Annotation of mandoc/man_validate.c, Revision 1.149
1.149 ! schwarze 1: /* $Id: man_validate.c,v 1.148 2019/03/13 18:29:18 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:
1.148 schwarze 44: static void check_abort(CHKARGS) __attribute__((__noreturn__));
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("");
1.149 ! schwarze 188: man->meta.date = mandoc_normdate(man, NULL, n->line, n->pos);
1.34 kristaps 189: }
1.130 schwarze 190:
191: if (man->meta.os_e &&
192: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.141 schwarze 193: mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
1.131 schwarze 194: man->meta.os_e == MANDOC_OS_OPENBSD ?
195: "(OpenBSD)" : "(NetBSD)");
1.75 kristaps 196: }
197:
1.148 schwarze 198: static void
1.138 schwarze 199: check_abort(CHKARGS)
200: {
201: abort();
202: }
203:
204: static void
1.75 kristaps 205: check_text(CHKARGS)
206: {
207: char *cp, *p;
208:
1.145 schwarze 209: if (n->flags & NODE_NOFILL)
1.107 schwarze 210: return;
1.76 schwarze 211:
212: cp = n->string;
213: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.141 schwarze 214: mandoc_msg(MANDOCERR_FI_TAB,
215: n->line, n->pos + (int)(p - cp), NULL);
1.145 schwarze 216: }
217:
218: static void
219: post_EE(CHKARGS)
220: {
221: if ((n->flags & NODE_NOFILL) == 0)
222: mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
223: }
224:
225: static void
226: post_EX(CHKARGS)
227: {
228: if (n->flags & NODE_NOFILL)
229: mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
1.1 kristaps 230: }
231:
1.113 schwarze 232: static void
233: post_OP(CHKARGS)
234: {
235:
1.122 schwarze 236: if (n->child == NULL)
1.141 schwarze 237: mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
1.122 schwarze 238: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 schwarze 239: n = n->child->next->next;
1.141 schwarze 240: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.113 schwarze 241: n->line, n->pos, "OP ... %s", n->string);
242: }
1.1 kristaps 243: }
244:
1.107 schwarze 245: static void
1.139 schwarze 246: post_SH(CHKARGS)
247: {
248: struct roff_node *nc;
249:
250: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
251: return;
252:
253: if (nc->tok == MAN_PP && nc->body->child != NULL) {
254: while (nc->body->last != NULL) {
255: man->next = ROFF_NEXT_CHILD;
256: roff_node_relink(man, nc->body->last);
257: man->last = n;
258: }
259: }
260:
261: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
1.141 schwarze 262: mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
263: "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
1.139 schwarze 264: roff_node_delete(man, nc);
265: }
266:
267: /*
268: * Trailing PP is empty, so it is deleted by check_par().
269: * Trailing sp is significant.
270: */
271:
272: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
1.141 schwarze 273: mandoc_msg(MANDOCERR_PAR_SKIP,
1.139 schwarze 274: nc->line, nc->pos, "%s at the end of %s",
275: roff_name[nc->tok], roff_name[n->tok]);
276: roff_node_delete(man, nc);
277: }
278: }
279:
280: static void
1.104 schwarze 281: post_UR(CHKARGS)
1.86 schwarze 282: {
1.114 schwarze 283: if (n->type == ROFFT_HEAD && n->child == NULL)
1.141 schwarze 284: mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
285: "%s", roff_name[n->tok]);
1.107 schwarze 286: check_part(man, n);
1.55 kristaps 287: }
1.16 kristaps 288:
1.107 schwarze 289: static void
1.23 kristaps 290: check_part(CHKARGS)
291: {
292:
1.114 schwarze 293: if (n->type == ROFFT_BODY && n->child == NULL)
1.141 schwarze 294: mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
295: "%s", roff_name[n->tok]);
1.23 kristaps 296: }
297:
1.107 schwarze 298: static void
1.17 kristaps 299: check_par(CHKARGS)
300: {
301:
1.59 kristaps 302: switch (n->type) {
1.114 schwarze 303: case ROFFT_BLOCK:
1.122 schwarze 304: if (n->body->child == NULL)
1.118 schwarze 305: roff_node_delete(man, n);
1.59 kristaps 306: break;
1.114 schwarze 307: case ROFFT_BODY:
1.139 schwarze 308: if (n->child != NULL &&
309: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
1.141 schwarze 310: mandoc_msg(MANDOCERR_PAR_SKIP,
311: n->child->line, n->child->pos,
1.139 schwarze 312: "%s after %s", roff_name[n->child->tok],
313: roff_name[n->tok]);
314: roff_node_delete(man, n->child);
315: }
1.122 schwarze 316: if (n->child == NULL)
1.141 schwarze 317: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.123 schwarze 318: "%s empty", roff_name[n->tok]);
1.59 kristaps 319: break;
1.114 schwarze 320: case ROFFT_HEAD:
1.122 schwarze 321: if (n->child != NULL)
1.141 schwarze 322: mandoc_msg(MANDOCERR_ARG_SKIP,
323: n->line, n->pos, "%s %s%s",
1.123 schwarze 324: roff_name[n->tok], n->child->string,
1.122 schwarze 325: n->child->next != NULL ? " ..." : "");
1.59 kristaps 326: break;
327: default:
328: break;
329: }
1.17 kristaps 330: }
331:
1.107 schwarze 332: static void
1.83 schwarze 333: post_IP(CHKARGS)
334: {
335:
336: switch (n->type) {
1.114 schwarze 337: case ROFFT_BLOCK:
1.122 schwarze 338: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 339: roff_node_delete(man, n);
1.83 schwarze 340: break;
1.114 schwarze 341: case ROFFT_BODY:
1.122 schwarze 342: if (n->parent->head->child == NULL && n->child == NULL)
1.141 schwarze 343: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.123 schwarze 344: "%s empty", roff_name[n->tok]);
1.83 schwarze 345: break;
346: default:
347: break;
348: }
349: }
1.17 kristaps 350:
1.107 schwarze 351: static void
1.51 kristaps 352: post_TH(CHKARGS)
353: {
1.115 schwarze 354: struct roff_node *nb;
1.60 schwarze 355: const char *p;
1.51 kristaps 356:
1.85 schwarze 357: free(man->meta.title);
358: free(man->meta.vol);
1.116 schwarze 359: free(man->meta.os);
1.85 schwarze 360: free(man->meta.msec);
361: free(man->meta.date);
1.51 kristaps 362:
1.85 schwarze 363: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 364: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 365:
1.92 schwarze 366: nb = n;
367:
1.116 schwarze 368: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 369:
370: n = n->child;
1.149 ! schwarze 371: if (n != NULL && n->string != NULL) {
! 372: for (p = n->string; *p != '\0'; p++) {
1.60 schwarze 373: /* Only warn about this once... */
1.91 schwarze 374: if (isalpha((unsigned char)*p) &&
375: ! isupper((unsigned char)*p)) {
1.141 schwarze 376: mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
377: n->pos + (int)(p - n->string),
1.102 schwarze 378: "TH %s", n->string);
1.60 schwarze 379: break;
380: }
381: }
1.85 schwarze 382: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 383: } else {
1.85 schwarze 384: man->meta.title = mandoc_strdup("");
1.141 schwarze 385: mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
1.105 schwarze 386: }
1.51 kristaps 387:
1.116 schwarze 388: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 389:
1.149 ! schwarze 390: if (n != NULL)
1.60 schwarze 391: n = n->next;
1.149 ! schwarze 392: if (n != NULL && n->string != NULL)
1.85 schwarze 393: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 394: else {
1.85 schwarze 395: man->meta.msec = mandoc_strdup("");
1.141 schwarze 396: mandoc_msg(MANDOCERR_MSEC_MISSING,
1.105 schwarze 397: nb->line, nb->pos, "TH %s", man->meta.title);
398: }
1.51 kristaps 399:
1.116 schwarze 400: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 401:
1.149 ! schwarze 402: if (n != NULL)
1.60 schwarze 403: n = n->next;
1.149 ! schwarze 404: if (n != NULL && n->string != NULL && n->string[0] != '\0')
! 405: man->meta.date = mandoc_normdate(man,
! 406: n->string, n->line, n->pos);
! 407: else {
1.85 schwarze 408: man->meta.date = mandoc_strdup("");
1.141 schwarze 409: mandoc_msg(MANDOCERR_DATE_MISSING,
1.149 ! schwarze 410: n == NULL ? nb->line : n->line,
! 411: n == NULL ? nb->pos : n->pos, "TH");
1.92 schwarze 412: }
1.51 kristaps 413:
1.116 schwarze 414: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 415:
416: if (n && (n = n->next))
1.116 schwarze 417: man->meta.os = mandoc_strdup(n->string);
1.131 schwarze 418: else if (man->os_s != NULL)
419: man->meta.os = mandoc_strdup(man->os_s);
420: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
421: if (strstr(man->meta.os, "OpenBSD") != NULL)
422: man->meta.os_e = MANDOC_OS_OPENBSD;
423: else if (strstr(man->meta.os, "NetBSD") != NULL)
424: man->meta.os_e = MANDOC_OS_NETBSD;
425: }
1.51 kristaps 426:
1.116 schwarze 427: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 428: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 429:
430: if (n && (n = n->next))
1.85 schwarze 431: man->meta.vol = mandoc_strdup(n->string);
432: else if ('\0' != man->meta.msec[0] &&
433: (NULL != (p = mandoc_a2msec(man->meta.msec))))
434: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 435:
436: if (n != NULL && (n = n->next) != NULL)
1.141 schwarze 437: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.113 schwarze 438: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 439:
440: /*
441: * Remove the `TH' node after we've processed it for our
442: * meta-data.
443: */
1.118 schwarze 444: roff_node_delete(man, man->last);
1.51 kristaps 445: }
446:
1.107 schwarze 447: static void
1.51 kristaps 448: post_UC(CHKARGS)
449: {
450: static const char * const bsd_versions[] = {
451: "3rd Berkeley Distribution",
452: "4th Berkeley Distribution",
453: "4.2 Berkeley Distribution",
454: "4.3 Berkeley Distribution",
455: "4.4 Berkeley Distribution",
456: };
457:
458: const char *p, *s;
459:
460: n = n->child;
461:
1.114 schwarze 462: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 463: p = bsd_versions[0];
464: else {
465: s = n->string;
466: if (0 == strcmp(s, "3"))
467: p = bsd_versions[0];
468: else if (0 == strcmp(s, "4"))
469: p = bsd_versions[1];
470: else if (0 == strcmp(s, "5"))
471: p = bsd_versions[2];
472: else if (0 == strcmp(s, "6"))
473: p = bsd_versions[3];
474: else if (0 == strcmp(s, "7"))
475: p = bsd_versions[4];
476: else
477: p = bsd_versions[0];
478: }
479:
1.116 schwarze 480: free(man->meta.os);
481: man->meta.os = mandoc_strdup(p);
1.51 kristaps 482: }
483:
1.107 schwarze 484: static void
1.51 kristaps 485: post_AT(CHKARGS)
486: {
487: static const char * const unix_versions[] = {
488: "7th Edition",
489: "System III",
490: "System V",
491: "System V Release 2",
492: };
493:
1.115 schwarze 494: struct roff_node *nn;
1.51 kristaps 495: const char *p, *s;
496:
497: n = n->child;
498:
1.114 schwarze 499: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 500: p = unix_versions[0];
501: else {
502: s = n->string;
503: if (0 == strcmp(s, "3"))
504: p = unix_versions[0];
505: else if (0 == strcmp(s, "4"))
506: p = unix_versions[1];
507: else if (0 == strcmp(s, "5")) {
508: nn = n->next;
1.114 schwarze 509: if (nn != NULL &&
510: nn->type == ROFFT_TEXT &&
511: nn->string[0] != '\0')
1.51 kristaps 512: p = unix_versions[3];
513: else
514: p = unix_versions[2];
515: } else
516: p = unix_versions[0];
517: }
518:
1.116 schwarze 519: free(man->meta.os);
520: man->meta.os = mandoc_strdup(p);
1.129 schwarze 521: }
522:
523: static void
524: post_in(CHKARGS)
525: {
526: char *s;
527:
528: if (n->parent->tok != MAN_TP ||
529: n->parent->type != ROFFT_HEAD ||
530: n->child == NULL ||
531: *n->child->string == '+' ||
532: *n->child->string == '-')
533: return;
534: mandoc_asprintf(&s, "+%s", n->child->string);
535: free(n->child->string);
536: n->child->string = s;
1.51 kristaps 537: }
CVSweb