mdoc.c

   1 /*      $Id: mdoc.c,v 1.214 2014/04/25 14:11:30 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] = {
  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",
  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         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         /* No text before an initial macro. */
 231
 232         if (SEC_NONE == mdoc->lastnamed) {
 233                 mdoc_pmsg(mdoc, ep->ln, ep->pos, MANDOCERR_NOTEXT);
 234                 return(1);
 235         }
 236
 237         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 238         n->eqn = ep;
 239
 240         if ( ! node_append(mdoc, n))
 241                 return(0);
 242
 243         mdoc->next = MDOC_NEXT_SIBLING;
 244         return(1);
 245 }
 246
 247 int
 248 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 249 {
 250         struct mdoc_node *n;
 251
 252         assert( ! (MDOC_HALT & mdoc->flags));
 253
 254         /* No text before an initial macro. */
 255
 256         if (SEC_NONE == mdoc->lastnamed) {
 257                 mdoc_pmsg(mdoc, sp->line, 0, MANDOCERR_NOTEXT);
 258                 return(1);
 259         }
 260
 261         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 262         n->span = sp;
 263
 264         if ( ! node_append(mdoc, n))
 265                 return(0);
 266
 267         mdoc->next = MDOC_NEXT_SIBLING;
 268         return(1);
 269 }
 270
 271 /*
 272  * Main parse routine.  Parses a single line -- really just hands off to
 273  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 274  */
 275 int
 276 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 277 {
 278
 279         assert( ! (MDOC_HALT & mdoc->flags));
 280
 281         mdoc->flags |= MDOC_NEWLINE;
 282
 283         /*
 284          * Let the roff nS register switch SYNOPSIS mode early,
 285          * such that the parser knows at all times
 286          * whether this mode is on or off.
 287          * Note that this mode is also switched by the Sh macro.
 288          */
 289         if (roff_getreg(mdoc->roff, "nS"))
 290                 mdoc->flags |= MDOC_SYNOPSIS;
 291         else
 292                 mdoc->flags &= ~MDOC_SYNOPSIS;
 293
 294         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 295             mdoc_pmacro(mdoc, ln, buf, offs) :
 296             mdoc_ptext(mdoc, ln, buf, offs));
 297 }
 298
 299 int
 300 mdoc_macro(MACRO_PROT_ARGS)
 301 {
 302         assert(tok < MDOC_MAX);
 303
 304         /* If we're in the body, deny prologue calls. */
 305
 306         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 307             MDOC_PBODY & mdoc->flags) {
 308                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADBODY);
 309                 return(1);
 310         }
 311
 312         /* If we're in the prologue, deny "body" macros.  */
 313
 314         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 315              ! (MDOC_PBODY & mdoc->flags)) {
 316                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADPROLOG);
 317                 if (NULL == mdoc->meta.msec)
 318                         mdoc->meta.msec = mandoc_strdup("1");
 319                 if (NULL == mdoc->meta.title)
 320                         mdoc->meta.title = mandoc_strdup("UNKNOWN");
 321                 if (NULL == mdoc->meta.vol)
 322                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 323                 if (NULL == mdoc->meta.os)
 324                         mdoc->meta.os = mandoc_strdup("LOCAL");
 325                 if (NULL == mdoc->meta.date)
 326                         mdoc->meta.date = mandoc_normdate
 327                                 (mdoc->parse, NULL, line, ppos);
 328                 mdoc->flags |= MDOC_PBODY;
 329         }
 330
 331         return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
 332 }
 333
 334
 335 static int
 336 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 337 {
 338
 339         assert(mdoc->last);
 340         assert(mdoc->first);
 341         assert(MDOC_ROOT != p->type);
 342
 343         switch (mdoc->next) {
 344         case MDOC_NEXT_SIBLING:
 345                 mdoc->last->next = p;
 346                 p->prev = mdoc->last;
 347                 p->parent = mdoc->last->parent;
 348                 break;
 349         case MDOC_NEXT_CHILD:
 350                 mdoc->last->child = p;
 351                 p->parent = mdoc->last;
 352                 break;
 353         default:
 354                 abort();
 355                 /* NOTREACHED */
 356         }
 357
 358         p->parent->nchild++;
 359
 360         /*
 361          * Copy over the normalised-data pointer of our parent.  Not
 362          * everybody has one, but copying a null pointer is fine.
 363          */
 364
 365         switch (p->type) {
 366         case MDOC_BODY:
 367                 if (ENDBODY_NOT != p->end)
 368                         break;
 369                 /* FALLTHROUGH */
 370         case MDOC_TAIL:
 371                 /* FALLTHROUGH */
 372         case MDOC_HEAD:
 373                 p->norm = p->parent->norm;
 374                 break;
 375         default:
 376                 break;
 377         }
 378
 379         if ( ! mdoc_valid_pre(mdoc, p))
 380                 return(0);
 381
 382         switch (p->type) {
 383         case MDOC_HEAD:
 384                 assert(MDOC_BLOCK == p->parent->type);
 385                 p->parent->head = p;
 386                 break;
 387         case MDOC_TAIL:
 388                 assert(MDOC_BLOCK == p->parent->type);
 389                 p->parent->tail = p;
 390                 break;
 391         case MDOC_BODY:
 392                 if (p->end)
 393                         break;
 394                 assert(MDOC_BLOCK == p->parent->type);
 395                 p->parent->body = p;
 396                 break;
 397         default:
 398                 break;
 399         }
 400
 401         mdoc->last = p;
 402
 403         switch (p->type) {
 404         case MDOC_TBL:
 405                 /* FALLTHROUGH */
 406         case MDOC_TEXT:
 407                 if ( ! mdoc_valid_post(mdoc))
 408                         return(0);
 409                 break;
 410         default:
 411                 break;
 412         }
 413
 414         return(1);
 415 }
 416
 417 static struct mdoc_node *
 418 node_alloc(struct mdoc *mdoc, int line, int pos,
 419                 enum mdoct tok, enum mdoc_type type)
 420 {
 421         struct mdoc_node *p;
 422
 423         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 424         p->sec = mdoc->lastsec;
 425         p->line = line;
 426         p->pos = pos;
 427         p->lastline = line;
 428         p->tok = tok;
 429         p->type = type;
 430
 431         /* Flag analysis. */
 432
 433         if (MDOC_SYNOPSIS & mdoc->flags)
 434                 p->flags |= MDOC_SYNPRETTY;
 435         else
 436                 p->flags &= ~MDOC_SYNPRETTY;
 437         if (MDOC_NEWLINE & mdoc->flags)
 438                 p->flags |= MDOC_LINE;
 439         mdoc->flags &= ~MDOC_NEWLINE;
 440
 441         return(p);
 442 }
 443
 444 int
 445 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 446 {
 447         struct mdoc_node *p;
 448
 449         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 450         if ( ! node_append(mdoc, p))
 451                 return(0);
 452         mdoc->next = MDOC_NEXT_CHILD;
 453         return(1);
 454 }
 455
 456 int
 457 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 458 {
 459         struct mdoc_node *p;
 460
 461         assert(mdoc->first);
 462         assert(mdoc->last);
 463
 464         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 465         if ( ! node_append(mdoc, p))
 466                 return(0);
 467         mdoc->next = MDOC_NEXT_CHILD;
 468         return(1);
 469 }
 470
 471 int
 472 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 473 {
 474         struct mdoc_node *p;
 475
 476         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 477         if ( ! node_append(mdoc, p))
 478                 return(0);
 479         mdoc->next = MDOC_NEXT_CHILD;
 480         return(1);
 481 }
 482
 483 int
 484 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 485                 struct mdoc_node *body, enum mdoc_endbody end)
 486 {
 487         struct mdoc_node *p;
 488
 489         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 490         p->pending = body;
 491         p->norm = body->norm;
 492         p->end = end;
 493         if ( ! node_append(mdoc, p))
 494                 return(0);
 495         mdoc->next = MDOC_NEXT_SIBLING;
 496         return(1);
 497 }
 498
 499 int
 500 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 501                 enum mdoct tok, struct mdoc_arg *args)
 502 {
 503         struct mdoc_node *p;
 504
 505         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 506         p->args = args;
 507         if (p->args)
 508                 (args->refcnt)++;
 509
 510         switch (tok) {
 511         case MDOC_Bd:
 512                 /* FALLTHROUGH */
 513         case MDOC_Bf:
 514                 /* FALLTHROUGH */
 515         case MDOC_Bl:
 516                 /* FALLTHROUGH */
 517         case MDOC_Rs:
 518                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 519                 break;
 520         default:
 521                 break;
 522         }
 523
 524         if ( ! node_append(mdoc, p))
 525                 return(0);
 526         mdoc->next = MDOC_NEXT_CHILD;
 527         return(1);
 528 }
 529
 530 int
 531 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 532                 enum mdoct tok, struct mdoc_arg *args)
 533 {
 534         struct mdoc_node *p;
 535
 536         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 537         p->args = args;
 538         if (p->args)
 539                 (args->refcnt)++;
 540
 541         switch (tok) {
 542         case MDOC_An:
 543                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 544                 break;
 545         default:
 546                 break;
 547         }
 548
 549         if ( ! node_append(mdoc, p))
 550                 return(0);
 551         mdoc->next = MDOC_NEXT_CHILD;
 552         return(1);
 553 }
 554
 555 int
 556 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 557 {
 558         struct mdoc_node *n;
 559
 560         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 561         n->string = roff_strdup(mdoc->roff, p);
 562
 563         if ( ! node_append(mdoc, n))
 564                 return(0);
 565
 566         mdoc->next = MDOC_NEXT_SIBLING;
 567         return(1);
 568 }
 569
 570 void
 571 mdoc_word_append(struct mdoc *mdoc, const char *p)
 572 {
 573         struct mdoc_node        *n;
 574         char                    *addstr, *newstr;
 575
 576         n = mdoc->last;
 577         addstr = roff_strdup(mdoc->roff, p);
 578         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 579         free(addstr);
 580         free(n->string);
 581         n->string = newstr;
 582         mdoc->next = MDOC_NEXT_SIBLING;
 583 }
 584
 585 static void
 586 mdoc_node_free(struct mdoc_node *p)
 587 {
 588
 589         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 590                 free(p->norm);
 591         if (p->string)
 592                 free(p->string);
 593         if (p->args)
 594                 mdoc_argv_free(p->args);
 595         free(p);
 596 }
 597
 598 static void
 599 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 600 {
 601
 602         /* Adjust siblings. */
 603
 604         if (n->prev)
 605                 n->prev->next = n->next;
 606         if (n->next)
 607                 n->next->prev = n->prev;
 608
 609         /* Adjust parent. */
 610
 611         if (n->parent) {
 612                 n->parent->nchild--;
 613                 if (n->parent->child == n)
 614                         n->parent->child = n->prev ? n->prev : n->next;
 615                 if (n->parent->last == n)
 616                         n->parent->last = n->prev ? n->prev : NULL;
 617         }
 618
 619         /* Adjust parse point, if applicable. */
 620
 621         if (mdoc && mdoc->last == n) {
 622                 if (n->prev) {
 623                         mdoc->last = n->prev;
 624                         mdoc->next = MDOC_NEXT_SIBLING;
 625                 } else {
 626                         mdoc->last = n->parent;
 627                         mdoc->next = MDOC_NEXT_CHILD;
 628                 }
 629         }
 630
 631         if (mdoc && mdoc->first == n)
 632                 mdoc->first = NULL;
 633 }
 634
 635 void
 636 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 637 {
 638
 639         while (p->child) {
 640                 assert(p->nchild);
 641                 mdoc_node_delete(mdoc, p->child);
 642         }
 643         assert(0 == p->nchild);
 644
 645         mdoc_node_unlink(mdoc, p);
 646         mdoc_node_free(p);
 647 }
 648
 649 int
 650 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 651 {
 652
 653         mdoc_node_unlink(mdoc, p);
 654         return(node_append(mdoc, p));
 655 }
 656
 657 #if 0
 658 /*
 659  * Pre-treat a text line.
 660  * Text lines can consist of equations, which must be handled apart from
 661  * the regular text.
 662  * Thus, use this function to step through a line checking if it has any
 663  * equations embedded in it.
 664  * This must handle multiple equations AND equations that do not end at
 665  * the end-of-line, i.e., will re-enter in the next roff parse.
 666  */
 667 static int
 668 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
 669 {
 670         char            *start, *end;
 671         char             delim;
 672
 673         while ('\0' != buf[offs]) {
 674                 /* Mark starting position if eqn is set. */
 675                 start = NULL;
 676                 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
 677                         if (NULL != (start = strchr(buf + offs, delim)))
 678                                 *start++ = '\0';
 679
 680                 /* Parse text as normal. */
 681                 if ( ! mdoc_ptext(mdoc, line, buf, offs))
 682                         return(0);
 683
 684                 /* Continue only if an equation exists. */
 685                 if (NULL == start)
 686                         break;
 687
 688                 /* Read past the end of the equation. */
 689                 offs += start - (buf + offs);
 690                 assert(start == &buf[offs]);
 691                 if (NULL != (end = strchr(buf + offs, delim))) {
 692                         *end++ = '\0';
 693                         while (' ' == *end)
 694                                 end++;
 695                 }
 696
 697                 /* Parse the equation itself. */
 698                 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
 699
 700                 /* Process a finished equation? */
 701                 if (roff_closeeqn(mdoc->roff))
 702                         if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
 703                                 return(0);
 704                 offs += (end - (buf + offs));
 705         }
 706
 707         return(1);
 708 }
 709 #endif
 710
 711 /*
 712  * Parse free-form text, that is, a line that does not begin with the
 713  * control character.
 714  */
 715 static int
 716 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 717 {
 718         char             *c, *ws, *end;
 719         struct mdoc_node *n;
 720
 721         /* No text before an initial macro. */
 722
 723         if (SEC_NONE == mdoc->lastnamed) {
 724                 mdoc_pmsg(mdoc, line, offs, MANDOCERR_NOTEXT);
 725                 return(1);
 726         }
 727
 728         assert(mdoc->last);
 729         n = mdoc->last;
 730
 731         /*
 732          * Divert directly to list processing if we're encountering a
 733          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 734          * (a MDOC_BODY means it's already open, in which case we should
 735          * process within its context in the normal way).
 736          */
 737
 738         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 739             LIST_column == n->norm->Bl.type) {
 740                 /* `Bl' is open without any children. */
 741                 mdoc->flags |= MDOC_FREECOL;
 742                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 743         }
 744
 745         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 746             NULL != n->parent &&
 747             MDOC_Bl == n->parent->tok &&
 748             LIST_column == n->parent->norm->Bl.type) {
 749                 /* `Bl' has block-level `It' children. */
 750                 mdoc->flags |= MDOC_FREECOL;
 751                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 752         }
 753
 754         /*
 755          * Search for the beginning of unescaped trailing whitespace (ws)
 756          * and for the first character not to be output (end).
 757          */
 758
 759         /* FIXME: replace with strcspn(). */
 760         ws = NULL;
 761         for (c = end = buf + offs; *c; c++) {
 762                 switch (*c) {
 763                 case ' ':
 764                         if (NULL == ws)
 765                                 ws = c;
 766                         continue;
 767                 case '\t':
 768                         /*
 769                          * Always warn about trailing tabs,
 770                          * even outside literal context,
 771                          * where they should be put on the next line.
 772                          */
 773                         if (NULL == ws)
 774                                 ws = c;
 775                         /*
 776                          * Strip trailing tabs in literal context only;
 777                          * outside, they affect the next line.
 778                          */
 779                         if (MDOC_LITERAL & mdoc->flags)
 780                                 continue;
 781                         break;
 782                 case '\\':
 783                         /* Skip the escaped character, too, if any. */
 784                         if (c[1])
 785                                 c++;
 786                         /* FALLTHROUGH */
 787                 default:
 788                         ws = NULL;
 789                         break;
 790                 }
 791                 end = c + 1;
 792         }
 793         *end = '\0';
 794
 795         if (ws)
 796                 mdoc_pmsg(mdoc, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
 797
 798         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 799                 mdoc_pmsg(mdoc, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
 800
 801                 /*
 802                  * Insert a `sp' in the case of a blank line.  Technically,
 803                  * blank lines aren't allowed, but enough manuals assume this
 804                  * behaviour that we want to work around it.
 805                  */
 806                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 807                         return(0);
 808
 809                 mdoc->next = MDOC_NEXT_SIBLING;
 810
 811                 return(mdoc_valid_post(mdoc));
 812         }
 813
 814         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 815                 return(0);
 816
 817         if (MDOC_LITERAL & mdoc->flags)
 818                 return(1);
 819
 820         /*
 821          * End-of-sentence check.  If the last character is an unescaped
 822          * EOS character, then flag the node as being the end of a
 823          * sentence.  The front-end will know how to interpret this.
 824          */
 825
 826         assert(buf < end);
 827
 828         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 829                 mdoc->last->flags |= MDOC_EOS;
 830
 831         return(1);
 832 }
 833
 834 /*
 835  * Parse a macro line, that is, a line beginning with the control
 836  * character.
 837  */
 838 static int
 839 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 840 {
 841         enum mdoct        tok;
 842         int               i, sv;
 843         char              mac[5];
 844         struct mdoc_node *n;
 845
 846         /* Empty post-control lines are ignored. */
 847
 848         if ('"' == buf[offs]) {
 849                 mdoc_pmsg(mdoc, ln, offs, MANDOCERR_BADCOMMENT);
 850                 return(1);
 851         } else if ('\0' == buf[offs])
 852                 return(1);
 853
 854         sv = offs;
 855
 856         /*
 857          * Copy the first word into a nil-terminated buffer.
 858          * Stop copying when a tab, space, or eoln is encountered.
 859          */
 860
 861         i = 0;
 862         while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
 863             '\t' != buf[offs])
 864                 mac[i++] = buf[offs++];
 865
 866         mac[i] = '\0';
 867
 868         tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 869
 870         if (MDOC_MAX == tok) {
 871                 mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
 872                     ln, sv, "%s", buf + sv - 1);
 873                 return(1);
 874         }
 875
 876         /* Disregard the first trailing tab, if applicable. */
 877
 878         if ('\t' == buf[offs])
 879                 offs++;
 880
 881         /* Jump to the next non-whitespace word. */
 882
 883         while (buf[offs] && ' ' == buf[offs])
 884                 offs++;
 885
 886         /*
 887          * Trailing whitespace.  Note that tabs are allowed to be passed
 888          * into the parser as "text", so we only warn about spaces here.
 889          */
 890
 891         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 892                 mdoc_pmsg(mdoc, ln, offs - 1, MANDOCERR_EOLNSPACE);
 893
 894         /*
 895          * If an initial macro or a list invocation, divert directly
 896          * into macro processing.
 897          */
 898
 899         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 900                 if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 901                         goto err;
 902                 return(1);
 903         }
 904
 905         n = mdoc->last;
 906         assert(mdoc->last);
 907
 908         /*
 909          * If the first macro of a `Bl -column', open an `It' block
 910          * context around the parsed macro.
 911          */
 912
 913         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 914             LIST_column == n->norm->Bl.type) {
 915                 mdoc->flags |= MDOC_FREECOL;
 916                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 917                         goto err;
 918                 return(1);
 919         }
 920
 921         /*
 922          * If we're following a block-level `It' within a `Bl -column'
 923          * context (perhaps opened in the above block or in ptext()),
 924          * then open an `It' block context around the parsed macro.
 925          */
 926
 927         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 928             NULL != n->parent &&
 929             MDOC_Bl == n->parent->tok &&
 930             LIST_column == n->parent->norm->Bl.type) {
 931                 mdoc->flags |= MDOC_FREECOL;
 932                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 933                         goto err;
 934                 return(1);
 935         }
 936
 937         /* Normal processing of a macro. */
 938
 939         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 940                 goto err;
 941
 942         /* In quick mode (for mandocdb), abort after the NAME section. */
 943
 944         if (mdoc->quick && MDOC_Sh == tok &&
 945             SEC_NAME != mdoc->last->sec)
 946                 return(2);
 947
 948         return(1);
 949
 950 err:    /* Error out. */
 951
 952         mdoc->flags |= MDOC_HALT;
 953         return(0);
 954 }
 955
 956 enum mdelim
 957 mdoc_isdelim(const char *p)
 958 {
 959
 960         if ('\0' == p[0])
 961                 return(DELIM_NONE);
 962
 963         if ('\0' == p[1])
 964                 switch (p[0]) {
 965                 case '(':
 966                         /* FALLTHROUGH */
 967                 case '[':
 968                         return(DELIM_OPEN);
 969                 case '|':
 970                         return(DELIM_MIDDLE);
 971                 case '.':
 972                         /* FALLTHROUGH */
 973                 case ',':
 974                         /* FALLTHROUGH */
 975                 case ';':
 976                         /* FALLTHROUGH */
 977                 case ':':
 978                         /* FALLTHROUGH */
 979                 case '?':
 980                         /* FALLTHROUGH */
 981                 case '!':
 982                         /* FALLTHROUGH */
 983                 case ')':
 984                         /* FALLTHROUGH */
 985                 case ']':
 986                         return(DELIM_CLOSE);
 987                 default:
 988                         return(DELIM_NONE);
 989                 }
 990
 991         if ('\\' != p[0])
 992                 return(DELIM_NONE);
 993
 994         if (0 == strcmp(p + 1, "."))
 995                 return(DELIM_CLOSE);
 996         if (0 == strcmp(p + 1, "fR|\\fP"))
 997                 return(DELIM_MIDDLE);
 998
 999         return(DELIM_NONE);
1000 }
1001
1002 void
1003 mdoc_deroff(char **dest, const struct mdoc_node *n)
1004 {
1005         char    *cp;
1006         size_t   sz;
1007
1008         if (MDOC_TEXT != n->type) {
1009                 for (n = n->child; n; n = n->next)
1010                         mdoc_deroff(dest, n);
1011                 return;
1012         }
1013
1014         /* Skip leading whitespace. */
1015
1016         for (cp = n->string; '\0' != *cp; cp++)
1017                 if (0 == isspace((unsigned char)*cp))
1018                         break;
1019
1020         /* Skip trailing whitespace. */
1021
1022         for (sz = strlen(cp); sz; sz--)
1023                 if (0 == isspace((unsigned char)cp[sz-1]))
1024                         break;
1025
1026         /* Skip empty strings. */
1027
1028         if (0 == sz)
1029                 return;
1030
1031         if (NULL == *dest) {
1032                 *dest = mandoc_strndup(cp, sz);
1033                 return;
1034         }
1035
1036         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1037         free(*dest);
1038         *dest = cp;
1039 }