huvimal/Vietnamese-document-intelligence-platform

1import fitz
2from PIL import Image
3import os
4
5
6def pdf_to_images(pdf_path, output_dir="temp_pages"):
7
8    os.makedirs(output_dir, exist_ok=True)
9
10    pdf_document = fitz.open(pdf_path)
11    image_paths = []
12
13    for page_number in range(len(pdf_document)):
14
15        page = pdf_document.load_page(page_number)
16        pix = page.get_pixmap()
17
18        image_path = f"{output_dir}/page_{page_number}.png"
19        pix.save(image_path)
20
21        image_paths.append(image_path)
22
23    return image_paths