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. */
86 /* Variable types, for convenience. */
87 enum val_type src_type; /* src_vars[*] type. */
88 enum val_type dst_type; /* dst_vars[*] type. */
91 const struct variable **src_vars; /* Source variables. */
92 const struct variable **dst_vars; /* Destination variables. */
93 char **dst_names; /* Name of dest variables, if they're new. */
94 size_t var_cnt; /* Number of variables. */
97 struct mapping *mappings; /* Value mappings. */
98 size_t map_cnt; /* Number of mappings. */
99 int max_src_width; /* Maximum width of src_vars[*]. */
100 int max_dst_width; /* Maximum width of any map_out in mappings. */
103 static bool parse_src_vars (struct lexer *, struct recode_trns *, const struct dictionary *dict);
104 static bool parse_mappings (struct lexer *, struct recode_trns *);
105 static bool parse_dst_vars (struct lexer *, struct recode_trns *, const struct dictionary *dict);
107 static void add_mapping (struct recode_trns *,
108 size_t *map_allocated, const struct map_in *);
110 static bool parse_map_in (struct lexer *lexer, struct map_in *, struct pool *,
111 enum val_type src_type, size_t max_src_width);
112 static void set_map_in_generic (struct map_in *, enum map_in_type);
113 static void set_map_in_num (struct map_in *, enum map_in_type, double, double);
114 static void set_map_in_str (struct map_in *, struct pool *,
115 const struct string *, size_t width);
117 static bool parse_map_out (struct lexer *lexer, struct pool *, struct map_out *);
118 static void set_map_out_num (struct map_out *, double);
119 static void set_map_out_str (struct map_out *, struct pool *,
120 const struct string *);
122 static void enlarge_dst_widths (struct recode_trns *);
123 static void create_dst_vars (struct recode_trns *, struct dictionary *);
125 static trns_proc_func recode_trns_proc;
126 static trns_free_func recode_trns_free;
130 /* Parses the RECODE transformation. */
132 cmd_recode (struct lexer *lexer, struct dataset *ds)
136 struct recode_trns *trns
137 = pool_create_container (struct recode_trns, pool);
139 /* Parse source variable names,
140 then input to output mappings,
141 then destintation variable names. */
142 if (!parse_src_vars (lexer, trns, dataset_dict (ds) )
143 || !parse_mappings (lexer, trns)
144 || !parse_dst_vars (lexer, trns, dataset_dict (ds)))
146 recode_trns_free (trns);
150 /* Ensure that all the output strings are at least as wide
151 as the widest destination variable. */
152 if (trns->dst_type == VAL_STRING)
153 enlarge_dst_widths (trns);
155 /* Create destination variables, if needed.
156 This must be the final step; otherwise we'd have to
157 delete destination variables on failure. */
158 if (trns->src_vars != trns->dst_vars)
159 create_dst_vars (trns, dataset_dict (ds));
162 add_transformation (ds,
163 recode_trns_proc, recode_trns_free, trns);
165 while (lex_match (lexer, '/'));
167 return lex_end_of_command (lexer);
170 /* Parses a set of variables to recode into TRNS->src_vars and
171 TRNS->var_cnt. Sets TRNS->src_type. Returns true if
172 successful, false on parse error. */
174 parse_src_vars (struct lexer *lexer,
175 struct recode_trns *trns, const struct dictionary *dict)
177 if (!parse_variables_const (lexer, dict, &trns->src_vars, &trns->var_cnt,
180 pool_register (trns->pool, free, trns->src_vars);
181 trns->src_type = var_get_type (trns->src_vars[0]);
185 /* Parses a set of mappings, which take the form (input=output),
186 into TRNS->mappings and TRNS->map_cnt. Sets TRNS->dst_type.
187 Returns true if successful, false on parse error. */
189 parse_mappings (struct lexer *lexer, struct recode_trns *trns)
191 size_t map_allocated;
195 /* Find length of longest source variable. */
196 trns->max_src_width = var_get_width (trns->src_vars[0]);
197 for (i = 1; i < trns->var_cnt; i++)
199 size_t var_width = var_get_width (trns->src_vars[i]);
200 if (var_width > trns->max_src_width)
201 trns->max_src_width = var_width;
204 /* Parse the mappings in parentheses. */
205 trns->mappings = NULL;
208 have_dst_type = false;
209 if (!lex_force_match (lexer, '('))
213 enum val_type dst_type;
215 if (!lex_match_id (lexer, "CONVERT"))
218 size_t first_map_idx;
221 first_map_idx = trns->map_cnt;
223 /* Parse source specifications. */
228 if (!parse_map_in (lexer, &in, trns->pool,
229 trns->src_type, trns->max_src_width))
231 add_mapping (trns, &map_allocated, &in);
232 lex_match (lexer, ',');
234 while (!lex_match (lexer, '='));
236 if (!parse_map_out (lexer, trns->pool, &out))
240 dst_type = trns->src_type;
242 dst_type = val_type_from_width (out.width);
243 if (have_dst_type && dst_type != trns->dst_type)
245 msg (SE, _("Inconsistent target variable types. "
247 "must be all numeric or all string."));
251 for (i = first_map_idx; i < trns->map_cnt; i++)
252 trns->mappings[i].out = out;
256 /* Parse CONVERT as a special case. */
258 set_map_in_generic (&in, MAP_CONVERT);
259 add_mapping (trns, &map_allocated, &in);
260 set_map_out_num (&trns->mappings[trns->map_cnt - 1].out, 0.0);
262 dst_type = VAL_NUMERIC;
263 if (trns->src_type != VAL_STRING
264 || (have_dst_type && trns->dst_type != VAL_NUMERIC))
266 msg (SE, _("CONVERT requires string input values and "
267 "numeric output values."));
271 trns->dst_type = dst_type;
272 have_dst_type = true;
274 if (!lex_force_match (lexer, ')'))
277 while (lex_match (lexer, '('));
282 /* Parses a mapping input value into IN, allocating memory from
283 POOL. The source value type must be provided as SRC_TYPE and,
284 if string, the maximum width of a string source variable must
285 be provided in MAX_SRC_WIDTH. Returns true if successful,
286 false on parse error. */
288 parse_map_in (struct lexer *lexer, struct map_in *in, struct pool *pool,
289 enum val_type src_type, size_t max_src_width)
292 if (lex_match_id (lexer, "ELSE"))
293 set_map_in_generic (in, MAP_ELSE);
294 else if (src_type == VAL_NUMERIC)
296 if (lex_match_id (lexer, "MISSING"))
297 set_map_in_generic (in, MAP_MISSING);
298 else if (lex_match_id (lexer, "SYSMIS"))
299 set_map_in_generic (in, MAP_SYSMIS);
303 if (!parse_num_range (lexer, &x, &y, NULL))
305 set_map_in_num (in, x == y ? MAP_SINGLE : MAP_RANGE, x, y);
310 if (lex_match_id (lexer, "MISSING"))
311 set_map_in_generic (in, MAP_MISSING);
312 else if (!lex_force_string (lexer))
316 set_map_in_str (in, pool, lex_tokstr (lexer), max_src_width);
318 if (lex_token (lexer) == T_ID
319 && lex_id_match (ss_cstr ("THRU"), ss_cstr (lex_tokid (lexer))))
321 msg (SE, _("THRU is not allowed with string variables."));
330 /* Adds IN to the list of mappings in TRNS.
331 MAP_ALLOCATED is the current number of allocated mappings,
332 which is updated as needed. */
334 add_mapping (struct recode_trns *trns,
335 size_t *map_allocated, const struct map_in *in)
338 if (trns->map_cnt >= *map_allocated)
339 trns->mappings = pool_2nrealloc (trns->pool, trns->mappings,
341 sizeof *trns->mappings);
342 m = &trns->mappings[trns->map_cnt++];
346 /* Sets IN as a mapping of the given TYPE. */
348 set_map_in_generic (struct map_in *in, enum map_in_type type)
353 /* Sets IN as a numeric mapping of the given TYPE,
354 with X and Y as the two numeric values. */
356 set_map_in_num (struct map_in *in, enum map_in_type type, double x, double y)
363 /* Sets IN as a string mapping, with STRING as the string,
364 allocated from POOL. The string is padded with spaces on the
365 right to WIDTH characters long. */
367 set_map_in_str (struct map_in *in, struct pool *pool,
368 const struct string *string, size_t width)
370 in->type = MAP_SINGLE;
371 value_init_pool (pool, &in->x, width);
372 value_copy_buf_rpad (&in->x, width,
373 ds_data (string), ds_length (string), ' ');
376 /* Parses a mapping output value into OUT, allocating memory from
377 POOL. Returns true if successful, false on parse error. */
379 parse_map_out (struct lexer *lexer, struct pool *pool, struct map_out *out)
381 if (lex_is_number (lexer))
383 set_map_out_num (out, lex_number (lexer));
386 else if (lex_match_id (lexer, "SYSMIS"))
387 set_map_out_num (out, SYSMIS);
388 else if (lex_token (lexer) == T_STRING)
390 set_map_out_str (out, pool, lex_tokstr (lexer));
393 else if (lex_match_id (lexer, "COPY"))
395 out->copy_input = true;
400 lex_error (lexer, _("expecting output value"));
406 /* Sets OUT as a numeric mapping output with the given VALUE. */
408 set_map_out_num (struct map_out *out, double value)
410 out->copy_input = false;
411 out->value.f = value;
415 /* Sets OUT as a string mapping output with the given VALUE. */
417 set_map_out_str (struct map_out *out, struct pool *pool,
418 const struct string *value)
420 const char *string = ds_data (value);
421 size_t length = ds_length (value);
425 /* A length of 0 will yield a numeric value, which is not
431 out->copy_input = false;
432 value_init_pool (pool, &out->value, length);
433 memcpy (value_str_rw (&out->value, length), string, length);
437 /* Parses a set of target variables into TRNS->dst_vars and
440 parse_dst_vars (struct lexer *lexer, struct recode_trns *trns,
441 const struct dictionary *dict)
445 if (lex_match_id (lexer, "INTO"))
450 if (!parse_mixed_vars_pool (lexer, dict, trns->pool,
451 &trns->dst_names, &name_cnt,
455 if (name_cnt != trns->var_cnt)
457 msg (SE, _("%zu variable(s) cannot be recoded into "
458 "%zu variable(s). Specify the same number "
459 "of variables as source and target variables."),
460 trns->var_cnt, name_cnt);
464 trns->dst_vars = pool_nalloc (trns->pool,
465 trns->var_cnt, sizeof *trns->dst_vars);
466 for (i = 0; i < trns->var_cnt; i++)
468 const struct variable *v;
469 v = trns->dst_vars[i] = dict_lookup_var (dict, trns->dst_names[i]);
470 if (v == NULL && trns->dst_type == VAL_STRING)
472 msg (SE, _("There is no variable named "
473 "%s. (All string variables specified "
474 "on INTO must already exist. Use the "
475 "STRING command to create a string "
485 trns->dst_vars = trns->src_vars;
486 if (trns->src_type != trns->dst_type)
488 msg (SE, _("INTO is required with %s input values "
489 "and %s output values."),
490 trns->src_type == VAL_NUMERIC ? _("numeric") : _("string"),
491 trns->dst_type == VAL_NUMERIC ? _("numeric") : _("string"));
496 for (i = 0; i < trns->var_cnt; i++)
498 const struct variable *v = trns->dst_vars[i];
499 if (v != NULL && var_get_type (v) != trns->dst_type)
501 msg (SE, _("Type mismatch. Cannot store %s data in "
503 trns->dst_type == VAL_STRING ? _("string") : _("numeric"),
504 var_is_alpha (v) ? _("string") : _("numeric"),
513 /* Ensures that all the output values in TRNS are as wide as the
514 widest destination variable. */
516 enlarge_dst_widths (struct recode_trns *trns)
520 trns->max_dst_width = 0;
521 for (i = 0; i < trns->var_cnt; i++)
523 const struct variable *v = trns->dst_vars[i];
524 if (var_get_width (v) > trns->max_dst_width)
525 trns->max_dst_width = var_get_width (v);
528 for (i = 0; i < trns->map_cnt; i++)
530 struct map_out *out = &trns->mappings[i].out;
531 if (!out->copy_input)
532 value_resize_pool (trns->pool, &out->value,
533 out->width, trns->max_dst_width);
537 /* Creates destination variables that don't already exist. */
539 create_dst_vars (struct recode_trns *trns, struct dictionary *dict)
543 for (i = 0; i < trns->var_cnt; i++)
545 const struct variable **var = &trns->dst_vars[i];
546 const char *name = trns->dst_names[i];
548 *var = dict_lookup_var (dict, name);
550 *var = dict_create_var_assert (dict, name, 0);
551 assert (var_get_type (*var) == trns->dst_type);
555 /* Data transformation. */
557 /* Returns the output mapping in TRNS for an input of VALUE on
558 variable V, or a null pointer if there is no mapping. */
559 static const struct map_out *
560 find_src_numeric (struct recode_trns *trns, double value, const struct variable *v)
564 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
566 const struct map_in *in = &m->in;
567 const struct map_out *out = &m->out;
573 match = value == in->x.f;
576 match = var_is_num_missing (v, value, MV_ANY);
579 match = value >= in->x.f && value <= in->y.f;
582 match = value == SYSMIS;
598 /* Returns the output mapping in TRNS for an input of VALUE with
599 the given WIDTH, or a null pointer if there is no mapping. */
600 static const struct map_out *
601 find_src_string (struct recode_trns *trns, const uint8_t *value,
602 const struct variable *src_var)
605 int width = var_get_width (src_var);
607 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
609 const struct map_in *in = &m->in;
610 struct map_out *out = &m->out;
616 match = !memcmp (value, value_str (&in->x, trns->max_src_width),
627 match = data_in (ss_buffer (value, width), LEGACY_NATIVE,
628 FMT_F, 0, 0, 0, &uv, 0);
634 match = var_is_str_missing (src_var, value, MV_ANY);
647 /* Performs RECODE transformation. */
649 recode_trns_proc (void *trns_, struct ccase **c, casenumber case_idx UNUSED)
651 struct recode_trns *trns = trns_;
654 *c = case_unshare (*c);
655 for (i = 0; i < trns->var_cnt; i++)
657 const struct variable *src_var = trns->src_vars[i];
658 const struct variable *dst_var = trns->dst_vars[i];
659 const struct map_out *out;
661 if (trns->src_type == VAL_NUMERIC)
662 out = find_src_numeric (trns, case_num (*c, src_var), src_var);
664 out = find_src_string (trns, case_str (*c, src_var), src_var);
666 if (trns->dst_type == VAL_NUMERIC)
668 double *dst = &case_data_rw (*c, dst_var)->f;
670 *dst = !out->copy_input ? out->value.f : case_num (*c, src_var);
671 else if (trns->src_vars != trns->dst_vars)
676 char *dst = case_str_rw (*c, dst_var);
679 if (!out->copy_input)
680 memcpy (dst, value_str (&out->value, trns->max_dst_width),
681 var_get_width (dst_var));
682 else if (trns->src_vars != trns->dst_vars)
684 union value *dst_data = case_data_rw (*c, dst_var);
685 const union value *src_data = case_data (*c, src_var);
686 value_copy_rpad (dst_data, var_get_width (dst_var),
687 src_data, var_get_width (src_var), ' ');
690 else if (trns->src_vars != trns->dst_vars)
691 memset (dst, ' ', var_get_width (dst_var));
695 return TRNS_CONTINUE;
698 /* Frees a RECODE transformation. */
700 recode_trns_free (void *trns_)
702 struct recode_trns *trns = trns_;
703 pool_destroy (trns->pool);