File
Blob: src/workerd/api/node/tests/punycode-nodejs-test.js
| 1 | // Copyright (c) 2025 Cloudflare, Inc. |
| 2 | // Licensed under the Apache 2.0 license found in the LICENSE file or at: |
| 3 | // https://opensource.org/licenses/Apache-2.0 |
| 4 | import assert from 'node:assert'; |
| 5 | import punycode from 'node:punycode'; |
| 6 | |
| 7 | export const testPunycode = { |
| 8 | test() { |
| 9 | assert.strictEqual(punycode.encode('ü'), 'tda'); |
| 10 | assert.strictEqual(punycode.encode('Goethe'), 'Goethe-'); |
| 11 | assert.strictEqual(punycode.encode('Bücher'), 'Bcher-kva'); |
| 12 | assert.strictEqual( |
| 13 | punycode.encode( |
| 14 | 'Willst du die Blüthe des frühen, die Früchte des späteren Jahres' |
| 15 | ), |
| 16 | 'Willst du die Blthe des frhen, die Frchte des spteren Jahres-x9e96lkal' |
| 17 | ); |
| 18 | assert.strictEqual(punycode.encode('日本語'), 'wgv71a119e'); |
| 19 | assert.strictEqual(punycode.encode('𩸽'), 'x73l'); |
| 20 | |
| 21 | assert.strictEqual(punycode.decode('tda'), 'ü'); |
| 22 | assert.strictEqual(punycode.decode('Goethe-'), 'Goethe'); |
| 23 | assert.strictEqual(punycode.decode('Bcher-kva'), 'Bücher'); |
| 24 | assert.strictEqual( |
| 25 | punycode.decode( |
| 26 | 'Willst du die Blthe des frhen, die Frchte des spteren Jahres-x9e96lkal' |
| 27 | ), |
| 28 | 'Willst du die Blüthe des frühen, die Früchte des späteren Jahres' |
| 29 | ); |
| 30 | assert.strictEqual(punycode.decode('wgv71a119e'), '日本語'); |
| 31 | assert.strictEqual(punycode.decode('x73l'), '𩸽'); |
| 32 | assert.throws(() => { |
| 33 | punycode.decode(' '); |
| 34 | }, /^RangeError: Invalid input$/); |
| 35 | assert.throws(() => { |
| 36 | punycode.decode('α-'); |
| 37 | }, /^RangeError: Illegal input >= 0x80 \(not a basic code point\)$/); |
| 38 | assert.throws(() => { |
| 39 | punycode.decode('あ'); |
| 40 | }, /^RangeError: Invalid input$/); |
| 41 | |
| 42 | // http://tools.ietf.org/html/rfc3492#section-7.1 |
| 43 | const tests = [ |
| 44 | // (A) Arabic (Egyptian) |
| 45 | { |
| 46 | encoded: 'egbpdaj6bu4bxfgehfvwxn', |
| 47 | decoded: |
| 48 | '\u0644\u064A\u0647\u0645\u0627\u0628\u062A\u0643\u0644\u0645' + |
| 49 | '\u0648\u0634\u0639\u0631\u0628\u064A\u061F', |
| 50 | }, |
| 51 | |
| 52 | // (B) Chinese (simplified) |
| 53 | { |
| 54 | encoded: 'ihqwcrb4cv8a8dqg056pqjye', |
| 55 | decoded: '\u4ED6\u4EEC\u4E3A\u4EC0\u4E48\u4E0D\u8BF4\u4E2D\u6587', |
| 56 | }, |
| 57 | |
| 58 | // (C) Chinese (traditional) |
| 59 | { |
| 60 | encoded: 'ihqwctvzc91f659drss3x8bo0yb', |
| 61 | decoded: '\u4ED6\u5011\u7232\u4EC0\u9EBD\u4E0D\u8AAA\u4E2D\u6587', |
| 62 | }, |
| 63 | |
| 64 | // (D) Czech: Pro<ccaron>prost<ecaron>nemluv<iacute><ccaron>esky |
| 65 | { |
| 66 | encoded: 'Proprostnemluvesky-uyb24dma41a', |
| 67 | decoded: |
| 68 | '\u0050\u0072\u006F\u010D\u0070\u0072\u006F\u0073\u0074\u011B' + |
| 69 | '\u006E\u0065\u006D\u006C\u0075\u0076\u00ED\u010D\u0065\u0073\u006B\u0079', |
| 70 | }, |
| 71 | |
| 72 | // (E) Hebrew |
| 73 | { |
| 74 | encoded: '4dbcagdahymbxekheh6e0a7fei0b', |
| 75 | decoded: |
| 76 | '\u05DC\u05DE\u05D4\u05D4\u05DD\u05E4\u05E9\u05D5\u05D8\u05DC' + |
| 77 | '\u05D0\u05DE\u05D3\u05D1\u05E8\u05D9\u05DD\u05E2\u05D1\u05E8\u05D9\u05EA', |
| 78 | }, |
| 79 | |
| 80 | // (F) Hindi (Devanagari) |
| 81 | { |
| 82 | encoded: 'i1baa7eci9glrd9b2ae1bj0hfcgg6iyaf8o0a1dig0cd', |
| 83 | decoded: |
| 84 | '\u092F\u0939\u0932\u094B\u0917\u0939\u093F\u0928\u094D\u0926' + |
| 85 | '\u0940\u0915\u094D\u092F\u094B\u0902\u0928\u0939\u0940\u0902\u092C' + |
| 86 | '\u094B\u0932\u0938\u0915\u0924\u0947\u0939\u0948\u0902', |
| 87 | }, |
| 88 | |
| 89 | // (G) Japanese (kanji and hiragana) |
| 90 | { |
| 91 | encoded: 'n8jok5ay5dzabd5bym9f0cm5685rrjetr6pdxa', |
| 92 | decoded: |
| 93 | '\u306A\u305C\u307F\u3093\u306A\u65E5\u672C\u8A9E\u3092\u8A71' + |
| 94 | '\u3057\u3066\u304F\u308C\u306A\u3044\u306E\u304B', |
| 95 | }, |
| 96 | |
| 97 | // (H) Korean (Hangul syllables) |
| 98 | { |
| 99 | encoded: |
| 100 | '989aomsvi5e83db1d2a355cv1e0vak1dwrv93d5xbh15a0dt30a5jpsd879' + |
| 101 | 'ccm6fea98c', |
| 102 | decoded: |
| 103 | '\uC138\uACC4\uC758\uBAA8\uB4E0\uC0AC\uB78C\uB4E4\uC774\uD55C' + |
| 104 | '\uAD6D\uC5B4\uB97C\uC774\uD574\uD55C\uB2E4\uBA74\uC5BC\uB9C8\uB098' + |
| 105 | '\uC88B\uC744\uAE4C', |
| 106 | }, |
| 107 | |
| 108 | // (I) Russian (Cyrillic) |
| 109 | { |
| 110 | encoded: 'b1abfaaepdrnnbgefbadotcwatmq2g4l', |
| 111 | decoded: |
| 112 | '\u043F\u043E\u0447\u0435\u043C\u0443\u0436\u0435\u043E\u043D' + |
| 113 | '\u0438\u043D\u0435\u0433\u043E\u0432\u043E\u0440\u044F\u0442\u043F' + |
| 114 | '\u043E\u0440\u0443\u0441\u0441\u043A\u0438', |
| 115 | }, |
| 116 | |
| 117 | // (J) Spanish: Porqu<eacute>nopuedensimplementehablarenEspa<ntilde>ol |
| 118 | { |
| 119 | encoded: 'PorqunopuedensimplementehablarenEspaol-fmd56a', |
| 120 | decoded: |
| 121 | '\u0050\u006F\u0072\u0071\u0075\u00E9\u006E\u006F\u0070\u0075' + |
| 122 | '\u0065\u0064\u0065\u006E\u0073\u0069\u006D\u0070\u006C\u0065\u006D' + |
| 123 | '\u0065\u006E\u0074\u0065\u0068\u0061\u0062\u006C\u0061\u0072\u0065' + |
| 124 | '\u006E\u0045\u0073\u0070\u0061\u00F1\u006F\u006C', |
| 125 | }, |
| 126 | |
| 127 | // (K) Vietnamese: T<adotbelow>isaoh<odotbelow>kh<ocirc>ngth |
| 128 | // <ecirchookabove>ch<ihookabove>n<oacute>iti<ecircacute>ngVi<ecircdotbelow>t |
| 129 | { |
| 130 | encoded: 'TisaohkhngthchnitingVit-kjcr8268qyxafd2f1b9g', |
| 131 | decoded: |
| 132 | '\u0054\u1EA1\u0069\u0073\u0061\u006F\u0068\u1ECD\u006B\u0068' + |
| 133 | '\u00F4\u006E\u0067\u0074\u0068\u1EC3\u0063\u0068\u1EC9\u006E\u00F3' + |
| 134 | '\u0069\u0074\u0069\u1EBF\u006E\u0067\u0056\u0069\u1EC7\u0074', |
| 135 | }, |
| 136 | |
| 137 | // (L) 3<nen>B<gumi><kinpachi><sensei> |
| 138 | { |
| 139 | encoded: '3B-ww4c5e180e575a65lsy2b', |
| 140 | decoded: '\u0033\u5E74\u0042\u7D44\u91D1\u516B\u5148\u751F', |
| 141 | }, |
| 142 | |
| 143 | // (M) <amuro><namie>-with-SUPER-MONKEYS |
| 144 | { |
| 145 | encoded: '-with-SUPER-MONKEYS-pc58ag80a8qai00g7n9n', |
| 146 | decoded: |
| 147 | '\u5B89\u5BA4\u5948\u7F8E\u6075\u002D\u0077\u0069\u0074\u0068' + |
| 148 | '\u002D\u0053\u0055\u0050\u0045\u0052\u002D\u004D\u004F\u004E\u004B' + |
| 149 | '\u0045\u0059\u0053', |
| 150 | }, |
| 151 | |
| 152 | // (N) Hello-Another-Way-<sorezore><no><basho> |
| 153 | { |
| 154 | encoded: 'Hello-Another-Way--fc4qua05auwb3674vfr0b', |
| 155 | decoded: |
| 156 | '\u0048\u0065\u006C\u006C\u006F\u002D\u0041\u006E\u006F\u0074' + |
| 157 | '\u0068\u0065\u0072\u002D\u0057\u0061\u0079\u002D\u305D\u308C\u305E' + |
| 158 | '\u308C\u306E\u5834\u6240', |
| 159 | }, |
| 160 | |
| 161 | // (O) <hitotsu><yane><no><shita>2 |
| 162 | { |
| 163 | encoded: '2-u9tlzr9756bt3uc0v', |
| 164 | decoded: '\u3072\u3068\u3064\u5C4B\u6839\u306E\u4E0B\u0032', |
| 165 | }, |
| 166 | |
| 167 | // (P) Maji<de>Koi<suru>5<byou><mae> |
| 168 | { |
| 169 | encoded: 'MajiKoi5-783gue6qz075azm5e', |
| 170 | decoded: |
| 171 | '\u004D\u0061\u006A\u0069\u3067\u004B\u006F\u0069\u3059\u308B' + |
| 172 | '\u0035\u79D2\u524D', |
| 173 | }, |
| 174 | |
| 175 | // (Q) <pafii>de<runba> |
| 176 | { |
| 177 | encoded: 'de-jg4avhby1noc0d', |
| 178 | decoded: '\u30D1\u30D5\u30A3\u30FC\u0064\u0065\u30EB\u30F3\u30D0', |
| 179 | }, |
| 180 | |
| 181 | // (R) <sono><supiido><de> |
| 182 | { |
| 183 | encoded: 'd9juau41awczczp', |
| 184 | decoded: '\u305D\u306E\u30B9\u30D4\u30FC\u30C9\u3067', |
| 185 | }, |
| 186 | |
| 187 | // (S) -> $1.00 <- |
| 188 | { |
| 189 | encoded: '-> $1.00 <--', |
| 190 | decoded: |
| 191 | '\u002D\u003E\u0020\u0024\u0031\u002E\u0030\u0030\u0020\u003C' + |
| 192 | '\u002D', |
| 193 | }, |
| 194 | ]; |
| 195 | |
| 196 | let errors = 0; |
| 197 | const handleError = (error, name) => { |
| 198 | console.error( |
| 199 | `FAIL: ${name} expected ${error.expected}, got ${error.actual}` |
| 200 | ); |
| 201 | errors++; |
| 202 | }; |
| 203 | |
| 204 | const regexNonASCII = /[^\x20-\x7E]/; |
| 205 | const testBattery = { |
| 206 | encode: (test) => |
| 207 | assert.strictEqual(punycode.encode(test.decoded), test.encoded), |
| 208 | decode: (test) => |
| 209 | assert.strictEqual(punycode.decode(test.encoded), test.decoded), |
| 210 | toASCII: (test) => |
| 211 | assert.strictEqual( |
| 212 | punycode.toASCII(test.decoded), |
| 213 | regexNonASCII.test(test.decoded) |
| 214 | ? `xn--${test.encoded}` |
| 215 | : test.decoded |
| 216 | ), |
| 217 | toUnicode: (test) => |
| 218 | assert.strictEqual( |
| 219 | punycode.toUnicode( |
| 220 | regexNonASCII.test(test.decoded) |
| 221 | ? `xn--${test.encoded}` |
| 222 | : test.decoded |
| 223 | ), |
| 224 | regexNonASCII.test(test.decoded) |
| 225 | ? test.decoded.toLowerCase() |
| 226 | : test.decoded |
| 227 | ), |
| 228 | }; |
| 229 | |
| 230 | tests.forEach((testCase) => { |
| 231 | Object.keys(testBattery).forEach((key) => { |
| 232 | try { |
| 233 | testBattery[key](testCase); |
| 234 | } catch (error) { |
| 235 | handleError(error, key); |
| 236 | } |
| 237 | }); |
| 238 | }); |
| 239 | |
| 240 | // BMP code point |
| 241 | assert.strictEqual(punycode.ucs2.encode([0x61]), 'a'); |
| 242 | // Supplementary code point (surrogate pair) |
| 243 | assert.strictEqual(punycode.ucs2.encode([0x1d306]), '\uD834\uDF06'); |
| 244 | // high surrogate |
| 245 | assert.strictEqual(punycode.ucs2.encode([0xd800]), '\uD800'); |
| 246 | // High surrogate followed by non-surrogates |
| 247 | assert.strictEqual(punycode.ucs2.encode([0xd800, 0x61, 0x62]), '\uD800ab'); |
| 248 | // low surrogate |
| 249 | assert.strictEqual(punycode.ucs2.encode([0xdc00]), '\uDC00'); |
| 250 | // Low surrogate followed by non-surrogates |
| 251 | assert.strictEqual(punycode.ucs2.encode([0xdc00, 0x61, 0x62]), '\uDC00ab'); |
| 252 | |
| 253 | assert.strictEqual(errors, 0); |
| 254 | |
| 255 | // test map domain |
| 256 | assert.strictEqual( |
| 257 | punycode.toASCII('Bücher@日本語.com'), |
| 258 | 'Bücher@xn--wgv71a119e.com' |
| 259 | ); |
| 260 | assert.strictEqual( |
| 261 | punycode.toUnicode('Bücher@xn--wgv71a119e.com'), |
| 262 | 'Bücher@日本語.com' |
| 263 | ); |
| 264 | }, |
| 265 | }; |