Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.94
1.94 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.93 2016/01/08 17:48:04 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.94 ! schwarze 4: * Copyright (c) 2010, 2012-2017 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_ft(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:
1.94 ! schwarze 55: static const v_check __man_valids[MAN_MAX - MAN_TH] = {
1.76 schwarze 56: post_TH, /* TH */
57: NULL, /* SH */
58: NULL, /* SS */
59: NULL, /* TP */
60: check_par, /* LP */
61: check_par, /* PP */
62: check_par, /* P */
63: post_IP, /* IP */
64: NULL, /* HP */
65: NULL, /* SM */
66: NULL, /* SB */
67: NULL, /* BI */
68: NULL, /* IB */
69: NULL, /* BR */
70: NULL, /* RB */
71: NULL, /* R */
72: NULL, /* B */
73: NULL, /* I */
74: NULL, /* IR */
75: NULL, /* RI */
1.94 ! schwarze 76: post_vs, /* br */
1.76 schwarze 77: post_vs, /* sp */
1.92 schwarze 78: NULL, /* nf */
79: NULL, /* fi */
1.76 schwarze 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 */
1.92 schwarze 89: NULL, /* EX */
90: NULL, /* EE */
1.76 schwarze 91: post_UR, /* UR */
92: NULL, /* UE */
93: NULL, /* ll */
1.1 kristaps 94: };
1.94 ! schwarze 95: static const v_check *man_valids = __man_valids - MAN_TH;
1.1 kristaps 96:
97:
1.78 schwarze 98: void
1.92 schwarze 99: man_node_validate(struct roff_man *man)
1.1 kristaps 100: {
1.86 schwarze 101: struct roff_node *n;
1.94 ! schwarze 102: const v_check *cp;
1.1 kristaps 103:
1.76 schwarze 104: n = man->last;
1.92 schwarze 105: man->last = man->last->child;
106: while (man->last != NULL) {
107: man_node_validate(man);
108: if (man->last == n)
109: man->last = man->last->child;
110: else
111: man->last = man->last->next;
112: }
1.1 kristaps 113:
1.92 schwarze 114: man->last = n;
115: man->next = ROFF_NEXT_SIBLING;
1.76 schwarze 116: switch (n->type) {
1.85 schwarze 117: case ROFFT_TEXT:
1.78 schwarze 118: check_text(man, n);
119: break;
1.85 schwarze 120: case ROFFT_ROOT:
1.78 schwarze 121: check_root(man, n);
122: break;
1.85 schwarze 123: case ROFFT_EQN:
124: case ROFFT_TBL:
1.78 schwarze 125: break;
1.1 kristaps 126: default:
1.76 schwarze 127: cp = man_valids + n->tok;
1.78 schwarze 128: if (*cp)
129: (*cp)(man, n);
1.92 schwarze 130: if (man->last == n)
131: man_state(man, n);
1.78 schwarze 132: break;
1.1 kristaps 133: }
134: }
135:
1.78 schwarze 136: static void
1.63 schwarze 137: check_root(CHKARGS)
1.4 schwarze 138: {
1.7 schwarze 139:
1.73 schwarze 140: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 141:
1.65 schwarze 142: if (NULL == man->first->child)
1.75 schwarze 143: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
144: n->line, n->pos, NULL);
1.65 schwarze 145: else
146: man->meta.hasbody = 1;
147:
148: if (NULL == man->meta.title) {
1.77 schwarze 149: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 150: n->line, n->pos, NULL);
1.35 schwarze 151:
1.18 schwarze 152: /*
153: * If a title hasn't been set, do so now (by
154: * implication, date and section also aren't set).
155: */
1.35 schwarze 156:
1.77 schwarze 157: man->meta.title = mandoc_strdup("");
158: man->meta.msec = mandoc_strdup("");
1.60 schwarze 159: man->meta.date = man->quick ? mandoc_strdup("") :
160: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 161: }
1.15 schwarze 162: }
163:
1.78 schwarze 164: static void
1.47 schwarze 165: check_text(CHKARGS)
1.4 schwarze 166: {
1.47 schwarze 167: char *cp, *p;
168:
1.57 schwarze 169: if (MAN_LITERAL & man->flags)
1.78 schwarze 170: return;
1.48 schwarze 171:
172: cp = n->string;
173: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 174: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
175: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 176: }
177:
1.84 schwarze 178: static void
179: post_OP(CHKARGS)
180: {
181:
1.93 schwarze 182: if (n->child == NULL)
1.84 schwarze 183: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
184: n->line, n->pos, "OP");
1.93 schwarze 185: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 186: n = n->child->next->next;
187: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
188: n->line, n->pos, "OP ... %s", n->string);
189: }
1.1 kristaps 190: }
191:
1.78 schwarze 192: static void
1.76 schwarze 193: post_UR(CHKARGS)
1.58 schwarze 194: {
195:
1.85 schwarze 196: if (n->type == ROFFT_HEAD && n->child == NULL)
1.84 schwarze 197: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
198: n->line, n->pos, "UR");
1.78 schwarze 199: check_part(man, n);
1.58 schwarze 200: }
1.32 schwarze 201:
1.78 schwarze 202: static void
1.47 schwarze 203: post_ft(CHKARGS)
1.32 schwarze 204: {
205: char *cp;
206: int ok;
207:
1.93 schwarze 208: if (n->child == NULL)
1.78 schwarze 209: return;
1.32 schwarze 210:
211: ok = 0;
212: cp = n->child->string;
213: switch (*cp) {
1.63 schwarze 214: case '1':
215: case '2':
216: case '3':
217: case '4':
218: case 'I':
219: case 'P':
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: }
1.7 schwarze 242:
1.78 schwarze 243: static void
1.8 schwarze 244: check_part(CHKARGS)
245: {
246:
1.85 schwarze 247: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 248: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.94 ! schwarze 249: n->line, n->pos, roff_name[n->tok]);
1.8 schwarze 250: }
251:
1.78 schwarze 252: static void
1.36 schwarze 253: check_par(CHKARGS)
254: {
255:
1.39 schwarze 256: switch (n->type) {
1.85 schwarze 257: case ROFFT_BLOCK:
1.93 schwarze 258: if (n->body->child == NULL)
1.89 schwarze 259: roff_node_delete(man, n);
1.39 schwarze 260: break;
1.85 schwarze 261: case ROFFT_BODY:
1.93 schwarze 262: if (n->child == NULL)
1.67 schwarze 263: mandoc_vmsg(MANDOCERR_PAR_SKIP,
264: man->parse, n->line, n->pos,
1.94 ! schwarze 265: "%s empty", roff_name[n->tok]);
1.39 schwarze 266: break;
1.85 schwarze 267: case ROFFT_HEAD:
1.93 schwarze 268: if (n->child != NULL)
1.69 schwarze 269: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.94 ! schwarze 270: man->parse, n->line, n->pos, "%s %s%s",
! 271: roff_name[n->tok], n->child->string,
1.93 schwarze 272: n->child->next != NULL ? " ..." : "");
1.39 schwarze 273: break;
274: default:
275: break;
276: }
1.36 schwarze 277: }
278:
1.78 schwarze 279: static void
1.55 schwarze 280: post_IP(CHKARGS)
281: {
282:
283: switch (n->type) {
1.85 schwarze 284: case ROFFT_BLOCK:
1.93 schwarze 285: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 286: roff_node_delete(man, n);
1.55 schwarze 287: break;
1.85 schwarze 288: case ROFFT_BODY:
1.93 schwarze 289: if (n->parent->head->child == NULL && n->child == NULL)
1.67 schwarze 290: mandoc_vmsg(MANDOCERR_PAR_SKIP,
291: man->parse, n->line, n->pos,
1.94 ! schwarze 292: "%s empty", roff_name[n->tok]);
1.55 schwarze 293: break;
294: default:
295: break;
296: }
297: }
1.36 schwarze 298:
1.78 schwarze 299: static void
1.34 schwarze 300: post_TH(CHKARGS)
301: {
1.86 schwarze 302: struct roff_node *nb;
1.40 schwarze 303: const char *p;
1.34 schwarze 304:
1.57 schwarze 305: free(man->meta.title);
306: free(man->meta.vol);
1.87 schwarze 307: free(man->meta.os);
1.57 schwarze 308: free(man->meta.msec);
309: free(man->meta.date);
1.34 schwarze 310:
1.57 schwarze 311: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 312: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 313:
1.64 schwarze 314: nb = n;
315:
1.87 schwarze 316: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 317:
318: n = n->child;
1.40 schwarze 319: if (n && n->string) {
320: for (p = n->string; '\0' != *p; p++) {
321: /* Only warn about this once... */
1.63 schwarze 322: if (isalpha((unsigned char)*p) &&
323: ! isupper((unsigned char)*p)) {
1.74 schwarze 324: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 325: man->parse, n->line,
326: n->pos + (p - n->string),
1.74 schwarze 327: "TH %s", n->string);
1.40 schwarze 328: break;
329: }
330: }
1.57 schwarze 331: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 332: } else {
1.57 schwarze 333: man->meta.title = mandoc_strdup("");
1.77 schwarze 334: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
335: nb->line, nb->pos, "TH");
336: }
1.34 schwarze 337:
1.87 schwarze 338: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 339:
1.40 schwarze 340: if (n)
341: n = n->next;
342: if (n && n->string)
1.57 schwarze 343: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 344: else {
1.57 schwarze 345: man->meta.msec = mandoc_strdup("");
1.77 schwarze 346: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
347: nb->line, nb->pos, "TH %s", man->meta.title);
348: }
1.34 schwarze 349:
1.87 schwarze 350: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 351:
1.40 schwarze 352: if (n)
353: n = n->next;
1.49 schwarze 354: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 355: man->meta.date = man->quick ?
356: mandoc_strdup(n->string) :
357: mandoc_normdate(man->parse, n->string,
358: n->line, n->pos);
1.64 schwarze 359: } else {
1.57 schwarze 360: man->meta.date = mandoc_strdup("");
1.74 schwarze 361: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
362: n ? n->line : nb->line,
363: n ? n->pos : nb->pos, "TH");
1.64 schwarze 364: }
1.34 schwarze 365:
1.87 schwarze 366: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 367:
368: if (n && (n = n->next))
1.87 schwarze 369: man->meta.os = mandoc_strdup(n->string);
1.79 schwarze 370: else if (man->defos != NULL)
1.87 schwarze 371: man->meta.os = mandoc_strdup(man->defos);
1.34 schwarze 372:
1.87 schwarze 373: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 374: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 375:
376: if (n && (n = n->next))
1.57 schwarze 377: man->meta.vol = mandoc_strdup(n->string);
378: else if ('\0' != man->meta.msec[0] &&
379: (NULL != (p = mandoc_a2msec(man->meta.msec))))
380: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 381:
382: if (n != NULL && (n = n->next) != NULL)
383: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
384: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 385:
386: /*
387: * Remove the `TH' node after we've processed it for our
388: * meta-data.
389: */
1.89 schwarze 390: roff_node_delete(man, man->last);
1.34 schwarze 391: }
392:
1.78 schwarze 393: static void
1.34 schwarze 394: post_UC(CHKARGS)
395: {
396: static const char * const bsd_versions[] = {
397: "3rd Berkeley Distribution",
398: "4th Berkeley Distribution",
399: "4.2 Berkeley Distribution",
400: "4.3 Berkeley Distribution",
401: "4.4 Berkeley Distribution",
402: };
403:
404: const char *p, *s;
405:
406: n = n->child;
407:
1.85 schwarze 408: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 409: p = bsd_versions[0];
410: else {
411: s = n->string;
412: if (0 == strcmp(s, "3"))
413: p = bsd_versions[0];
414: else if (0 == strcmp(s, "4"))
415: p = bsd_versions[1];
416: else if (0 == strcmp(s, "5"))
417: p = bsd_versions[2];
418: else if (0 == strcmp(s, "6"))
419: p = bsd_versions[3];
420: else if (0 == strcmp(s, "7"))
421: p = bsd_versions[4];
422: else
423: p = bsd_versions[0];
424: }
425:
1.87 schwarze 426: free(man->meta.os);
427: man->meta.os = mandoc_strdup(p);
1.34 schwarze 428: }
429:
1.78 schwarze 430: static void
1.34 schwarze 431: post_AT(CHKARGS)
432: {
433: static const char * const unix_versions[] = {
434: "7th Edition",
435: "System III",
436: "System V",
437: "System V Release 2",
438: };
439:
1.86 schwarze 440: struct roff_node *nn;
1.34 schwarze 441: const char *p, *s;
442:
443: n = n->child;
444:
1.85 schwarze 445: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 446: p = unix_versions[0];
447: else {
448: s = n->string;
449: if (0 == strcmp(s, "3"))
450: p = unix_versions[0];
451: else if (0 == strcmp(s, "4"))
452: p = unix_versions[1];
453: else if (0 == strcmp(s, "5")) {
454: nn = n->next;
1.85 schwarze 455: if (nn != NULL &&
456: nn->type == ROFFT_TEXT &&
457: nn->string[0] != '\0')
1.34 schwarze 458: p = unix_versions[3];
459: else
460: p = unix_versions[2];
461: } else
462: p = unix_versions[0];
463: }
464:
1.87 schwarze 465: free(man->meta.os);
466: man->meta.os = mandoc_strdup(p);
1.46 schwarze 467: }
468:
1.78 schwarze 469: static void
1.46 schwarze 470: post_vs(CHKARGS)
471: {
472:
1.54 schwarze 473: if (NULL != n->prev)
1.78 schwarze 474: return;
1.54 schwarze 475:
476: switch (n->parent->tok) {
1.63 schwarze 477: case MAN_SH:
478: case MAN_SS:
1.67 schwarze 479: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.94 ! schwarze 480: "%s after %s", roff_name[n->tok],
! 481: roff_name[n->parent->tok]);
1.54 schwarze 482: /* FALLTHROUGH */
1.90 schwarze 483: case TOKEN_NONE:
1.63 schwarze 484: /*
1.54 schwarze 485: * Don't warn about this because it occurs in pod2man
486: * and would cause considerable (unfixable) warnage.
487: */
1.89 schwarze 488: roff_node_delete(man, n);
1.54 schwarze 489: break;
490: default:
491: break;
492: }
1.34 schwarze 493: }