On Fri, Apr 16, 2004 at 02:19:50PM +0100, Richard Jones wrote:
> A user on one of the sites that I run has managed to create two
> user accounts for themselves:
>
> Yoshi
> %EF%BC%B9%EF%BD%8F%EF%BD%93%EF%BD%88%EF%BD%89 (UTF-8 using URL encoding)
Which, for reference, is "Ｙｏｓｈｉ" (double-width).
> When rendered in a web browser they both appear as "Yoshi", but from
> the point of view of my code and the database they are, of course,
> different. I allow people to have unrestricted usernames rather than
> restricting them to ASCII-printable-only characters because this makes
> sense on a Japanese site.
A simple fix for the above would be to convert U+FFxx characters; that's
all of those double-width roman characters, along with half-width
katakana and punctuation, which could present a similar problem.
http://www.unicode.org/charts/PDF/UFF00.pdf
Another approach would be to convert to NFKC or NFKD. This might do more
than you want, however; see
http://www.unicode.org/charts/normalization/chart_Katakana.html
--
Glenn Maynard
--
Linux-UTF8: i18n of Linux on all levels
Archive: http://mail.nl.linux.org/linux-utf8/
