R Job Notifications Using Twitter

[This article was first published on Deciphering life: One bit at a time :: R, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

R Job Notifications Using Twitter

There has been some interesting activity about getting R to send a notification somehow when a long running job is completed. The most notable entries I have seen in this category are RPushBullet for web notifications and pingr for audio notifications.

Although RPushBullet looks really cool (and Dirk does great work), I wondered if there was a way to do this using a free service that I already had access to, namely twitter. See, twitter will notify you when someone else uses your handle in a tweet. If you have twitter notifications on your device, it should also appear on a mobile device, and if you are using tweetdeck on a desktop or laptop machine, you can also set up to get notifications there.

However, the default method of authenticating and caching oauth tokens for twitteR does not seem to be really useful for job notifications. However, we will take some precautions so that it is not too big a deal, and still be really useful.


In a nutshell, using a specific twitter account and app that have credentials stored in an R package. Below are the steps I used to make this happen. You can see my job notification twitter user rmf_notifier and the package I created.

Install twitteR

Before you begin, you should have a modern version of R (all of this was initially done using v 3.0.1), devtools, and the twitteR package from github.

install_github('twitteR', username='goeffjentry')

Setup a new package

We are going to create a personal package solely for sending twitter notifications and storing the API credentials.

Warning: Hadley Wickham in the httr token caching documentation advises against storing token caches in a package, but this is part of the reason we are creating a twitter user and app solely for this purpose (this makes it easier to revoke tokens, or remove the app, delete the user, etc if something goes wrong). This package also should never be published with the .rdata file included, and that file should never be under version control. If any of these things happen, someone else may be able to hijack this twitter account.

You should create your package (create a directory with DESCRIPTION, R directory, data directory, and NAMESPACE file). Look up how to do this if you are not sure. Set your working directory to be your package base directory for all further steps.

All following steps assume that you are working in the base package directory.

Twitter Account

We are going to create a twitter account and app specifically for sending this one type of notification. You can register a new twitter account using a new email (if you use gmail you can add a dot (.) anywhere in your email address for a unique address that still reaches you) and setup a new user name.

After setting up your new account, log in to https://apps.twitter.com, and create new app. Fill in all the necessary details, and when you have it up, modify app permissions to be Read and write. This will allow it to actually send messages on the accounts behalf.


Click on the API Keys tab, and set up your api data by copying the values into the code below. Note if you dont see a token and a secret, try hitting test oauth to generate one.

apikey <- "" #API Key
apisecret <- "" #API Secret
token <- "" #Access Token
tokensecret <- "" #Access token secret

And now we will authorize our app and make sure that it can tweet.

setup_twitter_oauth(apikey, apisecret, token, tokensecret)
tweet("this is a test") # make sure you can see a tweet
tweet("@username this is a test") # check that you see notifications, change @username to your own username

And then save the cache for later use.

local_cache <- get("oauth_tken", twitteR:::oauth_cache) # saves the oauth token so we can reuse it
save(local_cache, file="data/oauth_cache.RData")

Using saved credentials

To make sure that our package uses the saved credentials every time, we will include a .onLoad function that sets the oauth cache up properly. This should go in the file R/zzz.R

.onLoad <- function(libname, pkgname){
  cachedToken <- new.env()
  dataFile <- system.file("data/oauth_cache.RData", package="packageName")
  load(dataFile, cachedToken)
  assign("oauth_token", cachedToken$local_cache, envir=twitteR:::oauth_cache)

Notice that this function loads the credentials into a particular environment, and then sets the oauth_token variable in the twitteR:::oauth_cache environment to our saved credentials.

Our Notifying Function

Finally, we need a function that we can use to notify us when something happens. One could simply importFrom(twitteR, tweet) and export(tweet) in the namespace, but why should we have to type the username every time? This should go in R/packageFunction.R.

#' notifies job status
#' sends a tweet to rmflight the job status
#' @param tweetText the text to include in the tweet
#' @export
#' @importFrom twitteR tweet
jobNotify <- function(tweetText){
  fullTweet <- paste("@username", tweetText, sep=" ") # change @username to where you want to recieve notifications

Test it

And there you go. After building and installing your new package and loading it, you should be able to do:

library(packageName) # change to the name of your own package
jobNotify("this is a test")

And get a notification. Now you can simply put the above code at the end of any long running jobs, and voila, you are getting notifications from twitter about your R jobs.

You are subject to twitters app limits, so be careful how you use this. If you had lots of mini jobs, you would want to put this after all of them were finished, not have each one call this.


I would like to find a way to extend this to watching jobs and sending a notice at particular levels of completion, and also be able to have a try-catch that catches an error and can tweet that the job error'd out. But this level is still rather useful I think.

To leave a comment for the author, please follow the link and comment on their blog: Deciphering life: One bit at a time :: R.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Never miss an update!
Subscribe to R-bloggers to receive
e-mails with the latest R posts.
(You will not see this message again.)

Click here to close (This popup will not appear again)