charset: numeric offset parameters

Replace dash shorthands by equivalent start-end options.

diff --git a/charset.plp b/charset.plp
index 92b918f8fd32452b1e6015032a6ac6b66d20ffd0..7794284f7ff685f4db1f6ef2279a249841dfc76b 100644 (file)
--- a/charset.plp
+++ b/charset.plp
@@ -30,19 +30,19 @@ sub tabinput {
        my $input = shift or return;
 
        state $ALIAS = {
-               default    => [qw(unicode- utf-8 iso-8859-1 -cp1252-- --iso-8859-15-- cp437 -cp850)],
-               us         => [qw( cp437 -cp863 AdobeStandardEncoding gsm0338-- )],
+               default    => [qw( unicode+0-192 utf-8 iso-8859-1 cp1252+128-159 iso-8859-15+160-191 cp437 cp850+128 )],
+               us         => [qw( cp437 cp863+128 AdobeStandardEncoding gsm0338+0-127 )],
                ebcdic     => [qw( cp37 cp500  cp875  cp1026  cp1047  posix-bc )],
-               westeur    => [qw( iso-8859-1 --iso-8859-15-- -cp1252-- --iso-8859-14 -cp850 -MacRoman -nextstep --hp-roman8 )],
-               centeur    => [qw( iso-8859-2 --iso-8859-16 -cp1250 -cp852 -MacCentralEurRoman -MacCroatian -MacRomanian )], # MacRumanian only for DB
-               turkish    => [qw( iso-8859-3 -iso-8859-9 -cp857 -cp1254 -MacTurkish )],
-               baltic     => [qw( iso-8859-4 --iso-8859-13 -cp775 -cp1257 )],
-               nordic     => [qw( iso-8859-10 -cp865 --cp861-- -MacIcelandic --MacSami )],
-               cyrillic   => [qw( koi8-f -koi8-r- --koi8-u- -iso-8859-5 -cp1251 -MacCyrillic -cp855 -cp866 )], # MacUkrainian is broken
-               arabic     => [qw( iso-8859-6 --cp1006 -cp864 -cp1256 MacArabic )], # MacFarsi same as MacArabic?
-               greek      => [qw( iso-8859-7 -cp1253 -MacGreek -cp737 -cp869 )],
-               hebrew     => [qw( iso-8859-8 -cp1255 -MacHebrew -cp862 )],
-               thai       => [qw( iso-8859-11 -cp874-- -MacThai )],
+               westeur    => [qw( iso-8859-1 iso-8859-15+160-191 cp1252+128-159 iso-8859-14+160 cp850+128 MacRoman+128 nextstep+128 hp-roman8+160 )],
+               centeur    => [qw( iso-8859-2 iso-8859-16+160 cp1250+128 cp852+128 MacCentralEurRoman+128 MacCroatian+128 MacRomanian+128 )], # MacRumanian only for DB
+               turkish    => [qw( iso-8859-3 iso-8859-9+128 cp857+128 cp1254+128 MacTurkish+128 )],
+               baltic     => [qw( iso-8859-4 iso-8859-13+160 cp775+128 cp1257+128 )],
+               nordic     => [qw( iso-8859-10 cp865+128 cp861+160-191 MacIcelandic+128 MacSami+160 )],
+               cyrillic   => [qw( koi8-f koi8-r+128-192 koi8-u+160-192 iso-8859-5+128 cp1251+128 MacCyrillic+128 cp855+128 cp866+128 )], # MacUkrainian is broken
+               arabic     => [qw( iso-8859-6 cp1006+160 cp864+128 cp1256+128 MacArabic )], # MacFarsi same as MacArabic?
+               greek      => [qw( iso-8859-7 cp1253+128 MacGreek+128 cp737+128 cp869+128 )],
+               hebrew     => [qw( iso-8859-8 cp1255+128 MacHebrew+128 cp862+128 )],
+               thai       => [qw( iso-8859-11 cp874+128-159 MacThai+128 )],
                vietnamese => [qw( viscii cp1258 MacVietnamese )],
                symbol     => [qw( symbol dingbats MacDingbats AdobeZdingbat AdobeSymbol )],
                # iso-code shorthand
@@ -64,17 +64,11 @@ sub tabinput {
 
        my %row = (offset => 0, cols => 16);
        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} ? $row{offset} < 160 ? 159 : 191 : 127;
-       }
-       elsif ($input =~ s/-$//) {
-               $endpoint = 192;
+       my $params = $input =~ s/[+](.*)\z// ? $1 : undef;
+
+       if (defined $params and $params =~ m/^ (\d+) (-\d+)? /x) {
+               $row{offset} = $1;
+               $endpoint = -$2 if $2;
        }
        if ($row{offset}) {
                $row{setnote} = 'over cp437' if $input eq 'cp850';