Annotation of mandoc/man_validate.c, Revision 1.54
1.54 ! kristaps 1: /* $Id: man_validate.c,v 1.53 2010/12/06 13:50:41 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);
1.25 kristaps 47: static int check_le1(CHKARGS);
1.17 kristaps 48: static int check_ge2(CHKARGS);
49: static int check_le5(CHKARGS);
50: static int check_par(CHKARGS);
1.23 kristaps 51: static int check_part(CHKARGS);
1.17 kristaps 52: static int check_root(CHKARGS);
53: static int check_sec(CHKARGS);
54: static int check_text(CHKARGS);
1.32 kristaps 55: static int check_title(CHKARGS);
1.17 kristaps 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 };
67: static v_check posts_nf[] = { check_eq0, post_nf, NULL };
1.17 kristaps 68: static v_check posts_par[] = { check_par, NULL };
1.23 kristaps 69: static v_check posts_part[] = { check_part, NULL };
1.17 kristaps 70: static v_check posts_sec[] = { check_sec, NULL };
1.51 kristaps 71: static v_check posts_th[] = { check_ge2, check_le5, check_title, post_TH, NULL };
72: static v_check posts_uc[] = { post_UC, NULL };
1.17 kristaps 73: static v_check pres_bline[] = { check_bline, NULL };
1.1 kristaps 74:
1.51 kristaps 75:
1.1 kristaps 76: static const struct man_valid man_valids[MAN_MAX] = {
1.29 kristaps 77: { NULL, posts_eq0 }, /* br */
1.32 kristaps 78: { pres_bline, posts_th }, /* TH */
1.17 kristaps 79: { pres_bline, posts_sec }, /* SH */
80: { pres_bline, posts_sec }, /* SS */
81: { pres_bline, posts_par }, /* TP */
82: { pres_bline, posts_par }, /* LP */
83: { pres_bline, posts_par }, /* PP */
84: { pres_bline, posts_par }, /* P */
85: { pres_bline, posts_par }, /* IP */
86: { pres_bline, posts_par }, /* HP */
1.22 kristaps 87: { NULL, NULL }, /* SM */
88: { NULL, NULL }, /* SB */
1.17 kristaps 89: { NULL, NULL }, /* BI */
90: { NULL, NULL }, /* IB */
91: { NULL, NULL }, /* BR */
92: { NULL, NULL }, /* RB */
1.22 kristaps 93: { NULL, NULL }, /* R */
94: { NULL, NULL }, /* B */
95: { NULL, NULL }, /* I */
1.17 kristaps 96: { NULL, NULL }, /* IR */
97: { NULL, NULL }, /* RI */
1.47 kristaps 98: { NULL, posts_eq0 }, /* na */ /* FIXME: should warn only. */
1.17 kristaps 99: { NULL, NULL }, /* i */
1.47 kristaps 100: { NULL, posts_le1 }, /* sp */ /* FIXME: should warn only. */
1.51 kristaps 101: { pres_bline, posts_nf }, /* nf */
102: { pres_bline, posts_fi }, /* fi */
1.17 kristaps 103: { NULL, NULL }, /* r */
1.19 kristaps 104: { NULL, NULL }, /* RE */
1.23 kristaps 105: { NULL, posts_part }, /* RS */
1.21 kristaps 106: { NULL, NULL }, /* DT */
1.51 kristaps 107: { NULL, posts_uc }, /* UC */
1.26 kristaps 108: { NULL, NULL }, /* PD */
1.51 kristaps 109: { NULL, posts_at }, /* AT */
1.47 kristaps 110: { NULL, NULL }, /* in */
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:
119: if (MAN_TEXT == n->type)
120: return(1);
121: if (MAN_ROOT == n->type)
122: return(1);
123:
124: if (NULL == (cp = man_valids[n->tok].pres))
125: return(1);
126: for ( ; *cp; cp++)
127: if ( ! (*cp)(m, n))
128: return(0);
129: return(1);
130: }
131:
132:
133: int
1.1 kristaps 134: man_valid_post(struct man *m)
135: {
1.17 kristaps 136: v_check *cp;
1.1 kristaps 137:
138: if (MAN_VALID & m->last->flags)
139: return(1);
140: m->last->flags |= MAN_VALID;
141:
142: switch (m->last->type) {
143: case (MAN_TEXT):
1.11 kristaps 144: return(check_text(m, m->last));
1.1 kristaps 145: case (MAN_ROOT):
1.14 kristaps 146: return(check_root(m, m->last));
1.1 kristaps 147: default:
148: break;
149: }
150:
151: if (NULL == (cp = man_valids[m->last->tok].posts))
152: return(1);
153: for ( ; *cp; cp++)
154: if ( ! (*cp)(m, m->last))
155: return(0);
156:
157: return(1);
158: }
159:
160:
1.11 kristaps 161: static int
1.17 kristaps 162: check_root(CHKARGS)
1.14 kristaps 163: {
1.17 kristaps 164:
165: if (MAN_BLINE & m->flags)
1.54 ! kristaps 166: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
! 167: else if (MAN_ELINE & m->flags)
! 168: man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
1.22 kristaps 169:
170: m->flags &= ~MAN_BLINE;
171: m->flags &= ~MAN_ELINE;
1.17 kristaps 172:
1.41 kristaps 173: if (NULL == m->first->child) {
174: man_nmsg(m, n, MANDOCERR_NODOCBODY);
175: return(0);
176: } else if (NULL == m->meta.title) {
1.54 ! kristaps 177: man_nmsg(m, n, MANDOCERR_NOTITLE);
! 178:
1.34 kristaps 179: /*
180: * If a title hasn't been set, do so now (by
181: * implication, date and section also aren't set).
182: */
1.54 ! kristaps 183:
1.34 kristaps 184: m->meta.title = mandoc_strdup("unknown");
185: m->meta.date = time(NULL);
1.37 kristaps 186: m->meta.msec = mandoc_strdup("1");
1.34 kristaps 187: }
1.32 kristaps 188:
189: return(1);
190: }
191:
192:
193: static int
194: check_title(CHKARGS)
195: {
196: const char *p;
197:
198: assert(n->child);
1.41 kristaps 199: /* FIXME: is this sufficient? */
200: if ('\0' == *n->child->string) {
201: man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
202: return(0);
203: }
1.32 kristaps 204:
205: for (p = n->child->string; '\0' != *p; p++)
1.54 ! kristaps 206: /* Only warn about this once... */
! 207: if (isalpha((u_char)*p) && ! isupper((u_char)*p)) {
! 208: man_nmsg(m, n, MANDOCERR_UPPERCASE);
! 209: break;
! 210: }
1.14 kristaps 211:
212: return(1);
213: }
214:
215:
216: static int
1.17 kristaps 217: check_text(CHKARGS)
1.11 kristaps 218: {
1.43 kristaps 219: char *p;
1.15 kristaps 220: int pos, c;
1.46 kristaps 221: size_t sz;
1.11 kristaps 222:
1.46 kristaps 223: for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
224: sz = strcspn(p, "\t\\");
225: p += (int)sz;
226:
227: if ('\0' == *p)
228: break;
229:
230: pos += (int)sz;
1.11 kristaps 231:
1.46 kristaps 232: if ('\t' == *p) {
233: if (MAN_LITERAL & m->flags)
234: continue;
235: if (man_pmsg(m, n->line, pos, MANDOCERR_BADTAB))
1.15 kristaps 236: continue;
1.46 kristaps 237: return(0);
1.15 kristaps 238: }
1.45 kristaps 239:
1.46 kristaps 240: /* Check the special character. */
1.15 kristaps 241:
1.46 kristaps 242: c = mandoc_special(p);
243: if (c) {
244: p += c - 1;
245: pos += c - 1;
1.49 schwarze 246: } else
247: man_pmsg(m, n->line, pos, MANDOCERR_BADESCAPE);
1.11 kristaps 248: }
249:
250: return(1);
1.1 kristaps 251: }
252:
253:
254: #define INEQ_DEFINE(x, ineq, name) \
255: static int \
1.17 kristaps 256: check_##name(CHKARGS) \
1.1 kristaps 257: { \
1.11 kristaps 258: if (n->nchild ineq (x)) \
1.1 kristaps 259: return(1); \
1.41 kristaps 260: man_vmsg(m, MANDOCERR_SYNTARGCOUNT, n->line, n->pos, \
261: "line arguments %s %d (have %d)", \
262: #ineq, (x), n->nchild); \
263: return(0); \
1.1 kristaps 264: }
265:
266: INEQ_DEFINE(0, ==, eq0)
1.25 kristaps 267: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 268: INEQ_DEFINE(2, >=, ge2)
269: INEQ_DEFINE(5, <=, le5)
270:
1.16 kristaps 271:
272: static int
1.17 kristaps 273: check_sec(CHKARGS)
274: {
1.16 kristaps 275:
1.41 kristaps 276: if (MAN_HEAD == n->type && 0 == n->nchild) {
277: man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
278: return(0);
279: } else if (MAN_BODY == n->type && 0 == n->nchild)
1.54 ! kristaps 280: man_nmsg(m, n, MANDOCERR_NOBODY);
1.41 kristaps 281:
1.16 kristaps 282: return(1);
283: }
1.17 kristaps 284:
285:
286: static int
1.23 kristaps 287: check_part(CHKARGS)
288: {
289:
290: if (MAN_BODY == n->type && 0 == n->nchild)
1.54 ! kristaps 291: man_nmsg(m, n, MANDOCERR_NOBODY);
! 292:
1.23 kristaps 293: return(1);
294: }
295:
296:
297: static int
1.17 kristaps 298: check_par(CHKARGS)
299: {
300:
301: if (MAN_BODY == n->type)
302: switch (n->tok) {
303: case (MAN_IP):
304: /* FALLTHROUGH */
305: case (MAN_HP):
306: /* FALLTHROUGH */
307: case (MAN_TP):
308: /* Body-less lists are ok. */
309: break;
310: default:
1.54 ! kristaps 311: if (0 == n->nchild)
! 312: man_nmsg(m, n, MANDOCERR_NOBODY);
! 313: break;
1.17 kristaps 314: }
315: if (MAN_HEAD == n->type)
316: switch (n->tok) {
317: case (MAN_PP):
318: /* FALLTHROUGH */
319: case (MAN_P):
320: /* FALLTHROUGH */
321: case (MAN_LP):
1.54 ! kristaps 322: if (n->nchild)
! 323: man_nmsg(m, n, MANDOCERR_ARGSLOST);
! 324: break;
1.17 kristaps 325: default:
1.53 kristaps 326: break;
1.17 kristaps 327: }
328:
329: return(1);
330: }
331:
332:
333: static int
334: check_bline(CHKARGS)
335: {
336:
1.22 kristaps 337: assert( ! (MAN_ELINE & m->flags));
1.41 kristaps 338: if (MAN_BLINE & m->flags) {
339: man_nmsg(m, n, MANDOCERR_SYNTLINESCOPE);
340: return(0);
341: }
1.31 kristaps 342:
1.18 kristaps 343: return(1);
1.17 kristaps 344: }
345:
1.51 kristaps 346: static int
347: post_TH(CHKARGS)
348: {
349:
350: if (m->meta.title)
351: free(m->meta.title);
352: if (m->meta.vol)
353: free(m->meta.vol);
354: if (m->meta.source)
355: free(m->meta.source);
356: if (m->meta.msec)
357: free(m->meta.msec);
358: if (m->meta.rawdate)
359: free(m->meta.rawdate);
360:
361: m->meta.title = m->meta.vol = m->meta.rawdate =
362: m->meta.msec = m->meta.source = NULL;
363: m->meta.date = 0;
364:
365: /* ->TITLE<- MSEC DATE SOURCE VOL */
366:
367: n = n->child;
368: assert(n);
369: m->meta.title = mandoc_strdup(n->string);
370:
371: /* TITLE ->MSEC<- DATE SOURCE VOL */
372:
373: n = n->next;
374: assert(n);
375: m->meta.msec = mandoc_strdup(n->string);
376:
377: /* TITLE MSEC ->DATE<- SOURCE VOL */
378:
379: /*
380: * Try to parse the date. If this works, stash the epoch (this
381: * is optimal because we can reformat it in the canonical form).
382: * If it doesn't parse, isn't specified at all, or is an empty
383: * string, then use the current date.
384: */
385:
386: n = n->next;
387: if (n && n->string && *n->string) {
388: m->meta.date = mandoc_a2time
389: (MTIME_ISO_8601, n->string);
390: if (0 == m->meta.date) {
391: man_nmsg(m, n, MANDOCERR_BADDATE);
392: m->meta.rawdate = mandoc_strdup(n->string);
393: }
394: } else
395: m->meta.date = time(NULL);
396:
397: /* TITLE MSEC DATE ->SOURCE<- VOL */
398:
399: if (n && (n = n->next))
400: m->meta.source = mandoc_strdup(n->string);
401:
402: /* TITLE MSEC DATE SOURCE ->VOL<- */
403:
404: if (n && (n = n->next))
405: m->meta.vol = mandoc_strdup(n->string);
406:
407: /*
408: * Remove the `TH' node after we've processed it for our
409: * meta-data.
410: */
411: man_node_delete(m, m->last);
412: return(1);
413: }
414:
415: static int
416: post_nf(CHKARGS)
417: {
418:
419: if (MAN_LITERAL & m->flags)
420: man_nmsg(m, n, MANDOCERR_SCOPEREP);
421:
422: m->flags |= MAN_LITERAL;
423: return(1);
424: }
425:
426: static int
427: post_fi(CHKARGS)
428: {
429:
430: if ( ! (MAN_LITERAL & m->flags))
431: man_nmsg(m, n, MANDOCERR_NOSCOPE);
432:
433: m->flags &= ~MAN_LITERAL;
434: return(1);
435: }
436:
437: static int
438: post_UC(CHKARGS)
439: {
440: static const char * const bsd_versions[] = {
441: "3rd Berkeley Distribution",
442: "4th Berkeley Distribution",
443: "4.2 Berkeley Distribution",
444: "4.3 Berkeley Distribution",
445: "4.4 Berkeley Distribution",
446: };
447:
448: const char *p, *s;
449:
450: n = n->child;
451: n = m->last->child;
452:
453: if (NULL == n || MAN_TEXT != n->type)
454: p = bsd_versions[0];
455: else {
456: s = n->string;
457: if (0 == strcmp(s, "3"))
458: p = bsd_versions[0];
459: else if (0 == strcmp(s, "4"))
460: p = bsd_versions[1];
461: else if (0 == strcmp(s, "5"))
462: p = bsd_versions[2];
463: else if (0 == strcmp(s, "6"))
464: p = bsd_versions[3];
465: else if (0 == strcmp(s, "7"))
466: p = bsd_versions[4];
467: else
468: p = bsd_versions[0];
469: }
470:
471: if (m->meta.source)
472: free(m->meta.source);
473:
474: m->meta.source = mandoc_strdup(p);
475: return(1);
476: }
477:
478: static int
479: post_AT(CHKARGS)
480: {
481: static const char * const unix_versions[] = {
482: "7th Edition",
483: "System III",
484: "System V",
485: "System V Release 2",
486: };
487:
488: const char *p, *s;
489: struct man_node *nn;
490:
491: n = n->child;
492:
493: if (NULL == n || MAN_TEXT != n->type)
494: p = unix_versions[0];
495: else {
496: s = n->string;
497: if (0 == strcmp(s, "3"))
498: p = unix_versions[0];
499: else if (0 == strcmp(s, "4"))
500: p = unix_versions[1];
501: else if (0 == strcmp(s, "5")) {
502: nn = n->next;
503: if (nn && MAN_TEXT == nn->type && nn->string[0])
504: p = unix_versions[3];
505: else
506: p = unix_versions[2];
507: } else
508: p = unix_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: }
CVSweb