Finding names in R packages which could be domain names (using rvest and the tidyverse)

Introduction

Twitter likes to turn things which look like URLs into URLS.

I tweeted some rstats code last week and twitter decided that seq.int was actually http://seq.int.

According to wikipedia, the .int TLD has the “strictest application policies of all TLDs”, so there’s no chance of registering that particular domain.

Out of the other standard R packages in base in the tidyverse, which ones could be domain names?

Valid Top Level Domains (TLDs)

According to wikipedia the list changes frequently, so I’m just going to grab them today as a snapshot in time.

library(tidyverse)
library(xml2)
library(rvest)

#-----------------------------------------------------------------------------
# - Read in HTML
# - Extract tables with TLDs
#
# - https://en.wikipedia.org/wiki/List_of_Internet_top-level_domains
#-----------------------------------------------------------------------------
html   <- xml2::read_html("../../static/img/TLDs/List_of_Internet_top-level_domains")
tables <- rvest::html_nodes(html, 'table')


#-----------------------------------------------------------------------------
# Extract all the TLDs from the html tables
#-----------------------------------------------------------------------------
tlds <- tables[7:65] %>%
  purrr::map_dfr(~rvest::html_table(.x, fill=TRUE)[,1:2] %>% 
                   set_names(c('name', 'tld_info'))) %>%
  as.tbl() %>%
  filter(grepl("^\\.", name)) %>%
  mutate(
    name     = stringr::str_sub(name, start=2),
    tld_info = gsub("\n", ' ', tld_info)
  ) %>%
  replace_na(list(tld_info=''))

Names in base, tidyverse and misc packages

#-----------------------------------------------------------------------------
# Packages from which to extract names
#-----------------------------------------------------------------------------
packages <- c(
  'base', 'graphics', 'grDevices', 
  'stats', 'stats4', 'utils', "broom", 
  "cli", "crayon", "dplyr", "dbplyr", "forcats", "ggplot2", 
  "haven", "hms", "httr", "jsonlite", "lubridate", "magrittr", 
  "modelr", "purrr", "readr", "reprex", "rlang", 
  "rstudioapi", "rvest", "stringr", "tibble", "tidyr", "xml2", 
  'Hmisc', 'zoo', 'knitr', 'Rcpp', 'shiny', 'xtable', 
  'data.table', 'devtools', 'testthat', 'roxygen2'
)


#-----------------------------------------------------------------------------
# Helper function for extracting names from a package
#-----------------------------------------------------------------------------
extract_names_from_package <- function(package_name) {
  data_frame(
    package   = package_name,
    function_ = ls(asNamespace(package_name))
  )
}


#-----------------------------------------------------------------------------
# Extract names from all named packages
#-----------------------------------------------------------------------------
all_names <- packages %>% 
  purrr::map_dfr(extract_names_from_package)

Warning: `data_frame()` is deprecated, use `tibble()`.
This warning is displayed once per session.

#-----------------------------------------------------------------------------
# Keep only those which contain a "."
#-----------------------------------------------------------------------------
possible_tlds <- all_names %>%
  filter(grepl("\\.", function_)) %>%
  mutate(
    tld = purrr::map_chr(strsplit(function_, '\\.'), tail, 1)
  )


#-----------------------------------------------------------------------------
# Keep only those whose last part of their name matches a TLD
#-----------------------------------------------------------------------------
possible_tlds %>% 
  inner_join(tlds, by=c(tld='name')) %>%
  knitr::kable()

Notes

• Impossible
  • is.na is not supported as the registry requires at least 3 characters. in the first part
  • .call is by amazon registry and i can’t see how to even try and register
  • .off - I couldn’t find a registry that sold it.
  • .test (like .example) is purely for testing
  • .map and .new seems to be associated with google registry.
• Expensive
  • lm.fit was quoted as available for $1200
  • match.fun quoted as available for $7500
• Taken
  • as.name
  • is.name
  • sys.info
  • na.fail
• Possible
  • glm.fit for just $64
  • POSIXct.date and POSIXlt.date may be available. $30
  • spline.fit for $33