Decode XML metadata as UTF-8

XML uses UTF-8 by default, which needs to be decoded to a Javascript
String prior to feeding it to the DOMParser.
In an ideal world, the XML would actually be analyzed and the specified
charset would be used, however that does not seem feasible unless JS
engines get iconv bindings.

Fixes GH-1692

diff --git a/src/obj.js b/src/obj.js
index 3432ac68dd6b3686c3f0813111945ee4153715b6..acc9e1284818ab79e29f54ede0e8acd4fa37e4cc 100644 (file)
--- a/src/obj.js
+++ b/src/obj.js
@@ -140,7 +140,12 @@ var Catalog = (function CatalogClosure() {
 
         if (isName(type) && isName(subtype) &&
             type.name === 'Metadata' && subtype.name === 'XML') {
-          metadata = stringToPDFString(bytesToString(stream.getBytes()));
+          // XXX: This should examine the charset the XML document defines,
+          // however since there are currently no real means to decode
+          // arbitrary charsets, let's just hope that the author of the PDF
+          // was reasonable enough to stick with the XML default charset,
+          // which is UTF-8.
+          metadata = stringToUTF8String(bytesToString(stream.getBytes()));
         }
       }
 

diff --git a/src/util.js b/src/util.js
index 90e6cee5d62737dac80205a61d29340f3d0c8b86..fe5d895e3312a9ac47b20a58b027d2873acf4dd8 100644 (file)
--- a/src/util.js
+++ b/src/util.js
@@ -302,6 +302,10 @@ function stringToPDFString(str) {
   return str2;
 }
 
+function stringToUTF8String(str) {
+  return decodeURIComponent(escape(str));
+}
+
 function isBool(v) {
   return typeof v == 'boolean';
 }