Sanitize a string using [unicode_set](https://hex.pm/packages/unicode_set)

sanitize_string.ex

defmodule Sanitize do
  # Unicode sets are defined at https://unicode-org.github.io/icu/userguide/strings/unicodeset.html
  require Unicode.Set
  
  # Defines a guard that is the intersection of alphanumerics and the latin script plus the 
  # space and underscore characters. Note that the set is resolved at compile time into an
  # integer expression and is therefore acceptably performant at runtime.
  defguard latin_alphanum(c) when Unicode.Set.match?(c, "[[:Alnum:]&[:script=Latin:][_\\ ]]")
  
  def sanitize_string(<<"">>), do: ""
  def sanitize_string(<<c::utf8, rest::binary>>) when latin_alphanum(c), do: <<c::utf8, sanitize_string(rest)::binary>>
  def sanitize_string(<<_c::utf8, rest::binary>>), do: sanitize_string(rest)
end

Example

iex> Sanitize.sanitize_string("this is a ๓ thai _char !!!")
"this is a  thai _char "