# utils/models.py
import os
import joblib
import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd
from huggingface_hub import hf_hub_download
import joblib




# ------------------------------------------------------------
# 🧠 Universal Model Loader
# ------------------------------------------------------------
def load_model(model_path: str = "app_best.joblib"):
    """
    Loads a trained scikit-learn model (.pkl or .joblib) from disk.
    Automatically searches in the /models folder if a relative path is provided.
    """
    
    # Download model dynamically from Hugging Face model repo
    if os.getenv("SPACE_ID"):  # Running inside a Hugging Face Space
        model_path_hf = hf_hub_download(
            repo_id="VasTk/user-churn-models",
            filename=model_path
        )
        model = joblib.load(model_path_hf)
        return model
    else:
        # Try direct path first
        if os.path.exists(model_path):
            return joblib.load(model_path)

        # Try inside models/ folder
        candidate_path = os.path.join("models", model_path)
        if os.path.exists(candidate_path):
            return joblib.load(candidate_path)
        raise FileNotFoundError(
            f"❌ Model file not found. Tried: {model_path_hf} and {model_path_hf}"
        )
        
    

# ------------------------------------------------------------
# 📊 Example placeholder metrics and visuals
# ------------------------------------------------------------
metrics = pd.DataFrame({
    "Model": ["Random Forest (App)", "Logistic Regression (App)"],
    "Accuracy": [0.82, 0.75],
    "AUC": [0.88, 0.80],
    "F1": [0.79, 0.72]
})

feature_importance = pd.DataFrame({
    "Feature": ["Recency", "Session Count"],
    "Importance": [0.7, 0.3]
})

fairness = pd.DataFrame({
    "Group": ["Male", "Female"],
    "Accuracy": [0.81, 0.83],
    "Precision": [0.78, 0.76],
    "Recall": [0.80, 0.85]
})

def show_metrics_table():
    """Returns model comparison metrics as a table."""
    return metrics

def plot_feature_importance():
    """Returns a Matplotlib bar plot of feature importance."""
    fig, ax = plt.subplots()
    sns.barplot(data=feature_importance, x="Importance", y="Feature", ax=ax)
    ax.set_title("Feature Importance")
    return fig

def show_fairness_table():
    """Returns fairness comparison metrics."""
    return fairness