import streamlit as st
import faiss
import numpy as np
from sentence_transformers import SentenceTransformer

# Load resume data
resume_data = {
    "name": "Pradeep Singh Sengar",
    "linkedin": "www.linkedin.com/in/ipradeepsengarr",
    "email": "pradeep19sengar@gmail.com",
    "github": "github.com/pradeepsengar",
    "mobile": "+91-7898367211",
    "education": "Bachelor of Engineering (Hons.) - Information Technology; CGPA: 8.31 (Oriental College Of Technology, Bhopal, 2019-2023)",
    "skills": "Python, HTML/CSS, Django, Reactjs, Node.js, Git, Web Scraping, Generative AI, Machine Learning (LLM)",
    "experience": "Graduate Engineer Trainee at Jio Platform Limited (Dec. 2023 - Present). Implemented chatbots with Docker, used Git/GitHub, worked with LLM concepts and Hugging Face.",
    "projects": "Room Rental System, Text to Image Generator, Fitness Tracker, Movie Recommendation System",
    "honors_awards": "Qualified for Round 1B of SnackDown (CodeChef), Startup Challenge (Top 10 teams)",
    "certifications": "Web Development (Internshala), The Complete Python Pro Bootcamp (Udemy), Data Science (LinkedIn Learning), Web Scraping (LinkedIn Learning)"
}

# Convert data to list of sentences for retrieval
resume_keys = list(resume_data.keys())
resume_values = list(resume_data.values())

# Load embedding model
model = SentenceTransformer('all-MiniLM-L6-v2')
embeddings = model.encode(resume_values)

# Store embeddings in FAISS index
index = faiss.IndexFlatL2(embeddings.shape[1])
index.add(np.array(embeddings))

def get_response(query):
    query_embedding = model.encode([query])
    D, I = index.search(query_embedding, 1)
    confidence = 1 - D[0][0] / 10  # Normalize confidence score
    
    # Keyword-based accuracy improvement
    query_lower = query.lower()
    for key, value in resume_data.items():
        if key in query_lower or any(word in query_lower for word in key.split("_")):
            return f"**{key.capitalize()}**: {value}"
    
    if confidence > 0.5:
        key = resume_keys[I[0][0]]
        answer = resume_data[key]
        return f"**{key.capitalize()}**: {answer}"
    else:
        return "I'm not sure about this. Please try asking differently or be more specific."

# Streamlit UI
st.title("📝 Resume Chatbot")
st.write("Ask anything about Pradeep's resume!")

# Suggested Questions
suggested_questions = [
    "What is your name?",
    "What is your LinkedIn profile?",
    "What skills do you have?",
    "Tell me about your experience?",
    "What are your certifications?",
    "List your projects.",
    "What is your mobile number?"
]

st.write("### Quick Questions")
col1, col2 = st.columns(2)

# Ensure session state exists
if "user_input" not in st.session_state:
    st.session_state["user_input"] = ""

def set_question(question):
    st.session_state["user_input"] = question
    st.rerun()

for i, question in enumerate(suggested_questions):
    if i % 2 == 0:
        with col1:
            st.button(question, key=f"btn_{i}", on_click=set_question, args=(question,))
    else:
        with col2:
            st.button(question, key=f"btn_{i}", on_click=set_question, args=(question,))

# User Input & Response
user_input = st.text_input("Your question:", value=st.session_state["user_input"])
if user_input:
    response = get_response(user_input)
    st.success(response)