Changeset 3181 for trunk/gsdl/perllib/ghtml.pm

Timestamp:

2002-06-25T20:15:26+12:00 (22 years ago)

Author:

sjboddie

Message:

Altered the getcharequiv() function so it now converts entities to raw
utf-8 characters. This should now work for entities like ∏ (i.e.
characters outside of the latin 1 character set).

File:

• trunk/gsdl/perllib/ghtml.pm (modified) (2 diffs)

trunk/gsdl/perllib/ghtml.pm

@@ -27 +27 @@
 
 package ghtml;
+
+use unicode;
 
 # htmlsafe(TEXT)
@@ -142 +144 @@
 
 
-# returns the character in the standard html font. It assumes that the
+# returns the character as a raw utf-8 character. It assumes that the
 # & and ; have been stripped off the string.
 sub getcharequiv {
     my ($entity, $convertsymbols) = @_;
 
-    # replace &#8218 with comma (",")
-    $entity =~ s/8218/044/;
-
     # a numeric entity
     if ($entity =~ /^\#0*(\d+)/) {
-    return pack("c", $1);
+    return &unicode::unicode2utf8([$1]);
     }
     
     # a named character entity
     if (defined $charnetosf{$entity}) {
-    return pack("c", $charnetosf{$entity});
+    return &unicode::unicode2utf8([$charnetosf{$entity}]);
     }
 
     # a named symbol entity
     if ($convertsymbols && defined $symnetosf{$entity}) {
-    return pack("c", $symnetosf{$entity});
+    return &unicode::unicode2utf8([$symnetosf{$entity}]);
     }