digraphs: separate rfc parser from include generator

[sheet.git] / tools / mkdigraphs-rfc

diff --git a/tools/mkdigraphs-rfc b/tools/mkdigraphs-rfc
new file mode 100755 (executable)
index 0000000..ab03a42
--- /dev/null
+++ b/tools/mkdigraphs-rfc
@@ -0,0 +1,140 @@
+#!/usr/bin/env perl
+
+use strict;
+use warnings;
+use utf8;
+use open OUT => ':utf8', ':std';
+
+our $VERSION = '1.00';
+
+# expect input data source at command line
+@ARGV or die "Specify input source file or - for STDIN\n";
+
+# skip everything until a character indented by 1 space (table start)
+do {
+       $_ = readline;
+       defined or die "Premature input end";
+} until /^\s\S/;
+
+my @t = $_;  # add first line (already read, assume it's ok)
+
+# read the rest of the character table
+while ($_ = readline) {
+       # check for table end (chapter 4)
+       last if /^\d/;
+
+       # parse table lines (ignore (unindented) page break)
+       next unless s/^ //;
+       chomp;
+
+       # add the line to @t
+       if (s/^ {15}/ /) {
+               # continuation line (add to last entry)
+               $t[-1] .= $_;
+       }
+       else {
+               # add a new entry
+               push @t, $_;
+       }
+}
+
+# create a hash of desired input
+my %di;
+for (@t) {
+       my ($mnem, $char, $name) = split / +/, $_, 3;
+       next if length $mnem != 2;
+       $di{$mnem} = hex $char;
+}
+
+# XXX
+my %trans = (
+       0xE001 => 0,  # join lines: not accepted
+       0xE004 => 0,  # umlaut is no different from diaeresis 0x0308
+       0xE005 => 0x0344, # discouraged
+       0xE006 => 0x0300,
+       0xE007 => 0x0301,
+       0xE008 => 0x0302,
+       0xE009 => 0x0303,
+       0xE00A => 0x0304,
+       0xE00B => 0x0306,
+       0xE00C => 0x0307,
+       0xE00D => 0x0308,
+       0xE00E => 0x030A,
+       0xE00F => 0x030B,
+       0xE010 => 0x030C,
+       0xE011 => 0x0327,
+       0xE012 => 0x0328,
+       0xE013 => 0x0332,
+       0xE014 => 0x0333,
+       0xE015 => 0x0338,
+       0xE016 => 0x0345,
+       0xE017 => 0x0314,
+       0xE018 => 0x0313,
+       0xE019 => 0x1FFE,
+       0xE01A => 0x1FBF,
+       0xE01B => 0x03D0,  # middle beta = curled beta?
+       0xE01C => 0x25CB,
+       0xE01D => 0x0192,
+       0xE01E => 0x0292,
+       0xE01F => 0x33C2,  # am, compatibility char
+       0xE020 => 0x33D8,  # pm, compatibility char
+       0xE021 => 0x2121,
+       0xE022 => 0xFE8E,
+       0xE023 => 0,  # dutch guilder 0192 is already encoded, and not very useful anyway
+       0xE024 => 0x0393,
+       0xE025 => 0x20D7,  # also 20D1; non-spacing
+       0xE026 => 0x1FEF,
+       0xE027 => 0x1FC0,
+       0xE028 => 0x01F0, #but uppercase
+);
+for (values %di) {
+       $_ >= 0xE000 or next;
+       $_ = $trans{$_} if defined $trans{$_};
+}
+
+# output perl code of hash
+# (assume no backslashes or curlies, so we can just q{} w/o escaping)
+print "# automatically generated by $0\n";
+print "use utf8;\n";
+print "+{\n";
+printf "q{%s}=>%s,\n", $_, $di{$_} for sort keys %di;
+print "}\n";
+
+__END__
+
+=head1 NAME
+
+mkdigraphs-rfc - Output digraph data from RFC-1345
+
+=head1 SYNOPSIS
+
+Extract digraphs from text specifications as a perl hash:
+
+    mkdigraphs-rfc rfc1345.txt >digraphs-rfc.inc.pl
+
+Input can be the literal RFC (or similar) document:
+
+    curl http://www.ietf.org/rfc/rfc1345.txt | mkdigraphlist -
+
+Test by printing the character for DO (should be a dollar sign):
+
+    perl -e'$di = do "digraphs-rfc.inc.pl"; print chr $di->{DO}'
+
+=head1 DESCRIPTION
+
+Parses the official RFC-1345 document, searching the
+'character mnemonic table' for all digraph definitions.
+If successful, Perl code is output resulting in a hash
+with Unicode code points keyed by digraph.
+Obsolete values (references to private use area)
+are converted to modern alternatives.
+Any errors and warnings are given at STDERR.
+
+=head1 AUTHOR
+
+Mischa POSLAWSKY <perl@shiar.org>
+
+=head1 LICENSE
+
+Licensed under the GNU Affero General Public License version 3.
+