Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.108
1.108 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.107 2018/08/16 23:40:19 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.105 schwarze 4: * Copyright (c) 2010, 2012-2018 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);
1.84 schwarze 48: static void post_OP(CHKARGS);
1.78 schwarze 49: static void post_TH(CHKARGS);
50: static void post_UC(CHKARGS);
51: static void post_UR(CHKARGS);
1.100 schwarze 52: static void post_in(CHKARGS);
53: static void post_vs(CHKARGS);
1.34 schwarze 54:
1.106 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 */
1.107 schwarze 60: NULL, /* TQ */
1.76 schwarze 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 */
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 */
1.100 schwarze 85: post_in, /* in */
1.108 ! schwarze 86: NULL, /* SY */
! 87: NULL, /* YS */
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 */
1.103 bentley 93: post_UR, /* MT */
94: NULL, /* ME */
1.1 kristaps 95: };
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;
1.105 schwarze 123: case ROFFT_COMMENT:
1.85 schwarze 124: case ROFFT_EQN:
125: case ROFFT_TBL:
1.78 schwarze 126: break;
1.1 kristaps 127: default:
1.95 schwarze 128: if (n->tok < ROFF_MAX) {
129: switch (n->tok) {
130: case ROFF_br:
1.98 schwarze 131: case ROFF_sp:
1.95 schwarze 132: post_vs(man, n);
133: break;
134: default:
1.96 schwarze 135: roff_validate(man);
136: break;
1.95 schwarze 137: }
138: break;
139: }
140: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.106 schwarze 141: cp = man_valids + (n->tok - MAN_TH);
1.78 schwarze 142: if (*cp)
143: (*cp)(man, n);
1.92 schwarze 144: if (man->last == n)
145: man_state(man, n);
1.78 schwarze 146: break;
1.1 kristaps 147: }
148: }
149:
1.78 schwarze 150: static void
1.63 schwarze 151: check_root(CHKARGS)
1.4 schwarze 152: {
1.73 schwarze 153: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 154:
1.105 schwarze 155: if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1.75 schwarze 156: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
157: n->line, n->pos, NULL);
1.65 schwarze 158: else
159: man->meta.hasbody = 1;
160:
161: if (NULL == man->meta.title) {
1.77 schwarze 162: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 163: n->line, n->pos, NULL);
1.35 schwarze 164:
1.18 schwarze 165: /*
166: * If a title hasn't been set, do so now (by
167: * implication, date and section also aren't set).
168: */
1.35 schwarze 169:
1.77 schwarze 170: man->meta.title = mandoc_strdup("");
171: man->meta.msec = mandoc_strdup("");
1.60 schwarze 172: man->meta.date = man->quick ? mandoc_strdup("") :
1.99 schwarze 173: mandoc_normdate(man, NULL, n->line, n->pos);
1.17 schwarze 174: }
1.101 schwarze 175:
176: if (man->meta.os_e &&
177: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.102 schwarze 178: mandoc_msg(MANDOCERR_RCS_MISSING, man->parse, 0, 0,
179: man->meta.os_e == MANDOC_OS_OPENBSD ?
180: "(OpenBSD)" : "(NetBSD)");
1.15 schwarze 181: }
182:
1.78 schwarze 183: static void
1.47 schwarze 184: check_text(CHKARGS)
1.4 schwarze 185: {
1.47 schwarze 186: char *cp, *p;
187:
1.57 schwarze 188: if (MAN_LITERAL & man->flags)
1.78 schwarze 189: return;
1.48 schwarze 190:
191: cp = n->string;
192: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 193: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
194: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 195: }
196:
1.84 schwarze 197: static void
198: post_OP(CHKARGS)
199: {
200:
1.93 schwarze 201: if (n->child == NULL)
1.84 schwarze 202: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
203: n->line, n->pos, "OP");
1.93 schwarze 204: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 205: n = n->child->next->next;
206: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
207: n->line, n->pos, "OP ... %s", n->string);
208: }
1.1 kristaps 209: }
210:
1.78 schwarze 211: static void
1.76 schwarze 212: post_UR(CHKARGS)
1.58 schwarze 213: {
1.85 schwarze 214: if (n->type == ROFFT_HEAD && n->child == NULL)
1.104 schwarze 215: mandoc_msg(MANDOCERR_UR_NOHEAD, man->parse,
1.103 bentley 216: n->line, n->pos, roff_name[n->tok]);
1.78 schwarze 217: check_part(man, n);
1.32 schwarze 218: }
1.7 schwarze 219:
1.78 schwarze 220: static void
1.8 schwarze 221: check_part(CHKARGS)
222: {
223:
1.85 schwarze 224: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 225: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.94 schwarze 226: n->line, n->pos, roff_name[n->tok]);
1.8 schwarze 227: }
228:
1.78 schwarze 229: static void
1.36 schwarze 230: check_par(CHKARGS)
231: {
232:
1.39 schwarze 233: switch (n->type) {
1.85 schwarze 234: case ROFFT_BLOCK:
1.93 schwarze 235: if (n->body->child == NULL)
1.89 schwarze 236: roff_node_delete(man, n);
1.39 schwarze 237: break;
1.85 schwarze 238: case ROFFT_BODY:
1.93 schwarze 239: if (n->child == NULL)
1.67 schwarze 240: mandoc_vmsg(MANDOCERR_PAR_SKIP,
241: man->parse, n->line, n->pos,
1.94 schwarze 242: "%s empty", roff_name[n->tok]);
1.39 schwarze 243: break;
1.85 schwarze 244: case ROFFT_HEAD:
1.93 schwarze 245: if (n->child != NULL)
1.69 schwarze 246: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.94 schwarze 247: man->parse, n->line, n->pos, "%s %s%s",
248: roff_name[n->tok], n->child->string,
1.93 schwarze 249: n->child->next != NULL ? " ..." : "");
1.39 schwarze 250: break;
251: default:
252: break;
253: }
1.36 schwarze 254: }
255:
1.78 schwarze 256: static void
1.55 schwarze 257: post_IP(CHKARGS)
258: {
259:
260: switch (n->type) {
1.85 schwarze 261: case ROFFT_BLOCK:
1.93 schwarze 262: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 263: roff_node_delete(man, n);
1.55 schwarze 264: break;
1.85 schwarze 265: case ROFFT_BODY:
1.93 schwarze 266: if (n->parent->head->child == NULL && n->child == NULL)
1.67 schwarze 267: mandoc_vmsg(MANDOCERR_PAR_SKIP,
268: man->parse, n->line, n->pos,
1.94 schwarze 269: "%s empty", roff_name[n->tok]);
1.55 schwarze 270: break;
271: default:
272: break;
273: }
274: }
1.36 schwarze 275:
1.78 schwarze 276: static void
1.34 schwarze 277: post_TH(CHKARGS)
278: {
1.86 schwarze 279: struct roff_node *nb;
1.40 schwarze 280: const char *p;
1.34 schwarze 281:
1.57 schwarze 282: free(man->meta.title);
283: free(man->meta.vol);
1.87 schwarze 284: free(man->meta.os);
1.57 schwarze 285: free(man->meta.msec);
286: free(man->meta.date);
1.34 schwarze 287:
1.57 schwarze 288: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 289: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 290:
1.64 schwarze 291: nb = n;
292:
1.87 schwarze 293: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 294:
295: n = n->child;
1.40 schwarze 296: if (n && n->string) {
297: for (p = n->string; '\0' != *p; p++) {
298: /* Only warn about this once... */
1.63 schwarze 299: if (isalpha((unsigned char)*p) &&
300: ! isupper((unsigned char)*p)) {
1.74 schwarze 301: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 302: man->parse, n->line,
303: n->pos + (p - n->string),
1.74 schwarze 304: "TH %s", n->string);
1.40 schwarze 305: break;
306: }
307: }
1.57 schwarze 308: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 309: } else {
1.57 schwarze 310: man->meta.title = mandoc_strdup("");
1.77 schwarze 311: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
312: nb->line, nb->pos, "TH");
313: }
1.34 schwarze 314:
1.87 schwarze 315: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 316:
1.40 schwarze 317: if (n)
318: n = n->next;
319: if (n && n->string)
1.57 schwarze 320: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 321: else {
1.57 schwarze 322: man->meta.msec = mandoc_strdup("");
1.77 schwarze 323: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
324: nb->line, nb->pos, "TH %s", man->meta.title);
325: }
1.34 schwarze 326:
1.87 schwarze 327: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 328:
1.40 schwarze 329: if (n)
330: n = n->next;
1.49 schwarze 331: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 332: man->meta.date = man->quick ?
333: mandoc_strdup(n->string) :
1.99 schwarze 334: mandoc_normdate(man, n->string, n->line, n->pos);
1.64 schwarze 335: } else {
1.57 schwarze 336: man->meta.date = mandoc_strdup("");
1.74 schwarze 337: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
338: n ? n->line : nb->line,
339: n ? n->pos : nb->pos, "TH");
1.64 schwarze 340: }
1.34 schwarze 341:
1.87 schwarze 342: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 343:
344: if (n && (n = n->next))
1.87 schwarze 345: man->meta.os = mandoc_strdup(n->string);
1.102 schwarze 346: else if (man->os_s != NULL)
347: man->meta.os = mandoc_strdup(man->os_s);
348: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
349: if (strstr(man->meta.os, "OpenBSD") != NULL)
350: man->meta.os_e = MANDOC_OS_OPENBSD;
351: else if (strstr(man->meta.os, "NetBSD") != NULL)
352: man->meta.os_e = MANDOC_OS_NETBSD;
353: }
1.34 schwarze 354:
1.87 schwarze 355: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 356: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 357:
358: if (n && (n = n->next))
1.57 schwarze 359: man->meta.vol = mandoc_strdup(n->string);
360: else if ('\0' != man->meta.msec[0] &&
361: (NULL != (p = mandoc_a2msec(man->meta.msec))))
362: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 363:
364: if (n != NULL && (n = n->next) != NULL)
365: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
366: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 367:
368: /*
369: * Remove the `TH' node after we've processed it for our
370: * meta-data.
371: */
1.89 schwarze 372: roff_node_delete(man, man->last);
1.34 schwarze 373: }
374:
1.78 schwarze 375: static void
1.34 schwarze 376: post_UC(CHKARGS)
377: {
378: static const char * const bsd_versions[] = {
379: "3rd Berkeley Distribution",
380: "4th Berkeley Distribution",
381: "4.2 Berkeley Distribution",
382: "4.3 Berkeley Distribution",
383: "4.4 Berkeley Distribution",
384: };
385:
386: const char *p, *s;
387:
388: n = n->child;
389:
1.85 schwarze 390: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 391: p = bsd_versions[0];
392: else {
393: s = n->string;
394: if (0 == strcmp(s, "3"))
395: p = bsd_versions[0];
396: else if (0 == strcmp(s, "4"))
397: p = bsd_versions[1];
398: else if (0 == strcmp(s, "5"))
399: p = bsd_versions[2];
400: else if (0 == strcmp(s, "6"))
401: p = bsd_versions[3];
402: else if (0 == strcmp(s, "7"))
403: p = bsd_versions[4];
404: else
405: p = bsd_versions[0];
406: }
407:
1.87 schwarze 408: free(man->meta.os);
409: man->meta.os = mandoc_strdup(p);
1.34 schwarze 410: }
411:
1.78 schwarze 412: static void
1.34 schwarze 413: post_AT(CHKARGS)
414: {
415: static const char * const unix_versions[] = {
416: "7th Edition",
417: "System III",
418: "System V",
419: "System V Release 2",
420: };
421:
1.86 schwarze 422: struct roff_node *nn;
1.34 schwarze 423: const char *p, *s;
424:
425: n = n->child;
426:
1.85 schwarze 427: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 428: p = unix_versions[0];
429: else {
430: s = n->string;
431: if (0 == strcmp(s, "3"))
432: p = unix_versions[0];
433: else if (0 == strcmp(s, "4"))
434: p = unix_versions[1];
435: else if (0 == strcmp(s, "5")) {
436: nn = n->next;
1.85 schwarze 437: if (nn != NULL &&
438: nn->type == ROFFT_TEXT &&
439: nn->string[0] != '\0')
1.34 schwarze 440: p = unix_versions[3];
441: else
442: p = unix_versions[2];
443: } else
444: p = unix_versions[0];
445: }
446:
1.87 schwarze 447: free(man->meta.os);
448: man->meta.os = mandoc_strdup(p);
1.100 schwarze 449: }
450:
451: static void
452: post_in(CHKARGS)
453: {
454: char *s;
455:
456: if (n->parent->tok != MAN_TP ||
457: n->parent->type != ROFFT_HEAD ||
458: n->child == NULL ||
459: *n->child->string == '+' ||
460: *n->child->string == '-')
461: return;
462: mandoc_asprintf(&s, "+%s", n->child->string);
463: free(n->child->string);
464: n->child->string = s;
1.46 schwarze 465: }
466:
1.78 schwarze 467: static void
1.46 schwarze 468: post_vs(CHKARGS)
469: {
470:
1.54 schwarze 471: if (NULL != n->prev)
1.78 schwarze 472: return;
1.54 schwarze 473:
474: switch (n->parent->tok) {
1.63 schwarze 475: case MAN_SH:
476: case MAN_SS:
1.98 schwarze 477: case MAN_PP:
478: case MAN_LP:
479: case MAN_P:
1.67 schwarze 480: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.94 schwarze 481: "%s after %s", roff_name[n->tok],
482: roff_name[n->parent->tok]);
1.54 schwarze 483: /* FALLTHROUGH */
1.90 schwarze 484: case TOKEN_NONE:
1.63 schwarze 485: /*
1.54 schwarze 486: * Don't warn about this because it occurs in pod2man
487: * and would cause considerable (unfixable) warnage.
488: */
1.89 schwarze 489: roff_node_delete(man, n);
1.54 schwarze 490: break;
491: default:
492: break;
493: }
1.34 schwarze 494: }