from transformers import pipeline  # For loading Hugging Face pre-trained models
import gradio as gr               # For creating the user interface
import pandas as pd              # For handling CSV and Excel files
import matplotlib.pyplot as plt   # For plotting the pie chart

# Load the pre-trained sentiment-analysis model from Hugging Face
# This is a binary sentiment model (Positive/Negative)
sentiment_analyzer = pipeline(
    "sentiment-analysis",
    model="distilbert-base-uncased-finetuned-sst-2-english"  # Model fine-tuned on SST-2 for sentiment analysis
)

# Function to analyze sentiment of a single text input
def analyze_sentiment(text):
    # Run the text through the sentiment analyzer
    result = sentiment_analyzer(text)
    
    # Return the sentiment label (Positive/Negative)
    return result[0]['label']

# Function to analyze a file (CSV or Excel) and return sentiment results + pie chart
def analyze_file(file):
    # Read the uploaded file based on its extension (CSV or Excel)
    if file.name.endswith(".csv"):
        df = pd.read_csv(file.name)  # Read as CSV
    elif file.name.endswith(".xlsx"):
        df = pd.read_excel(file.name)  # Read as Excel
    else:
        return "Unsupported file format", None  # Return error message if the file format is unsupported

    # Check if 'Text' column is present in the file
    if "Text" not in df.columns:
        return "Missing 'Text' column", None  # Return error if 'Text' column is missing

    # Predict sentiment for each row in the 'Text' column
    sentiments = [sentiment_analyzer(t)[0]['label'] for t in df["Text"].fillna("")]  # Handle empty cells with fillna("")
    
    # Create a DataFrame with 'Text' and corresponding 'Sentiment'
    df_result = pd.DataFrame({
        "Text": df["Text"],  # Original text column
        "Sentiment": sentiments  # Sentiment predictions
    })

    # Generate a pie chart to show the distribution of sentiments
    sentiment_counts = df_result["Sentiment"].value_counts()  # Count the occurrences of each sentiment
    fig, ax = plt.subplots()  # Create a figure and axis for plotting
    ax.pie(
        sentiment_counts.values,  # Data for the pie chart
        labels=sentiment_counts.index,  # Sentiment labels (e.g., 'POSITIVE', 'NEGATIVE')
        autopct="%1.1f%%",  # Show percentage on the pie chart
        colors=["#66bb6a", "#ef5350", "#ffee58"][:len(sentiment_counts)]  # Color palette for the pie chart
    )
    ax.set_title("Sentiment Distribution")  # Title of the pie chart

    return df_result, fig  # Return the DataFrame and the pie chart figure

# Creating the Gradio interface with two tabs: one for text input and one for file upload
with gr.Blocks() as demo:
    gr.Markdown("## 🧠 Text Sentiment Analyzer with Pie Chart")  # Title
    gr.Markdown("Enter text manually or upload a file with a **Text** column.")  # Description

    # Tab for analyzing a single text input
    with gr.Tab("🔤 Single Text"):
        txt = gr.Textbox(label="Text", lines=4)  # Textbox for text input
        out = gr.Textbox(label="Sentiment")  # Textbox to show sentiment output
        btn = gr.Button("Analyze")  # Button to trigger analysis
        btn.click(analyze_sentiment, inputs=txt, outputs=out)  # Trigger analysis when button is clicked

    # Tab for uploading a file (CSV or Excel)
    with gr.Tab("📁 Upload File"):
        file_input = gr.File(label="Upload CSV or Excel File")  # File uploader
        file_output = gr.Dataframe(label="Sentiment Results")  # Table to display the sentiment results
        chart_output = gr.Plot(label="Sentiment Distribution Pie Chart")  # Pie chart to display sentiment distribution
        btn2 = gr.Button("Analyze File")  # Button to trigger file analysis
        btn2.click(analyze_file, inputs=file_input, outputs=[file_output, chart_output])  # Trigger file analysis when button is clicked

# Launch the app
demo.launch()