Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.61
1.61 ! schwarze 1: /* $Id: man_validate.c,v 1.60 2014/01/06 22:39:19 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 */
1.1 kristaps 123: };
124:
125:
126: int
1.57 schwarze 127: man_valid_pre(struct man *man, struct man_node *n)
1.7 schwarze 128: {
129: v_check *cp;
130:
1.38 schwarze 131: switch (n->type) {
132: case (MAN_TEXT):
133: /* FALLTHROUGH */
134: case (MAN_ROOT):
135: /* FALLTHROUGH */
1.42 schwarze 136: case (MAN_EQN):
137: /* FALLTHROUGH */
1.38 schwarze 138: case (MAN_TBL):
1.7 schwarze 139: return(1);
1.38 schwarze 140: default:
141: break;
142: }
1.7 schwarze 143:
144: if (NULL == (cp = man_valids[n->tok].pres))
145: return(1);
146: for ( ; *cp; cp++)
1.57 schwarze 147: if ( ! (*cp)(man, n))
1.7 schwarze 148: return(0);
149: return(1);
150: }
151:
152:
153: int
1.57 schwarze 154: man_valid_post(struct man *man)
1.1 kristaps 155: {
1.7 schwarze 156: v_check *cp;
1.1 kristaps 157:
1.57 schwarze 158: if (MAN_VALID & man->last->flags)
1.1 kristaps 159: return(1);
1.57 schwarze 160: man->last->flags |= MAN_VALID;
1.1 kristaps 161:
1.57 schwarze 162: switch (man->last->type) {
1.1 kristaps 163: case (MAN_TEXT):
1.57 schwarze 164: check_text(man, man->last);
1.45 schwarze 165: return(1);
1.1 kristaps 166: case (MAN_ROOT):
1.57 schwarze 167: return(check_root(man, man->last));
1.42 schwarze 168: case (MAN_EQN):
169: /* FALLTHROUGH */
1.38 schwarze 170: case (MAN_TBL):
171: return(1);
1.1 kristaps 172: default:
173: break;
174: }
175:
1.57 schwarze 176: if (NULL == (cp = man_valids[man->last->tok].posts))
1.1 kristaps 177: return(1);
178: for ( ; *cp; cp++)
1.57 schwarze 179: if ( ! (*cp)(man, man->last))
1.1 kristaps 180: return(0);
181:
182: return(1);
183: }
184:
185:
1.4 schwarze 186: static int
1.7 schwarze 187: check_root(CHKARGS)
1.4 schwarze 188: {
1.7 schwarze 189:
1.57 schwarze 190: if (MAN_BLINE & man->flags)
191: man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
192: else if (MAN_ELINE & man->flags)
193: man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
1.8 schwarze 194:
1.57 schwarze 195: man->flags &= ~MAN_BLINE;
196: man->flags &= ~MAN_ELINE;
1.7 schwarze 197:
1.57 schwarze 198: if (NULL == man->first->child) {
199: man_nmsg(man, n, MANDOCERR_NODOCBODY);
1.25 schwarze 200: return(0);
1.57 schwarze 201: } else if (NULL == man->meta.title) {
202: man_nmsg(man, n, MANDOCERR_NOTITLE);
1.35 schwarze 203:
1.18 schwarze 204: /*
205: * If a title hasn't been set, do so now (by
206: * implication, date and section also aren't set).
207: */
1.35 schwarze 208:
1.57 schwarze 209: man->meta.title = mandoc_strdup("unknown");
210: man->meta.msec = mandoc_strdup("1");
1.60 schwarze 211: man->meta.date = man->quick ? mandoc_strdup("") :
212: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 213: }
1.15 schwarze 214:
215: return(1);
216: }
217:
1.45 schwarze 218: static void
1.47 schwarze 219: check_text(CHKARGS)
1.4 schwarze 220: {
1.47 schwarze 221: char *cp, *p;
222:
1.57 schwarze 223: if (MAN_LITERAL & man->flags)
1.48 schwarze 224: return;
225:
226: cp = n->string;
227: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.57 schwarze 228: man_pmsg(man, n->line, (int)(p - cp), MANDOCERR_BADTAB);
1.1 kristaps 229: }
230:
231: #define INEQ_DEFINE(x, ineq, name) \
232: static int \
1.7 schwarze 233: check_##name(CHKARGS) \
1.1 kristaps 234: { \
1.4 schwarze 235: if (n->nchild ineq (x)) \
1.1 kristaps 236: return(1); \
1.57 schwarze 237: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line, n->pos, \
1.25 schwarze 238: "line arguments %s %d (have %d)", \
239: #ineq, (x), n->nchild); \
1.40 schwarze 240: return(1); \
1.1 kristaps 241: }
242:
243: INEQ_DEFINE(0, ==, eq0)
1.52 schwarze 244: INEQ_DEFINE(2, ==, eq2)
1.10 schwarze 245: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 246: INEQ_DEFINE(2, >=, ge2)
247: INEQ_DEFINE(5, <=, le5)
1.58 schwarze 248:
249: static int
250: check_head1(CHKARGS)
251: {
252:
253: if (MAN_HEAD == n->type && 1 != n->nchild)
254: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
255: n->pos, "line arguments eq 1 (have %d)", n->nchild);
256:
257: return(1);
258: }
1.32 schwarze 259:
260: static int
1.47 schwarze 261: post_ft(CHKARGS)
1.32 schwarze 262: {
263: char *cp;
264: int ok;
265:
266: if (0 == n->nchild)
267: return(1);
268:
269: ok = 0;
270: cp = n->child->string;
271: switch (*cp) {
272: case ('1'):
273: /* FALLTHROUGH */
274: case ('2'):
275: /* FALLTHROUGH */
276: case ('3'):
277: /* FALLTHROUGH */
278: case ('4'):
279: /* FALLTHROUGH */
280: case ('I'):
281: /* FALLTHROUGH */
282: case ('P'):
283: /* FALLTHROUGH */
284: case ('R'):
285: if ('\0' == cp[1])
286: ok = 1;
287: break;
288: case ('B'):
289: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
290: ok = 1;
291: break;
292: case ('C'):
293: if ('W' == cp[1] && '\0' == cp[2])
294: ok = 1;
295: break;
296: default:
297: break;
298: }
299:
300: if (0 == ok) {
1.44 schwarze 301: mandoc_vmsg
1.57 schwarze 302: (MANDOCERR_BADFONT, man->parse,
1.44 schwarze 303: n->line, n->pos, "%s", cp);
1.32 schwarze 304: *cp = '\0';
305: }
306:
307: if (1 < n->nchild)
1.44 schwarze 308: mandoc_vmsg
1.57 schwarze 309: (MANDOCERR_ARGCOUNT, man->parse, n->line,
1.44 schwarze 310: n->pos, "want one child (have %d)",
311: n->nchild);
1.32 schwarze 312:
313: return(1);
314: }
1.7 schwarze 315:
316: static int
1.47 schwarze 317: pre_sec(CHKARGS)
318: {
319:
320: if (MAN_BLOCK == n->type)
1.57 schwarze 321: man->flags &= ~MAN_LITERAL;
1.47 schwarze 322: return(1);
323: }
324:
325: static int
326: post_sec(CHKARGS)
1.7 schwarze 327: {
1.6 schwarze 328:
1.44 schwarze 329: if ( ! (MAN_HEAD == n->type && 0 == n->nchild))
330: return(1);
1.25 schwarze 331:
1.57 schwarze 332: man_nmsg(man, n, MANDOCERR_SYNTARGCOUNT);
1.44 schwarze 333: return(0);
1.6 schwarze 334: }
1.7 schwarze 335:
336: static int
1.8 schwarze 337: check_part(CHKARGS)
338: {
339:
340: if (MAN_BODY == n->type && 0 == n->nchild)
1.57 schwarze 341: mandoc_msg(MANDOCERR_ARGCWARN, man->parse, n->line,
1.44 schwarze 342: n->pos, "want children (have none)");
1.35 schwarze 343:
1.8 schwarze 344: return(1);
345: }
346:
1.36 schwarze 347:
348: static int
349: check_par(CHKARGS)
350: {
351:
1.39 schwarze 352: switch (n->type) {
353: case (MAN_BLOCK):
354: if (0 == n->body->nchild)
1.57 schwarze 355: man_node_delete(man, n);
1.39 schwarze 356: break;
357: case (MAN_BODY):
358: if (0 == n->nchild)
1.57 schwarze 359: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.39 schwarze 360: break;
361: case (MAN_HEAD):
362: if (n->nchild)
1.57 schwarze 363: man_nmsg(man, n, MANDOCERR_ARGSLOST);
1.39 schwarze 364: break;
365: default:
366: break;
367: }
1.36 schwarze 368:
369: return(1);
370: }
371:
1.55 schwarze 372: static int
373: post_IP(CHKARGS)
374: {
375:
376: switch (n->type) {
377: case (MAN_BLOCK):
378: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 schwarze 379: man_node_delete(man, n);
1.55 schwarze 380: break;
381: case (MAN_BODY):
382: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.57 schwarze 383: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.55 schwarze 384: break;
385: default:
386: break;
387: }
388: return(1);
389: }
1.36 schwarze 390:
1.34 schwarze 391: static int
392: post_TH(CHKARGS)
393: {
1.40 schwarze 394: const char *p;
1.34 schwarze 395:
1.57 schwarze 396: free(man->meta.title);
397: free(man->meta.vol);
398: free(man->meta.source);
399: free(man->meta.msec);
400: free(man->meta.date);
1.34 schwarze 401:
1.57 schwarze 402: man->meta.title = man->meta.vol = man->meta.date =
403: man->meta.msec = man->meta.source = NULL;
1.34 schwarze 404:
405: /* ->TITLE<- MSEC DATE SOURCE VOL */
406:
407: n = n->child;
1.40 schwarze 408: if (n && n->string) {
409: for (p = n->string; '\0' != *p; p++) {
410: /* Only warn about this once... */
1.47 schwarze 411: if (isalpha((unsigned char)*p) &&
412: ! isupper((unsigned char)*p)) {
1.57 schwarze 413: man_nmsg(man, n, MANDOCERR_UPPERCASE);
1.40 schwarze 414: break;
415: }
416: }
1.57 schwarze 417: man->meta.title = mandoc_strdup(n->string);
1.40 schwarze 418: } else
1.57 schwarze 419: man->meta.title = mandoc_strdup("");
1.34 schwarze 420:
421: /* TITLE ->MSEC<- DATE SOURCE VOL */
422:
1.40 schwarze 423: if (n)
424: n = n->next;
425: if (n && n->string)
1.57 schwarze 426: man->meta.msec = mandoc_strdup(n->string);
1.40 schwarze 427: else
1.57 schwarze 428: man->meta.msec = mandoc_strdup("");
1.34 schwarze 429:
430: /* TITLE MSEC ->DATE<- SOURCE VOL */
431:
1.40 schwarze 432: if (n)
433: n = n->next;
1.49 schwarze 434: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 435: man->meta.date = man->quick ?
436: mandoc_strdup(n->string) :
437: mandoc_normdate(man->parse, n->string,
438: n->line, n->pos);
1.49 schwarze 439: } else
1.57 schwarze 440: man->meta.date = mandoc_strdup("");
1.34 schwarze 441:
442: /* TITLE MSEC DATE ->SOURCE<- VOL */
443:
444: if (n && (n = n->next))
1.57 schwarze 445: man->meta.source = mandoc_strdup(n->string);
1.34 schwarze 446:
447: /* TITLE MSEC DATE SOURCE ->VOL<- */
1.51 schwarze 448: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 449:
450: if (n && (n = n->next))
1.57 schwarze 451: man->meta.vol = mandoc_strdup(n->string);
452: else if ('\0' != man->meta.msec[0] &&
453: (NULL != (p = mandoc_a2msec(man->meta.msec))))
454: man->meta.vol = mandoc_strdup(p);
1.34 schwarze 455:
456: /*
457: * Remove the `TH' node after we've processed it for our
458: * meta-data.
459: */
1.57 schwarze 460: man_node_delete(man, man->last);
1.34 schwarze 461: return(1);
462: }
463:
464: static int
465: post_nf(CHKARGS)
466: {
467:
1.57 schwarze 468: if (MAN_LITERAL & man->flags)
469: man_nmsg(man, n, MANDOCERR_SCOPEREP);
1.34 schwarze 470:
1.57 schwarze 471: man->flags |= MAN_LITERAL;
1.34 schwarze 472: return(1);
473: }
474:
475: static int
476: post_fi(CHKARGS)
477: {
478:
1.57 schwarze 479: if ( ! (MAN_LITERAL & man->flags))
480: man_nmsg(man, n, MANDOCERR_WNOSCOPE);
1.34 schwarze 481:
1.57 schwarze 482: man->flags &= ~MAN_LITERAL;
1.34 schwarze 483: return(1);
484: }
485:
486: static int
487: post_UC(CHKARGS)
488: {
489: static const char * const bsd_versions[] = {
490: "3rd Berkeley Distribution",
491: "4th Berkeley Distribution",
492: "4.2 Berkeley Distribution",
493: "4.3 Berkeley Distribution",
494: "4.4 Berkeley Distribution",
495: };
496:
497: const char *p, *s;
498:
499: n = n->child;
500:
501: if (NULL == n || MAN_TEXT != n->type)
502: p = bsd_versions[0];
503: else {
504: s = n->string;
505: if (0 == strcmp(s, "3"))
506: p = bsd_versions[0];
507: else if (0 == strcmp(s, "4"))
508: p = bsd_versions[1];
509: else if (0 == strcmp(s, "5"))
510: p = bsd_versions[2];
511: else if (0 == strcmp(s, "6"))
512: p = bsd_versions[3];
513: else if (0 == strcmp(s, "7"))
514: p = bsd_versions[4];
515: else
516: p = bsd_versions[0];
517: }
518:
1.57 schwarze 519: free(man->meta.source);
520: man->meta.source = mandoc_strdup(p);
1.34 schwarze 521: return(1);
522: }
523:
524: static int
525: post_AT(CHKARGS)
526: {
527: static const char * const unix_versions[] = {
528: "7th Edition",
529: "System III",
530: "System V",
531: "System V Release 2",
532: };
533:
534: const char *p, *s;
535: struct man_node *nn;
536:
537: n = n->child;
538:
539: if (NULL == n || MAN_TEXT != n->type)
540: p = unix_versions[0];
541: else {
542: s = n->string;
543: if (0 == strcmp(s, "3"))
544: p = unix_versions[0];
545: else if (0 == strcmp(s, "4"))
546: p = unix_versions[1];
547: else if (0 == strcmp(s, "5")) {
548: nn = n->next;
549: if (nn && MAN_TEXT == nn->type && nn->string[0])
550: p = unix_versions[3];
551: else
552: p = unix_versions[2];
553: } else
554: p = unix_versions[0];
555: }
556:
1.57 schwarze 557: free(man->meta.source);
558: man->meta.source = mandoc_strdup(p);
1.46 schwarze 559: return(1);
560: }
561:
562: static int
563: post_vs(CHKARGS)
564: {
565:
1.54 schwarze 566: if (NULL != n->prev)
567: return(1);
568:
569: switch (n->parent->tok) {
570: case (MAN_SH):
571: /* FALLTHROUGH */
572: case (MAN_SS):
1.57 schwarze 573: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.54 schwarze 574: /* FALLTHROUGH */
575: case (MAN_MAX):
576: /*
577: * Don't warn about this because it occurs in pod2man
578: * and would cause considerable (unfixable) warnage.
579: */
1.57 schwarze 580: man_node_delete(man, n);
1.54 schwarze 581: break;
582: default:
583: break;
584: }
1.46 schwarze 585:
1.34 schwarze 586: return(1);
587: }