Windows: try to be smarter about encoding detection

[bpasero]

Instead of hardcoding the encoding to cp850, should actually check the output of running chcp and find some well supported encodings we could use instead.

Refs: https://ss64.com/nt/chcp.html

[roblourens]

Verification steps? Is this for the integrated terminal?

[bpasero]

@roblourens this is for #39330

[roblourens]

I don't think I'm doing it right, here's what I've tried

• chcp 866 for Russian
• echo 'глаголь добро есть' | code-insiders
• Change encoding in vscode to CP 866
• Result is random characters

...

• Have a file encoded as CP 866 with russian text
• type russian.txt | code-insiders
• Change encoding in vscode to CP 866
• Result is random characters

Same with Greek, 1253

[bpasero]

@roblourens we are not carrying over the encoding into VS Code, so the file in VS Code will be whatever your default encoding is (most likely UTF-8). If you see the characters you type showing up properly in VS Code, you have verified this bug.

[roblourens]

I fix the encoding in vscode manually, but even then, the characters are not correct.

[bpasero]

@roblourens I see. I think the issue here is that you are only changing the encoding for the current running terminal session and when VS Code starts, I run chcp again but in a different terminal. I think this only works if the terminal encoding is set more permanently and not per the current session.

[roblourens]

You're right, it works great after changing my system locale to Russian.

[bpasero]

@roblourens thanks for testing this!

[heroboy]

@bpasero please add 936:"cp936" for Simplified Chinese.
I modified cli.js, and tested dir|node, it works.

[bpasero]

@heroboy ok I can add it.