Foreign Characters and Low Search

Question

We have a site where some entry titles are company names including a German umlaut. When we perform a search using Low Search and type in the name 'without' the umlaut (as you would be most likely to do if you have a english language keyboard) then no results are returned.

However when performing the same search from 'within' the ExpressionEngine control panel channel entries listing screen searching without the umlaut does return results.

Is there some fundamental difference between how internal entry searches are performed in EE and how Low Search works and is there any workaround we can implement to enable the return of results when we do perform a search on the front end minus the umlaut?

Answer 1

Keyword searches with Low Search are diacritic insensitive. That is to say, both the collection index and given search terms go through an algorithm to strip them out or replace them with unaccented equivalents. LS uses EE's native $foreign_characters array for this, located in /system/expressionengine/config/foreign_chars.php.

For example, take the word München. LS will change this, using the foreign_chars map, to muenchen. That means if you use the search term "münchen", LS will transform the search term to "muenchen" too. However, if you search for "munchen", LS will not change the terms since that doesn't contain diacritics.

So, it's best to change the values in the foreign_chars.php file to suit your needs. You'll need to update your indexes if you do.