get_texttrace returned a incorrect character bbox

[little-polka-dot]

Describe the bug (mandatory)

get_texttrace returned a incorrect character bbox

To Reproduce (mandatory)

盐城高新区投资集团有限公司2023年度第四期超短期融资券募集说明书.pdf

d = fitz.Document('temp/盐城高新区投资集团有限公司2023年度第四期超短期融资券募集说明书.pdf')
page = d[250]
for span in page.get_texttrace():
    for char in span['chars']:
        if chr(char[0]) == '民':
            print(char)

The above code will print this message:

(27665, 8775, (114.0, 634.72998046875), (114.0, 633.2604370117188, 124.44999694824219, 643.71044921875)

Origin y (634.72998046875) is too close to bbox y0 (633.2604370117188), and this is obviously not right.

Screenshot of the incorrect character bbox (red).

Screenshot code:

page.get_pixmap(matrix=fitz.Matrix(2, 2), alpha=False, clip=fitz.Rect(114.0, 633.2604370117188, 124.44999694824219, 643.71044921875)).save(f'test.png')

Your configuration (mandatory)

print(sys.version, "\n", sys.platform, "\n", fitz.doc)
3.7.9 (tags/v3.7.9:13c94747c7, Aug 17 2020, 18:58:18) [MSC v.1900 64 bit (AMD64)]
win32

PyMuPDF 1.22.5: Python bindings for the MuPDF 1.22.2 library.
Version date: 2023-06-21 00:00:01.
Built for Python 3.7 on win32 (64-bit).

[little-polka-dot]

Is this a bug or just a special case?

[JorjMcKie]

It is a bug! I am working on this already.

[little-polka-dot]

😀Haha, thank you for answer.
Looking forward to your good news.

[JorjMcKie]

Your PDF nevertheless IS a special case in the way it is made - which is the reason why this bug was revealed.
Thank you!

[JorjMcKie]

Just implemented a fix. Should roll out in one of the next versions.

[little-polka-dot]

Haha, nice😀

[JorjMcKie]

We do no longer support Python 3.7 because this has been retired end of June.
So you need to upgrade. I recommend doing this now. We will not generate wheels for Python 3.7 for the next version.

[little-polka-dot]

I see, thanks