man_validate.c

   1 /*      $Id: man_validate.c,v 1.57 2011/01/01 12:59:17 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
   4  *
   5  * Permission to use, copy, modify, and distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   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.
  16  */
  17 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/types.h>
  22
  23 #include <assert.h>
  24 #include <ctype.h>
  25 #include <errno.h>
  26 #include <limits.h>
  27 #include <stdarg.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <time.h>
  31
  32 #include "mandoc.h"
  33 #include "libman.h"
  34 #include "libmandoc.h"
  35
  36 #define CHKARGS   struct man *m, struct man_node *n
  37
  38 typedef int     (*v_check)(CHKARGS);
  39
  40 struct  man_valid {
  41         v_check  *pres;
  42         v_check  *posts;
  43 };
  44
  45 static  int       check_bline(CHKARGS);
  46 static  int       check_eq0(CHKARGS);
  47 static  int       check_ft(CHKARGS);
  48 static  int       check_le1(CHKARGS);
  49 static  int       check_ge2(CHKARGS);
  50 static  int       check_le5(CHKARGS);
  51 static  int       check_par(CHKARGS);
  52 static  int       check_part(CHKARGS);
  53 static  int       check_root(CHKARGS);
  54 static  int       check_sec(CHKARGS);
  55 static  int       check_text(CHKARGS);
  56 static  int       check_title(CHKARGS);
  57
  58 static  int       post_AT(CHKARGS);
  59 static  int       post_fi(CHKARGS);
  60 static  int       post_nf(CHKARGS);
  61 static  int       post_TH(CHKARGS);
  62 static  int       post_UC(CHKARGS);
  63
  64 static  v_check   posts_at[] = { post_AT, NULL };
  65 static  v_check   posts_eq0[] = { check_eq0, NULL };
  66 static  v_check   posts_fi[] = { check_eq0, post_fi, NULL };
  67 static  v_check   posts_le1[] = { check_le1, NULL };
  68 static  v_check   posts_ft[] = { check_ft, NULL };
  69 static  v_check   posts_nf[] = { check_eq0, post_nf, NULL };
  70 static  v_check   posts_par[] = { check_par, NULL };
  71 static  v_check   posts_part[] = { check_part, NULL };
  72 static  v_check   posts_sec[] = { check_sec, NULL };
  73 static  v_check   posts_th[] = { check_ge2, check_le5, check_title, post_TH, NULL };
  74 static  v_check   posts_uc[] = { post_UC, NULL };
  75 static  v_check   pres_bline[] = { check_bline, NULL };
  76
  77
  78 static  const struct man_valid man_valids[MAN_MAX] = {
  79         { NULL, posts_eq0 }, /* br */
  80         { pres_bline, posts_th }, /* TH */
  81         { pres_bline, posts_sec }, /* SH */
  82         { pres_bline, posts_sec }, /* SS */
  83         { pres_bline, posts_par }, /* TP */
  84         { pres_bline, posts_par }, /* LP */
  85         { pres_bline, posts_par }, /* PP */
  86         { pres_bline, posts_par }, /* P */
  87         { pres_bline, posts_par }, /* IP */
  88         { pres_bline, posts_par }, /* HP */
  89         { NULL, NULL }, /* SM */
  90         { NULL, NULL }, /* SB */
  91         { NULL, NULL }, /* BI */
  92         { NULL, NULL }, /* IB */
  93         { NULL, NULL }, /* BR */
  94         { NULL, NULL }, /* RB */
  95         { NULL, NULL }, /* R */
  96         { NULL, NULL }, /* B */
  97         { NULL, NULL }, /* I */
  98         { NULL, NULL }, /* IR */
  99         { NULL, NULL }, /* RI */
 100         { NULL, posts_eq0 }, /* na */ /* FIXME: should warn only. */
 101         { NULL, posts_le1 }, /* sp */ /* FIXME: should warn only. */
 102         { pres_bline, posts_nf }, /* nf */
 103         { pres_bline, posts_fi }, /* fi */
 104         { NULL, NULL }, /* RE */
 105         { NULL, posts_part }, /* RS */
 106         { NULL, NULL }, /* DT */
 107         { NULL, posts_uc }, /* UC */
 108         { NULL, NULL }, /* PD */
 109         { NULL, posts_at }, /* AT */
 110         { NULL, NULL }, /* in */
 111         { NULL, posts_ft }, /* ft */
 112 };
 113
 114
 115 int
 116 man_valid_pre(struct man *m, struct man_node *n)
 117 {
 118         v_check         *cp;
 119
 120         switch (n->type) {
 121         case (MAN_TEXT):
 122                 /* FALLTHROUGH */
 123         case (MAN_ROOT):
 124                 /* FALLTHROUGH */
 125         case (MAN_TBL):
 126                 return(1);
 127         default:
 128                 break;
 129         }
 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
 141 man_valid_post(struct man *m)
 142 {
 143         v_check         *cp;
 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):
 151                 return(check_text(m, m->last));
 152         case (MAN_ROOT):
 153                 return(check_root(m, m->last));
 154         case (MAN_TBL):
 155                 return(1);
 156         default:
 157                 break;
 158         }
 159
 160         if (NULL == (cp = man_valids[m->last->tok].posts))
 161                 return(1);
 162         for ( ; *cp; cp++)
 163                 if ( ! (*cp)(m, m->last))
 164                         return(0);
 165
 166         return(1);
 167 }
 168
 169
 170 static int
 171 check_root(CHKARGS)
 172 {
 173
 174         if (MAN_BLINE & m->flags)
 175                 man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
 176         else if (MAN_ELINE & m->flags)
 177                 man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
 178
 179         m->flags &= ~MAN_BLINE;
 180         m->flags &= ~MAN_ELINE;
 181
 182         if (NULL == m->first->child) {
 183                 man_nmsg(m, n, MANDOCERR_NODOCBODY);
 184                 return(0);
 185         } else if (NULL == m->meta.title) {
 186                 man_nmsg(m, n, MANDOCERR_NOTITLE);
 187
 188                 /*
 189                  * If a title hasn't been set, do so now (by
 190                  * implication, date and section also aren't set).
 191                  */
 192
 193                 m->meta.title = mandoc_strdup("unknown");
 194                 m->meta.date = time(NULL);
 195                 m->meta.msec = mandoc_strdup("1");
 196         }
 197
 198         return(1);
 199 }
 200
 201
 202 static int
 203 check_title(CHKARGS)
 204 {
 205         const char      *p;
 206
 207         assert(n->child);
 208         /* FIXME: is this sufficient? */
 209         if ('\0' == *n->child->string) {
 210                 man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
 211                 return(0);
 212         }
 213
 214         for (p = n->child->string; '\0' != *p; p++)
 215                 /* Only warn about this once... */
 216                 if (isalpha((u_char)*p) && ! isupper((u_char)*p)) {
 217                         man_nmsg(m, n, MANDOCERR_UPPERCASE);
 218                         break;
 219                 }
 220
 221         return(1);
 222 }
 223
 224
 225 static int
 226 check_text(CHKARGS)
 227 {
 228         char            *p;
 229         int              pos, c;
 230         size_t           sz;
 231
 232         for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
 233                 sz = strcspn(p, "\t\\");
 234                 p += (int)sz;
 235
 236                 if ('\0' == *p)
 237                         break;
 238
 239                 pos += (int)sz;
 240
 241                 if ('\t' == *p) {
 242                         if (MAN_LITERAL & m->flags)
 243                                 continue;
 244                         if (man_pmsg(m, n->line, pos, MANDOCERR_BADTAB))
 245                                 continue;
 246                         return(0);
 247                 }
 248
 249                 /* Check the special character. */
 250
 251                 c = mandoc_special(p);
 252                 if (c) {
 253                         p += c - 1;
 254                         pos += c - 1;
 255                 } else
 256                         man_pmsg(m, n->line, pos, MANDOCERR_BADESCAPE);
 257         }
 258
 259         return(1);
 260 }
 261
 262
 263 #define INEQ_DEFINE(x, ineq, name) \
 264 static int \
 265 check_##name(CHKARGS) \
 266 { \
 267         if (n->nchild ineq (x)) \
 268                 return(1); \
 269         man_vmsg(m, MANDOCERR_SYNTARGCOUNT, n->line, n->pos, \
 270                         "line arguments %s %d (have %d)", \
 271                         #ineq, (x), n->nchild); \
 272         return(0); \
 273 }
 274
 275 INEQ_DEFINE(0, ==, eq0)
 276 INEQ_DEFINE(1, <=, le1)
 277 INEQ_DEFINE(2, >=, ge2)
 278 INEQ_DEFINE(5, <=, le5)
 279
 280 static int
 281 check_ft(CHKARGS)
 282 {
 283         char    *cp;
 284         int      ok;
 285
 286         if (0 == n->nchild)
 287                 return(1);
 288
 289         ok = 0;
 290         cp = n->child->string;
 291         switch (*cp) {
 292         case ('1'):
 293                 /* FALLTHROUGH */
 294         case ('2'):
 295                 /* FALLTHROUGH */
 296         case ('3'):
 297                 /* FALLTHROUGH */
 298         case ('4'):
 299                 /* FALLTHROUGH */
 300         case ('I'):
 301                 /* FALLTHROUGH */
 302         case ('P'):
 303                 /* FALLTHROUGH */
 304         case ('R'):
 305                 if ('\0' == cp[1])
 306                         ok = 1;
 307                 break;
 308         case ('B'):
 309                 if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
 310                         ok = 1;
 311                 break;
 312         case ('C'):
 313                 if ('W' == cp[1] && '\0' == cp[2])
 314                         ok = 1;
 315                 break;
 316         default:
 317                 break;
 318         }
 319
 320         if (0 == ok) {
 321                 man_vmsg(m, MANDOCERR_BADFONT,
 322                                 n->line, n->pos, "%s", cp);
 323                 *cp = '\0';
 324         }
 325
 326         if (1 < n->nchild)
 327                 man_vmsg(m, MANDOCERR_ARGCOUNT, n->line, n->pos,
 328                                 "want one child (have %d)", n->nchild);
 329
 330         return(1);
 331 }
 332
 333 static int
 334 check_sec(CHKARGS)
 335 {
 336
 337         if (MAN_HEAD == n->type && 0 == n->nchild) {
 338                 man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
 339                 return(0);
 340         } else if (MAN_BODY == n->type && 0 == n->nchild)
 341                 man_nmsg(m, n, MANDOCERR_NOBODY);
 342
 343         return(1);
 344 }
 345
 346
 347 static int
 348 check_part(CHKARGS)
 349 {
 350
 351         if (MAN_BODY == n->type && 0 == n->nchild)
 352                 man_nmsg(m, n, MANDOCERR_NOBODY);
 353
 354         return(1);
 355 }
 356
 357
 358 static int
 359 check_par(CHKARGS)
 360 {
 361
 362         if (MAN_BODY == n->type)
 363                 switch (n->tok) {
 364                 case (MAN_IP):
 365                         /* FALLTHROUGH */
 366                 case (MAN_HP):
 367                         /* FALLTHROUGH */
 368                 case (MAN_TP):
 369                         /* Body-less lists are ok. */
 370                         break;
 371                 default:
 372                         if (0 == n->nchild)
 373                                 man_nmsg(m, n, MANDOCERR_NOBODY);
 374                         break;
 375                 }
 376         if (MAN_HEAD == n->type)
 377                 switch (n->tok) {
 378                 case (MAN_PP):
 379                         /* FALLTHROUGH */
 380                 case (MAN_P):
 381                         /* FALLTHROUGH */
 382                 case (MAN_LP):
 383                         if (n->nchild)
 384                                 man_nmsg(m, n, MANDOCERR_ARGSLOST);
 385                         break;
 386                 default:
 387                         break;
 388                 }
 389
 390         return(1);
 391 }
 392
 393
 394 static int
 395 check_bline(CHKARGS)
 396 {
 397
 398         assert( ! (MAN_ELINE & m->flags));
 399         if (MAN_BLINE & m->flags) {
 400                 man_nmsg(m, n, MANDOCERR_SYNTLINESCOPE);
 401                 return(0);
 402         }
 403
 404         return(1);
 405 }
 406
 407 static int
 408 post_TH(CHKARGS)
 409 {
 410
 411         if (m->meta.title)
 412                 free(m->meta.title);
 413         if (m->meta.vol)
 414                 free(m->meta.vol);
 415         if (m->meta.source)
 416                 free(m->meta.source);
 417         if (m->meta.msec)
 418                 free(m->meta.msec);
 419         if (m->meta.rawdate)
 420                 free(m->meta.rawdate);
 421
 422         m->meta.title = m->meta.vol = m->meta.rawdate =
 423                 m->meta.msec = m->meta.source = NULL;
 424         m->meta.date = 0;
 425
 426         /* ->TITLE<- MSEC DATE SOURCE VOL */
 427
 428         n = n->child;
 429         assert(n);
 430         m->meta.title = mandoc_strdup(n->string);
 431
 432         /* TITLE ->MSEC<- DATE SOURCE VOL */
 433
 434         n = n->next;
 435         assert(n);
 436         m->meta.msec = mandoc_strdup(n->string);
 437
 438         /* TITLE MSEC ->DATE<- SOURCE VOL */
 439
 440         /*
 441          * Try to parse the date.  If this works, stash the epoch (this
 442          * is optimal because we can reformat it in the canonical form).
 443          * If it doesn't parse, isn't specified at all, or is an empty
 444          * string, then use the current date.
 445          */
 446
 447         n = n->next;
 448         if (n && n->string && *n->string) {
 449                 m->meta.date = mandoc_a2time
 450                         (MTIME_ISO_8601, n->string);
 451                 if (0 == m->meta.date) {
 452                         man_nmsg(m, n, MANDOCERR_BADDATE);
 453                         m->meta.rawdate = mandoc_strdup(n->string);
 454                 }
 455         } else
 456                 m->meta.date = time(NULL);
 457
 458         /* TITLE MSEC DATE ->SOURCE<- VOL */
 459
 460         if (n && (n = n->next))
 461                 m->meta.source = mandoc_strdup(n->string);
 462
 463         /* TITLE MSEC DATE SOURCE ->VOL<- */
 464
 465         if (n && (n = n->next))
 466                 m->meta.vol = mandoc_strdup(n->string);
 467
 468         /*
 469          * Remove the `TH' node after we've processed it for our
 470          * meta-data.
 471          */
 472         man_node_delete(m, m->last);
 473         return(1);
 474 }
 475
 476 static int
 477 post_nf(CHKARGS)
 478 {
 479
 480         if (MAN_LITERAL & m->flags)
 481                 man_nmsg(m, n, MANDOCERR_SCOPEREP);
 482
 483         m->flags |= MAN_LITERAL;
 484         return(1);
 485 }
 486
 487 static int
 488 post_fi(CHKARGS)
 489 {
 490
 491         if ( ! (MAN_LITERAL & m->flags))
 492                 man_nmsg(m, n, MANDOCERR_NOSCOPE);
 493
 494         m->flags &= ~MAN_LITERAL;
 495         return(1);
 496 }
 497
 498 static int
 499 post_UC(CHKARGS)
 500 {
 501         static const char * const bsd_versions[] = {
 502             "3rd Berkeley Distribution",
 503             "4th Berkeley Distribution",
 504             "4.2 Berkeley Distribution",
 505             "4.3 Berkeley Distribution",
 506             "4.4 Berkeley Distribution",
 507         };
 508
 509         const char      *p, *s;
 510
 511         n = n->child;
 512         n = m->last->child;
 513
 514         if (NULL == n || MAN_TEXT != n->type)
 515                 p = bsd_versions[0];
 516         else {
 517                 s = n->string;
 518                 if (0 == strcmp(s, "3"))
 519                         p = bsd_versions[0];
 520                 else if (0 == strcmp(s, "4"))
 521                         p = bsd_versions[1];
 522                 else if (0 == strcmp(s, "5"))
 523                         p = bsd_versions[2];
 524                 else if (0 == strcmp(s, "6"))
 525                         p = bsd_versions[3];
 526                 else if (0 == strcmp(s, "7"))
 527                         p = bsd_versions[4];
 528                 else
 529                         p = bsd_versions[0];
 530         }
 531
 532         if (m->meta.source)
 533                 free(m->meta.source);
 534
 535         m->meta.source = mandoc_strdup(p);
 536         return(1);
 537 }
 538
 539 static int
 540 post_AT(CHKARGS)
 541 {
 542         static const char * const unix_versions[] = {
 543             "7th Edition",
 544             "System III",
 545             "System V",
 546             "System V Release 2",
 547         };
 548
 549         const char      *p, *s;
 550         struct man_node *nn;
 551
 552         n = n->child;
 553
 554         if (NULL == n || MAN_TEXT != n->type)
 555                 p = unix_versions[0];
 556         else {
 557                 s = n->string;
 558                 if (0 == strcmp(s, "3"))
 559                         p = unix_versions[0];
 560                 else if (0 == strcmp(s, "4"))
 561                         p = unix_versions[1];
 562                 else if (0 == strcmp(s, "5")) {
 563                         nn = n->next;
 564                         if (nn && MAN_TEXT == nn->type && nn->string[0])
 565                                 p = unix_versions[3];
 566                         else
 567                                 p = unix_versions[2];
 568                 } else
 569                         p = unix_versions[0];
 570         }
 571
 572         if (m->meta.source)
 573                 free(m->meta.source);
 574
 575         m->meta.source = mandoc_strdup(p);
 576         return(1);
 577 }