1 /* $Id: man_validate.c,v 1.159 2023/10/24 20:53:12 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 "config.h"
21
22 #include <sys/types.h>
23
24 #include <assert.h>
25 #include <ctype.h>
26 #include <errno.h>
27 #include <limits.h>
28 #include <stdarg.h>
29 #include <stdio.h>
30 #include <stdlib.h>
31 #include <string.h>
32 #include <time.h>
33
34 #include "mandoc_aux.h"
35 #include "mandoc.h"
36 #include "mandoc_xr.h"
37 #include "roff.h"
38 #include "man.h"
39 #include "libmandoc.h"
40 #include "roff_int.h"
41 #include "libman.h"
42 #include "tag.h"
43
44 #define CHKARGS struct roff_man *man, struct roff_node *n
45
46 typedef void (*v_check)(CHKARGS);
47
48 static void check_par(CHKARGS);
49 static void check_part(CHKARGS);
50 static void check_root(CHKARGS);
51 static void check_tag(struct roff_node *, struct roff_node *);
52 static void check_text(CHKARGS);
53
54 static void post_AT(CHKARGS);
55 static void post_EE(CHKARGS);
56 static void post_EX(CHKARGS);
57 static void post_IP(CHKARGS);
58 static void post_MR(CHKARGS);
59 static void post_OP(CHKARGS);
60 static void post_SH(CHKARGS);
61 static void post_TH(CHKARGS);
62 static void post_TP(CHKARGS);
63 static void post_UC(CHKARGS);
64 static void post_UR(CHKARGS);
65 static void post_in(CHKARGS);
66
67 static const v_check man_valids[MAN_MAX - MAN_TH] = {
68 post_TH, /* TH */
69 post_SH, /* SH */
70 post_SH, /* SS */
71 post_TP, /* TP */
72 post_TP, /* TQ */
73 check_par, /* LP */
74 check_par, /* PP */
75 check_par, /* P */
76 post_IP, /* IP */
77 NULL, /* HP */
78 NULL, /* SM */
79 NULL, /* SB */
80 NULL, /* BI */
81 NULL, /* IB */
82 NULL, /* BR */
83 NULL, /* RB */
84 NULL, /* R */
85 NULL, /* B */
86 NULL, /* I */
87 NULL, /* IR */
88 NULL, /* RI */
89 NULL, /* RE */
90 check_part, /* RS */
91 NULL, /* DT */
92 post_UC, /* UC */
93 NULL, /* PD */
94 post_AT, /* AT */
95 post_in, /* in */
96 NULL, /* SY */
97 NULL, /* YS */
98 post_OP, /* OP */
99 post_EX, /* EX */
100 post_EE, /* EE */
101 post_UR, /* UR */
102 NULL, /* UE */
103 post_UR, /* MT */
104 NULL, /* ME */
105 post_MR, /* MR */
106 };
107
108
109 /* Validate the subtree rooted at man->last. */
110 void
man_validate(struct roff_man * man)111 man_validate(struct roff_man *man)
112 {
113 struct roff_node *n;
114 const v_check *cp;
115
116 /*
117 * Iterate over all children, recursing into each one
118 * in turn, depth-first.
119 */
120
121 n = man->last;
122 man->last = man->last->child;
123 while (man->last != NULL) {
124 man_validate(man);
125 if (man->last == n)
126 man->last = man->last->child;
127 else
128 man->last = man->last->next;
129 }
130
131 /* Finally validate the macro itself. */
132
133 man->last = n;
134 man->next = ROFF_NEXT_SIBLING;
135 switch (n->type) {
136 case ROFFT_TEXT:
137 check_text(man, n);
138 break;
139 case ROFFT_ROOT:
140 check_root(man, n);
141 break;
142 case ROFFT_COMMENT:
143 case ROFFT_EQN:
144 case ROFFT_TBL:
145 break;
146 default:
147 if (n->tok < ROFF_MAX) {
148 roff_validate(man);
149 break;
150 }
151 assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
152 cp = man_valids + (n->tok - MAN_TH);
153 if (*cp)
154 (*cp)(man, n);
155 if (man->last == n)
156 n->flags |= NODE_VALID;
157 break;
158 }
159 }
160
161 static void
check_root(CHKARGS)162 check_root(CHKARGS)
163 {
164 assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
165
166 if (n->last == NULL || n->last->type == ROFFT_COMMENT)
167 mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
168 else
169 man->meta.hasbody = 1;
170
171 if (NULL == man->meta.title) {
172 mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
173
174 /*
175 * If a title hasn't been set, do so now (by
176 * implication, date and section also aren't set).
177 */
178
179 man->meta.title = mandoc_strdup("");
180 man->meta.msec = mandoc_strdup("");
181 man->meta.date = mandoc_normdate(NULL, NULL);
182 }
183
184 if (man->meta.os_e &&
185 (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
186 mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
187 man->meta.os_e == MANDOC_OS_OPENBSD ?
188 "(OpenBSD)" : "(NetBSD)");
189 }
190
191 /*
192 * Skip leading whitespace, dashes, backslashes, and font escapes,
193 * then create a tag if the first following byte is a letter.
194 * Priority is high unless whitespace is present.
195 */
196 static void
check_tag(struct roff_node * n,struct roff_node * nt)197 check_tag(struct roff_node *n, struct roff_node *nt)
198 {
199 const char *cp, *arg;
200 int prio, sz;
201
202 if (nt == NULL || nt->type != ROFFT_TEXT)
203 return;
204
205 cp = nt->string;
206 prio = TAG_STRONG;
207 for (;;) {
208 switch (*cp) {
209 case ' ':
210 case '\t':
211 prio = TAG_WEAK;
212 /* FALLTHROUGH */
213 case '-':
214 cp++;
215 break;
216 case '\\':
217 cp++;
218 switch (mandoc_escape(&cp, &arg, &sz)) {
219 case ESCAPE_FONT:
220 case ESCAPE_FONTBOLD:
221 case ESCAPE_FONTITALIC:
222 case ESCAPE_FONTBI:
223 case ESCAPE_FONTROMAN:
224 case ESCAPE_FONTCR:
225 case ESCAPE_FONTCB:
226 case ESCAPE_FONTCI:
227 case ESCAPE_FONTPREV:
228 case ESCAPE_IGNORE:
229 break;
230 case ESCAPE_SPECIAL:
231 if (sz != 1)
232 return;
233 switch (*arg) {
234 case '-':
235 case 'e':
236 break;
237 default:
238 return;
239 }
240 break;
241 default:
242 return;
243 }
244 break;
245 default:
246 if (isalpha((unsigned char)*cp))
247 tag_put(cp, prio, n);
248 return;
249 }
250 }
251 }
252
253 static void
check_text(CHKARGS)254 check_text(CHKARGS)
255 {
256 char *cp, *p;
257
258 if (n->flags & NODE_NOFILL)
259 return;
260
261 cp = n->string;
262 for (p = cp; NULL != (p = strchr(p, '\t')); p++)
263 mandoc_msg(MANDOCERR_FI_TAB,
264 n->line, n->pos + (int)(p - cp), NULL);
265 }
266
267 static void
post_EE(CHKARGS)268 post_EE(CHKARGS)
269 {
270 if ((n->flags & NODE_NOFILL) == 0)
271 mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
272 }
273
274 static void
post_EX(CHKARGS)275 post_EX(CHKARGS)
276 {
277 if (n->flags & NODE_NOFILL)
278 mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
279 }
280
281 static void
post_OP(CHKARGS)282 post_OP(CHKARGS)
283 {
284
285 if (n->child == NULL)
286 mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
287 else if (n->child->next != NULL && n->child->next->next != NULL) {
288 n = n->child->next->next;
289 mandoc_msg(MANDOCERR_ARG_EXCESS,
290 n->line, n->pos, "OP ... %s", n->string);
291 }
292 }
293
294 static void
post_SH(CHKARGS)295 post_SH(CHKARGS)
296 {
297 struct roff_node *nc;
298 char *cp, *tag;
299
300 nc = n->child;
301 switch (n->type) {
302 case ROFFT_HEAD:
303 tag = NULL;
304 deroff(&tag, n);
305 if (tag != NULL) {
306 for (cp = tag; *cp != '\0'; cp++)
307 if (*cp == ' ')
308 *cp = '_';
309 if (nc != NULL && nc->type == ROFFT_TEXT &&
310 strcmp(nc->string, tag) == 0)
311 tag_put(NULL, TAG_STRONG, n);
312 else
313 tag_put(tag, TAG_FALLBACK, n);
314 free(tag);
315 }
316 return;
317 case ROFFT_BODY:
318 if (nc != NULL)
319 break;
320 return;
321 default:
322 return;
323 }
324
325 if ((nc->tok == MAN_LP || nc->tok == MAN_PP || nc->tok == MAN_P) &&
326 nc->body->child != NULL) {
327 while (nc->body->last != NULL) {
328 man->next = ROFF_NEXT_CHILD;
329 roff_node_relink(man, nc->body->last);
330 man->last = n;
331 }
332 }
333
334 if (nc->tok == MAN_LP || nc->tok == MAN_PP || nc->tok == MAN_P ||
335 nc->tok == ROFF_sp || nc->tok == ROFF_br) {
336 mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
337 "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
338 roff_node_delete(man, nc);
339 }
340
341 /*
342 * Trailing PP is empty, so it is deleted by check_par().
343 * Trailing sp is significant.
344 */
345
346 if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
347 mandoc_msg(MANDOCERR_PAR_SKIP,
348 nc->line, nc->pos, "%s at the end of %s",
349 roff_name[nc->tok], roff_name[n->tok]);
350 roff_node_delete(man, nc);
351 }
352 }
353
354 static void
post_UR(CHKARGS)355 post_UR(CHKARGS)
356 {
357 if (n->type == ROFFT_HEAD && n->child == NULL)
358 mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
359 "%s", roff_name[n->tok]);
360 }
361
362 static void
check_part(CHKARGS)363 check_part(CHKARGS)
364 {
365 if (n->type == ROFFT_BODY && n->child == NULL)
366 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
367 "%s", roff_name[n->tok]);
368 }
369
370 static void
check_par(CHKARGS)371 check_par(CHKARGS)
372 {
373
374 switch (n->type) {
375 case ROFFT_BLOCK:
376 if (n->body->child == NULL)
377 roff_node_delete(man, n);
378 break;
379 case ROFFT_BODY:
380 if (n->child != NULL &&
381 (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
382 mandoc_msg(MANDOCERR_PAR_SKIP,
383 n->child->line, n->child->pos,
384 "%s after %s", roff_name[n->child->tok],
385 roff_name[n->tok]);
386 roff_node_delete(man, n->child);
387 }
388 if (n->child == NULL)
389 mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
390 "%s empty", roff_name[n->tok]);
391 break;
392 case ROFFT_HEAD:
393 if (n->child != NULL)
394 mandoc_msg(MANDOCERR_ARG_SKIP,
395 n->line, n->pos, "%s %s%s",
396 roff_name[n->tok], n->child->string,
397 n->child->next != NULL ? " ..." : "");
398 break;
399 default:
400 break;
401 }
402 }
403
404 static void
post_IP(CHKARGS)405 post_IP(CHKARGS)
406 {
407 switch (n->type) {
408 case ROFFT_BLOCK:
409 if (n->head->child == NULL && n->body->child == NULL)
410 roff_node_delete(man, n);
411 break;
412 case ROFFT_HEAD:
413 check_tag(n, n->child);
414 break;
415 case ROFFT_BODY:
416 if (n->parent->head->child == NULL && n->child == NULL)
417 mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
418 "%s empty", roff_name[n->tok]);
419 break;
420 default:
421 break;
422 }
423 }
424
425 /*
426 * The first next-line element in the head is the tag.
427 * If that's a font macro, use its first child instead.
428 */
429 static void
post_TP(CHKARGS)430 post_TP(CHKARGS)
431 {
432 struct roff_node *nt;
433
434 if (n->type != ROFFT_HEAD || (nt = n->child) == NULL)
435 return;
436
437 while ((nt->flags & NODE_LINE) == 0)
438 if ((nt = nt->next) == NULL)
439 return;
440
441 switch (nt->tok) {
442 case MAN_B:
443 case MAN_BI:
444 case MAN_BR:
445 case MAN_I:
446 case MAN_IB:
447 case MAN_IR:
448 nt = nt->child;
449 break;
450 default:
451 break;
452 }
453 check_tag(n, nt);
454 }
455
456 static void
post_TH(CHKARGS)457 post_TH(CHKARGS)
458 {
459 struct roff_node *nb;
460 const char *p;
461
462 free(man->meta.title);
463 free(man->meta.vol);
464 free(man->meta.os);
465 free(man->meta.msec);
466 free(man->meta.date);
467
468 man->meta.title = man->meta.vol = man->meta.date =
469 man->meta.msec = man->meta.os = NULL;
470
471 nb = n;
472
473 /* ->TITLE<- MSEC DATE OS VOL */
474
475 n = n->child;
476 if (n != NULL && n->string != NULL) {
477 for (p = n->string; *p != '\0'; p++) {
478 /* Only warn about this once... */
479 if (isalpha((unsigned char)*p) &&
480 ! isupper((unsigned char)*p)) {
481 mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
482 n->pos + (int)(p - n->string),
483 "TH %s", n->string);
484 break;
485 }
486 }
487 man->meta.title = mandoc_strdup(n->string);
488 } else {
489 man->meta.title = mandoc_strdup("");
490 mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
491 }
492
493 /* TITLE ->MSEC<- DATE OS VOL */
494
495 if (n != NULL)
496 n = n->next;
497 if (n != NULL && n->string != NULL) {
498 man->meta.msec = mandoc_strdup(n->string);
499 if (man->filesec != '\0' &&
500 man->filesec != *n->string &&
501 *n->string >= '1' && *n->string <= '9')
502 mandoc_msg(MANDOCERR_MSEC_FILE, n->line, n->pos,
503 "*.%c vs TH ... %c", man->filesec, *n->string);
504 } else {
505 man->meta.msec = mandoc_strdup("");
506 mandoc_msg(MANDOCERR_MSEC_MISSING,
507 nb->line, nb->pos, "TH %s", man->meta.title);
508 }
509
510 /* TITLE MSEC ->DATE<- OS VOL */
511
512 if (n != NULL)
513 n = n->next;
514 if (man->quick && n != NULL)
515 man->meta.date = mandoc_strdup("");
516 else
517 man->meta.date = mandoc_normdate(n, nb);
518
519 /* TITLE MSEC DATE ->OS<- VOL */
520
521 if (n && (n = n->next))
522 man->meta.os = mandoc_strdup(n->string);
523 else if (man->os_s != NULL)
524 man->meta.os = mandoc_strdup(man->os_s);
525 if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
526 if (strstr(man->meta.os, "OpenBSD") != NULL)
527 man->meta.os_e = MANDOC_OS_OPENBSD;
528 else if (strstr(man->meta.os, "NetBSD") != NULL)
529 man->meta.os_e = MANDOC_OS_NETBSD;
530 }
531
532 /* TITLE MSEC DATE OS ->VOL<- */
533 /* If missing, use the default VOL name for MSEC. */
534
535 if (n && (n = n->next))
536 man->meta.vol = mandoc_strdup(n->string);
537 else if ('\0' != man->meta.msec[0] &&
538 (NULL != (p = mandoc_a2msec(man->meta.msec))))
539 man->meta.vol = mandoc_strdup(p);
540
541 if (n != NULL && (n = n->next) != NULL)
542 mandoc_msg(MANDOCERR_ARG_EXCESS,
543 n->line, n->pos, "TH ... %s", n->string);
544
545 /*
546 * Remove the `TH' node after we've processed it for our
547 * meta-data.
548 */
549 roff_node_delete(man, man->last);
550 }
551
552 static void
post_MR(CHKARGS)553 post_MR(CHKARGS)
554 {
555 struct roff_node *nch;
556
557 if ((nch = n->child) == NULL) {
558 mandoc_msg(MANDOCERR_NM_NONAME, n->line, n->pos, "MR");
559 return;
560 }
561 if (nch->next == NULL) {
562 mandoc_msg(MANDOCERR_XR_NOSEC,
563 n->line, n->pos, "MR %s", nch->string);
564 return;
565 }
566 if (mandoc_xr_add(nch->next->string, nch->string, nch->line, nch->pos))
567 mandoc_msg(MANDOCERR_XR_SELF, nch->line, nch->pos,
568 "MR %s %s", nch->string, nch->next->string);
569 if ((nch = nch->next->next) == NULL || nch->next == NULL)
570 return;
571
572 mandoc_msg(MANDOCERR_ARG_EXCESS, nch->next->line, nch->next->pos,
573 "MR ... %s", nch->next->string);
574 while (nch->next != NULL)
575 roff_node_delete(man, nch->next);
576 }
577
578 static void
post_UC(CHKARGS)579 post_UC(CHKARGS)
580 {
581 static const char * const bsd_versions[] = {
582 "3rd Berkeley Distribution",
583 "4th Berkeley Distribution",
584 "4.2 Berkeley Distribution",
585 "4.3 Berkeley Distribution",
586 "4.4 Berkeley Distribution",
587 };
588
589 const char *p, *s;
590
591 n = n->child;
592
593 if (n == NULL || n->type != ROFFT_TEXT)
594 p = bsd_versions[0];
595 else {
596 s = n->string;
597 if (0 == strcmp(s, "3"))
598 p = bsd_versions[0];
599 else if (0 == strcmp(s, "4"))
600 p = bsd_versions[1];
601 else if (0 == strcmp(s, "5"))
602 p = bsd_versions[2];
603 else if (0 == strcmp(s, "6"))
604 p = bsd_versions[3];
605 else if (0 == strcmp(s, "7"))
606 p = bsd_versions[4];
607 else
608 p = bsd_versions[0];
609 }
610
611 free(man->meta.os);
612 man->meta.os = mandoc_strdup(p);
613 }
614
615 static void
post_AT(CHKARGS)616 post_AT(CHKARGS)
617 {
618 static const char * const unix_versions[] = {
619 "7th Edition",
620 "System III",
621 "System V",
622 "System V Release 2",
623 };
624
625 struct roff_node *nn;
626 const char *p, *s;
627
628 n = n->child;
629
630 if (n == NULL || n->type != ROFFT_TEXT)
631 p = unix_versions[0];
632 else {
633 s = n->string;
634 if (0 == strcmp(s, "3"))
635 p = unix_versions[0];
636 else if (0 == strcmp(s, "4"))
637 p = unix_versions[1];
638 else if (0 == strcmp(s, "5")) {
639 nn = n->next;
640 if (nn != NULL &&
641 nn->type == ROFFT_TEXT &&
642 nn->string[0] != '\0')
643 p = unix_versions[3];
644 else
645 p = unix_versions[2];
646 } else
647 p = unix_versions[0];
648 }
649
650 free(man->meta.os);
651 man->meta.os = mandoc_strdup(p);
652 }
653
654 static void
post_in(CHKARGS)655 post_in(CHKARGS)
656 {
657 char *s;
658
659 if (n->parent->tok != MAN_TP ||
660 n->parent->type != ROFFT_HEAD ||
661 n->child == NULL ||
662 *n->child->string == '+' ||
663 *n->child->string == '-')
664 return;
665 mandoc_asprintf(&s, "+%s", n->child->string);
666 free(n->child->string);
667 n->child->string = s;
668 }
669