Upload app.py

app.py

@@ -1,24 +1,30 @@
 import gradio as gr
 from sentence_transformers import SentenceTransformer, util
-from transformers import pipeline, GPT2Tokenizer
+import openai
 import os
+import os
+os.environ["TOKENIZERS_PARALLELISM"] = "false"
+
 
-# Define paths and models
-filename = "output_country_details.txt"  # Adjust the filename as needed
+# 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/'
-gpt2_model_name = "gpt2"  # GPT-2 model
-tokenizer = GPT2Tokenizer.from_pretrained("gpt2")
 
-# Load models
+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)
-    gpt_model = pipeline("text-generation", model=gpt2_model_name)
     print("Models loaded successfully.")
 except Exception as e:
     print(f"Failed to load models: {e}")
 
-# Load and preprocess text from the country details file
 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()]
@@ -31,61 +37,82 @@ def load_and_preprocess_text(filename):
 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:
-        query_embedding = retrieval_model.encode(user_query)
-        segment_embeddings = retrieval_model.encode(segments)
+        # 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()
-        print("Relevant segment found:", segments[best_idx])
-        return segments[best_idx]
+        return country_segments[best_idx]
     except Exception as e:
-        print(f"Error finding relevant segment: {e}")
+        print(f"Error in finding relevant segment: {e}")
         return ""
 
+
 def generate_response(user_query, relevant_segment):
+    """
+    Generate a response emphasizing the bot's capability in providing country-specific visa information.
+    """
     try:
-        # Construct the prompt with the user query
-        prompt = f"Thank you for your question! this is an additional fact about your topic: {relevant_segment}"
-        
-        # Generate response with adjusted max_length for completeness
-        max_tokens = len(tokenizer(prompt)['input_ids']) + 50
-        response = gpt_model(prompt, max_length=max_tokens, temperature=0.25)[0]['generated_text']
-        
-        # Clean and format the response
-        response_cleaned = clean_up_response(response, relevant_segment)
-        return response_cleaned
+        system_message = "You are a visa chatbot specialized in providing country-specific visa requirement information."
+        user_message = f"Here's the information on visa requirements for your query: {relevant_segment}"
+        messages = [
+            {"role": "system", "content": system_message},
+            {"role": "user", "content": user_message}
+        ]
+        response = openai.ChatCompletion.create(
+            model="gpt-4-turbo",  # Verify model name
+            messages=messages,
+            max_tokens=150,
+            temperature=0.7,
+            top_p=1,
+            frequency_penalty=0,
+            presence_penalty=0
+        )
+        return response['choices'][0]['message']['content'].strip()
     except Exception as e:
-        print(f"Error generating response: {e}")
-        return ""
+        print(f"Error in generating response: {e}")
+        return f"Error in generating response: {e}"
 
-def clean_up_response(response, segments):
-    # Split the response into sentences
-    sentences = response.split('.')
-    
-    # Remove empty sentences and any repetitive parts
-    cleaned_sentences = []
-    for sentence in sentences:
-        if sentence.strip() and sentence.strip() not in segments and sentence.strip() not in cleaned_sentences:
-            cleaned_sentences.append(sentence.strip())
-    
-    # Join the sentences back together
-    cleaned_response = '. '.join(cleaned_sentences).strip()
-    
-    # Check if the last sentence ends with a complete sentence
-    if cleaned_response and not cleaned_response.endswith((".", "!", "?")):
-        cleaned_response += "."
-    
-    return cleaned_response
 
-# Define the welcome message with markdown for formatting and larger fonts
+
+
+
+
+# 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.
-
 """
 
-# Define topics and countries with flag emojis
 topics = """
 ### Feel Free to ask me anything from the topics below!
 - Visa issuance
@@ -114,24 +141,33 @@ countries = """
 - 🇻🇳 Vietnam
 """
 
-# Define the Gradio app interface
+# Define and configure the Gradio application interface to interact with users.
 def query_model(question):
-    if question == "":  # If there's no input, the bot will display the greeting message.
+    """
+    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
 
-# Create Gradio Blocks interface for custom layout
+# Setup the Gradio Blocks interface with custom layout components
 with gr.Blocks() as demo:
-    gr.Markdown(welcome_message)  # Display the welcome message with large fonts
+    gr.Markdown(welcome_message)  # Display the formatted welcome message
     with gr.Row():
         with gr.Column():
-            gr.Markdown(topics)  # Display the topics on the left
+            gr.Markdown(topics)  # Show the topics on the left side
         with gr.Column():
-            gr.Markdown(countries)  # Display the countries with flag emojis on the right
+            gr.Markdown(countries)  # Display the list of countries on the right side
     with gr.Row():
-        img = gr.Image(os.path.join(os.getcwd(), "final.png"), width=500)  # Adjust width as needed
+        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?")
@@ -139,5 +175,6 @@ with gr.Blocks() as demo:
             submit_button = gr.Button("Submit")
             submit_button.click(fn=query_model, inputs=question, outputs=answer)
 
-# Launch the app
-demo.launch()
+# Launch the Gradio app to allow user interaction
+demo.launch(share= True)
+