Dockerize R, expose as API

A few month ago, there was a requirement to use some R lang code in production. Porting it to other languages (c#) was not only time consuming but not reliable to be used in production, because we were worried new code may generate different result. Here is my solution, hope it helps you as it did help me.

#plumberapi #rlanguage #dockerfile


imagine you have an peice of code developed by your data scientist/ Mathematician/Statistician or whoever else that is happy to work with R, now your devs are required to plug in that peice of code to your project. i do not need to imagine it cuz it happened for me and this is how i solved it

step 0

lets start from creating a folder for this project$ mkdir exposemyRcodeasAPI $ cd  exposemyRcodeasAPI/

create a text file name it plumber.r$ touch plumber.r

then open plumber file with your text editor$ nano plumber.r

step 1

add following to head of your filelibrary(fitdistrplus) library(scales) library(actuar) library(plumber)

step 2

source the R code (lets call it func.R)source("funcs.R")

step 3

at run time your plumber file will generate swagger docs by adding following line you will set your apititle (append following text to plumber.R file after sourcing the R code)#* @apiTitle yourapinamehere

step 4

wrapp whichever function that you want to expose in a function with following declaration

#* Echo back the input
#* @param msg The message to echo
#* @get /echo
function(msg = "") {

if you want to have post method use (#* @post instead of get) to read more click here

step 5

now everything should be fine, you can dockerise your code and ship it anywhere you want

$ touch Dockerfile

append following lines

FROM trestletech/plumber

#to install packages uncoment following line
#RUN R -e "install.packages('whatever it is used in the code')"

#for apt-get uncomment & modify following line 
#RUN apt-get update -qq && apt-get install whatever is needed 

#expose a port 

#make a folder and copy your files 
RUN mkdir /app
COPY funcs.R /app
COPY plumber.R /app

#set entrypoint
ENTRYPOINT ["R", "-e", "pr <- plumber::plumb('plumber.R'); pr$run(host='', port=8000,swagger=TRUE)"]

build your docker and forward port 8000 from your container to any port you want on host