Agent Word Filters

Note: changes to reflect the release of Agent 1.9 have not yet been
made.

The following is a list of Agent 'classic' filters: they are (word)
oriented, no {regular expression} syntax. They are meant to give
ideas, NOT to be adopted as a working set. For this material to make
sense, you must be familiar with the basic information in the help
file.

Generally, (word) filters are quicker to create, faster to run and
easier
to understand. If you absolutely have to filter on things that (word)
filters can't handle [punctuation or sub-strings, for example], see
the
extensive list of {regular expression} filter examples and related
advice/tips given in
Agent {regular expression} filters mini-FAQ

To point out an error, suggest an editorial correction, or submit a
new
filter [or expression] for inclusion, post to
news:alt.usenet.offline-reader.forte-agent
or use the guest book on my home page.

Changes in V1.02:
V1.02.01 Added new 'money' filter term
Changes in V1.01:
V1.01.01 Added some From: notes and examples

Watch out for side effects when adapting these filters for your use.
The 'money' filters, for example, may not be useful in groups where
subjects containing numbers in other contexts normally occur, and the
'spam word' filters may need to be used judiciously or overridden in
certain groups [to avoid hitting more messages than intended].

The overall priorities I have in mind when setting up multiple filters
on similar subjects/authors so that they will interact correctly is as
follows:
1000 watch global specific posters [such as yourself]
900 kill global kill certain spammers even if they use 'allowed'
words
800 watch local allow certain spam words in specific groups
700 kill global spam words/posters
600 watch local specific watches [usually size or author]
500 watch local generic watches [usually subject]
400 kill local kills [usally size or author]; use <I>gnore for
threads
300 watch local 'default' processing
200 kill local 'left-over' processing
The functions assigned to the priorities are mine [not a requirement
of Agent]. In theory, all filters could have the same priority, but
that arrangement is scarcely practical [because filters of the same
priority are applied in a sequence which can neither be controlled by
the user nor predicted from the order in which they are listed in the
Usenet filters window]. A sort of middle ground [Agent's defaults of
watch=600 and kill=400] works well for users with few filters, none of
which are 'overlapping' or 'conflicting'.

Kill filters may be "skip" or "mark read"
until you are sure that the filter[s] are doing what you want, you
may prefer the latter
Watch filters may be "retrieve" "mark for retrieval" or "watch thread"
the last [in conjunction with the View feature] will highlight all
messages in threads to which you have contributed.

FWIW I follow this pattern for setting expirations:
watch filters
generally, no expiration
kill filters
Subject:
no expiration
From: 'constant' names of 'real' individuals
no expiration
From: 'pattern' names of frequent pests
no expiration
From: 'hit-and-run' accounts
10 days from last activity

Watch your own posts
» 1000 watch from: your.name@here.com
» 1000 watch from: (your.name@here.com or your.nym@elsewhere.com)
Note: using action "watch thread" will highlight posts in threads to
which you have contributed.
Also: by selecting an appropriate View you can see all messages in
those threads and nothing else.

Get rid of all posts by an annoying author
» 900 kill from: porno@star.org
+ +++ Added in V1.01
+ Note: the entire From field is tested by the filter, not just the
+ part selected to display in the message list pane
+ Also: if there is more than one header of a given type in a message
+ [rare with From, common with others] the filter will apply to all of
+ them as if they had been concatenated
+ + Get rid of posts by an annoying author without affecting posts by
+ authors with similar names
+ » 900 kill from: =pls.see.addr@my.sig
+ Hits that address without affecting
+ pls.see.addr@my.sig (name)
+ or
+ "handle" <pls.see.addr@my.sig>
+ Note: since punctuation is ignored in (word) filters
+ (pls) see addr @ my-sig
+ would also be hit.
+ + Get rid of posts by an annoying author without affecting posts by
+ authors with similar names
+ » 900 kill from: pls.see.addr@my.sig (name)
+ Hits that address without affecting
+ pls.see.addr@my.sig
+ or
+ "handle" <pls.see.addr@my.sig>
+ Note: since punctuation is ignored in (word) filters
+ "pls" see 'addr' - my-sig [name]
+ would also be hit.
+ +++

Allow certain keywords in some [or all] groups
» 800 watch subject: ( free agent | freeagent )
» 800 watch subject: ( freed* | freew* | freez* | ... )
» 800 watch subject: ( cashew* | cashier* | cashmere* | ... )
To override
subject: free*
subject: cash*
kill filters farther down in the priority list given here.

Allow certain posters
» 800 watch from: my.friend@earthlink.net
» 800 watch from: ( good.guy@earthlink.net or my.friend )
Overrides the spammer 'domain' list, below.

The spammer 'domain' list
» 700 kill from: (earthlink.net or savetrees.com or ...)
» 700 kill from: (earthlink or savetrees or ...)
Only the 'keywords' from the 'domain' are really needed.

Get rid of certain types of "offers"
» 700 kill http | ftp | www | url | 900 | 800 | 888 | 011 ...
Note: since no qualifier is specified in the filter, those items are
found in any filterable header; in usenet filters, that means either
subject: or from:.

Messages of a certain minimum size
» 600 watch [501,*]
Caution: all 'size' parameters assume that the option
[x] Join multipart messages when retrieving headers
is enabled; if it isn't, the size test will be performed separately
on each part, possibly with unintended consequences.
Note: there is an interesting 'gotcha'; when the parts are joined
automagically, the size used to determine whether a [low,*] filter
should apply is arrived at by multiplying the size of the first part
that happens to download by the number of parts stated in the tag.
In some cases that results in quite odd-looking behaviour [especially
when the final part, which is often much smaller than the others, is
encountered first].
Also: on some servers that filter will miss all messages posted with
[Free] Agent [because they all appear to have zero lines regardless of
their actual lengths].

Messages of a certain maximum size
» 600 watch [*,1587]
Note: on some servers that filter will hit all messages posted with
[Free] Agent [because they all appear to have zero lines regardless of
their actual lengths]; in a 'watch' that quirk may be a lesser problem
than it is in a kill.
Also: assuming 'lines' of 63 bytes [the 'standard' for uuencoding]
that filter cuts off at approximately 100kb; for messages posted in
other formats, the upper limit may not work as expected [and must be
adjusted in any case to your desired cutoff point].

Messages of a certain range of sizes
» 600 watch [501,50000]
Neither too small nor too large.

Messages of a certain size
» 600 watch [0,0]
Picks up all [Free] Agent posts on certain servers.

Messages of a certain minimum size
» 600 watch [501,*] & ! from: ( dumb | stupid ) & ! subject ( MMF |
sex* )
But prevents some authors and some subjects from being watched.

Messages of a certain minimum size
» 600 watch [501,*] and subject: (jpg or gif or ...)
But only those indicating the desired file type.

Messages of a certain maximum size
» 600 watch [*,300] and subject: Colorado
But only for a specific subject.

Messages meeting certain criteria
» 600 watch subject: ( senior & software & SE & (San Francisco |
Sunnyvale | Palo Alto) )
All of the first three elements must appear in conjunction with at
least one of the last three.

Messages meeting certain criteria
» 600 watch subject: ( ( grow* or pick* ) and ( apple* or orange* ) )
At least one of the first two and at least one of the last two must
appear.

Watch for certain domains
» 600 watch from: ( uk or nz or au )
Assumes that 'real' domains actually appear in the From: header.

Messages on a certain topic [regardless of author]
» 500 watch subject: filter*

Messages of a certain size [usually in binary groups]
» 400 kill [1,199]
Caution: as explained in the help file, the seemingly equivalent
[*,199] <-- bad
is NOT generally recommended [because on some servers that would kill
every post made with [Free] Agent].
Note: even with the option
[x] Join multipart messages when retrieving headers
enabled, some messages may not be joined automagically; to allow for
that possibility, the "mark read" action is strongly recommended in
this context.
Also: the AGENT.INI file setting
RequireFilenameWithTag=
also affects automagic joining.