Python: replace inplace proprietary format encoding only for characters recognised and ignoring the rest

Question

I am having trouble with a proprietary format file. My final goal is to read the file, substitute inplace some id number that are clearly displayed, and save the file. I can read almost everything using "utf-16le" decoding, but I have some troublesome characters mixed in. Luckily, I don't care about them because the id numbers I have to changed are correctly displayed.

Here it comes the issue:

If I use this code:

    with open(r'filepath', 'r', encoding='UTF-16LE') as file:
                content = file.read()
                content = content.replace(search,replace)
    with open(fr'filepath', 'w', encoding='UTF-16LE') as file:
               file.write(content)

The variable content is copying the unrecognised characters wrongly, so it is not recognised back by the proprietary software.

I manually tried to open my Notepad, to change the id numbers and to upload the file on the software and it works. So, I understood that I have to do it inplace, without touching anything else on the file.

Probably fileinput.FileInput() could help me, but I can't go any further when one of the characters is not recognised because it raise and error.

Can you please help me to identify which library and function to use?