Hi,

I'm using utf-8 charset and I'm a bit unhappy with the eZSearchEngine::normalizeText method.

The first line of the method is
$text =& strToLower( $text );

This breaks up utf-8 chars, so one of the following line
$unicodeValueArray =& $codec->convertString( $text );

won't give a reasonable result.

Example:

If you have the german word "für" (a preposition) - it will be converted to "f" after running through normalizeText. So the character "r" is also stripped off.

"Österreich" ( = "Austria" in english) -> "terreich"
"Schönfärberei" (= "garment dyeing" in english) => "schfberei"

At the moment this isn't big problem because the search-string of a user will also run through the method. So, if someone search "Österreich", he will find content which contains "Österreich".

But you give the feature of a wildcard search and I think this could be a problem because the search may find too much (or not expected) content.

Kind regards,
Emil.