Cygwin: locale(1): drop using LCID, use Windows locale names

LCIDs are deprecated since Windows Vista.  Worse, lots of new locales
have been added in the meantime which have no LCID attached.  They
are only available by locale name.

As first step, rearrange the locale(1) tool to use Windows locale
names, rather than LCIDs, so we can now enumerate *all* locales
available in more recent Windows versions.

Signed-off-by: Corinna Vinschen <corinna@vinschen.de>

winsup/utils/locale.cc

@@ -30,6 +30,7 @@
 #include <getopt.h>
 #include <string.h>
 #include <wchar.h>
+#include <wctype.h>
 #include <locale.h>
 #include <langinfo.h>
 #include <limits.h>
@@ -107,17 +108,42 @@ struct option longopts[] = {
 const char *opts = "acfhikmnsuUvV";
 
 int
-getlocale (LCID lcid, char *name)
+getlocale (PWCHAR loc_name, wchar_t *iso639, wchar_t *iso3166,
+	   wchar_t *iso15924 = NULL)
 {
-  char iso639[10];
+  wchar_t *cp;
-  char iso3166[10];
 
-  iso3166[0] = '\0';
+  /* Skip language-only locales, e. g. "en" */
-  if (!GetLocaleInfo (lcid, LOCALE_SISO639LANGNAME, iso639, 10))
+  if (!(cp = wcschr (loc_name, L'-')))
     return 0;
-  GetLocaleInfo (lcid, LOCALE_SISO3166CTRYNAME, iso3166, 10);
+  ++cp;
-  sprintf (name, "%s%s%s", iso639, lcid > 0x3ff ? "_" : "",
+  /* Script inside?  Scripts are Upper/Lower, e. g. "Latn" */
-			   lcid > 0x3ff ? iso3166 : "");
+  if (iswupper (cp[0]) && iswlower (cp[1]))
+    {
+      wchar_t *cp2;
+
+      /* Skip language-Script locales, missing country  */
+      if (!(cp2 = wcschr (cp + 2, L'-')))
+	return 0;
+      /* Otherwise, store in iso15924 */
+      if (iso15924)
+	wcpcpy (wcpncpy (iso15924, cp, cp2 - cp), L";");
+    }
+  cp = wcsrchr (loc_name, L'-');
+  if (cp)
+    {
+      /* Skip numeric iso3166 country name. */
+      if (iswdigit (cp[1]))
+	return 0;
+      /* Special case postfix after iso3166 country name: ca-ES-valencia.
+	 Use the postfix thingy as script so it will become a @modifier */
+      if (iswlower (cp[1]))
+	wcpcpy (iso15924, cp + 1);
+    }
+
+  if (!GetLocaleInfoEx (loc_name, LOCALE_SISO639LANGNAME, iso639, 10))
+    return 0;
+  GetLocaleInfoEx (loc_name, LOCALE_SISO3166CTRYNAME, iso3166, 10);
   return 1;
 }
 
@@ -132,14 +158,6 @@ loc_t *locale;
 size_t loc_max;
 size_t loc_num;
 
-void
-print_codeset (const char *codeset)
-{
-  for (; *codeset; ++codeset)
-    if (*codeset != '-')
-      putc (tolower ((int)(unsigned char) *codeset), stdout);
-}
-
 void
 print_locale_with_codeset (int verbose, loc_t *locale, bool utf8,
 			   const char *modifier)
@@ -187,7 +205,7 @@ print_locale (int verbose, loc_t *locale)
     {
       if (!modifier)
 	print_locale_with_codeset (verbose, locale, true, NULL);
-      else if (!strcmp (modifier, "@cjknarrow"))
+      else if (strcmp (modifier, "@euro"))
 	{
 	  *modifier++ = '\0';
 	  print_locale_with_codeset (verbose, locale, true, modifier);
@@ -266,110 +284,147 @@ add_locale_alias_locales ()
       c = strchr (replace, '.');
       if (c)
 	*c = '\0';
+      /* Ignore "ja_JP" and "ko_KR" locales from here, they are in the Windows
+	 DB anyway. */
+      if (!strcmp (alias, "ja_JP") || !strcmp (alias, "ko_KR"))
+	continue;
       search.name = replace;
       loc = (loc_t *) bsearch (&search, locale, orig_loc_num, sizeof (loc_t),
 			       compare_locales);
+
       add_locale (alias, loc ? loc->language : L"", loc ? loc->territory : L"",
 		  true);
     }
   fclose (fp);
 }
 
-void
+BOOL
-print_all_locales (int verbose)
+print_all_locales_proc (LPWSTR loc_name, DWORD info, LPARAM param)
 {
-  LCID lcid = 0;
+  wchar_t iso639[32] = { 0 };
-  char name[32];
+  wchar_t iso3166[32] = { 0 };
+  wchar_t iso15924[32] = { 0 };
   DWORD cp;
 
-  unsigned lang, sublang;
+#if 0
-
-  add_locale ("C", L"C", L"POSIX");
-  add_locale ("POSIX", L"C", L"POSIX", true);
-  for (lang = 1; lang <= 0xff; ++lang)
-    {
   struct {
     wchar_t language[256];
     wchar_t country[256];
     char loc[32];
   } loc_list[32];
   int lcnt = 0;
+#endif
 
-      for (sublang = 1; sublang <= 0x3f; ++sublang)
+  if (getlocale (loc_name, iso639, iso3166, iso15924))
-	{
-	  lcid = (sublang << 10) | lang;
-	  if (getlocale (lcid, name))
     {
+      char *c, posix_loc[32];
       wchar_t language[256];
       wchar_t country[256];
-	      int i;
+      wchar_t currency[9];
-	      char *c, loc[32];
-	      wchar_t wbuf[9];
 
-	      /* Go figure.  Even the English name of a language or
+      c = posix_loc + snprintf (posix_loc, sizeof posix_loc, "%ls_%ls",
-		 locale might contain native characters. */
+				iso639, iso3166);
-	      GetLocaleInfoW (lcid, LOCALE_SENGLANGUAGE, language, 256);
+      /* Inuktitut: equivalent @latin due to lack of info on Linux */
-	      GetLocaleInfoW (lcid, LOCALE_SENGCOUNTRY, country, 256);
+      if (!wcscmp (iso639, L"iu"))
-	      /* Avoid dups */
-	      for (i = 0; i < lcnt; ++ i)
-		if (!wcscmp (loc_list[i].language, language)
-		    && !wcscmp (loc_list[i].country, country))
-		  break;
-	      if (i < lcnt)
-		continue;
-	      if (lcnt < 32)
 	{
-		  wcscpy (loc_list[lcnt].language, language);
+	  if (wcscmp (iso15924, L"Latn;"))
-		  wcscpy (loc_list[lcnt].country, country);
+	    return TRUE;
 	}
-	      c = stpcpy (loc, name);
+      /* Javanese: only use @latin locale. */
-	      /* Now check certain conditions to figure out if that
+      else if (!wcscmp (iso639, L"jv"))
-		 locale requires a modifier. */
+	{
-	      if (lang == LANG_SERBIAN && !strncmp (loc, "sr_", 3)
+	  if (wcscmp (iso15924, L"Latn;"))
-		  && wcsstr (language, L"(Latin)"))
+	    return TRUE;
+	}
+      /* Mongolian: only use @mongolian locale. */
+      else if (!wcscmp (iso639, L"mn"))
+	{
+	  if (wcscmp (iso15924, L"Mong;"))
+	    return TRUE;
+	}
+      /* Serbian: Windows default is Latin, Linux default is Cyrillic.
+	 We want the Linux default and attach @latin otherwise */
+      else if (!wcscmp (iso639, L"sr")  && !wcscmp (iso15924, L"Latn;"))
 	stpcpy (c, "@latin");
-	      else if (lang == LANG_UZBEK
+      /* Tamazight: no modifier, iso639 is "ber" on Linux.
-		       && sublang == SUBLANG_UZBEK_CYRILLIC)
+	 "zgh-Tfng-MA" is equivalent to "ber_MA". */
+      else if (!wcscmp (iso639, L"zgh"))
+	snprintf (posix_loc, sizeof posix_loc, "ber_%.27ls", iso3166);
+      /* Tamazight: "tzm-Latn-DZ" is equivalent to "ber_DZ",
+		    skip everything else. */
+      else if (!wcscmp (iso639, L"tzm"))
+	{
+	  if (!wcscmp (iso3166, L"DZ") && !wcscmp (iso15924, L"Latn;"))
+	    snprintf (posix_loc, sizeof posix_loc, "ber_%.27ls", iso3166);
+	  else
+	    return TRUE;
+	}
+      /* In all other cases, we check if the script from the Windows
+	 locale is the default locale in that language.  If not, we
+	 add it as modifier if possible, or skip it */
+      else if (iso15924[0])
+	{
+	  wchar_t scriptless_win_locale[32];
+	  wchar_t default_iso15924[32];
+
+	  wcpcpy (wcpcpy (wcpcpy (scriptless_win_locale, iso639), L"_"),
+		  iso3166);
+	  if ((GetLocaleInfoEx (scriptless_win_locale, LOCALE_SSCRIPTS,
+				default_iso15924, 32)
+	       || GetLocaleInfoEx (iso639, LOCALE_SSCRIPTS,
+				   default_iso15924, 32))
+	      && !wcsstr (default_iso15924, iso15924))
+	    {
+	      if (!wcscmp (iso15924, L"Latn;"))
+		stpcpy (c, "@latin");
+	      else if (!wcscmp (iso15924, L"Cyrl;"))
 		stpcpy (c, "@cyrillic");
-	      /* Avoid more dups */
+	      else if (!wcscmp (iso15924, L"Deva;"))
-	      for (i = 0; i < lcnt; ++ i)
+		stpcpy (c, "@devanagari");
-		if (!strcmp (loc_list[i].loc, loc))
+	      else if (!wcscmp (iso15924, L"Adlm;"))
-		  {
+		stpcpy (c, "@adlam");
-		    lcnt++;
+	      else
-		    break;
+		return TRUE;
 	    }
-	      if (i < lcnt)
+	}
-		continue;
+
-	      if (lcnt < 32)
-		strcpy (loc_list[lcnt++].loc, loc);
       /* Print */
-	      add_locale (loc, language, country);
+      GetLocaleInfoEx (loc_name, LOCALE_SENGLISHLANGUAGENAME, language, 256);
-	      /* Check for locales which sport a modifier for
+      GetLocaleInfoEx (loc_name, LOCALE_SENGLISHCOUNTRYNAME, country, 256);
+      add_locale (posix_loc, language, country);
+      /* Check for locales sporting an additional modifier for
 	 changing the codeset and other stuff. */
-	      if (lang == LANG_BELARUSIAN
+      if (!wcscmp (iso639, L"be") && !wcscmp (iso3166, L"BY"))
-		  && sublang == SUBLANG_BELARUSIAN_BELARUS)
 	stpcpy (c, "@latin");
-	      else if (lang == LANG_TATAR
+      if (!wcscmp (iso639, L"tt") && !wcscmp (iso3166, L"RU"))
-		       && sublang == SUBLANG_TATAR_RUSSIA)
 	stpcpy (c, "@iqtelif");
-	      else if (GetLocaleInfoW (lcid,
+      else if (GetLocaleInfoEx (loc_name,
 				LOCALE_IDEFAULTANSICODEPAGE
 				| LOCALE_RETURN_NUMBER,
 				(PWCHAR) &cp, sizeof cp)
 	       && cp == 1252 /* Latin1*/
-		       && GetLocaleInfoW (lcid, LOCALE_SINTLSYMBOL, wbuf, 9)
+	       && GetLocaleInfoEx (loc_name, LOCALE_SINTLSYMBOL, currency, 9)
-		       && !wcsncmp (wbuf, L"EUR", 3))
+	       && !wcsncmp (currency, L"EUR", 3))
 	stpcpy (c, "@euro");
-	      else if (lang == LANG_JAPANESE
+      else if (!wcscmp (iso639, L"ja")
-		       || lang == LANG_KOREAN
+	       || !wcscmp (iso639, L"ko")
-		       || lang == LANG_CHINESE)
+	       || !wcscmp (iso639, L"zh"))
 	stpcpy (c, "@cjknarrow");
       else
-		continue;
+	return TRUE;
-	      add_locale (loc, language, country);
+      add_locale (posix_loc, language, country);
-	    }
     }
+  return TRUE;
 }
+
+void
+print_all_locales (int verbose)
+{
+  add_locale ("C", L"C", L"POSIX");
+  add_locale ("POSIX", L"C", L"POSIX", true);
+  EnumSystemLocalesEx (print_all_locales_proc,
+		       LOCALE_WINDOWS | LOCALE_SUPPLEMENTAL,
+		       0, NULL);
   /* First sort allows add_locale_alias_locales to bsearch in locales. */
   qsort (locale, loc_num, sizeof (loc_t), compare_locales);
   add_locale_alias_locales ();
@@ -739,14 +794,13 @@ int
 main (int argc, char **argv)
 {
   int opt;
-  LCID lcid = 0;
+  wchar_t loc_name[256] = { 0 };
   int all = 0;
   int cat = 0;
   int key = 0;
   int maps = 0;
   int verbose = 0;
   const char *utf = "";
-  char name[32];
 
   setlocale (LC_ALL, "");
   while ((opt = getopt_long (argc, argv, opts, longopts, NULL)) != -1)
@@ -765,19 +819,22 @@ main (int argc, char **argv)
 	maps = 1;
 	break;
       case 'i':
-	lcid = (UINT_PTR) GetKeyboardLayout (0) & 0xffff;
+	GetLocaleInfoW ((UINT_PTR) GetKeyboardLayout (0) & 0xffff, LOCALE_SNAME,
+			loc_name, 256);
 	break;
       case 's':
-	lcid = GetSystemDefaultUILanguage ();
+	GetLocaleInfoW (GetSystemDefaultUILanguage (), LOCALE_SNAME,
+			loc_name, 256);
 	break;
       case 'u':
-	lcid = GetUserDefaultUILanguage ();
+	GetLocaleInfoW (GetUserDefaultUILanguage (), LOCALE_SNAME,
+			loc_name, 256);
 	break;
       case 'f':
-	lcid = GetUserDefaultLCID ();
+	GetUserDefaultLocaleName (loc_name, 256);
 	break;
       case 'n':
-	lcid = GetSystemDefaultLCID ();
+	GetSystemDefaultLocaleName (loc_name, 256);
 	break;
       case 'U':
 	utf = ".UTF-8";
@@ -799,10 +856,13 @@ main (int argc, char **argv)
     print_all_locales (verbose);
   else if (maps)
     print_charmaps ();
-  else if (lcid)
+  else if (loc_name[0])
     {
-      if (getlocale (lcid, name))
+      wchar_t iso639[10];
-	printf ("%s%s\n", name, utf);
+      wchar_t iso3166[10];
+
+      if (getlocale (loc_name, iso639, iso3166, NULL))
+	printf ("%ls_%ls%s", iso639, iso3166, utf);
     }
   else if (optind < argc)
     while (optind < argc)