charset.plp

   1 <:
   2 use utf8;
   3 use strict;
   4 use warnings;
   5 use open IO => ':utf8';
   6
   7 our $VERSION = 'v1.0';
   8
   9 $header{content_type} = 'text/html; charset=utf-8';
  10
  11 :><!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"
  12  "http://www.w3.org/TR/html4/loose.dtd">
  13 <html>
  14
  15 <head>
  16 <meta http-equiv="content-type" content="<:= $header{content_type} :>">
  17 <title>charset cheat sheet</title>
  18 <link rel="stylesheet" type="text/css" media="all" href="/base.css">
  19 </head>
  20
  21 <body>
  22 <h1>Character encoding</h1>
  23
  24 <:
  25 my $diinfo = do 'digraphs.inc.pl';
  26 my %di = map { $diinfo->{$_}->[0] => $_ } grep { ref $diinfo->{$_} }
  27         keys %$diinfo;
  28
  29 use Encode qw(decode resolve_alias);
  30 # generate character table(s)
  31 # (~16x faster than decoding in loop;
  32 #  substr strings is twice as fast as splitting to an array)
  33 my %ALIAS = (
  34 #       default => [qw(unicode utf-8 iso-8859-1 cp437 -cp1252- --iso-8859-15- -koi8-f)],
  35         default => [qw(unicode- utf-8 iso-8859-1 -cp1252- --iso-8859-15- cp437 -cp850)],
  36         0 => [qw(cp437 cp863)],
  37         1 => [qw(iso-8859-1 cp1252 MacRoman cp850)],
  38         2 => [qw(iso-8859-2 cp1250 cp852 MacCentralEurRoman MacCroatian MacRumanian)],
  39         5 => [qw(koi8-f iso-8859-5 cp1251 MacCyrillic cp855 cp866)],
  40         7 => [qw(iso-8859-7 cp1253 MacGreek cp737 cp869)],
  41         8 => [qw(iso-8859-8 cp1255 MacHebrew cp862)],
  42 );
  43 my @request = map {
  44         if (my $input = $_) {
  45                 my %row = (offset => 0);
  46                 my $endpoint = 255;
  47                 if ($input =~ s/^--//) {
  48                         $row{offset} = $endpoint > 160 ? 160 : 48;
  49                 }
  50                 elsif ($input =~ s/^-//) {
  51                         $row{offset} = $endpoint > 128 ? 128 : 32;
  52                 }
  53                 if ($input =~ s/-$//) {
  54                         $endpoint = $row{offset} ? $row{offset} < 160 ? 159 : 191 : 127;
  55                 }
  56
  57                 if ($input =~ /^U([0-9a-f]+)(?:-([0-9a-f]+))?/) {
  58                         my $start = hex($1) << ($2 ? 4 : 8);
  59                         my $end = $2 ? hex($2) << 4 : $start + 240;
  60                         $row{table} = join '', map { chr } $start .. $end+15;
  61                         utf8::upgrade($row{table});  # prevent latin1 output
  62                         $row{set} = sprintf 'Unicode block U+%02Xxx', $start >> 8;
  63                 }
  64                 elsif ($input eq 'U') {
  65                         $row{table} = ' ' x 512;
  66                         $row{set} = 'Unicode planes';
  67                         $row{cell} = do 'charset-ucplanes.inc.pl';
  68                 }
  69                 elsif ($row{set} = resolve_alias($input)) {
  70                         if ($row{set} eq 'Internal') {
  71                                 $row{table} = ' ' x ($endpoint < 255 ? 640 : 4096);
  72                                 $row{set} = 'Unicode BMP';
  73                                 $row{cell} = do 'charset-unicode.inc.pl';
  74                         }
  75                         elsif ($row{set} eq 'utf-8-strict') {
  76                                 $row{table} = undef;
  77                                 $row{set} = 'UTF-8';
  78                                 $row{cell} = do 'charset-utf8.inc.pl';
  79                         }
  80                         else {
  81                                 $row{table} = decode($row{set}, pack 'C*', $row{offset} .. $endpoint);
  82                         }
  83                 }
  84                 else {
  85                         print "<p>Encoding $input unknown</p>\n";
  86                 }
  87                 \%row;
  88         }
  89         else {
  90                 ();
  91         }
  92 } map { defined $ALIAS{$_} ? @{ $ALIAS{$_} } : $_ }
  93         $ENV{PATH_INFO} =~ /\w/ ? split(m{[/+\s]}, $ENV{PATH_INFO}) : 'default';
  94 my $NOCHAR = chr 0xFFFD;
  95
  96 for my $cp437 (grep {$request[$_]->{set} eq 'cp437'} 0 .. $#request) {
  97         substr($request[$cp437]->{table}, 237, 1) = pack 'U*', 0x3D5; # phi sign
  98         substr($request[$cp437]->{table}, 0, 32) = pack 'U*', map {hex} qw(
  99                 2007 263A 263B 2665 2666 2663 2660 2022 25D8 25CB 25D9 2642 2640 266A 266B 263C
 100                 25BA 25C4 2195 203C 00B6 00A7 25AC 21A8 2191 2193 2192 2190 221F 2194 25B2 25BC
 101         );
 102 }
 103
 104 sub quote {
 105         local $_ = shift;
 106         s/"/&quot;/g;
 107         s/</&lt;/g;
 108         s/>/&gt;/g;
 109         return $_;
 110 }
 111
 112 print "<ul>\n";
 113
 114 my @nibble = (0..9, 'A'..'F');
 115 for my $row (@request) {
 116         printf '<li><table class="glyphs%s">', !$row->{cell} && ' charmap';
 117         printf '<caption>%s</caption>', $row->{set};
 118         print '<col>' x 17;
 119         for my $section (qw{thead}) {
 120                 print "<$section><tr><th>↱";
 121                 print '<th>', $_ for @nibble;
 122                 print "\n";
 123         }
 124         print '<tbody>';
 125         for my $msb (0 .. (length($row->{table}) || 256) - 1 >> 4) {
 126                 printf '<tr><th>%X', $msb + ($row->{offset} >> 4);
 127                 for my $lsb (0 .. $#nibble) {
 128                         if ($row->{cell}) {
 129                                 print $row->{cell}->(($msb<<4) + $lsb);
 130                                 next;
 131                         }
 132
 133                         my $glyph = substr $row->{table}, ($msb<<4) + $lsb, 1;
 134                         if ($glyph eq $NOCHAR) {
 135                                 print '<td>';
 136                                 next;
 137                         }
 138
 139                         my $info = [ord $glyph];
 140                         if (defined (my $mnem = $di{ord $glyph})) {
 141                                 $info = $diinfo->{$mnem};
 142                         }
 143                         else {
 144                                 require Unicode::UCD;
 145                                 my $fullinfo = Unicode::UCD::charinfo(ord $glyph);
 146                                 $info = [@$fullinfo{qw/code name category script string/}] if $fullinfo;
 147                         }
 148                         my ($codepoint, $name, $prop, $script, $string) = @$info;
 149
 150                         $glyph = quote($string || $glyph);
 151                         my $desc = sprintf 'U+%04X%s', $codepoint, $name && " ($name)";
 152                         my @class = ('X', grep {$_} $prop, $script);
 153
 154                         $glyph = "<span>$glyph</span>" if $prop eq 'Zs';
 155
 156                         printf "\n".'<td class="%s" title="%s">%s',
 157                                 join(' ', @class), quote($desc), $glyph;
 158                 }
 159                 print "\n";
 160         }
 161         print "</table>\n";
 162 }
 163
 164 print "</ul>\n";
 165
 166 :>
 167 <hr>
 168
 169 <div class="legend">
 170         <table class="glyphs"><tr>
 171         <td class="X Cc">control
 172         <td class="X Zs"><span>whitespace</span>
 173         <td class="X Mn">diacritic<table class="glyphs"><tr>
 174                 <td class="X Sk">letter
 175                 </table>
 176         <td class="X Po">punctuation<table class="glyphs"><tr>
 177                 <td class="X Pf">quote
 178                 </table>
 179         <td class="X So">symbol<table class="glyphs"><tr>
 180                 <td class="X Sm">math
 181                 <td class="X Sc">currency
 182                 </table>
 183         <td class="X No">numeric
 184         <td class="X Greek">greek<table class="glyphs"><tr>
 185                 <td class="X Latin">latin
 186                 <td class="X Cyrillic">cyrillic
 187                 </table>
 188         <td class="X Aramaic">aramaic<table class="glyphs"><tr>
 189                 <td class="X Brahmic">brahmic
 190                 <td class="X Arabic">arabic
 191                 </table>
 192         <td class="X Syllabic">syllabic<table class="glyphs"><tr>
 193                 <td class="X African">african
 194                 <td class="X Hiragana">japanese
 195                 <td class="X Han">cjk
 196                 <td class="X Bopomofo">chinese
 197                 </table>
 198         <td class="X Alpha">alphabetic
 199         </table>
 200
 201         <table class="glyphs"><tr>
 202         <td class="X">unicode 5.0
 203         <td class="X Xr">proposed
 204         <td class="X Xd">deprecated
 205         <td class="">unassigned
 206         <td class="X Xi">invalid
 207         </table>
 208 </div>
 209
 210 <p class="footer">
 211         <a href="/" rel="home">sheet.shiar.nl</a>/charset
 212         <a href="git://git.shiar.nl/sheet" rel="vcs-git" title="Git repository"><:= $VERSION :></a>
 213         created by <a href="http://shiar.nl/" rel="author">Shiar</a> •
 214         <a title="Licensed under the GNU Affero General Public License, version 3" rel="copyright"
 215            href="http://www.fsf.org/licensing/licenses/agpl-3.0.html">AGPLv3</a>
 216 </p>
 217
 218 </html>