Thanks a lot Cameron! I was just sitting down to write my lib, planning on
doing as you suggest below.

On Thu, May 20, 2010 at 5:06 PM, Cameron Kaiser <spec...@floodgap.com>wrote:

> > I will write my own if I have to. But before I do, I'd like to understand
> as
> > many details as possible about the specifics of Twitter's RFC 3986
> behavior.
>
> This is the regex I'm using, which is known to work:
>
>        $x =~ s/([^-0-9a-zA-Z._~])/"%".uc(unpack("H2",$1))/eg;
>
> In short, letters, numbers, and the set of -._~ are NOT URL encoded.
> Everything else is.
>
> Note this routine is not 100% UTF-8 safe as written; I have other code
> that handles that, so you may need to do that as your library warrants.
>
> --
> ------------------------------------ personal:
> http://www.cameronkaiser.com/ --
>  Cameron Kaiser * Floodgap Systems * www.floodgap.com *
> ckai...@floodgap.com
> -- People are weird. -- Law & Order SVU
> ---------------------------------------
>

Reply via email to