Wordpress search issue with searching html tags

Question

I am working on a wordpress site and I am using wordpress native search. When we search for a string like "example" it searches inside html tags. ie if there is any html tag like <a href="http://example.com"> in the page or post then the search result will show that particular page or post. I dont want to search inside html tags.

I already tried using the plugin Search Exclude HTML Tags But it uses MySql stored function which is not working for me. I think it may be issue with the permission of my mysql user.

Is there any alternate way to fix this.

Thanks Pramod

Answer 1

You could try using several regexps to only search outside angle brackets:

function wpse159789_posts_search( $search, $query ) {
    global $wpdb;
    if ( ! preg_match( '/' . $wpdb->posts . '\.post_content LIKE \'%(.+)%\'/', $search, $matches, PREG_OFFSET_CAPTURE ) ) {
        return $search;
    }
    $search_str = stripslashes( $matches[1][0] );
    // Cater for closed angle pairs embedded in the search string.
    for ( $i = 0, $len = mb_strlen( $search_str ); $i < $len; $i++ ) {
        $q_searches[] = '(<[^>]*>)?' . preg_quote( mb_substr( $search_str, $i, 1 ) );
    }
    $q_search = implode( '', $q_searches );
    $regexs[] = '^[^<]*' . $q_search; // Before any angle bracket.
    $regexs[] = '(<[^>]*>)[^<]*' . $q_search; // After any closed angle bracket pair.
    array_unshift( $regexs, implode( ' OR ', array_fill( 0, count( $regexs ), $wpdb->posts . '.post_content RLIKE %s' ) ) );
    $search_replace = call_user_func_array( array( $wpdb, 'prepare' ), $regexs );
    $search = substr( $search, 0, $matches[0][1] ) . $search_replace . substr( $search, $matches[0][1] + strlen( $matches[0][0] ) );
    return $search;
}

I'm not sure this covers all cases, but seems to work on limited testing ... (it'd be a lot simpler if MySQL regexp supported look behind/ahead).

Answer 2

I created a tiny plugin for this :) and fixed the issue.

https://wordpress.org/plugins/wp-search-ignore-html-tags/

Just one or 2 lines of code. Worked in most cases.