Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.80
1.80 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.79 2014/12/28 14:39:08 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.60 schwarze 4: * Copyright (c) 2010, 2012, 2013, 2014 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: *
1.2 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
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:
1.44 schwarze 29: #include "man.h"
1.25 schwarze 30: #include "mandoc.h"
1.61 schwarze 31: #include "mandoc_aux.h"
1.1 kristaps 32: #include "libman.h"
1.5 schwarze 33: #include "libmandoc.h"
1.1 kristaps 34:
1.57 schwarze 35: #define CHKARGS struct man *man, struct man_node *n
1.1 kristaps 36:
1.78 schwarze 37: typedef void (*v_check)(CHKARGS);
1.1 kristaps 38:
1.78 schwarze 39: static void check_eq0(CHKARGS);
40: static void check_eq2(CHKARGS);
41: static void check_le1(CHKARGS);
42: static void check_le5(CHKARGS);
43: static void check_par(CHKARGS);
44: static void check_part(CHKARGS);
45: static void check_root(CHKARGS);
46: static void check_text(CHKARGS);
47:
48: static void post_AT(CHKARGS);
49: static void post_IP(CHKARGS);
50: static void post_vs(CHKARGS);
51: static void post_fi(CHKARGS);
52: static void post_ft(CHKARGS);
53: static void post_nf(CHKARGS);
54: static void post_TH(CHKARGS);
55: static void post_UC(CHKARGS);
56: static void post_UR(CHKARGS);
1.34 schwarze 57:
1.76 schwarze 58: static v_check man_valids[MAN_MAX] = {
59: post_vs, /* br */
60: post_TH, /* TH */
61: NULL, /* SH */
62: NULL, /* SS */
63: NULL, /* TP */
64: check_par, /* LP */
65: check_par, /* PP */
66: check_par, /* P */
67: post_IP, /* IP */
68: NULL, /* HP */
69: NULL, /* SM */
70: NULL, /* SB */
71: NULL, /* BI */
72: NULL, /* IB */
73: NULL, /* BR */
74: NULL, /* RB */
75: NULL, /* R */
76: NULL, /* B */
77: NULL, /* I */
78: NULL, /* IR */
79: NULL, /* RI */
80: post_vs, /* sp */
81: post_nf, /* nf */
82: post_fi, /* fi */
83: NULL, /* RE */
84: check_part, /* RS */
85: NULL, /* DT */
86: post_UC, /* UC */
87: check_le1, /* PD */
88: post_AT, /* AT */
89: NULL, /* in */
90: post_ft, /* ft */
91: check_eq2, /* OP */
92: post_nf, /* EX */
93: post_fi, /* EE */
94: post_UR, /* UR */
95: NULL, /* UE */
96: NULL, /* ll */
1.1 kristaps 97: };
98:
99:
1.78 schwarze 100: void
1.57 schwarze 101: man_valid_post(struct man *man)
1.1 kristaps 102: {
1.76 schwarze 103: struct man_node *n;
1.7 schwarze 104: v_check *cp;
1.1 kristaps 105:
1.76 schwarze 106: n = man->last;
107: if (n->flags & MAN_VALID)
1.78 schwarze 108: return;
1.76 schwarze 109: n->flags |= MAN_VALID;
1.1 kristaps 110:
1.76 schwarze 111: switch (n->type) {
1.63 schwarze 112: case MAN_TEXT:
1.78 schwarze 113: check_text(man, n);
114: break;
1.63 schwarze 115: case MAN_ROOT:
1.78 schwarze 116: check_root(man, n);
117: break;
1.63 schwarze 118: case MAN_EQN:
1.42 schwarze 119: /* FALLTHROUGH */
1.63 schwarze 120: case MAN_TBL:
1.78 schwarze 121: break;
1.1 kristaps 122: default:
1.76 schwarze 123: cp = man_valids + n->tok;
1.78 schwarze 124: if (*cp)
125: (*cp)(man, n);
126: break;
1.1 kristaps 127: }
128: }
129:
1.78 schwarze 130: static void
1.63 schwarze 131: check_root(CHKARGS)
1.4 schwarze 132: {
1.7 schwarze 133:
1.73 schwarze 134: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 135:
1.65 schwarze 136: if (NULL == man->first->child)
1.75 schwarze 137: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
138: n->line, n->pos, NULL);
1.65 schwarze 139: else
140: man->meta.hasbody = 1;
141:
142: if (NULL == man->meta.title) {
1.77 schwarze 143: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 144: n->line, n->pos, NULL);
1.35 schwarze 145:
1.18 schwarze 146: /*
147: * If a title hasn't been set, do so now (by
148: * implication, date and section also aren't set).
149: */
1.35 schwarze 150:
1.77 schwarze 151: man->meta.title = mandoc_strdup("");
152: man->meta.msec = mandoc_strdup("");
1.60 schwarze 153: man->meta.date = man->quick ? mandoc_strdup("") :
154: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 155: }
1.15 schwarze 156: }
157:
1.78 schwarze 158: static void
1.47 schwarze 159: check_text(CHKARGS)
1.4 schwarze 160: {
1.47 schwarze 161: char *cp, *p;
162:
1.57 schwarze 163: if (MAN_LITERAL & man->flags)
1.78 schwarze 164: return;
1.48 schwarze 165:
166: cp = n->string;
167: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 168: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
169: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 170: }
171:
172: #define INEQ_DEFINE(x, ineq, name) \
1.78 schwarze 173: static void \
1.7 schwarze 174: check_##name(CHKARGS) \
1.1 kristaps 175: { \
1.4 schwarze 176: if (n->nchild ineq (x)) \
1.78 schwarze 177: return; \
1.57 schwarze 178: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line, n->pos, \
1.63 schwarze 179: "line arguments %s %d (have %d)", \
180: #ineq, (x), n->nchild); \
1.1 kristaps 181: }
182:
183: INEQ_DEFINE(0, ==, eq0)
1.52 schwarze 184: INEQ_DEFINE(2, ==, eq2)
1.10 schwarze 185: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 186: INEQ_DEFINE(5, <=, le5)
1.58 schwarze 187:
1.78 schwarze 188: static void
1.76 schwarze 189: post_UR(CHKARGS)
1.58 schwarze 190: {
191:
192: if (MAN_HEAD == n->type && 1 != n->nchild)
193: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
194: n->pos, "line arguments eq 1 (have %d)", n->nchild);
1.78 schwarze 195: check_part(man, n);
1.58 schwarze 196: }
1.32 schwarze 197:
1.78 schwarze 198: static void
1.47 schwarze 199: post_ft(CHKARGS)
1.32 schwarze 200: {
201: char *cp;
202: int ok;
203:
204: if (0 == n->nchild)
1.78 schwarze 205: return;
1.32 schwarze 206:
207: ok = 0;
208: cp = n->child->string;
209: switch (*cp) {
1.63 schwarze 210: case '1':
1.32 schwarze 211: /* FALLTHROUGH */
1.63 schwarze 212: case '2':
1.32 schwarze 213: /* FALLTHROUGH */
1.63 schwarze 214: case '3':
1.32 schwarze 215: /* FALLTHROUGH */
1.63 schwarze 216: case '4':
1.32 schwarze 217: /* FALLTHROUGH */
1.63 schwarze 218: case 'I':
1.32 schwarze 219: /* FALLTHROUGH */
1.63 schwarze 220: case 'P':
1.32 schwarze 221: /* FALLTHROUGH */
1.63 schwarze 222: case 'R':
1.32 schwarze 223: if ('\0' == cp[1])
224: ok = 1;
225: break;
1.63 schwarze 226: case 'B':
1.32 schwarze 227: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
228: ok = 1;
229: break;
1.63 schwarze 230: case 'C':
1.32 schwarze 231: if ('W' == cp[1] && '\0' == cp[2])
232: ok = 1;
233: break;
234: default:
235: break;
236: }
237:
238: if (0 == ok) {
1.70 schwarze 239: mandoc_vmsg(MANDOCERR_FT_BAD, man->parse,
240: n->line, n->pos, "ft %s", cp);
1.32 schwarze 241: *cp = '\0';
242: }
243:
244: if (1 < n->nchild)
1.63 schwarze 245: mandoc_vmsg(MANDOCERR_ARGCOUNT, man->parse, n->line,
246: n->pos, "want one child (have %d)", n->nchild);
1.32 schwarze 247: }
1.7 schwarze 248:
1.78 schwarze 249: static void
1.8 schwarze 250: check_part(CHKARGS)
251: {
252:
253: if (MAN_BODY == n->type && 0 == n->nchild)
1.63 schwarze 254: mandoc_msg(MANDOCERR_ARGCWARN, man->parse, n->line,
255: n->pos, "want children (have none)");
1.8 schwarze 256: }
257:
1.78 schwarze 258: static void
1.36 schwarze 259: check_par(CHKARGS)
260: {
261:
1.39 schwarze 262: switch (n->type) {
1.63 schwarze 263: case MAN_BLOCK:
1.39 schwarze 264: if (0 == n->body->nchild)
1.57 schwarze 265: man_node_delete(man, n);
1.39 schwarze 266: break;
1.63 schwarze 267: case MAN_BODY:
1.39 schwarze 268: if (0 == n->nchild)
1.67 schwarze 269: mandoc_vmsg(MANDOCERR_PAR_SKIP,
270: man->parse, n->line, n->pos,
271: "%s empty", man_macronames[n->tok]);
1.39 schwarze 272: break;
1.63 schwarze 273: case MAN_HEAD:
1.39 schwarze 274: if (n->nchild)
1.69 schwarze 275: mandoc_vmsg(MANDOCERR_ARG_SKIP,
276: man->parse, n->line, n->pos,
277: "%s %s%s", man_macronames[n->tok],
278: n->child->string,
279: n->nchild > 1 ? " ..." : "");
1.39 schwarze 280: break;
281: default:
282: break;
283: }
1.36 schwarze 284: }
285:
1.78 schwarze 286: static void
1.55 schwarze 287: post_IP(CHKARGS)
288: {
289:
290: switch (n->type) {
1.63 schwarze 291: case MAN_BLOCK:
1.55 schwarze 292: if (0 == n->head->nchild && 0 == n->body->nchild)
1.57 schwarze 293: man_node_delete(man, n);
1.55 schwarze 294: break;
1.63 schwarze 295: case MAN_BODY:
1.55 schwarze 296: if (0 == n->parent->head->nchild && 0 == n->nchild)
1.67 schwarze 297: mandoc_vmsg(MANDOCERR_PAR_SKIP,
298: man->parse, n->line, n->pos,
299: "%s empty", man_macronames[n->tok]);
1.55 schwarze 300: break;
301: default:
302: break;
303: }
304: }
1.36 schwarze 305:
1.78 schwarze 306: static void
1.34 schwarze 307: post_TH(CHKARGS)
308: {
1.64 schwarze 309: struct man_node *nb;
1.40 schwarze 310: const char *p;
1.34 schwarze 311:
1.76 schwarze 312: check_le5(man, n);
313:
1.57 schwarze 314: free(man->meta.title);
315: free(man->meta.vol);
316: free(man->meta.source);
317: free(man->meta.msec);
318: free(man->meta.date);
1.34 schwarze 319:
1.57 schwarze 320: man->meta.title = man->meta.vol = man->meta.date =
1.63 schwarze 321: man->meta.msec = man->meta.source = NULL;
1.34 schwarze 322:
1.64 schwarze 323: nb = n;
324:
1.34 schwarze 325: /* ->TITLE<- MSEC DATE SOURCE VOL */
326:
327: n = n->child;
1.40 schwarze 328: if (n && n->string) {
329: for (p = n->string; '\0' != *p; p++) {
330: /* Only warn about this once... */
1.63 schwarze 331: if (isalpha((unsigned char)*p) &&
332: ! isupper((unsigned char)*p)) {
1.74 schwarze 333: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 334: man->parse, n->line,
335: n->pos + (p - n->string),
1.74 schwarze 336: "TH %s", n->string);
1.40 schwarze 337: break;
338: }
339: }
1.57 schwarze 340: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 341: } else {
1.57 schwarze 342: man->meta.title = mandoc_strdup("");
1.77 schwarze 343: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
344: nb->line, nb->pos, "TH");
345: }
1.34 schwarze 346:
347: /* TITLE ->MSEC<- DATE SOURCE VOL */
348:
1.40 schwarze 349: if (n)
350: n = n->next;
351: if (n && n->string)
1.57 schwarze 352: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 353: else {
1.57 schwarze 354: man->meta.msec = mandoc_strdup("");
1.77 schwarze 355: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
356: nb->line, nb->pos, "TH %s", man->meta.title);
357: }
1.34 schwarze 358:
359: /* TITLE MSEC ->DATE<- SOURCE VOL */
360:
1.40 schwarze 361: if (n)
362: n = n->next;
1.49 schwarze 363: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 364: man->meta.date = man->quick ?
365: mandoc_strdup(n->string) :
366: mandoc_normdate(man->parse, n->string,
367: n->line, n->pos);
1.64 schwarze 368: } else {
1.57 schwarze 369: man->meta.date = mandoc_strdup("");
1.74 schwarze 370: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
371: n ? n->line : nb->line,
372: n ? n->pos : nb->pos, "TH");
1.64 schwarze 373: }
1.34 schwarze 374:
375: /* TITLE MSEC DATE ->SOURCE<- VOL */
376:
377: if (n && (n = n->next))
1.57 schwarze 378: man->meta.source = mandoc_strdup(n->string);
1.79 schwarze 379: else if (man->defos != NULL)
380: man->meta.source = mandoc_strdup(man->defos);
1.34 schwarze 381:
382: /* TITLE MSEC DATE SOURCE ->VOL<- */
1.51 schwarze 383: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 384:
385: if (n && (n = n->next))
1.57 schwarze 386: man->meta.vol = mandoc_strdup(n->string);
387: else if ('\0' != man->meta.msec[0] &&
388: (NULL != (p = mandoc_a2msec(man->meta.msec))))
389: man->meta.vol = mandoc_strdup(p);
1.34 schwarze 390:
391: /*
392: * Remove the `TH' node after we've processed it for our
393: * meta-data.
394: */
1.57 schwarze 395: man_node_delete(man, man->last);
1.34 schwarze 396: }
397:
1.78 schwarze 398: static void
1.34 schwarze 399: post_nf(CHKARGS)
400: {
401:
1.76 schwarze 402: check_eq0(man, n);
403:
1.57 schwarze 404: if (MAN_LITERAL & man->flags)
1.74 schwarze 405: mandoc_msg(MANDOCERR_NF_SKIP, man->parse,
406: n->line, n->pos, "nf");
1.34 schwarze 407:
1.57 schwarze 408: man->flags |= MAN_LITERAL;
1.34 schwarze 409: }
410:
1.78 schwarze 411: static void
1.34 schwarze 412: post_fi(CHKARGS)
413: {
414:
1.76 schwarze 415: check_eq0(man, n);
416:
1.57 schwarze 417: if ( ! (MAN_LITERAL & man->flags))
1.74 schwarze 418: mandoc_msg(MANDOCERR_FI_SKIP, man->parse,
419: n->line, n->pos, "fi");
1.34 schwarze 420:
1.57 schwarze 421: man->flags &= ~MAN_LITERAL;
1.34 schwarze 422: }
423:
1.78 schwarze 424: static void
1.34 schwarze 425: post_UC(CHKARGS)
426: {
427: static const char * const bsd_versions[] = {
428: "3rd Berkeley Distribution",
429: "4th Berkeley Distribution",
430: "4.2 Berkeley Distribution",
431: "4.3 Berkeley Distribution",
432: "4.4 Berkeley Distribution",
433: };
434:
435: const char *p, *s;
436:
437: n = n->child;
438:
439: if (NULL == n || MAN_TEXT != n->type)
440: p = bsd_versions[0];
441: else {
442: s = n->string;
443: if (0 == strcmp(s, "3"))
444: p = bsd_versions[0];
445: else if (0 == strcmp(s, "4"))
446: p = bsd_versions[1];
447: else if (0 == strcmp(s, "5"))
448: p = bsd_versions[2];
449: else if (0 == strcmp(s, "6"))
450: p = bsd_versions[3];
451: else if (0 == strcmp(s, "7"))
452: p = bsd_versions[4];
453: else
454: p = bsd_versions[0];
455: }
456:
1.57 schwarze 457: free(man->meta.source);
458: man->meta.source = mandoc_strdup(p);
1.34 schwarze 459: }
460:
1.78 schwarze 461: static void
1.34 schwarze 462: post_AT(CHKARGS)
463: {
464: static const char * const unix_versions[] = {
465: "7th Edition",
466: "System III",
467: "System V",
468: "System V Release 2",
469: };
470:
471: const char *p, *s;
472: struct man_node *nn;
473:
474: n = n->child;
475:
476: if (NULL == n || MAN_TEXT != n->type)
477: p = unix_versions[0];
478: else {
479: s = n->string;
480: if (0 == strcmp(s, "3"))
481: p = unix_versions[0];
482: else if (0 == strcmp(s, "4"))
483: p = unix_versions[1];
484: else if (0 == strcmp(s, "5")) {
485: nn = n->next;
486: if (nn && MAN_TEXT == nn->type && nn->string[0])
487: p = unix_versions[3];
488: else
489: p = unix_versions[2];
490: } else
491: p = unix_versions[0];
492: }
493:
1.57 schwarze 494: free(man->meta.source);
495: man->meta.source = mandoc_strdup(p);
1.46 schwarze 496: }
497:
1.78 schwarze 498: static void
1.46 schwarze 499: post_vs(CHKARGS)
500: {
1.76 schwarze 501:
502: if (n->tok == MAN_br)
503: check_eq0(man, n);
504: else
505: check_le1(man, n);
1.46 schwarze 506:
1.54 schwarze 507: if (NULL != n->prev)
1.78 schwarze 508: return;
1.54 schwarze 509:
510: switch (n->parent->tok) {
1.63 schwarze 511: case MAN_SH:
1.54 schwarze 512: /* FALLTHROUGH */
1.63 schwarze 513: case MAN_SS:
1.67 schwarze 514: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
515: "%s after %s", man_macronames[n->tok],
516: man_macronames[n->parent->tok]);
1.54 schwarze 517: /* FALLTHROUGH */
1.63 schwarze 518: case MAN_MAX:
519: /*
1.54 schwarze 520: * Don't warn about this because it occurs in pod2man
521: * and would cause considerable (unfixable) warnage.
522: */
1.57 schwarze 523: man_node_delete(man, n);
1.54 schwarze 524: break;
525: default:
526: break;
527: }
1.34 schwarze 528: }