Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.82
1.82 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.81 2015/02/06 07:12:34 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.60 schwarze 4: * Copyright (c) 2010, 2012, 2013, 2014 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: *
1.2 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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:
1.44 schwarze 29: #include "man.h"
1.25 schwarze 30: #include "mandoc.h"
1.61 schwarze 31: #include "mandoc_aux.h"
1.1 kristaps 32: #include "libman.h"
1.5 schwarze 33: #include "libmandoc.h"
1.1 kristaps 34:
1.57 schwarze 35: #define CHKARGS struct man *man, struct man_node *n
1.1 kristaps 36:
1.78 schwarze 37: typedef void (*v_check)(CHKARGS);
1.1 kristaps 38:
1.78 schwarze 39: static void check_eq2(CHKARGS);
40: static void check_le1(CHKARGS);
41: static void check_le5(CHKARGS);
42: static void check_par(CHKARGS);
43: static void check_part(CHKARGS);
44: static void check_root(CHKARGS);
45: static void check_text(CHKARGS);
46:
47: static void post_AT(CHKARGS);
48: static void post_IP(CHKARGS);
49: static void post_vs(CHKARGS);
50: static void post_fi(CHKARGS);
51: static void post_ft(CHKARGS);
52: static void post_nf(CHKARGS);
53: static void post_TH(CHKARGS);
54: static void post_UC(CHKARGS);
55: static void post_UR(CHKARGS);
1.34 schwarze 56:
1.76 schwarze 57: static v_check man_valids[MAN_MAX] = {
58: post_vs, /* br */
59: post_TH, /* TH */
60: NULL, /* SH */
61: NULL, /* SS */
62: NULL, /* TP */
63: check_par, /* LP */
64: check_par, /* PP */
65: check_par, /* P */
66: post_IP, /* IP */
67: NULL, /* HP */
68: NULL, /* SM */
69: NULL, /* SB */
70: NULL, /* BI */
71: NULL, /* IB */
72: NULL, /* BR */
73: NULL, /* RB */
74: NULL, /* R */
75: NULL, /* B */
76: NULL, /* I */
77: NULL, /* IR */
78: NULL, /* RI */
79: post_vs, /* sp */
80: post_nf, /* nf */
81: post_fi, /* fi */
82: NULL, /* RE */
83: check_part, /* RS */
84: NULL, /* DT */
85: post_UC, /* UC */
86: check_le1, /* PD */
87: post_AT, /* AT */
88: NULL, /* in */
89: post_ft, /* ft */
90: check_eq2, /* OP */
91: post_nf, /* EX */
92: post_fi, /* EE */
93: post_UR, /* UR */
94: NULL, /* UE */
95: NULL, /* ll */
1.1 kristaps 96: };
97:
98:
1.78 schwarze 99: void
1.57 schwarze 100: man_valid_post(struct man *man)
1.1 kristaps 101: {
1.76 schwarze 102: struct man_node *n;
1.7 schwarze 103: v_check *cp;
1.1 kristaps 104:
1.76 schwarze 105: n = man->last;
106: if (n->flags & MAN_VALID)
1.78 schwarze 107: return;
1.76 schwarze 108: n->flags |= MAN_VALID;
1.1 kristaps 109:
1.76 schwarze 110: switch (n->type) {
1.63 schwarze 111: case MAN_TEXT:
1.78 schwarze 112: check_text(man, n);
113: break;
1.63 schwarze 114: case MAN_ROOT:
1.78 schwarze 115: check_root(man, n);
116: break;
1.63 schwarze 117: case MAN_EQN:
1.42 schwarze 118: /* FALLTHROUGH */
1.63 schwarze 119: case MAN_TBL:
1.78 schwarze 120: break;
1.1 kristaps 121: default:
1.76 schwarze 122: cp = man_valids + n->tok;
1.78 schwarze 123: if (*cp)
124: (*cp)(man, n);
125: break;
1.1 kristaps 126: }
127: }
128:
1.78 schwarze 129: static void
1.63 schwarze 130: check_root(CHKARGS)
1.4 schwarze 131: {
1.7 schwarze 132:
1.73 schwarze 133: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 134:
1.65 schwarze 135: if (NULL == man->first->child)
1.75 schwarze 136: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
137: n->line, n->pos, NULL);
1.65 schwarze 138: else
139: man->meta.hasbody = 1;
140:
141: if (NULL == man->meta.title) {
1.77 schwarze 142: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 143: n->line, n->pos, NULL);
1.35 schwarze 144:
1.18 schwarze 145: /*
146: * If a title hasn't been set, do so now (by
147: * implication, date and section also aren't set).
148: */
1.35 schwarze 149:
1.77 schwarze 150: man->meta.title = mandoc_strdup("");
151: man->meta.msec = mandoc_strdup("");
1.60 schwarze 152: man->meta.date = man->quick ? mandoc_strdup("") :
153: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 154: }
1.15 schwarze 155: }
156:
1.78 schwarze 157: static void
1.47 schwarze 158: check_text(CHKARGS)
1.4 schwarze 159: {
1.47 schwarze 160: char *cp, *p;
161:
1.57 schwarze 162: if (MAN_LITERAL & man->flags)
1.78 schwarze 163: return;
1.48 schwarze 164:
165: cp = n->string;
166: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 167: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
168: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 169: }
170:
171: #define INEQ_DEFINE(x, ineq, name) \
1.78 schwarze 172: static void \
1.7 schwarze 173: check_##name(CHKARGS) \
1.1 kristaps 174: { \
1.4 schwarze 175: if (n->nchild ineq (x)) \
1.78 schwarze 176: return; \
1.57 schwarze 177: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line, n->pos, \
1.63 schwarze 178: "line arguments %s %d (have %d)", \
179: #ineq, (x), n->nchild); \
1.1 kristaps 180: }
181:
1.52 schwarze 182: INEQ_DEFINE(2, ==, eq2)
1.10 schwarze 183: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 184: INEQ_DEFINE(5, <=, le5)
1.58 schwarze 185:
1.78 schwarze 186: static void
1.76 schwarze 187: post_UR(CHKARGS)
1.58 schwarze 188: {
189:
190: if (MAN_HEAD == n->type && 1 != n->nchild)
191: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
192: n->pos, "line arguments eq 1 (have %d)", n->nchild);
1.78 schwarze 193: check_part(man, n);
1.58 schwarze 194: }
1.32 schwarze 195:
1.78 schwarze 196: static void
1.47 schwarze 197: post_ft(CHKARGS)
1.32 schwarze 198: {
199: char *cp;
200: int ok;
201:
202: if (0 == n->nchild)
1.78 schwarze 203: return;
1.32 schwarze 204:
205: ok = 0;
206: cp = n->child->string;
207: switch (*cp) {
1.63 schwarze 208: case '1':
1.32 schwarze 209: /* FALLTHROUGH */
1.63 schwarze 210: case '2':
1.32 schwarze 211: /* FALLTHROUGH */
1.63 schwarze 212: case '3':
1.32 schwarze 213: /* FALLTHROUGH */
1.63 schwarze 214: case '4':
1.32 schwarze 215: /* FALLTHROUGH */
1.63 schwarze 216: case 'I':
1.32 schwarze 217: /* FALLTHROUGH */
1.63 schwarze 218: case 'P':
1.32 schwarze 219: /* FALLTHROUGH */
1.63 schwarze 220: case 'R':
1.32 schwarze 221: if ('\0' == cp[1])
222: ok = 1;
223: break;
1.63 schwarze 224: case 'B':
1.32 schwarze 225: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
226: ok = 1;
227: break;
1.63 schwarze 228: case 'C':
1.32 schwarze 229: if ('W' == cp[1] && '\0' == cp[2])
230: ok = 1;
231: break;
232: default:
233: break;
234: }
235:
236: if (0 == ok) {
1.70 schwarze 237: mandoc_vmsg(MANDOCERR_FT_BAD, man->parse,
238: n->line, n->pos, "ft %s", cp);
1.32 schwarze 239: *cp = '\0';
240: }
241:
242: if (1 < n->nchild)
1.63 schwarze 243: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
244: n->pos, "want one child (have %d)", n->nchild);
1.32 schwarze 245: }
1.7 schwarze 246:
1.78 schwarze 247: static void
1.8 schwarze 248: check_part(CHKARGS)
249: {
250:
1.81 schwarze 251: if (n->type == MAN_BODY && n->child == NULL)
252: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
253: n->line, n->pos, man_macronames[n->tok]);
1.8 schwarze 254: }
255:
1.78 schwarze 256: static void
1.36 schwarze 257: check_par(CHKARGS)
258: {
259:
1.39 schwarze 260: switch (n->type) {
1.63 schwarze 261: case MAN_BLOCK:
1.39 schwarze 262: if (0 == n->body->nchild)
1.57 schwarze 263: man_node_delete(man, n);
1.39 schwarze 264: break;
1.63 schwarze 265: case MAN_BODY:
1.39 schwarze 266: if (0 == n->nchild)
1.67 schwarze 267: mandoc_vmsg(MANDOCERR_PAR_SKIP,
268: man->parse, n->line, n->pos,
269: "%s empty", man_macronames[n->tok]);
1.39 schwarze 270: break;
1.63 schwarze 271: case MAN_HEAD:
1.39 schwarze 272: if (n->nchild)
1.69 schwarze 273: mandoc_vmsg(MANDOCERR_ARG_SKIP,
274: man->parse, n->line, n->pos,
275: "%s %s%s", man_macronames[n->tok],
276: n->child->string,
277: n->nchild > 1 ? " ..." : "");
1.39 schwarze 278: break;
279: default:
280: break;
281: }
1.36 schwarze 282: }
283:
1.78 schwarze 284: static void
1.55 schwarze 285: post_IP(CHKARGS)
286: {
287:
288: switch (n->type) {
1.63 schwarze 289: case MAN_BLOCK:
1.55 schwarze 290: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 schwarze 291: man_node_delete(man, n);
1.55 schwarze 292: break;
1.63 schwarze 293: case MAN_BODY:
1.55 schwarze 294: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.67 schwarze 295: mandoc_vmsg(MANDOCERR_PAR_SKIP,
296: man->parse, n->line, n->pos,
297: "%s empty", man_macronames[n->tok]);
1.55 schwarze 298: break;
299: default:
300: break;
301: }
302: }
1.36 schwarze 303:
1.78 schwarze 304: static void
1.34 schwarze 305: post_TH(CHKARGS)
306: {
1.64 schwarze 307: struct man_node *nb;
1.40 schwarze 308: const char *p;
1.34 schwarze 309:
1.76 schwarze 310: check_le5(man, n);
311:
1.57 schwarze 312: free(man->meta.title);
313: free(man->meta.vol);
314: free(man->meta.source);
315: free(man->meta.msec);
316: free(man->meta.date);
1.34 schwarze 317:
1.57 schwarze 318: man->meta.title = man->meta.vol = man->meta.date =
1.63 schwarze 319: man->meta.msec = man->meta.source = NULL;
1.34 schwarze 320:
1.64 schwarze 321: nb = n;
322:
1.34 schwarze 323: /* ->TITLE<- MSEC DATE SOURCE VOL */
324:
325: n = n->child;
1.40 schwarze 326: if (n && n->string) {
327: for (p = n->string; '\0' != *p; p++) {
328: /* Only warn about this once... */
1.63 schwarze 329: if (isalpha((unsigned char)*p) &&
330: ! isupper((unsigned char)*p)) {
1.74 schwarze 331: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 332: man->parse, n->line,
333: n->pos + (p - n->string),
1.74 schwarze 334: "TH %s", n->string);
1.40 schwarze 335: break;
336: }
337: }
1.57 schwarze 338: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 339: } else {
1.57 schwarze 340: man->meta.title = mandoc_strdup("");
1.77 schwarze 341: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
342: nb->line, nb->pos, "TH");
343: }
1.34 schwarze 344:
345: /* TITLE ->MSEC<- DATE SOURCE VOL */
346:
1.40 schwarze 347: if (n)
348: n = n->next;
349: if (n && n->string)
1.57 schwarze 350: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 351: else {
1.57 schwarze 352: man->meta.msec = mandoc_strdup("");
1.77 schwarze 353: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
354: nb->line, nb->pos, "TH %s", man->meta.title);
355: }
1.34 schwarze 356:
357: /* TITLE MSEC ->DATE<- SOURCE VOL */
358:
1.40 schwarze 359: if (n)
360: n = n->next;
1.49 schwarze 361: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 362: man->meta.date = man->quick ?
363: mandoc_strdup(n->string) :
364: mandoc_normdate(man->parse, n->string,
365: n->line, n->pos);
1.64 schwarze 366: } else {
1.57 schwarze 367: man->meta.date = mandoc_strdup("");
1.74 schwarze 368: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
369: n ? n->line : nb->line,
370: n ? n->pos : nb->pos, "TH");
1.64 schwarze 371: }
1.34 schwarze 372:
373: /* TITLE MSEC DATE ->SOURCE<- VOL */
374:
375: if (n && (n = n->next))
1.57 schwarze 376: man->meta.source = mandoc_strdup(n->string);
1.79 schwarze 377: else if (man->defos != NULL)
378: man->meta.source = mandoc_strdup(man->defos);
1.34 schwarze 379:
380: /* TITLE MSEC DATE SOURCE ->VOL<- */
1.51 schwarze 381: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 382:
383: if (n && (n = n->next))
1.57 schwarze 384: man->meta.vol = mandoc_strdup(n->string);
385: else if ('\0' != man->meta.msec[0] &&
386: (NULL != (p = mandoc_a2msec(man->meta.msec))))
387: man->meta.vol = mandoc_strdup(p);
1.34 schwarze 388:
389: /*
390: * Remove the `TH' node after we've processed it for our
391: * meta-data.
392: */
1.57 schwarze 393: man_node_delete(man, man->last);
1.34 schwarze 394: }
395:
1.78 schwarze 396: static void
1.34 schwarze 397: post_nf(CHKARGS)
398: {
399:
1.82 ! schwarze 400: if (man->flags & MAN_LITERAL)
1.74 schwarze 401: mandoc_msg(MANDOCERR_NF_SKIP, man->parse,
402: n->line, n->pos, "nf");
1.34 schwarze 403:
1.57 schwarze 404: man->flags |= MAN_LITERAL;
1.34 schwarze 405: }
406:
1.78 schwarze 407: static void
1.34 schwarze 408: post_fi(CHKARGS)
409: {
410:
1.57 schwarze 411: if ( ! (MAN_LITERAL & man->flags))
1.74 schwarze 412: mandoc_msg(MANDOCERR_FI_SKIP, man->parse,
413: n->line, n->pos, "fi");
1.34 schwarze 414:
1.57 schwarze 415: man->flags &= ~MAN_LITERAL;
1.34 schwarze 416: }
417:
1.78 schwarze 418: static void
1.34 schwarze 419: post_UC(CHKARGS)
420: {
421: static const char * const bsd_versions[] = {
422: "3rd Berkeley Distribution",
423: "4th Berkeley Distribution",
424: "4.2 Berkeley Distribution",
425: "4.3 Berkeley Distribution",
426: "4.4 Berkeley Distribution",
427: };
428:
429: const char *p, *s;
430:
431: n = n->child;
432:
433: if (NULL == n || MAN_TEXT != n->type)
434: p = bsd_versions[0];
435: else {
436: s = n->string;
437: if (0 == strcmp(s, "3"))
438: p = bsd_versions[0];
439: else if (0 == strcmp(s, "4"))
440: p = bsd_versions[1];
441: else if (0 == strcmp(s, "5"))
442: p = bsd_versions[2];
443: else if (0 == strcmp(s, "6"))
444: p = bsd_versions[3];
445: else if (0 == strcmp(s, "7"))
446: p = bsd_versions[4];
447: else
448: p = bsd_versions[0];
449: }
450:
1.57 schwarze 451: free(man->meta.source);
452: man->meta.source = mandoc_strdup(p);
1.34 schwarze 453: }
454:
1.78 schwarze 455: static void
1.34 schwarze 456: post_AT(CHKARGS)
457: {
458: static const char * const unix_versions[] = {
459: "7th Edition",
460: "System III",
461: "System V",
462: "System V Release 2",
463: };
464:
465: const char *p, *s;
466: struct man_node *nn;
467:
468: n = n->child;
469:
470: if (NULL == n || MAN_TEXT != n->type)
471: p = unix_versions[0];
472: else {
473: s = n->string;
474: if (0 == strcmp(s, "3"))
475: p = unix_versions[0];
476: else if (0 == strcmp(s, "4"))
477: p = unix_versions[1];
478: else if (0 == strcmp(s, "5")) {
479: nn = n->next;
480: if (nn && MAN_TEXT == nn->type && nn->string[0])
481: p = unix_versions[3];
482: else
483: p = unix_versions[2];
484: } else
485: p = unix_versions[0];
486: }
487:
1.57 schwarze 488: free(man->meta.source);
489: man->meta.source = mandoc_strdup(p);
1.46 schwarze 490: }
491:
1.78 schwarze 492: static void
1.46 schwarze 493: post_vs(CHKARGS)
494: {
1.76 schwarze 495:
1.82 ! schwarze 496: if (n->tok == MAN_sp)
1.76 schwarze 497: check_le1(man, n);
1.46 schwarze 498:
1.54 schwarze 499: if (NULL != n->prev)
1.78 schwarze 500: return;
1.54 schwarze 501:
502: switch (n->parent->tok) {
1.63 schwarze 503: case MAN_SH:
1.54 schwarze 504: /* FALLTHROUGH */
1.63 schwarze 505: case MAN_SS:
1.67 schwarze 506: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
507: "%s after %s", man_macronames[n->tok],
508: man_macronames[n->parent->tok]);
1.54 schwarze 509: /* FALLTHROUGH */
1.63 schwarze 510: case MAN_MAX:
511: /*
1.54 schwarze 512: * Don't warn about this because it occurs in pod2man
513: * and would cause considerable (unfixable) warnage.
514: */
1.57 schwarze 515: man_node_delete(man, n);
1.54 schwarze 516: break;
517: default:
518: break;
519: }
1.34 schwarze 520: }