Annotation of mandoc/man_validate.c, Revision 1.139
1.107 schwarze 1: /* $OpenBSD$ */
1.1 kristaps 2: /*
1.63 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.134 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
1.8 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:
1.138 schwarze 43: static void check_abort(CHKARGS);
1.107 schwarze 44: static void check_par(CHKARGS);
45: static void check_part(CHKARGS);
46: static void check_root(CHKARGS);
47: static void check_text(CHKARGS);
48:
49: static void post_AT(CHKARGS);
50: static void post_IP(CHKARGS);
1.113 schwarze 51: static void post_OP(CHKARGS);
1.139 ! schwarze 52: static void post_SH(CHKARGS);
1.107 schwarze 53: static void post_TH(CHKARGS);
54: static void post_UC(CHKARGS);
55: static void post_UR(CHKARGS);
1.129 schwarze 56: static void post_in(CHKARGS);
1.51 kristaps 57:
1.135 schwarze 58: static const v_check man_valids[MAN_MAX - MAN_TH] = {
1.104 schwarze 59: post_TH, /* TH */
1.139 ! schwarze 60: post_SH, /* SH */
! 61: post_SH, /* SS */
1.104 schwarze 62: NULL, /* TP */
1.136 schwarze 63: NULL, /* TQ */
1.138 schwarze 64: check_abort,/* LP */
1.104 schwarze 65: check_par, /* PP */
1.138 schwarze 66: check_abort,/* P */
1.104 schwarze 67: post_IP, /* IP */
68: NULL, /* HP */
69: NULL, /* SM */
70: NULL, /* SB */
71: NULL, /* BI */
72: NULL, /* IB */
73: NULL, /* BR */
74: NULL, /* RB */
75: NULL, /* R */
76: NULL, /* B */
77: NULL, /* I */
78: NULL, /* IR */
79: NULL, /* RI */
1.121 schwarze 80: NULL, /* nf */
81: NULL, /* fi */
1.104 schwarze 82: NULL, /* RE */
83: check_part, /* RS */
84: NULL, /* DT */
85: post_UC, /* UC */
1.112 schwarze 86: NULL, /* PD */
1.104 schwarze 87: post_AT, /* AT */
1.129 schwarze 88: post_in, /* in */
1.137 schwarze 89: NULL, /* SY */
90: NULL, /* YS */
1.113 schwarze 91: post_OP, /* OP */
1.121 schwarze 92: NULL, /* EX */
93: NULL, /* EE */
1.104 schwarze 94: post_UR, /* UR */
95: NULL, /* UE */
1.132 schwarze 96: post_UR, /* MT */
97: NULL, /* ME */
1.1 kristaps 98: };
99:
100:
1.138 schwarze 101: /* Validate the subtree rooted at man->last. */
1.107 schwarze 102: void
1.121 schwarze 103: man_node_validate(struct roff_man *man)
1.1 kristaps 104: {
1.115 schwarze 105: struct roff_node *n;
1.123 schwarze 106: const v_check *cp;
1.1 kristaps 107:
1.138 schwarze 108: /*
109: * Translate obsolete macros such that later code
110: * does not need to look for them.
111: */
112:
1.104 schwarze 113: n = man->last;
1.138 schwarze 114: switch (n->tok) {
115: case MAN_LP:
116: case MAN_P:
117: n->tok = MAN_PP;
118: break;
119: default:
120: break;
121: }
122:
123: /*
124: * Iterate over all children, recursing into each one
125: * in turn, depth-first.
126: */
127:
1.121 schwarze 128: man->last = man->last->child;
129: while (man->last != NULL) {
130: man_node_validate(man);
131: if (man->last == n)
132: man->last = man->last->child;
133: else
134: man->last = man->last->next;
135: }
1.1 kristaps 136:
1.138 schwarze 137: /* Finally validate the macro itself. */
138:
1.121 schwarze 139: man->last = n;
140: man->next = ROFF_NEXT_SIBLING;
1.104 schwarze 141: switch (n->type) {
1.114 schwarze 142: case ROFFT_TEXT:
1.107 schwarze 143: check_text(man, n);
144: break;
1.114 schwarze 145: case ROFFT_ROOT:
1.107 schwarze 146: check_root(man, n);
147: break;
1.134 schwarze 148: case ROFFT_COMMENT:
1.114 schwarze 149: case ROFFT_EQN:
150: case ROFFT_TBL:
1.107 schwarze 151: break;
1.1 kristaps 152: default:
1.124 schwarze 153: if (n->tok < ROFF_MAX) {
1.139 ! schwarze 154: roff_validate(man);
1.124 schwarze 155: break;
156: }
157: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.135 schwarze 158: cp = man_valids + (n->tok - MAN_TH);
1.107 schwarze 159: if (*cp)
160: (*cp)(man, n);
1.121 schwarze 161: if (man->last == n)
162: man_state(man, n);
1.107 schwarze 163: break;
1.1 kristaps 164: }
165: }
166:
1.107 schwarze 167: static void
1.91 schwarze 168: check_root(CHKARGS)
1.14 kristaps 169: {
1.101 schwarze 170: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.17 kristaps 171:
1.134 schwarze 172: if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1.103 schwarze 173: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
174: n->line, n->pos, NULL);
1.93 schwarze 175: else
176: man->meta.hasbody = 1;
177:
178: if (NULL == man->meta.title) {
1.105 schwarze 179: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.103 schwarze 180: n->line, n->pos, NULL);
1.54 kristaps 181:
1.34 kristaps 182: /*
183: * If a title hasn't been set, do so now (by
184: * implication, date and section also aren't set).
185: */
1.54 kristaps 186:
1.105 schwarze 187: man->meta.title = mandoc_strdup("");
188: man->meta.msec = mandoc_strdup("");
1.88 schwarze 189: man->meta.date = man->quick ? mandoc_strdup("") :
1.128 schwarze 190: mandoc_normdate(man, NULL, n->line, n->pos);
1.34 kristaps 191: }
1.130 schwarze 192:
193: if (man->meta.os_e &&
194: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.131 schwarze 195: mandoc_msg(MANDOCERR_RCS_MISSING, man->parse, 0, 0,
196: man->meta.os_e == MANDOC_OS_OPENBSD ?
197: "(OpenBSD)" : "(NetBSD)");
1.75 kristaps 198: }
199:
1.107 schwarze 200: static void
1.138 schwarze 201: check_abort(CHKARGS)
202: {
203: abort();
204: }
205:
206: static void
1.75 kristaps 207: check_text(CHKARGS)
208: {
209: char *cp, *p;
210:
1.85 schwarze 211: if (MAN_LITERAL & man->flags)
1.107 schwarze 212: return;
1.76 schwarze 213:
214: cp = n->string;
215: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.99 schwarze 216: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
217: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 218: }
219:
1.113 schwarze 220: static void
221: post_OP(CHKARGS)
222: {
223:
1.122 schwarze 224: if (n->child == NULL)
1.113 schwarze 225: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
226: n->line, n->pos, "OP");
1.122 schwarze 227: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 schwarze 228: n = n->child->next->next;
229: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
230: n->line, n->pos, "OP ... %s", n->string);
231: }
1.1 kristaps 232: }
233:
1.107 schwarze 234: static void
1.139 ! schwarze 235: post_SH(CHKARGS)
! 236: {
! 237: struct roff_node *nc;
! 238:
! 239: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
! 240: return;
! 241:
! 242: if (nc->tok == MAN_PP && nc->body->child != NULL) {
! 243: while (nc->body->last != NULL) {
! 244: man->next = ROFF_NEXT_CHILD;
! 245: roff_node_relink(man, nc->body->last);
! 246: man->last = n;
! 247: }
! 248: }
! 249:
! 250: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
! 251: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse,
! 252: nc->line, nc->pos, "%s after %s",
! 253: roff_name[nc->tok], roff_name[n->tok]);
! 254: roff_node_delete(man, nc);
! 255: }
! 256:
! 257: /*
! 258: * Trailing PP is empty, so it is deleted by check_par().
! 259: * Trailing sp is significant.
! 260: */
! 261:
! 262: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
! 263: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse,
! 264: nc->line, nc->pos, "%s at the end of %s",
! 265: roff_name[nc->tok], roff_name[n->tok]);
! 266: roff_node_delete(man, nc);
! 267: }
! 268: }
! 269:
! 270: static void
1.104 schwarze 271: post_UR(CHKARGS)
1.86 schwarze 272: {
1.114 schwarze 273: if (n->type == ROFFT_HEAD && n->child == NULL)
1.133 schwarze 274: mandoc_msg(MANDOCERR_UR_NOHEAD, man->parse,
1.132 schwarze 275: n->line, n->pos, roff_name[n->tok]);
1.107 schwarze 276: check_part(man, n);
1.55 kristaps 277: }
1.16 kristaps 278:
1.107 schwarze 279: static void
1.23 kristaps 280: check_part(CHKARGS)
281: {
282:
1.114 schwarze 283: if (n->type == ROFFT_BODY && n->child == NULL)
1.110 schwarze 284: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.123 schwarze 285: n->line, n->pos, roff_name[n->tok]);
1.23 kristaps 286: }
287:
1.107 schwarze 288: static void
1.17 kristaps 289: check_par(CHKARGS)
290: {
291:
1.59 kristaps 292: switch (n->type) {
1.114 schwarze 293: case ROFFT_BLOCK:
1.122 schwarze 294: if (n->body->child == NULL)
1.118 schwarze 295: roff_node_delete(man, n);
1.59 kristaps 296: break;
1.114 schwarze 297: case ROFFT_BODY:
1.139 ! schwarze 298: if (n->child != NULL &&
! 299: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
! 300: mandoc_vmsg(MANDOCERR_PAR_SKIP,
! 301: man->parse, n->child->line, n->child->pos,
! 302: "%s after %s", roff_name[n->child->tok],
! 303: roff_name[n->tok]);
! 304: roff_node_delete(man, n->child);
! 305: }
1.122 schwarze 306: if (n->child == NULL)
1.95 schwarze 307: mandoc_vmsg(MANDOCERR_PAR_SKIP,
308: man->parse, n->line, n->pos,
1.123 schwarze 309: "%s empty", roff_name[n->tok]);
1.59 kristaps 310: break;
1.114 schwarze 311: case ROFFT_HEAD:
1.122 schwarze 312: if (n->child != NULL)
1.97 schwarze 313: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.123 schwarze 314: man->parse, n->line, n->pos, "%s %s%s",
315: roff_name[n->tok], n->child->string,
1.122 schwarze 316: n->child->next != NULL ? " ..." : "");
1.59 kristaps 317: break;
318: default:
319: break;
320: }
1.17 kristaps 321: }
322:
1.107 schwarze 323: static void
1.83 schwarze 324: post_IP(CHKARGS)
325: {
326:
327: switch (n->type) {
1.114 schwarze 328: case ROFFT_BLOCK:
1.122 schwarze 329: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 330: roff_node_delete(man, n);
1.83 schwarze 331: break;
1.114 schwarze 332: case ROFFT_BODY:
1.122 schwarze 333: if (n->parent->head->child == NULL && n->child == NULL)
1.95 schwarze 334: mandoc_vmsg(MANDOCERR_PAR_SKIP,
335: man->parse, n->line, n->pos,
1.123 schwarze 336: "%s empty", roff_name[n->tok]);
1.83 schwarze 337: break;
338: default:
339: break;
340: }
341: }
1.17 kristaps 342:
1.107 schwarze 343: static void
1.51 kristaps 344: post_TH(CHKARGS)
345: {
1.115 schwarze 346: struct roff_node *nb;
1.60 schwarze 347: const char *p;
1.51 kristaps 348:
1.85 schwarze 349: free(man->meta.title);
350: free(man->meta.vol);
1.116 schwarze 351: free(man->meta.os);
1.85 schwarze 352: free(man->meta.msec);
353: free(man->meta.date);
1.51 kristaps 354:
1.85 schwarze 355: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 356: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 357:
1.92 schwarze 358: nb = n;
359:
1.116 schwarze 360: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 361:
362: n = n->child;
1.60 schwarze 363: if (n && n->string) {
364: for (p = n->string; '\0' != *p; p++) {
365: /* Only warn about this once... */
1.91 schwarze 366: if (isalpha((unsigned char)*p) &&
367: ! isupper((unsigned char)*p)) {
1.102 schwarze 368: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.94 schwarze 369: man->parse, n->line,
370: n->pos + (p - n->string),
1.102 schwarze 371: "TH %s", n->string);
1.60 schwarze 372: break;
373: }
374: }
1.85 schwarze 375: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 376: } else {
1.85 schwarze 377: man->meta.title = mandoc_strdup("");
1.105 schwarze 378: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
379: nb->line, nb->pos, "TH");
380: }
1.51 kristaps 381:
1.116 schwarze 382: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 383:
1.60 schwarze 384: if (n)
385: n = n->next;
386: if (n && n->string)
1.85 schwarze 387: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 388: else {
1.85 schwarze 389: man->meta.msec = mandoc_strdup("");
1.105 schwarze 390: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
391: nb->line, nb->pos, "TH %s", man->meta.title);
392: }
1.51 kristaps 393:
1.116 schwarze 394: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 395:
1.60 schwarze 396: if (n)
397: n = n->next;
1.77 schwarze 398: if (n && n->string && '\0' != n->string[0]) {
1.88 schwarze 399: man->meta.date = man->quick ?
400: mandoc_strdup(n->string) :
1.128 schwarze 401: mandoc_normdate(man, n->string, n->line, n->pos);
1.92 schwarze 402: } else {
1.85 schwarze 403: man->meta.date = mandoc_strdup("");
1.102 schwarze 404: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
405: n ? n->line : nb->line,
406: n ? n->pos : nb->pos, "TH");
1.92 schwarze 407: }
1.51 kristaps 408:
1.116 schwarze 409: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 410:
411: if (n && (n = n->next))
1.116 schwarze 412: man->meta.os = mandoc_strdup(n->string);
1.131 schwarze 413: else if (man->os_s != NULL)
414: man->meta.os = mandoc_strdup(man->os_s);
415: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
416: if (strstr(man->meta.os, "OpenBSD") != NULL)
417: man->meta.os_e = MANDOC_OS_OPENBSD;
418: else if (strstr(man->meta.os, "NetBSD") != NULL)
419: man->meta.os_e = MANDOC_OS_NETBSD;
420: }
1.51 kristaps 421:
1.116 schwarze 422: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 423: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 424:
425: if (n && (n = n->next))
1.85 schwarze 426: man->meta.vol = mandoc_strdup(n->string);
427: else if ('\0' != man->meta.msec[0] &&
428: (NULL != (p = mandoc_a2msec(man->meta.msec))))
429: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 430:
431: if (n != NULL && (n = n->next) != NULL)
432: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
433: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 434:
435: /*
436: * Remove the `TH' node after we've processed it for our
437: * meta-data.
438: */
1.118 schwarze 439: roff_node_delete(man, man->last);
1.51 kristaps 440: }
441:
1.107 schwarze 442: static void
1.51 kristaps 443: post_UC(CHKARGS)
444: {
445: static const char * const bsd_versions[] = {
446: "3rd Berkeley Distribution",
447: "4th Berkeley Distribution",
448: "4.2 Berkeley Distribution",
449: "4.3 Berkeley Distribution",
450: "4.4 Berkeley Distribution",
451: };
452:
453: const char *p, *s;
454:
455: n = n->child;
456:
1.114 schwarze 457: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 458: p = bsd_versions[0];
459: else {
460: s = n->string;
461: if (0 == strcmp(s, "3"))
462: p = bsd_versions[0];
463: else if (0 == strcmp(s, "4"))
464: p = bsd_versions[1];
465: else if (0 == strcmp(s, "5"))
466: p = bsd_versions[2];
467: else if (0 == strcmp(s, "6"))
468: p = bsd_versions[3];
469: else if (0 == strcmp(s, "7"))
470: p = bsd_versions[4];
471: else
472: p = bsd_versions[0];
473: }
474:
1.116 schwarze 475: free(man->meta.os);
476: man->meta.os = mandoc_strdup(p);
1.51 kristaps 477: }
478:
1.107 schwarze 479: static void
1.51 kristaps 480: post_AT(CHKARGS)
481: {
482: static const char * const unix_versions[] = {
483: "7th Edition",
484: "System III",
485: "System V",
486: "System V Release 2",
487: };
488:
1.115 schwarze 489: struct roff_node *nn;
1.51 kristaps 490: const char *p, *s;
491:
492: n = n->child;
493:
1.114 schwarze 494: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 495: p = unix_versions[0];
496: else {
497: s = n->string;
498: if (0 == strcmp(s, "3"))
499: p = unix_versions[0];
500: else if (0 == strcmp(s, "4"))
501: p = unix_versions[1];
502: else if (0 == strcmp(s, "5")) {
503: nn = n->next;
1.114 schwarze 504: if (nn != NULL &&
505: nn->type == ROFFT_TEXT &&
506: nn->string[0] != '\0')
1.51 kristaps 507: p = unix_versions[3];
508: else
509: p = unix_versions[2];
510: } else
511: p = unix_versions[0];
512: }
513:
1.116 schwarze 514: free(man->meta.os);
515: man->meta.os = mandoc_strdup(p);
1.129 schwarze 516: }
517:
518: static void
519: post_in(CHKARGS)
520: {
521: char *s;
522:
523: if (n->parent->tok != MAN_TP ||
524: n->parent->type != ROFFT_HEAD ||
525: n->child == NULL ||
526: *n->child->string == '+' ||
527: *n->child->string == '-')
528: return;
529: mandoc_asprintf(&s, "+%s", n->child->string);
530: free(n->child->string);
531: n->child->string = s;
1.51 kristaps 532: }
CVSweb