Add regexp support to spamfilter

modules/spamfilter/lang/en.php

@@ -12,9 +12,12 @@ $lang->check_trackback = 'Check Trackbacks';
 $lang->word = 'Keyword';
 $lang->hit = 'Hit';
 $lang->latest_hit = 'Latest Hits';
+$lang->custom_message = 'Error Message';
+$lang->about_custom_message = 'You can customize the error message that will be displayed if a spam keyword is found.<br>%s can be used as a placeholder for the keyword. If not used, the keyword will be hidden.';
 $lang->about_interval = 'All articles attempted for posting within the assigned time will be blocked.';
 $lang->about_denied_ip = 'Please enter one IP address (e.g. 127.0.0.1) or range (e.g. 127.0.0.0/24) per line. Comments may start with //.';
-$lang->about_denied_word = 'Please enter one keyword per line. Keywords may contain 2 to 40 characters.';
+$lang->about_denied_word = 'Please enter one keyword per line. Keywords may contain 2 to 180 characters.<br>Formats such as /spam(key|word)?/ will be treated as a regular expression, and must use the proper syntax.<br>Spam keywords are not case sensitive.';
+$lang->msg_denied_word_is_regexp = 'REGEXP';
 $lang->msg_alert_limited_by_config = 'Please do not post repeatedly within %d seconds. If you keep trying, your IP address will be blocked.';
 $lang->msg_alert_limited_message_by_config = 'Please do not send messages repeatedly within %d seconds. If you keep trying, your IP address will be blocked.';
 $lang->msg_alert_denied_word = 'The word "%s" is not allowed on this site.';
@@ -37,5 +40,5 @@ $lang->unit_write_count = 'times';
 $lang->add = 'Add';
 $lang->msg_duplicate = 'Duplicate';
 $lang->msg_invalid_ip = 'Invalid IP address format.';
-$lang->msg_invalid_word = 'Spam keywords must be between 2 and 40 characters.';
+$lang->msg_invalid_word = 'Spam keywords must be between 2 and 180 characters.';
 $lang->msg_faillist = '<br />Error (already blocked)<br /> %s ';

modules/spamfilter/lang/ja.php

@@ -29,5 +29,5 @@ $lang->yes = 'はい';
 $lang->no = 'いいえ';
 $lang->msg_duplicate = '既に存在します。';
 $lang->msg_invalid_ip = 'IPアドレスの形式が正しくありません。';
-$lang->msg_invalid_word = 'スパムキーワードは2〜40文字の範囲で指定します。';
+$lang->msg_invalid_word = 'スパムキーワードは2〜180文字の範囲で指定します。';
 $lang->msg_faillist = '<br />失敗（既に存在します。） <br /> %s ';

modules/spamfilter/lang/ko.php

@@ -12,11 +12,12 @@ $lang->check_trackback = '트랙백 검사';
 $lang->word = '키워드';
 $lang->hit = '히트';
 $lang->latest_hit = '최근 히트';
-$lang->custom_message = '에러메세지 설정';
+$lang->custom_message = '차단 메시지 설정';
-$lang->about_custom_message = '키워드를 출력하지 않을시 에러메세지를 설정합니다.';
+$lang->about_custom_message = '스팸 키워드 발견시 표시할 에러 메시지를 지정할 수 있습니다.<br>%s를 넣으면 그 자리에 해당 키워드를 표시하고, 그렇지 않으면 키워드를 숨깁니다.';
 $lang->about_interval = '지정된 시간 내에 글을 등록하지 못하게 합니다.';
 $lang->about_denied_ip = '한 줄에 하나씩 IP 주소 또는 대역을 입력하세요. &quot;//&quot; 또는 &quot;#&quot; 뒷부분은 설명으로 저장됩니다. 예: 127.0.0.1 //설명, 127.0.0.1 #설명<br>IP 대역 표기법은 <a href="https://github.com/rhymix/rhymix-docs/blob/master/ko/misc/ipfilter.md" target="_blank">매뉴얼</a>을 참고하십시오.';
-$lang->about_denied_word = '한 줄에 하나씩 스팸 키워드를 입력하세요. (2~40자)';
+$lang->about_denied_word = '한 줄에 하나씩 스팸 키워드를 입력하세요. (2~180자)<br>/스팸(키+|워드)?/ 와 같은 형태로 입력하면 정규식으로 간주하며, 올바른 정규식 문법을 사용해야 합니다.<br>대소문자는 구분하지 않습니다.';
+$lang->msg_denied_word_is_regexp = '정규식';
 $lang->msg_alert_limited_by_config = '%d초 이내에 연속 글 작성은 금지됩니다. 계속 시도하면 IP가 차단될 수 있습니다.';
 $lang->msg_alert_limited_message_by_config = '%d초 이내에 연속 쪽지 발송은 금지됩니다. 계속 시도하면 IP가 차단될 수 있습니다.';
 $lang->msg_alert_denied_word = '"%s"은(는) 사용이 금지된 단어입니다.';
@@ -39,5 +40,5 @@ $lang->unit_write_count = '회';
 $lang->add = '추가';
 $lang->msg_duplicate = '이미 존재합니다.';
 $lang->msg_invalid_ip = 'IP 주소 형식이 올바르지 않습니다.';
-$lang->msg_invalid_word = '스팸 키워드는 2~40자 사이여야 합니다.';
+$lang->msg_invalid_word = '스팸 키워드는 2~180자 사이여야 합니다.';
 $lang->msg_faillist = '<br />실패 (이미 차단되어 있습니다)<br /> %s ';

modules/spamfilter/schemas/spamfilter_denied_word.xml

@@ -1,5 +1,5 @@
 <table name="spamfilter_denied_word">
-    <column name="word" type="varchar" size="250" utf8mb4="false" notnull="notnull" primary_key="primary_key" />
+    <column name="word" type="varchar" size="180" utf8mb4="false" notnull="notnull" primary_key="primary_key" />
     <column name="hit" type="number" notnull="notnull" default="0" index="idx_hit" />
     <column name="latest_hit" type="date" index="idx_latest_hit" />
     <column name="regdate" type="date" index="idx_regdate" />

modules/spamfilter/spamfilter.admin.controller.php

@@ -149,7 +149,7 @@ class spamfilterAdminController extends spamfilter
 				continue;
 			}
 			
-			if (mb_strlen($word, 'UTF-8') < 2 || mb_strlen($word, 'UTF-8') > 40)
+			if (mb_strlen($word, 'UTF-8') < 2 || mb_strlen($word, 'UTF-8') > 180)
 			{
 				return $this->setError('msg_invalid_word');
 			}

modules/spamfilter/spamfilter.model.php

@@ -66,10 +66,8 @@ class spamfilterModel extends spamfilter
 	{
 		$args = new stdClass();
 		$args->sort_index = "hit";
-		$output = executeQuery('spamfilter.getDeniedWordList', $args);
+		$output = executeQueryArray('spamfilter.getDeniedWordList', $args);
-		if(!$output->data) return;
+		return $output->data ?: array();
-		if(!is_array($output->data)) return array($output->data);
-		return $output->data;
 	}
 
 	/**
@@ -84,7 +82,16 @@ class spamfilterModel extends spamfilter
 		foreach ($word_list as $word_item)
 		{
 			$word = $word_item->word;
-			if (strpos($text, strtolower($word)) !== false)
+			$hit = false;
+			if (preg_match('#^/.+/$#', $word))
+			{
+				$hit = preg_match($word . 'iu', $text, $matches) ? $matches[0] : false;
+			}
+			if ($hit === false)
+			{
+				$hit = (strpos($text, strtolower($word)) !== false) ? $word : false;
+			}
+			if ($hit !== false)
 			{
 				$args = new stdClass();
 				$args->word = $word;
@@ -106,7 +113,12 @@ class spamfilterModel extends spamfilter
 				}
 				else
 				{
-					$custom_message = sprintf(lang('msg_alert_denied_word'), $word);
+					$custom_message = lang('msg_alert_denied_word');
+				}
+
+				if (strpos($custom_message, '%s') !== false)
+				{
+					$custom_message = sprintf($custom_message, escape($hit, false));
 				}
 				
 				return new BaseObject(-1, $custom_message);

modules/spamfilter/tpl/css/spamfilter_admin.css

@@ -0,0 +1,5 @@
+span.is_regexp {
+	display: inline-block;
+	margin-left: 4px;
+	color: red;
+}

modules/spamfilter/tpl/denied_word_list.html

@@ -25,7 +25,7 @@
 			</thead>
 			<tbody>
 				<tr loop="$word_list => $word_info">
-					<td>{$word_info->word}</td>
+					<td>{$word_info->word} <!--@if(preg_match('#^/.+/$#', $word_info->word))--><span class="is_regexp">[{$lang->msg_denied_word_is_regexp}]</span><!--@end--></td>
 					<td><!--@if($word_info->latest_hit)-->{zdate($word_info->latest_hit,'Y-m-d H:i')}<!--@else-->-<!--@end--></td>
 					<td>{$word_info->hit}</td>
 					<td>{zdate($word_info->regdate,'Y-m-d')}</td>

modules/spamfilter/tpl/header.html

@@ -1,3 +1,4 @@
+<load target="css/spamfilter_admin.css" />
 <load target="js/spamfilter_admin.js" />
 <div class="x_page-header">
 	<h1>{$lang->spamfilter} <a class="x_icon-question-sign" href="./common/manual/admin/index.html#UMAN_content_spamfilter" target="_blank">{$lang->help}</a></h1>