Merge branch 'develop'

mecha · mecha · commit 34f842c00433 · 2022-07-20T13:52:36.000+02:00
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -4,6 +4,10 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](http://keepachangelog.com/)
 and this project adheres to [Semantic Versioning](http://semver.org/).
 
+## [4.21.1] - 2022-07-20
+### Fixed
+* An out-of-memory PHP error when importing items.
+
 ## [4.21] - 2022-07-13
 ### Change
 * Updated Twig to v1.42.2, to support PHP 8 or later.
diff --git a/composer.json b/composer.json
@@ -14,6 +14,7 @@
     "require": {
         "php": "^5.4 | ^7.0",
         "ext-curl": "*",
+        "ext-json": "*",
         "ext-simplexml": "*",
         "psr/log": "^1.1",
         "twig/twig": "^1.0 | ^2.10",
diff --git a/includes/feed-importing.php b/includes/feed-importing.php
@@ -23,6 +23,8 @@
  * Called on hook 'wprss_fetch_single_feed_hook'.
  *
  * @since 3.2
+ *
+ * @throws Exception
  */
 function wprss_fetch_insert_single_feed_items( $feed_ID ) {
     set_transient('wpra/feeds/importing/' . $feed_ID, true, 0);
@@ -104,56 +106,72 @@ function wprss_fetch_insert_single_feed_items( $feed_ID ) {
                 ? wprss_get_general_setting('unique_titles')
                 : $unique_titles_only;
             $unique_titles_only = filter_var($unique_titles_only, FILTER_VALIDATE_BOOLEAN);
-            // Gather the titles of the items that are imported
-            // The import process will check not only the titles in the DB but the titles currently in the feed
-            $existing_titles = [];
 
             // Gather the existing feed item IDs for this feed source
             $useGuids = get_post_meta($feed_ID, 'wprss_use_guids', true);
             $useGuids = filter_var($useGuids, FILTER_VALIDATE_BOOLEAN);
-            $existingIds = $useGuids
-                ? wprss_get_existing_guids()
-                : wprss_get_existing_permalinks();
+
+            // Gather the IDs and titles of the items that are imported
+            // The import process will not only check the IDs and titles against the DB, but also against the feed
+            // itself. This prevents duplicate items in the feed from importing duplicates.
+            $existingIds = [];
+            $existingTitles = [];
 
             // Generate a list of items fetched, that are not already in the DB
             $new_items = array();
             foreach ( $items_to_insert as $item ) {
-                $item_title = $item->get_title();
+                $itemTitle = $item->get_title();
                 $guid = $item->get_id();
                 $permalink = $item->get_permalink();
                 $permalink = wprss_normalize_permalink( $permalink, $item, $feed_ID );
 
                 // Check if blacklisted
                 if (wprss_is_blacklisted($permalink)) {
-                    $logger->debug('Item "{0}" is blacklisted', [$item_title]);
-
+                    $logger->debug('Item "{0}" is blacklisted', [$itemTitle]);
                     continue;
                 }
 
-                // Check if already imported
-                $idToCheck = $useGuids ? $guid : $permalink;
-                if (array_key_exists($idToCheck, $existingIds)) {
-                    $logger->debug('Item "{0}" already exists in the database', [$item_title]);
+                $itemId = $useGuids ? $guid : $permalink;
+
+                // Check if already imported in database
+                //-----------------------------------------
+                $itemIdExists = $useGuids ? wprss_guid_exists($guid) : wprss_permalink_exists($permalink);
+                $itemsTitleExists = $unique_titles_only && wprss_item_title_exists($item->get_title());
+
+                if ($itemIdExists || $itemsTitleExists) {
+                    $reason = $itemIdExists
+                        ? ($useGuids ? 'GUID' : 'permalink')
+                        : 'Non-unique title';
+
+                    $logger->debug('Item "{title}" already exists in the database. Reason: {reason}', [
+                        'title' => $itemTitle,
+                        'reason' => $reason
+                    ]);
 
                     continue;
                 }
 
-                // Check if title exists (if the option is enabled)
-                if ($unique_titles_only) {
-                    $title_exists_db = wprss_item_title_exists($item->get_title());
-                    $title_exists_feed = array_key_exists($item_title, $existing_titles);
-                    $title_exists = $title_exists_db || $title_exists_feed;
-                    // Add this item's title to the list to check against
-                    $existing_titles[$item_title] = 1;
+                // Check if item is duplicated in the feed
+                //-----------------------------------------
+                $itemIdIsDuped = array_key_exists($itemId, $existingIds);
+                $itemTitleIsDuped = $unique_titles_only && array_key_exists($itemTitle, $existingTitles);
 
-                    if ($title_exists) {
-                        $logger->debug('An item with the title "{0}" already exists', [$item_title]);
+                if ($itemIdIsDuped || $itemTitleIsDuped) {
+                    $reason = $itemIdIsDuped
+                        ? ($useGuids ? 'GUID' : 'permalink')
+                        : 'Non-unique title';
 
-                        continue;
-                    }
+                    $logger->debug('Item "{title}" is duplicated in the feed. Reason: {reason}', [
+                        'title' => $itemTitle,
+                        'reason' => $reason,
+                    ]);
+
+                    continue;
+                } else {
+                    $existingIds[$itemId] = 1;
+                    $existingTitles[$itemTitle] = 1;
                 }
 
-                $existingIds[$idToCheck] = 1;
                 $new_items[] = $item;
             }
 
@@ -184,14 +202,14 @@ function wprss_fetch_insert_single_feed_items( $feed_ID ) {
                         ? 0
                         : $num_new_items - $num_can_insert;
 
-                // Get an array with the DB feed items in reverse order (oldest first)
+                // Get an array with the DB feed items in reverse order (the oldest first)
                 $db_feed_items_reversed = array_reverse( $db_feed_items->posts );
                 // Cut the array to get only the first few that are to be deleted ( equal to $num_feed_items_to_delete )
                 $feed_items_to_delete = array_slice( $db_feed_items_reversed, 0, $num_feed_items_to_delete );
 
                 // Iterate the feed items and delete them
                 $num_items_deleted = 0;
-                foreach ( $feed_items_to_delete as $key => $post ) {
+                foreach ( $feed_items_to_delete as $post ) {
                     wp_delete_post( $post->ID, TRUE );
                     $num_items_deleted++;
                 }
@@ -201,7 +219,7 @@ function wprss_fetch_insert_single_feed_items( $feed_ID ) {
                 }
             }
 
-            update_post_meta( $feed_ID, 'wprss_last_update', $last_update_time = time() );
+            update_post_meta( $feed_ID, 'wprss_last_update', time() );
             update_post_meta( $feed_ID, 'wprss_last_update_items', 0 );
 
             // Insert the items into the db
@@ -270,34 +288,31 @@ function wprss_get_feed_items( $feed_url, $source, $force_feed = FALSE ) {
 }
 
 if (defined('WP_DEBUG') && WP_DEBUG) {
-    add_action('cron_request', 'wprss_cron_add_xdebug_cookie', 10);
-}
-
-/**
- * Allow debugging of wp_cron jobs using xDebug.
- *
- * This is done by taking the XDEBUG cookie received from the browser (which enables an xDebug session) and passing it
- * to WP Cron. That way, code initiated from a cron job will be debuggable.
- *
- * @param array $cronRequest
- *
- * @return array $cron_request_array with the current XDEBUG_SESSION cookie added if set
- */
-function wprss_cron_add_xdebug_cookie($cronRequest)
-{
-    if (empty($_COOKIE['XDEBUG_SESSION'])) {
-        return ($cronRequest);
-    }
+    /**
+     * Allow debugging of wp_cron jobs using xDebug.
+     *
+     * This is done by taking the XDEBUG cookie received from the browser (which enables an xDebug session) and passing it
+     * to WP Cron. That way, code initiated from a cron job will be debuggable.
+     *
+     * @param array $cronRequest
+     *
+     * @return array $cron_request_array with the current XDEBUG_SESSION cookie added if set
+     */
+    add_action('cron_request', function($cronRequest) {
+        if (empty($_COOKIE['XDEBUG_SESSION'])) {
+            return ($cronRequest);
+        }
 
-    $cookie = filter_var($_COOKIE['XDEBUG_SESSION'], FILTER_SANITIZE_STRING);
+        $cookie = filter_var($_COOKIE['XDEBUG_SESSION'], FILTER_SANITIZE_STRING);
 
-    if (empty($cronRequest['args']['cookies'])) {
-        $cronRequest['args']['cookies'] = [];
-    }
+        if (empty($cronRequest['args']['cookies'])) {
+            $cronRequest['args']['cookies'] = [];
+        }
 
-    $cronRequest['args']['cookies']['XDEBUG_SESSION'] = $cookie;
+        $cronRequest['args']['cookies']['XDEBUG_SESSION'] = $cookie;
 
-    return $cronRequest;
+        return $cronRequest;
+    });
 }
 
 /**
@@ -364,7 +379,7 @@ function wprss_fetch_feed($url, $source = null, $param_force_feed = false)
 
     // If a feed source was passed
     if ($source !== null || $param_force_feed) {
-        // Get the force feed option for the feed source
+        // Get the force-feed option for the feed source
         $force_feed = get_post_meta($source, 'wprss_force_feed', true);
         // If turned on, force the feed
         if ($force_feed == 'true' || $param_force_feed) {
diff --git a/includes/feed-processing.php b/includes/feed-processing.php
@@ -120,39 +120,36 @@ function wprss_get_feed_items_for_source($source_id)
     return wprss_get_imported_items($source_id);
 }
 
-/**
- * Queries the DB to get the GUIDs for existing feed items.
- */
-function wprss_get_existing_guids() {
+/** Checks if a permalink exists in the database. */
+function wprss_permalink_exists($permalink) {
     global $wpdb;
 
-    $cols = $wpdb->get_col(
-        "SELECT q.`meta_value`
-        FROM {$wpdb->postmeta} AS p
-        JOIN {$wpdb->postmeta} AS q
-            ON (q.`meta_key` = 'wprss_item_guid' AND p.`post_id` = q.`post_id`)"
+    $numRows = $wpdb->query(
+        $wpdb->prepare(
+            "SELECT `meta_value`
+            FROM {$wpdb->postmeta}
+            WHERE (`meta_key` = 'wprss_item_permalink' AND `meta_value` = %s)",
+            $permalink
+        )
     );
 
-    return @array_flip($cols);
+    return !!$numRows;
 }
 
-/**
- * Database query to get existing permalinks
- *
- * @since 3.0
- */
-function wprss_get_existing_permalinks()
-{
+/** Checks if a GUID exists in the database. */
+function wprss_guid_exists($guid) {
     global $wpdb;
 
-    $cols = $wpdb->get_col(
-        "SELECT q.`meta_value`
-        FROM {$wpdb->postmeta} AS p
-        JOIN {$wpdb->postmeta} AS q
-            ON (q.`meta_key` = 'wprss_item_permalink' AND p.`post_id` = q.`post_id`)"
+    $numRows = $wpdb->query(
+        $wpdb->prepare(
+            "SELECT `meta_value`
+            FROM {$wpdb->postmeta}
+            WHERE (`meta_key` = 'wprss_item_guid' AND `meta_value` = %s)",
+            $guid
+        )
     );
 
-    return @array_flip($cols);
+    return !!$numRows;
 }
 
 /**
@@ -168,37 +165,17 @@ function wprss_item_title_exists($title)
 {
     global $wpdb;
 
-    $query = $wpdb->prepare(
-        "SELECT *
-        FROM `{$wpdb->posts}` AS p
-        JOIN `{$wpdb->postmeta}` AS q
-            ON p.`ID` = q.`post_id`
-        WHERE q.`meta_key` = 'wprss_feed_id' AND p.`post_title` = %s",
-        [html_entity_decode($title)]
-    );
-
-    $cols = $wpdb->get_col($query);
-
-    return count($cols) > 0;
-}
-
-/**
- * Database query to get existing titles
- *
- * @since 4.7
- */
-function wprss_get_existing_titles()
-{
-    global $wpdb;
-
-    $cols = $wpdb->get_col(
-        "SELECT p.`post_title`
-        FROM `{$wpdb->posts}` AS p
-        JOIN `{$wpdb->postmeta}` AS q
-            ON p.`ID` = q.`post_id`"
+    $numRows = $wpdb->query(
+        $wpdb->prepare(
+            "SELECT p.`post_title`
+            FROM {$wpdb->posts} AS p
+            JOIN {$wpdb->postmeta} AS q ON p.`ID` = q.`post_id`
+            WHERE q.`meta_key` = 'wprss_feed_id' AND p.`post_title` = %s",
+            [html_entity_decode($title)]
+        )
     );
 
-    return @array_flip($cols);
+    return !!$numRows;
 }
 
 add_action('publish_wprss_feed', 'wprss_fetch_insert_feed_items', 10);
diff --git a/wp-rss-aggregator.php b/wp-rss-aggregator.php
@@ -4,7 +4,7 @@
  * Plugin Name: WP RSS Aggregator
  * Plugin URI: https://www.wprssaggregator.com/#utm_source=wpadmin&utm_medium=plugin&utm_campaign=wpraplugin
  * Description: Imports and aggregates multiple RSS Feeds.
- * Version: 4.21
+ * Version: 4.21.1
  * Author: RebelCode
  * Author URI: https://www.wprssaggregator.com
  * Text Domain: wprss
@@ -76,7 +76,7 @@
 
 // Set the version number of the plugin.
 if( !defined( 'WPRSS_VERSION' ) )
-    define( 'WPRSS_VERSION', '4.21' );
+    define( 'WPRSS_VERSION', '4.21.1' );
 
 if( !defined( 'WPRSS_WP_MIN_VERSION' ) )
     define( 'WPRSS_WP_MIN_VERSION', '4.8' );
