From fd7ee9207ba3ace3d1172b0472750d3e2d84d322 Mon Sep 17 00:00:00 2001
From: Alex Ioannidis <a.ioannidis@cern.ch>
Date: Mon, 26 Aug 2024 18:27:45 +0200
Subject: [PATCH] processors: allow filtering out robots/machines

---
 invenio_stats/processors.py | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/invenio_stats/processors.py b/invenio_stats/processors.py
index 8174a5b..83e7ad8 100644
--- a/invenio_stats/processors.py
+++ b/invenio_stats/processors.py
@@ -88,8 +88,8 @@ def anonymize_user(doc):
     return doc
 
 
-def flag_robots(doc):
-    """Flag events which are created by robots.
+def flag_robots(doc, exclude=False):
+    """Flag and filter events which are created by robots.
 
     The list of robots is defined by the `COUNTER-robots Python package
     <https://github.com/inveniosoftware/counter-robots>`_ , which follows the
@@ -99,11 +99,13 @@ def flag_robots(doc):
     <https://github.com/CDLUC3/Make-Data-Count/tree/master/user-agents>`_.
     """
     doc["is_robot"] = "user_agent" in doc and is_robot(doc["user_agent"])
+    if exclude and doc["is_robot"]:
+        return None
     return doc
 
 
-def flag_machines(doc):
-    """Flag events which are created by machines.
+def flag_machines(doc, exclude=False):
+    """Flag and filter events which are created by machines.
 
     The list of machines is defined by the `COUNTER-robots Python package
     <https://github.com/inveniosoftware/counter-robots>`_ , which follows the
@@ -114,6 +116,8 @@ def flag_machines(doc):
 
     """
     doc["is_machine"] = "user_agent" in doc and is_machine(doc["user_agent"])
+    if exclude and doc["is_machine"]:
+        return None
     return doc