1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000, 2009, 2010 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/format.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/cast.h"
36 #include "libpspp/compiler.h"
37 #include "libpspp/message.h"
38 #include "libpspp/pool.h"
39 #include "libpspp/str.h"
41 #include "gl/xalloc.h"
44 #define _(msgid) gettext (msgid)
48 /* Type of source value for RECODE. */
51 MAP_SINGLE, /* Specific value. */
52 MAP_RANGE, /* Range of values. */
53 MAP_SYSMIS, /* System missing value. */
54 MAP_MISSING, /* Any missing value. */
55 MAP_ELSE, /* Any value. */
56 MAP_CONVERT /* "123" => 123. */
59 /* Describes input values to be mapped. */
62 enum map_in_type type; /* One of MAP_*. */
63 union value x, y; /* Source values. */
66 /* Describes the value used as output from a mapping. */
69 bool copy_input; /* If true, copy input to output. */
70 union value value; /* If copy_input false, recoded value. */
71 int width; /* If copy_input false, output value width. */
74 /* Describes how to recode a single value or range of values into a
78 struct map_in in; /* Input values. */
79 struct map_out out; /* Output value. */
82 /* RECODE transformation. */
89 /* Variable types, for convenience. */
90 enum val_type src_type; /* src_vars[*] type. */
91 enum val_type dst_type; /* dst_vars[*] type. */
94 const struct variable **src_vars; /* Source variables. */
95 const struct variable **dst_vars; /* Destination variables. */
96 const struct dictionary *dst_dict; /* Dictionary of dst_vars */
97 char **dst_names; /* Name of dest variables, if they're new. */
98 size_t var_cnt; /* Number of variables. */
101 struct mapping *mappings; /* Value mappings. */
102 size_t map_cnt; /* Number of mappings. */
103 int max_src_width; /* Maximum width of src_vars[*]. */
104 int max_dst_width; /* Maximum width of any map_out in 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 val_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 == VAL_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 map_allocated;
199 /* Find length of longest source variable. */
200 trns->max_src_width = var_get_width (trns->src_vars[0]);
201 for (i = 1; i < trns->var_cnt; i++)
203 size_t var_width = var_get_width (trns->src_vars[i]);
204 if (var_width > trns->max_src_width)
205 trns->max_src_width = var_width;
208 /* Parse the mappings in parentheses. */
209 trns->mappings = NULL;
212 have_dst_type = false;
213 if (!lex_force_match (lexer, '('))
217 enum val_type dst_type;
219 if (!lex_match_id (lexer, "CONVERT"))
222 size_t first_map_idx;
225 first_map_idx = trns->map_cnt;
227 /* Parse source specifications. */
232 if (!parse_map_in (lexer, &in, trns->pool,
233 trns->src_type, trns->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))
244 dst_type = trns->src_type;
246 dst_type = val_type_from_width (out.width);
247 if (have_dst_type && dst_type != trns->dst_type)
249 msg (SE, _("Inconsistent target variable types. "
251 "must be all numeric or all string."));
255 for (i = first_map_idx; i < trns->map_cnt; i++)
256 trns->mappings[i].out = out;
260 /* Parse CONVERT as a special case. */
262 set_map_in_generic (&in, MAP_CONVERT);
263 add_mapping (trns, &map_allocated, &in);
264 set_map_out_num (&trns->mappings[trns->map_cnt - 1].out, 0.0);
266 dst_type = VAL_NUMERIC;
267 if (trns->src_type != VAL_STRING
268 || (have_dst_type && trns->dst_type != VAL_NUMERIC))
270 msg (SE, _("CONVERT requires string input values and "
271 "numeric output values."));
275 trns->dst_type = dst_type;
276 have_dst_type = true;
278 if (!lex_force_match (lexer, ')'))
281 while (lex_match (lexer, '('));
286 /* Parses a mapping input value into IN, allocating memory from
287 POOL. The source value type must be provided as SRC_TYPE and,
288 if string, the maximum width of a string source variable must
289 be provided in MAX_SRC_WIDTH. Returns true if successful,
290 false on parse error. */
292 parse_map_in (struct lexer *lexer, struct map_in *in, struct pool *pool,
293 enum val_type src_type, size_t max_src_width)
296 if (lex_match_id (lexer, "ELSE"))
297 set_map_in_generic (in, MAP_ELSE);
298 else if (src_type == VAL_NUMERIC)
300 if (lex_match_id (lexer, "MISSING"))
301 set_map_in_generic (in, MAP_MISSING);
302 else if (lex_match_id (lexer, "SYSMIS"))
303 set_map_in_generic (in, MAP_SYSMIS);
307 if (!parse_num_range (lexer, &x, &y, NULL))
309 set_map_in_num (in, x == y ? MAP_SINGLE : MAP_RANGE, x, y);
314 if (lex_match_id (lexer, "MISSING"))
315 set_map_in_generic (in, MAP_MISSING);
316 else if (!lex_force_string (lexer))
320 set_map_in_str (in, pool, lex_tokstr (lexer), max_src_width);
322 if (lex_token (lexer) == T_ID
323 && lex_id_match (ss_cstr ("THRU"), ss_cstr (lex_tokid (lexer))))
325 msg (SE, _("THRU is not allowed with string variables."));
334 /* Adds IN to the list of mappings in TRNS.
335 MAP_ALLOCATED is the current number of allocated mappings,
336 which is updated as needed. */
338 add_mapping (struct recode_trns *trns,
339 size_t *map_allocated, const struct map_in *in)
342 if (trns->map_cnt >= *map_allocated)
343 trns->mappings = pool_2nrealloc (trns->pool, trns->mappings,
345 sizeof *trns->mappings);
346 m = &trns->mappings[trns->map_cnt++];
350 /* Sets IN as a mapping of the given TYPE. */
352 set_map_in_generic (struct map_in *in, enum map_in_type type)
357 /* Sets IN as a numeric mapping of the given TYPE,
358 with X and Y as the two numeric values. */
360 set_map_in_num (struct map_in *in, enum map_in_type type, double x, double y)
367 /* Sets IN as a string mapping, with STRING as the string,
368 allocated from POOL. The string is padded with spaces on the
369 right to WIDTH characters long. */
371 set_map_in_str (struct map_in *in, struct pool *pool,
372 const struct string *string, size_t width)
374 in->type = MAP_SINGLE;
375 value_init_pool (pool, &in->x, width);
376 value_copy_buf_rpad (&in->x, width,
377 CHAR_CAST_BUG (uint8_t *, ds_data (string)),
378 ds_length (string), ' ');
381 /* Parses a mapping output value into OUT, allocating memory from
382 POOL. Returns true if successful, false on parse error. */
384 parse_map_out (struct lexer *lexer, struct pool *pool, struct map_out *out)
386 if (lex_is_number (lexer))
388 set_map_out_num (out, lex_number (lexer));
391 else if (lex_match_id (lexer, "SYSMIS"))
392 set_map_out_num (out, SYSMIS);
393 else if (lex_is_string (lexer))
395 set_map_out_str (out, pool, lex_tokstr (lexer));
398 else if (lex_match_id (lexer, "COPY"))
400 out->copy_input = true;
405 lex_error (lexer, _("expecting output value"));
411 /* Sets OUT as a numeric mapping output with the given VALUE. */
413 set_map_out_num (struct map_out *out, double value)
415 out->copy_input = false;
416 out->value.f = value;
420 /* Sets OUT as a string mapping output with the given VALUE. */
422 set_map_out_str (struct map_out *out, struct pool *pool,
423 const struct string *value)
425 const char *string = ds_data (value);
426 size_t length = ds_length (value);
430 /* A length of 0 will yield a numeric value, which is not
436 out->copy_input = false;
437 value_init_pool (pool, &out->value, length);
438 memcpy (value_str_rw (&out->value, length), string, length);
442 /* Parses a set of target variables into TRNS->dst_vars and
445 parse_dst_vars (struct lexer *lexer, struct recode_trns *trns,
446 const struct dictionary *dict)
450 if (lex_match_id (lexer, "INTO"))
455 if (!parse_mixed_vars_pool (lexer, dict, trns->pool,
456 &trns->dst_names, &name_cnt,
460 if (name_cnt != trns->var_cnt)
462 msg (SE, _("%zu variable(s) cannot be recoded into "
463 "%zu variable(s). Specify the same number "
464 "of variables as source and target variables."),
465 trns->var_cnt, name_cnt);
469 trns->dst_vars = pool_nalloc (trns->pool,
470 trns->var_cnt, sizeof *trns->dst_vars);
471 for (i = 0; i < trns->var_cnt; i++)
473 const struct variable *v;
474 v = trns->dst_vars[i] = dict_lookup_var (dict, trns->dst_names[i]);
475 if (v == NULL && trns->dst_type == VAL_STRING)
477 msg (SE, _("There is no variable named "
478 "%s. (All string variables specified "
479 "on INTO must already exist. Use the "
480 "STRING command to create a string "
490 trns->dst_vars = trns->src_vars;
491 if (trns->src_type != trns->dst_type)
493 msg (SE, _("INTO is required with %s input values "
494 "and %s output values."),
495 trns->src_type == VAL_NUMERIC ? _("numeric") : _("string"),
496 trns->dst_type == VAL_NUMERIC ? _("numeric") : _("string"));
501 for (i = 0; i < trns->var_cnt; i++)
503 const struct variable *v = trns->dst_vars[i];
504 if (v != NULL && var_get_type (v) != trns->dst_type)
506 msg (SE, _("Type mismatch. Cannot store %s data in "
508 trns->dst_type == VAL_STRING ? _("string") : _("numeric"),
509 var_is_alpha (v) ? _("string") : _("numeric"),
518 /* Ensures that all the output values in TRNS are as wide as the
519 widest destination variable. */
521 enlarge_dst_widths (struct recode_trns *trns)
525 trns->max_dst_width = 0;
526 for (i = 0; i < trns->var_cnt; i++)
528 const struct variable *v = trns->dst_vars[i];
529 if (var_get_width (v) > trns->max_dst_width)
530 trns->max_dst_width = var_get_width (v);
533 for (i = 0; i < trns->map_cnt; i++)
535 struct map_out *out = &trns->mappings[i].out;
536 if (!out->copy_input)
537 value_resize_pool (trns->pool, &out->value,
538 out->width, trns->max_dst_width);
542 /* Creates destination variables that don't already exist. */
544 create_dst_vars (struct recode_trns *trns, struct dictionary *dict)
548 trns->dst_dict = dict;
550 for (i = 0; i < trns->var_cnt; i++)
552 const struct variable **var = &trns->dst_vars[i];
553 const char *name = trns->dst_names[i];
555 *var = dict_lookup_var (dict, name);
557 *var = dict_create_var_assert (dict, name, 0);
558 assert (var_get_type (*var) == trns->dst_type);
562 /* Data transformation. */
564 /* Returns the output mapping in TRNS for an input of VALUE on
565 variable V, or a null pointer if there is no mapping. */
566 static const struct map_out *
567 find_src_numeric (struct recode_trns *trns, double value, const struct variable *v)
571 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
573 const struct map_in *in = &m->in;
574 const struct map_out *out = &m->out;
580 match = value == in->x.f;
583 match = var_is_num_missing (v, value, MV_ANY);
586 match = value >= in->x.f && value <= in->y.f;
589 match = value == SYSMIS;
605 /* Returns the output mapping in TRNS for an input of VALUE with
606 the given WIDTH, or a null pointer if there is no mapping. */
607 static const struct map_out *
608 find_src_string (struct recode_trns *trns, const uint8_t *value,
609 const struct variable *src_var)
612 int width = var_get_width (src_var);
614 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
616 const struct map_in *in = &m->in;
617 struct map_out *out = &m->out;
623 match = !memcmp (value, value_str (&in->x, trns->max_src_width),
634 match = data_in (ss_buffer (CHAR_CAST_BUG (char *, value), width),
635 LEGACY_NATIVE, FMT_F, 0, 0, trns->dst_dict,
642 match = var_is_str_missing (src_var, value, MV_ANY);
655 /* Performs RECODE transformation. */
657 recode_trns_proc (void *trns_, struct ccase **c, casenumber case_idx UNUSED)
659 struct recode_trns *trns = trns_;
662 *c = case_unshare (*c);
663 for (i = 0; i < trns->var_cnt; i++)
665 const struct variable *src_var = trns->src_vars[i];
666 const struct variable *dst_var = trns->dst_vars[i];
667 const struct map_out *out;
669 if (trns->src_type == VAL_NUMERIC)
670 out = find_src_numeric (trns, case_num (*c, src_var), src_var);
672 out = find_src_string (trns, case_str (*c, src_var), src_var);
674 if (trns->dst_type == VAL_NUMERIC)
676 double *dst = &case_data_rw (*c, dst_var)->f;
678 *dst = !out->copy_input ? out->value.f : case_num (*c, src_var);
679 else if (trns->src_vars != trns->dst_vars)
684 char *dst = CHAR_CAST_BUG (char *, case_str_rw (*c, dst_var));
687 if (!out->copy_input)
688 memcpy (dst, value_str (&out->value, trns->max_dst_width),
689 var_get_width (dst_var));
690 else if (trns->src_vars != trns->dst_vars)
692 union value *dst_data = case_data_rw (*c, dst_var);
693 const union value *src_data = case_data (*c, src_var);
694 value_copy_rpad (dst_data, var_get_width (dst_var),
695 src_data, var_get_width (src_var), ' ');
698 else if (trns->src_vars != trns->dst_vars)
699 memset (dst, ' ', var_get_width (dst_var));
703 return TRNS_CONTINUE;
706 /* Frees a RECODE transformation. */
708 recode_trns_free (void *trns_)
710 struct recode_trns *trns = trns_;
711 pool_destroy (trns->pool);