X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/6c08f60465c441baff92074c3122fcb62457658c..5f257afb85ce61ae765dffad95ce755acb760234:/Shiar_Sheet/FormatChar.pm

diff --git a/Shiar_Sheet/FormatChar.pm b/Shiar_Sheet/FormatChar.pm
index 62f6791..0a8a50a 100644
--- a/Shiar_Sheet/FormatChar.pm
+++ b/Shiar_Sheet/FormatChar.pm
@@ -1,16 +1,16 @@
 package Shiar_Sheet::FormatChar;
 
+use 5.010;
 use strict;
 use warnings;
+use utf8;
 
 use Data::Dump 'pp';
 use PLP::Functions 'EscapeHTML';
 
-our $VERSION = '1.04';
+our $VERSION = '1.08';
 
-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 +19,27 @@ 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, $entity, $string) = @$info;
 
 	my $cell = EscapeHTML($string || $char);
-	my $title = sprintf 'U+%04X%s', $codepoint, $name && " ($name)";
-	my @class = ('X', grep {$_} $prop, $script);
+	my $title = sprintf 'U+%04X%s', $codepoint, !!$name && " ($name)";
 
-	$cell = "<span>$cell</span>" if $prop and $prop eq 'Zs';
+	$cell = "<span>$cell</span>" if $class and $class =~ /\bZs\b/;
 	$cell = '&nbsp;' if $cell eq '';
 
-	return ($cell, EscapeHTML($title), join(' ', @class), $mnem);
+	return ($cell, EscapeHTML($title), !!$class && "X $class", $mnem, $entity);
 }
 
 sub glyphs_html {
@@ -54,7 +52,7 @@ sub glyphs_html {
 		EscapeHTML($_[0]), # cell
 		join(' | ', map { $_->[1] } @chars), # title
 		$chars[0][2], # class
-		join(' ',  grep { defined } map { $_->[3] } @chars), # digraph
+		join(' ',  map { $_->[3] // '…' } @chars), # digraph
 	);
 }
 
@@ -65,7 +63,7 @@ sub glyph_cell {
 
 sub cell {
 	my ($self, $input, $html) = @_;
-	my (@class, $title, $cell, $mnem);
+	my (@class, $title, $cell, $mnem, $entity);
 
 	if ($input eq '-') {
 		$cell = '';
@@ -74,110 +72,140 @@ sub cell {
 		push @class, 'u-invalid';
 		$cell = '';
 	}
-	else {
+	else {{
 		push @class, 'X';
 
 		if ($input =~ s/^-//) {
 			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);
+		my $codepoint = ord $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') {
+			if ($input =~ /\p{age=unassigned}/) {
+				# check include for assignments after unicode 6.0 (perl v5.14)
+				state $agemap = do 'unicode-age.inc.pl';
+				my $version = $agemap->{$codepoint};
+				push @class, $version ? 'l2' : 'l1';
+			}
+			elsif ($input =~ /^\p{in=1.1}*$/) {
+				push @class, 'l5';  # first release 1993
+			}
+			elsif ($input =~ /^\p{in=3.0}*$/) {
+				push @class, 'l4';  # 20th century
+			}
+			elsif ($input =~ /^\p{in=4.1}*$/) {
+				push @class, 'l4';  # over 10 years ago
+			}
+			elsif ($input =~ /^\p{in=6.0}*$/) {
+				push @class, 'l3';  # before 2012
+			}
+			else {
+				push @class, 'l2';  # more recent
+			}
+			next;
+		}
+
+		if ($self->{style} eq 'di') {
+			if ($mnem and $mnem =~ /…/) {
+				# incomplete representation, usually partial
+			}
+			elsif ($class =~ /\bu-di\b/) {
+				push @class, ('l4', 'u-di'); # standard digraph
+			}
+			elsif ($class =~ /\bu-prop\b/) {
+				push @class, ('l3', 'u-prop'); # unofficial
+			}
+		}
+		elsif ($self->{style} eq 'html') {
+			if (defined $entity) {
+				push @class, ($codepoint <= 0xFF ? 'l4' : 'l3', 'u-html');
 			}
 		}
 		else {
-			my $codepoint = ord(substr $input, 0, 1);
 			if ($codepoint <= 0xFF) {
-				push @class, 'l3', 'u-lat1';  # latin1
+				push @class, 'l4', 'u-lat1';  # latin1
 			}
 			elsif ($codepoint <= 0xD7FF) {
-				push @class, 'l2', 'u-bmp';  # bmp
+				push @class, 'l3', 'u-bmp';  # bmp
 			}
 		}
 
 		if ($input =~ /[ -~]/) {
-			push @class, 'l4', 'u-ascii'; # ascii
+			push @class, 'l5', 'u-ascii'; # ascii
+		}
+		elsif ($input =~ /^\p{in=6.0}+$/) {
+			push @class, 'l2'; # in unicode 6.0
 		}
 		else {
-			push @class, 'l1'; # basic unicode
+			push @class, 'l1'; # any unicode
 		}
-	}
+	}}
 
 	my $anno = '';
-	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;
+	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));
+			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;
+				}
+			}
 		}
 	}
 
-	return sprintf('<td%s%s%s>%s%s',
-		defined $title  ? qq{ title="$title"}  : '',
-		@class ? sprintf(' class="%s"', join ' ', @class) : '',
-		$html || '',
+	return sprintf('<%s>%s%s',
+		join(' ', 'td',
+			defined $title ? qq{title="$title"}  : (),
+			@class ? sprintf('class="%s"', join ' ', @class) : (),
+			$html || (),
+		),
 		$cell eq '' ? '&nbsp;' : $cell,
 		$anno,
 	);
 }
 
-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)
-				$rows[-1] .= '<th>'.($cell || '&nbsp;');
+				my $class = $cell =~ s/^-// && ' class="ex"';
+				$html[-1] .= "<th$class>".($cell || '&nbsp;');
 			}
 			next;
 		}
@@ -186,27 +214,63 @@ sub table {
 			$colspan++;
 			next;
 		}
+		elsif ($cell eq '>-') {
+			$html[-1] .= '<th>';
+			next;
+		}
+		elsif ($cell =~ m/^</) {
+			$html[-1] .= '<td>'.$cell;
+			next;
+		}
 
-		$rows[-1] .= $self->cell($cell,
-			$colspan > 1 && qq{ colspan="$colspan"},
+		$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;
+		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;
+	push @rows, $self->row($digraphs);
+
+	return join '', map {"$_\n"} $self->tabletag, @rows, '</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>";
 	}
 }