bugfix: Regex removing all special characters (#9)

Closes #5

Author: ftonato

__tests__/utils.test.js

@@ -10,7 +10,7 @@ describe('Utils class', () => {
     expect(utils.findForbiddenWords(post)).toMatchObject({post, error: null});
   });
 
-  test.skip('it should return a list of used words filtered by valid characters', () => {
+  test('it should return a list of used words filtered by valid characters', () => {
     const post = `Hello, do you know some forbidden words?
       Take a look in our list: (ad-man), black-list, whitelist, {---}, hello_world, FIANCÉE, bOy.!
       [BYe]

src/suggestions.js

@@ -1,7 +1,7 @@
 const roles = {
   man_woman: [
     {
-      words: ['ad-man', 'ad-woman'],
+      words: ['ad-man', 'ad-woman', 'adman', 'adwoman'],
       suggestion: 'ad-person',
       others: 'advertising executive, promoter',
     },
@@ -529,7 +529,7 @@ const roles = {
   ],
   technology: [
     {
-      words: ['white-list', 'black-list'],
+      words: ['white-list', 'black-list', 'whitelist', 'blacklist'],
       suggestion: 'accept/reject list',
       others: 'include/exclude, allow/disallow',
     },

src/utils.js

@@ -25,9 +25,16 @@ export default class Utils {
   }
 
   static getWords(post) {
-    const regex = /[^a-z0-9]/gi;
+    /**
+     * \w Word = Matches any word character (alphanumeric & underscore)
+     * \- Escaped character = Matches a "-" character (char code 45)
+     * À-ú Range = Matches a character in the range "À" to "ú"
+     *             (char code 192 to 250). Case sensitive
+     */
+    const regex = /[^\w\-À-ú]+/gi;
+
     const words = post.split(regex);
-    return words.filter(word => word.toLowerCase()).filter(word => word.length > 0);
+    return words.filter(word => word.toLowerCase()).filter(word => word.length > 1);
   }
 
   findForbiddenWords(post) {