X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fdictionary.c;h=2d99f6d72cfe2a7e6dc442102f16a4dc687cbe32;hb=refs%2Fbuilds%2F20110914030501%2Fpspp;hp=79d36374fc42767660234911ed533c5732270b92;hpb=9ade26c8349b4434008c46cf09bc7473ec743972;p=pspp

diff --git a/src/data/dictionary.c b/src/data/dictionary.c
index 79d36374fc..2d99f6d72c 100644
--- a/src/data/dictionary.c
+++ b/src/data/dictionary.c
@@ -87,16 +87,8 @@ struct dictionary
 static void dict_unset_split_var (struct dictionary *, struct variable *);
 static void dict_unset_mrset_var (struct dictionary *, struct variable *);
 
-void
-dict_set_encoding (struct dictionary *d, const char *enc)
-{
-  if (enc)
-    {
-      free (d->encoding);
-      d->encoding = xstrdup (enc);
-    }
-}
-
+/* Returns the encoding for data in dictionary D.  The return value is a
+   nonnull string that contains an IANA character set name. */
 const char *
 dict_get_encoding (const struct dictionary *d)
 {
@@ -171,14 +163,16 @@ dict_copy_callbacks (struct dictionary *dest,
   dest->cb_data = src->cb_data;
 }
 
-/* Creates and returns a new dictionary. */
+/* Creates and returns a new dictionary with the specified ENCODING. */
 struct dictionary *
-dict_create (void)
+dict_create (const char *encoding)
 {
   struct dictionary *d = xzalloc (sizeof *d);
 
+  d->encoding = xstrdup (encoding);
   hmap_init (&d->name_map);
   attrset_init (&d->attributes);
+
   return d;
 }
 
@@ -189,14 +183,20 @@ dict_create (void)
    dictionary.  If the new dictionary won't be used to access
    cases produced with the old dictionary, then the new
    dictionary's case indexes should be compacted with
-   dict_compact_values to save space. */
+   dict_compact_values to save space.
+
+   Callbacks are not cloned. */
 struct dictionary *
 dict_clone (const struct dictionary *s)
 {
   struct dictionary *d;
   size_t i;
 
-  d = dict_create ();
+  d = dict_create (s->encoding);
+
+  /* Set the new dictionary's encoding early so that string length limitations
+     are interpreted correctly. */
+  d->encoding = xstrdup (s->encoding);
 
   for (i = 0; i < s->var_cnt; i++)
     {
@@ -235,9 +235,6 @@ dict_clone (const struct dictionary *s)
   for (i = 0; i < s->vector_cnt; i++)
     d->vector[i] = vector_clone (s->vector[i], s, d);
 
-  if ( s->encoding)
-    d->encoding = xstrdup (s->encoding);
-
   dict_set_attributes (d, dict_get_attributes (s));
 
   for (i = 0; i < s->n_mrsets; i++)
@@ -588,7 +585,7 @@ reindex_vars (struct dictionary *d, size_t from, size_t to)
 /* Deletes variable V from dictionary D and frees V.
 
    This is a very bad idea if there might be any pointers to V
-   from outside D.  In general, no variable in the active file's
+   from outside D.  In general, no variable in the active dataset's
    dictionary should be deleted when any transformations are
    active on the dictionary's dataset, because those
    transformations might reference the deleted variable.  The
@@ -1658,7 +1655,7 @@ struct variable *
 dict_create_internal_var (int case_idx, int width)
 {
   if (internal_dict == NULL)
-    internal_dict = dict_create ();
+    internal_dict = dict_create ("UTF-8");
 
   for (;;)
     {