Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.91
1.91 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.90 2015/04/19 13:59:37 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"
1.89 schwarze 34: #include "roff_int.h"
1.1 kristaps 35: #include "libman.h"
36:
1.88 schwarze 37: #define CHKARGS struct roff_man *man, struct roff_node *n
1.1 kristaps 38:
1.78 schwarze 39: typedef void (*v_check)(CHKARGS);
1.1 kristaps 40:
1.78 schwarze 41: static void check_par(CHKARGS);
42: static void check_part(CHKARGS);
43: static void check_root(CHKARGS);
44: static void check_text(CHKARGS);
45:
46: static void post_AT(CHKARGS);
47: static void post_IP(CHKARGS);
48: static void post_vs(CHKARGS);
49: static void post_fi(CHKARGS);
50: static void post_ft(CHKARGS);
51: static void post_nf(CHKARGS);
1.84 schwarze 52: static void post_OP(CHKARGS);
1.78 schwarze 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 */
1.83 schwarze 86: NULL, /* PD */
1.76 schwarze 87: post_AT, /* AT */
88: NULL, /* in */
89: post_ft, /* ft */
1.84 schwarze 90: post_OP, /* OP */
1.76 schwarze 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
1.88 schwarze 100: man_valid_post(struct roff_man *man)
1.1 kristaps 101: {
1.86 schwarze 102: struct roff_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) {
1.85 schwarze 111: case ROFFT_TEXT:
1.78 schwarze 112: check_text(man, n);
113: break;
1.85 schwarze 114: case ROFFT_ROOT:
1.78 schwarze 115: check_root(man, n);
116: break;
1.85 schwarze 117: case ROFFT_EQN:
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':
207: case '2':
208: case '3':
209: case '4':
210: case 'I':
211: case 'P':
212: case 'R':
1.32 schwarze 213: if ('\0' == cp[1])
214: ok = 1;
215: break;
1.63 schwarze 216: case 'B':
1.32 schwarze 217: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
218: ok = 1;
219: break;
1.63 schwarze 220: case 'C':
1.32 schwarze 221: if ('W' == cp[1] && '\0' == cp[2])
222: ok = 1;
223: break;
224: default:
225: break;
226: }
227:
228: if (0 == ok) {
1.70 schwarze 229: mandoc_vmsg(MANDOCERR_FT_BAD, man->parse,
230: n->line, n->pos, "ft %s", cp);
1.32 schwarze 231: *cp = '\0';
232: }
233: }
1.7 schwarze 234:
1.78 schwarze 235: static void
1.8 schwarze 236: check_part(CHKARGS)
237: {
238:
1.85 schwarze 239: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 240: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
241: n->line, n->pos, man_macronames[n->tok]);
1.8 schwarze 242: }
243:
1.78 schwarze 244: static void
1.36 schwarze 245: check_par(CHKARGS)
246: {
247:
1.39 schwarze 248: switch (n->type) {
1.85 schwarze 249: case ROFFT_BLOCK:
1.39 schwarze 250: if (0 == n->body->nchild)
1.89 schwarze 251: roff_node_delete(man, n);
1.39 schwarze 252: break;
1.85 schwarze 253: case ROFFT_BODY:
1.39 schwarze 254: if (0 == n->nchild)
1.67 schwarze 255: mandoc_vmsg(MANDOCERR_PAR_SKIP,
256: man->parse, n->line, n->pos,
257: "%s empty", man_macronames[n->tok]);
1.39 schwarze 258: break;
1.85 schwarze 259: case ROFFT_HEAD:
1.39 schwarze 260: if (n->nchild)
1.69 schwarze 261: mandoc_vmsg(MANDOCERR_ARG_SKIP,
262: man->parse, n->line, n->pos,
263: "%s %s%s", man_macronames[n->tok],
264: n->child->string,
265: n->nchild > 1 ? " ..." : "");
1.39 schwarze 266: break;
267: default:
268: break;
269: }
1.36 schwarze 270: }
271:
1.78 schwarze 272: static void
1.55 schwarze 273: post_IP(CHKARGS)
274: {
275:
276: switch (n->type) {
1.85 schwarze 277: case ROFFT_BLOCK:
1.55 schwarze 278: if (0 == n->head->nchild && 0 == n->body->nchild)
1.89 schwarze 279: roff_node_delete(man, n);
1.55 schwarze 280: break;
1.85 schwarze 281: case ROFFT_BODY:
1.55 schwarze 282: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.67 schwarze 283: mandoc_vmsg(MANDOCERR_PAR_SKIP,
284: man->parse, n->line, n->pos,
285: "%s empty", man_macronames[n->tok]);
1.55 schwarze 286: break;
287: default:
288: break;
289: }
290: }
1.36 schwarze 291:
1.78 schwarze 292: static void
1.34 schwarze 293: post_TH(CHKARGS)
294: {
1.86 schwarze 295: struct roff_node *nb;
1.40 schwarze 296: const char *p;
1.34 schwarze 297:
1.57 schwarze 298: free(man->meta.title);
299: free(man->meta.vol);
1.87 schwarze 300: free(man->meta.os);
1.57 schwarze 301: free(man->meta.msec);
302: free(man->meta.date);
1.34 schwarze 303:
1.57 schwarze 304: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 305: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 306:
1.64 schwarze 307: nb = n;
308:
1.87 schwarze 309: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 310:
311: n = n->child;
1.40 schwarze 312: if (n && n->string) {
313: for (p = n->string; '\0' != *p; p++) {
314: /* Only warn about this once... */
1.63 schwarze 315: if (isalpha((unsigned char)*p) &&
316: ! isupper((unsigned char)*p)) {
1.74 schwarze 317: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 318: man->parse, n->line,
319: n->pos + (p - n->string),
1.74 schwarze 320: "TH %s", n->string);
1.40 schwarze 321: break;
322: }
323: }
1.57 schwarze 324: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 325: } else {
1.57 schwarze 326: man->meta.title = mandoc_strdup("");
1.77 schwarze 327: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
328: nb->line, nb->pos, "TH");
329: }
1.34 schwarze 330:
1.87 schwarze 331: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 332:
1.40 schwarze 333: if (n)
334: n = n->next;
335: if (n && n->string)
1.57 schwarze 336: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 337: else {
1.57 schwarze 338: man->meta.msec = mandoc_strdup("");
1.77 schwarze 339: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
340: nb->line, nb->pos, "TH %s", man->meta.title);
341: }
1.34 schwarze 342:
1.87 schwarze 343: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 344:
1.40 schwarze 345: if (n)
346: n = n->next;
1.49 schwarze 347: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 348: man->meta.date = man->quick ?
349: mandoc_strdup(n->string) :
350: mandoc_normdate(man->parse, n->string,
351: n->line, n->pos);
1.64 schwarze 352: } else {
1.57 schwarze 353: man->meta.date = mandoc_strdup("");
1.74 schwarze 354: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
355: n ? n->line : nb->line,
356: n ? n->pos : nb->pos, "TH");
1.64 schwarze 357: }
1.34 schwarze 358:
1.87 schwarze 359: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 360:
361: if (n && (n = n->next))
1.87 schwarze 362: man->meta.os = mandoc_strdup(n->string);
1.79 schwarze 363: else if (man->defos != NULL)
1.87 schwarze 364: man->meta.os = mandoc_strdup(man->defos);
1.34 schwarze 365:
1.87 schwarze 366: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 367: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 368:
369: if (n && (n = n->next))
1.57 schwarze 370: man->meta.vol = mandoc_strdup(n->string);
371: else if ('\0' != man->meta.msec[0] &&
372: (NULL != (p = mandoc_a2msec(man->meta.msec))))
373: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 374:
375: if (n != NULL && (n = n->next) != NULL)
376: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
377: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 378:
379: /*
380: * Remove the `TH' node after we've processed it for our
381: * meta-data.
382: */
1.89 schwarze 383: roff_node_delete(man, man->last);
1.34 schwarze 384: }
385:
1.78 schwarze 386: static void
1.34 schwarze 387: post_nf(CHKARGS)
388: {
389:
1.82 schwarze 390: if (man->flags & MAN_LITERAL)
1.74 schwarze 391: mandoc_msg(MANDOCERR_NF_SKIP, man->parse,
392: n->line, n->pos, "nf");
1.34 schwarze 393:
1.57 schwarze 394: man->flags |= MAN_LITERAL;
1.34 schwarze 395: }
396:
1.78 schwarze 397: static void
1.34 schwarze 398: post_fi(CHKARGS)
399: {
400:
1.57 schwarze 401: if ( ! (MAN_LITERAL & man->flags))
1.74 schwarze 402: mandoc_msg(MANDOCERR_FI_SKIP, man->parse,
403: n->line, n->pos, "fi");
1.34 schwarze 404:
1.57 schwarze 405: man->flags &= ~MAN_LITERAL;
1.34 schwarze 406: }
407:
1.78 schwarze 408: static void
1.34 schwarze 409: post_UC(CHKARGS)
410: {
411: static const char * const bsd_versions[] = {
412: "3rd Berkeley Distribution",
413: "4th Berkeley Distribution",
414: "4.2 Berkeley Distribution",
415: "4.3 Berkeley Distribution",
416: "4.4 Berkeley Distribution",
417: };
418:
419: const char *p, *s;
420:
421: n = n->child;
422:
1.85 schwarze 423: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 424: p = bsd_versions[0];
425: else {
426: s = n->string;
427: if (0 == strcmp(s, "3"))
428: p = bsd_versions[0];
429: else if (0 == strcmp(s, "4"))
430: p = bsd_versions[1];
431: else if (0 == strcmp(s, "5"))
432: p = bsd_versions[2];
433: else if (0 == strcmp(s, "6"))
434: p = bsd_versions[3];
435: else if (0 == strcmp(s, "7"))
436: p = bsd_versions[4];
437: else
438: p = bsd_versions[0];
439: }
440:
1.87 schwarze 441: free(man->meta.os);
442: man->meta.os = mandoc_strdup(p);
1.34 schwarze 443: }
444:
1.78 schwarze 445: static void
1.34 schwarze 446: post_AT(CHKARGS)
447: {
448: static const char * const unix_versions[] = {
449: "7th Edition",
450: "System III",
451: "System V",
452: "System V Release 2",
453: };
454:
1.86 schwarze 455: struct roff_node *nn;
1.34 schwarze 456: const char *p, *s;
457:
458: n = n->child;
459:
1.85 schwarze 460: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 461: p = unix_versions[0];
462: else {
463: s = n->string;
464: if (0 == strcmp(s, "3"))
465: p = unix_versions[0];
466: else if (0 == strcmp(s, "4"))
467: p = unix_versions[1];
468: else if (0 == strcmp(s, "5")) {
469: nn = n->next;
1.85 schwarze 470: if (nn != NULL &&
471: nn->type == ROFFT_TEXT &&
472: nn->string[0] != '\0')
1.34 schwarze 473: p = unix_versions[3];
474: else
475: p = unix_versions[2];
476: } else
477: p = unix_versions[0];
478: }
479:
1.87 schwarze 480: free(man->meta.os);
481: man->meta.os = mandoc_strdup(p);
1.46 schwarze 482: }
483:
1.78 schwarze 484: static void
1.46 schwarze 485: post_vs(CHKARGS)
486: {
487:
1.54 schwarze 488: if (NULL != n->prev)
1.78 schwarze 489: return;
1.54 schwarze 490:
491: switch (n->parent->tok) {
1.63 schwarze 492: case MAN_SH:
493: case MAN_SS:
1.67 schwarze 494: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
495: "%s after %s", man_macronames[n->tok],
496: man_macronames[n->parent->tok]);
1.54 schwarze 497: /* FALLTHROUGH */
1.90 schwarze 498: case TOKEN_NONE:
1.63 schwarze 499: /*
1.54 schwarze 500: * Don't warn about this because it occurs in pod2man
501: * and would cause considerable (unfixable) warnage.
502: */
1.89 schwarze 503: roff_node_delete(man, n);
1.54 schwarze 504: break;
505: default:
506: break;
507: }
1.34 schwarze 508: }