writing: mkcharver script to parse unicode version history

[sheet.git] / tools / mkcharver

diff --git a/tools/mkcharver b/tools/mkcharver
new file mode 100755 (executable)
index 0000000..f8fc3e8
--- /dev/null
+++ b/tools/mkcharver
@@ -0,0 +1,43 @@
+#!/usr/bin/env perl
+use 5.012;
+use warnings;
+use utf8;
+
+our $VERSION = '1.00';
+
+say '# automatically generated by tools/mkcharver';
+say '+{';
+
+while (my $line = readline) {
+       $line =~ m{
+               \A (?<start> [0-9A-F]+)
+               (?: \.\. (?<end> [0-9A-F]+) )?
+               \h+ ; \h+ (?<major> [0-9]+) \. (?<minor> [0-9])
+       }x or next;
+
+       my $start   = hex $+{start};
+       my $end     = defined $+{end} ? hex $+{end} : $start;
+       my $version = $+{major} . $+{minor};
+       say "(map {\$_ => $version} $start .. $end),";
+}
+
+say '}';
+
+__END__
+
+=head1 NAME
+
+mkcharver - Create Perl include of Unicode character versions
+
+=head1 SYNOPSIS
+
+    mkcharver DerivedAge.txt >unicode-age.inc.pl
+
+=head1 AUTHOR
+
+Mischa POSLAWSKY <perl@shiar.org>
+
+=head1 LICENSE
+
+Licensed under the GNU Affero General Public License version 3.
+