[Rcpp-devel] Unicode on windows

Ned Harding ned at alteryx.com
Thu Aug 1 18:57:07 CEST 2013


Just to follow up in case anyone other than me is using Unicode in R:  Rcpp does not support Unicode, or really any encoding other than 7 bit ascii.  Internally R marks every string with an encoding, typically UTF8, Latin1 or ASCII.  When using as<string> Rcpp just copies the bytes over ignoring the encoding.  This means that if you take a string that was utf8 and then later wrap it again, the encoding info is lost and the characters get corrupted.  In particular, never use Rcpp::as<std::wstring> because the string gets widened without being converted to Unicode.

If you want (or need) to support Unicode text in an R plugin, you need to use Rf_translateCharUTF8(...) to get a string.  Regardless of what encoding it was originally, R will make sure it is encoded as UTF-8.  In order to set a string into a R object you have to use the corresponding Rf_mkCharLenCE(p, len, CE_UTF8) function - which tells R that the data you have is UTF-8.

Ned.

From: rcpp-devel-bounces at lists.r-forge.r-project.org [mailto:rcpp-devel-bounces at lists.r-forge.r-project.org] On Behalf Of Ned Harding
Sent: Wednesday, June 26, 2013 11:54 AM
To: rcpp-devel at lists.r-forge.r-project.org
Subject: [Rcpp-devel] Unicode on windows

I am having issues with the wide string conversion to and from Rcpp.  When taking in a string from R that is encoding UTF-8, I would expect as<wstring> to have converted the utf-8 to a wide string.  Instead, it is just widening all the characters and leaving the UTF-8 encoding.  I have no issue with UTF-8, but my issue is that Rcpp doesn't seem to be able to tell me what encoding the source is so I don't know if I should convert or not.

Similarly, I would expect that wrap<wstring> would produce a UTF-8 encoding SEXP, but instead the encoding in R comes back "Unknown" and the data can't print.  See The C++ & R sources below along with the output.

C++ function
----------------------------------------
RcppExport SEXP TestWide(SEXP _strIn)
{
                std::wstring strIn = Rcpp::as<std::wstring>(_strIn);
                for (const wchar_t *p = strIn.c_str(); *p; ++p)
                                Rprintf("%x\n", *p);
                std::wstring str = L"a\x02a5c";
                return Rcpp::wrap(str);
}

R Script
----------------------------------------
test <- "a\u02a5b"
a<-.Call( "TestWide", test, PACKAGE = "AlteryxRDataX" )
print(Encoding(a))
print(a)

R Output
----------------------------------------
R version 3.0.0 (2013-04-03) - x86_64
rgeos version: 0.2-16, (SVN revision 389)
GEOS runtime version: 3.3.6-CAPI-1.7.6
Polygon checking: TRUE
61
ffca
ffa5
62
"unknown"
"a?"

Thanks,

Ned Harding
Alteryx
CTO
3825 Iris Avenue, Suite 150
Boulder, CO 80301
Phone:  720-259-0541
eMail:   ned at alteryx.com<mailto:ned at alteryx.com>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rcpp-devel/attachments/20130801/a2fe3710/attachment.html>


More information about the Rcpp-devel mailing list