app.py

import gradio as gr
import pandas as pd
import openai
import os
import io
import time

# Language dictionary for translations
LANGUAGES = {
    "English": {
        "title": "📊 Sentiment Analysis of Comments",
        "upload_prompt": "Upload an Excel (.xlsx), CSV (.csv), or text (.txt) file with comments and get sentiment analysis!",
        "upload_button": "Upload a file",
        "error_column": "The file must contain text in the first column",
        "results": "### Analysis Results:",
        "download": "📥 Download Report Excel",
        "unsupported_format": "Unsupported format. Please upload a .xlsx, .csv, or .txt file."
    }
}

# Function to get API Key securely
def get_openai_api_key():
    return os.getenv("OPENAI_API_KEY")

# OpenAI API Key setup
openai.api_key = os.environ.get("OPENAI_API_KEY", None)

# Cost tracking variables
MAX_DAILY_COST = 1.0  # Maximum daily cost in USD
TOKEN_COST = 0.002 / 1000  # Approximate cost per input token (GPT-3.5-turbo)
TOKEN_LIMIT = int(MAX_DAILY_COST / TOKEN_COST)  # Max tokens per day

daily_token_usage = 0  # Track daily token usage
last_reset_time = time.time()

def analyze_sentiment(text):
    global daily_token_usage, last_reset_time
    
    # Reset usage if a new day starts
    if time.time() - last_reset_time > 86400:
        daily_token_usage = 0
        last_reset_time = time.time()
    
    if daily_token_usage >= TOKEN_LIMIT:
        return "Daily limit reached"
    
    if not openai.api_key:
        return "API Key Missing"
    
    prompt = f"Classifica il sentiment del seguente commento come 'Positive', 'Negative' o 'Neutral': \"{text}\""
    
    try:
        response = openai.ChatCompletion.create(
            model="gpt-4",
            messages=[{"role": "system", "content": "Sei un assistente che esegue sentiment analysis."},
                      {"role": "user", "content": prompt}],
            temperature=0.0
        )
        
        sentiment = response["choices"][0]["message"]["content"].strip()
        token_used = response["usage"]["total_tokens"]
        daily_token_usage += token_used
        
        return sentiment if sentiment in ["Positive", "Negative", "Neutral"] else "Neutral"
    except Exception as e:
        return f"Error: {str(e)}"

def analyze_file(uploaded_file):
    try:
        # Load the file depending on the format
        if uploaded_file.name.endswith(".xlsx"):
            df = pd.read_excel(uploaded_file, engine='openpyxl')
        elif uploaded_file.name.endswith(".csv"):
            df = pd.read_csv(uploaded_file)
        else:
            return "Unsupported file format. Please upload a .xlsx or .csv file.", None
        
        # Check if the first column contains text
        if df.shape[1] == 0:
            return "Error: No valid columns found in the file.", None
        
        df = df.rename(columns={df.columns[0]: "Comment"})
        df["Sentiment"] = df["Comment"].astype(str).apply(analyze_sentiment)
        
        excel_output = io.BytesIO()
        with pd.ExcelWriter(excel_output, engine='openpyxl') as writer:
            df.to_excel(writer, index=False)
        excel_output.seek(0)
        
        return df, excel_output.getvalue()
    except Exception as e:
        return f"Error processing file: {str(e)}", None

def sentiment_analysis_interface():
    language = gr.Radio(["English"], label="🌍 Select Language", value="English")
    file_input = gr.File(label="📥 Upload a file (CSV, XLSX)")
    
    results_output = gr.Dataframe()
    download_button = gr.File(label="📥 Download Excel")
    
    def process_file(language, uploaded_file):
        df, excel_data = analyze_file(uploaded_file)
        if isinstance(df, str):
            return df, None
        with open("sentiment_analysis.xlsx", "wb") as f:
            f.write(excel_data)
        return df, "sentiment_analysis.xlsx"
    
    return gr.Interface(
        fn=process_file,
        inputs=[language, file_input],
        outputs=[results_output, download_button],
        title="📊 Sentiment Analysis",
        description="Upload a file with comments and get sentiment analysis using GPT-3.5-turbo!"
    )

iface = sentiment_analysis_interface()

if __name__ == "__main__":
    iface.launch()