I believe this might be a little buggy, unless I misunderstand the requirement. But take a look at:
use strict;
my $input='
This is sample <This is sample text>
This is sample <Th<a> </a>is is sa<not a>mple text <a>this is samp
+le text</a> sample text >';
my @arr=$input=~m#<[^<>]+(?:<a>.+?</a>[^<>]*)?>#g;
print join "\n",@arr;
I'm wondering about what might be inside <a>..</a>, and whether they should be balanced. In case not, you could simply go for:
qr {<(?:[^<>]+|<a>|</a>)*>}
Or perhaps:
qr {<(?:[^<>]+|<a>|</a>)*[^>]*>}
Otherwise, you might try:
my $re;
$re=qr{(?:[^<>]+|<a>(??{$re})?</a>)*};
my @arr=$input=~m#<$re>#g;
