config: Rework handling of extra-languages to change locale format

Accept the locale format as documented by `setlocale(3)`, rather than
another arbitrary format.

This reworks the validation code, and was tested to accept all the
locales on my F30 system using:
```
flatpak config --user --set extra-languages $(locale -a | tr -s '\n' ';' | head -c -1)
```

Signed-off-by: Philip Withnall <withnall@endlessm.com>

app/flatpak-builtins-config.c

@@ -68,49 +68,73 @@ looks_like_a_language (const char *s)
 }
 
 static gboolean
-looks_like_a_region (const char *s)
+looks_like_a_territory (const char *s)
 {
-  g_auto(GStrv) locale = g_strsplit (s, "_", 3);
-  int i;
-  int len;
+  gsize len = strlen (s);
+  gsize i;
 
-  if (!looks_like_a_language(locale[0]))
+  if (len < 2)
     return FALSE;
 
-  if (locale[1] != NULL)
+  for (i = 0; i < len; i++)
     {
-      len = strlen (locale[1]);
-
-      // This can be either GB or Latn
-      if (len < 2 || len > 4)
+      if (!g_ascii_isalpha (s[i]) || !g_ascii_isupper (s[i]))
         return FALSE;
-
-      for (i = 0; i < len; i++)
-        {
-          if ((locale[2] == NULL) && (!g_ascii_isalpha (locale[1][i]) || !g_ascii_isupper (locale[1][i])))
-            return FALSE;
-          else if (!g_ascii_isalpha (locale[1][i]))
-            return FALSE;
-        }
     }
 
-  if (g_strv_length (locale) > 2)
+  return TRUE;
+}
+
+static gboolean
+looks_like_a_codeset_or_modifier (const char *s)
+{
+  gsize len = strlen (s);
+  gsize i;
+
+  if (len < 1)
+    return FALSE;
+
+  for (i = 0; i < len; i++)
     {
-      len = strlen (locale[2]);
-
-      if (len < 2 || len > 3)
+      if (!g_ascii_isalnum (s[i]) && s[i] != '-')
         return FALSE;
-
-      for (i = 0; i < len; i++)
-        {
-          if (!g_ascii_isalpha (locale[2][i]) || !g_ascii_isupper (locale[2][i]))
-            return FALSE;
-        }
     }
 
   return TRUE;
 }
 
+static gboolean
+looks_like_a_locale (const char *s)
+{
+  g_autofree gchar *locale = g_strdup (s);
+  gchar *language, *territory, *codeset, *modifier;
+
+  modifier = strchr (locale, '@');
+  if (modifier != NULL)
+    *modifier++ = '\0';
+
+  codeset = strchr (locale, '.');
+  if (codeset != NULL)
+    *codeset++ = '\0';
+
+  territory = strchr (locale, '_');
+  if (territory != NULL)
+    *territory++ = '\0';
+
+  language = locale;
+
+  if (!looks_like_a_language (language))
+    return FALSE;
+  if (territory != NULL && !looks_like_a_territory (territory))
+    return FALSE;
+  if (codeset != NULL && !looks_like_a_codeset_or_modifier (codeset))
+    return FALSE;
+  if (modifier != NULL && !looks_like_a_codeset_or_modifier (modifier))
+    return FALSE;
+
+  return TRUE;
+}
+
 static char *
 parse_locale (const char *value, GError **error)
 {
@@ -120,7 +144,7 @@ parse_locale (const char *value, GError **error)
   strs = g_strsplit (value, ";", 0);
   for (i = 0; strs[i]; i++)
     {
-      if (!looks_like_a_language (strs[i]) && !looks_like_a_region (strs[i]))
+      if (!looks_like_a_language (strs[i]) && !looks_like_a_locale (strs[i]))
         {
           flatpak_fail (error, _("'%s' does not look like a language/locale code"), strs[i]);
           return NULL;

common/flatpak-dir.c

@@ -14607,9 +14607,10 @@ flatpak_dir_get_default_locale_languages (FlatpakDir *self)
   extra_languages = flatpak_dir_get_config_strv (self, "xa.extra-languages");
   for (i = 0; extra_languages != NULL && extra_languages[i] != NULL; i++)
     {
-       g_auto(GStrv) locales = g_strsplit (extra_languages[i], "_", -1);
-       g_free (extra_languages[i]);
-       extra_languages[i] = g_strdup (locales[0]);
+      /* Strip the locale, modifier or codeset, if present. */
+      gchar *match = strpbrk (extra_languages[i], "._@");
+      if (match != NULL)
+        *match = '\0';
     }
 
   if (flatpak_dir_is_user (self))

common/flatpak-installation.c

@@ -1687,12 +1687,15 @@ flatpak_installation_get_default_languages (FlatpakInstallation  *self,
  * @self: a #FlatpakInstallation
  * @error: return location for a #GError
  *
- * Like flatpak_installation_get_default_languages() but includes region 
- * information (e.g. en_US rather than en) which may be included in the 
- * xa.extra-languages configuration.
+ * Like flatpak_installation_get_default_languages() but includes territory
+ * information (e.g. `en_US` rather than `en`) which may be included in the
+ * `xa.extra-languages` configuration.
+ *
+ * Strings returned by this function are in the format specified by
+ * [`setlocale()`](man:setlocale): `language[_territory][.codeset][@modifier]`.
  *
  * Returns: (array zero-terminated=1) (element-type utf8) (transfer full):
- *   A possibly empty array of language and locale strings, or %NULL on error.
+ *   A possibly empty array of locale strings, or %NULL on error.
  * Since: 1.5.1
  */
 char **