escape semicolons when serializing aggregated activities

Author: izhan

stream_framework/serializers/aggregated_activity_serializer.py

@@ -5,6 +5,7 @@
 from stream_framework.utils import epoch_to_datetime, datetime_to_epoch
 from stream_framework.serializers.base import BaseAggregatedSerializer
 import six
+import re
 
 
 class AggregatedActivitySerializer(BaseAggregatedSerializer):
@@ -23,7 +24,6 @@ class AggregatedActivitySerializer(BaseAggregatedSerializer):
     #: indicates if dumps returns dehydrated aggregated activities
     dehydrate = True
     identifier = 'v3'
-    reserved_characters = [';', ',', ';;']
     date_fields = ['created_at', 'updated_at', 'seen_at', 'read_at']
 
     activity_serializer_class = ActivitySerializer
@@ -55,7 +55,8 @@ def dumps(self, aggregated):
         else:
             for activity in aggregated.activities:
                 serialized = activity_serializer.dumps(activity)
-                check_reserved(serialized, [';', ';;'])
+                # we use semicolons as delimiter, so need to escape
+                serialized = serialized.replace(";", "\;")
                 serialized_activities.append(serialized)
 
         serialized_activities_part = ';'.join(serialized_activities)
@@ -71,41 +72,42 @@ def dumps(self, aggregated):
 
     def loads(self, serialized_aggregated):
         activity_serializer = self.activity_serializer_class(Activity)
-        try:
-            serialized_aggregated = serialized_aggregated[2:]
-            parts = serialized_aggregated.split(';;')
-            # start with the group
-            group = parts[0]
-            aggregated = self.aggregated_activity_class(group)
-
-            # get the date and activities
-            date_dict = dict(zip(self.date_fields, parts[1:5]))
-            for k, v in date_dict.items():
-                date_value = None
-                if v != '-1':
-                    date_value = epoch_to_datetime(float(v))
-                setattr(aggregated, k, date_value)
-
-            # write the activities
-            serializations = parts[5].split(';')
-            if self.dehydrate:
-                activity_ids = list(map(int, serializations))
-                aggregated._activity_ids = activity_ids
-                aggregated.dehydrated = True
-            else:
-                activities = [activity_serializer.loads(s)
-                              for s in serializations]
-                aggregated.activities = activities
-                aggregated.dehydrated = False
-
-            # write the minimized activities
-            minimized = int(parts[6])
-            aggregated.minimized_activities = minimized
-
-            return aggregated
-        except Exception as e:
-            msg = six.text_type(e)
-            raise SerializationException(msg)
+        serialized_aggregated = serialized_aggregated[2:]
+        parts = serialized_aggregated.split(';;')
+        # start with the group
+        group = parts[0]
+        aggregated = self.aggregated_activity_class(group)
+
+        # get the date and activities
+        date_dict = dict(zip(self.date_fields, parts[1:5]))
+        for k, v in date_dict.items():
+            date_value = None
+            if v != '-1':
+                date_value = epoch_to_datetime(float(v))
+            setattr(aggregated, k, date_value)
+
+        # looks for ; not \;
+        unescaped_semicolons_regex = re.compile("(?<=[^\\\]);")
+        # write the activities
+        serializations = unescaped_semicolons_regex.split(parts[5])
+        if self.dehydrate:
+            activity_ids = list(map(int, serializations))
+            aggregated._activity_ids = activity_ids
+            aggregated.dehydrated = True
+        else:
+            activities = []
+            for s in serializations:
+                s = s.replace("\;", ";")
+                deserialized = activity_serializer.loads(s)
+                activities.append(deserialized)
+            aggregated.activities = activities
+            aggregated.dehydrated = False
+
+        # write the minimized activities
+        minimized = int(parts[6])
+        aggregated.minimized_activities = minimized
+
+        return aggregated
 
 
 class NotificationSerializer(AggregatedActivitySerializer):