X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/7d1044e5e4df51459cafb742090ce62b52d9e087..d7c1207fb1cf7dcac45dd455a78d4c8c106ca4a3:/rfc1345convert

diff --git a/rfc1345convert b/rfc1345convert
index 412bac1..3157e04 100644
--- a/rfc1345convert
+++ b/rfc1345convert
@@ -46,6 +46,52 @@ for (@t) {
 	$di{$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 $trans{$_};
+}
+
 # personal addendums
 my @extra;
 if (-r 'shiar.inc.txt') {
@@ -77,13 +123,35 @@ for (values %info) {
 # mark unofficial extras as such
 $info{$_}->{category} .= ' Xz' for @extra;
 
+for (keys %di) {
+	$info{$_}->{string} = chr(9676) . chr($di{$_}) if $info{$_}->{combining};
+	# find control characters (first 32 chars from 0 and 128)
+	next unless ($di{$_} & ~0b1001_1111) == 0 or $di{$_} == 127;
+	# rename to something more descriptive
+	$info{$_}->{name} = $info{$_}->{unicode10}
+		? '<'.$info{$_}->{unicode10}.'>'  # the old name was much more useful
+		: sprintf('<control U+%04X>', $di{$_});  # at least identify by value
+	# show descriptive symbols instead of control chars themselves
+	$info{$_}->{string} = $di{$_} < 32 ? chr($di{$_} + 0x2400) : chr(0xFFFD);
+}
+
+# convert info hashes into arrays of strings to output in display order
+for my $row (values %info) {
+	$row = [ map { $row->{$_} } qw/name category script string/ ];
+	# strip off trailing missing values (especially string may be unknown)
+	defined $row->[-1] ? last : pop @$row for 1 .. @$row;
+}
+
 # output perl code of hash
 # (assume no backslashes or curlies, so we can just q{} w/o escaping)
-print "{\n";
+print "+{\n";
+printf '(map {$_=>0} qw{%s}),'."\n", join(' ',
+	map { substr($_, 1, 1).substr($_, 0, 1) } sort keys %di
+);
 printf "q{%s}=>[%s],\n", $_, join(',',
-	$di{$_},       # glyph code point
+	$di{$_},   # original code point
 	$info{$_}  # optional additional arguments
-		? map {"'$_'"} @{ $info{$_} }{qw/name category script/}
+		? map {"'$_'"} @{ $info{$_} }
 		: ()
 ) for sort keys %di;
 print "}\n";