X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fcategoricals.c;h=c544f97e98000ade273bd78785bd9a9a4b924e1b;hb=43b96ca660fb4c57674f1625aad48b83c0170de0;hp=e4feda2896645c0452133199563b3924249e1b1e;hpb=fc4133a6cd4b508fd61571a71cb19ccaea65a7f8;p=pspp

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index e4feda2896..c544f97e98 100644
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -27,6 +27,7 @@
 #include <libpspp/hmap.h>
 #include <libpspp/pool.h>
 
+#include <libpspp/str.h>
 
 struct value_node
 {
@@ -43,6 +44,8 @@ struct var_params
   /* A map indexed by a union values */
   struct hmap map;
 
+  const struct variable *var;
+
   int base_subscript;
 
   /* The number of distinct values of this variable */
@@ -50,25 +53,37 @@ struct var_params
 
   /* A map of values indexed by subscript */
   const struct value_node **reverse_value_map;
+
+  /* Total of the weights of this variable */
+  double cc; 
 };
 
 
 struct categoricals
 {
-  const struct variable **vars;
-  size_t n_vars;
-
+  /* The weight variable */
   const struct variable *wv;
 
+
   /* An array of var_params */
   struct var_params *vp;
 
+  /* The size of VP. (ie, the number of variables involved.) */
+  size_t n_vp;
+
+  /* The number of categorical variables which contain entries.
+     In the absence of missing values, this will be equal to N_VP */
+  size_t n_vars;
+
   /* A map to enable the lookup of variables indexed by subscript */
   int *reverse_variable_map;
 
   size_t n_cats_total;
 
   struct pool *pool;
+
+  /* Missing values to be excluded */
+  enum mv_class exclude;
 };
 
 
@@ -76,47 +91,72 @@ void
 categoricals_destroy ( struct categoricals *cat)
 {
   int i;
-  for (i = 0 ; i < cat->n_vars; ++i)
-    hmap_destroy (&cat->vp[i].map);
-
-  pool_destroy (cat->pool);
-  free (cat);
+  if (cat != NULL)
+    {
+      for (i = 0 ; i < cat->n_vp; ++i)
+	hmap_destroy (&cat->vp[i].map);
+      
+      pool_destroy (cat->pool);
+      free (cat);
+    }
 }
 
 
+#if 0
 void
 categoricals_dump (const struct categoricals *cat)
 {
   int v;
 
-  for (v = 0 ; v < cat->n_vars; ++v)
+  for (v = 0 ; v < cat->n_vp; ++v)
     {
       const struct var_params *vp = &cat->vp[v];
       const struct hmap *m = &vp->map;
-      size_t width = var_get_width (cat->vars[v]);
       struct hmap_node *node ;
       int x;
      
-      printf ("\n%s (%d):\n", var_get_name (cat->vars[v]), vp->base_subscript);
-
-      assert (vp->reverse_value_map);
+      printf ("\n%s (%d)  CC=%g n_cats=%d:\n", 
+	      var_get_name (vp->var), vp->base_subscript, vp->cc, vp->n_cats);
 
       printf ("Reverse map\n");
       for (x = 0 ; x < vp->n_cats; ++x)
 	{
+	  struct string s;
 	  const struct value_node *vn = vp->reverse_value_map[x];
-	  printf ("Value for %d is %s\n", x, value_str (&vn->value, width));
+	  ds_init_empty (&s);
+	  var_append_value_name (vp->var, &vn->value, &s);
+	  printf ("Value for %d is %s\n", x, ds_cstr(&s));
+	  ds_destroy (&s);
 	}
 
       printf ("\nForward map\n");
       for (node = hmap_first (m); node; node = hmap_next (m, node))
 	{
+	  struct string 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",
-		  value_str (&vn->value, width),  vn->subscript, vn->cc);
+		  ds_cstr (&s),
+		  vn->subscript, vn->cc);
+	  ds_destroy (&s);
 	}
     }
+
+  assert (cat->n_vars <= cat->n_vp);
+
+  printf ("\n");
+  printf ("Number of categorical variables: %d\n", cat->n_vp);
+  printf ("Number of non-empty categorical variables: %d\n", cat->n_vars);
+  printf ("Total number of categories: %d\n", cat->n_cats_total);
+
+  printf ("\nReverse variable map:\n");
+
+  for (v = 0 ; v < cat->n_cats_total; ++v)
+    printf ("%d ", cat->reverse_variable_map[v]);
+  printf ("\n");
 }
+#endif
 
 
 
@@ -141,22 +181,27 @@ 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)
+categoricals_create (const struct variable **v, size_t n_vars,
+		     const struct variable *wv, enum mv_class exclude)
 {
   size_t i;
   struct categoricals *cat = xmalloc (sizeof *cat);
   
-  cat->vars = v;
-  cat->n_vars = n_vars;
+  cat->n_vp = n_vars;
   cat->wv = wv;
   cat->n_cats_total = 0;
+  cat->n_vars = 0;
   cat->reverse_variable_map = NULL;
   cat->pool = pool_create ();
+  cat->exclude = exclude;
 
-  cat->vp = pool_calloc (cat->pool, sizeof *cat->vp, n_vars);
+  cat->vp = pool_calloc (cat->pool, cat->n_vp, sizeof *cat->vp);
 
-  for (i = 0 ; i < cat->n_vars; ++i)
-    hmap_init (&cat->vp[i].map);
+  for (i = 0 ; i < cat->n_vp; ++i)
+    {
+      hmap_init (&cat->vp[i].map);
+      cat->vp[i].var = v[i];
+    }
 
   return cat;
 }
@@ -172,27 +217,39 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 
   assert (NULL == cat->reverse_variable_map);
 
-  for (i = 0 ; i < cat->n_vars; ++i)
+  for (i = 0 ; i < cat->n_vp; ++i)
     {
-      unsigned int width = var_get_width (cat->vars[i]);
-      const union value *val = case_data (c, cat->vars[i]);
-      size_t hash = value_hash (val, width, 0);
+      const struct variable *var = cat->vp[i].var;
+      unsigned int width = var_get_width (var);
+      const union value *val = case_data (c, var);
+      size_t hash;
+      struct value_node *node ;
+
+      if ( var_is_value_missing (var, val, cat->exclude))
+	continue;
+
+      hash = value_hash (val, width, 0);
+      node = lookup_value (&cat->vp[i].map, var, val);
 
-      struct value_node  *node = lookup_value (&cat->vp[i].map, cat->vars[i], val);
       if ( NULL == node)
 	{
-	  node = pool_calloc (cat->pool, sizeof *node, 1);
+	  node = pool_malloc (cat->pool, sizeof *node);
 
 	  value_init (&node->value, width);
 	  value_copy (&node->value, val, width);
 	  node->cc = 0.0;
 
 	  hmap_insert (&cat->vp[i].map, &node->node,  hash);
-	  cat->n_cats_total ++;
+	  cat->n_cats_total++;
+	  
+	  if ( 0 == cat->vp[i].n_cats)
+	    cat->n_vars++;
+
 	  node->subscript = cat->vp[i].n_cats++ ;
 	}
 
       node->cc += weight;
+      cat->vp[i].cc += weight;
     }
 }
 
@@ -204,19 +261,6 @@ categoricals_n_count (const struct categoricals *cat, size_t n)
 }
 
 
-/* Return the index for value VAL in the Nth variable */
-int
-categoricals_index (const struct categoricals *cat, size_t n, const union value *val)
-{
-  struct value_node *vn = lookup_value (&cat->vp[n].map, cat->vars[n], val);
-
-  if ( vn == NULL)
-    return -1;
-
-  return vn->subscript;
-}
-
-
 /* Return the total number of categories */
 size_t
 categoricals_total (const struct categoricals *cat)
@@ -238,16 +282,18 @@ categoricals_done (struct categoricals *cat)
   */
   int v;
   int idx = 0;
-  cat->reverse_variable_map = pool_calloc (cat->pool, sizeof *cat->reverse_variable_map, cat->n_cats_total);
+  cat->reverse_variable_map = pool_calloc (cat->pool,
+					   cat->n_cats_total,
+					   sizeof *cat->reverse_variable_map);
   
-  for (v = 0 ; v < cat->n_vars; ++v)
+  for (v = 0 ; v < cat->n_vp; ++v)
     {
       int i;
       struct var_params *vp = &cat->vp[v];
       int n_cats_total = categoricals_n_count (cat, v);
       struct hmap_node *node ;
 
-      vp->reverse_value_map = pool_calloc (cat->pool, sizeof *vp->reverse_value_map,  n_cats_total);
+      vp->reverse_value_map = pool_calloc (cat->pool, n_cats_total, sizeof *vp->reverse_value_map);
 
       vp->base_subscript = idx;
 
@@ -257,32 +303,41 @@ categoricals_done (struct categoricals *cat)
 	  vp->reverse_value_map[vn->subscript] = vn;
 	}
 
+      /* Populate the reverse variable map.
+       */
       for (i = 0; i < vp->n_cats; ++i)
 	cat->reverse_variable_map[idx++] = v;
     }
+
+  assert (cat->n_vars <= cat->n_vp);
 }
 
 
+static int
+reverse_variable_lookup (const struct categoricals *cat, int subscript)
+{
+  assert (cat->reverse_variable_map);
+  assert (subscript >= 0);
+  assert (subscript < cat->n_cats_total);
+
+  return cat->reverse_variable_map[subscript];
+}
+
 
 /* Return the categorical variable corresponding to SUBSCRIPT */
 const struct variable *
 categoricals_get_variable_by_subscript (const struct categoricals *cat, int subscript)
 {
-  int index;
-
-  assert (cat->reverse_variable_map);
-  
-  index = cat->reverse_variable_map[subscript];
+  int index = reverse_variable_lookup (cat, subscript);
 
-  return cat->vars[index];
+  return cat->vp[index].var;
 }
 
-
 /* Return the value corresponding to SUBSCRIPT */
 const union value *
 categoricals_get_value_by_subscript (const struct categoricals *cat, int subscript)
 {
-  int vindex = cat->reverse_variable_map[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];
 
@@ -290,6 +345,26 @@ categoricals_get_value_by_subscript (const struct categoricals *cat, int subscri
 }
 
 
+double
+categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscript)
+{
+  int vindex = reverse_variable_lookup (cat, subscript);
+  const struct var_params *vp = &cat->vp[vindex];
+
+  return vp->cc;
+}
+
+double
+categoricals_get_sum_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->cc;
+}
+
+
 /* Returns unity if the value in case C at SUBSCRIPT is equal to the category
    for that subscript */
 double
@@ -303,3 +378,10 @@ categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscr
 
   return value_equal (val, categoricals_get_value_by_subscript (cat, subscript), width);
 }
+
+
+size_t
+categoricals_get_n_variables (const struct categoricals *cat)
+{
+  return cat->n_vars;
+}