Annotation of mandoc/man_validate.c, Revision 1.131
1.107 schwarze 1: /* $OpenBSD$ */
1.1 kristaps 2: /*
1.63 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.123 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
1.8 kristaps 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: *
1.114 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.8 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.114 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.8 kristaps 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: */
1.28 kristaps 18: #include "config.h"
19:
1.1 kristaps 20: #include <sys/types.h>
21:
22: #include <assert.h>
23: #include <ctype.h>
1.16 kristaps 24: #include <errno.h>
25: #include <limits.h>
1.1 kristaps 26: #include <stdarg.h>
27: #include <stdlib.h>
1.46 kristaps 28: #include <string.h>
1.50 kristaps 29: #include <time.h>
1.1 kristaps 30:
1.114 schwarze 31: #include "mandoc_aux.h"
32: #include "mandoc.h"
33: #include "roff.h"
1.66 kristaps 34: #include "man.h"
1.114 schwarze 35: #include "libmandoc.h"
1.118 schwarze 36: #include "roff_int.h"
1.1 kristaps 37: #include "libman.h"
38:
1.117 schwarze 39: #define CHKARGS struct roff_man *man, struct roff_node *n
1.1 kristaps 40:
1.107 schwarze 41: typedef void (*v_check)(CHKARGS);
1.1 kristaps 42:
1.107 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);
49: static void post_IP(CHKARGS);
1.113 schwarze 50: static void post_OP(CHKARGS);
1.107 schwarze 51: static void post_TH(CHKARGS);
52: static void post_UC(CHKARGS);
53: static void post_UR(CHKARGS);
1.129 schwarze 54: static void post_in(CHKARGS);
55: static void post_vs(CHKARGS);
1.51 kristaps 56:
1.123 schwarze 57: static const v_check __man_valids[MAN_MAX - MAN_TH] = {
1.104 schwarze 58: post_TH, /* TH */
59: NULL, /* SH */
60: NULL, /* SS */
61: NULL, /* TP */
62: check_par, /* LP */
63: check_par, /* PP */
64: check_par, /* P */
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 */
1.121 schwarze 78: NULL, /* nf */
79: NULL, /* fi */
1.104 schwarze 80: NULL, /* RE */
81: check_part, /* RS */
82: NULL, /* DT */
83: post_UC, /* UC */
1.112 schwarze 84: NULL, /* PD */
1.104 schwarze 85: post_AT, /* AT */
1.129 schwarze 86: post_in, /* in */
1.113 schwarze 87: post_OP, /* OP */
1.121 schwarze 88: NULL, /* EX */
89: NULL, /* EE */
1.104 schwarze 90: post_UR, /* UR */
91: NULL, /* UE */
1.1 kristaps 92: };
1.123 schwarze 93: static const v_check *man_valids = __man_valids - MAN_TH;
1.1 kristaps 94:
95:
1.107 schwarze 96: void
1.121 schwarze 97: man_node_validate(struct roff_man *man)
1.1 kristaps 98: {
1.115 schwarze 99: struct roff_node *n;
1.123 schwarze 100: const v_check *cp;
1.1 kristaps 101:
1.104 schwarze 102: n = man->last;
1.121 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:
1.121 schwarze 112: man->last = n;
113: man->next = ROFF_NEXT_SIBLING;
1.104 schwarze 114: switch (n->type) {
1.114 schwarze 115: case ROFFT_TEXT:
1.107 schwarze 116: check_text(man, n);
117: break;
1.114 schwarze 118: case ROFFT_ROOT:
1.107 schwarze 119: check_root(man, n);
120: break;
1.114 schwarze 121: case ROFFT_EQN:
122: case ROFFT_TBL:
1.107 schwarze 123: break;
1.1 kristaps 124: default:
1.124 schwarze 125: if (n->tok < ROFF_MAX) {
126: switch (n->tok) {
127: case ROFF_br:
1.127 schwarze 128: case ROFF_sp:
1.124 schwarze 129: post_vs(man, n);
130: break;
131: default:
1.125 schwarze 132: roff_validate(man);
133: break;
1.124 schwarze 134: }
135: break;
136: }
137: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.104 schwarze 138: cp = man_valids + n->tok;
1.107 schwarze 139: if (*cp)
140: (*cp)(man, n);
1.121 schwarze 141: if (man->last == n)
142: man_state(man, n);
1.107 schwarze 143: break;
1.1 kristaps 144: }
145: }
146:
1.107 schwarze 147: static void
1.91 schwarze 148: check_root(CHKARGS)
1.14 kristaps 149: {
1.17 kristaps 150:
1.101 schwarze 151: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.17 kristaps 152:
1.93 schwarze 153: if (NULL == man->first->child)
1.103 schwarze 154: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
155: n->line, n->pos, NULL);
1.93 schwarze 156: else
157: man->meta.hasbody = 1;
158:
159: if (NULL == man->meta.title) {
1.105 schwarze 160: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.103 schwarze 161: n->line, n->pos, NULL);
1.54 kristaps 162:
1.34 kristaps 163: /*
164: * If a title hasn't been set, do so now (by
165: * implication, date and section also aren't set).
166: */
1.54 kristaps 167:
1.105 schwarze 168: man->meta.title = mandoc_strdup("");
169: man->meta.msec = mandoc_strdup("");
1.88 schwarze 170: man->meta.date = man->quick ? mandoc_strdup("") :
1.128 schwarze 171: mandoc_normdate(man, NULL, n->line, n->pos);
1.34 kristaps 172: }
1.130 schwarze 173:
174: if (man->meta.os_e &&
175: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.131 ! schwarze 176: mandoc_msg(MANDOCERR_RCS_MISSING, man->parse, 0, 0,
! 177: man->meta.os_e == MANDOC_OS_OPENBSD ?
! 178: "(OpenBSD)" : "(NetBSD)");
1.75 kristaps 179: }
180:
1.107 schwarze 181: static void
1.75 kristaps 182: check_text(CHKARGS)
183: {
184: char *cp, *p;
185:
1.85 schwarze 186: if (MAN_LITERAL & man->flags)
1.107 schwarze 187: return;
1.76 schwarze 188:
189: cp = n->string;
190: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.99 schwarze 191: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
192: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 193: }
194:
1.113 schwarze 195: static void
196: post_OP(CHKARGS)
197: {
198:
1.122 schwarze 199: if (n->child == NULL)
1.113 schwarze 200: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
201: n->line, n->pos, "OP");
1.122 schwarze 202: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 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:
1.107 schwarze 209: static void
1.104 schwarze 210: post_UR(CHKARGS)
1.86 schwarze 211: {
212:
1.114 schwarze 213: if (n->type == ROFFT_HEAD && n->child == NULL)
1.113 schwarze 214: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
215: n->line, n->pos, "UR");
1.107 schwarze 216: check_part(man, n);
1.55 kristaps 217: }
1.16 kristaps 218:
1.107 schwarze 219: static void
1.23 kristaps 220: check_part(CHKARGS)
221: {
222:
1.114 schwarze 223: if (n->type == ROFFT_BODY && n->child == NULL)
1.110 schwarze 224: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.123 schwarze 225: n->line, n->pos, roff_name[n->tok]);
1.23 kristaps 226: }
227:
1.107 schwarze 228: static void
1.17 kristaps 229: check_par(CHKARGS)
230: {
231:
1.59 kristaps 232: switch (n->type) {
1.114 schwarze 233: case ROFFT_BLOCK:
1.122 schwarze 234: if (n->body->child == NULL)
1.118 schwarze 235: roff_node_delete(man, n);
1.59 kristaps 236: break;
1.114 schwarze 237: case ROFFT_BODY:
1.122 schwarze 238: if (n->child == NULL)
1.95 schwarze 239: mandoc_vmsg(MANDOCERR_PAR_SKIP,
240: man->parse, n->line, n->pos,
1.123 schwarze 241: "%s empty", roff_name[n->tok]);
1.59 kristaps 242: break;
1.114 schwarze 243: case ROFFT_HEAD:
1.122 schwarze 244: if (n->child != NULL)
1.97 schwarze 245: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.123 schwarze 246: man->parse, n->line, n->pos, "%s %s%s",
247: roff_name[n->tok], n->child->string,
1.122 schwarze 248: n->child->next != NULL ? " ..." : "");
1.59 kristaps 249: break;
250: default:
251: break;
252: }
1.17 kristaps 253: }
254:
1.107 schwarze 255: static void
1.83 schwarze 256: post_IP(CHKARGS)
257: {
258:
259: switch (n->type) {
1.114 schwarze 260: case ROFFT_BLOCK:
1.122 schwarze 261: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 262: roff_node_delete(man, n);
1.83 schwarze 263: break;
1.114 schwarze 264: case ROFFT_BODY:
1.122 schwarze 265: if (n->parent->head->child == NULL && n->child == NULL)
1.95 schwarze 266: mandoc_vmsg(MANDOCERR_PAR_SKIP,
267: man->parse, n->line, n->pos,
1.123 schwarze 268: "%s empty", roff_name[n->tok]);
1.83 schwarze 269: break;
270: default:
271: break;
272: }
273: }
1.17 kristaps 274:
1.107 schwarze 275: static void
1.51 kristaps 276: post_TH(CHKARGS)
277: {
1.115 schwarze 278: struct roff_node *nb;
1.60 schwarze 279: const char *p;
1.51 kristaps 280:
1.85 schwarze 281: free(man->meta.title);
282: free(man->meta.vol);
1.116 schwarze 283: free(man->meta.os);
1.85 schwarze 284: free(man->meta.msec);
285: free(man->meta.date);
1.51 kristaps 286:
1.85 schwarze 287: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 288: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 289:
1.92 schwarze 290: nb = n;
291:
1.116 schwarze 292: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 293:
294: n = n->child;
1.60 schwarze 295: if (n && n->string) {
296: for (p = n->string; '\0' != *p; p++) {
297: /* Only warn about this once... */
1.91 schwarze 298: if (isalpha((unsigned char)*p) &&
299: ! isupper((unsigned char)*p)) {
1.102 schwarze 300: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.94 schwarze 301: man->parse, n->line,
302: n->pos + (p - n->string),
1.102 schwarze 303: "TH %s", n->string);
1.60 schwarze 304: break;
305: }
306: }
1.85 schwarze 307: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 308: } else {
1.85 schwarze 309: man->meta.title = mandoc_strdup("");
1.105 schwarze 310: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
311: nb->line, nb->pos, "TH");
312: }
1.51 kristaps 313:
1.116 schwarze 314: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 315:
1.60 schwarze 316: if (n)
317: n = n->next;
318: if (n && n->string)
1.85 schwarze 319: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 320: else {
1.85 schwarze 321: man->meta.msec = mandoc_strdup("");
1.105 schwarze 322: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
323: nb->line, nb->pos, "TH %s", man->meta.title);
324: }
1.51 kristaps 325:
1.116 schwarze 326: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 327:
1.60 schwarze 328: if (n)
329: n = n->next;
1.77 schwarze 330: if (n && n->string && '\0' != n->string[0]) {
1.88 schwarze 331: man->meta.date = man->quick ?
332: mandoc_strdup(n->string) :
1.128 schwarze 333: mandoc_normdate(man, n->string, n->line, n->pos);
1.92 schwarze 334: } else {
1.85 schwarze 335: man->meta.date = mandoc_strdup("");
1.102 schwarze 336: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
337: n ? n->line : nb->line,
338: n ? n->pos : nb->pos, "TH");
1.92 schwarze 339: }
1.51 kristaps 340:
1.116 schwarze 341: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 342:
343: if (n && (n = n->next))
1.116 schwarze 344: man->meta.os = mandoc_strdup(n->string);
1.131 ! schwarze 345: else if (man->os_s != NULL)
! 346: man->meta.os = mandoc_strdup(man->os_s);
! 347: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
! 348: if (strstr(man->meta.os, "OpenBSD") != NULL)
! 349: man->meta.os_e = MANDOC_OS_OPENBSD;
! 350: else if (strstr(man->meta.os, "NetBSD") != NULL)
! 351: man->meta.os_e = MANDOC_OS_NETBSD;
! 352: }
1.51 kristaps 353:
1.116 schwarze 354: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 355: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 356:
357: if (n && (n = n->next))
1.85 schwarze 358: man->meta.vol = mandoc_strdup(n->string);
359: else if ('\0' != man->meta.msec[0] &&
360: (NULL != (p = mandoc_a2msec(man->meta.msec))))
361: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 362:
363: if (n != NULL && (n = n->next) != NULL)
364: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
365: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 366:
367: /*
368: * Remove the `TH' node after we've processed it for our
369: * meta-data.
370: */
1.118 schwarze 371: roff_node_delete(man, man->last);
1.51 kristaps 372: }
373:
1.107 schwarze 374: static void
1.51 kristaps 375: post_UC(CHKARGS)
376: {
377: static const char * const bsd_versions[] = {
378: "3rd Berkeley Distribution",
379: "4th Berkeley Distribution",
380: "4.2 Berkeley Distribution",
381: "4.3 Berkeley Distribution",
382: "4.4 Berkeley Distribution",
383: };
384:
385: const char *p, *s;
386:
387: n = n->child;
388:
1.114 schwarze 389: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 390: p = bsd_versions[0];
391: else {
392: s = n->string;
393: if (0 == strcmp(s, "3"))
394: p = bsd_versions[0];
395: else if (0 == strcmp(s, "4"))
396: p = bsd_versions[1];
397: else if (0 == strcmp(s, "5"))
398: p = bsd_versions[2];
399: else if (0 == strcmp(s, "6"))
400: p = bsd_versions[3];
401: else if (0 == strcmp(s, "7"))
402: p = bsd_versions[4];
403: else
404: p = bsd_versions[0];
405: }
406:
1.116 schwarze 407: free(man->meta.os);
408: man->meta.os = mandoc_strdup(p);
1.51 kristaps 409: }
410:
1.107 schwarze 411: static void
1.51 kristaps 412: post_AT(CHKARGS)
413: {
414: static const char * const unix_versions[] = {
415: "7th Edition",
416: "System III",
417: "System V",
418: "System V Release 2",
419: };
420:
1.115 schwarze 421: struct roff_node *nn;
1.51 kristaps 422: const char *p, *s;
423:
424: n = n->child;
425:
1.114 schwarze 426: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 427: p = unix_versions[0];
428: else {
429: s = n->string;
430: if (0 == strcmp(s, "3"))
431: p = unix_versions[0];
432: else if (0 == strcmp(s, "4"))
433: p = unix_versions[1];
434: else if (0 == strcmp(s, "5")) {
435: nn = n->next;
1.114 schwarze 436: if (nn != NULL &&
437: nn->type == ROFFT_TEXT &&
438: nn->string[0] != '\0')
1.51 kristaps 439: p = unix_versions[3];
440: else
441: p = unix_versions[2];
442: } else
443: p = unix_versions[0];
444: }
445:
1.116 schwarze 446: free(man->meta.os);
447: man->meta.os = mandoc_strdup(p);
1.129 schwarze 448: }
449:
450: static void
451: post_in(CHKARGS)
452: {
453: char *s;
454:
455: if (n->parent->tok != MAN_TP ||
456: n->parent->type != ROFFT_HEAD ||
457: n->child == NULL ||
458: *n->child->string == '+' ||
459: *n->child->string == '-')
460: return;
461: mandoc_asprintf(&s, "+%s", n->child->string);
462: free(n->child->string);
463: n->child->string = s;
1.70 kristaps 464: }
465:
1.107 schwarze 466: static void
1.70 kristaps 467: post_vs(CHKARGS)
468: {
469:
1.82 schwarze 470: if (NULL != n->prev)
1.107 schwarze 471: return;
1.82 schwarze 472:
473: switch (n->parent->tok) {
1.91 schwarze 474: case MAN_SH:
475: case MAN_SS:
1.127 schwarze 476: case MAN_PP:
477: case MAN_LP:
478: case MAN_P:
1.95 schwarze 479: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.123 schwarze 480: "%s after %s", roff_name[n->tok],
481: roff_name[n->parent->tok]);
1.82 schwarze 482: /* FALLTHROUGH */
1.119 schwarze 483: case TOKEN_NONE:
1.91 schwarze 484: /*
1.82 schwarze 485: * Don't warn about this because it occurs in pod2man
486: * and would cause considerable (unfixable) warnage.
487: */
1.118 schwarze 488: roff_node_delete(man, n);
1.82 schwarze 489: break;
490: default:
491: break;
492: }
1.51 kristaps 493: }
CVSweb