1 /* PSPP - a program for statistical analysis.
2 Copyright (C) 2021 Free Software Foundation, Inc.
4 This program is free software: you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation, either version 3 of the License, or
7 (at your option) any later version.
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program. If not, see <http://www.gnu.org/licenses/>. */
21 #include "data/casereader.h"
22 #include "data/dataset.h"
23 #include "data/dictionary.h"
24 #include "data/mrset.h"
25 #include "language/command.h"
26 #include "language/lexer/format-parser.h"
27 #include "language/lexer/lexer.h"
28 #include "language/lexer/variable-parser.h"
29 #include "libpspp/array.h"
30 #include "libpspp/assertion.h"
31 #include "libpspp/hash-functions.h"
32 #include "libpspp/hmap.h"
33 #include "libpspp/message.h"
34 #include "libpspp/string-array.h"
35 #include "math/moments.h"
36 #include "output/pivot-table.h"
38 #include "gl/minmax.h"
39 #include "gl/xalloc.h"
42 #define _(msgid) gettext (msgid)
43 #define N_(msgid) (msgid)
47 CTVL_NONE = SETTINGS_VALUE_SHOW_DEFAULT,
48 CTVL_NAME = SETTINGS_VALUE_SHOW_VALUE,
49 CTVL_LABEL = SETTINGS_VALUE_SHOW_LABEL,
50 CTVL_BOTH = SETTINGS_VALUE_SHOW_BOTH,
54 - unweighted summaries (U*)
55 - lower confidence limits (*.LCL)
56 - upper confidence limits (*.UCL)
57 - standard error (*.SE)
60 /* All variables. */ \
61 S(CTSF_COUNT, "COUNT", N_("Count"), CTF_COUNT, CTFA_ALL) \
62 S(CTSF_ECOUNT, "ECOUNT", N_("Adjusted Count"), CTF_COUNT, CTFA_ALL) \
63 S(CTSF_ROWPCT_COUNT, "ROWPCT.COUNT", N_("Row %"), CTF_PERCENT, CTFA_ALL) \
64 S(CTSF_COLPCT_COUNT, "COLPCT.COUNT", N_("Column %"), CTF_PERCENT, CTFA_ALL) \
65 S(CTSF_TABLEPCT_COUNT, "TABLEPCT.COUNT", N_("Table %"), CTF_PERCENT, CTFA_ALL) \
66 S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT.COUNT", N_("Subtable %"), CTF_PERCENT, CTFA_ALL) \
67 S(CTSF_LAYERPCT_COUNT, "LAYERPCT.COUNT", N_("Layer %"), CTF_PERCENT, CTFA_ALL) \
68 S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT.COUNT", N_("Layer Row %"), CTF_PERCENT, CTFA_ALL) \
69 S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT.COUNT", N_("Layer Column %"), CTF_PERCENT, CTFA_ALL) \
70 S(CTSF_ROWPCT_VALIDN, "ROWPCT.VALIDN", N_("Row Valid N %"), CTF_PERCENT, CTFA_ALL) \
71 S(CTSF_COLPCT_VALIDN, "COLPCT.VALIDN", N_("Column Valid N %"), CTF_PERCENT, CTFA_ALL) \
72 S(CTSF_TABLEPCT_VALIDN, "TABLEPCT.VALIDN", N_("Table Valid N %"), CTF_PERCENT, CTFA_ALL) \
73 S(CTSF_SUBTABLEPCT_VALIDN, "SUBTABLEPCT.VALIDN", N_("Subtable Valid N %"), CTF_PERCENT, CTFA_ALL) \
74 S(CTSF_LAYERPCT_VALIDN, "LAYERPCT.VALIDN", N_("Layer Valid N %"), CTF_PERCENT, CTFA_ALL) \
75 S(CTSF_LAYERROWPCT_VALIDN, "LAYERROWPCT.VALIDN", N_("Layer Row Valid N %"), CTF_PERCENT, CTFA_ALL) \
76 S(CTSF_LAYERCOLPCT_VALIDN, "LAYERCOLPCT.VALIDN", N_("Layer Column Valid N %"), CTF_PERCENT, CTFA_ALL) \
77 S(CTSF_ROWPCT_TOTALN, "ROWPCT.TOTALN", N_("Row Total N %"), CTF_PERCENT, CTFA_ALL) \
78 S(CTSF_COLPCT_TOTALN, "COLPCT.TOTALN", N_("Column Total N %"), CTF_PERCENT, CTFA_ALL) \
79 S(CTSF_TABLEPCT_TOTALN, "TABLEPCT.TOTALN", N_("Table Total N %"), CTF_PERCENT, CTFA_ALL) \
80 S(CTSF_SUBTABLEPCT_TOTALN, "SUBTABLEPCT.TOTALN", N_("Subtable Total N %"), CTF_PERCENT, CTFA_ALL) \
81 S(CTSF_LAYERPCT_TOTALN, "LAYERPCT.TOTALN", N_("Layer Total N %"), CTF_PERCENT, CTFA_ALL) \
82 S(CTSF_LAYERROWPCT_TOTALN, "LAYERROWPCT.TOTALN", N_("Layer Row Total N %"), CTF_PERCENT, CTFA_ALL) \
83 S(CTSF_LAYERCOLPCT_TOTALN, "LAYERCOLPCT.TOTALN", N_("Layer Column Total N %"), CTF_PERCENT, CTFA_ALL) \
85 /* Scale variables, totals, and subtotals. */ \
86 S(CTSF_MAXIMUM, "MAXIMUM", N_("Maximum"), CTF_GENERAL, CTFA_SCALE) \
87 S(CTSF_MEAN, "MEAN", N_("Mean"), CTF_GENERAL, CTFA_SCALE) \
88 S(CTSF_MEDIAN, "MEDIAN", N_("Median"), CTF_GENERAL, CTFA_SCALE) \
89 S(CTSF_MINIMUM, "MINIMUM", N_("Minimum"), CTF_GENERAL, CTFA_SCALE) \
90 S(CTSF_MISSING, "MISSING", N_("Missing"), CTF_GENERAL, CTFA_SCALE) \
91 S(CTSF_MODE, "MODE", N_("Mode"), CTF_GENERAL, CTFA_SCALE) \
92 S(CTSF_PTILE, "PTILE", N_("Percentile"), CTF_GENERAL, CTFA_SCALE) \
93 S(CTSF_RANGE, "RANGE", N_("Range"), CTF_GENERAL, CTFA_SCALE) \
94 S(CTSF_SEMEAN, "SEMEAN", N_("Std Error of Mean"), CTF_GENERAL, CTFA_SCALE) \
95 S(CTSF_STDDEV, "STDDEV", N_("Std Deviation"), CTF_GENERAL, CTFA_SCALE) \
96 S(CTSF_SUM, "SUM", N_("Sum"), CTF_GENERAL, CTFA_SCALE) \
97 S(CSTF_TOTALN, "TOTALN", N_("Total N"), CTF_COUNT, CTFA_SCALE) \
98 S(CTSF_ETOTALN, "ETOTALN", N_("Adjusted Total N"), CTF_COUNT, CTFA_SCALE) \
99 S(CTSF_VALIDN, "VALIDN", N_("Valid N"), CTF_COUNT, CTFA_SCALE) \
100 S(CTSF_EVALIDN, "EVALIDN", N_("Adjusted Valid N"), CTF_COUNT, CTFA_SCALE) \
101 S(CTSF_VARIANCE, "VARIANCE", N_("Variance"), CTF_GENERAL, CTFA_SCALE) \
102 S(CTSF_ROWPCT_SUM, "ROWPCT.SUM", N_("Row Sum %"), CTF_PERCENT, CTFA_SCALE) \
103 S(CTSF_COLPCT_SUM, "COLPCT.SUM", N_("Column Sum %"), CTF_PERCENT, CTFA_SCALE) \
104 S(CTSF_TABLEPCT_SUM, "TABLEPCT.SUM", N_("Table Sum %"), CTF_PERCENT, CTFA_SCALE) \
105 S(CTSF_SUBTABLEPCT_SUM, "SUBTABLEPCT.SUM", N_("Subtable Sum %"), CTF_PERCENT, CTFA_SCALE) \
106 S(CTSF_LAYERPCT_SUM, "LAYERPCT.SUM", N_("Layer Sum %"), CTF_PERCENT, CTFA_SCALE) \
107 S(CTSF_LAYERROWPCT_SUM, "LAYERROWPCT.SUM", N_("Layer Row Sum %"), CTF_PERCENT, CTFA_SCALE) \
108 S(CTSF_LAYERCOLPCT_SUM, "LAYERCOLPCT.SUM", N_("Layer Column Sum %"), CTF_PERCENT, CTFA_SCALE) \
110 /* Multiple response sets. */ \
111 S(CTSF_RESPONSES, "RESPONSES", N_("Responses"), CTF_COUNT, CTFA_MRSETS) \
112 S(CTSF_ROWPCT_RESPONSES, "ROWPCT.RESPONSES", N_("Row Responses %"), CTF_PERCENT, CTFA_MRSETS) \
113 S(CTSF_COLPCT_RESPONSES, "COLPCT.RESPONSES", N_("Column Responses %"), CTF_PERCENT, CTFA_MRSETS) \
114 S(CTSF_TABLEPCT_RESPONSES, "TABLEPCT.RESPONSES", N_("Table Responses %"), CTF_PERCENT, CTFA_MRSETS) \
115 S(CTSF_SUBTABLEPCT_RESPONSES, "SUBTABLEPCT.RESPONSES", N_("Subtable Responses %"), CTF_PERCENT, CTFA_MRSETS) \
116 S(CTSF_LAYERPCT_RESPONSES, "LAYERPCT.RESPONSES", N_("Layer Responses %"), CTF_PERCENT, CTFA_MRSETS) \
117 S(CTSF_LAYERROWPCT_RESPONSES, "LAYERROWPCT.RESPONSES", N_("Layer Row Responses %"), CTF_PERCENT, CTFA_MRSETS) \
118 S(CTSF_LAYERCOLPCT_RESPONSES, "LAYERCOLPCT.RESPONSES", N_("Layer Column Responses %"), CTF_PERCENT, CTFA_MRSETS) \
119 S(CTSF_ROWPCT_RESPONSES_COUNT, "ROWPCT.RESPONSES.COUNT", N_("Row Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
120 S(CTSF_COLPCT_RESPONSES_COUNT, "COLPCT.RESPONSES.COUNT", N_("Column Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
121 S(CTSF_TABLEPCT_RESPONSES_COUNT, "TABLEPCT.RESPONSES.COUNT", N_("Table Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
122 S(CTSF_SUBTABLEPCT_RESPONSES_COUNT, "SUBTABLEPCT.RESPONSES.COUNT", N_("Subtable Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
123 S(CTSF_LAYERPCT_RESPONSES_COUNT, "LAYERPCT.RESPONSES.COUNT", N_("Layer Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
124 S(CTSF_LAYERROWPCT_RESPONSES_COUNT, "LAYERROWPCT.RESPONSES.COUNT", N_("Layer Row Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
125 S(CTSF_LAYERCOLPCT_RESPONSES_COUNT, "LAYERCOLPCT.RESPONSES.COUNT", N_("Layer Column Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
126 S(CTSF_ROWPCT_COUNT_RESPONSES, "ROWPCT.COUNT.RESPONSES", N_("Row Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
127 S(CTSF_COLPCT_COUNT_RESPONSES, "COLPCT.COUNT.RESPONSES", N_("Column Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
128 S(CTSF_TABLEPCT_COUNT_RESPONSES, "TABLEPCT.COUNT.RESPONSES", N_("Table Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
129 S(CTSF_SUBTABLEPCT_COUNT_RESPONSES, "SUBTABLEPCT.COUNT.RESPONSES", N_("Subtable Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
130 S(CTSF_LAYERPCT_COUNT_RESPONSES, "LAYERPCT.COUNT.RESPONSES", N_("Layer Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
131 S(CTSF_LAYERROWPCT_COUNT_RESPONSES, "LAYERROWPCT.COUNT.RESPONSES", N_("Layer Row Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
132 S(CTSF_LAYERCOLPCT_COUNT_RESPONSES, "LAYERCOLPCT.RESPONSES.COUNT", N_("Layer Column Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS)
134 enum ctables_summary_function
136 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) ENUM,
142 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) +1
143 N_CTSF_FUNCTIONS = SUMMARIES
147 enum ctables_domain_type
149 /* Within a section, where stacked variables divide one section from
151 CTDT_TABLE, /* All layers of a whole section. */
152 CTDT_LAYER, /* One layer within a section. */
153 CTDT_LAYERROW, /* Row in one layer within a section. */
154 CTDT_LAYERCOL, /* Column in one layer within a section. */
156 /* Within a subtable, where a subtable pairs an innermost row variable with
157 an innermost column variable within a single layer. */
158 CTDT_SUBTABLE, /* Whole subtable. */
159 CTDT_ROW, /* Row within a subtable. */
160 CTDT_COL, /* Column within a subtable. */
164 struct ctables_domain
166 struct hmap_node node;
168 const struct ctables_cell *example;
174 enum ctables_summary_variant
183 /* In struct ctables's 'cells' hmap. Indexed by all the values in all the
184 axes (except the scalar variable, if any). */
185 struct hmap_node node;
187 /* The domains that contain this cell. */
188 struct ctables_domain *domains[N_CTDTS];
191 enum ctables_summary_variant sv;
196 struct ctables_cell_value
198 const struct ctables_category *category;
206 union ctables_summary *summaries;
211 struct pivot_table_look *look;
213 /* If this is NULL, zeros are displayed using the normal print format.
214 Otherwise, this string is displayed. */
217 /* If this is NULL, missing values are displayed using the normal print
218 format. Otherwise, this string is displayed. */
221 /* Indexed by variable dictionary index. */
222 enum ctables_vlabel *vlabels;
224 bool mrsets_count_duplicates; /* MRSETS. */
225 bool smissing_listwise; /* SMISSING. */
226 struct variable *base_weight; /* WEIGHT. */
227 int hide_threshold; /* HIDESMALLCOUNTS. */
229 struct ctables_table **tables;
233 struct ctables_postcompute
235 struct hmap_node hmap_node; /* In struct ctables's 'pcompute' hmap. */
236 const char *name; /* Name, without leading &. */
238 struct ctables_postcompute_expr *expr;
241 bool hide_source_cats;
244 struct ctables_postcompute_expr
246 enum ctables_postcompute_op
254 /* XXX SUBTOTAL and HSUBTOTAL */
267 /* CTPO_CAT_NUMBER, CTPO_NUMBER. */
272 XXX what about string ranges? */
275 /* CTPO_ADD, CTPO_SUB, CTPO_MUL, CTPO_DIV, CTPO_POW. */
276 struct ctables_postcompute_expr *subs[2];
280 enum ctables_label_position
287 struct ctables_summary_spec_set
289 struct ctables_summary_spec *specs;
293 struct variable *var;
296 static void ctables_summary_spec_set_clone (struct ctables_summary_spec_set *,
297 const struct ctables_summary_spec_set *);
298 static void ctables_summary_spec_set_uninit (struct ctables_summary_spec_set *);
300 /* A nested sequence of variables, e.g. a > b > c. */
303 struct variable **vars;
306 size_t *domains[N_CTDTS];
307 size_t n_domains[N_CTDTS];
309 struct ctables_summary_spec_set specs[N_CSVS];
312 /* A stack of nestings, e.g. nest1 + nest2 + ... + nestN. */
315 struct ctables_nest *nests;
321 struct ctables_axis *axes[PIVOT_N_AXES];
322 struct ctables_stack stacks[PIVOT_N_AXES];
323 enum pivot_axis_type summary_axis;
324 struct ctables_summary_spec_set summary_specs;
326 struct hmap domains[N_CTDTS];
328 enum pivot_axis_type slabels_position;
329 bool slabels_visible;
331 enum ctables_label_position row_labels;
332 enum ctables_label_position col_labels;
334 /* Indexed by variable dictionary index. */
335 struct ctables_categories **categories;
344 struct ctables_chisq *chisq;
345 struct ctables_pairwise *pairwise;
353 struct variable *var;
354 const struct mrset *mrset;
358 static const struct fmt_spec *
359 ctables_var_get_print_format (const struct ctables_var *var)
361 return (var->is_mrset
362 ? var_get_print_format (var->mrset->vars[0])
363 : var_get_print_format (var->var));
367 ctables_var_name (const struct ctables_var *var)
369 return var->is_mrset ? var->mrset->name : var_get_name (var->var);
372 struct ctables_categories
375 struct ctables_category *cats;
380 struct ctables_category
382 enum ctables_category_type
400 struct ctables_category *subtotal;
404 double number; /* CCT_NUMBER. */
405 char *string; /* CCT_STRING. */
406 double range[2]; /* CCT_RANGE. */
407 char *total_label; /* CCT_SUBTOTAL, CCT_HSUBTOTAL, CCT_TOTAL. */
409 /* CCT_VALUE, CCT_LABEL, CCT_FUNCTION. */
412 bool include_missing;
416 enum ctables_summary_function sort_function;
417 struct variable *sort_var;
424 ctables_category_uninit (struct ctables_category *cat)
444 free (cat->total_label);
455 ctables_categories_unref (struct ctables_categories *c)
460 assert (c->n_refs > 0);
464 for (size_t i = 0; i < c->n_cats; i++)
465 ctables_category_uninit (&c->cats[i]);
470 /* Chi-square test (SIGTEST). */
478 /* Pairwise comparison test (COMPARETEST). */
479 struct ctables_pairwise
481 enum { PROP, MEAN } type;
484 bool meansvariance_allcats;
486 enum { BONFERRONI = 1, BH } adjust;
510 struct ctables_var var;
512 struct ctables_summary_spec_set specs[N_CSVS];
516 struct ctables_axis *subs[2];
519 struct msg_location *loc;
522 static void ctables_axis_destroy (struct ctables_axis *);
531 enum ctables_function_availability
533 CTFA_ALL, /* Any variables. */
534 CTFA_SCALE, /* Only scale variables, totals, and subtotals. */
535 CTFA_MRSETS, /* Only multiple-response sets */
538 struct ctables_summary_spec
540 enum ctables_summary_function function;
541 double percentile; /* CTSF_PTILE only. */
543 struct fmt_spec format; /* XXX extra CTABLES formats */
548 ctables_summary_spec_clone (struct ctables_summary_spec *dst,
549 const struct ctables_summary_spec *src)
552 dst->label = xstrdup (src->label);
556 ctables_summary_spec_uninit (struct ctables_summary_spec *s)
563 ctables_summary_spec_set_clone (struct ctables_summary_spec_set *dst,
564 const struct ctables_summary_spec_set *src)
566 struct ctables_summary_spec *specs = xnmalloc (src->n, sizeof *specs);
567 for (size_t i = 0; i < src->n; i++)
568 ctables_summary_spec_clone (&specs[i], &src->specs[i]);
570 *dst = (struct ctables_summary_spec_set) {
579 ctables_summary_spec_set_uninit (struct ctables_summary_spec_set *set)
581 for (size_t i = 0; i < set->n; i++)
582 ctables_summary_spec_uninit (&set->specs[i]);
587 parse_col_width (struct lexer *lexer, const char *name, double *width)
589 lex_match (lexer, T_EQUALS);
590 if (lex_match_id (lexer, "DEFAULT"))
592 else if (lex_force_num_range_closed (lexer, name, 0, DBL_MAX))
594 *width = lex_number (lexer);
604 parse_bool (struct lexer *lexer, bool *b)
606 if (lex_match_id (lexer, "NO"))
608 else if (lex_match_id (lexer, "YES"))
612 lex_error_expecting (lexer, "YES", "NO");
618 static enum ctables_function_availability
619 ctables_function_availability (enum ctables_summary_function f)
621 static enum ctables_function_availability availability[] = {
622 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = AVAILABILITY,
627 return availability[f];
631 parse_ctables_summary_function (struct lexer *lexer,
632 enum ctables_summary_function *f)
636 enum ctables_summary_function function;
637 struct substring name;
639 static struct pair names[] = {
640 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) \
641 { ENUM, SS_LITERAL_INITIALIZER (NAME) },
644 /* The .COUNT suffix may be omitted. */
645 S(CTSF_ROWPCT_COUNT, "ROWPCT", _, _, _)
646 S(CTSF_COLPCT_COUNT, "COLPCT", _, _, _)
647 S(CTSF_TABLEPCT_COUNT, "TABLEPCT", _, _, _)
648 S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT", _, _, _)
649 S(CTSF_LAYERPCT_COUNT, "LAYERPCT", _, _, _)
650 S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT", _, _, _)
651 S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT", _, _, _)
655 if (!lex_force_id (lexer))
658 for (size_t i = 0; i < sizeof names / sizeof *names; i++)
659 if (ss_equals_case (names[i].name, lex_tokss (lexer)))
661 *f = names[i].function;
666 lex_error (lexer, _("Expecting summary function name."));
671 ctables_axis_destroy (struct ctables_axis *axis)
679 for (size_t i = 0; i < N_CSVS; i++)
680 ctables_summary_spec_set_uninit (&axis->specs[i]);
685 ctables_axis_destroy (axis->subs[0]);
686 ctables_axis_destroy (axis->subs[1]);
689 msg_location_destroy (axis->loc);
693 static struct ctables_axis *
694 ctables_axis_new_nonterminal (enum ctables_axis_op op,
695 struct ctables_axis *sub0,
696 struct ctables_axis *sub1,
697 struct lexer *lexer, int start_ofs)
699 struct ctables_axis *axis = xmalloc (sizeof *axis);
700 *axis = (struct ctables_axis) {
702 .subs = { sub0, sub1 },
703 .loc = lex_ofs_location (lexer, start_ofs, lex_ofs (lexer) - 1),
708 struct ctables_axis_parse_ctx
711 struct dictionary *dict;
713 struct ctables_table *t;
716 static struct fmt_spec
717 ctables_summary_default_format (enum ctables_summary_function function,
718 const struct ctables_var *var)
720 static const enum ctables_format default_formats[] = {
721 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = FORMAT,
725 switch (default_formats[function])
728 return (struct fmt_spec) { .type = FMT_F, .w = 40 };
731 return (struct fmt_spec) { .type = FMT_PCT, .w = 40, .d = 1 };
734 return *ctables_var_get_print_format (var);
742 ctables_summary_default_label (enum ctables_summary_function function,
745 static const char *default_labels[] = {
746 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = LABEL,
751 return (function == CTSF_PTILE
752 ? xasprintf (_("Percentile %.2f"), percentile)
753 : xstrdup (gettext (default_labels[function])));
757 ctables_summary_function_name (enum ctables_summary_function function)
759 static const char *names[] = {
760 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = NAME,
764 return names[function];
768 add_summary_spec (struct ctables_axis *axis,
769 enum ctables_summary_function function, double percentile,
770 const char *label, const struct fmt_spec *format,
771 const struct msg_location *loc, enum ctables_summary_variant sv)
773 if (axis->op == CTAO_VAR)
775 const char *function_name = ctables_summary_function_name (function);
776 const char *var_name = ctables_var_name (&axis->var);
777 switch (ctables_function_availability (function))
780 if (!axis->var.is_mrset)
782 msg_at (SE, loc, _("Summary function %s applies only to multiple "
783 "response sets."), function_name);
784 msg_at (SN, axis->loc, _("'%s' is not a multiple response set."),
794 _("Summary function %s applies only to scale variables."),
796 msg_at (SN, axis->loc, _("'%s' is not a scale variable."),
806 struct ctables_summary_spec_set *set = &axis->specs[sv];
807 if (set->n >= set->allocated)
808 set->specs = x2nrealloc (set->specs, &set->allocated,
811 struct ctables_summary_spec *dst = &set->specs[set->n++];
812 *dst = (struct ctables_summary_spec) {
813 .function = function,
814 .percentile = percentile,
815 .label = xstrdup (label),
816 .format = (format ? *format
817 : ctables_summary_default_format (function, &axis->var)),
823 for (size_t i = 0; i < 2; i++)
824 if (!add_summary_spec (axis->subs[i], function, percentile, label,
831 static struct ctables_axis *ctables_axis_parse_stack (
832 struct ctables_axis_parse_ctx *);
835 ctables_var_parse (struct lexer *lexer, struct dictionary *dict,
836 struct ctables_var *var)
838 if (ss_starts_with (lex_tokss (lexer), ss_cstr ("$")))
840 *var = (struct ctables_var) {
842 .mrset = dict_lookup_mrset (dict, lex_tokcstr (lexer))
846 lex_error (lexer, _("'%s' does not name a multiple-response set "
847 "in the active file dictionary."),
848 lex_tokcstr (lexer));
856 *var = (struct ctables_var) {
858 .var = parse_variable (lexer, dict),
860 return var->var != NULL;
864 static struct ctables_axis *
865 ctables_axis_parse_primary (struct ctables_axis_parse_ctx *ctx)
867 if (lex_match (ctx->lexer, T_LPAREN))
869 struct ctables_axis *sub = ctables_axis_parse_stack (ctx);
870 if (!sub || !lex_force_match (ctx->lexer, T_RPAREN))
872 ctables_axis_destroy (sub);
878 if (!lex_force_id (ctx->lexer))
881 int start_ofs = lex_ofs (ctx->lexer);
882 struct ctables_var var;
883 if (!ctables_var_parse (ctx->lexer, ctx->dict, &var))
886 struct ctables_axis *axis = xmalloc (sizeof *axis);
887 *axis = (struct ctables_axis) { .op = CTAO_VAR, .var = var };
889 /* XXX should figure out default measures by reading data */
890 axis->scale = (var.is_mrset ? false
891 : lex_match_phrase (ctx->lexer, "[S]") ? true
892 : lex_match_phrase (ctx->lexer, "[C]") ? false
893 : var_get_measure (var.var) == MEASURE_SCALE);
894 axis->loc = lex_ofs_location (ctx->lexer, start_ofs,
895 lex_ofs (ctx->lexer) - 1);
900 has_digit (const char *s)
902 return s[strcspn (s, "0123456789")] != '\0';
905 static struct ctables_axis *
906 ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
908 struct ctables_axis *sub = ctables_axis_parse_primary (ctx);
909 if (!sub || !lex_match (ctx->lexer, T_LBRACK))
912 enum ctables_summary_variant sv = CSV_CELL;
915 int start_ofs = lex_ofs (ctx->lexer);
917 /* Parse function. */
918 enum ctables_summary_function function;
919 if (!parse_ctables_summary_function (ctx->lexer, &function))
922 /* Parse percentile. */
923 double percentile = 0;
924 if (function == CTSF_PTILE)
926 if (!lex_force_num_range_closed (ctx->lexer, "PTILE", 0, 100))
928 percentile = lex_number (ctx->lexer);
929 lex_get (ctx->lexer);
934 if (lex_is_string (ctx->lexer))
936 label = ss_xstrdup (lex_tokss (ctx->lexer));
937 lex_get (ctx->lexer);
940 label = ctables_summary_default_label (function, percentile);
943 struct fmt_spec format;
944 const struct fmt_spec *formatp;
945 if (lex_token (ctx->lexer) == T_ID
946 && has_digit (lex_tokcstr (ctx->lexer)))
948 if (!parse_format_specifier (ctx->lexer, &format)
949 || !fmt_check_output (&format)
950 || !fmt_check_type_compat (&format, VAL_NUMERIC))
960 struct msg_location *loc = lex_ofs_location (ctx->lexer, start_ofs,
961 lex_ofs (ctx->lexer) - 1);
962 printf ("add %s\n", ctables_summary_function_name (function));
963 add_summary_spec (sub, function, percentile, label, formatp, loc, sv);
965 msg_location_destroy (loc);
967 lex_match (ctx->lexer, T_COMMA);
968 if (sv == CSV_CELL && lex_match_id (ctx->lexer, "TOTALS"))
970 if (!lex_force_match (ctx->lexer, T_LBRACK))
974 else if (lex_match (ctx->lexer, T_RBRACK))
976 if (sv == CSV_TOTAL && !lex_force_match (ctx->lexer, T_RBRACK))
983 ctables_axis_destroy (sub);
987 static const struct ctables_axis *
988 find_scale (const struct ctables_axis *axis)
992 else if (axis->op == CTAO_VAR)
996 assert (!axis->var.is_mrset);
1004 for (size_t i = 0; i < 2; i++)
1006 const struct ctables_axis *scale = find_scale (axis->subs[i]);
1014 static const struct ctables_axis *
1015 find_categorical_summary_spec (const struct ctables_axis *axis)
1019 else if (axis->op == CTAO_VAR)
1020 return !axis->scale && axis->specs[CSV_CELL].n ? axis : NULL;
1023 for (size_t i = 0; i < 2; i++)
1025 const struct ctables_axis *sum
1026 = find_categorical_summary_spec (axis->subs[i]);
1034 static struct ctables_axis *
1035 ctables_axis_parse_nest (struct ctables_axis_parse_ctx *ctx)
1037 int start_ofs = lex_ofs (ctx->lexer);
1038 struct ctables_axis *lhs = ctables_axis_parse_postfix (ctx);
1042 while (lex_match (ctx->lexer, T_GT))
1044 struct ctables_axis *rhs = ctables_axis_parse_postfix (ctx);
1048 struct ctables_axis *nest = ctables_axis_new_nonterminal (
1049 CTAO_NEST, lhs, rhs, ctx->lexer, start_ofs);
1051 const struct ctables_axis *outer_scale = find_scale (lhs);
1052 const struct ctables_axis *inner_scale = find_scale (rhs);
1053 if (outer_scale && inner_scale)
1055 msg_at (SE, nest->loc, _("Cannot nest scale variables."));
1056 msg_at (SN, outer_scale->loc, _("This is an outer scale variable."));
1057 msg_at (SN, inner_scale->loc, _("This is an inner scale variable."));
1058 ctables_axis_destroy (nest);
1062 const struct ctables_axis *outer_sum = find_categorical_summary_spec (lhs);
1065 msg_at (SE, nest->loc,
1066 _("Summaries may only be requested for categorical variables "
1067 "at the innermost nesting level."));
1068 msg_at (SN, outer_sum->loc,
1069 _("This outer categorical variable has a summary."));
1070 ctables_axis_destroy (nest);
1080 static struct ctables_axis *
1081 ctables_axis_parse_stack (struct ctables_axis_parse_ctx *ctx)
1083 int start_ofs = lex_ofs (ctx->lexer);
1084 struct ctables_axis *lhs = ctables_axis_parse_nest (ctx);
1088 while (lex_match (ctx->lexer, T_PLUS))
1090 struct ctables_axis *rhs = ctables_axis_parse_nest (ctx);
1094 lhs = ctables_axis_new_nonterminal (CTAO_STACK, lhs, rhs,
1095 ctx->lexer, start_ofs);
1102 ctables_axis_parse (struct lexer *lexer, struct dictionary *dict,
1103 struct ctables *ct, struct ctables_table *t,
1104 enum pivot_axis_type a)
1106 if (lex_token (lexer) == T_BY
1107 || lex_token (lexer) == T_SLASH
1108 || lex_token (lexer) == T_ENDCMD)
1111 struct ctables_axis_parse_ctx ctx = {
1117 t->axes[a] = ctables_axis_parse_stack (&ctx);
1118 return t->axes[a] != NULL;
1122 ctables_chisq_destroy (struct ctables_chisq *chisq)
1128 ctables_pairwise_destroy (struct ctables_pairwise *pairwise)
1134 ctables_table_destroy (struct ctables_table *t)
1139 for (size_t i = 0; i < t->n_categories; i++)
1140 ctables_categories_unref (t->categories[i]);
1141 free (t->categories);
1143 ctables_axis_destroy (t->axes[PIVOT_AXIS_COLUMN]);
1144 ctables_axis_destroy (t->axes[PIVOT_AXIS_ROW]);
1145 ctables_axis_destroy (t->axes[PIVOT_AXIS_LAYER]);
1149 ctables_chisq_destroy (t->chisq);
1150 ctables_pairwise_destroy (t->pairwise);
1155 ctables_destroy (struct ctables *ct)
1160 pivot_table_look_unref (ct->look);
1164 for (size_t i = 0; i < ct->n_tables; i++)
1165 ctables_table_destroy (ct->tables[i]);
1170 static struct ctables_category
1171 cct_range (double low, double high)
1173 return (struct ctables_category) {
1175 .range = { low, high }
1180 ctables_table_parse_categories (struct lexer *lexer, struct dictionary *dict,
1181 struct ctables_table *t)
1183 if (!lex_match_id (lexer, "VARIABLES"))
1185 lex_match (lexer, T_EQUALS);
1187 struct variable **vars;
1189 if (!parse_variables (lexer, dict, &vars, &n_vars, PV_NO_SCRATCH))
1192 struct ctables_categories *c = xmalloc (sizeof *c);
1193 *c = (struct ctables_categories) { .n_refs = n_vars };
1194 for (size_t i = 0; i < n_vars; i++)
1196 struct ctables_categories **cp
1197 = &t->categories[var_get_dict_index (vars[i])];
1198 ctables_categories_unref (*cp);
1203 size_t allocated_cats = 0;
1204 if (lex_match (lexer, T_LBRACK))
1208 if (c->n_cats >= allocated_cats)
1209 c->cats = x2nrealloc (c->cats, &allocated_cats, sizeof *c->cats);
1211 struct ctables_category *cat = &c->cats[c->n_cats];
1212 if (lex_match_id (lexer, "OTHERNM"))
1213 cat->type = CCT_OTHERNM;
1214 else if (lex_match_id (lexer, "MISSING"))
1215 cat->type = CCT_MISSING;
1216 else if (lex_match_id (lexer, "SUBTOTAL"))
1217 *cat = (struct ctables_category)
1218 { .type = CCT_SUBTOTAL, .total_label = NULL };
1219 else if (lex_match_id (lexer, "HSUBTOTAL"))
1220 *cat = (struct ctables_category)
1221 { .type = CCT_HSUBTOTAL, .total_label = NULL };
1222 else if (lex_match_id (lexer, "LO"))
1224 if (!lex_force_match_id (lexer, "THRU") || lex_force_num (lexer))
1226 *cat = cct_range (-DBL_MAX, lex_number (lexer));
1229 else if (lex_is_number (lexer))
1231 double number = lex_number (lexer);
1233 if (lex_match_id (lexer, "THRU"))
1235 cat->type = CCT_RANGE;
1236 cat->range[0] = number;
1237 if (lex_match_id (lexer, "HI"))
1238 *cat = cct_range (number, DBL_MAX);
1241 if (!lex_force_num (lexer))
1243 *cat = cct_range (number, lex_number (lexer));
1248 *cat = (struct ctables_category) {
1253 else if (lex_is_string (lexer))
1255 *cat = (struct ctables_category) {
1257 .string = ss_xstrdup (lex_tokss (lexer)),
1263 lex_error (lexer, NULL);
1267 if (cat->type == CCT_SUBTOTAL || cat->type == CCT_HSUBTOTAL)
1269 if (lex_match (lexer, T_EQUALS))
1271 if (!lex_force_string (lexer))
1274 cat->total_label = ss_xstrdup (lex_tokss (lexer));
1278 cat->total_label = xstrdup (_("Subtotal"));
1282 lex_match (lexer, T_COMMA);
1284 while (!lex_match (lexer, T_RBRACK));
1287 struct ctables_category cat = {
1289 .include_missing = false,
1290 .sort_ascending = true,
1292 bool show_totals = false;
1293 char *total_label = NULL;
1294 bool totals_before = false;
1295 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
1297 if (!c->n_cats && lex_match_id (lexer, "ORDER"))
1299 lex_match (lexer, T_EQUALS);
1300 if (lex_match_id (lexer, "A"))
1301 cat.sort_ascending = true;
1302 else if (lex_match_id (lexer, "D"))
1303 cat.sort_ascending = false;
1306 lex_error_expecting (lexer, "A", "D");
1310 else if (!c->n_cats && lex_match_id (lexer, "KEY"))
1312 lex_match (lexer, T_EQUALS);
1313 if (lex_match_id (lexer, "VALUE"))
1314 cat.type = CCT_VALUE;
1315 else if (lex_match_id (lexer, "LABEL"))
1316 cat.type = CCT_LABEL;
1319 cat.type = CCT_FUNCTION;
1320 if (!parse_ctables_summary_function (lexer, &cat.sort_function))
1323 if (lex_match (lexer, T_LPAREN))
1325 cat.sort_var = parse_variable (lexer, dict);
1329 if (cat.sort_function == CTSF_PTILE)
1331 lex_match (lexer, T_COMMA);
1332 if (!lex_force_num_range_closed (lexer, "PTILE", 0, 100))
1334 cat.percentile = lex_number (lexer);
1338 if (!lex_force_match (lexer, T_RPAREN))
1341 else if (ctables_function_availability (cat.sort_function)
1344 bool UNUSED b = lex_force_match (lexer, T_LPAREN);
1349 else if (!c->n_cats && lex_match_id (lexer, "MISSING"))
1351 lex_match (lexer, T_EQUALS);
1352 if (lex_match_id (lexer, "INCLUDE"))
1353 cat.include_missing = true;
1354 else if (lex_match_id (lexer, "EXCLUDE"))
1355 cat.include_missing = false;
1358 lex_error_expecting (lexer, "INCLUDE", "EXCLUDE");
1362 else if (lex_match_id (lexer, "TOTAL"))
1364 lex_match (lexer, T_EQUALS);
1365 if (!parse_bool (lexer, &show_totals))
1368 else if (lex_match_id (lexer, "LABEL"))
1370 lex_match (lexer, T_EQUALS);
1371 if (!lex_force_string (lexer))
1374 total_label = ss_xstrdup (lex_tokss (lexer));
1377 else if (lex_match_id (lexer, "POSITION"))
1379 lex_match (lexer, T_EQUALS);
1380 if (lex_match_id (lexer, "BEFORE"))
1381 totals_before = true;
1382 else if (lex_match_id (lexer, "AFTER"))
1383 totals_before = false;
1386 lex_error_expecting (lexer, "BEFORE", "AFTER");
1390 else if (lex_match_id (lexer, "EMPTY"))
1392 lex_match (lexer, T_EQUALS);
1393 if (lex_match_id (lexer, "INCLUDE"))
1394 c->show_empty = true;
1395 else if (lex_match_id (lexer, "EXCLUDE"))
1396 c->show_empty = false;
1399 lex_error_expecting (lexer, "INCLUDE", "EXCLUDE");
1406 lex_error_expecting (lexer, "ORDER", "KEY", "MISSING",
1407 "TOTAL", "LABEL", "POSITION", "EMPTY");
1409 lex_error_expecting (lexer, "TOTAL", "LABEL", "POSITION", "EMPTY");
1416 if (c->n_cats >= allocated_cats)
1417 c->cats = x2nrealloc (c->cats, &allocated_cats,
1419 c->cats[c->n_cats++] = cat;
1424 if (c->n_cats >= allocated_cats)
1425 c->cats = x2nrealloc (c->cats, &allocated_cats, sizeof *c->cats);
1427 struct ctables_category *totals;
1430 insert_element (c->cats, c->n_cats, sizeof *c->cats, 0);
1431 totals = &c->cats[0];
1434 totals = &c->cats[c->n_cats];
1437 *totals = (struct ctables_category) {
1439 .total_label = total_label ? total_label : xstrdup (_("Total")),
1443 struct ctables_category *subtotal = NULL;
1444 for (size_t i = totals_before ? 0 : c->n_cats;
1445 totals_before ? i < c->n_cats : i-- > 0;
1446 totals_before ? i++ : 0)
1448 struct ctables_category *cat = &c->cats[i];
1456 cat->subtotal = subtotal;
1476 ctables_nest_uninit (struct ctables_nest *nest)
1483 ctables_stack_uninit (struct ctables_stack *stack)
1487 for (size_t i = 0; i < stack->n; i++)
1488 ctables_nest_uninit (&stack->nests[i]);
1489 free (stack->nests);
1493 static struct ctables_stack
1494 nest_fts (struct ctables_stack s0, struct ctables_stack s1)
1501 struct ctables_stack stack = { .nests = xnmalloc (s0.n, s1.n * sizeof *stack.nests) };
1502 for (size_t i = 0; i < s0.n; i++)
1503 for (size_t j = 0; j < s1.n; j++)
1505 const struct ctables_nest *a = &s0.nests[i];
1506 const struct ctables_nest *b = &s1.nests[j];
1508 size_t allocate = a->n + b->n;
1509 struct variable **vars = xnmalloc (allocate, sizeof *vars);
1510 enum pivot_axis_type *axes = xnmalloc (allocate, sizeof *axes);
1512 for (size_t k = 0; k < a->n; k++)
1513 vars[n++] = a->vars[k];
1514 for (size_t k = 0; k < b->n; k++)
1515 vars[n++] = b->vars[k];
1516 assert (n == allocate);
1518 const struct ctables_nest *summary_src;
1519 if (!a->specs[CSV_CELL].var)
1521 else if (!b->specs[CSV_CELL].var)
1526 struct ctables_nest *new = &stack.nests[stack.n++];
1527 *new = (struct ctables_nest) {
1529 .scale_idx = (a->scale_idx != SIZE_MAX ? a->scale_idx
1530 : b->scale_idx != SIZE_MAX ? a->n + b->scale_idx
1534 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
1535 ctables_summary_spec_set_clone (&new->specs[sv], &summary_src->specs[sv]);
1537 ctables_stack_uninit (&s0);
1538 ctables_stack_uninit (&s1);
1542 static struct ctables_stack
1543 stack_fts (struct ctables_stack s0, struct ctables_stack s1)
1545 struct ctables_stack stack = { .nests = xnmalloc (s0.n + s1.n, sizeof *stack.nests) };
1546 for (size_t i = 0; i < s0.n; i++)
1547 stack.nests[stack.n++] = s0.nests[i];
1548 for (size_t i = 0; i < s1.n; i++)
1549 stack.nests[stack.n++] = s1.nests[i];
1550 assert (stack.n == s0.n + s1.n);
1556 static struct ctables_stack
1557 enumerate_fts (enum pivot_axis_type axis_type, const struct ctables_axis *a)
1560 return (struct ctables_stack) { .n = 0 };
1565 assert (!a->var.is_mrset);
1567 struct variable **vars = xmalloc (sizeof *vars);
1570 struct ctables_nest *nest = xmalloc (sizeof *nest);
1571 *nest = (struct ctables_nest) {
1574 .scale_idx = a->scale ? 0 : SIZE_MAX,
1576 if (a->specs[CSV_CELL].n || a->scale)
1577 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
1579 ctables_summary_spec_set_clone (&nest->specs[sv], &a->specs[sv]);
1580 nest->specs[sv].var = a->var.var;
1582 return (struct ctables_stack) { .nests = nest, .n = 1 };
1585 return stack_fts (enumerate_fts (axis_type, a->subs[0]),
1586 enumerate_fts (axis_type, a->subs[1]));
1589 return nest_fts (enumerate_fts (axis_type, a->subs[0]),
1590 enumerate_fts (axis_type, a->subs[1]));
1596 union ctables_summary
1598 /* COUNT, VALIDN, TOTALN. */
1605 /* MINIMUM, MAXIMUM, RANGE. */
1612 /* MEAN, SEMEAN, STDDEV, SUM, VARIANCE, *.SUM. */
1613 struct moments1 *moments;
1615 /* XXX percentiles, median, mode, multiple response */
1619 ctables_summary_init (union ctables_summary *s,
1620 const struct ctables_summary_spec *ss)
1622 switch (ss->function)
1626 case CTSF_ROWPCT_COUNT:
1627 case CTSF_COLPCT_COUNT:
1628 case CTSF_TABLEPCT_COUNT:
1629 case CTSF_SUBTABLEPCT_COUNT:
1630 case CTSF_LAYERPCT_COUNT:
1631 case CTSF_LAYERROWPCT_COUNT:
1632 case CTSF_LAYERCOLPCT_COUNT:
1633 case CTSF_ROWPCT_VALIDN:
1634 case CTSF_COLPCT_VALIDN:
1635 case CTSF_TABLEPCT_VALIDN:
1636 case CTSF_SUBTABLEPCT_VALIDN:
1637 case CTSF_LAYERPCT_VALIDN:
1638 case CTSF_LAYERROWPCT_VALIDN:
1639 case CTSF_LAYERCOLPCT_VALIDN:
1640 case CTSF_ROWPCT_TOTALN:
1641 case CTSF_COLPCT_TOTALN:
1642 case CTSF_TABLEPCT_TOTALN:
1643 case CTSF_SUBTABLEPCT_TOTALN:
1644 case CTSF_LAYERPCT_TOTALN:
1645 case CTSF_LAYERROWPCT_TOTALN:
1646 case CTSF_LAYERCOLPCT_TOTALN:
1651 s->missing = s->valid = 0;
1657 s->min = s->max = SYSMIS;
1665 case CTSF_ROWPCT_SUM:
1666 case CTSF_COLPCT_SUM:
1667 case CTSF_TABLEPCT_SUM:
1668 case CTSF_SUBTABLEPCT_SUM:
1669 case CTSF_LAYERPCT_SUM:
1670 case CTSF_LAYERROWPCT_SUM:
1671 case CTSF_LAYERCOLPCT_SUM:
1672 s->moments = moments1_create (MOMENT_VARIANCE);
1681 case CTSF_RESPONSES:
1682 case CTSF_ROWPCT_RESPONSES:
1683 case CTSF_COLPCT_RESPONSES:
1684 case CTSF_TABLEPCT_RESPONSES:
1685 case CTSF_SUBTABLEPCT_RESPONSES:
1686 case CTSF_LAYERPCT_RESPONSES:
1687 case CTSF_LAYERROWPCT_RESPONSES:
1688 case CTSF_LAYERCOLPCT_RESPONSES:
1689 case CTSF_ROWPCT_RESPONSES_COUNT:
1690 case CTSF_COLPCT_RESPONSES_COUNT:
1691 case CTSF_TABLEPCT_RESPONSES_COUNT:
1692 case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
1693 case CTSF_LAYERPCT_RESPONSES_COUNT:
1694 case CTSF_LAYERROWPCT_RESPONSES_COUNT:
1695 case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
1696 case CTSF_ROWPCT_COUNT_RESPONSES:
1697 case CTSF_COLPCT_COUNT_RESPONSES:
1698 case CTSF_TABLEPCT_COUNT_RESPONSES:
1699 case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
1700 case CTSF_LAYERPCT_COUNT_RESPONSES:
1701 case CTSF_LAYERROWPCT_COUNT_RESPONSES:
1702 case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
1708 ctables_summary_uninit (union ctables_summary *s,
1709 const struct ctables_summary_spec *ss)
1711 switch (ss->function)
1715 case CTSF_ROWPCT_COUNT:
1716 case CTSF_COLPCT_COUNT:
1717 case CTSF_TABLEPCT_COUNT:
1718 case CTSF_SUBTABLEPCT_COUNT:
1719 case CTSF_LAYERPCT_COUNT:
1720 case CTSF_LAYERROWPCT_COUNT:
1721 case CTSF_LAYERCOLPCT_COUNT:
1722 case CTSF_ROWPCT_VALIDN:
1723 case CTSF_COLPCT_VALIDN:
1724 case CTSF_TABLEPCT_VALIDN:
1725 case CTSF_SUBTABLEPCT_VALIDN:
1726 case CTSF_LAYERPCT_VALIDN:
1727 case CTSF_LAYERROWPCT_VALIDN:
1728 case CTSF_LAYERCOLPCT_VALIDN:
1729 case CTSF_ROWPCT_TOTALN:
1730 case CTSF_COLPCT_TOTALN:
1731 case CTSF_TABLEPCT_TOTALN:
1732 case CTSF_SUBTABLEPCT_TOTALN:
1733 case CTSF_LAYERPCT_TOTALN:
1734 case CTSF_LAYERROWPCT_TOTALN:
1735 case CTSF_LAYERCOLPCT_TOTALN:
1752 case CTSF_ROWPCT_SUM:
1753 case CTSF_COLPCT_SUM:
1754 case CTSF_TABLEPCT_SUM:
1755 case CTSF_SUBTABLEPCT_SUM:
1756 case CTSF_LAYERPCT_SUM:
1757 case CTSF_LAYERROWPCT_SUM:
1758 case CTSF_LAYERCOLPCT_SUM:
1759 moments1_destroy (s->moments);
1768 case CTSF_RESPONSES:
1769 case CTSF_ROWPCT_RESPONSES:
1770 case CTSF_COLPCT_RESPONSES:
1771 case CTSF_TABLEPCT_RESPONSES:
1772 case CTSF_SUBTABLEPCT_RESPONSES:
1773 case CTSF_LAYERPCT_RESPONSES:
1774 case CTSF_LAYERROWPCT_RESPONSES:
1775 case CTSF_LAYERCOLPCT_RESPONSES:
1776 case CTSF_ROWPCT_RESPONSES_COUNT:
1777 case CTSF_COLPCT_RESPONSES_COUNT:
1778 case CTSF_TABLEPCT_RESPONSES_COUNT:
1779 case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
1780 case CTSF_LAYERPCT_RESPONSES_COUNT:
1781 case CTSF_LAYERROWPCT_RESPONSES_COUNT:
1782 case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
1783 case CTSF_ROWPCT_COUNT_RESPONSES:
1784 case CTSF_COLPCT_COUNT_RESPONSES:
1785 case CTSF_TABLEPCT_COUNT_RESPONSES:
1786 case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
1787 case CTSF_LAYERPCT_COUNT_RESPONSES:
1788 case CTSF_LAYERROWPCT_COUNT_RESPONSES:
1789 case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
1795 ctables_summary_add (union ctables_summary *s,
1796 const struct ctables_summary_spec *ss,
1797 const struct variable *var, const union value *value,
1800 switch (ss->function)
1804 case CTSF_ROWPCT_COUNT:
1805 case CTSF_COLPCT_COUNT:
1806 case CTSF_TABLEPCT_COUNT:
1807 case CTSF_SUBTABLEPCT_COUNT:
1808 case CTSF_LAYERPCT_COUNT:
1809 case CTSF_LAYERROWPCT_COUNT:
1810 case CTSF_LAYERCOLPCT_COUNT:
1811 case CTSF_ROWPCT_VALIDN:
1812 case CTSF_COLPCT_VALIDN:
1813 case CTSF_TABLEPCT_VALIDN:
1814 case CTSF_SUBTABLEPCT_VALIDN:
1815 case CTSF_LAYERPCT_VALIDN:
1816 case CTSF_LAYERROWPCT_VALIDN:
1817 case CTSF_LAYERCOLPCT_VALIDN:
1818 case CTSF_ROWPCT_TOTALN:
1819 case CTSF_COLPCT_TOTALN:
1820 case CTSF_TABLEPCT_TOTALN:
1821 case CTSF_SUBTABLEPCT_TOTALN:
1822 case CTSF_LAYERPCT_TOTALN:
1823 case CTSF_LAYERROWPCT_TOTALN:
1824 case CTSF_LAYERCOLPCT_TOTALN:
1829 if (var_is_value_missing (var, value))
1830 s->missing += weight;
1838 if (!var_is_value_missing (var, value))
1840 assert (!var_is_alpha (var)); /* XXX? */
1841 if (s->min == SYSMIS || value->f < s->min)
1843 if (s->max == SYSMIS || value->f > s->max)
1853 case CTSF_ROWPCT_SUM:
1854 case CTSF_COLPCT_SUM:
1855 case CTSF_TABLEPCT_SUM:
1856 case CTSF_SUBTABLEPCT_SUM:
1857 case CTSF_LAYERPCT_SUM:
1858 case CTSF_LAYERROWPCT_SUM:
1859 case CTSF_LAYERCOLPCT_SUM:
1860 moments1_add (s->moments, value->f, weight);
1869 case CTSF_RESPONSES:
1870 case CTSF_ROWPCT_RESPONSES:
1871 case CTSF_COLPCT_RESPONSES:
1872 case CTSF_TABLEPCT_RESPONSES:
1873 case CTSF_SUBTABLEPCT_RESPONSES:
1874 case CTSF_LAYERPCT_RESPONSES:
1875 case CTSF_LAYERROWPCT_RESPONSES:
1876 case CTSF_LAYERCOLPCT_RESPONSES:
1877 case CTSF_ROWPCT_RESPONSES_COUNT:
1878 case CTSF_COLPCT_RESPONSES_COUNT:
1879 case CTSF_TABLEPCT_RESPONSES_COUNT:
1880 case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
1881 case CTSF_LAYERPCT_RESPONSES_COUNT:
1882 case CTSF_LAYERROWPCT_RESPONSES_COUNT:
1883 case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
1884 case CTSF_ROWPCT_COUNT_RESPONSES:
1885 case CTSF_COLPCT_COUNT_RESPONSES:
1886 case CTSF_TABLEPCT_COUNT_RESPONSES:
1887 case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
1888 case CTSF_LAYERPCT_COUNT_RESPONSES:
1889 case CTSF_LAYERROWPCT_COUNT_RESPONSES:
1890 case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
1896 ctables_summary_value (const struct ctables_cell *cell,
1897 union ctables_summary *s,
1898 const struct ctables_summary_spec *ss)
1900 switch (ss->function)
1906 case CTSF_SUBTABLEPCT_COUNT:
1907 return cell->domains[CTDT_SUBTABLE]->valid ? s->valid / cell->domains[CTDT_SUBTABLE]->valid * 100 : SYSMIS;
1909 case CTSF_ROWPCT_COUNT:
1910 return cell->domains[CTDT_ROW]->valid ? s->valid / cell->domains[CTDT_ROW]->valid * 100 : SYSMIS;
1912 case CTSF_COLPCT_COUNT:
1913 return cell->domains[CTDT_COL]->valid ? s->valid / cell->domains[CTDT_COL]->valid * 100 : SYSMIS;
1915 case CTSF_TABLEPCT_COUNT:
1916 return cell->domains[CTDT_TABLE]->valid ? s->valid / cell->domains[CTDT_TABLE]->valid * 100 : SYSMIS;
1918 case CTSF_LAYERPCT_COUNT:
1919 return cell->domains[CTDT_LAYER]->valid ? s->valid / cell->domains[CTDT_LAYER]->valid * 100 : SYSMIS;
1921 case CTSF_LAYERROWPCT_COUNT:
1922 return cell->domains[CTDT_LAYERROW]->valid ? s->valid / cell->domains[CTDT_LAYERROW]->valid * 100 : SYSMIS;
1924 case CTSF_LAYERCOLPCT_COUNT:
1925 return cell->domains[CTDT_LAYERCOL]->valid ? s->valid / cell->domains[CTDT_LAYERCOL]->valid * 100 : SYSMIS;
1927 case CTSF_ROWPCT_VALIDN:
1928 case CTSF_COLPCT_VALIDN:
1929 case CTSF_TABLEPCT_VALIDN:
1930 case CTSF_SUBTABLEPCT_VALIDN:
1931 case CTSF_LAYERPCT_VALIDN:
1932 case CTSF_LAYERROWPCT_VALIDN:
1933 case CTSF_LAYERCOLPCT_VALIDN:
1934 case CTSF_ROWPCT_TOTALN:
1935 case CTSF_COLPCT_TOTALN:
1936 case CTSF_TABLEPCT_TOTALN:
1937 case CTSF_SUBTABLEPCT_TOTALN:
1938 case CTSF_LAYERPCT_TOTALN:
1939 case CTSF_LAYERROWPCT_TOTALN:
1940 case CTSF_LAYERCOLPCT_TOTALN:
1945 return s->valid + s->missing;
1958 return s->max != SYSMIS && s->min != SYSMIS ? s->max - s->min : SYSMIS;
1963 moments1_calculate (s->moments, NULL, &mean, NULL, NULL, NULL);
1969 double weight, variance;
1970 moments1_calculate (s->moments, &weight, NULL, &variance, NULL, NULL);
1971 return calc_semean (variance, weight);
1977 moments1_calculate (s->moments, NULL, NULL, &variance, NULL, NULL);
1978 return variance != SYSMIS ? sqrt (variance) : SYSMIS;
1983 double weight, mean;
1984 moments1_calculate (s->moments, &weight, &mean, NULL, NULL, NULL);
1985 return weight != SYSMIS && mean != SYSMIS ? weight * mean : SYSMIS;
1991 moments1_calculate (s->moments, NULL, NULL, &variance, NULL, NULL);
1995 case CTSF_ROWPCT_SUM:
1996 case CTSF_COLPCT_SUM:
1997 case CTSF_TABLEPCT_SUM:
1998 case CTSF_SUBTABLEPCT_SUM:
1999 case CTSF_LAYERPCT_SUM:
2000 case CTSF_LAYERROWPCT_SUM:
2001 case CTSF_LAYERCOLPCT_SUM:
2010 case CTSF_RESPONSES:
2011 case CTSF_ROWPCT_RESPONSES:
2012 case CTSF_COLPCT_RESPONSES:
2013 case CTSF_TABLEPCT_RESPONSES:
2014 case CTSF_SUBTABLEPCT_RESPONSES:
2015 case CTSF_LAYERPCT_RESPONSES:
2016 case CTSF_LAYERROWPCT_RESPONSES:
2017 case CTSF_LAYERCOLPCT_RESPONSES:
2018 case CTSF_ROWPCT_RESPONSES_COUNT:
2019 case CTSF_COLPCT_RESPONSES_COUNT:
2020 case CTSF_TABLEPCT_RESPONSES_COUNT:
2021 case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
2022 case CTSF_LAYERPCT_RESPONSES_COUNT:
2023 case CTSF_LAYERROWPCT_RESPONSES_COUNT:
2024 case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
2025 case CTSF_ROWPCT_COUNT_RESPONSES:
2026 case CTSF_COLPCT_COUNT_RESPONSES:
2027 case CTSF_TABLEPCT_COUNT_RESPONSES:
2028 case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
2029 case CTSF_LAYERPCT_COUNT_RESPONSES:
2030 case CTSF_LAYERROWPCT_COUNT_RESPONSES:
2031 case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
2038 struct ctables_cell_sort_aux
2040 const struct ctables_table *t;
2041 enum pivot_axis_type a;
2045 ctables_cell_compare_3way (const void *a_, const void *b_, const void *aux_)
2047 const struct ctables_cell_sort_aux *aux = aux_;
2048 struct ctables_cell *const *ap = a_;
2049 struct ctables_cell *const *bp = b_;
2050 const struct ctables_cell *a = *ap;
2051 const struct ctables_cell *b = *bp;
2053 size_t a_idx = a->axes[aux->a].stack_idx;
2054 size_t b_idx = b->axes[aux->a].stack_idx;
2056 return a_idx < b_idx ? -1 : 1;
2058 const struct ctables_nest *nest = &aux->t->stacks[aux->a].nests[a_idx];
2059 for (size_t i = 0; i < nest->n; i++)
2060 if (i != nest->scale_idx)
2062 const struct variable *var = nest->vars[i];
2063 const struct ctables_cell_value *a_cv = &a->axes[aux->a].cvs[i];
2064 const struct ctables_cell_value *b_cv = &b->axes[aux->a].cvs[i];
2065 if (a_cv->category != b_cv->category)
2066 return a_cv->category > b_cv->category ? 1 : -1;
2068 const union value *a_val = &a_cv->value;
2069 const union value *b_val = &b_cv->value;
2070 switch (a_cv->category->type)
2077 /* Must be equal. */
2084 int cmp = value_compare_3way (a_val, b_val, var_get_width (var));
2092 int cmp = value_compare_3way (a_val, b_val, var_get_width (var));
2094 return a_cv->category->sort_ascending ? cmp : -cmp;
2100 const char *a_label = var_lookup_value_label (var, a_val);
2101 const char *b_label = var_lookup_value_label (var, b_val);
2103 ? (b_label ? strcmp (a_label, b_label) : 1)
2104 : (b_label ? -1 : value_compare_3way (
2105 a_val, b_val, var_get_width (var))));
2107 return a_cv->category->sort_ascending ? cmp : -cmp;
2121 For each ctables_table:
2122 For each combination of row vars:
2123 For each combination of column vars:
2124 For each combination of layer vars:
2126 Make a table of row values:
2127 Sort entries by row values
2128 Assign a 0-based index to each actual value
2129 Construct a dimension
2130 Make a table of column values
2131 Make a table of layer values
2133 Fill the table entry using the indexes from before.
2136 static struct ctables_domain *
2137 ctables_domain_insert (struct ctables_table *t, struct ctables_cell *cell,
2138 enum ctables_domain_type domain)
2141 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2143 size_t idx = cell->axes[a].stack_idx;
2144 const struct ctables_nest *nest = &t->stacks[a].nests[idx];
2145 hash = hash_int (idx, hash);
2146 for (size_t i = 0; i < nest->n_domains[domain]; i++)
2148 size_t v_idx = nest->domains[domain][i];
2149 hash = value_hash (&cell->axes[a].cvs[v_idx].value,
2150 var_get_width (nest->vars[v_idx]), hash);
2154 struct ctables_domain *d;
2155 HMAP_FOR_EACH_WITH_HASH (d, struct ctables_domain, node, hash, &t->domains[domain])
2157 const struct ctables_cell *df = d->example;
2158 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2160 size_t idx = cell->axes[a].stack_idx;
2161 if (idx != df->axes[a].stack_idx)
2164 const struct ctables_nest *nest = &t->stacks[a].nests[idx];
2165 for (size_t i = 0; i < nest->n_domains[domain]; i++)
2167 size_t v_idx = nest->domains[domain][i];
2168 if (!value_equal (&df->axes[a].cvs[v_idx].value,
2169 &cell->axes[a].cvs[v_idx].value,
2170 var_get_width (nest->vars[v_idx])))
2179 d = xmalloc (sizeof *d);
2180 *d = (struct ctables_domain) { .example = cell };
2181 hmap_insert (&t->domains[domain], &d->node, hash);
2185 static const struct ctables_category *
2186 ctables_categories_match (const struct ctables_categories *c,
2187 const union value *v, const struct variable *var)
2189 const struct ctables_category *othernm = NULL;
2190 for (size_t i = c->n_cats; i-- > 0; )
2192 const struct ctables_category *cat = &c->cats[i];
2196 if (cat->number == v->f)
2204 if ((cat->range[0] == -DBL_MAX || v->f >= cat->range[0])
2205 && (cat->range[1] == DBL_MAX || v->f <= cat->range[1]))
2210 if (var_is_value_missing (var, v))
2227 return (cat->include_missing || !var_is_value_missing (var, v) ? cat
2232 return var_is_value_missing (var, v) ? NULL : othernm;
2235 static const struct ctables_category *
2236 ctables_categories_total (const struct ctables_categories *c)
2238 const struct ctables_category *first = &c->cats[0];
2239 const struct ctables_category *last = &c->cats[c->n_cats - 1];
2240 return (first->type == CCT_TOTAL ? first
2241 : last->type == CCT_TOTAL ? last
2245 static struct ctables_cell *
2246 ctables_cell_insert__ (struct ctables_table *t, const struct ccase *c,
2247 size_t ix[PIVOT_N_AXES],
2248 const struct ctables_category *cats[PIVOT_N_AXES][10])
2250 const struct ctables_nest *ss = &t->stacks[t->summary_axis].nests[ix[t->summary_axis]];
2253 enum ctables_summary_variant sv = CSV_CELL;
2254 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2256 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2257 hash = hash_int (ix[a], hash);
2258 for (size_t i = 0; i < nest->n; i++)
2259 if (i != nest->scale_idx)
2261 hash = hash_pointer (cats[a][i], hash);
2262 if (cats[a][i]->type != CCT_TOTAL
2263 && cats[a][i]->type != CCT_SUBTOTAL
2264 && cats[a][i]->type != CCT_HSUBTOTAL)
2265 hash = value_hash (case_data (c, nest->vars[i]),
2266 var_get_width (nest->vars[i]), hash);
2272 struct ctables_cell *cell;
2273 HMAP_FOR_EACH_WITH_HASH (cell, struct ctables_cell, node, hash, &t->cells)
2275 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2277 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2278 if (cell->axes[a].stack_idx != ix[a])
2280 for (size_t i = 0; i < nest->n; i++)
2281 if (i != nest->scale_idx
2282 && (cats[a][i] != cell->axes[a].cvs[i].category
2283 || (cats[a][i]->type != CCT_TOTAL
2284 && cats[a][i]->type != CCT_SUBTOTAL
2285 && cats[a][i]->type != CCT_HSUBTOTAL
2286 && !value_equal (case_data (c, nest->vars[i]),
2287 &cell->axes[a].cvs[i].value,
2288 var_get_width (nest->vars[i])))))
2297 cell = xmalloc (sizeof *cell);
2300 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2302 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2303 cell->axes[a].stack_idx = ix[a];
2304 cell->axes[a].cvs = (nest->n
2305 ? xnmalloc (nest->n, sizeof *cell->axes[a].cvs)
2307 for (size_t i = 0; i < nest->n; i++)
2309 if (i != nest->scale_idx)
2311 const struct ctables_category *subtotal = cats[a][i]->subtotal;
2312 if (subtotal && subtotal->type == CCT_HSUBTOTAL)
2316 cell->axes[a].cvs[i].category = cats[a][i];
2317 value_clone (&cell->axes[a].cvs[i].value, case_data (c, nest->vars[i]),
2318 var_get_width (nest->vars[i]));
2322 const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv];
2323 cell->summaries = xmalloc (specs->n * sizeof *cell->summaries);
2324 for (size_t i = 0; i < specs->n; i++)
2325 ctables_summary_init (&cell->summaries[i], &specs->specs[i]);
2326 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
2327 cell->domains[dt] = ctables_domain_insert (t, cell, dt);
2328 hmap_insert (&t->cells, &cell->node, hash);
2333 ctables_cell_add__ (struct ctables_table *t, const struct ccase *c,
2334 size_t ix[PIVOT_N_AXES],
2335 const struct ctables_category *cats[PIVOT_N_AXES][10],
2338 struct ctables_cell *cell = ctables_cell_insert__ (t, c, ix, cats);
2339 const struct ctables_nest *ss = &t->stacks[t->summary_axis].nests[ix[t->summary_axis]];
2341 const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv];
2342 for (size_t i = 0; i < specs->n; i++)
2343 ctables_summary_add (&cell->summaries[i], &specs->specs[i], specs->var,
2344 case_data (c, specs->var), weight);
2345 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
2346 cell->domains[dt]->valid += weight;
2350 recurse_totals (struct ctables_table *t, const struct ccase *c,
2351 size_t ix[PIVOT_N_AXES],
2352 const struct ctables_category *cats[PIVOT_N_AXES][10],
2354 enum pivot_axis_type start_axis, size_t start_nest)
2356 for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
2358 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2359 for (size_t i = start_nest; i < nest->n; i++)
2361 if (i == nest->scale_idx)
2364 const struct variable *var = nest->vars[i];
2366 const struct ctables_category *total = ctables_categories_total (
2367 t->categories[var_get_dict_index (var)]);
2370 const struct ctables_category *save = cats[a][i];
2372 ctables_cell_add__ (t, c, ix, cats, weight);
2373 recurse_totals (t, c, ix, cats, weight, a, i + 1);
2382 ctables_cell_insert (struct ctables_table *t,
2383 const struct ccase *c,
2384 size_t ir, size_t ic, size_t il,
2387 size_t ix[PIVOT_N_AXES] = {
2388 [PIVOT_AXIS_ROW] = ir,
2389 [PIVOT_AXIS_COLUMN] = ic,
2390 [PIVOT_AXIS_LAYER] = il,
2393 const struct ctables_category *cats[PIVOT_N_AXES][10];
2394 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2396 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2397 for (size_t i = 0; i < nest->n; i++)
2399 if (i == nest->scale_idx)
2402 const struct variable *var = nest->vars[i];
2403 const union value *value = case_data (c, var);
2405 if (var_is_numeric (var) && value->f == SYSMIS)
2408 cats[a][i] = ctables_categories_match (
2409 t->categories[var_get_dict_index (var)], value, var);
2415 ctables_cell_add__ (t, c, ix, cats, weight);
2417 recurse_totals (t, c, ix, cats, weight, 0, 0);
2419 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2421 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2422 for (size_t i = 0; i < nest->n; i++)
2424 if (i == nest->scale_idx)
2427 const struct ctables_category *save = cats[a][i];
2430 cats[a][i] = save->subtotal;
2431 ctables_cell_add__ (t, c, ix, cats, weight);
2441 const struct ctables_summary_spec_set *set;
2446 merge_item_compare_3way (const struct merge_item *a, const struct merge_item *b)
2448 const struct ctables_summary_spec *as = &a->set->specs[a->ofs];
2449 const struct ctables_summary_spec *bs = &b->set->specs[b->ofs];
2450 if (as->function != bs->function)
2451 return as->function > bs->function ? 1 : -1;
2452 else if (as->percentile != bs->percentile)
2453 return as->percentile < bs->percentile ? 1 : -1;
2454 return strcmp (as->label, bs->label);
2458 ctables_table_output_same_axis (struct ctables *ct, struct ctables_table *t)
2460 struct pivot_table *pt = pivot_table_create__ (
2462 ? pivot_value_new_user_text (t->title, SIZE_MAX)
2463 : pivot_value_new_text (N_("Custom Tables"))),
2466 pivot_table_set_caption (
2467 pt, pivot_value_new_user_text (t->caption, SIZE_MAX));
2469 pivot_table_set_caption (
2470 pt, pivot_value_new_user_text (t->corner, SIZE_MAX));
2472 pivot_table_set_look (pt, ct->look);
2473 struct pivot_dimension *d[PIVOT_N_AXES];
2474 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2476 static const char *names[] = {
2477 [PIVOT_AXIS_ROW] = N_("Rows"),
2478 [PIVOT_AXIS_COLUMN] = N_("Columns"),
2479 [PIVOT_AXIS_LAYER] = N_("Layers"),
2481 d[a] = (t->axes[a] || a == t->summary_axis
2482 ? pivot_dimension_create (pt, a, names[a])
2487 assert (t->axes[a]);
2489 struct ctables_cell **sorted = xnmalloc (t->cells.count, sizeof *sorted);
2491 struct ctables_cell *cell;
2493 HMAP_FOR_EACH (cell, struct ctables_cell, node, &t->cells)
2496 assert (n <= t->cells.count);
2498 struct ctables_cell_sort_aux aux = { .t = t, .a = a };
2499 sort (sorted, n, sizeof *sorted, ctables_cell_compare_3way, &aux);
2501 size_t max_depth = 0;
2502 for (size_t j = 0; j < t->stacks[a].n; j++)
2503 if (t->stacks[a].nests[j].n > max_depth)
2504 max_depth = t->stacks[a].nests[j].n;
2506 struct pivot_category **groups = xnmalloc (max_depth, sizeof *groups);
2507 struct pivot_category *top = NULL;
2509 for (size_t j = 0; j < n; j++)
2511 struct ctables_cell *cell = sorted[j];
2512 const struct ctables_nest *nest = &t->stacks[a].nests[cell->axes[a].stack_idx];
2514 size_t n_common = 0;
2515 bool new_subtable = false;
2518 struct ctables_cell *prev = sorted[j - 1];
2519 if (prev->axes[a].stack_idx == cell->axes[a].stack_idx)
2521 for (; n_common < nest->n; n_common++)
2522 if (n_common != nest->scale_idx
2523 && (prev->axes[a].cvs[n_common].category
2524 != cell->axes[a].cvs[n_common].category
2525 || !value_equal (&prev->axes[a].cvs[n_common].value,
2526 &cell->axes[a].cvs[n_common].value,
2527 var_get_type (nest->vars[n_common]))))
2531 new_subtable = true;
2534 new_subtable = true;
2538 enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (nest->vars[0])];
2540 if (vlabel != CTVL_NONE)
2541 top = pivot_category_create_group__ (
2542 top, pivot_value_new_variable (nest->vars[0]));
2544 if (n_common == nest->n)
2546 cell->axes[a].leaf = prev_leaf;
2550 for (size_t k = n_common; k < nest->n; k++)
2552 struct pivot_category *parent = k > 0 ? groups[k - 1] : top;
2554 struct pivot_value *label
2555 = (k == nest->scale_idx ? NULL
2556 : (cell->axes[a].cvs[k].category->type == CCT_TOTAL
2557 || cell->axes[a].cvs[k].category->type == CCT_SUBTOTAL
2558 || cell->axes[a].cvs[k].category->type == CCT_HSUBTOTAL)
2559 ? pivot_value_new_user_text (cell->axes[a].cvs[k].category->total_label,
2561 : pivot_value_new_var_value (nest->vars[k],
2562 &cell->axes[a].cvs[k].value));
2563 if (k == nest->n - 1)
2565 if (a == t->summary_axis)
2568 parent = pivot_category_create_group__ (parent, label);
2569 const struct ctables_summary_spec_set *specs = &nest->specs[cell->sv];
2570 for (size_t m = 0; m < specs->n; m++)
2572 int leaf = pivot_category_create_leaf (
2573 parent, pivot_value_new_text (specs->specs[m].label));
2580 /* This assertion is true as long as the summary axis
2581 is the axis where the summaries are displayed. */
2584 prev_leaf = pivot_category_create_leaf (parent, label);
2590 parent = pivot_category_create_group__ (parent, label);
2592 enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (nest->vars[k + 1])];
2593 if (vlabel != CTVL_NONE)
2594 parent = pivot_category_create_group__ (
2595 parent, pivot_value_new_variable (nest->vars[k + 1]));
2599 cell->axes[a].leaf = prev_leaf;
2604 struct ctables_cell *cell;
2605 HMAP_FOR_EACH (cell, struct ctables_cell, node, &t->cells)
2610 const struct ctables_nest *nest = &t->stacks[t->summary_axis].nests[cell->axes[t->summary_axis].stack_idx];
2611 const struct ctables_summary_spec_set *specs = &nest->specs[cell->sv];
2612 for (size_t j = 0; j < specs->n; j++)
2615 size_t n_dindexes = 0;
2617 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2620 int leaf = cell->axes[a].leaf;
2621 if (a == t->summary_axis)
2623 dindexes[n_dindexes++] = leaf;
2626 double d = ctables_summary_value (cell, &cell->summaries[j], &specs->specs[j]);
2627 struct pivot_value *value = pivot_value_new_number (d);
2628 value->numeric.format = specs->specs[j].format;
2629 pivot_table_put (pt, dindexes, n_dindexes, value);
2633 pivot_table_submit (pt);
2638 ctables_table_output_different_axis (struct ctables *ct, struct ctables_table *t)
2640 struct pivot_table *pt = pivot_table_create__ (
2642 ? pivot_value_new_user_text (t->title, SIZE_MAX)
2643 : pivot_value_new_text (N_("Custom Tables"))),
2646 pivot_table_set_caption (
2647 pt, pivot_value_new_user_text (t->caption, SIZE_MAX));
2649 pivot_table_set_caption (
2650 pt, pivot_value_new_user_text (t->corner, SIZE_MAX));
2652 pivot_table_set_look (pt, ct->look);
2653 struct pivot_dimension *d[PIVOT_N_AXES];
2654 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2656 static const char *names[] = {
2657 [PIVOT_AXIS_ROW] = N_("Rows"),
2658 [PIVOT_AXIS_COLUMN] = N_("Columns"),
2659 [PIVOT_AXIS_LAYER] = N_("Layers"),
2661 d[a] = (t->axes[a] || a == t->summary_axis
2662 ? pivot_dimension_create (pt, a, names[a])
2667 assert (t->axes[a]);
2669 struct ctables_cell **sorted = xnmalloc (t->cells.count, sizeof *sorted);
2671 struct ctables_cell *cell;
2673 HMAP_FOR_EACH (cell, struct ctables_cell, node, &t->cells)
2676 assert (n <= t->cells.count);
2678 struct ctables_cell_sort_aux aux = { .t = t, .a = a };
2679 sort (sorted, n, sizeof *sorted, ctables_cell_compare_3way, &aux);
2681 size_t max_depth = 0;
2682 for (size_t j = 0; j < t->stacks[a].n; j++)
2683 if (t->stacks[a].nests[j].n > max_depth)
2684 max_depth = t->stacks[a].nests[j].n;
2686 struct pivot_category **groups = xnmalloc (max_depth, sizeof *groups);
2687 struct pivot_category *top = NULL;
2689 for (size_t j = 0; j < n; j++)
2691 struct ctables_cell *cell = sorted[j];
2692 const struct ctables_nest *nest = &t->stacks[a].nests[cell->axes[a].stack_idx];
2694 size_t n_common = 0;
2695 bool new_subtable = false;
2698 struct ctables_cell *prev = sorted[j - 1];
2699 if (prev->axes[a].stack_idx == cell->axes[a].stack_idx)
2701 for (; n_common < nest->n; n_common++)
2702 if (n_common != nest->scale_idx
2703 && (prev->axes[a].cvs[n_common].category
2704 != cell->axes[a].cvs[n_common].category
2705 || !value_equal (&prev->axes[a].cvs[n_common].value,
2706 &cell->axes[a].cvs[n_common].value,
2707 var_get_type (nest->vars[n_common]))))
2711 new_subtable = true;
2714 new_subtable = true;
2718 enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (nest->vars[0])];
2720 if (vlabel != CTVL_NONE)
2721 top = pivot_category_create_group__ (
2722 top, pivot_value_new_variable (nest->vars[0]));
2724 if (n_common == nest->n)
2726 cell->axes[a].leaf = prev_leaf;
2730 for (size_t k = n_common; k < nest->n; k++)
2732 struct pivot_category *parent = k > 0 ? groups[k - 1] : top;
2734 struct pivot_value *label
2735 = (k == nest->scale_idx ? NULL
2736 : (cell->axes[a].cvs[k].category->type == CCT_TOTAL
2737 || cell->axes[a].cvs[k].category->type == CCT_SUBTOTAL
2738 || cell->axes[a].cvs[k].category->type == CCT_HSUBTOTAL)
2739 ? pivot_value_new_user_text (cell->axes[a].cvs[k].category->total_label,
2741 : pivot_value_new_var_value (nest->vars[k],
2742 &cell->axes[a].cvs[k].value));
2743 if (k == nest->n - 1)
2745 if (a == t->summary_axis)
2748 parent = pivot_category_create_group__ (parent, label);
2749 const struct ctables_summary_spec_set *specs = &nest->specs[cell->sv];
2750 for (size_t m = 0; m < specs->n; m++)
2752 int leaf = pivot_category_create_leaf (
2753 parent, pivot_value_new_text (specs->specs[m].label));
2760 /* This assertion is true as long as the summary axis
2761 is the axis where the summaries are displayed. */
2764 prev_leaf = pivot_category_create_leaf (parent, label);
2770 parent = pivot_category_create_group__ (parent, label);
2772 enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (nest->vars[k + 1])];
2773 if (vlabel != CTVL_NONE)
2774 parent = pivot_category_create_group__ (
2775 parent, pivot_value_new_variable (nest->vars[k + 1]));
2779 cell->axes[a].leaf = prev_leaf;
2784 struct ctables_cell *cell;
2785 HMAP_FOR_EACH (cell, struct ctables_cell, node, &t->cells)
2790 const struct ctables_nest *nest = &t->stacks[t->summary_axis].nests[cell->axes[t->summary_axis].stack_idx];
2791 const struct ctables_summary_spec_set *specs = &nest->specs[cell->sv];
2792 for (size_t j = 0; j < specs->n; j++)
2795 size_t n_dindexes = 0;
2797 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2800 int leaf = cell->axes[a].leaf;
2801 if (a == t->summary_axis)
2803 dindexes[n_dindexes++] = leaf;
2806 double d = ctables_summary_value (cell, &cell->summaries[j], &specs->specs[j]);
2807 struct pivot_value *value = pivot_value_new_number (d);
2808 value->numeric.format = specs->specs[j].format;
2809 pivot_table_put (pt, dindexes, n_dindexes, value);
2813 pivot_table_submit (pt);
2818 ctables_execute (struct dataset *ds, struct ctables *ct)
2820 for (size_t i = 0; i < ct->n_tables; i++)
2822 struct ctables_table *t = ct->tables[i];
2823 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2826 t->stacks[a] = enumerate_fts (a, t->axes[a]);
2828 for (size_t j = 0; j < t->stacks[a].n; j++)
2830 struct ctables_nest *nest = &t->stacks[a].nests[j];
2831 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
2833 nest->domains[dt] = xmalloc (nest->n * sizeof *nest->domains[dt]);
2834 nest->n_domains[dt] = 0;
2836 for (size_t k = 0; k < nest->n; k++)
2838 if (k == nest->scale_idx)
2847 if (a != PIVOT_AXIS_LAYER)
2854 if (dt == CTDT_SUBTABLE ? a != PIVOT_AXIS_LAYER
2855 : dt == CTDT_ROW ? a == PIVOT_AXIS_COLUMN
2856 : a == PIVOT_AXIS_ROW)
2858 if (k == nest->n - 1
2859 || (nest->scale_idx == nest->n - 1
2860 && k == nest->n - 2))
2866 if (a == PIVOT_AXIS_COLUMN)
2871 if (a == PIVOT_AXIS_ROW)
2876 nest->domains[dt][nest->n_domains[dt]++] = k;
2883 struct ctables_nest *nest = xmalloc (sizeof *nest);
2884 *nest = (struct ctables_nest) { .n = 0 };
2885 t->stacks[a] = (struct ctables_stack) { .nests = nest, .n = 1 };
2888 struct ctables_stack *stack = &t->stacks[t->summary_axis];
2889 for (size_t i = 0; i < stack->n; i++)
2891 struct ctables_nest *nest = &stack->nests[i];
2892 if (!nest->specs[CSV_CELL].n)
2894 struct ctables_summary_spec_set *specs = &nest->specs[CSV_CELL];
2895 specs->specs = xmalloc (sizeof *specs->specs);
2898 enum ctables_summary_function function
2899 = specs->var ? CTSF_MEAN : CTSF_COUNT;
2900 struct ctables_var var = { .is_mrset = false, .var = specs->var };
2902 *specs->specs = (struct ctables_summary_spec) {
2903 .function = function,
2904 .format = ctables_summary_default_format (function, &var),
2905 .label = ctables_summary_default_label (function, 0),
2908 specs->var = nest->vars[0];
2910 ctables_summary_spec_set_clone (&nest->specs[CSV_TOTAL],
2911 &nest->specs[CSV_CELL]);
2913 else if (!nest->specs[CSV_TOTAL].n)
2914 ctables_summary_spec_set_clone (&nest->specs[CSV_TOTAL],
2915 &nest->specs[CSV_CELL]);
2918 struct ctables_summary_spec_set *merged = &t->summary_specs;
2919 struct merge_item *items = xnmalloc (2 * stack->n, sizeof *items);
2921 for (size_t j = 0; j < stack->n; j++)
2923 const struct ctables_nest *nest = &stack->nests[j];
2927 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
2929 items[n_left] = (struct merge_item) {
2930 .tiebreaker = n_left,
2931 .set = &nest->specs[sv]
2939 struct merge_item min = items[0];
2940 for (size_t j = 1; j < n_left; j++)
2941 if (merge_item_compare_3way (&items[j], &min) < 0)
2944 /* XXX Add to 'merged' */
2945 if (merged->n >= merged->allocated)
2946 merged->specs = x2nrealloc (merged->specs, &merged->allocated,
2947 sizeof *merged->specs);
2948 merged->specs[merged->n++] = min.set->specs[min.ofs];
2950 for (size_t j = 0; j < n_left; )
2952 if (merge_item_compare_3way (&items[j], &min) == 0)
2954 struct merge_item *item = &items[j];
2955 item->set->specs[item->ofs].axis_idx = merged->n - 1;
2956 if (++item->ofs >= item->set->n)
2958 items[j] = items[--n_left];
2966 for (size_t j = 0; j < merged->n; j++)
2967 printf ("%s\n", ctables_summary_function_name (merged->specs[j].function));
2969 for (size_t j = 0; j < stack->n; j++)
2971 const struct ctables_nest *nest = &stack->nests[j];
2972 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
2974 const struct ctables_summary_spec_set *specs = &nest->specs[sv];
2975 for (size_t k = 0; k < specs->n; k++)
2976 printf ("(%s, %zu) ", ctables_summary_function_name (specs->specs[k].function),
2977 specs->specs[k].axis_idx);
2983 struct casereader *input = casereader_create_filter_weight (proc_open (ds),
2986 bool warn_on_invalid = true;
2987 double total_weight = 0;
2988 for (struct ccase *c = casereader_read (input); c;
2989 case_unref (c), c = casereader_read (input))
2991 double weight = dict_get_case_weight (dataset_dict (ds), c,
2993 total_weight += weight;
2995 for (size_t i = 0; i < ct->n_tables; i++)
2997 struct ctables_table *t = ct->tables[i];
2999 for (size_t ir = 0; ir < t->stacks[PIVOT_AXIS_ROW].n; ir++)
3000 for (size_t ic = 0; ic < t->stacks[PIVOT_AXIS_COLUMN].n; ic++)
3001 for (size_t il = 0; il < t->stacks[PIVOT_AXIS_LAYER].n; il++)
3002 ctables_cell_insert (t, c, ir, ic, il, weight);
3005 casereader_destroy (input);
3007 for (size_t i = 0; i < ct->n_tables; i++)
3009 struct ctables_table *t = ct->tables[i];
3010 if (t->summary_axis == t->slabels_position)
3011 ctables_table_output_same_axis (ct, ct->tables[i]);
3013 ctables_table_output_different_axis (ct, ct->tables[i]);
3015 return proc_commit (ds);
3019 cmd_ctables (struct lexer *lexer, struct dataset *ds)
3021 size_t n_vars = dict_get_n_vars (dataset_dict (ds));
3022 enum ctables_vlabel *vlabels = xnmalloc (n_vars, sizeof *vlabels);
3023 enum settings_value_show tvars = settings_get_show_variables ();
3024 for (size_t i = 0; i < n_vars; i++)
3025 vlabels[i] = (enum ctables_vlabel) tvars;
3027 struct ctables *ct = xmalloc (sizeof *ct);
3028 *ct = (struct ctables) {
3029 .look = pivot_table_look_unshare (pivot_table_look_ref (
3030 pivot_table_look_get_default ())),
3032 .hide_threshold = 5,
3034 ct->look->omit_empty = false;
3036 if (!lex_force_match (lexer, T_SLASH))
3039 while (!lex_match_id (lexer, "TABLE"))
3041 if (lex_match_id (lexer, "FORMAT"))
3043 double widths[2] = { SYSMIS, SYSMIS };
3044 double units_per_inch = 72.0;
3046 while (lex_token (lexer) != T_SLASH)
3048 if (lex_match_id (lexer, "MINCOLWIDTH"))
3050 if (!parse_col_width (lexer, "MINCOLWIDTH", &widths[0]))
3053 else if (lex_match_id (lexer, "MAXCOLWIDTH"))
3055 if (!parse_col_width (lexer, "MAXCOLWIDTH", &widths[1]))
3058 else if (lex_match_id (lexer, "UNITS"))
3060 lex_match (lexer, T_EQUALS);
3061 if (lex_match_id (lexer, "POINTS"))
3062 units_per_inch = 72.0;
3063 else if (lex_match_id (lexer, "INCHES"))
3064 units_per_inch = 1.0;
3065 else if (lex_match_id (lexer, "CM"))
3066 units_per_inch = 2.54;
3069 lex_error_expecting (lexer, "POINTS", "INCHES", "CM");
3073 else if (lex_match_id (lexer, "EMPTY"))
3078 lex_match (lexer, T_EQUALS);
3079 if (lex_match_id (lexer, "ZERO"))
3081 /* Nothing to do. */
3083 else if (lex_match_id (lexer, "BLANK"))
3084 ct->zero = xstrdup ("");
3085 else if (lex_force_string (lexer))
3087 ct->zero = ss_xstrdup (lex_tokss (lexer));
3093 else if (lex_match_id (lexer, "MISSING"))
3095 lex_match (lexer, T_EQUALS);
3096 if (!lex_force_string (lexer))
3100 ct->missing = (strcmp (lex_tokcstr (lexer), ".")
3101 ? ss_xstrdup (lex_tokss (lexer))
3107 lex_error_expecting (lexer, "MINCOLWIDTH", "MAXCOLWIDTH",
3108 "UNITS", "EMPTY", "MISSING");
3113 if (widths[0] != SYSMIS && widths[1] != SYSMIS
3114 && widths[0] > widths[1])
3116 msg (SE, _("MINCOLWIDTH must not be greater than MAXCOLWIDTH."));
3120 for (size_t i = 0; i < 2; i++)
3121 if (widths[i] != SYSMIS)
3123 int *wr = ct->look->width_ranges[TABLE_HORZ];
3124 wr[i] = widths[i] / units_per_inch * 96.0;
3129 else if (lex_match_id (lexer, "VLABELS"))
3131 if (!lex_force_match_id (lexer, "VARIABLES"))
3133 lex_match (lexer, T_EQUALS);
3135 struct variable **vars;
3137 if (!parse_variables (lexer, dataset_dict (ds), &vars, &n_vars,
3141 if (!lex_force_match_id (lexer, "DISPLAY"))
3146 lex_match (lexer, T_EQUALS);
3148 enum ctables_vlabel vlabel;
3149 if (lex_match_id (lexer, "DEFAULT"))
3150 vlabel = (enum ctables_vlabel) settings_get_show_variables ();
3151 else if (lex_match_id (lexer, "NAME"))
3153 else if (lex_match_id (lexer, "LABEL"))
3154 vlabel = CTVL_LABEL;
3155 else if (lex_match_id (lexer, "BOTH"))
3157 else if (lex_match_id (lexer, "NONE"))
3161 lex_error_expecting (lexer, "DEFAULT", "NAME", "LABEL",
3167 for (size_t i = 0; i < n_vars; i++)
3168 ct->vlabels[var_get_dict_index (vars[i])] = vlabel;
3171 else if (lex_match_id (lexer, "MRSETS"))
3173 if (!lex_force_match_id (lexer, "COUNTDUPLICATES"))
3175 lex_match (lexer, T_EQUALS);
3176 if (!parse_bool (lexer, &ct->mrsets_count_duplicates))
3179 else if (lex_match_id (lexer, "SMISSING"))
3181 if (lex_match_id (lexer, "VARIABLE"))
3182 ct->smissing_listwise = false;
3183 else if (lex_match_id (lexer, "LISTWISE"))
3184 ct->smissing_listwise = true;
3187 lex_error_expecting (lexer, "VARIABLE", "LISTWISE");
3192 else if (lex_match_id (lexer, "WEIGHT"))
3194 if (!lex_force_match_id (lexer, "VARIABLE"))
3196 lex_match (lexer, T_EQUALS);
3197 ct->base_weight = parse_variable (lexer, dataset_dict (ds));
3198 if (!ct->base_weight)
3201 else if (lex_match_id (lexer, "HIDESMALLCOUNTS"))
3203 if (!lex_force_match_id (lexer, "COUNT"))
3205 lex_match (lexer, T_EQUALS);
3206 if (!lex_force_int_range (lexer, "HIDESMALLCOUNTS COUNT", 2, INT_MAX))
3208 ct->hide_threshold = lex_integer (lexer);
3213 lex_error_expecting (lexer, "FORMAT", "VLABELS", "MRSETS",
3214 "SMISSING", "PCOMPUTE", "PPROPERTIES",
3215 "WEIGHT", "HIDESMALLCOUNTS", "TABLE");
3219 if (!lex_force_match (lexer, T_SLASH))
3223 size_t allocated_tables = 0;
3226 if (ct->n_tables >= allocated_tables)
3227 ct->tables = x2nrealloc (ct->tables, &allocated_tables,
3228 sizeof *ct->tables);
3230 struct ctables_category *cat = xmalloc (sizeof *cat);
3231 *cat = (struct ctables_category) {
3233 .include_missing = false,
3234 .sort_ascending = true,
3237 struct ctables_categories *c = xmalloc (sizeof *c);
3238 size_t n_vars = dict_get_n_vars (dataset_dict (ds));
3239 *c = (struct ctables_categories) {
3245 struct ctables_categories **categories = xnmalloc (n_vars,
3246 sizeof *categories);
3247 for (size_t i = 0; i < n_vars; i++)
3250 struct ctables_table *t = xmalloc (sizeof *t);
3251 *t = (struct ctables_table) {
3252 .cells = HMAP_INITIALIZER (t->cells),
3253 .slabels_position = PIVOT_AXIS_COLUMN,
3254 .slabels_visible = true,
3255 .row_labels = CTLP_NORMAL,
3256 .col_labels = CTLP_NORMAL,
3257 .categories = categories,
3258 .n_categories = n_vars,
3261 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
3262 hmap_init (&t->domains[dt]);
3263 ct->tables[ct->n_tables++] = t;
3265 lex_match (lexer, T_EQUALS);
3266 if (!ctables_axis_parse (lexer, dataset_dict (ds), ct, t, PIVOT_AXIS_ROW))
3268 if (lex_match (lexer, T_BY))
3270 if (!ctables_axis_parse (lexer, dataset_dict (ds),
3271 ct, t, PIVOT_AXIS_COLUMN))
3274 if (lex_match (lexer, T_BY))
3276 if (!ctables_axis_parse (lexer, dataset_dict (ds),
3277 ct, t, PIVOT_AXIS_LAYER))
3282 if (!t->axes[PIVOT_AXIS_ROW] && !t->axes[PIVOT_AXIS_COLUMN]
3283 && !t->axes[PIVOT_AXIS_LAYER])
3285 lex_error (lexer, _("At least one variable must be specified."));
3289 const struct ctables_axis *scales[PIVOT_N_AXES];
3290 size_t n_scales = 0;
3291 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3293 scales[a] = find_scale (t->axes[a]);
3299 msg (SE, _("Scale variables may appear only on one axis."));
3300 if (scales[PIVOT_AXIS_ROW])
3301 msg_at (SN, scales[PIVOT_AXIS_ROW]->loc,
3302 _("This scale variable appears on the rows axis."));
3303 if (scales[PIVOT_AXIS_COLUMN])
3304 msg_at (SN, scales[PIVOT_AXIS_COLUMN]->loc,
3305 _("This scale variable appears on the columns axis."));
3306 if (scales[PIVOT_AXIS_LAYER])
3307 msg_at (SN, scales[PIVOT_AXIS_LAYER]->loc,
3308 _("This scale variable appears on the layer axis."));
3312 const struct ctables_axis *summaries[PIVOT_N_AXES];
3313 size_t n_summaries = 0;
3314 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3316 summaries[a] = (scales[a]
3318 : find_categorical_summary_spec (t->axes[a]));
3322 if (n_summaries > 1)
3324 msg (SE, _("Summaries may appear only on one axis."));
3325 if (summaries[PIVOT_AXIS_ROW])
3326 msg_at (SN, summaries[PIVOT_AXIS_ROW]->loc,
3327 _("This variable on the rows axis has a summary."));
3328 if (summaries[PIVOT_AXIS_COLUMN])
3329 msg_at (SN, summaries[PIVOT_AXIS_COLUMN]->loc,
3330 _("This variable on the columns axis has a summary."));
3331 if (summaries[PIVOT_AXIS_LAYER])
3332 msg_at (SN, summaries[PIVOT_AXIS_LAYER]->loc,
3333 _("This variable on the layers axis has a summary."));
3336 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3337 if (n_summaries ? summaries[a] : t->axes[a])
3339 t->summary_axis = a;
3343 if (lex_token (lexer) == T_ENDCMD)
3345 if (!lex_force_match (lexer, T_SLASH))
3348 while (!lex_match_id (lexer, "TABLE") && lex_token (lexer) != T_ENDCMD)
3350 if (lex_match_id (lexer, "SLABELS"))
3352 while (lex_token (lexer) != T_SLASH)
3354 if (lex_match_id (lexer, "POSITION"))
3356 lex_match (lexer, T_EQUALS);
3357 if (lex_match_id (lexer, "COLUMN"))
3358 t->slabels_position = PIVOT_AXIS_COLUMN;
3359 else if (lex_match_id (lexer, "ROW"))
3360 t->slabels_position = PIVOT_AXIS_ROW;
3361 else if (lex_match_id (lexer, "LAYER"))
3362 t->slabels_position = PIVOT_AXIS_LAYER;
3365 lex_error_expecting (lexer, "COLUMN", "ROW", "LAYER");
3369 else if (lex_match_id (lexer, "VISIBLE"))
3371 lex_match (lexer, T_EQUALS);
3372 if (!parse_bool (lexer, &t->slabels_visible))
3377 lex_error_expecting (lexer, "POSITION", "VISIBLE");
3382 else if (lex_match_id (lexer, "CLABELS"))
3384 while (lex_token (lexer) != T_SLASH)
3386 if (lex_match_id (lexer, "AUTO"))
3387 t->row_labels = t->col_labels = CTLP_NORMAL;
3388 else if (lex_match_id (lexer, "ROWLABELS"))
3390 lex_match (lexer, T_EQUALS);
3391 if (lex_match_id (lexer, "OPPOSITE"))
3392 t->row_labels = CTLP_OPPOSITE;
3393 else if (lex_match_id (lexer, "LAYER"))
3394 t->row_labels = CTLP_LAYER;
3397 lex_error_expecting (lexer, "OPPOSITE", "LAYER");
3401 else if (lex_match_id (lexer, "COLLABELS"))
3403 lex_match (lexer, T_EQUALS);
3404 if (lex_match_id (lexer, "OPPOSITE"))
3405 t->col_labels = CTLP_OPPOSITE;
3406 else if (lex_match_id (lexer, "LAYER"))
3407 t->col_labels = CTLP_LAYER;
3410 lex_error_expecting (lexer, "OPPOSITE", "LAYER");
3416 lex_error_expecting (lexer, "AUTO", "ROWLABELS",
3422 else if (lex_match_id (lexer, "CRITERIA"))
3424 if (!lex_force_match_id (lexer, "CILEVEL"))
3426 lex_match (lexer, T_EQUALS);
3428 if (!lex_force_num_range_halfopen (lexer, "CILEVEL", 0, 100))
3430 t->cilevel = lex_number (lexer);
3433 else if (lex_match_id (lexer, "CATEGORIES"))
3435 if (!ctables_table_parse_categories (lexer, dataset_dict (ds), t))
3438 else if (lex_match_id (lexer, "TITLES"))
3443 if (lex_match_id (lexer, "CAPTION"))
3444 textp = &t->caption;
3445 else if (lex_match_id (lexer, "CORNER"))
3447 else if (lex_match_id (lexer, "TITLE"))
3451 lex_error_expecting (lexer, "CAPTION", "CORNER", "TITLE");
3454 lex_match (lexer, T_EQUALS);
3456 struct string s = DS_EMPTY_INITIALIZER;
3457 while (lex_is_string (lexer))
3459 if (!ds_is_empty (&s))
3460 ds_put_byte (&s, ' ');
3461 ds_put_substring (&s, lex_tokss (lexer));
3465 *textp = ds_steal_cstr (&s);
3467 while (lex_token (lexer) != T_SLASH
3468 && lex_token (lexer) != T_ENDCMD);
3470 else if (lex_match_id (lexer, "SIGTEST"))
3474 t->chisq = xmalloc (sizeof *t->chisq);
3475 *t->chisq = (struct ctables_chisq) {
3477 .include_mrsets = true,
3478 .all_visible = true,
3484 if (lex_match_id (lexer, "TYPE"))
3486 lex_match (lexer, T_EQUALS);
3487 if (!lex_force_match_id (lexer, "CHISQUARE"))
3490 else if (lex_match_id (lexer, "ALPHA"))
3492 lex_match (lexer, T_EQUALS);
3493 if (!lex_force_num_range_halfopen (lexer, "ALPHA", 0, 1))
3495 t->chisq->alpha = lex_number (lexer);
3498 else if (lex_match_id (lexer, "INCLUDEMRSETS"))
3500 lex_match (lexer, T_EQUALS);
3501 if (parse_bool (lexer, &t->chisq->include_mrsets))
3504 else if (lex_match_id (lexer, "CATEGORIES"))
3506 lex_match (lexer, T_EQUALS);
3507 if (lex_match_id (lexer, "ALLVISIBLE"))
3508 t->chisq->all_visible = true;
3509 else if (lex_match_id (lexer, "SUBTOTALS"))
3510 t->chisq->all_visible = false;
3513 lex_error_expecting (lexer,
3514 "ALLVISIBLE", "SUBTOTALS");
3520 lex_error_expecting (lexer, "TYPE", "ALPHA",
3521 "INCLUDEMRSETS", "CATEGORIES");
3525 while (lex_token (lexer) != T_SLASH
3526 && lex_token (lexer) != T_ENDCMD);
3528 else if (lex_match_id (lexer, "COMPARETEST"))
3532 t->pairwise = xmalloc (sizeof *t->pairwise);
3533 *t->pairwise = (struct ctables_pairwise) {
3535 .alpha = { .05, .05 },
3536 .adjust = BONFERRONI,
3537 .include_mrsets = true,
3538 .meansvariance_allcats = true,
3539 .all_visible = true,
3548 if (lex_match_id (lexer, "TYPE"))
3550 lex_match (lexer, T_EQUALS);
3551 if (lex_match_id (lexer, "PROP"))
3552 t->pairwise->type = PROP;
3553 else if (lex_match_id (lexer, "MEAN"))
3554 t->pairwise->type = MEAN;
3557 lex_error_expecting (lexer, "PROP", "MEAN");
3561 else if (lex_match_id (lexer, "ALPHA"))
3563 lex_match (lexer, T_EQUALS);
3565 if (!lex_force_num_range_open (lexer, "ALPHA", 0, 1))
3567 double a0 = lex_number (lexer);
3570 lex_match (lexer, T_COMMA);
3571 if (lex_is_number (lexer))
3573 if (!lex_force_num_range_open (lexer, "ALPHA", 0, 1))
3575 double a1 = lex_number (lexer);
3578 t->pairwise->alpha[0] = MIN (a0, a1);
3579 t->pairwise->alpha[1] = MAX (a0, a1);
3582 t->pairwise->alpha[0] = t->pairwise->alpha[1] = a0;
3584 else if (lex_match_id (lexer, "ADJUST"))
3586 lex_match (lexer, T_EQUALS);
3587 if (lex_match_id (lexer, "BONFERRONI"))
3588 t->pairwise->adjust = BONFERRONI;
3589 else if (lex_match_id (lexer, "BH"))
3590 t->pairwise->adjust = BH;
3591 else if (lex_match_id (lexer, "NONE"))
3592 t->pairwise->adjust = 0;
3595 lex_error_expecting (lexer, "BONFERRONI", "BH",
3600 else if (lex_match_id (lexer, "INCLUDEMRSETS"))
3602 lex_match (lexer, T_EQUALS);
3603 if (!parse_bool (lexer, &t->pairwise->include_mrsets))
3606 else if (lex_match_id (lexer, "MEANSVARIANCE"))
3608 lex_match (lexer, T_EQUALS);
3609 if (lex_match_id (lexer, "ALLCATS"))
3610 t->pairwise->meansvariance_allcats = true;
3611 else if (lex_match_id (lexer, "TESTEDCATS"))
3612 t->pairwise->meansvariance_allcats = false;
3615 lex_error_expecting (lexer, "ALLCATS", "TESTEDCATS");
3619 else if (lex_match_id (lexer, "CATEGORIES"))
3621 lex_match (lexer, T_EQUALS);
3622 if (lex_match_id (lexer, "ALLVISIBLE"))
3623 t->pairwise->all_visible = true;
3624 else if (lex_match_id (lexer, "SUBTOTALS"))
3625 t->pairwise->all_visible = false;
3628 lex_error_expecting (lexer, "ALLVISIBLE",
3633 else if (lex_match_id (lexer, "MERGE"))
3635 lex_match (lexer, T_EQUALS);
3636 if (!parse_bool (lexer, &t->pairwise->merge))
3639 else if (lex_match_id (lexer, "STYLE"))
3641 lex_match (lexer, T_EQUALS);
3642 if (lex_match_id (lexer, "APA"))
3643 t->pairwise->apa_style = true;
3644 else if (lex_match_id (lexer, "SIMPLE"))
3645 t->pairwise->apa_style = false;
3648 lex_error_expecting (lexer, "APA", "SIMPLE");
3652 else if (lex_match_id (lexer, "SHOWSIG"))
3654 lex_match (lexer, T_EQUALS);
3655 if (!parse_bool (lexer, &t->pairwise->show_sig))
3660 lex_error_expecting (lexer, "TYPE", "ALPHA", "ADJUST",
3661 "INCLUDEMRSETS", "MEANSVARIANCE",
3662 "CATEGORIES", "MERGE", "STYLE",
3667 while (lex_token (lexer) != T_SLASH
3668 && lex_token (lexer) != T_ENDCMD);
3672 lex_error_expecting (lexer, "TABLE", "SLABELS", "CLABELS",
3673 "CRITERIA", "CATEGORIES", "TITLES",
3674 "SIGTEST", "COMPARETEST");
3678 if (!lex_match (lexer, T_SLASH))
3682 if (t->row_labels != CTLP_NORMAL && t->col_labels != CTLP_NORMAL)
3684 msg (SE, _("ROWLABELS and COLLABELS may not both be specified."));
3689 while (lex_token (lexer) != T_ENDCMD);
3691 bool ok = ctables_execute (ds, ct);
3692 ctables_destroy (ct);
3693 return ok ? CMD_SUCCESS : CMD_FAILURE;
3696 ctables_destroy (ct);