Making of nivocal

nivocal was built to be a working package, but while I was at it I wanted to record each step in the creation journey as a reference for future authors of reactR-based htmlwidgets. For reference, the package in its initial working state required less than 30 minutes of effort. I never left my RStudio window, and I only had to write 1.5 lines of JavaScript.

One Time on the Internet I Found …

There are some amazing React comoponents out there. If one day on the Internet, you find something you like then it can be ready to use in R in less than an hour (hopefully shorter if you read this document). Take for example the @nivo set of components. I’d like to use the Github-style calendar.

Starting the Package

usethis allows us to create a package in one line of code. Let’s begin our journey here.

usethis::create_package("nivocal")

Scaffolding

Once we have a package, we’ll open it up and then build a scaffold. Sometimes finding the npmPkgs argument can be a little tricky. Usually, the best clues are in the docs, but we can also use unpkg.com–the CRAN of Node JavaScript–for some help. End the url with / to see the contents of the package and find the most recent version. For the calender, we do https://unpkg.com/@nivo/calendar/. Remember the /. Try https://unpkg.com/@nivo/calendar to see the difference.

scaffoldReactWidget(
  "nivocal",
  npmPkgs = c("@nivo/calendar" = "0.52.1")
)

Now we have all the files we need for a working htmlwidget but unfortunately not working in the way we want.

1.5 Lines of JavaScript and Build

In the JavaScript, we will need to import the module we want to use. For nivocal we want ResponsiveCalendar. import in JavaScript is very similar to library() in R.

import { ResponsiveCalendar } from '@nivo/calendar'

The JavaScript build toolchain can get complicated, but fortunately reactR takes care of much of this for us. I hate to tell you, but you will need to install node and yarn. I promise this is not hard or scary though. Once you have both installed, we will build/webpack our JavaScript in the RStudio terminal or other terminal/console.

yarn install
yarn run webpack

The built JavaScript will be copied into the /inst/htmlwidgets directory ready for use in our R htmlwidget.

Build R Package

We have a couple more things to do on the R side. For now, let’s see if the package builds. In RStudio, we can CTRL + Shift + B or

devtools::document()
devtools::install(quick = TRUE)

If all goes well, then our package is working, but as I said just not quite in the way we want.

Add Some Arguments

Now we need a way to go from R to JavaScript. We’ll add arguments for the data, from, and to component props in our R function.

Change the Tag

The scaffold uses div, but we want to use the ResponsiveCalendar component. React components are always capitalized.

Add More Props/options and Do Some R Work

There are a lot of other options for the calendar. For a well-built R package, I think each of these should be dcoumented arguments, but for now we’ll use ... to pass other options from R to JavaScript.

data, from, and to are required for the calendar component. Eventually, we want to accept various forms of data from the user, but for now we will assume the user provides a data.frame with two columns day and value. htmlwidgets communicate data.frames as an array of arrays but ResponsiveCalendar wants the equivalent of dataframe = "rows" in jsonlite::toJSON(). We’ll use mapply to do this, but as described in the data transformation article we have other methods to achieve this. The most common form – using JavaScript HTMLWidgets.dataframeToD3() – does not currently work well with reactR-based htmlwidgets.

Without from and to, the calendar will not render, so let’s assume the user wants from to be the first row of the data and to to be the last row.

It’s Working

Now we have a working htmlwidget. Build the package with CTRL+Shift+B or

devtools::document()
devtools::install(quick = TRUE)

Give it some data and see an interactive calendar.

library(nivocal)

# fake data of 500 records/days starting 2017-03-15
df <- data.frame(
  day = seq.Date(
    from = as.Date("2017-03-15"),
    length.out = 500,
    by = "days"
  ),
  value = round(runif(500)*1000, 0)
)

nivocal(df)

Customize

Remember we added ... for further customization. Let’s see how this works.

library(nivocal)

# fake data of 500 records/days starting 2017-03-15
df <- data.frame(
  day = seq.Date(
    from = as.Date("2017-03-15"),
    length.out = 500,
    by = "days"
  ),
  value = round(runif(500)*1000, 0)
)

nivocal(
  df,
  direction = "vertical",
  colors = RColorBrewer::brewer.pal(n=9, "Blues")
)

More Resources

Even though all of this is fairly new, we have tried to offer examples and resources to ease the learning curve. The react-R Github organization is intended to be a friendly central location for all things R + React. Please join in the fun.

• Alan Dipert rstudio::conf 2019 Video hear from the creator himself

• Authoring reactR htmlwidgets tutorial

• react-sparklines example package great first example package

• office-fabric-ui example package widget example with no JavaScript build steps

• MapboxGL example package package discussed in the rstudio::conf 2019 video

• forcer react-force-graph example package for 2d, 3d, and virtual reality force directed network graphs.

Vue

We’d like to do the same for Vue. Please let us know if you have interest. vueR would be a good starting point.