[Mediawiki-l] import db: charset problem

Platonides Platonides at gmail.com
Mon May 14 19:44:57 UTC 2007


Klaus Becker wrote:
> Hi,
> 
> with mediawiki 1.9.3, I have a big charset problem with german and french 
> special caracters. 
> I begin with a basic question:
> 
> in the database, a word like "Université" ist stored as "Université" 
> in "page_title - varchar(255) - latin1_bin". Is that correct ? I suppose no. 
> Why is collation "latin1_bin" but phpmyadmin says "MySQL charset:  UTF-8 
> Unicode (utf8)" ?

Storing in latin1_bin is correct. mysql utf8 support is not "as good as 
it should" (and wasn't always available) so MediaWiki stores the utf8 
characters in a latin1 table. Université should be stored as Université

On your case, Université has been encoded again as utf8, as i warned 
you a week ago Likely caused by "intelligent" db dumpers.

> In another datase of a personal php-site, the same word is stored 
> as "Université" in "varchar(100)  latin1_swedish_ci" and this works.
> 
> In the wiki, all internal links with caracters like "é", "ä" and so on don't 
> work after export/import by phpmyadmin.
Because the titles on page table are broken.

> cheers
> 
> Klaus







More information about the MediaWiki-l mailing list