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))
167 /* Create destination variables, if needed.
168 This must be the final step; otherwise we'd have to
169 delete destination variables on failure. */
170 trns->dst_dict = dict;
171 if (trns->src_vars != trns->dst_vars)
172 create_dst_vars (trns, dict);
175 add_transformation (ds,
176 recode_trns_proc, recode_trns_free, trns);
178 while (lex_match (lexer, T_SLASH));
183 /* Parses a set of variables to recode into TRNS->src_vars and
184 TRNS->var_cnt. Sets TRNS->src_type. Returns true if
185 successful, false on parse error. */
187 parse_src_vars (struct lexer *lexer,
188 struct recode_trns *trns, const struct dictionary *dict)
190 if (!parse_variables_const (lexer, dict, &trns->src_vars, &trns->var_cnt,
193 pool_register (trns->pool, free, trns->src_vars);
194 trns->src_type = var_get_type (trns->src_vars[0]);
198 /* Parses a set of mappings, which take the form (input=output),
199 into TRNS->mappings and TRNS->map_cnt. Sets TRNS->dst_type.
200 Returns true if successful, false on parse error. */
202 parse_mappings (struct lexer *lexer, struct recode_trns *trns,
203 const char *dict_encoding)
205 size_t map_allocated;
209 /* Find length of longest source variable. */
210 trns->max_src_width = var_get_width (trns->src_vars[0]);
211 for (i = 1; i < trns->var_cnt; i++)
213 size_t var_width = var_get_width (trns->src_vars[i]);
214 if (var_width > trns->max_src_width)
215 trns->max_src_width = var_width;
218 /* Parse the mappings in parentheses. */
219 trns->mappings = NULL;
222 have_dst_type = false;
223 if (!lex_force_match (lexer, T_LPAREN))
227 enum val_type dst_type;
229 if (!lex_match_id (lexer, "CONVERT"))
232 size_t first_map_idx;
235 first_map_idx = trns->map_cnt;
237 /* Parse source specifications. */
242 if (!parse_map_in (lexer, &in, trns->pool,
243 trns->src_type, trns->max_src_width,
246 add_mapping (trns, &map_allocated, &in);
247 lex_match (lexer, T_COMMA);
249 while (!lex_match (lexer, T_EQUALS));
251 if (!parse_map_out (lexer, trns->pool, &out))
255 dst_type = trns->src_type;
257 dst_type = val_type_from_width (out.width);
258 if (have_dst_type && dst_type != trns->dst_type)
260 msg (SE, _("Inconsistent target variable types. "
262 "must be all numeric or all string."));
266 for (i = first_map_idx; i < trns->map_cnt; i++)
267 trns->mappings[i].out = out;
271 /* Parse CONVERT as a special case. */
273 set_map_in_generic (&in, MAP_CONVERT);
274 add_mapping (trns, &map_allocated, &in);
275 set_map_out_num (&trns->mappings[trns->map_cnt - 1].out, 0.0);
277 dst_type = VAL_NUMERIC;
278 if (trns->src_type != VAL_STRING
279 || (have_dst_type && trns->dst_type != VAL_NUMERIC))
281 msg (SE, _("CONVERT requires string input values and "
282 "numeric output values."));
286 trns->dst_type = dst_type;
287 have_dst_type = true;
289 if (!lex_force_match (lexer, T_RPAREN))
292 while (lex_match (lexer, T_LPAREN));
297 /* Parses a mapping input value into IN, allocating memory from
298 POOL. The source value type must be provided as SRC_TYPE and,
299 if string, the maximum width of a string source variable must
300 be provided in MAX_SRC_WIDTH. Returns true if successful,
301 false on parse error. */
303 parse_map_in (struct lexer *lexer, struct map_in *in, struct pool *pool,
304 enum val_type src_type, size_t max_src_width,
305 const char *dict_encoding)
308 if (lex_match_id (lexer, "ELSE"))
309 set_map_in_generic (in, MAP_ELSE);
310 else if (src_type == VAL_NUMERIC)
312 if (lex_match_id (lexer, "MISSING"))
313 set_map_in_generic (in, MAP_MISSING);
314 else if (lex_match_id (lexer, "SYSMIS"))
315 set_map_in_generic (in, MAP_SYSMIS);
319 if (!parse_num_range (lexer, &x, &y, NULL))
321 set_map_in_num (in, x == y ? MAP_SINGLE : MAP_RANGE, x, y);
326 if (lex_match_id (lexer, "MISSING"))
327 set_map_in_generic (in, MAP_MISSING);
328 else if (!lex_force_string (lexer))
332 set_map_in_str (in, pool, lex_tokss (lexer), max_src_width,
335 if (lex_token (lexer) == T_ID
336 && lex_id_match (ss_cstr ("THRU"), lex_tokss (lexer)))
338 msg (SE, _("THRU is not allowed with string variables."));
347 /* Adds IN to the list of mappings in TRNS.
348 MAP_ALLOCATED is the current number of allocated mappings,
349 which is updated as needed. */
351 add_mapping (struct recode_trns *trns,
352 size_t *map_allocated, const struct map_in *in)
355 if (trns->map_cnt >= *map_allocated)
356 trns->mappings = pool_2nrealloc (trns->pool, trns->mappings,
358 sizeof *trns->mappings);
359 m = &trns->mappings[trns->map_cnt++];
363 /* Sets IN as a mapping of the given TYPE. */
365 set_map_in_generic (struct map_in *in, enum map_in_type type)
370 /* Sets IN as a numeric mapping of the given TYPE,
371 with X and Y as the two numeric values. */
373 set_map_in_num (struct map_in *in, enum map_in_type type, double x, double y)
380 /* Sets IN as a string mapping, with STRING as the string,
381 allocated from POOL. The string is padded with spaces on the
382 right to WIDTH characters long. */
384 set_map_in_str (struct map_in *in, struct pool *pool,
385 struct substring string, size_t width,
386 const char *dict_encoding)
388 char *s = recode_string (dict_encoding, "UTF-8",
389 ss_data (string), ss_length (string));
390 in->type = MAP_SINGLE;
391 value_init_pool (pool, &in->x, width);
392 value_copy_buf_rpad (&in->x, width,
393 CHAR_CAST (uint8_t *, s), strlen (s), ' ');
397 /* Parses a mapping output value into OUT, allocating memory from
398 POOL. Returns true if successful, false on parse error. */
400 parse_map_out (struct lexer *lexer, struct pool *pool, struct map_out *out)
402 if (lex_is_number (lexer))
404 set_map_out_num (out, lex_number (lexer));
407 else if (lex_match_id (lexer, "SYSMIS"))
408 set_map_out_num (out, SYSMIS);
409 else if (lex_is_string (lexer))
411 set_map_out_str (out, pool, lex_tokss (lexer));
414 else if (lex_match_id (lexer, "COPY"))
416 out->copy_input = true;
421 lex_error (lexer, _("expecting output value"));
427 /* Sets OUT as a numeric mapping output with the given VALUE. */
429 set_map_out_num (struct map_out *out, double value)
431 out->copy_input = false;
432 out->value.f = value;
436 /* Sets OUT as a string mapping output with the given VALUE. */
438 set_map_out_str (struct map_out *out, struct pool *pool,
439 const struct substring value)
441 const char *string = ss_data (value);
442 size_t length = ss_length (value);
446 /* A length of 0 will yield a numeric value, which is not
452 out->copy_input = false;
453 value_init_pool (pool, &out->value, length);
454 memcpy (value_str_rw (&out->value, length), string, length);
458 /* Parses a set of target variables into TRNS->dst_vars and
461 parse_dst_vars (struct lexer *lexer, struct recode_trns *trns,
462 const struct dictionary *dict)
466 if (lex_match_id (lexer, "INTO"))
471 if (!parse_mixed_vars_pool (lexer, dict, trns->pool,
472 &trns->dst_names, &name_cnt,
476 if (name_cnt != trns->var_cnt)
478 msg (SE, _("%zu variable(s) cannot be recoded into "
479 "%zu variable(s). Specify the same number "
480 "of variables as source and target variables."),
481 trns->var_cnt, name_cnt);
485 trns->dst_vars = pool_nalloc (trns->pool,
486 trns->var_cnt, sizeof *trns->dst_vars);
487 for (i = 0; i < trns->var_cnt; i++)
489 const struct variable *v;
490 v = trns->dst_vars[i] = dict_lookup_var (dict, trns->dst_names[i]);
491 if (v == NULL && trns->dst_type == VAL_STRING)
493 msg (SE, _("There is no variable named "
494 "%s. (All string variables specified "
495 "on INTO must already exist. Use the "
496 "STRING command to create a string "
506 trns->dst_vars = trns->src_vars;
507 if (trns->src_type != trns->dst_type)
509 msg (SE, _("INTO is required with %s input values "
510 "and %s output values."),
511 trns->src_type == VAL_NUMERIC ? _("numeric") : _("string"),
512 trns->dst_type == VAL_NUMERIC ? _("numeric") : _("string"));
517 for (i = 0; i < trns->var_cnt; i++)
519 const struct variable *v = trns->dst_vars[i];
520 if (v != NULL && var_get_type (v) != trns->dst_type)
522 msg (SE, _("Type mismatch. Cannot store %s data in "
524 trns->dst_type == VAL_STRING ? _("string") : _("numeric"),
525 var_is_alpha (v) ? _("string") : _("numeric"),
534 /* Ensures that all the output values in TRNS are as wide as the
535 widest destination variable. */
537 enlarge_dst_widths (struct recode_trns *trns)
540 const struct variable *narrow_var = NULL;
541 int min_dst_width = INT_MAX;
542 trns->max_dst_width = 0;
544 for (i = 0; i < trns->var_cnt; i++)
546 const struct variable *v = trns->dst_vars[i];
547 if (var_get_width (v) > trns->max_dst_width)
548 trns->max_dst_width = var_get_width (v);
550 if (var_get_width (v) < min_dst_width)
552 min_dst_width = var_get_width (v);
557 for (i = 0; i < trns->map_cnt; i++)
559 struct map_out *out = &trns->mappings[i].out;
560 if (!out->copy_input)
562 if (out->width > min_dst_width)
565 _("Cannot recode because the variable %s would require a width of %d bytes or greater, but it has a width of only %d bytes."),
566 var_get_name (narrow_var), out->width, min_dst_width);
570 value_resize_pool (trns->pool, &out->value,
571 out->width, trns->max_dst_width);
578 /* Creates destination variables that don't already exist. */
580 create_dst_vars (struct recode_trns *trns, struct dictionary *dict)
584 for (i = 0; i < trns->var_cnt; i++)
586 const struct variable **var = &trns->dst_vars[i];
587 const char *name = trns->dst_names[i];
589 *var = dict_lookup_var (dict, name);
591 *var = dict_create_var_assert (dict, name, 0);
592 assert (var_get_type (*var) == trns->dst_type);
596 /* Data transformation. */
598 /* Returns the output mapping in TRNS for an input of VALUE on
599 variable V, or a null pointer if there is no mapping. */
600 static const struct map_out *
601 find_src_numeric (struct recode_trns *trns, double value, const struct variable *v)
605 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
607 const struct map_in *in = &m->in;
608 const struct map_out *out = &m->out;
614 match = value == in->x.f;
617 match = var_is_num_missing (v, value, MV_ANY);
620 match = value >= in->x.f && value <= in->y.f;
623 match = value == SYSMIS;
639 /* Returns the output mapping in TRNS for an input of VALUE with
640 the given WIDTH, or a null pointer if there is no mapping. */
641 static const struct map_out *
642 find_src_string (struct recode_trns *trns, const uint8_t *value,
643 const struct variable *src_var)
645 const char *encoding = dict_get_encoding (trns->dst_dict);
646 int width = var_get_width (src_var);
649 for (m = trns->mappings; m < trns->mappings + trns->map_cnt; m++)
651 const struct map_in *in = &m->in;
652 struct map_out *out = &m->out;
658 match = !memcmp (value, value_str (&in->x, trns->max_src_width),
669 error = data_in (ss_buffer (CHAR_CAST_BUG (char *, value), width),
670 C_ENCODING, FMT_F, &uv, 0, encoding);
671 match = error == NULL;
678 match = var_is_str_missing (src_var, value, MV_ANY);
691 /* Performs RECODE transformation. */
693 recode_trns_proc (void *trns_, struct ccase **c, casenumber case_idx UNUSED)
695 struct recode_trns *trns = trns_;
698 *c = case_unshare (*c);
699 for (i = 0; i < trns->var_cnt; i++)
701 const struct variable *src_var = trns->src_vars[i];
702 const struct variable *dst_var = trns->dst_vars[i];
703 const struct map_out *out;
705 if (trns->src_type == VAL_NUMERIC)
706 out = find_src_numeric (trns, case_num (*c, src_var), src_var);
708 out = find_src_string (trns, case_str (*c, src_var), src_var);
710 if (trns->dst_type == VAL_NUMERIC)
712 double *dst = &case_data_rw (*c, dst_var)->f;
714 *dst = !out->copy_input ? out->value.f : case_num (*c, src_var);
715 else if (trns->src_vars != trns->dst_vars)
720 char *dst = CHAR_CAST_BUG (char *, case_str_rw (*c, dst_var));
723 if (!out->copy_input)
724 memcpy (dst, value_str (&out->value, trns->max_dst_width),
725 var_get_width (dst_var));
726 else if (trns->src_vars != trns->dst_vars)
728 union value *dst_data = case_data_rw (*c, dst_var);
729 const union value *src_data = case_data (*c, src_var);
730 value_copy_rpad (dst_data, var_get_width (dst_var),
731 src_data, var_get_width (src_var), ' ');
734 else if (trns->src_vars != trns->dst_vars)
735 memset (dst, ' ', var_get_width (dst_var));
739 return TRNS_CONTINUE;
742 /* Frees a RECODE transformation. */
744 recode_trns_free (void *trns_)
746 struct recode_trns *trns = trns_;
747 pool_destroy (trns->pool);