Matching filenames can lead to catastrophic backtracking

[nbargnesi]

Similar to #17, matching filenames will result in a lot of backtracking as Python tries to match the .+ in strings without a match.

Here's an example using a large string containing no matches:

import re
test = ('ABCDEF' * 100000)
re.findall("(.+(\.([a-z]{2,3}$)|\/.+\/|\\\.+\\\))+", test)

Matching the max path length on Windows, and putting a 260 character limit in place of the .+ seems a better solution. It will extract filenames and not consume as many cycles when no matches are found.