1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000 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/dictionary.h>
26 #include <data/procedure.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 <language/lexer/range-parser.h>
33 #include <libpspp/assertion.h>
34 #include <libpspp/compiler.h>
35 #include <libpspp/message.h>
36 #include <libpspp/pool.h>
37 #include <libpspp/str.h>
42 #define _(msgid) gettext (msgid)
46 /* Type of source value for RECODE. */
49 MAP_SINGLE, /* Specific value. */
50 MAP_RANGE, /* Range of values. */
51 MAP_SYSMIS, /* System missing value. */
52 MAP_MISSING, /* Any missing value. */
53 MAP_ELSE, /* Any value. */
54 MAP_CONVERT /* "123" => 123. */
57 /* A value involved in a RECODE mapping. */
60 double f; /* Numeric. */
61 char *c; /* Short or long string. */
64 /* Describes input values to be mapped. */
67 enum map_in_type type; /* One of MAP_*. */
68 union recode_value x, y; /* Source values. */
71 /* Describes the value used as output from a mapping. */
74 bool copy_input; /* If true, copy input to output. */
75 union recode_value value; /* If copy_input false, recoded value. */
76 int width; /* If copy_input false, output value width. */
79 /* Describes how to recode a single value or range of values into a
83 struct map_in in; /* Input values. */
84 struct map_out out; /* Output value. */
87 /* RECODE transformation. */
92 /* Variable types, for convenience. */
93 enum var_type src_type; /* src_vars[*]->type. */
94 enum var_type dst_type; /* dst_vars[*]->type. */
97 const struct variable **src_vars; /* Source variables. */
98 const struct variable **dst_vars; /* Destination variables. */
99 char **dst_names; /* Name of dest variables, if they're new. */
100 size_t var_cnt; /* Number of variables. */
103 struct mapping *mappings; /* Value mappings. */
104 size_t map_cnt; /* Number of mappings. */
107 static bool parse_src_vars (struct lexer *, struct recode_trns *, const struct dictionary *dict);
108 static bool parse_mappings (struct lexer *, struct recode_trns *);
109 static bool parse_dst_vars (struct lexer *, struct recode_trns *, const struct dictionary *dict);
111 static void add_mapping (struct recode_trns *,
112 size_t *map_allocated, const struct map_in *);
114 static bool parse_map_in (struct lexer *lexer, struct map_in *, struct pool *,
115 enum var_type src_type, size_t max_src_width);
116 static void set_map_in_generic (struct map_in *, enum map_in_type);
117 static void set_map_in_num (struct map_in *, enum map_in_type, double, double);
118 static void set_map_in_str (struct map_in *, struct pool *,
119 const struct string *, size_t width);
121 static bool parse_map_out (struct lexer *lexer, struct pool *, struct map_out *);
122 static void set_map_out_num (struct map_out *, double);
123 static void set_map_out_str (struct map_out *, struct pool *,
124 const struct string *);
126 static void enlarge_dst_widths (struct recode_trns *);
127 static void create_dst_vars (struct recode_trns *, struct dictionary *);
129 static trns_proc_func recode_trns_proc;
130 static trns_free_func recode_trns_free;
134 /* Parses the RECODE transformation. */
136 cmd_recode (struct lexer *lexer, struct dataset *ds)
140 struct recode_trns *trns
141 = pool_create_container (struct recode_trns, pool);
143 /* Parse source variable names,
144 then input to output mappings,
145 then destintation variable names. */
146 if (!parse_src_vars (lexer, trns, dataset_dict (ds) )
147 || !parse_mappings (lexer, trns)
148 || !parse_dst_vars (lexer, trns, dataset_dict (ds)))
150 recode_trns_free (trns);
154 /* Ensure that all the output strings are at least as wide
155 as the widest destination variable. */
156 if (trns->dst_type == VAR_STRING)
157 enlarge_dst_widths (trns);
159 /* Create destination variables, if needed.
160 This must be the final step; otherwise we'd have to
161 delete destination variables on failure. */
162 if (trns->src_vars != trns->dst_vars)
163 create_dst_vars (trns, dataset_dict (ds));
166 add_transformation (ds,
167 recode_trns_proc, recode_trns_free, trns);
169 while (lex_match (lexer, '/'));
171 return lex_end_of_command (lexer);
174 /* Parses a set of variables to recode into TRNS->src_vars and
175 TRNS->var_cnt. Sets TRNS->src_type. Returns true if
176 successful, false on parse error. */
178 parse_src_vars (struct lexer *lexer,
179 struct recode_trns *trns, const struct dictionary *dict)
181 if (!parse_variables_const (lexer, dict, &trns->src_vars, &trns->var_cnt,
184 pool_register (trns->pool, free, trns->src_vars);
185 trns->src_type = var_get_type (trns->src_vars[0]);
189 /* Parses a set of mappings, which take the form (input=output),
190 into TRNS->mappings and TRNS->map_cnt. Sets TRNS->dst_type.
191 Returns true if successful, false on parse error. */
193 parse_mappings (struct lexer *lexer, struct recode_trns *trns)
195 size_t max_src_width;
196 size_t map_allocated;
200 /* Find length of longest source variable. */
201 max_src_width = var_get_width (trns->src_vars[0]);
202 for (i = 1; i < trns->var_cnt; i++)
204 size_t var_width = var_get_width (trns->src_vars[i]);
205 if (var_width > max_src_width)
206 max_src_width = var_width;
209 /* Parse the mappings in parentheses. */
210 trns->mappings = NULL;
213 have_dst_type = false;
214 if (!lex_force_match (lexer, '('))
218 enum var_type dst_type;
220 if (!lex_match_id (lexer, "CONVERT"))
223 size_t first_map_idx;
226 first_map_idx = trns->map_cnt;
228 /* Parse source specifications. */
232 if (!parse_map_in (lexer, &in, trns->pool,
233 trns->src_type, max_src_width))
235 add_mapping (trns, &map_allocated, &in);
236 lex_match (lexer, ',');
238 while (!lex_match (lexer, '='));
240 if (!parse_map_out (lexer, trns->pool, &out))
242 dst_type = var_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 = VAR_NUMERIC;
263 if (trns->src_type != VAR_STRING
264 || (have_dst_type && trns->dst_type != VAR_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 var_type src_type, size_t max_src_width)
291 if (lex_match_id (lexer, "ELSE"))
292 set_map_in_generic (in, MAP_ELSE);
293 else if (src_type == VAR_NUMERIC)
295 if (lex_match_id (lexer, "MISSING"))
296 set_map_in_generic (in, MAP_MISSING);
297 else if (lex_match_id (lexer, "SYSMIS"))
298 set_map_in_generic (in, MAP_SYSMIS);
302 if (!parse_num_range (lexer, &x, &y, NULL))
304 set_map_in_num (in, x == y ? MAP_SINGLE : MAP_RANGE, x, y);
309 if (!lex_force_string (lexer))
311 set_map_in_str (in, pool, lex_tokstr (lexer), max_src_width);
313 if (lex_token (lexer) == T_ID
314 && lex_id_match (ss_cstr ("THRU"), ss_cstr (lex_tokid (lexer))))
316 msg (SE, _("THRU is not allowed with string variables."));
324 /* Adds IN to the list of mappings in TRNS.
325 MAP_ALLOCATED is the current number of allocated mappings,
326 which is updated as needed. */
328 add_mapping (struct recode_trns *trns,
329 size_t *map_allocated, const struct map_in *in)
332 if (trns->map_cnt >= *map_allocated)
333 trns->mappings = pool_2nrealloc (trns->pool, trns->mappings,
335 sizeof *trns->mappings);
336 m = &trns->mappings[trns->map_cnt++];
340 /* Sets IN as a mapping of the given TYPE. */
342 set_map_in_generic (struct map_in *in, enum map_in_type type)
347 /* Sets IN as a numeric mapping of the given TYPE,
348 with X and Y as the two numeric values. */
350 set_map_in_num (struct map_in *in, enum map_in_type type, double x, double y)
357 /* Sets IN as a string mapping, with STRING as the string,
358 allocated from POOL. The string is padded with spaces on the
359 right to WIDTH characters long. */
361 set_map_in_str (struct map_in *in, struct pool *pool,
362 const struct string *string, size_t width)
364 in->type = MAP_SINGLE;
365 in->x.c = pool_alloc_unaligned (pool, width);
366 buf_copy_rpad (in->x.c, width, ds_data (string), ds_length (string));
369 /* Parses a mapping output value into OUT, allocating memory from
370 POOL. Returns true if successful, false on parse error. */
372 parse_map_out (struct lexer *lexer, struct pool *pool, struct map_out *out)
374 if (lex_is_number (lexer))
376 set_map_out_num (out, lex_number (lexer));
379 else if (lex_match_id (lexer, "SYSMIS"))
380 set_map_out_num (out, SYSMIS);
381 else if (lex_token (lexer) == T_STRING)
383 set_map_out_str (out, pool, lex_tokstr (lexer));
386 else if (lex_match_id (lexer, "COPY"))
387 out->copy_input = true;
390 lex_error (lexer, _("expecting output value"));
396 /* Sets OUT as a numeric mapping output with the given VALUE. */
398 set_map_out_num (struct map_out *out, double value)
400 out->copy_input = false;
401 out->value.f = value;
405 /* Sets OUT as a string mapping output with the given VALUE. */
407 set_map_out_str (struct map_out *out, struct pool *pool,
408 const struct string *value)
410 const char *string = ds_data (value);
411 size_t length = ds_length (value);
413 out->copy_input = false;
414 out->value.c = pool_alloc_unaligned (pool, length);
415 memcpy (out->value.c, string, length);
419 /* Parses a set of target variables into TRNS->dst_vars and
422 parse_dst_vars (struct lexer *lexer, struct recode_trns *trns,
423 const struct dictionary *dict)
427 if (lex_match_id (lexer, "INTO"))
432 if (!parse_mixed_vars_pool (lexer, dict, trns->pool,
433 &trns->dst_names, &name_cnt,
437 if (name_cnt != trns->var_cnt)
439 msg (SE, _("%zu variable(s) cannot be recoded into "
440 "%zu variable(s). Specify the same number "
441 "of variables as source and target variables."),
442 trns->var_cnt, name_cnt);
446 trns->dst_vars = pool_nalloc (trns->pool,
447 trns->var_cnt, sizeof *trns->dst_vars);
448 for (i = 0; i < trns->var_cnt; i++)
450 const struct variable *v;
451 v = trns->dst_vars[i] = dict_lookup_var (dict, trns->dst_names[i]);
452 if (v == NULL && trns->dst_type == VAR_STRING)
454 msg (SE, _("There is no variable named "
455 "%s. (All string variables specified "
456 "on INTO must already exist. Use the "
457 "STRING command to create a string "
466 trns->dst_vars = trns->src_vars;
467 if (trns->src_type != trns->dst_type)
469 msg (SE, _("INTO is required with %s input values "
470 "and %s output values."),
471 trns->src_type == VAR_NUMERIC ? _("numeric") : _("string"),
472 trns->dst_type == VAR_NUMERIC ? _("numeric") : _("string"));
477 for (i = 0; i < trns->var_cnt; i++)
479 const struct variable *v = trns->dst_vars[i];
480 if (v != NULL && var_get_type (v) != trns->dst_type)
482 msg (SE, _("Type mismatch. Cannot store %s data in "
484 trns->dst_type == VAR_STRING ? _("string") : _("numeric"),
485 var_is_alpha (v) ? _("string") : _("numeric"),
494 /* Ensures that all the output values in TRNS are as wide as the
495 widest destination variable. */
497 enlarge_dst_widths (struct recode_trns *trns)
499 size_t max_dst_width;
503 for (i = 0; i < trns->var_cnt; i++)
505 const struct variable *v = trns->dst_vars[i];
506 if (var_get_width (v) > max_dst_width)
507 max_dst_width = var_get_width (v);
510 for (i = 0; i < trns->map_cnt; i++)
512 struct map_out *out = &trns->mappings[i].out;
513 if (!out->copy_input && out->width < max_dst_width)
515 char *s = pool_alloc_unaligned (trns->pool, max_dst_width + 1);
516 buf_copy_rpad (s, max_dst_width + 1, out->value.c, out->width);
522 /* Creates destination variables that don't already exist. */
524 create_dst_vars (struct recode_trns *trns, struct dictionary *dict)
528 for (i = 0; i < trns->var_cnt; i++)
530 const struct variable **var = &trns->dst_vars[i];
531 const char *name = trns->dst_names[i];
533 *var = dict_lookup_var (dict, name);
535 *var = dict_create_var_assert (dict, name, 0);
536 assert (var_get_type (*var) == trns->dst_type);
540 /* Data transformation. */
542 /* Returns the output mapping in TRNS for an input of VALUE on
543 variable V, or a null pointer if there is no mapping. */
544 static const struct map_out *
545 find_src_numeric (struct recode_trns *trns, double value, const struct variable *v)
549 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
551 const struct map_in *in = &m->in;
552 const struct map_out *out = &m->out;
558 match = value == in->x.f;
561 match = var_is_num_missing (v, value, MV_ANY);
564 match = value >= in->x.f && value <= in->y.f;
567 match = value == SYSMIS;
583 /* Returns the output mapping in TRNS for an input of VALUE with
584 the given WIDTH, or a null pointer if there is no mapping. */
585 static const struct map_out *
586 find_src_string (struct recode_trns *trns, const char *value, int width)
590 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
592 const struct map_in *in = &m->in;
593 struct map_out *out = &m->out;
599 match = !memcmp (value, in->x.c, width);
609 match = data_in (ss_buffer (value, width), FMT_F, 0, 0, &uv, 0);
625 /* Performs RECODE transformation. */
627 recode_trns_proc (void *trns_, struct ccase *c, casenumber case_idx UNUSED)
629 struct recode_trns *trns = trns_;
632 for (i = 0; i < trns->var_cnt; i++)
634 const struct variable *src_var = trns->src_vars[i];
635 const struct variable *dst_var = trns->dst_vars[i];
637 const union value *src_data = case_data (c, src_var);
638 union value *dst_data = case_data_rw (c, dst_var);
640 const struct map_out *out;
642 if (trns->src_type == VAR_NUMERIC)
643 out = find_src_numeric (trns, src_data->f, src_var);
645 out = find_src_string (trns, src_data->s, var_get_width (src_var));
647 if (trns->dst_type == VAR_NUMERIC)
650 dst_data->f = !out->copy_input ? out->value.f : src_data->f;
651 else if (trns->src_vars != trns->dst_vars)
652 dst_data->f = SYSMIS;
658 if (!out->copy_input)
659 memcpy (dst_data->s, out->value.c, var_get_width (dst_var));
660 else if (trns->src_vars != trns->dst_vars)
661 buf_copy_rpad (dst_data->s, var_get_width (dst_var),
662 src_data->s, var_get_width (src_var));
664 else if (trns->src_vars != trns->dst_vars)
665 memset (dst_data->s, ' ', var_get_width (dst_var));
669 return TRNS_CONTINUE;
672 /* Frees a RECODE transformation. */
674 recode_trns_free (void *trns_)
676 struct recode_trns *trns = trns_;
677 pool_destroy (trns->pool);