Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.58
1.58 ! schwarze 1: /* $Id: man_validate.c,v 1.57 2012/11/17 00:25:20 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.58 ! schwarze 4: * Copyright (c) 2010, 2012, 2013 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"
1.1 kristaps 31: #include "libman.h"
1.5 schwarze 32: #include "libmandoc.h"
1.1 kristaps 33:
1.57 schwarze 34: #define CHKARGS struct man *man, struct man_node *n
1.1 kristaps 35:
1.7 schwarze 36: typedef int (*v_check)(CHKARGS);
1.1 kristaps 37:
38: struct man_valid {
1.7 schwarze 39: v_check *pres;
40: v_check *posts;
1.1 kristaps 41: };
42:
1.7 schwarze 43: static int check_eq0(CHKARGS);
1.52 schwarze 44: static int check_eq2(CHKARGS);
1.10 schwarze 45: static int check_le1(CHKARGS);
1.7 schwarze 46: static int check_ge2(CHKARGS);
47: static int check_le5(CHKARGS);
1.58 ! schwarze 48: static int check_head1(CHKARGS);
1.7 schwarze 49: static int check_par(CHKARGS);
1.8 schwarze 50: static int check_part(CHKARGS);
1.7 schwarze 51: static int check_root(CHKARGS);
1.45 schwarze 52: static void check_text(CHKARGS);
1.7 schwarze 53:
1.34 schwarze 54: static int post_AT(CHKARGS);
1.55 schwarze 55: static int post_IP(CHKARGS);
1.46 schwarze 56: static int post_vs(CHKARGS);
1.34 schwarze 57: static int post_fi(CHKARGS);
1.47 schwarze 58: static int post_ft(CHKARGS);
1.34 schwarze 59: static int post_nf(CHKARGS);
1.47 schwarze 60: static int post_sec(CHKARGS);
1.34 schwarze 61: static int post_TH(CHKARGS);
62: static int post_UC(CHKARGS);
1.47 schwarze 63: static int pre_sec(CHKARGS);
1.34 schwarze 64:
65: static v_check posts_at[] = { post_AT, NULL };
1.46 schwarze 66: static v_check posts_br[] = { post_vs, check_eq0, NULL };
1.7 schwarze 67: static v_check posts_eq0[] = { check_eq0, NULL };
1.52 schwarze 68: static v_check posts_eq2[] = { check_eq2, NULL };
1.34 schwarze 69: static v_check posts_fi[] = { check_eq0, post_fi, NULL };
1.47 schwarze 70: static v_check posts_ft[] = { post_ft, NULL };
1.55 schwarze 71: static v_check posts_ip[] = { post_IP, NULL };
1.56 schwarze 72: static v_check posts_le1[] = { check_le1, NULL };
1.34 schwarze 73: static v_check posts_nf[] = { check_eq0, post_nf, NULL };
1.7 schwarze 74: static v_check posts_par[] = { check_par, NULL };
1.8 schwarze 75: static v_check posts_part[] = { check_part, NULL };
1.47 schwarze 76: static v_check posts_sec[] = { post_sec, NULL };
1.46 schwarze 77: static v_check posts_sp[] = { post_vs, check_le1, NULL };
1.40 schwarze 78: static v_check posts_th[] = { check_ge2, check_le5, post_TH, NULL };
1.34 schwarze 79: static v_check posts_uc[] = { post_UC, NULL };
1.58 ! schwarze 80: static v_check posts_ur[] = { check_head1, check_part, NULL };
1.50 schwarze 81: static v_check pres_sec[] = { pre_sec, NULL };
1.1 kristaps 82:
83: static const struct man_valid man_valids[MAN_MAX] = {
1.46 schwarze 84: { NULL, posts_br }, /* br */
1.50 schwarze 85: { NULL, posts_th }, /* TH */
1.47 schwarze 86: { pres_sec, posts_sec }, /* SH */
87: { pres_sec, posts_sec }, /* SS */
1.50 schwarze 88: { NULL, NULL }, /* TP */
89: { NULL, posts_par }, /* LP */
90: { NULL, posts_par }, /* PP */
91: { NULL, posts_par }, /* P */
1.55 schwarze 92: { NULL, posts_ip }, /* IP */
1.50 schwarze 93: { NULL, NULL }, /* HP */
1.8 schwarze 94: { NULL, NULL }, /* SM */
95: { NULL, NULL }, /* SB */
1.7 schwarze 96: { NULL, NULL }, /* BI */
97: { NULL, NULL }, /* IB */
98: { NULL, NULL }, /* BR */
99: { NULL, NULL }, /* RB */
1.8 schwarze 100: { NULL, NULL }, /* R */
101: { NULL, NULL }, /* B */
102: { NULL, NULL }, /* I */
1.7 schwarze 103: { NULL, NULL }, /* IR */
104: { NULL, NULL }, /* RI */
1.52 schwarze 105: { NULL, posts_eq0 }, /* na */
106: { NULL, posts_sp }, /* sp */
1.50 schwarze 107: { NULL, posts_nf }, /* nf */
108: { NULL, posts_fi }, /* fi */
1.8 schwarze 109: { NULL, NULL }, /* RE */
110: { NULL, posts_part }, /* RS */
111: { NULL, NULL }, /* DT */
1.34 schwarze 112: { NULL, posts_uc }, /* UC */
1.56 schwarze 113: { NULL, posts_le1 }, /* PD */
1.34 schwarze 114: { NULL, posts_at }, /* AT */
1.28 schwarze 115: { NULL, NULL }, /* in */
1.32 schwarze 116: { NULL, posts_ft }, /* ft */
1.52 schwarze 117: { NULL, posts_eq2 }, /* OP */
1.53 schwarze 118: { NULL, posts_nf }, /* EX */
119: { NULL, posts_fi }, /* EE */
1.58 ! schwarze 120: { NULL, posts_ur }, /* UR */
! 121: { NULL, NULL }, /* UE */
1.1 kristaps 122: };
123:
124:
125: int
1.57 schwarze 126: man_valid_pre(struct man *man, struct man_node *n)
1.7 schwarze 127: {
128: v_check *cp;
129:
1.38 schwarze 130: switch (n->type) {
131: case (MAN_TEXT):
132: /* FALLTHROUGH */
133: case (MAN_ROOT):
134: /* FALLTHROUGH */
1.42 schwarze 135: case (MAN_EQN):
136: /* FALLTHROUGH */
1.38 schwarze 137: case (MAN_TBL):
1.7 schwarze 138: return(1);
1.38 schwarze 139: default:
140: break;
141: }
1.7 schwarze 142:
143: if (NULL == (cp = man_valids[n->tok].pres))
144: return(1);
145: for ( ; *cp; cp++)
1.57 schwarze 146: if ( ! (*cp)(man, n))
1.7 schwarze 147: return(0);
148: return(1);
149: }
150:
151:
152: int
1.57 schwarze 153: man_valid_post(struct man *man)
1.1 kristaps 154: {
1.7 schwarze 155: v_check *cp;
1.1 kristaps 156:
1.57 schwarze 157: if (MAN_VALID & man->last->flags)
1.1 kristaps 158: return(1);
1.57 schwarze 159: man->last->flags |= MAN_VALID;
1.1 kristaps 160:
1.57 schwarze 161: switch (man->last->type) {
1.1 kristaps 162: case (MAN_TEXT):
1.57 schwarze 163: check_text(man, man->last);
1.45 schwarze 164: return(1);
1.1 kristaps 165: case (MAN_ROOT):
1.57 schwarze 166: return(check_root(man, man->last));
1.42 schwarze 167: case (MAN_EQN):
168: /* FALLTHROUGH */
1.38 schwarze 169: case (MAN_TBL):
170: return(1);
1.1 kristaps 171: default:
172: break;
173: }
174:
1.57 schwarze 175: if (NULL == (cp = man_valids[man->last->tok].posts))
1.1 kristaps 176: return(1);
177: for ( ; *cp; cp++)
1.57 schwarze 178: if ( ! (*cp)(man, man->last))
1.1 kristaps 179: return(0);
180:
181: return(1);
182: }
183:
184:
1.4 schwarze 185: static int
1.7 schwarze 186: check_root(CHKARGS)
1.4 schwarze 187: {
1.7 schwarze 188:
1.57 schwarze 189: if (MAN_BLINE & man->flags)
190: man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
191: else if (MAN_ELINE & man->flags)
192: man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
1.8 schwarze 193:
1.57 schwarze 194: man->flags &= ~MAN_BLINE;
195: man->flags &= ~MAN_ELINE;
1.7 schwarze 196:
1.57 schwarze 197: if (NULL == man->first->child) {
198: man_nmsg(man, n, MANDOCERR_NODOCBODY);
1.25 schwarze 199: return(0);
1.57 schwarze 200: } else if (NULL == man->meta.title) {
201: man_nmsg(man, n, MANDOCERR_NOTITLE);
1.35 schwarze 202:
1.18 schwarze 203: /*
204: * If a title hasn't been set, do so now (by
205: * implication, date and section also aren't set).
206: */
1.35 schwarze 207:
1.57 schwarze 208: man->meta.title = mandoc_strdup("unknown");
209: man->meta.msec = mandoc_strdup("1");
210: man->meta.date = mandoc_normdate
211: (man->parse, NULL, n->line, n->pos);
1.17 schwarze 212: }
1.15 schwarze 213:
214: return(1);
215: }
216:
1.45 schwarze 217: static void
1.47 schwarze 218: check_text(CHKARGS)
1.4 schwarze 219: {
1.47 schwarze 220: char *cp, *p;
221:
1.57 schwarze 222: if (MAN_LITERAL & man->flags)
1.48 schwarze 223: return;
224:
225: cp = n->string;
226: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.57 schwarze 227: man_pmsg(man, n->line, (int)(p - cp), MANDOCERR_BADTAB);
1.1 kristaps 228: }
229:
230: #define INEQ_DEFINE(x, ineq, name) \
231: static int \
1.7 schwarze 232: check_##name(CHKARGS) \
1.1 kristaps 233: { \
1.4 schwarze 234: if (n->nchild ineq (x)) \
1.1 kristaps 235: return(1); \
1.57 schwarze 236: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line, n->pos, \
1.25 schwarze 237: "line arguments %s %d (have %d)", \
238: #ineq, (x), n->nchild); \
1.40 schwarze 239: return(1); \
1.1 kristaps 240: }
241:
242: INEQ_DEFINE(0, ==, eq0)
1.52 schwarze 243: INEQ_DEFINE(2, ==, eq2)
1.10 schwarze 244: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 245: INEQ_DEFINE(2, >=, ge2)
246: INEQ_DEFINE(5, <=, le5)
1.58 ! schwarze 247:
! 248: static int
! 249: check_head1(CHKARGS)
! 250: {
! 251:
! 252: if (MAN_HEAD == n->type && 1 != n->nchild)
! 253: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
! 254: n->pos, "line arguments eq 1 (have %d)", n->nchild);
! 255:
! 256: return(1);
! 257: }
1.32 schwarze 258:
259: static int
1.47 schwarze 260: post_ft(CHKARGS)
1.32 schwarze 261: {
262: char *cp;
263: int ok;
264:
265: if (0 == n->nchild)
266: return(1);
267:
268: ok = 0;
269: cp = n->child->string;
270: switch (*cp) {
271: case ('1'):
272: /* FALLTHROUGH */
273: case ('2'):
274: /* FALLTHROUGH */
275: case ('3'):
276: /* FALLTHROUGH */
277: case ('4'):
278: /* FALLTHROUGH */
279: case ('I'):
280: /* FALLTHROUGH */
281: case ('P'):
282: /* FALLTHROUGH */
283: case ('R'):
284: if ('\0' == cp[1])
285: ok = 1;
286: break;
287: case ('B'):
288: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
289: ok = 1;
290: break;
291: case ('C'):
292: if ('W' == cp[1] && '\0' == cp[2])
293: ok = 1;
294: break;
295: default:
296: break;
297: }
298:
299: if (0 == ok) {
1.44 schwarze 300: mandoc_vmsg
1.57 schwarze 301: (MANDOCERR_BADFONT, man->parse,
1.44 schwarze 302: n->line, n->pos, "%s", cp);
1.32 schwarze 303: *cp = '\0';
304: }
305:
306: if (1 < n->nchild)
1.44 schwarze 307: mandoc_vmsg
1.57 schwarze 308: (MANDOCERR_ARGCOUNT, man->parse, n->line,
1.44 schwarze 309: n->pos, "want one child (have %d)",
310: n->nchild);
1.32 schwarze 311:
312: return(1);
313: }
1.7 schwarze 314:
315: static int
1.47 schwarze 316: pre_sec(CHKARGS)
317: {
318:
319: if (MAN_BLOCK == n->type)
1.57 schwarze 320: man->flags &= ~MAN_LITERAL;
1.47 schwarze 321: return(1);
322: }
323:
324: static int
325: post_sec(CHKARGS)
1.7 schwarze 326: {
1.6 schwarze 327:
1.44 schwarze 328: if ( ! (MAN_HEAD == n->type && 0 == n->nchild))
329: return(1);
1.25 schwarze 330:
1.57 schwarze 331: man_nmsg(man, n, MANDOCERR_SYNTARGCOUNT);
1.44 schwarze 332: return(0);
1.6 schwarze 333: }
1.7 schwarze 334:
335: static int
1.8 schwarze 336: check_part(CHKARGS)
337: {
338:
339: if (MAN_BODY == n->type && 0 == n->nchild)
1.57 schwarze 340: mandoc_msg(MANDOCERR_ARGCWARN, man->parse, n->line,
1.44 schwarze 341: n->pos, "want children (have none)");
1.35 schwarze 342:
1.8 schwarze 343: return(1);
344: }
345:
1.36 schwarze 346:
347: static int
348: check_par(CHKARGS)
349: {
350:
1.39 schwarze 351: switch (n->type) {
352: case (MAN_BLOCK):
353: if (0 == n->body->nchild)
1.57 schwarze 354: man_node_delete(man, n);
1.39 schwarze 355: break;
356: case (MAN_BODY):
357: if (0 == n->nchild)
1.57 schwarze 358: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.39 schwarze 359: break;
360: case (MAN_HEAD):
361: if (n->nchild)
1.57 schwarze 362: man_nmsg(man, n, MANDOCERR_ARGSLOST);
1.39 schwarze 363: break;
364: default:
365: break;
366: }
1.36 schwarze 367:
368: return(1);
369: }
370:
1.55 schwarze 371: static int
372: post_IP(CHKARGS)
373: {
374:
375: switch (n->type) {
376: case (MAN_BLOCK):
377: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 schwarze 378: man_node_delete(man, n);
1.55 schwarze 379: break;
380: case (MAN_BODY):
381: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.57 schwarze 382: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.55 schwarze 383: break;
384: default:
385: break;
386: }
387: return(1);
388: }
1.36 schwarze 389:
1.34 schwarze 390: static int
391: post_TH(CHKARGS)
392: {
1.40 schwarze 393: const char *p;
1.41 schwarze 394: int line, pos;
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:
1.41 schwarze 402: line = n->line;
403: pos = n->pos;
1.57 schwarze 404: man->meta.title = man->meta.vol = man->meta.date =
405: man->meta.msec = man->meta.source = NULL;
1.34 schwarze 406:
407: /* ->TITLE<- MSEC DATE SOURCE VOL */
408:
409: n = n->child;
1.40 schwarze 410: if (n && n->string) {
411: for (p = n->string; '\0' != *p; p++) {
412: /* Only warn about this once... */
1.47 schwarze 413: if (isalpha((unsigned char)*p) &&
414: ! isupper((unsigned char)*p)) {
1.57 schwarze 415: man_nmsg(man, n, MANDOCERR_UPPERCASE);
1.40 schwarze 416: break;
417: }
418: }
1.57 schwarze 419: man->meta.title = mandoc_strdup(n->string);
1.40 schwarze 420: } else
1.57 schwarze 421: man->meta.title = mandoc_strdup("");
1.34 schwarze 422:
423: /* TITLE ->MSEC<- DATE SOURCE VOL */
424:
1.40 schwarze 425: if (n)
426: n = n->next;
427: if (n && n->string)
1.57 schwarze 428: man->meta.msec = mandoc_strdup(n->string);
1.40 schwarze 429: else
1.57 schwarze 430: man->meta.msec = mandoc_strdup("");
1.34 schwarze 431:
432: /* TITLE MSEC ->DATE<- SOURCE VOL */
433:
1.40 schwarze 434: if (n)
435: n = n->next;
1.49 schwarze 436: if (n && n->string && '\0' != n->string[0]) {
1.41 schwarze 437: pos = n->pos;
1.57 schwarze 438: man->meta.date = mandoc_normdate
439: (man->parse, n->string, line, 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: }