Filter

New Major Version is out - 1.1.0

CHANGES:

• Rewritten to Rust (Performance boost)
• .find() does not return every element uniquely anymore ['foo', 'foo', 'bar'] is no longer compressed to ['foo', 'bar']
• Arrays of matches are now sets, hence not sorted chronologically
• Built in support for arrays removed.

A regex based adaptable profanity filter.

npm i regexcensor

This module uses Rust's Fancy Regex, read more:

• Rust Regex
• Fancy Regex

Importing the Filter

import Filter from 'regexcensor'

Predefined Filter Sets

If the input is a string, the filter will be set to a predefined swear preset.

const filter = Filter('PG13')

| Field | Description | | ------ | -------------------------------------------------------------- | | * | Include all possible negatively-associated words. | | PG13 | Less words filtered, minimum for a child-friendly environment. |

Configuring the Filter

If you want to configure a certain preset of of censored words, you can pass a configuring object like follows. All configurative attributes are optional.

const filter = Filter({
    fields: ['*'],
    mild: 3,
    patterns: [],
    words: [],
})

Fields

| Field | Description | | ------------ | ------------------------------------------------------------- | | * | Include all fields. | | sexual | General set of sexually-oriented swear words. | | sexuality | List of sexual orientations. | | racist | Set of racist swear words. | | historical | Historical or political ideas or terms, for example from WW2. | | religious | Religious terminilogy. |

Severity (Word mild)

Sets how severe the filter should be. 4 and 5 will censor almost all bad words and 0 and 1 only illegal ones.

Patterns

Define your own set of regex-defined patterns.

Words

RegexCensor provides an oversimplified regex construction.

For example, hello will automatically filter hellllllooooooo, he11o, but also h3110. End and/or start a word with * if the substring after or before doesn't matter. hello* will censor many hellos and hellowwwwwwssssss.

Methods

Filter.add(RegExp)

Adds the patterns to the filter, to quickly add regex arrays to the filter, use Filter.add_many(RegExp[])

Filter.add_word(string)

Adds a word to the filter, to quickly add arrays of words to the filter, use Filter.add_many_words(RegExp[])

Filter.check(string) : boolean

Checks one or several strings for profanity defined by the Filter. Returns true if any string triggers the patterns and false otherwise, to quickly check arrays with the filter, use Filter.check_many(string[]) : boolean

Filter.find(string) : string[]

Checks one or several strings for profanity defined by the Filter. Returns all emitted profane words.

Filter.find_patterns(string) : string[]

Checks one or several strings for profanity defined by the Filter. Returns all emitted regex pattern sources.

Filter.replace(string) : string

Replaces profanity in the string with asterisks.

Testing the Repository

• Run npm run test to test the code with mocha.
• Run npm run check to test different phrases in the console.
• Run npm run codegen to generate rust code files from latin.txt and universal.txt
• Run npm run build to build web assembly from rust code in /lib to /build
• Run npm run wat to build web assembly textfile from wasm in /build

Sources

Inspired from: https://www.cs.cmu.edu/~biglou/resources/bad-words.txt Inspired from: https://github.com/Minehut/MinehutBOT/blob/master/src/util/censorRules.ts

No content was directly cloned or copied, I took my time to rethink every regex. The only thing that are similar are the words. This statement does not apply to latin.txt which is copied from https://github.com/dundalek/latinize/blob/master/latinize.js