xref: /openbsd-src/gnu/usr.bin/perl/cpan/Unicode-Collate/t/cjkrange.t (revision 50b7afb2c2c0993b0894d4e34bf857cb13ed9c80)
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..421\n"; } # 1 + 42 x @Versions
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 $coll = Unicode::Collate->new(
34    table => 'keys.txt',
35    normalization => undef,
36);
37
38# CJK UI Ext > CJK UI.
39# [ UCA_Version 8: Ext.A < UI and BMP < Ext.B (code point order) ]
40
41# 4E00..9FA5 are CJK UI.
42# 9FA6..9FBB are CJK UI since UCA_Version 14 (Unicode 4.1).
43# 9FBC..9FC3 are CJK UI since UCA_Version 18 (Unicode 5.1).
44# 9FC4..9FCB are CJK UI since UCA_Version 20 (Unicode 5.2).
45# 9FCC       is  CJK UI since UCA_Version 24 (Unicode 6.1).
46
47# 3400..4DB5   are CJK UI Ext.A since UCA_Version 8  (Unicode 3.0).
48# 20000..2A6D6 are CJK UI Ext.B since UCA_Version 8  (Unicode 3.1).
49# 2A700..2B734 are CJK UI Ext.C since UCA_Version 20 (Unicode 5.2).
50# 2B740..2B81D are CJK UI Ext.D since UCA_Version 22 (Unicode 6.0).
51
52my @Versions = (8, 9, 11, 14, 16, 18, 20, 22, 24, 26);
53
54for my $v (@Versions) {
55    $coll->change(UCA_Version => $v);
56
57    # Ext.A > UI
58    ok($coll->cmp("\x{3400}", "\x{4E00}") == ($v >=  9 ? 1 : -1)); # UI
59    ok($coll->cmp("\x{3400}", "\x{9FA5}") == ($v >=  9 ? 1 : -1)); # UI
60    ok($coll->cmp("\x{3400}", "\x{9FA6}") == ($v >= 14 ? 1 : -1)); # new
61    ok($coll->cmp("\x{3400}", "\x{9FBB}") == ($v >= 14 ? 1 : -1)); # new
62    ok($coll->cmp("\x{3400}", "\x{9FBC}") == ($v >= 18 ? 1 : -1)); # new
63    ok($coll->cmp("\x{3400}", "\x{9FC3}") == ($v >= 18 ? 1 : -1)); # new
64    ok($coll->cmp("\x{3400}", "\x{9FC4}") == ($v >= 20 ? 1 : -1)); # new
65    ok($coll->cmp("\x{3400}", "\x{9FCB}") == ($v >= 20 ? 1 : -1)); # new
66    ok($coll->cmp("\x{3400}", "\x{9FCC}") == ($v >= 24 ? 1 : -1)); # new
67    ok($coll->cmp("\x{3400}", "\x{9FCD}") == -1); # na
68    ok($coll->cmp("\x{3400}", "\x{9FFF}") == -1); # na
69
70    # UI < UI
71    ok($coll->cmp("\x{4E00}", "\x{9FA5}") == -1); # UI < UI
72    ok($coll->cmp("\x{9FA5}", "\x{9FA6}") == -1); # UI < new
73    ok($coll->cmp("\x{9FA6}", "\x{9FBB}") == -1); # new < new
74    ok($coll->cmp("\x{9FBB}", "\x{9FBC}") == -1); # new < new
75    ok($coll->cmp("\x{9FBC}", "\x{9FC3}") == -1); # new < new
76    ok($coll->cmp("\x{9FC3}", "\x{9FC4}") == -1); # new < new
77    ok($coll->cmp("\x{9FC4}", "\x{9FCB}") == -1); # new < new
78    ok($coll->cmp("\x{9FCB}", "\x{9FCC}") == -1); # new < new
79    ok($coll->cmp("\x{9FCC}", "\x{9FCD}") == -1); # new < na
80    ok($coll->cmp("\x{9FCD}", "\x{9FFF}") == -1); # na < na
81
82    # Ext.A < Ext.B
83    ok($coll->cmp("\x{3400}", "\x{20000}") == -1);
84
85    # Ext.A
86    ok($coll->cmp("\x{3400}", "\x{4DB5}") == -1); # A < A
87    ok($coll->cmp("\x{2FFF}", "\x{3400}") == ($v >= 8 ? 1 : -1)); # na > A
88    ok($coll->cmp("\x{2FFF}", "\x{4DB5}") == ($v >= 8 ? 1 : -1)); # na > A
89    ok($coll->cmp("\x{2FFF}", "\x{4DB6}") == -1); # na < na
90    ok($coll->cmp("\x{2FFF}", "\x{4DBF}") == -1); # na < na
91
92    # Ext.B
93    ok($coll->cmp("\x{20000}","\x{2A6D6}") == -1); # B < B
94    ok($coll->cmp("\x{2FFF}", "\x{20000}") == ($v >= 9 ? 1 : -1)); # na > B
95    ok($coll->cmp("\x{2FFF}", "\x{2A6D6}") == ($v >= 9 ? 1 : -1)); # na > B
96    ok($coll->cmp("\x{2FFF}", "\x{2A6D7}") == -1); # na < na
97    ok($coll->cmp("\x{2FFF}", "\x{2A6DF}") == -1); # na < na
98
99    # Ext.C
100    ok($coll->cmp("\x{2A700}","\x{2B734}") == -1); # C < C
101    ok($coll->cmp("\x{2FFF}", "\x{2A700}") == ($v >= 20 ? 1 : -1)); # na > C
102    ok($coll->cmp("\x{2FFF}", "\x{2B734}") == ($v >= 20 ? 1 : -1)); # na > C
103    ok($coll->cmp("\x{2FFF}", "\x{2B735}") == -1); # na < na
104    ok($coll->cmp("\x{2FFF}", "\x{2B73F}") == -1); # na < na
105
106    # Ext.D
107    ok($coll->cmp("\x{2B740}","\x{2B81D}") == -1); # D < D
108    ok($coll->cmp("\x{2FFF}", "\x{2B740}") == ($v >= 22 ? 1 : -1)); # na > D
109    ok($coll->cmp("\x{2FFF}", "\x{2B81D}") == ($v >= 22 ? 1 : -1)); # na > D
110    ok($coll->cmp("\x{2FFF}", "\x{2B81E}") == -1); # na < na
111    ok($coll->cmp("\x{2FFF}", "\x{2B81F}") == -1); # na < na
112}
113
114