import gradio as gr
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import io

# --- 1. Function to process the uploaded file and generate stats/dataframe ---
def analyze_csv(file):
    """Reads a CSV file, calculates statistics, and prepares components for plotting."""
    if file is None:
        # Clear previous outputs if a file is unuploaded or if this is the initial state
        return None, "Please upload a CSV file.", gr.Dropdown(choices=[], label="Select Column to Plot"), None

    try:
        # Read the uploaded file object (it's a temporary file path)
        df = pd.read_csv(file.name)

        # Basic descriptive statistics
        stats = df.describe().round(2).T
        
        # Convert stats to markdown string for display
        stats_markdown = stats.to_markdown()

        # Get list of numeric columns for the dropdown
        numeric_cols = df.select_dtypes(include=['number']).columns.tolist()
        
        # Return the full DataFrame, the statistics table, the updated column dropdown, and the DataFrame state
        return df, stats_markdown, gr.Dropdown(choices=numeric_cols, label="Select Column to Plot"), df
        
    except Exception as e:
        error_message = f"Error processing file: {e}"
        # Return error message and clear other components
        return None, error_message, gr.Dropdown(choices=[], label="Select Column to Plot"), None


# --- 2. Function to generate a plot for a selected column ---
def generate_plot(df_state, column_name):
    """Generates a distribution plot (histogram) for the selected column."""
    if df_state is None or column_name is None or column_name == "":
        return None

    try:
        # Reset Matplotlib figure for a clean slate
        plt.figure(figsize=(8, 5))
        
        # Use a Histogram for distribution of the selected numeric column
        sns.histplot(df_state[column_name], kde=True)
        
        plt.title(f'Distribution of {column_name}')
        plt.xlabel(column_name)
        plt.ylabel('Frequency')
        
        # Save plot to an in-memory buffer
        buf = io.BytesIO()
        plt.savefig(buf, format='png')
        plt.close() # Close figure to free memory
        buf.seek(0)
        
        return buf.read()

    except Exception as e:
        print(f"Plotting Error: {e}")
        return None


# --- 3. Gradio Interface Definition ---
with gr.Blocks(title="CSV Data Analyzer") as demo:
    gr.Markdown("## 📈 CSV Data Analyzer")
    gr.Markdown("Upload your CSV file and see instant statistics and visualizations.")

    # State component to hold the DataFrame across function calls
    # This is essential for passing the DataFrame from `analyze_csv` to `generate_plot`
    df_state = gr.State(None)

    # Input components
    csv_file = gr.File(label="Upload CSV File (.csv)", file_types=[".csv"])
    
    with gr.Row():
        # Outputs
        # **FIXED:** Removed the unsupported 'height=200' argument.
        df_output = gr.Dataframe(label="Uploaded Data Preview", interactive=False)
        stats_output = gr.Markdown(label="Descriptive Statistics")
    
    gr.HTML("<hr>")

    # Plotting section
    with gr.Row():
        # Component that will be populated with column names
        column_dropdown = gr.Dropdown(label="Select Column to Plot", interactive=True)
        plot_button = gr.Button("Generate Plot")
    
    plot_output = gr.Plot(label="Column Distribution Plot")


    # --- 4. Event Handling (Interactions) ---
    
    # Triggered when a file is uploaded (or cleared)
    csv_file.upload(
        analyze_csv, 
        inputs=[csv_file], 
        # Note: The output for the dropdown component is updated with new choices here.
        outputs=[df_output, stats_output, column_dropdown, df_state]
    )

    # Triggered when the plot button is clicked
    plot_button.click(
        generate_plot, 
        inputs=[df_state, column_dropdown], 
        outputs=[plot_output]
    )


# Launch the app for the Hugging Face Space environment
if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)