1BEGIN {
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    }
7    if ($ENV{PERL_CORE}) {
8	chdir('t') if -d 't';
9	@INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib);
10    }
11}
12
13use Test;
14BEGIN { plan tests => 41 };
15
16use strict;
17use warnings;
18use Unicode::Collate;
19
20ok(1);
21
22my $trad = Unicode::Collate->new(
23  table => 'keys.txt',
24  normalization => undef,
25  ignoreName => qr/HANGUL|HIRAGANA|KATAKANA|BOPOMOFO/,
26  level => 3,
27  entry => << 'ENTRIES',
28 0063 0068 ; [.0A3F.0020.0002.0063] % "ch" in traditional Spanish
29 0043 0068 ; [.0A3F.0020.0007.0043] # "Ch" in traditional Spanish
30 0043 0048 ; [.0A3F.0020.0008.0043] # "CH" in traditional Spanish
31ENTRIES
32);
33# 0063  ; [.0A3D.0020.0002.0063] # LATIN SMALL LETTER C
34# 0064  ; [.0A49.0020.0002.0064] # LATIN SMALL LETTER D
35
36##### 2..3
37
38ok(
39  join(':', $trad->sort( qw/ acha aca ada acia acka / ) ),
40  join(':',              qw/ aca acia acka acha ada / ),
41);
42
43ok(
44  join(':', $trad->sort( qw/ ACHA ACA ADA ACIA ACKA / ) ),
45  join(':',              qw/ ACA ACIA ACKA ACHA ADA / ),
46);
47
48##### 4..7
49
50ok($trad->gt("ocho", "oc\cAho")); # UCA v14
51ok($trad->gt("ocho", "oc\0\cA\0\cBho"));  # UCA v14
52ok($trad->eq("-", ""));
53ok($trad->gt("ocho", "oc-ho"));
54
55##### 8..11
56
57$trad->change(UCA_Version => 9);
58
59ok($trad->eq("ocho", "oc\cAho")); # UCA v9
60ok($trad->eq("ocho", "oc\0\cA\0\cBho")); # UCA v9
61ok($trad->eq("-", ""));
62ok($trad->gt("ocho", "oc-ho"));
63
64##### 12..15
65
66$trad->change(UCA_Version => 8);
67
68ok($trad->gt("ocho", "oc\cAho"));
69ok($trad->gt("ocho", "oc\0\cA\0\cBho"));
70ok($trad->eq("-", ""));
71ok($trad->gt("ocho", "oc-ho"));
72
73
74##### 16..19
75
76$trad->change(UCA_Version => 9);
77
78my $hiragana = "\x{3042}\x{3044}";
79my $katakana = "\x{30A2}\x{30A4}";
80
81# HIRAGANA and KATAKANA are ignorable via ignoreName
82ok($trad->eq($hiragana, ""));
83ok($trad->eq("", $katakana));
84ok($trad->eq($hiragana, $katakana));
85ok($trad->eq($katakana, $hiragana));
86
87
88##### 20..31
89
90# According to Conformance Test (UCA_Version == 9 or 11),
91# a L3-ignorable is treated as a completely ignorable.
92
93my $L3ignorable = Unicode::Collate->new(
94  alternate => 'Non-ignorable',
95  level => 3,
96  table => undef,
97  normalization => undef,
98  UCA_Version => 9,
99  entry => <<'ENTRIES',
1000000  ; [.0000.0000.0000.0000] # [0000] NULL (in 6429)
1010001  ; [.0000.0000.0000.0000] # [0001] START OF HEADING (in 6429)
1020591  ; [.0000.0000.0000.0591] # HEBREW ACCENT ETNAHTA
1031D165 ; [.0000.0000.0000.1D165] # MUSICAL SYMBOL COMBINING STEM
1040021  ; [*024B.0020.0002.0021] # EXCLAMATION MARK
10509BE  ; [.114E.0020.0002.09BE] # BENGALI VOWEL SIGN AA
10609C7  ; [.1157.0020.0002.09C7] # BENGALI VOWEL SIGN E
10709CB  ; [.1159.0020.0002.09CB] # BENGALI VOWEL SIGN O
10809C7 09BE ; [.1159.0020.0002.09CB] # BENGALI VOWEL SIGN O
1091D1B9 ; [*098A.0020.0002.1D1B9] # MUSICAL SYMBOL SEMIBREVIS WHITE
1101D1BA ; [*098B.0020.0002.1D1BA] # MUSICAL SYMBOL SEMIBREVIS BLACK
1111D1BB ; [*098A.0020.0002.1D1B9][.0000.0000.0000.1D165] # M.S. MINIMA
1121D1BC ; [*098B.0020.0002.1D1BA][.0000.0000.0000.1D165] # M.S. MINIMA BLACK
113ENTRIES
114);
115
116ok($L3ignorable->lt("\cA", "!"));
117ok($L3ignorable->lt("\x{591}", "!"));
118ok($L3ignorable->eq("\cA", "\x{591}"));
119ok($L3ignorable->eq("\x{09C7}\x{09BE}A", "\x{09C7}\cA\x{09BE}A"));
120ok($L3ignorable->eq("\x{09C7}\x{09BE}A", "\x{09C7}\x{0591}\x{09BE}A"));
121ok($L3ignorable->eq("\x{09C7}\x{09BE}A", "\x{09C7}\x{1D165}\x{09BE}A"));
122ok($L3ignorable->eq("\x{09C7}\x{09BE}A", "\x{09CB}A"));
123ok($L3ignorable->lt("\x{1D1BB}", "\x{1D1BC}"));
124ok($L3ignorable->eq("\x{1D1BB}", "\x{1D1B9}"));
125ok($L3ignorable->eq("\x{1D1BC}", "\x{1D1BA}"));
126ok($L3ignorable->eq("\x{1D1BB}", "\x{1D1B9}\x{1D165}"));
127ok($L3ignorable->eq("\x{1D1BC}", "\x{1D1BA}\x{1D165}"));
128
129##### 32..41
130
131my $c = Unicode::Collate->new(
132  table => 'keys.txt',
133  normalization => undef,
134  level => 1,
135  UCA_Version => 14,
136  entry => << 'ENTRIES',
137034F  ; [.0000.0000.0000.034F] # COMBINING GRAPHEME JOINER
1380063 0068 ; [.0A3F.0020.0002.0063] % "ch" in traditional Spanish
1390043 0068 ; [.0A3F.0020.0007.0043] # "Ch" in traditional Spanish
1400043 0048 ; [.0A3F.0020.0008.0043] # "CH" in traditional Spanish
141ENTRIES
142);
143# 0063  ; [.0A3D.0020.0002.0063] # LATIN SMALL LETTER C
144# 0064  ; [.0A49.0020.0002.0064] # LATIN SMALL LETTER D
145
146ok($c->gt("ocho", "oc\x00\x00ho"));
147ok($c->gt("ocho", "oc\cAho"));
148ok($c->gt("ocho", "oc\x{034F}ho"));
149ok($c->gt("ocio", "oc\x{034F}ho"));
150ok($c->lt("ocgo", "oc\x{034F}ho"));
151ok($c->lt("oceo", "oc\x{034F}ho"));
152
153ok($c->viewSortKey("ocho"),         "[0B4B 0A3F 0B4B | | |]");
154ok($c->viewSortKey("oc\x00\x00ho"), "[0B4B 0A3D 0AB9 0B4B | | |]");
155ok($c->viewSortKey("oc\cAho"),      "[0B4B 0A3D 0AB9 0B4B | | |]");
156ok($c->viewSortKey("oc\x{034F}ho"), "[0B4B 0A3D 0AB9 0B4B | | |]");
157
158
159