X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/378d119f5791fea807f36749ce9ecb5a5c60952e..ad81ac24b07376ca3ca8899e2fa4eed19f7f32c7:/Shiar_Sheet/FormatChar.pm

diff --git a/Shiar_Sheet/FormatChar.pm b/Shiar_Sheet/FormatChar.pm
index a501b94..f676ead 100644
--- a/Shiar_Sheet/FormatChar.pm
+++ b/Shiar_Sheet/FormatChar.pm
@@ -6,42 +6,38 @@ use warnings;
 use Data::Dump 'pp';
 use PLP::Functions 'EscapeHTML';
 
-our $VERSION = '1.00';
+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) = @_;
-	bless {}, $class;
+	bless { anno => ['di', 0], style => 'di' }, $class;
 }
 
 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 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 {
@@ -65,34 +61,90 @@ sub glyph_cell {
 
 sub cell {
 	my ($self, $input, $html) = @_;
-	my (@class, $title, $cell, $mnem);
+	my (@class, $title, $cell, $mnem, $entity);
 
 	if ($input eq '-') {
 		$cell = '';
 	}
 	elsif ($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
 		}
 
-		($cell, $title, my $class, $mnem) = $self->glyphs_html($input);
+		$input =~ s/^\\//;  # escaped char
+		($cell, $title, my $class, $mnem, $entity) = $self->glyphs_html($input);
 
-		if (defined $mnem) {
-			push @class, 'di-d'; # digraph
-			push @class, 'di-prop' if $class =~ /\bXz\b/; # unofficial
+		if ($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 {
+			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 (@{ $self->{anno} }) {
+			if (/html$/) {
+				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;
+			}
+			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;
+				}
+			}
 		}
 	}
 
@@ -101,11 +153,7 @@ sub cell {
 		@class ? sprintf(' class="%s"', join ' ', @class) : '',
 		$html || '',
 		$cell eq '' ? '&nbsp;' : $cell,
-		defined $mnem && length $mnem
-			? sprintf(' <small class="digraph">%s</small>', EscapeHTML($mnem))
-			: $cell =~ /^[^a-zA-Z]$/
-				? sprintf(' <small class="%s">%04X</small>', 'value', ord $cell)
-				: '',
+		$anno,
 	);
 }
 
@@ -130,7 +178,7 @@ sub table {
 	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)
@@ -151,8 +199,22 @@ sub table {
 		$colspan = 1;
 	}
 
-	return sprintf qq{<table class="glyphs dilabel">\n%s</table>\n},
-		join '', map {"<tr>$_\n"} @rows;
+	return sprintf qq{<table class="glyphs%s">\n%s</table>\n},
+		@{ $self->{anno} } ? ' dilabel' : '',
+		join '', map {"$_\n"} @rows;
+}
+
+sub print {
+	my $self = shift;
+	while (@_) {
+		print '<div class="section">';
+		printf '<h2>%s</h2>', shift unless ref $_[0];
+		print "\n\n";
+		while (ref $_[0] and $_ = shift) {
+			print $self->table($_);
+		}
+		print "\n</div>";
+	}
 }
 
 1;