+global\Jetpack_Search::convert_wp_es_to_es_args()

Complexity

Conditions	36
Paths	> 20000

Size

Total Lines

347

Duplication

Lines	32
Ratio	9.22 %

Importance

Changes

0

<?php
	public function convert_wp_es_to_es_args( array $args ) {
		jetpack_require_lib( 'jetpack-wpes-query-builder/jetpack-wpes-query-parser' );

		$defaults = array(
			'blog_id'        => get_current_blog_id(),
			'query'          => null,    // Search phrase
			'query_fields'   => array(), // list of fields to search
			'excess_boost'   => array(), // map of field to excess boost values (multiply)
			'post_type'      => null,    // string or an array
			'terms'          => array(), // ex: array( 'taxonomy-1' => array( 'slug' ), 'taxonomy-2' => array( 'slug-a', 'slug-b' ) )
			'author'         => null,    // id or an array of ids
			'author_name'    => array(), // string or an array
			'date_range'     => null,    // array( 'field' => 'date', 'gt' => 'YYYY-MM-dd', 'lte' => 'YYYY-MM-dd' ); date formats: 'YYYY-MM-dd' or 'YYYY-MM-dd HH:MM:SS'
			'orderby'        => null,    // Defaults to 'relevance' if query is set, otherwise 'date'. Pass an array for multiple orders.
			'order'          => 'DESC',
			'posts_per_page' => 10,
			'offset'         => null,
			'paged'          => null,
			/**
			 * Aggregations. Examples:
			 * array(
			 *     'Tag'       => array( 'type' => 'taxonomy', 'taxonomy' => 'post_tag', 'count' => 10 ) ),
			 *     'Post Type' => array( 'type' => 'post_type', 'count' => 10 ) ),
			 * );
			 */
			'aggregations'   => null,
		);

		$args = wp_parse_args( $args, $defaults );

		$parser = new Jetpack_WPES_Search_Query_Parser( $args['query'], array( get_locale() ) );

		if ( empty( $args['query_fields'] ) ) {
			if ( defined( 'JETPACK_SEARCH_VIP_INDEX' ) && JETPACK_SEARCH_VIP_INDEX ) {
				// VIP indices do not have per language fields
				$match_fields = $this->_get_caret_boosted_fields(
					array(
						'title'         => 0.1,
						'content'       => 0.1,
						'excerpt'       => 0.1,
						'tag.name'      => 0.1,
						'category.name' => 0.1,
						'author_login'  => 0.1,
						'author'        => 0.1,
					)
				);

				$boost_fields = $this->_get_caret_boosted_fields(
					$this->_apply_boosts_multiplier( array(
						'title'         => 2,
						'tag.name'      => 1,
						'category.name' => 1,
						'author_login'  => 1,
						'author'        => 1,
					), $args['excess_boost'] )
				);

				$boost_phrase_fields = $this->_get_caret_boosted_fields(
					array(
						'title'         => 1,
						'content'       => 1,
						'excerpt'       => 1,
						'tag.name'      => 1,
						'category.name' => 1,
						'author'        => 1,
					)
				);
			} else {
				$match_fields = $parser->merge_ml_fields(
					array(
						'title'         => 0.1,
						'content'       => 0.1,
						'excerpt'       => 0.1,
						'tag.name'      => 0.1,
						'category.name' => 0.1,
					),
					$this->_get_caret_boosted_fields( array(
						'author_login'  => 0.1,
						'author'        => 0.1,
					) )
				);

				$boost_fields = $parser->merge_ml_fields(
					$this->_apply_boosts_multiplier( array(
						'title'         => 2,
						'tag.name'      => 1,
						'category.name' => 1,
					), $args['excess_boost'] ),
					$this->_get_caret_boosted_fields( $this->_apply_boosts_multiplier( array(
						'author_login'  => 1,
						'author'        => 1,
					), $args['excess_boost'] ) )
				);

				$boost_phrase_fields = $parser->merge_ml_fields(
					array(
						'title'         => 1,
						'content'       => 1,
						'excerpt'       => 1,
						'tag.name'      => 1,
						'category.name' => 1,
					),
					$this->_get_caret_boosted_fields( array(
						'author'        => 1,
					) )
				);
			}
		} else {
			// If code is overriding the fields, then use that. Important for backwards compatibility.
			$match_fields        = $args['query_fields'];
			$boost_phrase_fields = $match_fields;
			$boost_fields        = null;
		}

		$parser->phrase_filter( array(
			'must_query_fields'  => $match_fields,
			'boost_query_fields' => null,
		) );
		$parser->remaining_query( array(
			'must_query_fields'  => $match_fields,
			'boost_query_fields' => $boost_fields,
		) );

		// Boost on phrase matches
		$parser->remaining_query( array(
			'boost_query_fields' => $boost_phrase_fields,
			'boost_query_type'   => 'phrase',
		) );

		/**
		 * Modify the recency decay parameters for the search query.
		 *
		 * The recency decay lowers the search scores based on the age of a post relative to an origin date. Basic adjustments:
		 *  - origin: A date. Posts with this date will have the highest score and no decay applied. Default is today.
		 *  - offset: Number of days/months/years (eg 30d). All posts within this time range of the origin (before and after) will have no decay applied. Default is no offset.
		 *  - scale: The number of days/months/years from the origin+offset at which the decay will equal the decay param. Default 360d
		 *  - decay: The amount of decay applied at offset+scale. Default 0.9.
		 *
		 * The curve applied is a Gaussian. More details available at {@see https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-function-score-query.html#function-decay}
		 *
		 * @module search
		 *
		 * @since  5.8.0
		 *
		 * @param array $decay_params The decay parameters.
		 * @param array $args         The WP query parameters.
		 */
		$decay_params = apply_filters(
			'jetpack_search_recency_score_decay',
			array(
				'origin' => date( 'Y-m-d' ),
				'scale'  => '360d',
				'decay'  => 0.9,
			),
			$args
		);

		if ( ! empty( $decay_params ) ) {
			// Newer content gets weighted slightly higher
			$parser->add_decay( 'gauss', array(
				'date_gmt' => $decay_params
			) );
		}

		$es_query_args = array(
			'blog_id' => absint( $args['blog_id'] ),
			'size'    => absint( $args['posts_per_page'] ),
		);

		// ES "from" arg (offset)
		if ( $args['offset'] ) {
			$es_query_args['from'] = absint( $args['offset'] );
		} elseif ( $args['paged'] ) {
			$es_query_args['from'] = max( 0, ( absint( $args['paged'] ) - 1 ) * $es_query_args['size'] );
		}

		$es_query_args['from'] = min( $es_query_args['from'], Jetpack_Search_Helpers::get_max_offset() );

		if ( ! is_array( $args['author_name'] ) ) {
			$args['author_name'] = array( $args['author_name'] );
		}

		// ES stores usernames, not IDs, so transform
		if ( ! empty( $args['author'] ) ) {
			if ( ! is_array( $args['author'] ) ) {
				$args['author'] = array( $args['author'] );
			}

			foreach ( $args['author'] as $author ) {
				$user = get_user_by( 'id', $author );

				if ( $user && ! empty( $user->user_login ) ) {
					$args['author_name'][] = $user->user_login;
				}
			}
		}

		//////////////////////////////////////////////////
		// Build the filters from the query elements.
		// Filters rock because they are cached from one query to the next
		// but they are cached as individual filters, rather than all combined together.
		// May get performance boost by also caching the top level boolean filter too.

		if ( $args['post_type'] ) {
			if ( ! is_array( $args['post_type'] ) ) {
				$args['post_type'] = array( $args['post_type'] );
			}

			$parser->add_filter( array(
				'terms' => array(
					'post_type' => $args['post_type'],
				),
			) );
		}

		if ( $args['author_name'] ) {
			$parser->add_filter( array(
				'terms' => array(
					'author_login' => $args['author_name'],
				),
			) );
		}

		if ( ! empty( $args['date_range'] ) && isset( $args['date_range']['field'] ) ) {
			$field = $args['date_range']['field'];

			unset( $args['date_range']['field'] );

			$parser->add_filter( array(
				'range' => array(
					$field => $args['date_range'],
				),
			) );
		}

		if ( is_array( $args['terms'] ) ) {
			foreach ( $args['terms'] as $tax => $terms ) {
				$terms = (array) $terms;

				if ( count( $terms ) && mb_strlen( $tax ) ) {
					switch ( $tax ) {
						case 'post_tag':
							$tax_fld = 'tag.slug';

							break;

						case 'category':
							$tax_fld = 'category.slug';

							break;

						default:
							$tax_fld = 'taxonomy.' . $tax . '.slug';

							break;
					}

					foreach ( $terms as $term ) {
						$parser->add_filter( array(
							'term' => array(
								$tax_fld => $term,
							),
						) );
					}
				}
			}
		}

		if ( ! $args['orderby'] ) {
			if ( $args['query'] ) {
				$args['orderby'] = array( 'relevance' );
			} else {
				$args['orderby'] = array( 'date' );
			}
		}

		// Validate the "order" field
		switch ( strtolower( $args['order'] ) ) {
			case 'asc':
				$args['order'] = 'asc';
				break;

			case 'desc':
			default:
				$args['order'] = 'desc';
				break;
		}

		$es_query_args['sort'] = array();

		foreach ( (array) $args['orderby'] as $orderby ) {
			// Translate orderby from WP field to ES field
			switch ( $orderby ) {
				case 'relevance' :
					//never order by score ascending
					$es_query_args['sort'][] = array(
						'_score' => array(
							'order' => 'desc',
						),
					);

					break;

				case 'date' :
					$es_query_args['sort'][] = array(
						'date' => array(
							'order' => $args['order'],
						),
					);

					break;

				case 'ID' :
					$es_query_args['sort'][] = array(
						'id' => array(
							'order' => $args['order'],
						),
					);

					break;

				case 'author' :
					$es_query_args['sort'][] = array(
						'author.raw' => array(
							'order' => $args['order'],
						),
					);

					break;
			} // End switch().
		} // End foreach().

		if ( empty( $es_query_args['sort'] ) ) {
			unset( $es_query_args['sort'] );
		}

		// Aggregations
		if ( ! empty( $args['aggregations'] ) ) {
			$this->add_aggregations_to_es_query_builder( $args['aggregations'], $parser );

$args['aggregations'] is of type string, but the function expects a array.

		}

		$es_query_args['filter']       = $parser->build_filter();
		$es_query_args['query']        = $parser->build_query();
		$es_query_args['aggregations'] = $parser->build_aggregation();

		return $es_query_args;
	}