app.py

import os
import gradio as gr
import google.generativeai as genai
from langchain_google_genai import ChatGoogleGenerativeAI, GoogleGenerativeAIEmbeddings
from langchain_community.document_loaders import PyPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.vectorstores import FAISS
from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain
from datetime import datetime
import pytz
import time
import shutil
import numpy as np
import cv2
from PIL import Image, ImageEnhance, ImageFilter

# Get API key from Hugging Face Spaces secrets
google_api_key = os.environ.get("GOOGLE_API_KEY")

if not google_api_key:
    raise ValueError("GOOGLE_API_KEY not found in environment variables. Please add it to Hugging Face Space secrets.")

# Configure Google Generative AI
genai.configure(api_key=google_api_key)

# Function to get current date and time
def get_current_datetime():
    # Using UTC as default, but you can change to any timezone
    utc_now = datetime.now(pytz.UTC)
    
    # Convert to IST (Indian Standard Time) - modify as needed
    ist_timezone = pytz.timezone('Asia/Kolkata')
    ist_now = utc_now.astimezone(ist_timezone)
    
    # Format the datetime
    formatted_date = ist_now.strftime("%B %d, %Y")
    formatted_time = ist_now.strftime("%I:%M:%S %p")
    
    return formatted_date, formatted_time

# Load PDF and create vector store
def initialize_retriever():
    try:
        # Get current directory
        current_dir = os.getcwd()
        print(f"Current working directory: {current_dir}")
        
        # List files in current directory for debugging
        print(f"Files in directory: {os.listdir(current_dir)}")
        
        # Use absolute path for the PDF
        pdf_path = os.path.join(current_dir, "Team1.pdf")
        print(f"Attempting to load PDF from: {pdf_path}")
        
        # Check if file exists
        if not os.path.exists(pdf_path):
            raise FileNotFoundError(f"The file {pdf_path} does not exist")
            
        # Load PDF
        loader = PyPDFLoader(pdf_path)
        documents = loader.load()
        
        print(f"Successfully loaded {len(documents)} pages from the PDF")

        # Split text into chunks
        text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=150)
        text_chunks = text_splitter.split_documents(documents)
        print(f"Split into {len(text_chunks)} text chunks")

        # Generate embeddings
        embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001")

        # Store embeddings in FAISS index
        vectorstore = FAISS.from_documents(text_chunks, embeddings)
        print("Successfully created vector store")
        return vectorstore.as_retriever(search_kwargs={"k": 10})
    
    except Exception as e:
        print(f"Error in initialize_retriever: {str(e)}")
        # Return a dummy retriever for graceful failure
        class DummyRetriever:
            def get_relevant_documents(self, query):
                return []
        
        print("Returning dummy retriever due to error")
        return DummyRetriever()

# Initialize LLM
def get_llm():
    try:
        return ChatGoogleGenerativeAI(model="gemini-2.0-flash", temperature=0)
    except Exception as e:
        print(f"Error initializing LLM: {str(e)}")
        return None

llm = get_llm()

# RAG query function
def rag_query(query, retriever):
    if retriever is None:
        return "Error: Could not initialize document retriever. Please check if Team1.pdf exists."
    
    # Get current date and time for context
    current_date, current_time = get_current_datetime()
    
    try:
        # Retrieve relevant documents
        docs = retriever.get_relevant_documents(query)
        
        if not docs:
            return "No relevant information found in the document. Try a general query instead."
        
        # Create context from retrieved documents
        context = "\n".join([doc.page_content for doc in docs])
        prompt = f"""Context:\n{context}
        Current Date: {current_date}
        Current Time: {current_time}
        Question: {query}
        Answer directly and concisely, using the current date and time information if relevant:"""

        response = llm.invoke(prompt)
        return response.content
    except Exception as e:
        return f"Error in RAG processing: {str(e)}"

# General query function
def general_query(query):
    if llm is None:
        return "Error: Could not initialize language model. Please check your API key."
    
    # Get current date and time for context
    current_date, current_time = get_current_datetime()
    
    try:
        # Define the prompt with date and time context
        prompt_template = """Current Date: {date}
        Current Time: {time}
        Answer the following query, using the current date and time information if relevant: {query}"""
            
        prompt = PromptTemplate.from_template(prompt_template)
        
        # Create an LLM Chain
        chain = LLMChain(llm=llm, prompt=prompt)
        
        # Run chatbot and get response
        response = chain.run(date=current_date, time=current_time, query=query)
        return response
    
    except Exception as e:
        return f"Error in general query: {str(e)}"

# Function to make a person look younger in an image
def make_younger(input_image, youth_level=50):
    try:
        if input_image is None:
            return None, "No image uploaded. Please upload an image first."
        
        # Convert to PIL Image if necessary
        if isinstance(input_image, np.ndarray):
            input_image = Image.fromarray(input_image.astype('uint8'))
        
        # Youth level should be between 0 and 100
        youth_level = max(0, min(100, youth_level))
        
        # Apply a series of transformations to make the person look younger
        
        # 1. Smooth skin (reduce wrinkles)
        smoothing_factor = youth_level / 100
        smoothed = input_image.filter(ImageFilter.GaussianBlur(radius=smoothing_factor * 1.5))
        
        # 2. Enhance brightness slightly (younger skin tends to be brighter)
        brightness_enhancer = ImageEnhance.Brightness(smoothed)
        brightened = brightness_enhancer.enhance(1 + (smoothing_factor * 0.2))
        
        # 3. Enhance color (more vibrant)
        color_enhancer = ImageEnhance.Color(brightened)
        colored = color_enhancer.enhance(1 + (smoothing_factor * 0.3))
        
        # 4. Adjust contrast (younger skin has better contrast)
        contrast_enhancer = ImageEnhance.Contrast(colored)
        contrasted = contrast_enhancer.enhance(1 + (smoothing_factor * 0.1))
        
        # 5. Sharpen to maintain some details
        sharpened = contrasted.filter(ImageFilter.SHARPEN)
        
        return sharpened, f"Image processed successfully! Youth level applied: {youth_level}%"
    
    except Exception as e:
        return None, f"Error processing image: {str(e)}"

# Function to handle the case when no PDF is found
def file_not_found_message():
    return ("The Team1.pdf file could not be found. Team Query mode will not work properly. "
            "Please ensure the PDF is correctly uploaded to the Hugging Face Space.")

# Query router function
def query_router(query, method, retriever):
    if method == "Team Query":
        if isinstance(retriever, type) or retriever is None:
            return file_not_found_message()
        return rag_query(query, retriever)
    elif method == "General Query":
        return general_query(query)
    return "Invalid selection!"

# Function to reset input and output
def reset_query_field():
    return "", ""  # Reset only the query input

# Function to update the clock
def update_datetime():
    date, time = get_current_datetime()
    return date, time

# Main function to create and launch the Gradio interface
def main():
    # Initialize retriever
    print("Initializing retriever...")
    retriever = initialize_retriever()
    
    # Define local image paths
    logo_path = "Equinix-LOGO.jpeg"  # Ensure this file exists
        
    # Custom CSS for background styling
    custom_css = """
    .gradio-container {
        background-color: #f0f0f0;
        text-align: center;
    }
    #logo img {
        display: block;
        margin: 0 auto;
        max-width: 200px; /* Adjust size */
    }
    /* Hide download buttons and controls */
    .download-button {
        display: none !important;
    }
    /* Hide other download options */
    .file-preview .download {
        display: none !important;
    }
    /* Hide the three dots menu that might contain download options */
    .icon-button.secondary {
        display: none !important;
    }
    .tab-selected {
        background-color: #e6f7ff;
        border-bottom: 2px solid #1890ff;
    }
    """
    
    # Create the Gradio interface using Blocks
    with gr.Blocks(css=custom_css) as demo:
        gr.Image(logo_path, elem_id="logo", show_label=False, height=100, width=400, show_download_button=False)
        
        # Title & Description
        gr.Markdown("<h1 style='text-align: center; color: black;'>Equinix Chatbot for Automation Team</h1>")
        
        # Create tabs for different functionalities
        with gr.Tabs() as tabs:
            with gr.TabItem("Chat Assistant", id="chat_tab"):
                # Date and Time Display
                with gr.Row(elem_classes="datetime-display"):
                    date_display = gr.Textbox(label="Date", interactive=False)
                    time_display = gr.Textbox(label="Time", interactive=False)
                
                # Add refresh button for time
                refresh_btn = gr.Button("Update Date & Time")
                refresh_btn.click(fn=update_datetime, inputs=[], outputs=[date_display, time_display])
                
                gr.Markdown("<p style='text-align: center; color: black;'>Ask me anything!</p>")

                # Input & Dropdown Section
                with gr.Row():
                    query_input = gr.Textbox(label="Enter your query")
                    query_method = gr.Dropdown(["Team Query", "General Query"], label="Select Query Type", value="Team Query")

                # Output Textbox
                output_box = gr.Textbox(label="Response", interactive=False)
                
                # Buttons Section
                with gr.Row():
                    submit_button = gr.Button("Submit")
                    reset_button = gr.Button("Reset Query")
                
                # Button Click Events
                submit_button.click(
                    lambda query, method: query_router(query, method, retriever), 
                    inputs=[query_input, query_method], 
                    outputs=output_box
                )
                    
                # Reset only the query input
                reset_button.click(reset_query_field, inputs=[], outputs=[query_input, output_box])
               
                # Update date and time on submission
                submit_button.click(
                    fn=update_datetime,
                    inputs=[],
                    outputs=[date_display, time_display]
                )
                
                # Initialize date and time values
                date_val, time_val = get_current_datetime()
                date_display.value = date_val
                time_display.value = time_val
            
            # Add a new tab for the image age modification feature
            with gr.TabItem("Age Modification", id="age_mod_tab"):
                gr.Markdown("<h2 style='text-align: center; color: black;'>Make Person Look Younger</h2>")
                gr.Markdown("<p style='text-align: center; color: black;'>Upload an image to make the person look younger.</p>")
                
                with gr.Row():
                    input_image = gr.Image(label="Upload Image", type="pil")
                    output_image = gr.Image(label="Younger Version", show_download_button=False)
                
                with gr.Row():
                    youth_slider = gr.Slider(minimum=0, maximum=100, value=50, step=5, label="Youth Level (%)")
                    process_button = gr.Button("Make Younger")
                
                result_text = gr.Textbox(label="Processing Result", interactive=False)
                
                process_button.click(
                    fn=make_younger,
                    inputs=[input_image, youth_slider],
                    outputs=[output_image, result_text]
                )

    # Launch the interface
    demo.launch(share=True)

if __name__ == "__main__":
    main()