Fix problem ID validation.

app.py

@@ -156,7 +156,10 @@ with blocks:
                     if len(submission_path) == 0:
                         return styled_error("Please upload JSONL submission file.")
 
-                    if not is_submission_file_valid(submission_path):
+                    if not is_submission_file_valid(
+                        submission_path,
+                        is_warmup_dataset=(SPLIT == "warmup"),
+                    ):
                         return styled_error("Failed to read JSONL submission file. Please try again later.")
 
                     # Validating all user-supplied arguments.
@@ -185,6 +188,7 @@ with blocks:
                     sys_type,
                     submission_path,
                     ensure_all_present=ENSURE_ALL_PRESENT,
+                    is_warmup_dataset=(SPLIT == "warmup"),
                 )
 
             submit_button.click(

src/submission/submit.py

@@ -47,6 +47,7 @@ def add_new_solutions(
     sys_type: str,
     submission_path: str,
     ensure_all_present: bool = False,
+    is_warmup_dataset: bool,
 ):
     logger.info(
         f"Adding new submission! {system_name=}, {org=}, {sys_type=} and {submission_path=}",
@@ -55,7 +56,7 @@ def add_new_solutions(
     # Double-checking.
     for val in [system_name, org, sys_type]:
         assert is_valid(val)
-    assert is_submission_file_valid(submission_path)
+    assert is_submission_file_valid(submission_path, is_warmup_dataset=is_warmup_dataset)
 
     sys_type = ModelType.from_str(sys_type).name
 

src/validation/validate.py

@@ -4,6 +4,7 @@ import string
 
 from src.logger import get_logger
 
+WARMUP_DATASET_SIZE = 100
 DATASET_SIZE = 120
 
 MIN_INPUT_LENGTH = 2
@@ -46,10 +47,14 @@ def is_valid(
     return True
 
 
-def is_submission_file_valid(submission_path: str) -> bool:
+def is_submission_file_valid(
+    submission_path: str,
+    is_warmup_dataset: bool,
+) -> bool:
     """
     @brief Checks whether the given submission file is valid.
     @param submission_path The path to the submission file.
+    @param is_warmup_dataset Whether we are working on the regular or the warmup dataset.
     @return True iff the file is within the size constraints, a JSONL, and every line is no longer than
             the fixed maximum bound.
     """
@@ -93,9 +98,14 @@ def is_submission_file_valid(submission_path: str) -> bool:
                 logger.warning("Could not convert problem ID to int")
                 return False
 
-            if problem_id < 0 or problem_id >= DATASET_SIZE:
-                logger.warning(f"Problem ID {problem_id} is beyond allowed bounds")
-                return False
+            if is_warmup_dataset:
+                if problem_id < DATASET_SIZE or problem_id >= DATASET_SIZE + WARMUP_DATASET_SIZE:
+                    logger.warning(f"Problem ID {problem_id} is beyond allowed bounds")
+                    return False
+            else:
+                if problem_id < 0 or problem_id >= DATASET_SIZE:
+                    logger.warning(f"Problem ID {problem_id} is beyond allowed bounds")
+                    return False
 
             if problem_id in seen_ids:
                 logger.warning(f"Got duplicate submission -- ID {problem_id} appears twice")