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);
119 var_set_destroy (vs);
123 /* Parses a set of variables from dictionary D given options
124 OPTS. Resulting list of variables stored in *VARS and the
125 number of variables into *VAR_CNT. Returns true only if
126 successful. Same behavior as parse_variables, except that all
127 allocations are taken from the given POOL. */
129 parse_variables_pool (struct lexer *lexer, struct pool *pool,
130 const struct dictionary *dict,
131 struct variable ***vars, size_t *var_cnt, int opts)
135 /* PV_APPEND is unsafe because parse_variables would free the
136 existing names on failure, but those names are presumably
137 already in the pool, which would attempt to re-free it
139 assert (!(opts & PV_APPEND));
141 retval = parse_variables (lexer, dict, vars, var_cnt, opts);
143 pool_register (pool, free, *vars);
147 /* Parses a variable name from VS. If successful, sets *IDX to
148 the variable's index in VS, *CLASS to the variable's
149 dictionary class, and returns true. Returns false on
152 parse_var_idx_class (struct lexer *lexer, const struct var_set *vs,
154 enum dict_class *class)
156 if (!parse_vs_variable_idx (lexer, vs, idx))
159 *class = dict_class_from_id (var_get_name (var_set_get_var (vs, *idx)));
163 /* Add the variable from VS with index IDX to the list of
164 variables V that has *NV elements and room for *MV.
165 Uses and updates INCLUDED to avoid duplicates if indicated by
166 PV_OPTS, which also affects what variables are allowed in
169 add_variable (struct variable ***v, size_t *nv, size_t *mv,
170 char *included, int pv_opts,
171 const struct var_set *vs, size_t idx)
173 struct variable *add = var_set_get_var (vs, idx);
174 const char *add_name = var_get_name (add);
176 if ((pv_opts & PV_NUMERIC) && !var_is_numeric (add))
177 msg (SW, _("%s is not a numeric variable. It will not be "
178 "included in the variable list."), add_name);
179 else if ((pv_opts & PV_STRING) && !var_is_alpha (add))
180 msg (SE, _("%s is not a string variable. It will not be "
181 "included in the variable list."), add_name);
182 else if ((pv_opts & PV_NO_SCRATCH)
183 && dict_class_from_id (add_name) == DC_SCRATCH)
184 msg (SE, _("Scratch variables (such as %s) are not allowed "
186 else if ((pv_opts & (PV_SAME_TYPE | PV_SAME_WIDTH)) && *nv
187 && var_get_type (add) != var_get_type ((*v)[0]))
188 msg (SE, _("%s and %s are not the same type. All variables in "
189 "this variable list must be of the same type. %s "
190 "will be omitted from the list."),
191 var_get_name ((*v)[0]), add_name, add_name);
192 else if ((pv_opts & PV_SAME_WIDTH) && *nv
193 && var_get_width (add) != var_get_width ((*v)[0]))
194 msg (SE, _("%s and %s are string variables with different widths. "
195 "All variables in this variable list must have the "
196 "same width. %s will be omttied from the list."),
197 var_get_name ((*v)[0]), add_name, add_name);
198 else if ((pv_opts & PV_NO_DUPLICATE) && included[idx])
199 msg (SE, _("Variable %s appears twice in variable list."), add_name);
200 else if ((pv_opts & PV_DUPLICATE) || !included[idx])
205 *v = xnrealloc (*v, *mv, sizeof **v);
208 if (included != NULL)
213 /* Adds the variables in VS with indexes FIRST_IDX through
214 LAST_IDX, inclusive, to the list of variables V that has *NV
215 elements and room for *MV. Uses and updates INCLUDED to avoid
216 duplicates if indicated by PV_OPTS, which also affects what
217 variables are allowed in appropriate ways. */
219 add_variables (struct variable ***v, size_t *nv, size_t *mv, char *included,
221 const struct var_set *vs, int first_idx, int last_idx,
222 enum dict_class class)
226 for (i = first_idx; i <= last_idx; i++)
227 if (dict_class_from_id (var_get_name (var_set_get_var (vs, i))) == class)
228 add_variable (v, nv, mv, included, pv_opts, vs, i);
231 /* Note that if parse_variables() returns false, *v is free()'d.
232 Conversely, if parse_variables() returns true, then *nv is
233 nonzero and *v is non-NULL. */
235 parse_var_set_vars (struct lexer *lexer, const struct var_set *vs,
236 struct variable ***v, size_t *nv,
246 /* At most one of PV_NUMERIC, PV_STRING, PV_SAME_TYPE,
247 PV_SAME_WIDTH may be specified. */
248 assert (((pv_opts & PV_NUMERIC) != 0)
249 + ((pv_opts & PV_STRING) != 0)
250 + ((pv_opts & PV_SAME_TYPE) != 0)
251 + ((pv_opts & PV_SAME_WIDTH) != 0) <= 1);
253 /* PV_DUPLICATE and PV_NO_DUPLICATE are incompatible. */
254 assert (!(pv_opts & PV_DUPLICATE) || !(pv_opts & PV_NO_DUPLICATE));
256 if (!(pv_opts & PV_APPEND))
265 if (!(pv_opts & PV_DUPLICATE))
269 included = xcalloc (var_set_get_cnt (vs), sizeof *included);
270 for (i = 0; i < *nv; i++)
273 if (!var_set_lookup_var_idx (vs, var_get_name ((*v)[i]), &index))
283 if (lex_match (lexer, T_ALL))
284 add_variables (v, nv, &mv, included, pv_opts,
285 vs, 0, var_set_get_cnt (vs) - 1, DC_ORDINARY);
288 enum dict_class class;
291 if (!parse_var_idx_class (lexer, vs, &first_idx, &class))
294 if (!lex_match (lexer, T_TO))
295 add_variable (v, nv, &mv, included, pv_opts, vs, first_idx);
299 enum dict_class last_class;
300 struct variable *first_var, *last_var;
302 if (!parse_var_idx_class (lexer, vs, &last_idx, &last_class))
305 first_var = var_set_get_var (vs, first_idx);
306 last_var = var_set_get_var (vs, last_idx);
308 if (last_idx < first_idx)
310 const char *first_name = var_get_name (first_var);
311 const char *last_name = var_get_name (last_var);
312 msg (SE, _("%s TO %s is not valid syntax since %s "
313 "precedes %s in the dictionary."),
314 first_name, last_name, first_name, last_name);
318 if (class != last_class)
320 msg (SE, _("When using the TO keyword to specify several "
321 "variables, both variables must be from "
322 "the same variable dictionaries, of either "
323 "ordinary, scratch, or system variables. "
324 "%s is a %s variable, whereas %s is %s."),
325 var_get_name (first_var), dict_class_to_name (class),
326 var_get_name (last_var),
327 dict_class_to_name (last_class));
331 add_variables (v, nv, &mv, included, pv_opts,
332 vs, first_idx, last_idx, class);
336 if (pv_opts & PV_SINGLE)
338 lex_match (lexer, ',');
340 while (lex_token (lexer) == T_ALL
341 || (lex_token (lexer) == T_ID && var_set_lookup_var (vs, lex_tokid (lexer)) != NULL));
357 /* Extracts a numeric suffix from variable name S, copying it
358 into string R. Sets *D to the length of R and *N to its
361 extract_num (char *s, char *r, int *n, int *d)
365 /* Find first digit. */
366 cp = s + strlen (s) - 1;
367 while (isdigit ((unsigned char) *cp) && cp > s)
372 strncpy (r, s, cp - s);
375 /* Count initial zeros. */
384 while (isdigit ((unsigned char) *cp))
387 *n = (*n * 10) + (*cp - '0');
392 if (*n == 0 && *d == 0)
394 msg (SE, _("incorrect use of TO convention"));
400 /* Parses a list of variable names according to the DATA LIST version
401 of the TO convention. */
403 parse_DATA_LIST_vars (struct lexer *lexer, char ***names, size_t *nnames, int pv_opts)
409 char name1[LONG_NAME_LEN + 1], name2[LONG_NAME_LEN + 1];
410 char root1[LONG_NAME_LEN + 1], root2[LONG_NAME_LEN + 1];
413 assert (names != NULL);
414 assert (nnames != NULL);
415 assert ((pv_opts & ~(PV_APPEND | PV_SINGLE
416 | PV_NO_SCRATCH | PV_NO_DUPLICATE)) == 0);
417 /* FIXME: PV_NO_DUPLICATE is not implemented. */
419 if (pv_opts & PV_APPEND)
420 nvar = mvar = *nnames;
429 if (lex_token (lexer) != T_ID)
431 lex_error (lexer, "expecting variable name");
434 if (dict_class_from_id (lex_tokid (lexer)) == DC_SCRATCH
435 && (pv_opts & PV_NO_SCRATCH))
437 msg (SE, _("Scratch variables not allowed here."));
440 strcpy (name1, lex_tokid (lexer));
442 if (lex_token (lexer) == T_TO)
445 if (lex_token (lexer) != T_ID)
447 lex_error (lexer, "expecting variable name");
450 strcpy (name2, lex_tokid (lexer));
453 if (!extract_num (name1, root1, &n1, &d1)
454 || !extract_num (name2, root2, &n2, &d2))
457 if (strcasecmp (root1, root2))
459 msg (SE, _("Prefixes don't match in use of TO convention."));
464 msg (SE, _("Bad bounds in use of TO convention."));
470 if (mvar < nvar + (n2 - n1 + 1))
472 mvar += ROUND_UP (n2 - n1 + 1, 16);
473 *names = xnrealloc (*names, mvar, sizeof **names);
476 for (n = n1; n <= n2; n++)
478 char name[LONG_NAME_LEN + 1];
479 sprintf (name, "%s%0*d", root1, d1, n);
480 (*names)[nvar] = xstrdup (name);
489 *names = xnrealloc (*names, mvar, sizeof **names);
491 (*names)[nvar++] = xstrdup (name1);
494 lex_match (lexer, ',');
496 if (pv_opts & PV_SINGLE)
499 while (lex_token (lexer) == T_ID);
507 for (i = 0; i < nvar; i++)
516 /* Registers each of the NAMES[0...NNAMES - 1] in POOL, as well
519 register_vars_pool (struct pool *pool, char **names, size_t nnames)
523 for (i = 0; i < nnames; i++)
524 pool_register (pool, free, names[i]);
525 pool_register (pool, free, names);
528 /* Parses a list of variable names according to the DATA LIST
529 version of the TO convention. Same args as
530 parse_DATA_LIST_vars(), except that all allocations are taken
531 from the given POOL. */
533 parse_DATA_LIST_vars_pool (struct lexer *lexer, struct pool *pool,
534 char ***names, size_t *nnames, int pv_opts)
538 /* PV_APPEND is unsafe because parse_DATA_LIST_vars would free
539 the existing names on failure, but those names are
540 presumably already in the pool, which would attempt to
542 assert (!(pv_opts & PV_APPEND));
544 retval = parse_DATA_LIST_vars (lexer, names, nnames, pv_opts);
546 register_vars_pool (pool, *names, *nnames);
550 /* Parses a list of variables where some of the variables may be
551 existing and the rest are to be created. Same args as
552 parse_DATA_LIST_vars(). */
554 parse_mixed_vars (struct lexer *lexer, const struct dictionary *dict,
555 char ***names, size_t *nnames, int pv_opts)
559 assert (names != NULL);
560 assert (nnames != NULL);
561 assert ((pv_opts & ~PV_APPEND) == 0);
563 if (!(pv_opts & PV_APPEND))
568 while (lex_token (lexer) == T_ID || lex_token (lexer) == T_ALL)
570 if (lex_token (lexer) == T_ALL || dict_lookup_var (dict, lex_tokid (lexer)) != NULL)
575 if (!parse_variables (lexer, dict, &v, &nv, PV_NONE))
577 *names = xnrealloc (*names, *nnames + nv, sizeof **names);
578 for (i = 0; i < nv; i++)
579 (*names)[*nnames + i] = xstrdup (var_get_name (v[i]));
583 else if (!parse_DATA_LIST_vars (lexer, names, nnames, PV_APPEND))
589 for (i = 0; i < *nnames; i++)
597 /* Parses a list of variables where some of the variables may be
598 existing and the rest are to be created. Same args as
599 parse_mixed_vars(), except that all allocations are taken
600 from the given POOL. */
602 parse_mixed_vars_pool (struct lexer *lexer, const struct dictionary *dict, struct pool *pool,
603 char ***names, size_t *nnames, int pv_opts)
607 /* PV_APPEND is unsafe because parse_mixed_vars_pool would free
608 the existing names on failure, but those names are
609 presumably already in the pool, which would attempt to
611 assert (!(pv_opts & PV_APPEND));
613 retval = parse_mixed_vars (lexer, dict, names, nnames, pv_opts);
615 register_vars_pool (pool, *names, *nnames);
619 /* A set of variables. */
622 size_t (*get_cnt) (const struct var_set *);
623 struct variable *(*get_var) (const struct var_set *, size_t idx);
624 bool (*lookup_var_idx) (const struct var_set *, const char *, size_t *);
625 void (*destroy) (struct var_set *);
629 /* Returns the number of variables in VS. */
631 var_set_get_cnt (const struct var_set *vs)
635 return vs->get_cnt (vs);
638 /* Return variable with index IDX in VS.
639 IDX must be less than the number of variables in VS. */
640 static struct variable *
641 var_set_get_var (const struct var_set *vs, size_t idx)
644 assert (idx < var_set_get_cnt (vs));
646 return vs->get_var (vs, idx);
649 /* Returns the variable in VS named NAME, or a null pointer if VS
650 contains no variable with that name. */
652 var_set_lookup_var (const struct var_set *vs, const char *name)
655 return (var_set_lookup_var_idx (vs, name, &idx)
656 ? var_set_get_var (vs, idx)
660 /* If VS contains a variable named NAME, sets *IDX to its index
661 and returns true. Otherwise, returns false. */
663 var_set_lookup_var_idx (const struct var_set *vs, const char *name,
667 assert (name != NULL);
668 assert (strlen (name) <= LONG_NAME_LEN);
670 return vs->lookup_var_idx (vs, name, idx);
675 var_set_destroy (struct var_set *vs)
681 /* Returns the number of variables in VS. */
683 dict_var_set_get_cnt (const struct var_set *vs)
685 struct dictionary *d = vs->aux;
687 return dict_get_var_cnt (d);
690 /* Return variable with index IDX in VS.
691 IDX must be less than the number of variables in VS. */
692 static struct variable *
693 dict_var_set_get_var (const struct var_set *vs, size_t idx)
695 struct dictionary *d = vs->aux;
697 return dict_get_var (d, idx);
700 /* If VS contains a variable named NAME, sets *IDX to its index
701 and returns true. Otherwise, returns false. */
703 dict_var_set_lookup_var_idx (const struct var_set *vs, const char *name,
706 struct dictionary *d = vs->aux;
707 struct variable *v = dict_lookup_var (d, name);
710 *idx = var_get_dict_index (v);
719 dict_var_set_destroy (struct var_set *vs)
724 /* Returns a variable set based on D. */
726 var_set_create_from_dict (const struct dictionary *d)
728 struct var_set *vs = xmalloc (sizeof *vs);
729 vs->get_cnt = dict_var_set_get_cnt;
730 vs->get_var = dict_var_set_get_var;
731 vs->lookup_var_idx = dict_var_set_lookup_var_idx;
732 vs->destroy = dict_var_set_destroy;
733 vs->aux = (void *) d;
737 /* A variable set based on an array. */
740 struct variable *const *var;/* Array of variables. */
741 size_t var_cnt; /* Number of elements in var. */
742 struct hsh_table *name_tab; /* Hash from variable names to variables. */
745 /* Returns the number of variables in VS. */
747 array_var_set_get_cnt (const struct var_set *vs)
749 struct array_var_set *avs = vs->aux;
754 /* Return variable with index IDX in VS.
755 IDX must be less than the number of variables in VS. */
756 static struct variable *
757 array_var_set_get_var (const struct var_set *vs, size_t idx)
759 struct array_var_set *avs = vs->aux;
761 return (struct variable *) avs->var[idx];
764 /* If VS contains a variable named NAME, sets *IDX to its index
765 and returns true. Otherwise, returns false. */
767 array_var_set_lookup_var_idx (const struct var_set *vs, const char *name,
770 struct array_var_set *avs = vs->aux;
771 struct variable *v, *const *vpp;
773 v = var_create (name, 0);
774 vpp = hsh_find (avs->name_tab, &v);
779 *idx = vpp - avs->var;
788 array_var_set_destroy (struct var_set *vs)
790 struct array_var_set *avs = vs->aux;
792 hsh_destroy (avs->name_tab);
797 /* Returns a variable set based on the VAR_CNT variables in
800 var_set_create_from_array (struct variable *const *var, size_t var_cnt)
803 struct array_var_set *avs;
806 vs = xmalloc (sizeof *vs);
807 vs->get_cnt = array_var_set_get_cnt;
808 vs->get_var = array_var_set_get_var;
809 vs->lookup_var_idx = array_var_set_lookup_var_idx;
810 vs->destroy = array_var_set_destroy;
811 vs->aux = avs = xmalloc (sizeof *avs);
813 avs->var_cnt = var_cnt;
814 avs->name_tab = hsh_create (2 * var_cnt,
815 compare_var_ptrs_by_name, hash_var_ptr_by_name,
817 for (i = 0; i < var_cnt; i++)
818 if (hsh_insert (avs->name_tab, (void *) &var[i]) != NULL)
820 var_set_destroy (vs);