support arrays in any2utf8

2 files changed, 19 insertions, 9 deletions

diff --git a/it.class b/it.class
index 839e457..8ce1d65 100644
--- a/it.class
+++ b/it.class
@@ -410,19 +410,27 @@ static function grep($pattern, $array, $p = array())
 
 /**
  * Convert string to utf8 if it was not already utf-8 before. Also handles double encoding
- * @param $value String to convert
+ * @param $value String or array to convert
  * @param $errmsg Error message to output if anything needed to be done
  * @return Same string in utf-8 encoding
  */
 function any2utf8($value, $errmsg = "")
 {
-	if (grapheme_strlen($value) === null)
-		list($value, $error) = array(utf8_encode($value), utf8_encode("$errmsg: incorrect utf8-encoding. input=" . trim($value)));
-	if (preg_match('/\xc3\x83\xc2([\xbc\xa9\xa4\xb6\xa8\xa2\xa0\xb4\xaa\xa7\x84\xab\xae\x9c\xaf\x96\xb2\xbb\xb9\x9f])/', $value))
-		list($value, $error) = array(preg_replace('/\xc3\x83\xc2([\xbc\xa9\xa4\xb6\xa8\xa2\xa0\xb4\xaa\xa7\x84\xab\xae\x9c\xaf\x96\xb2\xbb\xb9\x9f])/', "\xc3\$1", $value), utf8_encode("$errmsg: double utf8-encoding. input=" . trim($value)));
-
-	if ($error && $errmsg)
-		it::error(array('title' => $error, 'skipfiles' => "it_html"));
+	if (is_array($value))
+	{
+		foreach ($value as $idx => $v)
+			if (is_string($v) || is_array($v))
+				$value[$idx] = self::any2utf8($v, $errmsg);
+	}
+	else
+	{
+		if (grapheme_strlen($value) === null)
+			list($value, $error) = array(utf8_encode($value), utf8_encode("$errmsg: incorrect utf8-encoding. input=" . trim($value)));
+		if (preg_match('/\xc3\x83\xc2([\xbc\xa9\xa4\xb6\xa8\xa2\xa0\xb4\xaa\xa7\x84\xab\xae\x9c\xaf\x96\xb2\xbb\xb9\x9f])/', $value))
+			list($value, $error) = array(preg_replace('/\xc3\x83\xc2([\xbc\xa9\xa4\xb6\xa8\xa2\xa0\xb4\xaa\xa7\x84\xab\xae\x9c\xaf\x96\xb2\xbb\xb9\x9f])/', "\xc3\$1", $value), utf8_encode("$errmsg: double utf8-encoding. input=" . trim($value)));
+		if ($error && $errmsg)
+			it::error(array('title' => $error, 'skipfiles' => "it_html"));
+	}
 
 	return $value;
 }
diff --git a/tests/it.t b/tests/it.t
index 68ea398..335c520 100755
--- a/tests/it.t
+++ b/tests/it.t
@@ -379,4 +379,6 @@ is(it::any2utf8(utf8_encode("Müller")), "Müller", "it::any2utf8 double encoded
 is(it::any2utf8(utf8_decode("Müller")), "Müller", "it::any2utf8 incorrectly encoded latin1");
 is(it::any2utf8("a💚b"), "a💚b", "it::any2utf8 correctly handles 4-byte utf-8 character GREEN HEART");
 
-?>
+is(it::any2utf8(array("foo", utf8_decode("bär"))), array("foo", "bär"), "any2utf8 on arrays");
+is(it::any2utf8(array("foo", array(utf8_decode("bär")))), array("foo", array("bär")), "any2utf8 on recursive arrays");
+is(it::any2utf8(array(1, true, false, null)), array(1, true, false, null), "any2utf8 should leave types alone");