server-structure is better than auto-discovered, self-documented, dumb[1] clients,
thinking that json is better that xml,
thinking that raw-data structures worth anything without its semantics,
thinking that hardcoding URLs client-side and API versioning is better than letting the server do the job.
/rant

[1] dumb clients are actually smart: they know as much about your API than firefox knows about just another mother fucking website.
Yet this very same browser is capable of displaying it and any other website. Now who's smart?