X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Ffrequencies.c;h=8fb96b0d57e2316d59ecda71099d317d189e10db;hb=3368d7a7d2e5fd6c284781aa327daff250922756;hp=da1a3828d1d13a2ca4f42768537258b67fc08eff;hpb=e1335ea67114822554e0d51bd5e0b49401b17dc4;p=pspp

diff --git a/src/language/stats/frequencies.c b/src/language/stats/frequencies.c
index da1a3828d1..8fb96b0d57 100644
--- a/src/language/stats/frequencies.c
+++ b/src/language/stats/frequencies.c
@@ -1,6 +1,6 @@
 /*
   PSPP - a program for statistical analysis.
-  Copyright (C) 1997-9, 2000, 2007, 2009, 2010, 2011, 2014 Free Software Foundation, Inc.
+  Copyright (C) 1997-9, 2000, 2007, 2009, 2010, 2011, 2014, 2015 Free Software Foundation, Inc.
    
   This program is free software: you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
@@ -53,6 +53,7 @@
 
 
 #include "output/chart-item.h"
+#include "output/charts/barchart.h"
 #include "output/charts/piechart.h"
 #include "output/charts/plot-hist.h"
 #include "output/tab.h"
@@ -82,6 +83,14 @@ ptile_3way (const void *_p1, const void *_p2)
   if (p1->p < p2->p)
     return -1;
 
+  if (p1->p == p2->p)
+    {
+      if (p1->show > p2->show)
+	return -1;
+
+      return (p1->show < p2->show);
+    }
+
   return (p1->p > p2->p);
 }
 
@@ -205,7 +214,7 @@ struct frq_proc
     int n_percentiles, n_show_percentiles;
 
     /* Frequency table display. */
-    int max_categories;         /* Maximum categories to show. */
+    long int max_categories;         /* Maximum categories to show. */
     int sort;                   /* FRQ_AVALUE or FRQ_DVALUE
                                    or FRQ_AFREQ or FRQ_DFREQ. */
 
@@ -214,7 +223,7 @@ struct frq_proc
     int n_stats;
 
     /* Histogram and pie chart settings. */
-    struct frq_chart *hist, *pie;
+    struct frq_chart *hist, *pie, *bar;
   };
 
 
@@ -233,6 +242,10 @@ static void do_piechart(const struct frq_chart *pie,
 			const struct variable *var,
 			const struct freq_tab *frq_tab);
 
+static void do_barchart(const struct frq_chart *bar,
+			const struct variable **var,
+			const struct freq_tab *frq_tab);
+
 static void dump_statistics (const struct frq_proc *frq, 
 			     const struct var_freqs *vf,
 			     const struct variable *wv);
@@ -251,7 +264,7 @@ compare_freq (const void *a_, const void *b_, const void *aux_)
     }
   else
     {
-      int cmp = value_compare_3way (&a->value, &b->value, aux->width);
+      int cmp = value_compare_3way (a->values, b->values, aux->width);
       return aux->ascending_value ? cmp : -cmp;
     }
 }
@@ -304,11 +317,11 @@ dump_freq_table (const struct var_freqs *vf, const struct variable *wv)
       valid_percent = f->count / ft->valid_cases * 100.0;
       cum_total += valid_percent;
 
-      label = var_lookup_value_label (vf->var, &f->value);
+      label = var_lookup_value_label (vf->var, f->values);
       if (label != NULL)
         tab_text (t, 0, r, TAB_LEFT, label);
 
-      tab_value (t, 1, r, TAB_NONE, &f->value, vf->var, NULL);
+      tab_value (t, 1, r, TAB_NONE, f->values, vf->var, NULL);
       tab_double (t, 2, r, TAB_NONE, f->count, NULL, RC_WEIGHT);
       tab_double (t, 3, r, TAB_NONE, percent, NULL, RC_OTHER);
       tab_double (t, 4, r, TAB_NONE, valid_percent, NULL, RC_OTHER);
@@ -321,11 +334,11 @@ dump_freq_table (const struct var_freqs *vf, const struct variable *wv)
 
       cum_freq += f->count;
 
-      label = var_lookup_value_label (vf->var, &f->value);
+      label = var_lookup_value_label (vf->var, f->values);
       if (label != NULL)
         tab_text (t, 0, r, TAB_LEFT, label);
 
-      tab_value (t, 1, r, TAB_NONE, &f->value, vf->var, NULL);
+      tab_value (t, 1, r, TAB_NONE, f->values, vf->var, NULL);
       tab_double (t, 2, r, TAB_NONE, f->count, NULL, RC_WEIGHT);
       tab_double (t, 3, r, TAB_NONE,
 		  f->count / ft->total_cases * 100.0, NULL, RC_OTHER);
@@ -367,13 +380,9 @@ calc_percentiles (const struct frq_proc *frq, const struct var_freqs *vf)
   const struct freq_tab *ft = &vf->tab;
   double W = ft->valid_cases;
   const struct freq *f;
-  int percentile_idx;
-  double rank;
-
-  assert (ft->n_valid > 0);
+  int percentile_idx = 0;
+  double  rank = 0;
 
-  rank = 0;
-  percentile_idx = 0;
   for (f = ft->valid; f < ft->missing; f++)
     {
       rank += f->count;
@@ -390,15 +399,15 @@ calc_percentiles (const struct frq_proc *frq, const struct var_freqs *vf)
             break;
 
           if (tp + 1 < rank || f + 1 >= ft->missing)
-            pc->value = f->value.f;
+            pc->value = f->values[0].f;
           else
-            pc->value = calc_percentile (pc->p, W, f->value.f, f[1].value.f);
+            pc->value = calc_percentile (pc->p, W, f->values[0].f, f[1].values[0].f);
         }
     }
   for (; percentile_idx < frq->n_percentiles; percentile_idx++)
     {
       struct percentile *pc = &frq->percentiles[percentile_idx];
-      pc->value = ft->valid[ft->n_valid - 1].value.f;
+      pc->value = ft->valid[ft->n_valid - 1].values[0].f;
     }
 }
 
@@ -410,7 +419,7 @@ not_missing (const void *f_, const void *v_)
   const struct freq *f = f_;
   const struct variable *v = v_;
 
-  return !var_is_value_missing (v, &f->value, MV_ANY);
+  return !var_is_value_missing (v, f->values, MV_ANY);
 }
 
 
@@ -559,6 +568,9 @@ postcalc (struct frq_proc *frq, const struct dataset *ds)
       if (frq->pie)
         do_piechart(frq->pie, vf->var, &vf->tab);
 
+      if (frq->bar)
+        do_barchart(frq->bar, &vf->var, &vf->tab);
+
       cleanup_freq_tab (vf);
     }
 }
@@ -576,7 +588,11 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 
   double pie_min = -DBL_MAX;
   double pie_max = DBL_MAX;
-  bool pie_missing = false;
+  bool pie_missing = true;
+
+  double bar_min = -DBL_MAX;
+  double bar_max = DBL_MAX;
+  bool bar_freq = true;
 
   double hi_min = -DBL_MAX;
   double hi_max = DBL_MAX;
@@ -598,7 +614,7 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 
   frq.n_stats = 4;
 
-  frq.max_categories = INT_MAX;
+  frq.max_categories = LONG_MAX;
 
   frq.percentiles = NULL;
   frq.n_percentiles = 0;
@@ -606,6 +622,7 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 
   frq.hist = NULL;
   frq.pie = NULL;
+  frq.bar = NULL;
 
 
   /* Accept an optional, completely pointless "/VARIABLES=" */
@@ -770,6 +787,7 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 		  lex_error (lexer, NULL);
 		  goto error;
 		}
+              lex_match (lexer, T_COMMA);
 	    }
 	}
       else if (lex_match_id (lexer, "FORMAT"))
@@ -780,12 +798,23 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 	    {
 	      if (lex_match_id (lexer, "TABLE"))
 		{
-		  
 		}
 	      else if (lex_match_id (lexer, "NOTABLE"))
 		{
 		  frq.max_categories = 0;
 		}
+              else if (lex_match_id (lexer, "LIMIT"))
+                {
+                  if (!lex_force_match (lexer, T_LPAREN)
+                      || !lex_force_int (lexer))
+                    goto error;
+
+                  frq.max_categories = lex_integer (lexer);
+                  lex_get (lexer);
+
+                  if (!lex_force_match (lexer, T_RPAREN))
+                    goto error;
+                }
 	      else if (lex_match_id (lexer, "AVALUE"))
 		{
 		  frq.sort = FRQ_AVALUE;
@@ -975,6 +1004,66 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 	    }
 	  sbc_piechart = true;
 	}
+      else if (lex_match_id (lexer, "BARCHART"))
+        {
+	  lex_match (lexer, T_EQUALS);
+	  while (lex_token (lexer) != T_ENDCMD
+		 && lex_token (lexer) != T_SLASH)
+	    {
+	      if (lex_match_id (lexer, "MINIMUM"))
+		{
+		  lex_force_match (lexer, T_LPAREN);
+		  if (lex_force_num (lexer))
+		    {
+		      bar_min = lex_number (lexer);
+		      lex_get (lexer);
+		    }
+		  lex_force_match (lexer, T_RPAREN);
+		}
+	      else if (lex_match_id (lexer, "MAXIMUM"))
+		{
+		  lex_force_match (lexer, T_LPAREN);
+		  if (lex_force_num (lexer))
+		    {
+		      bar_max = lex_number (lexer);
+		      lex_get (lexer);
+		    }
+ 		  lex_force_match (lexer, T_RPAREN);
+		}
+	      else if (lex_match_id (lexer, "FREQ"))
+		{
+		  if ( lex_match (lexer, T_LPAREN))
+		    {
+		      if (lex_force_num (lexer))
+			{
+			  lex_number (lexer);
+			  lex_get (lexer);
+			}
+		      lex_force_match (lexer, T_RPAREN);
+		    }
+		  bar_freq = true;
+		}
+	      else if (lex_match_id (lexer, "PERCENT"))
+		{
+		  if ( lex_match (lexer, T_LPAREN))
+		    {
+		      if (lex_force_num (lexer))
+			{
+			  lex_number (lexer);
+			  lex_get (lexer);
+			}
+		      lex_force_match (lexer, T_RPAREN);
+		    }
+		  bar_freq = false;
+		}
+	      else
+		{
+		  lex_error (lexer, NULL);
+		  goto error;
+		}
+	    }
+	  sbc_barchart = true;
+	}
       else if (lex_match_id (lexer, "MISSING"))
         {
 	  lex_match (lexer, T_EQUALS);
@@ -995,6 +1084,12 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
                 }
             }
         }
+      else if (lex_match_id (lexer, "ORDER"))
+        {
+          lex_match (lexer, T_EQUALS);
+          if (!lex_match_id (lexer, "ANALYSIS"))
+            lex_match_id (lexer, "VARIABLE");
+        }
       else
         {
           lex_error (lexer, NULL);
@@ -1019,9 +1114,6 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 /* Figure out which charts the user requested.  */
 
   {
-    if (sbc_barchart)
-      msg (SW, _("Bar charts are not implemented."));
-
     if (sbc_histogram)
       {
 	struct frq_chart *hist;
@@ -1061,6 +1153,15 @@ cmd_frequencies (struct lexer *lexer, struct dataset *ds)
 	frq.n_percentiles+=2;
       }
 
+    if (sbc_barchart)
+      {
+	frq.bar = xmalloc (sizeof *frq.bar);
+	frq.bar->x_min = bar_min;
+	frq.bar->x_max = bar_max;
+	frq.bar->include_missing = false;
+	frq.bar->y_scale = bar_freq ? FRQ_FREQ : FRQ_PERCENT;
+      }
+
     if (sbc_piechart)
       {
 	struct frq_chart *pie;
@@ -1197,10 +1298,10 @@ freq_tab_to_hist (const struct frq_proc *frq, const struct freq_tab *ft,
   for (i = 0; i < ft->n_valid; i++)
     {
       const struct freq *f = &ft->valid[i];
-      if (chart_includes_value (frq->hist, var, &f->value))
+      if (chart_includes_value (frq->hist, var, f->values))
         {
-          x_min = MIN (x_min, f->value.f);
-          x_max = MAX (x_max, f->value.f);
+          x_min = MIN (x_min, f->values[0].f);
+          x_max = MAX (x_max, f->values[0].f);
           valid_freq += f->count;
         }
     }
@@ -1224,63 +1325,118 @@ freq_tab_to_hist (const struct frq_proc *frq, const struct freq_tab *ft,
   for (i = 0; i < ft->n_valid; i++)
     {
       const struct freq *f = &ft->valid[i];
-      if (chart_includes_value (frq->hist, var, &f->value))
-        histogram_add (histogram, f->value.f, f->count);
+      if (chart_includes_value (frq->hist, var, f->values))
+        histogram_add (histogram, f->values[0].f, f->count);
     }
 
   return histogram;
 }
 
-static int
-add_slice (const struct frq_chart *pie, const struct freq *freq,
-           const struct variable *var, struct slice *slice)
+
+/* Allocate an array of struct freqs and fill them from the data in FRQ_TAB,
+   according to the parameters of CATCHART
+   N_SLICES will contain the number of slices allocated.
+   The caller is responsible for freeing slices
+*/
+static struct freq *
+pick_cat_counts (const struct frq_chart *catchart,
+		 const struct freq_tab *frq_tab,
+		 int *n_slicesp)
 {
-  if (chart_includes_value (pie, var, &freq->value))
+  int n_slices = 0;
+  int i;
+  struct freq *slices = xnmalloc (frq_tab->n_valid + frq_tab->n_missing, sizeof *slices);
+  
+  for (i = 0; i < frq_tab->n_valid; i++)
     {
-      ds_init_empty (&slice->label);
-      var_append_value_name (var, &freq->value, &slice->label);
-      slice->magnitude = freq->count;
-      return 1;
+      const struct freq *f = &frq_tab->valid[i];
+      if (f->count > catchart->x_max)
+	continue;
+
+      if (f->count < catchart->x_min)
+	continue;
+      
+      slices[n_slices] = *f;
+      
+      n_slices++;
     }
-  else
-    return 0;
+
+  if (catchart->include_missing)
+    {
+      for (i = 0; i < frq_tab->n_missing; i++)
+	{
+	  const struct freq *f = &frq_tab->missing[i];
+	  slices[n_slices].count += f->count;
+	  
+	  if (i == 0)
+	    slices[n_slices].values[0] = f->values[0];
+	}
+      
+      if (frq_tab->n_missing > 0)
+	n_slices++;
+    }
+
+  *n_slicesp = n_slices;
+  return slices;
 }
 
-/* Allocate an array of slices and fill them from the data in frq_tab
-   n_slices will contain the number of slices allocated.
+
+/* Allocate an array of struct freqs and fill them from the data in FRQ_TAB,
+   according to the parameters of CATCHART
+   N_SLICES will contain the number of slices allocated.
    The caller is responsible for freeing slices
 */
-static struct slice *
-freq_tab_to_slice_array(const struct frq_chart *pie,
-                        const struct freq_tab *frq_tab,
-			const struct variable *var,
-			int *n_slicesp)
+static struct freq **
+pick_cat_counts_ptr (const struct frq_chart *catchart,
+		     const struct freq_tab *frq_tab,
+		     int *n_slicesp)
 {
-  struct slice *slices;
-  int n_slices;
+  int n_slices = 0;
   int i;
+  struct freq **slices = xnmalloc (frq_tab->n_valid + frq_tab->n_missing, sizeof *slices);
+  
+  for (i = 0; i < frq_tab->n_valid; i++)
+    {
+      struct freq *f = &frq_tab->valid[i];
+      if (f->count > catchart->x_max)
+	continue;
 
-  slices = xnmalloc (frq_tab->n_valid + frq_tab->n_missing, sizeof *slices);
-  n_slices = 0;
+      if (f->count < catchart->x_min)
+	continue;
+      
+      slices[n_slices] = f;
+      
+      n_slices++;
+    }
 
-  for (i = 0; i < frq_tab->n_valid; i++)
-    n_slices += add_slice (pie, &frq_tab->valid[i], var, &slices[n_slices]);
-  for (i = 0; i < frq_tab->n_missing; i++)
-    n_slices += add_slice (pie, &frq_tab->missing[i], var, &slices[n_slices]);
+  if (catchart->include_missing)
+    {
+      for (i = 0; i < frq_tab->n_missing; i++)
+	{
+	  const struct freq *f = &frq_tab->missing[i];
+	  if (i == 0)
+	    {
+	      slices[n_slices] = xmalloc (sizeof (struct freq));
+	      slices[n_slices]->values[0] = f->values[0];
+	    }
+
+	  slices[n_slices]->count += f->count;
+	  
+	}
+    }
 
   *n_slicesp = n_slices;
   return slices;
 }
 
 
+
 static void
 do_piechart(const struct frq_chart *pie, const struct variable *var,
             const struct freq_tab *frq_tab)
 {
-  struct slice *slices;
-  int n_slices, i;
-
-  slices = freq_tab_to_slice_array (pie, frq_tab, var, &n_slices);
+  int n_slices;
+  struct freq *slices = pick_cat_counts (pie, frq_tab, &n_slices);
 
   if (n_slices < 2)
     msg (SW, _("Omitting pie chart for %s, which has only %d unique values."),
@@ -1289,13 +1445,26 @@ do_piechart(const struct frq_chart *pie, const struct variable *var,
     msg (SW, _("Omitting pie chart for %s, which has over 50 unique values."),
          var_get_name (var));
   else
-    chart_item_submit (piechart_create (var_to_string(var), slices, n_slices));
+    chart_item_submit (piechart_create (var, slices, n_slices));
+
+  free (slices);
+}
+
+
+static void
+do_barchart(const struct frq_chart *bar, const struct variable **var,
+            const struct freq_tab *frq_tab)
+{
+  int n_slices;
+  struct freq **slices = pick_cat_counts_ptr (bar, frq_tab, &n_slices);
 
-  for (i = 0; i < n_slices; i++)
-    ds_destroy (&slices[i].label);
+  chart_item_submit (barchart_create (var, 1,
+				      (bar->y_scale == FRQ_FREQ) ? _("Count") : _("Percent"),
+				      slices, n_slices));
   free (slices);
 }
 
+
 /* Calculates all the pertinent statistics for VF, putting them in array
    D[]. */
 static void
@@ -1305,20 +1474,16 @@ calc_stats (const struct var_freqs *vf, double d[FRQ_ST_count])
   double W = ft->valid_cases;
   const struct freq *f;
   struct moments *m;
-  int most_often;
-  double X_mode;
-
-  assert (ft->n_valid > 0);
+  int most_often = -1;
+  double X_mode = SYSMIS;
 
   /* Calculate the mode. */
-  most_often = -1;
-  X_mode = SYSMIS;
   for (f = ft->valid; f < ft->missing; f++)
     {
       if (most_often < f->count)
         {
           most_often = f->count;
-          X_mode = f->value.f;
+          X_mode = f->values[0].f;
         }
       else if (most_often == f->count)
         {
@@ -1331,16 +1496,16 @@ calc_stats (const struct var_freqs *vf, double d[FRQ_ST_count])
   /* Calculate moments. */
   m = moments_create (MOMENT_KURTOSIS);
   for (f = ft->valid; f < ft->missing; f++)
-    moments_pass_one (m, f->value.f, f->count);
+    moments_pass_one (m, f->values[0].f, f->count);
   for (f = ft->valid; f < ft->missing; f++)
-    moments_pass_two (m, f->value.f, f->count);
+    moments_pass_two (m, f->values[0].f, f->count);
   moments_calculate (m, NULL, &d[FRQ_ST_MEAN], &d[FRQ_ST_VARIANCE],
                      &d[FRQ_ST_SKEWNESS], &d[FRQ_ST_KURTOSIS]);
   moments_destroy (m);
 
-  /* Formulas below are taken from _SPSS Statistical Algorithms_. */
-  d[FRQ_ST_MINIMUM] = ft->valid[0].value.f;
-  d[FRQ_ST_MAXIMUM] = ft->valid[ft->n_valid - 1].value.f;
+  /* Formulae below are taken from _SPSS Statistical Algorithms_. */
+  d[FRQ_ST_MINIMUM] = ft->valid[0].values[0].f;
+  d[FRQ_ST_MAXIMUM] = ft->valid[ft->n_valid - 1].values[0].f;
   d[FRQ_ST_MODE] = X_mode;
   d[FRQ_ST_RANGE] = d[FRQ_ST_MAXIMUM] - d[FRQ_ST_MINIMUM];
   d[FRQ_ST_SUM] = d[FRQ_ST_MEAN] * W;
@@ -1359,30 +1524,22 @@ dump_statistics (const struct frq_proc *frq, const struct var_freqs *vf,
   const struct freq_tab *ft = &vf->tab;
   double stat_value[FRQ_ST_count];
   struct tab_table *t;
-  int i, r;
+  int i, r = 2; /* N missing and N valid are always dumped */
 
   if (var_is_alpha (vf->var))
     return;
 
-  if (ft->n_valid == 0)
-    {
-      msg (SW, _("No valid data for variable %s; statistics not displayed."),
-	   var_get_name (vf->var));
-      return;
-    }
   calc_stats (vf, stat_value);
 
   t = tab_create (3, ((frq->stats & BIT_INDEX (FRQ_ST_MEDIAN)) ? frq->n_stats - 1 : frq->n_stats)
-		  + frq->n_show_percentiles + 2);
+		                + frq->n_show_percentiles + 2);
+
   tab_set_format (t, RC_WEIGHT, wfmt);
   tab_box (t, TAL_1, TAL_1, -1, -1 , 0 , 0 , 2, tab_nr(t) - 1) ;
 
-
   tab_vline (t, TAL_1 , 2, 0, tab_nr(t) - 1);
   tab_vline (t, TAL_GAP , 1, 0, tab_nr(t) - 1 ) ;
 
-  r = 2; /* N missing and N valid are always dumped */
-
   for (i = 0; i < FRQ_ST_count; i++)
     {
       if (FRQ_ST_MEDIAN == i)
@@ -1392,7 +1549,11 @@ dump_statistics (const struct frq_proc *frq, const struct var_freqs *vf,
       {
 	tab_text (t, 0, r, TAB_LEFT | TAT_TITLE,
 		      gettext (st_name[i]));
-	tab_double (t, 2, r, TAB_NONE, stat_value[i], NULL, RC_OTHER);
+
+	if (vf->tab.n_valid <= 0 && r >= 2)
+	  tab_text (t, 2, r, 0,   ".");
+	else
+	  tab_double (t, 2, r, TAB_NONE, stat_value[i], NULL, RC_OTHER);
 	r++;
       }
     }
@@ -1416,13 +1577,21 @@ dump_statistics (const struct frq_proc *frq, const struct var_freqs *vf,
 	  tab_text (t, 0, r, TAB_LEFT | TAT_TITLE, _("Percentiles"));
 	}
 
+      if (vf->tab.n_valid <= 0)
+	{
+	  tab_text (t, 2, r, 0,   ".");
+	  ++r;
+	  continue;
+	}
+
       if (pc->p == 0.5)
         tab_text (t, 1, r, TAB_LEFT, _("50 (Median)"));
       else
         tab_double (t, 1, r, TAB_LEFT, pc->p * 100, NULL, RC_INTEGER);
       tab_double (t, 2, r, TAB_NONE, pc->value,
                   var_get_print_format (vf->var), RC_OTHER);
-      r++;
+
+      ++r;
     }
 
   tab_title (t, "%s", var_to_string (vf->var));