[baseformat] Fix handling of string when packing

The size used for packing the string was wrong because it used the
original string lenght rather than the encoded.

beat/backend/python/baseformat.py

@@ -267,8 +267,10 @@ def pack_array(dtype, value, fd):
         fd.write(value.tostring())
 
     elif issubclass(dtype, str):  # it is a string
-        for o in value.flat:
-            fd.write(struct.pack(STRING % len(o), len(o), o.encode('utf-8')))
+        for item in value.flat:
+            encoded = item.encode('utf-8')
+            length = len(encoded)
+            fd.write(struct.pack(STRING % length, length, encoded))
 
     else:  # it is a dataformat
         for o in value.flat:
@@ -298,8 +300,9 @@ def pack_scalar(dtype, value, fd):
             fd.write(struct.pack(ENDIANNESS + BINCODE[dtype], value))
 
     elif issubclass(dtype, str):  # it is a string
-        fd.write(struct.pack(STRING % len(value), len(value),
-            value.encode('utf8')))
+        encoded = value.encode('utf-8')
+        length = len(encoded)
+        fd.write(struct.pack(STRING % length, length, encoded))
 
     else:  # it is a dataformat
         value.pack_into(fd)