Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.107
1.107 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.106 2018/08/16 15:04:45 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:
1.78 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:
1.106 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 */
1.107 ! schwarze 60: NULL, /* TQ */
1.76 schwarze 61: check_par, /* LP */
62: check_par, /* PP */
63: check_par, /* P */
64: post_IP, /* IP */
65: NULL, /* HP */
66: NULL, /* SM */
67: NULL, /* SB */
68: NULL, /* BI */
69: NULL, /* IB */
70: NULL, /* BR */
71: NULL, /* RB */
72: NULL, /* R */
73: NULL, /* B */
74: NULL, /* I */
75: NULL, /* IR */
76: NULL, /* RI */
1.92 schwarze 77: NULL, /* nf */
78: NULL, /* fi */
1.76 schwarze 79: NULL, /* RE */
80: check_part, /* RS */
81: NULL, /* DT */
82: post_UC, /* UC */
1.83 schwarze 83: NULL, /* PD */
1.76 schwarze 84: post_AT, /* AT */
1.100 schwarze 85: post_in, /* in */
1.84 schwarze 86: post_OP, /* OP */
1.92 schwarze 87: NULL, /* EX */
88: NULL, /* EE */
1.76 schwarze 89: post_UR, /* UR */
90: NULL, /* UE */
1.103 bentley 91: post_UR, /* MT */
92: NULL, /* ME */
1.1 kristaps 93: };
94:
95:
1.78 schwarze 96: void
1.92 schwarze 97: man_node_validate(struct roff_man *man)
1.1 kristaps 98: {
1.86 schwarze 99: struct roff_node *n;
1.94 schwarze 100: const v_check *cp;
1.1 kristaps 101:
1.76 schwarze 102: n = man->last;
1.92 schwarze 103: man->last = man->last->child;
104: while (man->last != NULL) {
105: man_node_validate(man);
106: if (man->last == n)
107: man->last = man->last->child;
108: else
109: man->last = man->last->next;
110: }
1.1 kristaps 111:
1.92 schwarze 112: man->last = n;
113: man->next = ROFF_NEXT_SIBLING;
1.76 schwarze 114: switch (n->type) {
1.85 schwarze 115: case ROFFT_TEXT:
1.78 schwarze 116: check_text(man, n);
117: break;
1.85 schwarze 118: case ROFFT_ROOT:
1.78 schwarze 119: check_root(man, n);
120: break;
1.105 schwarze 121: case ROFFT_COMMENT:
1.85 schwarze 122: case ROFFT_EQN:
123: case ROFFT_TBL:
1.78 schwarze 124: break;
1.1 kristaps 125: default:
1.95 schwarze 126: if (n->tok < ROFF_MAX) {
127: switch (n->tok) {
128: case ROFF_br:
1.98 schwarze 129: case ROFF_sp:
1.95 schwarze 130: post_vs(man, n);
131: break;
132: default:
1.96 schwarze 133: roff_validate(man);
134: break;
1.95 schwarze 135: }
136: break;
137: }
138: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.106 schwarze 139: cp = man_valids + (n->tok - MAN_TH);
1.78 schwarze 140: if (*cp)
141: (*cp)(man, n);
1.92 schwarze 142: if (man->last == n)
143: man_state(man, n);
1.78 schwarze 144: break;
1.1 kristaps 145: }
146: }
147:
1.78 schwarze 148: static void
1.63 schwarze 149: check_root(CHKARGS)
1.4 schwarze 150: {
1.73 schwarze 151: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 152:
1.105 schwarze 153: if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1.75 schwarze 154: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
155: n->line, n->pos, NULL);
1.65 schwarze 156: else
157: man->meta.hasbody = 1;
158:
159: if (NULL == man->meta.title) {
1.77 schwarze 160: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 161: n->line, n->pos, NULL);
1.35 schwarze 162:
1.18 schwarze 163: /*
164: * If a title hasn't been set, do so now (by
165: * implication, date and section also aren't set).
166: */
1.35 schwarze 167:
1.77 schwarze 168: man->meta.title = mandoc_strdup("");
169: man->meta.msec = mandoc_strdup("");
1.60 schwarze 170: man->meta.date = man->quick ? mandoc_strdup("") :
1.99 schwarze 171: mandoc_normdate(man, NULL, n->line, n->pos);
1.17 schwarze 172: }
1.101 schwarze 173:
174: if (man->meta.os_e &&
175: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.102 schwarze 176: mandoc_msg(MANDOCERR_RCS_MISSING, man->parse, 0, 0,
177: man->meta.os_e == MANDOC_OS_OPENBSD ?
178: "(OpenBSD)" : "(NetBSD)");
1.15 schwarze 179: }
180:
1.78 schwarze 181: static void
1.47 schwarze 182: check_text(CHKARGS)
1.4 schwarze 183: {
1.47 schwarze 184: char *cp, *p;
185:
1.57 schwarze 186: if (MAN_LITERAL & man->flags)
1.78 schwarze 187: return;
1.48 schwarze 188:
189: cp = n->string;
190: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 191: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
192: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 193: }
194:
1.84 schwarze 195: static void
196: post_OP(CHKARGS)
197: {
198:
1.93 schwarze 199: if (n->child == NULL)
1.84 schwarze 200: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
201: n->line, n->pos, "OP");
1.93 schwarze 202: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 203: n = n->child->next->next;
204: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
205: n->line, n->pos, "OP ... %s", n->string);
206: }
1.1 kristaps 207: }
208:
1.78 schwarze 209: static void
1.76 schwarze 210: post_UR(CHKARGS)
1.58 schwarze 211: {
1.85 schwarze 212: if (n->type == ROFFT_HEAD && n->child == NULL)
1.104 schwarze 213: mandoc_msg(MANDOCERR_UR_NOHEAD, man->parse,
1.103 bentley 214: n->line, n->pos, roff_name[n->tok]);
1.78 schwarze 215: check_part(man, n);
1.32 schwarze 216: }
1.7 schwarze 217:
1.78 schwarze 218: static void
1.8 schwarze 219: check_part(CHKARGS)
220: {
221:
1.85 schwarze 222: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 223: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.94 schwarze 224: n->line, n->pos, roff_name[n->tok]);
1.8 schwarze 225: }
226:
1.78 schwarze 227: static void
1.36 schwarze 228: check_par(CHKARGS)
229: {
230:
1.39 schwarze 231: switch (n->type) {
1.85 schwarze 232: case ROFFT_BLOCK:
1.93 schwarze 233: if (n->body->child == NULL)
1.89 schwarze 234: roff_node_delete(man, n);
1.39 schwarze 235: break;
1.85 schwarze 236: case ROFFT_BODY:
1.93 schwarze 237: if (n->child == NULL)
1.67 schwarze 238: mandoc_vmsg(MANDOCERR_PAR_SKIP,
239: man->parse, n->line, n->pos,
1.94 schwarze 240: "%s empty", roff_name[n->tok]);
1.39 schwarze 241: break;
1.85 schwarze 242: case ROFFT_HEAD:
1.93 schwarze 243: if (n->child != NULL)
1.69 schwarze 244: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.94 schwarze 245: man->parse, n->line, n->pos, "%s %s%s",
246: roff_name[n->tok], n->child->string,
1.93 schwarze 247: n->child->next != NULL ? " ..." : "");
1.39 schwarze 248: break;
249: default:
250: break;
251: }
1.36 schwarze 252: }
253:
1.78 schwarze 254: static void
1.55 schwarze 255: post_IP(CHKARGS)
256: {
257:
258: switch (n->type) {
1.85 schwarze 259: case ROFFT_BLOCK:
1.93 schwarze 260: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 261: roff_node_delete(man, n);
1.55 schwarze 262: break;
1.85 schwarze 263: case ROFFT_BODY:
1.93 schwarze 264: if (n->parent->head->child == NULL && n->child == NULL)
1.67 schwarze 265: mandoc_vmsg(MANDOCERR_PAR_SKIP,
266: man->parse, n->line, n->pos,
1.94 schwarze 267: "%s empty", roff_name[n->tok]);
1.55 schwarze 268: break;
269: default:
270: break;
271: }
272: }
1.36 schwarze 273:
1.78 schwarze 274: static void
1.34 schwarze 275: post_TH(CHKARGS)
276: {
1.86 schwarze 277: struct roff_node *nb;
1.40 schwarze 278: const char *p;
1.34 schwarze 279:
1.57 schwarze 280: free(man->meta.title);
281: free(man->meta.vol);
1.87 schwarze 282: free(man->meta.os);
1.57 schwarze 283: free(man->meta.msec);
284: free(man->meta.date);
1.34 schwarze 285:
1.57 schwarze 286: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 287: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 288:
1.64 schwarze 289: nb = n;
290:
1.87 schwarze 291: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 292:
293: n = n->child;
1.40 schwarze 294: if (n && n->string) {
295: for (p = n->string; '\0' != *p; p++) {
296: /* Only warn about this once... */
1.63 schwarze 297: if (isalpha((unsigned char)*p) &&
298: ! isupper((unsigned char)*p)) {
1.74 schwarze 299: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 300: man->parse, n->line,
301: n->pos + (p - n->string),
1.74 schwarze 302: "TH %s", n->string);
1.40 schwarze 303: break;
304: }
305: }
1.57 schwarze 306: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 307: } else {
1.57 schwarze 308: man->meta.title = mandoc_strdup("");
1.77 schwarze 309: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
310: nb->line, nb->pos, "TH");
311: }
1.34 schwarze 312:
1.87 schwarze 313: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 314:
1.40 schwarze 315: if (n)
316: n = n->next;
317: if (n && n->string)
1.57 schwarze 318: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 319: else {
1.57 schwarze 320: man->meta.msec = mandoc_strdup("");
1.77 schwarze 321: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
322: nb->line, nb->pos, "TH %s", man->meta.title);
323: }
1.34 schwarze 324:
1.87 schwarze 325: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 326:
1.40 schwarze 327: if (n)
328: n = n->next;
1.49 schwarze 329: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 330: man->meta.date = man->quick ?
331: mandoc_strdup(n->string) :
1.99 schwarze 332: mandoc_normdate(man, n->string, n->line, n->pos);
1.64 schwarze 333: } else {
1.57 schwarze 334: man->meta.date = mandoc_strdup("");
1.74 schwarze 335: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
336: n ? n->line : nb->line,
337: n ? n->pos : nb->pos, "TH");
1.64 schwarze 338: }
1.34 schwarze 339:
1.87 schwarze 340: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 341:
342: if (n && (n = n->next))
1.87 schwarze 343: man->meta.os = mandoc_strdup(n->string);
1.102 schwarze 344: else if (man->os_s != NULL)
345: man->meta.os = mandoc_strdup(man->os_s);
346: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
347: if (strstr(man->meta.os, "OpenBSD") != NULL)
348: man->meta.os_e = MANDOC_OS_OPENBSD;
349: else if (strstr(man->meta.os, "NetBSD") != NULL)
350: man->meta.os_e = MANDOC_OS_NETBSD;
351: }
1.34 schwarze 352:
1.87 schwarze 353: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 354: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 355:
356: if (n && (n = n->next))
1.57 schwarze 357: man->meta.vol = mandoc_strdup(n->string);
358: else if ('\0' != man->meta.msec[0] &&
359: (NULL != (p = mandoc_a2msec(man->meta.msec))))
360: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 361:
362: if (n != NULL && (n = n->next) != NULL)
363: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
364: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 365:
366: /*
367: * Remove the `TH' node after we've processed it for our
368: * meta-data.
369: */
1.89 schwarze 370: roff_node_delete(man, man->last);
1.34 schwarze 371: }
372:
1.78 schwarze 373: static void
1.34 schwarze 374: post_UC(CHKARGS)
375: {
376: static const char * const bsd_versions[] = {
377: "3rd Berkeley Distribution",
378: "4th Berkeley Distribution",
379: "4.2 Berkeley Distribution",
380: "4.3 Berkeley Distribution",
381: "4.4 Berkeley Distribution",
382: };
383:
384: const char *p, *s;
385:
386: n = n->child;
387:
1.85 schwarze 388: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 389: p = bsd_versions[0];
390: else {
391: s = n->string;
392: if (0 == strcmp(s, "3"))
393: p = bsd_versions[0];
394: else if (0 == strcmp(s, "4"))
395: p = bsd_versions[1];
396: else if (0 == strcmp(s, "5"))
397: p = bsd_versions[2];
398: else if (0 == strcmp(s, "6"))
399: p = bsd_versions[3];
400: else if (0 == strcmp(s, "7"))
401: p = bsd_versions[4];
402: else
403: p = bsd_versions[0];
404: }
405:
1.87 schwarze 406: free(man->meta.os);
407: man->meta.os = mandoc_strdup(p);
1.34 schwarze 408: }
409:
1.78 schwarze 410: static void
1.34 schwarze 411: post_AT(CHKARGS)
412: {
413: static const char * const unix_versions[] = {
414: "7th Edition",
415: "System III",
416: "System V",
417: "System V Release 2",
418: };
419:
1.86 schwarze 420: struct roff_node *nn;
1.34 schwarze 421: const char *p, *s;
422:
423: n = n->child;
424:
1.85 schwarze 425: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 426: p = unix_versions[0];
427: else {
428: s = n->string;
429: if (0 == strcmp(s, "3"))
430: p = unix_versions[0];
431: else if (0 == strcmp(s, "4"))
432: p = unix_versions[1];
433: else if (0 == strcmp(s, "5")) {
434: nn = n->next;
1.85 schwarze 435: if (nn != NULL &&
436: nn->type == ROFFT_TEXT &&
437: nn->string[0] != '\0')
1.34 schwarze 438: p = unix_versions[3];
439: else
440: p = unix_versions[2];
441: } else
442: p = unix_versions[0];
443: }
444:
1.87 schwarze 445: free(man->meta.os);
446: man->meta.os = mandoc_strdup(p);
1.100 schwarze 447: }
448:
449: static void
450: post_in(CHKARGS)
451: {
452: char *s;
453:
454: if (n->parent->tok != MAN_TP ||
455: n->parent->type != ROFFT_HEAD ||
456: n->child == NULL ||
457: *n->child->string == '+' ||
458: *n->child->string == '-')
459: return;
460: mandoc_asprintf(&s, "+%s", n->child->string);
461: free(n->child->string);
462: n->child->string = s;
1.46 schwarze 463: }
464:
1.78 schwarze 465: static void
1.46 schwarze 466: post_vs(CHKARGS)
467: {
468:
1.54 schwarze 469: if (NULL != n->prev)
1.78 schwarze 470: return;
1.54 schwarze 471:
472: switch (n->parent->tok) {
1.63 schwarze 473: case MAN_SH:
474: case MAN_SS:
1.98 schwarze 475: case MAN_PP:
476: case MAN_LP:
477: case MAN_P:
1.67 schwarze 478: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.94 schwarze 479: "%s after %s", roff_name[n->tok],
480: roff_name[n->parent->tok]);
1.54 schwarze 481: /* FALLTHROUGH */
1.90 schwarze 482: case TOKEN_NONE:
1.63 schwarze 483: /*
1.54 schwarze 484: * Don't warn about this because it occurs in pod2man
485: * and would cause considerable (unfixable) warnage.
486: */
1.89 schwarze 487: roff_node_delete(man, n);
1.54 schwarze 488: break;
489: default:
490: break;
491: }
1.34 schwarze 492: }