Romain Francois, Professional R Enthusiast

To content | To menu | To search

Tag - parser

Entries feed - Comments feed

Sunday, November 28 2010

parser 0.0-13

I've pushed a new version of the parser package to CRAN.

This is the first release that depends on Rcpp, which allowed me to reduce the code size and increase its maintainability.

This also features a faster version of nlines, a function that retrieves the number of lines of a text file.

Sunday, November 22 2009

new R package : highlight

I finally pushed highlight to CRAN, which should be available in a few days. The package uses the information gathered by the parser package to perform syntax highlighting of R code

The main function of the package is highlight, which takes a number of argument including :

  • file : the file in which the R code is
  • output : some output connection or file name where to write the result (The default is standard output)
  • renderer : a collection of function controlling how to render code into a given markup language

The package ships three functions that create such renderers

  • renderer_html : renders in html/css
  • renderer_latex: renders in latex
  • renderer_verbatim: does nothing

And additionally, the xterm256 package defines a renderer that allows syntax highlighting directly in the console (if the console knows xterm 256 colors)

Let's assume we have this code file (/tmp/code.R)

f <- function( x){
        x + rnorm(1)
}

g <- function(x){}
h <- function(x){}

Then we can syntax highlight it like this :

> highlight( "/tmp/code.R", renderer = renderer_html(), output = "/tmp/code.R.html" )
> highlight( "/tmp/code.R", renderer = renderer_latex(), output = "/tmp/code.R.latex" )

which makes these files : code.R.html and code.R.latex

The package also ships a sweave driver that can highlight code chunks in a sweave document, but I'll talk about this in another post

Tuesday, August 4 2009

R parser package on CRAN

The parser package has been released to CRAN, the package mainly defines a function parser that is similar to the usual R function parse, with the few following differences:

  • The information about the location of each token is structured differently, in a data frame
  • location is gathered for all symbols from the source code, including terminal symbols (tokens), comments
  • An equal sign is identified to be either an assignment, the declaration of a formal argument or the use of an argument

Here is an example file containing R source code that we are going to parse with parser

#' a roxygen comment
f <- function( x = 3 ){
	
	# a regular comment
	rnorm(10 ) + runif( 10 )
	
}

It is a very simple file, for illustration purpose. Let's look what to do with it with the parser package

The parser generates a list of expressions, just like the regular parse function, but the gain is the data attribute. This is a data frame where each token of the parse tree is a line. The id column identifies each line, and the parent column identifies the parent of the current line.

At the moment, only the forthcoming highlight package uses the parser package (actually the parser package has been factored out of highlight), but some anticipated uses of the package include:

  • rework the codetools package so that it tells source location of potential problems
  • code coverage in RUnit or svUnit
  • rework the roxygen parser