lib/Unicode/UCD.t

   1 BEGIN {
   2     if (ord("A") == 193) {
   3         print "1..0 # Skip: EBCDIC\n";
   4         exit 0;
   5     }
   6     chdir 't' if -d 't';
   7     @INC = '../lib';
   8 }
   9
  10 use strict;
  11 use Unicode::UCD;
  12 use Test::More;
  13
  14 BEGIN { plan tests => 162 };
  15
  16 use Unicode::UCD 'charinfo';
  17
  18 my $charinfo;
  19
  20 $charinfo = charinfo(0x41);
  21
  22 is($charinfo->{code},           '0041', 'LATIN CAPITAL LETTER A');
  23 is($charinfo->{name},           'LATIN CAPITAL LETTER A');
  24 is($charinfo->{category},       'Lu');
  25 is($charinfo->{combining},      '0');
  26 is($charinfo->{bidi},           'L');
  27 is($charinfo->{decomposition},  '');
  28 is($charinfo->{decimal},        '');
  29 is($charinfo->{digit},          '');
  30 is($charinfo->{numeric},        '');
  31 is($charinfo->{mirrored},       'N');
  32 is($charinfo->{unicode10},      '');
  33 is($charinfo->{comment},        '');
  34 is($charinfo->{upper},          '');
  35 is($charinfo->{lower},          '0061');
  36 is($charinfo->{title},          '');
  37 is($charinfo->{block},          'Basic Latin');
  38 is($charinfo->{script},         'Latin');
  39
  40 $charinfo = charinfo(0x100);
  41
  42 is($charinfo->{code},           '0100', 'LATIN CAPITAL LETTER A WITH MACRON');
  43 is($charinfo->{name},           'LATIN CAPITAL LETTER A WITH MACRON');
  44 is($charinfo->{category},       'Lu');
  45 is($charinfo->{combining},      '0');
  46 is($charinfo->{bidi},           'L');
  47 is($charinfo->{decomposition},  '0041 0304');
  48 is($charinfo->{decimal},        '');
  49 is($charinfo->{digit},          '');
  50 is($charinfo->{numeric},        '');
  51 is($charinfo->{mirrored},       'N');
  52 is($charinfo->{unicode10},      'LATIN CAPITAL LETTER A MACRON');
  53 is($charinfo->{comment},        '');
  54 is($charinfo->{upper},          '');
  55 is($charinfo->{lower},          '0101');
  56 is($charinfo->{title},          '');
  57 is($charinfo->{block},          'Latin Extended-A');
  58 is($charinfo->{script},         'Latin');
  59
  60 # 0x0590 is in the Hebrew block but unused.
  61
  62 $charinfo = charinfo(0x590);
  63
  64 is($charinfo->{code},          undef,   '0x0590 - unused Hebrew');
  65 is($charinfo->{name},          undef);
  66 is($charinfo->{category},      undef);
  67 is($charinfo->{combining},     undef);
  68 is($charinfo->{bidi},          undef);
  69 is($charinfo->{decomposition}, undef);
  70 is($charinfo->{decimal},       undef);
  71 is($charinfo->{digit},         undef);
  72 is($charinfo->{numeric},       undef);
  73 is($charinfo->{mirrored},      undef);
  74 is($charinfo->{unicode10},     undef);
  75 is($charinfo->{comment},       undef);
  76 is($charinfo->{upper},         undef);
  77 is($charinfo->{lower},         undef);
  78 is($charinfo->{title},         undef);
  79 is($charinfo->{block},         undef);
  80 is($charinfo->{script},        undef);
  81
  82 # 0x05d0 is in the Hebrew block and used.
  83
  84 $charinfo = charinfo(0x5d0);
  85
  86 is($charinfo->{code},           '05D0', '05D0 - used Hebrew');
  87 is($charinfo->{name},           'HEBREW LETTER ALEF');
  88 is($charinfo->{category},       'Lo');
  89 is($charinfo->{combining},      '0');
  90 is($charinfo->{bidi},           'R');
  91 is($charinfo->{decomposition},  '');
  92 is($charinfo->{decimal},        '');
  93 is($charinfo->{digit},          '');
  94 is($charinfo->{numeric},        '');
  95 is($charinfo->{mirrored},       'N');
  96 is($charinfo->{unicode10},      '');
  97 is($charinfo->{comment},        '');
  98 is($charinfo->{upper},          '');
  99 is($charinfo->{lower},          '');
 100 is($charinfo->{title},          '');
 101 is($charinfo->{block},          'Hebrew');
 102 is($charinfo->{script},         'Hebrew');
 103
 104 # An open syllable in Hangul.
 105
 106 $charinfo = charinfo(0xAC00);
 107
 108 is($charinfo->{code},           'AC00', 'HANGUL SYLLABLE-AC00');
 109 is($charinfo->{name},           'HANGUL SYLLABLE-AC00');
 110 is($charinfo->{category},       'Lo');
 111 is($charinfo->{combining},      '0');
 112 is($charinfo->{bidi},           'L');
 113 is($charinfo->{decomposition},  undef);
 114 is($charinfo->{decimal},        '');
 115 is($charinfo->{digit},          '');
 116 is($charinfo->{numeric},        '');
 117 is($charinfo->{mirrored},       'N');
 118 is($charinfo->{unicode10},      '');
 119 is($charinfo->{comment},        '');
 120 is($charinfo->{upper},          '');
 121 is($charinfo->{lower},          '');
 122 is($charinfo->{title},          '');
 123 is($charinfo->{block},          'Hangul Syllables');
 124 is($charinfo->{script},         'Hangul');
 125
 126 # A closed syllable in Hangul.
 127
 128 $charinfo = charinfo(0xAE00);
 129
 130 is($charinfo->{code},           'AE00', 'HANGUL SYLLABLE-AE00');
 131 is($charinfo->{name},           'HANGUL SYLLABLE-AE00');
 132 is($charinfo->{category},       'Lo');
 133 is($charinfo->{combining},      '0');
 134 is($charinfo->{bidi},           'L');
 135 is($charinfo->{decomposition},  undef);
 136 is($charinfo->{decimal},        '');
 137 is($charinfo->{digit},          '');
 138 is($charinfo->{numeric},        '');
 139 is($charinfo->{mirrored},       'N');
 140 is($charinfo->{unicode10},      '');
 141 is($charinfo->{comment},        '');
 142 is($charinfo->{upper},          '');
 143 is($charinfo->{lower},          '');
 144 is($charinfo->{title},          '');
 145 is($charinfo->{block},          'Hangul Syllables');
 146 is($charinfo->{script},         'Hangul');
 147
 148 $charinfo = charinfo(0x1D400);
 149
 150 is($charinfo->{code},           '1D400', 'MATHEMATICAL BOLD CAPITAL A');
 151 is($charinfo->{name},           'MATHEMATICAL BOLD CAPITAL A');
 152 is($charinfo->{category},       'Lu');
 153 is($charinfo->{combining},      '0');
 154 is($charinfo->{bidi},           'L');
 155 is($charinfo->{decomposition},  '<font> 0041');
 156 is($charinfo->{decimal},        '');
 157 is($charinfo->{digit},          '');
 158 is($charinfo->{numeric},        '');
 159 is($charinfo->{mirrored},       'N');
 160 is($charinfo->{unicode10},      '');
 161 is($charinfo->{comment},        '');
 162 is($charinfo->{upper},          '');
 163 is($charinfo->{lower},          '');
 164 is($charinfo->{title},          '');
 165 is($charinfo->{block},          'Mathematical Alphanumeric Symbols');
 166 is($charinfo->{script},         undef);
 167
 168 use Unicode::UCD qw(charblock charscript);
 169
 170 # 0x0590 is in the Hebrew block but unused.
 171
 172 is(charblock(0x590),          'Hebrew', '0x0590 - Hebrew unused charblock');
 173 is(charscript(0x590),         undef,    '0x0590 - Hebrew unused charscript');
 174
 175 $charinfo = charinfo(0xbe);
 176
 177 is($charinfo->{code},           '00BE', 'VULGAR FRACTION THREE QUARTERS');
 178 is($charinfo->{name},           'VULGAR FRACTION THREE QUARTERS');
 179 is($charinfo->{category},       'No');
 180 is($charinfo->{combining},      '0');
 181 is($charinfo->{bidi},           'ON');
 182 is($charinfo->{decomposition},  '<fraction> 0033 2044 0034');
 183 is($charinfo->{decimal},        '');
 184 is($charinfo->{digit},          '');
 185 is($charinfo->{numeric},        '3/4');
 186 is($charinfo->{mirrored},       'N');
 187 is($charinfo->{unicode10},      'FRACTION THREE QUARTERS');
 188 is($charinfo->{comment},        '');
 189 is($charinfo->{upper},          '');
 190 is($charinfo->{lower},          '');
 191 is($charinfo->{title},          '');
 192 is($charinfo->{block},          'Latin-1 Supplement');
 193 is($charinfo->{script},         undef);
 194
 195 use Unicode::UCD qw(charblocks charscripts);
 196
 197 my $charblocks = charblocks();
 198
 199 ok(exists $charblocks->{Thai}, 'Thai charblock exists');
 200 is($charblocks->{Thai}->[0]->[0], hex('0e00'));
 201 ok(!exists $charblocks->{PigLatin}, 'PigLatin charblock does not exist');
 202
 203 my $charscripts = charscripts();
 204
 205 ok(exists $charscripts->{Armenian}, 'Armenian charscript exists');
 206 is($charscripts->{Armenian}->[0]->[0], hex('0531'));
 207 ok(!exists $charscripts->{PigLatin}, 'PigLatin charscript does not exist');
 208
 209 my $charscript;
 210
 211 $charscript = charscript("12ab");
 212 is($charscript, 'Ethiopic', 'Ethiopic charscript');
 213
 214 $charscript = charscript("0x12ab");
 215 is($charscript, 'Ethiopic');
 216
 217 $charscript = charscript("U+12ab");
 218 is($charscript, 'Ethiopic');
 219
 220 my $ranges;
 221
 222 $ranges = charscript('Ogham');
 223 is($ranges->[0]->[0], hex('1681'), 'Ogham charscript');
 224 is($ranges->[0]->[1], hex('169a'));
 225
 226 use Unicode::UCD qw(charinrange);
 227
 228 $ranges = charscript('Cherokee');
 229 ok(!charinrange($ranges, "139f"), 'Cherokee charscript');
 230 ok( charinrange($ranges, "13a0"));
 231 ok( charinrange($ranges, "13f4"));
 232 ok(!charinrange($ranges, "13f5"));
 233
 234 is(Unicode::UCD::UnicodeVersion, '3.1.1', 'UnicodeVersion');
 235
 236 use Unicode::UCD qw(compexcl);
 237
 238 ok(!compexcl(0x0100), 'compexcl');
 239 ok( compexcl(0x0958));
 240
 241 use Unicode::UCD qw(casefold);
 242
 243 my $casefold;
 244
 245 $casefold = casefold(0x41);
 246
 247 ok($casefold->{code} eq '0041' &&
 248    $casefold->{status} eq 'C'  &&
 249    $casefold->{mapping} eq '0061', 'casefold 0x41');
 250
 251 $casefold = casefold(0xdf);
 252
 253 ok($casefold->{code} eq '00DF' &&
 254    $casefold->{status} eq 'F'  &&
 255    $casefold->{mapping} eq '0073 0073', 'casefold 0xDF');
 256
 257 ok(!casefold(0x20));
 258
 259 use Unicode::UCD qw(casespec);
 260
 261 my $casespec;
 262
 263 ok(!casespec(0x41));
 264
 265 $casespec = casespec(0xdf);
 266
 267 ok($casespec->{code} eq '00DF' &&
 268    $casespec->{lower} eq '00DF'  &&
 269    $casespec->{title} eq '0053 0073'  &&
 270    $casespec->{upper} eq '0053 0053' &&
 271    $casespec->{condition} eq undef, 'casespec 0xDF');
 272
 273 $casespec = casespec(0x307);
 274
 275 ok($casespec->{az}->{code} eq '0307' &&
 276    $casespec->{az}->{lower} eq ''  &&
 277    $casespec->{az}->{title} eq '0307'  &&
 278    $casespec->{az}->{upper} eq '0307' &&
 279    $casespec->{az}->{condition} eq 'az AFTER_i NOT_MORE_ABOVE',
 280    'casespec 0x307');