1 2BEGIN { 3 if ($ENV{PERL_CORE}) { 4 chdir('t') if -d 't'; 5 @INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib); 6 } 7} 8 9use strict; 10use warnings; 11BEGIN { $| = 1; print "1..1070\n"; } # 6 + 56 x @Versions 12my $count = 0; 13sub ok ($;$) { 14 my $p = my $r = shift; 15 if (@_) { 16 my $x = shift; 17 $p = !defined $x ? !defined $r : !defined $r ? 0 : $r eq $x; 18 } 19 print $p ? "ok" : "not ok", ' ', ++$count, "\n"; 20} 21 22use Unicode::Collate; 23 24ok(1); 25 26sub _pack_U { Unicode::Collate::pack_U(@_) } 27sub _unpack_U { Unicode::Collate::unpack_U(@_) } 28 29######################### 30 31# 2..6 32 33my $ignoreCJK = Unicode::Collate->new( 34 table => undef, 35 normalization => undef, 36 overrideCJK => sub {()}, 37 entry => <<'ENTRIES', 385B57 ; [.0107.0020.0002.5B57] # CJK Ideograph "Letter" 39ENTRIES 40); 41 42# All CJK Unified Ideographs except U+5B57 are ignored. 43 44ok($ignoreCJK->eq("\x{4E00}", "")); 45ok($ignoreCJK->lt("\x{4E00}", "\0")); 46ok($ignoreCJK->eq("Pe\x{4E00}rl", "Perl")); # U+4E00 is a CJK. 47ok($ignoreCJK->gt("\x{4DFF}", "\x{4E00}")); # U+4DFF is not CJK. 48ok($ignoreCJK->lt("Pe\x{5B57}rl", "Perl")); # 'r' is unassigned. 49 50##### 51 52# 4E00..9FA5 are CJK UI. 53# 9FA6..9FBB are CJK UI since UCA_Version 14 (Unicode 4.1). 54# 9FBC..9FC3 are CJK UI since UCA_Version 18 (Unicode 5.1). 55# 9FC4..9FCB are CJK UI since UCA_Version 20 (Unicode 5.2). 56# 9FCC is CJK UI since UCA_Version 24 (Unicode 6.1). 57# 9FCD..9FD5 are CJK UI since UCA_Version 32 (Unicode 8.0). 58# 9FD6..9FEA are CJK UI since UCA_Version 36 (Unicode 10.0). 59# 9FEB..9FEF are CJK UI since UCA_Version 38 (Unicode 11.0). 60# 9FF0..9FFC are CJK UI since UCA_Version 43 (Unicode 13.0). 61 62# 3400..4DB5 are CJK UI Ext.A since UCA_Version 8 (Unicode 3.0). 63# 4DB6..4DBF are CJK UI Ext.A since UCA_Version 43 (Unicode 13.0). 64# 20000..2A6D6 are CJK UI Ext.B since UCA_Version 8 (Unicode 3.1). 65# 2A6D7..2A6DD are CJK UI Ext.B since UCA_Version 43 (Unicode 13.0). 66# 2A700..2B734 are CJK UI Ext.C since UCA_Version 20 (Unicode 5.2). 67# 2B740..2B81D are CJK UI Ext.D since UCA_Version 22 (Unicode 6.0). 68# 2B820..2CEA1 are CJK UI Ext.E since UCA_Version 32 (Unicode 8.0). 69# 2CEB0..2EBE0 are CJK UI Ext.F since UCA_Version 36 (Unicode 10.0). 70# 30000..3134A are CJK UI Ext.G since UCA_Version 43 (Unicode 13.0). 71 72my @Versions = ( 8, 9, 11, 14, 16, 18, 20, 22, 24, 26, 73 28, 30, 32, 34, 36, 38, 40, 41, 43); 74 75for my $v (@Versions) { 76 $ignoreCJK->change(UCA_Version => $v); 77 78 # UI 79 ok($ignoreCJK->cmp("\x{4E00}", "") == 0); 80 ok($ignoreCJK->cmp("\x{9FA5}", "") == 0); 81 ok($ignoreCJK->cmp("\x{9FA6}", "") == ($v >= 14 ? 0 : 1)); 82 ok($ignoreCJK->cmp("\x{9FAF}", "") == ($v >= 14 ? 0 : 1)); 83 ok($ignoreCJK->cmp("\x{9FB0}", "") == ($v >= 14 ? 0 : 1)); 84 ok($ignoreCJK->cmp("\x{9FBB}", "") == ($v >= 14 ? 0 : 1)); 85 ok($ignoreCJK->cmp("\x{9FBC}", "") == ($v >= 18 ? 0 : 1)); 86 ok($ignoreCJK->cmp("\x{9FBF}", "") == ($v >= 18 ? 0 : 1)); 87 ok($ignoreCJK->cmp("\x{9FC0}", "") == ($v >= 18 ? 0 : 1)); 88 ok($ignoreCJK->cmp("\x{9FC3}", "") == ($v >= 18 ? 0 : 1)); 89 ok($ignoreCJK->cmp("\x{9FC4}", "") == ($v >= 20 ? 0 : 1)); 90 ok($ignoreCJK->cmp("\x{9FCB}", "") == ($v >= 20 ? 0 : 1)); 91 ok($ignoreCJK->cmp("\x{9FCC}", "") == ($v >= 24 ? 0 : 1)); 92 ok($ignoreCJK->cmp("\x{9FCD}", "") == ($v >= 32 ? 0 : 1)); 93 ok($ignoreCJK->cmp("\x{9FCF}", "") == ($v >= 32 ? 0 : 1)); 94 ok($ignoreCJK->cmp("\x{9FD0}", "") == ($v >= 32 ? 0 : 1)); 95 ok($ignoreCJK->cmp("\x{9FD5}", "") == ($v >= 32 ? 0 : 1)); 96 ok($ignoreCJK->cmp("\x{9FD6}", "") == ($v >= 36 ? 0 : 1)); 97 ok($ignoreCJK->cmp("\x{9FDF}", "") == ($v >= 36 ? 0 : 1)); 98 ok($ignoreCJK->cmp("\x{9FEA}", "") == ($v >= 36 ? 0 : 1)); 99 ok($ignoreCJK->cmp("\x{9FEB}", "") == ($v >= 38 ? 0 : 1)); 100 ok($ignoreCJK->cmp("\x{9FEF}", "") == ($v >= 38 ? 0 : 1)); 101 ok($ignoreCJK->cmp("\x{9FF0}", "") == ($v >= 43 ? 0 : 1)); 102 ok($ignoreCJK->cmp("\x{9FFC}", "") == ($v >= 43 ? 0 : 1)); 103 ok($ignoreCJK->cmp("\x{9FFD}", "") == 1); 104 ok($ignoreCJK->cmp("\x{9FFF}", "") == 1); 105 106 # Ext.A 107 ok($ignoreCJK->cmp("\x{3400}", "") == 0); 108 ok($ignoreCJK->cmp("\x{4DB5}", "") == 0); 109 ok($ignoreCJK->cmp("\x{4DB6}", "") == ($v >= 43 ? 0 : 1)); 110 ok($ignoreCJK->cmp("\x{4DBF}", "") == ($v >= 43 ? 0 : 1)); 111 112 # Ext.B 113 ok($ignoreCJK->cmp("\x{20000}","") == 0); 114 ok($ignoreCJK->cmp("\x{2A6D6}","") == 0); 115 ok($ignoreCJK->cmp("\x{2A6D7}","") == ($v >= 43 ? 0 : 1)); 116 ok($ignoreCJK->cmp("\x{2A6DD}","") == ($v >= 43 ? 0 : 1)); 117 ok($ignoreCJK->cmp("\x{2A6DE}","") == 1); 118 ok($ignoreCJK->cmp("\x{2A6DF}","") == 1); 119 120 # Ext.C 121 ok($ignoreCJK->cmp("\x{2A700}","") == ($v >= 20 ? 0 : 1)); 122 ok($ignoreCJK->cmp("\x{2B734}","") == ($v >= 20 ? 0 : 1)); 123 ok($ignoreCJK->cmp("\x{2B735}","") == 1); 124 ok($ignoreCJK->cmp("\x{2B73F}","") == 1); 125 126 # Ext.D 127 ok($ignoreCJK->cmp("\x{2B740}","") == ($v >= 22 ? 0 : 1)); 128 ok($ignoreCJK->cmp("\x{2B81D}","") == ($v >= 22 ? 0 : 1)); 129 ok($ignoreCJK->cmp("\x{2B81E}","") == 1); 130 ok($ignoreCJK->cmp("\x{2B81F}","") == 1); 131 132 # Ext.E 133 ok($ignoreCJK->cmp("\x{2B820}","") == ($v >= 32 ? 0 : 1)); 134 ok($ignoreCJK->cmp("\x{2CEA1}","") == ($v >= 32 ? 0 : 1)); 135 ok($ignoreCJK->cmp("\x{2CEA2}","") == 1); 136 ok($ignoreCJK->cmp("\x{2CEAF}","") == 1); 137 138 # Ext.F 139 ok($ignoreCJK->cmp("\x{2CEB0}","") == ($v >= 36 ? 0 : 1)); 140 ok($ignoreCJK->cmp("\x{2EBE0}","") == ($v >= 36 ? 0 : 1)); 141 ok($ignoreCJK->cmp("\x{2EBE1}","") == 1); 142 ok($ignoreCJK->cmp("\x{2EBEF}","") == 1); 143 144 # Ext.G 145 ok($ignoreCJK->cmp("\x{30000}","") == ($v >= 43 ? 0 : 1)); 146 ok($ignoreCJK->cmp("\x{3134A}","") == ($v >= 43 ? 0 : 1)); 147 ok($ignoreCJK->cmp("\x{3134B}","") == 1); 148 ok($ignoreCJK->cmp("\x{3134F}","") == 1); 149} 150 151