app.py

import os
import zipfile
import torch
import faiss
import logging
import numpy as np
import gradio as gr

from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
from sentence_transformers import SentenceTransformer
from langchain.document_loaders import TextLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import FAISS as LangChainFAISS
from langchain.docstore import InMemoryDocstore
from langchain.schema import Document
from langchain.llms import HuggingFacePipeline

# === Logger ===
logging.basicConfig(level=logging.ERROR)

# === 1. Extract the Knowledge Base ZIP ===
if os.path.exists("roosevelt_knowledge_base.zip"):
    with zipfile.ZipFile("roosevelt_knowledge_base.zip", "r") as zip_ref:
        zip_ref.extractall("roosevelt_knowledge_base")
    print("✅ Knowledge base extracted.")

# === 2. Load Markdown Files ===
KB_PATH = "/Users/toajibul/Documents/Data_Science/Amdari/Chatbot Uni/roosevelt_knowledge_base"
files = [os.path.join(dp, f) for dp, _, fn in os.walk(KB_PATH) for f in fn if f.endswith(".md")]
docs = [doc for f in files for doc in TextLoader(f, encoding="utf-8").load()]
print(f"✅ Loaded {len(docs)} documents.")

# === 3. Chunking ===
def get_dynamic_chunk_size(text):
    if len(text) < 1000:
        return 300
    elif len(text) < 5000:
        return 500
    else:
        return 1000

chunks = []
for doc in docs:
    chunk_size = get_dynamic_chunk_size(doc.page_content)
    chunk_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size, chunk_overlap=100)
    chunks.extend(chunk_splitter.split_documents([doc]))
texts = [chunk.page_content for chunk in chunks]

# === 4. Vectorstore (FAISS) ===
embed_model_id = "sentence-transformers/all-mpnet-base-v2" #"sentence-transformers/all-mpnet-base-v2"
embedder = SentenceTransformer(embed_model_id)
embeddings = embedder.encode(texts, show_progress_bar=False)

dim = embeddings.shape[1]
index = faiss.IndexFlatL2(dim)
index.add(np.array(embeddings, dtype="float32"))

docs = [Document(page_content=t) for t in texts]
docstore = InMemoryDocstore({str(i): docs[i] for i in range(len(docs))})
id_map = {i: str(i) for i in range(len(docs))}
embed_fn = HuggingFaceEmbeddings(model_name=embed_model_id)

vectorstore = LangChainFAISS(
    index=index,
    docstore=docstore,
    index_to_docstore_id=id_map,
    embedding_function=embed_fn
)

print("✅ FAISS vectorstore ready.")

# === 5. Load Falcon-e-1B-Instruct ===
print("🔁 Loading Falcon-1B model...")
model_id = "tiiuae/falcon-rw-1b"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.float16,
    device_map="auto",
    offload_folder="./offload",
    low_cpu_mem_usage=True
)


text_gen_pipeline = pipeline(
    "text-generation",
    model=model,
    tokenizer=tokenizer,
    torch_dtype=torch.float16,
    return_full_text=False,
    do_sample=False,
    max_new_tokens=200,
    pad_token_id=tokenizer.eos_token_id
)

llm = HuggingFacePipeline(pipeline=text_gen_pipeline)

# === 6. Prompt Format and Q&A ===
def truncate_context(context, max_tokens=1024):
    input_ids = tokenizer(context, truncation=True, max_length=max_tokens, return_tensors="pt").input_ids
    return tokenizer.decode(input_ids[0], skip_special_tokens=True)

def format_prompt(context, question):
    return (
        "You are the Roosevelt University Assistant.\n\n"
        "Use only the information provided in the context to answer the student's question. "
        "If the answer is not found in the context, respond: "
        "\"I'm sorry, but I don’t have that information available right now.\"\n\n"
        f"Context:\n{context.strip()}\n\n"
        f"Student Question: {question.strip()}\n"
        "Assistant Answer:"
    )

def answer_fn(question):
    if not question.strip():
        return "Please enter a valid question."
    try:
        docs = vectorstore.similarity_search(question, k=5)
        if not docs:
            return "I'm sorry, I couldn't find any relevant information for your query."
        context = "\n\n".join(d.page_content for d in docs)
        prompt = format_prompt(context, question)
        response = llm.invoke(prompt).strip()
        return response
    except Exception as e:
        logging.error(f"Error: {e}")
        return "An internal error occurred while processing your question."

# === 7. Gradio Interface ===
def chat_fn(user_message, history):
    bot_response = answer_fn(user_message)
    history = history + [(user_message, bot_response)]
    print("Returning from chat_fn:", history)
    return history, history


with gr.Blocks(theme=gr.themes.Soft()) as demo:
    with gr.Column(elem_id="main-col", scale=1):
        gr.Markdown(
            """
            <div style="text-align:center">
                <h1 style="color:#003366;">🎓 Roosevelt University Assistant</h1>
                <p style="font-size:16px; color:#444;">
                Ask questions about Roosevelt courses, admissions, student life, or tuition fees. 
                The assistant replies using only the university's official documentation.
                </p>
            </div>
            """,
            elem_id="title"
        )

        chatbot = gr.Chatbot(label="📘 Roosevelt Assistant Chat", height=400)
        state = gr.State([])
        user_input = gr.Textbox(
            placeholder="Ask a question about Roosevelt...",
            show_label=False,
            container=False
        )

        submit_btn = gr.Button("Ask")
        submit_btn.click(chat_fn, inputs=[user_input, state], outputs=[chatbot, state]).then(
            lambda: "", None, user_input  # Clear input box
        )

demo.launch()