Commit df09d854 authored by Aymeric Augustin's avatar Aymeric Augustin
Browse files

Fixed #17552 -- Removed a hack for IE6 and earlier. 

It prevented the GZipMiddleware from compressing some data types even on
more recent version of IE where the corresponding bug was fixed.

Thanks Aaron Cannon for the report and Tim Graham for the review.
parent fea8ccdc

django/middleware/gzip.py

+1 −7
Original line number Diff line number Diff line
@@ -17,17 +17,11 @@ class GZipMiddleware(object): 
        if not response.streaming and len(response.content) < 200:

            return response



        patch_vary_headers(response, ('Accept-Encoding',))



        # Avoid gzipping if we've already got a content-encoding.

        if response.has_header('Content-Encoding'):

            return response



        # MSIE have issues with gzipped response of various content types.

        if "msie" in request.META.get('HTTP_USER_AGENT', '').lower():

            ctype = response.get('Content-Type', '').lower()

            if not ctype.startswith("text/") or "javascript" in ctype:

                return response

        patch_vary_headers(response, ('Accept-Encoding',))



        ae = request.META.get('HTTP_ACCEPT_ENCODING', '')

        if not re_accepts_gzip.search(ae):

docs/ref/middleware.txt

+0 −5
Original line number Diff line number Diff line
@@ -108,11 +108,6 @@ It will NOT compress content if any of the following are true: 
* The request (the browser) hasn't sent an ``Accept-Encoding`` header

  containing ``gzip``.



* The request is from Internet Explorer and the ``Content-Type`` header

  contains ``javascript`` or starts with anything other than ``text/``.

  We do this to avoid a bug in early versions of IE that caused decompression

  not to be performed on certain content types.



You can apply GZip compression to individual views using the

:func:`~django.views.decorators.gzip.gzip_page()` decorator.

docs/releases/1.8.txt

+5 −0
Original line number Diff line number Diff line
@@ -273,6 +273,11 @@ Miscellaneous 
* Database connections are considered equal only if they're the same object.

  They aren't hashable any more.



* :class:`~django.middleware.gzip.GZipMiddleware` used to disable compression

  for some content types when the request is from Internet Explorer, in order

  to work around a bug in IE6 and earlier. This behavior could affect

  performance on IE7 and later. It was removed.



* ``URLField.to_python`` no longer adds a trailing slash to pathless URLs.



* ``django.contrib.gis`` dropped support for GEOS 3.1 and GDAL 1.6.

tests/middleware/tests.py

+0 −10
Original line number Diff line number Diff line
@@ -613,16 +613,6 @@ class GZipMiddlewareTest(TestCase): 
        self.assertEqual(r.content, self.compressible_string)

        self.assertEqual(r.get('Content-Encoding'), 'deflate')



    def test_no_compress_ie_js_requests(self):

        """

        Tests that compression isn't performed on JavaScript requests from Internet Explorer.

        """

        self.req.META['HTTP_USER_AGENT'] = 'Mozilla/4.0 (compatible; MSIE 5.00; Windows 98)'

        self.resp['Content-Type'] = 'application/javascript; charset=UTF-8'

        r = GZipMiddleware().process_response(self.req, self.resp)

        self.assertEqual(r.content, self.compressible_string)

        self.assertEqual(r.get('Content-Encoding'), None)



    def test_no_compress_uncompressible_response(self):

        """

        Tests that compression isn't performed on responses with uncompressible content.