Regex use ucp flag because they use unicode flag (#357)

Author: inoas

CHANGELOG.md

@@ -2,6 +2,9 @@
 
 ## v0.24.1 - unreleased
 
+- For `regexp.compile` unicode character properties are now used when
+  resolving `\B`, `\b`, `\D`, `\d`, `\S`, `\s`, `\W`, and `\w` on target
+  Erlang.
 - `list.sort` is now tail recursive and will no longer exceed the stack size
   on large inputs on target JavaScript.
 - `list.sort` is now a "stable" sort, meaning equal elements are sorted in
@@ -11,7 +14,7 @@
 - Fixed a bug where `regex.scan` would not work correctly on utf8.
 - The performance of `list.flatten` has been greatly improved.
 - The `string_builder` module gains the `join` function.
-- The `list` module gains the  `shuffle` function.
+- The `list` module gains the `shuffle` function.
 
 ## v0.24.0 - 2022-10-15
 

src/gleam_stdlib.erl

@@ -184,6 +184,7 @@ compile_regex(String, Options) ->
     {options, Caseless, Multiline} = Options,
     OptionsList = [
         unicode,
+        ucp,
         Caseless andalso caseless,
         Multiline andalso multiline
     ],

test/gleam/regex_test.gleam

@@ -26,6 +26,12 @@ pub fn compile_test() {
 
   regex.check(re, "abc\n123")
   |> should.be_true
+
+  // For Erlang: This test will only passes if unicode and ucp flags are set
+  assert Ok(re) = regex.compile("\\s", options)
+  // Em space == U+2003 == " " == used below
+  regex.check(re, " ")
+  |> should.be_true
 }
 
 pub fn check_test() {