utils.py

import os
from langchain_community.document_loaders import PyPDFLoader
from langchain_community.vectorstores import FAISS
from langchain_google_genai import GoogleGenerativeAIEmbeddings

VECTORSTORE_PATH = "vectorstore"

def load_documents(file_paths):
    """Load PDFs and return a list of documents with metadata."""
    all_docs = []
    for path in file_paths:
        loader = PyPDFLoader(path)
        docs = loader.load()
        # Add page/file metadata
        for i, doc in enumerate(docs):
            doc.metadata["source_page"] = i + 1
            doc.metadata["source_file"] = os.path.basename(path)
        all_docs.extend(docs)
    return all_docs

def build_vectorstore(docs, persist_path=VECTORSTORE_PATH):
    """Build FAISS vectorstore from documents."""
    embeddings = GoogleGenerativeAIEmbeddings(
        model="models/embedding-001"
    )
    db = FAISS.from_documents(docs, embeddings)
    os.makedirs(persist_path, exist_ok=True)
    db.save_local(persist_path)
    return db

def load_vectorstore(persist_path=VECTORSTORE_PATH):
    """Load existing FAISS vectorstore safely."""
    embeddings = GoogleGenerativeAIEmbeddings(
        model="models/embedding-001"
    )
    return FAISS.load_local(
        persist_path,
        embeddings,
        allow_dangerous_deserialization=True
    )