X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fdictionary.c;h=0856df93618b4569e2809457a5617eec73efeb32;hb=e8f3c723e255ca91fe94512a809762223e6edcc3;hp=0f3e25e3285d0ac76fe6814c1a29436ed5ac196c;hpb=103646b53527f82c291f317b660ca395f0b50f0c;p=pspp-builds.git diff --git a/src/data/dictionary.c b/src/data/dictionary.c index 0f3e25e3..0856df93 100644 --- a/src/data/dictionary.c +++ b/src/data/dictionary.c @@ -23,12 +23,14 @@ #include "case.h" #include "category.h" +#include "identifier.h" #include "settings.h" #include "value-labels.h" #include "vardict.h" #include "variable.h" #include "vector.h" #include +#include #include #include #include @@ -36,6 +38,7 @@ #include #include +#include "intprops.h" #include "minmax.h" #include "xalloc.h" @@ -53,7 +56,7 @@ struct dictionary 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. */ struct string documents; /* Documents, as a string. */ struct vector **vector; /* Vectors of variables. */ @@ -257,21 +260,26 @@ dict_get_var (const struct dictionary *d, size_t idx) return d->var[idx]; } -inline void +/* 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 (const struct dictionary *d, const struct variable ***vars, - size_t *cnt, unsigned exclude_classes) + size_t *cnt, enum dict_class exclude) { - dict_get_vars_mutable (d, (struct variable ***) vars, cnt, exclude_classes); + 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 - 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_mutable (const struct dictionary *d, struct variable ***vars, - size_t *cnt, unsigned exclude_classes) + size_t *cnt, enum dict_class exclude) { size_t count; size_t i; @@ -279,15 +287,13 @@ dict_get_vars_mutable (const struct dictionary *d, struct variable ***vars, 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++) { - enum dict_class class = dict_class_from_id (var_get_name (d->var[i])); - if (!(exclude_classes & (1u << class))) + enum dict_class class = var_get_dict_class (d->var[i]); + if (!(class & exclude)) count++; } @@ -295,8 +301,8 @@ dict_get_vars_mutable (const struct dictionary *d, struct variable ***vars, *cnt = 0; for (i = 0; i < d->var_cnt; i++) { - enum dict_class class = dict_class_from_id (var_get_name (d->var[i])); - if (!(exclude_classes & (1u << class))) + enum dict_class class = var_get_dict_class (d->var[i]); + if (!(class & exclude)) (*vars)[(*cnt)++] = d->var[i]; } assert (*cnt == count); @@ -480,7 +486,7 @@ reindex_vars (struct dictionary *d, size_t from, size_t to) 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 MODIFY VARS. + 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, @@ -567,7 +573,7 @@ dict_delete_scratch_vars (struct dictionary *d) assert (d != NULL); for (i = 0; i < d->var_cnt; ) - if (dict_class_from_id (var_get_name (d->var[i])) == DC_SCRATCH) + if (var_get_dict_class (d->var[i]) == DC_SCRATCH) dict_delete_var (d, d->var[i]); else i++; @@ -650,7 +656,7 @@ dict_rename_var (struct dictionary *d, struct variable *v, rename_var (d, v, new_name); hsh_force_insert (d->name_tab, v); - if (get_algorithm () == ENHANCED) + if (settings_get_algorithm () == ENHANCED) var_clear_short_names (v); if ( d->callbacks && d->callbacks->var_changed ) @@ -715,7 +721,7 @@ dict_rename_vars (struct dictionary *d, } /* Clear short names. */ - if (get_algorithm () == ENHANCED) + if (settings_get_algorithm () == ENHANCED) for (i = 0; i < count; i++) var_clear_short_names (vars[i]); @@ -723,6 +729,111 @@ dict_rename_vars (struct dictionary *d, return true; } +static bool +make_hinted_name (const struct dictionary *dict, const char *hint, + char name[VAR_NAME_LEN + 1]) +{ + bool dropped = false; + char *cp; + + for (cp = name; *hint && cp < name + VAR_NAME_LEN; hint++) + { + if (cp == name + ? lex_is_id1 (*hint) && *hint != '$' + : lex_is_idn (*hint)) + { + if (dropped) + { + *cp++ = '_'; + dropped = false; + } + if (cp < name + VAR_NAME_LEN) + *cp++ = *hint; + } + else if (cp > name) + dropped = true; + } + *cp = '\0'; + + if (name[0] != '\0') + { + size_t len = strlen (name); + unsigned long int i; + + if (dict_lookup_var (dict, name) == NULL) + 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 (dict_lookup_var (dict, name) == NULL) + return true; + } + } + + return false; +} + +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. + + 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 * @@ -734,11 +845,12 @@ dict_get_weight (const struct dictionary *d) 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 true. The function will set warn_on_invalid to false - 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, bool *warn_on_invalid) @@ -798,6 +910,7 @@ dict_set_filter (struct dictionary *d, struct variable *v) { assert (d != NULL); assert (v == NULL || dict_contains_var (d, v)); + assert (v == NULL || var_is_numeric (v)); d->filter = v; @@ -809,7 +922,7 @@ dict_set_filter (struct dictionary *d, struct variable *v) /* Returns the case limit for dictionary D, or zero if the number of cases is unlimited. */ -size_t +casenumber dict_get_case_limit (const struct dictionary *d) { assert (d != NULL); @@ -820,7 +933,7 @@ dict_get_case_limit (const struct dictionary *d) /* 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); @@ -913,7 +1026,7 @@ dict_count_values (const struct dictionary *d, unsigned int exclude_classes) cnt = 0; for (i = 0; i < d->var_cnt; i++) { - enum dict_class class = dict_class_from_id (var_get_name (d->var[i])); + enum dict_class class = var_get_dict_class (d->var[i]); if (!(exclude_classes & (1u << class))) cnt += var_get_value_cnt (d->var[i]); } @@ -941,20 +1054,25 @@ dict_get_split_cnt (const struct dictionary *d) return d->split_cnt; } -/* Removes variable V from the set of split variables in dictionary D */ +/* 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) +dict_unset_split_var (struct dictionary *d, struct variable *v) { - const int count = d->split_cnt; - d->split_cnt = remove_equal (d->split, d->split_cnt, sizeof *d->split, - &v, compare_var_ptrs, NULL); + int orig_count; - if ( count == d->split_cnt) - return; + assert (dict_contains_var (d, v)); - if ( d->callbacks && d->callbacks->split_changed ) - d->callbacks->split_changed (d, d->cb_data); + 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) + { + /* 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. */ @@ -999,16 +1117,7 @@ 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