X-Git-Url: http://git.shiar.nl/sheet.git/blobdiff_plain/70e3b473d744176d9692900114c8beefc0527150..22df5af9280e24267be789edcff52be9c43fd8ca:/Shiar_Sheet/FormatChar.pm?ds=inline
diff --git a/Shiar_Sheet/FormatChar.pm b/Shiar_Sheet/FormatChar.pm
index 35f6076..f471497 100644
--- a/Shiar_Sheet/FormatChar.pm
+++ b/Shiar_Sheet/FormatChar.pm
@@ -3,42 +3,54 @@ 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 = '1.10';
-our $uc = do 'unicode-char.inc.pl';
+our $uc = do 'data/unicode-char.inc.pl';
sub new {
my ($class) = @_;
bless { anno => ['di', 0], style => 'di' }, $class;
}
-sub glyph_info {
+sub glyph_mkinfo {
my ($self, $codepoint) = @_;
- return $uc->{chr $codepoint} || eval {
+ # attempt to get unicode character information
+ my $info = eval {
require Unicode::UCD;
- if (my $fullinfo = Unicode::UCD::charinfo($codepoint)) {
- return [@$fullinfo{qw/category name - string/}];
- }
- } || [];
+ Unicode::UCD::charinfo($codepoint)
+ || { category => 'Xn', name => '' };
+ } or return;
+ my $string;
+ if ($info->{combining}) {
+ # overlay combining diacritics
+ $string = chr(9676) . chr($codepoint);
+ }
+ return [@$info{qw( category name )}, undef, $string];
+}
+
+sub glyph_info {
+ my ($self, $codepoint) = @_;
+ return $uc->{chr $codepoint} || $self->glyph_mkinfo($codepoint) || [];
}
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, $entity);
}
sub glyphs_html {
@@ -51,7 +63,7 @@ sub glyphs_html {
EscapeHTML($_[0]), # cell
join(' | ', map { $_->[1] } @chars), # title
$chars[0][2], # class
- join(' ', grep { defined } map { $_->[3] } @chars), # digraph
+ join(' ', map { $_->[3] // 'â¦' } @chars), # digraph
);
}
@@ -60,6 +72,31 @@ sub glyph_cell {
return sprintf('
%s', $self->glyph_html($char));
}
+sub glyph_level_univer {
+ my ($self, $input) = @_;
+ if ($input =~ /\p{age=unassigned}/) {
+ # check include for assignments after unicode 6.0 (perl v5.14)
+ state $agemap = do 'data/unicode-age.inc.pl';
+ my $version = $agemap->{ord $input};
+ return $version ? 'l2' : 'l1';
+ }
+ elsif ($input =~ /^\p{in=1.1}*$/) {
+ return 'l5'; # first release 1993
+ }
+ elsif ($input =~ /^\p{in=3.0}*$/) {
+ return 'l4'; # 20th century
+ }
+ elsif ($input =~ /^\p{in=4.1}*$/) {
+ return 'l4'; # over 10 years ago
+ }
+ elsif ($input =~ /^\p{in=6.0}*$/) {
+ return 'l3'; # before 2012
+ }
+ else {
+ return 'l2'; # more recent
+ }
+}
+
sub cell {
my ($self, $input, $html) = @_;
my (@class, $title, $cell, $mnem, $entity);
@@ -80,60 +117,46 @@ sub cell {
$input =~ s/^\\//; # escaped char
($cell, $title, my $class, $mnem, $entity) = $self->glyphs_html($input);
- my $codepoint = ord(substr $input, 0, 1);
if ($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
- }
+ push @class, $self->glyph_level_univer($input);
next;
}
+ my $codepoint = ord $input;
if ($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 {
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}+$/ and $input !~ /\p{Co}/) {
+ push @class, 'l2'; # in unicode 6.0
}
else {
- push @class, 'l1'; # basic unicode
+ push @class, 'l1'; # any unicode
}
}}
@@ -166,50 +189,39 @@ sub cell {
}
}
else {
- if ($_ eq 'hex' or $cell =~ /^[^a-zA-Z]$/) {
- $anno = sprintf(' %04X', 'value', ord $cell);
+ if ($_ eq 'hex' or $input =~ /^[^a-zA-Z]$/) {
+ $anno = sprintf(' %04X', 'value', ord $input);
last;
}
}
}
}
- 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)
my $class = $cell =~ s/^-// && ' class="ex"';
- $rows[-1] .= "".($cell || ' ');
+ $html[-1] .= " | ".($cell || ' ');
}
next;
}
@@ -218,17 +230,57 @@ sub table {
$colspan++;
next;
}
+ elsif ($cell eq '>-') {
+ $html[-1] .= ' | ';
+ next;
+ }
+ elsif ($cell =~ m/^) {
+ $html[-1] .= ' | '.$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;
+ if ($cell eq '>') {
+ push @colheads, '';
+ next;
+ }
+ push @colheads, join('',
+ '<',
+ $cell =~ s/^-// ? 'td' : 'th',
+ $cell =~ s/:(.*)// && qq{ title="$1"},
+ $cell =~ s/^(>+)// && ' colspan='.(length($1) + 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 {
@@ -244,5 +296,30 @@ 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 => $self->{style} eq 'di' ? 'proposed' : 'HTML4'],
+ [l2 => 'unicode â¤6.0'],
+ [l1 => 'other unicode'],
+ [ex => 'discouraged'],
+ );
+
+ return (
+ '',
+ (map { sprintf '%s', @{$_} } @classes),
+ ' |
',
+ );
+}
+
1;
|