X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/193d10e6034b48611e9ce97d07bcfdf0eef08a65..5f257afb85ce61ae765dffad95ce755acb760234:/Shiar_Sheet/FormatChar.pm
diff --git a/Shiar_Sheet/FormatChar.pm b/Shiar_Sheet/FormatChar.pm
index f676ead..0a8a50a 100644
--- a/Shiar_Sheet/FormatChar.pm
+++ b/Shiar_Sheet/FormatChar.pm
@@ -1,12 +1,14 @@
package Shiar_Sheet::FormatChar;
+use 5.010;
use strict;
use warnings;
+use utf8;
use Data::Dump 'pp';
use PLP::Functions 'EscapeHTML';
-our $VERSION = '1.06';
+our $VERSION = '1.08';
our $uc = do 'unicode-char.inc.pl';
@@ -29,15 +31,15 @@ sub glyph_html {
my ($self, $char) = @_;
my $codepoint = ord $char;
my $info = $self->glyph_info($codepoint);
- my ($class, $name, $mnem, $html, $string) = @$info;
+ 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)";
- $cell = "$cell" if $class =~ /\bZs\b/;
+ $cell = "$cell" if $class and $class =~ /\bZs\b/;
$cell = ' ' if $cell eq '';
- return ($cell, EscapeHTML($title), "X $class", $mnem, $html);
+ return ($cell, EscapeHTML($title), !!$class && "X $class", $mnem, $entity);
}
sub glyphs_html {
@@ -50,7 +52,7 @@ sub glyphs_html {
EscapeHTML($_[0]), # cell
join(' | ', map { $_->[1] } @chars), # title
$chars[0][2], # class
- join(' ', grep { defined } map { $_->[3] } @chars), # digraph
+ join(' ', map { $_->[3] // 'â¦' } @chars), # digraph
);
}
@@ -70,7 +72,7 @@ sub cell {
push @class, 'u-invalid';
$cell = '';
}
- else {
+ else {{
push @class, 'X';
if ($input =~ s/^-//) {
@@ -79,37 +81,68 @@ 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') {
+ 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 ($class =~ /\bu-di\b/) {
- push @class, ('l3', 'u-di'); # standard digraph
+ if ($mnem and $mnem =~ /â¦/) {
+ # incomplete representation, usually partial
+ }
+ elsif ($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
}
- }
+ }}
my $anno = '';
if ($cell ne '') {
@@ -148,41 +181,31 @@ sub cell {
}
}
- return sprintf('%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 '' ? ' ' : $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 '_' ? ' ' : $cell
- );
- }
- push @rows, sprintf '%s
', join '', @colheads if @colheads;
+sub row {
+ my ($self, $cells) = @_;
+ my @html;
my $colspan = 1;
- for my $cell (@$digraphs) {
+ for my $cell (@{$cells}) {
if ($cell =~ s/^\.//) {
# dot indicates start of a new row
- push @rows, '';
+ push @html, '
';
if ($cell =~ s/^>//) {
# header cell text follows
$cell =~ s/_/ /g; # underscores may be used instead of whitespace (for qw//ability)
- $rows[-1] .= ''.($cell || ' ');
+ my $class = $cell =~ s/^-// && ' class="ex"';
+ $html[-1] .= "".($cell || ' ');
}
next;
}
@@ -191,17 +214,51 @@ sub table {
$colspan++;
next;
}
+ elsif ($cell eq '>-') {
+ $html[-1] .= '';
+ next;
+ }
+ elsif ($cell =~ m/^'.$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{\n},
- @{ $self->{anno} } ? ' dilabel' : '',
- join '', map {"$_\n"} @rows;
+ return @html;
+}
+
+sub tabletag {
+ my ($self) = @_;
+ my $class = 'glyphs';
+ $class .= ' dilabel' if @{ $self->{anno} };
+ return sprintf '', $class;
+}
+
+sub table {
+ my ($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 '_' ? ' ' : $cell
+ );
+ }
+ push @rows, sprintf '%s ', join '', @colheads if @colheads;
+ push @rows, $self->row($digraphs);
+
+ return join '', map {"$_\n"} $self->tabletag, @rows, ' ';
}
sub print {
| |