Commit inicial - WordPress Análisis de Precios Unitarios

- WordPress core y plugins
- Tema Twenty Twenty-Four configurado
- Plugin allow-unfiltered-html.php simplificado
- .gitignore configurado para excluir wp-config.php y uploads

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

wp-content/plugins/ajax-search-pro/src/server/Index/Content.php

@@ -0,0 +1,96 @@
+<?php
+namespace WPDRMS\ASP\Index;
+
+
+defined('ABSPATH') or die("You can't access this file directly.");
+
+if ( !class_exists(__NAMESPACE__ . '\Content') ) {
+	class Content {
+		public static function mySQLFixes( $str ) {
+			return str_replace(array(
+				// SELECT œ LIKE oe => FALSE but SELECT œ = oe => TRUE
+				// When doing INSERT with "oe" or "œ" in keyword, it is ignored as it is considered already existing
+				'œ'
+			),array(
+				'oe'
+			), $str);
+		}
+
+		public static function arabicRemoveDiacritics( $str ) {
+			if ( is_array($str) ) {
+				foreach ($str as &$v) {
+					$v = self::arabicRemoveDiacritics($v);
+				}
+				return $str;
+			}
+
+			$characters = array(
+				"~[\x{0600}-\x{061F}]~u",
+				"~[\x{063B}-\x{063F}]~u",
+				"~[\x{064B}-\x{065E}]~u",
+				"~[\x{066A}-\x{06FF}]~u",
+			);
+			return preg_replace($characters, "", $str);
+		}
+		public static function hebrewUnvocalize( $str ) {
+			if ( is_array($str) ) {
+				foreach ($str as &$v) {
+					$v = self::hebrewUnvocalize($v);
+				}
+				return $str;
+			}
+			if ( preg_match("/[\x{0591}-\x{05F4}]/u", $str) ) {
+				$hebrew_common_ligatures = array(
+					'ײַ' => 'ײ',
+					'ﬠ' => 'ע',
+					'ﬡ' => 'א',
+					'ﬢ' => 'ד',
+					'ﬣ' => 'ה',
+					'ﬤ' => 'כ',
+					'ﬥ' => 'ל',
+					'ﬦ' => 'ם',
+					'ﬧ' => 'ר',
+					'ﬨ' => 'ת',
+					'שׁ' => 'ש',
+					'שׂ' => 'ש',
+					'שּׁ' => 'ש',
+					'שּׂ' => 'ש',
+					'אַ' => 'א',
+					'אָ' => 'א',
+					'אּ' => 'א',
+					'בּ' => 'ב',
+					'גּ' => 'ג',
+					'דּ' => 'ד',
+					'הּ' => 'ה',
+					'וּ' => 'ו',
+					'זּ' => 'ז',
+					'טּ' => 'ט',
+					'יּ' => 'י',
+					'ךּ' => 'ך',
+					'כּ' => 'כ',
+					'לּ' => 'ל',
+					'מּ' => 'מ',
+					'נּ' => 'נ',
+					'סּ' => 'ס',
+					'ףּ' => 'ף',
+					'פּ' => 'פ',
+					'צּ' => 'צ',
+					'קּ' => 'ק',
+					'רּ' => 'ר',
+					'שּ' => 'ש',
+					'תּ' => 'ת',
+					'וֹ' => 'ו',
+					'בֿ' => 'ב',
+					'כֿ' => 'כ',
+					'פֿ' => 'פ',
+					'ﭏ' => 'אל'
+				);
+				$str = trim(preg_replace('/\p{Mn}/u', '', $str));
+				foreach ($hebrew_common_ligatures as $word1 => $word2) {
+					$str = trim(str_replace($word1, $word2, $str));
+				}
+			}
+			return $str;
+		}
+	}
+}

wp-content/plugins/ajax-search-pro/src/server/Index/Database.php

@@ -0,0 +1,563 @@
+<?php
+namespace WPDRMS\ASP\Index;
+
+use WPDRMS\ASP\Utils\Plugin;
+use WPDRMS\ASP\Utils\Str;
+
+defined('ABSPATH') or die("You can't access this file directly.");
+
+if ( !class_exists(__NAMESPACE__ . '\Database') ) {
+	class Database {
+		private $table_name;
+
+		function __construct() {
+			require_once ABSPATH . 'wp-admin/includes/upgrade.php';
+			$this->table_name = wd_asp()->db->table('index');
+		}
+
+		function create(): array {
+			global $wpdb;
+			$return = array();
+
+			$charset_collate = '';
+
+			if ( ! empty( $wpdb->charset ) ) {
+				$charset_collate_bin_column = "CHARACTER SET $wpdb->charset";
+				$charset_collate            = "DEFAULT $charset_collate_bin_column";
+			}
+			if ( strpos( $wpdb->collate, '_' ) > 0 ) {
+				$charset_collate .= " COLLATE $wpdb->collate";
+			}
+
+			$query = '
+				CREATE TABLE IF NOT EXISTS ' . $this->table_name . " (
+					doc bigint(20) UNSIGNED NOT NULL DEFAULT '0',
+					term varchar(150) NOT NULL DEFAULT '0',
+					term_reverse varchar(150) NOT NULL DEFAULT '0',
+					blogid mediumint(9) UNSIGNED NOT NULL DEFAULT '0',
+					content smallint(9) UNSIGNED NOT NULL DEFAULT '0',
+					title tinyint(3) UNSIGNED NOT NULL DEFAULT '0',
+					comment tinyint(3) UNSIGNED NOT NULL DEFAULT '0',
+					tag tinyint(3) UNSIGNED NOT NULL DEFAULT '0',
+					link tinyint(3) UNSIGNED NOT NULL DEFAULT '0',
+					author tinyint(3) UNSIGNED NOT NULL DEFAULT '0',
+					excerpt tinyint(3) UNSIGNED NOT NULL DEFAULT '0',
+					customfield smallint(9) UNSIGNED NOT NULL DEFAULT '0',
+					post_type varchar(50) NOT NULL DEFAULT 'post',
+					lang varchar(20) NOT NULL DEFAULT '0',
+			    PRIMARY KEY doctermitem (doc, term, blogid)) $charset_collate";
+
+			dbDelta( $query );
+
+			$return[]         = $query;
+			$query            = "SHOW INDEX FROM $this->table_name";
+			$indices          = $wpdb->get_results( $query );
+			$existing_indices = array();
+
+			foreach ( $indices as $index ) {
+				if ( isset( $index->Key_name ) ) {
+					$existing_indices[] = $index->Key_name;
+				}
+			}
+
+			// Worst case scenario optimal indexes
+			if ( ! in_array( 'term_ptype_bid_lang', $existing_indices ) ) {
+				$sql = "CREATE INDEX term_ptype_bid_lang ON $this->table_name (term(20), post_type(20), blogid, lang(10))";
+				$wpdb->query( $sql );
+				$return[] = $sql;
+			}
+			if ( ! in_array( 'rterm_ptype_bid_lang', $existing_indices ) ) {
+				$sql = "CREATE INDEX rterm_ptype_bid_lang ON $this->table_name (term_reverse(20), post_type(20), blogid, lang(10))";
+				$wpdb->query( $sql );
+				$return[] = $sql;
+			}
+			if ( !in_array( 'doc', $existing_indices ) ) {
+				$sql = "CREATE INDEX `doc` ON $this->table_name (`doc`)";
+				$wpdb->query( $sql );
+				$return[] = $sql;
+			}
+
+			return $return;
+		}
+
+		public function scheduled() {
+			global $wpdb;
+
+			// 4.20.3
+			if ( Plugin::previousVersion('4.20.2') ) {
+				if ( $wpdb->get_var( "SHOW COLUMNS FROM `$this->table_name` LIKE 'taxonomy';" ) ) {
+					$query = "ALTER TABLE `$this->table_name` 
+                    DROP COLUMN `taxonomy`,
+                    DROP COLUMN `category`,
+                    DROP COLUMN `item`";
+					$wpdb->query($query);
+
+					$query = "ALTER TABLE `$this->table_name` 
+                    MODIFY COLUMN `content` smallint(9) UNSIGNED,
+                    MODIFY COLUMN `title` tinyint(3) UNSIGNED,
+                    MODIFY COLUMN `comment` tinyint(3) UNSIGNED,
+                    MODIFY COLUMN `tag` tinyint(3) UNSIGNED,
+                    MODIFY COLUMN `link` tinyint(3) UNSIGNED,
+                    MODIFY COLUMN `author` tinyint(3) UNSIGNED,
+                    MODIFY COLUMN `excerpt` tinyint(3) UNSIGNED,
+                    MODIFY COLUMN `customfield` smallint(9) UNSIGNED";
+					$wpdb->query( $query );
+
+					$query = "OPTIMIZE TABLE `$this->table_name`";
+					$wpdb->query( $query );
+				}
+			}
+		}
+
+		/**
+		 * Runs a table optimize query on the index table
+		 *
+		 * @return bool|false|int
+		 */
+		public function optimize() {
+			global $wpdb;
+			// In innoDB this is mapped to "ALTER TABLE .. FORCE", aka. defragmenting
+			// OPTIMIZE only needs SELECT and INSERT privileges
+			return $wpdb->query( 'OPTIMIZE TABLE ' . $this->table_name );
+		}
+
+		public function truncate() {
+			global $wpdb;
+			$wpdb->query( 'TRUNCATE TABLE ' . $this->table_name );
+		}
+
+		function removeDocument( $post_id ) {
+			global $wpdb;
+
+			if ( is_array($post_id) ) {
+				foreach ( $post_id as &$v ) {
+					$v = $v + 0;
+				}
+				$post_ids = implode(', ', $post_id);
+				$wpdb->query( "DELETE FROM $this->table_name WHERE doc IN ($post_ids)"  );
+			} else {
+				$wpdb->query(
+					$wpdb->prepare(
+						"DELETE FROM $this->table_name WHERE doc = %d",
+						$post_id
+					) 
+				);
+			}
+
+			/*
+			DO NOT call finishOperation() here, it would switch back the blog too early.
+			Calling this function from an action hooks does not require switching the blog,
+			as the correct one is in use there.
+			*/
+		}
+
+		/**
+		 * Generates the query based on the post and the token array and inserts into DB
+		 *
+		 * @return int
+		 */
+		function insertTokensToDB( $the_post, $tokens, $blog_id, $lang ) {
+			global $wpdb;
+			$values = array();
+
+			if ( count( $tokens ) <= 0 ) {
+				return false;
+			}
+
+			foreach ( $tokens as $d ) {
+				// If it's numeric, delete the leading space
+				$term = trim( $d['_keyword'] );
+
+				if ( isset($d['_no_reverse']) && $d['_no_reverse'] === true ) {
+					$value = $wpdb->prepare(
+						'(%d, %s, %s, %d, %d, %d, %d, %d, %d, %d, %d, %d, %s, %s)',
+						$the_post->ID,
+						$term,
+						'',
+						$blog_id,
+						$d['content'],
+						$d['title'],
+						$d['comment'],
+						$d['tag'],
+						$d['link'],
+						$d['author'],
+						$d['excerpt'],
+						$d['customfield'],
+						$the_post->post_type,
+						$lang
+					);
+				} else {
+					$value = $wpdb->prepare(
+						'(%d, %s, REVERSE(%s), %d, %d, %d, %d, %d, %d, %d, %d, %d, %s, %s)',
+						$the_post->ID,
+						$term,
+						$term,
+						$blog_id,
+						$d['content'],
+						$d['title'],
+						$d['comment'],
+						$d['tag'],
+						$d['link'],
+						$d['author'],
+						$d['excerpt'],
+						$d['customfield'],
+						$the_post->post_type,
+						$lang
+					);
+				}
+
+				$values[] = $value;
+
+				// Split INSERT at every 200 records
+				if ( count( $values ) > 199 ) {
+					$values = implode( ', ', $values );
+					$query  = "INSERT IGNORE INTO $this->table_name
+                    (`doc`, `term`, `term_reverse`, `blogid`, `content`, `title`, `comment`, `tag`, `link`, `author`,
+                     `excerpt`, `customfield`, `post_type`, `lang`)
+                    VALUES $values";
+					$wpdb->query( $query );
+					$values = array();
+				}
+			}
+
+			// Add the remaining as well
+			if ( count( $values ) > 0 ) {
+				$values = implode( ', ', $values );
+				$query  = "INSERT IGNORE INTO $this->table_name
+				(`doc`, `term`, `term_reverse`, `blogid`, `content`, `title`, `comment`, `tag`, `link`, `author`,
+				 `excerpt`, `customfield`, `post_type`, `lang`)
+				VALUES $values";
+				$wpdb->query( $query );
+			}
+
+			return count( $tokens );
+		}
+
+		/**
+		 * Gets the post IDs to index
+		 *
+		 * @return array of post IDs
+		 */
+		function getPostIdsToIndex( $args, $posts_to_ignore ): array {
+			global $wpdb;
+			$parent_join = '';
+
+			$_statuses = explode(',', $args['post_statuses']);
+			foreach ( $_statuses as &$sv ) {
+				$sv = trim($sv);
+			}
+			$valid_status = "'" . implode("', '", $_statuses ) . "'";
+
+			if ( count($args['post_types']) > 0 ) {
+				$post_types = $args['post_types'];
+				if ( class_exists('WooCommerce') && in_array('product_variation', $post_types) ) { // Special case for Woo variations
+					$post_types = array_diff($post_types, array( 'product_variation' ));
+					$rest       = '';
+					if ( count($post_types) > 0 ) {
+						$rest = " OR post.post_type IN('" . implode("', '", $post_types) . "') ";
+					}
+					// In case of product variation the parent post status must also match, otherwise it is not relevant
+					$parent_join = "LEFT JOIN $wpdb->posts parent ON (post.post_parent = parent.ID)";
+					$restriction = " AND ( (post.post_type = 'product_variation' AND parent.post_status IN($valid_status) ) $rest )";
+				} else {
+					$restriction = " AND post.post_type IN ('" . implode("', '", $post_types) . "')";
+				}
+			} else {
+				return array();
+			}
+
+			$mimes_restrict = '';
+
+			if ( in_array('attachment', $args['post_types'], true) ) {
+				$restriction .= $this->getAttachmentDirRestrictionQueryPart( $args );
+
+				if ( $args['attachment_mime_types'] != '' ) {
+					$mimes_arr = wpd_comma_separated_to_array($args['attachment_mime_types']);
+					if ( count($mimes_arr) > 0 ) {
+						$mimes_restrict = "OR ( post.post_status = 'inherit' AND post.post_mime_type IN ('" . implode("','", $mimes_arr) . "') )";
+					}
+				}
+			}
+
+			$post_password = '';
+			if ( $args['post_password_protected'] == 0 ) {
+				$post_password = " AND (post.post_password = '') ";
+			}
+
+			$ignore_posts = '';
+			if ( !empty($posts_to_ignore[ $args['blog_id'] ]) ) {
+				$ignore_posts = ' AND post.ID NOT IN( ' . implode(',', $posts_to_ignore[ $args['blog_id'] ]) . ' )';
+			}
+
+			$limit = $args['limit'] > 1000 ? 1000 : ( $args['limit'] + 0 );
+
+			$add_where             = apply_filters('asp/index/database/get_posts_to_index/query/add_where', '', $args);
+			$add_where_post_status = apply_filters('asp/index/database/get_posts_to_index/query/add_where_post_status', '', $args);
+
+			if ( $args['extend'] ) {
+				// We are extending, so keep the existing
+				$q = "SELECT post.ID
+						FROM $wpdb->posts post
+						$parent_join
+						LEFT JOIN $this->table_name r ON (post.ID = r.doc AND r.blogid = " . $args['blog_id'] . ")
+						WHERE
+								r.doc is null
+						AND
+                            (
+                                post.post_status IN ($valid_status)
+                                $mimes_restrict
+                                $add_where_post_status
+                            )
+						$restriction
+						$ignore_posts
+						$post_password
+						$add_where
+						ORDER BY post.ID ASC
+						LIMIT $limit";
+			} else {
+				$q = "SELECT post.ID
+						FROM $wpdb->posts post
+						$parent_join
+						WHERE
+                            (
+                                post.post_status IN ($valid_status)
+                                $mimes_restrict
+                                $add_where_post_status
+                            )
+						$restriction
+						$ignore_posts
+						$post_password
+						$add_where
+						ORDER BY post.ID ASC
+						LIMIT $limit";
+
+			}
+			return $wpdb->get_results( $q );
+		}
+
+		/**
+		 * Gets the number documents to index
+		 */
+		public function getPostIdsToIndexCount( $args, $posts_to_ignore, $check_only = false ): int {
+			if ( defined('ASP_INDEX_BYPASS_COUNT') ) {
+				return 9999;
+			}
+			global $wpdb;
+			$parent_join = '';
+
+			$_statuses = explode(',', $args['post_statuses']);
+			foreach ( $_statuses as &$sv ) {
+				$sv = trim($sv);
+			}
+			$valid_status   = "'" . implode("', '", $_statuses ) . "'";
+			$mimes_restrict = '';
+
+			if ( count($args['post_types']) > 0 ) {
+				$post_types = $args['post_types'];
+				if ( class_exists('WooCommerce') && in_array('product_variation', $post_types) ) { // Special case for Woo variations
+					$post_types = array_diff($post_types, array( 'product_variation' ));
+					$rest       = '';
+					if ( count($post_types) > 0 ) { // are there any left?
+						$rest = " OR post.post_type IN('" . implode("', '", $post_types) . "') ";
+					}
+					// In case of product variation the parent post status must also match, otherwise it is not relevant
+					$parent_join = "LEFT JOIN $wpdb->posts parent ON (post.post_parent = parent.ID)";
+					$restriction = " AND ( (post.post_type = 'product_variation' AND parent.post_status IN($valid_status) ) $rest )";
+				} else {
+					$restriction = " AND post.post_type IN ('" . implode("', '", $post_types) . "')";
+				}
+			} else {
+				return 0;
+			}
+
+			if ( in_array('attachment', $args['post_types'], true) ) {
+				$restriction .= $this->getAttachmentDirRestrictionQueryPart( $args );
+
+				if ( $args['attachment_mime_types'] != '' ) {
+					$mimes_arr = wpd_comma_separated_to_array($args['attachment_mime_types']);
+					if ( count($mimes_arr) > 0 ) {
+						$mimes_restrict = "OR ( post.post_status = 'inherit' AND post.post_mime_type IN ('" . implode("','", $mimes_arr) . "') )";
+					}
+				}
+			}
+
+			$post_password = '';
+			if ( $args['post_password_protected'] == 0 ) {
+				$post_password = " AND (post.post_password = '') ";
+			}
+
+			$ignore_posts = '';
+			if ( !empty($posts_to_ignore[ $args['blog_id'] ]) ) {
+				$ignore_posts = ' AND post.ID NOT IN( ' . implode(',', $posts_to_ignore[ $args['blog_id'] ]) . ' )';
+			}
+
+			$add_where             = apply_filters('asp/index/database/get_posts_to_index/query/add_where', '', $args);
+			$add_where_post_status = apply_filters('asp/index/database/get_posts_to_index/query/add_where_post_status', '', $args);
+
+			if ( $check_only ) {
+				$q = "SELECT 1
+                        FROM $wpdb->posts post
+                        $parent_join
+                        LEFT JOIN $this->table_name r ON (post.ID = r.doc AND r.blogid = " . $args['blog_id'] . ")
+                        WHERE
+                            r.doc is null
+                        AND
+                            (
+                                post.post_status IN ($valid_status)
+                                $mimes_restrict
+                                $add_where_post_status
+                            )
+                        $restriction
+                        $ignore_posts
+                        $post_password
+                        $add_where
+                        LIMIT 1";
+			} else {
+				$q = "SELECT COUNT(DISTINCT post.ID)
+                        FROM $wpdb->posts post
+                        $parent_join
+                        LEFT JOIN $this->table_name r ON (post.ID = r.doc AND r.blogid = " . $args['blog_id'] . ")
+                        WHERE
+                                r.doc is null
+                        AND
+                            (
+                                post.post_status IN ($valid_status)
+                                $mimes_restrict
+                                $add_where_post_status
+                            )
+                        $restriction
+                        $ignore_posts
+                        $post_password
+                        $add_where";
+			}
+
+			return intval( $wpdb->get_var( $q ) );
+		}
+
+		private function getAttachmentDirRestrictionQueryPart( array $args ): string {
+			global $wpdb;
+			$attachment_dir_query = '';
+
+			$uploads = wp_get_upload_dir();
+			if ( false !== $uploads['error'] ) {
+				return '';
+			}
+
+			/**
+			 * This parts makes corrections to "trim" the absolute path to the upload directory
+			 * as _wp_attached_file stores the directory part only after ../wp-content/uploads/
+			 */
+			$uploads_relative_dir =
+				trailingslashit( str_replace(ABSPATH, '', $uploads['basedir']) );
+			$exclude_directories  = array_filter(
+				array_map(
+					function ( $dir ) use ( $uploads_relative_dir ) {
+						return str_replace($uploads_relative_dir, '', $dir);
+					},
+					$args['attachment_exclude_directories']
+				)
+			);
+			$include_directories  = array_filter(
+				array_map(
+					function ( $dir ) use ( $uploads_relative_dir ) {
+						return str_replace($uploads_relative_dir, '', $dir);
+					},
+					$args['attachment_include_directories']
+				)
+			);
+
+			if ( count($exclude_directories) > 0 ) {
+				$not_like_values       = implode(
+					'AND',
+					array_map(
+						function ( $directory ) use ( $wpdb ) {
+							return " $wpdb->postmeta.meta_value NOT LIKE '" . Str::escape($directory) . "%'";
+						},
+						$exclude_directories
+					)
+				);
+				$attachment_dir_query .= " AND ( 
+			(
+			  (
+				SELECT IF((meta_key IS NULL OR meta_value = ''), -1, COUNT(meta_id))
+				FROM $wpdb->postmeta
+				WHERE $wpdb->postmeta.post_id = post.ID AND $wpdb->postmeta.meta_key='_wp_attached_file'
+				LIMIT 1
+			  ) = -1
+			 OR
+			  (
+				SELECT COUNT(meta_id) as mtc
+				FROM $wpdb->postmeta
+				WHERE $wpdb->postmeta.post_id = post.ID AND $wpdb->postmeta.meta_key='_wp_attached_file' AND 
+				($not_like_values)
+				ORDER BY mtc
+				LIMIT 1
+			  )  >= 1
+			) )";
+			}
+
+			if ( count($include_directories) > 0 ) {
+				$not_like_values       = implode(
+					'OR',
+					array_map(
+						function ( $directory ) use ( $wpdb ) {
+							return " $wpdb->postmeta.meta_value LIKE '" . Str::escape($directory) . "%'";
+						},
+						$include_directories
+					)
+				);
+				$attachment_dir_query .= " AND ( 
+			(
+			  (
+				SELECT IF((meta_key IS NULL OR meta_value = ''), -1, COUNT(meta_id))
+				FROM $wpdb->postmeta
+				WHERE $wpdb->postmeta.post_id = post.ID AND $wpdb->postmeta.meta_key='_wp_attached_file'
+				LIMIT 1
+			  ) = -1
+			 OR
+			  (
+				SELECT COUNT(meta_id) as mtc
+				FROM $wpdb->postmeta
+				WHERE $wpdb->postmeta.post_id = post.ID AND $wpdb->postmeta.meta_key='_wp_attached_file' AND 
+				($not_like_values)
+				ORDER BY mtc
+				LIMIT 1
+			  )  >= 1
+			) )";
+			}
+
+			return $attachment_dir_query;
+		}
+
+
+		public function getPostsIndexed() {
+			if ( defined('ASP_INDEX_BYPASS_COUNT') ) {
+				return 9999;
+			}
+			global $wpdb;
+			// Tested faster as a regular single query count
+			$sql = 'SELECT COUNT(count) FROM (SELECT 1 as count FROM ' . wd_asp()->db->table('index') . ' GROUP BY doc) as A';
+			return $wpdb->get_var($sql);
+		}
+
+		public function getTotalKeywords() {
+			if ( defined('ASP_INDEX_BYPASS_COUNT') ) {
+				return 9999;
+			}
+			global $wpdb;
+
+			if ( is_multisite() ) {
+				$sql = 'SELECT COUNT(doc) FROM ' . wd_asp()->db->table('index');
+			} else {
+				$sql = 'SELECT COUNT(doc) FROM ' . wd_asp()->db->table('index') . ' WHERE blogid = ' . get_current_blog_id();
+			}
+
+			return $wpdb->get_var($sql);
+		}
+
+		public function isEmpty(): bool {
+			global $wpdb;
+			return $wpdb->query('SELECT 1 FROM ' . wd_asp()->db->table('index') . ' LIMIT 1') == 0;
+		}
+	}
+}

wp-content/plugins/ajax-search-pro/src/server/Index/Shortcode.php

@@ -0,0 +1,166 @@
+<?php /** @noinspection PhpComposerExtensionStubsInspection */
+
+/** @noinspection RegExpRedundantEscape */
+
+namespace WPDRMS\ASP\Index;
+
+use WPDRMS\ASP\Utils\Str;
+
+defined('ABSPATH') or die("You can't access this file directly.");
+
+if ( !class_exists(__NAMESPACE__ . '\Shortcode') ) {
+	class Shortcode {
+
+		private $temporary_shortcode_tags = array();
+
+		/**
+		 * Executes the shortcodes within the given string
+		 *
+		 * @param $content
+		 * @param $post
+		 * @param $exclude
+		 * @return string
+		 * @noinspection PhpUndefinedClassInspection
+		 */
+		function execute($content, $post, $exclude): string {
+			$content = apply_filters( 'asp_index_before_shortcode_execution', $content, $post );
+
+			// WP Table Reloaded support
+			if ( defined( 'WP_TABLE_RELOADED_ABSPATH' ) ) {
+				/** @noinspection PhpIncludeInspection */
+				include_once( WP_TABLE_RELOADED_ABSPATH . 'controllers/controller-frontend.php' );
+				if ( class_exists('\\WP_Table_Reloaded_Controller_Frontend') ) {
+					/** @noinspection PhpFullyQualifiedNameUsageInspection */
+					$wpt_reloaded = new \WP_Table_Reloaded_Controller_Frontend();
+				}
+			}
+			// TablePress support
+			if ( defined( 'TABLEPRESS_ABSPATH' ) && class_exists('\\TablePress') ) {
+				$content .= ' ' . $this->parseTablePressShortcodes($content);
+			}
+
+			// Remove user defined shortcodes
+			$shortcodes = explode( ',',$exclude );
+			$try_getting_sc_content = apply_filters('asp_it_try_getting_sc_content', true);
+			foreach ( $shortcodes as $shortcode ) {
+				$shortcode = trim($shortcode);
+				if ( $shortcode == '' )
+					continue;
+				// First let us try to get any contents from the shortcode itself
+				if ( $try_getting_sc_content ) {
+					$content = preg_replace(
+						'/(?:\[' . $shortcode . '[ ]+.*?\]|\[' . $shortcode . '[ ]*\])(.*?)\[\/' . $shortcode . '[ ]*]/su',
+						' $1 ',
+						$content
+					);
+				}
+				// Then remove the shortcode completely
+				$this->temporaryDisableShortcode($shortcode);
+			}
+
+			// Try extracting the content of these shortcodes, but do not execute
+			$more_shortcodes = array(
+				'cws-widget', 'cws-row', 'cws-column', 'col', 'row', 'item'
+			);
+			foreach ( $more_shortcodes as $shortcode ) {
+				// First let us try to get any contents from the shortcode itself
+				$content = preg_replace(
+					'/(?:\[' . $shortcode . '[ ]+.*?\]|\[' . $shortcode . '[ ]*\])(.*?)\[\/' . $shortcode . '[ ]*]/su',
+					' $1 ',
+					$content
+				);
+				/*remove_shortcode( $shortcode );
+				add_shortcode( $shortcode, array( $this, 'return_empty_string' ) );*/
+				$this->temporaryDisableShortcode($shortcode);
+			}
+
+			// These shortcodes are completely ignored, and removed with content
+			$ignore_shortcodes = array(
+				'vc_asp_search',
+				'ts_products_in_category_tabs',
+				'wd_asp',
+				'wpdreams_ajaxsearchpro',
+				'wpdreams_ajaxsearchpro_results',
+				'wpdreams_asp_settings',
+				'contact-form',
+				'starrater',
+				'responsive-flipbook',
+				'avatar_upload',
+				'product_categories',
+				'recent_products',
+				'templatera',
+				'bsf-info-box', 'logo-slider',
+				'ourteam', 'embedyt', 'gallery', 'bsf-info-box', 'tweet', 'blog', 'portfolio',
+				'peepso_activity', 'peepso_profile', 'peepso_group'
+			);
+			if ( defined( 'TABLEPRESS_ABSPATH' ) && class_exists('\\TablePress') ) {
+				$ignore_shortcodes[] = 'table';
+			}
+
+			foreach ( $ignore_shortcodes as $shortcode ) {
+				$this->temporaryDisableShortcode($shortcode);
+			}
+
+			$content = do_shortcode( $content );
+
+			// WP 4.2 emoji strip
+			if ( function_exists( 'wp_encode_emoji' ) ) {
+				$content = wp_encode_emoji( $content );
+			}
+
+			if ( defined( 'TABLEPRESS_ABSPATH' ) ) {
+				unset( $tp_controller );
+			}
+
+			if ( defined( 'WP_TABLE_RELOADED_ABSPATH' ) ) {
+				unset( $wpt_reloaded );
+			}
+
+			$this->enableDisabledShortcodes();
+
+			return apply_filters( 'asp_index_after_shortcode_execution', $content, $post );
+		}
+
+		private function temporaryDisableShortcode($tag) {
+			global $shortcode_tags;
+			if ( array_key_exists( $tag, $shortcode_tags ) ) {
+				$this->temporary_shortcode_tags[$tag] = $shortcode_tags[$tag];
+				$shortcode_tags[ $tag ] = array( $this, 'return_empty_string' );
+			}
+		}
+
+		private function enableDisabledShortcodes() {
+			global $shortcode_tags;
+			foreach ($this->temporary_shortcode_tags as $tag => $callback) {
+				$shortcode_tags[$tag] = $callback;
+			}
+		}
+
+		private function parseTablePressShortcodes( $content ): string {
+			$regex = '/\[table[^\]]*id\=[\'"]{0,1}([0-9]+)[\'"]{0,1}?[^\]]*\]/';
+			$tables = json_decode(get_option('tablepress_tables'), true);
+			if ( !is_null($tables) && isset($tables['table_post']) && preg_match_all($regex, $content, $matches) > 0) {
+				$return = array();
+				foreach ( $matches[1] as $table_id ) {
+					$data = json_decode( get_post_field('post_content', $tables['table_post'][$table_id]), true );
+					if ( $data !== null ) {
+						$return[] = Str::anyToString($data);
+					}
+				}
+				return implode(' ', $return);
+			}
+
+			return '';
+		}
+
+		/**
+		 * An empty function to override individual shortcodes. This must be a public method.
+		 *
+		 * @return string
+		 * @noinspection PhpUnused
+		 */
+		function return_empty_string(): string {
+			return "";
+		}
+	}
+}

wp-content/plugins/ajax-search-pro/src/server/Index/Tokenizer.php

@@ -0,0 +1,590 @@
+<?php
+namespace WPDRMS\ASP\Index;
+
+use WPDRMS\ASP\Utils\Html;
+use WPDRMS\ASP\Utils\Inflect\InflectController;
+use WPDRMS\ASP\Utils\MB;
+use WPDRMS\ASP\Utils\Str;
+
+/**
+ * @phpstan-type TokenizerArgs array{
+ *      min_word_length: int,
+ *      use_stopwords: bool,
+ *      stopwords: string[],
+ *      synonyms_as_keywords: bool,
+ *      inflect: bool,
+ *      language: string,
+ *  }
+ */
+class Tokenizer {
+	/**
+	 * @var string unique random string for special replacements
+	 */
+	private string $randstr = 'wpivdny3htnydqd6mlyg';
+
+	/**
+	 * @var TokenizerArgs
+	 */
+	private array $args;
+
+	/**
+	 * @var string[]
+	 */
+	public static array $additional_keywords_pattern = array(
+		'"',
+		"'",
+		'`',
+		'’',
+		'‘',
+		'”',
+		'“',
+		'«',
+		'»',
+		'+',
+		'.',
+		',',
+		':',
+		'-',
+		'_',
+		'=',
+		'%',
+		'(',
+		')',
+		'{',
+		'}',
+		'*',
+		'[',
+		']',
+		'|',
+		'&',
+		'/',
+	);
+
+	/**
+	 * @param TokenizerArgs $args
+	 */
+	public function __construct( array $args ) {
+		$defaults   = array(
+			'min_word_length'      => 2,
+			'use_stopwords'        => false,
+			'stopwords'            => array(),
+			'synonyms_as_keywords' => false,
+			'inflect'              => false,
+			'language'             => 'english',
+		);
+		$this->args = wp_parse_args( $args, $defaults ); // @phpstan-ignore-line
+	}
+
+	public function setLanguage(string $language): void {
+		$this->args['language'] = $language;
+	}
+
+	/**
+	 * Performs a simple trimming, character removal on a string
+	 *
+	 * @param $str
+	 * @param $post
+	 * @return string
+	 */
+	function tokenizeSimple( $str, $post ): string {
+		if ( function_exists( 'mb_internal_encoding' ) ) {
+			mb_internal_encoding( 'UTF-8' );
+		}
+
+		$str = Str::anyToString($str);
+		$str = Html::toTxt( $str );
+		$str = strip_tags( $str );
+		$str = stripslashes( $str );
+		// Non breakable spaces to regular spaces
+		$str = preg_replace('/\xc2\xa0/', ' ', $str);
+		$str = preg_replace( '/[[:space:]]+/', ' ', $str );
+
+		$str = str_replace( array( "\n", "\r", '  ' ), ' ', $str );
+		// Turkish uppercase I does not lowercase correctly
+		$str = str_replace( 'İ', 'i', $str );
+		$str = MB::strtolower( $str );
+		$str = trim($str);
+
+		$str = Content::hebrewUnvocalize($str);
+		$str = Content::arabicRemoveDiacritics($str);
+
+		$negative_keywords = $this->getNegativeWords($post);
+		foreach ( $negative_keywords as $negative_keyword ) {
+			// If there is a negative keyword within, this case is over
+			if ( strpos($str, $negative_keyword) !== false ) {
+				return '';
+			}
+		}
+
+		$stop_words = $this->getStopWords();
+		foreach ( $stop_words as $stop_word ) {
+			// whole word matches only
+			if ( preg_match('/\b' . $stop_word . '\b/', $str, $match, PREG_OFFSET_CAPTURE) ) {
+				return '';
+			}
+		}
+
+		return $str;
+	}
+
+	/**
+	 * Performs a simple trimming, character removal on a string, but returns array of keywords
+	 * by the separator character
+	 *
+	 * @param $str
+	 * @param $post
+	 * @return array
+	 */
+	function tokenizePhrases( $str, $post, $word_separator = ',' ): array {
+		if ( function_exists( 'mb_internal_encoding' ) ) {
+			mb_internal_encoding( 'UTF-8' );
+		}
+
+		$args = $this->args;
+
+		$str = Str::anyToString($str);
+		$str = Html::toTxt( $str );
+		$str = strip_tags( $str );
+		$str = stripslashes( $str );
+		// Non breakable spaces to regular spaces
+		$str = preg_replace('/\xc2\xa0/', ' ', $str);
+		$str = preg_replace( '/[[:space:]]+/', ' ', $str );
+
+		$str = str_replace( array( "\n", "\r", '  ' ), ' ', $str );
+		// Turkish uppercase I does not lowercase correctly
+		$str = str_replace( 'İ', 'i', $str );
+		$str = MB::strtolower( $str );
+		$str = trim($str);
+
+		$str = Content::hebrewUnvocalize($str);
+		$str = Content::arabicRemoveDiacritics($str);
+
+		$words = explode($word_separator, $str);
+		$words = array_map('trim', $words);
+		$words = array_filter(
+			$words,
+			function ( $word ) {
+				return \WPDRMS\ASP\Utils\MB::strlen($word);
+			}
+		);
+
+		$keywords = array();
+
+		while ( ( $c_word = array_shift($words) ) !== null ) {
+			$c_word = trim($c_word);
+
+			if ( $c_word == '' || MB::strlen($c_word) < $args['min_word_length'] ) {
+				continue;
+			}
+
+			// Numerics won't work otherwise, need to trim that later
+			if ( is_numeric($c_word) ) {
+				$c_word = ' ' . $c_word;
+			}
+
+			if ( array_key_exists($c_word, $keywords) ) {
+				++$keywords[ $c_word ][1];
+			} else {
+				$keywords[ $c_word ] = array( $c_word, 1 );
+			}
+		}
+		unset($c_word);
+
+		return $keywords;
+	}
+
+	/**
+	 * Performs a keyword extraction on the given content string.
+	 *
+	 * @return array of keywords $keyword = array( 'keyword', {count} )
+	 */
+	function tokenize( $str, $post = false, $lang = '' ): array {
+
+		if ( is_array( $str ) ) {
+			$str = Str::anyToString( $str );
+		}
+		if ( function_exists('mb_strlen') ) {
+			$fn_strlen = 'mb_strlen';
+		} else {
+			$fn_strlen = 'strlen';
+		}
+
+		$args = $this->args;
+
+		if ( function_exists( 'mb_internal_encoding' ) ) {
+			mb_internal_encoding( 'UTF-8' );
+		}
+
+		$str = apply_filters( 'asp_indexing_string_pre_process', $str );
+
+		$str = Html::toTxt( $str );
+		$str = wp_specialchars_decode( $str );
+
+		$str = strip_tags( $str );
+		$str = stripslashes( $str );
+
+		// Replace non-word boundary dots with a unique string + 'd'
+		/** @noinspection RegExpRedundantEscape */
+		$str = preg_replace('/([0-9])[\.]([0-9])/', '$1' . $this->randstr . 'd$2', $str);
+
+		// Remove potentially dangerous or unusable characters
+		$str = str_replace(
+			array(
+				'·',
+				'…',
+				'€',
+				'&shy;',
+				'·',
+				'…',
+				'®',
+				'©',
+				'™',
+				"\xC2\xAD",
+			),
+			'',
+			$str 
+		);
+		$str = str_replace(
+			array(
+				'. ', // dot followed by space as boundary, otherwise it might be a part of the word
+				', ', // comma followed by space only, otherwise it might be a word part
+				'<',
+				'>',
+				'†',
+				'‡',
+				'‰',
+				'‹',
+				'™',
+				'¡',
+				'¢',
+				'¤',
+				'¥',
+				'¦',
+				'§',
+				'¨',
+				'©',
+				'ª',
+				'«',
+				'¬',
+				'®',
+				'¯',
+				'°',
+				'±',
+				'¹',
+				'²',
+				'³',
+				'¶',
+				'·',
+				'º',
+				'»',
+				'¼',
+				'½',
+				'¾',
+				'¿',
+				'÷',
+				'•',
+				'…',
+				'←',
+				'←',
+				'↑',
+				'→',
+				'↓',
+				'↔',
+				'↵',
+				'⇐',
+				'⇑',
+				'⇒',
+				'⇓',
+				'⇔',
+				'√',
+				'∝',
+				'∞',
+				'∠',
+				'∧',
+				'∨',
+				'∂',
+				'∃',
+				'∅',
+				'∗',
+				'∩',
+				'∪',
+				'∫',
+				'∴',
+				'∼',
+				'≅',
+				'≈',
+				'≠',
+				'≡',
+				'≤',
+				'≥',
+				'⊂',
+				'⊃',
+				'⊄',
+				'⊆',
+				'⊇',
+				'⊕',
+				'⊗',
+				'⊥',
+				'◊',
+				'♠',
+				'♣',
+				'♥',
+				'♦',
+				'🔴',
+				'​',
+				'◊',
+				'〈',
+				'〉',
+				'⌊',
+				'⌋',
+				'⌈',
+				'⌉',
+				'⋅',
+				'ˇ',
+				'°',
+				'~',
+				'Ë›',
+				'Ă‹ĹĄ',
+				'¸',
+				'§',
+				'¨',
+				'’',
+				'‘',
+				'”',
+				'“',
+				'„',
+				'´',
+				'—',
+				'–',
+				'Ă—',
+				'&#8217;',
+				'&#128308;',
+				'&nbsp;',
+				"\n",
+				"\r",
+				'& ',
+				'\\',
+				'^',
+				'?',
+				'!',
+				';',
+				chr( 194 ) . chr( 160 ),
+			),
+			' ',
+			$str 
+		);
+		$str = str_replace( 'Ăź', 'ss', $str );
+
+		// Turkish uppercase I does not lowercase correctly
+		$special_replace = array(
+			'İ' => 'i',
+			'—' => '-',
+		);
+		$str             = str_replace( array_keys($special_replace), array_values($special_replace), $str );
+
+		// Any yet undefined punctuation
+		// $str = preg_replace( '/[[:punct:]]+/u', ' ', $str );
+		// Non breakable spaces to regular spaces
+		$str = preg_replace('/\xc2\xa0/', ' ', $str);
+		// Any remaining multiple space characters
+		$str = preg_replace( '/[[:space:]]+/', ' ', $str );
+
+		$str = MB::strtolower($str);
+
+		$str = Content::hebrewUnvocalize($str);
+		$str = Content::arabicRemoveDiacritics($str);
+
+		// $str = preg_replace('/[^\p{L}0-9 ]/', ' ', $str);
+		$str = str_replace( "\xEF\xBB\xBF", '', $str );
+
+		$str = trim( preg_replace( '/\s+/', ' ', $str ) );
+
+		// Set back the non-word boundary dots
+		$str = str_replace( $this->randstr . 'd', '.', $str );
+
+		$str = apply_filters( 'asp_indexing_string_post_process', $str );
+
+		$words = explode( ' ', $str );
+
+		// Remove punctuation marks + some extra from the start and the end of words
+
+		// Characters, which should not be standalone (but can be in start on end)
+		$non_standalone_strings = array( '$', '€', '£', '%' );
+		// Additional keywords, should not be standalone
+		$additional_keywords_string = implode('', array_diff(self::$additional_keywords_pattern, $non_standalone_strings));
+		foreach ( $words as $wk => &$ww ) {
+			$ww = MB::trim($ww, $additional_keywords_string);
+			if ( $ww == '' || in_array($ww, $non_standalone_strings ) ) {
+				unset($words[ $wk ]);
+			}
+		}
+		unset($wk);
+		unset($ww);
+
+		// Get additional words if available
+		$additional_words = array();
+		$started          = microtime(true);
+		foreach ( $words as $ww ) {
+			// This operation can be costly, so limit to 3 seconds just to be sure
+			if ( ( microtime(true) - $started ) > 3 ) {
+				break;
+			}
+
+			// ex.: 123-45-678 to 123, 45, 678
+			$ww1 = str_replace(self::$additional_keywords_pattern, ' ', $ww);
+			$wa  = explode(' ', $ww1);
+			if ( count($wa) > 1 ) {
+				foreach ( $wa as $wak => $wav ) {
+					$wav = trim(preg_replace( '/[[:space:]]+/', ' ', $wav ));
+					if ( $wav != '' && !in_array($wav, $words) ) {
+						$wa[ $wak ] = $wav;
+					} else {
+						unset($wa[ $wak ]);
+					}
+				}
+				$additional_words = array_merge($additional_words, $wa);
+			}
+			// ex.: 123-45-678 to 12345678
+			$ww2 = str_replace(self::$additional_keywords_pattern, '', $ww);
+			if ( $ww2 !== '' && $ww2 != $ww && !in_array($ww2, $words) && !in_array($ww2, $additional_words) ) {
+				$additional_words[] = $ww2;
+			}
+
+			// Accent removal and transliteration
+			$transliterated = str_replace(array( 'ʾ', "'", '"' ), '', Str::removeAccents($ww));
+			if (
+				$transliterated !== '' &&
+				$transliterated !== $ww &&
+				!in_array($transliterated, $words, true) &&
+				!in_array($transliterated, $additional_words, true)
+			) {
+				$additional_words[] = $transliterated;
+			}
+		}
+
+		// Append them after the words array
+		$words = array_merge($words, $additional_words);
+		// Inflections
+		if ( $this->args['inflect'] && !empty($words) ) {
+			$words = array_merge(
+				$words,
+				InflectController::instance()->get($words, $this->args['language'])
+			);
+		}
+
+		/**
+		 * Apply synonyms for the whole string instead of the words, because
+		 * synonyms can be multi-keyword phrases too
+		 */
+		$syn_inst = \WPDRMS\ASP\Synonyms\Manager::getInstance();
+		if ( $syn_inst->exists() ) {
+			if ( $this->args['synonyms_as_keywords'] == 1 ) {
+				$syn_inst->synonymsAsKeywords();
+			}
+			$additional_words_by_synonyms = array();
+			$synonyms                     = $syn_inst->get();
+
+			// If the langauge is set
+			if ( $lang != '' && isset($synonyms[ $lang ]) ) {
+				foreach ( $synonyms[ $lang ] as $keyword => $synonyms_arr ) {
+					if ( preg_match('/\b' . preg_quote($keyword) . '\b/u', $str) ) {
+						$additional_words_by_synonyms = array_merge($additional_words_by_synonyms, $synonyms_arr);
+					}
+				}
+			}
+			unset($keyword, $synonyms_arr);
+
+			// Also for the "default" aka "any"
+			if ( isset($synonyms['default']) ) {
+				foreach ( $synonyms['default'] as $keyword => $synonyms_arr ) {
+					if ( preg_match('/\b' . preg_quote($keyword) . '\b/u', $str) ) {
+						$additional_words_by_synonyms = array_merge($additional_words_by_synonyms, $synonyms_arr);
+					}
+				}
+			}
+
+			if ( count($additional_words_by_synonyms) > 0 ) {
+				$words = array_merge($words, $additional_words_by_synonyms);
+			}
+		}
+
+		$stop_words        = $this->getStopWords();
+		$negative_keywords = $this->getNegativeWords($post);
+
+		$keywords = array();
+
+		foreach ( $words as $c_word ) {
+			$c_word = trim($c_word);
+
+			if ( $c_word == '' || $fn_strlen($c_word) < $args['min_word_length'] ) {
+				continue;
+			}
+
+			// Only whole word matches for stop-words
+			if ( !empty($stop_words) && in_array($c_word, $stop_words) ) {
+				continue;
+			}
+
+			// Partial matches for negative keywords
+			foreach ( $negative_keywords as $negative_keyword ) {
+				if ( strpos($c_word, $negative_keyword) !== false ) {
+					continue 2;
+				}
+			}
+
+			// Numerics won't work otherwise, need to trim that later
+			if ( is_numeric($c_word) ) {
+				$c_word = ' ' . $c_word;
+			}
+
+			if ( array_key_exists($c_word, $keywords) ) {
+				++$keywords[ $c_word ][1];
+			} else {
+				$keywords[ $c_word ] = array( $c_word, 1 );
+			}
+		}
+		unset($c_word);
+
+		return apply_filters( 'asp_indexing_keywords', $keywords );
+	}
+
+
+	/**
+	 * Returns the stop words
+	 */
+	private function getStopWords(): array {
+		$stop_words = array();
+		// Only compare to common words if $restrict is set to false
+		if ( $this->args['use_stopwords'] == 1 && $this->args['stopwords'] != '' ) {
+			$this->args['stopwords'] = str_replace(' ', '', $this->args['stopwords']);
+			$stop_words              = explode( ',', $this->args['stopwords'] );
+		}
+		$stop_words = array_unique( $stop_words );
+		foreach ( $stop_words as $sk => &$sv ) {
+			$sv = trim($sv);
+			if ( $sv == '' || MB::strlen($sv) < $this->args['min_word_length'] ) {
+				unset($stop_words[ $sk ]);
+			}
+		}
+
+		return $stop_words;
+	}
+
+	/**
+	 * Returns negative keywords for the current post object
+	 */
+	private function getNegativeWords( $post ): array {
+		// Post level stop-words, negative keywords
+		if ( $post !== false ) {
+			$negative_keywords = get_post_meta($post->ID, '_asp_negative_keywords', true);
+			if ( !empty($negative_keywords) ) {
+				$negative_keywords = trim( preg_replace('/\s+/', ' ', $negative_keywords) );
+				$negative_keywords = explode(' ', $negative_keywords);
+				$negative_keywords = array_filter($negative_keywords, fn( $keyword )=>$keyword !=='');
+				return array_unique($negative_keywords);
+			}
+		}
+
+		return array();
+	}
+}