LuaSnip eating long unicodes

[aloispichler]

Consider the snippet

    "emphasize": {
        "prefix": ["\\emph", "*"], 
        "body": "\\emph{${1:$TM_SELECTED_TEXT}}$0",
        "description": "Emphasize text"
    },

to emphasize selected text (this is an example from LaTeX).
If the selected text is 𝔼f-𝔼abc, e.g., then the result is \emph{𝔼f-𝔼a} instead of \emph{𝔼f-𝔼abc}. The reason seems to be that 𝔼 is a long unicode character (u+1d53c). As there are two 𝔼s, LuaSnip ‘eats’ 2 characters. The same happens with other long unicode characters, or with 3 long unicode characters in the selected text, etc.
It seems that LuaSnip somewhere messes the number of characters with the number of bytes.

[L3MON4D3]

Ohhhh yup, that's a bug.
I'll look into it

[L3MON4D3]

Alriiight, so multibyte characters work now :)
What doesn't is block-select with virtualedit, but I think that we can put off until there is some api-support for just retrieving the visual selection on neovims side.

[aloispichler]

I've tested it, it works well now, thank you!

[aloispichler]

Thank you again.
I have tested the behavior today with combined unicodes, as ‖͚ or ∫₀᪲ , and found the same erroneous behavior as described above.

[L3MON4D3]

Dammit :|

[L3MON4D3]

Okay, #1013 should work

[L3MON4D3]

Finally got the PR into master, we should now have correct behaviour for all kinds of weird edgecases

[aloispichler]

Seems to work indeed now. Thank you!