Annotation of mandoc/man_validate.c, Revision 1.140
1.107 schwarze 1: /* $OpenBSD$ */
1.1 kristaps 2: /*
1.63 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.134 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
1.8 kristaps 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: *
1.114 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.8 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.114 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.8 kristaps 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: */
1.28 kristaps 18: #include "config.h"
19:
1.1 kristaps 20: #include <sys/types.h>
21:
22: #include <assert.h>
23: #include <ctype.h>
1.16 kristaps 24: #include <errno.h>
25: #include <limits.h>
1.1 kristaps 26: #include <stdarg.h>
1.140 ! schwarze 27: #include <stdio.h>
1.1 kristaps 28: #include <stdlib.h>
1.46 kristaps 29: #include <string.h>
1.50 kristaps 30: #include <time.h>
1.1 kristaps 31:
1.114 schwarze 32: #include "mandoc_aux.h"
33: #include "mandoc.h"
34: #include "roff.h"
1.66 kristaps 35: #include "man.h"
1.114 schwarze 36: #include "libmandoc.h"
1.118 schwarze 37: #include "roff_int.h"
1.1 kristaps 38: #include "libman.h"
39:
1.117 schwarze 40: #define CHKARGS struct roff_man *man, struct roff_node *n
1.1 kristaps 41:
1.107 schwarze 42: typedef void (*v_check)(CHKARGS);
1.1 kristaps 43:
1.138 schwarze 44: static void check_abort(CHKARGS);
1.107 schwarze 45: static void check_par(CHKARGS);
46: static void check_part(CHKARGS);
47: static void check_root(CHKARGS);
48: static void check_text(CHKARGS);
49:
50: static void post_AT(CHKARGS);
51: static void post_IP(CHKARGS);
1.113 schwarze 52: static void post_OP(CHKARGS);
1.139 schwarze 53: static void post_SH(CHKARGS);
1.107 schwarze 54: static void post_TH(CHKARGS);
55: static void post_UC(CHKARGS);
56: static void post_UR(CHKARGS);
1.129 schwarze 57: static void post_in(CHKARGS);
1.51 kristaps 58:
1.135 schwarze 59: static const v_check man_valids[MAN_MAX - MAN_TH] = {
1.104 schwarze 60: post_TH, /* TH */
1.139 schwarze 61: post_SH, /* SH */
62: post_SH, /* SS */
1.104 schwarze 63: NULL, /* TP */
1.136 schwarze 64: NULL, /* TQ */
1.138 schwarze 65: check_abort,/* LP */
1.104 schwarze 66: check_par, /* PP */
1.138 schwarze 67: check_abort,/* P */
1.104 schwarze 68: post_IP, /* IP */
69: NULL, /* HP */
70: NULL, /* SM */
71: NULL, /* SB */
72: NULL, /* BI */
73: NULL, /* IB */
74: NULL, /* BR */
75: NULL, /* RB */
76: NULL, /* R */
77: NULL, /* B */
78: NULL, /* I */
79: NULL, /* IR */
80: NULL, /* RI */
1.121 schwarze 81: NULL, /* nf */
82: NULL, /* fi */
1.104 schwarze 83: NULL, /* RE */
84: check_part, /* RS */
85: NULL, /* DT */
86: post_UC, /* UC */
1.112 schwarze 87: NULL, /* PD */
1.104 schwarze 88: post_AT, /* AT */
1.129 schwarze 89: post_in, /* in */
1.137 schwarze 90: NULL, /* SY */
91: NULL, /* YS */
1.113 schwarze 92: post_OP, /* OP */
1.121 schwarze 93: NULL, /* EX */
94: NULL, /* EE */
1.104 schwarze 95: post_UR, /* UR */
96: NULL, /* UE */
1.132 schwarze 97: post_UR, /* MT */
98: NULL, /* ME */
1.1 kristaps 99: };
100:
101:
1.138 schwarze 102: /* Validate the subtree rooted at man->last. */
1.107 schwarze 103: void
1.121 schwarze 104: man_node_validate(struct roff_man *man)
1.1 kristaps 105: {
1.115 schwarze 106: struct roff_node *n;
1.123 schwarze 107: const v_check *cp;
1.1 kristaps 108:
1.138 schwarze 109: /*
110: * Translate obsolete macros such that later code
111: * does not need to look for them.
112: */
113:
1.104 schwarze 114: n = man->last;
1.138 schwarze 115: switch (n->tok) {
116: case MAN_LP:
117: case MAN_P:
118: n->tok = MAN_PP;
119: break;
120: default:
121: break;
122: }
123:
124: /*
125: * Iterate over all children, recursing into each one
126: * in turn, depth-first.
127: */
128:
1.121 schwarze 129: man->last = man->last->child;
130: while (man->last != NULL) {
131: man_node_validate(man);
132: if (man->last == n)
133: man->last = man->last->child;
134: else
135: man->last = man->last->next;
136: }
1.1 kristaps 137:
1.138 schwarze 138: /* Finally validate the macro itself. */
139:
1.121 schwarze 140: man->last = n;
141: man->next = ROFF_NEXT_SIBLING;
1.104 schwarze 142: switch (n->type) {
1.114 schwarze 143: case ROFFT_TEXT:
1.107 schwarze 144: check_text(man, n);
145: break;
1.114 schwarze 146: case ROFFT_ROOT:
1.107 schwarze 147: check_root(man, n);
148: break;
1.134 schwarze 149: case ROFFT_COMMENT:
1.114 schwarze 150: case ROFFT_EQN:
151: case ROFFT_TBL:
1.107 schwarze 152: break;
1.1 kristaps 153: default:
1.124 schwarze 154: if (n->tok < ROFF_MAX) {
1.139 schwarze 155: roff_validate(man);
1.124 schwarze 156: break;
157: }
158: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.135 schwarze 159: cp = man_valids + (n->tok - MAN_TH);
1.107 schwarze 160: if (*cp)
161: (*cp)(man, n);
1.121 schwarze 162: if (man->last == n)
163: man_state(man, n);
1.107 schwarze 164: break;
1.1 kristaps 165: }
166: }
167:
1.107 schwarze 168: static void
1.91 schwarze 169: check_root(CHKARGS)
1.14 kristaps 170: {
1.101 schwarze 171: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.17 kristaps 172:
1.134 schwarze 173: if (n->last == NULL || n->last->type == ROFFT_COMMENT)
1.103 schwarze 174: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
175: n->line, n->pos, NULL);
1.93 schwarze 176: else
177: man->meta.hasbody = 1;
178:
179: if (NULL == man->meta.title) {
1.105 schwarze 180: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.103 schwarze 181: n->line, n->pos, NULL);
1.54 kristaps 182:
1.34 kristaps 183: /*
184: * If a title hasn't been set, do so now (by
185: * implication, date and section also aren't set).
186: */
1.54 kristaps 187:
1.105 schwarze 188: man->meta.title = mandoc_strdup("");
189: man->meta.msec = mandoc_strdup("");
1.88 schwarze 190: man->meta.date = man->quick ? mandoc_strdup("") :
1.128 schwarze 191: mandoc_normdate(man, NULL, n->line, n->pos);
1.34 kristaps 192: }
1.130 schwarze 193:
194: if (man->meta.os_e &&
195: (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
1.131 schwarze 196: mandoc_msg(MANDOCERR_RCS_MISSING, man->parse, 0, 0,
197: man->meta.os_e == MANDOC_OS_OPENBSD ?
198: "(OpenBSD)" : "(NetBSD)");
1.75 kristaps 199: }
200:
1.107 schwarze 201: static void
1.138 schwarze 202: check_abort(CHKARGS)
203: {
204: abort();
205: }
206:
207: static void
1.75 kristaps 208: check_text(CHKARGS)
209: {
210: char *cp, *p;
211:
1.85 schwarze 212: if (MAN_LITERAL & man->flags)
1.107 schwarze 213: return;
1.76 schwarze 214:
215: cp = n->string;
216: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.99 schwarze 217: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
218: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 219: }
220:
1.113 schwarze 221: static void
222: post_OP(CHKARGS)
223: {
224:
1.122 schwarze 225: if (n->child == NULL)
1.113 schwarze 226: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
227: n->line, n->pos, "OP");
1.122 schwarze 228: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.113 schwarze 229: n = n->child->next->next;
230: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
231: n->line, n->pos, "OP ... %s", n->string);
232: }
1.1 kristaps 233: }
234:
1.107 schwarze 235: static void
1.139 schwarze 236: post_SH(CHKARGS)
237: {
238: struct roff_node *nc;
239:
240: if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
241: return;
242:
243: if (nc->tok == MAN_PP && nc->body->child != NULL) {
244: while (nc->body->last != NULL) {
245: man->next = ROFF_NEXT_CHILD;
246: roff_node_relink(man, nc->body->last);
247: man->last = n;
248: }
249: }
250:
251: if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
252: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse,
253: nc->line, nc->pos, "%s after %s",
254: roff_name[nc->tok], roff_name[n->tok]);
255: roff_node_delete(man, nc);
256: }
257:
258: /*
259: * Trailing PP is empty, so it is deleted by check_par().
260: * Trailing sp is significant.
261: */
262:
263: if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
264: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse,
265: nc->line, nc->pos, "%s at the end of %s",
266: roff_name[nc->tok], roff_name[n->tok]);
267: roff_node_delete(man, nc);
268: }
269: }
270:
271: static void
1.104 schwarze 272: post_UR(CHKARGS)
1.86 schwarze 273: {
1.114 schwarze 274: if (n->type == ROFFT_HEAD && n->child == NULL)
1.133 schwarze 275: mandoc_msg(MANDOCERR_UR_NOHEAD, man->parse,
1.132 schwarze 276: n->line, n->pos, roff_name[n->tok]);
1.107 schwarze 277: check_part(man, n);
1.55 kristaps 278: }
1.16 kristaps 279:
1.107 schwarze 280: static void
1.23 kristaps 281: check_part(CHKARGS)
282: {
283:
1.114 schwarze 284: if (n->type == ROFFT_BODY && n->child == NULL)
1.110 schwarze 285: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.123 schwarze 286: n->line, n->pos, roff_name[n->tok]);
1.23 kristaps 287: }
288:
1.107 schwarze 289: static void
1.17 kristaps 290: check_par(CHKARGS)
291: {
292:
1.59 kristaps 293: switch (n->type) {
1.114 schwarze 294: case ROFFT_BLOCK:
1.122 schwarze 295: if (n->body->child == NULL)
1.118 schwarze 296: roff_node_delete(man, n);
1.59 kristaps 297: break;
1.114 schwarze 298: case ROFFT_BODY:
1.139 schwarze 299: if (n->child != NULL &&
300: (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
301: mandoc_vmsg(MANDOCERR_PAR_SKIP,
302: man->parse, n->child->line, n->child->pos,
303: "%s after %s", roff_name[n->child->tok],
304: roff_name[n->tok]);
305: roff_node_delete(man, n->child);
306: }
1.122 schwarze 307: if (n->child == NULL)
1.95 schwarze 308: mandoc_vmsg(MANDOCERR_PAR_SKIP,
309: man->parse, n->line, n->pos,
1.123 schwarze 310: "%s empty", roff_name[n->tok]);
1.59 kristaps 311: break;
1.114 schwarze 312: case ROFFT_HEAD:
1.122 schwarze 313: if (n->child != NULL)
1.97 schwarze 314: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.123 schwarze 315: man->parse, n->line, n->pos, "%s %s%s",
316: roff_name[n->tok], n->child->string,
1.122 schwarze 317: n->child->next != NULL ? " ..." : "");
1.59 kristaps 318: break;
319: default:
320: break;
321: }
1.17 kristaps 322: }
323:
1.107 schwarze 324: static void
1.83 schwarze 325: post_IP(CHKARGS)
326: {
327:
328: switch (n->type) {
1.114 schwarze 329: case ROFFT_BLOCK:
1.122 schwarze 330: if (n->head->child == NULL && n->body->child == NULL)
1.118 schwarze 331: roff_node_delete(man, n);
1.83 schwarze 332: break;
1.114 schwarze 333: case ROFFT_BODY:
1.122 schwarze 334: if (n->parent->head->child == NULL && n->child == NULL)
1.95 schwarze 335: mandoc_vmsg(MANDOCERR_PAR_SKIP,
336: man->parse, n->line, n->pos,
1.123 schwarze 337: "%s empty", roff_name[n->tok]);
1.83 schwarze 338: break;
339: default:
340: break;
341: }
342: }
1.17 kristaps 343:
1.107 schwarze 344: static void
1.51 kristaps 345: post_TH(CHKARGS)
346: {
1.115 schwarze 347: struct roff_node *nb;
1.60 schwarze 348: const char *p;
1.51 kristaps 349:
1.85 schwarze 350: free(man->meta.title);
351: free(man->meta.vol);
1.116 schwarze 352: free(man->meta.os);
1.85 schwarze 353: free(man->meta.msec);
354: free(man->meta.date);
1.51 kristaps 355:
1.85 schwarze 356: man->meta.title = man->meta.vol = man->meta.date =
1.116 schwarze 357: man->meta.msec = man->meta.os = NULL;
1.51 kristaps 358:
1.92 schwarze 359: nb = n;
360:
1.116 schwarze 361: /* ->TITLE<- MSEC DATE OS VOL */
1.51 kristaps 362:
363: n = n->child;
1.60 schwarze 364: if (n && n->string) {
365: for (p = n->string; '\0' != *p; p++) {
366: /* Only warn about this once... */
1.91 schwarze 367: if (isalpha((unsigned char)*p) &&
368: ! isupper((unsigned char)*p)) {
1.102 schwarze 369: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.94 schwarze 370: man->parse, n->line,
371: n->pos + (p - n->string),
1.102 schwarze 372: "TH %s", n->string);
1.60 schwarze 373: break;
374: }
375: }
1.85 schwarze 376: man->meta.title = mandoc_strdup(n->string);
1.105 schwarze 377: } else {
1.85 schwarze 378: man->meta.title = mandoc_strdup("");
1.105 schwarze 379: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
380: nb->line, nb->pos, "TH");
381: }
1.51 kristaps 382:
1.116 schwarze 383: /* TITLE ->MSEC<- DATE OS VOL */
1.51 kristaps 384:
1.60 schwarze 385: if (n)
386: n = n->next;
387: if (n && n->string)
1.85 schwarze 388: man->meta.msec = mandoc_strdup(n->string);
1.105 schwarze 389: else {
1.85 schwarze 390: man->meta.msec = mandoc_strdup("");
1.105 schwarze 391: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
392: nb->line, nb->pos, "TH %s", man->meta.title);
393: }
1.51 kristaps 394:
1.116 schwarze 395: /* TITLE MSEC ->DATE<- OS VOL */
1.51 kristaps 396:
1.60 schwarze 397: if (n)
398: n = n->next;
1.77 schwarze 399: if (n && n->string && '\0' != n->string[0]) {
1.88 schwarze 400: man->meta.date = man->quick ?
401: mandoc_strdup(n->string) :
1.128 schwarze 402: mandoc_normdate(man, n->string, n->line, n->pos);
1.92 schwarze 403: } else {
1.85 schwarze 404: man->meta.date = mandoc_strdup("");
1.102 schwarze 405: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
406: n ? n->line : nb->line,
407: n ? n->pos : nb->pos, "TH");
1.92 schwarze 408: }
1.51 kristaps 409:
1.116 schwarze 410: /* TITLE MSEC DATE ->OS<- VOL */
1.51 kristaps 411:
412: if (n && (n = n->next))
1.116 schwarze 413: man->meta.os = mandoc_strdup(n->string);
1.131 schwarze 414: else if (man->os_s != NULL)
415: man->meta.os = mandoc_strdup(man->os_s);
416: if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
417: if (strstr(man->meta.os, "OpenBSD") != NULL)
418: man->meta.os_e = MANDOC_OS_OPENBSD;
419: else if (strstr(man->meta.os, "NetBSD") != NULL)
420: man->meta.os_e = MANDOC_OS_NETBSD;
421: }
1.51 kristaps 422:
1.116 schwarze 423: /* TITLE MSEC DATE OS ->VOL<- */
1.79 schwarze 424: /* If missing, use the default VOL name for MSEC. */
1.51 kristaps 425:
426: if (n && (n = n->next))
1.85 schwarze 427: man->meta.vol = mandoc_strdup(n->string);
428: else if ('\0' != man->meta.msec[0] &&
429: (NULL != (p = mandoc_a2msec(man->meta.msec))))
430: man->meta.vol = mandoc_strdup(p);
1.113 schwarze 431:
432: if (n != NULL && (n = n->next) != NULL)
433: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
434: n->line, n->pos, "TH ... %s", n->string);
1.51 kristaps 435:
436: /*
437: * Remove the `TH' node after we've processed it for our
438: * meta-data.
439: */
1.118 schwarze 440: roff_node_delete(man, man->last);
1.51 kristaps 441: }
442:
1.107 schwarze 443: static void
1.51 kristaps 444: post_UC(CHKARGS)
445: {
446: static const char * const bsd_versions[] = {
447: "3rd Berkeley Distribution",
448: "4th Berkeley Distribution",
449: "4.2 Berkeley Distribution",
450: "4.3 Berkeley Distribution",
451: "4.4 Berkeley Distribution",
452: };
453:
454: const char *p, *s;
455:
456: n = n->child;
457:
1.114 schwarze 458: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 459: p = bsd_versions[0];
460: else {
461: s = n->string;
462: if (0 == strcmp(s, "3"))
463: p = bsd_versions[0];
464: else if (0 == strcmp(s, "4"))
465: p = bsd_versions[1];
466: else if (0 == strcmp(s, "5"))
467: p = bsd_versions[2];
468: else if (0 == strcmp(s, "6"))
469: p = bsd_versions[3];
470: else if (0 == strcmp(s, "7"))
471: p = bsd_versions[4];
472: else
473: p = bsd_versions[0];
474: }
475:
1.116 schwarze 476: free(man->meta.os);
477: man->meta.os = mandoc_strdup(p);
1.51 kristaps 478: }
479:
1.107 schwarze 480: static void
1.51 kristaps 481: post_AT(CHKARGS)
482: {
483: static const char * const unix_versions[] = {
484: "7th Edition",
485: "System III",
486: "System V",
487: "System V Release 2",
488: };
489:
1.115 schwarze 490: struct roff_node *nn;
1.51 kristaps 491: const char *p, *s;
492:
493: n = n->child;
494:
1.114 schwarze 495: if (n == NULL || n->type != ROFFT_TEXT)
1.51 kristaps 496: p = unix_versions[0];
497: else {
498: s = n->string;
499: if (0 == strcmp(s, "3"))
500: p = unix_versions[0];
501: else if (0 == strcmp(s, "4"))
502: p = unix_versions[1];
503: else if (0 == strcmp(s, "5")) {
504: nn = n->next;
1.114 schwarze 505: if (nn != NULL &&
506: nn->type == ROFFT_TEXT &&
507: nn->string[0] != '\0')
1.51 kristaps 508: p = unix_versions[3];
509: else
510: p = unix_versions[2];
511: } else
512: p = unix_versions[0];
513: }
514:
1.116 schwarze 515: free(man->meta.os);
516: man->meta.os = mandoc_strdup(p);
1.129 schwarze 517: }
518:
519: static void
520: post_in(CHKARGS)
521: {
522: char *s;
523:
524: if (n->parent->tok != MAN_TP ||
525: n->parent->type != ROFFT_HEAD ||
526: n->child == NULL ||
527: *n->child->string == '+' ||
528: *n->child->string == '-')
529: return;
530: mandoc_asprintf(&s, "+%s", n->child->string);
531: free(n->child->string);
532: n->child->string = s;
1.51 kristaps 533: }
CVSweb