import gradio as gr
import pandas as pd
import joblib
from huggingface_hub import hf_hub_download
import numpy as np

# Download model and feature names from Hugging Face
model_path = hf_hub_download(repo_id="alperugurcan/mercedes", filename="mercedes_model.joblib")
feature_names_path = hf_hub_download(repo_id="alperugurcan/mercedes", filename="feature_names.joblib")

# Load the saved model and feature names
model = joblib.load(model_path)
feature_names = joblib.load(feature_names_path)

# Most common X0 values with their frequencies
FEATURE_OPTIONS = {
    "z (Most Common - 360 cases)": "z",
    "ak (349 cases)": "ak",
    "y (324 cases)": "y",
    "ay (313 cases)": "ay",
    "t (306 cases)": "t",
    "x (300 cases)": "x",
    "o (269 cases)": "o",
    "f (227 cases)": "f",
    "n (195 cases)": "n",
    "w (182 cases)": "w"
}

# Default values for other features
DEFAULT_VALUES = {name: 0.0 for name in feature_names}

def predict(selected_option):
    try:
        # Create a dictionary with all features set to default values
        input_dict = DEFAULT_VALUES.copy()
        
        # Get the actual value from the selected option
        selected_value = FEATURE_OPTIONS[selected_option]
        
        # Create dummy variable columns for X0
        for val in set(FEATURE_OPTIONS.values()):
            col_name = f'X0_{val}'
            input_dict[col_name] = 1 if val == selected_value else 0
        
        # Create DataFrame with all features
        df = pd.DataFrame([input_dict])
        
        # Make prediction
        if hasattr(model, '_Booster'):
            booster = model._Booster
            prediction = booster.predict(df)[0]
        else:
            prediction = model.predict(df)[0]
            
        return f"Predicted manufacturing time: {prediction:.2f} seconds"
    except Exception as e:
        return f"Error in prediction: {str(e)}"

# Create interface with dropdown
interface = gr.Interface(
    fn=predict,
    inputs=gr.Dropdown(
        choices=list(FEATURE_OPTIONS.keys()),
        label="Select Manufacturing Configuration (X0)",
        value=list(FEATURE_OPTIONS.keys())[0]
    ),
    outputs=gr.Textbox(label="Prediction Result"),
    title="Mercedes-Benz Manufacturing Time Predictor",
    description="Select one of the most common manufacturing configurations to predict the production time. The options are sorted by frequency of occurrence in the training data.",
    examples=[[list(FEATURE_OPTIONS.keys())[0]]],
    cache_examples=True,
    theme=gr.themes.Soft()
)

interface.launch(debug=True)