allow a 'message' field in an event schema (#72)

* allow a 'message' field in an event schema
* add unit test for nested message field; update private docstring

jupyter_events/logger.py

@@ -149,16 +149,21 @@ def register_handler(self, handler: logging.Handler) -> None:
         All outgoing messages will be formatted as a JSON string.
         """
 
-        def _skip_message(record, **kwargs):
+        def _handle_message_field(record, **kwargs):
+            """Python's logger always emits the "message" field with
+            the value as "null" unless it's present in the schema/data.
+            Message happens to be a common field for event logs,
+            so special case it here and only emit it if "message"
+            is found the in the schema's property list.
             """
-            Remove 'message' from log record.
-            It is always emitted with 'null', and we do not want it,
-            since we are always emitting events only
-            """
-            del record["message"]
+            schema = self.schemas.get(record["__schema__"])
+            if "message" not in schema.properties:
+                del record["message"]
             return json.dumps(record, **kwargs)
 
-        formatter = jsonlogger.JsonFormatter(json_serializer=_skip_message)
+        formatter = jsonlogger.JsonFormatter(
+            json_serializer=_handle_message_field,
+        )
         handler.setFormatter(formatter)
         self._logger.addHandler(handler)
         if handler not in self.handlers:

jupyter_events/schema.py

@@ -144,6 +144,10 @@ def version(self) -> int:
         """Schema's version."""
         return self._schema["version"]
 
+    @property
+    def properties(self) -> dict:
+        return self._schema["properties"]
+
     def validate(self, data: dict) -> None:
         """Validate an incoming instance of this event schema."""
         self._validator.validate(data)

tests/test_logger.py

@@ -136,7 +136,7 @@ def test_emit():
             "something": {
                 "type": "string",
                 "title": "test",
-            },
+            }
         },
     }
 
@@ -166,6 +166,97 @@ def test_emit():
     }
 
 
+def test_message_field():
+    """
+    Simple test for emitting an event with
+    the literal property "message".
+    """
+    schema = {
+        "$id": "http://test/test",
+        "version": 1,
+        "properties": {
+            "something": {
+                "type": "string",
+                "title": "test",
+            },
+            "message": {
+                "type": "string",
+                "title": "test",
+            },
+        },
+    }
+
+    output = io.StringIO()
+    handler = logging.StreamHandler(output)
+    el = EventLogger(handlers=[handler])
+    el.register_event_schema(schema)
+
+    el.emit(
+        schema_id="http://test/test",
+        data={"something": "blah", "message": "a message was seen"},
+    )
+    handler.flush()
+
+    event_capsule = json.loads(output.getvalue())
+
+    assert "__timestamp__" in event_capsule
+    # Remove timestamp from capsule when checking equality, since it is gonna vary
+    del event_capsule["__timestamp__"]
+    assert event_capsule == {
+        "__schema__": "http://test/test",
+        "__schema_version__": 1,
+        "__metadata_version__": 1,
+        "something": "blah",
+        "message": "a message was seen",
+    }
+
+
+def test_nested_message_field():
+    """
+    Simple test for emitting an event with
+    the literal property "message".
+    """
+    schema = {
+        "$id": "http://test/test",
+        "version": 1,
+        "properties": {
+            "thing": {
+                "type": "object",
+                "title": "thing",
+                "properties": {
+                    "message": {
+                        "type": "string",
+                        "title": "message",
+                    },
+                },
+            },
+        },
+    }
+
+    output = io.StringIO()
+    handler = logging.StreamHandler(output)
+    el = EventLogger(handlers=[handler])
+    el.register_event_schema(schema)
+
+    el.emit(
+        schema_id="http://test/test",
+        data={"thing": {"message": "a nested message was seen"}},
+    )
+    handler.flush()
+
+    event_capsule = json.loads(output.getvalue())
+
+    assert "__timestamp__" in event_capsule
+    # Remove timestamp from capsule when checking equality, since it is gonna vary
+    del event_capsule["__timestamp__"]
+    assert event_capsule == {
+        "__schema__": "http://test/test",
+        "__schema_version__": 1,
+        "__metadata_version__": 1,
+        "thing": {"message": "a nested message was seen"},
+    }
+
+
 def test_register_event_schema(tmp_path):
     """
     Register schema from a file