ejabberd-contrib/mod_pottymouth
Badlop a95e657a21 Add simple mod_doc/0 to prevent warnings; they won't be used anyway 2021-07-06 21:02:29 +02:00
..
conf filter all body elemements in message 2016-09-13 17:13:32 -04:00
src Add simple mod_doc/0 to prevent warnings; they won't be used anyway 2021-07-06 21:02:29 +02:00
COPYING adding mod_pottymouth 2016-06-24 15:34:37 +00:00
README.txt Use banword instead of bloom + etbloom + bitarray (thanks to ytkang) 2021-04-16 16:37:11 +02:00
mod_pottymouth.spec code cleanup. add optional normalize_leet to do dynamic substitution of one-to-one mappings (reduce blacklist size). 2016-07-19 21:34:04 +00:00

README.txt

The 'mod_pottymouth' ejabberd module aims to fill the void left by 'mod_shit'
which has disappeared from the net. It allows individual whole words of a
message to be filtered against a blacklist. It allows multiple blacklists
sharded by language. The internal bloomfilter can support arbitrary blacklist
sizes. Using a large list (say, 87M terms) will slow down the initial server
boot time (to about 15 minutes respectively), but once loaded lookups are very
speedy.

Configuration file is ~/.ejabberd-modules/mod_pottymouth/conf/mod_pottymouth.yml

modules:
    mod_pottymouth:
        blacklists:
            default: /home/your_user/blacklist_en.txt
            en: /home/your_user/blacklist_en.txt
            cn: /home/your_user/blacklist_cn.txt
            fr: /home/your_user/blacklist_fr.txt
        charmaps:
            default: /etc/ejabberd/modules/mod_pottymouth/charmap_en.txt
            en: /etc/ejabberd/modules/mod_pottymouth/charmap_en.txt

For each language (en,cn,fr,...whatever) provide a full path to a backlist file.
The blacklist file is a plain text file with blacklisted words listed one per
line.

You can also provide an optional 'charmap' for each language. This allows you
to specify simple substitutions that will be made on the fly so you don't need
to include those permutations in the blacklist. This keeps the blacklist small
and reduces server startup time. For example, if you included the word:
'xyza' in the blacklist, adding the following substitutions in the charmap
would filter permutations such as 'XYZA', 'xYz4', or 'Xyz@' automatically.

charmap format:

[
 {"X", "x"},
 {"Y", "y"},
 {"Z", "z"},
 {"@", "a"},
 {"4", "a"}
].

Gotchas:

The language will be looked up by whatever value is passed in the xml:lang
attribute of the xml message. So, any xml:lang value to be supported will need
a corresponding entry/blacklist in the config file. If xml:lang is missing,
the 'default' entry in config will be used.

For xml:lang attribute docs, see:
http://wiki.xmpp.org/web/Programming_XMPP_Clients#Sending_a_message

Blacklist helper

Thinking of a bunch of swear words and all the permutations can be tough. We made
a helper script to take a bare wordlist and generate permutations given a
dictionary of substitution characters: https://github.com/madglory/permute_wordlist