m/\bDE\d+/g
[download]

m/\b(DE\d+)/g
[download]

# Would like to see the page source but this will work
# for one target string per line.

# $_ contains mypage.html
 
print $1 if /\s(DE\d{6})/;
[download]

I came up with this code:

perl -n -e 'while(m/\b(DE\d{6})/g){print "$1\n";}' mypage.html
[download]

There is also this slight variation:

perl -n -e 'print "$1\n" while(/\b(DE\d{6})/g)' mypage.html
[download]

but I prefer the first because as a general rule I don't like suffixing statments with conditionals except for error handling.

It is slightly longer but does handle multiple matches in a line and it also breaks apart the output to one "DE" per line.

that is exactly what I mean... without a \n or other separater the output is harder to read because it runs together.

Well, sorry :-) I thought the lack of \n was kinda obvious.
So if thats the case, I'd say your "$1\n" really should be $1,"\n" on the grounds of efficiency. 25% faster. D'oh, I must admit your solution is better than mine. :-)