[R] parsing - input buffer overflow

[Prof Brian Ripley]

On Fri, 13 Jun 2008, Daniel Malter wrote:

> Hi,
>
> I am trying to parse a large amount of text using gregexpr(). Unfortunately,
> I get an "input buffer overflow" message when I attempt that with too large
> an amount of text. The error messages occurs before the parsing. The problem
> is that I cannot assign the text to a variable (an object) if the text is
> too large.

R does have limits on the command line length (1024 bytes up to R-devel, 
4096 bytes there).  What happens if you exceed that depends on the 
interface you are using (and you have not told us).  Beyond that, the 
parser has a limit of MAXELTSIZE (8192 bytes) on strings.

I don't see any need for 'improvement' though: why are you entering very 
long strings as part of the R program?  They are data, and e.g. 
readLines() and scan() have no limits on string length beyond those 
imposed by R's internals (2^31-1 bytes).

> This problem has been mentioned before, which I found using the RSiteSearch.
> However, the post is from 2006, and I thought it might have improved by now.
> Is there any way to increase the limit or to get around this problem?
>
> x="Saint Lucia, Saint Kitts and Nevis, Saint Helena, Clipperton Island,
> Tristan da Cunha"

I presume that is not an example?  It looks like a character vector which 
has been collapsed by paste(x, ", ") and would be better strsplit() into 
its components than using gregexpr.

> #What I want to achieve is to parse the text for the number of occurrences
> of a certain character string within the text.
>
> #This is done using:
>
> n=100 #choose n large enough
> length(which(is.na(gregexpr("Saint",x,ignore.case=TRUE)[[1]][1:n])==FALSE))
>
> But again, if the text is large, I cannot assign it to x. I'd be grateful
> for any suggestions.
>
> Cheers,
> Daniel
>
>
> -------------------------
> cuncta stricte discussurus
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595