X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/bb412b576d883e11a4ec09d01c34a933d894af6c..f1614fb2f16c1e75c8dcef8a9d3e1e46cc236bee:/charset.plp diff --git a/charset.plp b/charset.plp index f546f2d..73f9377 100644 --- a/charset.plp +++ b/charset.plp @@ -4,7 +4,7 @@ use strict; use warnings; use open IO => ':utf8'; -our $VERSION = '1.0'; +our $VERSION = 'v1.0'; $header{content_type} = 'text/html; charset=utf-8'; @@ -13,12 +13,12 @@ $header{content_type} = 'text/html; charset=utf-8'; + charset cheat sheet - - +

Character encoding

<: @@ -53,7 +53,20 @@ my @request = map { if ($input =~ s/-$//) { $endpoint = $row{offset} ? $row{offset} < 160 ? 159 : 191 : 127; } - if ($row{set} = resolve_alias($input)) { + + if ($input =~ /^U([0-9a-f]+)(?:-([0-9a-f]+))?/) { + my $start = hex($1) << ($2 ? 4 : 8); + my $end = $2 ? hex($2) << 4 : $start + 240; + $row{table} = join '', map { chr } $start .. $end+15; + utf8::upgrade($row{table}); # prevent latin1 output + $row{set} = sprintf 'Unicode block U+%02Xxx', $start >> 8; + } + elsif ($input eq 'U') { + $row{table} = ' ' x 512; + $row{set} = 'Unicode planes'; + $row{cell} = do 'charset-ucplanes.inc.pl'; + } + elsif ($row{set} = resolve_alias($input)) { if ($row{set} eq 'Internal') { $row{table} = ' ' x ($endpoint < 255 ? 640 : 4096); $row{set} = 'Unicode BMP'; @@ -96,13 +109,11 @@ sub quote { return $_; } -print "

'; + printf '

', !$row->{cell} && ' charmap'; printf '', $row->{set}; - print ''; + print '' x 17; for my $section (qw{thead}) { print "<$section>

%s
â±"; print '	', $_ for @nibble; @@ -127,6 +138,11 @@ for my $row (@request) { if (defined (my $mnem = $di{ord $glyph})) { $info = $diinfo->{$mnem}; } + else { + require Unicode::UCD; + my $fullinfo = Unicode::UCD::charinfo(ord $glyph); + $info = [@$fullinfo{qw/code name category script string/}] if $fullinfo; + } my ($codepoint, $name, $prop, $script, $string) = @$info; $glyph = quote($string \|\| $glyph); @@ -140,24 +156,61 @@ for my $row (@request) { } print "\n"; } - print "

\n"; + print "\n"; } -print "

\n"; - :>

+ +

control +

whitespace +

diacritic +

letter +

punctuation +

quote +

symbol +

math +

currency +

numeric +

greek +

latin +

cyrillic +

aramaic +

brahmic +

arabic +

syllabic +

african +

japanese +

cjk +

chinese +

alphabetic +

+ + +

unicode 5.0 +

proposed +

deprecated +

unassigned +

invalid +