Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.95
1.95 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.94 2017/04/24 23:06:09 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 */
76: post_vs, /* sp */
1.92 schwarze 77: NULL, /* nf */
78: NULL, /* fi */
1.76 schwarze 79: NULL, /* RE */
80: check_part, /* RS */
81: NULL, /* DT */
82: post_UC, /* UC */
1.83 schwarze 83: NULL, /* PD */
1.76 schwarze 84: post_AT, /* AT */
85: NULL, /* in */
86: post_ft, /* ft */
1.84 schwarze 87: post_OP, /* OP */
1.92 schwarze 88: NULL, /* EX */
89: NULL, /* EE */
1.76 schwarze 90: post_UR, /* UR */
91: NULL, /* UE */
92: NULL, /* ll */
1.1 kristaps 93: };
1.94 schwarze 94: static const v_check *man_valids = __man_valids - MAN_TH;
1.1 kristaps 95:
96:
1.78 schwarze 97: void
1.92 schwarze 98: man_node_validate(struct roff_man *man)
1.1 kristaps 99: {
1.86 schwarze 100: struct roff_node *n;
1.94 schwarze 101: const v_check *cp;
1.1 kristaps 102:
1.76 schwarze 103: n = man->last;
1.92 schwarze 104: man->last = man->last->child;
105: while (man->last != NULL) {
106: man_node_validate(man);
107: if (man->last == n)
108: man->last = man->last->child;
109: else
110: man->last = man->last->next;
111: }
1.1 kristaps 112:
1.92 schwarze 113: man->last = n;
114: man->next = ROFF_NEXT_SIBLING;
1.76 schwarze 115: switch (n->type) {
1.85 schwarze 116: case ROFFT_TEXT:
1.78 schwarze 117: check_text(man, n);
118: break;
1.85 schwarze 119: case ROFFT_ROOT:
1.78 schwarze 120: check_root(man, n);
121: break;
1.85 schwarze 122: case ROFFT_EQN:
123: case ROFFT_TBL:
1.78 schwarze 124: break;
1.1 kristaps 125: default:
1.95 ! schwarze 126: if (n->tok < ROFF_MAX) {
! 127: switch (n->tok) {
! 128: case ROFF_br:
! 129: post_vs(man, n);
! 130: break;
! 131: default:
! 132: abort();
! 133: }
! 134: break;
! 135: }
! 136: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.76 schwarze 137: cp = man_valids + n->tok;
1.78 schwarze 138: if (*cp)
139: (*cp)(man, n);
1.92 schwarze 140: if (man->last == n)
141: man_state(man, n);
1.78 schwarze 142: break;
1.1 kristaps 143: }
144: }
145:
1.78 schwarze 146: static void
1.63 schwarze 147: check_root(CHKARGS)
1.4 schwarze 148: {
1.7 schwarze 149:
1.73 schwarze 150: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 151:
1.65 schwarze 152: if (NULL == man->first->child)
1.75 schwarze 153: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
154: n->line, n->pos, NULL);
1.65 schwarze 155: else
156: man->meta.hasbody = 1;
157:
158: if (NULL == man->meta.title) {
1.77 schwarze 159: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 160: n->line, n->pos, NULL);
1.35 schwarze 161:
1.18 schwarze 162: /*
163: * If a title hasn't been set, do so now (by
164: * implication, date and section also aren't set).
165: */
1.35 schwarze 166:
1.77 schwarze 167: man->meta.title = mandoc_strdup("");
168: man->meta.msec = mandoc_strdup("");
1.60 schwarze 169: man->meta.date = man->quick ? mandoc_strdup("") :
170: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 171: }
1.15 schwarze 172: }
173:
1.78 schwarze 174: static void
1.47 schwarze 175: check_text(CHKARGS)
1.4 schwarze 176: {
1.47 schwarze 177: char *cp, *p;
178:
1.57 schwarze 179: if (MAN_LITERAL & man->flags)
1.78 schwarze 180: return;
1.48 schwarze 181:
182: cp = n->string;
183: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 184: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
185: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 186: }
187:
1.84 schwarze 188: static void
189: post_OP(CHKARGS)
190: {
191:
1.93 schwarze 192: if (n->child == NULL)
1.84 schwarze 193: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
194: n->line, n->pos, "OP");
1.93 schwarze 195: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 196: n = n->child->next->next;
197: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
198: n->line, n->pos, "OP ... %s", n->string);
199: }
1.1 kristaps 200: }
201:
1.78 schwarze 202: static void
1.76 schwarze 203: post_UR(CHKARGS)
1.58 schwarze 204: {
205:
1.85 schwarze 206: if (n->type == ROFFT_HEAD && n->child == NULL)
1.84 schwarze 207: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
208: n->line, n->pos, "UR");
1.78 schwarze 209: check_part(man, n);
1.58 schwarze 210: }
1.32 schwarze 211:
1.78 schwarze 212: static void
1.47 schwarze 213: post_ft(CHKARGS)
1.32 schwarze 214: {
215: char *cp;
216: int ok;
217:
1.93 schwarze 218: if (n->child == NULL)
1.78 schwarze 219: return;
1.32 schwarze 220:
221: ok = 0;
222: cp = n->child->string;
223: switch (*cp) {
1.63 schwarze 224: case '1':
225: case '2':
226: case '3':
227: case '4':
228: case 'I':
229: case 'P':
230: case 'R':
1.32 schwarze 231: if ('\0' == cp[1])
232: ok = 1;
233: break;
1.63 schwarze 234: case 'B':
1.32 schwarze 235: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
236: ok = 1;
237: break;
1.63 schwarze 238: case 'C':
1.32 schwarze 239: if ('W' == cp[1] && '\0' == cp[2])
240: ok = 1;
241: break;
242: default:
243: break;
244: }
245:
246: if (0 == ok) {
1.70 schwarze 247: mandoc_vmsg(MANDOCERR_FT_BAD, man->parse,
248: n->line, n->pos, "ft %s", cp);
1.32 schwarze 249: *cp = '\0';
250: }
251: }
1.7 schwarze 252:
1.78 schwarze 253: static void
1.8 schwarze 254: check_part(CHKARGS)
255: {
256:
1.85 schwarze 257: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 258: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.94 schwarze 259: n->line, n->pos, roff_name[n->tok]);
1.8 schwarze 260: }
261:
1.78 schwarze 262: static void
1.36 schwarze 263: check_par(CHKARGS)
264: {
265:
1.39 schwarze 266: switch (n->type) {
1.85 schwarze 267: case ROFFT_BLOCK:
1.93 schwarze 268: if (n->body->child == NULL)
1.89 schwarze 269: roff_node_delete(man, n);
1.39 schwarze 270: break;
1.85 schwarze 271: case ROFFT_BODY:
1.93 schwarze 272: if (n->child == NULL)
1.67 schwarze 273: mandoc_vmsg(MANDOCERR_PAR_SKIP,
274: man->parse, n->line, n->pos,
1.94 schwarze 275: "%s empty", roff_name[n->tok]);
1.39 schwarze 276: break;
1.85 schwarze 277: case ROFFT_HEAD:
1.93 schwarze 278: if (n->child != NULL)
1.69 schwarze 279: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.94 schwarze 280: man->parse, n->line, n->pos, "%s %s%s",
281: roff_name[n->tok], n->child->string,
1.93 schwarze 282: n->child->next != NULL ? " ..." : "");
1.39 schwarze 283: break;
284: default:
285: break;
286: }
1.36 schwarze 287: }
288:
1.78 schwarze 289: static void
1.55 schwarze 290: post_IP(CHKARGS)
291: {
292:
293: switch (n->type) {
1.85 schwarze 294: case ROFFT_BLOCK:
1.93 schwarze 295: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 296: roff_node_delete(man, n);
1.55 schwarze 297: break;
1.85 schwarze 298: case ROFFT_BODY:
1.93 schwarze 299: if (n->parent->head->child == NULL && n->child == NULL)
1.67 schwarze 300: mandoc_vmsg(MANDOCERR_PAR_SKIP,
301: man->parse, n->line, n->pos,
1.94 schwarze 302: "%s empty", roff_name[n->tok]);
1.55 schwarze 303: break;
304: default:
305: break;
306: }
307: }
1.36 schwarze 308:
1.78 schwarze 309: static void
1.34 schwarze 310: post_TH(CHKARGS)
311: {
1.86 schwarze 312: struct roff_node *nb;
1.40 schwarze 313: const char *p;
1.34 schwarze 314:
1.57 schwarze 315: free(man->meta.title);
316: free(man->meta.vol);
1.87 schwarze 317: free(man->meta.os);
1.57 schwarze 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 =
1.87 schwarze 322: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 323:
1.64 schwarze 324: nb = n;
325:
1.87 schwarze 326: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 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:
1.87 schwarze 348: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 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:
1.87 schwarze 360: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 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:
1.87 schwarze 376: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 377:
378: if (n && (n = n->next))
1.87 schwarze 379: man->meta.os = mandoc_strdup(n->string);
1.79 schwarze 380: else if (man->defos != NULL)
1.87 schwarze 381: man->meta.os = mandoc_strdup(man->defos);
1.34 schwarze 382:
1.87 schwarze 383: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 384: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 385:
386: if (n && (n = n->next))
1.57 schwarze 387: man->meta.vol = mandoc_strdup(n->string);
388: else if ('\0' != man->meta.msec[0] &&
389: (NULL != (p = mandoc_a2msec(man->meta.msec))))
390: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 391:
392: if (n != NULL && (n = n->next) != NULL)
393: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
394: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 395:
396: /*
397: * Remove the `TH' node after we've processed it for our
398: * meta-data.
399: */
1.89 schwarze 400: roff_node_delete(man, man->last);
1.34 schwarze 401: }
402:
1.78 schwarze 403: static void
1.34 schwarze 404: post_UC(CHKARGS)
405: {
406: static const char * const bsd_versions[] = {
407: "3rd Berkeley Distribution",
408: "4th Berkeley Distribution",
409: "4.2 Berkeley Distribution",
410: "4.3 Berkeley Distribution",
411: "4.4 Berkeley Distribution",
412: };
413:
414: const char *p, *s;
415:
416: n = n->child;
417:
1.85 schwarze 418: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 419: p = bsd_versions[0];
420: else {
421: s = n->string;
422: if (0 == strcmp(s, "3"))
423: p = bsd_versions[0];
424: else if (0 == strcmp(s, "4"))
425: p = bsd_versions[1];
426: else if (0 == strcmp(s, "5"))
427: p = bsd_versions[2];
428: else if (0 == strcmp(s, "6"))
429: p = bsd_versions[3];
430: else if (0 == strcmp(s, "7"))
431: p = bsd_versions[4];
432: else
433: p = bsd_versions[0];
434: }
435:
1.87 schwarze 436: free(man->meta.os);
437: man->meta.os = mandoc_strdup(p);
1.34 schwarze 438: }
439:
1.78 schwarze 440: static void
1.34 schwarze 441: post_AT(CHKARGS)
442: {
443: static const char * const unix_versions[] = {
444: "7th Edition",
445: "System III",
446: "System V",
447: "System V Release 2",
448: };
449:
1.86 schwarze 450: struct roff_node *nn;
1.34 schwarze 451: const char *p, *s;
452:
453: n = n->child;
454:
1.85 schwarze 455: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 456: p = unix_versions[0];
457: else {
458: s = n->string;
459: if (0 == strcmp(s, "3"))
460: p = unix_versions[0];
461: else if (0 == strcmp(s, "4"))
462: p = unix_versions[1];
463: else if (0 == strcmp(s, "5")) {
464: nn = n->next;
1.85 schwarze 465: if (nn != NULL &&
466: nn->type == ROFFT_TEXT &&
467: nn->string[0] != '\0')
1.34 schwarze 468: p = unix_versions[3];
469: else
470: p = unix_versions[2];
471: } else
472: p = unix_versions[0];
473: }
474:
1.87 schwarze 475: free(man->meta.os);
476: man->meta.os = mandoc_strdup(p);
1.46 schwarze 477: }
478:
1.78 schwarze 479: static void
1.46 schwarze 480: post_vs(CHKARGS)
481: {
482:
1.54 schwarze 483: if (NULL != n->prev)
1.78 schwarze 484: return;
1.54 schwarze 485:
486: switch (n->parent->tok) {
1.63 schwarze 487: case MAN_SH:
488: case MAN_SS:
1.67 schwarze 489: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.94 schwarze 490: "%s after %s", roff_name[n->tok],
491: roff_name[n->parent->tok]);
1.54 schwarze 492: /* FALLTHROUGH */
1.90 schwarze 493: case TOKEN_NONE:
1.63 schwarze 494: /*
1.54 schwarze 495: * Don't warn about this because it occurs in pod2man
496: * and would cause considerable (unfixable) warnage.
497: */
1.89 schwarze 498: roff_node_delete(man, n);
1.54 schwarze 499: break;
500: default:
501: break;
502: }
1.34 schwarze 503: }