Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.90
1.90 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.89 2015/04/19 13:50:10 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:
1.42 schwarze 118: /* FALLTHROUGH */
1.85 schwarze 119: case ROFFT_TBL:
1.78 schwarze 120: break;
1.1 kristaps 121: default:
1.76 schwarze 122: cp = man_valids + n->tok;
1.78 schwarze 123: if (*cp)
124: (*cp)(man, n);
125: break;
1.1 kristaps 126: }
127: }
128:
1.78 schwarze 129: static void
1.63 schwarze 130: check_root(CHKARGS)
1.4 schwarze 131: {
1.7 schwarze 132:
1.73 schwarze 133: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 134:
1.65 schwarze 135: if (NULL == man->first->child)
1.75 schwarze 136: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
137: n->line, n->pos, NULL);
1.65 schwarze 138: else
139: man->meta.hasbody = 1;
140:
141: if (NULL == man->meta.title) {
1.77 schwarze 142: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 143: n->line, n->pos, NULL);
1.35 schwarze 144:
1.18 schwarze 145: /*
146: * If a title hasn't been set, do so now (by
147: * implication, date and section also aren't set).
148: */
1.35 schwarze 149:
1.77 schwarze 150: man->meta.title = mandoc_strdup("");
151: man->meta.msec = mandoc_strdup("");
1.60 schwarze 152: man->meta.date = man->quick ? mandoc_strdup("") :
153: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 154: }
1.15 schwarze 155: }
156:
1.78 schwarze 157: static void
1.47 schwarze 158: check_text(CHKARGS)
1.4 schwarze 159: {
1.47 schwarze 160: char *cp, *p;
161:
1.57 schwarze 162: if (MAN_LITERAL & man->flags)
1.78 schwarze 163: return;
1.48 schwarze 164:
165: cp = n->string;
166: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 167: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
168: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 169: }
170:
1.84 schwarze 171: static void
172: post_OP(CHKARGS)
173: {
174:
175: if (n->nchild == 0)
176: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
177: n->line, n->pos, "OP");
178: else if (n->nchild > 2) {
179: n = n->child->next->next;
180: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
181: n->line, n->pos, "OP ... %s", n->string);
182: }
1.1 kristaps 183: }
184:
1.78 schwarze 185: static void
1.76 schwarze 186: post_UR(CHKARGS)
1.58 schwarze 187: {
188:
1.85 schwarze 189: if (n->type == ROFFT_HEAD && n->child == NULL)
1.84 schwarze 190: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
191: n->line, n->pos, "UR");
1.78 schwarze 192: check_part(man, n);
1.58 schwarze 193: }
1.32 schwarze 194:
1.78 schwarze 195: static void
1.47 schwarze 196: post_ft(CHKARGS)
1.32 schwarze 197: {
198: char *cp;
199: int ok;
200:
201: if (0 == n->nchild)
1.78 schwarze 202: return;
1.32 schwarze 203:
204: ok = 0;
205: cp = n->child->string;
206: switch (*cp) {
1.63 schwarze 207: case '1':
1.32 schwarze 208: /* FALLTHROUGH */
1.63 schwarze 209: case '2':
1.32 schwarze 210: /* FALLTHROUGH */
1.63 schwarze 211: case '3':
1.32 schwarze 212: /* FALLTHROUGH */
1.63 schwarze 213: case '4':
1.32 schwarze 214: /* FALLTHROUGH */
1.63 schwarze 215: case 'I':
1.32 schwarze 216: /* FALLTHROUGH */
1.63 schwarze 217: case 'P':
1.32 schwarze 218: /* FALLTHROUGH */
1.63 schwarze 219: case 'R':
1.32 schwarze 220: if ('\0' == cp[1])
221: ok = 1;
222: break;
1.63 schwarze 223: case 'B':
1.32 schwarze 224: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
225: ok = 1;
226: break;
1.63 schwarze 227: case 'C':
1.32 schwarze 228: if ('W' == cp[1] && '\0' == cp[2])
229: ok = 1;
230: break;
231: default:
232: break;
233: }
234:
235: if (0 == ok) {
1.70 schwarze 236: mandoc_vmsg(MANDOCERR_FT_BAD, man->parse,
237: n->line, n->pos, "ft %s", cp);
1.32 schwarze 238: *cp = '\0';
239: }
240: }
1.7 schwarze 241:
1.78 schwarze 242: static void
1.8 schwarze 243: check_part(CHKARGS)
244: {
245:
1.85 schwarze 246: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 247: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
248: n->line, n->pos, man_macronames[n->tok]);
1.8 schwarze 249: }
250:
1.78 schwarze 251: static void
1.36 schwarze 252: check_par(CHKARGS)
253: {
254:
1.39 schwarze 255: switch (n->type) {
1.85 schwarze 256: case ROFFT_BLOCK:
1.39 schwarze 257: if (0 == n->body->nchild)
1.89 schwarze 258: roff_node_delete(man, n);
1.39 schwarze 259: break;
1.85 schwarze 260: case ROFFT_BODY:
1.39 schwarze 261: if (0 == n->nchild)
1.67 schwarze 262: mandoc_vmsg(MANDOCERR_PAR_SKIP,
263: man->parse, n->line, n->pos,
264: "%s empty", man_macronames[n->tok]);
1.39 schwarze 265: break;
1.85 schwarze 266: case ROFFT_HEAD:
1.39 schwarze 267: if (n->nchild)
1.69 schwarze 268: mandoc_vmsg(MANDOCERR_ARG_SKIP,
269: man->parse, n->line, n->pos,
270: "%s %s%s", man_macronames[n->tok],
271: n->child->string,
272: n->nchild > 1 ? " ..." : "");
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:
1.55 schwarze 285: if (0 == n->head->nchild && 0 == n->body->nchild)
1.89 schwarze 286: roff_node_delete(man, n);
1.55 schwarze 287: break;
1.85 schwarze 288: case ROFFT_BODY:
1.55 schwarze 289: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.67 schwarze 290: mandoc_vmsg(MANDOCERR_PAR_SKIP,
291: man->parse, n->line, n->pos,
292: "%s empty", man_macronames[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_nf(CHKARGS)
395: {
396:
1.82 schwarze 397: if (man->flags & MAN_LITERAL)
1.74 schwarze 398: mandoc_msg(MANDOCERR_NF_SKIP, man->parse,
399: n->line, n->pos, "nf");
1.34 schwarze 400:
1.57 schwarze 401: man->flags |= MAN_LITERAL;
1.34 schwarze 402: }
403:
1.78 schwarze 404: static void
1.34 schwarze 405: post_fi(CHKARGS)
406: {
407:
1.57 schwarze 408: if ( ! (MAN_LITERAL & man->flags))
1.74 schwarze 409: mandoc_msg(MANDOCERR_FI_SKIP, man->parse,
410: n->line, n->pos, "fi");
1.34 schwarze 411:
1.57 schwarze 412: man->flags &= ~MAN_LITERAL;
1.34 schwarze 413: }
414:
1.78 schwarze 415: static void
1.34 schwarze 416: post_UC(CHKARGS)
417: {
418: static const char * const bsd_versions[] = {
419: "3rd Berkeley Distribution",
420: "4th Berkeley Distribution",
421: "4.2 Berkeley Distribution",
422: "4.3 Berkeley Distribution",
423: "4.4 Berkeley Distribution",
424: };
425:
426: const char *p, *s;
427:
428: n = n->child;
429:
1.85 schwarze 430: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 431: p = bsd_versions[0];
432: else {
433: s = n->string;
434: if (0 == strcmp(s, "3"))
435: p = bsd_versions[0];
436: else if (0 == strcmp(s, "4"))
437: p = bsd_versions[1];
438: else if (0 == strcmp(s, "5"))
439: p = bsd_versions[2];
440: else if (0 == strcmp(s, "6"))
441: p = bsd_versions[3];
442: else if (0 == strcmp(s, "7"))
443: p = bsd_versions[4];
444: else
445: p = bsd_versions[0];
446: }
447:
1.87 schwarze 448: free(man->meta.os);
449: man->meta.os = mandoc_strdup(p);
1.34 schwarze 450: }
451:
1.78 schwarze 452: static void
1.34 schwarze 453: post_AT(CHKARGS)
454: {
455: static const char * const unix_versions[] = {
456: "7th Edition",
457: "System III",
458: "System V",
459: "System V Release 2",
460: };
461:
1.86 schwarze 462: struct roff_node *nn;
1.34 schwarze 463: const char *p, *s;
464:
465: n = n->child;
466:
1.85 schwarze 467: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 468: p = unix_versions[0];
469: else {
470: s = n->string;
471: if (0 == strcmp(s, "3"))
472: p = unix_versions[0];
473: else if (0 == strcmp(s, "4"))
474: p = unix_versions[1];
475: else if (0 == strcmp(s, "5")) {
476: nn = n->next;
1.85 schwarze 477: if (nn != NULL &&
478: nn->type == ROFFT_TEXT &&
479: nn->string[0] != '\0')
1.34 schwarze 480: p = unix_versions[3];
481: else
482: p = unix_versions[2];
483: } else
484: p = unix_versions[0];
485: }
486:
1.87 schwarze 487: free(man->meta.os);
488: man->meta.os = mandoc_strdup(p);
1.46 schwarze 489: }
490:
1.78 schwarze 491: static void
1.46 schwarze 492: post_vs(CHKARGS)
493: {
494:
1.54 schwarze 495: if (NULL != n->prev)
1.78 schwarze 496: return;
1.54 schwarze 497:
498: switch (n->parent->tok) {
1.63 schwarze 499: case MAN_SH:
1.54 schwarze 500: /* FALLTHROUGH */
1.63 schwarze 501: case MAN_SS:
1.67 schwarze 502: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
503: "%s after %s", man_macronames[n->tok],
504: man_macronames[n->parent->tok]);
1.54 schwarze 505: /* FALLTHROUGH */
1.90 ! schwarze 506: case TOKEN_NONE:
1.63 schwarze 507: /*
1.54 schwarze 508: * Don't warn about this because it occurs in pod2man
509: * and would cause considerable (unfixable) warnage.
510: */
1.89 schwarze 511: roff_node_delete(man, n);
1.54 schwarze 512: break;
513: default:
514: break;
515: }
1.34 schwarze 516: }