id	summary	reporter	owner	description	type	status	component	version	severity	resolution	keywords	cc	stage	has_patch	needs_docs	needs_tests	needs_better_patch	easy	ui_ux
11012	Memcached cache module fails when retrieving binary strings via cache.get, during forcing it unicode	erny	nobody	"E.g.
{{{
from zlib import compress
cache_val = compress(""sdf sdlf sdlfj sldkfj alsdkjf gallksr glasrljit rweioj tasdj gfapsdopjof ps"")
cache.set('key', cache_val)
res = cache.get('key')

DjangoUnicodeDecodeError: 'utf8' codec can't decode byte ... in position ...: ...
}}}

In [5718] ticket #4845, the following is introduced:
{{{
...
--- memcached.py	(revisión: 5717)
+++ memcached.py	(revisión: 5718)
@@ -1,6 +1,7 @@
 ""Memcached cache backend""
 
 from django.core.cache.backends.base import BaseCache, InvalidCacheBackendError
+from django.utils.encoding import smart_unicode, smart_str
 
 try:
     import cmemcache as memcache
@@ -16,17 +17,22 @@
         self._cache = memcache.Client(server.split(';'))
 
     def get(self, key, default=None):
-        val = self._cache.get(key)
+        val = self._cache.get(smart_str(key))
         if val is None:
             return default
         else:
-            return val
+            if isinstance(val, basestring):
+                return smart_unicode(val)
+            else:
+                return val
 
     def set(self, key, value, timeout=0):
-        self._cache.set(key, value, timeout or self.default_timeout)
+        if isinstance(value, unicode):
+            value = value.encode('utf-8')
+        self._cache.set(smart_str(key), value, timeout or self.default_timeout)
 
     def delete(self, key):
-        self._cache.delete(key)
+        self._cache.delete(smart_str(key))
 
     def get_many(self, keys):
-        return self._cache.get_multi(keys)
+        return self._cache.get_multi(map(smart_str,keys))
}}}

The problem is in:
{{{
-            return val
+            if isinstance(val, basestring):
+                return smart_unicode(val)
+            else:
+                return val
}}}
which makes it impossible to store binary strings. You have to encapsulate it in dummy objects, otherwise.

I still don't know why is it necessary to convert it to str (encode utf8) and back. Python memcached 1.43 pickles anything different to int, long, str, which should work fine with unicode.

"		closed	Core (Cache system)	dev		fixed		josh@… gjost@… mcroydon@…	Accepted	1	0	0	0	0	0