Fix stale DB redirects appearing as new assessments in quality logs

wp1/logic/page.py

@@ -4,7 +4,7 @@
 import requests
 
 import wp1.logic.util as logic_util
-from wp1.constants import GLOBAL_TIMESTAMP, TS_FORMAT
+from wp1.constants import GLOBAL_TIMESTAMP, TS_FORMAT, TS_FORMAT_WP10
 from wp1.logic import log as logic_log
 from wp1.logic import move as logic_move
 from wp1.logic.api import page as api_page
@@ -87,25 +87,29 @@ def update_page_moved(
 def _get_redirects_from_db(wikidb, namespace, title, timestamp_dt):
     wiki_db_title = title.decode("utf-8").replace(" ", "_")
     wikidb.ping()
-    args_dict = {"title": wiki_db_title, "namespace": namespace}
+    args_dict = {
+        "title": wiki_db_title,
+        "namespace": namespace,
+        "timestamp": timestamp_dt.strftime(TS_FORMAT_WP10),
+    }
     with wikidb.cursor() as cursor:
         cursor.execute(
             """
         SELECT rd_namespace, rd_title, page_touched FROM page
         JOIN redirect ON page_id = rd_from AND
              page_title = %(title)s AND page_namespace = %(namespace)s
+        WHERE page_touched > %(timestamp)s
     """,
             args_dict,
         )
         row = cursor.fetchone()
         if row:
-            timestamp_dt = datetime.strptime(
-                row["page_touched"].decode("utf-8"), "%Y%m%d%H%M%S"
-            )
             return {
                 "dest_ns": row["rd_namespace"],
                 "dest_title": row["rd_title"],
-                "timestamp_dt": timestamp_dt,
+                "timestamp_dt": datetime.strptime(
+                    row["page_touched"].decode("utf-8"), TS_FORMAT_WP10
+                ),
             }
         return None
 

wp1/logic/page_test.py

@@ -235,6 +235,55 @@ def test_get_single_move_too_old_from_api(self, patched_site):
         )
         self.assertIsNone(move_data)
 
+    def test_get_redirect_from_db_stale(self):
+        # Insert a redirect row with a 2016 timestamp which is older than last run
+        with self.wikidb.cursor() as cursor:
+            cursor.execute(
+                """
+                INSERT INTO page (page_id, page_namespace, page_title, page_touched)
+                VALUES (200, 0, 'Some_Moved_Article', '20160315142300')
+            """
+            )
+            cursor.execute(
+                """
+                INSERT INTO redirect (rd_from, rd_namespace, rd_title)
+                VALUES (200, 0, 'Destination_Article')
+            """
+            )
+        self.wikidb.commit()
+
+        # Last run was 2022, redirect is from 2016 so should be discarded
+        move_data = logic_page._get_redirects_from_db(
+            self.wikidb, 0, b"Some Moved Article", datetime(2022, 1, 1)
+        )
+        self.assertIsNone(move_data)
+
+    def test_get_redirect_from_db_fresh(self):
+        # Insert a redirect row with a 2023 timestamp which newer than last run
+        with self.wikidb.cursor() as cursor:
+            cursor.execute(
+                """
+                INSERT INTO page (page_id, page_namespace, page_title, page_touched)
+                VALUES (201, 0, 'Some_Moved_Article', '20230315142300')
+            """
+            )
+            cursor.execute(
+                """
+                INSERT INTO redirect (rd_from, rd_namespace, rd_title)
+                VALUES (201, 0, 'Destination_Article')
+            """
+            )
+        self.wikidb.commit()
+
+        # Last run was 2022, redirect is from 2023 so should return
+        move_data = logic_page._get_redirects_from_db(
+            self.wikidb, 0, b"Some Moved Article", datetime(2022, 1, 1)
+        )
+        self.assertIsNotNone(move_data)
+        self.assertEqual(0, move_data["dest_ns"])
+        self.assertEqual(b"Destination_Article", move_data["dest_title"])
+        self.assertEqual(datetime(2023, 3, 15, 14, 23, 0), move_data["timestamp_dt"])
+
 
 class LogicPageMoveDbTest(BaseWpOneDbTest):