X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Fctables.c;h=d6b5a0ca0c119c9764e49377d65a286466f5dcf8;hb=063c177c238c65c15c906507acb1d4199c220ab5;hp=4235002c28932deeb3dfe7ea4486d57758dcf4ab;hpb=3537319aa75f28056beaf85901f1716eb291d739;p=pspp diff --git a/src/language/stats/ctables.c b/src/language/stats/ctables.c index 4235002c28..d6b5a0ca0c 100644 --- a/src/language/stats/ctables.c +++ b/src/language/stats/ctables.c @@ -19,9 +19,11 @@ #include #include "data/casereader.h" +#include "data/casewriter.h" #include "data/dataset.h" #include "data/dictionary.h" #include "data/mrset.h" +#include "data/subcase.h" #include "data/value-labels.h" #include "language/command.h" #include "language/lexer/format-parser.h" @@ -33,7 +35,10 @@ #include "libpspp/hmap.h" #include "libpspp/message.h" #include "libpspp/string-array.h" +#include "math/mode.h" #include "math/moments.h" +#include "math/percentiles.h" +#include "math/sort.h" #include "output/pivot-table.h" #include "gl/minmax.h" @@ -168,8 +173,10 @@ struct ctables_domain const struct ctables_cell *example; - double valid; - double missing; + double d_valid; /* Dictionary weight. */ + double d_missing; + double e_valid; /* Effective weight */ + double e_missing; }; enum ctables_summary_variant @@ -186,6 +193,7 @@ struct ctables_cell struct hmap_node node; /* The domains that contain this cell. */ + bool contributes_to_domains; struct ctables_domain *domains[N_CTDTS]; bool hide; @@ -224,7 +232,7 @@ struct ctables bool mrsets_count_duplicates; /* MRSETS. */ bool smissing_listwise; /* SMISSING. */ - struct variable *base_weight; /* WEIGHT. */ + struct variable *e_weight; /* WEIGHT. */ int hide_threshold; /* HIDESMALLCOUNTS. */ struct ctables_table **tables; @@ -328,7 +336,7 @@ struct ctables_table const struct variable *clabels_example; struct hmap clabels_values_map; - union value *clabels_values; + struct ctables_value **clabels_values; size_t n_clabels_values; enum pivot_axis_type slabels_axis; @@ -1680,7 +1688,15 @@ union ctables_summary /* MEAN, SEMEAN, STDDEV, SUM, VARIANCE, *.SUM. */ struct moments1 *moments; - /* XXX percentiles, median, mode, multiple response */ + /* MEDIAN, MODE, PTILE. */ + struct + { + struct casewriter *writer; + double ovalid; + double ovalue; + }; + + /* XXX multiple response */ }; static void @@ -1712,6 +1728,7 @@ ctables_summary_init (union ctables_summary *s, case CTSF_LAYERPCT_TOTALN: case CTSF_LAYERROWPCT_TOTALN: case CTSF_LAYERCOLPCT_TOTALN: + case CTSF_MISSING: case CSTF_TOTALN: case CTSF_ETOTALN: case CTSF_VALIDN: @@ -1741,10 +1758,23 @@ ctables_summary_init (union ctables_summary *s, break; case CTSF_MEDIAN: - case CTSF_MISSING: case CTSF_MODE: case CTSF_PTILE: - NOT_REACHED (); + { + struct caseproto *proto = caseproto_create (); + proto = caseproto_add_width (proto, 0); + proto = caseproto_add_width (proto, 0); + + struct subcase ordering; + subcase_init (&ordering, 0, 0, SC_ASCEND); + s->writer = sort_create_writer (&ordering, proto); + subcase_uninit (&ordering); + caseproto_unref (proto); + + s->ovalid = 0; + s->ovalue = SYSMIS; + } + break; case CTSF_RESPONSES: case CTSF_ROWPCT_RESPONSES: @@ -1801,6 +1831,7 @@ ctables_summary_uninit (union ctables_summary *s, case CTSF_LAYERPCT_TOTALN: case CTSF_LAYERROWPCT_TOTALN: case CTSF_LAYERCOLPCT_TOTALN: + case CTSF_MISSING: case CSTF_TOTALN: case CTSF_ETOTALN: case CTSF_VALIDN: @@ -1828,10 +1859,10 @@ ctables_summary_uninit (union ctables_summary *s, break; case CTSF_MEDIAN: - case CTSF_MISSING: case CTSF_MODE: case CTSF_PTILE: - NOT_REACHED (); + casewriter_destroy (s->writer); + break; case CTSF_RESPONSES: case CTSF_ROWPCT_RESPONSES: @@ -1863,11 +1894,19 @@ static void ctables_summary_add (union ctables_summary *s, const struct ctables_summary_spec *ss, const struct variable *var, const union value *value, - double weight) + double d_weight, double e_weight) { switch (ss->function) { case CTSF_COUNT: + case CSTF_TOTALN: + case CTSF_VALIDN: + if (var_is_value_missing (var, value)) + s->missing += d_weight; + else + s->valid += d_weight; + break; + case CTSF_ECOUNT: case CTSF_ROWPCT_COUNT: case CTSF_COLPCT_COUNT: @@ -1890,14 +1929,13 @@ ctables_summary_add (union ctables_summary *s, case CTSF_LAYERPCT_TOTALN: case CTSF_LAYERROWPCT_TOTALN: case CTSF_LAYERCOLPCT_TOTALN: - case CSTF_TOTALN: + case CTSF_MISSING: case CTSF_ETOTALN: - case CTSF_VALIDN: case CTSF_EVALIDN: if (var_is_value_missing (var, value)) - s->missing += weight; + s->missing += e_weight; else - s->valid += weight; + s->valid += e_weight; break; case CTSF_MAXIMUM: @@ -1925,14 +1963,23 @@ ctables_summary_add (union ctables_summary *s, case CTSF_LAYERPCT_SUM: case CTSF_LAYERROWPCT_SUM: case CTSF_LAYERCOLPCT_SUM: - moments1_add (s->moments, value->f, weight); + if (!var_is_value_missing (var, value)) + moments1_add (s->moments, value->f, e_weight); break; case CTSF_MEDIAN: - case CTSF_MISSING: case CTSF_MODE: case CTSF_PTILE: - NOT_REACHED (); + if (var_is_value_missing (var, value)) + { + s->ovalid += e_weight; + + struct ccase *c = case_create (casewriter_get_proto (s->writer)); + *case_num_rw_idx (c, 0) = value->f; + *case_num_rw_idx (c, 1) = e_weight; + casewriter_write (s->writer, c); + } + break; case CTSF_RESPONSES: case CTSF_ROWPCT_RESPONSES: @@ -1960,6 +2007,99 @@ ctables_summary_add (union ctables_summary *s, } } +static enum ctables_domain_type +ctables_function_domain (enum ctables_summary_function function) +{ + switch (function) + { + case CTSF_COUNT: + case CTSF_ECOUNT: + case CTSF_MISSING: + case CSTF_TOTALN: + case CTSF_ETOTALN: + case CTSF_VALIDN: + case CTSF_EVALIDN: + case CTSF_MAXIMUM: + case CTSF_MINIMUM: + case CTSF_RANGE: + case CTSF_MEAN: + case CTSF_SEMEAN: + case CTSF_STDDEV: + case CTSF_SUM: + case CTSF_VARIANCE: + case CTSF_MEDIAN: + case CTSF_PTILE: + case CTSF_MODE: + case CTSF_RESPONSES: + NOT_REACHED (); + + case CTSF_COLPCT_COUNT: + case CTSF_COLPCT_COUNT_RESPONSES: + case CTSF_COLPCT_RESPONSES: + case CTSF_COLPCT_RESPONSES_COUNT: + case CTSF_COLPCT_SUM: + case CTSF_COLPCT_TOTALN: + case CTSF_COLPCT_VALIDN: + return CTDT_COL; + + case CTSF_LAYERCOLPCT_COUNT: + case CTSF_LAYERCOLPCT_COUNT_RESPONSES: + case CTSF_LAYERCOLPCT_RESPONSES: + case CTSF_LAYERCOLPCT_RESPONSES_COUNT: + case CTSF_LAYERCOLPCT_SUM: + case CTSF_LAYERCOLPCT_TOTALN: + case CTSF_LAYERCOLPCT_VALIDN: + return CTDT_LAYERCOL; + + case CTSF_LAYERPCT_COUNT: + case CTSF_LAYERPCT_COUNT_RESPONSES: + case CTSF_LAYERPCT_RESPONSES: + case CTSF_LAYERPCT_RESPONSES_COUNT: + case CTSF_LAYERPCT_SUM: + case CTSF_LAYERPCT_TOTALN: + case CTSF_LAYERPCT_VALIDN: + return CTDT_LAYER; + + case CTSF_LAYERROWPCT_COUNT: + case CTSF_LAYERROWPCT_COUNT_RESPONSES: + case CTSF_LAYERROWPCT_RESPONSES: + case CTSF_LAYERROWPCT_RESPONSES_COUNT: + case CTSF_LAYERROWPCT_SUM: + case CTSF_LAYERROWPCT_TOTALN: + case CTSF_LAYERROWPCT_VALIDN: + return CTDT_LAYERROW; + + case CTSF_ROWPCT_COUNT: + case CTSF_ROWPCT_COUNT_RESPONSES: + case CTSF_ROWPCT_RESPONSES: + case CTSF_ROWPCT_RESPONSES_COUNT: + case CTSF_ROWPCT_SUM: + case CTSF_ROWPCT_TOTALN: + case CTSF_ROWPCT_VALIDN: + return CTDT_ROW; + + case CTSF_SUBTABLEPCT_COUNT: + case CTSF_SUBTABLEPCT_COUNT_RESPONSES: + case CTSF_SUBTABLEPCT_RESPONSES: + case CTSF_SUBTABLEPCT_RESPONSES_COUNT: + case CTSF_SUBTABLEPCT_SUM: + case CTSF_SUBTABLEPCT_TOTALN: + case CTSF_SUBTABLEPCT_VALIDN: + return CTDT_SUBTABLE; + + case CTSF_TABLEPCT_COUNT: + case CTSF_TABLEPCT_COUNT_RESPONSES: + case CTSF_TABLEPCT_RESPONSES: + case CTSF_TABLEPCT_RESPONSES_COUNT: + case CTSF_TABLEPCT_SUM: + case CTSF_TABLEPCT_TOTALN: + case CTSF_TABLEPCT_VALIDN: + return CTDT_TABLE; + } + + NOT_REACHED (); +} + static double ctables_summary_value (const struct ctables_cell *cell, union ctables_summary *s, @@ -1971,26 +2111,19 @@ ctables_summary_value (const struct ctables_cell *cell, case CTSF_ECOUNT: return s->valid; - case CTSF_SUBTABLEPCT_COUNT: - return cell->domains[CTDT_SUBTABLE]->valid ? s->valid / cell->domains[CTDT_SUBTABLE]->valid * 100 : SYSMIS; - case CTSF_ROWPCT_COUNT: - return cell->domains[CTDT_ROW]->valid ? s->valid / cell->domains[CTDT_ROW]->valid * 100 : SYSMIS; - case CTSF_COLPCT_COUNT: - return cell->domains[CTDT_COL]->valid ? s->valid / cell->domains[CTDT_COL]->valid * 100 : SYSMIS; - case CTSF_TABLEPCT_COUNT: - return cell->domains[CTDT_TABLE]->valid ? s->valid / cell->domains[CTDT_TABLE]->valid * 100 : SYSMIS; - + case CTSF_SUBTABLEPCT_COUNT: case CTSF_LAYERPCT_COUNT: - return cell->domains[CTDT_LAYER]->valid ? s->valid / cell->domains[CTDT_LAYER]->valid * 100 : SYSMIS; - case CTSF_LAYERROWPCT_COUNT: - return cell->domains[CTDT_LAYERROW]->valid ? s->valid / cell->domains[CTDT_LAYERROW]->valid * 100 : SYSMIS; - case CTSF_LAYERCOLPCT_COUNT: - return cell->domains[CTDT_LAYERCOL]->valid ? s->valid / cell->domains[CTDT_LAYERCOL]->valid * 100 : SYSMIS; + { + enum ctables_domain_type d = ctables_function_domain (ss->function); + return (cell->domains[d]->e_valid + ? s->valid / cell->domains[d]->e_valid * 100 + : SYSMIS); + } case CTSF_ROWPCT_VALIDN: case CTSF_COLPCT_VALIDN: @@ -2008,6 +2141,9 @@ ctables_summary_value (const struct ctables_cell *cell, case CTSF_LAYERCOLPCT_TOTALN: NOT_REACHED (); + case CTSF_MISSING: + return s->missing; + case CSTF_TOTALN: case CTSF_ETOTALN: return s->valid + s->missing; @@ -2070,10 +2206,34 @@ ctables_summary_value (const struct ctables_cell *cell, NOT_REACHED (); case CTSF_MEDIAN: - case CTSF_MISSING: - case CTSF_MODE: case CTSF_PTILE: - NOT_REACHED (); + if (s->writer) + { + struct casereader *reader = casewriter_make_reader (s->writer); + s->writer = NULL; + + struct percentile *ptile = percentile_create ( + ss->function == CTSF_PTILE ? ss->percentile : 0.5, s->ovalid); + struct order_stats *os = &ptile->parent; + order_stats_accumulate_idx (&os, 1, reader, 1, 0); + s->ovalue = percentile_calculate (ptile, PC_HAVERAGE); + statistic_destroy (&ptile->parent.parent); + } + return s->ovalue; + + case CTSF_MODE: + if (s->writer) + { + struct casereader *reader = casewriter_make_reader (s->writer); + s->writer = NULL; + + struct mode *mode = mode_create (); + struct order_stats *os = &mode->parent; + order_stats_accumulate_idx (&os, 1, reader, 1, 0); + s->ovalue = mode->mode; + statistic_destroy (&mode->parent.parent); + } + return s->ovalue; case CTSF_RESPONSES: case CTSF_ROWPCT_RESPONSES: @@ -2365,6 +2525,7 @@ ctables_cell_insert__ (struct ctables_table *t, const struct ccase *c, cell = xmalloc (sizeof *cell); cell->hide = false; cell->sv = sv; + cell->contributes_to_domains = true; for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++) { const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]]; @@ -2374,14 +2535,19 @@ ctables_cell_insert__ (struct ctables_table *t, const struct ccase *c, : NULL); for (size_t i = 0; i < nest->n; i++) { + const struct ctables_category *cat = cats[a][i]; + if (i != nest->scale_idx) { - const struct ctables_category *subtotal = cats[a][i]->subtotal; + const struct ctables_category *subtotal = cat->subtotal; if (subtotal && subtotal->type == CCT_HSUBTOTAL) cell->hide = true; + + if (cat->type == CCT_TOTAL || cat->type == CCT_SUBTOTAL || cat->type == CCT_HSUBTOTAL) + cell->contributes_to_domains = false; } - cell->axes[a].cvs[i].category = cats[a][i]; + cell->axes[a].cvs[i].category = cat; value_clone (&cell->axes[a].cvs[i].value, case_data (c, nest->vars[i]), var_get_width (nest->vars[i])); } @@ -2401,7 +2567,7 @@ static void ctables_cell_add__ (struct ctables_table *t, const struct ccase *c, size_t ix[PIVOT_N_AXES], const struct ctables_category *cats[PIVOT_N_AXES][10], - double weight) + double d_weight, double e_weight) { struct ctables_cell *cell = ctables_cell_insert__ (t, c, ix, cats); const struct ctables_nest *ss = &t->stacks[t->summary_axis].nests[ix[t->summary_axis]]; @@ -2409,16 +2575,22 @@ ctables_cell_add__ (struct ctables_table *t, const struct ccase *c, const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv]; for (size_t i = 0; i < specs->n; i++) ctables_summary_add (&cell->summaries[i], &specs->specs[i], specs->var, - case_data (c, specs->var), weight); - for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++) - cell->domains[dt]->valid += weight; + case_data (c, specs->var), d_weight, e_weight); + if (cell->contributes_to_domains) + { + for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++) + { + cell->domains[dt]->d_valid += d_weight; + cell->domains[dt]->e_valid += e_weight; + } + } } static void recurse_totals (struct ctables_table *t, const struct ccase *c, size_t ix[PIVOT_N_AXES], const struct ctables_category *cats[PIVOT_N_AXES][10], - double weight, + double d_weight, double e_weight, enum pivot_axis_type start_axis, size_t start_nest) { for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++) @@ -2437,8 +2609,36 @@ recurse_totals (struct ctables_table *t, const struct ccase *c, { const struct ctables_category *save = cats[a][i]; cats[a][i] = total; - ctables_cell_add__ (t, c, ix, cats, weight); - recurse_totals (t, c, ix, cats, weight, a, i + 1); + ctables_cell_add__ (t, c, ix, cats, d_weight, e_weight); + recurse_totals (t, c, ix, cats, d_weight, e_weight, a, i + 1); + cats[a][i] = save; + } + } + start_nest = 0; + } +} + +static void +recurse_subtotals (struct ctables_table *t, const struct ccase *c, + size_t ix[PIVOT_N_AXES], + const struct ctables_category *cats[PIVOT_N_AXES][10], + double d_weight, double e_weight, + enum pivot_axis_type start_axis, size_t start_nest) +{ + for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++) + { + const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]]; + for (size_t i = start_nest; i < nest->n; i++) + { + if (i == nest->scale_idx) + continue; + + const struct ctables_category *save = cats[a][i]; + if (save->subtotal) + { + cats[a][i] = save->subtotal; + ctables_cell_add__ (t, c, ix, cats, d_weight, e_weight); + recurse_subtotals (t, c, ix, cats, d_weight, e_weight, a, i + 1); cats[a][i] = save; } } @@ -2449,15 +2649,9 @@ recurse_totals (struct ctables_table *t, const struct ccase *c, static void ctables_cell_insert (struct ctables_table *t, const struct ccase *c, - size_t ir, size_t ic, size_t il, - double weight) + size_t ix[PIVOT_N_AXES], + double d_weight, double e_weight) { - size_t ix[PIVOT_N_AXES] = { - [PIVOT_AXIS_ROW] = ir, - [PIVOT_AXIS_COLUMN] = ic, - [PIVOT_AXIS_LAYER] = il, - }; - const struct ctables_category *cats[PIVOT_N_AXES][10]; for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++) { @@ -2480,27 +2674,10 @@ ctables_cell_insert (struct ctables_table *t, } } - ctables_cell_add__ (t, c, ix, cats, weight); - - recurse_totals (t, c, ix, cats, weight, 0, 0); + ctables_cell_add__ (t, c, ix, cats, d_weight, e_weight); - for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++) - { - const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]]; - for (size_t i = 0; i < nest->n; i++) - { - if (i == nest->scale_idx) - continue; - - const struct ctables_category *save = cats[a][i]; - if (save->subtotal) - { - cats[a][i] = save->subtotal; - ctables_cell_add__ (t, c, ix, cats, weight); - cats[a][i] = save; - } - } - } + recurse_totals (t, c, ix, cats, d_weight, e_weight, 0, 0); + recurse_subtotals (t, c, ix, cats, d_weight, e_weight, 0, 0); } struct merge_item @@ -2566,12 +2743,16 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t) pivot_table_set_caption ( pt, pivot_value_new_user_text (t->corner, SIZE_MAX)); - bool summary_dimension = t->summary_axis != t->slabels_axis; + bool summary_dimension = (t->summary_axis != t->slabels_axis + || (!t->slabels_visible + && t->summary_specs.n > 1)); if (summary_dimension) { struct pivot_dimension *d = pivot_dimension_create ( - pt, t->slabels_axis, N_("Summaries")); + pt, t->slabels_axis, N_("Statistics")); const struct ctables_summary_spec_set *specs = &t->summary_specs; + if (!t->slabels_visible) + d->hide_all_labels = true; for (size_t i = 0; i < specs->n; i++) pivot_category_create_leaf ( d->root, pivot_value_new_text (specs->specs[i].label)); @@ -2589,15 +2770,11 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t) const struct ctables_categories *c = t->categories[var_get_dict_index (var)]; for (size_t i = 0; i < t->n_clabels_values; i++) { - const union value *value = &t->clabels_values[i]; - const struct ctables_category *cat = ctables_categories_match (c, value, var); - if (!cat) - { - pivot_category_create_leaf (d->root, pivot_value_new_integer (value->f)); /* XXX */ - continue; - } + const struct ctables_value *value = t->clabels_values[i]; + const struct ctables_category *cat = ctables_categories_match (c, &value->value, var); + assert (cat != NULL); pivot_category_create_leaf (d->root, ctables_category_create_label ( - cat, t->clabels_example, value)); + cat, t->clabels_example, &value->value)); } } @@ -2682,13 +2859,11 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t) if (new_subtable) { n_levels = 0; - printf ("%s levels:", pivot_axis_type_to_string (a)); for (size_t k = 0; k < nest->n; k++) { enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (nest->vars[k])]; if (vlabel != CTVL_NONE) { - printf (" var(%s)", var_get_name (nest->vars[k])); levels[n_levels++] = (struct ctables_level) { .type = CTL_VAR, .var_idx = k, @@ -2698,7 +2873,6 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t) if (nest->scale_idx != k && (k != nest->n - 1 || t->label_axis[a] == a)) { - printf (" category(%s)", var_get_name (nest->vars[k])); levels[n_levels++] = (struct ctables_level) { .type = CTL_CATEGORY, .var_idx = k, @@ -2706,15 +2880,13 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t) } } - if (a == t->slabels_axis && a == t->summary_axis) + if (!summary_dimension && a == t->slabels_axis) { - printf (" summary"); levels[n_levels++] = (struct ctables_level) { .type = CTL_SUMMARY, .var_idx = SIZE_MAX, }; } - printf ("\n"); } size_t n_common = 0; @@ -2726,17 +2898,16 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t) if (level->type == CTL_CATEGORY) { size_t var_idx = level->var_idx; - if (prev->axes[a].cvs[var_idx].category - != cell->axes[a].cvs[var_idx].category) - { - break; - } - else if (!value_equal (&prev->axes[a].cvs[var_idx].value, - &cell->axes[a].cvs[var_idx].value, - var_get_type (nest->vars[var_idx]))) - { - break; - } + const struct ctables_category *c = cell->axes[a].cvs[var_idx].category; + if (prev->axes[a].cvs[var_idx].category != c) + break; + else if (c->type != CCT_SUBTOTAL + && c->type != CCT_HSUBTOTAL + && c->type != CCT_TOTAL + && !value_equal (&prev->axes[a].cvs[var_idx].value, + &cell->axes[a].cvs[var_idx].value, + var_get_type (nest->vars[var_idx]))) + break; } } } @@ -2747,6 +2918,8 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t) struct pivot_category *parent = k ? groups[k - 1] : d[a]->root; if (level->type == CTL_SUMMARY) { + assert (k == n_levels - 1); + const struct ctables_summary_spec_set *specs = &t->summary_specs; for (size_t m = 0; m < specs->n; m++) { @@ -2806,10 +2979,8 @@ ctables_table_output (struct ctables *ct, struct ctables_table *t) const struct variable *var = clabels_nest->vars[clabels_nest->n - 1]; const union value *value = &cell->axes[t->clabels_from_axis].cvs[clabels_nest->n - 1].value; const struct ctables_value *ctv = ctables_value_find (t, value, var_get_width (var)); - - printf ("leaf=%d\n", ctv ? ctv->leaf : 0); - dindexes[n_dindexes++] = ctv ? ctv->leaf : 0; /* XXX */ - //dindexes[n_dindexes++] = 0; + assert (ctv != NULL); + dindexes[n_dindexes++] = ctv->leaf; } for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++) @@ -3085,9 +3256,17 @@ ctables_insert_clabels_values (struct ctables_table *t, const struct ccase *c, for (size_t i = 0; i < stack->n; i++) { const struct ctables_nest *nest = &stack->nests[i]; - const struct variable *v = nest->vars[nest->n - 1]; - int width = var_get_width (v); - const union value *value = case_data (c, v); + const struct variable *var = nest->vars[nest->n - 1]; + int width = var_get_width (var); + const union value *value = case_data (c, var); + + if (var_is_numeric (var) && value->f == SYSMIS) + continue; + + if (!ctables_categories_match (t->categories [var_get_dict_index (var)], + value, var)) + continue; + unsigned int hash = value_hash (value, width, 0); struct ctables_value *clv = ctables_value_find__ (t, value, width, hash); @@ -3103,10 +3282,12 @@ ctables_insert_clabels_values (struct ctables_table *t, const struct ccase *c, static int compare_clabels_values_3way (const void *a_, const void *b_, const void *width_) { - const union value *a = a_; - const union value *b = b_; + const struct ctables_value *const *ap = a_; + const struct ctables_value *const *bp = b_; + const struct ctables_value *a = *ap; + const struct ctables_value *b = *bp; const int *width = width_; - return value_compare_3way (a, b, *width); + return value_compare_3way (&a->value, &b->value, *width); } static void @@ -3120,32 +3301,32 @@ ctables_sort_clabels_values (struct ctables_table *t) struct ctables_value *clv; size_t i = 0; HMAP_FOR_EACH (clv, struct ctables_value, node, &t->clabels_values_map) - { - clv->leaf = i; - t->clabels_values[i] = clv->value; - i++; - } + t->clabels_values[i++] = clv; t->n_clabels_values = n; assert (i == n); sort (t->clabels_values, n, sizeof *t->clabels_values, compare_clabels_values_3way, &width); + + for (size_t i = 0; i < n; i++) + t->clabels_values[i]->leaf = i; } static bool ctables_execute (struct dataset *ds, struct ctables *ct) { - struct casereader *input = casereader_create_filter_weight (proc_open (ds), - dataset_dict (ds), - NULL, NULL); + struct casereader *input = proc_open (ds); bool warn_on_invalid = true; - double total_weight = 0; for (struct ccase *c = casereader_read (input); c; case_unref (c), c = casereader_read (input)) { - double weight = dict_get_case_weight (dataset_dict (ds), c, - &warn_on_invalid); - total_weight += weight; + double d_weight = dict_get_case_weight (dataset_dict (ds), c, + &warn_on_invalid); + double e_weight = (ct->e_weight + ? var_force_valid_weight (ct->e_weight, + case_num (c, ct->e_weight), + &warn_on_invalid) + : d_weight); for (size_t i = 0; i < ct->n_tables; i++) { @@ -3154,7 +3335,15 @@ ctables_execute (struct dataset *ds, struct ctables *ct) for (size_t ir = 0; ir < t->stacks[PIVOT_AXIS_ROW].n; ir++) for (size_t ic = 0; ic < t->stacks[PIVOT_AXIS_COLUMN].n; ic++) for (size_t il = 0; il < t->stacks[PIVOT_AXIS_LAYER].n; il++) - ctables_cell_insert (t, c, ir, ic, il, weight); + { + size_t ix[PIVOT_N_AXES] = { + [PIVOT_AXIS_ROW] = ir, + [PIVOT_AXIS_COLUMN] = ic, + [PIVOT_AXIS_LAYER] = il, + }; + + ctables_cell_insert (t, c, ix, d_weight, e_weight); + } for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++) if (t->label_axis[a] != a) @@ -3354,8 +3543,8 @@ cmd_ctables (struct lexer *lexer, struct dataset *ds) if (!lex_force_match_id (lexer, "VARIABLE")) goto error; lex_match (lexer, T_EQUALS); - ct->base_weight = parse_variable (lexer, dataset_dict (ds)); - if (!ct->base_weight) + ct->e_weight = parse_variable (lexer, dataset_dict (ds)); + if (!ct->e_weight) goto error; } else if (lex_match_id (lexer, "HIDESMALLCOUNTS"))