Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.57
1.57 ! schwarze 1: /* $Id: man_validate.c,v 1.56 2012/07/29 12:35:05 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.54 schwarze 4: * Copyright (c) 2010, 2012 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);
48: static int check_par(CHKARGS);
1.8 schwarze 49: static int check_part(CHKARGS);
1.7 schwarze 50: static int check_root(CHKARGS);
1.45 schwarze 51: static void check_text(CHKARGS);
1.7 schwarze 52:
1.34 schwarze 53: static int post_AT(CHKARGS);
1.55 schwarze 54: static int post_IP(CHKARGS);
1.46 schwarze 55: static int post_vs(CHKARGS);
1.34 schwarze 56: static int post_fi(CHKARGS);
1.47 schwarze 57: static int post_ft(CHKARGS);
1.34 schwarze 58: static int post_nf(CHKARGS);
1.47 schwarze 59: static int post_sec(CHKARGS);
1.34 schwarze 60: static int post_TH(CHKARGS);
61: static int post_UC(CHKARGS);
1.47 schwarze 62: static int pre_sec(CHKARGS);
1.34 schwarze 63:
64: static v_check posts_at[] = { post_AT, NULL };
1.46 schwarze 65: static v_check posts_br[] = { post_vs, check_eq0, NULL };
1.7 schwarze 66: static v_check posts_eq0[] = { check_eq0, NULL };
1.52 schwarze 67: static v_check posts_eq2[] = { check_eq2, NULL };
1.34 schwarze 68: static v_check posts_fi[] = { check_eq0, post_fi, NULL };
1.47 schwarze 69: static v_check posts_ft[] = { post_ft, NULL };
1.55 schwarze 70: static v_check posts_ip[] = { post_IP, NULL };
1.56 schwarze 71: static v_check posts_le1[] = { check_le1, NULL };
1.34 schwarze 72: static v_check posts_nf[] = { check_eq0, post_nf, NULL };
1.7 schwarze 73: static v_check posts_par[] = { check_par, NULL };
1.8 schwarze 74: static v_check posts_part[] = { check_part, NULL };
1.47 schwarze 75: static v_check posts_sec[] = { post_sec, NULL };
1.46 schwarze 76: static v_check posts_sp[] = { post_vs, check_le1, NULL };
1.40 schwarze 77: static v_check posts_th[] = { check_ge2, check_le5, post_TH, NULL };
1.34 schwarze 78: static v_check posts_uc[] = { post_UC, NULL };
1.50 schwarze 79: static v_check pres_sec[] = { pre_sec, NULL };
1.1 kristaps 80:
81: static const struct man_valid man_valids[MAN_MAX] = {
1.46 schwarze 82: { NULL, posts_br }, /* br */
1.50 schwarze 83: { NULL, posts_th }, /* TH */
1.47 schwarze 84: { pres_sec, posts_sec }, /* SH */
85: { pres_sec, posts_sec }, /* SS */
1.50 schwarze 86: { NULL, NULL }, /* TP */
87: { NULL, posts_par }, /* LP */
88: { NULL, posts_par }, /* PP */
89: { NULL, posts_par }, /* P */
1.55 schwarze 90: { NULL, posts_ip }, /* IP */
1.50 schwarze 91: { NULL, NULL }, /* HP */
1.8 schwarze 92: { NULL, NULL }, /* SM */
93: { NULL, NULL }, /* SB */
1.7 schwarze 94: { NULL, NULL }, /* BI */
95: { NULL, NULL }, /* IB */
96: { NULL, NULL }, /* BR */
97: { NULL, NULL }, /* RB */
1.8 schwarze 98: { NULL, NULL }, /* R */
99: { NULL, NULL }, /* B */
100: { NULL, NULL }, /* I */
1.7 schwarze 101: { NULL, NULL }, /* IR */
102: { NULL, NULL }, /* RI */
1.52 schwarze 103: { NULL, posts_eq0 }, /* na */
104: { NULL, posts_sp }, /* sp */
1.50 schwarze 105: { NULL, posts_nf }, /* nf */
106: { NULL, posts_fi }, /* fi */
1.8 schwarze 107: { NULL, NULL }, /* RE */
108: { NULL, posts_part }, /* RS */
109: { NULL, NULL }, /* DT */
1.34 schwarze 110: { NULL, posts_uc }, /* UC */
1.56 schwarze 111: { NULL, posts_le1 }, /* PD */
1.34 schwarze 112: { NULL, posts_at }, /* AT */
1.28 schwarze 113: { NULL, NULL }, /* in */
1.32 schwarze 114: { NULL, posts_ft }, /* ft */
1.52 schwarze 115: { NULL, posts_eq2 }, /* OP */
1.53 schwarze 116: { NULL, posts_nf }, /* EX */
117: { NULL, posts_fi }, /* EE */
1.1 kristaps 118: };
119:
120:
121: int
1.57 ! schwarze 122: man_valid_pre(struct man *man, struct man_node *n)
1.7 schwarze 123: {
124: v_check *cp;
125:
1.38 schwarze 126: switch (n->type) {
127: case (MAN_TEXT):
128: /* FALLTHROUGH */
129: case (MAN_ROOT):
130: /* FALLTHROUGH */
1.42 schwarze 131: case (MAN_EQN):
132: /* FALLTHROUGH */
1.38 schwarze 133: case (MAN_TBL):
1.7 schwarze 134: return(1);
1.38 schwarze 135: default:
136: break;
137: }
1.7 schwarze 138:
139: if (NULL == (cp = man_valids[n->tok].pres))
140: return(1);
141: for ( ; *cp; cp++)
1.57 ! schwarze 142: if ( ! (*cp)(man, n))
1.7 schwarze 143: return(0);
144: return(1);
145: }
146:
147:
148: int
1.57 ! schwarze 149: man_valid_post(struct man *man)
1.1 kristaps 150: {
1.7 schwarze 151: v_check *cp;
1.1 kristaps 152:
1.57 ! schwarze 153: if (MAN_VALID & man->last->flags)
1.1 kristaps 154: return(1);
1.57 ! schwarze 155: man->last->flags |= MAN_VALID;
1.1 kristaps 156:
1.57 ! schwarze 157: switch (man->last->type) {
1.1 kristaps 158: case (MAN_TEXT):
1.57 ! schwarze 159: check_text(man, man->last);
1.45 schwarze 160: return(1);
1.1 kristaps 161: case (MAN_ROOT):
1.57 ! schwarze 162: return(check_root(man, man->last));
1.42 schwarze 163: case (MAN_EQN):
164: /* FALLTHROUGH */
1.38 schwarze 165: case (MAN_TBL):
166: return(1);
1.1 kristaps 167: default:
168: break;
169: }
170:
1.57 ! schwarze 171: if (NULL == (cp = man_valids[man->last->tok].posts))
1.1 kristaps 172: return(1);
173: for ( ; *cp; cp++)
1.57 ! schwarze 174: if ( ! (*cp)(man, man->last))
1.1 kristaps 175: return(0);
176:
177: return(1);
178: }
179:
180:
1.4 schwarze 181: static int
1.7 schwarze 182: check_root(CHKARGS)
1.4 schwarze 183: {
1.7 schwarze 184:
1.57 ! schwarze 185: if (MAN_BLINE & man->flags)
! 186: man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
! 187: else if (MAN_ELINE & man->flags)
! 188: man_nmsg(man, n, MANDOCERR_SCOPEEXIT);
1.8 schwarze 189:
1.57 ! schwarze 190: man->flags &= ~MAN_BLINE;
! 191: man->flags &= ~MAN_ELINE;
1.7 schwarze 192:
1.57 ! schwarze 193: if (NULL == man->first->child) {
! 194: man_nmsg(man, n, MANDOCERR_NODOCBODY);
1.25 schwarze 195: return(0);
1.57 ! schwarze 196: } else if (NULL == man->meta.title) {
! 197: man_nmsg(man, n, MANDOCERR_NOTITLE);
1.35 schwarze 198:
1.18 schwarze 199: /*
200: * If a title hasn't been set, do so now (by
201: * implication, date and section also aren't set).
202: */
1.35 schwarze 203:
1.57 ! schwarze 204: man->meta.title = mandoc_strdup("unknown");
! 205: man->meta.msec = mandoc_strdup("1");
! 206: man->meta.date = mandoc_normdate
! 207: (man->parse, NULL, n->line, n->pos);
1.17 schwarze 208: }
1.15 schwarze 209:
210: return(1);
211: }
212:
1.45 schwarze 213: static void
1.47 schwarze 214: check_text(CHKARGS)
1.4 schwarze 215: {
1.47 schwarze 216: char *cp, *p;
217:
1.57 ! schwarze 218: if (MAN_LITERAL & man->flags)
1.48 schwarze 219: return;
220:
221: cp = n->string;
222: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.57 ! schwarze 223: man_pmsg(man, n->line, (int)(p - cp), MANDOCERR_BADTAB);
1.1 kristaps 224: }
225:
226: #define INEQ_DEFINE(x, ineq, name) \
227: static int \
1.7 schwarze 228: check_##name(CHKARGS) \
1.1 kristaps 229: { \
1.4 schwarze 230: if (n->nchild ineq (x)) \
1.1 kristaps 231: return(1); \
1.57 ! schwarze 232: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line, n->pos, \
1.25 schwarze 233: "line arguments %s %d (have %d)", \
234: #ineq, (x), n->nchild); \
1.40 schwarze 235: return(1); \
1.1 kristaps 236: }
237:
238: INEQ_DEFINE(0, ==, eq0)
1.52 schwarze 239: INEQ_DEFINE(2, ==, eq2)
1.10 schwarze 240: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 241: INEQ_DEFINE(2, >=, ge2)
242: INEQ_DEFINE(5, <=, le5)
1.32 schwarze 243:
244: static int
1.47 schwarze 245: post_ft(CHKARGS)
1.32 schwarze 246: {
247: char *cp;
248: int ok;
249:
250: if (0 == n->nchild)
251: return(1);
252:
253: ok = 0;
254: cp = n->child->string;
255: switch (*cp) {
256: case ('1'):
257: /* FALLTHROUGH */
258: case ('2'):
259: /* FALLTHROUGH */
260: case ('3'):
261: /* FALLTHROUGH */
262: case ('4'):
263: /* FALLTHROUGH */
264: case ('I'):
265: /* FALLTHROUGH */
266: case ('P'):
267: /* FALLTHROUGH */
268: case ('R'):
269: if ('\0' == cp[1])
270: ok = 1;
271: break;
272: case ('B'):
273: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
274: ok = 1;
275: break;
276: case ('C'):
277: if ('W' == cp[1] && '\0' == cp[2])
278: ok = 1;
279: break;
280: default:
281: break;
282: }
283:
284: if (0 == ok) {
1.44 schwarze 285: mandoc_vmsg
1.57 ! schwarze 286: (MANDOCERR_BADFONT, man->parse,
1.44 schwarze 287: n->line, n->pos, "%s", cp);
1.32 schwarze 288: *cp = '\0';
289: }
290:
291: if (1 < n->nchild)
1.44 schwarze 292: mandoc_vmsg
1.57 ! schwarze 293: (MANDOCERR_ARGCOUNT, man->parse, n->line,
1.44 schwarze 294: n->pos, "want one child (have %d)",
295: n->nchild);
1.32 schwarze 296:
297: return(1);
298: }
1.7 schwarze 299:
300: static int
1.47 schwarze 301: pre_sec(CHKARGS)
302: {
303:
304: if (MAN_BLOCK == n->type)
1.57 ! schwarze 305: man->flags &= ~MAN_LITERAL;
1.47 schwarze 306: return(1);
307: }
308:
309: static int
310: post_sec(CHKARGS)
1.7 schwarze 311: {
1.6 schwarze 312:
1.44 schwarze 313: if ( ! (MAN_HEAD == n->type && 0 == n->nchild))
314: return(1);
1.25 schwarze 315:
1.57 ! schwarze 316: man_nmsg(man, n, MANDOCERR_SYNTARGCOUNT);
1.44 schwarze 317: return(0);
1.6 schwarze 318: }
1.7 schwarze 319:
320: static int
1.8 schwarze 321: check_part(CHKARGS)
322: {
323:
324: if (MAN_BODY == n->type && 0 == n->nchild)
1.57 ! schwarze 325: mandoc_msg(MANDOCERR_ARGCWARN, man->parse, n->line,
1.44 schwarze 326: n->pos, "want children (have none)");
1.35 schwarze 327:
1.8 schwarze 328: return(1);
329: }
330:
1.36 schwarze 331:
332: static int
333: check_par(CHKARGS)
334: {
335:
1.39 schwarze 336: switch (n->type) {
337: case (MAN_BLOCK):
338: if (0 == n->body->nchild)
1.57 ! schwarze 339: man_node_delete(man, n);
1.39 schwarze 340: break;
341: case (MAN_BODY):
342: if (0 == n->nchild)
1.57 ! schwarze 343: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.39 schwarze 344: break;
345: case (MAN_HEAD):
346: if (n->nchild)
1.57 ! schwarze 347: man_nmsg(man, n, MANDOCERR_ARGSLOST);
1.39 schwarze 348: break;
349: default:
350: break;
351: }
1.36 schwarze 352:
353: return(1);
354: }
355:
1.55 schwarze 356: static int
357: post_IP(CHKARGS)
358: {
359:
360: switch (n->type) {
361: case (MAN_BLOCK):
362: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 ! schwarze 363: man_node_delete(man, n);
1.55 schwarze 364: break;
365: case (MAN_BODY):
366: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.57 ! schwarze 367: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.55 schwarze 368: break;
369: default:
370: break;
371: }
372: return(1);
373: }
1.36 schwarze 374:
1.34 schwarze 375: static int
376: post_TH(CHKARGS)
377: {
1.40 schwarze 378: const char *p;
1.41 schwarze 379: int line, pos;
1.34 schwarze 380:
1.57 ! schwarze 381: free(man->meta.title);
! 382: free(man->meta.vol);
! 383: free(man->meta.source);
! 384: free(man->meta.msec);
! 385: free(man->meta.date);
1.34 schwarze 386:
1.41 schwarze 387: line = n->line;
388: pos = n->pos;
1.57 ! schwarze 389: man->meta.title = man->meta.vol = man->meta.date =
! 390: man->meta.msec = man->meta.source = NULL;
1.34 schwarze 391:
392: /* ->TITLE<- MSEC DATE SOURCE VOL */
393:
394: n = n->child;
1.40 schwarze 395: if (n && n->string) {
396: for (p = n->string; '\0' != *p; p++) {
397: /* Only warn about this once... */
1.47 schwarze 398: if (isalpha((unsigned char)*p) &&
399: ! isupper((unsigned char)*p)) {
1.57 ! schwarze 400: man_nmsg(man, n, MANDOCERR_UPPERCASE);
1.40 schwarze 401: break;
402: }
403: }
1.57 ! schwarze 404: man->meta.title = mandoc_strdup(n->string);
1.40 schwarze 405: } else
1.57 ! schwarze 406: man->meta.title = mandoc_strdup("");
1.34 schwarze 407:
408: /* TITLE ->MSEC<- DATE SOURCE VOL */
409:
1.40 schwarze 410: if (n)
411: n = n->next;
412: if (n && n->string)
1.57 ! schwarze 413: man->meta.msec = mandoc_strdup(n->string);
1.40 schwarze 414: else
1.57 ! schwarze 415: man->meta.msec = mandoc_strdup("");
1.34 schwarze 416:
417: /* TITLE MSEC ->DATE<- SOURCE VOL */
418:
1.40 schwarze 419: if (n)
420: n = n->next;
1.49 schwarze 421: if (n && n->string && '\0' != n->string[0]) {
1.41 schwarze 422: pos = n->pos;
1.57 ! schwarze 423: man->meta.date = mandoc_normdate
! 424: (man->parse, n->string, line, pos);
1.49 schwarze 425: } else
1.57 ! schwarze 426: man->meta.date = mandoc_strdup("");
1.34 schwarze 427:
428: /* TITLE MSEC DATE ->SOURCE<- VOL */
429:
430: if (n && (n = n->next))
1.57 ! schwarze 431: man->meta.source = mandoc_strdup(n->string);
1.34 schwarze 432:
433: /* TITLE MSEC DATE SOURCE ->VOL<- */
1.51 schwarze 434: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 435:
436: if (n && (n = n->next))
1.57 ! schwarze 437: man->meta.vol = mandoc_strdup(n->string);
! 438: else if ('\0' != man->meta.msec[0] &&
! 439: (NULL != (p = mandoc_a2msec(man->meta.msec))))
! 440: man->meta.vol = mandoc_strdup(p);
1.34 schwarze 441:
442: /*
443: * Remove the `TH' node after we've processed it for our
444: * meta-data.
445: */
1.57 ! schwarze 446: man_node_delete(man, man->last);
1.34 schwarze 447: return(1);
448: }
449:
450: static int
451: post_nf(CHKARGS)
452: {
453:
1.57 ! schwarze 454: if (MAN_LITERAL & man->flags)
! 455: man_nmsg(man, n, MANDOCERR_SCOPEREP);
1.34 schwarze 456:
1.57 ! schwarze 457: man->flags |= MAN_LITERAL;
1.34 schwarze 458: return(1);
459: }
460:
461: static int
462: post_fi(CHKARGS)
463: {
464:
1.57 ! schwarze 465: if ( ! (MAN_LITERAL & man->flags))
! 466: man_nmsg(man, n, MANDOCERR_WNOSCOPE);
1.34 schwarze 467:
1.57 ! schwarze 468: man->flags &= ~MAN_LITERAL;
1.34 schwarze 469: return(1);
470: }
471:
472: static int
473: post_UC(CHKARGS)
474: {
475: static const char * const bsd_versions[] = {
476: "3rd Berkeley Distribution",
477: "4th Berkeley Distribution",
478: "4.2 Berkeley Distribution",
479: "4.3 Berkeley Distribution",
480: "4.4 Berkeley Distribution",
481: };
482:
483: const char *p, *s;
484:
485: n = n->child;
486:
487: if (NULL == n || MAN_TEXT != n->type)
488: p = bsd_versions[0];
489: else {
490: s = n->string;
491: if (0 == strcmp(s, "3"))
492: p = bsd_versions[0];
493: else if (0 == strcmp(s, "4"))
494: p = bsd_versions[1];
495: else if (0 == strcmp(s, "5"))
496: p = bsd_versions[2];
497: else if (0 == strcmp(s, "6"))
498: p = bsd_versions[3];
499: else if (0 == strcmp(s, "7"))
500: p = bsd_versions[4];
501: else
502: p = bsd_versions[0];
503: }
504:
1.57 ! schwarze 505: free(man->meta.source);
! 506: man->meta.source = mandoc_strdup(p);
1.34 schwarze 507: return(1);
508: }
509:
510: static int
511: post_AT(CHKARGS)
512: {
513: static const char * const unix_versions[] = {
514: "7th Edition",
515: "System III",
516: "System V",
517: "System V Release 2",
518: };
519:
520: const char *p, *s;
521: struct man_node *nn;
522:
523: n = n->child;
524:
525: if (NULL == n || MAN_TEXT != n->type)
526: p = unix_versions[0];
527: else {
528: s = n->string;
529: if (0 == strcmp(s, "3"))
530: p = unix_versions[0];
531: else if (0 == strcmp(s, "4"))
532: p = unix_versions[1];
533: else if (0 == strcmp(s, "5")) {
534: nn = n->next;
535: if (nn && MAN_TEXT == nn->type && nn->string[0])
536: p = unix_versions[3];
537: else
538: p = unix_versions[2];
539: } else
540: p = unix_versions[0];
541: }
542:
1.57 ! schwarze 543: free(man->meta.source);
! 544: man->meta.source = mandoc_strdup(p);
1.46 schwarze 545: return(1);
546: }
547:
548: static int
549: post_vs(CHKARGS)
550: {
551:
1.54 schwarze 552: if (NULL != n->prev)
553: return(1);
554:
555: switch (n->parent->tok) {
556: case (MAN_SH):
557: /* FALLTHROUGH */
558: case (MAN_SS):
1.57 ! schwarze 559: man_nmsg(man, n, MANDOCERR_IGNPAR);
1.54 schwarze 560: /* FALLTHROUGH */
561: case (MAN_MAX):
562: /*
563: * Don't warn about this because it occurs in pod2man
564: * and would cause considerable (unfixable) warnage.
565: */
1.57 ! schwarze 566: man_node_delete(man, n);
1.54 schwarze 567: break;
568: default:
569: break;
570: }
1.46 schwarze 571:
1.34 schwarze 572: return(1);
573: }