Travel Forum / Travel Types / RV Travel / November 2008

On Nov 14, 9:04 am, Elliot Richmond <xmrichm...@xaustin.xrr.xcom>
wrote:

She is my "record holder", with 20 entries in the kill file... sigh.
Ya know she actually used to post things about RVs, but then so did a
bunch of others (like Carl A, I recently checked and couldn't find an
RV posts from him since 2006, but then he's in the kill-file too)

-------- crap snipped -----------

Most of us filter her new identity after reading the opening sentence of her
latest post.

Unfortunately there is no way we can filter to eliminate someone (who is
usually worth reading) when he re-posts her crap.

Please don't ever do it again <VBG>

Sure.  You ain't been payin' attention.  I posted one just the other day.  The
key is using regular expressions that have to be enclosed in curly braces
instead of Agent's native language.  Here it is again.

Author: {.*[Cc]arol(ine)?.*}

Explanation:

".*" means match zero or more characters before the next character.

[Cc] in the square brackets mean match any single character.  In this case "C"
or "c".

The next letters are literals.

(line)?

matches "Caroline", "Carol", "caroline" and "carol"

the last ".*" means to match zero or more characters after the last specified
character or pattern.

In a nutshel, this nukes Caroline or Carol anywhere in the "Author: string

She's too narcissistic to post without something resembling "Caroline" so that
filter should work well.  It has for me for weeks since I figured out the
difference between Agents's language and its regular expression language.

If she tries some BS like cArOlInE, this very paranoid regular expression will
handle it.

Author: {.*[Cc][Aa][Rr][Oo][Ll]([Ii][Nn][Ee])?.*}

I can make this one even more aggressive but I don't want to give away all my
tactics up front.  Let the bitchlet figure out how to beat these (if she can)
before we send out the second wave.

If she tries to use extended/ISO characters such as "Ä", simply put it in the
square brackets like this:

Author: {.*[Cc][AaÀÁÂÃÄÅÆàáâäãåæ][Rr][Oo][Ll]([Ii][Nn][Ee])?.*}

Actually I went ahead and selected all the "A"-like characters in the lower
128 for your convenience.  Remember that ANY character at that position that
is inside the square brackets matches.

Basically, if she wastes her time trying to defeat this rule set (what waste
of a tiny intellect), we can make a regular expression that will handle it.
Her vanity won't allow her to post completely anonymously so we can always
snack something identifiable

Here's a really brutal one:

Author: {.*[CckK][AaÀÁÂÃÄÅÆàáâäãåæ][Rr][Oo].*}

That nukes anything with a C or K, followed by anything "A-like" and then
"O-like"

I added the "O" so that it wouldn't nuke people with "Car" in their names or
handles.

To modify an old expression, "The regular expression is mightier than the
fool".

If you set the filter action to "mark read", you can occasionally glance and
make sure it isn't nuking something unrelated to the bitchlet.  That's
generally what I do - make widely encompassing regular expressions and then
set the action to "mark read".  I'd rather there be a few false positives that
I can detect by visual scanning than there be false negatives that let her
kind of spewage through.

I highly suggest you go back to my previous regular expression post and find
the URL to the tutorial.  It is very good and covers many of the subtleties of
REs that I wasn't aware of.

Lastly, one can have more than one line of regular expression in a filter rule
and they can be linked by logical operators.  RTFM on the details.  I'm still
learning myself.  You could have a three line one that matched any of the
above AND "princess" )OR ("democratic") OR ("democrat")

If you want to see some VERY impressive REs that take hours to understand,
install Firefox's adblocker and then the AdBlock Filterset.G.  They be some
impressive REGEX talkers in that bunch!

What I REALLY wish is that Agent included some sort of semantic analysis,
perhaps with that nice Bayesian spam filter code that they already have
working.  Impressively, I might add.  That way writing styles could be
identified and filtered and we'd not have to worry about REGEXes.  Just feed
the a.shole Bayesian filter enough BS and it'll develop its own nose for the
stuff.

I gotta write that up and send it in as a feature request.

BTW, if you want to practice with REGEX off-line, I highly suggest installing
the GNU core utilities and the work with either Awk, Grep or one of the higher
languages like Pearl.  Grep is kinda the reference implementation (the name
stands for "generalized regular expression processor") so that's what I'd work
with.  Capture some of the bitchlet's headers, write your expressions (grep
will accept them in a separate file) and see what happens.  You can work out
your expressions on your own computer without having to wait for more spew
from her to appear.

There are some essential features like being able to cause all incoming
traffic to be piped through grep filters, that we had 20+ years ago with
Unix/BSD mail and news readers that we STILL don't have with all this "modern"
code.  I could not give a sh*t less whether I can process HTML in a mail
message but I DO care if I can run the message content through filters,
something that I CAN do with Agent and email but CAN NOT do with Agent's news
reader but COULD do 20 years ago with the NN newsreader and the ELM mail
reader.

Other nice "features" that we've lost include being able to use your own
familiar editor to write the messages (vi in my case) and the ability to edit
received messages.  I could, for example, change the subject line to something
more relevant to the content, cut out excessively quoted stuff and other crap
before archiving it.  Were I on the receiving of this message, for instance,
I'd call it up in vi (one keystroke) and change the subject to "regular
expression filtering" or something more appropriate and archive it with a few
more keystrokes.

Winders.  One tiny tippy-toe forward, 1000 giant leaps backwards.

John

--
John De Armond
See my website for my current email address
http://www.neon-john.com
http://www.johndearmond.com <-- best little blog on the net!
Tellico Plains, Occupied TN
I love cats ... they taste just like chicken.

Why?