[R] Find out what "native.enc" corresponds to

[Milan Bouchet-Valat]

Le dimanche 05 août 2012 à 10:04 +0100, Prof Brian Ripley a écrit :
> On 05/08/2012 09:54, Milan Bouchet-Valat wrote:
> > Hi!
> >
> > I'm using R2HTML in my RcmdrPlugin.temis package to output localized
> > strings to a HTML file. Thus, I insert a simple header at the top of the
> > file to specify what encoding is used; if I don't do that, Web browsers
> > assume it is latin1, which is not always true.
> >
> > My problem is, I could not find a way to detect what encoding is used by
> > R2HTML in the most general case. R2HTML simply calls cat() with the file
> > name, which means the text connection is opened using file(encoding =
> > getOption("encoding")). This is fine, except that when
> > getOption("encoding")) is set to "native.enc", I'm not able to find out
> > the real encoding that was used for output.
> >
> > Of course, ideally I would tell R2HTML to output everything as UTF-8,
> > and I would add this information to the header. But AFAICT this is not
> > possible in the current state of this package. So I would be very
> > grateful if somebody could provide me with a solution to resolve
> > "native.enc" to the encoding name.
> 
> ?options points you to ?connections, which does explain this.  See 
> Sys.getlocale("LC_CTYPE") to see
> 
> 'the internal encoding of the current locale'
> 
> (or at least, what the OS claims it to be: e.g. some lie about 'C' locales).
Thanks for the pointers, but the issue is/was that LC_CTYPE does not
provide a valid encoding name. But your reply prompted me to read ?iconv
again, and I discovered the existence of localeToCharset(), which seems
to provide me with the encoding name I'm looking for.

> As for a name, iconv() knows this as "" (and some OSes do make it rather 
> hard to find a name if it is not part of the locale name).
I'm afraid I don't understand what you mean. Do you suggest I encode
data to/from the current encoding?


Regards