Annotation of mandoc/man_validate.c, Revision 1.145
1.145 ! schwarze 1: /* $Id: man_validate.c,v 1.144 2018/12/31 07:08:12 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);
1.144 schwarze 156: man_state(man, n);
1.124 schwarze 157: break;
158: }
159: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.135 schwarze 160: cp = man_valids + (n->tok - MAN_TH);
1.107 schwarze 161: if (*cp)
162: (*cp)(man, n);
1.121 schwarze 163: if (man->last == n)
164: man_state(man, n);
1.107 schwarze 165: break;
1.1 kristaps 166: }
167: }
168:
1.107 schwarze 169: static void
1.91 schwarze 170: check_root(CHKARGS)
1.14 kristaps 171: {
1.101 schwarze 172: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.17 kristaps 173:
1.134 schwarze 174: if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1.141 schwarze 175: mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
1.93 schwarze 176: else
177: man->meta.hasbody = 1;
178:
179: if (NULL == man->meta.title) {
1.141 schwarze 180: mandoc_msg(MANDOCERR_TH_NOTITLE, 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)
1.141 schwarze 195: mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
1.131 schwarze 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:
1.145 ! schwarze 211: if (n->flags & NODE_NOFILL)
1.107 schwarze 212: return;
1.76 schwarze 213:
214: cp = n->string;
215: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.141 schwarze 216: mandoc_msg(MANDOCERR_FI_TAB,
217: n->line, n->pos + (int)(p - cp), NULL);
1.145 ! schwarze 218: }
! 219:
! 220: static void
! 221: post_EE(CHKARGS)
! 222: {
! 223: if ((n->flags & NODE_NOFILL) == 0)
! 224: mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
! 225: }
! 226:
! 227: static void
! 228: post_EX(CHKARGS)
! 229: {
! 230: if (n->flags & NODE_NOFILL)
! 231: mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
1.1 kristaps 232: }
233:
1.113 schwarze 234: static void
235: post_OP(CHKARGS)
236: {
237:
1.122 schwarze 238: if (n->child == NULL)
1.141 schwarze 239: mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
1.122 schwarze 240: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 schwarze 241: n = n->child->next->next;
1.141 schwarze 242: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.113 schwarze 243: n->line, n->pos, "OP ... %s", n->string);
244: }
1.1 kristaps 245: }
246:
1.107 schwarze 247: static void
1.139 schwarze 248: post_SH(CHKARGS)
249: {
250: struct roff_node *nc;
251:
252: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
253: return;
254:
255: if (nc->tok == MAN_PP && nc->body->child != NULL) {
256: while (nc->body->last != NULL) {
257: man->next = ROFF_NEXT_CHILD;
258: roff_node_relink(man, nc->body->last);
259: man->last = n;
260: }
261: }
262:
263: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
1.141 schwarze 264: mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
265: "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
1.139 schwarze 266: roff_node_delete(man, nc);
267: }
268:
269: /*
270: * Trailing PP is empty, so it is deleted by check_par().
271: * Trailing sp is significant.
272: */
273:
274: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
1.141 schwarze 275: mandoc_msg(MANDOCERR_PAR_SKIP,
1.139 schwarze 276: nc->line, nc->pos, "%s at the end of %s",
277: roff_name[nc->tok], roff_name[n->tok]);
278: roff_node_delete(man, nc);
279: }
280: }
281:
282: static void
1.104 schwarze 283: post_UR(CHKARGS)
1.86 schwarze 284: {
1.114 schwarze 285: if (n->type == ROFFT_HEAD && n->child == NULL)
1.141 schwarze 286: mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
287: "%s", roff_name[n->tok]);
1.107 schwarze 288: check_part(man, n);
1.55 kristaps 289: }
1.16 kristaps 290:
1.107 schwarze 291: static void
1.23 kristaps 292: check_part(CHKARGS)
293: {
294:
1.114 schwarze 295: if (n->type == ROFFT_BODY && n->child == NULL)
1.141 schwarze 296: mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
297: "%s", roff_name[n->tok]);
1.23 kristaps 298: }
299:
1.107 schwarze 300: static void
1.17 kristaps 301: check_par(CHKARGS)
302: {
303:
1.59 kristaps 304: switch (n->type) {
1.114 schwarze 305: case ROFFT_BLOCK:
1.122 schwarze 306: if (n->body->child == NULL)
1.118 schwarze 307: roff_node_delete(man, n);
1.59 kristaps 308: break;
1.114 schwarze 309: case ROFFT_BODY:
1.139 schwarze 310: if (n->child != NULL &&
311: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
1.141 schwarze 312: mandoc_msg(MANDOCERR_PAR_SKIP,
313: n->child->line, n->child->pos,
1.139 schwarze 314: "%s after %s", roff_name[n->child->tok],
315: roff_name[n->tok]);
316: roff_node_delete(man, n->child);
317: }
1.122 schwarze 318: if (n->child == NULL)
1.141 schwarze 319: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.123 schwarze 320: "%s empty", roff_name[n->tok]);
1.59 kristaps 321: break;
1.114 schwarze 322: case ROFFT_HEAD:
1.122 schwarze 323: if (n->child != NULL)
1.141 schwarze 324: mandoc_msg(MANDOCERR_ARG_SKIP,
325: n->line, n->pos, "%s %s%s",
1.123 schwarze 326: roff_name[n->tok], n->child->string,
1.122 schwarze 327: n->child->next != NULL ? " ..." : "");
1.59 kristaps 328: break;
329: default:
330: break;
331: }
1.17 kristaps 332: }
333:
1.107 schwarze 334: static void
1.83 schwarze 335: post_IP(CHKARGS)
336: {
337:
338: switch (n->type) {
1.114 schwarze 339: case ROFFT_BLOCK:
1.122 schwarze 340: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 341: roff_node_delete(man, n);
1.83 schwarze 342: break;
1.114 schwarze 343: case ROFFT_BODY:
1.122 schwarze 344: if (n->parent->head->child == NULL && n->child == NULL)
1.141 schwarze 345: mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
1.123 schwarze 346: "%s empty", roff_name[n->tok]);
1.83 schwarze 347: break;
348: default:
349: break;
350: }
351: }
1.17 kristaps 352:
1.107 schwarze 353: static void
1.51 kristaps 354: post_TH(CHKARGS)
355: {
1.115 schwarze 356: struct roff_node *nb;
1.60 schwarze 357: const char *p;
1.51 kristaps 358:
1.85 schwarze 359: free(man->meta.title);
360: free(man->meta.vol);
1.116 schwarze 361: free(man->meta.os);
1.85 schwarze 362: free(man->meta.msec);
363: free(man->meta.date);
1.51 kristaps 364:
1.85 schwarze 365: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 366: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 367:
1.92 schwarze 368: nb = n;
369:
1.116 schwarze 370: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 371:
372: n = n->child;
1.60 schwarze 373: if (n && n->string) {
374: for (p = n->string; '\0' != *p; p++) {
375: /* Only warn about this once... */
1.91 schwarze 376: if (isalpha((unsigned char)*p) &&
377: ! isupper((unsigned char)*p)) {
1.141 schwarze 378: mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
379: n->pos + (int)(p - n->string),
1.102 schwarze 380: "TH %s", n->string);
1.60 schwarze 381: break;
382: }
383: }
1.85 schwarze 384: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 385: } else {
1.85 schwarze 386: man->meta.title = mandoc_strdup("");
1.141 schwarze 387: mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
1.105 schwarze 388: }
1.51 kristaps 389:
1.116 schwarze 390: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 391:
1.60 schwarze 392: if (n)
393: n = n->next;
394: if (n && n->string)
1.85 schwarze 395: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 396: else {
1.85 schwarze 397: man->meta.msec = mandoc_strdup("");
1.141 schwarze 398: mandoc_msg(MANDOCERR_MSEC_MISSING,
1.105 schwarze 399: nb->line, nb->pos, "TH %s", man->meta.title);
400: }
1.51 kristaps 401:
1.116 schwarze 402: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 403:
1.60 schwarze 404: if (n)
405: n = n->next;
1.77 schwarze 406: if (n && n->string && '\0' != n->string[0]) {
1.88 schwarze 407: man->meta.date = man->quick ?
408: mandoc_strdup(n->string) :
1.128 schwarze 409: mandoc_normdate(man, n->string, n->line, n->pos);
1.92 schwarze 410: } else {
1.85 schwarze 411: man->meta.date = mandoc_strdup("");
1.141 schwarze 412: mandoc_msg(MANDOCERR_DATE_MISSING,
1.102 schwarze 413: n ? n->line : nb->line,
414: n ? n->pos : nb->pos, "TH");
1.92 schwarze 415: }
1.51 kristaps 416:
1.116 schwarze 417: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 418:
419: if (n && (n = n->next))
1.116 schwarze 420: man->meta.os = mandoc_strdup(n->string);
1.131 schwarze 421: else if (man->os_s != NULL)
422: man->meta.os = mandoc_strdup(man->os_s);
423: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
424: if (strstr(man->meta.os, "OpenBSD") != NULL)
425: man->meta.os_e = MANDOC_OS_OPENBSD;
426: else if (strstr(man->meta.os, "NetBSD") != NULL)
427: man->meta.os_e = MANDOC_OS_NETBSD;
428: }
1.51 kristaps 429:
1.116 schwarze 430: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 431: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 432:
433: if (n && (n = n->next))
1.85 schwarze 434: man->meta.vol = mandoc_strdup(n->string);
435: else if ('\0' != man->meta.msec[0] &&
436: (NULL != (p = mandoc_a2msec(man->meta.msec))))
437: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 438:
439: if (n != NULL && (n = n->next) != NULL)
1.141 schwarze 440: mandoc_msg(MANDOCERR_ARG_EXCESS,
1.113 schwarze 441: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 442:
443: /*
444: * Remove the `TH' node after we've processed it for our
445: * meta-data.
446: */
1.118 schwarze 447: roff_node_delete(man, man->last);
1.51 kristaps 448: }
449:
1.107 schwarze 450: static void
1.51 kristaps 451: post_UC(CHKARGS)
452: {
453: static const char * const bsd_versions[] = {
454: "3rd Berkeley Distribution",
455: "4th Berkeley Distribution",
456: "4.2 Berkeley Distribution",
457: "4.3 Berkeley Distribution",
458: "4.4 Berkeley Distribution",
459: };
460:
461: const char *p, *s;
462:
463: n = n->child;
464:
1.114 schwarze 465: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 466: p = bsd_versions[0];
467: else {
468: s = n->string;
469: if (0 == strcmp(s, "3"))
470: p = bsd_versions[0];
471: else if (0 == strcmp(s, "4"))
472: p = bsd_versions[1];
473: else if (0 == strcmp(s, "5"))
474: p = bsd_versions[2];
475: else if (0 == strcmp(s, "6"))
476: p = bsd_versions[3];
477: else if (0 == strcmp(s, "7"))
478: p = bsd_versions[4];
479: else
480: p = bsd_versions[0];
481: }
482:
1.116 schwarze 483: free(man->meta.os);
484: man->meta.os = mandoc_strdup(p);
1.51 kristaps 485: }
486:
1.107 schwarze 487: static void
1.51 kristaps 488: post_AT(CHKARGS)
489: {
490: static const char * const unix_versions[] = {
491: "7th Edition",
492: "System III",
493: "System V",
494: "System V Release 2",
495: };
496:
1.115 schwarze 497: struct roff_node *nn;
1.51 kristaps 498: const char *p, *s;
499:
500: n = n->child;
501:
1.114 schwarze 502: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 503: p = unix_versions[0];
504: else {
505: s = n->string;
506: if (0 == strcmp(s, "3"))
507: p = unix_versions[0];
508: else if (0 == strcmp(s, "4"))
509: p = unix_versions[1];
510: else if (0 == strcmp(s, "5")) {
511: nn = n->next;
1.114 schwarze 512: if (nn != NULL &&
513: nn->type == ROFFT_TEXT &&
514: nn->string[0] != '\0')
1.51 kristaps 515: p = unix_versions[3];
516: else
517: p = unix_versions[2];
518: } else
519: p = unix_versions[0];
520: }
521:
1.116 schwarze 522: free(man->meta.os);
523: man->meta.os = mandoc_strdup(p);
1.129 schwarze 524: }
525:
526: static void
527: post_in(CHKARGS)
528: {
529: char *s;
530:
531: if (n->parent->tok != MAN_TP ||
532: n->parent->type != ROFFT_HEAD ||
533: n->child == NULL ||
534: *n->child->string == '+' ||
535: *n->child->string == '-')
536: return;
537: mandoc_asprintf(&s, "+%s", n->child->string);
538: free(n->child->string);
539: n->child->string = s;
1.51 kristaps 540: }
CVSweb