X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fautorecode.c;h=e7d06a818ddfa7c8d860943d82d492696d0d9111;hb=d416357a9e9c23251ff1357f0f1c3ba5ac48223e;hp=3331d74569ad2a4ba475bdc9cb57aba0b68c3dca;hpb=8e7b07bbd3cd28e02aeb5014a5e86a0e9b8f2411;p=pspp

diff --git a/src/language/stats/autorecode.c b/src/language/stats/autorecode.c
index 3331d74569..e7d06a818d 100644
--- a/src/language/stats/autorecode.c
+++ b/src/language/stats/autorecode.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 1997-9, 2000, 2009, 2010 Free Software Foundation, Inc.
+   Copyright (C) 1997-9, 2000, 2009, 2010, 2012, 2013, 2014 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
@@ -16,6 +16,7 @@
 
 #include <config.h>
 
+#include <float.h>
 #include <stdlib.h>
 
 #include "data/case.h"
@@ -37,9 +38,10 @@
 #include "libpspp/str.h"
 
 #include "gl/xalloc.h"
-#include "gl/vasnprintf.h"
+#include "gl/c-xvasprintf.h"
 #include "gl/mbiter.h"
 
+
 #include "gettext.h"
 #define _(msgid) gettext (msgid)
 
@@ -58,9 +60,10 @@ struct arc_item
 /* Explains how to recode an AUTORECODE variable. */
 struct arc_spec
   {
-    const struct variable *src;	/* Source variable. */
+    int width;                  /* Variable width. */
+    int src_idx;                /* Case index of 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. */
@@ -70,6 +73,14 @@ enum arc_direction
     DESCENDING
   };
 
+struct rec_items
+{
+  struct hmap ht;         /* Hash table of "struct arc_item"s. */
+  int refcnt;
+};
+
+
+
 /* AUTORECODE data. */
 struct autorecode_pgm
 {
@@ -77,10 +88,9 @@ 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;
@@ -95,7 +105,7 @@ 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));
+  const char *str = CHAR_CAST_BUG (const char *, val->s);
   char *text = recode_string (UTF8, dict_get_encoding (dict), str, width);
 
   for (mbi_init (iter, text, width); mbi_avail (iter); mbi_advance (iter))
@@ -126,29 +136,27 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
   size_t n_dsts = 0;
 
   enum arc_direction direction = ASCENDING;
-  bool print = true;
 
   struct casereader *input;
   struct ccase *c;
 
-  size_t i;
   bool ok;
 
   /* 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, arc->dict, &src_vars, &n_srcs,
-                              PV_NO_DUPLICATE))
+  if (!parse_variables_const (lexer, dict, &src_vars, &n_srcs,
+                              PV_NO_DUPLICATE | PV_NO_SCRATCH))
     goto error;
+  lex_match (lexer, T_SLASH);
   if (!lex_force_match_id (lexer, "INTO"))
     goto error;
   lex_match (lexer, T_EQUALS);
-  if (!parse_DATA_LIST_vars (lexer, arc->dict, &dst_names, &n_dsts,
+  if (!parse_DATA_LIST_vars (lexer, dict, &dst_names, &n_dsts,
                              PV_NO_DUPLICATE))
     goto error;
   if (n_dsts != n_srcs)
@@ -159,11 +167,11 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
 
       goto error;
     }
-  for (i = 0; i < n_dsts; i++)
+  for (size_t i = 0; i < n_dsts; i++)
     {
       const char *name = dst_names[i];
 
-      if (dict_lookup_var (arc->dict, name) != NULL)
+      if (dict_lookup_var (dict, name) != NULL)
         {
           msg (SE, _("Target variable %s duplicates existing variable %s."),
                name, name);
@@ -177,11 +185,14 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
       if (lex_match_id (lexer, "DESCENDING"))
 	direction = DESCENDING;
       else if (lex_match_id (lexer, "PRINT"))
-	print = true;
+        {
+          /* Not yet implemented. */
+        }
       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"))
 	{
@@ -195,10 +206,16 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
 	      arc->blank_valid = false;
 	    }
 	  else
-	    goto error;
+            {
+              lex_error_expecting (lexer, "VALID", "MISSING");
+              goto error;
+            }
 	}
       else
-	goto error;
+        {
+          lex_error_expecting (lexer, "DESCENDING", "PRINT", "GROUP", "BLANK");
+          goto error;
+        }
     }
 
   if (lex_token (lexer) != T_ENDCMD)
@@ -211,19 +228,23 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
   arc->n_specs = n_dsts;
 
 
-  for (i = 0; i < n_dsts; i++)
+  for (size_t i = 0; i < n_dsts; i++)
     {
       struct arc_spec *spec = &arc->specs[i];
 
-      spec->src = src_vars[i];
+      spec->width = var_get_width (src_vars[i]);
+      spec->src_idx = var_get_case_index (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);
 	}
     }
 
@@ -231,31 +252,37 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
   /* Execute procedure. */
   input = proc_open (ds);
   for (; (c = casereader_read (input)) != NULL; case_unref (c))
-    for (i = 0; i < arc->n_specs; i++)
+    for (size_t i = 0; i < arc->n_specs; i++)
       {
         struct arc_spec *spec = &arc->specs[i];
-        int width = var_get_width (spec->src);
-        const union value *value = case_data (c, spec->src);
+        int width = spec->width;
+        const union value *value = case_data_idx (c, spec->src_idx);
         size_t hash = value_hash (value, width, 0);
         struct arc_item *item;
 
         item = find_arc_item (spec, value, hash);
         if ( (item == NULL)
-	     &&  
-	     ( arc->blank_valid || var_is_numeric (spec->src) || ! value_is_blank (value, width, arc->dict))
+	     &&
+	     ( arc->blank_valid
+               || val_type_from_width (spec->width) == VAL_NUMERIC
+               || ! value_is_blank (value, width, 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);
   ok = proc_commit (ds) && ok;
 
+  /* Re-fetch dictionary because it might have changed (if TEMPORARY was in
+     use). */
+  dict = dataset_dict (ds);
+
   /* Create transformation. */
-  for (i = 0; i < arc->n_specs; i++)
+  for (size_t i = 0; i < arc->n_specs; i++)
     {
       struct arc_spec *spec = &arc->specs[i];
       struct arc_item **items;
@@ -267,10 +294,10 @@ 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);
@@ -282,35 +309,33 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
       for (j = 0; j < n_items; j++)
 	{
 	  const union value *from = &items[j]->from;
-	  size_t len;
 	  char *recoded_value  = NULL;
-	  char *c;
 	  const int src_width = items[j]->width;
 	  union value to_val;
+	  size_t len;
+
 	  value_init (&to_val, 0);
 
 	  items[j]->to = direction == ASCENDING ? j + 1 : n_items - j;
-	  
+
 	  to_val.f = items[j]->to;
 
 	  /* Add value labels to the destination variable which indicate
 	     the source value from whence the new value comes. */
 	  if (src_width > 0)
 	    {
-	      const char *str = CHAR_CAST_BUG (const char*, value_str (from, src_width));
+	      const char *str = CHAR_CAST_BUG (const char *, from->s);
 
-	      recoded_value = recode_string (UTF8, dict_get_encoding (arc->dict), str, src_width);
+	      recoded_value = recode_string (UTF8, dict_get_encoding (dict),
+                                             str, src_width);
 	    }
 	  else
-	    recoded_value = asnprintf (NULL, &len, "%g", from->f);
-	  
+	    recoded_value = c_xasprintf ("%.*g", DBL_DIG + 1, from->f);
+
 	  /* Remove trailing whitespace */
-	  for (c = recoded_value; *c != '\0'; c++)
-	    if ( *c == ' ')
-	      {
-		*c = '\0';
-		break;
-	      }
+          len = strlen (recoded_value);
+          while (len > 0 && recoded_value[len - 1] == ' ')
+            recoded_value[--len] = '\0';
 
 	  var_add_value_label (spec->dst, &to_val, recoded_value);
 	  value_destroy (&to_val, 0);
@@ -322,7 +347,7 @@ 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++)
+  for (size_t i = 0; i < n_dsts; i++)
     free (dst_names[i]);
   free (dst_names);
   free (src_vars);
@@ -330,7 +355,7 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
   return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;
 
 error:
-  for (i = 0; i < n_dsts; i++)
+  for (size_t i = 0; i < n_dsts; i++)
     free (dst_names[i]);
   free (dst_names);
   free (src_vars);
@@ -343,39 +368,37 @@ arc_free (struct autorecode_pgm *arc)
 {
   if (arc != NULL)
     {
-      size_t i;
-
-      for (i = 0; i < arc->n_specs; i++)
+      for (size_t i = 0; i < arc->n_specs; i++)
         {
           struct arc_spec *spec = &arc->specs[i];
           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 (size_t 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);
     }
@@ -387,8 +410,8 @@ 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)
-    if (value_equal (value, &item->from, var_get_width (spec->src)))
+  HMAP_FOR_EACH_WITH_HASH (item, struct arc_item, hmap_node, hash, &spec->items->ht)
+    if (value_equal (value, &item->from, spec->width))
       return item;
   return NULL;
 }
@@ -410,8 +433,8 @@ compare_arc_items (const void *a_, const void *b_, const void *aux UNUSED)
   if ( width_b == 0 && width_a != 0)
     return +1;
 
-  return buf_compare_rpad (CHAR_CAST_BUG (const char *, value_str (&(*a)->from, width_a)), width_a,
-			   CHAR_CAST_BUG (const char *, value_str (&(*b)->from, width_b)), width_b);
+  return buf_compare_rpad (CHAR_CAST_BUG (const char *, (*a)->from.s), width_a,
+			   CHAR_CAST_BUG (const char *, (*b)->from.s), width_b);
 }
 
 static int
@@ -419,15 +442,14 @@ autorecode_trns_proc (void *arc_, struct ccase **c,
                       casenumber case_idx UNUSED)
 {
   struct autorecode_pgm *arc = arc_;
-  size_t i;
 
   *c = case_unshare (*c);
-  for (i = 0; i < arc->n_specs; i++)
+  for (size_t i = 0; i < arc->n_specs; i++)
     {
       const struct arc_spec *spec = &arc->specs[i];
-      int width = var_get_width (spec->src);
-      const union value *value = case_data (*c, spec->src);
-      const struct arc_item *item = find_arc_item (spec, value, value_hash (value, width, 0));
+      const union value *value = case_data_idx (*c, spec->src_idx);
+      unsigned int hash = value_hash (value, spec->width, 0);
+      const struct arc_item *item = find_arc_item (spec, value, hash);
 
       case_data_rw (*c, spec->dst)->f = item ? item->to : SYSMIS;
     }