X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/6ae2e63f9d7f8944b507b7396686a733c6422739..a605211aaba5d956789c1d30c09a55494858d086:/Shiar_Sheet/FormatChar.pm

diff --git a/Shiar_Sheet/FormatChar.pm b/Shiar_Sheet/FormatChar.pm
index 4668c55..6b5fb6c 100644
--- a/Shiar_Sheet/FormatChar.pm
+++ b/Shiar_Sheet/FormatChar.pm
@@ -1,16 +1,15 @@
 package Shiar_Sheet::FormatChar;
 
+use 5.010;
 use strict;
 use warnings;
 
 use Data::Dump 'pp';
 use PLP::Functions 'EscapeHTML';
 
-our $VERSION = '1.05';
+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 +18,27 @@ 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 +62,7 @@ sub glyph_cell {
 
 sub cell {
 	my ($self, $input, $html) = @_;
-	my (@class, $title, $cell, $mnem);
+	my (@class, $title, $cell, $mnem, $entity);
 
 	if ($input eq '-') {
 		$cell = '';
@@ -74,7 +71,7 @@ sub cell {
 		push @class, 'u-invalid';
 		$cell = '';
 	}
-	else {
+	else {{
 		push @class, 'X';
 
 		if ($input =~ s/^-//) {
@@ -82,12 +79,43 @@ sub cell {
 		}
 
 		$input =~ s/^\\//;  # escaped char
-		($cell, $title, my $class, $mnem) = $self->glyphs_html($input);
+		($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 'univer') {
+			state $agemap = do 'unicode-age.inc.pl' or die $!;
+			my $version = $agemap->{ord $input};
+			if (!$version) {
+				push @class, 'l1';  # no known unicode assignment
+			}
+			elsif ($version < 20) {
+				push @class, 'l5';  # first release 1993
+			}
+			elsif ($version < 31) {
+				push @class, 'l4';  # 20th century
+			}
+			elsif ($version < 50) {
+				push @class, 'l4';  # over 10 years ago
+			}
+			elsif ($version < 61) {
+				push @class, 'l3';  # before 2012
+			}
+			else {
+				push @class, 'l2';  # more recent
+			}
+			next;
+		}
+
+		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 {
@@ -106,15 +134,14 @@ sub cell {
 		else {
 			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 /^&/;
+				if (defined $entity) {
+					$entity = "&$entity;" if /^&/;
 					$anno = sprintf(' <small class="digraph">%s</small>', EscapeHTML($entity));
 					last;
 				}
@@ -180,7 +207,8 @@ sub table {
 			if ($cell =~ s/^>//) {
 				# header cell text follows
 				$cell =~ s/_/ /g;  # underscores may be used instead of whitespace (for qw//ability)
-				$rows[-1] .= '<th>'.($cell || '&nbsp;');
+				my $class = $cell =~ s/^-// && ' class="ex"';
+				$rows[-1] .= "<th$class>".($cell || '&nbsp;');
 			}
 			next;
 		}