Annotation of mandoc/man_validate.c, Revision 1.53
1.53 ! kristaps 1: /* $Id: man_validate.c,v 1.52 2010/12/05 16:14:16 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)
1.41 kristaps 166: return(man_nmsg(m, n, MANDOCERR_SCOPEEXIT));
1.17 kristaps 167: if (MAN_ELINE & m->flags)
1.41 kristaps 168: return(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) {
177: if ( ! man_nmsg(m, n, MANDOCERR_NOTITLE))
1.34 kristaps 178: return(0);
179: /*
180: * If a title hasn't been set, do so now (by
181: * implication, date and section also aren't set).
182: *
183: * FIXME: this should be in man_action.c.
184: */
185: m->meta.title = mandoc_strdup("unknown");
186: m->meta.date = time(NULL);
1.37 kristaps 187: m->meta.msec = mandoc_strdup("1");
1.34 kristaps 188: }
1.32 kristaps 189:
190: return(1);
191: }
192:
193:
194: static int
195: check_title(CHKARGS)
196: {
197: const char *p;
198:
199: assert(n->child);
1.41 kristaps 200: /* FIXME: is this sufficient? */
201: if ('\0' == *n->child->string) {
202: man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
203: return(0);
204: }
1.32 kristaps 205:
206: for (p = n->child->string; '\0' != *p; p++)
207: if (isalpha((u_char)*p) && ! isupper((u_char)*p))
1.41 kristaps 208: if ( ! man_nmsg(m, n, MANDOCERR_UPPERCASE))
1.32 kristaps 209: return(0);
1.14 kristaps 210:
211: return(1);
212: }
213:
214:
215: static int
1.17 kristaps 216: check_text(CHKARGS)
1.11 kristaps 217: {
1.43 kristaps 218: char *p;
1.15 kristaps 219: int pos, c;
1.46 kristaps 220: size_t sz;
1.11 kristaps 221:
1.46 kristaps 222: for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
223: sz = strcspn(p, "\t\\");
224: p += (int)sz;
225:
226: if ('\0' == *p)
227: break;
228:
229: pos += (int)sz;
1.11 kristaps 230:
1.46 kristaps 231: if ('\t' == *p) {
232: if (MAN_LITERAL & m->flags)
233: continue;
234: if (man_pmsg(m, n->line, pos, MANDOCERR_BADTAB))
1.15 kristaps 235: continue;
1.46 kristaps 236: return(0);
1.15 kristaps 237: }
1.45 kristaps 238:
1.46 kristaps 239: /* Check the special character. */
1.15 kristaps 240:
1.46 kristaps 241: c = mandoc_special(p);
242: if (c) {
243: p += c - 1;
244: pos += c - 1;
1.49 schwarze 245: } else
246: man_pmsg(m, n->line, pos, MANDOCERR_BADESCAPE);
1.11 kristaps 247: }
248:
249: return(1);
1.1 kristaps 250: }
251:
252:
253: #define INEQ_DEFINE(x, ineq, name) \
254: static int \
1.17 kristaps 255: check_##name(CHKARGS) \
1.1 kristaps 256: { \
1.11 kristaps 257: if (n->nchild ineq (x)) \
1.1 kristaps 258: return(1); \
1.41 kristaps 259: man_vmsg(m, MANDOCERR_SYNTARGCOUNT, n->line, n->pos, \
260: "line arguments %s %d (have %d)", \
261: #ineq, (x), n->nchild); \
262: return(0); \
1.1 kristaps 263: }
264:
265: INEQ_DEFINE(0, ==, eq0)
1.25 kristaps 266: INEQ_DEFINE(1, <=, le1)
1.1 kristaps 267: INEQ_DEFINE(2, >=, ge2)
268: INEQ_DEFINE(5, <=, le5)
269:
1.16 kristaps 270:
271: static int
1.17 kristaps 272: check_sec(CHKARGS)
273: {
1.16 kristaps 274:
1.41 kristaps 275: if (MAN_HEAD == n->type && 0 == n->nchild) {
276: man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
277: return(0);
278: } else if (MAN_BODY == n->type && 0 == n->nchild)
279: return(man_nmsg(m, n, MANDOCERR_NOBODY));
280:
1.16 kristaps 281: return(1);
282: }
1.17 kristaps 283:
284:
285: static int
1.23 kristaps 286: check_part(CHKARGS)
287: {
288:
289: if (MAN_BODY == n->type && 0 == n->nchild)
1.41 kristaps 290: return(man_nmsg(m, n, MANDOCERR_NOBODY));
1.23 kristaps 291: return(1);
292: }
293:
294:
295: static int
1.17 kristaps 296: check_par(CHKARGS)
297: {
298:
299: if (MAN_BODY == n->type)
300: switch (n->tok) {
301: case (MAN_IP):
302: /* FALLTHROUGH */
303: case (MAN_HP):
304: /* FALLTHROUGH */
305: case (MAN_TP):
306: /* Body-less lists are ok. */
307: break;
308: default:
309: if (n->nchild)
310: break;
1.41 kristaps 311: return(man_nmsg(m, n, MANDOCERR_NOBODY));
1.17 kristaps 312: }
313: if (MAN_HEAD == n->type)
314: switch (n->tok) {
315: case (MAN_PP):
316: /* FALLTHROUGH */
317: case (MAN_P):
318: /* FALLTHROUGH */
319: case (MAN_LP):
320: if (0 == n->nchild)
321: break;
1.41 kristaps 322: return(man_nmsg(m, n, MANDOCERR_ARGSLOST));
1.17 kristaps 323: default:
1.53 ! kristaps 324: break;
1.17 kristaps 325: }
326:
327: return(1);
328: }
329:
330:
331: static int
332: check_bline(CHKARGS)
333: {
334:
1.22 kristaps 335: assert( ! (MAN_ELINE & m->flags));
1.41 kristaps 336: if (MAN_BLINE & m->flags) {
337: man_nmsg(m, n, MANDOCERR_SYNTLINESCOPE);
338: return(0);
339: }
1.31 kristaps 340:
1.18 kristaps 341: return(1);
1.17 kristaps 342: }
343:
1.51 kristaps 344: static int
345: post_TH(CHKARGS)
346: {
347:
348: if (m->meta.title)
349: free(m->meta.title);
350: if (m->meta.vol)
351: free(m->meta.vol);
352: if (m->meta.source)
353: free(m->meta.source);
354: if (m->meta.msec)
355: free(m->meta.msec);
356: if (m->meta.rawdate)
357: free(m->meta.rawdate);
358:
359: m->meta.title = m->meta.vol = m->meta.rawdate =
360: m->meta.msec = m->meta.source = NULL;
361: m->meta.date = 0;
362:
363: /* ->TITLE<- MSEC DATE SOURCE VOL */
364:
365: n = n->child;
366: assert(n);
367: m->meta.title = mandoc_strdup(n->string);
368:
369: /* TITLE ->MSEC<- DATE SOURCE VOL */
370:
371: n = n->next;
372: assert(n);
373: m->meta.msec = mandoc_strdup(n->string);
374:
375: /* TITLE MSEC ->DATE<- SOURCE VOL */
376:
377: /*
378: * Try to parse the date. If this works, stash the epoch (this
379: * is optimal because we can reformat it in the canonical form).
380: * If it doesn't parse, isn't specified at all, or is an empty
381: * string, then use the current date.
382: */
383:
384: n = n->next;
385: if (n && n->string && *n->string) {
386: m->meta.date = mandoc_a2time
387: (MTIME_ISO_8601, n->string);
388: if (0 == m->meta.date) {
389: man_nmsg(m, n, MANDOCERR_BADDATE);
390: m->meta.rawdate = mandoc_strdup(n->string);
391: }
392: } else
393: m->meta.date = time(NULL);
394:
395: /* TITLE MSEC DATE ->SOURCE<- VOL */
396:
397: if (n && (n = n->next))
398: m->meta.source = mandoc_strdup(n->string);
399:
400: /* TITLE MSEC DATE SOURCE ->VOL<- */
401:
402: if (n && (n = n->next))
403: m->meta.vol = mandoc_strdup(n->string);
404:
405: /*
406: * Remove the `TH' node after we've processed it for our
407: * meta-data.
408: */
409: man_node_delete(m, m->last);
410: return(1);
411: }
412:
413: static int
414: post_nf(CHKARGS)
415: {
416:
417: if (MAN_LITERAL & m->flags)
418: man_nmsg(m, n, MANDOCERR_SCOPEREP);
419:
420: m->flags |= MAN_LITERAL;
421: return(1);
422: }
423:
424: static int
425: post_fi(CHKARGS)
426: {
427:
428: if ( ! (MAN_LITERAL & m->flags))
429: man_nmsg(m, n, MANDOCERR_NOSCOPE);
430:
431: m->flags &= ~MAN_LITERAL;
432: return(1);
433: }
434:
435: static int
436: post_UC(CHKARGS)
437: {
438: static const char * const bsd_versions[] = {
439: "3rd Berkeley Distribution",
440: "4th Berkeley Distribution",
441: "4.2 Berkeley Distribution",
442: "4.3 Berkeley Distribution",
443: "4.4 Berkeley Distribution",
444: };
445:
446: const char *p, *s;
447:
448: n = n->child;
449: n = m->last->child;
450:
451: if (NULL == n || MAN_TEXT != n->type)
452: p = bsd_versions[0];
453: else {
454: s = n->string;
455: if (0 == strcmp(s, "3"))
456: p = bsd_versions[0];
457: else if (0 == strcmp(s, "4"))
458: p = bsd_versions[1];
459: else if (0 == strcmp(s, "5"))
460: p = bsd_versions[2];
461: else if (0 == strcmp(s, "6"))
462: p = bsd_versions[3];
463: else if (0 == strcmp(s, "7"))
464: p = bsd_versions[4];
465: else
466: p = bsd_versions[0];
467: }
468:
469: if (m->meta.source)
470: free(m->meta.source);
471:
472: m->meta.source = mandoc_strdup(p);
473: return(1);
474: }
475:
476: static int
477: post_AT(CHKARGS)
478: {
479: static const char * const unix_versions[] = {
480: "7th Edition",
481: "System III",
482: "System V",
483: "System V Release 2",
484: };
485:
486: const char *p, *s;
487: struct man_node *nn;
488:
489: n = n->child;
490:
491: if (NULL == n || MAN_TEXT != n->type)
492: p = unix_versions[0];
493: else {
494: s = n->string;
495: if (0 == strcmp(s, "3"))
496: p = unix_versions[0];
497: else if (0 == strcmp(s, "4"))
498: p = unix_versions[1];
499: else if (0 == strcmp(s, "5")) {
500: nn = n->next;
501: if (nn && MAN_TEXT == nn->type && nn->string[0])
502: p = unix_versions[3];
503: else
504: p = unix_versions[2];
505: } else
506: p = unix_versions[0];
507: }
508:
509: if (m->meta.source)
510: free(m->meta.source);
511:
512: m->meta.source = mandoc_strdup(p);
513: return(1);
514: }
CVSweb