1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/netwerk/test/unit/test_idn_urls.js Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,337 @@
1.4 +// Test algorithm for unicode display of IDNA URL (bug 722299)
1.5 +const testcases = [
1.6 + // Original Punycode or Expected UTF-8 by profile
1.7 + // URL normalized form ASCII-Only, High, Moderate
1.8 + //
1.9 + // Latin script
1.10 + ["cuillère", "xn--cuillre-6xa", false, true, true],
1.11 +
1.12 + // repeated non-spacing marks
1.13 + ["gruz̀̀ere", "xn--gruzere-ogea", false, false, false],
1.14 +
1.15 + // non-XID character
1.16 + ["I♥NY", "xn--iny-zx5a", false, false, false],
1.17 +
1.18 + // new non-XID character in Unicode 6.3
1.19 + ["حلا\u061cل", "xn--bgbvr6gc", false, false, false],
1.20 +
1.21 + // U+30FB KATAKANA MIDDLE DOT is excluded from non-XID characters (bug 857490)
1.22 + ["乾燥肌・石けん", "xn--08j4gylj12hz80b0uhfup", false, true, true],
1.23 +
1.24 + // Cyrillic alone
1.25 + ["толсто́й", "xn--lsa83dealbred", false, true, true],
1.26 +
1.27 + // Mixed script Cyrillic/Latin
1.28 + ["толсто́й-in-Russian",
1.29 + "xn---in-russian-1jg071b0a8bb4cpd", false, false, false],
1.30 +
1.31 + // Mixed script Latin/Cyrillic
1.32 + ["war-and-миръ", "xn--war-and--b9g3b7b3h", false, false, false],
1.33 +
1.34 + // Cherokee (Restricted script)
1.35 + ["ᏣᎳᎩ", "xn--f9dt7l", false, false, false],
1.36 +
1.37 + // Yi (Aspirational script)
1.38 + ["ꆈꌠꁱꂷ", "xn--4o7a6e1x64c", false, true, true],
1.39 +
1.40 + // Greek alone
1.41 + ["πλάτων", "xn--hxa3ahjw4a", false, true, true],
1.42 +
1.43 + // Mixed script Greek/Latin
1.44 + ["πλάτωνicrelationship",
1.45 + "xn--icrelationship-96j4t9a3cwe2e", false, false, false],
1.46 +
1.47 + // Mixed script Latin/Greek
1.48 + ["spaceὈδύσσεια", "xn--space-h9dui0b0ga2j1562b", false, false, false],
1.49 +
1.50 + // Devanagari alone
1.51 + ["मराठी", "xn--d2b1ag0dl", false, true, true],
1.52 +
1.53 + // Devanagari with Armenian
1.54 + ["मराठीՀայաստան",
1.55 + "xn--y9aaa1d0ai1cq964f8dwa2o1a", false, false, false],
1.56 +
1.57 + // Devanagari with common
1.58 + ["मराठी123", "xn--123-mhh3em2hra", false, true, true],
1.59 +
1.60 + // Common with Devanagari
1.61 + ["123मराठी", "xn--123-phh3em2hra", false, true, true],
1.62 +
1.63 + // Latin with Han
1.64 + ["chairman毛",
1.65 + "xn--chairman-k65r", false, true, true],
1.66 +
1.67 + // Han with Latin
1.68 + ["山葵sauce", "xn--sauce-6j9ii40v", false, true, true],
1.69 +
1.70 + // Latin with Han, Hiragana and Katakana
1.71 + ["van語ではドイ", "xn--van-ub4bpb6w0in486d", false, true, true],
1.72 +
1.73 + // Latin with Han, Katakana and Hiragana
1.74 + ["van語ドイでは", "xn--van-ub4bpb4w0ip486d", false, true, true],
1.75 +
1.76 + // Latin with Hiragana, Han and Katakana
1.77 + ["vanでは語ドイ", "xn--van-ub4bpb6w0ip486d", false, true, true],
1.78 +
1.79 + // Latin with Hiragana, Katakana and Han
1.80 + ["vanではドイ語", "xn--van-ub4bpb6w0ir486d", false, true, true],
1.81 +
1.82 + // Latin with Katakana, Han and Hiragana
1.83 + ["vanドイ語では", "xn--van-ub4bpb4w0ir486d", false, true, true],
1.84 +
1.85 + // Latin with Katakana, Hiragana and Han
1.86 + ["vanドイでは語", "xn--van-ub4bpb4w0it486d", false, true, true],
1.87 +
1.88 + // Han with Latin, Hiragana and Katakana
1.89 + ["語vanではドイ", "xn--van-ub4bpb6w0ik486d", false, true, true],
1.90 +
1.91 + // Han with Latin, Katakana and Hiragana
1.92 + ["語vanドイでは", "xn--van-ub4bpb4w0im486d", false, true, true],
1.93 +
1.94 + // Han with Hiragana, Latin and Katakana
1.95 + ["語ではvanドイ", "xn--van-rb4bpb9w0ik486d", false, true, true],
1.96 +
1.97 + // Han with Hiragana, Katakana and Latin
1.98 + ["語ではドイvan", "xn--van-rb4bpb6w0in486d", false, true, true],
1.99 +
1.100 + // Han with Katakana, Latin and Hiragana
1.101 + ["語ドイvanでは", "xn--van-ub4bpb1w0ip486d", false, true, true],
1.102 +
1.103 + // Han with Katakana, Hiragana and Latin
1.104 + ["語ドイではvan", "xn--van-rb4bpb4w0ip486d", false, true, true],
1.105 +
1.106 + // Hiragana with Latin, Han and Katakana
1.107 + ["イツvan語ではド", "xn--van-ub4bpb1wvhsbx330n", false, true, true],
1.108 +
1.109 + // Hiragana with Latin, Katakana and Han
1.110 + ["ではvanドイ語", "xn--van-rb4bpb9w0ir486d", false, true, true],
1.111 +
1.112 + // Hiragana with Han, Latin and Katakana
1.113 + ["では語vanドイ", "xn--van-rb4bpb9w0im486d", false, true, true],
1.114 +
1.115 + // Hiragana with Han, Katakana and Latin
1.116 + ["では語ドイvan", "xn--van-rb4bpb6w0ip486d", false, true, true],
1.117 +
1.118 + // Hiragana with Katakana, Latin and Han
1.119 + ["ではドイvan語", "xn--van-rb4bpb6w0iu486d", false, true, true],
1.120 +
1.121 + // Hiragana with Katakana, Han and Latin
1.122 + ["ではドイ語van", "xn--van-rb4bpb6w0ir486d", false, true, true],
1.123 +
1.124 + // Katakana with Latin, Han and Hiragana
1.125 + ["ドイvan語では", "xn--van-ub4bpb1w0iu486d", false, true, true],
1.126 +
1.127 + // Katakana with Latin, Hiragana and Han
1.128 + ["ドイvanでは語", "xn--van-ub4bpb1w0iw486d", false, true, true],
1.129 +
1.130 + // Katakana with Han, Latin and Hiragana
1.131 + ["ドイ語vanでは", "xn--van-ub4bpb1w0ir486d", false, true, true],
1.132 +
1.133 + // Katakana with Han, Hiragana and Latin
1.134 + ["ドイ語ではvan", "xn--van-rb4bpb4w0ir486d", false, true, true],
1.135 +
1.136 + // Katakana with Hiragana, Latin and Han
1.137 + ["ドイではvan語", "xn--van-rb4bpb4w0iw486d", false, true, true],
1.138 +
1.139 + // Katakana with Hiragana, Han and Latin
1.140 + ["ドイでは語van", "xn--van-rb4bpb4w0it486d", false, true, true],
1.141 +
1.142 + // Han with common
1.143 + ["中国123", "xn--123-u68dy61b", false, true, true],
1.144 +
1.145 + // common with Han
1.146 + ["123中国", "xn--123-x68dy61b", false, true, true],
1.147 +
1.148 + // Characters that normalize to permitted characters
1.149 + // (also tests Plane 1 supplementary characters)
1.150 + ["super𝟖", "super8", true, true, true],
1.151 +
1.152 + // Han from Plane 2
1.153 + ["𠀀𠀁𠀂", "xn--j50icd", false, true, true],
1.154 +
1.155 + // Han from Plane 2 with js (UTF-16) escapes
1.156 + ["\uD840\uDC00\uD840\uDC01\uD840\uDC02",
1.157 + "xn--j50icd", false, true, true],
1.158 +
1.159 + // Same with a lone high surrogate at the end
1.160 + ["\uD840\uDC00\uD840\uDC01\uD840", "", false, false, false],
1.161 +
1.162 + // Latin text and Bengali digits
1.163 + ["super৪", "xn--super-k2l", false, false, true],
1.164 +
1.165 + // Bengali digits and Latin text
1.166 + ["৫ab", "xn--ab-x5f", false, false, true],
1.167 +
1.168 + // Bengali text and Latin digits
1.169 + ["অঙ্কুর8", "xn--8-70d2cp0j6dtd", false, true, true],
1.170 +
1.171 + // Latin digits and Bengali text
1.172 + ["5াব", "xn--5-h3d7c", false, true, true],
1.173 +
1.174 + // Mixed numbering systems
1.175 + ["٢٠۰٠", "xn--8hbae38c", false, false, false],
1.176 +
1.177 + // Traditional Chinese
1.178 + ["萬城", "xn--uis754h", false, true, true],
1.179 +
1.180 + // Simplified Chinese
1.181 + ["万城", "xn--chq31v", false, true, true],
1.182 +
1.183 + // Simplified-only and Traditional-only Chinese in the same label
1.184 + ["万萬城", "xn--chq31vsl1b", false, true, true],
1.185 +
1.186 + // Traditional-only and Simplified-only Chinese in the same label
1.187 + ["萬万城", "xn--chq31vrl1b", false, true, true],
1.188 +
1.189 + // Han and Latin and Bopomofo
1.190 + ["注音符号bopomofoㄅㄆㄇㄈ",
1.191 + "xn--bopomofo-hj5gkalm1637i876cuw0brk5f",
1.192 + false, true, true],
1.193 +
1.194 + // Han, bopomofo, Latin
1.195 + ["注音符号ㄅㄆㄇㄈbopomofo",
1.196 + "xn--bopomofo-8i5gkalm9637i876cuw0brk5f",
1.197 + false, true, true],
1.198 +
1.199 + // Latin, Han, Bopomofo
1.200 + ["bopomofo注音符号ㄅㄆㄇㄈ",
1.201 + "xn--bopomofo-hj5gkalm9637i876cuw0brk5f",
1.202 + false, true, true],
1.203 +
1.204 + // Latin, Bopomofo, Han
1.205 + ["bopomofoㄅㄆㄇㄈ注音符号",
1.206 + "xn--bopomofo-hj5gkalm3737i876cuw0brk5f",
1.207 + false, true, true],
1.208 +
1.209 + // Bopomofo, Han, Latin
1.210 + ["ㄅㄆㄇㄈ注音符号bopomofo",
1.211 + "xn--bopomofo-8i5gkalm3737i876cuw0brk5f",
1.212 + false, true, true],
1.213 +
1.214 + // Bopomofo, Latin, Han
1.215 + ["ㄅㄆㄇㄈbopomofo注音符号",
1.216 + "xn--bopomofo-8i5gkalm1837i876cuw0brk5f",
1.217 + false, true, true],
1.218 +
1.219 + // Han, bopomofo and katakana
1.220 + ["注音符号ㄅㄆㄇㄈボポモフォ",
1.221 + "xn--jckteuaez1shij0450gylvccz9asi4e",
1.222 + false, false, false],
1.223 +
1.224 + // Han, katakana, bopomofo
1.225 + ["注音符号ボポモフォㄅㄆㄇㄈ",
1.226 + "xn--jckteuaez6shij5350gylvccz9asi4e",
1.227 + false, false, false],
1.228 +
1.229 + // bopomofo, han, katakana
1.230 + ["ㄅㄆㄇㄈ注音符号ボポモフォ",
1.231 + "xn--jckteuaez1shij4450gylvccz9asi4e",
1.232 + false, false, false],
1.233 +
1.234 + // bopomofo, katakana, han
1.235 + ["ㄅㄆㄇㄈボポモフォ注音符号",
1.236 + "xn--jckteuaez1shij9450gylvccz9asi4e",
1.237 + false, false, false],
1.238 +
1.239 + // katakana, Han, bopomofo
1.240 + ["ボポモフォ注音符号ㄅㄆㄇㄈ",
1.241 + "xn--jckteuaez6shij0450gylvccz9asi4e",
1.242 + false, false, false],
1.243 +
1.244 + // katakana, bopomofo, Han
1.245 + ["ボポモフォㄅㄆㄇㄈ注音符号",
1.246 + "xn--jckteuaez6shij4450gylvccz9asi4e",
1.247 + false, false, false],
1.248 +
1.249 + // Han, Hangul and Latin
1.250 + ["韓한글hangul",
1.251 + "xn--hangul-2m5ti09k79ze", false, true, true],
1.252 +
1.253 + // Han, Latin and Hangul
1.254 + ["韓hangul한글",
1.255 + "xn--hangul-2m5to09k79ze", false, true, true],
1.256 +
1.257 + // Hangul, Han and Latin
1.258 + ["한글韓hangul",
1.259 + "xn--hangul-2m5th09k79ze", false, true, true],
1.260 +
1.261 + // Hangul, Latin and Han
1.262 + ["한글hangul韓",
1.263 + "xn--hangul-8m5t898k79ze", false, true, true],
1.264 +
1.265 + // Latin, Han and Hangul
1.266 + ["hangul韓한글",
1.267 + "xn--hangul-8m5ti09k79ze", false, true, true],
1.268 +
1.269 + // Latin, Hangul and Han
1.270 + ["hangul한글韓",
1.271 + "xn--hangul-8m5th09k79ze", false, true, true],
1.272 +
1.273 + // Hangul and katakana
1.274 + ["한글ハングル",
1.275 + "xn--qck1c2d4a9266lkmzb", false, false, false],
1.276 +
1.277 + // Katakana and Hangul
1.278 + ["ハングル한글",
1.279 + "xn--qck1c2d4a2366lkmzb", false, false, false],
1.280 +
1.281 + // Thai (also tests that node with over 63 UTF-8 octets doesn't fail)
1.282 + ["เครื่องทําน้ําทําน้ําแข็ง",
1.283 + "xn--22cdjb2fanb9fyepcbbb9dwh4a3igze4fdcd",
1.284 + false, true, true]
1.285 +];
1.286 +
1.287 +
1.288 +const profiles = ["ASCII", "high", "moderate"];
1.289 +
1.290 +function run_test() {
1.291 + var pbi = Cc["@mozilla.org/preferences-service;1"].getService(Ci.nsIPrefBranch);
1.292 + var oldProfile = pbi.getCharPref("network.IDN.restriction_profile", "moderate");
1.293 + var oldWhiteListCom;
1.294 + try {
1.295 + oldWhitelistCom = pbi.getBoolPref("network.IDN.whitelist.com");
1.296 + } catch(e) {
1.297 + oldWhitelistCom = false;
1.298 + }
1.299 + var idnService = Cc["@mozilla.org/network/idn-service;1"].getService(Ci.nsIIDNService);
1.300 +
1.301 + for (var i = 0; i < profiles.length; ++i) {
1.302 + pbi.setCharPref("network.IDN.restriction_profile", profiles[i]);
1.303 + pbi.setBoolPref("network.IDN.whitelist.com", false);
1.304 +
1.305 + dump("testing " + profiles[i] + " profile");
1.306 +
1.307 + for (var j = 0; j < testcases.length; ++j) {
1.308 + var test = testcases[j];
1.309 + var URL = test[0] + ".com";
1.310 + var punycodeURL = test[1] + ".com";
1.311 + var expectedUnicode = test[2 + i];
1.312 + var isASCII = {};
1.313 +
1.314 + var result;
1.315 + try {
1.316 + result = idnService.convertToDisplayIDN(URL, isASCII);
1.317 + } catch(e) {
1.318 + result = ".com";
1.319 + }
1.320 + if (punycodeURL.substr(0, 4) == "xn--") {
1.321 + // test convertToDisplayIDN with a Unicode URL and with a
1.322 + // Punycode URL if we have one
1.323 + do_check_eq(escape(result),
1.324 + expectedUnicode ? escape(URL) : escape(punycodeURL));
1.325 +
1.326 + result = idnService.convertToDisplayIDN(punycodeURL, isASCII);
1.327 + do_check_eq(escape(result),
1.328 + expectedUnicode ? escape(URL) : escape(punycodeURL));
1.329 + } else {
1.330 + // The "punycode" URL isn't punycode. This happens in testcases
1.331 + // where the Unicode URL has become normalized to an ASCII URL,
1.332 + // so, even though expectedUnicode is true, the expected result
1.333 + // is equal to punycodeURL
1.334 + do_check_eq(escape(result), escape(punycodeURL));
1.335 + }
1.336 + }
1.337 + }
1.338 + pbi.setBoolPref("network.IDN.whitelist.com", oldWhitelistCom);
1.339 + pbi.setCharPref("network.IDN.restriction_profile", oldProfile);
1.340 +}
