From: Ben Pfaff <blp@cs.stanford.edu>
Date: Sat, 15 Jan 2022 07:24:53 +0000 (-0800)
Subject: merging works
X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?p=pspp;a=commitdiff_plain;h=125b43a54108a29ecaa52886c928b76ddceba1db

merging works
---

diff --git a/src/language/stats/ctables.c b/src/language/stats/ctables.c
index 99efc09c5a..169f8631e7 100644
--- a/src/language/stats/ctables.c
+++ b/src/language/stats/ctables.c
@@ -540,6 +540,7 @@ struct ctables_summary_spec
     double percentile;          /* CTSF_PTILE only. */
     char *label;
     struct fmt_spec format;     /* XXX extra CTABLES formats */
+    size_t axis_idx;
   };
 
 static void
@@ -894,6 +895,12 @@ ctables_axis_parse_primary (struct ctables_axis_parse_ctx *ctx)
   return axis;
 }
 
+static bool
+has_digit (const char *s)
+{
+  return s[strcspn (s, "0123456789")] != '\0';
+}
+
 static struct ctables_axis *
 ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
 {
@@ -934,7 +941,8 @@ ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
       /* Parse format. */
       struct fmt_spec format;
       const struct fmt_spec *formatp;
-      if (lex_token (ctx->lexer) == T_ID)
+      if (lex_token (ctx->lexer) == T_ID
+          && has_digit (lex_tokcstr (ctx->lexer)))
         {
           if (!parse_format_specifier (ctx->lexer, &format)
               || !fmt_check_output (&format)
@@ -950,6 +958,7 @@ ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
 
       struct msg_location *loc = lex_ofs_location (ctx->lexer, start_ofs,
                                                    lex_ofs (ctx->lexer) - 1);
+      printf ("add %s\n", ctables_summary_function_name (function));
       add_summary_spec (sub, function, percentile, label, formatp, loc, sv);
       free (label);
       msg_location_destroy (loc);
@@ -961,14 +970,12 @@ ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
             goto error;
           sv = CSV_TOTAL;
         }
-      else if (lex_force_match (ctx->lexer, T_RBRACK))
+      else if (lex_match (ctx->lexer, T_RBRACK))
         {
           if (sv == CSV_TOTAL && !lex_force_match (ctx->lexer, T_RBRACK))
             goto error;
           return sub;
         }
-      else
-        goto error;
     }
 
 error:
@@ -2427,6 +2434,25 @@ ctables_cell_insert (struct ctables_table *t,
     }
 }
 
+struct merge_item
+  {
+    size_t tiebreaker;
+    const struct ctables_summary_spec_set *set;
+    size_t ofs;
+  };
+
+static int
+merge_item_compare_3way (const struct merge_item *a, const struct merge_item *b)
+{
+  const struct ctables_summary_spec *as = &a->set->specs[a->ofs];
+  const struct ctables_summary_spec *bs = &b->set->specs[b->ofs];
+  if (as->function != bs->function)
+    return as->function > bs->function ? 1 : -1;
+  else if (as->percentile != bs->percentile)
+    return as->percentile < bs->percentile ? 1 : -1;
+  return strcmp (as->label, bs->label);
+}
+
 static bool
 ctables_execute (struct dataset *ds, struct ctables *ct)
 {
@@ -2498,9 +2524,10 @@ ctables_execute (struct dataset *ds, struct ctables *ct)
             t->stacks[a] = (struct ctables_stack) { .nests = nest, .n = 1 };
           }
 
-      for (size_t i = 0; i < t->stacks[t->summary_axis].n; i++)
+      struct ctables_stack *stack = &t->stacks[t->summary_axis];
+      for (size_t i = 0; i < stack->n; i++)
         {
-          struct ctables_nest *nest = &t->stacks[t->summary_axis].nests[i];
+          struct ctables_nest *nest = &stack->nests[i];
           if (!nest->specs[CSV_CELL].n)
             {
               struct ctables_summary_spec_set *specs = &nest->specs[CSV_CELL];
@@ -2526,6 +2553,57 @@ ctables_execute (struct dataset *ds, struct ctables *ct)
             ctables_summary_spec_set_clone (&nest->specs[CSV_TOTAL],
                                             &nest->specs[CSV_CELL]);
         }
+
+      struct ctables_summary_spec_set merged = { .n = 0 };
+      struct merge_item *items = xnmalloc (2 * stack->n, sizeof *items);
+      size_t n_left = 0;
+      for (size_t j = 0; j < stack->n; j++)
+        {
+          const struct ctables_nest *nest = &stack->nests[j];
+          if (!nest->n)
+            continue;
+
+          for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
+            {
+              items[n_left] = (struct merge_item) {
+                .tiebreaker = n_left,
+                .set = &nest->specs[sv]
+              };
+              n_left++;
+            }
+        }
+
+      while (n_left > 0)
+        {
+          struct merge_item min = items[0];
+          for (size_t j = 1; j < n_left; j++)
+            if (merge_item_compare_3way (&items[j], &min) < 0)
+              min = items[j];
+
+          /* XXX Add to 'merged' */
+          if (merged.n >= merged.allocated)
+            merged.specs = x2nrealloc (merged.specs, &merged.allocated,
+                                       sizeof *merged.specs);
+          merged.specs[merged.n++] = min.set->specs[min.ofs];
+
+          for (size_t j = 0; j < n_left; )
+            {
+              if (merge_item_compare_3way (&items[j], &min) == 0)
+                {
+                  struct merge_item *item = &items[j];
+                  item->set->specs[item->ofs].axis_idx = merged.n - 1;
+                  if (++item->ofs >= item->set->n)
+                    {
+                      items[j] = items[--n_left];
+                      continue;
+                    }
+                }
+              j++;
+            }
+        }
+
+      for (size_t j = 0; j < merged.n; j++)
+        printf ("%s\n", ctables_summary_function_name (merged.specs[j].function));
     }
 
   struct casereader *input = casereader_create_filter_weight (proc_open (ds),