Strip invalid characters from base64 strings (fix #33)

diff --git a/index.js b/index.js
index 533f8f002d2a70739a49cff9a46822c3ad3ce551..9526e880f32132d05a281d4d2c1363151fb743e5 100644 (file)
--- a/index.js
+++ b/index.js
@@ -53,13 +53,8 @@ function Buffer (subject, encoding, noZero) {
 
   var type = typeof subject
 
-  // Workaround: node's base64 implementation allows for non-padded strings
-  // while base64-js does not.
   if (encoding === 'base64' && type === 'string') {
-    subject = stringtrim(subject)
-    while (subject.length % 4 !== 0) {
-      subject = subject + '='
-    }
+    subject = base64clean(subject)
   }
 
   // Find the length
@@ -1016,6 +1011,18 @@ Buffer._augment = function (arr) {
   return arr
 }
 
+var INVALID_BASE64_RE = /[^+\/0-9A-z]/g
+
+function base64clean (str) {
+  // Node strips out invalid characters like \n and \t from the string, base64-js does not
+  str = stringtrim(str).replace(INVALID_BASE64_RE, '')
+  // Node allows for non-padded base64 strings (missing trailing ===), base64-js does not
+  while (str.length % 4 !== 0) {
+    str = str + '='
+  }
+  return str
+}
+
 function stringtrim (str) {
   if (str.trim) return str.trim()
   return str.replace(/^\s+|\s+$/g, '')