How to extract text from rectangles larger than CropBox

[gregoriopellegrino]

Describe the bug

I am trying to extract text from a PDF for offset print that in the mediabox (outside the trimbox) contains text for printing worflow. I would like this extra-page information not to be extracted with the get_text, so I've set the clip argument equal to the bleed box, but the method always returns text that is outside the page as well.

To Reproduce

Sample file: p5.pdf

pdf_file = fitz.open("p5.pdf")
page = pdf_file.load_page(0)
print(page.get_text("text", clip=page.bleedbox))

Expected behavior

Extract only visible text.

Screenshots

Your configuration

• macOS 12.1, 64bit
• Python 3.9.7, 64bit
• PyMuPDF 1.19.5, 64bit

[JorjMcKie]

Both rectangles cropbox, bleedbox contain all the text!

[JorjMcKie]

To confirm draw the rectangles, set cropbox to mediabox and save to a new file.

[JorjMcKie]

No - revoking the previous! Forget it.
Text extraction does not / cannot "know" that you are handing in the /BleedBox. It has no choice but to assume that clip is a rectangle relative to the cropbox - it will not be regarded as being relative to the mediabox.
So what always works is choosing the cropbox as clip.
If you want a rectangle larger than that, you must

• either first enlarge the cropbox to that value and then use cropbox as clip again,
• or set cropbox to mediabox, and then use any rectangle you want as the clip