X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fvariable.c;h=971154368cb7ae5ede7b21d4f88374e6f1e4b752;hb=01cbd30a9b6d38fae395620d2fcc49052d8e8b0f;hp=dcd91800022cd26a55ea10b719ee2cdb620a1898;hpb=679e253a4777564db48ccec70b8ce2f6e5b7e46b;p=pspp diff --git a/src/data/variable.c b/src/data/variable.c index dcd9180002..971154368c 100644 --- a/src/data/variable.c +++ b/src/data/variable.c @@ -1,5 +1,5 @@ /* PSPP - a program for statistical analysis. - Copyright (C) 1997-9, 2000, 2006 Free Software Foundation, Inc. + Copyright (C) 1997-9, 2000, 2006, 2009, 2010, 2011, 2012, 2013, 2014 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 @@ -15,26 +15,30 @@ along with this program. If not, see . */ #include -#include "variable.h" + +#include "data/variable.h" #include -#include "category.h" -#include "data-out.h" -#include "format.h" -#include "dictionary.h" -#include "identifier.h" -#include "missing-values.h" -#include "value-labels.h" -#include "vardict.h" - -#include -#include -#include -#include -#include -#include -#include +#include "data/attributes.h" +#include "data/data-out.h" +#include "data/dictionary.h" +#include "data/format.h" +#include "data/identifier.h" +#include "data/missing-values.h" +#include "data/settings.h" +#include "data/value-labels.h" +#include "data/vardict.h" +#include "libpspp/assertion.h" +#include "libpspp/compiler.h" +#include "libpspp/hash-functions.h" +#include "libpspp/i18n.h" +#include "libpspp/message.h" +#include "libpspp/misc.h" +#include "libpspp/str.h" + +#include "gl/minmax.h" +#include "gl/xalloc.h" #include "gettext.h" #define _(msgid) gettext (msgid) @@ -43,16 +47,18 @@ struct variable { /* Dictionary information. */ - char name[LONG_NAME_LEN + 1]; /* Variable name. Mixed case. */ + char *name; /* Variable name. Mixed case. */ int width; /* 0 for numeric, otherwise string width. */ struct missing_values miss; /* Missing values. */ struct fmt_spec print; /* Default format for PRINT. */ struct fmt_spec write; /* Default format for WRITE. */ struct val_labs *val_labs; /* Value labels. */ char *label; /* Variable label. */ + struct string name_and_label; /* The name and label in the same string */ /* GUI information. */ enum measure measure; /* Nominal, ordinal, or continuous. */ + enum var_role role; /* Intended use. */ int display_width; /* Width of data editor column. */ enum alignment alignment; /* Alignment of data in GUI. */ @@ -60,37 +66,23 @@ struct variable bool leave; /* Leave value from case to case? */ /* Data for use by containing dictionary. */ - struct vardict_info vardict; + struct vardict_info *vardict; /* Used only for system and portable file input and output. See short-names.h. */ char **short_names; size_t short_name_cnt; - /* Each command may use these fields as needed. */ - void *aux; - void (*aux_dtor) (struct variable *); - - /* Values of a categorical variable. Procedures need - vectors with binary entries, so any variable of type ALPHA will - have its values stored here. */ - struct cat_vals *obs_vals; + /* Custom attributes. */ + struct attrset attributes; }; + -/* Returns true if VAR_TYPE is a valid variable type. */ -bool -var_type_is_valid (enum var_type var_type) -{ - return var_type == VAR_NUMERIC || var_type == VAR_STRING; -} +static void var_set_print_format_quiet (struct variable *v, const struct fmt_spec *print); +static void var_set_write_format_quiet (struct variable *v, const struct fmt_spec *write); +static void var_set_label_quiet (struct variable *v, const char *label); +static void var_set_name_quiet (struct variable *v, const char *name); -/* Returns the variable type for the given width. */ -enum var_type -var_type_from_width (int width) -{ - return width != 0 ? VAR_STRING : VAR_NUMERIC; -} - /* Creates and returns a new variable with the given NAME and WIDTH and other fields initialized to default values. The variable is not added to a dictionary; for that, use @@ -99,70 +91,27 @@ struct variable * var_create (const char *name, int width) { struct variable *v; + enum val_type type; assert (width >= 0 && width <= MAX_STRING); - v = xmalloc (sizeof *v); - v->vardict.dict_index = v->vardict.case_index = -1; - var_set_name (v, name); + v = xzalloc (sizeof *v); + var_set_name_quiet (v, name); v->width = width; mv_init (&v->miss, width); v->leave = var_must_leave (v); - if (var_is_numeric (v)) - { - v->print = fmt_for_output (FMT_F, 8, 2); - v->alignment = ALIGN_RIGHT; - v->measure = MEASURE_SCALE; - } - else - { - v->print = fmt_for_output (FMT_A, var_get_width (v), 0); - v->alignment = ALIGN_LEFT; - v->measure = MEASURE_NOMINAL; - } + type = val_type_from_width (width); + v->alignment = var_default_alignment (type); + v->measure = var_default_measure (type); + v->role = ROLE_INPUT; v->display_width = var_default_display_width (width); - v->write = v->print; - v->val_labs = NULL; - v->label = NULL; - v->short_names = NULL; - v->short_name_cnt = 0; - v->aux = NULL; - v->aux_dtor = NULL; - v->obs_vals = NULL; + v->print = v->write = var_default_formats (width); + attrset_init (&v->attributes); + ds_init_empty (&v->name_and_label); return v; } -/* Creates and returns a clone of OLD_VAR. Most properties of - the new variable are copied from OLD_VAR, except: - - - The variable's short name is not copied, because there is - no reason to give a new variable with potentially a new - name the same short name. - - - The new variable is not added to OLD_VAR's dictionary by - default. Use dict_clone_var, instead, to do that. - - - Auxiliary data and obs_vals are not copied. */ -struct variable * -var_clone (const struct variable *old_var) -{ - struct variable *new_var = var_create (var_get_name (old_var), - var_get_width (old_var)); - - var_set_missing_values (new_var, var_get_missing_values (old_var)); - var_set_print_format (new_var, var_get_print_format (old_var)); - var_set_write_format (new_var, var_get_write_format (old_var)); - var_set_value_labels (new_var, var_get_value_labels (old_var)); - var_set_label (new_var, var_get_label (old_var)); - var_set_measure (new_var, var_get_measure (old_var)); - var_set_display_width (new_var, var_get_display_width (old_var)); - var_set_alignment (new_var, var_get_alignment (old_var)); - var_set_leave (new_var, var_get_leave (old_var)); - - return new_var; -} - /* Destroys variable V. V must not belong to a dictionary. If it does, use dict_delete_var instead. */ @@ -172,120 +121,59 @@ var_destroy (struct variable *v) if (v != NULL) { assert (!var_has_vardict (v)); - cat_stored_values_destroy (v->obs_vals); - var_clear_aux (v); + mv_destroy (&v->miss); + var_clear_short_names (v); val_labs_destroy (v->val_labs); - var_clear_label (v); + var_set_label_quiet (v, NULL); + attrset_destroy (var_get_attributes (v)); + free (v->name); + ds_destroy (&v->name_and_label); free (v); } } /* Variable names. */ -/* Return variable V's name. */ +/* Return variable V's name, as a UTF-8 encoded string. */ const char * var_get_name (const struct variable *v) { return v->name; } -/* Sets V's name to NAME. + + +/* Sets V's name to NAME, a UTF-8 encoded string. Do not use this function for a variable in a dictionary. Use dict_rename_var instead. */ -void -var_set_name (struct variable *v, const char *name) +static void +var_set_name_quiet (struct variable *v, const char *name) { - assert (v->vardict.dict_index == -1); - assert (var_is_plausible_name (name, false)); + assert (!var_has_vardict (v)); + assert (id_is_plausible (name, false)); - str_copy_trunc (v->name, sizeof v->name, name); - dict_var_changed (v); + free (v->name); + v->name = xstrdup (name); + ds_destroy (&v->name_and_label); + ds_init_empty (&v->name_and_label); } -/* Returns true if NAME is an acceptable name for a variable, - false otherwise. If ISSUE_ERROR is true, issues an - explanatory error message on failure. */ -bool -var_is_valid_name (const char *name, bool issue_error) +/* Sets V's name to NAME, a UTF-8 encoded string. + Do not use this function for a variable in a dictionary. Use + dict_rename_var instead. */ +void +var_set_name (struct variable *v, const char *name) { - bool plausible; - size_t length, i; - - assert (name != NULL); - - /* Note that strlen returns number of BYTES, not the number of - CHARACTERS */ - length = strlen (name); - - plausible = var_is_plausible_name(name, issue_error); - - if ( ! plausible ) - return false; - - - if (!lex_is_id1 (name[0])) - { - if (issue_error) - msg (SE, _("Character `%c' (in %s) may not appear " - "as the first character in a variable name."), - name[0], name); - return false; - } - - - for (i = 0; i < length; i++) - { - if (!lex_is_idn (name[i])) - { - if (issue_error) - msg (SE, _("Character `%c' (in %s) may not appear in " - "a variable name."), - name[i], name); - return false; - } - } - - return true; + struct variable *ov = var_clone (v); + var_set_name_quiet (v, name); + dict_var_changed (v, VAR_TRAIT_NAME, ov); } -/* Returns true if NAME is an plausible name for a variable, - false otherwise. If ISSUE_ERROR is true, issues an - explanatory error message on failure. - This function makes no use of LC_CTYPE. -*/ -bool -var_is_plausible_name (const char *name, bool issue_error) +/* Returns VAR's dictionary class. */ +enum dict_class +var_get_dict_class (const struct variable *var) { - size_t length; - - assert (name != NULL); - - /* Note that strlen returns number of BYTES, not the number of - CHARACTERS */ - length = strlen (name); - if (length < 1) - { - if (issue_error) - msg (SE, _("Variable name cannot be empty string.")); - return false; - } - else if (length > LONG_NAME_LEN) - { - if (issue_error) - msg (SE, _("Variable name %s exceeds %d-character limit."), - name, (int) LONG_NAME_LEN); - return false; - } - - if (lex_id_to_token (ss_cstr (name)) != T_ID) - { - if (issue_error) - msg (SE, _("`%s' may not be used as a variable name because it " - "is a reserved word."), name); - return false; - } - - return true; + return dict_class_from_id (var->name); } /* A hsh_compare_func that orders variables A and B by their @@ -296,7 +184,7 @@ compare_vars_by_name (const void *a_, const void *b_, const void *aux UNUSED) const struct variable *a = a_; const struct variable *b = b_; - return strcasecmp (a->name, b->name); + return utf8_strcasecmp (a->name, b->name); } /* A hsh_hash_func that hashes variable V based on its name. */ @@ -305,7 +193,7 @@ hash_var_by_name (const void *v_, const void *aux UNUSED) { const struct variable *v = v_; - return hsh_hash_case_string (v->name); + return utf8_hash_case_string (v->name, 0); } /* A hsh_compare_func that orders pointers to variables A and B @@ -317,7 +205,21 @@ compare_var_ptrs_by_name (const void *a_, const void *b_, struct variable *const *a = a_; struct variable *const *b = b_; - return strcasecmp (var_get_name (*a), var_get_name (*b)); + return utf8_strcasecmp (var_get_name (*a), var_get_name (*b)); +} + +/* A hsh_compare_func that orders pointers to variables A and B + by their dictionary indexes. */ +int +compare_var_ptrs_by_dict_index (const void *a_, const void *b_, + const void *aux UNUSED) +{ + struct variable *const *a = a_; + struct variable *const *b = b_; + size_t a_index = var_get_dict_index (*a); + size_t b_index = var_get_dict_index (*b); + + return a_index < b_index ? -1 : a_index > b_index; } /* A hsh_hash_func that hashes pointer to variable V based on its @@ -327,14 +229,14 @@ hash_var_ptr_by_name (const void *v_, const void *aux UNUSED) { struct variable *const *v = v_; - return hsh_hash_case_string (var_get_name (*v)); + return utf8_hash_case_string (var_get_name (*v), 0); } /* Returns the type of variable V. */ -enum var_type +enum val_type var_get_type (const struct variable *v) { - return var_type_from_width (v->width); + return val_type_from_width (v->width); } /* Returns the width of variable V. */ @@ -344,16 +246,26 @@ var_get_width (const struct variable *v) return v->width; } -/* Sets the width of V to WIDTH. */ void -var_set_width (struct variable *v, int new_width) +var_set_width_and_formats (struct variable *v, int new_width, + const struct fmt_spec *print, const struct fmt_spec *write) { - const int old_width = v->width; + struct variable *ov; + unsigned int traits = 0; - if (mv_is_resizable (&v->miss, new_width)) - mv_resize (&v->miss, new_width); - else - mv_init (&v->miss, new_width); + ov = var_clone (v); + + if (var_has_missing_values (v)) + { + if (mv_is_resizable (&v->miss, new_width)) + mv_resize (&v->miss, new_width); + else + { + mv_destroy (&v->miss); + mv_init (&v->miss, new_width); + } + traits |= VAR_TRAIT_MISSING_VALUES; + } if (v->val_labs != NULL) { @@ -364,61 +276,66 @@ var_set_width (struct variable *v, int new_width) val_labs_destroy (v->val_labs); v->val_labs = NULL; } + traits |= VAR_TRAIT_VALUE_LABELS; } - fmt_resize (&v->print, new_width); - fmt_resize (&v->write, new_width); + if (fmt_resize (&v->print, new_width)) + traits |= VAR_TRAIT_PRINT_FORMAT; - v->width = new_width; + if (fmt_resize (&v->write, new_width)) + traits |= VAR_TRAIT_WRITE_FORMAT; - { - const int old_val_count = value_cnt_from_width (old_width); - const int new_val_count = value_cnt_from_width (new_width); + if (v->width != new_width) + { + v->width = new_width; + traits |= VAR_TRAIT_WIDTH; + } - if ( old_val_count != new_val_count) - dict_var_resized (v, new_val_count - old_val_count); - } + if (print) + { + var_set_print_format_quiet (v, print); + traits |= VAR_TRAIT_PRINT_FORMAT; + } - dict_var_changed (v); -} + if (write) + { + var_set_write_format_quiet (v, write); + traits |= VAR_TRAIT_WRITE_FORMAT; + } -/* Returns true if variable V is numeric, false otherwise. */ -bool -var_is_numeric (const struct variable *v) -{ - return var_get_type (v) == VAR_NUMERIC; + if (traits != 0) + dict_var_changed (v, traits, ov); } -/* Returns true if variable V is a string variable, false - otherwise. */ -bool -var_is_alpha (const struct variable *v) +/* Changes the width of V to NEW_WIDTH. + This function should be used cautiously. */ +void +var_set_width (struct variable *v, int new_width) { - return var_get_type (v) == VAR_STRING; + const int old_width = v->width; + + if (old_width == new_width) + return; + + var_set_width_and_formats (v, new_width, NULL, NULL); } -/* Returns true if variable V is a short string variable, false - otherwise. */ + + + +/* Returns true if variable V is numeric, false otherwise. */ bool -var_is_short_string (const struct variable *v) +var_is_numeric (const struct variable *v) { - return v->width > 0 && v->width <= MAX_SHORT_STRING; + return var_get_type (v) == VAL_NUMERIC; } -/* Returns true if variable V is a long string variable, false +/* Returns true if variable V is a string variable, false otherwise. */ bool -var_is_long_string (const struct variable *v) -{ - return v->width > MAX_SHORT_STRING; -} - -/* Returns the number of "union value"s need to store a value of - variable V. */ -size_t -var_get_value_cnt (const struct variable *v) +var_is_alpha (const struct variable *v) { - return value_cnt_from_width (v->width); + return var_get_type (v) == VAL_STRING; } /* Returns variable V's missing values. */ @@ -432,19 +349,30 @@ var_get_missing_values (const struct variable *v) width or at least resizable to V's width. If MISS is null, then V's missing values, if any, are cleared. */ -void -var_set_missing_values (struct variable *v, const struct missing_values *miss) +static void +var_set_missing_values_quiet (struct variable *v, const struct missing_values *miss) { if (miss != NULL) { assert (mv_is_resizable (miss, v->width)); + mv_destroy (&v->miss); mv_copy (&v->miss, miss); mv_resize (&v->miss, v->width); } else - mv_init (&v->miss, v->width); + mv_clear (&v->miss); +} - dict_var_changed (v); +/* Sets variable V's missing values to MISS, which must be of V's + width or at least resizable to V's width. + If MISS is null, then V's missing values, if any, are + cleared. */ +void +var_set_missing_values (struct variable *v, const struct missing_values *miss) +{ + struct variable *ov = var_clone (v); + var_set_missing_values_quiet (v, miss); + dict_var_changed (v, VAR_TRAIT_MISSING_VALUES, ov); } /* Sets variable V to have no user-missing values. */ @@ -484,7 +412,7 @@ var_is_num_missing (const struct variable *v, double d, enum mv_class class) S[] must contain exactly as many characters as V's width. V must be a string variable. */ bool -var_is_str_missing (const struct variable *v, const char s[], +var_is_str_missing (const struct variable *v, const uint8_t s[], enum mv_class class) { return mv_is_str_missing (&v->miss, s, class); @@ -509,8 +437,8 @@ var_has_value_labels (const struct variable *v) which must have a width equal to V's width or one that can be changed to V's width. If VLS is null, then V's value labels, if any, are removed. */ -void -var_set_value_labels (struct variable *v, const struct val_labs *vls) +static void +var_set_value_labels_quiet (struct variable *v, const struct val_labs *vls) { val_labs_destroy (v->val_labs); v->val_labs = NULL; @@ -518,43 +446,57 @@ var_set_value_labels (struct variable *v, const struct val_labs *vls) if (vls != NULL) { assert (val_labs_can_set_width (vls, v->width)); - v->val_labs = val_labs_copy (vls); + v->val_labs = val_labs_clone (vls); val_labs_set_width (v->val_labs, v->width); - dict_var_changed (v); } } + +/* Sets variable V's value labels to a copy of VLS, + which must have a width equal to V's width or one that can be + changed to V's width. + If VLS is null, then V's value labels, if any, are removed. */ +void +var_set_value_labels (struct variable *v, const struct val_labs *vls) +{ + struct variable *ov = var_clone (v); + var_set_value_labels_quiet (v, vls); + dict_var_changed (v, VAR_TRAIT_LABEL, ov); +} + + /* Makes sure that V has a set of value labels, by assigning one to it if necessary. */ static void alloc_value_labels (struct variable *v) { - assert (!var_is_long_string (v)); if (v->val_labs == NULL) v->val_labs = val_labs_create (v->width); } -/* Attempts to add a value label with the given VALUE and LABEL - to V. Returns true if successful, false if VALUE has an - existing label. - V must not be a long string variable. */ +/* Attempts to add a value label with the given VALUE and UTF-8 encoded LABEL + to V. Returns true if successful, false otherwise (probably due to an + existing label). + + In LABEL, the two-byte sequence "\\n" is interpreted as a new-line. */ bool var_add_value_label (struct variable *v, const union value *value, const char *label) { alloc_value_labels (v); - return val_labs_add (v->val_labs, *value, label); + return val_labs_add (v->val_labs, value, label); } -/* Adds or replaces a value label with the given VALUE and LABEL +/* Adds or replaces a value label with the given VALUE and UTF-8 encoded LABEL to V. - V must not be a long string variable. */ + + In LABEL, the two-byte sequence "\\n" is interpreted as a new-line. */ void var_replace_value_label (struct variable *v, const union value *value, const char *label) { alloc_value_labels (v); - val_labs_replace (v->val_labs, *value, label); + val_labs_replace (v->val_labs, value, label); } /* Removes V's value labels, if any. */ @@ -564,29 +506,63 @@ var_clear_value_labels (struct variable *v) var_set_value_labels (v, NULL); } -/* Returns the label associated with VALUE for variable V, - or a null pointer if none. */ +/* Returns the label associated with VALUE for variable V, as a UTF-8 string in + a format suitable for output, or a null pointer if none. */ const char * var_lookup_value_label (const struct variable *v, const union value *value) { - return val_labs_find (v->val_labs, *value); + return val_labs_find (v->val_labs, value); } -/* Get a string representing VALUE for variable V. - That is, if VALUE has a label, return that label, - otherwise format VALUE and return the formatted string. */ -const char * -var_get_value_name (const struct variable *v, const union value *value) +/* + Append to STR the string representation of VALUE for variable V. + STR must be a pointer to an initialised struct string. +*/ +static void +append_value (const struct variable *v, const union value *value, + struct string *str) { + char *s = data_out (value, var_get_encoding (v), &v->print); + ds_put_cstr (str, s); + free (s); +} + +/* Append STR with a string representing VALUE for variable V. + That is, if VALUE has a label, append that label, + otherwise format VALUE and append the formatted string. + STR must be a pointer to an initialised struct string. +*/ +void +var_append_value_name (const struct variable *v, const union value *value, + struct string *str) +{ + enum settings_value_style style = settings_get_value_style (); const char *name = var_lookup_value_label (v, value); - if (name == NULL) + + switch (style) { - static char buf[MAX_STRING + 1]; - data_out (value, &v->print, buf); - buf[v->print.w] = '\0'; - name = buf; - } - return name; + case SETTINGS_VAL_STYLE_VALUES: + append_value (v, value, str); + break; + + case SETTINGS_VAL_STYLE_LABELS: + if (name == NULL) + append_value (v, value, str); + else + ds_put_cstr (str, name); + break; + + case SETTINGS_VAL_STYLE_BOTH: + default: + append_value (v, value, str); + if (name != NULL) + { + ds_put_cstr (str, " ("); + ds_put_cstr (str, name); + ds_put_cstr (str, ")"); + } + break; + }; } /* Print and write formats. */ @@ -599,14 +575,29 @@ var_get_print_format (const struct variable *v) } /* Sets V's print format specification to PRINT, which must be a - valid format specification for outputting a variable of V's - width. */ + valid format specification for a variable of V's width + (ordinarily an output format, but input formats are not + rejected). */ +static void +var_set_print_format_quiet (struct variable *v, const struct fmt_spec *print) +{ + if (!fmt_equal (&v->print, print)) + { + assert (fmt_check_width_compat (print, v->width)); + v->print = *print; + } +} + +/* Sets V's print format specification to PRINT, which must be a + valid format specification for a variable of V's width + (ordinarily an output format, but input formats are not + rejected). */ void var_set_print_format (struct variable *v, const struct fmt_spec *print) { - assert (fmt_check_width_compat (print, v->width)); - v->print = *print; - dict_var_changed (v); + struct variable *ov = var_clone (v); + var_set_print_format_quiet (v, print); + dict_var_changed (v, VAR_TRAIT_PRINT_FORMAT, ov); } /* Returns V's write format specification. */ @@ -617,33 +608,101 @@ var_get_write_format (const struct variable *v) } /* Sets V's write format specification to WRITE, which must be a - valid format specification for outputting a variable of V's - width. */ + valid format specification for a variable of V's width + (ordinarily an output format, but input formats are not + rejected). */ +static void +var_set_write_format_quiet (struct variable *v, const struct fmt_spec *write) +{ + if (!fmt_equal (&v->write, write)) + { + assert (fmt_check_width_compat (write, v->width)); + v->write = *write; + } +} + +/* Sets V's write format specification to WRITE, which must be a + valid format specification for a variable of V's width + (ordinarily an output format, but input formats are not + rejected). */ void var_set_write_format (struct variable *v, const struct fmt_spec *write) { - assert (fmt_check_width_compat (write, v->width)); - v->write = *write; - dict_var_changed (v); + struct variable *ov = var_clone (v); + var_set_write_format_quiet (v, write); + dict_var_changed (v, VAR_TRAIT_WRITE_FORMAT, ov); } + /* Sets V's print and write format specifications to FORMAT, - which must be a valid format specification for outputting a - variable of V's width. */ + which must be a valid format specification for a variable of + V's width (ordinarily an output format, but input formats are + not rejected). */ void var_set_both_formats (struct variable *v, const struct fmt_spec *format) { - var_set_print_format (v, format); - var_set_write_format (v, format); + struct variable *ov = var_clone (v); + var_set_print_format_quiet (v, format); + var_set_write_format_quiet (v, format); + dict_var_changed (v, VAR_TRAIT_PRINT_FORMAT | VAR_TRAIT_WRITE_FORMAT, ov); } + +/* Returns the default print and write format for a variable of + the given TYPE, as set by var_create. The return value can be + used to reset a variable's print and write formats to the + default. */ +struct fmt_spec +var_default_formats (int width) +{ + return (width == 0 + ? fmt_for_output (FMT_F, 8, 2) + : fmt_for_output (FMT_A, width, 0)); +} + + + +/* Update the combined name and label string if necessary */ +static void +update_vl_string (const struct variable *v) +{ + /* Cast away const! */ + struct string *str = (struct string *) &v->name_and_label; + + if (ds_is_empty (str)) + { + if (v->label) + ds_put_format (str, _("%s (%s)"), v->label, v->name); + else + ds_put_cstr (str, v->name); + } +} + + /* Return a string representing this variable, in the form most appropriate from a human factors perspective, that is, its variable label if it has one, otherwise its name. */ const char * var_to_string (const struct variable *v) { - return v->label != NULL ? v->label : v->name; + enum settings_var_style style = settings_get_var_style (); + + switch (style) + { + case SETTINGS_VAR_STYLE_NAMES: + return v->name; + break; + case SETTINGS_VAR_STYLE_LABELS: + return v->label != NULL ? v->label : v->name; + break; + case SETTINGS_VAR_STYLE_BOTH: + update_vl_string (v); + return ds_cstr (&v->name_and_label); + break; + default: + NOT_REACHED (); + break; + }; } /* Returns V's variable label, or a null pointer if it has none. */ @@ -653,28 +712,38 @@ var_get_label (const struct variable *v) return v->label; } -/* Sets V's variable label to LABEL, stripping off leading and - trailing white space and truncating to 255 characters. - If LABEL is a null pointer or if LABEL is an empty string - (after stripping white space), then V's variable label (if - any) is removed. */ -void -var_set_label (struct variable *v, const char *label) +/* Sets V's variable label to UTF-8 encoded string LABEL, stripping off leading + and trailing white space. If LABEL is a null pointer or if LABEL is an + empty string (after stripping white space), then V's variable label (if any) + is removed. */ +static void +var_set_label_quiet (struct variable *v, const char *label) { free (v->label); v->label = NULL; - if (label != NULL) - { - struct substring s = ss_cstr (label); - ss_trim (&s, ss_cstr (CC_SPACES)); - ss_truncate (&s, 255); - if (!ss_is_empty (s)) - v->label = ss_xstrdup (s); - dict_var_changed (v); - } + if (label != NULL && label[strspn (label, CC_SPACES)]) + v->label = xstrdup (label); + + ds_destroy (&v->name_and_label); + ds_init_empty (&v->name_and_label); +} + + + +/* Sets V's variable label to UTF-8 encoded string LABEL, stripping off leading + and trailing white space. If LABEL is a null pointer or if LABEL is an + empty string (after stripping white space), then V's variable label (if any) + is removed. */ +void +var_set_label (struct variable *v, const char *label) +{ + struct variable *ov = var_clone (v); + var_set_label_quiet (v, label); + dict_var_changed (v, VAR_TRAIT_LABEL, ov); } + /* Removes any variable label from V. */ void var_clear_label (struct variable *v) @@ -698,6 +767,47 @@ measure_is_valid (enum measure m) return m == MEASURE_NOMINAL || m == MEASURE_ORDINAL || m == MEASURE_SCALE; } +/* Returns a string version of measurement level M, for display to a user. */ +const char * +measure_to_string (enum measure m) +{ + switch (m) + { + case MEASURE_NOMINAL: + return _("Nominal"); + + case MEASURE_ORDINAL: + return _("Ordinal"); + + case MEASURE_SCALE: + return _("Scale"); + + default: + return "Invalid"; + } +} + +/* Returns a string version of measurement level M, for use in PSPP command + syntax. */ +const char * +measure_to_syntax (enum measure m) +{ + switch (m) + { + case MEASURE_NOMINAL: + return "NOMINAL"; + + case MEASURE_ORDINAL: + return "ORDINAL"; + + case MEASURE_SCALE: + return "SCALE"; + + default: + return "Invalid"; + } +} + /* Returns V's measurement level. */ enum measure var_get_measure (const struct variable *v) @@ -706,12 +816,135 @@ var_get_measure (const struct variable *v) } /* Sets V's measurement level to MEASURE. */ -void -var_set_measure (struct variable *v, enum measure measure) +static void +var_set_measure_quiet (struct variable *v, enum measure measure) { assert (measure_is_valid (measure)); v->measure = measure; - dict_var_changed (v); +} + + +/* Sets V's measurement level to MEASURE. */ +void +var_set_measure (struct variable *v, enum measure measure) +{ + struct variable *ov = var_clone (v); + var_set_measure_quiet (v, measure); + dict_var_changed (v, VAR_TRAIT_MEASURE, ov); +} + + +/* Returns the default measurement level for a variable of the + given TYPE, as set by var_create. The return value can be + used to reset a variable's measurement level to the + default. */ +enum measure +var_default_measure (enum val_type type) +{ + return type == VAL_NUMERIC ? MEASURE_SCALE : MEASURE_NOMINAL; +} + +/* Returns true if M is a valid variable role, + false otherwise. */ +bool +var_role_is_valid (enum var_role role) +{ + switch (role) + { + case ROLE_NONE: + case ROLE_INPUT: + case ROLE_TARGET: + case ROLE_BOTH: + case ROLE_PARTITION: + case ROLE_SPLIT: + return true; + + default: + return false; + } +} + +/* Returns a string version of ROLE, for display to a user. */ +const char * +var_role_to_string (enum var_role role) +{ + switch (role) + { + case ROLE_INPUT: + return _("Input"); + + case ROLE_TARGET: + return _("Output"); + + case ROLE_BOTH: + return _("Both"); + + case ROLE_NONE: + return _("None"); + + case ROLE_PARTITION: + return _("Partition"); + + case ROLE_SPLIT: + return _("Split"); + + default: + return "Invalid"; + } +} + +/* Returns a string version of ROLE, for use in PSPP comamnd syntax. */ +const char * +var_role_to_syntax (enum var_role role) +{ + switch (role) + { + case ROLE_INPUT: + return "INPUT"; + + case ROLE_TARGET: + return "TARGET"; + + case ROLE_BOTH: + return "BOTH"; + + case ROLE_NONE: + return "NONE"; + + case ROLE_PARTITION: + return "PARTITION"; + + case ROLE_SPLIT: + return "SPLIT"; + + default: + return ""; + } +} + +/* Returns V's role. */ +enum var_role +var_get_role (const struct variable *v) +{ + return v->role; +} + +/* Sets V's role to ROLE. */ +static void +var_set_role_quiet (struct variable *v, enum var_role role) +{ + assert (var_role_is_valid (role)); + v->role = role; +} + + +/* Sets V's role to ROLE. */ +void +var_set_role (struct variable *v, enum var_role role) +{ + struct variable *ov = var_clone (v); + var_set_role_quiet (v, role); + dict_var_changed (v, VAR_TRAIT_ROLE, ov); } /* Returns V's display width, which applies only to GUIs. */ @@ -722,11 +955,24 @@ var_get_display_width (const struct variable *v) } /* Sets V's display width to DISPLAY_WIDTH. */ +static void +var_set_display_width_quiet (struct variable *v, int new_width) +{ + if (v->display_width != new_width) + { + v->display_width = new_width; + } +} + void -var_set_display_width (struct variable *v, int display_width) +var_set_display_width (struct variable *v, int new_width) { - v->display_width = display_width; - dict_var_changed (v); + if (v->display_width != new_width) + { + struct variable *ov = var_clone (v); + var_set_display_width_quiet (v, new_width); + dict_var_changed (v, VAR_TRAIT_DISPLAY_WIDTH, ov); + } } /* Returns the default display width for a variable of the given @@ -746,6 +992,46 @@ alignment_is_valid (enum alignment a) return a == ALIGN_LEFT || a == ALIGN_RIGHT || a == ALIGN_CENTRE; } +/* Returns a string version of alignment A, for display to a user. */ +const char * +alignment_to_string (enum alignment a) +{ + switch (a) + { + case ALIGN_LEFT: + return _("Left"); + + case ALIGN_RIGHT: + return _("Right"); + + case ALIGN_CENTRE: + return _("Center"); + + default: + return "Invalid"; + } +} + +/* Returns a string version of alignment A, for use in PSPP command syntax. */ +const char * +alignment_to_syntax (enum alignment a) +{ + switch (a) + { + case ALIGN_LEFT: + return "LEFT"; + + case ALIGN_RIGHT: + return "RIGHT"; + + case ALIGN_CENTRE: + return "CENTER"; + + default: + return "Invalid"; + } +} + /* Returns V's display alignment, which applies only to GUIs. */ enum alignment var_get_alignment (const struct variable *v) @@ -754,19 +1040,37 @@ var_get_alignment (const struct variable *v) } /* Sets V's display alignment to ALIGNMENT. */ -void -var_set_alignment (struct variable *v, enum alignment alignment) +static void +var_set_alignment_quiet (struct variable *v, enum alignment alignment) { assert (alignment_is_valid (alignment)); v->alignment = alignment; - dict_var_changed (v); +} + +/* Sets V's display alignment to ALIGNMENT. */ +void +var_set_alignment (struct variable *v, enum alignment alignment) +{ + struct variable *ov = var_clone (v); + var_set_alignment_quiet (v, alignment); + dict_var_changed (v, VAR_TRAIT_ALIGNMENT, ov); +} + + +/* Returns the default display alignment for a variable of the + given TYPE, as set by var_create. The return value can be + used to reset a variable's display alignment to the default. */ +enum alignment +var_default_alignment (enum val_type type) +{ + return type == VAL_NUMERIC ? ALIGN_RIGHT : ALIGN_LEFT; } /* Whether variables' values should be preserved from case to case. */ /* Returns true if variable V's value should be left from case to - case, instead of being reset to 0, system-missing, or blanks. */ + case, instead of being reset to system-missing or blanks. */ bool var_get_leave (const struct variable *v) { @@ -774,20 +1078,30 @@ var_get_leave (const struct variable *v) } /* Sets V's leave setting to LEAVE. */ -void -var_set_leave (struct variable *v, bool leave) +static void +var_set_leave_quiet (struct variable *v, bool leave) { assert (leave || !var_must_leave (v)); v->leave = leave; - dict_var_changed (v); } + +/* Sets V's leave setting to LEAVE. */ +void +var_set_leave (struct variable *v, bool leave) +{ + struct variable *ov = var_clone (v); + var_set_leave_quiet (v, leave); + dict_var_changed (v, VAR_TRAIT_LEAVE, ov); +} + + /* Returns true if V must be left from case to case, false if it can be set either way. */ bool var_must_leave (const struct variable *v) { - return dict_class_from_id (v->name) == DC_SCRATCH; + return var_get_dict_class (v) == DC_SCRATCH; } /* Returns the number of short names stored in VAR. @@ -816,15 +1130,18 @@ var_get_short_name (const struct variable *var, size_t idx) return idx < var->short_name_cnt ? var->short_names[idx] : NULL; } -/* Sets VAR's short name with the given IDX to SHORT_NAME, - truncating it to SHORT_NAME_LEN characters and converting it - to uppercase in the process. Specifying a null pointer for - SHORT_NAME clears the specified short name. */ +/* Sets VAR's short name with the given IDX to the UTF-8 string SHORT_NAME. + The caller must already have checked that, in the dictionary encoding, + SHORT_NAME is no more than SHORT_NAME_LEN bytes long. The new short name + will be converted to uppercase. + + Specifying a null pointer for SHORT_NAME clears the specified short name. */ void var_set_short_name (struct variable *var, size_t idx, const char *short_name) { - assert (var != NULL); - assert (short_name == NULL || var_is_plausible_name (short_name, false)); + struct variable *ov = var_clone (var); + + assert (short_name == NULL || id_is_plausible (short_name, false)); /* Clear old short name numbered IDX, if any. */ if (idx < var->short_name_cnt) @@ -840,18 +1157,16 @@ var_set_short_name (struct variable *var, size_t idx, const char *short_name) { size_t old_cnt = var->short_name_cnt; size_t i; - var->short_name_cnt = MAX (idx * 2, 1); var->short_names = xnrealloc (var->short_names, var->short_name_cnt, sizeof *var->short_names); for (i = old_cnt; i < var->short_name_cnt; i++) var->short_names[i] = NULL; } - var->short_names[idx] = xstrndup (short_name, MAX_SHORT_STRING); - str_uppercase (var->short_names[idx]); + var->short_names[idx] = utf8_to_upper (short_name); } - dict_var_changed (var); + dict_var_changed (var, VAR_TRAIT_NAME, ov); } /* Clears V's short names. */ @@ -859,9 +1174,10 @@ void var_clear_short_names (struct variable *v) { size_t i; - + for (i = 0; i < v->short_name_cnt; i++) free (v->short_names[i]); + free (v->short_names); v->short_names = NULL; v->short_name_cnt = 0; } @@ -874,8 +1190,8 @@ var_clear_short_names (struct variable *v) size_t var_get_dict_index (const struct variable *v) { - assert (v->vardict.dict_index != -1); - return v->vardict.dict_index; + assert (var_has_vardict (v)); + return vardict_get_dict_index (v->vardict); } /* Returns V's index within the case represented by its @@ -885,156 +1201,114 @@ var_get_dict_index (const struct variable *v) size_t var_get_case_index (const struct variable *v) { - assert (v->vardict.case_index != -1); - return v->vardict.case_index; + assert (var_has_vardict (v)); + return vardict_get_case_index (v->vardict); } -/* Returns V's auxiliary data, or a null pointer if none has been - attached. */ -void * -var_get_aux (const struct variable *v) +/* Returns variable V's attribute set. The caller may examine or + modify the attribute set, but must not destroy it. Destroying + V, or calling var_set_attributes() on V, will also destroy its + attribute set. */ +struct attrset * +var_get_attributes (const struct variable *v) { - return v->aux; + return CONST_CAST (struct attrset *, &v->attributes); } -/* Assign auxiliary data AUX to variable V, which must not - already have auxiliary data. Before V's auxiliary data is - cleared, AUX_DTOR(V) will be called. (var_dtor_free, below, - may be appropriate for use as AUX_DTOR.) */ -void * -var_attach_aux (const struct variable *v_, - void *aux, void (*aux_dtor) (struct variable *)) -{ - struct variable *v = (struct variable *) v_ ; /* cast away const */ - assert (v->aux == NULL); - assert (aux != NULL); - v->aux = aux; - v->aux_dtor = aux_dtor; - return aux; -} - -/* Remove auxiliary data, if any, from V, and return it, without - calling any associated destructor. */ -void * -var_detach_aux (struct variable *v) +/* Replaces variable V's attributes set by a copy of ATTRS. */ +static void +var_set_attributes_quiet (struct variable *v, const struct attrset *attrs) { - void *aux = v->aux; - assert (aux != NULL); - v->aux = NULL; - return aux; + attrset_destroy (&v->attributes); + attrset_clone (&v->attributes, attrs); } -/* Clears auxiliary data, if any, from V, and calls any - associated destructor. */ +/* Replaces variable V's attributes set by a copy of ATTRS. */ void -var_clear_aux (struct variable *v) +var_set_attributes (struct variable *v, const struct attrset *attrs) { - assert (v != NULL); - if (v->aux != NULL) - { - if (v->aux_dtor != NULL) - v->aux_dtor (v); - v->aux = NULL; - } + struct variable *ov = var_clone (v); + var_set_attributes_quiet (v, attrs); + dict_var_changed (v, VAR_TRAIT_ATTRIBUTES, ov); } -/* This function is appropriate for use an auxiliary data - destructor (passed as AUX_DTOR to var_attach_aux()) for the - case where the auxiliary data should be passed to free(). */ -void -var_dtor_free (struct variable *v) + +/* Returns true if V has any custom attributes, false if it has none. */ +bool +var_has_attributes (const struct variable *v) { - free (v->aux); + return attrset_count (&v->attributes) > 0; } -/* Observed categorical values. */ -/* Returns V's observed categorical values, - which V must have. */ -struct cat_vals * -var_get_obs_vals (const struct variable *v) -{ - assert (v->obs_vals != NULL); - return v->obs_vals; -} +/* Creates and returns a clone of OLD_VAR. Most properties of + the new variable are copied from OLD_VAR, except: -/* Sets V's observed categorical values to CAT_VALS. */ -void -var_set_obs_vals (const struct variable *v_, struct cat_vals *cat_vals) -{ - struct variable *v = (struct variable *) v_ ; /* cast away const */ - cat_stored_values_destroy (v->obs_vals); - v->obs_vals = cat_vals; -} + - The variable's short name is not copied, because there is + no reason to give a new variable with potentially a new + name the same short name. -/* Returns true if V has observed categorical values, - false otherwise. */ -bool -var_has_obs_vals (const struct variable *v) -{ - return v->obs_vals != NULL; -} - -/* Returns the dictionary class corresponding to a variable named - NAME. */ -enum dict_class -dict_class_from_id (const char *name) + - The new variable is not added to OLD_VAR's dictionary by + default. Use dict_clone_var, instead, to do that. +*/ +struct variable * +var_clone (const struct variable *old_var) { - switch (name[0]) - { - default: - return DC_ORDINARY; - case '$': - return DC_SYSTEM; - case '#': - return DC_SCRATCH; - } + struct variable *new_var = var_create (var_get_name (old_var), + var_get_width (old_var)); + + var_set_missing_values_quiet (new_var, var_get_missing_values (old_var)); + var_set_print_format_quiet (new_var, var_get_print_format (old_var)); + var_set_write_format_quiet (new_var, var_get_write_format (old_var)); + var_set_value_labels_quiet (new_var, var_get_value_labels (old_var)); + var_set_label_quiet (new_var, var_get_label (old_var)); + var_set_measure_quiet (new_var, var_get_measure (old_var)); + var_set_role_quiet (new_var, var_get_role (old_var)); + var_set_display_width_quiet (new_var, var_get_display_width (old_var)); + var_set_alignment_quiet (new_var, var_get_alignment (old_var)); + var_set_leave_quiet (new_var, var_get_leave (old_var)); + var_set_attributes_quiet (new_var, var_get_attributes (old_var)); + + return new_var; } -/* Returns the name of dictionary class DICT_CLASS. */ + + +/* Returns the encoding of values of variable VAR. (This is actually a + property of the dictionary.) Returns null if no specific encoding has been + set. */ const char * -dict_class_to_name (enum dict_class dict_class) +var_get_encoding (const struct variable *var) { - switch (dict_class) - { - case DC_ORDINARY: - return _("ordinary"); - case DC_SYSTEM: - return _("system"); - case DC_SCRATCH: - return _("scratch"); - default: - NOT_REACHED (); - } + return (var_has_vardict (var) + ? dict_get_encoding (vardict_get_dictionary (var->vardict)) + : NULL); } /* Returns V's vardict structure. */ -const struct vardict_info * +struct vardict_info * var_get_vardict (const struct variable *v) { - assert (var_has_vardict (v)); - return &v->vardict; + return CONST_CAST (struct vardict_info *, v->vardict); } /* Sets V's vardict data to VARDICT. */ void -var_set_vardict (struct variable *v, const struct vardict_info *vardict) +var_set_vardict (struct variable *v, struct vardict_info *vardict) { - assert (vardict->dict_index >= 0); - assert (vardict->case_index >= 0); - v->vardict = *vardict; + v->vardict = vardict; } /* Returns true if V has vardict data. */ bool var_has_vardict (const struct variable *v) { - return v->vardict.dict_index != -1; + return v->vardict != NULL; } /* Clears V's vardict data. */ void var_clear_vardict (struct variable *v) { - v->vardict.dict_index = v->vardict.case_index = -1; + v->vardict = NULL; }