1 2BEGIN { 3 unless ('A' eq pack('U', 0x41)) { 4 print "1..0 # Unicode::Collate cannot pack a Unicode code point\n"; 5 exit 0; 6 } 7 unless (0x41 == unpack('U', 'A')) { 8 print "1..0 # Unicode::Collate cannot get a Unicode code point\n"; 9 exit 0; 10 } 11 if ($ENV{PERL_CORE}) { 12 chdir('t') if -d 't'; 13 @INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib); 14 } 15} 16 17use strict; 18use warnings; 19BEGIN { $| = 1; print "1..901\n"; } # 1 + 60 x @Versions 20my $count = 0; 21sub ok ($;$) { 22 my $p = my $r = shift; 23 if (@_) { 24 my $x = shift; 25 $p = !defined $x ? !defined $r : !defined $r ? 0 : $r eq $x; 26 } 27 print $p ? "ok" : "not ok", ' ', ++$count, "\n"; 28} 29 30use Unicode::Collate; 31 32ok(1); 33 34######################### 35 36my $coll = Unicode::Collate->new( 37 table => 'keys.txt', 38 normalization => undef, 39); 40 41# CJK UI Ext > CJK UI. 42# [ UCA_Version 8: Ext.A < UI and BMP < Ext.B (code point order) ] 43 44# 4E00..9FA5 are CJK UI. 45# 9FA6..9FBB are CJK UI since UCA_Version 14 (Unicode 4.1). 46# 9FBC..9FC3 are CJK UI since UCA_Version 18 (Unicode 5.1). 47# 9FC4..9FCB are CJK UI since UCA_Version 20 (Unicode 5.2). 48# 9FCC is CJK UI since UCA_Version 24 (Unicode 6.1). 49# 9FCD..9FD5 are CJK UI since UCA_Version 32 (Unicode 8.0). 50# 9FD6..9FEA are CJK UI since UCA_Version 36 (Unicode 10.0). 51 52# 3400..4DB5 are CJK UI Ext.A since UCA_Version 8 (Unicode 3.0). 53# 20000..2A6D6 are CJK UI Ext.B since UCA_Version 8 (Unicode 3.1). 54# 2A700..2B734 are CJK UI Ext.C since UCA_Version 20 (Unicode 5.2). 55# 2B740..2B81D are CJK UI Ext.D since UCA_Version 22 (Unicode 6.0). 56# 2B820..2CEA1 are CJK UI Ext.E since UCA_Version 32 (Unicode 8.0). 57# 2CEB0..2EBE0 are CJK UI Ext.F since UCA_Version 36 (Unicode 10.0). 58 59my @Versions = (8, 9, 11, 14, 16, 18, 20, 22, 24, 26, 28, 30, 32, 34, 36); 60 61for my $v (@Versions) { 62 $coll->change(UCA_Version => $v); 63 64 # Ext.A > UI 65 ok($coll->cmp("\x{3400}", "\x{4E00}") == ($v >= 9 ? 1 : -1)); # UI 66 ok($coll->cmp("\x{3400}", "\x{9FA5}") == ($v >= 9 ? 1 : -1)); # UI 67 ok($coll->cmp("\x{3400}", "\x{9FA6}") == ($v >= 14 ? 1 : -1)); # new 68 ok($coll->cmp("\x{3400}", "\x{9FBB}") == ($v >= 14 ? 1 : -1)); # new 69 ok($coll->cmp("\x{3400}", "\x{9FBC}") == ($v >= 18 ? 1 : -1)); # new 70 ok($coll->cmp("\x{3400}", "\x{9FC3}") == ($v >= 18 ? 1 : -1)); # new 71 ok($coll->cmp("\x{3400}", "\x{9FC4}") == ($v >= 20 ? 1 : -1)); # new 72 ok($coll->cmp("\x{3400}", "\x{9FCB}") == ($v >= 20 ? 1 : -1)); # new 73 ok($coll->cmp("\x{3400}", "\x{9FCC}") == ($v >= 24 ? 1 : -1)); # new 74 ok($coll->cmp("\x{3400}", "\x{9FCD}") == ($v >= 32 ? 1 : -1)); # new 75 ok($coll->cmp("\x{3400}", "\x{9FD5}") == ($v >= 32 ? 1 : -1)); # new 76 ok($coll->cmp("\x{3400}", "\x{9FD6}") == ($v >= 36 ? 1 : -1)); # new 77 ok($coll->cmp("\x{3400}", "\x{9FEA}") == ($v >= 36 ? 1 : -1)); # new 78 ok($coll->cmp("\x{3400}", "\x{9FEB}") == -1); # na 79 ok($coll->cmp("\x{3400}", "\x{9FFF}") == -1); # na 80 81 # UI < UI 82 ok($coll->cmp("\x{4E00}", "\x{9FA5}") == -1); # UI < UI 83 ok($coll->cmp("\x{9FA5}", "\x{9FA6}") == -1); # UI < new 84 ok($coll->cmp("\x{9FA6}", "\x{9FBB}") == -1); # new < new 85 ok($coll->cmp("\x{9FBB}", "\x{9FBC}") == -1); # new < new 86 ok($coll->cmp("\x{9FBC}", "\x{9FC3}") == -1); # new < new 87 ok($coll->cmp("\x{9FC3}", "\x{9FC4}") == -1); # new < new 88 ok($coll->cmp("\x{9FC4}", "\x{9FCB}") == -1); # new < new 89 ok($coll->cmp("\x{9FCB}", "\x{9FCC}") == -1); # new < new 90 ok($coll->cmp("\x{9FCC}", "\x{9FCD}") == -1); # new < new 91 ok($coll->cmp("\x{9FCD}", "\x{9FD5}") == -1); # new < new 92 ok($coll->cmp("\x{9FD5}", "\x{9FD6}") == -1); # new < new 93 ok($coll->cmp("\x{9FD6}", "\x{9FEA}") == -1); # new < new 94 ok($coll->cmp("\x{9FEA}", "\x{9FEB}") == -1); # new < na 95 ok($coll->cmp("\x{9FEB}", "\x{9FFF}") == -1); # na < na 96 97 # Ext.A < Ext.B 98 ok($coll->cmp("\x{3400}", "\x{20000}") == -1); 99 100 # Ext.A 101 ok($coll->cmp("\x{3400}", "\x{4DB5}") == -1); # A < A 102 ok($coll->cmp("\x{2FFF}", "\x{3400}") == ($v >= 8 ? 1 : -1)); # na > A 103 ok($coll->cmp("\x{2FFF}", "\x{4DB5}") == ($v >= 8 ? 1 : -1)); # na > A 104 ok($coll->cmp("\x{2FFF}", "\x{4DB6}") == -1); # na < na 105 ok($coll->cmp("\x{2FFF}", "\x{4DBF}") == -1); # na < na 106 107 # Ext.B 108 ok($coll->cmp("\x{20000}","\x{2A6D6}") == -1); # B < B 109 ok($coll->cmp("\x{2FFF}", "\x{20000}") == ($v >= 9 ? 1 : -1)); # na > B 110 ok($coll->cmp("\x{2FFF}", "\x{2A6D6}") == ($v >= 9 ? 1 : -1)); # na > B 111 ok($coll->cmp("\x{2FFF}", "\x{2A6D7}") == -1); # na < na 112 ok($coll->cmp("\x{2FFF}", "\x{2A6DF}") == -1); # na < na 113 114 # Ext.C 115 ok($coll->cmp("\x{2A700}","\x{2B734}") == -1); # C < C 116 ok($coll->cmp("\x{2FFF}", "\x{2A700}") == ($v >= 20 ? 1 : -1)); # na > C 117 ok($coll->cmp("\x{2FFF}", "\x{2B734}") == ($v >= 20 ? 1 : -1)); # na > C 118 ok($coll->cmp("\x{2FFF}", "\x{2B735}") == -1); # na < na 119 ok($coll->cmp("\x{2FFF}", "\x{2B73F}") == -1); # na < na 120 121 # Ext.D 122 ok($coll->cmp("\x{2B740}","\x{2B81D}") == -1); # D < D 123 ok($coll->cmp("\x{2FFF}", "\x{2B740}") == ($v >= 22 ? 1 : -1)); # na > D 124 ok($coll->cmp("\x{2FFF}", "\x{2B81D}") == ($v >= 22 ? 1 : -1)); # na > D 125 ok($coll->cmp("\x{2FFF}", "\x{2B81E}") == -1); # na < na 126 ok($coll->cmp("\x{2FFF}", "\x{2B81F}") == -1); # na < na 127 128 # Ext.E 129 ok($coll->cmp("\x{2B820}","\x{2CEA1}") == -1); # E < E 130 ok($coll->cmp("\x{2FFF}", "\x{2B820}") == ($v >= 32 ? 1 : -1)); # na > E 131 ok($coll->cmp("\x{2FFF}", "\x{2CEA1}") == ($v >= 32 ? 1 : -1)); # na > E 132 ok($coll->cmp("\x{2FFF}", "\x{2CEA2}") == -1); # na < na 133 ok($coll->cmp("\x{2FFF}", "\x{2CEAF}") == -1); # na < na 134 135 # Ext.F 136 ok($coll->cmp("\x{2CEB0}","\x{2EBE0}") == -1); # F < F 137 ok($coll->cmp("\x{2FFF}", "\x{2CEB0}") == ($v >= 36 ? 1 : -1)); # na > F 138 ok($coll->cmp("\x{2FFF}", "\x{2EBE0}") == ($v >= 36 ? 1 : -1)); # na > F 139 ok($coll->cmp("\x{2FFF}", "\x{2EBE1}") == -1); # na < na 140 ok($coll->cmp("\x{2FFF}", "\x{2EBEF}") == -1); # na < na 141} 142 143