So it didn't work, here's what's interesting:
Code:

   $line =~ m/(.*?)(\d+\/\d+\/\d+)?\s*(.*)/;
   print "Input    : $line";
   print "Output \$1: -$1-\n";
   print "Output \$2: -$2-\n";
   print "Output \$3: -$3-\n\n";
[download]

Results:

Input    :                          301 S. MAPLE STREET
Output $1: --
Output $2: --
-utput $3: -301 S. MAPLE STREET

Input    :  09/09/2009              301 S. MAPLE STREET
Output $1: --
Output $2: --
-utput $3: -09/09/2009              301 S. MAPLE STREET
[download]

I included the "neat" flag (x), but decided to take it away. For all purposes, the above should be the same. Note: I put the first "ungreedy capture anything" (.*?) in $1 for debugging purposes. The results are interesting.

One interesting thing is the termination '-' that is being outputted to the front of the line, which makes it look like more is going on here than meets the eye.

What I really need is a lookbehind. Note: even if I get rid of the first .*, there are still problems.

The same thing is throwing it off (the date capture). If I remove or change the '?' to a '+', then it works for the second input, but not the first input. The first input reuses $1 and $2 from a previous regex (not shown) $3 empty; hence, it doesn't match at all.

Example:

Input    :                          301 S. MAPLE STREET
Output $1: -000001-
-utput $2: -                  JOHN SMITH, III
Output $3: --

Input    :  09/09/2009              301 S. MAPLE STREET
Output $1: - -
Output $2: -09/09/2009-
-utput $3: -301 S. MAPLE STREET
[download]