X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/3a76788894b92930175ec23149c0793c6b0b1100..b16ebb61e918ea6b67674f515814f4fae6fcc37f:/charset.plp

diff --git a/charset.plp b/charset.plp
index 9a9586a..9dd98a4 100644
--- a/charset.plp
+++ b/charset.plp
@@ -43,7 +43,7 @@ sub optionlink {
 print join " •\n", (
 	map {
 		join " ·\n", pairmap {
-			optionlink($b || ucfirst $a, "/charset/$a?compare", $a eq $Request);
+			optionlink($b || ucfirst $a, '/charset'.($a && "/$a?compare"), $a eq $Request);
 		} @{$_}
 	}
 	[
@@ -51,6 +51,7 @@ print join " •\n", (
 		dos      => 'DOS',
 		mac      => 'Apple',
 		ebcdic   => 'EBCDIC',
+		$tablist[0] eq 'default' ? () : ('' => 'common'),
 	],
 	[
 		westeur  => 'West',
@@ -77,8 +78,8 @@ sub tabinput {
 	my $input = shift or return;
 
 	state $ALIAS = {
-		default    => [qw( unicode+0-639 utf-8 iso-8859-1 iso-8859-15 cp1252 cp437 cp850 )],
-		uc         => [qw( U++ unicode+0-4095 unicode+4096-6319  unicode+6320-8191 )],
+		default    => [qw( unicode+0-639 utf-8+realsize iso-8859-1 iso-8859-15 cp1252 cp437 cp850 )],
+		uc         => [qw( U+cols=32 unicode+0-4095 unicode+4096-6319  unicode+6320-8191 )],
 		us         => [qw( cp437 cp863 gsm0338 AdobeStandardEncoding )],
 		ebcdic     => [qw( cp37 cp500 cp1047 posix-bc cp1026 cp875 )],
 		iso        => [map {"iso-8859-$_"} 1 .. 11, 13 .. 16],
@@ -183,9 +184,9 @@ sub tabinput {
 			'MacRumanian' => ['MacRomanian' => '160-191+208-223', 'MacRoman' => '160-191+208-223'],
 			'MacCroatian' => ['MacRoman' => '160'],
 			'MacCentralEurRoman' => ['MacRoman' => '128'],
-			'MacIcelandic'=> ['MacRoman' => '160-175+208-239'], #TODO: gaps at C/E
+			'MacIcelandic'=> ['MacRoman' => '160-175+208-239'],
 			'MacTurkish'  => ['MacRoman' => '208-223'], # F5 is unassigned
-			'MacSami'     => ['MacIcelandic' => '144', 'MacRoman' => '144'],
+			'MacSami'     => ['MacIcelandic' => '144-191+208-223+240', 'MacRoman' => '144'],
 			'MacGreek'    => ['MacRoman' => '128'],
 			'MacCyrillic' => ['MacRoman' => '128'],
 			'MacHebrew'   => ['iso-8859-8' => '128', 'MacRoman' => '128-143+160'], # partial ascii
@@ -195,7 +196,7 @@ sub tabinput {
 			'cp37'        => ['posix-bc' => '0'],
 			'posix-bc'    => ['cp1047' => '64'],
 			'cp500'       => ['cp37' => '64-95+176-191'],
-			'cp1047'      => ['cp37' => '16-95+160-191'], #TODO: gap at 3/4
+			'cp1047'      => ['cp37' => '16-47+80-95+160-191'],
 			'cp1026'      => ['cp37' => '64'],
 			'cp875'       => ['cp37' => '48'],
 
@@ -221,28 +222,28 @@ sub tabinput {
 		$visible->{$_} //= 0 for $row{parent} || ();
 	}
 
-	if (length $params) {
-		# manual option to double table width
-		$row{cols} *= 2 if $params =~ s/[+]\z//;
-	}
-
-	if (length $params) {
-		$params =~ m{
-			\A (?<offset> \d+)
-			(?: (?: [-] (?<stop> \d+) )? (?: [+] (?<restart> \d+) ) )?
-			(?: [-] (?<endpoint> \d+) )? \z
-		}x or Alert([
-			"Unknown range parameters for $input",
-			"<q>$params</q> is not in format start(-stop)(+restart(-end))",
-		]);
-
-		$row{offset} = $+{offset};
-		$row{endpoint} = $+{endpoint} if $+{endpoint};
-		if (my $restart = $+{restart}) {
-			my $skip = int(($+{stop} || $row{offset}) / $row{cols});
-			for ($skip + 1 .. ($restart / $row{cols}) - 1) {
-				$row{skip}->{ $_ * $row{cols} - $row{offset} }++;
+	for my $param (split /[+]+/, $params // '') {
+		if ($param eq 'realsize') {
+			$row{realsize}++;
+		}
+		elsif ($param =~ m{ \A cols = (\d+) \z }x) {
+			$row{cols} = $1;
+		}
+		elsif ($param =~ m{ \A (?<offset> \d+) (?: [-] (?<endpoint> \d+) )? \z }x) {
+			if (defined $row{endpoint}) {
+				# extend earlier range
+				my $skip = int(($row{endpoint} || $row{offset}) / $row{cols});
+				for ($skip + 1 .. ($+{offset} / $row{cols}) - 1) {
+					$row{skip}->{ $_ * $row{cols} - $row{offset} }++;
+				}
+			}
+			else {
+				$row{offset} = $+{offset};
 			}
+			$row{endpoint} = $+{endpoint} || 0;
+		}
+		else {
+			Alert("Unknown option <q>$param</q> for charset $input");
 		}
 	}
 
@@ -297,14 +298,21 @@ sub tabinput {
 			}
 			$row{endpoint} -= $row{offset};
 
-			if ($row{set} eq 'cp437' and !$row{offset}) {
-				substr($row{table}, 237, 1) = pack 'U*', 0x3D5; # phi sign
-				substr($row{table}, 0, 32) = pack 'U*', map {hex} qw(
-					2007 263A 263B 2665 2666 2663 2660 2022
-					25D8 25CB 25D9 2642 2640 266A 266B 263C
-					25BA 25C4 2195 203C 00B6 00A7 25AC 21A8
-					2191 2193 2192 2190 221F 2194 25B2 25BC
-				);
+			if ($row{set} eq 'cp437') {
+				for my $phipos (237 - $row{offset}) {
+					next if $phipos < 0 or $phipos > $row{endpoint};
+					# replace phi glyph
+					substr($row{table}, $phipos, 1) = pack 'U*', 0x3D5;
+				}
+				if ($row{offset} == 0) {
+					# replace control characters by visible variants
+					substr($row{table}, 0, 32) = pack 'U*', map {hex} qw(
+						2007 263A 263B 2665 2666 2663 2660 2022
+						25D8 25CB 25D9 2642 2640 266A 266B 263C
+						25BA 25C4 2195 203C 00B6 00A7 25AC 21A8
+						2191 2193 2192 2190 221F 2194 25B2 25BC
+					);
+				}
 			}
 
 			$visible->{ascii} =  # assume common base
@@ -364,13 +372,15 @@ sub range_cell {
 			$class .= ' joind';
 		}
 
-		# coalesce multiple rows
-		while ($rows > 3) {
-			$info->{skip}->{$offset += $rowsize}++;
-			$rows--;
-		}
-		if ($rows > 2) {
-			$info->{skip}->{$offset += $rowsize} = 0;
+		unless ($info->{realsize}) {
+			# coalesce multiple rows
+			while ($rows > 3) {
+				$info->{skip}->{$offset += $rowsize}++;
+				$rows--;
+			}
+			if ($rows > 2) {
+				$info->{skip}->{$offset += $rowsize} = 0;
+			}
 		}
 
 		$attr .= sprintf ' rowspan=%d', $rows;