Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.62
1.62 ! schwarze 1: /* $Id: man_validate.c,v 1.61 2014/03/21 22:17:01 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.60 schwarze 4: * Copyright (c) 2010, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.2 schwarze 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.2 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR 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.
1.1 kristaps 17: */
18: #include <sys/types.h>
19:
20: #include <assert.h>
21: #include <ctype.h>
1.6 schwarze 22: #include <errno.h>
23: #include <limits.h>
1.1 kristaps 24: #include <stdarg.h>
25: #include <stdlib.h>
1.28 schwarze 26: #include <string.h>
1.36 schwarze 27: #include <time.h>
1.1 kristaps 28:
1.44 schwarze 29: #include "man.h"
1.25 schwarze 30: #include "mandoc.h"
1.61 schwarze 31: #include "mandoc_aux.h"
1.1 kristaps 32: #include "libman.h"
1.5 schwarze 33: #include "libmandoc.h"
1.1 kristaps 34:
1.57 schwarze 35: #define CHKARGS struct man *man, struct man_node *n
1.1 kristaps 36:
1.7 schwarze 37: typedef int (*v_check)(CHKARGS);
1.1 kristaps 38:
39: struct man_valid {
1.7 schwarze 40: v_check *pres;
41: v_check *posts;
1.1 kristaps 42: };
43:
1.7 schwarze 44: static int check_eq0(CHKARGS);
1.52 schwarze 45: static int check_eq2(CHKARGS);
1.10 schwarze 46: static int check_le1(CHKARGS);
1.7 schwarze 47: static int check_ge2(CHKARGS);
48: static int check_le5(CHKARGS);
1.58 schwarze 49: static int check_head1(CHKARGS);
1.7 schwarze 50: static int check_par(CHKARGS);
1.8 schwarze 51: static int check_part(CHKARGS);
1.7 schwarze 52: static int check_root(CHKARGS);
1.45 schwarze 53: static void check_text(CHKARGS);
1.7 schwarze 54:
1.34 schwarze 55: static int post_AT(CHKARGS);
1.55 schwarze 56: static int post_IP(CHKARGS);
1.46 schwarze 57: static int post_vs(CHKARGS);
1.34 schwarze 58: static int post_fi(CHKARGS);
1.47 schwarze 59: static int post_ft(CHKARGS);
1.34 schwarze 60: static int post_nf(CHKARGS);
1.47 schwarze 61: static int post_sec(CHKARGS);
1.34 schwarze 62: static int post_TH(CHKARGS);
63: static int post_UC(CHKARGS);
1.47 schwarze 64: static int pre_sec(CHKARGS);
1.34 schwarze 65:
66: static v_check posts_at[] = { post_AT, NULL };
1.46 schwarze 67: static v_check posts_br[] = { post_vs, check_eq0, NULL };
1.7 schwarze 68: static v_check posts_eq0[] = { check_eq0, NULL };
1.52 schwarze 69: static v_check posts_eq2[] = { check_eq2, NULL };
1.34 schwarze 70: static v_check posts_fi[] = { check_eq0, post_fi, NULL };
1.47 schwarze 71: static v_check posts_ft[] = { post_ft, NULL };
1.55 schwarze 72: static v_check posts_ip[] = { post_IP, NULL };
1.56 schwarze 73: static v_check posts_le1[] = { check_le1, NULL };
1.34 schwarze 74: static v_check posts_nf[] = { check_eq0, post_nf, NULL };
1.7 schwarze 75: static v_check posts_par[] = { check_par, NULL };
1.8 schwarze 76: static v_check posts_part[] = { check_part, NULL };
1.47 schwarze 77: static v_check posts_sec[] = { post_sec, NULL };
1.46 schwarze 78: static v_check posts_sp[] = { post_vs, check_le1, NULL };
1.40 schwarze 79: static v_check posts_th[] = { check_ge2, check_le5, post_TH, NULL };
1.34 schwarze 80: static v_check posts_uc[] = { post_UC, NULL };
1.58 schwarze 81: static v_check posts_ur[] = { check_head1, check_part, NULL };
1.50 schwarze 82: static v_check pres_sec[] = { pre_sec, NULL };
1.1 kristaps 83:
84: static const struct man_valid man_valids[MAN_MAX] = {
1.46 schwarze 85: { NULL, posts_br }, /* br */
1.50 schwarze 86: { NULL, posts_th }, /* TH */
1.47 schwarze 87: { pres_sec, posts_sec }, /* SH */
88: { pres_sec, posts_sec }, /* SS */
1.50 schwarze 89: { NULL, NULL }, /* TP */
90: { NULL, posts_par }, /* LP */
91: { NULL, posts_par }, /* PP */
92: { NULL, posts_par }, /* P */
1.55 schwarze 93: { NULL, posts_ip }, /* IP */
1.50 schwarze 94: { NULL, NULL }, /* HP */
1.8 schwarze 95: { NULL, NULL }, /* SM */
96: { NULL, NULL }, /* SB */
1.7 schwarze 97: { NULL, NULL }, /* BI */
98: { NULL, NULL }, /* IB */
99: { NULL, NULL }, /* BR */
100: { NULL, NULL }, /* RB */
1.8 schwarze 101: { NULL, NULL }, /* R */
102: { NULL, NULL }, /* B */
103: { NULL, NULL }, /* I */
1.7 schwarze 104: { NULL, NULL }, /* IR */
105: { NULL, NULL }, /* RI */
1.52 schwarze 106: { NULL, posts_eq0 }, /* na */
107: { NULL, posts_sp }, /* sp */
1.50 schwarze 108: { NULL, posts_nf }, /* nf */
109: { NULL, posts_fi }, /* fi */
1.8 schwarze 110: { NULL, NULL }, /* RE */
111: { NULL, posts_part }, /* RS */
112: { NULL, NULL }, /* DT */
1.34 schwarze 113: { NULL, posts_uc }, /* UC */
1.56 schwarze 114: { NULL, posts_le1 }, /* PD */
1.34 schwarze 115: { NULL, posts_at }, /* AT */
1.28 schwarze 116: { NULL, NULL }, /* in */
1.32 schwarze 117: { NULL, posts_ft }, /* ft */
1.52 schwarze 118: { NULL, posts_eq2 }, /* OP */
1.53 schwarze 119: { NULL, posts_nf }, /* EX */
120: { NULL, posts_fi }, /* EE */
1.58 schwarze 121: { NULL, posts_ur }, /* UR */
122: { NULL, NULL }, /* UE */
1.62 ! schwarze 123: { NULL, NULL }, /* ll */
1.1 kristaps 124: };
125:
126:
127: int
1.57 schwarze 128: man_valid_pre(struct man *man, struct man_node *n)
1.7 schwarze 129: {
130: v_check *cp;
131:
1.38 schwarze 132: switch (n->type) {
133: case (MAN_TEXT):
134: /* FALLTHROUGH */
135: case (MAN_ROOT):
136: /* FALLTHROUGH */
1.42 schwarze 137: case (MAN_EQN):
138: /* FALLTHROUGH */
1.38 schwarze 139: case (MAN_TBL):
1.7 schwarze 140: return(1);
1.38 schwarze 141: default:
142: break;
143: }
1.7 schwarze 144:
145: if (NULL == (cp = man_valids[n->tok].pres))
146: return(1);
147: for ( ; *cp; cp++)
1.57 schwarze 148: if ( ! (*cp)(man, n))
1.7 schwarze 149: return(0);
150: return(1);
151: }
152:
153:
154: int
1.57 schwarze 155: man_valid_post(struct man *man)
1.1 kristaps 156: {
1.7 schwarze 157: v_check *cp;
1.1 kristaps 158:
1.57 schwarze 159: if (MAN_VALID & man->last->flags)
1.1 kristaps 160: return(1);
1.57 schwarze 161: man->last->flags |= MAN_VALID;
1.1 kristaps 162:
1.57 schwarze 163: switch (man->last->type) {
1.1 kristaps 164: case (MAN_TEXT):
1.57 schwarze 165: check_text(man, man->last);
1.45 schwarze 166: return(1);
1.1 kristaps 167: case (MAN_ROOT):
1.57 schwarze 168: return(check_root(man, man->last));
1.42 schwarze 169: case (MAN_EQN):
170: /* FALLTHROUGH */
1.38 schwarze 171: case (MAN_TBL):
172: return(1);
1.1 kristaps 173: default:
174: break;
175: }
176:
1.57 schwarze 177: if (NULL == (cp = man_valids[man->last->tok].posts))
1.1 kristaps 178: return(1);
179: for ( ; *cp; cp++)
1.57 schwarze 180: if ( ! (*cp)(man, man->last))
1.1 kristaps 181: return(0);
182:
183: return(1);
184: }
185:
186:
1.4 schwarze 187: static int
1.7 schwarze 188: check_root(CHKARGS)
1.4 schwarze 189: {
1.7 schwarze 190:
1.57 schwarze 191: if (MAN_BLINE & man->flags)
192: man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
193: else if (MAN_ELINE & man->flags)
194: man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
1.8 schwarze 195:
1.57 schwarze 196: man->flags &= ~MAN_BLINE;
197: man->flags &= ~MAN_ELINE;
1.7 schwarze 198:
1.57 schwarze 199: if (NULL == man->first->child) {
200: man_nmsg(man, n, MANDOCERR_NODOCBODY);
1.25 schwarze 201: return(0);
1.57 schwarze 202: } else if (NULL == man->meta.title) {
203: man_nmsg(man, n, MANDOCERR_NOTITLE);
1.35 schwarze 204:
1.18 schwarze 205: /*
206: * If a title hasn't been set, do so now (by
207: * implication, date and section also aren't set).
208: */
1.35 schwarze 209:
1.57 schwarze 210: man->meta.title = mandoc_strdup("unknown");
211: man->meta.msec = mandoc_strdup("1");
1.60 schwarze 212: man->meta.date = man->quick ? mandoc_strdup("") :
213: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 214: }
1.15 schwarze 215:
216: return(1);
217: }
218:
1.45 schwarze 219: static void
1.47 schwarze 220: check_text(CHKARGS)
1.4 schwarze 221: {
1.47 schwarze 222: char *cp, *p;
223:
1.57 schwarze 224: if (MAN_LITERAL & man->flags)
1.48 schwarze 225: return;
226:
227: cp = n->string;
228: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.57 schwarze 229: man_pmsg(man, n->line, (int)(p - cp), MANDOCERR_BADTAB);
1.1 kristaps 230: }
231:
232: #define INEQ_DEFINE(x, ineq, name) \
233: static int \
1.7 schwarze 234: check_##name(CHKARGS) \
1.1 kristaps 235: { \
1.4 schwarze 236: if (n->nchild ineq (x)) \
1.1 kristaps 237: return(1); \
1.57 schwarze 238: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line, n->pos, \
1.25 schwarze 239: "line arguments %s %d (have %d)", \
240: #ineq, (x), n->nchild); \
1.40 schwarze 241: return(1); \
1.1 kristaps 242: }
243:
244: INEQ_DEFINE(0, ==, eq0)
1.52 schwarze 245: INEQ_DEFINE(2, ==, eq2)
1.10 schwarze 246: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 247: INEQ_DEFINE(2, >=, ge2)
248: INEQ_DEFINE(5, <=, le5)
1.58 schwarze 249:
250: static int
251: check_head1(CHKARGS)
252: {
253:
254: if (MAN_HEAD == n->type && 1 != n->nchild)
255: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
256: n->pos, "line arguments eq 1 (have %d)", n->nchild);
257:
258: return(1);
259: }
1.32 schwarze 260:
261: static int
1.47 schwarze 262: post_ft(CHKARGS)
1.32 schwarze 263: {
264: char *cp;
265: int ok;
266:
267: if (0 == n->nchild)
268: return(1);
269:
270: ok = 0;
271: cp = n->child->string;
272: switch (*cp) {
273: case ('1'):
274: /* FALLTHROUGH */
275: case ('2'):
276: /* FALLTHROUGH */
277: case ('3'):
278: /* FALLTHROUGH */
279: case ('4'):
280: /* FALLTHROUGH */
281: case ('I'):
282: /* FALLTHROUGH */
283: case ('P'):
284: /* FALLTHROUGH */
285: case ('R'):
286: if ('\0' == cp[1])
287: ok = 1;
288: break;
289: case ('B'):
290: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
291: ok = 1;
292: break;
293: case ('C'):
294: if ('W' == cp[1] && '\0' == cp[2])
295: ok = 1;
296: break;
297: default:
298: break;
299: }
300:
301: if (0 == ok) {
1.44 schwarze 302: mandoc_vmsg
1.57 schwarze 303: (MANDOCERR_BADFONT, man->parse,
1.44 schwarze 304: n->line, n->pos, "%s", cp);
1.32 schwarze 305: *cp = '\0';
306: }
307:
308: if (1 < n->nchild)
1.44 schwarze 309: mandoc_vmsg
1.57 schwarze 310: (MANDOCERR_ARGCOUNT, man->parse, n->line,
1.44 schwarze 311: n->pos, "want one child (have %d)",
312: n->nchild);
1.32 schwarze 313:
314: return(1);
315: }
1.7 schwarze 316:
317: static int
1.47 schwarze 318: pre_sec(CHKARGS)
319: {
320:
321: if (MAN_BLOCK == n->type)
1.57 schwarze 322: man->flags &= ~MAN_LITERAL;
1.47 schwarze 323: return(1);
324: }
325:
326: static int
327: post_sec(CHKARGS)
1.7 schwarze 328: {
1.6 schwarze 329:
1.44 schwarze 330: if ( ! (MAN_HEAD == n->type && 0 == n->nchild))
331: return(1);
1.25 schwarze 332:
1.57 schwarze 333: man_nmsg(man, n, MANDOCERR_SYNTARGCOUNT);
1.44 schwarze 334: return(0);
1.6 schwarze 335: }
1.7 schwarze 336:
337: static int
1.8 schwarze 338: check_part(CHKARGS)
339: {
340:
341: if (MAN_BODY == n->type && 0 == n->nchild)
1.57 schwarze 342: mandoc_msg(MANDOCERR_ARGCWARN, man->parse, n->line,
1.44 schwarze 343: n->pos, "want children (have none)");
1.35 schwarze 344:
1.8 schwarze 345: return(1);
346: }
347:
1.36 schwarze 348:
349: static int
350: check_par(CHKARGS)
351: {
352:
1.39 schwarze 353: switch (n->type) {
354: case (MAN_BLOCK):
355: if (0 == n->body->nchild)
1.57 schwarze 356: man_node_delete(man, n);
1.39 schwarze 357: break;
358: case (MAN_BODY):
359: if (0 == n->nchild)
1.57 schwarze 360: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.39 schwarze 361: break;
362: case (MAN_HEAD):
363: if (n->nchild)
1.57 schwarze 364: man_nmsg(man, n, MANDOCERR_ARGSLOST);
1.39 schwarze 365: break;
366: default:
367: break;
368: }
1.36 schwarze 369:
370: return(1);
371: }
372:
1.55 schwarze 373: static int
374: post_IP(CHKARGS)
375: {
376:
377: switch (n->type) {
378: case (MAN_BLOCK):
379: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 schwarze 380: man_node_delete(man, n);
1.55 schwarze 381: break;
382: case (MAN_BODY):
383: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.57 schwarze 384: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.55 schwarze 385: break;
386: default:
387: break;
388: }
389: return(1);
390: }
1.36 schwarze 391:
1.34 schwarze 392: static int
393: post_TH(CHKARGS)
394: {
1.40 schwarze 395: const char *p;
1.34 schwarze 396:
1.57 schwarze 397: free(man->meta.title);
398: free(man->meta.vol);
399: free(man->meta.source);
400: free(man->meta.msec);
401: free(man->meta.date);
1.34 schwarze 402:
1.57 schwarze 403: man->meta.title = man->meta.vol = man->meta.date =
404: man->meta.msec = man->meta.source = NULL;
1.34 schwarze 405:
406: /* ->TITLE<- MSEC DATE SOURCE VOL */
407:
408: n = n->child;
1.40 schwarze 409: if (n && n->string) {
410: for (p = n->string; '\0' != *p; p++) {
411: /* Only warn about this once... */
1.47 schwarze 412: if (isalpha((unsigned char)*p) &&
413: ! isupper((unsigned char)*p)) {
1.57 schwarze 414: man_nmsg(man, n, MANDOCERR_UPPERCASE);
1.40 schwarze 415: break;
416: }
417: }
1.57 schwarze 418: man->meta.title = mandoc_strdup(n->string);
1.40 schwarze 419: } else
1.57 schwarze 420: man->meta.title = mandoc_strdup("");
1.34 schwarze 421:
422: /* TITLE ->MSEC<- DATE SOURCE VOL */
423:
1.40 schwarze 424: if (n)
425: n = n->next;
426: if (n && n->string)
1.57 schwarze 427: man->meta.msec = mandoc_strdup(n->string);
1.40 schwarze 428: else
1.57 schwarze 429: man->meta.msec = mandoc_strdup("");
1.34 schwarze 430:
431: /* TITLE MSEC ->DATE<- SOURCE VOL */
432:
1.40 schwarze 433: if (n)
434: n = n->next;
1.49 schwarze 435: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 436: man->meta.date = man->quick ?
437: mandoc_strdup(n->string) :
438: mandoc_normdate(man->parse, n->string,
439: n->line, n->pos);
1.49 schwarze 440: } else
1.57 schwarze 441: man->meta.date = mandoc_strdup("");
1.34 schwarze 442:
443: /* TITLE MSEC DATE ->SOURCE<- VOL */
444:
445: if (n && (n = n->next))
1.57 schwarze 446: man->meta.source = mandoc_strdup(n->string);
1.34 schwarze 447:
448: /* TITLE MSEC DATE SOURCE ->VOL<- */
1.51 schwarze 449: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 450:
451: if (n && (n = n->next))
1.57 schwarze 452: man->meta.vol = mandoc_strdup(n->string);
453: else if ('\0' != man->meta.msec[0] &&
454: (NULL != (p = mandoc_a2msec(man->meta.msec))))
455: man->meta.vol = mandoc_strdup(p);
1.34 schwarze 456:
457: /*
458: * Remove the `TH' node after we've processed it for our
459: * meta-data.
460: */
1.57 schwarze 461: man_node_delete(man, man->last);
1.34 schwarze 462: return(1);
463: }
464:
465: static int
466: post_nf(CHKARGS)
467: {
468:
1.57 schwarze 469: if (MAN_LITERAL & man->flags)
470: man_nmsg(man, n, MANDOCERR_SCOPEREP);
1.34 schwarze 471:
1.57 schwarze 472: man->flags |= MAN_LITERAL;
1.34 schwarze 473: return(1);
474: }
475:
476: static int
477: post_fi(CHKARGS)
478: {
479:
1.57 schwarze 480: if ( ! (MAN_LITERAL & man->flags))
481: man_nmsg(man, n, MANDOCERR_WNOSCOPE);
1.34 schwarze 482:
1.57 schwarze 483: man->flags &= ~MAN_LITERAL;
1.34 schwarze 484: return(1);
485: }
486:
487: static int
488: post_UC(CHKARGS)
489: {
490: static const char * const bsd_versions[] = {
491: "3rd Berkeley Distribution",
492: "4th Berkeley Distribution",
493: "4.2 Berkeley Distribution",
494: "4.3 Berkeley Distribution",
495: "4.4 Berkeley Distribution",
496: };
497:
498: const char *p, *s;
499:
500: n = n->child;
501:
502: if (NULL == n || MAN_TEXT != n->type)
503: p = bsd_versions[0];
504: else {
505: s = n->string;
506: if (0 == strcmp(s, "3"))
507: p = bsd_versions[0];
508: else if (0 == strcmp(s, "4"))
509: p = bsd_versions[1];
510: else if (0 == strcmp(s, "5"))
511: p = bsd_versions[2];
512: else if (0 == strcmp(s, "6"))
513: p = bsd_versions[3];
514: else if (0 == strcmp(s, "7"))
515: p = bsd_versions[4];
516: else
517: p = bsd_versions[0];
518: }
519:
1.57 schwarze 520: free(man->meta.source);
521: man->meta.source = mandoc_strdup(p);
1.34 schwarze 522: return(1);
523: }
524:
525: static int
526: post_AT(CHKARGS)
527: {
528: static const char * const unix_versions[] = {
529: "7th Edition",
530: "System III",
531: "System V",
532: "System V Release 2",
533: };
534:
535: const char *p, *s;
536: struct man_node *nn;
537:
538: n = n->child;
539:
540: if (NULL == n || MAN_TEXT != n->type)
541: p = unix_versions[0];
542: else {
543: s = n->string;
544: if (0 == strcmp(s, "3"))
545: p = unix_versions[0];
546: else if (0 == strcmp(s, "4"))
547: p = unix_versions[1];
548: else if (0 == strcmp(s, "5")) {
549: nn = n->next;
550: if (nn && MAN_TEXT == nn->type && nn->string[0])
551: p = unix_versions[3];
552: else
553: p = unix_versions[2];
554: } else
555: p = unix_versions[0];
556: }
557:
1.57 schwarze 558: free(man->meta.source);
559: man->meta.source = mandoc_strdup(p);
1.46 schwarze 560: return(1);
561: }
562:
563: static int
564: post_vs(CHKARGS)
565: {
566:
1.54 schwarze 567: if (NULL != n->prev)
568: return(1);
569:
570: switch (n->parent->tok) {
571: case (MAN_SH):
572: /* FALLTHROUGH */
573: case (MAN_SS):
1.57 schwarze 574: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.54 schwarze 575: /* FALLTHROUGH */
576: case (MAN_MAX):
577: /*
578: * Don't warn about this because it occurs in pod2man
579: * and would cause considerable (unfixable) warnage.
580: */
1.57 schwarze 581: man_node_delete(man, n);
1.54 schwarze 582: break;
583: default:
584: break;
585: }
1.46 schwarze 586:
1.34 schwarze 587: return(1);
588: }