| @@ -7,9 +7,9 @@ | ||
| 7 | 7 | */ |
| 8 | 8 | |
| 9 | 9 | namespace ElasticPress\Indexable\Comment; |
| 10 | 10 | |
| 11 | -use \WP_Comment_Query; | |
| 11 | +use WP_Comment_Query; | |
| 12 | 12 | use ElasticPress\Elasticsearch; |
| 13 | 13 | use ElasticPress\Features; |
| 14 | 14 | use ElasticPress\Indexable; |
| 15 | 15 | use ElasticPress\Indexable\Post\DateQuery; |
| @@ -32,18 +32,15 @@ | ||
| 32 | 32 | */ |
| 33 | 33 | public $slug = 'comment'; |
| 34 | 34 | |
| 35 | 35 | /** |
| 36 | - * Create indexable and initialize dependencies | |
| 36 | + * Flag to indicate if the indexable has support for | |
| 37 | + * `id_range` pagination method during a sync. | |
| 37 | 38 | * |
| 38 | - * @since 3.6.0 | |
| 39 | + * @var boolean | |
| 40 | + * @since 5.2.0 | |
| 39 | 41 | */ |
| 40 | - public function __construct() { | |
| 41 | - $this->labels = [ | |
| 42 | - 'plural' => esc_html__( 'Comments', 'elasticpress' ), | |
| 43 | - 'singular' => esc_html__( 'Comment', 'elasticpress' ), | |
| 44 | - ]; | |
| 45 | - } | |
| 42 | + public $support_indexing_advanced_pagination = true; | |
| 46 | 43 | |
| 47 | 44 | /** |
| 48 | 45 | * Instantiate the indexable SyncManager and QueryIntegration, the main responsibles for the WP integration. |
| 49 | 46 | * |
| @@ -50,8 +47,13 @@ | ||
| 50 | 47 | * @since 4.5.0 |
| 51 | 48 | * @return void |
| 52 | 49 | */ |
| 53 | 50 | public function setup() { |
| 51 | + $this->labels = [ | |
| 52 | + 'plural' => esc_html__( 'Comments', 'elasticpress' ), | |
| 53 | + 'singular' => esc_html__( 'Comment', 'elasticpress' ), | |
| 54 | + ]; | |
| 55 | + | |
| 54 | 56 | $this->sync_manager = new SyncManager( $this->slug ); |
| 55 | 57 | $this->query_integration = new QueryIntegration(); |
| 56 | 58 | } |
| 57 | 59 | |
| @@ -100,9 +102,9 @@ | ||
| 100 | 102 | |
| 101 | 103 | /** |
| 102 | 104 | * Support `paged` query var |
| 103 | 105 | * |
| 104 | - * If `offset` is used, that takes precendence | |
| 106 | + * If `offset` is used, that takes precedence | |
| 105 | 107 | * over this. |
| 106 | 108 | */ |
| 107 | 109 | if ( isset( $query_vars['paged'] ) && empty( $query_vars['offset'] ) && $query_vars['paged'] > 1 ) { |
| 108 | 110 | $formatted_args['from'] = $number * ( $query_vars['paged'] - 1 ); |
| @@ -702,14 +704,12 @@ | ||
| 702 | 704 | } |
| 703 | 705 | |
| 704 | 706 | $es_version = (string) $es_version; |
| 705 | 707 | |
| 706 | - $mapping_file = 'initial.php'; | |
| 708 | + $mapping_file = '7-0.php'; | |
| 707 | 709 | |
| 708 | - if ( version_compare( $es_version, '5.0', '<' ) ) { | |
| 709 | - $mapping_file = 'pre-5-0.php'; | |
| 710 | - } elseif ( version_compare( $es_version, '7.0', '>=' ) ) { | |
| 711 | - $mapping_file = '7-0.php'; | |
| 710 | + if ( version_compare( $es_version, '7.0', '<' ) ) { | |
| 711 | + $mapping_file = 'initial.php'; | |
| 712 | 712 | } |
| 713 | 713 | |
| 714 | 714 | /** |
| 715 | 715 | * Filter comment indexable mapping file |
| @@ -785,18 +785,19 @@ | ||
| 785 | 785 | * @since 3.6.0 |
| 786 | 786 | * @return array |
| 787 | 787 | */ |
| 788 | 788 | public function query_db( $args ) { |
| 789 | - | |
| 790 | 789 | $defaults = [ |
| 791 | - 'type' => $this->get_indexable_comment_types(), | |
| 792 | - 'status' => $this->get_indexable_comment_status(), | |
| 793 | - 'post_type' => Indexables::factory()->get( 'post' )->get_indexable_post_types(), | |
| 794 | - 'post_status' => Indexables::factory()->get( 'post' )->get_indexable_post_status(), | |
| 795 | - 'number' => $this->get_bulk_items_per_page(), | |
| 796 | - 'offset' => 0, | |
| 797 | - 'orderby' => 'comment_ID', | |
| 798 | - 'order' => 'desc', | |
| 790 | + 'type' => $this->get_indexable_comment_types(), | |
| 791 | + 'status' => $this->get_indexable_comment_status(), | |
| 792 | + 'post_type' => Indexables::factory()->get( 'post' )->get_indexable_post_types(), | |
| 793 | + 'post_status' => Indexables::factory()->get( 'post' )->get_indexable_post_status(), | |
| 794 | + 'number' => $this->get_bulk_items_per_page(), | |
| 795 | + 'offset' => 0, | |
| 796 | + 'orderby' => 'comment_ID', | |
| 797 | + 'order' => 'desc', | |
| 798 | + 'ep_indexing_advanced_pagination' => true, | |
| 799 | + 'no_found_rows' => false, | |
| 799 | 800 | ]; |
| 800 | 801 | |
| 801 | 802 | if ( isset( $args['per_page'] ) ) { |
| 802 | 803 | $args['number'] = $args['per_page']; |
| @@ -801,8 +802,16 @@ | ||
| 801 | 802 | if ( isset( $args['per_page'] ) ) { |
| 802 | 803 | $args['number'] = $args['per_page']; |
| 803 | 804 | } |
| 804 | 805 | |
| 806 | + if ( isset( $args['include'] ) ) { | |
| 807 | + $args['comment__in'] = $args['include']; | |
| 808 | + } | |
| 809 | + | |
| 810 | + if ( isset( $args['exclude'] ) ) { | |
| 811 | + $args['comment__not_in'] = $args['exclude']; | |
| 812 | + } | |
| 813 | + | |
| 805 | 814 | /** |
| 806 | 815 | * Filter database arguments for comment query |
| 807 | 816 | * |
| 808 | 817 | * @hook ep_comment_query_db_args |
| @@ -817,25 +826,44 @@ | ||
| 817 | 826 | unset( $all_query_args['number'] ); |
| 818 | 827 | unset( $all_query_args['offset'] ); |
| 819 | 828 | $all_query_args['count'] = true; |
| 820 | 829 | |
| 821 | - /** | |
| 822 | - * Filter database arguments for comment count query | |
| 823 | - * | |
| 824 | - * @hook ep_comment_all_query_db_args | |
| 825 | - * @param {array} $args Query arguments based to WP_Comment_Query | |
| 826 | - * @since 3.6.0 | |
| 827 | - * @return {array} New arguments | |
| 828 | - */ | |
| 829 | - $total_objects = get_comments( apply_filters( 'ep_comment_all_query_db_args', $all_query_args, $args ) ); | |
| 830 | + if ( isset( $args['comment__in'] ) || 0 < $args['offset'] ) { | |
| 831 | + // Disable advanced pagination. Not useful if only indexing specific IDs. | |
| 832 | + $args['ep_indexing_advanced_pagination'] = false; | |
| 833 | + } | |
| 830 | 834 | |
| 831 | - if ( ! empty( $args['offset'] ) ) { | |
| 832 | - if ( (int) $args['offset'] >= $total_objects ) { | |
| 833 | - $total_objects = 0; | |
| 834 | - } | |
| 835 | + // Explicitly set the orderby to ID to prevent accidental modifications by other code. | |
| 836 | + add_filter( 'comments_clauses', [ $this, 'set_orderby' ], 9999, 2 ); | |
| 837 | + | |
| 838 | + // Enforce the following query args during advanced pagination to ensure things work correctly. | |
| 839 | + if ( $args['ep_indexing_advanced_pagination'] ) { | |
| 840 | + $args = array_merge( | |
| 841 | + $args, | |
| 842 | + [ | |
| 843 | + 'suppress_filters' => false, | |
| 844 | + 'orderby' => 'comment_ID', | |
| 845 | + 'order' => 'desc', | |
| 846 | + 'paged' => 1, | |
| 847 | + 'offset' => 0, | |
| 848 | + ] | |
| 849 | + ); | |
| 850 | + | |
| 851 | + // It's important to pass a custom cache domain. By default, WordPress caches results based on the default query arguments and doesn't account for custom arguments. @see \WP_Comment_Query::get_comments() | |
| 852 | + $cache_key = md5( get_current_blog_id() . wp_json_encode( $args ) ); | |
| 853 | + $args['cache_domain'] = 'elasticpress-comment-indexable-' . $cache_key; | |
| 854 | + | |
| 855 | + add_filter( 'comments_clauses', array( $this, 'bulk_indexing_filter_comments_where' ), 9999, 2 ); | |
| 856 | + | |
| 857 | + $query = new WP_Comment_Query( $args ); | |
| 858 | + $total_objects = $this->get_total_objects_for_query( $args ); | |
| 859 | + remove_filter( 'comments_clauses', array( $this, 'bulk_indexing_filter_comments_where' ), 9999, 2 ); | |
| 860 | + } else { | |
| 861 | + $query = new WP_Comment_Query( $args ); | |
| 862 | + $total_objects = $query->found_comments; | |
| 835 | 863 | } |
| 836 | 864 | |
| 837 | - $query = new WP_Comment_Query( $args ); | |
| 865 | + remove_filter( 'comments_clauses', [ $this, 'set_orderby' ], 9999, 2 ); | |
| 838 | 866 | |
| 839 | 867 | if ( is_array( $query->comments ) ) { |
| 840 | 868 | array_walk( $query->comments, [ $this, 'remap_comments' ] ); |
| 841 | 869 | } |
| @@ -846,8 +874,79 @@ | ||
| 846 | 874 | ]; |
| 847 | 875 | } |
| 848 | 876 | |
| 849 | 877 | /** |
| 878 | + * Filters the WHERE clause of the SQL query used for bulk indexing comments by modifying it to include a range | |
| 879 | + * of comment IDs based on advanced pagination parameters. | |
| 880 | + * | |
| 881 | + * @param array $clauses Associative array of the clauses for the query. | |
| 882 | + * @param \WP_Comment_Query $query The current WP_Comment_Query instance. | |
| 883 | + * | |
| 884 | + * @return array Modified SQL query clauses. | |
| 885 | + */ | |
| 886 | + public function bulk_indexing_filter_comments_where( $clauses, $query ) { | |
| 887 | + global $wpdb; | |
| 888 | + | |
| 889 | + $using_advanced_pagination = $this->get_query_var( $query, 'ep_indexing_advanced_pagination', false ); | |
| 890 | + | |
| 891 | + if ( $using_advanced_pagination ) { | |
| 892 | + $requested_upper_limit_id = $this->get_query_var( $query, 'ep_indexing_upper_limit_object_id', PHP_INT_MAX ); | |
| 893 | + $requested_lower_limit_object_id = $this->get_query_var( $query, 'ep_indexing_lower_limit_object_id', 0 ); | |
| 894 | + $last_processed_id = $this->get_query_var( $query, 'ep_indexing_last_processed_object_id', null ); | |
| 895 | + | |
| 896 | + // On the first loopthrough we begin with the requested upper limit ID. Afterwards, use the last processed ID to paginate. | |
| 897 | + $upper_limit_range_object_id = $requested_upper_limit_id; | |
| 898 | + if ( is_numeric( $last_processed_id ) ) { | |
| 899 | + $upper_limit_range_object_id = $last_processed_id - 1; | |
| 900 | + } | |
| 901 | + | |
| 902 | + // Sanitize. Abort if unexpected data at this point. | |
| 903 | + if ( ! is_numeric( $upper_limit_range_object_id ) || ! is_numeric( $requested_lower_limit_object_id ) ) { | |
| 904 | + return $clauses; | |
| 905 | + } | |
| 906 | + | |
| 907 | + $range = [ | |
| 908 | + 'upper_limit' => "{$wpdb->comments}.comment_ID <= {$upper_limit_range_object_id}", | |
| 909 | + 'lower_limit' => "{$wpdb->comments}.comment_ID >= {$requested_lower_limit_object_id}", | |
| 910 | + ]; | |
| 911 | + | |
| 912 | + // Skip the end range if it's unnecessary. | |
| 913 | + $skip_ending_range = 0 === $requested_lower_limit_object_id; | |
| 914 | + $where = $clauses['where']; | |
| 915 | + $where = $skip_ending_range ? " {$range['upper_limit']} AND {$where}" : " {$range['upper_limit']} AND {$range['lower_limit']} AND {$where}"; | |
| 916 | + | |
| 917 | + $clauses['where'] = $where; | |
| 918 | + } | |
| 919 | + | |
| 920 | + return $clauses; | |
| 921 | + } | |
| 922 | + | |
| 923 | + /** | |
| 924 | + * Get the total number of comments for a given query. | |
| 925 | + * | |
| 926 | + * @param array $query_args The query args. | |
| 927 | + * @return int The query result's found_comments. | |
| 928 | + */ | |
| 929 | + protected function get_total_objects_for_query( $query_args ) { | |
| 930 | + $normalized_query_args = array_merge( | |
| 931 | + $query_args, | |
| 932 | + [ | |
| 933 | + 'offset' => 0, | |
| 934 | + 'paged' => 1, | |
| 935 | + 'posts_per_page' => 1, | |
| 936 | + 'no_found_rows' => false, | |
| 937 | + 'ep_indexing_last_processed_object_id' => null, | |
| 938 | + ] | |
| 939 | + ); | |
| 940 | + | |
| 941 | + $cache_key = md5( get_current_blog_id() . wp_json_encode( $normalized_query_args ) ); | |
| 942 | + | |
| 943 | + $normalized_query_args['cache_domain'] = 'elasticpress-comment-indexable-' . $cache_key; | |
| 944 | + | |
| 945 | + return ( new WP_Comment_Query( $normalized_query_args ) )->found_comments; | |
| 946 | + } | |
| 947 | + | |
| 948 | + /** | |
| 850 | 949 | * Prepare a comment document for indexing |
| 851 | 950 | * |
| 852 | 951 | * @param int $comment_id Comment ID |
| 853 | 952 | * @since 3.6.0 |
| @@ -988,13 +1087,11 @@ | ||
| 988 | 1087 | |
| 989 | 1088 | if ( true === $allowed_protected_keys || in_array( $key, $allowed_protected_keys, true ) ) { |
| 990 | 1089 | $allow_index = true; |
| 991 | 1090 | } |
| 992 | - } else { | |
| 1091 | + } elseif ( true !== $excluded_public_keys && ! in_array( $key, $excluded_public_keys, true ) ) { | |
| 993 | 1092 | |
| 994 | - if ( true !== $excluded_public_keys && ! in_array( $key, $excluded_public_keys, true ) ) { | |
| 995 | 1093 | $allow_index = true; |
| 996 | - } | |
| 997 | 1094 | } |
| 998 | 1095 | |
| 999 | 1096 | /** |
| 1000 | 1097 | * Filter force allow a meta key |
| @@ -1073,8 +1170,17 @@ | ||
| 1073 | 1170 | $from_to['meta_value_num'] = 'meta.' . $args['meta_key'] . '.long'; |
| 1074 | 1171 | } |
| 1075 | 1172 | } |
| 1076 | 1173 | |
| 1174 | + /** | |
| 1175 | + * If `orderby` is 'none', WordPress will let the database decide on what should be used to order. | |
| 1176 | + * It will use the primary key ASC. | |
| 1177 | + */ | |
| 1178 | + if ( 'none' === $orderby ) { | |
| 1179 | + $orderby = 'ID'; | |
| 1180 | + $order = 'asc'; | |
| 1181 | + } | |
| 1182 | + | |
| 1077 | 1183 | $orderby = $from_to[ $orderby ] ?? $orderby; |
| 1078 | 1184 | |
| 1079 | 1185 | $sort[] = array( |
| 1080 | 1186 | $orderby => array( |
| @@ -1084,5 +1190,32 @@ | ||
| 1084 | 1190 | |
| 1085 | 1191 | return $sort; |
| 1086 | 1192 | } |
| 1087 | 1193 | |
| 1194 | + /** | |
| 1195 | + * Retrieve a specific query variable from the query object. | |
| 1196 | + * | |
| 1197 | + * @param \WP_Comment_Query $query The query object. | |
| 1198 | + * @param string $query_var The name of the query variable to retrieve. | |
| 1199 | + * @param string $default_value The default value to return if the query variable is not set. Default is an empty string. | |
| 1200 | + * | |
| 1201 | + * @return mixed The value of the query variable if set, otherwise the default value. | |
| 1202 | + */ | |
| 1203 | + public function get_query_var( $query, $query_var, $default_value = '' ) { | |
| 1204 | + return $query->query_vars[ $query_var ] ?? $default_value; | |
| 1205 | + } | |
| 1206 | + | |
| 1207 | + /** | |
| 1208 | + * Sets the ORDER BY clause for comment queries to order comments by their ID. | |
| 1209 | + * | |
| 1210 | + * @param array $clauses The SQL clauses array to modify. | |
| 1211 | + * @return array The modified SQL clauses array with the ORDER BY clause set. | |
| 1212 | + * | |
| 1213 | + * @since 5.2.0 | |
| 1214 | + */ | |
| 1215 | + public function set_orderby( $clauses ) { | |
| 1216 | + global $wpdb; | |
| 1217 | + | |
| 1218 | + $clauses['orderby'] = "{$wpdb->comments}.comment_ID DESC"; | |
| 1219 | + return $clauses; | |
| 1220 | + } | |
| 1088 | 1221 | } |
