Incorrect block detection

[mxav1111]

Description of the bug

It seems block detection has some issue in certain cases.

I am trying to extract the blocks (not text) and it seems it isn't detecting block boundary coordinates correctly.

pip list -o confirms that version is latest.

Attaching files here with :- Source pdf , Output pdf as well as code (as .txt file).

output_check_last_page_for_issue.pdf
source.pdf
block_detection_issue_on_last_page.py.txt

Thanks for your help.
-M

How to reproduce the bug

To reproduce this issue, please see attached files :- source pdf , output pdf as well as code.

To run the code rename the .txt file to python file:- block_detection_issue_on_last_page.py
and run .....

$ python block_detection_issue_on_last_page.py source.pdf output_new.pdf

PyMuPDF version

1.25.2

Operating system

Linux

Python version

3.12

[JorjMcKie]

Please be specific: against which block on which page do you have objections?
There seems to be a misconception WRT how blocks are determined:

• there is no guarantee that text block rectangles are disjoint
• it may even happen that text blocks are contained within each other (which is the case here)
• text blocks do not follow reading sequence

If you are looking for text in the right (reading) sequence, do not rely on blocks, but go down at least one hierarchy level, i.e. lines or wven text spans.

[JorjMcKie]

I am going to transfer this to Discussions. I reviewed the example file I am failing to see any issue.