[Bug]: sandwich renders differently than hocr

[femifrak]

Describe the bug

I like the hocr renderer very much as it is the only way by which you have influence on the outcome.
However, since a while it has rendered some letters wrongly, especially the "s" in old texts.

Steps to reproduce

Running these two commands:

1.) ocrmypdf -l frk --pdf-renderer hocr inmidst.pdf inmidst_hocr.pdf
2.) ocrmypdf -l frk inmidst.pdf inmidst_sandwich.pdf

leads to these two underlying texts:
1.) That's a tent with „3“ inmidnt a word.
2.) That's a test with „3“ inmidst a word.

There is no difference when using an older version of hocrtransform.py (Don't know how old it must be.)
Could you pls. have a look at this? Thanks a lot!

Files

inmidst.pdf

How did you download and install the software?

PyPI (pip, poetry, pipx, etc.)

OCRmyPDF version

15.4.2

Relevant log output

No response

[jbarlow83]

See #1194 - your issue is fixed by the new renderer, but curious if you have other input.

[femifrak]

Works great :) Thanks a lot for all your work!! Tested it with some other pdfs as well. 👍