$ perl -wle'
use HTML::TreeBuilder::XPath;
my $tree = HTML::TreeBuilder::XPath->new;

$tree->parse(<<EOM);
<h3 class="blue-space">D-Lib Magazine</h3>
<p class="blue">March/April 2010<br />
Volume 16, Number 3/4<br />

<a href="../03contents.html">Table of Contents</a>
</p>
EOM

my $href = $tree->findvalue(q{//p[@class="blue"]/a/@href});
print $href'
../03contents.html