summaries work

[pspp] / src / language / stats / ctables.c

diff --git a/src/language/stats/ctables.c b/src/language/stats/ctables.c
index 9f8223cbb6936a1afe3da25c8c265d965acc756e..26837f1b93c61fae51bd946c5f2c6b4b7238e242 100644 (file)
--- a/src/language/stats/ctables.c
+++ b/src/language/stats/ctables.c
@@ -1406,7 +1406,6 @@ union ctables_summary
     /* XXX percentiles, median, mode, multiple response */
   };
 
-#if 0
 static void
 ctables_summary_init (union ctables_summary *s,
                       const struct ctables_summary_spec *ss)
@@ -1496,7 +1495,7 @@ ctables_summary_init (union ctables_summary *s,
     }
 }
 
-static void
+static void UNUSED
 ctables_summary_uninit (union ctables_summary *s,
                         const struct ctables_summary_spec *ss)
 {
@@ -1811,7 +1810,6 @@ ctables_summary_value (union ctables_summary *s,
 
   NOT_REACHED ();
 }
-#endif
 
 struct ctables_freq
   {
@@ -1821,11 +1819,11 @@ struct ctables_freq
       {
         size_t vaa_idx;
         union value *values;
+        int leaf;
       }
     axes[PIVOT_N_AXES];
 
-    //union ctables_summary *summaries;
-    double count;
+    union ctables_summary *summaries;
   };
 
 #if 0
@@ -1911,6 +1909,7 @@ ctables_freqtab_insert (struct ctables_table *t,
     [PIVOT_AXIS_COLUMN] = ic,
     [PIVOT_AXIS_LAYER] = il,
   };
+  const struct var_array *ss = &t->vaas[t->summary_axis].vas[ix[t->summary_axis]];
 
   size_t hash = 0;
   for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
@@ -1937,8 +1936,7 @@ ctables_freqtab_insert (struct ctables_table *t,
               goto not_equal;
         }
 
-      f->count += weight;
-      return;
+      goto summarize;
 
     not_equal: ;
     }
@@ -1955,8 +1953,15 @@ ctables_freqtab_insert (struct ctables_table *t,
         value_clone (&f->axes[a].values[i], case_data (c, va->vars[i]),
                      var_get_width (va->vars[i]));
     }
-  f->count = weight;
+  f->summaries = xmalloc (ss->n_summaries * sizeof *f->summaries);
+  for (size_t i = 0; i < ss->n_summaries; i++)
+    ctables_summary_init (&f->summaries[i], &ss->summaries[i]);
   hmap_insert (&t->ft, &f->node, hash);
+
+summarize:
+  for (size_t i = 0; i < ss->n_summaries; i++)
+    ctables_summary_add (&f->summaries[i], &ss->summaries[i], ss->summary_var,
+                         case_data (c, ss->summary_var), weight);
 }
 
 static bool
@@ -2051,7 +2056,7 @@ ctables_execute (struct dataset *ds, struct ctables *ct)
           assert (n == t->ft.count);
 
           struct ctables_freq_sort_aux aux = { .t = t, .a = a };
-          n = sort_unique (sorted, n, sizeof *sorted, ctables_freq_compare_3way, &aux);
+          sort (sorted, n, sizeof *sorted, ctables_freq_compare_3way, &aux);
 
           size_t max_depth = 0;
           for (size_t j = 0; j < t->vaas[a].n; j++)
@@ -2060,6 +2065,7 @@ ctables_execute (struct dataset *ds, struct ctables *ct)
 
           struct pivot_category **groups = xnmalloc (max_depth, sizeof *groups);
           struct pivot_category *top = NULL;
+          int prev_leaf = 0;
           for (size_t j = 0; j < n; j++)
             {
               struct ctables_freq *f = sorted[j];
@@ -2083,10 +2089,20 @@ ctables_execute (struct dataset *ds, struct ctables *ct)
                 }
               else
                 new_subtable = true;
+
               if (new_subtable)
-                top = pivot_category_create_group__ (
-                  d[a]->root, pivot_value_new_variable (va->vars[0]));
-              printf ("n_common=%zu\n", n_common);
+                {
+                  enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (va->vars[0])];
+                  top = d[a]->root;
+                  if (vlabel != CTVL_NONE)
+                    top = pivot_category_create_group__ (
+                      top, pivot_value_new_variable (va->vars[0]));
+                }
+              if (n_common == va->n)
+                {
+                  f->axes[a].leaf = prev_leaf;
+                  continue;
+                }
 
               for (size_t k = n_common; k < va->n; k++)
                 {
@@ -2101,23 +2117,56 @@ ctables_execute (struct dataset *ds, struct ctables *ct)
                         {
                           parent = pivot_category_create_group__ (parent, label);
                           for (size_t m = 0; m < va->n_summaries; m++)
-                            pivot_category_create_leaf (
-                              parent, pivot_value_new_text (va->summaries[m].label));
+                            {
+                              int leaf = pivot_category_create_leaf (
+                                parent, pivot_value_new_text (va->summaries[m].label));
+                              if (m == 0)
+                                prev_leaf = leaf;
+                            }
                         }
                       else
-                        pivot_category_create_leaf (parent, label);
+                        prev_leaf = pivot_category_create_leaf (parent, label);
                       break;
                     }
 
                   parent = pivot_category_create_group__ (parent, label);
 
-                  parent = pivot_category_create_group__ (
-                    parent, pivot_value_new_variable (va->vars[k]));
+                  enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (va->vars[k + 1])];
+                  if (vlabel != CTVL_NONE)
+                    parent = pivot_category_create_group__ (
+                      parent, pivot_value_new_variable (va->vars[k + 1]));
                   groups[k] = parent;
-                    }
                 }
+
+              f->axes[a].leaf = prev_leaf;
+            }
+          free (sorted);
           free (groups);
         }
+      struct ctables_freq *f;
+      HMAP_FOR_EACH (f, struct ctables_freq, node, &t->ft)
+        {
+          const struct var_array *ss = &t->vaas[t->summary_axis].vas[f->axes[t->summary_axis].vaa_idx];
+          for (size_t j = 0; j < ss->n_summaries; j++)
+            {
+              size_t dindexes[3];
+              size_t n_dindexes = 0;
+
+              for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
+                if (d[a])
+                  {
+                    int leaf = f->axes[a].leaf;
+                    if (a == t->summary_axis)
+                      leaf += j;
+                    dindexes[n_dindexes++] = leaf;
+                  }
+
+              double value = ctables_summary_value (&f->summaries[j], &ss->summaries[j]);
+              pivot_table_put (pt, dindexes, n_dindexes,
+                               pivot_value_new_number (value));
+            }
+        }
+
       pivot_table_submit (pt);
     }