So first a solution using XML::Parser:

#!/bin/perl -w
use strict;
use XML::Parser;

# global, but could be attached to the parser or passed to the handler
+s
my @results;

# we need those to hold info about the parsing
my( $elt_id, $child_seen, $in_child, $child_text);  

my $p= new XML::Parser( Handlers => 
                          { Start => \&start, # called when a start ta
+g is found
                            End   => \&end,   # called when an end tag
+ is found
                            Char  => \&char,  # called when characters
+ are found
                          },
                       );
$p->parse( \*DATA);                           # use parsefile to parse
+... a file

print join "\n", @results;
print "\n";


sub start
  { my( $p, $elt, %atts)= @_;                 # set by XML::Parser::Ex
+pat

    if( $elt eq 'elt')                        # we found an elt start 
+tag
      { $child_seen= 0;                       # reset the flag, no chi
+ld found yet
        $elt_id= $atts{id};                   # store it in case we ne
+ed it
      }
    elsif( $elt eq 'child')                   # found a child start ta
+g
      { $child_seen= 1;                       # we've seen a child
        $in_child= 1;                         # we are in the child
        $child_text= '';                      # reset the child text
      }
  }

sub end
  { my( $p, $elt)= @_;
    if( $elt eq 'elt')                        # found and elt end tag
      { if( $child_seen)                      
          { push @results, $child_text; }     
        else
          { push @results, "missing child for elt $elt_id"; }
      }
    elsif( $elt eq 'child')                   # found a child end tag
      { $in_child= 0; }                       # Toto, I guess we are n
+ot in the child any more
  }

sub char                                      # called for all non mar
+k-up text
  { my( $p, $string)= @_;
    $child_text .= $string if( $in_child);    # see the docs for why y
+ou can't
  }                                           # just write $child_text
+ = $string

__DATA__
<doc>
  <dummy>I am a </dummy>
  <elt id="elt1"><child_1/><child>child 1</child></elt>
  <elt id="elt2"><child_1/></elt>
  <elt id="elt3"><child_1/><child>child 2</child></elt>
  <elt id="elt4"><child_1/></elt>
</doc>
[download]

Then the obligatory XML::Twig plug:

#!/bin/perl -w
use strict;
use XML::Twig;

my @results;                                                # does not
+ have to be global, it's just easier

# create the twig see the docs for why to use TwigRoots
my $t= new XML::Twig( TwigRoots => { elt => \&check_elt }); # call che
+ck_elt every time an element elt is parsed
$t->parse( \*DATA);                                         # parse th
+e XML (use parsefile to parse... a file)

print join "\n", @results;
print "\n";

sub check_elt                                               
  { my( $t, $elt)= @_;)                                     # $t is th
+e XML::Twig object
                                                            # $elt is 
+an XML::Twig::Elt object
    if( my $child= $elt->first_child( 'child'))             # that's h
+ow you navigate the element
      { push @results, $child->text; }                      # text inc
+ludes sub elements of child
    else
      { push @results, "missing child for elt " . $elt->att( 'id'); }
    $t->purge;                                              # call onl
+y if your document is huge
  }                                                         # to free 
+the memory


__DATA__
<doc>
  <dummy>I am a </dummy>
  <elt id="elt1"><child_1/><child>child 1</child></elt>
  <elt id="elt2"><child_1/></elt>
  <elt id="elt3"><child_1/><child>child 2</child></elt>
  <elt id="elt4"><child_1/></elt>
</doc>
[download]

Now the XML::PYX way. To run this one you need to have the document in a file, say doc.xml and to run pyx doc.xml | perl test_pyx .

#!/bin/perl -n -w
use strict;

# global, but could be attached to the parser or passed to the handler
+s
use vars qw( @results);

# we need those to hold info about the parsing
# @in_element is a stack of open elements, 
# the current element is $in_element[-1]
use vars qw( @in_element $elt_id $child_seen $child_text);  

if(m/^\((.*)$/)                # element start tag (tag
  { push @in_element, $1;
    if( $1 eq 'elt')          # elt start tag
      { $child_seen= 0; }     #   reset the flag
    elsif( $1 eq 'child')     # child start tag
      { $child_seen= 1;       #   set the flag
        $child_text= '';      #   reset the text
      }
  }
elsif( m/^A([^\s]*) (.*)$/)    # attribute Aatt value
  {                           # store the id for elt elements
    $elt_id= $2 if( ($in_element[-1] eq 'elt') && ($1 eq 'id')); 
  }
elsif( m/^-(.*)\n/)            # text -text
  { $child_text.= $1 if( $in_element[-1] eq 'child'); }
elsif( m/\)(.*)$/)             # end tag )tag
  { if( $1 eq 'elt')          
      { if( $child_seen)                      
          { push @results, $child_text; }     
        else
          { push @results, "missing child for elt $elt_id"; }
      }
  }

END
  { print join "\n", @results;
    print "\n";
  }
[download]

And finally the XML::XPath version. I am not very familiar with this module so it is probably not the most elegant way to do it, but hey, it works!

#!/bin/perl -w
use strict;

use XML::XPath;
              
my @results;

# create the xpath object from the DATA filehandle
my $xp = XML::XPath->new( ioref => \*DATA);
              
# find all paragraphs
my $elts = $xp->find('/doc/elt'); 
              
foreach my $elt ($elts->get_nodelist) 
  { # there is probably a more elegant way to get the child children
    # but I don't know XML::XPath enough
    my $children= $elt->getChildNodes;                         # get a
+ll children
    my @children= grep { $_->getName eq 'child' } @$children;  # grep 
+only the relevant ones
    if( @children)
      { push @results, $children[0]->string_value; }           # that'
+s how you get the text
    else
      { push @results, "missing child for elt " . $elt->getAttribute( 
+'id'); }
   } 
print join "\n", @results;
print "\n";

__DATA__
<doc>
  <dummy>I am a </dummy>
  <elt id="elt1"><child_1/><child>child 1</child></elt>
  <elt id="elt2"><child_1/></elt>
  <elt id="elt3"><child_1/><child>child 2</child></elt>
  <elt id="elt4"><child_1/></elt>
</doc>
[download]