1 /* $OpenBSD: man_validate.c,v 1.129 2023/10/24 20:30:49 schwarze Exp $ */
2 /*
3 * Copyright (c) 2010, 2012-2020, 2023 Ingo Schwarze <schwarze@openbsd.org>
4 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
5 *
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
9 *
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 *
18 * Validation module for man(7) syntax trees used by mandoc(1).
19 */
20 #include <sys/types.h>
21
22 #include <assert.h>
23 #include <ctype.h>
24 #include <errno.h>
25 #include <limits.h>
26 #include <stdarg.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 #include <time.h>
31
32 #include "mandoc_aux.h"
33 #include "mandoc.h"
34 #include "mandoc_xr.h"
35 #include "roff.h"
36 #include "man.h"
37 #include "libmandoc.h"
38 #include "roff_int.h"
39 #include "libman.h"
40 #include "tag.h"
41
42 #define CHKARGS struct roff_man *man, struct roff_node *n
43
44 typedef void (*v_check)(CHKARGS);
45
46 static void check_par(CHKARGS);
47 static void check_part(CHKARGS);
48 static void check_root(CHKARGS);
49 static void check_tag(struct roff_node *, struct roff_node *);
50 static void check_text(CHKARGS);
51
52 static void post_AT(CHKARGS);
53 static void post_EE(CHKARGS);
54 static void post_EX(CHKARGS);
55 static void post_IP(CHKARGS);
56 static void post_MR(CHKARGS);
57 static void post_OP(CHKARGS);
58 static void post_SH(CHKARGS);
59 static void post_TH(CHKARGS);
60 static void post_TP(CHKARGS);
61 static void post_UC(CHKARGS);
62 static void post_UR(CHKARGS);
63 static void post_in(CHKARGS);
64
65 static const v_check man_valids[MAN_MAX - MAN_TH] = {
66 post_TH, /* TH */
67 post_SH, /* SH */
68 post_SH, /* SS */
69 post_TP, /* TP */
70 post_TP, /* TQ */
71 check_par, /* LP */
72 check_par, /* PP */
73 check_par, /* P */
74 post_IP, /* IP */
75 NULL, /* HP */
76 NULL, /* SM */
77 NULL, /* SB */
78 NULL, /* BI */
79 NULL, /* IB */
80 NULL, /* BR */
81 NULL, /* RB */
82 NULL, /* R */
83 NULL, /* B */
84 NULL, /* I */
85 NULL, /* IR */
86 NULL, /* RI */
87 NULL, /* RE */
88 check_part, /* RS */
89 NULL, /* DT */
90 post_UC, /* UC */
91 NULL, /* PD */
92 post_AT, /* AT */
93 post_in, /* in */
94 NULL, /* SY */
95 NULL, /* YS */
96 post_OP, /* OP */
97 post_EX, /* EX */
98 post_EE, /* EE */
99 post_UR, /* UR */
100 NULL, /* UE */
101 post_UR, /* MT */
102 NULL, /* ME */
103 post_MR, /* MR */
104 };
105
106
107 /* Validate the subtree rooted at man->last. */
108 void
man_validate(struct roff_man * man)109 man_validate(struct roff_man *man)
110 {
111 struct roff_node *n;
112 const v_check *cp;
113
114 /*
115 * Iterate over all children, recursing into each one
116 * in turn, depth-first.
117 */
118
119 n = man->last;
120 man->last = man->last->child;
121 while (man->last != NULL) {
122 man_validate(man);
123 if (man->last == n)
124 man->last = man->last->child;
125 else
126 man->last = man->last->next;
127 }
128
129 /* Finally validate the macro itself. */
130
131 man->last = n;
132 man->next = ROFF_NEXT_SIBLING;
133 switch (n->type) {
134 case ROFFT_TEXT:
135 check_text(man, n);
136 break;
137 case ROFFT_ROOT:
138 check_root(man, n);
139 break;
140 case ROFFT_COMMENT:
141 case ROFFT_EQN:
142 case ROFFT_TBL:
143 break;
144 default:
145 if (n->tok < ROFF_MAX) {
146 roff_validate(man);
147 break;
148 }
149 assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
150 cp = man_valids + (n->tok - MAN_TH);
151 if (*cp)
152 (*cp)(man, n);
153 if (man->last == n)
154 n->flags |= NODE_VALID;
155 break;
156 }
157 }
158
159 static void
check_root(CHKARGS)160 check_root(CHKARGS)
161 {
162 assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
163
164 if (n->last == NULL || n->last->type == ROFFT_COMMENT)
165 mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
166 else
167 man->meta.hasbody = 1;
168
169 if (NULL == man->meta.title) {
170 mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
171
172 /*
173 * If a title hasn't been set, do so now (by
174 * implication, date and section also aren't set).
175 */
176
177 man->meta.title = mandoc_strdup("");
178 man->meta.msec = mandoc_strdup("");
179 man->meta.date = mandoc_normdate(NULL, NULL);
180 }
181
182 if (man->meta.os_e &&
183 (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
184 mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
185 man->meta.os_e == MANDOC_OS_OPENBSD ?
186 "(OpenBSD)" : "(NetBSD)");
187 }
188
189 /*
190 * Skip leading whitespace, dashes, backslashes, and font escapes,
191 * then create a tag if the first following byte is a letter.
192 * Priority is high unless whitespace is present.
193 */
194 static void
check_tag(struct roff_node * n,struct roff_node * nt)195 check_tag(struct roff_node *n, struct roff_node *nt)
196 {
197 const char *cp, *arg;
198 int prio, sz;
199
200 if (nt == NULL || nt->type != ROFFT_TEXT)
201 return;
202
203 cp = nt->string;
204 prio = TAG_STRONG;
205 for (;;) {
206 switch (*cp) {
207 case ' ':
208 case '\t':
209 prio = TAG_WEAK;
210 /* FALLTHROUGH */
211 case '-':
212 cp++;
213 break;
214 case '\\':
215 cp++;
216 switch (mandoc_escape(&cp, &arg, &sz)) {
217 case ESCAPE_FONT:
218 case ESCAPE_FONTBOLD:
219 case ESCAPE_FONTITALIC:
220 case ESCAPE_FONTBI:
221 case ESCAPE_FONTROMAN:
222 case ESCAPE_FONTCR:
223 case ESCAPE_FONTCB:
224 case ESCAPE_FONTCI:
225 case ESCAPE_FONTPREV:
226 case ESCAPE_IGNORE:
227 break;
228 case ESCAPE_SPECIAL:
229 if (sz != 1)
230 return;
231 switch (*arg) {
232 case '-':
233 case 'e':
234 break;
235 default:
236 return;
237 }
238 break;
239 default:
240 return;
241 }
242 break;
243 default:
244 if (isalpha((unsigned char)*cp))
245 tag_put(cp, prio, n);
246 return;
247 }
248 }
249 }
250
251 static void
check_text(CHKARGS)252 check_text(CHKARGS)
253 {
254 char *cp, *p;
255
256 if (n->flags & NODE_NOFILL)
257 return;
258
259 cp = n->string;
260 for (p = cp; NULL != (p = strchr(p, '\t')); p++)
261 mandoc_msg(MANDOCERR_FI_TAB,
262 n->line, n->pos + (int)(p - cp), NULL);
263 }
264
265 static void
post_EE(CHKARGS)266 post_EE(CHKARGS)
267 {
268 if ((n->flags & NODE_NOFILL) == 0)
269 mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
270 }
271
272 static void
post_EX(CHKARGS)273 post_EX(CHKARGS)
274 {
275 if (n->flags & NODE_NOFILL)
276 mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
277 }
278
279 static void
post_OP(CHKARGS)280 post_OP(CHKARGS)
281 {
282
283 if (n->child == NULL)
284 mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
285 else if (n->child->next != NULL && n->child->next->next != NULL) {
286 n = n->child->next->next;
287 mandoc_msg(MANDOCERR_ARG_EXCESS,
288 n->line, n->pos, "OP ... %s", n->string);
289 }
290 }
291
292 static void
post_SH(CHKARGS)293 post_SH(CHKARGS)
294 {
295 struct roff_node *nc;
296 char *cp, *tag;
297
298 nc = n->child;
299 switch (n->type) {
300 case ROFFT_HEAD:
301 tag = NULL;
302 deroff(&tag, n);
303 if (tag != NULL) {
304 for (cp = tag; *cp != '\0'; cp++)
305 if (*cp == ' ')
306 *cp = '_';
307 if (nc != NULL && nc->type == ROFFT_TEXT &&
308 strcmp(nc->string, tag) == 0)
309 tag_put(NULL, TAG_STRONG, n);
310 else
311 tag_put(tag, TAG_FALLBACK, n);
312 free(tag);
313 }
314 return;
315 case ROFFT_BODY:
316 if (nc != NULL)
317 break;
318 return;
319 default:
320 return;
321 }
322
323 if ((nc->tok == MAN_LP || nc->tok == MAN_PP || nc->tok == MAN_P) &&
324 nc->body->child != NULL) {
325 while (nc->body->last != NULL) {
326 man->next = ROFF_NEXT_CHILD;
327 roff_node_relink(man, nc->body->last);
328 man->last = n;
329 }
330 }
331
332 if (nc->tok == MAN_LP || nc->tok == MAN_PP || nc->tok == MAN_P ||
333 nc->tok == ROFF_sp || nc->tok == ROFF_br) {
334 mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
335 "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
336 roff_node_delete(man, nc);
337 }
338
339 /*
340 * Trailing PP is empty, so it is deleted by check_par().
341 * Trailing sp is significant.
342 */
343
344 if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
345 mandoc_msg(MANDOCERR_PAR_SKIP,
346 nc->line, nc->pos, "%s at the end of %s",
347 roff_name[nc->tok], roff_name[n->tok]);
348 roff_node_delete(man, nc);
349 }
350 }
351
352 static void
post_UR(CHKARGS)353 post_UR(CHKARGS)
354 {
355 if (n->type == ROFFT_HEAD && n->child == NULL)
356 mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
357 "%s", roff_name[n->tok]);
358 }
359
360 static void
check_part(CHKARGS)361 check_part(CHKARGS)
362 {
363 if (n->type == ROFFT_BODY && n->child == NULL)
364 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
365 "%s", roff_name[n->tok]);
366 }
367
368 static void
check_par(CHKARGS)369 check_par(CHKARGS)
370 {
371
372 switch (n->type) {
373 case ROFFT_BLOCK:
374 if (n->body->child == NULL)
375 roff_node_delete(man, n);
376 break;
377 case ROFFT_BODY:
378 if (n->child != NULL &&
379 (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
380 mandoc_msg(MANDOCERR_PAR_SKIP,
381 n->child->line, n->child->pos,
382 "%s after %s", roff_name[n->child->tok],
383 roff_name[n->tok]);
384 roff_node_delete(man, n->child);
385 }
386 if (n->child == NULL)
387 mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
388 "%s empty", roff_name[n->tok]);
389 break;
390 case ROFFT_HEAD:
391 if (n->child != NULL)
392 mandoc_msg(MANDOCERR_ARG_SKIP,
393 n->line, n->pos, "%s %s%s",
394 roff_name[n->tok], n->child->string,
395 n->child->next != NULL ? " ..." : "");
396 break;
397 default:
398 break;
399 }
400 }
401
402 static void
post_IP(CHKARGS)403 post_IP(CHKARGS)
404 {
405 switch (n->type) {
406 case ROFFT_BLOCK:
407 if (n->head->child == NULL && n->body->child == NULL)
408 roff_node_delete(man, n);
409 break;
410 case ROFFT_HEAD:
411 check_tag(n, n->child);
412 break;
413 case ROFFT_BODY:
414 if (n->parent->head->child == NULL && n->child == NULL)
415 mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
416 "%s empty", roff_name[n->tok]);
417 break;
418 default:
419 break;
420 }
421 }
422
423 /*
424 * The first next-line element in the head is the tag.
425 * If that's a font macro, use its first child instead.
426 */
427 static void
post_TP(CHKARGS)428 post_TP(CHKARGS)
429 {
430 struct roff_node *nt;
431
432 if (n->type != ROFFT_HEAD || (nt = n->child) == NULL)
433 return;
434
435 while ((nt->flags & NODE_LINE) == 0)
436 if ((nt = nt->next) == NULL)
437 return;
438
439 switch (nt->tok) {
440 case MAN_B:
441 case MAN_BI:
442 case MAN_BR:
443 case MAN_I:
444 case MAN_IB:
445 case MAN_IR:
446 nt = nt->child;
447 break;
448 default:
449 break;
450 }
451 check_tag(n, nt);
452 }
453
454 static void
post_TH(CHKARGS)455 post_TH(CHKARGS)
456 {
457 struct roff_node *nb;
458 const char *p;
459
460 free(man->meta.title);
461 free(man->meta.vol);
462 free(man->meta.os);
463 free(man->meta.msec);
464 free(man->meta.date);
465
466 man->meta.title = man->meta.vol = man->meta.date =
467 man->meta.msec = man->meta.os = NULL;
468
469 nb = n;
470
471 /* ->TITLE<- MSEC DATE OS VOL */
472
473 n = n->child;
474 if (n != NULL && n->string != NULL) {
475 for (p = n->string; *p != '\0'; p++) {
476 /* Only warn about this once... */
477 if (isalpha((unsigned char)*p) &&
478 ! isupper((unsigned char)*p)) {
479 mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
480 n->pos + (int)(p - n->string),
481 "TH %s", n->string);
482 break;
483 }
484 }
485 man->meta.title = mandoc_strdup(n->string);
486 } else {
487 man->meta.title = mandoc_strdup("");
488 mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
489 }
490
491 /* TITLE ->MSEC<- DATE OS VOL */
492
493 if (n != NULL)
494 n = n->next;
495 if (n != NULL && n->string != NULL) {
496 man->meta.msec = mandoc_strdup(n->string);
497 if (man->filesec != '\0' &&
498 man->filesec != *n->string &&
499 *n->string >= '1' && *n->string <= '9')
500 mandoc_msg(MANDOCERR_MSEC_FILE, n->line, n->pos,
501 "*.%c vs TH ... %c", man->filesec, *n->string);
502 } else {
503 man->meta.msec = mandoc_strdup("");
504 mandoc_msg(MANDOCERR_MSEC_MISSING,
505 nb->line, nb->pos, "TH %s", man->meta.title);
506 }
507
508 /* TITLE MSEC ->DATE<- OS VOL */
509
510 if (n != NULL)
511 n = n->next;
512 if (man->quick && n != NULL)
513 man->meta.date = mandoc_strdup("");
514 else
515 man->meta.date = mandoc_normdate(n, nb);
516
517 /* TITLE MSEC DATE ->OS<- VOL */
518
519 if (n && (n = n->next))
520 man->meta.os = mandoc_strdup(n->string);
521 else if (man->os_s != NULL)
522 man->meta.os = mandoc_strdup(man->os_s);
523 if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
524 if (strstr(man->meta.os, "OpenBSD") != NULL)
525 man->meta.os_e = MANDOC_OS_OPENBSD;
526 else if (strstr(man->meta.os, "NetBSD") != NULL)
527 man->meta.os_e = MANDOC_OS_NETBSD;
528 }
529
530 /* TITLE MSEC DATE OS ->VOL<- */
531 /* If missing, use the default VOL name for MSEC. */
532
533 if (n && (n = n->next))
534 man->meta.vol = mandoc_strdup(n->string);
535 else if ('\0' != man->meta.msec[0] &&
536 (NULL != (p = mandoc_a2msec(man->meta.msec))))
537 man->meta.vol = mandoc_strdup(p);
538
539 if (n != NULL && (n = n->next) != NULL)
540 mandoc_msg(MANDOCERR_ARG_EXCESS,
541 n->line, n->pos, "TH ... %s", n->string);
542
543 /*
544 * Remove the `TH' node after we've processed it for our
545 * meta-data.
546 */
547 roff_node_delete(man, man->last);
548 }
549
550 static void
post_MR(CHKARGS)551 post_MR(CHKARGS)
552 {
553 struct roff_node *nch;
554
555 if ((nch = n->child) == NULL) {
556 mandoc_msg(MANDOCERR_NM_NONAME, n->line, n->pos, "MR");
557 return;
558 }
559 if (nch->next == NULL) {
560 mandoc_msg(MANDOCERR_XR_NOSEC,
561 n->line, n->pos, "MR %s", nch->string);
562 return;
563 }
564 if (mandoc_xr_add(nch->next->string, nch->string, nch->line, nch->pos))
565 mandoc_msg(MANDOCERR_XR_SELF, nch->line, nch->pos,
566 "MR %s %s", nch->string, nch->next->string);
567 if ((nch = nch->next->next) == NULL || nch->next == NULL)
568 return;
569
570 mandoc_msg(MANDOCERR_ARG_EXCESS, nch->next->line, nch->next->pos,
571 "MR ... %s", nch->next->string);
572 while (nch->next != NULL)
573 roff_node_delete(man, nch->next);
574 }
575
576 static void
post_UC(CHKARGS)577 post_UC(CHKARGS)
578 {
579 static const char * const bsd_versions[] = {
580 "3rd Berkeley Distribution",
581 "4th Berkeley Distribution",
582 "4.2 Berkeley Distribution",
583 "4.3 Berkeley Distribution",
584 "4.4 Berkeley Distribution",
585 };
586
587 const char *p, *s;
588
589 n = n->child;
590
591 if (n == NULL || n->type != ROFFT_TEXT)
592 p = bsd_versions[0];
593 else {
594 s = n->string;
595 if (0 == strcmp(s, "3"))
596 p = bsd_versions[0];
597 else if (0 == strcmp(s, "4"))
598 p = bsd_versions[1];
599 else if (0 == strcmp(s, "5"))
600 p = bsd_versions[2];
601 else if (0 == strcmp(s, "6"))
602 p = bsd_versions[3];
603 else if (0 == strcmp(s, "7"))
604 p = bsd_versions[4];
605 else
606 p = bsd_versions[0];
607 }
608
609 free(man->meta.os);
610 man->meta.os = mandoc_strdup(p);
611 }
612
613 static void
post_AT(CHKARGS)614 post_AT(CHKARGS)
615 {
616 static const char * const unix_versions[] = {
617 "7th Edition",
618 "System III",
619 "System V",
620 "System V Release 2",
621 };
622
623 struct roff_node *nn;
624 const char *p, *s;
625
626 n = n->child;
627
628 if (n == NULL || n->type != ROFFT_TEXT)
629 p = unix_versions[0];
630 else {
631 s = n->string;
632 if (0 == strcmp(s, "3"))
633 p = unix_versions[0];
634 else if (0 == strcmp(s, "4"))
635 p = unix_versions[1];
636 else if (0 == strcmp(s, "5")) {
637 nn = n->next;
638 if (nn != NULL &&
639 nn->type == ROFFT_TEXT &&
640 nn->string[0] != '\0')
641 p = unix_versions[3];
642 else
643 p = unix_versions[2];
644 } else
645 p = unix_versions[0];
646 }
647
648 free(man->meta.os);
649 man->meta.os = mandoc_strdup(p);
650 }
651
652 static void
post_in(CHKARGS)653 post_in(CHKARGS)
654 {
655 char *s;
656
657 if (n->parent->tok != MAN_TP ||
658 n->parent->type != ROFFT_HEAD ||
659 n->child == NULL ||
660 *n->child->string == '+' ||
661 *n->child->string == '-')
662 return;
663 mandoc_asprintf(&s, "+%s", n->child->string);
664 free(n->child->string);
665 n->child->string = s;
666 }
667