Normalize combining diaeresis to umlaut in it::any2utf8()

1 files changed, 3 insertions, 1 deletions

diff --git a/it.class b/it.class
index 4c57b2c..83fcf0e 100644
--- a/it.class
+++ b/it.class
@@ -519,7 +519,9 @@ static function any2utf8($value, $errprefix = "")
 			list($value, $error) = array(it::any2utf8(preg_replace_callback('/\xc3[\x82\x83]\xc2[\x82\x83\xbc\xa9\xa4\xb6\xa8\xa2\xa0\xb4\xaa\xa7\x84\xab\xae\x9c\xaf\x96\xb2\xbb\xb9\x9f]/', function($m) {return utf8_decode($m[0]);}, $value)), $errprefix ? "$errprefix: double utf8-encoding. input=$value" : "");
 		if (preg_match('/\xef\xb7[\x90-\xaf]|\xef\xbf[\xbe\xbf]/', $value))
 			list($value, $error) = array(preg_replace('/\xef\xb7[\x90-\xaf]|\xef\xbf[\xbe\xbf]/', " ", $value), "forbidden utf-8 character. input=$value");
-		$value = preg_replace('/\xc2\xad/', '', $value);
+		$value = preg_replace('/\xc2\xad/', '', $value);	# Kill invisible soft hyphens
+		if (preg_match('/\xcc\x88/', $value))	# Normalize combining diaeresis to umlaut
+			$value = strtr($value, [ "a\xcc\x88" => 'ä', "A\xcc\x88" => 'Ä', "e\xcc\x88" => 'ë', "E\xcc\x88" => 'Ë', "i\xcc\x88" => 'ï', "I\xcc\x88" => 'Ï', "o\xcc\x88" => 'ö', "O\xcc\x88" => 'Ö', "u\xcc\x88" => 'ü', "U\xcc\x88" => 'Ü' ]);
 		if ($error && $errprefix)
 			it::error(array('title' => "$errprefix: " . trim($error)));
 	}