The R Language is really good at data and statistical analysis, but when
it comes to working with information security data it has a few holes
that need plugging up. Bob has been doing a couple of posts using Rcpp
to do things like Basic DNS
Lookups,
TXT
lookups,
and IPv4
Conversions.
I wanted to add to some of that work with a quick package for parsing
domain names.

While *.com, *.net and *.org top-level domains are easy
to parse, the rest of the world gets messy rather quick. Just taking the
entry after the last dot creates problems for top-level domains like
anything in *.com.uk. Or to make things even more complicated, the
name of “us-west-1.compute.amazonaws.com” is considered (for name
parsing) to be a top-level domain and the domain name we’d want to
process is the name that would appear before the us-west-1 in that name.

Introducing TLD Extract (the R version)

It’s always easier to imitate rather than reinvent, so I took some time
to read through the
tldextract Python
package, and used that to test my code was executing properly during
development so I used the same name for the R pacakge. The data for the
package is drawn from the same source as the python package, the Public
Suffix List from the Mozilla Foundation.
For convenience, I include a cached version of the data so it can run offline.

Installation

To install this package, use the devtools package:

devtools::install_github("jayjacobs/tldextract")

Usage

Using the package is fairly straight forward, it will return a data frame with the
original name and seperate columns for each parsed component.

library(tldextract)
# use the cached lookup data, simple call
tldextract("www.google.com")

##             host subdomain domain tld
## 1 www.google.com       www google com

# it can take multiple domains at the same time
tldextract(c("www.google.com", "www.google.com.ar", "googlemaps.ca", "tbn0.google.cn"))

##                host subdomain     domain    tld
## 1    www.google.com       www     google    com
## 2 www.google.com.ar       www     google com.ar
## 3     googlemaps.ca       googlemaps     ca
## 4    tbn0.google.cn      tbn0     google     cn

The specification for the top-level domains is cached in the package and
is viewable.

# view and update the TLD domains list in the tldnames data
data(tldnames)
head(tldnames)

## [1] "ac"     "com.ac" "edu.ac" "gov.ac" "net.ac" "mil.ac"

If the cached version is out of data and the package isn’t updated, the
data can be manually loaded, and then passed into the function.

# get most recent TLD listings
tld <- getTLD() # optionally pass in a different URL than the default
manyhosts <- c("pages.parts.marionautomotive.com", "www.embroiderypassion.com", 
               "fsbusiness.co.uk", "www.vmm.adv.br", "ttfc.cn", "carole.co.il",
               "visiontravail.qc.ca", "mail.space-hoppers.co.uk", "chilton.k12.pa.us")
tldextract(manyhosts, tldnames=tld)

##                               host   subdomain            domain       tld
## 1 pages.parts.marionautomotive.com pages.parts  marionautomotive       com
## 2        www.embroiderypassion.com         www embroiderypassion       com
## 3                 fsbusiness.co.uk                fsbusiness     co.uk
## 4                   www.vmm.adv.br         www               vmm    adv.br
## 5                          ttfc.cn                      ttfc        cn
## 6                     carole.co.il                    carole     co.il
## 7              visiontravail.qc.ca             visiontravail     qc.ca
## 8         mail.space-hoppers.co.uk        mail     space-hoppers     co.uk
## 9                chilton.k12.pa.us                   chilton k12.pa.us

And there we have it!

One last thing, this is the first package I created with unit tests.
This package is really simple and adding in unit tests seamed like a
no-brainer. After reading through Hadley Wickham’s Advanced
R online book and exploring how
other packages implement the
testthat package, I implemented a
few simple tests. If you are creating (or about to create) R packages,
look at the source for the tldextract package
for the incredibly simple unit tests included with it!