Choosing the correct language analyzer

[comzeradd]

Hi,

I have two websites that have identical language configuration. Both are using the WPML-Elasticpress add-on. Both are bilingual for Arabic (default) and English.

On the first one, if I go to Status Report, I can see two indices. The first one is for Arabic and the language analyzer is set correctly to arabic.

But on the second one I noticed that the analyzer was set to english. So I tried adding this to its functions file:

add_filter('ep_analyzer_language', static function (): string {
	return 'arabic';
});

That did change the analyzer to arabic and surprisingly didn’t affect the English one.

But I was curious on this inconsistency. Is there a better way to have the proper language analyzer per language?

[decodekult]

Hi Nikos

The way we generate the indices and the mapping applied to each of them, including the analyzer, happens after this filter that you used is applied. We get data for each language during the synchronization, and apply the right mapping analyzer for it.

In theory, ElasticPress shoudl eb setting Arabic as the site default language for mappings and analysis, if that is the language set as default in the WordPress settings, and if the ElasticPress settings are set as to use the site default language as its own language - I assume that both of those settings are right on both of your sites?

If the conditions above apply, then the content in the default language (Arabix) shoudl use the analyzed in the default language (Arabc) and WPML ElasticPress should not be modifying it - we would only modify the analyzed when dealing with English content.

If despite both conditions above (the default language of the site is Arabic under the WordPress Settings -> General page, and ElasticPress setting is set to use the default site language, then it shoudl be working fine out of the box. The code that youa dded woudl do no harm, but if it is needed there might be something that I am missing.

If despite the checks above the issue still persists when you remove your code snippet, I would need some additional information to try and find out what is happenign here:

• The WordPress default language. When using WPML, it should be the WPML default language.
• The ElasticPress default language. It lives under the ep_language option name in the WordPress options database table.
• The mechanism used to run the synchronization: GUI or CLI.

Thanks in advance.