X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;ds=sidebyside;f=src%2Fdictionary.c;h=86076f9cc761fd6e8479fdd290c6d370b735944f;hb=084280e4bfd6b94c26320e72553a9a38201610a3;hp=e03c429eabd3d206920b6b6898a7f5a37c1b9c67;hpb=7b98b3a4f58f6dc5a8e9cbc188b627966d5e652d;p=pspp-builds.git

diff --git a/src/dictionary.c b/src/dictionary.c
index e03c429e..86076f9c 100644
--- a/src/dictionary.c
+++ b/src/dictionary.c
@@ -18,10 +18,13 @@
    02111-1307, USA. */
 
 #include <config.h>
-#include <assert.h>
+#include "dictionary.h"
 #include <stdlib.h>
+#include <ctype.h>
 #include "algorithm.h"
 #include "alloc.h"
+#include "case.h"
+#include "error.h"
 #include "hash.h"
 #include "misc.h"
 #include "str.h"
@@ -34,7 +37,8 @@ struct dictionary
     struct variable **var;	/* Variables. */
     size_t var_cnt, var_cap;    /* Number of variables, capacity. */
     struct hsh_table *name_tab;	/* Variable index by name. */
-    int value_cnt;              /* Number of `union value's per case. */
+    struct hsh_table *long_name_tab; /* Variable indexed by long name */
+    int next_value_idx;         /* Index of next `union value' to allocate. */
     struct variable **split;    /* SPLIT FILE vars. */
     size_t split_cnt;           /* SPLIT FILE count. */
     struct variable *weight;    /* WEIGHT variable. */
@@ -46,6 +50,44 @@ struct dictionary
     size_t vector_cnt;          /* Number of vectors. */
   };
 
+
+
+
+
+int
+compare_long_names(const void *a_, const void *b_, void *aux UNUSED)
+{
+  const struct name_table_entry *a = a_;
+  const struct name_table_entry *b = b_;
+
+  return strcasecmp(a->longname, b->longname);
+}
+
+
+/* Long names use case insensitive comparison */
+unsigned int
+hash_long_name (const void *e_, void *aux UNUSED) 
+{
+  const struct name_table_entry *e = e_;
+  unsigned int hash;
+  int i;
+
+  char *s = strdup(e->longname);
+
+  for ( i = 0 ; i < strlen(s) ; ++i ) 
+    s[i] = toupper(s[i]);
+
+  hash = hsh_hash_string (s);
+  
+  free (s);
+
+  return hash;
+}
+
+static char *make_short_name(struct dictionary *dict, const char *longname) ;
+
+
+/* Creates and returns a new dictionary. */
 struct dictionary *
 dict_create (void) 
 {
@@ -53,8 +95,9 @@ dict_create (void)
   
   d->var = NULL;
   d->var_cnt = d->var_cap = 0;
-  d->name_tab = hsh_create (8, compare_variables, hash_variable, NULL, NULL);
-  d->value_cnt = 0;
+  d->name_tab = hsh_create (8, compare_var_names, hash_var_name, NULL, NULL);
+  d->long_name_tab = hsh_create (8, compare_long_names, hash_long_name, NULL, NULL);
+  d->next_value_idx = 0;
   d->split = NULL;
   d->split_cnt = 0;
   d->weight = NULL;
@@ -68,41 +111,36 @@ dict_create (void)
   return d;
 }
 
+/* Creates and returns a (deep) copy of an existing
+   dictionary. */
 struct dictionary *
 dict_clone (const struct dictionary *s) 
 {
   struct dictionary *d;
   size_t i;
-  
+
   assert (s != NULL);
-  
+
   d = dict_create ();
+
   for (i = 0; i < s->var_cnt; i++)
-    dict_clone_var (d, s->var[i], s->var[i]->name);
-  d->value_cnt = s->value_cnt;
+    dict_clone_var (d, s->var[i], s->var[i]->name, s->var[i]->longname);
+
+  d->next_value_idx = s->next_value_idx;
 
   d->split_cnt = s->split_cnt;
   if (d->split_cnt > 0) 
     {
       d->split = xmalloc (d->split_cnt * sizeof *d->split);
       for (i = 0; i < d->split_cnt; i++) 
-        {
-          d->split[i] = dict_lookup_var (d, s->split[i]->name);
-          assert (d->split[i] != NULL);
-        }
+        d->split[i] = dict_lookup_var_assert (d, s->split[i]->name);
     }
 
   if (s->weight != NULL) 
-    {
-      d->weight = dict_lookup_var (d, s->weight->name);
-      assert (d->weight != NULL); 
-    }
+    d->weight = dict_lookup_var_assert (d, s->weight->name);
 
   if (s->filter != NULL) 
-    {
-      d->filter = dict_lookup_var (d, s->filter->name);
-      assert (d->filter != NULL);
-    }
+    d->filter = dict_lookup_var_assert (d, s->filter->name);
 
   d->case_limit = s->case_limit;
   dict_set_label (d, dict_get_label (s));
@@ -115,6 +153,8 @@ dict_clone (const struct dictionary *s)
   return d;
 }
 
+/* Clears the contents from a dictionary without destroying the
+   dictionary itself. */
 void
 dict_clear (struct dictionary *d) 
 {
@@ -127,6 +167,7 @@ dict_clear (struct dictionary *d)
   for (i = 0; i < d->var_cnt; i++) 
     {
       struct variable *v = d->var[i];
+      var_clear_aux (v);
       val_labs_destroy (v->val_labs);
       free (v->label);
       free (v); 
@@ -135,7 +176,9 @@ dict_clear (struct dictionary *d)
   d->var = NULL;
   d->var_cnt = d->var_cap = 0;
   hsh_clear (d->name_tab);
-  d->value_cnt = 0;
+  if ( d->long_name_tab) 
+    hsh_clear (d->long_name_tab);
+  d->next_value_idx = 0;
   free (d->split);
   d->split = NULL;
   d->split_cnt = 0;
@@ -149,6 +192,92 @@ dict_clear (struct dictionary *d)
   dict_clear_vectors (d);
 }
 
+/* Allocate the pointer TEXT and fill it with text representing the
+   long variable name buffer.  SIZE will contain the size of TEXT.
+   TEXT must be freed by the caller when no longer required.
+*/
+void
+dict_get_varname_block(const struct dictionary *dict, char **text, int *size)
+{
+  char *buf = 0;
+  int bufsize = 0;
+  struct hsh_iterator hi;
+  struct name_table_entry *nte;
+  short first = 1;
+
+  for ( nte = hsh_first(dict->long_name_tab, &hi);
+	nte; 
+	nte = hsh_next(dict->long_name_tab, &hi))
+    {
+      bufsize += strlen(nte->name) + strlen(nte->longname) + 2;
+      buf = xrealloc(buf, bufsize + 1);
+      if ( first ) 
+	strcpy(buf, "");
+      first = 0;
+
+      strcat(buf, nte->name);
+      strcat(buf, "=");
+      strcat(buf, nte->longname);
+      strcat(buf, "\t");
+    }
+
+  if ( bufsize > 0 ) 
+    {
+      /* Loose the final delimiting TAB */
+      buf[bufsize]='\0';
+      bufsize--;
+    }
+  
+  *text = buf;
+  *size = bufsize;
+}
+
+
+/* Add a new entry into the dictionary's long name table, and update the 
+   corresponding varible with the relevant long name.
+*/
+void
+dict_add_longvar_entry(struct dictionary *d, 
+		       const char *name, 
+		       const char *longname)
+{
+  struct variable *v;
+  assert ( name ) ;
+  assert ( longname );
+  struct name_table_entry *nte = xmalloc (sizeof (struct name_table_entry));
+  nte->longname = strdup(longname);
+  nte->name = strdup(name);
+
+
+  /* Look up the name in name_tab */
+  v = hsh_find ( d->name_tab, name);
+  if ( !v ) 
+    {
+      msg (FE, _("The entry \"%s\" in the variable name map, has no corresponding variable"), name);
+      return ;
+    }
+  assert ( 0 == strcmp(v->name, name) );
+  v->longname = nte->longname;
+
+  hsh_insert(d->long_name_tab, nte);
+  
+
+}
+
+/* Destroys the aux data for every variable in D, by calling
+   var_clear_aux() for each variable. */
+void
+dict_clear_aux (struct dictionary *d) 
+{
+  int i;
+  
+  assert (d != NULL);
+  
+  for (i = 0; i < d->var_cnt; i++)
+    var_clear_aux (d->var[i]);
+}
+
+/* Clears a dictionary and destroys it. */
 void
 dict_destroy (struct dictionary *d)
 {
@@ -156,10 +285,12 @@ dict_destroy (struct dictionary *d)
     {
       dict_clear (d);
       hsh_destroy (d->name_tab);
+      hsh_destroy (d->long_name_tab);
       free (d);
     }
 }
 
+/* Returns the number of variables in D. */
 size_t
 dict_get_var_cnt (const struct dictionary *d) 
 {
@@ -168,6 +299,9 @@ dict_get_var_cnt (const struct dictionary *d)
   return d->var_cnt;
 }
 
+/* Returns the variable in D with index IDX, which must be
+   between 0 and the count returned by dict_get_var_cnt(),
+   exclusive. */
 struct variable *
 dict_get_var (const struct dictionary *d, size_t idx) 
 {
@@ -177,6 +311,10 @@ dict_get_var (const struct dictionary *d, size_t idx)
   return d->var[idx];
 }
 
+/* Sets *VARS to an array of pointers to variables in D and *CNT
+   to the number of variables in *D.  By default all variables
+   are returned, but bits may be set in EXCLUDE_CLASSES to
+   exclude ordinary, system, and/or scratch variables. */
 void
 dict_get_vars (const struct dictionary *d, struct variable ***vars,
                size_t *cnt, unsigned exclude_classes)
@@ -204,15 +342,59 @@ dict_get_vars (const struct dictionary *d, struct variable ***vars,
   assert (*cnt == count);
 }
 
+
+static struct variable * dict_create_var_x (struct dictionary *d, 
+					    const char *name, int width, 
+					    short name_is_short) ;
+
+/* Creates and returns a new variable in D with the given LONGNAME
+   and WIDTH.  Returns a null pointer if the given LONGNAME would
+   duplicate that of an existing variable in the dictionary.
+*/
+struct variable *
+dict_create_var (struct dictionary *d, const char *longname, int width)
+{
+  return dict_create_var_x(d, longname, width, 0);
+}
+
+
+/* Creates and returns a new variable in D with the given SHORTNAME and 
+   WIDTH.  The long name table is not updated */
 struct variable *
-dict_create_var (struct dictionary *d, const char *name, int width) 
+dict_create_var_from_short (struct dictionary *d, const char *shortname, 
+			    int width)
+{
+  return dict_create_var_x(d, shortname, width, 1);
+}
+
+
+
+/* Creates and returns a new variable in D with the given NAME
+   and WIDTH.  
+   If NAME_IS_SHORT, assume NAME is the short name.  Otherwise assumes
+   NAME is the long name, and creates the corresponding entry in the 
+   Dictionary's lookup name table .
+   Returns a null pointer if the given NAME would
+   duplicate that of an existing variable in the dictionary. 
+   
+*/
+static struct variable *
+dict_create_var_x (struct dictionary *d, const char *name, int width, 
+		 short name_is_short) 
 {
   struct variable *v;
 
   assert (d != NULL);
   assert (name != NULL);
-  assert (strlen (name) >= 1 && strlen (name) <= 8);
+
+  assert (strlen (name) >= 1);
+
   assert (width >= 0 && width < 256);
+    
+  if ( name_is_short ) 
+    assert(strlen (name) <= SHORT_NAME_LEN);
+  else
+    assert(strlen (name) <= LONG_NAME_LEN);
 
   /* Make sure there's not already a variable by that name. */
   if (dict_lookup_var (d, name) != NULL)
@@ -220,29 +402,49 @@ dict_create_var (struct dictionary *d, const char *name, int width)
 
   /* Allocate and initialize variable. */
   v = xmalloc (sizeof *v);
-  strncpy (v->name, name, sizeof v->name);
-  v->name[8] = '\0';
+
+  if ( name_is_short )
+    {
+      strncpy (v->name, name, sizeof v->name);
+      v->name[SHORT_NAME_LEN] = '\0';
+    }
+  else
+    strcpy(v->name,make_short_name(d, name));
+  
+
   v->index = d->var_cnt;
   v->type = width == 0 ? NUMERIC : ALPHA;
   v->width = width;
-  v->fv = d->value_cnt;
+  v->fv = d->next_value_idx;
   v->nv = width == 0 ? 1 : DIV_RND_UP (width, 8);
-  v->left = name[0] == '#';
+  v->init = 1;
+  v->reinit = dict_class_from_id (v->name) != DC_SCRATCH;
   v->miss_type = MISSING_NONE;
   if (v->type == NUMERIC)
     {
       v->print.type = FMT_F;
       v->print.w = 8;
       v->print.d = 2;
+
+      v->alignment = ALIGN_RIGHT;
+      v->display_width = 8;
+      v->measure = MEASURE_SCALE;
     }
   else
     {
       v->print.type = FMT_A;
       v->print.w = v->width;
       v->print.d = 0;
+
+      v->alignment = ALIGN_LEFT;
+      v->display_width = 8;
+      v->measure = MEASURE_NOMINAL;
     }
+  v->write = v->print;
   v->val_labs = val_labs_create (v->width);
   v->label = NULL;
+  v->aux = NULL;
+  v->aux_dtor = NULL;
 
   /* Update dictionary. */
   if (d->var_cnt >= d->var_cap) 
@@ -253,27 +455,64 @@ dict_create_var (struct dictionary *d, const char *name, int width)
   d->var[v->index] = v;
   d->var_cnt++;
   hsh_force_insert (d->name_tab, v);
-  d->value_cnt += v->nv;
 
+  if ( ! name_is_short) 
+    dict_add_longvar_entry(d, v->name, name);
+
+  d->next_value_idx += v->nv;
+
+  return v;
+}
+
+/* Creates and returns a new variable in D with the given NAME
+   and WIDTH.  Assert-fails if the given NAME would duplicate
+   that of an existing variable in the dictionary. */
+struct variable *
+dict_create_var_assert (struct dictionary *d, const char *longname, int width)
+{
+  struct variable *v = dict_create_var (d, longname, width);
+  assert (v != NULL);
   return v;
 }
 
+/* Creates a new variable in D with longname LONGNAME, as a copy of
+   existing  variable OV, which need not be in D or in any
+   dictionary. 
+   If SHORTNAME is non null, it will be used as the short name
+   otherwise a new short name will be generated.
+*/
 struct variable *
 dict_clone_var (struct dictionary *d, const struct variable *ov,
-                const char *name)
+                const char *name, const char *longname)
 {
   struct variable *nv;
 
   assert (d != NULL);
   assert (ov != NULL);
-  assert (name != NULL);
-  assert (strlen (name) >= 1 && strlen (name) <= 8);
+  assert (strlen (longname) <= LONG_NAME_LEN);
 
-  nv = dict_create_var (d, name, ov->width);
+  struct name_table_entry *nte = xmalloc (sizeof (struct name_table_entry));
+
+  nte->longname = strdup(longname);
+  if ( name ) 
+    {
+      assert (strlen (name) >= 1);
+      assert (strlen (name) <= SHORT_NAME_LEN);
+      nte->name = strdup(name);
+    }
+  else 
+    nte->name = make_short_name(d, longname);
+
+
+  nv = dict_create_var_from_short (d, nte->name, ov->width);
   if (nv == NULL)
     return NULL;
 
-  nv->left = ov->left;
+  hsh_insert(d->long_name_tab, nte);
+  nv->longname = nte->longname;
+
+  nv->init = 1;
+  nv->reinit = ov->reinit;
   nv->miss_type = ov->miss_type;
   memcpy (nv->missing, ov->missing, sizeof nv->missing);
   nv->print = ov->print;
@@ -283,9 +522,16 @@ dict_clone_var (struct dictionary *d, const struct variable *ov,
   if (ov->label != NULL)
     nv->label = xstrdup (ov->label);
 
+  nv->alignment = ov->alignment;
+  nv->measure = ov->measure;
+  nv->display_width = ov->display_width;
+
   return nv;
 }
 
+/* Changes the name of V in D to name NEW_NAME.  Assert-fails if
+   a variable named NEW_NAME is already in D, except that
+   NEW_NAME may be the same as V's existing name. */
 void 
 dict_rename_var (struct dictionary *d, struct variable *v,
                  const char *new_name) 
@@ -293,7 +539,8 @@ dict_rename_var (struct dictionary *d, struct variable *v,
   assert (d != NULL);
   assert (v != NULL);
   assert (new_name != NULL);
-  assert (strlen (new_name) >= 1 && strlen (new_name) <= 8);
+  assert (strlen (new_name) >= 1 && strlen (new_name) <= SHORT_NAME_LEN);
+  assert (dict_contains_var (d, v));
 
   if (!strcmp (v->name, new_name))
     return;
@@ -302,73 +549,113 @@ dict_rename_var (struct dictionary *d, struct variable *v,
 
   hsh_force_delete (d->name_tab, v);
   strncpy (v->name, new_name, sizeof v->name);
-  v->name[8] = '\0';
+  v->name[SHORT_NAME_LEN] = '\0';
   hsh_force_insert (d->name_tab, v);
 }
 
+/* Returns the variable named NAME in D, or a null pointer if no
+   variable has that name. */
 struct variable *
 dict_lookup_var (const struct dictionary *d, const char *name)
 {
   struct variable v;
+  struct variable *vr;
   
+  char *short_name;
+  struct name_table_entry key;
+  struct name_table_entry *nte;
+
   assert (d != NULL);
   assert (name != NULL);
-  assert (strlen (name) >= 1 && strlen (name) <= 8);
+  assert (strlen (name) >= 1 && strlen (name) <= LONG_NAME_LEN);
 
-  strncpy (v.name, name, sizeof v.name);
-  v.name[8] = '\0';
+  key.longname = name;
+  nte = hsh_find (d->long_name_tab, &key);
+  
+  if ( ! nte ) 
+  {
+    return 0;
+  }
 
-  return hsh_find (d->name_tab, &v);
+  short_name = nte->name ;
+
+  strncpy (v.name, short_name, sizeof v.name);
+  v.name[SHORT_NAME_LEN] = '\0';
+
+  vr = hsh_find (d->name_tab, &v);
+
+  return vr; 
+}
+
+
+/* Returns the variable named NAME in D.  Assert-fails if no
+   variable has that name. */
+struct variable *
+dict_lookup_var_assert (const struct dictionary *d, const char *name)
+{
+  struct variable *v = dict_lookup_var (d, name);
+  assert (v != NULL);
+  return v;
 }
 
+/* Returns nonzero if variable V is in dictionary D. */
 int
 dict_contains_var (const struct dictionary *d, const struct variable *v)
 {
   assert (d != NULL);
   assert (v != NULL);
 
-  return dict_lookup_var (d, v->name) == v;
+  return v->index >= 0 && v->index < d->var_cnt && d->var[v->index] == v;
 }
 
+/* Compares two double pointers to variables, which should point
+   to elements of a struct dictionary's `var' member array. */
 static int
-compare_variable_dblptrs (const void *a_, const void *b_, void *aux UNUSED) 
+compare_var_ptrs (const void *a_, const void *b_, void *aux UNUSED) 
 {
   struct variable *const *a = a_;
   struct variable *const *b = b_;
 
-  if (a > b)
-    return 1;
-  else if (a < b)
-    return -1;
-  else
-    return 0;
+  return *a < *b ? -1 : *a > *b;
 }
 
+/* Deletes variable V from dictionary D and frees V.
+
+   This is a very bad idea if there might be any pointers to V
+   from outside D.  In general, no variable in default_dict
+   should be deleted when any transformations are active, because
+   those transformations might reference the deleted variable.
+   The safest time to delete a variable is just after a procedure
+   has been executed, as done by MODIFY VARS.
+
+   Pointers to V within D are not a problem, because
+   dict_delete_var() knows to remove V from split variables,
+   weights, filters, etc. */
 void
 dict_delete_var (struct dictionary *d, struct variable *v) 
 {
   size_t i;
 
-  /* FIXME?  Does not sync d->value_cnt. */
   assert (d != NULL);
   assert (v != NULL);
   assert (dict_contains_var (d, v));
   assert (d->var[v->index] == v);
 
-  /* Remove v from splits, weight, filter variables. */
+  /* Delete aux data. */
+  var_clear_aux (v);
+
+  /* Remove V from splits, weight, filter variables. */
   d->split_cnt = remove_equal (d->split, d->split_cnt, sizeof *d->split,
-                               &v,
-                               compare_variable_dblptrs, NULL);
+                               &v, compare_var_ptrs, NULL);
   if (d->weight == v)
     d->weight = NULL;
   if (d->filter == v)
     d->filter = NULL;
   dict_clear_vectors (d);
 
-  /* Remove v from var array. */
+  /* Remove V from var array. */
+  remove_element (d->var, d->var_cnt, sizeof *d->var, v->index);
   d->var_cnt--;
-  memmove (d->var + v->index, d->var + v->index + 1,
-           (d->var_cnt - v->index) * sizeof *d->var);
 
   /* Update index. */
   for (i = v->index; i < d->var_cnt; i++)
@@ -381,9 +668,10 @@ dict_delete_var (struct dictionary *d, struct variable *v)
   val_labs_destroy (v->val_labs);
   free (v->label);
   free (v);
-
 }
 
+/* Deletes the COUNT variables listed in VARS from D.  This is
+   unsafe; see the comment on dict_delete_var() for details. */
 void 
 dict_delete_vars (struct dictionary *d,
                   struct variable *const *vars, size_t count) 
@@ -397,6 +685,10 @@ dict_delete_vars (struct dictionary *d,
     dict_delete_var (d, *vars++);
 }
 
+/* Reorders the variables in D, placing the COUNT variables
+   listed in ORDER in that order at the beginning of D.  The
+   other variables in D, if any, retain their relative
+   positions. */
 void 
 dict_reorder_vars (struct dictionary *d,
                    struct variable *const *order, size_t count) 
@@ -428,6 +720,12 @@ dict_reorder_vars (struct dictionary *d,
   d->var = new_var;
 }
 
+/* Renames COUNT variables specified in VARS to the names given
+   in NEW_NAMES within dictionary D.  If the renaming would
+   result in a duplicate variable name, returns zero and stores a
+   name that would be duplicated into *ERR_NAME (if ERR_NAME is
+   non-null).  Otherwise, the renaming is successful, and nonzero
+   is returned. */
 int
 dict_rename_vars (struct dictionary *d,
                   struct variable **vars, char **new_names,
@@ -437,10 +735,12 @@ dict_rename_vars (struct dictionary *d,
   size_t i;
   int success = 1;
 
+
   assert (d != NULL);
   assert (count == 0 || vars != NULL);
   assert (count == 0 || new_names != NULL);
 
+
   old_names = xmalloc (count * sizeof *old_names);
   for (i = 0; i < count; i++) 
     {
@@ -451,10 +751,22 @@ dict_rename_vars (struct dictionary *d,
   
   for (i = 0; i < count; i++)
     {
+      struct name_table_entry key;
+      struct name_table_entry *nte;
       assert (new_names[i] != NULL);
-      assert (strlen (new_names[i]) < 9);
-      strcpy (vars[i]->name, new_names[i]);
-      if (hsh_insert (d->name_tab, vars[i]) != NULL) 
+      assert (*new_names[i] != '\0');
+      assert (strlen (new_names[i]) <= LONG_NAME_LEN );
+      strcpy (vars[i]->name, make_short_name(d, new_names[i]));
+
+
+      key.longname = vars[i]->longname;
+      nte = hsh_find (d->long_name_tab, &key);
+      
+      free( nte->longname ) ;
+      nte->longname = strdup ( new_names[i]);
+      vars[i]->longname = nte->longname;
+
+      if (hsh_insert (d->name_tab, vars[i]) != NULL )
         {
           size_t fail_idx = i;
           if (err_name != NULL) 
@@ -467,6 +779,14 @@ dict_rename_vars (struct dictionary *d,
             {
               strcpy (vars[i]->name, old_names[i]);
               hsh_force_insert (d->name_tab, vars[i]);
+ 
+      key.longname = vars[i]->longname;
+      nte = hsh_find (d->long_name_tab, &key);
+      
+      free( nte->longname ) ;
+      nte->longname = strdup ( old_names[i]);
+      vars[i]->longname = nte->longname;
+
             }
 
           success = 0;
@@ -481,6 +801,8 @@ dict_rename_vars (struct dictionary *d,
   return success;
 }
 
+/* Returns the weighting variable in dictionary D, or a null
+   pointer if the dictionary is unweighted. */
 struct variable *
 dict_get_weight (const struct dictionary *d) 
 {
@@ -490,8 +812,14 @@ dict_get_weight (const struct dictionary *d)
   return d->weight;
 }
 
+/* Returns the value of D's weighting variable in case C, except that a
+   negative weight is returned as 0.  Returns 1 if the dictionary is
+   unweighted. Will warn about missing, negative, or zero values if
+   warn_on_invalid is nonzero. The function will set warn_on_invalid to zero
+   if an invalid weight is found. */
 double
-dict_get_case_weight (const struct dictionary *d, const struct ccase *c)
+dict_get_case_weight (const struct dictionary *d, const struct ccase *c, 
+		      int *warn_on_invalid)
 {
   assert (d != NULL);
   assert (c != NULL);
@@ -500,13 +828,21 @@ dict_get_case_weight (const struct dictionary *d, const struct ccase *c)
     return 1.0;
   else 
     {
-      double w = c->data[d->weight->fv].f;
-      if (w < 0.0)
+      double w = case_num (c, d->weight->fv);
+      if ( w < 0.0 || w == SYSMIS || is_num_user_missing(w, d->weight) )
         w = 0.0;
+      if ( w == 0.0 && *warn_on_invalid ) {
+	  *warn_on_invalid = 0;
+	  msg (SW, _("At least one case in the data file had a weight value "
+		     "that was user-missing, system-missing, zero, or "
+		     "negative.  These case(s) were ignored."));
+      }
       return w;
     }
 }
 
+/* Sets the weighting variable of D to V, or turning off
+   weighting if V is a null pointer. */
 void
 dict_set_weight (struct dictionary *d, struct variable *v) 
 {
@@ -517,6 +853,8 @@ dict_set_weight (struct dictionary *d, struct variable *v)
   d->weight = v;
 }
 
+/* Returns the filter variable in dictionary D (see cmd_filter())
+   or a null pointer if the dictionary is unfiltered. */
 struct variable *
 dict_get_filter (const struct dictionary *d) 
 {
@@ -526,6 +864,8 @@ dict_get_filter (const struct dictionary *d)
   return d->filter;
 }
 
+/* Sets V as the filter variable for dictionary D.  Passing a
+   null pointer for V turn off filtering. */
 void
 dict_set_filter (struct dictionary *d, struct variable *v)
 {
@@ -535,6 +875,8 @@ dict_set_filter (struct dictionary *d, struct variable *v)
   d->filter = v;
 }
 
+/* Returns the case limit for dictionary D, or zero if the number
+   of cases is unlimited (see cmd_n()). */
 int
 dict_get_case_limit (const struct dictionary *d) 
 {
@@ -543,6 +885,8 @@ dict_get_case_limit (const struct dictionary *d)
   return d->case_limit;
 }
 
+/* Sets CASE_LIMIT as the case limit for dictionary D.  Zero for
+   CASE_LIMIT indicates no limit. */
 void
 dict_set_case_limit (struct dictionary *d, int case_limit) 
 {
@@ -552,29 +896,121 @@ dict_set_case_limit (struct dictionary *d, int case_limit)
   d->case_limit = case_limit;
 }
 
+/* Returns the index of the next value to be added to D.  This
+   value is the number of `union value's that need to be
+   allocated to store a case for dictionary D. */
 int
-dict_get_value_cnt (const struct dictionary *d) 
+dict_get_next_value_idx (const struct dictionary *d) 
+{
+  assert (d != NULL);
+
+  return d->next_value_idx;
+}
+
+/* Returns the number of bytes needed to store a case for
+   dictionary D. */
+size_t
+dict_get_case_size (const struct dictionary *d) 
 {
   assert (d != NULL);
 
-  return d->value_cnt;
+  return sizeof (union value) * dict_get_next_value_idx (d);
 }
 
+/* Deletes scratch variables in dictionary D and reassigns values
+   so that fragmentation is eliminated. */
 void
 dict_compact_values (struct dictionary *d) 
 {
   size_t i;
 
-  d->value_cnt = 0;
+  d->next_value_idx = 0;
+  for (i = 0; i < d->var_cnt; )
+    {
+      struct variable *v = d->var[i];
+
+      if (dict_class_from_id (v->name) != DC_SCRATCH) 
+        {
+          v->fv = d->next_value_idx;
+          d->next_value_idx += v->nv;
+          i++;
+        }
+      else
+        dict_delete_var (default_dict, v);
+    }
+}
+
+/* Copies values from SRC, which represents a case arranged
+   according to dictionary D, to DST, which represents a case
+   arranged according to the dictionary that will be produced by
+   dict_compact_values(D). */
+void
+dict_compact_case (const struct dictionary *d,
+                   struct ccase *dst, const struct ccase *src)
+{
+  size_t i;
+  size_t value_idx;
+
+  value_idx = 0;
+  for (i = 0; i < d->var_cnt; i++) 
+    {
+      struct variable *v = d->var[i];
+
+      if (dict_class_from_id (v->name) != DC_SCRATCH)
+        {
+          case_copy (dst, value_idx, src, v->fv, v->nv);
+          value_idx += v->nv;
+        }
+    }
+}
+
+/* Returns the number of values that would be used by a case if
+   dict_compact_values() were called. */
+size_t
+dict_get_compacted_value_cnt (const struct dictionary *d) 
+{
+  size_t i;
+  size_t cnt;
+
+  cnt = 0;
+  for (i = 0; i < d->var_cnt; i++)
+    if (dict_class_from_id (d->var[i]->name) != DC_SCRATCH) 
+      cnt += d->var[i]->nv;
+  return cnt;
+}
+
+/* Creates and returns an array mapping from a dictionary index
+   to the `fv' that the corresponding variable will have after
+   calling dict_compact_values().  Scratch variables receive -1
+   for `fv' because dict_compact_values() will delete them. */
+int *
+dict_get_compacted_idx_to_fv (const struct dictionary *d) 
+{
+  size_t i;
+  size_t next_value_idx;
+  int *idx_to_fv;
+  
+  idx_to_fv = xmalloc (d->var_cnt * sizeof *idx_to_fv);
+  next_value_idx = 0;
   for (i = 0; i < d->var_cnt; i++)
     {
       struct variable *v = d->var[i];
 
-      v->fv = d->value_cnt;
-      d->value_cnt += v->nv;
+      if (dict_class_from_id (v->name) != DC_SCRATCH) 
+        {
+          idx_to_fv[i] = next_value_idx;
+          next_value_idx += v->nv;
+        }
+      else 
+        idx_to_fv[i] = -1;
     }
+  return idx_to_fv;
 }
 
+/* Returns the SPLIT FILE vars (see cmd_split_file()).  Call
+   dict_get_split_cnt() to determine how many SPLIT FILE vars
+   there are.  Returns a null pointer if and only if there are no
+   SPLIT FILE vars. */
 struct variable *const *
 dict_get_split_vars (const struct dictionary *d) 
 {
@@ -583,6 +1019,7 @@ dict_get_split_vars (const struct dictionary *d)
   return d->split;
 }
 
+/* Returns the number of SPLIT FILE vars. */
 size_t
 dict_get_split_cnt (const struct dictionary *d) 
 {
@@ -591,6 +1028,7 @@ dict_get_split_cnt (const struct dictionary *d)
   return d->split_cnt;
 }
 
+/* Sets CNT split vars SPLIT in dictionary D. */
 void
 dict_set_split_vars (struct dictionary *d,
                      struct variable *const *split, size_t cnt)
@@ -603,6 +1041,8 @@ dict_set_split_vars (struct dictionary *d,
   memcpy (d->split, split, cnt * sizeof *d->split);
 }
 
+/* Returns the file label for D, or a null pointer if D is
+   unlabeled (see cmd_file_label()). */
 const char *
 dict_get_label (const struct dictionary *d) 
 {
@@ -611,6 +1051,8 @@ dict_get_label (const struct dictionary *d)
   return d->label;
 }
 
+/* Sets D's file label to LABEL, truncating it to a maximum of 60
+   characters. */
 void
 dict_set_label (struct dictionary *d, const char *label) 
 {
@@ -629,6 +1071,8 @@ dict_set_label (struct dictionary *d, const char *label)
     }
 }
 
+/* Returns the documents for D, or a null pointer if D has no
+   documents (see cmd_document()).. */
 const char *
 dict_get_documents (const struct dictionary *d) 
 {
@@ -637,6 +1081,8 @@ dict_get_documents (const struct dictionary *d)
   return d->documents;
 }
 
+/* Sets the documents for D to DOCUMENTS, or removes D's
+   documents if DOCUMENT is a null pointer. */
 void
 dict_set_documents (struct dictionary *d, const char *documents)
 {
@@ -649,6 +1095,9 @@ dict_set_documents (struct dictionary *d, const char *documents)
     d->documents = xstrdup (documents);
 }
 
+/* Creates in D a vector named NAME that contains CNT variables
+   VAR (see cmd_vector()).  Returns nonzero if successful, or
+   zero if a vector named NAME already exists in D. */
 int
 dict_create_vector (struct dictionary *d,
                     const char *name,
@@ -658,7 +1107,7 @@ dict_create_vector (struct dictionary *d,
 
   assert (d != NULL);
   assert (name != NULL);
-  assert (strlen (name) > 0 && strlen (name) < 9);
+  assert (strlen (name) > 0 && strlen (name) <= SHORT_NAME_LEN );
   assert (var != NULL);
   assert (cnt > 0);
   
@@ -668,8 +1117,8 @@ dict_create_vector (struct dictionary *d,
   d->vector = xrealloc (d->vector, (d->vector_cnt + 1) * sizeof *d->vector);
   vector = d->vector[d->vector_cnt] = xmalloc (sizeof *vector);
   vector->idx = d->vector_cnt++;
-  strncpy (vector->name, name, 8);
-  vector->name[8] = '\0';
+  strncpy (vector->name, name, SHORT_NAME_LEN);
+  vector->name[SHORT_NAME_LEN] = '\0';
   vector->var = xmalloc (cnt * sizeof *var);
   memcpy (vector->var, var, cnt * sizeof *var);
   vector->cnt = cnt;
@@ -677,6 +1126,8 @@ dict_create_vector (struct dictionary *d,
   return 1;
 }
 
+/* Returns the vector in D with index IDX, which must be less
+   than dict_get_vector_cnt (D). */
 const struct vector *
 dict_get_vector (const struct dictionary *d, size_t idx) 
 {
@@ -686,6 +1137,7 @@ dict_get_vector (const struct dictionary *d, size_t idx)
   return d->vector[idx];
 }
 
+/* Returns the number of vectors in D. */
 size_t
 dict_get_vector_cnt (const struct dictionary *d) 
 {
@@ -694,6 +1146,8 @@ dict_get_vector_cnt (const struct dictionary *d)
   return d->vector_cnt;
 }
 
+/* Looks up and returns the vector within D with the given
+   NAME. */
 const struct vector *
 dict_lookup_vector (const struct dictionary *d, const char *name) 
 {
@@ -708,6 +1162,7 @@ dict_lookup_vector (const struct dictionary *d, const char *name)
   return NULL;
 }
 
+/* Deletes all vectors from D. */
 void
 dict_clear_vectors (struct dictionary *d) 
 {
@@ -724,3 +1179,82 @@ dict_clear_vectors (struct dictionary *d)
   d->vector = NULL;
   d->vector_cnt = 0;
 }
+
+
+static const char * quasi_base27(int i);
+
+
+/* Convert I to quasi base 27 
+   The result is a staticly allocated string.
+*/
+static const char *
+quasi_base27(int i)
+{
+  static char result[SHORT_NAME_LEN + 1];
+  static char reverse[SHORT_NAME_LEN + 1];
+
+  /* FIXME: check the result cant overflow these arrays */
+
+  char *s = result ;
+  const int radix = 27;
+  int units;
+
+  /* and here's the quasi-ness of this routine */
+  i = i + ( i / radix );
+
+  strcpy(result,"");
+  do {
+    units = i % radix;
+    *s++ = (units > 0 ) ? units + 'A' - 1 : 'A';
+    i = i / radix; 
+  } while (i > 0 ) ;
+  *s = '\0';
+
+  /* Reverse the result */
+  i = strlen(result);
+  s = reverse;
+  while(i >= 0)
+	*s++ = result[--i];
+  *s = '\0';
+	
+  return reverse;
+}
+
+
+/* Generate a short name, given a long name.
+   The return value of this function must be freed by the caller. 
+*/
+static char *
+make_short_name(struct dictionary *dict, const char *longname)
+{
+  int i = 0;
+  char *p;
+ 
+
+  char *d = xmalloc ( SHORT_NAME_LEN + 1);
+
+  /* Truncate the name */
+  strncpy(d, longname, SHORT_NAME_LEN);
+  d[SHORT_NAME_LEN] = '\0';
+
+  /* Convert to upper case */
+  for ( p = d; *p ; ++p )
+	*p = toupper(*p);
+
+  /* If a variable with that name already exists, then munge it
+     until there's no conflict */
+  while (0 != hsh_find (dict->name_tab, d))
+  {
+	const char *suffix = quasi_base27(i++);
+
+        d[SHORT_NAME_LEN -  strlen(suffix) - 1 ] = '_';
+        d[SHORT_NAME_LEN -  strlen(suffix)  ] = '\0';
+	strcat(d, suffix);
+  }
+
+  return d;
+}
+
+
+
+