Beginning of VFM cleanup.

[pspp-builds.git] / src / frequencies.q

diff --git a/src/frequencies.q b/src/frequencies.q
index 98f5b5802586048530278c4d119358dff0677bc5..0d896b4711f3a6f2f7f9db6ccc67a5da7eff191b 100644 (file)
--- a/src/frequencies.q
+++ b/src/frequencies.q
@@ -160,9 +160,9 @@ static struct pool *gen_pool;       /* General mode. */
 
 static void determine_charts (void);
 
-static void precalc (void);
-static int calc (struct ccase *);
-static void postcalc (void);
+static void precalc (void *);
+static int calc (struct ccase *, void *);
+static void postcalc (void *);
 
 static void postprocess_freq_tab (struct variable *);
 static void dump_full (struct variable *);
@@ -208,8 +208,8 @@ internal_cmd_frequencies (void)
   n_variables = 0;
   v_variables = NULL;
 
-  for (i = 0; i < default_dict.nvar; i++)
-    default_dict.var[i]->foo = 0;
+  for (i = 0; i < dict_get_var_cnt (default_dict); i++)
+    dict_get_var(default_dict, i)->p.frq.used = 0;
 
   lex_match_id ("FREQUENCIES");
   if (!parse_frequencies (&cmd))
@@ -246,8 +246,7 @@ internal_cmd_frequencies (void)
     cmd.sort = FRQ_AVALUE;
 
   /* Do it! */
-  update_weighting (&default_dict);
-  procedure (precalc, calc, postcalc);
+  procedure (precalc, calc, postcalc, NULL);
 
   return CMD_SUCCESS;
 }
@@ -356,15 +355,12 @@ determine_charts (void)
 
 /* Add data from case C to the frequency table. */
 static int
-calc (struct ccase *c)
+calc (struct ccase *c, void *aux UNUSED)
 {
   double weight;
   int i;
 
-  if (default_dict.weight_index == -1)
-    weight = 1.0;
-  else
-    weight = c->data[default_dict.var[default_dict.weight_index]->fv].f;
+  weight = dict_get_case_weight (default_dict, c);
 
   for (i = 0; i < n_variables; i++)
     {
@@ -412,7 +408,7 @@ calc (struct ccase *c)
 /* Prepares each variable that is the target of FREQUENCIES by setting
    up its hash table. */
 static void
-precalc (void)
+precalc (void *aux UNUSED)
 {
   int i;
 
@@ -455,7 +451,7 @@ precalc (void)
 /* Finishes up with the variables after frequencies have been
    calculated.  Displays statistics, percentiles, ... */
 static void
-postcalc (void)
+postcalc (void *aux UNUSED)
 {
   int i;
 
@@ -518,6 +514,8 @@ get_freq_comparator (int frq_sort, int var_type)
     case FRQ_DFREQ | (ALPHA << 16):     return compare_freq_alpha_d;
     default: assert (0);
     }
+
+  return 0;
 }
 
 static int
@@ -582,12 +580,13 @@ cleanup_freq_tab (struct variable *v)
 {
   assert (v->p.frq.tab.mode == FRQM_GENERAL);
   free (v->p.frq.tab.valid);
+  hsh_destroy (v->p.frq.tab.data);
 }
 
 /* Parses the VARIABLES subcommand, adding to
    {n_variables,v_variables}. */
 static int
-frq_custom_variables (struct cmd_frequencies *cmd unused)
+frq_custom_variables (struct cmd_frequencies *cmd UNUSED)
 {
   int mode;
   int min, max;
@@ -596,10 +595,11 @@ frq_custom_variables (struct cmd_frequencies *cmd unused)
   int i;
 
   lex_match ('=');
-  if (token != T_ALL && (token != T_ID || !is_varname (tokid)))
+  if (token != T_ALL && (token != T_ID
+                         || dict_lookup_var (default_dict, tokid) == NULL))
     return 2;
 
-  if (!parse_variables (NULL, &v_variables, &n_variables,
+  if (!parse_variables (default_dict, &v_variables, &n_variables,
                        PV_APPEND | PV_NO_SCRATCH))
     return 0;
 
@@ -635,14 +635,14 @@ frq_custom_variables (struct cmd_frequencies *cmd unused)
     {
       struct variable *v = v_variables[i];
 
-      if (v->foo != 0)
+      if (v->p.frq.used != 0)
        {
          msg (SE, _("Variable %s specified multiple times on VARIABLES "
                     "subcommand."), v->name);
          return 0;
        }
       
-      v->foo = 1;              /* Used simply as a marker. */
+      v->p.frq.used = 1;               /* Used simply as a marker. */
 
       v->p.frq.tab.valid = v->p.frq.tab.missing = NULL;
 
@@ -672,10 +672,11 @@ frq_custom_variables (struct cmd_frequencies *cmd unused)
 /* Parses the GROUPED subcommand, setting the frq.{n_grouped,grouped}
    fields of specified variables. */
 static int
-frq_custom_grouped (struct cmd_frequencies *cmd unused)
+frq_custom_grouped (struct cmd_frequencies *cmd UNUSED)
 {
   lex_match ('=');
-  if ((token == T_ID && is_varname (tokid)) || token == T_ID)
+  if ((token == T_ID && dict_lookup_var (default_dict, tokid) != NULL)
+      || token == T_ID)
     for (;;)
       {
        int i;
@@ -688,7 +689,8 @@ frq_custom_grouped (struct cmd_frequencies *cmd unused)
        int n;
        struct variable **v;
 
-       if (!parse_variables (NULL, &v, &n, PV_NO_DUPLICATE | PV_NUMERIC))
+       if (!parse_variables (default_dict, &v, &n,
+                              PV_NO_DUPLICATE | PV_NUMERIC))
          return 0;
        if (lex_match ('('))
          {
@@ -719,7 +721,7 @@ frq_custom_grouped (struct cmd_frequencies *cmd unused)
 
        for (i = 0; i < n; i++)
          {
-           if (v[i]->foo == 0)
+           if (v[i]->p.frq.used == 0)
              msg (SE, _("Variables %s specified on GROUPED but not on "
                   "VARIABLES."), v[i]->name);
            if (v[i]->p.frq.groups != NULL)
@@ -734,7 +736,8 @@ frq_custom_grouped (struct cmd_frequencies *cmd unused)
        free (v);
        if (!lex_match ('/'))
          break;
-       if ((token != T_ID || !is_varname (tokid)) && token != T_ALL)
+       if ((token != T_ID || dict_lookup_var (default_dict, tokid) != NULL)
+            && token != T_ALL)
          {
            lex_put_back ('/');
            break;
@@ -777,7 +780,7 @@ add_percentile (double x)
 /* Parses the PERCENTILES subcommand, adding user-specified
    percentiles to the list. */
 static int
-frq_custom_percentiles (struct cmd_frequencies *cmd unused)
+frq_custom_percentiles (struct cmd_frequencies *cmd UNUSED)
 {
   lex_match ('=');
   if (token != T_NUM)
@@ -806,7 +809,7 @@ frq_custom_percentiles (struct cmd_frequencies *cmd unused)
 /* Parses the NTILES subcommand, adding the percentiles that
    correspond to the specified evenly-distributed ntiles. */
 static int
-frq_custom_ntiles (struct cmd_frequencies *cmd unused)
+frq_custom_ntiles (struct cmd_frequencies *cmd UNUSED)
 {
   int i;
 
@@ -823,7 +826,7 @@ frq_custom_ntiles (struct cmd_frequencies *cmd unused)
 
 /* Hash of numeric values. */
 static unsigned
-hash_value_numeric (const void *value_, void *foo unused)
+hash_value_numeric (const void *value_, void *foo UNUSED)
 {
   const struct freq *value = value_;
   return hsh_hash_double (value->v.f);
@@ -831,17 +834,17 @@ hash_value_numeric (const void *value_, void *foo unused)
 
 /* Hash of string values. */
 static unsigned
-hash_value_alpha (const void *value_, void *len_)
+hash_value_alpha (const void *value_, void *v_)
 {
   const struct freq *value = value_;
-  int *len = len_;
+  struct variable *v = v_;
 
-  return hsh_hash_bytes (value->v.s, *len);
+  return hsh_hash_bytes (value->v.s, v->width);
 }
 
 /* Ascending numeric compare of values. */
 static int
-compare_value_numeric_a (const void *a_, const void *b_, void *foo unused)
+compare_value_numeric_a (const void *a_, const void *b_, void *foo UNUSED)
 {
   const struct freq *a = a_;
   const struct freq *b = b_;
@@ -867,7 +870,7 @@ compare_value_alpha_a (const void *a_, const void *b_, void *v_)
 
 /* Descending numeric compare of values. */
 static int
-compare_value_numeric_d (const void *a, const void *b, void *foo unused)
+compare_value_numeric_d (const void *a, const void *b, void *foo UNUSED)
 {
   return -compare_value_numeric_a (a, b, foo);
 }
@@ -882,14 +885,14 @@ compare_value_alpha_d (const void *a, const void *b, void *v)
 /* Ascending numeric compare of frequency;
    secondary key on ascending numeric value. */
 static int
-compare_freq_numeric_a (const void *a_, const void *b_, void *foo unused)
+compare_freq_numeric_a (const void *a_, const void *b_, void *foo UNUSED)
 {
   const struct freq *a = a_;
   const struct freq *b = b_;
 
-  if (a->v.c > b->v.c)
+  if (a->c > b->c)
     return 1;
-  else if (a->v.c < b->v.c)
+  else if (a->c < b->c)
     return -1;
 
   if (a->v.f > b->v.f)
@@ -909,9 +912,9 @@ compare_freq_alpha_a (const void *a_, const void *b_, void *v_)
   const struct freq *b = b_;
   const struct variable *v = v_;
 
-  if (a->v.c > b->v.c)
+  if (a->c > b->c)
     return 1;
-  else if (a->v.c < b->v.c)
+  else if (a->c < b->c)
     return -1;
   else
     return memcmp (a->v.s, b->v.s, v->width);
@@ -920,14 +923,14 @@ compare_freq_alpha_a (const void *a_, const void *b_, void *v_)
 /* Descending numeric compare of frequency;
    secondary key on ascending numeric value. */
 static int
-compare_freq_numeric_d (const void *a_, const void *b_, void *foo unused)
+compare_freq_numeric_d (const void *a_, const void *b_, void *foo UNUSED)
 {
   const struct freq *a = a_;
   const struct freq *b = b_;
 
-  if (a->v.c > b->v.c)
+  if (a->c > b->c)
     return -1;
-  else if (a->v.c < b->v.c)
+  else if (a->c < b->c)
     return 1;
 
   if (a->v.f > b->v.f)
@@ -947,9 +950,9 @@ compare_freq_alpha_d (const void *a_, const void *b_, void *v_)
   const struct freq *b = b_;
   const struct variable *v = v_;
 
-  if (a->v.c > b->v.c)
+  if (a->c > b->c)
     return -1;
-  else if (a->v.c < b->v.c)
+  else if (a->c < b->c)
     return 1;
   else
     return memcmp (a->v.s, b->v.s, v->width);
@@ -981,7 +984,7 @@ dump_full (struct variable * v)
   struct freq *f;
   struct tab_table *t;
   int r;
-  double cum_percent = 0.0;
+  double cum_total = 0.0;
   double cum_freq = 0.0;
 
   struct init
@@ -1030,7 +1033,7 @@ dump_full (struct variable * v)
 
       percent = f->c / v->p.frq.tab.total_cases * 100.0;
       valid_percent = f->c / v->p.frq.tab.valid_cases * 100.0;
-      cum_percent += valid_percent;
+      cum_total += valid_percent;
 
       if (lab)
        {
@@ -1043,7 +1046,7 @@ dump_full (struct variable * v)
       tab_float (t, 1 + lab, r, TAB_NONE, f->c, 8, 0);
       tab_float (t, 2 + lab, r, TAB_NONE, percent, 5, 1);
       tab_float (t, 3 + lab, r, TAB_NONE, valid_percent, 5, 1);
-      tab_float (t, 4 + lab, r, TAB_NONE, cum_percent, 5, 1);
+      tab_float (t, 4 + lab, r, TAB_NONE, cum_total, 5, 1);
       r++;
     }
   for (; f < &v->p.frq.tab.valid[n_categories]; f++)
@@ -1107,7 +1110,7 @@ dump_condensed (struct variable * v)
   struct freq *f;
   struct tab_table *t;
   int r;
-  double cum_percent = 0.0;
+  double cum_total = 0.0;
 
   n_categories = v->p.frq.tab.n_valid + v->p.frq.tab.n_missing;
   t = tab_create (4, n_categories + 2, 0);
@@ -1126,12 +1129,12 @@ dump_condensed (struct variable * v)
       double percent;
 
       percent = f->c / v->p.frq.tab.total_cases * 100.0;
-      cum_percent += f->c / v->p.frq.tab.valid_cases * 100.0;
+      cum_total += f->c / v->p.frq.tab.valid_cases * 100.0;
 
       tab_value (t, 0, r, TAB_NONE, &f->v, &v->print);
       tab_float (t, 1, r, TAB_NONE, f->c, 8, 0);
       tab_float (t, 2, r, TAB_NONE, percent, 3, 0);
-      tab_float (t, 3, r, TAB_NONE, cum_percent, 3, 0);
+      tab_float (t, 3, r, TAB_NONE, cum_total, 3, 0);
       r++;
     }
   for (; f < &v->p.frq.tab.valid[n_categories]; f++)
@@ -1164,7 +1167,7 @@ calc_stats (struct variable * v, double d[frq_n_stats])
   struct freq *f;
   int most_often;
 
-  double cum_percent;
+  double cum_total;
   int i = 0;
   double previous_value;
 
@@ -1175,14 +1178,14 @@ calc_stats (struct variable * v, double d[frq_n_stats])
   X_bar /= W;
 
   /* Calculate percentiles. */
-  cum_percent = 0;
+  cum_total = 0;
   previous_value = SYSMIS;
   for (f = v->p.frq.tab.valid; f < v->p.frq.tab.missing; f++)
     {
-      cum_percent += f->c / v->p.frq.tab.valid_cases;
+      cum_total += f->c ;
       for (; i < n_percentiles; i++) 
         {
-          if (cum_percent <= percentiles[i])
+          if (cum_total / v->p.frq.tab.valid_cases < percentiles[i])
             break;
 
           percentile_values[i] = previous_value;