Redaction is not simple. What should the software do if the user's black box is drawn over an image, some text and part of a spline on a page? Modify the image

Redaction is not simple. What should the software do if the user's black box is drawn over an image, some text and part of a spline on a page? Modify the image and recompress it even if it's lossy?, find the text run and remove the obscured characters (what about a half covered letter?), somehow modify spline and possibly replace it with several smaller ones if the redaction is in the middle of it? All while taking into account any transforms that have been applied to the page. Also what if the black box is over part of a pdf form? Is it just the appearance that gets redacted or the form data (which is separate). Same questions if the black box is drawn over annotations. And what about other page content like embedded videos or 3d content, good luck redacting those. But if you only redact some things the user might then justifiably expect it to redact everything.