Is the tm element the only child of another node? If so, you can get the parent of the matching node, remove it's children, then insert the text node:

use strict;
use warnings;
use XML::LibXML;

my $xml_str = <<_XML_STR;
<foo>
  <bar>
    <tm tmclass="ibm" tmowner="IBM Corporation" 
        tmtype="reg" trademark="AIX">AIX</tm>
  </bar>
</foo>
_XML_STR

my $parser = XML::LibXML->new();
my $doc    = $parser->parse_string( $xml_str );
my $query  = "//tm";

foreach my $node ( $doc->findnodes($query) ) {
  my $val = $node->findvalue( '@tmowner' );
  my $txt = $node->textContent();

  if( $val !~ /my_company/ ) {
    my $parent = $node->parentNode();
       $parent->removeChildNodes();
       $parent->appendTextNode( $txt );
  }
}

print $doc->toString;
[download]

<foo>
  <bar>AIX</bar>
</foo>
[download]

Thanks, Derby, this is a step in the right direction.

Unfortunately, it is not the only child of another node. In fact, there may be multiple <tm> or other children within a given parent.

This <tm> element is being used to trigger inline formatting during output processing. In this example, a superscript "TM" would be added automatically like AIX^TM. So there may be more than 1 in the same sentence, paragraph, etc. within a single parent.

Testing your suggestion, it certainly removes the <tm> child, but appends the "AIX" to the very end of the parent, as one would expect (and as you insinuated).

I am currently looking into the XML::LibXML::Document class. It has some methods that look promising, perhaps I can use a fragment or text node somehow.

Thanks for looking into this!

SC

open file.xml ;
for //tm mv text() replace . ;
save :b ;
[download]

Note: Does not work if <tm>'s can be nested.

Update: If <tm> can contain something more than just a text, you have to use the more general "unwrap" method:

for //tm xmv node() replace . ;
[download]

Tested with

<tm>a<b>c</b>c<?run test?><!-- comment --></tm>
[download]

Thanks for the reply. I will look into XML::XSH2, as I am unfamiliar with that module.
SC

1. Build a text-node containing the new content.
2. Navigate to the parent node and instruct it to replaceChild the existing node with the new text-node.

Your step #1 was the ticket! Thanks AM. I was not first constructing a new text node.
The simple fix:

  $parentnode = $node -> parentNode;
  $txtnode = XML::LibXML::Text ->new($tmtext);
  $node = $parentnode -> replaceChild($txtnode, $node);
[download]

SC

#!/usr/bin/perl

use strict;
use warnings;
use XML::Twig;

my $twig= XML::Twig->new(
                        pretty_print => 'indented',
                        twig_roots    => { 'tm' => 1 },
                        twig_print_outside_roots => 1,
                        twig_handlers => {   tm=>sub{my $text = $_->te
+xt(); $_->cut();print $text;  },
                        },
);
$twig->parse('<?xml version="1.0"?><stats><tm tmclass="ibm" tmowner="I
+BM Corporation" tmtype="reg" trademark="AIX">AIX</tm></stats>');
[download]

#declare a global:
my $only_one=0;

#same as before, then:
twig_handlers => {   tm=>sub{my $text = $_->text();   $only_one ? $_->
+cut()  : ( $_->cut() and print $text and $only_one++ ) },

__DATA__
<?xml version="1.0"?>
                <stats>
                  <tm tmclass="ibm" tmowner="IBM Corporation" tmtype="
+reg" trademark="AIX">AIX</tm>
                  <tm tmclass="ibm" tmowner="IBM Corporation" tmtype="
+reg" trademark="AIX">AIX</tm>
                  <tm tmclass="ibm" tmowner="IBM Corporation" tmtype="
+reg" trademark="AIX">AIX</tm>
                </stats>'

__OUTPUT__
<?xml version="1.0"?>
                <stats>
                  AIX


                </stats>
[download]

This can certainly be accomplished via regexps (usually my default mode for this sort of thing). In this case, however, I'm dealing with thousands of files authored by an assortment of tools and people, and would prefer to use a method that helps ensure that syntax and context remain intact 100% of the time when passed chaotic code. If I can't find an XPath-style solution, I will go the regexp route.
SC