X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fautorecode.c;h=7c1b766750fb4319218c34a7ff8a30f47e9cc6fe;hb=3b9dfd387fdcbf5c2eb3c1a578990b58f4157ea5;hp=68571cb2ef4d8f96265824c5c9968f47662f7977;hpb=f550aee00a62fe1d8baf62d83cd7efef6cc2ee92;p=pspp

diff --git a/src/language/stats/autorecode.c b/src/language/stats/autorecode.c
index 68571cb2ef..7c1b766750 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 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
@@ -20,8 +20,8 @@
 
 #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"
@@ -31,11 +31,15 @@
 #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/c-xvasprintf.h"
+#include "gl/mbiter.h"
+
 
 #include "gettext.h"
 #define _(msgid) gettext (msgid)
@@ -47,15 +51,18 @@ struct arc_item
   {
     struct hmap_node hmap_node; /* Element in "struct arc_spec" hash table. */
     union value from;           /* Original value. */
+    int width;                  /* Width of the original value */
+
     double to;			/* Recoded value. */
   };
 
 /* 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. */
@@ -65,34 +72,69 @@ enum arc_direction
     DESCENDING
   };
 
+struct rec_items
+{
+  struct hmap ht;         /* Hash table of "struct arc_item"s. */
+  int refcnt;
+};
+
+
+
 /* AUTORECODE data. */
 struct autorecode_pgm
-  {
-    struct arc_spec *specs;
-    size_t n_specs;
-  };
+{
+  struct arc_spec *specs;
+  size_t n_specs;
+
+  /* Hash table of "struct arc_item"s. */
+  struct rec_items *global_items;
+
+  bool blank_valid;
+};
 
 static trns_proc_func autorecode_trns_proc;
 static trns_free_func autorecode_trns_free;
 
-static int compare_arc_items (const void *, const void *, const void *width);
+static int compare_arc_items (const void *, const void *, const void *aux);
 static void arc_free (struct autorecode_pgm *);
-static struct arc_item *find_arc_item (struct arc_spec *, const union value *,
+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);
   const struct variable **src_vars = NULL;
   char **dst_names = NULL;
   size_t n_srcs = 0;
   size_t n_dsts = 0;
 
-  enum arc_direction direction;
-  bool print;
+  enum arc_direction direction = ASCENDING;
 
   struct casereader *input;
   struct ccase *c;
@@ -100,16 +142,21 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
   size_t i;
   bool ok;
 
+  /* Create procedure. */
+  arc = xzalloc (sizeof *arc);
+  arc->blank_valid = true;
+
   /* Parse variable lists. */
   lex_match_id (lexer, "VARIABLES");
-  lex_match (lexer, '=');
-  if (!parse_variables_const (lexer, dataset_dict (ds), &src_vars, &n_srcs,
-                              PV_NO_DUPLICATE))
+  lex_match (lexer, T_EQUALS);
+  if (!parse_variables_const (lexer, dict, &src_vars, &n_srcs,
+                              PV_NO_DUPLICATE | PV_NO_SCRATCH))
     goto error;
   if (!lex_force_match_id (lexer, "INTO"))
     goto error;
-  lex_match (lexer, '=');
-  if (!parse_DATA_LIST_vars (lexer, &dst_names, &n_dsts, PV_NO_DUPLICATE))
+  lex_match (lexer, T_EQUALS);
+  if (!parse_DATA_LIST_vars (lexer, dict, &dst_names, &n_dsts,
+                             PV_NO_DUPLICATE))
     goto error;
   if (n_dsts != n_srcs)
     {
@@ -123,7 +170,7 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
     {
       const char *name = dst_names[i];
 
-      if (dict_lookup_var (dataset_dict (ds), name) != NULL)
+      if (dict_lookup_var (dict, name) != NULL)
         {
           msg (SE, _("Target variable %s duplicates existing variable %s."),
                name, name);
@@ -132,53 +179,101 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
     }
 
   /* Parse options. */
-  direction = ASCENDING;
-  print = false;
-  while (lex_match (lexer, '/'))
-    if (lex_match_id (lexer, "DESCENDING"))
-      direction = DESCENDING;
-    else if (lex_match_id (lexer, "PRINT"))
-      print = true;
-  if (lex_token (lexer) != '.')
+  while (lex_match (lexer, T_SLASH))
+    {
+      if (lex_match_id (lexer, "DESCENDING"))
+	direction = DESCENDING;
+      else if (lex_match_id (lexer, "PRINT"))
+        {
+          /* Not yet implemented. */
+        }
+      else if (lex_match_id (lexer, "GROUP"))
+	{
+	  arc->global_items = xmalloc (sizeof (*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)
     {
       lex_error (lexer, _("expecting end of command"));
       goto error;
     }
 
-  /* Create procedure. */
-  arc = xmalloc (sizeof *arc);
   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];
-      hmap_init (&spec->items);
+      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 = xzalloc (sizeof (*spec->items));
+	  spec->items->refcnt = 1;
+	  hmap_init (&spec->items->ht);
+	}
     }
 
+
   /* Execute procedure. */
   input = proc_open (ds);
   for (; (c = casereader_read (input)) != NULL; case_unref (c))
     for (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)
+        if ( (item == NULL)
+	     &&  
+	     ( 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++)
     {
@@ -186,33 +281,70 @@ cmd_autorecode (struct lexer *lexer, struct dataset *ds)
       struct arc_item **items;
       struct arc_item *item;
       size_t n_items;
-      int src_width;
       size_t j;
 
       /* Create destination variable. */
-      spec->dst = dict_create_var_assert (dataset_dict (ds), dst_names[i], 0);
+      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);
 
       /* Sort array by value. */
-      src_width = var_get_width (spec->src);
-      sort (items, n_items, sizeof *items, compare_arc_items, &src_width);
+      sort (items, n_items, sizeof *items, compare_arc_items, NULL);
 
       /* Assign recoded values in sorted order. */
       for (j = 0; j < n_items; j++)
-        items[j]->to = direction == ASCENDING ? j + 1 : n_items - j;
+	{
+	  const union value *from = &items[j]->from;
+	  char *recoded_value  = NULL;
+	  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));
+
+	      recoded_value = recode_string (UTF8, dict_get_encoding (dict),
+                                             str, src_width);
+	    }
+	  else
+	    recoded_value = c_xasprintf ("%g", from->f);
+	  
+	  /* Remove trailing whitespace */
+          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);
+	  free (recoded_value);
+	}
 
       /* Free array. */
       free (items);
     }
   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:
@@ -234,44 +366,70 @@ arc_free (struct autorecode_pgm *arc)
       for (i = 0; i < arc->n_specs; i++)
         {
           struct arc_spec *spec = &arc->specs[i];
-          int width = var_get_width (spec->src);
           struct arc_item *item, *next;
 
-          HMAP_FOR_EACH_SAFE (item, next, struct arc_item, hmap_node,
-                              &spec->items)
-            {
-              value_destroy (&item->from, width);
-              hmap_delete (&spec->items, &item->hmap_node);
-              free (item);
-            }
-          hmap_destroy (&spec->items);
+	  HMAP_FOR_EACH_SAFE (item, next, struct arc_item, hmap_node,
+			      &spec->items->ht)
+	    {
+	      value_destroy (&item->from, item->width);
+	      hmap_delete (&spec->items->ht, &item->hmap_node);
+	      free (item);
+	    }
         }
+
+      for (i = 0; i < arc->n_specs; i++)
+	{
+	  struct arc_spec *spec = &arc->specs[i];
+	  
+	  if (--spec->items->refcnt == 0)
+	    {
+	      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);
     }
 }
 
 static struct arc_item *
-find_arc_item (struct arc_spec *spec, const union value *value,
+find_arc_item (const struct arc_spec *spec, const union value *value,
                size_t hash)
 {
   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;
 }
 
 static int
-compare_arc_items (const void *a_, const void *b_, const void *width_)
+compare_arc_items (const void *a_, const void *b_, const void *aux UNUSED)
 {
   const struct arc_item *const *a = a_;
   const struct arc_item *const *b = b_;
-  const int *width = width_;
+  int width_a = (*a)->width;
+  int width_b = (*b)->width;
 
-  return value_compare_3way (&(*a)->from, &(*b)->from, *width);
+  if ( width_a == width_b)
+    return value_compare_3way (&(*a)->from, &(*b)->from, width_a);
+
+  if ( width_a == 0 && width_b != 0)
+    return -1;
+
+  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);
 }
 
 static int
@@ -284,12 +442,11 @@ autorecode_trns_proc (void *arc_, struct ccase **c,
   *c = case_unshare (*c);
   for (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);
-      struct arc_item *item;
+      const struct arc_spec *spec = &arc->specs[i];
+      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);
 
-      item = find_arc_item (spec, value, value_hash (value, width, 0));
       case_data_rw (*c, spec->dst)->f = item ? item->to : SYSMIS;
     }