Fix markup again. And add the option to repair fucked up nesting (and more) with HTML Tidy ($config['markup_repair_tidy'])

10 years ago · 7b817eea11
2 changed files with 18 additions and 5 deletions
--- a/inc/config.php
+++ b/inc/config.php
@ -459,10 +459,10 @@
 */

 	// "Wiki" markup syntax ($config['wiki_markup'] in pervious versions):
-	$config['markup'][] = array("/'''([^<]+?)'''/", "<strong>\$1</strong>");
-	$config['markup'][] = array("/''([^<]+?)''/", "<em>\$1</em>");
-	$config['markup'][] = array("/\*\*([^<]+?)\*\*/", "<span class=\"spoiler\">\$1</span>");
-	// $config['markup'][] = array("/^[ |\t]*==([^<]+?)==[ |\t]*$/m", "<span class=\"heading\">\$1</span>");
+	$config['markup'][] = array("/'''(.+?)'''/", "<strong>\$1</strong>");
+	$config['markup'][] = array("/''(.+?)''/", "<em>\$1</em>");
+	$config['markup'][] = array("/\*\*(.+?)\*\*/", "<span class=\"spoiler\">\$1</span>");
+	// $config['markup'][] = array("/^[ |\t]*==(.+?)==[ |\t]*$/m", "<span class=\"heading\">\$1</span>");

 	// Highlight PHP code wrapped in <code> tags (PHP 5.3+)
 	// $config['markup'][] = array(
@ -472,6 +472,10 @@
 	// 	}
 	// );

+	// Repair markup with HTML Tidy. This may be slower, but most if the time it solves nesting mistakes.
+	// Tinyboad, at the time of writing this, can not prevent out-of-order markup tags (eg. "**''test**'').
+	$config['markup_repair_tidy'] = false;
+
 	// Always regenerate markup. This isn't recommended and should only be used for debugging; by default,
 	// Tinyboard only parses post markup when it needs to, and keeps post-markup HTML in the database. This
 	// will significantly impact performance when enabled.
--- a/inc/functions.php
+++ b/inc/functions.php
@ -1728,7 +1728,16 @@ function markup(&$body, $track_cites = false) {
 		$body = preg_replace('/\s+$/', '', $body);

 	$body = preg_replace("/\n/", '<br/>', $body);
-
+	
+	if ($config['markup_repair_tidy']) {
+		$tidy = new tidy();
+		$body = $tidy->repairString($body, array(
+			'doctype' => 'omit'
+		));
+		$body = str_replace("\n", '', $body);
+		$body = preg_replace('@^.+<body>|</body>[email protected]', '', $body);
+	}
+	
 	return $tracked_cites;
 }