From 0be44910cfb177aaec2fc3786f10f3fcfdd37423 Mon Sep 17 00:00:00 2001
From: db0 <mail@dbzer0.com>
Date: Wed, 21 Jun 2023 01:57:30 +0200
Subject: [PATCH] activity_suspicion

---
 overseer/apis/models/v1.py |  3 ++-
 overseer/apis/v1/base.py   |  4 ++--
 overseer/observer.py       | 14 ++++++++------
 3 files changed, 12 insertions(+), 9 deletions(-)

diff --git a/overseer/apis/models/v1.py b/overseer/apis/models/v1.py
index 6c9e643..8aa5520 100644
--- a/overseer/apis/models/v1.py
+++ b/overseer/apis/models/v1.py
@@ -9,10 +9,11 @@ class Models:
             'domain': fields.String(description="The instance domain"),
             'uptime_alltime': fields.Float(description="The instance uptime pct. 100% and thousand of users is unlikely"),
             'local_posts': fields.Integer(description="The amount of local posts in that instance"),
+            'comment_counts': fields.Integer(description="The amount of comments in that instance"),
             '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"),
-            'user_post_ratio': fields.Float(description="Users to Post Ratio"),
+            'activity_suspicion': fields.Float(description="Local Comments+Posts per User. Higher is worse"),
         })
         self.response_model_model_SusInstances_get = api.model('SuspiciousInstancesDomainList', {
             'instances': fields.List(fields.Nested(self.response_model_suspicious_instances)),
diff --git a/overseer/apis/v1/base.py b/overseer/apis/v1/base.py
index b00c2f4..84a3354 100644
--- a/overseer/apis/v1/base.py
+++ b/overseer/apis/v1/base.py
@@ -19,7 +19,7 @@ def get_request_path():
 class SusInstances(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("user_to_post_ratio", required=False, default=20, type=int, help="The amount of local users / amount of local posts to consider suspicious", location="args")
+    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("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")
 
@@ -31,7 +31,7 @@ class SusInstances(Resource):
         '''A List with the details of all suspicious instances
         '''
         self.args = self.get_parser.parse_args()
-        sus_instances = retrieve_suspicious_instances(self.args.user_to_post_ratio)
+        sus_instances = retrieve_suspicious_instances(self.args.activity_suspicion)
         if self.args.csv:
             return {"csv": ",".join([instance["domain"] for instance in sus_instances])},200
         if self.args.domains:
diff --git a/overseer/observer.py b/overseer/observer.py
index dcc9540..164bf7a 100644
--- a/overseer/observer.py
+++ b/overseer/observer.py
@@ -2,7 +2,7 @@ import requests
 from loguru import logger
 
 
-def retrieve_suspicious_instances(users_to_posts_ratio = 20):
+def retrieve_suspicious_instances(activity_suspicion = 20):
     # GraphQL query
     query = '''
     {
@@ -19,6 +19,7 @@ def retrieve_suspicious_instances(users_to_posts_ratio = 20):
         active_users_halfyear
         signup
         local_posts
+        comment_counts
     }
     }
     '''
@@ -58,12 +59,12 @@ def retrieve_suspicious_instances(users_to_posts_ratio = 20):
         bad_nodes = []
         for node in data["data"]["nodes"]:
             is_bad = False
-            local_posts = node["local_posts"]
+            local_activity = node["local_posts"] + node["comment_counts"]
             if node["total_users"] < 300:
                 continue
-            if local_posts == 0:
-                local_posts= 1
-            if node["total_users"] / local_posts > users_to_posts_ratio:
+            if local_activity == 0:
+                local_activity= 1
+            if node["total_users"] / local_activity > activity_suspicion:
                 is_bad = True
                 # print(node)
             if is_bad:
@@ -71,10 +72,11 @@ def retrieve_suspicious_instances(users_to_posts_ratio = 20):
                     "domain": node["domain"],
                     "uptime_alltime": node["uptime_alltime"],
                     "local_posts": node["local_posts"],
+                    "comment_counts": node["comment_counts"],
                     "total_users": node["total_users"],
                     "active_users_monthly": node["active_users_monthly"],
                     "signup": node["signup"],
-                    "user_post_ratio": node["total_users"] / local_posts,
+                    "activity_suspicion": node["total_users"] / local_activity,
                 }
                 bad_nodes.append(bad_node)
         return bad_nodes