Upgrade to Unicode-Normalize-1.00
[p5sagit/p5-mst-13.2.git] / ext / Unicode / Normalize / t / norm.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 => 64 };
45394607 23use Unicode::Normalize qw(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(normalize('D', ""), "");
fe067ad9 32ok(normalize('C', ""), "");
33ok(normalize('KD',""), "");
34ok(normalize('KC',""), "");
35
36ok(normalize('D', "A"), "A");
37ok(normalize('C', "A"), "A");
38ok(normalize('KD',"A"), "A");
39ok(normalize('KC',"A"), "A");
40
41ok(normalize('NFD', ""), "");
42ok(normalize('NFC', ""), "");
43ok(normalize('NFKD',""), "");
44ok(normalize('NFKC',""), "");
45
46ok(normalize('NFD', "A"), "A");
47ok(normalize('NFC', "A"), "A");
48ok(normalize('NFKD',"A"), "A");
49ok(normalize('NFKC',"A"), "A");
50
51# don't modify the source
52my $sNFD = "\x{FA19}";
53ok(normalize('NFD', $sNFD), "\x{795E}");
54ok($sNFD, "\x{FA19}");
55
56my $sNFC = "\x{FA1B}";
57ok(normalize('NFC', $sNFC), "\x{798F}");
58ok($sNFC, "\x{FA1B}");
59
60my $sNFKD = "\x{FA1E}";
61ok(normalize('NFKD', $sNFKD), "\x{7FBD}");
62ok($sNFKD, "\x{FA1E}");
63
64my $sNFKC = "\x{FA26}";
65ok(normalize('NFKC', $sNFKC), "\x{90FD}");
66ok($sNFKC, "\x{FA26}");
45394607 67
68sub hexNFC {
69 join " ", map sprintf("%04X", $_),
6c941e0c 70 _unpack_U normalize 'C', _pack_U map hex, split ' ', shift;
45394607 71}
72sub hexNFD {
73 join " ", map sprintf("%04X", $_),
6c941e0c 74 _unpack_U normalize 'D', _pack_U map hex, split ' ', shift;
45394607 75}
76
fe067ad9 77ok(hexNFD("1E14 AC01"), "0045 0304 0300 1100 1161 11A8");
78ok(hexNFD("AC00 AE00"), "1100 1161 1100 1173 11AF");
79
7ac4cfc1 80ok(hexNFC("0061 0315 0300 05AE 05C4 0062"), "00E0 05AE 05C4 0315 0062");
81ok(hexNFC("00E0 05AE 05C4 0315 0062"), "00E0 05AE 05C4 0315 0062");
82ok(hexNFC("0061 05AE 0300 05C4 0315 0062"), "00E0 05AE 05C4 0315 0062");
ac5ea531 83ok(hexNFC("0045 0304 0300 AC00 11A8"), "1E14 AC01");
84ok(hexNFC("1100 1161 1100 1173 11AF"), "AC00 AE00");
85ok(hexNFC("1100 0300 1161 1173 11AF"), "1100 0300 1161 1173 11AF");
45394607 86
87ok(hexNFD("0061 0315 0300 05AE 05C4 0062"), "0061 05AE 0300 05C4 0315 0062");
88ok(hexNFD("00E0 05AE 05C4 0315 0062"), "0061 05AE 0300 05C4 0315 0062");
89ok(hexNFD("0061 05AE 0300 05C4 0315 0062"), "0061 05AE 0300 05C4 0315 0062");
7ac4cfc1 90ok(hexNFC("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
91ok(hexNFC("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
45394607 92ok(hexNFD("0061 05C4 0315 0300 05AE 0062"), "0061 05AE 05C4 0300 0315 0062");
93ok(hexNFD("0061 05AE 05C4 0300 0315 0062"), "0061 05AE 05C4 0300 0315 0062");
7ac4cfc1 94ok(hexNFC("0000 0041 0000 0000"), "0000 0041 0000 0000");
45394607 95ok(hexNFD("0000 0041 0000 0000"), "0000 0041 0000 0000");
96
628bbff0 97ok(hexNFC("AC00 11A7"), "AC00 11A7");
98ok(hexNFC("AC00 11A8"), "AC01");
99ok(hexNFC("AC00 11A9"), "AC02");
100ok(hexNFC("AC00 11C2"), "AC1B");
101ok(hexNFC("AC00 11C3"), "AC00 11C3");
102
103# Test Cases from Public Review Issue #29: Normalization Issue
104# cf. http://www.unicode.org/review/pr-29.html
105ok(hexNFC("0B47 0300 0B3E"), "0B47 0300 0B3E");
106ok(hexNFC("1100 0300 1161"), "1100 0300 1161");
107
108ok(hexNFC("0B47 0B3E 0300"), "0B4B 0300");
109ok(hexNFC("1100 1161 0300"), "AC00 0300");
110
111ok(hexNFC("0B47 0300 0B3E 0327"), "0B47 0300 0B3E 0327");
112ok(hexNFC("1100 0300 1161 0327"), "1100 0300 1161 0327");
fe067ad9 113
114ok(hexNFC("0300 0041"), "0300 0041");
115ok(hexNFC("0300 0301 0041"), "0300 0301 0041");
116ok(hexNFC("0301 0300 0041"), "0301 0300 0041");
117ok(hexNFC("0000 0300 0000 0301"), "0000 0300 0000 0301");
118ok(hexNFC("0000 0301 0000 0300"), "0000 0301 0000 0300");
119
120ok(hexNFC("0327 0061 0300"), "0327 00E0");
121ok(hexNFC("0301 0061 0300"), "0301 00E0");
122ok(hexNFC("0315 0061 0300"), "0315 00E0");
123ok(hexNFC("0000 0327 0061 0300"), "0000 0327 00E0");
124ok(hexNFC("0000 0301 0061 0300"), "0000 0301 00E0");
125ok(hexNFC("0000 0315 0061 0300"), "0000 0315 00E0");