fix usage report pagination (#4183)

* early work in progress

* rename utility script

* move actual data seeding to a shareable function

* add test

* make the test pass with the fix

* fix comment

---------

Co-authored-by: Richard Kuo (Danswer) <rkuo@onyx.app>

backend/tests/integration/tests/query_history/test_usage_reports.py

@@ -0,0 +1,46 @@
+from datetime import datetime
+from datetime import timedelta
+from datetime import timezone
+
+from ee.onyx.db.usage_export import get_all_empty_chat_message_entries
+from onyx.db.engine import get_session_with_current_tenant
+from onyx.db.seeding.chat_history_seeding import seed_chat_history
+
+
+def test_usage_reports(reset: None) -> None:
+    EXPECTED_SESSIONS = 2048
+    MESSAGES_PER_SESSION = 4
+    EXPECTED_MESSAGES = EXPECTED_SESSIONS * MESSAGES_PER_SESSION
+
+    seed_chat_history(EXPECTED_SESSIONS, MESSAGES_PER_SESSION, 90)
+
+    with get_session_with_current_tenant() as db_session:
+        # count of all entries should be exact
+        period = (
+            datetime.fromtimestamp(0, tz=timezone.utc),
+            datetime.now(tz=timezone.utc),
+        )
+
+        count = 0
+        for entry_batch in get_all_empty_chat_message_entries(db_session, period):
+            for entry in entry_batch:
+                count += 1
+
+        assert count == EXPECTED_MESSAGES
+
+        # count in a one month time range should be within a certain range statistically
+        # this can be improved if we seed the chat history data deterministically
+        period = (
+            datetime.now(tz=timezone.utc) - timedelta(days=30),
+            datetime.now(tz=timezone.utc),
+        )
+
+        count = 0
+        for entry_batch in get_all_empty_chat_message_entries(db_session, period):
+            for entry in entry_batch:
+                count += 1
+
+        lower = EXPECTED_MESSAGES // 3 - (EXPECTED_MESSAGES // (3 * 3))
+        upper = EXPECTED_MESSAGES // 3 + (EXPECTED_MESSAGES // (3 * 3))
+        assert count > lower
+        assert count < upper