Annotation of mandoc/man_validate.c, Revision 1.60
1.60 ! schwarze 1: /* $Id: man_validate.c,v 1.59 2011/01/12 16:55:22 kristaps 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):
123: /* FALLTHROUGH */
124: case (MAN_TBL):
1.17 kristaps 125: return(1);
1.57 kristaps 126: default:
127: break;
128: }
1.17 kristaps 129:
130: if (NULL == (cp = man_valids[n->tok].pres))
131: return(1);
132: for ( ; *cp; cp++)
133: if ( ! (*cp)(m, n))
134: return(0);
135: return(1);
136: }
137:
138:
139: int
1.1 kristaps 140: man_valid_post(struct man *m)
141: {
1.17 kristaps 142: v_check *cp;
1.1 kristaps 143:
144: if (MAN_VALID & m->last->flags)
145: return(1);
146: m->last->flags |= MAN_VALID;
147:
148: switch (m->last->type) {
149: case (MAN_TEXT):
1.11 kristaps 150: return(check_text(m, m->last));
1.1 kristaps 151: case (MAN_ROOT):
1.14 kristaps 152: return(check_root(m, m->last));
1.57 kristaps 153: case (MAN_TBL):
154: return(1);
1.1 kristaps 155: default:
156: break;
157: }
158:
159: if (NULL == (cp = man_valids[m->last->tok].posts))
160: return(1);
161: for ( ; *cp; cp++)
162: if ( ! (*cp)(m, m->last))
163: return(0);
164:
165: return(1);
166: }
167:
168:
1.11 kristaps 169: static int
1.17 kristaps 170: check_root(CHKARGS)
1.14 kristaps 171: {
1.17 kristaps 172:
173: if (MAN_BLINE & m->flags)
1.54 kristaps 174: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
175: else if (MAN_ELINE & m->flags)
176: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
1.22 kristaps 177:
178: m->flags &= ~MAN_BLINE;
179: m->flags &= ~MAN_ELINE;
1.17 kristaps 180:
1.41 kristaps 181: if (NULL == m->first->child) {
182: man_nmsg(m, n, MANDOCERR_NODOCBODY);
183: return(0);
184: } else if (NULL == m->meta.title) {
1.54 kristaps 185: man_nmsg(m, n, MANDOCERR_NOTITLE);
186:
1.34 kristaps 187: /*
188: * If a title hasn't been set, do so now (by
189: * implication, date and section also aren't set).
190: */
1.54 kristaps 191:
1.34 kristaps 192: m->meta.title = mandoc_strdup("unknown");
193: m->meta.date = time(NULL);
1.37 kristaps 194: m->meta.msec = mandoc_strdup("1");
1.34 kristaps 195: }
1.32 kristaps 196:
197: return(1);
198: }
199:
200:
201: static int
1.17 kristaps 202: check_text(CHKARGS)
1.11 kristaps 203: {
1.43 kristaps 204: char *p;
1.15 kristaps 205: int pos, c;
1.46 kristaps 206: size_t sz;
1.11 kristaps 207:
1.46 kristaps 208: for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
209: sz = strcspn(p, "\t\\");
210: p += (int)sz;
211:
212: if ('\0' == *p)
213: break;
214:
215: pos += (int)sz;
1.11 kristaps 216:
1.46 kristaps 217: if ('\t' == *p) {
218: if (MAN_LITERAL & m->flags)
219: continue;
220: if (man_pmsg(m, n->line, pos, MANDOCERR_BADTAB))
1.15 kristaps 221: continue;
1.46 kristaps 222: return(0);
1.15 kristaps 223: }
1.45 kristaps 224:
1.46 kristaps 225: /* Check the special character. */
1.15 kristaps 226:
1.46 kristaps 227: c = mandoc_special(p);
228: if (c) {
229: p += c - 1;
230: pos += c - 1;
1.49 schwarze 231: } else
232: man_pmsg(m, n->line, pos, MANDOCERR_BADESCAPE);
1.11 kristaps 233: }
234:
235: return(1);
1.1 kristaps 236: }
237:
238:
239: #define INEQ_DEFINE(x, ineq, name) \
240: static int \
1.17 kristaps 241: check_##name(CHKARGS) \
1.1 kristaps 242: { \
1.11 kristaps 243: if (n->nchild ineq (x)) \
1.1 kristaps 244: return(1); \
1.60 ! schwarze 245: man_vmsg(m, MANDOCERR_ARGCOUNT, n->line, n->pos, \
1.41 kristaps 246: "line arguments %s %d (have %d)", \
247: #ineq, (x), n->nchild); \
1.60 ! schwarze 248: return(1); \
1.1 kristaps 249: }
250:
251: INEQ_DEFINE(0, ==, eq0)
1.25 kristaps 252: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 253: INEQ_DEFINE(2, >=, ge2)
254: INEQ_DEFINE(5, <=, le5)
255:
1.55 kristaps 256: static int
257: check_ft(CHKARGS)
258: {
259: char *cp;
260: int ok;
261:
262: if (0 == n->nchild)
263: return(1);
264:
265: ok = 0;
266: cp = n->child->string;
267: switch (*cp) {
268: case ('1'):
269: /* FALLTHROUGH */
270: case ('2'):
271: /* FALLTHROUGH */
272: case ('3'):
273: /* FALLTHROUGH */
274: case ('4'):
275: /* FALLTHROUGH */
276: case ('I'):
277: /* FALLTHROUGH */
278: case ('P'):
279: /* FALLTHROUGH */
280: case ('R'):
281: if ('\0' == cp[1])
282: ok = 1;
283: break;
284: case ('B'):
285: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
286: ok = 1;
287: break;
288: case ('C'):
289: if ('W' == cp[1] && '\0' == cp[2])
290: ok = 1;
291: break;
292: default:
293: break;
294: }
295:
296: if (0 == ok) {
297: man_vmsg(m, MANDOCERR_BADFONT,
298: n->line, n->pos, "%s", cp);
299: *cp = '\0';
300: }
301:
302: if (1 < n->nchild)
303: man_vmsg(m, MANDOCERR_ARGCOUNT, n->line, n->pos,
304: "want one child (have %d)", n->nchild);
305:
306: return(1);
307: }
1.16 kristaps 308:
309: static int
1.17 kristaps 310: check_sec(CHKARGS)
311: {
1.16 kristaps 312:
1.41 kristaps 313: if (MAN_HEAD == n->type && 0 == n->nchild) {
314: man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
315: return(0);
316: } else if (MAN_BODY == n->type && 0 == n->nchild)
1.54 kristaps 317: man_nmsg(m, n, MANDOCERR_NOBODY);
1.41 kristaps 318:
1.16 kristaps 319: return(1);
320: }
1.17 kristaps 321:
322:
323: static int
1.23 kristaps 324: check_part(CHKARGS)
325: {
326:
327: if (MAN_BODY == n->type && 0 == n->nchild)
1.54 kristaps 328: man_nmsg(m, n, MANDOCERR_NOBODY);
329:
1.23 kristaps 330: return(1);
331: }
332:
333:
334: static int
1.17 kristaps 335: check_par(CHKARGS)
336: {
337:
1.59 kristaps 338: switch (n->type) {
339: case (MAN_BLOCK):
340: if (0 == n->body->nchild)
341: man_node_delete(m, n);
342: break;
343: case (MAN_BODY):
344: if (0 == n->nchild)
345: man_nmsg(m, n, MANDOCERR_IGNPAR);
346: break;
347: case (MAN_HEAD):
348: if (n->nchild)
349: man_nmsg(m, n, MANDOCERR_ARGSLOST);
350: break;
351: default:
352: break;
353: }
1.17 kristaps 354:
355: return(1);
356: }
357:
358:
359: static int
360: check_bline(CHKARGS)
361: {
362:
1.22 kristaps 363: assert( ! (MAN_ELINE & m->flags));
1.41 kristaps 364: if (MAN_BLINE & m->flags) {
365: man_nmsg(m, n, MANDOCERR_SYNTLINESCOPE);
366: return(0);
367: }
1.31 kristaps 368:
1.18 kristaps 369: return(1);
1.17 kristaps 370: }
371:
1.51 kristaps 372: static int
373: post_TH(CHKARGS)
374: {
1.60 ! schwarze 375: const char *p;
1.51 kristaps 376:
377: if (m->meta.title)
378: free(m->meta.title);
379: if (m->meta.vol)
380: free(m->meta.vol);
381: if (m->meta.source)
382: free(m->meta.source);
383: if (m->meta.msec)
384: free(m->meta.msec);
385: if (m->meta.rawdate)
386: free(m->meta.rawdate);
387:
388: m->meta.title = m->meta.vol = m->meta.rawdate =
389: m->meta.msec = m->meta.source = NULL;
390: m->meta.date = 0;
391:
392: /* ->TITLE<- MSEC DATE SOURCE VOL */
393:
394: n = n->child;
1.60 ! schwarze 395: if (n && n->string) {
! 396: for (p = n->string; '\0' != *p; p++) {
! 397: /* Only warn about this once... */
! 398: if (isalpha((u_char)*p) && ! isupper((u_char)*p)) {
! 399: man_nmsg(m, n, MANDOCERR_UPPERCASE);
! 400: break;
! 401: }
! 402: }
! 403: m->meta.title = mandoc_strdup(n->string);
! 404: } else
! 405: m->meta.title = mandoc_strdup("");
1.51 kristaps 406:
407: /* TITLE ->MSEC<- DATE SOURCE VOL */
408:
1.60 ! schwarze 409: if (n)
! 410: n = n->next;
! 411: if (n && n->string)
! 412: m->meta.msec = mandoc_strdup(n->string);
! 413: else
! 414: m->meta.msec = mandoc_strdup("");
1.51 kristaps 415:
416: /* TITLE MSEC ->DATE<- SOURCE VOL */
417:
418: /*
419: * Try to parse the date. If this works, stash the epoch (this
420: * is optimal because we can reformat it in the canonical form).
421: * If it doesn't parse, isn't specified at all, or is an empty
422: * string, then use the current date.
423: */
424:
1.60 ! schwarze 425: if (n)
! 426: n = n->next;
1.51 kristaps 427: if (n && n->string && *n->string) {
428: m->meta.date = mandoc_a2time
429: (MTIME_ISO_8601, n->string);
430: if (0 == m->meta.date) {
431: man_nmsg(m, n, MANDOCERR_BADDATE);
432: m->meta.rawdate = mandoc_strdup(n->string);
433: }
434: } else
435: m->meta.date = time(NULL);
436:
437: /* TITLE MSEC DATE ->SOURCE<- VOL */
438:
439: if (n && (n = n->next))
440: m->meta.source = mandoc_strdup(n->string);
441:
442: /* TITLE MSEC DATE SOURCE ->VOL<- */
443:
444: if (n && (n = n->next))
445: m->meta.vol = mandoc_strdup(n->string);
446:
447: /*
448: * Remove the `TH' node after we've processed it for our
449: * meta-data.
450: */
451: man_node_delete(m, m->last);
452: return(1);
453: }
454:
455: static int
456: post_nf(CHKARGS)
457: {
458:
459: if (MAN_LITERAL & m->flags)
460: man_nmsg(m, n, MANDOCERR_SCOPEREP);
461:
462: m->flags |= MAN_LITERAL;
463: return(1);
464: }
465:
466: static int
467: post_fi(CHKARGS)
468: {
469:
470: if ( ! (MAN_LITERAL & m->flags))
1.58 kristaps 471: man_nmsg(m, n, MANDOCERR_WNOSCOPE);
1.51 kristaps 472:
473: m->flags &= ~MAN_LITERAL;
474: return(1);
475: }
476:
477: static int
478: post_UC(CHKARGS)
479: {
480: static const char * const bsd_versions[] = {
481: "3rd Berkeley Distribution",
482: "4th Berkeley Distribution",
483: "4.2 Berkeley Distribution",
484: "4.3 Berkeley Distribution",
485: "4.4 Berkeley Distribution",
486: };
487:
488: const char *p, *s;
489:
490: n = n->child;
491: n = m->last->child;
492:
493: if (NULL == n || MAN_TEXT != n->type)
494: p = bsd_versions[0];
495: else {
496: s = n->string;
497: if (0 == strcmp(s, "3"))
498: p = bsd_versions[0];
499: else if (0 == strcmp(s, "4"))
500: p = bsd_versions[1];
501: else if (0 == strcmp(s, "5"))
502: p = bsd_versions[2];
503: else if (0 == strcmp(s, "6"))
504: p = bsd_versions[3];
505: else if (0 == strcmp(s, "7"))
506: p = bsd_versions[4];
507: else
508: p = bsd_versions[0];
509: }
510:
511: if (m->meta.source)
512: free(m->meta.source);
513:
514: m->meta.source = mandoc_strdup(p);
515: return(1);
516: }
517:
518: static int
519: post_AT(CHKARGS)
520: {
521: static const char * const unix_versions[] = {
522: "7th Edition",
523: "System III",
524: "System V",
525: "System V Release 2",
526: };
527:
528: const char *p, *s;
529: struct man_node *nn;
530:
531: n = n->child;
532:
533: if (NULL == n || MAN_TEXT != n->type)
534: p = unix_versions[0];
535: else {
536: s = n->string;
537: if (0 == strcmp(s, "3"))
538: p = unix_versions[0];
539: else if (0 == strcmp(s, "4"))
540: p = unix_versions[1];
541: else if (0 == strcmp(s, "5")) {
542: nn = n->next;
543: if (nn && MAN_TEXT == nn->type && nn->string[0])
544: p = unix_versions[3];
545: else
546: p = unix_versions[2];
547: } else
548: p = unix_versions[0];
549: }
550:
551: if (m->meta.source)
552: free(m->meta.source);
553:
554: m->meta.source = mandoc_strdup(p);
555: return(1);
556: }
CVSweb