Annotation of mandoc/man_validate.c, Revision 1.61
1.61 ! kristaps 1: /* $Id: man_validate.c,v 1.60 2011/01/17 00:21:29 schwarze Exp $ */
1.1 kristaps 2: /*
1.48 schwarze 3: * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
1.1 kristaps 4: *
5: * Permission to use, copy, modify, and distribute this software for any
1.8 kristaps 6: * purpose with or without fee is hereby granted, provided that the above
7: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 8: *
1.8 kristaps 9: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 16: */
1.28 kristaps 17: #ifdef HAVE_CONFIG_H
18: #include "config.h"
19: #endif
20:
1.1 kristaps 21: #include <sys/types.h>
22:
23: #include <assert.h>
24: #include <ctype.h>
1.16 kristaps 25: #include <errno.h>
26: #include <limits.h>
1.1 kristaps 27: #include <stdarg.h>
28: #include <stdlib.h>
1.46 kristaps 29: #include <string.h>
1.50 kristaps 30: #include <time.h>
1.1 kristaps 31:
1.41 kristaps 32: #include "mandoc.h"
1.1 kristaps 33: #include "libman.h"
1.15 kristaps 34: #include "libmandoc.h"
1.1 kristaps 35:
1.43 kristaps 36: #define CHKARGS struct man *m, struct man_node *n
1.1 kristaps 37:
1.17 kristaps 38: typedef int (*v_check)(CHKARGS);
1.1 kristaps 39:
40: struct man_valid {
1.17 kristaps 41: v_check *pres;
42: v_check *posts;
1.1 kristaps 43: };
44:
1.17 kristaps 45: static int check_bline(CHKARGS);
46: static int check_eq0(CHKARGS);
1.55 kristaps 47: static int check_ft(CHKARGS);
1.25 kristaps 48: static int check_le1(CHKARGS);
1.17 kristaps 49: static int check_ge2(CHKARGS);
50: static int check_le5(CHKARGS);
51: static int check_par(CHKARGS);
1.23 kristaps 52: static int check_part(CHKARGS);
1.17 kristaps 53: static int check_root(CHKARGS);
54: static int check_sec(CHKARGS);
55: static int check_text(CHKARGS);
56:
1.51 kristaps 57: static int post_AT(CHKARGS);
58: static int post_fi(CHKARGS);
59: static int post_nf(CHKARGS);
60: static int post_TH(CHKARGS);
61: static int post_UC(CHKARGS);
62:
63: static v_check posts_at[] = { post_AT, NULL };
1.17 kristaps 64: static v_check posts_eq0[] = { check_eq0, NULL };
1.51 kristaps 65: static v_check posts_fi[] = { check_eq0, post_fi, NULL };
66: static v_check posts_le1[] = { check_le1, NULL };
1.55 kristaps 67: static v_check posts_ft[] = { check_ft, NULL };
1.51 kristaps 68: static v_check posts_nf[] = { check_eq0, post_nf, NULL };
1.17 kristaps 69: static v_check posts_par[] = { check_par, NULL };
1.23 kristaps 70: static v_check posts_part[] = { check_part, NULL };
1.17 kristaps 71: static v_check posts_sec[] = { check_sec, NULL };
1.60 schwarze 72: static v_check posts_th[] = { check_ge2, check_le5, post_TH, NULL };
1.51 kristaps 73: static v_check posts_uc[] = { post_UC, NULL };
1.17 kristaps 74: static v_check pres_bline[] = { check_bline, NULL };
1.1 kristaps 75:
1.51 kristaps 76:
1.1 kristaps 77: static const struct man_valid man_valids[MAN_MAX] = {
1.29 kristaps 78: { NULL, posts_eq0 }, /* br */
1.32 kristaps 79: { pres_bline, posts_th }, /* TH */
1.17 kristaps 80: { pres_bline, posts_sec }, /* SH */
81: { pres_bline, posts_sec }, /* SS */
1.59 kristaps 82: { pres_bline, NULL }, /* TP */
1.17 kristaps 83: { pres_bline, posts_par }, /* LP */
84: { pres_bline, posts_par }, /* PP */
85: { pres_bline, posts_par }, /* P */
1.59 kristaps 86: { pres_bline, NULL }, /* IP */
87: { pres_bline, NULL }, /* HP */
1.22 kristaps 88: { NULL, NULL }, /* SM */
89: { NULL, NULL }, /* SB */
1.17 kristaps 90: { NULL, NULL }, /* BI */
91: { NULL, NULL }, /* IB */
92: { NULL, NULL }, /* BR */
93: { NULL, NULL }, /* RB */
1.22 kristaps 94: { NULL, NULL }, /* R */
95: { NULL, NULL }, /* B */
96: { NULL, NULL }, /* I */
1.17 kristaps 97: { NULL, NULL }, /* IR */
98: { NULL, NULL }, /* RI */
1.47 kristaps 99: { NULL, posts_eq0 }, /* na */ /* FIXME: should warn only. */
100: { NULL, posts_le1 }, /* sp */ /* FIXME: should warn only. */
1.51 kristaps 101: { pres_bline, posts_nf }, /* nf */
102: { pres_bline, posts_fi }, /* fi */
1.19 kristaps 103: { NULL, NULL }, /* RE */
1.23 kristaps 104: { NULL, posts_part }, /* RS */
1.21 kristaps 105: { NULL, NULL }, /* DT */
1.51 kristaps 106: { NULL, posts_uc }, /* UC */
1.26 kristaps 107: { NULL, NULL }, /* PD */
1.51 kristaps 108: { NULL, posts_at }, /* AT */
1.47 kristaps 109: { NULL, NULL }, /* in */
1.55 kristaps 110: { NULL, posts_ft }, /* ft */
1.1 kristaps 111: };
112:
113:
114: int
1.43 kristaps 115: man_valid_pre(struct man *m, struct man_node *n)
1.17 kristaps 116: {
117: v_check *cp;
118:
1.57 kristaps 119: switch (n->type) {
120: case (MAN_TEXT):
121: /* FALLTHROUGH */
122: case (MAN_ROOT):
1.61 ! kristaps 123: /* FALLTHROUGH */
! 124: case (MAN_EQN):
1.57 kristaps 125: /* FALLTHROUGH */
126: case (MAN_TBL):
1.17 kristaps 127: return(1);
1.57 kristaps 128: default:
129: break;
130: }
1.17 kristaps 131:
132: if (NULL == (cp = man_valids[n->tok].pres))
133: return(1);
134: for ( ; *cp; cp++)
135: if ( ! (*cp)(m, n))
136: return(0);
137: return(1);
138: }
139:
140:
141: int
1.1 kristaps 142: man_valid_post(struct man *m)
143: {
1.17 kristaps 144: v_check *cp;
1.1 kristaps 145:
146: if (MAN_VALID & m->last->flags)
147: return(1);
148: m->last->flags |= MAN_VALID;
149:
150: switch (m->last->type) {
151: case (MAN_TEXT):
1.11 kristaps 152: return(check_text(m, m->last));
1.1 kristaps 153: case (MAN_ROOT):
1.14 kristaps 154: return(check_root(m, m->last));
1.57 kristaps 155: case (MAN_TBL):
156: return(1);
1.1 kristaps 157: default:
158: break;
159: }
160:
161: if (NULL == (cp = man_valids[m->last->tok].posts))
162: return(1);
163: for ( ; *cp; cp++)
164: if ( ! (*cp)(m, m->last))
165: return(0);
166:
167: return(1);
168: }
169:
170:
1.11 kristaps 171: static int
1.17 kristaps 172: check_root(CHKARGS)
1.14 kristaps 173: {
1.17 kristaps 174:
175: if (MAN_BLINE & m->flags)
1.54 kristaps 176: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
177: else if (MAN_ELINE & m->flags)
178: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
1.22 kristaps 179:
180: m->flags &= ~MAN_BLINE;
181: m->flags &= ~MAN_ELINE;
1.17 kristaps 182:
1.41 kristaps 183: if (NULL == m->first->child) {
184: man_nmsg(m, n, MANDOCERR_NODOCBODY);
185: return(0);
186: } else if (NULL == m->meta.title) {
1.54 kristaps 187: man_nmsg(m, n, MANDOCERR_NOTITLE);
188:
1.34 kristaps 189: /*
190: * If a title hasn't been set, do so now (by
191: * implication, date and section also aren't set).
192: */
1.54 kristaps 193:
1.34 kristaps 194: m->meta.title = mandoc_strdup("unknown");
195: m->meta.date = time(NULL);
1.37 kristaps 196: m->meta.msec = mandoc_strdup("1");
1.34 kristaps 197: }
1.32 kristaps 198:
199: return(1);
200: }
201:
202:
203: static int
1.17 kristaps 204: check_text(CHKARGS)
1.11 kristaps 205: {
1.43 kristaps 206: char *p;
1.15 kristaps 207: int pos, c;
1.46 kristaps 208: size_t sz;
1.11 kristaps 209:
1.46 kristaps 210: for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
211: sz = strcspn(p, "\t\\");
212: p += (int)sz;
213:
214: if ('\0' == *p)
215: break;
216:
217: pos += (int)sz;
1.11 kristaps 218:
1.46 kristaps 219: if ('\t' == *p) {
220: if (MAN_LITERAL & m->flags)
221: continue;
222: if (man_pmsg(m, n->line, pos, MANDOCERR_BADTAB))
1.15 kristaps 223: continue;
1.46 kristaps 224: return(0);
1.15 kristaps 225: }
1.45 kristaps 226:
1.46 kristaps 227: /* Check the special character. */
1.15 kristaps 228:
1.46 kristaps 229: c = mandoc_special(p);
230: if (c) {
231: p += c - 1;
232: pos += c - 1;
1.49 schwarze 233: } else
234: man_pmsg(m, n->line, pos, MANDOCERR_BADESCAPE);
1.11 kristaps 235: }
236:
237: return(1);
1.1 kristaps 238: }
239:
240:
241: #define INEQ_DEFINE(x, ineq, name) \
242: static int \
1.17 kristaps 243: check_##name(CHKARGS) \
1.1 kristaps 244: { \
1.11 kristaps 245: if (n->nchild ineq (x)) \
1.1 kristaps 246: return(1); \
1.60 schwarze 247: man_vmsg(m, MANDOCERR_ARGCOUNT, n->line, n->pos, \
1.41 kristaps 248: "line arguments %s %d (have %d)", \
249: #ineq, (x), n->nchild); \
1.60 schwarze 250: return(1); \
1.1 kristaps 251: }
252:
253: INEQ_DEFINE(0, ==, eq0)
1.25 kristaps 254: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 255: INEQ_DEFINE(2, >=, ge2)
256: INEQ_DEFINE(5, <=, le5)
257:
1.55 kristaps 258: static int
259: check_ft(CHKARGS)
260: {
261: char *cp;
262: int ok;
263:
264: if (0 == n->nchild)
265: return(1);
266:
267: ok = 0;
268: cp = n->child->string;
269: switch (*cp) {
270: case ('1'):
271: /* FALLTHROUGH */
272: case ('2'):
273: /* FALLTHROUGH */
274: case ('3'):
275: /* FALLTHROUGH */
276: case ('4'):
277: /* FALLTHROUGH */
278: case ('I'):
279: /* FALLTHROUGH */
280: case ('P'):
281: /* FALLTHROUGH */
282: case ('R'):
283: if ('\0' == cp[1])
284: ok = 1;
285: break;
286: case ('B'):
287: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
288: ok = 1;
289: break;
290: case ('C'):
291: if ('W' == cp[1] && '\0' == cp[2])
292: ok = 1;
293: break;
294: default:
295: break;
296: }
297:
298: if (0 == ok) {
299: man_vmsg(m, MANDOCERR_BADFONT,
300: n->line, n->pos, "%s", cp);
301: *cp = '\0';
302: }
303:
304: if (1 < n->nchild)
305: man_vmsg(m, MANDOCERR_ARGCOUNT, n->line, n->pos,
306: "want one child (have %d)", n->nchild);
307:
308: return(1);
309: }
1.16 kristaps 310:
311: static int
1.17 kristaps 312: check_sec(CHKARGS)
313: {
1.16 kristaps 314:
1.41 kristaps 315: if (MAN_HEAD == n->type && 0 == n->nchild) {
316: man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
317: return(0);
318: } else if (MAN_BODY == n->type && 0 == n->nchild)
1.54 kristaps 319: man_nmsg(m, n, MANDOCERR_NOBODY);
1.41 kristaps 320:
1.16 kristaps 321: return(1);
322: }
1.17 kristaps 323:
324:
325: static int
1.23 kristaps 326: check_part(CHKARGS)
327: {
328:
329: if (MAN_BODY == n->type && 0 == n->nchild)
1.54 kristaps 330: man_nmsg(m, n, MANDOCERR_NOBODY);
331:
1.23 kristaps 332: return(1);
333: }
334:
335:
336: static int
1.17 kristaps 337: check_par(CHKARGS)
338: {
339:
1.59 kristaps 340: switch (n->type) {
341: case (MAN_BLOCK):
342: if (0 == n->body->nchild)
343: man_node_delete(m, n);
344: break;
345: case (MAN_BODY):
346: if (0 == n->nchild)
347: man_nmsg(m, n, MANDOCERR_IGNPAR);
348: break;
349: case (MAN_HEAD):
350: if (n->nchild)
351: man_nmsg(m, n, MANDOCERR_ARGSLOST);
352: break;
353: default:
354: break;
355: }
1.17 kristaps 356:
357: return(1);
358: }
359:
360:
361: static int
362: check_bline(CHKARGS)
363: {
364:
1.22 kristaps 365: assert( ! (MAN_ELINE & m->flags));
1.41 kristaps 366: if (MAN_BLINE & m->flags) {
367: man_nmsg(m, n, MANDOCERR_SYNTLINESCOPE);
368: return(0);
369: }
1.31 kristaps 370:
1.18 kristaps 371: return(1);
1.17 kristaps 372: }
373:
1.51 kristaps 374: static int
375: post_TH(CHKARGS)
376: {
1.60 schwarze 377: const char *p;
1.51 kristaps 378:
379: if (m->meta.title)
380: free(m->meta.title);
381: if (m->meta.vol)
382: free(m->meta.vol);
383: if (m->meta.source)
384: free(m->meta.source);
385: if (m->meta.msec)
386: free(m->meta.msec);
387: if (m->meta.rawdate)
388: free(m->meta.rawdate);
389:
390: m->meta.title = m->meta.vol = m->meta.rawdate =
391: m->meta.msec = m->meta.source = NULL;
392: m->meta.date = 0;
393:
394: /* ->TITLE<- MSEC DATE SOURCE VOL */
395:
396: n = n->child;
1.60 schwarze 397: if (n && n->string) {
398: for (p = n->string; '\0' != *p; p++) {
399: /* Only warn about this once... */
400: if (isalpha((u_char)*p) && ! isupper((u_char)*p)) {
401: man_nmsg(m, n, MANDOCERR_UPPERCASE);
402: break;
403: }
404: }
405: m->meta.title = mandoc_strdup(n->string);
406: } else
407: m->meta.title = mandoc_strdup("");
1.51 kristaps 408:
409: /* TITLE ->MSEC<- DATE SOURCE VOL */
410:
1.60 schwarze 411: if (n)
412: n = n->next;
413: if (n && n->string)
414: m->meta.msec = mandoc_strdup(n->string);
415: else
416: m->meta.msec = mandoc_strdup("");
1.51 kristaps 417:
418: /* TITLE MSEC ->DATE<- SOURCE VOL */
419:
420: /*
421: * Try to parse the date. If this works, stash the epoch (this
422: * is optimal because we can reformat it in the canonical form).
423: * If it doesn't parse, isn't specified at all, or is an empty
424: * string, then use the current date.
425: */
426:
1.60 schwarze 427: if (n)
428: n = n->next;
1.51 kristaps 429: if (n && n->string && *n->string) {
430: m->meta.date = mandoc_a2time
431: (MTIME_ISO_8601, n->string);
432: if (0 == m->meta.date) {
433: man_nmsg(m, n, MANDOCERR_BADDATE);
434: m->meta.rawdate = mandoc_strdup(n->string);
435: }
436: } else
437: m->meta.date = time(NULL);
438:
439: /* TITLE MSEC DATE ->SOURCE<- VOL */
440:
441: if (n && (n = n->next))
442: m->meta.source = mandoc_strdup(n->string);
443:
444: /* TITLE MSEC DATE SOURCE ->VOL<- */
445:
446: if (n && (n = n->next))
447: m->meta.vol = mandoc_strdup(n->string);
448:
449: /*
450: * Remove the `TH' node after we've processed it for our
451: * meta-data.
452: */
453: man_node_delete(m, m->last);
454: return(1);
455: }
456:
457: static int
458: post_nf(CHKARGS)
459: {
460:
461: if (MAN_LITERAL & m->flags)
462: man_nmsg(m, n, MANDOCERR_SCOPEREP);
463:
464: m->flags |= MAN_LITERAL;
465: return(1);
466: }
467:
468: static int
469: post_fi(CHKARGS)
470: {
471:
472: if ( ! (MAN_LITERAL & m->flags))
1.58 kristaps 473: man_nmsg(m, n, MANDOCERR_WNOSCOPE);
1.51 kristaps 474:
475: m->flags &= ~MAN_LITERAL;
476: return(1);
477: }
478:
479: static int
480: post_UC(CHKARGS)
481: {
482: static const char * const bsd_versions[] = {
483: "3rd Berkeley Distribution",
484: "4th Berkeley Distribution",
485: "4.2 Berkeley Distribution",
486: "4.3 Berkeley Distribution",
487: "4.4 Berkeley Distribution",
488: };
489:
490: const char *p, *s;
491:
492: n = n->child;
493: n = m->last->child;
494:
495: if (NULL == n || MAN_TEXT != n->type)
496: p = bsd_versions[0];
497: else {
498: s = n->string;
499: if (0 == strcmp(s, "3"))
500: p = bsd_versions[0];
501: else if (0 == strcmp(s, "4"))
502: p = bsd_versions[1];
503: else if (0 == strcmp(s, "5"))
504: p = bsd_versions[2];
505: else if (0 == strcmp(s, "6"))
506: p = bsd_versions[3];
507: else if (0 == strcmp(s, "7"))
508: p = bsd_versions[4];
509: else
510: p = bsd_versions[0];
511: }
512:
513: if (m->meta.source)
514: free(m->meta.source);
515:
516: m->meta.source = mandoc_strdup(p);
517: return(1);
518: }
519:
520: static int
521: post_AT(CHKARGS)
522: {
523: static const char * const unix_versions[] = {
524: "7th Edition",
525: "System III",
526: "System V",
527: "System V Release 2",
528: };
529:
530: const char *p, *s;
531: struct man_node *nn;
532:
533: n = n->child;
534:
535: if (NULL == n || MAN_TEXT != n->type)
536: p = unix_versions[0];
537: else {
538: s = n->string;
539: if (0 == strcmp(s, "3"))
540: p = unix_versions[0];
541: else if (0 == strcmp(s, "4"))
542: p = unix_versions[1];
543: else if (0 == strcmp(s, "5")) {
544: nn = n->next;
545: if (nn && MAN_TEXT == nn->type && nn->string[0])
546: p = unix_versions[3];
547: else
548: p = unix_versions[2];
549: } else
550: p = unix_versions[0];
551: }
552:
553: if (m->meta.source)
554: free(m->meta.source);
555:
556: m->meta.source = mandoc_strdup(p);
557: return(1);
558: }
CVSweb