Rewrite PSPP output engine.

[pspp-builds.git] / src / language / stats / chisquare.c

diff --git a/src/language/stats/chisquare.c b/src/language/stats/chisquare.c
index 6cb7fc0b192ee2bfce5e50a620a26bd466a3ee6e..1f2df8d29a2417395836e4f9b81a9f84e722a959 100644 (file)
--- a/src/language/stats/chisquare.c
+++ b/src/language/stats/chisquare.c
@@ -31,11 +31,12 @@
 #include <language/stats/freq.h>
 #include <language/stats/npar.h>
 #include <libpspp/assertion.h>
+#include <libpspp/cast.h>
 #include <libpspp/compiler.h>
 #include <libpspp/hash.h>
 #include <libpspp/message.h>
 #include <libpspp/taint.h>
-#include <output/table.h>
+#include <output/tab.h>
 
 #include <gsl/gsl_cdf.h>
 
@@ -68,47 +69,22 @@ create_freq_hash_with_range (const struct dictionary *dict,
   /* Populate the hash with zero entries */
   for (i_d = trunc (lo); i_d <= trunc (hi); i_d += 1.0 )
     {
-      union value the_value;
       struct freq_mutable *fr = xmalloc (sizeof (*fr));
-
-      the_value.f = i_d;
-
-      fr->value = value_dup (&the_value, 0);
+      value_init (&fr->value, 0);
+      fr->value.f = i_d;
       fr->count = 0;
-
       hsh_insert (freq_hash, fr);
     }
 
-  while ((c = casereader_read (input)) != NULL)
+  for (; (c = casereader_read (input)) != NULL; case_unref (c))
     {
-      union value obs_value;
-      struct freq **existing_fr;
-      struct freq *fr = xmalloc(sizeof  (*fr));
-      fr->value = case_data (c, var);
-
-      fr->count = dict_get_case_weight (dict, c, &warn);
-
-      obs_value.f = trunc (fr->value->f);
-
-      if ( obs_value.f < lo || obs_value.f > hi)
-       {
-         free (fr);
-         case_unref (c);
-         continue;
-       }
-
-      fr->value = &obs_value;
-
-      existing_fr = (struct freq **) hsh_probe (freq_hash, fr);
-
-      /* This must exist in the hash, because we previously populated it
-        with zero counts */
-      assert (*existing_fr);
-
-      (*existing_fr)->count += fr->count;
-      free (fr);
-
-      case_unref (c);
+      struct freq_mutable fr;
+      fr.value.f = trunc (case_num (c, var));
+      if (fr.value.f >= lo && fr.value.f <= hi)
+        {
+          struct freq_mutable *existing_fr = hsh_force_find (freq_hash, &fr);
+          existing_fr->count += dict_get_case_weight (dict, c, &warn);
+        }
     }
   if (casereader_destroy (input))
     return freq_hash;
@@ -130,6 +106,7 @@ create_freq_hash (const struct dictionary *dict,
                  struct casereader *input,
                  const struct variable *var)
 {
+  int width = var_get_width (var);
   bool warn = true;
   struct ccase *c;
 
@@ -140,23 +117,25 @@ create_freq_hash (const struct dictionary *dict,
 
   for (; (c = casereader_read (input)) != NULL; case_unref (c))
     {
-      struct freq **existing_fr;
-      struct freq *fr = xmalloc(sizeof  (*fr));
-      fr->value = case_data (c, var);
+      struct freq_mutable fr;
+      void **p;
 
-      fr->count = dict_get_case_weight (dict, c, &warn);
+      fr.value = *case_data (c, var);
+      fr.count = dict_get_case_weight (dict, c, &warn);
 
-      existing_fr = (struct freq **) hsh_probe (freq_hash, fr);
-      if ( *existing_fr)
-       {
-         (*existing_fr)->count += fr->count;
-         free (fr);
-       }
+      p = hsh_probe (freq_hash, &fr);
+      if (*p == NULL)
+        {
+          struct freq_mutable *new_fr = *p = xmalloc (sizeof *new_fr);
+          value_init (&new_fr->value, width);
+          value_copy (&new_fr->value, &fr.value, width);
+          new_fr->count = fr.count;
+        }
       else
-       {
-          *existing_fr = fr;
-          fr->value = value_dup (fr->value, var_get_width (var));
-       }
+        {
+          struct freq *existing_fr = *p;
+          existing_fr->count += fr.count;
+        }
     }
   if (casereader_destroy (input))
     return freq_hash;
@@ -201,8 +180,7 @@ create_variable_frequency_table (const struct dictionary *dict,
       return NULL;
     }
 
-  table = tab_create(4, n_cells + 2, 0);
-  tab_dim (table, tab_natural_dimensions);
+  table = tab_create(4, n_cells + 2);
 
   tab_title (table, var_to_string(var));
   tab_text (table, 1, 0, TAB_LEFT, _("Observed N"));
@@ -212,7 +190,7 @@ create_variable_frequency_table (const struct dictionary *dict,
   tab_headers (table, 1, 0, 1, 0);
 
   tab_box (table, TAL_1, TAL_1, -1, -1,
-          0, 0, table->nc - 1, tab_nr(table) - 1 );
+          0, 0, tab_nc (table) - 1, tab_nr(table) - 1 );
 
   tab_hline (table, TAL_1, 0, tab_nc(table) - 1, 1);
 
@@ -221,7 +199,7 @@ create_variable_frequency_table (const struct dictionary *dict,
     tab_vline (table, TAL_1, i, 0, tab_nr(table) - 1);
 
 
-  tab_text (table, 0, table->nr - 1, TAB_LEFT, _("Total"));
+  tab_text (table, 0, tab_nr (table) - 1, TAB_LEFT, _("Total"));
 
   return table;
 }
@@ -237,8 +215,7 @@ create_combo_frequency_table (const struct chisquare_test *test)
 
   int n_cells = test->hi - test->lo + 1;
 
-  table = tab_create(1 + ost->n_vars * 4, n_cells + 3, 0);
-  tab_dim (table, tab_natural_dimensions);
+  table = tab_create(1 + ost->n_vars * 4, n_cells + 3);
 
   tab_title (table, _("Frequencies"));
   for ( i = 0 ; i < ost->n_vars ; ++i )
@@ -276,12 +253,12 @@ create_combo_frequency_table (const struct chisquare_test *test)
   tab_headers (table, 1, 0, 2, 0);
 
   tab_box (table, TAL_1, TAL_1, -1, -1,
-          0, 0, table->nc - 1, tab_nr(table) - 1 );
+          0, 0, tab_nc (table) - 1, tab_nr(table) - 1 );
 
   tab_hline (table, TAL_1, 1, tab_nc(table) - 1, 1);
   tab_hline (table, TAL_1, 0, tab_nc(table) - 1, 2);
 
-  tab_text (table, 0, table->nr - 1, TAB_LEFT, _("Total"));
+  tab_text (table, 0, tab_nr (table) - 1, TAB_LEFT, _("Total"));
 
   return table;
 }
@@ -293,8 +270,7 @@ create_stats_table (const struct chisquare_test *test)
   const struct one_sample_test *ost = (const struct one_sample_test*) test;
 
   struct tab_table *table;
-  table = tab_create (1 + ost->n_vars, 4, 0);
-  tab_dim (table, tab_natural_dimensions);
+  table = tab_create (1 + ost->n_vars, 4);
   tab_title (table, _("Test Statistics"));
   tab_headers (table, 1, 0, 1, 0);
 
@@ -327,8 +303,9 @@ chisquare_execute (const struct dataset *ds,
 {
   const struct dictionary *dict = dataset_dict (ds);
   int v, i;
-  struct one_sample_test *ost = (struct one_sample_test *) test;
-  struct chisquare_test *cst = (struct chisquare_test *) test;
+  struct chisquare_test *cst = UP_CAST (test, struct chisquare_test,
+                                        parent.parent);
+  struct one_sample_test *ost = &cst->parent;
   int n_cells = 0;
   double total_expected = 0.0;
   const struct variable *wvar = dict_get_weight (dict);
@@ -371,7 +348,7 @@ chisquare_execute (const struct dataset *ds,
            {
              struct string str;
              double exp;
-             const union value *observed_value = ff[i]->value;
+             const union value *observed_value = &ff[i]->value;
 
              ds_init_empty (&str);
              var_append_value_name (ost->vars[v], observed_value, &str);
@@ -444,7 +421,7 @@ chisquare_execute (const struct dataset *ds,
              struct string str;
              double exp;
 
-             const union value *observed_value = ff[i]->value;
+             const union value *observed_value = &ff[i]->value;
 
              ds_init_empty (&str);
              var_append_value_name (ost->vars[v], observed_value, &str);