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/>. */
22 #include "data/casereader.h"
23 #include "data/casewriter.h"
24 #include "data/data-out.h"
25 #include "data/dataset.h"
26 #include "data/dictionary.h"
27 #include "data/mrset.h"
28 #include "data/subcase.h"
29 #include "data/value-labels.h"
30 #include "language/command.h"
31 #include "language/lexer/format-parser.h"
32 #include "language/lexer/lexer.h"
33 #include "language/lexer/variable-parser.h"
34 #include "libpspp/array.h"
35 #include "libpspp/assertion.h"
36 #include "libpspp/hash-functions.h"
37 #include "libpspp/hmap.h"
38 #include "libpspp/i18n.h"
39 #include "libpspp/message.h"
40 #include "libpspp/string-array.h"
41 #include "math/mode.h"
42 #include "math/moments.h"
43 #include "math/percentiles.h"
44 #include "math/sort.h"
45 #include "output/pivot-table.h"
47 #include "gl/minmax.h"
48 #include "gl/xalloc.h"
51 #define _(msgid) gettext (msgid)
52 #define N_(msgid) (msgid)
56 CTVL_NONE = SETTINGS_VALUE_SHOW_DEFAULT,
57 CTVL_NAME = SETTINGS_VALUE_SHOW_VALUE,
58 CTVL_LABEL = SETTINGS_VALUE_SHOW_LABEL,
59 CTVL_BOTH = SETTINGS_VALUE_SHOW_BOTH,
63 - unweighted summaries (U*)
64 - lower confidence limits (*.LCL)
65 - upper confidence limits (*.UCL)
66 - standard error (*.SE)
69 /* All variables. */ \
70 S(CTSF_COUNT, "COUNT", N_("Count"), CTF_COUNT, CTFA_ALL) \
71 S(CTSF_ECOUNT, "ECOUNT", N_("Adjusted Count"), CTF_COUNT, CTFA_ALL) \
72 S(CTSF_ROWPCT_COUNT, "ROWPCT.COUNT", N_("Row %"), CTF_PERCENT, CTFA_ALL) \
73 S(CTSF_COLPCT_COUNT, "COLPCT.COUNT", N_("Column %"), CTF_PERCENT, CTFA_ALL) \
74 S(CTSF_TABLEPCT_COUNT, "TABLEPCT.COUNT", N_("Table %"), CTF_PERCENT, CTFA_ALL) \
75 S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT.COUNT", N_("Subtable %"), CTF_PERCENT, CTFA_ALL) \
76 S(CTSF_LAYERPCT_COUNT, "LAYERPCT.COUNT", N_("Layer %"), CTF_PERCENT, CTFA_ALL) \
77 S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT.COUNT", N_("Layer Row %"), CTF_PERCENT, CTFA_ALL) \
78 S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT.COUNT", N_("Layer Column %"), CTF_PERCENT, CTFA_ALL) \
79 S(CTSF_ROWPCT_VALIDN, "ROWPCT.VALIDN", N_("Row Valid N %"), CTF_PERCENT, CTFA_ALL) \
80 S(CTSF_COLPCT_VALIDN, "COLPCT.VALIDN", N_("Column Valid N %"), CTF_PERCENT, CTFA_ALL) \
81 S(CTSF_TABLEPCT_VALIDN, "TABLEPCT.VALIDN", N_("Table Valid N %"), CTF_PERCENT, CTFA_ALL) \
82 S(CTSF_SUBTABLEPCT_VALIDN, "SUBTABLEPCT.VALIDN", N_("Subtable Valid N %"), CTF_PERCENT, CTFA_ALL) \
83 S(CTSF_LAYERPCT_VALIDN, "LAYERPCT.VALIDN", N_("Layer Valid N %"), CTF_PERCENT, CTFA_ALL) \
84 S(CTSF_LAYERROWPCT_VALIDN, "LAYERROWPCT.VALIDN", N_("Layer Row Valid N %"), CTF_PERCENT, CTFA_ALL) \
85 S(CTSF_LAYERCOLPCT_VALIDN, "LAYERCOLPCT.VALIDN", N_("Layer Column Valid N %"), CTF_PERCENT, CTFA_ALL) \
86 S(CTSF_ROWPCT_TOTALN, "ROWPCT.TOTALN", N_("Row Total N %"), CTF_PERCENT, CTFA_ALL) \
87 S(CTSF_COLPCT_TOTALN, "COLPCT.TOTALN", N_("Column Total N %"), CTF_PERCENT, CTFA_ALL) \
88 S(CTSF_TABLEPCT_TOTALN, "TABLEPCT.TOTALN", N_("Table Total N %"), CTF_PERCENT, CTFA_ALL) \
89 S(CTSF_SUBTABLEPCT_TOTALN, "SUBTABLEPCT.TOTALN", N_("Subtable Total N %"), CTF_PERCENT, CTFA_ALL) \
90 S(CTSF_LAYERPCT_TOTALN, "LAYERPCT.TOTALN", N_("Layer Total N %"), CTF_PERCENT, CTFA_ALL) \
91 S(CTSF_LAYERROWPCT_TOTALN, "LAYERROWPCT.TOTALN", N_("Layer Row Total N %"), CTF_PERCENT, CTFA_ALL) \
92 S(CTSF_LAYERCOLPCT_TOTALN, "LAYERCOLPCT.TOTALN", N_("Layer Column Total N %"), CTF_PERCENT, CTFA_ALL) \
94 /* Scale variables, totals, and subtotals. */ \
95 S(CTSF_MAXIMUM, "MAXIMUM", N_("Maximum"), CTF_GENERAL, CTFA_SCALE) \
96 S(CTSF_MEAN, "MEAN", N_("Mean"), CTF_GENERAL, CTFA_SCALE) \
97 S(CTSF_MEDIAN, "MEDIAN", N_("Median"), CTF_GENERAL, CTFA_SCALE) \
98 S(CTSF_MINIMUM, "MINIMUM", N_("Minimum"), CTF_GENERAL, CTFA_SCALE) \
99 S(CTSF_MISSING, "MISSING", N_("Missing"), CTF_GENERAL, CTFA_SCALE) \
100 S(CTSF_MODE, "MODE", N_("Mode"), CTF_GENERAL, CTFA_SCALE) \
101 S(CTSF_PTILE, "PTILE", N_("Percentile"), CTF_GENERAL, CTFA_SCALE) \
102 S(CTSF_RANGE, "RANGE", N_("Range"), CTF_GENERAL, CTFA_SCALE) \
103 S(CTSF_SEMEAN, "SEMEAN", N_("Std Error of Mean"), CTF_GENERAL, CTFA_SCALE) \
104 S(CTSF_STDDEV, "STDDEV", N_("Std Deviation"), CTF_GENERAL, CTFA_SCALE) \
105 S(CTSF_SUM, "SUM", N_("Sum"), CTF_GENERAL, CTFA_SCALE) \
106 S(CSTF_TOTALN, "TOTALN", N_("Total N"), CTF_COUNT, CTFA_SCALE) \
107 S(CTSF_ETOTALN, "ETOTALN", N_("Adjusted Total N"), CTF_COUNT, CTFA_SCALE) \
108 S(CTSF_VALIDN, "VALIDN", N_("Valid N"), CTF_COUNT, CTFA_SCALE) \
109 S(CTSF_EVALIDN, "EVALIDN", N_("Adjusted Valid N"), CTF_COUNT, CTFA_SCALE) \
110 S(CTSF_VARIANCE, "VARIANCE", N_("Variance"), CTF_GENERAL, CTFA_SCALE) \
111 S(CTSF_ROWPCT_SUM, "ROWPCT.SUM", N_("Row Sum %"), CTF_PERCENT, CTFA_SCALE) \
112 S(CTSF_COLPCT_SUM, "COLPCT.SUM", N_("Column Sum %"), CTF_PERCENT, CTFA_SCALE) \
113 S(CTSF_TABLEPCT_SUM, "TABLEPCT.SUM", N_("Table Sum %"), CTF_PERCENT, CTFA_SCALE) \
114 S(CTSF_SUBTABLEPCT_SUM, "SUBTABLEPCT.SUM", N_("Subtable Sum %"), CTF_PERCENT, CTFA_SCALE) \
115 S(CTSF_LAYERPCT_SUM, "LAYERPCT.SUM", N_("Layer Sum %"), CTF_PERCENT, CTFA_SCALE) \
116 S(CTSF_LAYERROWPCT_SUM, "LAYERROWPCT.SUM", N_("Layer Row Sum %"), CTF_PERCENT, CTFA_SCALE) \
117 S(CTSF_LAYERCOLPCT_SUM, "LAYERCOLPCT.SUM", N_("Layer Column Sum %"), CTF_PERCENT, CTFA_SCALE) \
119 #if 0 /* Multiple response sets not yet implemented. */
120 S(CTSF_RESPONSES, "RESPONSES", N_("Responses"), CTF_COUNT, CTFA_MRSETS) \
121 S(CTSF_ROWPCT_RESPONSES, "ROWPCT.RESPONSES", N_("Row Responses %"), CTF_PERCENT, CTFA_MRSETS) \
122 S(CTSF_COLPCT_RESPONSES, "COLPCT.RESPONSES", N_("Column Responses %"), CTF_PERCENT, CTFA_MRSETS) \
123 S(CTSF_TABLEPCT_RESPONSES, "TABLEPCT.RESPONSES", N_("Table Responses %"), CTF_PERCENT, CTFA_MRSETS) \
124 S(CTSF_SUBTABLEPCT_RESPONSES, "SUBTABLEPCT.RESPONSES", N_("Subtable Responses %"), CTF_PERCENT, CTFA_MRSETS) \
125 S(CTSF_LAYERPCT_RESPONSES, "LAYERPCT.RESPONSES", N_("Layer Responses %"), CTF_PERCENT, CTFA_MRSETS) \
126 S(CTSF_LAYERROWPCT_RESPONSES, "LAYERROWPCT.RESPONSES", N_("Layer Row Responses %"), CTF_PERCENT, CTFA_MRSETS) \
127 S(CTSF_LAYERCOLPCT_RESPONSES, "LAYERCOLPCT.RESPONSES", N_("Layer Column Responses %"), CTF_PERCENT, CTFA_MRSETS) \
128 S(CTSF_ROWPCT_RESPONSES_COUNT, "ROWPCT.RESPONSES.COUNT", N_("Row Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
129 S(CTSF_COLPCT_RESPONSES_COUNT, "COLPCT.RESPONSES.COUNT", N_("Column Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
130 S(CTSF_TABLEPCT_RESPONSES_COUNT, "TABLEPCT.RESPONSES.COUNT", N_("Table Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
131 S(CTSF_SUBTABLEPCT_RESPONSES_COUNT, "SUBTABLEPCT.RESPONSES.COUNT", N_("Subtable Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
132 S(CTSF_LAYERPCT_RESPONSES_COUNT, "LAYERPCT.RESPONSES.COUNT", N_("Layer Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
133 S(CTSF_LAYERROWPCT_RESPONSES_COUNT, "LAYERROWPCT.RESPONSES.COUNT", N_("Layer Row Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
134 S(CTSF_LAYERCOLPCT_RESPONSES_COUNT, "LAYERCOLPCT.RESPONSES.COUNT", N_("Layer Column Responses % (Base: Count)"), CTF_PERCENT, CTFA_MRSETS) \
135 S(CTSF_ROWPCT_COUNT_RESPONSES, "ROWPCT.COUNT.RESPONSES", N_("Row Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
136 S(CTSF_COLPCT_COUNT_RESPONSES, "COLPCT.COUNT.RESPONSES", N_("Column Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
137 S(CTSF_TABLEPCT_COUNT_RESPONSES, "TABLEPCT.COUNT.RESPONSES", N_("Table Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
138 S(CTSF_SUBTABLEPCT_COUNT_RESPONSES, "SUBTABLEPCT.COUNT.RESPONSES", N_("Subtable Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
139 S(CTSF_LAYERPCT_COUNT_RESPONSES, "LAYERPCT.COUNT.RESPONSES", N_("Layer Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
140 S(CTSF_LAYERROWPCT_COUNT_RESPONSES, "LAYERROWPCT.COUNT.RESPONSES", N_("Layer Row Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS) \
141 S(CTSF_LAYERCOLPCT_COUNT_RESPONSES, "LAYERCOLPCT.RESPONSES.COUNT", N_("Layer Column Count % (Base: Responses)"), CTF_PERCENT, CTFA_MRSETS)
144 enum ctables_summary_function
146 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) ENUM,
152 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) +1
153 N_CTSF_FUNCTIONS = SUMMARIES
157 static bool ctables_summary_function_is_count (enum ctables_summary_function);
159 enum ctables_domain_type
161 /* Within a section, where stacked variables divide one section from
163 CTDT_TABLE, /* All layers of a whole section. */
164 CTDT_LAYER, /* One layer within a section. */
165 CTDT_LAYERROW, /* Row in one layer within a section. */
166 CTDT_LAYERCOL, /* Column in one layer within a section. */
168 /* Within a subtable, where a subtable pairs an innermost row variable with
169 an innermost column variable within a single layer. */
170 CTDT_SUBTABLE, /* Whole subtable. */
171 CTDT_ROW, /* Row within a subtable. */
172 CTDT_COL, /* Column within a subtable. */
176 struct ctables_domain
178 struct hmap_node node;
180 const struct ctables_cell *example;
182 double d_valid; /* Dictionary weight. */
184 double e_valid; /* Effective weight */
188 enum ctables_summary_variant
197 /* In struct ctables_section's 'cells' hmap. Indexed by all the values in
198 all the axes (except the scalar variable, if any). */
199 struct hmap_node node;
201 /* The domains that contain this cell. */
202 bool contributes_to_domains;
203 struct ctables_domain *domains[N_CTDTS];
207 enum ctables_summary_variant sv;
209 struct ctables_cell_axis
211 struct ctables_cell_value
213 const struct ctables_category *category;
221 union ctables_summary *summaries;
226 const struct dictionary *dict;
227 struct pivot_table_look *look;
229 /* CTABLES has a number of extra formats that we implement via custom
230 currency specifications on an alternate fmt_settings. */
231 #define CTEF_NEGPAREN FMT_CCA
232 #define CTEF_NEQUAL FMT_CCB
233 #define CTEF_PAREN FMT_CCC
234 #define CTEF_PCTPAREN FMT_CCD
235 struct fmt_settings ctables_formats;
237 /* If this is NULL, zeros are displayed using the normal print format.
238 Otherwise, this string is displayed. */
241 /* If this is NULL, missing values are displayed using the normal print
242 format. Otherwise, this string is displayed. */
245 /* Indexed by variable dictionary index. */
246 enum ctables_vlabel *vlabels;
248 struct hmap postcomputes; /* Contains "struct ctables_postcompute"s. */
250 bool mrsets_count_duplicates; /* MRSETS. */
251 bool smissing_listwise; /* SMISSING. */
252 struct variable *e_weight; /* WEIGHT. */
253 int hide_threshold; /* HIDESMALLCOUNTS. */
255 struct ctables_table **tables;
259 static struct ctables_postcompute *ctables_find_postcompute (struct ctables *,
262 struct ctables_postcompute
264 struct hmap_node hmap_node; /* In struct ctables's 'pcompute' hmap. */
265 char *name; /* Name, without leading &. */
267 struct msg_location *location; /* Location of definition. */
268 struct ctables_pcexpr *expr;
270 struct ctables_summary_spec_set *specs;
271 bool hide_source_cats;
274 struct ctables_pcexpr
284 enum ctables_postcompute_op
287 CTPO_CONSTANT, /* 5 */
288 CTPO_CAT_NUMBER, /* [5] */
289 CTPO_CAT_STRING, /* ["STRING"] */
290 CTPO_CAT_RANGE, /* [LO THRU 5] */
291 CTPO_CAT_MISSING, /* MISSING */
292 CTPO_CAT_OTHERNM, /* OTHERNM */
293 CTPO_CAT_SUBTOTAL, /* SUBTOTAL */
294 CTPO_CAT_TOTAL, /* TOTAL */
308 /* CTPO_CAT_NUMBER. */
311 /* CTPO_CAT_STRING. */
314 /* CTPO_CAT_RANGE. */
317 /* CTPO_CAT_SUBTOTAL. */
318 size_t subtotal_index;
320 /* Two elements: CTPO_ADD, CTPO_SUB, CTPO_MUL, CTPO_DIV, CTPO_POW.
321 One element: CTPO_NEG. */
322 struct ctables_pcexpr *subs[2];
325 /* Source location. */
326 struct msg_location *location;
329 static void ctables_pcexpr_destroy (struct ctables_pcexpr *);
330 static struct ctables_pcexpr *ctables_pcexpr_allocate_binary (
331 enum ctables_postcompute_op, struct ctables_pcexpr *sub0,
332 struct ctables_pcexpr *sub1);
334 struct ctables_summary_spec_set
336 struct ctables_summary_spec *specs;
340 struct variable *var;
343 static void ctables_summary_spec_set_clone (struct ctables_summary_spec_set *,
344 const struct ctables_summary_spec_set *);
345 static void ctables_summary_spec_set_uninit (struct ctables_summary_spec_set *);
347 /* A nested sequence of variables, e.g. a > b > c. */
350 struct variable **vars;
353 size_t *domains[N_CTDTS];
354 size_t n_domains[N_CTDTS];
356 struct ctables_summary_spec_set specs[N_CSVS];
359 /* A stack of nestings, e.g. nest1 + nest2 + ... + nestN. */
362 struct ctables_nest *nests;
368 struct hmap_node node;
373 struct ctables_occurrence
375 struct hmap_node node;
379 struct ctables_section
381 struct ctables_table *table;
382 struct ctables_nest *nests[PIVOT_N_AXES];
383 struct hmap *occurrences[PIVOT_N_AXES];
384 struct hmap cells; /* Contains "struct ctable_cell"s. */
385 struct hmap domains[N_CTDTS]; /* Contains "struct ctable_domain"s. */
390 struct ctables *ctables;
391 struct ctables_axis *axes[PIVOT_N_AXES];
392 struct ctables_stack stacks[PIVOT_N_AXES];
393 struct ctables_section *sections;
395 enum pivot_axis_type summary_axis;
396 struct ctables_summary_spec_set summary_specs;
398 const struct variable *clabels_example;
399 struct hmap clabels_values_map;
400 struct ctables_value **clabels_values;
401 size_t n_clabels_values;
403 enum pivot_axis_type slabels_axis;
404 bool slabels_visible;
406 /* The innermost category labels for axis 'a' appear on axis label_axis[a].
408 Most commonly, label_axis[a] == a, and in particular we always have
409 label_axis{PIVOT_AXIS_LAYER] == PIVOT_AXIS_LAYER.
411 If ROWLABELS or COLLABELS is specified, then one of
412 label_axis[PIVOT_AXIS_ROW] or label_axis[PIVOT_AXIS_COLUMN] can be the
413 opposite axis or PIVOT_AXIS_LAYER. Only one of them will differ.
415 enum pivot_axis_type label_axis[PIVOT_N_AXES];
416 enum pivot_axis_type clabels_from_axis;
418 /* Indexed by variable dictionary index. */
419 struct ctables_categories **categories;
428 struct ctables_chisq *chisq;
429 struct ctables_pairwise *pairwise;
437 struct variable *var;
438 const struct mrset *mrset;
442 static const struct fmt_spec *
443 ctables_var_get_print_format (const struct ctables_var *var)
445 return (var->is_mrset
446 ? var_get_print_format (var->mrset->vars[0])
447 : var_get_print_format (var->var));
451 ctables_var_name (const struct ctables_var *var)
453 return var->is_mrset ? var->mrset->name : var_get_name (var->var);
456 struct ctables_categories
459 struct ctables_category *cats;
464 struct ctables_category
466 enum ctables_category_type
468 /* Explicit category lists. */
476 /* Totals and subtotals. */
480 /* Implicit category lists. */
487 struct ctables_category *subtotal;
493 double number; /* CCT_NUMBER. */
494 char *string; /* CCT_STRING. */
495 double range[2]; /* CCT_RANGE. */
499 char *total_label; /* CCT_SUBTOTAL, CCT_TOTAL. */
500 bool hide_subcategories; /* CCT_SUBTOTAL. */
503 const struct ctables_postcompute *pc; /* CCT_POSTCOMPUTE. */
505 /* CCT_VALUE, CCT_LABEL, CCT_FUNCTION. */
508 bool include_missing;
512 enum ctables_summary_function sort_function;
513 struct variable *sort_var;
518 /* Source location. This is null for CCT_TOTAL, CCT_VALUE, CCT_LABEL,
520 struct msg_location *location;
524 ctables_category_uninit (struct ctables_category *cat)
535 case CCT_POSTCOMPUTE:
544 free (cat->total_label);
555 ctables_category_equal (const struct ctables_category *a,
556 const struct ctables_category *b)
558 if (a->type != b->type)
564 return a->number == b->number;
567 return strcmp (a->string, b->string);
570 return a->range[0] == b->range[0] && a->range[1] == b->range[1];
576 case CCT_POSTCOMPUTE:
577 return a->pc == b->pc;
581 return !strcmp (a->total_label, b->total_label);
586 return (a->include_missing == b->include_missing
587 && a->sort_ascending == b->sort_ascending
588 && a->sort_function == b->sort_function
589 && a->sort_var == b->sort_var
590 && a->percentile == b->percentile);
597 ctables_categories_unref (struct ctables_categories *c)
602 assert (c->n_refs > 0);
606 for (size_t i = 0; i < c->n_cats; i++)
607 ctables_category_uninit (&c->cats[i]);
613 ctables_categories_equal (const struct ctables_categories *a,
614 const struct ctables_categories *b)
616 if (a->n_cats != b->n_cats || a->show_empty != b->show_empty)
619 for (size_t i = 0; i < a->n_cats; i++)
620 if (!ctables_category_equal (&a->cats[i], &b->cats[i]))
626 /* Chi-square test (SIGTEST). */
634 /* Pairwise comparison test (COMPARETEST). */
635 struct ctables_pairwise
637 enum { PROP, MEAN } type;
640 bool meansvariance_allcats;
642 enum { BONFERRONI = 1, BH } adjust;
666 struct ctables_var var;
668 struct ctables_summary_spec_set specs[N_CSVS];
672 struct ctables_axis *subs[2];
675 struct msg_location *loc;
678 static void ctables_axis_destroy (struct ctables_axis *);
687 enum ctables_function_availability
689 CTFA_ALL, /* Any variables. */
690 CTFA_SCALE, /* Only scale variables, totals, and subtotals. */
691 CTFA_MRSETS, /* Only multiple-response sets */
694 struct ctables_summary_spec
696 enum ctables_summary_function function;
697 double percentile; /* CTSF_PTILE only. */
700 struct fmt_spec format;
701 bool is_ctables_format; /* Is 'format' one of CTEF_*? */
707 ctables_summary_spec_clone (struct ctables_summary_spec *dst,
708 const struct ctables_summary_spec *src)
711 dst->label = xstrdup (src->label);
715 ctables_summary_spec_uninit (struct ctables_summary_spec *s)
722 ctables_summary_spec_set_clone (struct ctables_summary_spec_set *dst,
723 const struct ctables_summary_spec_set *src)
725 struct ctables_summary_spec *specs = xnmalloc (src->n, sizeof *specs);
726 for (size_t i = 0; i < src->n; i++)
727 ctables_summary_spec_clone (&specs[i], &src->specs[i]);
729 *dst = (struct ctables_summary_spec_set) {
738 ctables_summary_spec_set_uninit (struct ctables_summary_spec_set *set)
740 for (size_t i = 0; i < set->n; i++)
741 ctables_summary_spec_uninit (&set->specs[i]);
746 parse_col_width (struct lexer *lexer, const char *name, double *width)
748 lex_match (lexer, T_EQUALS);
749 if (lex_match_id (lexer, "DEFAULT"))
751 else if (lex_force_num_range_closed (lexer, name, 0, DBL_MAX))
753 *width = lex_number (lexer);
763 parse_bool (struct lexer *lexer, bool *b)
765 if (lex_match_id (lexer, "NO"))
767 else if (lex_match_id (lexer, "YES"))
771 lex_error_expecting (lexer, "YES", "NO");
777 static enum ctables_function_availability
778 ctables_function_availability (enum ctables_summary_function f)
780 static enum ctables_function_availability availability[] = {
781 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = AVAILABILITY,
786 return availability[f];
790 ctables_summary_function_is_count (enum ctables_summary_function f)
796 case CTSF_ROWPCT_COUNT:
797 case CTSF_COLPCT_COUNT:
798 case CTSF_TABLEPCT_COUNT:
799 case CTSF_SUBTABLEPCT_COUNT:
800 case CTSF_LAYERPCT_COUNT:
801 case CTSF_LAYERROWPCT_COUNT:
802 case CTSF_LAYERCOLPCT_COUNT:
805 case CTSF_ROWPCT_VALIDN:
806 case CTSF_COLPCT_VALIDN:
807 case CTSF_TABLEPCT_VALIDN:
808 case CTSF_SUBTABLEPCT_VALIDN:
809 case CTSF_LAYERPCT_VALIDN:
810 case CTSF_LAYERROWPCT_VALIDN:
811 case CTSF_LAYERCOLPCT_VALIDN:
812 case CTSF_ROWPCT_TOTALN:
813 case CTSF_COLPCT_TOTALN:
814 case CTSF_TABLEPCT_TOTALN:
815 case CTSF_SUBTABLEPCT_TOTALN:
816 case CTSF_LAYERPCT_TOTALN:
817 case CTSF_LAYERROWPCT_TOTALN:
818 case CTSF_LAYERCOLPCT_TOTALN:
835 case CTSF_ROWPCT_SUM:
836 case CTSF_COLPCT_SUM:
837 case CTSF_TABLEPCT_SUM:
838 case CTSF_SUBTABLEPCT_SUM:
839 case CTSF_LAYERPCT_SUM:
840 case CTSF_LAYERROWPCT_SUM:
841 case CTSF_LAYERCOLPCT_SUM:
849 parse_ctables_summary_function (struct lexer *lexer,
850 enum ctables_summary_function *f)
854 enum ctables_summary_function function;
855 struct substring name;
857 static struct pair names[] = {
858 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) \
859 { ENUM, SS_LITERAL_INITIALIZER (NAME) },
862 /* The .COUNT suffix may be omitted. */
863 S(CTSF_ROWPCT_COUNT, "ROWPCT", _, _, _)
864 S(CTSF_COLPCT_COUNT, "COLPCT", _, _, _)
865 S(CTSF_TABLEPCT_COUNT, "TABLEPCT", _, _, _)
866 S(CTSF_SUBTABLEPCT_COUNT, "SUBTABLEPCT", _, _, _)
867 S(CTSF_LAYERPCT_COUNT, "LAYERPCT", _, _, _)
868 S(CTSF_LAYERROWPCT_COUNT, "LAYERROWPCT", _, _, _)
869 S(CTSF_LAYERCOLPCT_COUNT, "LAYERCOLPCT", _, _, _)
873 if (!lex_force_id (lexer))
876 for (size_t i = 0; i < sizeof names / sizeof *names; i++)
877 if (ss_equals_case (names[i].name, lex_tokss (lexer)))
879 *f = names[i].function;
884 lex_error (lexer, _("Expecting summary function name."));
889 ctables_axis_destroy (struct ctables_axis *axis)
897 for (size_t i = 0; i < N_CSVS; i++)
898 ctables_summary_spec_set_uninit (&axis->specs[i]);
903 ctables_axis_destroy (axis->subs[0]);
904 ctables_axis_destroy (axis->subs[1]);
907 msg_location_destroy (axis->loc);
911 static struct ctables_axis *
912 ctables_axis_new_nonterminal (enum ctables_axis_op op,
913 struct ctables_axis *sub0,
914 struct ctables_axis *sub1,
915 struct lexer *lexer, int start_ofs)
917 struct ctables_axis *axis = xmalloc (sizeof *axis);
918 *axis = (struct ctables_axis) {
920 .subs = { sub0, sub1 },
921 .loc = lex_ofs_location (lexer, start_ofs, lex_ofs (lexer) - 1),
926 struct ctables_axis_parse_ctx
929 struct dictionary *dict;
931 struct ctables_table *t;
934 static struct fmt_spec
935 ctables_summary_default_format (enum ctables_summary_function function,
936 const struct ctables_var *var)
938 static const enum ctables_format default_formats[] = {
939 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = FORMAT,
943 switch (default_formats[function])
946 return (struct fmt_spec) { .type = FMT_F, .w = 40 };
949 return (struct fmt_spec) { .type = FMT_PCT, .w = 40, .d = 1 };
952 return *ctables_var_get_print_format (var);
960 ctables_summary_default_label (enum ctables_summary_function function,
963 static const char *default_labels[] = {
964 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = LABEL,
969 return (function == CTSF_PTILE
970 ? xasprintf (_("Percentile %.2f"), percentile)
971 : xstrdup (gettext (default_labels[function])));
975 ctables_summary_function_name (enum ctables_summary_function function)
977 static const char *names[] = {
978 #define S(ENUM, NAME, LABEL, FORMAT, AVAILABILITY) [ENUM] = NAME,
982 return names[function];
986 add_summary_spec (struct ctables_axis *axis,
987 enum ctables_summary_function function, double percentile,
988 const char *label, const struct fmt_spec *format,
989 bool is_ctables_format, const struct msg_location *loc,
990 enum ctables_summary_variant sv)
992 if (axis->op == CTAO_VAR)
994 const char *function_name = ctables_summary_function_name (function);
995 const char *var_name = ctables_var_name (&axis->var);
996 switch (ctables_function_availability (function))
999 if (!axis->var.is_mrset)
1001 msg_at (SE, loc, _("Summary function %s applies only to multiple "
1002 "response sets."), function_name);
1003 msg_at (SN, axis->loc, _("'%s' is not a multiple response set."),
1013 _("Summary function %s applies only to scale variables."),
1015 msg_at (SN, axis->loc, _("'%s' is not a scale variable."),
1025 struct ctables_summary_spec_set *set = &axis->specs[sv];
1026 if (set->n >= set->allocated)
1027 set->specs = x2nrealloc (set->specs, &set->allocated,
1028 sizeof *set->specs);
1030 struct ctables_summary_spec *dst = &set->specs[set->n++];
1031 *dst = (struct ctables_summary_spec) {
1032 .function = function,
1033 .percentile = percentile,
1034 .label = xstrdup (label),
1035 .format = (format ? *format
1036 : ctables_summary_default_format (function, &axis->var)),
1037 .is_ctables_format = is_ctables_format,
1043 for (size_t i = 0; i < 2; i++)
1044 if (!add_summary_spec (axis->subs[i], function, percentile, label,
1045 format, is_ctables_format, loc, sv))
1051 static struct ctables_axis *ctables_axis_parse_stack (
1052 struct ctables_axis_parse_ctx *);
1055 ctables_var_parse (struct lexer *lexer, struct dictionary *dict,
1056 struct ctables_var *var)
1058 if (ss_starts_with (lex_tokss (lexer), ss_cstr ("$")))
1060 *var = (struct ctables_var) {
1062 .mrset = dict_lookup_mrset (dict, lex_tokcstr (lexer))
1066 lex_error (lexer, _("'%s' does not name a multiple-response set "
1067 "in the active file dictionary."),
1068 lex_tokcstr (lexer));
1076 *var = (struct ctables_var) {
1078 .var = parse_variable (lexer, dict),
1080 return var->var != NULL;
1084 static struct ctables_axis *
1085 ctables_axis_parse_primary (struct ctables_axis_parse_ctx *ctx)
1087 if (lex_match (ctx->lexer, T_LPAREN))
1089 struct ctables_axis *sub = ctables_axis_parse_stack (ctx);
1090 if (!sub || !lex_force_match (ctx->lexer, T_RPAREN))
1092 ctables_axis_destroy (sub);
1098 if (!lex_force_id (ctx->lexer))
1101 int start_ofs = lex_ofs (ctx->lexer);
1102 struct ctables_var var;
1103 if (!ctables_var_parse (ctx->lexer, ctx->dict, &var))
1106 struct ctables_axis *axis = xmalloc (sizeof *axis);
1107 *axis = (struct ctables_axis) { .op = CTAO_VAR, .var = var };
1109 /* XXX should figure out default measures by reading data */
1110 axis->scale = (var.is_mrset ? false
1111 : lex_match_phrase (ctx->lexer, "[S]") ? true
1112 : lex_match_phrase (ctx->lexer, "[C]") ? false
1113 : var_get_measure (var.var) == MEASURE_SCALE);
1114 axis->loc = lex_ofs_location (ctx->lexer, start_ofs,
1115 lex_ofs (ctx->lexer) - 1);
1120 has_digit (const char *s)
1122 return s[strcspn (s, "0123456789")] != '\0';
1126 parse_ctables_format_specifier (struct lexer *lexer, struct fmt_spec *format,
1127 bool *is_ctables_format)
1129 char type[FMT_TYPE_LEN_MAX + 1];
1130 if (!parse_abstract_format_specifier__ (lexer, type, &format->w, &format->d))
1133 if (!strcasecmp (type, "NEGPAREN"))
1134 format->type = CTEF_NEGPAREN;
1135 else if (!strcasecmp (type, "NEQUAL"))
1136 format->type = CTEF_NEQUAL;
1137 else if (!strcasecmp (type, "PAREN"))
1138 format->type = CTEF_PAREN;
1139 else if (!strcasecmp (type, "PCTPAREN"))
1140 format->type = CTEF_PCTPAREN;
1143 *is_ctables_format = false;
1144 return (parse_format_specifier (lexer, format)
1145 && fmt_check_output (format)
1146 && fmt_check_type_compat (format, VAL_NUMERIC));
1151 msg (SE, _("Output format %s requires width 2 or greater."), type);
1154 else if (format->d > format->w - 1)
1156 msg (SE, _("Output format %s requires width greater than decimals."),
1162 *is_ctables_format = true;
1167 static struct ctables_axis *
1168 ctables_axis_parse_postfix (struct ctables_axis_parse_ctx *ctx)
1170 struct ctables_axis *sub = ctables_axis_parse_primary (ctx);
1171 if (!sub || !lex_match (ctx->lexer, T_LBRACK))
1174 enum ctables_summary_variant sv = CSV_CELL;
1177 int start_ofs = lex_ofs (ctx->lexer);
1179 /* Parse function. */
1180 enum ctables_summary_function function;
1181 if (!parse_ctables_summary_function (ctx->lexer, &function))
1184 /* Parse percentile. */
1185 double percentile = 0;
1186 if (function == CTSF_PTILE)
1188 if (!lex_force_num_range_closed (ctx->lexer, "PTILE", 0, 100))
1190 percentile = lex_number (ctx->lexer);
1191 lex_get (ctx->lexer);
1196 if (lex_is_string (ctx->lexer))
1198 label = ss_xstrdup (lex_tokss (ctx->lexer));
1199 lex_get (ctx->lexer);
1202 label = ctables_summary_default_label (function, percentile);
1205 struct fmt_spec format;
1206 const struct fmt_spec *formatp;
1207 bool is_ctables_format = false;
1208 if (lex_token (ctx->lexer) == T_ID
1209 && has_digit (lex_tokcstr (ctx->lexer)))
1211 if (!parse_ctables_format_specifier (ctx->lexer, &format,
1212 &is_ctables_format))
1222 struct msg_location *loc = lex_ofs_location (ctx->lexer, start_ofs,
1223 lex_ofs (ctx->lexer) - 1);
1224 add_summary_spec (sub, function, percentile, label, formatp,
1225 is_ctables_format, loc, sv);
1227 msg_location_destroy (loc);
1229 lex_match (ctx->lexer, T_COMMA);
1230 if (sv == CSV_CELL && lex_match_id (ctx->lexer, "TOTALS"))
1232 if (!lex_force_match (ctx->lexer, T_LBRACK))
1236 else if (lex_match (ctx->lexer, T_RBRACK))
1238 if (sv == CSV_TOTAL && !lex_force_match (ctx->lexer, T_RBRACK))
1245 ctables_axis_destroy (sub);
1249 static const struct ctables_axis *
1250 find_scale (const struct ctables_axis *axis)
1254 else if (axis->op == CTAO_VAR)
1258 assert (!axis->var.is_mrset);
1266 for (size_t i = 0; i < 2; i++)
1268 const struct ctables_axis *scale = find_scale (axis->subs[i]);
1276 static const struct ctables_axis *
1277 find_categorical_summary_spec (const struct ctables_axis *axis)
1281 else if (axis->op == CTAO_VAR)
1282 return !axis->scale && axis->specs[CSV_CELL].n ? axis : NULL;
1285 for (size_t i = 0; i < 2; i++)
1287 const struct ctables_axis *sum
1288 = find_categorical_summary_spec (axis->subs[i]);
1296 static struct ctables_axis *
1297 ctables_axis_parse_nest (struct ctables_axis_parse_ctx *ctx)
1299 int start_ofs = lex_ofs (ctx->lexer);
1300 struct ctables_axis *lhs = ctables_axis_parse_postfix (ctx);
1304 while (lex_match (ctx->lexer, T_GT))
1306 struct ctables_axis *rhs = ctables_axis_parse_postfix (ctx);
1310 struct ctables_axis *nest = ctables_axis_new_nonterminal (
1311 CTAO_NEST, lhs, rhs, ctx->lexer, start_ofs);
1313 const struct ctables_axis *outer_scale = find_scale (lhs);
1314 const struct ctables_axis *inner_scale = find_scale (rhs);
1315 if (outer_scale && inner_scale)
1317 msg_at (SE, nest->loc, _("Cannot nest scale variables."));
1318 msg_at (SN, outer_scale->loc, _("This is an outer scale variable."));
1319 msg_at (SN, inner_scale->loc, _("This is an inner scale variable."));
1320 ctables_axis_destroy (nest);
1324 const struct ctables_axis *outer_sum = find_categorical_summary_spec (lhs);
1327 msg_at (SE, nest->loc,
1328 _("Summaries may only be requested for categorical variables "
1329 "at the innermost nesting level."));
1330 msg_at (SN, outer_sum->loc,
1331 _("This outer categorical variable has a summary."));
1332 ctables_axis_destroy (nest);
1342 static struct ctables_axis *
1343 ctables_axis_parse_stack (struct ctables_axis_parse_ctx *ctx)
1345 int start_ofs = lex_ofs (ctx->lexer);
1346 struct ctables_axis *lhs = ctables_axis_parse_nest (ctx);
1350 while (lex_match (ctx->lexer, T_PLUS))
1352 struct ctables_axis *rhs = ctables_axis_parse_nest (ctx);
1356 lhs = ctables_axis_new_nonterminal (CTAO_STACK, lhs, rhs,
1357 ctx->lexer, start_ofs);
1364 ctables_axis_parse (struct lexer *lexer, struct dictionary *dict,
1365 struct ctables *ct, struct ctables_table *t,
1366 enum pivot_axis_type a)
1368 if (lex_token (lexer) == T_BY
1369 || lex_token (lexer) == T_SLASH
1370 || lex_token (lexer) == T_ENDCMD)
1373 struct ctables_axis_parse_ctx ctx = {
1379 t->axes[a] = ctables_axis_parse_stack (&ctx);
1380 return t->axes[a] != NULL;
1384 ctables_chisq_destroy (struct ctables_chisq *chisq)
1390 ctables_pairwise_destroy (struct ctables_pairwise *pairwise)
1396 ctables_table_destroy (struct ctables_table *t)
1401 for (size_t i = 0; i < t->n_categories; i++)
1402 ctables_categories_unref (t->categories[i]);
1403 free (t->categories);
1405 ctables_axis_destroy (t->axes[PIVOT_AXIS_COLUMN]);
1406 ctables_axis_destroy (t->axes[PIVOT_AXIS_ROW]);
1407 ctables_axis_destroy (t->axes[PIVOT_AXIS_LAYER]);
1411 ctables_chisq_destroy (t->chisq);
1412 ctables_pairwise_destroy (t->pairwise);
1417 ctables_destroy (struct ctables *ct)
1422 pivot_table_look_unref (ct->look);
1426 for (size_t i = 0; i < ct->n_tables; i++)
1427 ctables_table_destroy (ct->tables[i]);
1432 static struct ctables_category
1433 cct_range (double low, double high)
1435 return (struct ctables_category) {
1437 .range = { low, high }
1442 ctables_table_parse_subtotal (struct lexer *lexer, bool hide_subcategories,
1443 struct ctables_category *cat)
1446 if (lex_match (lexer, T_EQUALS))
1448 if (!lex_force_string (lexer))
1451 total_label = ss_xstrdup (lex_tokss (lexer));
1455 total_label = xstrdup (_("Subtotal"));
1457 *cat = (struct ctables_category) {
1458 .type = CCT_SUBTOTAL,
1459 .hide_subcategories = hide_subcategories,
1460 .total_label = total_label
1466 ctables_table_parse_explicit_category (struct lexer *lexer, struct ctables *ct,
1467 struct ctables_category *cat)
1469 if (lex_match_id (lexer, "OTHERNM"))
1470 *cat = (struct ctables_category) { .type = CCT_OTHERNM };
1471 else if (lex_match_id (lexer, "MISSING"))
1472 *cat = (struct ctables_category) { .type = CCT_MISSING };
1473 else if (lex_match_id (lexer, "SUBTOTAL"))
1474 return ctables_table_parse_subtotal (lexer, false, cat);
1475 else if (lex_match_id (lexer, "HSUBTOTAL"))
1476 return ctables_table_parse_subtotal (lexer, true, cat);
1477 else if (lex_match_id (lexer, "LO"))
1479 if (!lex_force_match_id (lexer, "THRU") || lex_force_num (lexer))
1481 *cat = cct_range (-DBL_MAX, lex_number (lexer));
1484 else if (lex_is_number (lexer))
1486 double number = lex_number (lexer);
1488 if (lex_match_id (lexer, "THRU"))
1490 if (lex_match_id (lexer, "HI"))
1491 *cat = cct_range (number, DBL_MAX);
1494 if (!lex_force_num (lexer))
1496 *cat = cct_range (number, lex_number (lexer));
1501 *cat = (struct ctables_category) {
1506 else if (lex_is_string (lexer))
1508 *cat = (struct ctables_category) {
1510 .string = ss_xstrdup (lex_tokss (lexer)),
1514 else if (lex_match (lexer, T_AND))
1516 if (!lex_force_id (lexer))
1518 struct ctables_postcompute *pc = ctables_find_postcompute (
1519 ct, lex_tokcstr (lexer));
1522 struct msg_location *loc = lex_get_location (lexer, -1, 0);
1523 msg_at (SE, loc, _("Unknown postcompute &%s."),
1524 lex_tokcstr (lexer));
1525 msg_location_destroy (loc);
1530 *cat = (struct ctables_category) { .type = CCT_POSTCOMPUTE, .pc = pc };
1534 lex_error (lexer, NULL);
1541 static struct ctables_category *
1542 ctables_find_category_for_postcompute (const struct ctables_categories *cats,
1543 const struct ctables_pcexpr *e)
1545 struct ctables_category *best = NULL;
1546 size_t n_subtotals = 0;
1547 for (size_t i = 0; i < cats->n_cats; i++)
1549 struct ctables_category *cat = &cats->cats[i];
1552 case CTPO_CAT_NUMBER:
1553 if (cat->type == CCT_NUMBER && cat->number == e->number)
1557 case CTPO_CAT_STRING:
1558 if (cat->type == CCT_STRING && !strcmp (cat->string, e->string))
1562 case CTPO_CAT_RANGE:
1563 if (cat->type == CCT_RANGE
1564 && cat->range[0] == e->range[0]
1565 && cat->range[1] == e->range[1])
1569 case CTPO_CAT_MISSING:
1570 if (cat->type == CCT_MISSING)
1574 case CTPO_CAT_OTHERNM:
1575 if (cat->type == CCT_OTHERNM)
1579 case CTPO_CAT_SUBTOTAL:
1580 if (cat->type == CCT_SUBTOTAL)
1583 if (e->subtotal_index == n_subtotals)
1585 else if (e->subtotal_index == 0)
1590 case CTPO_CAT_TOTAL:
1591 if (cat->type == CCT_TOTAL)
1605 if (e->op == CTPO_CAT_SUBTOTAL && e->subtotal_index == 0 && n_subtotals > 1)
1611 ctables_recursive_check_postcompute (const struct ctables_pcexpr *e,
1612 struct ctables_category *pc_cat,
1613 const struct ctables_categories *cats,
1614 const struct msg_location *cats_location)
1618 case CTPO_CAT_NUMBER:
1619 case CTPO_CAT_STRING:
1620 case CTPO_CAT_RANGE:
1621 case CTPO_CAT_MISSING:
1622 case CTPO_CAT_OTHERNM:
1623 case CTPO_CAT_SUBTOTAL:
1624 case CTPO_CAT_TOTAL:
1626 struct ctables_category *cat = ctables_find_category_for_postcompute (
1630 if (e->op == CTPO_CAT_SUBTOTAL && e->subtotal_index == 0)
1632 size_t n_subtotals = 0;
1633 for (size_t i = 0; i < cats->n_cats; i++)
1634 n_subtotals += cats->cats[i].type == CCT_SUBTOTAL;
1635 if (n_subtotals > 1)
1637 msg_at (SE, cats_location,
1638 ngettext ("These categories include %zu instance "
1639 "of SUBTOTAL or HSUBTOTAL, so references "
1640 "from computed categories must refer to "
1641 "subtotals by position.",
1642 "These categories include %zu instances "
1643 "of SUBTOTAL or HSUBTOTAL, so references "
1644 "from computed categories must refer to "
1645 "subtotals by position.",
1648 msg_at (SN, e->location,
1649 _("This is the reference that lacks a position."));
1654 msg_at (SE, pc_cat->location,
1655 _("Computed category &%s references a category not included "
1656 "in the category list."),
1658 msg_at (SN, e->location, _("This is the missing category."));
1659 msg_at (SN, cats_location,
1660 _("To fix the problem, add the missing category to the "
1661 "list of categories here."));
1664 if (pc_cat->pc->hide_source_cats)
1678 for (size_t i = 0; i < 2; i++)
1679 if (e->subs[i] && !ctables_recursive_check_postcompute (
1680 e->subs[i], pc_cat, cats, cats_location))
1690 ctables_table_parse_categories (struct lexer *lexer, struct dictionary *dict,
1691 struct ctables *ct, struct ctables_table *t)
1693 if (!lex_match_id (lexer, "VARIABLES"))
1695 lex_match (lexer, T_EQUALS);
1697 struct variable **vars;
1699 if (!parse_variables (lexer, dict, &vars, &n_vars, PV_NO_SCRATCH))
1702 struct ctables_categories *c = xmalloc (sizeof *c);
1703 *c = (struct ctables_categories) { .n_refs = n_vars, .show_empty = true };
1704 for (size_t i = 0; i < n_vars; i++)
1706 struct ctables_categories **cp
1707 = &t->categories[var_get_dict_index (vars[i])];
1708 ctables_categories_unref (*cp);
1713 size_t allocated_cats = 0;
1714 if (lex_match (lexer, T_LBRACK))
1716 int cats_start_ofs = lex_ofs (lexer);
1719 if (c->n_cats >= allocated_cats)
1720 c->cats = x2nrealloc (c->cats, &allocated_cats, sizeof *c->cats);
1722 int start_ofs = lex_ofs (lexer);
1723 struct ctables_category *cat = &c->cats[c->n_cats];
1724 if (!ctables_table_parse_explicit_category (lexer, ct, cat))
1726 cat->location = lex_ofs_location (lexer, start_ofs, lex_ofs (lexer) - 1);
1729 lex_match (lexer, T_COMMA);
1731 while (!lex_match (lexer, T_RBRACK));
1733 struct msg_location *cats_location
1734 = lex_ofs_location (lexer, cats_start_ofs, lex_ofs (lexer) - 1);
1735 for (size_t i = 0; i < c->n_cats; i++)
1737 struct ctables_category *cat = &c->cats[i];
1738 if (cat->type == CCT_POSTCOMPUTE
1739 && !ctables_recursive_check_postcompute (cat->pc->expr, cat,
1745 struct ctables_category cat = {
1747 .include_missing = false,
1748 .sort_ascending = true,
1750 bool show_totals = false;
1751 char *total_label = NULL;
1752 bool totals_before = false;
1753 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
1755 if (!c->n_cats && lex_match_id (lexer, "ORDER"))
1757 lex_match (lexer, T_EQUALS);
1758 if (lex_match_id (lexer, "A"))
1759 cat.sort_ascending = true;
1760 else if (lex_match_id (lexer, "D"))
1761 cat.sort_ascending = false;
1764 lex_error_expecting (lexer, "A", "D");
1768 else if (!c->n_cats && lex_match_id (lexer, "KEY"))
1770 lex_match (lexer, T_EQUALS);
1771 if (lex_match_id (lexer, "VALUE"))
1772 cat.type = CCT_VALUE;
1773 else if (lex_match_id (lexer, "LABEL"))
1774 cat.type = CCT_LABEL;
1777 cat.type = CCT_FUNCTION;
1778 if (!parse_ctables_summary_function (lexer, &cat.sort_function))
1781 if (lex_match (lexer, T_LPAREN))
1783 cat.sort_var = parse_variable (lexer, dict);
1787 if (cat.sort_function == CTSF_PTILE)
1789 lex_match (lexer, T_COMMA);
1790 if (!lex_force_num_range_closed (lexer, "PTILE", 0, 100))
1792 cat.percentile = lex_number (lexer);
1796 if (!lex_force_match (lexer, T_RPAREN))
1799 else if (ctables_function_availability (cat.sort_function)
1802 bool UNUSED b = lex_force_match (lexer, T_LPAREN);
1807 else if (!c->n_cats && lex_match_id (lexer, "MISSING"))
1809 lex_match (lexer, T_EQUALS);
1810 if (lex_match_id (lexer, "INCLUDE"))
1811 cat.include_missing = true;
1812 else if (lex_match_id (lexer, "EXCLUDE"))
1813 cat.include_missing = false;
1816 lex_error_expecting (lexer, "INCLUDE", "EXCLUDE");
1820 else if (lex_match_id (lexer, "TOTAL"))
1822 lex_match (lexer, T_EQUALS);
1823 if (!parse_bool (lexer, &show_totals))
1826 else if (lex_match_id (lexer, "LABEL"))
1828 lex_match (lexer, T_EQUALS);
1829 if (!lex_force_string (lexer))
1832 total_label = ss_xstrdup (lex_tokss (lexer));
1835 else if (lex_match_id (lexer, "POSITION"))
1837 lex_match (lexer, T_EQUALS);
1838 if (lex_match_id (lexer, "BEFORE"))
1839 totals_before = true;
1840 else if (lex_match_id (lexer, "AFTER"))
1841 totals_before = false;
1844 lex_error_expecting (lexer, "BEFORE", "AFTER");
1848 else if (lex_match_id (lexer, "EMPTY"))
1850 lex_match (lexer, T_EQUALS);
1851 if (lex_match_id (lexer, "INCLUDE"))
1852 c->show_empty = true;
1853 else if (lex_match_id (lexer, "EXCLUDE"))
1854 c->show_empty = false;
1857 lex_error_expecting (lexer, "INCLUDE", "EXCLUDE");
1864 lex_error_expecting (lexer, "ORDER", "KEY", "MISSING",
1865 "TOTAL", "LABEL", "POSITION", "EMPTY");
1867 lex_error_expecting (lexer, "TOTAL", "LABEL", "POSITION", "EMPTY");
1874 if (c->n_cats >= allocated_cats)
1875 c->cats = x2nrealloc (c->cats, &allocated_cats, sizeof *c->cats);
1876 c->cats[c->n_cats++] = cat;
1881 if (c->n_cats >= allocated_cats)
1882 c->cats = x2nrealloc (c->cats, &allocated_cats, sizeof *c->cats);
1884 struct ctables_category *totals;
1887 insert_element (c->cats, c->n_cats, sizeof *c->cats, 0);
1888 totals = &c->cats[0];
1891 totals = &c->cats[c->n_cats];
1894 *totals = (struct ctables_category) {
1896 .total_label = total_label ? total_label : xstrdup (_("Total")),
1900 struct ctables_category *subtotal = NULL;
1901 for (size_t i = totals_before ? 0 : c->n_cats;
1902 totals_before ? i < c->n_cats : i-- > 0;
1903 totals_before ? i++ : 0)
1905 struct ctables_category *cat = &c->cats[i];
1913 cat->subtotal = subtotal;
1916 case CCT_POSTCOMPUTE:
1935 ctables_nest_uninit (struct ctables_nest *nest)
1942 ctables_stack_uninit (struct ctables_stack *stack)
1946 for (size_t i = 0; i < stack->n; i++)
1947 ctables_nest_uninit (&stack->nests[i]);
1948 free (stack->nests);
1952 static struct ctables_stack
1953 nest_fts (struct ctables_stack s0, struct ctables_stack s1)
1960 struct ctables_stack stack = { .nests = xnmalloc (s0.n, s1.n * sizeof *stack.nests) };
1961 for (size_t i = 0; i < s0.n; i++)
1962 for (size_t j = 0; j < s1.n; j++)
1964 const struct ctables_nest *a = &s0.nests[i];
1965 const struct ctables_nest *b = &s1.nests[j];
1967 size_t allocate = a->n + b->n;
1968 struct variable **vars = xnmalloc (allocate, sizeof *vars);
1969 enum pivot_axis_type *axes = xnmalloc (allocate, sizeof *axes);
1971 for (size_t k = 0; k < a->n; k++)
1972 vars[n++] = a->vars[k];
1973 for (size_t k = 0; k < b->n; k++)
1974 vars[n++] = b->vars[k];
1975 assert (n == allocate);
1977 const struct ctables_nest *summary_src;
1978 if (!a->specs[CSV_CELL].var)
1980 else if (!b->specs[CSV_CELL].var)
1985 struct ctables_nest *new = &stack.nests[stack.n++];
1986 *new = (struct ctables_nest) {
1988 .scale_idx = (a->scale_idx != SIZE_MAX ? a->scale_idx
1989 : b->scale_idx != SIZE_MAX ? a->n + b->scale_idx
1993 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
1994 ctables_summary_spec_set_clone (&new->specs[sv], &summary_src->specs[sv]);
1996 ctables_stack_uninit (&s0);
1997 ctables_stack_uninit (&s1);
2001 static struct ctables_stack
2002 stack_fts (struct ctables_stack s0, struct ctables_stack s1)
2004 struct ctables_stack stack = { .nests = xnmalloc (s0.n + s1.n, sizeof *stack.nests) };
2005 for (size_t i = 0; i < s0.n; i++)
2006 stack.nests[stack.n++] = s0.nests[i];
2007 for (size_t i = 0; i < s1.n; i++)
2008 stack.nests[stack.n++] = s1.nests[i];
2009 assert (stack.n == s0.n + s1.n);
2015 static struct ctables_stack
2016 enumerate_fts (enum pivot_axis_type axis_type, const struct ctables_axis *a)
2019 return (struct ctables_stack) { .n = 0 };
2024 assert (!a->var.is_mrset);
2026 struct variable **vars = xmalloc (sizeof *vars);
2029 struct ctables_nest *nest = xmalloc (sizeof *nest);
2030 *nest = (struct ctables_nest) {
2033 .scale_idx = a->scale ? 0 : SIZE_MAX,
2035 if (a->specs[CSV_CELL].n || a->scale)
2036 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
2038 ctables_summary_spec_set_clone (&nest->specs[sv], &a->specs[sv]);
2039 nest->specs[sv].var = a->var.var;
2041 return (struct ctables_stack) { .nests = nest, .n = 1 };
2044 return stack_fts (enumerate_fts (axis_type, a->subs[0]),
2045 enumerate_fts (axis_type, a->subs[1]));
2048 return nest_fts (enumerate_fts (axis_type, a->subs[0]),
2049 enumerate_fts (axis_type, a->subs[1]));
2055 union ctables_summary
2057 /* COUNT, VALIDN, TOTALN. */
2064 /* MINIMUM, MAXIMUM, RANGE. */
2071 /* MEAN, SEMEAN, STDDEV, SUM, VARIANCE, *.SUM. */
2072 struct moments1 *moments;
2074 /* MEDIAN, MODE, PTILE. */
2077 struct casewriter *writer;
2082 /* XXX multiple response */
2086 ctables_summary_init (union ctables_summary *s,
2087 const struct ctables_summary_spec *ss)
2089 switch (ss->function)
2093 case CTSF_ROWPCT_COUNT:
2094 case CTSF_COLPCT_COUNT:
2095 case CTSF_TABLEPCT_COUNT:
2096 case CTSF_SUBTABLEPCT_COUNT:
2097 case CTSF_LAYERPCT_COUNT:
2098 case CTSF_LAYERROWPCT_COUNT:
2099 case CTSF_LAYERCOLPCT_COUNT:
2100 case CTSF_ROWPCT_VALIDN:
2101 case CTSF_COLPCT_VALIDN:
2102 case CTSF_TABLEPCT_VALIDN:
2103 case CTSF_SUBTABLEPCT_VALIDN:
2104 case CTSF_LAYERPCT_VALIDN:
2105 case CTSF_LAYERROWPCT_VALIDN:
2106 case CTSF_LAYERCOLPCT_VALIDN:
2107 case CTSF_ROWPCT_TOTALN:
2108 case CTSF_COLPCT_TOTALN:
2109 case CTSF_TABLEPCT_TOTALN:
2110 case CTSF_SUBTABLEPCT_TOTALN:
2111 case CTSF_LAYERPCT_TOTALN:
2112 case CTSF_LAYERROWPCT_TOTALN:
2113 case CTSF_LAYERCOLPCT_TOTALN:
2119 s->missing = s->valid = 0;
2125 s->min = s->max = SYSMIS;
2133 case CTSF_ROWPCT_SUM:
2134 case CTSF_COLPCT_SUM:
2135 case CTSF_TABLEPCT_SUM:
2136 case CTSF_SUBTABLEPCT_SUM:
2137 case CTSF_LAYERPCT_SUM:
2138 case CTSF_LAYERROWPCT_SUM:
2139 case CTSF_LAYERCOLPCT_SUM:
2140 s->moments = moments1_create (MOMENT_VARIANCE);
2147 struct caseproto *proto = caseproto_create ();
2148 proto = caseproto_add_width (proto, 0);
2149 proto = caseproto_add_width (proto, 0);
2151 struct subcase ordering;
2152 subcase_init (&ordering, 0, 0, SC_ASCEND);
2153 s->writer = sort_create_writer (&ordering, proto);
2154 subcase_uninit (&ordering);
2155 caseproto_unref (proto);
2165 ctables_summary_uninit (union ctables_summary *s,
2166 const struct ctables_summary_spec *ss)
2168 switch (ss->function)
2172 case CTSF_ROWPCT_COUNT:
2173 case CTSF_COLPCT_COUNT:
2174 case CTSF_TABLEPCT_COUNT:
2175 case CTSF_SUBTABLEPCT_COUNT:
2176 case CTSF_LAYERPCT_COUNT:
2177 case CTSF_LAYERROWPCT_COUNT:
2178 case CTSF_LAYERCOLPCT_COUNT:
2179 case CTSF_ROWPCT_VALIDN:
2180 case CTSF_COLPCT_VALIDN:
2181 case CTSF_TABLEPCT_VALIDN:
2182 case CTSF_SUBTABLEPCT_VALIDN:
2183 case CTSF_LAYERPCT_VALIDN:
2184 case CTSF_LAYERROWPCT_VALIDN:
2185 case CTSF_LAYERCOLPCT_VALIDN:
2186 case CTSF_ROWPCT_TOTALN:
2187 case CTSF_COLPCT_TOTALN:
2188 case CTSF_TABLEPCT_TOTALN:
2189 case CTSF_SUBTABLEPCT_TOTALN:
2190 case CTSF_LAYERPCT_TOTALN:
2191 case CTSF_LAYERROWPCT_TOTALN:
2192 case CTSF_LAYERCOLPCT_TOTALN:
2210 case CTSF_ROWPCT_SUM:
2211 case CTSF_COLPCT_SUM:
2212 case CTSF_TABLEPCT_SUM:
2213 case CTSF_SUBTABLEPCT_SUM:
2214 case CTSF_LAYERPCT_SUM:
2215 case CTSF_LAYERROWPCT_SUM:
2216 case CTSF_LAYERCOLPCT_SUM:
2217 moments1_destroy (s->moments);
2223 casewriter_destroy (s->writer);
2229 ctables_summary_add (union ctables_summary *s,
2230 const struct ctables_summary_spec *ss,
2231 const struct variable *var, const union value *value,
2232 double d_weight, double e_weight)
2234 switch (ss->function)
2239 if (var_is_value_missing (var, value))
2240 s->missing += d_weight;
2242 s->valid += d_weight;
2246 case CTSF_ROWPCT_COUNT:
2247 case CTSF_COLPCT_COUNT:
2248 case CTSF_TABLEPCT_COUNT:
2249 case CTSF_SUBTABLEPCT_COUNT:
2250 case CTSF_LAYERPCT_COUNT:
2251 case CTSF_LAYERROWPCT_COUNT:
2252 case CTSF_LAYERCOLPCT_COUNT:
2253 case CTSF_ROWPCT_VALIDN:
2254 case CTSF_COLPCT_VALIDN:
2255 case CTSF_TABLEPCT_VALIDN:
2256 case CTSF_SUBTABLEPCT_VALIDN:
2257 case CTSF_LAYERPCT_VALIDN:
2258 case CTSF_LAYERROWPCT_VALIDN:
2259 case CTSF_LAYERCOLPCT_VALIDN:
2260 case CTSF_ROWPCT_TOTALN:
2261 case CTSF_COLPCT_TOTALN:
2262 case CTSF_TABLEPCT_TOTALN:
2263 case CTSF_SUBTABLEPCT_TOTALN:
2264 case CTSF_LAYERPCT_TOTALN:
2265 case CTSF_LAYERROWPCT_TOTALN:
2266 case CTSF_LAYERCOLPCT_TOTALN:
2270 if (var_is_value_missing (var, value))
2271 s->missing += e_weight;
2273 s->valid += e_weight;
2279 if (!var_is_value_missing (var, value))
2281 assert (!var_is_alpha (var)); /* XXX? */
2282 if (s->min == SYSMIS || value->f < s->min)
2284 if (s->max == SYSMIS || value->f > s->max)
2294 case CTSF_ROWPCT_SUM:
2295 case CTSF_COLPCT_SUM:
2296 case CTSF_TABLEPCT_SUM:
2297 case CTSF_SUBTABLEPCT_SUM:
2298 case CTSF_LAYERPCT_SUM:
2299 case CTSF_LAYERROWPCT_SUM:
2300 case CTSF_LAYERCOLPCT_SUM:
2301 if (!var_is_value_missing (var, value))
2302 moments1_add (s->moments, value->f, e_weight);
2308 if (var_is_value_missing (var, value))
2310 s->ovalid += e_weight;
2312 struct ccase *c = case_create (casewriter_get_proto (s->writer));
2313 *case_num_rw_idx (c, 0) = value->f;
2314 *case_num_rw_idx (c, 1) = e_weight;
2315 casewriter_write (s->writer, c);
2321 static enum ctables_domain_type
2322 ctables_function_domain (enum ctables_summary_function function)
2346 case CTSF_COLPCT_COUNT:
2347 case CTSF_COLPCT_SUM:
2348 case CTSF_COLPCT_TOTALN:
2349 case CTSF_COLPCT_VALIDN:
2352 case CTSF_LAYERCOLPCT_COUNT:
2353 case CTSF_LAYERCOLPCT_SUM:
2354 case CTSF_LAYERCOLPCT_TOTALN:
2355 case CTSF_LAYERCOLPCT_VALIDN:
2356 return CTDT_LAYERCOL;
2358 case CTSF_LAYERPCT_COUNT:
2359 case CTSF_LAYERPCT_SUM:
2360 case CTSF_LAYERPCT_TOTALN:
2361 case CTSF_LAYERPCT_VALIDN:
2364 case CTSF_LAYERROWPCT_COUNT:
2365 case CTSF_LAYERROWPCT_SUM:
2366 case CTSF_LAYERROWPCT_TOTALN:
2367 case CTSF_LAYERROWPCT_VALIDN:
2368 return CTDT_LAYERROW;
2370 case CTSF_ROWPCT_COUNT:
2371 case CTSF_ROWPCT_SUM:
2372 case CTSF_ROWPCT_TOTALN:
2373 case CTSF_ROWPCT_VALIDN:
2376 case CTSF_SUBTABLEPCT_COUNT:
2377 case CTSF_SUBTABLEPCT_SUM:
2378 case CTSF_SUBTABLEPCT_TOTALN:
2379 case CTSF_SUBTABLEPCT_VALIDN:
2380 return CTDT_SUBTABLE;
2382 case CTSF_TABLEPCT_COUNT:
2383 case CTSF_TABLEPCT_SUM:
2384 case CTSF_TABLEPCT_TOTALN:
2385 case CTSF_TABLEPCT_VALIDN:
2393 ctables_summary_value (const struct ctables_cell *cell,
2394 union ctables_summary *s,
2395 const struct ctables_summary_spec *ss)
2397 switch (ss->function)
2403 case CTSF_ROWPCT_COUNT:
2404 case CTSF_COLPCT_COUNT:
2405 case CTSF_TABLEPCT_COUNT:
2406 case CTSF_SUBTABLEPCT_COUNT:
2407 case CTSF_LAYERPCT_COUNT:
2408 case CTSF_LAYERROWPCT_COUNT:
2409 case CTSF_LAYERCOLPCT_COUNT:
2411 enum ctables_domain_type d = ctables_function_domain (ss->function);
2412 return (cell->domains[d]->e_valid
2413 ? s->valid / cell->domains[d]->e_valid * 100
2417 case CTSF_ROWPCT_VALIDN:
2418 case CTSF_COLPCT_VALIDN:
2419 case CTSF_TABLEPCT_VALIDN:
2420 case CTSF_SUBTABLEPCT_VALIDN:
2421 case CTSF_LAYERPCT_VALIDN:
2422 case CTSF_LAYERROWPCT_VALIDN:
2423 case CTSF_LAYERCOLPCT_VALIDN:
2424 case CTSF_ROWPCT_TOTALN:
2425 case CTSF_COLPCT_TOTALN:
2426 case CTSF_TABLEPCT_TOTALN:
2427 case CTSF_SUBTABLEPCT_TOTALN:
2428 case CTSF_LAYERPCT_TOTALN:
2429 case CTSF_LAYERROWPCT_TOTALN:
2430 case CTSF_LAYERCOLPCT_TOTALN:
2438 return s->valid + s->missing;
2451 return s->max != SYSMIS && s->min != SYSMIS ? s->max - s->min : SYSMIS;
2456 moments1_calculate (s->moments, NULL, &mean, NULL, NULL, NULL);
2462 double weight, variance;
2463 moments1_calculate (s->moments, &weight, NULL, &variance, NULL, NULL);
2464 return calc_semean (variance, weight);
2470 moments1_calculate (s->moments, NULL, NULL, &variance, NULL, NULL);
2471 return variance != SYSMIS ? sqrt (variance) : SYSMIS;
2476 double weight, mean;
2477 moments1_calculate (s->moments, &weight, &mean, NULL, NULL, NULL);
2478 return weight != SYSMIS && mean != SYSMIS ? weight * mean : SYSMIS;
2484 moments1_calculate (s->moments, NULL, NULL, &variance, NULL, NULL);
2488 case CTSF_ROWPCT_SUM:
2489 case CTSF_COLPCT_SUM:
2490 case CTSF_TABLEPCT_SUM:
2491 case CTSF_SUBTABLEPCT_SUM:
2492 case CTSF_LAYERPCT_SUM:
2493 case CTSF_LAYERROWPCT_SUM:
2494 case CTSF_LAYERCOLPCT_SUM:
2501 struct casereader *reader = casewriter_make_reader (s->writer);
2504 struct percentile *ptile = percentile_create (
2505 ss->function == CTSF_PTILE ? ss->percentile : 0.5, s->ovalid);
2506 struct order_stats *os = &ptile->parent;
2507 order_stats_accumulate_idx (&os, 1, reader, 1, 0);
2508 s->ovalue = percentile_calculate (ptile, PC_HAVERAGE);
2509 statistic_destroy (&ptile->parent.parent);
2516 struct casereader *reader = casewriter_make_reader (s->writer);
2519 struct mode *mode = mode_create ();
2520 struct order_stats *os = &mode->parent;
2521 order_stats_accumulate_idx (&os, 1, reader, 1, 0);
2522 s->ovalue = mode->mode;
2523 statistic_destroy (&mode->parent.parent);
2531 struct ctables_cell_sort_aux
2533 const struct ctables_nest *nest;
2534 enum pivot_axis_type a;
2538 ctables_cell_compare_3way (const void *a_, const void *b_, const void *aux_)
2540 const struct ctables_cell_sort_aux *aux = aux_;
2541 struct ctables_cell *const *ap = a_;
2542 struct ctables_cell *const *bp = b_;
2543 const struct ctables_cell *a = *ap;
2544 const struct ctables_cell *b = *bp;
2546 const struct ctables_nest *nest = aux->nest;
2547 for (size_t i = 0; i < nest->n; i++)
2548 if (i != nest->scale_idx)
2550 const struct variable *var = nest->vars[i];
2551 const struct ctables_cell_value *a_cv = &a->axes[aux->a].cvs[i];
2552 const struct ctables_cell_value *b_cv = &b->axes[aux->a].cvs[i];
2553 if (a_cv->category != b_cv->category)
2554 return a_cv->category > b_cv->category ? 1 : -1;
2556 const union value *a_val = &a_cv->value;
2557 const union value *b_val = &b_cv->value;
2558 switch (a_cv->category->type)
2564 case CCT_POSTCOMPUTE:
2565 /* Must be equal. */
2572 int cmp = value_compare_3way (a_val, b_val, var_get_width (var));
2580 int cmp = value_compare_3way (a_val, b_val, var_get_width (var));
2582 return a_cv->category->sort_ascending ? cmp : -cmp;
2588 const char *a_label = var_lookup_value_label (var, a_val);
2589 const char *b_label = var_lookup_value_label (var, b_val);
2591 ? (b_label ? strcmp (a_label, b_label) : 1)
2592 : (b_label ? -1 : value_compare_3way (
2593 a_val, b_val, var_get_width (var))));
2595 return a_cv->category->sort_ascending ? cmp : -cmp;
2609 For each ctables_table:
2610 For each combination of row vars:
2611 For each combination of column vars:
2612 For each combination of layer vars:
2614 Make a table of row values:
2615 Sort entries by row values
2616 Assign a 0-based index to each actual value
2617 Construct a dimension
2618 Make a table of column values
2619 Make a table of layer values
2621 Fill the table entry using the indexes from before.
2624 static struct ctables_domain *
2625 ctables_domain_insert (struct ctables_section *s, struct ctables_cell *cell,
2626 enum ctables_domain_type domain)
2629 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2631 const struct ctables_nest *nest = s->nests[a];
2632 for (size_t i = 0; i < nest->n_domains[domain]; i++)
2634 size_t v_idx = nest->domains[domain][i];
2635 hash = value_hash (&cell->axes[a].cvs[v_idx].value,
2636 var_get_width (nest->vars[v_idx]), hash);
2640 struct ctables_domain *d;
2641 HMAP_FOR_EACH_WITH_HASH (d, struct ctables_domain, node, hash, &s->domains[domain])
2643 const struct ctables_cell *df = d->example;
2644 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2646 const struct ctables_nest *nest = s->nests[a];
2647 for (size_t i = 0; i < nest->n_domains[domain]; i++)
2649 size_t v_idx = nest->domains[domain][i];
2650 if (!value_equal (&df->axes[a].cvs[v_idx].value,
2651 &cell->axes[a].cvs[v_idx].value,
2652 var_get_width (nest->vars[v_idx])))
2661 d = xmalloc (sizeof *d);
2662 *d = (struct ctables_domain) { .example = cell };
2663 hmap_insert (&s->domains[domain], &d->node, hash);
2667 static const struct ctables_category *
2668 ctables_categories_match (const struct ctables_categories *c,
2669 const union value *v, const struct variable *var)
2671 const struct ctables_category *othernm = NULL;
2672 for (size_t i = c->n_cats; i-- > 0; )
2674 const struct ctables_category *cat = &c->cats[i];
2678 if (cat->number == v->f)
2686 if ((cat->range[0] == -DBL_MAX || v->f >= cat->range[0])
2687 && (cat->range[1] == DBL_MAX || v->f <= cat->range[1]))
2692 if (var_is_value_missing (var, v))
2696 case CCT_POSTCOMPUTE:
2711 return (cat->include_missing || !var_is_value_missing (var, v) ? cat
2716 return var_is_value_missing (var, v) ? NULL : othernm;
2719 static const struct ctables_category *
2720 ctables_categories_total (const struct ctables_categories *c)
2722 const struct ctables_category *first = &c->cats[0];
2723 const struct ctables_category *last = &c->cats[c->n_cats - 1];
2724 return (first->type == CCT_TOTAL ? first
2725 : last->type == CCT_TOTAL ? last
2729 static struct ctables_cell *
2730 ctables_cell_insert__ (struct ctables_section *s, const struct ccase *c,
2731 const struct ctables_category *cats[PIVOT_N_AXES][10])
2734 enum ctables_summary_variant sv = CSV_CELL;
2735 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2737 const struct ctables_nest *nest = s->nests[a];
2738 for (size_t i = 0; i < nest->n; i++)
2739 if (i != nest->scale_idx)
2741 hash = hash_pointer (cats[a][i], hash);
2742 if (cats[a][i]->type != CCT_TOTAL
2743 && cats[a][i]->type != CCT_SUBTOTAL
2744 && cats[a][i]->type != CCT_POSTCOMPUTE)
2745 hash = value_hash (case_data (c, nest->vars[i]),
2746 var_get_width (nest->vars[i]), hash);
2752 struct ctables_cell *cell;
2753 HMAP_FOR_EACH_WITH_HASH (cell, struct ctables_cell, node, hash, &s->cells)
2755 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2757 const struct ctables_nest *nest = s->nests[a];
2758 for (size_t i = 0; i < nest->n; i++)
2759 if (i != nest->scale_idx
2760 && (cats[a][i] != cell->axes[a].cvs[i].category
2761 || (cats[a][i]->type != CCT_TOTAL
2762 && cats[a][i]->type != CCT_SUBTOTAL
2763 && cats[a][i]->type != CCT_POSTCOMPUTE
2764 && !value_equal (case_data (c, nest->vars[i]),
2765 &cell->axes[a].cvs[i].value,
2766 var_get_width (nest->vars[i])))))
2775 cell = xmalloc (sizeof *cell);
2778 cell->contributes_to_domains = true;
2779 cell->postcompute = false;
2780 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2782 const struct ctables_nest *nest = s->nests[a];
2783 cell->axes[a].cvs = (nest->n
2784 ? xnmalloc (nest->n, sizeof *cell->axes[a].cvs)
2786 for (size_t i = 0; i < nest->n; i++)
2788 const struct ctables_category *cat = cats[a][i];
2789 if (i != nest->scale_idx)
2791 const struct ctables_category *subtotal = cat->subtotal;
2792 if (cat->hide || (subtotal && subtotal->hide_subcategories))
2795 if (cat->type == CCT_TOTAL
2796 || cat->type == CCT_SUBTOTAL
2797 || cat->type == CCT_POSTCOMPUTE)
2798 cell->contributes_to_domains = false;
2799 if (cat->type == CCT_POSTCOMPUTE)
2800 cell->postcompute = true;
2803 cell->axes[a].cvs[i].category = cat;
2804 value_clone (&cell->axes[a].cvs[i].value, case_data (c, nest->vars[i]),
2805 var_get_width (nest->vars[i]));
2809 const struct ctables_nest *ss = s->nests[s->table->summary_axis];
2810 const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv];
2811 cell->summaries = xmalloc (specs->n * sizeof *cell->summaries);
2812 for (size_t i = 0; i < specs->n; i++)
2813 ctables_summary_init (&cell->summaries[i], &specs->specs[i]);
2814 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
2815 cell->domains[dt] = ctables_domain_insert (s, cell, dt);
2816 hmap_insert (&s->cells, &cell->node, hash);
2821 ctables_cell_add__ (struct ctables_section *s, const struct ccase *c,
2822 const struct ctables_category *cats[PIVOT_N_AXES][10],
2823 double d_weight, double e_weight)
2825 struct ctables_cell *cell = ctables_cell_insert__ (s, c, cats);
2826 const struct ctables_nest *ss = s->nests[s->table->summary_axis];
2828 const struct ctables_summary_spec_set *specs = &ss->specs[cell->sv];
2829 for (size_t i = 0; i < specs->n; i++)
2830 ctables_summary_add (&cell->summaries[i], &specs->specs[i], specs->var,
2831 case_data (c, specs->var), d_weight, e_weight);
2832 if (cell->contributes_to_domains)
2834 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
2836 cell->domains[dt]->d_valid += d_weight;
2837 cell->domains[dt]->e_valid += e_weight;
2843 recurse_totals (struct ctables_section *s, const struct ccase *c,
2844 const struct ctables_category *cats[PIVOT_N_AXES][10],
2845 double d_weight, double e_weight,
2846 enum pivot_axis_type start_axis, size_t start_nest)
2848 for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
2850 const struct ctables_nest *nest = s->nests[a];
2851 for (size_t i = start_nest; i < nest->n; i++)
2853 if (i == nest->scale_idx)
2856 const struct variable *var = nest->vars[i];
2858 const struct ctables_category *total = ctables_categories_total (
2859 s->table->categories[var_get_dict_index (var)]);
2862 const struct ctables_category *save = cats[a][i];
2864 ctables_cell_add__ (s, c, cats, d_weight, e_weight);
2865 recurse_totals (s, c, cats, d_weight, e_weight, a, i + 1);
2874 recurse_subtotals (struct ctables_section *s, const struct ccase *c,
2875 const struct ctables_category *cats[PIVOT_N_AXES][10],
2876 double d_weight, double e_weight,
2877 enum pivot_axis_type start_axis, size_t start_nest)
2879 for (enum pivot_axis_type a = start_axis; a < PIVOT_N_AXES; a++)
2881 const struct ctables_nest *nest = s->nests[a];
2882 for (size_t i = start_nest; i < nest->n; i++)
2884 if (i == nest->scale_idx)
2887 const struct ctables_category *save = cats[a][i];
2890 cats[a][i] = save->subtotal;
2891 ctables_cell_add__ (s, c, cats, d_weight, e_weight);
2892 recurse_subtotals (s, c, cats, d_weight, e_weight, a, i + 1);
2901 ctables_add_occurrence (const struct variable *var,
2902 const union value *value,
2903 struct hmap *occurrences)
2905 int width = var_get_width (var);
2906 unsigned int hash = value_hash (value, width, 0);
2908 struct ctables_occurrence *o;
2909 HMAP_FOR_EACH_WITH_HASH (o, struct ctables_occurrence, node, hash,
2911 if (value_equal (value, &o->value, width))
2914 o = xmalloc (sizeof *o);
2915 value_clone (&o->value, value, width);
2916 hmap_insert (occurrences, &o->node, hash);
2920 ctables_cell_insert (struct ctables_section *s,
2921 const struct ccase *c,
2922 double d_weight, double e_weight)
2924 const struct ctables_category *cats[PIVOT_N_AXES][10]; /* XXX */
2925 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2927 const struct ctables_nest *nest = s->nests[a];
2928 for (size_t i = 0; i < nest->n; i++)
2930 if (i == nest->scale_idx)
2933 const struct variable *var = nest->vars[i];
2934 const union value *value = case_data (c, var);
2936 if (var_is_numeric (var) && value->f == SYSMIS)
2939 cats[a][i] = ctables_categories_match (
2940 s->table->categories[var_get_dict_index (var)], value, var);
2946 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
2948 const struct ctables_nest *nest = s->nests[a];
2949 for (size_t i = 0; i < nest->n; i++)
2950 if (i != nest->scale_idx)
2952 const struct variable *var = nest->vars[i];
2953 const union value *value = case_data (c, var);
2954 ctables_add_occurrence (var, value, &s->occurrences[a][i]);
2958 ctables_cell_add__ (s, c, cats, d_weight, e_weight);
2960 recurse_totals (s, c, cats, d_weight, e_weight, 0, 0);
2961 recurse_subtotals (s, c, cats, d_weight, e_weight, 0, 0);
2966 const struct ctables_summary_spec_set *set;
2971 merge_item_compare_3way (const struct merge_item *a, const struct merge_item *b)
2973 const struct ctables_summary_spec *as = &a->set->specs[a->ofs];
2974 const struct ctables_summary_spec *bs = &b->set->specs[b->ofs];
2975 if (as->function != bs->function)
2976 return as->function > bs->function ? 1 : -1;
2977 else if (as->percentile != bs->percentile)
2978 return as->percentile < bs->percentile ? 1 : -1;
2979 return strcmp (as->label, bs->label);
2982 static struct pivot_value *
2983 ctables_category_create_label (const struct ctables_category *cat,
2984 const struct variable *var,
2985 const union value *value)
2987 return (cat->type == CCT_TOTAL || cat->type == CCT_SUBTOTAL
2988 ? pivot_value_new_user_text (cat->total_label, SIZE_MAX)
2989 : cat->type == CCT_POSTCOMPUTE && cat->pc->label
2990 ? pivot_value_new_user_text (cat->pc->label, SIZE_MAX)
2991 : pivot_value_new_var_value (var, value));
2994 static struct ctables_value *
2995 ctables_value_find__ (struct ctables_table *t, const union value *value,
2996 int width, unsigned int hash)
2998 struct ctables_value *clv;
2999 HMAP_FOR_EACH_WITH_HASH (clv, struct ctables_value, node,
3000 hash, &t->clabels_values_map)
3001 if (value_equal (value, &clv->value, width))
3007 ctables_value_insert (struct ctables_table *t, const union value *value,
3010 unsigned int hash = value_hash (value, width, 0);
3011 struct ctables_value *clv = ctables_value_find__ (t, value, width, hash);
3014 clv = xmalloc (sizeof *clv);
3015 value_clone (&clv->value, value, width);
3016 hmap_insert (&t->clabels_values_map, &clv->node, hash);
3020 static struct ctables_value *
3021 ctables_value_find (struct ctables_table *t,
3022 const union value *value, int width)
3024 return ctables_value_find__ (t, value, width,
3025 value_hash (value, width, 0));
3029 ctables_table_add_section (struct ctables_table *t, enum pivot_axis_type a,
3030 size_t ix[PIVOT_N_AXES])
3032 if (a < PIVOT_N_AXES)
3034 size_t limit = MAX (t->stacks[a].n, 1);
3035 for (ix[a] = 0; ix[a] < limit; ix[a]++)
3036 ctables_table_add_section (t, a + 1, ix);
3040 struct ctables_section *s = &t->sections[t->n_sections++];
3041 *s = (struct ctables_section) {
3043 .cells = HMAP_INITIALIZER (s->cells),
3045 for (a = 0; a < PIVOT_N_AXES; a++)
3048 struct ctables_nest *nest = &t->stacks[a].nests[ix[a]];
3050 s->occurrences[a] = xnmalloc (nest->n, sizeof *s->occurrences[a]);
3051 for (size_t i = 0; i < nest->n; i++)
3052 hmap_init (&s->occurrences[a][i]);
3054 for (size_t i = 0; i < N_CTDTS; i++)
3055 hmap_init (&s->domains[i]);
3060 ctpo_add (double a, double b)
3066 ctpo_sub (double a, double b)
3072 ctpo_mul (double a, double b)
3078 ctpo_div (double a, double b)
3080 return b ? a / b : SYSMIS;
3084 ctpo_pow (double a, double b)
3086 int save_errno = errno;
3088 double result = pow (a, b);
3096 ctpo_neg (double a, double b UNUSED)
3101 struct ctables_pcexpr_evaluate_ctx
3103 const struct ctables_cell *cell;
3104 const struct ctables_section *section;
3105 const struct ctables_categories *cats;
3106 enum pivot_axis_type pc_a;
3110 static double ctables_pcexpr_evaluate (
3111 const struct ctables_pcexpr_evaluate_ctx *, const struct ctables_pcexpr *);
3114 ctables_pcexpr_evaluate_nonterminal (
3115 const struct ctables_pcexpr_evaluate_ctx *ctx,
3116 const struct ctables_pcexpr *e, size_t n_args,
3117 double evaluate (double, double))
3119 double args[2] = { 0, 0 };
3120 for (size_t i = 0; i < n_args; i++)
3122 args[i] = ctables_pcexpr_evaluate (ctx, e->subs[i]);
3123 if (!isfinite (args[i]) || args[i] == SYSMIS)
3126 return evaluate (args[0], args[1]);
3130 ctables_pcexpr_evaluate_category (const struct ctables_pcexpr_evaluate_ctx *ctx,
3131 const struct ctables_cell_value *pc_cv)
3133 const struct ctables_section *s = ctx->section;
3136 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3138 const struct ctables_nest *nest = s->nests[a];
3139 for (size_t i = 0; i < nest->n; i++)
3140 if (i != nest->scale_idx)
3142 const struct ctables_cell_value *cv
3143 = (a == ctx->pc_a && i == ctx->pc_a_idx ? pc_cv
3144 : &ctx->cell->axes[a].cvs[i]);
3145 hash = hash_pointer (cv->category, hash);
3146 if (cv->category->type != CCT_TOTAL
3147 && cv->category->type != CCT_SUBTOTAL
3148 && cv->category->type != CCT_POSTCOMPUTE)
3149 hash = value_hash (&cv->value,
3150 var_get_width (nest->vars[i]), hash);
3154 struct ctables_cell *tc;
3155 HMAP_FOR_EACH_WITH_HASH (tc, struct ctables_cell, node, hash, &s->cells)
3157 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3159 const struct ctables_nest *nest = s->nests[a];
3160 for (size_t i = 0; i < nest->n; i++)
3161 if (i != nest->scale_idx)
3163 const struct ctables_cell_value *p_cv
3164 = (a == ctx->pc_a && i == ctx->pc_a_idx ? pc_cv
3165 : &ctx->cell->axes[a].cvs[i]);
3166 const struct ctables_cell_value *t_cv = &tc->axes[a].cvs[i];
3167 if (p_cv->category != t_cv->category
3168 || (p_cv->category->type != CCT_TOTAL
3169 && p_cv->category->type != CCT_SUBTOTAL
3170 && p_cv->category->type != CCT_POSTCOMPUTE
3171 && !value_equal (&p_cv->value,
3173 var_get_width (nest->vars[i]))))
3185 const struct ctables_table *t = s->table;
3186 const struct ctables_nest *specs_nest = s->nests[t->summary_axis];
3187 const struct ctables_summary_spec_set *specs = &specs_nest->specs[tc->sv];
3188 size_t j = 0 /* XXX */;
3189 return ctables_summary_value (tc, &tc->summaries[j], &specs->specs[j]);
3193 ctables_pcexpr_evaluate (const struct ctables_pcexpr_evaluate_ctx *ctx,
3194 const struct ctables_pcexpr *e)
3201 case CTPO_CAT_RANGE:
3203 struct ctables_cell_value cv = {
3204 .category = ctables_find_category_for_postcompute (ctx->cats, e)
3206 assert (cv.category != NULL);
3208 struct hmap *occurrences = &ctx->section->occurrences[ctx->pc_a][ctx->pc_a_idx];
3209 const struct ctables_occurrence *o;
3212 const struct variable *var = ctx->section->nests[ctx->pc_a]->vars[ctx->pc_a_idx];
3213 HMAP_FOR_EACH (o, struct ctables_occurrence, node, occurrences)
3214 if (ctables_categories_match (ctx->cats, &o->value, var) == cv.category)
3216 cv.value = o->value;
3217 sum += ctables_pcexpr_evaluate_category (ctx, &cv);
3222 case CTPO_CAT_NUMBER:
3223 case CTPO_CAT_STRING:
3224 case CTPO_CAT_MISSING:
3225 case CTPO_CAT_OTHERNM:
3226 case CTPO_CAT_SUBTOTAL:
3227 case CTPO_CAT_TOTAL:
3229 struct ctables_cell_value cv = {
3230 .category = ctables_find_category_for_postcompute (ctx->cats, e),
3231 .value = { .f = e->number },
3233 assert (cv.category != NULL);
3234 return ctables_pcexpr_evaluate_category (ctx, &cv);
3238 return ctables_pcexpr_evaluate_nonterminal (ctx, e, 2, ctpo_add);
3241 return ctables_pcexpr_evaluate_nonterminal (ctx, e, 2, ctpo_sub);
3244 return ctables_pcexpr_evaluate_nonterminal (ctx, e, 2, ctpo_mul);
3247 return ctables_pcexpr_evaluate_nonterminal (ctx, e, 2, ctpo_div);
3250 return ctables_pcexpr_evaluate_nonterminal (ctx, e, 2, ctpo_pow);
3253 return ctables_pcexpr_evaluate_nonterminal (ctx, e, 1, ctpo_neg);
3260 ctables_cell_calculate_postcompute (const struct ctables_section *s,
3261 const struct ctables_cell *cell)
3263 enum pivot_axis_type pc_a;
3265 const struct ctables_postcompute *pc;
3266 for (pc_a = 0; ; pc_a++)
3268 assert (pc_a < PIVOT_N_AXES);
3269 for (pc_a_idx = 0; pc_a_idx < s->nests[pc_a]->n; pc_a_idx++)
3271 const struct ctables_cell_value *cv = &cell->axes[pc_a].cvs[pc_a_idx];
3272 if (cv->category->type == CCT_POSTCOMPUTE)
3274 pc = cv->category->pc;
3281 const struct variable *var = s->nests[pc_a]->vars[pc_a_idx];
3282 const struct ctables_categories *cats = s->table->categories[
3283 var_get_dict_index (var)];
3284 struct ctables_pcexpr_evaluate_ctx ctx = {
3289 .pc_a_idx = pc_a_idx,
3291 return ctables_pcexpr_evaluate (&ctx, pc->expr);
3295 ctables_table_output (struct ctables *ct, struct ctables_table *t)
3297 struct pivot_table *pt = pivot_table_create__ (
3299 ? pivot_value_new_user_text (t->title, SIZE_MAX)
3300 : pivot_value_new_text (N_("Custom Tables"))),
3303 pivot_table_set_caption (
3304 pt, pivot_value_new_user_text (t->caption, SIZE_MAX));
3306 pivot_table_set_caption (
3307 pt, pivot_value_new_user_text (t->corner, SIZE_MAX));
3309 bool summary_dimension = (t->summary_axis != t->slabels_axis
3310 || (!t->slabels_visible
3311 && t->summary_specs.n > 1));
3312 if (summary_dimension)
3314 struct pivot_dimension *d = pivot_dimension_create (
3315 pt, t->slabels_axis, N_("Statistics"));
3316 const struct ctables_summary_spec_set *specs = &t->summary_specs;
3317 if (!t->slabels_visible)
3318 d->hide_all_labels = true;
3319 for (size_t i = 0; i < specs->n; i++)
3320 pivot_category_create_leaf (
3321 d->root, pivot_value_new_text (specs->specs[i].label));
3324 bool categories_dimension = t->clabels_example != NULL;
3325 if (categories_dimension)
3327 struct pivot_dimension *d = pivot_dimension_create (
3328 pt, t->label_axis[t->clabels_from_axis],
3329 t->clabels_from_axis == PIVOT_AXIS_ROW
3330 ? N_("Row Categories")
3331 : N_("Column Categories"));
3332 const struct variable *var = t->clabels_example;
3333 const struct ctables_categories *c = t->categories[var_get_dict_index (var)];
3334 for (size_t i = 0; i < t->n_clabels_values; i++)
3336 const struct ctables_value *value = t->clabels_values[i];
3337 const struct ctables_category *cat = ctables_categories_match (c, &value->value, var);
3338 assert (cat != NULL);
3339 pivot_category_create_leaf (d->root, ctables_category_create_label (
3340 cat, t->clabels_example, &value->value));
3344 pivot_table_set_look (pt, ct->look);
3345 struct pivot_dimension *d[PIVOT_N_AXES];
3346 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3348 static const char *names[] = {
3349 [PIVOT_AXIS_ROW] = N_("Rows"),
3350 [PIVOT_AXIS_COLUMN] = N_("Columns"),
3351 [PIVOT_AXIS_LAYER] = N_("Layers"),
3353 d[a] = (t->axes[a] || a == t->summary_axis
3354 ? pivot_dimension_create (pt, a, names[a])
3359 assert (t->axes[a]);
3361 for (size_t i = 0; i < t->stacks[a].n; i++)
3363 struct ctables_nest *nest = &t->stacks[a].nests[i];
3364 struct ctables_section **sections = xnmalloc (t->n_sections,
3366 size_t n_sections = 0;
3368 size_t n_total_cells = 0;
3369 size_t max_depth = 0;
3370 for (size_t j = 0; j < t->n_sections; j++)
3371 if (t->sections[j].nests[a] == nest)
3373 struct ctables_section *s = &t->sections[j];
3374 sections[n_sections++] = s;
3375 n_total_cells += s->cells.count;
3377 size_t depth = s->nests[a]->n;
3378 max_depth = MAX (depth, max_depth);
3381 struct ctables_cell **sorted = xnmalloc (n_total_cells,
3383 size_t n_sorted = 0;
3385 for (size_t j = 0; j < n_sections; j++)
3387 struct ctables_section *s = sections[j];
3389 struct ctables_cell *cell;
3390 HMAP_FOR_EACH (cell, struct ctables_cell, node, &s->cells)
3392 sorted[n_sorted++] = cell;
3393 assert (n_sorted <= n_total_cells);
3396 struct ctables_cell_sort_aux aux = { .nest = nest, .a = a };
3397 sort (sorted, n_sorted, sizeof *sorted, ctables_cell_compare_3way, &aux);
3399 struct ctables_level
3401 enum ctables_level_type
3403 CTL_VAR, /* Variable label for nest->vars[var_idx]. */
3404 CTL_CATEGORY, /* Category for nest->vars[var_idx]. */
3405 CTL_SUMMARY, /* Summary functions. */
3409 enum settings_value_show vlabel; /* CTL_VAR only. */
3412 struct ctables_level *levels = xnmalloc (1 + 2 * max_depth, sizeof *levels);
3413 size_t n_levels = 0;
3414 for (size_t k = 0; k < nest->n; k++)
3416 enum ctables_vlabel vlabel = ct->vlabels[var_get_dict_index (nest->vars[k])];
3417 if (vlabel != CTVL_NONE)
3419 levels[n_levels++] = (struct ctables_level) {
3421 .vlabel = (enum settings_value_show) vlabel,
3426 if (nest->scale_idx != k
3427 && (k != nest->n - 1 || t->label_axis[a] == a))
3429 levels[n_levels++] = (struct ctables_level) {
3430 .type = CTL_CATEGORY,
3436 if (!summary_dimension && a == t->slabels_axis)
3438 levels[n_levels++] = (struct ctables_level) {
3439 .type = CTL_SUMMARY,
3440 .var_idx = SIZE_MAX,
3444 /* Pivot categories:
3446 - variable label for nest->vars[0], if vlabel != CTVL_NONE
3447 - category for nest->vars[0], if nest->scale_idx != 0
3448 - variable label for nest->vars[1], if vlabel != CTVL_NONE
3449 - category for nest->vars[1], if nest->scale_idx != 1
3451 - variable label for nest->vars[n - 1], if vlabel != CTVL_NONE
3452 - category for nest->vars[n - 1], if t->label_axis[a] == a && nest->scale_idx != n - 1.
3453 - summary function, if 'a == t->slabels_axis && a ==
3456 Additional dimensions:
3458 - If 'a == t->slabels_axis && a != t->summary_axis', add a summary
3460 - If 't->label_axis[b] == a' for some 'b != a', add a category
3465 struct pivot_category **groups = xnmalloc (1 + 2 * max_depth, sizeof *groups);
3467 for (size_t j = 0; j < n_sorted; j++)
3469 struct ctables_cell *cell = sorted[j];
3470 struct ctables_cell *prev = j > 0 ? sorted[j - 1] : NULL;
3472 size_t n_common = 0;
3475 for (; n_common < n_levels; n_common++)
3477 const struct ctables_level *level = &levels[n_common];
3478 if (level->type == CTL_CATEGORY)
3480 size_t var_idx = level->var_idx;
3481 const struct ctables_category *c = cell->axes[a].cvs[var_idx].category;
3482 if (prev->axes[a].cvs[var_idx].category != c)
3484 else if (c->type != CCT_SUBTOTAL
3485 && c->type != CCT_TOTAL
3486 && c->type != CCT_POSTCOMPUTE
3487 && !value_equal (&prev->axes[a].cvs[var_idx].value,
3488 &cell->axes[a].cvs[var_idx].value,
3489 var_get_type (nest->vars[var_idx])))
3495 for (size_t k = n_common; k < n_levels; k++)
3497 const struct ctables_level *level = &levels[k];
3498 struct pivot_category *parent = k ? groups[k - 1] : d[a]->root;
3499 if (level->type == CTL_SUMMARY)
3501 assert (k == n_levels - 1);
3503 const struct ctables_summary_spec_set *specs = &t->summary_specs;
3504 for (size_t m = 0; m < specs->n; m++)
3506 int leaf = pivot_category_create_leaf (
3507 parent, pivot_value_new_text (specs->specs[m].label));
3514 const struct variable *var = nest->vars[level->var_idx];
3515 struct pivot_value *label;
3516 if (level->type == CTL_VAR)
3518 label = pivot_value_new_variable (var);
3519 label->variable.show = level->vlabel;
3521 else if (level->type == CTL_CATEGORY)
3523 const struct ctables_cell_value *cv = &cell->axes[a].cvs[level->var_idx];
3524 label = ctables_category_create_label (cv->category,
3530 if (k == n_levels - 1)
3531 prev_leaf = pivot_category_create_leaf (parent, label);
3533 groups[k] = pivot_category_create_group__ (parent, label);
3537 cell->axes[a].leaf = prev_leaf;
3544 for (size_t i = 0; i < t->n_sections; i++)
3546 struct ctables_section *s = &t->sections[i];
3548 struct ctables_cell *cell;
3549 HMAP_FOR_EACH (cell, struct ctables_cell, node, &s->cells)
3554 const struct ctables_nest *specs_nest = s->nests[t->summary_axis];
3555 const struct ctables_summary_spec_set *specs = &specs_nest->specs[cell->sv];
3556 for (size_t j = 0; j < specs->n; j++)
3559 size_t n_dindexes = 0;
3561 if (summary_dimension)
3562 dindexes[n_dindexes++] = specs->specs[j].axis_idx;
3564 if (categories_dimension)
3566 const struct ctables_nest *clabels_nest = s->nests[t->clabels_from_axis];
3567 const struct variable *var = clabels_nest->vars[clabels_nest->n - 1];
3568 const union value *value = &cell->axes[t->clabels_from_axis].cvs[clabels_nest->n - 1].value;
3569 const struct ctables_value *ctv = ctables_value_find (t, value, var_get_width (var));
3572 dindexes[n_dindexes++] = ctv->leaf;
3575 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3578 int leaf = cell->axes[a].leaf;
3579 if (a == t->summary_axis && !summary_dimension)
3581 dindexes[n_dindexes++] = leaf;
3584 const struct ctables_summary_spec *ss = &specs->specs[j];
3586 double d = (cell->postcompute
3587 ? ctables_cell_calculate_postcompute (s, cell)
3588 : ctables_summary_value (cell, &cell->summaries[j], ss));
3589 struct pivot_value *value;
3590 if (ct->hide_threshold != 0
3591 && d < ct->hide_threshold
3592 && (cell->postcompute
3594 : ctables_summary_function_is_count (ss->function)))
3596 value = pivot_value_new_user_text_nocopy (
3597 xasprintf ("<%d", ct->hide_threshold));
3599 else if (d == 0 && ct->zero)
3600 value = pivot_value_new_user_text (ct->zero, SIZE_MAX);
3601 else if (d == SYSMIS && ct->missing)
3602 value = pivot_value_new_user_text (ct->missing, SIZE_MAX);
3603 else if (specs->specs[j].is_ctables_format)
3605 char *s = data_out_stretchy (&(union value) { .f = d },
3607 &specs->specs[j].format,
3608 &ct->ctables_formats, NULL);
3609 value = pivot_value_new_user_text_nocopy (s);
3613 value = pivot_value_new_number (d);
3614 value->numeric.format = specs->specs[j].format;
3616 pivot_table_put (pt, dindexes, n_dindexes, value);
3621 pivot_table_submit (pt);
3625 ctables_check_label_position (struct ctables_table *t, enum pivot_axis_type a)
3627 enum pivot_axis_type label_pos = t->label_axis[a];
3631 t->clabels_from_axis = a;
3633 const char *subcommand_name = a == PIVOT_AXIS_ROW ? "ROWLABELS" : "COLLABELS";
3634 const char *pos_name = label_pos == PIVOT_AXIS_LAYER ? "LAYER" : "OPPOSITE";
3636 const struct ctables_stack *stack = &t->stacks[a];
3640 const struct ctables_nest *n0 = &stack->nests[0];
3642 const struct variable *v0 = n0->vars[n0->n - 1];
3643 struct ctables_categories *c0 = t->categories[var_get_dict_index (v0)];
3644 t->clabels_example = v0;
3646 for (size_t i = 0; i < c0->n_cats; i++)
3647 if (c0->cats[i].type == CCT_FUNCTION)
3649 msg (SE, _("%s=%s is not allowed with sorting based "
3650 "on a summary function."),
3651 subcommand_name, pos_name);
3654 if (n0->n - 1 == n0->scale_idx)
3656 msg (SE, _("%s=%s requires the variables to be moved to be categorical, "
3657 "but %s is a scale variable."),
3658 subcommand_name, pos_name, var_get_name (v0));
3662 for (size_t i = 1; i < stack->n; i++)
3664 const struct ctables_nest *ni = &stack->nests[i];
3666 const struct variable *vi = ni->vars[ni->n - 1];
3667 struct ctables_categories *ci = t->categories[var_get_dict_index (vi)];
3669 if (ni->n - 1 == ni->scale_idx)
3671 msg (SE, _("%s=%s requires the variables to be moved to be "
3672 "categorical, but %s is a scale variable."),
3673 subcommand_name, pos_name, var_get_name (vi));
3676 if (var_get_width (v0) != var_get_width (vi))
3678 msg (SE, _("%s=%s requires the variables to be "
3679 "moved to have the same width, but %s has "
3680 "width %d and %s has width %d."),
3681 subcommand_name, pos_name,
3682 var_get_name (v0), var_get_width (v0),
3683 var_get_name (vi), var_get_width (vi));
3686 if (!val_labs_equal (var_get_value_labels (v0),
3687 var_get_value_labels (vi)))
3689 msg (SE, _("%s=%s requires the variables to be "
3690 "moved to have the same value labels, but %s "
3691 "and %s have different value labels."),
3692 subcommand_name, pos_name,
3693 var_get_name (v0), var_get_name (vi));
3696 if (!ctables_categories_equal (c0, ci))
3698 msg (SE, _("%s=%s requires the variables to be "
3699 "moved to have the same category "
3700 "specifications, but %s and %s have different "
3701 "category specifications."),
3702 subcommand_name, pos_name,
3703 var_get_name (v0), var_get_name (vi));
3712 ctables_prepare_table (struct ctables_table *t)
3714 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
3717 t->stacks[a] = enumerate_fts (a, t->axes[a]);
3719 for (size_t j = 0; j < t->stacks[a].n; j++)
3721 struct ctables_nest *nest = &t->stacks[a].nests[j];
3722 for (enum ctables_domain_type dt = 0; dt < N_CTDTS; dt++)
3724 nest->domains[dt] = xmalloc (nest->n * sizeof *nest->domains[dt]);
3725 nest->n_domains[dt] = 0;
3727 for (size_t k = 0; k < nest->n; k++)
3729 if (k == nest->scale_idx)
3738 if (a != PIVOT_AXIS_LAYER)
3745 if (dt == CTDT_SUBTABLE ? a != PIVOT_AXIS_LAYER
3746 : dt == CTDT_ROW ? a == PIVOT_AXIS_COLUMN
3747 : a == PIVOT_AXIS_ROW)
3749 if (k == nest->n - 1
3750 || (nest->scale_idx == nest->n - 1
3751 && k == nest->n - 2))
3757 if (a == PIVOT_AXIS_COLUMN)
3762 if (a == PIVOT_AXIS_ROW)
3767 nest->domains[dt][nest->n_domains[dt]++] = k;
3774 struct ctables_nest *nest = xmalloc (sizeof *nest);
3775 *nest = (struct ctables_nest) { .n = 0 };
3776 t->stacks[a] = (struct ctables_stack) { .nests = nest, .n = 1 };
3779 struct ctables_stack *stack = &t->stacks[t->summary_axis];
3780 for (size_t i = 0; i < stack->n; i++)
3782 struct ctables_nest *nest = &stack->nests[i];
3783 if (!nest->specs[CSV_CELL].n)
3785 struct ctables_summary_spec_set *specs = &nest->specs[CSV_CELL];
3786 specs->specs = xmalloc (sizeof *specs->specs);
3789 enum ctables_summary_function function
3790 = specs->var ? CTSF_MEAN : CTSF_COUNT;
3791 struct ctables_var var = { .is_mrset = false, .var = specs->var };
3793 *specs->specs = (struct ctables_summary_spec) {
3794 .function = function,
3795 .format = ctables_summary_default_format (function, &var),
3796 .label = ctables_summary_default_label (function, 0),
3799 specs->var = nest->vars[0];
3801 ctables_summary_spec_set_clone (&nest->specs[CSV_TOTAL],
3802 &nest->specs[CSV_CELL]);
3804 else if (!nest->specs[CSV_TOTAL].n)
3805 ctables_summary_spec_set_clone (&nest->specs[CSV_TOTAL],
3806 &nest->specs[CSV_CELL]);
3809 struct ctables_summary_spec_set *merged = &t->summary_specs;
3810 struct merge_item *items = xnmalloc (2 * stack->n, sizeof *items);
3812 for (size_t j = 0; j < stack->n; j++)
3814 const struct ctables_nest *nest = &stack->nests[j];
3816 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
3817 items[n_left++] = (struct merge_item) { .set = &nest->specs[sv] };
3822 struct merge_item min = items[0];
3823 for (size_t j = 1; j < n_left; j++)
3824 if (merge_item_compare_3way (&items[j], &min) < 0)
3827 if (merged->n >= merged->allocated)
3828 merged->specs = x2nrealloc (merged->specs, &merged->allocated,
3829 sizeof *merged->specs);
3830 merged->specs[merged->n++] = min.set->specs[min.ofs];
3832 for (size_t j = 0; j < n_left; )
3834 if (merge_item_compare_3way (&items[j], &min) == 0)
3836 struct merge_item *item = &items[j];
3837 item->set->specs[item->ofs].axis_idx = merged->n - 1;
3838 if (++item->ofs >= item->set->n)
3840 items[j] = items[--n_left];
3849 for (size_t j = 0; j < merged->n; j++)
3850 printf ("%s\n", ctables_summary_function_name (merged->specs[j].function));
3852 for (size_t j = 0; j < stack->n; j++)
3854 const struct ctables_nest *nest = &stack->nests[j];
3855 for (enum ctables_summary_variant sv = 0; sv < N_CSVS; sv++)
3857 const struct ctables_summary_spec_set *specs = &nest->specs[sv];
3858 for (size_t k = 0; k < specs->n; k++)
3859 printf ("(%s, %zu) ", ctables_summary_function_name (specs->specs[k].function),
3860 specs->specs[k].axis_idx);
3866 return (ctables_check_label_position (t, PIVOT_AXIS_ROW)
3867 && ctables_check_label_position (t, PIVOT_AXIS_COLUMN));
3871 ctables_insert_clabels_values (struct ctables_table *t, const struct ccase *c,
3872 enum pivot_axis_type a)
3874 struct ctables_stack *stack = &t->stacks[a];
3875 for (size_t i = 0; i < stack->n; i++)
3877 const struct ctables_nest *nest = &stack->nests[i];
3878 const struct variable *var = nest->vars[nest->n - 1];
3879 const union value *value = case_data (c, var);
3881 if (var_is_numeric (var) && value->f == SYSMIS)
3884 if (ctables_categories_match (t->categories [var_get_dict_index (var)],
3886 ctables_value_insert (t, value, var_get_width (var));
3891 compare_clabels_values_3way (const void *a_, const void *b_, const void *width_)
3893 const struct ctables_value *const *ap = a_;
3894 const struct ctables_value *const *bp = b_;
3895 const struct ctables_value *a = *ap;
3896 const struct ctables_value *b = *bp;
3897 const int *width = width_;
3898 return value_compare_3way (&a->value, &b->value, *width);
3902 ctables_sort_clabels_values (struct ctables_table *t)
3904 const struct variable *v0 = t->clabels_example;
3905 int width = var_get_width (v0);
3907 struct ctables_categories *c0 = t->categories[var_get_dict_index (v0)];
3910 const struct val_labs *val_labs = var_get_value_labels (v0);
3911 for (const struct val_lab *vl = val_labs_first (val_labs); vl;
3912 vl = val_labs_next (val_labs, vl))
3913 if (ctables_categories_match (c0, &vl->value, v0))
3914 ctables_value_insert (t, &vl->value, width);
3917 size_t n = hmap_count (&t->clabels_values_map);
3918 t->clabels_values = xnmalloc (n, sizeof *t->clabels_values);
3920 struct ctables_value *clv;
3922 HMAP_FOR_EACH (clv, struct ctables_value, node, &t->clabels_values_map)
3923 t->clabels_values[i++] = clv;
3924 t->n_clabels_values = n;
3927 sort (t->clabels_values, n, sizeof *t->clabels_values,
3928 compare_clabels_values_3way, &width);
3930 for (size_t i = 0; i < n; i++)
3931 t->clabels_values[i]->leaf = i;
3935 ctables_add_category_occurrences (const struct variable *var,
3936 struct hmap *occurrences,
3937 const struct ctables_categories *cats)
3939 const struct val_labs *val_labs = var_get_value_labels (var);
3941 for (size_t i = 0; i < cats->n_cats; i++)
3943 const struct ctables_category *c = &cats->cats[i];
3947 ctables_add_occurrence (var, &(const union value) { .f = c->number },
3955 assert (var_is_numeric (var));
3956 for (const struct val_lab *vl = val_labs_first (val_labs); vl;
3957 vl = val_labs_next (val_labs, vl))
3958 if (vl->value.f >= c->range[0] && vl->value.f <= c->range[1])
3959 ctables_add_occurrence (var, &vl->value, occurrences);
3963 for (const struct val_lab *vl = val_labs_first (val_labs); vl;
3964 vl = val_labs_next (val_labs, vl))
3965 if (var_is_value_missing (var, &vl->value))
3966 ctables_add_occurrence (var, &vl->value, occurrences);
3970 for (const struct val_lab *vl = val_labs_first (val_labs); vl;
3971 vl = val_labs_next (val_labs, vl))
3972 ctables_add_occurrence (var, &vl->value, occurrences);
3975 case CCT_POSTCOMPUTE:
3985 for (const struct val_lab *vl = val_labs_first (val_labs); vl;
3986 vl = val_labs_next (val_labs, vl))
3987 if (c->include_missing || !var_is_value_missing (var, &vl->value))
3988 ctables_add_occurrence (var, &vl->value, occurrences);
3995 ctables_section_recurse_add_empty_categories (
3996 struct ctables_section *s,
3997 const struct ctables_category *cats[PIVOT_N_AXES][10], struct ccase *c,
3998 enum pivot_axis_type a, size_t a_idx)
4000 if (a >= PIVOT_N_AXES)
4001 ctables_cell_insert__ (s, c, cats);
4002 else if (!s->nests[a] || a_idx >= s->nests[a]->n)
4003 ctables_section_recurse_add_empty_categories (s, cats, c, a + 1, 0);
4006 const struct variable *var = s->nests[a]->vars[a_idx];
4007 const struct ctables_categories *categories = s->table->categories[
4008 var_get_dict_index (var)];
4009 int width = var_get_width (var);
4010 const struct hmap *occurrences = &s->occurrences[a][a_idx];
4011 const struct ctables_occurrence *o;
4012 HMAP_FOR_EACH (o, struct ctables_occurrence, node, occurrences)
4014 union value *value = case_data_rw (c, var);
4015 value_destroy (value, width);
4016 value_clone (value, &o->value, width);
4017 cats[a][a_idx] = ctables_categories_match (categories, value, var);
4018 assert (cats[a][a_idx] != NULL);
4019 ctables_section_recurse_add_empty_categories (s, cats, c, a, a_idx + 1);
4022 for (size_t i = 0; i < categories->n_cats; i++)
4024 const struct ctables_category *cat = &categories->cats[i];
4025 if (cat->type == CCT_POSTCOMPUTE)
4027 cats[a][a_idx] = cat;
4028 ctables_section_recurse_add_empty_categories (s, cats, c, a, a_idx + 1);
4035 ctables_section_add_empty_categories (struct ctables_section *s)
4037 bool show_empty = false;
4038 for (size_t a = 0; a < PIVOT_N_AXES; a++)
4040 for (size_t k = 0; k < s->nests[a]->n; k++)
4041 if (k != s->nests[a]->scale_idx)
4043 const struct variable *var = s->nests[a]->vars[k];
4044 const struct ctables_categories *cats = s->table->categories[
4045 var_get_dict_index (var)];
4046 if (cats->show_empty)
4049 ctables_add_category_occurrences (var, &s->occurrences[a][k], cats);
4055 const struct ctables_category *cats[PIVOT_N_AXES][10]; /* XXX */
4056 struct ccase *c = case_create (dict_get_proto (s->table->ctables->dict));
4057 ctables_section_recurse_add_empty_categories (s, cats, c, 0, 0);
4062 ctables_execute (struct dataset *ds, struct ctables *ct)
4064 for (size_t i = 0; i < ct->n_tables; i++)
4066 struct ctables_table *t = ct->tables[i];
4067 t->sections = xnmalloc (MAX (1, t->stacks[PIVOT_AXIS_ROW].n) *
4068 MAX (1, t->stacks[PIVOT_AXIS_COLUMN].n) *
4069 MAX (1, t->stacks[PIVOT_AXIS_LAYER].n),
4070 sizeof *t->sections);
4071 size_t ix[PIVOT_N_AXES];
4072 ctables_table_add_section (t, 0, ix);
4075 struct casereader *input = proc_open (ds);
4076 bool warn_on_invalid = true;
4077 for (struct ccase *c = casereader_read (input); c;
4078 case_unref (c), c = casereader_read (input))
4080 double d_weight = dict_get_case_weight (dataset_dict (ds), c,
4082 double e_weight = (ct->e_weight
4083 ? var_force_valid_weight (ct->e_weight,
4084 case_num (c, ct->e_weight),
4088 for (size_t i = 0; i < ct->n_tables; i++)
4090 struct ctables_table *t = ct->tables[i];
4092 for (size_t j = 0; j < t->n_sections; j++)
4093 ctables_cell_insert (&t->sections[j], c, d_weight, e_weight);
4095 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
4096 if (t->label_axis[a] != a)
4097 ctables_insert_clabels_values (t, c, a);
4100 casereader_destroy (input);
4102 for (size_t i = 0; i < ct->n_tables; i++)
4104 struct ctables_table *t = ct->tables[i];
4106 if (t->clabels_example)
4107 ctables_sort_clabels_values (t);
4109 for (size_t j = 0; j < t->n_sections; j++)
4110 ctables_section_add_empty_categories (&t->sections[j]);
4112 ctables_table_output (ct, ct->tables[i]);
4114 return proc_commit (ds);
4119 typedef struct ctables_pcexpr *parse_recursively_func (struct lexer *);
4122 ctables_pcexpr_destroy (struct ctables_pcexpr *e)
4128 case CTPO_CAT_STRING:
4138 for (size_t i = 0; i < 2; i++)
4139 ctables_pcexpr_destroy (e->subs[i]);
4143 case CTPO_CAT_NUMBER:
4144 case CTPO_CAT_RANGE:
4145 case CTPO_CAT_MISSING:
4146 case CTPO_CAT_OTHERNM:
4147 case CTPO_CAT_SUBTOTAL:
4148 case CTPO_CAT_TOTAL:
4152 msg_location_destroy (e->location);
4157 static struct ctables_pcexpr *
4158 ctables_pcexpr_allocate_binary (enum ctables_postcompute_op op,
4159 struct ctables_pcexpr *sub0,
4160 struct ctables_pcexpr *sub1)
4162 struct ctables_pcexpr *e = xmalloc (sizeof *e);
4163 *e = (struct ctables_pcexpr) {
4165 .subs = { sub0, sub1 },
4166 .location = msg_location_merged (sub0->location, sub1->location),
4171 /* How to parse an operator. */
4174 enum token_type token;
4175 enum ctables_postcompute_op op;
4178 static const struct operator *
4179 match_operator (struct lexer *lexer, const struct operator ops[], size_t n_ops)
4181 for (const struct operator *op = ops; op < ops + n_ops; op++)
4182 if (lex_token (lexer) == op->token)
4184 if (op->token != T_NEG_NUM)
4193 static struct ctables_pcexpr *
4194 parse_binary_operators__ (struct lexer *lexer,
4195 const struct operator ops[], size_t n_ops,
4196 parse_recursively_func *parse_next_level,
4197 const char *chain_warning,
4198 struct ctables_pcexpr *lhs)
4200 for (int op_count = 0; ; op_count++)
4202 const struct operator *op = match_operator (lexer, ops, n_ops);
4205 if (op_count > 1 && chain_warning)
4206 msg_at (SW, lhs->location, "%s", chain_warning);
4211 struct ctables_pcexpr *rhs = parse_next_level (lexer);
4214 ctables_pcexpr_destroy (lhs);
4218 lhs = ctables_pcexpr_allocate_binary (op->op, lhs, rhs);
4222 static struct ctables_pcexpr *
4223 parse_binary_operators (struct lexer *lexer,
4224 const struct operator ops[], size_t n_ops,
4225 parse_recursively_func *parse_next_level,
4226 const char *chain_warning)
4228 struct ctables_pcexpr *lhs = parse_next_level (lexer);
4232 return parse_binary_operators__ (lexer, ops, n_ops, parse_next_level,
4233 chain_warning, lhs);
4236 static struct ctables_pcexpr *parse_add (struct lexer *);
4238 static struct ctables_pcexpr
4239 ctpo_cat_range (double low, double high)
4241 return (struct ctables_pcexpr) {
4242 .op = CTPO_CAT_RANGE,
4243 .range = { low, high },
4247 static struct ctables_pcexpr *
4248 parse_primary (struct lexer *lexer)
4250 int start_ofs = lex_ofs (lexer);
4251 struct ctables_pcexpr e;
4252 if (lex_is_number (lexer))
4254 e = (struct ctables_pcexpr) { .op = CTPO_CONSTANT,
4255 .number = lex_number (lexer) };
4258 else if (lex_match_id (lexer, "MISSING"))
4259 e = (struct ctables_pcexpr) { .op = CTPO_CAT_MISSING };
4260 else if (lex_match_id (lexer, "OTHERNM"))
4261 e = (struct ctables_pcexpr) { .op = CTPO_CAT_OTHERNM };
4262 else if (lex_match_id (lexer, "TOTAL"))
4263 e = (struct ctables_pcexpr) { .op = CTPO_CAT_TOTAL };
4264 else if (lex_match_id (lexer, "SUBTOTAL"))
4266 size_t subtotal_index = 0;
4267 if (lex_match (lexer, T_LBRACK))
4269 if (!lex_force_int_range (lexer, "SUBTOTAL", 1, LONG_MAX))
4271 subtotal_index = lex_integer (lexer);
4273 if (!lex_force_match (lexer, T_RBRACK))
4276 e = (struct ctables_pcexpr) { .op = CTPO_CAT_SUBTOTAL,
4277 .subtotal_index = subtotal_index };
4279 else if (lex_match (lexer, T_LBRACK))
4281 if (lex_match_id (lexer, "LO"))
4283 if (!lex_force_match_id (lexer, "THRU") || lex_force_num (lexer))
4285 e = ctpo_cat_range (-DBL_MAX, lex_number (lexer));
4288 else if (lex_is_number (lexer))
4290 double number = lex_number (lexer);
4292 if (lex_match_id (lexer, "THRU"))
4294 if (lex_match_id (lexer, "HI"))
4295 e = ctpo_cat_range (number, DBL_MAX);
4298 if (!lex_force_num (lexer))
4300 e = ctpo_cat_range (number, lex_number (lexer));
4305 e = (struct ctables_pcexpr) { .op = CTPO_CAT_NUMBER,
4308 else if (lex_is_string (lexer))
4310 e = (struct ctables_pcexpr) {
4311 .op = CTPO_CAT_STRING,
4312 .string = ss_xstrdup (lex_tokss (lexer)),
4318 lex_error (lexer, NULL);
4322 if (!lex_force_match (lexer, T_RBRACK))
4324 if (e.op == CTPO_CAT_STRING)
4329 else if (lex_match (lexer, T_LPAREN))
4331 struct ctables_pcexpr *ep = parse_add (lexer);
4334 if (!lex_force_match (lexer, T_RPAREN))
4336 ctables_pcexpr_destroy (ep);
4343 lex_error (lexer, NULL);
4347 e.location = lex_ofs_location (lexer, start_ofs, lex_ofs (lexer) - 1);
4348 return xmemdup (&e, sizeof e);
4351 static struct ctables_pcexpr *
4352 ctables_pcexpr_allocate_neg (struct ctables_pcexpr *sub,
4353 struct lexer *lexer, int start_ofs)
4355 struct ctables_pcexpr *e = xmalloc (sizeof *e);
4356 *e = (struct ctables_pcexpr) {
4359 .location = lex_ofs_location (lexer, start_ofs, lex_ofs (lexer) - 1),
4364 static struct ctables_pcexpr *
4365 parse_exp (struct lexer *lexer)
4367 static const struct operator op = { T_EXP, CTPO_POW };
4369 const char *chain_warning =
4370 _("The exponentiation operator (`**') is left-associative: "
4371 "`a**b**c' equals `(a**b)**c', not `a**(b**c)'. "
4372 "To disable this warning, insert parentheses.");
4374 if (lex_token (lexer) != T_NEG_NUM || lex_next_token (lexer, 1) != T_EXP)
4375 return parse_binary_operators (lexer, &op, 1,
4376 parse_primary, chain_warning);
4378 /* Special case for situations like "-5**6", which must be parsed as
4381 int start_ofs = lex_ofs (lexer);
4382 struct ctables_pcexpr *lhs = xmalloc (sizeof *lhs);
4383 *lhs = (struct ctables_pcexpr) {
4384 .op = CTPO_CONSTANT,
4385 .number = -lex_tokval (lexer),
4386 .location = lex_ofs_location (lexer, start_ofs, lex_ofs (lexer)),
4390 struct ctables_pcexpr *node = parse_binary_operators__ (
4391 lexer, &op, 1, parse_primary, chain_warning, lhs);
4395 return ctables_pcexpr_allocate_neg (node, lexer, start_ofs);
4398 /* Parses the unary minus level. */
4399 static struct ctables_pcexpr *
4400 parse_neg (struct lexer *lexer)
4402 int start_ofs = lex_ofs (lexer);
4403 if (!lex_match (lexer, T_DASH))
4404 return parse_exp (lexer);
4406 struct ctables_pcexpr *inner = parse_neg (lexer);
4410 return ctables_pcexpr_allocate_neg (inner, lexer, start_ofs);
4413 /* Parses the multiplication and division level. */
4414 static struct ctables_pcexpr *
4415 parse_mul (struct lexer *lexer)
4417 static const struct operator ops[] =
4419 { T_ASTERISK, CTPO_MUL },
4420 { T_SLASH, CTPO_DIV },
4423 return parse_binary_operators (lexer, ops, sizeof ops / sizeof *ops,
4427 /* Parses the addition and subtraction level. */
4428 static struct ctables_pcexpr *
4429 parse_add (struct lexer *lexer)
4431 static const struct operator ops[] =
4433 { T_PLUS, CTPO_ADD },
4434 { T_DASH, CTPO_SUB },
4435 { T_NEG_NUM, CTPO_ADD },
4438 return parse_binary_operators (lexer, ops, sizeof ops / sizeof *ops,
4442 static struct ctables_postcompute *
4443 ctables_find_postcompute (struct ctables *ct, const char *name)
4445 struct ctables_postcompute *pc;
4446 HMAP_FOR_EACH_WITH_HASH (pc, struct ctables_postcompute, hmap_node,
4447 utf8_hash_case_string (name, 0), &ct->postcomputes)
4448 if (!utf8_strcasecmp (pc->name, name))
4454 ctables_parse_pcompute (struct lexer *lexer, struct ctables *ct)
4456 int pcompute_start = lex_ofs (lexer) - 1;
4458 if (!lex_force_match (lexer, T_AND) || !lex_force_id (lexer))
4461 char *name = ss_xstrdup (lex_tokss (lexer));
4464 if (!lex_force_match (lexer, T_EQUALS)
4465 || !lex_force_match_id (lexer, "EXPR")
4466 || !lex_force_match (lexer, T_LPAREN))
4472 int expr_start = lex_ofs (lexer);
4473 struct ctables_pcexpr *expr = parse_add (lexer);
4474 int expr_end = lex_ofs (lexer) - 1;
4475 if (!expr || !lex_force_match (lexer, T_RPAREN))
4480 int pcompute_end = lex_ofs (lexer) - 1;
4482 struct msg_location *location = lex_ofs_location (lexer, pcompute_start,
4485 struct ctables_postcompute *pc = ctables_find_postcompute (ct, name);
4488 msg_at (SW, location, _("New definition of &%s will override the "
4489 "previous definition."),
4491 msg_at (SN, pc->location, _("This is the previous definition."));
4493 ctables_pcexpr_destroy (pc->expr);
4494 msg_location_destroy (pc->location);
4499 pc = xmalloc (sizeof *pc);
4500 *pc = (struct ctables_postcompute) { .name = name };
4501 hmap_insert (&ct->postcomputes, &pc->hmap_node,
4502 utf8_hash_case_string (pc->name, 0));
4505 pc->location = location;
4507 pc->label = lex_ofs_representation (lexer, expr_start, expr_end);
4512 ctables_parse_pproperties_format (struct lexer *lexer,
4513 struct ctables_summary_spec_set *sss)
4515 *sss = (struct ctables_summary_spec_set) { .n = 0 };
4517 while (lex_token (lexer) != T_ENDCMD && lex_token (lexer) != T_SLASH
4518 && !(lex_token (lexer) == T_ID
4519 && (lex_id_match (ss_cstr ("LABEL"), lex_tokss (lexer))
4520 || lex_id_match (ss_cstr ("HIDESOURCECATS"),
4521 lex_tokss (lexer)))))
4523 /* Parse function. */
4524 enum ctables_summary_function function;
4525 if (!parse_ctables_summary_function (lexer, &function))
4528 /* Parse percentile. */
4529 double percentile = 0;
4530 if (function == CTSF_PTILE)
4532 if (!lex_force_num_range_closed (lexer, "PTILE", 0, 100))
4534 percentile = lex_number (lexer);
4539 struct fmt_spec format;
4540 if (!parse_format_specifier (lexer, &format)
4541 || !fmt_check_output (&format)
4542 || !fmt_check_type_compat (&format, VAL_NUMERIC))
4545 if (sss->n >= sss->allocated)
4546 sss->specs = x2nrealloc (sss->specs, &sss->allocated,
4547 sizeof *sss->specs);
4548 sss->specs[sss->n++] = (struct ctables_summary_spec) {
4549 .function = function,
4550 .percentile = percentile,
4557 ctables_summary_spec_set_uninit (sss);
4562 ctables_parse_pproperties (struct lexer *lexer, struct ctables *ct)
4564 struct ctables_postcompute **pcs = NULL;
4566 size_t allocated_pcs = 0;
4568 while (lex_match (lexer, T_AND))
4570 if (!lex_force_id (lexer))
4572 struct ctables_postcompute *pc
4573 = ctables_find_postcompute (ct, lex_tokcstr (lexer));
4576 msg (SE, _("Unknown computed category &%s."), lex_tokcstr (lexer));
4581 if (n_pcs >= allocated_pcs)
4582 pcs = x2nrealloc (pcs, &allocated_pcs, sizeof *pcs);
4586 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
4588 if (lex_match_id (lexer, "LABEL"))
4590 lex_match (lexer, T_EQUALS);
4591 if (!lex_force_string (lexer))
4594 for (size_t i = 0; i < n_pcs; i++)
4596 free (pcs[i]->label);
4597 pcs[i]->label = ss_xstrdup (lex_tokss (lexer));
4602 else if (lex_match_id (lexer, "FORMAT"))
4604 lex_match (lexer, T_EQUALS);
4606 struct ctables_summary_spec_set sss;
4607 if (!ctables_parse_pproperties_format (lexer, &sss))
4610 for (size_t i = 0; i < n_pcs; i++)
4613 ctables_summary_spec_set_uninit (pcs[i]->specs);
4615 pcs[i]->specs = xmalloc (sizeof *pcs[i]->specs);
4616 ctables_summary_spec_set_clone (pcs[i]->specs, &sss);
4618 ctables_summary_spec_set_uninit (&sss);
4620 else if (lex_match_id (lexer, "HIDESOURCECATS"))
4622 lex_match (lexer, T_EQUALS);
4623 bool hide_source_cats;
4624 if (!parse_bool (lexer, &hide_source_cats))
4626 for (size_t i = 0; i < n_pcs; i++)
4627 pcs[i]->hide_source_cats = hide_source_cats;
4631 lex_error_expecting (lexer, "LABEL", "FORMAT", "HIDESOURCECATS");
4644 cmd_ctables (struct lexer *lexer, struct dataset *ds)
4646 size_t n_vars = dict_get_n_vars (dataset_dict (ds));
4647 enum ctables_vlabel *vlabels = xnmalloc (n_vars, sizeof *vlabels);
4648 enum settings_value_show tvars = settings_get_show_variables ();
4649 for (size_t i = 0; i < n_vars; i++)
4650 vlabels[i] = (enum ctables_vlabel) tvars;
4652 struct pivot_table_look *look = pivot_table_look_unshare (
4653 pivot_table_look_ref (pivot_table_look_get_default ()));
4654 look->omit_empty = false;
4656 struct ctables *ct = xmalloc (sizeof *ct);
4657 *ct = (struct ctables) {
4658 .dict = dataset_dict (ds),
4660 .ctables_formats = FMT_SETTINGS_INIT,
4662 .postcomputes = HMAP_INITIALIZER (ct->postcomputes),
4668 const char *dot_string;
4669 const char *comma_string;
4671 static const struct ctf ctfs[4] = {
4672 { CTEF_NEGPAREN, "(,,,)", "(...)" },
4673 { CTEF_NEQUAL, "-,N=,,", "-.N=.." },
4674 { CTEF_PAREN, "-,(,),", "-.(.)." },
4675 { CTEF_PCTPAREN, "-,(,%),", "-.(.%)." },
4677 bool is_dot = settings_get_fmt_settings ()->decimal == '.';
4678 for (size_t i = 0; i < 4; i++)
4680 const char *s = is_dot ? ctfs[i].dot_string : ctfs[i].comma_string;
4681 fmt_settings_set_cc (&ct->ctables_formats, ctfs[i].type,
4682 fmt_number_style_from_string (s));
4685 if (!lex_force_match (lexer, T_SLASH))
4688 while (!lex_match_id (lexer, "TABLE"))
4690 if (lex_match_id (lexer, "FORMAT"))
4692 double widths[2] = { SYSMIS, SYSMIS };
4693 double units_per_inch = 72.0;
4695 while (lex_token (lexer) != T_SLASH)
4697 if (lex_match_id (lexer, "MINCOLWIDTH"))
4699 if (!parse_col_width (lexer, "MINCOLWIDTH", &widths[0]))
4702 else if (lex_match_id (lexer, "MAXCOLWIDTH"))
4704 if (!parse_col_width (lexer, "MAXCOLWIDTH", &widths[1]))
4707 else if (lex_match_id (lexer, "UNITS"))
4709 lex_match (lexer, T_EQUALS);
4710 if (lex_match_id (lexer, "POINTS"))
4711 units_per_inch = 72.0;
4712 else if (lex_match_id (lexer, "INCHES"))
4713 units_per_inch = 1.0;
4714 else if (lex_match_id (lexer, "CM"))
4715 units_per_inch = 2.54;
4718 lex_error_expecting (lexer, "POINTS", "INCHES", "CM");
4722 else if (lex_match_id (lexer, "EMPTY"))
4727 lex_match (lexer, T_EQUALS);
4728 if (lex_match_id (lexer, "ZERO"))
4730 /* Nothing to do. */
4732 else if (lex_match_id (lexer, "BLANK"))
4733 ct->zero = xstrdup ("");
4734 else if (lex_force_string (lexer))
4736 ct->zero = ss_xstrdup (lex_tokss (lexer));
4742 else if (lex_match_id (lexer, "MISSING"))
4744 lex_match (lexer, T_EQUALS);
4745 if (!lex_force_string (lexer))
4749 ct->missing = (strcmp (lex_tokcstr (lexer), ".")
4750 ? ss_xstrdup (lex_tokss (lexer))
4756 lex_error_expecting (lexer, "MINCOLWIDTH", "MAXCOLWIDTH",
4757 "UNITS", "EMPTY", "MISSING");
4762 if (widths[0] != SYSMIS && widths[1] != SYSMIS
4763 && widths[0] > widths[1])
4765 msg (SE, _("MINCOLWIDTH must not be greater than MAXCOLWIDTH."));
4769 for (size_t i = 0; i < 2; i++)
4770 if (widths[i] != SYSMIS)
4772 int *wr = ct->look->width_ranges[TABLE_HORZ];
4773 wr[i] = widths[i] / units_per_inch * 96.0;
4778 else if (lex_match_id (lexer, "VLABELS"))
4780 if (!lex_force_match_id (lexer, "VARIABLES"))
4782 lex_match (lexer, T_EQUALS);
4784 struct variable **vars;
4786 if (!parse_variables (lexer, dataset_dict (ds), &vars, &n_vars,
4790 if (!lex_force_match_id (lexer, "DISPLAY"))
4795 lex_match (lexer, T_EQUALS);
4797 enum ctables_vlabel vlabel;
4798 if (lex_match_id (lexer, "DEFAULT"))
4799 vlabel = (enum ctables_vlabel) settings_get_show_variables ();
4800 else if (lex_match_id (lexer, "NAME"))
4802 else if (lex_match_id (lexer, "LABEL"))
4803 vlabel = CTVL_LABEL;
4804 else if (lex_match_id (lexer, "BOTH"))
4806 else if (lex_match_id (lexer, "NONE"))
4810 lex_error_expecting (lexer, "DEFAULT", "NAME", "LABEL",
4816 for (size_t i = 0; i < n_vars; i++)
4817 ct->vlabels[var_get_dict_index (vars[i])] = vlabel;
4820 else if (lex_match_id (lexer, "MRSETS"))
4822 if (!lex_force_match_id (lexer, "COUNTDUPLICATES"))
4824 lex_match (lexer, T_EQUALS);
4825 if (!parse_bool (lexer, &ct->mrsets_count_duplicates))
4828 else if (lex_match_id (lexer, "SMISSING"))
4830 if (lex_match_id (lexer, "VARIABLE"))
4831 ct->smissing_listwise = false;
4832 else if (lex_match_id (lexer, "LISTWISE"))
4833 ct->smissing_listwise = true;
4836 lex_error_expecting (lexer, "VARIABLE", "LISTWISE");
4840 else if (lex_match_id (lexer, "PCOMPUTE"))
4842 if (!ctables_parse_pcompute (lexer, ct))
4845 else if (lex_match_id (lexer, "PPROPERTIES"))
4847 if (!ctables_parse_pproperties (lexer, ct))
4850 else if (lex_match_id (lexer, "WEIGHT"))
4852 if (!lex_force_match_id (lexer, "VARIABLE"))
4854 lex_match (lexer, T_EQUALS);
4855 ct->e_weight = parse_variable (lexer, dataset_dict (ds));
4859 else if (lex_match_id (lexer, " HIDESMALLCOUNTS"))
4861 if (lex_match_id (lexer, "COUNT"))
4863 lex_match (lexer, T_EQUALS);
4864 if (!lex_force_int_range (lexer, "HIDESMALLCOUNTS COUNT",
4867 ct->hide_threshold = lex_integer (lexer);
4870 else if (ct->hide_threshold == 0)
4871 ct->hide_threshold = 5;
4875 lex_error_expecting (lexer, "FORMAT", "VLABELS", "MRSETS",
4876 "SMISSING", "PCOMPUTE", "PPROPERTIES",
4877 "WEIGHT", "HIDESMALLCOUNTS", "TABLE");
4881 if (!lex_force_match (lexer, T_SLASH))
4885 size_t allocated_tables = 0;
4888 if (ct->n_tables >= allocated_tables)
4889 ct->tables = x2nrealloc (ct->tables, &allocated_tables,
4890 sizeof *ct->tables);
4892 struct ctables_category *cat = xmalloc (sizeof *cat);
4893 *cat = (struct ctables_category) {
4895 .include_missing = false,
4896 .sort_ascending = true,
4899 struct ctables_categories *c = xmalloc (sizeof *c);
4900 size_t n_vars = dict_get_n_vars (dataset_dict (ds));
4901 *c = (struct ctables_categories) {
4908 struct ctables_categories **categories = xnmalloc (n_vars,
4909 sizeof *categories);
4910 for (size_t i = 0; i < n_vars; i++)
4913 struct ctables_table *t = xmalloc (sizeof *t);
4914 *t = (struct ctables_table) {
4916 .slabels_axis = PIVOT_AXIS_COLUMN,
4917 .slabels_visible = true,
4918 .clabels_values_map = HMAP_INITIALIZER (t->clabels_values_map),
4920 [PIVOT_AXIS_ROW] = PIVOT_AXIS_ROW,
4921 [PIVOT_AXIS_COLUMN] = PIVOT_AXIS_COLUMN,
4922 [PIVOT_AXIS_LAYER] = PIVOT_AXIS_LAYER,
4924 .clabels_from_axis = PIVOT_AXIS_LAYER,
4925 .categories = categories,
4926 .n_categories = n_vars,
4929 ct->tables[ct->n_tables++] = t;
4931 lex_match (lexer, T_EQUALS);
4932 if (!ctables_axis_parse (lexer, dataset_dict (ds), ct, t, PIVOT_AXIS_ROW))
4934 if (lex_match (lexer, T_BY))
4936 if (!ctables_axis_parse (lexer, dataset_dict (ds),
4937 ct, t, PIVOT_AXIS_COLUMN))
4940 if (lex_match (lexer, T_BY))
4942 if (!ctables_axis_parse (lexer, dataset_dict (ds),
4943 ct, t, PIVOT_AXIS_LAYER))
4948 if (!t->axes[PIVOT_AXIS_ROW] && !t->axes[PIVOT_AXIS_COLUMN]
4949 && !t->axes[PIVOT_AXIS_LAYER])
4951 lex_error (lexer, _("At least one variable must be specified."));
4955 const struct ctables_axis *scales[PIVOT_N_AXES];
4956 size_t n_scales = 0;
4957 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
4959 scales[a] = find_scale (t->axes[a]);
4965 msg (SE, _("Scale variables may appear only on one axis."));
4966 if (scales[PIVOT_AXIS_ROW])
4967 msg_at (SN, scales[PIVOT_AXIS_ROW]->loc,
4968 _("This scale variable appears on the rows axis."));
4969 if (scales[PIVOT_AXIS_COLUMN])
4970 msg_at (SN, scales[PIVOT_AXIS_COLUMN]->loc,
4971 _("This scale variable appears on the columns axis."));
4972 if (scales[PIVOT_AXIS_LAYER])
4973 msg_at (SN, scales[PIVOT_AXIS_LAYER]->loc,
4974 _("This scale variable appears on the layer axis."));
4978 const struct ctables_axis *summaries[PIVOT_N_AXES];
4979 size_t n_summaries = 0;
4980 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
4982 summaries[a] = (scales[a]
4984 : find_categorical_summary_spec (t->axes[a]));
4988 if (n_summaries > 1)
4990 msg (SE, _("Summaries may appear only on one axis."));
4991 if (summaries[PIVOT_AXIS_ROW])
4992 msg_at (SN, summaries[PIVOT_AXIS_ROW]->loc,
4993 _("This variable on the rows axis has a summary."));
4994 if (summaries[PIVOT_AXIS_COLUMN])
4995 msg_at (SN, summaries[PIVOT_AXIS_COLUMN]->loc,
4996 _("This variable on the columns axis has a summary."));
4997 if (summaries[PIVOT_AXIS_LAYER])
4998 msg_at (SN, summaries[PIVOT_AXIS_LAYER]->loc,
4999 _("This variable on the layers axis has a summary."));
5002 for (enum pivot_axis_type a = 0; a < PIVOT_N_AXES; a++)
5003 if (n_summaries ? summaries[a] : t->axes[a])
5005 t->summary_axis = a;
5009 if (lex_token (lexer) == T_ENDCMD)
5011 if (!ctables_prepare_table (t))
5015 if (!lex_force_match (lexer, T_SLASH))
5018 while (!lex_match_id (lexer, "TABLE") && lex_token (lexer) != T_ENDCMD)
5020 if (lex_match_id (lexer, "SLABELS"))
5022 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
5024 if (lex_match_id (lexer, "POSITION"))
5026 lex_match (lexer, T_EQUALS);
5027 if (lex_match_id (lexer, "COLUMN"))
5028 t->slabels_axis = PIVOT_AXIS_COLUMN;
5029 else if (lex_match_id (lexer, "ROW"))
5030 t->slabels_axis = PIVOT_AXIS_ROW;
5031 else if (lex_match_id (lexer, "LAYER"))
5032 t->slabels_axis = PIVOT_AXIS_LAYER;
5035 lex_error_expecting (lexer, "COLUMN", "ROW", "LAYER");
5039 else if (lex_match_id (lexer, "VISIBLE"))
5041 lex_match (lexer, T_EQUALS);
5042 if (!parse_bool (lexer, &t->slabels_visible))
5047 lex_error_expecting (lexer, "POSITION", "VISIBLE");
5052 else if (lex_match_id (lexer, "CLABELS"))
5054 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
5056 if (lex_match_id (lexer, "AUTO"))
5058 t->label_axis[PIVOT_AXIS_ROW] = PIVOT_AXIS_ROW;
5059 t->label_axis[PIVOT_AXIS_COLUMN] = PIVOT_AXIS_COLUMN;
5061 else if (lex_match_id (lexer, "ROWLABELS"))
5063 lex_match (lexer, T_EQUALS);
5064 if (lex_match_id (lexer, "OPPOSITE"))
5065 t->label_axis[PIVOT_AXIS_ROW] = PIVOT_AXIS_COLUMN;
5066 else if (lex_match_id (lexer, "LAYER"))
5067 t->label_axis[PIVOT_AXIS_ROW] = PIVOT_AXIS_LAYER;
5070 lex_error_expecting (lexer, "OPPOSITE", "LAYER");
5074 else if (lex_match_id (lexer, "COLLABELS"))
5076 lex_match (lexer, T_EQUALS);
5077 if (lex_match_id (lexer, "OPPOSITE"))
5078 t->label_axis[PIVOT_AXIS_COLUMN] = PIVOT_AXIS_ROW;
5079 else if (lex_match_id (lexer, "LAYER"))
5080 t->label_axis[PIVOT_AXIS_COLUMN] = PIVOT_AXIS_LAYER;
5083 lex_error_expecting (lexer, "OPPOSITE", "LAYER");
5089 lex_error_expecting (lexer, "AUTO", "ROWLABELS",
5095 else if (lex_match_id (lexer, "CRITERIA"))
5097 if (!lex_force_match_id (lexer, "CILEVEL"))
5099 lex_match (lexer, T_EQUALS);
5101 if (!lex_force_num_range_halfopen (lexer, "CILEVEL", 0, 100))
5103 t->cilevel = lex_number (lexer);
5106 else if (lex_match_id (lexer, "CATEGORIES"))
5108 if (!ctables_table_parse_categories (lexer, dataset_dict (ds),
5112 else if (lex_match_id (lexer, "TITLES"))
5117 if (lex_match_id (lexer, "CAPTION"))
5118 textp = &t->caption;
5119 else if (lex_match_id (lexer, "CORNER"))
5121 else if (lex_match_id (lexer, "TITLE"))
5125 lex_error_expecting (lexer, "CAPTION", "CORNER", "TITLE");
5128 lex_match (lexer, T_EQUALS);
5130 struct string s = DS_EMPTY_INITIALIZER;
5131 while (lex_is_string (lexer))
5133 if (!ds_is_empty (&s))
5134 ds_put_byte (&s, ' ');
5135 ds_put_substring (&s, lex_tokss (lexer));
5139 *textp = ds_steal_cstr (&s);
5141 while (lex_token (lexer) != T_SLASH
5142 && lex_token (lexer) != T_ENDCMD);
5144 else if (lex_match_id (lexer, "SIGTEST"))
5148 t->chisq = xmalloc (sizeof *t->chisq);
5149 *t->chisq = (struct ctables_chisq) {
5151 .include_mrsets = true,
5152 .all_visible = true,
5158 if (lex_match_id (lexer, "TYPE"))
5160 lex_match (lexer, T_EQUALS);
5161 if (!lex_force_match_id (lexer, "CHISQUARE"))
5164 else if (lex_match_id (lexer, "ALPHA"))
5166 lex_match (lexer, T_EQUALS);
5167 if (!lex_force_num_range_halfopen (lexer, "ALPHA", 0, 1))
5169 t->chisq->alpha = lex_number (lexer);
5172 else if (lex_match_id (lexer, "INCLUDEMRSETS"))
5174 lex_match (lexer, T_EQUALS);
5175 if (parse_bool (lexer, &t->chisq->include_mrsets))
5178 else if (lex_match_id (lexer, "CATEGORIES"))
5180 lex_match (lexer, T_EQUALS);
5181 if (lex_match_id (lexer, "ALLVISIBLE"))
5182 t->chisq->all_visible = true;
5183 else if (lex_match_id (lexer, "SUBTOTALS"))
5184 t->chisq->all_visible = false;
5187 lex_error_expecting (lexer,
5188 "ALLVISIBLE", "SUBTOTALS");
5194 lex_error_expecting (lexer, "TYPE", "ALPHA",
5195 "INCLUDEMRSETS", "CATEGORIES");
5199 while (lex_token (lexer) != T_SLASH
5200 && lex_token (lexer) != T_ENDCMD);
5202 else if (lex_match_id (lexer, "COMPARETEST"))
5206 t->pairwise = xmalloc (sizeof *t->pairwise);
5207 *t->pairwise = (struct ctables_pairwise) {
5209 .alpha = { .05, .05 },
5210 .adjust = BONFERRONI,
5211 .include_mrsets = true,
5212 .meansvariance_allcats = true,
5213 .all_visible = true,
5222 if (lex_match_id (lexer, "TYPE"))
5224 lex_match (lexer, T_EQUALS);
5225 if (lex_match_id (lexer, "PROP"))
5226 t->pairwise->type = PROP;
5227 else if (lex_match_id (lexer, "MEAN"))
5228 t->pairwise->type = MEAN;
5231 lex_error_expecting (lexer, "PROP", "MEAN");
5235 else if (lex_match_id (lexer, "ALPHA"))
5237 lex_match (lexer, T_EQUALS);
5239 if (!lex_force_num_range_open (lexer, "ALPHA", 0, 1))
5241 double a0 = lex_number (lexer);
5244 lex_match (lexer, T_COMMA);
5245 if (lex_is_number (lexer))
5247 if (!lex_force_num_range_open (lexer, "ALPHA", 0, 1))
5249 double a1 = lex_number (lexer);
5252 t->pairwise->alpha[0] = MIN (a0, a1);
5253 t->pairwise->alpha[1] = MAX (a0, a1);
5256 t->pairwise->alpha[0] = t->pairwise->alpha[1] = a0;
5258 else if (lex_match_id (lexer, "ADJUST"))
5260 lex_match (lexer, T_EQUALS);
5261 if (lex_match_id (lexer, "BONFERRONI"))
5262 t->pairwise->adjust = BONFERRONI;
5263 else if (lex_match_id (lexer, "BH"))
5264 t->pairwise->adjust = BH;
5265 else if (lex_match_id (lexer, "NONE"))
5266 t->pairwise->adjust = 0;
5269 lex_error_expecting (lexer, "BONFERRONI", "BH",
5274 else if (lex_match_id (lexer, "INCLUDEMRSETS"))
5276 lex_match (lexer, T_EQUALS);
5277 if (!parse_bool (lexer, &t->pairwise->include_mrsets))
5280 else if (lex_match_id (lexer, "MEANSVARIANCE"))
5282 lex_match (lexer, T_EQUALS);
5283 if (lex_match_id (lexer, "ALLCATS"))
5284 t->pairwise->meansvariance_allcats = true;
5285 else if (lex_match_id (lexer, "TESTEDCATS"))
5286 t->pairwise->meansvariance_allcats = false;
5289 lex_error_expecting (lexer, "ALLCATS", "TESTEDCATS");
5293 else if (lex_match_id (lexer, "CATEGORIES"))
5295 lex_match (lexer, T_EQUALS);
5296 if (lex_match_id (lexer, "ALLVISIBLE"))
5297 t->pairwise->all_visible = true;
5298 else if (lex_match_id (lexer, "SUBTOTALS"))
5299 t->pairwise->all_visible = false;
5302 lex_error_expecting (lexer, "ALLVISIBLE",
5307 else if (lex_match_id (lexer, "MERGE"))
5309 lex_match (lexer, T_EQUALS);
5310 if (!parse_bool (lexer, &t->pairwise->merge))
5313 else if (lex_match_id (lexer, "STYLE"))
5315 lex_match (lexer, T_EQUALS);
5316 if (lex_match_id (lexer, "APA"))
5317 t->pairwise->apa_style = true;
5318 else if (lex_match_id (lexer, "SIMPLE"))
5319 t->pairwise->apa_style = false;
5322 lex_error_expecting (lexer, "APA", "SIMPLE");
5326 else if (lex_match_id (lexer, "SHOWSIG"))
5328 lex_match (lexer, T_EQUALS);
5329 if (!parse_bool (lexer, &t->pairwise->show_sig))
5334 lex_error_expecting (lexer, "TYPE", "ALPHA", "ADJUST",
5335 "INCLUDEMRSETS", "MEANSVARIANCE",
5336 "CATEGORIES", "MERGE", "STYLE",
5341 while (lex_token (lexer) != T_SLASH
5342 && lex_token (lexer) != T_ENDCMD);
5346 lex_error_expecting (lexer, "TABLE", "SLABELS", "CLABELS",
5347 "CRITERIA", "CATEGORIES", "TITLES",
5348 "SIGTEST", "COMPARETEST");
5352 if (!lex_match (lexer, T_SLASH))
5356 if (t->label_axis[PIVOT_AXIS_ROW] != PIVOT_AXIS_ROW
5357 && t->label_axis[PIVOT_AXIS_COLUMN] != PIVOT_AXIS_COLUMN)
5359 msg (SE, _("ROWLABELS and COLLABELS may not both be specified."));
5363 if (!ctables_prepare_table (t))
5366 while (lex_token (lexer) != T_ENDCMD);
5368 bool ok = ctables_execute (ds, ct);
5369 ctables_destroy (ct);
5370 return ok ? CMD_SUCCESS : CMD_FAILURE;
5373 ctables_destroy (ct);