[Evolution-hackers] Non-UTF8 encodings in LDAP data

[Scott MacLean <scott nit ca>]

I've recently found a problem with the Evolution address book's 
handling of LDAP data that is not UTF8-encoded: any LDAP attribute
string containing non-UTF8 characters is truncated at the first such
character.

To see this in action, create an LDAP entry with a name containing
non-UTF8, non-ASCII characters. Pointing Evolution to this directory
server will result in the name being truncated - this affects both the
book and autocomplete views.

I have built (and attached) a patch against Evolution 1.4.5 that works
(read 'kludges') around this by assuming that an invalid UTF8 string is
ISO8859-1-encoded and converting it to UTF8, but I am unsure how to go
about determining a string's encoding and converting it appropriately in
the general case.

Any advice or pointers would be much appreciated.

Thanks,

Scott MacLean
<scott nit ca>

diff -uNr evolution-1.4.5.orig/addressbook/backend/ebook/e-card-simple.c evolution-1.4.5/addressbook/backend/ebook/e-card-simple.c
--- evolution-1.4.5.orig/addressbook/backend/ebook/e-card-simple.c	2004-02-02 16:39:52.000000000 -0500
+++ evolution-1.4.5/addressbook/backend/ebook/e-card-simple.c	2004-02-02 16:43:29.000000000 -0500
@@ -1077,8 +1077,14 @@
 	ECardSimpleInternalType type = field_data[field].type;
 	ECardAddrLabel *address;
 	ECardPhone *phone;
+	char *data_cache = data;
 	int style;
 	simple->changed = TRUE;
+
+	/* assume non-UTF strings are Western European for LDAP's sake? */
+	if (!g_utf8_validate(data, -1, NULL))
+		data = e_utf8_from_charset_string("iso-8859-1", data);
+	
 	switch (field) {
 	case E_CARD_SIMPLE_FIELD_FULL_NAME:
 	case E_CARD_SIMPLE_FIELD_ORG:
@@ -1135,6 +1141,9 @@
 		}
 		break;
 	}
+
+	if (data != data_cache)
+		g_free(data);
 }
 					     
 ECardSimpleType e_card_simple_type       (ECardSimple          *simple,