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 struct variable *dst; /* 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 arc_spec *specs;
99 static trns_proc_func autorecode_trns_proc;
100 static trns_free_func autorecode_trns_free;
102 static int compare_arc_items (const void *, const void *, const void *aux);
103 static void arc_free (struct autorecode_pgm *);
104 static struct arc_item *find_arc_item (
105 const struct rec_items *, const union value *, int width,
108 /* Returns WIDTH with any trailing spaces in VALUE trimmed off (except that a
109 minimum width of 1 is always returned because otherwise the width would
110 indicate a numeric type). */
112 value_trim_spaces (const union value *value, int width)
114 while (width > 1 && value->s[width - 1] == ' ')
119 /* Performs the AUTORECODE procedure. */
121 cmd_autorecode (struct lexer *lexer, struct dataset *ds)
123 struct dictionary *dict = dataset_dict (ds);
125 const struct variable **src_vars = NULL;
128 char **dst_names = NULL;
131 enum arc_direction direction = ASCENDING;
134 /* Create procedure. */
135 struct autorecode_pgm *arc = xzalloc (sizeof *arc);
136 arc->blank_valid = true;
138 /* Parse variable lists. */
139 lex_match_id (lexer, "VARIABLES");
140 lex_match (lexer, T_EQUALS);
141 if (!parse_variables_const (lexer, dict, &src_vars, &n_srcs,
142 PV_NO_DUPLICATE | PV_NO_SCRATCH))
144 lex_match (lexer, T_SLASH);
145 if (!lex_force_match_id (lexer, "INTO"))
147 lex_match (lexer, T_EQUALS);
148 if (!parse_DATA_LIST_vars (lexer, dict, &dst_names, &n_dsts,
151 if (n_dsts != n_srcs)
153 msg (SE, _("Source variable count (%zu) does not match "
154 "target variable count (%zu)."),
159 for (size_t i = 0; i < n_dsts; i++)
161 const char *name = dst_names[i];
163 if (dict_lookup_var (dict, name) != NULL)
165 msg (SE, _("Target variable %s duplicates existing variable %s."),
173 while (lex_match (lexer, T_SLASH))
175 if (lex_match_id (lexer, "DESCENDING"))
176 direction = DESCENDING;
177 else if (lex_match_id (lexer, "PRINT"))
179 else if (lex_match_id (lexer, "GROUP"))
181 else if (lex_match_id (lexer, "BLANK"))
183 lex_match (lexer, T_EQUALS);
184 if (lex_match_id (lexer, "VALID"))
186 arc->blank_valid = true;
188 else if (lex_match_id (lexer, "MISSING"))
190 arc->blank_valid = false;
194 lex_error_expecting (lexer, "VALID", "MISSING");
200 lex_error_expecting (lexer, "DESCENDING", "PRINT", "GROUP", "BLANK");
205 if (lex_token (lexer) != T_ENDCMD)
207 lex_error (lexer, _("expecting end of command"));
211 /* If GROUP is specified, verify that the variables are all string or all
215 enum val_type type = var_get_type (src_vars[0]);
216 for (size_t i = 1; i < n_dsts; i++)
218 if (var_get_type (src_vars[i]) != type)
220 size_t string_idx = type == VAL_STRING ? 0 : i;
221 size_t numeric_idx = type == VAL_STRING ? i : 0;
222 lex_error (lexer, _("With GROUP, variables may not mix string "
223 "variables (such as %s) and numeric "
224 "variables (such as %s)."),
225 var_get_name (src_vars[string_idx]),
226 var_get_name (src_vars[numeric_idx]));
232 /* Allocate all the specs and the rec_items that they point to.
234 If GROUP is specified, there is only a single global rec_items, with the
235 maximum width 'width', and all of the specs point to it; otherwise each
236 spec has its own rec_items. */
237 arc->specs = xmalloc (n_dsts * sizeof *arc->specs);
238 arc->n_specs = n_dsts;
239 for (size_t i = 0; i < n_dsts; i++)
241 struct arc_spec *spec = &arc->specs[i];
243 spec->width = var_get_width (src_vars[i]);
244 spec->src_idx = var_get_case_index (src_vars[i]);
245 spec->src_name = xstrdup (var_get_name (src_vars[i]));
246 spec->format = *var_get_print_format (src_vars[i]);
248 const char *label = var_get_label (src_vars[i]);
249 spec->label = xstrdup_if_nonnull (label);
252 spec->items = arc->specs[0].items;
255 spec->items = xzalloc (sizeof (*spec->items));
256 hmap_init (&spec->items->ht);
260 /* Initialize specs[*]->mv to the user-missing values for each
264 /* Use the first source variable that has any user-missing values. */
266 for (size_t i = 0; i < n_dsts; i++)
267 if (var_has_missing_values (src_vars[i]))
273 for (size_t i = 0; i < n_dsts; i++)
274 mv_copy (&arc->specs[i].mv, var_get_missing_values (src_vars[mv_idx]));
278 /* Each variable uses its own user-missing values. */
279 for (size_t i = 0; i < n_dsts; i++)
280 mv_copy (&arc->specs[i].mv, var_get_missing_values (src_vars[i]));
283 /* Execute procedure. */
284 struct casereader *input = proc_open (ds);
286 for (; (c = casereader_read (input)) != NULL; case_unref (c))
287 for (size_t i = 0; i < arc->n_specs; i++)
289 struct arc_spec *spec = &arc->specs[i];
290 const union value *value = case_data_idx (c, spec->src_idx);
291 if (spec->width == 0 && value->f == SYSMIS)
293 /* AUTORECODE never changes the system-missing value.
294 (Leaving it out of the translation table has this
295 effect automatically because values not found in the
296 translation table get translated to system-missing.) */
300 int width = value_trim_spaces (value, spec->width);
301 if (width == 1 && value->s[0] == ' ' && !arc->blank_valid)
304 size_t hash = value_hash (value, width, 0);
305 if (find_arc_item (spec->items, value, width, hash))
308 struct string value_label = DS_EMPTY_INITIALIZER;
309 var_append_value_name__ (src_vars[i], value,
310 SETTINGS_VALUE_SHOW_LABEL, &value_label);
312 struct arc_item *item = xmalloc (sizeof *item);
314 value_clone (&item->from, value, width);
315 item->missing = mv_is_value_missing_varwidth (&spec->mv, value, spec->width,
317 item->value_label = ds_steal_cstr (&value_label);
318 hmap_insert (&spec->items->ht, &item->hmap_node, hash);
320 ds_destroy (&value_label);
322 bool ok = casereader_destroy (input);
323 ok = proc_commit (ds) && ok;
325 /* Re-fetch dictionary because it might have changed (if TEMPORARY was in
327 dict = dataset_dict (ds);
329 /* Create transformation. */
330 for (size_t i = 0; i < arc->n_specs; i++)
332 struct arc_spec *spec = &arc->specs[i];
334 /* Create destination variable. */
335 spec->dst = dict_create_var_assert (dict, dst_names[i], 0);
336 var_set_label (spec->dst, spec->label);
338 /* Set print format. */
339 size_t n_items = hmap_count (&spec->items->ht);
340 char *longest_value = xasprintf ("%zu", n_items);
341 struct fmt_spec format = { .type = FMT_F, .w = strlen (longest_value) };
342 var_set_both_formats (spec->dst, &format);
343 free (longest_value);
345 /* Create array of pointers to items. */
346 struct arc_item **items = xmalloc (n_items * sizeof *items);
347 struct arc_item *item;
349 HMAP_FOR_EACH (item, struct arc_item, hmap_node, &spec->items->ht)
351 assert (j == n_items);
353 /* Sort array by value. */
354 sort (items, n_items, sizeof *items, compare_arc_items, &direction);
356 /* Assign recoded values in sorted order. */
357 for (j = 0; j < n_items; j++)
358 items[j]->to = j + 1;
360 if (print && (!group || i == 0))
362 struct pivot_value *title
364 ? pivot_value_new_text (N_("Recoding grouped variables."))
365 : spec->label && spec->label[0]
366 ? pivot_value_new_text_format (N_("Recoding %s into %s (%s)."),
368 var_get_name (spec->dst),
370 : pivot_value_new_text_format (N_("Recoding %s into %s."),
372 var_get_name (spec->dst)));
373 struct pivot_table *table = pivot_table_create__ (title, "Recoding");
375 pivot_dimension_create (
376 table, PIVOT_AXIS_COLUMN, N_("Attributes"),
377 N_("New Value"), N_("Value Label"));
379 struct pivot_dimension *old_values = pivot_dimension_create (
380 table, PIVOT_AXIS_ROW, N_("Old Value"));
381 old_values->root->show_label = true;
383 for (size_t k = 0; k < n_items; k++)
385 const struct arc_item *item = items[k];
386 int old_value_idx = pivot_category_create_leaf (
387 old_values->root, pivot_value_new_value (
388 &item->from, item->width,
390 ? &(struct fmt_spec) { .type = FMT_F, .w = item->width }
392 dict_get_encoding (dict)));
393 pivot_table_put2 (table, 0, old_value_idx,
394 pivot_value_new_integer (item->to));
396 const char *value_label = item->value_label;
397 if (value_label && value_label[0])
398 pivot_table_put2 (table, 1, old_value_idx,
399 pivot_value_new_user_text (value_label, -1));
402 pivot_table_submit (table);
405 /* Assign user-missing values.
407 User-missing values in the source variable(s) must be marked
408 as user-missing values in the destination variable. There
409 might be an arbitrary number of missing values, since the
410 source variable might have a range. Our sort function always
411 puts missing values together at the top of the range, so that
412 means that we can use a missing value range to cover all of
413 the user-missing values in any case (but we avoid it unless
414 necessary because user-missing value ranges are an obscure
416 size_t n_missing = n_items;
417 for (size_t k = 0; k < n_items; k++)
418 if (!items[n_items - k - 1]->missing)
425 size_t lo = n_items - (n_missing - 1);
428 struct missing_values mv;
431 mv_add_range (&mv, lo, hi);
433 for (size_t k = 0; k < n_missing; k++)
434 mv_add_num (&mv, lo + k);
435 var_set_missing_values (spec->dst, &mv);
439 /* Add value labels to the destination variable. */
440 for (j = 0; j < n_items; j++)
442 const char *value_label = items[j]->value_label;
443 if (value_label && value_label[0])
445 union value to_val = { .f = items[j]->to };
446 var_add_value_label (spec->dst, &to_val, value_label);
453 add_transformation (ds, autorecode_trns_proc, autorecode_trns_free, arc);
455 for (size_t i = 0; i < n_dsts; i++)
460 return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;
463 for (size_t i = 0; i < n_dsts; i++)
468 return CMD_CASCADING_FAILURE;
472 arc_free (struct autorecode_pgm *arc)
476 for (size_t i = 0; i < arc->n_specs; i++)
478 struct arc_spec *spec = &arc->specs[i];
479 struct arc_item *item, *next;
481 HMAP_FOR_EACH_SAFE (item, next, struct arc_item, hmap_node,
484 value_destroy (&item->from, item->width);
485 free (item->value_label);
486 hmap_delete (&spec->items->ht, &item->hmap_node);
490 free (spec->src_name);
491 mv_destroy (&spec->mv);
495 (arc->n_specs >= 2 && arc->specs[0].items == arc->specs[1].items
499 for (size_t i = 0; i < n_rec_items; i++)
501 struct arc_spec *spec = &arc->specs[i];
502 hmap_destroy (&spec->items->ht);
511 static struct arc_item *
512 find_arc_item (const struct rec_items *items,
513 const union value *value, int width,
516 struct arc_item *item;
518 HMAP_FOR_EACH_WITH_HASH (item, struct arc_item, hmap_node, hash, &items->ht)
519 if (item->width == width && value_equal (value, &item->from, width))
525 compare_arc_items (const void *a_, const void *b_, const void *direction_)
527 const struct arc_item *const *ap = a_;
528 const struct arc_item *const *bp = b_;
529 const struct arc_item *a = *ap;
530 const struct arc_item *b = *bp;
532 /* User-missing values always sort to the highest target values
533 (regardless of sort direction). */
534 if (a->missing != b->missing)
535 return a->missing < b->missing ? -1 : 1;
537 /* Otherwise, compare the data. */
542 cmp = value_compare_3way (&a->from, &b->from, aw);
546 cmp = buf_compare_rpad (CHAR_CAST_BUG (const char *, a->from.s), aw,
547 CHAR_CAST_BUG (const char *, b->from.s), bw);
550 /* Then apply sort direction. */
551 const enum arc_direction *directionp = direction_;
552 enum arc_direction direction = *directionp;
553 return direction == ASCENDING ? cmp : -cmp;
557 autorecode_trns_proc (void *arc_, struct ccase **c,
558 casenumber case_idx UNUSED)
560 struct autorecode_pgm *arc = arc_;
562 *c = case_unshare (*c);
563 for (size_t i = 0; i < arc->n_specs; i++)
565 const struct arc_spec *spec = &arc->specs[i];
566 const union value *value = case_data_idx (*c, spec->src_idx);
567 int width = value_trim_spaces (value, spec->width);
568 size_t hash = value_hash (value, width, 0);
569 const struct arc_item *item = find_arc_item (spec->items, value, width,
571 *case_num_rw (*c, spec->dst) = item ? item->to : SYSMIS;
574 return TRNS_CONTINUE;
578 autorecode_trns_free (void *arc_)
580 struct autorecode_pgm *arc = arc_;