Update CLDR to v27

+ A bunch of fixes in the locale data
+ New scripts from Unicode 7.0
+ New locales
- Some locales disappeared (aa_DJ, aa_ER, st_LS, ss_SZ, swc_CD, tn_BW)
- Some locales lost their contents (i.e. en_Dsrt_US)

[ChangeLog][QtCore] QLocale data updated to CLDR v.27

Change-Id: Iba8c7884f8087e577cbb25a8fc106dd7bd3ebb5d
Reviewed-by: Lars Knoll <lars.knoll@digia.com>

src/corelib/tools/qlocale.cpp

@@ -2376,15 +2376,21 @@ Qt::LayoutDirection QLocale::textDirection() const
     case QLocale::KharoshthiScript:
     case QLocale::LydianScript:
     case QLocale::MandaeanScript:
+    case QLocale::ManichaeanScript:
+    case QLocale::MendeKikakuiScript:
     case QLocale::MeroiticCursiveScript:
     case QLocale::MeroiticScript:
+    case QLocale::NabataeanScript:
+    case QLocale::NkoScript:
+    case QLocale::OldNorthArabianScript:
+    case QLocale::OldSouthArabianScript:
+    case QLocale::OrkhonScript:
+    case QLocale::PalmyreneScript:
+    case QLocale::PhoenicianScript:
+    case QLocale::PsalterPahlaviScript:
     case QLocale::SamaritanScript:
     case QLocale::SyriacScript:
     case QLocale::ThaanaScript:
-    case QLocale::NkoScript:
-    case QLocale::OldSouthArabianScript:
-    case QLocale::OrkhonScript:
-    case QLocale::PhoenicianScript:
         return Qt::RightToLeft;
     default:
         break;

src/corelib/tools/qlocale.h

@@ -382,6 +382,31 @@ public:
         Akoose = 312,
         Lakota = 313,
         StandardMoroccanTamazight = 314,
+        Mapuche = 315,
+        CentralKurdish = 316,
+        LowerSorbian = 317,
+        UpperSorbian = 318,
+        Kenyang = 319,
+        Mohawk = 320,
+        Nko = 321,
+        Prussian = 322,
+        Kiche = 323,
+        SouthernSami = 324,
+        LuleSami = 325,
+        InariSami = 326,
+        SkoltSami = 327,
+        Warlpiri = 328,
+        ManichaeanMiddlePersian = 329,
+        Mende = 330,
+        AncientNorthArabian = 331,
+        LinearA = 332,
+        HmongNjua = 333,
+        Ho = 334,
+        Lezghian = 335,
+        Bassa = 336,
+        Mono = 337,
+        TedimChin = 338,
+        Maithili = 339,
         Norwegian = NorwegianBokmal,
         Moldavian = Romanian,
         SerboCroatian = Serbian,
@@ -396,7 +421,7 @@ public:
         Chewa = Nyanja,
         Frisian = WesternFrisian,
         Uigur = Uighur,
-        LastLanguage = StandardMoroccanTamazight
+        LastLanguage = Maithili
     };
 
     enum Script {
@@ -505,9 +530,32 @@ public:
         UgariticScript = 102,
         BrailleScript = 103,
         HiraganaScript = 104,
+        CaucasianAlbanianScript = 105,
+        BassaVahScript = 106,
+        DuployanScript = 107,
+        ElbasanScript = 108,
+        GranthaScript = 109,
+        PahawhHmongScript = 110,
+        KhojkiScript = 111,
+        LinearAScript = 112,
+        MahajaniScript = 113,
+        ManichaeanScript = 114,
+        MendeKikakuiScript = 115,
+        ModiScript = 116,
+        MroScript = 117,
+        OldNorthArabianScript = 118,
+        NabataeanScript = 119,
+        PalmyreneScript = 120,
+        PauCinHauScript = 121,
+        OldPermicScript = 122,
+        PsalterPahlaviScript = 123,
+        SiddhamScript = 124,
+        KhudawadiScript = 125,
+        TirhutaScript = 126,
+        VarangKshitiScript = 127,
         SimplifiedChineseScript = SimplifiedHanScript,
         TraditionalChineseScript = TraditionalHanScript,
-        LastScript = HiraganaScript
+        LastScript = VarangKshitiScript
     };
     enum Country {
         AnyCountry = 0,

src/corelib/tools/qlocale.qdoc

@@ -92,7 +92,7 @@
     \note For the current keyboard input locale take a look at
     QInputMethod::locale().
 
-    QLocale's data is based on Common Locale Data Repository v24.
+    QLocale's data is based on Common Locale Data Repository v27.
 
     The double-to-string and string-to-double conversion functions are
     covered by the following licenses:
@@ -456,6 +456,33 @@
     \value TaiDam
     \value TaiNua
     \value Ugaritic
+    \value Mapuche Since Qt 5.5
+    \value CentralKurdish Since Qt 5.5
+    \value LowerSorbian Since Qt 5.5
+    \value UpperSorbian Since Qt 5.5
+    \value Kenyang Since Qt 5.5
+    \value Mohawk Since Qt 5.5
+    \value Nko Since Qt 5.5
+    \value Prussian Since Qt 5.5
+    \value Kiche Since Qt 5.5
+    \value SouthernSami Since Qt 5.5
+    \value LuleSami Since Qt 5.5
+    \value InariSami Since Qt 5.5
+    \value SkoltSami Since Qt 5.5
+    \value Warlpiri Since Qt 5.5
+    \value ManichaeanMiddlePersian Since Qt 5.5
+    \value Mende Since Qt 5.5
+    \value AncientNorthArabian Since Qt 5.5
+    \value LinearA Since Qt 5.5
+    \value HmongNjua Since Qt 5.5
+    \value Ho Since Qt 5.5
+    \value Lezghian Since Qt 5.5
+    \value Bassa Since Qt 5.5
+    \value Mono Since Qt 5.5
+    \value TedimChin Since Qt 5.5
+    \value Maithili Since Qt 5.5
+    \value LowerSorbian Since Qt 5.5
+    \value UpperSorbian Since Qt 5.5
     \omitvalue LastLanguage
 
     \sa language(), languageToString()
@@ -746,6 +773,7 @@
     \value AvestanScript
     \value BalineseScript
     \value BamumScript
+    \value BassaVahScript Since Qt 5.5
     \value BatakScript
     \value BengaliScript
     \value BopomofoScript
@@ -755,6 +783,7 @@
     \value BuhidScript
     \value CanadianAboriginalScript
     \value CarianScript
+    \value CaucasianAlbanianScript Since Qt 5.5
     \value ChakmaScript
     \value ChamScript
     \value CherokeeScript
@@ -763,12 +792,15 @@
     \value CyrillicScript
     \value DeseretScript
     \value DevanagariScript
+    \value DuployanScript Since Qt 5.5
     \value EgyptianHieroglyphsScript
+    \value ElbasanScript Since Qt 5.5
     \value EthiopicScript
     \value FraserScript
     \value GeorgianScript
     \value GlagoliticScript
     \value GothicScript
+    \value GranthaScript Since Qt 5.5
     \value GreekScript
     \value GujaratiScript
     \value GurmukhiScript
@@ -788,41 +820,57 @@
     \value KayahLiScript
     \value KharoshthiScript
     \value KhmerScript
+    \value KhojkiScript Since Qt 5.5
+    \value KhudawadiScript Since Qt 5.5
     \value KoreanScript
     \value LannaScript
     \value LaoScript
     \value LatinScript
     \value LepchaScript
     \value LimbuScript
+    \value LinearAScript Since Qt 5.5
     \value LinearBScript
     \value LycianScript
     \value LydianScript
+    \value MahajaniScript Since Qt 5.5
     \value MalayalamScript
     \value MandaeanScript
+    \value ManichaeanScript Since Qt 5.5
     \value MeiteiMayekScript
+    \value MendeKikakuiScript Since Qt 5.5
     \value MeroiticScript
     \value MeroiticCursiveScript
+    \value ModiScript Since Qt 5.5
     \value MongolianScript
+    \value MroScript Since Qt 5.5
     \value MyanmarScript
+    \value NabataeanScript Since Qt 5.5
     \value NkoScript
     \value NewTaiLueScript
     \value OghamScript
     \value OlChikiScript
     \value OldItalicScript
+    \value OldNorthArabianScript Since Qt 5.5
+    \value OldPermicScript Since Qt 5.5
     \value OldPersianScript
     \value OldSouthArabianScript
     \value OriyaScript
     \value OrkhonScript
     \value OsmanyaScript
+    \value PahawhHmongScript Since Qt 5.5
+    \value PalmyreneScript Since Qt 5.5
+    \value PauCinHauScript Since Qt 5.5
     \value PhagsPaScript
     \value PhoenicianScript
     \value PollardPhoneticScript
+    \value PsalterPahlaviScript Since Qt 5.5
     \value RejangScript
     \value RunicScript
     \value SamaritanScript
     \value SaurashtraScript
     \value SharadaScript
     \value ShavianScript
+    \value SiddhamScript Since Qt 5.5
     \value SimplifiedHanScript same as SimplifiedChineseScript
     \value SimplifiedChineseScript same as SimplifiedHanScript
     \value SinhalaScript
@@ -842,10 +890,12 @@
     \value ThaiScript
     \value TibetanScript
     \value TifinaghScript
+    \value TirhutaScript Since Qt 5.5
     \value TraditionalHanScript same as TraditionalChineseScript
     \value TraditionalChineseScript same as TraditionalHanScript
     \value UgariticScript
     \value VaiScript
+    \value VarangKshitiScript Since Qt 5.5
     \value YiScript
     \omitvalue LastScript
 

tests/auto/corelib/tools/qlocale/tst_qlocale.cpp

@@ -1889,8 +1889,8 @@ void tst_QLocale::ampm()
     QCOMPARE(sv.pmText(), QLatin1String("em"));
 
     QLocale nn("nl_NL");
-    QCOMPARE(nn.amText(), QLatin1String("AM"));
+    QCOMPARE(nn.amText(), QLatin1String("a.m."));
-    QCOMPARE(nn.pmText(), QLatin1String("PM"));
+    QCOMPARE(nn.pmText(), QLatin1String("p.m."));
 
     QLocale ua("uk_UA");
     QCOMPARE(ua.amText(), QString::fromUtf8("\320\264\320\277"));
@@ -1916,12 +1916,12 @@ void tst_QLocale::dateFormat()
     QCOMPARE(c.dateFormat(QLocale::NarrowFormat), c.dateFormat(QLocale::ShortFormat));
 
     const QLocale no("no_NO");
-    QCOMPARE(no.dateFormat(QLocale::NarrowFormat), QLatin1String("dd.MM.yy"));
+    QCOMPARE(no.dateFormat(QLocale::NarrowFormat), QLatin1String("dd.MM.yyyy"));
-    QCOMPARE(no.dateFormat(QLocale::ShortFormat), QLatin1String("dd.MM.yy"));
+    QCOMPARE(no.dateFormat(QLocale::ShortFormat), QLatin1String("dd.MM.yyyy"));
     QCOMPARE(no.dateFormat(QLocale::LongFormat), QLatin1String("dddd d. MMMM yyyy"));
 
     const QLocale ca("en_CA");
-    QCOMPARE(ca.dateFormat(QLocale::ShortFormat), QLatin1String("M/d/yy"));
+    QCOMPARE(ca.dateFormat(QLocale::ShortFormat), QLatin1String("yyyy-MM-dd"));
     QCOMPARE(ca.dateFormat(QLocale::LongFormat), QLatin1String("dddd, MMMM d, yyyy"));
 
     const QLocale ja("ja_JP");
@@ -1947,8 +1947,8 @@ void tst_QLocale::timeFormat()
     QCOMPARE(id.timeFormat(QLocale::LongFormat), QLatin1String("HH.mm.ss t"));
 
     const QLocale cat("ca_ES");
-    QCOMPARE(cat.timeFormat(QLocale::ShortFormat), QLatin1String("H.mm"));
+    QCOMPARE(cat.timeFormat(QLocale::ShortFormat), QLatin1String("H:mm"));
-    QCOMPARE(cat.timeFormat(QLocale::LongFormat), QLatin1String("H.mm.ss t"));
+    QCOMPARE(cat.timeFormat(QLocale::LongFormat), QLatin1String("H:mm:ss t"));
 
     const QLocale bra("pt_BR");
     QCOMPARE(bra.timeFormat(QLocale::ShortFormat), QLatin1String("HH:mm"));
@@ -1962,8 +1962,8 @@ void tst_QLocale::dateTimeFormat()
     QCOMPARE(c.dateTimeFormat(QLocale::NarrowFormat), c.dateTimeFormat(QLocale::ShortFormat));
 
     const QLocale no("no_NO");
-    QCOMPARE(no.dateTimeFormat(QLocale::NarrowFormat), QLatin1String("dd.MM.yy HH.mm"));
+    QCOMPARE(no.dateTimeFormat(QLocale::NarrowFormat), QLatin1String("dd.MM.yyyy HH.mm"));
-    QCOMPARE(no.dateTimeFormat(QLocale::ShortFormat), QLatin1String("dd.MM.yy HH.mm"));
+    QCOMPARE(no.dateTimeFormat(QLocale::ShortFormat), QLatin1String("dd.MM.yyyy HH.mm"));
     QCOMPARE(no.dateTimeFormat(QLocale::LongFormat), QLatin1String("dddd d. MMMM yyyy HH.mm.ss t"));
 }
 
@@ -1992,10 +1992,6 @@ void tst_QLocale::monthName()
     QCOMPARE(ru.monthName(1, QLocale::ShortFormat), QString::fromUtf8("\321\217\320\275\320\262\56"));
     QCOMPARE(ru.monthName(1, QLocale::NarrowFormat), QString::fromUtf8("\320\257"));
 
-    // check that our CLDR scripts handle surrogate pairs correctly
-    QLocale dsrt("en-Dsrt-US");
-    QCOMPARE(dsrt.monthName(1, QLocale::LongFormat), QString::fromUtf8("\xf0\x90\x90\x96\xf0\x90\x90\xb0\xf0\x90\x91\x8c\xf0\x90\x90\xb7\xf0\x90\x90\xad\xf0\x90\x90\xaf\xf0\x90\x91\x89\xf0\x90\x90\xa8"));
-
     QLocale ir("ga_IE");
     QCOMPARE(ir.monthName(1, QLocale::ShortFormat), QLatin1String("Ean"));
     QCOMPARE(ir.monthName(12, QLocale::ShortFormat), QLatin1String("Noll"));
@@ -2027,9 +2023,9 @@ void tst_QLocale::standaloneMonthName()
     QCOMPARE(de.standaloneMonthName(12, QLocale::NarrowFormat), QLatin1String("D"));
 
     QLocale ru("ru_RU");
-    QCOMPARE(ru.standaloneMonthName(1, QLocale::LongFormat), QString::fromUtf8("\320\257\320\275\320\262\320\260\321\200\321\214"));
+    QCOMPARE(ru.standaloneMonthName(1, QLocale::LongFormat), QString::fromUtf8("\xd1\x8f\xd0\xbd\xd0\xb2\xd0\xb0\xd1\x80\xd1\x8c"));
-    QCOMPARE(ru.standaloneMonthName(1, QLocale::ShortFormat), QString::fromUtf8("\320\257\320\275\320\262\56"));
+    QCOMPARE(ru.standaloneMonthName(1, QLocale::ShortFormat), QString::fromUtf8("\xd1\x8f\xd0\xbd\xd0\xb2."));
-    QCOMPARE(ru.standaloneMonthName(1, QLocale::NarrowFormat), QString::fromUtf8("\320\257"));
+    QCOMPARE(ru.standaloneMonthName(1, QLocale::NarrowFormat), QString::fromUtf8("\xd0\xaf"));
 }
 
 void tst_QLocale::currency()
@@ -2073,8 +2069,8 @@ void tst_QLocale::quoteString()
     QCOMPARE(c.quoteString(someText, QLocale::AlternateQuotation), QString::fromUtf8("\x27" "text" "\x27"));
 
     const QLocale de_CH("de_CH");
-    QCOMPARE(de_CH.quoteString(someText), QString::fromUtf8("\xc2\xab" "text" "\xc2\xbb"));
+    QCOMPARE(de_CH.quoteString(someText), QString::fromUtf8("\xe2\x80\x9e" "text" "\xe2\x80\x9c"));
-    QCOMPARE(de_CH.quoteString(someText, QLocale::AlternateQuotation), QString::fromUtf8("\xe2\x80\xb9" "text" "\xe2\x80\xba"));
+    QCOMPARE(de_CH.quoteString(someText, QLocale::AlternateQuotation), QString::fromUtf8("\xe2\x80\x9a" "text" "\xe2\x80\x98"));
 
 }
 
@@ -2201,15 +2197,36 @@ void tst_QLocale::textDirection_data()
     for (int language = QLocale::C; language <= QLocale::LastLanguage; ++language) {
         bool rightToLeft = false;
         switch (language) {
+        // based on likelySubtags for RTL scripts
+        case QLocale::AncientGreek:
+        case QLocale::AncientNorthArabian:
         case QLocale::Arabic:
+        case QLocale::Aramaic:
+        case QLocale::Avestan:
+        case QLocale::CentralKurdish:
+        case QLocale::ClassicalMandaic:
+        case QLocale::Divehi:
         case QLocale::Hebrew:
         case QLocale::Kashmiri:
-        case QLocale::Persian:
+        case QLocale::Lydian:
+        case QLocale::Mandingo:
+        case QLocale::ManichaeanMiddlePersian:
+        case QLocale::Mende:
+        case QLocale::Meroitic:
+        case QLocale::Nko:
+        case QLocale::OldTurkish:
+        case QLocale::Pahlavi:
+        case QLocale::Parthian:
         case QLocale::Pashto:
+        case QLocale::Persian:
+        case QLocale::Phoenician:
+        case QLocale::PrakritLanguage:
+        case QLocale::Sabaean:
+        case QLocale::Samaritan:
+        case QLocale::Sindhi:
+        case QLocale::Syriac:
         case QLocale::Uighur:
         case QLocale::Urdu:
-        case QLocale::Syriac:
-        case QLocale::Divehi:
             rightToLeft = QLocale(QLocale::Language(language)).language() == QLocale::Language(language); // false if there is no locale data for language
             break;
         default:

util/local_database/cldr2qlocalexml.py

@@ -803,42 +803,42 @@ for key in locale_keys:
     print "            <minus>"    + fixOrdStrMinus(l['minus'])   + "</minus>"
     print "            <plus>"     + fixOrdStrPlus(l['plus'])   + "</plus>"
     print "            <exp>"      + fixOrdStrExp(l['exp'])     + "</exp>"
-    print "            <quotationStart>" + l['quotationStart'].encode('utf-8') + "</quotationStart>"
+    print "            <quotationStart>" + escape(l['quotationStart']).encode('utf-8') + "</quotationStart>"
-    print "            <quotationEnd>" + l['quotationEnd'].encode('utf-8')   + "</quotationEnd>"
+    print "            <quotationEnd>" + escape(l['quotationEnd']).encode('utf-8')   + "</quotationEnd>"
-    print "            <alternateQuotationStart>" + l['alternateQuotationStart'].encode('utf-8') + "</alternateQuotationStart>"
+    print "            <alternateQuotationStart>" + escape(l['alternateQuotationStart']).encode('utf-8') + "</alternateQuotationStart>"
-    print "            <alternateQuotationEnd>" + l['alternateQuotationEnd'].encode('utf-8')   + "</alternateQuotationEnd>"
+    print "            <alternateQuotationEnd>" + escape(l['alternateQuotationEnd']).encode('utf-8')   + "</alternateQuotationEnd>"
-    print "            <listPatternPartStart>" + l['listPatternPartStart'].encode('utf-8')   + "</listPatternPartStart>"
+    print "            <listPatternPartStart>" + escape(l['listPatternPartStart']).encode('utf-8')   + "</listPatternPartStart>"
-    print "            <listPatternPartMiddle>" + l['listPatternPartMiddle'].encode('utf-8')   + "</listPatternPartMiddle>"
+    print "            <listPatternPartMiddle>" + escape(l['listPatternPartMiddle']).encode('utf-8')   + "</listPatternPartMiddle>"
-    print "            <listPatternPartEnd>" + l['listPatternPartEnd'].encode('utf-8')   + "</listPatternPartEnd>"
+    print "            <listPatternPartEnd>" + escape(l['listPatternPartEnd']).encode('utf-8')   + "</listPatternPartEnd>"
-    print "            <listPatternPartTwo>" + l['listPatternPartTwo'].encode('utf-8')   + "</listPatternPartTwo>"
+    print "            <listPatternPartTwo>" + escape(l['listPatternPartTwo']).encode('utf-8')   + "</listPatternPartTwo>"
-    print "            <am>"       + l['am'].encode('utf-8') + "</am>"
+    print "            <am>"       + escape(l['am']).encode('utf-8') + "</am>"
-    print "            <pm>"       + l['pm'].encode('utf-8') + "</pm>"
+    print "            <pm>"       + escape(l['pm']).encode('utf-8') + "</pm>"
-    print "            <firstDayOfWeek>"  + l['firstDayOfWeek'].encode('utf-8') + "</firstDayOfWeek>"
+    print "            <firstDayOfWeek>"  + escape(l['firstDayOfWeek']).encode('utf-8') + "</firstDayOfWeek>"
-    print "            <weekendStart>"  + l['weekendStart'].encode('utf-8') + "</weekendStart>"
+    print "            <weekendStart>"  + escape(l['weekendStart']).encode('utf-8') + "</weekendStart>"
-    print "            <weekendEnd>"  + l['weekendEnd'].encode('utf-8') + "</weekendEnd>"
+    print "            <weekendEnd>"  + escape(l['weekendEnd']).encode('utf-8') + "</weekendEnd>"
-    print "            <longDateFormat>"  + l['longDateFormat'].encode('utf-8')  + "</longDateFormat>"
+    print "            <longDateFormat>"  + escape(l['longDateFormat']).encode('utf-8')  + "</longDateFormat>"
-    print "            <shortDateFormat>" + l['shortDateFormat'].encode('utf-8') + "</shortDateFormat>"
+    print "            <shortDateFormat>" + escape(l['shortDateFormat']).encode('utf-8') + "</shortDateFormat>"
-    print "            <longTimeFormat>"  + l['longTimeFormat'].encode('utf-8')  + "</longTimeFormat>"
+    print "            <longTimeFormat>"  + escape(l['longTimeFormat']).encode('utf-8')  + "</longTimeFormat>"
-    print "            <shortTimeFormat>" + l['shortTimeFormat'].encode('utf-8') + "</shortTimeFormat>"
+    print "            <shortTimeFormat>" + escape(l['shortTimeFormat']).encode('utf-8') + "</shortTimeFormat>"
-    print "            <standaloneLongMonths>" + l['standaloneLongMonths'].encode('utf-8')      + "</standaloneLongMonths>"
+    print "            <standaloneLongMonths>" + escape(l['standaloneLongMonths']).encode('utf-8')      + "</standaloneLongMonths>"
-    print "            <standaloneShortMonths>"+ l['standaloneShortMonths'].encode('utf-8')      + "</standaloneShortMonths>"
+    print "            <standaloneShortMonths>"+ escape(l['standaloneShortMonths']).encode('utf-8')      + "</standaloneShortMonths>"
-    print "            <standaloneNarrowMonths>"+ l['standaloneNarrowMonths'].encode('utf-8')      + "</standaloneNarrowMonths>"
+    print "            <standaloneNarrowMonths>"+ escape(l['standaloneNarrowMonths']).encode('utf-8')      + "</standaloneNarrowMonths>"
-    print "            <longMonths>"      + l['longMonths'].encode('utf-8')      + "</longMonths>"
+    print "            <longMonths>"      + escape(l['longMonths']).encode('utf-8')      + "</longMonths>"
-    print "            <shortMonths>"     + l['shortMonths'].encode('utf-8')     + "</shortMonths>"
+    print "            <shortMonths>"     + escape(l['shortMonths']).encode('utf-8')     + "</shortMonths>"
-    print "            <narrowMonths>"     + l['narrowMonths'].encode('utf-8')     + "</narrowMonths>"
+    print "            <narrowMonths>"     + escape(l['narrowMonths']).encode('utf-8')     + "</narrowMonths>"
-    print "            <longDays>"        + l['longDays'].encode('utf-8')        + "</longDays>"
+    print "            <longDays>"        + escape(l['longDays']).encode('utf-8')        + "</longDays>"
-    print "            <shortDays>"       + l['shortDays'].encode('utf-8')       + "</shortDays>"
+    print "            <shortDays>"       + escape(l['shortDays']).encode('utf-8')       + "</shortDays>"
-    print "            <narrowDays>"       + l['narrowDays'].encode('utf-8')       + "</narrowDays>"
+    print "            <narrowDays>"       + escape(l['narrowDays']).encode('utf-8')       + "</narrowDays>"
-    print "            <standaloneLongDays>" + l['standaloneLongDays'].encode('utf-8')        + "</standaloneLongDays>"
+    print "            <standaloneLongDays>" + escape(l['standaloneLongDays']).encode('utf-8')        + "</standaloneLongDays>"
-    print "            <standaloneShortDays>" + l['standaloneShortDays'].encode('utf-8')       + "</standaloneShortDays>"
+    print "            <standaloneShortDays>" + escape(l['standaloneShortDays']).encode('utf-8')       + "</standaloneShortDays>"
-    print "            <standaloneNarrowDays>" + l['standaloneNarrowDays'].encode('utf-8')       + "</standaloneNarrowDays>"
+    print "            <standaloneNarrowDays>" + escape(l['standaloneNarrowDays']).encode('utf-8')       + "</standaloneNarrowDays>"
-    print "            <currencyIsoCode>" + l['currencyIsoCode'].encode('utf-8') + "</currencyIsoCode>"
+    print "            <currencyIsoCode>" + escape(l['currencyIsoCode']).encode('utf-8') + "</currencyIsoCode>"
-    print "            <currencySymbol>" + l['currencySymbol'].encode('utf-8') + "</currencySymbol>"
+    print "            <currencySymbol>" + escape(l['currencySymbol']).encode('utf-8') + "</currencySymbol>"
-    print "            <currencyDisplayName>" + l['currencyDisplayName'].encode('utf-8') + "</currencyDisplayName>"
+    print "            <currencyDisplayName>" + escape(l['currencyDisplayName']).encode('utf-8') + "</currencyDisplayName>"
     print "            <currencyDigits>" + str(l['currencyDigits']) + "</currencyDigits>"
     print "            <currencyRounding>" + str(l['currencyRounding']) + "</currencyRounding>"
-    print "            <currencyFormat>" + l['currencyFormat'].encode('utf-8') + "</currencyFormat>"
+    print "            <currencyFormat>" + escape(l['currencyFormat']).encode('utf-8') + "</currencyFormat>"
-    print "            <currencyNegativeFormat>" + l['currencyNegativeFormat'].encode('utf-8') + "</currencyNegativeFormat>"
+    print "            <currencyNegativeFormat>" + escape(l['currencyNegativeFormat']).encode('utf-8') + "</currencyNegativeFormat>"
     print "        </locale>"
 print "    </localeList>"
 print "</localeDatabase>"

util/local_database/enumdata.py

@@ -351,7 +351,32 @@ language_list = {
     311 : [ "Ugaritic",           "uga" ],
     312 : [ "Akoose",             "bss" ],
     313 : [ "Lakota",             "lkt" ],
-    314 : [ "Standard Moroccan Tamazight", "zgh" ]
+    314 : [ "Standard Moroccan Tamazight", "zgh" ],
+    315 : [ "Mapuche",            "arn" ],
+    316 : [ "Central Kurdish",    "ckb" ],
+    317 : [ "LowerSorbian",       "dsb" ],
+    318 : [ "UpperSorbian",       "hsb" ],
+    319 : [ "Kenyang",            "ken" ],
+    320 : [ "Mohawk",             "moh" ],
+    321 : [ "Nko",                "nqo" ],
+    322 : [ "Prussian",           "prg" ],
+    323 : [ "Kiche",              "quc" ],
+    324 : [ "Southern Sami",      "sma" ],
+    325 : [ "Lule Sami",          "smj" ],
+    326 : [ "Inari Sami",         "smn" ],
+    327 : [ "Skolt Sami",         "sms" ],
+    328 : [ "Warlpiri",           "wbp" ],
+    329 : [ "Manichaean Middle Persian", "xmn" ],
+    330 : [ "Mende",              "men" ],
+    331 : [ "Ancient North Arabian", "xna" ],
+    332 : [ "Linear A",           "lab" ],
+    333 : [ "Hmong Njua",         "hnj" ],
+    334 : [ "Ho",                 "hoc" ],
+    335 : [ "Lezghian",           "lez" ],
+    336 : [ "Bassa",              "bsq" ],
+    337 : [ "Mono",               "mru" ],
+    338 : [ "Tedim Chin",         "ctd" ],
+    339 : [ "Maithili",           "mai" ]
 }
 
 country_list = {
@@ -720,13 +745,33 @@ script_list = {
     101 : [ "Takri",                  "Takr" ],
     102 : [ "Ugaritic",               "Ugar" ],
     103 : [ "Braille",                "Brai" ],
-    104 : [ "Hiragana",               "Hira" ]
+    104 : [ "Hiragana",               "Hira" ],
+    105 : [ "Caucasian Albanian",     "Aghb" ],
+    106 : [ "Bassa Vah",              "Bass" ],
+    107 : [ "Duployan",               "Dupl" ],
+    108 : [ "Elbasan",                "Elba" ],
+    109 : [ "Grantha",                "Gran" ],
+    110 : [ "Pahawh Hmong",           "Hmng" ],
+    111 : [ "Khojki",                 "Khoj" ],
+    112 : [ "Linear A",               "Lina" ],
+    113 : [ "Mahajani",               "Mahj" ],
+    114 : [ "Manichaean",             "Mani" ],
+    115 : [ "Mende Kikakui",          "Mend" ],
+    116 : [ "Modi",                   "Modi" ],
+    117 : [ "Mro",                    "Mroo" ],
+    118 : [ "Old North Arabian",      "Narb" ],
+    119 : [ "Nabataean",              "Nbat" ],
+    120 : [ "Palmyrene",              "Palm" ],
+    121 : [ "Pau Cin Hau",            "Pauc" ],
+    122 : [ "Old Permic",             "Perm" ],
+    123 : [ "Psalter Pahlavi",        "Phlp" ],
+    124 : [ "Siddham",                "Sidd" ],
+    125 : [ "Khudawadi",              "Sind" ],
+    126 : [ "Tirhuta",                "Tirh" ],
+    127 : [ "Varang Kshiti",          "Wara" ]
 
 #    ### : [ "Blissymbols",            "Blis" ],
-#    ### : [ "Linear A",               "Lina" ],
 #    ### : [ "Naxi Geba",              "Nkgb" ],
-#    ### : [ "Pahawh Hmong",           "Hmng" ],
-#    ### : [ "Varang Kshiti",          "Wara" ],
 }
 
 def countryCodeToId(code):