Upgrade to Unicode-Normalize-1.00
[p5sagit/p5-mst-13.2.git] / ext / Unicode / Normalize / t / test.t
CommitLineData
45394607 1
4a2e806c 2BEGIN {
1efaba7f 3 unless ("A" eq pack('U', 0x41)) {
9f1f04a1 4 print "1..0 # Unicode::Normalize " .
5 "cannot stringify a Unicode code point\n";
4a2e806c 6 exit 0;
7 }
8}
9
6c941e0c 10BEGIN {
11 if ($ENV{PERL_CORE}) {
12 chdir('t') if -d 't';
9f1f04a1 13 @INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib);
6c941e0c 14 }
15}
16
45394607 17#########################
18
19use Test;
20use strict;
21use warnings;
fe067ad9 22BEGIN { plan tests => 58 };
45394607 23use Unicode::Normalize;
24ok(1); # If we made it this far, we're ok.
25
9f1f04a1 26sub _pack_U { Unicode::Normalize::pack_U(@_) }
27sub _unpack_U { Unicode::Normalize::unpack_U(@_) }
6c941e0c 28
45394607 29#########################
30
45394607 31ok(NFD(""), "");
fe067ad9 32ok(NFC(""), "");
33ok(NFKD(""), "");
34ok(NFKC(""), "");
35
36ok(NFD("A"), "A");
37ok(NFC("A"), "A");
38ok(NFKD("A"), "A");
39ok(NFKC("A"), "A");
40
41# don't modify the source
42# don't modify the source
43my $sNFD = "\x{FA19}";
44ok(NFD($sNFD), "\x{795E}");
45ok($sNFD, "\x{FA19}");
46
47my $sNFC = "\x{FA1B}";
48ok(NFC($sNFC), "\x{798F}");
49ok($sNFC, "\x{FA1B}");
50
51my $sNFKD = "\x{FA1E}";
52ok(NFKD($sNFKD), "\x{7FBD}");
53ok($sNFKD, "\x{FA1E}");
54
55my $sNFKC = "\x{FA26}";
56ok(NFKC($sNFKC), "\x{90FD}");
57ok($sNFKC, "\x{FA26}");
58
45394607 59
60sub hexNFC {
61 join " ", map sprintf("%04X", $_),
6c941e0c 62 _unpack_U NFC _pack_U map hex, split ' ', shift;
45394607 63}
64sub hexNFD {
65 join " ", map sprintf("%04X", $_),
6c941e0c 66 _unpack_U NFD _pack_U map hex, split ' ', shift;
45394607 67}
68
fe067ad9 69ok(hexNFD("1E14 AC01"), "0045 0304 0300 1100 1161 11A8");
70ok(hexNFD("AC00 AE00"), "1100 1161 1100 1173 11AF");
71
4a2e806c 72ok(hexNFC("0061 0315 0300 05AE 05C4 0062"), "00E0 05AE 05C4 0315 0062");
73ok(hexNFC("00E0 05AE 05C4 0315 0062"), "00E0 05AE 05C4 0315 0062");
74ok(hexNFC("0061 05AE 0300 05C4 0315 0062"), "00E0 05AE 05C4 0315 0062");
ac5ea531 75ok(hexNFC("0045 0304 0300 AC00 11A8"), "1E14 AC01");
76ok(hexNFC("1100 1161 1100 1173 11AF"), "AC00 AE00");
77ok(hexNFC("1100 0300 1161 1173 11AF"), "1100 0300 1161 1173 11AF");
45394607 78
79ok(hexNFD("0061 0315 0300 05AE 05C4 0062"), "0061 05AE 0300 05C4 0315 0062");
80ok(hexNFD("00E0 05AE 05C4 0315 0062"), "0061 05AE 0300 05C4 0315 0062");
81ok(hexNFD("0061 05AE 0300 05C4 0315 0062"), "0061 05AE 0300 05C4 0315 0062");
4a2e806c 82ok(hexNFC("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
83ok(hexNFC("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
45394607 84ok(hexNFD("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
85ok(hexNFD("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
4a2e806c 86ok(hexNFC("0000 0041 0000 0000"), "0000 0041 0000 0000");
45394607 87ok(hexNFD("0000 0041 0000 0000"), "0000 0041 0000 0000");
88
628bbff0 89ok(hexNFC("AC00 11A7"), "AC00 11A7");
90ok(hexNFC("AC00 11A8"), "AC01");
91ok(hexNFC("AC00 11A9"), "AC02");
92ok(hexNFC("AC00 11C2"), "AC1B");
93ok(hexNFC("AC00 11C3"), "AC00 11C3");
94
95# Test Cases from Public Review Issue #29: Normalization Issue
96# cf. http://www.unicode.org/review/pr-29.html
97ok(hexNFC("0B47 0300 0B3E"), "0B47 0300 0B3E");
98ok(hexNFC("1100 0300 1161"), "1100 0300 1161");
628bbff0 99ok(hexNFC("0B47 0B3E 0300"), "0B4B 0300");
100ok(hexNFC("1100 1161 0300"), "AC00 0300");
628bbff0 101ok(hexNFC("0B47 0300 0B3E 0327"), "0B47 0300 0B3E 0327");
102ok(hexNFC("1100 0300 1161 0327"), "1100 0300 1161 0327");
103
fe067ad9 104ok(hexNFC("0300 0041"), "0300 0041");
105ok(hexNFC("0300 0301 0041"), "0300 0301 0041");
106ok(hexNFC("0301 0300 0041"), "0301 0300 0041");
107ok(hexNFC("0000 0300 0000 0301"), "0000 0300 0000 0301");
108ok(hexNFC("0000 0301 0000 0300"), "0000 0301 0000 0300");
109
110ok(hexNFC("0327 0061 0300"), "0327 00E0");
111ok(hexNFC("0301 0061 0300"), "0301 00E0");
112ok(hexNFC("0315 0061 0300"), "0315 00E0");
113ok(hexNFC("0000 0327 0061 0300"), "0000 0327 00E0");
114ok(hexNFC("0000 0301 0061 0300"), "0000 0301 00E0");
115ok(hexNFC("0000 0315 0061 0300"), "0000 0315 00E0");
116
628bbff0 117# NFC() should be unary.
6c941e0c 118my $str11 = _pack_U(0x41, 0x0302, 0x0301, 0x62);
119my $str12 = _pack_U(0x1EA4, 0x62);
ab8fe378 120ok(NFC $str11 eq $str12);
121
628bbff0 122# NFD() should be unary.
6c941e0c 123my $str21 = _pack_U(0xE0, 0xAC00);
124my $str22 = _pack_U(0x61, 0x0300, 0x1100, 0x1161);
ab8fe378 125ok(NFD $str21 eq $str22);
6c941e0c 126