gistfile1.diff

--------------- quintagroup/transmogrifier/adapters/importing.py ---------------
index 70f21ae..57e3bcd 100644
@@ -11,9 +11,23 @@ from collective.transmogrifier.interfaces import ITransmogrifier
 
 from quintagroup.transmogrifier.interfaces import IImportDataCorrector
 
+from xml.parsers.expat import ExpatError
+
 EXISTING_UIDS = {}
 REFERENCE_QUEUE = {}
 
+def guess_encoding(text):
+    for best_enc in ['utf-8', 'us-ascii', 'iso-8859-1', 'iso-8859-2',
+                     'windows-1250', 'windows-1252']:
+        try:
+            unicode(text,best_enc,"strict")
+        except:
+            pass
+        else:
+            break
+    return best_enc
+
+
 class ReferenceImporter(object):
     """ Demarshall content from xml file by using of Marshall product.
     """
@@ -45,14 +59,20 @@ class ReferenceImporter(object):
     def importReferences(self, data):
         """ Marshall 1.0.0 doesn't import references, do it manually.
         """
-        doc = minidom.parseString(data)
+        try:
+            doc = minidom.parseString(data)
+        except ExpatError:
+            doc = minidom.parseString(data.decode(
+                    guess_encoding(data)).encode('utf-8'))
         root = doc.documentElement
         for fname in self.context.Schema().keys():
-            if not isinstance(self.context.Schema()[fname], atapi.ReferenceField):
+            if not isinstance(self.context.Schema()[fname],
+                              atapi.ReferenceField):
                 continue
             uids = []
             validUIDs = True
-            elements = [i for i in root.getElementsByTagName('field') if i.getAttribute('name') == fname]
+            elements = [i for i in root.getElementsByTagName('field')
+                        if i.getAttribute('name') == fname]
             if not elements:
                 # if needed elements are absent skip this field
                 # update as much as posible fields and don't raise exceptions
@@ -67,7 +87,8 @@ class ReferenceImporter(object):
                 mutator = self.context.Schema()[fname].getMutator(self.context)
                 mutator(uids)
             else:
-                suid = str(root.getElementsByTagName('uid')[0].firstChild.nodeValue.strip())
+                suid = str(root.getElementsByTagName('uid')\
+                               [0].firstChild.nodeValue.strip())
                 REFERENCE_QUEUE[suid] = {}
                 REFERENCE_QUEUE[suid][fname] = uids
             root.removeChild(elem)