From 5f8e3e558b9cc55ad90eb290fb6a829764a974b7 Mon Sep 17 00:00:00 2001
From: db0 <mail@dbzer0.com>
Date: Mon, 26 Jun 2023 00:32:41 +0200
Subject: [PATCH] feat: report sus per monthly active user

---
 fediseer/apis/models/v1.py | 10 ++--------
 fediseer/apis/v1/base.py   |  3 ++-
 fediseer/observer.py       |  7 ++++++-
 3 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/fediseer/apis/models/v1.py b/fediseer/apis/models/v1.py
index 64e8fe4..1fe369e 100644
--- a/fediseer/apis/models/v1.py
+++ b/fediseer/apis/models/v1.py
@@ -16,8 +16,8 @@ class Models:
             'total_users': fields.Integer(description="The total amount of users registered in that instance"),
             'active_users_monthly': fields.Integer(description="The amount of active users monthly."),
             'signup': fields.Boolean(default=False,description="True when subscriptions are open, else False"),
-            'activity_suspicion': fields.Float(description="Local Comments+Posts per User. Higher is worse"),
-            'activity_suspicion': fields.Float(description="Local Comments+Posts per User. Higher is worse"),
+            'activity_suspicion': fields.Float(description="Local Comments+Posts per user. Higher is worse"),
+            'active_users_suspicion': fields.Float(description="Monthly active users per user. Higher is worse"),
         })
         self.response_model_model_Suspicions_get = api.model('SuspiciousInstances', {
             'instances': fields.List(fields.Nested(self.response_model_suspicious_instances)),
@@ -40,9 +40,3 @@ class Models:
             'domains': fields.List(fields.String(description="The instance domains as a list.")),
             'csv': fields.String(description="The instance domains as a csv."),
         })
-        self.response_model_model_Suspicions_get = api.model('SuspiciousInstances', {
-            'self': fields.Nested(self.response_model_instances),
-            'guaranteed': fields.List(fields.Nested(self.response_model_instances)),
-            'endorsed': fields.List(fields.String(description="The suspicious domains as a list.")),
-            'endorsing': fields.String(description="The suspicious domains as a csv."),
-        })
diff --git a/fediseer/apis/v1/base.py b/fediseer/apis/v1/base.py
index 3f4abde..ccb4e2c 100644
--- a/fediseer/apis/v1/base.py
+++ b/fediseer/apis/v1/base.py
@@ -34,6 +34,7 @@ class Suspicions(Resource):
     get_parser = reqparse.RequestParser()
     get_parser.add_argument("Client-Agent", default="unknown:0:unknown", type=str, required=False, help="The client name and version.", location="headers")
     get_parser.add_argument("activity_suspicion", required=False, default=20, type=int, help="How many users per local post+comment to consider suspicious", location="args")
+    get_parser.add_argument("active_suspicion", required=False, default=500, type=int, help="How many users per active users to consider suspicious", location="args")
     get_parser.add_argument("csv", required=False, type=bool, help="Set to true to return just the domains as a csv. Mutually exclusive with domains", location="args")
     get_parser.add_argument("domains", required=False, type=bool, help="Set to true to return just the domains as a list. Mutually exclusive with csv", location="args")
 
@@ -45,7 +46,7 @@ class Suspicions(Resource):
         '''A List with the details of all suspicious instances
         '''
         self.args = self.get_parser.parse_args()
-        sus_instances = retrieve_suspicious_instances(self.args.activity_suspicion)
+        sus_instances = retrieve_suspicious_instances(self.args.activity_suspicion, self.args.active_suspicion)
         if self.args.csv:
             return {"csv": ",".join([instance["domain"] for instance in sus_instances])},200
         if self.args.domains:
diff --git a/fediseer/observer.py b/fediseer/observer.py
index aae71c2..24dc6af 100644
--- a/fediseer/observer.py
+++ b/fediseer/observer.py
@@ -69,7 +69,11 @@ def retrieve_suspicious_instances(activity_suspicion = 20, active_suspicious = 5
                 # print(node)
 
             # check active users (monthly is a lot lower than total users)
-            if node["total_users"] / node["active_users_monthly"] > active_suspicious:
+            local_active_monthly_users = node["active_users_monthly"]
+            # Avoid division by 0
+            if local_active_monthly_users == 0:
+                local_active_monthly_users= 0.5
+            if node["total_users"] / local_active_monthly_users > active_suspicious:
                 is_bad = True
                 # print(node)
 
@@ -83,6 +87,7 @@ def retrieve_suspicious_instances(activity_suspicion = 20, active_suspicious = 5
                     "active_users_monthly": node["active_users_monthly"],
                     "signup": node["signup"],
                     "activity_suspicion": node["total_users"] / local_activity,
+                    "active_users_suspicion": node["total_users"] / local_active_monthly_users,
                 }
                 bad_nodes.append(bad_node)
         return bad_nodes