XML::Filter::SAXT - Replicates SAX events to several SAX event handlers

 $saxt = new XML::Filter::SAXT ( { Handler => $out1 },
				 { DocumentHandler => $out2 },
				 { DTDHandler => $out3,
				   Handler => $out4 
				 }
			       );

 $perlsax = new XML::Parser::PerlSAX ( Handler => $saxt );
 $perlsax->parse ( [OPTIONS] );

SAXT is like the Unix 'tee' command in that it multiplexes the input stream to several output streams. In this case, the input stream is a PerlSAX event producer (like XML::Parser::PerlSAX) and the output streams are PerlSAX handlers or filters.

The SAXT constructor takes a list of hash references. Each hash specifies an output handler. The hash keys can be: DocumentHandler, DTDHandler, EntityResolver or Handler, where Handler is a combination of the previous three and acts as the default handler. E.g. if DocumentHandler is not specified, it will try to use Handler.

In this example we use XML::Parser::PerlSAX to parse an XML file and to invoke the PerlSAX callbacks of our SAXT object. The SAXT object then forwards the callbacks to XML::Checker, which will 'die' if it encounters an error, and to XML::Handler::BuildDOM, which will store the XML in an XML::DOM::Document.

 use XML::Parser::PerlSAX;
 use XML::Filter::SAXT;
 use XML::Handler::BuildDOM;
 use XML::Checker;

 my $checker = new XML::Checker;
 my $builder = new XML::Handler::BuildDOM (KeepCDATA => 1);
 my $tee = new XML::Filter::SAXT ( { Handler => $checker },
				   { Handler => $builder } );

 my $parser = new XML::Parser::PerlSAX (Handler => $tee);
 eval
 {
    # This is how you set the error handler for XML::Checker
    local $XML::Checker::FAIL = \&my_fail;

    my $dom_document = $parser->parsefile ("file.xml");
    ... your code here ...
 };
 if ($@)
 {
    # Either XML::Parser::PerlSAX threw an exception (bad XML)
    # or XML::Checker found an error and my_fail died.
    ... your error handling code here ...
 }

 # XML::Checker error handler
 sub my_fail
 {
   my $code = shift;
   die XML::Checker::error_string ($code, @_)
	if $code < 200;	  # warnings and info messages are >= 200
 }

This is still alpha software. Package names and interfaces are subject to change.

Enno Dersken is the original author.

Send bug reports, hints, tips, suggestions to T.J. Mather at <tjmather@tjmather.com>.