Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.110
1.110 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.109 2018/12/03 21:00:06 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.105 schwarze 4: * Copyright (c) 2010, 2012-2018 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: *
1.85 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.2 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.85 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.2 schwarze 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:
1.85 schwarze 29: #include "mandoc_aux.h"
30: #include "mandoc.h"
31: #include "roff.h"
1.44 schwarze 32: #include "man.h"
1.85 schwarze 33: #include "libmandoc.h"
1.89 schwarze 34: #include "roff_int.h"
1.1 kristaps 35: #include "libman.h"
36:
1.88 schwarze 37: #define CHKARGS struct roff_man *man, struct roff_node *n
1.1 kristaps 38:
1.78 schwarze 39: typedef void (*v_check)(CHKARGS);
1.1 kristaps 40:
1.109 schwarze 41: static void check_abort(CHKARGS);
1.78 schwarze 42: static void check_par(CHKARGS);
43: static void check_part(CHKARGS);
44: static void check_root(CHKARGS);
45: static void check_text(CHKARGS);
46:
47: static void post_AT(CHKARGS);
48: static void post_IP(CHKARGS);
1.84 schwarze 49: static void post_OP(CHKARGS);
1.110 ! schwarze 50: static void post_SH(CHKARGS);
1.78 schwarze 51: static void post_TH(CHKARGS);
52: static void post_UC(CHKARGS);
53: static void post_UR(CHKARGS);
1.100 schwarze 54: static void post_in(CHKARGS);
1.34 schwarze 55:
1.106 schwarze 56: static const v_check man_valids[MAN_MAX - MAN_TH] = {
1.76 schwarze 57: post_TH, /* TH */
1.110 ! schwarze 58: post_SH, /* SH */
! 59: post_SH, /* SS */
1.76 schwarze 60: NULL, /* TP */
1.107 schwarze 61: NULL, /* TQ */
1.109 schwarze 62: check_abort,/* LP */
1.76 schwarze 63: check_par, /* PP */
1.109 schwarze 64: check_abort,/* P */
1.76 schwarze 65: post_IP, /* IP */
66: NULL, /* HP */
67: NULL, /* SM */
68: NULL, /* SB */
69: NULL, /* BI */
70: NULL, /* IB */
71: NULL, /* BR */
72: NULL, /* RB */
73: NULL, /* R */
74: NULL, /* B */
75: NULL, /* I */
76: NULL, /* IR */
77: NULL, /* RI */
1.92 schwarze 78: NULL, /* nf */
79: NULL, /* fi */
1.76 schwarze 80: NULL, /* RE */
81: check_part, /* RS */
82: NULL, /* DT */
83: post_UC, /* UC */
1.83 schwarze 84: NULL, /* PD */
1.76 schwarze 85: post_AT, /* AT */
1.100 schwarze 86: post_in, /* in */
1.108 schwarze 87: NULL, /* SY */
88: NULL, /* YS */
1.84 schwarze 89: post_OP, /* OP */
1.92 schwarze 90: NULL, /* EX */
91: NULL, /* EE */
1.76 schwarze 92: post_UR, /* UR */
93: NULL, /* UE */
1.103 bentley 94: post_UR, /* MT */
95: NULL, /* ME */
1.1 kristaps 96: };
97:
98:
1.109 schwarze 99: /* Validate the subtree rooted at man->last. */
1.78 schwarze 100: void
1.92 schwarze 101: man_node_validate(struct roff_man *man)
1.1 kristaps 102: {
1.86 schwarze 103: struct roff_node *n;
1.94 schwarze 104: const v_check *cp;
1.1 kristaps 105:
1.109 schwarze 106: /*
107: * Translate obsolete macros such that later code
108: * does not need to look for them.
109: */
110:
1.76 schwarze 111: n = man->last;
1.109 schwarze 112: switch (n->tok) {
113: case MAN_LP:
114: case MAN_P:
115: n->tok = MAN_PP;
116: break;
117: default:
118: break;
119: }
120:
121: /*
122: * Iterate over all children, recursing into each one
123: * in turn, depth-first.
124: */
125:
1.92 schwarze 126: man->last = man->last->child;
127: while (man->last != NULL) {
128: man_node_validate(man);
129: if (man->last == n)
130: man->last = man->last->child;
131: else
132: man->last = man->last->next;
133: }
1.1 kristaps 134:
1.109 schwarze 135: /* Finally validate the macro itself. */
136:
1.92 schwarze 137: man->last = n;
138: man->next = ROFF_NEXT_SIBLING;
1.76 schwarze 139: switch (n->type) {
1.85 schwarze 140: case ROFFT_TEXT:
1.78 schwarze 141: check_text(man, n);
142: break;
1.85 schwarze 143: case ROFFT_ROOT:
1.78 schwarze 144: check_root(man, n);
145: break;
1.105 schwarze 146: case ROFFT_COMMENT:
1.85 schwarze 147: case ROFFT_EQN:
148: case ROFFT_TBL:
1.78 schwarze 149: break;
1.1 kristaps 150: default:
1.95 schwarze 151: if (n->tok < ROFF_MAX) {
1.110 ! schwarze 152: roff_validate(man);
1.95 schwarze 153: break;
154: }
155: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.106 schwarze 156: cp = man_valids + (n->tok - MAN_TH);
1.78 schwarze 157: if (*cp)
158: (*cp)(man, n);
1.92 schwarze 159: if (man->last == n)
160: man_state(man, n);
1.78 schwarze 161: break;
1.1 kristaps 162: }
163: }
164:
1.78 schwarze 165: static void
1.63 schwarze 166: check_root(CHKARGS)
1.4 schwarze 167: {
1.73 schwarze 168: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 169:
1.105 schwarze 170: if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1.75 schwarze 171: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
172: n->line, n->pos, NULL);
1.65 schwarze 173: else
174: man->meta.hasbody = 1;
175:
176: if (NULL == man->meta.title) {
1.77 schwarze 177: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 178: n->line, n->pos, NULL);
1.35 schwarze 179:
1.18 schwarze 180: /*
181: * If a title hasn't been set, do so now (by
182: * implication, date and section also aren't set).
183: */
1.35 schwarze 184:
1.77 schwarze 185: man->meta.title = mandoc_strdup("");
186: man->meta.msec = mandoc_strdup("");
1.60 schwarze 187: man->meta.date = man->quick ? mandoc_strdup("") :
1.99 schwarze 188: mandoc_normdate(man, NULL, n->line, n->pos);
1.17 schwarze 189: }
1.101 schwarze 190:
191: if (man->meta.os_e &&
192: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.102 schwarze 193: mandoc_msg(MANDOCERR_RCS_MISSING, man->parse, 0, 0,
194: man->meta.os_e == MANDOC_OS_OPENBSD ?
195: "(OpenBSD)" : "(NetBSD)");
1.15 schwarze 196: }
197:
1.78 schwarze 198: static void
1.109 schwarze 199: check_abort(CHKARGS)
200: {
201: abort();
202: }
203:
204: static void
1.47 schwarze 205: check_text(CHKARGS)
1.4 schwarze 206: {
1.47 schwarze 207: char *cp, *p;
208:
1.57 schwarze 209: if (MAN_LITERAL & man->flags)
1.78 schwarze 210: return;
1.48 schwarze 211:
212: cp = n->string;
213: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 214: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
215: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 216: }
217:
1.84 schwarze 218: static void
219: post_OP(CHKARGS)
220: {
221:
1.93 schwarze 222: if (n->child == NULL)
1.84 schwarze 223: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
224: n->line, n->pos, "OP");
1.93 schwarze 225: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 226: n = n->child->next->next;
227: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
228: n->line, n->pos, "OP ... %s", n->string);
229: }
1.1 kristaps 230: }
231:
1.78 schwarze 232: static void
1.110 ! schwarze 233: post_SH(CHKARGS)
! 234: {
! 235: struct roff_node *nc;
! 236:
! 237: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
! 238: return;
! 239:
! 240: if (nc->tok == MAN_PP && nc->body->child != NULL) {
! 241: while (nc->body->last != NULL) {
! 242: man->next = ROFF_NEXT_CHILD;
! 243: roff_node_relink(man, nc->body->last);
! 244: man->last = n;
! 245: }
! 246: }
! 247:
! 248: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
! 249: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse,
! 250: nc->line, nc->pos, "%s after %s",
! 251: roff_name[nc->tok], roff_name[n->tok]);
! 252: roff_node_delete(man, nc);
! 253: }
! 254:
! 255: /*
! 256: * Trailing PP is empty, so it is deleted by check_par().
! 257: * Trailing sp is significant.
! 258: */
! 259:
! 260: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
! 261: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse,
! 262: nc->line, nc->pos, "%s at the end of %s",
! 263: roff_name[nc->tok], roff_name[n->tok]);
! 264: roff_node_delete(man, nc);
! 265: }
! 266: }
! 267:
! 268: static void
1.76 schwarze 269: post_UR(CHKARGS)
1.58 schwarze 270: {
1.85 schwarze 271: if (n->type == ROFFT_HEAD && n->child == NULL)
1.104 schwarze 272: mandoc_msg(MANDOCERR_UR_NOHEAD, man->parse,
1.103 bentley 273: n->line, n->pos, roff_name[n->tok]);
1.78 schwarze 274: check_part(man, n);
1.32 schwarze 275: }
1.7 schwarze 276:
1.78 schwarze 277: static void
1.8 schwarze 278: check_part(CHKARGS)
279: {
280:
1.85 schwarze 281: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 282: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.94 schwarze 283: n->line, n->pos, roff_name[n->tok]);
1.8 schwarze 284: }
285:
1.78 schwarze 286: static void
1.36 schwarze 287: check_par(CHKARGS)
288: {
289:
1.39 schwarze 290: switch (n->type) {
1.85 schwarze 291: case ROFFT_BLOCK:
1.93 schwarze 292: if (n->body->child == NULL)
1.89 schwarze 293: roff_node_delete(man, n);
1.39 schwarze 294: break;
1.85 schwarze 295: case ROFFT_BODY:
1.110 ! schwarze 296: if (n->child != NULL &&
! 297: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
! 298: mandoc_vmsg(MANDOCERR_PAR_SKIP,
! 299: man->parse, n->child->line, n->child->pos,
! 300: "%s after %s", roff_name[n->child->tok],
! 301: roff_name[n->tok]);
! 302: roff_node_delete(man, n->child);
! 303: }
1.93 schwarze 304: if (n->child == NULL)
1.67 schwarze 305: mandoc_vmsg(MANDOCERR_PAR_SKIP,
306: man->parse, n->line, n->pos,
1.94 schwarze 307: "%s empty", roff_name[n->tok]);
1.39 schwarze 308: break;
1.85 schwarze 309: case ROFFT_HEAD:
1.93 schwarze 310: if (n->child != NULL)
1.69 schwarze 311: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.94 schwarze 312: man->parse, n->line, n->pos, "%s %s%s",
313: roff_name[n->tok], n->child->string,
1.93 schwarze 314: n->child->next != NULL ? " ..." : "");
1.39 schwarze 315: break;
316: default:
317: break;
318: }
1.36 schwarze 319: }
320:
1.78 schwarze 321: static void
1.55 schwarze 322: post_IP(CHKARGS)
323: {
324:
325: switch (n->type) {
1.85 schwarze 326: case ROFFT_BLOCK:
1.93 schwarze 327: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 328: roff_node_delete(man, n);
1.55 schwarze 329: break;
1.85 schwarze 330: case ROFFT_BODY:
1.93 schwarze 331: if (n->parent->head->child == NULL && n->child == NULL)
1.67 schwarze 332: mandoc_vmsg(MANDOCERR_PAR_SKIP,
333: man->parse, n->line, n->pos,
1.94 schwarze 334: "%s empty", roff_name[n->tok]);
1.55 schwarze 335: break;
336: default:
337: break;
338: }
339: }
1.36 schwarze 340:
1.78 schwarze 341: static void
1.34 schwarze 342: post_TH(CHKARGS)
343: {
1.86 schwarze 344: struct roff_node *nb;
1.40 schwarze 345: const char *p;
1.34 schwarze 346:
1.57 schwarze 347: free(man->meta.title);
348: free(man->meta.vol);
1.87 schwarze 349: free(man->meta.os);
1.57 schwarze 350: free(man->meta.msec);
351: free(man->meta.date);
1.34 schwarze 352:
1.57 schwarze 353: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 354: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 355:
1.64 schwarze 356: nb = n;
357:
1.87 schwarze 358: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 359:
360: n = n->child;
1.40 schwarze 361: if (n && n->string) {
362: for (p = n->string; '\0' != *p; p++) {
363: /* Only warn about this once... */
1.63 schwarze 364: if (isalpha((unsigned char)*p) &&
365: ! isupper((unsigned char)*p)) {
1.74 schwarze 366: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 367: man->parse, n->line,
368: n->pos + (p - n->string),
1.74 schwarze 369: "TH %s", n->string);
1.40 schwarze 370: break;
371: }
372: }
1.57 schwarze 373: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 374: } else {
1.57 schwarze 375: man->meta.title = mandoc_strdup("");
1.77 schwarze 376: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
377: nb->line, nb->pos, "TH");
378: }
1.34 schwarze 379:
1.87 schwarze 380: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 381:
1.40 schwarze 382: if (n)
383: n = n->next;
384: if (n && n->string)
1.57 schwarze 385: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 386: else {
1.57 schwarze 387: man->meta.msec = mandoc_strdup("");
1.77 schwarze 388: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
389: nb->line, nb->pos, "TH %s", man->meta.title);
390: }
1.34 schwarze 391:
1.87 schwarze 392: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 393:
1.40 schwarze 394: if (n)
395: n = n->next;
1.49 schwarze 396: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 397: man->meta.date = man->quick ?
398: mandoc_strdup(n->string) :
1.99 schwarze 399: mandoc_normdate(man, n->string, n->line, n->pos);
1.64 schwarze 400: } else {
1.57 schwarze 401: man->meta.date = mandoc_strdup("");
1.74 schwarze 402: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
403: n ? n->line : nb->line,
404: n ? n->pos : nb->pos, "TH");
1.64 schwarze 405: }
1.34 schwarze 406:
1.87 schwarze 407: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 408:
409: if (n && (n = n->next))
1.87 schwarze 410: man->meta.os = mandoc_strdup(n->string);
1.102 schwarze 411: else if (man->os_s != NULL)
412: man->meta.os = mandoc_strdup(man->os_s);
413: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
414: if (strstr(man->meta.os, "OpenBSD") != NULL)
415: man->meta.os_e = MANDOC_OS_OPENBSD;
416: else if (strstr(man->meta.os, "NetBSD") != NULL)
417: man->meta.os_e = MANDOC_OS_NETBSD;
418: }
1.34 schwarze 419:
1.87 schwarze 420: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 421: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 422:
423: if (n && (n = n->next))
1.57 schwarze 424: man->meta.vol = mandoc_strdup(n->string);
425: else if ('\0' != man->meta.msec[0] &&
426: (NULL != (p = mandoc_a2msec(man->meta.msec))))
427: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 428:
429: if (n != NULL && (n = n->next) != NULL)
430: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
431: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 432:
433: /*
434: * Remove the `TH' node after we've processed it for our
435: * meta-data.
436: */
1.89 schwarze 437: roff_node_delete(man, man->last);
1.34 schwarze 438: }
439:
1.78 schwarze 440: static void
1.34 schwarze 441: post_UC(CHKARGS)
442: {
443: static const char * const bsd_versions[] = {
444: "3rd Berkeley Distribution",
445: "4th Berkeley Distribution",
446: "4.2 Berkeley Distribution",
447: "4.3 Berkeley Distribution",
448: "4.4 Berkeley Distribution",
449: };
450:
451: const char *p, *s;
452:
453: n = n->child;
454:
1.85 schwarze 455: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 456: p = bsd_versions[0];
457: else {
458: s = n->string;
459: if (0 == strcmp(s, "3"))
460: p = bsd_versions[0];
461: else if (0 == strcmp(s, "4"))
462: p = bsd_versions[1];
463: else if (0 == strcmp(s, "5"))
464: p = bsd_versions[2];
465: else if (0 == strcmp(s, "6"))
466: p = bsd_versions[3];
467: else if (0 == strcmp(s, "7"))
468: p = bsd_versions[4];
469: else
470: p = bsd_versions[0];
471: }
472:
1.87 schwarze 473: free(man->meta.os);
474: man->meta.os = mandoc_strdup(p);
1.34 schwarze 475: }
476:
1.78 schwarze 477: static void
1.34 schwarze 478: post_AT(CHKARGS)
479: {
480: static const char * const unix_versions[] = {
481: "7th Edition",
482: "System III",
483: "System V",
484: "System V Release 2",
485: };
486:
1.86 schwarze 487: struct roff_node *nn;
1.34 schwarze 488: const char *p, *s;
489:
490: n = n->child;
491:
1.85 schwarze 492: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 493: p = unix_versions[0];
494: else {
495: s = n->string;
496: if (0 == strcmp(s, "3"))
497: p = unix_versions[0];
498: else if (0 == strcmp(s, "4"))
499: p = unix_versions[1];
500: else if (0 == strcmp(s, "5")) {
501: nn = n->next;
1.85 schwarze 502: if (nn != NULL &&
503: nn->type == ROFFT_TEXT &&
504: nn->string[0] != '\0')
1.34 schwarze 505: p = unix_versions[3];
506: else
507: p = unix_versions[2];
508: } else
509: p = unix_versions[0];
510: }
511:
1.87 schwarze 512: free(man->meta.os);
513: man->meta.os = mandoc_strdup(p);
1.100 schwarze 514: }
515:
516: static void
517: post_in(CHKARGS)
518: {
519: char *s;
520:
521: if (n->parent->tok != MAN_TP ||
522: n->parent->type != ROFFT_HEAD ||
523: n->child == NULL ||
524: *n->child->string == '+' ||
525: *n->child->string == '-')
526: return;
527: mandoc_asprintf(&s, "+%s", n->child->string);
528: free(n->child->string);
529: n->child->string = s;
1.34 schwarze 530: }