Hi All,

It is probably my misunderstanding, your advices are very much appreciated. I made an xml file to show the problem, the real file(s) are much more complex.

<?xml version="1.0" encoding="utf-8"?>
<root>
  <departments>
    <department>
      <name>A</name>
      <articles>
        <article>
          <production>
            <country>France</country>
            <year>1989</year>
          </production>
          <key>001</key>
        </article>
        <article>
          <production>
            <country>Italy</country>
            <year>1991</year>
          </production>
          <key>002</key>
        </article>
        <article>
          <extra>
            <production>
              <country>Germany</country>
              <year>1995</year>
            </production>
            <extrakey>003</extrakey>
          </extra>
        </article>
      </articles>
    </department>
    <department>
      <name>B</name>
      <articles>
        <article>
          <key>004</key>
        </article>
        <article>
          <key>005</key>
        </article>
      </articles>
    </department>
    <department>
      <name>C</name>
    </department>
  </departments>
</root>
[download]

I parse the file with XML::Rabbit using the following library file (made for this example as well).

 
package House
{
    use XML::Rabbit::Root;
    
    has_xpath_object_list depts => '/root/departments/department' =>
        'House::Depts';
        
    finalize_class();
}

package House::Depts
{
    use XML::Rabbit;
    
    has_xpath_object_list articles => './articles/article[key]|./artic
+les/article/extra' => 'House::Article';
    has_xpath_value name => './name';
    
    finalize_class();
}

package House::Article
{
    use XML::Rabbit;
    
    has_xpath_value key => './key|./extrakey';
    has_xpath_object prod => './production' => 'Article::Production';
    finalize_class();
}

package Article::Production
{
    use XML::Rabbit;
    has_xpath_value country     => './country';
    has_xpath_value year    => './year';
    finalize_class();
}

1;
[download]

Now I use this library file in the following script.

 
#!/perl
use strict;
use warnings FATAL => qw(all);

use Encode;
use FindBin qw($Bin);
use lib $Bin;
use testlib;
use Try::Tiny;

my $path = shift or die "No source!\n";
my $string = do { local $/ = undef; 
   open my $in, "<", "$path" or die "$!";<$in>;};

my $xml = decode('utf8', $string); 
my $h = House->new(xml => $xml);

for my $d ( @{$h->depts} )
{
    for my $art ( @{$d->articles} )
    {
        my $prod = try {$art->prod} catch {0}; # $art->prod // 0;
        my $country = $prod ? $prod->country : '';
        print join("\t", $d->name, $art->key, $country), $/;
    }
}
[download]

You see that not all the articles have the "production" child. The line my $prod = $art->prod//0; (commented) would produce an error: "Attribute (prod) does not pass the type constraint because: Validation failed for 'Article::Production' with value undef at reader House::Article::prod (unknown origin)..." If I use try {...} catch {...}, then I get the awaited output.

A       001     France
A       002     Italy
A       003     Germany
B       004     
B       005
[download]

Now two questions.

• First, is the use of Try::Tiny here "decent" or is it just a hack and the output should be achieved without it by changing the code (probably in the library file)?
• Second, the script complains over missing xpath_object ("production") but does not seem to bother if xpath_object_list is not there: the loop over articles does not produce an error though there are not articles in the department C. Why does it make a difference?