work
authorBen Pfaff <blp@cs.stanford.edu>
Sat, 1 Jan 2022 18:09:12 +0000 (10:09 -0800)
committerBen Pfaff <blp@cs.stanford.edu>
Thu, 13 Jan 2022 05:52:27 +0000 (21:52 -0800)
src/language/stats/ctables.c

index 30ce9db8ab010cbee13227a983516d8779ad1752..8cf79c01230d3c87165e24e475d1f98c5de3e29c 100644 (file)
@@ -1672,7 +1672,7 @@ ctables_summary_add (union ctables_summary *s,
 }
 
 
-static double
+static double UNUSED
 ctables_summary_value (union ctables_summary *s,
                        const struct ctables_summary_spec *ss)
 {
@@ -1814,7 +1814,6 @@ struct ctables_freqtab
     const struct ctables_summary_spec *summaries;
     size_t n_summaries;
     const struct variable *summary_var;
-    struct ctables_freq **sorted;
   };
 
 static struct ctables_freq *
@@ -1837,17 +1836,28 @@ ctables_freq_add (struct ctables_freqtab *ft, struct ctables_freq *f,
                          var, value, weight);
 }
 
+struct ctables_axis_render
+  {
+    size_t ofs;
+    struct variable **vars;
+    size_t n_vars;
+
+    struct ctables_freq **freqs;
+    size_t n_freqs;
+  };
+
 static int
-ctables_freq_compare_3way (const void *a_, const void *b_, const void *vars_)
+ctables_freq_compare_3way (const void *a_, const void *b_, const void *ar_)
 {
-  const struct var_array *vars = vars_;
+  const struct ctables_axis_render *ar = ar_;
   struct ctables_freq *const *a = a_;
   struct ctables_freq *const *b = b_;
 
-  for (size_t i = 0; i < vars->n; i++)
+  for (size_t i = 0; i < ar->n_vars; i++)
     {
-      int cmp = value_compare_3way (&(*a)->values[i], &(*b)->values[i],
-                                    var_get_width (vars->vars[i]));
+      int cmp = value_compare_3way (&(*a)->values[i + ar->ofs],
+                                    &(*b)->values[i + ar->ofs],
+                                    var_get_width (ar->vars[i]));
       if (cmp)
         return cmp;
     }
@@ -2011,70 +2021,119 @@ ctables_execute (struct dataset *ds, struct ctables *ct)
       struct ctables_table *t = &ct->tables[i];
 
       struct pivot_table *pt = pivot_table_create (N_("Custom Tables"));
-      struct pivot_dimension *d = pivot_dimension_create (
-        pt, PIVOT_AXIS_ROW, N_("Rows"));
+      pivot_table_set_look (pt, ct->look);
+      struct pivot_dimension *d[PIVOT_N_AXES];
+      for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
+        {
+          static const char *names[] = {
+            [PIVOT_AXIS_ROW] = N_("Rows"),
+            [PIVOT_AXIS_COLUMN] = N_("Columns"),
+            [PIVOT_AXIS_LAYER] = N_("Layers"),
+          };
+          d[a] = (t->axes[a] || a == t->slabels_position
+                  ? pivot_dimension_create (pt, a, names[a])
+                  : NULL);
+        }
       for (size_t j = 0; j < t->n_fts; j++)
         {
           struct ctables_freqtab *ft = t->fts[j];
-          ft->sorted = xnmalloc (ft->data.count, sizeof *ft->sorted);
-
-          struct ctables_freq *f;
-          size_t n = 0;
-          HMAP_FOR_EACH (f, struct ctables_freq, node, &ft->data)
-            ft->sorted[n++] = f;
-          assert (n == ft->data.count);
-          sort (ft->sorted, n, sizeof *ft->sorted,
-                ctables_freq_compare_3way, &ft->vars);
-
-          struct pivot_category **groups = xnmalloc (ft->vars.n,
-                                                     sizeof *groups);
-          for (size_t k = 0; k < n; k++)
+
+          struct ctables_axis_render axis_renders[PIVOT_N_AXES];
+
+          for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
             {
-              struct ctables_freq *prev = k > 0 ? ft->sorted[k - 1] : NULL;
-              struct ctables_freq *f = ft->sorted[k];
-
-              size_t n_common = 0;
-              if (prev)
-                for (; n_common + 1 < ft->vars.n; n_common++)
-                  if (!value_equal (&prev->values[n_common],
-                                    &f->values[n_common],
-                                    var_get_type (ft->vars.vars[n_common])))
+              size_t ofs = 0;
+              struct variable **vars = NULL;
+              size_t n = 0;
+              for (size_t k = 0; k < ft->vars.n; k++)
+                if (ft->vars.axes[k] == a)
+                  {
+                    ofs = k;
+                    vars = &ft->vars.vars[k];
+                    for (n = 1; k + n < ft->vars.n; n++)
+                      if (ft->vars.axes[k + n] != a)
+                        break;
                     break;
+                  }
 
-              for (size_t m = n_common; m < ft->vars.n; m++)
-                {
-                  struct pivot_category *parent = m > 0 ? groups[m - 1] : d->root;
-                  const struct variable *var = ft->vars.vars[m];
-                  enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (var)];
+              struct ctables_axis_render *ar = &axis_renders[a];
+              *ar = (struct ctables_axis_render) {
+                .ofs = ofs,
+                .vars = vars,
+                .n_vars = n
+              };
+              printf ("axis %s: %zu variables\n", pivot_axis_type_to_string (a), n);
+              if (!n)
+                continue;
 
-                  if (vlabel != CTVL_NONE)
-                    parent = pivot_category_create_group__ (
-                      parent, pivot_value_new_variable (ft->vars.vars[m]));
+              ar->freqs = xnmalloc (ft->data.count, sizeof *ar->freqs);
 
-                  if (m + 1 < ft->vars.n)
-                    parent = pivot_category_create_group__ (
-                      parent,
-                      pivot_value_new_var_value (ft->vars.vars[m], &f->values[m]));
-                  groups[m] = parent;
+              struct ctables_freq *f;
+              size_t n_freqs = 0;
+              HMAP_FOR_EACH (f, struct ctables_freq, node, &ft->data)
+                ar->freqs[n_freqs++] = f;
+              assert (n_freqs == ft->data.count);
+              ar->n_freqs = sort_unique (ar->freqs, n_freqs, sizeof *ar->freqs,
+                                         ctables_freq_compare_3way, ar);
+            }
 
-                  if (m == ft->vars.n - 1)
+          for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
+            {
+              struct ctables_axis_render *ar = &axis_renders[a];
+              if (!ar->n_vars && a != t->slabels_position)
+                continue;
+
+              struct pivot_category **groups = xnmalloc (ar->n_vars,
+                                                         sizeof *groups);
+              for (size_t k = 0; k < ar->n_freqs; k++)
+                {
+                  struct ctables_freq *prev = k > 0 ? ar->freqs[k - 1] : NULL;
+                  struct ctables_freq *f = ar->freqs[k];
+
+                  size_t n_common = 0;
+                  if (prev)
+                    for (; n_common + 1 < ar->n_vars; n_common++)
+                      if (!value_equal (&prev->values[ar->ofs + n_common],
+                                        &f->values[ar->ofs + n_common],
+                                        var_get_type (ar->vars[n_common])))
+                        break;
+
+                  for (size_t m = n_common; m < ar->n_vars; m++)
                     {
-                      struct pivot_category *c = pivot_category_create_group__ (
-                        parent,
-                        pivot_value_new_var_value (ft->vars.vars[ft->vars.n - 1],
-                                                   &f->values[ft->vars.n - 1]));
-                      for (size_t p = 0; p < ft->n_summaries; p++)
+                      struct pivot_category *parent = m > 0 ? groups[m - 1] : d[a]->root;
+                      const struct variable *var = ar->vars[m];
+                      enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (var)];
+
+                      if (vlabel != CTVL_NONE)
+                        parent = pivot_category_create_group__ (
+                          parent, pivot_value_new_variable (ar->vars[m]));
+
+                      if (m + 1 < ar->n_vars)
+                        parent = pivot_category_create_group__ (
+                          parent,
+                          pivot_value_new_var_value (ar->vars[m], &f->values[m]));
+                      groups[m] = parent;
+
+                      if (m == ar->n_vars - 1)
                         {
-                          double value = ctables_summary_value (
-                            &f->summaries[p], &ft->summaries[p]);
-                          int leaf = pivot_category_create_leaf (
-                            c, pivot_value_new_text (ft->summaries[p].label));
-                          pivot_table_put1 (pt, leaf, pivot_value_new_number (value));
+                          pivot_category_create_leaf (
+                            parent,
+                            pivot_value_new_var_value (ar->vars[ar->n_vars - 1],
+                                                       &f->values[ar->ofs + ar->n_vars - 1]));
+#if 0
+                          for (size_t p = 0; p < ft->n_summaries; p++)
+                            {
+                              if (a == t->slabels_position)
+                                pivot_category_create_leaf (
+                                  c, pivot_value_new_text (ft->summaries[p].label));
+                              //pivot_table_put1 (pt, leaf, pivot_value_new_number (value));
+                            }
+#endif
                         }
                     }
                 }
+              free (groups);
             }
-          free (groups);
         }
       pivot_table_submit (pt);
     }
@@ -2101,7 +2160,6 @@ ctables_execute (struct dataset *ds, struct ctables *ct)
               free (f);
             }
           hmap_destroy (&ft->data);
-          free (ft->sorted);
           var_array_uninit (&ft->vars);
           free (ft);
         }
@@ -2127,6 +2185,7 @@ cmd_ctables (struct lexer *lexer, struct dataset *ds)
     .vlabels = vlabels,
     .hide_threshold = 5,
   };
+  ct->look->omit_empty = false;
 
   if (!lex_force_match (lexer, T_SLASH))
     goto error;