1 /* PSPP - computes sample statistics.
2 Copyright (C) 1997-9, 2000 Free Software Foundation, Inc.
4 This program is free software; you can redistribute it and/or
5 modify it under the terms of the GNU General Public License as
6 published by the Free Software Foundation; either version 2 of the
7 License, or (at your option) any later version.
9 This program is distributed in the hope that it will be useful, but
10 WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 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, write to the Free Software
16 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
21 #include <language/lexer/variable-parser.h>
28 #include <data/dictionary.h>
29 #include <data/procedure.h>
30 #include <data/variable.h>
31 #include <libpspp/alloc.h>
32 #include <libpspp/assertion.h>
33 #include <libpspp/bit-vector.h>
34 #include <libpspp/hash.h>
35 #include <libpspp/message.h>
36 #include <libpspp/misc.h>
37 #include <libpspp/pool.h>
38 #include <libpspp/str.h>
41 #define _(msgid) gettext (msgid)
43 static struct variable * var_set_get_var (const struct var_set *, size_t );
45 static struct variable *var_set_lookup_var (const struct var_set *,
48 static bool var_set_lookup_var_idx (const struct var_set *, const char *,
53 /* Parses a name as a variable within VS. Sets *IDX to the
54 variable's index and returns true if successful. On failure
55 emits an error message and returns false. */
57 parse_vs_variable_idx (struct lexer *lexer, const struct var_set *vs,
62 if (lex_token (lexer) != T_ID)
64 lex_error (lexer, _("expecting variable name"));
67 else if (var_set_lookup_var_idx (vs, lex_tokid (lexer), idx))
74 msg (SE, _("%s is not a variable name."), lex_tokid (lexer));
79 /* Parses a name as a variable within VS and returns the variable
80 if successful. On failure emits an error message and returns
82 static struct variable *
83 parse_vs_variable (struct lexer *lexer, const struct var_set *vs)
86 return parse_vs_variable_idx (lexer, vs, &idx) ? var_set_get_var (vs, idx) : NULL;
89 /* Parses a variable name in dictionary D and returns the
90 variable if successful. On failure emits an error message and
91 returns a null pointer. */
93 parse_variable (struct lexer *lexer, const struct dictionary *d)
95 struct var_set *vs = var_set_create_from_dict (d);
96 struct variable *var = parse_vs_variable (lexer, vs);
101 /* Parses a set of variables from dictionary D given options
102 OPTS. Resulting list of variables stored in *VAR and the
103 number of variables into *CNT. Returns true only if
106 parse_variables (struct lexer *lexer, const struct dictionary *d,
107 struct variable ***var,
108 size_t *cnt, int opts)
114 assert (var != NULL);
115 assert (cnt != NULL);
117 vs = var_set_create_from_dict (d);
118 success = parse_var_set_vars (lexer, vs, var, cnt, opts);
125 var_set_destroy (vs);
129 /* Parses a set of variables from dictionary D given options
130 OPTS. Resulting list of variables stored in *VARS and the
131 number of variables into *VAR_CNT. Returns true only if
132 successful. Same behavior as parse_variables, except that all
133 allocations are taken from the given POOL. */
135 parse_variables_pool (struct lexer *lexer, struct pool *pool,
136 const struct dictionary *dict,
137 struct variable ***vars, size_t *var_cnt, int opts)
141 /* PV_APPEND is unsafe because parse_variables would free the
142 existing names on failure, but those names are presumably
143 already in the pool, which would attempt to re-free it
145 assert (!(opts & PV_APPEND));
147 retval = parse_variables (lexer, dict, vars, var_cnt, opts);
149 pool_register (pool, free, *vars);
153 /* Parses a variable name from VS. If successful, sets *IDX to
154 the variable's index in VS, *CLASS to the variable's
155 dictionary class, and returns true. Returns false on
158 parse_var_idx_class (struct lexer *lexer, const struct var_set *vs,
160 enum dict_class *class)
162 if (!parse_vs_variable_idx (lexer, vs, idx))
165 *class = dict_class_from_id (var_get_name (var_set_get_var (vs, *idx)));
169 /* Add the variable from VS with index IDX to the list of
170 variables V that has *NV elements and room for *MV.
171 Uses and updates INCLUDED to avoid duplicates if indicated by
172 PV_OPTS, which also affects what variables are allowed in
175 add_variable (struct variable ***v, size_t *nv, size_t *mv,
176 char *included, int pv_opts,
177 const struct var_set *vs, size_t idx)
179 struct variable *add = var_set_get_var (vs, idx);
180 const char *add_name = var_get_name (add);
182 if ((pv_opts & PV_NUMERIC) && !var_is_numeric (add))
183 msg (SW, _("%s is not a numeric variable. It will not be "
184 "included in the variable list."), add_name);
185 else if ((pv_opts & PV_STRING) && !var_is_alpha (add))
186 msg (SE, _("%s is not a string variable. It will not be "
187 "included in the variable list."), add_name);
188 else if ((pv_opts & PV_NO_SCRATCH)
189 && dict_class_from_id (add_name) == DC_SCRATCH)
190 msg (SE, _("Scratch variables (such as %s) are not allowed "
192 else if ((pv_opts & (PV_SAME_TYPE | PV_SAME_WIDTH)) && *nv
193 && var_get_type (add) != var_get_type ((*v)[0]))
194 msg (SE, _("%s and %s are not the same type. All variables in "
195 "this variable list must be of the same type. %s "
196 "will be omitted from the list."),
197 var_get_name ((*v)[0]), add_name, add_name);
198 else if ((pv_opts & PV_SAME_WIDTH) && *nv
199 && var_get_width (add) != var_get_width ((*v)[0]))
200 msg (SE, _("%s and %s are string variables with different widths. "
201 "All variables in this variable list must have the "
202 "same width. %s will be omttied from the list."),
203 var_get_name ((*v)[0]), add_name, add_name);
204 else if ((pv_opts & PV_NO_DUPLICATE) && included[idx])
205 msg (SE, _("Variable %s appears twice in variable list."), add_name);
206 else if ((pv_opts & PV_DUPLICATE) || !included[idx])
211 *v = xnrealloc (*v, *mv, sizeof **v);
214 if (included != NULL)
219 /* Adds the variables in VS with indexes FIRST_IDX through
220 LAST_IDX, inclusive, to the list of variables V that has *NV
221 elements and room for *MV. Uses and updates INCLUDED to avoid
222 duplicates if indicated by PV_OPTS, which also affects what
223 variables are allowed in appropriate ways. */
225 add_variables (struct variable ***v, size_t *nv, size_t *mv, char *included,
227 const struct var_set *vs, int first_idx, int last_idx,
228 enum dict_class class)
232 for (i = first_idx; i <= last_idx; i++)
233 if (dict_class_from_id (var_get_name (var_set_get_var (vs, i))) == class)
234 add_variable (v, nv, mv, included, pv_opts, vs, i);
237 /* Note that if parse_variables() returns false, *v is free()'d.
238 Conversely, if parse_variables() returns true, then *nv is
239 nonzero and *v is non-NULL. */
241 parse_var_set_vars (struct lexer *lexer, const struct var_set *vs,
242 struct variable ***v, size_t *nv,
252 /* At most one of PV_NUMERIC, PV_STRING, PV_SAME_TYPE,
253 PV_SAME_WIDTH may be specified. */
254 assert (((pv_opts & PV_NUMERIC) != 0)
255 + ((pv_opts & PV_STRING) != 0)
256 + ((pv_opts & PV_SAME_TYPE) != 0)
257 + ((pv_opts & PV_SAME_WIDTH) != 0) <= 1);
259 /* PV_DUPLICATE and PV_NO_DUPLICATE are incompatible. */
260 assert (!(pv_opts & PV_DUPLICATE) || !(pv_opts & PV_NO_DUPLICATE));
262 if (!(pv_opts & PV_APPEND))
271 if (!(pv_opts & PV_DUPLICATE))
275 included = xcalloc (var_set_get_cnt (vs), sizeof *included);
276 for (i = 0; i < *nv; i++)
279 if (!var_set_lookup_var_idx (vs, var_get_name ((*v)[i]), &index))
289 if (lex_match (lexer, T_ALL))
290 add_variables (v, nv, &mv, included, pv_opts,
291 vs, 0, var_set_get_cnt (vs) - 1, DC_ORDINARY);
294 enum dict_class class;
297 if (!parse_var_idx_class (lexer, vs, &first_idx, &class))
300 if (!lex_match (lexer, T_TO))
301 add_variable (v, nv, &mv, included, pv_opts, vs, first_idx);
305 enum dict_class last_class;
306 struct variable *first_var, *last_var;
308 if (!parse_var_idx_class (lexer, vs, &last_idx, &last_class))
311 first_var = var_set_get_var (vs, first_idx);
312 last_var = var_set_get_var (vs, last_idx);
314 if (last_idx < first_idx)
316 const char *first_name = var_get_name (first_var);
317 const char *last_name = var_get_name (last_var);
318 msg (SE, _("%s TO %s is not valid syntax since %s "
319 "precedes %s in the dictionary."),
320 first_name, last_name, first_name, last_name);
324 if (class != last_class)
326 msg (SE, _("When using the TO keyword to specify several "
327 "variables, both variables must be from "
328 "the same variable dictionaries, of either "
329 "ordinary, scratch, or system variables. "
330 "%s is a %s variable, whereas %s is %s."),
331 var_get_name (first_var), dict_class_to_name (class),
332 var_get_name (last_var),
333 dict_class_to_name (last_class));
337 add_variables (v, nv, &mv, included, pv_opts,
338 vs, first_idx, last_idx, class);
342 if (pv_opts & PV_SINGLE)
344 lex_match (lexer, ',');
346 while (lex_token (lexer) == T_ALL
347 || (lex_token (lexer) == T_ID && var_set_lookup_var (vs, lex_tokid (lexer)) != NULL));
363 /* Extracts a numeric suffix from variable name S, copying it
364 into string R. Sets *D to the length of R and *N to its
367 extract_num (char *s, char *r, int *n, int *d)
371 /* Find first digit. */
372 cp = s + strlen (s) - 1;
373 while (isdigit ((unsigned char) *cp) && cp > s)
378 strncpy (r, s, cp - s);
381 /* Count initial zeros. */
390 while (isdigit ((unsigned char) *cp))
393 *n = (*n * 10) + (*cp - '0');
398 if (*n == 0 && *d == 0)
400 msg (SE, _("incorrect use of TO convention"));
406 /* Parses a list of variable names according to the DATA LIST version
407 of the TO convention. */
409 parse_DATA_LIST_vars (struct lexer *lexer, char ***names, size_t *nnames, int pv_opts)
415 char name1[LONG_NAME_LEN + 1], name2[LONG_NAME_LEN + 1];
416 char root1[LONG_NAME_LEN + 1], root2[LONG_NAME_LEN + 1];
419 assert (names != NULL);
420 assert (nnames != NULL);
421 assert ((pv_opts & ~(PV_APPEND | PV_SINGLE
422 | PV_NO_SCRATCH | PV_NO_DUPLICATE)) == 0);
423 /* FIXME: PV_NO_DUPLICATE is not implemented. */
425 if (pv_opts & PV_APPEND)
426 nvar = mvar = *nnames;
435 if (lex_token (lexer) != T_ID)
437 lex_error (lexer, "expecting variable name");
440 if (dict_class_from_id (lex_tokid (lexer)) == DC_SCRATCH
441 && (pv_opts & PV_NO_SCRATCH))
443 msg (SE, _("Scratch variables not allowed here."));
446 strcpy (name1, lex_tokid (lexer));
448 if (lex_token (lexer) == T_TO)
451 if (lex_token (lexer) != T_ID)
453 lex_error (lexer, "expecting variable name");
456 strcpy (name2, lex_tokid (lexer));
459 if (!extract_num (name1, root1, &n1, &d1)
460 || !extract_num (name2, root2, &n2, &d2))
463 if (strcasecmp (root1, root2))
465 msg (SE, _("Prefixes don't match in use of TO convention."));
470 msg (SE, _("Bad bounds in use of TO convention."));
476 if (mvar < nvar + (n2 - n1 + 1))
478 mvar += ROUND_UP (n2 - n1 + 1, 16);
479 *names = xnrealloc (*names, mvar, sizeof **names);
482 for (n = n1; n <= n2; n++)
484 char name[LONG_NAME_LEN + 1];
485 sprintf (name, "%s%0*d", root1, d1, n);
486 (*names)[nvar] = xstrdup (name);
495 *names = xnrealloc (*names, mvar, sizeof **names);
497 (*names)[nvar++] = xstrdup (name1);
500 lex_match (lexer, ',');
502 if (pv_opts & PV_SINGLE)
505 while (lex_token (lexer) == T_ID);
513 for (i = 0; i < nvar; i++)
522 /* Registers each of the NAMES[0...NNAMES - 1] in POOL, as well
525 register_vars_pool (struct pool *pool, char **names, size_t nnames)
529 for (i = 0; i < nnames; i++)
530 pool_register (pool, free, names[i]);
531 pool_register (pool, free, names);
534 /* Parses a list of variable names according to the DATA LIST
535 version of the TO convention. Same args as
536 parse_DATA_LIST_vars(), except that all allocations are taken
537 from the given POOL. */
539 parse_DATA_LIST_vars_pool (struct lexer *lexer, struct pool *pool,
540 char ***names, size_t *nnames, int pv_opts)
544 /* PV_APPEND is unsafe because parse_DATA_LIST_vars would free
545 the existing names on failure, but those names are
546 presumably already in the pool, which would attempt to
548 assert (!(pv_opts & PV_APPEND));
550 retval = parse_DATA_LIST_vars (lexer, names, nnames, pv_opts);
552 register_vars_pool (pool, *names, *nnames);
556 /* Parses a list of variables where some of the variables may be
557 existing and the rest are to be created. Same args as
558 parse_DATA_LIST_vars(). */
560 parse_mixed_vars (struct lexer *lexer, const struct dictionary *dict,
561 char ***names, size_t *nnames, int pv_opts)
565 assert (names != NULL);
566 assert (nnames != NULL);
567 assert ((pv_opts & ~PV_APPEND) == 0);
569 if (!(pv_opts & PV_APPEND))
574 while (lex_token (lexer) == T_ID || lex_token (lexer) == T_ALL)
576 if (lex_token (lexer) == T_ALL || dict_lookup_var (dict, lex_tokid (lexer)) != NULL)
581 if (!parse_variables (lexer, dict, &v, &nv, PV_NONE))
583 *names = xnrealloc (*names, *nnames + nv, sizeof **names);
584 for (i = 0; i < nv; i++)
585 (*names)[*nnames + i] = xstrdup (var_get_name (v[i]));
589 else if (!parse_DATA_LIST_vars (lexer, names, nnames, PV_APPEND))
595 for (i = 0; i < *nnames; i++)
603 /* Parses a list of variables where some of the variables may be
604 existing and the rest are to be created. Same args as
605 parse_mixed_vars(), except that all allocations are taken
606 from the given POOL. */
608 parse_mixed_vars_pool (struct lexer *lexer, const struct dictionary *dict, struct pool *pool,
609 char ***names, size_t *nnames, int pv_opts)
613 /* PV_APPEND is unsafe because parse_mixed_vars_pool would free
614 the existing names on failure, but those names are
615 presumably already in the pool, which would attempt to
617 assert (!(pv_opts & PV_APPEND));
619 retval = parse_mixed_vars (lexer, dict, names, nnames, pv_opts);
621 register_vars_pool (pool, *names, *nnames);
625 /* A set of variables. */
628 size_t (*get_cnt) (const struct var_set *);
629 struct variable *(*get_var) (const struct var_set *, size_t idx);
630 bool (*lookup_var_idx) (const struct var_set *, const char *, size_t *);
631 void (*destroy) (struct var_set *);
635 /* Returns the number of variables in VS. */
637 var_set_get_cnt (const struct var_set *vs)
641 return vs->get_cnt (vs);
644 /* Return variable with index IDX in VS.
645 IDX must be less than the number of variables in VS. */
646 static struct variable *
647 var_set_get_var (const struct var_set *vs, size_t idx)
650 assert (idx < var_set_get_cnt (vs));
652 return vs->get_var (vs, idx);
655 /* Returns the variable in VS named NAME, or a null pointer if VS
656 contains no variable with that name. */
658 var_set_lookup_var (const struct var_set *vs, const char *name)
661 return (var_set_lookup_var_idx (vs, name, &idx)
662 ? var_set_get_var (vs, idx)
666 /* If VS contains a variable named NAME, sets *IDX to its index
667 and returns true. Otherwise, returns false. */
669 var_set_lookup_var_idx (const struct var_set *vs, const char *name,
673 assert (name != NULL);
674 assert (strlen (name) <= LONG_NAME_LEN);
676 return vs->lookup_var_idx (vs, name, idx);
681 var_set_destroy (struct var_set *vs)
687 /* Returns the number of variables in VS. */
689 dict_var_set_get_cnt (const struct var_set *vs)
691 struct dictionary *d = vs->aux;
693 return dict_get_var_cnt (d);
696 /* Return variable with index IDX in VS.
697 IDX must be less than the number of variables in VS. */
698 static struct variable *
699 dict_var_set_get_var (const struct var_set *vs, size_t idx)
701 struct dictionary *d = vs->aux;
703 return dict_get_var (d, idx);
706 /* If VS contains a variable named NAME, sets *IDX to its index
707 and returns true. Otherwise, returns false. */
709 dict_var_set_lookup_var_idx (const struct var_set *vs, const char *name,
712 struct dictionary *d = vs->aux;
713 struct variable *v = dict_lookup_var (d, name);
716 *idx = var_get_dict_index (v);
725 dict_var_set_destroy (struct var_set *vs)
730 /* Returns a variable set based on D. */
732 var_set_create_from_dict (const struct dictionary *d)
734 struct var_set *vs = xmalloc (sizeof *vs);
735 vs->get_cnt = dict_var_set_get_cnt;
736 vs->get_var = dict_var_set_get_var;
737 vs->lookup_var_idx = dict_var_set_lookup_var_idx;
738 vs->destroy = dict_var_set_destroy;
739 vs->aux = (void *) d;
743 /* A variable set based on an array. */
746 struct variable *const *var;/* Array of variables. */
747 size_t var_cnt; /* Number of elements in var. */
748 struct hsh_table *name_tab; /* Hash from variable names to variables. */
751 /* Returns the number of variables in VS. */
753 array_var_set_get_cnt (const struct var_set *vs)
755 struct array_var_set *avs = vs->aux;
760 /* Return variable with index IDX in VS.
761 IDX must be less than the number of variables in VS. */
762 static struct variable *
763 array_var_set_get_var (const struct var_set *vs, size_t idx)
765 struct array_var_set *avs = vs->aux;
767 return (struct variable *) avs->var[idx];
770 /* If VS contains a variable named NAME, sets *IDX to its index
771 and returns true. Otherwise, returns false. */
773 array_var_set_lookup_var_idx (const struct var_set *vs, const char *name,
776 struct array_var_set *avs = vs->aux;
777 struct variable *v, *const *vpp;
779 v = var_create (name, 0);
780 vpp = hsh_find (avs->name_tab, &v);
785 *idx = vpp - avs->var;
794 array_var_set_destroy (struct var_set *vs)
796 struct array_var_set *avs = vs->aux;
798 hsh_destroy (avs->name_tab);
803 /* Returns a variable set based on the VAR_CNT variables in
806 var_set_create_from_array (struct variable *const *var, size_t var_cnt)
809 struct array_var_set *avs;
812 vs = xmalloc (sizeof *vs);
813 vs->get_cnt = array_var_set_get_cnt;
814 vs->get_var = array_var_set_get_var;
815 vs->lookup_var_idx = array_var_set_lookup_var_idx;
816 vs->destroy = array_var_set_destroy;
817 vs->aux = avs = xmalloc (sizeof *avs);
819 avs->var_cnt = var_cnt;
820 avs->name_tab = hsh_create (2 * var_cnt,
821 compare_var_ptrs_by_name, hash_var_ptr_by_name,
823 for (i = 0; i < var_cnt; i++)
824 if (hsh_insert (avs->name_tab, (void *) &var[i]) != NULL)
826 var_set_destroy (vs);