import gradio as gr import pandas as pd from transformers import ( normalize_open_invoices, normalize_payment_notification, ) from invoice_schema import invoice_pipeline_schema from validator import ColumnValidator def process_files(open_invoices_file, payment_notification_file): results = {} if open_invoices_file is not None: # HF uploads are temp files with a .name attribute df_open = pd.read_excel(open_invoices_file.name) df_open_norm = normalize_open_invoices( df_open, schema=invoice_pipeline_schema, dataset_name="Open Invoices", ) ColumnValidator(invoice_pipeline_schema).validate( df_open_norm, dataset_name="Open Invoices (normalized)" ) results["Open Invoices (normalized) head"] = df_open_norm.head().to_markdown() if payment_notification_file is not None: df_pay = pd.read_excel(payment_notification_file.name) df_pay_norm = normalize_payment_notification( df_pay, schema=invoice_pipeline_schema, dataset_name="Payment Notification", ) ColumnValidator(invoice_pipeline_schema).validate( df_pay_norm, dataset_name="Payment Notification (normalized)" ) results["Payment Notification (normalized) head"] = ( df_pay_norm.head().to_markdown() ) if not results: return "Upload at least one Excel file to see normalized output." parts = [] for title, md in results.items(): parts.append(f"## {title}\n\n{md}") return "\n\n".join(parts) with gr.Blocks() as demo: gr.Markdown("# Column Matching & Normalization Demo") gr.Markdown( "Upload your **Open Invoices** and/or **Payment Notification** Excel files.\n" "The app will normalize them into the canonical schema and show the first few rows." ) with gr.Row(): open_file = gr.File(label="Open Invoices (.xlsx)", file_types=[".xlsx"]) pay_file = gr.File(label="Payment Notification (.xlsx)", file_types=[".xlsx"]) output = gr.Markdown() run_btn = gr.Button("Normalize & Validate") run_btn.click(process_files, inputs=[open_file, pay_file], outputs=output) if __name__ == "__main__": demo.launch()