Sorry, couldn't resist :) : Once again a standard application of the nice data structure suffix trees/arrays. It is in general easier to explain this with suffix trees, but it also works quite analogous with arrays (which have several advantages).
Take a look at the mississippi
example in http://www.csse.monash.edu.au/~lloyd/tildeAlgDS/Tree/Suffix/
. Now just enumerate all edge labels of nodes with more than one leaf: issi, i, ssi, si, p
. These are all repeats because they are common prefixes of different suffixes of the string. There are algorithms that output only the maximal (overlapping) repeats issi
Construction and calculation are both possible in linear time, but obviously with a lot of overhead (so the string must be quite large that this pays off - and then you don't want a Perl implementation).
And btw. it is not an "artificial problem". The human genome consists of many, many repeats (this is in fact the reason why the assembly is so hard) and we don't know much about most of them.
See also: http://en.wikipedia.org/wiki/Suffix_tree