| @@ -7,14 +7,14 @@ | ||
| 7 | 7 | */ |
| 8 | 8 | |
| 9 | 9 | namespace ElasticPress\Indexable\Comment; |
| 10 | 10 | |
| 11 | -use ElasticPress\Indexable as Indexable; | |
| 12 | -use ElasticPress\Indexables as Indexables; | |
| 13 | -use ElasticPress\Elasticsearch as Elasticsearch; | |
| 14 | -use ElasticPress\Indexable\Post\DateQuery as DateQuery; | |
| 15 | -use \WP_Comment_Query as WP_Comment_Query; | |
| 16 | -use ElasticPress\Features as Features; | |
| 11 | +use WP_Comment_Query; | |
| 12 | +use ElasticPress\Elasticsearch; | |
| 13 | +use ElasticPress\Features; | |
| 14 | +use ElasticPress\Indexable; | |
| 15 | +use ElasticPress\Indexable\Post\DateQuery; | |
| 16 | +use ElasticPress\Indexables; | |
| 17 | 17 | |
| 18 | 18 | if ( ! defined( 'ABSPATH' ) ) { |
| 19 | 19 | exit; // Exit if accessed directly. |
| 20 | 20 | } |
| @@ -32,13 +32,23 @@ | ||
| 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() { | |
| 42 | + public $support_indexing_advanced_pagination = true; | |
| 43 | + | |
| 44 | + /** | |
| 45 | + * Instantiate the indexable SyncManager and QueryIntegration, the main responsibles for the WP integration. | |
| 46 | + * | |
| 47 | + * @since 4.5.0 | |
| 48 | + * @return void | |
| 49 | + */ | |
| 50 | + public function setup() { | |
| 41 | 51 | $this->labels = [ |
| 42 | 52 | 'plural' => esc_html__( 'Comments', 'elasticpress' ), |
| 43 | 53 | 'singular' => esc_html__( 'Comment', 'elasticpress' ), |
| 44 | 54 | ]; |
| @@ -92,9 +102,9 @@ | ||
| 92 | 102 | |
| 93 | 103 | /** |
| 94 | 104 | * Support `paged` query var |
| 95 | 105 | * |
| 96 | - * If `offset` is used, that takes precendence | |
| 106 | + * If `offset` is used, that takes precedence | |
| 97 | 107 | * over this. |
| 98 | 108 | */ |
| 99 | 109 | if ( isset( $query_vars['paged'] ) && empty( $query_vars['offset'] ) && $query_vars['paged'] > 1 ) { |
| 100 | 110 | $formatted_args['from'] = $number * ( $query_vars['paged'] - 1 ); |
| @@ -442,10 +452,10 @@ | ||
| 442 | 452 | * Support `post_type` query var. |
| 443 | 453 | */ |
| 444 | 454 | if ( ! empty( $query_vars['post_type'] ) ) { |
| 445 | 455 | $filter['bool']['must'][]['bool']['must'] = [ |
| 446 | - 'term' => [ | |
| 447 | - 'comment_post_type.raw' => $query_vars['post_type'], | |
| 456 | + 'terms' => [ | |
| 457 | + 'comment_post_type.raw' => array_values( (array) $query_vars['post_type'] ), | |
| 448 | 458 | ], |
| 449 | 459 | ]; |
| 450 | 460 | |
| 451 | 461 | $use_filters = true; |
| @@ -692,14 +702,14 @@ | ||
| 692 | 702 | */ |
| 693 | 703 | $es_version = apply_filters( 'ep_fallback_elasticsearch_version', '2.0' ); |
| 694 | 704 | } |
| 695 | 705 | |
| 696 | - $mapping_file = 'initial.php'; | |
| 706 | + $es_version = (string) $es_version; | |
| 697 | 707 | |
| 698 | - if ( version_compare( $es_version, '5.0', '<' ) ) { | |
| 699 | - $mapping_file = 'pre-5-0.php'; | |
| 700 | - } elseif ( version_compare( $es_version, '7.0', '>=' ) ) { | |
| 701 | - $mapping_file = '7-0.php'; | |
| 708 | + $mapping_file = '7-0.php'; | |
| 709 | + | |
| 710 | + if ( version_compare( $es_version, '7.0', '<' ) ) { | |
| 711 | + $mapping_file = 'initial.php'; | |
| 702 | 712 | } |
| 703 | 713 | |
| 704 | 714 | /** |
| 705 | 715 | * Filter comment indexable mapping file |
| @@ -775,18 +785,19 @@ | ||
| 775 | 785 | * @since 3.6.0 |
| 776 | 786 | * @return array |
| 777 | 787 | */ |
| 778 | 788 | public function query_db( $args ) { |
| 779 | - | |
| 780 | 789 | $defaults = [ |
| 781 | - 'type' => $this->get_indexable_comment_types(), | |
| 782 | - 'status' => $this->get_indexable_comment_status(), | |
| 783 | - 'post_type' => Indexables::factory()->get( 'post' )->get_indexable_post_types(), | |
| 784 | - 'post_status' => Indexables::factory()->get( 'post' )->get_indexable_post_status(), | |
| 785 | - 'number' => $this->get_bulk_items_per_page(), | |
| 786 | - 'offset' => 0, | |
| 787 | - 'orderby' => 'comment_ID', | |
| 788 | - '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, | |
| 789 | 800 | ]; |
| 790 | 801 | |
| 791 | 802 | if ( isset( $args['per_page'] ) ) { |
| 792 | 803 | $args['number'] = $args['per_page']; |
| @@ -791,8 +802,16 @@ | ||
| 791 | 802 | if ( isset( $args['per_page'] ) ) { |
| 792 | 803 | $args['number'] = $args['per_page']; |
| 793 | 804 | } |
| 794 | 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 | + | |
| 795 | 814 | /** |
| 796 | 815 | * Filter database arguments for comment query |
| 797 | 816 | * |
| 798 | 817 | * @hook ep_comment_query_db_args |
| @@ -807,25 +826,44 @@ | ||
| 807 | 826 | unset( $all_query_args['number'] ); |
| 808 | 827 | unset( $all_query_args['offset'] ); |
| 809 | 828 | $all_query_args['count'] = true; |
| 810 | 829 | |
| 811 | - /** | |
| 812 | - * Filter database arguments for comment count query | |
| 813 | - * | |
| 814 | - * @hook ep_comment_all_query_db_args | |
| 815 | - * @param {array} $args Query arguments based to WP_Comment_Query | |
| 816 | - * @since 3.6.0 | |
| 817 | - * @return {array} New arguments | |
| 818 | - */ | |
| 819 | - $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 | + } | |
| 820 | 834 | |
| 821 | - if ( ! empty( $args['offset'] ) ) { | |
| 822 | - if ( (int) $args['offset'] >= $total_objects ) { | |
| 823 | - $total_objects = 0; | |
| 824 | - } | |
| 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; | |
| 825 | 863 | } |
| 826 | 864 | |
| 827 | - $query = new WP_Comment_Query( $args ); | |
| 865 | + remove_filter( 'comments_clauses', [ $this, 'set_orderby' ], 9999, 2 ); | |
| 828 | 866 | |
| 829 | 867 | if ( is_array( $query->comments ) ) { |
| 830 | 868 | array_walk( $query->comments, [ $this, 'remap_comments' ] ); |
| 831 | 869 | } |
| @@ -836,8 +874,79 @@ | ||
| 836 | 874 | ]; |
| 837 | 875 | } |
| 838 | 876 | |
| 839 | 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 | + /** | |
| 840 | 949 | * Prepare a comment document for indexing |
| 841 | 950 | * |
| 842 | 951 | * @param int $comment_id Comment ID |
| 843 | 952 | * @since 3.6.0 |
| @@ -978,13 +1087,11 @@ | ||
| 978 | 1087 | |
| 979 | 1088 | if ( true === $allowed_protected_keys || in_array( $key, $allowed_protected_keys, true ) ) { |
| 980 | 1089 | $allow_index = true; |
| 981 | 1090 | } |
| 982 | - } else { | |
| 1091 | + } elseif ( true !== $excluded_public_keys && ! in_array( $key, $excluded_public_keys, true ) ) { | |
| 983 | 1092 | |
| 984 | - if ( true !== $excluded_public_keys && ! in_array( $key, $excluded_public_keys, true ) ) { | |
| 985 | 1093 | $allow_index = true; |
| 986 | - } | |
| 987 | 1094 | } |
| 988 | 1095 | |
| 989 | 1096 | /** |
| 990 | 1097 | * Filter force allow a meta key |
| @@ -1042,98 +1149,73 @@ | ||
| 1042 | 1149 | if ( empty( $orderby ) ) { |
| 1043 | 1150 | return $sort; |
| 1044 | 1151 | } |
| 1045 | 1152 | |
| 1046 | - switch ( $orderby ) { | |
| 1047 | - case 'comment_agent': | |
| 1048 | - $orderby_field = 'comment_agent.raw'; | |
| 1049 | - break; | |
| 1153 | + $from_to = [ | |
| 1154 | + 'comment_agent' => 'comment_agent.raw', | |
| 1155 | + 'comment_approved' => 'comment_approved.raw', | |
| 1156 | + 'comment_author' => 'comment_author.raw', | |
| 1157 | + 'comment_author_email' => 'comment_author_email.raw', | |
| 1158 | + 'comment_author_IP' => 'comment_author_IP.raw', | |
| 1159 | + 'comment_author_url' => 'comment_author_url.raw', | |
| 1160 | + 'comment_content' => 'comment_content.raw', | |
| 1161 | + 'comment_type' => 'comment_type.raw', | |
| 1162 | + 'comment_post_type' => 'comment_post_type.raw', | |
| 1163 | + ]; | |
| 1050 | 1164 | |
| 1051 | - case 'comment_approved': | |
| 1052 | - $orderby_field = 'comment_approved.raw'; | |
| 1053 | - break; | |
| 1165 | + if ( in_array( $orderby, [ 'meta_value', 'meta_value_num' ], true ) ) { | |
| 1166 | + if ( empty( $args['meta_key'] ) ) { | |
| 1167 | + return $sort; | |
| 1168 | + } else { | |
| 1169 | + $from_to['meta_value'] = 'meta.' . $args['meta_key'] . '.raw'; | |
| 1170 | + $from_to['meta_value_num'] = 'meta.' . $args['meta_key'] . '.long'; | |
| 1171 | + } | |
| 1172 | + } | |
| 1054 | 1173 | |
| 1055 | - case 'comment_author': | |
| 1056 | - $orderby_field = 'comment_author.raw'; | |
| 1057 | - break; | |
| 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 | + } | |
| 1058 | 1182 | |
| 1059 | - case 'comment_author_email': | |
| 1060 | - $orderby_field = 'comment_author_email.raw'; | |
| 1061 | - break; | |
| 1183 | + $orderby = $from_to[ $orderby ] ?? $orderby; | |
| 1062 | 1184 | |
| 1063 | - case 'comment_author_IP': | |
| 1064 | - $orderby_field = 'comment_author_IP.raw'; | |
| 1065 | - break; | |
| 1185 | + $sort[] = array( | |
| 1186 | + $orderby => array( | |
| 1187 | + 'order' => $order, | |
| 1188 | + ), | |
| 1189 | + ); | |
| 1066 | 1190 | |
| 1067 | - case 'comment_author_url': | |
| 1068 | - $orderby_field = 'comment_author_url.raw'; | |
| 1069 | - break; | |
| 1191 | + return $sort; | |
| 1192 | + } | |
| 1070 | 1193 | |
| 1071 | - case 'comment_content': | |
| 1072 | - $orderby_field = 'comment_content.raw'; | |
| 1073 | - break; | |
| 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 | + } | |
| 1074 | 1206 | |
| 1075 | - case 'comment_date': | |
| 1076 | - $orderby_field = 'comment_date'; | |
| 1077 | - break; | |
| 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; | |
| 1078 | 1217 | |
| 1079 | - case 'comment_date_gmt': | |
| 1080 | - $orderby_field = 'comment_date_gmt'; | |
| 1081 | - break; | |
| 1082 | - | |
| 1083 | - case 'comment_ID': | |
| 1084 | - $orderby_field = 'comment_ID'; | |
| 1085 | - break; | |
| 1086 | - | |
| 1087 | - case 'comment_karma': | |
| 1088 | - $orderby_field = 'comment_karma'; | |
| 1089 | - break; | |
| 1090 | - | |
| 1091 | - case 'comment_parent': | |
| 1092 | - $orderby_field = 'comment_parent'; | |
| 1093 | - break; | |
| 1094 | - | |
| 1095 | - case 'comment_post_ID': | |
| 1096 | - $orderby_field = 'comment_post_ID'; | |
| 1097 | - break; | |
| 1098 | - | |
| 1099 | - case 'comment_type': | |
| 1100 | - $orderby_field = 'comment_type.raw'; | |
| 1101 | - break; | |
| 1102 | - | |
| 1103 | - case 'comment_post_type': | |
| 1104 | - $orderby_field = 'comment_post_type.raw'; | |
| 1105 | - break; | |
| 1106 | - | |
| 1107 | - case 'user_id': | |
| 1108 | - $orderby_field = 'user_id'; | |
| 1109 | - break; | |
| 1110 | - | |
| 1111 | - case 'meta_value': | |
| 1112 | - if ( ! empty( $args['meta_key'] ) ) { | |
| 1113 | - $orderby_field = 'meta.' . $args['meta_key'] . '.value'; | |
| 1114 | - } | |
| 1115 | - break; | |
| 1116 | - | |
| 1117 | - case 'meta_value_num': | |
| 1118 | - if ( ! empty( $args['meta_key'] ) ) { | |
| 1119 | - $orderby_field = 'meta.' . $args['meta_key'] . '.long'; | |
| 1120 | - } | |
| 1121 | - break; | |
| 1122 | - | |
| 1123 | - default: | |
| 1124 | - $orderby_field = $orderby; | |
| 1125 | - break; | |
| 1126 | - } | |
| 1127 | - | |
| 1128 | - if ( ! empty( $orderby_field ) ) { | |
| 1129 | - $sort[] = [ | |
| 1130 | - $orderby_field => [ | |
| 1131 | - 'order' => $order, | |
| 1132 | - ], | |
| 1133 | - ]; | |
| 1134 | - } | |
| 1135 | - | |
| 1136 | - return $sort; | |
| 1218 | + $clauses['orderby'] = "{$wpdb->comments}.comment_ID DESC"; | |
| 1219 | + return $clauses; | |
| 1137 | 1220 | } |
| 1138 | - | |
| 1139 | 1221 | } |
