Fix support for nested styles

[farnabaz]

As I recovered both bold and dim style has same closing chars and this cause the problem

bold: { open: '\u001b[1m', close: '\u001b[22m' }
dim: { open: '\u001b[2m', close: '\u001b[22m' }

On line 165 all closing chars on current style replace by opening chars, so in this specific situation closing tag of dim replaced by bold opening tag and bar seen as dim

chalk/index.js

Line 165 in 7b9211b

string = code.open + string.replace(code.closeRe, code.open) + code.close;

Fixes #334

[Qix-]

Unfortunately this doesn't solve it. This is intentional. The history of escape codes is annoying and unclear, but 22m is the only widely compatible way to close either bold or dim. 21m is not used as frequently since it has been confused as a double-underline, which isn't commonly supported in rendering libraries and thus is usually just silently ignored.

Also, "dim" is usually the analog of "bold", so having both modes enabled at once wouldn't make much sense under such an interpretation, hence most implementations keeping it simply 22m.

I wish it was this easy.

[farnabaz]

Thank you @Qix- for information. I didn't know much about escape code, I've checked ansi-colors and compared its result with chalk:

'\u001b[1mfoo \u001b[31m\u001b[2mbar\u001b[22m\u001b[1m\u001b[39m baz\u001b[22m' ansi-colors
'\u001b[1mfoo \u001b[31m\u001b[2mbar          \u001b[1m\u001b[39m baz\u001b[22m' chalk

As you can see the only difference between chalk and ansi-colors is 22m that closes dim, So I thought I found problem, I've change line 165 to prevent removing close char of dim and it works.

Also, all test passed except the one that compares the nesting style result.
Here is result after this change:

[mirefly]

This is how ansi-colors solve the nested styling bug. To me, it looks like a good way to do it without introducing a global.

[Qix-]

Yeah, this seems right. @sindresorhus at some point we need to condense codes down into a single CSI character and not repeat it because that output is ugly (not your problem @farnabaz).

[sindresorhus]

@mirefly Can you fix the merge conflict?

[sindresorhus]

Can you also try to write a more descriptive PR title of what this fixes?

[mirefly]

@sindresorhus I guess you mean the author @farnabaz ?

[sindresorhus]

Yes, sorry.

[farnabaz]

@sindresorhus Done