Code style to convert byte arrays to UTF8 strings

[davidwengier]

Relates to #58848
Part of #60582

I've never created an analyzer and fixer from scratch before, nor a code style analyzer, so please let me know where, not if, I've gone wrong.

[Youssef1313]

roslyn/src/Compilers/CSharp/Portable/SymbolDisplay/ObjectDisplay.cs

Lines 144 to 190 in 218232b

	private static bool TryReplaceChar(char c, out string replaceWith)
	{
	replaceWith = null;
	switch (c)
	{
	case '\\':
	replaceWith = "\\\\";
	break;
	case '\0':
	replaceWith = "\\0";
	break;
	case '\a':
	replaceWith = "\\a";
	break;
	case '\b':
	replaceWith = "\\b";
	break;
	case '\f':
	replaceWith = "\\f";
	break;
	case '\n':
	replaceWith = "\\n";
	break;
	case '\r':
	replaceWith = "\\r";
	break;
	case '\t':
	replaceWith = "\\t";
	break;
	case '\v':
	replaceWith = "\\v";
	break;
	}
	if (replaceWith != null)
	{
	return true;
	}
	if (NeedsEscaping(CharUnicodeInfo.GetUnicodeCategory(c)))
	{
	replaceWith = "\\u" + ((int)c).ToString("x4");
	return true;
	}
	return false;
	}

This is a similar method (not public from the compiler layer, so it cannot be reused). But it does more work:

     if (NeedsEscaping(CharUnicodeInfo.GetUnicodeCategory(c))) 
     { 
         replaceWith = "\\u" + ((int)c).ToString("x4"); 
         return true; 
     } 

So wondering if the extra work is necessary for this feature? I think the answer is no (I think difference is whether we output things like emojis as \uXXXX vs the character itself). I'm not sure so noting that you can confirm.

[CyrusNajmabadi]

i do think there is merit in using either \u or \U for things like control/formatting characters. Those will likely show up very weird once in string-literal form.

[davidwengier]

I deliberately didnt do anything (yet?) about \u because it's not that simple. My HalfEmoji test is a good example. That byte array should be "\uD83D", but whilst that is a valid string literal, "\uD83D"u8 is not a valid UTF8 string literal.

[davidwengier]

Also worth mentioning similar code to this exists in IVirtualCharService.TryGetEscapeCharacter but it was slightly painful to use. We could definitely expand that system if we wanted to (allow it to take in strings, not just syntax, and produce escaped code as well as consuming it) and use it here instead of direct Rune usage potentially

[AlekseyTs]

Changes under Compilers LGTM (commit 45)

[cston]

Compiler changes LGTM.