Add latin1 decoding to HTTP Headers

common/djangoapps/track/middleware.py

@@ -140,12 +140,11 @@ class TrackMiddleware(object):
             'ip': self.get_request_ip_address(request),
         }
         for header_name, context_key in META_KEY_TO_CONTEXT_KEY.iteritems():
-            context[context_key] = request.META.get(header_name, '')
+            # HTTP headers may contain Latin1 characters. Decoding using Latin1 encoding here
+            # avoids encountering UnicodeDecodeError exceptions when these header strings are
+            # output to tracking logs.
+            context[context_key] = request.META.get(header_name, '').decode('latin1')
 
-        # HTTP_USER_AGENT user might can contain the information that include latin1 characters
-        # decoding this using latin1 scheme will prevent to raise UnicodeDecodeError when using
-        # json.dumps for tracking purpose.
-        context['agent'] = context['agent'].decode('latin1')
         # Google Analytics uses the clientId to keep track of unique visitors. A GA cookie looks like
         # this: _ga=GA1.2.1033501218.1368477899. The clientId is this part: 1033501218.1368477899.
         google_analytics_cookie = request.COOKIES.get('_ga')

common/djangoapps/track/tests/test_middleware.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 """Tests for tracking middleware."""
+import ddt
 from mock import patch
 from mock import sentinel
 
@@ -13,7 +14,9 @@ from eventtracking import tracker
 from track.middleware import TrackMiddleware
 
 
+@ddt.ddt
 class TrackMiddlewareTestCase(TestCase):
+    """  Class for checking tracking requests """
 
     def setUp(self):
         super(TrackMiddlewareTestCase, self).setUp()
@@ -29,17 +32,25 @@ class TrackMiddlewareTestCase(TestCase):
         self.track_middleware.process_request(request)
         self.assertTrue(self.mock_server_track.called)
 
-    def test_request_with_latin1_characters(self):
+    @ddt.unpack
+    @ddt.data(
+        ('HTTP_USER_AGENT', 'agent'),
+        ('PATH_INFO', 'path'),
+        ('HTTP_REFERER', 'referer'),
+        ('HTTP_ACCEPT_LANGUAGE', 'accept_language'),
+    )
+    def test_request_with_latin1_characters(self, meta_key, context_key):
         """
-        When HTTP_USER_AGENT in request.META contains latin1 characters.
+        When HTTP headers contains latin1 characters.
         """
         request = self.request_factory.get('/somewhere')
-        request.META['HTTP_USER_AGENT'] = 'test latin1 \xd3 \xe9 \xf1'  # pylint: disable=no-member
+        # pylint: disable=no-member
+        request.META[meta_key] = 'test latin1 \xd3 \xe9 \xf1'  # pylint: disable=no-member
 
         context = self.get_context_for_request(request)
         # The bytes in the string on the right are utf8 encoded in the source file, so we decode them to construct
         # a valid unicode string.
-        self.assertEqual(context['agent'], 'test latin1 Ó é ñ'.decode('utf8'))
+        self.assertEqual(context[context_key], 'test latin1 Ó é ñ'.decode('utf8'))
 
     def test_default_filters_do_not_render_view(self):
         for url in ['/event', '/event/1', '/login', '/heartbeat']: