man_validate.c

   1 /*      $Id: man_validate.c,v 1.149 2019/06/27 15:07:30 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010, 2012-2018 Ingo Schwarze <schwarze@openbsd.org>
   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 #include "config.h"
  19
  20 #include <sys/types.h>
  21
  22 #include <assert.h>
  23 #include <ctype.h>
  24 #include <errno.h>
  25 #include <limits.h>
  26 #include <stdarg.h>
  27 #include <stdio.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <time.h>
  31
  32 #include "mandoc_aux.h"
  33 #include "mandoc.h"
  34 #include "roff.h"
  35 #include "man.h"
  36 #include "libmandoc.h"
  37 #include "roff_int.h"
  38 #include "libman.h"
  39
  40 #define CHKARGS   struct roff_man *man, struct roff_node *n
  41
  42 typedef void    (*v_check)(CHKARGS);
  43
  44 static  void      check_abort(CHKARGS) __attribute__((__noreturn__));
  45 static  void      check_par(CHKARGS);
  46 static  void      check_part(CHKARGS);
  47 static  void      check_root(CHKARGS);
  48 static  void      check_text(CHKARGS);
  49
  50 static  void      post_AT(CHKARGS);
  51 static  void      post_EE(CHKARGS);
  52 static  void      post_EX(CHKARGS);
  53 static  void      post_IP(CHKARGS);
  54 static  void      post_OP(CHKARGS);
  55 static  void      post_SH(CHKARGS);
  56 static  void      post_TH(CHKARGS);
  57 static  void      post_UC(CHKARGS);
  58 static  void      post_UR(CHKARGS);
  59 static  void      post_in(CHKARGS);
  60
  61 static  const v_check man_valids[MAN_MAX - MAN_TH] = {
  62         post_TH,    /* TH */
  63         post_SH,    /* SH */
  64         post_SH,    /* SS */
  65         NULL,       /* TP */
  66         NULL,       /* TQ */
  67         check_abort,/* LP */
  68         check_par,  /* PP */
  69         check_abort,/* P */
  70         post_IP,    /* IP */
  71         NULL,       /* HP */
  72         NULL,       /* SM */
  73         NULL,       /* SB */
  74         NULL,       /* BI */
  75         NULL,       /* IB */
  76         NULL,       /* BR */
  77         NULL,       /* RB */
  78         NULL,       /* R */
  79         NULL,       /* B */
  80         NULL,       /* I */
  81         NULL,       /* IR */
  82         NULL,       /* RI */
  83         NULL,       /* RE */
  84         check_part, /* RS */
  85         NULL,       /* DT */
  86         post_UC,    /* UC */
  87         NULL,       /* PD */
  88         post_AT,    /* AT */
  89         post_in,    /* in */
  90         NULL,       /* SY */
  91         NULL,       /* YS */
  92         post_OP,    /* OP */
  93         post_EX,    /* EX */
  94         post_EE,    /* EE */
  95         post_UR,    /* UR */
  96         NULL,       /* UE */
  97         post_UR,    /* MT */
  98         NULL,       /* ME */
  99 };
 100
 101
 102 /* Validate the subtree rooted at man->last. */
 103 void
 104 man_validate(struct roff_man *man)
 105 {
 106         struct roff_node *n;
 107         const v_check    *cp;
 108
 109         /*
 110          * Translate obsolete macros such that later code
 111          * does not need to look for them.
 112          */
 113
 114         n = man->last;
 115         switch (n->tok) {
 116         case MAN_LP:
 117         case MAN_P:
 118                 n->tok = MAN_PP;
 119                 break;
 120         default:
 121                 break;
 122         }
 123
 124         /*
 125          * Iterate over all children, recursing into each one
 126          * in turn, depth-first.
 127          */
 128
 129         man->last = man->last->child;
 130         while (man->last != NULL) {
 131                 man_validate(man);
 132                 if (man->last == n)
 133                         man->last = man->last->child;
 134                 else
 135                         man->last = man->last->next;
 136         }
 137
 138         /* Finally validate the macro itself. */
 139
 140         man->last = n;
 141         man->next = ROFF_NEXT_SIBLING;
 142         switch (n->type) {
 143         case ROFFT_TEXT:
 144                 check_text(man, n);
 145                 break;
 146         case ROFFT_ROOT:
 147                 check_root(man, n);
 148                 break;
 149         case ROFFT_COMMENT:
 150         case ROFFT_EQN:
 151         case ROFFT_TBL:
 152                 break;
 153         default:
 154                 if (n->tok < ROFF_MAX) {
 155                         roff_validate(man);
 156                         break;
 157                 }
 158                 assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
 159                 cp = man_valids + (n->tok - MAN_TH);
 160                 if (*cp)
 161                         (*cp)(man, n);
 162                 if (man->last == n)
 163                         n->flags |= NODE_VALID;
 164                 break;
 165         }
 166 }
 167
 168 static void
 169 check_root(CHKARGS)
 170 {
 171         assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
 172
 173         if (n->last == NULL || n->last->type == ROFFT_COMMENT)
 174                 mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL);
 175         else
 176                 man->meta.hasbody = 1;
 177
 178         if (NULL == man->meta.title) {
 179                 mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL);
 180
 181                 /*
 182                  * If a title hasn't been set, do so now (by
 183                  * implication, date and section also aren't set).
 184                  */
 185
 186                 man->meta.title = mandoc_strdup("");
 187                 man->meta.msec = mandoc_strdup("");
 188                 man->meta.date = mandoc_normdate(man, NULL, n->line, n->pos);
 189         }
 190
 191         if (man->meta.os_e &&
 192             (man->meta.rcsids & (1 << man->meta.os_e)) == 0)
 193                 mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0,
 194                     man->meta.os_e == MANDOC_OS_OPENBSD ?
 195                     "(OpenBSD)" : "(NetBSD)");
 196 }
 197
 198 static void
 199 check_abort(CHKARGS)
 200 {
 201         abort();
 202 }
 203
 204 static void
 205 check_text(CHKARGS)
 206 {
 207         char            *cp, *p;
 208
 209         if (n->flags & NODE_NOFILL)
 210                 return;
 211
 212         cp = n->string;
 213         for (p = cp; NULL != (p = strchr(p, '\t')); p++)
 214                 mandoc_msg(MANDOCERR_FI_TAB,
 215                     n->line, n->pos + (int)(p - cp), NULL);
 216 }
 217
 218 static void
 219 post_EE(CHKARGS)
 220 {
 221         if ((n->flags & NODE_NOFILL) == 0)
 222                 mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE");
 223 }
 224
 225 static void
 226 post_EX(CHKARGS)
 227 {
 228         if (n->flags & NODE_NOFILL)
 229                 mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX");
 230 }
 231
 232 static void
 233 post_OP(CHKARGS)
 234 {
 235
 236         if (n->child == NULL)
 237                 mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP");
 238         else if (n->child->next != NULL && n->child->next->next != NULL) {
 239                 n = n->child->next->next;
 240                 mandoc_msg(MANDOCERR_ARG_EXCESS,
 241                     n->line, n->pos, "OP ... %s", n->string);
 242         }
 243 }
 244
 245 static void
 246 post_SH(CHKARGS)
 247 {
 248         struct roff_node        *nc;
 249
 250         if (n->type != ROFFT_BODY || (nc = n->child) == NULL)
 251                 return;
 252
 253         if (nc->tok == MAN_PP && nc->body->child != NULL) {
 254                 while (nc->body->last != NULL) {
 255                         man->next = ROFF_NEXT_CHILD;
 256                         roff_node_relink(man, nc->body->last);
 257                         man->last = n;
 258                 }
 259         }
 260
 261         if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) {
 262                 mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos,
 263                     "%s after %s", roff_name[nc->tok], roff_name[n->tok]);
 264                 roff_node_delete(man, nc);
 265         }
 266
 267         /*
 268          * Trailing PP is empty, so it is deleted by check_par().
 269          * Trailing sp is significant.
 270          */
 271
 272         if ((nc = n->last) != NULL && nc->tok == ROFF_br) {
 273                 mandoc_msg(MANDOCERR_PAR_SKIP,
 274                     nc->line, nc->pos, "%s at the end of %s",
 275                     roff_name[nc->tok], roff_name[n->tok]);
 276                 roff_node_delete(man, nc);
 277         }
 278 }
 279
 280 static void
 281 post_UR(CHKARGS)
 282 {
 283         if (n->type == ROFFT_HEAD && n->child == NULL)
 284                 mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos,
 285                     "%s", roff_name[n->tok]);
 286         check_part(man, n);
 287 }
 288
 289 static void
 290 check_part(CHKARGS)
 291 {
 292
 293         if (n->type == ROFFT_BODY && n->child == NULL)
 294                 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos,
 295                     "%s", roff_name[n->tok]);
 296 }
 297
 298 static void
 299 check_par(CHKARGS)
 300 {
 301
 302         switch (n->type) {
 303         case ROFFT_BLOCK:
 304                 if (n->body->child == NULL)
 305                         roff_node_delete(man, n);
 306                 break;
 307         case ROFFT_BODY:
 308                 if (n->child != NULL &&
 309                     (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) {
 310                         mandoc_msg(MANDOCERR_PAR_SKIP,
 311                             n->child->line, n->child->pos,
 312                             "%s after %s", roff_name[n->child->tok],
 313                             roff_name[n->tok]);
 314                         roff_node_delete(man, n->child);
 315                 }
 316                 if (n->child == NULL)
 317                         mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
 318                             "%s empty", roff_name[n->tok]);
 319                 break;
 320         case ROFFT_HEAD:
 321                 if (n->child != NULL)
 322                         mandoc_msg(MANDOCERR_ARG_SKIP,
 323                             n->line, n->pos, "%s %s%s",
 324                             roff_name[n->tok], n->child->string,
 325                             n->child->next != NULL ? " ..." : "");
 326                 break;
 327         default:
 328                 break;
 329         }
 330 }
 331
 332 static void
 333 post_IP(CHKARGS)
 334 {
 335
 336         switch (n->type) {
 337         case ROFFT_BLOCK:
 338                 if (n->head->child == NULL && n->body->child == NULL)
 339                         roff_node_delete(man, n);
 340                 break;
 341         case ROFFT_BODY:
 342                 if (n->parent->head->child == NULL && n->child == NULL)
 343                         mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos,
 344                             "%s empty", roff_name[n->tok]);
 345                 break;
 346         default:
 347                 break;
 348         }
 349 }
 350
 351 static void
 352 post_TH(CHKARGS)
 353 {
 354         struct roff_node *nb;
 355         const char      *p;
 356
 357         free(man->meta.title);
 358         free(man->meta.vol);
 359         free(man->meta.os);
 360         free(man->meta.msec);
 361         free(man->meta.date);
 362
 363         man->meta.title = man->meta.vol = man->meta.date =
 364             man->meta.msec = man->meta.os = NULL;
 365
 366         nb = n;
 367
 368         /* ->TITLE<- MSEC DATE OS VOL */
 369
 370         n = n->child;
 371         if (n != NULL && n->string != NULL) {
 372                 for (p = n->string; *p != '\0'; p++) {
 373                         /* Only warn about this once... */
 374                         if (isalpha((unsigned char)*p) &&
 375                             ! isupper((unsigned char)*p)) {
 376                                 mandoc_msg(MANDOCERR_TITLE_CASE, n->line,
 377                                     n->pos + (int)(p - n->string),
 378                                     "TH %s", n->string);
 379                                 break;
 380                         }
 381                 }
 382                 man->meta.title = mandoc_strdup(n->string);
 383         } else {
 384                 man->meta.title = mandoc_strdup("");
 385                 mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH");
 386         }
 387
 388         /* TITLE ->MSEC<- DATE OS VOL */
 389
 390         if (n != NULL)
 391                 n = n->next;
 392         if (n != NULL && n->string != NULL)
 393                 man->meta.msec = mandoc_strdup(n->string);
 394         else {
 395                 man->meta.msec = mandoc_strdup("");
 396                 mandoc_msg(MANDOCERR_MSEC_MISSING,
 397                     nb->line, nb->pos, "TH %s", man->meta.title);
 398         }
 399
 400         /* TITLE MSEC ->DATE<- OS VOL */
 401
 402         if (n != NULL)
 403                 n = n->next;
 404         if (n != NULL && n->string != NULL && n->string[0] != '\0')
 405                 man->meta.date = mandoc_normdate(man,
 406                     n->string, n->line, n->pos);
 407         else {
 408                 man->meta.date = mandoc_strdup("");
 409                 mandoc_msg(MANDOCERR_DATE_MISSING,
 410                     n == NULL ? nb->line : n->line,
 411                     n == NULL ? nb->pos : n->pos, "TH");
 412         }
 413
 414         /* TITLE MSEC DATE ->OS<- VOL */
 415
 416         if (n && (n = n->next))
 417                 man->meta.os = mandoc_strdup(n->string);
 418         else if (man->os_s != NULL)
 419                 man->meta.os = mandoc_strdup(man->os_s);
 420         if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) {
 421                 if (strstr(man->meta.os, "OpenBSD") != NULL)
 422                         man->meta.os_e = MANDOC_OS_OPENBSD;
 423                 else if (strstr(man->meta.os, "NetBSD") != NULL)
 424                         man->meta.os_e = MANDOC_OS_NETBSD;
 425         }
 426
 427         /* TITLE MSEC DATE OS ->VOL<- */
 428         /* If missing, use the default VOL name for MSEC. */
 429
 430         if (n && (n = n->next))
 431                 man->meta.vol = mandoc_strdup(n->string);
 432         else if ('\0' != man->meta.msec[0] &&
 433             (NULL != (p = mandoc_a2msec(man->meta.msec))))
 434                 man->meta.vol = mandoc_strdup(p);
 435
 436         if (n != NULL && (n = n->next) != NULL)
 437                 mandoc_msg(MANDOCERR_ARG_EXCESS,
 438                     n->line, n->pos, "TH ... %s", n->string);
 439
 440         /*
 441          * Remove the `TH' node after we've processed it for our
 442          * meta-data.
 443          */
 444         roff_node_delete(man, man->last);
 445 }
 446
 447 static void
 448 post_UC(CHKARGS)
 449 {
 450         static const char * const bsd_versions[] = {
 451             "3rd Berkeley Distribution",
 452             "4th Berkeley Distribution",
 453             "4.2 Berkeley Distribution",
 454             "4.3 Berkeley Distribution",
 455             "4.4 Berkeley Distribution",
 456         };
 457
 458         const char      *p, *s;
 459
 460         n = n->child;
 461
 462         if (n == NULL || n->type != ROFFT_TEXT)
 463                 p = bsd_versions[0];
 464         else {
 465                 s = n->string;
 466                 if (0 == strcmp(s, "3"))
 467                         p = bsd_versions[0];
 468                 else if (0 == strcmp(s, "4"))
 469                         p = bsd_versions[1];
 470                 else if (0 == strcmp(s, "5"))
 471                         p = bsd_versions[2];
 472                 else if (0 == strcmp(s, "6"))
 473                         p = bsd_versions[3];
 474                 else if (0 == strcmp(s, "7"))
 475                         p = bsd_versions[4];
 476                 else
 477                         p = bsd_versions[0];
 478         }
 479
 480         free(man->meta.os);
 481         man->meta.os = mandoc_strdup(p);
 482 }
 483
 484 static void
 485 post_AT(CHKARGS)
 486 {
 487         static const char * const unix_versions[] = {
 488             "7th Edition",
 489             "System III",
 490             "System V",
 491             "System V Release 2",
 492         };
 493
 494         struct roff_node *nn;
 495         const char      *p, *s;
 496
 497         n = n->child;
 498
 499         if (n == NULL || n->type != ROFFT_TEXT)
 500                 p = unix_versions[0];
 501         else {
 502                 s = n->string;
 503                 if (0 == strcmp(s, "3"))
 504                         p = unix_versions[0];
 505                 else if (0 == strcmp(s, "4"))
 506                         p = unix_versions[1];
 507                 else if (0 == strcmp(s, "5")) {
 508                         nn = n->next;
 509                         if (nn != NULL &&
 510                             nn->type == ROFFT_TEXT &&
 511                             nn->string[0] != '\0')
 512                                 p = unix_versions[3];
 513                         else
 514                                 p = unix_versions[2];
 515                 } else
 516                         p = unix_versions[0];
 517         }
 518
 519         free(man->meta.os);
 520         man->meta.os = mandoc_strdup(p);
 521 }
 522
 523 static void
 524 post_in(CHKARGS)
 525 {
 526         char    *s;
 527
 528         if (n->parent->tok != MAN_TP ||
 529             n->parent->type != ROFFT_HEAD ||
 530             n->child == NULL ||
 531             *n->child->string == '+' ||
 532             *n->child->string == '-')
 533                 return;
 534         mandoc_asprintf(&s, "+%s", n->child->string);
 535         free(n->child->string);
 536         n->child->string = s;
 537 }