Merge pull request #106 from sigmavirus24/bug/43

Fix query matcher handling unicode

betamax/matchers/query.py

@@ -1,11 +1,15 @@
 # -*- coding: utf-8 -*-
+import sys
+
 from .base import BaseMatcher
-from requests.compat import urlparse
 
 try:
-    from urlparse import parse_qs
+    from urlparse import parse_qs, urlparse
 except ImportError:
-    from urllib.parse import parse_qs
+    from urllib.parse import parse_qs, urlparse
+
+
+isPY2 = (2, 6) <= sys.version_info < (3, 0)
 
 
 class QueryMatcher(BaseMatcher):
@@ -17,8 +21,14 @@ def to_dict(self, query):
         return parse_qs(query or '')  # Protect against None
 
     def match(self, request, recorded_request):
-        request_query = self.to_dict(urlparse(request.url).query)
-        recorded_query = self.to_dict(
-            urlparse(recorded_request['uri']).query
-        )
-        return request_query == recorded_query
+        request_query_dict = self.to_dict(urlparse(request.url).query)
+        recorded_query = urlparse(recorded_request['uri']).query
+        if recorded_query and isPY2:
+            # NOTE(sigmavirus24): If we're on Python 2, the request.url will
+            # be str/bytes and the recorded_request['uri'] will be unicode.
+            # For the comparison to work for high unicode characters, we need
+            # to encode the recorded query string before parsing it. See also
+            # GitHub bug #43.
+            recorded_query = recorded_query.encode('utf-8')
+        recorded_query_dict = self.to_dict(recorded_query)
+        return request_query_dict == recorded_query_dict