X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fvariable.c;h=707b2813eccb172bdbee017ff92f2bea009ac8f2;hb=46e213d759e32e607cc2c04c6feb75809572d0cf;hp=0b8079ca534762c2a776739c6001b8327be2d335;hpb=c41f14854e73ad44824b54933ae96eb52f781fc2;p=pspp

diff --git a/src/data/variable.c b/src/data/variable.c
index 0b8079ca53..707b2813ec 100644
--- a/src/data/variable.c
+++ b/src/data/variable.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 1997-9, 2000, 2006 Free Software Foundation, Inc.
+   Copyright (C) 1997-9, 2000, 2006, 2009, 2010 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
@@ -19,22 +19,23 @@
 
 #include <stdlib.h>
 
-#include "category.h"
-#include "data-out.h"
-#include "format.h"
-#include "dictionary.h"
-#include "identifier.h"
-#include "missing-values.h"
-#include "value-labels.h"
-#include "vardict.h"
+#include <data/attributes.h>
+#include <data/data-out.h>
+#include <data/format.h>
+#include <data/dictionary.h>
+#include <data/identifier.h>
+#include <data/missing-values.h>
+#include <data/value-labels.h>
+#include <data/vardict.h>
 
 #include <libpspp/misc.h>
 #include <libpspp/assertion.h>
 #include <libpspp/compiler.h>
-#include <libpspp/hash.h>
+#include <libpspp/hash-functions.h>
 #include <libpspp/message.h>
 #include <libpspp/str.h>
 
+#include "minmax.h"
 #include "xalloc.h"
 
 #include "gettext.h"
@@ -61,7 +62,7 @@ struct variable
     bool leave;                 /* Leave value from case to case? */
 
     /* Data for use by containing dictionary. */
-    struct vardict_info vardict;
+    struct vardict_info *vardict;
 
     /* Used only for system and portable file input and output.
        See short-names.h. */
@@ -72,10 +73,8 @@ struct variable
     void *aux;
     void (*aux_dtor) (struct variable *);
 
-    /* Values of a categorical variable.  Procedures need
-       vectors with binary entries, so any variable of type ALPHA will
-       have its values stored here. */
-    struct cat_vals *obs_vals;
+    /* Custom attributes. */
+    struct attrset attributes;
   };
 
 /* Creates and returns a new variable with the given NAME and
@@ -86,36 +85,28 @@ struct variable *
 var_create (const char *name, int width)
 {
   struct variable *v;
+  enum val_type type;
 
   assert (width >= 0 && width <= MAX_STRING);
 
   v = xmalloc (sizeof *v);
-  v->vardict.dict_index = v->vardict.case_index = -1;
+  v->vardict = NULL;
   var_set_name (v, name);
   v->width = width;
   mv_init (&v->miss, width);
   v->leave = var_must_leave (v);
-  if (var_is_numeric (v))
-    {
-      v->print = fmt_for_output (FMT_F, 8, 2);
-      v->alignment = ALIGN_RIGHT;
-      v->measure = MEASURE_SCALE;
-    }
-  else
-    {
-      v->print = fmt_for_output (FMT_A, var_get_width (v), 0);
-      v->alignment = ALIGN_LEFT;
-      v->measure = MEASURE_NOMINAL;
-    }
+  type = val_type_from_width (width);
+  v->alignment = var_default_alignment (type);
+  v->measure = var_default_measure (type);
   v->display_width = var_default_display_width (width);
-  v->write = v->print;
+  v->print = v->write = var_default_formats (width);
   v->val_labs = NULL;
   v->label = NULL;
   v->short_names = NULL;
   v->short_name_cnt = 0;
   v->aux = NULL;
   v->aux_dtor = NULL;
-  v->obs_vals = NULL;
+  attrset_init (&v->attributes);
 
   return v;
 }
@@ -129,8 +120,7 @@ var_create (const char *name, int width)
 
     - The new variable is not added to OLD_VAR's dictionary by
       default.  Use dict_clone_var, instead, to do that.
-
-    - Auxiliary data and obs_vals are not copied. */
+*/
 struct variable *
 var_clone (const struct variable *old_var)
 {
@@ -146,6 +136,7 @@ var_clone (const struct variable *old_var)
   var_set_display_width (new_var, var_get_display_width (old_var));
   var_set_alignment (new_var, var_get_alignment (old_var));
   var_set_leave (new_var, var_get_leave (old_var));
+  var_set_attributes (new_var, var_get_attributes (old_var));
 
   return new_var;
 }
@@ -159,7 +150,7 @@ var_destroy (struct variable *v)
   if (v != NULL)
     {
       assert (!var_has_vardict (v));
-      cat_stored_values_destroy (v->obs_vals);
+      mv_destroy (&v->miss);
       var_clear_short_names (v);
       var_clear_aux (v);
       val_labs_destroy (v->val_labs);
@@ -183,7 +174,7 @@ var_get_name (const struct variable *v)
 void
 var_set_name (struct variable *v, const char *name)
 {
-  assert (v->vardict.dict_index == -1);
+  assert (!var_has_vardict (v));
   assert (var_is_plausible_name (name, false));
 
   str_copy_trunc (v->name, sizeof v->name, name);
@@ -199,8 +190,6 @@ var_is_valid_name (const char *name, bool issue_error)
   bool plausible;
   size_t length, i;
 
-  assert (name != NULL);
-
   /* Note that strlen returns number of BYTES, not the number of
      CHARACTERS */
   length = strlen (name);
@@ -246,8 +235,6 @@ var_is_plausible_name (const char *name, bool issue_error)
 {
   size_t length;
 
-  assert (name != NULL);
-
   /* Note that strlen returns number of BYTES, not the number of
      CHARACTERS */
   length = strlen (name);
@@ -300,7 +287,7 @@ hash_var_by_name (const void *v_, const void *aux UNUSED)
 {
   const struct variable *v = v_;
 
-  return hsh_hash_case_string (v->name);
+  return hash_case_string (v->name, 0);
 }
 
 /* A hsh_compare_func that orders pointers to variables A and B
@@ -315,6 +302,20 @@ compare_var_ptrs_by_name (const void *a_, const void *b_,
   return strcasecmp (var_get_name (*a), var_get_name (*b));
 }
 
+/* A hsh_compare_func that orders pointers to variables A and B
+   by their dictionary indexes. */
+int
+compare_var_ptrs_by_dict_index (const void *a_, const void *b_,
+                                const void *aux UNUSED)
+{
+  struct variable *const *a = a_;
+  struct variable *const *b = b_;
+  size_t a_index = var_get_dict_index (*a);
+  size_t b_index = var_get_dict_index (*b);
+
+  return a_index < b_index ? -1 : a_index > b_index;
+}
+
 /* A hsh_hash_func that hashes pointer to variable V based on its
    name. */
 unsigned
@@ -322,7 +323,7 @@ hash_var_ptr_by_name (const void *v_, const void *aux UNUSED)
 {
   struct variable *const *v = v_;
 
-  return hsh_hash_case_string (var_get_name (*v));
+  return hash_case_string (var_get_name (*v), 0);
 }
 
 /* Returns the type of variable V. */
@@ -346,10 +347,16 @@ var_set_width (struct variable *v, int new_width)
 {
   const int old_width = v->width;
 
+  if (old_width == new_width)
+    return;
+
   if (mv_is_resizable (&v->miss, new_width))
     mv_resize (&v->miss, new_width);
   else
-    mv_init (&v->miss, new_width);
+    {
+      mv_destroy (&v->miss);
+      mv_init (&v->miss, new_width);
+    }
 
   if (v->val_labs != NULL)
     {
@@ -366,15 +373,7 @@ var_set_width (struct variable *v, int new_width)
   fmt_resize (&v->write, new_width);
 
   v->width = new_width;
-
-  {
-    const int old_val_count = value_cnt_from_width (old_width);
-    const int new_val_count = value_cnt_from_width (new_width);
-
-    if ( old_val_count != new_val_count)
-	 dict_var_resized (v, new_val_count - old_val_count);
-  }
-
+  dict_var_resized (v, old_width);
   dict_var_changed (v);
 }
 
@@ -392,30 +391,6 @@ var_is_alpha (const struct variable *v)
 {
   return var_get_type (v) == VAL_STRING;
 }
-
-/* Returns true if variable V is a short string variable, false
-   otherwise. */
-bool
-var_is_short_string (const struct variable *v)
-{
-  return v->width > 0 && v->width <= MAX_SHORT_STRING;
-}
-
-/* Returns true if variable V is a long string variable, false
-   otherwise. */
-bool
-var_is_long_string (const struct variable *v)
-{
-  return v->width > MAX_SHORT_STRING;
-}
-
-/* Returns the number of "union value"s need to store a value of
-   variable V. */
-size_t
-var_get_value_cnt (const struct variable *v)
-{
-  return value_cnt_from_width (v->width);
-}
 
 /* Returns variable V's missing values. */
 const struct missing_values *
@@ -434,11 +409,12 @@ var_set_missing_values (struct variable *v, const struct missing_values *miss)
   if (miss != NULL)
     {
       assert (mv_is_resizable (miss, v->width));
+      mv_destroy (&v->miss);
       mv_copy (&v->miss, miss);
       mv_resize (&v->miss, v->width);
     }
   else
-    mv_init (&v->miss, v->width);
+    mv_clear (&v->miss);
 
   dict_var_changed (v);
 }
@@ -480,7 +456,7 @@ var_is_num_missing (const struct variable *v, double d, enum mv_class class)
    S[] must contain exactly as many characters as V's width.
    V must be a string variable. */
 bool
-var_is_str_missing (const struct variable *v, const char s[],
+var_is_str_missing (const struct variable *v, const uint8_t s[],
                     enum mv_class class)
 {
   return mv_is_str_missing (&v->miss, s, class);
@@ -525,31 +501,30 @@ var_set_value_labels (struct variable *v, const struct val_labs *vls)
 static void
 alloc_value_labels (struct variable *v)
 {
-  assert (!var_is_long_string (v));
   if (v->val_labs == NULL)
     v->val_labs = val_labs_create (v->width);
 }
 
 /* Attempts to add a value label with the given VALUE and LABEL
-   to V.  Returns true if successful, false if VALUE has an
-   existing label or if V is a long string variable. */
+   to V.  Returns true if successful, false otherwise (probably
+   due to an existing label). */
 bool
 var_add_value_label (struct variable *v,
                      const union value *value, const char *label)
 {
   alloc_value_labels (v);
-  return val_labs_add (v->val_labs, *value, label);
+  return val_labs_add (v->val_labs, value, label);
 }
 
 /* Adds or replaces a value label with the given VALUE and LABEL
    to V.
-   Has no effect if V is a long string variable. */
+*/
 void
 var_replace_value_label (struct variable *v,
                          const union value *value, const char *label)
 {
   alloc_value_labels (v);
-  val_labs_replace (v->val_labs, *value, label);
+  val_labs_replace (v->val_labs, value, label);
 }
 
 /* Removes V's value labels, if any. */
@@ -564,7 +539,7 @@ var_clear_value_labels (struct variable *v)
 const char *
 var_lookup_value_label (const struct variable *v, const union value *value)
 {
-  return val_labs_find (v->val_labs, *value);
+  return val_labs_find (v->val_labs, value);
 }
 
 /* Append STR with a string representing VALUE for variable V.
@@ -579,14 +554,14 @@ var_append_value_name (const struct variable *v, const union value *value,
   const char *name = var_lookup_value_label (v, value);
   if (name == NULL)
     {
-      char *s = ds_put_uninit (str, v->print.w);
-      data_out (value, &v->print, s);
+      char *s = data_out (value, var_get_encoding (v), &v->print);
+      ds_put_cstr (str, s);
+      free (s);
     }
   else
     ds_put_cstr (str, name);
 }
 
-
 /* Print and write formats. */
 
 /* Returns V's print format specification. */
@@ -597,8 +572,9 @@ var_get_print_format (const struct variable *v)
 }
 
 /* Sets V's print format specification to PRINT, which must be a
-   valid format specification for outputting a variable of V's
-   width. */
+   valid format specification for a variable of V's width
+   (ordinarily an output format, but input formats are not
+   rejected). */
 void
 var_set_print_format (struct variable *v, const struct fmt_spec *print)
 {
@@ -615,8 +591,9 @@ var_get_write_format (const struct variable *v)
 }
 
 /* Sets V's write format specification to WRITE, which must be a
-   valid format specification for outputting a variable of V's
-   width. */
+   valid format specification for a variable of V's width
+   (ordinarily an output format, but input formats are not
+   rejected). */
 void
 var_set_write_format (struct variable *v, const struct fmt_spec *write)
 {
@@ -626,14 +603,27 @@ var_set_write_format (struct variable *v, const struct fmt_spec *write)
 }
 
 /* Sets V's print and write format specifications to FORMAT,
-   which must be a valid format specification for outputting a
-   variable of V's width. */
+   which must be a valid format specification for a variable of
+   V's width (ordinarily an output format, but input formats are
+   not rejected). */
 void
 var_set_both_formats (struct variable *v, const struct fmt_spec *format)
 {
   var_set_print_format (v, format);
   var_set_write_format (v, format);
 }
+
+/* Returns the default print and write format for a variable of
+   the given TYPE, as set by var_create.  The return value can be
+   used to reset a variable's print and write formats to the
+   default. */
+struct fmt_spec
+var_default_formats (int width)
+{
+  return (width == 0
+          ? fmt_for_output (FMT_F, 8, 2)
+          : fmt_for_output (FMT_A, width, 0));
+}
 
 /* Return a string representing this variable, in the form most
    appropriate from a human factors perspective, that is, its
@@ -669,8 +659,8 @@ var_set_label (struct variable *v, const char *label)
       ss_truncate (&s, 255);
       if (!ss_is_empty (s))
         v->label = ss_xstrdup (s);
-      dict_var_changed (v);
     }
+  dict_var_changed (v);
 }
 
 /* Removes any variable label from V. */
@@ -711,6 +701,16 @@ var_set_measure (struct variable *v, enum measure measure)
   v->measure = measure;
   dict_var_changed (v);
 }
+
+/* Returns the default measurement level for a variable of the
+   given TYPE, as set by var_create.  The return value can be
+   used to reset a variable's measurement level to the
+   default. */
+enum measure
+var_default_measure (enum val_type type)
+{
+  return type == VAL_NUMERIC ? MEASURE_SCALE : MEASURE_NOMINAL;
+}
 
 /* Returns V's display width, which applies only to GUIs. */
 int
@@ -721,9 +721,15 @@ var_get_display_width (const struct variable *v)
 
 /* Sets V's display width to DISPLAY_WIDTH. */
 void
-var_set_display_width (struct variable *v, int display_width)
+var_set_display_width (struct variable *v, int new_width)
 {
-  v->display_width = display_width;
+  int old_width = v->display_width;
+
+  v->display_width = new_width;
+
+  if ( old_width != new_width)
+    dict_var_display_width_changed (v);
+
   dict_var_changed (v);
 }
 
@@ -759,6 +765,15 @@ var_set_alignment (struct variable *v, enum alignment alignment)
   v->alignment = alignment;
   dict_var_changed (v);
 }
+
+/* Returns the default display alignment for a variable of the
+   given TYPE, as set by var_create.  The return value can be
+   used to reset a variable's display alignment to the default. */
+enum alignment
+var_default_alignment (enum val_type type)
+{
+  return type == VAL_NUMERIC ? ALIGN_RIGHT : ALIGN_LEFT;
+}
 
 /* Whether variables' values should be preserved from case to
    case. */
@@ -821,7 +836,6 @@ var_get_short_name (const struct variable *var, size_t idx)
 void
 var_set_short_name (struct variable *var, size_t idx, const char *short_name)
 {
-  assert (var != NULL);
   assert (short_name == NULL || var_is_plausible_name (short_name, false));
 
   /* Clear old short name numbered IDX, if any. */
@@ -872,8 +886,8 @@ var_clear_short_names (struct variable *v)
 size_t
 var_get_dict_index (const struct variable *v)
 {
-  assert (v->vardict.dict_index != -1);
-  return v->vardict.dict_index;
+  assert (var_has_vardict (v));
+  return vardict_get_dict_index (v->vardict);
 }
 
 /* Returns V's index within the case represented by its
@@ -883,8 +897,8 @@ var_get_dict_index (const struct variable *v)
 size_t
 var_get_case_index (const struct variable *v)
 {
-  assert (v->vardict.case_index != -1);
-  return v->vardict.case_index;
+  assert (var_has_vardict (v));
+  return vardict_get_case_index (v->vardict);
 }
 
 /* Returns V's auxiliary data, or a null pointer if none has been
@@ -903,7 +917,7 @@ void *
 var_attach_aux (const struct variable *v_,
                 void *aux, void (*aux_dtor) (struct variable *))
 {
-  struct variable *v = (struct variable *) v_ ; /* cast away const  */
+  struct variable *v = CONST_CAST (struct variable *, v_);
   assert (v->aux == NULL);
   assert (aux != NULL);
   v->aux = aux;
@@ -927,7 +941,6 @@ var_detach_aux (struct variable *v)
 void
 var_clear_aux (struct variable *v)
 {
-  assert (v != NULL);
   if (v->aux != NULL)
     {
       if (v->aux_dtor != NULL)
@@ -945,62 +958,66 @@ var_dtor_free (struct variable *v)
   free (v->aux);
 }
 
-/* Observed categorical values. */
-
-/* Returns V's observed categorical values,
-   which V must have. */
-struct cat_vals *
-var_get_obs_vals (const struct variable *v)
+/* Returns variable V's attribute set.  The caller may examine or
+   modify the attribute set, but must not destroy it.  Destroying
+   V, or calling var_set_attributes() on V, will also destroy its
+   attribute set. */
+struct attrset *
+var_get_attributes (const struct variable *v) 
 {
-  assert (v->obs_vals != NULL);
-  return v->obs_vals;
+  return CONST_CAST (struct attrset *, &v->attributes);
 }
 
-/* Sets V's observed categorical values to CAT_VALS.
-   V becomes the owner of CAT_VALS. */
+/* Replaces variable V's attributes set by a copy of ATTRS. */
 void
-var_set_obs_vals (const struct variable *v_, struct cat_vals *cat_vals)
+var_set_attributes (struct variable *v, const struct attrset *attrs) 
 {
-  struct variable *v = (struct variable *) v_ ; /* cast away const */
-  cat_stored_values_destroy (v->obs_vals);
-  v->obs_vals = cat_vals;
+  attrset_destroy (&v->attributes);
+  attrset_clone (&v->attributes, attrs);
 }
 
-/* Returns true if V has observed categorical values,
-   false otherwise. */
+/* Returns true if V has any custom attributes, false if it has none. */
 bool
-var_has_obs_vals (const struct variable *v)
+var_has_attributes (const struct variable *v)
+{
+  return attrset_count (&v->attributes) > 0;
+}
+
+/* Returns the encoding of values of variable VAR.  (This is actually a
+   property of the dictionary.)  Returns null if no specific encoding has been
+   set.  */
+const char *
+var_get_encoding (const struct variable *var)
 {
-  return v->obs_vals != NULL;
+  return (var_has_vardict (var)
+          ? dict_get_encoding (vardict_get_dictionary (var->vardict))
+          : NULL);
 }
 
 /* Returns V's vardict structure. */
-const struct vardict_info *
+struct vardict_info *
 var_get_vardict (const struct variable *v)
 {
-  assert (var_has_vardict (v));
-  return &v->vardict;
+  return CONST_CAST (struct vardict_info *, v->vardict);
 }
 
 /* Sets V's vardict data to VARDICT. */
 void
-var_set_vardict (struct variable *v, const struct vardict_info *vardict)
+var_set_vardict (struct variable *v, struct vardict_info *vardict)
 {
-  assert (vardict->dict_index >= 0);
-  assert (vardict->case_index >= 0);
-  v->vardict = *vardict;
+  v->vardict = vardict;
 }
 
 /* Returns true if V has vardict data. */
 bool
 var_has_vardict (const struct variable *v)
 {
-  return v->vardict.dict_index != -1;
+  return v->vardict != NULL;
 }
 
 /* Clears V's vardict data. */
 void
 var_clear_vardict (struct variable *v)
 {
-  v->vardict.dict_index = v->vardict.case_index = -1;
+  v->vardict = NULL;
 }