michelebertoli / banbuilder

Source lists for swear words for web applications, forums, etc.

Banbuilder

Banbuilder is a PHP function and bad word database for profanity filtering. The PHP script uses regex to intelligently look for "leetspeak"-style numeric or symbol replacements.

The databases of profanity are located in the lang/*.wordlist-regex.php files, and are in a simple PHP array. So far we have translations for:

• US English
• British English
• Spain Spanish
• France French
• Korean (South)
• Netherlands Dutch
• Norwegian (Bokmål & various dialects)

We are actively looking for translation files!

Usage

Simply require the database file and the function file, and invoke the function:

 $badwords = array();
 include('lang/en-us.wordlist-regex.php');
 include('censor.function.php');
 $censored = censorString($input, $badwords);

To use multiple language dictionaries, just include them.

 $badwords = array();
 include('lang/en-us.wordlist-regex.php');
 include('lang/fr.wordlist-regex.php');
 include('lang/es.wordlist-regex.php');
 include('censor.function.php');
 $censored = censorString($input, $badwords);

Make sure you set the $badwords = array(); before including the dictionaries.

You end up with an array called $censored. You can access the original, uncensored string as $censored['orig'] and the newly censored string as $censored['clean'].

There is an optional string parameter that you can pass to use a different replacement character. For example:

 $censored = censorString($input, $badwords,'X');

Will replace "bitch" to "XXXXX" instead of the default "*****".

If the optional parameter is more than 1 character long, a bad words will be replaced by random strings which are composed by characters taken from the string. For example:

 $censored = censorString($input, $badwords,'X!°#%$@');

Could replace "bitch" to something like "%#@#@" or "%@!X#".

Summary

In a nutshell, this code takes your array of bad words and compares it to an array of common filter-evasion tactics. It then does a string replacement to insert regex parameters into your badwords array, and then evaluates your input string to that expanded banned word list.

So in your bad words array, you might have:

 [0] => 'ass'

The preg_replace functions replace all of the possible shenaningan letters with regex patterns (in lieu of adding the variants onto the end of the array), so the 'ass' in your array gets turned into this, right before the preg_replace checks for matches:

 [0] => /(a|a\.|a\-|4|@|Á|á|À|Â|à|Â|â|Ä|ä|Ã|ã|Å|å|α)(s|s\.|s\-|5|\$|§)(s|s\.|s\-|5|\$|§)/i

This means that a word can have none, one or any variety of leet replacements and it will still trip the trigger. Part of the leet filter includes stripping out letter-dash and letter-dots.

This means that the following all evaluate to the "bitch":

• B1tch
• bi7tch
• b.i.t.c.h.
• b-i-t-c-h
• b.1.t.c.h.
• ßitch
• and so on....

Legacy Database

When this project was first started, it was used to compile a database of swear words in every permutation for scenarios where regex wasn't possible for whatever reason). While the regex method is much better, the legacy full (non-regexy) databases contain over 800 words ready to use in banned words lists for projects. You can find those badword lists in the word-dbs directory in the repo.

These files are extra and are not required for the PHP function to run.

Current file types are:

• SQL
• CSV
• LaTex
• YML
• XML
• PHP Array