wip

src/submission/submit.py

@@ -3,6 +3,7 @@ import os
 from datetime import datetime, timezone
 import time
 
+from datasets import Dataset
 import pandas as pd
 
 from src.datamodel.data import F1Data
@@ -18,6 +19,22 @@ from src.logger import get_logger
 
 logger = get_logger(__name__)
 
+def validate_submission(lbdb: F1Data, pd_ds: pd.DataFrame) -> str | None:
+    if set(pd_ds.columns) != set(["formula_name", "solution"]):
+        return "Bad format of submission"
+    if any(type(v) != str for v in pd_ds["formula_name"]):
+        return "Not all formula_name values are of type str"
+    if any(type(v) != str for v in pd_ds["solution"]):
+        return "Not all solution values are of type str"
+    submitted_formulas = set(pd_ds["formula_name"])
+    if submitted_formulas != lbdb.code_problem_formulas:
+        missing = lbdb.code_problem_formulas - submitted_formulas
+        unknown = submitted_formulas - lbdb.code_problem_formulas
+        return f"Mismatched formula names: missing {len(missing)} unknown {len(unknown)}"
+    if len(pd_ds) > len(lbdb.code_problem_formulas):
+        return "Duplicate formula solutions exist in uploaded file"
+    return None
+
 def add_new_solutions(
     lbdb: F1Data,
     submitter: str,
@@ -31,17 +48,13 @@ def add_new_solutions(
         return styled_error("Please upload JSONL solutions file")
 
     try:
-        ds = pd.read_json(submission_path, lines=True)
+        pd_ds = pd.read_json(submission_path, lines=True)
     except Exception as e:
         return styled_error(f"Cannot read uploaded JSONL file: {str(e)}")
 
-    submitted_formulas = set(ds["formula_name"])
-    if submitted_formulas != lbdb.code_problem_formulas:
-        missing = lbdb.code_problem_formulas - submitted_formulas
-        unknown = submitted_formulas - lbdb.code_problem_formulas
-        return styled_error(f"Mismatched formula names: missing {len(missing)} unknown {len(unknown)}")
-    if len(ds) > len(lbdb.code_problem_formulas):
-        return styled_error("Duplicate formula solutions exist in uploaded file")
+    validation_error = validate_submission(lbdb, pd_ds)
+    if validation_error:
+        return styled_error(validation_error)
 
     submission_id = datetime.now().strftime("%Y%m%d%H%M%S")
 
@@ -54,7 +67,7 @@ def add_new_solutions(
         row["submission_id"] = submission_id
         row["submission_ts"] = submission_ts
 
-    ds = ds.map(add_info)
+    ds = Dataset.from_pandas(pd_ds).map(add_info)
 
     ds.push_to_hub(SUBMISSIONS_REPO, submission_id, private=True)
     # print("Creating eval file")