X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/bba0d5b6fb0b1bbe6b5bb4a84c72de2152d8dfe3..77ac3d52d89913a79cca834d882c54a58c15a488:/Shiar_Sheet/FormatChar.pm

diff --git a/Shiar_Sheet/FormatChar.pm b/Shiar_Sheet/FormatChar.pm
index 8054655..c93a48a 100644
--- a/Shiar_Sheet/FormatChar.pm
+++ b/Shiar_Sheet/FormatChar.pm
@@ -8,9 +8,9 @@ use utf8;
 use Data::Dump 'pp';
 use PLP::Functions 'EscapeHTML';
 
-our $VERSION = '1.06';
+our $VERSION = '1.08';
 
-our $uc = do 'unicode-char.inc.pl';
+our $uc = do 'data/unicode-char.inc.pl';
 
 sub new {
 	my ($class) = @_;
@@ -34,12 +34,12 @@ sub glyph_html {
 	my ($class, $name, $mnem, $entity, $string) = @$info;
 
 	my $cell = EscapeHTML($string || $char);
-	my $title = sprintf 'U+%04X%s', $codepoint, $name && " ($name)";
+	my $title = sprintf 'U+%04X%s', $codepoint, !!$name && " ($name)";
 
-	$cell = "<span>$cell</span>" if $class =~ /\bZs\b/;
+	$cell = "<span>$cell</span>" if $class and $class =~ /\bZs\b/;
 	$cell = '&nbsp;' if $cell eq '';
 
-	return ($cell, EscapeHTML($title), "X $class", $mnem, $entity);
+	return ($cell, EscapeHTML($title), !!$class && "X $class", $mnem, $entity);
 }
 
 sub glyphs_html {
@@ -86,7 +86,7 @@ sub cell {
 		if ($self->{style} eq 'univer') {
 			if ($input =~ /\p{age=unassigned}/) {
 				# check include for assignments after unicode 6.0 (perl v5.14)
-				state $agemap = do 'unicode-age.inc.pl';
+				state $agemap = do 'data/unicode-age.inc.pl';
 				my $version = $agemap->{$codepoint};
 				push @class, $version ? 'l2' : 'l1';
 			}
@@ -109,7 +109,7 @@ sub cell {
 		}
 
 		if ($self->{style} eq 'di') {
-			if ($mnem =~ /…/) {
+			if ($mnem and $mnem =~ /…/) {
 				# incomplete representation, usually partial
 			}
 			elsif ($class =~ /\bu-di\b/) {
@@ -136,7 +136,7 @@ sub cell {
 		if ($input =~ /[ -~]/) {
 			push @class, 'l5', 'u-ascii'; # ascii
 		}
-		elsif ($input =~ /^\p{in=6.0}+$/) {
+		elsif ($input =~ /^\p{in=6.0}+$/ and $input !~ /\p{Co}/) {
 			push @class, 'l2'; # in unicode 6.0
 		}
 		else {
@@ -173,8 +173,8 @@ sub cell {
 				}
 			}
 			else {
-				if ($_ eq 'hex' or $cell =~ /^[^a-zA-Z]$/) {
-					$anno = sprintf(' <small class="%s">%04X</small>', 'value', ord $cell);
+				if ($_ eq 'hex' or $input =~ /^[^a-zA-Z]$/) {
+					$anno = sprintf(' <small class="%s">%04X</small>', 'value', ord $input);
 					last;
 				}
 			}
@@ -183,7 +183,7 @@ sub cell {
 
 	return sprintf('<%s>%s%s',
 		join(' ', 'td',
-			defined $title  ? qq{ title="$title"}  : (),
+			defined $title ? qq{title="$title"}  : (),
 			@class ? sprintf('class="%s"', join ' ', @class) : (),
 			$html || (),
 		),
@@ -192,33 +192,20 @@ sub cell {
 	);
 }
 
-sub table {
-	my ($self, $digraphs) = @_;
-
-	my @rows;
-
-	my @colheads;
-	while ($digraphs->[0] !~ /^\./) {
-		my $cell = shift @$digraphs or last;
-		push @colheads, sprintf(
-			'<%s%s>%s',
-			$cell =~ s/^-// ? 'td' : 'th',
-			$cell =~ s/:(.*)// ? qq{ title="$1"} : '',
-			$cell eq '_' ? '&nbsp;' : $cell
-		);
-	}
-	push @rows, sprintf '<thead><tr>%s<tbody>', join '', @colheads if @colheads;
+sub row {
+	my ($self, $cells) = @_;
+	my @html;
 
 	my $colspan = 1;
-	for my $cell (@$digraphs) {
+	for my $cell (@{$cells}) {
 		if ($cell =~ s/^\.//) {
 			# dot indicates start of a new row
-			push @rows, '<tr>';
+			push @html, '<tr>';
 			if ($cell =~ s/^>//) {
 				# header cell text follows
 				$cell =~ s/_/ /g;  # underscores may be used instead of whitespace (for qw//ability)
 				my $class = $cell =~ s/^-// && ' class="ex"';
-				$rows[-1] .= "<th$class>".($cell || '&nbsp;');
+				$html[-1] .= "<th$class>".($cell || '&nbsp;');
 			}
 			next;
 		}
@@ -228,24 +215,56 @@ sub table {
 			next;
 		}
 		elsif ($cell eq '>-') {
-			$rows[-1] .= '<th>';
+			$html[-1] .= '<th>';
 			next;
 		}
 		elsif ($cell =~ m/^</) {
-			$rows[-1] .= '<td>'.$cell;
+			$html[-1] .= '<td>'.$cell;
 			next;
 		}
 
-		$rows[-1] .= $self->cell($cell,
+		$html[-1] .= $self->cell($cell,
 			$colspan > 1 && qq{colspan="$colspan"},
 		);
 
 		$colspan = 1;
 	}
 
-	return sprintf qq{<table class="glyphs%s">\n%s</table>\n},
-		@{ $self->{anno} } ? ' dilabel' : '',
-		join '', map {"$_\n"} @rows;
+	return @html;
+}
+
+sub tabletag {
+	my ($self) = @_;
+	my $class = 'glyphs';
+	$class .= ' dilabel' if @{ $self->{anno} };
+	return sprintf '<table class="%s">', $class;
+}
+
+sub table {
+	my ($self, $digraphs) = @_;
+
+	my @rows;
+
+	my @colheads;
+	while ($digraphs->[0] !~ /^\./) {
+		my $cell = shift @$digraphs or last;
+		if ($cell eq '>') {
+			push @colheads, '<tr>';
+			next;
+		}
+		push @colheads, join('',
+			'<',
+			$cell =~ s/^-// ? 'td' : 'th',
+			$cell =~ s/:(.*)// && qq{ title="$1"},
+			$cell =~ s/^(>+)// && ' colspan='.(length($1) + 1),
+			'>',
+			$cell eq '_' ? '&nbsp;' : $cell
+		);
+	}
+	push @rows, sprintf '<thead><tr>%s<tbody>', join '', @colheads if @colheads;
+	push @rows, $self->row($digraphs);
+
+	return join '', map {"$_\n"} $self->tabletag, @rows, '</table>';
 }
 
 sub print {
@@ -261,5 +280,30 @@ sub print {
 	}
 }
 
+sub legend {
+	my $self = shift;
+	my @classes = $self->{style} eq 'univer' ? (
+		[l5 => 'unicode 1.1'],
+		[l4 => '20th century'],
+		[l3 => 'in 6.0 (2010)'],
+		[l2 => 'recent assignments'],
+		[l1 => 'proposed'],
+		[ex => 'irregular'],
+	) : (
+		[l5 => 'ascii'],
+		[l4 => $self->{style} eq 'di' ? 'digraph' : 'latin1'],
+		[l3 => $self->{style} eq 'di' ? 'proposed' : 'HTML4'],
+		[l2 => 'unicode ≤6.0'],
+		[l1 => 'other unicode'],
+		[ex => 'discouraged'],
+	);
+
+	return (
+		'<div class="legend"><table class="glyphs"><tr>',
+		(map { sprintf '<td class="X %s">%s', @{$_} } @classes),
+		'</table></div>',
+	);
+}
+
 1;