Categoricals: Use moments instead of keeping cc count ourselves

[pspp] / src / math / categoricals.c

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 33e3b51c62f99cc2290ce7918dbba6aedb111878..576e19b1d322be30f2ee878aeda65e33083347d2 100644 (file)
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -34,6 +34,9 @@ struct value_node
   struct hmap_node node;      /* Node in hash map. */
   union value value;          /* The value being labeled. */
   double cc;                  /* The total of the weights of cases with this value */
+
+  void *user_data;            /* A pointer to data which the caller can store stuff */
+
   int subscript;              /* A zero based integer, unique within the variable.
                                 Can be used as an index into an array */
 };
@@ -84,6 +87,16 @@ struct categoricals
 
   /* Missing values to be excluded */
   enum mv_class exclude;
+
+  /* Function to be called on each update */
+  update_func *update;
+
+  /* Function specified by the caller to create user_data */
+  user_data_create_func *user_data_create;
+
+  /* Auxilliary data to be passed to update and user_data_create_func*/
+  void *aux1;
+  void *aux2;
 };
 
 
@@ -102,6 +115,7 @@ categoricals_destroy ( struct categoricals *cat)
 }
 
 
+#if 1
 void
 categoricals_dump (const struct categoricals *cat)
 {
@@ -132,8 +146,8 @@ categoricals_dump (const struct categoricals *cat)
       for (node = hmap_first (m); node; node = hmap_next (m, node))
        {
          struct string s;
-         ds_init_empty (&s);
          const struct value_node *vn = HMAP_DATA (node, struct value_node, node);
+         ds_init_empty (&s);
          var_append_value_name (vp->var, &vn->value, &s);
          printf ("Value: %s; Index %d; CC %g\n",
                  ds_cstr (&s),
@@ -151,10 +165,11 @@ categoricals_dump (const struct categoricals *cat)
 
   printf ("\nReverse variable map:\n");
 
-  for (v = 0 ; v < cat->n_cats_total - cat->n_vars; ++v)
+  for (v = 0 ; v < cat->n_cats_total; ++v)
     printf ("%d ", cat->reverse_variable_map[v]);
   printf ("\n");
 }
+#endif
 
 
 
@@ -177,10 +192,12 @@ lookup_value (const struct hmap *map, const struct variable *var, const union va
 }
 
 
-
 struct categoricals *
-categoricals_create (const struct variable **v, size_t n_vars,
-                    const struct variable *wv, enum mv_class exclude)
+categoricals_create (const struct variable *const *v, size_t n_vars,
+                    const struct variable *wv, enum mv_class exclude,
+                    user_data_create_func *udf,
+                    update_func *update, void *aux1, void *aux2
+                    )
 {
   size_t i;
   struct categoricals *cat = xmalloc (sizeof *cat);
@@ -192,6 +209,12 @@ categoricals_create (const struct variable **v, size_t n_vars,
   cat->reverse_variable_map = NULL;
   cat->pool = pool_create ();
   cat->exclude = exclude;
+  cat->update = update;
+  cat->user_data_create = udf;
+
+  cat->aux1 = aux1;
+  cat->aux2 = aux2;
+
 
   cat->vp = pool_calloc (cat->pool, cat->n_vp, sizeof *cat->vp);
 
@@ -244,10 +267,16 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
            cat->n_vars++;
 
          node->subscript = cat->vp[i].n_cats++ ;
+
+         if (cat->user_data_create)
+           node->user_data = cat->user_data_create (cat->aux1, cat->aux2);
        }
 
       node->cc += weight;
       cat->vp[i].cc += weight;
+
+      if (cat->update)
+       cat->update (node->user_data, cat->exclude, cat->wv, var, c, cat->aux1, cat->aux2);
     }
 }
 
@@ -281,7 +310,7 @@ categoricals_done (struct categoricals *cat)
   int v;
   int idx = 0;
   cat->reverse_variable_map = pool_calloc (cat->pool,
-                                          cat->n_cats_total - cat->n_vars,
+                                          cat->n_cats_total,
                                           sizeof *cat->reverse_variable_map);
   
   for (v = 0 ; v < cat->n_vp; ++v)
@@ -302,13 +331,14 @@ categoricals_done (struct categoricals *cat)
        }
 
       /* Populate the reverse variable map.
-        This implementation considers the first value of each categorical variable
-        as the basis.  Therefore, this loop starts from 1 instead of 0 */
-      for (i = 1; i < vp->n_cats; ++i)
+       */
+      for (i = 0; i < vp->n_cats; ++i)
        cat->reverse_variable_map[idx++] = v;
     }
 
   assert (cat->n_vars <= cat->n_vp);
+
+  //  categoricals_dump (cat);
 }
 
 
@@ -317,7 +347,7 @@ reverse_variable_lookup (const struct categoricals *cat, int subscript)
 {
   assert (cat->reverse_variable_map);
   assert (subscript >= 0);
-  assert (subscript < cat->n_cats_total - cat->n_vars);
+  assert (subscript < cat->n_cats_total);
 
   return cat->reverse_variable_map[subscript];
 }
@@ -384,3 +414,15 @@ categoricals_get_n_variables (const struct categoricals *cat)
 {
   return cat->n_vars;
 }
+
+
+
+void *
+categoricals_get_user_data_by_subscript (const struct categoricals *cat, int subscript)
+{
+  int vindex = reverse_variable_lookup (cat, subscript);
+  const struct var_params *vp = &cat->vp[vindex];
+
+  const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript];
+  return vn->user_data;
+}