X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fautorecode.c;h=66afb13afaba92e5df8791048292b279539c98ce;hb=b1b2d40a74005efbc4aa09609cc3bea4e2c16410;hp=de9416821c15bf5fa8c37d5e9a241d7a647fd804;hpb=9a331fe64eb814ae5c1322e21717a04fb254bf65;p=pspp
diff --git a/src/language/stats/autorecode.c b/src/language/stats/autorecode.c
index de9416821c..66afb13afa 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 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
@@ -15,362 +15,440 @@
along with this program. If not, see . */
#include
+
+#include
#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-#include
-
-#include "xalloc.h"
+#include "data/case.h"
+#include "data/casereader.h"
+#include "data/dataset.h"
+#include "data/dictionary.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/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)
/* FIXME: Implement PRINT subcommand. */
-/* An AUTORECODE variable's original value. */
-union arc_value
- {
- double f; /* Numeric. */
- char *c; /* Short or long string. */
- };
-
-/* Explains how to recode one value. `from' must be first element. */
+/* Explains how to recode one value. */
struct arc_item
{
- union arc_value from; /* Original value. */
+ 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. */
- struct variable *dest; /* Target variable. */
- struct hsh_table *items; /* Hash table of `freq's. */
- };
-
-/* AUTORECODE transformation. */
-struct autorecode_trns
- {
- struct pool *pool; /* Contains AUTORECODE specs. */
- struct arc_spec *specs; /* AUTORECODE specifications. */
- size_t spec_cnt; /* Number of specifications. */
+ int width; /* Variable width. */
+ int src_idx; /* Case index of source variable. */
+ struct variable *dst; /* Target variable. */
+ struct rec_items *items;
};
/* Descending or ascending sort order. */
-enum direction
+enum arc_direction
{
ASCENDING,
DESCENDING
};
+struct rec_items
+{
+ struct hmap ht; /* Hash table of "struct arc_item"s. */
+};
+
+
+
/* AUTORECODE data. */
struct autorecode_pgm
- {
- const struct variable **src_vars; /* Source variables. */
- char **dst_names; /* Target variable names. */
- struct variable **dst_vars; /* Target variables. */
- struct hsh_table **src_values; /* `union arc_value's of source vars. */
- size_t var_cnt; /* Number of variables. */
- struct pool *src_values_pool; /* Pool used by src_values. */
- enum direction direction; /* Sort order. */
- int print; /* Print mapping table if nonzero. */
- };
+{
+ struct arc_spec *specs;
+ size_t n_specs;
+
+ bool blank_valid;
+};
static trns_proc_func autorecode_trns_proc;
static trns_free_func autorecode_trns_free;
-static hsh_compare_func compare_alpha_value, compare_numeric_value;
-static hsh_hash_func hash_alpha_value, hash_numeric_value;
-static void recode (struct dataset *, const struct autorecode_pgm *);
+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 rec_items *, const union value *, int width,
+ size_t hash);
+
+/* 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)
+{
+ 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;
- struct casereader *input;
- struct ccase *c;
- size_t dst_cnt;
- size_t i;
- bool ok;
-
- arc.src_vars = NULL;
- arc.dst_names = NULL;
- arc.dst_vars = NULL;
- arc.src_values = NULL;
- arc.var_cnt = 0;
- arc.src_values_pool = NULL;
- arc.direction = ASCENDING;
- arc.print = 0;
- dst_cnt = 0;
+ struct dictionary *dict = dataset_dict (ds);
+
+ const struct variable **src_vars = NULL;
+ size_t n_srcs = 0;
+
+ char **dst_names = NULL;
+ size_t n_dsts = 0;
+ enum arc_direction direction = ASCENDING;
+
+ /* Create procedure. */
+ struct autorecode_pgm *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), &arc.src_vars,
- &arc.var_cnt,
- PV_NO_DUPLICATE))
- goto lossage;
+ lex_match (lexer, T_EQUALS);
+ 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 lossage;
- lex_match (lexer, '=');
- if (!parse_DATA_LIST_vars (lexer, &arc.dst_names, &dst_cnt, PV_NONE))
- goto lossage;
- if (dst_cnt != arc.var_cnt)
+ goto error;
+ 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)
{
- size_t i;
-
msg (SE, _("Source variable count (%zu) does not match "
"target variable count (%zu)."),
- arc.var_cnt, dst_cnt);
+ n_srcs, n_dsts);
- for (i = 0; i < dst_cnt; i++)
- free (arc.dst_names[i]);
- free (arc.dst_names);
- arc.dst_names = NULL;
+ goto error;
+ }
+ for (size_t i = 0; i < n_dsts; i++)
+ {
+ const char *name = dst_names[i];
+
+ if (dict_lookup_var (dict, name) != NULL)
+ {
+ msg (SE, _("Target variable %s duplicates existing variable %s."),
+ name, name);
+ goto error;
+ }
+ }
- goto lossage;
+ /* 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"))
+ {
+ /* Not yet implemented. */
+ }
+ else if (lex_match_id (lexer, "GROUP"))
+ group = true;
+ 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
+ {
+ lex_error_expecting (lexer, "VALID", "MISSING");
+ goto error;
+ }
+ }
+ else
+ {
+ lex_error_expecting (lexer, "DESCENDING", "PRINT", "GROUP", "BLANK");
+ goto error;
+ }
}
- while (lex_match (lexer, '/'))
- if (lex_match_id (lexer, "DESCENDING"))
- arc.direction = DESCENDING;
- else if (lex_match_id (lexer, "PRINT"))
- arc.print = 1;
- if (lex_token (lexer) != '.')
+
+ if (lex_token (lexer) != T_ENDCMD)
{
lex_error (lexer, _("expecting end of command"));
- goto lossage;
+ goto error;
}
- for (i = 0; i < arc.var_cnt; i++)
+ /* If GROUP is specified, verify that the variables are all string or all
+ numeric. */
+ if (group)
{
- int j;
+ 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;
+ }
+ }
+ }
- if (dict_lookup_var (dataset_dict (ds), arc.dst_names[i]) != NULL)
+ /* Allocate all the specs and the rec_items that they point to.
+
+ 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->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
{
- msg (SE, _("Target variable %s duplicates existing variable %s."),
- arc.dst_names[i], arc.dst_names[i]);
- goto lossage;
+ spec->items = xzalloc (sizeof (*spec->items));
+ hmap_init (&spec->items->ht);
}
- for (j = 0; j < i; j++)
- if (!strcasecmp (arc.dst_names[i], arc.dst_names[j]))
- {
- msg (SE, _("Duplicate variable name %s among target variables."),
- arc.dst_names[i]);
- goto lossage;
- }
}
- arc.src_values_pool = pool_create ();
- arc.dst_vars = xnmalloc (arc.var_cnt, sizeof *arc.dst_vars);
- arc.src_values = xnmalloc (arc.var_cnt, sizeof *arc.src_values);
- for (i = 0; i < dst_cnt; i++)
- {
- /* FIXME: consolodate this hsh_create */
- if (var_is_alpha (arc.src_vars[i]))
- arc.src_values[i] = hsh_create (10, compare_alpha_value,
- hash_alpha_value, NULL, arc.src_vars[i]);
- else
- arc.src_values[i] = hsh_create (10, compare_numeric_value,
- hash_numeric_value, NULL, NULL);
- }
-
- input = proc_open (ds);
+ /* Execute procedure. */
+ struct casereader *input = proc_open (ds);
+ struct ccase *c;
for (; (c = casereader_read (input)) != NULL; case_unref (c))
- for (i = 0; i < arc.var_cnt; i++)
+ for (size_t i = 0; i < arc->n_specs; i++)
{
- union arc_value v, *vp, **vpp;
-
- if (var_is_numeric (arc.src_vars[i]))
- v.f = case_num (c, arc.src_vars[i]);
- else
- v.c = (char *) case_str (c, arc.src_vars[i]);
-
- vpp = (union arc_value **) hsh_probe (arc.src_values[i], &v);
- if (*vpp == NULL)
- {
- vp = pool_alloc (arc.src_values_pool, sizeof *vp);
- if (var_is_numeric (arc.src_vars[i]))
- vp->f = v.f;
- else
- vp->c = pool_clone (arc.src_values_pool,
- v.c, var_get_width (arc.src_vars[i]));
- *vpp = vp;
- }
+ struct arc_spec *spec = &arc->specs[i];
+ 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);
+ 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;
- for (i = 0; i < arc.var_cnt; i++)
- arc.dst_vars[i] = dict_create_var_assert (dataset_dict (ds),
- arc.dst_names[i], 0);
-
- recode (ds, &arc);
- arc_free (&arc);
- return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;
+ /* Re-fetch dictionary because it might have changed (if TEMPORARY was in
+ use). */
+ dict = dataset_dict (ds);
-lossage:
- arc_free (&arc);
- return CMD_CASCADING_FAILURE;
-}
-
-static void
-arc_free (struct autorecode_pgm *arc)
-{
- free (arc->src_vars);
- if (arc->dst_names != NULL)
+ /* Create transformation. */
+ for (size_t i = 0; i < arc->n_specs; i++)
{
- size_t i;
+ struct arc_spec *spec = &arc->specs[i];
+ struct arc_item **items;
+ struct arc_item *item;
+ size_t n_items;
+ size_t j;
- for (i = 0; i < arc->var_cnt; i++)
- free (arc->dst_names[i]);
- free (arc->dst_names);
- }
- free (arc->dst_vars);
- if (arc->src_values != NULL)
- {
- size_t i;
+ /* Create destination variable. */
+ spec->dst = dict_create_var_assert (dict, dst_names[i], 0);
- for (i = 0; i < arc->var_cnt; i++)
- hsh_destroy (arc->src_values[i]);
- free (arc->src_values);
- }
- pool_destroy (arc->src_values_pool);
-}
+ /* Create array of pointers to 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->ht)
+ items[j++] = item;
-
-/* AUTORECODE transformation. */
+ assert (j == n_items);
-static void
-recode (struct dataset *ds, const struct autorecode_pgm *arc)
-{
- struct autorecode_trns *trns;
- size_t i;
+ /* Sort array by value. */
+ sort (items, n_items, sizeof *items, compare_arc_items, NULL);
- trns = pool_create_container (struct autorecode_trns, pool);
- trns->specs = pool_nalloc (trns->pool, arc->var_cnt, sizeof *trns->specs);
- trns->spec_cnt = arc->var_cnt;
- for (i = 0; i < arc->var_cnt; i++)
- {
- struct arc_spec *spec = &trns->specs[i];
- void *const *p = hsh_sort (arc->src_values[i]);
- int count = hsh_count (arc->src_values[i]);
- int j;
+ /* Assign recoded values in sorted order. */
+ for (j = 0; j < n_items; j++)
+ items[j]->to = direction == ASCENDING ? j + 1 : n_items - j;
- spec->src = arc->src_vars[i];
- spec->dest = arc->dst_vars[i];
-
- if (var_is_alpha (arc->src_vars[i]))
- spec->items = hsh_create (2 * count, compare_alpha_value,
- hash_alpha_value, NULL, arc->src_vars[i]);
- else
- spec->items = hsh_create (2 * count, compare_numeric_value,
- hash_numeric_value, NULL, NULL);
-
- for (j = 0; *p; p++, 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++)
{
- struct arc_item *item = pool_alloc (trns->pool, sizeof *item);
- union arc_value *vp = *p;
-
- if (var_is_numeric (arc->src_vars[i]))
- item->from.f = vp->f;
- else
- item->from.c = pool_clone (trns->pool, vp->c,
- var_get_width (arc->src_vars[i]));
- item->to = arc->direction == ASCENDING ? j + 1 : count - j;
- hsh_force_insert (spec->items, item);
+ const union value *from = &items[j]->from;
+ const int src_width = items[j]->width;
+ char *recoded_value;
+ if (src_width > 0)
+ {
+ const char *str = CHAR_CAST_BUG (const char *, from->s);
+
+ recoded_value = recode_string (UTF8, dict_get_encoding (dict),
+ str, src_width);
+ }
+ else
+ 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);
}
+
+ /* Free array. */
+ free (items);
}
- add_transformation (ds,
- autorecode_trns_proc, autorecode_trns_free, trns);
+ add_transformation (ds, autorecode_trns_proc, autorecode_trns_free, arc);
+
+ 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 (size_t i = 0; i < n_dsts; i++)
+ free (dst_names[i]);
+ free (dst_names);
+ free (src_vars);
+ arc_free (arc);
+ return CMD_CASCADING_FAILURE;
}
-/* Executes an AUTORECODE transformation. */
-static int
-autorecode_trns_proc (void *trns_, struct ccase **c,
- casenumber case_idx UNUSED)
+static void
+arc_free (struct autorecode_pgm *arc)
{
- struct autorecode_trns *trns = trns_;
- size_t i;
-
- *c = case_unshare (*c);
- for (i = 0; i < trns->spec_cnt; i++)
+ if (arc != NULL)
{
- struct arc_spec *spec = &trns->specs[i];
- struct arc_item *item;
- union arc_value v;
-
- if (var_is_numeric (spec->src))
- v.f = case_num (*c, spec->src);
- else
- v.c = (char *) case_str (*c, spec->src);
- item = hsh_force_find (spec->items, &v);
+ 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->ht)
+ {
+ value_destroy (&item->from, item->width);
+ hmap_delete (&spec->items->ht, &item->hmap_node);
+ free (item);
+ }
+ }
+
+ 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++)
+ {
+ struct arc_spec *spec = &arc->specs[i];
+ hmap_destroy (&spec->items->ht);
+ free (spec->items);
+ }
- case_data_rw (*c, spec->dest)->f = item->to;
+ free (arc->specs);
+ free (arc);
}
- return TRNS_CONTINUE;
}
-/* Frees an AUTORECODE transformation. */
-static bool
-autorecode_trns_free (void *trns_)
+static struct arc_item *
+find_arc_item (const struct rec_items *items,
+ const union value *value, int width,
+ size_t hash)
{
- struct autorecode_trns *trns = trns_;
- size_t i;
+ struct arc_item *item;
- for (i = 0; i < trns->spec_cnt; i++)
- hsh_destroy (trns->specs[i].items);
- pool_destroy (trns->pool);
- return true;
+ 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;
}
-
-/* AUTORECODE procedure. */
static int
-compare_alpha_value (const void *a_, const void *b_, const void *v_)
+compare_arc_items (const void *a_, const void *b_, const void *aux UNUSED)
{
- const union arc_value *a = a_;
- const union arc_value *b = b_;
- const struct variable *v = v_;
+ const struct arc_item *const *a = a_;
+ const struct arc_item *const *b = b_;
+ int width_a = (*a)->width;
+ int width_b = (*b)->width;
- return memcmp (a->c, b->c, var_get_width (v));
-}
+ if ( width_a == width_b)
+ return value_compare_3way (&(*a)->from, &(*b)->from, width_a);
-static unsigned
-hash_alpha_value (const void *a_, const void *v_)
-{
- const union arc_value *a = a_;
- const struct variable *v = v_;
+ if ( width_a == 0 && width_b != 0)
+ return -1;
+
+ if ( width_b == 0 && width_a != 0)
+ return +1;
- return hsh_hash_bytes (a->c, var_get_width (v));
+ 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
-compare_numeric_value (const void *a_, const void *b_, const void *aux UNUSED)
+autorecode_trns_proc (void *arc_, struct ccase **c,
+ casenumber case_idx UNUSED)
{
- const union arc_value *a = a_;
- const union arc_value *b = b_;
+ struct autorecode_pgm *arc = arc_;
- return a->f < b->f ? -1 : a->f > b->f;
+ *c = case_unshare (*c);
+ for (size_t i = 0; i < arc->n_specs; i++)
+ {
+ const struct arc_spec *spec = &arc->specs[i];
+ 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;
+ }
+
+ return TRNS_CONTINUE;
}
-static unsigned
-hash_numeric_value (const void *a_, const void *aux UNUSED)
+static bool
+autorecode_trns_free (void *arc_)
{
- const union arc_value *a = a_;
+ struct autorecode_pgm *arc = arc_;
- return hsh_hash_double (a->f);
+ arc_free (arc);
+ return true;
}