custom config SearchWP

custom.php

<?php

/*
// Add additional content to Custom Field during indexing in SearchWP.
add_filter( 'searchwp\source\post\attributes\meta', function( $meta_value, $args ) {
	//error_log($args['meta_key']);
  if ( 'searchwp_gtin' !== $args['meta_key'] ) {
    return $meta_value;
  }
  error_log('gtin');
  // Add 'coffee' to Custom Field value to make it searchable.
  return 'coffee';
}, 20, 2 );
*/
// Reduce SearchWP's batch size for token generation.
/*
add_filter( 'searchwp\index\tokens_max', function( $limit ) {
  return 10;
} );
*/

// Remove company name "ACME" from tokens in SearchWP.
add_filter( 'searchwp\tokens', function( $tokens ) {
	$newtokens = array();
		
	
	foreach ($tokens as $key => $token)
	{		
		if ( str_contains($token, " ") )
		{
			array_merge( $newtokens, explode(' ', $tokens));
		}
		else
		{
			$newtokens[] = $token;
		}
	}
	//$newtokens[0] = 'lantern';
	return $newtokens;
		
/*	
  return array_filter( $tokens, function( $token ) {
    return 'nan' !== $token;
  } );
*/
} );

// Programmatically enable partial matching in SearchWP.
//add_filter( 'searchwp\query\partial_matches', '__return_false' );

// make sure woocommerce doesnt redirect
add_filter( 'woocommerce_redirect_single_search_result', '__return_false' );

//
//
////add_filter( 'facetwp_use_search_relevancy', '__return_true' );
// Customize the results returned by SearchWP.
/*
add_filter( 'searchwp\query\results', function( $results, $query ) {
    // This is where you can manipulate the $results array of results.
    if (is_array($results))
	{
		//error_log(print_r($results,true));	
		if (count($results) == 1)
		{
			//$productid = array_pop($results);
			$product = wc_get_product( $results[0] );
			$permalink = $product->get_permalink();
			//error_log($permalink);
			//wp_redirect( $permalink );
			
		}
	}
    
    return $results;
}, 30, 2 );
*/
/*
add_filter('searchwp\query\partial_matches\fuzzy\threshold', function ( $threshold ) {
	return 100;
});
*/

// Disable SearchWP's fuzzy logic when performing partial matches.
//add_filter( 'searchwp\query\partial_matches\fuzzy', '__return_false' ); //avoid lots of 2 char strings on FZ-60EZ for example
//
// Prevent partial matching on the 'default' SearchWP engine only.
// NOTE: the engine name in this hook tag is: default
//add_filter( 'searchwp_partial_matching_default', '__return_false' );

// Disable keyword stemming for all Engines in SearchWP.
//add_filter( 'searchwp\query\tokens\use_stems', '__return_false' );

// Disable SearchWP's exact match buoy when performing partial match logic.
//add_filter( 'searchwp\query\partial_matches\buoy', '__return_true' );

// // &amp;
// Control SearchWP AND logic token threshold.
// @link https://searchwp.com/documentation/hooks/searchwp-query-logic-and-token_threshold/
add_filter( 'searchwp\query\logic\and\token_threshold', function( $threshold, $args ) {
	
	/*
	// Control SearchWP AND logic token threshold.
	// @link https://searchwp.com/documentation/hooks/searchwp-query-logic-and-token_threshold/
	add_filter( 'searchwp\query\logic\and\token_threshold', function( $threshold, $args ) {
	// If the search contains 'coffee' allow up to 10 tokens for AND logic.
	if ( in_array( 'coffee', $args['tokens'], true ) ) {
		$threshold = 10;
	}
      
	// If the search contains 'soccer' disable AND logic token threshold.
	if ( in_array( 'soccer', $args['tokens'], true ) ) {
		$threshold = false;
	}
	*/
	$threshold = 5;
  	return $threshold;
}, 20, 2 );
  
// Prevent specific Category from being indexed with posts in SearchWP.
// 

add_filter( 'searchwp\source\post\attributes\taxonomy\terms', function( $terms, $args ) {
  // Do not index the '_sap' term.
  if (!is_array($terms))
	  return $terms;
	
  return array_filter( $terms, function( $term ) {
	return false === stripos($term->name, '_sap');
    //return '_sap-tripod-photo' !== $term->name;
  } );
}, 20, 2 );

//Control whether exact matches are given extra relevance weight when performing partial match logic.
// // Enable SearchWP's exact match buoy when performing partial match logic.
//add_filter( 'searchwp\query\partial_matches\buoy', '__return_true' );

// Tell SearchWP to skip rule application for regex pattern matches.
//add_filter( 'searchwp\tokens\apply_rules_to_pattern_matches', '__return_false' );
//this generated way too many results for search term Nano4S


// Remove company name "ACME" from tokens in SearchWP.

add_filter( 'searchwp\tokens', function( $tokens ) {
	foreach ($tokens as $key => $token)
	{
		if ($token[0] == '-')	
		{
			//$tokens[$key] = str_replace('-','',$token);
			$tokens[$key] = ltrim($token, '-');
		}
		
		if ($token[strlen($token)-1] == '-')
		{
			//$tokens[$key] = str_replace('-','',$token);
			$tokens[$key] = rtrim($token, '-');
		}
		
	}

	return $tokens;
  
  //return array_filter( $tokens, function( $token ) {
  //  return 'ii' !== $token;
  //} );
  
} );



/*
// Force partial matches in SearchWP even when matches are found using provided search.
add_filter( 'searchwp\query\partial_matches\force', function( $force, $args ) {
	// BY ADDING MORE PATTERNS WE CAN SAFELY TURN THIS OFF!
	return false;
}, 10, 2 );
*/
// Tell SearchWP to use a wildcard prefix when performing partial match logic.
//add_filter( 'searchwp\query\partial_matches\wildcard_before', '__return_true' );

// Tell SearchWP to use a wildcard prefix when performing partial match logic.
add_filter( 'searchwp\query\partial_matches\wildcard_after', '__return_true' );


// Force SearchWP to use AND logic and not fall back to OR logic.
//add_filter( 'searchwp\query\logic\and\strict', '__return_true' );


// Enable regex pattern match tokenization in SearchWP.
//add_filter( 'searchwp\tokens\tokenize_pattern_matches', '__return_false' );

// Reduce SearchWP's minimum character length to 2 (default is 3).
add_filter( 'searchwp\tokens\minimum_length', function( $min ) {
	return 2;
} );

// Accept up to 12 search terms in SearchWP.
add_filter( 'searchwp\query\logic\and\token_threshold', function( $max, $query ) {
	return 8;
	//at 70 pavotube 30x doesnt return correct results
	//at 400 si-t returns a lot of lenses!
}, 30, 2 );

	
// Accept up to 12 search terms in SearchWP.
add_filter( 'searchwp\query\tokens\limit', function( $max, $query ) {
	return 100; 
	//at 70 pavotube 30x doesnt return correct results
	//at 400 si-t returns a lot of lenses!
}, 30, 2 );



// Allow partial regex pattern matches to be considered in SearchWP.
//add_filter( 'searchwp\tokens\regex_patterns\only_full_matches', '__return_true' );

add_filter( 'searchwp\tokens\regex_patterns', function( $patterns ){
	$my_patterns = array(	
		
		//"/(?!^[a-zA-Z]{2,3}[\-])([a-zA-Z0-9]{2,20})/UsA", // PT6CII
		//
		"/(coffee)/",
		"/(?!:^[a-zA-Z]{2,3})([-][a-zA-Z0-9\.\-\/]{2}?)/",
		"/(?!:^[a-zA-Z]{2,3})([-][a-zA-Z0-9\.\-\/]{3}?)/",
		"/(?!:^[a-zA-Z]{2,3})([-][a-zA-Z0-9\.\-\/]{4}?)/",
		"/(?!:^[a-zA-Z]{2,3})([-][a-zA-Z0-9\.\-\/]{5}?)/",
		"/(?!:^[a-zA-Z]{2,3})([-][a-zA-Z0-9\.\-\/]{6}?)/",
		"/(?!:^[a-zA-Z]{2,3})([-][a-zA-Z0-9\.\-\/]{7}?)/",
		"/(?!:^[a-zA-Z]{2,3})([-][a-zA-Z0-9\.\-\/]{8}?)/",
		"/(?!:^[a-zA-Z]{2,3})([-][a-zA-Z0-9\.\-\/]{15,30}?)/",
		
		"/(^[a-zA-Z]{2,3}[-])/U", // HO- NL-
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\.\-\/]{1})/", // HO-UV, NL-FZ
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\.\-\/]{2})/", // HO-UV, NL-FZ
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\.\-\/]{3})/", // HO-PLC
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\.\-\/]{4})/", // HO-PLC
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\.\-\/]{5})/", // HO-PLC
		"/^[a-zA-Z]{2,3}[-]([a-zA-Z0-9\.\-\/]{6})/", // HO-PLC
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\.\-\/]{7})/", // HO-PLC
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\.\-\/]{8})/", // HO-PLC
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\.\-\/]{15,30})/", // HO-PLC
		
		//"/(?!^[a-zA-Z]{2,3}[-])([a-zA-Z0-9]{2,20})/", // HO- NL-
		
		"/(^[a-zA-Z]{3})/", // DZO
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z]{2,4}[0-9]{1,2})/", // SI-VA5
		"/([a-zA-Z]{2,4}[0-9]{1,2})/", // VA5
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z0-9\-]{4,20})/", // HO-IRND1000P82, NL-FZ
		"/([a-zA-Z]{1,2})([0-9]{1,4})/", //FZ720		
		
		"/(?<=[0-9].)([a-zA-Z]{2,6}\b)/", // 2-4 chars if lookahead is number and end of word
		"/([0-9]{1,2}\/[0-9]{1,2})/", // 1/4 3/8 etc
		//"/(^[a-zA-Z]{2}[-][a-zA-Z]{6})(?=[0-9].)/", // full code if followed by number // with lookahead
		//"/(^[a-zA-Z]{2}[-][a-zA-Z]{5})(?=[0-9].)/", // full code if followed by number // with lookahead
		//"/(^[a-zA-Z]{2}[-][a-zA-Z]{4})(?=[0-9].)/", // full code if followed by number // with lookahead
		//"/(^[a-zA-Z]{2}[-][a-zA-Z]{3})(?=[0-9].)/", // full code if followed by number // with lookahead
		//"/(^[a-zA-Z]{2}[-][a-zA-Z]{2})(?=[0-9].)/", // full code if followed by number // with lookahead
		//"/(^[a-zA-Z]{2}[-][a-zA-Z]{1})(?=[0-9].)/", // full code if followed by number // with lookahead
		"/([a-zA-Z]{2,4}[0-9]{1}[a-zA-Z]{1})/", //NANOxS / VA5X
		"/([a-zA-Z]{2,4}[0-9]{1})/", //NANOx (VA5 WERKT HIER NIET!?)
		"/(^[a-zA-Z]{2,3}[-][a-zA-Z]{1,2}[0-9]{2,4})/", // SI-T1004 / 
		"/([a-zA-Z]{1}[0-9]{4})/", //T1004 N2004 
		"/([0-9]{2,4})/", //324 / 1004 / 30
		"/([0-9]{2,4}[a-zA-Z]{2,3})/", //300BII / 300II / 60BII
		"/([0-9]{2,4}[a-zA-Z]{1,1})/", //30C
		//"(\b[a-z]{1,2}[0-9]{1,4})", //FZ500 - NOT WORKING
		//"([a-z]{1,2}[0-9]{1,4})", //FZ500 - NOT WORKING
		"/([a-zA-Z]{1,2})([0-9]{1,4})([a-zA-Z]{1,3})/", //FZ500BII
		//"/(?<![a-zA-Z]{1,2})([0-9]{1,4})/" //FZ720
		#"([0-9]{1,}[A-Za-z]{1,})" #/iu  #for 600w etc
		
	);
	
	//unset($patterns[9]);
	/*
	foreach ($patterns as $key => $value)
	{
    	if ($value == '/([A-Za-z0-9]{1,}\s[0-9]{1,}[A-Za-z0-9]*)/iu'
		   		||
		   	$value == '(?=\S*[\-\_])([[:alnum:]\-\_]+)') 
		{
        	unset($patterns[$key]);
    	}
	}
	*/
	//unset($patterm)
	//print_r($patterns,false);
	// we want our pattern to be considered the most specific
	// so that false positive matches do not interfere
	return array_merge( $my_patterns, $patterns );
	//return $my_patterns;
	
	//return array_merge( $my_patterns );
} );



add_filter( 'searchwp\valid_html_tags', function( $tags ) {
	if( isset( $tags['img'] ) ){
		//Remove whole image tags from the index
		unset( $tags['img'] );
	}
	return $tags;
} );



// Exclude CPL / PLC group filters
// Exclude products that are part of a grouped product

add_filter( 'searchwp\post__not_in', function( $ids ) {
	global $wpdb;


	$products = $wpdb->get_results( "
		SELECT DISTINCT
			pm.meta_value as child_ids
		FROM
			wp_tse_posts AS post
		INNER JOIN wp_tse_postmeta AS pm ON post.ID = pm.post_id AND pm.meta_key = '_children'
		INNER JOIN wp_tse_postmeta AS pm2 ON post.ID = pm2.post_id AND pm2.meta_key = '_sku'		
		WHERE
			post.post_type = 'product'		
			AND pm.meta_value > ''
			AND pm2.meta_value LIKE 'HO-%%'
		ORDER BY
			post.id;
	");
	
	$children = [];
	
	foreach( $products as $product ){
		$ids = maybe_unserialize( $product->child_ids );
		if (count($ids) > 1)
			$children = array_merge( $children, $ids );
    }
	return array_merge( $ids, array_merge( $children , [ 2717 , 2368 ] ) );

}, 900, 1 );


// Add WooCommerce Product (and Variation) SKUs to SearchWP.
// @link https://searchwp.com/documentation/knowledge-base/search-woocommerce-skus-and-variation-skus/
add_filter( 'searchwp\entry\data', function( $data, \SearchWP\Entry $entry ) {

	// If this is not a Product, there's nothing to do.
	if ( 'product' !== get_post_type( $entry->get_id() ) ) {
		return $data;
	}

	$my_extra_meta_key = 'searchwp_skus';
	$my_extra_ean_key = 'searchwp_gtin';
	
	// Retrieve this Product SKU.
	$data['meta'][ $my_extra_meta_key ] = [
		get_post_meta( $entry->get_id(), '_sku', true )		
	];

	$data['meta'][ $my_extra_ean_key ] = [
		get_post_meta( $entry->get_id(), '_gtin', true )		
	];
		
	// Retrieve all grouped product for this parent.
	$product = wc_get_product( $entry->get_id() );
	$grouped_products = $product->get_children();

	foreach ( $grouped_products as $grouped_product ) {
		$sku = get_post_meta( $grouped_product, '_sku', true );
		$gtin = get_post_meta( $grouped_product, '_gtin', true );
		$data['meta'][ $my_extra_meta_key ][] = $sku;
		$data['meta'][ $my_extra_ean_key ][] = $gtin;
	}

	return $data;

}, 20, 2 );


// Add our Extra Meta entry to SearchWP's UI.
// @link https://searchwp.com/documentation/knowledge-base/search-woocommerce-skus-and-variation-skus/
add_filter( 'searchwp\source\attribute\options', function( $keys, $args ) {
	if ( $args['attribute'] !== 'meta' ) {
		return $keys;
	}

	// This key is the same as the one used in the searchwp\entry\data hook above, they must be the same.
	$my_extra_meta_key = 'searchwp_skus';
	$option = new \SearchWP\Option( $my_extra_meta_key, 'SearchWP WooCommerce SKUs' );

	// If there's already a match, remove it because we want ours there.
	$keys = array_filter( $keys, function( $option ) use ( $my_extra_meta_key ) {
		return $my_extra_meta_key !== $option->get_value();
	} );
	
	$keys[] = $option;
	
	$my_extra_ean_key = 'searchwp_gtin';
	$option = new \SearchWP\Option( $my_extra_ean_key, 'SearchWP WooCommerce EANs' );	
	
	// If there's already a match, remove it because we want ours there.
	$keys = array_filter( $keys, function( $option ) use ( $my_extra_ean_key ) {
		return $my_extra_ean_key !== $option->get_value();
	} );

	// Add "SearchWP WooCommerce SKUs" Option
	$keys[] = $option;

	return $keys;
}, 20, 2 );

/*
function my_searchwp_term_pattern_whitelist( $whitelist ) {
	$my_whitelist = array(
			"/\\b(IT)\\b/u", // always keep "IT" (all caps only!)
		);

	// we want our pattern to be considered the most specific 
	// so that false positive matches do not interfere
	$whitelist = array_merge( $my_whitelist, $whitelist );

	return $whitelist;
}
*/
//add_filter( 'searchwp_term_pattern_whitelist', 'my_searchwp_term_pattern_whitelist' );


/*
	$product_query = new WC_Product_Query([
		'type' => ['grouped'],
		'limit' => -1
	]);
	$grouped_products = $product_query->get_products();
*/


	// The SQL query
	// 
	/*
    $products = $wpdb->get_results( "
		SELECT  pm.meta_value as child_ids, pm.post_id
        FROM {$wpdb->prefix}postmeta as pm
		INNER JOIN {$wpdb->prefix}posts as p ON pm.post_id = p.ID
        INNER JOIN {$wpdb->prefix}term_relationships as tr ON pm.post_id = tr.object_id
        INNER JOIN {$wpdb->prefix}terms as t ON tr.term_taxonomy_id = t.term_id		
        WHERE p.post_type LIKE 'product'
        AND p.post_status LIKE 'publish'
        AND t.slug LIKE 'grouped'
        AND pm.meta_key LIKE '_children'		
		AND pm.meta_value NOT LIKE '%%63250%%'		
        ORDER BY p.ID
	" );
	*/
	
/*
		SELECT  pm.meta_value as child_ids, pm.post_id
        FROM {$wpdb->prefix}postmeta as pm
		INNER JOIN {$wpdb->prefix}posts as p ON pm.post_id = p.ID
        INNER JOIN {$wpdb->prefix}term_relationships as tr ON pm.post_id = tr.object_id
        INNER JOIN {$wpdb->prefix}terms as t ON tr.term_taxonomy_id = t.term_id
		INNER JOIN {$wpdb->prefix}postmeta as pm2 ON pm2.post_id = p.ID AND pm2.meta_key = '_sku'
        WHERE p.post_type LIKE 'product'
        AND p.post_status LIKE 'publish'
        AND t.slug LIKE 'grouped'
        AND pm.meta_key LIKE '_children'
		AND pm2.meta_value LIKE 'HO-%%'
        ORDER BY p.ID
*/

	
		/*
        foreach( maybe_unserialize( $result->child_ids ) as $child_id )
            if( $child_id == $post_id ){
                $parent_grouped_id = $result->post_id;
                break;
            }
        if( $parent_grouped_id != 0 ) break;
		*/


	//AND t.slug LIKE 'grouped'
	//AND p.post_type = 'product_grouped'


	/*
	foreach ($children as $key => $child)
	{
		$product = wc_get_product( $child );
		if ( substr_compare($product->get_sku, "HO-", 0) !== 0 )
		{
			unset($children[$key]);
		}
	}
	*/
	/*
	foreach ($products as $productid)
	{
		$product = wc_get_product( $productid );
		//$children = array_merge( $children, $product->get_children() );
	}
	*/
	/*
	foreach( $grouped_products as $grouped_product ){
		$children = array_merge( $children, $grouped_product->get_children() );
	}
	*/
	// Search value and delete
	//if(($key = array_search(63250, $children)) !== false) {
	//    unset($children[$key]);
	//}
?>