Implement redaction for search results

core/lib/opensearch.py

@@ -1,6 +1,10 @@
+import pprint
+
 from django.conf import settings
 from opensearchpy import OpenSearch
 
+pp = pprint.PrettyPrinter(indent=4)
+
 
 def initialise_opensearch():
     auth = (settings.OPENSEARCH_USERNAME, settings.OPENSEARCH_PASSWORD)
@@ -57,9 +61,33 @@ def construct_query(query, fields, results):
     return query
 
 
+def filter_blacklisted(response):
+    pp.pprint(response["hits"]["hits"])
+    print("LEN", len(response["hits"]["hits"]))
+    response["redacted"] = 0
+    # For every hit from ES
+    for item in list(response["hits"]["hits"]):
+        # For every blacklisted type
+        for blacklisted_type in settings.OPENSEARCH_BLACKLISTED.keys():
+            # Check this field we are matching exists
+            if blacklisted_type in item["_source"].keys():
+                content = item["_source"][blacklisted_type]
+                # For every item in the blacklisted array for the type
+                for blacklisted_item in settings.OPENSEARCH_BLACKLISTED[
+                    blacklisted_type
+                ]:
+                    if blacklisted_item in str(content):
+                        # Remove the item
+                        if item in response["hits"]["hits"]:
+                            response["hits"]["hits"].remove(item)
+                            # Let the UI know something was redacted
+                            response["redacted"] += 1
+
+
 def run_main_query(client, query, fields=None, results=None):
     search_query = construct_query(query, fields, results)
     # fmt: off
     response = client.search(body=search_query,
                              index=settings.OPENSEARCH_INDEX_MAIN)
+    filter_blacklisted(response)
     return response