[feature] Add support for dictionary selection, add a racist words list

.gitignore

@@ -3,3 +3,4 @@ node_modules
 npm-debug.log
 *.pem
 package-lock.json
+jsconfig.json

dictionaries/racist.json

@@ -0,0 +1,26 @@
+{
+  "en": [
+    "nigger",
+    "gook",
+    "ape",
+    "coon",
+    "jigaboo",
+    "negro",
+    "sambo",
+    "sooty",
+    "teapot",
+    "cracker",
+    "gringo",
+    "gweilo",
+    "gwailo",
+    "kwai lo",
+    "honky",
+    "redskin",
+    "white trash",
+    "slave",
+    "whitelist",
+    "white list",
+    "blacklist",
+    "black list"
+  ]
+}

index.js

@@ -67,7 +67,7 @@ module.exports = async robot => {
   }
 
   async function forRepository (context) {
-    let config = await getConfig(context, 'profanity.yml')
+    let config = await getConfig(context, 'profanity.yml', null)
 
     if (!config) {
       scheduler.stop(context.payload.repository)

lib/profanity.js

@@ -1,4 +1,3 @@
-const blacklist = require('../config/blacklist.json')
 const Filter = require('./filter')
 const schema = require('./schema')
 
@@ -27,11 +26,12 @@ module.exports = class Profanity {
   getFilter () {
     const {
       language,
+      dictionaries,
       extraWords,
       exemptWords,
       placeholder
     } = this.config
-    const list = blacklist[language]
+    const list = this.getForbiddenWordList(dictionaries, language)
 
     return new Filter({
       list: list,
@@ -41,6 +41,27 @@ module.exports = class Profanity {
     })
   }
 
+  /**
+   * Get the flattened forbidden words list for the specified dictionaries and language
+   *
+   * @param {string[]} dictionaries The list of dictionaries to use
+   * @param {string} language The language code to use
+   * @returns {string[]} The flattened list of forbidden words
+   */
+  getForbiddenWordList (dictionaries, language) {
+    if (!dictionaries || !dictionaries.length || !language) {
+      return []
+    }
+
+    return Array.from(new Set(dictionaries.reduce((acc, dictName) => {
+      const dictionary = require('../dictionaries/' + dictName)
+      if (!dictionary || !dictionary[language]) {
+        return acc
+      }
+      return acc.concat(dictionary[language])
+    }, [])))
+  }
+
   async markAndSweep (type) {
     const {only} = this.config
     if (only && only !== type) {

lib/schema.js

@@ -7,6 +7,9 @@ const fields = {
   censor: Joi.boolean()
     .description('Set to true to censor issues (defaults to false)'),
 
+  dictionaries: Joi.array().items(Joi.string().valid('profanity', 'racist')).min(1).single()
+    .description('The dictionaries of forbidden words to use'),
+
   placeholder: Joi.string().max(1)
     .description('A letter to replace those of a forbidden word'),
 
@@ -52,6 +55,7 @@ const fields = {
 const schema = Joi.object().keys({
   language: fields.language.default('en'),
   censor: fields.censor.default(false),
+  dictionaries: fields.dictionaries.default(['profanity', 'racist']),
   placeholder: fields.placeholder.default('*'),
   extraWords: fields.extraWords.default([]),
   exemptWords: fields.exemptWords.default([]),

test/schema.test.js

@@ -4,6 +4,8 @@ const validConfigs = [
   [{language: 'en'}],
   [{censor: true}],
   [{censor: false}],
+  [{dictionaries: ['profanity']}],
+  [{dictionaries: ['racist', 'profanity']}],
   [{placeholder: '*'}],
   [{extraWords: ['duck']}],
   [{extraWords: 'duck'}, {extraWords: ['duck']}],
@@ -44,6 +46,8 @@ const validConfigs = [
 const invalidConfigs = [
   [{language: 'bananas'}, 'must be one of [de, en, es, fr, it, nl, pt, ru]'],
   [{censor: 'nope'}, 'must be a boolean'],
+  [{dictionaries: ['fake']}, 'must be one of [profanity, racist]'],
+  [{dictionaries: []}, 'must contain at least 1 items'],
   [{placeholder: ''}, 'not allowed to be empty'],
   [{placeholder: false}, 'must be a string'],
   [{placeholder: ['a', 'b']}, 'must be a string'],
@@ -72,6 +76,7 @@ describe('schema', () => {
     expect(schema.validate({}).value).toEqual({
       language: 'en',
       censor: false,
+      dictionaries: ['profanity', 'racist'],
       placeholder: '*',
       extraWords: [],
       exemptWords: [],