Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.40
1.40 ! schwarze 1: /* $Id: man_validate.c,v 1.39 2011/01/16 19:27:25 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.32 schwarze 4: * Copyright (c) 2010 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:
1.25 schwarze 29: #include "mandoc.h"
1.1 kristaps 30: #include "libman.h"
1.5 schwarze 31: #include "libmandoc.h"
1.1 kristaps 32:
1.27 schwarze 33: #define CHKARGS struct man *m, struct man_node *n
1.1 kristaps 34:
1.7 schwarze 35: typedef int (*v_check)(CHKARGS);
1.1 kristaps 36:
37: struct man_valid {
1.7 schwarze 38: v_check *pres;
39: v_check *posts;
1.1 kristaps 40: };
41:
1.7 schwarze 42: static int check_bline(CHKARGS);
43: static int check_eq0(CHKARGS);
1.36 schwarze 44: static int check_ft(CHKARGS);
1.10 schwarze 45: static int check_le1(CHKARGS);
1.7 schwarze 46: static int check_ge2(CHKARGS);
47: static int check_le5(CHKARGS);
48: static int check_par(CHKARGS);
1.8 schwarze 49: static int check_part(CHKARGS);
1.7 schwarze 50: static int check_root(CHKARGS);
51: static int check_sec(CHKARGS);
52: static int check_text(CHKARGS);
53:
1.34 schwarze 54: static int post_AT(CHKARGS);
55: static int post_fi(CHKARGS);
56: static int post_nf(CHKARGS);
57: static int post_TH(CHKARGS);
58: static int post_UC(CHKARGS);
59:
60: static v_check posts_at[] = { post_AT, NULL };
1.7 schwarze 61: static v_check posts_eq0[] = { check_eq0, NULL };
1.34 schwarze 62: static v_check posts_fi[] = { check_eq0, post_fi, NULL };
63: static v_check posts_le1[] = { check_le1, NULL };
1.36 schwarze 64: static v_check posts_ft[] = { check_ft, NULL };
1.34 schwarze 65: static v_check posts_nf[] = { check_eq0, post_nf, NULL };
1.7 schwarze 66: static v_check posts_par[] = { check_par, NULL };
1.8 schwarze 67: static v_check posts_part[] = { check_part, NULL };
1.7 schwarze 68: static v_check posts_sec[] = { check_sec, NULL };
1.40 ! schwarze 69: static v_check posts_th[] = { check_ge2, check_le5, post_TH, NULL };
1.34 schwarze 70: static v_check posts_uc[] = { post_UC, NULL };
1.7 schwarze 71: static v_check pres_bline[] = { check_bline, NULL };
1.1 kristaps 72:
1.34 schwarze 73:
1.1 kristaps 74: static const struct man_valid man_valids[MAN_MAX] = {
1.12 schwarze 75: { NULL, posts_eq0 }, /* br */
1.15 schwarze 76: { pres_bline, posts_th }, /* TH */
1.7 schwarze 77: { pres_bline, posts_sec }, /* SH */
78: { pres_bline, posts_sec }, /* SS */
1.39 schwarze 79: { pres_bline, NULL }, /* TP */
1.7 schwarze 80: { pres_bline, posts_par }, /* LP */
81: { pres_bline, posts_par }, /* PP */
82: { pres_bline, posts_par }, /* P */
1.39 schwarze 83: { pres_bline, NULL }, /* IP */
84: { pres_bline, NULL }, /* HP */
1.8 schwarze 85: { NULL, NULL }, /* SM */
86: { NULL, NULL }, /* SB */
1.7 schwarze 87: { NULL, NULL }, /* BI */
88: { NULL, NULL }, /* IB */
89: { NULL, NULL }, /* BR */
90: { NULL, NULL }, /* RB */
1.8 schwarze 91: { NULL, NULL }, /* R */
92: { NULL, NULL }, /* B */
93: { NULL, NULL }, /* I */
1.7 schwarze 94: { NULL, NULL }, /* IR */
95: { NULL, NULL }, /* RI */
1.28 schwarze 96: { NULL, posts_eq0 }, /* na */ /* FIXME: should warn only. */
97: { NULL, posts_le1 }, /* sp */ /* FIXME: should warn only. */
1.34 schwarze 98: { pres_bline, posts_nf }, /* nf */
99: { pres_bline, posts_fi }, /* fi */
1.8 schwarze 100: { NULL, NULL }, /* RE */
101: { NULL, posts_part }, /* RS */
102: { NULL, NULL }, /* DT */
1.34 schwarze 103: { NULL, posts_uc }, /* UC */
1.11 schwarze 104: { NULL, NULL }, /* PD */
1.34 schwarze 105: { NULL, posts_at }, /* AT */
1.28 schwarze 106: { NULL, NULL }, /* in */
1.32 schwarze 107: { NULL, posts_ft }, /* ft */
1.1 kristaps 108: };
109:
110:
111: int
1.27 schwarze 112: man_valid_pre(struct man *m, struct man_node *n)
1.7 schwarze 113: {
114: v_check *cp;
115:
1.38 schwarze 116: switch (n->type) {
117: case (MAN_TEXT):
118: /* FALLTHROUGH */
119: case (MAN_ROOT):
120: /* FALLTHROUGH */
121: case (MAN_TBL):
1.7 schwarze 122: return(1);
1.38 schwarze 123: default:
124: break;
125: }
1.7 schwarze 126:
127: if (NULL == (cp = man_valids[n->tok].pres))
128: return(1);
129: for ( ; *cp; cp++)
130: if ( ! (*cp)(m, n))
131: return(0);
132: return(1);
133: }
134:
135:
136: int
1.1 kristaps 137: man_valid_post(struct man *m)
138: {
1.7 schwarze 139: v_check *cp;
1.1 kristaps 140:
141: if (MAN_VALID & m->last->flags)
142: return(1);
143: m->last->flags |= MAN_VALID;
144:
145: switch (m->last->type) {
146: case (MAN_TEXT):
1.4 schwarze 147: return(check_text(m, m->last));
1.1 kristaps 148: case (MAN_ROOT):
1.4 schwarze 149: return(check_root(m, m->last));
1.38 schwarze 150: case (MAN_TBL):
151: return(1);
1.1 kristaps 152: default:
153: break;
154: }
155:
156: if (NULL == (cp = man_valids[m->last->tok].posts))
157: return(1);
158: for ( ; *cp; cp++)
159: if ( ! (*cp)(m, m->last))
160: return(0);
161:
162: return(1);
163: }
164:
165:
1.4 schwarze 166: static int
1.7 schwarze 167: check_root(CHKARGS)
1.4 schwarze 168: {
1.7 schwarze 169:
170: if (MAN_BLINE & m->flags)
1.35 schwarze 171: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
172: else if (MAN_ELINE & m->flags)
173: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
1.8 schwarze 174:
175: m->flags &= ~MAN_BLINE;
176: m->flags &= ~MAN_ELINE;
1.7 schwarze 177:
1.25 schwarze 178: if (NULL == m->first->child) {
179: man_nmsg(m, n, MANDOCERR_NODOCBODY);
180: return(0);
181: } else if (NULL == m->meta.title) {
1.35 schwarze 182: man_nmsg(m, n, MANDOCERR_NOTITLE);
183:
1.18 schwarze 184: /*
185: * If a title hasn't been set, do so now (by
186: * implication, date and section also aren't set).
187: */
1.35 schwarze 188:
1.17 schwarze 189: m->meta.title = mandoc_strdup("unknown");
1.18 schwarze 190: m->meta.date = time(NULL);
1.21 schwarze 191: m->meta.msec = mandoc_strdup("1");
1.17 schwarze 192: }
1.15 schwarze 193:
194: return(1);
195: }
196:
197:
198: static int
1.7 schwarze 199: check_text(CHKARGS)
1.4 schwarze 200: {
1.27 schwarze 201: char *p;
1.5 schwarze 202: int pos, c;
1.28 schwarze 203: size_t sz;
1.4 schwarze 204:
1.28 schwarze 205: for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
206: sz = strcspn(p, "\t\\");
207: p += (int)sz;
208:
209: if ('\0' == *p)
210: break;
211:
212: pos += (int)sz;
1.4 schwarze 213:
1.28 schwarze 214: if ('\t' == *p) {
215: if (MAN_LITERAL & m->flags)
216: continue;
217: if (man_pmsg(m, n->line, pos, MANDOCERR_BADTAB))
1.5 schwarze 218: continue;
1.28 schwarze 219: return(0);
220: }
221:
222: /* Check the special character. */
1.25 schwarze 223:
1.28 schwarze 224: c = mandoc_special(p);
225: if (c) {
226: p += c - 1;
227: pos += c - 1;
1.29 schwarze 228: } else
229: man_pmsg(m, n->line, pos, MANDOCERR_BADESCAPE);
1.4 schwarze 230: }
231:
232: return(1);
1.1 kristaps 233: }
234:
235:
236: #define INEQ_DEFINE(x, ineq, name) \
237: static int \
1.7 schwarze 238: check_##name(CHKARGS) \
1.1 kristaps 239: { \
1.4 schwarze 240: if (n->nchild ineq (x)) \
1.1 kristaps 241: return(1); \
1.40 ! schwarze 242: man_vmsg(m, MANDOCERR_ARGCOUNT, n->line, n->pos, \
1.25 schwarze 243: "line arguments %s %d (have %d)", \
244: #ineq, (x), n->nchild); \
1.40 ! schwarze 245: return(1); \
1.1 kristaps 246: }
247:
248: INEQ_DEFINE(0, ==, eq0)
1.10 schwarze 249: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 250: INEQ_DEFINE(2, >=, ge2)
251: INEQ_DEFINE(5, <=, le5)
1.32 schwarze 252:
253: static int
254: check_ft(CHKARGS)
255: {
256: char *cp;
257: int ok;
258:
259: if (0 == n->nchild)
260: return(1);
261:
262: ok = 0;
263: cp = n->child->string;
264: switch (*cp) {
265: case ('1'):
266: /* FALLTHROUGH */
267: case ('2'):
268: /* FALLTHROUGH */
269: case ('3'):
270: /* FALLTHROUGH */
271: case ('4'):
272: /* FALLTHROUGH */
273: case ('I'):
274: /* FALLTHROUGH */
275: case ('P'):
276: /* FALLTHROUGH */
277: case ('R'):
278: if ('\0' == cp[1])
279: ok = 1;
280: break;
281: case ('B'):
282: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
283: ok = 1;
284: break;
285: case ('C'):
286: if ('W' == cp[1] && '\0' == cp[2])
287: ok = 1;
288: break;
289: default:
290: break;
291: }
292:
293: if (0 == ok) {
1.36 schwarze 294: man_vmsg(m, MANDOCERR_BADFONT,
295: n->line, n->pos, "%s", cp);
1.32 schwarze 296: *cp = '\0';
297: }
298:
299: if (1 < n->nchild)
300: man_vmsg(m, MANDOCERR_ARGCOUNT, n->line, n->pos,
1.36 schwarze 301: "want one child (have %d)", n->nchild);
1.32 schwarze 302:
303: return(1);
304: }
1.7 schwarze 305:
306: static int
307: check_sec(CHKARGS)
308: {
1.6 schwarze 309:
1.25 schwarze 310: if (MAN_HEAD == n->type && 0 == n->nchild) {
311: man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
312: return(0);
313: } else if (MAN_BODY == n->type && 0 == n->nchild)
1.35 schwarze 314: man_nmsg(m, n, MANDOCERR_NOBODY);
1.25 schwarze 315:
1.6 schwarze 316: return(1);
317: }
1.7 schwarze 318:
319:
320: static int
1.8 schwarze 321: check_part(CHKARGS)
322: {
323:
324: if (MAN_BODY == n->type && 0 == n->nchild)
1.35 schwarze 325: man_nmsg(m, n, MANDOCERR_NOBODY);
326:
1.8 schwarze 327: return(1);
328: }
329:
1.36 schwarze 330:
331: static int
332: check_par(CHKARGS)
333: {
334:
1.39 schwarze 335: switch (n->type) {
336: case (MAN_BLOCK):
337: if (0 == n->body->nchild)
338: man_node_delete(m, n);
339: break;
340: case (MAN_BODY):
341: if (0 == n->nchild)
342: man_nmsg(m, n, MANDOCERR_IGNPAR);
343: break;
344: case (MAN_HEAD):
345: if (n->nchild)
346: man_nmsg(m, n, MANDOCERR_ARGSLOST);
347: break;
348: default:
349: break;
350: }
1.36 schwarze 351:
352: return(1);
353: }
354:
355:
356: static int
357: check_bline(CHKARGS)
358: {
359:
360: assert( ! (MAN_ELINE & m->flags));
361: if (MAN_BLINE & m->flags) {
362: man_nmsg(m, n, MANDOCERR_SYNTLINESCOPE);
363: return(0);
364: }
365:
366: return(1);
367: }
368:
1.34 schwarze 369: static int
370: post_TH(CHKARGS)
371: {
1.40 ! schwarze 372: const char *p;
1.34 schwarze 373:
374: if (m->meta.title)
375: free(m->meta.title);
376: if (m->meta.vol)
377: free(m->meta.vol);
378: if (m->meta.source)
379: free(m->meta.source);
380: if (m->meta.msec)
381: free(m->meta.msec);
382: if (m->meta.rawdate)
383: free(m->meta.rawdate);
384:
385: m->meta.title = m->meta.vol = m->meta.rawdate =
386: m->meta.msec = m->meta.source = NULL;
387: m->meta.date = 0;
388:
389: /* ->TITLE<- MSEC DATE SOURCE VOL */
390:
391: n = n->child;
1.40 ! schwarze 392: if (n && n->string) {
! 393: for (p = n->string; '\0' != *p; p++) {
! 394: /* Only warn about this once... */
! 395: if (isalpha((u_char)*p) && ! isupper((u_char)*p)) {
! 396: man_nmsg(m, n, MANDOCERR_UPPERCASE);
! 397: break;
! 398: }
! 399: }
! 400: m->meta.title = mandoc_strdup(n->string);
! 401: } else
! 402: m->meta.title = mandoc_strdup("");
1.34 schwarze 403:
404: /* TITLE ->MSEC<- DATE SOURCE VOL */
405:
1.40 ! schwarze 406: if (n)
! 407: n = n->next;
! 408: if (n && n->string)
! 409: m->meta.msec = mandoc_strdup(n->string);
! 410: else
! 411: m->meta.msec = mandoc_strdup("");
1.34 schwarze 412:
413: /* TITLE MSEC ->DATE<- SOURCE VOL */
414:
415: /*
416: * Try to parse the date. If this works, stash the epoch (this
417: * is optimal because we can reformat it in the canonical form).
418: * If it doesn't parse, isn't specified at all, or is an empty
419: * string, then use the current date.
420: */
421:
1.40 ! schwarze 422: if (n)
! 423: n = n->next;
1.34 schwarze 424: if (n && n->string && *n->string) {
425: m->meta.date = mandoc_a2time
426: (MTIME_ISO_8601, n->string);
427: if (0 == m->meta.date) {
428: man_nmsg(m, n, MANDOCERR_BADDATE);
429: m->meta.rawdate = mandoc_strdup(n->string);
430: }
431: } else
432: m->meta.date = time(NULL);
433:
434: /* TITLE MSEC DATE ->SOURCE<- VOL */
435:
436: if (n && (n = n->next))
437: m->meta.source = mandoc_strdup(n->string);
438:
439: /* TITLE MSEC DATE SOURCE ->VOL<- */
440:
441: if (n && (n = n->next))
442: m->meta.vol = mandoc_strdup(n->string);
443:
444: /*
445: * Remove the `TH' node after we've processed it for our
446: * meta-data.
447: */
448: man_node_delete(m, m->last);
449: return(1);
450: }
451:
452: static int
453: post_nf(CHKARGS)
454: {
455:
456: if (MAN_LITERAL & m->flags)
457: man_nmsg(m, n, MANDOCERR_SCOPEREP);
458:
459: m->flags |= MAN_LITERAL;
460: return(1);
461: }
462:
463: static int
464: post_fi(CHKARGS)
465: {
466:
467: if ( ! (MAN_LITERAL & m->flags))
1.39 schwarze 468: man_nmsg(m, n, MANDOCERR_WNOSCOPE);
1.34 schwarze 469:
470: m->flags &= ~MAN_LITERAL;
471: return(1);
472: }
473:
474: static int
475: post_UC(CHKARGS)
476: {
477: static const char * const bsd_versions[] = {
478: "3rd Berkeley Distribution",
479: "4th Berkeley Distribution",
480: "4.2 Berkeley Distribution",
481: "4.3 Berkeley Distribution",
482: "4.4 Berkeley Distribution",
483: };
484:
485: const char *p, *s;
486:
487: n = n->child;
488: n = m->last->child;
489:
490: if (NULL == n || MAN_TEXT != n->type)
491: p = bsd_versions[0];
492: else {
493: s = n->string;
494: if (0 == strcmp(s, "3"))
495: p = bsd_versions[0];
496: else if (0 == strcmp(s, "4"))
497: p = bsd_versions[1];
498: else if (0 == strcmp(s, "5"))
499: p = bsd_versions[2];
500: else if (0 == strcmp(s, "6"))
501: p = bsd_versions[3];
502: else if (0 == strcmp(s, "7"))
503: p = bsd_versions[4];
504: else
505: p = bsd_versions[0];
506: }
507:
508: if (m->meta.source)
509: free(m->meta.source);
510:
511: m->meta.source = mandoc_strdup(p);
512: return(1);
513: }
514:
515: static int
516: post_AT(CHKARGS)
517: {
518: static const char * const unix_versions[] = {
519: "7th Edition",
520: "System III",
521: "System V",
522: "System V Release 2",
523: };
524:
525: const char *p, *s;
526: struct man_node *nn;
527:
528: n = n->child;
529:
530: if (NULL == n || MAN_TEXT != n->type)
531: p = unix_versions[0];
532: else {
533: s = n->string;
534: if (0 == strcmp(s, "3"))
535: p = unix_versions[0];
536: else if (0 == strcmp(s, "4"))
537: p = unix_versions[1];
538: else if (0 == strcmp(s, "5")) {
539: nn = n->next;
540: if (nn && MAN_TEXT == nn->type && nn->string[0])
541: p = unix_versions[3];
542: else
543: p = unix_versions[2];
544: } else
545: p = unix_versions[0];
546: }
547:
548: if (m->meta.source)
549: free(m->meta.source);
550:
551: m->meta.source = mandoc_strdup(p);
552: return(1);
553: }