Bug #15772 for WWW-CheckSite: Wish: optional validation using XML::LibXML/xmllint for XHTML pages

On Sat Jun 02 11:31:15 2007, ABELTJE wrote: Show quoted text
Thanks. I have an alternative implementation which uses XML::LibXML directly, see the attachment. Maybe this implementation could be added to Validator.pm, probably under another name? Note also that this implementation uses an additional member "xmlparser". This member should probably be passed by WWW::CheckSite to WWW::CheckSite::Validator and may be set to a self-defined xmlparser object, e.g. to add a catalog path to XML::LibXML. Another quesion: I see tempfile(..., UNLINK => 0) calls in your code. Shouldn't UNLINK be set to 1 instead, to have temporary files cleaned up auatomatically in case of fatal errors? Regards, Slaven

sub WWW::CheckSite::Validator::validate_by_xmllint { my( $self, $stats ) = @_; eval "use File::Temp"; return if $@; my( $mech, $xmlparser ) = @{ $self }{qw( _agent xmlparser )}; $self->{v} and printf "HTML-Validate_xmllint(%s) ", $mech->uri; if (!$xmlparser) { Carp::cluck("Fallback to standard XML::LibXML parser..."); require XML::LibXML; my $xmlparser = XML::LibXML->new; $xmlparser->validation(1); $xmlparser->load_ext_dtd(1); } eval { $xmlparser->parse_string($mech->content); }; my $err = $@; if ($err ne "" && $self->{v} >= 2) { # XXX hmmm never set to anything > 1 print "$err\n"; } $stats->{valid} = $err eq "" ? 1 : -1; $self->{v} and printf " done(%sok)\n", $stats->{valid} == 1 ? "" : "not "; }