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/>. */
23 #include <data/case.h>
24 #include <data/data-in.h>
25 #include <data/format.h>
26 #include <data/dictionary.h>
27 #include <data/procedure.h>
28 #include <data/transformations.h>
29 #include <data/variable.h>
30 #include <language/command.h>
31 #include <language/lexer/lexer.h>
32 #include <language/lexer/value-parser.h>
33 #include <language/lexer/variable-parser.h>
34 #include <libpspp/assertion.h>
35 #include <libpspp/compiler.h>
36 #include <libpspp/message.h>
37 #include <libpspp/pool.h>
38 #include <libpspp/str.h>
43 #define _(msgid) gettext (msgid)
47 /* Type of source value for RECODE. */
50 MAP_SINGLE, /* Specific value. */
51 MAP_RANGE, /* Range of values. */
52 MAP_SYSMIS, /* System missing value. */
53 MAP_MISSING, /* Any missing value. */
54 MAP_ELSE, /* Any value. */
55 MAP_CONVERT /* "123" => 123. */
58 /* Describes input values to be mapped. */
61 enum map_in_type type; /* One of MAP_*. */
62 union value x, y; /* Source values. */
65 /* Describes the value used as output from a mapping. */
68 bool copy_input; /* If true, copy input to output. */
69 union value value; /* If copy_input false, recoded value. */
70 int width; /* If copy_input false, output value width. */
73 /* Describes how to recode a single value or range of values into a
77 struct map_in in; /* Input values. */
78 struct map_out out; /* Output value. */
81 /* RECODE transformation. */
88 /* Variable types, for convenience. */
89 enum val_type src_type; /* src_vars[*] type. */
90 enum val_type dst_type; /* dst_vars[*] type. */
93 const struct variable **src_vars; /* Source variables. */
94 const struct variable **dst_vars; /* Destination variables. */
95 const struct dictionary *dst_dict; /* Dictionary of dst_vars */
96 char **dst_names; /* Name of dest variables, if they're new. */
97 size_t var_cnt; /* Number of variables. */
100 struct mapping *mappings; /* Value mappings. */
101 size_t map_cnt; /* Number of mappings. */
102 int max_src_width; /* Maximum width of src_vars[*]. */
103 int max_dst_width; /* Maximum width of any map_out in mappings. */
106 static bool parse_src_vars (struct lexer *, struct recode_trns *, const struct dictionary *dict);
107 static bool parse_mappings (struct lexer *, struct recode_trns *);
108 static bool parse_dst_vars (struct lexer *, struct recode_trns *, const struct dictionary *dict);
110 static void add_mapping (struct recode_trns *,
111 size_t *map_allocated, const struct map_in *);
113 static bool parse_map_in (struct lexer *lexer, struct map_in *, struct pool *,
114 enum val_type src_type, size_t max_src_width);
115 static void set_map_in_generic (struct map_in *, enum map_in_type);
116 static void set_map_in_num (struct map_in *, enum map_in_type, double, double);
117 static void set_map_in_str (struct map_in *, struct pool *,
118 const struct string *, size_t width);
120 static bool parse_map_out (struct lexer *lexer, struct pool *, struct map_out *);
121 static void set_map_out_num (struct map_out *, double);
122 static void set_map_out_str (struct map_out *, struct pool *,
123 const struct string *);
125 static void enlarge_dst_widths (struct recode_trns *);
126 static void create_dst_vars (struct recode_trns *, struct dictionary *);
128 static trns_proc_func recode_trns_proc;
129 static trns_free_func recode_trns_free;
133 /* Parses the RECODE transformation. */
135 cmd_recode (struct lexer *lexer, struct dataset *ds)
139 struct recode_trns *trns
140 = pool_create_container (struct recode_trns, pool);
142 /* Parse source variable names,
143 then input to output mappings,
144 then destintation variable names. */
145 if (!parse_src_vars (lexer, trns, dataset_dict (ds) )
146 || !parse_mappings (lexer, trns)
147 || !parse_dst_vars (lexer, trns, dataset_dict (ds)))
149 recode_trns_free (trns);
153 /* Ensure that all the output strings are at least as wide
154 as the widest destination variable. */
155 if (trns->dst_type == VAL_STRING)
156 enlarge_dst_widths (trns);
158 /* Create destination variables, if needed.
159 This must be the final step; otherwise we'd have to
160 delete destination variables on failure. */
161 if (trns->src_vars != trns->dst_vars)
162 create_dst_vars (trns, dataset_dict (ds));
165 add_transformation (ds,
166 recode_trns_proc, recode_trns_free, trns);
168 while (lex_match (lexer, '/'));
170 return lex_end_of_command (lexer);
173 /* Parses a set of variables to recode into TRNS->src_vars and
174 TRNS->var_cnt. Sets TRNS->src_type. Returns true if
175 successful, false on parse error. */
177 parse_src_vars (struct lexer *lexer,
178 struct recode_trns *trns, const struct dictionary *dict)
180 if (!parse_variables_const (lexer, dict, &trns->src_vars, &trns->var_cnt,
183 pool_register (trns->pool, free, trns->src_vars);
184 trns->src_type = var_get_type (trns->src_vars[0]);
188 /* Parses a set of mappings, which take the form (input=output),
189 into TRNS->mappings and TRNS->map_cnt. Sets TRNS->dst_type.
190 Returns true if successful, false on parse error. */
192 parse_mappings (struct lexer *lexer, struct recode_trns *trns)
194 size_t map_allocated;
198 /* Find length of longest source variable. */
199 trns->max_src_width = var_get_width (trns->src_vars[0]);
200 for (i = 1; i < trns->var_cnt; i++)
202 size_t var_width = var_get_width (trns->src_vars[i]);
203 if (var_width > trns->max_src_width)
204 trns->max_src_width = var_width;
207 /* Parse the mappings in parentheses. */
208 trns->mappings = NULL;
211 have_dst_type = false;
212 if (!lex_force_match (lexer, '('))
216 enum val_type dst_type;
218 if (!lex_match_id (lexer, "CONVERT"))
221 size_t first_map_idx;
224 first_map_idx = trns->map_cnt;
226 /* Parse source specifications. */
231 if (!parse_map_in (lexer, &in, trns->pool,
232 trns->src_type, trns->max_src_width))
234 add_mapping (trns, &map_allocated, &in);
235 lex_match (lexer, ',');
237 while (!lex_match (lexer, '='));
239 if (!parse_map_out (lexer, trns->pool, &out))
243 dst_type = trns->src_type;
245 dst_type = val_type_from_width (out.width);
246 if (have_dst_type && dst_type != trns->dst_type)
248 msg (SE, _("Inconsistent target variable types. "
250 "must be all numeric or all string."));
254 for (i = first_map_idx; i < trns->map_cnt; i++)
255 trns->mappings[i].out = out;
259 /* Parse CONVERT as a special case. */
261 set_map_in_generic (&in, MAP_CONVERT);
262 add_mapping (trns, &map_allocated, &in);
263 set_map_out_num (&trns->mappings[trns->map_cnt - 1].out, 0.0);
265 dst_type = VAL_NUMERIC;
266 if (trns->src_type != VAL_STRING
267 || (have_dst_type && trns->dst_type != VAL_NUMERIC))
269 msg (SE, _("CONVERT requires string input values and "
270 "numeric output values."));
274 trns->dst_type = dst_type;
275 have_dst_type = true;
277 if (!lex_force_match (lexer, ')'))
280 while (lex_match (lexer, '('));
285 /* Parses a mapping input value into IN, allocating memory from
286 POOL. The source value type must be provided as SRC_TYPE and,
287 if string, the maximum width of a string source variable must
288 be provided in MAX_SRC_WIDTH. Returns true if successful,
289 false on parse error. */
291 parse_map_in (struct lexer *lexer, struct map_in *in, struct pool *pool,
292 enum val_type src_type, size_t max_src_width)
295 if (lex_match_id (lexer, "ELSE"))
296 set_map_in_generic (in, MAP_ELSE);
297 else if (src_type == VAL_NUMERIC)
299 if (lex_match_id (lexer, "MISSING"))
300 set_map_in_generic (in, MAP_MISSING);
301 else if (lex_match_id (lexer, "SYSMIS"))
302 set_map_in_generic (in, MAP_SYSMIS);
306 if (!parse_num_range (lexer, &x, &y, NULL))
308 set_map_in_num (in, x == y ? MAP_SINGLE : MAP_RANGE, x, y);
313 if (lex_match_id (lexer, "MISSING"))
314 set_map_in_generic (in, MAP_MISSING);
315 else if (!lex_force_string (lexer))
319 set_map_in_str (in, pool, lex_tokstr (lexer), max_src_width);
321 if (lex_token (lexer) == T_ID
322 && lex_id_match (ss_cstr ("THRU"), ss_cstr (lex_tokid (lexer))))
324 msg (SE, _("THRU is not allowed with string variables."));
333 /* Adds IN to the list of mappings in TRNS.
334 MAP_ALLOCATED is the current number of allocated mappings,
335 which is updated as needed. */
337 add_mapping (struct recode_trns *trns,
338 size_t *map_allocated, const struct map_in *in)
341 if (trns->map_cnt >= *map_allocated)
342 trns->mappings = pool_2nrealloc (trns->pool, trns->mappings,
344 sizeof *trns->mappings);
345 m = &trns->mappings[trns->map_cnt++];
349 /* Sets IN as a mapping of the given TYPE. */
351 set_map_in_generic (struct map_in *in, enum map_in_type type)
356 /* Sets IN as a numeric mapping of the given TYPE,
357 with X and Y as the two numeric values. */
359 set_map_in_num (struct map_in *in, enum map_in_type type, double x, double y)
366 /* Sets IN as a string mapping, with STRING as the string,
367 allocated from POOL. The string is padded with spaces on the
368 right to WIDTH characters long. */
370 set_map_in_str (struct map_in *in, struct pool *pool,
371 const struct string *string, size_t width)
373 in->type = MAP_SINGLE;
374 value_init_pool (pool, &in->x, width);
375 value_copy_buf_rpad (&in->x, width,
376 ds_data (string), ds_length (string), ' ');
379 /* Parses a mapping output value into OUT, allocating memory from
380 POOL. Returns true if successful, false on parse error. */
382 parse_map_out (struct lexer *lexer, struct pool *pool, struct map_out *out)
384 if (lex_is_number (lexer))
386 set_map_out_num (out, lex_number (lexer));
389 else if (lex_match_id (lexer, "SYSMIS"))
390 set_map_out_num (out, SYSMIS);
391 else if (lex_token (lexer) == T_STRING)
393 set_map_out_str (out, pool, lex_tokstr (lexer));
396 else if (lex_match_id (lexer, "COPY"))
398 out->copy_input = true;
403 lex_error (lexer, _("expecting output value"));
409 /* Sets OUT as a numeric mapping output with the given VALUE. */
411 set_map_out_num (struct map_out *out, double value)
413 out->copy_input = false;
414 out->value.f = value;
418 /* Sets OUT as a string mapping output with the given VALUE. */
420 set_map_out_str (struct map_out *out, struct pool *pool,
421 const struct string *value)
423 const char *string = ds_data (value);
424 size_t length = ds_length (value);
428 /* A length of 0 will yield a numeric value, which is not
434 out->copy_input = false;
435 value_init_pool (pool, &out->value, length);
436 memcpy (value_str_rw (&out->value, length), string, length);
440 /* Parses a set of target variables into TRNS->dst_vars and
443 parse_dst_vars (struct lexer *lexer, struct recode_trns *trns,
444 const struct dictionary *dict)
448 if (lex_match_id (lexer, "INTO"))
453 if (!parse_mixed_vars_pool (lexer, dict, trns->pool,
454 &trns->dst_names, &name_cnt,
458 if (name_cnt != trns->var_cnt)
460 msg (SE, _("%zu variable(s) cannot be recoded into "
461 "%zu variable(s). Specify the same number "
462 "of variables as source and target variables."),
463 trns->var_cnt, name_cnt);
467 trns->dst_vars = pool_nalloc (trns->pool,
468 trns->var_cnt, sizeof *trns->dst_vars);
469 for (i = 0; i < trns->var_cnt; i++)
471 const struct variable *v;
472 v = trns->dst_vars[i] = dict_lookup_var (dict, trns->dst_names[i]);
473 if (v == NULL && trns->dst_type == VAL_STRING)
475 msg (SE, _("There is no variable named "
476 "%s. (All string variables specified "
477 "on INTO must already exist. Use the "
478 "STRING command to create a string "
488 trns->dst_vars = trns->src_vars;
489 if (trns->src_type != trns->dst_type)
491 msg (SE, _("INTO is required with %s input values "
492 "and %s output values."),
493 trns->src_type == VAL_NUMERIC ? _("numeric") : _("string"),
494 trns->dst_type == VAL_NUMERIC ? _("numeric") : _("string"));
499 for (i = 0; i < trns->var_cnt; i++)
501 const struct variable *v = trns->dst_vars[i];
502 if (v != NULL && var_get_type (v) != trns->dst_type)
504 msg (SE, _("Type mismatch. Cannot store %s data in "
506 trns->dst_type == VAL_STRING ? _("string") : _("numeric"),
507 var_is_alpha (v) ? _("string") : _("numeric"),
516 /* Ensures that all the output values in TRNS are as wide as the
517 widest destination variable. */
519 enlarge_dst_widths (struct recode_trns *trns)
523 trns->max_dst_width = 0;
524 for (i = 0; i < trns->var_cnt; i++)
526 const struct variable *v = trns->dst_vars[i];
527 if (var_get_width (v) > trns->max_dst_width)
528 trns->max_dst_width = var_get_width (v);
531 for (i = 0; i < trns->map_cnt; i++)
533 struct map_out *out = &trns->mappings[i].out;
534 if (!out->copy_input)
535 value_resize_pool (trns->pool, &out->value,
536 out->width, trns->max_dst_width);
540 /* Creates destination variables that don't already exist. */
542 create_dst_vars (struct recode_trns *trns, struct dictionary *dict)
546 trns->dst_dict = dict;
548 for (i = 0; i < trns->var_cnt; i++)
550 const struct variable **var = &trns->dst_vars[i];
551 const char *name = trns->dst_names[i];
553 *var = dict_lookup_var (dict, name);
555 *var = dict_create_var_assert (dict, name, 0);
556 assert (var_get_type (*var) == trns->dst_type);
560 /* Data transformation. */
562 /* Returns the output mapping in TRNS for an input of VALUE on
563 variable V, or a null pointer if there is no mapping. */
564 static const struct map_out *
565 find_src_numeric (struct recode_trns *trns, double value, const struct variable *v)
569 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
571 const struct map_in *in = &m->in;
572 const struct map_out *out = &m->out;
578 match = value == in->x.f;
581 match = var_is_num_missing (v, value, MV_ANY);
584 match = value >= in->x.f && value <= in->y.f;
587 match = value == SYSMIS;
603 /* Returns the output mapping in TRNS for an input of VALUE with
604 the given WIDTH, or a null pointer if there is no mapping. */
605 static const struct map_out *
606 find_src_string (struct recode_trns *trns, const uint8_t *value,
607 const struct variable *src_var)
610 int width = var_get_width (src_var);
612 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
614 const struct map_in *in = &m->in;
615 struct map_out *out = &m->out;
621 match = !memcmp (value, value_str (&in->x, trns->max_src_width),
632 match = data_in (ss_buffer (value, width), LEGACY_NATIVE,
633 FMT_F, 0, 0, 0, trns->dst_dict, &uv, 0);
639 match = var_is_str_missing (src_var, value, MV_ANY);
652 /* Performs RECODE transformation. */
654 recode_trns_proc (void *trns_, struct ccase **c, casenumber case_idx UNUSED)
656 struct recode_trns *trns = trns_;
659 *c = case_unshare (*c);
660 for (i = 0; i < trns->var_cnt; i++)
662 const struct variable *src_var = trns->src_vars[i];
663 const struct variable *dst_var = trns->dst_vars[i];
664 const struct map_out *out;
666 if (trns->src_type == VAL_NUMERIC)
667 out = find_src_numeric (trns, case_num (*c, src_var), src_var);
669 out = find_src_string (trns, case_str (*c, src_var), src_var);
671 if (trns->dst_type == VAL_NUMERIC)
673 double *dst = &case_data_rw (*c, dst_var)->f;
675 *dst = !out->copy_input ? out->value.f : case_num (*c, src_var);
676 else if (trns->src_vars != trns->dst_vars)
681 char *dst = case_str_rw (*c, dst_var);
684 if (!out->copy_input)
685 memcpy (dst, value_str (&out->value, trns->max_dst_width),
686 var_get_width (dst_var));
687 else if (trns->src_vars != trns->dst_vars)
689 union value *dst_data = case_data_rw (*c, dst_var);
690 const union value *src_data = case_data (*c, src_var);
691 value_copy_rpad (dst_data, var_get_width (dst_var),
692 src_data, var_get_width (src_var), ' ');
695 else if (trns->src_vars != trns->dst_vars)
696 memset (dst, ' ', var_get_width (dst_var));
700 return TRNS_CONTINUE;
703 /* Frees a RECODE transformation. */
705 recode_trns_free (void *trns_)
707 struct recode_trns *trns = trns_;
708 pool_destroy (trns->pool);