Tue, 28 Apr 2015

RcppTOML 0.0.3: A New Approach to Configuration Files

A small project I worked on during the last few weeks has now come together in new package RcppTOML which arrived on CRAN yesterday.

It provides R with a reader for TOML files. TOML stands for Tom's Obvious Markup Language. And before you roll your eyes, glance at the TOML site. It really is different, and has a number of rather wonderful features:

  • free-format indentation as you please
  • comments anywhere, even on the same line
  • actual types such as string, integer, float, bool and datetime (!!) which are all native
  • vectors, of course, of the above
  • arbitrary nesting of tables

Here is a simple illustration where we parse the TOML example file derived from what is part of the main TOML README

R> p <- parseTOML(system.file("toml", "example.toml", package="RcppTOML"))
R> summary(p)
toml object with top-level slots:
   clients, database, owner, servers, title 
read from /usr/local/lib/R/site-library/RcppTOML/toml/example.toml 
R> p
List of 5
 $ clients :List of 2
  ..$ data :List of 2
  .. ..$ : chr [1:2] "gamma" "delta"
  .. ..$ : int [1:2] 1 2
  ..$ hosts: chr [1:2] "alpha" "omega"
 $ database:List of 4
  ..$ connection_max: int 5000
  ..$ enabled       : logi TRUE
  ..$ ports         : int [1:3] 8001 8001 8002
  ..$ server        : chr ""
 $ owner   :List of 4
  ..$ bio         : chr "GitHub Cofounder & CEO\\nLikes tater tots and beer."
  ..$ dob         : POSIXct[1:1], format: "1979-05-27 07:32:00"
  ..$ name        : chr "Tom Preston-Werner"
  ..$ organization: chr "GitHub"
 $ servers :List of 2
  ..$ alpha:List of 2
  .. ..$ dc: chr "eqdc10"
  .. ..$ ip: chr ""
  ..$ beta :List of 2
  .. ..$ dc: chr "eqdc10"
  .. ..$ ip: chr ""
 $ title   : chr "TOML Example"

See much more at the TOML site. I converted one first project at work to this and it really rocks. Point to a file, get a list back and index all components by their names.

We also added really simple S3 classes to the default print() method uses str() for a more compact presentation of what (in R) is of course nested list types.

Internally, the RcppTOML packages use the splendid cpptoml parser by Chase Geigle. This brings in modern C++11 and makes it that CRAN simply cannot build a binary for R on Windows as the g++ version (still, as of April 2015) in Rtools is too old. There is word of an update to Rtools and that point should we able to support Windows as well. Until then, no mas.

A bit more information is on the package page here as well as as the GitHub repo.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

/code/rcpp | permanent link