Changeset 22952

Timestamp:

2010-09-23T17:22:49+12:00 (13 years ago)

Author:

davidb

Message:

Encode::decode cannot be applied to all characters returned by ghtml::getcharequiv(). If getcharequiv does not recognize a character then it does not encode it, and so we cannot apply decode() to it. getcharequiv() upgraded to include an optional extra param that says whether or not to decode the equiv char should there be one it can map it to.

File:

• main/trunk/greenstone2/perllib/ghtml.pm (modified) (2 diffs)

main/trunk/greenstone2/perllib/ghtml.pm

@@ -202 +202 @@
 # & and ; have been stripped off the string.
 sub getcharequiv {
-    my ($entity, $convertsymbols) = @_;
+    my ($entity, $convertsymbols, $and_decode) = @_;
+
+    my $char_equiv = undef;
 
     # a numeric entity
@@ -214 +216 @@
         elsif ($code == 0x94) {$code=0x201d} # 148 = double right quote
         # ...
-    }
-    return &unicode::unicode2utf8([$code]);
+    }   
+    $char_equiv = &unicode::unicode2utf8([$code]);
     }
     
     # a named character entity
-    if (defined $charnetosf{$entity}) {
-    return &unicode::unicode2utf8([$charnetosf{$entity}]);
+    elsif (defined $charnetosf{$entity}) {
+    $char_equiv = &unicode::unicode2utf8([$charnetosf{$entity}]);
     }
 
     # a named symbol entity
-    if ($convertsymbols && defined $symnetosf{$entity}) {
-    return &unicode::unicode2utf8([$symnetosf{$entity}]);
-    }
-
-    return "&$entity;"; # unknown character
+    elsif ($convertsymbols && defined $symnetosf{$entity}) {
+    $char_equiv = &unicode::unicode2utf8([$symnetosf{$entity}]);
+    }
+
+    if (!defined $char_equiv) {
+    return "&$entity;"; # unknown character
+    }
+    else {
+    if ((defined $and_decode) && ($and_decode)) {
+        $char_equiv = Encode::decode("utf8",$char_equiv);
+    }
+    return $char_equiv;
+    }
 }