Added logging to validation logic.

src/validation/validate.py

@@ -2,6 +2,8 @@ import json
 import os
 import string
 
+from src.logger import get_logger
+
 DATASET_SIZE = 120
 
 MIN_INPUT_LENGTH = 2
@@ -12,6 +14,8 @@ MAX_SUBMISSION_SIZE = 1024 * 1024 * 120  # 120 MB.
 MAX_SINGLE_SUBMISSION_SIZE = 1024 * 1024  # 1MB.
 MAX_SUBMISSION_LINES = DATASET_SIZE + 1  # Allow empty line.
 
+logger = get_logger()
+
 
 def is_valid(
     s: str,
@@ -51,10 +55,12 @@ def is_submission_file_valid(submission_path: str) -> bool:
     """
 
     if not os.path.exists(submission_path):
+        logger.warning(f"Could not find submission file {submission_path=}")
         return False
 
     submission_size = os.stat(submission_path).st_size
     if submission_size < MIN_SUBMISSION_SIZE or submission_size > MAX_SUBMISSION_SIZE:
+        logger.warning(f"Submission size was {submission_size}, exceeding [{MIN_SUBMISSION_SIZE, MAX_SUBMISSION_SIZE}]")
         return False
 
     with open(submission_path, "r") as f:
@@ -65,24 +71,34 @@ def is_submission_file_valid(submission_path: str) -> bool:
         while len(line := f.readline(MAX_SINGLE_SUBMISSION_SIZE)) > 0:
             n_lines += 1
             if n_lines > MAX_SUBMISSION_LINES:
+                logger.warning(f"Got submission with more than {MAX_SUBMISSION_LINES} lines")
                 return False
 
             if not line.startswith("{") or not line.endswith("}"):
+                logger.warning("Submission has line that does not appear to be a JSONL")
                 return False
 
             d = json.loads(line)
             if set(d.keys()) != set(["problem_id", "solution"]):
+                logger.warning("Found unexpected keys")
                 return False
 
             if not ((type(d["problem_id"]) is str or type(d["problem_id"]) is int) and type(d["solution"] is str)):
+                logger.warning("Found unexpected types")
                 return False
-            if not d["problem_id"].isdigit():
+
+            try:
+                problem_id = int(d["problem_id"])
+            except Exception:
+                logger.warning("Could not convert problem ID to int")
                 return False
-            problem_id = int(d["problem_id"])
+
             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")
                 return False  # Duplicate submission.
             seen_ids.add(problem_id)