app.py

import gradio as gr
from sentence_transformers import SentenceTransformer, util
import openai
import os
import os
os.environ["TOKENIZERS_PARALLELISM"] = "false"


# Initialize paths and model identifiers for easy configuration and maintenance
filename = "output_country_details.txt"  # Path to the file storing country-specific details
retrieval_model_name = 'output/sentence-transformer-finetuned/'

openai.api_key = 'sk-proj-BVO7g5ig8PKdlQwDCZSeT3BlbkFJAvilYAEcPFbA0XOjz7ce'



# Attempt to load the necessary models and provide feedback on success or failure
try:
    retrieval_model = SentenceTransformer(retrieval_model_name)
    print("Models loaded successfully.")
except Exception as e:
    print(f"Failed to load models: {e}")

def load_and_preprocess_text(filename):
    """
    Load and preprocess text from a file, removing empty lines and stripping whitespace.
    """
    try:
        with open(filename, 'r', encoding='utf-8') as file:
            segments = [line.strip() for line in file if line.strip()]
        print("Text loaded and preprocessed successfully.")
        return segments
    except Exception as e:
        print(f"Failed to load or preprocess text: {e}")
        return []

segments = load_and_preprocess_text(filename)

def find_relevant_segment(user_query, segments):
    """
    Find the most relevant text segment for a user's query using cosine similarity among sentence embeddings.
    This version tries to match country names in the query with those in the segments.
    """
    try:
        # Lowercase the query for better matching
        lower_query = user_query.lower()
        # Filter segments to include only those containing country names mentioned in the query
        country_segments = [seg for seg in segments if any(country.lower() in seg.lower() for country in ['Guatemala', 'Mexico', 'U.S.', 'United States'])]
        
        # If no specific country segments found, default to general matching
        if not country_segments:
            country_segments = segments
        
        query_embedding = retrieval_model.encode(lower_query)
        segment_embeddings = retrieval_model.encode(country_segments)
        similarities = util.pytorch_cos_sim(query_embedding, segment_embeddings)[0]
        best_idx = similarities.argmax()
        return country_segments[best_idx]
    except Exception as e:
        print(f"Error in finding relevant segment: {e}")
        return ""


def generate_response(user_query, relevant_segment):
    """
    Generate a response using the latest GPT-3 model available via OpenAI's API.
    """
    try:
        prompt = f"Thank you for your question! Here's additional information: {relevant_segment}"
        response = openai.Completion.create(
            engine="gpt-3.5-turbo-instruct",  # Updated to a currently supported engine
            prompt=prompt,
            max_tokens=150,
            temperature=0.7,
            top_p=1,
            frequency_penalty=0,
            presence_penalty=0
        )
        return response.choices[0].text.strip()
    except Exception as e:
        print(f"Error in generating response: {e}")
        return f"Error in generating response: {e}"



# Define and configure the Gradio application interface to interact with users.
# Define and configure the Gradio application interface to interact with users.
def query_model(question):
    """
    Process a question, find relevant information, and generate a response, specifically for U.S. visa questions.
    """
    if question == "":
        return "Welcome to VisaBot! Ask me anything about U.S. visa processes."
    relevant_segment = find_relevant_segment(question, segments)
    if not relevant_segment:
        return "Could not find U.S.-specific information. Please refine your question."
    response = generate_response(question, relevant_segment)
    return response


# Define the welcome message and specific topics and countries the chatbot can provide information about.
welcome_message = """
# Welcome to VISABOT!

## Your AI-driven visa assistant for all travel-related queries.
"""

topics = """
### Feel Free to ask me anything from the topics below!
- Visa issuance
- Documents needed
- Application process
- Processing time
- Recommended Vaccines
- Health Risks
- Healthcare Facilities
- Currency Information
- Embassy Information 
- Allowed stay
"""

countries = """
### Our chatbot can currently answer questions for these countries!
- 🇨🇳 China
- 🇫🇷 France
- 🇬🇹 Guatemala
- 🇱🇧 Lebanon
- 🇲🇽 Mexico
- 🇵🇭 Philippines
- 🇷🇸 Serbia
- 🇸🇱 Sierra Leone
- 🇿🇦 South Africa
- 🇻🇳 Vietnam
"""

# Define and configure the Gradio application interface to interact with users.
def query_model(question):
    """
    Process a question, find relevant information, and generate a response.
    
    Args:
        question (str): User's input question.

    Returns:
        str: Generated response or a default welcome message if no question is provided.
    """
    if question == "":
        return welcome_message
    relevant_segment = find_relevant_segment(question, segments)
    response = generate_response(question, relevant_segment)
    return response

# Setup the Gradio Blocks interface with custom layout components
with gr.Blocks() as demo:
    gr.Markdown(welcome_message)  # Display the formatted welcome message
    with gr.Row():
        with gr.Column():
            gr.Markdown(topics)  # Show the topics on the left side
        with gr.Column():
            gr.Markdown(countries)  # Display the list of countries on the right side
    with gr.Row():
        img = gr.Image(os.path.join(os.getcwd(), "poster.png"), width=500)  # Include an image for visual appeal
    with gr.Row():
        with gr.Column():
            question = gr.Textbox(label="Your question", placeholder="What do you want to ask about?")
            answer = gr.Textbox(label="VisaBot Response", placeholder="VisaBot will respond here...", interactive=False, lines=10)
            submit_button = gr.Button("Submit")
            submit_button.click(fn=query_model, inputs=question, outputs=answer)

# Launch the Gradio app to allow user interaction
demo.launch(share= True)