Annotation of mandoc/man_validate.c, Revision 1.132
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 */
1.132 ! schwarze 92: post_UR, /* MT */
! 93: NULL, /* ME */
1.1 kristaps 94: };
1.123 schwarze 95: static const v_check *man_valids = __man_valids - MAN_TH;
1.1 kristaps 96:
97:
1.107 schwarze 98: void
1.121 schwarze 99: man_node_validate(struct roff_man *man)
1.1 kristaps 100: {
1.115 schwarze 101: struct roff_node *n;
1.123 schwarze 102: const v_check *cp;
1.1 kristaps 103:
1.104 schwarze 104: n = man->last;
1.121 schwarze 105: man->last = man->last->child;
106: while (man->last != NULL) {
107: man_node_validate(man);
108: if (man->last == n)
109: man->last = man->last->child;
110: else
111: man->last = man->last->next;
112: }
1.1 kristaps 113:
1.121 schwarze 114: man->last = n;
115: man->next = ROFF_NEXT_SIBLING;
1.104 schwarze 116: switch (n->type) {
1.114 schwarze 117: case ROFFT_TEXT:
1.107 schwarze 118: check_text(man, n);
119: break;
1.114 schwarze 120: case ROFFT_ROOT:
1.107 schwarze 121: check_root(man, n);
122: break;
1.114 schwarze 123: case ROFFT_EQN:
124: case ROFFT_TBL:
1.107 schwarze 125: break;
1.1 kristaps 126: default:
1.124 schwarze 127: if (n->tok < ROFF_MAX) {
128: switch (n->tok) {
129: case ROFF_br:
1.127 schwarze 130: case ROFF_sp:
1.124 schwarze 131: post_vs(man, n);
132: break;
133: default:
1.125 schwarze 134: roff_validate(man);
135: break;
1.124 schwarze 136: }
137: break;
138: }
139: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.104 schwarze 140: cp = man_valids + n->tok;
1.107 schwarze 141: if (*cp)
142: (*cp)(man, n);
1.121 schwarze 143: if (man->last == n)
144: man_state(man, n);
1.107 schwarze 145: break;
1.1 kristaps 146: }
147: }
148:
1.107 schwarze 149: static void
1.91 schwarze 150: check_root(CHKARGS)
1.14 kristaps 151: {
1.17 kristaps 152:
1.101 schwarze 153: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.17 kristaps 154:
1.93 schwarze 155: if (NULL == man->first->child)
1.103 schwarze 156: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
157: n->line, n->pos, NULL);
1.93 schwarze 158: else
159: man->meta.hasbody = 1;
160:
161: if (NULL == man->meta.title) {
1.105 schwarze 162: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.103 schwarze 163: n->line, n->pos, NULL);
1.54 kristaps 164:
1.34 kristaps 165: /*
166: * If a title hasn't been set, do so now (by
167: * implication, date and section also aren't set).
168: */
1.54 kristaps 169:
1.105 schwarze 170: man->meta.title = mandoc_strdup("");
171: man->meta.msec = mandoc_strdup("");
1.88 schwarze 172: man->meta.date = man->quick ? mandoc_strdup("") :
1.128 schwarze 173: mandoc_normdate(man, NULL, n->line, n->pos);
1.34 kristaps 174: }
1.130 schwarze 175:
176: if (man->meta.os_e &&
177: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.131 schwarze 178: mandoc_msg(MANDOCERR_RCS_MISSING, man->parse, 0, 0,
179: man->meta.os_e == MANDOC_OS_OPENBSD ?
180: "(OpenBSD)" : "(NetBSD)");
1.75 kristaps 181: }
182:
1.107 schwarze 183: static void
1.75 kristaps 184: check_text(CHKARGS)
185: {
186: char *cp, *p;
187:
1.85 schwarze 188: if (MAN_LITERAL & man->flags)
1.107 schwarze 189: return;
1.76 schwarze 190:
191: cp = n->string;
192: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.99 schwarze 193: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
194: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 195: }
196:
1.113 schwarze 197: static void
198: post_OP(CHKARGS)
199: {
200:
1.122 schwarze 201: if (n->child == NULL)
1.113 schwarze 202: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
203: n->line, n->pos, "OP");
1.122 schwarze 204: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 schwarze 205: n = n->child->next->next;
206: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
207: n->line, n->pos, "OP ... %s", n->string);
208: }
1.1 kristaps 209: }
210:
1.107 schwarze 211: static void
1.104 schwarze 212: post_UR(CHKARGS)
1.86 schwarze 213: {
214:
1.114 schwarze 215: if (n->type == ROFFT_HEAD && n->child == NULL)
1.113 schwarze 216: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
1.132 ! schwarze 217: n->line, n->pos, roff_name[n->tok]);
1.107 schwarze 218: check_part(man, n);
1.55 kristaps 219: }
1.16 kristaps 220:
1.107 schwarze 221: static void
1.23 kristaps 222: check_part(CHKARGS)
223: {
224:
1.114 schwarze 225: if (n->type == ROFFT_BODY && n->child == NULL)
1.110 schwarze 226: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.123 schwarze 227: n->line, n->pos, roff_name[n->tok]);
1.23 kristaps 228: }
229:
1.107 schwarze 230: static void
1.17 kristaps 231: check_par(CHKARGS)
232: {
233:
1.59 kristaps 234: switch (n->type) {
1.114 schwarze 235: case ROFFT_BLOCK:
1.122 schwarze 236: if (n->body->child == NULL)
1.118 schwarze 237: roff_node_delete(man, n);
1.59 kristaps 238: break;
1.114 schwarze 239: case ROFFT_BODY:
1.122 schwarze 240: if (n->child == NULL)
1.95 schwarze 241: mandoc_vmsg(MANDOCERR_PAR_SKIP,
242: man->parse, n->line, n->pos,
1.123 schwarze 243: "%s empty", roff_name[n->tok]);
1.59 kristaps 244: break;
1.114 schwarze 245: case ROFFT_HEAD:
1.122 schwarze 246: if (n->child != NULL)
1.97 schwarze 247: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.123 schwarze 248: man->parse, n->line, n->pos, "%s %s%s",
249: roff_name[n->tok], n->child->string,
1.122 schwarze 250: n->child->next != NULL ? " ..." : "");
1.59 kristaps 251: break;
252: default:
253: break;
254: }
1.17 kristaps 255: }
256:
1.107 schwarze 257: static void
1.83 schwarze 258: post_IP(CHKARGS)
259: {
260:
261: switch (n->type) {
1.114 schwarze 262: case ROFFT_BLOCK:
1.122 schwarze 263: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 264: roff_node_delete(man, n);
1.83 schwarze 265: break;
1.114 schwarze 266: case ROFFT_BODY:
1.122 schwarze 267: if (n->parent->head->child == NULL && n->child == NULL)
1.95 schwarze 268: mandoc_vmsg(MANDOCERR_PAR_SKIP,
269: man->parse, n->line, n->pos,
1.123 schwarze 270: "%s empty", roff_name[n->tok]);
1.83 schwarze 271: break;
272: default:
273: break;
274: }
275: }
1.17 kristaps 276:
1.107 schwarze 277: static void
1.51 kristaps 278: post_TH(CHKARGS)
279: {
1.115 schwarze 280: struct roff_node *nb;
1.60 schwarze 281: const char *p;
1.51 kristaps 282:
1.85 schwarze 283: free(man->meta.title);
284: free(man->meta.vol);
1.116 schwarze 285: free(man->meta.os);
1.85 schwarze 286: free(man->meta.msec);
287: free(man->meta.date);
1.51 kristaps 288:
1.85 schwarze 289: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 290: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 291:
1.92 schwarze 292: nb = n;
293:
1.116 schwarze 294: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 295:
296: n = n->child;
1.60 schwarze 297: if (n && n->string) {
298: for (p = n->string; '\0' != *p; p++) {
299: /* Only warn about this once... */
1.91 schwarze 300: if (isalpha((unsigned char)*p) &&
301: ! isupper((unsigned char)*p)) {
1.102 schwarze 302: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.94 schwarze 303: man->parse, n->line,
304: n->pos + (p - n->string),
1.102 schwarze 305: "TH %s", n->string);
1.60 schwarze 306: break;
307: }
308: }
1.85 schwarze 309: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 310: } else {
1.85 schwarze 311: man->meta.title = mandoc_strdup("");
1.105 schwarze 312: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
313: nb->line, nb->pos, "TH");
314: }
1.51 kristaps 315:
1.116 schwarze 316: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 317:
1.60 schwarze 318: if (n)
319: n = n->next;
320: if (n && n->string)
1.85 schwarze 321: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 322: else {
1.85 schwarze 323: man->meta.msec = mandoc_strdup("");
1.105 schwarze 324: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
325: nb->line, nb->pos, "TH %s", man->meta.title);
326: }
1.51 kristaps 327:
1.116 schwarze 328: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 329:
1.60 schwarze 330: if (n)
331: n = n->next;
1.77 schwarze 332: if (n && n->string && '\0' != n->string[0]) {
1.88 schwarze 333: man->meta.date = man->quick ?
334: mandoc_strdup(n->string) :
1.128 schwarze 335: mandoc_normdate(man, n->string, n->line, n->pos);
1.92 schwarze 336: } else {
1.85 schwarze 337: man->meta.date = mandoc_strdup("");
1.102 schwarze 338: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
339: n ? n->line : nb->line,
340: n ? n->pos : nb->pos, "TH");
1.92 schwarze 341: }
1.51 kristaps 342:
1.116 schwarze 343: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 344:
345: if (n && (n = n->next))
1.116 schwarze 346: man->meta.os = mandoc_strdup(n->string);
1.131 schwarze 347: else if (man->os_s != NULL)
348: man->meta.os = mandoc_strdup(man->os_s);
349: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
350: if (strstr(man->meta.os, "OpenBSD") != NULL)
351: man->meta.os_e = MANDOC_OS_OPENBSD;
352: else if (strstr(man->meta.os, "NetBSD") != NULL)
353: man->meta.os_e = MANDOC_OS_NETBSD;
354: }
1.51 kristaps 355:
1.116 schwarze 356: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 357: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 358:
359: if (n && (n = n->next))
1.85 schwarze 360: man->meta.vol = mandoc_strdup(n->string);
361: else if ('\0' != man->meta.msec[0] &&
362: (NULL != (p = mandoc_a2msec(man->meta.msec))))
363: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 364:
365: if (n != NULL && (n = n->next) != NULL)
366: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
367: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 368:
369: /*
370: * Remove the `TH' node after we've processed it for our
371: * meta-data.
372: */
1.118 schwarze 373: roff_node_delete(man, man->last);
1.51 kristaps 374: }
375:
1.107 schwarze 376: static void
1.51 kristaps 377: post_UC(CHKARGS)
378: {
379: static const char * const bsd_versions[] = {
380: "3rd Berkeley Distribution",
381: "4th Berkeley Distribution",
382: "4.2 Berkeley Distribution",
383: "4.3 Berkeley Distribution",
384: "4.4 Berkeley Distribution",
385: };
386:
387: const char *p, *s;
388:
389: n = n->child;
390:
1.114 schwarze 391: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 392: p = bsd_versions[0];
393: else {
394: s = n->string;
395: if (0 == strcmp(s, "3"))
396: p = bsd_versions[0];
397: else if (0 == strcmp(s, "4"))
398: p = bsd_versions[1];
399: else if (0 == strcmp(s, "5"))
400: p = bsd_versions[2];
401: else if (0 == strcmp(s, "6"))
402: p = bsd_versions[3];
403: else if (0 == strcmp(s, "7"))
404: p = bsd_versions[4];
405: else
406: p = bsd_versions[0];
407: }
408:
1.116 schwarze 409: free(man->meta.os);
410: man->meta.os = mandoc_strdup(p);
1.51 kristaps 411: }
412:
1.107 schwarze 413: static void
1.51 kristaps 414: post_AT(CHKARGS)
415: {
416: static const char * const unix_versions[] = {
417: "7th Edition",
418: "System III",
419: "System V",
420: "System V Release 2",
421: };
422:
1.115 schwarze 423: struct roff_node *nn;
1.51 kristaps 424: const char *p, *s;
425:
426: n = n->child;
427:
1.114 schwarze 428: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 429: p = unix_versions[0];
430: else {
431: s = n->string;
432: if (0 == strcmp(s, "3"))
433: p = unix_versions[0];
434: else if (0 == strcmp(s, "4"))
435: p = unix_versions[1];
436: else if (0 == strcmp(s, "5")) {
437: nn = n->next;
1.114 schwarze 438: if (nn != NULL &&
439: nn->type == ROFFT_TEXT &&
440: nn->string[0] != '\0')
1.51 kristaps 441: p = unix_versions[3];
442: else
443: p = unix_versions[2];
444: } else
445: p = unix_versions[0];
446: }
447:
1.116 schwarze 448: free(man->meta.os);
449: man->meta.os = mandoc_strdup(p);
1.129 schwarze 450: }
451:
452: static void
453: post_in(CHKARGS)
454: {
455: char *s;
456:
457: if (n->parent->tok != MAN_TP ||
458: n->parent->type != ROFFT_HEAD ||
459: n->child == NULL ||
460: *n->child->string == '+' ||
461: *n->child->string == '-')
462: return;
463: mandoc_asprintf(&s, "+%s", n->child->string);
464: free(n->child->string);
465: n->child->string = s;
1.70 kristaps 466: }
467:
1.107 schwarze 468: static void
1.70 kristaps 469: post_vs(CHKARGS)
470: {
471:
1.82 schwarze 472: if (NULL != n->prev)
1.107 schwarze 473: return;
1.82 schwarze 474:
475: switch (n->parent->tok) {
1.91 schwarze 476: case MAN_SH:
477: case MAN_SS:
1.127 schwarze 478: case MAN_PP:
479: case MAN_LP:
480: case MAN_P:
1.95 schwarze 481: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.123 schwarze 482: "%s after %s", roff_name[n->tok],
483: roff_name[n->parent->tok]);
1.82 schwarze 484: /* FALLTHROUGH */
1.119 schwarze 485: case TOKEN_NONE:
1.91 schwarze 486: /*
1.82 schwarze 487: * Don't warn about this because it occurs in pod2man
488: * and would cause considerable (unfixable) warnage.
489: */
1.118 schwarze 490: roff_node_delete(man, n);
1.82 schwarze 491: break;
492: default:
493: break;
494: }
1.51 kristaps 495: }
CVSweb