Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.50
1.50 ! schwarze 1: /* $Id: man_validate.c,v 1.49 2011/11/03 20:46:40 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:
1.44 schwarze 29: #include "man.h"
1.25 schwarze 30: #include "mandoc.h"
1.1 kristaps 31: #include "libman.h"
1.5 schwarze 32: #include "libmandoc.h"
1.1 kristaps 33:
1.27 schwarze 34: #define CHKARGS struct man *m, struct man_node *n
1.1 kristaps 35:
1.7 schwarze 36: typedef int (*v_check)(CHKARGS);
1.1 kristaps 37:
38: struct man_valid {
1.7 schwarze 39: v_check *pres;
40: v_check *posts;
1.1 kristaps 41: };
42:
1.7 schwarze 43: static int check_eq0(CHKARGS);
1.10 schwarze 44: static int check_le1(CHKARGS);
1.7 schwarze 45: static int check_ge2(CHKARGS);
46: static int check_le5(CHKARGS);
47: static int check_par(CHKARGS);
1.8 schwarze 48: static int check_part(CHKARGS);
1.7 schwarze 49: static int check_root(CHKARGS);
1.45 schwarze 50: static void check_text(CHKARGS);
1.7 schwarze 51:
1.34 schwarze 52: static int post_AT(CHKARGS);
1.46 schwarze 53: static int post_vs(CHKARGS);
1.34 schwarze 54: static int post_fi(CHKARGS);
1.47 schwarze 55: static int post_ft(CHKARGS);
1.34 schwarze 56: static int post_nf(CHKARGS);
1.47 schwarze 57: static int post_sec(CHKARGS);
1.34 schwarze 58: static int post_TH(CHKARGS);
59: static int post_UC(CHKARGS);
1.47 schwarze 60: static int pre_sec(CHKARGS);
1.34 schwarze 61:
62: static v_check posts_at[] = { post_AT, NULL };
1.46 schwarze 63: static v_check posts_br[] = { post_vs, check_eq0, NULL };
1.7 schwarze 64: static v_check posts_eq0[] = { check_eq0, NULL };
1.34 schwarze 65: static v_check posts_fi[] = { check_eq0, post_fi, NULL };
1.47 schwarze 66: static v_check posts_ft[] = { post_ft, NULL };
1.34 schwarze 67: static v_check posts_nf[] = { check_eq0, post_nf, NULL };
1.7 schwarze 68: static v_check posts_par[] = { check_par, NULL };
1.8 schwarze 69: static v_check posts_part[] = { check_part, NULL };
1.47 schwarze 70: static v_check posts_sec[] = { post_sec, NULL };
1.46 schwarze 71: static v_check posts_sp[] = { post_vs, check_le1, NULL };
1.40 schwarze 72: static v_check posts_th[] = { check_ge2, check_le5, post_TH, NULL };
1.34 schwarze 73: static v_check posts_uc[] = { post_UC, NULL };
1.50 ! schwarze 74: static v_check pres_sec[] = { pre_sec, NULL };
1.1 kristaps 75:
76: static const struct man_valid man_valids[MAN_MAX] = {
1.46 schwarze 77: { NULL, posts_br }, /* br */
1.50 ! schwarze 78: { NULL, posts_th }, /* TH */
1.47 schwarze 79: { pres_sec, posts_sec }, /* SH */
80: { pres_sec, posts_sec }, /* SS */
1.50 ! schwarze 81: { NULL, NULL }, /* TP */
! 82: { NULL, posts_par }, /* LP */
! 83: { NULL, posts_par }, /* PP */
! 84: { NULL, posts_par }, /* P */
! 85: { NULL, NULL }, /* IP */
! 86: { NULL, NULL }, /* HP */
1.8 schwarze 87: { NULL, NULL }, /* SM */
88: { NULL, NULL }, /* SB */
1.7 schwarze 89: { NULL, NULL }, /* BI */
90: { NULL, NULL }, /* IB */
91: { NULL, NULL }, /* BR */
92: { NULL, NULL }, /* RB */
1.8 schwarze 93: { NULL, NULL }, /* R */
94: { NULL, NULL }, /* B */
95: { NULL, NULL }, /* I */
1.7 schwarze 96: { NULL, NULL }, /* IR */
97: { NULL, NULL }, /* RI */
1.28 schwarze 98: { NULL, posts_eq0 }, /* na */ /* FIXME: should warn only. */
1.46 schwarze 99: { NULL, posts_sp }, /* sp */ /* FIXME: should warn only. */
1.50 ! schwarze 100: { NULL, posts_nf }, /* nf */
! 101: { NULL, posts_fi }, /* fi */
1.8 schwarze 102: { NULL, NULL }, /* RE */
103: { NULL, posts_part }, /* RS */
104: { NULL, NULL }, /* DT */
1.34 schwarze 105: { NULL, posts_uc }, /* UC */
1.11 schwarze 106: { NULL, NULL }, /* PD */
1.34 schwarze 107: { NULL, posts_at }, /* AT */
1.28 schwarze 108: { NULL, NULL }, /* in */
1.32 schwarze 109: { NULL, posts_ft }, /* ft */
1.1 kristaps 110: };
111:
112:
113: int
1.27 schwarze 114: man_valid_pre(struct man *m, struct man_node *n)
1.7 schwarze 115: {
116: v_check *cp;
117:
1.38 schwarze 118: switch (n->type) {
119: case (MAN_TEXT):
120: /* FALLTHROUGH */
121: case (MAN_ROOT):
122: /* FALLTHROUGH */
1.42 schwarze 123: case (MAN_EQN):
124: /* FALLTHROUGH */
1.38 schwarze 125: case (MAN_TBL):
1.7 schwarze 126: return(1);
1.38 schwarze 127: default:
128: break;
129: }
1.7 schwarze 130:
131: if (NULL == (cp = man_valids[n->tok].pres))
132: return(1);
133: for ( ; *cp; cp++)
134: if ( ! (*cp)(m, n))
135: return(0);
136: return(1);
137: }
138:
139:
140: int
1.1 kristaps 141: man_valid_post(struct man *m)
142: {
1.7 schwarze 143: v_check *cp;
1.1 kristaps 144:
145: if (MAN_VALID & m->last->flags)
146: return(1);
147: m->last->flags |= MAN_VALID;
148:
149: switch (m->last->type) {
150: case (MAN_TEXT):
1.45 schwarze 151: check_text(m, m->last);
152: return(1);
1.1 kristaps 153: case (MAN_ROOT):
1.4 schwarze 154: return(check_root(m, m->last));
1.42 schwarze 155: case (MAN_EQN):
156: /* FALLTHROUGH */
1.38 schwarze 157: case (MAN_TBL):
158: return(1);
1.1 kristaps 159: default:
160: break;
161: }
162:
163: if (NULL == (cp = man_valids[m->last->tok].posts))
164: return(1);
165: for ( ; *cp; cp++)
166: if ( ! (*cp)(m, m->last))
167: return(0);
168:
169: return(1);
170: }
171:
172:
1.4 schwarze 173: static int
1.7 schwarze 174: check_root(CHKARGS)
1.4 schwarze 175: {
1.7 schwarze 176:
177: if (MAN_BLINE & m->flags)
1.35 schwarze 178: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
179: else if (MAN_ELINE & m->flags)
180: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
1.8 schwarze 181:
182: m->flags &= ~MAN_BLINE;
183: m->flags &= ~MAN_ELINE;
1.7 schwarze 184:
1.25 schwarze 185: if (NULL == m->first->child) {
186: man_nmsg(m, n, MANDOCERR_NODOCBODY);
187: return(0);
188: } else if (NULL == m->meta.title) {
1.35 schwarze 189: man_nmsg(m, n, MANDOCERR_NOTITLE);
190:
1.18 schwarze 191: /*
192: * If a title hasn't been set, do so now (by
193: * implication, date and section also aren't set).
194: */
1.35 schwarze 195:
1.17 schwarze 196: m->meta.title = mandoc_strdup("unknown");
1.21 schwarze 197: m->meta.msec = mandoc_strdup("1");
1.44 schwarze 198: m->meta.date = mandoc_normdate
199: (m->parse, NULL, n->line, n->pos);
1.17 schwarze 200: }
1.15 schwarze 201:
202: return(1);
203: }
204:
1.45 schwarze 205: static void
1.47 schwarze 206: check_text(CHKARGS)
1.4 schwarze 207: {
1.47 schwarze 208: char *cp, *p;
209:
1.48 schwarze 210: if (MAN_LITERAL & m->flags)
211: return;
212:
213: cp = n->string;
214: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.47 schwarze 215: man_pmsg(m, n->line, (int)(p - cp), MANDOCERR_BADTAB);
1.1 kristaps 216: }
217:
218: #define INEQ_DEFINE(x, ineq, name) \
219: static int \
1.7 schwarze 220: check_##name(CHKARGS) \
1.1 kristaps 221: { \
1.4 schwarze 222: if (n->nchild ineq (x)) \
1.1 kristaps 223: return(1); \
1.44 schwarze 224: mandoc_vmsg(MANDOCERR_ARGCOUNT, m->parse, n->line, n->pos, \
1.25 schwarze 225: "line arguments %s %d (have %d)", \
226: #ineq, (x), n->nchild); \
1.40 schwarze 227: return(1); \
1.1 kristaps 228: }
229:
230: INEQ_DEFINE(0, ==, eq0)
1.10 schwarze 231: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 232: INEQ_DEFINE(2, >=, ge2)
233: INEQ_DEFINE(5, <=, le5)
1.32 schwarze 234:
235: static int
1.47 schwarze 236: post_ft(CHKARGS)
1.32 schwarze 237: {
238: char *cp;
239: int ok;
240:
241: if (0 == n->nchild)
242: return(1);
243:
244: ok = 0;
245: cp = n->child->string;
246: switch (*cp) {
247: case ('1'):
248: /* FALLTHROUGH */
249: case ('2'):
250: /* FALLTHROUGH */
251: case ('3'):
252: /* FALLTHROUGH */
253: case ('4'):
254: /* FALLTHROUGH */
255: case ('I'):
256: /* FALLTHROUGH */
257: case ('P'):
258: /* FALLTHROUGH */
259: case ('R'):
260: if ('\0' == cp[1])
261: ok = 1;
262: break;
263: case ('B'):
264: if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
265: ok = 1;
266: break;
267: case ('C'):
268: if ('W' == cp[1] && '\0' == cp[2])
269: ok = 1;
270: break;
271: default:
272: break;
273: }
274:
275: if (0 == ok) {
1.44 schwarze 276: mandoc_vmsg
277: (MANDOCERR_BADFONT, m->parse,
278: n->line, n->pos, "%s", cp);
1.32 schwarze 279: *cp = '\0';
280: }
281:
282: if (1 < n->nchild)
1.44 schwarze 283: mandoc_vmsg
284: (MANDOCERR_ARGCOUNT, m->parse, n->line,
285: n->pos, "want one child (have %d)",
286: n->nchild);
1.32 schwarze 287:
288: return(1);
289: }
1.7 schwarze 290:
291: static int
1.47 schwarze 292: pre_sec(CHKARGS)
293: {
294:
295: if (MAN_BLOCK == n->type)
296: m->flags &= ~MAN_LITERAL;
297: return(1);
298: }
299:
300: static int
301: post_sec(CHKARGS)
1.7 schwarze 302: {
1.6 schwarze 303:
1.44 schwarze 304: if ( ! (MAN_HEAD == n->type && 0 == n->nchild))
305: return(1);
1.25 schwarze 306:
1.44 schwarze 307: man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
308: return(0);
1.6 schwarze 309: }
1.7 schwarze 310:
311: static int
1.8 schwarze 312: check_part(CHKARGS)
313: {
314:
315: if (MAN_BODY == n->type && 0 == n->nchild)
1.44 schwarze 316: mandoc_msg(MANDOCERR_ARGCWARN, m->parse, n->line,
317: n->pos, "want children (have none)");
1.35 schwarze 318:
1.8 schwarze 319: return(1);
320: }
321:
1.36 schwarze 322:
323: static int
324: check_par(CHKARGS)
325: {
326:
1.39 schwarze 327: switch (n->type) {
328: case (MAN_BLOCK):
329: if (0 == n->body->nchild)
330: man_node_delete(m, n);
331: break;
332: case (MAN_BODY):
333: if (0 == n->nchild)
334: man_nmsg(m, n, MANDOCERR_IGNPAR);
335: break;
336: case (MAN_HEAD):
337: if (n->nchild)
338: man_nmsg(m, n, MANDOCERR_ARGSLOST);
339: break;
340: default:
341: break;
342: }
1.36 schwarze 343:
344: return(1);
345: }
346:
347:
1.34 schwarze 348: static int
349: post_TH(CHKARGS)
350: {
1.40 schwarze 351: const char *p;
1.41 schwarze 352: int line, pos;
1.34 schwarze 353:
354: if (m->meta.title)
355: free(m->meta.title);
356: if (m->meta.vol)
357: free(m->meta.vol);
358: if (m->meta.source)
359: free(m->meta.source);
360: if (m->meta.msec)
361: free(m->meta.msec);
1.41 schwarze 362: if (m->meta.date)
363: free(m->meta.date);
1.34 schwarze 364:
1.41 schwarze 365: line = n->line;
366: pos = n->pos;
367: m->meta.title = m->meta.vol = m->meta.date =
1.34 schwarze 368: m->meta.msec = m->meta.source = NULL;
369:
370: /* ->TITLE<- MSEC DATE SOURCE VOL */
371:
372: n = n->child;
1.40 schwarze 373: if (n && n->string) {
374: for (p = n->string; '\0' != *p; p++) {
375: /* Only warn about this once... */
1.47 schwarze 376: if (isalpha((unsigned char)*p) &&
377: ! isupper((unsigned char)*p)) {
1.40 schwarze 378: man_nmsg(m, n, MANDOCERR_UPPERCASE);
379: break;
380: }
381: }
382: m->meta.title = mandoc_strdup(n->string);
383: } else
384: m->meta.title = mandoc_strdup("");
1.34 schwarze 385:
386: /* TITLE ->MSEC<- DATE SOURCE VOL */
387:
1.40 schwarze 388: if (n)
389: n = n->next;
390: if (n && n->string)
391: m->meta.msec = mandoc_strdup(n->string);
392: else
393: m->meta.msec = mandoc_strdup("");
1.34 schwarze 394:
395: /* TITLE MSEC ->DATE<- SOURCE VOL */
396:
1.40 schwarze 397: if (n)
398: n = n->next;
1.49 schwarze 399: if (n && n->string && '\0' != n->string[0]) {
1.41 schwarze 400: pos = n->pos;
1.49 schwarze 401: m->meta.date = mandoc_normdate
402: (m->parse, n->string, line, pos);
403: } else
404: m->meta.date = mandoc_strdup("");
1.34 schwarze 405:
406: /* TITLE MSEC DATE ->SOURCE<- VOL */
407:
408: if (n && (n = n->next))
409: m->meta.source = mandoc_strdup(n->string);
410:
411: /* TITLE MSEC DATE SOURCE ->VOL<- */
412:
413: if (n && (n = n->next))
414: m->meta.vol = mandoc_strdup(n->string);
415:
416: /*
417: * Remove the `TH' node after we've processed it for our
418: * meta-data.
419: */
420: man_node_delete(m, m->last);
421: return(1);
422: }
423:
424: static int
425: post_nf(CHKARGS)
426: {
427:
428: if (MAN_LITERAL & m->flags)
429: man_nmsg(m, n, MANDOCERR_SCOPEREP);
430:
431: m->flags |= MAN_LITERAL;
432: return(1);
433: }
434:
435: static int
436: post_fi(CHKARGS)
437: {
438:
439: if ( ! (MAN_LITERAL & m->flags))
1.39 schwarze 440: man_nmsg(m, n, MANDOCERR_WNOSCOPE);
1.34 schwarze 441:
442: m->flags &= ~MAN_LITERAL;
443: return(1);
444: }
445:
446: static int
447: post_UC(CHKARGS)
448: {
449: static const char * const bsd_versions[] = {
450: "3rd Berkeley Distribution",
451: "4th Berkeley Distribution",
452: "4.2 Berkeley Distribution",
453: "4.3 Berkeley Distribution",
454: "4.4 Berkeley Distribution",
455: };
456:
457: const char *p, *s;
458:
459: n = n->child;
460:
461: if (NULL == n || MAN_TEXT != n->type)
462: p = bsd_versions[0];
463: else {
464: s = n->string;
465: if (0 == strcmp(s, "3"))
466: p = bsd_versions[0];
467: else if (0 == strcmp(s, "4"))
468: p = bsd_versions[1];
469: else if (0 == strcmp(s, "5"))
470: p = bsd_versions[2];
471: else if (0 == strcmp(s, "6"))
472: p = bsd_versions[3];
473: else if (0 == strcmp(s, "7"))
474: p = bsd_versions[4];
475: else
476: p = bsd_versions[0];
477: }
478:
479: if (m->meta.source)
480: free(m->meta.source);
481:
482: m->meta.source = mandoc_strdup(p);
483: return(1);
484: }
485:
486: static int
487: post_AT(CHKARGS)
488: {
489: static const char * const unix_versions[] = {
490: "7th Edition",
491: "System III",
492: "System V",
493: "System V Release 2",
494: };
495:
496: const char *p, *s;
497: struct man_node *nn;
498:
499: n = n->child;
500:
501: if (NULL == n || MAN_TEXT != n->type)
502: p = unix_versions[0];
503: else {
504: s = n->string;
505: if (0 == strcmp(s, "3"))
506: p = unix_versions[0];
507: else if (0 == strcmp(s, "4"))
508: p = unix_versions[1];
509: else if (0 == strcmp(s, "5")) {
510: nn = n->next;
511: if (nn && MAN_TEXT == nn->type && nn->string[0])
512: p = unix_versions[3];
513: else
514: p = unix_versions[2];
515: } else
516: p = unix_versions[0];
517: }
518:
519: if (m->meta.source)
520: free(m->meta.source);
521:
522: m->meta.source = mandoc_strdup(p);
1.46 schwarze 523: return(1);
524: }
525:
526: static int
527: post_vs(CHKARGS)
528: {
529:
530: /*
531: * Don't warn about this because it occurs in pod2man and would
532: * cause considerable (unfixable) warnage.
533: */
534: if (NULL == n->prev && MAN_ROOT == n->parent->type)
535: man_node_delete(m, n);
536:
1.34 schwarze 537: return(1);
538: }