One way to make the matching a little more "fuzzy", would be to strip
all leading and trailing white space on each line of the body, and also
compress the intervening whitespace. I will investigate this the next
time I visit (and edit) dupcheck.rc, unless someone else volunteers.
Add a couple more steps:
Expand tabs, and use fmt(1) to undo the bad habit of Netscape (and others)
of breaking lines when the sender doesn't (or when the sender broke the
lines one column past where the tool thought he should have #$%^%#^)
Another heuristic is to check if the entire body has been quoted, and,
if so, remove the leading quote string (ie: "> ") and do the checksum on
that (or throw it way :^).
Irritating, but doesn't occur often enough to justify the extra coding.
Trailer recognition is another "feature" which might be accomplished,
with some heuristics, but, in general, is extremely difficult to get
right.
And shouldn't do it at all if can't do it reliably. Unfortunately (for
me) this would mean I still would get lots of duplicates.
I've included the "dupcheck.rc" file here, for your convenience:
Thanks, I may try it.