Sweave,R,Beamer : How to convert the LaTex text in an Rnw file to R comments?

Question

Say I have a .Rnw file containing the usual LaTex mixed in with R code chunks. (I'm especially interested in converting a .Rnw slides document, but this question applies to any .Rnw document). Now I want to convert this to a file which contains all of the R code, plus all of the text that would normally be generated by LaTex, as R comments. In other words, the functionality I want is similar to what Stangle() does, but I also want all the text part of the LaTex converted to plain text that's commented out in the resulting .R file.

This would be a very convenient way to automatically generate a commented R file that's easy to look at in your favorite syntax-highlighting editor (e.g. emacs). This might not sound like a great idea for an Sweave document that's a long article with only a bit of R code, but it starts to look appealing when the .Rnw document is actually a slide presentation (e.g. using beamer) -- then the text portion of the slides would make perfect comments for the R code.

Anyone have any ideas on how to do this? Thanks in advance.

Tachyphylaxis

Answer 1

Here is one approach using regex. There are still some issues that remain, and I will maintain a list which will be updated with resolutions.

# READ LINES FROM RNW FILE
lines <- readLines('http://users.stat.umn.edu/~charlie/Sweave/foo.Rnw')

# DETECT CODE LINES USING SWEAVE CHUNK DEFINITIONS
start_chunk <- grep("^<<.*=$", lines)
end_chunk   <- grep("^@" , lines)
r_lines     <- unlist(mapply(seq, start_chunk + 1, end_chunk - 1))

# COMMENT OUT NON CODE LINES AND WRITE TO FILE
lines[-r_lines] <- paste("##", lines[-r_lines])
writeLines(lines, con='codefile.R')

ISSUES REMAINING:

1. Does not deal well with chunks called inside other chunks using <<chunk_name>>

Phosphate

You may want to define more complex regex to mend the chunk naming issue, like: "^<<.*>>=?$" for start chunk. But it's very neat approach, anyway... nice usage of mapply. – Lassalle

That's a neat approach, and yes, nice use of mapply. One more thing that would be nice is to get rid of all the LaTeX markup (things like \begin{frame}, \frametitle, ...) to produce clean, purely textual comments -- At least, getting rid of all LaTeX keywords would be a start. I suppose one could write a regex to replace all the reserved words of LaTeX with empty strings. That would be a start, but I'm hoping there's some way to leverage the LaTeX parser, and somehow capture the text that latex would have generated. – Tachyphylaxis

Combining @Ramnath's idea with one of the LaTeX-to-text solutions from another SO question (#530621) may get me what I want. – Tachyphylaxis