Make javascript delivery unicode-safe

2 files changed, 21 insertions, 12 deletions

diff --git a/itjs.class b/itjs.class
index a7a6a15..e41ffe5 100644
--- a/itjs.class
+++ b/itjs.class
@@ -23,12 +23,18 @@
 
 class itjs
 {
+static $charset;
 
 /**
  * Send HTTP headers (content-type) to transmit javascript code
  */
-function send_headers($charset = 'iso-8859-1')
+function send_headers($charset = null)
 {
+	if (!$charset)
+		$charset = ini_get('default_charset') ?: 'iso-8859-1';
+
+	self::$charset = $charset;
+
 	if (!preg_match('/Opera/', $_SERVER['HTTP_USER_AGENT']) && !$_REQUEST['itjs_iframe'])	# text/plain breaks Opera 8.51/Linux and IFrame fallback
 		header("Content-Type: text/plain; charset=$charset");	# Berni reported some Firewalls to require this
 
@@ -67,16 +73,17 @@ function serialize($values, $envelope = false)
  */
 function encode($values)
 {
-	$texts = ($values === array_values($values)) ? "[]0 " : "{}1\n";	# Numerical or associative array
 	static $jskeyword = array("abstract" => 1, "boolean" => 1, "break" => 1, "byte" => 1, "case" => 1, "catch" => 1, "char" => 1, "class" => 1, "const" => 1, "continue" => 1, "debugger" => 1, "default" => 1, "delete" => 1, "do" => 1, "double" => 1, "each" => 1, "else" => 1, "enum" => 1, "export" => 1, "extends" => 1, "false" => 1, "final" => 1, "finally" => 1, "float" => 1, "for" => 1, "function" => 1, "goto" => 1, "if" => 1, "implements" => 1, "import" => 1, "in" => 1, "instanceof" => 1, "int" => 1, "interface" => 1, "long" => 1, "namespace" => 1, "native" => 1, "new" => 1, "null" => 1, "package" => 1, "private" => 1, "protected" => 1, "public" => 1, "return" => 1, "short" => 1, "static" => 1, "super" => 1, "switch" => 1, "synchronized" => 1, "this" => 1, "throw" => 1, "throws" => 1, "transient" => 1, "true" => 1, "try" => 1, "typeof" => 1, "var" => 1, "void" => 1, "volatile" => 1, "while" => 1, "with" => 1, "xml" => 1);
 
-	$result = $texts{0};
+	$charset = self::$charset ?: ini_get('default_charset');
+	$texts = ($values === array_values($values)) ? "[]0 " : "{}1\n";	# Numerical or associative array
+	$result = $texts[0];
 
 	foreach ($values as $key => $value)
 	{
 		$result .= $separator;
 
-		if ($texts{2})
+		if ($texts[2])
 		{
 			if ($jskeyword[$key] || !preg_match('/^[a-z_]\w*$/i', $key))
 				$key = "'$key'";
@@ -91,17 +98,19 @@ function encode($values)
 		else if (!is_array($value))
 		{
 			$quote = (strval(intval($value)) === strval($value)) ? "" : '"';
-			$string = strtr($value, array("\0" => '\\0', "\x84" => '\\"', "\x93" => '\\"',"\x94" => '\\"', '"' => '\\"', "</"=>"<\\/", "\n" => '\\n', "\r" => '\\r', "\t" => '\\t', "\\" => '\\\\'));
+			if (strtolower($charset) != "utf-8")
+				$value = strtr($value, array("\x84" => '"', "\x93" => '"', "\x94" => '"'));
+			$string = strtr($value, array("\0" => '\\0', '"' => '\\"', "</"=>"<\\/", "\n" => '\\n', "\r" => '\\r', "\t" => '\\t', "\\" => '\\\\'));
 			$string = $GLOBALS['itjs_defaultconfig']['latin2unicode'] ? preg_replace_callback('/([\xa0-\xff])/', function($m) { return sprintf("\\u%04x", ord($m[1])); }, $string) : $string;
 			$result .= $quote . $string . $quote;
 		}
 		else
 			$result .= itjs::encode($value);
 
-		$separator = "," . $texts{3};
+		$separator = "," . $texts[3];
 	}
 
-	$result .= $texts{1};
+	$result .= $texts[1];
 
 	return $result;
 }
diff --git a/itjs.php b/itjs.php
index 03073e9..f819cab 100644
--- a/itjs.php
+++ b/itjs.php
@@ -33,7 +33,7 @@ foreach ($files as $file)
 {
 	ob_start();	# Needs to capture inside loop to guarantee file order
 	if (!(it::match('W3C_CSS_Validator', $_SERVER['HTTP_USER_AGENT']) && it::match('jquery-ui\.css', $file)))
-		$data .= it::replace(array('^1$' => ""), @include($file));
+		$data .= it::replace(array('^1$' => ""), @include($file), array('utf8' => false));
 	$data .= ob_get_clean();
 }
 
@@ -50,9 +50,8 @@ else if (it::match('\.css', $_SERVER['PHP_SELF']))
 	$data .= "\n#it_boot_dom { display:none }\n";	# Append magic style for it_boot
 	if (!it::match('^devel', $GLOBALS['ULTRASERVERTYPE']))
 		$data = it::replace(array('[ \t]*([{};])[ \t]*' => '$1', '/\*.*?\*/' => ""), $data);
-	if (it::match('W3C_CSS_Validator', $_SERVER['HTTP_USER_AGENT']) || EDC('w3c')) {
+	if (it::match('W3C_CSS_Validator', $_SERVER['HTTP_USER_AGENT']) || EDC('w3c'))
 		$data = it::replace(array('@-.*' => "", 'background[^;}]*(gradient|rgba)[^;}]*;?' => "", '(filter:\s*(progid|none)|text-overflow:|zoom:|-webkit-|display:-moz-|-moz-|-o-|cursor:|border-radius:|behavior:|\w+:expression)[^;}]*;?' => "", 'html\.ie6.*' => "", '([^/])\*(\w)' => '$1$2'), $data);
-	}
 }
 else if (it::match('\.htc$', $file))
 {
@@ -64,7 +63,8 @@ else if (!it::match('\.html$', $file))
 	if ($_REQUEST['boot'] && !$_REQUEST['retry'])
 		ob_start('ob_gzhandler');
 
-	header("Content-Type: application/x-javascript; charset=iso-8859-1");
+	$charset = ini_get('default_charset') ?: 'iso-8859-1';
+	header("Content-Type: application/x-javascript; charset=$charset");
 }
 
 @header("Etag: $checksum");
@@ -80,7 +80,7 @@ if ($checksum != $_SERVER['HTTP_IF_NONE_MATCH'])
 	{
 		$data .= "window.it_boot_init();\n";
 		if (!$_REQUEST['script'])
-			$data = sprintf("/*sln:%d*/\n%s/*eln:%d*/", strlen($data), strtr($data, array('%' => "%25", '.' => "%2e", 'e' => "%65", 'i' => "%69")), strlen($data));	# Protect from Firewalls/Proxies altering Javascript source code
+			$data = sprintf("/*sln:%d*/\n%s/*eln:%d*/", grapheme_strlen($data), strtr($data, array('%' => "%25", '.' => "%2e", 'e' => "%65", 'i' => "%69")), grapheme_strlen($data));	# Protect from Firewalls/Proxies altering Javascript source code
 	}
 
 	echo it_untaint($data);