Annotation of mandoc/man_validate.c, Revision 1.143
1.143 ! schwarze 1: /* $Id: man_validate.c,v 1.142 2018/12/16 00:21:05 schwarze Exp $ */
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>
1.140 schwarze 27: #include <stdio.h>
1.1 kristaps 28: #include <stdlib.h>
1.46 kristaps 29: #include <string.h>
1.50 kristaps 30: #include <time.h>
1.1 kristaps 31:
1.114 schwarze 32: #include "mandoc_aux.h"
33: #include "mandoc.h"
34: #include "roff.h"
1.66 kristaps 35: #include "man.h"
1.114 schwarze 36: #include "libmandoc.h"
1.118 schwarze 37: #include "roff_int.h"
1.1 kristaps 38: #include "libman.h"
39:
1.117 schwarze 40: #define CHKARGS struct roff_man *man, struct roff_node *n
1.1 kristaps 41:
1.107 schwarze 42: typedef void (*v_check)(CHKARGS);
1.1 kristaps 43:
1.138 schwarze 44: static void check_abort(CHKARGS);
1.107 schwarze 45: static void check_par(CHKARGS);
46: static void check_part(CHKARGS);
47: static void check_root(CHKARGS);
48: static void check_text(CHKARGS);
49:
50: static void post_AT(CHKARGS);
51: static void post_IP(CHKARGS);
1.113 schwarze 52: static void post_OP(CHKARGS);
1.139 schwarze 53: static void post_SH(CHKARGS);
1.107 schwarze 54: static void post_TH(CHKARGS);
55: static void post_UC(CHKARGS);
56: static void post_UR(CHKARGS);
1.129 schwarze 57: static void post_in(CHKARGS);
1.51 kristaps 58:
1.135 schwarze 59: static const v_check man_valids[MAN_MAX - MAN_TH] = {
1.104 schwarze 60: post_TH, /* TH */
1.139 schwarze 61: post_SH, /* SH */
62: post_SH, /* SS */
1.104 schwarze 63: NULL, /* TP */
1.136 schwarze 64: NULL, /* TQ */
1.138 schwarze 65: check_abort,/* LP */
1.104 schwarze 66: check_par, /* PP */
1.138 schwarze 67: check_abort,/* P */
1.104 schwarze 68: post_IP, /* IP */
69: NULL, /* HP */
70: NULL, /* SM */
71: NULL, /* SB */
72: NULL, /* BI */
73: NULL, /* IB */
74: NULL, /* BR */
75: NULL, /* RB */
76: NULL, /* R */
77: NULL, /* B */
78: NULL, /* I */
79: NULL, /* IR */
80: NULL, /* RI */
1.121 schwarze 81: NULL, /* nf */
82: NULL, /* fi */
1.104 schwarze 83: NULL, /* RE */
84: check_part, /* RS */
85: NULL, /* DT */
86: post_UC, /* UC */
1.112 schwarze 87: NULL, /* PD */
1.104 schwarze 88: post_AT, /* AT */
1.129 schwarze 89: post_in, /* in */
1.137 schwarze 90: NULL, /* SY */
91: NULL, /* YS */
1.113 schwarze 92: post_OP, /* OP */
1.121 schwarze 93: NULL, /* EX */
94: NULL, /* EE */
1.104 schwarze 95: post_UR, /* UR */
96: NULL, /* UE */
1.132 schwarze 97: post_UR, /* MT */
98: NULL, /* ME */
1.1 kristaps 99: };
100:
101:
1.138 schwarze 102: /* Validate the subtree rooted at man->last. */
1.107 schwarze 103: void
1.143 ! schwarze 104: man_validate(struct roff_man *man)
1.1 kristaps 105: {
1.115 schwarze 106: struct roff_node *n;
1.123 schwarze 107: const v_check *cp;
1.1 kristaps 108:
1.138 schwarze 109: /*
110: * Translate obsolete macros such that later code
111: * does not need to look for them.
112: */
113:
1.104 schwarze 114: n = man->last;
1.138 schwarze 115: switch (n->tok) {
116: case MAN_LP:
117: case MAN_P:
118: n->tok = MAN_PP;
119: break;
120: default:
121: break;
122: }
123:
124: /*
125: * Iterate over all children, recursing into each one
126: * in turn, depth-first.
127: */
128:
1.121 schwarze 129: man->last = man->last->child;
130: while (man->last != NULL) {
1.143 ! schwarze 131: man_validate(man);
1.121 schwarze 132: if (man->last == n)
133: man->last = man->last->child;
134: else
135: man->last = man->last->next;
136: }
1.1 kristaps 137:
1.138 schwarze 138: /* Finally validate the macro itself. */
139:
1.121 schwarze 140: man->last = n;
141: man->next = ROFF_NEXT_SIBLING;
1.104 schwarze 142: switch (n->type) {
1.114 schwarze 143: case ROFFT_TEXT:
1.107 schwarze 144: check_text(man, n);
145: break;
1.114 schwarze 146: case ROFFT_ROOT:
1.107 schwarze 147: check_root(man, n);
148: break;
1.134 schwarze 149: case ROFFT_COMMENT:
1.114 schwarze 150: case ROFFT_EQN:
151: case ROFFT_TBL:
1.107 schwarze 152: break;
1.1 kristaps 153: default:
1.124 schwarze 154: if (n->tok < ROFF_MAX) {
1.139 schwarze 155: roff_validate(man);
1.124 schwarze 156: break;
157: }
158: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.135 schwarze 159: cp = man_valids + (n->tok - MAN_TH);
1.107 schwarze 160: if (*cp)
161: (*cp)(man, n);
1.121 schwarze 162: if (man->last == n)
163: man_state(man, n);
1.107 schwarze 164: break;
1.1 kristaps 165: }
166: }
167:
1.107 schwarze 168: static void
1.91 schwarze 169: check_root(CHKARGS)
1.14 kristaps 170: {
1.101 schwarze 171: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.17 kristaps 172:
1.134 schwarze 173: if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1.141 schwarze 174: mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
1.93 schwarze 175: else
176: man->meta.hasbody = 1;
177:
178: if (NULL == man->meta.title) {
1.141 schwarze 179: mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
1.54 kristaps 180:
1.34 kristaps 181: /*
182: * If a title hasn't been set, do so now (by
183: * implication, date and section also aren't set).
184: */
1.54 kristaps 185:
1.105 schwarze 186: man->meta.title = mandoc_strdup("");
187: man->meta.msec = mandoc_strdup("");
1.88 schwarze 188: man->meta.date = man->quick ? mandoc_strdup("") :
1.128 schwarze 189: mandoc_normdate(man, NULL, n->line, n->pos);
1.34 kristaps 190: }
1.130 schwarze 191:
192: if (man->meta.os_e &&
193: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.141 schwarze 194: mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
1.131 schwarze 195: man->meta.os_e == MANDOC_OS_OPENBSD ?
196: "(OpenBSD)" : "(NetBSD)");
1.75 kristaps 197: }
198:
1.107 schwarze 199: static void
1.138 schwarze 200: check_abort(CHKARGS)
201: {
202: abort();
203: }
204:
205: static void
1.75 kristaps 206: check_text(CHKARGS)
207: {
208: char *cp, *p;
209:
1.85 schwarze 210: if (MAN_LITERAL & man->flags)
1.107 schwarze 211: return;
1.76 schwarze 212:
213: cp = n->string;
214: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.141 schwarze 215: mandoc_msg(MANDOCERR_FI_TAB,
216: n->line, n->pos + (int)(p - cp), NULL);
1.1 kristaps 217: }
218:
1.113 schwarze 219: static void
220: post_OP(CHKARGS)
221: {
222:
1.122 schwarze 223: if (n->child == NULL)
1.141 schwarze 224: mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
1.122 schwarze 225: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 schwarze 226: n = n->child->next->next;
1.141 schwarze 227: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.113 schwarze 228: n->line, n->pos, "OP ... %s", n->string);
229: }
1.1 kristaps 230: }
231:
1.107 schwarze 232: static void
1.139 schwarze 233: post_SH(CHKARGS)
234: {
235: struct roff_node *nc;
236:
237: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
238: return;
239:
240: if (nc->tok == MAN_PP && nc->body->child != NULL) {
241: while (nc->body->last != NULL) {
242: man->next = ROFF_NEXT_CHILD;
243: roff_node_relink(man, nc->body->last);
244: man->last = n;
245: }
246: }
247:
248: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
1.141 schwarze 249: mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
250: "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
1.139 schwarze 251: roff_node_delete(man, nc);
252: }
253:
254: /*
255: * Trailing PP is empty, so it is deleted by check_par().
256: * Trailing sp is significant.
257: */
258:
259: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
1.141 schwarze 260: mandoc_msg(MANDOCERR_PAR_SKIP,
1.139 schwarze 261: nc->line, nc->pos, "%s at the end of %s",
262: roff_name[nc->tok], roff_name[n->tok]);
263: roff_node_delete(man, nc);
264: }
265: }
266:
267: static void
1.104 schwarze 268: post_UR(CHKARGS)
1.86 schwarze 269: {
1.114 schwarze 270: if (n->type == ROFFT_HEAD && n->child == NULL)
1.141 schwarze 271: mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
272: "%s", roff_name[n->tok]);
1.107 schwarze 273: check_part(man, n);
1.55 kristaps 274: }
1.16 kristaps 275:
1.107 schwarze 276: static void
1.23 kristaps 277: check_part(CHKARGS)
278: {
279:
1.114 schwarze 280: if (n->type == ROFFT_BODY && n->child == NULL)
1.141 schwarze 281: mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
282: "%s", roff_name[n->tok]);
1.23 kristaps 283: }
284:
1.107 schwarze 285: static void
1.17 kristaps 286: check_par(CHKARGS)
287: {
288:
1.59 kristaps 289: switch (n->type) {
1.114 schwarze 290: case ROFFT_BLOCK:
1.122 schwarze 291: if (n->body->child == NULL)
1.118 schwarze 292: roff_node_delete(man, n);
1.59 kristaps 293: break;
1.114 schwarze 294: case ROFFT_BODY:
1.139 schwarze 295: if (n->child != NULL &&
296: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
1.141 schwarze 297: mandoc_msg(MANDOCERR_PAR_SKIP,
298: n->child->line, n->child->pos,
1.139 schwarze 299: "%s after %s", roff_name[n->child->tok],
300: roff_name[n->tok]);
301: roff_node_delete(man, n->child);
302: }
1.122 schwarze 303: if (n->child == NULL)
1.141 schwarze 304: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.123 schwarze 305: "%s empty", roff_name[n->tok]);
1.59 kristaps 306: break;
1.114 schwarze 307: case ROFFT_HEAD:
1.122 schwarze 308: if (n->child != NULL)
1.141 schwarze 309: mandoc_msg(MANDOCERR_ARG_SKIP,
310: n->line, n->pos, "%s %s%s",
1.123 schwarze 311: roff_name[n->tok], n->child->string,
1.122 schwarze 312: n->child->next != NULL ? " ..." : "");
1.59 kristaps 313: break;
314: default:
315: break;
316: }
1.17 kristaps 317: }
318:
1.107 schwarze 319: static void
1.83 schwarze 320: post_IP(CHKARGS)
321: {
322:
323: switch (n->type) {
1.114 schwarze 324: case ROFFT_BLOCK:
1.122 schwarze 325: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 326: roff_node_delete(man, n);
1.83 schwarze 327: break;
1.114 schwarze 328: case ROFFT_BODY:
1.122 schwarze 329: if (n->parent->head->child == NULL && n->child == NULL)
1.141 schwarze 330: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.123 schwarze 331: "%s empty", roff_name[n->tok]);
1.83 schwarze 332: break;
333: default:
334: break;
335: }
336: }
1.17 kristaps 337:
1.107 schwarze 338: static void
1.51 kristaps 339: post_TH(CHKARGS)
340: {
1.115 schwarze 341: struct roff_node *nb;
1.60 schwarze 342: const char *p;
1.51 kristaps 343:
1.85 schwarze 344: free(man->meta.title);
345: free(man->meta.vol);
1.116 schwarze 346: free(man->meta.os);
1.85 schwarze 347: free(man->meta.msec);
348: free(man->meta.date);
1.51 kristaps 349:
1.85 schwarze 350: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 351: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 352:
1.92 schwarze 353: nb = n;
354:
1.116 schwarze 355: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 356:
357: n = n->child;
1.60 schwarze 358: if (n && n->string) {
359: for (p = n->string; '\0' != *p; p++) {
360: /* Only warn about this once... */
1.91 schwarze 361: if (isalpha((unsigned char)*p) &&
362: ! isupper((unsigned char)*p)) {
1.141 schwarze 363: mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
364: n->pos + (int)(p - n->string),
1.102 schwarze 365: "TH %s", n->string);
1.60 schwarze 366: break;
367: }
368: }
1.85 schwarze 369: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 370: } else {
1.85 schwarze 371: man->meta.title = mandoc_strdup("");
1.141 schwarze 372: mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
1.105 schwarze 373: }
1.51 kristaps 374:
1.116 schwarze 375: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 376:
1.60 schwarze 377: if (n)
378: n = n->next;
379: if (n && n->string)
1.85 schwarze 380: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 381: else {
1.85 schwarze 382: man->meta.msec = mandoc_strdup("");
1.141 schwarze 383: mandoc_msg(MANDOCERR_MSEC_MISSING,
1.105 schwarze 384: nb->line, nb->pos, "TH %s", man->meta.title);
385: }
1.51 kristaps 386:
1.116 schwarze 387: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 388:
1.60 schwarze 389: if (n)
390: n = n->next;
1.77 schwarze 391: if (n && n->string && '\0' != n->string[0]) {
1.88 schwarze 392: man->meta.date = man->quick ?
393: mandoc_strdup(n->string) :
1.128 schwarze 394: mandoc_normdate(man, n->string, n->line, n->pos);
1.92 schwarze 395: } else {
1.85 schwarze 396: man->meta.date = mandoc_strdup("");
1.141 schwarze 397: mandoc_msg(MANDOCERR_DATE_MISSING,
1.102 schwarze 398: n ? n->line : nb->line,
399: n ? n->pos : nb->pos, "TH");
1.92 schwarze 400: }
1.51 kristaps 401:
1.116 schwarze 402: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 403:
404: if (n && (n = n->next))
1.116 schwarze 405: man->meta.os = mandoc_strdup(n->string);
1.131 schwarze 406: else if (man->os_s != NULL)
407: man->meta.os = mandoc_strdup(man->os_s);
408: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
409: if (strstr(man->meta.os, "OpenBSD") != NULL)
410: man->meta.os_e = MANDOC_OS_OPENBSD;
411: else if (strstr(man->meta.os, "NetBSD") != NULL)
412: man->meta.os_e = MANDOC_OS_NETBSD;
413: }
1.51 kristaps 414:
1.116 schwarze 415: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 416: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 417:
418: if (n && (n = n->next))
1.85 schwarze 419: man->meta.vol = mandoc_strdup(n->string);
420: else if ('\0' != man->meta.msec[0] &&
421: (NULL != (p = mandoc_a2msec(man->meta.msec))))
422: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 423:
424: if (n != NULL && (n = n->next) != NULL)
1.141 schwarze 425: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.113 schwarze 426: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 427:
428: /*
429: * Remove the `TH' node after we've processed it for our
430: * meta-data.
431: */
1.118 schwarze 432: roff_node_delete(man, man->last);
1.51 kristaps 433: }
434:
1.107 schwarze 435: static void
1.51 kristaps 436: post_UC(CHKARGS)
437: {
438: static const char * const bsd_versions[] = {
439: "3rd Berkeley Distribution",
440: "4th Berkeley Distribution",
441: "4.2 Berkeley Distribution",
442: "4.3 Berkeley Distribution",
443: "4.4 Berkeley Distribution",
444: };
445:
446: const char *p, *s;
447:
448: n = n->child;
449:
1.114 schwarze 450: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 451: p = bsd_versions[0];
452: else {
453: s = n->string;
454: if (0 == strcmp(s, "3"))
455: p = bsd_versions[0];
456: else if (0 == strcmp(s, "4"))
457: p = bsd_versions[1];
458: else if (0 == strcmp(s, "5"))
459: p = bsd_versions[2];
460: else if (0 == strcmp(s, "6"))
461: p = bsd_versions[3];
462: else if (0 == strcmp(s, "7"))
463: p = bsd_versions[4];
464: else
465: p = bsd_versions[0];
466: }
467:
1.116 schwarze 468: free(man->meta.os);
469: man->meta.os = mandoc_strdup(p);
1.51 kristaps 470: }
471:
1.107 schwarze 472: static void
1.51 kristaps 473: post_AT(CHKARGS)
474: {
475: static const char * const unix_versions[] = {
476: "7th Edition",
477: "System III",
478: "System V",
479: "System V Release 2",
480: };
481:
1.115 schwarze 482: struct roff_node *nn;
1.51 kristaps 483: const char *p, *s;
484:
485: n = n->child;
486:
1.114 schwarze 487: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 488: p = unix_versions[0];
489: else {
490: s = n->string;
491: if (0 == strcmp(s, "3"))
492: p = unix_versions[0];
493: else if (0 == strcmp(s, "4"))
494: p = unix_versions[1];
495: else if (0 == strcmp(s, "5")) {
496: nn = n->next;
1.114 schwarze 497: if (nn != NULL &&
498: nn->type == ROFFT_TEXT &&
499: nn->string[0] != '\0')
1.51 kristaps 500: p = unix_versions[3];
501: else
502: p = unix_versions[2];
503: } else
504: p = unix_versions[0];
505: }
506:
1.116 schwarze 507: free(man->meta.os);
508: man->meta.os = mandoc_strdup(p);
1.129 schwarze 509: }
510:
511: static void
512: post_in(CHKARGS)
513: {
514: char *s;
515:
516: if (n->parent->tok != MAN_TP ||
517: n->parent->type != ROFFT_HEAD ||
518: n->child == NULL ||
519: *n->child->string == '+' ||
520: *n->child->string == '-')
521: return;
522: mandoc_asprintf(&s, "+%s", n->child->string);
523: free(n->child->string);
524: n->child->string = s;
1.51 kristaps 525: }
CVSweb