replace multi-level approach to tagging approach

jupyter_telemetry/eventlog.py

@@ -116,17 +116,13 @@ def register_schema(self, schema):
         jsonschema.validators.validator_for(schema).check_schema(schema)
 
         # Check that the properties we require are present
-        required_schema_fields = {'$id', 'version'}
+        required_schema_fields = {'$id', 'version', 'properties'}
         for rsf in required_schema_fields:
             if rsf not in schema:
                 raise ValueError(
                     '{} is required in schema specification'.format(rsf)
                 )
 
-        # Verify that all properties have a sensitivity level.
-        if
-
-
         # Make sure reserved, auto-added fields are not in schema
         if any([p.startswith('__') for p in schema['properties']]):
             raise ValueError(

jupyter_telemetry/formatter.py

@@ -1,62 +1,58 @@
+from traitlets import HasTraits, validate, Set
 from pythonjsonlogger import jsonlogger
 
 
-EVENT_MAP = {
-    'unrestricted': 0,
-    'user-identifier': 10,
-    'user-identifiable-information': 20
-}
-
-EVENT_LEVELS = list(EVENT_MAP.keys())
-
-
 class JsonEventFormatter(jsonlogger.JsonFormatter):
     """Patch the jsonlogger formatter to include levels for telemetry.
 
     Properties in a logged event that has a level less than
     the handler's event_level will be dropped from the emitted event.
     """
-
     def __init__(self, logger, handler, *args, **kwargs):
         self.logger = logger
         self.handler = handler
-        # Set the event logging level
-        self.event_level = getattr(handler, 'event_level', None)
         super(JsonEventFormatter, self).__init__(*args, **kwargs)
 
-    # Protect the event_level attribute.
     @property
-    def event_level(self):
-        """Event Log security level."""
-        return self._event_level
+    def allowed_tags(self):
+        return getattr(self.handler, 'allowed_tags', {})
 
-    @event_level.setter
-    def event_level(self, event_level):
-        # Check that the event level makes sense.
-        if event_level not in EVENT_LEVELS:
-            raise Exception("Event level '{}' not understood.".format(event_level))
-        self._event_level = self.handler.event_level
+    @property
+    def hashed_tags(self):
+        return getattr(self.handler, 'hashed_tags', {})
 
     def process_log_record(self, log_record):
         log_record = super(JsonEventFormatter, self).process_log_record(log_record)
-        return self.process_event_levels(log_record)
+        return self.process_tags(log_record)
 
-    def process_event_levels(self, log_record):
-        """Removes any properties in a log_record that have an attribute `pii = True`.
+    def drop_property(self, key, record):
+        del record[key]
+        return record
+
+    def hash_property(self, key, record):
+        hash_function = lambda x: x
+        record[key] = hash_function(record[key])
+        return record
+
+    def process_tags(self, log_record):
         """
-        # Get schema for this log_record
-        key = (log_record['__schema__'], log_record['__version__'])
+        """
+        # Registered schemas are identified by their name and version.
+        key = (log_record['__schema__'], log_record['__schema_version__'])
         schema = self.logger.schemas[key]['properties']
-
-        # Find all properties that have a level less than the handler.
-        keys = list(log_record.keys())
-        for key in keys:
-            # Ignore any keys that start with __
-            if not key.startswith('__'):
-                # Check if security level is listed in the schema.
-                if EVENT_MAP[schema[key]['level']] > EVENT_MAP[self.event_level]:
-                    # If property's level is less than handler's level,
-                    # delete this property from the log record.
-                    del log_record[key]
+        props = [key for key in log_record.keys()
+                    if not key.startswith('__') and key != 'message']
+
+        # Walk through the recorded event and handle each key
+        # based on its tag/category.
+        for key in props:
+            tag = schema[key]['tag']
+            if tag in self.allowed_tags or tag == "unrestricted":
+                # If the tag is found in the allowed_tags trait, do nothing.
+                if tag in self.hashed_tags:
+                    log_record = self.hash_property(key, log_record)
+            # Drop tags not listed in allowed_tags
+            else:
+                log_record = self.drop_property(key, log_record)
 
         return log_record

tests/test_levels.py → tests/test_tags.py

@@ -24,39 +24,39 @@ def schema(schema_id, version):
         'description': 'Test Event.',
         'type': 'object',
         'properties': {
+            'nothing-exciting': {
+                'description': 'a property with nothing exciting happening',
+                'tag': 'unrestricted',
+                'type': 'string'
+            },
             'id': {
-                'description': 'user ID', 
-                'level': 'confidential',
+                'description': 'user ID',
+                'tag': 'user-identifier',
                 'type': 'string'
             },
             'email': {
                 'description': 'email address',
-                'level': 'secret',
+                'tag': 'user-identifiable-information',
                 'type': 'string'
             },
-            'name': { 
-                'description': 'name of user',
-                'level': 'top_secret',
-                'type': 'string'
-            }
         }
     }
 
 
 @pytest.mark.parametrize(
-    'level,expected_props',
+    'tags,expected_props',
     [
-        ('unclassified', set()),
-        ('confidential', {'id'}),
-        ('secret', {'id', 'email'}),
-        ('top_secret', {'id', 'email', 'name'})
+        ({'unrestricted'}, {'nothing-exciting'}),
+        ({'user-identifier'}, {'nothing-exciting', 'id'}),
+        ({'user-identifiable-information'}, {'nothing-exciting', 'email'})
     ]
 )
-def test_drop_sensitive_properties(schema, schema_id, version, level, expected_props):
+def test_properties_tags(schema, schema_id, version, tags, expected_props):
     sink = io.StringIO()
+
+    # Create a handler that captures+records events with allowed tags.
     handler = logging.StreamHandler(sink)
-    # Set the event level
-    handler.event_level = level
+    handler.allowed_tags = tags
 
     e = EventLog(
         handlers=[handler],
@@ -65,19 +65,15 @@ def test_drop_sensitive_properties(schema, schema_id, version, level, expected_p
     e.register_schema(schema)
 
     event = {
+        'nothing-exciting': 'hello, world',
         'id': 'test id',
         'email': '[email protected]',
-        'name': 'test name'
     }
 
-    # Get a set of keys that should be missing
-    dropped_props = set(event.keys()) - expected_props
-
     # Record event and read output
     e.record_event(schema_id, version, event)
     recorded_event = json.loads(sink.getvalue())
-    recorded_props = set(recorded_event.keys())
+    recorded_props = set([key for key in recorded_event if not key.startswith('__')])
+
+    assert expected_props == recorded_props
 
-    # Assert that sensitive properties are dropped.
-    assert len(dropped_props.intersection(recorded_props)) == 0
-