problem finding certain phrase

[sujag]

I have a strange problem with finding a certain (german) text in my page. On kulturfeste.de a search for "Festival für Freunde" https://kulturfeste.de/suche/?q=festival+für+freunde&submit= doesn't get results, omitting "für" https://kulturfeste.de/suche/?q=festival+freunde&submit= get's beside others the expected result https://kulturfeste.de/feste/festival-fuer-freunde/

The used selectors for $pages->find are

'title|headline|Intro~*=festival für freunde, limit=50, template=fest'

and

'title|headline|body~*=festival für freunde, limit=150, template=veranstaltung,sort=eventTime'

The search text appears multiple times in the relevant fields, it seems not to be the word "für" as possible stopword and not the umlaut 'ü' Now I'm running out of ideas. What else could I check for?

 

[ottogal]

1 hour ago, sujag said:

a search for "Festival für Freunde" https://kulturfeste.de/suche/?q=festival+für+freunde&submit= doesn't get results,

On that page it says

Suche nach "festival für freunde"

It might be a case issue. Usually a search term in double quotes is case sensitive.

[sujag]

Thanks, but the quotes are added by the template, they are not part of the search term and this wouldn't explain a difference between the search phrases "Festival für Freunde" and "Festival Freunde" (both submitted without quotes)

[aagd]

I'm just guessing, but I think "für" is simply too short to get indexed, but you expect to match the exact phrase by using the `*=` selector operator. You might try `%=` instead.

See also @BitPoet's post in Help understanding search results - General Support - ProcessWire Support Forums

[sujag]

It does indeed work with the *= operator, but I don't quite understand why and I get more than the wanted results. 

I think, too short can't be the reason, cause the search without "für" works as well another search after "Gesänge der Mönche" where the article "der" doesn't interfere with the result.

[aagd]

Do you have a custom MySQL config? Because it seems to me that by default it shouldn't find für or der:

Quote

The *= and ~= rely upon MySQL fulltext indexes, which only index words of at least a certain length (configurable, but typically 4 characters)

Source: Using selectors in ProcessWire CMS

[sujag]

I'm on a shared hosting, don't know about a custom config.

But anyway, shouldn't stopwords just be ignored? In the search phrase and the result? Then there shouldn't be a different result searching for "Festival Freunde" and "Festival für Freunde"

[aagd]

Is "für" a really stopword? Try $database->getStopwords() method - ProcessWire API to find out.

[sujag]

To be honest I don't think that stopwords are the problem. I checked this a few days ago and got only an english stopword list. Then I switched to testing different phrases and decided to ask others.

Another try to explain the problem:

I get the same results when searching for "Gesang der Mönche" and "Gesang Mönche" but different when searching for "Festival für Freunde" und "Festival Freunde"

Why?