I am looking forward to your answer to the OP's question based on XML::Rules.

I've posted about XML::Rules as a replacement for XML::Simple before: Re: What is correct way to reference?, Re^2: hashref with(out) arrays. The following at least generates a consistent Perl data structure, and can be adapted as needed.

use warnings;
use strict;
use XML::Rules;
use Data::Dump;

my $parser = XML::Rules->new(
        stripspaces => 3|4,
        rules => [ _default => 'raw' ],
    );

dd $parser->parse(<<'ENDXML');
<xmltop>
    <xmlbody>
        <tag1>jdjdjdjd</tag1>
        <tag2>dkdkkd</tag2>
    </xmlbody>
</xmltop>
ENDXML

__END__

[
  "xmltop",
  {
    _content => [
      [
        "xmlbody",
        {
          _content => [
            ["tag1", { _content => "jdjdjdjd" }],
            ["tag2", { _content => "dkdkkd" }],
          ],
        },
      ],
    ],
  },
]
[download]

Update: One possible caveat here is that if an XML element has an attribute _content, then its value will be prepended to the node's text content.