charset: utf-8 byte function table
authorMischa POSLAWSKY <perl@shiar.org>
Sun, 19 Apr 2009 14:31:06 +0000 (14:31 +0000)
committerMischa POSLAWSKY <perl@shiar.org>
Wed, 22 Apr 2009 00:01:21 +0000 (00:01 +0000)
charset.plp

index c83508b..c9ef2b6 100644 (file)
@@ -91,6 +91,80 @@ for my $tablenum (0 .. $#tables) {
        print "</table>\n";
 }
 
+{
+       print '<li><table class="glyphs"><caption>UTF-8</caption><col>';
+       for my $section (qw{thead}) {
+               print "<$section><tr><th>↱";
+               print '<th>', $_ for @nibble;
+               print "\n";
+       }
+       print '<tbody>';
+       print '<tr rowspan="8">';
+       for my $msb (0 .. $#nibble) {
+               print '<tr><th>', $nibble[$msb];
+               for my $lsb (0 .. $#nibble) {
+                       my $value = ($msb<<4) + $lsb;
+                       if ($value <= 0x7F) {
+                               print '<td rowspan="8" colspan="16" class="X di-a"',
+                                     ' title="U+0000 – U+007F">Single byte ASCII'
+                                       if $value == 0;
+                       }
+                       elsif ($value <= 0xBF) {
+                               print '<td rowspan="4" colspan="16" class="X di-d"',
+                                     '>Multi-byte continuation'
+                                       if $value == 0x80;
+                       }
+                       elsif ($value <= 0xC1) {
+                               print '<td colspan="2" class="X di-b" style="border-right:none; border-bottom:none"',
+                                     ' title="U+0000 – U+007F">(Overl.)'
+                                       if $value == 0xC0;
+                       }
+                       elsif ($value <= 0xDF) {
+                               print '<td rowspan="2" colspan="14" class="X di-prop" style="border-left:none"',
+                                     ' title="U+0080 – U+03FF">2-byte sequence start'
+                                       if $value == 0xC2;
+                               print '<td rowspan="1" colspan="16" class="X di-prop" style="border-top:none"',
+                                     ' title="U+0400 – U+07FF">'
+                                       if $value == 0xD0;
+                       }
+                       elsif ($value <= 0xEF) {
+                               print '<td colspan="16" class="X di-prop"',
+                                     ' title="U+0800 – U+FFFF">3-byte sequence start'
+                                       if $value == 0xE0;
+                       }
+                       elsif ($value <= 0xF4) {
+                               print '<td colspan="5" class="X di-prop" style="border-right:none"',
+                                     ' title="U+1·0000 – U+10·FFFF">4-byte sequence'
+                                       if $value == 0xF0;
+                       }
+                       elsif ($value <= 0xF7) {
+                               print '<td colspan="3" class="X di-b" style="border-left:none"',
+                                     ' title="U+11·0000 – U+1FF·FFFF">(Overflow)'
+                                       if $value == 0xF5;
+                       }
+                       elsif ($value <= 0xFB) {
+                               print '<td colspan="4" class="X di-b"',
+                                     ' title="U+200·0000 – U+3FFF·FFFF">5-byte'
+                                       if $value == 0xF8;
+                       }
+                       elsif ($value <= 0xFD) {
+                               print '<td colspan="2" class="X di-b"',
+                                     ' title="U+4000·0000 – 7FFFF·FFFF">6-byte'
+                                       if $value == 0xFC;
+                       }
+                       elsif ($value <= 0xFF) {
+                               print '<td colspan="2" class="di-invalid">Invalid'
+                                       if $value == 0xFE;
+                       }
+                       else {
+                               print "\n".'<td class="X">?';
+                       }
+               }
+               print "\n";
+       }
+       print "</table>\n";
+}
+
 print "</ul>\n";
 
 :>