MEANS: implemented the /CELLS={ALL,NONE,DEFAULT} options and added a test

[pspp-builds.git] / src / language / stats / means.c

diff --git a/src/language/stats/means.c b/src/language/stats/means.c
index 5312ace3a5424e05fd447929806574ba34c47e07..874ddda8b256d8030ebdf42558608dde68ce89d3 100644 (file)
--- a/src/language/stats/means.c
+++ b/src/language/stats/means.c
@@ -16,7 +16,6 @@
 
 #include <config.h>
 
-
 #include "data/case.h"
 #include "data/casegrouper.h"
 #include "data/casereader.h"
@@ -54,12 +53,9 @@ struct per_var_data
 };
 
 
-typedef void *stat_create (const struct means *, struct pool *pool);
-
-typedef void stat_update (const struct means *, void *stat, double w, double x);
-
-typedef double stat_get (const struct means *, struct per_var_data *, void *aux);
-
+typedef void *stat_create (struct pool *pool);
+typedef void stat_update  (void *stat, double w, double x);
+typedef double stat_get   (const struct per_var_data *, void *aux);
 
 struct cell_spec
 {
@@ -74,16 +70,14 @@ struct cell_spec
   stat_get *sd;
 };
 
-
 struct harmonic_mean
 {
   double rsum;
   double n;
 };
 
-
 static void *
-harmonic_create (const struct means *means UNUSED, struct pool *pool)
+harmonic_create (struct pool *pool)
 {
   struct harmonic_mean *hm = pool_alloc (pool, sizeof *hm);
 
@@ -95,7 +89,7 @@ harmonic_create (const struct means *means UNUSED, struct pool *pool)
 
 
 static void
-harmonic_update (const struct means *means UNUSED, void *stat, double w, double x)
+harmonic_update (void *stat, double w, double x)
 {
   struct harmonic_mean *hm = stat;
   hm->rsum  += w / x;
@@ -104,7 +98,7 @@ harmonic_update (const struct means *means UNUSED, void *stat, double w, double
 
 
 static double
-harmonic_get (const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+harmonic_get (const struct per_var_data *pvd UNUSED, void *stat)
 {
   struct harmonic_mean *hm = stat;
 
@@ -121,7 +115,7 @@ struct geometric_mean
 
 
 static void *
-geometric_create (const struct means *means UNUSED, struct pool *pool)
+geometric_create (struct pool *pool)
 {
   struct geometric_mean *gm = pool_alloc (pool, sizeof *gm);
 
@@ -133,7 +127,7 @@ geometric_create (const struct means *means UNUSED, struct pool *pool)
 
 
 static void
-geometric_update (const struct means *means UNUSED, void *stat, double w, double x)
+geometric_update (void *stat, double w, double x)
 {
   struct geometric_mean *gm = stat;
   gm->prod  *=  pow (x, w);
@@ -142,7 +136,7 @@ geometric_update (const struct means *means UNUSED, void *stat, double w, double
 
 
 static double
-geometric_get (const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+geometric_get (const struct per_var_data *pvd UNUSED, void *stat)
 {
   struct geometric_mean *gm = stat;
 
@@ -151,14 +145,8 @@ geometric_get (const struct means *means UNUSED, struct per_var_data *pvd, void
 
 \f
 
-static void *
-sum_create (const struct means *means UNUSED, struct pool *pool)
-{
-  return NULL;
-}
-
 static double
-sum_get (const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+sum_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double n, mean;
 
@@ -168,14 +156,8 @@ sum_get (const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
 }
 
 
-static void *
-n_create (const struct means *means UNUSED, struct pool *pool)
-{
-  return NULL;
-}
-
 static double
-n_get (const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+n_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double n;
 
@@ -184,14 +166,8 @@ n_get (const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
   return n;
 }
 
-static void *
-arithmean_create (const struct means *means UNUSED, struct pool *pool)
-{
-  return NULL;
-}
-
 static double
-arithmean_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+arithmean_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double n, mean;
 
@@ -200,14 +176,8 @@ arithmean_get (const const struct means *means UNUSED, struct per_var_data *pvd,
   return mean;
 }
 
-static void *
-stddev_create (const struct means *means UNUSED, struct pool *pool)
-{
-  return NULL;
-}
-
 static double
-variance_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+variance_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double n, mean, variance;
 
@@ -218,22 +188,16 @@ variance_get (const const struct means *means UNUSED, struct per_var_data *pvd,
 
 
 static double
-stddev_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+stddev_get (const struct per_var_data *pvd, void *stat)
 {
-  return sqrt (variance_get (means, pvd, stat));
+  return sqrt (variance_get (pvd, stat));
 }
 
 
 \f
 
-static void *
-skew_create (const struct means *means UNUSED, struct pool *pool)
-{
-  return NULL;
-}
-
 static double
-skew_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+skew_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double skew;
 
@@ -243,7 +207,7 @@ skew_get (const const struct means *means UNUSED, struct per_var_data *pvd, void
 }
 
 static double
-sekurt_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+sekurt_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double n;
 
@@ -252,10 +216,8 @@ sekurt_get (const const struct means *means UNUSED, struct per_var_data *pvd, vo
   return calc_sekurt (n);
 }
 
-
-
 static double
-seskew_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+seskew_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double n;
 
@@ -264,17 +226,8 @@ seskew_get (const const struct means *means UNUSED, struct per_var_data *pvd, vo
   return calc_seskew (n);
 }
 
-
-
-static void *
-kurt_create (const struct means *means UNUSED, struct pool *pool)
-{
-  return NULL;
-}
-
-
 static double
-kurt_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+kurt_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double kurt;
 
@@ -283,9 +236,8 @@ kurt_get (const const struct means *means UNUSED, struct per_var_data *pvd, void
   return kurt;
 }
 
-
 static double
-semean_get (const struct means *means, struct per_var_data *pvd, void *stat)
+semean_get (const struct per_var_data *pvd, void *stat UNUSED)
 {
   double n, var;
 
@@ -294,12 +246,10 @@ semean_get (const struct means *means, struct per_var_data *pvd, void *stat)
   return sqrt (var / n);
 }
 
-
-
 \f
 
 static void *
-min_create (const struct means *means UNUSED, struct pool *pool)
+min_create (struct pool *pool)
 {
   double *r = pool_alloc (pool, sizeof *r);
 
@@ -309,7 +259,7 @@ min_create (const struct means *means UNUSED, struct pool *pool)
 }
 
 static void
-min_update (const struct means *means UNUSED, void *stat, double w UNUSED, double x)
+min_update (void *stat, double w UNUSED, double x)
 {
   double *r = stat;
 
@@ -318,7 +268,7 @@ min_update (const struct means *means UNUSED, void *stat, double w UNUSED, doubl
 }
 
 static double
-min_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+min_get (const struct per_var_data *pvd UNUSED, void *stat)
 {
   double *r = stat;
 
@@ -326,7 +276,7 @@ min_get (const const struct means *means UNUSED, struct per_var_data *pvd, void
 }
 
 static void *
-max_create (const struct means *means UNUSED, struct pool *pool)
+max_create (struct pool *pool)
 {
   double *r = pool_alloc (pool, sizeof *r);
 
@@ -336,7 +286,7 @@ max_create (const struct means *means UNUSED, struct pool *pool)
 }
 
 static void
-max_update (const struct means *means UNUSED, void *stat, double w UNUSED, double x)
+max_update (void *stat, double w UNUSED, double x)
 {
   double *r = stat;
 
@@ -345,7 +295,7 @@ max_update (const struct means *means UNUSED, void *stat, double w UNUSED, doubl
 }
 
 static double
-max_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+max_get (const struct per_var_data *pvd UNUSED, void *stat)
 {
   double *r = stat;
 
@@ -361,7 +311,7 @@ struct range
 };
 
 static void *
-range_create (const struct means *means UNUSED, struct pool *pool)
+range_create (struct pool *pool)
 {
   struct range *r = pool_alloc (pool, sizeof *r);
 
@@ -372,7 +322,7 @@ range_create (const struct means *means UNUSED, struct pool *pool)
 }
 
 static void
-range_update (const struct means *means UNUSED, void *stat, double w UNUSED, double x)
+range_update (void *stat, double w UNUSED, double x)
 {
   struct range *r = stat;
 
@@ -384,7 +334,7 @@ range_update (const struct means *means UNUSED, void *stat, double w UNUSED, dou
 }
 
 static double
-range_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+range_get (const struct per_var_data *pvd UNUSED, void *stat)
 {
   struct range *r = stat;
 
@@ -394,7 +344,7 @@ range_get (const const struct means *means UNUSED, struct per_var_data *pvd, voi
 \f
 
 static void *
-last_create (const struct means *means UNUSED, struct pool *pool)
+last_create (struct pool *pool)
 {
   double *l = pool_alloc (pool, sizeof *l);
 
@@ -402,7 +352,7 @@ last_create (const struct means *means UNUSED, struct pool *pool)
 }
 
 static void
-last_update (const struct means *means UNUSED, void *stat, double w UNUSED, double x)
+last_update (void *stat, double w UNUSED, double x)
 {
   double *l = stat;
 
@@ -410,7 +360,7 @@ last_update (const struct means *means UNUSED, void *stat, double w UNUSED, doub
 }
 
 static double
-last_get (const const struct means *means UNUSED, struct per_var_data *pvd, void *stat)
+last_get (const struct per_var_data *pvd UNUSED, void *stat)
 {
   double *l = stat;
 
@@ -419,17 +369,17 @@ last_get (const const struct means *means UNUSED, struct per_var_data *pvd, void
 
 
 static void *
-first_create (const struct means *means UNUSED, struct pool *pool)
+first_create (struct pool *pool)
 {
   double *f = pool_alloc (pool, sizeof *f);
 
-   *f = SYSMIS;
+  *f = SYSMIS;
 
   return f;
 }
 
 static void
-first_update (const struct means *means UNUSED, void *stat, double w UNUSED, double x)
+first_update (void *stat, double w UNUSED, double x)
 {
   double *f = stat;
 
@@ -438,14 +388,19 @@ first_update (const struct means *means UNUSED, void *stat, double w UNUSED, dou
 }
 
 static double
-first_get (const const struct means *means UNUSED, struct per_var_data *pvd,  void *stat)
+first_get (const struct per_var_data *pvd UNUSED,  void *stat)
 {
   double *f = stat;
 
   return *f;
 }
 
-
+enum 
+  {
+    MEANS_MEAN = 0,
+    MEANS_N,
+    MEANS_STDDEV
+  };
 
 /* Table of cell_specs */
 static const struct cell_spec cell_spec[] = {
@@ -751,17 +706,44 @@ cmd_means (struct lexer *lexer, struct dataset *ds)
          while (lex_token (lexer) != T_ENDCMD
                 && lex_token (lexer) != T_SLASH)
            {
-             int k;
-             for (k = 0; k < n_C; ++k)
+             int k = 0;
+             if (lex_match (lexer, T_ALL))
                {
-                 if (lex_match_id (lexer, cell_spec[k].keyword))
-                   {
-                     means.cells =
-                       xrealloc (means.cells,
-                                 ++means.n_cells * sizeof (*means.cells));
+                 int x;
+                 means.cells =
+                   xrealloc (means.cells,
+                             (means.n_cells += n_C) * sizeof (*means.cells));
 
-                     means.cells[means.n_cells - 1] = k;
-                     break;
+                 for (x = 0; x < n_C; ++x)
+                   means.cells[means.n_cells - (n_C - 1 - x) - 1] = x;
+               }
+             else if (lex_match_id (lexer, "NONE"))
+               {
+                 /* Do nothing */
+               }
+             else if (lex_match_id (lexer, "DEFAULT"))
+               {
+                 means.cells =
+                   xrealloc (means.cells,
+                             (means.n_cells += 3) * sizeof (*means.cells));
+                 
+                 means.cells[means.n_cells - 2 - 1] = MEANS_MEAN;
+                 means.cells[means.n_cells - 1 - 1] = MEANS_N;
+                 means.cells[means.n_cells - 0 - 1] = MEANS_STDDEV;
+               }
+             else
+               {
+                 for (; k < n_C; ++k)
+                   {
+                     if (lex_match_id (lexer, cell_spec[k].keyword))
+                       {
+                         means.cells =
+                           xrealloc (means.cells,
+                                     ++means.n_cells * sizeof (*means.cells));
+                       
+                         means.cells[means.n_cells - 1] = k;
+                         break;
+                       }
                    }
                }
              if (k >= n_C)
@@ -883,7 +865,7 @@ create_n (const void *aux1, void *aux2)
          const struct cell_spec *cs = &cell_spec[csi];
          if (cs->sc)
            {
-             pp->cell_stats[i] = cs->sc (means, means->pool);
+             pp->cell_stats[i] = cs->sc (means->pool);
            }
        }
       pp->mom = moments1_create (maxmom);
@@ -896,7 +878,7 @@ create_n (const void *aux1, void *aux2)
 }
 
 static void
-update_n (const void *aux1, void *aux2, void *user_data, const struct ccase *c)
+update_n (const void *aux1, void *aux2, void *user_data, const struct ccase *c, double weight)
 {
   int i;
   int v = 0;
@@ -904,8 +886,6 @@ update_n (const void *aux1, void *aux2, void *user_data, const struct ccase *c)
   struct mtable *table = aux2;
   struct per_cat_data *per_cat_data = user_data;
 
-  double weight = dict_get_case_weight (means->dict, c, &per_cat_data->warn);
-
   for (v = 0; v < table->n_dep_vars; ++v)
     {
       struct per_var_data *pvd = &per_cat_data->pvd[v];
@@ -925,8 +905,7 @@ update_n (const void *aux1, void *aux2, void *user_data, const struct ccase *c)
 
 
          if (cs->su)
-           cs->su (means,
-                   pvd->cell_stats[i],
+           cs->su (pvd->cell_stats[i],
                    weight, x);
        }
 
@@ -955,7 +934,7 @@ calculate_n (const void *aux1, void *aux2, void *user_data)
          const struct cell_spec *cs = &cell_spec[csi];
 
          if (cs->su)
-           cs->sd (means, pvd, pvd->cell_stats[i]);
+           cs->sd (pvd, pvd->cell_stats[i]);
        }
     }
 }
@@ -963,7 +942,7 @@ calculate_n (const void *aux1, void *aux2, void *user_data)
 
 static void
 run_means (struct means *cmd, struct casereader *input,
-          const struct dataset *ds)
+          const struct dataset *ds UNUSED)
 {
   int i,t;
   const struct variable *wv = dict_get_weight (cmd->dict);
@@ -1225,7 +1204,7 @@ output_report (const struct means *cmd,  int iact_idx,
              const int csi = cmd->cells[i];
              const struct cell_spec *cs = &cell_spec[csi];
 
-             double result = cs->sd (cmd, pvd, pvd->cell_stats[i]);
+             double result = cs->sd (pvd, pvd->cell_stats[i]);
 
              tab_double (t, heading_columns + i,
                          heading_rows + grp + dv * n_cats,