| Commit | Line | Data |
|---|
| 91ae00cb |
1 | BEGIN { |
| 2 | unless ("A" eq pack('U', 0x41)) { |
| 3 | print "1..0 # Unicode::Collate " . |
| 4 | "cannot stringify a Unicode code point\n"; |
| 5 | exit 0; |
| 6 | } |
| 91ae00cb |
7 | if ($ENV{PERL_CORE}) { |
| 3756e7ca |
8 | chdir('t') if -d 't'; |
| 9 | @INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib); |
| 91ae00cb |
10 | } |
| 11 | } |
| 12 | |
| 13 | use Test; |
| 14 | BEGIN { plan tests => 58 }; |
| 15 | |
| 16 | use strict; |
| 17 | use warnings; |
| 18 | use Unicode::Collate; |
| 19 | |
| 20 | ######################### |
| 21 | |
| 22 | ok(1); |
| 23 | |
| 24 | # a standard collator (3.1.1) |
| 25 | my $Collator = Unicode::Collate->new( |
| 26 | level => 1, |
| 27 | table => 'keys.txt', |
| 28 | normalization => undef, |
| 29 | |
| 30 | entry => <<'ENTRIES', |
| 31 | 326E ; [.1831.0020.0006.326E][.188D.0020.0006.326E] # c.h.s. GA |
| 32 | 326F ; [.1833.0020.0006.326F][.188D.0020.0006.326F] # c.h.s. NA |
| 33 | 3270 ; [.1834.0020.0006.3270][.188D.0020.0006.3270] # c.h.s. DA |
| 34 | 3271 ; [.1836.0020.0006.3271][.188D.0020.0006.3271] # c.h.s. RA |
| 35 | 3272 ; [.1837.0020.0006.3272][.188D.0020.0006.3272] # c.h.s. MA |
| 36 | 3273 ; [.1838.0020.0006.3273][.188D.0020.0006.3273] # c.h.s. BA |
| 37 | 3274 ; [.183A.0020.0006.3274][.188D.0020.0006.3274] # c.h.s. SA |
| 38 | 3275 ; [.183C.0020.0006.3275][.188D.0020.0006.3275] # c.h.s. A |
| 39 | 3276 ; [.183D.0020.0006.3276][.188D.0020.0006.3276] # c.h.s. JA |
| 40 | 3277 ; [.183F.0020.0006.3277][.188D.0020.0006.3277] # c.h.s. CA |
| 41 | 3278 ; [.1840.0020.0006.3278][.188D.0020.0006.3278] # c.h.s. KA |
| 42 | 3279 ; [.1841.0020.0006.3279][.188D.0020.0006.3279] # c.h.s. TA |
| 43 | 327A ; [.1842.0020.0006.327A][.188D.0020.0006.327A] # c.h.s. PA |
| 44 | 327B ; [.1843.0020.0006.327B][.188D.0020.0006.327B] # c.h.s. HA |
| 45 | ENTRIES |
| 46 | ); |
| 47 | |
| 48 | my $hangul = Unicode::Collate->new( |
| 49 | level => 1, |
| 50 | table => 'keys.txt', |
| 51 | normalization => undef, |
| 52 | hangul_terminator => 16, |
| 53 | |
| 54 | entry => <<'ENTRIES', |
| 55 | 326E ; [.1831.0020.0006.326E][.188D.0020.0006.326E] # c.h.s. GA |
| 56 | 326F ; [.1833.0020.0006.326F][.188D.0020.0006.326F] # c.h.s. NA |
| 57 | 3270 ; [.1834.0020.0006.3270][.188D.0020.0006.3270] # c.h.s. DA |
| 58 | 3271 ; [.1836.0020.0006.3271][.188D.0020.0006.3271] # c.h.s. RA |
| 59 | 3272 ; [.1837.0020.0006.3272][.188D.0020.0006.3272] # c.h.s. MA |
| 60 | 3273 ; [.1838.0020.0006.3273][.188D.0020.0006.3273] # c.h.s. BA |
| 61 | 3274 ; [.183A.0020.0006.3274][.188D.0020.0006.3274] # c.h.s. SA |
| 62 | 3275 ; [.183C.0020.0006.3275][.188D.0020.0006.3275] # c.h.s. A |
| 63 | 3276 ; [.183D.0020.0006.3276][.188D.0020.0006.3276] # c.h.s. JA |
| 64 | 3277 ; [.183F.0020.0006.3277][.188D.0020.0006.3277] # c.h.s. CA |
| 65 | 3278 ; [.1840.0020.0006.3278][.188D.0020.0006.3278] # c.h.s. KA |
| 66 | 3279 ; [.1841.0020.0006.3279][.188D.0020.0006.3279] # c.h.s. TA |
| 67 | 327A ; [.1842.0020.0006.327A][.188D.0020.0006.327A] # c.h.s. PA |
| 68 | 327B ; [.1843.0020.0006.327B][.188D.0020.0006.327B] # c.h.s. HA |
| 69 | ENTRIES |
| 70 | ); |
| 71 | |
| 72 | ok(ref $hangul, "Unicode::Collate"); |
| 73 | |
| 74 | ######################### |
| 75 | |
| 76 | # LVX vs LVV: /GAA/ vs /GA/.latinA |
| 77 | ok($Collator->gt("\x{1100}\x{1161}\x{1161}", "\x{1100}\x{1161}A")); |
| 78 | ok($hangul ->gt("\x{1100}\x{1161}\x{1161}", "\x{1100}\x{1161}A")); |
| 79 | |
| 80 | # LVX vs LVV: /GAA/ vs /GA/.hiraganaA |
| 81 | ok($Collator->lt("\x{1100}\x{1161}\x{1161}", "\x{1100}\x{1161}\x{3042}")); |
| 82 | ok($hangul ->gt("\x{1100}\x{1161}\x{1161}", "\x{1100}\x{1161}\x{3042}")); |
| 83 | |
| 84 | # LVX vs LVV: /GAA/ vs /GA/.hanja |
| 85 | ok($Collator->lt("\x{1100}\x{1161}\x{1161}", "\x{1100}\x{1161}\x{4E00}")); |
| 86 | ok($hangul ->gt("\x{1100}\x{1161}\x{1161}", "\x{1100}\x{1161}\x{4E00}")); |
| 87 | |
| 88 | # LVL vs LVT: /GA/./G/ vs /GAG/ |
| 89 | ok($Collator->lt("\x{1100}\x{1161}\x{1100}", "\x{1100}\x{1161}\x{11A8}")); |
| 90 | ok($hangul ->lt("\x{1100}\x{1161}\x{1100}", "\x{1100}\x{1161}\x{11A8}")); |
| 91 | |
| 92 | # LVT vs LVX: /GAG/ vs /GA/.latinA |
| 93 | ok($Collator->gt("\x{1100}\x{1161}\x{11A8}", "\x{1100}\x{1161}A")); |
| 94 | ok($hangul ->gt("\x{1100}\x{1161}\x{11A8}", "\x{1100}\x{1161}A")); |
| 95 | |
| 96 | # LVT vs LVX: /GAG/ vs /GA/.hiraganaA |
| 97 | ok($Collator->lt("\x{1100}\x{1161}\x{11A8}", "\x{1100}\x{1161}\x{3042}")); |
| 98 | ok($hangul ->gt("\x{1100}\x{1161}\x{11A8}", "\x{1100}\x{1161}\x{3042}")); |
| 99 | |
| 100 | # LVT vs LVX: /GAG/ vs /GA/.hanja |
| 101 | ok($Collator->lt("\x{1100}\x{1161}\x{11A8}", "\x{1100}\x{1161}\x{4E00}")); |
| 102 | ok($hangul ->gt("\x{1100}\x{1161}\x{11A8}", "\x{1100}\x{1161}\x{4E00}")); |
| 103 | |
| 104 | # LV vs Syl(LV): /GA/ vs /[GA]/ |
| 105 | ok($Collator->eq("\x{1100}\x{1161}", "\x{AC00}")); |
| 106 | ok($hangul ->eq("\x{1100}\x{1161}", "\x{AC00}")); |
| 107 | |
| 108 | # LVT vs Syl(LV)T: /GAG/ vs /[GA]G/ |
| 109 | ok($Collator->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC00}\x{11A8}")); |
| 110 | ok($hangul ->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC00}\x{11A8}")); |
| 111 | |
| 112 | # LVT vs Syl(LVT): /GAG/ vs /[GAG]/ |
| 113 | ok($Collator->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC01}")); |
| 114 | ok($hangul ->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC01}")); |
| 115 | |
| 116 | # LVTT vs Syl(LVTT): /GAGG/ vs /[GAGG]/ |
| 117 | ok($Collator->eq("\x{1100}\x{1161}\x{11A9}", "\x{AC02}")); |
| 118 | ok($hangul ->eq("\x{1100}\x{1161}\x{11A9}", "\x{AC02}")); |
| 119 | |
| 120 | # Syl(LVT) vs : /GAG/ vs /[GAG]/ |
| 121 | ok($Collator->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC01}")); |
| 122 | ok($hangul ->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC01}")); |
| 123 | |
| 124 | ######################### |
| 125 | |
| 126 | my $hangcirc = Unicode::Collate->new( |
| 127 | level => 1, |
| 128 | table => 'keys.txt', |
| 129 | normalization => undef, |
| 130 | hangul_terminator => 16, |
| 131 | |
| 132 | entry => <<'ENTRIES', |
| 133 | 326E ; [.1831.0020.0006.326E][.188D.0020.0006.326E][.10.0.0.0] # c.h.s. GA |
| 134 | 326F ; [.1833.0020.0006.326F][.188D.0020.0006.326F][.10.0.0.0] # c.h.s. NA |
| 135 | 3270 ; [.1834.0020.0006.3270][.188D.0020.0006.3270][.10.0.0.0] # c.h.s. DA |
| 136 | 3271 ; [.1836.0020.0006.3271][.188D.0020.0006.3271][.10.0.0.0] # c.h.s. RA |
| 137 | 3272 ; [.1837.0020.0006.3272][.188D.0020.0006.3272][.10.0.0.0] # c.h.s. MA |
| 138 | 3273 ; [.1838.0020.0006.3273][.188D.0020.0006.3273][.10.0.0.0] # c.h.s. BA |
| 139 | 3274 ; [.183A.0020.0006.3274][.188D.0020.0006.3274][.10.0.0.0] # c.h.s. SA |
| 140 | 3275 ; [.183C.0020.0006.3275][.188D.0020.0006.3275][.10.0.0.0] # c.h.s. A |
| 141 | 3276 ; [.183D.0020.0006.3276][.188D.0020.0006.3276][.10.0.0.0] # c.h.s. JA |
| 142 | 3277 ; [.183F.0020.0006.3277][.188D.0020.0006.3277][.10.0.0.0] # c.h.s. CA |
| 143 | 3278 ; [.1840.0020.0006.3278][.188D.0020.0006.3278][.10.0.0.0] # c.h.s. KA |
| 144 | 3279 ; [.1841.0020.0006.3279][.188D.0020.0006.3279][.10.0.0.0] # c.h.s. TA |
| 145 | 327A ; [.1842.0020.0006.327A][.188D.0020.0006.327A][.10.0.0.0] # c.h.s. PA |
| 146 | 327B ; [.1843.0020.0006.327B][.188D.0020.0006.327B][.10.0.0.0] # c.h.s. HA |
| 147 | ENTRIES |
| 148 | ); |
| 149 | |
| 150 | # LV vs Circled Syl(LV): /GA/ vs /(GA)/ |
| 151 | ok($Collator->eq("\x{1100}\x{1161}", "\x{326E}")); |
| 152 | ok($hangul ->gt("\x{1100}\x{1161}", "\x{326E}")); |
| 153 | ok($hangcirc->eq("\x{1100}\x{1161}", "\x{326E}")); |
| 154 | |
| 155 | # LV vs Circled Syl(LV): followed by latin A |
| 156 | ok($Collator->eq("\x{1100}\x{1161}A", "\x{326E}A")); |
| 157 | ok($hangul ->lt("\x{1100}\x{1161}A", "\x{326E}A")); |
| 158 | ok($hangcirc->eq("\x{1100}\x{1161}A", "\x{326E}A")); |
| 159 | |
| 160 | # LV vs Circled Syl(LV): followed by hiragana A |
| 161 | ok($Collator->eq("\x{1100}\x{1161}\x{3042}", "\x{326E}\x{3042}")); |
| 162 | ok($hangul ->lt("\x{1100}\x{1161}\x{3042}", "\x{326E}\x{3042}")); |
| 163 | ok($hangcirc->eq("\x{1100}\x{1161}\x{3042}", "\x{326E}\x{3042}")); |
| 164 | |
| 165 | # LVT vs LVX: /GAG/ vs /GA/.hanja |
| 166 | ok($Collator->eq("\x{1100}\x{1161}\x{4E00}", "\x{326E}\x{4E00}")); |
| 167 | ok($hangul ->lt("\x{1100}\x{1161}\x{4E00}", "\x{326E}\x{4E00}")); |
| 168 | ok($hangcirc->eq("\x{1100}\x{1161}\x{4E00}", "\x{326E}\x{4E00}")); |
| 169 | |
| 170 | ######################### |
| 171 | |
| 172 | # checks contraction in LVT: |
| 173 | # weights of these contractions may be non-sense. |
| 174 | |
| 175 | my $hangcont = Unicode::Collate->new( |
| 176 | level => 1, |
| 177 | table => 'keys.txt', |
| 178 | normalization => undef, |
| 179 | hangul_terminator => 16, |
| 180 | |
| 181 | entry => <<'ENTRIES', |
| 182 | 1100 1161 ; [.1831.0020.0002.1100][.188D.0020.0002.1161] # KIYEOK+A |
| 183 | 1161 11A8 ; [.188D.0020.0002.1161][.18CF.0020.0002.11A8] # A+KIYEOK |
| 184 | ENTRIES |
| 185 | ); |
| 186 | |
| 187 | # cont<LV> vs Syl(LV): /<GA>/ vs /[GA]/ |
| 188 | ok($Collator->eq("\x{1100}\x{1161}", "\x{AC00}")); |
| 189 | ok($hangcont->eq("\x{1100}\x{1161}", "\x{AC00}")); |
| 190 | |
| 191 | # cont<LV>.T vs Syl(LV).T: /<GA>G/ vs /[GA]G/ |
| 192 | ok($Collator->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC00}\x{11A8}")); |
| 193 | ok($hangcont->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC00}\x{11A8}")); |
| 194 | |
| 195 | # cont<LV>.T vs Syl(LVT): /<GA>G/ vs /[GAG]/ |
| 196 | ok($Collator->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC01}")); |
| 197 | ok($hangcont->eq("\x{1100}\x{1161}\x{11A8}", "\x{AC01}")); |
| 198 | |
| 199 | # L.cont<VT> vs Syl(LV).T: /D<AG>/ vs /[DA]G/ |
| 200 | ok($Collator->eq("\x{1103}\x{1161}\x{11A8}", "\x{B2E4}\x{11A8}")); |
| 201 | ok($hangcont->eq("\x{1103}\x{1161}\x{11A8}", "\x{B2E4}\x{11A8}")); |
| 202 | |
| 203 | # L.cont<VT> vs Syl(LVT): /D<AG>/ vs /[DAG]/ |
| 204 | ok($Collator->eq("\x{1103}\x{1161}\x{11A8}", "\x{B2E5}")); |
| 205 | ok($hangcont->eq("\x{1103}\x{1161}\x{11A8}", "\x{B2E5}")); |
| 206 | |
| 207 | ##### |
| 208 | |
| 209 | $Collator->change(hangul_terminator => 16); |
| 210 | |
| 211 | ok($Collator->gt("\x{1100}\x{1161}\x{11A8}", "\x{1100}\x{1161}\x{4E00}")); |
| 212 | ok($Collator->gt("\x{1100}\x{1161}", "\x{326E}")); |
| 213 | ok($Collator->lt("\x{1100}\x{1161}A", "\x{326E}A")); |
| 214 | ok($Collator->lt("\x{1100}\x{1161}\x{3042}", "\x{326E}\x{3042}")); |
| 215 | ok($Collator->lt("\x{1100}\x{1161}\x{4E00}", "\x{326E}\x{4E00}")); |
| 216 | |
| 217 | $Collator->change(hangul_terminator => 0); |
| 218 | |
| 219 | ok($Collator->lt("\x{1100}\x{1161}\x{11A8}", "\x{1100}\x{1161}\x{4E00}")); |
| 220 | ok($Collator->eq("\x{1100}\x{1161}", "\x{326E}")); |
| 221 | ok($Collator->eq("\x{1100}\x{1161}A", "\x{326E}A")); |
| 222 | ok($Collator->eq("\x{1100}\x{1161}\x{3042}", "\x{326E}\x{3042}")); |
| 223 | ok($Collator->eq("\x{1100}\x{1161}\x{4E00}", "\x{326E}\x{4E00}")); |
| 224 | |
| 225 | 1; |
| 226 | __END__ |
projects / p5sagit/p5-mst-13.2.git / blame