Added some error checking to casereader_create_append_rank

Thanks to Ben Pfaff for this suggestion.

diff --git a/src/data/casereader-translator.c b/src/data/casereader-translator.c
index 14fcd9ab70b6e3be60df6a585cfda094e48c844a..ae22f1297e19dbe5a12e6d9e2ec49f75ce8de087 100644 (file)
--- a/src/data/casereader-translator.c
+++ b/src/data/casereader-translator.c
@@ -17,7 +17,6 @@
 #include <config.h>
 #include <data/val-type.h>
 #include <data/casereader.h>
-#include <assert.h>
 #include <stdlib.h>
 
 #include <data/casereader-provider.h>
@@ -131,7 +130,7 @@ static void can_translate (struct ccase *input, struct ccase *output,
    by reading from SUBREADER and appending an additional value,
    generated by FUNC.  AUX is an optional parameter which
    gets passed to FUNC. FUNC will also receive N as it, which is
-   the ordinal number of the case in the reader.  DESTROY is an 
+   the ordinal number of the case in the reader.  DESTROY is an
    optional parameter used to destroy AUX.
 
    After this function is called, SUBREADER must not ever again
@@ -226,6 +225,8 @@ struct casereader_append_rank
   double cc;
   distinct_func *distinct;
   void *aux;
+  enum rank_error *err;
+  double prev_value;
 };
 
 static bool car_destroy (void *car_);
@@ -235,9 +236,19 @@ static void car_translate (struct ccase *input, struct ccase *output,
 
 /* Creates and returns a new casereader whose cases are produced
    by reading from SUBREADER and appending an additional value,
-   which is the rank of the observation.  SUBREADER must be sorted
-   on V.  W is the weight variable of the dictionary containing V,
-   or NULL if there is no weight variable.
+   which is the rank of the observation.   W is the weight variable
+   of the dictionary containing V, or NULL if there is no weight
+   variable.
+
+   The following preconditions must be met:
+
+   1.    SUBREADER must be sorted on V.
+
+   2.    The weight variables, must be non-negative.
+
+   If either of these preconditions are not satisfied, then the rank
+   variables may not be correct.  In this case, if ERR is non-null,
+   it will be set according to the erroneous conditions encountered.
 
    If DISTINCT_CALLBACK is non-null, then  it will be called exactly
    once for every case containing a distinct value of V.  AUX is
@@ -250,6 +261,7 @@ struct casereader *
 casereader_create_append_rank (struct casereader *subreader,
                               const struct variable *v,
                               const struct variable *w,
+                              enum rank_error *err,
                               distinct_func *distinct_callback,
                               void *aux
                               )
@@ -264,6 +276,8 @@ casereader_create_append_rank (struct casereader *subreader,
   car->clone = casereader_clone (subreader);
   car->distinct = distinct_callback;
   car->aux = aux;
+  car->err = err;
+  car->prev_value = SYSMIS;
 
   return casereader_create_translator (subreader, car->value_ofs + 1,
                                        car_translate, car_destroy, car);
@@ -285,7 +299,13 @@ car_translate (struct ccase *input, struct ccase *output,  void *car_)
 {
   struct casereader_append_rank *car = car_;
 
-  double value = case_data (input, car->var)->f;
+  const double value = case_data (input, car->var)->f;
+
+  if ( car->prev_value != SYSMIS)
+    {
+      if (car->err && value < car->prev_value)
+       *car->err |= RANK_ERR_UNSORTED;
+    }
 
   if ( car->n_common == 1)
     {
@@ -293,7 +313,11 @@ car_translate (struct ccase *input, struct ccase *output,  void *car_)
       casenumber k = 0;
       double weight = 1.0;
       if (car->weight)
-       weight = case_data (input, car->weight)->f;
+       {
+         weight = case_data (input, car->weight)->f;
+         if ( car->err && weight < 0 )
+           *car->err |= RANK_ERR_NEGATIVE_WEIGHT;
+       }
 
       do
        {
@@ -305,7 +329,14 @@ car_translate (struct ccase *input, struct ccase *output,  void *car_)
          if ( vxx == value)
            {
              if (car->weight)
-               weight += case_data (&c, car->weight)->f;
+               {
+                 double w = case_data (&c, car->weight)->f;
+
+                 if ( car->err && w < 0 )
+                   *car->err |= RANK_ERR_NEGATIVE_WEIGHT;
+
+                 weight += w;
+               }
              else
                weight += 1.0;
              car->n_common++;
@@ -328,6 +359,7 @@ car_translate (struct ccase *input, struct ccase *output,  void *car_)
   case_move (output, input);
   case_resize (output, car->value_ofs + 1);
   case_data_rw_idx (output, car->value_ofs)->f = car->mean_rank ;
+  car->prev_value = value;
 }
 
 

diff --git a/src/data/casereader.h b/src/data/casereader.h
index a2d9403bb03bce10ede5a638f459261889d373ad..3df80cb08fb8d38acc302e0654a0217dd666287a 100644 (file)
--- a/src/data/casereader.h
+++ b/src/data/casereader.h
@@ -125,12 +125,20 @@ struct casereader *
 casereader_create_arithmetic_sequence (struct casereader *,
                                        double first, double increment);
 
+enum rank_error
+  {
+    RANK_ERR_NONE = 0,
+    RANK_ERR_NEGATIVE_WEIGHT = 0x01,
+    RANK_ERR_UNSORTED = 0x02
+  };
+
 
 typedef void distinct_func (double v, casenumber n, double w, void *aux);
 
 struct casereader *
 casereader_create_append_rank (struct casereader *,
                               const struct variable *v, const struct variable *w,
+                              enum rank_error *err,
                               distinct_func *distinct_callback, void *aux);