[bug] logging line reduced with ... still wrapped

[mo-han]

Although these lines are already reduced by replacing the middle part with ellipsis ("..."), they are still wrapped by the terminal.
I'm not sure but maybe it's caused by some full-width chars 〇〇, which make the line occupying more length than what len() tells.

If my guess was correct, then to fix it we need to re-calculate the "visual length" of the text string, e.g. via a function like this:

def visual_len(s: str):
    eaw = unicodedata.east_asian_width
    n = len(s)
    n += sum([1 for c in s if eaw(c) == 'W'])
    return n

[mikf]

I'm not sure but maybe it's caused by some full-width chars 〇〇, which make the line occupying more length than what len() tells.

That's exactly the problem here. Happens for all paths containing characters wider than a regular ASCII character.

I've known about this for a long time, but never found a way to compute the display size / width of a string or character until you mentioned unicodedata.east_asian_width. Thank you for that!

The next problem is cmd.exe (maybe Powershell as well?), which usually can't display Japanese / East Asian characters and replaces them with a ?. At least on Windows 7 this is only 1 character wide instead of the expected 2, resulting in shortened lines that only span half the terminal: ??????? foo... ??????????

[mo-han]

which usually can't display Japanese / East Asian characters and replaces them with a `?``

If it's me, I'll just leave them as they are, right now can't figure out a soultion for that.

BTW, in my env those chars are usually displayed as blocks □ - which can still be selected to copy / paste the correct text - or, as question marks. Situation vary with different fonts for cmd.exe, or when something like cygwin is involved.

[mikf]

This was fixed in bd84530 by implementing the ability to set output.shorten to "eaw" (East Asian Width).

It's not enabled by default since it produces the wrong results without the needed fonts being available and it is at least an order of magnitude slower than the other, simpler algorithm.