charset: move group references into metadata include

diff --git a/charset-encoding.inc.pl b/charset-encoding.inc.pl
index d7d1461fc37471b4511a7daea38920a610775c54..15b303d31bad97e37142d9991c2224201043acf8 100644 (file)
--- a/charset-encoding.inc.pl
+++ b/charset-encoding.inc.pl
@@ -2,6 +2,48 @@ use 5.014;
 use warnings;
 
 +{
+       default    => [qw( u+0-27F utf-8+realsize iso-8859-1 iso-8859-15 cp1252 cp437 cp850 )],
+       unicode    => [qw( uu+cols=32+realsize u+0-FFF u+1000-18AF u+18B0-1FFF )],
+       us         => [qw( cp437 cp863 gsm0338 AdobeStandardEncoding )],
+       ebcdic     => [qw( cp37 cp500 cp1047 posix-bc cp1026 cp875 )],
+       iso        => [map {"iso-8859-$_"} 1 .. 11, 13 .. 16],
+       dos        => [qw( cp437 cp865 cp861 cp860 cp863 cp850 cp857 cp852 cp775
+                          cp737 cp869 cp866 cp855 cp862 cp864 )],
+       aix        => [qw( cp1006 )],
+       win        => [qw( cp1252 cp1250 cp1254 cp1257 cp1258 cp1253 cp1251 cp1255 cp1256 cp874 )],
+       mac        => [qw( MacRoman MacRomanian MacRumanian MacCroatian MacCentralEurRoman MacTurkish MacIcelandic MacSami
+                          MacGreek MacCyrillic MacHebrew MacArabic MacFarsi MacThai )],
+
+       # languages
+       westeur    => [qw( iso-8859-1 iso-8859-15 cp1252 iso-8859-14 cp850 hp-roman8 nextstep MacRoman )],
+       centeur    => [qw( iso-8859-2 iso-8859-16 cp1250 cp852 MacRomanian MacCroatian MacCentralEurRoman )], # MacRumanian only for DB
+       turkish    => [qw( iso-8859-9 iso-8859-3 cp1254 cp857 MacTurkish )],
+       norteur    => [qw( baltic nordic )],
+       baltic     => [qw( iso-8859-4 iso-8859-13 cp1257 cp775 )],
+       nordic     => [qw( iso-8859-10 cp865 cp861 MacIcelandic MacSami )],
+       cyrillic   => [qw( koi8-r koi8-u koi8-f iso-8859-5 cp1251 MacCyrillic cp866 cp855
+                          +400 +2DE0 +A640-A69F +500-52F )], # MacUkrainian is broken
+       arabic     => [qw( iso-8859-6 cp1256 MacArabic cp864 cp1006 MacFarsi
+                          +600 +8A0-8BF+8E0 +750-77F )],
+       greek      => [qw( iso-8859-7 cp1253 MacGreek cp737 cp869  +370 +1F00 )],
+       hebrew     => [qw( iso-8859-8 cp1255 MacHebrew cp862  +590 )],
+       thai       => [qw( iso-8859-11 cp874 MacThai )],
+       vietnamese => [qw( viscii cp1258 MacVietnamese )],
+       symbols    => [qw( symbol dingbats MacDingbats wingdings wingdings2 wingdings3 webdings )],
+
+       # iso-code shorthand
+       1 => 'westeur',
+       2 => 'centeur',
+       3 => 'turkish',
+       4 => 'baltic',
+       5 => 'cyrillic',
+       6 => 'arabic',
+       7 => 'greek',
+       8 => 'hebrew',
+       9 => 'turkish',
+       10 => 'nordic',
+       11 => 'thai',
+
        'ascii'        => {inherit => ['' => '00-7F']},
        'cp437'        => {inherit => ['cp850' => 0, 'ascii' => '00-1F+80']}, # ascii range overridden later
        'gsm0338'      => {inherit => ['ascii' => '00-7F']},

diff --git a/charset.plp b/charset.plp
index d2c92c5378c4227e8917ba419e7d3f2cee6c723d..49ee3eba48dafc17365a6216e7253d0c1f1755bc 100644 (file)
--- a/charset.plp
+++ b/charset.plp
@@ -83,61 +83,21 @@ my $charsets = do 'charset-encoding.inc.pl'
 sub tabinput {
        # generate character table(s)
        my $input = shift or return;
+       my $params = $input =~ s/[+](.*)\z// ? $1 : undef;
+       my $charset = $charsets->{$input} || {};
 
-       state $ALIAS = {
-               default    => [qw( u+0-27F utf-8+realsize iso-8859-1 iso-8859-15 cp1252 cp437 cp850 )],
-               unicode    => [qw( uu+cols=32+realsize u+0-FFF u+1000-18AF u+18B0-1FFF )],
-               us         => [qw( cp437 cp863 gsm0338 AdobeStandardEncoding )],
-               ebcdic     => [qw( cp37 cp500 cp1047 posix-bc cp1026 cp875 )],
-               iso        => [map {"iso-8859-$_"} 1 .. 11, 13 .. 16],
-               dos        => [qw( cp437 cp865 cp861 cp860 cp863 cp850 cp857 cp852 cp775
-                                  cp737 cp869 cp866 cp855 cp862 cp864 )],
-               aix        => [qw( cp1006 )],
-               win        => [qw( cp1252 cp1250 cp1254 cp1257 cp1258 cp1253 cp1251 cp1255 cp1256 cp874 )],
-               mac        => [qw( MacRoman MacRomanian MacRumanian MacCroatian MacCentralEurRoman MacTurkish MacIcelandic MacSami
-                                  MacGreek MacCyrillic MacHebrew MacArabic MacFarsi MacThai )],
-
-               # languages
-               westeur    => [qw( iso-8859-1 iso-8859-15 cp1252 iso-8859-14 cp850 hp-roman8 nextstep MacRoman )],
-               centeur    => [qw( iso-8859-2 iso-8859-16 cp1250 cp852 MacRomanian MacCroatian MacCentralEurRoman )], # MacRumanian only for DB
-               turkish    => [qw( iso-8859-9 iso-8859-3 cp1254 cp857 MacTurkish )],
-               norteur    => [qw( baltic nordic )],
-               baltic     => [qw( iso-8859-4 iso-8859-13 cp1257 cp775 )],
-               nordic     => [qw( iso-8859-10 cp865 cp861 MacIcelandic MacSami )],
-               cyrillic   => [qw( koi8-r koi8-u koi8-f iso-8859-5 cp1251 MacCyrillic cp866 cp855
-                                  +400 +2DE0 +A640-A69F +500-52F )], # MacUkrainian is broken
-               arabic     => [qw( iso-8859-6 cp1256 MacArabic cp864 cp1006 MacFarsi
-                                  +600 +8A0-8BF+8E0 +750-77F )],
-               greek      => [qw( iso-8859-7 cp1253 MacGreek cp737 cp869  +370 +1F00 )],
-               hebrew     => [qw( iso-8859-8 cp1255 MacHebrew cp862  +590 )],
-               thai       => [qw( iso-8859-11 cp874 MacThai )],
-               vietnamese => [qw( viscii cp1258 MacVietnamese )],
-               symbols    => [qw( symbol dingbats MacDingbats wingdings wingdings2 wingdings3 webdings )],
-
-               # iso-code shorthand
-               1 => 'westeur',
-               2 => 'centeur',
-               3 => 'turkish',
-               4 => 'baltic',
-               5 => 'cyrillic',
-               6 => 'arabic',
-               7 => 'greek',
-               8 => 'hebrew',
-               9 => 'turkish',
-               10 => 'nordic',
-               11 => 'thai',
-       };
-       if (my $follow = $ALIAS->{$input}) {
-               tabinput($_) for ref $follow ? @{$follow} : $follow;
+       if (ref $charset ne 'HASH') {
+               $params and Alert("Parameters ignored for $input",
+                       "Cannot apply <q>$params</q> to multiple charsets.",
+               );
+               tabinput($_) for ref $charset ? @{$charset} : $charset;
                return;
        }
 
        state $visible = {'' => 1};  # all present tables
        my %row = (offset => 0, cols => 16);
-       my $params = $input =~ s/[+](.*)\z// ? $1 : undef;
 
        if (not defined $params) {
-               my $charset = $charsets->{$input} || {};
                my @parents = @{ $charset->{inherit} || [] };
 
                if (my ($parent, $part) = pairfirst { defined $visible->{$a} } @parents) {