Polylang compatibility

Alas, there is no way to use the locale filter… the news sitemap does not use a query loop so I cannot detect which post (URL) is being treated. So no way to determine with which language to overwrite the locale…

So… any way RankMath would consider adding a useful language string filter somewhere?

Excellent news! Thank you 🙂

Testing the new filter in the beta version, I fear I’m no succeeding to make it work.

I see the new filter in class-news-sitemap.php:

$lang = $this->do_filter( 'sitemap/news/language', $lang, $entity );

But even a crude filter like this does absolutely nothing:

add_filter( 'sitemap/news/language', function ( $lang, $entity ) {
    error_log( 'filtering' );
    return 'xx';
}, 10, 2 );

Nothing in debug.log, no ‘xx’ news language XML node in the news sitemap. It’s as if the hook is ignored and the filter is not even run… Very strange.

Note: I have set add_filter( 'rank_math/sitemap/enable_caching', '__return_false' ); at the same time.

Ah… now I see my mistake: I forgot to prepend the hook slug with rank_math/. Thanks for the reply!

UPDATE: it appears the filter hook is useless… I’m currently working with this filter:

add_filter(
    'rank_math/sitemap/news/language',
    function ( $lang, $entity ) {
        $post_id = url_to_postid( $entity['loc'] );
        if ( $post_id ) {
            $lang = pll_get_post_language( $post_id );
        }
        return $lang;
    },
    10,
    2
);

And it does return the correct language but… only for the first post in the news sitemap!

I found these lines at the start of the method get_news_publication() in class-news-sitemap.php and commented them out:

if ( ! is_null( $this->news_publication ) ) {
    return $this->news_publication;
}

After doing that, the news post language starts reflecting individual post languages correctly…

So another feature request: could the static news publication nodes caching be dependant of the rank_math/sitemap/enable_caching filter as well?

Hi,

With the news source name I mean the news:name node inside the news:publication node. For example:

<url>
	<loc>https://ontherhone.com/fr/naviguer-traversee-de-lyon-confluence-saone/</loc>
	<news:news>
		<news:publication>
			<news:name>Au fil du fleuve Rhône</news:name>
			<news:language>fr</news:language>
		</news:publication>
		<news:publication_date>2024-04-28T19:07:40+00:00</news:publication_date>
		<news:title><![CDATA[Naviguer sur le Rhône : traversée de Lyon et confluence de la Saône]]></news:title>
	</news:news>
</url>

About the filter: I fear there is a misunderstanding.

What I’m saying is that the filter, inside the class get_news_publication() method, is only run ONCE per news sitemap request. This happens because the news publication node is saved to the class $news_publication property, on the first URL node, then on all subsequent URL nodes, the stored property value is returned, skipping the filter from then on…

This means that in a news sitemap with multiple news items in different languages, ALL publication nodes will be the same as the first URL node’s.

Please take a slightly adapted version of your example code, to see it happen:

add_filter( 'rank_math/sitemap/news/language', function( $lang, $entity ) { static $count = 1; echo "<pre>"; print_r($count); echo "</pre>"; $count++; return $lang; }, 10, 2 );

You’ll notice I added a static counter, which should increment each time the filter is run, but on a live news sitemap with multiple entries, it will repeat <pre>1</pre> for each node.

Hope that explains it more clearly 🙂

For example, you won’t see Search Engine Land publication being translated into other languages.

A good example. But on the other hand, it is Google themselves that advise to create distinct publication sources in Publisher Center for each language (whether on a different domain, subdomain or subdirectory) and that does allow distinct publication names… So I’m not so sure that it is strictly not allowed, or just not common (yet).

Anyway, that is of lesser importance than the rank_math/sitemap/news/language filter:

I fear I’m still am not getting my point across…

Your example code is for WPML, and I have a similar routine that works for Polylang ready. But…

Please test your own example filter above on a site where there are multiple (recent news) posts in multiple languages. You will notice that ALL posts in the news sitemap will have the same language as the first post, irrespective of their true language.

You see these lines at the start of the method get_news_publication() in class-news-sitemap.php?

if ( ! is_null( $this->news_publication ) ) {
    return $this->news_publication;
}

This makes it so that the filter (applied after these lines) is only occurring ONCE per news sitemap request, not on each URL node. So there will be only ONE language mentioned in the news sitemap.

But, at least with Polylang when languages are set to subdirectories, there will be multiple language posts in the same news sitemap. I would suspect this is the same for WPML but I’m not

This means: the filter is useless for Polylang with languages not set to domain or subdomain 🙁

Unless… I could ask for this addition (or something similar) :

if ( false !== apply_filters( 'rank_math/sitemap/enable_caching' ) && ! is_null( $this->news_publication ) ) {
    return $this->news_publication;
}