Strip the language prefix from Search Console URLs before matching

strip-the-language-prefix-from-search-console-urls-before-matching.php

<?php
/**
 * SEOPress PRO — Strip the language prefix from Search Console URLs before matching.
 *
 * Multilingual plugins (Polylang, WPML, TranslatePress…) prepend a language segment
 * (e.g. /en/, /fr/) to public URLs but do not register that segment in WP_Rewrite.
 * As a result, url_to_postid() cannot match the raw URL returned by Google Search
 * Console. This snippet removes a known list of language prefixes so SEOPress can
 * resolve the WordPress post and store GSC impressions / clicks / CTR / position
 * against it.
 *
 * Hook: seopress_search_console_match_url
 * Plugin: SEOPress PRO 9.9+
 */

add_filter( 'seopress_search_console_match_url', 'sp_search_console_match_url', 10, 2 );

function sp_search_console_match_url( $clean_url, $url ) {
	// Adjust to the languages enabled on your site.
	$language_prefixes = array( 'en', 'fr', 'es', 'de', 'it' );

	$parts = wp_parse_url( $clean_url );
	if ( empty( $parts['path'] ) ) {
		return $clean_url;
	}

	// Detect /xx/ at the very start of the path.
	if ( preg_match( '#^/(' . implode( '|', $language_prefixes ) . ')(/|$)#', $parts['path'], $matches ) ) {
		$parts['path'] = substr( $parts['path'], strlen( $matches[1] ) + 1 );
		if ( '' === $parts['path'] ) {
			$parts['path'] = '/';
		}

		$clean_url = $parts['scheme'] . '://' . $parts['host'] . $parts['path'];
	}

	return $clean_url;
}