Page header gets removed after applying annotations.

[varun-rf]

Hi all,

I need help with annotations. I'm trying to redact some texts using add_redact_annot. It successfully redacts the area but also removes the header image that I had added in the previous step.

Simplified version of the code I used to add header and footer images.

img_xref = 0
for page in doc:
   prect = page.rect
   if hdr_logo:
           myrect = fitz.Rect(0, 0, prect.width, h_height)
            page.set_mediabox(fitz.Rect(0, 0, prect.width, prect.height + h_height))
            img_xref = page.insert_image(myrect, stream=img, height=h_height, width=h_width, xref=img_xref)
   
   if footer_logo:
            cb = page.cropbox
            myrect = fitz.Rect(0, prect.height, prect.width, prect.height + f_height)
            page.set_mediabox(cb + fitz.Rect(0, -1*f_height, 0, 0))
            img_xref = page.insert_image(myrect, stream=img, height=f_height, width=f_width, xref=img_xref)

doc.save(pdfpath, incremental=True, deflate=True, encryption=fitz.PDF_ENCRYPT_KEEP)

Code for redaction

doc = fitz.open(tempf.filename)
pg = doc[0]
redact_coordinates = [211.4375, 174.203125, 399.4375, 204.203125]
rgb_color = {red: 9, green: 47, blue: 86}
# convering RGB to 0-1 scale
redact_color = (rgb_color['red']/255, rgb_color['green']/255, rgb_color['blue']/255)
pg.add_redact_annot(fitz.Rect(redact_coordinates[0], redact_coordinates[1], redact_coordinates[2], redact_coordinates[3]), fill=redact_color)
pg.apply_redactions()

doc.save(tempf.filename, incremental=True, deflate=True, encryption=fitz.PDF_ENCRYPT_KEEP)
doc.close()

Header coordinates: Rect(0.0, 0.0, 612.0, 107.87050359712231)
Footer coordinates: Rect(0.0, 899.8704833984375, 612.0, 974.7194042617468)

Any help will be greatly appreciated.

[JorjMcKie]

Please provide a reproducing file!
It is quite probable this is a duplicate of a known error.

[varun-rf]

Hey, thanks for replying. I have attached a sample file. Let me know if it helps.

Source File: sample_resume.pdf

File after redaction: sample_resume(Formatted).pdf

[JorjMcKie]

I didn't follow all the coordinate computations in depth, but:
If you apply redaction without parameters like you did, then into all images overlapping a redaction holes will be punched.
That may be the reason here.
Us option images=pymupdf.PDF_REDACT_IMAGE_NONE to keep images untouched

[varun-rf]

Thank you for your suggestion. I had already tried using the images=pymupdf.PDF_REDACT_IMAGE_NONE option before posting here, but the issue still persisted. Additionally, I'm not sure how the header image could overlap with the redaction area, as both have different coordinates.

[JorjMcKie]

Can you please provide me with a file

1. containing these images
2. your redactions added - but not applied

So that example file should show the header / footer images and a number of red, crossed-out rectangles at places where you added redactions.

[varun-rf]

I've encountered an issue with a specific header image—it creates an empty space at the top but doesn't display. This isn't related to redaction; I apologize for the confusion. Any idea why this particular image isn't showing up? Thanks for your help!

Header Image - https://drive.google.com/file/d/1uFezFAMNCYUdZIZQdNHbkL9sx2wcUhaK/view?usp=sharing

[JorjMcKie]

Ok, this is a moving target then: so we are not dealing with redaction problems.
When using page.insert_image you do a lot of things that I find hard to understand: why do you modify the page's MediaBox at all? This has all sorts of consequences that can easily cause things to become invisible?

What is the purpose of setting width and height in insert_image - why don't you start with using the default parameters?

[JorjMcKie]

Just some side remarks:

• You can create PyMuPDF's geometry objects also by directly passing in a list / tuple with the right length. So in your example above you could create a Rect directly via fitz.Rect(redact_coordinates). No need to name all 4 number separately as you did.
• In all functions with geometry objects as parameters, you can directly use the corresponding list / tuple with the correct length. So for instance you can add a redaction annotation without first creating a Rect like this: add_redact_annot(redact_coordinates, ...). All methods understand this and internally do the right things.

This will save you a lot of coding effort!

[varun-rf]

Thank you for sharing this. I'll keep this in mind.