chore: Add tests for relative date parsing

posthog/models/filters/test/__snapshots__/test_filter.ambr

@@ -353,7 +353,7 @@
   INNER JOIN "posthog_persondistinctid" ON ("posthog_person"."id" = "posthog_persondistinctid"."person_id")
   WHERE ("posthog_persondistinctid"."distinct_id" = 'example_id'
          AND "posthog_person"."team_id" = 2
-         AND ("posthog_person"."properties" -> 'created_at') > '"2021-02-06T10:00:00+00:00"')
+         AND "posthog_person"."id" = -1)
   LIMIT 1
   '
 ---

posthog/models/filters/test/test_filter.py

@@ -854,9 +854,8 @@ def test_person_relative_date_parsing_with_invalid_date(self):
                 .filter(properties_to_Q(filter.property_groups.flat))
                 .exists()
             )
-            # matches '2m'
-            # TODO: Should this not match instead?
-            self.assertTrue(matched_person)
+            # needs an exact match
+            self.assertFalse(matched_person)
 
         filter = Filter(
             data={

posthog/queries/base.py

@@ -435,11 +435,16 @@ def is_truthy_or_falsy_property_value(value: Any) -> bool:
 
 
 def relative_date_parse_for_feature_flag_matching(value: str) -> Optional[datetime.datetime]:
-    regex = r"(?P<number>[0-9]+)(?P<interval>[a-z])"
+    regex = r"^(?P<number>[0-9]+)(?P<interval>[a-z])$"
     match = re.search(regex, value)
     parsed_dt = datetime.datetime.now(tz=ZoneInfo("UTC"))
     if match:
         number = int(match.group("number"))
+
+        if number >= 10_000:
+            # Guard against overflow, disallow numbers greater than 10_000
+            return None
+
         interval = match.group("interval")
         if interval == "h":
             parsed_dt = parsed_dt - relativedelta(hours=number)

posthog/queries/test/test_base.py

@@ -1,5 +1,6 @@
 import datetime
 import re
+import unittest
 from unittest.mock import patch
 
 from dateutil import parser, tz
@@ -10,7 +11,7 @@
 
 from posthog.models.filters.path_filter import PathFilter
 from posthog.models.property.property import Property
-from posthog.queries.base import match_property, sanitize_property_key
+from posthog.queries.base import match_property, relative_date_parse_for_feature_flag_matching, sanitize_property_key
 from posthog.test.base import APIBaseTest
 
 
@@ -406,3 +407,151 @@ def test_sanitize_keys(key, expected):
     sanitized_key = sanitize_property_key(key)
 
     assert sanitized_key == expected
+
+
+class TestRelativeDateParsing(unittest.TestCase):
+    def test_invalid_input(self):
+        with freeze_time("2020-01-01T12:01:20.1340Z"):
+            assert relative_date_parse_for_feature_flag_matching("1") is None
+            assert relative_date_parse_for_feature_flag_matching("1x") is None
+            assert relative_date_parse_for_feature_flag_matching("1.2y") is None
+            assert relative_date_parse_for_feature_flag_matching("1z") is None
+            assert relative_date_parse_for_feature_flag_matching("1s") is None
+            assert relative_date_parse_for_feature_flag_matching("123344000.134m") is None
+            assert relative_date_parse_for_feature_flag_matching("bazinga") is None
+            assert relative_date_parse_for_feature_flag_matching("000bello") is None
+            assert relative_date_parse_for_feature_flag_matching("000hello") is None
+
+            assert relative_date_parse_for_feature_flag_matching("000h") is not None
+            assert relative_date_parse_for_feature_flag_matching("1000h") is not None
+
+    def test_overflow(self):
+        assert relative_date_parse_for_feature_flag_matching("1000000h") is None
+        assert relative_date_parse_for_feature_flag_matching("100000000000000000y") is None
+
+    def test_hour_parsing(self):
+        with freeze_time("2020-01-01T12:01:20.1340Z"):
+            assert relative_date_parse_for_feature_flag_matching("1h") == datetime.datetime(
+                2020, 1, 1, 11, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("2h") == datetime.datetime(
+                2020, 1, 1, 10, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("24h") == datetime.datetime(
+                2019, 12, 31, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("30h") == datetime.datetime(
+                2019, 12, 31, 6, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("48h") == datetime.datetime(
+                2019, 12, 30, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+
+            assert relative_date_parse_for_feature_flag_matching(
+                "24h"
+            ) == relative_date_parse_for_feature_flag_matching("1d")
+            assert relative_date_parse_for_feature_flag_matching(
+                "48h"
+            ) == relative_date_parse_for_feature_flag_matching("2d")
+
+    def test_day_parsing(self):
+        with freeze_time("2020-01-01T12:01:20.1340Z"):
+            assert relative_date_parse_for_feature_flag_matching("1d") == datetime.datetime(
+                2019, 12, 31, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("2d") == datetime.datetime(
+                2019, 12, 30, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("7d") == datetime.datetime(
+                2019, 12, 25, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("14d") == datetime.datetime(
+                2019, 12, 18, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("30d") == datetime.datetime(
+                2019, 12, 2, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+
+            assert relative_date_parse_for_feature_flag_matching("7d") == relative_date_parse_for_feature_flag_matching(
+                "1w"
+            )
+
+    def test_week_parsing(self):
+        with freeze_time("2020-01-01T12:01:20.1340Z"):
+            assert relative_date_parse_for_feature_flag_matching("1w") == datetime.datetime(
+                2019, 12, 25, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("2w") == datetime.datetime(
+                2019, 12, 18, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("4w") == datetime.datetime(
+                2019, 12, 4, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("8w") == datetime.datetime(
+                2019, 11, 6, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+
+            assert relative_date_parse_for_feature_flag_matching("1m") == datetime.datetime(
+                2019, 12, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("4w") != relative_date_parse_for_feature_flag_matching(
+                "1m"
+            )
+
+    def test_month_parsing(self):
+        with freeze_time("2020-01-01T12:01:20.1340Z"):
+            assert relative_date_parse_for_feature_flag_matching("1m") == datetime.datetime(
+                2019, 12, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("2m") == datetime.datetime(
+                2019, 11, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("4m") == datetime.datetime(
+                2019, 9, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("8m") == datetime.datetime(
+                2019, 5, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+
+            assert relative_date_parse_for_feature_flag_matching("1y") == datetime.datetime(
+                2019, 1, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching(
+                "12m"
+            ) == relative_date_parse_for_feature_flag_matching("1y")
+
+        with freeze_time("2020-04-03T00:00:00"):
+            assert relative_date_parse_for_feature_flag_matching("1m") == datetime.datetime(
+                2020, 3, 3, 0, 0, 0, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("2m") == datetime.datetime(
+                2020, 2, 3, 0, 0, 0, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("4m") == datetime.datetime(
+                2019, 12, 3, 0, 0, 0, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("8m") == datetime.datetime(
+                2019, 8, 3, 0, 0, 0, tzinfo=tz.gettz("UTC")
+            )
+
+            assert relative_date_parse_for_feature_flag_matching("1y") == datetime.datetime(
+                2019, 4, 3, 0, 0, 0, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching(
+                "12m"
+            ) == relative_date_parse_for_feature_flag_matching("1y")
+
+    def test_year_parsing(self):
+        with freeze_time("2020-01-01T12:01:20.1340Z"):
+            assert relative_date_parse_for_feature_flag_matching("1y") == datetime.datetime(
+                2019, 1, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("2y") == datetime.datetime(
+                2018, 1, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("4y") == datetime.datetime(
+                2016, 1, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )
+            assert relative_date_parse_for_feature_flag_matching("8y") == datetime.datetime(
+                2012, 1, 1, 12, 1, 20, 134000, tzinfo=tz.gettz("UTC")
+            )