Rename ext/Unicode/Normalize to ext/Unicode-Normalize
[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
fe067ad9 42my $sNFD = "\x{FA19}";
43ok(NFD($sNFD), "\x{795E}");
44ok($sNFD, "\x{FA19}");
45
46my $sNFC = "\x{FA1B}";
47ok(NFC($sNFC), "\x{798F}");
48ok($sNFC, "\x{FA1B}");
49
50my $sNFKD = "\x{FA1E}";
51ok(NFKD($sNFKD), "\x{7FBD}");
52ok($sNFKD, "\x{FA1E}");
53
54my $sNFKC = "\x{FA26}";
55ok(NFKC($sNFKC), "\x{90FD}");
56ok($sNFKC, "\x{FA26}");
57
45394607 58
59sub hexNFC {
60 join " ", map sprintf("%04X", $_),
6c941e0c 61 _unpack_U NFC _pack_U map hex, split ' ', shift;
45394607 62}
63sub hexNFD {
64 join " ", map sprintf("%04X", $_),
6c941e0c 65 _unpack_U NFD _pack_U map hex, split ' ', shift;
45394607 66}
67
fe067ad9 68ok(hexNFD("1E14 AC01"), "0045 0304 0300 1100 1161 11A8");
69ok(hexNFD("AC00 AE00"), "1100 1161 1100 1173 11AF");
70
4a2e806c 71ok(hexNFC("0061 0315 0300 05AE 05C4 0062"), "00E0 05AE 05C4 0315 0062");
72ok(hexNFC("00E0 05AE 05C4 0315 0062"), "00E0 05AE 05C4 0315 0062");
73ok(hexNFC("0061 05AE 0300 05C4 0315 0062"), "00E0 05AE 05C4 0315 0062");
ac5ea531 74ok(hexNFC("0045 0304 0300 AC00 11A8"), "1E14 AC01");
75ok(hexNFC("1100 1161 1100 1173 11AF"), "AC00 AE00");
76ok(hexNFC("1100 0300 1161 1173 11AF"), "1100 0300 1161 1173 11AF");
45394607 77
78ok(hexNFD("0061 0315 0300 05AE 05C4 0062"), "0061 05AE 0300 05C4 0315 0062");
79ok(hexNFD("00E0 05AE 05C4 0315 0062"), "0061 05AE 0300 05C4 0315 0062");
80ok(hexNFD("0061 05AE 0300 05C4 0315 0062"), "0061 05AE 0300 05C4 0315 0062");
4a2e806c 81ok(hexNFC("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
82ok(hexNFC("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
45394607 83ok(hexNFD("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
84ok(hexNFD("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
4a2e806c 85ok(hexNFC("0000 0041 0000 0000"), "0000 0041 0000 0000");
45394607 86ok(hexNFD("0000 0041 0000 0000"), "0000 0041 0000 0000");
87
628bbff0 88ok(hexNFC("AC00 11A7"), "AC00 11A7");
89ok(hexNFC("AC00 11A8"), "AC01");
90ok(hexNFC("AC00 11A9"), "AC02");
91ok(hexNFC("AC00 11C2"), "AC1B");
92ok(hexNFC("AC00 11C3"), "AC00 11C3");
93
94# Test Cases from Public Review Issue #29: Normalization Issue
95# cf. http://www.unicode.org/review/pr-29.html
96ok(hexNFC("0B47 0300 0B3E"), "0B47 0300 0B3E");
97ok(hexNFC("1100 0300 1161"), "1100 0300 1161");
628bbff0 98ok(hexNFC("0B47 0B3E 0300"), "0B4B 0300");
99ok(hexNFC("1100 1161 0300"), "AC00 0300");
628bbff0 100ok(hexNFC("0B47 0300 0B3E 0327"), "0B47 0300 0B3E 0327");
101ok(hexNFC("1100 0300 1161 0327"), "1100 0300 1161 0327");
102
fe067ad9 103ok(hexNFC("0300 0041"), "0300 0041");
104ok(hexNFC("0300 0301 0041"), "0300 0301 0041");
105ok(hexNFC("0301 0300 0041"), "0301 0300 0041");
106ok(hexNFC("0000 0300 0000 0301"), "0000 0300 0000 0301");
107ok(hexNFC("0000 0301 0000 0300"), "0000 0301 0000 0300");
108
109ok(hexNFC("0327 0061 0300"), "0327 00E0");
110ok(hexNFC("0301 0061 0300"), "0301 00E0");
111ok(hexNFC("0315 0061 0300"), "0315 00E0");
112ok(hexNFC("0000 0327 0061 0300"), "0000 0327 00E0");
113ok(hexNFC("0000 0301 0061 0300"), "0000 0301 00E0");
114ok(hexNFC("0000 0315 0061 0300"), "0000 0315 00E0");
115
628bbff0 116# NFC() should be unary.
6c941e0c 117my $str11 = _pack_U(0x41, 0x0302, 0x0301, 0x62);
118my $str12 = _pack_U(0x1EA4, 0x62);
ab8fe378 119ok(NFC $str11 eq $str12);
120
628bbff0 121# NFD() should be unary.
6c941e0c 122my $str21 = _pack_U(0xE0, 0xAC00);
123my $str22 = _pack_U(0x61, 0x0300, 0x1100, 0x1161);
ab8fe378 124ok(NFD $str21 eq $str22);
6c941e0c 125