Ticket #3783: mc-3783-mcview-invalid-utf8.patch

src/viewer/ascii.c

@@ -341 +341 @@
  * Normally: stores c, updates state, returns TRUE.
  * At EOF: state is unchanged, c is undefined, returns FALSE.
  *
+ * Just as with mcview_get_utf(), invalid UTF-8 is reported using negative integers.
+ *
  * Also, temporary hack: handle force_max here.
  * TODO: move it to lower layers (datasource.c)?
  */

src/viewer/datasource.c

@@ -147 +147 @@
 
 /* --------------------------------------------------------------------------------------------- */
 
+/* Invalid UTF-8 is reported as negative integers (one for each byte),
+ * see ticket 3783. */
 gboolean
 mcview_get_utf (WView * view, off_t byte_index, int *ch, int *ch_len)
 {
@@ -200 +202 @@
 
     if (res < 0)
     {
-        *ch = (unsigned char) (*str);
+        /* Implicit conversion from signed char to signed int keeps negative values. */
+        *ch = *str;
         *ch_len = 1;
     }
     else