Merge pull request #1954 from edx/fix/will/md5-unicode-error

Fix UnicodeEncodeError when generating cache keys

common/lib/capa/capa/safe_exec/safe_exec.py

@@ -58,7 +58,7 @@ def safe_exec(code, globals_dict, random_seed=None, python_path=None, cache=None
     if cache:
         canonical_globals = sorted(json_safe(globals_dict).iteritems())
         md5er = hashlib.md5()
-        md5er.update(code)
+        md5er.update(repr(code))
         md5er.update(repr(canonical_globals))
         key = "safe_exec.%r.%s" % (random_seed, md5er.hexdigest())
         cached = cache.get(key)

common/lib/capa/capa/safe_exec/tests/test_safe_exec.py

@@ -143,6 +143,19 @@ class TestSafeExecCaching(unittest.TestCase):
         safe_exec(code, g, cache=DictCache(cache))
         self.assertEqual(g['a'], 17)
 
+    def test_unicode_submission(self):
+        # Check that using non-ASCII unicode does not raise an encoding error.
+    
+        # Try several non-ASCII unicode characters
+        for code in [129, 500, 2**8 - 1, 2**16 - 1]:
+
+            code_with_unichr = unicode("# ") + unichr(code)
+
+            try:
+                safe_exec(code_with_unichr, {}, cache=DictCache({}))
+            except UnicodeEncodeError:
+                self.fail("Tried executing code with non-ASCII unicode: {0}".format(code))
+
 
 class TestRealProblems(unittest.TestCase):
     def test_802x(self):