This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Fix bug #37628 (both lcfirst and ucfirst)
[perl5.git] / t / uni / case.pl
CommitLineData
e49298ea
JH
1use File::Spec;
2
3require "test.pl";
4
e486a898
JH
5sub unidump {
6 join " ", map { sprintf "%04X", $_ } unpack "U*", $_[0];
7}
8
e49298ea 9sub casetest {
6f9b16a7 10 my ($base, $spec, @funcs) = @_;
e49298ea
JH
11 my $file = File::Spec->catfile(File::Spec->catdir(File::Spec->updir,
12 "lib", "unicore", "To"),
13 "$base.pl");
14 my $simple = do $file;
15 my %simple;
16 for my $i (split(/\n/, $simple)) {
17 my ($k, $v) = split(' ', $i);
18 $simple{$k} = $v;
19 }
20 my %seen;
21
22 for my $i (sort keys %simple) {
b08cf34e 23 $seen{$i}++;
e49298ea
JH
24 }
25 print "# ", scalar keys %simple, " simple mappings\n";
26
27 my $both;
28
29 for my $i (sort keys %$spec) {
b08cf34e
JH
30 if (++$seen{$i} == 2) {
31 warn sprintf "$base: $i seen twice\n";
e49298ea
JH
32 $both++;
33 }
34 }
35 print "# ", scalar keys %$spec, " special mappings\n";
36
37 exit(1) if $both;
38
39 my %none;
40 for my $i (map { ord } split //,
41 "\e !\"#\$%&'()+,-./0123456789:;<=>?\@[\\]^_{|}~\b") {
42 next if pack("U0U", $i) =~ /\w/;
43 $none{$i}++ unless $seen{$i};
44 }
45 print "# ", scalar keys %none, " noncase mappings\n";
46
47 my $tests =
6f9b16a7
NC
48 ((scalar keys %simple) +
49 (scalar keys %$spec) +
50 (scalar keys %none)) * @funcs;
e49298ea
JH
51 print "1..$tests\n";
52
53 my $test = 1;
54
b08cf34e 55 for my $i (sort keys %simple) {
e486a898 56 my $w = $simple{$i};
e49298ea 57 my $c = pack "U0U", hex $i;
6f9b16a7
NC
58 foreach my $func (@funcs) {
59 my $d = $func->($c);
60 my $e = unidump($d);
61 print $d eq pack("U0U", hex $simple{$i}) ?
62 "ok $test # $i -> $w\n" : "not ok $test # $i -> $e ($w)\n";
63 $test++;
64 }
e49298ea
JH
65 }
66
b08cf34e 67 for my $i (sort keys %$spec) {
e486a898 68 my $w = unidump($spec->{$i});
08ca2aa3 69 my $u = unpack "C0U", $i;
b08cf34e
JH
70 my $h = sprintf "%04X", $u;
71 my $c = chr($u); $c .= chr(0x100); chop $c;
6f9b16a7
NC
72 foreach my $func (@funcs) {
73 my $d = $func->($c);
74 my $e = unidump($d);
75 if (ord "A" == 193) { # EBCDIC
76 # We need to a little bit of remapping.
77 #
78 # For example, in titlecase (ucfirst) mapping
79 # of U+0149 the Unicode mapping is U+02BC U+004E.
80 # The 4E is N, which in EBCDIC is 2B--
81 # and the ucfirst() does that right.
82 # The problem is that our reference
83 # data is in Unicode code points.
84 #
85 # The Right Way here would be to use, say,
86 # Encode, to remap the less-than 0x100 code points,
87 # but let's try to be Encode-independent here.
88 #
89 # These are the titlecase exceptions:
90 #
91 # Unicode Unicode+EBCDIC
92 #
93 # 0149 -> 02BC 004E (02BC 002B)
94 # 01F0 -> 004A 030C (00A2 030C)
95 # 1E96 -> 0048 0331 (00E7 0331)
96 # 1E97 -> 0054 0308 (00E8 0308)
97 # 1E98 -> 0057 030A (00EF 030A)
98 # 1E99 -> 0059 030A (00DF 030A)
99 # 1E9A -> 0041 02BE (00A0 02BE)
100 #
101 # The uppercase exceptions are identical.
102 #
103 # The lowercase has one more:
104 #
105 # Unicode Unicode+EBCDIC
106 #
107 # 0130 -> 0069 0307 (00D1 0307)
108 #
109 if ($i =~ /^(0130|0149|01F0|1E96|1E97|1E98|1E99|1E9A)$/) {
110 $e =~ s/004E/002B/; # N
111 $e =~ s/004A/00A2/; # J
112 $e =~ s/0048/00E7/; # H
113 $e =~ s/0054/00E8/; # T
114 $e =~ s/0057/00EF/; # W
115 $e =~ s/0059/00DF/; # Y
116 $e =~ s/0041/00A0/; # A
117 $e =~ s/0069/00D1/; # i
118 }
119 # We have to map the output, not the input, because
120 # pack/unpack U has been EBCDICified, too, it would
121 # just undo our remapping.
0c23d0e0 122 }
6f9b16a7
NC
123 print $w eq $e ?
124 "ok $test # $i -> $w\n" : "not ok $test # $h -> $e ($w)\n";
125 $test++;
0c23d0e0 126 }
e49298ea
JH
127 }
128
e49298ea 129 for my $i (sort { $a <=> $b } keys %none) {
e486a898
JH
130 my $w = $i = sprintf "%04X", $i;
131 my $c = pack "U0U", hex $i;
6f9b16a7
NC
132 foreach my $func (@funcs) {
133 my $d = $func->($c);
134 my $e = unidump($d);
135 print $d eq $c ?
136 "ok $test # $i -> $w\n" : "not ok $test # $i -> $e ($w)\n";
137 $test++;
138 }
e49298ea
JH
139 }
140}
141
1421;