1
2BEGIN {
3    unless ('A' eq pack('U', 0x41)) {
4	print "1..0 # Unicode::Normalize cannot pack a Unicode code point\n";
5	exit 0;
6    }
7    unless (0x41 == unpack('U', 'A')) {
8	print "1..0 # Unicode::Normalize cannot get a Unicode code point\n";
9	exit 0;
10    }
11}
12
13BEGIN {
14    if ($ENV{PERL_CORE}) {
15        chdir('t') if -d 't';
16        @INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib);
17    }
18}
19
20#########################
21
22use strict;
23use warnings;
24BEGIN { $| = 1; print "1..64\n"; }
25my $count = 0;
26sub ok ($;$) {
27    my $p = my $r = shift;
28    if (@_) {
29	my $x = shift;
30	$p = !defined $x ? !defined $r : !defined $r ? 0 : $r eq $x;
31    }
32    print $p ? "ok" : "not ok", ' ', ++$count, "\n";
33}
34
35use Unicode::Normalize qw(normalize);
36
37ok(1);
38
39sub _pack_U   { Unicode::Normalize::pack_U(@_) }
40sub _unpack_U { Unicode::Normalize::unpack_U(@_) }
41
42#########################
43
44ok(normalize('D', ""), "");
45ok(normalize('C', ""), "");
46ok(normalize('KD',""), "");
47ok(normalize('KC',""), "");
48
49ok(normalize('D', "A"), "A");
50ok(normalize('C', "A"), "A");
51ok(normalize('KD',"A"), "A");
52ok(normalize('KC',"A"), "A");
53
54ok(normalize('NFD', ""), "");
55ok(normalize('NFC', ""), "");
56ok(normalize('NFKD',""), "");
57ok(normalize('NFKC',""), "");
58
59ok(normalize('NFD', "A"), "A");
60ok(normalize('NFC', "A"), "A");
61ok(normalize('NFKD',"A"), "A");
62ok(normalize('NFKC',"A"), "A");
63
64# 17
65
66# don't modify the source
67my $sNFD = "\x{FA19}";
68ok(normalize('NFD', $sNFD), "\x{795E}");
69ok($sNFD, "\x{FA19}");
70
71my $sNFC = "\x{FA1B}";
72ok(normalize('NFC', $sNFC), "\x{798F}");
73ok($sNFC, "\x{FA1B}");
74
75my $sNFKD = "\x{FA1E}";
76ok(normalize('NFKD', $sNFKD), "\x{7FBD}");
77ok($sNFKD, "\x{FA1E}");
78
79my $sNFKC = "\x{FA26}";
80ok(normalize('NFKC', $sNFKC), "\x{90FD}");
81ok($sNFKC, "\x{FA26}");
82
83# 25
84
85sub hexNFC {
86  join " ", map sprintf("%04X", $_),
87  _unpack_U normalize 'C', _pack_U map hex, split ' ', shift;
88}
89sub hexNFD {
90  join " ", map sprintf("%04X", $_),
91  _unpack_U normalize 'D', _pack_U map hex, split ' ', shift;
92}
93
94ok(hexNFD("1E14 AC01"), "0045 0304 0300 1100 1161 11A8");
95ok(hexNFD("AC00 AE00"), "1100 1161 1100 1173 11AF");
96
97ok(hexNFC("0061 0315 0300 05AE 05C4 0062"), "00E0 05AE 05C4 0315 0062");
98ok(hexNFC("00E0 05AE 05C4 0315 0062"),      "00E0 05AE 05C4 0315 0062");
99ok(hexNFC("0061 05AE 0300 05C4 0315 0062"), "00E0 05AE 05C4 0315 0062");
100ok(hexNFC("0045 0304 0300 AC00 11A8"), "1E14 AC01");
101ok(hexNFC("1100 1161 1100 1173 11AF"), "AC00 AE00");
102ok(hexNFC("1100 0300 1161 1173 11AF"), "1100 0300 1161 1173 11AF");
103
104ok(hexNFD("0061 0315 0300 05AE 05C4 0062"), "0061 05AE 0300 05C4 0315 0062");
105ok(hexNFD("00E0 05AE 05C4 0315 0062"),      "0061 05AE 0300 05C4 0315 0062");
106ok(hexNFD("0061 05AE 0300 05C4 0315 0062"), "0061 05AE 0300 05C4 0315 0062");
107ok(hexNFC("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
108ok(hexNFC("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
109ok(hexNFD("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
110ok(hexNFD("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
111ok(hexNFC("0000 0041 0000 0000"), "0000 0041 0000 0000");
112ok(hexNFD("0000 0041 0000 0000"), "0000 0041 0000 0000");
113
114ok(hexNFC("AC00 11A7"), "AC00 11A7");
115ok(hexNFC("AC00 11A8"), "AC01");
116ok(hexNFC("AC00 11A9"), "AC02");
117ok(hexNFC("AC00 11C2"), "AC1B");
118ok(hexNFC("AC00 11C3"), "AC00 11C3");
119
120# 47
121
122# Test Cases from Public Review Issue #29: Normalization Issue
123# cf. http://www.unicode.org/review/pr-29.html
124ok(hexNFC("0B47 0300 0B3E"), "0B47 0300 0B3E");
125ok(hexNFC("1100 0300 1161"), "1100 0300 1161");
126ok(hexNFC("0B47 0B3E 0300"), "0B4B 0300");
127ok(hexNFC("1100 1161 0300"), "AC00 0300");
128ok(hexNFC("0B47 0300 0B3E 0327"), "0B47 0300 0B3E 0327");
129ok(hexNFC("1100 0300 1161 0327"), "1100 0300 1161 0327");
130
131ok(hexNFC("0300 0041"), "0300 0041");
132ok(hexNFC("0300 0301 0041"), "0300 0301 0041");
133ok(hexNFC("0301 0300 0041"), "0301 0300 0041");
134ok(hexNFC("0000 0300 0000 0301"), "0000 0300 0000 0301");
135ok(hexNFC("0000 0301 0000 0300"), "0000 0301 0000 0300");
136
137ok(hexNFC("0327 0061 0300"), "0327 00E0");
138ok(hexNFC("0301 0061 0300"), "0301 00E0");
139ok(hexNFC("0315 0061 0300"), "0315 00E0");
140ok(hexNFC("0000 0327 0061 0300"), "0000 0327 00E0");
141ok(hexNFC("0000 0301 0061 0300"), "0000 0301 00E0");
142ok(hexNFC("0000 0315 0061 0300"), "0000 0315 00E0");
143
144# 64
145
146