codepoints = {}

IO.readlines('UnicodeData.txt').each do |l|
  a = l.chomp.split ';'
  codepoints[a.first] = a.drop 1
end


((0x0000..0xD7FF).to_a+(0xE000..0x10FFFF).to_a).each do |i|
  c = i.chr 'UTF-8'
  if c =~ /\p{Age=12.0}/ and c !~ /\p{Age=11.0}/
    catA = c =~ /\p{Ll}|\p{Lu}|\p{Lo}|\p{Nd}|\p{Lm}|\p{Mn}|\p{Mc}/  # LetterDigits (A)
    catB = c != c.unicode_normalize(:nfkc).downcase(:fold).unicode_normalize(:nfkc) # Unstable (B)
    catC = c =~ /p{Default_Ignorable_Code_Point}|p{White_Space}|p{Noncharacter_Code_Point}/ # IgnorableProperties (C)
    catD = c =~ /p{In_Combining_Diacritical_Marks_for_Symbols}|p{In_Musical_Symbols}|p{In_Ancient_Greek_Musical_Notation}/ # IgnorableBlocks (D)
    # LDH (E): This is fixed and so can be ignored
    # Exceptions (F): There are no exceptions in new codepoints yet, so this can be ignored
    # BackwardCompatible (G): Empty, so ignorable
    catH = c =~ /p{Join_Control}/ # JoinControl (H)
    catI = c =~ /p{Grapheme_Cluster_Break=L}|\p{Grapheme_Cluster_Break=V}|\p{Grapheme_Cluster_Break=T}/ # OldHangulJamo (I), not completely sure Grapheme_Cluster_Break is right, but should be
    # Unassigned (J): not possible if we check for Age
    s = i.to_s(16).upcase.rjust 4, '0'
    print s
    print ' ', if catH then 'CONTEXTJ'
               elsif catB or catC or catD or catI then 'DISALLAWED'
               elsif catA then 'PVALID'
               else 'DISALLOWED'
               end
    v = codepoints[s]
    print ' ', v.first if v
    puts
  end
end