encoding checker

1 files changed, 27 insertions, 9 deletions

diff --git a/it_html.class b/it_html.class
index 1aeed94..41d799f 100644
--- a/it_html.class
+++ b/it_html.class
@@ -258,6 +258,24 @@ function _parse_args($args)
 }
 
 
+# internal
+function fix_encoding($string)
+{
+	if (preg_match('/[\x20-\x7f][\x80-\xff][\x20-\x7f]/', $string))
+	{
+		it::error(array('title' => utf8_encode("incorrectly utf8-encoded: " . trim($string)), 'skipfiles' => "it_html"));
+		$string = utf8_encode($string);
+	}
+	else if ($string && preg_match('/[\x80-\xff]/', $string) && htmlspecialchars(utf8_decode($string), ENT_COMPAT, "utf-8") !== "")
+	{
+		it::error(array('title' => utf8_encode("doubly utf8-encoded: " . trim($string)), 'skipfiles' => "it_html"));
+		$string = utf8_decode($string);
+	}
+
+	return $string;
+}
+
+
 /**
  * function div($args...)
  * Return a <div>...</div> element
@@ -312,16 +330,12 @@ function _tag($name, $args)
 	else
 		$result .= " />$newline";
 
-	if ($GLOBALS['debug_utf8check'] && $GLOBALS['it_html']->p['charset'] == "utf-8" && preg_match('/[\x20-\x7f][\x80-\xff][\x20-\x7f]/', $result))
-	{
-		it::error(array('title' => utf8_encode("incorrectly utf8-encoded: " . trim($result)), 'skipfiles' => "it_html"));
-		$result = utf8_encode($result);
-	}
+	if ($GLOBALS['debug_utf8check'] && $GLOBALS['it_html']->p['charset'] == "utf-8")
+		$result = self::fix_encoding($result);
 
 	return $result;
 }
 
-
 /**
  * Return a <tag> containing optional data.
  * @param $name tag name ('style', etc.)
@@ -482,9 +496,13 @@ function latinize($string)
  */
 function Q($string)
 {
-	if (preg_match('/[<>&"\x00-\x08\x0a-\x0c\x0e-\x1f\x80-\xff]/', $origstring = $string)) # WARNING: copy/pasted to _tag()
-		if (($string = htmlspecialchars($GLOBALS['it_html']->p['charset'] == "iso-8859-1" ? it_html::latinize($string) : $string, ENT_COMPAT, $GLOBALS['it_html']->p['charset'])) === "" && $GLOBALS['debug_utf8check'])
-			it::error(array('title' => utf8_encode("incorrectly utf8-encoded: " . trim($origstring)), 'skipfiles' => "it_html"));
+	if (preg_match('/[<>&"\x00-\x08\x0a-\x0c\x0e-\x1f\x80-\xff]/', $string)) # WARNING: copy/pasted to _tag()
+	{
+		if ($GLOBALS['debug_utf8check'] && $GLOBALS['it_html']->p['charset'] == "utf-8")
+			$string = self::fix_encoding($string);
+
+		$string = htmlspecialchars($GLOBALS['it_html']->p['charset'] == "iso-8859-1" ? it_html::latinize($string) : $string, ENT_COMPAT, $GLOBALS['it_html']->p['charset']);
+	}
 
 	return $GLOBALS['debug_q'] && $string ? "<span style='background:#8FF'>$string</span>" : $string;
 }