I would very much recommend not parsing HTML with regular expressions - see Parsing HTML/XML with Regular Expressions. For example, see HTML::LinkExtor:

use HTML::LinkExtor;

my $p = HTML::LinkExtor->new();
$p->parse_file("test.html");
my @hrefs = map { {@$_[1..$#$_]}->{href} } $p->links;
[download]

On your sample data, @hrefs is ("test1.htm", "test2.htm", "test3.htm").