Make sure scripts in imports with unicode characters are kept in the …

…correct encoding

Fixes Polymer/polymer#498

add test for unicode scripts

src/Parser.js

@@ -234,15 +234,16 @@ function nodeIsImport(elt) {
 }
 
 function generateScriptDataUrl(script) {
-  var scriptContent = generateScriptContent(script), b64;
+  var scriptContent = generateScriptContent(script);
+  var b64 = 'data:text/javascript';
+  // base64 may be smaller, but does not handle unicode characters
+  // attempt base64 first, fall back to escaped text
   try {
-    b64 = btoa(scriptContent);
+    b64 += (';base64,' + btoa(scriptContent));
   } catch(e) {
-    b64 = btoa(unescape(encodeURIComponent(scriptContent)));
-    console.warn('Script contained non-latin characters that were forced ' +
-      'to latin. Some characters may be wrong.', script);
+    b64 += (';charset=utf-8,' + encodeURIComponent(scriptContent));
   }
-  return 'data:text/javascript;base64,' + b64;
+  return b64;
 }
 
 function generateScriptContent(script) {

test/html/imports/parsed-import-1.html

@@ -1,4 +1,7 @@
 <link rel="import" href="parsed-import-2.html">
 <script>
   window.inlineScriptParsed = new Date().getTime();
-</script>
+</script>
+<script>
+  window.uc = '♡ 㒨';
+</script>

test/html/imports/parsed-import-2.html

@@ -1 +1 @@
-<script src="external-script.js"></script>
+<script src="external-script.js"></script>

test/html/parser.html

@@ -12,6 +12,7 @@
       addEventListener('HTMLImportsLoaded', function() {
         chai.assert.ok(window.inlineScriptParsed, 'inlineScriptParsed');
         chai.assert.ok(window.externalScriptParsed, 'externalScriptParsed');
+        chai.assert.equal(window.uc, '\u2661 \u34A8', 'unicode matches');
         done();
       });
     </script>