use Data::Dump 'pp';
use PLP::Functions 'EscapeHTML';
-our $VERSION = '1.01';
+our $VERSION = '1.04';
our $diinfo = do 'digraphs.inc.pl';
our %di = map { $diinfo->{$_}->[0] => $_ } grep { ref $diinfo->{$_} }
sub new {
my ($class) = @_;
- bless { digraph => 1, unicode => 0 }, $class;
+ bless { anno => ['di', 0], style => 'di' }, $class;
}
sub glyph_info {
$cell = '';
}
elsif ($input eq '=') {
- push @class, 'di-invalid';
+ push @class, 'u-invalid';
$cell = '';
}
else {
push @class, 'X';
if ($input =~ s/^-//) {
- push @class, 'di-rare'; # discouraged
+ push @class, 'ex'; # discouraged
}
+ $input =~ s/^\\//; # escaped char
($cell, $title, my $class, $mnem) = $self->glyphs_html($input);
- if (defined $mnem) {
- push @class, 'di-d'; # digraph
- push @class, 'di-prop' if $class =~ /\bXz\b/; # unofficial
+ if ($self->{style} = 'di') {
+ if (defined $mnem) {
+ push @class, $class =~ /\bXz\b/ ? ('l2', 'u-prop') # unofficial
+ : ('l3', 'u-di'); # standard digraph
+ }
+ }
+ else {
+ my $codepoint = ord(substr $input, 0, 1);
+ if ($codepoint <= 0xFF) {
+ push @class, 'l3', 'u-lat1'; # latin1
+ }
+ elsif ($codepoint <= 0xD7FF) {
+ push @class, 'l2', 'u-bmp'; # bmp
+ }
}
if ($input =~ /[ -~]/) {
- push @class, 'di-a'; # ascii
+ push @class, 'l4', 'u-ascii'; # ascii
}
else {
- push @class, 'di-b'; # basic unicode
+ push @class, 'l1'; # basic unicode
+ }
+ }
+
+ my $anno = '';
+ if ($cell ne '') {
+ for (@{ $self->{anno} }) {
+ if (/html$/) {
+ require HTML::Entities;
+ if (my $entity = $HTML::Entities::char2entity{$cell}) {
+ $entity = substr($entity, 1, -1) unless /^&/;
+ $anno = sprintf(' <small class="digraph">%s</small>', EscapeHTML($entity));
+ last;
+ }
+ }
+ elsif ($_ eq 'xml') {
+ $anno = sprintf(' <small class="digraph">%s</small>',
+ sprintf '#%d', ord($cell)
+ );
+ last;
+ }
+ elsif ($_ eq '&xml') {
+ $anno = sprintf(' <small class="digraph">%s</small>',
+ sprintf '&#%d;', ord($cell)
+ );
+ last;
+ }
+ elsif ($_ eq 'di') {
+ if (defined $mnem and length $mnem) {
+ $anno = sprintf(' <small class="digraph">%s</small>', EscapeHTML($mnem));
+ last;
+ }
+ }
+ else {
+ if ($_ eq 'hex' or $cell =~ /^[^a-zA-Z]$/) {
+ $anno = sprintf(' <small class="%s">%04X</small>', 'value', ord $cell);
+ last;
+ }
+ }
}
}
@class ? sprintf(' class="%s"', join ' ', @class) : '',
$html || '',
$cell eq '' ? ' ' : $cell,
- $self->{digraph} && defined $mnem && length $mnem
- ? sprintf(' <small class="digraph">%s</small>', EscapeHTML($mnem))
- : $self->{unicode} + $cell =~ /^[^a-zA-Z]$/ > 0
- ? sprintf(' <small class="%s">%04X</small>', 'value', ord $cell)
- : '',
+ $anno,
);
}
for my $cell (@$digraphs) {
if ($cell =~ s/^\.//) {
# dot indicates start of a new row
- push @rows, '';
+ push @rows, '<tr>';
if ($cell =~ s/^>//) {
# header cell text follows
$cell =~ s/_/ /g; # underscores may be used instead of whitespace (for qw//ability)
}
return sprintf qq{<table class="glyphs%s">\n%s</table>\n},
- $self->{digraph} || $self->{unicode} >= 0 ? ' dilabel' : '',
- join '', map {"<tr>$_\n"} @rows;
+ @{ $self->{anno} } ? ' dilabel' : '',
+ join '', map {"$_\n"} @rows;
}
sub print {