1 /* PSPP - computes sample statistics.
2 Copyright (C) 1997-9, 2000 Free Software Foundation, Inc.
3 Written by Ben Pfaff <blp@gnu.org>.
5 This program is free software; you can redistribute it and/or
6 modify it under the terms of the GNU General Public License as
7 published by the Free Software Foundation; either version 2 of the
8 License, or (at your option) any later version.
10 This program is distributed in the hope that it will be useful, but
11 WITHOUT ANY WARRANTY; without even the implied warranty of
12 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 General Public License for more details.
15 You should have received a copy of the GNU General Public License
16 along with this program; if not, write to the Free Software
17 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
22 #include <language/lexer/variable-parser.h>
29 #include <data/dictionary.h>
30 #include <data/procedure.h>
31 #include <data/variable.h>
32 #include <libpspp/alloc.h>
33 #include <libpspp/assertion.h>
34 #include <libpspp/bit-vector.h>
35 #include <libpspp/hash.h>
36 #include <libpspp/message.h>
37 #include <libpspp/misc.h>
38 #include <libpspp/pool.h>
39 #include <libpspp/str.h>
42 #define _(msgid) gettext (msgid)
44 /* Parses a name as a variable within VS. Sets *IDX to the
45 variable's index and returns true if successful. On failure
46 emits an error message and returns false. */
48 parse_vs_variable_idx (struct lexer *lexer, const struct var_set *vs,
53 if (lex_token (lexer) != T_ID)
55 lex_error (lexer, _("expecting variable name"));
58 else if (var_set_lookup_var_idx (vs, lex_tokid (lexer), idx))
65 msg (SE, _("%s is not a variable name."), lex_tokid (lexer));
70 /* Parses a name as a variable within VS and returns the variable
71 if successful. On failure emits an error message and returns
73 static struct variable *
74 parse_vs_variable (struct lexer *lexer, const struct var_set *vs)
77 return parse_vs_variable_idx (lexer, vs, &idx) ? var_set_get_var (vs, idx) : NULL;
80 /* Parses a variable name in dictionary D and returns the
81 variable if successful. On failure emits an error message and
82 returns a null pointer. */
84 parse_variable (struct lexer *lexer, const struct dictionary *d)
86 struct var_set *vs = var_set_create_from_dict (d);
87 struct variable *var = parse_vs_variable (lexer, vs);
92 /* Parses a set of variables from dictionary D given options
93 OPTS. Resulting list of variables stored in *VAR and the
94 number of variables into *CNT. Returns true only if
97 parse_variables (struct lexer *lexer, const struct dictionary *d,
98 struct variable ***var,
99 size_t *cnt, int opts)
105 assert (var != NULL);
106 assert (cnt != NULL);
108 vs = var_set_create_from_dict (d);
109 success = parse_var_set_vars (lexer, vs, var, cnt, opts);
112 var_set_destroy (vs);
116 /* Parses a set of variables from dictionary D given options
117 OPTS. Resulting list of variables stored in *VARS and the
118 number of variables into *VAR_CNT. Returns true only if
119 successful. Same behavior as parse_variables, except that all
120 allocations are taken from the given POOL. */
122 parse_variables_pool (struct lexer *lexer, struct pool *pool,
123 const struct dictionary *dict,
124 struct variable ***vars, size_t *var_cnt, int opts)
128 /* PV_APPEND is unsafe because parse_variables would free the
129 existing names on failure, but those names are presumably
130 already in the pool, which would attempt to re-free it
132 assert (!(opts & PV_APPEND));
134 retval = parse_variables (lexer, dict, vars, var_cnt, opts);
136 pool_register (pool, free, *vars);
140 /* Parses a variable name from VS. If successful, sets *IDX to
141 the variable's index in VS, *CLASS to the variable's
142 dictionary class, and returns true. Returns false on
145 parse_var_idx_class (struct lexer *lexer, const struct var_set *vs,
147 enum dict_class *class)
149 if (!parse_vs_variable_idx (lexer, vs, idx))
152 *class = dict_class_from_id (var_get_name (var_set_get_var (vs, *idx)));
156 /* Add the variable from VS with index IDX to the list of
157 variables V that has *NV elements and room for *MV.
158 Uses and updates INCLUDED to avoid duplicates if indicated by
159 PV_OPTS, which also affects what variables are allowed in
162 add_variable (struct variable ***v, size_t *nv, size_t *mv,
163 char *included, int pv_opts,
164 const struct var_set *vs, size_t idx)
166 struct variable *add = var_set_get_var (vs, idx);
167 const char *add_name = var_get_name (add);
169 if ((pv_opts & PV_NUMERIC) && !var_is_numeric (add))
170 msg (SW, _("%s is not a numeric variable. It will not be "
171 "included in the variable list."), add_name);
172 else if ((pv_opts & PV_STRING) && !var_is_alpha (add))
173 msg (SE, _("%s is not a string variable. It will not be "
174 "included in the variable list."), add_name);
175 else if ((pv_opts & PV_NO_SCRATCH)
176 && dict_class_from_id (add_name) == DC_SCRATCH)
177 msg (SE, _("Scratch variables (such as %s) are not allowed "
179 else if ((pv_opts & (PV_SAME_TYPE | PV_SAME_WIDTH)) && *nv
180 && var_get_type (add) != var_get_type ((*v)[0]))
181 msg (SE, _("%s and %s are not the same type. All variables in "
182 "this variable list must be of the same type. %s "
183 "will be omitted from the list."),
184 var_get_name ((*v)[0]), add_name, add_name);
185 else if ((pv_opts & PV_SAME_WIDTH) && *nv
186 && var_get_width (add) != var_get_width ((*v)[0]))
187 msg (SE, _("%s and %s are string variables with different widths. "
188 "All variables in this variable list must have the "
189 "same width. %s will be omttied from the list."),
190 var_get_name ((*v)[0]), add_name, add_name);
191 else if ((pv_opts & PV_NO_DUPLICATE) && included[idx])
192 msg (SE, _("Variable %s appears twice in variable list."), add_name);
193 else if ((pv_opts & PV_DUPLICATE) || !included[idx])
198 *v = xnrealloc (*v, *mv, sizeof **v);
201 if (included != NULL)
206 /* Adds the variables in VS with indexes FIRST_IDX through
207 LAST_IDX, inclusive, to the list of variables V that has *NV
208 elements and room for *MV. Uses and updates INCLUDED to avoid
209 duplicates if indicated by PV_OPTS, which also affects what
210 variables are allowed in appropriate ways. */
212 add_variables (struct variable ***v, size_t *nv, size_t *mv, char *included,
214 const struct var_set *vs, int first_idx, int last_idx,
215 enum dict_class class)
219 for (i = first_idx; i <= last_idx; i++)
220 if (dict_class_from_id (var_get_name (var_set_get_var (vs, i))) == class)
221 add_variable (v, nv, mv, included, pv_opts, vs, i);
224 /* Note that if parse_variables() returns false, *v is free()'d.
225 Conversely, if parse_variables() returns true, then *nv is
226 nonzero and *v is non-NULL. */
228 parse_var_set_vars (struct lexer *lexer, const struct var_set *vs,
229 struct variable ***v, size_t *nv,
239 /* At most one of PV_NUMERIC, PV_STRING, PV_SAME_TYPE,
240 PV_SAME_WIDTH may be specified. */
241 assert (((pv_opts & PV_NUMERIC) != 0)
242 + ((pv_opts & PV_STRING) != 0)
243 + ((pv_opts & PV_SAME_TYPE) != 0)
244 + ((pv_opts & PV_SAME_WIDTH) != 0) <= 1);
246 /* PV_DUPLICATE and PV_NO_DUPLICATE are incompatible. */
247 assert (!(pv_opts & PV_DUPLICATE) || !(pv_opts & PV_NO_DUPLICATE));
249 if (!(pv_opts & PV_APPEND))
258 if (!(pv_opts & PV_DUPLICATE))
262 included = xcalloc (var_set_get_cnt (vs), sizeof *included);
263 for (i = 0; i < *nv; i++)
266 if (!var_set_lookup_var_idx (vs, var_get_name ((*v)[i]), &index))
276 if (lex_match (lexer, T_ALL))
277 add_variables (v, nv, &mv, included, pv_opts,
278 vs, 0, var_set_get_cnt (vs) - 1, DC_ORDINARY);
281 enum dict_class class;
284 if (!parse_var_idx_class (lexer, vs, &first_idx, &class))
287 if (!lex_match (lexer, T_TO))
288 add_variable (v, nv, &mv, included, pv_opts, vs, first_idx);
292 enum dict_class last_class;
293 struct variable *first_var, *last_var;
295 if (!parse_var_idx_class (lexer, vs, &last_idx, &last_class))
298 first_var = var_set_get_var (vs, first_idx);
299 last_var = var_set_get_var (vs, last_idx);
301 if (last_idx < first_idx)
303 const char *first_name = var_get_name (first_var);
304 const char *last_name = var_get_name (last_var);
305 msg (SE, _("%s TO %s is not valid syntax since %s "
306 "precedes %s in the dictionary."),
307 first_name, last_name, first_name, last_name);
311 if (class != last_class)
313 msg (SE, _("When using the TO keyword to specify several "
314 "variables, both variables must be from "
315 "the same variable dictionaries, of either "
316 "ordinary, scratch, or system variables. "
317 "%s is a %s variable, whereas %s is %s."),
318 var_get_name (first_var), dict_class_to_name (class),
319 var_get_name (last_var),
320 dict_class_to_name (last_class));
324 add_variables (v, nv, &mv, included, pv_opts,
325 vs, first_idx, last_idx, class);
329 if (pv_opts & PV_SINGLE)
331 lex_match (lexer, ',');
333 while (lex_token (lexer) == T_ALL
334 || (lex_token (lexer) == T_ID && var_set_lookup_var (vs, lex_tokid (lexer)) != NULL));
350 /* Extracts a numeric suffix from variable name S, copying it
351 into string R. Sets *D to the length of R and *N to its
354 extract_num (char *s, char *r, int *n, int *d)
358 /* Find first digit. */
359 cp = s + strlen (s) - 1;
360 while (isdigit ((unsigned char) *cp) && cp > s)
365 strncpy (r, s, cp - s);
368 /* Count initial zeros. */
377 while (isdigit ((unsigned char) *cp))
380 *n = (*n * 10) + (*cp - '0');
385 if (*n == 0 && *d == 0)
387 msg (SE, _("incorrect use of TO convention"));
393 /* Parses a list of variable names according to the DATA LIST version
394 of the TO convention. */
396 parse_DATA_LIST_vars (struct lexer *lexer, char ***names, size_t *nnames, int pv_opts)
402 char name1[LONG_NAME_LEN + 1], name2[LONG_NAME_LEN + 1];
403 char root1[LONG_NAME_LEN + 1], root2[LONG_NAME_LEN + 1];
406 assert (names != NULL);
407 assert (nnames != NULL);
408 assert ((pv_opts & ~(PV_APPEND | PV_SINGLE
409 | PV_NO_SCRATCH | PV_NO_DUPLICATE)) == 0);
410 /* FIXME: PV_NO_DUPLICATE is not implemented. */
412 if (pv_opts & PV_APPEND)
413 nvar = mvar = *nnames;
422 if (lex_token (lexer) != T_ID)
424 lex_error (lexer, "expecting variable name");
427 if (dict_class_from_id (lex_tokid (lexer)) == DC_SCRATCH
428 && (pv_opts & PV_NO_SCRATCH))
430 msg (SE, _("Scratch variables not allowed here."));
433 strcpy (name1, lex_tokid (lexer));
435 if (lex_token (lexer) == T_TO)
438 if (lex_token (lexer) != T_ID)
440 lex_error (lexer, "expecting variable name");
443 strcpy (name2, lex_tokid (lexer));
446 if (!extract_num (name1, root1, &n1, &d1)
447 || !extract_num (name2, root2, &n2, &d2))
450 if (strcasecmp (root1, root2))
452 msg (SE, _("Prefixes don't match in use of TO convention."));
457 msg (SE, _("Bad bounds in use of TO convention."));
463 if (mvar < nvar + (n2 - n1 + 1))
465 mvar += ROUND_UP (n2 - n1 + 1, 16);
466 *names = xnrealloc (*names, mvar, sizeof **names);
469 for (n = n1; n <= n2; n++)
471 char name[LONG_NAME_LEN + 1];
472 sprintf (name, "%s%0*d", root1, d1, n);
473 (*names)[nvar] = xstrdup (name);
482 *names = xnrealloc (*names, mvar, sizeof **names);
484 (*names)[nvar++] = xstrdup (name1);
487 lex_match (lexer, ',');
489 if (pv_opts & PV_SINGLE)
492 while (lex_token (lexer) == T_ID);
500 for (i = 0; i < nvar; i++)
509 /* Registers each of the NAMES[0...NNAMES - 1] in POOL, as well
512 register_vars_pool (struct pool *pool, char **names, size_t nnames)
516 for (i = 0; i < nnames; i++)
517 pool_register (pool, free, names[i]);
518 pool_register (pool, free, names);
521 /* Parses a list of variable names according to the DATA LIST
522 version of the TO convention. Same args as
523 parse_DATA_LIST_vars(), except that all allocations are taken
524 from the given POOL. */
526 parse_DATA_LIST_vars_pool (struct lexer *lexer, struct pool *pool,
527 char ***names, size_t *nnames, int pv_opts)
531 /* PV_APPEND is unsafe because parse_DATA_LIST_vars would free
532 the existing names on failure, but those names are
533 presumably already in the pool, which would attempt to
535 assert (!(pv_opts & PV_APPEND));
537 retval = parse_DATA_LIST_vars (lexer, names, nnames, pv_opts);
539 register_vars_pool (pool, *names, *nnames);
543 /* Parses a list of variables where some of the variables may be
544 existing and the rest are to be created. Same args as
545 parse_DATA_LIST_vars(). */
547 parse_mixed_vars (struct lexer *lexer, const struct dictionary *dict,
548 char ***names, size_t *nnames, int pv_opts)
552 assert (names != NULL);
553 assert (nnames != NULL);
554 assert ((pv_opts & ~PV_APPEND) == 0);
556 if (!(pv_opts & PV_APPEND))
561 while (lex_token (lexer) == T_ID || lex_token (lexer) == T_ALL)
563 if (lex_token (lexer) == T_ALL || dict_lookup_var (dict, lex_tokid (lexer)) != NULL)
568 if (!parse_variables (lexer, dict, &v, &nv, PV_NONE))
570 *names = xnrealloc (*names, *nnames + nv, sizeof **names);
571 for (i = 0; i < nv; i++)
572 (*names)[*nnames + i] = xstrdup (var_get_name (v[i]));
576 else if (!parse_DATA_LIST_vars (lexer, names, nnames, PV_APPEND))
582 for (i = 0; i < *nnames; i++)
590 /* Parses a list of variables where some of the variables may be
591 existing and the rest are to be created. Same args as
592 parse_mixed_vars(), except that all allocations are taken
593 from the given POOL. */
595 parse_mixed_vars_pool (struct lexer *lexer, const struct dictionary *dict, struct pool *pool,
596 char ***names, size_t *nnames, int pv_opts)
600 /* PV_APPEND is unsafe because parse_mixed_vars_pool would free
601 the existing names on failure, but those names are
602 presumably already in the pool, which would attempt to
604 assert (!(pv_opts & PV_APPEND));
606 retval = parse_mixed_vars (lexer, dict, names, nnames, pv_opts);
608 register_vars_pool (pool, *names, *nnames);
612 /* A set of variables. */
615 size_t (*get_cnt) (const struct var_set *);
616 struct variable *(*get_var) (const struct var_set *, size_t idx);
617 bool (*lookup_var_idx) (const struct var_set *, const char *, size_t *);
618 void (*destroy) (struct var_set *);
622 /* Returns the number of variables in VS. */
624 var_set_get_cnt (const struct var_set *vs)
628 return vs->get_cnt (vs);
631 /* Return variable with index IDX in VS.
632 IDX must be less than the number of variables in VS. */
634 var_set_get_var (const struct var_set *vs, size_t idx)
637 assert (idx < var_set_get_cnt (vs));
639 return vs->get_var (vs, idx);
642 /* Returns the variable in VS named NAME, or a null pointer if VS
643 contains no variable with that name. */
645 var_set_lookup_var (const struct var_set *vs, const char *name)
648 return (var_set_lookup_var_idx (vs, name, &idx)
649 ? var_set_get_var (vs, idx)
653 /* If VS contains a variable named NAME, sets *IDX to its index
654 and returns true. Otherwise, returns false. */
656 var_set_lookup_var_idx (const struct var_set *vs, const char *name,
660 assert (name != NULL);
661 assert (strlen (name) <= LONG_NAME_LEN);
663 return vs->lookup_var_idx (vs, name, idx);
668 var_set_destroy (struct var_set *vs)
674 /* Returns the number of variables in VS. */
676 dict_var_set_get_cnt (const struct var_set *vs)
678 struct dictionary *d = vs->aux;
680 return dict_get_var_cnt (d);
683 /* Return variable with index IDX in VS.
684 IDX must be less than the number of variables in VS. */
685 static struct variable *
686 dict_var_set_get_var (const struct var_set *vs, size_t idx)
688 struct dictionary *d = vs->aux;
690 return dict_get_var (d, idx);
693 /* If VS contains a variable named NAME, sets *IDX to its index
694 and returns true. Otherwise, returns false. */
696 dict_var_set_lookup_var_idx (const struct var_set *vs, const char *name,
699 struct dictionary *d = vs->aux;
700 struct variable *v = dict_lookup_var (d, name);
703 *idx = var_get_dict_index (v);
712 dict_var_set_destroy (struct var_set *vs)
717 /* Returns a variable set based on D. */
719 var_set_create_from_dict (const struct dictionary *d)
721 struct var_set *vs = xmalloc (sizeof *vs);
722 vs->get_cnt = dict_var_set_get_cnt;
723 vs->get_var = dict_var_set_get_var;
724 vs->lookup_var_idx = dict_var_set_lookup_var_idx;
725 vs->destroy = dict_var_set_destroy;
726 vs->aux = (void *) d;
730 /* A variable set based on an array. */
733 struct variable *const *var;/* Array of variables. */
734 size_t var_cnt; /* Number of elements in var. */
735 struct hsh_table *name_tab; /* Hash from variable names to variables. */
738 /* Returns the number of variables in VS. */
740 array_var_set_get_cnt (const struct var_set *vs)
742 struct array_var_set *avs = vs->aux;
747 /* Return variable with index IDX in VS.
748 IDX must be less than the number of variables in VS. */
749 static struct variable *
750 array_var_set_get_var (const struct var_set *vs, size_t idx)
752 struct array_var_set *avs = vs->aux;
754 return (struct variable *) avs->var[idx];
757 /* If VS contains a variable named NAME, sets *IDX to its index
758 and returns true. Otherwise, returns false. */
760 array_var_set_lookup_var_idx (const struct var_set *vs, const char *name,
763 struct array_var_set *avs = vs->aux;
764 struct variable *v, *const *vpp;
766 v = var_create (name, 0);
767 vpp = hsh_find (avs->name_tab, &v);
772 *idx = vpp - avs->var;
781 array_var_set_destroy (struct var_set *vs)
783 struct array_var_set *avs = vs->aux;
785 hsh_destroy (avs->name_tab);
790 /* Returns a variable set based on the VAR_CNT variables in
793 var_set_create_from_array (struct variable *const *var, size_t var_cnt)
796 struct array_var_set *avs;
799 vs = xmalloc (sizeof *vs);
800 vs->get_cnt = array_var_set_get_cnt;
801 vs->get_var = array_var_set_get_var;
802 vs->lookup_var_idx = array_var_set_lookup_var_idx;
803 vs->destroy = array_var_set_destroy;
804 vs->aux = avs = xmalloc (sizeof *avs);
806 avs->var_cnt = var_cnt;
807 avs->name_tab = hsh_create (2 * var_cnt,
808 compare_var_ptrs_by_name, hash_var_ptr_by_name,
810 for (i = 0; i < var_cnt; i++)
811 if (hsh_insert (avs->name_tab, (void *) &var[i]) != NULL)
813 var_set_destroy (vs);