WP.com related posts query building

gistfile1.txt

	public static function post( $blog_id, $post_id, $mlt_fields = array( 'mlt_content' ), $analyzer = false ) {
		switch_to_blog( $blog_id );
		$post = get_post( $post_id );

		$fld_bldr = new ES_WP_Field_Builder();

		$tax_data = $fld_bldr->taxonomy( $post );
		$mlt_content = $fld_bldr->mlt_content( array(
			'title' => $fld_bldr->clean_string( $post->post_title ),
			'content' => $fld_bldr->clean_string( $post->post_content ),
			'excerpt' => $fld_bldr->clean_string( $post->post_excerpt ),
			'tags' => isset( $tax_data['tag'] ) ? wp_list_pluck( $tax_data['tag'], 'name' ) : array(),
			'cats' => isset( $tax_data['category'] ) ? wp_list_pluck( $tax_data['category'], 'name' ) : array(),
			'max_bytes' => 1000, //~200 English words
		) );
		restore_current_blog();

		if ( ! preg_match( '/\p{L}/', $mlt_content ) )
			return false; //no utf-8 letters, we can't detect anything

		$query = array();
		//For short content (< 25 words), just use a match query
		if ( strlen( $mlt_content ) < 125 ) {
			$query = array(
				'multi_match' => array(
					'query' =>  $mlt_content,
					'fields' => $mlt_fields,
				)
			);
			if ( $analyzer )
				$query['multi_match']['analyzer'] = $analyzer;
		} else {
			$query =  array(
				'more_like_this' => array_filter(
					array(
						'fields' => $mlt_fields,
						'like_text' => $mlt_content,
						'percent_terms_to_match' => 0.08, // Default .3, lower to match 2 terms out of 25 max_query_terms
						'min_term_freq' => null, // Default 2
						'max_query_terms' => null, // Default 25
						'stop_words' => null,
						'min_doc_freq' => null, // Default 5
						'max_doc_freq' => null, // Default inf
						'min_word_len' => null, // Default 0
						'max_word_len' => null, // Default inf
						'boost_terms' => 5, // Default 1 (This is the max boost value per term, actual boost values for each term is porportional to how often it appears in docs.)
					)
				)
			);
			if ( $analyzer )
				$query['more_like_this']['analyzer'] = $analyzer;
		}
		return $query;
	}