1 2BEGIN { 3 unless ("A" eq pack('U', 0x41)) { 4 print "1..0 # Unicode::Normalize " . 5 "cannot stringify a Unicode code point\n"; 6 exit 0; 7 } 8} 9 10BEGIN { 11 if ($ENV{PERL_CORE}) { 12 chdir('t') if -d 't'; 13 @INC = $^O eq 'MacOS' ? qw(::lib) : qw(../lib); 14 } 15} 16 17BEGIN { 18 unless (5.006001 <= $]) { 19 print "1..0 # skipped: Perl 5.6.1 or later". 20 " needed for this test\n"; 21 exit; 22 } 23} 24 25######################### 26 27use Test; 28use strict; 29use warnings; 30BEGIN { plan tests => 14 }; 31use Unicode::Normalize qw(:all); 32ok(1); # If we made it this far, we're ok. 33 34sub _pack_U { Unicode::Normalize::pack_U(@_) } 35sub _unpack_U { Unicode::Normalize::unpack_U(@_) } 36 37######################### 38 39our $proc; # before the last starter 40our $unproc; # the last starter and after 41# If string has no starter, entire string is set to $unproc. 42 43# When you have $normalized string and $unnormalized string following, 44# a simple concatenation 45# C<$concat = $normalized . normalize($form, $unnormalized)> 46# is wrong. Instead of it, like this: 47# 48# ($processed, $unprocessed) = splitOnLastStarter($normalized); 49# $concat = $processed . normalize($form, $unprocessed.$unnormalized); 50 51($proc, $unproc) = splitOnLastStarter(""); 52ok($proc, ""); 53ok($unproc, ""); 54 55($proc, $unproc) = splitOnLastStarter("A"); 56ok($proc, ""); 57ok($unproc, "A"); 58 59($proc, $unproc) = splitOnLastStarter(_pack_U(0x41, 0x300, 0x327, 0x42)); 60ok($proc, _pack_U(0x41, 0x300, 0x327)); 61ok($unproc, "B"); 62 63($proc, $unproc) = splitOnLastStarter(_pack_U(0x4E00, 0x41, 0x301)); 64ok($proc, _pack_U(0x4E00)); 65ok($unproc, _pack_U(0x41, 0x301)); 66 67($proc, $unproc) = splitOnLastStarter(_pack_U(0x302, 0x301, 0x300)); 68ok($proc, ""); 69ok($unproc, _pack_U(0x302, 0x301, 0x300)); 70 71our $ka_grave = _pack_U(0x41, 0, 0x42, 0x304B, 0x300); 72our $dakuten = _pack_U(0x3099); 73our $ga_grave = _pack_U(0x41, 0, 0x42, 0x304C, 0x300); 74 75our ($p, $u) = splitOnLastStarter($ka_grave); 76our $concat = $p . NFC($u.$dakuten); 77 78ok(NFC($ka_grave.$dakuten) eq $ga_grave); 79ok(NFC($ka_grave).NFC($dakuten) ne $ga_grave); 80ok($concat eq $ga_grave); 81 82