page.add_redact_annot(rect, fill=False) still has white cover. how can i delete it?

[Bkhxq]

doc = pymupdf.open("D:\1.pdf")
page = doc[0]
page.add_redact_annot(quad=pymupdf.Rect([560, 225, 740, 570]), fill=False)
page.apply_redactions(images=2, graphics=1)
doc.save("d:\x.pdf")

i want to remove all elements in the red circle. it does well. but a cover ([560, 225, 740, 570]) leaves here. How can i delete this cover? tks.

1.pdf
x.pdf

[JorjMcKie]

This is not an issue, but a Discussions item.

[JorjMcKie]

Your page is a wild mixture of images and graphics and text overlapping each other in different constellation.
Draw the image and graphics rectangles in different colors to understand the situation:

for p in page.get_drawings():
    page.draw_rect(p["rect"], color=(1, 0, 0))
for img in page.get_image_info():
    page.draw_rect(img["bbox"], color=(0, 1, 0))

Then decide whether you have a chance at all - at least using redactions.
But for me looking at the situation, this is a hopeless case.

[Bkhxq]

i have lots of more complex pages to deal with. it's disappointing. why it depends on complexity to do redactions?

[JorjMcKie]

The problem is your misconception of what a "redaction" is!

By PDF specification, a redaction must erase everything underneath its rectangle - independent from how many content objects overlap each other in that area.
It is a concept invented by PDF specification and also has legal implications.
So by strict definition, redactions do not even allow you to make exceptions there.
In PyMuPDF we nevertheless have introduced some selectiveness by enabling to differentiate between content classes text, images and vectors.
But still: if e.g. two vectors overlap in the redaction rectangle, they will (and must be!) treated in the same way.

[Bkhxq]

thanks for your explanation. my question is why there is a white rectangle on top, even i set reaction fill transparent. i want to convert page to picture, but how can i remove it.

[JorjMcKie]

Why didn't you follow my suggestion and first look at how the different content categories are overlapping?
From there you can derive what is possible with redactions and what is not.

Images=green, vectors=red, your rectangle=blue.

Your white rectangle is an image ... which you can remove.
But not without also removing (or damaging) all the other images (green objects) overlapping into the blue rectangle.

[Bkhxq]

this is not a image. i also add another redaction, and modify first one. you can see both are not images:

doc = pymupdf.open("D:\1.pdf")
page = doc[0]
for p in page.get_drawings():
page.draw_rect(p["rect"], color=(1, 0, 0))
for img in page.get_image_info():
page.draw_rect(img["bbox"], color=(0, 1, 0))
page.add_redact_annot(quad=pymupdf.Rect([560, 225, 740, 570]), fill=False)
page.apply_redactions(images=2, graphics=1)
page.add_redact_annot(quad=pymupdf.Rect([0, 0, 500, 500]), fill=False)
page.apply_redactions(images=2, graphics=1)
doc.save("d:\x.pdf")

[jamie-lemon]

@Bkhxq If required please continue this discussion on our forum at https://forum.pymupdf.com , we are trying to move all discussions there! :)