perf: Use cityhash to filter events (#25409)

PostHog · Oct 5, 2024 · 1f100bc · 1f100bc
1 parent dc9752d
commit 1f100bc
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 3 deletions.
diff --git a/posthog/hogql_queries/events_query_runner.py b/posthog/hogql_queries/events_query_runner.py
@@ -115,8 +115,15 @@ def to_query(self) -> ast.SelectQuery:
                         ).first()
                         where_exprs.append(
                             parse_expr(
-                                "distinct_id in {list}",
-                                {"list": ast.Constant(value=get_distinct_ids_for_subquery(person, self.team))},
+                                "cityHash64(distinct_id) in {list}",  # Because the events table is partitioned by cityHash64(distinct_ids), using cityhash for the comparison is much quicker,
+                                {
+                                    "list": ast.Constant(
+                                        value=[
+                                            ast.Call(name="cityHash64", args=[ast.Constant(value=id)])
+                                            for id in get_distinct_ids_for_subquery(person, self.team)
+                                        ]
+                                    )
+                                },
                                 timings=self.timings,
                             )
                         )

diff --git a/posthog/hogql_queries/test/test_events_query_runner.py b/posthog/hogql_queries/test/test_events_query_runner.py
@@ -131,7 +131,7 @@ def test_person_id_expands_to_distinct_ids(self):
         query_ast = EventsQueryRunner(query=query, team=self.team).to_query()
         where_expr = cast(ast.CompareOperation, cast(ast.And, query_ast.where).exprs[0])
         right_expr = cast(ast.Constant, where_expr.right)
-        self.assertEqual(right_expr.value, ["id1", "id2"])
+        self.assertEqual([x.args[0].value for x in right_expr.value], ["id1", "id2"])
 
         # another team
         another_team = Team.objects.create(organization=Organization.objects.create())