Changeset 22636

Timestamp:

2010-08-16T11:41:22+12:00 (14 years ago)

Author:

davidb

Message:

Using -utf8 as options to html-tidy leads to wrong encoding for HTML docs in demo collection. Using '-raw' works better.

File:

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

main/trunk/greenstone2/perllib/plugins/HTMLPlugin.pm

@@ -1577 +1577 @@
 
 
-    my ($language, $encoding) = $self->textcat_get_language_encoding ($input_filename);
-    #print STDERR "*** encoding = $encoding\n";
+###    my ($language, $encoding) = $self->textcat_get_language_encoding ($input_filename);
+###    #print STDERR "*** encoding = $encoding\n";
 
     # run html-tidy on the tmp file to make it a proper XML file
@@ -1587 +1587 @@
     my $tidy_cmd = "tidy";
     $tidy_cmd .= " -q" if ($self->{'verbosity'} <= 2);
-    $tidy_cmd .= " -utf8 -wrap 0 -asxml \"$tmp_filename\"";
+##    $tidy_cmd .= " -utf8 -wrap 0 -asxml \"$tmp_filename\"";
+    $tidy_cmd .= " -raw -wrap 0 -asxml \"$tmp_filename\"";
     if ($self->{'verbosity'} <= 2) {
     if ($ENV{'GSDLOS'} =~ m/^windows/i) {