backend/chatbot.py

import os
import pdfplumber #type:ignore 
from sentence_transformers import SentenceTransformer #type:ignore
import faiss #type:ignore
import numpy as np #type:ignore
from gpt4all import GPT4All #type:ignore

# -------------------------------
# Step 1: Load PDFs and extract text
# -------------------------------
papers_folder = "papers"  # folder where you store PDFs
pdf_texts = []

for filename in os.listdir(papers_folder):
    if filename.endswith(".pdf"):
        path = os.path.join(papers_folder, filename)
        with pdfplumber.open(path) as pdf:
            text = ""
            for page in pdf.pages:
                page_text = page.extract_text()
                if page_text:
                    text += page_text + "\n"
            pdf_texts.append(text)

print(f"Loaded {len(pdf_texts)} PDFs.")

# -------------------------------
# Step 2: Chunk texts (optional but recommended)
# -------------------------------
def chunk_text(text, chunk_size=500):
    words = text.split()
    return [" ".join(words[i:i+chunk_size]) for i in range(0, len(words), chunk_size)]

chunks = []
for text in pdf_texts:
    chunks.extend(chunk_text(text))

print(f"Total text chunks: {len(chunks)}")

# -------------------------------
# Step 3: Create embeddings with SentenceTransformers
# -------------------------------
embedding_model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
embeddings = embedding_model.encode(chunks, show_progress_bar=True)

# -------------------------------
# Step 4: Setup FAISS index
# -------------------------------
dimension = embeddings.shape[1]
index = faiss.IndexFlatL2(dimension)
index.add(np.array(embeddings))

print("FAISS index created with all chunks.")

# -------------------------------
# Step 5: Load GPT4All offline model
# -------------------------------
gpt_model = GPT4All(r"C:\Users\ACER\Desktop\Llama-3.2-3B-Instruct-Q4_0.gguf")
print("GPT4All model loaded.")

# -------------------------------
# Step 6: Function to answer questions
# -------------------------------
def ask_question(query, top_k=3):
    # Embed the query
    query_vec = embedding_model.encode([query])
    
    # Search FAISS for top_k relevant chunks
    distances, indices = index.search(np.array(query_vec), top_k)
    
    # Combine top chunks as context
    context = "\n".join([chunks[i] for i in indices[0]])
    
    prompt = f"Answer the question using ONLY the information from the following context:\n{context}\n\nQuestion: {query}\nAnswer:"
    
    # Generate response
    response = gpt_model.generate(prompt, max_tokens=200, temp=0.7)
    return response

# -------------------------------
# Step 7: Interactive loop
# -------------------------------
print("\nResearch-Paper Chatbot (type 'exit' to quit)")
while True:
    question = input("\nYour question: ")
    if question.lower() in ["exit", "quit"]:
        break
    answer = ask_question(question)
    print("\nAnswer:\n", answer)