charset: crop shared character ranges

diff --git a/charset.plp b/charset.plp
index 25b69b99b47c618048b1b460d008360d7d4ddb75..e5ca8bfb5e09bef07ae77820d0a27b3e69c7150e 100644 (file)
--- a/charset.plp
+++ b/charset.plp
@@ -31,7 +31,8 @@ use Encode qw(decode resolve_alias);
 # (~16x faster than decoding in loop;
 #  substr strings is twice as fast as splitting to an array)
 my %ALIAS = (
 # (~16x faster than decoding in loop;
 #  substr strings is twice as fast as splitting to an array)
 my %ALIAS = (

-       default => [qw(utf-8 iso-8859-1 cp437)],
+#      default => [qw(unicode utf-8 iso-8859-1 cp437 -cp1252- --iso-8859-15- -koi8-f)],
+       default => [qw(unicode utf-8 iso-8859-1 -cp1252- --iso-8859-15- cp437 -cp850)],

        0 => [qw(cp437 cp863)],
        1 => [qw(iso-8859-1 cp1252 MacRoman cp850)],
        2 => [qw(iso-8859-2 cp1250 cp852 MacCentralEurRoman MacCroatian MacRumanian)],
        0 => [qw(cp437 cp863)],
        1 => [qw(iso-8859-1 cp1252 MacRoman cp850)],
        2 => [qw(iso-8859-2 cp1250 cp852 MacCentralEurRoman MacCroatian MacRumanian)],


@@ -39,16 +40,47 @@ my %ALIAS = (
        7 => [qw(iso-8859-7 cp1253 MacGreek cp737 cp869)],
        8 => [qw(iso-8859-8 cp1255 MacHebrew cp862)],
 );
        7 => [qw(iso-8859-7 cp1253 MacGreek cp737 cp869)],
        8 => [qw(iso-8859-8 cp1255 MacHebrew cp862)],
 );

-my @request = grep { defined } map {
-       $_ ? (resolve_alias($_) or print("Encoding $_ unknown") && ()) : ();
+my @request = map {
+       if (my $input = $_) {
+               my %row = (offset => 0);
+               my $endpoint = 255;
+               if ($input =~ s/^--//) {
+                       $row{offset} = $endpoint > 160 ? 160 : 48;
+               }
+               elsif ($input =~ s/^-//) {
+                       $row{offset} = $endpoint > 128 ? 128 : 32;
+               }
+               if ($input =~ s/-$//) {
+                       $endpoint = $row{offset} ? $row{offset} < 160 ? 159 : 191 : 127;
+               }
+               if ($row{set} = resolve_alias($input)) {
+                       if ($row{set} eq 'Internal') {
+                               $row{table} = ' 'x512;
+                               $row{set} = 'Unicode BMP';
+                       }
+                       elsif ($row{set} eq 'utf-8-strict') {
+                               $row{table} = undef;
+                               $row{set} = 'UTF-8';
+                       }
+                       else {
+                               $row{table} = decode($row{set}, pack 'C*', $row{offset} .. $endpoint);
+                       }
+               }
+               else {
+                       print "<p>Encoding $input unknown</p>\n";
+               }
+               \%row;
+       }
+       else {
+               ();
+       }

 } map { defined $ALIAS{$_} ? @{ $ALIAS{$_} } : $_ }
        $ENV{PATH_INFO} =~ /\w/ ? split(m{[/+\s]}, $ENV{PATH_INFO}) : 'default';
 } map { defined $ALIAS{$_} ? @{ $ALIAS{$_} } : $_ }
        $ENV{PATH_INFO} =~ /\w/ ? split(m{[/+\s]}, $ENV{PATH_INFO}) : 'default';

-my @tables = map { $_ eq 'utf-8-strict' ? undef : $_ eq 'Internal' ? ' 'x512 : decode($_, pack 'C*', 0..255) } @request;

 my $NOCHAR = chr 0xFFFD;
 
 my $NOCHAR = chr 0xFFFD;
 

-for my $cp437 (grep {$request[$_] eq 'cp437'} 0 .. $#request) {
-       substr($tables[$cp437], 237, 1) = pack 'U*', 0x3D5; # phi sign
-       substr($tables[$cp437], 0, 32) = pack 'U*', map {hex} qw(
+for my $cp437 (grep {$request[$_]->{set} eq 'cp437'} 0 .. $#request) {
+       substr($request[$cp437]->{table}, 237, 1) = pack 'U*', 0x3D5; # phi sign
+       substr($request[$cp437]->{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
        );
                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
        );


@@ -350,9 +382,9 @@ sub printcell_utf8 {
 print "<ul>\n";
 
 my @nibble = (0..9, 'A'..'F');
 print "<ul>\n";
 
 my @nibble = (0..9, 'A'..'F');

-for my $tablenum (0 .. $#tables) {
+for my $row (@request) {

        print '<li><table class="glyphs">';
        print '<li><table class="glyphs">';

-       printf '<caption>%s</caption>', $request[$tablenum] eq 'utf-8-strict' ? 'UTF-8' : $request[$tablenum];
+       printf '<caption>%s</caption>', $row->{set};

        print '<col>';
        for my $section (qw{thead}) {
                print "<$section><tr><th>↱";
        print '<col>';
        for my $section (qw{thead}) {
                print "<$section><tr><th>↱";


@@ -360,19 +392,19 @@ for my $tablenum (0 .. $#tables) {
                print "\n";
        }
        print '<tbody>';
                print "\n";
        }
        print '<tbody>';

-       for my $msb (0 .. (length($tables[$tablenum]) || 256) - 1 >> 4) {
-               printf '<tr><th>%X', $msb;
+       for my $msb (0 .. (length($row->{table}) || 256) - 1 >> 4) {
+               printf '<tr><th>%X', $msb + ($row->{offset} >> 4);

                for my $lsb (0 .. $#nibble) {
                for my $lsb (0 .. $#nibble) {

-                       if ($request[$tablenum] eq 'utf-8-strict') {
+                       if ($row->{set} eq 'UTF-8') {

                                printcell_utf8(($msb<<4) + $lsb);
                                next;
                        }
                                printcell_utf8(($msb<<4) + $lsb);
                                next;
                        }

-                       elsif ($request[$tablenum] eq 'Internal') {
+                       elsif ($row->{set} eq 'Unicode BMP') {

                                printcell_unicode(($msb<<4) + $lsb);
                                next;
                        }
 
                                printcell_unicode(($msb<<4) + $lsb);
                                next;
                        }
 

-                       my $glyph = substr $tables[$tablenum], ($msb<<4) + $lsb, 1;
+                       my $glyph = substr $row->{table}, ($msb<<4) + $lsb, 1;

                        if ($glyph eq $NOCHAR) {
                                print '<td>';
                                next;
                        if ($glyph eq $NOCHAR) {
                                print '<td>';
                                next;