exchange

commit a369995a792e3a9618c5c8f8e6b257f406df9f5f
parent f194a76d391d9a3c7c0d7a0070759fc1809ad082
Author: Christian Grothoff <grothoff@gnunet.org>
Date:   Sun, 16 Aug 2026 08:28:56 +0200

better language code check

Diffstat:
M
src/json/i18n.c
 | 
92
++++++++++++++++++++++++++++++++++++++++----------------------------------------

1 file changed, 46 insertions(+), 46 deletions(-)
diff --git a/src/json/i18n.c b/src/json/i18n.c
@@ -67,6 +67,51 @@ TALER_JSON_extract_i18n (const json_t *object,
 }
 
 
+/**
+ * Check that @a tag is a valid language tag.  Accepted are a
+ * two- or three-letter language code (ISO 639-1, 639-2 or 639-3),
+ * optionally followed by '_' or '-' and a two-letter country code
+ * (ISO 3166-1 alpha-2) or a three-digit region code (UN M.49).
+ * We do not care about capitalization.  For the syntax, see the
+ * GNU Gettext manual, including appendix A for rare language
+ * codes.
+ *
+ * @param tag language tag to check
+ * @return true if @a tag is well-formed
+ */
+static bool
+check_language_tag (const char *tag)
+{
+  size_t len = strlen (tag);
+  size_t off = 0;
+
+  while ( (off < len) &&
+          (0 != isalpha ((unsigned char) tag[off])) )
+    off++;
+  if ( (off < 2) ||
+       (off > 3) )
+    return false; /* language code must have 2 or 3 letters */
+  if (off == len)
+    return true; /* language code without country code */
+  if ( ('_' != tag[off]) &&
+       ('-' != tag[off]) )
+    return false;
+  off++;
+  if (2 == len - off)
+  {
+    return ( (0 != isalpha ((unsigned char) tag[off])) &&
+             (0 != isalpha ((unsigned char) tag[off + 1])) );
+  }
+  if (3 == len - off)
+  {
+    return ( (0 != isdigit ((unsigned char) tag[off])) &&
+             (0 != isdigit ((unsigned char) tag[off + 1])) &&
+             (0 != isdigit ((unsigned char) tag[off + 2])) );
+  }
+  return false;
+}
+
+
 bool
 TALER_JSON_check_i18n (const json_t *i18n)
 {
@@ -79,53 +124,8 @@ TALER_JSON_check_i18n (const json_t *i18n)
   {
     if (! json_is_string (member))
       return false;
-    /* Field name must be either of format "en_UK"
-       or just "en"; we do not care about capitalization;
-       for syntax, see GNU Gettext manual, including
-       appendix A for rare language codes. */
-    switch (strlen (field))
-    {
-    case 0:
-    case 1:
+    if (! check_language_tag (field))
       return false;
-    case 2:
-      if (! isalpha ((unsigned char) field[0]))
-        return false;
-      if (! isalpha ((unsigned char) field[1]))
-        return false;
-      break;
-    case 3:
-    case 4:
-      return false;
-    case 5:
-      if (! isalpha ((unsigned char) field[0]))
-        return false;
-      if (! isalpha ((unsigned char) field[1]))
-        return false;
-      if ('_' != field[2])
-        return false;
-      if (! isalpha ((unsigned char) field[3]))
-        return false;
-      if (! isalpha ((unsigned char) field[4]))
-        return false;
-      break;
-    case 6:
-      if (! isalpha ((unsigned char) field[0]))
-        return false;
-      if (! isalpha ((unsigned char) field[1]))
-        return false;
-      if ('_' != field[2])
-        return false;
-      if (! isalpha ((unsigned char) field[3]))
-        return false;
-      if (! isalpha ((unsigned char) field[4]))
-        return false;
-      if (! isalpha ((unsigned char) field[5]))
-        return false;
-      break;
-    default:
-      return false;
-    }
   }
   return true;
 }