Pasting 🤦🏼‍♂️character into repl throws exception and exits repl

[MikeSchurman]

Version

0.6.5

What happened?

I tried pasting the 🤦🏼‍♂️character into the repl and it throws an exception:

> Unhandled exception. System.NullReferenceException: Object reference not set to an instance of an object.──────────────┬──────────────────┐
 at PrettyPrompt.Rendering.Screen.PositionCursor(Screen screen, ConsoleCoordinate cursor)
 at PrettyPrompt.Rendering.Screen..ctor(Int32 width, Int32 height, ConsoleCoordinate cursor, ScreenArea[] screenAreas) │
 at PrettyPrompt.Renderer.<RenderOutput>g__Redraw|7_0(<>c__DisplayClass7_0&)
 at PrettyPrompt.Renderer.RenderOutput(PromptResult result, CodePane codePane, OverloadPane overloadPane, CompletionPane completionPane, IReadOnlyCollection`1 highlights, KeyPress key)│ AppContext │
 at PrettyPrompt.Prompt.RenderSyntaxHighlightedOutput(Renderer renderer, CodePane codePane, OverloadPane overloadPane, CompletionPane completionPane, KeyPress key, String inputText, PromptResult result)─────────┘
 at PrettyPrompt.Prompt.ReadLineAsync()
 at CSharpRepl.ReadEvalPrintLoop.RunAsync(Configuration config) in D:\a\CSharpRepl\CSharpRepl\CSharpRepl\ReadEvalPrintLoop.cs:line 49
 at CSharpRepl.Program.Main(String[] args) in D:\a\CSharpRepl\CSharpRepl\CSharpRepl\Program.cs:line 68
 at CSharpRepl.Program.Main(String[] args) in D:\a\CSharpRepl\CSharpRepl\CSharpRepl\Program.cs:line 74
 at CSharpRepl.Program.<Main>(String[] args)

[waf]

Thanks for reporting this! It was definitely working before, and I thought we even had good integration tests around this. I'll see what happened.

[waf]

Just a note for myself, 🤦🏼‍♂️ is composed of the following codepoints:

Render	Codepoint	Name
🤦	U+1F926	FACE PALM
🏼	U+1F3FC	EMOJI MODIFIER FITZPATRICK TYPE-3
‍	U+200D	ZERO WIDTH JOINER
♂	U+2642	MALE SIGN
	U+FE0F	VARIATION SELECTOR-16

I'm transferring this to the PrettyPrompt underlying library repository where it should be fixed.

[tats-u]

• ⚠️ (U+26A0 U+FE0F) → No crash
• 🤦 (U+1F926) → No crash
• 🤦🏼 (U+1F926 U+1F3FC) → CRASH
• 🤦‍♂ (U+1F926 U+200D U+2642) → CRASH
• 🤦️️ (U+1F926 U+FE0F) → No crash
• 𠮟 (U+20B9F; means "scold") → No crash
• 禰 (U+79B0) → No crash
• 禰󠄀 (U+79B0 U+E0100; used in the name of the sister of the main character of Daemon Slayer: Kimetsu no Yaiba) → CRASH
• 侮︀ (U+4FAE U+FE00) → No crash
• 👩‍💻 (U+1F469 U+200D U+1F4BB) → No crash but not joinned

[tats-u]

PrettyPrompt/src/PrettyPrompt/Rendering/Screen.cs

Line 100 in 4245c19

Debug.Assert(previousCell.ElementWidth == 2);

This assertion fails.

PrettyPrompt/src/PrettyPrompt/Rendering/UnicodeWidth.cs

Line 244 in 4245c19

public static int GetWidth(ReadOnlySpan<char> text)

This returns more than two and the returned value is set to .ElementWidth.

UnicodeWidth does not consider 🂡🂢🂣 or 🀀🀁, which are in the SMP (not BMP) and whose width should be treated as 1.

No Rune available only in .NET is used there. We can use StringInfo.GetTextElementEnumerator to split string into grapheme clusters now.

[waf]

I've released a version that improves this by quite a lot. There are still a few gremlins to sort out, but I think the latest release's handling is much more correct:

It looks like there may still be some cursor navigation issues on ⚠️, but I should be able to figure that out shortly.

[tats-u]

Great news, but U+FF9E–U+FE9F in "ﾊﾟｸﾞ" (U+FF8A U+FF9F U+FF78 U+FF9E) (= パグ; means pug dog in Japanese) seem not to be counted as 1 character width respectively. They are grapheme extenders but letters (Lm). This word should be counted as 4 character width, not 2.

microsoft/terminal#18087