A new filter has been added for books. This can be enabled with the following config.yml setting:
bookfilter: true
A new file called "book.txt" in the rules directory will be applied to books. Both the title and the pages will be filtered.
then broadcast <string>|<<END
This action is the same as "then respond", but instead of sending the message to the player, it is broadcast to all players on the server.
then broadcastfile <filename.txt>
This is the same as "then respondfile", but instead of sending the message to the player, it sends it to all players on the server.
The "events chat,command, etc.." option in the rules.txt has been deprecated. If you have rules that you want to apply to multiple types of events, put them in their own file, eg:
common.txt:
match blah then replace
And then, in the event rules, use include...
chat.txt:
include common.txt
New action: then points <##>
New config: warning thresholds. drain rate
Idea:
Think of a bucket with holes in the bottom, and multiple lines on it:
\ / -- threshold3 \ / -- threshold2 \ / -- threshold1 - - - -- Leak rate: points / s
Given rules like this:
match fuck rule S1 Fuck then points 20 match asshole rule S2 Asshole then points 5
The following will happen:
A user will have 0 points by default. Every time they trip the filter, it will add the # of points (20 for 'fuck', 5 for 'asshole'). When they hit the threshold1 level, PwnFilter will execute the "ascending" commands at the threshold1 level. When they hit thresh2, same, thresh3, same. Every second or minute, depending on how configured, the configured leak rate number of points will be subtracted from the bucket. As the points balance crosses the threshold from above, it will execute the "descending" actions.
Thus, if a player swears once in a while, they will get no warning, no consequence. If they have a sailor's mouth, they might get a warning at threshold1 and 2, and a ban at threshold3. Once their points balance drops back below the ban threshold, they will be unbanned, and allowed back on the server.
A sample configuration for thresholds is below:
points:
enabled: true # 'false' disables the points-system
leak:
points: 1.0
interval: 30 # seconds
thresholds:
threshold1:
name: Warn
points: 10.0
actions:
ascending:
- respond You have hit our warning threshold for bad language!
descending:
- respond You are now back in our good books. Remember to keep it clean!
threshold2:
name: Kick
points: 20.0
actions:
ascending:
- kick You need to tone down your language!
- notify pwnfilter.admins %player% was kicked for hitting the kick threshold.
threshold3:
name: Tempban
points: 30.0
actions:
ascending:
- console ban %player% Your language is not acceptable. Take 15m to cool off!
- notify pwnfilter.admins %player% was given a temporary ban for bad language.
- 'notify console &4%player% was given a temporary ban for bad language. Points %points%'
descending:
- console pardon %player%
- notify pwnfilter.admins %player% points have fallen below Tempban threshold. %player% was unbanned
- notify console &4%player% points have fallen below Tempban threshold. %player% was unbanned
Enhance "then respond" action, by allowing a "here" tag function, to permit a multi-line response. Use "then respond <<END" to start the multiline message, and Terminate the response with a single line that has only 'END' (no quotes). Eg:
match ^/grue$ then respond <<END The grue is a sinister, lurking presence in the dark places of the earth. Its favorite diet is adventurers, but its insatiable appetite is tempered by its fear of light. No grue has ever been seen by the light of day, and few have survived its fearsome jaws to tell the tale. END
By default, commands will no longer be treated as "chat". They will not be decolored, nor will they be "spam filtered". There is a new config entry called cmdchat:. If commands are listed in cmdchat, those commands will be treated like chat events, and will be filtered by the chat.txt filter, instead of the command.txt filter. The reason for this change is that there are some commands (eg: /tell, /msg, /me) which most people feel should be filtered with the chat filter, and having to include chat filters in command.txt is cumbersome.
cmdchat: - me - nick - tell -whisper
Add then respondfile <filename.txt> which will be send to player. By default, text files for this command are stored in the PwnFilter/textfiles directory. This can be overridden with the config.yml textdir: config.
Please read these notes in their entirety. A lot have changes have been made since 3.1.x.
Your existing config may not work. At very least, please read the section about backward incompatible changes, and the new file structure. If you have questions, please join the #pwn9 channel on espernet and ask your question there. Please be patient. We're not always around.
*NOTE**
Any occurances of: &world ,&player, &string, &rawstring, &event, &ruleid, &ruledescr
will need to be replaced with: %world% ,%player%, %string%, %rawstring%, %event%, %ruleid%, %ruledescr%
You will get deprecation warnings if you use the old format, but it should still work for now.
ALSO...
A subtle, but important change has been made to the rules file format. If a blank line is detected, this will cause the parser to finish a rule. This used to be valid:
match blah then warn Hey! then deny
This is no longer valid, though, and the "then deny" will not ba attached to the rule.
Further, at least one blank line must separate all statement groups. eg:
VALID: match blah then action match foo then action NOT VALID: match blah then action match foo then action
Comments do not count as blank line. eg:
VALID: match blah #Now do an action. then action NOT VALID: match blah then action #Now another rule match foo then blah
Got it? :)
All of these changes (except the ones noted above) should be backwards compatible with the 3.1.x and lower versions.
By default, PwnFilter 3.2 will create a PwnFilter/rules directory, move your current rules.txt into it, and create one rules file for each handler, which links back to rules.txt. You do not need to keep all your rules in rules.txt. In fact, it is recommended that you create several rules files (in seperate subdirectories, if you prefer), and link them from each handler.
New folder structure:
plugins/PwnFilter
\->rules
|-> common --> tamewords.txt
| |-> badwords.txt
| |-> reallybadwords.txt
|-> sign.txt
|-> chat.txt
|-> item.txt
|-> command.txt
\-> console.txt
Each of the sign, chat, etc. are rulesets for specific event handlers. They can import from any of the files in the rules directory (or, in fact, any file that can be referred to relative to where it is, eg: common/tamewords.txt) and/or they can just have rules directly entered. Eg:
chat.txt:
include tamewords.txt include badwords.txt match derp then ...
and so on...
Adding a name / ID to a rule. eg:
match <matchstring> rule <id> [Optional description] ... etc...
Also, you can use &ruleid and &ruledescr in "then command" and "then console" messages. Eg:
match badword rule BW1 Badword Rule then console ban &player 1d (&ruleid) &ruledescr
would cause the following command to be run:
/ban PlayerName 1d (BW1) Badword Rule
Writing regex's can be tedious. Shortcuts allow the use of configurable "variables" that can are replaced in the regex. Eg:
match ((http)*(\w|\W|\d|_)*(www)*(\w|\W|\d|_)*[a-zA-Z0-9\.\-\*_\^\+\~\`\=\,\&*]{3,}(\W|\d|_|dot|\(dot\))+(com\b|org\b|net\b|edu\b|co\b|uk\b|de\b|cc\b|biz\b|mobi\b|xxx\b|tv\b))
could be replaced with:
shortcuts words.vars
match ((http)*<chr>*(www)*<chr>*<xta>{3,}<dot>+<dom>)
shortcuts
# ^ This will disable the shortcuts for future rules.
Internally, this would be expanded out to the regex above.
In a file called words.vars, you would specify:
chr (\w|\W|\d|_) dom (com\b|org\b|net\b|edu\b|co\b|uk\b|de\b|cc\b|biz\b|mobi\b|xxx\b|tv\b) dot (\W|\d|_|dot|\(dot\)) xta [a-zA-Z0-9\.\-\*_\^\+\~\`\=\,\&*]
You can surround up to 3 characters with <> and they will be replaced with whatever is defined in that varset.yml file.
Another example:
This file is called letters.vars:
_ (\W|\d|_) E [eu] K [ck] matchusing letters.var j+<_>*<E>+<_>*r+<_>*<K>+<_>*s*
If you want to match an actual less-than (<) or greater-than (>), use a backslash (\).
Allowed Characters in shortcut names: [_a-zA-z]
Sometimes, you want to have multiple rules that all do the same actions. An Action Group allows you to predefine a set of actions which you can then apply to a rule. Eg:
actiongroup swearactions then warn "Don't say that!" then fine 50 Pay $50 to the swear jar! .. later in the rules.txt .. match jerk then replace meanie then actions swearactions
Just as with action groups, condition groups let you specify common conditions you wish to apply to multiple rules. Eg:
conditiongroup ignoreAdmins ignore user Sage905 ignore user tremor77 ignore user DreamPhreak ignore user EpicATrain ... later in the rules.txt ... rule L3 Match jerk matchusing varset j+<_>*<E>+<_>*r+<_>*<K>+<_>*s* conditions ignoreAdmins then replace meanie then actions swearactions
An enhancement to the Regex which will automatically time-out if a Regex takes more than 500ms to execute. Upon triggering the timeout, PwnFilter will log an error showing the failed rule as well as the text that triggered the timeout. This should be a big help in troubleshooting runaway regexes.