output: github_document
editor_options:
chunk_output_type: console

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-"
)
library(ntfy)

ntfy

ntfy (pronounce: notify) is a simple HTTP-based pub-sub notification service. It allows you to send notifications to your phone or desktop via scripts from any computer, entirely without signup, cost or setup. It's also open source if you want to run your own. Visit ntfy.sh for more details.

{ntfy} is a lightweight R wrapper for this service. The magic sauce is just POST and GET calls equivalent to

curl -d "Process Complete 😀" ntfy.sh/yourSecretTopic

but made to work nicely in an R workflow.

Installation

You can install the development version of ntfy like so:

# install.packages("remotes")
remotes::install_github("jonocarroll/ntfy")

Functionality

Follow the instructions at ntfy.sh to install any of the mobile apps or use the web app. No sign-up or account is necessary.

Choose a topic (note: this isn't a password-protected service, so choose something obscure) and subscribe to it on your device.

Add the topic as an environment variable, e.g.

usethis::edit_r_environ()

[...]

NTFY_TOPIC='yourSecretTopic'
#NTFY_SERVER='https://ntfy.sh'

The server will automatically be set to https://ntfy.sh unless you specify another.

This can be confirmed with

ntfy_server()

With the package loaded, you can now send notifications which should appear on your device

library(ntfy)
ntfy_send("test from R!")

This can be used in many ways. One would be to notify the completion of a process. The ntfy_done() function sends a notification with the (default) body

Process completed at <Sys.time()>

slow_process <- function(x) {
  Sys.sleep(8) # sleep for 8 seconds
  x
}

mtcars |> 
  head() |> 
  slow_process() |> 
  ntfy_done()

which results in a notification on subscribed devices

Process completed at 2023-07-04 17:00

When using the base R pipe |> the piped commands are composed together by the parser, so

f() |> 
  g() |> 
    h()

becomes

h(g(f()))

We can use this fact to time the running of a process if the last function (above, h()) is system.time(). The ntfy_done_with_timing() function does exactly this

mtcars |> 
  head() |> 
  slow_process() |> 
  ntfy_done_with_timing()

which sends the notification

Process completed in 8.003s

Note: the {magrittr} pipe %>% works differently and does not compose the same way, so this will result in a very short time report. Wrapping an entire pipeline with ntfy_done_with_timing() will work, though

library(magrittr)
ntfy_done_with_timing(
  mtcars %>%
    head() %>% 
    slow_process()
)

sends

Process completed in 8.004s

If you're using a topic on a server that requires authentication, you can pass auth = TRUE, along with a username and password:

ntfy_send(
  "test from R!", 
  auth = TRUE, 
  username = "example", 
  password = "super-secret-password"
)

Alternatively, you can set these as environment variables and they'll get used by ntfy_send() automatically:

usethis::edit_r_environ()

[...]

NTFY_AUTH='TRUE'
NTFY_USERNAME='example'
NTFY_PASSWORD='super-secret-password'

ntfy_send("test from R!")

The history of the notifications sent can be retrieved as well, with control over how far back to search

ntfy_history(since = "1h")

structure(list(id = c("0oDpk4oisfNO", "4Fcy9kIL0m6Z", "AGXn4q0CirFT"
), time = c(1667988383L, 1667988413L, 1667990983L), event = c("message", 
"message", "message"), topic = c("yourSecretTopic", "yourSecretTopic", "yourSecretTopic"
), message = c("test from R!", "Process completed at 2022-11-09 17:31:03", "Process completed in 8.003s")),
row.names = c(NA, -3L
), class = "data.frame")

API

The full ntfy.sh API should be supported, including sending a title and tags

{width=300}

Images

Images can be sent within notifications by specifying as image either the filename or a ggplot2 object (which will be saved to a temporary file)

library(ggplot2)
p <- ggplot(mtcars, (aes(mpg, wt))) + 
  geom_point() + 
  geom_smooth() + 
  labs(title = "ggplot images in {ntfy}")
ntfy_send("ggplot2 images in notifications!", 
          tags = c("tada", "chart"),
          image = p)

Emoji

Supported tags (emoji) can be sent with the tags argument (one or more). These can be searched or shown with show_emoji() which will look for a given name in the compatible values, or search for it in the compatible metadata.

The compatible data is stored as emoji

data("emoji")
head(emoji)

with the tags stored as tags for easy auto-complete

ntfy_send(message = "sending with tags!", 
          tags = c(tags$cat, tags$dog)
)

The compatible emoji can be shown with

show_emoji("rofl")

If the name is not found in aliases (the compatible names) it will be searched in tags

show_emoji("lol")

show_emoji("pet")

You can force this behaviour with

show_emoji("dog", search = TRUE)

Similar Services

{Rpushbullet} offers similar functionality, but requires sign-up / an API key
{beepr} can play a sound when a process completes
IFTTT has an API and can be configured to send messages with e.g. nifffty
This blog post details many ways to send notifications, via email, text, Slack, and MS Teams

Privacy

Q: "Will you know what topics exist, can you spy on me?"

A: Refer to the FAQ

Contributing

If this service is useful to you, consider donating to the developer via GitHub sponsors. If this package is useful to you, I also accept donations via GitHub sponsors.