#!\c:\apps\perl\bin\perl.exe
#program makes an htm file that shows all
#the characters in the charmap.sgm file.
#include tcp character map subroutines
require "C:/code/Perl/Eebo/tcp-charmap-include.pm";
#file containing character map
my $charactermap = "C:/code/charents/charmap.sgm";
#put character map into a perl array
charmap_array($charactermap);
# create file headers
print "\n";
print "\n\nText Creation Partnership Character Entity List\n";
print "\n\n\n";
print "\n";
print "Text Creation Partnership Character Entity List generated by tcp-charmap-htm.pl on " . localtime(time) ."\n";
print "| Entity | Displays as | U+ | Set | Description or Display Equivalent |
|---|
\n";
#cycle through entities
foreach $entity (@entities) {
my $entsgm = content_attr('ENT','TCP');
my $entset = content_attr('ENT','SRC');
my $unic = substr(content_attr('EQUIV','UNIC'),1); #remove initial U
#use Arial Unicode MS entity
my $entview = content_attr_known('REPL','TXT','SUP','Arial Unicode MS');
#if no Arial entry, use default entity
if (! $entview) {
$entview = content_attr_known('REPL','TXT','SUP','default');
}
my $equivdesc = content_attr('EQUIV','DESC');
if ($equivdesc) {
$equivdesc = "" . $equivdesc . "";
}
my $repldesc = content_attr('REPL','DESC');
if ($repldesc) {
$repldesc = "display " . $repldesc . "";
}
my $comment = content_tag('COMMENT');
my $description = join(" ", $equivdesc, $repldesc, $comment);
#TODO check for multiple EQUIV and COMMENT tags
#put a character in to show combining usage
#for entities beginning with cmb.
if ($entsgm =~ /^cmb/) {
$entview = 'o' . $entview;
}
print "| $entsgm | $entview | $unic | $entset | $description |
\n";
}
print "
\n\n";