man_validate.c

   1 /* $Id: man_validate.c,v 1.151 2020/03/13 15:32:28 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2010, 2012-2020 Ingo Schwarze <schwarze@openbsd.org>
   4  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   5  *
   6  * Permission to use, copy, modify, and distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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.
  17  *
  18  * Validation module for man(7) syntax trees used by mandoc(1).
  19  */
  20 #include "config.h"
  21
  22 #include <sys/types.h>
  23
  24 #include <assert.h>
  25 #include <ctype.h>
  26 #include <errno.h>
  27 #include <limits.h>
  28 #include <stdarg.h>
  29 #include <stdio.h>
  30 #include <stdlib.h>
  31 #include <string.h>
  32 #include <time.h>
  33
  34 #include "mandoc_aux.h"
  35 #include "mandoc.h"
  36 #include "roff.h"
  37 #include "tag.h"
  38 #include "man.h"
  39 #include "libmandoc.h"
  40 #include "roff_int.h"
  41 #include "libman.h"
  42
  43 #define CHKARGS   struct roff_man *man, struct roff_node *n
  44
  45 typedef void    (*v_check)(CHKARGS);
  46
  47 static  void      check_abort(CHKARGS) __attribute__((__noreturn__));
  48 static  void      check_par(CHKARGS);
  49 static  void      check_part(CHKARGS);
  50 static  void      check_root(CHKARGS);
  51 static  void      check_tag(struct roff_node *, struct roff_node *);
  52 static  void      check_text(CHKARGS);
  53
  54 static  void      post_AT(CHKARGS);
  55 static  void      post_EE(CHKARGS);
  56 static  void      post_EX(CHKARGS);
  57 static  void      post_IP(CHKARGS);
  58 static  void      post_OP(CHKARGS);
  59 static  void      post_SH(CHKARGS);
  60 static  void      post_TH(CHKARGS);
  61 static  void      post_TP(CHKARGS);
  62 static  void      post_UC(CHKARGS);
  63 static  void      post_UR(CHKARGS);
  64 static  void      post_in(CHKARGS);
  65
  66 static  const v_check man_valids[MAN_MAX - MAN_TH] = {
  67         post_TH,    /* TH */
  68         post_SH,    /* SH */
  69         post_SH,    /* SS */
  70         post_TP,    /* TP */
  71         post_TP,    /* TQ */
  72         check_abort,/* LP */
  73         check_par,  /* PP */
  74         check_abort,/* P */
  75         post_IP,    /* IP */
  76         NULL,       /* HP */
  77         NULL,       /* SM */
  78         NULL,       /* SB */
  79         NULL,       /* BI */
  80         NULL,       /* IB */
  81         NULL,       /* BR */
  82         NULL,       /* RB */
  83         NULL,       /* R */
  84         NULL,       /* B */
  85         NULL,       /* I */
  86         NULL,       /* IR */
  87         NULL,       /* RI */
  88         NULL,       /* RE */
  89         check_part, /* RS */
  90         NULL,       /* DT */
  91         post_UC,    /* UC */
  92         NULL,       /* PD */
  93         post_AT,    /* AT */
  94         post_in,    /* in */
  95         NULL,       /* SY */
  96         NULL,       /* YS */
  97         post_OP,    /* OP */
  98         post_EX,    /* EX */
  99         post_EE,    /* EE */
 100         post_UR,    /* UR */
 101         NULL,       /* UE */
 102         post_UR,    /* MT */
 103         NULL,       /* ME */
 104 };
 105
 106
 107 /* Validate the subtree rooted at man->last. */
 108 void
 109 man_validate(struct roff_man *man)
 110 {
 111         struct roff_node *n;
 112         const v_check    *cp;
 113
 114         /*
 115          * Translate obsolete macros such that later code
 116          * does not need to look for them.
 117          */
 118
 119         n = man->last;
 120         switch (n->tok) {
 121         case MAN_LP:
 122         case MAN_P:
 123                 n->tok = MAN_PP;
 124                 break;
 125         default:
 126                 break;
 127         }
 128
 129         /*
 130          * Iterate over all children, recursing into each one
 131          * in turn, depth-first.
 132          */
 133
 134         man->last = man->last->child;
 135         while (man->last != NULL) {
 136                 man_validate(man);
 137                 if (man->last == n)
 138                         man->last = man->last->child;
 139                 else
 140                         man->last = man->last->next;
 141         }
 142
 143         /* Finally validate the macro itself. */
 144
 145         man->last = n;
 146         man->next = ROFF_NEXT_SIBLING;
 147         switch (n->type) {
 148         case ROFFT_TEXT:
 149                 check_text(man, n);
 150                 break;
 151         case ROFFT_ROOT:
 152                 check_root(man, n);
 153                 break;
 154         case ROFFT_COMMENT:
 155         case ROFFT_EQN:
 156         case ROFFT_TBL:
 157                 break;
 158         default:
 159                 if (n->tok < ROFF_MAX) {
 160                         roff_validate(man);
 161                         break;
 162                 }
 163                 assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
 164                 cp = man_valids + (n->tok - MAN_TH);
 165                 if (*cp)
 166                         (*cp)(man, n);
 167                 if (man->last == n)
 168                         n->flags |= NODE_VALID;
 169                 break;
 170         }
 171 }
 172
 173 static void
 174 check_root(CHKARGS)
 175 {
 176         assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
 177
 178         if (n->last == NULL || n->last->type == ROFFT_COMMENT)
 179                 mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
 180         else
 181                 man->meta.hasbody = 1;
 182
 183         if (NULL == man->meta.title) {
 184                 mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
 185
 186                 /*
 187                  * If a title hasn't been set, do so now (by
 188                  * implication, date and section also aren't set).
 189                  */
 190
 191                 man->meta.title = mandoc_strdup("");
 192                 man->meta.msec = mandoc_strdup("");
 193                 man->meta.date = mandoc_normdate(NULL, NULL);
 194         }
 195
 196         if (man->meta.os_e &&
 197             (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
 198                 mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
 199                     man->meta.os_e == MANDOC_OS_OPENBSD ?
 200                     "(OpenBSD)" : "(NetBSD)");
 201 }
 202
 203 static void
 204 check_abort(CHKARGS)
 205 {
 206         abort();
 207 }
 208
 209 /*
 210  * Skip leading whitespace, dashes, backslashes, and font escapes,
 211  * then create a tag if the first following byte is a letter.
 212  * Priority is high unless whitespace is present.
 213  */
 214 static void
 215 check_tag(struct roff_node *n, struct roff_node *nt)
 216 {
 217         const char      *cp, *arg;
 218         int              prio, sz;
 219
 220         if (nt == NULL || nt->type != ROFFT_TEXT)
 221                 return;
 222
 223         cp = nt->string;
 224         prio = TAG_STRONG;
 225         for (;;) {
 226                 switch (*cp) {
 227                 case ' ':
 228                 case '\t':
 229                         prio = TAG_WEAK;
 230                         /* FALLTHROUGH */
 231                 case '-':
 232                         cp++;
 233                         break;
 234                 case '\\':
 235                         cp++;
 236                         switch (mandoc_escape(&cp, &arg, &sz)) {
 237                         case ESCAPE_FONT:
 238                         case ESCAPE_FONTBOLD:
 239                         case ESCAPE_FONTITALIC:
 240                         case ESCAPE_FONTBI:
 241                         case ESCAPE_FONTROMAN:
 242                         case ESCAPE_FONTCW:
 243                         case ESCAPE_FONTPREV:
 244                         case ESCAPE_IGNORE:
 245                                 break;
 246                         case ESCAPE_SPECIAL:
 247                                 if (sz != 1)
 248                                         return;
 249                                 switch (*arg) {
 250                                 case '-':
 251                                 case 'e':
 252                                         break;
 253                                 default:
 254                                         return;
 255                                 }
 256                                 break;
 257                         default:
 258                                 return;
 259                         }
 260                         break;
 261                 default:
 262                         if (isalpha((unsigned char)*cp))
 263                                 tag_put(cp, prio, n);
 264                         return;
 265                 }
 266         }
 267 }
 268
 269 static void
 270 check_text(CHKARGS)
 271 {
 272         char            *cp, *p;
 273
 274         if (n->flags & NODE_NOFILL)
 275                 return;
 276
 277         cp = n->string;
 278         for (p = cp; NULL != (p = strchr(p, '\t')); p++)
 279                 mandoc_msg(MANDOCERR_FI_TAB,
 280                     n->line, n->pos + (int)(p - cp), NULL);
 281 }
 282
 283 static void
 284 post_EE(CHKARGS)
 285 {
 286         if ((n->flags & NODE_NOFILL) == 0)
 287                 mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
 288 }
 289
 290 static void
 291 post_EX(CHKARGS)
 292 {
 293         if (n->flags & NODE_NOFILL)
 294                 mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
 295 }
 296
 297 static void
 298 post_OP(CHKARGS)
 299 {
 300
 301         if (n->child == NULL)
 302                 mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
 303         else if (n->child->next != NULL && n->child->next->next != NULL) {
 304                 n = n->child->next->next;
 305                 mandoc_msg(MANDOCERR_ARG_EXCESS,
 306                     n->line, n->pos, "OP ... %s", n->string);
 307         }
 308 }
 309
 310 static void
 311 post_SH(CHKARGS)
 312 {
 313         struct roff_node        *nc;
 314
 315         if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
 316                 return;
 317
 318         if (nc->tok == MAN_PP && nc->body->child != NULL) {
 319                 while (nc->body->last != NULL) {
 320                         man->next = ROFF_NEXT_CHILD;
 321                         roff_node_relink(man, nc->body->last);
 322                         man->last = n;
 323                 }
 324         }
 325
 326         if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
 327                 mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
 328                     "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
 329                 roff_node_delete(man, nc);
 330         }
 331
 332         /*
 333          * Trailing PP is empty, so it is deleted by check_par().
 334          * Trailing sp is significant.
 335          */
 336
 337         if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
 338                 mandoc_msg(MANDOCERR_PAR_SKIP,
 339                     nc->line, nc->pos, "%s at the end of %s",
 340                     roff_name[nc->tok], roff_name[n->tok]);
 341                 roff_node_delete(man, nc);
 342         }
 343 }
 344
 345 static void
 346 post_UR(CHKARGS)
 347 {
 348         if (n->type == ROFFT_HEAD && n->child == NULL)
 349                 mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
 350                     "%s", roff_name[n->tok]);
 351         check_part(man, n);
 352 }
 353
 354 static void
 355 check_part(CHKARGS)
 356 {
 357
 358         if (n->type == ROFFT_BODY && n->child == NULL)
 359                 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
 360                     "%s", roff_name[n->tok]);
 361 }
 362
 363 static void
 364 check_par(CHKARGS)
 365 {
 366
 367         switch (n->type) {
 368         case ROFFT_BLOCK:
 369                 if (n->body->child == NULL)
 370                         roff_node_delete(man, n);
 371                 break;
 372         case ROFFT_BODY:
 373                 if (n->child != NULL &&
 374                     (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
 375                         mandoc_msg(MANDOCERR_PAR_SKIP,
 376                             n->child->line, n->child->pos,
 377                             "%s after %s", roff_name[n->child->tok],
 378                             roff_name[n->tok]);
 379                         roff_node_delete(man, n->child);
 380                 }
 381                 if (n->child == NULL)
 382                         mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
 383                             "%s empty", roff_name[n->tok]);
 384                 break;
 385         case ROFFT_HEAD:
 386                 if (n->child != NULL)
 387                         mandoc_msg(MANDOCERR_ARG_SKIP,
 388                             n->line, n->pos, "%s %s%s",
 389                             roff_name[n->tok], n->child->string,
 390                             n->child->next != NULL ? " ..." : "");
 391                 break;
 392         default:
 393                 break;
 394         }
 395 }
 396
 397 static void
 398 post_IP(CHKARGS)
 399 {
 400         switch (n->type) {
 401         case ROFFT_BLOCK:
 402                 if (n->head->child == NULL && n->body->child == NULL)
 403                         roff_node_delete(man, n);
 404                 break;
 405         case ROFFT_HEAD:
 406                 check_tag(n, n->child);
 407                 break;
 408         case ROFFT_BODY:
 409                 if (n->parent->head->child == NULL && n->child == NULL)
 410                         mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
 411                             "%s empty", roff_name[n->tok]);
 412                 break;
 413         default:
 414                 break;
 415         }
 416 }
 417
 418 /*
 419  * The first next-line element in the head is the tag.
 420  * If that's a font macro, use its first child instead.
 421  */
 422 static void
 423 post_TP(CHKARGS)
 424 {
 425         struct roff_node *nt;
 426
 427         if (n->type != ROFFT_HEAD || (nt = n->child) == NULL)
 428                 return;
 429
 430         while ((nt->flags & NODE_LINE) == 0)
 431                 if ((nt = nt->next) == NULL)
 432                         return;
 433
 434         switch (nt->tok) {
 435         case MAN_B:
 436         case MAN_BI:
 437         case MAN_BR:
 438         case MAN_I:
 439         case MAN_IB:
 440         case MAN_IR:
 441                 nt = nt->child;
 442                 break;
 443         default:
 444                 break;
 445         }
 446         check_tag(n, nt);
 447 }
 448
 449 static void
 450 post_TH(CHKARGS)
 451 {
 452         struct roff_node *nb;
 453         const char      *p;
 454
 455         free(man->meta.title);
 456         free(man->meta.vol);
 457         free(man->meta.os);
 458         free(man->meta.msec);
 459         free(man->meta.date);
 460
 461         man->meta.title = man->meta.vol = man->meta.date =
 462             man->meta.msec = man->meta.os = NULL;
 463
 464         nb = n;
 465
 466         /* ->TITLE<- MSEC DATE OS VOL */
 467
 468         n = n->child;
 469         if (n != NULL && n->string != NULL) {
 470                 for (p = n->string; *p != '\0'; p++) {
 471                         /* Only warn about this once... */
 472                         if (isalpha((unsigned char)*p) &&
 473                             ! isupper((unsigned char)*p)) {
 474                                 mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
 475                                     n->pos + (int)(p - n->string),
 476                                     "TH %s", n->string);
 477                                 break;
 478                         }
 479                 }
 480                 man->meta.title = mandoc_strdup(n->string);
 481         } else {
 482                 man->meta.title = mandoc_strdup("");
 483                 mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
 484         }
 485
 486         /* TITLE ->MSEC<- DATE OS VOL */
 487
 488         if (n != NULL)
 489                 n = n->next;
 490         if (n != NULL && n->string != NULL)
 491                 man->meta.msec = mandoc_strdup(n->string);
 492         else {
 493                 man->meta.msec = mandoc_strdup("");
 494                 mandoc_msg(MANDOCERR_MSEC_MISSING,
 495                     nb->line, nb->pos, "TH %s", man->meta.title);
 496         }
 497
 498         /* TITLE MSEC ->DATE<- OS VOL */
 499
 500         if (n != NULL)
 501                 n = n->next;
 502         if (man->quick && n != NULL)
 503                 man->meta.date = mandoc_strdup("");
 504         else
 505                 man->meta.date = mandoc_normdate(n, nb);
 506
 507         /* TITLE MSEC DATE ->OS<- VOL */
 508
 509         if (n && (n = n->next))
 510                 man->meta.os = mandoc_strdup(n->string);
 511         else if (man->os_s != NULL)
 512                 man->meta.os = mandoc_strdup(man->os_s);
 513         if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
 514                 if (strstr(man->meta.os, "OpenBSD") != NULL)
 515                         man->meta.os_e = MANDOC_OS_OPENBSD;
 516                 else if (strstr(man->meta.os, "NetBSD") != NULL)
 517                         man->meta.os_e = MANDOC_OS_NETBSD;
 518         }
 519
 520         /* TITLE MSEC DATE OS ->VOL<- */
 521         /* If missing, use the default VOL name for MSEC. */
 522
 523         if (n && (n = n->next))
 524                 man->meta.vol = mandoc_strdup(n->string);
 525         else if ('\0' != man->meta.msec[0] &&
 526             (NULL != (p = mandoc_a2msec(man->meta.msec))))
 527                 man->meta.vol = mandoc_strdup(p);
 528
 529         if (n != NULL && (n = n->next) != NULL)
 530                 mandoc_msg(MANDOCERR_ARG_EXCESS,
 531                     n->line, n->pos, "TH ... %s", n->string);
 532
 533         /*
 534          * Remove the `TH' node after we've processed it for our
 535          * meta-data.
 536          */
 537         roff_node_delete(man, man->last);
 538 }
 539
 540 static void
 541 post_UC(CHKARGS)
 542 {
 543         static const char * const bsd_versions[] = {
 544             "3rd Berkeley Distribution",
 545             "4th Berkeley Distribution",
 546             "4.2 Berkeley Distribution",
 547             "4.3 Berkeley Distribution",
 548             "4.4 Berkeley Distribution",
 549         };
 550
 551         const char      *p, *s;
 552
 553         n = n->child;
 554
 555         if (n == NULL || n->type != ROFFT_TEXT)
 556                 p = bsd_versions[0];
 557         else {
 558                 s = n->string;
 559                 if (0 == strcmp(s, "3"))
 560                         p = bsd_versions[0];
 561                 else if (0 == strcmp(s, "4"))
 562                         p = bsd_versions[1];
 563                 else if (0 == strcmp(s, "5"))
 564                         p = bsd_versions[2];
 565                 else if (0 == strcmp(s, "6"))
 566                         p = bsd_versions[3];
 567                 else if (0 == strcmp(s, "7"))
 568                         p = bsd_versions[4];
 569                 else
 570                         p = bsd_versions[0];
 571         }
 572
 573         free(man->meta.os);
 574         man->meta.os = mandoc_strdup(p);
 575 }
 576
 577 static void
 578 post_AT(CHKARGS)
 579 {
 580         static const char * const unix_versions[] = {
 581             "7th Edition",
 582             "System III",
 583             "System V",
 584             "System V Release 2",
 585         };
 586
 587         struct roff_node *nn;
 588         const char      *p, *s;
 589
 590         n = n->child;
 591
 592         if (n == NULL || n->type != ROFFT_TEXT)
 593                 p = unix_versions[0];
 594         else {
 595                 s = n->string;
 596                 if (0 == strcmp(s, "3"))
 597                         p = unix_versions[0];
 598                 else if (0 == strcmp(s, "4"))
 599                         p = unix_versions[1];
 600                 else if (0 == strcmp(s, "5")) {
 601                         nn = n->next;
 602                         if (nn != NULL &&
 603                             nn->type == ROFFT_TEXT &&
 604                             nn->string[0] != '\0')
 605                                 p = unix_versions[3];
 606                         else
 607                                 p = unix_versions[2];
 608                 } else
 609                         p = unix_versions[0];
 610         }
 611
 612         free(man->meta.os);
 613         man->meta.os = mandoc_strdup(p);
 614 }
 615
 616 static void
 617 post_in(CHKARGS)
 618 {
 619         char    *s;
 620
 621         if (n->parent->tok != MAN_TP ||
 622             n->parent->type != ROFFT_HEAD ||
 623             n->child == NULL ||
 624             *n->child->string == '+' ||
 625             *n->child->string == '-')
 626                 return;
 627         mandoc_asprintf(&s, "+%s", n->child->string);
 628         free(n->child->string);
 629         n->child->string = s;
 630 }