--- cvs-head/lib/WAIT/Document/Ora.pm	2002/01/23 12:22:54	65
+++ cvs-head/lib/WAIT/Document/Ora.pm	2002/01/25 07:27:30	69
@@ -14,9 +14,10 @@
 
 package WAIT::Document::Ora;
 @ISA = qw(WAIT::Document::Base);
-require WAIT::Document::Base;
+use WAIT::Document::Base;
 
 use IO::File;
+use Encode;
 use strict;
 use Carp;
 
@@ -51,13 +52,13 @@
   local($/) = undef;
 
   my $fh = IO::File->new(join('/',$self->{Dir},$file,'desc.html'));
-  my $desc = $fh->getline();
+  my $desc = conv_getline($fh);
   $fh = IO::File->new(join('/',$self->{Dir},$file,'author.html'));
-  my $author = $fh->getline() if $fh;
+  my $author = conv_getline($fh) if $fh;
   $fh = IO::File->new(join('/',$self->{Dir},$file,'index.html'));
-  my $index = $fh->getline() if $fh;
+  my $index = conv_getline($fh) if $fh;
   $fh = IO::File->new(join('/',$self->{Dir},$file,'colophon.html'));
-  my $colophon = $fh->getline() if $fh;
+  my $colophon = conv_getline($fh) if $fh;
   return {
           desc => $desc,
           author => $author,
@@ -66,6 +67,28 @@
          };
 }
 
+# WAIT::Document::Ora::conv_getline
+sub conv_getline ($) {
+  my($fh) = shift;
+  local $/ = "\n";
+  my $firstline = <$fh>;
+  my $src_enc;
+  # \042 is double quote, \047 is single quote. I avoid single quotes
+  # here just for easier copy and paste to the terminal (I need to
+  # debug here frequently)
+  if ($firstline =~ /<\?xml[^>]+encoding\s*=([\042\047])([\w\-]+)\1/) {
+    $src_enc = $2;
+  } else {
+    $src_enc = "ISO-8859-1";
+  }
+  seek $fh, 0, 0;
+  undef $/;
+  my $content = <$fh>;
+  $content =~ s/\s+/ /gs; # eliminate TABs and CRs for easier debugging
+  my $dcontent = Encode::decode($src_enc,$content);
+  $dcontent;
+}
+
 sub FIRSTKEY {
   my $self = shift;
   $self->{fno} = 0;