ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/AnyEvent/util/gen_uts46data
(Generate patch)

Comparing AnyEvent/util/gen_uts46data (file contents):
Revision 1.7 by root, Tue Jul 30 23:31:56 2013 UTC vs.
Revision 1.10 by root, Wed Sep 30 07:48:47 2020 UTC

6use utf8; 6use utf8;
7no warnings 'utf8'; 7no warnings 'utf8';
8 8
9binmode STDOUT, ":utf8"; 9binmode STDOUT, ":utf8";
10 10
11open my $fh, "GET http://www.unicode.org/Public/idna/6.2.0/IdnaMappingTable.txt |" 11open my $fh, "GET http://www.unicode.org/Public/idna/13.0.0/IdnaMappingTable.txt |"
12 or die; 12 or die;
13 13
14my $valid; 14my $valid;
15my $imap; # index map \x00 char replacement 15my $imap; # index map \x00 char replacement
16 16
31 31
32 my $R1 = chr $r1; 32 my $R1 = chr $r1;
33 my $R2 = chr $r2; 33 my $R2 = chr $r2;
34 34
35 $map = join "", map chr hex, split ' ', $map; 35 $map = join "", map chr hex, split ' ', $map;
36
37 $type = "valid" if $type eq "deviation"; # use non-transitional behaviour for deviation characters
36 38
37 given ($type) { 39 given ($type) {
38 when (/^(?:disallowed|disallowed_STD3_valid|disallowed_STD3_mapped)$/) { 40 when (/^(?:disallowed|disallowed_STD3_valid|disallowed_STD3_mapped)$/) {
39 # nop 41 # nop
40 } 42 }
630 > index $imap, "\x02" # it's not supposed to be anywhere in there 650 > index $imap, "\x02" # it's not supposed to be anywhere in there
64 or die "imap contains \\x02"; 66 or die "imap contains \\x02";
65print $fh "\$uts46_imap = q\x02$imap\x00\x02;\n"; 67print $fh "\$uts46_imap = q\x02$imap\x00\x02;\n";
66 68
67# try to find a valid quoting character - there usually are many legal combos 69# try to find a valid quoting character - there usually are many legal combos
68for (1..127) { # stay out of utf-8 range 70for (33..112, 1..31) { # stay out of utf-8 range, prefer printable things
69 if (0 >= index $valid, chr) { 71 if (0 >= index $valid, chr) {
72 my $q = chr;
73
74 # primitive compression
75 $valid =~ s/(\x00{32,})/"$q.(\"\x00\"x" . (length $1) . ").$q"/ge;
76 $valid =~ s/(\xff{32,})/"$q.(\"\xff\"x" . (length $1) . ").$q"/ge;
77
70 print $fh "\$uts46_valid = q", chr, $valid, chr, ";\n"; 78 print $fh "\$uts46_valid = q$q$valid$q;\n";
71 goto valid_ok; 79 goto valid_ok;
72 } 80 }
73} 81}
74die "unable to found valid quoting character"; 82die "unable to found valid quoting character";
75valid_ok:; 83valid_ok:;

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines