Str.graphemes incorrectly groups grapheme clusters

roc-lang / roc

A fast, friendly, functional language.

https://roc-lang.org

Universal Permissive License v1.0

4.47k stars 315 forks source link

Str.graphemes incorrectly groups grapheme clusters #4779

Open rtfeldman opened 1 year ago

rtfeldman commented 1 year ago

To reproduce:

» "👩‍👩‍👦‍👦" |> Str.toScalars

[128105, 8205, 128105, 8205, 128102, 8205, 128102] : List U32

» "👩‍👩‍👦‍👦" |> Str.graphemes
["👩‍", "👩‍", "👦‍", "👦"] : List Str

The toScalars part is correct, but toGraphemes is incorrect. It should return ["👩‍👩‍👦‍👦"] : List Str - just one element in the list.

Swift does this correctly.

yawaramin commented 11 months ago

OCaml REPL session using the uuseg (linked to source code) library:

# #require "uuseg.string";;
# List.rev (Uuseg_string.fold_utf_8 `Grapheme_cluster (fun list segment -> segment :: list) [] "👩‍👩‍👦‍👦");;
- : string list = ["👩‍👩‍👦‍👦"]

lukewilliamboswell commented 11 months ago

For some additional context to this issue, the plan is to remove unicode text segmentation from builtins and move to a library over at roc-lang/unicode. That is a work in progress.