X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;ds=sidebyside;f=src%2Fmath%2Fcategoricals.c;h=218ac6c180701c78ea424eaa25b6a46fa3f8c1e6;hb=dc32b2366bbb293a81b57a9a19285188db28c881;hp=78944fca44803aabe3da896e8ac23e3f430257b0;hpb=ed38ada34331b3b1e0167c350b375a3fb38099a2;p=pspp-builds.git

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 78944fca..218ac6c1 100644
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -17,6 +17,7 @@
 #include <config.h>
 
 #include "math/categoricals.h"
+#include "math/interaction.h"
 
 #include <stdio.h>
 
@@ -33,7 +34,7 @@
 struct value_node
 {
   struct hmap_node node;      /* Node in hash map. */
-  union value value;          /* The value being labeled. */
+  struct ccase *ccase;
   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 */
@@ -42,12 +43,12 @@ struct value_node
 				 Can be used as an index into an array */
 };
 
-struct var_params
+struct interact_params
 {
   /* A map indexed by a union values */
   struct hmap map;
 
-  const struct variable *var;
+  const struct interaction *iact;
 
   int base_subscript_short;
   int base_subscript_long;
@@ -69,9 +70,9 @@ compare_value_node (const void *vn1_, const void *vn2_, const void *aux)
 {
   const struct value_node * const *vn1 = vn1_;
   const struct value_node * const *vn2 = vn2_;
-  const struct var_params *vp = aux;
+  const struct interact_params *vp = aux;
 
-  return value_compare_3way (&(*vn1)->value, &(*vn2)->value, var_get_width (vp->var));
+  return interaction_case_cmp_3way (vp->iact, (*vn1)->ccase, (*vn2)->ccase);
 }
 
 
@@ -80,15 +81,14 @@ struct categoricals
   /* The weight variable */
   const struct variable *wv;
 
+  /* An array of interact_params */
+  struct interact_params *iap;
 
-  /* An array of var_params */
-  struct var_params *vp;
-
-  /* The size of VP. (ie, the number of variables involved.) */
-  size_t n_vp;
+  /* The size of IAP. (ie, the number of interactions involved.) */
+  size_t n_iap;
 
   /* The number of categorical variables which contain entries.
-     In the absence of missing values, this will be equal to N_VP */
+     In the absence of missing values, this will be equal to N_IAP */
   size_t n_vars;
 
   /* A map to enable the lookup of variables indexed by subscript.
@@ -124,8 +124,18 @@ categoricals_destroy ( struct categoricals *cat)
   int i;
   if (cat != NULL)
     {
-      for (i = 0 ; i < cat->n_vp; ++i)
-	hmap_destroy (&cat->vp[i].map);
+      for (i = 0 ; i < cat->n_iap; ++i)
+	{
+	  struct hmap *map = &cat->iap[i].map;
+	  struct value_node *nn;
+
+	  HMAP_FOR_EACH (nn, struct value_node, node, map)
+	    {
+	      case_unref (nn->ccase);
+	    }	  
+	  
+	  hmap_destroy (map);
+	}
       
       pool_destroy (cat->pool);
       free (cat);
@@ -139,16 +149,21 @@ categoricals_dump (const struct categoricals *cat)
 {
   int v;
 
-  for (v = 0 ; v < cat->n_vp; ++v)
+  for (v = 0 ; v < cat->n_iap; ++v)
     {
-      const struct var_params *vp = &cat->vp[v];
+      const struct interact_params *vp = &cat->iap[v];
       const struct hmap *m = &vp->map;
       struct hmap_node *node ;
       int x;
-     
+      struct string str ;
+      ds_init_empty (&str);
+
+      interaction_to_string (vp->iact, &str);
       printf ("\n%s (%d)  CC=%g n_cats=%d:\n", 
-	      var_get_name (vp->var), vp->base_subscript_long, vp->cc, vp->n_cats);
+	      ds_cstr (&str),
+	      vp->base_subscript_long, vp->cc, vp->n_cats);
 
+#if 0
       printf ("Reverse map\n");
       for (x = 0 ; x < vp->n_cats; ++x)
 	{
@@ -172,12 +187,13 @@ categoricals_dump (const struct categoricals *cat)
 		  vn->subscript, vn->cc);
 	  ds_destroy (&s);
 	}
+#endif
     }
 
-  assert (cat->n_vars <= cat->n_vp);
+  assert (cat->n_vars <= cat->n_iap);
 
   printf ("\n");
-  printf ("Number of categorical variables: %d\n", cat->n_vp);
+  printf ("Number of interactions: %d\n", cat->n_iap);
   printf ("Number of non-empty categorical variables: %d\n", cat->n_vars);
   printf ("Total number of categories: %d\n", cat->n_cats_total);
 
@@ -194,28 +210,26 @@ categoricals_dump (const struct categoricals *cat)
 #endif
 
 
-
 static struct value_node *
-lookup_value (const struct hmap *map, const struct variable *var, const union value *val)
+lookup_case (const struct hmap *map, const struct interaction *iact, const struct ccase *c)
 {
-  struct value_node *foo;
-  unsigned int width = var_get_width (var);
-  size_t hash = value_hash (val, width, 0);
+  struct value_node *nn;
+  size_t hash = interaction_case_hash (iact, c);
 
-  HMAP_FOR_EACH_WITH_HASH (foo, struct value_node, node, hash, map)
+  HMAP_FOR_EACH_WITH_HASH (nn, struct value_node, node, hash, map)
     {
-      if (value_equal (val, &foo->value, width))
+      if (interaction_case_equal (iact, c, nn->ccase))
 	break;
 
       fprintf (stderr, "Warning: Hash table collision\n");
     }
 
-  return foo;
+  return nn;
 }
 
 
 struct categoricals *
-categoricals_create (const struct variable *const *v, size_t n_vars,
+categoricals_create (const struct interaction **inter, size_t n_inter,
 		     const struct variable *wv, enum mv_class exclude,
 		     user_data_create_func *udf,
 		     update_func *update, void *aux1, void *aux2
@@ -224,7 +238,7 @@ categoricals_create (const struct variable *const *v, size_t n_vars,
   size_t i;
   struct categoricals *cat = xmalloc (sizeof *cat);
   
-  cat->n_vp = n_vars;
+  cat->n_iap = n_inter;
   cat->wv = wv;
   cat->n_cats_total = 0;
   cat->n_vars = 0;
@@ -239,12 +253,12 @@ categoricals_create (const struct variable *const *v, size_t n_vars,
   cat->aux2 = aux2;
 
 
-  cat->vp = pool_calloc (cat->pool, cat->n_vp, sizeof *cat->vp);
+  cat->iap = pool_calloc (cat->pool, cat->n_iap, sizeof *cat->iap);
 
-  for (i = 0 ; i < cat->n_vp; ++i)
+  for (i = 0 ; i < cat->n_iap; ++i)
     {
-      hmap_init (&cat->vp[i].map);
-      cat->vp[i].var = v[i];
+      hmap_init (&cat->iap[i].map);
+      cat->iap[i].iact = inter[i];
     }
 
   return cat;
@@ -262,45 +276,42 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
   assert (NULL == cat->reverse_variable_map_short);
   assert (NULL == cat->reverse_variable_map_long);
 
-  for (i = 0 ; i < cat->n_vp; ++i)
+  for (i = 0 ; i < cat->n_iap; ++i)
     {
-      const struct variable *var = cat->vp[i].var;
-      unsigned int width = var_get_width (var);
-      const union value *val = case_data (c, var);
+      const struct interaction *iact = cat->iap[i].iact;
       size_t hash;
       struct value_node *node ;
 
-      if ( var_is_value_missing (var, val, cat->exclude))
+      if ( interaction_case_is_missing (iact, c, cat->exclude))
 	continue;
 
-      hash = value_hash (val, width, 0);
-      node = lookup_value (&cat->vp[i].map, var, val);
+      hash = interaction_case_hash (iact, c);
+      node = lookup_case (&cat->iap[i].map, iact, c);
 
       if ( NULL == node)
 	{
 	  node = pool_malloc (cat->pool, sizeof *node);
 
-	  value_init (&node->value, width);
-	  value_copy (&node->value, val, width);
+	  node->ccase = case_ref (c);
 	  node->cc = 0.0;
 
-	  hmap_insert (&cat->vp[i].map, &node->node,  hash);
+	  hmap_insert (&cat->iap[i].map, &node->node,  hash);
 	  cat->n_cats_total++;
 	  
-	  if ( 0 == cat->vp[i].n_cats)
+	  if ( 0 == cat->iap[i].n_cats)
 	    cat->n_vars++;
 
-	  node->subscript = cat->vp[i].n_cats++ ;
+	  node->subscript = cat->iap[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;
+      cat->iap[i].cc += weight;
 
       if (cat->update)
-	cat->update (node->user_data, cat->exclude, cat->wv, var, c, cat->aux1, cat->aux2);
+	cat->update (node->user_data, cat->exclude, cat->wv, NULL, c, cat->aux1, cat->aux2);
     }
 }
 
@@ -308,7 +319,7 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 size_t
 categoricals_n_count (const struct categoricals *cat, size_t n)
 {
-  return hmap_count (&cat->vp[n].map);
+  return hmap_count (&cat->iap[n].map);
 }
 
 
@@ -343,10 +354,10 @@ categoricals_done (const struct categoricals *cat_)
 						cat->n_cats_total,
 						sizeof *cat->reverse_variable_map_long);
   
-  for (v = 0 ; v < cat->n_vp; ++v)
+  for (v = 0 ; v < cat->n_iap; ++v)
     {
       int i;
-      struct var_params *vp = &cat->vp[v];
+      struct interact_params *vp = &cat->iap[v];
       int n_cats_total = categoricals_n_count (cat, v);
       struct hmap_node *node ;
 
@@ -373,7 +384,7 @@ categoricals_done (const struct categoricals *cat_)
 	cat->reverse_variable_map_long[idx_long++] = v;
     }
 
-  assert (cat->n_vars <= cat->n_vp);
+  assert (cat->n_vars <= cat->n_iap);
 }
 
 
@@ -398,25 +409,25 @@ reverse_variable_lookup_long (const struct categoricals *cat, int subscript)
 }
 
 
-
-/* Return the categorical variable corresponding to SUBSCRIPT */
-const struct variable *
-categoricals_get_variable_by_subscript (const struct categoricals *cat, int subscript)
+/* Return the interaction corresponding to SUBSCRIPT */
+const struct interaction *
+categoricals_get_interaction_by_subscript (const struct categoricals *cat, int subscript)
 {
   int index = reverse_variable_lookup_short (cat, subscript);
 
-  return cat->vp[index].var;
+  return cat->iap[index].iact;
 }
 
-/* Return the value corresponding to SUBSCRIPT */
-static const union value *
-categoricals_get_value_by_subscript (const struct categoricals *cat, int subscript)
+
+/* Return the case corresponding to SUBSCRIPT */
+static const struct ccase *
+categoricals_get_case_by_subscript (const struct categoricals *cat, int subscript)
 {
   int vindex = reverse_variable_lookup_short (cat, subscript);
-  const struct var_params *vp = &cat->vp[vindex];
+  const struct interact_params *vp = &cat->iap[vindex];
   const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_short];
 
-  return &vn->value;
+  return vn->ccase;
 }
 
 
@@ -424,7 +435,7 @@ double
 categoricals_get_weight_by_subscript (const struct categoricals *cat, int subscript)
 {
   int vindex = reverse_variable_lookup_short (cat, subscript);
-  const struct var_params *vp = &cat->vp[vindex];
+  const struct interact_params *vp = &cat->iap[vindex];
 
   return vp->cc;
 }
@@ -433,7 +444,7 @@ double
 categoricals_get_sum_by_subscript (const struct categoricals *cat, int subscript)
 {
   int vindex = reverse_variable_lookup_short (cat, subscript);
-  const struct var_params *vp = &cat->vp[vindex];
+  const struct interact_params *vp = &cat->iap[vindex];
 
   const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_short];
   return vn->cc;
@@ -446,12 +457,11 @@ double
 categoricals_get_binary_by_subscript (const struct categoricals *cat, int subscript,
 				      const struct ccase *c)
 {
-  const struct variable *var = categoricals_get_variable_by_subscript (cat, subscript);
-  int width = var_get_width (var);
+  const struct interaction *iact = categoricals_get_interaction_by_subscript (cat, subscript);
 
-  const union value *val = case_data (c, var);
+  const struct ccase *c2 =  categoricals_get_case_by_subscript (cat, subscript);
 
-  return value_equal (val, categoricals_get_value_by_subscript (cat, subscript), width);
+  return interaction_case_equal (iact, c, c2);
 }
 
 
@@ -462,16 +472,15 @@ categoricals_get_n_variables (const struct categoricals *cat)
 }
 
 
-
-/* Return the value corresponding to SUBSCRIPT */
-const union value *
-categoricals_get_value_by_category (const struct categoricals *cat, int subscript)
+/* Return a case containing the set of values corresponding to SUBSCRIPT */
+const struct ccase *
+categoricals_get_case_by_category (const struct categoricals *cat, int subscript)
 {
   int vindex = reverse_variable_lookup_long (cat, subscript);
-  const struct var_params *vp = &cat->vp[vindex];
+  const struct interact_params *vp = &cat->iap[vindex];
   const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_long];
 
-  return &vn->value;
+  return vn->ccase;
 }
 
 
@@ -479,7 +488,7 @@ void *
 categoricals_get_user_data_by_category (const struct categoricals *cat, int subscript)
 {
   int vindex = reverse_variable_lookup_long (cat, subscript);
-  const struct var_params *vp = &cat->vp[vindex];
+  const struct interact_params *vp = &cat->iap[vindex];
 
   const struct value_node *vn = vp->reverse_value_map [subscript - vp->base_subscript_long];
   return vn->user_data;