Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.78
1.78 ! schwarze 1: /* $OpenBSD$ */
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_eq0(CHKARGS);
! 40: static void check_eq2(CHKARGS);
! 41: static void check_le1(CHKARGS);
! 42: static void check_le5(CHKARGS);
! 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);
! 50: static void post_vs(CHKARGS);
! 51: static void post_fi(CHKARGS);
! 52: static void post_ft(CHKARGS);
! 53: static void post_nf(CHKARGS);
! 54: static void post_TH(CHKARGS);
! 55: static void post_UC(CHKARGS);
! 56: static void post_UR(CHKARGS);
1.34 schwarze 57:
1.76 schwarze 58: static v_check man_valids[MAN_MAX] = {
59: post_vs, /* br */
60: post_TH, /* TH */
61: NULL, /* SH */
62: NULL, /* SS */
63: NULL, /* TP */
64: check_par, /* LP */
65: check_par, /* PP */
66: check_par, /* P */
67: post_IP, /* IP */
68: NULL, /* HP */
69: NULL, /* SM */
70: NULL, /* SB */
71: NULL, /* BI */
72: NULL, /* IB */
73: NULL, /* BR */
74: NULL, /* RB */
75: NULL, /* R */
76: NULL, /* B */
77: NULL, /* I */
78: NULL, /* IR */
79: NULL, /* RI */
80: check_eq0, /* na */
81: post_vs, /* sp */
82: post_nf, /* nf */
83: post_fi, /* fi */
84: NULL, /* RE */
85: check_part, /* RS */
86: NULL, /* DT */
87: post_UC, /* UC */
88: check_le1, /* PD */
89: post_AT, /* AT */
90: NULL, /* in */
91: post_ft, /* ft */
92: check_eq2, /* OP */
93: post_nf, /* EX */
94: post_fi, /* EE */
95: post_UR, /* UR */
96: NULL, /* UE */
97: NULL, /* ll */
1.1 kristaps 98: };
99:
100:
1.78 ! schwarze 101: void
1.57 schwarze 102: man_valid_post(struct man *man)
1.1 kristaps 103: {
1.76 schwarze 104: struct man_node *n;
1.7 schwarze 105: v_check *cp;
1.1 kristaps 106:
1.76 schwarze 107: n = man->last;
108: if (n->flags & MAN_VALID)
1.78 ! schwarze 109: return;
1.76 schwarze 110: n->flags |= MAN_VALID;
1.1 kristaps 111:
1.76 schwarze 112: switch (n->type) {
1.63 schwarze 113: case MAN_TEXT:
1.78 ! schwarze 114: check_text(man, n);
! 115: break;
1.63 schwarze 116: case MAN_ROOT:
1.78 ! schwarze 117: check_root(man, n);
! 118: break;
1.63 schwarze 119: case MAN_EQN:
1.42 schwarze 120: /* FALLTHROUGH */
1.63 schwarze 121: case MAN_TBL:
1.78 ! schwarze 122: break;
1.1 kristaps 123: default:
1.76 schwarze 124: cp = man_valids + n->tok;
1.78 ! schwarze 125: if (*cp)
! 126: (*cp)(man, n);
! 127: break;
1.1 kristaps 128: }
129: }
130:
1.78 ! schwarze 131: static void
1.63 schwarze 132: check_root(CHKARGS)
1.4 schwarze 133: {
1.7 schwarze 134:
1.73 schwarze 135: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 136:
1.65 schwarze 137: if (NULL == man->first->child)
1.75 schwarze 138: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
139: n->line, n->pos, NULL);
1.65 schwarze 140: else
141: man->meta.hasbody = 1;
142:
143: if (NULL == man->meta.title) {
1.77 schwarze 144: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 145: n->line, n->pos, NULL);
1.35 schwarze 146:
1.18 schwarze 147: /*
148: * If a title hasn't been set, do so now (by
149: * implication, date and section also aren't set).
150: */
1.35 schwarze 151:
1.77 schwarze 152: man->meta.title = mandoc_strdup("");
153: man->meta.msec = mandoc_strdup("");
1.60 schwarze 154: man->meta.date = man->quick ? mandoc_strdup("") :
155: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 156: }
1.15 schwarze 157: }
158:
1.78 ! schwarze 159: static void
1.47 schwarze 160: check_text(CHKARGS)
1.4 schwarze 161: {
1.47 schwarze 162: char *cp, *p;
163:
1.57 schwarze 164: if (MAN_LITERAL & man->flags)
1.78 ! schwarze 165: return;
1.48 schwarze 166:
167: cp = n->string;
168: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 169: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
170: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 171: }
172:
173: #define INEQ_DEFINE(x, ineq, name) \
1.78 ! schwarze 174: static void \
1.7 schwarze 175: check_##name(CHKARGS) \
1.1 kristaps 176: { \
1.4 schwarze 177: if (n->nchild ineq (x)) \
1.78 ! schwarze 178: return; \
1.57 schwarze 179: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line, n->pos, \
1.63 schwarze 180: "line arguments %s %d (have %d)", \
181: #ineq, (x), n->nchild); \
1.1 kristaps 182: }
183:
184: INEQ_DEFINE(0, ==, eq0)
1.52 schwarze 185: INEQ_DEFINE(2, ==, eq2)
1.10 schwarze 186: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 187: INEQ_DEFINE(5, <=, le5)
1.58 schwarze 188:
1.78 ! schwarze 189: static void
1.76 schwarze 190: post_UR(CHKARGS)
1.58 schwarze 191: {
192:
193: if (MAN_HEAD == n->type && 1 != n->nchild)
194: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
195: n->pos, "line arguments eq 1 (have %d)", n->nchild);
1.78 ! schwarze 196: check_part(man, n);
1.58 schwarze 197: }
1.32 schwarze 198:
1.78 ! schwarze 199: static void
1.47 schwarze 200: post_ft(CHKARGS)
1.32 schwarze 201: {
202: char *cp;
203: int ok;
204:
205: if (0 == n->nchild)
1.78 ! schwarze 206: return;
1.32 schwarze 207:
208: ok = 0;
209: cp = n->child->string;
210: switch (*cp) {
1.63 schwarze 211: case '1':
1.32 schwarze 212: /* FALLTHROUGH */
1.63 schwarze 213: case '2':
1.32 schwarze 214: /* FALLTHROUGH */
1.63 schwarze 215: case '3':
1.32 schwarze 216: /* FALLTHROUGH */
1.63 schwarze 217: case '4':
1.32 schwarze 218: /* FALLTHROUGH */
1.63 schwarze 219: case 'I':
1.32 schwarze 220: /* FALLTHROUGH */
1.63 schwarze 221: case 'P':
1.32 schwarze 222: /* FALLTHROUGH */
1.63 schwarze 223: case 'R':
1.32 schwarze 224: if ('\0' == cp[1])
225: ok = 1;
226: break;
1.63 schwarze 227: case 'B':
1.32 schwarze 228: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
229: ok = 1;
230: break;
1.63 schwarze 231: case 'C':
1.32 schwarze 232: if ('W' == cp[1] && '\0' == cp[2])
233: ok = 1;
234: break;
235: default:
236: break;
237: }
238:
239: if (0 == ok) {
1.70 schwarze 240: mandoc_vmsg(MANDOCERR_FT_BAD, man->parse,
241: n->line, n->pos, "ft %s", cp);
1.32 schwarze 242: *cp = '\0';
243: }
244:
245: if (1 < n->nchild)
1.63 schwarze 246: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
247: n->pos, "want one child (have %d)", n->nchild);
1.32 schwarze 248: }
1.7 schwarze 249:
1.78 ! schwarze 250: static void
1.8 schwarze 251: check_part(CHKARGS)
252: {
253:
254: if (MAN_BODY == n->type && 0 == n->nchild)
1.63 schwarze 255: mandoc_msg(MANDOCERR_ARGCWARN, man->parse, n->line,
256: n->pos, "want children (have none)");
1.8 schwarze 257: }
258:
1.78 ! schwarze 259: static void
1.36 schwarze 260: check_par(CHKARGS)
261: {
262:
1.39 schwarze 263: switch (n->type) {
1.63 schwarze 264: case MAN_BLOCK:
1.39 schwarze 265: if (0 == n->body->nchild)
1.57 schwarze 266: man_node_delete(man, n);
1.39 schwarze 267: break;
1.63 schwarze 268: case MAN_BODY:
1.39 schwarze 269: if (0 == n->nchild)
1.67 schwarze 270: mandoc_vmsg(MANDOCERR_PAR_SKIP,
271: man->parse, n->line, n->pos,
272: "%s empty", man_macronames[n->tok]);
1.39 schwarze 273: break;
1.63 schwarze 274: case MAN_HEAD:
1.39 schwarze 275: if (n->nchild)
1.69 schwarze 276: mandoc_vmsg(MANDOCERR_ARG_SKIP,
277: man->parse, n->line, n->pos,
278: "%s %s%s", man_macronames[n->tok],
279: n->child->string,
280: n->nchild > 1 ? " ..." : "");
1.39 schwarze 281: break;
282: default:
283: break;
284: }
1.36 schwarze 285: }
286:
1.78 ! schwarze 287: static void
1.55 schwarze 288: post_IP(CHKARGS)
289: {
290:
291: switch (n->type) {
1.63 schwarze 292: case MAN_BLOCK:
1.55 schwarze 293: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 schwarze 294: man_node_delete(man, n);
1.55 schwarze 295: break;
1.63 schwarze 296: case MAN_BODY:
1.55 schwarze 297: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.67 schwarze 298: mandoc_vmsg(MANDOCERR_PAR_SKIP,
299: man->parse, n->line, n->pos,
300: "%s empty", man_macronames[n->tok]);
1.55 schwarze 301: break;
302: default:
303: break;
304: }
305: }
1.36 schwarze 306:
1.78 ! schwarze 307: static void
1.34 schwarze 308: post_TH(CHKARGS)
309: {
1.64 schwarze 310: struct man_node *nb;
1.40 schwarze 311: const char *p;
1.34 schwarze 312:
1.76 schwarze 313: check_le5(man, n);
314:
1.57 schwarze 315: free(man->meta.title);
316: free(man->meta.vol);
317: free(man->meta.source);
318: free(man->meta.msec);
319: free(man->meta.date);
1.34 schwarze 320:
1.57 schwarze 321: man->meta.title = man->meta.vol = man->meta.date =
1.63 schwarze 322: man->meta.msec = man->meta.source = NULL;
1.34 schwarze 323:
1.64 schwarze 324: nb = n;
325:
1.34 schwarze 326: /* ->TITLE<- MSEC DATE SOURCE VOL */
327:
328: n = n->child;
1.40 schwarze 329: if (n && n->string) {
330: for (p = n->string; '\0' != *p; p++) {
331: /* Only warn about this once... */
1.63 schwarze 332: if (isalpha((unsigned char)*p) &&
333: ! isupper((unsigned char)*p)) {
1.74 schwarze 334: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 335: man->parse, n->line,
336: n->pos + (p - n->string),
1.74 schwarze 337: "TH %s", n->string);
1.40 schwarze 338: break;
339: }
340: }
1.57 schwarze 341: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 342: } else {
1.57 schwarze 343: man->meta.title = mandoc_strdup("");
1.77 schwarze 344: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
345: nb->line, nb->pos, "TH");
346: }
1.34 schwarze 347:
348: /* TITLE ->MSEC<- DATE SOURCE VOL */
349:
1.40 schwarze 350: if (n)
351: n = n->next;
352: if (n && n->string)
1.57 schwarze 353: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 354: else {
1.57 schwarze 355: man->meta.msec = mandoc_strdup("");
1.77 schwarze 356: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
357: nb->line, nb->pos, "TH %s", man->meta.title);
358: }
1.34 schwarze 359:
360: /* TITLE MSEC ->DATE<- SOURCE VOL */
361:
1.40 schwarze 362: if (n)
363: n = n->next;
1.49 schwarze 364: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 365: man->meta.date = man->quick ?
366: mandoc_strdup(n->string) :
367: mandoc_normdate(man->parse, n->string,
368: n->line, n->pos);
1.64 schwarze 369: } else {
1.57 schwarze 370: man->meta.date = mandoc_strdup("");
1.74 schwarze 371: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
372: n ? n->line : nb->line,
373: n ? n->pos : nb->pos, "TH");
1.64 schwarze 374: }
1.34 schwarze 375:
376: /* TITLE MSEC DATE ->SOURCE<- VOL */
377:
378: if (n && (n = n->next))
1.57 schwarze 379: man->meta.source = mandoc_strdup(n->string);
1.34 schwarze 380:
381: /* TITLE MSEC DATE SOURCE ->VOL<- */
1.51 schwarze 382: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 383:
384: if (n && (n = n->next))
1.57 schwarze 385: man->meta.vol = mandoc_strdup(n->string);
386: else if ('\0' != man->meta.msec[0] &&
387: (NULL != (p = mandoc_a2msec(man->meta.msec))))
388: man->meta.vol = mandoc_strdup(p);
1.34 schwarze 389:
390: /*
391: * Remove the `TH' node after we've processed it for our
392: * meta-data.
393: */
1.57 schwarze 394: man_node_delete(man, man->last);
1.34 schwarze 395: }
396:
1.78 ! schwarze 397: static void
1.34 schwarze 398: post_nf(CHKARGS)
399: {
400:
1.76 schwarze 401: check_eq0(man, n);
402:
1.57 schwarze 403: if (MAN_LITERAL & man->flags)
1.74 schwarze 404: mandoc_msg(MANDOCERR_NF_SKIP, man->parse,
405: n->line, n->pos, "nf");
1.34 schwarze 406:
1.57 schwarze 407: man->flags |= MAN_LITERAL;
1.34 schwarze 408: }
409:
1.78 ! schwarze 410: static void
1.34 schwarze 411: post_fi(CHKARGS)
412: {
413:
1.76 schwarze 414: check_eq0(man, n);
415:
1.57 schwarze 416: if ( ! (MAN_LITERAL & man->flags))
1.74 schwarze 417: mandoc_msg(MANDOCERR_FI_SKIP, man->parse,
418: n->line, n->pos, "fi");
1.34 schwarze 419:
1.57 schwarze 420: man->flags &= ~MAN_LITERAL;
1.34 schwarze 421: }
422:
1.78 ! schwarze 423: static void
1.34 schwarze 424: post_UC(CHKARGS)
425: {
426: static const char * const bsd_versions[] = {
427: "3rd Berkeley Distribution",
428: "4th Berkeley Distribution",
429: "4.2 Berkeley Distribution",
430: "4.3 Berkeley Distribution",
431: "4.4 Berkeley Distribution",
432: };
433:
434: const char *p, *s;
435:
436: n = n->child;
437:
438: if (NULL == n || MAN_TEXT != n->type)
439: p = bsd_versions[0];
440: else {
441: s = n->string;
442: if (0 == strcmp(s, "3"))
443: p = bsd_versions[0];
444: else if (0 == strcmp(s, "4"))
445: p = bsd_versions[1];
446: else if (0 == strcmp(s, "5"))
447: p = bsd_versions[2];
448: else if (0 == strcmp(s, "6"))
449: p = bsd_versions[3];
450: else if (0 == strcmp(s, "7"))
451: p = bsd_versions[4];
452: else
453: p = bsd_versions[0];
454: }
455:
1.57 schwarze 456: free(man->meta.source);
457: man->meta.source = mandoc_strdup(p);
1.34 schwarze 458: }
459:
1.78 ! schwarze 460: static void
1.34 schwarze 461: post_AT(CHKARGS)
462: {
463: static const char * const unix_versions[] = {
464: "7th Edition",
465: "System III",
466: "System V",
467: "System V Release 2",
468: };
469:
470: const char *p, *s;
471: struct man_node *nn;
472:
473: n = n->child;
474:
475: if (NULL == n || MAN_TEXT != n->type)
476: p = unix_versions[0];
477: else {
478: s = n->string;
479: if (0 == strcmp(s, "3"))
480: p = unix_versions[0];
481: else if (0 == strcmp(s, "4"))
482: p = unix_versions[1];
483: else if (0 == strcmp(s, "5")) {
484: nn = n->next;
485: if (nn && MAN_TEXT == nn->type && nn->string[0])
486: p = unix_versions[3];
487: else
488: p = unix_versions[2];
489: } else
490: p = unix_versions[0];
491: }
492:
1.57 schwarze 493: free(man->meta.source);
494: man->meta.source = mandoc_strdup(p);
1.46 schwarze 495: }
496:
1.78 ! schwarze 497: static void
1.46 schwarze 498: post_vs(CHKARGS)
499: {
1.76 schwarze 500:
501: if (n->tok == MAN_br)
502: check_eq0(man, n);
503: else
504: check_le1(man, n);
1.46 schwarze 505:
1.54 schwarze 506: if (NULL != n->prev)
1.78 ! schwarze 507: return;
1.54 schwarze 508:
509: switch (n->parent->tok) {
1.63 schwarze 510: case MAN_SH:
1.54 schwarze 511: /* FALLTHROUGH */
1.63 schwarze 512: case MAN_SS:
1.67 schwarze 513: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
514: "%s after %s", man_macronames[n->tok],
515: man_macronames[n->parent->tok]);
1.54 schwarze 516: /* FALLTHROUGH */
1.63 schwarze 517: case MAN_MAX:
518: /*
1.54 schwarze 519: * Don't warn about this because it occurs in pod2man
520: * and would cause considerable (unfixable) warnage.
521: */
1.57 schwarze 522: man_node_delete(man, n);
1.54 schwarze 523: break;
524: default:
525: break;
526: }
1.34 schwarze 527: }