datalang

The goal is to help people whose first, or preferred, language is not English. This package aids in translating the data and help files. The hope is to lower the barrier of learning R, and also data analysis.

Even though this package can translate data on the fly, this package is meant to be used inside another R package.

Installation

The development version is available on GitHub:

# install.packages("devtools")
devtools::install_github("edgararuiz/datalang")

How it works

The datalang package uses a translation spec file for translation. The expected file format is YAML. datalang uses the YAML file to know which column names, and data values, to change. The same YAML spec file can contain the information needed to produce a help file.

Here is a sample spec file:

df:
  source: thisweek
  name: esta_semana
variables:
  day:
    trans: dia
    desc: Dia de la semana
    values:
      friday: viernes
      saturday: sábado
      sunday: domingo
  morning:
    trans: manana
    desc: Temperatura a las 10 AM
  afternoon:
    trans: tarde
    desc: Temperatura a las 3 PM
help:
  name: esta_semana
  alias: esta_semana
  title: Un data set de ejemplo
  description: Tres días de temperaturas tomadas en la mañana y la tarde

So a the data set can go from this:

#>        day morning afternoon
#> 1   friday      76        88
#> 2 saturday      71        85
#> 3   sunday      70        83

To this:

#>       dia maniana tarde
#> 1 viernes      76    88
#> 2  sábado      71    85
#> 3 domingo      70    83

Basics

The datalang package includes a simple data set called: thisweek. The package also has a YAML spec file with the translation.

my_spec <- system.file("specs/thisweek.yml", package = "datalang")

The translate_data() function can ready the YAML spec file found in the my_spec file path. It uses that spec to locate the data set, in this case thisweek, modifies a copy of the data set based on the spec, and outputs a new, translated, data set.

translate_data(my_spec)
#>       dia maniana tarde
#> 1 viernes      76    88
#> 2  sábado      71    85
#> 3 domingo      70    83

Not all of the spec’s entries were needed to translate the data. The other entries are useful for creating the help documentation. In datalang there are two functions that output the help:

1. create_rd() - It outputs a character vector with contents that can be saved as an rd file.
2. create_html_help() - It outputs a character vector with contents that can be saved as an HTML file.

Using

There are two options to use datalang to provide accessibility to your package and data:

1. Use in a new R package dedicated to a translation to a single language
2. Use to extend, and provide translations from inside an existing R package

Option 1 - New package dedicated to a single translation

A package can be built to focus on translations to a single language. Here are some of the features of this approach:

1. The package contains copies of the data sets in the target language
2. The created help files are R “friendly”

The translate_folder() function is a higher level wrapper function that produces the translations and help files. The procedure would be to run translate_folder() on the package’s project folder, and then rebuilt the package. The default arguments have the current expectations:

• The translation spec YAML files will be located in the R project’s inst/specs folder.
• The target folder to save the new data sets to will be data.
• The location of the new help files will be man.

For a working example of this approach, check out the datos package. This package is dedicated to translating data sets into Spanish: https://github.com/cienciadedatos/datos.

Option 2 - Provide translations as part of your existing package

The goal of this approach will be for the existing package to provide the translated data, and the help files, without becoming bulky with copies of the data. For this approach, datalang provides a function, called on_attach, that can be run when the package is attached to the R session.

When the package is loaded, via library(), then datalang can check for a language setting, and then use that setting as both a trigger, and the pointer to which language should datalang translate to. At that point, the translated data set is created and loaded into the R session.

The on_attach() function can be added to the package’s .onAttach() function, so it’s processed when the library is loaded. This is what is currently on a fork of gapminder:

.onAttach <- function(...) {
  datalang::on_attach("gapminder")
}

Using this fork as an example. The language is set to Spanish by using the LANGUAGE environment variable (Sys.setenv(LANGUAGE = "es")):

Sys.setenv(LANGUAGE = "es")

library(gapminder)

#> El lenguaje asignado actualmente es español  
#>   Cargando datos y/o ayuda de los siguientes objetos : 
#>      gapminder  >->  paises 
#>    Una nueva función de ayuda fue cargada: ayuda() 
#>      Utilice asi: ayuda(paises)

Help function

Because the help documentation is created on-the-fly, it cannot be added to the regular data file that functions such as help() use. So a new help wrapper function is created that creates an HTML version of the data set, or function, help. If the function, data set, or general topic is not one created by datalang then the requested topic is passed through to the help() function in the background.

The help’s function name will be set based on the language. So in Spanish, the name will be: ayuda.