Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1 | // © 2016 and later: Unicode, Inc. and others. |
| 2 | // License & terms of use: http://www.unicode.org/copyright.html |
| 3 | /* |
| 4 | ******************************************************************************* |
| 5 | * |
| 6 | * Copyright (C) 2003-2016, International Business Machines |
| 7 | * Corporation and others. All Rights Reserved. |
| 8 | * |
| 9 | ******************************************************************************* |
| 10 | * file name: testidna.cpp |
| 11 | * encoding: UTF-8 |
| 12 | * tab size: 8 (not used) |
| 13 | * indentation:4 |
| 14 | * |
| 15 | * created on: 2003feb1 |
| 16 | * created by: Ram Viswanadha |
| 17 | */ |
| 18 | |
| 19 | #include "unicode/utypes.h" |
| 20 | |
| 21 | #if !UCONFIG_NO_IDNA && !UCONFIG_NO_TRANSLITERATION |
| 22 | |
| 23 | #include <time.h> |
| 24 | #include <limits.h> |
| 25 | #include <stdlib.h> |
| 26 | #include <string.h> |
| 27 | #include "unicode/localpointer.h" |
| 28 | #include "unicode/ustring.h" |
| 29 | #include "unicode/usprep.h" |
| 30 | #include "unicode/uniset.h" |
| 31 | #include "unicode/utf16.h" |
| 32 | #include "cmemory.h" |
| 33 | #include "testidna.h" |
| 34 | #include "idnaref.h" |
| 35 | #include "nptrans.h" |
| 36 | #include "unicode/putil.h" |
| 37 | #include "idnaconf.h" |
| 38 | |
| 39 | static const UChar unicodeIn[][41] ={ |
| 40 | { |
| 41 | 0x0644, 0x064A, 0x0647, 0x0645, 0x0627, 0x0628, 0x062A, 0x0643, 0x0644, |
| 42 | 0x0645, 0x0648, 0x0634, 0x0639, 0x0631, 0x0628, 0x064A, 0x061F, 0x0000 |
| 43 | }, |
| 44 | { |
| 45 | 0x4ED6, 0x4EEC, 0x4E3A, 0x4EC0, 0x4E48, 0x4E0D, 0x8BF4, 0x4E2D, 0x6587, |
| 46 | 0x0000 |
| 47 | }, |
| 48 | { |
| 49 | 0x0050, 0x0072, 0x006F, 0x010D, 0x0070, 0x0072, 0x006F, 0x0073, 0x0074, |
| 50 | 0x011B, 0x006E, 0x0065, 0x006D, 0x006C, 0x0075, 0x0076, 0x00ED, 0x010D, |
| 51 | 0x0065, 0x0073, 0x006B, 0x0079, 0x0000 |
| 52 | }, |
| 53 | { |
| 54 | 0x05DC, 0x05DE, 0x05D4, 0x05D4, 0x05DD, 0x05E4, 0x05E9, 0x05D5, 0x05D8, |
| 55 | 0x05DC, 0x05D0, 0x05DE, 0x05D3, 0x05D1, 0x05E8, 0x05D9, 0x05DD, 0x05E2, |
| 56 | 0x05D1, 0x05E8, 0x05D9, 0x05EA, 0x0000 |
| 57 | }, |
| 58 | { |
| 59 | 0x092F, 0x0939, 0x0932, 0x094B, 0x0917, 0x0939, 0x093F, 0x0928, 0x094D, |
| 60 | 0x0926, 0x0940, 0x0915, 0x094D, 0x092F, 0x094B, 0x0902, 0x0928, 0x0939, |
| 61 | 0x0940, 0x0902, 0x092C, 0x094B, 0x0932, 0x0938, 0x0915, 0x0924, 0x0947, |
| 62 | 0x0939, 0x0948, 0x0902, 0x0000 |
| 63 | }, |
| 64 | { |
| 65 | 0x306A, 0x305C, 0x307F, 0x3093, 0x306A, 0x65E5, 0x672C, 0x8A9E, 0x3092, |
| 66 | 0x8A71, 0x3057, 0x3066, 0x304F, 0x308C, 0x306A, 0x3044, 0x306E, 0x304B, |
| 67 | 0x0000 |
| 68 | }, |
| 69 | /* |
| 70 | { |
| 71 | 0xC138, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774, |
| 72 | 0xD55C, 0xAD6D, 0xC5B4, 0xB97C, 0xC774, 0xD574, 0xD55C, 0xB2E4, 0xBA74, |
| 73 | 0xC5BC, 0xB9C8, 0xB098, 0xC88B, 0xC744, 0xAE4C, 0x0000 |
| 74 | }, |
| 75 | */ |
| 76 | { |
| 77 | 0x043F, 0x043E, 0x0447, 0x0435, 0x043C, 0x0443, 0x0436, 0x0435, 0x043E, |
| 78 | 0x043D, 0x0438, 0x043D, 0x0435, 0x0433, 0x043E, 0x0432, 0x043E, 0x0440, |
| 79 | 0x044F, 0x0442, 0x043F, 0x043E, 0x0440, 0x0443, 0x0441, 0x0441, 0x043A, |
| 80 | 0x0438, 0x0000 |
| 81 | }, |
| 82 | { |
| 83 | 0x0050, 0x006F, 0x0072, 0x0071, 0x0075, 0x00E9, 0x006E, 0x006F, 0x0070, |
| 84 | 0x0075, 0x0065, 0x0064, 0x0065, 0x006E, 0x0073, 0x0069, 0x006D, 0x0070, |
| 85 | 0x006C, 0x0065, 0x006D, 0x0065, 0x006E, 0x0074, 0x0065, 0x0068, 0x0061, |
| 86 | 0x0062, 0x006C, 0x0061, 0x0072, 0x0065, 0x006E, 0x0045, 0x0073, 0x0070, |
| 87 | 0x0061, 0x00F1, 0x006F, 0x006C, 0x0000 |
| 88 | }, |
| 89 | { |
| 90 | 0x4ED6, 0x5011, 0x7232, 0x4EC0, 0x9EBD, 0x4E0D, 0x8AAA, 0x4E2D, 0x6587, |
| 91 | 0x0000 |
| 92 | }, |
| 93 | { |
| 94 | 0x0054, 0x1EA1, 0x0069, 0x0073, 0x0061, 0x006F, 0x0068, 0x1ECD, 0x006B, |
| 95 | 0x0068, 0x00F4, 0x006E, 0x0067, 0x0074, 0x0068, 0x1EC3, 0x0063, 0x0068, |
| 96 | 0x1EC9, 0x006E, 0x00F3, 0x0069, 0x0074, 0x0069, 0x1EBF, 0x006E, 0x0067, |
| 97 | 0x0056, 0x0069, 0x1EC7, 0x0074, 0x0000 |
| 98 | }, |
| 99 | { |
| 100 | 0x0033, 0x5E74, 0x0042, 0x7D44, 0x91D1, 0x516B, 0x5148, 0x751F, 0x0000 |
| 101 | }, |
| 102 | { |
| 103 | 0x5B89, 0x5BA4, 0x5948, 0x7F8E, 0x6075, 0x002D, 0x0077, 0x0069, 0x0074, |
| 104 | 0x0068, 0x002D, 0x0053, 0x0055, 0x0050, 0x0045, 0x0052, 0x002D, 0x004D, |
| 105 | 0x004F, 0x004E, 0x004B, 0x0045, 0x0059, 0x0053, 0x0000 |
| 106 | }, |
| 107 | { |
| 108 | 0x0048, 0x0065, 0x006C, 0x006C, 0x006F, 0x002D, 0x0041, 0x006E, 0x006F, |
| 109 | 0x0074, 0x0068, 0x0065, 0x0072, 0x002D, 0x0057, 0x0061, 0x0079, 0x002D, |
| 110 | 0x305D, 0x308C, 0x305E, 0x308C, 0x306E, 0x5834, 0x6240, 0x0000 |
| 111 | }, |
| 112 | { |
| 113 | 0x3072, 0x3068, 0x3064, 0x5C4B, 0x6839, 0x306E, 0x4E0B, 0x0032, 0x0000 |
| 114 | }, |
| 115 | { |
| 116 | 0x004D, 0x0061, 0x006A, 0x0069, 0x3067, 0x004B, 0x006F, 0x0069, 0x3059, |
| 117 | 0x308B, 0x0035, 0x79D2, 0x524D, 0x0000 |
| 118 | }, |
| 119 | { |
| 120 | 0x30D1, 0x30D5, 0x30A3, 0x30FC, 0x0064, 0x0065, 0x30EB, 0x30F3, 0x30D0, |
| 121 | 0x0000 |
| 122 | }, |
| 123 | { |
| 124 | 0x305D, 0x306E, 0x30B9, 0x30D4, 0x30FC, 0x30C9, 0x3067, 0x0000 |
| 125 | }, |
| 126 | // test non-BMP code points |
| 127 | { |
| 128 | 0xD800, 0xDF00, 0xD800, 0xDF01, 0xD800, 0xDF02, 0xD800, 0xDF03, 0xD800, 0xDF05, |
| 129 | 0xD800, 0xDF06, 0xD800, 0xDF07, 0xD800, 0xDF09, 0xD800, 0xDF0A, 0xD800, 0xDF0B, |
| 130 | 0x0000 |
| 131 | }, |
| 132 | { |
| 133 | 0xD800, 0xDF0D, 0xD800, 0xDF0C, 0xD800, 0xDF1E, 0xD800, 0xDF0F, 0xD800, 0xDF16, |
| 134 | 0xD800, 0xDF15, 0xD800, 0xDF14, 0xD800, 0xDF12, 0xD800, 0xDF10, 0xD800, 0xDF20, |
| 135 | 0xD800, 0xDF21, |
| 136 | 0x0000 |
| 137 | }, |
| 138 | // Greek |
| 139 | { |
| 140 | 0x03b5, 0x03bb, 0x03bb, 0x03b7, 0x03bd, 0x03b9, 0x03ba, 0x03ac |
| 141 | }, |
| 142 | // Maltese |
| 143 | { |
| 144 | 0x0062, 0x006f, 0x006e, 0x0121, 0x0075, 0x0073, 0x0061, 0x0127, |
| 145 | 0x0127, 0x0061 |
| 146 | }, |
| 147 | // Russian |
| 148 | { |
| 149 | 0x043f, 0x043e, 0x0447, 0x0435, 0x043c, 0x0443, 0x0436, 0x0435, |
| 150 | 0x043e, 0x043d, 0x0438, 0x043d, 0x0435, 0x0433, 0x043e, 0x0432, |
| 151 | 0x043e, 0x0440, 0x044f, 0x0442, 0x043f, 0x043e, 0x0440, 0x0443, |
| 152 | 0x0441, 0x0441, 0x043a, 0x0438 |
| 153 | }, |
| 154 | { |
| 155 | 0xFB00, 0xFB01 |
| 156 | } |
| 157 | |
| 158 | }; |
| 159 | |
| 160 | static const char *asciiIn[] = { |
| 161 | "xn--egbpdaj6bu4bxfgehfvwxn", |
| 162 | "xn--ihqwcrb4cv8a8dqg056pqjye", |
| 163 | "xn--Proprostnemluvesky-uyb24dma41a", |
| 164 | "xn--4dbcagdahymbxekheh6e0a7fei0b", |
| 165 | "xn--i1baa7eci9glrd9b2ae1bj0hfcgg6iyaf8o0a1dig0cd", |
| 166 | "xn--n8jok5ay5dzabd5bym9f0cm5685rrjetr6pdxa", |
| 167 | /* "xn--989aomsvi5e83db1d2a355cv1e0vak1dwrv93d5xbh15a0dt30a5jpsd879ccm6fea98c",*/ |
| 168 | "xn--b1abfaaepdrnnbgefbaDotcwatmq2g4l", |
| 169 | "xn--PorqunopuedensimplementehablarenEspaol-fmd56a", |
| 170 | "xn--ihqwctvzc91f659drss3x8bo0yb", |
| 171 | "xn--TisaohkhngthchnitingVit-kjcr8268qyxafd2f1b9g", |
| 172 | "xn--3B-ww4c5e180e575a65lsy2b", |
| 173 | "xn---with-SUPER-MONKEYS-pc58ag80a8qai00g7n9n", |
| 174 | "xn--Hello-Another-Way--fc4qua05auwb3674vfr0b", |
| 175 | "xn--2-u9tlzr9756bt3uc0v", |
| 176 | "xn--MajiKoi5-783gue6qz075azm5e", |
| 177 | "xn--de-jg4avhby1noc0d", |
| 178 | "xn--d9juau41awczczp", |
| 179 | "XN--097CCDEKGHQJK", |
| 180 | "XN--db8CBHEJLGH4E0AL", |
| 181 | "xn--hxargifdar", // Greek |
| 182 | "xn--bonusaa-5bb1da", // Maltese |
| 183 | "xn--b1abfaaepdrnnbgefbadotcwatmq2g4l", // Russian (Cyrillic) |
| 184 | "fffi" |
| 185 | }; |
| 186 | |
| 187 | static const char *domainNames[] = { |
| 188 | "slip129-37-118-146.nc.us.ibm.net", |
| 189 | "saratoga.pe.utexas.edu", |
| 190 | "dial-120-45.ots.utexas.edu", |
| 191 | "woo-085.dorms.waller.net", |
| 192 | "hd30-049.hil.compuserve.com", |
| 193 | "pem203-31.pe.ttu.edu", |
| 194 | "56K-227.MaxTNT3.pdq.net", |
| 195 | "dial-36-2.ots.utexas.edu", |
| 196 | "slip129-37-23-152.ga.us.ibm.net", |
| 197 | "ts45ip119.cadvision.com", |
| 198 | "sdn-ts-004txaustP05.dialsprint.net", |
| 199 | "bar-tnt1s66.erols.com", |
| 200 | "101.st-louis-15.mo.dial-access.att.net", |
| 201 | "h92-245.Arco.COM", |
| 202 | "dial-13-2.ots.utexas.edu", |
| 203 | "net-redynet29.datamarkets.com.ar", |
| 204 | "ccs-shiva28.reacciun.net.ve", |
| 205 | "7.houston-11.tx.dial-access.att.net", |
| 206 | "ingw129-37-120-26.mo.us.ibm.net", |
| 207 | "dialup6.austintx.com", |
| 208 | "dns2.tpao.gov.tr", |
| 209 | "slip129-37-119-194.nc.us.ibm.net", |
| 210 | "cs7.dillons.co.uk.203.119.193.in-addr.arpa", |
| 211 | "swprd1.innovplace.saskatoon.sk.ca", |
| 212 | "bikini.bologna.maraut.it", |
| 213 | "node91.subnet159-198-79.baxter.com", |
| 214 | "cust19.max5.new-york.ny.ms.uu.net", |
| 215 | "balexander.slip.andrew.cmu.edu", |
| 216 | "pool029.max2.denver.co.dynip.alter.net", |
| 217 | "cust49.max9.new-york.ny.ms.uu.net", |
| 218 | "s61.abq-dialin2.hollyberry.com", |
| 219 | "\\u0917\\u0928\\u0947\\u0936.sanjose.ibm.com", //':'(0x003a) produces U_IDNA_STD3_ASCII_RULES_ERROR |
| 220 | "www.xn--vea.com", |
| 221 | // "www.\\u00E0\\u00B3\\u00AF.com",//' ' (0x0020) produces U_IDNA_STD3_ASCII_RULES_ERROR |
| 222 | "www.\\u00C2\\u00A4.com", |
| 223 | "www.\\u00C2\\u00A3.com", |
| 224 | // "\\u0025", //'%' (0x0025) produces U_IDNA_STD3_ASCII_RULES_ERROR |
| 225 | // "\\u005C\\u005C", //'\' (0x005C) produces U_IDNA_STD3_ASCII_RULES_ERROR |
| 226 | //"@", |
| 227 | //"\\u002F", |
| 228 | //"www.\\u0021.com", |
| 229 | //"www.\\u0024.com", |
| 230 | //"\\u003f", |
| 231 | // These yield U_IDNA_PROHIBITED_ERROR |
| 232 | //"\\u00CF\\u0082.com", |
| 233 | //"\\u00CE\\u00B2\\u00C3\\u009Fss.com", |
| 234 | //"\\u00E2\\u0098\\u00BA.com", |
| 235 | "\\u00C3\\u00BC.com", |
| 236 | |
| 237 | }; |
| 238 | |
| 239 | typedef struct ErrorCases ErrorCases; |
| 240 | |
| 241 | static const struct ErrorCases{ |
| 242 | |
| 243 | UChar unicode[100]; |
| 244 | const char *ascii; |
| 245 | UErrorCode expected; |
| 246 | UBool useSTD3ASCIIRules; |
| 247 | UBool testToUnicode; |
| 248 | UBool testLabel; |
| 249 | } errorCases[] = { |
| 250 | { |
| 251 | |
| 252 | { |
| 253 | 0x0077, 0x0077, 0x0077, 0x002e, /* www. */ |
| 254 | 0xC138, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774, |
| 255 | 0x070F,/*prohibited*/ |
| 256 | 0xD55C, 0xAD6D, 0xC5B4, 0xB97C, 0xC774, 0xD574, 0xD55C, 0xB2E4, 0xBA74, |
| 257 | 0x002e, 0x0063, 0x006f, 0x006d, /* com. */ |
| 258 | 0x0000 |
| 259 | }, |
| 260 | "www.XN--8mb5595fsoa28orucya378bqre2tcwop06c5qbw82a1rffmae0361dea96b.com", |
| 261 | U_IDNA_PROHIBITED_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 262 | false, false, true |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 263 | }, |
| 264 | |
| 265 | { |
| 266 | { |
| 267 | 0x0077, 0x0077, 0x0077, 0x002e, /* www. */ |
| 268 | 0xC138, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774, |
| 269 | 0x0221, 0x0234/*Unassigned code points*/, |
| 270 | 0x002e, 0x0063, 0x006f, 0x006d, /* com. */ |
| 271 | 0x0000 |
| 272 | }, |
| 273 | "www.XN--6lA2Bz548Fj1GuA391Bf1Gb1N59Ab29A7iA.com", |
| 274 | |
| 275 | U_IDNA_UNASSIGNED_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 276 | false, false, true |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 277 | }, |
| 278 | { |
| 279 | { |
| 280 | 0x0077, 0x0077, 0x0077, 0x002e, /* www. */ |
| 281 | 0xC138, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774, |
| 282 | 0x0644, 0x064A, 0x0647,/*Arabic code points. Cannot mix RTL with LTR*/ |
| 283 | 0xD55C, 0xAD6D, 0xC5B4, 0xB97C, 0xC774, 0xD574, 0xD55C, 0xB2E4, 0xBA74, |
| 284 | 0x002e, 0x0063, 0x006f, 0x006d, /* com. */ |
| 285 | 0x0000 |
| 286 | }, |
| 287 | "www.xn--ghBGI4851OiyA33VqrD6Az86C4qF83CtRv93D5xBk15AzfG0nAgA0578DeA71C.com", |
| 288 | U_IDNA_CHECK_BIDI_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 289 | false, false, true |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 290 | }, |
| 291 | { |
| 292 | { |
| 293 | 0x0077, 0x0077, 0x0077, 0x002e, /* www. */ |
| 294 | /* labels cannot begin with an HYPHEN */ |
| 295 | 0x002D, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774, |
| 296 | 0x002E, |
| 297 | 0xD55C, 0xAD6D, 0xC5B4, 0xB97C, 0xC774, 0xD574, 0xD55C, 0xB2E4, 0xBA74, |
| 298 | 0x002e, 0x0063, 0x006f, 0x006d, /* com. */ |
| 299 | 0x0000 |
| 300 | |
| 301 | }, |
| 302 | "www.xn----b95Ew8SqA315Ao5FbuMlnNmhA.com", |
| 303 | U_IDNA_STD3_ASCII_RULES_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 304 | true, false, false |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 305 | }, |
| 306 | { |
| 307 | { |
| 308 | /* correct ACE-prefix followed by unicode */ |
| 309 | 0x0077, 0x0077, 0x0077, 0x002e, /* www. */ |
| 310 | 0x0078, 0x006e, 0x002d,0x002d, /* ACE Prefix */ |
| 311 | 0x002D, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774, |
| 312 | 0x002D, |
| 313 | 0xD55C, 0xAD6D, 0xC5B4, 0xB97C, 0xC774, 0xD574, 0xD55C, 0xB2E4, 0xBA74, |
| 314 | 0x002e, 0x0063, 0x006f, 0x006d, /* com. */ |
| 315 | 0x0000 |
| 316 | |
| 317 | }, |
| 318 | /* wrong ACE-prefix followed by valid ACE-encoded ASCII */ |
| 319 | "www.XY-----b91I0V65S96C2A355Cw1E5yCeQr19CsnP1mFfmAE0361DeA96B.com", |
| 320 | U_IDNA_ACE_PREFIX_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 321 | false, false, false |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 322 | }, |
| 323 | /* cannot verify U_IDNA_VERIFICATION_ERROR */ |
| 324 | |
| 325 | { |
| 326 | { |
| 327 | 0x0077, 0x0077, 0x0077, 0x002e, /* www. */ |
| 328 | 0xC138, 0xACC4, 0xC758, 0xBAA8, 0xB4E0, 0xC0AC, 0xB78C, 0xB4E4, 0xC774, |
| 329 | 0xD55C, 0xAD6D, 0xC5B4, 0xB97C, 0xC774, 0xD574, 0xD55C, 0xB2E4, 0xBA74, |
| 330 | 0xC5BC, 0xB9C8, 0xB098, 0xC88B, 0xC744, 0xAE4C, |
| 331 | 0x002e, 0x0063, 0x006f, 0x006d, /* com. */ |
| 332 | 0x0000 |
| 333 | }, |
| 334 | "www.xn--989AoMsVi5E83Db1D2A355Cv1E0vAk1DwRv93D5xBh15A0Dt30A5JpSD879Ccm6FeA98C.com", |
| 335 | U_IDNA_LABEL_TOO_LONG_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 336 | false, false, true |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 337 | }, |
| 338 | |
| 339 | { |
| 340 | { |
| 341 | 0x0077, 0x0077, 0x0077, 0x002e, /* www. */ |
| 342 | 0x0030, 0x0644, 0x064A, 0x0647, 0x0031, /* Arabic code points squashed between EN codepoints */ |
| 343 | 0x002e, 0x0063, 0x006f, 0x006d, /* com. */ |
| 344 | 0x0000 |
| 345 | }, |
| 346 | "www.xn--01-tvdmo.com", |
| 347 | U_IDNA_CHECK_BIDI_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 348 | false, false, true |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 349 | }, |
| 350 | |
| 351 | { |
| 352 | { |
| 353 | 0x0077, 0x0077, 0x0077, 0x002e, // www. |
| 354 | 0x206C, 0x0644, 0x064A, 0x0647, 0x206D, // Arabic code points squashed between BN codepoints |
| 355 | 0x002e, 0x0063, 0x006f, 0x006d, // com. |
| 356 | 0x0000 |
| 357 | }, |
| 358 | "www.XN--ghbgi278xia.com", |
| 359 | U_IDNA_PROHIBITED_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 360 | false, false, true |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 361 | }, |
| 362 | { |
| 363 | { |
| 364 | 0x0077, 0x0077, 0x0077, 0x002e, // www. |
| 365 | 0x002D, 0x0041, 0x0042, 0x0043, 0x0044, 0x0045, // HYPHEN at the start of label |
| 366 | 0x002e, 0x0063, 0x006f, 0x006d, // com. |
| 367 | 0x0000 |
| 368 | }, |
| 369 | "www.-abcde.com", |
| 370 | U_IDNA_STD3_ASCII_RULES_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 371 | true, false, false |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 372 | }, |
| 373 | { |
| 374 | { |
| 375 | 0x0077, 0x0077, 0x0077, 0x002e, // www. |
| 376 | 0x0041, 0x0042, 0x0043, 0x0044, 0x0045,0x002D, // HYPHEN at the end of the label |
| 377 | 0x002e, 0x0063, 0x006f, 0x006d, // com. |
| 378 | 0x0000 |
| 379 | }, |
| 380 | "www.abcde-.com", |
| 381 | U_IDNA_STD3_ASCII_RULES_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 382 | true, false, false |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 383 | }, |
| 384 | { |
| 385 | { |
| 386 | 0x0077, 0x0077, 0x0077, 0x002e, // www. |
| 387 | 0x0041, 0x0042, 0x0043, 0x0044, 0x0045,0x0040, // Containing non LDH code point |
| 388 | 0x002e, 0x0063, 0x006f, 0x006d, // com. |
| 389 | 0x0000 |
| 390 | }, |
| 391 | "www.abcde@.com", |
| 392 | U_IDNA_STD3_ASCII_RULES_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 393 | true, false, false |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 394 | }, |
| 395 | { |
| 396 | { |
| 397 | 0x0077, 0x0077, 0x0077, 0x002e, // www. |
| 398 | // zero length label |
| 399 | 0x002e, 0x0063, 0x006f, 0x006d, // com. |
| 400 | 0x0000 |
| 401 | }, |
| 402 | "www..com", |
| 403 | U_IDNA_ZERO_LENGTH_LABEL_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 404 | true, false, false |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 405 | }, |
| 406 | { |
| 407 | {0}, |
| 408 | NULL, |
| 409 | U_ILLEGAL_ARGUMENT_ERROR, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 410 | true, true, false |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 411 | } |
| 412 | }; |
| 413 | |
| 414 | |
| 415 | |
| 416 | |
| 417 | #define MAX_DEST_SIZE 300 |
| 418 | |
| 419 | void TestIDNA::debug(const UChar* src, int32_t srcLength, int32_t options){ |
| 420 | UParseError parseError; |
| 421 | UErrorCode transStatus = U_ZERO_ERROR; |
| 422 | UErrorCode prepStatus = U_ZERO_ERROR; |
| 423 | NamePrepTransform* trans = NamePrepTransform::createInstance(parseError,transStatus); |
| 424 | int32_t prepOptions = (((options & UIDNA_ALLOW_UNASSIGNED) != 0) ? USPREP_ALLOW_UNASSIGNED: 0); |
| 425 | LocalUStringPrepProfilePointer prep(usprep_openByType(USPREP_RFC3491_NAMEPREP,&prepStatus)); |
| 426 | UChar *transOut=NULL, *prepOut=NULL; |
| 427 | int32_t transOutLength=0, prepOutLength=0; |
| 428 | |
| 429 | |
| 430 | transOutLength = trans->process(src,srcLength,transOut, 0, prepOptions>0, &parseError, transStatus); |
| 431 | if( transStatus == U_BUFFER_OVERFLOW_ERROR){ |
| 432 | transStatus = U_ZERO_ERROR; |
| 433 | transOut = (UChar*) malloc(U_SIZEOF_UCHAR * transOutLength); |
| 434 | transOutLength = trans->process(src,srcLength,transOut, transOutLength, prepOptions>0, &parseError, transStatus); |
| 435 | } |
| 436 | |
| 437 | prepOutLength = usprep_prepare(prep.getAlias(), src, srcLength, prepOut, 0, prepOptions, &parseError, &prepStatus); |
| 438 | |
| 439 | if( prepStatus == U_BUFFER_OVERFLOW_ERROR){ |
| 440 | prepStatus = U_ZERO_ERROR; |
| 441 | prepOut = (UChar*) malloc(U_SIZEOF_UCHAR * prepOutLength); |
| 442 | prepOutLength = usprep_prepare(prep.getAlias(), src, srcLength, prepOut, prepOutLength, prepOptions, &parseError, &prepStatus); |
| 443 | } |
| 444 | |
| 445 | if(UnicodeString(transOut,transOutLength)!= UnicodeString(prepOut, prepOutLength)){ |
| 446 | errln("Failed. Expected: " + prettify(UnicodeString(transOut, transOutLength)) |
| 447 | + " Got: " + prettify(UnicodeString(prepOut,prepOutLength))); |
| 448 | } |
| 449 | free(transOut); |
| 450 | free(prepOut); |
| 451 | delete trans; |
| 452 | } |
| 453 | |
| 454 | void TestIDNA::testAPI(const UChar* src, const UChar* expected, const char* testName, |
| 455 | UBool useSTD3ASCIIRules,UErrorCode expectedStatus, |
| 456 | UBool doCompare, UBool testUnassigned, TestFunc func, UBool testSTD3ASCIIRules){ |
| 457 | |
| 458 | UErrorCode status = U_ZERO_ERROR; |
| 459 | UChar destStack[MAX_DEST_SIZE]; |
| 460 | int32_t destLen = 0; |
| 461 | UChar* dest = NULL; |
| 462 | int32_t expectedLen = (expected != NULL) ? u_strlen(expected) : 0; |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 463 | int32_t options = (useSTD3ASCIIRules == true) ? UIDNA_USE_STD3_RULES : UIDNA_DEFAULT; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 464 | UParseError parseError; |
| 465 | int32_t tSrcLen = 0; |
| 466 | UChar* tSrc = NULL; |
| 467 | |
| 468 | if(src != NULL){ |
| 469 | tSrcLen = u_strlen(src); |
| 470 | tSrc =(UChar*) malloc( U_SIZEOF_UCHAR * tSrcLen ); |
| 471 | memcpy(tSrc,src,tSrcLen * U_SIZEOF_UCHAR); |
| 472 | } |
| 473 | |
| 474 | // test null-terminated source and return value of number of UChars required |
| 475 | destLen = func(src,-1,NULL,0,options, &parseError , &status); |
| 476 | if (status == U_FILE_ACCESS_ERROR) { |
| 477 | dataerrln("U_FILE_ACCESS_ERROR. Skipping the remainder of this test."); |
| 478 | return; |
| 479 | } |
| 480 | if(status == U_BUFFER_OVERFLOW_ERROR){ |
| 481 | status = U_ZERO_ERROR; // reset error code |
| 482 | if(destLen+1 < MAX_DEST_SIZE){ |
| 483 | dest = destStack; |
| 484 | destLen = func(src,-1,dest,destLen+1,options, &parseError, &status); |
| 485 | // TODO : compare output with expected |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 486 | if(U_SUCCESS(status) && expectedStatus != U_IDNA_STD3_ASCII_RULES_ERROR&& (doCompare==true) && u_strCaseCompare(dest,destLen, expected,expectedLen,0,&status)!=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 487 | errln("Did not get the expected result for "+UnicodeString(testName) +" null terminated source. Expected : " |
| 488 | + prettify(UnicodeString(expected,expectedLen)) |
| 489 | + " Got: " + prettify(UnicodeString(dest,destLen)) |
| 490 | ); |
| 491 | } |
| 492 | }else{ |
| 493 | errln( "%s null terminated source failed. Requires destCapacity > 300\n",testName); |
| 494 | } |
| 495 | } |
| 496 | |
| 497 | if(status != expectedStatus){ |
| 498 | errcheckln(status, "Did not get the expected error for "+ |
| 499 | UnicodeString(testName)+ |
| 500 | " null terminated source. Expected: " +UnicodeString(u_errorName(expectedStatus)) |
| 501 | + " Got: "+ UnicodeString(u_errorName(status)) |
| 502 | + " Source: " + prettify(UnicodeString(src)) |
| 503 | ); |
| 504 | free(tSrc); |
| 505 | return; |
| 506 | } |
| 507 | if(testUnassigned ){ |
| 508 | status = U_ZERO_ERROR; |
| 509 | destLen = func(src,-1,NULL,0,options | UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 510 | if(status == U_BUFFER_OVERFLOW_ERROR){ |
| 511 | status = U_ZERO_ERROR; // reset error code |
| 512 | if(destLen+1 < MAX_DEST_SIZE){ |
| 513 | dest = destStack; |
| 514 | destLen = func(src,-1,dest,destLen+1,options | UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 515 | // TODO : compare output with expected |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 516 | if(U_SUCCESS(status) && (doCompare==true) && u_strCaseCompare(dest,destLen, expected,expectedLen,0,&status)!=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 517 | //errln("Did not get the expected result for %s null terminated source with both options set.\n",testName); |
| 518 | errln("Did not get the expected result for "+UnicodeString(testName) + |
| 519 | " null terminated source "+ prettify(src) + |
| 520 | " with both options set. Expected: "+ prettify(UnicodeString(expected,expectedLen))+ |
| 521 | "Got: " + prettify(UnicodeString(dest,destLen))); |
| 522 | |
| 523 | debug(src,-1,options | UIDNA_ALLOW_UNASSIGNED); |
| 524 | |
| 525 | } |
| 526 | }else{ |
| 527 | errln( "%s null terminated source failed. Requires destCapacity > 300\n",testName); |
| 528 | } |
| 529 | } |
| 530 | //testing query string |
| 531 | if(status != expectedStatus && expectedStatus != U_IDNA_UNASSIGNED_ERROR){ |
| 532 | errln( "Did not get the expected error for "+ |
| 533 | UnicodeString(testName)+ |
| 534 | " null terminated source with options set. Expected: " +UnicodeString(u_errorName(expectedStatus)) |
| 535 | + " Got: "+ UnicodeString(u_errorName(status)) |
| 536 | + " Source: " + prettify(UnicodeString(src)) |
| 537 | ); |
| 538 | } |
| 539 | } |
| 540 | |
| 541 | status = U_ZERO_ERROR; |
| 542 | |
| 543 | // test source with lengthand return value of number of UChars required |
| 544 | destLen = func(tSrc, tSrcLen, NULL,0,options, &parseError, &status); |
| 545 | if(status == U_BUFFER_OVERFLOW_ERROR){ |
| 546 | status = U_ZERO_ERROR; // reset error code |
| 547 | if(destLen+1 < MAX_DEST_SIZE){ |
| 548 | dest = destStack; |
| 549 | destLen = func(src,u_strlen(src),dest,destLen+1,options, &parseError, &status); |
| 550 | // TODO : compare output with expected |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 551 | if(U_SUCCESS(status) && (doCompare==true) && u_strCaseCompare(dest,destLen, expected,expectedLen,0,&status)!=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 552 | errln("Did not get the expected result for %s with source length.\n",testName); |
| 553 | } |
| 554 | }else{ |
| 555 | errln( "%s with source length failed. Requires destCapacity > 300\n",testName); |
| 556 | } |
| 557 | } |
| 558 | |
| 559 | if(status != expectedStatus){ |
| 560 | errln( "Did not get the expected error for "+ |
| 561 | UnicodeString(testName)+ |
| 562 | " with source length. Expected: " +UnicodeString(u_errorName(expectedStatus)) |
| 563 | + " Got: "+ UnicodeString(u_errorName(status)) |
| 564 | + " Source: " + prettify(UnicodeString(src)) |
| 565 | ); |
| 566 | } |
| 567 | if(testUnassigned){ |
| 568 | status = U_ZERO_ERROR; |
| 569 | |
| 570 | destLen = func(tSrc,tSrcLen,NULL,0,options | UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 571 | |
| 572 | if(status == U_BUFFER_OVERFLOW_ERROR){ |
| 573 | status = U_ZERO_ERROR; // reset error code |
| 574 | if(destLen+1 < MAX_DEST_SIZE){ |
| 575 | dest = destStack; |
| 576 | destLen = func(src,u_strlen(src),dest,destLen+1,options | UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 577 | // TODO : compare output with expected |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 578 | if(U_SUCCESS(status) && (doCompare==true) && u_strCaseCompare(dest,destLen, expected,expectedLen,0,&status)!=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 579 | errln("Did not get the expected result for %s with source length and both options set.\n",testName); |
| 580 | } |
| 581 | }else{ |
| 582 | errln( "%s with source length failed. Requires destCapacity > 300\n",testName); |
| 583 | } |
| 584 | } |
| 585 | //testing query string |
| 586 | if(status != expectedStatus && expectedStatus != U_IDNA_UNASSIGNED_ERROR){ |
| 587 | errln( "Did not get the expected error for "+ |
| 588 | UnicodeString(testName)+ |
| 589 | " with source length and options set. Expected: " +UnicodeString(u_errorName(expectedStatus)) |
| 590 | + " Got: "+ UnicodeString(u_errorName(status)) |
| 591 | + " Source: " + prettify(UnicodeString(src)) |
| 592 | ); |
| 593 | } |
| 594 | } |
| 595 | |
| 596 | status = U_ZERO_ERROR; |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 597 | if(testSTD3ASCIIRules==true){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 598 | destLen = func(src,-1,NULL,0,options | UIDNA_USE_STD3_RULES, &parseError, &status); |
| 599 | if(status == U_BUFFER_OVERFLOW_ERROR){ |
| 600 | status = U_ZERO_ERROR; // reset error code |
| 601 | if(destLen+1 < MAX_DEST_SIZE){ |
| 602 | dest = destStack; |
| 603 | destLen = func(src,-1,dest,destLen+1,options | UIDNA_USE_STD3_RULES, &parseError, &status); |
| 604 | // TODO : compare output with expected |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 605 | if(U_SUCCESS(status) && (doCompare==true) && u_strCaseCompare(dest,destLen, expected,expectedLen,0,&status)!=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 606 | //errln("Did not get the expected result for %s null terminated source with both options set.\n",testName); |
| 607 | errln("Did not get the expected result for "+UnicodeString(testName) +" null terminated source with both options set. Expected: "+ prettify(UnicodeString(expected,expectedLen))); |
| 608 | |
| 609 | } |
| 610 | }else{ |
| 611 | errln( "%s null terminated source failed. Requires destCapacity > 300\n",testName); |
| 612 | } |
| 613 | } |
| 614 | //testing query string |
| 615 | if(status != expectedStatus){ |
| 616 | errln( "Did not get the expected error for "+ |
| 617 | UnicodeString(testName)+ |
| 618 | " null terminated source with options set. Expected: " +UnicodeString(u_errorName(expectedStatus)) |
| 619 | + " Got: "+ UnicodeString(u_errorName(status)) |
| 620 | + " Source: " + prettify(UnicodeString(src)) |
| 621 | ); |
| 622 | } |
| 623 | |
| 624 | status = U_ZERO_ERROR; |
| 625 | |
| 626 | destLen = func(tSrc,tSrcLen,NULL,0,options | UIDNA_USE_STD3_RULES, &parseError, &status); |
| 627 | |
| 628 | if(status == U_BUFFER_OVERFLOW_ERROR){ |
| 629 | status = U_ZERO_ERROR; // reset error code |
| 630 | if(destLen+1 < MAX_DEST_SIZE){ |
| 631 | dest = destStack; |
| 632 | destLen = func(src,u_strlen(src),dest,destLen+1,options | UIDNA_USE_STD3_RULES, &parseError, &status); |
| 633 | // TODO : compare output with expected |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 634 | if(U_SUCCESS(status) && (doCompare==true) && u_strCaseCompare(dest,destLen, expected,expectedLen,0,&status)!=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 635 | errln("Did not get the expected result for %s with source length and both options set.\n",testName); |
| 636 | } |
| 637 | }else{ |
| 638 | errln( "%s with source length failed. Requires destCapacity > 300\n",testName); |
| 639 | } |
| 640 | } |
| 641 | //testing query string |
| 642 | if(status != expectedStatus && expectedStatus != U_IDNA_UNASSIGNED_ERROR){ |
| 643 | errln( "Did not get the expected error for "+ |
| 644 | UnicodeString(testName)+ |
| 645 | " with source length and options set. Expected: " +UnicodeString(u_errorName(expectedStatus)) |
| 646 | + " Got: "+ UnicodeString(u_errorName(status)) |
| 647 | + " Source: " + prettify(UnicodeString(src)) |
| 648 | ); |
| 649 | } |
| 650 | } |
| 651 | free(tSrc); |
| 652 | } |
| 653 | |
| 654 | void TestIDNA::testCompare(const UChar* s1, int32_t s1Len, |
| 655 | const UChar* s2, int32_t s2Len, |
| 656 | const char* testName, CompareFunc func, |
| 657 | UBool isEqual){ |
| 658 | |
| 659 | UErrorCode status = U_ZERO_ERROR; |
| 660 | int32_t retVal = func(s1,-1,s2,-1,UIDNA_DEFAULT,&status); |
| 661 | |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 662 | if(isEqual==true && retVal !=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 663 | errln("Did not get the expected result for %s with null termniated strings.\n",testName); |
| 664 | } |
| 665 | if(U_FAILURE(status)){ |
| 666 | errcheckln(status, "%s null terminated source failed. Error: %s", testName,u_errorName(status)); |
| 667 | } |
| 668 | |
| 669 | status = U_ZERO_ERROR; |
| 670 | retVal = func(s1,-1,s2,-1,UIDNA_ALLOW_UNASSIGNED,&status); |
| 671 | |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 672 | if(isEqual==true && retVal !=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 673 | errln("Did not get the expected result for %s with null termniated strings with options set.\n", testName); |
| 674 | } |
| 675 | if(U_FAILURE(status)){ |
| 676 | errcheckln(status, "%s null terminated source and options set failed. Error: %s",testName, u_errorName(status)); |
| 677 | } |
| 678 | |
| 679 | status = U_ZERO_ERROR; |
| 680 | retVal = func(s1,s1Len,s2,s2Len,UIDNA_DEFAULT,&status); |
| 681 | |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 682 | if(isEqual==true && retVal !=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 683 | errln("Did not get the expected result for %s with string length.\n",testName); |
| 684 | } |
| 685 | if(U_FAILURE(status)){ |
| 686 | errcheckln(status, "%s with string length. Error: %s",testName, u_errorName(status)); |
| 687 | } |
| 688 | |
| 689 | status = U_ZERO_ERROR; |
| 690 | retVal = func(s1,s1Len,s2,s2Len,UIDNA_ALLOW_UNASSIGNED,&status); |
| 691 | |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 692 | if(isEqual==true && retVal !=0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 693 | errln("Did not get the expected result for %s with string length and options set.\n",testName); |
| 694 | } |
| 695 | if(U_FAILURE(status)){ |
| 696 | errcheckln(status, "%s with string length and options set. Error: %s", u_errorName(status), testName); |
| 697 | } |
| 698 | } |
| 699 | |
| 700 | void TestIDNA::testToASCII(const char* testName, TestFunc func){ |
| 701 | |
| 702 | int32_t i; |
| 703 | UChar buf[MAX_DEST_SIZE]; |
| 704 | |
| 705 | for(i=0;i< UPRV_LENGTHOF(unicodeIn); i++){ |
| 706 | u_charsToUChars(asciiIn[i],buf, (int32_t)(strlen(asciiIn[i])+1)); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 707 | testAPI(unicodeIn[i], buf,testName, false,U_ZERO_ERROR, true, true, func); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 708 | |
| 709 | } |
| 710 | } |
| 711 | |
| 712 | void TestIDNA::testToUnicode(const char* testName, TestFunc func){ |
| 713 | |
| 714 | int32_t i; |
| 715 | UChar buf[MAX_DEST_SIZE]; |
| 716 | |
| 717 | for(i=0;i< UPRV_LENGTHOF(asciiIn); i++){ |
| 718 | u_charsToUChars(asciiIn[i],buf, (int32_t)(strlen(asciiIn[i])+1)); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 719 | testAPI(buf,unicodeIn[i],testName,false,U_ZERO_ERROR, true, true, func); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 720 | } |
| 721 | } |
| 722 | |
| 723 | |
| 724 | void TestIDNA::testIDNToUnicode(const char* testName, TestFunc func){ |
| 725 | int32_t i; |
| 726 | UChar buf[MAX_DEST_SIZE]; |
| 727 | UChar expected[MAX_DEST_SIZE]; |
| 728 | UErrorCode status = U_ZERO_ERROR; |
| 729 | int32_t bufLen = 0; |
| 730 | UParseError parseError; |
| 731 | for(i=0;i< UPRV_LENGTHOF(domainNames); i++){ |
| 732 | bufLen = (int32_t)strlen(domainNames[i]); |
| 733 | bufLen = u_unescape(domainNames[i],buf, bufLen+1); |
| 734 | func(buf,bufLen,expected,MAX_DEST_SIZE, UIDNA_ALLOW_UNASSIGNED, &parseError,&status); |
| 735 | if(U_FAILURE(status)){ |
| 736 | errcheckln(status, "%s failed to convert domainNames[%i].Error: %s",testName, i, u_errorName(status)); |
| 737 | break; |
| 738 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 739 | testAPI(buf,expected,testName,false,U_ZERO_ERROR, true, true, func); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 740 | //test toUnicode with all labels in the string |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 741 | testAPI(buf,expected,testName, false,U_ZERO_ERROR, true, true, func); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 742 | if(U_FAILURE(status)){ |
| 743 | errln( "%s failed to convert domainNames[%i].Error: %s \n",testName,i, u_errorName(status)); |
| 744 | break; |
| 745 | } |
| 746 | } |
| 747 | |
| 748 | } |
| 749 | |
| 750 | void TestIDNA::testIDNToASCII(const char* testName, TestFunc func){ |
| 751 | int32_t i; |
| 752 | UChar buf[MAX_DEST_SIZE]; |
| 753 | UChar expected[MAX_DEST_SIZE]; |
| 754 | UErrorCode status = U_ZERO_ERROR; |
| 755 | int32_t bufLen = 0; |
| 756 | UParseError parseError; |
| 757 | for(i=0;i< UPRV_LENGTHOF(domainNames); i++){ |
| 758 | bufLen = (int32_t)strlen(domainNames[i]); |
| 759 | bufLen = u_unescape(domainNames[i],buf, bufLen+1); |
| 760 | func(buf,bufLen,expected,MAX_DEST_SIZE, UIDNA_ALLOW_UNASSIGNED, &parseError,&status); |
| 761 | if(U_FAILURE(status)){ |
| 762 | errcheckln(status, "%s failed to convert domainNames[%i].Error: %s",testName,i, u_errorName(status)); |
| 763 | break; |
| 764 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 765 | testAPI(buf,expected,testName, false,U_ZERO_ERROR, true, true, func); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 766 | //test toASCII with all labels in the string |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 767 | testAPI(buf,expected,testName, false,U_ZERO_ERROR, false, true, func); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 768 | if(U_FAILURE(status)){ |
| 769 | errln( "%s failed to convert domainNames[%i].Error: %s \n",testName,i, u_errorName(status)); |
| 770 | break; |
| 771 | } |
| 772 | } |
| 773 | |
| 774 | } |
| 775 | |
| 776 | void TestIDNA::testCompare(const char* testName, CompareFunc func){ |
| 777 | int32_t i; |
| 778 | |
| 779 | |
| 780 | UChar www[] = {0x0057, 0x0057, 0x0057, 0x002E, 0x0000}; |
| 781 | UChar com[] = {0x002E, 0x0043, 0x004F, 0x004D, 0x0000}; |
| 782 | UChar buf[MAX_DEST_SIZE]={0x0057, 0x0057, 0x0057, 0x002E, 0x0000}; |
| 783 | |
| 784 | UnicodeString source(www), uni0(www),uni1(www), ascii0(www), ascii1(www); |
| 785 | |
| 786 | uni0.append(unicodeIn[0]); |
| 787 | uni0.append(com); |
| 788 | uni0.append((UChar)0x0000); |
| 789 | |
| 790 | uni1.append(unicodeIn[1]); |
| 791 | uni1.append(com); |
| 792 | uni1.append((UChar)0x0000); |
| 793 | |
| 794 | ascii0.append(asciiIn[0]); |
| 795 | ascii0.append(com); |
| 796 | ascii0.append((UChar)0x0000); |
| 797 | |
| 798 | ascii1.append(asciiIn[1]); |
| 799 | ascii1.append(com); |
| 800 | ascii1.append((UChar)0x0000); |
| 801 | |
| 802 | for(i=0;i< UPRV_LENGTHOF(unicodeIn); i++){ |
| 803 | |
| 804 | u_charsToUChars(asciiIn[i],buf+4, (int32_t)(strlen(asciiIn[i])+1)); |
| 805 | u_strcat(buf,com); |
| 806 | |
| 807 | // for every entry in unicodeIn array |
| 808 | // prepend www. and append .com |
| 809 | source.truncate(4); |
| 810 | source.append(unicodeIn[i]); |
| 811 | source.append(com); |
| 812 | source.append((UChar)0x0000); |
| 813 | // a) compare it with itself |
| 814 | const UChar* src = source.getBuffer(); |
| 815 | int32_t srcLen = u_strlen(src); //subtract null |
| 816 | |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 817 | testCompare(src,srcLen,src,srcLen,testName, func, true); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 818 | |
| 819 | // b) compare it with asciiIn equivalent |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 820 | testCompare(src,srcLen,buf,u_strlen(buf),testName, func,true); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 821 | |
| 822 | // c) compare it with unicodeIn not equivalent |
| 823 | if(i==0){ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 824 | testCompare(src,srcLen,uni1.getBuffer(),uni1.length()-1,testName, func,false); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 825 | }else{ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 826 | testCompare(src,srcLen,uni0.getBuffer(),uni0.length()-1,testName, func,false); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 827 | } |
| 828 | // d) compare it with asciiIn not equivalent |
| 829 | if(i==0){ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 830 | testCompare(src,srcLen,ascii1.getBuffer(),ascii1.length()-1,testName, func,false); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 831 | }else{ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 832 | testCompare(src,srcLen,ascii0.getBuffer(),ascii0.length()-1,testName, func,false); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 833 | } |
| 834 | |
| 835 | } |
| 836 | } |
| 837 | |
| 838 | #if 0 |
| 839 | |
| 840 | static int32_t |
| 841 | getNextSeperator(UChar *src,int32_t srcLength, |
| 842 | UChar **limit){ |
| 843 | if(srcLength == -1){ |
| 844 | int32_t i; |
| 845 | for(i=0 ; ;i++){ |
| 846 | if(src[i] == 0){ |
| 847 | *limit = src + i; // point to null |
| 848 | return i; |
| 849 | } |
| 850 | if(src[i]==0x002e){ |
| 851 | *limit = src + (i+1); // go past the delimiter |
| 852 | return i; |
| 853 | } |
| 854 | } |
| 855 | // we have not found the delimiter |
| 856 | if(i==srcLength){ |
| 857 | *limit = src+srcLength; |
| 858 | } |
| 859 | return i; |
| 860 | }else{ |
| 861 | int32_t i; |
| 862 | for(i=0;i<srcLength;i++){ |
| 863 | if(src[i]==0x002e){ |
| 864 | *limit = src + (i+1); // go past the delimiter |
| 865 | return i; |
| 866 | } |
| 867 | } |
| 868 | // we have not found the delimiter |
| 869 | if(i==srcLength){ |
| 870 | *limit = src+srcLength; |
| 871 | } |
| 872 | return i; |
| 873 | } |
| 874 | } |
| 875 | |
| 876 | void printPunycodeOutput(){ |
| 877 | |
| 878 | UChar dest[MAX_DEST_SIZE]; |
| 879 | int32_t destCapacity=MAX_DEST_SIZE; |
| 880 | UChar* start; |
| 881 | UChar* limit; |
| 882 | int32_t labelLen=0; |
| 883 | UBool caseFlags[MAX_DEST_SIZE]; |
| 884 | |
| 885 | for(int32_t i=0;i< UPRV_LENGTHOF(errorCases);i++){ |
| 886 | ErrorCases errorCase = errorCases[i]; |
| 887 | UErrorCode status = U_ZERO_ERROR; |
| 888 | start = errorCase.unicode; |
| 889 | int32_t srcLen = u_strlen(start); |
| 890 | labelLen = getNextSeperator(start,srcLen,&limit); |
| 891 | start = limit; |
| 892 | labelLen=getNextSeperator(start,srcLen-labelLen,&limit); |
| 893 | int32_t destLen = u_strToPunycode(dest,destCapacity,start,labelLen,caseFlags, &status); |
| 894 | if(U_FAILURE(status)){ |
| 895 | printf("u_strToPunycode failed for index %i\n",i); |
| 896 | continue; |
| 897 | } |
| 898 | for(int32_t j=0; j<destLen; j++){ |
| 899 | printf("%c",(char)dest[j]); |
| 900 | } |
| 901 | printf("\n"); |
| 902 | } |
| 903 | } |
| 904 | #endif |
| 905 | |
| 906 | void TestIDNA::testErrorCases(const char* IDNToASCIIName, TestFunc IDNToASCII, |
| 907 | const char* IDNToUnicodeName, TestFunc IDNToUnicode){ |
| 908 | UChar buf[MAX_DEST_SIZE]; |
| 909 | int32_t bufLen=0; |
| 910 | |
| 911 | for(int32_t i=0;i< UPRV_LENGTHOF(errorCases);i++){ |
| 912 | ErrorCases errorCase = errorCases[i]; |
| 913 | UChar* src =NULL; |
| 914 | if(errorCase.ascii != NULL){ |
| 915 | bufLen = (int32_t)strlen(errorCase.ascii); |
| 916 | u_charsToUChars(errorCase.ascii,buf, bufLen+1); |
| 917 | }else{ |
| 918 | bufLen = 1 ; |
| 919 | memset(buf,0,U_SIZEOF_UCHAR*MAX_DEST_SIZE); |
| 920 | } |
| 921 | |
| 922 | if(errorCase.unicode[0]!=0){ |
| 923 | src = errorCase.unicode; |
| 924 | } |
| 925 | // test toASCII |
| 926 | testAPI(src,buf, |
| 927 | IDNToASCIIName, errorCase.useSTD3ASCIIRules, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 928 | errorCase.expected, true, true, IDNToASCII); |
| 929 | if(errorCase.testLabel ==true){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 930 | testAPI(src,buf, |
| 931 | IDNToASCIIName, errorCase.useSTD3ASCIIRules, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 932 | errorCase.expected, false,true, IDNToASCII); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 933 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 934 | if(errorCase.testToUnicode ==true){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 935 | testAPI((src==NULL)? NULL : buf,src, |
| 936 | IDNToUnicodeName, errorCase.useSTD3ASCIIRules, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 937 | errorCase.expected, true, true, IDNToUnicode); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 938 | } |
| 939 | |
| 940 | } |
| 941 | |
| 942 | } |
| 943 | /* |
| 944 | void TestIDNA::testConformance(const char* toASCIIName, TestFunc toASCII, |
| 945 | const char* IDNToASCIIName, TestFunc IDNToASCII, |
| 946 | const char* IDNToUnicodeName, TestFunc IDNToUnicode, |
| 947 | const char* toUnicodeName, TestFunc toUnicode){ |
| 948 | UChar src[MAX_DEST_SIZE]; |
| 949 | int32_t srcLen=0; |
| 950 | UChar expected[MAX_DEST_SIZE]; |
| 951 | int32_t expectedLen = 0; |
| 952 | for(int32_t i=0;i< UPRV_LENGTHOF(conformanceTestCases);i++){ |
| 953 | const char* utf8Chars1 = conformanceTestCases[i].in; |
| 954 | int32_t utf8Chars1Len = (int32_t)strlen(utf8Chars1); |
| 955 | const char* utf8Chars2 = conformanceTestCases[i].out; |
| 956 | int32_t utf8Chars2Len = (utf8Chars2 == NULL) ? 0 : (int32_t)strlen(utf8Chars2); |
| 957 | |
| 958 | UErrorCode status = U_ZERO_ERROR; |
| 959 | u_strFromUTF8(src,MAX_DEST_SIZE,&srcLen,utf8Chars1,utf8Chars1Len,&status); |
| 960 | if(U_FAILURE(status)){ |
| 961 | errln(UnicodeString("Conversion of UTF8 source in conformanceTestCases[") + i +UnicodeString( "].in ( ")+prettify(utf8Chars1) +UnicodeString(" ) failed. Error: ")+ UnicodeString(u_errorName(status))); |
| 962 | continue; |
| 963 | } |
| 964 | if(utf8Chars2 != NULL){ |
| 965 | u_strFromUTF8(expected,MAX_DEST_SIZE,&expectedLen,utf8Chars2,utf8Chars2Len, &status); |
| 966 | if(U_FAILURE(status)){ |
| 967 | errln(UnicodeString("Conversion of UTF8 source in conformanceTestCases[") + i +UnicodeString( "].in ( ")+prettify(utf8Chars1) +UnicodeString(" ) failed. Error: ")+ UnicodeString(u_errorName(status))); |
| 968 | continue; |
| 969 | } |
| 970 | } |
| 971 | |
| 972 | if(conformanceTestCases[i].expectedStatus != U_ZERO_ERROR){ |
| 973 | // test toASCII |
| 974 | testAPI(src,expected, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 975 | IDNToASCIIName, false, |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 976 | conformanceTestCases[i].expectedStatus, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 977 | true, |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 978 | (conformanceTestCases[i].expectedStatus != U_IDNA_UNASSIGNED_ERROR), |
| 979 | IDNToASCII); |
| 980 | |
| 981 | testAPI(src,expected, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 982 | toASCIIName, false, |
| 983 | conformanceTestCases[i].expectedStatus, true, |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 984 | (conformanceTestCases[i].expectedStatus != U_IDNA_UNASSIGNED_ERROR), |
| 985 | toASCII); |
| 986 | } |
| 987 | |
| 988 | testAPI(src,src, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 989 | IDNToUnicodeName, false, |
| 990 | conformanceTestCases[i].expectedStatus, true, true, IDNToUnicode); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 991 | testAPI(src,src, |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 992 | toUnicodeName, false, |
| 993 | conformanceTestCases[i].expectedStatus, true, true, toUnicode); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 994 | |
| 995 | } |
| 996 | |
| 997 | } |
| 998 | */ |
| 999 | // test and ascertain |
| 1000 | // func(func(func(src))) == func(src) |
| 1001 | void TestIDNA::testChaining(const UChar* src,int32_t numIterations,const char* testName, |
| 1002 | UBool useSTD3ASCIIRules, UBool caseInsensitive, TestFunc func){ |
| 1003 | UChar even[MAX_DEST_SIZE]; |
| 1004 | UChar odd[MAX_DEST_SIZE]; |
| 1005 | UChar expected[MAX_DEST_SIZE]; |
| 1006 | int32_t i=0,evenLen=0,oddLen=0,expectedLen=0; |
| 1007 | UErrorCode status = U_ZERO_ERROR; |
| 1008 | int32_t srcLen = u_strlen(src); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1009 | int32_t options = (useSTD3ASCIIRules == true) ? UIDNA_USE_STD3_RULES : UIDNA_DEFAULT; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1010 | UParseError parseError; |
| 1011 | |
| 1012 | // test null-terminated source |
| 1013 | expectedLen = func(src,-1,expected,MAX_DEST_SIZE, options, &parseError, &status); |
| 1014 | if(U_FAILURE(status)){ |
| 1015 | errcheckln(status, "%s null terminated source failed. Error: %s",testName, u_errorName(status)); |
| 1016 | } |
| 1017 | memcpy(odd,expected,(expectedLen+1) * U_SIZEOF_UCHAR); |
| 1018 | memcpy(even,expected,(expectedLen+1) * U_SIZEOF_UCHAR); |
| 1019 | for(;i<=numIterations; i++){ |
| 1020 | if((i%2) ==0){ |
| 1021 | evenLen = func(odd,-1,even,MAX_DEST_SIZE,options, &parseError, &status); |
| 1022 | if(U_FAILURE(status)){ |
| 1023 | errcheckln(status, "%s null terminated source failed - %s",testName, u_errorName(status)); |
| 1024 | break; |
| 1025 | } |
| 1026 | }else{ |
| 1027 | oddLen = func(even,-1,odd,MAX_DEST_SIZE,options, &parseError, &status); |
| 1028 | if(U_FAILURE(status)){ |
| 1029 | errln("%s null terminated source failed\n",testName); |
| 1030 | break; |
| 1031 | } |
| 1032 | } |
| 1033 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1034 | if(caseInsensitive ==true){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1035 | if( u_strCaseCompare(even,evenLen, expected,expectedLen, 0, &status) !=0 || |
| 1036 | u_strCaseCompare(odd,oddLen, expected,expectedLen, 0, &status) !=0 ){ |
| 1037 | |
| 1038 | errln("Chaining for %s null terminated source failed\n",testName); |
| 1039 | } |
| 1040 | }else{ |
| 1041 | if( u_strncmp(even,expected,expectedLen) != 0 || |
| 1042 | u_strncmp(odd,expected,expectedLen) !=0 ){ |
| 1043 | |
| 1044 | errln("Chaining for %s null terminated source failed\n",testName); |
| 1045 | } |
| 1046 | } |
| 1047 | |
| 1048 | // test null-terminated source |
| 1049 | status = U_ZERO_ERROR; |
| 1050 | expectedLen = func(src,-1,expected,MAX_DEST_SIZE,options|UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 1051 | if(U_FAILURE(status)){ |
| 1052 | errcheckln(status, "%s null terminated source with options set failed. Error: %s",testName, u_errorName(status)); |
| 1053 | } |
| 1054 | memcpy(odd,expected,(expectedLen+1) * U_SIZEOF_UCHAR); |
| 1055 | memcpy(even,expected,(expectedLen+1) * U_SIZEOF_UCHAR); |
| 1056 | for(;i<=numIterations; i++){ |
| 1057 | if((i%2) ==0){ |
| 1058 | evenLen = func(odd,-1,even,MAX_DEST_SIZE,options|UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 1059 | if(U_FAILURE(status)){ |
| 1060 | errcheckln(status, "%s null terminated source with options set failed - %s",testName, u_errorName(status)); |
| 1061 | break; |
| 1062 | } |
| 1063 | }else{ |
| 1064 | oddLen = func(even,-1,odd,MAX_DEST_SIZE,options|UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 1065 | if(U_FAILURE(status)){ |
| 1066 | errln("%s null terminated source with options set failed\n",testName); |
| 1067 | break; |
| 1068 | } |
| 1069 | } |
| 1070 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1071 | if(caseInsensitive ==true){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1072 | if( u_strCaseCompare(even,evenLen, expected,expectedLen, 0, &status) !=0 || |
| 1073 | u_strCaseCompare(odd,oddLen, expected,expectedLen, 0, &status) !=0 ){ |
| 1074 | |
| 1075 | errln("Chaining for %s null terminated source with options set failed\n",testName); |
| 1076 | } |
| 1077 | }else{ |
| 1078 | if( u_strncmp(even,expected,expectedLen) != 0 || |
| 1079 | u_strncmp(odd,expected,expectedLen) !=0 ){ |
| 1080 | |
| 1081 | errln("Chaining for %s null terminated source with options set failed\n",testName); |
| 1082 | } |
| 1083 | } |
| 1084 | |
| 1085 | |
| 1086 | // test source with length |
| 1087 | status = U_ZERO_ERROR; |
| 1088 | expectedLen = func(src,srcLen,expected,MAX_DEST_SIZE,options, &parseError, &status); |
| 1089 | if(U_FAILURE(status)){ |
| 1090 | errcheckln(status, "%s null terminated source failed. Error: %s",testName, u_errorName(status)); |
| 1091 | } |
| 1092 | memcpy(odd,expected,(expectedLen+1) * U_SIZEOF_UCHAR); |
| 1093 | memcpy(even,expected,(expectedLen+1) * U_SIZEOF_UCHAR); |
| 1094 | for(;i<=numIterations; i++){ |
| 1095 | if((i%2) ==0){ |
| 1096 | evenLen = func(odd,oddLen,even,MAX_DEST_SIZE,options, &parseError, &status); |
| 1097 | if(U_FAILURE(status)){ |
| 1098 | errcheckln(status, "%s source with source length failed - %s",testName, u_errorName(status)); |
| 1099 | break; |
| 1100 | } |
| 1101 | }else{ |
| 1102 | oddLen = func(even,evenLen,odd,MAX_DEST_SIZE,options, &parseError, &status); |
| 1103 | if(U_FAILURE(status)){ |
| 1104 | errcheckln(status, "%s source with source length failed - %s",testName, u_errorName(status)); |
| 1105 | break; |
| 1106 | } |
| 1107 | } |
| 1108 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1109 | if(caseInsensitive ==true){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1110 | if( u_strCaseCompare(even,evenLen, expected,expectedLen, 0, &status) !=0 || |
| 1111 | u_strCaseCompare(odd,oddLen, expected,expectedLen, 0, &status) !=0 ){ |
| 1112 | |
| 1113 | errln("Chaining for %s source with source length failed\n",testName); |
| 1114 | } |
| 1115 | }else{ |
| 1116 | if( u_strncmp(even,expected,expectedLen) != 0 || |
| 1117 | u_strncmp(odd,expected,expectedLen) !=0 ){ |
| 1118 | |
| 1119 | errln("Chaining for %s source with source length failed\n",testName); |
| 1120 | } |
| 1121 | } |
| 1122 | status = U_ZERO_ERROR; |
| 1123 | expectedLen = func(src,srcLen,expected,MAX_DEST_SIZE,options|UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 1124 | if(U_FAILURE(status)){ |
| 1125 | errcheckln(status, "%s null terminated source with options set failed. Error: %s",testName, u_errorName(status)); |
| 1126 | } |
| 1127 | memcpy(odd,expected,(expectedLen+1) * U_SIZEOF_UCHAR); |
| 1128 | memcpy(even,expected,(expectedLen+1) * U_SIZEOF_UCHAR); |
| 1129 | for(;i<=numIterations; i++){ |
| 1130 | if((i%2) ==0){ |
| 1131 | evenLen = func(odd,oddLen,even,MAX_DEST_SIZE,options|UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 1132 | if(U_FAILURE(status)){ |
| 1133 | errcheckln(status, "%s source with source length and options set failed - %s",testName, u_errorName(status)); |
| 1134 | break; |
| 1135 | } |
| 1136 | }else{ |
| 1137 | oddLen = func(even,evenLen,odd,MAX_DEST_SIZE,options|UIDNA_ALLOW_UNASSIGNED, &parseError, &status); |
| 1138 | if(U_FAILURE(status)){ |
| 1139 | errcheckln(status, "%s source with source length and options set failed - %s",testName, u_errorName(status)); |
| 1140 | break; |
| 1141 | } |
| 1142 | } |
| 1143 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1144 | if(caseInsensitive ==true){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1145 | if( u_strCaseCompare(even,evenLen, expected,expectedLen, 0, &status) !=0 || |
| 1146 | u_strCaseCompare(odd,oddLen, expected,expectedLen, 0, &status) !=0 ){ |
| 1147 | |
| 1148 | errln("Chaining for %s source with source length and options set failed\n",testName); |
| 1149 | } |
| 1150 | }else{ |
| 1151 | if( u_strncmp(even,expected,expectedLen) != 0 || |
| 1152 | u_strncmp(odd,expected,expectedLen) !=0 ){ |
| 1153 | |
| 1154 | errln("Chaining for %s source with source length and options set failed\n",testName); |
| 1155 | } |
| 1156 | } |
| 1157 | } |
| 1158 | void TestIDNA::testChaining(const char* toASCIIName, TestFunc toASCII, |
| 1159 | const char* toUnicodeName, TestFunc toUnicode){ |
| 1160 | int32_t i; |
| 1161 | UChar buf[MAX_DEST_SIZE]; |
| 1162 | |
| 1163 | for(i=0;i< UPRV_LENGTHOF(asciiIn); i++){ |
| 1164 | u_charsToUChars(asciiIn[i],buf, (int32_t)(strlen(asciiIn[i])+1)); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1165 | testChaining(buf,5,toUnicodeName, false, false, toUnicode); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1166 | } |
| 1167 | for(i=0;i< UPRV_LENGTHOF(unicodeIn); i++){ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1168 | testChaining(unicodeIn[i], 5,toASCIIName, false, true, toASCII); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1169 | } |
| 1170 | } |
| 1171 | |
| 1172 | |
| 1173 | void TestIDNA::testRootLabelSeparator(const char* testName, CompareFunc func, |
| 1174 | const char* IDNToASCIIName, TestFunc IDNToASCII, |
| 1175 | const char* IDNToUnicodeName, TestFunc IDNToUnicode){ |
| 1176 | int32_t i; |
| 1177 | |
| 1178 | |
| 1179 | UChar www[] = {0x0057, 0x0057, 0x0057, 0x002E, 0x0000}; |
| 1180 | UChar com[] = {0x002E, 0x0043, 0x004F, 0x004D, 0x002E, /* root label separator */0x0000}; |
| 1181 | UChar buf[MAX_DEST_SIZE]={0x0057, 0x0057, 0x0057, 0x002E, 0x0000}; |
| 1182 | |
| 1183 | UnicodeString source(www), uni0(www),uni1(www), ascii0(www), ascii1(www); |
| 1184 | |
| 1185 | uni0.append(unicodeIn[0]); |
| 1186 | uni0.append(com); |
| 1187 | uni0.append((UChar)0x0000); |
| 1188 | |
| 1189 | uni1.append(unicodeIn[1]); |
| 1190 | uni1.append(com); |
| 1191 | uni1.append((UChar)0x0000); |
| 1192 | |
| 1193 | ascii0.append(asciiIn[0]); |
| 1194 | ascii0.append(com); |
| 1195 | ascii0.append((UChar)0x0000); |
| 1196 | |
| 1197 | ascii1.append(asciiIn[1]); |
| 1198 | ascii1.append(com); |
| 1199 | ascii1.append((UChar)0x0000); |
| 1200 | |
| 1201 | for(i=0;i< UPRV_LENGTHOF(unicodeIn); i++){ |
| 1202 | |
| 1203 | u_charsToUChars(asciiIn[i],buf+4, (int32_t)(strlen(asciiIn[i])+1)); |
| 1204 | u_strcat(buf,com); |
| 1205 | |
| 1206 | // for every entry in unicodeIn array |
| 1207 | // prepend www. and append .com |
| 1208 | source.truncate(4); |
| 1209 | source.append(unicodeIn[i]); |
| 1210 | source.append(com); |
| 1211 | source.append((UChar)0x0000); |
| 1212 | |
| 1213 | const UChar* src = source.getBuffer(); |
| 1214 | int32_t srcLen = u_strlen(src); //subtract null |
| 1215 | |
| 1216 | // b) compare it with asciiIn equivalent |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1217 | testCompare(src,srcLen,buf,u_strlen(buf),testName, func,true); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1218 | |
| 1219 | // a) compare it with itself |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1220 | testCompare(src,srcLen,src,srcLen,testName, func,true); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1221 | |
| 1222 | |
| 1223 | // IDNToASCII comparison |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1224 | testAPI(src,buf,IDNToASCIIName,false,U_ZERO_ERROR,true, true, IDNToASCII); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1225 | // IDNToUnicode comparison |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1226 | testAPI(buf,src,IDNToUnicodeName, false,U_ZERO_ERROR, true, true, IDNToUnicode); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1227 | |
| 1228 | // c) compare it with unicodeIn not equivalent |
| 1229 | if(i==0){ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1230 | testCompare(src,srcLen,uni1.getBuffer(),uni1.length()-1,testName, func,false); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1231 | }else{ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1232 | testCompare(src,srcLen,uni0.getBuffer(),uni0.length()-1,testName, func,false); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1233 | } |
| 1234 | // d) compare it with asciiIn not equivalent |
| 1235 | if(i==0){ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1236 | testCompare(src,srcLen,ascii1.getBuffer(),ascii1.length()-1,testName, func,false); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1237 | }else{ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1238 | testCompare(src,srcLen,ascii0.getBuffer(),ascii0.length()-1,testName, func,false); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1239 | } |
| 1240 | } |
| 1241 | } |
| 1242 | |
| 1243 | //--------------------------------------------- |
| 1244 | // runIndexedTest |
| 1245 | //--------------------------------------------- |
| 1246 | |
| 1247 | extern IntlTest *createUTS46Test(); |
| 1248 | |
| 1249 | void TestIDNA::runIndexedTest( int32_t index, UBool exec, const char* &name, char* par) |
| 1250 | { |
| 1251 | if (exec) logln((UnicodeString)"TestSuite IDNA API "); |
| 1252 | switch (index) { |
| 1253 | |
| 1254 | case 0: name = "TestToASCII"; if (exec) TestToASCII(); break; |
| 1255 | case 1: name = "TestToUnicode"; if (exec) TestToUnicode(); break; |
| 1256 | case 2: name = "TestIDNToASCII"; if (exec) TestIDNToASCII(); break; |
| 1257 | case 3: name = "TestIDNToUnicode"; if (exec) TestIDNToUnicode(); break; |
| 1258 | case 4: name = "TestCompare"; if (exec) TestCompare(); break; |
| 1259 | case 5: name = "TestErrorCases"; if (exec) TestErrorCases(); break; |
| 1260 | case 6: name = "TestChaining"; if (exec) TestChaining(); break; |
| 1261 | case 7: name = "TestRootLabelSeparator"; if(exec) TestRootLabelSeparator(); break; |
| 1262 | case 8: name = "TestCompareReferenceImpl"; if(exec) TestCompareReferenceImpl(); break; |
| 1263 | case 9: name = "TestDataFile"; if(exec) TestDataFile(); break; |
| 1264 | #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION |
| 1265 | case 10: name = "TestRefIDNA"; if(exec) TestRefIDNA(); break; |
| 1266 | case 11: name = "TestIDNAMonkeyTest"; if(exec) TestIDNAMonkeyTest(); break; |
| 1267 | #else |
| 1268 | case 10: case 11: name = "skip"; break; |
| 1269 | #endif |
| 1270 | case 12: |
| 1271 | { |
| 1272 | name = "TestConformanceTestVectors"; |
| 1273 | if(exec){ |
| 1274 | logln("TestSuite IDNA conf----"); logln(); |
| 1275 | IdnaConfTest test; |
| 1276 | callTest(test, par); |
| 1277 | } |
| 1278 | break; |
| 1279 | } |
| 1280 | case 13: |
| 1281 | name = "UTS46Test"; |
| 1282 | if (exec) { |
| 1283 | logln("TestSuite UTS46Test---"); logln(); |
| 1284 | LocalPointer<IntlTest> test(createUTS46Test()); |
| 1285 | callTest(*test, par); |
| 1286 | } |
| 1287 | break; |
| 1288 | default: name = ""; break; /*needed to end loop*/ |
| 1289 | } |
| 1290 | } |
| 1291 | void TestIDNA::TestToASCII(){ |
| 1292 | testToASCII("uidna_toASCII", uidna_toASCII); |
| 1293 | } |
| 1294 | void TestIDNA::TestToUnicode(){ |
| 1295 | testToUnicode("uidna_toUnicode", uidna_toUnicode); |
| 1296 | } |
| 1297 | void TestIDNA::TestIDNToASCII(){ |
| 1298 | testIDNToASCII("uidna_IDNToASCII", uidna_IDNToASCII); |
| 1299 | } |
| 1300 | void TestIDNA::TestIDNToUnicode(){ |
| 1301 | testIDNToUnicode("uidna_IDNToUnicode", uidna_IDNToUnicode); |
| 1302 | } |
| 1303 | void TestIDNA::TestCompare(){ |
| 1304 | UErrorCode status = U_ZERO_ERROR; |
| 1305 | uidna_close(uidna_openUTS46(0, &status)); // Fail quickly if no data. |
| 1306 | if (assertSuccess("", status, true, __FILE__, __LINE__)) { |
| 1307 | testCompare("uidna_compare",uidna_compare); |
| 1308 | } |
| 1309 | } |
| 1310 | void TestIDNA::TestErrorCases(){ |
| 1311 | testErrorCases( "uidna_IDNToASCII",uidna_IDNToASCII, |
| 1312 | "uidna_IDNToUnicode",uidna_IDNToUnicode); |
| 1313 | } |
| 1314 | void TestIDNA::TestRootLabelSeparator(){ |
| 1315 | UErrorCode status = U_ZERO_ERROR; |
| 1316 | uidna_close(uidna_openUTS46(0, &status)); // Fail quickly if no data. |
| 1317 | if (assertSuccess("", status, true, __FILE__, __LINE__)) { |
| 1318 | testRootLabelSeparator( "uidna_compare",uidna_compare, |
| 1319 | "uidna_IDNToASCII", uidna_IDNToASCII, |
| 1320 | "uidna_IDNToUnicode",uidna_IDNToUnicode |
| 1321 | ); |
| 1322 | } |
| 1323 | } |
| 1324 | void TestIDNA::TestChaining(){ |
| 1325 | testChaining("uidna_toASCII",uidna_toASCII, "uidna_toUnicode", uidna_toUnicode); |
| 1326 | } |
| 1327 | |
| 1328 | |
| 1329 | static const int loopCount = 100; |
| 1330 | static const int maxCharCount = 20; |
| 1331 | |
| 1332 | static uint32_t |
| 1333 | randul() |
| 1334 | { |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1335 | static UBool initialized = false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1336 | if (!initialized) |
| 1337 | { |
| 1338 | srand((unsigned)time(NULL)); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1339 | initialized = true; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1340 | } |
| 1341 | // Assume rand has at least 12 bits of precision |
| 1342 | uint32_t l = 0; |
| 1343 | for (uint32_t i=0; i<sizeof(l); ++i) |
| 1344 | ((char*)&l)[i] = (char)((rand() & 0x0FF0) >> 4); |
| 1345 | return l; |
| 1346 | } |
| 1347 | |
| 1348 | /** |
| 1349 | * Return a random integer i where 0 <= i < n. |
| 1350 | * A special function that gets random codepoints from planes 0,1,2 and 14 |
| 1351 | */ |
| 1352 | static int32_t rand_uni() |
| 1353 | { |
| 1354 | int32_t retVal = (int32_t)(randul()& 0x3FFFF); |
| 1355 | if(retVal >= 0x30000){ |
| 1356 | retVal+=0xB0000; |
| 1357 | } |
| 1358 | return retVal; |
| 1359 | } |
| 1360 | |
| 1361 | static int32_t randi(int32_t n){ |
| 1362 | return (int32_t) (randul() % (n+1)); |
| 1363 | } |
| 1364 | |
| 1365 | void getTestSource(UnicodeString& fillIn) { |
| 1366 | int32_t i = 0; |
| 1367 | int32_t charCount = (randi(maxCharCount) + 1); |
| 1368 | while (i <charCount ) { |
| 1369 | int32_t codepoint = rand_uni(); |
| 1370 | if(codepoint == 0x0000){ |
| 1371 | continue; |
| 1372 | } |
| 1373 | fillIn.append((UChar32)codepoint); |
| 1374 | i++; |
| 1375 | } |
| 1376 | |
| 1377 | } |
| 1378 | |
| 1379 | UnicodeString TestIDNA::testCompareReferenceImpl(UnicodeString& src, |
| 1380 | TestFunc refIDNA, const char* refIDNAName, |
| 1381 | TestFunc uIDNA, const char* uIDNAName, |
| 1382 | int32_t options){ |
| 1383 | |
| 1384 | const UChar* srcUChars = src.getBuffer(); |
| 1385 | UChar exp[MAX_DEST_SIZE]={0}; |
| 1386 | int32_t expCap = MAX_DEST_SIZE, expLen=0; |
| 1387 | UErrorCode expStatus = U_ZERO_ERROR; |
| 1388 | UParseError parseError; |
| 1389 | |
| 1390 | logln("Comparing "+ UnicodeString(refIDNAName) |
| 1391 | + " with "+ UnicodeString(uIDNAName) |
| 1392 | +" for input: " + prettify(srcUChars)); |
| 1393 | |
| 1394 | expLen = refIDNA(srcUChars, src.length()-1, exp, expCap, |
| 1395 | options, &parseError, &expStatus); |
| 1396 | |
| 1397 | UChar got[MAX_DEST_SIZE]={0}; |
| 1398 | int32_t gotCap = MAX_DEST_SIZE, gotLen=0; |
| 1399 | UErrorCode gotStatus = U_ZERO_ERROR; |
| 1400 | |
| 1401 | gotLen = uIDNA(srcUChars, src.length()-1, got, gotCap, |
| 1402 | options, &parseError, &gotStatus); |
| 1403 | |
| 1404 | if(expStatus != gotStatus){ |
| 1405 | errln("Did not get the expected status while comparing " + UnicodeString(refIDNAName) |
| 1406 | + " with " + UnicodeString(uIDNAName) |
| 1407 | + " Expected: " + UnicodeString(u_errorName(expStatus)) |
| 1408 | + " Got: " + UnicodeString(u_errorName(gotStatus)) |
| 1409 | + " for Source: "+ prettify(srcUChars) |
| 1410 | + " Options: " + options); |
| 1411 | return UnicodeString(""); |
| 1412 | } |
| 1413 | |
| 1414 | // now we know that both implementations yielded same error |
| 1415 | if(U_SUCCESS(expStatus)){ |
| 1416 | // compare the outputs if status == U_ZERO_ERROR |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1417 | if(u_strCompare(exp, expLen, got, gotLen, true) != 0){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1418 | errln("Did not get the expected output while comparing " + UnicodeString(refIDNAName) |
| 1419 | + " with " + UnicodeString(uIDNAName) |
| 1420 | + " Expected: " + prettify(UnicodeString(exp, expLen)) |
| 1421 | + " Got: " + prettify(UnicodeString(got, gotLen)) |
| 1422 | + " for Source: "+ prettify(srcUChars) |
| 1423 | + " Options: " + options); |
| 1424 | } |
| 1425 | return UnicodeString(exp, expLen); |
| 1426 | |
| 1427 | }else{ |
| 1428 | logln("Got the same error while comparing " |
| 1429 | + UnicodeString(refIDNAName) |
| 1430 | + " with "+ UnicodeString(uIDNAName) |
| 1431 | +" for input: " + prettify(srcUChars)); |
| 1432 | } |
| 1433 | return UnicodeString(""); |
| 1434 | } |
| 1435 | |
| 1436 | void TestIDNA::testCompareReferenceImpl(const UChar* src, int32_t srcLen){ |
| 1437 | UnicodeString label(src,srcLen); |
| 1438 | label.append((UChar)0x0000); |
| 1439 | |
| 1440 | //test idnaref_toASCII and idnare |
| 1441 | UnicodeString asciiLabel = testCompareReferenceImpl(label, |
| 1442 | idnaref_toASCII, "idnaref_toASCII", |
| 1443 | uidna_toASCII, "uidna_toASCII", |
| 1444 | UIDNA_ALLOW_UNASSIGNED); |
| 1445 | testCompareReferenceImpl(label, |
| 1446 | idnaref_toASCII, "idnaref_toASCII", |
| 1447 | uidna_toASCII, "uidna_toASCII", |
| 1448 | UIDNA_DEFAULT); |
| 1449 | testCompareReferenceImpl(label, |
| 1450 | idnaref_toASCII, "idnaref_toASCII", |
| 1451 | uidna_toASCII, "uidna_toASCII", |
| 1452 | UIDNA_USE_STD3_RULES); |
| 1453 | testCompareReferenceImpl(label, |
| 1454 | idnaref_toASCII, "idnaref_toASCII", |
| 1455 | uidna_toASCII, "uidna_toASCII", |
| 1456 | UIDNA_USE_STD3_RULES | UIDNA_ALLOW_UNASSIGNED); |
| 1457 | |
| 1458 | if(asciiLabel.length()!=0){ |
| 1459 | asciiLabel.append((UChar)0x0000); |
| 1460 | |
| 1461 | // test toUnciode |
| 1462 | testCompareReferenceImpl(asciiLabel, |
| 1463 | idnaref_toUnicode, "idnaref_toUnicode", |
| 1464 | uidna_toUnicode, "uidna_toUnicode", |
| 1465 | UIDNA_ALLOW_UNASSIGNED); |
| 1466 | testCompareReferenceImpl(asciiLabel, |
| 1467 | idnaref_toUnicode, "idnaref_toUnicode", |
| 1468 | uidna_toUnicode, "uidna_toUnicode", |
| 1469 | UIDNA_DEFAULT); |
| 1470 | testCompareReferenceImpl(asciiLabel, |
| 1471 | idnaref_toUnicode, "idnaref_toUnicode", |
| 1472 | uidna_toUnicode, "uidna_toUnicode", |
| 1473 | UIDNA_USE_STD3_RULES); |
| 1474 | testCompareReferenceImpl(asciiLabel, |
| 1475 | idnaref_toUnicode, "idnaref_toUnicode", |
| 1476 | uidna_toUnicode, "uidna_toUnicode", |
| 1477 | UIDNA_USE_STD3_RULES | UIDNA_ALLOW_UNASSIGNED); |
| 1478 | } |
| 1479 | |
| 1480 | } |
| 1481 | const char* failures[] ={ |
| 1482 | "\\uAA42\\U0001F8DD\\U00019D01\\U000149A3\\uD385\\U000EE0F5\\U00018B92\\U000179D1\\U00018624\\U0002227F\\U000E83C0\\U000E8DCD\\u5460\\U00017F34\\U0001570B\\u43D1\\U0002C9C9\\U000281EC\\u2105\\U000180AE\\uC5D4", |
| 1483 | "\\U0002F5A6\\uD638\\u0D0A\\u9E9C\\uFE5B\\U0001FCCB\\u66C4", |
| 1484 | }; |
| 1485 | |
| 1486 | void TestIDNA::TestIDNAMonkeyTest(){ |
| 1487 | UnicodeString source; |
| 1488 | UErrorCode status = U_ZERO_ERROR; |
| 1489 | int i; |
| 1490 | |
| 1491 | getInstance(status); // Init prep |
| 1492 | if (U_FAILURE(status)) { |
| 1493 | dataerrln("Test could not initialize. Got %s", u_errorName(status)); |
| 1494 | return; |
| 1495 | } |
| 1496 | |
| 1497 | for(i=0; i<loopCount; i++){ |
| 1498 | source.truncate(0); |
| 1499 | getTestSource(source); |
| 1500 | source.append((UChar)0x0000); |
| 1501 | const UChar* src = source.getBuffer(); |
| 1502 | testCompareReferenceImpl(src,source.length()-1); |
| 1503 | testCompareReferenceImpl(src,source.length()-1); |
| 1504 | } |
| 1505 | |
| 1506 | /* for debugging */ |
| 1507 | for (i=0; i<UPRV_LENGTHOF(failures); i++){ |
| 1508 | source.truncate(0); |
| 1509 | source.append( UnicodeString(failures[i], -1, US_INV) ); |
| 1510 | source = source.unescape(); |
| 1511 | source.append((UChar)0x0000); |
| 1512 | const UChar *src = source.getBuffer(); |
| 1513 | testCompareReferenceImpl(src,source.length()-1); |
| 1514 | //debug(source.getBuffer(),source.length(),UIDNA_ALLOW_UNASSIGNED); |
| 1515 | } |
| 1516 | |
| 1517 | |
| 1518 | source.truncate(0); |
| 1519 | source.append(UNICODE_STRING_SIMPLE("\\uCF18\\U00021161\\U000EEF11\\U0002BB82\\U0001D63C")); |
| 1520 | debug(source.getBuffer(),source.length(),UIDNA_ALLOW_UNASSIGNED); |
| 1521 | |
| 1522 | { // test deletion of code points |
| 1523 | UnicodeString source("\\u043f\\u00AD\\u034f\\u043e\\u0447\\u0435\\u043c\\u0443\\u0436\\u0435\\u043e\\u043d\\u0438\\u043d\\u0435\\u0433\\u043e\\u0432\\u043e\\u0440\\u044f\\u0442\\u043f\\u043e\\u0440\\u0443\\u0441\\u0441\\u043a\\u0438\\u0000", -1, US_INV); |
| 1524 | source = source.unescape(); |
| 1525 | UnicodeString expected("\\u043f\\u043e\\u0447\\u0435\\u043c\\u0443\\u0436\\u0435\\u043e\\u043d\\u0438\\u043d\\u0435\\u0433\\u043e\\u0432\\u043e\\u0440\\u044f\\u0442\\u043f\\u043e\\u0440\\u0443\\u0441\\u0441\\u043a\\u0438\\u0000", -1, US_INV); |
| 1526 | expected = expected.unescape(); |
| 1527 | UnicodeString ascii("xn--b1abfaaepdrnnbgefbadotcwatmq2g4l"); |
| 1528 | ascii.append((UChar)0x0000); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1529 | testAPI(source.getBuffer(),ascii.getBuffer(), "uidna_toASCII", false, U_ZERO_ERROR, true, true, uidna_toASCII); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1530 | |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1531 | testAPI(source.getBuffer(),ascii.getBuffer(), "idnaref_toASCII", false, U_ZERO_ERROR, true, true, idnaref_toASCII); |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1532 | |
| 1533 | testCompareReferenceImpl(source.getBuffer(), source.length()-1); |
| 1534 | } |
| 1535 | |
| 1536 | } |
| 1537 | |
| 1538 | void TestIDNA::TestCompareReferenceImpl(){ |
| 1539 | |
| 1540 | UChar src [2] = {0,0}; |
| 1541 | int32_t srcLen = 0; |
| 1542 | |
| 1543 | // data even OK? |
| 1544 | UErrorCode dataStatus = U_ZERO_ERROR; |
| 1545 | loadTestData(dataStatus); |
| 1546 | if(U_FAILURE(dataStatus)) { |
| 1547 | dataerrln("Couldn't load test data: %s\n", u_errorName(dataStatus)); // save us from thousands and thousands of errors |
| 1548 | return; |
| 1549 | } |
| 1550 | uidna_close(uidna_openUTS46(0, &dataStatus)); // Fail quickly if no data. |
| 1551 | if (!assertSuccess("", dataStatus, true, __FILE__, __LINE__)) { return; } |
| 1552 | |
| 1553 | for (int32_t i = 0; i <= 0x10FFFF; i++){ |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 1554 | if (quick == true && i > 0x0FFF){ |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 1555 | return; |
| 1556 | } |
| 1557 | if(i == 0x30000){ |
| 1558 | // jump to E0000, no characters assigned in plain 3 to plain 13 as of Unicode 6.0 |
| 1559 | i = 0xE0000; |
| 1560 | } |
| 1561 | if (i > 0xFFFF){ |
| 1562 | src[0] = U16_LEAD(i); |
| 1563 | src[1] = U16_TRAIL(i); |
| 1564 | srcLen =2; |
| 1565 | } else { |
| 1566 | src[0] = (UChar)i; |
| 1567 | src[1] = 0; |
| 1568 | srcLen = 1; |
| 1569 | } |
| 1570 | testCompareReferenceImpl(src, srcLen); |
| 1571 | } |
| 1572 | } |
| 1573 | |
| 1574 | void TestIDNA::TestRefIDNA(){ |
| 1575 | UErrorCode status = U_ZERO_ERROR; |
| 1576 | |
| 1577 | getInstance(status); // Init prep. Abort test early if no data. |
| 1578 | if (!assertSuccess("", status, true, __FILE__, __LINE__)) { return; } |
| 1579 | |
| 1580 | testToASCII("idnaref_toASCII", idnaref_toASCII); |
| 1581 | testToUnicode("idnaref_toUnicode", idnaref_toUnicode); |
| 1582 | testIDNToASCII("idnaref_IDNToASCII", idnaref_IDNToASCII); |
| 1583 | testIDNToUnicode("idnaref_IDNToUnicode", idnaref_IDNToUnicode); |
| 1584 | testCompare("idnaref_compare",idnaref_compare); |
| 1585 | testErrorCases( "idnaref_IDNToASCII",idnaref_IDNToASCII, |
| 1586 | "idnaref_IDNToUnicode",idnaref_IDNToUnicode); |
| 1587 | testChaining("idnaref_toASCII",idnaref_toASCII, "idnaref_toUnicode", idnaref_toUnicode); |
| 1588 | |
| 1589 | testRootLabelSeparator( "idnaref_compare",idnaref_compare, |
| 1590 | "idnaref_IDNToASCII", idnaref_IDNToASCII, |
| 1591 | "idnaref_IDNToUnicode",idnaref_IDNToUnicode |
| 1592 | ); |
| 1593 | testChaining("idnaref_toASCII",idnaref_toASCII, "idnaref_toUnicode", idnaref_toUnicode); |
| 1594 | } |
| 1595 | |
| 1596 | |
| 1597 | void TestIDNA::TestDataFile(){ |
| 1598 | testData(*this); |
| 1599 | } |
| 1600 | |
| 1601 | TestIDNA::~TestIDNA(){ |
| 1602 | delete gPrep; |
| 1603 | gPrep = NULL; |
| 1604 | } |
| 1605 | |
| 1606 | NamePrepTransform* TestIDNA::getInstance(UErrorCode& status){ |
| 1607 | if(gPrep == NULL){ |
| 1608 | UParseError parseError; |
| 1609 | gPrep = NamePrepTransform::createInstance(parseError, status); |
| 1610 | if(gPrep == NULL){ |
| 1611 | //status = U_MEMORY_ALLOCATION_ERROR; |
| 1612 | return NULL; |
| 1613 | } |
| 1614 | } |
| 1615 | return gPrep; |
| 1616 | |
| 1617 | } |
| 1618 | #endif /* #if !UCONFIG_NO_IDNA */ |