man_validate.c

   1 /*      $Id: man_validate.c,v 1.56 2010/12/08 10:58:22 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         if (MAN_TEXT == n->type)
 121                 return(1);
 122         if (MAN_ROOT == n->type)
 123                 return(1);
 124
 125         if (NULL == (cp = man_valids[n->tok].pres))
 126                 return(1);
 127         for ( ; *cp; cp++)
 128                 if ( ! (*cp)(m, n))
 129                         return(0);
 130         return(1);
 131 }
 132
 133
 134 int
 135 man_valid_post(struct man *m)
 136 {
 137         v_check         *cp;
 138
 139         if (MAN_VALID & m->last->flags)
 140                 return(1);
 141         m->last->flags |= MAN_VALID;
 142
 143         switch (m->last->type) {
 144         case (MAN_TEXT):
 145                 return(check_text(m, m->last));
 146         case (MAN_ROOT):
 147                 return(check_root(m, m->last));
 148         default:
 149                 break;
 150         }
 151
 152         if (NULL == (cp = man_valids[m->last->tok].posts))
 153                 return(1);
 154         for ( ; *cp; cp++)
 155                 if ( ! (*cp)(m, m->last))
 156                         return(0);
 157
 158         return(1);
 159 }
 160
 161
 162 static int
 163 check_root(CHKARGS)
 164 {
 165
 166         if (MAN_BLINE & m->flags)
 167                 man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
 168         else if (MAN_ELINE & m->flags)
 169                 man_nmsg(m, n, MANDOCERR_SCOPEEXIT);
 170
 171         m->flags &= ~MAN_BLINE;
 172         m->flags &= ~MAN_ELINE;
 173
 174         if (NULL == m->first->child) {
 175                 man_nmsg(m, n, MANDOCERR_NODOCBODY);
 176                 return(0);
 177         } else if (NULL == m->meta.title) {
 178                 man_nmsg(m, n, MANDOCERR_NOTITLE);
 179
 180                 /*
 181                  * If a title hasn't been set, do so now (by
 182                  * implication, date and section also aren't set).
 183                  */
 184
 185                 m->meta.title = mandoc_strdup("unknown");
 186                 m->meta.date = time(NULL);
 187                 m->meta.msec = mandoc_strdup("1");
 188         }
 189
 190         return(1);
 191 }
 192
 193
 194 static int
 195 check_title(CHKARGS)
 196 {
 197         const char      *p;
 198
 199         assert(n->child);
 200         /* FIXME: is this sufficient? */
 201         if ('\0' == *n->child->string) {
 202                 man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
 203                 return(0);
 204         }
 205
 206         for (p = n->child->string; '\0' != *p; p++)
 207                 /* Only warn about this once... */
 208                 if (isalpha((u_char)*p) && ! isupper((u_char)*p)) {
 209                         man_nmsg(m, n, MANDOCERR_UPPERCASE);
 210                         break;
 211                 }
 212
 213         return(1);
 214 }
 215
 216
 217 static int
 218 check_text(CHKARGS)
 219 {
 220         char            *p;
 221         int              pos, c;
 222         size_t           sz;
 223
 224         for (p = n->string, pos = n->pos + 1; *p; p++, pos++) {
 225                 sz = strcspn(p, "\t\\");
 226                 p += (int)sz;
 227
 228                 if ('\0' == *p)
 229                         break;
 230
 231                 pos += (int)sz;
 232
 233                 if ('\t' == *p) {
 234                         if (MAN_LITERAL & m->flags)
 235                                 continue;
 236                         if (man_pmsg(m, n->line, pos, MANDOCERR_BADTAB))
 237                                 continue;
 238                         return(0);
 239                 }
 240
 241                 /* Check the special character. */
 242
 243                 c = mandoc_special(p);
 244                 if (c) {
 245                         p += c - 1;
 246                         pos += c - 1;
 247                 } else
 248                         man_pmsg(m, n->line, pos, MANDOCERR_BADESCAPE);
 249         }
 250
 251         return(1);
 252 }
 253
 254
 255 #define INEQ_DEFINE(x, ineq, name) \
 256 static int \
 257 check_##name(CHKARGS) \
 258 { \
 259         if (n->nchild ineq (x)) \
 260                 return(1); \
 261         man_vmsg(m, MANDOCERR_SYNTARGCOUNT, n->line, n->pos, \
 262                         "line arguments %s %d (have %d)", \
 263                         #ineq, (x), n->nchild); \
 264         return(0); \
 265 }
 266
 267 INEQ_DEFINE(0, ==, eq0)
 268 INEQ_DEFINE(1, <=, le1)
 269 INEQ_DEFINE(2, >=, ge2)
 270 INEQ_DEFINE(5, <=, le5)
 271
 272 static int
 273 check_ft(CHKARGS)
 274 {
 275         char    *cp;
 276         int      ok;
 277
 278         if (0 == n->nchild)
 279                 return(1);
 280
 281         ok = 0;
 282         cp = n->child->string;
 283         switch (*cp) {
 284         case ('1'):
 285                 /* FALLTHROUGH */
 286         case ('2'):
 287                 /* FALLTHROUGH */
 288         case ('3'):
 289                 /* FALLTHROUGH */
 290         case ('4'):
 291                 /* FALLTHROUGH */
 292         case ('I'):
 293                 /* FALLTHROUGH */
 294         case ('P'):
 295                 /* FALLTHROUGH */
 296         case ('R'):
 297                 if ('\0' == cp[1])
 298                         ok = 1;
 299                 break;
 300         case ('B'):
 301                 if ('\0' == cp[1] || ('I' == cp[1] && '\0' == cp[2]))
 302                         ok = 1;
 303                 break;
 304         case ('C'):
 305                 if ('W' == cp[1] && '\0' == cp[2])
 306                         ok = 1;
 307                 break;
 308         default:
 309                 break;
 310         }
 311
 312         if (0 == ok) {
 313                 man_vmsg(m, MANDOCERR_BADFONT,
 314                                 n->line, n->pos, "%s", cp);
 315                 *cp = '\0';
 316         }
 317
 318         if (1 < n->nchild)
 319                 man_vmsg(m, MANDOCERR_ARGCOUNT, n->line, n->pos,
 320                                 "want one child (have %d)", n->nchild);
 321
 322         return(1);
 323 }
 324
 325 static int
 326 check_sec(CHKARGS)
 327 {
 328
 329         if (MAN_HEAD == n->type && 0 == n->nchild) {
 330                 man_nmsg(m, n, MANDOCERR_SYNTARGCOUNT);
 331                 return(0);
 332         } else if (MAN_BODY == n->type && 0 == n->nchild)
 333                 man_nmsg(m, n, MANDOCERR_NOBODY);
 334
 335         return(1);
 336 }
 337
 338
 339 static int
 340 check_part(CHKARGS)
 341 {
 342
 343         if (MAN_BODY == n->type && 0 == n->nchild)
 344                 man_nmsg(m, n, MANDOCERR_NOBODY);
 345
 346         return(1);
 347 }
 348
 349
 350 static int
 351 check_par(CHKARGS)
 352 {
 353
 354         if (MAN_BODY == n->type)
 355                 switch (n->tok) {
 356                 case (MAN_IP):
 357                         /* FALLTHROUGH */
 358                 case (MAN_HP):
 359                         /* FALLTHROUGH */
 360                 case (MAN_TP):
 361                         /* Body-less lists are ok. */
 362                         break;
 363                 default:
 364                         if (0 == n->nchild)
 365                                 man_nmsg(m, n, MANDOCERR_NOBODY);
 366                         break;
 367                 }
 368         if (MAN_HEAD == n->type)
 369                 switch (n->tok) {
 370                 case (MAN_PP):
 371                         /* FALLTHROUGH */
 372                 case (MAN_P):
 373                         /* FALLTHROUGH */
 374                 case (MAN_LP):
 375                         if (n->nchild)
 376                                 man_nmsg(m, n, MANDOCERR_ARGSLOST);
 377                         break;
 378                 default:
 379                         break;
 380                 }
 381
 382         return(1);
 383 }
 384
 385
 386 static int
 387 check_bline(CHKARGS)
 388 {
 389
 390         assert( ! (MAN_ELINE & m->flags));
 391         if (MAN_BLINE & m->flags) {
 392                 man_nmsg(m, n, MANDOCERR_SYNTLINESCOPE);
 393                 return(0);
 394         }
 395
 396         return(1);
 397 }
 398
 399 static int
 400 post_TH(CHKARGS)
 401 {
 402
 403         if (m->meta.title)
 404                 free(m->meta.title);
 405         if (m->meta.vol)
 406                 free(m->meta.vol);
 407         if (m->meta.source)
 408                 free(m->meta.source);
 409         if (m->meta.msec)
 410                 free(m->meta.msec);
 411         if (m->meta.rawdate)
 412                 free(m->meta.rawdate);
 413
 414         m->meta.title = m->meta.vol = m->meta.rawdate =
 415                 m->meta.msec = m->meta.source = NULL;
 416         m->meta.date = 0;
 417
 418         /* ->TITLE<- MSEC DATE SOURCE VOL */
 419
 420         n = n->child;
 421         assert(n);
 422         m->meta.title = mandoc_strdup(n->string);
 423
 424         /* TITLE ->MSEC<- DATE SOURCE VOL */
 425
 426         n = n->next;
 427         assert(n);
 428         m->meta.msec = mandoc_strdup(n->string);
 429
 430         /* TITLE MSEC ->DATE<- SOURCE VOL */
 431
 432         /*
 433          * Try to parse the date.  If this works, stash the epoch (this
 434          * is optimal because we can reformat it in the canonical form).
 435          * If it doesn't parse, isn't specified at all, or is an empty
 436          * string, then use the current date.
 437          */
 438
 439         n = n->next;
 440         if (n && n->string && *n->string) {
 441                 m->meta.date = mandoc_a2time
 442                         (MTIME_ISO_8601, n->string);
 443                 if (0 == m->meta.date) {
 444                         man_nmsg(m, n, MANDOCERR_BADDATE);
 445                         m->meta.rawdate = mandoc_strdup(n->string);
 446                 }
 447         } else
 448                 m->meta.date = time(NULL);
 449
 450         /* TITLE MSEC DATE ->SOURCE<- VOL */
 451
 452         if (n && (n = n->next))
 453                 m->meta.source = mandoc_strdup(n->string);
 454
 455         /* TITLE MSEC DATE SOURCE ->VOL<- */
 456
 457         if (n && (n = n->next))
 458                 m->meta.vol = mandoc_strdup(n->string);
 459
 460         /*
 461          * Remove the `TH' node after we've processed it for our
 462          * meta-data.
 463          */
 464         man_node_delete(m, m->last);
 465         return(1);
 466 }
 467
 468 static int
 469 post_nf(CHKARGS)
 470 {
 471
 472         if (MAN_LITERAL & m->flags)
 473                 man_nmsg(m, n, MANDOCERR_SCOPEREP);
 474
 475         m->flags |= MAN_LITERAL;
 476         return(1);
 477 }
 478
 479 static int
 480 post_fi(CHKARGS)
 481 {
 482
 483         if ( ! (MAN_LITERAL & m->flags))
 484                 man_nmsg(m, n, MANDOCERR_NOSCOPE);
 485
 486         m->flags &= ~MAN_LITERAL;
 487         return(1);
 488 }
 489
 490 static int
 491 post_UC(CHKARGS)
 492 {
 493         static const char * const bsd_versions[] = {
 494             "3rd Berkeley Distribution",
 495             "4th Berkeley Distribution",
 496             "4.2 Berkeley Distribution",
 497             "4.3 Berkeley Distribution",
 498             "4.4 Berkeley Distribution",
 499         };
 500
 501         const char      *p, *s;
 502
 503         n = n->child;
 504         n = m->last->child;
 505
 506         if (NULL == n || MAN_TEXT != n->type)
 507                 p = bsd_versions[0];
 508         else {
 509                 s = n->string;
 510                 if (0 == strcmp(s, "3"))
 511                         p = bsd_versions[0];
 512                 else if (0 == strcmp(s, "4"))
 513                         p = bsd_versions[1];
 514                 else if (0 == strcmp(s, "5"))
 515                         p = bsd_versions[2];
 516                 else if (0 == strcmp(s, "6"))
 517                         p = bsd_versions[3];
 518                 else if (0 == strcmp(s, "7"))
 519                         p = bsd_versions[4];
 520                 else
 521                         p = bsd_versions[0];
 522         }
 523
 524         if (m->meta.source)
 525                 free(m->meta.source);
 526
 527         m->meta.source = mandoc_strdup(p);
 528         return(1);
 529 }
 530
 531 static int
 532 post_AT(CHKARGS)
 533 {
 534         static const char * const unix_versions[] = {
 535             "7th Edition",
 536             "System III",
 537             "System V",
 538             "System V Release 2",
 539         };
 540
 541         const char      *p, *s;
 542         struct man_node *nn;
 543
 544         n = n->child;
 545
 546         if (NULL == n || MAN_TEXT != n->type)
 547                 p = unix_versions[0];
 548         else {
 549                 s = n->string;
 550                 if (0 == strcmp(s, "3"))
 551                         p = unix_versions[0];
 552                 else if (0 == strcmp(s, "4"))
 553                         p = unix_versions[1];
 554                 else if (0 == strcmp(s, "5")) {
 555                         nn = n->next;
 556                         if (nn && MAN_TEXT == nn->type && nn->string[0])
 557                                 p = unix_versions[3];
 558                         else
 559                                 p = unix_versions[2];
 560                 } else
 561                         p = unix_versions[0];
 562         }
 563
 564         if (m->meta.source)
 565                 free(m->meta.source);
 566
 567         m->meta.source = mandoc_strdup(p);
 568         return(1);
 569 }