Don't crash if all categorical variables are empty

[pspp-builds.git] / src / math / categoricals.c

diff --git a/src/math/categoricals.c b/src/math/categoricals.c
index 033b6a18a07c8bdb705c0bfb3ec44f9270f80cb5..b1f0ce84d9a397bdca0c71062dab79988d9eef0f 100644 (file)
--- 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 */
@@ -50,16 +53,18 @@ 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;
-
   const struct variable *wv;
 
+  size_t n_vars;
+
   /* An array of var_params */
   struct var_params *vp;
 
@@ -69,6 +74,9 @@ struct categoricals
   size_t n_cats_total;
 
   struct pool *pool;
+
+  /* Missing values to be excluded */
+  enum mv_class exclude;
 };
 
 
@@ -93,27 +101,33 @@ categoricals_dump (const struct categoricals *cat)
     {
       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;
+         ds_init_empty (&s);
          const struct value_node *vn = HMAP_DATA (node, struct value_node, node);
+         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);
        }
     }
 }
@@ -141,22 +155,26 @@ 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->wv = wv;
   cat->n_cats_total = 0;
   cat->reverse_variable_map = NULL;
   cat->pool = pool_create ();
+  cat->exclude = exclude;
 
   cat->vp = pool_calloc (cat->pool, n_vars, sizeof *cat->vp);
 
   for (i = 0 ; i < cat->n_vars; ++i)
-    hmap_init (&cat->vp[i].map);
+    {
+      hmap_init (&cat->vp[i].map);
+      cat->vp[i].var = v[i];
+    }
 
   return cat;
 }
@@ -174,11 +192,18 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
 
   for (i = 0 ; i < cat->n_vars; ++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_malloc (cat->pool, sizeof *node);
@@ -188,11 +213,12 @@ categoricals_update (struct categoricals *cat, const struct ccase *c)
          node->cc = 0.0;
 
          hmap_insert (&cat->vp[i].map, &node->node,  hash);
-         cat->n_cats_total ++;
+         cat->n_cats_total++;
          node->subscript = cat->vp[i].n_cats++ ;
        }
 
       node->cc += weight;
+      cat->vp[i].cc += weight;
     }
 }
 
@@ -208,7 +234,7 @@ categoricals_n_count (const struct categoricals *cat, size_t n)
 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);
+  struct value_node *vn = lookup_value (&cat->vp[n].map, cat->vp[n].var, val);
 
   if ( vn == NULL)
     return -1;
@@ -274,7 +300,7 @@ categoricals_get_variable_by_subscript (const struct categoricals *cat, int subs
   
   index = cat->reverse_variable_map[subscript];
 
-  return cat->vars[index];
+  return cat->vp[index].var;
 }
 
 
@@ -290,6 +316,26 @@ 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 = cat->reverse_variable_map[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 = cat->reverse_variable_map[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