Merge pull request #3194 from pwithnall/what-is-a-locale

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

app/flatpak-builtins-config.c

@@ -51,7 +51,11 @@ looks_like_a_language (const char *s)
   int len = strlen (s);
   int i;
 
-  if (len < 2 || len > 3)
+  if (g_str_equal (s, "C") ||
+      g_str_equal (s, "POSIX"))
+    return TRUE;
+
+  if (len < 2)
     return FALSE;
 
   for (i = 0; i < len; i++)
@@ -64,49 +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)
 {
@@ -116,7 +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 **

doc/flatpak-config.xml

@@ -61,8 +61,8 @@
                 <listitem><para>
                    The languages that are included when installing Locale extensions.
                    The value is a semicolon-separated list of two-letter language codes,
-                   or one of the special values * or all. If this key is unset, flatpak
-                   defaults to including the extra-languages key and the current locale.
+                   or one of the special values <literal>*</literal> or <literal>all</literal>. If this key is unset, flatpak
+                   defaults to including the <varname>extra-languages</varname> key and the current locale.
                 </para></listitem>
             </varlistentry>
             <varlistentry>
@@ -70,8 +70,10 @@
                 <listitem><para>
                    This key is used when languages is not set, and it defines extra locale
                    extensions on top of the system configured languages. The value is a
-                   semicolon-separated list of two-letter language codes or locale identifiers
-                   (eg. en;en_DK;az_Latn_AZ).
+                   semicolon-separated list of locale identifiers
+                   (language, optional locale, optional codeset, optional modifier) as documented by
+                   <citerefentry><refentrytitle>setlocale</refentrytitle><manvolnum>3</manvolnum></citerefentry>
+                   (for example, <literal>en;en_DK;zh_HK.big5hkscs;uz_UZ.utf8@cyrillic</literal>).
                 </para></listitem>
             </varlistentry>
         </variablelist>

tests/testlibrary.c

@@ -291,27 +291,30 @@ test_languages_config (void)
   g_assert_cmpstr (value[0], ==, "en");
   g_assert_null (value[1]);
 
-  res = flatpak_installation_set_config_sync (inst, "extra-languages", "pt_BR;az_Latn_AZ;es", NULL, &error);
+  res = flatpak_installation_set_config_sync (inst, "extra-languages", "pt_BR;uz_UZ.utf8@cyrillic;es;zh_HK.big5hkscs;uz_UZ@cyrillic", NULL, &error);
   g_assert_no_error (error);
   g_assert_true (res);
 
   value = flatpak_installation_get_default_languages (inst, &error);
   g_assert_no_error (error);
-  g_assert_cmpstr (value[0], ==, "az");
-  g_assert_cmpstr (value[1], ==, "en");
-  g_assert_cmpstr (value[2], ==, "es");
-  g_assert_cmpstr (value[3], ==, "pt");
-  g_assert_null (value[4]);
+  g_assert_cmpstr (value[0], ==, "en");
+  g_assert_cmpstr (value[1], ==, "es");
+  g_assert_cmpstr (value[2], ==, "pt");
+  g_assert_cmpstr (value[3], ==, "uz");
+  g_assert_cmpstr (value[4], ==, "zh");
+  g_assert_null (value[5]);
 
   g_clear_pointer (&value, g_strfreev);
 
   value = flatpak_installation_get_default_locales (inst, &error);
   g_assert_no_error (error);
-  g_assert_cmpstr (value[0], ==, "az_Latn_AZ");
-  g_assert_cmpstr (value[1], ==, "en");
-  g_assert_cmpstr (value[2], ==, "es");
-  g_assert_cmpstr (value[3], ==, "pt_BR");
-  g_assert_null (value[4]);
+  g_assert_cmpstr (value[0], ==, "en");
+  g_assert_cmpstr (value[1], ==, "es");
+  g_assert_cmpstr (value[2], ==, "pt_BR");
+  g_assert_cmpstr (value[3], ==, "uz_UZ.utf8@cyrillic");
+  g_assert_cmpstr (value[4], ==, "uz_UZ@cyrillic");
+  g_assert_cmpstr (value[5], ==, "zh_HK.big5hkscs");
+  g_assert_null (value[6]);
 
   g_clear_pointer (&value, g_strfreev);