Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.120
1.120 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.119 2019/06/27 15:05:14 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.120 ! schwarze 4: * Copyright (c) 2010, 2012-2020 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>
1.111 schwarze 25: #include <stdio.h>
1.1 kristaps 26: #include <stdlib.h>
1.28 schwarze 27: #include <string.h>
1.36 schwarze 28: #include <time.h>
1.1 kristaps 29:
1.85 schwarze 30: #include "mandoc_aux.h"
31: #include "mandoc.h"
32: #include "roff.h"
1.44 schwarze 33: #include "man.h"
1.85 schwarze 34: #include "libmandoc.h"
1.89 schwarze 35: #include "roff_int.h"
1.1 kristaps 36: #include "libman.h"
37:
1.88 schwarze 38: #define CHKARGS struct roff_man *man, struct roff_node *n
1.1 kristaps 39:
1.78 schwarze 40: typedef void (*v_check)(CHKARGS);
1.1 kristaps 41:
1.118 schwarze 42: static void check_abort(CHKARGS) __attribute__((__noreturn__));
1.78 schwarze 43: static void check_par(CHKARGS);
44: static void check_part(CHKARGS);
45: static void check_root(CHKARGS);
46: static void check_text(CHKARGS);
47:
48: static void post_AT(CHKARGS);
1.115 schwarze 49: static void post_EE(CHKARGS);
50: static void post_EX(CHKARGS);
1.78 schwarze 51: static void post_IP(CHKARGS);
1.84 schwarze 52: static void post_OP(CHKARGS);
1.110 schwarze 53: static void post_SH(CHKARGS);
1.78 schwarze 54: static void post_TH(CHKARGS);
55: static void post_UC(CHKARGS);
56: static void post_UR(CHKARGS);
1.100 schwarze 57: static void post_in(CHKARGS);
1.34 schwarze 58:
1.106 schwarze 59: static const v_check man_valids[MAN_MAX - MAN_TH] = {
1.76 schwarze 60: post_TH, /* TH */
1.110 schwarze 61: post_SH, /* SH */
62: post_SH, /* SS */
1.76 schwarze 63: NULL, /* TP */
1.107 schwarze 64: NULL, /* TQ */
1.109 schwarze 65: check_abort,/* LP */
1.76 schwarze 66: check_par, /* PP */
1.109 schwarze 67: check_abort,/* P */
1.76 schwarze 68: post_IP, /* IP */
69: NULL, /* HP */
70: NULL, /* SM */
71: NULL, /* SB */
72: NULL, /* BI */
73: NULL, /* IB */
74: NULL, /* BR */
75: NULL, /* RB */
76: NULL, /* R */
77: NULL, /* B */
78: NULL, /* I */
79: NULL, /* IR */
80: NULL, /* RI */
81: NULL, /* RE */
82: check_part, /* RS */
83: NULL, /* DT */
84: post_UC, /* UC */
1.83 schwarze 85: NULL, /* PD */
1.76 schwarze 86: post_AT, /* AT */
1.100 schwarze 87: post_in, /* in */
1.108 schwarze 88: NULL, /* SY */
89: NULL, /* YS */
1.84 schwarze 90: post_OP, /* OP */
1.115 schwarze 91: post_EX, /* EX */
92: post_EE, /* EE */
1.76 schwarze 93: post_UR, /* UR */
94: NULL, /* UE */
1.103 bentley 95: post_UR, /* MT */
96: NULL, /* ME */
1.1 kristaps 97: };
98:
99:
1.109 schwarze 100: /* Validate the subtree rooted at man->last. */
1.78 schwarze 101: void
1.113 schwarze 102: man_validate(struct roff_man *man)
1.1 kristaps 103: {
1.86 schwarze 104: struct roff_node *n;
1.94 schwarze 105: const v_check *cp;
1.1 kristaps 106:
1.109 schwarze 107: /*
108: * Translate obsolete macros such that later code
109: * does not need to look for them.
110: */
111:
1.76 schwarze 112: n = man->last;
1.109 schwarze 113: switch (n->tok) {
114: case MAN_LP:
115: case MAN_P:
116: n->tok = MAN_PP;
117: break;
118: default:
119: break;
120: }
121:
122: /*
123: * Iterate over all children, recursing into each one
124: * in turn, depth-first.
125: */
126:
1.92 schwarze 127: man->last = man->last->child;
128: while (man->last != NULL) {
1.113 schwarze 129: man_validate(man);
1.92 schwarze 130: if (man->last == n)
131: man->last = man->last->child;
132: else
133: man->last = man->last->next;
134: }
1.1 kristaps 135:
1.109 schwarze 136: /* Finally validate the macro itself. */
137:
1.92 schwarze 138: man->last = n;
139: man->next = ROFF_NEXT_SIBLING;
1.76 schwarze 140: switch (n->type) {
1.85 schwarze 141: case ROFFT_TEXT:
1.78 schwarze 142: check_text(man, n);
143: break;
1.85 schwarze 144: case ROFFT_ROOT:
1.78 schwarze 145: check_root(man, n);
146: break;
1.105 schwarze 147: case ROFFT_COMMENT:
1.85 schwarze 148: case ROFFT_EQN:
149: case ROFFT_TBL:
1.78 schwarze 150: break;
1.1 kristaps 151: default:
1.95 schwarze 152: if (n->tok < ROFF_MAX) {
1.110 schwarze 153: roff_validate(man);
1.95 schwarze 154: break;
155: }
156: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.106 schwarze 157: cp = man_valids + (n->tok - MAN_TH);
1.78 schwarze 158: if (*cp)
159: (*cp)(man, n);
1.92 schwarze 160: if (man->last == n)
1.116 schwarze 161: n->flags |= NODE_VALID;
1.78 schwarze 162: break;
1.1 kristaps 163: }
164: }
165:
1.78 schwarze 166: static void
1.63 schwarze 167: check_root(CHKARGS)
1.4 schwarze 168: {
1.73 schwarze 169: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 170:
1.105 schwarze 171: if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1.112 schwarze 172: mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
1.65 schwarze 173: else
174: man->meta.hasbody = 1;
175:
176: if (NULL == man->meta.title) {
1.112 schwarze 177: mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
1.35 schwarze 178:
1.18 schwarze 179: /*
180: * If a title hasn't been set, do so now (by
181: * implication, date and section also aren't set).
182: */
1.35 schwarze 183:
1.77 schwarze 184: man->meta.title = mandoc_strdup("");
185: man->meta.msec = mandoc_strdup("");
1.120 ! schwarze 186: man->meta.date = mandoc_normdate(NULL, NULL);
1.17 schwarze 187: }
1.101 schwarze 188:
189: if (man->meta.os_e &&
190: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.112 schwarze 191: mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
1.102 schwarze 192: man->meta.os_e == MANDOC_OS_OPENBSD ?
193: "(OpenBSD)" : "(NetBSD)");
1.15 schwarze 194: }
195:
1.118 schwarze 196: static void
1.109 schwarze 197: check_abort(CHKARGS)
198: {
199: abort();
200: }
201:
202: static void
1.47 schwarze 203: check_text(CHKARGS)
1.4 schwarze 204: {
1.47 schwarze 205: char *cp, *p;
206:
1.115 schwarze 207: if (n->flags & NODE_NOFILL)
1.78 schwarze 208: return;
1.48 schwarze 209:
210: cp = n->string;
211: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.112 schwarze 212: mandoc_msg(MANDOCERR_FI_TAB,
213: n->line, n->pos + (int)(p - cp), NULL);
1.115 schwarze 214: }
215:
216: static void
217: post_EE(CHKARGS)
218: {
219: if ((n->flags & NODE_NOFILL) == 0)
220: mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
221: }
222:
223: static void
224: post_EX(CHKARGS)
225: {
226: if (n->flags & NODE_NOFILL)
227: mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
1.1 kristaps 228: }
229:
1.84 schwarze 230: static void
231: post_OP(CHKARGS)
232: {
233:
1.93 schwarze 234: if (n->child == NULL)
1.112 schwarze 235: mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
1.93 schwarze 236: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 237: n = n->child->next->next;
1.112 schwarze 238: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.84 schwarze 239: n->line, n->pos, "OP ... %s", n->string);
240: }
1.1 kristaps 241: }
242:
1.78 schwarze 243: static void
1.110 schwarze 244: post_SH(CHKARGS)
245: {
246: struct roff_node *nc;
247:
248: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
249: return;
250:
251: if (nc->tok == MAN_PP && nc->body->child != NULL) {
252: while (nc->body->last != NULL) {
253: man->next = ROFF_NEXT_CHILD;
254: roff_node_relink(man, nc->body->last);
255: man->last = n;
256: }
257: }
258:
259: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
1.112 schwarze 260: mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
261: "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
1.110 schwarze 262: roff_node_delete(man, nc);
263: }
264:
265: /*
266: * Trailing PP is empty, so it is deleted by check_par().
267: * Trailing sp is significant.
268: */
269:
270: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
1.112 schwarze 271: mandoc_msg(MANDOCERR_PAR_SKIP,
1.110 schwarze 272: nc->line, nc->pos, "%s at the end of %s",
273: roff_name[nc->tok], roff_name[n->tok]);
274: roff_node_delete(man, nc);
275: }
276: }
277:
278: static void
1.76 schwarze 279: post_UR(CHKARGS)
1.58 schwarze 280: {
1.85 schwarze 281: if (n->type == ROFFT_HEAD && n->child == NULL)
1.112 schwarze 282: mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
283: "%s", roff_name[n->tok]);
1.78 schwarze 284: check_part(man, n);
1.32 schwarze 285: }
1.7 schwarze 286:
1.78 schwarze 287: static void
1.8 schwarze 288: check_part(CHKARGS)
289: {
290:
1.85 schwarze 291: if (n->type == ROFFT_BODY && n->child == NULL)
1.112 schwarze 292: mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
293: "%s", roff_name[n->tok]);
1.8 schwarze 294: }
295:
1.78 schwarze 296: static void
1.36 schwarze 297: check_par(CHKARGS)
298: {
299:
1.39 schwarze 300: switch (n->type) {
1.85 schwarze 301: case ROFFT_BLOCK:
1.93 schwarze 302: if (n->body->child == NULL)
1.89 schwarze 303: roff_node_delete(man, n);
1.39 schwarze 304: break;
1.85 schwarze 305: case ROFFT_BODY:
1.110 schwarze 306: if (n->child != NULL &&
307: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
1.112 schwarze 308: mandoc_msg(MANDOCERR_PAR_SKIP,
309: n->child->line, n->child->pos,
1.110 schwarze 310: "%s after %s", roff_name[n->child->tok],
311: roff_name[n->tok]);
312: roff_node_delete(man, n->child);
313: }
1.93 schwarze 314: if (n->child == NULL)
1.112 schwarze 315: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.94 schwarze 316: "%s empty", roff_name[n->tok]);
1.39 schwarze 317: break;
1.85 schwarze 318: case ROFFT_HEAD:
1.93 schwarze 319: if (n->child != NULL)
1.112 schwarze 320: mandoc_msg(MANDOCERR_ARG_SKIP,
321: n->line, n->pos, "%s %s%s",
1.94 schwarze 322: roff_name[n->tok], n->child->string,
1.93 schwarze 323: n->child->next != NULL ? " ..." : "");
1.39 schwarze 324: break;
325: default:
326: break;
327: }
1.36 schwarze 328: }
329:
1.78 schwarze 330: static void
1.55 schwarze 331: post_IP(CHKARGS)
332: {
333:
334: switch (n->type) {
1.85 schwarze 335: case ROFFT_BLOCK:
1.93 schwarze 336: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 337: roff_node_delete(man, n);
1.55 schwarze 338: break;
1.85 schwarze 339: case ROFFT_BODY:
1.93 schwarze 340: if (n->parent->head->child == NULL && n->child == NULL)
1.112 schwarze 341: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.94 schwarze 342: "%s empty", roff_name[n->tok]);
1.55 schwarze 343: break;
344: default:
345: break;
346: }
347: }
1.36 schwarze 348:
1.78 schwarze 349: static void
1.34 schwarze 350: post_TH(CHKARGS)
351: {
1.86 schwarze 352: struct roff_node *nb;
1.40 schwarze 353: const char *p;
1.34 schwarze 354:
1.57 schwarze 355: free(man->meta.title);
356: free(man->meta.vol);
1.87 schwarze 357: free(man->meta.os);
1.57 schwarze 358: free(man->meta.msec);
359: free(man->meta.date);
1.34 schwarze 360:
1.57 schwarze 361: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 362: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 363:
1.64 schwarze 364: nb = n;
365:
1.87 schwarze 366: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 367:
368: n = n->child;
1.119 schwarze 369: if (n != NULL && n->string != NULL) {
370: for (p = n->string; *p != '\0'; p++) {
1.40 schwarze 371: /* Only warn about this once... */
1.63 schwarze 372: if (isalpha((unsigned char)*p) &&
373: ! isupper((unsigned char)*p)) {
1.112 schwarze 374: mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
375: n->pos + (int)(p - n->string),
1.74 schwarze 376: "TH %s", n->string);
1.40 schwarze 377: break;
378: }
379: }
1.57 schwarze 380: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 381: } else {
1.57 schwarze 382: man->meta.title = mandoc_strdup("");
1.112 schwarze 383: mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
1.77 schwarze 384: }
1.34 schwarze 385:
1.87 schwarze 386: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 387:
1.119 schwarze 388: if (n != NULL)
1.40 schwarze 389: n = n->next;
1.119 schwarze 390: if (n != NULL && n->string != NULL)
1.57 schwarze 391: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 392: else {
1.57 schwarze 393: man->meta.msec = mandoc_strdup("");
1.112 schwarze 394: mandoc_msg(MANDOCERR_MSEC_MISSING,
1.77 schwarze 395: nb->line, nb->pos, "TH %s", man->meta.title);
396: }
1.34 schwarze 397:
1.87 schwarze 398: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 399:
1.119 schwarze 400: if (n != NULL)
1.40 schwarze 401: n = n->next;
1.120 ! schwarze 402: if (man->quick && n != NULL)
1.57 schwarze 403: man->meta.date = mandoc_strdup("");
1.120 ! schwarze 404: else
! 405: man->meta.date = mandoc_normdate(n, nb);
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)
1.112 schwarze 430: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.84 schwarze 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: }