*/
#include <config.h>
-#include <libpspp/message.h>
+
#include <math.h>
#include <stdlib.h>
#include <gsl/gsl_histogram.h>
-#include <libpspp/alloc.h>
-#include <libpspp/bit-vector.h>
#include <data/case.h>
-#include <libpspp/compiler.h>
#include <data/dictionary.h>
-#include <libpspp/hash.h>
-#include <libpspp/pool.h>
+#include <data/procedure.h>
+#include <data/settings.h>
+#include <data/value-labels.h>
+#include <data/variable.h>
#include <language/command.h>
+#include <language/dictionary/split-file.h>
#include <language/lexer/lexer.h>
-#include <math/moments.h>
-#include <libpspp/message.h>
+#include <libpspp/alloc.h>
#include <libpspp/array.h>
+#include <libpspp/bit-vector.h>
+#include <libpspp/compiler.h>
+#include <libpspp/hash.h>
#include <libpspp/magic.h>
+#include <libpspp/message.h>
+#include <libpspp/message.h>
#include <libpspp/misc.h>
-#include <output/output.h>
-#include <output/manager.h>
+#include <libpspp/pool.h>
#include <libpspp/str.h>
-#include <output/table.h>
-#include <data/value-labels.h>
-#include <data/variable.h>
-#include <procedure.h>
-#include <data/settings.h>
-#include <output/charts/piechart.h>
+#include <math/histogram.h>
+#include <math/moments.h>
#include <output/chart.h>
+#include <output/charts/piechart.h>
#include <output/charts/plot-hist.h>
+#include <output/manager.h>
+#include <output/output.h>
+#include <output/table.h>
+
+#include "minmax.h"
#include "gettext.h"
#define _(msgid) gettext (msgid)
/* (headers) */
-#include <libpspp/debug-print.h>
-
/* (specification)
FREQUENCIES (frq_):
- *variables=custom;
- format=cond:condense/onepage(*n:onepage_limit,"%s>=0")/!standard,
- table:limit(n:limit,"%s>0")/notable/!table,
- labels:!labels/nolabels,
- sort:!avalue/dvalue/afreq/dfreq,
- spaces:!single/double,
- paging:newpage/!oldpage;
+ *+variables=custom;
+ +format=cond:condense/onepage(*n:onepage_limit,"%s>=0")/!standard,
+ table:limit(n:limit,"%s>0")/notable/!table,
+ labels:!labels/nolabels,
+ sort:!avalue/dvalue/afreq/dfreq,
+ spaces:!single/double,
+ paging:newpage/!oldpage;
missing=miss:include/!exclude;
barchart(ba_)=:minimum(d:min),
:maximum(d:max),
scale:freq(*n:freq,"%s>0")/percent(*n:pcnt,"%s>0"),
norm:!nonormal/normal,
incr:increment(d:inc,"%s>0");
- grouped=custom;
- ntiles=integer;
+ +grouped=custom;
+ +ntiles=integer;
+percentiles = double list;
- statistics[st_]=1|mean,2|semean,3|median,4|mode,5|stddev,6|variance,
- 7|kurtosis,8|skewness,9|range,10|minimum,11|maximum,12|sum,
- 13|default,14|seskewness,15|sekurtosis,all,none.
+ +statistics[st_]=1|mean,2|semean,3|median,4|mode,5|stddev,6|variance,
+ 7|kurtosis,8|skewness,9|range,10|minimum,11|maximum,12|sum,
+ 13|default,14|seskewness,15|sekurtosis,all,none.
*/
/* (declarations) */
/* (functions) */
/* Frequency table entry. */
struct freq
{
- union value v; /* The value. */
+ union value *v; /* The value. */
double c; /* The number of occurrences of the value. */
};
/* 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. */
+ int width;
+ struct fmt_spec print;
};
static inline struct var_freqs *
static void calc_stats (struct variable *v, double d[frq_n_stats]);
-static void precalc (void *);
-static bool calc (struct ccase *, void *);
+static void precalc (const struct ccase *, void *);
+static bool calc (const struct ccase *, void *);
static void postcalc (void *);
static void postprocess_freq_tab (struct variable *);
n_variables = 0;
v_variables = NULL;
- if (!parse_frequencies (&cmd))
+ if (!parse_frequencies (&cmd, NULL))
return CMD_FAILURE;
if (cmd.onepage_limit == NOT_LONG)
/* Add data from case C to the frequency table. */
static bool
-calc (struct ccase *c, void *aux UNUSED)
+calc (const struct ccase *c, void *aux UNUSED)
{
double weight;
size_t i;
- int bad_warn = 1;
+ bool bad_warn = true;
weight = dict_get_case_weight (default_dict, c, &bad_warn);
{
struct variable *v = v_variables[i];
const union value *val = case_data (c, v->fv);
- struct freq_tab *ft = &get_var_freqs (v)->tab;
+ struct var_freqs *vf = get_var_freqs (v);
+ struct freq_tab *ft = &vf->tab;
switch (ft->mode)
{
case FRQM_GENERAL:
{
-
/* General mode. */
- struct freq **fpp = (struct freq **) hsh_probe (ft->data, val);
+ struct freq target;
+ struct freq **fpp;
+
+ target.v = (union value *) val;
+ fpp = (struct freq **) hsh_probe (ft->data, &target);
if (*fpp != NULL)
(*fpp)->c += weight;
else
{
- struct freq *fp = *fpp = pool_alloc (gen_pool, sizeof *fp);
- fp->v = *val;
- fp->c = weight;
+ struct freq *fp = pool_alloc (gen_pool, sizeof *fp);
+ fp->c = weight;
+ fp->v = pool_clone (gen_pool,
+ val, MAX (MAX_SHORT_STRING, vf->width));
+ *fpp = fp;
}
}
break;
ft->out_of_range += weight;
break;
default:
- assert (0);
+ NOT_REACHED ();
}
}
return true;
/* Prepares each variable that is the target of FREQUENCIES by setting
up its hash table. */
static void
-precalc (void *aux UNUSED)
+precalc (const struct ccase *first, void *aux UNUSED)
{
size_t i;
+ output_split_file_values (first);
+
pool_destroy (gen_pool);
gen_pool = pool_create ();
dump_full (v);
break;
default:
- assert (0);
+ NOT_REACHED ();
}
else
dumped_freq_tab = 0;
case FRQ_AFREQ | (ALPHA << 16): return compare_freq_alpha_a;
case FRQ_DFREQ | (NUMERIC << 16): return compare_freq_numeric_d;
case FRQ_DFREQ | (ALPHA << 16): return compare_freq_alpha_d;
- default: assert (0);
+ default: NOT_REACHED ();
}
return 0;
}
-/* Returns nonzero iff the value in struct freq F is non-missing
+/* Returns true iff the value in struct freq F is non-missing
for variable V. */
-static int
+static bool
not_missing (const void *f_, void *v_)
{
const struct freq *f = f_;
struct variable *v = v_;
- return !mv_is_value_missing (&v->miss, &f->v);
+ return !mv_is_value_missing (&v->miss, f->v);
}
/* Summarizes the frequency table data for variable V. */
/* Parses the VARIABLES subcommand, adding to
{n_variables,v_variables}. */
static int
-frq_custom_variables (struct cmd_frequencies *cmd UNUSED)
+frq_custom_variables (struct cmd_frequencies *cmd UNUSED, void *aux UNUSED)
{
int mode;
int min = 0, max = 0;
vf->tab.vector = pool_nalloc (int_pool,
max - min + 1, sizeof *vf->tab.vector);
}
- else
- vf->tab.vector = NULL;
+ else
+ vf->tab.vector = NULL;
vf->n_groups = 0;
vf->groups = NULL;
+ vf->width = v->width;
+ vf->print = v->print;
+ if (vf->width > MAX_SHORT_STRING && get_algorithm () == COMPATIBLE)
+ {
+ vf->width = MAX_SHORT_STRING;
+ vf->print.w = MAX_SHORT_STRING * (v->print.type == FMT_AHEX ? 2 : 1);
+ }
}
return 1;
}
/* Parses the GROUPED subcommand, setting the n_grouped, grouped
fields of specified variables. */
static int
-frq_custom_grouped (struct cmd_frequencies *cmd UNUSED)
+frq_custom_grouped (struct cmd_frequencies *cmd UNUSED, void *aux UNUSED)
{
lex_match ('=');
if ((token == T_ID && dict_lookup_var (default_dict, tokid) != NULL)
hash_value_numeric (const void *value_, void *foo UNUSED)
{
const struct freq *value = value_;
- return hsh_hash_double (value->v.f);
+ return hsh_hash_double (value->v[0].f);
}
/* Hash of string values. */
{
const struct freq *value = value_;
struct variable *v = v_;
+ struct var_freqs *vf = get_var_freqs (v);
- return hsh_hash_bytes (value->v.s, v->width);
+ return hsh_hash_bytes (value->v[0].s, vf->width);
}
/* Ascending numeric compare of values. */
const struct freq *a = a_;
const struct freq *b = b_;
- if (a->v.f > b->v.f)
+ if (a->v[0].f > b->v[0].f)
return 1;
- else if (a->v.f < b->v.f)
+ else if (a->v[0].f < b->v[0].f)
return -1;
else
return 0;
{
const struct freq *a = a_;
const struct freq *b = b_;
- const struct variable *v = v_;
+ struct variable *v = v_;
+ struct var_freqs *vf = get_var_freqs (v);
- return memcmp (a->v.s, b->v.s, v->width);
+ return memcmp (a->v[0].s, b->v[0].s, vf->width);
}
/* Descending numeric compare of values. */
else if (a->c < b->c)
return -1;
- if (a->v.f > b->v.f)
+ if (a->v[0].f > b->v[0].f)
return 1;
- else if (a->v.f < b->v.f)
+ else if (a->v[0].f < b->v[0].f)
return -1;
else
return 0;
{
const struct freq *a = a_;
const struct freq *b = b_;
- const struct variable *v = v_;
+ struct variable *v = v_;
+ struct var_freqs *vf = get_var_freqs (v);
if (a->c > b->c)
return 1;
else if (a->c < b->c)
return -1;
else
- return memcmp (a->v.s, b->v.s, v->width);
+ return memcmp (a->v[0].s, b->v[0].s, vf->width);
}
/* Descending numeric compare of frequency;
else if (a->c < b->c)
return 1;
- if (a->v.f > b->v.f)
+ if (a->v[0].f > b->v[0].f)
return 1;
- else if (a->v.f < b->v.f)
+ else if (a->v[0].f < b->v[0].f)
return -1;
else
return 0;
{
const struct freq *a = a_;
const struct freq *b = b_;
- const struct variable *v = v_;
+ struct variable *v = v_;
+ struct var_freqs *vf = get_var_freqs (v);
if (a->c > b->c)
return -1;
else if (a->c < b->c)
return 1;
else
- return memcmp (a->v.s, b->v.s, v->width);
+ return memcmp (a->v[0].s, b->v[0].s, vf->width);
}
\f
/* Frequency table display. */
dump_full (struct variable *v)
{
int n_categories;
+ struct var_freqs *vf;
struct freq_tab *ft;
struct freq *f;
struct tab_table *t;
int lab = cmd.labels == FRQ_LABELS;
- ft = &get_var_freqs (v)->tab;
+ vf = get_var_freqs (v);
+ ft = &vf->tab;
n_categories = ft->n_valid + ft->n_missing;
t = tab_create (5 + lab, n_categories + 3, 0);
tab_headers (t, 0, 0, 2, 0);
if (lab)
{
- const char *label = val_labs_find (v->val_labs, f->v);
+ const char *label = val_labs_find (v->val_labs, f->v[0]);
if (label != NULL)
tab_text (t, 0, r, TAB_LEFT, label);
}
- tab_value (t, 0 + lab, r, TAB_NONE, &f->v, &v->print);
+ tab_value (t, 0 + lab, r, TAB_NONE, f->v, &vf->print);
tab_float (t, 1 + lab, r, TAB_NONE, f->c, 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);
if (lab)
{
- const char *label = val_labs_find (v->val_labs, f->v);
+ const char *label = val_labs_find (v->val_labs, f->v[0]);
if (label != NULL)
tab_text (t, 0, r, TAB_LEFT, label);
}
- tab_value (t, 0 + lab, r, TAB_NONE, &f->v, &v->print);
+ tab_value (t, 0 + lab, r, TAB_NONE, f->v, &vf->print);
tab_float (t, 1 + lab, r, TAB_NONE, f->c, 8, 0);
tab_float (t, 2 + lab, r, TAB_NONE,
f->c / ft->total_cases * 100.0, 5, 1);
}
tab_box (t, TAL_1, TAL_1,
- cmd.spaces == FRQ_SINGLE ? -1 : (TAL_1 | TAL_SPACING), TAL_1,
+ cmd.spaces == FRQ_SINGLE ? -1 : TAL_GAP, TAL_1,
0, 0, 4 + lab, r);
tab_hline (t, TAL_2, 0, 4 + lab, 2);
tab_hline (t, TAL_2, 0, 4 + lab, r);
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_title (t, 1, "%s: %s", v->name, v->label ? v->label : "");
+ tab_title (t, "%s: %s", v->name, v->label ? v->label : "");
tab_submit (t);
}
static void
condensed_dim (struct tab_table *t, struct outp_driver *d)
{
- int cum_w = max (outp_string_width (d, _("Cum")),
- max (outp_string_width (d, _("Cum")),
- outp_string_width (d, "000")));
+ int cum_w = max (outp_string_width (d, _("Cum"), OUTP_PROPORTIONAL),
+ max (outp_string_width (d, _("Cum"), OUTP_PROPORTIONAL),
+ outp_string_width (d, "000", OUTP_PROPORTIONAL)));
int i;
dump_condensed (struct variable *v)
{
int n_categories;
+ struct var_freqs *vf;
struct freq_tab *ft;
struct freq *f;
struct tab_table *t;
int r;
double cum_total = 0.0;
- ft = &get_var_freqs (v)->tab;
+ vf = get_var_freqs (v);
+ ft = &vf->tab;
n_categories = ft->n_valid + ft->n_missing;
t = tab_create (4, n_categories + 2, 0);
percent = f->c / ft->total_cases * 100.0;
cum_total += f->c / ft->valid_cases * 100.0;
- tab_value (t, 0, r, TAB_NONE, &f->v, &v->print);
+ tab_value (t, 0, r, TAB_NONE, f->v, &vf->print);
tab_float (t, 1, r, TAB_NONE, f->c, 8, 0);
tab_float (t, 2, r, TAB_NONE, percent, 3, 0);
tab_float (t, 3, r, TAB_NONE, cum_total, 3, 0);
}
for (; f < &ft->valid[n_categories]; f++)
{
- tab_value (t, 0, r, TAB_NONE, &f->v, &v->print);
+ tab_value (t, 0, r, TAB_NONE, f->v, &vf->print);
tab_float (t, 1, r, TAB_NONE, f->c, 8, 0);
tab_float (t, 2, r, TAB_NONE,
f->c / ft->total_cases * 100.0, 3, 0);
}
tab_box (t, TAL_1, TAL_1,
- cmd.spaces == FRQ_SINGLE ? -1 : (TAL_1 | TAL_SPACING), TAL_1,
+ cmd.spaces == FRQ_SINGLE ? -1 : TAL_GAP, TAL_1,
0, 0, 3, r - 1);
tab_hline (t, TAL_2, 0, 3, 2);
- tab_title (t, 1, "%s: %s", v->name, v->label ? v->label : "");
+ tab_title (t, "%s: %s", v->name, v->label ? v->label : "");
tab_columns (t, SOM_COL_DOWN, 1);
tab_submit (t);
}
if ( percentiles[i].flag )
{
- percentiles[i].x2 = f->v.f;
+ percentiles[i].x2 = f->v[0].f;
percentiles[i].x1 = prev_value;
percentiles[i].flag2 = 1;
continue;
{
if ( f->c > 1 && rank - (f->c - 1) > tp )
{
- percentiles[i].x2 = percentiles[i].x1 = f->v.f;
+ percentiles[i].x2 = percentiles[i].x1 = f->v[0].f;
percentiles[i].flag2 = 1;
}
else
continue;
}
}
- prev_value = f->v.f;
+ prev_value = f->v[0].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->v.f;
+ percentiles[i].x1 = percentiles[i].x2 = f->v[0].f;
/*
printf("percentile %d (p==%.2f); X1 = %g; X2 = %g\n",
if (most_often < f->c)
{
most_often = f->c;
- X_mode = f->v.f;
+ X_mode = f->v[0].f;
}
else if (most_often == f->c)
{
/* Calculate moments. */
m = moments_create (MOMENT_KURTOSIS);
for (f = ft->valid; f < ft->missing; f++)
- moments_pass_one (m, f->v.f, f->c);
+ moments_pass_one (m, f->v[0].f, f->c);
for (f = ft->valid; f < ft->missing; f++)
- moments_pass_two (m, f->v.f, f->c);
+ moments_pass_two (m, f->v[0].f, f->c);
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].v.f;
- d[frq_max] = ft->valid[ft->n_valid - 1].v.f;
+ d[frq_min] = ft->valid[0].v[0].f;
+ d[frq_max] = ft->valid[ft->n_valid - 1].v[0].f;
d[frq_mode] = X_mode;
d[frq_range] = d[frq_max] - d[frq_min];
d[frq_median] = *median_value;
tab_vline (t, TAL_1 , 2, 0, tab_nr(t) - 1);
- tab_vline (t, TAL_1 | TAL_SPACING , 1, 0, tab_nr(t) - 1 ) ;
+ tab_vline (t, TAL_GAP , 1, 0, tab_nr(t) - 1 ) ;
r=2; /* N missing and N valid are always dumped */
if (show_varname)
{
if (v->label)
- tab_title (t, 1, "%s: %s", v->name, v->label);
+ tab_title (t, "%s: %s", v->name, v->label);
else
- tab_title (t, 0, v->name);
+ tab_title (t, "%s", v->name);
}
else
tab_flags (t, SOMF_NO_TITLE);
/* Find out the extremes of the x value */
for ( frq = hsh_first(fh, &hi); frq != 0; frq = hsh_next(fh, &hi) )
{
- if ( mv_is_value_missing(&var->miss, &frq->v))
+ if ( mv_is_value_missing(&var->miss, frq->v))
continue;
- if ( frq->v.f < x_min ) x_min = frq->v.f ;
- if ( frq->v.f > x_max ) x_max = frq->v.f ;
+ if ( frq->v[0].f < x_min ) x_min = frq->v[0].f ;
+ if ( frq->v[0].f > x_max ) x_max = frq->v[0].f ;
}
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->v.f, frq->c);
+ gsl_histogram_accumulate(hist, frq->v[0].f, frq->c);
}
return hist;
{
const struct freq *frq = &frq_tab->valid[i];
- slices[i].label = value_to_string(&frq->v, var);
+ slices[i].label = value_to_string(frq->v, var);
slices[i].magnetude = frq->c;
}
projects / pspp-builds.git / blobdiff