/* RFC 3491 IDN cleanup algorithm. */
static gchar *
nameprep (const gchar *hostname,
gint len)
{
gchar *name, *tmp = NULL, *p;
/* It would be nice if we could do this without repeatedly
* allocating strings and converting back and forth between
* gunichars and UTF-8... The code does at least avoid doing most of
* the sub-operations when they would just be equivalent to a
* g_strdup().
*/
/* Remove presentation-only characters */
name = remove_junk (hostname, len);
if (name)
{
tmp = name;
len = -1;
}
else
name = (gchar *)hostname;
/* Convert to lowercase */
if (contains_uppercase_letters (name, len))
{
name = g_utf8_strdown (name, len);
g_free (tmp);
tmp = name;
len = -1;
}
/* If there are no UTF8 characters, we're done. */
if (!contains_non_ascii (name, len))
{
if (name == (gchar *)hostname)
return len == -1 ? g_strdup (hostname) : g_strndup (hostname, len);
else
return name;
}
/* Normalize */
name = g_utf8_normalize (name, len, G_NORMALIZE_NFKC);
g_free (tmp);
tmp = name;
if (!name)
return NULL;
/* KC normalization may have created more capital letters (eg,
* angstrom -> capital A with ring). So we have to lowercasify a
* second time. (This is more-or-less how the nameprep algorithm
* does it. If tolower(nfkc(tolower(X))) is guaranteed to be the
* same as tolower(nfkc(X)), then we could skip the first tolower,
* but I'm not sure it is.)
*/
if (contains_uppercase_letters (name, -1))
{
name = g_utf8_strdown (name, -1);
g_free (tmp);
tmp = name;
}
/* Check for prohibited characters */
for (p = name; *p; p = g_utf8_next_char (p))
{
if (idna_is_prohibited (g_utf8_get_char (p)))
{
name = NULL;
g_free (tmp);
goto done;
}
}
/* FIXME: We're supposed to verify certain constraints on bidi
* characters, but glib does not appear to have that information.
*/
done:
return name;
}
static void remove_junk_on_signal(int signo)
{
remove_junk();
sigchain_pop(signo);
raise(signo);
}
static void remove_junk_on_signal(int signo)
{
remove_junk();
signal(SIGINT, SIG_DFL);
raise(signo);
}