test: more comprehensive IDNA test cases

- Split the tests out to a separate file
- Add invalid cases
- Add tests for url.domainTo*()
- Re-enable previously broken WPT URL parsing tests

PR-URL: https://github.com/nodejs/node/pull/11549
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>

4 files changed, 311 insertions, 89 deletions

diff --git a/test/fixtures/url-idna.js b/test/fixtures/url-idna.js
new file mode 100644
index 00000000000..af169bb04a8
--- /dev/null
+++ b/test/fixtures/url-idna.js
@@ -0,0 +1,217 @@
+'use strict';
+
+// Credit for list: http://www.i18nguy.com/markup/idna-examples.html
+module.exports = {
+  valid: [
+    { ascii: 'xn--mgbaal8b0b9b2b.icom.museum',
+      unicode: 'افغانستا.icom.museum'
+    },
+    {
+      ascii: 'xn--lgbbat1ad8j.icom.museum',
+      unicode: 'الجزائر.icom.museum'
+    },
+    {
+      ascii: 'xn--sterreich-z7a.icom.museum',
+      unicode: 'österreich.icom.museum'
+    },
+    {
+      ascii: 'xn--54b6eqazv8bc7e.icom.museum',
+      unicode: 'বাংলাদেশ.icom.museum'
+    },
+    {
+      ascii: 'xn--80abmy0agn7e.icom.museum',
+      unicode: 'беларусь.icom.museum'
+    },
+    {
+      ascii: 'xn--belgi-rsa.icom.museum',
+      unicode: 'belgië.icom.museum'
+    },
+    {
+      ascii: 'xn--80abgvm6a7d2b.icom.museum',
+      unicode: 'българия.icom.museum'
+    },
+    {
+      ascii: 'xn--mgbfqim.icom.museum',
+      unicode: 'تشادر.icom.museum'
+    },
+    {
+      ascii: 'xn--fiqs8s.icom.museum',
+      unicode: '中国.icom.museum'
+    },
+    {
+      ascii: 'xn--mgbu4chg.icom.museum',
+      unicode: 'القمر.icom.museum'
+    },
+    {
+      ascii: 'xn--vxakcego.icom.museum',
+      unicode: 'κυπρος.icom.museum'
+    },
+    {
+      ascii: 'xn--eskrepublika-ebb62d.icom.museum',
+      unicode: 'českárepublika.icom.museum'
+    },
+    {
+      ascii: 'xn--wgbh1c.icom.museum',
+      unicode: 'مصر.icom.museum'
+    },
+    {
+      ascii: 'xn--hxakic4aa.icom.museum',
+      unicode: 'ελλάδα.icom.museum'
+    },
+    {
+      ascii: 'xn--magyarorszg-t7a.icom.museum',
+      unicode: 'magyarország.icom.museum'
+    },
+    {
+      ascii: 'xn--sland-ysa.icom.museum',
+      unicode: 'ísland.icom.museum'
+    },
+    {
+      ascii: 'xn--h2brj9c.icom.museum',
+      unicode: 'भारत.icom.museum'
+    },
+    {
+      ascii: 'xn--mgba3a4fra.icom.museum',
+      unicode: 'ايران.icom.museum'
+    },
+    {
+      ascii: 'xn--ire-9la.icom.museum',
+      unicode: 'éire.icom.museum'
+    },
+    {
+      ascii: 'xn--4dbklr2c8d.xn--4dbrk0ce.museum',
+      unicode: 'איקו״ם.ישראל.museum'
+    },
+    {
+      ascii: 'xn--wgv71a.icom.museum',
+      unicode: '日本.icom.museum'
+    },
+    {
+      ascii: 'xn--igbhzh7gpa.icom.museum',
+      unicode: 'الأردن.icom.museum'
+    },
+    {
+      ascii: 'xn--80aaa0a6awh12ed.icom.museum',
+      unicode: 'қазақстан.icom.museum'
+    },
+    {
+      ascii: 'xn--3e0b707e.icom.museum',
+      unicode: '한국.icom.museum'
+    },
+    {
+      ascii: 'xn--80afmksoji0fc.icom.museum',
+      unicode: 'кыргызстан.icom.museum'
+    },
+    {
+      ascii: 'xn--q7ce6a.icom.museum',
+      unicode: 'ລາວ.icom.museum'
+    },
+    {
+      ascii: 'xn--mgbb7fjb.icom.museum',
+      unicode: 'لبنان.icom.museum'
+    },
+    {
+      ascii: 'xn--80aaldqjmmi6x.icom.museum',
+      unicode: 'македонија.icom.museum'
+    },
+    {
+      ascii: 'xn--mgbah1a3hjkrd.icom.museum',
+      unicode: 'موريتانيا.icom.museum'
+    },
+    {
+      ascii: 'xn--mxico-bsa.icom.museum',
+      unicode: 'méxico.icom.museum'
+    },
+    {
+      ascii: 'xn--c1aqabffc0aq.icom.museum',
+      unicode: 'монголулс.icom.museum'
+    },
+    {
+      ascii: 'xn--mgbc0a9azcg.icom.museum',
+      unicode: 'المغرب.icom.museum'
+    },
+    {
+      ascii: 'xn--l2bey1c2b.icom.museum',
+      unicode: 'नेपाल.icom.museum'
+    },
+    {
+      ascii: 'xn--mgb9awbf.icom.museum',
+      unicode: 'عمان.icom.museum'
+    },
+    {
+      ascii: 'xn--wgbl6a.icom.museum',
+      unicode: 'قطر.icom.museum'
+    },
+    {
+      ascii: 'xn--romnia-yta.icom.museum',
+      unicode: 'românia.icom.museum'
+    },
+    {
+      ascii: 'xn--h1alffa9f.xn--h1aegh.museum',
+      unicode: 'россия.иком.museum'
+    },
+    {
+      ascii: 'xn--80aaabm1ab4blmeec9e7n.xn--h1aegh.museum',
+      unicode: 'србијаицрнагора.иком.museum'
+    },
+    {
+      ascii: 'xn--xkc2al3hye2a.icom.museum',
+      unicode: 'இலங்கை.icom.museum'
+    },
+    {
+      ascii: 'xn--espaa-rta.icom.museum',
+      unicode: 'españa.icom.museum'
+    },
+    {
+      ascii: 'xn--o3cw4h.icom.museum',
+      unicode: 'ไทย.icom.museum'
+    },
+    {
+      ascii: 'xn--pgbs0dh.icom.museum',
+      unicode: 'تونس.icom.museum'
+    },
+    {
+      ascii: 'xn--trkiye-3ya.icom.museum',
+      unicode: 'türkiye.icom.museum'
+    },
+    {
+      ascii: 'xn--80aaxgrpt.icom.museum',
+      unicode: 'украина.icom.museum'
+    },
+    {
+      ascii: 'xn--vitnam-jk8b.icom.museum',
+      unicode: 'việtnam.icom.museum'
+    },
+    // long URL
+    {
+      ascii: `${`${'a'.repeat(63)}.`.repeat(3)}com`,
+      unicode: `${`${'a'.repeat(63)}.`.repeat(3)}com`
+    }
+  ],
+  invalid: [
+    // long label
+    {
+      url: `${'a'.repeat(64)}.com`,
+      mode: 'ascii'
+    },
+    // long URL
+    {
+      url: `${`${'a'.repeat(63)}.`.repeat(4)}com`,
+      mode: 'ascii'
+    },
+    // invalid character
+    {
+      url: '\ufffd.com',
+      mode: 'ascii'
+    },
+    {
+      url: '\ufffd.com',
+      mode: 'unicode'
+    },
+    // invalid Punycode
+    {
+      url: 'xn---abc.com',
+      mode: 'unicode'
+    }
+  ]
+}
diff --git a/test/fixtures/url-tests.js b/test/fixtures/url-tests.js
index 3b162391cd2..3e303910ca6 100644
--- a/test/fixtures/url-tests.js
+++ b/test/fixtures/url-tests.js
@@ -3589,17 +3589,17 @@ module.exports =
     "base": "http://other.com/",
     "failure": true
   },
-  // "U+FFFD",
-  // {
-  //   "input": "https://\ufffd",
-  //   "base": "about:blank",
-  //   "failure": true
-  // },
-  // {
-  //   "input": "https://%EF%BF%BD",
-  //   "base": "about:blank",
-  //   "failure": true
-  // },
+  "U+FFFD",
+  {
+    "input": "https://\ufffd",
+    "base": "about:blank",
+    "failure": true
+  },
+  {
+    "input": "https://%EF%BF%BD",
+    "base": "about:blank",
+    "failure": true
+  },
   {
     "input": "https://x/\ufffd?\ufffd#\ufffd",
     "base": "about:blank",
@@ -4497,26 +4497,26 @@ module.exports =
     "hash": ""
   },
   "# Hosts and percent-encoding",
-  // {
-  //   "input": "ftp://example.com%80/",
-  //   "base": "about:blank",
-  //   "failure": true
-  // },
-  // {
-  //   "input": "ftp://example.com%A0/",
-  //   "base": "about:blank",
-  //   "failure": true
-  // },
-  // {
-  //   "input": "https://example.com%80/",
-  //   "base": "about:blank",
-  //   "failure": true
-  // },
-  // {
-  //   "input": "https://example.com%A0/",
-  //   "base": "about:blank",
-  //   "failure": true
-  // },
+  {
+    "input": "ftp://example.com%80/",
+    "base": "about:blank",
+    "failure": true
+  },
+  {
+    "input": "ftp://example.com%A0/",
+    "base": "about:blank",
+    "failure": true
+  },
+  {
+    "input": "https://example.com%80/",
+    "base": "about:blank",
+    "failure": true
+  },
+  {
+    "input": "https://example.com%A0/",
+    "base": "about:blank",
+    "failure": true
+  },
   {
     "input": "ftp://%e2%98%83",
     "base": "about:blank",
diff --git a/test/parallel/test-icu-punycode.js b/test/parallel/test-icu-punycode.js
index 62508bc9f78..411704bb8f4 100644
--- a/test/parallel/test-icu-punycode.js
+++ b/test/parallel/test-icu-punycode.js
@@ -6,67 +6,36 @@ if (!common.hasIntl) {
   return;
 }
 
-const icu = getPunycode();
+const icu = process.binding('icu');
 const assert = require('assert');
 
-function getPunycode() {
-  try {
-    return process.binding('icu');
-  } catch (err) {
-    return undefined;
+const tests = require('../fixtures/url-idna.js');
+
+{
+  for (const [i, { ascii, unicode }] of tests.valid.entries()) {
+    assert.strictEqual(ascii, icu.toASCII(unicode), `toASCII(${i + 1})`);
+    assert.strictEqual(unicode, icu.toUnicode(ascii), `toUnicode(${i + 1})`);
+    assert.strictEqual(ascii, icu.toASCII(icu.toUnicode(ascii)),
+                       `toASCII(toUnicode(${i + 1}))`);
+    assert.strictEqual(unicode, icu.toUnicode(icu.toASCII(unicode)),
+                       `toUnicode(toASCII(${i + 1}))`);
   }
 }
 
-// Credit for list: http://www.i18nguy.com/markup/idna-examples.html
-const tests = [
-  'افغانستا.icom.museum',
-  'الجزائر.icom.museum',
-  'österreich.icom.museum',
-  'বাংলাদেশ.icom.museum',
-  'беларусь.icom.museum',
-  'belgië.icom.museum',
-  'българия.icom.museum',
-  'تشادر.icom.museum',
-  '中国.icom.museum',
-  'القمر.icom.museum',
-  'κυπρος.icom.museum',
-  'českárepublika.icom.museum',
-  'مصر.icom.museum',
-  'ελλάδα.icom.museum',
-  'magyarország.icom.museum',
-  'ísland.icom.museum',
-  'भारत.icom.museum',
-  'ايران.icom.museum',
-  'éire.icom.museum',
-  'איקו״ם.ישראל.museum',
-  '日本.icom.museum',
-  'الأردن.icom.museum',
-  'қазақстан.icom.museum',
-  '한국.icom.museum',
-  'кыргызстан.icom.museum',
-  'ລາວ.icom.museum',
-  'لبنان.icom.museum',
-  'македонија.icom.museum',
-  'موريتانيا.icom.museum',
-  'méxico.icom.museum',
-  'монголулс.icom.museum',
-  'المغرب.icom.museum',
-  'नेपाल.icom.museum',
-  'عمان.icom.museum',
-  'قطر.icom.museum',
-  'românia.icom.museum',
-  'россия.иком.museum',
-  'србијаицрнагора.иком.museum',
-  'இலங்கை.icom.museum',
-  'españa.icom.museum',
-  'ไทย.icom.museum',
-  'تونس.icom.museum',
-  'türkiye.icom.museum',
-  'украина.icom.museum',
-  'việtnam.icom.museum'
-];
+{
+  const errorRe = {
+    ascii: /^Error: Cannot convert name to ASCII$/,
+    unicode: /^Error: Cannot convert name to Unicode$/
+  };
+  const convertFunc = {
+    ascii: icu.toASCII,
+    unicode: icu.toUnicode
+  };
 
-// Testing the roundtrip
-tests.forEach((i) => {
-  assert.strictEqual(i, icu.toUnicode(icu.toASCII(i)));
-});
+  for (const [i, { url, mode }] of tests.invalid.entries()) {
+    assert.throws(() => convertFunc[mode](url), errorRe[mode],
+                  `Invalid case ${i + 1}`);
+    assert.doesNotThrow(() => convertFunc[mode](url, true),
+                        `Invalid case ${i + 1} in lenient mode`);
+  }
+}
diff --git a/test/parallel/test-whatwg-url-domainto.js b/test/parallel/test-whatwg-url-domainto.js
new file mode 100644
index 00000000000..f891f95a19c
--- /dev/null
+++ b/test/parallel/test-whatwg-url-domainto.js
@@ -0,0 +1,36 @@
+'use strict';
+const common = require('../common');
+
+if (!common.hasIntl) {
+  common.skip('missing Intl');
+  return;
+}
+
+const assert = require('assert');
+const { domainToASCII, domainToUnicode } = require('url');
+
+// Tests below are not from WPT.
+const tests = require('../fixtures/url-idna.js');
+
+{
+  for (const [i, { ascii, unicode }] of tests.valid.entries()) {
+    assert.strictEqual(ascii, domainToASCII(unicode),
+                       `domainToASCII(${i + 1})`);
+    assert.strictEqual(unicode, domainToUnicode(ascii),
+                       `domainToUnicode(${i + 1})`);
+    assert.strictEqual(ascii, domainToASCII(domainToUnicode(ascii)),
+                       `domainToASCII(domainToUnicode(${i + 1}))`);
+    assert.strictEqual(unicode, domainToUnicode(domainToASCII(unicode)),
+                       `domainToUnicode(domainToASCII(${i + 1}))`);
+  }
+}
+
+{
+  const convertFunc = {
+    ascii: domainToASCII,
+    unicode: domainToUnicode
+  };
+
+  for (const [i, { url, mode }] of tests.invalid.entries())
+    assert.strictEqual(convertFunc[mode](url), '', `Invalid case ${i + 1}`);
+}