I'm still not getting it, I'm afraid. Would this be defined at the top of a file, as in "here's a list of words and their associated links" or *within* the file, as in only certain occurrences of the words will get linked, and that link is determined by what follows the 'pipe' symbol.

Example #1

cult|http://www.slashdot.org
banana|/orange.html

Being in a cult can reduce your ability to eat a banana, doctors claim
+.
[download]

Example #2

Being in a cult|http://www.slashdot.org can reduce your ability to eat
+ bananas|/orange.html , doctors claim.
[download]

if the latter, it's so close to HTML already it's almost not worth the work of writing a script. But whatever. If the former, what you could to do is read the word/url pairs into a hash, and do something like what I suggested above. Philosophy can be made out of anything. Or less -- Jerry A. Fodor

Example #1 is what I want.

I don't expect the link file to be more than 50 or 60 pairs.

Any idea what kind of overhead on CPU this might have if the returned text file is about 7000 characters in size?

Basic code (assuming the file you're reading consists *ONLY* of word/link pairs on a single line ... I'd suggest separating them with something other than a pipe, too, but whatever.)

    # assuming the link data file is open
my $keywords;

while (<LINKDATA>) {

# regex matches (everything to the left of the pipe), puts 
# it in $1, then (everything to the right of the pipe) 
# and puts that in $2

        /([^|]+)\|(.*)/;
        $keywords{$1} = $2;
}
[download]

This should be reasonably efficient in terms of speed and not so bad on memory if your list of words isn't too large.
HTH,

Philosophy can be made out of anything. Or less -- Jerry A. Fodor