Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.84
1.84 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.83 2015/02/06 09:38:22 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: *
1.2 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_par(CHKARGS);
40: static void check_part(CHKARGS);
41: static void check_root(CHKARGS);
42: static void check_text(CHKARGS);
43:
44: static void post_AT(CHKARGS);
45: static void post_IP(CHKARGS);
46: static void post_vs(CHKARGS);
47: static void post_fi(CHKARGS);
48: static void post_ft(CHKARGS);
49: static void post_nf(CHKARGS);
1.84 ! schwarze 50: static void post_OP(CHKARGS);
1.78 schwarze 51: static void post_TH(CHKARGS);
52: static void post_UC(CHKARGS);
53: static void post_UR(CHKARGS);
1.34 schwarze 54:
1.76 schwarze 55: static v_check man_valids[MAN_MAX] = {
56: post_vs, /* br */
57: post_TH, /* TH */
58: NULL, /* SH */
59: NULL, /* SS */
60: NULL, /* TP */
61: check_par, /* LP */
62: check_par, /* PP */
63: check_par, /* P */
64: post_IP, /* IP */
65: NULL, /* HP */
66: NULL, /* SM */
67: NULL, /* SB */
68: NULL, /* BI */
69: NULL, /* IB */
70: NULL, /* BR */
71: NULL, /* RB */
72: NULL, /* R */
73: NULL, /* B */
74: NULL, /* I */
75: NULL, /* IR */
76: NULL, /* RI */
77: post_vs, /* sp */
78: post_nf, /* nf */
79: post_fi, /* fi */
80: NULL, /* RE */
81: check_part, /* RS */
82: NULL, /* DT */
83: post_UC, /* UC */
1.83 schwarze 84: NULL, /* PD */
1.76 schwarze 85: post_AT, /* AT */
86: NULL, /* in */
87: post_ft, /* ft */
1.84 ! schwarze 88: post_OP, /* OP */
1.76 schwarze 89: post_nf, /* EX */
90: post_fi, /* EE */
91: post_UR, /* UR */
92: NULL, /* UE */
93: NULL, /* ll */
1.1 kristaps 94: };
95:
96:
1.78 schwarze 97: void
1.57 schwarze 98: man_valid_post(struct man *man)
1.1 kristaps 99: {
1.76 schwarze 100: struct man_node *n;
1.7 schwarze 101: v_check *cp;
1.1 kristaps 102:
1.76 schwarze 103: n = man->last;
104: if (n->flags & MAN_VALID)
1.78 schwarze 105: return;
1.76 schwarze 106: n->flags |= MAN_VALID;
1.1 kristaps 107:
1.76 schwarze 108: switch (n->type) {
1.63 schwarze 109: case MAN_TEXT:
1.78 schwarze 110: check_text(man, n);
111: break;
1.63 schwarze 112: case MAN_ROOT:
1.78 schwarze 113: check_root(man, n);
114: break;
1.63 schwarze 115: case MAN_EQN:
1.42 schwarze 116: /* FALLTHROUGH */
1.63 schwarze 117: case MAN_TBL:
1.78 schwarze 118: break;
1.1 kristaps 119: default:
1.76 schwarze 120: cp = man_valids + n->tok;
1.78 schwarze 121: if (*cp)
122: (*cp)(man, n);
123: break;
1.1 kristaps 124: }
125: }
126:
1.78 schwarze 127: static void
1.63 schwarze 128: check_root(CHKARGS)
1.4 schwarze 129: {
1.7 schwarze 130:
1.73 schwarze 131: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 132:
1.65 schwarze 133: if (NULL == man->first->child)
1.75 schwarze 134: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
135: n->line, n->pos, NULL);
1.65 schwarze 136: else
137: man->meta.hasbody = 1;
138:
139: if (NULL == man->meta.title) {
1.77 schwarze 140: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 141: n->line, n->pos, NULL);
1.35 schwarze 142:
1.18 schwarze 143: /*
144: * If a title hasn't been set, do so now (by
145: * implication, date and section also aren't set).
146: */
1.35 schwarze 147:
1.77 schwarze 148: man->meta.title = mandoc_strdup("");
149: man->meta.msec = mandoc_strdup("");
1.60 schwarze 150: man->meta.date = man->quick ? mandoc_strdup("") :
151: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 152: }
1.15 schwarze 153: }
154:
1.78 schwarze 155: static void
1.47 schwarze 156: check_text(CHKARGS)
1.4 schwarze 157: {
1.47 schwarze 158: char *cp, *p;
159:
1.57 schwarze 160: if (MAN_LITERAL & man->flags)
1.78 schwarze 161: return;
1.48 schwarze 162:
163: cp = n->string;
164: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 165: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
166: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 167: }
168:
1.84 ! schwarze 169: static void
! 170: post_OP(CHKARGS)
! 171: {
! 172:
! 173: if (n->nchild == 0)
! 174: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
! 175: n->line, n->pos, "OP");
! 176: else if (n->nchild > 2) {
! 177: n = n->child->next->next;
! 178: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
! 179: n->line, n->pos, "OP ... %s", n->string);
! 180: }
1.1 kristaps 181: }
182:
1.78 schwarze 183: static void
1.76 schwarze 184: post_UR(CHKARGS)
1.58 schwarze 185: {
186:
1.84 ! schwarze 187: if (n->type == MAN_HEAD && n->child == NULL)
! 188: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
! 189: n->line, n->pos, "UR");
1.78 schwarze 190: check_part(man, n);
1.58 schwarze 191: }
1.32 schwarze 192:
1.78 schwarze 193: static void
1.47 schwarze 194: post_ft(CHKARGS)
1.32 schwarze 195: {
196: char *cp;
197: int ok;
198:
199: if (0 == n->nchild)
1.78 schwarze 200: return;
1.32 schwarze 201:
202: ok = 0;
203: cp = n->child->string;
204: switch (*cp) {
1.63 schwarze 205: case '1':
1.32 schwarze 206: /* FALLTHROUGH */
1.63 schwarze 207: case '2':
1.32 schwarze 208: /* FALLTHROUGH */
1.63 schwarze 209: case '3':
1.32 schwarze 210: /* FALLTHROUGH */
1.63 schwarze 211: case '4':
1.32 schwarze 212: /* FALLTHROUGH */
1.63 schwarze 213: case 'I':
1.32 schwarze 214: /* FALLTHROUGH */
1.63 schwarze 215: case 'P':
1.32 schwarze 216: /* FALLTHROUGH */
1.63 schwarze 217: case 'R':
1.32 schwarze 218: if ('\0' == cp[1])
219: ok = 1;
220: break;
1.63 schwarze 221: case 'B':
1.32 schwarze 222: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
223: ok = 1;
224: break;
1.63 schwarze 225: case 'C':
1.32 schwarze 226: if ('W' == cp[1] && '\0' == cp[2])
227: ok = 1;
228: break;
229: default:
230: break;
231: }
232:
233: if (0 == ok) {
1.70 schwarze 234: mandoc_vmsg(MANDOCERR_FT_BAD, man->parse,
235: n->line, n->pos, "ft %s", cp);
1.32 schwarze 236: *cp = '\0';
237: }
238: }
1.7 schwarze 239:
1.78 schwarze 240: static void
1.8 schwarze 241: check_part(CHKARGS)
242: {
243:
1.81 schwarze 244: if (n->type == MAN_BODY && n->child == NULL)
245: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
246: n->line, n->pos, man_macronames[n->tok]);
1.8 schwarze 247: }
248:
1.78 schwarze 249: static void
1.36 schwarze 250: check_par(CHKARGS)
251: {
252:
1.39 schwarze 253: switch (n->type) {
1.63 schwarze 254: case MAN_BLOCK:
1.39 schwarze 255: if (0 == n->body->nchild)
1.57 schwarze 256: man_node_delete(man, n);
1.39 schwarze 257: break;
1.63 schwarze 258: case MAN_BODY:
1.39 schwarze 259: if (0 == n->nchild)
1.67 schwarze 260: mandoc_vmsg(MANDOCERR_PAR_SKIP,
261: man->parse, n->line, n->pos,
262: "%s empty", man_macronames[n->tok]);
1.39 schwarze 263: break;
1.63 schwarze 264: case MAN_HEAD:
1.39 schwarze 265: if (n->nchild)
1.69 schwarze 266: mandoc_vmsg(MANDOCERR_ARG_SKIP,
267: man->parse, n->line, n->pos,
268: "%s %s%s", man_macronames[n->tok],
269: n->child->string,
270: n->nchild > 1 ? " ..." : "");
1.39 schwarze 271: break;
272: default:
273: break;
274: }
1.36 schwarze 275: }
276:
1.78 schwarze 277: static void
1.55 schwarze 278: post_IP(CHKARGS)
279: {
280:
281: switch (n->type) {
1.63 schwarze 282: case MAN_BLOCK:
1.55 schwarze 283: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 schwarze 284: man_node_delete(man, n);
1.55 schwarze 285: break;
1.63 schwarze 286: case MAN_BODY:
1.55 schwarze 287: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.67 schwarze 288: mandoc_vmsg(MANDOCERR_PAR_SKIP,
289: man->parse, n->line, n->pos,
290: "%s empty", man_macronames[n->tok]);
1.55 schwarze 291: break;
292: default:
293: break;
294: }
295: }
1.36 schwarze 296:
1.78 schwarze 297: static void
1.34 schwarze 298: post_TH(CHKARGS)
299: {
1.64 schwarze 300: struct man_node *nb;
1.40 schwarze 301: const char *p;
1.34 schwarze 302:
1.57 schwarze 303: free(man->meta.title);
304: free(man->meta.vol);
305: free(man->meta.source);
306: free(man->meta.msec);
307: free(man->meta.date);
1.34 schwarze 308:
1.57 schwarze 309: man->meta.title = man->meta.vol = man->meta.date =
1.63 schwarze 310: man->meta.msec = man->meta.source = NULL;
1.34 schwarze 311:
1.64 schwarze 312: nb = n;
313:
1.34 schwarze 314: /* ->TITLE<- MSEC DATE SOURCE VOL */
315:
316: n = n->child;
1.40 schwarze 317: if (n && n->string) {
318: for (p = n->string; '\0' != *p; p++) {
319: /* Only warn about this once... */
1.63 schwarze 320: if (isalpha((unsigned char)*p) &&
321: ! isupper((unsigned char)*p)) {
1.74 schwarze 322: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 323: man->parse, n->line,
324: n->pos + (p - n->string),
1.74 schwarze 325: "TH %s", n->string);
1.40 schwarze 326: break;
327: }
328: }
1.57 schwarze 329: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 330: } else {
1.57 schwarze 331: man->meta.title = mandoc_strdup("");
1.77 schwarze 332: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
333: nb->line, nb->pos, "TH");
334: }
1.34 schwarze 335:
336: /* TITLE ->MSEC<- DATE SOURCE VOL */
337:
1.40 schwarze 338: if (n)
339: n = n->next;
340: if (n && n->string)
1.57 schwarze 341: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 342: else {
1.57 schwarze 343: man->meta.msec = mandoc_strdup("");
1.77 schwarze 344: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
345: nb->line, nb->pos, "TH %s", man->meta.title);
346: }
1.34 schwarze 347:
348: /* TITLE MSEC ->DATE<- SOURCE VOL */
349:
1.40 schwarze 350: if (n)
351: n = n->next;
1.49 schwarze 352: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 353: man->meta.date = man->quick ?
354: mandoc_strdup(n->string) :
355: mandoc_normdate(man->parse, n->string,
356: n->line, n->pos);
1.64 schwarze 357: } else {
1.57 schwarze 358: man->meta.date = mandoc_strdup("");
1.74 schwarze 359: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
360: n ? n->line : nb->line,
361: n ? n->pos : nb->pos, "TH");
1.64 schwarze 362: }
1.34 schwarze 363:
364: /* TITLE MSEC DATE ->SOURCE<- VOL */
365:
366: if (n && (n = n->next))
1.57 schwarze 367: man->meta.source = mandoc_strdup(n->string);
1.79 schwarze 368: else if (man->defos != NULL)
369: man->meta.source = mandoc_strdup(man->defos);
1.34 schwarze 370:
371: /* TITLE MSEC DATE SOURCE ->VOL<- */
1.51 schwarze 372: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 373:
374: if (n && (n = n->next))
1.57 schwarze 375: man->meta.vol = mandoc_strdup(n->string);
376: else if ('\0' != man->meta.msec[0] &&
377: (NULL != (p = mandoc_a2msec(man->meta.msec))))
378: man->meta.vol = mandoc_strdup(p);
1.84 ! schwarze 379:
! 380: if (n != NULL && (n = n->next) != NULL)
! 381: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
! 382: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 383:
384: /*
385: * Remove the `TH' node after we've processed it for our
386: * meta-data.
387: */
1.57 schwarze 388: man_node_delete(man, man->last);
1.34 schwarze 389: }
390:
1.78 schwarze 391: static void
1.34 schwarze 392: post_nf(CHKARGS)
393: {
394:
1.82 schwarze 395: if (man->flags & MAN_LITERAL)
1.74 schwarze 396: mandoc_msg(MANDOCERR_NF_SKIP, man->parse,
397: n->line, n->pos, "nf");
1.34 schwarze 398:
1.57 schwarze 399: man->flags |= MAN_LITERAL;
1.34 schwarze 400: }
401:
1.78 schwarze 402: static void
1.34 schwarze 403: post_fi(CHKARGS)
404: {
405:
1.57 schwarze 406: if ( ! (MAN_LITERAL & man->flags))
1.74 schwarze 407: mandoc_msg(MANDOCERR_FI_SKIP, man->parse,
408: n->line, n->pos, "fi");
1.34 schwarze 409:
1.57 schwarze 410: man->flags &= ~MAN_LITERAL;
1.34 schwarze 411: }
412:
1.78 schwarze 413: static void
1.34 schwarze 414: post_UC(CHKARGS)
415: {
416: static const char * const bsd_versions[] = {
417: "3rd Berkeley Distribution",
418: "4th Berkeley Distribution",
419: "4.2 Berkeley Distribution",
420: "4.3 Berkeley Distribution",
421: "4.4 Berkeley Distribution",
422: };
423:
424: const char *p, *s;
425:
426: n = n->child;
427:
428: if (NULL == n || MAN_TEXT != n->type)
429: p = bsd_versions[0];
430: else {
431: s = n->string;
432: if (0 == strcmp(s, "3"))
433: p = bsd_versions[0];
434: else if (0 == strcmp(s, "4"))
435: p = bsd_versions[1];
436: else if (0 == strcmp(s, "5"))
437: p = bsd_versions[2];
438: else if (0 == strcmp(s, "6"))
439: p = bsd_versions[3];
440: else if (0 == strcmp(s, "7"))
441: p = bsd_versions[4];
442: else
443: p = bsd_versions[0];
444: }
445:
1.57 schwarze 446: free(man->meta.source);
447: man->meta.source = mandoc_strdup(p);
1.34 schwarze 448: }
449:
1.78 schwarze 450: static void
1.34 schwarze 451: post_AT(CHKARGS)
452: {
453: static const char * const unix_versions[] = {
454: "7th Edition",
455: "System III",
456: "System V",
457: "System V Release 2",
458: };
459:
460: const char *p, *s;
461: struct man_node *nn;
462:
463: n = n->child;
464:
465: if (NULL == n || MAN_TEXT != n->type)
466: p = unix_versions[0];
467: else {
468: s = n->string;
469: if (0 == strcmp(s, "3"))
470: p = unix_versions[0];
471: else if (0 == strcmp(s, "4"))
472: p = unix_versions[1];
473: else if (0 == strcmp(s, "5")) {
474: nn = n->next;
475: if (nn && MAN_TEXT == nn->type && nn->string[0])
476: p = unix_versions[3];
477: else
478: p = unix_versions[2];
479: } else
480: p = unix_versions[0];
481: }
482:
1.57 schwarze 483: free(man->meta.source);
484: man->meta.source = mandoc_strdup(p);
1.46 schwarze 485: }
486:
1.78 schwarze 487: static void
1.46 schwarze 488: post_vs(CHKARGS)
489: {
490:
1.54 schwarze 491: if (NULL != n->prev)
1.78 schwarze 492: return;
1.54 schwarze 493:
494: switch (n->parent->tok) {
1.63 schwarze 495: case MAN_SH:
1.54 schwarze 496: /* FALLTHROUGH */
1.63 schwarze 497: case MAN_SS:
1.67 schwarze 498: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
499: "%s after %s", man_macronames[n->tok],
500: man_macronames[n->parent->tok]);
1.54 schwarze 501: /* FALLTHROUGH */
1.63 schwarze 502: case MAN_MAX:
503: /*
1.54 schwarze 504: * Don't warn about this because it occurs in pod2man
505: * and would cause considerable (unfixable) warnage.
506: */
1.57 schwarze 507: man_node_delete(man, n);
1.54 schwarze 508: break;
509: default:
510: break;
511: }
1.34 schwarze 512: }