mdoc.c

   1 /*      $Id: mdoc.c,v 1.204 2013/10/05 22:08:12 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_getreg(mdoc->roff, "nS"))
 299                 mdoc->flags |= MDOC_SYNOPSIS;
 300         else
 301                 mdoc->flags &= ~MDOC_SYNOPSIS;
 302
 303         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 304                         mdoc_pmacro(mdoc, ln, buf, offs) :
 305                         mdoc_ptext(mdoc, ln, buf, offs));
 306 }
 307
 308 int
 309 mdoc_macro(MACRO_PROT_ARGS)
 310 {
 311         assert(tok < MDOC_MAX);
 312
 313         /* If we're in the body, deny prologue calls. */
 314
 315         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 316                         MDOC_PBODY & mdoc->flags) {
 317                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADBODY);
 318                 return(1);
 319         }
 320
 321         /* If we're in the prologue, deny "body" macros.  */
 322
 323         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 324                         ! (MDOC_PBODY & mdoc->flags)) {
 325                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADPROLOG);
 326                 if (NULL == mdoc->meta.msec)
 327                         mdoc->meta.msec = mandoc_strdup("1");
 328                 if (NULL == mdoc->meta.title)
 329                         mdoc->meta.title = mandoc_strdup("UNKNOWN");
 330                 if (NULL == mdoc->meta.vol)
 331                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 332                 if (NULL == mdoc->meta.os)
 333                         mdoc->meta.os = mandoc_strdup("LOCAL");
 334                 if (NULL == mdoc->meta.date)
 335                         mdoc->meta.date = mandoc_normdate
 336                                 (mdoc->parse, NULL, line, ppos);
 337                 mdoc->flags |= MDOC_PBODY;
 338         }
 339
 340         return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
 341 }
 342
 343
 344 static int
 345 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 346 {
 347
 348         assert(mdoc->last);
 349         assert(mdoc->first);
 350         assert(MDOC_ROOT != p->type);
 351
 352         switch (mdoc->next) {
 353         case (MDOC_NEXT_SIBLING):
 354                 mdoc->last->next = p;
 355                 p->prev = mdoc->last;
 356                 p->parent = mdoc->last->parent;
 357                 break;
 358         case (MDOC_NEXT_CHILD):
 359                 mdoc->last->child = p;
 360                 p->parent = mdoc->last;
 361                 break;
 362         default:
 363                 abort();
 364                 /* NOTREACHED */
 365         }
 366
 367         p->parent->nchild++;
 368
 369         /*
 370          * Copy over the normalised-data pointer of our parent.  Not
 371          * everybody has one, but copying a null pointer is fine.
 372          */
 373
 374         switch (p->type) {
 375         case (MDOC_BODY):
 376                 if (ENDBODY_NOT != p->end)
 377                         break;
 378                 /* FALLTHROUGH */
 379         case (MDOC_TAIL):
 380                 /* FALLTHROUGH */
 381         case (MDOC_HEAD):
 382                 p->norm = p->parent->norm;
 383                 break;
 384         default:
 385                 break;
 386         }
 387
 388         if ( ! mdoc_valid_pre(mdoc, p))
 389                 return(0);
 390
 391         switch (p->type) {
 392         case (MDOC_HEAD):
 393                 assert(MDOC_BLOCK == p->parent->type);
 394                 p->parent->head = p;
 395                 break;
 396         case (MDOC_TAIL):
 397                 assert(MDOC_BLOCK == p->parent->type);
 398                 p->parent->tail = p;
 399                 break;
 400         case (MDOC_BODY):
 401                 if (p->end)
 402                         break;
 403                 assert(MDOC_BLOCK == p->parent->type);
 404                 p->parent->body = p;
 405                 break;
 406         default:
 407                 break;
 408         }
 409
 410         mdoc->last = p;
 411
 412         switch (p->type) {
 413         case (MDOC_TBL):
 414                 /* FALLTHROUGH */
 415         case (MDOC_TEXT):
 416                 if ( ! mdoc_valid_post(mdoc))
 417                         return(0);
 418                 break;
 419         default:
 420                 break;
 421         }
 422
 423         return(1);
 424 }
 425
 426
 427 static struct mdoc_node *
 428 node_alloc(struct mdoc *mdoc, int line, int pos,
 429                 enum mdoct tok, enum mdoc_type type)
 430 {
 431         struct mdoc_node *p;
 432
 433         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 434         p->sec = mdoc->lastsec;
 435         p->line = line;
 436         p->pos = pos;
 437         p->tok = tok;
 438         p->type = type;
 439
 440         /* Flag analysis. */
 441
 442         if (MDOC_SYNOPSIS & mdoc->flags)
 443                 p->flags |= MDOC_SYNPRETTY;
 444         else
 445                 p->flags &= ~MDOC_SYNPRETTY;
 446         if (MDOC_NEWLINE & mdoc->flags)
 447                 p->flags |= MDOC_LINE;
 448         mdoc->flags &= ~MDOC_NEWLINE;
 449
 450         return(p);
 451 }
 452
 453
 454 int
 455 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 456 {
 457         struct mdoc_node *p;
 458
 459         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 460         if ( ! node_append(mdoc, p))
 461                 return(0);
 462         mdoc->next = MDOC_NEXT_CHILD;
 463         return(1);
 464 }
 465
 466
 467 int
 468 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 469 {
 470         struct mdoc_node *p;
 471
 472         assert(mdoc->first);
 473         assert(mdoc->last);
 474
 475         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 476         if ( ! node_append(mdoc, p))
 477                 return(0);
 478         mdoc->next = MDOC_NEXT_CHILD;
 479         return(1);
 480 }
 481
 482
 483 int
 484 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 485 {
 486         struct mdoc_node *p;
 487
 488         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 489         if ( ! node_append(mdoc, p))
 490                 return(0);
 491         mdoc->next = MDOC_NEXT_CHILD;
 492         return(1);
 493 }
 494
 495
 496 int
 497 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 498                 struct mdoc_node *body, enum mdoc_endbody end)
 499 {
 500         struct mdoc_node *p;
 501
 502         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 503         p->pending = body;
 504         p->norm = body->norm;
 505         p->end = end;
 506         if ( ! node_append(mdoc, p))
 507                 return(0);
 508         mdoc->next = MDOC_NEXT_SIBLING;
 509         return(1);
 510 }
 511
 512
 513 int
 514 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 515                 enum mdoct tok, struct mdoc_arg *args)
 516 {
 517         struct mdoc_node *p;
 518
 519         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 520         p->args = args;
 521         if (p->args)
 522                 (args->refcnt)++;
 523
 524         switch (tok) {
 525         case (MDOC_Bd):
 526                 /* FALLTHROUGH */
 527         case (MDOC_Bf):
 528                 /* FALLTHROUGH */
 529         case (MDOC_Bl):
 530                 /* FALLTHROUGH */
 531         case (MDOC_Rs):
 532                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 533                 break;
 534         default:
 535                 break;
 536         }
 537
 538         if ( ! node_append(mdoc, p))
 539                 return(0);
 540         mdoc->next = MDOC_NEXT_CHILD;
 541         return(1);
 542 }
 543
 544
 545 int
 546 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 547                 enum mdoct tok, struct mdoc_arg *args)
 548 {
 549         struct mdoc_node *p;
 550
 551         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 552         p->args = args;
 553         if (p->args)
 554                 (args->refcnt)++;
 555
 556         switch (tok) {
 557         case (MDOC_An):
 558                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 559                 break;
 560         default:
 561                 break;
 562         }
 563
 564         if ( ! node_append(mdoc, p))
 565                 return(0);
 566         mdoc->next = MDOC_NEXT_CHILD;
 567         return(1);
 568 }
 569
 570 int
 571 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 572 {
 573         struct mdoc_node *n;
 574
 575         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 576         n->string = roff_strdup(mdoc->roff, p);
 577
 578         if ( ! node_append(mdoc, n))
 579                 return(0);
 580
 581         mdoc->next = MDOC_NEXT_SIBLING;
 582         return(1);
 583 }
 584
 585
 586 static void
 587 mdoc_node_free(struct mdoc_node *p)
 588 {
 589
 590         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 591                 free(p->norm);
 592         if (p->string)
 593                 free(p->string);
 594         if (p->args)
 595                 mdoc_argv_free(p->args);
 596         free(p);
 597 }
 598
 599
 600 static void
 601 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 602 {
 603
 604         /* Adjust siblings. */
 605
 606         if (n->prev)
 607                 n->prev->next = n->next;
 608         if (n->next)
 609                 n->next->prev = n->prev;
 610
 611         /* Adjust parent. */
 612
 613         if (n->parent) {
 614                 n->parent->nchild--;
 615                 if (n->parent->child == n)
 616                         n->parent->child = n->prev ? n->prev : n->next;
 617                 if (n->parent->last == n)
 618                         n->parent->last = n->prev ? n->prev : NULL;
 619         }
 620
 621         /* Adjust parse point, if applicable. */
 622
 623         if (mdoc && mdoc->last == n) {
 624                 if (n->prev) {
 625                         mdoc->last = n->prev;
 626                         mdoc->next = MDOC_NEXT_SIBLING;
 627                 } else {
 628                         mdoc->last = n->parent;
 629                         mdoc->next = MDOC_NEXT_CHILD;
 630                 }
 631         }
 632
 633         if (mdoc && mdoc->first == n)
 634                 mdoc->first = NULL;
 635 }
 636
 637
 638 void
 639 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 640 {
 641
 642         while (p->child) {
 643                 assert(p->nchild);
 644                 mdoc_node_delete(mdoc, p->child);
 645         }
 646         assert(0 == p->nchild);
 647
 648         mdoc_node_unlink(mdoc, p);
 649         mdoc_node_free(p);
 650 }
 651
 652 int
 653 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 654 {
 655
 656         mdoc_node_unlink(mdoc, p);
 657         return(node_append(mdoc, p));
 658 }
 659
 660 #if 0
 661 /*
 662  * Pre-treat a text line.
 663  * Text lines can consist of equations, which must be handled apart from
 664  * the regular text.
 665  * Thus, use this function to step through a line checking if it has any
 666  * equations embedded in it.
 667  * This must handle multiple equations AND equations that do not end at
 668  * the end-of-line, i.e., will re-enter in the next roff parse.
 669  */
 670 static int
 671 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
 672 {
 673         char            *start, *end;
 674         char             delim;
 675
 676         while ('\0' != buf[offs]) {
 677                 /* Mark starting position if eqn is set. */
 678                 start = NULL;
 679                 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
 680                         if (NULL != (start = strchr(buf + offs, delim)))
 681                                 *start++ = '\0';
 682
 683                 /* Parse text as normal. */
 684                 if ( ! mdoc_ptext(mdoc, line, buf, offs))
 685                         return(0);
 686
 687                 /* Continue only if an equation exists. */
 688                 if (NULL == start)
 689                         break;
 690
 691                 /* Read past the end of the equation. */
 692                 offs += start - (buf + offs);
 693                 assert(start == &buf[offs]);
 694                 if (NULL != (end = strchr(buf + offs, delim))) {
 695                         *end++ = '\0';
 696                         while (' ' == *end)
 697                                 end++;
 698                 }
 699
 700                 /* Parse the equation itself. */
 701                 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
 702
 703                 /* Process a finished equation? */
 704                 if (roff_closeeqn(mdoc->roff))
 705                         if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
 706                                 return(0);
 707                 offs += (end - (buf + offs));
 708         }
 709
 710         return(1);
 711 }
 712 #endif
 713
 714 /*
 715  * Parse free-form text, that is, a line that does not begin with the
 716  * control character.
 717  */
 718 static int
 719 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 720 {
 721         char             *c, *ws, *end;
 722         struct mdoc_node *n;
 723
 724         /* No text before an initial macro. */
 725
 726         if (SEC_NONE == mdoc->lastnamed) {
 727                 mdoc_pmsg(mdoc, line, offs, MANDOCERR_NOTEXT);
 728                 return(1);
 729         }
 730
 731         assert(mdoc->last);
 732         n = mdoc->last;
 733
 734         /*
 735          * Divert directly to list processing if we're encountering a
 736          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 737          * (a MDOC_BODY means it's already open, in which case we should
 738          * process within its context in the normal way).
 739          */
 740
 741         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 742                         LIST_column == n->norm->Bl.type) {
 743                 /* `Bl' is open without any children. */
 744                 mdoc->flags |= MDOC_FREECOL;
 745                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 746         }
 747
 748         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 749                         NULL != n->parent &&
 750                         MDOC_Bl == n->parent->tok &&
 751                         LIST_column == n->parent->norm->Bl.type) {
 752                 /* `Bl' has block-level `It' children. */
 753                 mdoc->flags |= MDOC_FREECOL;
 754                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 755         }
 756
 757         /*
 758          * Search for the beginning of unescaped trailing whitespace (ws)
 759          * and for the first character not to be output (end).
 760          */
 761
 762         /* FIXME: replace with strcspn(). */
 763         ws = NULL;
 764         for (c = end = buf + offs; *c; c++) {
 765                 switch (*c) {
 766                 case ' ':
 767                         if (NULL == ws)
 768                                 ws = c;
 769                         continue;
 770                 case '\t':
 771                         /*
 772                          * Always warn about trailing tabs,
 773                          * even outside literal context,
 774                          * where they should be put on the next line.
 775                          */
 776                         if (NULL == ws)
 777                                 ws = c;
 778                         /*
 779                          * Strip trailing tabs in literal context only;
 780                          * outside, they affect the next line.
 781                          */
 782                         if (MDOC_LITERAL & mdoc->flags)
 783                                 continue;
 784                         break;
 785                 case '\\':
 786                         /* Skip the escaped character, too, if any. */
 787                         if (c[1])
 788                                 c++;
 789                         /* FALLTHROUGH */
 790                 default:
 791                         ws = NULL;
 792                         break;
 793                 }
 794                 end = c + 1;
 795         }
 796         *end = '\0';
 797
 798         if (ws)
 799                 mdoc_pmsg(mdoc, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
 800
 801         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 802                 mdoc_pmsg(mdoc, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
 803
 804                 /*
 805                  * Insert a `sp' in the case of a blank line.  Technically,
 806                  * blank lines aren't allowed, but enough manuals assume this
 807                  * behaviour that we want to work around it.
 808                  */
 809                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 810                         return(0);
 811
 812                 mdoc->next = MDOC_NEXT_SIBLING;
 813
 814                 return(mdoc_valid_post(mdoc));
 815         }
 816
 817         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 818                 return(0);
 819
 820         if (MDOC_LITERAL & mdoc->flags)
 821                 return(1);
 822
 823         /*
 824          * End-of-sentence check.  If the last character is an unescaped
 825          * EOS character, then flag the node as being the end of a
 826          * sentence.  The front-end will know how to interpret this.
 827          */
 828
 829         assert(buf < end);
 830
 831         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs), 0))
 832                 mdoc->last->flags |= MDOC_EOS;
 833
 834         return(1);
 835 }
 836
 837
 838 /*
 839  * Parse a macro line, that is, a line beginning with the control
 840  * character.
 841  */
 842 static int
 843 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 844 {
 845         enum mdoct        tok;
 846         int               i, sv;
 847         char              mac[5];
 848         struct mdoc_node *n;
 849
 850         /* Empty post-control lines are ignored. */
 851
 852         if ('"' == buf[offs]) {
 853                 mdoc_pmsg(mdoc, ln, offs, MANDOCERR_BADCOMMENT);
 854                 return(1);
 855         } else if ('\0' == buf[offs])
 856                 return(1);
 857
 858         sv = offs;
 859
 860         /*
 861          * Copy the first word into a nil-terminated buffer.
 862          * Stop copying when a tab, space, or eoln is encountered.
 863          */
 864
 865         i = 0;
 866         while (i < 4 && '\0' != buf[offs] &&
 867                         ' ' != buf[offs] && '\t' != buf[offs])
 868                 mac[i++] = buf[offs++];
 869
 870         mac[i] = '\0';
 871
 872         tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 873
 874         if (MDOC_MAX == tok) {
 875                 mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
 876                                 ln, sv, "%s", buf + sv - 1);
 877                 return(1);
 878         }
 879
 880         /* Disregard the first trailing tab, if applicable. */
 881
 882         if ('\t' == buf[offs])
 883                 offs++;
 884
 885         /* Jump to the next non-whitespace word. */
 886
 887         while (buf[offs] && ' ' == buf[offs])
 888                 offs++;
 889
 890         /*
 891          * Trailing whitespace.  Note that tabs are allowed to be passed
 892          * into the parser as "text", so we only warn about spaces here.
 893          */
 894
 895         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 896                 mdoc_pmsg(mdoc, ln, offs - 1, MANDOCERR_EOLNSPACE);
 897
 898         /*
 899          * If an initial macro or a list invocation, divert directly
 900          * into macro processing.
 901          */
 902
 903         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 904                 if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 905                         goto err;
 906                 return(1);
 907         }
 908
 909         n = mdoc->last;
 910         assert(mdoc->last);
 911
 912         /*
 913          * If the first macro of a `Bl -column', open an `It' block
 914          * context around the parsed macro.
 915          */
 916
 917         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 918                         LIST_column == n->norm->Bl.type) {
 919                 mdoc->flags |= MDOC_FREECOL;
 920                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 921                         goto err;
 922                 return(1);
 923         }
 924
 925         /*
 926          * If we're following a block-level `It' within a `Bl -column'
 927          * context (perhaps opened in the above block or in ptext()),
 928          * then open an `It' block context around the parsed macro.
 929          */
 930
 931         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 932                         NULL != n->parent &&
 933                         MDOC_Bl == n->parent->tok &&
 934                         LIST_column == n->parent->norm->Bl.type) {
 935                 mdoc->flags |= MDOC_FREECOL;
 936                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 937                         goto err;
 938                 return(1);
 939         }
 940
 941         /* Normal processing of a macro. */
 942
 943         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 944                 goto err;
 945
 946         return(1);
 947
 948 err:    /* Error out. */
 949
 950         mdoc->flags |= MDOC_HALT;
 951         return(0);
 952 }
 953
 954 enum mdelim
 955 mdoc_isdelim(const char *p)
 956 {
 957
 958         if ('\0' == p[0])
 959                 return(DELIM_NONE);
 960
 961         if ('\0' == p[1])
 962                 switch (p[0]) {
 963                 case('('):
 964                         /* FALLTHROUGH */
 965                 case('['):
 966                         return(DELIM_OPEN);
 967                 case('|'):
 968                         return(DELIM_MIDDLE);
 969                 case('.'):
 970                         /* FALLTHROUGH */
 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                         return(DELIM_CLOSE);
 985                 default:
 986                         return(DELIM_NONE);
 987                 }
 988
 989         if ('\\' != p[0])
 990                 return(DELIM_NONE);
 991
 992         if (0 == strcmp(p + 1, "."))
 993                 return(DELIM_CLOSE);
 994         if (0 == strcmp(p + 1, "fR|\\fP"))
 995                 return(DELIM_MIDDLE);
 996
 997         return(DELIM_NONE);
 998 }