/* 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
#include <config.h>
+#include <float.h>
#include <stdlib.h>
#include "data/case.h"
#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)
/* 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. */
DESCENDING
};
+struct rec_items
+{
+ struct hmap ht; /* Hash table of "struct arc_item"s. */
+};
+
+
+
/* AUTORECODE data. */
struct autorecode_pgm
{
struct arc_spec *specs;
size_t n_specs;
- /* Hash table of "struct arc_item"s. */
- struct hmap *global_items;
-
bool blank_valid;
- const struct dictionary *dict;
};
static trns_proc_func autorecode_trns_proc;
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 (const struct arc_spec *, const union value *,
- size_t hash);
+static struct arc_item *find_arc_item (
+ const struct rec_items *, const union value *, int width,
+ size_t hash);
-static bool
-value_is_blank (const union value *val, int width, const struct dictionary *dict)
+/* Returns WIDTH with any trailing spaces in VALUE trimmed off (except that a
+ minimum width of 1 is always returned because otherwise the width would
+ indicate a numeric type). */
+static int
+value_trim_spaces (const union value *value, int width)
{
- 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;
+ while (width > 1 && value->s[width - 1] == ' ')
+ width--;
+ return width;
}
/* 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;
+
+ char **dst_names = NULL;
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);
+ struct autorecode_pgm *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)
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);
}
/* Parse options. */
+ bool group = false;
while (lex_match (lexer, T_SLASH))
{
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);
- }
+ group = true;
else if (lex_match_id (lexer, "BLANK"))
{
lex_match (lexer, T_EQUALS);
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)
goto error;
}
- arc->specs = xmalloc (n_dsts * sizeof *arc->specs);
- arc->n_specs = n_dsts;
+ /* If GROUP is specified, verify that the variables are all string or all
+ numeric. */
+ if (group)
+ {
+ enum val_type type = var_get_type (src_vars[0]);
+ for (size_t i = 1; i < n_dsts; i++)
+ {
+ if (var_get_type (src_vars[i]) != type)
+ {
+ size_t string_idx = type == VAL_STRING ? 0 : i;
+ size_t numeric_idx = type == VAL_STRING ? i : 0;
+ lex_error (lexer, _("With GROUP, variables may not mix string "
+ "variables (such as %s) and numeric "
+ "variables (such as %s)."),
+ var_get_name (src_vars[string_idx]),
+ var_get_name (src_vars[numeric_idx]));
+ goto error;
+ }
+ }
+ }
+ /* Allocate all the specs and the rec_items that they point to.
- for (i = 0; i < n_dsts; i++)
+ If GROUP is specified, there is only a single global rec_items, with the
+ maximum width 'width', and all of the specs point to it; otherwise each
+ spec has its own rec_items. */
+ arc->specs = xmalloc (n_dsts * sizeof *arc->specs);
+ arc->n_specs = n_dsts;
+ for (size_t i = 0; i < n_dsts; i++)
{
struct arc_spec *spec = &arc->specs[i];
- spec->src = src_vars[i];
- if (arc->global_items)
- {
- spec->items = arc->global_items;
- }
+ spec->width = var_get_width (src_vars[i]);
+ spec->src_idx = var_get_case_index (src_vars[i]);
+
+ if (group && i > 0)
+ spec->items = arc->specs[0].items;
else
{
- spec->items = xmalloc (sizeof (*spec->items));
- hmap_init (spec->items);
+ spec->items = xzalloc (sizeof (*spec->items));
+ hmap_init (&spec->items->ht);
}
}
-
/* Execute procedure. */
- input = proc_open (ds);
+ struct casereader *input = proc_open (ds);
+ struct ccase *c;
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);
+ const union value *value = case_data_idx (c, spec->src_idx);
+ int width = value_trim_spaces (value, spec->width);
+ if (width == 1 && value->s[0] == ' ' && !arc->blank_valid)
+ continue;
+
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))
- )
- {
- item = xmalloc (sizeof *item);
- item->width = width;
- value_clone (&item->from, value, width);
- hmap_insert (spec->items, &item->hmap_node, hash);
- }
+ if (find_arc_item (spec->items, value, width, hash))
+ continue;
+
+ struct arc_item *item = xmalloc (sizeof *item);
+ item->width = width;
+ value_clone (&item->from, value, width);
+ hmap_insert (&spec->items->ht, &item->hmap_node, hash);
}
- ok = casereader_destroy (input);
+ bool 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;
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 (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;
+
+ /* Add value labels to the destination variable which indicate
+ the source value from whence the new value comes. */
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;
- 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. */
+ char *recoded_value;
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);
-
- /* Remove trailing whitespace */
- for (c = recoded_value; *c != '\0'; c++)
- if ( *c == ' ')
- {
- *c = '\0';
- break;
- }
-
- var_add_value_label (spec->dst, &to_val, recoded_value);
- value_destroy (&to_val, 0);
+ recoded_value = c_xasprintf ("%.*g", DBL_DIG + 1, from->f);
+
+ /* Remove trailing whitespace. */
+ size_t len = strlen (recoded_value);
+ while (len > 0 && recoded_value[len - 1] == ' ')
+ recoded_value[--len] = '\0';
+
+ /* Add value label, if it would be nonempty. */
+ if (len)
+ {
+ union value to_val = { .f = items[j]->to };
+ var_add_value_label (spec->dst, &to_val, recoded_value);
+ }
free (recoded_value);
}
}
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);
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);
{
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)
+ size_t n_rec_items =
+ (arc->n_specs == 1 || arc->specs[0].items == arc->specs[1].items
+ ? 1
+ : arc->n_specs);
+ for (size_t i = 0; i < n_rec_items; i++)
{
- free (arc->global_items);
- }
- else
- {
- for (i = 0; i < arc->n_specs; i++)
- {
- struct arc_spec *spec = &arc->specs[i];
- if (spec->items)
- {
- hmap_destroy (spec->items);
- free (spec->items);
- }
- }
+ struct arc_spec *spec = &arc->specs[i];
+ hmap_destroy (&spec->items->ht);
+ free (spec->items);
}
free (arc->specs);
}
static struct arc_item *
-find_arc_item (const struct arc_spec *spec, const union value *value,
+find_arc_item (const struct rec_items *items,
+ const union value *value, int width,
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, &items->ht)
+ if (item->width == width && value_equal (value, &item->from, width))
return item;
return NULL;
}
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
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);
+ int width = value_trim_spaces (value, spec->width);
+ size_t hash = value_hash (value, width, 0);
+ const struct arc_item *item = find_arc_item (spec->items, value, width,
+ hash);
case_data_rw (*c, spec->dst)->f = item ? item->to : SYSMIS;
}
projects / pspp / blobdiff