1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000, 2009, 2010, 2011 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/dataset.h"
26 #include "data/dictionary.h"
27 #include "data/format.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/i18n.h"
38 #include "libpspp/message.h"
39 #include "libpspp/pool.h"
40 #include "libpspp/str.h"
42 #include "gl/xalloc.h"
45 #define _(msgid) gettext (msgid)
49 /* Type of source value for RECODE. */
52 MAP_SINGLE, /* Specific value. */
53 MAP_RANGE, /* Range of values. */
54 MAP_SYSMIS, /* System missing value. */
55 MAP_MISSING, /* Any missing value. */
56 MAP_ELSE, /* Any value. */
57 MAP_CONVERT /* "123" => 123. */
60 /* Describes input values to be mapped. */
63 enum map_in_type type; /* One of MAP_*. */
64 union value x, y; /* Source values. */
67 /* Describes the value used as output from a mapping. */
70 bool copy_input; /* If true, copy input to output. */
71 union value value; /* If copy_input false, recoded value. */
72 int width; /* If copy_input false, output value width. */
75 /* Describes how to recode a single value or range of values into a
79 struct map_in in; /* Input values. */
80 struct map_out out; /* Output value. */
83 /* 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 n_vars; /* Number of variables. */
100 struct mapping *mappings; /* Value mappings. */
101 size_t n_maps; /* 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 const char *dict_encoding);
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 const char *dict_encoding);
117 static void set_map_in_generic (struct map_in *, enum map_in_type);
118 static void set_map_in_num (struct map_in *, enum map_in_type, double, double);
119 static void set_map_in_str (struct map_in *, struct pool *,
120 struct substring, size_t width,
121 const char *dict_encoding);
123 static bool parse_map_out (struct lexer *lexer, struct pool *, struct map_out *);
124 static void set_map_out_num (struct map_out *, double);
125 static void set_map_out_str (struct map_out *, struct pool *,
128 static bool enlarge_dst_widths (struct recode_trns *);
129 static void create_dst_vars (struct recode_trns *, struct dictionary *);
131 static bool recode_trns_free (void *trns_);
133 static const struct trns_class recode_trns_class;
137 /* Parses the RECODE transformation. */
139 cmd_recode (struct lexer *lexer, struct dataset *ds)
143 struct dictionary *dict = dataset_dict (ds);
144 struct recode_trns *trns
145 = pool_create_container (struct recode_trns, pool);
147 /* Parse source variable names,
148 then input to output mappings,
149 then destintation variable names. */
150 if (!parse_src_vars (lexer, trns, dict)
151 || !parse_mappings (lexer, trns, dict_get_encoding (dict))
152 || !parse_dst_vars (lexer, trns, dict))
154 recode_trns_free (trns);
158 /* Ensure that all the output strings are at least as wide
159 as the widest destination variable. */
160 if (trns->dst_type == VAL_STRING)
162 if (! enlarge_dst_widths (trns))
164 recode_trns_free (trns);
169 /* Create destination variables, if needed.
170 This must be the final step; otherwise we'd have to
171 delete destination variables on failure. */
172 trns->dst_dict = dict;
173 if (trns->src_vars != trns->dst_vars)
174 create_dst_vars (trns, dict);
177 add_transformation (ds, &recode_trns_class, trns);
179 while (lex_match (lexer, T_SLASH));
184 /* Parses a set of variables to recode into TRNS->src_vars and
185 TRNS->n_vars. Sets TRNS->src_type. Returns true if
186 successful, false on parse error. */
188 parse_src_vars (struct lexer *lexer,
189 struct recode_trns *trns, const struct dictionary *dict)
191 if (!parse_variables_const (lexer, dict, &trns->src_vars, &trns->n_vars,
194 pool_register (trns->pool, free, trns->src_vars);
195 trns->src_type = var_get_type (trns->src_vars[0]);
199 /* Parses a set of mappings, which take the form (input=output),
200 into TRNS->mappings and TRNS->n_maps. Sets TRNS->dst_type.
201 Returns true if successful, false on parse error. */
203 parse_mappings (struct lexer *lexer, struct recode_trns *trns,
204 const char *dict_encoding)
206 size_t map_allocated;
210 /* Find length of longest source variable. */
211 trns->max_src_width = var_get_width (trns->src_vars[0]);
212 for (i = 1; i < trns->n_vars; i++)
214 size_t var_width = var_get_width (trns->src_vars[i]);
215 if (var_width > trns->max_src_width)
216 trns->max_src_width = var_width;
219 /* Parse the mappings in parentheses. */
220 trns->mappings = NULL;
223 have_dst_type = false;
224 if (!lex_force_match (lexer, T_LPAREN))
228 enum val_type dst_type;
230 if (!lex_match_id (lexer, "CONVERT"))
233 size_t first_map_idx;
236 first_map_idx = trns->n_maps;
238 /* Parse source specifications. */
243 if (!parse_map_in (lexer, &in, trns->pool,
244 trns->src_type, trns->max_src_width,
247 add_mapping (trns, &map_allocated, &in);
248 lex_match (lexer, T_COMMA);
250 while (!lex_match (lexer, T_EQUALS));
252 if (!parse_map_out (lexer, trns->pool, &out))
256 dst_type = trns->src_type;
258 dst_type = val_type_from_width (out.width);
259 if (have_dst_type && dst_type != trns->dst_type)
261 msg (SE, _("Inconsistent target variable types. "
263 "must be all numeric or all string."));
267 for (i = first_map_idx; i < trns->n_maps; i++)
268 trns->mappings[i].out = out;
272 /* Parse CONVERT as a special case. */
274 set_map_in_generic (&in, MAP_CONVERT);
275 add_mapping (trns, &map_allocated, &in);
276 set_map_out_num (&trns->mappings[trns->n_maps - 1].out, 0.0);
278 dst_type = VAL_NUMERIC;
279 if (trns->src_type != VAL_STRING
280 || (have_dst_type && trns->dst_type != VAL_NUMERIC))
282 lex_next_error (lexer, -1, -1,
283 _("CONVERT requires string input values and "
284 "numeric output values."));
288 trns->dst_type = dst_type;
289 have_dst_type = true;
291 if (!lex_force_match (lexer, T_RPAREN))
294 while (lex_match (lexer, T_LPAREN));
299 /* Parses a mapping input value into IN, allocating memory from
300 POOL. The source value type must be provided as SRC_TYPE and,
301 if string, the maximum width of a string source variable must
302 be provided in MAX_SRC_WIDTH. Returns true if successful,
303 false on parse error. */
305 parse_map_in (struct lexer *lexer, struct map_in *in, struct pool *pool,
306 enum val_type src_type, size_t max_src_width,
307 const char *dict_encoding)
310 if (lex_match_id (lexer, "ELSE"))
311 set_map_in_generic (in, MAP_ELSE);
312 else if (src_type == VAL_NUMERIC)
314 if (lex_match_id (lexer, "MISSING"))
315 set_map_in_generic (in, MAP_MISSING);
316 else if (lex_match_id (lexer, "SYSMIS"))
317 set_map_in_generic (in, MAP_SYSMIS);
321 if (!parse_num_range (lexer, &x, &y, NULL))
323 set_map_in_num (in, x == y ? MAP_SINGLE : MAP_RANGE, x, y);
328 if (lex_match_id (lexer, "MISSING"))
329 set_map_in_generic (in, MAP_MISSING);
330 else if (!lex_force_string (lexer))
334 set_map_in_str (in, pool, lex_tokss (lexer), max_src_width,
337 if (lex_token (lexer) == T_ID
338 && lex_id_match (ss_cstr ("THRU"), lex_tokss (lexer)))
340 lex_error (lexer, _("%s is not allowed with string variables."),
350 /* Adds IN to the list of mappings in TRNS.
351 MAP_ALLOCATED is the current number of allocated mappings,
352 which is updated as needed. */
354 add_mapping (struct recode_trns *trns,
355 size_t *map_allocated, const struct map_in *in)
358 if (trns->n_maps >= *map_allocated)
359 trns->mappings = pool_2nrealloc (trns->pool, trns->mappings,
361 sizeof *trns->mappings);
362 m = &trns->mappings[trns->n_maps++];
366 /* Sets IN as a mapping of the given TYPE. */
368 set_map_in_generic (struct map_in *in, enum map_in_type type)
373 /* Sets IN as a numeric mapping of the given TYPE,
374 with X and Y as the two numeric values. */
376 set_map_in_num (struct map_in *in, enum map_in_type type, double x, double y)
383 /* Sets IN as a string mapping, with STRING as the string,
384 allocated from POOL. The string is padded with spaces on the
385 right to WIDTH characters long. */
387 set_map_in_str (struct map_in *in, struct pool *pool,
388 struct substring string, size_t width,
389 const char *dict_encoding)
391 char *s = recode_string (dict_encoding, "UTF-8",
392 ss_data (string), ss_length (string));
393 in->type = MAP_SINGLE;
394 value_init_pool (pool, &in->x, width);
395 value_copy_buf_rpad (&in->x, width,
396 CHAR_CAST (uint8_t *, s), strlen (s), ' ');
400 /* Parses a mapping output value into OUT, allocating memory from
401 POOL. Returns true if successful, false on parse error. */
403 parse_map_out (struct lexer *lexer, struct pool *pool, struct map_out *out)
405 if (lex_is_number (lexer))
407 set_map_out_num (out, lex_number (lexer));
410 else if (lex_match_id (lexer, "SYSMIS"))
411 set_map_out_num (out, SYSMIS);
412 else if (lex_is_string (lexer))
414 set_map_out_str (out, pool, lex_tokss (lexer));
417 else if (lex_match_id (lexer, "COPY"))
419 out->copy_input = true;
424 lex_error (lexer, _("Syntax error expecting output value."));
430 /* Sets OUT as a numeric mapping output with the given VALUE. */
432 set_map_out_num (struct map_out *out, double value)
434 out->copy_input = false;
435 out->value.f = value;
439 /* Sets OUT as a string mapping output with the given VALUE. */
441 set_map_out_str (struct map_out *out, struct pool *pool,
442 const struct substring value)
444 const char *string = ss_data (value);
445 size_t length = ss_length (value);
449 /* A length of 0 will yield a numeric value, which is not
455 out->copy_input = false;
456 value_init_pool (pool, &out->value, length);
457 memcpy (out->value.s, string, length);
461 /* Parses a set of target variables into TRNS->dst_vars and
464 parse_dst_vars (struct lexer *lexer, struct recode_trns *trns,
465 const struct dictionary *dict)
469 if (lex_match_id (lexer, "INTO"))
474 if (!parse_mixed_vars_pool (lexer, dict, trns->pool,
475 &trns->dst_names, &n_names,
479 if (n_names != trns->n_vars)
481 msg (SE, _("%zu variable(s) cannot be recoded into "
482 "%zu variable(s). Specify the same number "
483 "of variables as source and target variables."),
484 trns->n_vars, n_names);
488 trns->dst_vars = pool_nalloc (trns->pool,
489 trns->n_vars, sizeof *trns->dst_vars);
490 for (i = 0; i < trns->n_vars; i++)
492 const struct variable *v;
493 v = trns->dst_vars[i] = dict_lookup_var (dict, trns->dst_names[i]);
494 if (v == NULL && trns->dst_type == VAL_STRING)
496 msg (SE, _("There is no variable named "
497 "%s. (All string variables specified "
498 "on INTO must already exist. Use the "
499 "STRING command to create a string "
509 trns->dst_vars = trns->src_vars;
510 if (trns->src_type != trns->dst_type)
512 msg (SE, _("INTO is required with %s input values "
513 "and %s output values."),
514 trns->src_type == VAL_NUMERIC ? _("numeric") : _("string"),
515 trns->dst_type == VAL_NUMERIC ? _("numeric") : _("string"));
520 for (i = 0; i < trns->n_vars; i++)
522 const struct variable *v = trns->dst_vars[i];
523 if (v != NULL && var_get_type (v) != trns->dst_type)
525 if (trns->dst_type == VAL_STRING)
526 msg (SE, _("Type mismatch. Cannot store string data in "
527 "numeric variable %s."), var_get_name (v));
529 msg (SE, _("Type mismatch. Cannot store numeric data in "
530 "string variable %s."), var_get_name (v));
538 /* Ensures that all the output values in TRNS are as wide as the
539 widest destination variable. */
541 enlarge_dst_widths (struct recode_trns *trns)
544 const struct variable *narrow_var = NULL;
545 int min_dst_width = INT_MAX;
546 trns->max_dst_width = 0;
548 for (i = 0; i < trns->n_vars; i++)
550 const struct variable *v = trns->dst_vars[i];
551 if (var_get_width (v) > trns->max_dst_width)
552 trns->max_dst_width = var_get_width (v);
554 if (var_get_width (v) < min_dst_width)
556 min_dst_width = var_get_width (v);
561 for (i = 0; i < trns->n_maps; i++)
563 struct map_out *out = &trns->mappings[i].out;
564 if (!out->copy_input)
566 if (out->width > min_dst_width)
569 _("Cannot recode because the variable %s would require a width of %d bytes or greater, but it has a width of only %d bytes."),
570 var_get_name (narrow_var), out->width, min_dst_width);
574 value_resize_pool (trns->pool, &out->value,
575 out->width, trns->max_dst_width);
582 /* Creates destination variables that don't already exist. */
584 create_dst_vars (struct recode_trns *trns, struct dictionary *dict)
588 for (i = 0; i < trns->n_vars; i++)
590 const struct variable **var = &trns->dst_vars[i];
591 const char *name = trns->dst_names[i];
593 *var = dict_lookup_var (dict, name);
595 *var = dict_create_var_assert (dict, name, 0);
596 assert (var_get_type (*var) == trns->dst_type);
600 /* Data transformation. */
602 /* Returns the output mapping in TRNS for an input of VALUE on
603 variable V, or a null pointer if there is no mapping. */
604 static const struct map_out *
605 find_src_numeric (struct recode_trns *trns, double value, const struct variable *v)
609 for (m = trns->mappings; m < trns->mappings + trns->n_maps; m++)
611 const struct map_in *in = &m->in;
612 const struct map_out *out = &m->out;
618 match = value == in->x.f;
621 match = var_is_num_missing (v, value) != 0;
624 match = value >= in->x.f && value <= in->y.f;
627 match = value == SYSMIS;
643 /* Returns the output mapping in TRNS for an input of VALUE with
644 the given WIDTH, or a null pointer if there is no mapping. */
645 static const struct map_out *
646 find_src_string (struct recode_trns *trns, const uint8_t *value,
647 const struct variable *src_var)
649 const char *encoding = dict_get_encoding (trns->dst_dict);
650 int width = var_get_width (src_var);
653 for (m = trns->mappings; m < trns->mappings + trns->n_maps; m++)
655 const struct map_in *in = &m->in;
656 struct map_out *out = &m->out;
662 match = !memcmp (value, in->x.s, width);
672 error = data_in (ss_buffer (CHAR_CAST_BUG (char *, value), width),
673 C_ENCODING, FMT_F, settings_get_fmt_settings (),
675 match = error == NULL;
682 match = var_is_str_missing (src_var, value) != 0;
695 /* Performs RECODE transformation. */
696 static enum trns_result
697 recode_trns_proc (void *trns_, struct ccase **c, casenumber case_idx UNUSED)
699 struct recode_trns *trns = trns_;
702 *c = case_unshare (*c);
703 for (i = 0; i < trns->n_vars; i++)
705 const struct variable *src_var = trns->src_vars[i];
706 const struct variable *dst_var = trns->dst_vars[i];
707 const struct map_out *out;
709 if (trns->src_type == VAL_NUMERIC)
710 out = find_src_numeric (trns, case_num (*c, src_var), src_var);
712 out = find_src_string (trns, case_str (*c, src_var), src_var);
714 if (trns->dst_type == VAL_NUMERIC)
716 double *dst = case_num_rw (*c, dst_var);
718 *dst = !out->copy_input ? out->value.f : case_num (*c, src_var);
719 else if (trns->src_vars != trns->dst_vars)
724 char *dst = CHAR_CAST_BUG (char *, case_str_rw (*c, dst_var));
727 if (!out->copy_input)
728 memcpy (dst, out->value.s, var_get_width (dst_var));
729 else if (trns->src_vars != trns->dst_vars)
731 union value *dst_data = case_data_rw (*c, dst_var);
732 const union value *src_data = case_data (*c, src_var);
733 value_copy_rpad (dst_data, var_get_width (dst_var),
734 src_data, var_get_width (src_var), ' ');
737 else if (trns->src_vars != trns->dst_vars)
738 memset (dst, ' ', var_get_width (dst_var));
742 return TRNS_CONTINUE;
745 /* Frees a RECODE transformation. */
747 recode_trns_free (void *trns_)
749 struct recode_trns *trns = trns_;
750 pool_destroy (trns->pool);
754 static const struct trns_class recode_trns_class = {
756 .execute = recode_trns_proc,
757 .destroy = recode_trns_free,