Add support of unicode type

xml_marshaller/xml_marshaller.py

@@ -48,6 +48,7 @@ class Marshaller(object):
     self.tag_float = E.float
     self.tag_long = E.long
     self.tag_string = E.string
+    self.tag_unicode = E.unicode
     self.tag_tuple = E.tuple
     self.tag_list = E.list
     self.tag_dictionary = E.dictionary
@@ -136,6 +137,8 @@ class Marshaller(object):
   def m_string(self, value, kw):
     return self.tag_string(escape(str(value)))
 
+  def m_unicode(self, value, kw):
+    return self.tag_unicode(value)
   # Since Python 2.2, the string type has a name of 'str'
   # To avoid having to rewrite all classes that implement m_string
   # we delegate m_str to m_string.
@@ -259,6 +262,7 @@ class Unmarshaller(ElementTreeContentHandler):
       'float': ('um_start_float', 'um_end_float'),
       'long': ('um_start_long', 'um_end_long'),
       'string': ('um_start_string', 'um_end_string'),
+      'unicode': ('um_start_unicode', 'um_end_unicode'),
       'tuple': ('um_start_tuple', 'um_end_tuple'),
       'list': ('um_start_list', 'um_end_list'),
       'dictionary': ('um_start_dictionary', 'um_end_dictionary'),
@@ -408,7 +412,7 @@ class Unmarshaller(ElementTreeContentHandler):
     self.data_stack.append([])
     self.accumulating_chars = 1
 
-  um_start_float = um_start_long = um_start_string = um_start_generic
+  um_start_float = um_start_long = um_start_string = um_start_unicode = um_start_generic
   um_start_complex = um_start_code = um_start_none = um_start_generic
   um_start_int = um_start_generic
 
@@ -418,6 +422,12 @@ class Unmarshaller(ElementTreeContentHandler):
     ds[-1] = unescape(''.join(ds[-1]))
     self.accumulating_chars = 0
 
+  def um_end_unicode(self, name):
+    ds = self.data_stack
+    # might need to convert unicode string to byte string
+    ds[-1] = unescape(''.join(ds[-1]))
+    self.accumulating_chars = 0
+
   def um_end_int(self, name):
     ds = self.data_stack
     ds[-1] = ''.join(ds[-1])
@@ -599,7 +609,8 @@ def runtests(namespace_uri=None):
        "here is a string & a <fake tag>",
        (1, 2, 3),
        ['alpha', 'beta', 'gamma'],
-       {'key': 'value', 1: 2}
+       {'key': 'value', 1: 2},
+       'éàù^ç'.decode('utf-8'),
        ]
   if namespace_uri:
     test(load, loads, dump_ns, dumps_ns, L)