app.py

import os
import gradio as gr
import pandas as pd
import joblib

# Load your model and feature list
model = joblib.load("ar_overdue_model.joblib")
feature_names = joblib.load("ar_model_features.joblib")

def predict(company_code, document_type, amount, due_in_days):
    # Build the input DataFrame
    input_dict = {
        "company_code": company_code,
        "document_type": document_type,
        "amount": amount,
        "due_in_days": due_in_days
    }
    input_df = pd.DataFrame([input_dict])
    # One-hot encode and align columns
    input_df = pd.get_dummies(input_df)
    for col in feature_names:
        if col not in input_df.columns:
            input_df[col] = 0
    input_df = input_df[feature_names]
    # Predict
    proba = model.predict_proba(input_df)[0, 1]
    pred = model.predict(input_df)[0]
    return f"Overdue: {bool(pred)} (Probability: {proba:.2f})"

# Define the Gradio interface
iface = gr.Interface(
    fn=predict,
    inputs=[
        gr.Dropdown(['CompanyA', 'CompanyB', 'CompanyC'], label="Company Code"),
        gr.Dropdown(['INV', 'CRN', 'DBN'],      label="Document Type"),
        gr.Number(label="Amount"),
        gr.Number(label="Due In Days")
    ],
    outputs="text",
    title="AR Overdue Prediction",
    description="Enter invoice details to predict overdue probability."
)

if __name__ == "__main__":
    # 1) Turn on the async queue so the /api/* routes get mounted
    iface = iface.queue()
    # 2) Read the HF Spaces port (default to 7860 locally)
    port = int(os.environ.get("PORT", 7860))
    # 3) Launch on all interfaces
    iface.launch(server_name="0.0.0.0", server_port=port)