Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.116
1.116 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.115 2018/12/31 08:38:17 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>
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:
1.109 schwarze 42: static void check_abort(CHKARGS);
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("");
1.60 schwarze 186: man->meta.date = man->quick ? mandoc_strdup("") :
1.99 schwarze 187: mandoc_normdate(man, NULL, n->line, n->pos);
1.17 schwarze 188: }
1.101 schwarze 189:
190: if (man->meta.os_e &&
191: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.112 schwarze 192: mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
1.102 schwarze 193: man->meta.os_e == MANDOC_OS_OPENBSD ?
194: "(OpenBSD)" : "(NetBSD)");
1.15 schwarze 195: }
196:
1.78 schwarze 197: static void
1.109 schwarze 198: check_abort(CHKARGS)
199: {
200: abort();
201: }
202:
203: static void
1.47 schwarze 204: check_text(CHKARGS)
1.4 schwarze 205: {
1.47 schwarze 206: char *cp, *p;
207:
1.115 schwarze 208: if (n->flags & NODE_NOFILL)
1.78 schwarze 209: return;
1.48 schwarze 210:
211: cp = n->string;
212: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.112 schwarze 213: mandoc_msg(MANDOCERR_FI_TAB,
214: n->line, n->pos + (int)(p - cp), NULL);
1.115 schwarze 215: }
216:
217: static void
218: post_EE(CHKARGS)
219: {
220: if ((n->flags & NODE_NOFILL) == 0)
221: mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
222: }
223:
224: static void
225: post_EX(CHKARGS)
226: {
227: if (n->flags & NODE_NOFILL)
228: mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
1.1 kristaps 229: }
230:
1.84 schwarze 231: static void
232: post_OP(CHKARGS)
233: {
234:
1.93 schwarze 235: if (n->child == NULL)
1.112 schwarze 236: mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
1.93 schwarze 237: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 238: n = n->child->next->next;
1.112 schwarze 239: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.84 schwarze 240: n->line, n->pos, "OP ... %s", n->string);
241: }
1.1 kristaps 242: }
243:
1.78 schwarze 244: static void
1.110 schwarze 245: post_SH(CHKARGS)
246: {
247: struct roff_node *nc;
248:
249: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
250: return;
251:
252: if (nc->tok == MAN_PP && nc->body->child != NULL) {
253: while (nc->body->last != NULL) {
254: man->next = ROFF_NEXT_CHILD;
255: roff_node_relink(man, nc->body->last);
256: man->last = n;
257: }
258: }
259:
260: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
1.112 schwarze 261: mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
262: "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
1.110 schwarze 263: roff_node_delete(man, nc);
264: }
265:
266: /*
267: * Trailing PP is empty, so it is deleted by check_par().
268: * Trailing sp is significant.
269: */
270:
271: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
1.112 schwarze 272: mandoc_msg(MANDOCERR_PAR_SKIP,
1.110 schwarze 273: nc->line, nc->pos, "%s at the end of %s",
274: roff_name[nc->tok], roff_name[n->tok]);
275: roff_node_delete(man, nc);
276: }
277: }
278:
279: static void
1.76 schwarze 280: post_UR(CHKARGS)
1.58 schwarze 281: {
1.85 schwarze 282: if (n->type == ROFFT_HEAD && n->child == NULL)
1.112 schwarze 283: mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
284: "%s", roff_name[n->tok]);
1.78 schwarze 285: check_part(man, n);
1.32 schwarze 286: }
1.7 schwarze 287:
1.78 schwarze 288: static void
1.8 schwarze 289: check_part(CHKARGS)
290: {
291:
1.85 schwarze 292: if (n->type == ROFFT_BODY && n->child == NULL)
1.112 schwarze 293: mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
294: "%s", roff_name[n->tok]);
1.8 schwarze 295: }
296:
1.78 schwarze 297: static void
1.36 schwarze 298: check_par(CHKARGS)
299: {
300:
1.39 schwarze 301: switch (n->type) {
1.85 schwarze 302: case ROFFT_BLOCK:
1.93 schwarze 303: if (n->body->child == NULL)
1.89 schwarze 304: roff_node_delete(man, n);
1.39 schwarze 305: break;
1.85 schwarze 306: case ROFFT_BODY:
1.110 schwarze 307: if (n->child != NULL &&
308: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
1.112 schwarze 309: mandoc_msg(MANDOCERR_PAR_SKIP,
310: n->child->line, n->child->pos,
1.110 schwarze 311: "%s after %s", roff_name[n->child->tok],
312: roff_name[n->tok]);
313: roff_node_delete(man, n->child);
314: }
1.93 schwarze 315: if (n->child == NULL)
1.112 schwarze 316: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.94 schwarze 317: "%s empty", roff_name[n->tok]);
1.39 schwarze 318: break;
1.85 schwarze 319: case ROFFT_HEAD:
1.93 schwarze 320: if (n->child != NULL)
1.112 schwarze 321: mandoc_msg(MANDOCERR_ARG_SKIP,
322: n->line, n->pos, "%s %s%s",
1.94 schwarze 323: roff_name[n->tok], n->child->string,
1.93 schwarze 324: n->child->next != NULL ? " ..." : "");
1.39 schwarze 325: break;
326: default:
327: break;
328: }
1.36 schwarze 329: }
330:
1.78 schwarze 331: static void
1.55 schwarze 332: post_IP(CHKARGS)
333: {
334:
335: switch (n->type) {
1.85 schwarze 336: case ROFFT_BLOCK:
1.93 schwarze 337: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 338: roff_node_delete(man, n);
1.55 schwarze 339: break;
1.85 schwarze 340: case ROFFT_BODY:
1.93 schwarze 341: if (n->parent->head->child == NULL && n->child == NULL)
1.112 schwarze 342: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.94 schwarze 343: "%s empty", roff_name[n->tok]);
1.55 schwarze 344: break;
345: default:
346: break;
347: }
348: }
1.36 schwarze 349:
1.78 schwarze 350: static void
1.34 schwarze 351: post_TH(CHKARGS)
352: {
1.86 schwarze 353: struct roff_node *nb;
1.40 schwarze 354: const char *p;
1.34 schwarze 355:
1.57 schwarze 356: free(man->meta.title);
357: free(man->meta.vol);
1.87 schwarze 358: free(man->meta.os);
1.57 schwarze 359: free(man->meta.msec);
360: free(man->meta.date);
1.34 schwarze 361:
1.57 schwarze 362: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 363: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 364:
1.64 schwarze 365: nb = n;
366:
1.87 schwarze 367: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 368:
369: n = n->child;
1.40 schwarze 370: if (n && n->string) {
371: for (p = n->string; '\0' != *p; p++) {
372: /* Only warn about this once... */
1.63 schwarze 373: if (isalpha((unsigned char)*p) &&
374: ! isupper((unsigned char)*p)) {
1.112 schwarze 375: mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
376: n->pos + (int)(p - n->string),
1.74 schwarze 377: "TH %s", n->string);
1.40 schwarze 378: break;
379: }
380: }
1.57 schwarze 381: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 382: } else {
1.57 schwarze 383: man->meta.title = mandoc_strdup("");
1.112 schwarze 384: mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
1.77 schwarze 385: }
1.34 schwarze 386:
1.87 schwarze 387: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 388:
1.40 schwarze 389: if (n)
390: n = n->next;
391: if (n && n->string)
1.57 schwarze 392: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 393: else {
1.57 schwarze 394: man->meta.msec = mandoc_strdup("");
1.112 schwarze 395: mandoc_msg(MANDOCERR_MSEC_MISSING,
1.77 schwarze 396: nb->line, nb->pos, "TH %s", man->meta.title);
397: }
1.34 schwarze 398:
1.87 schwarze 399: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 400:
1.40 schwarze 401: if (n)
402: n = n->next;
1.49 schwarze 403: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 404: man->meta.date = man->quick ?
405: mandoc_strdup(n->string) :
1.99 schwarze 406: mandoc_normdate(man, n->string, n->line, n->pos);
1.64 schwarze 407: } else {
1.57 schwarze 408: man->meta.date = mandoc_strdup("");
1.112 schwarze 409: mandoc_msg(MANDOCERR_DATE_MISSING,
1.74 schwarze 410: n ? n->line : nb->line,
411: n ? n->pos : nb->pos, "TH");
1.64 schwarze 412: }
1.34 schwarze 413:
1.87 schwarze 414: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 415:
416: if (n && (n = n->next))
1.87 schwarze 417: man->meta.os = mandoc_strdup(n->string);
1.102 schwarze 418: else if (man->os_s != NULL)
419: man->meta.os = mandoc_strdup(man->os_s);
420: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
421: if (strstr(man->meta.os, "OpenBSD") != NULL)
422: man->meta.os_e = MANDOC_OS_OPENBSD;
423: else if (strstr(man->meta.os, "NetBSD") != NULL)
424: man->meta.os_e = MANDOC_OS_NETBSD;
425: }
1.34 schwarze 426:
1.87 schwarze 427: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 428: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 429:
430: if (n && (n = n->next))
1.57 schwarze 431: man->meta.vol = mandoc_strdup(n->string);
432: else if ('\0' != man->meta.msec[0] &&
433: (NULL != (p = mandoc_a2msec(man->meta.msec))))
434: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 435:
436: if (n != NULL && (n = n->next) != NULL)
1.112 schwarze 437: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.84 schwarze 438: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 439:
440: /*
441: * Remove the `TH' node after we've processed it for our
442: * meta-data.
443: */
1.89 schwarze 444: roff_node_delete(man, man->last);
1.34 schwarze 445: }
446:
1.78 schwarze 447: static void
1.34 schwarze 448: post_UC(CHKARGS)
449: {
450: static const char * const bsd_versions[] = {
451: "3rd Berkeley Distribution",
452: "4th Berkeley Distribution",
453: "4.2 Berkeley Distribution",
454: "4.3 Berkeley Distribution",
455: "4.4 Berkeley Distribution",
456: };
457:
458: const char *p, *s;
459:
460: n = n->child;
461:
1.85 schwarze 462: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 463: p = bsd_versions[0];
464: else {
465: s = n->string;
466: if (0 == strcmp(s, "3"))
467: p = bsd_versions[0];
468: else if (0 == strcmp(s, "4"))
469: p = bsd_versions[1];
470: else if (0 == strcmp(s, "5"))
471: p = bsd_versions[2];
472: else if (0 == strcmp(s, "6"))
473: p = bsd_versions[3];
474: else if (0 == strcmp(s, "7"))
475: p = bsd_versions[4];
476: else
477: p = bsd_versions[0];
478: }
479:
1.87 schwarze 480: free(man->meta.os);
481: man->meta.os = mandoc_strdup(p);
1.34 schwarze 482: }
483:
1.78 schwarze 484: static void
1.34 schwarze 485: post_AT(CHKARGS)
486: {
487: static const char * const unix_versions[] = {
488: "7th Edition",
489: "System III",
490: "System V",
491: "System V Release 2",
492: };
493:
1.86 schwarze 494: struct roff_node *nn;
1.34 schwarze 495: const char *p, *s;
496:
497: n = n->child;
498:
1.85 schwarze 499: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 500: p = unix_versions[0];
501: else {
502: s = n->string;
503: if (0 == strcmp(s, "3"))
504: p = unix_versions[0];
505: else if (0 == strcmp(s, "4"))
506: p = unix_versions[1];
507: else if (0 == strcmp(s, "5")) {
508: nn = n->next;
1.85 schwarze 509: if (nn != NULL &&
510: nn->type == ROFFT_TEXT &&
511: nn->string[0] != '\0')
1.34 schwarze 512: p = unix_versions[3];
513: else
514: p = unix_versions[2];
515: } else
516: p = unix_versions[0];
517: }
518:
1.87 schwarze 519: free(man->meta.os);
520: man->meta.os = mandoc_strdup(p);
1.100 schwarze 521: }
522:
523: static void
524: post_in(CHKARGS)
525: {
526: char *s;
527:
528: if (n->parent->tok != MAN_TP ||
529: n->parent->type != ROFFT_HEAD ||
530: n->child == NULL ||
531: *n->child->string == '+' ||
532: *n->child->string == '-')
533: return;
534: mandoc_asprintf(&s, "+%s", n->child->string);
535: free(n->child->string);
536: n->child->string = s;
1.34 schwarze 537: }