Further, it seems to me that it would not be hard to determine if
the text/plain and text/html parts are essentially identical.
This is a more difficult problem than you imagine. You'd have to render
the HTML, and then decide what "essentially identical" means. For many
it would be differences in whitespace only, but for others a different
definition may apply e.g; HTML is a superset of the plain text, and only
contains additions.
I "solve" this problem by using the MIME-hooks scripts I developed in
February. They create a backup of the original message and save it
as #.msg, then mhfixmsg is forced to replace the text/plain part with
a converted version of the text/html. This works well, although the
formatting is a bit different for the odd sender that actually makes
an effort to produce a useful text/plain. Most do not of course,
which is why this was adopted in the first place.
The MIME-hooks scripts work quite well. The only nit I have with them is
how they interact with my Trash folder, since I have rmm aliased to refile
things instead of remove them. I usually let that folder get quite large,
and packing it after culling the first thousand messages can be slow as
the remaining hundreds of files are renamed one at a time by repeated
invocations of mime-ref-hook.
_______________________________________________
Nmh-workers mailing list
Nmh-workers(_at_)nongnu(_dot_)org
https://lists.nongnu.org/mailman/listinfo/nmh-workers