Re: Re: Question to Slovak Wikipedia SW
Tomasz Wegrzanowski <[email protected]> Wed, 25 Feb 2004 02:45:59 +0100
| Newsgroups | gmane.science.linguistics.wikipedia.international |
|---|---|
| Message-ID | <[email protected]> |
On Tue, Feb 24, 2004 at 05:33:23PM -0800, Brion Vibber wrote: > >"Stanislav Valasek" <[email protected]> schrieb: > >> during a translation of LanguageSk.php we found out a variable: > >> "linktrail" => "/^([a-z]+)(.*)$/sD" > [snip] > On Feb 25, 2004, at 01:22, Andre Engels wrote: > >On nl: I have for example changed this to > >"/^([äöüïëéèàa-z]+)(.*)$/sD", so that > >äöüïëéèà are also all seen as letters for this purpose. > > Note that for the wikis using UTF-8 encoding, one non-ASCII character > is composed of more than one byte, and the regexp matching currently > works on bytes. So you'd have to make it something like this: > > >"/^((?:ä|ö|ü|ï|ë|é|è|à|[a-z])+)(.*)$/sD" > > The [] bits match single characters (really bytes), and (||) matches > longer sequences. The ?: is to avoid messing up the returned match > sequence. Yuck! Sorry, it's one of those ugly implementation details > that we shouldn't have to expose... Doesn't PHP have perl-style -C switch for unicode regular expressions ?