From 45d6f42623ce9ff9dacfb1c8578ae9271f8e1ea2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Petr=20Ja=C5=A1ek?= <jasekpetr@gmail.com>
Date: Tue, 1 Feb 2022 19:37:40 +0100
Subject: [PATCH] allow filtering of subjects before storing (#65)

based on config it will filter out subjects which
are not used and should not be visible.

NHUB-63
---
 newsroom/push.py                 |  6 ++++++
 newsroom/web/default_settings.py |  8 ++++++++
 tests/core/test_push.py          | 12 ++++++++++++
 3 files changed, 26 insertions(+)

diff --git a/newsroom/push.py b/newsroom/push.py
index 3be4b9d74..6eefc8d5d 100644
--- a/newsroom/push.py
+++ b/newsroom/push.py
@@ -146,6 +146,12 @@ def publish_item(doc, original):
         logger.info('Failed to notify new wire item for Agenda watches')
         logger.exception(ex)
 
+    if app.config.get("WIRE_SUBJECT_SCHEME_WHITELIST") and doc.get("subject"):
+        doc["subject"] = [
+            subject for subject in doc["subject"]
+            if subject.get("scheme") in app.config["WIRE_SUBJECT_SCHEME_WHITELIST"]
+        ]
+
     publish_item_signal.send(app._get_current_object(), item=doc, is_new=original is None)
     _id = service.create([doc])[0]
     if 'associations' not in doc and original is not None and bool(original.get('associations', {})):
diff --git a/newsroom/web/default_settings.py b/newsroom/web/default_settings.py
index dd82fe41e..bb533b070 100644
--- a/newsroom/web/default_settings.py
+++ b/newsroom/web/default_settings.py
@@ -412,6 +412,14 @@
 APM_SECRET_TOKEN = env("APM_SECRET_TOKEN")
 APM_SERVICE_NAME = env("APM_SERVICE_NAME") or SITE_NAME
 
+#: Filter out subjects with schema which is not in the whitelist
+#: before storing the item to avoid those being displayed in filter,
+#: preview and outputs.
+#:
+#: .. versionadded:: 2.1
+#:
+WIRE_SUBJECT_SCHEME_WHITELIST = []
+
 #: Agenda Filter groups (defaults set in ``newsroom.agenda.init_app``)
 #:
 #: .. versionadded:: 2.1.0
diff --git a/tests/core/test_push.py b/tests/core/test_push.py
index eff23a4c5..91672358f 100644
--- a/tests/core/test_push.py
+++ b/tests/core/test_push.py
@@ -46,6 +46,10 @@ def get_signature_headers(data, key):
     },
     'event_id': 'urn:event/1',
     'coverage_id': 'urn:coverage/1',
+    'subject': [
+        {"name": "a", "code": "a", "scheme": "a"},
+        {"name": "b", "code": "b", "scheme": "b"}
+    ],
 }
 
 
@@ -683,3 +687,11 @@ def test_push_event_coverage_info(client, app):
     parsed = get_entity_or_404(item['guid'], 'items')
     assert parsed['event_id'] == 'urn:event/1'
     assert parsed['coverage_id'] == 'urn:coverage/1'
+
+
+def test_push_wire_subject_whitelist(client, app):
+    app.config["WIRE_SUBJECT_SCHEME_WHITELIST"] = ['b']
+    client.post('/push', data=json.dumps(item), content_type='application/json')
+    parsed = get_entity_or_404(item['guid'], 'items')
+    assert 1 == len(parsed['subject'])
+    assert 'b' == parsed['subject'][0]['name']