How can I create the XML::Simple data structure using a Perl XML SAX parser?

Answer 1

XML::Twig has a simplify method which you can call on a XML element which according to docs says:

Return a data structure suspiciously similar to XML::Simple's

Here is an example:

use XML::Twig;
use Data::Dumper;

my $twig = XML::Twig->new(
    twig_handlers => {
        rec => \&rec,
    }
)->parsefile( 'data.xml' );


sub rec {
    my ($twig, $rec) = @_;
    my $data = $rec->simplify;
    say Dumper $data;
    $rec->purge;
}

NB. The $rec->purge cleans out the record immediately from memory.

Running this against your XML example produces this:

$VAR1 = {
          'f1' => 'v1',
          'f2' => 'v2'
        };

$VAR1 = {
          'f1' => 'v1b',
          'f2' => 'v2b'
        };

$VAR1 = {
          'f1' => 'v1c',
          'f2' => 'v2c'
        };

Which I hope is suspiciously like what comes out of XML::Simple :)

/I3az/

Answer 2

As the author of XML::Simple, I'd just like to correct some misconceptions in your question.

XML::Simple isn't a DOM parser, in fact it isn't a parser at all. It delegates all parsing duties to either a SAX parser or XML::Parser. The speed of parsing will depend on which parser module is the default on your system. When you run 'make test' for the XML::Simple distribution, the output will list the default parser.

If the default parser on your system is XML::SAX::PurePerl then it will be slow and more importantly buggy too. If that's the case then I'd recommend installing either XML::Expat or XML::ExpatXS for an immediate speed up. (Whichever SAX parser is installed last will be the default from that point).

Having said that, your requirements are a bit contradictory, you want something that returns your whole document as a hash and yet you don't want a parser that slurps the whole document into memory.

I understand your short-term goals, but as a longer term solution, I'd recommend migrating your code to XML::LibXML. It is a DOM parser but it's very fast because all the grunt work is done in C. Best of all the built-in XPath support makes it even simpler to use than XML::Simple - see this article.

Answer 3

Take a look at XML::LibXML::Reader.