-/* PSPP - computes sample statistics.
- Copyright (C) 1997-9, 2000, 2006 Free Software Foundation, Inc.
- Written by Ben Pfaff <blp@gnu.org>.
+/* PSPP - a program for statistical analysis.
+ Copyright (C) 1997-9, 2000, 2006, 2007, 2009 Free Software Foundation, Inc.
- This program is free software; you can redistribute it and/or
- modify it under the terms of the GNU General Public License as
- published by the Free Software Foundation; either version 2 of the
- License, or (at your option) any later version.
+ This program is free software: you can redistribute it and/or modify
+ it under the terms of the GNU General Public License as published by
+ the Free Software Foundation, either version 3 of the License, or
+ (at your option) any later version.
- This program is distributed in the hope that it will be useful, but
- WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
- General Public License for more details.
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ GNU General Public License for more details.
You should have received a copy of the GNU General Public License
- along with this program; if not, write to the Free Software
- Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
- 02110-1301, USA. */
+ along with this program. If not, see <http://www.gnu.org/licenses/>. */
#include <config.h>
#include <stdlib.h>
#include <ctype.h>
-#include "case.h"
-#include "cat-routines.h"
-#include "category.h"
-#include "settings.h"
-#include "value-labels.h"
-#include "variable.h"
-#include <libpspp/alloc.h>
+#include <data/attributes.h>
+#include <data/case.h>
+#include <data/category.h>
+#include <data/identifier.h>
+#include <data/settings.h>
+#include <data/value-labels.h>
+#include <data/vardict.h>
+#include <data/variable.h>
+#include <data/vector.h>
#include <libpspp/array.h>
+#include <libpspp/assertion.h>
#include <libpspp/compiler.h>
#include <libpspp/hash.h>
#include <libpspp/message.h>
#include <libpspp/misc.h>
+#include <libpspp/pool.h>
#include <libpspp/str.h>
+#include "intprops.h"
+#include "minmax.h"
+#include "xalloc.h"
+
#include "gettext.h"
#define _(msgid) gettext (msgid)
{
struct variable **var; /* Variables. */
size_t var_cnt, var_cap; /* Number of variables, capacity. */
+ struct caseproto *proto; /* Prototype for dictionary cases
+ (updated lazily). */
struct hsh_table *name_tab; /* Variable index by name. */
int next_value_idx; /* Index of next `union value' to allocate. */
- struct variable **split; /* SPLIT FILE vars. */
+ const struct variable **split; /* SPLIT FILE vars. */
size_t split_cnt; /* SPLIT FILE count. */
struct variable *weight; /* WEIGHT variable. */
struct variable *filter; /* FILTER variable. */
- size_t case_limit; /* Current case limit (N command). */
+ casenumber case_limit; /* Current case limit (N command). */
char *label; /* File label. */
- char *documents; /* Documents, as a string. */
+ struct string documents; /* Documents, as a string. */
struct vector **vector; /* Vectors of variables. */
size_t vector_cnt; /* Number of vectors. */
+ struct attrset attributes; /* Custom attributes. */
+
+ char *encoding; /* Character encoding of string data */
+
+ const struct dict_callbacks *callbacks; /* Callbacks on dictionary
+ modification */
+ void *cb_data ; /* Data passed to callbacks */
+
+ void (*changed) (struct dictionary *, void *); /* Generic change callback */
+ void *changed_data;
};
+
+void
+dict_set_encoding (struct dictionary *d, const char *enc)
+{
+ if (enc)
+ d->encoding = xstrdup (enc);
+}
+
+const char *
+dict_get_encoding (const struct dictionary *d)
+{
+ return d->encoding ;
+}
+
+
+void
+dict_set_change_callback (struct dictionary *d,
+ void (*changed) (struct dictionary *, void*),
+ void *data)
+{
+ d->changed = changed;
+ d->changed_data = data;
+}
+
+/* Discards dictionary D's caseproto. (It will be regenerated
+ lazily, on demand.) */
+static void
+invalidate_proto (struct dictionary *d)
+{
+ caseproto_unref (d->proto);
+ d->proto = NULL;
+}
+
+/* Print a representation of dictionary D to stdout, for
+ debugging purposes. */
+void
+dict_dump (const struct dictionary *d)
+{
+ int i;
+ for (i = 0 ; i < d->var_cnt ; ++i )
+ {
+ const struct variable *v =
+ d->var[i];
+ printf ("Name: %s;\tdict_idx: %d; case_idx: %d\n",
+ var_get_name (v),
+ var_get_dict_index (v),
+ var_get_case_index (v));
+
+ }
+}
+
+/* Associate CALLBACKS with DICT. Callbacks will be invoked whenever
+ the dictionary or any of the variables it contains are modified.
+ Each callback will get passed CALLBACK_DATA.
+ Any callback may be NULL, in which case it'll be ignored.
+*/
+void
+dict_set_callbacks (struct dictionary *dict,
+ const struct dict_callbacks *callbacks,
+ void *callback_data)
+{
+ dict->callbacks = callbacks;
+ dict->cb_data = callback_data;
+}
+
+/* Shallow copy the callbacks from SRC to DEST */
+void
+dict_copy_callbacks (struct dictionary *dest,
+ const struct dictionary *src)
+{
+ dest->callbacks = src->callbacks;
+ dest->cb_data = src->cb_data;
+}
+
/* Creates and returns a new dictionary. */
struct dictionary *
-dict_create (void)
+dict_create (void)
{
- struct dictionary *d = xmalloc (sizeof *d);
-
- d->var = NULL;
- d->var_cnt = d->var_cap = 0;
- d->name_tab = hsh_create (8, compare_var_names, hash_var_name, NULL, NULL);
- d->next_value_idx = 0;
- d->split = NULL;
- d->split_cnt = 0;
- d->weight = NULL;
- d->filter = NULL;
- d->case_limit = 0;
- d->label = NULL;
- d->documents = NULL;
- d->vector = NULL;
- d->vector_cnt = 0;
+ struct dictionary *d = xzalloc (sizeof *d);
+ d->name_tab = hsh_create (8, compare_vars_by_name, hash_var_by_name,
+ NULL, NULL);
+ attrset_init (&d->attributes);
return d;
}
/* Creates and returns a (deep) copy of an existing
- dictionary. */
+ dictionary.
+
+ The new dictionary's case indexes are copied from the old
+ dictionary. If the new dictionary won't be used to access
+ cases produced with the old dictionary, then the new
+ dictionary's case indexes should be compacted with
+ dict_compact_values to save space. */
struct dictionary *
-dict_clone (const struct dictionary *s)
+dict_clone (const struct dictionary *s)
{
struct dictionary *d;
size_t i;
d = dict_create ();
- for (i = 0; i < s->var_cnt; i++)
+ for (i = 0; i < s->var_cnt; i++)
{
+ const struct vardict_info *svdi;
+ struct vardict_info dvdi;
struct variable *sv = s->var[i];
- struct variable *dv = dict_clone_var_assert (d, sv, sv->name);
- var_set_short_name (dv, sv->short_name);
+ struct variable *dv = dict_clone_var_assert (d, sv, var_get_name (sv));
+ size_t i;
+
+ for (i = 0; i < var_get_short_name_cnt (sv); i++)
+ var_set_short_name (dv, i, var_get_short_name (sv, i));
+
+ svdi = var_get_vardict (sv);
+ dvdi = *svdi;
+ dvdi.dict = d;
+ var_set_vardict (dv, &dvdi);
}
d->next_value_idx = s->next_value_idx;
d->split_cnt = s->split_cnt;
- if (d->split_cnt > 0)
+ if (d->split_cnt > 0)
{
d->split = xnmalloc (d->split_cnt, sizeof *d->split);
- for (i = 0; i < d->split_cnt; i++)
- d->split[i] = dict_lookup_var_assert (d, s->split[i]->name);
+ for (i = 0; i < d->split_cnt; i++)
+ d->split[i] = dict_lookup_var_assert (d, var_get_name (s->split[i]));
}
- if (s->weight != NULL)
- d->weight = dict_lookup_var_assert (d, s->weight->name);
+ if (s->weight != NULL)
+ dict_set_weight (d, dict_lookup_var_assert (d, var_get_name (s->weight)));
- if (s->filter != NULL)
- d->filter = dict_lookup_var_assert (d, s->filter->name);
+ if (s->filter != NULL)
+ dict_set_filter (d, dict_lookup_var_assert (d, var_get_name (s->filter)));
d->case_limit = s->case_limit;
dict_set_label (d, dict_get_label (s));
d->vector_cnt = s->vector_cnt;
d->vector = xnmalloc (d->vector_cnt, sizeof *d->vector);
- for (i = 0; i < s->vector_cnt; i++)
- {
- struct vector *sv = s->vector[i];
- struct vector *dv = d->vector[i] = xmalloc (sizeof *dv);
- int j;
-
- dv->idx = i;
- strcpy (dv->name, sv->name);
- dv->cnt = sv->cnt;
- dv->var = xnmalloc (dv->cnt, sizeof *dv->var);
- for (j = 0; j < dv->cnt; j++)
- dv->var[j] = d->var[sv->var[j]->index];
- }
+ for (i = 0; i < s->vector_cnt; i++)
+ d->vector[i] = vector_clone (s->vector[i], s, d);
+
+ if ( s->encoding)
+ d->encoding = xstrdup (s->encoding);
+
+ dict_set_attributes (d, dict_get_attributes (s));
return d;
}
/* Clears the contents from a dictionary without destroying the
dictionary itself. */
void
-dict_clear (struct dictionary *d)
+dict_clear (struct dictionary *d)
{
/* FIXME? Should we really clear case_limit, label, documents?
Others are necessarily cleared by deleting all the variables.*/
- int i;
-
assert (d != NULL);
- for (i = 0; i < d->var_cnt; i++)
+ while (d->var_cnt > 0 )
{
- struct variable *v = d->var[i];
- var_clear_aux (v);
- val_labs_destroy (v->val_labs);
- free (v->label);
- free (v);
+ dict_delete_var (d, d->var[d->var_cnt - 1]);
}
+
free (d->var);
d->var = NULL;
d->var_cnt = d->var_cap = 0;
+ invalidate_proto (d);
hsh_clear (d->name_tab);
d->next_value_idx = 0;
- free (d->split);
- d->split = NULL;
- d->split_cnt = 0;
- d->weight = NULL;
- d->filter = NULL;
+ dict_set_split_vars (d, NULL, 0);
+ dict_set_weight (d, NULL);
+ dict_set_filter (d, NULL);
d->case_limit = 0;
free (d->label);
d->label = NULL;
- free (d->documents);
- d->documents = NULL;
+ ds_destroy (&d->documents);
dict_clear_vectors (d);
+ attrset_clear (&d->attributes);
}
/* Destroys the aux data for every variable in D, by calling
var_clear_aux() for each variable. */
void
-dict_clear_aux (struct dictionary *d)
+dict_clear_aux (struct dictionary *d)
{
int i;
-
+
assert (d != NULL);
-
+
for (i = 0; i < d->var_cnt; i++)
var_clear_aux (d->var[i]);
}
void
dict_destroy (struct dictionary *d)
{
- if (d != NULL)
+ if (d != NULL)
{
+ /* In general, we don't want callbacks occuring, if the dictionary
+ is being destroyed */
+ d->callbacks = NULL ;
+
dict_clear (d);
hsh_destroy (d->name_tab);
+ attrset_destroy (&d->attributes);
free (d);
}
}
/* Returns the number of variables in D. */
size_t
-dict_get_var_cnt (const struct dictionary *d)
+dict_get_var_cnt (const struct dictionary *d)
{
assert (d != NULL);
return d->var_cnt;
}
-/* Returns the variable in D with index IDX, which must be
- between 0 and the count returned by dict_get_var_cnt(),
- exclusive. */
+/* Returns the variable in D with dictionary index IDX, which
+ must be between 0 and the count returned by
+ dict_get_var_cnt(), exclusive. */
struct variable *
-dict_get_var (const struct dictionary *d, size_t idx)
+dict_get_var (const struct dictionary *d, size_t idx)
{
assert (d != NULL);
assert (idx < d->var_cnt);
/* Sets *VARS to an array of pointers to variables in D and *CNT
to the number of variables in *D. All variables are returned
- if EXCLUDE_CLASSES is 0, or it may contain one or more of (1u
- << DC_ORDINARY), (1u << DC_SYSTEM), or (1u << DC_SCRATCH) to
- exclude the corresponding type of variable. */
+ except for those, if any, in the classes indicated by EXCLUDE.
+ (There is no point in putting DC_SYSTEM in EXCLUDE as
+ dictionaries never include system variables.) */
void
-dict_get_vars (const struct dictionary *d, struct variable ***vars,
- size_t *cnt, unsigned exclude_classes)
+dict_get_vars (const struct dictionary *d, const struct variable ***vars,
+ size_t *cnt, enum dict_class exclude)
+{
+ dict_get_vars_mutable (d, (struct variable ***) vars, cnt, exclude);
+}
+
+/* Sets *VARS to an array of pointers to variables in D and *CNT
+ to the number of variables in *D. All variables are returned
+ except for those, if any, in the classes indicated by EXCLUDE.
+ (There is no point in putting DC_SYSTEM in EXCLUDE as
+ dictionaries never include system variables.) */
+void
+dict_get_vars_mutable (const struct dictionary *d, struct variable ***vars,
+ size_t *cnt, enum dict_class exclude)
{
size_t count;
size_t i;
-
+
assert (d != NULL);
assert (vars != NULL);
assert (cnt != NULL);
- assert ((exclude_classes & ~((1u << DC_ORDINARY)
- | (1u << DC_SYSTEM)
- | (1u << DC_SCRATCH))) == 0);
-
+ assert (exclude == (exclude & DC_ALL));
+
count = 0;
for (i = 0; i < d->var_cnt; i++)
- if (!(exclude_classes & (1u << dict_class_from_id (d->var[i]->name))))
- count++;
+ {
+ enum dict_class class = var_get_dict_class (d->var[i]);
+ if (!(class & exclude))
+ count++;
+ }
*vars = xnmalloc (count, sizeof **vars);
*cnt = 0;
for (i = 0; i < d->var_cnt; i++)
- if (!(exclude_classes & (1u << dict_class_from_id (d->var[i]->name))))
- (*vars)[(*cnt)++] = d->var[i];
+ {
+ enum dict_class class = var_get_dict_class (d->var[i]);
+ if (!(class & exclude))
+ (*vars)[(*cnt)++] = d->var[i];
+ }
assert (*cnt == count);
}
-
-/* Creates and returns a new variable in D with the given NAME
- and WIDTH. Returns a null pointer if the given NAME would
- duplicate that of an existing variable in the dictionary. */
-struct variable *
-dict_create_var (struct dictionary *d, const char *name, int width)
+static struct variable *
+add_var (struct dictionary *d, struct variable *v)
{
- struct variable *v;
-
- assert (d != NULL);
- assert (name != NULL);
-
- assert (width >= 0 && width <= MAX_STRING);
-
- assert (var_is_plausible_name(name,0));
-
- /* Make sure there's not already a variable by that name. */
- if (dict_lookup_var (d, name) != NULL)
- return NULL;
-
- /* Allocate and initialize variable. */
- v = xmalloc (sizeof *v);
- str_copy_trunc (v->name, sizeof v->name, name);
- v->type = width == 0 ? NUMERIC : ALPHA;
- v->width = width;
- v->fv = d->next_value_idx;
- v->nv = width == 0 ? 1 : DIV_RND_UP (width, MAX_SHORT_STRING);
- v->leave = dict_class_from_id (v->name) == DC_SCRATCH;
- v->index = d->var_cnt;
- mv_init (&v->miss, width);
- if (v->type == NUMERIC)
- {
- v->print = f8_2;
- v->alignment = ALIGN_RIGHT;
- v->display_width = 8;
- v->measure = MEASURE_SCALE;
- }
- else
- {
- v->print = make_output_format (FMT_A, v->width, 0);
- v->alignment = ALIGN_LEFT;
- v->display_width = 8;
- v->measure = MEASURE_NOMINAL;
- }
- v->write = v->print;
- v->val_labs = val_labs_create (v->width);
- v->label = NULL;
- var_clear_short_name (v);
- v->aux = NULL;
- v->aux_dtor = NULL;
- v->obs_vals = NULL;
+ /* Add dictionary info to variable. */
+ struct vardict_info vdi;
+ vdi.case_index = d->next_value_idx;
+ vdi.dict_index = d->var_cnt;
+ vdi.dict = d;
+ var_set_vardict (v, &vdi);
/* Update dictionary. */
- if (d->var_cnt >= d->var_cap)
+ if (d->var_cnt >= d->var_cap)
{
- d->var_cap = 8 + 2 * d->var_cap;
+ d->var_cap = 8 + 2 * d->var_cap;
d->var = xnrealloc (d->var, d->var_cap, sizeof *d->var);
}
- d->var[v->index] = v;
- d->var_cnt++;
+ d->var[d->var_cnt++] = v;
hsh_force_insert (d->name_tab, v);
- d->next_value_idx += v->nv;
+ if ( d->changed ) d->changed (d, d->changed_data);
+ if ( d->callbacks && d->callbacks->var_added )
+ d->callbacks->var_added (d, var_get_dict_index (v), d->cb_data);
+
+ d->next_value_idx++;
+ invalidate_proto (d);
return v;
}
+/* Creates and returns a new variable in D with the given NAME
+ and WIDTH. Returns a null pointer if the given NAME would
+ duplicate that of an existing variable in the dictionary. */
+struct variable *
+dict_create_var (struct dictionary *d, const char *name, int width)
+{
+ return (dict_lookup_var (d, name) == NULL
+ ? dict_create_var_assert (d, name, width)
+ : NULL);
+}
+
/* Creates and returns a new variable in D with the given NAME
and WIDTH. Assert-fails if the given NAME would duplicate
that of an existing variable in the dictionary. */
struct variable *
dict_create_var_assert (struct dictionary *d, const char *name, int width)
{
- struct variable *v = dict_create_var (d, name, width);
- assert (v != NULL);
- return v;
+ assert (dict_lookup_var (d, name) == NULL);
+ return add_var (d, var_create (name, width));
}
/* Creates and returns a new variable in D with name NAME, as a
- copy of existing variable OV, which need not be in D or in any
- dictionary. Returns a null pointer if the given NAME would
- duplicate that of an existing variable in the dictionary. */
+ copy of existing variable OLD_VAR, which need not be in D or
+ in any dictionary. Returns a null pointer if the given NAME
+ would duplicate that of an existing variable in the
+ dictionary. */
struct variable *
-dict_clone_var (struct dictionary *d, const struct variable *ov,
+dict_clone_var (struct dictionary *d, const struct variable *old_var,
const char *name)
{
- struct variable *nv;
-
- assert (d != NULL);
- assert (ov != NULL);
- assert (name != NULL);
-
- assert (strlen (name) >= 1);
- assert (strlen (name) <= LONG_NAME_LEN);
-
- nv = dict_create_var (d, name, ov->width);
- if (nv == NULL)
- return NULL;
-
- /* Copy most members not copied via dict_create_var().
- short_name[] is intentionally not copied, because there is
- no reason to give a new variable with potentially a new name
- the same short name. */
- nv->leave = ov->leave;
- mv_copy (&nv->miss, &ov->miss);
- nv->print = ov->print;
- nv->write = ov->write;
- val_labs_destroy (nv->val_labs);
- nv->val_labs = val_labs_copy (ov->val_labs);
- if (ov->label != NULL)
- nv->label = xstrdup (ov->label);
- nv->measure = ov->measure;
- nv->display_width = ov->display_width;
- nv->alignment = ov->alignment;
-
- return nv;
+ return (dict_lookup_var (d, name) == NULL
+ ? dict_clone_var_assert (d, old_var, name)
+ : NULL);
}
/* Creates and returns a new variable in D with name NAME, as a
- copy of existing variable OV, which need not be in D or in any
- dictionary. Assert-fails if the given NAME would duplicate
- that of an existing variable in the dictionary. */
+ copy of existing variable OLD_VAR, which need not be in D or
+ in any dictionary. Assert-fails if the given NAME would
+ duplicate that of an existing variable in the dictionary. */
struct variable *
-dict_clone_var_assert (struct dictionary *d, const struct variable *ov,
+dict_clone_var_assert (struct dictionary *d, const struct variable *old_var,
const char *name)
{
- struct variable *v = dict_clone_var (d, ov, name);
- assert (v != NULL);
- return v;
+ struct variable *new_var = var_clone (old_var);
+ assert (dict_lookup_var (d, name) == NULL);
+ var_set_name (new_var, name);
+ return add_var (d, new_var);
}
/* Returns the variable named NAME in D, or a null pointer if no
struct variable *
dict_lookup_var (const struct dictionary *d, const char *name)
{
- struct variable v;
-
- assert (d != NULL);
- assert (name != NULL);
+ struct variable *target ;
+ struct variable *result ;
+
+ if ( ! var_is_plausible_name (name, false))
+ return NULL;
+
+ target = var_create (name, 0);
+ result = hsh_find (d->name_tab, target);
+ var_destroy (target);
+
+ if ( result && var_has_vardict (result))
+ {
+ const struct vardict_info *vdi = var_get_vardict (result);
+ assert (vdi->dict == d);
+ }
- str_copy_trunc (v.name, sizeof v.name, name);
- return hsh_find (d->name_tab, &v);
+ return result;
}
/* Returns the variable named NAME in D. Assert-fails if no
bool
dict_contains_var (const struct dictionary *d, const struct variable *v)
{
- assert (d != NULL);
- assert (v != NULL);
-
- return v->index >= 0 && v->index < d->var_cnt && d->var[v->index] == v;
+ if (var_has_vardict (v))
+ {
+ const struct vardict_info *vdi = var_get_vardict (v);
+ return (vdi->dict_index >= 0
+ && vdi->dict_index < d->var_cnt
+ && d->var[vdi->dict_index] == v);
+ }
+ else
+ return false;
}
/* Compares two double pointers to variables, which should point
to elements of a struct dictionary's `var' member array. */
static int
-compare_var_ptrs (const void *a_, const void *b_, void *aux UNUSED)
+compare_var_ptrs (const void *a_, const void *b_, const void *aux UNUSED)
{
struct variable *const *a = a_;
struct variable *const *b = b_;
return *a < *b ? -1 : *a > *b;
}
+/* Sets the dict_index in V's vardict to DICT_INDEX. */
+static void
+set_var_dict_index (struct variable *v, int dict_index)
+{
+ struct vardict_info vdi = *var_get_vardict (v);
+ struct dictionary *d = vdi.dict;
+ vdi.dict_index = dict_index;
+ var_set_vardict (v, &vdi);
+
+ if ( d->changed ) d->changed (d, d->changed_data);
+ if ( d->callbacks && d->callbacks->var_changed )
+ d->callbacks->var_changed (d, dict_index, d->cb_data);
+}
+
+/* Sets the case_index in V's vardict to CASE_INDEX. */
+static void
+set_var_case_index (struct variable *v, int case_index)
+{
+ struct vardict_info vdi = *var_get_vardict (v);
+ vdi.case_index = case_index;
+ var_set_vardict (v, &vdi);
+}
+
+/* Re-sets the dict_index in the dictionary variables with
+ indexes from FROM to TO (exclusive). */
+static void
+reindex_vars (struct dictionary *d, size_t from, size_t to)
+{
+ size_t i;
+
+ for (i = from; i < to; i++)
+ set_var_dict_index (d->var[i], i);
+}
+
/* Deletes variable V from dictionary D and frees V.
This is a very bad idea if there might be any pointers to V
- from outside D. In general, no variable in default_dict
- should be deleted when any transformations are active, because
- those transformations might reference the deleted variable.
- The safest time to delete a variable is just after a procedure
- has been executed, as done by MODIFY VARS.
+ from outside D. In general, no variable in the active file's
+ dictionary should be deleted when any transformations are
+ active on the dictionary's dataset, because those
+ transformations might reference the deleted variable. The
+ safest time to delete a variable is just after a procedure has
+ been executed, as done by DELETE VARIABLES.
Pointers to V within D are not a problem, because
dict_delete_var() knows to remove V from split variables,
weights, filters, etc. */
void
-dict_delete_var (struct dictionary *d, struct variable *v)
+dict_delete_var (struct dictionary *d, struct variable *v)
{
- size_t i;
+ int dict_index = var_get_dict_index (v);
+ const int case_index = var_get_case_index (v);
+ const int width = var_get_width (v);
- assert (d != NULL);
- assert (v != NULL);
assert (dict_contains_var (d, v));
/* Delete aux data. */
var_clear_aux (v);
- /* Remove V from splits, weight, filter variables. */
- d->split_cnt = remove_equal (d->split, d->split_cnt, sizeof *d->split,
- &v, compare_var_ptrs, NULL);
+ dict_unset_split_var (d, v);
+
if (d->weight == v)
- d->weight = NULL;
+ dict_set_weight (d, NULL);
+
if (d->filter == v)
- d->filter = NULL;
+ dict_set_filter (d, NULL);
+
dict_clear_vectors (d);
/* Remove V from var array. */
- remove_element (d->var, d->var_cnt, sizeof *d->var, v->index);
+ remove_element (d->var, d->var_cnt, sizeof *d->var, dict_index);
d->var_cnt--;
- /* Update index. */
- for (i = v->index; i < d->var_cnt; i++)
- d->var[i]->index = i;
+ /* Update dict_index for each affected variable. */
+ reindex_vars (d, dict_index, d->var_cnt);
/* Update name hash. */
hsh_force_delete (d->name_tab, v);
+
/* Free memory. */
- val_labs_destroy (v->val_labs);
- cat_stored_values_destroy (v);
- free (v->label);
- free (v);
+ var_clear_vardict (v);
+ var_destroy (v);
+
+ if ( d->changed ) d->changed (d, d->changed_data);
+
+ invalidate_proto (d);
+ if (d->callbacks && d->callbacks->var_deleted )
+ d->callbacks->var_deleted (d, dict_index, case_index, width, d->cb_data);
}
/* Deletes the COUNT variables listed in VARS from D. This is
unsafe; see the comment on dict_delete_var() for details. */
-void
+void
dict_delete_vars (struct dictionary *d,
- struct variable *const *vars, size_t count)
+ struct variable *const *vars, size_t count)
{
/* FIXME: this can be done in O(count) time, but this algorithm
is O(count**2). */
dict_delete_var (d, *vars++);
}
+/* Deletes the COUNT variables in D starting at index IDX. This
+ is unsafe; see the comment on dict_delete_var() for
+ details. */
+void
+dict_delete_consecutive_vars (struct dictionary *d, size_t idx, size_t count)
+{
+ /* FIXME: this can be done in O(count) time, but this algorithm
+ is O(count**2). */
+ assert (idx + count <= d->var_cnt);
+
+ while (count-- > 0)
+ dict_delete_var (d, d->var[idx]);
+}
+
/* Deletes scratch variables from dictionary D. */
void
dict_delete_scratch_vars (struct dictionary *d)
assert (d != NULL);
for (i = 0; i < d->var_cnt; )
- if (dict_class_from_id (d->var[i]->name) == DC_SCRATCH)
+ if (var_get_dict_class (d->var[i]) == DC_SCRATCH)
dict_delete_var (d, d->var[i]);
else
i++;
if any, retain their relative positions. Runs in time linear
in the distance moved. */
void
-dict_reorder_var (struct dictionary *d, struct variable *v,
- size_t new_index)
+dict_reorder_var (struct dictionary *d, struct variable *v, size_t new_index)
{
- size_t min_idx, max_idx;
- size_t i;
-
- assert (d != NULL);
- assert (v != NULL);
- assert (dict_contains_var (d, v));
- assert (new_index < d->var_cnt);
+ size_t old_index = var_get_dict_index (v);
- move_element (d->var, d->var_cnt, sizeof *d->var, v->index, new_index);
-
- min_idx = min (v->index, new_index);
- max_idx = max (v->index, new_index);
- for (i = min_idx; i <= max_idx; i++)
- d->var[i]->index = i;
+ assert (new_index < d->var_cnt);
+ move_element (d->var, d->var_cnt, sizeof *d->var, old_index, new_index);
+ reindex_vars (d, MIN (old_index, new_index), MAX (old_index, new_index) + 1);
}
/* Reorders the variables in D, placing the COUNT variables
listed in ORDER in that order at the beginning of D. The
other variables in D, if any, retain their relative
positions. */
-void
+void
dict_reorder_vars (struct dictionary *d,
- struct variable *const *order, size_t count)
+ struct variable *const *order, size_t count)
{
struct variable **new_var;
size_t i;
-
+
assert (d != NULL);
assert (count == 0 || order != NULL);
assert (count <= d->var_cnt);
new_var = xnmalloc (d->var_cnt, sizeof *new_var);
memcpy (new_var, order, count * sizeof *new_var);
- for (i = 0; i < count; i++)
+ for (i = 0; i < count; i++)
{
- assert (d->var[order[i]->index] != NULL);
- d->var[order[i]->index] = NULL;
- order[i]->index = i;
+ size_t index = var_get_dict_index (order[i]);
+ assert (d->var[index] == order[i]);
+ d->var[index] = NULL;
+ set_var_dict_index (order[i], i);
}
for (i = 0; i < d->var_cnt; i++)
if (d->var[i] != NULL)
{
assert (count < d->var_cnt);
new_var[count] = d->var[i];
- new_var[count]->index = count;
+ set_var_dict_index (new_var[count], count);
count++;
}
free (d->var);
d->var = new_var;
}
+/* Changes the name of variable V in dictionary D to NEW_NAME. */
+static void
+rename_var (struct dictionary *d, struct variable *v, const char *new_name)
+{
+ struct vardict_info vdi;
+
+ assert (dict_contains_var (d, v));
+
+ vdi = *var_get_vardict (v);
+ var_clear_vardict (v);
+ var_set_name (v, new_name);
+ var_set_vardict (v, &vdi);
+}
+
/* Changes the name of V in D to name NEW_NAME. Assert-fails if
a variable named NEW_NAME is already in D, except that
NEW_NAME may be the same as V's existing name. */
-void
+void
dict_rename_var (struct dictionary *d, struct variable *v,
- const char *new_name)
+ const char *new_name)
{
- assert (d != NULL);
- assert (v != NULL);
- assert (new_name != NULL);
- assert (var_is_plausible_name (new_name, false));
- assert (dict_contains_var (d, v));
- assert (!compare_var_names (v->name, new_name, NULL)
+ assert (!strcasecmp (var_get_name (v), new_name)
|| dict_lookup_var (d, new_name) == NULL);
hsh_force_delete (d->name_tab, v);
- str_copy_trunc (v->name, sizeof v->name, new_name);
+ rename_var (d, v, new_name);
hsh_force_insert (d->name_tab, v);
- if (get_algorithm () == ENHANCED)
- var_clear_short_name (v);
+ if (settings_get_algorithm () == ENHANCED)
+ var_clear_short_names (v);
+
+ if ( d->changed ) d->changed (d, d->changed_data);
+ if ( d->callbacks && d->callbacks->var_changed )
+ d->callbacks->var_changed (d, var_get_dict_index (v), d->cb_data);
}
/* Renames COUNT variables specified in VARS to the names given
is returned. */
bool
dict_rename_vars (struct dictionary *d,
- struct variable **vars, char **new_names,
- size_t count, char **err_name)
+ struct variable **vars, char **new_names, size_t count,
+ char **err_name)
{
+ struct pool *pool;
char **old_names;
size_t i;
- bool success = true;
- assert (d != NULL);
assert (count == 0 || vars != NULL);
assert (count == 0 || new_names != NULL);
+ /* Save the names of the variables to be renamed. */
+ pool = pool_create ();
+ old_names = pool_nalloc (pool, count, sizeof *old_names);
+ for (i = 0; i < count; i++)
+ old_names[i] = pool_strdup (pool, var_get_name (vars[i]));
+
/* Remove the variables to be renamed from the name hash,
- save their names, and rename them. */
- old_names = xnmalloc (count, sizeof *old_names);
- for (i = 0; i < count; i++)
+ and rename them. */
+ for (i = 0; i < count; i++)
{
- assert (d->var[vars[i]->index] == vars[i]);
- assert (var_is_plausible_name (new_names[i], false));
hsh_force_delete (d->name_tab, vars[i]);
- old_names[i] = xstrdup (vars[i]->name);
- strcpy (vars[i]->name, new_names[i]);
+ rename_var (d, vars[i], new_names[i]);
}
/* Add the renamed variables back into the name hash,
checking for conflicts. */
for (i = 0; i < count; i++)
- {
- assert (new_names[i] != NULL);
- assert (*new_names[i] != '\0');
- assert (strlen (new_names[i]) >= 1);
- assert (strlen (new_names[i]) <= LONG_NAME_LEN);
+ if (hsh_insert (d->name_tab, vars[i]) != NULL)
+ {
+ /* There is a name conflict.
+ Back out all the name changes that have already
+ taken place, and indicate failure. */
+ size_t fail_idx = i;
+ if (err_name != NULL)
+ *err_name = new_names[i];
+
+ for (i = 0; i < fail_idx; i++)
+ hsh_force_delete (d->name_tab, vars[i]);
+
+ for (i = 0; i < count; i++)
+ {
+ rename_var (d, vars[i], old_names[i]);
+ hsh_force_insert (d->name_tab, vars[i]);
+ }
+
+ pool_destroy (pool);
+ return false;
+ }
+
+ /* Clear short names. */
+ if (settings_get_algorithm () == ENHANCED)
+ for (i = 0; i < count; i++)
+ var_clear_short_names (vars[i]);
+
+ pool_destroy (pool);
+ return true;
+}
+
+/* Returns true if a variable named NAME may be inserted in DICT;
+ that is, if there is not already a variable with that name in
+ DICT and if NAME is not a reserved word. (The caller's checks
+ have already verified that NAME is otherwise acceptable as a
+ variable name.) */
+static bool
+var_name_is_insertable (const struct dictionary *dict, const char *name)
+{
+ return (dict_lookup_var (dict, name) == NULL
+ && lex_id_to_token (ss_cstr (name)) == T_ID);
+}
+
+static bool
+make_hinted_name (const struct dictionary *dict, const char *hint,
+ char name[VAR_NAME_LEN + 1])
+{
+ bool dropped = false;
+ char *cp;
- if (hsh_insert (d->name_tab, vars[i]) != NULL)
+ for (cp = name; *hint && cp < name + VAR_NAME_LEN; hint++)
+ {
+ if (cp == name
+ ? lex_is_id1 (*hint) && *hint != '$'
+ : lex_is_idn (*hint))
{
- /* There is a name conflict.
- Back out all the name changes that have already
- taken place, and indicate failure. */
- size_t fail_idx = i;
- if (err_name != NULL)
- *err_name = new_names[i];
-
- for (i = 0; i < fail_idx; i++)
- hsh_force_delete (d->name_tab, vars[i]);
-
- for (i = 0; i < count; i++)
+ if (dropped)
{
- strcpy (vars[i]->name, old_names[i]);
- hsh_force_insert (d->name_tab, vars[i]);
+ *cp++ = '_';
+ dropped = false;
}
+ if (cp < name + VAR_NAME_LEN)
+ *cp++ = *hint;
+ }
+ else if (cp > name)
+ dropped = true;
+ }
+ *cp = '\0';
- success = false;
- goto done;
+ if (name[0] != '\0')
+ {
+ size_t len = strlen (name);
+ unsigned long int i;
+
+ if (var_name_is_insertable (dict, name))
+ return true;
+
+ for (i = 0; i < ULONG_MAX; i++)
+ {
+ char suffix[INT_BUFSIZE_BOUND (i) + 1];
+ int ofs;
+
+ suffix[0] = '_';
+ if (!str_format_26adic (i + 1, &suffix[1], sizeof suffix - 1))
+ NOT_REACHED ();
+
+ ofs = MIN (VAR_NAME_LEN - strlen (suffix), len);
+ strcpy (&name[ofs], suffix);
+
+ if (var_name_is_insertable (dict, name))
+ return true;
}
}
- /* Clear short names. */
- if (get_algorithm () == ENHANCED)
- for (i = 0; i < count; i++)
- var_clear_short_name (vars[i]);
+ return false;
+}
- done:
- /* Free the old names we kept around. */
- for (i = 0; i < count; i++)
- free (old_names[i]);
- free (old_names);
+static bool
+make_numeric_name (const struct dictionary *dict, unsigned long int *num_start,
+ char name[VAR_NAME_LEN + 1])
+{
+ unsigned long int number;
+
+ for (number = num_start != NULL ? MAX (*num_start, 1) : 1;
+ number < ULONG_MAX;
+ number++)
+ {
+ sprintf (name, "VAR%03lu", number);
+ if (dict_lookup_var (dict, name) == NULL)
+ {
+ if (num_start != NULL)
+ *num_start = number + 1;
+ return true;
+ }
+ }
+
+ if (num_start != NULL)
+ *num_start = ULONG_MAX;
+ return false;
+}
+
+
+/* Attempts to devise a variable name unique within DICT.
+ Returns true if successful, in which case the new variable
+ name is stored into NAME. Returns false if all names that can
+ be generated have already been taken. (Returning false is
+ quite unlikely: at least ULONG_MAX unique names can be
+ generated.)
+
+ HINT, if it is non-null, is used as a suggestion that will be
+ modified for suitability as a variable name and for
+ uniqueness.
- return success;
+ If HINT is null or entirely unsuitable, a name in the form
+ "VAR%03d" will be generated, where the smallest unused integer
+ value is used. If NUM_START is non-null, then its value is
+ used as the minimum numeric value to check, and it is updated
+ to the next value to be checked.
+ */
+bool
+dict_make_unique_var_name (const struct dictionary *dict, const char *hint,
+ unsigned long int *num_start,
+ char name[VAR_NAME_LEN + 1])
+{
+ return ((hint != NULL && make_hinted_name (dict, hint, name))
+ || make_numeric_name (dict, num_start, name));
}
/* Returns the weighting variable in dictionary D, or a null
pointer if the dictionary is unweighted. */
struct variable *
-dict_get_weight (const struct dictionary *d)
+dict_get_weight (const struct dictionary *d)
{
assert (d != NULL);
assert (d->weight == NULL || dict_contains_var (d, d->weight));
-
+
return d->weight;
}
-/* Returns the value of D's weighting variable in case C, except that a
- negative weight is returned as 0. Returns 1 if the dictionary is
- unweighted. Will warn about missing, negative, or zero values if
- warn_on_invalid is nonzero. The function will set warn_on_invalid to zero
- if an invalid weight is found. */
+/* Returns the value of D's weighting variable in case C, except
+ that a negative weight is returned as 0. Returns 1 if the
+ dictionary is unweighted. Will warn about missing, negative,
+ or zero values if *WARN_ON_INVALID is true. The function will
+ set *WARN_ON_INVALID to false if an invalid weight is
+ found. */
double
-dict_get_case_weight (const struct dictionary *d, const struct ccase *c,
- int *warn_on_invalid)
+dict_get_case_weight (const struct dictionary *d, const struct ccase *c,
+ bool *warn_on_invalid)
{
assert (d != NULL);
assert (c != NULL);
if (d->weight == NULL)
return 1.0;
- else
+ else
{
- double w = case_num (c, d->weight->fv);
- if (w < 0.0 || mv_is_num_missing (&d->weight->miss, w))
+ double w = case_num (c, d->weight);
+ if (w < 0.0 || var_is_num_missing (d->weight, w, MV_ANY))
w = 0.0;
- if ( w == 0.0 && *warn_on_invalid ) {
- *warn_on_invalid = 0;
+ if ( w == 0.0 && warn_on_invalid != NULL && *warn_on_invalid ) {
+ *warn_on_invalid = false;
msg (SW, _("At least one case in the data file had a weight value "
"that was user-missing, system-missing, zero, or "
"negative. These case(s) were ignored."));
/* Sets the weighting variable of D to V, or turning off
weighting if V is a null pointer. */
void
-dict_set_weight (struct dictionary *d, struct variable *v)
+dict_set_weight (struct dictionary *d, struct variable *v)
{
assert (d != NULL);
assert (v == NULL || dict_contains_var (d, v));
- assert (v == NULL || v->type == NUMERIC);
+ assert (v == NULL || var_is_numeric (v));
d->weight = v;
+
+ if (d->changed) d->changed (d, d->changed_data);
+ if ( d->callbacks && d->callbacks->weight_changed )
+ d->callbacks->weight_changed (d,
+ v ? var_get_dict_index (v) : -1,
+ d->cb_data);
}
/* Returns the filter variable in dictionary D (see cmd_filter())
or a null pointer if the dictionary is unfiltered. */
struct variable *
-dict_get_filter (const struct dictionary *d)
+dict_get_filter (const struct dictionary *d)
{
assert (d != NULL);
assert (d->filter == NULL || dict_contains_var (d, d->filter));
-
+
return d->filter;
}
{
assert (d != NULL);
assert (v == NULL || dict_contains_var (d, v));
+ assert (v == NULL || var_is_numeric (v));
d->filter = v;
+
+ if (d->changed) d->changed (d, d->changed_data);
+ if ( d->callbacks && d->callbacks->filter_changed )
+ d->callbacks->filter_changed (d,
+ v ? var_get_dict_index (v) : -1,
+ d->cb_data);
}
/* Returns the case limit for dictionary D, or zero if the number
of cases is unlimited. */
-size_t
-dict_get_case_limit (const struct dictionary *d)
+casenumber
+dict_get_case_limit (const struct dictionary *d)
{
assert (d != NULL);
/* Sets CASE_LIMIT as the case limit for dictionary D. Use
0 for CASE_LIMIT to indicate no limit. */
void
-dict_set_case_limit (struct dictionary *d, size_t case_limit)
+dict_set_case_limit (struct dictionary *d, casenumber case_limit)
{
assert (d != NULL);
d->case_limit = case_limit;
}
-/* Returns the index of the next value to be added to D. This
- value is the number of `union value's that need to be
+/* Returns the prototype used for cases created by dictionary D. */
+const struct caseproto *
+dict_get_proto (const struct dictionary *d_)
+{
+ struct dictionary *d = (struct dictionary *) d_;
+ if (d->proto == NULL)
+ {
+ size_t i;
+
+ d->proto = caseproto_create ();
+ d->proto = caseproto_reserve (d->proto, d->var_cnt);
+ for (i = 0; i < d->var_cnt; i++)
+ d->proto = caseproto_set_width (d->proto,
+ var_get_case_index (d->var[i]),
+ var_get_width (d->var[i]));
+ }
+ return d->proto;
+}
+
+/* Returns the case index of the next value to be added to D.
+ This value is the number of `union value's that need to be
allocated to store a case for dictionary D. */
int
-dict_get_next_value_idx (const struct dictionary *d)
+dict_get_next_value_idx (const struct dictionary *d)
{
assert (d != NULL);
/* Returns the number of bytes needed to store a case for
dictionary D. */
size_t
-dict_get_case_size (const struct dictionary *d)
+dict_get_case_size (const struct dictionary *d)
{
assert (d != NULL);
return sizeof (union value) * dict_get_next_value_idx (d);
}
-/* Deletes scratch variables in dictionary D and reassigns values
- so that fragmentation is eliminated. */
+/* Reassigns values in dictionary D so that fragmentation is
+ eliminated. */
void
-dict_compact_values (struct dictionary *d)
+dict_compact_values (struct dictionary *d)
{
size_t i;
d->next_value_idx = 0;
- for (i = 0; i < d->var_cnt; )
+ for (i = 0; i < d->var_cnt; i++)
{
struct variable *v = d->var[i];
-
- if (dict_class_from_id (v->name) != DC_SCRATCH)
- {
- v->fv = d->next_value_idx;
- d->next_value_idx += v->nv;
- i++;
- }
- else
- dict_delete_var (d, v);
+ set_var_case_index (v, d->next_value_idx++);
}
+ invalidate_proto (d);
}
-/* Returns the number of values that would be used by a case if
- dict_compact_values() were called. */
+/* Returns the number of values occupied by the variables in
+ dictionary D. All variables are considered if EXCLUDE_CLASSES
+ is 0, or it may contain one or more of (1u << DC_ORDINARY),
+ (1u << DC_SYSTEM), or (1u << DC_SCRATCH) to exclude the
+ corresponding type of variable.
+
+ The return value may be less than the number of values in one
+ of dictionary D's cases (as returned by
+ dict_get_next_value_idx) even if E is 0, because there may be
+ gaps in D's cases due to deleted variables. */
size_t
-dict_get_compacted_value_cnt (const struct dictionary *d)
+dict_count_values (const struct dictionary *d, unsigned int exclude_classes)
{
size_t i;
size_t cnt;
- cnt = 0;
- for (i = 0; i < d->var_cnt; i++)
- if (dict_class_from_id (d->var[i]->name) != DC_SCRATCH)
- cnt += d->var[i]->nv;
- return cnt;
-}
+ assert ((exclude_classes & ~((1u << DC_ORDINARY)
+ | (1u << DC_SYSTEM)
+ | (1u << DC_SCRATCH))) == 0);
-/* Creates and returns an array mapping from a dictionary index
- to the `fv' that the corresponding variable will have after
- calling dict_compact_values(). Scratch variables receive -1
- for `fv' because dict_compact_values() will delete them. */
-int *
-dict_get_compacted_idx_to_fv (const struct dictionary *d)
-{
- size_t i;
- size_t next_value_idx;
- int *idx_to_fv;
-
- idx_to_fv = xnmalloc (d->var_cnt, sizeof *idx_to_fv);
- next_value_idx = 0;
+ cnt = 0;
for (i = 0; i < d->var_cnt; i++)
{
- struct variable *v = d->var[i];
-
- if (dict_class_from_id (v->name) != DC_SCRATCH)
- {
- idx_to_fv[i] = next_value_idx;
- next_value_idx += v->nv;
- }
- else
- idx_to_fv[i] = -1;
+ enum dict_class class = var_get_dict_class (d->var[i]);
+ if (!(exclude_classes & (1u << class)))
+ cnt++;
}
- return idx_to_fv;
-}
-
-/* Returns true if a case for dictionary D would be smaller after
- compacting, false otherwise. Compacting a case eliminates
- "holes" between values and after the last value. Holes are
- created by deleting variables (or by scratch variables).
-
- The return value may differ from whether compacting a case
- from dictionary D would *change* the case: compacting could
- rearrange values even if it didn't reduce space
- requirements. */
-bool
-dict_compacting_would_shrink (const struct dictionary *d)
-{
- return dict_get_compacted_value_cnt (d) < dict_get_next_value_idx (d);
-}
-
-/* Returns true if a case for dictionary D would be smaller after
- compacting, false otherwise. Compacting a case eliminates
- "holes" between values and after the last value. Holes are
- created by deleting variables (or by scratch variables).
-
- The return value may differ from whether compacting a case
- from dictionary D would *shrink* the case: compacting could
- rearrange values without reducing space requirements. */
-bool
-dict_compacting_would_change (const struct dictionary *d)
-{
- size_t case_idx;
- size_t i;
-
- case_idx = 0;
- for (i = 0; i < dict_get_var_cnt (d); i++)
- {
- struct variable *v = dict_get_var (d, i);
- if (v->fv != case_idx)
- return true;
- case_idx += v->nv;
- }
- return false;
+ return cnt;
}
-\f
-/* How to copy a contiguous range of values between cases. */
-struct copy_map
- {
- size_t src_idx; /* Starting value index in source case. */
- size_t dst_idx; /* Starting value index in target case. */
- size_t cnt; /* Number of values. */
- };
-
-/* How to compact a case. */
-struct dict_compactor
- {
- struct copy_map *maps; /* Array of mappings. */
- size_t map_cnt; /* Number of mappings. */
- };
-/* Creates and returns a dict_compactor that can be used to
- compact cases for dictionary D.
+/* Returns the case prototype that would result after deleting
+ all variables from D that are not in one of the
+ EXCLUDE_CLASSES and compacting the dictionary with
+ dict_compact().
- Compacting a case eliminates "holes" between values and after
- the last value. Holes are created by deleting variables (or
- by scratch variables). */
-struct dict_compactor *
-dict_make_compactor (const struct dictionary *d)
+ The caller must unref the returned caseproto when it is no
+ longer needed. */
+struct caseproto *
+dict_get_compacted_proto (const struct dictionary *d,
+ unsigned int exclude_classes)
{
- struct dict_compactor *compactor;
- struct copy_map *map;
- size_t map_allocated;
- size_t value_idx;
+ struct caseproto *proto;
size_t i;
- compactor = xmalloc (sizeof *compactor);
- compactor->maps = NULL;
- compactor->map_cnt = 0;
- map_allocated = 0;
+ assert ((exclude_classes & ~((1u << DC_ORDINARY)
+ | (1u << DC_SYSTEM)
+ | (1u << DC_SCRATCH))) == 0);
- value_idx = 0;
- map = NULL;
- for (i = 0; i < d->var_cnt; i++)
+ proto = caseproto_create ();
+ for (i = 0; i < d->var_cnt; i++)
{
struct variable *v = d->var[i];
-
- if (dict_class_from_id (v->name) == DC_SCRATCH)
- continue;
- if (map != NULL && map->src_idx + map->cnt == v->fv)
- map->cnt += v->nv;
- else
- {
- if (compactor->map_cnt == map_allocated)
- compactor->maps = x2nrealloc (compactor->maps, &map_allocated,
- sizeof *compactor->maps);
- map = &compactor->maps[compactor->map_cnt++];
- map->src_idx = v->fv;
- map->dst_idx = value_idx;
- map->cnt = v->nv;
- }
- value_idx += v->nv;
- }
-
- return compactor;
-}
-
-/* Compacts SRC by copying it to DST according to the scheme in
- COMPACTOR.
-
- Compacting a case eliminates "holes" between values and after
- the last value. Holes are created by deleting variables (or
- by scratch variables). */
-void
-dict_compactor_compact (const struct dict_compactor *compactor,
- struct ccase *dst, const struct ccase *src)
-{
- size_t i;
-
- for (i = 0; i < compactor->map_cnt; i++)
- {
- const struct copy_map *map = &compactor->maps[i];
- case_copy (dst, map->dst_idx, src, map->src_idx, map->cnt);
+ if (!(exclude_classes & (1u << var_get_dict_class (v))))
+ proto = caseproto_add_width (proto, var_get_width (v));
}
+ return proto;
}
-
-/* Destroys COMPACTOR. */
-void
-dict_compactor_destroy (struct dict_compactor *compactor)
-{
- if (compactor != NULL)
- {
- free (compactor->maps);
- free (compactor);
- }
-}
-
+\f
/* Returns the SPLIT FILE vars (see cmd_split_file()). Call
dict_get_split_cnt() to determine how many SPLIT FILE vars
there are. Returns a null pointer if and only if there are no
SPLIT FILE vars. */
-struct variable *const *
-dict_get_split_vars (const struct dictionary *d)
+const struct variable *const *
+dict_get_split_vars (const struct dictionary *d)
{
assert (d != NULL);
-
+
return d->split;
}
/* Returns the number of SPLIT FILE vars. */
size_t
-dict_get_split_cnt (const struct dictionary *d)
+dict_get_split_cnt (const struct dictionary *d)
{
assert (d != NULL);
return d->split_cnt;
}
+/* Removes variable V, which must be in D, from D's set of split
+ variables. */
+void
+dict_unset_split_var (struct dictionary *d, struct variable *v)
+{
+ int orig_count;
+
+ assert (dict_contains_var (d, v));
+
+ orig_count = d->split_cnt;
+ d->split_cnt = remove_equal (d->split, d->split_cnt, sizeof *d->split,
+ &v, compare_var_ptrs, NULL);
+ if (orig_count != d->split_cnt)
+ {
+ if (d->changed) d->changed (d, d->changed_data);
+ /* We changed the set of split variables so invoke the
+ callback. */
+ if (d->callbacks && d->callbacks->split_changed)
+ d->callbacks->split_changed (d, d->cb_data);
+ }
+}
+
/* Sets CNT split vars SPLIT in dictionary D. */
void
dict_set_split_vars (struct dictionary *d,
assert (cnt == 0 || split != NULL);
d->split_cnt = cnt;
- d->split = xnrealloc (d->split, cnt, sizeof *d->split);
- memcpy (d->split, split, cnt * sizeof *d->split);
+ if ( cnt > 0 )
+ {
+ d->split = xnrealloc (d->split, cnt, sizeof *d->split) ;
+ memcpy (d->split, split, cnt * sizeof *d->split);
+ }
+ else
+ {
+ free (d->split);
+ d->split = NULL;
+ }
+
+ if (d->changed) d->changed (d, d->changed_data);
+ if ( d->callbacks && d->callbacks->split_changed )
+ d->callbacks->split_changed (d, d->cb_data);
}
/* Returns the file label for D, or a null pointer if D is
unlabeled (see cmd_file_label()). */
const char *
-dict_get_label (const struct dictionary *d)
+dict_get_label (const struct dictionary *d)
{
assert (d != NULL);
/* Sets D's file label to LABEL, truncating it to a maximum of 60
characters. */
void
-dict_set_label (struct dictionary *d, const char *label)
+dict_set_label (struct dictionary *d, const char *label)
{
assert (d != NULL);
free (d->label);
- if (label == NULL)
- d->label = NULL;
- else if (strlen (label) < 60)
- d->label = xstrdup (label);
- else
- {
- d->label = xmalloc (61);
- memcpy (d->label, label, 60);
- d->label[60] = '\0';
- }
+ d->label = label != NULL ? xstrndup (label, 60) : NULL;
}
/* Returns the documents for D, or a null pointer if D has no
- documents (see cmd_document()).. */
+ documents. If the return value is nonnull, then the string
+ will be an exact multiple of DOC_LINE_LENGTH bytes in length,
+ with each segment corresponding to one line. */
const char *
-dict_get_documents (const struct dictionary *d)
+dict_get_documents (const struct dictionary *d)
{
- assert (d != NULL);
-
- return d->documents;
+ return ds_is_empty (&d->documents) ? NULL : ds_cstr (&d->documents);
}
/* Sets the documents for D to DOCUMENTS, or removes D's
- documents if DOCUMENT is a null pointer. */
+ documents if DOCUMENT is a null pointer. If DOCUMENTS is
+ nonnull, then it should be an exact multiple of
+ DOC_LINE_LENGTH bytes in length, with each segment
+ corresponding to one line. */
void
dict_set_documents (struct dictionary *d, const char *documents)
{
- assert (d != NULL);
+ size_t remainder;
- free (d->documents);
- if (documents == NULL)
- d->documents = NULL;
- else
- d->documents = xstrdup (documents);
+ ds_assign_cstr (&d->documents, documents != NULL ? documents : "");
+
+ /* In case the caller didn't get it quite right, pad out the
+ final line with spaces. */
+ remainder = ds_length (&d->documents) % DOC_LINE_LENGTH;
+ if (remainder != 0)
+ ds_put_char_multiple (&d->documents, ' ', DOC_LINE_LENGTH - remainder);
}
-/* Creates in D a vector named NAME that contains CNT variables
- VAR (see cmd_vector()). Returns true if successful, or
- false if a vector named NAME already exists in D. */
+/* Drops the documents from dictionary D. */
+void
+dict_clear_documents (struct dictionary *d)
+{
+ ds_clear (&d->documents);
+}
+
+/* Appends LINE to the documents in D. LINE will be truncated or
+ padded on the right with spaces to make it exactly
+ DOC_LINE_LENGTH bytes long. */
+void
+dict_add_document_line (struct dictionary *d, const char *line)
+{
+ if (strlen (line) > DOC_LINE_LENGTH)
+ {
+ /* Note to translators: "bytes" is correct, not characters */
+ msg (SW, _("Truncating document line to %d bytes."), DOC_LINE_LENGTH);
+ }
+ buf_copy_str_rpad (ds_put_uninit (&d->documents, DOC_LINE_LENGTH),
+ DOC_LINE_LENGTH, line, ' ');
+}
+
+/* Returns the number of document lines in dictionary D. */
+size_t
+dict_get_document_line_cnt (const struct dictionary *d)
+{
+ return ds_length (&d->documents) / DOC_LINE_LENGTH;
+}
+
+/* Copies document line number IDX from dictionary D into
+ LINE, trimming off any trailing white space. */
+void
+dict_get_document_line (const struct dictionary *d,
+ size_t idx, struct string *line)
+{
+ assert (idx < dict_get_document_line_cnt (d));
+ ds_assign_substring (line, ds_substr (&d->documents, idx * DOC_LINE_LENGTH,
+ DOC_LINE_LENGTH));
+ ds_rtrim (line, ss_cstr (CC_SPACES));
+}
+
+/* Creates in D a vector named NAME that contains the CNT
+ variables in VAR. Returns true if successful, or false if a
+ vector named NAME already exists in D. */
bool
dict_create_vector (struct dictionary *d,
const char *name,
- struct variable **var, size_t cnt)
+ struct variable **var, size_t cnt)
{
- struct vector *vector;
size_t i;
- assert (d != NULL);
- assert (name != NULL);
- assert (var_is_plausible_name (name, false));
assert (var != NULL);
assert (cnt > 0);
-
- if (dict_lookup_vector (d, name) != NULL)
- return false;
-
- d->vector = xnrealloc (d->vector, d->vector_cnt + 1, sizeof *d->vector);
- vector = d->vector[d->vector_cnt] = xmalloc (sizeof *vector);
- vector->idx = d->vector_cnt++;
- str_copy_trunc (vector->name, sizeof vector->name, name);
- vector->var = xnmalloc (cnt, sizeof *var);
for (i = 0; i < cnt; i++)
+ assert (dict_contains_var (d, var[i]));
+
+ if (dict_lookup_vector (d, name) == NULL)
{
- assert (dict_contains_var (d, var[i]));
- vector->var[i] = var[i];
+ d->vector = xnrealloc (d->vector, d->vector_cnt + 1, sizeof *d->vector);
+ d->vector[d->vector_cnt++] = vector_create (name, var, cnt);
+ return true;
}
- vector->cnt = cnt;
-
- return true;
+ else
+ return false;
+}
+
+/* Creates in D a vector named NAME that contains the CNT
+ variables in VAR. A vector named NAME must not already exist
+ in D. */
+void
+dict_create_vector_assert (struct dictionary *d,
+ const char *name,
+ struct variable **var, size_t cnt)
+{
+ assert (dict_lookup_vector (d, name) == NULL);
+ dict_create_vector (d, name, var, cnt);
}
/* Returns the vector in D with index IDX, which must be less
than dict_get_vector_cnt (D). */
const struct vector *
-dict_get_vector (const struct dictionary *d, size_t idx)
+dict_get_vector (const struct dictionary *d, size_t idx)
{
assert (d != NULL);
assert (idx < d->vector_cnt);
/* Returns the number of vectors in D. */
size_t
-dict_get_vector_cnt (const struct dictionary *d)
+dict_get_vector_cnt (const struct dictionary *d)
{
assert (d != NULL);
/* Looks up and returns the vector within D with the given
NAME. */
const struct vector *
-dict_lookup_vector (const struct dictionary *d, const char *name)
+dict_lookup_vector (const struct dictionary *d, const char *name)
{
size_t i;
-
- assert (d != NULL);
- assert (name != NULL);
-
for (i = 0; i < d->vector_cnt; i++)
- if (!strcasecmp (d->vector[i]->name, name))
+ if (!strcasecmp (vector_get_name (d->vector[i]), name))
return d->vector[i];
return NULL;
}
/* Deletes all vectors from D. */
void
-dict_clear_vectors (struct dictionary *d)
+dict_clear_vectors (struct dictionary *d)
{
size_t i;
-
- assert (d != NULL);
- for (i = 0; i < d->vector_cnt; i++)
- {
- free (d->vector[i]->var);
- free (d->vector[i]);
- }
+ for (i = 0; i < d->vector_cnt; i++)
+ vector_destroy (d->vector[i]);
free (d->vector);
+
d->vector = NULL;
d->vector_cnt = 0;
}
-/* Compares two strings. */
-static int
-compare_strings (const void *a, const void *b, void *aux UNUSED)
+/* Returns D's attribute set. The caller may examine or modify
+ the attribute set, but must not destroy it. Destroying D or
+ calling dict_set_attributes for D will also destroy D's
+ attribute set. */
+struct attrset *
+dict_get_attributes (const struct dictionary *d)
{
- return strcmp (a, b);
+ return (struct attrset *) &d->attributes;
}
-/* Hashes a string. */
-static unsigned
-hash_string (const void *s, void *aux UNUSED)
+/* Replaces D's attributes set by a copy of ATTRS. */
+void
+dict_set_attributes (struct dictionary *d, const struct attrset *attrs)
{
- return hsh_hash_string (s);
+ attrset_destroy (&d->attributes);
+ attrset_clone (&d->attributes, attrs);
}
-/* Assigns a valid, unique short_name[] to each variable in D.
- Each variable whose actual name is short has highest priority
- for that short name. Otherwise, variables with an existing
- short_name[] have the next highest priority for a given short
- name; if it is already taken, then the variable is treated as
- if short_name[] had been empty. Otherwise, long names are
- truncated to form short names. If that causes conflicts,
- variables are renamed as PREFIX_A, PREFIX_B, and so on. */
-void
-dict_assign_short_names (struct dictionary *d)
+/* Returns true if D has at least one attribute in its attribute
+ set, false if D's attribute set is empty. */
+bool
+dict_has_attributes (const struct dictionary *d)
{
- struct hsh_table *short_names;
- size_t i;
+ return attrset_count (&d->attributes) > 0;
+}
- /* Give variables whose names are short the corresponding short
- names, and clear short_names[] that conflict with a variable
- name. */
- for (i = 0; i < d->var_cnt; i++)
+/* Called from variable.c to notify the dictionary that some property of
+ the variable has changed */
+void
+dict_var_changed (const struct variable *v)
+{
+ if ( var_has_vardict (v))
{
- struct variable *v = d->var[i];
- if (strlen (v->name) <= SHORT_NAME_LEN)
- var_set_short_name (v, v->name);
- else if (dict_lookup_var (d, v->short_name) != NULL)
- var_clear_short_name (v);
+ const struct vardict_info *vdi = var_get_vardict (v);
+ struct dictionary *d = vdi->dict;
+
+ if ( NULL == d)
+ return;
+
+ if (d->changed ) d->changed (d, d->changed_data);
+ if ( d->callbacks && d->callbacks->var_changed )
+ d->callbacks->var_changed (d, var_get_dict_index (v), d->cb_data);
}
+}
- /* Each variable with an assigned short_name[] now gets it
- unless there is a conflict. */
- short_names = hsh_create (d->var_cnt, compare_strings, hash_string,
- NULL, NULL);
- for (i = 0; i < d->var_cnt; i++)
+
+/* Called from variable.c to notify the dictionary that the variable's width
+ has changed */
+void
+dict_var_resized (const struct variable *v, int old_width)
+{
+ if ( var_has_vardict (v))
{
- struct variable *v = d->var[i];
- if (v->short_name[0] && hsh_insert (short_names, v->short_name) != NULL)
- var_clear_short_name (v);
+ const struct vardict_info *vdi = var_get_vardict (v);
+ struct dictionary *d;
+
+ d = vdi->dict;
+
+ if (d->changed) d->changed (d, d->changed_data);
+
+ invalidate_proto (d);
+ if ( d->callbacks && d->callbacks->var_resized )
+ d->callbacks->var_resized (d, var_get_dict_index (v), old_width,
+ d->cb_data);
}
-
- /* Now assign short names to remaining variables. */
- for (i = 0; i < d->var_cnt; i++)
+}
+
+/* Called from variable.c to notify the dictionary that the variable's display width
+ has changed */
+void
+dict_var_display_width_changed (const struct variable *v)
+{
+ if ( var_has_vardict (v))
{
- struct variable *v = d->var[i];
- if (v->short_name[0] == '\0')
- {
- int sfx;
+ const struct vardict_info *vdi = var_get_vardict (v);
+ struct dictionary *d;
- /* Form initial short_name. */
- var_set_short_name (v, v->name);
+ d = vdi->dict;
- /* Try _A, _B, ... _AA, _AB, etc., if needed. */
- for (sfx = 0; hsh_insert (short_names, v->short_name) != NULL; sfx++)
- var_set_short_name_suffix (v, v->name, sfx);
- }
+ if (d->changed) d->changed (d, d->changed_data);
+ if ( d->callbacks && d->callbacks->var_display_width_changed )
+ d->callbacks->var_display_width_changed (d, var_get_dict_index (v), d->cb_data);
}
-
- /* Get rid of hash table. */
- hsh_destroy (short_names);
}
+
projects / pspp-builds.git / blobdiff