Merge working-april-04 validation changes

validation.py

@@ -1,12 +1,11 @@
 import json
-import mlcroissant as mlc
 import traceback
-import asyncio
-from asyncio import timeout
+import mlcroissant as mlc
+import func_timeout
 
 WAIT_TIME = 5 * 60  # seconds
 
-async def validate_json(file_path):
+def validate_json(file_path):
     """Validate that the file is proper JSON."""
     try:
         with open(file_path, 'r') as f:
@@ -19,7 +18,7 @@ async def validate_json(file_path):
         error_message = f"❌ Error reading file: {str(e)}"
         return False, error_message, None
 
-async def validate_croissant(json_data):
+def validate_croissant(json_data):
     """Validate that the JSON follows Croissant schema."""
     try:
         dataset = mlc.Dataset(jsonld=json_data)
@@ -33,7 +32,7 @@ async def validate_croissant(json_data):
         error_message = f"❌ Unexpected error during validation: {str(e)}\n\n{error_details}"
         return False, error_message
 
-async def validate_records(json_data):
+def validate_records(json_data):
     """Validate that records can be generated within the time limit."""
     try:
         dataset = mlc.Dataset(jsonld=json_data)
@@ -46,13 +45,12 @@ async def validate_records(json_data):
         
         for record_set in record_sets:
             try:
-                async with timeout(WAIT_TIME):
-                    records = dataset.records(record_set=record_set.uuid)
-                    print(records)
-                    _ = next(iter(records))
-                    results.append(f"✅ Record set '{record_set.uuid}' passed validation.")
-            except asyncio.TimeoutError:
-                error_message = f"❌ Record set '{record_set.uuid}' generation took too long (>{WAIT_TIME}s)"
+                records = dataset.records(record_set=record_set.uuid)
+                print(records)
+                _ = func_timeout.func_timeout(WAIT_TIME, lambda: next(iter(records)))
+                results.append(f"✅ Record set '{record_set.uuid}' passed validation.")
+            except func_timeout.exceptions.FunctionTimedOut:
+                error_message = f"❌ Record set '{record_set.uuid}' generation took too long (>300s)"
                 return False, error_message
             except Exception as e:
                 error_details = traceback.format_exc()