mdoc.c

   1 /*      $Id: mdoc.c,v 1.220 2014/07/09 11:31:43 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010, 2012, 2013, 2014 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 AUTHOR DISCLAIMS ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR 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 #ifdef HAVE_CONFIG_H
  19 #include "config.h"
  20 #endif
  21
  22 #include <sys/types.h>
  23
  24 #include <assert.h>
  25 #include <ctype.h>
  26 #include <stdarg.h>
  27 #include <stdio.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <time.h>
  31
  32 #include "mdoc.h"
  33 #include "mandoc.h"
  34 #include "mandoc_aux.h"
  35 #include "libmdoc.h"
  36 #include "libmandoc.h"
  37
  38 const   char *const __mdoc_macronames[MDOC_MAX + 1] = {
  39         "Ap",           "Dd",           "Dt",           "Os",
  40         "Sh",           "Ss",           "Pp",           "D1",
  41         "Dl",           "Bd",           "Ed",           "Bl",
  42         "El",           "It",           "Ad",           "An",
  43         "Ar",           "Cd",           "Cm",           "Dv",
  44         "Er",           "Ev",           "Ex",           "Fa",
  45         "Fd",           "Fl",           "Fn",           "Ft",
  46         "Ic",           "In",           "Li",           "Nd",
  47         "Nm",           "Op",           "Ot",           "Pa",
  48         "Rv",           "St",           "Va",           "Vt",
  49         "Xr",           "%A",           "%B",           "%D",
  50         "%I",           "%J",           "%N",           "%O",
  51         "%P",           "%R",           "%T",           "%V",
  52         "Ac",           "Ao",           "Aq",           "At",
  53         "Bc",           "Bf",           "Bo",           "Bq",
  54         "Bsx",          "Bx",           "Db",           "Dc",
  55         "Do",           "Dq",           "Ec",           "Ef",
  56         "Em",           "Eo",           "Fx",           "Ms",
  57         "No",           "Ns",           "Nx",           "Ox",
  58         "Pc",           "Pf",           "Po",           "Pq",
  59         "Qc",           "Ql",           "Qo",           "Qq",
  60         "Re",           "Rs",           "Sc",           "So",
  61         "Sq",           "Sm",           "Sx",           "Sy",
  62         "Tn",           "Ux",           "Xc",           "Xo",
  63         "Fo",           "Fc",           "Oo",           "Oc",
  64         "Bk",           "Ek",           "Bt",           "Hf",
  65         "Fr",           "Ud",           "Lb",           "Lp",
  66         "Lk",           "Mt",           "Brq",          "Bro",
  67         "Brc",          "%C",           "Es",           "En",
  68         "Dx",           "%Q",           "br",           "sp",
  69         "%U",           "Ta",           "ll",           "text",
  70         };
  71
  72 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  73         "split",                "nosplit",              "ragged",
  74         "unfilled",             "literal",              "file",
  75         "offset",               "bullet",               "dash",
  76         "hyphen",               "item",                 "enum",
  77         "tag",                  "diag",                 "hang",
  78         "ohang",                "inset",                "column",
  79         "width",                "compact",              "std",
  80         "filled",               "words",                "emphasis",
  81         "symbolic",             "nested",               "centered"
  82         };
  83
  84 const   char * const *mdoc_macronames = __mdoc_macronames;
  85 const   char * const *mdoc_argnames = __mdoc_argnames;
  86
  87 static  void              mdoc_node_free(struct mdoc_node *);
  88 static  void              mdoc_node_unlink(struct mdoc *,
  89                                 struct mdoc_node *);
  90 static  void              mdoc_free1(struct mdoc *);
  91 static  void              mdoc_alloc1(struct mdoc *);
  92 static  struct mdoc_node *node_alloc(struct mdoc *, int, int,
  93                                 enum mdoct, enum mdoc_type);
  94 static  int               node_append(struct mdoc *,
  95                                 struct mdoc_node *);
  96 #if 0
  97 static  int               mdoc_preptext(struct mdoc *, int, char *, int);
  98 #endif
  99 static  int               mdoc_ptext(struct mdoc *, int, char *, int);
 100 static  int               mdoc_pmacro(struct mdoc *, int, char *, int);
 101
 102
 103 const struct mdoc_node *
 104 mdoc_node(const struct mdoc *mdoc)
 105 {
 106
 107         assert( ! (MDOC_HALT & mdoc->flags));
 108         return(mdoc->first);
 109 }
 110
 111 const struct mdoc_meta *
 112 mdoc_meta(const struct mdoc *mdoc)
 113 {
 114
 115         assert( ! (MDOC_HALT & mdoc->flags));
 116         return(&mdoc->meta);
 117 }
 118
 119 /*
 120  * Frees volatile resources (parse tree, meta-data, fields).
 121  */
 122 static void
 123 mdoc_free1(struct mdoc *mdoc)
 124 {
 125
 126         if (mdoc->first)
 127                 mdoc_node_delete(mdoc, mdoc->first);
 128         if (mdoc->meta.title)
 129                 free(mdoc->meta.title);
 130         if (mdoc->meta.os)
 131                 free(mdoc->meta.os);
 132         if (mdoc->meta.name)
 133                 free(mdoc->meta.name);
 134         if (mdoc->meta.arch)
 135                 free(mdoc->meta.arch);
 136         if (mdoc->meta.vol)
 137                 free(mdoc->meta.vol);
 138         if (mdoc->meta.msec)
 139                 free(mdoc->meta.msec);
 140         if (mdoc->meta.date)
 141                 free(mdoc->meta.date);
 142 }
 143
 144 /*
 145  * Allocate all volatile resources (parse tree, meta-data, fields).
 146  */
 147 static void
 148 mdoc_alloc1(struct mdoc *mdoc)
 149 {
 150
 151         memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
 152         mdoc->flags = 0;
 153         mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 154         mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
 155         mdoc->first = mdoc->last;
 156         mdoc->last->type = MDOC_ROOT;
 157         mdoc->last->tok = MDOC_MAX;
 158         mdoc->next = MDOC_NEXT_CHILD;
 159 }
 160
 161 /*
 162  * Free up volatile resources (see mdoc_free1()) then re-initialises the
 163  * data with mdoc_alloc1().  After invocation, parse data has been reset
 164  * and the parser is ready for re-invocation on a new tree; however,
 165  * cross-parse non-volatile data is kept intact.
 166  */
 167 void
 168 mdoc_reset(struct mdoc *mdoc)
 169 {
 170
 171         mdoc_free1(mdoc);
 172         mdoc_alloc1(mdoc);
 173 }
 174
 175 /*
 176  * Completely free up all volatile and non-volatile parse resources.
 177  * After invocation, the pointer is no longer usable.
 178  */
 179 void
 180 mdoc_free(struct mdoc *mdoc)
 181 {
 182
 183         mdoc_free1(mdoc);
 184         free(mdoc);
 185 }
 186
 187 /*
 188  * Allocate volatile and non-volatile parse resources.
 189  */
 190 struct mdoc *
 191 mdoc_alloc(struct roff *roff, struct mparse *parse,
 192         const char *defos, int quick)
 193 {
 194         struct mdoc     *p;
 195
 196         p = mandoc_calloc(1, sizeof(struct mdoc));
 197
 198         p->parse = parse;
 199         p->defos = defos;
 200         p->quick = quick;
 201         p->roff = roff;
 202
 203         mdoc_hash_init();
 204         mdoc_alloc1(p);
 205         return(p);
 206 }
 207
 208 /*
 209  * Climb back up the parse tree, validating open scopes.  Mostly calls
 210  * through to macro_end() in macro.c.
 211  */
 212 int
 213 mdoc_endparse(struct mdoc *mdoc)
 214 {
 215
 216         assert( ! (MDOC_HALT & mdoc->flags));
 217         if (mdoc_macroend(mdoc))
 218                 return(1);
 219         mdoc->flags |= MDOC_HALT;
 220         return(0);
 221 }
 222
 223 int
 224 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 225 {
 226         struct mdoc_node *n;
 227
 228         assert( ! (MDOC_HALT & mdoc->flags));
 229
 230         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 231         n->eqn = ep;
 232
 233         if ( ! node_append(mdoc, n))
 234                 return(0);
 235
 236         mdoc->next = MDOC_NEXT_SIBLING;
 237         return(1);
 238 }
 239
 240 int
 241 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 242 {
 243         struct mdoc_node *n;
 244
 245         assert( ! (MDOC_HALT & mdoc->flags));
 246
 247         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 248         n->span = sp;
 249
 250         if ( ! node_append(mdoc, n))
 251                 return(0);
 252
 253         mdoc->next = MDOC_NEXT_SIBLING;
 254         return(1);
 255 }
 256
 257 /*
 258  * Main parse routine.  Parses a single line -- really just hands off to
 259  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 260  */
 261 int
 262 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 263 {
 264
 265         assert( ! (MDOC_HALT & mdoc->flags));
 266
 267         mdoc->flags |= MDOC_NEWLINE;
 268
 269         /*
 270          * Let the roff nS register switch SYNOPSIS mode early,
 271          * such that the parser knows at all times
 272          * whether this mode is on or off.
 273          * Note that this mode is also switched by the Sh macro.
 274          */
 275         if (roff_getreg(mdoc->roff, "nS"))
 276                 mdoc->flags |= MDOC_SYNOPSIS;
 277         else
 278                 mdoc->flags &= ~MDOC_SYNOPSIS;
 279
 280         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 281             mdoc_pmacro(mdoc, ln, buf, offs) :
 282             mdoc_ptext(mdoc, ln, buf, offs));
 283 }
 284
 285 int
 286 mdoc_macro(MACRO_PROT_ARGS)
 287 {
 288         assert(tok < MDOC_MAX);
 289
 290         /* If we're in the body, deny prologue calls. */
 291
 292         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 293             MDOC_PBODY & mdoc->flags) {
 294                 mandoc_vmsg(MANDOCERR_PROLOG_ONLY, mdoc->parse,
 295                     line, ppos, "%s", mdoc_macronames[tok]);
 296                 return(1);
 297         }
 298
 299         /* If we're in the prologue, deny "body" macros.  */
 300
 301         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 302              ! (MDOC_PBODY & mdoc->flags)) {
 303                 mandoc_vmsg(MANDOCERR_PROLOG_BAD, mdoc->parse,
 304                     line, ppos, "%s", mdoc_macronames[tok]);
 305                 if (NULL == mdoc->meta.msec)
 306                         mdoc->meta.msec = mandoc_strdup("1");
 307                 if (NULL == mdoc->meta.title)
 308                         mdoc->meta.title = mandoc_strdup("UNKNOWN");
 309                 if (NULL == mdoc->meta.vol)
 310                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 311                 if (NULL == mdoc->meta.os)
 312                         mdoc->meta.os = mandoc_strdup("LOCAL");
 313                 if (NULL == mdoc->meta.date)
 314                         mdoc->meta.date = mandoc_normdate
 315                                 (mdoc->parse, NULL, line, ppos);
 316                 mdoc->flags |= MDOC_PBODY;
 317         }
 318
 319         return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
 320 }
 321
 322
 323 static int
 324 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 325 {
 326
 327         assert(mdoc->last);
 328         assert(mdoc->first);
 329         assert(MDOC_ROOT != p->type);
 330
 331         switch (mdoc->next) {
 332         case MDOC_NEXT_SIBLING:
 333                 mdoc->last->next = p;
 334                 p->prev = mdoc->last;
 335                 p->parent = mdoc->last->parent;
 336                 break;
 337         case MDOC_NEXT_CHILD:
 338                 mdoc->last->child = p;
 339                 p->parent = mdoc->last;
 340                 break;
 341         default:
 342                 abort();
 343                 /* NOTREACHED */
 344         }
 345
 346         p->parent->nchild++;
 347
 348         /*
 349          * Copy over the normalised-data pointer of our parent.  Not
 350          * everybody has one, but copying a null pointer is fine.
 351          */
 352
 353         switch (p->type) {
 354         case MDOC_BODY:
 355                 if (ENDBODY_NOT != p->end)
 356                         break;
 357                 /* FALLTHROUGH */
 358         case MDOC_TAIL:
 359                 /* FALLTHROUGH */
 360         case MDOC_HEAD:
 361                 p->norm = p->parent->norm;
 362                 break;
 363         default:
 364                 break;
 365         }
 366
 367         if ( ! mdoc_valid_pre(mdoc, p))
 368                 return(0);
 369
 370         switch (p->type) {
 371         case MDOC_HEAD:
 372                 assert(MDOC_BLOCK == p->parent->type);
 373                 p->parent->head = p;
 374                 break;
 375         case MDOC_TAIL:
 376                 assert(MDOC_BLOCK == p->parent->type);
 377                 p->parent->tail = p;
 378                 break;
 379         case MDOC_BODY:
 380                 if (p->end)
 381                         break;
 382                 assert(MDOC_BLOCK == p->parent->type);
 383                 p->parent->body = p;
 384                 break;
 385         default:
 386                 break;
 387         }
 388
 389         mdoc->last = p;
 390
 391         switch (p->type) {
 392         case MDOC_TBL:
 393                 /* FALLTHROUGH */
 394         case MDOC_TEXT:
 395                 if ( ! mdoc_valid_post(mdoc))
 396                         return(0);
 397                 break;
 398         default:
 399                 break;
 400         }
 401
 402         return(1);
 403 }
 404
 405 static struct mdoc_node *
 406 node_alloc(struct mdoc *mdoc, int line, int pos,
 407                 enum mdoct tok, enum mdoc_type type)
 408 {
 409         struct mdoc_node *p;
 410
 411         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 412         p->sec = mdoc->lastsec;
 413         p->line = line;
 414         p->pos = pos;
 415         p->lastline = line;
 416         p->tok = tok;
 417         p->type = type;
 418
 419         /* Flag analysis. */
 420
 421         if (MDOC_SYNOPSIS & mdoc->flags)
 422                 p->flags |= MDOC_SYNPRETTY;
 423         else
 424                 p->flags &= ~MDOC_SYNPRETTY;
 425         if (MDOC_NEWLINE & mdoc->flags)
 426                 p->flags |= MDOC_LINE;
 427         mdoc->flags &= ~MDOC_NEWLINE;
 428
 429         return(p);
 430 }
 431
 432 int
 433 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 434 {
 435         struct mdoc_node *p;
 436
 437         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 438         if ( ! node_append(mdoc, p))
 439                 return(0);
 440         mdoc->next = MDOC_NEXT_CHILD;
 441         return(1);
 442 }
 443
 444 int
 445 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 446 {
 447         struct mdoc_node *p;
 448
 449         assert(mdoc->first);
 450         assert(mdoc->last);
 451
 452         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 453         if ( ! node_append(mdoc, p))
 454                 return(0);
 455         mdoc->next = MDOC_NEXT_CHILD;
 456         return(1);
 457 }
 458
 459 int
 460 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 461 {
 462         struct mdoc_node *p;
 463
 464         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 465         if ( ! node_append(mdoc, p))
 466                 return(0);
 467         mdoc->next = MDOC_NEXT_CHILD;
 468         return(1);
 469 }
 470
 471 int
 472 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 473                 struct mdoc_node *body, enum mdoc_endbody end)
 474 {
 475         struct mdoc_node *p;
 476
 477         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 478         p->pending = body;
 479         p->norm = body->norm;
 480         p->end = end;
 481         if ( ! node_append(mdoc, p))
 482                 return(0);
 483         mdoc->next = MDOC_NEXT_SIBLING;
 484         return(1);
 485 }
 486
 487 int
 488 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 489                 enum mdoct tok, struct mdoc_arg *args)
 490 {
 491         struct mdoc_node *p;
 492
 493         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 494         p->args = args;
 495         if (p->args)
 496                 (args->refcnt)++;
 497
 498         switch (tok) {
 499         case MDOC_Bd:
 500                 /* FALLTHROUGH */
 501         case MDOC_Bf:
 502                 /* FALLTHROUGH */
 503         case MDOC_Bl:
 504                 /* FALLTHROUGH */
 505         case MDOC_En:
 506                 /* FALLTHROUGH */
 507         case MDOC_Rs:
 508                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 509                 break;
 510         default:
 511                 break;
 512         }
 513
 514         if ( ! node_append(mdoc, p))
 515                 return(0);
 516         mdoc->next = MDOC_NEXT_CHILD;
 517         return(1);
 518 }
 519
 520 int
 521 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 522                 enum mdoct tok, struct mdoc_arg *args)
 523 {
 524         struct mdoc_node *p;
 525
 526         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 527         p->args = args;
 528         if (p->args)
 529                 (args->refcnt)++;
 530
 531         switch (tok) {
 532         case MDOC_An:
 533                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 534                 break;
 535         default:
 536                 break;
 537         }
 538
 539         if ( ! node_append(mdoc, p))
 540                 return(0);
 541         mdoc->next = MDOC_NEXT_CHILD;
 542         return(1);
 543 }
 544
 545 int
 546 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 547 {
 548         struct mdoc_node *n;
 549
 550         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 551         n->string = roff_strdup(mdoc->roff, p);
 552
 553         if ( ! node_append(mdoc, n))
 554                 return(0);
 555
 556         mdoc->next = MDOC_NEXT_SIBLING;
 557         return(1);
 558 }
 559
 560 void
 561 mdoc_word_append(struct mdoc *mdoc, const char *p)
 562 {
 563         struct mdoc_node        *n;
 564         char                    *addstr, *newstr;
 565
 566         n = mdoc->last;
 567         addstr = roff_strdup(mdoc->roff, p);
 568         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 569         free(addstr);
 570         free(n->string);
 571         n->string = newstr;
 572         mdoc->next = MDOC_NEXT_SIBLING;
 573 }
 574
 575 static void
 576 mdoc_node_free(struct mdoc_node *p)
 577 {
 578
 579         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 580                 free(p->norm);
 581         if (p->string)
 582                 free(p->string);
 583         if (p->args)
 584                 mdoc_argv_free(p->args);
 585         free(p);
 586 }
 587
 588 static void
 589 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 590 {
 591
 592         /* Adjust siblings. */
 593
 594         if (n->prev)
 595                 n->prev->next = n->next;
 596         if (n->next)
 597                 n->next->prev = n->prev;
 598
 599         /* Adjust parent. */
 600
 601         if (n->parent) {
 602                 n->parent->nchild--;
 603                 if (n->parent->child == n)
 604                         n->parent->child = n->prev ? n->prev : n->next;
 605                 if (n->parent->last == n)
 606                         n->parent->last = n->prev ? n->prev : NULL;
 607         }
 608
 609         /* Adjust parse point, if applicable. */
 610
 611         if (mdoc && mdoc->last == n) {
 612                 if (n->prev) {
 613                         mdoc->last = n->prev;
 614                         mdoc->next = MDOC_NEXT_SIBLING;
 615                 } else {
 616                         mdoc->last = n->parent;
 617                         mdoc->next = MDOC_NEXT_CHILD;
 618                 }
 619         }
 620
 621         if (mdoc && mdoc->first == n)
 622                 mdoc->first = NULL;
 623 }
 624
 625 void
 626 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 627 {
 628
 629         while (p->child) {
 630                 assert(p->nchild);
 631                 mdoc_node_delete(mdoc, p->child);
 632         }
 633         assert(0 == p->nchild);
 634
 635         mdoc_node_unlink(mdoc, p);
 636         mdoc_node_free(p);
 637 }
 638
 639 int
 640 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 641 {
 642
 643         mdoc_node_unlink(mdoc, p);
 644         return(node_append(mdoc, p));
 645 }
 646
 647 #if 0
 648 /*
 649  * Pre-treat a text line.
 650  * Text lines can consist of equations, which must be handled apart from
 651  * the regular text.
 652  * Thus, use this function to step through a line checking if it has any
 653  * equations embedded in it.
 654  * This must handle multiple equations AND equations that do not end at
 655  * the end-of-line, i.e., will re-enter in the next roff parse.
 656  */
 657 static int
 658 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
 659 {
 660         char            *start, *end;
 661         char             delim;
 662
 663         while ('\0' != buf[offs]) {
 664                 /* Mark starting position if eqn is set. */
 665                 start = NULL;
 666                 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
 667                         if (NULL != (start = strchr(buf + offs, delim)))
 668                                 *start++ = '\0';
 669
 670                 /* Parse text as normal. */
 671                 if ( ! mdoc_ptext(mdoc, line, buf, offs))
 672                         return(0);
 673
 674                 /* Continue only if an equation exists. */
 675                 if (NULL == start)
 676                         break;
 677
 678                 /* Read past the end of the equation. */
 679                 offs += start - (buf + offs);
 680                 assert(start == &buf[offs]);
 681                 if (NULL != (end = strchr(buf + offs, delim))) {
 682                         *end++ = '\0';
 683                         while (' ' == *end)
 684                                 end++;
 685                 }
 686
 687                 /* Parse the equation itself. */
 688                 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
 689
 690                 /* Process a finished equation? */
 691                 if (roff_closeeqn(mdoc->roff))
 692                         if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
 693                                 return(0);
 694                 offs += (end - (buf + offs));
 695         }
 696
 697         return(1);
 698 }
 699 #endif
 700
 701 /*
 702  * Parse free-form text, that is, a line that does not begin with the
 703  * control character.
 704  */
 705 static int
 706 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 707 {
 708         char             *c, *ws, *end;
 709         struct mdoc_node *n;
 710
 711         assert(mdoc->last);
 712         n = mdoc->last;
 713
 714         /*
 715          * Divert directly to list processing if we're encountering a
 716          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 717          * (a MDOC_BODY means it's already open, in which case we should
 718          * process within its context in the normal way).
 719          */
 720
 721         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 722             LIST_column == n->norm->Bl.type) {
 723                 /* `Bl' is open without any children. */
 724                 mdoc->flags |= MDOC_FREECOL;
 725                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 726         }
 727
 728         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 729             NULL != n->parent &&
 730             MDOC_Bl == n->parent->tok &&
 731             LIST_column == n->parent->norm->Bl.type) {
 732                 /* `Bl' has block-level `It' children. */
 733                 mdoc->flags |= MDOC_FREECOL;
 734                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 735         }
 736
 737         /*
 738          * Search for the beginning of unescaped trailing whitespace (ws)
 739          * and for the first character not to be output (end).
 740          */
 741
 742         /* FIXME: replace with strcspn(). */
 743         ws = NULL;
 744         for (c = end = buf + offs; *c; c++) {
 745                 switch (*c) {
 746                 case ' ':
 747                         if (NULL == ws)
 748                                 ws = c;
 749                         continue;
 750                 case '\t':
 751                         /*
 752                          * Always warn about trailing tabs,
 753                          * even outside literal context,
 754                          * where they should be put on the next line.
 755                          */
 756                         if (NULL == ws)
 757                                 ws = c;
 758                         /*
 759                          * Strip trailing tabs in literal context only;
 760                          * outside, they affect the next line.
 761                          */
 762                         if (MDOC_LITERAL & mdoc->flags)
 763                                 continue;
 764                         break;
 765                 case '\\':
 766                         /* Skip the escaped character, too, if any. */
 767                         if (c[1])
 768                                 c++;
 769                         /* FALLTHROUGH */
 770                 default:
 771                         ws = NULL;
 772                         break;
 773                 }
 774                 end = c + 1;
 775         }
 776         *end = '\0';
 777
 778         if (ws)
 779                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 780                     line, (int)(ws-buf), NULL);
 781
 782         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 783                 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
 784                     line, (int)(c - buf), NULL);
 785
 786                 /*
 787                  * Insert a `sp' in the case of a blank line.  Technically,
 788                  * blank lines aren't allowed, but enough manuals assume this
 789                  * behaviour that we want to work around it.
 790                  */
 791                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 792                         return(0);
 793
 794                 mdoc->next = MDOC_NEXT_SIBLING;
 795
 796                 return(mdoc_valid_post(mdoc));
 797         }
 798
 799         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 800                 return(0);
 801
 802         if (MDOC_LITERAL & mdoc->flags)
 803                 return(1);
 804
 805         /*
 806          * End-of-sentence check.  If the last character is an unescaped
 807          * EOS character, then flag the node as being the end of a
 808          * sentence.  The front-end will know how to interpret this.
 809          */
 810
 811         assert(buf < end);
 812
 813         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 814                 mdoc->last->flags |= MDOC_EOS;
 815
 816         return(1);
 817 }
 818
 819 /*
 820  * Parse a macro line, that is, a line beginning with the control
 821  * character.
 822  */
 823 static int
 824 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 825 {
 826         enum mdoct        tok;
 827         int               i, sv;
 828         char              mac[5];
 829         struct mdoc_node *n;
 830
 831         /* Empty post-control lines are ignored. */
 832
 833         if ('"' == buf[offs]) {
 834                 mandoc_msg(MANDOCERR_COMMENT_BAD, mdoc->parse,
 835                     ln, offs, NULL);
 836                 return(1);
 837         } else if ('\0' == buf[offs])
 838                 return(1);
 839
 840         sv = offs;
 841
 842         /*
 843          * Copy the first word into a nil-terminated buffer.
 844          * Stop copying when a tab, space, or eoln is encountered.
 845          */
 846
 847         i = 0;
 848         while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
 849             '\t' != buf[offs])
 850                 mac[i++] = buf[offs++];
 851
 852         mac[i] = '\0';
 853
 854         tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 855
 856         if (MDOC_MAX == tok) {
 857                 mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
 858                     ln, sv, "%s", buf + sv - 1);
 859                 return(1);
 860         }
 861
 862         /* Disregard the first trailing tab, if applicable. */
 863
 864         if ('\t' == buf[offs])
 865                 offs++;
 866
 867         /* Jump to the next non-whitespace word. */
 868
 869         while (buf[offs] && ' ' == buf[offs])
 870                 offs++;
 871
 872         /*
 873          * Trailing whitespace.  Note that tabs are allowed to be passed
 874          * into the parser as "text", so we only warn about spaces here.
 875          */
 876
 877         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 878                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 879                     ln, offs - 1, NULL);
 880
 881         /*
 882          * If an initial macro or a list invocation, divert directly
 883          * into macro processing.
 884          */
 885
 886         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 887                 if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 888                         goto err;
 889                 return(1);
 890         }
 891
 892         n = mdoc->last;
 893         assert(mdoc->last);
 894
 895         /*
 896          * If the first macro of a `Bl -column', open an `It' block
 897          * context around the parsed macro.
 898          */
 899
 900         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 901             LIST_column == n->norm->Bl.type) {
 902                 mdoc->flags |= MDOC_FREECOL;
 903                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 904                         goto err;
 905                 return(1);
 906         }
 907
 908         /*
 909          * If we're following a block-level `It' within a `Bl -column'
 910          * context (perhaps opened in the above block or in ptext()),
 911          * then open an `It' block context around the parsed macro.
 912          */
 913
 914         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 915             NULL != n->parent &&
 916             MDOC_Bl == n->parent->tok &&
 917             LIST_column == n->parent->norm->Bl.type) {
 918                 mdoc->flags |= MDOC_FREECOL;
 919                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 920                         goto err;
 921                 return(1);
 922         }
 923
 924         /* Normal processing of a macro. */
 925
 926         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 927                 goto err;
 928
 929         /* In quick mode (for mandocdb), abort after the NAME section. */
 930
 931         if (mdoc->quick && MDOC_Sh == tok &&
 932             SEC_NAME != mdoc->last->sec)
 933                 return(2);
 934
 935         return(1);
 936
 937 err:    /* Error out. */
 938
 939         mdoc->flags |= MDOC_HALT;
 940         return(0);
 941 }
 942
 943 enum mdelim
 944 mdoc_isdelim(const char *p)
 945 {
 946
 947         if ('\0' == p[0])
 948                 return(DELIM_NONE);
 949
 950         if ('\0' == p[1])
 951                 switch (p[0]) {
 952                 case '(':
 953                         /* FALLTHROUGH */
 954                 case '[':
 955                         return(DELIM_OPEN);
 956                 case '|':
 957                         return(DELIM_MIDDLE);
 958                 case '.':
 959                         /* FALLTHROUGH */
 960                 case ',':
 961                         /* FALLTHROUGH */
 962                 case ';':
 963                         /* FALLTHROUGH */
 964                 case ':':
 965                         /* FALLTHROUGH */
 966                 case '?':
 967                         /* FALLTHROUGH */
 968                 case '!':
 969                         /* FALLTHROUGH */
 970                 case ')':
 971                         /* FALLTHROUGH */
 972                 case ']':
 973                         return(DELIM_CLOSE);
 974                 default:
 975                         return(DELIM_NONE);
 976                 }
 977
 978         if ('\\' != p[0])
 979                 return(DELIM_NONE);
 980
 981         if (0 == strcmp(p + 1, "."))
 982                 return(DELIM_CLOSE);
 983         if (0 == strcmp(p + 1, "fR|\\fP"))
 984                 return(DELIM_MIDDLE);
 985
 986         return(DELIM_NONE);
 987 }
 988
 989 void
 990 mdoc_deroff(char **dest, const struct mdoc_node *n)
 991 {
 992         char    *cp;
 993         size_t   sz;
 994
 995         if (MDOC_TEXT != n->type) {
 996                 for (n = n->child; n; n = n->next)
 997                         mdoc_deroff(dest, n);
 998                 return;
 999         }
1000
1001         /* Skip leading whitespace. */
1002
1003         for (cp = n->string; '\0' != *cp; cp++)
1004                 if (0 == isspace((unsigned char)*cp))
1005                         break;
1006
1007         /* Skip trailing whitespace. */
1008
1009         for (sz = strlen(cp); sz; sz--)
1010                 if (0 == isspace((unsigned char)cp[sz-1]))
1011                         break;
1012
1013         /* Skip empty strings. */
1014
1015         if (0 == sz)
1016                 return;
1017
1018         if (NULL == *dest) {
1019                 *dest = mandoc_strndup(cp, sz);
1020                 return;
1021         }
1022
1023         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1024         free(*dest);
1025         *dest = cp;
1026 }