unicode: Added handling for illegaly encoded form input.

git-svn-id: http://code.djangoproject.com/svn/django/branches/unicode@5197 bcc190cf-cafb-0310-a4f2-bffc1f526a37

django/http/__init__.py

@@ -360,12 +360,14 @@ def get_host(request):
 # this slightly more restricted function.
 def str_to_unicode(s, encoding):
     """
-    Convert basestring objects to unicode, using the given encoding.
+    Convert basestring objects to unicode, using the given encoding. Illegaly
+    encoded input characters are replaced with Unicode "unknown" codepoint
+    (\ufffd).
 
     Returns any non-basestring objects without change.
     """
     if isinstance(s, str):
-        return unicode(s, encoding)
+        return unicode(s, encoding, 'replace')
     else:
         return s
 

tests/regressiontests/httpwrappers/tests.py

@@ -367,6 +367,16 @@ AttributeError: This QueryDict instance is immutable
 >>> q.urlencode()
 'vote=yes&vote=no'
 
+# QueryDicts must be able to handle invalid input encoding (in this case, bad
+# UTF-8 encoding).
+>>> q = QueryDict('foo=bar&foo=\xff')
+
+>>> q['foo']
+u'\ufffd'
+
+>>> q.getlist('foo')
+[u'bar', u'\ufffd']
+
 """
 
 from django.http import QueryDict