X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/652fa60a059b983b901b344ea400230206bff8ff..a0ba9298856b2426c5c66b6d2f2b284d98cee594:/charset.plp

diff --git a/charset.plp b/charset.plp
index fd8d2ad..c9ef2b6 100644
--- a/charset.plp
+++ b/charset.plp
@@ -30,7 +30,8 @@ use Encode qw(decode);
 # generate character table(s)
 # (~16x faster than decoding in loop;
 #  substr strings is twice as fast as splitting to an array)
-my @tables = map { decode($_, pack 'C*', 0..255) } 'iso-8859-1', 'cp437';
+my @request = ('iso-8859-1', 'cp437');
+my @tables = map { decode($_, pack 'C*', 0..255) } @request;
 my $NOCHAR = chr 0xFFFD;
 
 for my $cp437 (grep {$request[$_] eq 'cp437'} 0 .. $#request) {
@@ -49,19 +50,23 @@ sub quote {
 	return $_;
 }
 
+print "<ul>\n";
+
 my @nibble = (0..9, 'A'..'F');
-for my $table (@tables) {
-	print '<table class="glyphs"><col>';
-	for my $section (qw{thead tfoot}) {
-		print "<$section><tr><th>↳";
+for my $tablenum (0 .. $#tables) {
+	print '<li><table class="glyphs">';
+	printf '<caption>%s</caption>', $request[$tablenum];
+	print '<col>';
+	for my $section (qw{thead}) {
+		print "<$section><tr><th>↱";
 		print '<th>', $_ for @nibble;
-		print "<th>&nbsp;\n";
+		print "\n";
 	}
 	print '<tbody>';
 	for my $msb (0 .. $#nibble) {
 		print '<tr><th>', $nibble[$msb];
 		for my $lsb (0 .. $#nibble) {
-			my $glyph = substr $table, ($msb<<4) + $lsb, 1;
+			my $glyph = substr $tables[$tablenum], ($msb<<4) + $lsb, 1;
 			if ($glyph eq $NOCHAR) {
 				print '<td>';
 				next;
@@ -81,8 +86,100 @@ for my $table (@tables) {
 			printf "\n".'<td class="%s" title="%s">%s',
 				join(' ', @class), quote($desc), $glyph;
 		}
-		print "\n<th>", $nibble[$msb], "\n";
+		print "\n";
+	}
+	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";
+
+:>
+<hr>
+
+<p class="footer">
+	<a href="http://sheet.shiar.nl/" rel="home">sheet.shiar.nl</a>/charset
+	<a href="git://git.shiar.nl/sheet" rel="vcs-git" title="Git repository"><:= "v$VERSION" :></a>
+	created by <a href="http://shiar.nl/" rel="author">Shiar</a> •
+	<a title="Licensed under the GNU Affero General Public License, version 3" rel="copyright"
+	   href="http://www.fsf.org/licensing/licenses/agpl-3.0.html">AGPLv3</a> •
+	last update <:
+		use Time::Format qw(time_format);
+		print time_format('yyyy-mm-dd', (stat $ENV{SCRIPT_FILENAME})[9]);
+	:>
+</p>
+
+</html>