mdoc.c

   1 /*      $Id: mdoc.c,v 1.203 2012/11/17 00:26:33 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010, 2012 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 <stdarg.h>
  26 #include <stdio.h>
  27 #include <stdlib.h>
  28 #include <string.h>
  29 #include <time.h>
  30
  31 #include "mdoc.h"
  32 #include "mandoc.h"
  33 #include "libmdoc.h"
  34 #include "libmandoc.h"
  35
  36 const   char *const __mdoc_macronames[MDOC_MAX] = {
  37         "Ap",           "Dd",           "Dt",           "Os",
  38         "Sh",           "Ss",           "Pp",           "D1",
  39         "Dl",           "Bd",           "Ed",           "Bl",
  40         "El",           "It",           "Ad",           "An",
  41         "Ar",           "Cd",           "Cm",           "Dv",
  42         "Er",           "Ev",           "Ex",           "Fa",
  43         "Fd",           "Fl",           "Fn",           "Ft",
  44         "Ic",           "In",           "Li",           "Nd",
  45         "Nm",           "Op",           "Ot",           "Pa",
  46         "Rv",           "St",           "Va",           "Vt",
  47         /* LINTED */
  48         "Xr",           "%A",           "%B",           "%D",
  49         /* LINTED */
  50         "%I",           "%J",           "%N",           "%O",
  51         /* LINTED */
  52         "%P",           "%R",           "%T",           "%V",
  53         "Ac",           "Ao",           "Aq",           "At",
  54         "Bc",           "Bf",           "Bo",           "Bq",
  55         "Bsx",          "Bx",           "Db",           "Dc",
  56         "Do",           "Dq",           "Ec",           "Ef",
  57         "Em",           "Eo",           "Fx",           "Ms",
  58         "No",           "Ns",           "Nx",           "Ox",
  59         "Pc",           "Pf",           "Po",           "Pq",
  60         "Qc",           "Ql",           "Qo",           "Qq",
  61         "Re",           "Rs",           "Sc",           "So",
  62         "Sq",           "Sm",           "Sx",           "Sy",
  63         "Tn",           "Ux",           "Xc",           "Xo",
  64         "Fo",           "Fc",           "Oo",           "Oc",
  65         "Bk",           "Ek",           "Bt",           "Hf",
  66         "Fr",           "Ud",           "Lb",           "Lp",
  67         "Lk",           "Mt",           "Brq",          "Bro",
  68         /* LINTED */
  69         "Brc",          "%C",           "Es",           "En",
  70         /* LINTED */
  71         "Dx",           "%Q",           "br",           "sp",
  72         /* LINTED */
  73         "%U",           "Ta"
  74         };
  75
  76 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  77         "split",                "nosplit",              "ragged",
  78         "unfilled",             "literal",              "file",
  79         "offset",               "bullet",               "dash",
  80         "hyphen",               "item",                 "enum",
  81         "tag",                  "diag",                 "hang",
  82         "ohang",                "inset",                "column",
  83         "width",                "compact",              "std",
  84         "filled",               "words",                "emphasis",
  85         "symbolic",             "nested",               "centered"
  86         };
  87
  88 const   char * const *mdoc_macronames = __mdoc_macronames;
  89 const   char * const *mdoc_argnames = __mdoc_argnames;
  90
  91 static  void              mdoc_node_free(struct mdoc_node *);
  92 static  void              mdoc_node_unlink(struct mdoc *,
  93                                 struct mdoc_node *);
  94 static  void              mdoc_free1(struct mdoc *);
  95 static  void              mdoc_alloc1(struct mdoc *);
  96 static  struct mdoc_node *node_alloc(struct mdoc *, int, int,
  97                                 enum mdoct, enum mdoc_type);
  98 static  int               node_append(struct mdoc *,
  99                                 struct mdoc_node *);
 100 #if 0
 101 static  int               mdoc_preptext(struct mdoc *, int, char *, int);
 102 #endif
 103 static  int               mdoc_ptext(struct mdoc *, int, char *, int);
 104 static  int               mdoc_pmacro(struct mdoc *, int, char *, int);
 105
 106 const struct mdoc_node *
 107 mdoc_node(const struct mdoc *mdoc)
 108 {
 109
 110         assert( ! (MDOC_HALT & mdoc->flags));
 111         return(mdoc->first);
 112 }
 113
 114
 115 const struct mdoc_meta *
 116 mdoc_meta(const struct mdoc *mdoc)
 117 {
 118
 119         assert( ! (MDOC_HALT & mdoc->flags));
 120         return(&mdoc->meta);
 121 }
 122
 123
 124 /*
 125  * Frees volatile resources (parse tree, meta-data, fields).
 126  */
 127 static void
 128 mdoc_free1(struct mdoc *mdoc)
 129 {
 130
 131         if (mdoc->first)
 132                 mdoc_node_delete(mdoc, mdoc->first);
 133         if (mdoc->meta.title)
 134                 free(mdoc->meta.title);
 135         if (mdoc->meta.os)
 136                 free(mdoc->meta.os);
 137         if (mdoc->meta.name)
 138                 free(mdoc->meta.name);
 139         if (mdoc->meta.arch)
 140                 free(mdoc->meta.arch);
 141         if (mdoc->meta.vol)
 142                 free(mdoc->meta.vol);
 143         if (mdoc->meta.msec)
 144                 free(mdoc->meta.msec);
 145         if (mdoc->meta.date)
 146                 free(mdoc->meta.date);
 147 }
 148
 149
 150 /*
 151  * Allocate all volatile resources (parse tree, meta-data, fields).
 152  */
 153 static void
 154 mdoc_alloc1(struct mdoc *mdoc)
 155 {
 156
 157         memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
 158         mdoc->flags = 0;
 159         mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 160         mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
 161         mdoc->first = mdoc->last;
 162         mdoc->last->type = MDOC_ROOT;
 163         mdoc->last->tok = MDOC_MAX;
 164         mdoc->next = MDOC_NEXT_CHILD;
 165 }
 166
 167
 168 /*
 169  * Free up volatile resources (see mdoc_free1()) then re-initialises the
 170  * data with mdoc_alloc1().  After invocation, parse data has been reset
 171  * and the parser is ready for re-invocation on a new tree; however,
 172  * cross-parse non-volatile data is kept intact.
 173  */
 174 void
 175 mdoc_reset(struct mdoc *mdoc)
 176 {
 177
 178         mdoc_free1(mdoc);
 179         mdoc_alloc1(mdoc);
 180 }
 181
 182
 183 /*
 184  * Completely free up all volatile and non-volatile parse resources.
 185  * After invocation, the pointer is no longer usable.
 186  */
 187 void
 188 mdoc_free(struct mdoc *mdoc)
 189 {
 190
 191         mdoc_free1(mdoc);
 192         free(mdoc);
 193 }
 194
 195
 196 /*
 197  * Allocate volatile and non-volatile parse resources.
 198  */
 199 struct mdoc *
 200 mdoc_alloc(struct roff *roff, struct mparse *parse, char *defos)
 201 {
 202         struct mdoc     *p;
 203
 204         p = mandoc_calloc(1, sizeof(struct mdoc));
 205
 206         p->parse = parse;
 207         p->defos = defos;
 208         p->roff = roff;
 209
 210         mdoc_hash_init();
 211         mdoc_alloc1(p);
 212         return(p);
 213 }
 214
 215
 216 /*
 217  * Climb back up the parse tree, validating open scopes.  Mostly calls
 218  * through to macro_end() in macro.c.
 219  */
 220 int
 221 mdoc_endparse(struct mdoc *mdoc)
 222 {
 223
 224         assert( ! (MDOC_HALT & mdoc->flags));
 225         if (mdoc_macroend(mdoc))
 226                 return(1);
 227         mdoc->flags |= MDOC_HALT;
 228         return(0);
 229 }
 230
 231 int
 232 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 233 {
 234         struct mdoc_node *n;
 235
 236         assert( ! (MDOC_HALT & mdoc->flags));
 237
 238         /* No text before an initial macro. */
 239
 240         if (SEC_NONE == mdoc->lastnamed) {
 241                 mdoc_pmsg(mdoc, ep->ln, ep->pos, MANDOCERR_NOTEXT);
 242                 return(1);
 243         }
 244
 245         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 246         n->eqn = ep;
 247
 248         if ( ! node_append(mdoc, n))
 249                 return(0);
 250
 251         mdoc->next = MDOC_NEXT_SIBLING;
 252         return(1);
 253 }
 254
 255 int
 256 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 257 {
 258         struct mdoc_node *n;
 259
 260         assert( ! (MDOC_HALT & mdoc->flags));
 261
 262         /* No text before an initial macro. */
 263
 264         if (SEC_NONE == mdoc->lastnamed) {
 265                 mdoc_pmsg(mdoc, sp->line, 0, MANDOCERR_NOTEXT);
 266                 return(1);
 267         }
 268
 269         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 270         n->span = sp;
 271
 272         if ( ! node_append(mdoc, n))
 273                 return(0);
 274
 275         mdoc->next = MDOC_NEXT_SIBLING;
 276         return(1);
 277 }
 278
 279
 280 /*
 281  * Main parse routine.  Parses a single line -- really just hands off to
 282  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 283  */
 284 int
 285 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 286 {
 287
 288         assert( ! (MDOC_HALT & mdoc->flags));
 289
 290         mdoc->flags |= MDOC_NEWLINE;
 291
 292         /*
 293          * Let the roff nS register switch SYNOPSIS mode early,
 294          * such that the parser knows at all times
 295          * whether this mode is on or off.
 296          * Note that this mode is also switched by the Sh macro.
 297          */
 298         if (roff_regisset(mdoc->roff, REG_nS)) {
 299                 if (roff_regget(mdoc->roff, REG_nS))
 300                         mdoc->flags |= MDOC_SYNOPSIS;
 301                 else
 302                         mdoc->flags &= ~MDOC_SYNOPSIS;
 303         }
 304
 305         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 306                         mdoc_pmacro(mdoc, ln, buf, offs) :
 307                         mdoc_ptext(mdoc, ln, buf, offs));
 308 }
 309
 310 int
 311 mdoc_macro(MACRO_PROT_ARGS)
 312 {
 313         assert(tok < MDOC_MAX);
 314
 315         /* If we're in the body, deny prologue calls. */
 316
 317         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 318                         MDOC_PBODY & mdoc->flags) {
 319                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADBODY);
 320                 return(1);
 321         }
 322
 323         /* If we're in the prologue, deny "body" macros.  */
 324
 325         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 326                         ! (MDOC_PBODY & mdoc->flags)) {
 327                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADPROLOG);
 328                 if (NULL == mdoc->meta.msec)
 329                         mdoc->meta.msec = mandoc_strdup("1");
 330                 if (NULL == mdoc->meta.title)
 331                         mdoc->meta.title = mandoc_strdup("UNKNOWN");
 332                 if (NULL == mdoc->meta.vol)
 333                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 334                 if (NULL == mdoc->meta.os)
 335                         mdoc->meta.os = mandoc_strdup("LOCAL");
 336                 if (NULL == mdoc->meta.date)
 337                         mdoc->meta.date = mandoc_normdate
 338                                 (mdoc->parse, NULL, line, ppos);
 339                 mdoc->flags |= MDOC_PBODY;
 340         }
 341
 342         return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
 343 }
 344
 345
 346 static int
 347 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 348 {
 349
 350         assert(mdoc->last);
 351         assert(mdoc->first);
 352         assert(MDOC_ROOT != p->type);
 353
 354         switch (mdoc->next) {
 355         case (MDOC_NEXT_SIBLING):
 356                 mdoc->last->next = p;
 357                 p->prev = mdoc->last;
 358                 p->parent = mdoc->last->parent;
 359                 break;
 360         case (MDOC_NEXT_CHILD):
 361                 mdoc->last->child = p;
 362                 p->parent = mdoc->last;
 363                 break;
 364         default:
 365                 abort();
 366                 /* NOTREACHED */
 367         }
 368
 369         p->parent->nchild++;
 370
 371         /*
 372          * Copy over the normalised-data pointer of our parent.  Not
 373          * everybody has one, but copying a null pointer is fine.
 374          */
 375
 376         switch (p->type) {
 377         case (MDOC_BODY):
 378                 if (ENDBODY_NOT != p->end)
 379                         break;
 380                 /* FALLTHROUGH */
 381         case (MDOC_TAIL):
 382                 /* FALLTHROUGH */
 383         case (MDOC_HEAD):
 384                 p->norm = p->parent->norm;
 385                 break;
 386         default:
 387                 break;
 388         }
 389
 390         if ( ! mdoc_valid_pre(mdoc, p))
 391                 return(0);
 392
 393         switch (p->type) {
 394         case (MDOC_HEAD):
 395                 assert(MDOC_BLOCK == p->parent->type);
 396                 p->parent->head = p;
 397                 break;
 398         case (MDOC_TAIL):
 399                 assert(MDOC_BLOCK == p->parent->type);
 400                 p->parent->tail = p;
 401                 break;
 402         case (MDOC_BODY):
 403                 if (p->end)
 404                         break;
 405                 assert(MDOC_BLOCK == p->parent->type);
 406                 p->parent->body = p;
 407                 break;
 408         default:
 409                 break;
 410         }
 411
 412         mdoc->last = p;
 413
 414         switch (p->type) {
 415         case (MDOC_TBL):
 416                 /* FALLTHROUGH */
 417         case (MDOC_TEXT):
 418                 if ( ! mdoc_valid_post(mdoc))
 419                         return(0);
 420                 break;
 421         default:
 422                 break;
 423         }
 424
 425         return(1);
 426 }
 427
 428
 429 static struct mdoc_node *
 430 node_alloc(struct mdoc *mdoc, int line, int pos,
 431                 enum mdoct tok, enum mdoc_type type)
 432 {
 433         struct mdoc_node *p;
 434
 435         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 436         p->sec = mdoc->lastsec;
 437         p->line = line;
 438         p->pos = pos;
 439         p->tok = tok;
 440         p->type = type;
 441
 442         /* Flag analysis. */
 443
 444         if (MDOC_SYNOPSIS & mdoc->flags)
 445                 p->flags |= MDOC_SYNPRETTY;
 446         else
 447                 p->flags &= ~MDOC_SYNPRETTY;
 448         if (MDOC_NEWLINE & mdoc->flags)
 449                 p->flags |= MDOC_LINE;
 450         mdoc->flags &= ~MDOC_NEWLINE;
 451
 452         return(p);
 453 }
 454
 455
 456 int
 457 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 458 {
 459         struct mdoc_node *p;
 460
 461         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 462         if ( ! node_append(mdoc, p))
 463                 return(0);
 464         mdoc->next = MDOC_NEXT_CHILD;
 465         return(1);
 466 }
 467
 468
 469 int
 470 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 471 {
 472         struct mdoc_node *p;
 473
 474         assert(mdoc->first);
 475         assert(mdoc->last);
 476
 477         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 478         if ( ! node_append(mdoc, p))
 479                 return(0);
 480         mdoc->next = MDOC_NEXT_CHILD;
 481         return(1);
 482 }
 483
 484
 485 int
 486 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 487 {
 488         struct mdoc_node *p;
 489
 490         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 491         if ( ! node_append(mdoc, p))
 492                 return(0);
 493         mdoc->next = MDOC_NEXT_CHILD;
 494         return(1);
 495 }
 496
 497
 498 int
 499 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 500                 struct mdoc_node *body, enum mdoc_endbody end)
 501 {
 502         struct mdoc_node *p;
 503
 504         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 505         p->pending = body;
 506         p->norm = body->norm;
 507         p->end = end;
 508         if ( ! node_append(mdoc, p))
 509                 return(0);
 510         mdoc->next = MDOC_NEXT_SIBLING;
 511         return(1);
 512 }
 513
 514
 515 int
 516 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 517                 enum mdoct tok, struct mdoc_arg *args)
 518 {
 519         struct mdoc_node *p;
 520
 521         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 522         p->args = args;
 523         if (p->args)
 524                 (args->refcnt)++;
 525
 526         switch (tok) {
 527         case (MDOC_Bd):
 528                 /* FALLTHROUGH */
 529         case (MDOC_Bf):
 530                 /* FALLTHROUGH */
 531         case (MDOC_Bl):
 532                 /* FALLTHROUGH */
 533         case (MDOC_Rs):
 534                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 535                 break;
 536         default:
 537                 break;
 538         }
 539
 540         if ( ! node_append(mdoc, p))
 541                 return(0);
 542         mdoc->next = MDOC_NEXT_CHILD;
 543         return(1);
 544 }
 545
 546
 547 int
 548 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 549                 enum mdoct tok, struct mdoc_arg *args)
 550 {
 551         struct mdoc_node *p;
 552
 553         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 554         p->args = args;
 555         if (p->args)
 556                 (args->refcnt)++;
 557
 558         switch (tok) {
 559         case (MDOC_An):
 560                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 561                 break;
 562         default:
 563                 break;
 564         }
 565
 566         if ( ! node_append(mdoc, p))
 567                 return(0);
 568         mdoc->next = MDOC_NEXT_CHILD;
 569         return(1);
 570 }
 571
 572 int
 573 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 574 {
 575         struct mdoc_node *n;
 576
 577         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 578         n->string = roff_strdup(mdoc->roff, p);
 579
 580         if ( ! node_append(mdoc, n))
 581                 return(0);
 582
 583         mdoc->next = MDOC_NEXT_SIBLING;
 584         return(1);
 585 }
 586
 587
 588 static void
 589 mdoc_node_free(struct mdoc_node *p)
 590 {
 591
 592         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 593                 free(p->norm);
 594         if (p->string)
 595                 free(p->string);
 596         if (p->args)
 597                 mdoc_argv_free(p->args);
 598         free(p);
 599 }
 600
 601
 602 static void
 603 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 604 {
 605
 606         /* Adjust siblings. */
 607
 608         if (n->prev)
 609                 n->prev->next = n->next;
 610         if (n->next)
 611                 n->next->prev = n->prev;
 612
 613         /* Adjust parent. */
 614
 615         if (n->parent) {
 616                 n->parent->nchild--;
 617                 if (n->parent->child == n)
 618                         n->parent->child = n->prev ? n->prev : n->next;
 619                 if (n->parent->last == n)
 620                         n->parent->last = n->prev ? n->prev : NULL;
 621         }
 622
 623         /* Adjust parse point, if applicable. */
 624
 625         if (mdoc && mdoc->last == n) {
 626                 if (n->prev) {
 627                         mdoc->last = n->prev;
 628                         mdoc->next = MDOC_NEXT_SIBLING;
 629                 } else {
 630                         mdoc->last = n->parent;
 631                         mdoc->next = MDOC_NEXT_CHILD;
 632                 }
 633         }
 634
 635         if (mdoc && mdoc->first == n)
 636                 mdoc->first = NULL;
 637 }
 638
 639
 640 void
 641 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 642 {
 643
 644         while (p->child) {
 645                 assert(p->nchild);
 646                 mdoc_node_delete(mdoc, p->child);
 647         }
 648         assert(0 == p->nchild);
 649
 650         mdoc_node_unlink(mdoc, p);
 651         mdoc_node_free(p);
 652 }
 653
 654 int
 655 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 656 {
 657
 658         mdoc_node_unlink(mdoc, p);
 659         return(node_append(mdoc, p));
 660 }
 661
 662 #if 0
 663 /*
 664  * Pre-treat a text line.
 665  * Text lines can consist of equations, which must be handled apart from
 666  * the regular text.
 667  * Thus, use this function to step through a line checking if it has any
 668  * equations embedded in it.
 669  * This must handle multiple equations AND equations that do not end at
 670  * the end-of-line, i.e., will re-enter in the next roff parse.
 671  */
 672 static int
 673 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
 674 {
 675         char            *start, *end;
 676         char             delim;
 677
 678         while ('\0' != buf[offs]) {
 679                 /* Mark starting position if eqn is set. */
 680                 start = NULL;
 681                 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
 682                         if (NULL != (start = strchr(buf + offs, delim)))
 683                                 *start++ = '\0';
 684
 685                 /* Parse text as normal. */
 686                 if ( ! mdoc_ptext(mdoc, line, buf, offs))
 687                         return(0);
 688
 689                 /* Continue only if an equation exists. */
 690                 if (NULL == start)
 691                         break;
 692
 693                 /* Read past the end of the equation. */
 694                 offs += start - (buf + offs);
 695                 assert(start == &buf[offs]);
 696                 if (NULL != (end = strchr(buf + offs, delim))) {
 697                         *end++ = '\0';
 698                         while (' ' == *end)
 699                                 end++;
 700                 }
 701
 702                 /* Parse the equation itself. */
 703                 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
 704
 705                 /* Process a finished equation? */
 706                 if (roff_closeeqn(mdoc->roff))
 707                         if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
 708                                 return(0);
 709                 offs += (end - (buf + offs));
 710         }
 711
 712         return(1);
 713 }
 714 #endif
 715
 716 /*
 717  * Parse free-form text, that is, a line that does not begin with the
 718  * control character.
 719  */
 720 static int
 721 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 722 {
 723         char             *c, *ws, *end;
 724         struct mdoc_node *n;
 725
 726         /* No text before an initial macro. */
 727
 728         if (SEC_NONE == mdoc->lastnamed) {
 729                 mdoc_pmsg(mdoc, line, offs, MANDOCERR_NOTEXT);
 730                 return(1);
 731         }
 732
 733         assert(mdoc->last);
 734         n = mdoc->last;
 735
 736         /*
 737          * Divert directly to list processing if we're encountering a
 738          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 739          * (a MDOC_BODY means it's already open, in which case we should
 740          * process within its context in the normal way).
 741          */
 742
 743         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 744                         LIST_column == n->norm->Bl.type) {
 745                 /* `Bl' is open without any children. */
 746                 mdoc->flags |= MDOC_FREECOL;
 747                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 748         }
 749
 750         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 751                         NULL != n->parent &&
 752                         MDOC_Bl == n->parent->tok &&
 753                         LIST_column == n->parent->norm->Bl.type) {
 754                 /* `Bl' has block-level `It' children. */
 755                 mdoc->flags |= MDOC_FREECOL;
 756                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 757         }
 758
 759         /*
 760          * Search for the beginning of unescaped trailing whitespace (ws)
 761          * and for the first character not to be output (end).
 762          */
 763
 764         /* FIXME: replace with strcspn(). */
 765         ws = NULL;
 766         for (c = end = buf + offs; *c; c++) {
 767                 switch (*c) {
 768                 case ' ':
 769                         if (NULL == ws)
 770                                 ws = c;
 771                         continue;
 772                 case '\t':
 773                         /*
 774                          * Always warn about trailing tabs,
 775                          * even outside literal context,
 776                          * where they should be put on the next line.
 777                          */
 778                         if (NULL == ws)
 779                                 ws = c;
 780                         /*
 781                          * Strip trailing tabs in literal context only;
 782                          * outside, they affect the next line.
 783                          */
 784                         if (MDOC_LITERAL & mdoc->flags)
 785                                 continue;
 786                         break;
 787                 case '\\':
 788                         /* Skip the escaped character, too, if any. */
 789                         if (c[1])
 790                                 c++;
 791                         /* FALLTHROUGH */
 792                 default:
 793                         ws = NULL;
 794                         break;
 795                 }
 796                 end = c + 1;
 797         }
 798         *end = '\0';
 799
 800         if (ws)
 801                 mdoc_pmsg(mdoc, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
 802
 803         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 804                 mdoc_pmsg(mdoc, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
 805
 806                 /*
 807                  * Insert a `sp' in the case of a blank line.  Technically,
 808                  * blank lines aren't allowed, but enough manuals assume this
 809                  * behaviour that we want to work around it.
 810                  */
 811                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 812                         return(0);
 813
 814                 mdoc->next = MDOC_NEXT_SIBLING;
 815
 816                 return(mdoc_valid_post(mdoc));
 817         }
 818
 819         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 820                 return(0);
 821
 822         if (MDOC_LITERAL & mdoc->flags)
 823                 return(1);
 824
 825         /*
 826          * End-of-sentence check.  If the last character is an unescaped
 827          * EOS character, then flag the node as being the end of a
 828          * sentence.  The front-end will know how to interpret this.
 829          */
 830
 831         assert(buf < end);
 832
 833         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
 834                 mdoc->last->flags |= MDOC_EOS;
 835
 836         return(1);
 837 }
 838
 839
 840 /*
 841  * Parse a macro line, that is, a line beginning with the control
 842  * character.
 843  */
 844 static int
 845 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 846 {
 847         enum mdoct        tok;
 848         int               i, sv;
 849         char              mac[5];
 850         struct mdoc_node *n;
 851
 852         /* Empty post-control lines are ignored. */
 853
 854         if ('"' == buf[offs]) {
 855                 mdoc_pmsg(mdoc, ln, offs, MANDOCERR_BADCOMMENT);
 856                 return(1);
 857         } else if ('\0' == buf[offs])
 858                 return(1);
 859
 860         sv = offs;
 861
 862         /*
 863          * Copy the first word into a nil-terminated buffer.
 864          * Stop copying when a tab, space, or eoln is encountered.
 865          */
 866
 867         i = 0;
 868         while (i < 4 && '\0' != buf[offs] &&
 869                         ' ' != buf[offs] && '\t' != buf[offs])
 870                 mac[i++] = buf[offs++];
 871
 872         mac[i] = '\0';
 873
 874         tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 875
 876         if (MDOC_MAX == tok) {
 877                 mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
 878                                 ln, sv, "%s", buf + sv - 1);
 879                 return(1);
 880         }
 881
 882         /* Disregard the first trailing tab, if applicable. */
 883
 884         if ('\t' == buf[offs])
 885                 offs++;
 886
 887         /* Jump to the next non-whitespace word. */
 888
 889         while (buf[offs] && ' ' == buf[offs])
 890                 offs++;
 891
 892         /*
 893          * Trailing whitespace.  Note that tabs are allowed to be passed
 894          * into the parser as "text", so we only warn about spaces here.
 895          */
 896
 897         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 898                 mdoc_pmsg(mdoc, ln, offs - 1, MANDOCERR_EOLNSPACE);
 899
 900         /*
 901          * If an initial macro or a list invocation, divert directly
 902          * into macro processing.
 903          */
 904
 905         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 906                 if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 907                         goto err;
 908                 return(1);
 909         }
 910
 911         n = mdoc->last;
 912         assert(mdoc->last);
 913
 914         /*
 915          * If the first macro of a `Bl -column', open an `It' block
 916          * context around the parsed macro.
 917          */
 918
 919         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 920                         LIST_column == n->norm->Bl.type) {
 921                 mdoc->flags |= MDOC_FREECOL;
 922                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 923                         goto err;
 924                 return(1);
 925         }
 926
 927         /*
 928          * If we're following a block-level `It' within a `Bl -column'
 929          * context (perhaps opened in the above block or in ptext()),
 930          * then open an `It' block context around the parsed macro.
 931          */
 932
 933         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 934                         NULL != n->parent &&
 935                         MDOC_Bl == n->parent->tok &&
 936                         LIST_column == n->parent->norm->Bl.type) {
 937                 mdoc->flags |= MDOC_FREECOL;
 938                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 939                         goto err;
 940                 return(1);
 941         }
 942
 943         /* Normal processing of a macro. */
 944
 945         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 946                 goto err;
 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 }