Filter Rules and HowTo Make them

The filter tries to find the "match" string then use the "From" and "To" as a substitution. You could do both operations simply with the substation this allows finer control on which strings you do the substitution. The match, from, and to are all perl regex expressions. Rules are evaluated in descending order. The top rule is first evaluated, then the second from the top, etc. All strings are evaluated as:

$<name> is expanded to the environment variable <name>. This is done immediately when you insert the rule, and will not change after that time. Therefor it's probably of limited value.

$$<name> is substituted with the perl $<name> variable during the match. This can be substrings such as $1, $2 in the substitution, or normal variables available under sirc (such as $nick, $channel, etc).

~<name>~ PREPENDED ONCE AND ONLY ONCE to the line will send the line of text to the window called <name>. If the window does not exist it will go to the last window which had focus. There are several special windows, all prefixed by a single "!":

!default

the current default window. Guaranteed to be.

!all

Send to every window. This might not show up on all windows, depending on how the parse the text. For example, channel windows won't show a part/quit unless the nick is on the channel.

!discard

discards the text.

Examples:

Want to convert all boren from boren to BoreN

Match: .*
From: boren
To: BoreN

Pretty straight forward, match anything, then substitute boren with BoreN.

You want to match everything with boren in it and send to the window called "boren"

Match: boren
From: ^
To: ~boren~

Looks for "boren" if found, substitutes the beginning of the string (^) with ~boren~.

Though #2 works, if the string already has ~somewindow~ on it, you'll now have two ~boren~~somewindow~... So you can do this instead.

Match: boren
From: ^(?:~\S+~)
To: ~boren~

Ok, the from line is a little bit more complicated. It says match 0 or 1 copies of "~\S+~". Which is 1 tilde, one or more none whitespaces, and then another tilde. The paranoid might do (*:~\S+~) which says match 0 or more channel directives in case prior rules are broken.

Server kill messages then to be long, ugly, annoying, etc. Basic message on dalnet looks like:

*** Notice -- Received KILL message for BOBO!ANDY@line82-basel.datacomm.ch 
from NickServ Path: empire.ny.us.dal.net[209.51.168.14]!trapdoor.ca.us.dal
.net[206.86.127.252]!caris.ca.us.dal.net[208.1.222.221]!services.dal.net[2
08.1.222.222]!services.dal.net (NickServ Enforcement)

When you're +s you get a tons of them, you don't want all of them flying across your screen. I'm going to make 3 rules to deal with them one bit at a time. You could do it in less rules, but it'll show you the basic rule structure, in nice steps, and how to use multiple rules to parse a message. The first step is to remove the Path: portion of the message, and will be example 4.

Match: ^\*\*\*.* KILL message for.*
From: Path: \S+
To: .

Match looks for the message starting with ***, the *'s have to be quoted with \ since by themselves they mean 0 or more of the prior character. .* the means match anything until you find " KILL message for". This allows us to avoid tying in "-- Reicevied..." etc. The trailing .* says match anything to the end of the line. (not needed I think)

The From like says substitute space Path: space and any non whitespace characters with the To. To is a "." therefor the entire path turns into a single period.

The message now looks like:

*** Notice -- Received KILL message for BOBO!ANDY@line82-basel.datacomm.ch
from NickServ. (NickServ Enforcement)

Notice the new "." after NickServ?

Ok, the message is a lot cleaner, but KILL's from nickserv aren't really that important, so let's forward them to the !discard window.

Match: ^\*\*\*.*KILL message.*\(NickServ Enforcement\)
From: ^(?:~\S+~)
To: ~!discard~

Match rule searches for the KILL message and makes sure it's from NickServ. Notice the \( and \) both () and used in regex, therefor we have to quote them. This is very similar to example 3.

We've now filtered out all the nickserv kills, but the message is still pretty hard to read by simply glancing at it. So let's reorder it to something like:

*** [KILL] <KILLER> killed <KILLED> (reason)

Match: \*\*\*.*KILL message
From: \*\*\*.*for (.*?) from (.*?)\. \((.*?)\).*
To: *** [KILL] $$2 killed $$1 ($$3)

Ok, the match looks for ***<something> KILL message. We can't use ^ since we may have just appended ~<window>~.

The from line get's little more interesting. The "for (.*?) " looks for the word for then some text. .*? says match zero or more of anything except newline, but isn't greedy. Stop when the first terminating condition is found, not the last. In other words it matches anything until a space. The surrounding () says to save the contents. Each () saves the matched data in $# where # starts at 1 for the first substring, etc. In this case, $1 gets the nick/user-info of the person killed. $2 is then filled with the name of the killer. Between the () we have the reason for the kill. Here the ( and \( get a little confusing. Remember \( matchin the actual character '('.

How to colourize your life. Ok, you want to add some colour to ksirc. See README.colours for colour info, but here's a filter ruls to highlight the nick between <NICK> on each line:

Match: ^(?:~\S+~)<\S+>
From: <(\S+)>
To: <~4$$1~c>

Takes the nick and adds colour #4 between the two <> ~c clears the colour.

---

Andrew Stanley-Jones
Last modified: Mon Jan 12 21:07:55 MST 1998