auth0/slack-robot

Name: slack-robot

Owner: Auth0

Description: Simple robot for your slack integration

Forked from: traveloka/slack-robot

Created: 2016-05-10 00:50:25.0

Updated: 2017-10-20 02:52:27.0

Pushed: 2017-08-15 18:07:31.0

Homepage:

Size: 195

Language: JavaScript

GitHub Committers

UserMost Recent Commit# Commits

Other Committers

UserEmailMost Recent Commit# Commits

README

slack-robot Build Status codecov.io Package Version

Simple robot for your slack integration

slack-robot initially built to respond message from user with an action, then replying with a response. While hubot will do the job, it's too generic (no slack specific response like reaction, snippet, attachment, or listening to slack specific event). Using hubot also mean learning hubot specific command, using the right adapter, using the right slack-client, and patching them with slack specific event

While modularity is good, we're not sure about using multiple module with exponentially larger bug possibility (and also broader learning material). Because we are using it in critical application, we need to make sure that the API surface is as little as possible, with near perfect test coverage. With that in mind, we built slack-robot by wrapping “official” slack-client into much easier to consume request-response handling library.

Features
Installation
m install slack-robot --save
Usage example
SlackRobot = require('slack-robot');
robot = new SlackRobot(process.env.SLACK_TOKEN);

ill post 'world' text as bot when receiving 'hello' message
n channel, group, or direct message
t.listen('hello', function (req, res) {
turn res.text('world').send();


gnore message from '#general' channel, even if it matches the listener
t.ignore('#general');

tart listening
t.start();
Parameterized message

You can also listen to dynamic message by using parameterized message (usually called named-regexp), by using :name(REGEXP) syntaxes. All parameters will be available via req.params

end 'get sheep from 2010' to your bot
t.listen('get :animal([a-z\-]+) from :year([0-9]{4})', function (req, res) {
nsole.log(req.params)
 { animal: 'sheep', year: 2010 }

Pure regular expression (regexp) listener

Aside from named regexp (which uses a partial regexp match), you can use pure regexp inside your listener. The difference is, instead of getting an object in req.params, you get Array of regexp matches in req.matches

end 'get sheep from 2010' to your bot
t.listen(/get ([a-z]+) from ([0-9]{4})/, function (req, res) {
 you don't have anything in named-param
nsole.log(req.params)
 {}

 you use req.matches instead
nsole.log(req.matches);
 ['sheep', '2010']

Request & response life cycle

In slack-robot, receiving and sending message is handled via Request and Response object. Every time your bot receive a chat, you get Request object with typedef below

 Request = {
ssage: {
type: string
value: {}

om: {
id: string,
name: string

: {
id: string,
type: string, (channel, group, or dm)
?name: string // missing if direct message

rams: {},
tches: []

For example when you @anonymous send the bot @hola message get sheep from 2010 @hola in #general channel

t.listen('get :animal([a-z\-]+) from :year([0-9]{4})', function (req, res) {
nsole.log(req);
 message: {
   type: 'text'
   value: {
     text: 'get sheep from 2010'
     mentioned: true
   }
 },
 from: {
   id: 'your_random_id',
   name: 'anonymous'
 },
 to: {
   id: 'random_channel_id',
   name: 'general',
   type: 'channel'
 },
 params: {
   animal: 'sheep',
   year: '2010',
 },
 matches: []

To respond a message, use res object. You can respond multiple times as you want

t.listen('yo', function (req, res) {
*
 Send text

 @param {string} text
/
s.text('what\'s up?');

*
 Send attachment

 @param {string} text in attachment
 @param {Array<Object>|Object}
 @see https://api.slack.com/docs/attachments
/
s.attachment(text, attachment);

*
 Send file

 @param {string} filename
 @param {string|ReadStream} content
 @see https://nodejs.org/api/fs.html
/
s.upload('snippet.txt', fs.createReadStream('snippet.txt'));

*
 Add reaction to the message

 @param {string} reaction emoji
/
s.reaction(':+1:');

*
 Always end your handler by returning res.send
/
turn res.send();

Until you call res.send(), your message will not be sent. By calling res.send(), it will queue all your response and send them in series. To change this behavior, change concurrency property from robot:

end 3 response in parallel (this will affect all listener)
eep in mind that enabling concurrency means the order of the
essage is not guaranteed
t.set('concurrency', 3);
Custom response target

To respond in another channel/im, simply pass the last argument to .text(), .attachment() or .upload(), with channel name #general, group name private-group, or user name @anon, or use an array of string to send multiple target:

t.listen('yo', function (req, res) {
s.attachment('here', attachment, '#general');
s.upload('document.doc', file, '@anon');
s.text('done!', ['#general', '@anon']);

 end your request
turn res.send();

NOTE: You cannot use custom response target when adding reaction

Asynchronous response

Sometimes you want to do some asynchronous task before sending back response, you can use res.async() which accept a callback that receive send function as argument. To end your asynchronous task call send() without any argument. If your asynchronous task failed, call send() with an error object:

t.listen('deploy', function (req, res) {
s.text('executing scripts..');

turn res.async(function (done) {
childProcess.exec('~/scripts/deploy --to production', function (err, stdout, stderr) {
  if (err) {
    // return to stop code for reaching res.text
    return done(err);
  }

  // use res.text like usual
  res.text('done, printing stdout:');
  res.text(stdout);      
  done();
});
 call .send() to send all previous response declared in asynchronous task
.send();

NOTE: Calling res.send() after res.async() doesn't send all the response, because res.send() is synchronous. Make sure you call res.async().send() to send the response

If you already use Promise, you can return your Promise chain instead of using res.async

s2015 code style
t.listen('deploy', (req, res) => {
turn deployer().then(output => {
res.text('done, printing output:');
res.text(output);
return res.send();


External trigger

You can also send message without having to listen to any message. This is particularly useful when combined with another service that run asynchronously (for example error reporting). Use robot.to() to get the response object you usually use when responding message

ws = require('websocket');

n('message', function (msg) {
 robot.to() is asynchronous by nature because we need to make sure
 the bot is connected before you able to send the message
 hence the use of the callback to get the response object
bot.to('@anon', function (res) {
res.text('Hi anon, you got a message');
res.text(msg);
return res.send();
;

NOTE: .reaction() and .async() cannot be used here

Handling the unexpected

slack-robot will emit event if something happened. Below is the list of events you can listen to:

To listen specific event, use robot.on(message, callback). Most event will receive message object, except response_failed and error event which receive error object instead, request_handled which receive request object, and message_no_sender which receive original message object from slack API

t.on('error', function (err) {
 print to stderr, or sent to error reporting service
nsole.error(err);

Help command generator

When you have many listener, you sometimes forget all your listeners. You can see it by enabling help generator which will sent you all your listeners. Enable it using robot.set('help_generator', true) (it's disabled by default). It will add another listener that will listen to all text message containing “help”. So if you send message to the bot “show help” or “help”, it will send you the command list.

License

MIT


This work is supported by the National Institutes of Health's National Center for Advancing Translational Sciences, Grant Number U24TR002306. This work is solely the responsibility of the creators and does not necessarily represent the official views of the National Institutes of Health.