X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fautorecode.c;h=c3ab81c93dabebf6967f9cf952e8cc22d3ce19da;hb=1a25b1d47d0f8d25830b8331f53749b887c075ce;hp=77385783549029af27c4e3ced65ba07c30e03482;hpb=691c25e36fd1ee722dd35419d6110e3876b99f9c;p=pspp

diff --git a/src/language/stats/autorecode.c b/src/language/stats/autorecode.c
index 7738578354..c3ab81c93d 100644
--- a/src/language/stats/autorecode.c
+++ b/src/language/stats/autorecode.c
@@ -20,24 +20,25 @@
 
 #include "data/case.h"
 #include "data/casereader.h"
+#include "data/dataset.h"
 #include "data/dictionary.h"
-#include "data/procedure.h"
 #include "data/transformations.h"
 #include "data/variable.h"
 #include "language/command.h"
 #include "language/lexer/lexer.h"
 #include "language/lexer/variable-parser.h"
 #include "libpspp/array.h"
-#include "libpspp/i18n.h"
 #include "libpspp/compiler.h"
 #include "libpspp/hash-functions.h"
 #include "libpspp/hmap.h"
+#include "libpspp/i18n.h"
 #include "libpspp/message.h"
 #include "libpspp/pool.h"
 #include "libpspp/str.h"
 
 #include "gl/xalloc.h"
 #include "gl/vasnprintf.h"
+#include "gl/mbiter.h"
 
 #include "gettext.h"
 #define _(msgid) gettext (msgid)
@@ -59,7 +60,7 @@ struct arc_spec
   {
     const struct variable *src;	/* Source variable. */
     struct variable *dst;	/* Target variable. */
-    struct hmap *items;         /* Hash table of "struct arc_item"s. */
+    struct rec_items *items;
   };
 
 /* Descending or ascending sort order. */
@@ -69,6 +70,14 @@ enum arc_direction
     DESCENDING
   };
 
+struct rec_items
+{
+  struct hmap ht;         /* Hash table of "struct arc_item"s. */
+  int refcnt;
+};
+
+
+
 /* AUTORECODE data. */
 struct autorecode_pgm
 {
@@ -76,7 +85,10 @@ struct autorecode_pgm
   size_t n_specs;
 
   /* Hash table of "struct arc_item"s. */
-  struct hmap *global_items;
+  struct rec_items *global_items;
+
+  bool blank_valid;
+  const struct dictionary *dict;
 };
 
 static trns_proc_func autorecode_trns_proc;
@@ -87,13 +99,35 @@ static void arc_free (struct autorecode_pgm *);
 static struct arc_item *find_arc_item (const struct arc_spec *, const union value *,
                                        size_t hash);
 
+static bool
+value_is_blank (const union value *val, int width, const struct dictionary *dict)
+{
+  mbi_iterator_t iter;
+  const char *str = CHAR_CAST_BUG (const char*, value_str (val, width));
+  char *text = recode_string (UTF8, dict_get_encoding (dict), str, width);
+
+  for (mbi_init (iter, text, width); mbi_avail (iter); mbi_advance (iter))
+    {
+      mbchar_t c = mbi_cur (iter);
+
+      if ( ! mb_isblank (c))
+	{
+	  free (text);
+	  return false;
+	}
+    }
+
+  free (text);
+  return true;
+}
+
 /* Performs the AUTORECODE procedure. */
 int
 cmd_autorecode (struct lexer *lexer, struct dataset *ds)
 {
   struct autorecode_pgm *arc = NULL;
-  struct dictionary *dict = dataset_dict (ds);
 
+  struct dictionary *dict = dataset_dict (ds);
   const struct variable **src_vars = NULL;
   char **dst_names = NULL;
   size_t n_srcs = 0;
@@ -110,17 +144,20 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
 
   /* Create procedure. */
   arc = xzalloc (sizeof *arc);
+  arc->blank_valid = true;
+  arc->dict = dataset_dict (ds);
 
   /* Parse variable lists. */
   lex_match_id (lexer, "VARIABLES");
   lex_match (lexer, T_EQUALS);
-  if (!parse_variables_const (lexer, dict, &src_vars, &n_srcs,
+  if (!parse_variables_const (lexer, arc->dict, &src_vars, &n_srcs,
                               PV_NO_DUPLICATE))
     goto error;
   if (!lex_force_match_id (lexer, "INTO"))
     goto error;
   lex_match (lexer, T_EQUALS);
-  if (!parse_DATA_LIST_vars (lexer, &dst_names, &n_dsts, PV_NO_DUPLICATE))
+  if (!parse_DATA_LIST_vars (lexer, arc->dict, &dst_names, &n_dsts,
+                             PV_NO_DUPLICATE))
     goto error;
   if (n_dsts != n_srcs)
     {
@@ -134,7 +171,7 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
     {
       const char *name = dst_names[i];
 
-      if (dict_lookup_var (dict, name) != NULL)
+      if (dict_lookup_var (arc->dict, name) != NULL)
         {
           msg (SE, _("Target variable %s duplicates existing variable %s."),
                name, name);
@@ -152,8 +189,25 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
       else if (lex_match_id (lexer, "GROUP"))
 	{
 	  arc->global_items = xmalloc (sizeof (*arc->global_items));
-	  hmap_init (arc->global_items);
+	  arc->global_items->refcnt = 1;
+	  hmap_init (&arc->global_items->ht);
+	}
+      else if (lex_match_id (lexer, "BLANK"))
+	{
+	  lex_match (lexer, T_EQUALS);
+	  if (lex_match_id (lexer, "VALID"))
+	    {
+	      arc->blank_valid = true;
+	    }
+	  else if (lex_match_id (lexer, "MISSING"))
+	    {
+	      arc->blank_valid = false;
+	    }
+	  else
+	    goto error;
 	}
+      else
+	goto error;
     }
 
   if (lex_token (lexer) != T_ENDCMD)
@@ -165,19 +219,23 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
   arc->specs = xmalloc (n_dsts * sizeof *arc->specs);
   arc->n_specs = n_dsts;
 
+
   for (i = 0; i < n_dsts; i++)
     {
       struct arc_spec *spec = &arc->specs[i];
 
       spec->src = src_vars[i];
+
       if (arc->global_items)
 	{
+	  arc->global_items->refcnt++;
 	  spec->items = arc->global_items;
 	}
       else
 	{
-	  spec->items = xmalloc (sizeof (*spec->items));
-	  hmap_init (spec->items);
+	  spec->items = xzalloc (sizeof (*spec->items));
+	  spec->items->refcnt = 1;
+	  hmap_init (&spec->items->ht);
 	}
     }
 
@@ -194,12 +252,15 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
         struct arc_item *item;
 
         item = find_arc_item (spec, value, hash);
-        if (item == NULL)
+        if ( (item == NULL)
+	     &&  
+	     ( arc->blank_valid || var_is_numeric (spec->src) || ! value_is_blank (value, width, arc->dict))
+	     )
           {
             item = xmalloc (sizeof *item);
 	    item->width = width;
             value_clone (&item->from, value, width);
-            hmap_insert (spec->items, &item->hmap_node, hash);
+            hmap_insert (&spec->items->ht, &item->hmap_node, hash);
           }
       }
   ok = casereader_destroy (input);
@@ -218,10 +279,10 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
       spec->dst = dict_create_var_assert (dict, dst_names[i], 0);
 
       /* Create array of pointers to items. */
-      n_items = hmap_count (spec->items);
+      n_items = hmap_count (&spec->items->ht);
       items = xmalloc (n_items * sizeof *items);
       j = 0;
-      HMAP_FOR_EACH (item, struct arc_item, hmap_node, spec->items)
+      HMAP_FOR_EACH (item, struct arc_item, hmap_node, &spec->items->ht)
         items[j++] = item;
 
       assert (j == n_items);
@@ -250,7 +311,7 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
 	    {
 	      const char *str = CHAR_CAST_BUG (const char*, value_str (from, src_width));
 
-	      recoded_value = recode_string (UTF8, dict_get_encoding (dict), str, src_width);
+	      recoded_value = recode_string (UTF8, dict_get_encoding (arc->dict), str, src_width);
 	    }
 	  else
 	    recoded_value = asnprintf (NULL, &len, "%g", from->f);
@@ -273,6 +334,11 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
     }
   add_transformation (ds, autorecode_trns_proc, autorecode_trns_free, arc);
 
+  for (i = 0; i < n_dsts; i++)
+    free (dst_names[i]);
+  free (dst_names);
+  free (src_vars);
+
   return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;
 
 error:
@@ -297,31 +363,31 @@ arc_free (struct autorecode_pgm *arc)
           struct arc_item *item, *next;
 
 	  HMAP_FOR_EACH_SAFE (item, next, struct arc_item, hmap_node,
-			      spec->items)
+			      &spec->items->ht)
 	    {
 	      value_destroy (&item->from, item->width);
-	      hmap_delete (spec->items, &item->hmap_node);
+	      hmap_delete (&spec->items->ht, &item->hmap_node);
 	      free (item);
 	    }
         }
 
-      if (arc->global_items)
-	{
-	  free (arc->global_items);
-	}
-      else
+      for (i = 0; i < arc->n_specs; i++)
 	{
-	  for (i = 0; i < arc->n_specs; i++)
+	  struct arc_spec *spec = &arc->specs[i];
+	  
+	  if (--spec->items->refcnt == 0)
 	    {
-	      struct arc_spec *spec = &arc->specs[i];
-	      if (spec->items)
-		{
-		  hmap_destroy (spec->items);
-		  free (spec->items);
-		}
+	      hmap_destroy (&spec->items->ht);
+	      free (spec->items);
 	    }
 	}
 
+      if (arc->global_items && --arc->global_items->refcnt == 0)
+	{
+	  hmap_destroy (&arc->global_items->ht);
+	  free (arc->global_items);
+	}
+      
       free (arc->specs);
       free (arc);
     }
@@ -333,7 +399,7 @@ find_arc_item (const struct arc_spec *spec, const union value *value,
 {
   struct arc_item *item;
 
-  HMAP_FOR_EACH_WITH_HASH (item, struct arc_item, hmap_node, hash, spec->items)
+  HMAP_FOR_EACH_WITH_HASH (item, struct arc_item, hmap_node, hash, &spec->items->ht)
     if (value_equal (value, &item->from, var_get_width (spec->src)))
       return item;
   return NULL;
@@ -373,9 +439,8 @@ autorecode_trns_proc (void *arc_, struct ccase **c,
       const struct arc_spec *spec = &arc->specs[i];
       int width = var_get_width (spec->src);
       const union value *value = case_data (*c, spec->src);
-      struct arc_item *item;
+      const struct arc_item *item = find_arc_item (spec, value, value_hash (value, width, 0));
 
-      item = find_arc_item (spec, value, value_hash (value, width, 0));
       case_data_rw (*c, spec->dst)->f = item ? item->to : SYSMIS;
     }