lib/unibits/symbolify.rb in unibits-1.1.0 vs lib/unibits/symbolify.rb in unibits-1.2.0

@@ -1,5 +1,7 @@
+require "unicode/categories"
+
 module Unibits
   module Symbolify
     ASCII_CONTROL_CODEPOINTS = "\x00-\x1F\x7F".freeze
     ASCII_CONTROL_SYMBOLS = "\u{2400}-\u{241F}\u{2421}".freeze
     ASCII_CHARS = "\x20-\x7E".freeze
@@ -72,24 +74,22 @@
       "\u{FE0D}" => "VS14",
       "\u{FE0E}" => "VS15",
       "\u{FE0F}" => "VS16",
     }.freeze
     COULD_BE_WHITESPACE = '[\p{Space}­᠎​‌‍⁠⁡⁢⁣⁤⠀𛲠𛲡𛲢𛲣𝅳𝅴𝅵𝅶𝅷𝅸𝅹𝅺]'.freeze
-    # UNASSIGNED = '\p{Cn}'.freeze
 
     def self.symbolify(char, encoding = char.encoding)
+      return "n/a" if Unicode::Categories.category(char) == "Cn"
+
       char.tr!(
         ASCII_CONTROL_CODEPOINTS.encode(encoding),
         ASCII_CONTROL_SYMBOLS.encode(encoding)
       )
       char.gsub!(
         Regexp.compile(COULD_BE_WHITESPACE.encode(encoding)),
         ']\0['.encode(encoding)
       )
-      # char.gsub!(
-      #   Regexp.compile(UNASSIGNED.encode(encoding)),
-      #   'n/a'.encode(encoding)
-      # )
+
       INTERESTING_CODEPOINTS.each{ |cp, desc|
         char.gsub! Regexp.compile(cp.encode(encoding)), desc.encode(encoding)
       }
       char.gsub! TAG_START.encode(encoding), TAG_START_SYMBOL.encode(encoding)
       char.gsub! TAG_SPACE.encode(encoding), TAG_SPACE_SYMBOL.encode(encoding)