1/* $Id: mdoc_validate.c,v 1.374 2019/06/27 15:07:30 schwarze Exp $ */ 2/* 3 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv> 4 * Copyright (c) 2010-2019 Ingo Schwarze <schwarze@openbsd.org> 5 * Copyright (c) 2010 Joerg Sonnenberger <joerg@netbsd.org> 6 * 7 * Permission to use, copy, modify, and distribute this software for any 8 * purpose with or without fee is hereby granted, provided that the above 9 * copyright notice and this permission notice appear in all copies. 10 * 11 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES 12 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 13 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR 14 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 15 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 16 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 17 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 18 */ 19#include "config.h" 20 21#include <sys/types.h> 22#ifndef OSNAME 23#include <sys/utsname.h> 24#endif 25 26#include <assert.h> 27#include <ctype.h> 28#include <limits.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 "mandoc_xr.h" 37#include "roff.h" 38#include "mdoc.h" 39#include "libmandoc.h" 40#include "roff_int.h" 41#include "libmdoc.h" 42 43/* FIXME: .Bl -diag can't have non-text children in HEAD. */ 44 45#define POST_ARGS struct roff_man *mdoc 46 47enum check_ineq { 48 CHECK_LT, 49 CHECK_GT, 50 CHECK_EQ 51}; 52 53typedef void (*v_post)(POST_ARGS); 54 55static int build_list(struct roff_man *, int); 56static void check_argv(struct roff_man *, 57 struct roff_node *, struct mdoc_argv *); 58static void check_args(struct roff_man *, struct roff_node *); 59static void check_text(struct roff_man *, int, int, char *); 60static void check_text_em(struct roff_man *, int, int, char *); 61static void check_toptext(struct roff_man *, int, int, const char *); 62static int child_an(const struct roff_node *); 63static size_t macro2len(enum roff_tok); 64static void rewrite_macro2len(struct roff_man *, char **); 65static int similar(const char *, const char *); 66 67static void post_abort(POST_ARGS) __attribute__((__noreturn__)); 68static void post_an(POST_ARGS); 69static void post_an_norm(POST_ARGS); 70static void post_at(POST_ARGS); 71static void post_bd(POST_ARGS); 72static void post_bf(POST_ARGS); 73static void post_bk(POST_ARGS); 74static void post_bl(POST_ARGS); 75static void post_bl_block(POST_ARGS); 76static void post_bl_head(POST_ARGS); 77static void post_bl_norm(POST_ARGS); 78static void post_bx(POST_ARGS); 79static void post_defaults(POST_ARGS); 80static void post_display(POST_ARGS); 81static void post_dd(POST_ARGS); 82static void post_delim(POST_ARGS); 83static void post_delim_nb(POST_ARGS); 84static void post_dt(POST_ARGS); 85static void post_en(POST_ARGS); 86static void post_es(POST_ARGS); 87static void post_eoln(POST_ARGS); 88static void post_ex(POST_ARGS); 89static void post_fa(POST_ARGS); 90static void post_fn(POST_ARGS); 91static void post_fname(POST_ARGS); 92static void post_fo(POST_ARGS); 93static void post_hyph(POST_ARGS); 94static void post_ignpar(POST_ARGS); 95static void post_it(POST_ARGS); 96static void post_lb(POST_ARGS); 97static void post_nd(POST_ARGS); 98static void post_nm(POST_ARGS); 99static void post_ns(POST_ARGS); 100static void post_obsolete(POST_ARGS); 101static void post_os(POST_ARGS); 102static void post_par(POST_ARGS); 103static void post_prevpar(POST_ARGS); 104static void post_root(POST_ARGS); 105static void post_rs(POST_ARGS); 106static void post_rv(POST_ARGS); 107static void post_sh(POST_ARGS); 108static void post_sh_head(POST_ARGS); 109static void post_sh_name(POST_ARGS); 110static void post_sh_see_also(POST_ARGS); 111static void post_sh_authors(POST_ARGS); 112static void post_sm(POST_ARGS); 113static void post_st(POST_ARGS); 114static void post_std(POST_ARGS); 115static void post_sx(POST_ARGS); 116static void post_useless(POST_ARGS); 117static void post_xr(POST_ARGS); 118static void post_xx(POST_ARGS); 119 120static const v_post mdoc_valids[MDOC_MAX - MDOC_Dd] = { 121 post_dd, /* Dd */ 122 post_dt, /* Dt */ 123 post_os, /* Os */ 124 post_sh, /* Sh */ 125 post_ignpar, /* Ss */ 126 post_par, /* Pp */ 127 post_display, /* D1 */ 128 post_display, /* Dl */ 129 post_display, /* Bd */ 130 NULL, /* Ed */ 131 post_bl, /* Bl */ 132 NULL, /* El */ 133 post_it, /* It */ 134 post_delim_nb, /* Ad */ 135 post_an, /* An */ 136 NULL, /* Ap */ 137 post_defaults, /* Ar */ 138 NULL, /* Cd */ 139 post_delim_nb, /* Cm */ 140 post_delim_nb, /* Dv */ 141 post_delim_nb, /* Er */ 142 post_delim_nb, /* Ev */ 143 post_ex, /* Ex */ 144 post_fa, /* Fa */ 145 NULL, /* Fd */ 146 post_delim_nb, /* Fl */ 147 post_fn, /* Fn */ 148 post_delim_nb, /* Ft */ 149 post_delim_nb, /* Ic */ 150 post_delim_nb, /* In */ 151 post_defaults, /* Li */ 152 post_nd, /* Nd */ 153 post_nm, /* Nm */ 154 post_delim_nb, /* Op */ 155 post_abort, /* Ot */ 156 post_defaults, /* Pa */ 157 post_rv, /* Rv */ 158 post_st, /* St */ 159 post_delim_nb, /* Va */ 160 post_delim_nb, /* Vt */ 161 post_xr, /* Xr */ 162 NULL, /* %A */ 163 post_hyph, /* %B */ /* FIXME: can be used outside Rs/Re. */ 164 NULL, /* %D */ 165 NULL, /* %I */ 166 NULL, /* %J */ 167 post_hyph, /* %N */ 168 post_hyph, /* %O */ 169 NULL, /* %P */ 170 post_hyph, /* %R */ 171 post_hyph, /* %T */ /* FIXME: can be used outside Rs/Re. */ 172 NULL, /* %V */ 173 NULL, /* Ac */ 174 NULL, /* Ao */ 175 post_delim_nb, /* Aq */ 176 post_at, /* At */ 177 NULL, /* Bc */ 178 post_bf, /* Bf */ 179 NULL, /* Bo */ 180 NULL, /* Bq */ 181 post_xx, /* Bsx */ 182 post_bx, /* Bx */ 183 post_obsolete, /* Db */ 184 NULL, /* Dc */ 185 NULL, /* Do */ 186 NULL, /* Dq */ 187 NULL, /* Ec */ 188 NULL, /* Ef */ 189 post_delim_nb, /* Em */ 190 NULL, /* Eo */ 191 post_xx, /* Fx */ 192 post_delim_nb, /* Ms */ 193 NULL, /* No */ 194 post_ns, /* Ns */ 195 post_xx, /* Nx */ 196 post_xx, /* Ox */ 197 NULL, /* Pc */ 198 NULL, /* Pf */ 199 NULL, /* Po */ 200 post_delim_nb, /* Pq */ 201 NULL, /* Qc */ 202 post_delim_nb, /* Ql */ 203 NULL, /* Qo */ 204 post_delim_nb, /* Qq */ 205 NULL, /* Re */ 206 post_rs, /* Rs */ 207 NULL, /* Sc */ 208 NULL, /* So */ 209 post_delim_nb, /* Sq */ 210 post_sm, /* Sm */ 211 post_sx, /* Sx */ 212 post_delim_nb, /* Sy */ 213 post_useless, /* Tn */ 214 post_xx, /* Ux */ 215 NULL, /* Xc */ 216 NULL, /* Xo */ 217 post_fo, /* Fo */ 218 NULL, /* Fc */ 219 NULL, /* Oo */ 220 NULL, /* Oc */ 221 post_bk, /* Bk */ 222 NULL, /* Ek */ 223 post_eoln, /* Bt */ 224 post_obsolete, /* Hf */ 225 post_obsolete, /* Fr */ 226 post_eoln, /* Ud */ 227 post_lb, /* Lb */ 228 post_abort, /* Lp */ 229 post_delim_nb, /* Lk */ 230 post_defaults, /* Mt */ 231 post_delim_nb, /* Brq */ 232 NULL, /* Bro */ 233 NULL, /* Brc */ 234 NULL, /* %C */ 235 post_es, /* Es */ 236 post_en, /* En */ 237 post_xx, /* Dx */ 238 NULL, /* %Q */ 239 NULL, /* %U */ 240 NULL, /* Ta */ 241}; 242 243#define RSORD_MAX 14 /* Number of `Rs' blocks. */ 244 245static const enum roff_tok rsord[RSORD_MAX] = { 246 MDOC__A, 247 MDOC__T, 248 MDOC__B, 249 MDOC__I, 250 MDOC__J, 251 MDOC__R, 252 MDOC__N, 253 MDOC__V, 254 MDOC__U, 255 MDOC__P, 256 MDOC__Q, 257 MDOC__C, 258 MDOC__D, 259 MDOC__O 260}; 261 262static const char * const secnames[SEC__MAX] = { 263 NULL, 264 "NAME", 265 "LIBRARY", 266 "SYNOPSIS", 267 "DESCRIPTION", 268 "CONTEXT", 269 "IMPLEMENTATION NOTES", 270 "RETURN VALUES", 271 "ENVIRONMENT", 272 "FILES", 273 "EXIT STATUS", 274 "EXAMPLES", 275 "DIAGNOSTICS", 276 "COMPATIBILITY", 277 "ERRORS", 278 "SEE ALSO", 279 "STANDARDS", 280 "HISTORY", 281 "AUTHORS", 282 "CAVEATS", 283 "BUGS", 284 "SECURITY CONSIDERATIONS", 285 NULL 286}; 287 288 289/* Validate the subtree rooted at mdoc->last. */ 290void 291mdoc_validate(struct roff_man *mdoc) 292{ 293 struct roff_node *n, *np; 294 const v_post *p; 295 296 /* 297 * Translate obsolete macros to modern macros first 298 * such that later code does not need to look 299 * for the obsolete versions. 300 */ 301 302 n = mdoc->last; 303 switch (n->tok) { 304 case MDOC_Lp: 305 n->tok = MDOC_Pp; 306 break; 307 case MDOC_Ot: 308 post_obsolete(mdoc); 309 n->tok = MDOC_Ft; 310 break; 311 default: 312 break; 313 } 314 315 /* 316 * Iterate over all children, recursing into each one 317 * in turn, depth-first. 318 */ 319 320 mdoc->last = mdoc->last->child; 321 while (mdoc->last != NULL) { 322 mdoc_validate(mdoc); 323 if (mdoc->last == n) 324 mdoc->last = mdoc->last->child; 325 else 326 mdoc->last = mdoc->last->next; 327 } 328 329 /* Finally validate the macro itself. */ 330 331 mdoc->last = n; 332 mdoc->next = ROFF_NEXT_SIBLING; 333 switch (n->type) { 334 case ROFFT_TEXT: 335 np = n->parent; 336 if (n->sec != SEC_SYNOPSIS || 337 (np->tok != MDOC_Cd && np->tok != MDOC_Fd)) 338 check_text(mdoc, n->line, n->pos, n->string); 339 if ((n->flags & NODE_NOFILL) == 0 && 340 (np->tok != MDOC_It || np->type != ROFFT_HEAD || 341 np->parent->parent->norm->Bl.type != LIST_diag)) 342 check_text_em(mdoc, n->line, n->pos, n->string); 343 if (np->tok == MDOC_It || (np->type == ROFFT_BODY && 344 (np->tok == MDOC_Sh || np->tok == MDOC_Ss))) 345 check_toptext(mdoc, n->line, n->pos, n->string); 346 break; 347 case ROFFT_COMMENT: 348 case ROFFT_EQN: 349 case ROFFT_TBL: 350 break; 351 case ROFFT_ROOT: 352 post_root(mdoc); 353 break; 354 default: 355 check_args(mdoc, mdoc->last); 356 357 /* 358 * Closing delimiters are not special at the 359 * beginning of a block, opening delimiters 360 * are not special at the end. 361 */ 362 363 if (n->child != NULL) 364 n->child->flags &= ~NODE_DELIMC; 365 if (n->last != NULL) 366 n->last->flags &= ~NODE_DELIMO; 367 368 /* Call the macro's postprocessor. */ 369 370 if (n->tok < ROFF_MAX) { 371 roff_validate(mdoc); 372 break; 373 } 374 375 assert(n->tok >= MDOC_Dd && n->tok < MDOC_MAX); 376 p = mdoc_valids + (n->tok - MDOC_Dd); 377 if (*p) 378 (*p)(mdoc); 379 if (mdoc->last == n) 380 mdoc_state(mdoc, n); 381 break; 382 } 383} 384 385static void 386check_args(struct roff_man *mdoc, struct roff_node *n) 387{ 388 int i; 389 390 if (NULL == n->args) 391 return; 392 393 assert(n->args->argc); 394 for (i = 0; i < (int)n->args->argc; i++) 395 check_argv(mdoc, n, &n->args->argv[i]); 396} 397 398static void 399check_argv(struct roff_man *mdoc, struct roff_node *n, struct mdoc_argv *v) 400{ 401 int i; 402 403 for (i = 0; i < (int)v->sz; i++) 404 check_text(mdoc, v->line, v->pos, v->value[i]); 405} 406 407static void 408check_text(struct roff_man *mdoc, int ln, int pos, char *p) 409{ 410 char *cp; 411 412 if (mdoc->last->flags & NODE_NOFILL) 413 return; 414 415 for (cp = p; NULL != (p = strchr(p, '\t')); p++) 416 mandoc_msg(MANDOCERR_FI_TAB, ln, pos + (int)(p - cp), NULL); 417} 418 419static void 420check_text_em(struct roff_man *mdoc, int ln, int pos, char *p) 421{ 422 const struct roff_node *np, *nn; 423 char *cp; 424 425 np = mdoc->last->prev; 426 nn = mdoc->last->next; 427 428 /* Look for em-dashes wrongly encoded as "--". */ 429 430 for (cp = p; *cp != '\0'; cp++) { 431 if (cp[0] != '-' || cp[1] != '-') 432 continue; 433 cp++; 434 435 /* Skip input sequences of more than two '-'. */ 436 437 if (cp[1] == '-') { 438 while (cp[1] == '-') 439 cp++; 440 continue; 441 } 442 443 /* Skip "--" directly attached to something else. */ 444 445 if ((cp - p > 1 && cp[-2] != ' ') || 446 (cp[1] != '\0' && cp[1] != ' ')) 447 continue; 448 449 /* Require a letter right before or right afterwards. */ 450 451 if ((cp - p > 2 ? 452 isalpha((unsigned char)cp[-3]) : 453 np != NULL && 454 np->type == ROFFT_TEXT && 455 *np->string != '\0' && 456 isalpha((unsigned char)np->string[ 457 strlen(np->string) - 1])) || 458 (cp[1] != '\0' && cp[2] != '\0' ? 459 isalpha((unsigned char)cp[2]) : 460 nn != NULL && 461 nn->type == ROFFT_TEXT && 462 isalpha((unsigned char)*nn->string))) { 463 mandoc_msg(MANDOCERR_DASHDASH, 464 ln, pos + (int)(cp - p) - 1, NULL); 465 break; 466 } 467 } 468} 469 470static void 471check_toptext(struct roff_man *mdoc, int ln, int pos, const char *p) 472{ 473 const char *cp, *cpr; 474 475 if (*p == '\0') 476 return; 477 478 if ((cp = strstr(p, "OpenBSD")) != NULL) 479 mandoc_msg(MANDOCERR_BX, ln, pos + (int)(cp - p), "Ox"); 480 if ((cp = strstr(p, "NetBSD")) != NULL) 481 mandoc_msg(MANDOCERR_BX, ln, pos + (int)(cp - p), "Nx"); 482 if ((cp = strstr(p, "FreeBSD")) != NULL) 483 mandoc_msg(MANDOCERR_BX, ln, pos + (int)(cp - p), "Fx"); 484 if ((cp = strstr(p, "DragonFly")) != NULL) 485 mandoc_msg(MANDOCERR_BX, ln, pos + (int)(cp - p), "Dx"); 486 487 cp = p; 488 while ((cp = strstr(cp + 1, "()")) != NULL) { 489 for (cpr = cp - 1; cpr >= p; cpr--) 490 if (*cpr != '_' && !isalnum((unsigned char)*cpr)) 491 break; 492 if ((cpr < p || *cpr == ' ') && cpr + 1 < cp) { 493 cpr++; 494 mandoc_msg(MANDOCERR_FUNC, ln, pos + (int)(cpr - p), 495 "%.*s()", (int)(cp - cpr), cpr); 496 } 497 } 498} 499 500static void 501post_abort(POST_ARGS) 502{ 503 abort(); 504} 505 506static void 507post_delim(POST_ARGS) 508{ 509 const struct roff_node *nch; 510 const char *lc; 511 enum mdelim delim; 512 enum roff_tok tok; 513 514 tok = mdoc->last->tok; 515 nch = mdoc->last->last; 516 if (nch == NULL || nch->type != ROFFT_TEXT) 517 return; 518 lc = strchr(nch->string, '\0') - 1; 519 if (lc < nch->string) 520 return; 521 delim = mdoc_isdelim(lc); 522 if (delim == DELIM_NONE || delim == DELIM_OPEN) 523 return; 524 if (*lc == ')' && (tok == MDOC_Nd || tok == MDOC_Sh || 525 tok == MDOC_Ss || tok == MDOC_Fo)) 526 return; 527 528 mandoc_msg(MANDOCERR_DELIM, nch->line, 529 nch->pos + (int)(lc - nch->string), "%s%s %s", roff_name[tok], 530 nch == mdoc->last->child ? "" : " ...", nch->string); 531} 532 533static void 534post_delim_nb(POST_ARGS) 535{ 536 const struct roff_node *nch; 537 const char *lc, *cp; 538 int nw; 539 enum mdelim delim; 540 enum roff_tok tok; 541 542 /* 543 * Find candidates: at least two bytes, 544 * the last one a closing or middle delimiter. 545 */ 546 547 tok = mdoc->last->tok; 548 nch = mdoc->last->last; 549 if (nch == NULL || nch->type != ROFFT_TEXT) 550 return; 551 lc = strchr(nch->string, '\0') - 1; 552 if (lc <= nch->string) 553 return; 554 delim = mdoc_isdelim(lc); 555 if (delim == DELIM_NONE || delim == DELIM_OPEN) 556 return; 557 558 /* 559 * Reduce false positives by allowing various cases. 560 */ 561 562 /* Escaped delimiters. */ 563 if (lc > nch->string + 1 && lc[-2] == '\\' && 564 (lc[-1] == '&' || lc[-1] == 'e')) 565 return; 566 567 /* Specific byte sequences. */ 568 switch (*lc) { 569 case ')': 570 for (cp = lc; cp >= nch->string; cp--) 571 if (*cp == '(') 572 return; 573 break; 574 case '.': 575 if (lc > nch->string + 1 && lc[-2] == '.' && lc[-1] == '.') 576 return; 577 if (lc[-1] == '.') 578 return; 579 break; 580 case ';': 581 if (tok == MDOC_Vt) 582 return; 583 break; 584 case '?': 585 if (lc[-1] == '?') 586 return; 587 break; 588 case ']': 589 for (cp = lc; cp >= nch->string; cp--) 590 if (*cp == '[') 591 return; 592 break; 593 case '|': 594 if (lc == nch->string + 1 && lc[-1] == '|') 595 return; 596 default: 597 break; 598 } 599 600 /* Exactly two non-alphanumeric bytes. */ 601 if (lc == nch->string + 1 && !isalnum((unsigned char)lc[-1])) 602 return; 603 604 /* At least three alphabetic words with a sentence ending. */ 605 if (strchr("!.:?", *lc) != NULL && (tok == MDOC_Em || 606 tok == MDOC_Li || tok == MDOC_Pq || tok == MDOC_Sy)) { 607 nw = 0; 608 for (cp = lc - 1; cp >= nch->string; cp--) { 609 if (*cp == ' ') { 610 nw++; 611 if (cp > nch->string && cp[-1] == ',') 612 cp--; 613 } else if (isalpha((unsigned int)*cp)) { 614 if (nw > 1) 615 return; 616 } else 617 break; 618 } 619 } 620 621 mandoc_msg(MANDOCERR_DELIM_NB, nch->line, 622 nch->pos + (int)(lc - nch->string), "%s%s %s", roff_name[tok], 623 nch == mdoc->last->child ? "" : " ...", nch->string); 624} 625 626static void 627post_bl_norm(POST_ARGS) 628{ 629 struct roff_node *n; 630 struct mdoc_argv *argv, *wa; 631 int i; 632 enum mdocargt mdoclt; 633 enum mdoc_list lt; 634 635 n = mdoc->last->parent; 636 n->norm->Bl.type = LIST__NONE; 637 638 /* 639 * First figure out which kind of list to use: bind ourselves to 640 * the first mentioned list type and warn about any remaining 641 * ones. If we find no list type, we default to LIST_item. 642 */ 643 644 wa = (n->args == NULL) ? NULL : n->args->argv; 645 mdoclt = MDOC_ARG_MAX; 646 for (i = 0; n->args && i < (int)n->args->argc; i++) { 647 argv = n->args->argv + i; 648 lt = LIST__NONE; 649 switch (argv->arg) { 650 /* Set list types. */ 651 case MDOC_Bullet: 652 lt = LIST_bullet; 653 break; 654 case MDOC_Dash: 655 lt = LIST_dash; 656 break; 657 case MDOC_Enum: 658 lt = LIST_enum; 659 break; 660 case MDOC_Hyphen: 661 lt = LIST_hyphen; 662 break; 663 case MDOC_Item: 664 lt = LIST_item; 665 break; 666 case MDOC_Tag: 667 lt = LIST_tag; 668 break; 669 case MDOC_Diag: 670 lt = LIST_diag; 671 break; 672 case MDOC_Hang: 673 lt = LIST_hang; 674 break; 675 case MDOC_Ohang: 676 lt = LIST_ohang; 677 break; 678 case MDOC_Inset: 679 lt = LIST_inset; 680 break; 681 case MDOC_Column: 682 lt = LIST_column; 683 break; 684 /* Set list arguments. */ 685 case MDOC_Compact: 686 if (n->norm->Bl.comp) 687 mandoc_msg(MANDOCERR_ARG_REP, 688 argv->line, argv->pos, "Bl -compact"); 689 n->norm->Bl.comp = 1; 690 break; 691 case MDOC_Width: 692 wa = argv; 693 if (0 == argv->sz) { 694 mandoc_msg(MANDOCERR_ARG_EMPTY, 695 argv->line, argv->pos, "Bl -width"); 696 n->norm->Bl.width = "0n"; 697 break; 698 } 699 if (NULL != n->norm->Bl.width) 700 mandoc_msg(MANDOCERR_ARG_REP, 701 argv->line, argv->pos, 702 "Bl -width %s", argv->value[0]); 703 rewrite_macro2len(mdoc, argv->value); 704 n->norm->Bl.width = argv->value[0]; 705 break; 706 case MDOC_Offset: 707 if (0 == argv->sz) { 708 mandoc_msg(MANDOCERR_ARG_EMPTY, 709 argv->line, argv->pos, "Bl -offset"); 710 break; 711 } 712 if (NULL != n->norm->Bl.offs) 713 mandoc_msg(MANDOCERR_ARG_REP, 714 argv->line, argv->pos, 715 "Bl -offset %s", argv->value[0]); 716 rewrite_macro2len(mdoc, argv->value); 717 n->norm->Bl.offs = argv->value[0]; 718 break; 719 default: 720 continue; 721 } 722 if (LIST__NONE == lt) 723 continue; 724 mdoclt = argv->arg; 725 726 /* Check: multiple list types. */ 727 728 if (LIST__NONE != n->norm->Bl.type) { 729 mandoc_msg(MANDOCERR_BL_REP, n->line, n->pos, 730 "Bl -%s", mdoc_argnames[argv->arg]); 731 continue; 732 } 733 734 /* The list type should come first. */ 735 736 if (n->norm->Bl.width || 737 n->norm->Bl.offs || 738 n->norm->Bl.comp) 739 mandoc_msg(MANDOCERR_BL_LATETYPE, 740 n->line, n->pos, "Bl -%s", 741 mdoc_argnames[n->args->argv[0].arg]); 742 743 n->norm->Bl.type = lt; 744 if (LIST_column == lt) { 745 n->norm->Bl.ncols = argv->sz; 746 n->norm->Bl.cols = (void *)argv->value; 747 } 748 } 749 750 /* Allow lists to default to LIST_item. */ 751 752 if (LIST__NONE == n->norm->Bl.type) { 753 mandoc_msg(MANDOCERR_BL_NOTYPE, n->line, n->pos, "Bl"); 754 n->norm->Bl.type = LIST_item; 755 mdoclt = MDOC_Item; 756 } 757 758 /* 759 * Validate the width field. Some list types don't need width 760 * types and should be warned about them. Others should have it 761 * and must also be warned. Yet others have a default and need 762 * no warning. 763 */ 764 765 switch (n->norm->Bl.type) { 766 case LIST_tag: 767 if (n->norm->Bl.width == NULL) 768 mandoc_msg(MANDOCERR_BL_NOWIDTH, 769 n->line, n->pos, "Bl -tag"); 770 break; 771 case LIST_column: 772 case LIST_diag: 773 case LIST_ohang: 774 case LIST_inset: 775 case LIST_item: 776 if (n->norm->Bl.width != NULL) 777 mandoc_msg(MANDOCERR_BL_SKIPW, wa->line, wa->pos, 778 "Bl -%s", mdoc_argnames[mdoclt]); 779 n->norm->Bl.width = NULL; 780 break; 781 case LIST_bullet: 782 case LIST_dash: 783 case LIST_hyphen: 784 if (n->norm->Bl.width == NULL) 785 n->norm->Bl.width = "2n"; 786 break; 787 case LIST_enum: 788 if (n->norm->Bl.width == NULL) 789 n->norm->Bl.width = "3n"; 790 break; 791 default: 792 break; 793 } 794} 795 796static void 797post_bd(POST_ARGS) 798{ 799 struct roff_node *n; 800 struct mdoc_argv *argv; 801 int i; 802 enum mdoc_disp dt; 803 804 n = mdoc->last; 805 for (i = 0; n->args && i < (int)n->args->argc; i++) { 806 argv = n->args->argv + i; 807 dt = DISP__NONE; 808 809 switch (argv->arg) { 810 case MDOC_Centred: 811 dt = DISP_centered; 812 break; 813 case MDOC_Ragged: 814 dt = DISP_ragged; 815 break; 816 case MDOC_Unfilled: 817 dt = DISP_unfilled; 818 break; 819 case MDOC_Filled: 820 dt = DISP_filled; 821 break; 822 case MDOC_Literal: 823 dt = DISP_literal; 824 break; 825 case MDOC_File: 826 mandoc_msg(MANDOCERR_BD_FILE, n->line, n->pos, NULL); 827 break; 828 case MDOC_Offset: 829 if (0 == argv->sz) { 830 mandoc_msg(MANDOCERR_ARG_EMPTY, 831 argv->line, argv->pos, "Bd -offset"); 832 break; 833 } 834 if (NULL != n->norm->Bd.offs) 835 mandoc_msg(MANDOCERR_ARG_REP, 836 argv->line, argv->pos, 837 "Bd -offset %s", argv->value[0]); 838 rewrite_macro2len(mdoc, argv->value); 839 n->norm->Bd.offs = argv->value[0]; 840 break; 841 case MDOC_Compact: 842 if (n->norm->Bd.comp) 843 mandoc_msg(MANDOCERR_ARG_REP, 844 argv->line, argv->pos, "Bd -compact"); 845 n->norm->Bd.comp = 1; 846 break; 847 default: 848 abort(); 849 } 850 if (DISP__NONE == dt) 851 continue; 852 853 if (DISP__NONE == n->norm->Bd.type) 854 n->norm->Bd.type = dt; 855 else 856 mandoc_msg(MANDOCERR_BD_REP, n->line, n->pos, 857 "Bd -%s", mdoc_argnames[argv->arg]); 858 } 859 860 if (DISP__NONE == n->norm->Bd.type) { 861 mandoc_msg(MANDOCERR_BD_NOTYPE, n->line, n->pos, "Bd"); 862 n->norm->Bd.type = DISP_ragged; 863 } 864} 865 866/* 867 * Stand-alone line macros. 868 */ 869 870static void 871post_an_norm(POST_ARGS) 872{ 873 struct roff_node *n; 874 struct mdoc_argv *argv; 875 size_t i; 876 877 n = mdoc->last; 878 if (n->args == NULL) 879 return; 880 881 for (i = 1; i < n->args->argc; i++) { 882 argv = n->args->argv + i; 883 mandoc_msg(MANDOCERR_AN_REP, argv->line, argv->pos, 884 "An -%s", mdoc_argnames[argv->arg]); 885 } 886 887 argv = n->args->argv; 888 if (argv->arg == MDOC_Split) 889 n->norm->An.auth = AUTH_split; 890 else if (argv->arg == MDOC_Nosplit) 891 n->norm->An.auth = AUTH_nosplit; 892 else 893 abort(); 894} 895 896static void 897post_eoln(POST_ARGS) 898{ 899 struct roff_node *n; 900 901 post_useless(mdoc); 902 n = mdoc->last; 903 if (n->child != NULL) 904 mandoc_msg(MANDOCERR_ARG_SKIP, n->line, 905 n->pos, "%s %s", roff_name[n->tok], n->child->string); 906 907 while (n->child != NULL) 908 roff_node_delete(mdoc, n->child); 909 910 roff_word_alloc(mdoc, n->line, n->pos, n->tok == MDOC_Bt ? 911 "is currently in beta test." : "currently under development."); 912 mdoc->last->flags |= NODE_EOS | NODE_NOSRC; 913 mdoc->last = n; 914} 915 916static int 917build_list(struct roff_man *mdoc, int tok) 918{ 919 struct roff_node *n; 920 int ic; 921 922 n = mdoc->last->next; 923 for (ic = 1;; ic++) { 924 roff_elem_alloc(mdoc, n->line, n->pos, tok); 925 mdoc->last->flags |= NODE_NOSRC; 926 roff_node_relink(mdoc, n); 927 n = mdoc->last = mdoc->last->parent; 928 mdoc->next = ROFF_NEXT_SIBLING; 929 if (n->next == NULL) 930 return ic; 931 if (ic > 1 || n->next->next != NULL) { 932 roff_word_alloc(mdoc, n->line, n->pos, ","); 933 mdoc->last->flags |= NODE_DELIMC | NODE_NOSRC; 934 } 935 n = mdoc->last->next; 936 if (n->next == NULL) { 937 roff_word_alloc(mdoc, n->line, n->pos, "and"); 938 mdoc->last->flags |= NODE_NOSRC; 939 } 940 } 941} 942 943static void 944post_ex(POST_ARGS) 945{ 946 struct roff_node *n; 947 int ic; 948 949 post_std(mdoc); 950 951 n = mdoc->last; 952 mdoc->next = ROFF_NEXT_CHILD; 953 roff_word_alloc(mdoc, n->line, n->pos, "The"); 954 mdoc->last->flags |= NODE_NOSRC; 955 956 if (mdoc->last->next != NULL) 957 ic = build_list(mdoc, MDOC_Nm); 958 else if (mdoc->meta.name != NULL) { 959 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Nm); 960 mdoc->last->flags |= NODE_NOSRC; 961 roff_word_alloc(mdoc, n->line, n->pos, mdoc->meta.name); 962 mdoc->last->flags |= NODE_NOSRC; 963 mdoc->last = mdoc->last->parent; 964 mdoc->next = ROFF_NEXT_SIBLING; 965 ic = 1; 966 } else { 967 mandoc_msg(MANDOCERR_EX_NONAME, n->line, n->pos, "Ex"); 968 ic = 0; 969 } 970 971 roff_word_alloc(mdoc, n->line, n->pos, 972 ic > 1 ? "utilities exit\\~0" : "utility exits\\~0"); 973 mdoc->last->flags |= NODE_NOSRC; 974 roff_word_alloc(mdoc, n->line, n->pos, 975 "on success, and\\~>0 if an error occurs."); 976 mdoc->last->flags |= NODE_EOS | NODE_NOSRC; 977 mdoc->last = n; 978} 979 980static void 981post_lb(POST_ARGS) 982{ 983 struct roff_node *n; 984 const char *p; 985 986 post_delim_nb(mdoc); 987 988 n = mdoc->last; 989 assert(n->child->type == ROFFT_TEXT); 990 mdoc->next = ROFF_NEXT_CHILD; 991 992 if ((p = mdoc_a2lib(n->child->string)) != NULL) { 993 n->child->flags |= NODE_NOPRT; 994 roff_word_alloc(mdoc, n->line, n->pos, p); 995 mdoc->last->flags = NODE_NOSRC; 996 mdoc->last = n; 997 return; 998 } 999 1000 mandoc_msg(MANDOCERR_LB_BAD, n->child->line, 1001 n->child->pos, "Lb %s", n->child->string); 1002 1003 roff_word_alloc(mdoc, n->line, n->pos, "library"); 1004 mdoc->last->flags = NODE_NOSRC; 1005 roff_word_alloc(mdoc, n->line, n->pos, "\\(lq"); 1006 mdoc->last->flags = NODE_DELIMO | NODE_NOSRC; 1007 mdoc->last = mdoc->last->next; 1008 roff_word_alloc(mdoc, n->line, n->pos, "\\(rq"); 1009 mdoc->last->flags = NODE_DELIMC | NODE_NOSRC; 1010 mdoc->last = n; 1011} 1012 1013static void 1014post_rv(POST_ARGS) 1015{ 1016 struct roff_node *n; 1017 int ic; 1018 1019 post_std(mdoc); 1020 1021 n = mdoc->last; 1022 mdoc->next = ROFF_NEXT_CHILD; 1023 if (n->child != NULL) { 1024 roff_word_alloc(mdoc, n->line, n->pos, "The"); 1025 mdoc->last->flags |= NODE_NOSRC; 1026 ic = build_list(mdoc, MDOC_Fn); 1027 roff_word_alloc(mdoc, n->line, n->pos, 1028 ic > 1 ? "functions return" : "function returns"); 1029 mdoc->last->flags |= NODE_NOSRC; 1030 roff_word_alloc(mdoc, n->line, n->pos, 1031 "the value\\~0 if successful;"); 1032 } else 1033 roff_word_alloc(mdoc, n->line, n->pos, "Upon successful " 1034 "completion, the value\\~0 is returned;"); 1035 mdoc->last->flags |= NODE_NOSRC; 1036 1037 roff_word_alloc(mdoc, n->line, n->pos, "otherwise " 1038 "the value\\~\\-1 is returned and the global variable"); 1039 mdoc->last->flags |= NODE_NOSRC; 1040 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Va); 1041 mdoc->last->flags |= NODE_NOSRC; 1042 roff_word_alloc(mdoc, n->line, n->pos, "errno"); 1043 mdoc->last->flags |= NODE_NOSRC; 1044 mdoc->last = mdoc->last->parent; 1045 mdoc->next = ROFF_NEXT_SIBLING; 1046 roff_word_alloc(mdoc, n->line, n->pos, 1047 "is set to indicate the error."); 1048 mdoc->last->flags |= NODE_EOS | NODE_NOSRC; 1049 mdoc->last = n; 1050} 1051 1052static void 1053post_std(POST_ARGS) 1054{ 1055 struct roff_node *n; 1056 1057 post_delim(mdoc); 1058 1059 n = mdoc->last; 1060 if (n->args && n->args->argc == 1) 1061 if (n->args->argv[0].arg == MDOC_Std) 1062 return; 1063 1064 mandoc_msg(MANDOCERR_ARG_STD, n->line, n->pos, 1065 "%s", roff_name[n->tok]); 1066} 1067 1068static void 1069post_st(POST_ARGS) 1070{ 1071 struct roff_node *n, *nch; 1072 const char *p; 1073 1074 n = mdoc->last; 1075 nch = n->child; 1076 assert(nch->type == ROFFT_TEXT); 1077 1078 if ((p = mdoc_a2st(nch->string)) == NULL) { 1079 mandoc_msg(MANDOCERR_ST_BAD, 1080 nch->line, nch->pos, "St %s", nch->string); 1081 roff_node_delete(mdoc, n); 1082 return; 1083 } 1084 1085 nch->flags |= NODE_NOPRT; 1086 mdoc->next = ROFF_NEXT_CHILD; 1087 roff_word_alloc(mdoc, nch->line, nch->pos, p); 1088 mdoc->last->flags |= NODE_NOSRC; 1089 mdoc->last= n; 1090} 1091 1092static void 1093post_obsolete(POST_ARGS) 1094{ 1095 struct roff_node *n; 1096 1097 n = mdoc->last; 1098 if (n->type == ROFFT_ELEM || n->type == ROFFT_BLOCK) 1099 mandoc_msg(MANDOCERR_MACRO_OBS, n->line, n->pos, 1100 "%s", roff_name[n->tok]); 1101} 1102 1103static void 1104post_useless(POST_ARGS) 1105{ 1106 struct roff_node *n; 1107 1108 n = mdoc->last; 1109 mandoc_msg(MANDOCERR_MACRO_USELESS, n->line, n->pos, 1110 "%s", roff_name[n->tok]); 1111} 1112 1113/* 1114 * Block macros. 1115 */ 1116 1117static void 1118post_bf(POST_ARGS) 1119{ 1120 struct roff_node *np, *nch; 1121 1122 /* 1123 * Unlike other data pointers, these are "housed" by the HEAD 1124 * element, which contains the goods. 1125 */ 1126 1127 np = mdoc->last; 1128 if (np->type != ROFFT_HEAD) 1129 return; 1130 1131 assert(np->parent->type == ROFFT_BLOCK); 1132 assert(np->parent->tok == MDOC_Bf); 1133 1134 /* Check the number of arguments. */ 1135 1136 nch = np->child; 1137 if (np->parent->args == NULL) { 1138 if (nch == NULL) { 1139 mandoc_msg(MANDOCERR_BF_NOFONT, 1140 np->line, np->pos, "Bf"); 1141 return; 1142 } 1143 nch = nch->next; 1144 } 1145 if (nch != NULL) 1146 mandoc_msg(MANDOCERR_ARG_EXCESS, 1147 nch->line, nch->pos, "Bf ... %s", nch->string); 1148 1149 /* Extract argument into data. */ 1150 1151 if (np->parent->args != NULL) { 1152 switch (np->parent->args->argv[0].arg) { 1153 case MDOC_Emphasis: 1154 np->norm->Bf.font = FONT_Em; 1155 break; 1156 case MDOC_Literal: 1157 np->norm->Bf.font = FONT_Li; 1158 break; 1159 case MDOC_Symbolic: 1160 np->norm->Bf.font = FONT_Sy; 1161 break; 1162 default: 1163 abort(); 1164 } 1165 return; 1166 } 1167 1168 /* Extract parameter into data. */ 1169 1170 if ( ! strcmp(np->child->string, "Em")) 1171 np->norm->Bf.font = FONT_Em; 1172 else if ( ! strcmp(np->child->string, "Li")) 1173 np->norm->Bf.font = FONT_Li; 1174 else if ( ! strcmp(np->child->string, "Sy")) 1175 np->norm->Bf.font = FONT_Sy; 1176 else 1177 mandoc_msg(MANDOCERR_BF_BADFONT, np->child->line, 1178 np->child->pos, "Bf %s", np->child->string); 1179} 1180 1181static void 1182post_fname(POST_ARGS) 1183{ 1184 const struct roff_node *n; 1185 const char *cp; 1186 size_t pos; 1187 1188 n = mdoc->last->child; 1189 pos = strcspn(n->string, "()"); 1190 cp = n->string + pos; 1191 if ( ! (cp[0] == '\0' || (cp[0] == '(' && cp[1] == '*'))) 1192 mandoc_msg(MANDOCERR_FN_PAREN, n->line, n->pos + pos, 1193 "%s", n->string); 1194} 1195 1196static void 1197post_fn(POST_ARGS) 1198{ 1199 1200 post_fname(mdoc); 1201 post_fa(mdoc); 1202} 1203 1204static void 1205post_fo(POST_ARGS) 1206{ 1207 const struct roff_node *n; 1208 1209 n = mdoc->last; 1210 1211 if (n->type != ROFFT_HEAD) 1212 return; 1213 1214 if (n->child == NULL) { 1215 mandoc_msg(MANDOCERR_FO_NOHEAD, n->line, n->pos, "Fo"); 1216 return; 1217 } 1218 if (n->child != n->last) { 1219 mandoc_msg(MANDOCERR_ARG_EXCESS, 1220 n->child->next->line, n->child->next->pos, 1221 "Fo ... %s", n->child->next->string); 1222 while (n->child != n->last) 1223 roff_node_delete(mdoc, n->last); 1224 } else 1225 post_delim(mdoc); 1226 1227 post_fname(mdoc); 1228} 1229 1230static void 1231post_fa(POST_ARGS) 1232{ 1233 const struct roff_node *n; 1234 const char *cp; 1235 1236 for (n = mdoc->last->child; n != NULL; n = n->next) { 1237 for (cp = n->string; *cp != '\0'; cp++) { 1238 /* Ignore callbacks and alterations. */ 1239 if (*cp == '(' || *cp == '{') 1240 break; 1241 if (*cp != ',') 1242 continue; 1243 mandoc_msg(MANDOCERR_FA_COMMA, n->line, 1244 n->pos + (int)(cp - n->string), "%s", n->string); 1245 break; 1246 } 1247 } 1248 post_delim_nb(mdoc); 1249} 1250 1251static void 1252post_nm(POST_ARGS) 1253{ 1254 struct roff_node *n; 1255 1256 n = mdoc->last; 1257 1258 if (n->sec == SEC_NAME && n->child != NULL && 1259 n->child->type == ROFFT_TEXT && mdoc->meta.msec != NULL) 1260 mandoc_xr_add(mdoc->meta.msec, n->child->string, -1, -1); 1261 1262 if (n->last != NULL && n->last->tok == MDOC_Pp) 1263 roff_node_relink(mdoc, n->last); 1264 1265 if (mdoc->meta.name == NULL) 1266 deroff(&mdoc->meta.name, n); 1267 1268 if (mdoc->meta.name == NULL || 1269 (mdoc->lastsec == SEC_NAME && n->child == NULL)) 1270 mandoc_msg(MANDOCERR_NM_NONAME, n->line, n->pos, "Nm"); 1271 1272 switch (n->type) { 1273 case ROFFT_ELEM: 1274 post_delim_nb(mdoc); 1275 break; 1276 case ROFFT_HEAD: 1277 post_delim(mdoc); 1278 break; 1279 default: 1280 return; 1281 } 1282 1283 if ((n->child != NULL && n->child->type == ROFFT_TEXT) || 1284 mdoc->meta.name == NULL) 1285 return; 1286 1287 mdoc->next = ROFF_NEXT_CHILD; 1288 roff_word_alloc(mdoc, n->line, n->pos, mdoc->meta.name); 1289 mdoc->last->flags |= NODE_NOSRC; 1290 mdoc->last = n; 1291} 1292 1293static void 1294post_nd(POST_ARGS) 1295{ 1296 struct roff_node *n; 1297 1298 n = mdoc->last; 1299 1300 if (n->type != ROFFT_BODY) 1301 return; 1302 1303 if (n->sec != SEC_NAME) 1304 mandoc_msg(MANDOCERR_ND_LATE, n->line, n->pos, "Nd"); 1305 1306 if (n->child == NULL) 1307 mandoc_msg(MANDOCERR_ND_EMPTY, n->line, n->pos, "Nd"); 1308 else 1309 post_delim(mdoc); 1310 1311 post_hyph(mdoc); 1312} 1313 1314static void 1315post_display(POST_ARGS) 1316{ 1317 struct roff_node *n, *np; 1318 1319 n = mdoc->last; 1320 switch (n->type) { 1321 case ROFFT_BODY: 1322 if (n->end != ENDBODY_NOT) { 1323 if (n->tok == MDOC_Bd && 1324 n->body->parent->args == NULL) 1325 roff_node_delete(mdoc, n); 1326 } else if (n->child == NULL) 1327 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos, 1328 "%s", roff_name[n->tok]); 1329 else if (n->tok == MDOC_D1) 1330 post_hyph(mdoc); 1331 break; 1332 case ROFFT_BLOCK: 1333 if (n->tok == MDOC_Bd) { 1334 if (n->args == NULL) { 1335 mandoc_msg(MANDOCERR_BD_NOARG, 1336 n->line, n->pos, "Bd"); 1337 mdoc->next = ROFF_NEXT_SIBLING; 1338 while (n->body->child != NULL) 1339 roff_node_relink(mdoc, 1340 n->body->child); 1341 roff_node_delete(mdoc, n); 1342 break; 1343 } 1344 post_bd(mdoc); 1345 post_prevpar(mdoc); 1346 } 1347 for (np = n->parent; np != NULL; np = np->parent) { 1348 if (np->type == ROFFT_BLOCK && np->tok == MDOC_Bd) { 1349 mandoc_msg(MANDOCERR_BD_NEST, n->line, 1350 n->pos, "%s in Bd", roff_name[n->tok]); 1351 break; 1352 } 1353 } 1354 break; 1355 default: 1356 break; 1357 } 1358} 1359 1360static void 1361post_defaults(POST_ARGS) 1362{ 1363 struct roff_node *nn; 1364 1365 if (mdoc->last->child != NULL) { 1366 post_delim_nb(mdoc); 1367 return; 1368 } 1369 1370 /* 1371 * The `Ar' defaults to "file ..." if no value is provided as an 1372 * argument; the `Mt' and `Pa' macros use "~"; the `Li' just 1373 * gets an empty string. 1374 */ 1375 1376 nn = mdoc->last; 1377 switch (nn->tok) { 1378 case MDOC_Ar: 1379 mdoc->next = ROFF_NEXT_CHILD; 1380 roff_word_alloc(mdoc, nn->line, nn->pos, "file"); 1381 mdoc->last->flags |= NODE_NOSRC; 1382 roff_word_alloc(mdoc, nn->line, nn->pos, "..."); 1383 mdoc->last->flags |= NODE_NOSRC; 1384 break; 1385 case MDOC_Pa: 1386 case MDOC_Mt: 1387 mdoc->next = ROFF_NEXT_CHILD; 1388 roff_word_alloc(mdoc, nn->line, nn->pos, "~"); 1389 mdoc->last->flags |= NODE_NOSRC; 1390 break; 1391 default: 1392 abort(); 1393 } 1394 mdoc->last = nn; 1395} 1396 1397static void 1398post_at(POST_ARGS) 1399{ 1400 struct roff_node *n, *nch; 1401 const char *att; 1402 1403 n = mdoc->last; 1404 nch = n->child; 1405 1406 /* 1407 * If we have a child, look it up in the standard keys. If a 1408 * key exist, use that instead of the child; if it doesn't, 1409 * prefix "AT&T UNIX " to the existing data. 1410 */ 1411 1412 att = NULL; 1413 if (nch != NULL && ((att = mdoc_a2att(nch->string)) == NULL)) 1414 mandoc_msg(MANDOCERR_AT_BAD, 1415 nch->line, nch->pos, "At %s", nch->string); 1416 1417 mdoc->next = ROFF_NEXT_CHILD; 1418 if (att != NULL) { 1419 roff_word_alloc(mdoc, nch->line, nch->pos, att); 1420 nch->flags |= NODE_NOPRT; 1421 } else 1422 roff_word_alloc(mdoc, n->line, n->pos, "AT&T UNIX"); 1423 mdoc->last->flags |= NODE_NOSRC; 1424 mdoc->last = n; 1425} 1426 1427static void 1428post_an(POST_ARGS) 1429{ 1430 struct roff_node *np, *nch; 1431 1432 post_an_norm(mdoc); 1433 1434 np = mdoc->last; 1435 nch = np->child; 1436 if (np->norm->An.auth == AUTH__NONE) { 1437 if (nch == NULL) 1438 mandoc_msg(MANDOCERR_MACRO_EMPTY, 1439 np->line, np->pos, "An"); 1440 else 1441 post_delim_nb(mdoc); 1442 } else if (nch != NULL) 1443 mandoc_msg(MANDOCERR_ARG_EXCESS, 1444 nch->line, nch->pos, "An ... %s", nch->string); 1445} 1446 1447static void 1448post_en(POST_ARGS) 1449{ 1450 1451 post_obsolete(mdoc); 1452 if (mdoc->last->type == ROFFT_BLOCK) 1453 mdoc->last->norm->Es = mdoc->last_es; 1454} 1455 1456static void 1457post_es(POST_ARGS) 1458{ 1459 1460 post_obsolete(mdoc); 1461 mdoc->last_es = mdoc->last; 1462} 1463 1464static void 1465post_xx(POST_ARGS) 1466{ 1467 struct roff_node *n; 1468 const char *os; 1469 char *v; 1470 1471 post_delim_nb(mdoc); 1472 1473 n = mdoc->last; 1474 switch (n->tok) { 1475 case MDOC_Bsx: 1476 os = "BSD/OS"; 1477 break; 1478 case MDOC_Dx: 1479 os = "DragonFly"; 1480 break; 1481 case MDOC_Fx: 1482 os = "FreeBSD"; 1483 break; 1484 case MDOC_Nx: 1485 os = "NetBSD"; 1486 if (n->child == NULL) 1487 break; 1488 v = n->child->string; 1489 if ((v[0] != '0' && v[0] != '1') || v[1] != '.' || 1490 v[2] < '0' || v[2] > '9' || 1491 v[3] < 'a' || v[3] > 'z' || v[4] != '\0') 1492 break; 1493 n->child->flags |= NODE_NOPRT; 1494 mdoc->next = ROFF_NEXT_CHILD; 1495 roff_word_alloc(mdoc, n->child->line, n->child->pos, v); 1496 v = mdoc->last->string; 1497 v[3] = toupper((unsigned char)v[3]); 1498 mdoc->last->flags |= NODE_NOSRC; 1499 mdoc->last = n; 1500 break; 1501 case MDOC_Ox: 1502 os = "OpenBSD"; 1503 break; 1504 case MDOC_Ux: 1505 os = "UNIX"; 1506 break; 1507 default: 1508 abort(); 1509 } 1510 mdoc->next = ROFF_NEXT_CHILD; 1511 roff_word_alloc(mdoc, n->line, n->pos, os); 1512 mdoc->last->flags |= NODE_NOSRC; 1513 mdoc->last = n; 1514} 1515 1516static void 1517post_it(POST_ARGS) 1518{ 1519 struct roff_node *nbl, *nit, *nch; 1520 int i, cols; 1521 enum mdoc_list lt; 1522 1523 post_prevpar(mdoc); 1524 1525 nit = mdoc->last; 1526 if (nit->type != ROFFT_BLOCK) 1527 return; 1528 1529 nbl = nit->parent->parent; 1530 lt = nbl->norm->Bl.type; 1531 1532 switch (lt) { 1533 case LIST_tag: 1534 case LIST_hang: 1535 case LIST_ohang: 1536 case LIST_inset: 1537 case LIST_diag: 1538 if (nit->head->child == NULL) 1539 mandoc_msg(MANDOCERR_IT_NOHEAD, 1540 nit->line, nit->pos, "Bl -%s It", 1541 mdoc_argnames[nbl->args->argv[0].arg]); 1542 break; 1543 case LIST_bullet: 1544 case LIST_dash: 1545 case LIST_enum: 1546 case LIST_hyphen: 1547 if (nit->body == NULL || nit->body->child == NULL) 1548 mandoc_msg(MANDOCERR_IT_NOBODY, 1549 nit->line, nit->pos, "Bl -%s It", 1550 mdoc_argnames[nbl->args->argv[0].arg]); 1551 /* FALLTHROUGH */ 1552 case LIST_item: 1553 if ((nch = nit->head->child) != NULL) 1554 mandoc_msg(MANDOCERR_ARG_SKIP, 1555 nit->line, nit->pos, "It %s", 1556 nch->string == NULL ? roff_name[nch->tok] : 1557 nch->string); 1558 break; 1559 case LIST_column: 1560 cols = (int)nbl->norm->Bl.ncols; 1561 1562 assert(nit->head->child == NULL); 1563 1564 if (nit->head->next->child == NULL && 1565 nit->head->next->next == NULL) { 1566 mandoc_msg(MANDOCERR_MACRO_EMPTY, 1567 nit->line, nit->pos, "It"); 1568 roff_node_delete(mdoc, nit); 1569 break; 1570 } 1571 1572 i = 0; 1573 for (nch = nit->child; nch != NULL; nch = nch->next) { 1574 if (nch->type != ROFFT_BODY) 1575 continue; 1576 if (i++ && nch->flags & NODE_LINE) 1577 mandoc_msg(MANDOCERR_TA_LINE, 1578 nch->line, nch->pos, "Ta"); 1579 } 1580 if (i < cols || i > cols + 1) 1581 mandoc_msg(MANDOCERR_BL_COL, nit->line, nit->pos, 1582 "%d columns, %d cells", cols, i); 1583 else if (nit->head->next->child != NULL && 1584 nit->head->next->child->flags & NODE_LINE) 1585 mandoc_msg(MANDOCERR_IT_NOARG, 1586 nit->line, nit->pos, "Bl -column It"); 1587 break; 1588 default: 1589 abort(); 1590 } 1591} 1592 1593static void 1594post_bl_block(POST_ARGS) 1595{ 1596 struct roff_node *n, *ni, *nc; 1597 1598 post_prevpar(mdoc); 1599 1600 n = mdoc->last; 1601 for (ni = n->body->child; ni != NULL; ni = ni->next) { 1602 if (ni->body == NULL) 1603 continue; 1604 nc = ni->body->last; 1605 while (nc != NULL) { 1606 switch (nc->tok) { 1607 case MDOC_Pp: 1608 case ROFF_br: 1609 break; 1610 default: 1611 nc = NULL; 1612 continue; 1613 } 1614 if (ni->next == NULL) { 1615 mandoc_msg(MANDOCERR_PAR_MOVE, nc->line, 1616 nc->pos, "%s", roff_name[nc->tok]); 1617 roff_node_relink(mdoc, nc); 1618 } else if (n->norm->Bl.comp == 0 && 1619 n->norm->Bl.type != LIST_column) { 1620 mandoc_msg(MANDOCERR_PAR_SKIP, 1621 nc->line, nc->pos, 1622 "%s before It", roff_name[nc->tok]); 1623 roff_node_delete(mdoc, nc); 1624 } else 1625 break; 1626 nc = ni->body->last; 1627 } 1628 } 1629} 1630 1631/* 1632 * If the argument of -offset or -width is a macro, 1633 * replace it with the associated default width. 1634 */ 1635static void 1636rewrite_macro2len(struct roff_man *mdoc, char **arg) 1637{ 1638 size_t width; 1639 enum roff_tok tok; 1640 1641 if (*arg == NULL) 1642 return; 1643 else if ( ! strcmp(*arg, "Ds")) 1644 width = 6; 1645 else if ((tok = roffhash_find(mdoc->mdocmac, *arg, 0)) == TOKEN_NONE) 1646 return; 1647 else 1648 width = macro2len(tok); 1649 1650 free(*arg); 1651 mandoc_asprintf(arg, "%zun", width); 1652} 1653 1654static void 1655post_bl_head(POST_ARGS) 1656{ 1657 struct roff_node *nbl, *nh, *nch, *nnext; 1658 struct mdoc_argv *argv; 1659 int i, j; 1660 1661 post_bl_norm(mdoc); 1662 1663 nh = mdoc->last; 1664 if (nh->norm->Bl.type != LIST_column) { 1665 if ((nch = nh->child) == NULL) 1666 return; 1667 mandoc_msg(MANDOCERR_ARG_EXCESS, 1668 nch->line, nch->pos, "Bl ... %s", nch->string); 1669 while (nch != NULL) { 1670 roff_node_delete(mdoc, nch); 1671 nch = nh->child; 1672 } 1673 return; 1674 } 1675 1676 /* 1677 * Append old-style lists, where the column width specifiers 1678 * trail as macro parameters, to the new-style ("normal-form") 1679 * lists where they're argument values following -column. 1680 */ 1681 1682 if (nh->child == NULL) 1683 return; 1684 1685 nbl = nh->parent; 1686 for (j = 0; j < (int)nbl->args->argc; j++) 1687 if (nbl->args->argv[j].arg == MDOC_Column) 1688 break; 1689 1690 assert(j < (int)nbl->args->argc); 1691 1692 /* 1693 * Accommodate for new-style groff column syntax. Shuffle the 1694 * child nodes, all of which must be TEXT, as arguments for the 1695 * column field. Then, delete the head children. 1696 */ 1697 1698 argv = nbl->args->argv + j; 1699 i = argv->sz; 1700 for (nch = nh->child; nch != NULL; nch = nch->next) 1701 argv->sz++; 1702 argv->value = mandoc_reallocarray(argv->value, 1703 argv->sz, sizeof(char *)); 1704 1705 nh->norm->Bl.ncols = argv->sz; 1706 nh->norm->Bl.cols = (void *)argv->value; 1707 1708 for (nch = nh->child; nch != NULL; nch = nnext) { 1709 argv->value[i++] = nch->string; 1710 nch->string = NULL; 1711 nnext = nch->next; 1712 roff_node_delete(NULL, nch); 1713 } 1714 nh->child = NULL; 1715} 1716 1717static void 1718post_bl(POST_ARGS) 1719{ 1720 struct roff_node *nparent, *nprev; /* of the Bl block */ 1721 struct roff_node *nblock, *nbody; /* of the Bl */ 1722 struct roff_node *nchild, *nnext; /* of the Bl body */ 1723 const char *prev_Er; 1724 int order; 1725 1726 nbody = mdoc->last; 1727 switch (nbody->type) { 1728 case ROFFT_BLOCK: 1729 post_bl_block(mdoc); 1730 return; 1731 case ROFFT_HEAD: 1732 post_bl_head(mdoc); 1733 return; 1734 case ROFFT_BODY: 1735 break; 1736 default: 1737 return; 1738 } 1739 if (nbody->end != ENDBODY_NOT) 1740 return; 1741 1742 nchild = nbody->child; 1743 if (nchild == NULL) { 1744 mandoc_msg(MANDOCERR_BLK_EMPTY, 1745 nbody->line, nbody->pos, "Bl"); 1746 return; 1747 } 1748 while (nchild != NULL) { 1749 nnext = nchild->next; 1750 if (nchild->tok == MDOC_It || 1751 (nchild->tok == MDOC_Sm && 1752 nnext != NULL && nnext->tok == MDOC_It)) { 1753 nchild = nnext; 1754 continue; 1755 } 1756 1757 /* 1758 * In .Bl -column, the first rows may be implicit, 1759 * that is, they may not start with .It macros. 1760 * Such rows may be followed by nodes generated on the 1761 * roff level, for example .TS, which cannot be moved 1762 * out of the list. In that case, wrap such roff nodes 1763 * into an implicit row. 1764 */ 1765 1766 if (nchild->prev != NULL) { 1767 mdoc->last = nchild; 1768 mdoc->next = ROFF_NEXT_SIBLING; 1769 roff_block_alloc(mdoc, nchild->line, 1770 nchild->pos, MDOC_It); 1771 roff_head_alloc(mdoc, nchild->line, 1772 nchild->pos, MDOC_It); 1773 mdoc->next = ROFF_NEXT_SIBLING; 1774 roff_body_alloc(mdoc, nchild->line, 1775 nchild->pos, MDOC_It); 1776 while (nchild->tok != MDOC_It) { 1777 roff_node_relink(mdoc, nchild); 1778 if ((nchild = nnext) == NULL) 1779 break; 1780 nnext = nchild->next; 1781 mdoc->next = ROFF_NEXT_SIBLING; 1782 } 1783 mdoc->last = nbody; 1784 continue; 1785 } 1786 1787 mandoc_msg(MANDOCERR_BL_MOVE, nchild->line, nchild->pos, 1788 "%s", roff_name[nchild->tok]); 1789 1790 /* 1791 * Move the node out of the Bl block. 1792 * First, collect all required node pointers. 1793 */ 1794 1795 nblock = nbody->parent; 1796 nprev = nblock->prev; 1797 nparent = nblock->parent; 1798 1799 /* 1800 * Unlink this child. 1801 */ 1802 1803 nbody->child = nnext; 1804 if (nnext == NULL) 1805 nbody->last = NULL; 1806 else 1807 nnext->prev = NULL; 1808 1809 /* 1810 * Relink this child. 1811 */ 1812 1813 nchild->parent = nparent; 1814 nchild->prev = nprev; 1815 nchild->next = nblock; 1816 1817 nblock->prev = nchild; 1818 if (nprev == NULL) 1819 nparent->child = nchild; 1820 else 1821 nprev->next = nchild; 1822 1823 nchild = nnext; 1824 } 1825 1826 if (mdoc->meta.os_e != MANDOC_OS_NETBSD) 1827 return; 1828 1829 prev_Er = NULL; 1830 for (nchild = nbody->child; nchild != NULL; nchild = nchild->next) { 1831 if (nchild->tok != MDOC_It) 1832 continue; 1833 if ((nnext = nchild->head->child) == NULL) 1834 continue; 1835 if (nnext->type == ROFFT_BLOCK) 1836 nnext = nnext->body->child; 1837 if (nnext == NULL || nnext->tok != MDOC_Er) 1838 continue; 1839 nnext = nnext->child; 1840 if (prev_Er != NULL) { 1841 order = strcmp(prev_Er, nnext->string); 1842 if (order > 0) 1843 mandoc_msg(MANDOCERR_ER_ORDER, 1844 nnext->line, nnext->pos, 1845 "Er %s %s (NetBSD)", 1846 prev_Er, nnext->string); 1847 else if (order == 0) 1848 mandoc_msg(MANDOCERR_ER_REP, 1849 nnext->line, nnext->pos, 1850 "Er %s (NetBSD)", prev_Er); 1851 } 1852 prev_Er = nnext->string; 1853 } 1854} 1855 1856static void 1857post_bk(POST_ARGS) 1858{ 1859 struct roff_node *n; 1860 1861 n = mdoc->last; 1862 1863 if (n->type == ROFFT_BLOCK && n->body->child == NULL) { 1864 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos, "Bk"); 1865 roff_node_delete(mdoc, n); 1866 } 1867} 1868 1869static void 1870post_sm(POST_ARGS) 1871{ 1872 struct roff_node *nch; 1873 1874 nch = mdoc->last->child; 1875 1876 if (nch == NULL) { 1877 mdoc->flags ^= MDOC_SMOFF; 1878 return; 1879 } 1880 1881 assert(nch->type == ROFFT_TEXT); 1882 1883 if ( ! strcmp(nch->string, "on")) { 1884 mdoc->flags &= ~MDOC_SMOFF; 1885 return; 1886 } 1887 if ( ! strcmp(nch->string, "off")) { 1888 mdoc->flags |= MDOC_SMOFF; 1889 return; 1890 } 1891 1892 mandoc_msg(MANDOCERR_SM_BAD, nch->line, nch->pos, 1893 "%s %s", roff_name[mdoc->last->tok], nch->string); 1894 roff_node_relink(mdoc, nch); 1895 return; 1896} 1897 1898static void 1899post_root(POST_ARGS) 1900{ 1901 struct roff_node *n; 1902 1903 /* Add missing prologue data. */ 1904 1905 if (mdoc->meta.date == NULL) 1906 mdoc->meta.date = mandoc_normdate(mdoc, NULL, 0, 0); 1907 1908 if (mdoc->meta.title == NULL) { 1909 mandoc_msg(MANDOCERR_DT_NOTITLE, 0, 0, "EOF"); 1910 mdoc->meta.title = mandoc_strdup("UNTITLED"); 1911 } 1912 1913 if (mdoc->meta.vol == NULL) 1914 mdoc->meta.vol = mandoc_strdup("LOCAL"); 1915 1916 if (mdoc->meta.os == NULL) { 1917 mandoc_msg(MANDOCERR_OS_MISSING, 0, 0, NULL); 1918 mdoc->meta.os = mandoc_strdup(""); 1919 } else if (mdoc->meta.os_e && 1920 (mdoc->meta.rcsids & (1 << mdoc->meta.os_e)) == 0) 1921 mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0, 1922 mdoc->meta.os_e == MANDOC_OS_OPENBSD ? 1923 "(OpenBSD)" : "(NetBSD)"); 1924 1925 if (mdoc->meta.arch != NULL && 1926 arch_valid(mdoc->meta.arch, mdoc->meta.os_e) == 0) { 1927 n = mdoc->meta.first->child; 1928 while (n->tok != MDOC_Dt || 1929 n->child == NULL || 1930 n->child->next == NULL || 1931 n->child->next->next == NULL) 1932 n = n->next; 1933 n = n->child->next->next; 1934 mandoc_msg(MANDOCERR_ARCH_BAD, n->line, n->pos, 1935 "Dt ... %s %s", mdoc->meta.arch, 1936 mdoc->meta.os_e == MANDOC_OS_OPENBSD ? 1937 "(OpenBSD)" : "(NetBSD)"); 1938 } 1939 1940 /* Check that we begin with a proper `Sh'. */ 1941 1942 n = mdoc->meta.first->child; 1943 while (n != NULL && 1944 (n->type == ROFFT_COMMENT || 1945 (n->tok >= MDOC_Dd && 1946 mdoc_macro(n->tok)->flags & MDOC_PROLOGUE))) 1947 n = n->next; 1948 1949 if (n == NULL) 1950 mandoc_msg(MANDOCERR_DOC_EMPTY, 0, 0, NULL); 1951 else if (n->tok != MDOC_Sh) 1952 mandoc_msg(MANDOCERR_SEC_BEFORE, n->line, n->pos, 1953 "%s", roff_name[n->tok]); 1954} 1955 1956static void 1957post_rs(POST_ARGS) 1958{ 1959 struct roff_node *np, *nch, *next, *prev; 1960 int i, j; 1961 1962 np = mdoc->last; 1963 1964 if (np->type != ROFFT_BODY) 1965 return; 1966 1967 if (np->child == NULL) { 1968 mandoc_msg(MANDOCERR_RS_EMPTY, np->line, np->pos, "Rs"); 1969 return; 1970 } 1971 1972 /* 1973 * The full `Rs' block needs special handling to order the 1974 * sub-elements according to `rsord'. Pick through each element 1975 * and correctly order it. This is an insertion sort. 1976 */ 1977 1978 next = NULL; 1979 for (nch = np->child->next; nch != NULL; nch = next) { 1980 /* Determine order number of this child. */ 1981 for (i = 0; i < RSORD_MAX; i++) 1982 if (rsord[i] == nch->tok) 1983 break; 1984 1985 if (i == RSORD_MAX) { 1986 mandoc_msg(MANDOCERR_RS_BAD, nch->line, nch->pos, 1987 "%s", roff_name[nch->tok]); 1988 i = -1; 1989 } else if (nch->tok == MDOC__J || nch->tok == MDOC__B) 1990 np->norm->Rs.quote_T++; 1991 1992 /* 1993 * Remove this child from the chain. This somewhat 1994 * repeats roff_node_unlink(), but since we're 1995 * just re-ordering, there's no need for the 1996 * full unlink process. 1997 */ 1998 1999 if ((next = nch->next) != NULL) 2000 next->prev = nch->prev; 2001 2002 if ((prev = nch->prev) != NULL) 2003 prev->next = nch->next; 2004 2005 nch->prev = nch->next = NULL; 2006 2007 /* 2008 * Scan back until we reach a node that's 2009 * to be ordered before this child. 2010 */ 2011 2012 for ( ; prev ; prev = prev->prev) { 2013 /* Determine order of `prev'. */ 2014 for (j = 0; j < RSORD_MAX; j++) 2015 if (rsord[j] == prev->tok) 2016 break; 2017 if (j == RSORD_MAX) 2018 j = -1; 2019 2020 if (j <= i) 2021 break; 2022 } 2023 2024 /* 2025 * Set this child back into its correct place 2026 * in front of the `prev' node. 2027 */ 2028 2029 nch->prev = prev; 2030 2031 if (prev == NULL) { 2032 np->child->prev = nch; 2033 nch->next = np->child; 2034 np->child = nch; 2035 } else { 2036 if (prev->next) 2037 prev->next->prev = nch; 2038 nch->next = prev->next; 2039 prev->next = nch; 2040 } 2041 } 2042} 2043 2044/* 2045 * For some arguments of some macros, 2046 * convert all breakable hyphens into ASCII_HYPH. 2047 */ 2048static void 2049post_hyph(POST_ARGS) 2050{ 2051 struct roff_node *nch; 2052 char *cp; 2053 2054 for (nch = mdoc->last->child; nch != NULL; nch = nch->next) { 2055 if (nch->type != ROFFT_TEXT) 2056 continue; 2057 cp = nch->string; 2058 if (*cp == '\0') 2059 continue; 2060 while (*(++cp) != '\0') 2061 if (*cp == '-' && 2062 isalpha((unsigned char)cp[-1]) && 2063 isalpha((unsigned char)cp[1])) 2064 *cp = ASCII_HYPH; 2065 } 2066} 2067 2068static void 2069post_ns(POST_ARGS) 2070{ 2071 struct roff_node *n; 2072 2073 n = mdoc->last; 2074 if (n->flags & NODE_LINE || 2075 (n->next != NULL && n->next->flags & NODE_DELIMC)) 2076 mandoc_msg(MANDOCERR_NS_SKIP, n->line, n->pos, NULL); 2077} 2078 2079static void 2080post_sx(POST_ARGS) 2081{ 2082 post_delim(mdoc); 2083 post_hyph(mdoc); 2084} 2085 2086static void 2087post_sh(POST_ARGS) 2088{ 2089 2090 post_ignpar(mdoc); 2091 2092 switch (mdoc->last->type) { 2093 case ROFFT_HEAD: 2094 post_sh_head(mdoc); 2095 break; 2096 case ROFFT_BODY: 2097 switch (mdoc->lastsec) { 2098 case SEC_NAME: 2099 post_sh_name(mdoc); 2100 break; 2101 case SEC_SEE_ALSO: 2102 post_sh_see_also(mdoc); 2103 break; 2104 case SEC_AUTHORS: 2105 post_sh_authors(mdoc); 2106 break; 2107 default: 2108 break; 2109 } 2110 break; 2111 default: 2112 break; 2113 } 2114} 2115 2116static void 2117post_sh_name(POST_ARGS) 2118{ 2119 struct roff_node *n; 2120 int hasnm, hasnd; 2121 2122 hasnm = hasnd = 0; 2123 2124 for (n = mdoc->last->child; n != NULL; n = n->next) { 2125 switch (n->tok) { 2126 case MDOC_Nm: 2127 if (hasnm && n->child != NULL) 2128 mandoc_msg(MANDOCERR_NAMESEC_PUNCT, 2129 n->line, n->pos, 2130 "Nm %s", n->child->string); 2131 hasnm = 1; 2132 continue; 2133 case MDOC_Nd: 2134 hasnd = 1; 2135 if (n->next != NULL) 2136 mandoc_msg(MANDOCERR_NAMESEC_ND, 2137 n->line, n->pos, NULL); 2138 break; 2139 case TOKEN_NONE: 2140 if (n->type == ROFFT_TEXT && 2141 n->string[0] == ',' && n->string[1] == '\0' && 2142 n->next != NULL && n->next->tok == MDOC_Nm) { 2143 n = n->next; 2144 continue; 2145 } 2146 /* FALLTHROUGH */ 2147 default: 2148 mandoc_msg(MANDOCERR_NAMESEC_BAD, 2149 n->line, n->pos, "%s", roff_name[n->tok]); 2150 continue; 2151 } 2152 break; 2153 } 2154 2155 if ( ! hasnm) 2156 mandoc_msg(MANDOCERR_NAMESEC_NONM, 2157 mdoc->last->line, mdoc->last->pos, NULL); 2158 if ( ! hasnd) 2159 mandoc_msg(MANDOCERR_NAMESEC_NOND, 2160 mdoc->last->line, mdoc->last->pos, NULL); 2161} 2162 2163static void 2164post_sh_see_also(POST_ARGS) 2165{ 2166 const struct roff_node *n; 2167 const char *name, *sec; 2168 const char *lastname, *lastsec, *lastpunct; 2169 int cmp; 2170 2171 n = mdoc->last->child; 2172 lastname = lastsec = lastpunct = NULL; 2173 while (n != NULL) { 2174 if (n->tok != MDOC_Xr || 2175 n->child == NULL || 2176 n->child->next == NULL) 2177 break; 2178 2179 /* Process one .Xr node. */ 2180 2181 name = n->child->string; 2182 sec = n->child->next->string; 2183 if (lastsec != NULL) { 2184 if (lastpunct[0] != ',' || lastpunct[1] != '\0') 2185 mandoc_msg(MANDOCERR_XR_PUNCT, n->line, 2186 n->pos, "%s before %s(%s)", 2187 lastpunct, name, sec); 2188 cmp = strcmp(lastsec, sec); 2189 if (cmp > 0) 2190 mandoc_msg(MANDOCERR_XR_ORDER, n->line, 2191 n->pos, "%s(%s) after %s(%s)", 2192 name, sec, lastname, lastsec); 2193 else if (cmp == 0 && 2194 strcasecmp(lastname, name) > 0) 2195 mandoc_msg(MANDOCERR_XR_ORDER, n->line, 2196 n->pos, "%s after %s", name, lastname); 2197 } 2198 lastname = name; 2199 lastsec = sec; 2200 2201 /* Process the following node. */ 2202 2203 n = n->next; 2204 if (n == NULL) 2205 break; 2206 if (n->tok == MDOC_Xr) { 2207 lastpunct = "none"; 2208 continue; 2209 } 2210 if (n->type != ROFFT_TEXT) 2211 break; 2212 for (name = n->string; *name != '\0'; name++) 2213 if (isalpha((const unsigned char)*name)) 2214 return; 2215 lastpunct = n->string; 2216 if (n->next == NULL || n->next->tok == MDOC_Rs) 2217 mandoc_msg(MANDOCERR_XR_PUNCT, n->line, 2218 n->pos, "%s after %s(%s)", 2219 lastpunct, lastname, lastsec); 2220 n = n->next; 2221 } 2222} 2223 2224static int 2225child_an(const struct roff_node *n) 2226{ 2227 2228 for (n = n->child; n != NULL; n = n->next) 2229 if ((n->tok == MDOC_An && n->child != NULL) || child_an(n)) 2230 return 1; 2231 return 0; 2232} 2233 2234static void 2235post_sh_authors(POST_ARGS) 2236{ 2237 2238 if ( ! child_an(mdoc->last)) 2239 mandoc_msg(MANDOCERR_AN_MISSING, 2240 mdoc->last->line, mdoc->last->pos, NULL); 2241} 2242 2243/* 2244 * Return an upper bound for the string distance (allowing 2245 * transpositions). Not a full Levenshtein implementation 2246 * because Levenshtein is quadratic in the string length 2247 * and this function is called for every standard name, 2248 * so the check for each custom name would be cubic. 2249 * The following crude heuristics is linear, resulting 2250 * in quadratic behaviour for checking one custom name, 2251 * which does not cause measurable slowdown. 2252 */ 2253static int 2254similar(const char *s1, const char *s2) 2255{ 2256 const int maxdist = 3; 2257 int dist = 0; 2258 2259 while (s1[0] != '\0' && s2[0] != '\0') { 2260 if (s1[0] == s2[0]) { 2261 s1++; 2262 s2++; 2263 continue; 2264 } 2265 if (++dist > maxdist) 2266 return INT_MAX; 2267 if (s1[1] == s2[1]) { /* replacement */ 2268 s1++; 2269 s2++; 2270 } else if (s1[0] == s2[1] && s1[1] == s2[0]) { 2271 s1 += 2; /* transposition */ 2272 s2 += 2; 2273 } else if (s1[0] == s2[1]) /* insertion */ 2274 s2++; 2275 else if (s1[1] == s2[0]) /* deletion */ 2276 s1++; 2277 else 2278 return INT_MAX; 2279 } 2280 dist += strlen(s1) + strlen(s2); 2281 return dist > maxdist ? INT_MAX : dist; 2282} 2283 2284static void 2285post_sh_head(POST_ARGS) 2286{ 2287 struct roff_node *nch; 2288 const char *goodsec; 2289 const char *const *testsec; 2290 int dist, mindist; 2291 enum roff_sec sec; 2292 2293 /* 2294 * Process a new section. Sections are either "named" or 2295 * "custom". Custom sections are user-defined, while named ones 2296 * follow a conventional order and may only appear in certain 2297 * manual sections. 2298 */ 2299 2300 sec = mdoc->last->sec; 2301 2302 /* The NAME should be first. */ 2303 2304 if (sec != SEC_NAME && mdoc->lastnamed == SEC_NONE) 2305 mandoc_msg(MANDOCERR_NAMESEC_FIRST, 2306 mdoc->last->line, mdoc->last->pos, "Sh %s", 2307 sec != SEC_CUSTOM ? secnames[sec] : 2308 (nch = mdoc->last->child) == NULL ? "" : 2309 nch->type == ROFFT_TEXT ? nch->string : 2310 roff_name[nch->tok]); 2311 2312 /* The SYNOPSIS gets special attention in other areas. */ 2313 2314 if (sec == SEC_SYNOPSIS) { 2315 roff_setreg(mdoc->roff, "nS", 1, '='); 2316 mdoc->flags |= MDOC_SYNOPSIS; 2317 } else { 2318 roff_setreg(mdoc->roff, "nS", 0, '='); 2319 mdoc->flags &= ~MDOC_SYNOPSIS; 2320 } 2321 2322 /* Mark our last section. */ 2323 2324 mdoc->lastsec = sec; 2325 2326 /* We don't care about custom sections after this. */ 2327 2328 if (sec == SEC_CUSTOM) { 2329 if ((nch = mdoc->last->child) == NULL || 2330 nch->type != ROFFT_TEXT || nch->next != NULL) 2331 return; 2332 goodsec = NULL; 2333 mindist = INT_MAX; 2334 for (testsec = secnames + 1; *testsec != NULL; testsec++) { 2335 dist = similar(nch->string, *testsec); 2336 if (dist < mindist) { 2337 goodsec = *testsec; 2338 mindist = dist; 2339 } 2340 } 2341 if (goodsec != NULL) 2342 mandoc_msg(MANDOCERR_SEC_TYPO, nch->line, nch->pos, 2343 "Sh %s instead of %s", nch->string, goodsec); 2344 return; 2345 } 2346 2347 /* 2348 * Check whether our non-custom section is being repeated or is 2349 * out of order. 2350 */ 2351 2352 if (sec == mdoc->lastnamed) 2353 mandoc_msg(MANDOCERR_SEC_REP, mdoc->last->line, 2354 mdoc->last->pos, "Sh %s", secnames[sec]); 2355 2356 if (sec < mdoc->lastnamed) 2357 mandoc_msg(MANDOCERR_SEC_ORDER, mdoc->last->line, 2358 mdoc->last->pos, "Sh %s", secnames[sec]); 2359 2360 /* Mark the last named section. */ 2361 2362 mdoc->lastnamed = sec; 2363 2364 /* Check particular section/manual conventions. */ 2365 2366 if (mdoc->meta.msec == NULL) 2367 return; 2368 2369 goodsec = NULL; 2370 switch (sec) { 2371 case SEC_ERRORS: 2372 if (*mdoc->meta.msec == '4') 2373 break; 2374 goodsec = "2, 3, 4, 9"; 2375 /* FALLTHROUGH */ 2376 case SEC_RETURN_VALUES: 2377 case SEC_LIBRARY: 2378 if (*mdoc->meta.msec == '2') 2379 break; 2380 if (*mdoc->meta.msec == '3') 2381 break; 2382 if (NULL == goodsec) 2383 goodsec = "2, 3, 9"; 2384 /* FALLTHROUGH */ 2385 case SEC_CONTEXT: 2386 if (*mdoc->meta.msec == '9') 2387 break; 2388 if (NULL == goodsec) 2389 goodsec = "9"; 2390 mandoc_msg(MANDOCERR_SEC_MSEC, 2391 mdoc->last->line, mdoc->last->pos, 2392 "Sh %s for %s only", secnames[sec], goodsec); 2393 break; 2394 default: 2395 break; 2396 } 2397} 2398 2399static void 2400post_xr(POST_ARGS) 2401{ 2402 struct roff_node *n, *nch; 2403 2404 n = mdoc->last; 2405 nch = n->child; 2406 if (nch->next == NULL) { 2407 mandoc_msg(MANDOCERR_XR_NOSEC, 2408 n->line, n->pos, "Xr %s", nch->string); 2409 } else { 2410 assert(nch->next == n->last); 2411 if(mandoc_xr_add(nch->next->string, nch->string, 2412 nch->line, nch->pos)) 2413 mandoc_msg(MANDOCERR_XR_SELF, 2414 nch->line, nch->pos, "Xr %s %s", 2415 nch->string, nch->next->string); 2416 } 2417 post_delim_nb(mdoc); 2418} 2419 2420static void 2421post_ignpar(POST_ARGS) 2422{ 2423 struct roff_node *np; 2424 2425 switch (mdoc->last->type) { 2426 case ROFFT_BLOCK: 2427 post_prevpar(mdoc); 2428 return; 2429 case ROFFT_HEAD: 2430 post_delim(mdoc); 2431 post_hyph(mdoc); 2432 return; 2433 case ROFFT_BODY: 2434 break; 2435 default: 2436 return; 2437 } 2438 2439 if ((np = mdoc->last->child) != NULL) 2440 if (np->tok == MDOC_Pp || 2441 np->tok == ROFF_br || np->tok == ROFF_sp) { 2442 mandoc_msg(MANDOCERR_PAR_SKIP, np->line, np->pos, 2443 "%s after %s", roff_name[np->tok], 2444 roff_name[mdoc->last->tok]); 2445 roff_node_delete(mdoc, np); 2446 } 2447 2448 if ((np = mdoc->last->last) != NULL) 2449 if (np->tok == MDOC_Pp || np->tok == ROFF_br) { 2450 mandoc_msg(MANDOCERR_PAR_SKIP, np->line, np->pos, 2451 "%s at the end of %s", roff_name[np->tok], 2452 roff_name[mdoc->last->tok]); 2453 roff_node_delete(mdoc, np); 2454 } 2455} 2456 2457static void 2458post_prevpar(POST_ARGS) 2459{ 2460 struct roff_node *n; 2461 2462 n = mdoc->last; 2463 if (NULL == n->prev) 2464 return; 2465 if (n->type != ROFFT_ELEM && n->type != ROFFT_BLOCK) 2466 return; 2467 2468 /* 2469 * Don't allow `Pp' prior to a paragraph-type 2470 * block: `Pp' or non-compact `Bd' or `Bl'. 2471 */ 2472 2473 if (n->prev->tok != MDOC_Pp && n->prev->tok != ROFF_br) 2474 return; 2475 if (n->tok == MDOC_Bl && n->norm->Bl.comp) 2476 return; 2477 if (n->tok == MDOC_Bd && n->norm->Bd.comp) 2478 return; 2479 if (n->tok == MDOC_It && n->parent->norm->Bl.comp) 2480 return; 2481 2482 mandoc_msg(MANDOCERR_PAR_SKIP, n->prev->line, n->prev->pos, 2483 "%s before %s", roff_name[n->prev->tok], roff_name[n->tok]); 2484 roff_node_delete(mdoc, n->prev); 2485} 2486 2487static void 2488post_par(POST_ARGS) 2489{ 2490 struct roff_node *np; 2491 2492 post_prevpar(mdoc); 2493 2494 np = mdoc->last; 2495 if (np->child != NULL) 2496 mandoc_msg(MANDOCERR_ARG_SKIP, np->line, np->pos, 2497 "%s %s", roff_name[np->tok], np->child->string); 2498} 2499 2500static void 2501post_dd(POST_ARGS) 2502{ 2503 struct roff_node *n; 2504 char *datestr; 2505 2506 n = mdoc->last; 2507 n->flags |= NODE_NOPRT; 2508 2509 if (mdoc->meta.date != NULL) { 2510 mandoc_msg(MANDOCERR_PROLOG_REP, n->line, n->pos, "Dd"); 2511 free(mdoc->meta.date); 2512 } else if (mdoc->flags & MDOC_PBODY) 2513 mandoc_msg(MANDOCERR_PROLOG_LATE, n->line, n->pos, "Dd"); 2514 else if (mdoc->meta.title != NULL) 2515 mandoc_msg(MANDOCERR_PROLOG_ORDER, 2516 n->line, n->pos, "Dd after Dt"); 2517 else if (mdoc->meta.os != NULL) 2518 mandoc_msg(MANDOCERR_PROLOG_ORDER, 2519 n->line, n->pos, "Dd after Os"); 2520 2521 datestr = NULL; 2522 deroff(&datestr, n); 2523 mdoc->meta.date = mandoc_normdate(mdoc, datestr, n->line, n->pos); 2524 free(datestr); 2525} 2526 2527static void 2528post_dt(POST_ARGS) 2529{ 2530 struct roff_node *nn, *n; 2531 const char *cp; 2532 char *p; 2533 2534 n = mdoc->last; 2535 n->flags |= NODE_NOPRT; 2536 2537 if (mdoc->flags & MDOC_PBODY) { 2538 mandoc_msg(MANDOCERR_DT_LATE, n->line, n->pos, "Dt"); 2539 return; 2540 } 2541 2542 if (mdoc->meta.title != NULL) 2543 mandoc_msg(MANDOCERR_PROLOG_REP, n->line, n->pos, "Dt"); 2544 else if (mdoc->meta.os != NULL) 2545 mandoc_msg(MANDOCERR_PROLOG_ORDER, 2546 n->line, n->pos, "Dt after Os"); 2547 2548 free(mdoc->meta.title); 2549 free(mdoc->meta.msec); 2550 free(mdoc->meta.vol); 2551 free(mdoc->meta.arch); 2552 2553 mdoc->meta.title = NULL; 2554 mdoc->meta.msec = NULL; 2555 mdoc->meta.vol = NULL; 2556 mdoc->meta.arch = NULL; 2557 2558 /* Mandatory first argument: title. */ 2559 2560 nn = n->child; 2561 if (nn == NULL || *nn->string == '\0') { 2562 mandoc_msg(MANDOCERR_DT_NOTITLE, n->line, n->pos, "Dt"); 2563 mdoc->meta.title = mandoc_strdup("UNTITLED"); 2564 } else { 2565 mdoc->meta.title = mandoc_strdup(nn->string); 2566 2567 /* Check that all characters are uppercase. */ 2568 2569 for (p = nn->string; *p != '\0'; p++) 2570 if (islower((unsigned char)*p)) { 2571 mandoc_msg(MANDOCERR_TITLE_CASE, nn->line, 2572 nn->pos + (int)(p - nn->string), 2573 "Dt %s", nn->string); 2574 break; 2575 } 2576 } 2577 2578 /* Mandatory second argument: section. */ 2579 2580 if (nn != NULL) 2581 nn = nn->next; 2582 2583 if (nn == NULL) { 2584 mandoc_msg(MANDOCERR_MSEC_MISSING, n->line, n->pos, 2585 "Dt %s", mdoc->meta.title); 2586 mdoc->meta.vol = mandoc_strdup("LOCAL"); 2587 return; /* msec and arch remain NULL. */ 2588 } 2589 2590 mdoc->meta.msec = mandoc_strdup(nn->string); 2591 2592 /* Infer volume title from section number. */ 2593 2594 cp = mandoc_a2msec(nn->string); 2595 if (cp == NULL) { 2596 mandoc_msg(MANDOCERR_MSEC_BAD, 2597 nn->line, nn->pos, "Dt ... %s", nn->string); 2598 mdoc->meta.vol = mandoc_strdup(nn->string); 2599 } else 2600 mdoc->meta.vol = mandoc_strdup(cp); 2601 2602 /* Optional third argument: architecture. */ 2603 2604 if ((nn = nn->next) == NULL) 2605 return; 2606 2607 for (p = nn->string; *p != '\0'; p++) 2608 *p = tolower((unsigned char)*p); 2609 mdoc->meta.arch = mandoc_strdup(nn->string); 2610 2611 /* Ignore fourth and later arguments. */ 2612 2613 if ((nn = nn->next) != NULL) 2614 mandoc_msg(MANDOCERR_ARG_EXCESS, 2615 nn->line, nn->pos, "Dt ... %s", nn->string); 2616} 2617 2618static void 2619post_bx(POST_ARGS) 2620{ 2621 struct roff_node *n, *nch; 2622 const char *macro; 2623 2624 post_delim_nb(mdoc); 2625 2626 n = mdoc->last; 2627 nch = n->child; 2628 2629 if (nch != NULL) { 2630 macro = !strcmp(nch->string, "Open") ? "Ox" : 2631 !strcmp(nch->string, "Net") ? "Nx" : 2632 !strcmp(nch->string, "Free") ? "Fx" : 2633 !strcmp(nch->string, "DragonFly") ? "Dx" : NULL; 2634 if (macro != NULL) 2635 mandoc_msg(MANDOCERR_BX, 2636 n->line, n->pos, "%s", macro); 2637 mdoc->last = nch; 2638 nch = nch->next; 2639 mdoc->next = ROFF_NEXT_SIBLING; 2640 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Ns); 2641 mdoc->last->flags |= NODE_NOSRC; 2642 mdoc->next = ROFF_NEXT_SIBLING; 2643 } else 2644 mdoc->next = ROFF_NEXT_CHILD; 2645 roff_word_alloc(mdoc, n->line, n->pos, "BSD"); 2646 mdoc->last->flags |= NODE_NOSRC; 2647 2648 if (nch == NULL) { 2649 mdoc->last = n; 2650 return; 2651 } 2652 2653 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Ns); 2654 mdoc->last->flags |= NODE_NOSRC; 2655 mdoc->next = ROFF_NEXT_SIBLING; 2656 roff_word_alloc(mdoc, n->line, n->pos, "-"); 2657 mdoc->last->flags |= NODE_NOSRC; 2658 roff_elem_alloc(mdoc, n->line, n->pos, MDOC_Ns); 2659 mdoc->last->flags |= NODE_NOSRC; 2660 mdoc->last = n; 2661 2662 /* 2663 * Make `Bx's second argument always start with an uppercase 2664 * letter. Groff checks if it's an "accepted" term, but we just 2665 * uppercase blindly. 2666 */ 2667 2668 *nch->string = (char)toupper((unsigned char)*nch->string); 2669} 2670 2671static void 2672post_os(POST_ARGS) 2673{ 2674#ifndef OSNAME 2675 struct utsname utsname; 2676 static char *defbuf; 2677#endif 2678 struct roff_node *n; 2679 2680 n = mdoc->last; 2681 n->flags |= NODE_NOPRT; 2682 2683 if (mdoc->meta.os != NULL) 2684 mandoc_msg(MANDOCERR_PROLOG_REP, n->line, n->pos, "Os"); 2685 else if (mdoc->flags & MDOC_PBODY) 2686 mandoc_msg(MANDOCERR_PROLOG_LATE, n->line, n->pos, "Os"); 2687 2688 post_delim(mdoc); 2689 2690 /* 2691 * Set the operating system by way of the `Os' macro. 2692 * The order of precedence is: 2693 * 1. the argument of the `Os' macro, unless empty 2694 * 2. the -Ios=foo command line argument, if provided 2695 * 3. -DOSNAME="\"foo\"", if provided during compilation 2696 * 4. "sysname release" from uname(3) 2697 */ 2698 2699 free(mdoc->meta.os); 2700 mdoc->meta.os = NULL; 2701 deroff(&mdoc->meta.os, n); 2702 if (mdoc->meta.os) 2703 goto out; 2704 2705 if (mdoc->os_s != NULL) { 2706 mdoc->meta.os = mandoc_strdup(mdoc->os_s); 2707 goto out; 2708 } 2709 2710#ifdef OSNAME 2711 mdoc->meta.os = mandoc_strdup(OSNAME); 2712#else /*!OSNAME */ 2713 if (defbuf == NULL) { 2714 if (uname(&utsname) == -1) { 2715 mandoc_msg(MANDOCERR_OS_UNAME, n->line, n->pos, "Os"); 2716 defbuf = mandoc_strdup("UNKNOWN"); 2717 } else 2718 mandoc_asprintf(&defbuf, "%s %s", 2719 utsname.sysname, utsname.release); 2720 } 2721 mdoc->meta.os = mandoc_strdup(defbuf); 2722#endif /*!OSNAME*/ 2723 2724out: 2725 if (mdoc->meta.os_e == MANDOC_OS_OTHER) { 2726 if (strstr(mdoc->meta.os, "OpenBSD") != NULL) 2727 mdoc->meta.os_e = MANDOC_OS_OPENBSD; 2728 else if (strstr(mdoc->meta.os, "NetBSD") != NULL) 2729 mdoc->meta.os_e = MANDOC_OS_NETBSD; 2730 } 2731 2732 /* 2733 * This is the earliest point where we can check 2734 * Mdocdate conventions because we don't know 2735 * the operating system earlier. 2736 */ 2737 2738 if (n->child != NULL) 2739 mandoc_msg(MANDOCERR_OS_ARG, n->child->line, n->child->pos, 2740 "Os %s (%s)", n->child->string, 2741 mdoc->meta.os_e == MANDOC_OS_OPENBSD ? 2742 "OpenBSD" : "NetBSD"); 2743 2744 while (n->tok != MDOC_Dd) 2745 if ((n = n->prev) == NULL) 2746 return; 2747 if ((n = n->child) == NULL) 2748 return; 2749 if (strncmp(n->string, "$" "Mdocdate", 9)) { 2750 if (mdoc->meta.os_e == MANDOC_OS_OPENBSD) 2751 mandoc_msg(MANDOCERR_MDOCDATE_MISSING, n->line, 2752 n->pos, "Dd %s (OpenBSD)", n->string); 2753 } else { 2754 if (mdoc->meta.os_e == MANDOC_OS_NETBSD) 2755 mandoc_msg(MANDOCERR_MDOCDATE, n->line, 2756 n->pos, "Dd %s (NetBSD)", n->string); 2757 } 2758} 2759 2760enum roff_sec 2761mdoc_a2sec(const char *p) 2762{ 2763 int i; 2764 2765 for (i = 0; i < (int)SEC__MAX; i++) 2766 if (secnames[i] && 0 == strcmp(p, secnames[i])) 2767 return (enum roff_sec)i; 2768 2769 return SEC_CUSTOM; 2770} 2771 2772static size_t 2773macro2len(enum roff_tok macro) 2774{ 2775 2776 switch (macro) { 2777 case MDOC_Ad: 2778 return 12; 2779 case MDOC_Ao: 2780 return 12; 2781 case MDOC_An: 2782 return 12; 2783 case MDOC_Aq: 2784 return 12; 2785 case MDOC_Ar: 2786 return 12; 2787 case MDOC_Bo: 2788 return 12; 2789 case MDOC_Bq: 2790 return 12; 2791 case MDOC_Cd: 2792 return 12; 2793 case MDOC_Cm: 2794 return 10; 2795 case MDOC_Do: 2796 return 10; 2797 case MDOC_Dq: 2798 return 12; 2799 case MDOC_Dv: 2800 return 12; 2801 case MDOC_Eo: 2802 return 12; 2803 case MDOC_Em: 2804 return 10; 2805 case MDOC_Er: 2806 return 17; 2807 case MDOC_Ev: 2808 return 15; 2809 case MDOC_Fa: 2810 return 12; 2811 case MDOC_Fl: 2812 return 10; 2813 case MDOC_Fo: 2814 return 16; 2815 case MDOC_Fn: 2816 return 16; 2817 case MDOC_Ic: 2818 return 10; 2819 case MDOC_Li: 2820 return 16; 2821 case MDOC_Ms: 2822 return 6; 2823 case MDOC_Nm: 2824 return 10; 2825 case MDOC_No: 2826 return 12; 2827 case MDOC_Oo: 2828 return 10; 2829 case MDOC_Op: 2830 return 14; 2831 case MDOC_Pa: 2832 return 32; 2833 case MDOC_Pf: 2834 return 12; 2835 case MDOC_Po: 2836 return 12; 2837 case MDOC_Pq: 2838 return 12; 2839 case MDOC_Ql: 2840 return 16; 2841 case MDOC_Qo: 2842 return 12; 2843 case MDOC_So: 2844 return 12; 2845 case MDOC_Sq: 2846 return 12; 2847 case MDOC_Sy: 2848 return 6; 2849 case MDOC_Sx: 2850 return 16; 2851 case MDOC_Tn: 2852 return 10; 2853 case MDOC_Va: 2854 return 12; 2855 case MDOC_Vt: 2856 return 12; 2857 case MDOC_Xr: 2858 return 10; 2859 default: 2860 break; 2861 }; 2862 return 0; 2863} 2864