Make memo _pad work with bytes, instead of strings.

This simplifies the code and removes the ambiguity with
"last byte" fetching. If the message contains unicode characters
in the end, they will be dealt with correctly now.

bitsharesbase/memo.py

@@ -63,8 +63,8 @@ def _pad(s, BS):
 
 
 def _unpad(s, BS):
-    count = int(struct.unpack('B', bytes(s[-1], 'ascii'))[0])
-    if bytes(s[-count::], 'ascii') == count * struct.pack('B', count):
+    count = s[-1]
+    if s[-count::] == count * struct.pack('B', count):
         return s[:-count]
     return s
 
@@ -116,6 +116,7 @@ def decode_memo(priv, pub, nonce, message):
     " TODO, verify checksum "
     message = cleartext[4:]
     try:
-        return _unpad(message.decode('utf8'), 16)
+        message = _unpad(message, 16)
     except Exception as e:
         raise ValueError(message)
+    return message.decode('utf8')

tests/test_memo.py

@@ -87,8 +87,8 @@ class Testcases(unittest.TestCase):
     def test_padding(self):
         for l in range(0, 255):
             s = bytes(l * chr(l), 'utf-8')
-            padded = _pad(s, 16).decode('utf-8')
-            self.assertEqual(s.decode('utf-8'), _unpad(padded, 16))
+            padded = _pad(s, 16)
+            self.assertEqual(s, _unpad(padded, 16))
 
     def test_decrypt(self):
         for memo in test_cases: