* lib: cleanup header fields a bit more; fixes #209

lib/mu-msg-file.c

@@ -592,28 +592,34 @@ get_tags (MuMsgFile *self)
 }
 
 
-/* wrongly encoded messages may cause GMime to return invalid
- * UTF8... we double check, and ensure our output is always correct
- * utf8 */
-gchar *
-maybe_cleanup (const char* str, const char *path, gboolean *do_free)
+static char*
+cleanup_maybe (const char *str, gboolean *do_free)
 {
-	if (!str || G_LIKELY(g_utf8_validate(str, -1, NULL)))
-		return (char*)str;
+	char *cur, *s;
 
-	g_debug ("invalid utf8 in %s", path);
+	if (!str)
+		return NULL;
 
+	if (!g_utf8_validate(str, -1, NULL)) {
 		if (*do_free)
-		return mu_str_asciify_in_place ((char*)str);
+			s = mu_str_asciify_in_place ((char*)str);
 		else {
-		gchar *ascii;
-		ascii = mu_str_asciify_in_place(g_strdup (str));
 			*do_free = TRUE;
-		return ascii;
+			s = mu_str_asciify_in_place(g_strdup (str));
 		}
+	} else
+		s = (char*)str;
+
+	/* strip control chars */
+	for (cur = s; *cur; ++cur)
+		if (iscntrl(*cur))
+			*cur = ' ';
+
+	return s;
 }
 
 
+
 G_GNUC_CONST static GMimeRecipientType
 recipient_type (MuMsgFieldId mfid)
 {
@@ -659,9 +665,8 @@ mu_msg_file_get_str_field (MuMsgFile *self, MuMsgFieldId mfid,
 		return get_recipient (self, recipient_type(mfid));
 
 	case MU_MSG_FIELD_ID_FROM:
-		return (char*)maybe_cleanup
-			(g_mime_message_get_sender (self->_mime_msg),
-			 self->_path, do_free);
+		return (char*)cleanup_maybe
+			(g_mime_message_get_sender (self->_mime_msg), do_free);
 
 	case MU_MSG_FIELD_ID_PATH: return self->_path;
 
@@ -670,9 +675,8 @@ mu_msg_file_get_str_field (MuMsgFile *self, MuMsgFieldId mfid,
 		return (char*)get_mailing_list (self);
 
 	case MU_MSG_FIELD_ID_SUBJECT:
-		return (char*)maybe_cleanup
-			(g_mime_message_get_subject (self->_mime_msg),
-			 self->_path, do_free);
+		return (char*)cleanup_maybe
+			(g_mime_message_get_subject (self->_mime_msg), do_free);
 
 	case MU_MSG_FIELD_ID_MSGID:
 		return get_msgid (self, do_free);

lib/mu-msg-sexp.c

@@ -17,6 +17,7 @@
 **
 */
 #include <string.h>
+#include <ctype.h>
 
 #include "mu-str.h"
 #include "mu-msg.h"
@@ -49,12 +50,18 @@ append_sexp_attr_list (GString *gstr, const char* elm, const GSList *lst)
 static void
 append_sexp_attr (GString *gstr, const char* elm, const char *str)
 {
-	gchar *esc;
+	gchar *esc, *utf8, *cur;
 
 	if (!str || strlen(str) == 0)
 		return; /* empty: don't include */
 
-	esc = mu_str_escape_c_literal (str, TRUE);
+	utf8 = mu_str_utf8ify (str);
+	for (cur = utf8; *cur; ++cur)
+		if (iscntrl(*cur))
+			*cur = ' ';
+
+	esc = mu_str_escape_c_literal (utf8, TRUE);
+	g_free (utf8);
 
 	g_string_append_printf (gstr, "\t:%s %s\n", elm, esc);
 	g_free (esc);