From 525fedc91bd02f53330cfb40fe228b148dcf562b Mon Sep 17 00:00:00 2001
From: Matt Mayer <152770+matthewmayer@users.noreply.github.com>
Date: Fri, 8 Nov 2024 17:57:53 +0700
Subject: [PATCH] fix(internet): ensure domainWord always returns a valid value
 in all locales (#3253)

---
 src/modules/internet/index.ts | 29 ++++++++++++++++++++++++++---
 test/modules/internet.spec.ts | 13 ++++++++++++-
 2 files changed, 38 insertions(+), 4 deletions(-)

diff --git a/src/modules/internet/index.ts b/src/modules/internet/index.ts
index 58c6e586aab..115bd21581d 100644
--- a/src/modules/internet/index.ts
+++ b/src/modules/internet/index.ts
@@ -595,9 +595,32 @@ export class InternetModule extends ModuleBase {
    * @since 2.0.1
    */
   domainWord(): string {
-    return this.faker.helpers
-      .slugify(`${this.faker.word.adjective()}-${this.faker.word.noun()}`)
-      .toLowerCase();
+    // Generate an ASCII "word" in the form `noun-adjective`
+    // For locales with non-ASCII characters, we fall back to lorem words, or a random string
+    const isValidSlug = (slug: string): boolean => {
+      return /^[a-z][a-z-]*[a-z]$/i.exec(slug) !== null;
+    };
+
+    const makeValidSlug = (word: string): string => {
+      const slug1 = this.faker.helpers.slugify(word);
+      if (isValidSlug(slug1)) {
+        return slug1;
+      }
+
+      const slug2 = this.faker.helpers.slugify(this.faker.lorem.word());
+      if (isValidSlug(slug2)) {
+        return slug2;
+      }
+
+      return this.faker.string.alpha({
+        casing: 'lower',
+        length: this.faker.number.int({ min: 4, max: 8 }),
+      });
+    };
+
+    const word1 = makeValidSlug(this.faker.word.adjective());
+    const word2 = makeValidSlug(this.faker.word.noun());
+    return `${word1}-${word2}`.toLowerCase();
   }
 
   /**
diff --git a/test/modules/internet.spec.ts b/test/modules/internet.spec.ts
index 84ded3d4480..cade8ddf102 100644
--- a/test/modules/internet.spec.ts
+++ b/test/modules/internet.spec.ts
@@ -1,6 +1,6 @@
 import validator from 'validator';
 import { describe, expect, it } from 'vitest';
-import { allFakers, faker } from '../../src';
+import { allFakers, faker, fakerKO } from '../../src';
 import { FakerError } from '../../src/errors/faker-error';
 import { IPv4Network } from '../../src/modules/internet';
 import { seededTests } from '../support/seeded-runs';
@@ -667,6 +667,17 @@ describe('internet', () => {
             validator.isFQDN(value, { require_tld: false })
           );
         });
+
+        it('should return a lower-case domain in non-ASCII locales', () => {
+          const domainWord = fakerKO.internet.domainWord();
+
+          expect(domainWord).toBeTruthy();
+          expect(domainWord).toBeTypeOf('string');
+          expect(domainWord).toSatisfy(validator.isSlug);
+          expect(domainWord).toSatisfy((value: string) =>
+            validator.isFQDN(value, { require_tld: false })
+          );
+        });
       });
 
       describe('ip()', () => {