Fix misleadingly indented statements.
[pspp.git] / src / language / stats / npar.c
blobe9e2e61b2f85613f7c401900cc782e31016d5054
1 /* PSPP - a program for statistical analysis. -*-c-*-
2 Copyright (C) 2006, 2008, 2009, 2010, 2011, 2016 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/>. */
17 #include <config.h>
19 #include "language/stats/npar.h"
21 #include <stdlib.h>
22 #include <math.h>
24 #include "data/case.h"
25 #include "data/casegrouper.h"
26 #include "data/casereader.h"
27 #include "data/dataset.h"
28 #include "data/dictionary.h"
29 #include "data/settings.h"
30 #include "data/variable.h"
31 #include "language/command.h"
32 #include "language/lexer/lexer.h"
33 #include "language/lexer/value-parser.h"
34 #include "language/lexer/variable-parser.h"
35 #include "language/stats/binomial.h"
36 #include "language/stats/chisquare.h"
37 #include "language/stats/ks-one-sample.h"
38 #include "language/stats/cochran.h"
39 #include "language/stats/friedman.h"
40 #include "language/stats/jonckheere-terpstra.h"
41 #include "language/stats/kruskal-wallis.h"
42 #include "language/stats/mann-whitney.h"
43 #include "language/stats/mcnemar.h"
44 #include "language/stats/median.h"
45 #include "language/stats/npar-summary.h"
46 #include "language/stats/runs.h"
47 #include "language/stats/sign.h"
48 #include "language/stats/wilcoxon.h"
49 #include "libpspp/array.h"
50 #include "libpspp/assertion.h"
51 #include "libpspp/cast.h"
52 #include "libpspp/hash-functions.h"
53 #include "libpspp/hmapx.h"
54 #include "libpspp/message.h"
55 #include "libpspp/pool.h"
56 #include "libpspp/str.h"
57 #include "libpspp/taint.h"
58 #include "math/moments.h"
60 #include "gl/xalloc.h"
62 #include "gettext.h"
63 #define _(msgid) gettext (msgid)
65 /* Settings for subcommand specifiers. */
66 enum missing_type
68 MISS_ANALYSIS,
69 MISS_LISTWISE,
72 /* Array indices for STATISTICS subcommand. */
73 enum
75 NPAR_ST_DESCRIPTIVES = 0,
76 NPAR_ST_QUARTILES = 1,
77 NPAR_ST_ALL = 2,
78 NPAR_ST_count
81 /* NPAR TESTS structure. */
82 struct cmd_npar_tests
84 /* Count variables indicating how many
85 of the subcommands have been given. */
86 int chisquare;
87 int cochran;
88 int binomial;
89 int ks_one_sample;
90 int wilcoxon;
91 int sign;
92 int runs;
93 int friedman;
94 int kendall;
95 int kruskal_wallis;
96 int mann_whitney;
97 int mcnemar;
98 int median;
99 int jonckheere_terpstra;
100 int missing;
101 int method;
102 int statistics;
104 /* How missing values should be treated */
105 long miss;
107 /* Which statistics have been requested */
108 int a_statistics[NPAR_ST_count];
112 struct npar_specs
114 struct pool *pool;
115 struct npar_test **test;
116 size_t n_tests;
118 const struct variable **vv; /* Compendium of all variables
119 (those mentioned on ANY subcommand */
120 int n_vars; /* Number of variables in vv */
122 enum mv_class filter; /* Missing values to filter. */
124 bool descriptives; /* Descriptive statistics should be calculated */
125 bool quartiles; /* Quartiles should be calculated */
127 bool exact; /* Whether exact calculations have been requested */
128 double timer; /* Maximum time (in minutes) to wait for exact calculations */
132 /* Prototype for custom subcommands of NPAR TESTS. */
133 static int npar_chisquare (struct lexer *, struct dataset *, struct npar_specs *);
134 static int npar_binomial (struct lexer *, struct dataset *, struct npar_specs *);
135 static int npar_ks_one_sample (struct lexer *, struct dataset *, struct npar_specs *);
136 static int npar_runs (struct lexer *, struct dataset *, struct npar_specs *);
137 static int npar_friedman (struct lexer *, struct dataset *, struct npar_specs *);
138 static int npar_kendall (struct lexer *, struct dataset *, struct npar_specs *);
139 static int npar_cochran (struct lexer *, struct dataset *, struct npar_specs *);
140 static int npar_wilcoxon (struct lexer *, struct dataset *, struct npar_specs *);
141 static int npar_sign (struct lexer *, struct dataset *, struct npar_specs *);
142 static int npar_kruskal_wallis (struct lexer *, struct dataset *, struct npar_specs *);
143 static int npar_jonckheere_terpstra (struct lexer *, struct dataset *, struct npar_specs *);
144 static int npar_mann_whitney (struct lexer *, struct dataset *, struct npar_specs *);
145 static int npar_mcnemar (struct lexer *, struct dataset *, struct npar_specs *);
146 static int npar_median (struct lexer *, struct dataset *, struct npar_specs *);
148 static int npar_method (struct lexer *, struct npar_specs *);
150 /* Command parsing functions. */
151 static int parse_npar_tests (struct lexer *lexer, struct dataset *ds, struct cmd_npar_tests *p,
152 struct npar_specs *npar_specs );
154 static int
155 parse_npar_tests (struct lexer *lexer, struct dataset *ds, struct cmd_npar_tests *npt,
156 struct npar_specs *nps)
158 npt->chisquare = 0;
159 npt->cochran = 0;
160 npt->binomial = 0;
161 npt->ks_one_sample = 0;
162 npt->wilcoxon = 0;
163 npt->sign = 0;
164 npt->runs = 0;
165 npt->friedman = 0;
166 npt->kendall = 0;
167 npt->kruskal_wallis = 0;
168 npt->mann_whitney = 0;
169 npt->mcnemar = 0;
170 npt->median = 0;
171 npt->jonckheere_terpstra = 0;
173 npt->miss = MISS_ANALYSIS;
174 npt->missing = 0;
175 npt->method = 0;
176 npt->statistics = 0;
178 memset (npt->a_statistics, 0, sizeof npt->a_statistics);
179 for (;;)
181 if (lex_match_id (lexer, "COCHRAN"))
183 npt->cochran++;
184 switch (npar_cochran (lexer, ds, nps))
186 case 0:
187 goto lossage;
188 case 1:
189 break;
190 case 2:
191 lex_error (lexer, NULL);
192 goto lossage;
193 default:
194 NOT_REACHED ();
197 else if (lex_match_id (lexer, "FRIEDMAN"))
199 npt->friedman++;
200 switch (npar_friedman (lexer, ds, nps))
202 case 0:
203 goto lossage;
204 case 1:
205 break;
206 case 2:
207 lex_error (lexer, NULL);
208 goto lossage;
209 default:
210 NOT_REACHED ();
213 else if (lex_match_id (lexer, "KENDALL"))
215 npt->kendall++;
216 switch (npar_kendall (lexer, ds, nps))
218 case 0:
219 goto lossage;
220 case 1:
221 break;
222 case 2:
223 lex_error (lexer, NULL);
224 goto lossage;
225 default:
226 NOT_REACHED ();
229 else if (lex_match_id (lexer, "RUNS"))
231 npt->runs++;
232 switch (npar_runs (lexer, ds, nps))
234 case 0:
235 goto lossage;
236 case 1:
237 break;
238 case 2:
239 lex_error (lexer, NULL);
240 goto lossage;
241 default:
242 NOT_REACHED ();
245 else if (lex_match_id (lexer, "CHISQUARE"))
247 lex_match (lexer, T_EQUALS);
248 npt->chisquare++;
249 switch (npar_chisquare (lexer, ds, nps))
251 case 0:
252 goto lossage;
253 case 1:
254 break;
255 case 2:
256 lex_error (lexer, NULL);
257 goto lossage;
258 case 3:
259 continue;
260 default:
261 NOT_REACHED ();
264 else if (lex_match_id (lexer, "BINOMIAL"))
266 lex_match (lexer, T_EQUALS);
267 npt->binomial++;
268 switch (npar_binomial (lexer, ds, nps))
270 case 0:
271 goto lossage;
272 case 1:
273 break;
274 case 2:
275 lex_error (lexer, NULL);
276 goto lossage;
277 default:
278 NOT_REACHED ();
281 else if (lex_match_phrase (lexer, "K-S") ||
282 lex_match_phrase (lexer, "KOLMOGOROV-SMIRNOV"))
284 lex_match (lexer, T_EQUALS);
285 npt->ks_one_sample++;
286 switch (npar_ks_one_sample (lexer, ds, nps))
288 case 0:
289 goto lossage;
290 case 1:
291 break;
292 case 2:
293 lex_error (lexer, NULL);
294 goto lossage;
295 default:
296 NOT_REACHED ();
299 else if (lex_match_phrase (lexer, "J-T") ||
300 lex_match_phrase (lexer, "JONCKHEERE-TERPSTRA"))
302 lex_match (lexer, T_EQUALS);
303 npt->jonckheere_terpstra++;
304 switch (npar_jonckheere_terpstra (lexer, ds, nps))
306 case 0:
307 goto lossage;
308 case 1:
309 break;
310 case 2:
311 lex_error (lexer, NULL);
312 goto lossage;
313 default:
314 NOT_REACHED ();
317 else if (lex_match_phrase (lexer, "K-W") ||
318 lex_match_phrase (lexer, "KRUSKAL-WALLIS"))
320 lex_match (lexer, T_EQUALS);
321 npt->kruskal_wallis++;
322 switch (npar_kruskal_wallis (lexer, ds, nps))
324 case 0:
325 goto lossage;
326 case 1:
327 break;
328 case 2:
329 lex_error (lexer, NULL);
330 goto lossage;
331 default:
332 NOT_REACHED ();
335 else if (lex_match_phrase (lexer, "MCNEMAR"))
337 lex_match (lexer, T_EQUALS);
338 npt->mcnemar++;
339 switch (npar_mcnemar (lexer, ds, nps))
341 case 0:
342 goto lossage;
343 case 1:
344 break;
345 case 2:
346 lex_error (lexer, NULL);
347 goto lossage;
348 default:
349 NOT_REACHED ();
352 else if (lex_match_phrase (lexer, "M-W") ||
353 lex_match_phrase (lexer, "MANN-WHITNEY"))
355 lex_match (lexer, T_EQUALS);
356 npt->mann_whitney++;
357 switch (npar_mann_whitney (lexer, ds, nps))
359 case 0:
360 goto lossage;
361 case 1:
362 break;
363 case 2:
364 lex_error (lexer, NULL);
365 goto lossage;
366 default:
367 NOT_REACHED ();
370 else if (lex_match_phrase (lexer, "MEDIAN"))
372 npt->median++;
374 switch (npar_median (lexer, ds, nps))
376 case 0:
377 goto lossage;
378 case 1:
379 break;
380 case 2:
381 lex_error (lexer, NULL);
382 goto lossage;
383 default:
384 NOT_REACHED ();
387 else if (lex_match_id (lexer, "WILCOXON"))
389 lex_match (lexer, T_EQUALS);
390 npt->wilcoxon++;
391 switch (npar_wilcoxon (lexer, ds, nps))
393 case 0:
394 goto lossage;
395 case 1:
396 break;
397 case 2:
398 lex_error (lexer, NULL);
399 goto lossage;
400 default:
401 NOT_REACHED ();
404 else if (lex_match_id (lexer, "SIGN"))
406 lex_match (lexer, T_EQUALS);
407 npt->sign++;
408 switch (npar_sign (lexer, ds, nps))
410 case 0:
411 goto lossage;
412 case 1:
413 break;
414 case 2:
415 lex_error (lexer, NULL);
416 goto lossage;
417 default:
418 NOT_REACHED ();
421 else if (lex_match_id (lexer, "MISSING"))
423 lex_match (lexer, T_EQUALS);
424 npt->missing++;
425 if (npt->missing > 1)
427 lex_sbc_only_once ("MISSING");
428 goto lossage;
430 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
432 if (lex_match_id (lexer, "ANALYSIS"))
433 npt->miss = MISS_ANALYSIS;
434 else if (lex_match_id (lexer, "LISTWISE"))
435 npt->miss = MISS_LISTWISE;
436 else if (lex_match_id (lexer, "INCLUDE"))
437 nps->filter = MV_SYSTEM;
438 else if (lex_match_id (lexer, "EXCLUDE"))
439 nps->filter = MV_ANY;
440 else
442 lex_error (lexer, NULL);
443 goto lossage;
445 lex_match (lexer, T_COMMA);
448 else if (lex_match_id (lexer, "METHOD"))
450 lex_match (lexer, T_EQUALS);
451 npt->method++;
452 if (npt->method > 1)
454 lex_sbc_only_once ("METHOD");
455 goto lossage;
457 switch (npar_method (lexer, nps))
459 case 0:
460 goto lossage;
461 case 1:
462 break;
463 case 2:
464 lex_error (lexer, NULL);
465 goto lossage;
466 default:
467 NOT_REACHED ();
470 else if (lex_match_id (lexer, "STATISTICS"))
472 lex_match (lexer, T_EQUALS);
473 npt->statistics++;
474 while (lex_token (lexer) != T_SLASH && lex_token (lexer) != T_ENDCMD)
476 if (lex_match_id (lexer, "DESCRIPTIVES"))
477 npt->a_statistics[NPAR_ST_DESCRIPTIVES] = 1;
478 else if (lex_match_id (lexer, "QUARTILES"))
479 npt->a_statistics[NPAR_ST_QUARTILES] = 1;
480 else if (lex_match (lexer, T_ALL))
481 npt->a_statistics[NPAR_ST_ALL] = 1;
482 else
484 lex_error (lexer, NULL);
485 goto lossage;
487 lex_match (lexer, T_COMMA);
490 else if ( settings_get_syntax () != COMPATIBLE && lex_match_id (lexer, "ALGORITHM"))
492 lex_match (lexer, T_EQUALS);
493 if (lex_match_id (lexer, "COMPATIBLE"))
494 settings_set_cmd_algorithm (COMPATIBLE);
495 else if (lex_match_id (lexer, "ENHANCED"))
496 settings_set_cmd_algorithm (ENHANCED);
498 if (!lex_match (lexer, T_SLASH))
499 break;
501 if (lex_token (lexer) != T_ENDCMD)
503 lex_error (lexer, _("expecting end of command"));
504 goto lossage;
507 return true;
509 lossage:
510 return false;
514 static void one_sample_insert_variables (const struct npar_test *test,
515 struct hmapx *);
517 static void two_sample_insert_variables (const struct npar_test *test,
518 struct hmapx *);
520 static void n_sample_insert_variables (const struct npar_test *test,
521 struct hmapx *);
523 static void
524 npar_execute (struct casereader *input,
525 const struct npar_specs *specs,
526 const struct dataset *ds)
528 int t;
529 struct descriptives *summary_descriptives = NULL;
531 for ( t = 0 ; t < specs->n_tests; ++t )
533 const struct npar_test *test = specs->test[t];
534 if ( NULL == test->execute )
536 msg (SW, _("%s subcommand not currently implemented."), "NPAR");
537 continue;
539 test->execute (ds, casereader_clone (input), specs->filter, test, specs->exact, specs->timer);
542 if (specs->descriptives && specs->n_vars > 0)
544 summary_descriptives = xnmalloc (sizeof (*summary_descriptives),
545 specs->n_vars);
547 npar_summary_calc_descriptives (summary_descriptives,
548 casereader_clone (input),
549 dataset_dict (ds),
550 specs->vv, specs->n_vars,
551 specs->filter);
554 if ( (specs->descriptives || specs->quartiles)
555 && !taint_has_tainted_successor (casereader_get_taint (input)) )
556 do_summary_box (summary_descriptives, specs->vv, specs->n_vars );
558 free (summary_descriptives);
559 casereader_destroy (input);
563 cmd_npar_tests (struct lexer *lexer, struct dataset *ds)
565 struct cmd_npar_tests cmd;
566 bool ok;
567 int i;
568 struct npar_specs npar_specs = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
569 struct casegrouper *grouper;
570 struct casereader *input, *group;
571 struct hmapx var_map = HMAPX_INITIALIZER (var_map);
574 npar_specs.pool = pool_create ();
575 npar_specs.filter = MV_ANY;
576 npar_specs.n_vars = -1;
577 npar_specs.vv = NULL;
579 if ( ! parse_npar_tests (lexer, ds, &cmd, &npar_specs) )
581 pool_destroy (npar_specs.pool);
582 return CMD_FAILURE;
585 for (i = 0; i < npar_specs.n_tests; ++i )
587 const struct npar_test *test = npar_specs.test[i];
588 test->insert_variables (test, &var_map);
592 struct hmapx_node *node;
593 struct variable *var;
594 npar_specs.n_vars = 0;
596 HMAPX_FOR_EACH (var, node, &var_map)
598 npar_specs.n_vars ++;
599 npar_specs.vv = pool_nrealloc (npar_specs.pool, npar_specs.vv, npar_specs.n_vars, sizeof (*npar_specs.vv));
600 npar_specs.vv[npar_specs.n_vars - 1] = var;
604 sort (npar_specs.vv, npar_specs.n_vars, sizeof (*npar_specs.vv),
605 compare_var_ptrs_by_name, NULL);
607 if ( cmd.statistics )
609 int i;
611 for ( i = 0 ; i < NPAR_ST_count; ++i )
613 if ( cmd.a_statistics[i] )
615 switch ( i )
617 case NPAR_ST_DESCRIPTIVES:
618 npar_specs.descriptives = true;
619 break;
620 case NPAR_ST_QUARTILES:
621 npar_specs.quartiles = true;
622 break;
623 case NPAR_ST_ALL:
624 npar_specs.quartiles = true;
625 npar_specs.descriptives = true;
626 break;
627 default:
628 NOT_REACHED ();
634 input = proc_open (ds);
635 if ( cmd.miss == MISS_LISTWISE )
637 input = casereader_create_filter_missing (input,
638 npar_specs.vv,
639 npar_specs.n_vars,
640 npar_specs.filter,
641 NULL, NULL);
645 grouper = casegrouper_create_splits (input, dataset_dict (ds));
646 while (casegrouper_get_next_group (grouper, &group))
647 npar_execute (group, &npar_specs, ds);
648 ok = casegrouper_destroy (grouper);
649 ok = proc_commit (ds) && ok;
651 pool_destroy (npar_specs.pool);
652 hmapx_destroy (&var_map);
654 return ok ? CMD_SUCCESS : CMD_CASCADING_FAILURE;
657 static int
658 npar_runs (struct lexer *lexer, struct dataset *ds,
659 struct npar_specs *specs)
661 struct runs_test *rt = pool_alloc (specs->pool, sizeof (*rt));
662 struct one_sample_test *tp = &rt->parent;
663 struct npar_test *nt = &tp->parent;
665 nt->execute = runs_execute;
666 nt->insert_variables = one_sample_insert_variables;
668 if ( lex_force_match (lexer, T_LPAREN) )
670 if ( lex_match_id (lexer, "MEAN"))
672 rt->cp_mode = CP_MEAN;
674 else if (lex_match_id (lexer, "MEDIAN"))
676 rt->cp_mode = CP_MEDIAN;
678 else if (lex_match_id (lexer, "MODE"))
680 rt->cp_mode = CP_MODE;
682 else if (lex_is_number (lexer))
684 rt->cutpoint = lex_number (lexer);
685 rt->cp_mode = CP_CUSTOM;
686 lex_get (lexer);
688 else
690 lex_error (lexer, _("Expecting %s, %s, %s or a number."), "MEAN", "MEDIAN", "MODE");
691 return 0;
694 if (! lex_force_match (lexer, T_RPAREN))
695 return 2;
697 if (! lex_force_match (lexer, T_EQUALS))
698 return 2;
700 if (!parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
701 &tp->vars, &tp->n_vars,
702 PV_NO_SCRATCH | PV_NO_DUPLICATE | PV_NUMERIC))
704 return 2;
708 specs->n_tests++;
709 specs->test = pool_realloc (specs->pool,
710 specs->test,
711 sizeof (*specs->test) * specs->n_tests);
713 specs->test[specs->n_tests - 1] = nt;
715 return 1;
718 static int
719 npar_friedman (struct lexer *lexer, struct dataset *ds,
720 struct npar_specs *specs)
722 struct friedman_test *ft = pool_alloc (specs->pool, sizeof (*ft));
723 struct one_sample_test *ost = &ft->parent;
724 struct npar_test *nt = &ost->parent;
726 ft->kendalls_w = false;
727 nt->execute = friedman_execute;
728 nt->insert_variables = one_sample_insert_variables;
730 lex_match (lexer, T_EQUALS);
732 if (!parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
733 &ost->vars, &ost->n_vars,
734 PV_NO_SCRATCH | PV_NO_DUPLICATE | PV_NUMERIC))
736 return 2;
739 specs->n_tests++;
740 specs->test = pool_realloc (specs->pool,
741 specs->test,
742 sizeof (*specs->test) * specs->n_tests);
744 specs->test[specs->n_tests - 1] = nt;
746 return 1;
749 static int
750 npar_kendall (struct lexer *lexer, struct dataset *ds,
751 struct npar_specs *specs)
753 struct friedman_test *kt = pool_alloc (specs->pool, sizeof (*kt));
754 struct one_sample_test *ost = &kt->parent;
755 struct npar_test *nt = &ost->parent;
757 kt->kendalls_w = true;
758 nt->execute = friedman_execute;
759 nt->insert_variables = one_sample_insert_variables;
761 lex_match (lexer, T_EQUALS);
763 if (!parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
764 &ost->vars, &ost->n_vars,
765 PV_NO_SCRATCH | PV_NO_DUPLICATE | PV_NUMERIC))
767 return 2;
770 specs->n_tests++;
771 specs->test = pool_realloc (specs->pool,
772 specs->test,
773 sizeof (*specs->test) * specs->n_tests);
775 specs->test[specs->n_tests - 1] = nt;
777 return 1;
781 static int
782 npar_cochran (struct lexer *lexer, struct dataset *ds,
783 struct npar_specs *specs)
785 struct one_sample_test *ft = pool_alloc (specs->pool, sizeof (*ft));
786 struct npar_test *nt = &ft->parent;
788 nt->execute = cochran_execute;
789 nt->insert_variables = one_sample_insert_variables;
791 lex_match (lexer, T_EQUALS);
793 if (!parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
794 &ft->vars, &ft->n_vars,
795 PV_NO_SCRATCH | PV_NO_DUPLICATE | PV_NUMERIC))
797 return 2;
800 specs->n_tests++;
801 specs->test = pool_realloc (specs->pool,
802 specs->test,
803 sizeof (*specs->test) * specs->n_tests);
805 specs->test[specs->n_tests - 1] = nt;
807 return 1;
811 static int
812 npar_chisquare (struct lexer *lexer, struct dataset *ds,
813 struct npar_specs *specs)
815 struct chisquare_test *cstp = pool_alloc (specs->pool, sizeof (*cstp));
816 struct one_sample_test *tp = &cstp->parent;
817 struct npar_test *nt = &tp->parent;
818 int retval = 1;
820 nt->execute = chisquare_execute;
821 nt->insert_variables = one_sample_insert_variables;
823 if (!parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
824 &tp->vars, &tp->n_vars,
825 PV_NO_SCRATCH | PV_NO_DUPLICATE))
827 return 2;
830 cstp->ranged = false;
832 if ( lex_match (lexer, T_LPAREN))
834 cstp->ranged = true;
835 if ( ! lex_force_num (lexer)) return 0;
836 cstp->lo = lex_number (lexer);
837 lex_get (lexer);
838 if (! lex_force_match (lexer, T_COMMA)) return 0;
839 if (! lex_force_num (lexer) ) return 0;
840 cstp->hi = lex_number (lexer);
841 if ( cstp->lo >= cstp->hi )
843 msg (ME,
844 _("The specified value of HI (%d) is "
845 "lower than the specified value of LO (%d)"),
846 cstp->hi, cstp->lo);
847 return 0;
849 lex_get (lexer);
850 if (! lex_force_match (lexer, T_RPAREN)) return 0;
853 cstp->n_expected = 0;
854 cstp->expected = NULL;
855 if (lex_match_phrase (lexer, "/EXPECTED"))
857 if (! lex_force_match (lexer, T_EQUALS)) return 0;
858 if (! lex_match_id (lexer, "EQUAL") )
860 double f;
861 int n;
862 while ( lex_is_number (lexer) )
864 int i;
865 n = 1;
866 f = lex_number (lexer);
867 lex_get (lexer);
868 if ( lex_match (lexer, T_ASTERISK))
870 n = f;
871 if (!lex_force_num (lexer))
872 return 0;
873 f = lex_number (lexer);
874 lex_get (lexer);
876 lex_match (lexer, T_COMMA);
878 cstp->n_expected += n;
879 cstp->expected = pool_realloc (specs->pool,
880 cstp->expected,
881 sizeof (double) *
882 cstp->n_expected);
883 for ( i = cstp->n_expected - n ;
884 i < cstp->n_expected;
885 ++i )
886 cstp->expected[i] = f;
892 if ( cstp->ranged && cstp->n_expected > 0 &&
893 cstp->n_expected != cstp->hi - cstp->lo + 1 )
895 msg (ME,
896 _("%d expected values were given, but the specified "
897 "range (%d-%d) requires exactly %d values."),
898 cstp->n_expected, cstp->lo, cstp->hi,
899 cstp->hi - cstp->lo +1);
900 return 0;
903 specs->n_tests++;
904 specs->test = pool_realloc (specs->pool,
905 specs->test,
906 sizeof (*specs->test) * specs->n_tests);
908 specs->test[specs->n_tests - 1] = nt;
910 return retval;
914 static int
915 npar_binomial (struct lexer *lexer, struct dataset *ds,
916 struct npar_specs *specs)
918 struct binomial_test *btp = pool_alloc (specs->pool, sizeof (*btp));
919 struct one_sample_test *tp = &btp->parent;
920 struct npar_test *nt = &tp->parent;
921 bool equals = false;
923 nt->execute = binomial_execute;
924 nt->insert_variables = one_sample_insert_variables;
926 btp->category1 = btp->category2 = btp->cutpoint = SYSMIS;
928 btp->p = 0.5;
930 if ( lex_match (lexer, T_LPAREN) )
932 equals = false;
933 if ( lex_force_num (lexer) )
935 btp->p = lex_number (lexer);
936 lex_get (lexer);
937 if (!lex_force_match (lexer, T_RPAREN))
938 return 0;
940 else
941 return 0;
943 else
944 equals = true;
946 if (!equals)
947 if (!lex_force_match (lexer, T_EQUALS))
948 return 0;
951 if (parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
952 &tp->vars, &tp->n_vars,
953 PV_NUMERIC | PV_NO_SCRATCH | PV_NO_DUPLICATE) )
955 if (lex_match (lexer, T_LPAREN))
957 if (! lex_force_num (lexer))
958 return 2;
959 btp->category1 = lex_number (lexer);
960 lex_get (lexer);
961 if ( lex_match (lexer, T_COMMA))
963 if ( ! lex_force_num (lexer) ) return 2;
964 btp->category2 = lex_number (lexer);
965 lex_get (lexer);
967 else
969 btp->cutpoint = btp->category1;
972 if (! lex_force_match (lexer, T_RPAREN))
973 return 0;
976 else
978 return 2;
982 specs->n_tests++;
983 specs->test = pool_realloc (specs->pool,
984 specs->test,
985 sizeof (*specs->test) * specs->n_tests);
987 specs->test[specs->n_tests - 1] = nt;
989 return 1;
994 static void
995 ks_one_sample_parse_params (struct lexer *lexer, struct ks_one_sample_test *kst, int params)
997 assert (params == 1 || params == 2);
999 if (lex_is_number (lexer))
1001 kst->p[0] = lex_number (lexer);
1003 lex_get (lexer);
1004 if ( params == 2)
1006 lex_match (lexer, T_COMMA);
1007 if (lex_force_num (lexer))
1009 kst->p[1] = lex_number (lexer);
1010 lex_get (lexer);
1016 static int
1017 npar_ks_one_sample (struct lexer *lexer, struct dataset *ds, struct npar_specs *specs)
1019 struct ks_one_sample_test *kst = pool_alloc (specs->pool, sizeof (*kst));
1020 struct one_sample_test *tp = &kst->parent;
1021 struct npar_test *nt = &tp->parent;
1023 nt->execute = ks_one_sample_execute;
1024 nt->insert_variables = one_sample_insert_variables;
1026 kst->p[0] = kst->p[1] = SYSMIS;
1028 if (! lex_force_match (lexer, T_LPAREN))
1029 return 2;
1031 if (lex_match_id (lexer, "NORMAL"))
1033 kst->dist = KS_NORMAL;
1034 ks_one_sample_parse_params (lexer, kst, 2);
1036 else if (lex_match_id (lexer, "POISSON"))
1038 kst->dist = KS_POISSON;
1039 ks_one_sample_parse_params (lexer, kst, 1);
1041 else if (lex_match_id (lexer, "UNIFORM"))
1043 kst->dist = KS_UNIFORM;
1044 ks_one_sample_parse_params (lexer, kst, 2);
1046 else if (lex_match_id (lexer, "EXPONENTIAL"))
1048 kst->dist = KS_EXPONENTIAL;
1049 ks_one_sample_parse_params (lexer, kst, 1);
1051 else
1052 return 2;
1054 if (! lex_force_match (lexer, T_RPAREN))
1055 return 2;
1057 lex_match (lexer, T_EQUALS);
1059 if (! parse_variables_const_pool (lexer, specs->pool, dataset_dict (ds),
1060 &tp->vars, &tp->n_vars,
1061 PV_NUMERIC | PV_NO_SCRATCH | PV_NO_DUPLICATE) )
1062 return 2;
1064 specs->n_tests++;
1065 specs->test = pool_realloc (specs->pool,
1066 specs->test,
1067 sizeof (*specs->test) * specs->n_tests);
1069 specs->test[specs->n_tests - 1] = nt;
1071 return 1;
1075 static bool
1076 parse_two_sample_related_test (struct lexer *lexer,
1077 const struct dictionary *dict,
1078 struct two_sample_test *test_parameters,
1079 struct pool *pool)
1081 int n = 0;
1082 bool paired = false;
1083 bool with = false;
1084 const struct variable **vlist1;
1085 size_t n_vlist1;
1087 const struct variable **vlist2;
1088 size_t n_vlist2;
1090 test_parameters->parent.insert_variables = two_sample_insert_variables;
1092 if (!parse_variables_const_pool (lexer, pool,
1093 dict,
1094 &vlist1, &n_vlist1,
1095 PV_NUMERIC | PV_NO_SCRATCH | PV_DUPLICATE) )
1096 return false;
1098 if ( lex_match (lexer, T_WITH))
1100 with = true;
1101 if ( !parse_variables_const_pool (lexer, pool, dict,
1102 &vlist2, &n_vlist2,
1103 PV_NUMERIC | PV_NO_SCRATCH | PV_DUPLICATE) )
1104 return false;
1106 paired = (lex_match (lexer, T_LPAREN) &&
1107 lex_match_id (lexer, "PAIRED") && lex_match (lexer, T_RPAREN));
1111 if ( with )
1113 if (paired)
1115 if ( n_vlist1 != n_vlist2)
1117 msg (SE, _("PAIRED was specified but the number of variables "
1118 "preceding WITH (%zu) did not match the number "
1119 "following (%zu)."), n_vlist1, n_vlist2);
1120 return false;
1123 test_parameters->n_pairs = n_vlist1 ;
1125 else
1127 test_parameters->n_pairs = n_vlist1 * n_vlist2;
1130 else
1132 test_parameters->n_pairs = (n_vlist1 * (n_vlist1 - 1)) / 2 ;
1135 test_parameters->pairs =
1136 pool_alloc (pool, sizeof ( variable_pair) * test_parameters->n_pairs);
1138 if ( with )
1140 if (paired)
1142 int i;
1143 assert (n_vlist1 == n_vlist2);
1144 for ( i = 0 ; i < n_vlist1; ++i )
1146 test_parameters->pairs[n][0] = vlist1[i];
1147 test_parameters->pairs[n][1] = vlist2[i];
1148 n++;
1151 else
1153 int i,j;
1154 for ( i = 0 ; i < n_vlist1; ++i )
1156 for ( j = 0 ; j < n_vlist2; ++j )
1158 test_parameters->pairs[n][0] = vlist1[i];
1159 test_parameters->pairs[n][1] = vlist2[j];
1160 n++;
1165 else
1167 int i,j;
1168 for ( i = 0 ; i < n_vlist1 - 1; ++i )
1170 for ( j = i + 1 ; j < n_vlist1; ++j )
1172 assert ( n < test_parameters->n_pairs);
1173 test_parameters->pairs[n][0] = vlist1[i];
1174 test_parameters->pairs[n][1] = vlist1[j];
1175 n++;
1180 assert ( n == test_parameters->n_pairs);
1182 return true;
1186 static bool
1187 parse_n_sample_related_test (struct lexer *lexer,
1188 const struct dictionary *dict,
1189 struct n_sample_test *nst,
1190 struct pool *pool
1193 if (!parse_variables_const_pool (lexer, pool,
1194 dict,
1195 &nst->vars, &nst->n_vars,
1196 PV_NUMERIC | PV_NO_SCRATCH | PV_NO_DUPLICATE) )
1197 return false;
1199 if ( ! lex_force_match (lexer, T_BY))
1200 return false;
1202 nst->indep_var = parse_variable_const (lexer, dict);
1203 if (!nst->indep_var)
1204 return false;
1206 if ( ! lex_force_match (lexer, T_LPAREN))
1207 return false;
1209 value_init (&nst->val1, var_get_width (nst->indep_var));
1210 if ( ! parse_value (lexer, &nst->val1, nst->indep_var))
1212 value_destroy (&nst->val1, var_get_width (nst->indep_var));
1213 return false;
1216 lex_match (lexer, T_COMMA);
1218 value_init (&nst->val2, var_get_width (nst->indep_var));
1219 if ( ! parse_value (lexer, &nst->val2, nst->indep_var))
1221 value_destroy (&nst->val2, var_get_width (nst->indep_var));
1222 return false;
1225 if ( ! lex_force_match (lexer, T_RPAREN))
1226 return false;
1228 return true;
1231 static int
1232 npar_wilcoxon (struct lexer *lexer,
1233 struct dataset *ds,
1234 struct npar_specs *specs )
1236 struct two_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
1237 struct npar_test *nt = &tp->parent;
1238 nt->execute = wilcoxon_execute;
1240 if (!parse_two_sample_related_test (lexer, dataset_dict (ds),
1241 tp, specs->pool) )
1242 return 0;
1244 specs->n_tests++;
1245 specs->test = pool_realloc (specs->pool,
1246 specs->test,
1247 sizeof (*specs->test) * specs->n_tests);
1248 specs->test[specs->n_tests - 1] = nt;
1250 return 1;
1254 static int
1255 npar_mann_whitney (struct lexer *lexer,
1256 struct dataset *ds,
1257 struct npar_specs *specs )
1259 struct n_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
1260 struct npar_test *nt = &tp->parent;
1262 nt->insert_variables = n_sample_insert_variables;
1263 nt->execute = mann_whitney_execute;
1265 if (!parse_n_sample_related_test (lexer, dataset_dict (ds),
1266 tp, specs->pool) )
1267 return 0;
1269 specs->n_tests++;
1270 specs->test = pool_realloc (specs->pool,
1271 specs->test,
1272 sizeof (*specs->test) * specs->n_tests);
1273 specs->test[specs->n_tests - 1] = nt;
1275 return 1;
1279 static int
1280 npar_median (struct lexer *lexer,
1281 struct dataset *ds,
1282 struct npar_specs *specs)
1284 struct median_test *mt = pool_alloc (specs->pool, sizeof (*mt));
1285 struct n_sample_test *tp = &mt->parent;
1286 struct npar_test *nt = &tp->parent;
1288 mt->median = SYSMIS;
1290 if ( lex_match (lexer, T_LPAREN) && lex_force_num (lexer))
1292 mt->median = lex_number (lexer);
1293 lex_get (lexer);
1294 if (! lex_force_match (lexer, T_RPAREN))
1295 return 0;
1298 lex_match (lexer, T_EQUALS);
1300 nt->insert_variables = n_sample_insert_variables;
1301 nt->execute = median_execute;
1303 if (!parse_n_sample_related_test (lexer, dataset_dict (ds),
1304 tp, specs->pool) )
1305 return 0;
1307 specs->n_tests++;
1308 specs->test = pool_realloc (specs->pool,
1309 specs->test,
1310 sizeof (*specs->test) * specs->n_tests);
1311 specs->test[specs->n_tests - 1] = nt;
1313 return 1;
1317 static int
1318 npar_sign (struct lexer *lexer, struct dataset *ds,
1319 struct npar_specs *specs)
1321 struct two_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
1322 struct npar_test *nt = &tp->parent;
1324 nt->execute = sign_execute;
1326 if (!parse_two_sample_related_test (lexer, dataset_dict (ds),
1327 tp, specs->pool) )
1328 return 0;
1330 specs->n_tests++;
1331 specs->test = pool_realloc (specs->pool,
1332 specs->test,
1333 sizeof (*specs->test) * specs->n_tests);
1334 specs->test[specs->n_tests - 1] = nt;
1336 return 1;
1340 static int
1341 npar_mcnemar (struct lexer *lexer, struct dataset *ds,
1342 struct npar_specs *specs)
1344 struct two_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
1345 struct npar_test *nt = &tp->parent;
1347 nt->execute = mcnemar_execute;
1349 if (!parse_two_sample_related_test (lexer, dataset_dict (ds),
1350 tp, specs->pool) )
1351 return 0;
1353 specs->n_tests++;
1354 specs->test = pool_realloc (specs->pool,
1355 specs->test,
1356 sizeof (*specs->test) * specs->n_tests);
1357 specs->test[specs->n_tests - 1] = nt;
1359 return 1;
1363 static int
1364 npar_jonckheere_terpstra (struct lexer *lexer, struct dataset *ds,
1365 struct npar_specs *specs)
1367 struct n_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
1368 struct npar_test *nt = &tp->parent;
1370 nt->insert_variables = n_sample_insert_variables;
1372 nt->execute = jonckheere_terpstra_execute;
1374 if (!parse_n_sample_related_test (lexer, dataset_dict (ds),
1375 tp, specs->pool) )
1376 return 0;
1378 specs->n_tests++;
1379 specs->test = pool_realloc (specs->pool,
1380 specs->test,
1381 sizeof (*specs->test) * specs->n_tests);
1382 specs->test[specs->n_tests - 1] = nt;
1384 return 1;
1387 static int
1388 npar_kruskal_wallis (struct lexer *lexer, struct dataset *ds,
1389 struct npar_specs *specs)
1391 struct n_sample_test *tp = pool_alloc (specs->pool, sizeof (*tp));
1392 struct npar_test *nt = &tp->parent;
1394 nt->insert_variables = n_sample_insert_variables;
1396 nt->execute = kruskal_wallis_execute;
1398 if (!parse_n_sample_related_test (lexer, dataset_dict (ds),
1399 tp, specs->pool) )
1400 return 0;
1402 specs->n_tests++;
1403 specs->test = pool_realloc (specs->pool,
1404 specs->test,
1405 sizeof (*specs->test) * specs->n_tests);
1406 specs->test[specs->n_tests - 1] = nt;
1408 return 1;
1411 static void
1412 insert_variable_into_map (struct hmapx *var_map, const struct variable *var)
1414 size_t hash = hash_pointer (var, 0);
1415 struct hmapx_node *node;
1416 const struct variable *v = NULL;
1418 HMAPX_FOR_EACH_WITH_HASH (v, node, hash, var_map)
1420 if ( v == var)
1421 return ;
1424 hmapx_insert (var_map, CONST_CAST (struct variable *, var), hash);
1427 /* Insert the variables for TEST into VAR_MAP */
1428 static void
1429 one_sample_insert_variables (const struct npar_test *test,
1430 struct hmapx *var_map)
1432 int i;
1433 const struct one_sample_test *ost = UP_CAST (test, const struct one_sample_test, parent);
1435 for ( i = 0 ; i < ost->n_vars ; ++i )
1436 insert_variable_into_map (var_map, ost->vars[i]);
1440 static void
1441 two_sample_insert_variables (const struct npar_test *test,
1442 struct hmapx *var_map)
1444 int i;
1445 const struct two_sample_test *tst = UP_CAST (test, const struct two_sample_test, parent);
1447 for ( i = 0 ; i < tst->n_pairs ; ++i )
1449 variable_pair *pair = &tst->pairs[i];
1451 insert_variable_into_map (var_map, (*pair)[0]);
1452 insert_variable_into_map (var_map, (*pair)[1]);
1456 static void
1457 n_sample_insert_variables (const struct npar_test *test,
1458 struct hmapx *var_map)
1460 int i;
1461 const struct n_sample_test *tst = UP_CAST (test, const struct n_sample_test, parent);
1463 for ( i = 0 ; i < tst->n_vars ; ++i )
1464 insert_variable_into_map (var_map, tst->vars[i]);
1466 insert_variable_into_map (var_map, tst->indep_var);
1470 static int
1471 npar_method (struct lexer *lexer, struct npar_specs *specs)
1473 if ( lex_match_id (lexer, "EXACT") )
1475 specs->exact = true;
1476 specs->timer = 0.0;
1477 if (lex_match_id (lexer, "TIMER"))
1479 specs->timer = 5.0;
1481 if ( lex_match (lexer, T_LPAREN))
1483 if (lex_force_num (lexer) )
1485 specs->timer = lex_number (lexer);
1486 lex_get (lexer);
1488 if (lex_force_match (lexer, T_RPAREN))
1489 return 0;
1494 return 1;