Given that you've read your file into a scalar:

my (@articles) = ($text =~ /(<a name.*?<!--start-->.*?<!--end-->)/gs);
[download]

Of course this is doing two stingy matches in one pattern which is probably godawful slow.

Update: Or use a hash instead

my (%hash) = ($text =~ /(<a name.*?)(<!--start-->.*?<!--end-->)/gs);
[download]