#include "libpspp/pool.h"
#include "libpspp/string-set.h"
#include "libpspp/taint.h"
-
-#include "output/tab.h"
+#include "output/pivot-table.h"
#include "gettext.h"
#define _(msgid) gettext (msgid)
};
static const struct fmt_spec dest_format[n_RANK_FUNCS] = {
- {FMT_F, 9, 3}, /* rank */
- {FMT_F, 6, 4}, /* normal */
- {FMT_F, 6, 2}, /* percent */
- {FMT_F, 6, 4}, /* rfraction */
- {FMT_F, 6, 4}, /* proportion */
- {FMT_F, 6, 0}, /* n */
- {FMT_F, 3, 0}, /* ntiles */
- {FMT_F, 8, 4} /* savage */
+ [RANK] = { .type = FMT_F, .w = 9, .d = 3 },
+ [NORMAL] = { .type = FMT_F, .w = 6, .d = 4 },
+ [PERCENT] = { .type = FMT_F, .w = 6, .d = 2 },
+ [RFRACTION] = { .type = FMT_F, .w = 6, .d = 4 },
+ [PROPORTION] = { .type = FMT_F, .w = 6, .d = 4 },
+ [N] = { .type = FMT_F, .w = 6, .d = 0 },
+ [NTILES] = { .type = FMT_F, .w = 3, .d = 0 },
+ [SAVAGE] = { .type = FMT_F, .w = 8, .d = 4 }
};
static const char * const function_name[n_RANK_FUNCS] = {
rank_savage
};
+static enum measure rank_measures[n_RANK_FUNCS] = {
+ [RANK] = MEASURE_ORDINAL,
+ [NORMAL] = MEASURE_ORDINAL,
+ [PERCENT] = MEASURE_ORDINAL,
+ [RFRACTION] = MEASURE_ORDINAL,
+ [PROPORTION] = MEASURE_ORDINAL,
+ [N] = MEASURE_SCALE,
+ [NTILES] = MEASURE_ORDINAL,
+ [SAVAGE] = MEASURE_ORDINAL,
+};
enum ties
{
}
/* Returns a variable name for storing ranks of a variable named SRC_NAME
- accoring to the rank function F. The name chosen will not be one already in
+ according to the rank function F. The name chosen will not be one already in
DICT or NEW_NAMES.
If successful, adds the new name to NEW_NAMES and returns the name added.
{
free (rank->vars);
free (rank->group_vars);
- subcase_destroy (&rank->sc);
+ subcase_uninit (&rank->sc);
pool_destroy (rank->pool);
}
cmd->rs = pool_realloc (cmd->pool, cmd->rs, sizeof (*cmd->rs) * (cmd->n_rs + 1));
rs = &cmd->rs[cmd->n_rs];
-
+
if (lex_match_id (lexer, "RANK"))
{
rs->rfunc = RANK;
}
else if (lex_match_id (lexer, "NTILES"))
{
- if ( !lex_force_match (lexer, T_LPAREN))
+ if (!lex_force_match (lexer, T_LPAREN))
return false;
-
- if (! lex_force_int (lexer) )
+
+ if (! lex_force_int_range (lexer, "NTILES", 1, INT_MAX))
return false;
-
+
cmd->k_ntiles = lex_integer (lexer);
lex_get (lexer);
-
- if ( !lex_force_match (lexer, T_RPAREN))
+
+ if (!lex_force_match (lexer, T_RPAREN))
return false;
rs->rfunc = NTILES;
if (lex_match_id (lexer, "INTO"))
{
- while( lex_token (lexer) == T_ID )
+ while(lex_token (lexer) == T_ID)
{
const char *name = lex_tokcstr (lexer);
- if ( var_count >= subcase_get_n_fields (&cmd->sc) )
+ if (var_count >= subcase_get_n_fields (&cmd->sc))
msg (SE, _("Too many variables in %s clause."), "INTO");
- else if ( dict_lookup_var (cmd->dict, name) != NULL )
+ else if (dict_lookup_var (cmd->dict, name) != NULL)
msg (SE, _("Variable %s already exists."), name);
else if (string_set_contains (new_names, name))
msg (SE, _("Duplicate variable name %s."), name);
{
double rank;
- if ( c >= 1.0 )
+ if (c >= 1.0)
{
switch (cmd->ties)
{
double f;
- switch ( cmd->fraction )
+ switch (cmd->fraction)
{
case FRAC_BLOM:
f = (r - 3.0/8.0) / (w + 0.25);
f = (r - 1.0/3.0) / (w + 1.0/3.0);
break;
case FRAC_VW:
- f = r / ( w + 1.0);
+ f = r / (w + 1.0);
break;
default:
NOT_REACHED ();
double r = rank_rank (cmd, c, cc, cc_1, i, w);
- return ( floor (( r * cmd->k_ntiles) / ( w + 1) ) + 1);
+ return (floor ((r * cmd->k_ntiles) / (w + 1)) + 1);
}
/* Expected value of the order statistics from an exponential distribution */
double sum = 0.0;
for (k = 1 ; k <= j; k++)
- sum += 1.0 / ( w_star + 1 - k );
+ sum += 1.0 / (w_star + 1 - k);
return sum;
}
const int i_1 = floor (cc_1);
const int i_2 = floor (cc);
- const double w_star = (modf (w, &int_part) == 0 ) ? w : floor (w) + 1;
+ const double w_star = (modf (w, &int_part) == 0) ? w : floor (w) + 1;
const double g_1 = cc_1 - i_1;
const double g_2 = cc - i_2;
/* The second factor is infinite, when the first is zero.
Therefore, evaluate the second, only when the first is non-zero */
- const double expr1 = (1 - g_1) ? (1 - g_1) * ee(i_1+1, w_star) : ( 1 - g_1);
+ const double expr1 = (1 - g_1) ? (1 - g_1) * ee(i_1+1, w_star) : (1 - g_1);
const double expr2 = g_2 ? g_2 * ee (i_2+1, w_star) : g_2 ;
- if ( i_1 == i_2 )
+ if (i_1 == i_2)
return ee (i_1 + 1, w_star) - 1;
- if ( i_1 + 1 == i_2 )
- return ( ( expr1 + expr2 )/c ) - 1;
+ if (i_1 + 1 == i_2)
+ return ((expr1 + expr2)/c) - 1;
- if ( i_1 + 2 <= i_2 )
+ if (i_1 + 2 <= i_2)
{
int j;
double sigma = 0.0;
- for (j = i_1 + 2 ; j <= i_2; ++j )
+ for (j = i_1 + 2 ; j <= i_2; ++j)
sigma += ee (j, w_star);
- return ( (expr1 + expr2 + sigma) / c) -1;
+ return ((expr1 + expr2 + sigma) / c) -1;
}
NOT_REACHED();
/* Do ranking. */
subcase_init (&input_var, 0, 0, SC_ASCEND);
tie_grouper = casegrouper_create_subcase (input, &input_var);
- subcase_destroy (&input_var);
+ subcase_uninit (&input_var);
for (; casegrouper_get_next_group (tie_grouper, &tied_cases);
casereader_destroy (tied_cases))
{
size_t i;
out_case = case_create (casewriter_get_proto (output));
- case_data_rw_idx (out_case, 0)->f = case_num_idx (c, 1);
+ *case_num_rw_idx (out_case, 0) = case_num_idx (c, 1);
for (i = 0; i < cmd->n_rs; ++i)
{
rank_function_t func = rank_func[cmd->rs[i].rfunc];
double rank = func (cmd, tw, cc, cc_1, tie_group, w);
- case_data_rw_idx (out_case, i + 1)->f = rank;
+ *case_num_rw_idx (out_case, i + 1) = rank;
}
casewriter_write (output, out_case);
static const char *
fraction_name (const struct rank *cmd)
{
- switch (cmd->fraction )
+ switch (cmd->fraction)
{
case FRAC_BLOM: return "BLOM";
case FRAC_RANKIT: return "RANKIT";
ds_init_empty (&label);
- if ( cmd->n_group_vars > 0 )
+ if (cmd->n_group_vars > 0)
{
struct string group_var_str;
int g;
ds_init_empty (&group_var_str);
- for (g = 0 ; g < cmd->n_group_vars ; ++g )
+ for (g = 0 ; g < cmd->n_group_vars ; ++g)
{
- if ( g > 0 ) ds_put_cstr (&group_var_str, " ");
+ if (g > 0) ds_put_cstr (&group_var_str, " ");
ds_put_cstr (&group_var_str, var_get_name (cmd->group_vars[g]));
}
function_name[f], var_get_name (src_var));
pool_label = pool_strdup (cmd->pool, ds_cstr (&label));
-
+
ds_destroy (&label);
return pool_label;
struct string_set new_names;
struct rank rank;
struct rank_spec *rs;
- int i;
subcase_init_empty (&rank.sc);
rank.ties = TIES_MEAN;
rank.fraction = FRAC_BLOM;
rank.print = true;
+ rank.vars = NULL;
rank.pool = pool_create ();
string_set_init (&new_names);
rank.n_vars = rank.sc.n_fields;
- if (lex_match (lexer, T_BY) )
+ if (lex_match (lexer, T_BY))
{
- if ( ! parse_variables_const (lexer, rank.dict,
+ if (! parse_variables_const (lexer, rank.dict,
&rank.group_vars, &rank.n_group_vars,
PV_NO_DUPLICATE | PV_NO_SCRATCH))
goto error;
}
- while (lex_token (lexer) != T_ENDCMD )
+ while (lex_token (lexer) != T_ENDCMD)
{
if (! lex_force_match (lexer, T_SLASH))
goto error;
/* If no rank specs are given, then apply a default */
- if ( rank.n_rs == 0)
+ if (rank.n_rs == 0)
{
struct rank_spec *rs;
created with INTO. */
for (rs = rank.rs; rs < &rank.rs[rank.n_rs]; rs++)
{
- int v;
-
rs->dest_labels = pool_calloc (rank.pool, rank.n_vars,
sizeof *rs->dest_labels);
- for ( v = 0 ; v < rank.n_vars ; v ++ )
+ for (int v = 0 ; v < rank.n_vars ; v ++)
{
const char **dst_name = &rs->dest_names[v];
- if ( *dst_name == NULL )
+ if (*dst_name == NULL)
{
*dst_name = rank_choose_dest_name (rank.dict, &new_names,
rs->rfunc,
}
}
- if ( rank.print )
+ if (rank.print)
{
- int v;
+ struct pivot_table *table = pivot_table_create (
+ N_("Variables Created by RANK"));
+
+ pivot_dimension_create (table, PIVOT_AXIS_COLUMN, N_("New Variable"),
+ N_("New Variable"), N_("Function"),
+ N_("Fraction"), N_("Grouping Variables"));
- tab_output_text_format (0, _("Variables Created By %s"), "RANK");
- tab_output_text (0, "");
+ struct pivot_dimension *variables = pivot_dimension_create (
+ table, PIVOT_AXIS_ROW, N_("Existing Variable"),
+ N_("Existing Variable"));
+ variables->root->show_label = true;
- for (i = 0 ; i < rank.n_rs ; ++i )
+ for (size_t i = 0 ; i < rank.n_rs ; ++i)
{
- for ( v = 0 ; v < rank.n_vars ; v ++ )
+ for (size_t v = 0 ; v < rank.n_vars ; v ++)
{
- if ( rank.n_group_vars > 0 )
- {
- struct string varlist;
- int g;
-
- ds_init_empty (&varlist);
- for ( g = 0 ; g < rank.n_group_vars ; ++g )
- {
- ds_put_cstr (&varlist, var_get_name (rank.group_vars[g]));
-
- if ( g < rank.n_group_vars - 1)
- ds_put_cstr (&varlist, " ");
- }
-
- if ( rank.rs[i].rfunc == NORMAL ||
- rank.rs[i].rfunc == PROPORTION )
- tab_output_text_format (0,
- _("%s into %s(%s of %s using %s BY %s)"),
- var_get_name (rank.vars[v]),
- rank.rs[i].dest_names[v],
- function_name[rank.rs[i].rfunc],
- var_get_name (rank.vars[v]),
- fraction_name (&rank),
- ds_cstr (&varlist));
-
- else
- tab_output_text_format (0,
- _("%s into %s(%s of %s BY %s)"),
- var_get_name (rank.vars[v]),
- rank.rs[i].dest_names[v],
- function_name[rank.rs[i].rfunc],
- var_get_name (rank.vars[v]),
- ds_cstr (&varlist));
- ds_destroy (&varlist);
- }
- else
- {
- if ( rank.rs[i].rfunc == NORMAL ||
- rank.rs[i].rfunc == PROPORTION )
- tab_output_text_format (0,
- _("%s into %s(%s of %s using %s)"),
- var_get_name (rank.vars[v]),
- rank.rs[i].dest_names[v],
- function_name[rank.rs[i].rfunc],
- var_get_name (rank.vars[v]),
- fraction_name (&rank));
-
- else
- tab_output_text_format (0,
- _("%s into %s(%s of %s)"),
- var_get_name (rank.vars[v]),
- rank.rs[i].dest_names[v],
- function_name[rank.rs[i].rfunc],
- var_get_name (rank.vars[v]));
- }
+ int row_idx = pivot_category_create_leaf (
+ variables->root, pivot_value_new_variable (rank.vars[v]));
+
+ struct string group_vars = DS_EMPTY_INITIALIZER;
+ for (int g = 0 ; g < rank.n_group_vars ; ++g)
+ {
+ if (g)
+ ds_put_byte (&group_vars, ' ');
+ ds_put_cstr (&group_vars, var_get_name (rank.group_vars[g]));
+ }
+
+ enum rank_func rfunc = rank.rs[i].rfunc;
+ bool has_fraction = rfunc == NORMAL || rfunc == PROPORTION;
+ const char *entries[] =
+ {
+ rank.rs[i].dest_names[v],
+ function_name[rank.rs[i].rfunc],
+ has_fraction ? fraction_name (&rank) : NULL,
+ rank.n_group_vars ? ds_cstr (&group_vars) : NULL,
+ };
+ for (size_t j = 0; j < sizeof entries / sizeof *entries; j++)
+ {
+ const char *entry = entries[j];
+ if (entry)
+ pivot_table_put2 (table, j, row_idx,
+ pivot_value_new_user_text (entry, -1));
+ }
+ ds_destroy (&group_vars);
}
}
+
+ pivot_table_submit (table);
}
/* Do the ranking */
iv->current = casereader_read (iv->input);
}
-static int
+static enum trns_result
rank_trns_proc (void *trns_, struct ccase **c, casenumber case_idx UNUSED)
{
struct rank_trns *trns = trns_;
size_t i;
for (i = 0; i < trns->n_funcs; i++)
- case_data_rw (*c, iv->output_vars[i])->f
+ *case_num_rw (*c, iv->output_vars[i])
= case_num_idx (iv->current, i + 1);
advance_ranking (iv);
break;
return true;
}
+static const struct trns_class rank_trns_class = {
+ .name = "RANK",
+ .execute = rank_trns_proc,
+ .destroy = rank_trns_free,
+};
+
static bool
rank_cmd (struct dataset *ds, const struct rank *cmd)
{
for (i = 0; i < cmd->n_vars; i++)
outputs[i] = sort_create_writer (&by_order, output_proto);
- subcase_destroy (&by_order);
+ subcase_uninit (&by_order);
caseproto_unref (output_proto);
}
/* Open the active file and make one pass per input variable. */
input = proc_open (ds);
input = casereader_create_filter_weight (input, d, NULL, NULL);
- for (i = 0 ; i < cmd->n_vars ; ++i )
+ for (i = 0 ; i < cmd->n_vars ; ++i)
{
const struct variable *input_var = cmd->vars[i];
struct casereader *input_pass;
subcase_add_vars_always (&projection,
cmd->group_vars, cmd->n_group_vars);
subcase_add_vars_always (&projection, dict_get_split_vars (d),
- dict_get_split_cnt (d));
+ dict_get_n_splits (d));
if (weight_var != NULL)
{
subcase_add_var_always (&projection, weight_var, SC_ASCEND);
- weight_idx = 2 + cmd->n_group_vars + dict_get_split_cnt (d);
+ weight_idx = 2 + cmd->n_group_vars + dict_get_n_splits (d);
}
else
weight_idx = -1;
input_pass = casereader_project (input_pass, &projection);
- subcase_destroy (&projection);
+ subcase_uninit (&projection);
/* Prepare 'group_vars' as the set of grouping variables. */
subcase_init_empty (&group_vars);
/* Group by split variables */
subcase_init_empty (&split_vars);
- for (j = 0; j < dict_get_split_cnt (d); j++)
+ for (j = 0; j < dict_get_n_splits (d); j++)
subcase_add_always (&split_vars, 2 + j + cmd->n_group_vars,
var_get_width (dict_get_split_vars (d)[j]),
SC_ASCEND);
split_grouper = casegrouper_create_subcase (input_pass, &split_vars);
- subcase_destroy (&split_vars);
+ subcase_uninit (&split_vars);
while (casegrouper_get_next_group (split_grouper, &split_group))
{
struct casereader *ordered;
rank_sorted_file (by_group, outputs[i], weight_idx, cmd);
ok = casegrouper_destroy (by_grouper) && ok;
}
- subcase_destroy (&group_vars);
- subcase_destroy (&rank_ordering);
+ subcase_uninit (&group_vars);
+ subcase_uninit (&rank_ordering);
ok = casegrouper_destroy (split_grouper) && ok;
}
var = dict_create_var_assert (d, rs->dest_names[i], 0);
var_set_both_formats (var, &dest_format[rs->rfunc]);
var_set_label (var, rs->dest_labels[i]);
+ var_set_measure (var, rank_measures[rs->rfunc]);
iv->output_vars[j] = var;
}
}
free (outputs);
- add_transformation (ds, rank_trns_proc, rank_trns_free, trns);
+ add_transformation (ds, &rank_trns_class, trns);
/* Delete our sort key, which we don't need anymore. */
dict_delete_var (d, order_var);
projects / pspp / blobdiff