Some Polish letters are eaten

[Danielsam66]

Is it some problem with Sumatra or other?

[GitHubRulesOK]

Commonly that is either a fail in not meeting PDF standards (Faulty PDF) OR a fail in SumatraPDF ignoring some font data, and can only be decided by decompiling the relevant parts of the PDF itself.
There are constantly different ways of writing a PDF and thus some better than others at including non Latin/American characters. Since that file looks like it should be PDF/A standard (Invoice) if meeting that standard it should work the same in all compliant readers as the Unicode mapped font should have been correctly included inside the PDF?.

SumatraPDF should have no problem with finding or showing such letters HOWEVER the use in annotation is a known current problem.

[Danielsam66]

I understand. But I also have another problem (only on one station), that copied words from Sumatra lose Polish characters when pasted. I have no idea how to get around this. From the same file on other station copy/paste is correct.

[GitHubRulesOK]

If it is per workstation difference then most likely the PDF does not carry all fonts and is asking for the fonts from that device. If so it is likely one device has the font but the other may not. Usually such fonts are named say Helvetica etc. For which windows would substitute Arial etc. Likewise Acrobat or a browser may substitute a different font name.
You may inspect the font names IF they are listed and look in the c:\windows\fonts folder

[GitHubRulesOK]

Is the file publicly suitable for posting or can you find a public one that is showing the same behaviour ?

[Danielsam66]

I'll try to generate a file without any personal data, but I'm leaving for a week's vacation in a few hours, so I'll come back to the topic later.

[GitHubRulesOK]

enjoy the break sounds like well deserved