import gradio as gr import json import time import traceback from validation import validate_json, validate_croissant, validate_records, generate_validation_report import requests def process_file(file): results = [] json_data = None filename = file.name.split("/")[-1] # Check 1: JSON validation json_valid, json_message, json_data = validate_json(file.name) json_message = json_message.replace("\nāœ“\n", "\n") results.append(("JSON Format Validation", json_valid, json_message, "pass" if json_valid else "error")) if not json_valid: return results, None # Check 2: Croissant validation croissant_valid, croissant_message = validate_croissant(json_data) croissant_message = croissant_message.replace("\nāœ“\n", "\n") results.append(("Croissant Schema Validation", croissant_valid, croissant_message, "pass" if croissant_valid else "error")) if not croissant_valid: return results, None # Check 3: Records validation (with timeout-safe and error-specific logic) records_valid, records_message, records_status = validate_records(json_data) records_message = records_message.replace("\nāœ“\n", "\n") results.append(("Records Generation Test", records_valid, records_message, records_status)) # Generate final report report = generate_validation_report(filename, json_data, results) return results, report def on_validate_stream(file): results = [] def add_and_yield(name, passed, message, status, progress_message=None): results.append((name, passed, message, status)) html = build_results_html(results) return ( gr.update(value=html, visible=True), gr.update(value=f"
{progress_message}
", visible=True) if progress_message else gr.update(), gr.update(visible=False), None, None ) if file is None: yield gr.update(visible=False), gr.update(value="No file selected", visible=True), gr.update(visible=False), None, None return filename = file.name.split("/")[-1] yield gr.update(visible=False), gr.update(value="
šŸ” Starting validation...
", visible=True), gr.update(visible=False), None, None # Step 1 json_valid, json_message, json_data = validate_json(file.name) yield add_and_yield("JSON Format Validation", json_valid, json_message, "pass" if json_valid else "error", "āœ… JSON check done. Checking Croissant schema...") if not json_valid: return # Step 2 croissant_valid, croissant_message = validate_croissant(json_data) yield add_and_yield("Croissant Schema Validation", croissant_valid, croissant_message, "pass" if croissant_valid else "error", "āœ… Croissant schema check done. Testing record generation...") if not croissant_valid: return # Step 3 records_valid, records_message, records_status = validate_records(json_data) yield add_and_yield("Records Generation Test", records_valid, records_message, records_status, "āœ… Record test complete.") # Generate and save report report = generate_validation_report(filename, json_data, results) report_filename = f"report_croissant-validation_{json_data.get('name', 'unnamed')}.md" with open(report_filename, "w") as f: f.write(report) # Final state yield gr.update(value=build_results_html(results), visible=True), \ gr.update(visible=False), \ gr.update(visible=True), \ report, \ report_filename def create_ui(): with gr.Blocks(theme=gr.themes.Soft()) as app: gr.HTML("

NeurIPS Logo

") gr.Markdown("# 🄐 Croissant Validator for NeurIPS") gr.Markdown(""" Upload your Croissant JSON-LD file or enter a URL to validate if it meets the requirements for NeurIPS submission. Read more about why this is required.. *It doesn't matter where your dataset is hosted. It can be Hugging Face, Kaggle, OpenML, Harvard Dataverse, self-hosted, or elsewhere.* """) gr.Markdown(""" The validator will check: 1. If the file is valid JSON 2. If it passes Croissant schema validation 3. If records can be generated within a reasonable time """) # Track the active tab for conditional UI updates active_tab = gr.State("upload") # Default to upload tab # Create a container for the entire input section with gr.Group(): # Input tabs with gr.Tabs() as tabs: with gr.TabItem("Upload File", id="upload_tab"): file_input = gr.File(label="Upload Croissant JSON-LD File", file_types=[".json", ".jsonld"]) validate_btn = gr.Button("Validate Uploaded File", variant="primary") with gr.TabItem("URL Input", id="url_tab"): url_input = gr.Textbox( label="Enter Croissant JSON-LD URL", placeholder="e.g. https://huggingface.co/api/datasets/facebook/natural_reasoning/croissant" ) fetch_btn = gr.Button("Fetch and Validate", variant="primary") # Change initial message to match upload tab upload_progress = gr.HTML( """
Ready for upload
""", visible=True) # Now create the validation results section in a separate group with gr.Group(): # Validation results validation_results = gr.HTML(visible=False) validation_progress = gr.HTML(visible=False) # Collapsible report section with gr.Accordion("Download full validation report", visible=False, open=False) as report_group: with gr.Column(): report_md = gr.File( label="Download Report", visible=True, file_types=[".md"] ) report_text = gr.Textbox( label="Report Content", visible=True, show_copy_button=True, lines=10 ) # Define CSS for the validation UI gr.HTML(""" """) # Update helper messages based on tab changes def on_tab_change(evt: gr.SelectData): tab_id = evt.value if tab_id == "Upload File": return [ "upload", """
Ready for upload
""", gr.update(visible=False), gr.update(visible=False), # Hide report group None, # Clear report text None, # Clear report file None, # Clear file input gr.update(value="") # Clear URL input ] else: return [ "url", """
Enter a URL to fetch
""", gr.update(visible=False), gr.update(visible=False), # Hide report group None, # Clear report text None, # Clear report file None, # Clear file input gr.update(value="") # Clear URL input ] def on_copy_click(report): return report def on_download_click(report, file_name): report_file = f"report_{file_name}.md" with open(report_file, "w") as f: f.write(report) return report_file def on_file_upload(file): if file is None: return [ """
Ready for upload
""", gr.update(visible=False), gr.update(visible=False), # Hide report group None, # Clear report text None # Clear report file ] return [ """
āœ… File uploaded successfully
""", gr.update(visible=False), gr.update(visible=False), # Hide report group None, # Clear report text None # Clear report file ] def fetch_from_url(url): if not url: return [ """
Please enter a URL
""", gr.update(visible=False), gr.update(visible=False), None, None ] try: # Fetch JSON from URL response = requests.get(url, timeout=10) response.raise_for_status() json_data = response.json() # Process validation progress_html = """
āœ… JSON fetched successfully from URL
""" # Validate the fetched JSON results = [] results.append(("JSON Format Validation", True, "The URL returned valid JSON.")) croissant_valid, croissant_message = validate_croissant(json_data) results.append(("Croissant Schema Validation", croissant_valid, croissant_message)) if not croissant_valid: return [ """
āœ… JSON fetched successfully from URL
""", build_results_html(results), gr.update(visible=False), None, None ] records_valid, records_message, records_status = validate_records(json_data) results.append(("Records Generation Test (Optional)", records_valid, records_message, records_status)) # Generate report report = generate_validation_report(url.split("/")[-1], json_data, results) report_filename = f"report_croissant-validation_{json_data.get('name', 'unnamed')}.md" if report: with open(report_filename, "w") as f: f.write(report) return [ """
āœ… JSON fetched successfully from URL
""", build_results_html(results), gr.update(visible=True), report, report_filename ] except requests.exceptions.RequestException as e: error_message = f"Error fetching URL: {str(e)}" return [ f"""
{error_message}
""", gr.update(visible=False), gr.update(visible=False), None, None ] except json.JSONDecodeError as e: error_message = f"URL did not return valid JSON: {str(e)}" return [ f"""
{error_message}
""", gr.update(visible=False), gr.update(visible=False), None, None ] except Exception as e: error_message = f"Unexpected error: {str(e)}" return [ f"""
{error_message}
""", gr.update(visible=False), gr.update(visible=False), None, None ] def build_results_html(results): html = '
' for i, result in enumerate(results): if len(result) == 4: test_name, passed, message, status = result else: test_name, passed, message = result status = "pass" if passed else "error" if status == "pass": status_class = "status-success" status_icon = "āœ“" message_with_emoji = "āœ… " + message elif status == "warning": status_class = "status-warning" status_icon = "?" message_with_emoji = "āš ļø Could not automatically generate records. This is oftentimes not an issue (e.g. datasets could be too large or too complex), and it's not required to pass this test to submit to NeurIPS.\n\n" + message else: # error status_class = "status-error" status_icon = "āœ—" message_with_emoji = "āŒ " + message message_with_emoji = message_with_emoji.replace("\n", "
") html += f'''
{status_icon}
{test_name} ā–¶
''' html += '
' return gr.update(value=html, visible=True) def on_validate(file): if file is None: return [ gr.update(visible=False), # validation_results gr.update(visible=False), # validation_progress gr.update(visible=False), # report_group None, # report_text None # report_md ] # Process the file and get results results, report = process_file(file) # Extract dataset name from the JSON for the report filename try: with open(file.name, 'r') as f: json_data = json.load(f) dataset_name = json_data.get('name', 'unnamed') except: dataset_name = 'unnamed' # Save report to file with new naming convention report_filename = f"report_croissant-validation_{dataset_name}.md" if report: with open(report_filename, "w") as f: f.write(report) # Return final state return [ build_results_html(results), # validation_results gr.update(visible=False), # validation_progress gr.update(visible=True) if report else gr.update(visible=False), # report_group report if report else None, # report_text report_filename if report else None # report_md ] # Connect UI events to functions with updated outputs tabs.select( on_tab_change, None, [active_tab, upload_progress, validation_results, report_group, report_text, report_md, file_input, url_input] ) file_input.change( on_file_upload, inputs=file_input, outputs=[upload_progress, validation_results, report_group, report_text, report_md] ) # Add progress state handling def show_progress(): progress_html = """
Validating file...
""" return [ gr.update(visible=False), # validation_results gr.update(visible=True, value=progress_html), # validation_progress gr.update(visible=False), # report_group None, # report_text None # report_md ] fetch_btn.click( fetch_from_url, inputs=url_input, outputs=[upload_progress, validation_results, report_group, report_text, report_md] ) # Footer gr.HTML("""

Learn more about Croissant.

""") def setup_events(): validate_btn.click( fn=on_validate_stream, inputs=[file_input], outputs=[validation_results, validation_progress, report_group, report_text, report_md], stream=True ) app.load(fn=setup_events) return app if __name__ == "__main__": app = create_ui() app.queue().launch()