# exclude Executive
import gradio as gr
import pandas as pd
import matplotlib.pyplot as plt
from transformers import pipeline

# Load the sentiment analysis model (supports neutral feedback)
pipe = pipeline("text-classification", model="cardiffnlp/twitter-roberta-base-sentiment-latest")

# Function to process CSV and generate dropdown options
# Function to process CSV and generate dropdown options
def process_csv(file):
    df = pd.read_csv(file.name)
    
    # Exclude the 'Executive' department from the department list
    departments = ['All'] + [dept for dept in df['Department'].unique().tolist() if dept != 'Executive']
    
    # Include all questions
    questions = ['All'] + df['Question'].unique().tolist()
    
    return gr.update(choices=departments), gr.update(choices=questions)


# Function to analyze sentiments and generate a styled HTML table
# Function to analyze sentiments and generate a styled HTML table
def get_analysis(department, question, file):
    df = pd.read_csv(file.name)

    # Exclude the Executive department
    df = df[df['Department'] != 'Executive']

    # Filter by department and question
    if department != 'All':
        df = df[df['Department'] == department]
    if question != 'All':
        df = df[df['Question'] == question]

    # Perform sentiment analysis
    df['Sentiment'] = df['Feedback'].apply(lambda x: pipe(x)[0]['label'])
    
    # Count employees per department
    employee_counts = df.groupby('Department')['Employee'].nunique().reset_index()
    employee_counts.columns = ['Department', 'Total Employees']

    # Count positive, neutral, and negative feedback per department
    sentiment_counts = df.groupby(['Department', 'Sentiment']).size().unstack(fill_value=0)
    sentiment_counts = sentiment_counts.reindex(columns=['positive', 'neutral', 'negative'], fill_value=0)

    # Merge employee counts with sentiment counts
    summary_df = employee_counts.merge(sentiment_counts, on="Department", how="left")

    # Add a row for total counts
    total_row = pd.DataFrame({
        'Department': ['Total'],
        'Total Employees': [df['Employee'].nunique()],
        'positive': [summary_df['positive'].sum()],
        'neutral': [summary_df['neutral'].sum()],
        'negative': [summary_df['negative'].sum()]
    })

    summary_df = pd.concat([summary_df, total_row], ignore_index=True)

    # **Generate HTML Table with Colors**
    def generate_html_table(df):
        html = """<style>
            table {width: 100%; border-collapse: collapse; font-family: Arial, sans-serif; text-align: center;}
            th {background-color: gray; padding: 10px;}
            td, th {border: 1px solid #ddd; padding: 8px;}
            tr:nth-child(even) {background-color: #f2f2f2;}
            .total {background-color: lightgray; font-weight: bold;}
            .positive {color: green; font-weight: bold;}
            .neutral {color: gray; font-weight: bold;}
            .negative {color: red; font-weight: bold;}
        </style>
        <table>
        <tr>
            <th>Department</th><th>Total Employees</th>
            <th>Positive</th><th>Neutral</th><th>Negative</th>
        </tr>"""
        for _, row in df.iterrows():
            row_class = "total" if row["Department"] == "Total" else ""
            html += f"""
            <tr class="{row_class}">
                <td>{row["Department"]}</td>
                <td>{row["Total Employees"]}</td>
                <td class="positive">{row["positive"]}</td>
                <td class="neutral">{row["neutral"]}</td>
                <td class="negative">{row["negative"]}</td>
            </tr>"""
        html += "</table>"
        return html

    table_html = generate_html_table(summary_df)

    # **Plot department-wise sentiment distribution**
    plt.figure(figsize=(10, 6))
    sentiment_counts.plot(kind='bar', stacked=False, color=['green', 'gray', 'red'], width=0.6)
    plt.xlabel("Department")
    plt.ylabel("Count")
    plt.title("Department-wise Sentiment Analysis")
    plt.xticks(rotation=45)
    plt.legend(title="Sentiment", loc="upper right")
    plt.grid(axis='y', linestyle='--', alpha=0.7)
    plt.tight_layout()

    return table_html, plt.gcf()

# **Gradio UI**
with gr.Blocks() as demo:
    gr.Markdown("### Employee Sentiment Analysis")

    file_input = gr.File(label="Upload CSV", file_types=[".csv"])
    process_button = gr.Button("Process CSV")

    dept_dropdown = gr.Dropdown(label="Department", choices=["All"], interactive=True)
    question_dropdown = gr.Dropdown(label="Question", choices=["All"], interactive=True)

    analyze_button = gr.Button("Get Analysis")

    output_table = gr.HTML(label="Department-wise Sentiment Summary")
    output_plot = gr.Plot(label="Department-wise Sentiment Distribution")

    process_button.click(process_csv, inputs=file_input, outputs=[dept_dropdown, question_dropdown])
    analyze_button.click(get_analysis, inputs=[dept_dropdown, question_dropdown, file_input], outputs=[output_table, output_plot])

demo.launch(share=True)