Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.101
1.101 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.100 2017/06/17 16:47:29 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.94 schwarze 4: * Copyright (c) 2010, 2012-2017 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:
1.78 schwarze 41: static void check_par(CHKARGS);
42: static void check_part(CHKARGS);
43: static void check_root(CHKARGS);
44: static void check_text(CHKARGS);
45:
46: static void post_AT(CHKARGS);
47: static void post_IP(CHKARGS);
1.84 schwarze 48: static void post_OP(CHKARGS);
1.78 schwarze 49: static void post_TH(CHKARGS);
50: static void post_UC(CHKARGS);
51: static void post_UR(CHKARGS);
1.100 schwarze 52: static void post_in(CHKARGS);
53: static void post_vs(CHKARGS);
1.34 schwarze 54:
1.94 schwarze 55: static const v_check __man_valids[MAN_MAX - MAN_TH] = {
1.76 schwarze 56: post_TH, /* TH */
57: NULL, /* SH */
58: NULL, /* SS */
59: NULL, /* TP */
60: check_par, /* LP */
61: check_par, /* PP */
62: check_par, /* P */
63: post_IP, /* IP */
64: NULL, /* HP */
65: NULL, /* SM */
66: NULL, /* SB */
67: NULL, /* BI */
68: NULL, /* IB */
69: NULL, /* BR */
70: NULL, /* RB */
71: NULL, /* R */
72: NULL, /* B */
73: NULL, /* I */
74: NULL, /* IR */
75: NULL, /* RI */
1.92 schwarze 76: NULL, /* nf */
77: NULL, /* fi */
1.76 schwarze 78: NULL, /* RE */
79: check_part, /* RS */
80: NULL, /* DT */
81: post_UC, /* UC */
1.83 schwarze 82: NULL, /* PD */
1.76 schwarze 83: post_AT, /* AT */
1.100 schwarze 84: post_in, /* in */
1.84 schwarze 85: post_OP, /* OP */
1.92 schwarze 86: NULL, /* EX */
87: NULL, /* EE */
1.76 schwarze 88: post_UR, /* UR */
89: NULL, /* UE */
1.1 kristaps 90: };
1.94 schwarze 91: static const v_check *man_valids = __man_valids - MAN_TH;
1.1 kristaps 92:
93:
1.78 schwarze 94: void
1.92 schwarze 95: man_node_validate(struct roff_man *man)
1.1 kristaps 96: {
1.86 schwarze 97: struct roff_node *n;
1.94 schwarze 98: const v_check *cp;
1.1 kristaps 99:
1.76 schwarze 100: n = man->last;
1.92 schwarze 101: man->last = man->last->child;
102: while (man->last != NULL) {
103: man_node_validate(man);
104: if (man->last == n)
105: man->last = man->last->child;
106: else
107: man->last = man->last->next;
108: }
1.1 kristaps 109:
1.92 schwarze 110: man->last = n;
111: man->next = ROFF_NEXT_SIBLING;
1.76 schwarze 112: switch (n->type) {
1.85 schwarze 113: case ROFFT_TEXT:
1.78 schwarze 114: check_text(man, n);
115: break;
1.85 schwarze 116: case ROFFT_ROOT:
1.78 schwarze 117: check_root(man, n);
118: break;
1.85 schwarze 119: case ROFFT_EQN:
120: case ROFFT_TBL:
1.78 schwarze 121: break;
1.1 kristaps 122: default:
1.95 schwarze 123: if (n->tok < ROFF_MAX) {
124: switch (n->tok) {
125: case ROFF_br:
1.98 schwarze 126: case ROFF_sp:
1.95 schwarze 127: post_vs(man, n);
128: break;
129: default:
1.96 schwarze 130: roff_validate(man);
131: break;
1.95 schwarze 132: }
133: break;
134: }
135: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.76 schwarze 136: cp = man_valids + n->tok;
1.78 schwarze 137: if (*cp)
138: (*cp)(man, n);
1.92 schwarze 139: if (man->last == n)
140: man_state(man, n);
1.78 schwarze 141: break;
1.1 kristaps 142: }
143: }
144:
1.78 schwarze 145: static void
1.63 schwarze 146: check_root(CHKARGS)
1.4 schwarze 147: {
1.7 schwarze 148:
1.73 schwarze 149: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 150:
1.65 schwarze 151: if (NULL == man->first->child)
1.75 schwarze 152: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
153: n->line, n->pos, NULL);
1.65 schwarze 154: else
155: man->meta.hasbody = 1;
156:
157: if (NULL == man->meta.title) {
1.77 schwarze 158: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 159: n->line, n->pos, NULL);
1.35 schwarze 160:
1.18 schwarze 161: /*
162: * If a title hasn't been set, do so now (by
163: * implication, date and section also aren't set).
164: */
1.35 schwarze 165:
1.77 schwarze 166: man->meta.title = mandoc_strdup("");
167: man->meta.msec = mandoc_strdup("");
1.60 schwarze 168: man->meta.date = man->quick ? mandoc_strdup("") :
1.99 schwarze 169: mandoc_normdate(man, NULL, n->line, n->pos);
1.17 schwarze 170: }
1.101 ! schwarze 171:
! 172: if (man->meta.os_e &&
! 173: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
! 174: mandoc_msg(MANDOCERR_RCS_MISSING, man->parse, 0, 0, NULL);
1.15 schwarze 175: }
176:
1.78 schwarze 177: static void
1.47 schwarze 178: check_text(CHKARGS)
1.4 schwarze 179: {
1.47 schwarze 180: char *cp, *p;
181:
1.57 schwarze 182: if (MAN_LITERAL & man->flags)
1.78 schwarze 183: return;
1.48 schwarze 184:
185: cp = n->string;
186: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 187: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
188: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 189: }
190:
1.84 schwarze 191: static void
192: post_OP(CHKARGS)
193: {
194:
1.93 schwarze 195: if (n->child == NULL)
1.84 schwarze 196: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
197: n->line, n->pos, "OP");
1.93 schwarze 198: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 199: n = n->child->next->next;
200: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
201: n->line, n->pos, "OP ... %s", n->string);
202: }
1.1 kristaps 203: }
204:
1.78 schwarze 205: static void
1.76 schwarze 206: post_UR(CHKARGS)
1.58 schwarze 207: {
208:
1.85 schwarze 209: if (n->type == ROFFT_HEAD && n->child == NULL)
1.84 schwarze 210: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
211: n->line, n->pos, "UR");
1.78 schwarze 212: check_part(man, n);
1.32 schwarze 213: }
1.7 schwarze 214:
1.78 schwarze 215: static void
1.8 schwarze 216: check_part(CHKARGS)
217: {
218:
1.85 schwarze 219: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 220: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.94 schwarze 221: n->line, n->pos, roff_name[n->tok]);
1.8 schwarze 222: }
223:
1.78 schwarze 224: static void
1.36 schwarze 225: check_par(CHKARGS)
226: {
227:
1.39 schwarze 228: switch (n->type) {
1.85 schwarze 229: case ROFFT_BLOCK:
1.93 schwarze 230: if (n->body->child == NULL)
1.89 schwarze 231: roff_node_delete(man, n);
1.39 schwarze 232: break;
1.85 schwarze 233: case ROFFT_BODY:
1.93 schwarze 234: if (n->child == NULL)
1.67 schwarze 235: mandoc_vmsg(MANDOCERR_PAR_SKIP,
236: man->parse, n->line, n->pos,
1.94 schwarze 237: "%s empty", roff_name[n->tok]);
1.39 schwarze 238: break;
1.85 schwarze 239: case ROFFT_HEAD:
1.93 schwarze 240: if (n->child != NULL)
1.69 schwarze 241: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.94 schwarze 242: man->parse, n->line, n->pos, "%s %s%s",
243: roff_name[n->tok], n->child->string,
1.93 schwarze 244: n->child->next != NULL ? " ..." : "");
1.39 schwarze 245: break;
246: default:
247: break;
248: }
1.36 schwarze 249: }
250:
1.78 schwarze 251: static void
1.55 schwarze 252: post_IP(CHKARGS)
253: {
254:
255: switch (n->type) {
1.85 schwarze 256: case ROFFT_BLOCK:
1.93 schwarze 257: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 258: roff_node_delete(man, n);
1.55 schwarze 259: break;
1.85 schwarze 260: case ROFFT_BODY:
1.93 schwarze 261: if (n->parent->head->child == NULL && n->child == NULL)
1.67 schwarze 262: mandoc_vmsg(MANDOCERR_PAR_SKIP,
263: man->parse, n->line, n->pos,
1.94 schwarze 264: "%s empty", roff_name[n->tok]);
1.55 schwarze 265: break;
266: default:
267: break;
268: }
269: }
1.36 schwarze 270:
1.78 schwarze 271: static void
1.34 schwarze 272: post_TH(CHKARGS)
273: {
1.86 schwarze 274: struct roff_node *nb;
1.40 schwarze 275: const char *p;
1.34 schwarze 276:
1.57 schwarze 277: free(man->meta.title);
278: free(man->meta.vol);
1.87 schwarze 279: free(man->meta.os);
1.57 schwarze 280: free(man->meta.msec);
281: free(man->meta.date);
1.34 schwarze 282:
1.57 schwarze 283: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 284: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 285:
1.64 schwarze 286: nb = n;
287:
1.87 schwarze 288: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 289:
290: n = n->child;
1.40 schwarze 291: if (n && n->string) {
292: for (p = n->string; '\0' != *p; p++) {
293: /* Only warn about this once... */
1.63 schwarze 294: if (isalpha((unsigned char)*p) &&
295: ! isupper((unsigned char)*p)) {
1.74 schwarze 296: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 297: man->parse, n->line,
298: n->pos + (p - n->string),
1.74 schwarze 299: "TH %s", n->string);
1.40 schwarze 300: break;
301: }
302: }
1.57 schwarze 303: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 304: } else {
1.57 schwarze 305: man->meta.title = mandoc_strdup("");
1.77 schwarze 306: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
307: nb->line, nb->pos, "TH");
308: }
1.34 schwarze 309:
1.87 schwarze 310: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 311:
1.40 schwarze 312: if (n)
313: n = n->next;
314: if (n && n->string)
1.57 schwarze 315: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 316: else {
1.57 schwarze 317: man->meta.msec = mandoc_strdup("");
1.77 schwarze 318: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
319: nb->line, nb->pos, "TH %s", man->meta.title);
320: }
1.34 schwarze 321:
1.87 schwarze 322: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 323:
1.40 schwarze 324: if (n)
325: n = n->next;
1.49 schwarze 326: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 327: man->meta.date = man->quick ?
328: mandoc_strdup(n->string) :
1.99 schwarze 329: mandoc_normdate(man, n->string, n->line, n->pos);
1.64 schwarze 330: } else {
1.57 schwarze 331: man->meta.date = mandoc_strdup("");
1.74 schwarze 332: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
333: n ? n->line : nb->line,
334: n ? n->pos : nb->pos, "TH");
1.64 schwarze 335: }
1.34 schwarze 336:
1.87 schwarze 337: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 338:
339: if (n && (n = n->next))
1.87 schwarze 340: man->meta.os = mandoc_strdup(n->string);
1.79 schwarze 341: else if (man->defos != NULL)
1.87 schwarze 342: man->meta.os = mandoc_strdup(man->defos);
1.101 ! schwarze 343: man->meta.os_e = man->meta.os == NULL ? MDOC_OS_OTHER :
! 344: strstr(man->meta.os, "OpenBSD") != NULL ? MDOC_OS_OPENBSD :
! 345: strstr(man->meta.os, "NetBSD") != NULL ? MDOC_OS_NETBSD :
! 346: MDOC_OS_OTHER;
1.34 schwarze 347:
1.87 schwarze 348: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 349: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 350:
351: if (n && (n = n->next))
1.57 schwarze 352: man->meta.vol = mandoc_strdup(n->string);
353: else if ('\0' != man->meta.msec[0] &&
354: (NULL != (p = mandoc_a2msec(man->meta.msec))))
355: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 356:
357: if (n != NULL && (n = n->next) != NULL)
358: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
359: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 360:
361: /*
362: * Remove the `TH' node after we've processed it for our
363: * meta-data.
364: */
1.89 schwarze 365: roff_node_delete(man, man->last);
1.34 schwarze 366: }
367:
1.78 schwarze 368: static void
1.34 schwarze 369: post_UC(CHKARGS)
370: {
371: static const char * const bsd_versions[] = {
372: "3rd Berkeley Distribution",
373: "4th Berkeley Distribution",
374: "4.2 Berkeley Distribution",
375: "4.3 Berkeley Distribution",
376: "4.4 Berkeley Distribution",
377: };
378:
379: const char *p, *s;
380:
381: n = n->child;
382:
1.85 schwarze 383: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 384: p = bsd_versions[0];
385: else {
386: s = n->string;
387: if (0 == strcmp(s, "3"))
388: p = bsd_versions[0];
389: else if (0 == strcmp(s, "4"))
390: p = bsd_versions[1];
391: else if (0 == strcmp(s, "5"))
392: p = bsd_versions[2];
393: else if (0 == strcmp(s, "6"))
394: p = bsd_versions[3];
395: else if (0 == strcmp(s, "7"))
396: p = bsd_versions[4];
397: else
398: p = bsd_versions[0];
399: }
400:
1.87 schwarze 401: free(man->meta.os);
402: man->meta.os = mandoc_strdup(p);
1.34 schwarze 403: }
404:
1.78 schwarze 405: static void
1.34 schwarze 406: post_AT(CHKARGS)
407: {
408: static const char * const unix_versions[] = {
409: "7th Edition",
410: "System III",
411: "System V",
412: "System V Release 2",
413: };
414:
1.86 schwarze 415: struct roff_node *nn;
1.34 schwarze 416: const char *p, *s;
417:
418: n = n->child;
419:
1.85 schwarze 420: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 421: p = unix_versions[0];
422: else {
423: s = n->string;
424: if (0 == strcmp(s, "3"))
425: p = unix_versions[0];
426: else if (0 == strcmp(s, "4"))
427: p = unix_versions[1];
428: else if (0 == strcmp(s, "5")) {
429: nn = n->next;
1.85 schwarze 430: if (nn != NULL &&
431: nn->type == ROFFT_TEXT &&
432: nn->string[0] != '\0')
1.34 schwarze 433: p = unix_versions[3];
434: else
435: p = unix_versions[2];
436: } else
437: p = unix_versions[0];
438: }
439:
1.87 schwarze 440: free(man->meta.os);
441: man->meta.os = mandoc_strdup(p);
1.100 schwarze 442: }
443:
444: static void
445: post_in(CHKARGS)
446: {
447: char *s;
448:
449: if (n->parent->tok != MAN_TP ||
450: n->parent->type != ROFFT_HEAD ||
451: n->child == NULL ||
452: *n->child->string == '+' ||
453: *n->child->string == '-')
454: return;
455: mandoc_asprintf(&s, "+%s", n->child->string);
456: free(n->child->string);
457: n->child->string = s;
1.46 schwarze 458: }
459:
1.78 schwarze 460: static void
1.46 schwarze 461: post_vs(CHKARGS)
462: {
463:
1.54 schwarze 464: if (NULL != n->prev)
1.78 schwarze 465: return;
1.54 schwarze 466:
467: switch (n->parent->tok) {
1.63 schwarze 468: case MAN_SH:
469: case MAN_SS:
1.98 schwarze 470: case MAN_PP:
471: case MAN_LP:
472: case MAN_P:
1.67 schwarze 473: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.94 schwarze 474: "%s after %s", roff_name[n->tok],
475: roff_name[n->parent->tok]);
1.54 schwarze 476: /* FALLTHROUGH */
1.90 schwarze 477: case TOKEN_NONE:
1.63 schwarze 478: /*
1.54 schwarze 479: * Don't warn about this because it occurs in pod2man
480: * and would cause considerable (unfixable) warnage.
481: */
1.89 schwarze 482: roff_node_delete(man, n);
1.54 schwarze 483: break;
484: default:
485: break;
486: }
1.34 schwarze 487: }