async test

app.py

@@ -4,26 +4,27 @@ import time
 import traceback
 from validation import validate_json, validate_croissant, validate_records
 import requests
+import aiohttp
 
-def process_file(file):
+async def process_file(file):
     results = []
     
     # Check 1: JSON validation
-    json_valid, json_message, json_data = validate_json(file.name)
+    json_valid, json_message, json_data = await validate_json(file.name)
     results.append(("JSON Format Validation", json_valid, json_message))
     
     if not json_valid:
         return results
     
     # Check 2: Croissant validation
-    croissant_valid, croissant_message = validate_croissant(json_data)
+    croissant_valid, croissant_message = await validate_croissant(json_data)
     results.append(("Croissant Schema Validation", croissant_valid, croissant_message))
     
     if not croissant_valid:
         return results
     
     # Check 3: Records validation
-    records_valid, records_message = validate_records(json_data)
+    records_valid, records_message = await validate_records(json_data)
     results.append(("Records Generation Test", records_valid, records_message))
     
     return results
@@ -181,15 +182,16 @@ def create_ui():
             
             return """<div class="progress-status">✅ File uploaded successfully</div>""", gr.update(visible=False)
         
-        def fetch_from_url(url):
+        async def fetch_from_url(url):
             if not url:
                 return """<div class="progress-status">Please enter a URL</div>""", gr.update(visible=False)
             
             try:
                 # Fetch JSON from URL
-                response = requests.get(url, timeout=10)
-                response.raise_for_status()
-                json_data = response.json()
+                async with aiohttp.ClientSession() as session:
+                    async with session.get(url, timeout=10) as response:
+                        response.raise_for_status()
+                        json_data = await response.json()
                 
                 # Show success message
                 progress_html = """<div class="progress-status">✅ JSON fetched successfully from URL</div>"""
@@ -198,18 +200,18 @@ def create_ui():
                 results = []
                 results.append(("JSON Format Validation", True, "✅ The URL returned valid JSON."))
                 
-                croissant_valid, croissant_message = validate_croissant(json_data)
+                croissant_valid, croissant_message = await validate_croissant(json_data)
                 results.append(("Croissant Schema Validation", croissant_valid, croissant_message))
                 
                 if not croissant_valid:
                     return progress_html, build_results_html(results)
                 
-                records_valid, records_message = validate_records(json_data)
+                records_valid, records_message = await validate_records(json_data)
                 results.append(("Records Generation Test", records_valid, records_message))
                 
                 return progress_html, build_results_html(results)
                 
-            except requests.exceptions.RequestException as e:
+            except aiohttp.ClientError as e:
                 error_message = f"❌ Error fetching URL: {str(e)}"
                 return f"""<div class="progress-status">{error_message}</div>""", gr.update(visible=False)
             except json.JSONDecodeError as e:
@@ -254,19 +256,19 @@ def create_ui():
             html += '</div>'
             return gr.update(value=html, visible=True)
         
-        def on_validate(file):
+        async def on_validate(file):
             if file is None:
                 return gr.update(visible=False)
             
             # Process the file and get results
-            results = process_file(file)
+            results = await process_file(file)
             return build_results_html(results)
         
         # Connect UI events to functions
         tabs.select(on_tab_change, None, [active_tab, upload_progress, validation_results])
         file_input.change(on_file_upload, inputs=file_input, outputs=[upload_progress, validation_results])
-        validate_btn.click(on_validate, inputs=file_input, outputs=validation_results)
-        fetch_btn.click(fetch_from_url, inputs=url_input, outputs=[upload_progress, validation_results])
+        validate_btn.click(fn=on_validate, inputs=file_input, outputs=validation_results)
+        fetch_btn.click(fn=fetch_from_url, inputs=url_input, outputs=[upload_progress, validation_results])
         
         # Footer
         gr.HTML("""
@@ -279,4 +281,9 @@ def create_ui():
 
 if __name__ == "__main__":
     app = create_ui()
-    app.launch() 
+    app.launch(
+        max_threads=1,  # Reduce thread conflicts
+        show_error=True,
+        queue_timeout=300,  # Match our 5-minute timeout
+        share=False
+    ) 

requirements.txt

@@ -4,8 +4,7 @@ gradio>=3.50.2
 func_timeout
 requests 
 huggingface-hub==0.30.1
-fsspec==2025.3.2
-gcsfs==2025.3.2
+fsspec==2023.10.0
 aiohttp==3.11.15
 aiohappyeyeballs==2.6.1
 pandas==2.2.2

validation.py

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