Annotation of mandoc/man_validate.c, Revision 1.129
1.107 schwarze 1: /* $OpenBSD$ */
1.1 kristaps 2: /*
1.63 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.123 schwarze 4: * Copyright (c) 2010, 2012-2017 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:
1.107 schwarze 43: static void check_par(CHKARGS);
44: static void check_part(CHKARGS);
45: static void check_root(CHKARGS);
46: static void check_text(CHKARGS);
47:
48: static void post_AT(CHKARGS);
49: static void post_IP(CHKARGS);
1.113 schwarze 50: static void post_OP(CHKARGS);
1.107 schwarze 51: static void post_TH(CHKARGS);
52: static void post_UC(CHKARGS);
53: static void post_UR(CHKARGS);
1.129 ! schwarze 54: static void post_in(CHKARGS);
! 55: static void post_vs(CHKARGS);
1.51 kristaps 56:
1.123 schwarze 57: static const v_check __man_valids[MAN_MAX - MAN_TH] = {
1.104 schwarze 58: post_TH, /* TH */
59: NULL, /* SH */
60: NULL, /* SS */
61: NULL, /* TP */
62: check_par, /* LP */
63: check_par, /* PP */
64: check_par, /* P */
65: post_IP, /* IP */
66: NULL, /* HP */
67: NULL, /* SM */
68: NULL, /* SB */
69: NULL, /* BI */
70: NULL, /* IB */
71: NULL, /* BR */
72: NULL, /* RB */
73: NULL, /* R */
74: NULL, /* B */
75: NULL, /* I */
76: NULL, /* IR */
77: NULL, /* RI */
1.121 schwarze 78: NULL, /* nf */
79: NULL, /* fi */
1.104 schwarze 80: NULL, /* RE */
81: check_part, /* RS */
82: NULL, /* DT */
83: post_UC, /* UC */
1.112 schwarze 84: NULL, /* PD */
1.104 schwarze 85: post_AT, /* AT */
1.129 ! schwarze 86: post_in, /* in */
1.113 schwarze 87: post_OP, /* OP */
1.121 schwarze 88: NULL, /* EX */
89: NULL, /* EE */
1.104 schwarze 90: post_UR, /* UR */
91: NULL, /* UE */
1.1 kristaps 92: };
1.123 schwarze 93: static const v_check *man_valids = __man_valids - MAN_TH;
1.1 kristaps 94:
95:
1.107 schwarze 96: void
1.121 schwarze 97: man_node_validate(struct roff_man *man)
1.1 kristaps 98: {
1.115 schwarze 99: struct roff_node *n;
1.123 schwarze 100: const v_check *cp;
1.1 kristaps 101:
1.104 schwarze 102: n = man->last;
1.121 schwarze 103: man->last = man->last->child;
104: while (man->last != NULL) {
105: man_node_validate(man);
106: if (man->last == n)
107: man->last = man->last->child;
108: else
109: man->last = man->last->next;
110: }
1.1 kristaps 111:
1.121 schwarze 112: man->last = n;
113: man->next = ROFF_NEXT_SIBLING;
1.104 schwarze 114: switch (n->type) {
1.114 schwarze 115: case ROFFT_TEXT:
1.107 schwarze 116: check_text(man, n);
117: break;
1.114 schwarze 118: case ROFFT_ROOT:
1.107 schwarze 119: check_root(man, n);
120: break;
1.114 schwarze 121: case ROFFT_EQN:
122: case ROFFT_TBL:
1.107 schwarze 123: break;
1.1 kristaps 124: default:
1.124 schwarze 125: if (n->tok < ROFF_MAX) {
126: switch (n->tok) {
127: case ROFF_br:
1.127 schwarze 128: case ROFF_sp:
1.124 schwarze 129: post_vs(man, n);
130: break;
131: default:
1.125 schwarze 132: roff_validate(man);
133: break;
1.124 schwarze 134: }
135: break;
136: }
137: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.104 schwarze 138: cp = man_valids + n->tok;
1.107 schwarze 139: if (*cp)
140: (*cp)(man, n);
1.121 schwarze 141: if (man->last == n)
142: man_state(man, n);
1.107 schwarze 143: break;
1.1 kristaps 144: }
145: }
146:
1.107 schwarze 147: static void
1.91 schwarze 148: check_root(CHKARGS)
1.14 kristaps 149: {
1.17 kristaps 150:
1.101 schwarze 151: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.17 kristaps 152:
1.93 schwarze 153: if (NULL == man->first->child)
1.103 schwarze 154: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
155: n->line, n->pos, NULL);
1.93 schwarze 156: else
157: man->meta.hasbody = 1;
158:
159: if (NULL == man->meta.title) {
1.105 schwarze 160: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.103 schwarze 161: n->line, n->pos, NULL);
1.54 kristaps 162:
1.34 kristaps 163: /*
164: * If a title hasn't been set, do so now (by
165: * implication, date and section also aren't set).
166: */
1.54 kristaps 167:
1.105 schwarze 168: man->meta.title = mandoc_strdup("");
169: man->meta.msec = mandoc_strdup("");
1.88 schwarze 170: man->meta.date = man->quick ? mandoc_strdup("") :
1.128 schwarze 171: mandoc_normdate(man, NULL, n->line, n->pos);
1.34 kristaps 172: }
1.75 kristaps 173: }
174:
1.107 schwarze 175: static void
1.75 kristaps 176: check_text(CHKARGS)
177: {
178: char *cp, *p;
179:
1.85 schwarze 180: if (MAN_LITERAL & man->flags)
1.107 schwarze 181: return;
1.76 schwarze 182:
183: cp = n->string;
184: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.99 schwarze 185: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
186: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 187: }
188:
1.113 schwarze 189: static void
190: post_OP(CHKARGS)
191: {
192:
1.122 schwarze 193: if (n->child == NULL)
1.113 schwarze 194: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
195: n->line, n->pos, "OP");
1.122 schwarze 196: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 schwarze 197: n = n->child->next->next;
198: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
199: n->line, n->pos, "OP ... %s", n->string);
200: }
1.1 kristaps 201: }
202:
1.107 schwarze 203: static void
1.104 schwarze 204: post_UR(CHKARGS)
1.86 schwarze 205: {
206:
1.114 schwarze 207: if (n->type == ROFFT_HEAD && n->child == NULL)
1.113 schwarze 208: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
209: n->line, n->pos, "UR");
1.107 schwarze 210: check_part(man, n);
1.55 kristaps 211: }
1.16 kristaps 212:
1.107 schwarze 213: static void
1.23 kristaps 214: check_part(CHKARGS)
215: {
216:
1.114 schwarze 217: if (n->type == ROFFT_BODY && n->child == NULL)
1.110 schwarze 218: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.123 schwarze 219: n->line, n->pos, roff_name[n->tok]);
1.23 kristaps 220: }
221:
1.107 schwarze 222: static void
1.17 kristaps 223: check_par(CHKARGS)
224: {
225:
1.59 kristaps 226: switch (n->type) {
1.114 schwarze 227: case ROFFT_BLOCK:
1.122 schwarze 228: if (n->body->child == NULL)
1.118 schwarze 229: roff_node_delete(man, n);
1.59 kristaps 230: break;
1.114 schwarze 231: case ROFFT_BODY:
1.122 schwarze 232: if (n->child == NULL)
1.95 schwarze 233: mandoc_vmsg(MANDOCERR_PAR_SKIP,
234: man->parse, n->line, n->pos,
1.123 schwarze 235: "%s empty", roff_name[n->tok]);
1.59 kristaps 236: break;
1.114 schwarze 237: case ROFFT_HEAD:
1.122 schwarze 238: if (n->child != NULL)
1.97 schwarze 239: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.123 schwarze 240: man->parse, n->line, n->pos, "%s %s%s",
241: roff_name[n->tok], n->child->string,
1.122 schwarze 242: n->child->next != NULL ? " ..." : "");
1.59 kristaps 243: break;
244: default:
245: break;
246: }
1.17 kristaps 247: }
248:
1.107 schwarze 249: static void
1.83 schwarze 250: post_IP(CHKARGS)
251: {
252:
253: switch (n->type) {
1.114 schwarze 254: case ROFFT_BLOCK:
1.122 schwarze 255: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 256: roff_node_delete(man, n);
1.83 schwarze 257: break;
1.114 schwarze 258: case ROFFT_BODY:
1.122 schwarze 259: if (n->parent->head->child == NULL && n->child == NULL)
1.95 schwarze 260: mandoc_vmsg(MANDOCERR_PAR_SKIP,
261: man->parse, n->line, n->pos,
1.123 schwarze 262: "%s empty", roff_name[n->tok]);
1.83 schwarze 263: break;
264: default:
265: break;
266: }
267: }
1.17 kristaps 268:
1.107 schwarze 269: static void
1.51 kristaps 270: post_TH(CHKARGS)
271: {
1.115 schwarze 272: struct roff_node *nb;
1.60 schwarze 273: const char *p;
1.51 kristaps 274:
1.85 schwarze 275: free(man->meta.title);
276: free(man->meta.vol);
1.116 schwarze 277: free(man->meta.os);
1.85 schwarze 278: free(man->meta.msec);
279: free(man->meta.date);
1.51 kristaps 280:
1.85 schwarze 281: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 282: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 283:
1.92 schwarze 284: nb = n;
285:
1.116 schwarze 286: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 287:
288: n = n->child;
1.60 schwarze 289: if (n && n->string) {
290: for (p = n->string; '\0' != *p; p++) {
291: /* Only warn about this once... */
1.91 schwarze 292: if (isalpha((unsigned char)*p) &&
293: ! isupper((unsigned char)*p)) {
1.102 schwarze 294: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.94 schwarze 295: man->parse, n->line,
296: n->pos + (p - n->string),
1.102 schwarze 297: "TH %s", n->string);
1.60 schwarze 298: break;
299: }
300: }
1.85 schwarze 301: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 302: } else {
1.85 schwarze 303: man->meta.title = mandoc_strdup("");
1.105 schwarze 304: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
305: nb->line, nb->pos, "TH");
306: }
1.51 kristaps 307:
1.116 schwarze 308: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 309:
1.60 schwarze 310: if (n)
311: n = n->next;
312: if (n && n->string)
1.85 schwarze 313: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 314: else {
1.85 schwarze 315: man->meta.msec = mandoc_strdup("");
1.105 schwarze 316: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
317: nb->line, nb->pos, "TH %s", man->meta.title);
318: }
1.51 kristaps 319:
1.116 schwarze 320: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 321:
1.60 schwarze 322: if (n)
323: n = n->next;
1.77 schwarze 324: if (n && n->string && '\0' != n->string[0]) {
1.88 schwarze 325: man->meta.date = man->quick ?
326: mandoc_strdup(n->string) :
1.128 schwarze 327: mandoc_normdate(man, n->string, n->line, n->pos);
1.92 schwarze 328: } else {
1.85 schwarze 329: man->meta.date = mandoc_strdup("");
1.102 schwarze 330: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
331: n ? n->line : nb->line,
332: n ? n->pos : nb->pos, "TH");
1.92 schwarze 333: }
1.51 kristaps 334:
1.116 schwarze 335: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 336:
337: if (n && (n = n->next))
1.116 schwarze 338: man->meta.os = mandoc_strdup(n->string);
1.108 schwarze 339: else if (man->defos != NULL)
1.116 schwarze 340: man->meta.os = mandoc_strdup(man->defos);
1.51 kristaps 341:
1.116 schwarze 342: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 343: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 344:
345: if (n && (n = n->next))
1.85 schwarze 346: man->meta.vol = mandoc_strdup(n->string);
347: else if ('\0' != man->meta.msec[0] &&
348: (NULL != (p = mandoc_a2msec(man->meta.msec))))
349: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 350:
351: if (n != NULL && (n = n->next) != NULL)
352: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
353: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 354:
355: /*
356: * Remove the `TH' node after we've processed it for our
357: * meta-data.
358: */
1.118 schwarze 359: roff_node_delete(man, man->last);
1.51 kristaps 360: }
361:
1.107 schwarze 362: static void
1.51 kristaps 363: post_UC(CHKARGS)
364: {
365: static const char * const bsd_versions[] = {
366: "3rd Berkeley Distribution",
367: "4th Berkeley Distribution",
368: "4.2 Berkeley Distribution",
369: "4.3 Berkeley Distribution",
370: "4.4 Berkeley Distribution",
371: };
372:
373: const char *p, *s;
374:
375: n = n->child;
376:
1.114 schwarze 377: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 378: p = bsd_versions[0];
379: else {
380: s = n->string;
381: if (0 == strcmp(s, "3"))
382: p = bsd_versions[0];
383: else if (0 == strcmp(s, "4"))
384: p = bsd_versions[1];
385: else if (0 == strcmp(s, "5"))
386: p = bsd_versions[2];
387: else if (0 == strcmp(s, "6"))
388: p = bsd_versions[3];
389: else if (0 == strcmp(s, "7"))
390: p = bsd_versions[4];
391: else
392: p = bsd_versions[0];
393: }
394:
1.116 schwarze 395: free(man->meta.os);
396: man->meta.os = mandoc_strdup(p);
1.51 kristaps 397: }
398:
1.107 schwarze 399: static void
1.51 kristaps 400: post_AT(CHKARGS)
401: {
402: static const char * const unix_versions[] = {
403: "7th Edition",
404: "System III",
405: "System V",
406: "System V Release 2",
407: };
408:
1.115 schwarze 409: struct roff_node *nn;
1.51 kristaps 410: const char *p, *s;
411:
412: n = n->child;
413:
1.114 schwarze 414: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 415: p = unix_versions[0];
416: else {
417: s = n->string;
418: if (0 == strcmp(s, "3"))
419: p = unix_versions[0];
420: else if (0 == strcmp(s, "4"))
421: p = unix_versions[1];
422: else if (0 == strcmp(s, "5")) {
423: nn = n->next;
1.114 schwarze 424: if (nn != NULL &&
425: nn->type == ROFFT_TEXT &&
426: nn->string[0] != '\0')
1.51 kristaps 427: p = unix_versions[3];
428: else
429: p = unix_versions[2];
430: } else
431: p = unix_versions[0];
432: }
433:
1.116 schwarze 434: free(man->meta.os);
435: man->meta.os = mandoc_strdup(p);
1.129 ! schwarze 436: }
! 437:
! 438: static void
! 439: post_in(CHKARGS)
! 440: {
! 441: char *s;
! 442:
! 443: if (n->parent->tok != MAN_TP ||
! 444: n->parent->type != ROFFT_HEAD ||
! 445: n->child == NULL ||
! 446: *n->child->string == '+' ||
! 447: *n->child->string == '-')
! 448: return;
! 449: mandoc_asprintf(&s, "+%s", n->child->string);
! 450: free(n->child->string);
! 451: n->child->string = s;
1.70 kristaps 452: }
453:
1.107 schwarze 454: static void
1.70 kristaps 455: post_vs(CHKARGS)
456: {
457:
1.82 schwarze 458: if (NULL != n->prev)
1.107 schwarze 459: return;
1.82 schwarze 460:
461: switch (n->parent->tok) {
1.91 schwarze 462: case MAN_SH:
463: case MAN_SS:
1.127 schwarze 464: case MAN_PP:
465: case MAN_LP:
466: case MAN_P:
1.95 schwarze 467: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.123 schwarze 468: "%s after %s", roff_name[n->tok],
469: roff_name[n->parent->tok]);
1.82 schwarze 470: /* FALLTHROUGH */
1.119 schwarze 471: case TOKEN_NONE:
1.91 schwarze 472: /*
1.82 schwarze 473: * Don't warn about this because it occurs in pod2man
474: * and would cause considerable (unfixable) warnage.
475: */
1.118 schwarze 476: roff_node_delete(man, n);
1.82 schwarze 477: break;
478: default:
479: break;
480: }
1.51 kristaps 481: }
CVSweb