/* PSPP - a program for statistical analysis.
- Copyright (C) 1997-9, 2000, 2007 Free Software Foundation, Inc.
+ Copyright (C) 1997-9, 2000, 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
struct freq_tab
{
struct hsh_table *data; /* Undifferentiated data. */
- struct freq *valid; /* Valid freqs. */
+ struct freq_mutable *valid; /* Valid freqs. */
int n_valid; /* Number of total freqs. */
+ const struct dictionary *dict; /* The dict from whence entries in the table
+ come */
- struct freq *missing; /* Missing freqs. */
+ struct freq_mutable *missing; /* Missing freqs. */
int n_missing; /* Number of missing freqs. */
/* Statistics. */
/* Statistics. */
double stat[frq_n_stats];
- /* Width and format for analysis and display.
- This is normally the same as "width" and "print" in struct
- variable, but in SPSS-compatible mode only the first
- MAX_SHORT_STRING bytes of long string variables are
- included. */
+ /* Variable attributes. */
int width;
struct fmt_spec print;
};
static void precalc (struct casereader *, struct dataset *);
static void calc (const struct ccase *, const struct dataset *);
-static void postcalc (void);
+static void postcalc (const struct dataset *);
static void postprocess_freq_tab (const struct variable *);
-static void dump_full (const struct variable *);
-static void dump_condensed (const struct variable *);
-static void dump_statistics (const struct variable *, int show_varname);
+static void dump_full ( const struct variable *, const struct variable *);
+static void dump_condensed (const struct variable *, const struct variable *);
+static void dump_statistics (const struct variable *, bool show_varname, const struct variable *);
static void cleanup_freq_tab (const struct variable *);
static hsh_compare_func compare_value_numeric_a, compare_value_alpha_a;
static void do_piechart(const struct variable *var,
const struct freq_tab *frq_tab);
-gsl_histogram *
+struct histogram *
freq_tab_to_hist(const struct freq_tab *ft, const struct variable *var);
for (; casegrouper_get_next_group (grouper, &group);
casereader_destroy (group))
{
- struct ccase c;
+ struct ccase *c;
precalc (group, ds);
- for (; casereader_read (group, &c); case_destroy (&c))
- calc (&c, ds);
- postcalc ();
+ for (; (c = casereader_read (group)) != NULL; case_unref (c))
+ calc (c, ds);
+ postcalc (ds);
}
ok = casegrouper_destroy (grouper);
ok = proc_commit (ds) && ok;
struct var_freqs *vf = get_var_freqs (v);
struct freq_tab *ft = &vf->tab;
- struct freq target;
- struct freq **fpp;
+ struct freq_mutable target;
+ struct freq_mutable **fpp;
- target.value = (union value *) val;
- fpp = (struct freq **) hsh_probe (ft->data, &target);
+ target.value = *val;
+ fpp = (struct freq_mutable **) hsh_probe (ft->data, &target);
if (*fpp != NULL)
(*fpp)->count += weight;
else
{
- struct freq *fp = pool_alloc (data_pool, sizeof *fp);
+ struct freq_mutable *fp = pool_alloc (data_pool, sizeof *fp);
fp->count = weight;
- fp->value = pool_clone (data_pool,
- val,
- MAX (MAX_SHORT_STRING, vf->width));
+ value_init_pool (data_pool, &fp->value, vf->width);
+ value_copy (&fp->value, val, vf->width);
*fpp = fp;
}
}
static void
precalc (struct casereader *input, struct dataset *ds)
{
- struct ccase c;
+ struct ccase *c;
size_t i;
- if (casereader_peek (input, 0, &c))
+ c = casereader_peek (input, 0);
+ if (c != NULL)
{
- output_split_file_values (ds, &c);
- case_destroy (&c);
+ output_split_file_values (ds, c);
+ case_unref (c);
}
pool_destroy (data_pool);
/* Finishes up with the variables after frequencies have been
calculated. Displays statistics, percentiles, ... */
static void
-postcalc (void)
+postcalc (const struct dataset *ds)
{
+ const struct dictionary *dict = dataset_dict (ds);
+ const struct variable *wv = dict_get_weight (dict);
size_t i;
for (i = 0; i < n_variables; i++)
switch (cmd.cond)
{
case FRQ_CONDENSE:
- dump_condensed (v);
+ dump_condensed (v, wv);
break;
case FRQ_STANDARD:
- dump_full (v);
+ dump_full (v, wv);
break;
case FRQ_ONEPAGE:
if (n_categories > cmd.onepage_limit)
- dump_condensed (v);
+ dump_condensed (v, wv);
else
- dump_full (v);
+ dump_full (v, wv);
break;
default:
NOT_REACHED ();
/* Statistics. */
if (n_stats)
- dump_statistics (v, !dumped_freq_tab);
+ dump_statistics (v, !dumped_freq_tab, wv);
- if ( chart == GFT_HIST)
+ if ( chart == GFT_HIST && var_is_numeric (v) )
{
double d[frq_n_stats];
- struct normal_curve norm;
- gsl_histogram *hist ;
-
-
- norm.N = vf->tab.valid_cases;
+ struct histogram *hist ;
calc_stats (v, d);
- norm.mean = d[frq_mean];
- norm.stddev = d[frq_stddev];
- hist = freq_tab_to_hist(ft,v);
+ hist = freq_tab_to_hist (ft,v);
- histogram_plot(hist, var_to_string(v), &norm, normal);
+ histogram_plot_n (hist, var_to_string(v),
+ vf->tab.valid_cases,
+ d[frq_mean],
+ d[frq_stddev],
+ normal);
- gsl_histogram_free(hist);
+ statistic_destroy ((struct statistic *)hist);
}
-
if ( chart == GFT_PIE)
{
do_piechart(v_variables[i], ft);
}
-
-
cleanup_freq_tab (v);
}
}
}
-/* Returns true iff the value in struct freq F is non-missing
+/* Returns true iff the value in struct freq_mutable F is non-missing
for variable V. */
static bool
not_missing (const void *f_, const void *v_)
{
- const struct freq *f = f_;
+ const struct freq_mutable *f = f_;
const struct variable *v = v_;
- return !var_is_value_missing (v, f->value, MV_ANY);
+ return !var_is_value_missing (v, &f->value, MV_ANY);
}
/* Summarizes the frequency table data for variable V. */
struct freq_tab *ft;
size_t count;
void *const *data;
- struct freq *freqs, *f;
+ struct freq_mutable *freqs, *f;
size_t i;
ft = &get_var_freqs (v)->tab;
freqs = xnmalloc (count, sizeof *freqs);
for (i = 0; i < count; i++)
{
- struct freq *f = data[i];
+ struct freq_mutable *f = data[i];
freqs[i] = *f;
}
}
vf = var_attach_aux (v, xmalloc (sizeof *vf), var_dtor_free);
vf->tab.valid = vf->tab.missing = NULL;
+ vf->tab.dict = dataset_dict (ds);
vf->n_groups = 0;
vf->groups = NULL;
vf->width = var_get_width (v);
vf->print = *var_get_print_format (v);
- if (vf->width > MAX_SHORT_STRING && get_algorithm () == COMPATIBLE)
- {
- enum fmt_type type = var_get_print_format (v)->type;
- vf->width = MAX_SHORT_STRING;
- vf->print.w = MAX_SHORT_STRING * (type == FMT_AHEX ? 2 : 1);
- }
}
return 1;
}
static int
compare_value_numeric_a (const void *a_, const void *b_, const void *aux UNUSED)
{
- const struct freq *a = a_;
- const struct freq *b = b_;
+ const struct freq_mutable *a = a_;
+ const struct freq_mutable *b = b_;
- if (a->value[0].f > b->value[0].f)
+ if (a->value.f > b->value.f)
return 1;
- else if (a->value[0].f < b->value[0].f)
+ else if (a->value.f < b->value.f)
return -1;
else
return 0;
static int
compare_value_alpha_a (const void *a_, const void *b_, const void *v_)
{
- const struct freq *a = a_;
- const struct freq *b = b_;
+ const struct freq_mutable *a = a_;
+ const struct freq_mutable *b = b_;
const struct variable *v = v_;
struct var_freqs *vf = get_var_freqs (v);
- return memcmp (a->value[0].s, b->value[0].s, vf->width);
+ return value_compare_3way (&a->value, &b->value, vf->width);
}
/* Descending numeric compare of values. */
static int
compare_freq_numeric_a (const void *a_, const void *b_, const void *aux UNUSED)
{
- const struct freq *a = a_;
- const struct freq *b = b_;
+ const struct freq_mutable *a = a_;
+ const struct freq_mutable *b = b_;
if (a->count > b->count)
return 1;
else if (a->count < b->count)
return -1;
- if (a->value[0].f > b->value[0].f)
+ if (a->value.f > b->value.f)
return 1;
- else if (a->value[0].f < b->value[0].f)
+ else if (a->value.f < b->value.f)
return -1;
else
return 0;
static int
compare_freq_alpha_a (const void *a_, const void *b_, const void *v_)
{
- const struct freq *a = a_;
- const struct freq *b = b_;
+ const struct freq_mutable *a = a_;
+ const struct freq_mutable *b = b_;
const struct variable *v = v_;
struct var_freqs *vf = get_var_freqs (v);
else if (a->count < b->count)
return -1;
else
- return memcmp (a->value[0].s, b->value[0].s, vf->width);
+ return value_compare_3way (&a->value, &b->value, vf->width);
}
/* Descending numeric compare of frequency;
static int
compare_freq_numeric_d (const void *a_, const void *b_, const void *aux UNUSED)
{
- const struct freq *a = a_;
- const struct freq *b = b_;
+ const struct freq_mutable *a = a_;
+ const struct freq_mutable *b = b_;
if (a->count > b->count)
return -1;
else if (a->count < b->count)
return 1;
- if (a->value[0].f > b->value[0].f)
+ if (a->value.f > b->value.f)
return 1;
- else if (a->value[0].f < b->value[0].f)
+ else if (a->value.f < b->value.f)
return -1;
else
return 0;
static int
compare_freq_alpha_d (const void *a_, const void *b_, const void *v_)
{
- const struct freq *a = a_;
- const struct freq *b = b_;
+ const struct freq_mutable *a = a_;
+ const struct freq_mutable *b = b_;
const struct variable *v = v_;
struct var_freqs *vf = get_var_freqs (v);
else if (a->count < b->count)
return 1;
else
- return memcmp (a->value[0].s, b->value[0].s, vf->width);
+ return value_compare_3way (&a->value, &b->value, vf->width);
}
\f
/* Frequency table display. */
/* Sets the widths of all the columns and heights of all the rows in
table T for driver D. */
static void
-full_dim (struct tab_table *t, struct outp_driver *d)
+full_dim (struct tab_table *t, struct outp_driver *d, void *aux UNUSED)
{
int i = 0;
int columns = 5;
/* Displays a full frequency table for variable V. */
static void
-dump_full (const struct variable *v)
+dump_full (const struct variable *v, const struct variable *wv)
{
+ const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : &F_8_0;
int n_categories;
struct var_freqs *vf;
struct freq_tab *ft;
- struct freq *f;
+ struct freq_mutable *f;
struct tab_table *t;
int r;
double cum_total = 0.0;
n_categories = ft->n_valid + ft->n_missing;
t = tab_create (5 + lab, n_categories + 3, 0);
tab_headers (t, 0, 0, 2, 0);
- tab_dim (t, full_dim);
+ tab_dim (t, full_dim, NULL);
if (lab)
tab_text (t, 0, 1, TAB_CENTER | TAT_TITLE, _("Value Label"));
if (lab)
{
- const char *label = var_lookup_value_label (v, &f->value[0]);
+ const char *label = var_lookup_value_label (v, &f->value);
if (label != NULL)
tab_text (t, 0, r, TAB_LEFT, label);
}
- tab_value (t, 0 + lab, r, TAB_NONE, f->value, &vf->print);
- tab_float (t, 1 + lab, r, TAB_NONE, f->count, 8, 0);
- tab_float (t, 2 + lab, r, TAB_NONE, percent, 5, 1);
- tab_float (t, 3 + lab, r, TAB_NONE, valid_percent, 5, 1);
- tab_float (t, 4 + lab, r, TAB_NONE, cum_total, 5, 1);
+ tab_value (t, 0 + lab, r, TAB_NONE, &f->value, ft->dict, &vf->print);
+ tab_double (t, 1 + lab, r, TAB_NONE, f->count, wfmt);
+ tab_double (t, 2 + lab, r, TAB_NONE, percent, NULL);
+ tab_double (t, 3 + lab, r, TAB_NONE, valid_percent, NULL);
+ tab_double (t, 4 + lab, r, TAB_NONE, cum_total, NULL);
r++;
}
for (; f < &ft->valid[n_categories]; f++)
if (lab)
{
- const char *label = var_lookup_value_label (v, &f->value[0]);
+ const char *label = var_lookup_value_label (v, &f->value);
if (label != NULL)
tab_text (t, 0, r, TAB_LEFT, label);
}
- tab_value (t, 0 + lab, r, TAB_NONE, f->value, &vf->print);
- tab_float (t, 1 + lab, r, TAB_NONE, f->count, 8, 0);
- tab_float (t, 2 + lab, r, TAB_NONE,
- f->count / ft->total_cases * 100.0, 5, 1);
+ tab_value (t, 0 + lab, r, TAB_NONE, &f->value, ft->dict, &vf->print);
+ tab_double (t, 1 + lab, r, TAB_NONE, f->count, wfmt);
+ tab_double (t, 2 + lab, r, TAB_NONE,
+ f->count / ft->total_cases * 100.0, NULL);
tab_text (t, 3 + lab, r, TAB_NONE, _("Missing"));
r++;
}
tab_hline (t, TAL_2, 0, 4 + lab, r);
tab_joint_text (t, 0, r, 0 + lab, r, TAB_RIGHT | TAT_TITLE, _("Total"));
tab_vline (t, TAL_0, 1, r, r);
- tab_float (t, 1 + lab, r, TAB_NONE, cum_freq, 8, 0);
- tab_float (t, 2 + lab, r, TAB_NONE, 100.0, 5, 1);
- tab_float (t, 3 + lab, r, TAB_NONE, 100.0, 5, 1);
+ tab_double (t, 1 + lab, r, TAB_NONE, cum_freq, wfmt);
+ tab_fixed (t, 2 + lab, r, TAB_NONE, 100.0, 5, 1);
+ tab_fixed (t, 3 + lab, r, TAB_NONE, 100.0, 5, 1);
tab_title (t, "%s", var_to_string (v));
tab_submit (t);
/* Sets the widths of all the columns and heights of all the rows in
table T for driver D. */
static void
-condensed_dim (struct tab_table *t, struct outp_driver *d)
+condensed_dim (struct tab_table *t, struct outp_driver *d, void *aux UNUSED)
{
int cum_w = MAX (outp_string_width (d, _("Cum"), OUTP_PROPORTIONAL),
MAX (outp_string_width (d, _("Cum"), OUTP_PROPORTIONAL),
/* Display condensed frequency table for variable V. */
static void
-dump_condensed (const struct variable *v)
+dump_condensed (const struct variable *v, const struct variable *wv)
{
+ const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : &F_8_0;
int n_categories;
struct var_freqs *vf;
struct freq_tab *ft;
- struct freq *f;
+ struct freq_mutable *f;
struct tab_table *t;
int r;
double cum_total = 0.0;
tab_text (t, 2, 1, TAB_CENTER | TAT_TITLE, _("Pct"));
tab_text (t, 3, 0, TAB_CENTER | TAT_TITLE, _("Cum"));
tab_text (t, 3, 1, TAB_CENTER | TAT_TITLE, _("Pct"));
- tab_dim (t, condensed_dim);
+ tab_dim (t, condensed_dim, NULL);
r = 2;
for (f = ft->valid; f < ft->missing; f++)
percent = f->count / ft->total_cases * 100.0;
cum_total += f->count / ft->valid_cases * 100.0;
- tab_value (t, 0, r, TAB_NONE, f->value, &vf->print);
- tab_float (t, 1, r, TAB_NONE, f->count, 8, 0);
- tab_float (t, 2, r, TAB_NONE, percent, 3, 0);
- tab_float (t, 3, r, TAB_NONE, cum_total, 3, 0);
+ tab_value (t, 0, r, TAB_NONE, &f->value, ft->dict, &vf->print);
+ tab_double (t, 1, r, TAB_NONE, f->count, wfmt);
+ tab_double (t, 2, r, TAB_NONE, percent, NULL);
+ tab_double (t, 3, r, TAB_NONE, cum_total, NULL);
r++;
}
for (; f < &ft->valid[n_categories]; f++)
{
- tab_value (t, 0, r, TAB_NONE, f->value, &vf->print);
- tab_float (t, 1, r, TAB_NONE, f->count, 8, 0);
- tab_float (t, 2, r, TAB_NONE,
- f->count / ft->total_cases * 100.0, 3, 0);
+ tab_value (t, 0, r, TAB_NONE, &f->value, ft->dict, &vf->print);
+ tab_double (t, 1, r, TAB_NONE, f->count, wfmt);
+ tab_double (t, 2, r, TAB_NONE,
+ f->count / ft->total_cases * 100.0, NULL);
r++;
}
struct freq_tab *ft = &get_var_freqs (v)->tab;
double W = ft->valid_cases;
struct moments *m;
- struct freq *f=0;
+ struct freq_mutable *f=0;
int most_often;
double X_mode;
double tp;
if ( percentiles[i].flag2 ) continue ;
- if ( get_algorithm() != COMPATIBLE )
+ if ( settings_get_algorithm () != COMPATIBLE )
tp =
(ft->valid_cases - 1) * percentiles[i].p;
else
if ( percentiles[i].flag )
{
- percentiles[i].x2 = f->value[0].f;
+ percentiles[i].x2 = f->value.f;
percentiles[i].x1 = prev_value;
percentiles[i].flag2 = 1;
continue;
{
if ( f->count > 1 && rank - (f->count - 1) > tp )
{
- percentiles[i].x2 = percentiles[i].x1 = f->value[0].f;
+ percentiles[i].x2 = percentiles[i].x1 = f->value.f;
percentiles[i].flag2 = 1;
}
else
continue;
}
}
- prev_value = f->value[0].f;
+ prev_value = f->value.f;
}
for (i = 0; i < n_percentiles; i++)
{
/* Catches the case when p == 100% */
if ( ! percentiles[i].flag2 )
- percentiles[i].x1 = percentiles[i].x2 = f->value[0].f;
+ percentiles[i].x1 = percentiles[i].x2 = f->value.f;
/*
printf("percentile %d (p==%.2f); X1 = %g; X2 = %g\n",
double s;
double dummy;
- if ( get_algorithm() != COMPATIBLE )
+ if ( settings_get_algorithm () != COMPATIBLE )
{
s = modf((ft->valid_cases - 1) * percentiles[i].p , &dummy);
}
if (most_often < f->count)
{
most_often = f->count;
- X_mode = f->value[0].f;
+ X_mode = f->value.f;
}
else if (most_often == f->count)
{
/* Calculate moments. */
m = moments_create (MOMENT_KURTOSIS);
for (f = ft->valid; f < ft->missing; f++)
- moments_pass_one (m, f->value[0].f, f->count);
+ moments_pass_one (m, f->value.f, f->count);
for (f = ft->valid; f < ft->missing; f++)
- moments_pass_two (m, f->value[0].f, f->count);
+ moments_pass_two (m, f->value.f, f->count);
moments_calculate (m, NULL, &d[frq_mean], &d[frq_variance],
&d[frq_skew], &d[frq_kurt]);
moments_destroy (m);
/* Formulas below are taken from _SPSS Statistical Algorithms_. */
- d[frq_min] = ft->valid[0].value[0].f;
- d[frq_max] = ft->valid[ft->n_valid - 1].value[0].f;
+ d[frq_min] = ft->valid[0].value.f;
+ d[frq_max] = ft->valid[ft->n_valid - 1].value.f;
d[frq_mode] = X_mode;
d[frq_range] = d[frq_max] - d[frq_min];
d[frq_sum] = d[frq_mean] * W;
/* Displays a table of all the statistics requested for variable V. */
static void
-dump_statistics (const struct variable *v, int show_varname)
+dump_statistics (const struct variable *v, bool show_varname,
+ const struct variable *wv)
{
+ const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : &F_8_0;
struct freq_tab *ft;
double stat_value[frq_n_stats];
struct tab_table *t;
calc_stats (v, stat_value);
t = tab_create (3, n_stats + n_percentiles + 2, 0);
- tab_dim (t, tab_natural_dimensions);
+ tab_dim (t, tab_natural_dimensions, NULL);
tab_box (t, TAL_1, TAL_1, -1, -1 , 0 , 0 , 2, tab_nr(t) - 1) ;
{
tab_text (t, 0, r, TAB_LEFT | TAT_TITLE,
gettext (st_name[i].s10));
- tab_float (t, 2, r, TAB_NONE, stat_value[i], 11, 3);
+ tab_double (t, 2, r, TAB_NONE, stat_value[i], NULL);
r++;
}
tab_text (t, 1, 0, TAB_LEFT | TAT_TITLE, _("Valid"));
tab_text (t, 1, 1, TAB_LEFT | TAT_TITLE, _("Missing"));
- tab_float(t, 2, 0, TAB_NONE, ft->valid_cases, 11, 0);
- tab_float(t, 2, 1, TAB_NONE, ft->total_cases - ft->valid_cases, 11, 0);
-
+ tab_double (t, 2, 0, TAB_NONE, ft->valid_cases, wfmt);
+ tab_double (t, 2, 1, TAB_NONE, ft->total_cases - ft->valid_cases, wfmt);
for (i = 0; i < n_percentiles; i++, r++)
{
if (percentiles[i].p == 0.5)
tab_text (t, 1, r, TAB_LEFT, _("50 (Median)"));
else
- tab_float (t, 1, r, TAB_LEFT, percentiles[i].p * 100, 3, 0);
- tab_float (t, 2, r, TAB_NONE, percentiles[i].value, 11, 3);
-
+ tab_fixed (t, 1, r, TAB_LEFT, percentiles[i].p * 100, 3, 0);
+ tab_double (t, 2, r, TAB_NONE, percentiles[i].value,
+ var_get_print_format (v));
}
tab_columns (t, SOM_COL_DOWN, 1);
/* Create a gsl_histogram from a freq_tab */
-gsl_histogram *
-freq_tab_to_hist(const struct freq_tab *ft, const struct variable *var)
+struct histogram *
+freq_tab_to_hist (const struct freq_tab *ft, const struct variable *var)
{
int i;
double x_min = DBL_MAX;
double x_max = -DBL_MAX;
- gsl_histogram *hist;
+ struct statistic *hist;
const double bins = 11;
struct hsh_iterator hi;
struct hsh_table *fh = ft->data;
- struct freq *frq;
+ struct freq_mutable *frq;
/* Find out the extremes of the x value */
for ( frq = hsh_first(fh, &hi); frq != 0; frq = hsh_next(fh, &hi) )
{
- if (var_is_value_missing(var, frq->value, MV_ANY))
+ if (var_is_value_missing(var, &frq->value, MV_ANY))
continue;
- if ( frq->value[0].f < x_min ) x_min = frq->value[0].f ;
- if ( frq->value[0].f > x_max ) x_max = frq->value[0].f ;
+ if ( frq->value.f < x_min ) x_min = frq->value.f ;
+ if ( frq->value.f > x_max ) x_max = frq->value.f ;
}
- hist = histogram_create(bins, x_min, x_max);
+ hist = histogram_create (bins, x_min, x_max);
for( i = 0 ; i < ft->n_valid ; ++i )
{
frq = &ft->valid[i];
- gsl_histogram_accumulate(hist, frq->value[0].f, frq->count);
+ histogram_add ((struct histogram *)hist, frq->value.f, frq->count);
}
- return hist;
+ return (struct histogram *)hist;
}
for (i = 0 ; i < *n_slices ; ++i )
{
- const struct freq *frq = &frq_tab->valid[i];
+ const struct freq_mutable *frq = &frq_tab->valid[i];
- var_append_value_name (var, frq->value, &slices[i].label);
+ ds_init_empty (&slices[i].label);
+ var_append_value_name (var, &frq->value, &slices[i].label);
slices[i].magnetude = frq->count;
}
projects / pspp-builds.git / blobdiff