Improved word censor 🤐 (#1393)

* Improved default word censor

* Updated

* Updated

* Updated

* Update Censor.php

* Update CHANGELOG.md

* Updated

CHANGELOG.md

@@ -11,6 +11,7 @@
 - Removed `wbr()` [\#1387](https://github.com/torrentpier/torrentpier/pull/1387) ([belomaxorka](https://github.com/belomaxorka))
 - Removed converting for legacy md5 passwords [\#1386](https://github.com/torrentpier/torrentpier/pull/1386) ([belomaxorka](https://github.com/belomaxorka))
 - Fixed issue with poll_users cleaning at every cron job startup [\#1390](https://github.com/torrentpier/torrentpier/pull/1390) ([belomaxorka](https://github.com/belomaxorka))
+- Improved word censor 🤐 [\#1393](https://github.com/torrentpier/torrentpier/pull/1393) ([belomaxorka](https://github.com/belomaxorka))
 - Used hashing for filenames generation [\#1385](https://github.com/torrentpier/torrentpier/pull/1385) ([belomaxorka](https://github.com/belomaxorka))
 - Minor improvements [\#1382](https://github.com/torrentpier/torrentpier/pull/1382), [\#1383](https://github.com/torrentpier/torrentpier/pull/1383), [\#1391](https://github.com/torrentpier/torrentpier/pull/1391) ([belomaxorka](https://github.com/belomaxorka))
 - Some bugfixes [\#1380](https://github.com/torrentpier/torrentpier/pull/1380) ([belomaxorka](https://github.com/belomaxorka))

index.php

@@ -327,7 +327,7 @@ if ($bb_cfg['show_latest_news']) {
     foreach ($latest_news as $news) {
         $template->assign_block_vars('news', [
             'NEWS_TOPIC_ID' => $news['topic_id'],
-            'NEWS_TITLE' => str_short($news['topic_title'], $bb_cfg['max_news_title']),
+            'NEWS_TITLE' => str_short($wordCensor->censorString($news['topic_title']), $bb_cfg['max_news_title']),
             'NEWS_TIME' => bb_date($news['topic_time'], 'd-M', false),
             'NEWS_IS_NEW' => is_unread($news['topic_time'], $news['topic_id'], $news['forum_id']),
         ]);
@@ -346,7 +346,7 @@ if ($bb_cfg['show_network_news']) {
     foreach ($network_news as $net) {
         $template->assign_block_vars('net', [
             'NEWS_TOPIC_ID' => $net['topic_id'],
-            'NEWS_TITLE' => str_short($net['topic_title'], $bb_cfg['max_net_title']),
+            'NEWS_TITLE' => str_short($wordCensor->censorString($net['topic_title']), $bb_cfg['max_net_title']),
             'NEWS_TIME' => bb_date($net['topic_time'], 'd-M', false),
             'NEWS_IS_NEW' => is_unread($net['topic_time'], $net['topic_id'], $net['forum_id']),
         ]);

library/ajax/posts.php

@@ -47,13 +47,6 @@ if (isset($this->request['post_id'])) {
     $is_auth = auth(AUTH_ALL, $post['forum_id'], $userdata, $post);
 }
 
-if (!defined('WORD_LIST_OBTAINED')) {
-    $orig_word = [];
-    $replace_word = [];
-    obtain_word_list($orig_word, $replace_word);
-    define('WORD_LIST_OBTAINED', true);
-}
-
 switch ($this->request['type']) {
     case 'delete':
         if ($post['post_id'] != $post['topic_first_post_id'] && $is_auth['auth_delete'] && ($is_auth['auth_mod'] || ($userdata['user_id'] == $post['poster_id'] && $post['topic_last_post_id'] == $post['post_id'] && $post['post_time'] + 3600 * 3 > TIMENOW))) {
@@ -83,13 +76,10 @@ switch ($this->request['type']) {
         $message = "[quote=\"" . $quote_username . "\"][qpost=" . $post['post_id'] . "]" . $post['post_text'] . "[/quote]\r";
 
         // hide user passkey
-        $message = preg_replace('#(?<=\?uk=)[a-zA-Z0-9]{10}(?=&)#', 'passkey', $message);
+        $message = preg_replace('#(?<=\?uk=)[a-zA-Z0-9](?=&)#', 'passkey', $message);
         // hide sid
-        $message = preg_replace('#(?<=[\?&;]sid=)[a-zA-Z0-9]{12}#', 'sid', $message);
+        $message = preg_replace('#(?<=[\?&;]sid=)[a-zA-Z0-9]#', 'sid', $message);
-
+        $message = $wordCensor->censorString($message);
-        if (!empty($orig_word)) {
-            $message = (!empty($message)) ? preg_replace($orig_word, $replace_word, $message) : '';
-        }
 
         if ($post['post_id'] == $post['topic_first_post_id']) {
             $message = "[quote]" . $post['topic_title'] . "[/quote]\r";

library/includes/bbcode.php

@@ -388,17 +388,12 @@ function add_search_words($post_id, $post_message, $topic_title = '', $only_retu
 
 function bbcode2html($text)
 {
-    global $bbcode;
+    global $bbcode, $wordCensor;
 
     if (!isset($bbcode)) {
         $bbcode = new TorrentPier\Legacy\BBCode();
     }
-    $orig_word = [];
+    $text = $wordCensor->censorString($text);
-    $replacement_word = [];
-    obtain_word_list($orig_word, $replacement_word);
-    if (count($orig_word)) {
-        $text = preg_replace($orig_word, $replacement_word, $text);
-    }
     return $bbcode->bbcode2html($text);
 }
 

library/includes/functions.php

@@ -1606,36 +1606,6 @@ function bb_preg_quote($str, $delimiter)
     return $text;
 }
 
-//
-// Obtain list of naughty words and build preg style replacement arrays for use by the
-// calling script, note that the vars are passed as references this just makes it easier
-// to return both sets of arrays
-//
-function obtain_word_list(&$orig_word, &$replacement_word)
-{
-    global $bb_cfg;
-
-    if (!$bb_cfg['use_word_censor']) {
-        return false;
-    }
-
-    if (!$sql = CACHE('bb_cache')->get('censored')) {
-        $sql = DB()->fetch_rowset("SELECT word, replacement FROM " . BB_WORDS);
-        if (!$sql) {
-            $sql = [['word' => 1, 'replacement' => 1]];
-        }
-        CACHE('bb_cache')->set('censored', $sql, 7200);
-    }
-
-    foreach ($sql as $row) {
-        //$orig_word[] = '#(?<!\S)(' . str_replace('\*', '\S*?', preg_quote($row['word'], '#')) . ')(?!\S)#iu';
-        $orig_word[] = '#(?<![\p{Nd}\p{L}_])(' . str_replace('\*', '[\p{Nd}\p{L}_]*?', preg_quote($row['word'], '#')) . ')(?![\p{Nd}\p{L}_])#iu';
-        $replacement_word[] = $row['replacement'];
-    }
-
-    return true;
-}
-
 function bb_die($msg_text, $status_code = null)
 {
     global $ajax, $bb_cfg, $lang, $template, $theme, $userdata, $user;

library/includes/init_bb.php

@@ -394,6 +394,11 @@ $user = new TorrentPier\Legacy\Common\User();
 
 $userdata =& $user->data;
 
+/**
+ * Word censor
+ */
+$wordCensor = new \TorrentPier\Censor();
+
 /**
  * Cron
  */

library/includes/ucp/topic_watch.php

@@ -81,7 +81,7 @@ if ($watch_count > 0) {
                 'ROW_CLASS' => (!($i % 2)) ? 'row1' : 'row2',
                 'POST_ID' => $watch[$i]['topic_first_post_id'],
                 'TOPIC_ID' => $watch[$i]['topic_id'],
-                'TOPIC_TITLE' => str_short($watch[$i]['topic_title'], 70),
+                'TOPIC_TITLE' => str_short($wordCensor->censorString($watch[$i]['topic_title']), 70),
                 'FULL_TOPIC_TITLE' => $watch[$i]['topic_title'],
                 'U_TOPIC' => TOPIC_URL . $watch[$i]['topic_id'],
                 'FORUM_TITLE' => $watch[$i]['forum_name'],

posting.php

@@ -18,20 +18,15 @@ $page_cfg['load_tpl_vars'] = [
 ];
 
 $submit = (bool)@$_REQUEST['post'];
-$preview = (bool)@$_REQUEST['preview'];
+$refresh = $preview = (bool)@$_REQUEST['preview'];
 $delete = (bool)@$_REQUEST['delete'];
+$mode = (string)@$_REQUEST['mode'];
+$confirm = isset($_POST['confirm']);
 
 $forum_id = (int)@$_REQUEST[POST_FORUM_URL];
 $topic_id = (int)@$_REQUEST[POST_TOPIC_URL];
 $post_id = (int)@$_REQUEST[POST_POST_URL];
 
-$mode = (string)@$_REQUEST['mode'];
-
-$confirm = isset($_POST['confirm']);
-
-$refresh = $preview;
-$orig_word = $replacement_word = [];
-
 // Set topic type
 $topic_type = (@$_POST['topictype']) ? (int)$_POST['topictype'] : POST_NORMAL;
 $topic_type = in_array($topic_type, [POST_NORMAL, POST_STICKY, POST_ANNOUNCE]) ? $topic_type : POST_NORMAL;
@@ -461,13 +456,6 @@ if ($refresh || $error_msg || ($submit && $topic_has_new_posts)) {
         $message = $post_info['post_text'];
 
         if ($mode == 'quote') {
-            if (!defined('WORD_LIST_OBTAINED')) {
-                $orig_word = [];
-                $replace_word = [];
-                obtain_word_list($orig_word, $replace_word);
-                define('WORD_LIST_OBTAINED', true);
-            }
-
             if ($post_info['post_attachment'] && !IS_AM) {
                 $message = $post_info['topic_title'];
             }
@@ -475,16 +463,14 @@ if ($refresh || $error_msg || ($submit && $topic_has_new_posts)) {
             $message = '[quote="' . $quote_username . '"][qpost=' . $post_info['post_id'] . ']' . $message . '[/quote]';
 
             // hide user passkey
-            $message = preg_replace('#(?<=\?uk=)[a-zA-Z0-9]{10}(?=&)#', 'passkey', $message);
+            $message = preg_replace('#(?<=\?uk=)[a-zA-Z0-9](?=&)#', 'passkey', $message);
             // hide sid
-            $message = preg_replace('#(?<=[\?&;]sid=)[a-zA-Z0-9]{12}#', 'sid', $message);
+            $message = preg_replace('#(?<=[\?&;]sid=)[a-zA-Z0-9]#', 'sid', $message);
 
-            if (!empty($orig_word)) {
+            $subject = $wordCensor->censorString($subject);
-                $subject = (!empty($subject)) ? preg_replace($orig_word, $replace_word, $subject) : '';
+            $message = $wordCensor->censorString($message);
-                $message = (!empty($message)) ? preg_replace($orig_word, $replace_word, $message) : '';
-            }
 
-            if (!preg_match('/^Re:/', $subject) && strlen($subject) > 0) {
+            if (!preg_match('/^Re:/', $subject) && !empty($subject)) {
                 $subject = 'Re: ' . $subject;
             }
 

privmsg.php

@@ -375,18 +375,9 @@ if ($mode == 'read') {
     // Processing of post
     //
     $post_subject = htmlCHR($privmsg['privmsgs_subject']);
-
     $private_message = $privmsg['privmsgs_text'];
-
+    $post_subject = $wordCensor->censorString($post_subject);
-    $orig_word = [];
+    $private_message = $wordCensor->censorString($private_message);
-    $replacement_word = [];
-    obtain_word_list($orig_word, $replacement_word);
-
-    if (count($orig_word)) {
-        $post_subject = preg_replace($orig_word, $replacement_word, $post_subject);
-        $private_message = preg_replace($orig_word, $replacement_word, $private_message);
-    }
-
     $private_message = bbcode2html($private_message);
 
     //
@@ -1052,18 +1043,9 @@ if ($mode == 'read') {
     }
 
     if ($preview && !$error) {
-        $orig_word = [];
-        $replacement_word = [];
-        obtain_word_list($orig_word, $replacement_word);
-
         $preview_message = bbcode2html($privmsg_message);
-
+        $preview_subject = $wordCensor->censorString($privmsg_subject);
-        if (count($orig_word)) {
+        $preview_message = $wordCensor->censorString($preview_message);
-            $preview_subject = preg_replace($orig_word, $replacement_word, $privmsg_subject);
-            $preview_message = preg_replace($orig_word, $replacement_word, $preview_message);
-        } else {
-            $preview_subject = $privmsg_subject;
-        }
 
         $s_hidden_fields = '<input type="hidden" name="folder" value="' . $folder . '" />';
         $s_hidden_fields .= '<input type="hidden" name="mode" value="' . $mode . '" />';
@@ -1188,9 +1170,6 @@ if ($mode == 'read') {
     //
     $template->set_filenames(['body' => 'privmsgs.tpl']);
 
-    $orig_word = $replacement_word = [];
-    obtain_word_list($orig_word, $replacement_word);
-
     //
     // New message
     //
@@ -1402,12 +1381,7 @@ if ($mode == 'read') {
 
             $msg_userid = $row['user_id'];
             $msg_user = profile_url($row);
-
+            $msg_subject = $wordCensor->censorString($row['privmsgs_subject']);
-            $msg_subject = $row['privmsgs_subject'];
-
-            if (count($orig_word)) {
-                $msg_subject = preg_replace($orig_word, $replacement_word, $msg_subject);
-            }
 
             $u_subject = PM_URL . "?folder=$folder&amp;mode=read&amp;" . POST_POST_URL . "=$privmsg_id";
 

search.php

@@ -59,12 +59,6 @@ if (isset($_POST['del_my_post'])) {
     redirect("search.php?u={$user->id}");
 }
 
-//
-// Define censored word matches
-//
-$orig_word = $replacement_word = [];
-obtain_word_list($orig_word, $replacement_word);
-
 $tracking_topics = get_tracks('topic');
 $tracking_forums = get_tracks('forum');
 
@@ -572,35 +566,27 @@ if ($post_mode) {
         $topic_id = (int)$topic_id;
         $forum_id = (int)$first_post['forum_id'];
         $is_unread_t = is_unread($first_post['topic_last_post_time'], $topic_id, $forum_id);
-        $topic_title = $first_post['topic_title'];
-
-        if (count($orig_word)) {
-            $topic_title = preg_replace($orig_word, $replacement_word, $topic_title);
-        }
 
         $template->assign_block_vars('t', array(
             'FORUM_ID' => $forum_id,
             'FORUM_NAME' => $forum_name_html[$forum_id],
             'TOPIC_ID' => $topic_id,
-            'TOPIC_TITLE' => $topic_title,
+            'TOPIC_TITLE' => $wordCensor->censorString($first_post['topic_title']),
             'TOPIC_ICON' => get_topic_icon($first_post, $is_unread_t),
         ));
 
         $quote_btn = $edit_btn = $ip_btn = '';
-        $delpost_btn = (IS_AM);
+        $delpost_btn = IS_AM;
 
         // Topic posts block
         foreach ($topic_posts as $row_num => $post) {
             if ($post['poster_id'] != BOT_UID) {
                 $quote_btn = true;
-                $edit_btn = $ip_btn = (IS_AM);
+                $edit_btn = $ip_btn = IS_AM;
             }
 
             $message = get_parsed_post($post);
-
+            $message = $wordCensor->censorString($message);
-            if (count($orig_word)) {
-                $message = preg_replace($orig_word, $replacement_word, $message);
-            }
 
             $template->assign_block_vars('t.p', array(
                 'ROW_NUM' => $row_num,
@@ -801,7 +787,7 @@ else {
             'FORUM_NAME' => $forum_name_html[$forum_id],
             'TOPIC_ID' => $topic_id,
             'HREF_TOPIC_ID' => $moved ? $topic['topic_moved_id'] : $topic['topic_id'],
-            'TOPIC_TITLE' => $topic['topic_title'],
+            'TOPIC_TITLE' => $wordCensor->censorString($topic['topic_title']),
             'IS_UNREAD' => $is_unread,
             'TOPIC_ICON' => get_topic_icon($topic, $is_unread),
             'PAGINATION' => $moved ? '' : build_topic_pagination(TOPIC_URL . $topic_id, $topic['topic_replies'], $bb_cfg['posts_per_page']),

src/Censor.php

@@ -0,0 +1,64 @@
+<?php
+/**
+ * TorrentPier – Bull-powered BitTorrent tracker engine
+ *
+ * @copyright Copyright (c) 2005-2024 TorrentPier (https://torrentpier.com)
+ * @link      https://github.com/torrentpier/torrentpier for the canonical source repository
+ * @license   https://github.com/torrentpier/torrentpier/blob/master/LICENSE MIT License
+ */
+
+namespace TorrentPier;
+
+/**
+ * Class Censor
+ * @package TorrentPier
+ */
+class Censor
+{
+    /**
+     * Word replacements
+     *
+     * @var array
+     */
+    public array $replacements = [];
+
+    /**
+     * All censored words (RegEx)
+     *
+     * @var array
+     */
+    public array $words = [];
+
+    /**
+     * Initialize word censor
+     */
+    public function __construct()
+    {
+        global $bb_cfg;
+
+        if (!$bb_cfg['use_word_censor']) {
+            return;
+        }
+
+        if (!$words = CACHE('bb_cache')->get('censored')) {
+            $words = DB()->fetch_rowset("SELECT word, replacement FROM " . BB_WORDS);
+            CACHE('bb_cache')->set('censored', $words, 7200);
+        }
+
+        foreach ($words as $word) {
+            $this->words[] = '#(?<![\p{Nd}\p{L}_])(' . str_replace('\*', '[\p{Nd}\p{L}_]*?', preg_quote($word['word'], '#')) . ')(?![\p{Nd}\p{L}_])#iu';
+            $this->replacements[] = $word['replacement'];
+        }
+    }
+
+    /**
+     * Word censor
+     *
+     * @param string $word
+     * @return string
+     */
+    public function censorString(string $word): string
+    {
+        return preg_replace($this->words, $this->replacements, $word);
+    }
+}

src/Legacy/Atom.php

@@ -169,7 +169,7 @@ class Atom
      */
     private static function create_atom($file_path, $mode, $id, $title, $topics)
     {
-        global $bb_cfg, $lang;
+        global $bb_cfg, $lang, $wordCensor;
         $date = null;
         $time = null;
         $dir = \dirname($file_path);
@@ -203,13 +203,7 @@ class Atom
             if (isset($topic['tor_status'])) {
                 $tor_status = " ({$lang['TOR_STATUS_NAME'][$topic['tor_status']]})";
             }
-            $topic_title = $topic['topic_title'];
+            $topic_title = $wordCensor->censorString($topic['topic_title']);
-            $orig_word = [];
-            $replacement_word = [];
-            obtain_word_list($orig_word, $replacement_word);
-            if (\count($orig_word)) {
-                $topic_title = preg_replace($orig_word, $replacement_word, $topic_title);
-            }
             $author_name = $topic['first_username'] ?: $lang['GUEST'];
             $last_time = $topic['topic_last_post_time'];
             if ($topic['topic_last_post_edit_time']) {

src/Legacy/Post.php

@@ -335,7 +335,7 @@ class Post
      */
     public static function user_notification($mode, &$post_data, &$topic_title, &$forum_id, &$topic_id, &$notify_user)
     {
-        global $bb_cfg, $lang, $userdata;
+        global $bb_cfg, $lang, $userdata, $wordCensor;
 
         if (!$bb_cfg['topic_notify_enabled']) {
             return;
@@ -358,12 +358,7 @@ class Post
 			");
 
                 if ($watch_list) {
-                    $orig_word = $replacement_word = [];
+                    $topic_title = $wordCensor->censorString($topic_title);
-                    obtain_word_list($orig_word, $replacement_word);
-
-                    if (\count($orig_word)) {
-                        $topic_title = preg_replace($orig_word, $replacement_word, $topic_title);
-                    }
 
                     $u_topic = make_url(TOPIC_URL . $topic_id . '&view=newest#newest');
                     $unwatch_topic = make_url(TOPIC_URL . "$topic_id&unwatch=topic");

viewforum.php

@@ -354,10 +354,6 @@ if ($topics_csv = implode(',', $topic_ids)) {
 	");
 }
 
-// Define censored word matches
-$orig_word = $replacement_word = [];
-obtain_word_list($orig_word, $replacement_word);
-
 if ($forum_data['allow_reg_tracker']) {
     $post_new_topic_url = POSTING_URL . "?mode=new_rel&f=$forum_id";
     $post_img = $images['release_new'];
@@ -444,7 +440,7 @@ foreach ($topic_rowset as $topic) {
         'FORUM_ID' => $forum_id,
         'TOPIC_ID' => $topic_id,
         'HREF_TOPIC_ID' => $moved ? $topic['topic_moved_id'] : $topic['topic_id'],
-        'TOPIC_TITLE' => $topic['topic_title'],
+        'TOPIC_TITLE' => $wordCensor->censorString($topic['topic_title']),
         'TOPICS_SEPARATOR' => $separator,
         'IS_UNREAD' => $is_unread,
         'TOPIC_ICON' => get_topic_icon($topic, $is_unread),

viewtopic.php

@@ -365,14 +365,8 @@ if (!$ranks = $datastore->get('ranks')) {
     $ranks = $datastore->get('ranks');
 }
 
-// Define censored word matches
-$orig_word = $replacement_word = [];
-obtain_word_list($orig_word, $replacement_word);
-
 // Censor topic title
-if (count($orig_word)) {
+$topic_title = $wordCensor->censorString($topic_title);
-    $topic_title = preg_replace($orig_word, $replacement_word, $topic_title);
-}
 
 // Post, reply and other URL generation for templating vars
 $new_topic_url = POSTING_URL . "?mode=newtopic&f=" . $forum_id;
@@ -627,28 +621,26 @@ for ($i = 0; $i < $total_posts; $i++) {
     }
 
     // Replace naughty words
-    if (count($orig_word)) {
+    if ($user_sig) {
-        if ($user_sig) {
+        $user_sig = str_replace(
-            $user_sig = str_replace(
-                '\"', '"',
-                substr(
-                    preg_replace_callback('#(\>(((?>([^><]+|(?R)))*)\<))#s', function ($matches) use ($orig_word, $replacement_word) {
-                        return preg_replace($orig_word, $replacement_word, reset($matches));
-                    }, '>' . $user_sig . '<'), 1, -1
-                )
-            );
-        }
-
-        $message = str_replace(
             '\"', '"',
             substr(
-                preg_replace_callback('#(\>(((?>([^><]+|(?R)))*)\<))#s', function ($matches) use ($orig_word, $replacement_word) {
+                preg_replace_callback('#(\>(((?>([^><]+|(?R)))*)\<))#s', function ($matches) use ($wordCensor) {
-                    return preg_replace($orig_word, $replacement_word, reset($matches));
+                    return $wordCensor->censorString(reset($matches));
-                }, '>' . $message . '<'), 1, -1
+                }, '>' . $user_sig . '<'), 1, -1
             )
         );
     }
 
+    $message = str_replace(
+        '\"', '"',
+        substr(
+            preg_replace_callback('#(\>(((?>([^><]+|(?R)))*)\<))#s', function ($matches) use ($wordCensor) {
+                return $wordCensor->censorString(reset($matches));
+            }, '>' . $message . '<'), 1, -1
+        )
+    );
+
     // Replace newlines (we use this rather than nl2br because till recently it wasn't XHTML compliant)
     if ($user_sig) {
         $user_sig = $bb_cfg['user_signature_start'] . $user_sig . $bb_cfg['user_signature_end'];