model.py

import os
import tempfile
import PyPDF2
from langchain.vectorstores import FAISS
from langchain_huggingface import HuggingFaceEmbeddings
from langchain.text_splitter import CharacterTextSplitter
from langchain.docstore.document import Document
from langchain.chains import RetrievalQA
from langchain_huggingface import HuggingFaceEndpoint

# Use /tmp for cache
CACHE_DIR = tempfile.gettempdir()
os.environ["TRANSFORMERS_CACHE"] = CACHE_DIR
os.environ["HF_HOME"] = CACHE_DIR

DATA_PATH = "/app/data"
VECTORSTORE_PATH = "/app/vectorstore"
EMBEDDING_MODEL_NAME = "sentence-transformers/paraphrase-MiniLM-L6-v2"

def load_embedding_model():
    """Load sentence transformer embeddings."""
    return HuggingFaceEmbeddings(model_name=EMBEDDING_MODEL_NAME)

def load_documents(pdf_path):
    """Extract text from PDF and split into documents."""
    try:
        with open(pdf_path, "rb") as f:
            pdf = PyPDF2.PdfReader(f)
            text = "".join(page.extract_text() or "" for page in pdf.pages)
        if not text.strip():
            raise ValueError("No text extracted from PDF")
        
        splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=100)
        docs = splitter.create_documents([text])
        return docs
    except Exception as e:
        raise ValueError(f"Failed to process PDF: {str(e)}")

def load_vectorstore(pdf_path):
    """Load or create FAISS vector store from PDF."""
    vectorstore_file = os.path.join(VECTORSTORE_PATH, "faiss_index")
    embedding_model = load_embedding_model()
    
    if os.path.exists(vectorstore_file):
        try:
            return FAISS.load_local(vectorstore_file, embedding_model, allow_dangerous_deserialization=True)
        except:
            pass  # Rebuild if loading fails
    
    docs = load_documents(pdf_path)
    vectorstore = FAISS.from_documents(docs, embedding_model)
    vectorstore.save_local(vectorstore_file)
    return vectorstore

def ask_question(query, pdf_path):
    """Run RAG query and return answer with contexts."""
    api_key = os.getenv("HUGGINGFACEHUB_API_TOKEN")
    if not api_key:
        raise ValueError("HUGGINGFACEHUB_API_TOKEN not set")
    
    vectorstore = load_vectorstore(pdf_path)
    llm = HuggingFaceEndpoint(
        repo_id="mistralai/Mistral-7B-Instruct-v0.2",
        huggingfacehub_api_token=api_key,
        temperature=0.5,
        max_new_tokens=256
    )
    
    qa = RetrievalQA.from_chain_type(
        llm=llm,
        retriever=vectorstore.as_retriever(search_kwargs={"k": 3}),
        return_source_documents=True
    )
    
    result = qa({"query": query})
    return {
        "answer": result["result"],
        "contexts": [doc.page_content for doc in result["source_documents"]]
    }