tm or cm matrix with a visitor for 'extract_text'?

[Apfelkuchenbemme]

I was testing this example from the documentation, but print(text_body) printed an empty string. I then changed the visitor function to this:

def visitor_body(text, cm, tm, font_dict, font_size):
    #y = cm[5]
    y = tm[5]
    print(cm, tm)
    if 50 < y < 720:
        parts.append(text)

and ran it again.This produced the expected outcome, i.e. that print(text_body) printed the text of the 4th page of the PDF without header and footer. All printed cm matrices looked like this: [1.0, 0.0, 0.0, 1.0, 0.0, 0.0]. I didn't look under the hood of pypdf itself to check where this might be used "incorrectly", but I think it's the same problem with the second example on that page.

See the SVG files created from the code of the 2nd example, that I have attached here. The first one uses the cm matrix in visitor_svg_text, and puts all of the text in the top left corner. The second one uses the tm matrix and while it won't win beauty contests, it looks a lot better:

Environment

Python 3.10.0, pypdf 5.7.0, Windows 10

Code + PDF

See the links to the examples in the documentation above.

Traceback

N/A

[stefan6419846]

See #2431 and the related PR #2432 where the discussions have not yet been resolved due to my knowledge being limited. TL;DR: Ideally, use mult(cm, tm).