1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000, 2009 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
20 #include <data/case.h>
21 #include <data/casereader.h>
22 #include <data/dictionary.h>
23 #include <data/procedure.h>
24 #include <data/transformations.h>
25 #include <data/variable.h>
26 #include <language/command.h>
27 #include <language/lexer/lexer.h>
28 #include <language/lexer/variable-parser.h>
29 #include <libpspp/compiler.h>
30 #include <libpspp/hash.h>
31 #include <libpspp/message.h>
32 #include <libpspp/pool.h>
33 #include <libpspp/str.h>
38 #define _(msgid) gettext (msgid)
40 /* FIXME: Implement PRINT subcommand. */
42 /* An AUTORECODE variable's original value. */
45 double f; /* Numeric. */
46 char *c; /* Short or long string. */
49 /* Explains how to recode one value. `from' must be first element. */
52 union arc_value from; /* Original value. */
53 double to; /* Recoded value. */
56 /* Explains how to recode an AUTORECODE variable. */
59 const struct variable *src; /* Source variable. */
60 struct variable *dest; /* Target variable. */
61 struct hsh_table *items; /* Hash table of `freq's. */
64 /* AUTORECODE transformation. */
65 struct autorecode_trns
67 struct pool *pool; /* Contains AUTORECODE specs. */
68 struct arc_spec *specs; /* AUTORECODE specifications. */
69 size_t spec_cnt; /* Number of specifications. */
72 /* Descending or ascending sort order. */
79 /* AUTORECODE data. */
82 const struct variable **src_vars; /* Source variables. */
83 char **dst_names; /* Target variable names. */
84 struct variable **dst_vars; /* Target variables. */
85 struct hsh_table **src_values; /* `union arc_value's of source vars. */
86 size_t var_cnt; /* Number of variables. */
87 struct pool *src_values_pool; /* Pool used by src_values. */
88 enum direction direction; /* Sort order. */
89 int print; /* Print mapping table if nonzero. */
92 static trns_proc_func autorecode_trns_proc;
93 static trns_free_func autorecode_trns_free;
94 static hsh_compare_func compare_alpha_value, compare_numeric_value;
95 static hsh_hash_func hash_alpha_value, hash_numeric_value;
97 static void recode (struct dataset *, const struct autorecode_pgm *);
98 static void arc_free (struct autorecode_pgm *);
100 /* Performs the AUTORECODE procedure. */
102 cmd_autorecode (struct lexer *lexer, struct dataset *ds)
104 struct autorecode_pgm arc;
105 struct casereader *input;
112 arc.dst_names = NULL;
114 arc.src_values = NULL;
116 arc.src_values_pool = NULL;
117 arc.direction = ASCENDING;
121 lex_match_id (lexer, "VARIABLES");
122 lex_match (lexer, '=');
123 if (!parse_variables_const (lexer, dataset_dict (ds), &arc.src_vars,
127 if (!lex_force_match_id (lexer, "INTO"))
129 lex_match (lexer, '=');
130 if (!parse_DATA_LIST_vars (lexer, &arc.dst_names, &dst_cnt, PV_NONE))
132 if (dst_cnt != arc.var_cnt)
136 msg (SE, _("Source variable count (%zu) does not match "
137 "target variable count (%zu)."),
138 arc.var_cnt, dst_cnt);
140 for (i = 0; i < dst_cnt; i++)
141 free (arc.dst_names[i]);
142 free (arc.dst_names);
143 arc.dst_names = NULL;
147 while (lex_match (lexer, '/'))
148 if (lex_match_id (lexer, "DESCENDING"))
149 arc.direction = DESCENDING;
150 else if (lex_match_id (lexer, "PRINT"))
152 if (lex_token (lexer) != '.')
154 lex_error (lexer, _("expecting end of command"));
158 for (i = 0; i < arc.var_cnt; i++)
162 if (dict_lookup_var (dataset_dict (ds), arc.dst_names[i]) != NULL)
164 msg (SE, _("Target variable %s duplicates existing variable %s."),
165 arc.dst_names[i], arc.dst_names[i]);
168 for (j = 0; j < i; j++)
169 if (!strcasecmp (arc.dst_names[i], arc.dst_names[j]))
171 msg (SE, _("Duplicate variable name %s among target variables."),
177 arc.src_values_pool = pool_create ();
178 arc.dst_vars = xnmalloc (arc.var_cnt, sizeof *arc.dst_vars);
179 arc.src_values = xnmalloc (arc.var_cnt, sizeof *arc.src_values);
180 for (i = 0; i < dst_cnt; i++)
182 /* FIXME: consolodate this hsh_create */
183 if (var_is_alpha (arc.src_vars[i]))
184 arc.src_values[i] = hsh_create (10, compare_alpha_value,
185 hash_alpha_value, NULL, arc.src_vars[i]);
187 arc.src_values[i] = hsh_create (10, compare_numeric_value,
188 hash_numeric_value, NULL, NULL);
191 input = proc_open (ds);
192 for (; (c = casereader_read (input)) != NULL; case_unref (c))
193 for (i = 0; i < arc.var_cnt; i++)
195 union arc_value v, *vp, **vpp;
197 if (var_is_numeric (arc.src_vars[i]))
198 v.f = case_num (c, arc.src_vars[i]);
200 v.c = (char *) case_str (c, arc.src_vars[i]);
202 vpp = (union arc_value **) hsh_probe (arc.src_values[i], &v);
205 vp = pool_alloc (arc.src_values_pool, sizeof *vp);
206 if (var_is_numeric (arc.src_vars[i]))
209 vp->c = pool_clone (arc.src_values_pool,
210 v.c, var_get_width (arc.src_vars[i]));
214 ok = casereader_destroy (input);
215 ok = proc_commit (ds) && ok;
217 for (i = 0; i < arc.var_cnt; i++)
218 arc.dst_vars[i] = dict_create_var_assert (dataset_dict (ds),
219 arc.dst_names[i], 0);
223 return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;
227 return CMD_CASCADING_FAILURE;
231 arc_free (struct autorecode_pgm *arc)
233 free (arc->src_vars);
234 if (arc->dst_names != NULL)
238 for (i = 0; i < arc->var_cnt; i++)
239 free (arc->dst_names[i]);
240 free (arc->dst_names);
242 free (arc->dst_vars);
243 if (arc->src_values != NULL)
247 for (i = 0; i < arc->var_cnt; i++)
248 hsh_destroy (arc->src_values[i]);
249 free (arc->src_values);
251 pool_destroy (arc->src_values_pool);
255 /* AUTORECODE transformation. */
258 recode (struct dataset *ds, const struct autorecode_pgm *arc)
260 struct autorecode_trns *trns;
263 trns = pool_create_container (struct autorecode_trns, pool);
264 trns->specs = pool_nalloc (trns->pool, arc->var_cnt, sizeof *trns->specs);
265 trns->spec_cnt = arc->var_cnt;
266 for (i = 0; i < arc->var_cnt; i++)
268 struct arc_spec *spec = &trns->specs[i];
269 void *const *p = hsh_sort (arc->src_values[i]);
270 int count = hsh_count (arc->src_values[i]);
273 spec->src = arc->src_vars[i];
274 spec->dest = arc->dst_vars[i];
276 if (var_is_alpha (arc->src_vars[i]))
277 spec->items = hsh_create (2 * count, compare_alpha_value,
278 hash_alpha_value, NULL, arc->src_vars[i]);
280 spec->items = hsh_create (2 * count, compare_numeric_value,
281 hash_numeric_value, NULL, NULL);
283 for (j = 0; *p; p++, j++)
285 struct arc_item *item = pool_alloc (trns->pool, sizeof *item);
286 union arc_value *vp = *p;
288 if (var_is_numeric (arc->src_vars[i]))
289 item->from.f = vp->f;
291 item->from.c = pool_clone (trns->pool, vp->c,
292 var_get_width (arc->src_vars[i]));
293 item->to = arc->direction == ASCENDING ? j + 1 : count - j;
294 hsh_force_insert (spec->items, item);
297 add_transformation (ds,
298 autorecode_trns_proc, autorecode_trns_free, trns);
301 /* Executes an AUTORECODE transformation. */
303 autorecode_trns_proc (void *trns_, struct ccase **c,
304 casenumber case_idx UNUSED)
306 struct autorecode_trns *trns = trns_;
309 *c = case_unshare (*c);
310 for (i = 0; i < trns->spec_cnt; i++)
312 struct arc_spec *spec = &trns->specs[i];
313 struct arc_item *item;
316 if (var_is_numeric (spec->src))
317 v.f = case_num (*c, spec->src);
319 v.c = (char *) case_str (*c, spec->src);
320 item = hsh_force_find (spec->items, &v);
322 case_data_rw (*c, spec->dest)->f = item->to;
324 return TRNS_CONTINUE;
327 /* Frees an AUTORECODE transformation. */
329 autorecode_trns_free (void *trns_)
331 struct autorecode_trns *trns = trns_;
334 for (i = 0; i < trns->spec_cnt; i++)
335 hsh_destroy (trns->specs[i].items);
336 pool_destroy (trns->pool);
340 /* AUTORECODE procedure. */
343 compare_alpha_value (const void *a_, const void *b_, const void *v_)
345 const union arc_value *a = a_;
346 const union arc_value *b = b_;
347 const struct variable *v = v_;
349 return memcmp (a->c, b->c, var_get_width (v));
353 hash_alpha_value (const void *a_, const void *v_)
355 const union arc_value *a = a_;
356 const struct variable *v = v_;
358 return hsh_hash_bytes (a->c, var_get_width (v));
362 compare_numeric_value (const void *a_, const void *b_, const void *aux UNUSED)
364 const union arc_value *a = a_;
365 const union arc_value *b = b_;
367 return a->f < b->f ? -1 : a->f > b->f;
371 hash_numeric_value (const void *a_, const void *aux UNUSED)
373 const union arc_value *a = a_;
375 return hsh_hash_double (a->f);