1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 1997-9, 2000, 2006, 2007 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/>. */
19 #include "dictionary.h"
26 #include "identifier.h"
28 #include "value-labels.h"
32 #include <libpspp/array.h>
33 #include <libpspp/assertion.h>
34 #include <libpspp/compiler.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>
46 #define _(msgid) gettext (msgid)
51 struct variable **var; /* Variables. */
52 size_t var_cnt, var_cap; /* Number of variables, capacity. */
53 struct hsh_table *name_tab; /* Variable index by name. */
54 int next_value_idx; /* Index of next `union value' to allocate. */
55 const struct variable **split; /* SPLIT FILE vars. */
56 size_t split_cnt; /* SPLIT FILE count. */
57 struct variable *weight; /* WEIGHT variable. */
58 struct variable *filter; /* FILTER variable. */
59 casenumber case_limit; /* Current case limit (N command). */
60 char *label; /* File label. */
61 struct string documents; /* Documents, as a string. */
62 struct vector **vector; /* Vectors of variables. */
63 size_t vector_cnt; /* Number of vectors. */
64 const struct dict_callbacks *callbacks; /* Callbacks on dictionary
66 void *cb_data ; /* Data passed to callbacks */
69 /* Print a representation of dictionary D to stdout, for
70 debugging purposes. */
72 dict_dump (const struct dictionary *d)
75 for (i = 0 ; i < d->var_cnt ; ++i )
77 const struct variable *v =
79 printf ("Name: %s;\tdict_idx: %d; case_idx: %d\n",
81 var_get_dict_index (v),
82 var_get_case_index (v));
87 /* Associate CALLBACKS with DICT. Callbacks will be invoked whenever
88 the dictionary or any of the variables it contains are modified.
89 Each callback will get passed CALLBACK_DATA.
90 Any callback may be NULL, in which case it'll be ignored.
93 dict_set_callbacks (struct dictionary *dict,
94 const struct dict_callbacks *callbacks,
97 dict->callbacks = callbacks;
98 dict->cb_data = callback_data;
101 /* Shallow copy the callbacks from SRC to DEST */
103 dict_copy_callbacks (struct dictionary *dest,
104 const struct dictionary *src)
106 dest->callbacks = src->callbacks;
107 dest->cb_data = src->cb_data;
110 /* Creates and returns a new dictionary. */
114 struct dictionary *d = xzalloc (sizeof *d);
116 d->name_tab = hsh_create (8, compare_vars_by_name, hash_var_by_name,
121 /* Creates and returns a (deep) copy of an existing
124 The new dictionary's case indexes are copied from the old
125 dictionary. If the new dictionary won't be used to access
126 cases produced with the old dictionary, then the new
127 dictionary's case indexes should be compacted with
128 dict_compact_values to save space. */
130 dict_clone (const struct dictionary *s)
132 struct dictionary *d;
139 for (i = 0; i < s->var_cnt; i++)
141 const struct vardict_info *svdi;
142 struct vardict_info dvdi;
143 struct variable *sv = s->var[i];
144 struct variable *dv = dict_clone_var_assert (d, sv, var_get_name (sv));
147 for (i = 0; i < var_get_short_name_cnt (sv); i++)
148 var_set_short_name (dv, i, var_get_short_name (sv, i));
150 svdi = var_get_vardict (sv);
153 var_set_vardict (dv, &dvdi);
156 d->next_value_idx = s->next_value_idx;
158 d->split_cnt = s->split_cnt;
159 if (d->split_cnt > 0)
161 d->split = xnmalloc (d->split_cnt, sizeof *d->split);
162 for (i = 0; i < d->split_cnt; i++)
163 d->split[i] = dict_lookup_var_assert (d, var_get_name (s->split[i]));
166 if (s->weight != NULL)
167 dict_set_weight (d, dict_lookup_var_assert (d, var_get_name (s->weight)));
169 if (s->filter != NULL)
170 dict_set_filter (d, dict_lookup_var_assert (d, var_get_name (s->filter)));
172 d->case_limit = s->case_limit;
173 dict_set_label (d, dict_get_label (s));
174 dict_set_documents (d, dict_get_documents (s));
176 d->vector_cnt = s->vector_cnt;
177 d->vector = xnmalloc (d->vector_cnt, sizeof *d->vector);
178 for (i = 0; i < s->vector_cnt; i++)
179 d->vector[i] = vector_clone (s->vector[i], s, d);
184 /* Clears the contents from a dictionary without destroying the
185 dictionary itself. */
187 dict_clear (struct dictionary *d)
189 /* FIXME? Should we really clear case_limit, label, documents?
190 Others are necessarily cleared by deleting all the variables.*/
193 while (d->var_cnt > 0 )
195 dict_delete_var (d, d->var[d->var_cnt - 1]);
200 d->var_cnt = d->var_cap = 0;
201 hsh_clear (d->name_tab);
202 d->next_value_idx = 0;
203 dict_set_split_vars (d, NULL, 0);
204 dict_set_weight (d, NULL);
205 dict_set_filter (d, NULL);
209 ds_destroy (&d->documents);
210 dict_clear_vectors (d);
213 /* Destroys the aux data for every variable in D, by calling
214 var_clear_aux() for each variable. */
216 dict_clear_aux (struct dictionary *d)
222 for (i = 0; i < d->var_cnt; i++)
223 var_clear_aux (d->var[i]);
226 /* Clears a dictionary and destroys it. */
228 dict_destroy (struct dictionary *d)
232 /* In general, we don't want callbacks occuring, if the dictionary
233 is being destroyed */
234 d->callbacks = NULL ;
237 hsh_destroy (d->name_tab);
242 /* Returns the number of variables in D. */
244 dict_get_var_cnt (const struct dictionary *d)
251 /* Returns the variable in D with dictionary index IDX, which
252 must be between 0 and the count returned by
253 dict_get_var_cnt(), exclusive. */
255 dict_get_var (const struct dictionary *d, size_t idx)
258 assert (idx < d->var_cnt);
263 /* Sets *VARS to an array of pointers to variables in D and *CNT
264 to the number of variables in *D. All variables are returned
265 except for those, if any, in the classes indicated by EXCLUDE.
266 (There is no point in putting DC_SYSTEM in EXCLUDE as
267 dictionaries never include system variables.) */
269 dict_get_vars (const struct dictionary *d, const struct variable ***vars,
270 size_t *cnt, enum dict_class exclude)
272 dict_get_vars_mutable (d, (struct variable ***) vars, cnt, exclude);
275 /* Sets *VARS to an array of pointers to variables in D and *CNT
276 to the number of variables in *D. All variables are returned
277 except for those, if any, in the classes indicated by EXCLUDE.
278 (There is no point in putting DC_SYSTEM in EXCLUDE as
279 dictionaries never include system variables.) */
281 dict_get_vars_mutable (const struct dictionary *d, struct variable ***vars,
282 size_t *cnt, enum dict_class exclude)
288 assert (vars != NULL);
289 assert (cnt != NULL);
290 assert (exclude == (exclude & DC_ALL));
293 for (i = 0; i < d->var_cnt; i++)
295 enum dict_class class = var_get_dict_class (d->var[i]);
296 if (!(class & exclude))
300 *vars = xnmalloc (count, sizeof **vars);
302 for (i = 0; i < d->var_cnt; i++)
304 enum dict_class class = var_get_dict_class (d->var[i]);
305 if (!(class & exclude))
306 (*vars)[(*cnt)++] = d->var[i];
308 assert (*cnt == count);
311 static struct variable *
312 add_var (struct dictionary *d, struct variable *v)
314 /* Add dictionary info to variable. */
315 struct vardict_info vdi;
316 vdi.case_index = d->next_value_idx;
317 vdi.dict_index = d->var_cnt;
319 var_set_vardict (v, &vdi);
321 /* Update dictionary. */
322 if (d->var_cnt >= d->var_cap)
324 d->var_cap = 8 + 2 * d->var_cap;
325 d->var = xnrealloc (d->var, d->var_cap, sizeof *d->var);
327 d->var[d->var_cnt++] = v;
328 hsh_force_insert (d->name_tab, v);
330 if ( d->callbacks && d->callbacks->var_added )
331 d->callbacks->var_added (d, var_get_dict_index (v), d->cb_data);
333 d->next_value_idx += var_get_value_cnt (v);
338 /* Creates and returns a new variable in D with the given NAME
339 and WIDTH. Returns a null pointer if the given NAME would
340 duplicate that of an existing variable in the dictionary. */
342 dict_create_var (struct dictionary *d, const char *name, int width)
344 return (dict_lookup_var (d, name) == NULL
345 ? dict_create_var_assert (d, name, width)
349 /* Creates and returns a new variable in D with the given NAME
350 and WIDTH. Assert-fails if the given NAME would duplicate
351 that of an existing variable in the dictionary. */
353 dict_create_var_assert (struct dictionary *d, const char *name, int width)
355 assert (dict_lookup_var (d, name) == NULL);
356 return add_var (d, var_create (name, width));
359 /* Creates and returns a new variable in D with name NAME, as a
360 copy of existing variable OLD_VAR, which need not be in D or
361 in any dictionary. Returns a null pointer if the given NAME
362 would duplicate that of an existing variable in the
365 dict_clone_var (struct dictionary *d, const struct variable *old_var,
368 return (dict_lookup_var (d, name) == NULL
369 ? dict_clone_var_assert (d, old_var, name)
373 /* Creates and returns a new variable in D with name NAME, as a
374 copy of existing variable OLD_VAR, which need not be in D or
375 in any dictionary. Assert-fails if the given NAME would
376 duplicate that of an existing variable in the dictionary. */
378 dict_clone_var_assert (struct dictionary *d, const struct variable *old_var,
381 struct variable *new_var = var_clone (old_var);
382 assert (dict_lookup_var (d, name) == NULL);
383 var_set_name (new_var, name);
384 return add_var (d, new_var);
387 /* Returns the variable named NAME in D, or a null pointer if no
388 variable has that name. */
390 dict_lookup_var (const struct dictionary *d, const char *name)
392 struct variable *target ;
393 struct variable *result ;
395 if ( ! var_is_plausible_name (name, false))
398 target = var_create (name, 0);
399 result = hsh_find (d->name_tab, target);
400 var_destroy (target);
402 if ( result && var_has_vardict (result))
404 const struct vardict_info *vdi = var_get_vardict (result);
405 assert (vdi->dict == d);
411 /* Returns the variable named NAME in D. Assert-fails if no
412 variable has that name. */
414 dict_lookup_var_assert (const struct dictionary *d, const char *name)
416 struct variable *v = dict_lookup_var (d, name);
421 /* Returns true if variable V is in dictionary D,
424 dict_contains_var (const struct dictionary *d, const struct variable *v)
426 if (var_has_vardict (v))
428 const struct vardict_info *vdi = var_get_vardict (v);
429 return (vdi->dict_index >= 0
430 && vdi->dict_index < d->var_cnt
431 && d->var[vdi->dict_index] == v);
437 /* Compares two double pointers to variables, which should point
438 to elements of a struct dictionary's `var' member array. */
440 compare_var_ptrs (const void *a_, const void *b_, const void *aux UNUSED)
442 struct variable *const *a = a_;
443 struct variable *const *b = b_;
445 return *a < *b ? -1 : *a > *b;
448 /* Sets the dict_index in V's vardict to DICT_INDEX. */
450 set_var_dict_index (struct variable *v, int dict_index)
452 struct vardict_info vdi = *var_get_vardict (v);
453 struct dictionary *d = vdi.dict;
454 vdi.dict_index = dict_index;
455 var_set_vardict (v, &vdi);
457 if ( d->callbacks && d->callbacks->var_changed )
458 d->callbacks->var_changed (d, dict_index, d->cb_data);
461 /* Sets the case_index in V's vardict to CASE_INDEX. */
463 set_var_case_index (struct variable *v, int case_index)
465 struct vardict_info vdi = *var_get_vardict (v);
466 vdi.case_index = case_index;
467 var_set_vardict (v, &vdi);
470 /* Re-sets the dict_index in the dictionary variables with
471 indexes from FROM to TO (exclusive). */
473 reindex_vars (struct dictionary *d, size_t from, size_t to)
477 for (i = from; i < to; i++)
478 set_var_dict_index (d->var[i], i);
481 /* Deletes variable V from dictionary D and frees V.
483 This is a very bad idea if there might be any pointers to V
484 from outside D. In general, no variable in the active file's
485 dictionary should be deleted when any transformations are
486 active on the dictionary's dataset, because those
487 transformations might reference the deleted variable. The
488 safest time to delete a variable is just after a procedure has
489 been executed, as done by DELETE VARIABLES.
491 Pointers to V within D are not a problem, because
492 dict_delete_var() knows to remove V from split variables,
493 weights, filters, etc. */
495 dict_delete_var (struct dictionary *d, struct variable *v)
497 int dict_index = var_get_dict_index (v);
498 const int case_index = var_get_case_index (v);
499 const int val_cnt = var_get_value_cnt (v);
501 assert (dict_contains_var (d, v));
503 /* Delete aux data. */
506 dict_unset_split_var (d, v);
509 dict_set_weight (d, NULL);
512 dict_set_filter (d, NULL);
514 dict_clear_vectors (d);
516 /* Remove V from var array. */
517 remove_element (d->var, d->var_cnt, sizeof *d->var, dict_index);
520 /* Update dict_index for each affected variable. */
521 reindex_vars (d, dict_index, d->var_cnt);
523 /* Update name hash. */
524 hsh_force_delete (d->name_tab, v);
528 var_clear_vardict (v);
532 if (d->callbacks && d->callbacks->var_deleted )
533 d->callbacks->var_deleted (d, dict_index, case_index, val_cnt, d->cb_data);
536 /* Deletes the COUNT variables listed in VARS from D. This is
537 unsafe; see the comment on dict_delete_var() for details. */
539 dict_delete_vars (struct dictionary *d,
540 struct variable *const *vars, size_t count)
542 /* FIXME: this can be done in O(count) time, but this algorithm
545 assert (count == 0 || vars != NULL);
548 dict_delete_var (d, *vars++);
551 /* Deletes the COUNT variables in D starting at index IDX. This
552 is unsafe; see the comment on dict_delete_var() for
555 dict_delete_consecutive_vars (struct dictionary *d, size_t idx, size_t count)
557 /* FIXME: this can be done in O(count) time, but this algorithm
559 assert (idx + count <= d->var_cnt);
562 dict_delete_var (d, d->var[idx]);
565 /* Deletes scratch variables from dictionary D. */
567 dict_delete_scratch_vars (struct dictionary *d)
571 /* FIXME: this can be done in O(count) time, but this algorithm
575 for (i = 0; i < d->var_cnt; )
576 if (var_get_dict_class (d->var[i]) == DC_SCRATCH)
577 dict_delete_var (d, d->var[i]);
582 /* Moves V to 0-based position IDX in D. Other variables in D,
583 if any, retain their relative positions. Runs in time linear
584 in the distance moved. */
586 dict_reorder_var (struct dictionary *d, struct variable *v, size_t new_index)
588 size_t old_index = var_get_dict_index (v);
590 assert (new_index < d->var_cnt);
591 move_element (d->var, d->var_cnt, sizeof *d->var, old_index, new_index);
592 reindex_vars (d, MIN (old_index, new_index), MAX (old_index, new_index) + 1);
595 /* Reorders the variables in D, placing the COUNT variables
596 listed in ORDER in that order at the beginning of D. The
597 other variables in D, if any, retain their relative
600 dict_reorder_vars (struct dictionary *d,
601 struct variable *const *order, size_t count)
603 struct variable **new_var;
607 assert (count == 0 || order != NULL);
608 assert (count <= d->var_cnt);
610 new_var = xnmalloc (d->var_cnt, sizeof *new_var);
611 memcpy (new_var, order, count * sizeof *new_var);
612 for (i = 0; i < count; i++)
614 size_t index = var_get_dict_index (order[i]);
615 assert (d->var[index] == order[i]);
616 d->var[index] = NULL;
617 set_var_dict_index (order[i], i);
619 for (i = 0; i < d->var_cnt; i++)
620 if (d->var[i] != NULL)
622 assert (count < d->var_cnt);
623 new_var[count] = d->var[i];
624 set_var_dict_index (new_var[count], count);
631 /* Changes the name of variable V in dictionary D to NEW_NAME. */
633 rename_var (struct dictionary *d, struct variable *v, const char *new_name)
635 struct vardict_info vdi;
637 assert (dict_contains_var (d, v));
639 vdi = *var_get_vardict (v);
640 var_clear_vardict (v);
641 var_set_name (v, new_name);
642 var_set_vardict (v, &vdi);
645 /* Changes the name of V in D to name NEW_NAME. Assert-fails if
646 a variable named NEW_NAME is already in D, except that
647 NEW_NAME may be the same as V's existing name. */
649 dict_rename_var (struct dictionary *d, struct variable *v,
650 const char *new_name)
652 assert (!strcasecmp (var_get_name (v), new_name)
653 || dict_lookup_var (d, new_name) == NULL);
655 hsh_force_delete (d->name_tab, v);
656 rename_var (d, v, new_name);
657 hsh_force_insert (d->name_tab, v);
659 if (settings_get_algorithm () == ENHANCED)
660 var_clear_short_names (v);
662 if ( d->callbacks && d->callbacks->var_changed )
663 d->callbacks->var_changed (d, var_get_dict_index (v), d->cb_data);
666 /* Renames COUNT variables specified in VARS to the names given
667 in NEW_NAMES within dictionary D. If the renaming would
668 result in a duplicate variable name, returns false and stores a
669 name that would be duplicated into *ERR_NAME (if ERR_NAME is
670 non-null). Otherwise, the renaming is successful, and true
673 dict_rename_vars (struct dictionary *d,
674 struct variable **vars, char **new_names, size_t count,
681 assert (count == 0 || vars != NULL);
682 assert (count == 0 || new_names != NULL);
684 /* Save the names of the variables to be renamed. */
685 pool = pool_create ();
686 old_names = pool_nalloc (pool, count, sizeof *old_names);
687 for (i = 0; i < count; i++)
688 old_names[i] = pool_strdup (pool, var_get_name (vars[i]));
690 /* Remove the variables to be renamed from the name hash,
692 for (i = 0; i < count; i++)
694 hsh_force_delete (d->name_tab, vars[i]);
695 rename_var (d, vars[i], new_names[i]);
698 /* Add the renamed variables back into the name hash,
699 checking for conflicts. */
700 for (i = 0; i < count; i++)
701 if (hsh_insert (d->name_tab, vars[i]) != NULL)
703 /* There is a name conflict.
704 Back out all the name changes that have already
705 taken place, and indicate failure. */
707 if (err_name != NULL)
708 *err_name = new_names[i];
710 for (i = 0; i < fail_idx; i++)
711 hsh_force_delete (d->name_tab, vars[i]);
713 for (i = 0; i < count; i++)
715 rename_var (d, vars[i], old_names[i]);
716 hsh_force_insert (d->name_tab, vars[i]);
723 /* Clear short names. */
724 if (settings_get_algorithm () == ENHANCED)
725 for (i = 0; i < count; i++)
726 var_clear_short_names (vars[i]);
733 make_hinted_name (const struct dictionary *dict, const char *hint,
734 char name[VAR_NAME_LEN + 1])
736 bool dropped = false;
739 for (cp = name; *hint && cp < name + VAR_NAME_LEN; hint++)
742 ? lex_is_id1 (*hint) && *hint != '$'
743 : lex_is_idn (*hint))
750 if (cp < name + VAR_NAME_LEN)
760 size_t len = strlen (name);
763 if (dict_lookup_var (dict, name) == NULL)
766 for (i = 0; i < ULONG_MAX; i++)
768 char suffix[INT_BUFSIZE_BOUND (i) + 1];
772 if (!str_format_26adic (i + 1, &suffix[1], sizeof suffix - 1))
775 ofs = MIN (VAR_NAME_LEN - strlen (suffix), len);
776 strcpy (&name[ofs], suffix);
778 if (dict_lookup_var (dict, name) == NULL)
787 make_numeric_name (const struct dictionary *dict, unsigned long int *num_start,
788 char name[VAR_NAME_LEN + 1])
790 unsigned long int number;
792 for (number = num_start != NULL ? MAX (*num_start, 1) : 1;
796 sprintf (name, "VAR%03lu", number);
797 if (dict_lookup_var (dict, name) == NULL)
799 if (num_start != NULL)
800 *num_start = number + 1;
805 if (num_start != NULL)
806 *num_start = ULONG_MAX;
811 /* Attempts to devise a variable name unique within DICT.
812 Returns true if successful, in which case the new variable
813 name is stored into NAME. Returns false if all names that can
814 be generated have already been taken. (Returning false is
815 quite unlikely: at least ULONG_MAX unique names can be
818 HINT, if it is non-null, is used as a suggestion that will be
819 modified for suitability as a variable name and for
822 If HINT is null or entirely unsuitable, a name in the form
823 "VAR%03d" will be generated, where the smallest unused integer
824 value is used. If NUM_START is non-null, then its value is
825 used as the minimum numeric value to check, and it is updated
826 to the next value to be checked.
829 dict_make_unique_var_name (const struct dictionary *dict, const char *hint,
830 unsigned long int *num_start,
831 char name[VAR_NAME_LEN + 1])
833 return ((hint != NULL && make_hinted_name (dict, hint, name))
834 || make_numeric_name (dict, num_start, name));
837 /* Returns the weighting variable in dictionary D, or a null
838 pointer if the dictionary is unweighted. */
840 dict_get_weight (const struct dictionary *d)
843 assert (d->weight == NULL || dict_contains_var (d, d->weight));
848 /* Returns the value of D's weighting variable in case C, except
849 that a negative weight is returned as 0. Returns 1 if the
850 dictionary is unweighted. Will warn about missing, negative,
851 or zero values if *WARN_ON_INVALID is true. The function will
852 set *WARN_ON_INVALID to false if an invalid weight is
855 dict_get_case_weight (const struct dictionary *d, const struct ccase *c,
856 bool *warn_on_invalid)
861 if (d->weight == NULL)
865 double w = case_num (c, d->weight);
866 if (w < 0.0 || var_is_num_missing (d->weight, w, MV_ANY))
868 if ( w == 0.0 && warn_on_invalid != NULL && *warn_on_invalid ) {
869 *warn_on_invalid = false;
870 msg (SW, _("At least one case in the data file had a weight value "
871 "that was user-missing, system-missing, zero, or "
872 "negative. These case(s) were ignored."));
878 /* Sets the weighting variable of D to V, or turning off
879 weighting if V is a null pointer. */
881 dict_set_weight (struct dictionary *d, struct variable *v)
884 assert (v == NULL || dict_contains_var (d, v));
885 assert (v == NULL || var_is_numeric (v));
889 if ( d->callbacks && d->callbacks->weight_changed )
890 d->callbacks->weight_changed (d,
891 v ? var_get_dict_index (v) : -1,
895 /* Returns the filter variable in dictionary D (see cmd_filter())
896 or a null pointer if the dictionary is unfiltered. */
898 dict_get_filter (const struct dictionary *d)
901 assert (d->filter == NULL || dict_contains_var (d, d->filter));
906 /* Sets V as the filter variable for dictionary D. Passing a
907 null pointer for V turn off filtering. */
909 dict_set_filter (struct dictionary *d, struct variable *v)
912 assert (v == NULL || dict_contains_var (d, v));
913 assert (v == NULL || var_is_numeric (v));
917 if ( d->callbacks && d->callbacks->filter_changed )
918 d->callbacks->filter_changed (d,
919 v ? var_get_dict_index (v) : -1,
923 /* Returns the case limit for dictionary D, or zero if the number
924 of cases is unlimited. */
926 dict_get_case_limit (const struct dictionary *d)
930 return d->case_limit;
933 /* Sets CASE_LIMIT as the case limit for dictionary D. Use
934 0 for CASE_LIMIT to indicate no limit. */
936 dict_set_case_limit (struct dictionary *d, casenumber case_limit)
940 d->case_limit = case_limit;
943 /* Returns the case index of the next value to be added to D.
944 This value is the number of `union value's that need to be
945 allocated to store a case for dictionary D. */
947 dict_get_next_value_idx (const struct dictionary *d)
951 return d->next_value_idx;
954 /* Returns the number of bytes needed to store a case for
957 dict_get_case_size (const struct dictionary *d)
961 return sizeof (union value) * dict_get_next_value_idx (d);
964 /* Reassigns values in dictionary D so that fragmentation is
967 dict_compact_values (struct dictionary *d)
971 d->next_value_idx = 0;
972 for (i = 0; i < d->var_cnt; i++)
974 struct variable *v = d->var[i];
975 set_var_case_index (v, d->next_value_idx);
976 d->next_value_idx += var_get_value_cnt (v);
981 Reassigns case indices for D, increasing each index above START by
985 dict_pad_values (struct dictionary *d, int start, int padding)
992 for (i = 0; i < d->var_cnt; ++i)
994 struct variable *v = d->var[i];
996 int index = var_get_case_index (v);
999 set_var_case_index (v, index + padding);
1002 d->next_value_idx += padding;
1006 /* Returns the number of values occupied by the variables in
1007 dictionary D. All variables are considered if EXCLUDE_CLASSES
1008 is 0, or it may contain one or more of (1u << DC_ORDINARY),
1009 (1u << DC_SYSTEM), or (1u << DC_SCRATCH) to exclude the
1010 corresponding type of variable.
1012 The return value may be less than the number of values in one
1013 of dictionary D's cases (as returned by
1014 dict_get_next_value_idx) even if E is 0, because there may be
1015 gaps in D's cases due to deleted variables. */
1017 dict_count_values (const struct dictionary *d, unsigned int exclude_classes)
1022 assert ((exclude_classes & ~((1u << DC_ORDINARY)
1024 | (1u << DC_SCRATCH))) == 0);
1027 for (i = 0; i < d->var_cnt; i++)
1029 enum dict_class class = var_get_dict_class (d->var[i]);
1030 if (!(exclude_classes & (1u << class)))
1031 cnt += var_get_value_cnt (d->var[i]);
1036 /* Returns the SPLIT FILE vars (see cmd_split_file()). Call
1037 dict_get_split_cnt() to determine how many SPLIT FILE vars
1038 there are. Returns a null pointer if and only if there are no
1040 const struct variable *const *
1041 dict_get_split_vars (const struct dictionary *d)
1048 /* Returns the number of SPLIT FILE vars. */
1050 dict_get_split_cnt (const struct dictionary *d)
1054 return d->split_cnt;
1057 /* Removes variable V, which must be in D, from D's set of split
1060 dict_unset_split_var (struct dictionary *d, struct variable *v)
1064 assert (dict_contains_var (d, v));
1066 orig_count = d->split_cnt;
1067 d->split_cnt = remove_equal (d->split, d->split_cnt, sizeof *d->split,
1068 &v, compare_var_ptrs, NULL);
1069 if (orig_count != d->split_cnt)
1071 /* We changed the set of split variables so invoke the
1073 if (d->callbacks && d->callbacks->split_changed)
1074 d->callbacks->split_changed (d, d->cb_data);
1078 /* Sets CNT split vars SPLIT in dictionary D. */
1080 dict_set_split_vars (struct dictionary *d,
1081 struct variable *const *split, size_t cnt)
1084 assert (cnt == 0 || split != NULL);
1089 d->split = xnrealloc (d->split, cnt, sizeof *d->split) ;
1090 memcpy (d->split, split, cnt * sizeof *d->split);
1098 if ( d->callbacks && d->callbacks->split_changed )
1099 d->callbacks->split_changed (d, d->cb_data);
1102 /* Returns the file label for D, or a null pointer if D is
1103 unlabeled (see cmd_file_label()). */
1105 dict_get_label (const struct dictionary *d)
1112 /* Sets D's file label to LABEL, truncating it to a maximum of 60
1115 dict_set_label (struct dictionary *d, const char *label)
1120 d->label = label != NULL ? xstrndup (label, 60) : NULL;
1123 /* Returns the documents for D, or a null pointer if D has no
1124 documents. If the return value is nonnull, then the string
1125 will be an exact multiple of DOC_LINE_LENGTH bytes in length,
1126 with each segment corresponding to one line. */
1128 dict_get_documents (const struct dictionary *d)
1130 return ds_is_empty (&d->documents) ? NULL : ds_cstr (&d->documents);
1133 /* Sets the documents for D to DOCUMENTS, or removes D's
1134 documents if DOCUMENT is a null pointer. If DOCUMENTS is
1135 nonnull, then it should be an exact multiple of
1136 DOC_LINE_LENGTH bytes in length, with each segment
1137 corresponding to one line. */
1139 dict_set_documents (struct dictionary *d, const char *documents)
1143 ds_assign_cstr (&d->documents, documents != NULL ? documents : "");
1145 /* In case the caller didn't get it quite right, pad out the
1146 final line with spaces. */
1147 remainder = ds_length (&d->documents) % DOC_LINE_LENGTH;
1149 ds_put_char_multiple (&d->documents, ' ', DOC_LINE_LENGTH - remainder);
1152 /* Drops the documents from dictionary D. */
1154 dict_clear_documents (struct dictionary *d)
1156 ds_clear (&d->documents);
1159 /* Appends LINE to the documents in D. LINE will be truncated or
1160 padded on the right with spaces to make it exactly
1161 DOC_LINE_LENGTH bytes long. */
1163 dict_add_document_line (struct dictionary *d, const char *line)
1165 if (strlen (line) > DOC_LINE_LENGTH)
1167 /* Note to translators: "bytes" is correct, not characters */
1168 msg (SW, _("Truncating document line to %d bytes."), DOC_LINE_LENGTH);
1170 buf_copy_str_rpad (ds_put_uninit (&d->documents, DOC_LINE_LENGTH),
1171 DOC_LINE_LENGTH, line);
1174 /* Returns the number of document lines in dictionary D. */
1176 dict_get_document_line_cnt (const struct dictionary *d)
1178 return ds_length (&d->documents) / DOC_LINE_LENGTH;
1181 /* Copies document line number IDX from dictionary D into
1182 LINE, trimming off any trailing white space. */
1184 dict_get_document_line (const struct dictionary *d,
1185 size_t idx, struct string *line)
1187 assert (idx < dict_get_document_line_cnt (d));
1188 ds_assign_substring (line, ds_substr (&d->documents, idx * DOC_LINE_LENGTH,
1190 ds_rtrim (line, ss_cstr (CC_SPACES));
1193 /* Creates in D a vector named NAME that contains the CNT
1194 variables in VAR. Returns true if successful, or false if a
1195 vector named NAME already exists in D. */
1197 dict_create_vector (struct dictionary *d,
1199 struct variable **var, size_t cnt)
1203 assert (var != NULL);
1205 for (i = 0; i < cnt; i++)
1206 assert (dict_contains_var (d, var[i]));
1208 if (dict_lookup_vector (d, name) == NULL)
1210 d->vector = xnrealloc (d->vector, d->vector_cnt + 1, sizeof *d->vector);
1211 d->vector[d->vector_cnt++] = vector_create (name, var, cnt);
1218 /* Creates in D a vector named NAME that contains the CNT
1219 variables in VAR. A vector named NAME must not already exist
1222 dict_create_vector_assert (struct dictionary *d,
1224 struct variable **var, size_t cnt)
1226 assert (dict_lookup_vector (d, name) == NULL);
1227 dict_create_vector (d, name, var, cnt);
1230 /* Returns the vector in D with index IDX, which must be less
1231 than dict_get_vector_cnt (D). */
1232 const struct vector *
1233 dict_get_vector (const struct dictionary *d, size_t idx)
1236 assert (idx < d->vector_cnt);
1238 return d->vector[idx];
1241 /* Returns the number of vectors in D. */
1243 dict_get_vector_cnt (const struct dictionary *d)
1247 return d->vector_cnt;
1250 /* Looks up and returns the vector within D with the given
1252 const struct vector *
1253 dict_lookup_vector (const struct dictionary *d, const char *name)
1256 for (i = 0; i < d->vector_cnt; i++)
1257 if (!strcasecmp (vector_get_name (d->vector[i]), name))
1258 return d->vector[i];
1262 /* Deletes all vectors from D. */
1264 dict_clear_vectors (struct dictionary *d)
1268 for (i = 0; i < d->vector_cnt; i++)
1269 vector_destroy (d->vector[i]);
1276 /* Called from variable.c to notify the dictionary that some property of
1277 the variable has changed */
1279 dict_var_changed (const struct variable *v)
1281 if ( var_has_vardict (v))
1283 const struct vardict_info *vdi = var_get_vardict (v);
1284 struct dictionary *d = vdi->dict;
1286 if ( d->callbacks && d->callbacks->var_changed )
1287 d->callbacks->var_changed (d, var_get_dict_index (v), d->cb_data);
1292 /* Called from variable.c to notify the dictionary that the variable's width
1295 dict_var_resized (const struct variable *v, int delta)
1297 if ( var_has_vardict (v))
1299 const struct vardict_info *vdi = var_get_vardict (v);
1300 struct dictionary *d;
1304 dict_pad_values (d, var_get_case_index(v) + 1, delta);
1306 if ( d->callbacks && d->callbacks->var_resized )
1307 d->callbacks->var_resized (d, var_get_dict_index (v), delta, d->cb_data);