fix(internet): userName, email and slugify return only ascii (#1554)

3 files changed, 178 insertions, 30 deletions

diff --git a/test/__snapshots__/internet.spec.ts.snap b/test/__snapshots__/internet.spec.ts.snap
index 4b87bff9..5b33c9e2 100644
--- a/test/__snapshots__/internet.spec.ts.snap
+++ b/test/__snapshots__/internet.spec.ts.snap
@@ -6,6 +6,24 @@ exports[`internet > 42 > color > noArgs 1`] = `"#30667a"`;
 
 exports[`internet > 42 > color > with color base 1`] = `"#6298ac"`;
 
+exports[`internet > 42 > displayName > noArgs 1`] = `"Garnet.Schinner73"`;
+
+exports[`internet > 42 > displayName > noArgs 2`] = `"Garnet.Schinner73"`;
+
+exports[`internet > 42 > displayName > noArgs 3`] = `"Garnet.Schinner73"`;
+
+exports[`internet > 42 > displayName > noArgs 4`] = `"Garnet.Schinner73"`;
+
+exports[`internet > 42 > displayName > with Chinese names 1`] = `"大羽_陳"`;
+
+exports[`internet > 42 > displayName > with Cyrillic names 1`] = `"Фёдор_Достоевский"`;
+
+exports[`internet > 42 > displayName > with Latin names 1`] = `"Jane_Doe"`;
+
+exports[`internet > 42 > displayName > with accented names 1`] = `"Hélene_Müller"`;
+
+exports[`internet > 42 > displayName > with names 1`] = `"Jane_Doe"`;
+
 exports[`internet > 42 > domainName 1`] = `"hasty-shin.org"`;
 
 exports[`internet > 42 > domainSuffix 1`] = `"info"`;
@@ -62,6 +80,20 @@ exports[`internet > 42 > userAgent 1`] = `"Mozilla/5.0 (Windows; U; Windows NT 6
 
 exports[`internet > 42 > userName > noArgs 1`] = `"Garnet.Schinner73"`;
 
+exports[`internet > 42 > userName > noArgs 2`] = `"Garnet.Schinner73"`;
+
+exports[`internet > 42 > userName > noArgs 3`] = `"Garnet.Schinner73"`;
+
+exports[`internet > 42 > userName > noArgs 4`] = `"Garnet.Schinner73"`;
+
+exports[`internet > 42 > userName > with Chinese names 1`] = `"hlzp8d_tpv"`;
+
+exports[`internet > 42 > userName > with Cyrillic names 1`] = `"Fedor_Dostoevskii"`;
+
+exports[`internet > 42 > userName > with Latin names 1`] = `"Jane_Doe"`;
+
+exports[`internet > 42 > userName > with accented names 1`] = `"Helene_Muller"`;
+
 exports[`internet > 42 > userName > with names 1`] = `"Jane_Doe"`;
 
 exports[`internet > 1211 > avatar 1`] = `"https://cloudflare-ipfs.com/ipfs/Qmd3W5DuhgHirLHGVixi6V76LhCkZUz6pnFt5AJBiyvHye/avatar/1160.jpg"`;
@@ -70,6 +102,24 @@ exports[`internet > 1211 > color > noArgs 1`] = `"#773a72"`;
 
 exports[`internet > 1211 > color > with color base 1`] = `"#a96ca4"`;
 
+exports[`internet > 1211 > displayName > noArgs 1`] = `"Tito_Koelpin22"`;
+
+exports[`internet > 1211 > displayName > noArgs 2`] = `"Tito_Koelpin22"`;
+
+exports[`internet > 1211 > displayName > noArgs 3`] = `"Tito_Koelpin22"`;
+
+exports[`internet > 1211 > displayName > noArgs 4`] = `"Tito_Koelpin22"`;
+
+exports[`internet > 1211 > displayName > with Chinese names 1`] = `"大羽.陳89"`;
+
+exports[`internet > 1211 > displayName > with Cyrillic names 1`] = `"Фёдор.Достоевский89"`;
+
+exports[`internet > 1211 > displayName > with Latin names 1`] = `"Jane.Doe89"`;
+
+exports[`internet > 1211 > displayName > with accented names 1`] = `"Hélene.Müller89"`;
+
+exports[`internet > 1211 > displayName > with names 1`] = `"Jane.Doe89"`;
+
 exports[`internet > 1211 > domainName 1`] = `"vibrant-infix.org"`;
 
 exports[`internet > 1211 > domainSuffix 1`] = `"org"`;
@@ -126,6 +176,20 @@ exports[`internet > 1211 > userAgent 1`] = `"Mozilla/5.0 (Macintosh; U; Intel Ma
 
 exports[`internet > 1211 > userName > noArgs 1`] = `"Tito_Koelpin22"`;
 
+exports[`internet > 1211 > userName > noArgs 2`] = `"Tito_Koelpin22"`;
+
+exports[`internet > 1211 > userName > noArgs 3`] = `"Tito_Koelpin22"`;
+
+exports[`internet > 1211 > userName > noArgs 4`] = `"Tito_Koelpin22"`;
+
+exports[`internet > 1211 > userName > with Chinese names 1`] = `"hlzp8d.tpv89"`;
+
+exports[`internet > 1211 > userName > with Cyrillic names 1`] = `"Fedor.Dostoevskii89"`;
+
+exports[`internet > 1211 > userName > with Latin names 1`] = `"Jane.Doe89"`;
+
+exports[`internet > 1211 > userName > with accented names 1`] = `"Helene.Muller89"`;
+
 exports[`internet > 1211 > userName > with names 1`] = `"Jane.Doe89"`;
 
 exports[`internet > 1337 > avatar 1`] = `"https://cloudflare-ipfs.com/ipfs/Qmd3W5DuhgHirLHGVixi6V76LhCkZUz6pnFt5AJBiyvHye/avatar/327.jpg"`;
@@ -134,6 +198,24 @@ exports[`internet > 1337 > color > noArgs 1`] = `"#214814"`;
 
 exports[`internet > 1337 > color > with color base 1`] = `"#537a46"`;
 
+exports[`internet > 1337 > displayName > noArgs 1`] = `"Devyn21"`;
+
+exports[`internet > 1337 > displayName > noArgs 2`] = `"Devyn21"`;
+
+exports[`internet > 1337 > displayName > noArgs 3`] = `"Devyn21"`;
+
+exports[`internet > 1337 > displayName > noArgs 4`] = `"Devyn21"`;
+
+exports[`internet > 1337 > displayName > with Chinese names 1`] = `"大羽56"`;
+
+exports[`internet > 1337 > displayName > with Cyrillic names 1`] = `"Фёдор56"`;
+
+exports[`internet > 1337 > displayName > with Latin names 1`] = `"Jane56"`;
+
+exports[`internet > 1337 > displayName > with accented names 1`] = `"Hélene56"`;
+
+exports[`internet > 1337 > displayName > with names 1`] = `"Jane56"`;
+
 exports[`internet > 1337 > domainName 1`] = `"fair-migration.com"`;
 
 exports[`internet > 1337 > domainSuffix 1`] = `"biz"`;
@@ -190,4 +272,18 @@ exports[`internet > 1337 > userAgent 1`] = `"Mozilla/5.0 (Windows; U; Windows NT
 
 exports[`internet > 1337 > userName > noArgs 1`] = `"Devyn21"`;
 
+exports[`internet > 1337 > userName > noArgs 2`] = `"Devyn21"`;
+
+exports[`internet > 1337 > userName > noArgs 3`] = `"Devyn21"`;
+
+exports[`internet > 1337 > userName > noArgs 4`] = `"Devyn21"`;
+
+exports[`internet > 1337 > userName > with Chinese names 1`] = `"hlzp8d56"`;
+
+exports[`internet > 1337 > userName > with Cyrillic names 1`] = `"Fedor56"`;
+
+exports[`internet > 1337 > userName > with Latin names 1`] = `"Jane56"`;
+
+exports[`internet > 1337 > userName > with accented names 1`] = `"Helene56"`;
+
 exports[`internet > 1337 > userName > with names 1`] = `"Jane56"`;
diff --git a/test/helpers.spec.ts b/test/helpers.spec.ts
index c56d2589..530c3b49 100644
--- a/test/helpers.spec.ts
+++ b/test/helpers.spec.ts
@@ -180,9 +180,19 @@ describe('helpers', () => {
       });
 
       describe('slugify()', () => {
-        it('removes unwanted characters from URI string', () => {
-          expect(faker.helpers.slugify('Aiden.Harªann')).toBe('Aiden.Harann');
+        it('removes non-word characters from strings except . and -', () => {
+          expect(faker.helpers.slugify('foo bar')).toBe('foo-bar');
+          expect(faker.helpers.slugify('Faker is cool')).toBe('Faker-is-cool');
+          expect(faker.helpers.slugify('super*star')).toBe('superstar');
           expect(faker.helpers.slugify("d'angelo.net")).toBe('dangelo.net');
+          expect(faker.helpers.slugify('hello你好')).toBe('hello');
+        });
+
+        it('strips simple diacritics from strings', () => {
+          expect(faker.helpers.slugify('Aiden.Harªann')).toBe('Aiden.Haraann');
+          expect(faker.helpers.slugify('Adèle.Argüello')).toBe(
+            'Adele.Arguello'
+          );
         });
       });
 
diff --git a/test/internet.spec.ts b/test/internet.spec.ts
index 02f0edaa..45ed1963 100644
--- a/test/internet.spec.ts
+++ b/test/internet.spec.ts
@@ -38,7 +38,17 @@ describe('internet', () => {
     });
 
     t.describe('userName', (t) => {
-      t.it('noArgs').it('with names', 'Jane', 'Doe');
+      t.it('noArgs').it('with Latin names', 'Jane', 'Doe');
+      t.it('noArgs').it('with accented names', 'Hélene', 'Müller');
+      t.it('noArgs').it('with Cyrillic names', 'Фёдор', 'Достоевский');
+      t.it('noArgs').it('with Chinese names', '大羽', '陳');
+    });
+
+    t.describe('displayName', (t) => {
+      t.it('noArgs').it('with Latin names', 'Jane', 'Doe');
+      t.it('noArgs').it('with accented names', 'Hélene', 'Müller');
+      t.it('noArgs').it('with Cyrillic names', 'Фёдор', 'Достоевский');
+      t.it('noArgs').it('with Chinese names', '大羽', '陳');
     });
 
     t.describe('password', (t) => {
@@ -98,6 +108,20 @@ describe('internet', () => {
           expect(faker.definitions.internet.free_email).toContain(suffix);
         });
 
+        it('should return a valid email in every locale', () => {
+          for (const locale of Object.keys(faker.locales)) {
+            faker.setLocale(locale);
+            const email = faker.internet.email();
+
+            expect(email).toBeTruthy();
+            expect(email).toBeTypeOf('string');
+            expect(email).toSatisfy(
+              validator.isEmail,
+              `locale: ${locale} has invalid email: ${email}`
+            );
+          }
+        });
+
         it('should return an email with given firstName', () => {
           const email = faker.internet.email('Aiden.Harann55');
 
@@ -130,19 +154,6 @@ describe('internet', () => {
           expect(faker.definitions.internet.free_email).toContain(suffix);
         });
 
-        it('should return an email with japanese characters', () => {
-          const email = faker.internet.email('思源_唐3');
-
-          expect(email).toBeTruthy();
-          expect(email).toBeTypeOf('string');
-          expect(email).toSatisfy(validator.isEmail);
-
-          const [prefix, suffix] = email.split('@');
-
-          expect(prefix).toMatch(/^思源_唐3/);
-          expect(faker.definitions.internet.free_email).toContain(suffix);
-        });
-
         it('should return an email with special characters', () => {
           const email = faker.internet.email('Mike', 'Smith', null, {
             allowSpecialCharacters: true,
@@ -203,20 +214,6 @@ describe('internet', () => {
           expect(prefix).toMatch(/^Aiden([._]Harann)?\d*/);
         });
 
-        it('should return an email with the example suffix and japanese characters', () => {
-          const email = faker.internet.exampleEmail('思源_唐3');
-
-          expect(email).toBeTruthy();
-          expect(email).toBeTypeOf('string');
-          expect(email).toSatisfy(validator.isEmail);
-
-          const [prefix, suffix] = email.split('@');
-
-          expect(suffix).toMatch(/^example\.(com|net|org)$/);
-          expect(faker.definitions.internet.example_email).toContain(suffix);
-          expect(prefix).toMatch(/^思源_唐3/);
-        });
-
         it('should return an email with special characters', () => {
           const email = faker.internet.exampleEmail('Mike', 'Smith', {
             allowSpecialCharacters: true,
@@ -262,6 +259,51 @@ describe('internet', () => {
             /^Aiden((\d{1,2})|([._]Harann\d{1,2})|([._](Harann)))/
           );
         });
+
+        it('should strip accents', () => {
+          const username = faker.internet.userName('Adèle', 'Smith');
+          expect(username).includes('Adele');
+        });
+
+        it('should transliterate Cyrillic', () => {
+          const username = faker.internet.userName('Амос', 'Васильев');
+          expect(username).includes('Amos');
+        });
+
+        it('should provide a fallback for Chinese etc', () => {
+          const username = faker.internet.userName('大羽', '陳');
+          expect(username).includes('hlzp8d');
+        });
+      });
+
+      describe('displayName()', () => {
+        it('should return a random display name', () => {
+          const displayName = faker.internet.displayName();
+
+          expect(displayName).toBeTruthy();
+          expect(displayName).toBeTypeOf('string');
+          expect(displayName).toMatch(/\w/);
+        });
+
+        it('should return a random display name with given firstName', () => {
+          const displayName = faker.internet.displayName('Aiden');
+
+          expect(displayName).toBeTruthy();
+          expect(displayName).toBeTypeOf('string');
+          expect(displayName).toMatch(/\w/);
+          expect(displayName).includes('Aiden');
+        });
+
+        it('should return a random display name with given firstName and lastName', () => {
+          const displayName = faker.internet.displayName('Aiden', 'Harann');
+
+          expect(displayName).toBeTruthy();
+          expect(displayName).toBeTypeOf('string');
+          expect(displayName).includes('Aiden');
+          expect(displayName).toMatch(
+            /^Aiden((\d{1,2})|([._]Harann\d{1,2})|([._](Harann)))/
+          );
+        });
       });
 
       describe('protocol()', () => {