enforce rate limits

src/submission/submit.py

@@ -46,32 +46,30 @@ def add_new_solutions(
 
     logger.info(f"Found {len(submitted_ids)} submissions")
 
-    user_last_submission_date = None
-    for sid in submitted_ids:
-        # Extract user ID (last part)
-        past_user_id = sid.rsplit("_", 1)[-1]
-        # Extract timestamp string (first two parts)
-        ts_str = "_".join(sid.split("_", 2)[:2])
-
-        logger.info(f"Comparing past user: {past_user_id} with current user ID: {user_id}")
-
-        ts = datetime.strptime(ts_str, "%Y%m%d_%H%M%S").replace(tzinfo=timezone.utc)
-        if past_user_id == user_id:
-            if user_last_submission_date is None:
-                user_last_submission_date = ts
-            else:
-                user_last_submission_date = max(user_last_submission_date, ts)
-
-    if user_last_submission_date is not None:
-        now = datetime.now(timezone.utc)
-        elapsed = now - user_last_submission_date
-        if elapsed < timedelta(hours=MIN_WAIT_TIME_PER_USER_HRS):
-            remaining_hrs = MIN_WAIT_TIME_PER_USER_HRS - elapsed.total_seconds() / 3600
-            logger.info(f"{username} must wait {remaining_hrs:.2f} more hours.")
-            return styled_error(
-                f"You must wait {MIN_WAIT_TIME_PER_USER_HRS} hours between submissions. "
-                f"Remaining wait time: {remaining_hrs:.2f} hours"
-            )
+    sub_df = pd.DataFrame.from_dict(
+        {
+            "submission_id": submitted_ids,
+            "user_id": map(submission_id_to_user_id, submitted_ids),
+            "timestamp": map(submission_id_to_timestamp, submitted_ids),
+        }
+    )
+
+    cutoff = datetime.now(timezone.utc) - timedelta(hours=RATE_LIMIT_WINDOW_HRS)
+    user_last_sub_date = sub_df[sub_df.user_id == user_id].timestamp.max()
+
+    if pd.notna(user_last_sub_date) and user_last_sub_date > cutoff:
+        remaining_hrs = MIN_WAIT_TIME_PER_USER_HRS - (user_last_sub_date - cutoff).total_seconds() / 3600
+        logger.info(f"{username} must wait {remaining_hrs:.2f} more hours.")
+        return styled_error(
+            f"You must wait {MIN_WAIT_TIME_PER_USER_HRS} hours between submissions. "
+            f"Remaining wait time: {remaining_hrs:.2f} hours"
+        )
+
+    if len(sub_df.timestamp > cutoff) >= MAX_SUBMISSIONS_PER_WINDOW:
+        logger.info(
+            f"Too many submissions in the last {RATE_LIMIT_WINDOW_HRS} hours: {len(sub_df.timestamp > cutoff)}."
+        )
+        return styled_error("The leaderboard has reached its submission capacity for now. Please try again later.")
 
     logger.info(
         f"Adding new submission: {system_name=}, {org=}, {sys_type=} and {submission_path=}",
@@ -172,3 +170,18 @@ def _validate_all_submissions_present(
         raise ValueError(f"Mismatched problem IDs: {len(missing)} missing, {len(unknown)} unknown")
     if len(pd_ds) > len(lbdb.code_problem_ids):
         return ValueError("Duplicate problem IDs exist in uploaded file")
+
+
+def submission_id_to_user_id(submission_id: str) -> str:
+    """
+    Extracts the user ID from the submission ID: "YYYYMMDD_HHMMSS_username_userid"
+    """
+    return submission_id.rsplit("_", 1)[-1]
+
+
+def submission_id_to_timestamp(submission_id: str) -> datetime:
+    """
+    Extracts the timestamp from the submission ID: "YYYYMMDD_HHMMSS_username_userid"
+    """
+    ts_str = "_".join(submission_id.split("_", 2)[:2])
+    return datetime.strptime(ts_str, "%Y%m%d_%H%M%S").replace(tzinfo=timezone.utc)