Annotation of mandoc/man_validate.c, Revision 1.146
1.146 ! schwarze 1: /* $Id: man_validate.c,v 1.145 2018/12/31 08:38:21 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);
1.145 schwarze 51: static void post_EE(CHKARGS);
52: static void post_EX(CHKARGS);
1.107 schwarze 53: static void post_IP(CHKARGS);
1.113 schwarze 54: static void post_OP(CHKARGS);
1.139 schwarze 55: static void post_SH(CHKARGS);
1.107 schwarze 56: static void post_TH(CHKARGS);
57: static void post_UC(CHKARGS);
58: static void post_UR(CHKARGS);
1.129 schwarze 59: static void post_in(CHKARGS);
1.51 kristaps 60:
1.135 schwarze 61: static const v_check man_valids[MAN_MAX - MAN_TH] = {
1.104 schwarze 62: post_TH, /* TH */
1.139 schwarze 63: post_SH, /* SH */
64: post_SH, /* SS */
1.104 schwarze 65: NULL, /* TP */
1.136 schwarze 66: NULL, /* TQ */
1.138 schwarze 67: check_abort,/* LP */
1.104 schwarze 68: check_par, /* PP */
1.138 schwarze 69: check_abort,/* P */
1.104 schwarze 70: post_IP, /* IP */
71: NULL, /* HP */
72: NULL, /* SM */
73: NULL, /* SB */
74: NULL, /* BI */
75: NULL, /* IB */
76: NULL, /* BR */
77: NULL, /* RB */
78: NULL, /* R */
79: NULL, /* B */
80: NULL, /* I */
81: NULL, /* IR */
82: NULL, /* RI */
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 */
1.145 schwarze 93: post_EX, /* EX */
94: post_EE, /* 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)
1.146 ! schwarze 163: n->flags |= NODE_VALID;
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:
1.145 schwarze 210: if (n->flags & NODE_NOFILL)
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);
1.145 schwarze 217: }
218:
219: static void
220: post_EE(CHKARGS)
221: {
222: if ((n->flags & NODE_NOFILL) == 0)
223: mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
224: }
225:
226: static void
227: post_EX(CHKARGS)
228: {
229: if (n->flags & NODE_NOFILL)
230: mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
1.1 kristaps 231: }
232:
1.113 schwarze 233: static void
234: post_OP(CHKARGS)
235: {
236:
1.122 schwarze 237: if (n->child == NULL)
1.141 schwarze 238: mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
1.122 schwarze 239: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 schwarze 240: n = n->child->next->next;
1.141 schwarze 241: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.113 schwarze 242: n->line, n->pos, "OP ... %s", n->string);
243: }
1.1 kristaps 244: }
245:
1.107 schwarze 246: static void
1.139 schwarze 247: post_SH(CHKARGS)
248: {
249: struct roff_node *nc;
250:
251: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
252: return;
253:
254: if (nc->tok == MAN_PP && nc->body->child != NULL) {
255: while (nc->body->last != NULL) {
256: man->next = ROFF_NEXT_CHILD;
257: roff_node_relink(man, nc->body->last);
258: man->last = n;
259: }
260: }
261:
262: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
1.141 schwarze 263: mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
264: "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
1.139 schwarze 265: roff_node_delete(man, nc);
266: }
267:
268: /*
269: * Trailing PP is empty, so it is deleted by check_par().
270: * Trailing sp is significant.
271: */
272:
273: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
1.141 schwarze 274: mandoc_msg(MANDOCERR_PAR_SKIP,
1.139 schwarze 275: nc->line, nc->pos, "%s at the end of %s",
276: roff_name[nc->tok], roff_name[n->tok]);
277: roff_node_delete(man, nc);
278: }
279: }
280:
281: static void
1.104 schwarze 282: post_UR(CHKARGS)
1.86 schwarze 283: {
1.114 schwarze 284: if (n->type == ROFFT_HEAD && n->child == NULL)
1.141 schwarze 285: mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
286: "%s", roff_name[n->tok]);
1.107 schwarze 287: check_part(man, n);
1.55 kristaps 288: }
1.16 kristaps 289:
1.107 schwarze 290: static void
1.23 kristaps 291: check_part(CHKARGS)
292: {
293:
1.114 schwarze 294: if (n->type == ROFFT_BODY && n->child == NULL)
1.141 schwarze 295: mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
296: "%s", roff_name[n->tok]);
1.23 kristaps 297: }
298:
1.107 schwarze 299: static void
1.17 kristaps 300: check_par(CHKARGS)
301: {
302:
1.59 kristaps 303: switch (n->type) {
1.114 schwarze 304: case ROFFT_BLOCK:
1.122 schwarze 305: if (n->body->child == NULL)
1.118 schwarze 306: roff_node_delete(man, n);
1.59 kristaps 307: break;
1.114 schwarze 308: case ROFFT_BODY:
1.139 schwarze 309: if (n->child != NULL &&
310: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
1.141 schwarze 311: mandoc_msg(MANDOCERR_PAR_SKIP,
312: n->child->line, n->child->pos,
1.139 schwarze 313: "%s after %s", roff_name[n->child->tok],
314: roff_name[n->tok]);
315: roff_node_delete(man, n->child);
316: }
1.122 schwarze 317: if (n->child == NULL)
1.141 schwarze 318: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.123 schwarze 319: "%s empty", roff_name[n->tok]);
1.59 kristaps 320: break;
1.114 schwarze 321: case ROFFT_HEAD:
1.122 schwarze 322: if (n->child != NULL)
1.141 schwarze 323: mandoc_msg(MANDOCERR_ARG_SKIP,
324: n->line, n->pos, "%s %s%s",
1.123 schwarze 325: roff_name[n->tok], n->child->string,
1.122 schwarze 326: n->child->next != NULL ? " ..." : "");
1.59 kristaps 327: break;
328: default:
329: break;
330: }
1.17 kristaps 331: }
332:
1.107 schwarze 333: static void
1.83 schwarze 334: post_IP(CHKARGS)
335: {
336:
337: switch (n->type) {
1.114 schwarze 338: case ROFFT_BLOCK:
1.122 schwarze 339: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 340: roff_node_delete(man, n);
1.83 schwarze 341: break;
1.114 schwarze 342: case ROFFT_BODY:
1.122 schwarze 343: if (n->parent->head->child == NULL && n->child == NULL)
1.141 schwarze 344: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.123 schwarze 345: "%s empty", roff_name[n->tok]);
1.83 schwarze 346: break;
347: default:
348: break;
349: }
350: }
1.17 kristaps 351:
1.107 schwarze 352: static void
1.51 kristaps 353: post_TH(CHKARGS)
354: {
1.115 schwarze 355: struct roff_node *nb;
1.60 schwarze 356: const char *p;
1.51 kristaps 357:
1.85 schwarze 358: free(man->meta.title);
359: free(man->meta.vol);
1.116 schwarze 360: free(man->meta.os);
1.85 schwarze 361: free(man->meta.msec);
362: free(man->meta.date);
1.51 kristaps 363:
1.85 schwarze 364: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 365: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 366:
1.92 schwarze 367: nb = n;
368:
1.116 schwarze 369: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 370:
371: n = n->child;
1.60 schwarze 372: if (n && n->string) {
373: for (p = n->string; '\0' != *p; p++) {
374: /* Only warn about this once... */
1.91 schwarze 375: if (isalpha((unsigned char)*p) &&
376: ! isupper((unsigned char)*p)) {
1.141 schwarze 377: mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
378: n->pos + (int)(p - n->string),
1.102 schwarze 379: "TH %s", n->string);
1.60 schwarze 380: break;
381: }
382: }
1.85 schwarze 383: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 384: } else {
1.85 schwarze 385: man->meta.title = mandoc_strdup("");
1.141 schwarze 386: mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
1.105 schwarze 387: }
1.51 kristaps 388:
1.116 schwarze 389: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 390:
1.60 schwarze 391: if (n)
392: n = n->next;
393: if (n && n->string)
1.85 schwarze 394: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 395: else {
1.85 schwarze 396: man->meta.msec = mandoc_strdup("");
1.141 schwarze 397: mandoc_msg(MANDOCERR_MSEC_MISSING,
1.105 schwarze 398: nb->line, nb->pos, "TH %s", man->meta.title);
399: }
1.51 kristaps 400:
1.116 schwarze 401: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 402:
1.60 schwarze 403: if (n)
404: n = n->next;
1.77 schwarze 405: if (n && n->string && '\0' != n->string[0]) {
1.88 schwarze 406: man->meta.date = man->quick ?
407: mandoc_strdup(n->string) :
1.128 schwarze 408: mandoc_normdate(man, n->string, n->line, n->pos);
1.92 schwarze 409: } else {
1.85 schwarze 410: man->meta.date = mandoc_strdup("");
1.141 schwarze 411: mandoc_msg(MANDOCERR_DATE_MISSING,
1.102 schwarze 412: n ? n->line : nb->line,
413: n ? n->pos : nb->pos, "TH");
1.92 schwarze 414: }
1.51 kristaps 415:
1.116 schwarze 416: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 417:
418: if (n && (n = n->next))
1.116 schwarze 419: man->meta.os = mandoc_strdup(n->string);
1.131 schwarze 420: else if (man->os_s != NULL)
421: man->meta.os = mandoc_strdup(man->os_s);
422: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
423: if (strstr(man->meta.os, "OpenBSD") != NULL)
424: man->meta.os_e = MANDOC_OS_OPENBSD;
425: else if (strstr(man->meta.os, "NetBSD") != NULL)
426: man->meta.os_e = MANDOC_OS_NETBSD;
427: }
1.51 kristaps 428:
1.116 schwarze 429: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 430: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 431:
432: if (n && (n = n->next))
1.85 schwarze 433: man->meta.vol = mandoc_strdup(n->string);
434: else if ('\0' != man->meta.msec[0] &&
435: (NULL != (p = mandoc_a2msec(man->meta.msec))))
436: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 437:
438: if (n != NULL && (n = n->next) != NULL)
1.141 schwarze 439: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.113 schwarze 440: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 441:
442: /*
443: * Remove the `TH' node after we've processed it for our
444: * meta-data.
445: */
1.118 schwarze 446: roff_node_delete(man, man->last);
1.51 kristaps 447: }
448:
1.107 schwarze 449: static void
1.51 kristaps 450: post_UC(CHKARGS)
451: {
452: static const char * const bsd_versions[] = {
453: "3rd Berkeley Distribution",
454: "4th Berkeley Distribution",
455: "4.2 Berkeley Distribution",
456: "4.3 Berkeley Distribution",
457: "4.4 Berkeley Distribution",
458: };
459:
460: const char *p, *s;
461:
462: n = n->child;
463:
1.114 schwarze 464: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 465: p = bsd_versions[0];
466: else {
467: s = n->string;
468: if (0 == strcmp(s, "3"))
469: p = bsd_versions[0];
470: else if (0 == strcmp(s, "4"))
471: p = bsd_versions[1];
472: else if (0 == strcmp(s, "5"))
473: p = bsd_versions[2];
474: else if (0 == strcmp(s, "6"))
475: p = bsd_versions[3];
476: else if (0 == strcmp(s, "7"))
477: p = bsd_versions[4];
478: else
479: p = bsd_versions[0];
480: }
481:
1.116 schwarze 482: free(man->meta.os);
483: man->meta.os = mandoc_strdup(p);
1.51 kristaps 484: }
485:
1.107 schwarze 486: static void
1.51 kristaps 487: post_AT(CHKARGS)
488: {
489: static const char * const unix_versions[] = {
490: "7th Edition",
491: "System III",
492: "System V",
493: "System V Release 2",
494: };
495:
1.115 schwarze 496: struct roff_node *nn;
1.51 kristaps 497: const char *p, *s;
498:
499: n = n->child;
500:
1.114 schwarze 501: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 502: p = unix_versions[0];
503: else {
504: s = n->string;
505: if (0 == strcmp(s, "3"))
506: p = unix_versions[0];
507: else if (0 == strcmp(s, "4"))
508: p = unix_versions[1];
509: else if (0 == strcmp(s, "5")) {
510: nn = n->next;
1.114 schwarze 511: if (nn != NULL &&
512: nn->type == ROFFT_TEXT &&
513: nn->string[0] != '\0')
1.51 kristaps 514: p = unix_versions[3];
515: else
516: p = unix_versions[2];
517: } else
518: p = unix_versions[0];
519: }
520:
1.116 schwarze 521: free(man->meta.os);
522: man->meta.os = mandoc_strdup(p);
1.129 schwarze 523: }
524:
525: static void
526: post_in(CHKARGS)
527: {
528: char *s;
529:
530: if (n->parent->tok != MAN_TP ||
531: n->parent->type != ROFFT_HEAD ||
532: n->child == NULL ||
533: *n->child->string == '+' ||
534: *n->child->string == '-')
535: return;
536: mandoc_asprintf(&s, "+%s", n->child->string);
537: free(n->child->string);
538: n->child->string = s;
1.51 kristaps 539: }
CVSweb