7 use open OUT => ':utf8', ':std';
11 # create a hash of desired input
12 my $di = do 'data/digraphs-rfc.inc.pl'
13 or die "error reading digraphs include: ", $@ // $!;
16 my $extra = do 'data/digraphs-shiar.inc.pl'
17 or warn "could not include shiar proposals: ", $@ // $!;
18 my $vim = do 'data/digraphs-vim.inc.pl'
19 or warn "could not include vim extensions ", $@ // $!;
20 $di = { %{$vim // {}}, %{$di}, %{$extra // {}} };
22 # optionally get unicode character information
23 my $uninfo = do 'unicode-char.inc.pl'
24 or warn "could not include unicode details: ", $@ // $!;
26 # convert info hashes into arrays of strings to output in display order
27 for my $row (values %{$uninfo}) {
28 my ($class, $name, $di, $html, $string) = @{$row};
29 $row = [$name, $class];
30 push @{$row}, '', $string if defined $string;
33 # output perl code of hash
34 # (assume no backslashes or curlies, so we can just q{} w/o escaping)
35 print "# automatically generated by $0\n";
38 printf '(map {$_=>0} qw{%s}),'."\n", join(' ',
39 grep { !defined $di->{$_} }
40 map { substr($_, 1, 1).substr($_, 0, 1) } sort keys %{$di}
42 printf "q{%s}=>[%s],\n", $_, join(',',
43 $di->{$_}, # original code point
44 (map {"'$_'"} @{ $uninfo->{ chr $di->{$_} } // [] }), # optional additional arguments
45 ) for sort keys %{$di};
52 mkdigraphlist - Output character list of combined digraph data
56 mkdigraphlist >digraphs.inc.pl
57 perl -e'$di = do "digraphs.inc.pl"; print chr $di->{DO}->[0]'
61 Combines precompiled digraph includes of rfc (1345), vim, and shiar
62 and outputs a complete map including character details and usage classes.
64 The value can either be a scalar string containing another
65 digraph which can be considered identical (usually inverted),
66 or an array ref containing at least the resulting character's
67 Unicode code point value. If available, the following UCD data
68 is appended: character name, category, script, and output string.
72 AE => [198, 'LATIN CAPITAL LETTER AE', 'Lu Xl', 'Latin'],
78 Mischa POSLAWSKY <perl@shiar.org>
82 Licensed under the GNU Affero General Public License version 3.