unescape surrogate pairs only (#2246)

fixes #2242

1 files changed, 11 insertions, 8 deletions

diff --git a/lib/output.js b/lib/output.js
index edb8d182..4c873f10 100644
--- a/lib/output.js
+++ b/lib/output.js
@@ -109,7 +109,7 @@ function OutputStream(options) {
     var current_pos = 0;
     var OUTPUT = "";
 
-    function to_ascii(str, identifier) {
+    var to_utf8 = options.ascii_only ? function(str, identifier) {
         return str.replace(/[\u0000-\u001f\u007f-\uffff]/g, function(ch) {
             var code = ch.charCodeAt(0).toString(16);
             if (code.length <= 2 && !identifier) {
@@ -120,6 +120,12 @@ function OutputStream(options) {
                 return "\\u" + code;
             }
         });
+    } : function(str) {
+        return str.replace(/[\ud800-\udbff](?![\udc00-\udfff])/g, function(ch) {
+            return "\\u" + ch.charCodeAt(0).toString(16);
+        }).replace(/(^|[^\ud800-\udbff])([\udc00-\udfff])/g, function(match, prefix, ch) {
+            return prefix + "\\u" + ch.charCodeAt(0).toString(16);
+        });
     };
 
     function make_string(str, quote) {
@@ -150,7 +156,7 @@ function OutputStream(options) {
         function quote_double() {
             return '"' + str.replace(/\x22/g, '\\"') + '"';
         }
-        if (options.ascii_only) str = to_ascii(str);
+        str = to_utf8(str);
         switch (options.quote_style) {
           case 1:
             return quote_single();
@@ -175,8 +181,7 @@ function OutputStream(options) {
 
     function make_name(name) {
         name = name.toString();
-        if (options.ascii_only)
-            name = to_ascii(name, true);
+        name = to_utf8(name, true);
         return name;
     };
 
@@ -433,7 +438,7 @@ function OutputStream(options) {
         last            : function() { return last },
         semicolon       : semicolon,
         force_semicolon : force_semicolon,
-        to_ascii        : to_ascii,
+        to_utf8         : to_utf8,
         print_name      : function(name) { print(make_name(name)) },
         print_string    : function(str, quote, escape_directive) {
             var encoded = encode_string(str, quote);
@@ -1318,9 +1323,7 @@ function OutputStream(options) {
         if (regexp.raw_source) {
             str = "/" + regexp.raw_source + str.slice(str.lastIndexOf("/"));
         }
-        if (output.option("ascii_only")) {
-            str = output.to_ascii(str);
-        }
+        str = output.to_utf8(str);
         output.print(str);
         var p = output.parent();
         if (p instanceof AST_Binary && /^in/.test(p.operator) && p.left === self)