Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.88
1.88 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.87 2015/04/02 23:47:43 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.84 schwarze 4: * Copyright (c) 2010, 2012-2015 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.2 schwarze 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: *
1.85 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.2 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.85 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.2 schwarze 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: */
18: #include <sys/types.h>
19:
20: #include <assert.h>
21: #include <ctype.h>
1.6 schwarze 22: #include <errno.h>
23: #include <limits.h>
1.1 kristaps 24: #include <stdarg.h>
25: #include <stdlib.h>
1.28 schwarze 26: #include <string.h>
1.36 schwarze 27: #include <time.h>
1.1 kristaps 28:
1.85 schwarze 29: #include "mandoc_aux.h"
30: #include "mandoc.h"
31: #include "roff.h"
1.44 schwarze 32: #include "man.h"
1.85 schwarze 33: #include "libmandoc.h"
1.1 kristaps 34: #include "libman.h"
35:
1.88 ! schwarze 36: #define CHKARGS struct roff_man *man, struct roff_node *n
1.1 kristaps 37:
1.78 schwarze 38: typedef void (*v_check)(CHKARGS);
1.1 kristaps 39:
1.78 schwarze 40: static void check_par(CHKARGS);
41: static void check_part(CHKARGS);
42: static void check_root(CHKARGS);
43: static void check_text(CHKARGS);
44:
45: static void post_AT(CHKARGS);
46: static void post_IP(CHKARGS);
47: static void post_vs(CHKARGS);
48: static void post_fi(CHKARGS);
49: static void post_ft(CHKARGS);
50: static void post_nf(CHKARGS);
1.84 schwarze 51: static void post_OP(CHKARGS);
1.78 schwarze 52: static void post_TH(CHKARGS);
53: static void post_UC(CHKARGS);
54: static void post_UR(CHKARGS);
1.34 schwarze 55:
1.76 schwarze 56: static v_check man_valids[MAN_MAX] = {
57: post_vs, /* br */
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 */
78: post_vs, /* sp */
79: post_nf, /* nf */
80: post_fi, /* fi */
81: NULL, /* RE */
82: check_part, /* RS */
83: NULL, /* DT */
84: post_UC, /* UC */
1.83 schwarze 85: NULL, /* PD */
1.76 schwarze 86: post_AT, /* AT */
87: NULL, /* in */
88: post_ft, /* ft */
1.84 schwarze 89: post_OP, /* OP */
1.76 schwarze 90: post_nf, /* EX */
91: post_fi, /* EE */
92: post_UR, /* UR */
93: NULL, /* UE */
94: NULL, /* ll */
1.1 kristaps 95: };
96:
97:
1.78 schwarze 98: void
1.88 ! schwarze 99: man_valid_post(struct roff_man *man)
1.1 kristaps 100: {
1.86 schwarze 101: struct roff_node *n;
1.7 schwarze 102: v_check *cp;
1.1 kristaps 103:
1.76 schwarze 104: n = man->last;
105: if (n->flags & MAN_VALID)
1.78 schwarze 106: return;
1.76 schwarze 107: n->flags |= MAN_VALID;
1.1 kristaps 108:
1.76 schwarze 109: switch (n->type) {
1.85 schwarze 110: case ROFFT_TEXT:
1.78 schwarze 111: check_text(man, n);
112: break;
1.85 schwarze 113: case ROFFT_ROOT:
1.78 schwarze 114: check_root(man, n);
115: break;
1.85 schwarze 116: case ROFFT_EQN:
1.42 schwarze 117: /* FALLTHROUGH */
1.85 schwarze 118: case ROFFT_TBL:
1.78 schwarze 119: break;
1.1 kristaps 120: default:
1.76 schwarze 121: cp = man_valids + n->tok;
1.78 schwarze 122: if (*cp)
123: (*cp)(man, n);
124: break;
1.1 kristaps 125: }
126: }
127:
1.78 schwarze 128: static void
1.63 schwarze 129: check_root(CHKARGS)
1.4 schwarze 130: {
1.7 schwarze 131:
1.73 schwarze 132: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 133:
1.65 schwarze 134: if (NULL == man->first->child)
1.75 schwarze 135: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
136: n->line, n->pos, NULL);
1.65 schwarze 137: else
138: man->meta.hasbody = 1;
139:
140: if (NULL == man->meta.title) {
1.77 schwarze 141: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 142: n->line, n->pos, NULL);
1.35 schwarze 143:
1.18 schwarze 144: /*
145: * If a title hasn't been set, do so now (by
146: * implication, date and section also aren't set).
147: */
1.35 schwarze 148:
1.77 schwarze 149: man->meta.title = mandoc_strdup("");
150: man->meta.msec = mandoc_strdup("");
1.60 schwarze 151: man->meta.date = man->quick ? mandoc_strdup("") :
152: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 153: }
1.15 schwarze 154: }
155:
1.78 schwarze 156: static void
1.47 schwarze 157: check_text(CHKARGS)
1.4 schwarze 158: {
1.47 schwarze 159: char *cp, *p;
160:
1.57 schwarze 161: if (MAN_LITERAL & man->flags)
1.78 schwarze 162: return;
1.48 schwarze 163:
164: cp = n->string;
165: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 166: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
167: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 168: }
169:
1.84 schwarze 170: static void
171: post_OP(CHKARGS)
172: {
173:
174: if (n->nchild == 0)
175: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
176: n->line, n->pos, "OP");
177: else if (n->nchild > 2) {
178: n = n->child->next->next;
179: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
180: n->line, n->pos, "OP ... %s", n->string);
181: }
1.1 kristaps 182: }
183:
1.78 schwarze 184: static void
1.76 schwarze 185: post_UR(CHKARGS)
1.58 schwarze 186: {
187:
1.85 schwarze 188: if (n->type == ROFFT_HEAD && n->child == NULL)
1.84 schwarze 189: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
190: n->line, n->pos, "UR");
1.78 schwarze 191: check_part(man, n);
1.58 schwarze 192: }
1.32 schwarze 193:
1.78 schwarze 194: static void
1.47 schwarze 195: post_ft(CHKARGS)
1.32 schwarze 196: {
197: char *cp;
198: int ok;
199:
200: if (0 == n->nchild)
1.78 schwarze 201: return;
1.32 schwarze 202:
203: ok = 0;
204: cp = n->child->string;
205: switch (*cp) {
1.63 schwarze 206: case '1':
1.32 schwarze 207: /* FALLTHROUGH */
1.63 schwarze 208: case '2':
1.32 schwarze 209: /* FALLTHROUGH */
1.63 schwarze 210: case '3':
1.32 schwarze 211: /* FALLTHROUGH */
1.63 schwarze 212: case '4':
1.32 schwarze 213: /* FALLTHROUGH */
1.63 schwarze 214: case 'I':
1.32 schwarze 215: /* FALLTHROUGH */
1.63 schwarze 216: case 'P':
1.32 schwarze 217: /* FALLTHROUGH */
1.63 schwarze 218: case 'R':
1.32 schwarze 219: if ('\0' == cp[1])
220: ok = 1;
221: break;
1.63 schwarze 222: case 'B':
1.32 schwarze 223: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
224: ok = 1;
225: break;
1.63 schwarze 226: case 'C':
1.32 schwarze 227: if ('W' == cp[1] && '\0' == cp[2])
228: ok = 1;
229: break;
230: default:
231: break;
232: }
233:
234: if (0 == ok) {
1.70 schwarze 235: mandoc_vmsg(MANDOCERR_FT_BAD, man->parse,
236: n->line, n->pos, "ft %s", cp);
1.32 schwarze 237: *cp = '\0';
238: }
239: }
1.7 schwarze 240:
1.78 schwarze 241: static void
1.8 schwarze 242: check_part(CHKARGS)
243: {
244:
1.85 schwarze 245: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 246: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
247: n->line, n->pos, man_macronames[n->tok]);
1.8 schwarze 248: }
249:
1.78 schwarze 250: static void
1.36 schwarze 251: check_par(CHKARGS)
252: {
253:
1.39 schwarze 254: switch (n->type) {
1.85 schwarze 255: case ROFFT_BLOCK:
1.39 schwarze 256: if (0 == n->body->nchild)
1.57 schwarze 257: man_node_delete(man, n);
1.39 schwarze 258: break;
1.85 schwarze 259: case ROFFT_BODY:
1.39 schwarze 260: if (0 == n->nchild)
1.67 schwarze 261: mandoc_vmsg(MANDOCERR_PAR_SKIP,
262: man->parse, n->line, n->pos,
263: "%s empty", man_macronames[n->tok]);
1.39 schwarze 264: break;
1.85 schwarze 265: case ROFFT_HEAD:
1.39 schwarze 266: if (n->nchild)
1.69 schwarze 267: mandoc_vmsg(MANDOCERR_ARG_SKIP,
268: man->parse, n->line, n->pos,
269: "%s %s%s", man_macronames[n->tok],
270: n->child->string,
271: n->nchild > 1 ? " ..." : "");
1.39 schwarze 272: break;
273: default:
274: break;
275: }
1.36 schwarze 276: }
277:
1.78 schwarze 278: static void
1.55 schwarze 279: post_IP(CHKARGS)
280: {
281:
282: switch (n->type) {
1.85 schwarze 283: case ROFFT_BLOCK:
1.55 schwarze 284: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 schwarze 285: man_node_delete(man, n);
1.55 schwarze 286: break;
1.85 schwarze 287: case ROFFT_BODY:
1.55 schwarze 288: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.67 schwarze 289: mandoc_vmsg(MANDOCERR_PAR_SKIP,
290: man->parse, n->line, n->pos,
291: "%s empty", man_macronames[n->tok]);
1.55 schwarze 292: break;
293: default:
294: break;
295: }
296: }
1.36 schwarze 297:
1.78 schwarze 298: static void
1.34 schwarze 299: post_TH(CHKARGS)
300: {
1.86 schwarze 301: struct roff_node *nb;
1.40 schwarze 302: const char *p;
1.34 schwarze 303:
1.57 schwarze 304: free(man->meta.title);
305: free(man->meta.vol);
1.87 schwarze 306: free(man->meta.os);
1.57 schwarze 307: free(man->meta.msec);
308: free(man->meta.date);
1.34 schwarze 309:
1.57 schwarze 310: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 311: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 312:
1.64 schwarze 313: nb = n;
314:
1.87 schwarze 315: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 316:
317: n = n->child;
1.40 schwarze 318: if (n && n->string) {
319: for (p = n->string; '\0' != *p; p++) {
320: /* Only warn about this once... */
1.63 schwarze 321: if (isalpha((unsigned char)*p) &&
322: ! isupper((unsigned char)*p)) {
1.74 schwarze 323: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 324: man->parse, n->line,
325: n->pos + (p - n->string),
1.74 schwarze 326: "TH %s", n->string);
1.40 schwarze 327: break;
328: }
329: }
1.57 schwarze 330: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 331: } else {
1.57 schwarze 332: man->meta.title = mandoc_strdup("");
1.77 schwarze 333: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
334: nb->line, nb->pos, "TH");
335: }
1.34 schwarze 336:
1.87 schwarze 337: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 338:
1.40 schwarze 339: if (n)
340: n = n->next;
341: if (n && n->string)
1.57 schwarze 342: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 343: else {
1.57 schwarze 344: man->meta.msec = mandoc_strdup("");
1.77 schwarze 345: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
346: nb->line, nb->pos, "TH %s", man->meta.title);
347: }
1.34 schwarze 348:
1.87 schwarze 349: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 350:
1.40 schwarze 351: if (n)
352: n = n->next;
1.49 schwarze 353: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 354: man->meta.date = man->quick ?
355: mandoc_strdup(n->string) :
356: mandoc_normdate(man->parse, n->string,
357: n->line, n->pos);
1.64 schwarze 358: } else {
1.57 schwarze 359: man->meta.date = mandoc_strdup("");
1.74 schwarze 360: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
361: n ? n->line : nb->line,
362: n ? n->pos : nb->pos, "TH");
1.64 schwarze 363: }
1.34 schwarze 364:
1.87 schwarze 365: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 366:
367: if (n && (n = n->next))
1.87 schwarze 368: man->meta.os = mandoc_strdup(n->string);
1.79 schwarze 369: else if (man->defos != NULL)
1.87 schwarze 370: man->meta.os = mandoc_strdup(man->defos);
1.34 schwarze 371:
1.87 schwarze 372: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 373: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 374:
375: if (n && (n = n->next))
1.57 schwarze 376: man->meta.vol = mandoc_strdup(n->string);
377: else if ('\0' != man->meta.msec[0] &&
378: (NULL != (p = mandoc_a2msec(man->meta.msec))))
379: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 380:
381: if (n != NULL && (n = n->next) != NULL)
382: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
383: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 384:
385: /*
386: * Remove the `TH' node after we've processed it for our
387: * meta-data.
388: */
1.57 schwarze 389: man_node_delete(man, man->last);
1.34 schwarze 390: }
391:
1.78 schwarze 392: static void
1.34 schwarze 393: post_nf(CHKARGS)
394: {
395:
1.82 schwarze 396: if (man->flags & MAN_LITERAL)
1.74 schwarze 397: mandoc_msg(MANDOCERR_NF_SKIP, man->parse,
398: n->line, n->pos, "nf");
1.34 schwarze 399:
1.57 schwarze 400: man->flags |= MAN_LITERAL;
1.34 schwarze 401: }
402:
1.78 schwarze 403: static void
1.34 schwarze 404: post_fi(CHKARGS)
405: {
406:
1.57 schwarze 407: if ( ! (MAN_LITERAL & man->flags))
1.74 schwarze 408: mandoc_msg(MANDOCERR_FI_SKIP, man->parse,
409: n->line, n->pos, "fi");
1.34 schwarze 410:
1.57 schwarze 411: man->flags &= ~MAN_LITERAL;
1.34 schwarze 412: }
413:
1.78 schwarze 414: static void
1.34 schwarze 415: post_UC(CHKARGS)
416: {
417: static const char * const bsd_versions[] = {
418: "3rd Berkeley Distribution",
419: "4th Berkeley Distribution",
420: "4.2 Berkeley Distribution",
421: "4.3 Berkeley Distribution",
422: "4.4 Berkeley Distribution",
423: };
424:
425: const char *p, *s;
426:
427: n = n->child;
428:
1.85 schwarze 429: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 430: p = bsd_versions[0];
431: else {
432: s = n->string;
433: if (0 == strcmp(s, "3"))
434: p = bsd_versions[0];
435: else if (0 == strcmp(s, "4"))
436: p = bsd_versions[1];
437: else if (0 == strcmp(s, "5"))
438: p = bsd_versions[2];
439: else if (0 == strcmp(s, "6"))
440: p = bsd_versions[3];
441: else if (0 == strcmp(s, "7"))
442: p = bsd_versions[4];
443: else
444: p = bsd_versions[0];
445: }
446:
1.87 schwarze 447: free(man->meta.os);
448: man->meta.os = mandoc_strdup(p);
1.34 schwarze 449: }
450:
1.78 schwarze 451: static void
1.34 schwarze 452: post_AT(CHKARGS)
453: {
454: static const char * const unix_versions[] = {
455: "7th Edition",
456: "System III",
457: "System V",
458: "System V Release 2",
459: };
460:
1.86 schwarze 461: struct roff_node *nn;
1.34 schwarze 462: const char *p, *s;
463:
464: n = n->child;
465:
1.85 schwarze 466: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 467: p = unix_versions[0];
468: else {
469: s = n->string;
470: if (0 == strcmp(s, "3"))
471: p = unix_versions[0];
472: else if (0 == strcmp(s, "4"))
473: p = unix_versions[1];
474: else if (0 == strcmp(s, "5")) {
475: nn = n->next;
1.85 schwarze 476: if (nn != NULL &&
477: nn->type == ROFFT_TEXT &&
478: nn->string[0] != '\0')
1.34 schwarze 479: p = unix_versions[3];
480: else
481: p = unix_versions[2];
482: } else
483: p = unix_versions[0];
484: }
485:
1.87 schwarze 486: free(man->meta.os);
487: man->meta.os = mandoc_strdup(p);
1.46 schwarze 488: }
489:
1.78 schwarze 490: static void
1.46 schwarze 491: post_vs(CHKARGS)
492: {
493:
1.54 schwarze 494: if (NULL != n->prev)
1.78 schwarze 495: return;
1.54 schwarze 496:
497: switch (n->parent->tok) {
1.63 schwarze 498: case MAN_SH:
1.54 schwarze 499: /* FALLTHROUGH */
1.63 schwarze 500: case MAN_SS:
1.67 schwarze 501: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
502: "%s after %s", man_macronames[n->tok],
503: man_macronames[n->parent->tok]);
1.54 schwarze 504: /* FALLTHROUGH */
1.63 schwarze 505: case MAN_MAX:
506: /*
1.54 schwarze 507: * Don't warn about this because it occurs in pod2man
508: * and would cause considerable (unfixable) warnage.
509: */
1.57 schwarze 510: man_node_delete(man, n);
1.54 schwarze 511: break;
512: default:
513: break;
514: }
1.34 schwarze 515: }