Hmm..., it looks like the problem is not in gnome-obex-send.
There is no character encoding conversion codes in the source code of
gnome-obex-send.
Perhaps, the problem is in libbtctl.
In the function btctl_obexclient_source_push, it calls OBEX_CharToUnicode to
convert a filename to Unicode as following.
libbtctl-0.10.0/src/obex-client-source.c:
gboolean
btctl_obexclient_source_push (BtctlObexclientSource *bc,
gchar *fname, const guchar * data, guint len)
{
...snip...
bfname = g_path_get_basename (fname);
uname_size = (strlen (bfname)+1)*2;
uname = g_malloc (uname_size);
OBEX_CharToUnicode (uname, bfname, uname_size);
OBEX_CharToUnicode converts a string to an Unicode string, however that function
seems not support non-ASCII or multibyte encoding.
Therefore it breaks non-ASCII characters.
I suggest to call g_convert instead of OBEX_CharToUnicode to support non-ASCII
or multibyte encoding.
I have attached a patch.
Regards,
--
Morita Sho <[email protected]>
--- libbtctl-0.10.0/src/obex-client-source.c.orig 2009-02-08 14:27:56.000000000 +0900
+++ libbtctl-0.10.0/src/obex-client-source.c 2009-02-08 14:28:03.000000000 +0900
@@ -307,6 +307,7 @@
guint uname_size;
gchar *uname;
gchar *bfname;
+ const gchar **charsets;
if (!bc->initialised) {
g_warning ("Not initialised.");
@@ -325,9 +326,12 @@
}
bfname = g_path_get_basename (fname);
- uname_size = (strlen (bfname)+1)*2;
- uname = g_malloc (uname_size);
- OBEX_CharToUnicode (uname, bfname, uname_size);
+ g_get_filename_charsets (&charsets);
+ uname = g_convert (bfname, -1, "UTF-16BE", charsets[0], NULL, &uname_size, NULL);
+ if (uname == NULL) {
+ g_warning ("Failed to convert the filename to UTF-16BE");
+ return FALSE;
+ }
hd.bs = uname;
OBEX_ObjectAddHeader (bc->handle, object,
