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 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 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 trns_proc_func recode_trns_proc;
132 static trns_free_func recode_trns_free;
136 /* Parses the RECODE transformation. */
138 cmd_recode (struct lexer *lexer, struct dataset *ds)
142 struct dictionary *dict = dataset_dict (ds);
143 struct recode_trns *trns
144 = pool_create_container (struct recode_trns, pool);
146 /* Parse source variable names,
147 then input to output mappings,
148 then destintation variable names. */
149 if (!parse_src_vars (lexer, trns, dict)
150 || !parse_mappings (lexer, trns, dict_get_encoding (dict))
151 || !parse_dst_vars (lexer, trns, dict))
153 recode_trns_free (trns);
157 /* Ensure that all the output strings are at least as wide
158 as the widest destination variable. */
159 if (trns->dst_type == VAL_STRING)
161 if ( ! enlarge_dst_widths (trns))
163 recode_trns_free (trns);
168 /* Create destination variables, if needed.
169 This must be the final step; otherwise we'd have to
170 delete destination variables on failure. */
171 trns->dst_dict = dict;
172 if (trns->src_vars != trns->dst_vars)
173 create_dst_vars (trns, dict);
176 add_transformation (ds,
177 recode_trns_proc, recode_trns_free, trns);
179 while (lex_match (lexer, T_SLASH));
184 /* Parses a set of variables to recode into TRNS->src_vars and
185 TRNS->var_cnt. 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->var_cnt,
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->map_cnt. 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->var_cnt; 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->map_cnt;
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->map_cnt; 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->map_cnt - 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 msg (SE, _("CONVERT requires string input values and "
283 "numeric output values."));
287 trns->dst_type = dst_type;
288 have_dst_type = true;
290 if (!lex_force_match (lexer, T_RPAREN))
293 while (lex_match (lexer, T_LPAREN));
298 /* Parses a mapping input value into IN, allocating memory from
299 POOL. The source value type must be provided as SRC_TYPE and,
300 if string, the maximum width of a string source variable must
301 be provided in MAX_SRC_WIDTH. Returns true if successful,
302 false on parse error. */
304 parse_map_in (struct lexer *lexer, struct map_in *in, struct pool *pool,
305 enum val_type src_type, size_t max_src_width,
306 const char *dict_encoding)
309 if (lex_match_id (lexer, "ELSE"))
310 set_map_in_generic (in, MAP_ELSE);
311 else if (src_type == VAL_NUMERIC)
313 if (lex_match_id (lexer, "MISSING"))
314 set_map_in_generic (in, MAP_MISSING);
315 else if (lex_match_id (lexer, "SYSMIS"))
316 set_map_in_generic (in, MAP_SYSMIS);
320 if (!parse_num_range (lexer, &x, &y, NULL))
322 set_map_in_num (in, x == y ? MAP_SINGLE : MAP_RANGE, x, y);
327 if (lex_match_id (lexer, "MISSING"))
328 set_map_in_generic (in, MAP_MISSING);
329 else if (!lex_force_string (lexer))
333 set_map_in_str (in, pool, lex_tokss (lexer), max_src_width,
336 if (lex_token (lexer) == T_ID
337 && lex_id_match (ss_cstr ("THRU"), lex_tokss (lexer)))
339 msg (SE, _("THRU is not allowed with string variables."));
348 /* Adds IN to the list of mappings in TRNS.
349 MAP_ALLOCATED is the current number of allocated mappings,
350 which is updated as needed. */
352 add_mapping (struct recode_trns *trns,
353 size_t *map_allocated, const struct map_in *in)
356 if (trns->map_cnt >= *map_allocated)
357 trns->mappings = pool_2nrealloc (trns->pool, trns->mappings,
359 sizeof *trns->mappings);
360 m = &trns->mappings[trns->map_cnt++];
364 /* Sets IN as a mapping of the given TYPE. */
366 set_map_in_generic (struct map_in *in, enum map_in_type type)
371 /* Sets IN as a numeric mapping of the given TYPE,
372 with X and Y as the two numeric values. */
374 set_map_in_num (struct map_in *in, enum map_in_type type, double x, double y)
381 /* Sets IN as a string mapping, with STRING as the string,
382 allocated from POOL. The string is padded with spaces on the
383 right to WIDTH characters long. */
385 set_map_in_str (struct map_in *in, struct pool *pool,
386 struct substring string, size_t width,
387 const char *dict_encoding)
389 char *s = recode_string (dict_encoding, "UTF-8",
390 ss_data (string), ss_length (string));
391 in->type = MAP_SINGLE;
392 value_init_pool (pool, &in->x, width);
393 value_copy_buf_rpad (&in->x, width,
394 CHAR_CAST (uint8_t *, s), strlen (s), ' ');
398 /* Parses a mapping output value into OUT, allocating memory from
399 POOL. Returns true if successful, false on parse error. */
401 parse_map_out (struct lexer *lexer, struct pool *pool, struct map_out *out)
403 if (lex_is_number (lexer))
405 set_map_out_num (out, lex_number (lexer));
408 else if (lex_match_id (lexer, "SYSMIS"))
409 set_map_out_num (out, SYSMIS);
410 else if (lex_is_string (lexer))
412 set_map_out_str (out, pool, lex_tokss (lexer));
415 else if (lex_match_id (lexer, "COPY"))
417 out->copy_input = true;
422 lex_error (lexer, _("expecting output value"));
428 /* Sets OUT as a numeric mapping output with the given VALUE. */
430 set_map_out_num (struct map_out *out, double value)
432 out->copy_input = false;
433 out->value.f = value;
437 /* Sets OUT as a string mapping output with the given VALUE. */
439 set_map_out_str (struct map_out *out, struct pool *pool,
440 const struct substring value)
442 const char *string = ss_data (value);
443 size_t length = ss_length (value);
447 /* A length of 0 will yield a numeric value, which is not
453 out->copy_input = false;
454 value_init_pool (pool, &out->value, length);
455 memcpy (value_str_rw (&out->value, length), string, length);
459 /* Parses a set of target variables into TRNS->dst_vars and
462 parse_dst_vars (struct lexer *lexer, struct recode_trns *trns,
463 const struct dictionary *dict)
467 if (lex_match_id (lexer, "INTO"))
472 if (!parse_mixed_vars_pool (lexer, dict, trns->pool,
473 &trns->dst_names, &name_cnt,
477 if (name_cnt != trns->var_cnt)
479 msg (SE, _("%zu variable(s) cannot be recoded into "
480 "%zu variable(s). Specify the same number "
481 "of variables as source and target variables."),
482 trns->var_cnt, name_cnt);
486 trns->dst_vars = pool_nalloc (trns->pool,
487 trns->var_cnt, sizeof *trns->dst_vars);
488 for (i = 0; i < trns->var_cnt; i++)
490 const struct variable *v;
491 v = trns->dst_vars[i] = dict_lookup_var (dict, trns->dst_names[i]);
492 if (v == NULL && trns->dst_type == VAL_STRING)
494 msg (SE, _("There is no variable named "
495 "%s. (All string variables specified "
496 "on INTO must already exist. Use the "
497 "STRING command to create a string "
507 trns->dst_vars = trns->src_vars;
508 if (trns->src_type != trns->dst_type)
510 msg (SE, _("INTO is required with %s input values "
511 "and %s output values."),
512 trns->src_type == VAL_NUMERIC ? _("numeric") : _("string"),
513 trns->dst_type == VAL_NUMERIC ? _("numeric") : _("string"));
518 for (i = 0; i < trns->var_cnt; i++)
520 const struct variable *v = trns->dst_vars[i];
521 if (v != NULL && var_get_type (v) != trns->dst_type)
523 msg (SE, _("Type mismatch. Cannot store %s data in "
525 trns->dst_type == VAL_STRING ? _("string") : _("numeric"),
526 var_is_alpha (v) ? _("string") : _("numeric"),
535 /* Ensures that all the output values in TRNS are as wide as the
536 widest destination variable. */
538 enlarge_dst_widths (struct recode_trns *trns)
541 const struct variable *narrow_var = NULL;
542 int min_dst_width = INT_MAX;
543 trns->max_dst_width = 0;
545 for (i = 0; i < trns->var_cnt; i++)
547 const struct variable *v = trns->dst_vars[i];
548 if (var_get_width (v) > trns->max_dst_width)
549 trns->max_dst_width = var_get_width (v);
551 if (var_get_width (v) < min_dst_width)
553 min_dst_width = var_get_width (v);
558 for (i = 0; i < trns->map_cnt; i++)
560 struct map_out *out = &trns->mappings[i].out;
561 if (!out->copy_input)
563 if (out->width > min_dst_width)
566 _("Cannot recode because the variable %s would require a width of %d bytes or greater, but it has a width of only %d bytes."),
567 var_get_name (narrow_var), out->width, min_dst_width);
571 value_resize_pool (trns->pool, &out->value,
572 out->width, trns->max_dst_width);
579 /* Creates destination variables that don't already exist. */
581 create_dst_vars (struct recode_trns *trns, struct dictionary *dict)
585 for (i = 0; i < trns->var_cnt; i++)
587 const struct variable **var = &trns->dst_vars[i];
588 const char *name = trns->dst_names[i];
590 *var = dict_lookup_var (dict, name);
592 *var = dict_create_var_assert (dict, name, 0);
593 assert (var_get_type (*var) == trns->dst_type);
597 /* Data transformation. */
599 /* Returns the output mapping in TRNS for an input of VALUE on
600 variable V, or a null pointer if there is no mapping. */
601 static const struct map_out *
602 find_src_numeric (struct recode_trns *trns, double value, const struct variable *v)
606 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
608 const struct map_in *in = &m->in;
609 const struct map_out *out = &m->out;
615 match = value == in->x.f;
618 match = var_is_num_missing (v, value, MV_ANY);
621 match = value >= in->x.f && value <= in->y.f;
624 match = value == SYSMIS;
640 /* Returns the output mapping in TRNS for an input of VALUE with
641 the given WIDTH, or a null pointer if there is no mapping. */
642 static const struct map_out *
643 find_src_string (struct recode_trns *trns, const uint8_t *value,
644 const struct variable *src_var)
646 const char *encoding = dict_get_encoding (trns->dst_dict);
647 int width = var_get_width (src_var);
650 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
652 const struct map_in *in = &m->in;
653 struct map_out *out = &m->out;
659 match = !memcmp (value, value_str (&in->x, trns->max_src_width),
670 error = data_in (ss_buffer (CHAR_CAST_BUG (char *, value), width),
671 C_ENCODING, FMT_F, &uv, 0, encoding);
672 match = error == NULL;
679 match = var_is_str_missing (src_var, value, MV_ANY);
692 /* Performs RECODE transformation. */
694 recode_trns_proc (void *trns_, struct ccase **c, casenumber case_idx UNUSED)
696 struct recode_trns *trns = trns_;
699 *c = case_unshare (*c);
700 for (i = 0; i < trns->var_cnt; i++)
702 const struct variable *src_var = trns->src_vars[i];
703 const struct variable *dst_var = trns->dst_vars[i];
704 const struct map_out *out;
706 if (trns->src_type == VAL_NUMERIC)
707 out = find_src_numeric (trns, case_num (*c, src_var), src_var);
709 out = find_src_string (trns, case_str (*c, src_var), src_var);
711 if (trns->dst_type == VAL_NUMERIC)
713 double *dst = &case_data_rw (*c, dst_var)->f;
715 *dst = !out->copy_input ? out->value.f : case_num (*c, src_var);
716 else if (trns->src_vars != trns->dst_vars)
721 char *dst = CHAR_CAST_BUG (char *, case_str_rw (*c, dst_var));
724 if (!out->copy_input)
725 memcpy (dst, value_str (&out->value, trns->max_dst_width),
726 var_get_width (dst_var));
727 else if (trns->src_vars != trns->dst_vars)
729 union value *dst_data = case_data_rw (*c, dst_var);
730 const union value *src_data = case_data (*c, src_var);
731 value_copy_rpad (dst_data, var_get_width (dst_var),
732 src_data, var_get_width (src_var), ' ');
735 else if (trns->src_vars != trns->dst_vars)
736 memset (dst, ' ', var_get_width (dst_var));
740 return TRNS_CONTINUE;
743 /* Frees a RECODE transformation. */
745 recode_trns_free (void *trns_)
747 struct recode_trns *trns = trns_;
748 pool_destroy (trns->pool);