This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
EBCDIC: special cases of special cases.
[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*", $_[0];
7}
8
e49298ea
JH
9sub casetest {
10 my ($base, $spec, $func) = @_;
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) {
23 $seen{hex $i}++;
24 }
25 print "# ", scalar keys %simple, " simple mappings\n";
26
27 my $both;
28
29 for my $i (sort keys %$spec) {
30 if (++$seen{hex $i} == 2) {
31 warn "$base: $i seen twice\n";
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 $tests =
48 (scalar keys %simple) +
49 (scalar keys %$spec) +
50 (scalar keys %none);
51 print "1..$tests\n";
52
53 my $test = 1;
54
55 for my $i (sort { hex $a <=> hex $b } keys %simple) {
e486a898 56 my $w = $simple{$i};
e49298ea
JH
57 my $c = pack "U0U", hex $i;
58 my $d = $func->($c);
e486a898 59 my $e = unidump($d);
e49298ea 60 print $d eq pack("U0U", hex $simple{$i}) ?
e486a898 61 "ok $test # $i -> $w\n" : "not ok $test # $i -> $e ($w)\n";
e49298ea
JH
62 $test++;
63 }
64
65 for my $i (sort { hex $a <=> hex $b } keys %$spec) {
e486a898 66 my $w = unidump($spec->{$i});
e49298ea
JH
67 my $c = pack "U0U", hex $i;
68 my $d = $func->($c);
e486a898 69 my $e = unidump($d);
0c23d0e0
JH
70 if (ord "A" == 193) { # EBCDIC
71 # We need to a little bit of remapping.
72 #
73 # For example, in titlecase (ucfirst) mapping
74 # of U+0149 the Unicode mapping is U+02BC U+004E.
75 # The 4E is N, which in EBCDIC is 2B--
76 # and the ucfirst() does that right.
77 # The problem is that our reference
78 # data is in Unicode code points.
79 #
80 # The Right Way here would be to use, say,
81 # Encode, to remap the less-than 0x100 code points,
82 # but let's try to be Encode-independent here.
83 #
84 # These are the titlecase exceptions:
85 #
86 # Unicode Unicode+EBCDIC
87 #
88 # 0149 -> 02BC 004E (02BC 002B)
89 # 01F0 -> 004A 030C (00A2 030C)
90 # 1E96 -> 0048 0331 (00E7 0331)
91 # 1E97 -> 0054 0308 (00E8 0308)
92 # 1E98 -> 0057 030A (00EF 030A)
93 # 1E99 -> 0059 030A (00DF 030A)
94 # 1E9A -> 0041 02BE (00A0 02BE)
95 #
96 # The uppercase exceptions are identical.
97 #
98 if ($i =~ /^(0149|01F0|1E96|1E97|1E98|1E99|1E9A)$/) {
99 $e =~ s/004E/002B/; # N
100 $e =~ s/004A/00A2/; # J
101 $e =~ s/0048/00E7/; # H
102 $e =~ s/0054/00E8/; # T
103 $e =~ s/0057/00EF/; # W
104 $e =~ s/0059/00DF/; # Y
105 $e =~ s/0041/00A0/; # A
106 }
107 # We have to map the output, not the input, because
108 # pack/unpack U has been EBCDICified, too, it would
109 # just undo our remapping.
110 }
111 print $w eq $e ?
e486a898 112 "ok $test # $i -> $w\n" : "not ok $test # $i -> $e ($w)\n";
e49298ea
JH
113 $test++;
114 }
115
e49298ea 116 for my $i (sort { $a <=> $b } keys %none) {
e486a898
JH
117 my $w = $i = sprintf "%04X", $i;
118 my $c = pack "U0U", hex $i;
e49298ea 119 my $d = $func->($c);
e486a898 120 my $e = unidump($d);
e49298ea 121 print $d eq $c ?
e486a898 122 "ok $test # $i -> $w\n" : "not ok $test # $i -> $e ($w)\n";
e49298ea
JH
123 $test++;
124 }
125}
126
1271;