Make the expression code a little nicer and fix bugs found

[pspp] / src / descript.q

diff --git a/src/descript.q b/src/descript.q
index 748655ca16cba1625776dc91ea1ecb05523328e6..9ba2b569c79dbf69e80e90e493c71579d30ff0de 100644 (file)
--- a/src/descript.q
+++ b/src/descript.q
@@ -20,7 +20,7 @@
 /* FIXME: Many possible optimizations. */
 
 #include <config.h>
-#include <assert.h>
+#include "error.h"
 #include <limits.h>
 #include <math.h>
 #include <stdlib.h>
@@ -30,7 +30,6 @@
 #include "command.h"
 #include "lexer.h"
 #include "error.h"
-#include "approx.h"
 #include "magic.h"
 #include "stats.h"
 #include "som.h"
@@ -178,9 +177,9 @@ static void dump_z_table (void);
 static void run_z_pass (void);
 
 /* Procedure execution functions. */
-static int calc (struct ccase *);
-static void precalc (void);
-static void postcalc (void);
+static int calc (struct ccase *, void *);
+static void precalc (void *);
+static void postcalc (void *);
 static void display (void);
 \f
 /* Parser and outline. */
@@ -194,8 +193,6 @@ cmd_descriptives (void)
   v_variables = NULL;
   n_variables = 0;
 
-  lex_match_id ("DESCRIPTIVES");
-  lex_match_id ("CONDESCRIPTIVES");
   if (!parse_descriptives (&cmd))
     goto lossage;
 
@@ -288,9 +285,8 @@ cmd_descriptives (void)
     }
 
   /* Data pass! */
-  update_weighting (&default_dict);
   bad_weight = 0;
-  procedure (precalc, calc, postcalc);
+  procedure_with_splits (precalc, calc, postcalc, NULL);
 
   if (bad_weight)
     msg (SW, _("At least one case in the data file had a weight value "
@@ -313,10 +309,10 @@ cmd_descriptives (void)
 
 /* Parses the VARIABLES subcommand. */
 static int
-dsc_custom_variables (struct cmd_descriptives *cmd unused)
+dsc_custom_variables (struct cmd_descriptives *cmd UNUSED)
 {
   if (!lex_match_id ("VARIABLES")
-      && (token != T_ID || !is_varname (tokid))
+      && (token != T_ID || dict_lookup_var (default_dict, tokid) == NULL)
       && token != T_ALL)
     return 2;
   lex_match ('=');
@@ -326,7 +322,7 @@ dsc_custom_variables (struct cmd_descriptives *cmd unused)
       int i, n;
 
       n = n_variables;
-      if (!parse_variables (NULL, &v_variables, &n_variables,
+      if (!parse_variables (default_dict, &v_variables, &n_variables,
                            PV_DUPLICATE | PV_SINGLE | PV_APPEND | PV_NUMERIC
                            | PV_NO_SCRATCH))
        return 0;
@@ -345,7 +341,7 @@ dsc_custom_variables (struct cmd_descriptives *cmd unused)
              msg (SE, _("Name for z-score variable expected."));
              return 0;
            }
-         if (is_varname (tokid))
+         if (dict_lookup_var (default_dict, tokid))
            {
              msg (SE, _("Z-score variable name `%s' is a "
                         "duplicate variable name with a current variable."),
@@ -384,7 +380,7 @@ try_name (char *name)
 {
   int i;
 
-  if (is_varname (name))
+  if (dict_lookup_var (default_dict, name) != NULL)
     return 0;
   for (i = 0; i < n_variables; i++)
     {
@@ -478,7 +474,8 @@ dump_z_table (void)
 
 /* Transformation function to calculate Z-scores. */
 static int
-descriptives_trns_proc (struct trns_header * trns, struct ccase * c)
+descriptives_trns_proc (struct trns_header * trns, struct ccase * c,
+                        int case_num UNUSED)
 {
   struct descriptives_trns *t = (struct descriptives_trns *) trns;
   struct dsc_z_score *z;
@@ -540,9 +537,9 @@ run_z_pass (void)
          struct variable *d;
 
          t->z[count].s = v;
-         t->z[count].d = d = force_create_variable (&default_dict,
-                                                    v->p.dsc.zname,
-                                                    NUMERIC, 0);
+         t->z[count].d = d = dict_create_var_assert (default_dict,
+                                                      v->p.dsc.zname, 0);
+          d->init = 0;
          if (v->label)
            {
              d->label = xmalloc (strlen (v->label) + 12);
@@ -557,8 +554,7 @@ run_z_pass (void)
            }
          t->z[count].mean = v->p.dsc.stats[dsc_mean];
          t->z[count].std_dev = v->p.dsc.stats[dsc_stddev];
-         if (t->z[count].std_dev == SYSMIS
-             || approx_eq (t->z[count].std_dev, 0.0))
+         if (t->z[count].std_dev == SYSMIS || t->z[count].std_dev == 0.0)
            t->z[count].mean = SYSMIS;
          count++;
        }
@@ -570,7 +566,7 @@ run_z_pass (void)
 /* Statistical calculation. */
 
 static void
-precalc (void)
+precalc (void *aux UNUSED)
 {
   int i;
 
@@ -594,7 +590,7 @@ precalc (void)
 }
 
 static int
-calc (struct ccase * c)
+calc (struct ccase * c, void *aux UNUSED)
 {
   int i;
 
@@ -602,20 +598,13 @@ calc (struct ccase * c)
   static int case_id;
 
   /* Get the weight for this case. */
-  double w;
+  double weight = dict_get_case_weight (default_dict, c);
 
-  if (default_dict.weight_index == -1)
-    w = 1.0;
-  else
+  if (weight <= 0.0)
     {
-      w = c->data[default_dict.weight_index].f;
-      if (w <= 0.0 || w == SYSMIS)
-       {
-         w = 0.0;
-         bad_weight = 1;
-       }
+      weight = 0.0;
+      bad_weight = 1;
     }
-
   case_id++;
 
   /* Handle missing values. */
@@ -628,17 +617,17 @@ calc (struct ccase * c)
        {
          if (opt[op_excl_miss])
            {
-             d_glob_missing += w;
+             d_glob_missing += weight;
              return 1;
            }
          else
            {
-             d_glob_miss_list += w;
+             d_glob_miss_list += weight;
              goto iterate;
            }
        }
     }
-  d_glob_valid += w;
+  d_glob_valid += weight;
 
 iterate:
   for (i = 0; i < n_variables; i++)
@@ -658,7 +647,7 @@ iterate:
       if (X == SYSMIS
          || (!opt[op_incl_miss] && is_num_user_missing (X, v_variables[i])))
        {
-         inf->miss += w;
+         inf->miss += weight;
          continue;
        }
 
@@ -670,19 +659,19 @@ iterate:
          I am happy that mathematical formulas may not be
          copyrighted. */
       W_old = inf->valid;
-      W_new = inf->valid += w;
-      v = (w / W_new) * (X - inf->X_bar);
+      W_new = inf->valid += weight;
+      v = (weight / W_new) * (X - inf->X_bar);
       v2 = v * v;
       v3 = v2 * v;
       v4 = v3 * v;
-      w2 = w * w;
-      w3 = w2 * w;
-      w4 = w3 * w;
+      w2 = weight * weight;
+      w3 = w2 * weight;
+      w4 = w3 * weight;
       inf->M4 += (-4.0 * v * inf->M3 + 6.0 * v2 * inf->M2
-              + (W_new * W_new - 3 * w * W_old / w3) * v4 * W_old * W_new);
+              + (W_new * W_new - 3 * weight * W_old / w3) * v4 * W_old * W_new);
       inf->M3 += (-3.0 * v * inf->M2 + W_new * W_old / w2
-                 * (W_new - 2 * w) * v3);
-      inf->M2 += W_new * W_old / w * v2;
+                 * (W_new - 2 * weight) * v3);
+      inf->M2 += W_new * W_old / weight * v2;
       inf->X_bar += v;
       if (X < inf->min)
        inf->min = X;
@@ -693,7 +682,7 @@ iterate:
 }
 
 static void
-postcalc (void)
+postcalc (void *aux UNUSED)
 {
   int i;