Better Unicode detection

[asmeurer]

My terminal supports Unicode, but pyinstrument doesn't use it.

SymPy's Unicode detection seems to be pretty robust (there are tons of users, and the only error reports about misusing Unicode when it isn't supported that I notice are from non-Terminals). It seems to be a combination of doing a unicodedata lookup and trying sys.stdout.encode. I also don't know if it makes sense to restrict unicode output to ttys. That generally only matters for colors, because the ANSI escape sequences might not be interpreted otherwise, and will just clutter the output.

The code's all BSD, so feel free to reuse it.

[joerick]

I've made some changes to the unicode detection, the code is on my develop branch. It should work better than the previous detection mechanism.

Does this work at your end?

[asmeurer]

Yes, it uses Unicode for me now.

[asmeurer]

And I just tried it in a terminal that only supports ASCII and it correctly doesn't use Unicode.