michael@0: // Test algorithm for unicode display of IDNA URL (bug 722299)
michael@0: const testcases = [
michael@0:     //  Original             Punycode or         Expected UTF-8 by profile
michael@0:     //    URL              normalized form      ASCII-Only, High, Moderate
michael@0:     //
michael@0:     // Latin script
michael@0:     ["cuillère", "xn--cuillre-6xa",                  false, true,  true],
michael@0: 
michael@0:     // repeated non-spacing marks
michael@0:     ["gruz̀̀ere",  "xn--gruzere-ogea",                 false, false, false],
michael@0: 
michael@0:     // non-XID character
michael@0:     ["I♥NY",     "xn--iny-zx5a",                     false, false, false],
michael@0: 
michael@0:     // new non-XID character in Unicode 6.3
michael@0:     ["حلا\u061cل", "xn--bgbvr6gc",                    false, false, false],
michael@0: 
michael@0:     // U+30FB KATAKANA MIDDLE DOT is excluded from non-XID characters (bug 857490)
michael@0:     ["乾燥肌・石けん", "xn--08j4gylj12hz80b0uhfup",     false, true,  true],
michael@0: 
michael@0:     // Cyrillic alone
michael@0:     ["толсто́й",  "xn--lsa83dealbred",                false, true,  true],
michael@0: 
michael@0:     // Mixed script Cyrillic/Latin
michael@0:     ["толсто́й-in-Russian",
michael@0:                  "xn---in-russian-1jg071b0a8bb4cpd", false, false, false],
michael@0: 
michael@0:     // Mixed script Latin/Cyrillic
michael@0:     ["war-and-миръ", "xn--war-and--b9g3b7b3h",       false, false, false],
michael@0: 
michael@0:     // Cherokee (Restricted script)
michael@0:     ["ᏣᎳᎩ",     "xn--f9dt7l",                        false, false, false],
michael@0: 
michael@0:     // Yi (Aspirational script)
michael@0:     ["ꆈꌠꁱꂷ", "xn--4o7a6e1x64c",                  false, true,  true],
michael@0: 
michael@0:     // Greek alone
michael@0:     ["πλάτων",   "xn--hxa3ahjw4a",                   false, true,  true],
michael@0: 
michael@0:     // Mixed script Greek/Latin
michael@0:     ["πλάτωνicrelationship",
michael@0:                  "xn--icrelationship-96j4t9a3cwe2e", false, false, false],
michael@0: 
michael@0:     // Mixed script Latin/Greek
michael@0:     ["spaceὈδύσσεια", "xn--space-h9dui0b0ga2j1562b", false, false, false],
michael@0: 
michael@0:     // Devanagari alone
michael@0:     ["मराठी",    "xn--d2b1ag0dl",                    false, true,  true],
michael@0: 
michael@0:     // Devanagari with Armenian
michael@0:     ["मराठीՀայաստան",
michael@0:                  "xn--y9aaa1d0ai1cq964f8dwa2o1a",    false, false, false],
michael@0: 
michael@0:     // Devanagari with common
michael@0:     ["मराठी123", "xn--123-mhh3em2hra",               false, true,  true],
michael@0: 
michael@0:     // Common with Devanagari
michael@0:     ["123मराठी", "xn--123-phh3em2hra",               false, true,  true],
michael@0: 
michael@0:     // Latin with Han
michael@0:     ["chairman毛",
michael@0:                  "xn--chairman-k65r",                false, true,  true],
michael@0: 
michael@0:     // Han with Latin
michael@0:     ["山葵sauce", "xn--sauce-6j9ii40v",              false, true,  true],
michael@0: 
michael@0:     // Latin with Han, Hiragana and Katakana
michael@0:     ["van語ではドイ", "xn--van-ub4bpb6w0in486d",     false, true,  true],
michael@0: 
michael@0:     // Latin with Han, Katakana and Hiragana
michael@0:     ["van語ドイでは", "xn--van-ub4bpb4w0ip486d",     false, true,  true],
michael@0: 
michael@0:     // Latin with Hiragana, Han and Katakana
michael@0:     ["vanでは語ドイ", "xn--van-ub4bpb6w0ip486d",     false, true,  true],
michael@0: 
michael@0:     // Latin with Hiragana, Katakana and Han
michael@0:     ["vanではドイ語", "xn--van-ub4bpb6w0ir486d",     false, true,  true],
michael@0: 
michael@0:     // Latin with Katakana, Han and Hiragana
michael@0:     ["vanドイ語では", "xn--van-ub4bpb4w0ir486d",     false, true,  true],
michael@0: 
michael@0:     // Latin with Katakana, Hiragana and Han
michael@0:     ["vanドイでは語", "xn--van-ub4bpb4w0it486d",     false, true,  true],
michael@0: 
michael@0:     // Han with Latin, Hiragana and Katakana
michael@0:     ["語vanではドイ", "xn--van-ub4bpb6w0ik486d",     false, true,  true],
michael@0: 
michael@0:     // Han with Latin, Katakana and Hiragana
michael@0:     ["語vanドイでは", "xn--van-ub4bpb4w0im486d",     false, true,  true],
michael@0: 
michael@0:     // Han with Hiragana, Latin and Katakana
michael@0:     ["語ではvanドイ", "xn--van-rb4bpb9w0ik486d",     false, true,  true],
michael@0: 
michael@0:     // Han with Hiragana, Katakana and Latin
michael@0:     ["語ではドイvan", "xn--van-rb4bpb6w0in486d",     false, true,  true],
michael@0: 
michael@0:     // Han with Katakana, Latin and Hiragana
michael@0:     ["語ドイvanでは", "xn--van-ub4bpb1w0ip486d",     false, true,  true],
michael@0: 
michael@0:     // Han with Katakana, Hiragana and Latin
michael@0:     ["語ドイではvan", "xn--van-rb4bpb4w0ip486d",     false, true,  true],
michael@0: 
michael@0:     // Hiragana with Latin, Han and Katakana
michael@0:     ["イツvan語ではド", "xn--van-ub4bpb1wvhsbx330n", false, true,  true],
michael@0: 
michael@0:     // Hiragana with Latin, Katakana and Han
michael@0:     ["ではvanドイ語", "xn--van-rb4bpb9w0ir486d",     false, true,  true],
michael@0: 
michael@0:     // Hiragana with Han, Latin and Katakana
michael@0:     ["では語vanドイ", "xn--van-rb4bpb9w0im486d",     false, true,  true],
michael@0: 
michael@0:     // Hiragana with Han, Katakana and Latin
michael@0:     ["では語ドイvan", "xn--van-rb4bpb6w0ip486d",     false, true,  true],
michael@0: 
michael@0:     // Hiragana with Katakana, Latin and Han
michael@0:     ["ではドイvan語", "xn--van-rb4bpb6w0iu486d",     false, true,  true],
michael@0: 
michael@0:     // Hiragana with Katakana, Han and Latin
michael@0:     ["ではドイ語van", "xn--van-rb4bpb6w0ir486d",     false, true,  true],
michael@0: 
michael@0:     // Katakana with Latin, Han and Hiragana
michael@0:     ["ドイvan語では", "xn--van-ub4bpb1w0iu486d",     false, true,  true],
michael@0: 
michael@0:     // Katakana with Latin, Hiragana and Han
michael@0:     ["ドイvanでは語", "xn--van-ub4bpb1w0iw486d",     false, true,  true],
michael@0: 
michael@0:     // Katakana with Han, Latin and Hiragana
michael@0:     ["ドイ語vanでは", "xn--van-ub4bpb1w0ir486d",     false, true,  true],
michael@0: 
michael@0:     // Katakana with Han, Hiragana and Latin
michael@0:     ["ドイ語ではvan", "xn--van-rb4bpb4w0ir486d",     false, true,  true],
michael@0: 
michael@0:     // Katakana with Hiragana, Latin and Han
michael@0:     ["ドイではvan語", "xn--van-rb4bpb4w0iw486d",     false, true,  true],
michael@0: 
michael@0:     // Katakana with Hiragana, Han and Latin
michael@0:     ["ドイでは語van", "xn--van-rb4bpb4w0it486d",     false, true,  true],
michael@0: 
michael@0:     // Han with common
michael@0:     ["中国123",   "xn--123-u68dy61b",                false, true,  true],
michael@0: 
michael@0:     // common with Han
michael@0:     ["123中国",   "xn--123-x68dy61b",                false, true,  true],
michael@0: 
michael@0:     // Characters that normalize to permitted characters
michael@0:     //  (also tests Plane 1 supplementary characters)
michael@0:     ["super𝟖",   "super8",                           true,  true,  true],
michael@0: 
michael@0:     // Han from Plane 2
michael@0:     ["𠀀𠀁𠀂", "xn--j50icd",                         false, true,  true],
michael@0: 
michael@0:     // Han from Plane 2 with js (UTF-16) escapes
michael@0:     ["\uD840\uDC00\uD840\uDC01\uD840\uDC02",
michael@0:             "xn--j50icd",                            false, true,  true],
michael@0: 
michael@0:     // Same with a lone high surrogate at the end
michael@0:     ["\uD840\uDC00\uD840\uDC01\uD840", "",           false, false, false],
michael@0: 
michael@0:     // Latin text and Bengali digits
michael@0:     ["super৪",   "xn--super-k2l",                    false, false, true],
michael@0: 
michael@0:     // Bengali digits and Latin text
michael@0:     ["৫ab",   "xn--ab-x5f",                          false, false, true],
michael@0: 
michael@0:     // Bengali text and Latin digits
michael@0:     ["অঙ্কুর8",    "xn--8-70d2cp0j6dtd",               false, true,  true],
michael@0: 
michael@0:     // Latin digits and Bengali text
michael@0:     ["5াব",        "xn--5-h3d7c",                    false, true,  true],
michael@0: 
michael@0:     // Mixed numbering systems
michael@0:     ["٢٠۰٠",     "xn--8hbae38c",                     false, false, false],
michael@0: 
michael@0:     // Traditional Chinese
michael@0:     ["萬城",     "xn--uis754h",                      false, true,  true],
michael@0: 
michael@0:     // Simplified Chinese
michael@0:     ["万城",     "xn--chq31v",                       false, true,  true],
michael@0: 
michael@0:     // Simplified-only and Traditional-only Chinese in the same label
michael@0:     ["万萬城",   "xn--chq31vsl1b",                   false, true,  true],
michael@0: 
michael@0:     // Traditional-only and Simplified-only Chinese in the same label
michael@0:     ["萬万城",   "xn--chq31vrl1b",                   false, true,  true],
michael@0: 
michael@0:     // Han and Latin and Bopomofo
michael@0:     ["注音符号bopomofoㄅㄆㄇㄈ",
michael@0:                  "xn--bopomofo-hj5gkalm1637i876cuw0brk5f",
michael@0:                                                      false, true,  true],
michael@0: 
michael@0:     // Han, bopomofo, Latin
michael@0:     ["注音符号ㄅㄆㄇㄈbopomofo",
michael@0:                  "xn--bopomofo-8i5gkalm9637i876cuw0brk5f",
michael@0:                                                      false, true,  true],
michael@0: 
michael@0:     // Latin, Han, Bopomofo
michael@0:     ["bopomofo注音符号ㄅㄆㄇㄈ",
michael@0:                  "xn--bopomofo-hj5gkalm9637i876cuw0brk5f",
michael@0:                                                      false, true,  true],
michael@0: 
michael@0:     // Latin, Bopomofo, Han
michael@0:     ["bopomofoㄅㄆㄇㄈ注音符号",
michael@0:                  "xn--bopomofo-hj5gkalm3737i876cuw0brk5f",
michael@0:                                                      false, true,  true],
michael@0: 
michael@0:     // Bopomofo, Han, Latin
michael@0:     ["ㄅㄆㄇㄈ注音符号bopomofo",
michael@0:                  "xn--bopomofo-8i5gkalm3737i876cuw0brk5f",
michael@0:                                                      false, true,  true],
michael@0: 
michael@0:     // Bopomofo, Latin, Han
michael@0:     ["ㄅㄆㄇㄈbopomofo注音符号",
michael@0:                  "xn--bopomofo-8i5gkalm1837i876cuw0brk5f",
michael@0:                                                      false, true,  true],
michael@0: 
michael@0:     // Han, bopomofo and katakana
michael@0:     ["注音符号ㄅㄆㄇㄈボポモフォ",
michael@0:                  "xn--jckteuaez1shij0450gylvccz9asi4e",
michael@0:                                                      false, false, false],
michael@0: 
michael@0:     // Han, katakana, bopomofo
michael@0:     ["注音符号ボポモフォㄅㄆㄇㄈ",
michael@0:                  "xn--jckteuaez6shij5350gylvccz9asi4e",
michael@0:                                                      false, false, false],
michael@0: 
michael@0:     // bopomofo, han, katakana
michael@0:     ["ㄅㄆㄇㄈ注音符号ボポモフォ",
michael@0:                  "xn--jckteuaez1shij4450gylvccz9asi4e",
michael@0:                                                      false, false, false],
michael@0: 
michael@0:     // bopomofo, katakana, han
michael@0:     ["ㄅㄆㄇㄈボポモフォ注音符号",
michael@0:                  "xn--jckteuaez1shij9450gylvccz9asi4e",
michael@0:                                                      false, false, false],
michael@0: 
michael@0:     // katakana, Han, bopomofo
michael@0:     ["ボポモフォ注音符号ㄅㄆㄇㄈ",
michael@0:                  "xn--jckteuaez6shij0450gylvccz9asi4e",
michael@0:                                                      false, false, false],
michael@0: 
michael@0:     // katakana, bopomofo, Han
michael@0:     ["ボポモフォㄅㄆㄇㄈ注音符号",
michael@0:                  "xn--jckteuaez6shij4450gylvccz9asi4e",
michael@0:                                                      false, false, false],
michael@0: 
michael@0:     // Han, Hangul and Latin
michael@0:     ["韓한글hangul",
michael@0:                  "xn--hangul-2m5ti09k79ze",          false, true,  true],
michael@0: 
michael@0:     // Han, Latin and Hangul
michael@0:     ["韓hangul한글",
michael@0:                  "xn--hangul-2m5to09k79ze",          false, true,  true],
michael@0: 
michael@0:     // Hangul, Han and Latin
michael@0:     ["한글韓hangul",
michael@0:                  "xn--hangul-2m5th09k79ze",          false, true,  true],
michael@0: 
michael@0:     // Hangul, Latin and Han
michael@0:     ["한글hangul韓",
michael@0:                  "xn--hangul-8m5t898k79ze",          false, true,  true],
michael@0: 
michael@0:     // Latin, Han and Hangul
michael@0:     ["hangul韓한글",
michael@0:                  "xn--hangul-8m5ti09k79ze",          false, true,  true],
michael@0: 
michael@0:     // Latin, Hangul and Han
michael@0:     ["hangul한글韓",
michael@0:                  "xn--hangul-8m5th09k79ze",          false, true,  true],
michael@0: 
michael@0:     // Hangul and katakana
michael@0:     ["한글ハングル",
michael@0:                  "xn--qck1c2d4a9266lkmzb",           false, false, false],
michael@0: 
michael@0:     // Katakana and Hangul
michael@0:     ["ハングル한글",
michael@0:                  "xn--qck1c2d4a2366lkmzb",           false, false, false],
michael@0: 
michael@0:     // Thai (also tests that node with over 63 UTF-8 octets doesn't fail)
michael@0:     ["เครื่องทําน้ําทําน้ําแข็ง",
michael@0:                  "xn--22cdjb2fanb9fyepcbbb9dwh4a3igze4fdcd",
michael@0:                                                      false, true, true]
michael@0: ];
michael@0: 
michael@0: 
michael@0: const profiles = ["ASCII", "high", "moderate"];
michael@0: 
michael@0: function run_test() {
michael@0:     var pbi = Cc["@mozilla.org/preferences-service;1"].getService(Ci.nsIPrefBranch);
michael@0:     var oldProfile = pbi.getCharPref("network.IDN.restriction_profile", "moderate");
michael@0:     var oldWhiteListCom;
michael@0:     try {
michael@0:         oldWhitelistCom = pbi.getBoolPref("network.IDN.whitelist.com");
michael@0:     } catch(e) {
michael@0:         oldWhitelistCom = false;
michael@0:     }
michael@0:     var idnService = Cc["@mozilla.org/network/idn-service;1"].getService(Ci.nsIIDNService);
michael@0: 
michael@0:     for (var i = 0; i < profiles.length; ++i) {
michael@0:         pbi.setCharPref("network.IDN.restriction_profile", profiles[i]);
michael@0:         pbi.setBoolPref("network.IDN.whitelist.com", false);
michael@0: 
michael@0:         dump("testing " + profiles[i] + " profile");
michael@0: 
michael@0:         for (var j = 0; j < testcases.length; ++j) {
michael@0:             var test = testcases[j];
michael@0:             var URL = test[0] + ".com";
michael@0:             var punycodeURL = test[1] + ".com";
michael@0:             var expectedUnicode = test[2 + i];
michael@0:             var isASCII = {};
michael@0: 
michael@0: 	    var result;
michael@0: 	    try {
michael@0: 		result = idnService.convertToDisplayIDN(URL, isASCII);
michael@0: 	    } catch(e) {
michael@0: 		result = ".com";
michael@0: 	    }
michael@0:             if (punycodeURL.substr(0, 4) == "xn--") {
michael@0:                 // test convertToDisplayIDN with a Unicode URL and with a
michael@0:                 //  Punycode URL if we have one
michael@0:                 do_check_eq(escape(result),
michael@0:                             expectedUnicode ? escape(URL) : escape(punycodeURL));
michael@0: 
michael@0:                 result = idnService.convertToDisplayIDN(punycodeURL, isASCII);
michael@0:                 do_check_eq(escape(result),
michael@0:                             expectedUnicode ? escape(URL) : escape(punycodeURL));
michael@0:             } else {
michael@0:                 // The "punycode" URL isn't punycode. This happens in testcases
michael@0:                 // where the Unicode URL has become normalized to an ASCII URL,
michael@0:                 // so, even though expectedUnicode is true, the expected result
michael@0:                 // is equal to punycodeURL
michael@0:                 do_check_eq(escape(result), escape(punycodeURL));
michael@0:             }
michael@0:         }
michael@0:     }
michael@0:     pbi.setBoolPref("network.IDN.whitelist.com", oldWhitelistCom);
michael@0:     pbi.setCharPref("network.IDN.restriction_profile", oldProfile);
michael@0: }