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 "data/value-labels.h"
26 #include "language/command.h"
27 #include "language/lexer/format-parser.h"
28 #include "language/lexer/lexer.h"
29 #include "language/lexer/variable-parser.h"
30 #include "libpspp/array.h"
31 #include "libpspp/assertion.h"
32 #include "libpspp/hash-functions.h"
33 #include "libpspp/hmap.h"
34 #include "libpspp/message.h"
35 #include "libpspp/string-array.h"
36 #include "math/moments.h"
37 #include "output/pivot-table.h"
39 #include "gl/minmax.h"
40 #include "gl/xalloc.h"
43 #define _(msgid) gettext (msgid)
44 #define N_(msgid) (msgid)
48 CTVL_NONE = SETTINGS_VALUE_SHOW_DEFAULT,
49 CTVL_NAME = SETTINGS_VALUE_SHOW_VALUE,
50 CTVL_LABEL = SETTINGS_VALUE_SHOW_LABEL,
51 CTVL_BOTH = SETTINGS_VALUE_SHOW_BOTH,
55 - unweighted summaries (U*)
56 - lower confidence limits (*.LCL)
57 - upper confidence limits (*.UCL)
58 - standard error (*.SE)
61 /* All variables. */ \
62 S(CTSF_COUNT, "COUNT", N_("Count"), CTF_COUNT, CTFA_ALL) \
63 S(CTSF_ECOUNT, "ECOUNT", N_("Adjusted Count"), CTF_COUNT, CTFA_ALL) \
64 S(CTSF_ROWPCT_COUNT, "ROWPCT.COUNT", N_("Row %"), CTF_PERCENT, CTFA_ALL) \
65 S(CTSF_COLPCT_COUNT, "COLPCT.COUNT", N_("Column %"), CTF_PERCENT, CTFA_ALL) \
66 S(CTSF_TABLEPCT_COUNT, "TABLEPCT.COUNT", N_("Table %"), CTF_PERCENT, CTFA_ALL) \
67 S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT.COUNT", N_("Subtable %"), CTF_PERCENT, CTFA_ALL) \
68 S(CTSF_LAYERPCT_COUNT, "LAYERPCT.COUNT", N_("Layer %"), CTF_PERCENT, CTFA_ALL) \
69 S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT.COUNT", N_("Layer Row %"), CTF_PERCENT, CTFA_ALL) \
70 S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT.COUNT", N_("Layer Column %"), CTF_PERCENT, CTFA_ALL) \
71 S(CTSF_ROWPCT_VALIDN, "ROWPCT.VALIDN", N_("Row Valid N %"), CTF_PERCENT, CTFA_ALL) \
72 S(CTSF_COLPCT_VALIDN, "COLPCT.VALIDN", N_("Column Valid N %"), CTF_PERCENT, CTFA_ALL) \
73 S(CTSF_TABLEPCT_VALIDN, "TABLEPCT.VALIDN", N_("Table Valid N %"), CTF_PERCENT, CTFA_ALL) \
74 S(CTSF_SUBTABLEPCT_VALIDN, "SUBTABLEPCT.VALIDN", N_("Subtable Valid N %"), CTF_PERCENT, CTFA_ALL) \
75 S(CTSF_LAYERPCT_VALIDN, "LAYERPCT.VALIDN", N_("Layer Valid N %"), CTF_PERCENT, CTFA_ALL) \
76 S(CTSF_LAYERROWPCT_VALIDN, "LAYERROWPCT.VALIDN", N_("Layer Row Valid N %"), CTF_PERCENT, CTFA_ALL) \
77 S(CTSF_LAYERCOLPCT_VALIDN, "LAYERCOLPCT.VALIDN", N_("Layer Column Valid N %"), CTF_PERCENT, CTFA_ALL) \
78 S(CTSF_ROWPCT_TOTALN, "ROWPCT.TOTALN", N_("Row Total N %"), CTF_PERCENT, CTFA_ALL) \
79 S(CTSF_COLPCT_TOTALN, "COLPCT.TOTALN", N_("Column Total N %"), CTF_PERCENT, CTFA_ALL) \
80 S(CTSF_TABLEPCT_TOTALN, "TABLEPCT.TOTALN", N_("Table Total N %"), CTF_PERCENT, CTFA_ALL) \
81 S(CTSF_SUBTABLEPCT_TOTALN, "SUBTABLEPCT.TOTALN", N_("Subtable Total N %"), CTF_PERCENT, CTFA_ALL) \
82 S(CTSF_LAYERPCT_TOTALN, "LAYERPCT.TOTALN", N_("Layer Total N %"), CTF_PERCENT, CTFA_ALL) \
83 S(CTSF_LAYERROWPCT_TOTALN, "LAYERROWPCT.TOTALN", N_("Layer Row Total N %"), CTF_PERCENT, CTFA_ALL) \
84 S(CTSF_LAYERCOLPCT_TOTALN, "LAYERCOLPCT.TOTALN", N_("Layer Column Total N %"), CTF_PERCENT, CTFA_ALL) \
86 /* Scale variables, totals, and subtotals. */ \
87 S(CTSF_MAXIMUM, "MAXIMUM", N_("Maximum"), CTF_GENERAL, CTFA_SCALE) \
88 S(CTSF_MEAN, "MEAN", N_("Mean"), CTF_GENERAL, CTFA_SCALE) \
89 S(CTSF_MEDIAN, "MEDIAN", N_("Median"), CTF_GENERAL, CTFA_SCALE) \
90 S(CTSF_MINIMUM, "MINIMUM", N_("Minimum"), CTF_GENERAL, CTFA_SCALE) \
91 S(CTSF_MISSING, "MISSING", N_("Missing"), CTF_GENERAL, CTFA_SCALE) \
92 S(CTSF_MODE, "MODE", N_("Mode"), CTF_GENERAL, CTFA_SCALE) \
93 S(CTSF_PTILE, "PTILE", N_("Percentile"), CTF_GENERAL, CTFA_SCALE) \
94 S(CTSF_RANGE, "RANGE", N_("Range"), CTF_GENERAL, CTFA_SCALE) \
95 S(CTSF_SEMEAN, "SEMEAN", N_("Std Error of Mean"), CTF_GENERAL, CTFA_SCALE) \
96 S(CTSF_STDDEV, "STDDEV", N_("Std Deviation"), CTF_GENERAL, CTFA_SCALE) \
97 S(CTSF_SUM, "SUM", N_("Sum"), CTF_GENERAL, CTFA_SCALE) \
98 S(CSTF_TOTALN, "TOTALN", N_("Total N"), CTF_COUNT, CTFA_SCALE) \
99 S(CTSF_ETOTALN, "ETOTALN", N_("Adjusted Total N"), CTF_COUNT, CTFA_SCALE) \
100 S(CTSF_VALIDN, "VALIDN", N_("Valid N"), CTF_COUNT, CTFA_SCALE) \
101 S(CTSF_EVALIDN, "EVALIDN", N_("Adjusted Valid N"), CTF_COUNT, CTFA_SCALE) \
102 S(CTSF_VARIANCE, "VARIANCE", N_("Variance"), CTF_GENERAL, CTFA_SCALE) \
103 S(CTSF_ROWPCT_SUM, "ROWPCT.SUM", N_("Row Sum %"), CTF_PERCENT, CTFA_SCALE) \
104 S(CTSF_COLPCT_SUM, "COLPCT.SUM", N_("Column Sum %"), CTF_PERCENT, CTFA_SCALE) \
105 S(CTSF_TABLEPCT_SUM, "TABLEPCT.SUM", N_("Table Sum %"), CTF_PERCENT, CTFA_SCALE) \
106 S(CTSF_SUBTABLEPCT_SUM, "SUBTABLEPCT.SUM", N_("Subtable Sum %"), CTF_PERCENT, CTFA_SCALE) \
107 S(CTSF_LAYERPCT_SUM, "LAYERPCT.SUM", N_("Layer Sum %"), CTF_PERCENT, CTFA_SCALE) \
108 S(CTSF_LAYERROWPCT_SUM, "LAYERROWPCT.SUM", N_("Layer Row Sum %"), CTF_PERCENT, CTFA_SCALE) \
109 S(CTSF_LAYERCOLPCT_SUM, "LAYERCOLPCT.SUM", N_("Layer Column Sum %"), CTF_PERCENT, CTFA_SCALE) \
111 /* Multiple response sets. */ \
112 S(CTSF_RESPONSES, "RESPONSES", N_("Responses"), CTF_COUNT, CTFA_MRSETS) \
113 S(CTSF_ROWPCT_RESPONSES, "ROWPCT.RESPONSES", N_("Row Responses %"), CTF_PERCENT, CTFA_MRSETS) \
114 S(CTSF_COLPCT_RESPONSES, "COLPCT.RESPONSES", N_("Column Responses %"), CTF_PERCENT, CTFA_MRSETS) \
115 S(CTSF_TABLEPCT_RESPONSES, "TABLEPCT.RESPONSES", N_("Table Responses %"), CTF_PERCENT, CTFA_MRSETS) \
116 S(CTSF_SUBTABLEPCT_RESPONSES, "SUBTABLEPCT.RESPONSES", N_("Subtable Responses %"), CTF_PERCENT, CTFA_MRSETS) \
117 S(CTSF_LAYERPCT_RESPONSES, "LAYERPCT.RESPONSES", N_("Layer Responses %"), CTF_PERCENT, CTFA_MRSETS) \
118 S(CTSF_LAYERROWPCT_RESPONSES, "LAYERROWPCT.RESPONSES", N_("Layer Row Responses %"), CTF_PERCENT, CTFA_MRSETS) \
119 S(CTSF_LAYERCOLPCT_RESPONSES, "LAYERCOLPCT.RESPONSES", N_("Layer Column Responses %"), CTF_PERCENT, CTFA_MRSETS) \
120 S(CTSF_ROWPCT_RESPONSES_COUNT, "ROWPCT.RESPONSES.COUNT", N_("Row Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
121 S(CTSF_COLPCT_RESPONSES_COUNT, "COLPCT.RESPONSES.COUNT", N_("Column Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
122 S(CTSF_TABLEPCT_RESPONSES_COUNT, "TABLEPCT.RESPONSES.COUNT", N_("Table Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
123 S(CTSF_SUBTABLEPCT_RESPONSES_COUNT, "SUBTABLEPCT.RESPONSES.COUNT", N_("Subtable Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
124 S(CTSF_LAYERPCT_RESPONSES_COUNT, "LAYERPCT.RESPONSES.COUNT", N_("Layer Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
125 S(CTSF_LAYERROWPCT_RESPONSES_COUNT, "LAYERROWPCT.RESPONSES.COUNT", N_("Layer Row Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
126 S(CTSF_LAYERCOLPCT_RESPONSES_COUNT, "LAYERCOLPCT.RESPONSES.COUNT", N_("Layer Column Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
127 S(CTSF_ROWPCT_COUNT_RESPONSES, "ROWPCT.COUNT.RESPONSES", N_("Row Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
128 S(CTSF_COLPCT_COUNT_RESPONSES, "COLPCT.COUNT.RESPONSES", N_("Column Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
129 S(CTSF_TABLEPCT_COUNT_RESPONSES, "TABLEPCT.COUNT.RESPONSES", N_("Table Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
130 S(CTSF_SUBTABLEPCT_COUNT_RESPONSES, "SUBTABLEPCT.COUNT.RESPONSES", N_("Subtable Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
131 S(CTSF_LAYERPCT_COUNT_RESPONSES, "LAYERPCT.COUNT.RESPONSES", N_("Layer Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
132 S(CTSF_LAYERROWPCT_COUNT_RESPONSES, "LAYERROWPCT.COUNT.RESPONSES", N_("Layer Row Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
133 S(CTSF_LAYERCOLPCT_COUNT_RESPONSES, "LAYERCOLPCT.RESPONSES.COUNT", N_("Layer Column Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS)
135 enum ctables_summary_function
137 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) ENUM,
143 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) +1
144 N_CTSF_FUNCTIONS = SUMMARIES
148 enum ctables_domain_type
150 /* Within a section, where stacked variables divide one section from
152 CTDT_TABLE, /* All layers of a whole section. */
153 CTDT_LAYER, /* One layer within a section. */
154 CTDT_LAYERROW, /* Row in one layer within a section. */
155 CTDT_LAYERCOL, /* Column in one layer within a section. */
157 /* Within a subtable, where a subtable pairs an innermost row variable with
158 an innermost column variable within a single layer. */
159 CTDT_SUBTABLE, /* Whole subtable. */
160 CTDT_ROW, /* Row within a subtable. */
161 CTDT_COL, /* Column within a subtable. */
165 struct ctables_domain
167 struct hmap_node node;
169 const struct ctables_cell *example;
175 enum ctables_summary_variant
184 /* In struct ctables's 'cells' hmap. Indexed by all the values in all the
185 axes (except the scalar variable, if any). */
186 struct hmap_node node;
188 /* The domains that contain this cell. */
189 struct ctables_domain *domains[N_CTDTS];
192 enum ctables_summary_variant sv;
197 struct ctables_cell_value
199 const struct ctables_category *category;
207 union ctables_summary *summaries;
212 struct pivot_table_look *look;
214 /* If this is NULL, zeros are displayed using the normal print format.
215 Otherwise, this string is displayed. */
218 /* If this is NULL, missing values are displayed using the normal print
219 format. Otherwise, this string is displayed. */
222 /* Indexed by variable dictionary index. */
223 enum ctables_vlabel *vlabels;
225 bool mrsets_count_duplicates; /* MRSETS. */
226 bool smissing_listwise; /* SMISSING. */
227 struct variable *base_weight; /* WEIGHT. */
228 int hide_threshold; /* HIDESMALLCOUNTS. */
230 struct ctables_table **tables;
234 struct ctables_postcompute
236 struct hmap_node hmap_node; /* In struct ctables's 'pcompute' hmap. */
237 const char *name; /* Name, without leading &. */
239 struct ctables_postcompute_expr *expr;
242 bool hide_source_cats;
245 struct ctables_postcompute_expr
247 enum ctables_postcompute_op
255 /* XXX SUBTOTAL and HSUBTOTAL */
268 /* CTPO_CAT_NUMBER, CTPO_NUMBER. */
273 XXX what about string ranges? */
276 /* CTPO_ADD, CTPO_SUB, CTPO_MUL, CTPO_DIV, CTPO_POW. */
277 struct ctables_postcompute_expr *subs[2];
281 struct ctables_summary_spec_set
283 struct ctables_summary_spec *specs;
287 struct variable *var;
290 static void ctables_summary_spec_set_clone (struct ctables_summary_spec_set *,
291 const struct ctables_summary_spec_set *);
292 static void ctables_summary_spec_set_uninit (struct ctables_summary_spec_set *);
294 /* A nested sequence of variables, e.g. a > b > c. */
297 struct variable **vars;
300 size_t *domains[N_CTDTS];
301 size_t n_domains[N_CTDTS];
303 struct ctables_summary_spec_set specs[N_CSVS];
306 /* A stack of nestings, e.g. nest1 + nest2 + ... + nestN. */
309 struct ctables_nest *nests;
315 struct hmap_node node;
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 const struct variable *clabels_example;
329 struct hmap clabels_values_map;
330 union value *clabels_values;
331 size_t n_clabels_values;
333 enum pivot_axis_type slabels_axis;
334 bool slabels_visible;
336 /* The innermost category labels for axis 'a' appear on axis label_axis[a].
338 Most commonly, label_axis[a] == a, and in particular we always have
339 label_axis{PIVOT_AXIS_LAYER] == PIVOT_AXIS_LAYER.
341 If ROWLABELS or COLLABELS is specified, then one of
342 label_axis[PIVOT_AXIS_ROW] or label_axis[PIVOT_AXIS_COLUMN] can be the
343 opposite axis or PIVOT_AXIS_LAYER. Only one of them will differ.
345 enum pivot_axis_type label_axis[PIVOT_N_AXES];
346 enum pivot_axis_type clabels_from_axis;
348 /* Indexed by variable dictionary index. */
349 struct ctables_categories **categories;
358 struct ctables_chisq *chisq;
359 struct ctables_pairwise *pairwise;
367 struct variable *var;
368 const struct mrset *mrset;
372 static const struct fmt_spec *
373 ctables_var_get_print_format (const struct ctables_var *var)
375 return (var->is_mrset
376 ? var_get_print_format (var->mrset->vars[0])
377 : var_get_print_format (var->var));
381 ctables_var_name (const struct ctables_var *var)
383 return var->is_mrset ? var->mrset->name : var_get_name (var->var);
386 struct ctables_categories
389 struct ctables_category *cats;
394 struct ctables_category
396 enum ctables_category_type
414 struct ctables_category *subtotal;
418 double number; /* CCT_NUMBER. */
419 char *string; /* CCT_STRING. */
420 double range[2]; /* CCT_RANGE. */
421 char *total_label; /* CCT_SUBTOTAL, CCT_HSUBTOTAL, CCT_TOTAL. */
423 /* CCT_VALUE, CCT_LABEL, CCT_FUNCTION. */
426 bool include_missing;
430 enum ctables_summary_function sort_function;
431 struct variable *sort_var;
438 ctables_category_uninit (struct ctables_category *cat)
458 free (cat->total_label);
469 ctables_category_equal (const struct ctables_category *a,
470 const struct ctables_category *b)
472 if (a->type != b->type)
478 return a->number == b->number;
481 return strcmp (a->string, b->string);
484 return a->range[0] == b->range[0] && a->range[1] == b->range[1];
493 return !strcmp (a->total_label, b->total_label);
498 return (a->include_missing == b->include_missing
499 && a->sort_ascending == b->sort_ascending
500 && a->sort_function == b->sort_function
501 && a->sort_var == b->sort_var
502 && a->percentile == b->percentile);
509 ctables_categories_unref (struct ctables_categories *c)
514 assert (c->n_refs > 0);
518 for (size_t i = 0; i < c->n_cats; i++)
519 ctables_category_uninit (&c->cats[i]);
525 ctables_categories_equal (const struct ctables_categories *a,
526 const struct ctables_categories *b)
528 if (a->n_cats != b->n_cats || a->show_empty != b->show_empty)
531 for (size_t i = 0; i < a->n_cats; i++)
532 if (!ctables_category_equal (&a->cats[i], &b->cats[i]))
538 /* Chi-square test (SIGTEST). */
546 /* Pairwise comparison test (COMPARETEST). */
547 struct ctables_pairwise
549 enum { PROP, MEAN } type;
552 bool meansvariance_allcats;
554 enum { BONFERRONI = 1, BH } adjust;
578 struct ctables_var var;
580 struct ctables_summary_spec_set specs[N_CSVS];
584 struct ctables_axis *subs[2];
587 struct msg_location *loc;
590 static void ctables_axis_destroy (struct ctables_axis *);
599 enum ctables_function_availability
601 CTFA_ALL, /* Any variables. */
602 CTFA_SCALE, /* Only scale variables, totals, and subtotals. */
603 CTFA_MRSETS, /* Only multiple-response sets */
606 struct ctables_summary_spec
608 enum ctables_summary_function function;
609 double percentile; /* CTSF_PTILE only. */
611 struct fmt_spec format; /* XXX extra CTABLES formats */
616 ctables_summary_spec_clone (struct ctables_summary_spec *dst,
617 const struct ctables_summary_spec *src)
620 dst->label = xstrdup (src->label);
624 ctables_summary_spec_uninit (struct ctables_summary_spec *s)
631 ctables_summary_spec_set_clone (struct ctables_summary_spec_set *dst,
632 const struct ctables_summary_spec_set *src)
634 struct ctables_summary_spec *specs = xnmalloc (src->n, sizeof *specs);
635 for (size_t i = 0; i < src->n; i++)
636 ctables_summary_spec_clone (&specs[i], &src->specs[i]);
638 *dst = (struct ctables_summary_spec_set) {
647 ctables_summary_spec_set_uninit (struct ctables_summary_spec_set *set)
649 for (size_t i = 0; i < set->n; i++)
650 ctables_summary_spec_uninit (&set->specs[i]);
655 parse_col_width (struct lexer *lexer, const char *name, double *width)
657 lex_match (lexer, T_EQUALS);
658 if (lex_match_id (lexer, "DEFAULT"))
660 else if (lex_force_num_range_closed (lexer, name, 0, DBL_MAX))
662 *width = lex_number (lexer);
672 parse_bool (struct lexer *lexer, bool *b)
674 if (lex_match_id (lexer, "NO"))
676 else if (lex_match_id (lexer, "YES"))
680 lex_error_expecting (lexer, "YES", "NO");
686 static enum ctables_function_availability
687 ctables_function_availability (enum ctables_summary_function f)
689 static enum ctables_function_availability availability[] = {
690 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = AVAILABILITY,
695 return availability[f];
699 parse_ctables_summary_function (struct lexer *lexer,
700 enum ctables_summary_function *f)
704 enum ctables_summary_function function;
705 struct substring name;
707 static struct pair names[] = {
708 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) \
709 { ENUM, SS_LITERAL_INITIALIZER (NAME) },
712 /* The .COUNT suffix may be omitted. */
713 S(CTSF_ROWPCT_COUNT, "ROWPCT", _, _, _)
714 S(CTSF_COLPCT_COUNT, "COLPCT", _, _, _)
715 S(CTSF_TABLEPCT_COUNT, "TABLEPCT", _, _, _)
716 S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT", _, _, _)
717 S(CTSF_LAYERPCT_COUNT, "LAYERPCT", _, _, _)
718 S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT", _, _, _)
719 S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT", _, _, _)
723 if (!lex_force_id (lexer))
726 for (size_t i = 0; i < sizeof names / sizeof *names; i++)
727 if (ss_equals_case (names[i].name, lex_tokss (lexer)))
729 *f = names[i].function;
734 lex_error (lexer, _("Expecting summary function name."));
739 ctables_axis_destroy (struct ctables_axis *axis)
747 for (size_t i = 0; i < N_CSVS; i++)
748 ctables_summary_spec_set_uninit (&axis->specs[i]);
753 ctables_axis_destroy (axis->subs[0]);
754 ctables_axis_destroy (axis->subs[1]);
757 msg_location_destroy (axis->loc);
761 static struct ctables_axis *
762 ctables_axis_new_nonterminal (enum ctables_axis_op op,
763 struct ctables_axis *sub0,
764 struct ctables_axis *sub1,
765 struct lexer *lexer, int start_ofs)
767 struct ctables_axis *axis = xmalloc (sizeof *axis);
768 *axis = (struct ctables_axis) {
770 .subs = { sub0, sub1 },
771 .loc = lex_ofs_location (lexer, start_ofs, lex_ofs (lexer) - 1),
776 struct ctables_axis_parse_ctx
779 struct dictionary *dict;
781 struct ctables_table *t;
784 static struct fmt_spec
785 ctables_summary_default_format (enum ctables_summary_function function,
786 const struct ctables_var *var)
788 static const enum ctables_format default_formats[] = {
789 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = FORMAT,
793 switch (default_formats[function])
796 return (struct fmt_spec) { .type = FMT_F, .w = 40 };
799 return (struct fmt_spec) { .type = FMT_PCT, .w = 40, .d = 1 };
802 return *ctables_var_get_print_format (var);
810 ctables_summary_default_label (enum ctables_summary_function function,
813 static const char *default_labels[] = {
814 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = LABEL,
819 return (function == CTSF_PTILE
820 ? xasprintf (_("Percentile %.2f"), percentile)
821 : xstrdup (gettext (default_labels[function])));
825 ctables_summary_function_name (enum ctables_summary_function function)
827 static const char *names[] = {
828 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = NAME,
832 return names[function];
836 add_summary_spec (struct ctables_axis *axis,
837 enum ctables_summary_function function, double percentile,
838 const char *label, const struct fmt_spec *format,
839 const struct msg_location *loc, enum ctables_summary_variant sv)
841 if (axis->op == CTAO_VAR)
843 const char *function_name = ctables_summary_function_name (function);
844 const char *var_name = ctables_var_name (&axis->var);
845 switch (ctables_function_availability (function))
848 if (!axis->var.is_mrset)
850 msg_at (SE, loc, _("Summary function %s applies only to multiple "
851 "response sets."), function_name);
852 msg_at (SN, axis->loc, _("'%s' is not a multiple response set."),
862 _("Summary function %s applies only to scale variables."),
864 msg_at (SN, axis->loc, _("'%s' is not a scale variable."),
874 struct ctables_summary_spec_set *set = &axis->specs[sv];
875 if (set->n >= set->allocated)
876 set->specs = x2nrealloc (set->specs, &set->allocated,
879 struct ctables_summary_spec *dst = &set->specs[set->n++];
880 *dst = (struct ctables_summary_spec) {
881 .function = function,
882 .percentile = percentile,
883 .label = xstrdup (label),
884 .format = (format ? *format
885 : ctables_summary_default_format (function, &axis->var)),
891 for (size_t i = 0; i < 2; i++)
892 if (!add_summary_spec (axis->subs[i], function, percentile, label,
899 static struct ctables_axis *ctables_axis_parse_stack (
900 struct ctables_axis_parse_ctx *);
903 ctables_var_parse (struct lexer *lexer, struct dictionary *dict,
904 struct ctables_var *var)
906 if (ss_starts_with (lex_tokss (lexer), ss_cstr ("$")))
908 *var = (struct ctables_var) {
910 .mrset = dict_lookup_mrset (dict, lex_tokcstr (lexer))
914 lex_error (lexer, _("'%s' does not name a multiple-response set "
915 "in the active file dictionary."),
916 lex_tokcstr (lexer));
924 *var = (struct ctables_var) {
926 .var = parse_variable (lexer, dict),
928 return var->var != NULL;
932 static struct ctables_axis *
933 ctables_axis_parse_primary (struct ctables_axis_parse_ctx *ctx)
935 if (lex_match (ctx->lexer, T_LPAREN))
937 struct ctables_axis *sub = ctables_axis_parse_stack (ctx);
938 if (!sub || !lex_force_match (ctx->lexer, T_RPAREN))
940 ctables_axis_destroy (sub);
946 if (!lex_force_id (ctx->lexer))
949 int start_ofs = lex_ofs (ctx->lexer);
950 struct ctables_var var;
951 if (!ctables_var_parse (ctx->lexer, ctx->dict, &var))
954 struct ctables_axis *axis = xmalloc (sizeof *axis);
955 *axis = (struct ctables_axis) { .op = CTAO_VAR, .var = var };
957 /* XXX should figure out default measures by reading data */
958 axis->scale = (var.is_mrset ? false
959 : lex_match_phrase (ctx->lexer, "[S]") ? true
960 : lex_match_phrase (ctx->lexer, "[C]") ? false
961 : var_get_measure (var.var) == MEASURE_SCALE);
962 axis->loc = lex_ofs_location (ctx->lexer, start_ofs,
963 lex_ofs (ctx->lexer) - 1);
968 has_digit (const char *s)
970 return s[strcspn (s, "0123456789")] != '\0';
973 static struct ctables_axis *
974 ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
976 struct ctables_axis *sub = ctables_axis_parse_primary (ctx);
977 if (!sub || !lex_match (ctx->lexer, T_LBRACK))
980 enum ctables_summary_variant sv = CSV_CELL;
983 int start_ofs = lex_ofs (ctx->lexer);
985 /* Parse function. */
986 enum ctables_summary_function function;
987 if (!parse_ctables_summary_function (ctx->lexer, &function))
990 /* Parse percentile. */
991 double percentile = 0;
992 if (function == CTSF_PTILE)
994 if (!lex_force_num_range_closed (ctx->lexer, "PTILE", 0, 100))
996 percentile = lex_number (ctx->lexer);
997 lex_get (ctx->lexer);
1002 if (lex_is_string (ctx->lexer))
1004 label = ss_xstrdup (lex_tokss (ctx->lexer));
1005 lex_get (ctx->lexer);
1008 label = ctables_summary_default_label (function, percentile);
1011 struct fmt_spec format;
1012 const struct fmt_spec *formatp;
1013 if (lex_token (ctx->lexer) == T_ID
1014 && has_digit (lex_tokcstr (ctx->lexer)))
1016 if (!parse_format_specifier (ctx->lexer, &format)
1017 || !fmt_check_output (&format)
1018 || !fmt_check_type_compat (&format, VAL_NUMERIC))
1028 struct msg_location *loc = lex_ofs_location (ctx->lexer, start_ofs,
1029 lex_ofs (ctx->lexer) - 1);
1030 add_summary_spec (sub, function, percentile, label, formatp, loc, sv);
1032 msg_location_destroy (loc);
1034 lex_match (ctx->lexer, T_COMMA);
1035 if (sv == CSV_CELL && lex_match_id (ctx->lexer, "TOTALS"))
1037 if (!lex_force_match (ctx->lexer, T_LBRACK))
1041 else if (lex_match (ctx->lexer, T_RBRACK))
1043 if (sv == CSV_TOTAL && !lex_force_match (ctx->lexer, T_RBRACK))
1050 ctables_axis_destroy (sub);
1054 static const struct ctables_axis *
1055 find_scale (const struct ctables_axis *axis)
1059 else if (axis->op == CTAO_VAR)
1063 assert (!axis->var.is_mrset);
1071 for (size_t i = 0; i < 2; i++)
1073 const struct ctables_axis *scale = find_scale (axis->subs[i]);
1081 static const struct ctables_axis *
1082 find_categorical_summary_spec (const struct ctables_axis *axis)
1086 else if (axis->op == CTAO_VAR)
1087 return !axis->scale && axis->specs[CSV_CELL].n ? axis : NULL;
1090 for (size_t i = 0; i < 2; i++)
1092 const struct ctables_axis *sum
1093 = find_categorical_summary_spec (axis->subs[i]);
1101 static struct ctables_axis *
1102 ctables_axis_parse_nest (struct ctables_axis_parse_ctx *ctx)
1104 int start_ofs = lex_ofs (ctx->lexer);
1105 struct ctables_axis *lhs = ctables_axis_parse_postfix (ctx);
1109 while (lex_match (ctx->lexer, T_GT))
1111 struct ctables_axis *rhs = ctables_axis_parse_postfix (ctx);
1115 struct ctables_axis *nest = ctables_axis_new_nonterminal (
1116 CTAO_NEST, lhs, rhs, ctx->lexer, start_ofs);
1118 const struct ctables_axis *outer_scale = find_scale (lhs);
1119 const struct ctables_axis *inner_scale = find_scale (rhs);
1120 if (outer_scale && inner_scale)
1122 msg_at (SE, nest->loc, _("Cannot nest scale variables."));
1123 msg_at (SN, outer_scale->loc, _("This is an outer scale variable."));
1124 msg_at (SN, inner_scale->loc, _("This is an inner scale variable."));
1125 ctables_axis_destroy (nest);
1129 const struct ctables_axis *outer_sum = find_categorical_summary_spec (lhs);
1132 msg_at (SE, nest->loc,
1133 _("Summaries may only be requested for categorical variables "
1134 "at the innermost nesting level."));
1135 msg_at (SN, outer_sum->loc,
1136 _("This outer categorical variable has a summary."));
1137 ctables_axis_destroy (nest);
1147 static struct ctables_axis *
1148 ctables_axis_parse_stack (struct ctables_axis_parse_ctx *ctx)
1150 int start_ofs = lex_ofs (ctx->lexer);
1151 struct ctables_axis *lhs = ctables_axis_parse_nest (ctx);
1155 while (lex_match (ctx->lexer, T_PLUS))
1157 struct ctables_axis *rhs = ctables_axis_parse_nest (ctx);
1161 lhs = ctables_axis_new_nonterminal (CTAO_STACK, lhs, rhs,
1162 ctx->lexer, start_ofs);
1169 ctables_axis_parse (struct lexer *lexer, struct dictionary *dict,
1170 struct ctables *ct, struct ctables_table *t,
1171 enum pivot_axis_type a)
1173 if (lex_token (lexer) == T_BY
1174 || lex_token (lexer) == T_SLASH
1175 || lex_token (lexer) == T_ENDCMD)
1178 struct ctables_axis_parse_ctx ctx = {
1184 t->axes[a] = ctables_axis_parse_stack (&ctx);
1185 return t->axes[a] != NULL;
1189 ctables_chisq_destroy (struct ctables_chisq *chisq)
1195 ctables_pairwise_destroy (struct ctables_pairwise *pairwise)
1201 ctables_table_destroy (struct ctables_table *t)
1206 for (size_t i = 0; i < t->n_categories; i++)
1207 ctables_categories_unref (t->categories[i]);
1208 free (t->categories);
1210 ctables_axis_destroy (t->axes[PIVOT_AXIS_COLUMN]);
1211 ctables_axis_destroy (t->axes[PIVOT_AXIS_ROW]);
1212 ctables_axis_destroy (t->axes[PIVOT_AXIS_LAYER]);
1216 ctables_chisq_destroy (t->chisq);
1217 ctables_pairwise_destroy (t->pairwise);
1222 ctables_destroy (struct ctables *ct)
1227 pivot_table_look_unref (ct->look);
1231 for (size_t i = 0; i < ct->n_tables; i++)
1232 ctables_table_destroy (ct->tables[i]);
1237 static struct ctables_category
1238 cct_range (double low, double high)
1240 return (struct ctables_category) {
1242 .range = { low, high }
1247 ctables_table_parse_categories (struct lexer *lexer, struct dictionary *dict,
1248 struct ctables_table *t)
1250 if (!lex_match_id (lexer, "VARIABLES"))
1252 lex_match (lexer, T_EQUALS);
1254 struct variable **vars;
1256 if (!parse_variables (lexer, dict, &vars, &n_vars, PV_NO_SCRATCH))
1259 struct ctables_categories *c = xmalloc (sizeof *c);
1260 *c = (struct ctables_categories) { .n_refs = n_vars };
1261 for (size_t i = 0; i < n_vars; i++)
1263 struct ctables_categories **cp
1264 = &t->categories[var_get_dict_index (vars[i])];
1265 ctables_categories_unref (*cp);
1270 size_t allocated_cats = 0;
1271 if (lex_match (lexer, T_LBRACK))
1275 if (c->n_cats >= allocated_cats)
1276 c->cats = x2nrealloc (c->cats, &allocated_cats, sizeof *c->cats);
1278 struct ctables_category *cat = &c->cats[c->n_cats];
1279 if (lex_match_id (lexer, "OTHERNM"))
1280 cat->type = CCT_OTHERNM;
1281 else if (lex_match_id (lexer, "MISSING"))
1282 cat->type = CCT_MISSING;
1283 else if (lex_match_id (lexer, "SUBTOTAL"))
1284 *cat = (struct ctables_category)
1285 { .type = CCT_SUBTOTAL, .total_label = NULL };
1286 else if (lex_match_id (lexer, "HSUBTOTAL"))
1287 *cat = (struct ctables_category)
1288 { .type = CCT_HSUBTOTAL, .total_label = NULL };
1289 else if (lex_match_id (lexer, "LO"))
1291 if (!lex_force_match_id (lexer, "THRU") || lex_force_num (lexer))
1293 *cat = cct_range (-DBL_MAX, lex_number (lexer));
1296 else if (lex_is_number (lexer))
1298 double number = lex_number (lexer);
1300 if (lex_match_id (lexer, "THRU"))
1302 cat->type = CCT_RANGE;
1303 cat->range[0] = number;
1304 if (lex_match_id (lexer, "HI"))
1305 *cat = cct_range (number, DBL_MAX);
1308 if (!lex_force_num (lexer))
1310 *cat = cct_range (number, lex_number (lexer));
1315 *cat = (struct ctables_category) {
1320 else if (lex_is_string (lexer))
1322 *cat = (struct ctables_category) {
1324 .string = ss_xstrdup (lex_tokss (lexer)),
1330 lex_error (lexer, NULL);
1334 if (cat->type == CCT_SUBTOTAL || cat->type == CCT_HSUBTOTAL)
1336 if (lex_match (lexer, T_EQUALS))
1338 if (!lex_force_string (lexer))
1341 cat->total_label = ss_xstrdup (lex_tokss (lexer));
1345 cat->total_label = xstrdup (_("Subtotal"));
1349 lex_match (lexer, T_COMMA);
1351 while (!lex_match (lexer, T_RBRACK));
1354 struct ctables_category cat = {
1356 .include_missing = false,
1357 .sort_ascending = true,
1359 bool show_totals = false;
1360 char *total_label = NULL;
1361 bool totals_before = false;
1362 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
1364 if (!c->n_cats && lex_match_id (lexer, "ORDER"))
1366 lex_match (lexer, T_EQUALS);
1367 if (lex_match_id (lexer, "A"))
1368 cat.sort_ascending = true;
1369 else if (lex_match_id (lexer, "D"))
1370 cat.sort_ascending = false;
1373 lex_error_expecting (lexer, "A", "D");
1377 else if (!c->n_cats && lex_match_id (lexer, "KEY"))
1379 lex_match (lexer, T_EQUALS);
1380 if (lex_match_id (lexer, "VALUE"))
1381 cat.type = CCT_VALUE;
1382 else if (lex_match_id (lexer, "LABEL"))
1383 cat.type = CCT_LABEL;
1386 cat.type = CCT_FUNCTION;
1387 if (!parse_ctables_summary_function (lexer, &cat.sort_function))
1390 if (lex_match (lexer, T_LPAREN))
1392 cat.sort_var = parse_variable (lexer, dict);
1396 if (cat.sort_function == CTSF_PTILE)
1398 lex_match (lexer, T_COMMA);
1399 if (!lex_force_num_range_closed (lexer, "PTILE", 0, 100))
1401 cat.percentile = lex_number (lexer);
1405 if (!lex_force_match (lexer, T_RPAREN))
1408 else if (ctables_function_availability (cat.sort_function)
1411 bool UNUSED b = lex_force_match (lexer, T_LPAREN);
1416 else if (!c->n_cats && lex_match_id (lexer, "MISSING"))
1418 lex_match (lexer, T_EQUALS);
1419 if (lex_match_id (lexer, "INCLUDE"))
1420 cat.include_missing = true;
1421 else if (lex_match_id (lexer, "EXCLUDE"))
1422 cat.include_missing = false;
1425 lex_error_expecting (lexer, "INCLUDE", "EXCLUDE");
1429 else if (lex_match_id (lexer, "TOTAL"))
1431 lex_match (lexer, T_EQUALS);
1432 if (!parse_bool (lexer, &show_totals))
1435 else if (lex_match_id (lexer, "LABEL"))
1437 lex_match (lexer, T_EQUALS);
1438 if (!lex_force_string (lexer))
1441 total_label = ss_xstrdup (lex_tokss (lexer));
1444 else if (lex_match_id (lexer, "POSITION"))
1446 lex_match (lexer, T_EQUALS);
1447 if (lex_match_id (lexer, "BEFORE"))
1448 totals_before = true;
1449 else if (lex_match_id (lexer, "AFTER"))
1450 totals_before = false;
1453 lex_error_expecting (lexer, "BEFORE", "AFTER");
1457 else if (lex_match_id (lexer, "EMPTY"))
1459 lex_match (lexer, T_EQUALS);
1460 if (lex_match_id (lexer, "INCLUDE"))
1461 c->show_empty = true;
1462 else if (lex_match_id (lexer, "EXCLUDE"))
1463 c->show_empty = false;
1466 lex_error_expecting (lexer, "INCLUDE", "EXCLUDE");
1473 lex_error_expecting (lexer, "ORDER", "KEY", "MISSING",
1474 "TOTAL", "LABEL", "POSITION", "EMPTY");
1476 lex_error_expecting (lexer, "TOTAL", "LABEL", "POSITION", "EMPTY");
1483 if (c->n_cats >= allocated_cats)
1484 c->cats = x2nrealloc (c->cats, &allocated_cats,
1486 c->cats[c->n_cats++] = cat;
1491 if (c->n_cats >= allocated_cats)
1492 c->cats = x2nrealloc (c->cats, &allocated_cats, sizeof *c->cats);
1494 struct ctables_category *totals;
1497 insert_element (c->cats, c->n_cats, sizeof *c->cats, 0);
1498 totals = &c->cats[0];
1501 totals = &c->cats[c->n_cats];
1504 *totals = (struct ctables_category) {
1506 .total_label = total_label ? total_label : xstrdup (_("Total")),
1510 struct ctables_category *subtotal = NULL;
1511 for (size_t i = totals_before ? 0 : c->n_cats;
1512 totals_before ? i < c->n_cats : i-- > 0;
1513 totals_before ? i++ : 0)
1515 struct ctables_category *cat = &c->cats[i];
1523 cat->subtotal = subtotal;
1543 ctables_nest_uninit (struct ctables_nest *nest)
1550 ctables_stack_uninit (struct ctables_stack *stack)
1554 for (size_t i = 0; i < stack->n; i++)
1555 ctables_nest_uninit (&stack->nests[i]);
1556 free (stack->nests);
1560 static struct ctables_stack
1561 nest_fts (struct ctables_stack s0, struct ctables_stack s1)
1568 struct ctables_stack stack = { .nests = xnmalloc (s0.n, s1.n * sizeof *stack.nests) };
1569 for (size_t i = 0; i < s0.n; i++)
1570 for (size_t j = 0; j < s1.n; j++)
1572 const struct ctables_nest *a = &s0.nests[i];
1573 const struct ctables_nest *b = &s1.nests[j];
1575 size_t allocate = a->n + b->n;
1576 struct variable **vars = xnmalloc (allocate, sizeof *vars);
1577 enum pivot_axis_type *axes = xnmalloc (allocate, sizeof *axes);
1579 for (size_t k = 0; k < a->n; k++)
1580 vars[n++] = a->vars[k];
1581 for (size_t k = 0; k < b->n; k++)
1582 vars[n++] = b->vars[k];
1583 assert (n == allocate);
1585 const struct ctables_nest *summary_src;
1586 if (!a->specs[CSV_CELL].var)
1588 else if (!b->specs[CSV_CELL].var)
1593 struct ctables_nest *new = &stack.nests[stack.n++];
1594 *new = (struct ctables_nest) {
1596 .scale_idx = (a->scale_idx != SIZE_MAX ? a->scale_idx
1597 : b->scale_idx != SIZE_MAX ? a->n + b->scale_idx
1601 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
1602 ctables_summary_spec_set_clone (&new->specs[sv], &summary_src->specs[sv]);
1604 ctables_stack_uninit (&s0);
1605 ctables_stack_uninit (&s1);
1609 static struct ctables_stack
1610 stack_fts (struct ctables_stack s0, struct ctables_stack s1)
1612 struct ctables_stack stack = { .nests = xnmalloc (s0.n + s1.n, sizeof *stack.nests) };
1613 for (size_t i = 0; i < s0.n; i++)
1614 stack.nests[stack.n++] = s0.nests[i];
1615 for (size_t i = 0; i < s1.n; i++)
1616 stack.nests[stack.n++] = s1.nests[i];
1617 assert (stack.n == s0.n + s1.n);
1623 static struct ctables_stack
1624 enumerate_fts (enum pivot_axis_type axis_type, const struct ctables_axis *a)
1627 return (struct ctables_stack) { .n = 0 };
1632 assert (!a->var.is_mrset);
1634 struct variable **vars = xmalloc (sizeof *vars);
1637 struct ctables_nest *nest = xmalloc (sizeof *nest);
1638 *nest = (struct ctables_nest) {
1641 .scale_idx = a->scale ? 0 : SIZE_MAX,
1643 if (a->specs[CSV_CELL].n || a->scale)
1644 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
1646 ctables_summary_spec_set_clone (&nest->specs[sv], &a->specs[sv]);
1647 nest->specs[sv].var = a->var.var;
1649 return (struct ctables_stack) { .nests = nest, .n = 1 };
1652 return stack_fts (enumerate_fts (axis_type, a->subs[0]),
1653 enumerate_fts (axis_type, a->subs[1]));
1656 return nest_fts (enumerate_fts (axis_type, a->subs[0]),
1657 enumerate_fts (axis_type, a->subs[1]));
1663 union ctables_summary
1665 /* COUNT, VALIDN, TOTALN. */
1672 /* MINIMUM, MAXIMUM, RANGE. */
1679 /* MEAN, SEMEAN, STDDEV, SUM, VARIANCE, *.SUM. */
1680 struct moments1 *moments;
1682 /* XXX percentiles, median, mode, multiple response */
1686 ctables_summary_init (union ctables_summary *s,
1687 const struct ctables_summary_spec *ss)
1689 switch (ss->function)
1693 case CTSF_ROWPCT_COUNT:
1694 case CTSF_COLPCT_COUNT:
1695 case CTSF_TABLEPCT_COUNT:
1696 case CTSF_SUBTABLEPCT_COUNT:
1697 case CTSF_LAYERPCT_COUNT:
1698 case CTSF_LAYERROWPCT_COUNT:
1699 case CTSF_LAYERCOLPCT_COUNT:
1700 case CTSF_ROWPCT_VALIDN:
1701 case CTSF_COLPCT_VALIDN:
1702 case CTSF_TABLEPCT_VALIDN:
1703 case CTSF_SUBTABLEPCT_VALIDN:
1704 case CTSF_LAYERPCT_VALIDN:
1705 case CTSF_LAYERROWPCT_VALIDN:
1706 case CTSF_LAYERCOLPCT_VALIDN:
1707 case CTSF_ROWPCT_TOTALN:
1708 case CTSF_COLPCT_TOTALN:
1709 case CTSF_TABLEPCT_TOTALN:
1710 case CTSF_SUBTABLEPCT_TOTALN:
1711 case CTSF_LAYERPCT_TOTALN:
1712 case CTSF_LAYERROWPCT_TOTALN:
1713 case CTSF_LAYERCOLPCT_TOTALN:
1718 s->missing = s->valid = 0;
1724 s->min = s->max = SYSMIS;
1732 case CTSF_ROWPCT_SUM:
1733 case CTSF_COLPCT_SUM:
1734 case CTSF_TABLEPCT_SUM:
1735 case CTSF_SUBTABLEPCT_SUM:
1736 case CTSF_LAYERPCT_SUM:
1737 case CTSF_LAYERROWPCT_SUM:
1738 case CTSF_LAYERCOLPCT_SUM:
1739 s->moments = moments1_create (MOMENT_VARIANCE);
1748 case CTSF_RESPONSES:
1749 case CTSF_ROWPCT_RESPONSES:
1750 case CTSF_COLPCT_RESPONSES:
1751 case CTSF_TABLEPCT_RESPONSES:
1752 case CTSF_SUBTABLEPCT_RESPONSES:
1753 case CTSF_LAYERPCT_RESPONSES:
1754 case CTSF_LAYERROWPCT_RESPONSES:
1755 case CTSF_LAYERCOLPCT_RESPONSES:
1756 case CTSF_ROWPCT_RESPONSES_COUNT:
1757 case CTSF_COLPCT_RESPONSES_COUNT:
1758 case CTSF_TABLEPCT_RESPONSES_COUNT:
1759 case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
1760 case CTSF_LAYERPCT_RESPONSES_COUNT:
1761 case CTSF_LAYERROWPCT_RESPONSES_COUNT:
1762 case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
1763 case CTSF_ROWPCT_COUNT_RESPONSES:
1764 case CTSF_COLPCT_COUNT_RESPONSES:
1765 case CTSF_TABLEPCT_COUNT_RESPONSES:
1766 case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
1767 case CTSF_LAYERPCT_COUNT_RESPONSES:
1768 case CTSF_LAYERROWPCT_COUNT_RESPONSES:
1769 case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
1775 ctables_summary_uninit (union ctables_summary *s,
1776 const struct ctables_summary_spec *ss)
1778 switch (ss->function)
1782 case CTSF_ROWPCT_COUNT:
1783 case CTSF_COLPCT_COUNT:
1784 case CTSF_TABLEPCT_COUNT:
1785 case CTSF_SUBTABLEPCT_COUNT:
1786 case CTSF_LAYERPCT_COUNT:
1787 case CTSF_LAYERROWPCT_COUNT:
1788 case CTSF_LAYERCOLPCT_COUNT:
1789 case CTSF_ROWPCT_VALIDN:
1790 case CTSF_COLPCT_VALIDN:
1791 case CTSF_TABLEPCT_VALIDN:
1792 case CTSF_SUBTABLEPCT_VALIDN:
1793 case CTSF_LAYERPCT_VALIDN:
1794 case CTSF_LAYERROWPCT_VALIDN:
1795 case CTSF_LAYERCOLPCT_VALIDN:
1796 case CTSF_ROWPCT_TOTALN:
1797 case CTSF_COLPCT_TOTALN:
1798 case CTSF_TABLEPCT_TOTALN:
1799 case CTSF_SUBTABLEPCT_TOTALN:
1800 case CTSF_LAYERPCT_TOTALN:
1801 case CTSF_LAYERROWPCT_TOTALN:
1802 case CTSF_LAYERCOLPCT_TOTALN:
1819 case CTSF_ROWPCT_SUM:
1820 case CTSF_COLPCT_SUM:
1821 case CTSF_TABLEPCT_SUM:
1822 case CTSF_SUBTABLEPCT_SUM:
1823 case CTSF_LAYERPCT_SUM:
1824 case CTSF_LAYERROWPCT_SUM:
1825 case CTSF_LAYERCOLPCT_SUM:
1826 moments1_destroy (s->moments);
1835 case CTSF_RESPONSES:
1836 case CTSF_ROWPCT_RESPONSES:
1837 case CTSF_COLPCT_RESPONSES:
1838 case CTSF_TABLEPCT_RESPONSES:
1839 case CTSF_SUBTABLEPCT_RESPONSES:
1840 case CTSF_LAYERPCT_RESPONSES:
1841 case CTSF_LAYERROWPCT_RESPONSES:
1842 case CTSF_LAYERCOLPCT_RESPONSES:
1843 case CTSF_ROWPCT_RESPONSES_COUNT:
1844 case CTSF_COLPCT_RESPONSES_COUNT:
1845 case CTSF_TABLEPCT_RESPONSES_COUNT:
1846 case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
1847 case CTSF_LAYERPCT_RESPONSES_COUNT:
1848 case CTSF_LAYERROWPCT_RESPONSES_COUNT:
1849 case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
1850 case CTSF_ROWPCT_COUNT_RESPONSES:
1851 case CTSF_COLPCT_COUNT_RESPONSES:
1852 case CTSF_TABLEPCT_COUNT_RESPONSES:
1853 case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
1854 case CTSF_LAYERPCT_COUNT_RESPONSES:
1855 case CTSF_LAYERROWPCT_COUNT_RESPONSES:
1856 case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
1862 ctables_summary_add (union ctables_summary *s,
1863 const struct ctables_summary_spec *ss,
1864 const struct variable *var, const union value *value,
1867 switch (ss->function)
1871 case CTSF_ROWPCT_COUNT:
1872 case CTSF_COLPCT_COUNT:
1873 case CTSF_TABLEPCT_COUNT:
1874 case CTSF_SUBTABLEPCT_COUNT:
1875 case CTSF_LAYERPCT_COUNT:
1876 case CTSF_LAYERROWPCT_COUNT:
1877 case CTSF_LAYERCOLPCT_COUNT:
1878 case CTSF_ROWPCT_VALIDN:
1879 case CTSF_COLPCT_VALIDN:
1880 case CTSF_TABLEPCT_VALIDN:
1881 case CTSF_SUBTABLEPCT_VALIDN:
1882 case CTSF_LAYERPCT_VALIDN:
1883 case CTSF_LAYERROWPCT_VALIDN:
1884 case CTSF_LAYERCOLPCT_VALIDN:
1885 case CTSF_ROWPCT_TOTALN:
1886 case CTSF_COLPCT_TOTALN:
1887 case CTSF_TABLEPCT_TOTALN:
1888 case CTSF_SUBTABLEPCT_TOTALN:
1889 case CTSF_LAYERPCT_TOTALN:
1890 case CTSF_LAYERROWPCT_TOTALN:
1891 case CTSF_LAYERCOLPCT_TOTALN:
1896 if (var_is_value_missing (var, value))
1897 s->missing += weight;
1905 if (!var_is_value_missing (var, value))
1907 assert (!var_is_alpha (var)); /* XXX? */
1908 if (s->min == SYSMIS || value->f < s->min)
1910 if (s->max == SYSMIS || value->f > s->max)
1920 case CTSF_ROWPCT_SUM:
1921 case CTSF_COLPCT_SUM:
1922 case CTSF_TABLEPCT_SUM:
1923 case CTSF_SUBTABLEPCT_SUM:
1924 case CTSF_LAYERPCT_SUM:
1925 case CTSF_LAYERROWPCT_SUM:
1926 case CTSF_LAYERCOLPCT_SUM:
1927 moments1_add (s->moments, value->f, weight);
1936 case CTSF_RESPONSES:
1937 case CTSF_ROWPCT_RESPONSES:
1938 case CTSF_COLPCT_RESPONSES:
1939 case CTSF_TABLEPCT_RESPONSES:
1940 case CTSF_SUBTABLEPCT_RESPONSES:
1941 case CTSF_LAYERPCT_RESPONSES:
1942 case CTSF_LAYERROWPCT_RESPONSES:
1943 case CTSF_LAYERCOLPCT_RESPONSES:
1944 case CTSF_ROWPCT_RESPONSES_COUNT:
1945 case CTSF_COLPCT_RESPONSES_COUNT:
1946 case CTSF_TABLEPCT_RESPONSES_COUNT:
1947 case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
1948 case CTSF_LAYERPCT_RESPONSES_COUNT:
1949 case CTSF_LAYERROWPCT_RESPONSES_COUNT:
1950 case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
1951 case CTSF_ROWPCT_COUNT_RESPONSES:
1952 case CTSF_COLPCT_COUNT_RESPONSES:
1953 case CTSF_TABLEPCT_COUNT_RESPONSES:
1954 case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
1955 case CTSF_LAYERPCT_COUNT_RESPONSES:
1956 case CTSF_LAYERROWPCT_COUNT_RESPONSES:
1957 case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
1962 static double UNUSED
1963 ctables_summary_value (const struct ctables_cell *cell,
1964 union ctables_summary *s,
1965 const struct ctables_summary_spec *ss)
1967 switch (ss->function)
1973 case CTSF_SUBTABLEPCT_COUNT:
1974 return cell->domains[CTDT_SUBTABLE]->valid ? s->valid / cell->domains[CTDT_SUBTABLE]->valid * 100 : SYSMIS;
1976 case CTSF_ROWPCT_COUNT:
1977 return cell->domains[CTDT_ROW]->valid ? s->valid / cell->domains[CTDT_ROW]->valid * 100 : SYSMIS;
1979 case CTSF_COLPCT_COUNT:
1980 return cell->domains[CTDT_COL]->valid ? s->valid / cell->domains[CTDT_COL]->valid * 100 : SYSMIS;
1982 case CTSF_TABLEPCT_COUNT:
1983 return cell->domains[CTDT_TABLE]->valid ? s->valid / cell->domains[CTDT_TABLE]->valid * 100 : SYSMIS;
1985 case CTSF_LAYERPCT_COUNT:
1986 return cell->domains[CTDT_LAYER]->valid ? s->valid / cell->domains[CTDT_LAYER]->valid * 100 : SYSMIS;
1988 case CTSF_LAYERROWPCT_COUNT:
1989 return cell->domains[CTDT_LAYERROW]->valid ? s->valid / cell->domains[CTDT_LAYERROW]->valid * 100 : SYSMIS;
1991 case CTSF_LAYERCOLPCT_COUNT:
1992 return cell->domains[CTDT_LAYERCOL]->valid ? s->valid / cell->domains[CTDT_LAYERCOL]->valid * 100 : SYSMIS;
1994 case CTSF_ROWPCT_VALIDN:
1995 case CTSF_COLPCT_VALIDN:
1996 case CTSF_TABLEPCT_VALIDN:
1997 case CTSF_SUBTABLEPCT_VALIDN:
1998 case CTSF_LAYERPCT_VALIDN:
1999 case CTSF_LAYERROWPCT_VALIDN:
2000 case CTSF_LAYERCOLPCT_VALIDN:
2001 case CTSF_ROWPCT_TOTALN:
2002 case CTSF_COLPCT_TOTALN:
2003 case CTSF_TABLEPCT_TOTALN:
2004 case CTSF_SUBTABLEPCT_TOTALN:
2005 case CTSF_LAYERPCT_TOTALN:
2006 case CTSF_LAYERROWPCT_TOTALN:
2007 case CTSF_LAYERCOLPCT_TOTALN:
2012 return s->valid + s->missing;
2025 return s->max != SYSMIS && s->min != SYSMIS ? s->max - s->min : SYSMIS;
2030 moments1_calculate (s->moments, NULL, &mean, NULL, NULL, NULL);
2036 double weight, variance;
2037 moments1_calculate (s->moments, &weight, NULL, &variance, NULL, NULL);
2038 return calc_semean (variance, weight);
2044 moments1_calculate (s->moments, NULL, NULL, &variance, NULL, NULL);
2045 return variance != SYSMIS ? sqrt (variance) : SYSMIS;
2050 double weight, mean;
2051 moments1_calculate (s->moments, &weight, &mean, NULL, NULL, NULL);
2052 return weight != SYSMIS && mean != SYSMIS ? weight * mean : SYSMIS;
2058 moments1_calculate (s->moments, NULL, NULL, &variance, NULL, NULL);
2062 case CTSF_ROWPCT_SUM:
2063 case CTSF_COLPCT_SUM:
2064 case CTSF_TABLEPCT_SUM:
2065 case CTSF_SUBTABLEPCT_SUM:
2066 case CTSF_LAYERPCT_SUM:
2067 case CTSF_LAYERROWPCT_SUM:
2068 case CTSF_LAYERCOLPCT_SUM:
2077 case CTSF_RESPONSES:
2078 case CTSF_ROWPCT_RESPONSES:
2079 case CTSF_COLPCT_RESPONSES:
2080 case CTSF_TABLEPCT_RESPONSES:
2081 case CTSF_SUBTABLEPCT_RESPONSES:
2082 case CTSF_LAYERPCT_RESPONSES:
2083 case CTSF_LAYERROWPCT_RESPONSES:
2084 case CTSF_LAYERCOLPCT_RESPONSES:
2085 case CTSF_ROWPCT_RESPONSES_COUNT:
2086 case CTSF_COLPCT_RESPONSES_COUNT:
2087 case CTSF_TABLEPCT_RESPONSES_COUNT:
2088 case CTSF_SUBTABLEPCT_RESPONSES_COUNT:
2089 case CTSF_LAYERPCT_RESPONSES_COUNT:
2090 case CTSF_LAYERROWPCT_RESPONSES_COUNT:
2091 case CTSF_LAYERCOLPCT_RESPONSES_COUNT:
2092 case CTSF_ROWPCT_COUNT_RESPONSES:
2093 case CTSF_COLPCT_COUNT_RESPONSES:
2094 case CTSF_TABLEPCT_COUNT_RESPONSES:
2095 case CTSF_SUBTABLEPCT_COUNT_RESPONSES:
2096 case CTSF_LAYERPCT_COUNT_RESPONSES:
2097 case CTSF_LAYERROWPCT_COUNT_RESPONSES:
2098 case CTSF_LAYERCOLPCT_COUNT_RESPONSES:
2105 struct ctables_cell_sort_aux
2107 const struct ctables_table *t;
2108 enum pivot_axis_type a;
2112 ctables_cell_compare_3way (const void *a_, const void *b_, const void *aux_)
2114 const struct ctables_cell_sort_aux *aux = aux_;
2115 struct ctables_cell *const *ap = a_;
2116 struct ctables_cell *const *bp = b_;
2117 const struct ctables_cell *a = *ap;
2118 const struct ctables_cell *b = *bp;
2120 size_t a_idx = a->axes[aux->a].stack_idx;
2121 size_t b_idx = b->axes[aux->a].stack_idx;
2123 return a_idx < b_idx ? -1 : 1;
2125 const struct ctables_nest *nest = &aux->t->stacks[aux->a].nests[a_idx];
2126 for (size_t i = 0; i < nest->n; i++)
2127 if (i != nest->scale_idx)
2129 const struct variable *var = nest->vars[i];
2130 const struct ctables_cell_value *a_cv = &a->axes[aux->a].cvs[i];
2131 const struct ctables_cell_value *b_cv = &b->axes[aux->a].cvs[i];
2132 if (a_cv->category != b_cv->category)
2133 return a_cv->category > b_cv->category ? 1 : -1;
2135 const union value *a_val = &a_cv->value;
2136 const union value *b_val = &b_cv->value;
2137 switch (a_cv->category->type)
2144 /* Must be equal. */
2151 int cmp = value_compare_3way (a_val, b_val, var_get_width (var));
2159 int cmp = value_compare_3way (a_val, b_val, var_get_width (var));
2161 return a_cv->category->sort_ascending ? cmp : -cmp;
2167 const char *a_label = var_lookup_value_label (var, a_val);
2168 const char *b_label = var_lookup_value_label (var, b_val);
2170 ? (b_label ? strcmp (a_label, b_label) : 1)
2171 : (b_label ? -1 : value_compare_3way (
2172 a_val, b_val, var_get_width (var))));
2174 return a_cv->category->sort_ascending ? cmp : -cmp;
2188 For each ctables_table:
2189 For each combination of row vars:
2190 For each combination of column vars:
2191 For each combination of layer vars:
2193 Make a table of row values:
2194 Sort entries by row values
2195 Assign a 0-based index to each actual value
2196 Construct a dimension
2197 Make a table of column values
2198 Make a table of layer values
2200 Fill the table entry using the indexes from before.
2203 static struct ctables_domain *
2204 ctables_domain_insert (struct ctables_table *t, struct ctables_cell *cell,
2205 enum ctables_domain_type domain)
2208 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2210 size_t idx = cell->axes[a].stack_idx;
2211 const struct ctables_nest *nest = &t->stacks[a].nests[idx];
2212 hash = hash_int (idx, hash);
2213 for (size_t i = 0; i < nest->n_domains[domain]; i++)
2215 size_t v_idx = nest->domains[domain][i];
2216 hash = value_hash (&cell->axes[a].cvs[v_idx].value,
2217 var_get_width (nest->vars[v_idx]), hash);
2221 struct ctables_domain *d;
2222 HMAP_FOR_EACH_WITH_HASH (d, struct ctables_domain, node, hash, &t->domains[domain])
2224 const struct ctables_cell *df = d->example;
2225 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2227 size_t idx = cell->axes[a].stack_idx;
2228 if (idx != df->axes[a].stack_idx)
2231 const struct ctables_nest *nest = &t->stacks[a].nests[idx];
2232 for (size_t i = 0; i < nest->n_domains[domain]; i++)
2234 size_t v_idx = nest->domains[domain][i];
2235 if (!value_equal (&df->axes[a].cvs[v_idx].value,
2236 &cell->axes[a].cvs[v_idx].value,
2237 var_get_width (nest->vars[v_idx])))
2246 d = xmalloc (sizeof *d);
2247 *d = (struct ctables_domain) { .example = cell };
2248 hmap_insert (&t->domains[domain], &d->node, hash);
2252 static const struct ctables_category *
2253 ctables_categories_match (const struct ctables_categories *c,
2254 const union value *v, const struct variable *var)
2256 const struct ctables_category *othernm = NULL;
2257 for (size_t i = c->n_cats; i-- > 0; )
2259 const struct ctables_category *cat = &c->cats[i];
2263 if (cat->number == v->f)
2271 if ((cat->range[0] == -DBL_MAX || v->f >= cat->range[0])
2272 && (cat->range[1] == DBL_MAX || v->f <= cat->range[1]))
2277 if (var_is_value_missing (var, v))
2294 return (cat->include_missing || !var_is_value_missing (var, v) ? cat
2299 return var_is_value_missing (var, v) ? NULL : othernm;
2302 static const struct ctables_category *
2303 ctables_categories_total (const struct ctables_categories *c)
2305 const struct ctables_category *first = &c->cats[0];
2306 const struct ctables_category *last = &c->cats[c->n_cats - 1];
2307 return (first->type == CCT_TOTAL ? first
2308 : last->type == CCT_TOTAL ? last
2312 static struct ctables_cell *
2313 ctables_cell_insert__ (struct ctables_table *t, const struct ccase *c,
2314 size_t ix[PIVOT_N_AXES],
2315 const struct ctables_category *cats[PIVOT_N_AXES][10])
2317 const struct ctables_nest *ss = &t->stacks[t->summary_axis].nests[ix[t->summary_axis]];
2320 enum ctables_summary_variant sv = CSV_CELL;
2321 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2323 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2324 hash = hash_int (ix[a], hash);
2325 for (size_t i = 0; i < nest->n; i++)
2326 if (i != nest->scale_idx)
2328 hash = hash_pointer (cats[a][i], hash);
2329 if (cats[a][i]->type != CCT_TOTAL
2330 && cats[a][i]->type != CCT_SUBTOTAL
2331 && cats[a][i]->type != CCT_HSUBTOTAL)
2332 hash = value_hash (case_data (c, nest->vars[i]),
2333 var_get_width (nest->vars[i]), hash);
2339 struct ctables_cell *cell;
2340 HMAP_FOR_EACH_WITH_HASH (cell, struct ctables_cell, node, hash, &t->cells)
2342 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2344 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2345 if (cell->axes[a].stack_idx != ix[a])
2347 for (size_t i = 0; i < nest->n; i++)
2348 if (i != nest->scale_idx
2349 && (cats[a][i] != cell->axes[a].cvs[i].category
2350 || (cats[a][i]->type != CCT_TOTAL
2351 && cats[a][i]->type != CCT_SUBTOTAL
2352 && cats[a][i]->type != CCT_HSUBTOTAL
2353 && !value_equal (case_data (c, nest->vars[i]),
2354 &cell->axes[a].cvs[i].value,
2355 var_get_width (nest->vars[i])))))
2364 cell = xmalloc (sizeof *cell);
2367 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2369 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2370 cell->axes[a].stack_idx = ix[a];
2371 cell->axes[a].cvs = (nest->n
2372 ? xnmalloc (nest->n, sizeof *cell->axes[a].cvs)
2374 for (size_t i = 0; i < nest->n; i++)
2376 if (i != nest->scale_idx)
2378 const struct ctables_category *subtotal = cats[a][i]->subtotal;
2379 if (subtotal && subtotal->type == CCT_HSUBTOTAL)
2383 cell->axes[a].cvs[i].category = cats[a][i];
2384 value_clone (&cell->axes[a].cvs[i].value, case_data (c, nest->vars[i]),
2385 var_get_width (nest->vars[i]));
2389 const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv];
2390 cell->summaries = xmalloc (specs->n * sizeof *cell->summaries);
2391 for (size_t i = 0; i < specs->n; i++)
2392 ctables_summary_init (&cell->summaries[i], &specs->specs[i]);
2393 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
2394 cell->domains[dt] = ctables_domain_insert (t, cell, dt);
2395 hmap_insert (&t->cells, &cell->node, hash);
2400 ctables_cell_add__ (struct ctables_table *t, const struct ccase *c,
2401 size_t ix[PIVOT_N_AXES],
2402 const struct ctables_category *cats[PIVOT_N_AXES][10],
2405 struct ctables_cell *cell = ctables_cell_insert__ (t, c, ix, cats);
2406 const struct ctables_nest *ss = &t->stacks[t->summary_axis].nests[ix[t->summary_axis]];
2408 const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv];
2409 for (size_t i = 0; i < specs->n; i++)
2410 ctables_summary_add (&cell->summaries[i], &specs->specs[i], specs->var,
2411 case_data (c, specs->var), weight);
2412 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
2413 cell->domains[dt]->valid += weight;
2417 recurse_totals (struct ctables_table *t, const struct ccase *c,
2418 size_t ix[PIVOT_N_AXES],
2419 const struct ctables_category *cats[PIVOT_N_AXES][10],
2421 enum pivot_axis_type start_axis, size_t start_nest)
2423 for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
2425 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2426 for (size_t i = start_nest; i < nest->n; i++)
2428 if (i == nest->scale_idx)
2431 const struct variable *var = nest->vars[i];
2433 const struct ctables_category *total = ctables_categories_total (
2434 t->categories[var_get_dict_index (var)]);
2437 const struct ctables_category *save = cats[a][i];
2439 ctables_cell_add__ (t, c, ix, cats, weight);
2440 recurse_totals (t, c, ix, cats, weight, a, i + 1);
2449 ctables_cell_insert (struct ctables_table *t,
2450 const struct ccase *c,
2451 size_t ir, size_t ic, size_t il,
2454 size_t ix[PIVOT_N_AXES] = {
2455 [PIVOT_AXIS_ROW] = ir,
2456 [PIVOT_AXIS_COLUMN] = ic,
2457 [PIVOT_AXIS_LAYER] = il,
2460 const struct ctables_category *cats[PIVOT_N_AXES][10];
2461 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2463 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2464 for (size_t i = 0; i < nest->n; i++)
2466 if (i == nest->scale_idx)
2469 const struct variable *var = nest->vars[i];
2470 const union value *value = case_data (c, var);
2472 if (var_is_numeric (var) && value->f == SYSMIS)
2475 cats[a][i] = ctables_categories_match (
2476 t->categories[var_get_dict_index (var)], value, var);
2482 ctables_cell_add__ (t, c, ix, cats, weight);
2484 recurse_totals (t, c, ix, cats, weight, 0, 0);
2486 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2488 const struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
2489 for (size_t i = 0; i < nest->n; i++)
2491 if (i == nest->scale_idx)
2494 const struct ctables_category *save = cats[a][i];
2497 cats[a][i] = save->subtotal;
2498 ctables_cell_add__ (t, c, ix, cats, weight);
2507 const struct ctables_summary_spec_set *set;
2512 merge_item_compare_3way (const struct merge_item *a, const struct merge_item *b)
2514 const struct ctables_summary_spec *as = &a->set->specs[a->ofs];
2515 const struct ctables_summary_spec *bs = &b->set->specs[b->ofs];
2516 if (as->function != bs->function)
2517 return as->function > bs->function ? 1 : -1;
2518 else if (as->percentile != bs->percentile)
2519 return as->percentile < bs->percentile ? 1 : -1;
2520 return strcmp (as->label, bs->label);
2523 static struct pivot_value *
2524 ctables_category_create_label (const struct ctables_category *cat,
2525 const struct variable *var,
2526 const union value *value)
2528 return (cat->type == CCT_TOTAL || cat->type == CCT_SUBTOTAL || cat->type == CCT_HSUBTOTAL
2529 ? pivot_value_new_user_text (cat->total_label, SIZE_MAX)
2530 : pivot_value_new_var_value (var, value));
2534 ctables_table_output_different_axis (struct ctables *ct, struct ctables_table *t)
2536 struct pivot_table *pt = pivot_table_create__ (
2538 ? pivot_value_new_user_text (t->title, SIZE_MAX)
2539 : pivot_value_new_text (N_("Custom Tables"))),
2542 pivot_table_set_caption (
2543 pt, pivot_value_new_user_text (t->caption, SIZE_MAX));
2545 pivot_table_set_caption (
2546 pt, pivot_value_new_user_text (t->corner, SIZE_MAX));
2548 if (t->summary_axis != t->slabels_axis)
2550 struct pivot_dimension *d = pivot_dimension_create (
2551 pt, t->slabels_axis, N_("Summaries"));
2552 const struct ctables_summary_spec_set *specs = &t->summary_specs;
2553 for (size_t i = 0; i < specs->n; i++)
2554 pivot_category_create_leaf (
2555 d->root, pivot_value_new_text (specs->specs[i].label));
2558 if (t->clabels_example)
2560 struct pivot_dimension *d = pivot_dimension_create (
2561 pt, t->label_axis[t->clabels_from_axis],
2562 t->clabels_from_axis == PIVOT_AXIS_ROW
2563 ? N_("Row Categories")
2564 : N_("Column Categories"));
2565 const struct variable *var = t->clabels_example;
2566 const struct ctables_categories *c = t->categories[var_get_dict_index (var)];
2567 for (size_t i = 0; i < t->n_clabels_values; i++)
2569 const union value *value = &t->clabels_values[i];
2570 const struct ctables_category *cat = ctables_categories_match (c, value, var);
2573 /* XXX probably missing */
2576 pivot_category_create_leaf (d->root, ctables_category_create_label (
2577 cat, t->clabels_example, value));
2581 pivot_table_set_look (pt, ct->look);
2582 struct pivot_dimension *d[PIVOT_N_AXES];
2583 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2585 static const char *names[] = {
2586 [PIVOT_AXIS_ROW] = N_("Rows"),
2587 [PIVOT_AXIS_COLUMN] = N_("Columns"),
2588 [PIVOT_AXIS_LAYER] = N_("Layers"),
2590 d[a] = (t->axes[a] || a == t->summary_axis
2591 ? pivot_dimension_create (pt, a, names[a])
2596 assert (t->axes[a]);
2598 struct ctables_cell **sorted = xnmalloc (t->cells.count, sizeof *sorted);
2599 size_t n_sorted = 0;
2601 struct ctables_cell *cell;
2602 HMAP_FOR_EACH (cell, struct ctables_cell, node, &t->cells)
2604 sorted[n_sorted++] = cell;
2605 assert (n_sorted <= t->cells.count);
2607 struct ctables_cell_sort_aux aux = { .t = t, .a = a };
2608 sort (sorted, n_sorted, sizeof *sorted, ctables_cell_compare_3way, &aux);
2610 size_t max_depth = 0;
2611 for (size_t j = 0; j < t->stacks[a].n; j++)
2612 if (t->stacks[a].nests[j].n > max_depth)
2613 max_depth = t->stacks[a].nests[j].n;
2615 /* Pivot categories:
2617 - variable label for nest->vars[0], if vlabel != CTVL_NONE
2618 - category for nest->vars[0], if nest->scale_idx != 0
2619 - variable label for nest->vars[1], if vlabel != CTVL_NONE
2620 - category for nest->vars[1], if nest->scale_idx != 1
2622 - variable label for nest->vars[n - 1], if vlabel != CTVL_NONE
2623 - category for nest->vars[n - 1], if t->label_axis[a] == a && nest->scale_idx != n - 1.
2624 - summary function, if 'a == t->slabels_axis && a ==
2627 Additional dimensions:
2629 - If 'a == t->slabels_axis && a != t->summary_axis', add a summary
2631 - If 't->label_axis[b] == a' for some 'b != a', add a category
2635 struct ctables_level
2637 enum ctables_level_type
2639 CTL_VAR, /* Variable label for nest->vars[var_idx]. */
2640 CTL_CATEGORY, /* Category for nest->vars[var_idx]. */
2641 CTL_SUMMARY, /* Summary functions. */
2647 struct ctables_level *levels = xnmalloc (1 + 2 * max_depth, sizeof *levels);
2648 size_t n_levels = 0;
2650 struct pivot_category **groups = xnmalloc (1 + 2 * max_depth, sizeof *groups);
2652 for (size_t j = 0; j < n_sorted; j++)
2654 struct ctables_cell *cell = sorted[j];
2655 struct ctables_cell *prev = j > 0 ? sorted[j - 1] : NULL;
2656 const struct ctables_nest *nest = &t->stacks[a].nests[cell->axes[a].stack_idx];
2658 bool new_subtable = !prev || prev->axes[a].stack_idx != cell->axes[a].stack_idx;
2662 printf ("%s levels:", pivot_axis_type_to_string (a));
2663 for (size_t k = 0; k < nest->n; k++)
2665 enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (nest->vars[k])];
2666 if (vlabel != CTVL_NONE)
2668 printf (" var(%s)", var_get_name (nest->vars[k]));
2669 levels[n_levels++] = (struct ctables_level) {
2675 if (nest->scale_idx != k
2676 && (k != nest->n - 1 || t->label_axis[a] == a))
2678 printf (" category(%s)", var_get_name (nest->vars[k]));
2679 levels[n_levels++] = (struct ctables_level) {
2680 .type = CTL_CATEGORY,
2686 if (a == t->slabels_axis && a == t->summary_axis)
2688 printf (" summary");
2689 levels[n_levels++] = (struct ctables_level) {
2690 .type = CTL_SUMMARY,
2691 .var_idx = SIZE_MAX,
2697 size_t n_common = 0;
2700 for (; n_common < n_levels; n_common++)
2702 const struct ctables_level *level = &levels[n_common];
2703 if (level->type == CTL_CATEGORY)
2705 size_t var_idx = level->var_idx;
2706 if (prev->axes[a].cvs[var_idx].category
2707 != cell->axes[a].cvs[var_idx].category)
2711 else if (!value_equal (&prev->axes[a].cvs[var_idx].value,
2712 &cell->axes[a].cvs[var_idx].value,
2713 var_get_type (nest->vars[var_idx])))
2721 for (size_t k = n_common; k < n_levels; k++)
2723 const struct ctables_level *level = &levels[k];
2724 struct pivot_category *parent = k ? groups[k - 1] : d[a]->root;
2725 if (level->type == CTL_SUMMARY)
2727 const struct ctables_summary_spec_set *specs = &t->summary_specs;
2728 for (size_t m = 0; m < specs->n; m++)
2729 pivot_category_create_leaf (
2730 parent, pivot_value_new_text (specs->specs[m].label));
2734 const struct variable *var = nest->vars[level->var_idx];
2735 struct pivot_value *label;
2736 if (level->type == CTL_VAR)
2737 label = pivot_value_new_variable (var);
2738 else if (level->type == CTL_CATEGORY)
2740 const struct ctables_cell_value *cv = &cell->axes[a].cvs[level->var_idx];
2741 label = ctables_category_create_label (cv->category,
2747 if (k == n_levels - 1)
2748 pivot_category_create_leaf (parent, label);
2750 groups[k] = pivot_category_create_group__ (parent, label);
2754 cell->axes[a].leaf = prev_leaf;
2759 pivot_table_submit (pt);
2764 ctables_prepare_table (struct ctables_table *t)
2766 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2769 t->stacks[a] = enumerate_fts (a, t->axes[a]);
2771 for (size_t j = 0; j < t->stacks[a].n; j++)
2773 struct ctables_nest *nest = &t->stacks[a].nests[j];
2774 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
2776 nest->domains[dt] = xmalloc (nest->n * sizeof *nest->domains[dt]);
2777 nest->n_domains[dt] = 0;
2779 for (size_t k = 0; k < nest->n; k++)
2781 if (k == nest->scale_idx)
2790 if (a != PIVOT_AXIS_LAYER)
2797 if (dt == CTDT_SUBTABLE ? a != PIVOT_AXIS_LAYER
2798 : dt == CTDT_ROW ? a == PIVOT_AXIS_COLUMN
2799 : a == PIVOT_AXIS_ROW)
2801 if (k == nest->n - 1
2802 || (nest->scale_idx == nest->n - 1
2803 && k == nest->n - 2))
2809 if (a == PIVOT_AXIS_COLUMN)
2814 if (a == PIVOT_AXIS_ROW)
2819 nest->domains[dt][nest->n_domains[dt]++] = k;
2826 struct ctables_nest *nest = xmalloc (sizeof *nest);
2827 *nest = (struct ctables_nest) { .n = 0 };
2828 t->stacks[a] = (struct ctables_stack) { .nests = nest, .n = 1 };
2831 struct ctables_stack *stack = &t->stacks[t->summary_axis];
2832 for (size_t i = 0; i < stack->n; i++)
2834 struct ctables_nest *nest = &stack->nests[i];
2835 if (!nest->specs[CSV_CELL].n)
2837 struct ctables_summary_spec_set *specs = &nest->specs[CSV_CELL];
2838 specs->specs = xmalloc (sizeof *specs->specs);
2841 enum ctables_summary_function function
2842 = specs->var ? CTSF_MEAN : CTSF_COUNT;
2843 struct ctables_var var = { .is_mrset = false, .var = specs->var };
2845 *specs->specs = (struct ctables_summary_spec) {
2846 .function = function,
2847 .format = ctables_summary_default_format (function, &var),
2848 .label = ctables_summary_default_label (function, 0),
2851 specs->var = nest->vars[0];
2853 ctables_summary_spec_set_clone (&nest->specs[CSV_TOTAL],
2854 &nest->specs[CSV_CELL]);
2856 else if (!nest->specs[CSV_TOTAL].n)
2857 ctables_summary_spec_set_clone (&nest->specs[CSV_TOTAL],
2858 &nest->specs[CSV_CELL]);
2861 struct ctables_summary_spec_set *merged = &t->summary_specs;
2862 struct merge_item *items = xnmalloc (2 * stack->n, sizeof *items);
2864 for (size_t j = 0; j < stack->n; j++)
2866 const struct ctables_nest *nest = &stack->nests[j];
2868 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
2869 items[n_left++] = (struct merge_item) { .set = &nest->specs[sv] };
2874 struct merge_item min = items[0];
2875 for (size_t j = 1; j < n_left; j++)
2876 if (merge_item_compare_3way (&items[j], &min) < 0)
2879 if (merged->n >= merged->allocated)
2880 merged->specs = x2nrealloc (merged->specs, &merged->allocated,
2881 sizeof *merged->specs);
2882 merged->specs[merged->n++] = min.set->specs[min.ofs];
2884 for (size_t j = 0; j < n_left; )
2886 if (merge_item_compare_3way (&items[j], &min) == 0)
2888 struct merge_item *item = &items[j];
2889 item->set->specs[item->ofs].axis_idx = merged->n - 1;
2890 if (++item->ofs >= item->set->n)
2892 items[j] = items[--n_left];
2901 for (size_t j = 0; j < merged->n; j++)
2902 printf ("%s\n", ctables_summary_function_name (merged->specs[j].function));
2904 for (size_t j = 0; j < stack->n; j++)
2906 const struct ctables_nest *nest = &stack->nests[j];
2907 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
2909 const struct ctables_summary_spec_set *specs = &nest->specs[sv];
2910 for (size_t k = 0; k < specs->n; k++)
2911 printf ("(%s, %zu) ", ctables_summary_function_name (specs->specs[k].function),
2912 specs->specs[k].axis_idx);
2920 ctables_insert_clabels_values (struct ctables_table *t, const struct ccase *c,
2921 enum pivot_axis_type a)
2923 struct ctables_stack *stack = &t->stacks[a];
2924 for (size_t i = 0; i < stack->n; i++)
2926 const struct ctables_nest *nest = &stack->nests[i];
2927 const struct variable *v = nest->vars[nest->n - 1];
2928 int width = var_get_width (v);
2929 const union value *value = case_data (c, v);
2930 unsigned int hash = value_hash (value, width, 0);
2932 struct ctables_value *clv;
2933 HMAP_FOR_EACH_WITH_HASH (clv, struct ctables_value, node, hash,
2934 &t->clabels_values_map)
2935 if (value_equal (value, &clv->value, width))
2938 clv = xmalloc (sizeof *clv);
2939 value_clone (&clv->value, value, width);
2940 hmap_insert (&t->clabels_values_map, &clv->node, hash);
2947 compare_clabels_values_3way (const void *a_, const void *b_, const void *width_)
2949 const union value *a = a_;
2950 const union value *b = b_;
2951 const int *width = width_;
2952 return value_compare_3way (a, b, *width);
2956 ctables_sort_clabels_values (struct ctables_table *t)
2958 int width = var_get_width (t->clabels_example);
2960 size_t n = hmap_count (&t->clabels_values_map);
2961 t->clabels_values = xnmalloc (n, sizeof *t->clabels_values);
2963 const struct ctables_value *clv;
2965 HMAP_FOR_EACH (clv, struct ctables_value, node, &t->clabels_values_map)
2966 t->clabels_values[i++] = clv->value;
2967 t->n_clabels_values = n;
2970 sort (t->clabels_values, n, sizeof *t->clabels_values,
2971 compare_clabels_values_3way, &width);
2975 ctables_execute (struct dataset *ds, struct ctables *ct)
2977 struct casereader *input = casereader_create_filter_weight (proc_open (ds),
2980 bool warn_on_invalid = true;
2981 double total_weight = 0;
2982 for (struct ccase *c = casereader_read (input); c;
2983 case_unref (c), c = casereader_read (input))
2985 double weight = dict_get_case_weight (dataset_dict (ds), c,
2987 total_weight += weight;
2989 for (size_t i = 0; i < ct->n_tables; i++)
2991 struct ctables_table *t = ct->tables[i];
2993 for (size_t ir = 0; ir < t->stacks[PIVOT_AXIS_ROW].n; ir++)
2994 for (size_t ic = 0; ic < t->stacks[PIVOT_AXIS_COLUMN].n; ic++)
2995 for (size_t il = 0; il < t->stacks[PIVOT_AXIS_LAYER].n; il++)
2996 ctables_cell_insert (t, c, ir, ic, il, weight);
2998 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2999 if (t->label_axis[a] != a)
3000 ctables_insert_clabels_values (t, c, a);
3003 casereader_destroy (input);
3005 for (size_t i = 0; i < ct->n_tables; i++)
3007 struct ctables_table *t = ct->tables[i];
3009 if (t->clabels_example)
3010 ctables_sort_clabels_values (t);
3012 ctables_table_output_different_axis (ct, ct->tables[i]);
3014 return proc_commit (ds);
3018 ctables_check_label_position (struct ctables_table *t, enum pivot_axis_type a)
3020 enum pivot_axis_type label_pos = t->label_axis[a];
3024 t->clabels_from_axis = a;
3026 const char *subcommand_name = a == PIVOT_AXIS_ROW ? "ROWLABELS" : "COLLABELS";
3027 const char *pos_name = label_pos == PIVOT_AXIS_LAYER ? "LAYER" : "OPPOSITE";
3029 const struct ctables_stack *stack = &t->stacks[a];
3033 const struct ctables_nest *n0 = &stack->nests[0];
3035 const struct variable *v0 = n0->vars[n0->n - 1];
3036 struct ctables_categories *c0 = t->categories[var_get_dict_index (v0)];
3037 t->clabels_example = v0;
3039 for (size_t i = 0; i < c0->n_cats; i++)
3040 if (c0->cats[i].type == CCT_FUNCTION)
3042 msg (SE, _("%s=%s is not allowed with sorting based "
3043 "on a summary function."),
3044 subcommand_name, pos_name);
3047 if (n0->n - 1 == n0->scale_idx)
3049 msg (SE, _("%s=%s requires the variables to be moved to be categorical, "
3050 "but %s is a scale variable."),
3051 subcommand_name, pos_name, var_get_name (v0));
3055 for (size_t i = 1; i < stack->n; i++)
3057 const struct ctables_nest *ni = &stack->nests[i];
3059 const struct variable *vi = ni->vars[ni->n - 1];
3060 struct ctables_categories *ci = t->categories[var_get_dict_index (vi)];
3062 if (ni->n - 1 == ni->scale_idx)
3064 msg (SE, _("%s=%s requires the variables to be moved to be "
3065 "categorical, but %s is a scale variable."),
3066 subcommand_name, pos_name, var_get_name (vi));
3069 if (var_get_width (v0) != var_get_width (vi))
3071 msg (SE, _("%s=%s requires the variables to be "
3072 "moved to have the same width, but %s has "
3073 "width %d and %s has width %d."),
3074 subcommand_name, pos_name,
3075 var_get_name (v0), var_get_width (v0),
3076 var_get_name (vi), var_get_width (vi));
3079 if (!val_labs_equal (var_get_value_labels (v0),
3080 var_get_value_labels (vi)))
3082 msg (SE, _("%s=%s requires the variables to be "
3083 "moved to have the same value labels, but %s "
3084 "and %s have different value labels."),
3085 subcommand_name, pos_name,
3086 var_get_name (v0), var_get_name (vi));
3089 if (!ctables_categories_equal (c0, ci))
3091 msg (SE, _("%s=%s requires the variables to be "
3092 "moved to have the same category "
3093 "specifications, but %s and %s have different "
3094 "category specifications."),
3095 subcommand_name, pos_name,
3096 var_get_name (v0), var_get_name (vi));
3105 cmd_ctables (struct lexer *lexer, struct dataset *ds)
3107 size_t n_vars = dict_get_n_vars (dataset_dict (ds));
3108 enum ctables_vlabel *vlabels = xnmalloc (n_vars, sizeof *vlabels);
3109 enum settings_value_show tvars = settings_get_show_variables ();
3110 for (size_t i = 0; i < n_vars; i++)
3111 vlabels[i] = (enum ctables_vlabel) tvars;
3113 struct ctables *ct = xmalloc (sizeof *ct);
3114 *ct = (struct ctables) {
3115 .look = pivot_table_look_unshare (pivot_table_look_ref (
3116 pivot_table_look_get_default ())),
3118 .hide_threshold = 5,
3120 ct->look->omit_empty = false;
3122 if (!lex_force_match (lexer, T_SLASH))
3125 while (!lex_match_id (lexer, "TABLE"))
3127 if (lex_match_id (lexer, "FORMAT"))
3129 double widths[2] = { SYSMIS, SYSMIS };
3130 double units_per_inch = 72.0;
3132 while (lex_token (lexer) != T_SLASH)
3134 if (lex_match_id (lexer, "MINCOLWIDTH"))
3136 if (!parse_col_width (lexer, "MINCOLWIDTH", &widths[0]))
3139 else if (lex_match_id (lexer, "MAXCOLWIDTH"))
3141 if (!parse_col_width (lexer, "MAXCOLWIDTH", &widths[1]))
3144 else if (lex_match_id (lexer, "UNITS"))
3146 lex_match (lexer, T_EQUALS);
3147 if (lex_match_id (lexer, "POINTS"))
3148 units_per_inch = 72.0;
3149 else if (lex_match_id (lexer, "INCHES"))
3150 units_per_inch = 1.0;
3151 else if (lex_match_id (lexer, "CM"))
3152 units_per_inch = 2.54;
3155 lex_error_expecting (lexer, "POINTS", "INCHES", "CM");
3159 else if (lex_match_id (lexer, "EMPTY"))
3164 lex_match (lexer, T_EQUALS);
3165 if (lex_match_id (lexer, "ZERO"))
3167 /* Nothing to do. */
3169 else if (lex_match_id (lexer, "BLANK"))
3170 ct->zero = xstrdup ("");
3171 else if (lex_force_string (lexer))
3173 ct->zero = ss_xstrdup (lex_tokss (lexer));
3179 else if (lex_match_id (lexer, "MISSING"))
3181 lex_match (lexer, T_EQUALS);
3182 if (!lex_force_string (lexer))
3186 ct->missing = (strcmp (lex_tokcstr (lexer), ".")
3187 ? ss_xstrdup (lex_tokss (lexer))
3193 lex_error_expecting (lexer, "MINCOLWIDTH", "MAXCOLWIDTH",
3194 "UNITS", "EMPTY", "MISSING");
3199 if (widths[0] != SYSMIS && widths[1] != SYSMIS
3200 && widths[0] > widths[1])
3202 msg (SE, _("MINCOLWIDTH must not be greater than MAXCOLWIDTH."));
3206 for (size_t i = 0; i < 2; i++)
3207 if (widths[i] != SYSMIS)
3209 int *wr = ct->look->width_ranges[TABLE_HORZ];
3210 wr[i] = widths[i] / units_per_inch * 96.0;
3215 else if (lex_match_id (lexer, "VLABELS"))
3217 if (!lex_force_match_id (lexer, "VARIABLES"))
3219 lex_match (lexer, T_EQUALS);
3221 struct variable **vars;
3223 if (!parse_variables (lexer, dataset_dict (ds), &vars, &n_vars,
3227 if (!lex_force_match_id (lexer, "DISPLAY"))
3232 lex_match (lexer, T_EQUALS);
3234 enum ctables_vlabel vlabel;
3235 if (lex_match_id (lexer, "DEFAULT"))
3236 vlabel = (enum ctables_vlabel) settings_get_show_variables ();
3237 else if (lex_match_id (lexer, "NAME"))
3239 else if (lex_match_id (lexer, "LABEL"))
3240 vlabel = CTVL_LABEL;
3241 else if (lex_match_id (lexer, "BOTH"))
3243 else if (lex_match_id (lexer, "NONE"))
3247 lex_error_expecting (lexer, "DEFAULT", "NAME", "LABEL",
3253 for (size_t i = 0; i < n_vars; i++)
3254 ct->vlabels[var_get_dict_index (vars[i])] = vlabel;
3257 else if (lex_match_id (lexer, "MRSETS"))
3259 if (!lex_force_match_id (lexer, "COUNTDUPLICATES"))
3261 lex_match (lexer, T_EQUALS);
3262 if (!parse_bool (lexer, &ct->mrsets_count_duplicates))
3265 else if (lex_match_id (lexer, "SMISSING"))
3267 if (lex_match_id (lexer, "VARIABLE"))
3268 ct->smissing_listwise = false;
3269 else if (lex_match_id (lexer, "LISTWISE"))
3270 ct->smissing_listwise = true;
3273 lex_error_expecting (lexer, "VARIABLE", "LISTWISE");
3278 else if (lex_match_id (lexer, "WEIGHT"))
3280 if (!lex_force_match_id (lexer, "VARIABLE"))
3282 lex_match (lexer, T_EQUALS);
3283 ct->base_weight = parse_variable (lexer, dataset_dict (ds));
3284 if (!ct->base_weight)
3287 else if (lex_match_id (lexer, "HIDESMALLCOUNTS"))
3289 if (!lex_force_match_id (lexer, "COUNT"))
3291 lex_match (lexer, T_EQUALS);
3292 if (!lex_force_int_range (lexer, "HIDESMALLCOUNTS COUNT", 2, INT_MAX))
3294 ct->hide_threshold = lex_integer (lexer);
3299 lex_error_expecting (lexer, "FORMAT", "VLABELS", "MRSETS",
3300 "SMISSING", "PCOMPUTE", "PPROPERTIES",
3301 "WEIGHT", "HIDESMALLCOUNTS", "TABLE");
3305 if (!lex_force_match (lexer, T_SLASH))
3309 size_t allocated_tables = 0;
3312 if (ct->n_tables >= allocated_tables)
3313 ct->tables = x2nrealloc (ct->tables, &allocated_tables,
3314 sizeof *ct->tables);
3316 struct ctables_category *cat = xmalloc (sizeof *cat);
3317 *cat = (struct ctables_category) {
3319 .include_missing = false,
3320 .sort_ascending = true,
3323 struct ctables_categories *c = xmalloc (sizeof *c);
3324 size_t n_vars = dict_get_n_vars (dataset_dict (ds));
3325 *c = (struct ctables_categories) {
3331 struct ctables_categories **categories = xnmalloc (n_vars,
3332 sizeof *categories);
3333 for (size_t i = 0; i < n_vars; i++)
3336 struct ctables_table *t = xmalloc (sizeof *t);
3337 *t = (struct ctables_table) {
3338 .cells = HMAP_INITIALIZER (t->cells),
3339 .slabels_axis = PIVOT_AXIS_COLUMN,
3340 .slabels_visible = true,
3341 .clabels_values_map = HMAP_INITIALIZER (t->clabels_values_map),
3343 [PIVOT_AXIS_ROW] = PIVOT_AXIS_ROW,
3344 [PIVOT_AXIS_COLUMN] = PIVOT_AXIS_COLUMN,
3345 [PIVOT_AXIS_LAYER] = PIVOT_AXIS_LAYER,
3347 .clabels_from_axis = PIVOT_AXIS_LAYER,
3348 .categories = categories,
3349 .n_categories = n_vars,
3352 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
3353 hmap_init (&t->domains[dt]);
3354 ct->tables[ct->n_tables++] = t;
3356 lex_match (lexer, T_EQUALS);
3357 if (!ctables_axis_parse (lexer, dataset_dict (ds), ct, t, PIVOT_AXIS_ROW))
3359 if (lex_match (lexer, T_BY))
3361 if (!ctables_axis_parse (lexer, dataset_dict (ds),
3362 ct, t, PIVOT_AXIS_COLUMN))
3365 if (lex_match (lexer, T_BY))
3367 if (!ctables_axis_parse (lexer, dataset_dict (ds),
3368 ct, t, PIVOT_AXIS_LAYER))
3373 if (!t->axes[PIVOT_AXIS_ROW] && !t->axes[PIVOT_AXIS_COLUMN]
3374 && !t->axes[PIVOT_AXIS_LAYER])
3376 lex_error (lexer, _("At least one variable must be specified."));
3380 const struct ctables_axis *scales[PIVOT_N_AXES];
3381 size_t n_scales = 0;
3382 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3384 scales[a] = find_scale (t->axes[a]);
3390 msg (SE, _("Scale variables may appear only on one axis."));
3391 if (scales[PIVOT_AXIS_ROW])
3392 msg_at (SN, scales[PIVOT_AXIS_ROW]->loc,
3393 _("This scale variable appears on the rows axis."));
3394 if (scales[PIVOT_AXIS_COLUMN])
3395 msg_at (SN, scales[PIVOT_AXIS_COLUMN]->loc,
3396 _("This scale variable appears on the columns axis."));
3397 if (scales[PIVOT_AXIS_LAYER])
3398 msg_at (SN, scales[PIVOT_AXIS_LAYER]->loc,
3399 _("This scale variable appears on the layer axis."));
3403 const struct ctables_axis *summaries[PIVOT_N_AXES];
3404 size_t n_summaries = 0;
3405 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3407 summaries[a] = (scales[a]
3409 : find_categorical_summary_spec (t->axes[a]));
3413 if (n_summaries > 1)
3415 msg (SE, _("Summaries may appear only on one axis."));
3416 if (summaries[PIVOT_AXIS_ROW])
3417 msg_at (SN, summaries[PIVOT_AXIS_ROW]->loc,
3418 _("This variable on the rows axis has a summary."));
3419 if (summaries[PIVOT_AXIS_COLUMN])
3420 msg_at (SN, summaries[PIVOT_AXIS_COLUMN]->loc,
3421 _("This variable on the columns axis has a summary."));
3422 if (summaries[PIVOT_AXIS_LAYER])
3423 msg_at (SN, summaries[PIVOT_AXIS_LAYER]->loc,
3424 _("This variable on the layers axis has a summary."));
3427 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3428 if (n_summaries ? summaries[a] : t->axes[a])
3430 t->summary_axis = a;
3434 if (lex_token (lexer) == T_ENDCMD)
3436 if (!lex_force_match (lexer, T_SLASH))
3439 while (!lex_match_id (lexer, "TABLE") && lex_token (lexer) != T_ENDCMD)
3441 if (lex_match_id (lexer, "SLABELS"))
3443 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
3445 if (lex_match_id (lexer, "POSITION"))
3447 lex_match (lexer, T_EQUALS);
3448 if (lex_match_id (lexer, "COLUMN"))
3449 t->slabels_axis = PIVOT_AXIS_COLUMN;
3450 else if (lex_match_id (lexer, "ROW"))
3451 t->slabels_axis = PIVOT_AXIS_ROW;
3452 else if (lex_match_id (lexer, "LAYER"))
3453 t->slabels_axis = PIVOT_AXIS_LAYER;
3456 lex_error_expecting (lexer, "COLUMN", "ROW", "LAYER");
3460 else if (lex_match_id (lexer, "VISIBLE"))
3462 lex_match (lexer, T_EQUALS);
3463 if (!parse_bool (lexer, &t->slabels_visible))
3468 lex_error_expecting (lexer, "POSITION", "VISIBLE");
3473 else if (lex_match_id (lexer, "CLABELS"))
3475 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
3477 if (lex_match_id (lexer, "AUTO"))
3479 t->label_axis[PIVOT_AXIS_ROW] = PIVOT_AXIS_ROW;
3480 t->label_axis[PIVOT_AXIS_COLUMN] = PIVOT_AXIS_COLUMN;
3482 else if (lex_match_id (lexer, "ROWLABELS"))
3484 lex_match (lexer, T_EQUALS);
3485 if (lex_match_id (lexer, "OPPOSITE"))
3486 t->label_axis[PIVOT_AXIS_ROW] = PIVOT_AXIS_COLUMN;
3487 else if (lex_match_id (lexer, "LAYER"))
3488 t->label_axis[PIVOT_AXIS_ROW] = PIVOT_AXIS_LAYER;
3491 lex_error_expecting (lexer, "OPPOSITE", "LAYER");
3495 else if (lex_match_id (lexer, "COLLABELS"))
3497 lex_match (lexer, T_EQUALS);
3498 if (lex_match_id (lexer, "OPPOSITE"))
3499 t->label_axis[PIVOT_AXIS_COLUMN] = PIVOT_AXIS_ROW;
3500 else if (lex_match_id (lexer, "LAYER"))
3501 t->label_axis[PIVOT_AXIS_COLUMN] = PIVOT_AXIS_LAYER;
3504 lex_error_expecting (lexer, "OPPOSITE", "LAYER");
3510 lex_error_expecting (lexer, "AUTO", "ROWLABELS",
3516 else if (lex_match_id (lexer, "CRITERIA"))
3518 if (!lex_force_match_id (lexer, "CILEVEL"))
3520 lex_match (lexer, T_EQUALS);
3522 if (!lex_force_num_range_halfopen (lexer, "CILEVEL", 0, 100))
3524 t->cilevel = lex_number (lexer);
3527 else if (lex_match_id (lexer, "CATEGORIES"))
3529 if (!ctables_table_parse_categories (lexer, dataset_dict (ds), t))
3532 else if (lex_match_id (lexer, "TITLES"))
3537 if (lex_match_id (lexer, "CAPTION"))
3538 textp = &t->caption;
3539 else if (lex_match_id (lexer, "CORNER"))
3541 else if (lex_match_id (lexer, "TITLE"))
3545 lex_error_expecting (lexer, "CAPTION", "CORNER", "TITLE");
3548 lex_match (lexer, T_EQUALS);
3550 struct string s = DS_EMPTY_INITIALIZER;
3551 while (lex_is_string (lexer))
3553 if (!ds_is_empty (&s))
3554 ds_put_byte (&s, ' ');
3555 ds_put_substring (&s, lex_tokss (lexer));
3559 *textp = ds_steal_cstr (&s);
3561 while (lex_token (lexer) != T_SLASH
3562 && lex_token (lexer) != T_ENDCMD);
3564 else if (lex_match_id (lexer, "SIGTEST"))
3568 t->chisq = xmalloc (sizeof *t->chisq);
3569 *t->chisq = (struct ctables_chisq) {
3571 .include_mrsets = true,
3572 .all_visible = true,
3578 if (lex_match_id (lexer, "TYPE"))
3580 lex_match (lexer, T_EQUALS);
3581 if (!lex_force_match_id (lexer, "CHISQUARE"))
3584 else if (lex_match_id (lexer, "ALPHA"))
3586 lex_match (lexer, T_EQUALS);
3587 if (!lex_force_num_range_halfopen (lexer, "ALPHA", 0, 1))
3589 t->chisq->alpha = lex_number (lexer);
3592 else if (lex_match_id (lexer, "INCLUDEMRSETS"))
3594 lex_match (lexer, T_EQUALS);
3595 if (parse_bool (lexer, &t->chisq->include_mrsets))
3598 else if (lex_match_id (lexer, "CATEGORIES"))
3600 lex_match (lexer, T_EQUALS);
3601 if (lex_match_id (lexer, "ALLVISIBLE"))
3602 t->chisq->all_visible = true;
3603 else if (lex_match_id (lexer, "SUBTOTALS"))
3604 t->chisq->all_visible = false;
3607 lex_error_expecting (lexer,
3608 "ALLVISIBLE", "SUBTOTALS");
3614 lex_error_expecting (lexer, "TYPE", "ALPHA",
3615 "INCLUDEMRSETS", "CATEGORIES");
3619 while (lex_token (lexer) != T_SLASH
3620 && lex_token (lexer) != T_ENDCMD);
3622 else if (lex_match_id (lexer, "COMPARETEST"))
3626 t->pairwise = xmalloc (sizeof *t->pairwise);
3627 *t->pairwise = (struct ctables_pairwise) {
3629 .alpha = { .05, .05 },
3630 .adjust = BONFERRONI,
3631 .include_mrsets = true,
3632 .meansvariance_allcats = true,
3633 .all_visible = true,
3642 if (lex_match_id (lexer, "TYPE"))
3644 lex_match (lexer, T_EQUALS);
3645 if (lex_match_id (lexer, "PROP"))
3646 t->pairwise->type = PROP;
3647 else if (lex_match_id (lexer, "MEAN"))
3648 t->pairwise->type = MEAN;
3651 lex_error_expecting (lexer, "PROP", "MEAN");
3655 else if (lex_match_id (lexer, "ALPHA"))
3657 lex_match (lexer, T_EQUALS);
3659 if (!lex_force_num_range_open (lexer, "ALPHA", 0, 1))
3661 double a0 = lex_number (lexer);
3664 lex_match (lexer, T_COMMA);
3665 if (lex_is_number (lexer))
3667 if (!lex_force_num_range_open (lexer, "ALPHA", 0, 1))
3669 double a1 = lex_number (lexer);
3672 t->pairwise->alpha[0] = MIN (a0, a1);
3673 t->pairwise->alpha[1] = MAX (a0, a1);
3676 t->pairwise->alpha[0] = t->pairwise->alpha[1] = a0;
3678 else if (lex_match_id (lexer, "ADJUST"))
3680 lex_match (lexer, T_EQUALS);
3681 if (lex_match_id (lexer, "BONFERRONI"))
3682 t->pairwise->adjust = BONFERRONI;
3683 else if (lex_match_id (lexer, "BH"))
3684 t->pairwise->adjust = BH;
3685 else if (lex_match_id (lexer, "NONE"))
3686 t->pairwise->adjust = 0;
3689 lex_error_expecting (lexer, "BONFERRONI", "BH",
3694 else if (lex_match_id (lexer, "INCLUDEMRSETS"))
3696 lex_match (lexer, T_EQUALS);
3697 if (!parse_bool (lexer, &t->pairwise->include_mrsets))
3700 else if (lex_match_id (lexer, "MEANSVARIANCE"))
3702 lex_match (lexer, T_EQUALS);
3703 if (lex_match_id (lexer, "ALLCATS"))
3704 t->pairwise->meansvariance_allcats = true;
3705 else if (lex_match_id (lexer, "TESTEDCATS"))
3706 t->pairwise->meansvariance_allcats = false;
3709 lex_error_expecting (lexer, "ALLCATS", "TESTEDCATS");
3713 else if (lex_match_id (lexer, "CATEGORIES"))
3715 lex_match (lexer, T_EQUALS);
3716 if (lex_match_id (lexer, "ALLVISIBLE"))
3717 t->pairwise->all_visible = true;
3718 else if (lex_match_id (lexer, "SUBTOTALS"))
3719 t->pairwise->all_visible = false;
3722 lex_error_expecting (lexer, "ALLVISIBLE",
3727 else if (lex_match_id (lexer, "MERGE"))
3729 lex_match (lexer, T_EQUALS);
3730 if (!parse_bool (lexer, &t->pairwise->merge))
3733 else if (lex_match_id (lexer, "STYLE"))
3735 lex_match (lexer, T_EQUALS);
3736 if (lex_match_id (lexer, "APA"))
3737 t->pairwise->apa_style = true;
3738 else if (lex_match_id (lexer, "SIMPLE"))
3739 t->pairwise->apa_style = false;
3742 lex_error_expecting (lexer, "APA", "SIMPLE");
3746 else if (lex_match_id (lexer, "SHOWSIG"))
3748 lex_match (lexer, T_EQUALS);
3749 if (!parse_bool (lexer, &t->pairwise->show_sig))
3754 lex_error_expecting (lexer, "TYPE", "ALPHA", "ADJUST",
3755 "INCLUDEMRSETS", "MEANSVARIANCE",
3756 "CATEGORIES", "MERGE", "STYLE",
3761 while (lex_token (lexer) != T_SLASH
3762 && lex_token (lexer) != T_ENDCMD);
3766 lex_error_expecting (lexer, "TABLE", "SLABELS", "CLABELS",
3767 "CRITERIA", "CATEGORIES", "TITLES",
3768 "SIGTEST", "COMPARETEST");
3772 if (!lex_match (lexer, T_SLASH))
3776 if (t->label_axis[PIVOT_AXIS_ROW] != PIVOT_AXIS_ROW
3777 && t->label_axis[PIVOT_AXIS_COLUMN] != PIVOT_AXIS_COLUMN)
3779 msg (SE, _("ROWLABELS and COLLABELS may not both be specified."));
3783 ctables_prepare_table (t);
3785 ctables_check_label_position (t, PIVOT_AXIS_ROW);
3786 ctables_check_label_position (t, PIVOT_AXIS_COLUMN);
3788 while (lex_token (lexer) != T_ENDCMD);
3790 bool ok = ctables_execute (ds, ct);
3791 ctables_destroy (ct);
3792 return ok ? CMD_SUCCESS : CMD_FAILURE;
3795 ctables_destroy (ct);