Fix #741 do not allow empty HTML content in document and comment

common/functions.php

@@ -608,3 +608,18 @@ function utf8_trim($str)
 {
 	return preg_replace('/^[\s\pZ\pC]+|[\s\pZ\pC]+$/u', '', $str);
 }
+
+/**
+ * Check if HTML content is empty.
+ * This function checks whether any printable characters remain
+ * after removing all tags except images, videos, iframes, etc.
+ * 
+ * @param string $str The input string
+ * @return bool
+ */
+function is_empty_html_content($str)
+{
+	$str = strip_tags($str, '<img><audio><video><iframe><object><embed>');
+	$str = utf8_trim(utf8_clean(html_entity_decode($str, ENT_QUOTES, 'UTF-8')));
+	return $str === '';
+}

common/lang/en.php

@@ -241,6 +241,7 @@ $lang->msg_module_is_not_exists = 'Cannot find the page you requested. Ask your
 $lang->msg_module_is_not_standalone = 'Requested page cannot be executed independently.';
 $lang->msg_empty_search_target = 'Cannot find the Search target.';
 $lang->msg_empty_search_keyword = 'Cannot find the Keyword.';
+$lang->msg_empty_content = 'The content is empty.';
 $lang->msg_server_error = 'Server Error';
 $lang->msg_server_error_see_log = 'Your server is configured to hide error messages. Please see your server\'s error log for details.';
 $lang->comment_to_be_approved = 'Your comment must be approved by admin before being published.';

common/lang/ko.php

@@ -241,6 +241,7 @@ $lang->msg_module_is_not_exists = '요청한 페이지를 찾을 수 없습니
 $lang->msg_module_is_not_standalone = '요청한 페이지는 독립적으로 동작할 수 없습니다.';
 $lang->msg_empty_search_target = '검색대상이 없습니다.';
 $lang->msg_empty_search_keyword = '검색어가 없습니다.';
+$lang->msg_empty_content = '내용이 없습니다.';
 $lang->msg_server_error = '서버 오류';
 $lang->msg_server_error_see_log = '오류 메시지를 표시하지 않도록 설정되어 있습니다. 서버의 에러 로그에서 자세한 내용을 확인해 주십시오.';
 $lang->comment_to_be_approved = '관리자의 확인이 필요한 댓글입니다.';

modules/comment/comment.controller.php

@@ -425,6 +425,11 @@ class commentController extends comment
 
 		// remove Rhymix's own tags from the contents
 		$obj->content = preg_replace('!<\!--(Before|After)(Document|Comment)\(([0-9]+),([0-9]+)\)-->!is', '', $obj->content);
+		// Return error if content is empty.
+		if (!$manual_inserted && is_empty_html_content($obj->content))
+		{
+			return new Object(-1, 'msg_empty_content');
+		}
 		
 		// if use editor of nohtml, Remove HTML tags from the contents.
 		if(!$manual_inserted)
@@ -851,6 +856,11 @@ class commentController extends comment
 
 		// remove Rhymix's wn tags from contents
 		$obj->content = preg_replace('!<\!--(Before|After)(Document|Comment)\(([0-9]+),([0-9]+)\)-->!is', '', $obj->content);
+		// Return error if content is empty.
+		if (!$manual_inserted && is_empty_html_content($obj->content))
+		{
+			return new Object(-1, 'msg_empty_content');
+		}
 
 		// if use editor of nohtml, Remove HTML tags from the contents.
 		if(!$manual_updated)

modules/document/document.controller.php

@@ -435,6 +435,11 @@ class documentController extends document
 		if($obj->title == '') $obj->title = 'Untitled';
 		// Remove XE's own tags from the contents.
 		$obj->content = preg_replace('!<\!--(Before|After)(Document|Comment)\(([0-9]+),([0-9]+)\)-->!is', '', $obj->content);
+		// Return error if content is empty.
+		if (!$manual_inserted && is_empty_html_content($obj->content))
+		{
+			return new Object(-1, 'msg_empty_content');
+		}
 		// if use editor of nohtml, Remove HTML tags from the contents.
 		if(!$manual_inserted)
 		{
@@ -670,6 +675,11 @@ class documentController extends document
 		if($obj->title == '') $obj->title = 'Untitled';
 		// Remove XE's own tags from the contents.
 		$obj->content = preg_replace('!<\!--(Before|After)(Document|Comment)\(([0-9]+),([0-9]+)\)-->!is', '', $obj->content);
+		// Return error if content is empty.
+		if (!$manual_inserted && is_empty_html_content($obj->content))
+		{
+			return new Object(-1, 'msg_empty_content');
+		}
 		// if use editor of nohtml, Remove HTML tags from the contents.
 		if(!$manual_updated)
 		{

tests/unit/functions/FunctionsTest.php

@@ -159,4 +159,16 @@ class FunctionsTest extends \Codeception\TestCase\Test
 		$this->assertEquals("Trimmed", utf8_trim("\x20\xe2\x80\x80Trimmed\xe2\x80\x84\xe2\x80\x86\xe2\x80\x8b"));
 		$this->assertEquals("Trimmed", utf8_trim("\x20\xe2\x80\x80Trimmed\x0a\x0c\x07\x09"));
 	}
+	
+	public function testIsEmptyHTMLContent()
+	{
+		$this->assertTrue(is_empty_html_content('<p>&nbsp;<br><br></p>'));
+		$this->assertTrue(is_empty_html_content('<p>&nbsp;</p>' . "\n\n" . '<p><span> </span></p>'));
+		$this->assertTrue(is_empty_html_content('<p>&#8194; &#8203; &#8205;</p>'));
+		$this->assertFalse(is_empty_html_content('<p>&nbsp;</p>' . "\n\n" . '<p>Hello world</p>'));
+		$this->assertFalse(is_empty_html_content('<p><img src="foobar.jpg"></p>'));
+		$this->assertFalse(is_empty_html_content('<p><iframe src="http://www.youtube.com/" /></p>'));
+		$this->assertFalse(is_empty_html_content('<p><video src="rickroll.webm" /></p>'));
+		$this->assertFalse(is_empty_html_content('<p><object></object></p>'));
+	}
 }