I mean the text is already on the PDF, not embedded on images inside, so no OCR needed. If not, they could be screenshotted for an image-based service