# (~16x faster than decoding in loop;
# substr strings is twice as fast as splitting to an array)
my %ALIAS = (
- default => [qw(utf-8 iso-8859-1 cp437)],
+# default => [qw(unicode utf-8 iso-8859-1 cp437 -cp1252- --iso-8859-15- -koi8-f)],
+ default => [qw(unicode utf-8 iso-8859-1 -cp1252- --iso-8859-15- cp437 -cp850)],
0 => [qw(cp437 cp863)],
1 => [qw(iso-8859-1 cp1252 MacRoman cp850)],
2 => [qw(iso-8859-2 cp1250 cp852 MacCentralEurRoman MacCroatian MacRumanian)],
7 => [qw(iso-8859-7 cp1253 MacGreek cp737 cp869)],
8 => [qw(iso-8859-8 cp1255 MacHebrew cp862)],
);
-my @request = grep { defined } map {
- $_ ? (resolve_alias($_) or print("Encoding $_ unknown") && ()) : ();
+my @request = map {
+ if (my $input = $_) {
+ my %row = (offset => 0);
+ 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} ? $row{offset} < 160 ? 159 : 191 : 127;
+ }
+ if ($row{set} = resolve_alias($input)) {
+ if ($row{set} eq 'Internal') {
+ $row{table} = ' 'x640;
+ $row{set} = 'Unicode BMP';
+ }
+ elsif ($row{set} eq 'utf-8-strict') {
+ $row{table} = undef;
+ $row{set} = 'UTF-8';
+ }
+ else {
+ $row{table} = decode($row{set}, pack 'C*', $row{offset} .. $endpoint);
+ }
+ }
+ else {
+ print "<p>Encoding $input unknown</p>\n";
+ }
+ \%row;
+ }
+ else {
+ ();
+ }
} map { defined $ALIAS{$_} ? @{ $ALIAS{$_} } : $_ }
$ENV{PATH_INFO} =~ /\w/ ? split(m{[/+\s]}, $ENV{PATH_INFO}) : 'default';
-my @tables = map { $_ eq 'utf-8-strict' ? undef : decode($_, pack 'C*', 0..255) } @request;
my $NOCHAR = chr 0xFFFD;
-for my $cp437 (grep {$request[$_] eq 'cp437'} 0 .. $#request) {
- substr($tables[$cp437], 237, 1) = pack 'U*', 0x3D5; # phi sign
- substr($tables[$cp437], 0, 32) = pack 'U*', map {hex} qw(
+for my $cp437 (grep {$request[$_]->{set} eq 'cp437'} 0 .. $#request) {
+ substr($request[$cp437]->{table}, 237, 1) = pack 'U*', 0x3D5; # phi sign
+ substr($request[$cp437]->{table}, 0, 32) = pack 'U*', map {hex} qw(
2007 263A 263B 2665 2666 2663 2660 2022 25D8 25CB 25D9 2642 2640 266A 266B 263C
25BA 25C4 2195 203C 00B6 00A7 25AC 21A8 2191 2193 2192 2190 221F 2194 25B2 25BC
);
return $_;
}
+sub printcell_unicode {
+ my ($value) = @_;
+ if ($value > 0x27F) {
+ print "\n".'<td class="X">?';
+ }
+ elsif ($value == 0) {
+ print '<td colspan="2" class="X Cc">control';
+ }
+ elsif ($value == 2) {
+ print '<td colspan="6" class="X Ll Latin">latin';
+ }
+ elsif ($value == 8) {
+ print '<td colspan="2" class="X Cc">control';
+ }
+ elsif ($value == 10) {
+ print '<td colspan="6" class="X Ll Latin">latin supplement';
+ }
+ elsif ($value == 0x10) {
+ print '<td colspan="8" class="X Ll Latin">latin ext-A';
+ }
+ elsif ($value == 0x18) {
+ print '<td colspan="8" class="X Ll Latin">latin ext-B';
+ }
+ elsif ($value == 0x20) {
+ print '<td colspan="5" class="X Ll Latin">latin ext-B';
+ }
+ elsif ($value == 0x25) {
+ print '<td colspan="6" class="X Ll Latin">IPA';
+ }
+ elsif ($value == 0x2B) {
+ print '<td colspan="5" class="X Sk">spacing modifier';
+ }
+ elsif ($value == 0x30) {
+ print '<td colspan="8" class="X Mn">diacritics';
+ }
+ elsif ($value == 0x38) {
+ print '<td colspan="8" class="X Ll Greek">greek';
+ }
+ elsif ($value == 0x40) {
+ print '<td colspan="16" class="X Ll Cyrillic">cyrillic';
+ }
+ elsif ($value == 0x50) {
+ print '<td colspan="3" class="X Ll Cyrillic">cyrillic+';
+ }
+ elsif ($value == 0x53) {
+ print '<td colspan="5" class="X Ll Armenian">armenian';
+ }
+ elsif ($value == 0x58) {
+ print '<td colspan="8" class="X Ll Hebrew">hebrew';
+ }
+ elsif ($value == 0x60) {
+ print '<td colspan="16" class="X Ll Arabic">arabic';
+ }
+ elsif ($value == 0x70) {
+ print '<td colspan="5" class="X Ll Aramaic">syriac';
+ }
+ elsif ($value == 0x75) {
+ print '<td colspan="3" class="X Ll Arabic">arabic+';
+ }
+ elsif ($value == 0x78) {
+ print '<td colspan="4" class="X Ll African">thaana';
+ }
+ elsif ($value == 0x7C) {
+ print '<td colspan="4" class="X Ll African">nko';
+ }
+ elsif ($value == 0x80) {
+ print '<td colspan="4" class="X di-rare">samaritan';
+ }
+ elsif ($value == 0x84) {
+ print '<td colspan="2" class="X di-rare Ll Aramaic">manda';
+ }
+ elsif ($value == 0x86) {
+ print '<td colspan="12" class="di-invalid">reserved';
+ }
+ elsif ($value == 0x90) {
+ print '<td colspan="8" class="X Ll Brahmic">devanagari';
+ }
+ elsif ($value == 0x98) {
+ print '<td colspan="8" class="X Ll Brahmic">bengali';
+ }
+ elsif ($value == 0xA0) {
+ print '<td colspan="8" class="X Ll Brahmic">gurmukhi';
+ }
+ elsif ($value == 0xA8) {
+ print '<td colspan="8" class="X Ll Brahmic">gujarati';
+ }
+ elsif ($value == 0xB0) {
+ print '<td colspan="8" class="X Ll Brahmic">oriya';
+ }
+ elsif ($value == 0xB8) {
+ print '<td colspan="8" class="X Ll Brahmic">tamil';
+ }
+ elsif ($value == 0xC0) {
+ print '<td colspan="8" class="X Ll Brahmic">telugu';
+ }
+ elsif ($value == 0xC8) {
+ print '<td colspan="8" class="X Ll Brahmic">kannada';
+ }
+ elsif ($value == 0xD0) {
+ print '<td colspan="8" class="X Ll Brahmic">malayalam';
+ }
+ elsif ($value == 0xD8) {
+ print '<td colspan="8" class="X Ll Brahmic">sinhala';
+ }
+ elsif ($value == 0xE0) {
+ print '<td colspan="8" class="X Ll Brahmic Khmer">thai';
+ }
+ elsif ($value == 0xE8) {
+ print '<td colspan="8" class="X Ll Brahmic Khmer">lao';
+ }
+ elsif ($value == 0xF0) {
+ print '<td colspan="16" class="X Ll Brahmic">tibetan';
+ }
+ elsif ($value == 0x100) {
+ print '<td colspan="10" class="X Ll Brahmic">myanmar';
+ }
+ elsif ($value == 0x10A) {
+ print '<td colspan="6" class="X Ll Aramaic">georgian';
+ }
+ elsif ($value == 0x110) {
+ print '<td colspan="16" class="X Ll Hangul">hangeul jamo';
+ }
+ elsif ($value == 0x120) {
+ print '<td colspan="16" class="X Ll African">ethiopic';
+ }
+ elsif ($value == 0x130) {
+ print '<td colspan="8" class="X Ll African">ethiopic';
+ }
+ elsif ($value == 0x138) {
+ print '<td colspan="2" class="X Ll African">eth+';
+ }
+ elsif ($value == 0x13A) {
+ print '<td colspan="6" class="X Ll Syllabic">cherokee';
+ }
+ elsif ($value == 0x140) {
+ print '<td colspan="16" rowspan="2" class="X Ll Syllabic">unified canadian aboriginal syllabics';
+ }
+ elsif ($value == 0x160) {
+ print '<td colspan="8" class="X Ll Syllabic">unified canadian syllabics';
+ }
+ elsif ($value == 0x168) {
+ print '<td colspan="2" class="X Ll X">ogham';
+ }
+ elsif ($value == 0x16A) {
+ print '<td colspan="6" class="X Ll X">runic';
+ }
+ elsif ($value == 0x170) {
+ print '<td colspan="2" class="X Ll Brahmic">tagalog';
+ }
+ elsif ($value == 0x172) {
+ print '<td colspan="2" class="X Ll Brahmic">hanun';
+ }
+ elsif ($value == 0x174) {
+ print '<td colspan="2" class="X Ll Brahmic">buhid';
+ }
+ elsif ($value == 0x176) {
+ print '<td colspan="2" class="X Ll Brahmic" title="tagbanwa">tagb';
+ }
+ elsif ($value == 0x178) {
+ print '<td colspan="8" class="X Ll Brahmic Khmer">khmer';
+ }
+ elsif ($value == 0x180) {
+ print '<td colspan="11" class="X Ll Aramaic">mongolian';
+ }
+ elsif ($value == 0x18B) {
+ print '<td colspan="5" class="X Ll Syllabic di-rare">canadian+';
+ }
+ elsif ($value == 0x190) {
+ print '<td colspan="5" class="X Ll Brahmic">limbu';
+ }
+ elsif ($value == 0x195) {
+ print '<td colspan="4" class="X Ll Brahmic">tai le';
+ }
+ elsif ($value == 0x198) {
+ print '<td colspan="6" class="X Ll Brahmic">new tai lue';
+ }
+ elsif ($value == 0x19E) {
+ print '<td colspan="2" class="X Ll Brahmic Khmer" title="khmer symbols">km';
+ }
+ elsif ($value == 0x1A0) {
+ print '<td colspan="2" class="X Ll Brahmic">lontara';
+ }
+ elsif ($value == 0x1A2) {
+ print '<td colspan="9" class="X Ll Brahmic di-rare">tai tham';
+ }
+ elsif ($value == 0x1AB) {
+ print '<td colspan="5" class="di-invalid">reserved';
+ }
+ elsif ($value == 0x1B0) {
+ print '<td colspan="8" class="X Ll ">balinese';
+ }
+ elsif ($value == 0x1B8) {
+ print '<td colspan="4" class="X Ll ">sundanese';
+ }
+ elsif ($value == 0x1BC) {
+ print '<td colspan="4" class="X Ll di-rare">batak';
+ }
+ elsif ($value == 0x1C0) {
+ print '<td colspan="5" class="X Ll ">lepcha';
+ }
+ elsif ($value == 0x1C5) {
+ print '<td colspan="3" class="X Ll ">ol chiki';
+ }
+ elsif ($value == 0x1C8) {
+ print '<td colspan="5" class="di-invalid">reserved';
+ }
+ elsif ($value == 0x1CD) {
+ print '<td colspan="3" class="X Ll di-rare">vedic';
+ }
+ elsif ($value == 0x1D0) {
+ print '<td colspan="8" class="X Ll Latin">phonetic';
+ }
+ elsif ($value == 0x1D8) {
+ print '<td colspan="4" class="X Ll Latin">phonetic+';
+ }
+ elsif ($value == 0x1DC) {
+ print '<td colspan="4" class="X Mn">combining';
+ }
+ elsif ($value == 0x1E0) {
+ print '<td colspan="16" class="X Ll Latin">latin extended additional';
+ }
+ elsif ($value == 0x1F0) {
+ print '<td colspan="16" class="X Ll Greek">greek+';
+ }
+ elsif ($value == 0x200) {
+ print '<td colspan="7" class="X Pd">general punctuation';
+ }
+ elsif ($value == 0x207) {
+ print '<td colspan="3" class="X Latin">su[bp]script';
+ }
+ elsif ($value == 0x20A) {
+ print '<td colspan="3" class="X Sc">currency';
+ }
+ elsif ($value == 0x20D) {
+ print '<td colspan="3" class="X Mn">overlay';
+ }
+ elsif ($value == 0x210) {
+ print '<td colspan="5" class="X So">letterlike';
+ }
+ elsif ($value == 0x215) {
+ print '<td colspan="4" class="X Latin">number';
+ }
+ elsif ($value == 0x219) {
+ print '<td colspan="7" class="X So">arrows';
+ }
+ elsif ($value == 0x220) {
+ print '<td colspan="16" class="X Sm">mathematical symbols';
+ }
+ elsif ($value == 0x230) {
+ print '<td colspan="16" class="X So">miscellaneous technical';
+ }
+ elsif ($value == 0x240) {
+ print '<td colspan="4" class="X So">control';
+ }
+ elsif ($value == 0x244) {
+ print '<td colspan="2" class="X So">OCR';
+ }
+ elsif ($value == 0x246) {
+ print '<td colspan="10" class="X Latin">enclosed alphanumerics';
+ }
+ elsif ($value == 0x250) {
+ print '<td colspan="8" class="X So">box drawing';
+ }
+ elsif ($value == 0x258) {
+ print '<td colspan="2" class="X So">blocks';
+ }
+ elsif ($value == 0x25A) {
+ print '<td colspan="6" class="X So">geometric shapes';
+ }
+ elsif ($value == 0x260) {
+ print '<td colspan="16" class="X So">miscellaneous symbols';
+ }
+ elsif ($value == 0x270) {
+ print '<td colspan="12" class="X So">dingbats';
+ }
+ elsif ($value == 0x27C) {
+ print '<td colspan="3" class="X So">maths-A';
+ }
+ elsif ($value == 0x27F) {
+ print '<td colspan="1" class="X So" title="supplemental arrows-A">arr';
+ }
+}
+
sub printcell_utf8 {
my ($value) = @_;
if ($value <= 0x7F) {
print "<ul>\n";
my @nibble = (0..9, 'A'..'F');
-for my $tablenum (0 .. $#tables) {
+for my $row (@request) {
print '<li><table class="glyphs">';
- printf '<caption>%s</caption>', $request[$tablenum] eq 'utf-8-strict' ? 'UTF-8' : $request[$tablenum];
+ printf '<caption>%s</caption>', $row->{set};
print '<col>';
for my $section (qw{thead}) {
print "<$section><tr><th>↱";
print "\n";
}
print '<tbody>';
- for my $msb (0 .. $#nibble) {
- print '<tr><th>', $nibble[$msb];
+ for my $msb (0 .. (length($row->{table}) || 256) - 1 >> 4) {
+ printf '<tr><th>%X', $msb + ($row->{offset} >> 4);
for my $lsb (0 .. $#nibble) {
- if ($request[$tablenum] eq 'utf-8-strict') {
+ if ($row->{set} eq 'UTF-8') {
printcell_utf8(($msb<<4) + $lsb);
next;
}
+ elsif ($row->{set} eq 'Unicode BMP') {
+ printcell_unicode(($msb<<4) + $lsb);
+ next;
+ }
- my $glyph = substr $tables[$tablenum], ($msb<<4) + $lsb, 1;
+ my $glyph = substr $row->{table}, ($msb<<4) + $lsb, 1;
if ($glyph eq $NOCHAR) {
print '<td>';
next;
git.shiar.nl / sheet.git / blobdiff