Bounced emails: automatically deactivating bad addresses...

Question

Just wanted a little feedback on what strategy might be best for automating the deactivation process. I have a database with a list of addresses that needs to be frequently updated-- any messages that I receive in a certain inbox are typically bounced emails and I wish to deactivate these addresses in my database so I no longer send to them.
The best method I can think of, short of going through and manually identifying the original address I sent to (I've got better things to do!), seems to be to develop a method which parses the entire header and finds any email addresses (rebuilding any tokens that contain '@') and through brute force, simply tries to deactivate any and all reconstructed email addresses that I find. Not a very elegant method, I know. Since I won't be needing to automatically send emails to myself (return addresses and original sender) and anyone else of the 'postmaster' or 'MAILER-DAEMON' variety, I assume that any other email addresses contained within the header (somewhere)will be the one(s) that I'm looking for. Is this a fair assumption?
Has any got a suggestion on how to improve this crude parsing? Everything I've read online suggests that a 100% success rate is impossible due to the lack of standards imposed on header content. Anything I send out is not guaranteed to come back to me, according to my understanding, so what other options may be available?
Cheers,
Sean

Answer 1

Here's the first bit of code for testing my brute-force method... It seems to work fine as long for generating possible addresses as there are no instances of '@' at the beginning or end of a parsable-string-- ie. " @xxx", "xxx@ ", etc.. I'm working on that part right now... shouldn't be too hard. Can anyone see possible/definite flaws or improvements/issues I may want to consider? Any comments on this code are much appreciated.
<code>
String content = new String("a string generated from email content");
               StringTokenizer st1 = new StringTokenizer( content, "@" );     //finds possible instances of email addresses
               int numberTokens=st1.countTokens();
               int record=0;
               String garbage;
if(numberTokens>1){
               // create array to hold user and hostname values for all addresses
               String address[] = new String[ 2*(numberTokens-1) ];
               for( int i=0; i<numberTokens; i++ ){
                    String content2 = new String( st1.nextToken() );
                    StringTokenizer st2 = new StringTokenizer( content2, " <>()" );     //set characters which may denote/separate addresses
                    int numberTokens2 = st2.countTokens();
                    for( int j=0; j<numberTokens2; j++ ){
                         if( ((i==0)&&(j==numberTokens2-1)) ||     //to get first username     (first '@' token)
                              ((i==numberTokens-1)&&(j==0)) ||     //to get last domain name (last '@' token)
                              ((i>0&&i<numberTokens-1)&&(j==0||j==(numberTokens2-1))) ){     //get username/domain (encapuslated '@' tokens)
                                   address[record] = st2.nextToken();
                                   record++;
                         }else if( (st2.hasMoreTokens()) )     //move parser to next token if exists
                                                                      //or let for-loop expire.                          
                                   garbage = st2.nextToken();
               for(int k=0;k<address.length;k++ )     
                    System.out.println(address[k]);     //let's see how we did
</code>