1 2BEGIN { 3 unless ("A" eq pack('U', 0x41)) { 4 print "1..0 # Unicode::Collate " . 5 "cannot stringify a Unicode code point\n"; 6 exit 0; 7 } 8 if ($ENV{PERL_CORE}) { 9 chdir('t') if -d 't'; 10 @INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib); 11 } 12} 13 14use strict; 15use warnings; 16BEGIN { $| = 1; print "1..41\n"; } 17my $count = 0; 18sub ok ($;$) { 19 my $p = my $r = shift; 20 if (@_) { 21 my $x = shift; 22 $p = !defined $x ? !defined $r : !defined $r ? 0 : $r eq $x; 23 } 24 print $p ? "ok" : "not ok", ' ', ++$count, "\n"; 25} 26 27use Unicode::Collate; 28 29ok(1); 30 31######################### 32 33my $trad = Unicode::Collate->new( 34 table => 'keys.txt', 35 normalization => undef, 36 ignoreName => qr/HANGUL|HIRAGANA|KATAKANA|BOPOMOFO/, 37 level => 3, 38 entry => << 'ENTRIES', 39 0063 0068 ; [.0A3F.0020.0002.0063] % "ch" in traditional Spanish 40 0043 0068 ; [.0A3F.0020.0007.0043] # "Ch" in traditional Spanish 41 0043 0048 ; [.0A3F.0020.0008.0043] # "CH" in traditional Spanish 42ENTRIES 43); 44# 0063 ; [.0A3D.0020.0002.0063] # LATIN SMALL LETTER C 45# 0064 ; [.0A49.0020.0002.0064] # LATIN SMALL LETTER D 46 47##### 2..3 48 49ok( 50 join(':', $trad->sort( qw/ acha aca ada acia acka / ) ), 51 join(':', qw/ aca acia acka acha ada / ), 52); 53 54ok( 55 join(':', $trad->sort( qw/ ACHA ACA ADA ACIA ACKA / ) ), 56 join(':', qw/ ACA ACIA ACKA ACHA ADA / ), 57); 58 59##### 4..7 60 61ok($trad->gt("ocho", "oc\cAho")); # UCA v14 62ok($trad->gt("ocho", "oc\0\cA\0\cBho")); # UCA v14 63ok($trad->eq("-", "")); 64ok($trad->gt("ocho", "oc-ho")); 65 66##### 8..11 67 68$trad->change(UCA_Version => 9); 69 70ok($trad->eq("ocho", "oc\cAho")); # UCA v9 71ok($trad->eq("ocho", "oc\0\cA\0\cBho")); # UCA v9 72ok($trad->eq("-", "")); 73ok($trad->gt("ocho", "oc-ho")); 74 75##### 12..15 76 77$trad->change(UCA_Version => 8); 78 79ok($trad->gt("ocho", "oc\cAho")); 80ok($trad->gt("ocho", "oc\0\cA\0\cBho")); 81ok($trad->eq("-", "")); 82ok($trad->gt("ocho", "oc-ho")); 83 84 85##### 16..19 86 87$trad->change(UCA_Version => 9); 88 89my $hiragana = "\x{3042}\x{3044}"; 90my $katakana = "\x{30A2}\x{30A4}"; 91 92# HIRAGANA and KATAKANA are ignorable via ignoreName 93ok($trad->eq($hiragana, "")); 94ok($trad->eq("", $katakana)); 95ok($trad->eq($hiragana, $katakana)); 96ok($trad->eq($katakana, $hiragana)); 97 98 99##### 20..31 100 101# According to Conformance Test (UCA_Version == 9 or 11), 102# a L3-ignorable is treated as a completely ignorable. 103 104my $L3ignorable = Unicode::Collate->new( 105 alternate => 'Non-ignorable', 106 level => 3, 107 table => undef, 108 normalization => undef, 109 UCA_Version => 9, 110 entry => <<'ENTRIES', 1110000 ; [.0000.0000.0000.0000] # [0000] NULL (in 6429) 1120001 ; [.0000.0000.0000.0000] # [0001] START OF HEADING (in 6429) 1130591 ; [.0000.0000.0000.0591] # HEBREW ACCENT ETNAHTA 1141D165 ; [.0000.0000.0000.1D165] # MUSICAL SYMBOL COMBINING STEM 1150021 ; [*024B.0020.0002.0021] # EXCLAMATION MARK 11609BE ; [.114E.0020.0002.09BE] # BENGALI VOWEL SIGN AA 11709C7 ; [.1157.0020.0002.09C7] # BENGALI VOWEL SIGN E 11809CB ; [.1159.0020.0002.09CB] # BENGALI VOWEL SIGN O 11909C7 09BE ; [.1159.0020.0002.09CB] # BENGALI VOWEL SIGN O 1201D1B9 ; [*098A.0020.0002.1D1B9] # MUSICAL SYMBOL SEMIBREVIS WHITE 1211D1BA ; [*098B.0020.0002.1D1BA] # MUSICAL SYMBOL SEMIBREVIS BLACK 1221D1BB ; [*098A.0020.0002.1D1B9][.0000.0000.0000.1D165] # M.S. MINIMA 1231D1BC ; [*098B.0020.0002.1D1BA][.0000.0000.0000.1D165] # M.S. MINIMA BLACK 124ENTRIES 125); 126 127ok($L3ignorable->lt("\cA", "!")); 128ok($L3ignorable->lt("\x{591}", "!")); 129ok($L3ignorable->eq("\cA", "\x{591}")); 130ok($L3ignorable->eq("\x{9C7}\x{9BE}A", "\x{9C7}\cA\x{9BE}A")); 131ok($L3ignorable->eq("\x{9C7}\x{9BE}A", "\x{9C7}\x{591}\x{9BE}A")); 132ok($L3ignorable->eq("\x{9C7}\x{9BE}A", "\x{9C7}\x{1D165}\x{9BE}A")); 133ok($L3ignorable->eq("\x{9C7}\x{9BE}A", "\x{9CB}A")); 134ok($L3ignorable->lt("\x{1D1BB}", "\x{1D1BC}")); 135ok($L3ignorable->eq("\x{1D1BB}", "\x{1D1B9}")); 136ok($L3ignorable->eq("\x{1D1BC}", "\x{1D1BA}")); 137ok($L3ignorable->eq("\x{1D1BB}", "\x{1D1B9}\x{1D165}")); 138ok($L3ignorable->eq("\x{1D1BC}", "\x{1D1BA}\x{1D165}")); 139 140##### 32..41 141 142my $c = Unicode::Collate->new( 143 table => 'keys.txt', 144 normalization => undef, 145 level => 1, 146 UCA_Version => 14, 147 entry => << 'ENTRIES', 148034F ; [.0000.0000.0000.034F] # COMBINING GRAPHEME JOINER 1490063 0068 ; [.0A3F.0020.0002.0063] % "ch" in traditional Spanish 1500043 0068 ; [.0A3F.0020.0007.0043] # "Ch" in traditional Spanish 1510043 0048 ; [.0A3F.0020.0008.0043] # "CH" in traditional Spanish 152ENTRIES 153); 154# 0063 ; [.0A3D.0020.0002.0063] # LATIN SMALL LETTER C 155# 0064 ; [.0A49.0020.0002.0064] # LATIN SMALL LETTER D 156 157ok($c->gt("ocho", "oc\x00\x00ho")); 158ok($c->gt("ocho", "oc\cAho")); 159ok($c->gt("ocho", "oc\x{34F}ho")); 160ok($c->gt("ocio", "oc\x{34F}ho")); 161ok($c->lt("ocgo", "oc\x{34F}ho")); 162ok($c->lt("oceo", "oc\x{34F}ho")); 163 164ok($c->viewSortKey("ocho"), "[0B4B 0A3F 0B4B | | |]"); 165ok($c->viewSortKey("oc\x00\x00ho"), "[0B4B 0A3D 0AB9 0B4B | | |]"); 166ok($c->viewSortKey("oc\cAho"), "[0B4B 0A3D 0AB9 0B4B | | |]"); 167ok($c->viewSortKey("oc\x{34F}ho"), "[0B4B 0A3D 0AB9 0B4B | | |]"); 168 169 170