unicode: include harvey balls in symbols

[sheet.git] / Shiar_Sheet / FormatChar.pm

diff --git a/Shiar_Sheet/FormatChar.pm b/Shiar_Sheet/FormatChar.pm
index 647280e8b878bca433512dad57c2e9d062bd44df..f676ead1da9cb2a1f7951545f532a39bf513f86a 100644 (file)
--- a/Shiar_Sheet/FormatChar.pm
+++ b/Shiar_Sheet/FormatChar.pm
@@ -6,11 +6,9 @@ use warnings;
 use Data::Dump 'pp';
 use PLP::Functions 'EscapeHTML';
 
-our $VERSION = '1.04';
+our $VERSION = '1.06';
 
-our $diinfo = do 'digraphs.inc.pl';
-our %di = map { $diinfo->{$_}->[0] => $_ } grep { ref $diinfo->{$_} }
-       sort { length $a <=> length $b } keys %$diinfo;
+our $uc = do 'unicode-char.inc.pl';
 
 sub new {
        my ($class) = @_;
@@ -19,29 +17,27 @@ sub new {
 
 sub glyph_info {
        my ($self, $codepoint) = @_;
-       if (defined (my $mnem = $di{$codepoint})) {
-               return ($diinfo->{$mnem}, length $mnem == 2 ? $mnem : undef);
-       }
-       require Unicode::UCD;
-       if (my $fullinfo = Unicode::UCD::charinfo($codepoint)) {
-               return [$codepoint, @$fullinfo{qw/name category script string/}];
-       }
-       return [$codepoint];
+       return $uc->{chr $codepoint} || eval {
+               require Unicode::UCD;
+               if (my $fullinfo = Unicode::UCD::charinfo($codepoint)) {
+                       return [@$fullinfo{qw/category name - string/}];
+               }
+       } || [];
 }
 
 sub glyph_html {
        my ($self, $char) = @_;
-       my ($info, $mnem) = $self->glyph_info(ord $char);
-       my ($codepoint, $name, $prop, $script, $string) = @$info;
+       my $codepoint = ord $char;
+       my $info = $self->glyph_info($codepoint);
+       my ($class, $name, $mnem, $html, $string) = @$info;
 
        my $cell = EscapeHTML($string || $char);
        my $title = sprintf 'U+%04X%s', $codepoint, $name && " ($name)";
-       my @class = ('X', grep {$_} $prop, $script);
 
-       $cell = "<span>$cell</span>" if $prop and $prop eq 'Zs';
+       $cell = "<span>$cell</span>" if $class =~ /\bZs\b/;
        $cell = '&nbsp;' if $cell eq '';
 
-       return ($cell, EscapeHTML($title), join(' ', @class), $mnem);
+       return ($cell, EscapeHTML($title), "X $class", $mnem, $html);
 }
 
 sub glyphs_html {
@@ -65,7 +61,7 @@ sub glyph_cell {
 
 sub cell {
        my ($self, $input, $html) = @_;
-       my (@class, $title, $cell, $mnem);
+       my (@class, $title, $cell, $mnem, $entity);
 
        if ($input eq '-') {
                $cell = '';
@@ -81,12 +77,20 @@ sub cell {
                        push @class, 'ex'; # discouraged
                }
 
-               ($cell, $title, my $class, $mnem) = $self->glyphs_html($input);
+               $input =~ s/^\\//;  # escaped char
+               ($cell, $title, my $class, $mnem, $entity) = $self->glyphs_html($input);
 
-               if ($self->{style} = 'di') {
-                       if (defined $mnem) {
-                               push @class, $class =~ /\bXz\b/ ? ('l2', 'u-prop') # unofficial
-                                       : ('l3', 'u-di'); # standard digraph
+               if ($self->{style} eq 'di') {
+                       if ($class =~ /\bu-di\b/) {
+                               push @class, ('l3', 'u-di'); # standard digraph
+                       }
+                       elsif ($class =~ /\bu-prop\b/) {
+                               push @class, ('l2', 'u-prop'); # unofficial
+                       }
+               }
+               elsif ($self->{style} eq 'html') {
+                       if (defined $entity) {
+                               push @class, ('l3', 'u-html');
                        }
                }
                else {
@@ -108,18 +112,39 @@ sub cell {
        }
 
        my $anno = '';
-       for (@{ $self->{anno} }) {
-               if ($_ eq 'di') {
-                       if (defined $mnem and length $mnem) {
-                               $anno = sprintf(' <small class="digraph">%s</small>', EscapeHTML($mnem));
+       if ($cell ne '') {
+               for (@{ $self->{anno} }) {
+                       if (/html$/) {
+                               if (defined $entity) {
+                                       $entity = "&$entity;" if /^&/;
+                                       $anno = sprintf(' <small class="digraph">%s</small>', EscapeHTML($entity));
+                                       last;
+                               }
+                       }
+                       elsif ($_ eq 'xml') {
+                               $anno = sprintf(' <small class="digraph">%s</small>',
+                                       sprintf '#%d', ord($cell)
+                               );
                                last;
                        }
-               }
-               else {
-                       if ($_ eq 'hex' or $cell =~ /^[^a-zA-Z]$/) {
-                               $anno = sprintf(' <small class="%s">%04X</small>', 'value', ord $cell);
+                       elsif ($_ eq '&xml') {
+                               $anno = sprintf(' <small class="digraph">%s</small>',
+                                       sprintf '&amp;#%d;', ord($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;
+                               }
+                       }
                }
        }
 
@@ -182,11 +207,13 @@ sub table {
 sub print {
        my $self = shift;
        while (@_) {
-               printf '<div class="section"><h2>%s</h2>'."\n\n", shift;
+               print '<div class="section">';
+               printf '<h2>%s</h2>', shift unless ref $_[0];
+               print "\n\n";
                while (ref $_[0] and $_ = shift) {
                        print $self->table($_);
                }
-               print '</div>';
+               print "\n</div>";
        }
 }