app/vector_store.py

from PIL import Image
import pytesseract
from pdf2image import convert_from_path
import os
from langchain.vectorstores import Chroma
from langchain.embeddings import SentenceTransformerEmbeddings
from langchain.document_loaders import PyMuPDFLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.docstore.document import Document
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain.vectorstores import Chroma
from langchain.embeddings import HuggingFaceEmbeddings

CHROMA_DIR = "data/chroma_db"
CHROMA_IMG_DIR = "data/image_db"


def store_pdf(pdf_path):
    loader = PyMuPDFLoader(pdf_path)
    docs = loader.load()
    splitter = RecursiveCharacterTextSplitter(
        chunk_size=500, chunk_overlap=100)
    chunks = splitter.split_documents(docs)
    embeddings = SentenceTransformerEmbeddings(model_name='thenlper/gte-large')
    Chroma.from_documents(chunks, embeddings, persist_directory=CHROMA_DIR)


def store_pdf_image(pdf_path):
    text = extract_text_from_scanned_pdf(pdf_path)
    doc = Document(page_content=text)
    splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
    chunks = splitter.split_documents([doc])
    embeddings = SentenceTransformerEmbeddings(model_name='thenlper/gte-large')
    Chroma.from_documents(chunks, embeddings, persist_directory=CHROMA_IMG_DIR)


def store_pdf_image_text(text):
    doc = Document(page_content=text)
    splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
    chunks = splitter.split_documents([doc])
    embeddings = SentenceTransformerEmbeddings(model_name='thenlper/gte-large')
    Chroma.from_documents(chunks, embeddings, persist_directory=CHROMA_IMG_DIR)


# images = convert_from_path("your_file.pdf", poppler_path="/opt/homebrew/bin")


def extract_text_from_scanned_pdf(pdf_path):
    pages = convert_from_path(pdf_path, dpi=300)
    all_text = ""
    for i, page in enumerate(pages):
        text = pytesseract.image_to_string(page, lang="eng")
        all_text += f"\n--- Page {i+1} ---\n{text}"
    return all_text