Fix handling of unicode in XML in Python 2

Fixes #996
Signed-off-by: Michal Čihař <michal@cihar.com>

weblate/trans/checks/markup.py

@@ -25,6 +25,8 @@ import re
 
 from django.utils.translation import ugettext_lazy as _
 
+import six
+
 from weblate.trans.checks.base import TargetCheck
 
 BBCODE_MATCH = re.compile(
@@ -95,8 +97,14 @@ class XMLTagsCheck(TargetCheck):
         '''
         Wrapper for parsing XML.
         '''
-        text = strip_entities(text)
-        return cElementTree.fromstring('<weblate>%s</weblate>' % text)
+        text = ''.join(
+            ('<weblate>', strip_entities(text), '</weblate>')
+        )
+
+        if six.PY2:
+            text = text.encode('utf-8')
+
+        return cElementTree.fromstring(text)
 
     def check_single(self, source, target, unit):
         # Quick check if source looks like XML

weblate/trans/tests/test_markup_checks.py

@@ -22,6 +22,7 @@
 Tests for quality checks.
 """
 
+from __future__ import unicode_literals
 from weblate.trans.checks.markup import (
     BBCodeCheck,
     XMLTagsCheck,
@@ -63,3 +64,6 @@ class XMLTagsCheckTest(CheckTestCase):
                 (24, 28, '</b>'),
             ]
         )
+
+    def test_unicode(self):
+        self.do_test(False, ('<a>zkouška</a>', '<a>zkouška</a>', ''))