1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000, 2009, 2010, 2012, 2013, 2014
3 2021, Free Software Foundation, Inc.
5 This program is free software: you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published by
7 the Free Software Foundation, either version 3 of the License, or
8 (at your option) any later version.
10 This program is distributed in the hope that it will be useful,
11 but WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 GNU General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program. If not, see <http://www.gnu.org/licenses/>. */
23 #include "data/case.h"
24 #include "data/casereader.h"
25 #include "data/dataset.h"
26 #include "data/dictionary.h"
27 #include "data/transformations.h"
28 #include "data/variable.h"
29 #include "language/command.h"
30 #include "language/lexer/lexer.h"
31 #include "language/lexer/variable-parser.h"
32 #include "libpspp/array.h"
33 #include "libpspp/compiler.h"
34 #include "libpspp/hash-functions.h"
35 #include "libpspp/hmap.h"
36 #include "libpspp/i18n.h"
37 #include "libpspp/message.h"
38 #include "libpspp/pool.h"
39 #include "libpspp/str.h"
40 #include "output/pivot-table.h"
42 #include "gl/xalloc.h"
43 #include "gl/c-xvasprintf.h"
44 #include "gl/mbiter.h"
45 #include "gl/size_max.h"
48 #define _(msgid) gettext (msgid)
49 #define N_(msgid) (msgid)
51 /* Explains how to recode one value. */
54 struct hmap_node hmap_node; /* Element in "struct arc_spec" hash table. */
55 union value from; /* Original value. */
56 int width; /* Width of the original value */
57 bool missing; /* Is 'from' missing in its source varible? */
58 char *value_label; /* Value label in source variable, if any. */
60 double to; /* Recoded value. */
63 /* Explains how to recode an AUTORECODE variable. */
66 int width; /* Variable width. */
67 int src_idx; /* Case index of source variable. */
68 char *src_name; /* Name of source variable. */
69 struct fmt_spec format; /* Print format in source variable. */
70 size_t dst_idx; /* Target variable. */
71 struct missing_values mv; /* Missing values of source variable. */
72 char *label; /* Variable label of source variable. */
73 struct rec_items *items;
76 /* Descending or ascending sort order. */
85 struct hmap ht; /* Hash table of "struct arc_item"s. */
90 /* AUTORECODE data. */
93 struct caseproto *proto;
95 struct arc_spec *specs;
101 static const struct casereader_translator_class autorecode_trns_class;
103 static int compare_arc_items (const void *, const void *, const void *aux);
104 static void arc_free (struct autorecode_pgm *);
105 static struct arc_item *find_arc_item (
106 const struct rec_items *, const union value *, int width,
109 /* Returns WIDTH with any trailing spaces in VALUE trimmed off (except that a
110 minimum width of 1 is always returned because otherwise the width would
111 indicate a numeric type). */
113 value_trim_spaces (const union value *value, int width)
115 while (width > 1 && value->s[width - 1] == ' ')
120 /* Performs the AUTORECODE procedure. */
122 cmd_autorecode (struct lexer *lexer, struct dataset *ds)
124 struct dictionary *dict = dataset_dict (ds);
126 const struct variable **src_vars = NULL;
129 char **dst_names = NULL;
132 enum arc_direction direction = ASCENDING;
135 /* Create procedure. */
136 struct autorecode_pgm *arc = xmalloc (sizeof *arc);
137 *arc = (struct autorecode_pgm) {
141 /* Parse variable lists. */
142 lex_match_id (lexer, "VARIABLES");
143 lex_match (lexer, T_EQUALS);
144 if (!parse_variables_const (lexer, dict, &src_vars, &n_srcs,
145 PV_NO_DUPLICATE | PV_NO_SCRATCH))
147 lex_match (lexer, T_SLASH);
148 if (!lex_force_match_id (lexer, "INTO"))
150 lex_match (lexer, T_EQUALS);
151 if (!parse_DATA_LIST_vars (lexer, dict, &dst_names, &n_dsts,
154 if (n_dsts != n_srcs)
156 msg (SE, _("Source variable count (%zu) does not match "
157 "target variable count (%zu)."),
162 for (size_t i = 0; i < n_dsts; i++)
164 const char *name = dst_names[i];
166 if (dict_lookup_var (dict, name) != NULL)
168 msg (SE, _("Target variable %s duplicates existing variable %s."),
176 while (lex_match (lexer, T_SLASH))
178 if (lex_match_id (lexer, "DESCENDING"))
179 direction = DESCENDING;
180 else if (lex_match_id (lexer, "PRINT"))
182 else if (lex_match_id (lexer, "GROUP"))
184 else if (lex_match_id (lexer, "BLANK"))
186 lex_match (lexer, T_EQUALS);
187 if (lex_match_id (lexer, "VALID"))
189 arc->blank_valid = true;
191 else if (lex_match_id (lexer, "MISSING"))
193 arc->blank_valid = false;
197 lex_error_expecting (lexer, "VALID", "MISSING");
203 lex_error_expecting (lexer, "DESCENDING", "PRINT", "GROUP", "BLANK");
208 if (lex_token (lexer) != T_ENDCMD)
210 lex_error (lexer, _("Syntax error expecting end of command."));
214 /* If GROUP is specified, verify that the variables are all string or all
218 enum val_type type = var_get_type (src_vars[0]);
219 for (size_t i = 1; i < n_dsts; i++)
221 if (var_get_type (src_vars[i]) != type)
223 size_t string_idx = type == VAL_STRING ? 0 : i;
224 size_t numeric_idx = type == VAL_STRING ? i : 0;
225 lex_error (lexer, _("With GROUP, variables may not mix string "
226 "variables (such as %s) and numeric "
227 "variables (such as %s)."),
228 var_get_name (src_vars[string_idx]),
229 var_get_name (src_vars[numeric_idx]));
235 /* Allocate all the specs and the rec_items that they point to.
237 If GROUP is specified, there is only a single global rec_items, with the
238 maximum width 'width', and all of the specs point to it; otherwise each
239 spec has its own rec_items. */
240 arc->specs = xmalloc (n_dsts * sizeof *arc->specs);
241 arc->n_specs = n_dsts;
242 for (size_t i = 0; i < n_dsts; i++)
244 struct arc_spec *spec = &arc->specs[i];
246 spec->width = var_get_width (src_vars[i]);
247 spec->src_idx = var_get_dict_index (src_vars[i]);
248 spec->src_name = xstrdup (var_get_name (src_vars[i]));
249 spec->format = var_get_print_format (src_vars[i]);
251 const char *label = var_get_label (src_vars[i]);
252 spec->label = xstrdup_if_nonnull (label);
255 spec->items = arc->specs[0].items;
258 spec->items = xzalloc (sizeof (*spec->items));
259 hmap_init (&spec->items->ht);
263 /* Initialize specs[*]->mv to the user-missing values for each
267 /* Use the first source variable that has any user-missing values. */
269 for (size_t i = 0; i < n_dsts; i++)
270 if (var_has_missing_values (src_vars[i]))
276 for (size_t i = 0; i < n_dsts; i++)
277 mv_copy (&arc->specs[i].mv, var_get_missing_values (src_vars[mv_idx]));
281 /* Each variable uses its own user-missing values. */
282 for (size_t i = 0; i < n_dsts; i++)
283 mv_copy (&arc->specs[i].mv, var_get_missing_values (src_vars[i]));
286 /* Execute procedure. */
287 struct casereader *input = proc_open (ds);
289 for (; (c = casereader_read (input)) != NULL; case_unref (c))
290 for (size_t i = 0; i < arc->n_specs; i++)
292 struct arc_spec *spec = &arc->specs[i];
293 const union value *value = case_data_idx (c, spec->src_idx);
294 if (spec->width == 0 && value->f == SYSMIS)
296 /* AUTORECODE never changes the system-missing value.
297 (Leaving it out of the translation table has this
298 effect automatically because values not found in the
299 translation table get translated to system-missing.) */
303 int width = value_trim_spaces (value, spec->width);
304 if (width == 1 && value->s[0] == ' ' && !arc->blank_valid)
307 size_t hash = value_hash (value, width, 0);
308 if (find_arc_item (spec->items, value, width, hash))
311 struct string value_label = DS_EMPTY_INITIALIZER;
312 var_append_value_name__ (src_vars[i], value,
313 SETTINGS_VALUE_SHOW_LABEL, &value_label);
315 struct arc_item *item = xmalloc (sizeof *item);
317 value_clone (&item->from, value, width);
318 item->missing = mv_is_value_missing_varwidth (&spec->mv, value,
320 item->value_label = ds_steal_cstr (&value_label);
321 hmap_insert (&spec->items->ht, &item->hmap_node, hash);
323 ds_destroy (&value_label);
325 bool ok = casereader_destroy (input);
326 ok = proc_commit (ds) && ok;
328 /* Re-fetch dictionary because it might have changed (if TEMPORARY was in
330 dict = dataset_dict (ds);
332 /* Create transformation. */
333 for (size_t i = 0; i < arc->n_specs; i++)
335 struct arc_spec *spec = &arc->specs[i];
337 /* Create destination variable. */
338 struct variable *dst = dict_create_var_assert (dict, dst_names[i], 0);
339 spec->dst_idx = var_get_dict_index (dst);
340 var_set_label (dst, spec->label);
342 /* Set print format. */
343 size_t n_items = hmap_count (&spec->items->ht);
344 char *longest_value = xasprintf ("%zu", n_items);
345 struct fmt_spec format = { .type = FMT_F, .w = strlen (longest_value) };
346 var_set_both_formats (dst, format);
347 free (longest_value);
349 /* Create array of pointers to items. */
350 struct arc_item **items = xmalloc (n_items * sizeof *items);
351 struct arc_item *item;
353 HMAP_FOR_EACH (item, struct arc_item, hmap_node, &spec->items->ht)
355 assert (j == n_items);
357 /* Sort array by value. */
358 sort (items, n_items, sizeof *items, compare_arc_items, &direction);
360 /* Assign recoded values in sorted order. */
361 for (j = 0; j < n_items; j++)
362 items[j]->to = j + 1;
364 if (print && (!group || i == 0))
366 struct pivot_value *title
368 ? pivot_value_new_text (N_("Recoding grouped variables."))
369 : spec->label && spec->label[0]
370 ? pivot_value_new_text_format (N_("Recoding %s into %s (%s)."),
374 : pivot_value_new_text_format (N_("Recoding %s into %s."),
376 var_get_name (dst)));
377 struct pivot_table *table = pivot_table_create__ (title, "Recoding");
379 pivot_dimension_create (
380 table, PIVOT_AXIS_COLUMN, N_("Attributes"),
381 N_("New Value"), N_("Value Label"));
383 struct pivot_dimension *old_values = pivot_dimension_create (
384 table, PIVOT_AXIS_ROW, N_("Old Value"));
385 old_values->root->show_label = true;
387 for (size_t k = 0; k < n_items; k++)
389 const struct arc_item *item = items[k];
390 int old_value_idx = pivot_category_create_leaf (
391 old_values->root, pivot_value_new_value (
392 &item->from, item->width,
394 ? (struct fmt_spec) { .type = FMT_F, .w = item->width }
396 dict_get_encoding (dict)));
397 pivot_table_put2 (table, 0, old_value_idx,
398 pivot_value_new_integer (item->to));
400 const char *value_label = item->value_label;
401 if (value_label && value_label[0])
402 pivot_table_put2 (table, 1, old_value_idx,
403 pivot_value_new_user_text (value_label, -1));
406 pivot_table_submit (table);
409 /* Assign user-missing values.
411 User-missing values in the source variable(s) must be marked
412 as user-missing values in the destination variable. There
413 might be an arbitrary number of missing values, since the
414 source variable might have a range. Our sort function always
415 puts missing values together at the top of the range, so that
416 means that we can use a missing value range to cover all of
417 the user-missing values in any case (but we avoid it unless
418 necessary because user-missing value ranges are an obscure
420 size_t n_missing = n_items;
421 for (size_t k = 0; k < n_items; k++)
422 if (!items[n_items - k - 1]->missing)
429 size_t lo = n_items - (n_missing - 1);
432 struct missing_values mv;
435 mv_add_range (&mv, lo, hi);
437 for (size_t k = 0; k < n_missing; k++)
438 mv_add_num (&mv, lo + k);
439 var_set_missing_values (dst, &mv);
443 /* Add value labels to the destination variable. */
444 for (j = 0; j < n_items; j++)
446 const char *value_label = items[j]->value_label;
447 if (value_label && value_label[0])
449 union value to_val = { .f = items[j]->to };
450 var_add_value_label (dst, &to_val, value_label);
457 arc->proto = caseproto_ref (dict_get_proto (dict));
458 dataset_set_source (ds, casereader_translate_stateless (
459 dataset_steal_source (ds), arc->proto,
460 &autorecode_trns_class, arc));
462 for (size_t i = 0; i < n_dsts; i++)
467 return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;
470 for (size_t i = 0; i < n_dsts; i++)
475 return CMD_CASCADING_FAILURE;
479 arc_free (struct autorecode_pgm *arc)
483 for (size_t i = 0; i < arc->n_specs; i++)
485 struct arc_spec *spec = &arc->specs[i];
486 struct arc_item *item, *next;
488 HMAP_FOR_EACH_SAFE (item, next, struct arc_item, hmap_node,
491 value_destroy (&item->from, item->width);
492 free (item->value_label);
493 hmap_delete (&spec->items->ht, &item->hmap_node);
497 free (spec->src_name);
498 mv_destroy (&spec->mv);
502 (arc->n_specs >= 2 && arc->specs[0].items == arc->specs[1].items
506 for (size_t i = 0; i < n_rec_items; i++)
508 struct arc_spec *spec = &arc->specs[i];
509 hmap_destroy (&spec->items->ht);
518 static struct arc_item *
519 find_arc_item (const struct rec_items *items,
520 const union value *value, int width,
523 struct arc_item *item;
525 HMAP_FOR_EACH_WITH_HASH (item, struct arc_item, hmap_node, hash, &items->ht)
526 if (item->width == width && value_equal (value, &item->from, width))
532 compare_arc_items (const void *a_, const void *b_, const void *direction_)
534 const struct arc_item *const *ap = a_;
535 const struct arc_item *const *bp = b_;
536 const struct arc_item *a = *ap;
537 const struct arc_item *b = *bp;
539 /* User-missing values always sort to the highest target values
540 (regardless of sort direction). */
541 if (a->missing != b->missing)
542 return a->missing < b->missing ? -1 : 1;
544 /* Otherwise, compare the data. */
549 cmp = value_compare_3way (&a->from, &b->from, aw);
553 cmp = buf_compare_rpad (CHAR_CAST_BUG (const char *, a->from.s), aw,
554 CHAR_CAST_BUG (const char *, b->from.s), bw);
557 /* Then apply sort direction. */
558 const enum arc_direction *directionp = direction_;
559 enum arc_direction direction = *directionp;
560 return direction == ASCENDING ? cmp : -cmp;
563 static struct ccase *
564 autorecode_translate (struct ccase *c, void *arc_)
566 struct autorecode_pgm *arc = arc_;
568 c = case_unshare_and_resize (c, arc->proto);
569 for (size_t i = 0; i < arc->n_specs; i++)
571 const struct arc_spec *spec = &arc->specs[i];
572 const union value *value = case_data_idx (c, spec->src_idx);
573 int width = value_trim_spaces (value, spec->width);
574 size_t hash = value_hash (value, width, 0);
575 const struct arc_item *item = find_arc_item (spec->items, value, width,
577 *case_num_rw_idx (c, spec->dst_idx) = item ? item->to : SYSMIS;
584 autorecode_trns_free (void *arc_)
586 struct autorecode_pgm *arc = arc_;
588 caseproto_unref (arc->proto);
593 static const struct casereader_translator_class autorecode_trns_class = {
594 .translate = autorecode_translate,
595 .destroy = autorecode_trns_free,