From 0cbd2d8a6fc6e1f2213c6276fd105bf9a7bef680 Mon Sep 17 00:00:00 2001
From: Mark Veidemanis <m@zm.is>
Date: Thu, 2 Feb 2023 18:58:40 +0000
Subject: [PATCH] Check hash of message when determining if it is a new match

---
 core/lib/rules.py | 21 +++++++++++++++++----
 core/models.py    |  3 ++-
 2 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/core/lib/rules.py b/core/lib/rules.py
index 16bee98..7ad0300 100644
--- a/core/lib/rules.py
+++ b/core/lib/rules.py
@@ -10,6 +10,7 @@ except ImportError:
 
 import orjson
 from asgiref.sync import async_to_sync
+from siphashc import siphash
 
 from core.lib.notify import sendmsg
 from core.lib.parsing import parse_index, parse_source
@@ -205,8 +206,13 @@ class NotificationRuleData(object):
         Store a match result.
         Accepts None for the index to set all indices.
         :param index: the index to store the match for, can be None
-        :param match: True or False, indicating if the rule matched
+        :param match: the object that matched
         """
+        if match is not False:
+            # Dump match to JSON while sorting the keys
+            match_normalised = orjson.dumps(match, option=orjson.OPT_SORT_KEYS)
+            match = siphash(self.db.hash_key, match_normalised)
+
         if self.object.match is None:
             self.object.match = {}
         if not isinstance(self.object.match, dict):
@@ -220,7 +226,7 @@ class NotificationRuleData(object):
         self.object.save()
         log.debug(f"Stored match: {index} - {match}")
 
-    def get_match(self, index=None):
+    def get_match(self, index=None, match=None):
         """
         Get a match result for an index.
         If the index is None, it will return True if any index has a match.
@@ -237,6 +243,13 @@ class NotificationRuleData(object):
             # Check if we have any matches on all indices
             return any(self.object.match.values())
 
+        # Check if it's the same hash
+        if match is not None:
+            match_normalised = orjson.dumps(match, option=orjson.OPT_SORT_KEYS)
+            match = siphash(self.db.hash_key, match_normalised)
+            hash_matches = self.object.match.get(index) == match
+            return hash_matches
+
         return self.object.match.get(index)
 
     def format_aggs(self, aggs):
@@ -268,13 +281,13 @@ class NotificationRuleData(object):
         :param message: the message object that matched
         :param aggs: the aggregations that matched
         """
-        current_match = self.get_match(index)
+        current_match = self.get_match(index, message)
         log.debug(f"Rule matched: {index} - current match: {current_match}")
         if current_match is False:
             # Matched now, but not before
             meta["matched"] = self.format_aggs(meta["aggs"])
             rule_notify(self.object, index, message, meta)
-            self.store_match(index, True)
+            self.store_match(index, message)
 
     def rule_no_match(self, index=None):
         """
diff --git a/core/models.py b/core/models.py
index 4bca3f0..a878437 100644
--- a/core/models.py
+++ b/core/models.py
@@ -203,7 +203,8 @@ class NotificationRule(models.Model):
         Get the total number of matches for this rule.
         """
         if isinstance(self.match, dict):
-            return f"{sum(list(self.match.values()))}/{len(self.match)}"
+            truthy_values = [x for x in self.match.values() if x is not False]
+            return f"{len(truthy_values)}/{len(self.match)}"
 
     def get_notification_settings(self, check=True):
         """