Added mappings to canonical locale identifers. The NSLocale class can now, for example, map from AmericanEnglish to en_US. The -preferredLanguages method now works, as well.

git-svn-id: svn+ssh://svn.gna.org/svn/gnustep/libs/base/trunk@31809 72102866-910b-0410-8b05-ffd578937521

ChangeLog

@@ -1,3 +1,10 @@
+2010-12-31 Stefan Bidigaray <stefanbidi@gmail.com>
+
+	* Resources/GNUmakefile:
+	* Resources/Languages/Locale.canonical: Added mapping to go from long
+	locale identifiers to short identifiers used by ICU.
+	* Source/NSLocale.m: Implemented -canonical* and -preferredLanaguges methods.
+
 2010-12-31 14:03  David Chisnall <theraven@gna.org>
 
 	* libs/base/trunk/Headers/Foundation/NSBundle.h: Add __attribute__

Resources/GNUmakefile

@@ -74,7 +74,7 @@ Languages/Ukrainian \
 Languages/Esperanto \
 Languages/Korean 
 LANGUAGE_INSTALL_FILES = $(LANGUAGE_FILES) Languages/Locale.aliases \
-	Languages/Locale.encodings
+	Languages/Locale.encodings Languages/Locale.canonical
 
 include $(GNUSTEP_MAKEFILES)/resource-set.make
 

Resources/Languages/Locale.canonical

@@ -0,0 +1,391 @@
+{
+  Abkhazian = ab;
+  Achinese = ace;
+  Acoli = ach;
+  Adangme = ada;
+  Adyghe = ady;
+  Afar = aa;
+  Afrikaans = af;
+  Akan = ak;
+  Albanian = sq;
+  AlgeriaArabic = ar_DZ;
+  AmericanEnglish = en_US;
+  Amharic = am;
+  Angika = anp;
+  Arabic = ar;
+  Aragonese = an;
+  ArgentinaSpanish = es_AR;
+  Armenian = hy;
+  Aromanian = rup;
+  Assamese = as;
+  Asturian = ast;
+  AustraliaEnglish = en_AU;
+  AustriaGerman = de_AT;
+  Avaric = av;
+  Awadhi = awa;
+  Aymara = ay;
+  Azerbaijani = az; /* -Cyrl, or -Latn ?*/
+  BahrainArabic = ar_BH;
+  Balinese = ban;
+  Baluchi = bal;
+  Bambara = bm;
+  BangladeshBengali = bn_BD;
+  Basa = bas;
+  Bashkir = ba;
+  Basque = eu;
+  Beja = bej;
+  Belarusian = be;
+  BelgiumDutch = nl_BE;
+  BelgiumFrench = fr_BE;
+  BelgiumGerman = de_BE;
+  Bemba = bem;
+  Bengali = bn;
+  Bhojpuri = bho;
+  Bhutani = dz;
+  Bihari = bh;
+  Bikol = bik;
+  Bini = bin;
+  Bislama = bi;
+  Blin = byn;
+  BoliviaSpanish = es_BO;
+  Bosnian = bs;
+  BotswanaEnglish = en_BW;
+  BrasilPortuguese = pt_BR;
+  Breton = br;
+  BritainEnglish = en_GB;
+  Buginese = bug;
+  Bulgarian = bg;
+  Buriat = bua;
+  Burmese = my;
+  Cambodian = km;
+  CanadaEnglish = en_CA;
+  CanadaFrench = fr_CA;
+  Catalan = ca;
+  Cebuano = ceb;
+  Chamorro = ch;
+  Chechen = ce;
+  Chichewa = ny;
+  ChileSpanish = es_CL;
+  Chinese = zh;
+  ChurchSlavic = cu;
+  Chuvash = cv;
+  ColombiaSpanish = es_CO;
+  Cornish = kw;
+  Corsican = co;
+  CostaRicaSpanish = es_CR;
+  Cree = cr;
+  CrimeanTatar = crh;
+  Croatian = hr;
+  Czech = cs;
+  Danish = da;
+  Dargwa = dar;
+  DenmarkEnglish = en_DK;
+  Dinka = din;
+  DjiboutiAfar = aa_DJ;
+  DjiboutiSomali = so_DJ;
+  Dogri = doi;
+  DominicanRepublicSpanish = es_DO;
+  Dutch = nl;
+  Dyula = dyu;
+  EcuadorSpanish = es_EC;
+  Efik = efi;
+  EgyptArabic = ar_EG;
+  ElSalvadorSpanish = es_SV;
+  English = en;
+  EritreaAfar = aa_ER;
+  EritreaGeez = gez_ER;
+  EritreaTigrinya = ti_ER;
+  Erzya = myv;
+  Esperanto = eo;
+  EstoniaEstonian = et_EE;
+  Estonian = et;
+  EthiopiaAfar = aa_ET;
+  EthiopiaGeez = gez_ET;
+  EthiopiaOromo = om_ET;
+  EthiopiaSomali = so_ET;
+  EthiopiaTigrinya = ti_ET;
+  Ewondo = ewo;
+  Fang = fan;
+  Faroese = fo;
+  Fijian = fj;
+  Filipino = fil;
+  FinlandSwedish = sv_FI;
+  Finnish = fi;
+  Fon = fon;
+  French = fr;
+  Friulian = fur;
+  Fulah = ff;
+  Ga = gaa;
+  Gaelic = gd;
+  Gallegan = gl;
+  Ganda = lg;
+  Gayo = gay;
+  Gbaya = gba;
+  Geez = gez;
+  Georgian = ka;
+  German = de;
+  Gondi = gon;
+  Gorontalo = gor;
+  Grebo = grb;
+  GreeceGreek = el_GR;
+  Greek = el;
+  Greenlandic = kl;
+  Guarani = gn;
+  GuatemalaSpanish = es_GT;
+  Gujarati = gu;
+  Haitian = ht;
+  Hausa = ha-Latn;
+  Hebrew = he;
+  Hebrew = iw;
+  Herero = hz;
+  Hiligaynon = hil;
+  Hindi = hi;
+  HiriMotu = ho;
+  Hmong = hmn;
+  HondurasSpanish = es_HN;
+  HongKongChinese = zh-Hant_HK;
+  HongKongEnglish = en_HK;
+  Hungarian = hu;
+  Iban = iba;
+  Icelandic = is;
+  Ido = io;
+  Igbo = ig;
+  Iloko = ilo;
+  IndiaArabic = ar_IN;
+  IndiaBengali = bn_IN;
+  IndiaEnglish = en_IN;
+  Indonesian = id;
+  Ingush = inh;
+  Interlingua = ia;
+  Interlingue = ie;
+  Inuktitut = iu;
+  Inupiak = ik;
+  IraqArabic = ar_IQ;
+  IrelandEnglish = en_IE;
+  Irish = ga;
+  Italian = it;
+  Japanese = ja;
+  Javanese = jv;
+  JordanArabic = ar_JO;
+  Kabardian = kbd;
+  Kabyle = kab;
+  Kachin = kac;
+  Kalmyk = xal;
+  Kamba = kam;
+  Kannada = kn;
+  Kanuri = kr;
+  KarachayBalkar = krc;
+  KaraKalpak = kaa;
+  Karelian = krl;
+  Kashmiri = ks;
+  Kazakh = kk-Cyrl;
+  KenyaOromo = om_KE;
+  KenyaSomali = so_KE;
+  Khasi = kha;
+  Kikuyu = ki;
+  Kimbundu = kmb;
+  Kinyarwanda = rw;
+  Kirghiz = ky;
+  Komi = kv;
+  Kongo = kg;
+  Korean = ko;
+  Kumyk = kum;
+  Kurdish = ku;
+  Kurukh = kru;
+  KuwaitArabic = ar_KW;
+  Kwanyama = kj;
+  Ladino = lad;
+  Lamba = lam;
+  Lao = lo;
+  Latin = la;
+  Latvian = lv;
+  LebanonArabic = ar_LB;
+  Lezghian = lez;
+  LibyaArabic = ar_LY;
+  Limburgish = li;
+  Lingala = ln;
+  Lithuanian = lt;
+  Lozi = loz;
+  LubaKatanga = lu;
+  LubaLulua = lua;
+  Luganda = lug;
+  Lunda = lun;
+  Luo = luo;
+  Lushai = lus;
+  Luxembourgish = lb;
+  LuxemburgFrench = fr_LU;
+  LuxemburgGerman = de_LU;
+  Macedonian = mk;
+  Madurese = mad;
+  Magahi = mag;
+  Maithili = mai;
+  Makasar = mak;
+  Malagasy = mg;
+  Malay = ms;
+  Malayalam = ml;
+  Maldivian = dv;
+  Maltese = mt;
+  Mandar = mdr;
+  Mandingo = man;
+  Manipuri = mni;
+  Manx = gv;
+  Maori = mi;
+  Mapudungun = arn;
+  Marathi = mr;
+  Mari = chm;
+  Marshallese = mh;
+  Masai = mas;
+  Mende = men;
+  MexicoSpanish = es_MX;
+  Minangkabau = min;
+  Moksha = mdf;
+  Moldavian = mo;
+  Mongo = lol;
+  Mongolian = mn;
+  MoroccoArabic = ar_MA;
+  Mossi = mos;
+  Nauru = na;
+  Navajo = nv;
+  NdebeleNorth = nd;
+  NdebeleSouth = nr;
+  Ndonga = ng;
+  Neapolitan = nap;
+  NepalBhasa = new;
+  Nepali = ne;
+  NetherlandsDutch = nl_NL;
+  NewZealandEnglish = en_NZ;
+  Nias = nia;
+  NicaraguaSpanish = es_NI;
+  NorthernSaami = se;
+  Norwegian = no;
+  NorwegianBokmal = nb;
+  NorwegianNynorsk = nn;
+  Nyamwezi = nym;
+  Nyankole = nyn;
+  Nyoro = nyo;
+  Nzima = nzi;
+  Occitan = oc;
+  Ojibwa = oj;
+  OmanArabic = ar_OM;
+  Oriya = or;
+  Oromo = om;
+  Ossetian = os;
+  Pali = pi;
+  Pampanga = pam;
+  PanamaSpanish = es_PA;
+  Pangasinan = pag;
+  Papiamento = pap;
+  ParaguaySpanish = es_PY;
+  Pashto = ps;
+  Pedi = nso;
+  Persian = fa;
+  PeruSpanish = es_PE;
+  PhilippinesEnglish = en_PH;
+  Polish = pl;
+  Portuguese = pt;
+  PuertoRicoSpanish = es_PR;
+  Punjabi = pa; /* -Arab, or -Guru ?*/
+  QatarArabic = ar_QA;
+  Quechua = qu;
+  Rajasthani = raj;
+  Romanian = ro;
+  Romansh = rm;
+  Rundi = rn;
+  Russian = ru;
+  Samoan = sm;
+  Sango = sg;
+  Sanskrit = sa;
+  Santali = sat;
+  Sardinian = sc;
+  Sasak = sas;
+  SaudiArabiaArabic = ar_SA;
+  Scots = sco;
+  ScotsGaelic = gd_GB;
+  Serbian = sr; /* -Cyrl, or -Latn ?*/
+  Serer = srr;
+  Shan = shn;
+  Shona = sn;
+  SichuanYi = ii;
+  Sicilian = scn;
+  Sidamo = sid;
+  SimplifiedChinese = zh-Hans;
+  Sindhi = sd;
+  SingaporeChinese = zh-Hans_SG;
+  SingaporeEnglish = en_SG;
+  Sinhala = si;
+  Slovak = sk;
+  Slovenian = sl;
+  Somali = so;
+  SomaliaSomali = so_SO;
+  Soninke = snk;
+  Sotho = st;
+  SouthAfricaEnglish = en_ZA;
+  SpainBasque = eu_ES;
+  SpainSpanish = es_ES;
+  Spanish = es;
+  SudanArabic = ar_SD;
+  Sukuma = suk;
+  Sundanese = su;
+  Susu = sus;
+  Swahili = sw;
+  Swati = ss;
+  Swedish = sv;
+  SwitzerlandFrench = fr_CH;
+  SwitzerlandGerman = de_CH;
+  SwitzerlandItalian = it_CH;
+  SyriaArabic = ar_SY;
+  Tagalog = tl;
+  Tahitian = ty;
+  Tajik = tg;
+  Tamil = ta;
+  Tatar = tt;
+  Telugu = te;
+  Tetum = tet;
+  Thai = th;
+  Tibetan = bo;
+  Tigre = tig;
+  Tigrinya = ti;
+  Timne = tem;
+  Tiv = tiv;
+  TokPisin = tpi;
+  Tonga = to;
+  TongaNyasa = tog;
+  TraditionalChinese = zh-Hant;
+  Tsonga = ts;
+  Tswana = tn;
+  Tumbuka = tum;
+  TunisiaArabic = ar_TN;
+  Turkish = tr;
+  Turkmen = tk;
+  Tuvinian = tyv;
+  Twi = tw;
+  Udmurt = udm;
+  Uighur = ug;
+  UkraineRussian = ru_UA;
+  Ukrainian = uk;
+  Umbundu = umb;
+  UnitedArabEmiratesArabic = ar_AE;
+  Urdu = ur;
+  UruguaySpanish = es_UY;
+  USASpanish = es_US;
+  Uzbek = uz; /* -Arab, -Cyrl, or -Latn ?*/
+  Vai = vai;
+  Venda = ve;
+  VenezuelaSpanish = es_VE;
+  Vietnamese = vi;
+  Walamo = wal;
+  Walloon = wa;
+  Waray = war;
+  Welsh = cy;
+  WesternFrisian = fy;
+  Wolof = wo;
+  Xhosa = xh;
+  Yakut = sah;
+  Yao = yao;
+  YemenArabic = ar_YE;
+  Yiddish = yi;
+  Yoruba = yo;
+  Zhuang = za;
+  ZimbabweEnglish = en_ZW;
+  Zulu = zu;
+}

Source/NSLocale.m

@@ -25,6 +25,7 @@
 #import "common.h"
 #import "Foundation/NSLocale.h"
 #import "Foundation/NSArray.h"
+#import "Foundation/NSBundle.h"
 #import "Foundation/NSCalendar.h"
 #import "Foundation/NSCoder.h"
 #import "Foundation/NSCharacterSet.h"
@@ -316,6 +317,7 @@ static NSLocale *autoupdatingLocale = nil;
 static NSLocale *currentLocale = nil;
 static NSLocale *systemLocale = nil;
 static NSMutableDictionary *allLocales = nil;
+static NSDictionary *canonicalLocales = nil;
 static NSRecursiveLock *classLock = nil;
 
 + (void) initialize
@@ -375,14 +377,45 @@ static NSRecursiveLock *classLock = nil;
 
 + (NSString *) canonicalLanguageIdentifierFromString: (NSString *) string
 {
-  // FIXME
+  NSString *result;
-  return string;
+  NSString *localeId;
+  NSArray *localeComps;
+  
+  /* Can't use the ICU functions here because, according to Apple locale docs,
+     the language has a format like "zh-Hant".  ICU, however, uses an
+     underscore to separate Scripts "zh_Hant". */
+  localeId = [self canonicalLocaleIdentifierFromString: string];
+  localeComps = [localeId componentsSeparatedByString: @"_"];
+  result = [localeComps objectAtIndex: 0];
+  
+  return result;
 }
 
 + (NSString *) canonicalLocaleIdentifierFromString: (NSString *) string
 {
-  // FIXME
+  /* The way this works, according to Apple docs, is a mess.  It seems
-  return string;
+     that both BCP 47's "-" and ICU's "_" separators are used.  According to
+     "Language and Locale Designations" (Apple docs) Taiwan, for example, has
+     zh-Hant_TW as it's locale identifier (was zh_TW on 10.3.9 and below).
+     Since ICU doesn't use "-" as a separator it will modify that identifier
+     to zh_Hant_TW. */
+  NSString *result;
+  
+  if (canonicalLocales == nil)
+    {
+      NSBundle *gbundle = [NSBundle bundleForLibrary: @"gnustep-base"];
+      NSString *file = [gbundle pathForResource: @"Locale"
+                                         ofType: @"canonical"
+                                    inDirectory: @"Languages"];
+      if (file != nil)
+        canonicalLocales = [[NSDictionary alloc] initWithContentsOfFile: file];
+    }
+  
+  result = [canonicalLocales objectForKey: string];
+  if (result == nil)
+    result = string;
+  
+  return result;
 }
 
 + (NSLocaleLanguageDirection) characterDirectionForLanguage:
@@ -590,8 +623,26 @@ static NSRecursiveLock *classLock = nil;
 
 + (NSArray *) preferredLanguages
 {
-  // FIXME
+  NSArray *result;
-  return [NSUserDefaults userLanguages];
+  NSMutableArray *mArray;
+  NSUInteger cnt;
+  NSUInteger idx = 0;
+  NSArray *languages = [NSUserDefaults userLanguages];
+  
+  if (languages == nil)
+    return [NSArray arrayWithObject: @"en"];
+  
+  mArray = [NSMutableArray array];
+  cnt = [languages count];
+  while (idx < cnt)
+    {
+      NSString *lang = [languages objectAtIndex: idx];
+      [mArray addObject: [self canonicalLanguageIdentifierFromString: lang]];
+      ++idx;
+    }
+  
+  result = [NSArray arrayWithArray: mArray];
+  return result;
 }
 
 + (id) systemLocale
@@ -662,7 +713,8 @@ static NSRecursiveLock *classLock = nil;
 
   return [NSString stringWithUTF8String: buffer];
 #else
-  return nil;	// FIXME
+  return nil;	// FIXME Check
+              // msdn.microsoft.com/en-us/library/0h88fahh%28v=vs.85%29.aspx
 #endif
 }
 
@@ -671,7 +723,8 @@ static NSRecursiveLock *classLock = nil;
 #if	GS_USE_ICU == 1
   return uloc_getLCID ([localeIdentifier UTF8String]);
 #else
-  return 0;	// FIXME
+  return 0;	// FIXME: Check
+            // msdn.microsoft.com/en-us/library/0h88fahh%28v=vs.85%29.aspx
 #endif
 }
 
@@ -753,14 +806,17 @@ static NSRecursiveLock *classLock = nil;
   int32_t	length;
   char cLocaleId[ULOC_FULLNAME_CAPACITY];
   UErrorCode error = U_ZERO_ERROR;
-
+  
-  length = uloc_canonicalize ([string UTF8String], cLocaleId,
+  localeId = [NSLocale canonicalLocaleIdentifierFromString: string];
+  // Normalize locale ID
+  length = uloc_canonicalize ([localeId UTF8String], cLocaleId,
     ULOC_FULLNAME_CAPACITY, &error);
   if (U_FAILURE(error))
     {
       [self release];
       return nil;
     }
+  
   localeId = [NSString stringWithUTF8String: cLocaleId];
 #else
   localeId = string;