X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/a605211aaba5d956789c1d30c09a55494858d086..0dd2b5ce328daf14b51f94c951acef2ca40a5fd4:/Shiar_Sheet/FormatChar.pm

diff --git a/Shiar_Sheet/FormatChar.pm b/Shiar_Sheet/FormatChar.pm
index 6b5fb6c..78e4a21 100644
--- a/Shiar_Sheet/FormatChar.pm
+++ b/Shiar_Sheet/FormatChar.pm
@@ -3,6 +3,7 @@ package Shiar_Sheet::FormatChar;
 use 5.010;
 use strict;
 use warnings;
+use utf8;
 
 use Data::Dump 'pp';
 use PLP::Functions 'EscapeHTML';
@@ -51,7 +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
 	);
 }
 
@@ -80,23 +81,25 @@ sub cell {
 
 		$input =~ s/^\\//;  # escaped char
 		($cell, $title, my $class, $mnem, $entity) = $self->glyphs_html($input);
+		my $codepoint = ord $input;
 
 		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
+			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, $version ? 'l2' : 'l1';
 			}
-			elsif ($version < 20) {
+			elsif ($input =~ /^\p{in=1.1}*$/) {
 				push @class, 'l5';  # first release 1993
 			}
-			elsif ($version < 31) {
+			elsif ($input =~ /^\p{in=3.0}*$/) {
 				push @class, 'l4';  # 20th century
 			}
-			elsif ($version < 50) {
+			elsif ($input =~ /^\p{in=4.1}*$/) {
 				push @class, 'l4';  # over 10 years ago
 			}
-			elsif ($version < 61) {
+			elsif ($input =~ /^\p{in=6.0}*$/) {
 				push @class, 'l3';  # before 2012
 			}
 			else {
@@ -106,33 +109,38 @@ sub cell {
 		}
 
 		if ($self->{style} eq 'di') {
-			if ($class =~ /\bu-di\b/) {
-				push @class, ('l3', 'u-di'); # standard digraph
+			if ($mnem =~ /…/) {
+				# incomplete representation, usually partial
+			}
+			elsif ($class =~ /\bu-di\b/) {
+				push @class, ('l4', 'u-di'); # standard digraph
 			}
 			elsif ($class =~ /\bu-prop\b/) {
-				push @class, ('l2', 'u-prop'); # unofficial
+				push @class, ('l3', 'u-prop'); # unofficial
 			}
 		}
 		elsif ($self->{style} eq 'html') {
 			if (defined $entity) {
-				push @class, ('l3', 'u-html');
+				push @class, ($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
 		}
 	}}
 
@@ -173,10 +181,12 @@ sub cell {
 		}
 	}
 
-	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,
 	);
@@ -217,9 +227,17 @@ sub table {
 			$colspan++;
 			next;
 		}
+		elsif ($cell eq '>-') {
+			$rows[-1] .= '<th>';
+			next;
+		}
+		elsif ($cell =~ m/^</) {
+			$rows[-1] .= '<td>'.$cell;
+			next;
+		}
 
 		$rows[-1] .= $self->cell($cell,
-			$colspan > 1 && qq{ colspan="$colspan"},
+			$colspan > 1 && qq{colspan="$colspan"},
 		);
 
 		$colspan = 1;