X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fmath%2Fcategoricals.c;h=83326c3f006789814c4930feed8d85d15db95155;hb=03384f9680feba65ee79b9680f366435a7a60442;hp=3f16c9437b40d3efdf49270c77df9c6ff992f955;hpb=6590347b1a0ed5037743f671ab846813c89f9112;p=pspp

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 3f16c9437b..83326c3f00 100644
--- a/src/math/categoricals.c
+++ b/src/math/categoricals.c
@@ -59,6 +59,7 @@ struct variable_node
     struct hmap_node node;      /* In struct categorical's 'varmap'. */
     const struct variable *var; /* The variable. */
     struct hmap valmap;         /* Contains "struct value_node"s. */
+    union value *values;        /* Values in valmap, as a sorted array. */
   };
 
 static int
@@ -74,10 +75,11 @@ compare_value_node_3way (const void *vn1_, const void *vn2_, const void *aux)
 }
 
 static struct variable_node *
-lookup_variable (const struct hmap *map, const struct variable *var, unsigned int hash)
+lookup_variable (const struct hmap *map, const struct variable *var)
 {
   struct variable_node *vn;
-  HMAP_FOR_EACH_WITH_HASH (vn, struct variable_node, node, hash, map)
+  HMAP_FOR_EACH_WITH_HASH (vn, struct variable_node, node,
+                           hash_pointer (var, 0), map)
     if (vn->var == var)
       return vn;
   return NULL;
@@ -180,13 +182,13 @@ categoricals_isbalanced (const struct categoricals *cat)
 
       double oval = -1.0;
       for (v = 0; v < hmap_count (&iap->ivmap); ++v)
-	{
-	  const struct interaction_value *iv = iap->ivs[v];
-	  if (oval == -1.0)
-	    oval = iv->cc;
-	  if (oval != iv->cc)
-	    return false;
-	}
+        {
+          const struct interaction_value *iv = iap->ivs[v];
+          if (oval == -1.0)
+            oval = iv->cc;
+          if (oval != iv->cc)
+            return false;
+        }
     }
   return true;
 }
@@ -211,46 +213,46 @@ categoricals_dump (const struct categoricals *cat)
 
       printf ("Number of interactions %zu\n", cat->n_iap);
       for (i = 0 ; i < cat->n_iap; ++i)
-	{
-	  int v;
-	  struct string str;
-	  const struct interact_params *iap = &cat->iap[i];
-	  const struct interaction *iact = iap->iact;
-
-	  ds_init_empty (&str);
-	  interaction_to_string (iact, &str);
-
-	  printf ("\nInteraction: \"%s\" (number of categories: %d); ", ds_cstr (&str), iap->n_cats);
-	  ds_destroy (&str);
-	  printf ("Base index (df/categories): %d/%d\n", iap->base_df, iap->base_cats);
-
-	  printf ("\t(");
-	  for (v = 0; v < hmap_count (&iap->ivmap); ++v)
-	    {
-	      int vv;
-	      const struct interaction_value *iv = iap->ivs[v];
-
-	      if (v > 0)  printf ("   ");
-	      printf ("{");
-	      for (vv = 0; vv < iact->n_vars; ++vv)
-		{
-		  const struct variable *var = iact->vars[vv];
-		  const union value *val = case_data (iv->ccase, var);
-		  struct variable_node *vn = iap->varnodes[vv];
-		  const int width = var_get_width (var);
-		  unsigned int valhash = value_hash (val, width, 0);
-		  struct value_node *valn = lookup_value (&vn->valmap, val, valhash, width);
-
-		  assert (vn->var == var);
-
-		  printf ("%.*g(%d)", DBL_DIG + 1, val->f, valn->index);
-		  if (vv < iact->n_vars - 1)
-		    printf (", ");
-		}
-	      printf ("}");
-	    }
-	  printf (")\n");
-	}
+        {
+          int v;
+          struct string str;
+          const struct interact_params *iap = &cat->iap[i];
+          const struct interaction *iact = iap->iact;
+
+          ds_init_empty (&str);
+          interaction_to_string (iact, &str);
+
+          printf ("\nInteraction: \"%s\" (number of categories: %d); ", ds_cstr (&str), iap->n_cats);
+          ds_destroy (&str);
+          printf ("Base index (df/categories): %d/%d\n", iap->base_df, iap->base_cats);
+
+          printf ("\t(");
+          for (v = 0; v < hmap_count (&iap->ivmap); ++v)
+            {
+              int vv;
+              const struct interaction_value *iv = iap->ivs[v];
+
+              if (v > 0)  printf ("   ");
+              printf ("{");
+              for (vv = 0; vv < iact->n_vars; ++vv)
+                {
+                  const struct variable *var = iact->vars[vv];
+                  const union value *val = case_data (iv->ccase, var);
+                  struct variable_node *vn = iap->varnodes[vv];
+                  const int width = var_get_width (var);
+                  unsigned int valhash = value_hash (val, width, 0);
+                  struct value_node *valn = lookup_value (&vn->valmap, val, valhash, width);
+
+                  assert (vn->var == var);
+
+                  printf ("%.*g(%d)", DBL_DIG + 1, val->f, valn->index);
+                  if (vv < iact->n_vars - 1)
+                    printf (", ");
+                }
+              printf ("}");
+            }
+          printf (")\n");
+        }
     }
 }
 
@@ -265,15 +267,16 @@ categoricals_destroy (struct categoricals *cat)
       /* Unref any cases that we reffed. */
       struct interaction_value *iv;
       HMAP_FOR_EACH (iv, struct interaction_value, node, &cat->iap[i].ivmap)
-	{
-	  if (cat->payload && cat->payload->destroy)
-	    cat->payload->destroy (cat->aux1, cat->aux2, iv->user_data);
-	  case_unref (iv->ccase);
-	}
+        {
+          if (cat->payload && cat->payload->destroy)
+            cat->payload->destroy (cat->aux1, cat->aux2, iv->user_data);
+          case_unref (iv->ccase);
+        }
 
       free (cat->iap[i].enc_sum);
       hmap_destroy (&cat->iap[i].ivmap);
     }
+  free (cat->iap);
 
   /* Interate over each variable and delete its value map.
 
@@ -318,9 +321,9 @@ categoricals_sane (const struct categoricals *cat)
    categoricals_update(). */
 struct categoricals *
 categoricals_create (struct interaction *const *inter, size_t n_inter,
-		     const struct variable *wv, enum mv_class fctr_excl)
+                     const struct variable *wv, enum mv_class fctr_excl)
 {
-  struct categoricals *cat = xzalloc (sizeof *cat);
+  struct categoricals *cat = XZALLOC (struct categoricals);
   cat->iap = pool_calloc (cat->pool, n_inter, sizeof *cat->iap);
   cat->n_iap = n_inter;
   cat->wv = wv;
@@ -337,19 +340,19 @@ categoricals_create (struct interaction *const *inter, size_t n_inter,
       iap->varnodes = pool_nmalloc (cat->pool, iap->iact->n_vars,
                                     sizeof *iap->varnodes);
       for (size_t v = 0; v < inter[i]->n_vars; ++v)
-	{
-	  const struct variable *var = inter[i]->vars[v];
-	  unsigned int hash = hash_pointer (var, 0);
-	  struct variable_node *vn = lookup_variable (&cat->varmap, var, hash);
-	  if (!vn)
-	    {
-	      vn = pool_malloc (cat->pool, sizeof *vn);
-	      vn->var = var;
-	      hmap_init (&vn->valmap);
-	      hmap_insert (&cat->varmap, &vn->node,  hash);
-	    }
+        {
+          const struct variable *var = inter[i]->vars[v];
+          struct variable_node *vn = lookup_variable (&cat->varmap, var);
+          if (!vn)
+            {
+              vn = pool_malloc (cat->pool, sizeof *vn);
+              vn->var = var;
+              vn->values = NULL;
+              hmap_init (&vn->valmap);
+              hmap_insert (&cat->varmap, &vn->node, hash_pointer (var, 0));
+            }
           iap->varnodes[v] = vn;
-	}
+        }
     }
 
   return cat;
@@ -364,7 +367,7 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
   assert (!cat->cat_to_iact);
 
   double weight;
-  weight = cat->wv ? case_data (c, cat->wv)->f : 1.0;
+  weight = cat->wv ? case_num (c, cat->wv) : 1.0;
   weight = var_force_valid_weight (cat->wv, weight, NULL);
 
   /* Update the frequency table for each variable. */
@@ -377,13 +380,13 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 
       struct value_node *valn = lookup_value (&vn->valmap, val, hash, width);
       if (valn == NULL)
-	{
-	  valn = pool_malloc (cat->pool, sizeof *valn);
-	  valn->index = -1;
-	  value_init (&valn->val, width);
-	  value_copy (&valn->val, val, width);
-	  hmap_insert (&vn->valmap, &valn->node, hash);
-	}
+        {
+          valn = pool_malloc (cat->pool, sizeof *valn);
+          valn->index = -1;
+          value_init_pool (cat->pool, &valn->val, width);
+          value_copy (&valn->val, val, width);
+          hmap_insert (&vn->valmap, &valn->node, hash);
+        }
     }
 
   /* Update the frequency table for full interactions. */
@@ -392,21 +395,21 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
       struct interact_params *iap = &cat->iap[i];
       const struct interaction *iact = iap->iact;
       if (interaction_case_is_missing (iact, c, cat->fctr_excl))
-	continue;
+        continue;
 
       unsigned int hash = interaction_case_hash (iact, c, 0);
       struct interaction_value *node = lookup_case (&iap->ivmap, iact, c);
       if (!node)
-	{
-	  node = pool_malloc (cat->pool, sizeof *node);
-	  node->ccase = case_ref (c);
-	  node->cc = weight;
+        {
+          node = pool_malloc (cat->pool, sizeof *node);
+          node->ccase = case_ref (c);
+          node->cc = weight;
 
-	  hmap_insert (&iap->ivmap, &node->node, hash);
+          hmap_insert (&iap->ivmap, &node->node, hash);
 
-	  if (cat->payload)
+          if (cat->payload)
             node->user_data = cat->payload->create (cat->aux1, cat->aux2);
-	}
+        }
       else
         node->cc += weight;
       iap->cc += weight;
@@ -477,7 +480,7 @@ categoricals_done (const struct categoricals *cat_)
           return;
         }
 
-      struct value_node **nodes = xcalloc (sizeof *nodes, n_vals);
+      struct value_node **nodes = XCALLOC (n_vals,  struct value_node *);
       int x = 0;
       struct value_node *valnd;
       HMAP_FOR_EACH (valnd, struct value_node, node, &vn->valmap)
@@ -499,15 +502,15 @@ categoricals_done (const struct categoricals *cat_)
       iap->df_prod = 1;
       iap->n_cats = 1;
       for (int v = 0 ; v < iact->n_vars; ++v)
-	{
+        {
           size_t n_vals = hmap_count (&iap->varnodes[v]->valmap);
 
           iap->df_prod *= n_vals - 1;
-	  iap->n_cats *= n_vals;
-	}
+          iap->n_cats *= n_vals;
+        }
 
       if (iact->n_vars > 0)
-	cat->df_sum += iap->df_prod;
+        cat->df_sum += iap->df_prod;
       cat->n_cats_total += iap->n_cats;
     }
 
@@ -544,7 +547,7 @@ categoricals_done (const struct categoricals *cat_)
           cat->df_to_iact[idx_df++] = i;
 
       for (int j = 0; j < iap->n_cats; ++j)
-	cat->cat_to_iact[idx_cat++] = i;
+        cat->cat_to_iact[idx_cat++] = i;
     }
 
   categoricals_dump (cat);
@@ -560,23 +563,40 @@ categoricals_done (const struct categoricals *cat_)
       iap->enc_sum = xcalloc (df, sizeof *iap->enc_sum);
 
       for (int y = 0; y < hmap_count (&iap->ivmap); ++y)
-	{
-	  struct interaction_value *iv = iap->ivs[y];
-	  for (int x = iap->base_df;
+        {
+          struct interaction_value *iv = iap->ivs[y];
+          for (int x = iap->base_df;
                x < iap->base_df + df; ++x)
-	    {
-	      const double bin = categoricals_get_effects_code_for_case (
+            {
+              const double bin = categoricals_get_effects_code_for_case (
                 cat, x, iv->ccase);
-	      iap->enc_sum[x - iap->base_df] += bin * iv->cc;
-	    }
-	  if (cat->payload && cat->payload->calculate)
-	    cat->payload->calculate (cat->aux1, cat->aux2, iv->user_data);
-	}
+              iap->enc_sum[x - iap->base_df] += bin * iv->cc;
+            }
+          if (cat->payload && cat->payload->calculate)
+            cat->payload->calculate (cat->aux1, cat->aux2, iv->user_data);
+        }
     }
 
   cat->sane = true;
 }
 
+union value *
+categoricals_get_var_values (const struct categoricals *cat,
+                             const struct variable *var, size_t *np)
+{
+  struct variable_node *vn = lookup_variable (&cat->varmap, var);
+  *np = hmap_count (&vn->valmap);
+  if (!vn->values)
+    {
+      vn->values = pool_nalloc (cat->pool, *np, sizeof *vn->values);
+
+      struct value_node *valnd;
+      HMAP_FOR_EACH (valnd, struct value_node, node, &vn->valmap)
+        vn->values[valnd->index] = valnd->val;
+    }
+  return vn->values;
+}
+
 static struct interact_params *
 df_to_iap (const struct categoricals *cat, int subscript)
 {
@@ -623,8 +643,8 @@ categoricals_get_sum_by_subscript (const struct categoricals *cat,
    for that subscript */
 static double
 categoricals_get_code_for_case (const struct categoricals *cat, int subscript,
-				const struct ccase *c,
-				bool effects_coding)
+                                const struct ccase *c,
+                                bool effects_coding)
 {
   const struct interaction *iact
     = categoricals_get_interaction_by_subscript (cat, subscript);
@@ -648,7 +668,7 @@ categoricals_get_code_for_case (const struct categoricals *cat, int subscript,
       const int dfpn = dfp * df;
 
       if (effects_coding && valn->index == df)
-	result = -result;
+        result = -result;
       else
         {
           /* Translate subscript into an index for the individual variable. */
@@ -704,6 +724,21 @@ categoricals_get_user_data_by_category_real (const struct categoricals *cat,
   return n < hmap_count (&iap->ivmap) ? iap->ivs[n]->user_data : NULL;
 }
 
+int
+categoricals_get_value_index_by_category_real (const struct categoricals *cat,
+                                               int iact_idx, int cat_idx,
+                                               int var_idx)
+{
+  const struct interact_params *iap = &cat->iap[iact_idx];
+  const struct interaction_value *ivn = iap->ivs[cat_idx];
+  const struct variable *var = iap->iact->vars[var_idx];
+  const struct variable_node *vn = iap->varnodes[var_idx];
+  const union value *val = case_data (ivn->ccase, var);
+  int width = var_get_width (var);
+  unsigned int hash = value_hash (val, width, 0);
+  return lookup_value (&vn->valmap, val, hash, width)->index;
+}
+
 /* Return a case containing the set of values corresponding to CAT_INDEX. */
 const struct ccase *
 categoricals_get_case_by_category (const struct categoricals *cat,
@@ -725,7 +760,7 @@ categoricals_get_user_data_by_category (const struct categoricals *cat,
 
 void
 categoricals_set_payload (struct categoricals *cat, const struct payload *p,
-			  const void *aux1, void *aux2)
+                          const void *aux1, void *aux2)
 {
   cat->payload = p;
   cat->aux1 = aux1;