mdoc.c

   1 /*      $Id: mdoc.c,v 1.208 2014/01/05 20:26:36 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 <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,
 201         char *defos, int quick)
 202 {
 203         struct mdoc     *p;
 204
 205         p = mandoc_calloc(1, sizeof(struct mdoc));
 206
 207         p->parse = parse;
 208         p->defos = defos;
 209         p->quick = quick;
 210         p->roff = roff;
 211
 212         mdoc_hash_init();
 213         mdoc_alloc1(p);
 214         return(p);
 215 }
 216
 217
 218 /*
 219  * Climb back up the parse tree, validating open scopes.  Mostly calls
 220  * through to macro_end() in macro.c.
 221  */
 222 int
 223 mdoc_endparse(struct mdoc *mdoc)
 224 {
 225
 226         assert( ! (MDOC_HALT & mdoc->flags));
 227         if (mdoc_macroend(mdoc))
 228                 return(1);
 229         mdoc->flags |= MDOC_HALT;
 230         return(0);
 231 }
 232
 233 int
 234 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 235 {
 236         struct mdoc_node *n;
 237
 238         assert( ! (MDOC_HALT & mdoc->flags));
 239
 240         /* No text before an initial macro. */
 241
 242         if (SEC_NONE == mdoc->lastnamed) {
 243                 mdoc_pmsg(mdoc, ep->ln, ep->pos, MANDOCERR_NOTEXT);
 244                 return(1);
 245         }
 246
 247         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 248         n->eqn = ep;
 249
 250         if ( ! node_append(mdoc, n))
 251                 return(0);
 252
 253         mdoc->next = MDOC_NEXT_SIBLING;
 254         return(1);
 255 }
 256
 257 int
 258 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 259 {
 260         struct mdoc_node *n;
 261
 262         assert( ! (MDOC_HALT & mdoc->flags));
 263
 264         /* No text before an initial macro. */
 265
 266         if (SEC_NONE == mdoc->lastnamed) {
 267                 mdoc_pmsg(mdoc, sp->line, 0, MANDOCERR_NOTEXT);
 268                 return(1);
 269         }
 270
 271         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 272         n->span = sp;
 273
 274         if ( ! node_append(mdoc, n))
 275                 return(0);
 276
 277         mdoc->next = MDOC_NEXT_SIBLING;
 278         return(1);
 279 }
 280
 281
 282 /*
 283  * Main parse routine.  Parses a single line -- really just hands off to
 284  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 285  */
 286 int
 287 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 288 {
 289
 290         assert( ! (MDOC_HALT & mdoc->flags));
 291
 292         mdoc->flags |= MDOC_NEWLINE;
 293
 294         /*
 295          * Let the roff nS register switch SYNOPSIS mode early,
 296          * such that the parser knows at all times
 297          * whether this mode is on or off.
 298          * Note that this mode is also switched by the Sh macro.
 299          */
 300         if (roff_getreg(mdoc->roff, "nS"))
 301                 mdoc->flags |= MDOC_SYNOPSIS;
 302         else
 303                 mdoc->flags &= ~MDOC_SYNOPSIS;
 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->lastline = line;
 440         p->tok = tok;
 441         p->type = type;
 442
 443         /* Flag analysis. */
 444
 445         if (MDOC_SYNOPSIS & mdoc->flags)
 446                 p->flags |= MDOC_SYNPRETTY;
 447         else
 448                 p->flags &= ~MDOC_SYNPRETTY;
 449         if (MDOC_NEWLINE & mdoc->flags)
 450                 p->flags |= MDOC_LINE;
 451         mdoc->flags &= ~MDOC_NEWLINE;
 452
 453         return(p);
 454 }
 455
 456
 457 int
 458 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 459 {
 460         struct mdoc_node *p;
 461
 462         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 463         if ( ! node_append(mdoc, p))
 464                 return(0);
 465         mdoc->next = MDOC_NEXT_CHILD;
 466         return(1);
 467 }
 468
 469
 470 int
 471 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 472 {
 473         struct mdoc_node *p;
 474
 475         assert(mdoc->first);
 476         assert(mdoc->last);
 477
 478         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 479         if ( ! node_append(mdoc, p))
 480                 return(0);
 481         mdoc->next = MDOC_NEXT_CHILD;
 482         return(1);
 483 }
 484
 485
 486 int
 487 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 488 {
 489         struct mdoc_node *p;
 490
 491         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 492         if ( ! node_append(mdoc, p))
 493                 return(0);
 494         mdoc->next = MDOC_NEXT_CHILD;
 495         return(1);
 496 }
 497
 498
 499 int
 500 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 501                 struct mdoc_node *body, enum mdoc_endbody end)
 502 {
 503         struct mdoc_node *p;
 504
 505         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 506         p->pending = body;
 507         p->norm = body->norm;
 508         p->end = end;
 509         if ( ! node_append(mdoc, p))
 510                 return(0);
 511         mdoc->next = MDOC_NEXT_SIBLING;
 512         return(1);
 513 }
 514
 515
 516 int
 517 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 518                 enum mdoct tok, struct mdoc_arg *args)
 519 {
 520         struct mdoc_node *p;
 521
 522         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 523         p->args = args;
 524         if (p->args)
 525                 (args->refcnt)++;
 526
 527         switch (tok) {
 528         case (MDOC_Bd):
 529                 /* FALLTHROUGH */
 530         case (MDOC_Bf):
 531                 /* FALLTHROUGH */
 532         case (MDOC_Bl):
 533                 /* FALLTHROUGH */
 534         case (MDOC_Rs):
 535                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 536                 break;
 537         default:
 538                 break;
 539         }
 540
 541         if ( ! node_append(mdoc, p))
 542                 return(0);
 543         mdoc->next = MDOC_NEXT_CHILD;
 544         return(1);
 545 }
 546
 547
 548 int
 549 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 550                 enum mdoct tok, struct mdoc_arg *args)
 551 {
 552         struct mdoc_node *p;
 553
 554         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 555         p->args = args;
 556         if (p->args)
 557                 (args->refcnt)++;
 558
 559         switch (tok) {
 560         case (MDOC_An):
 561                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 562                 break;
 563         default:
 564                 break;
 565         }
 566
 567         if ( ! node_append(mdoc, p))
 568                 return(0);
 569         mdoc->next = MDOC_NEXT_CHILD;
 570         return(1);
 571 }
 572
 573 int
 574 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 575 {
 576         struct mdoc_node *n;
 577
 578         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 579         n->string = roff_strdup(mdoc->roff, p);
 580
 581         if ( ! node_append(mdoc, n))
 582                 return(0);
 583
 584         mdoc->next = MDOC_NEXT_SIBLING;
 585         return(1);
 586 }
 587
 588 void
 589 mdoc_word_append(struct mdoc *mdoc, const char *p)
 590 {
 591         struct mdoc_node        *n;
 592         char                    *addstr, *newstr;
 593
 594         n = mdoc->last;
 595         addstr = roff_strdup(mdoc->roff, p);
 596         if (-1 == asprintf(&newstr, "%s %s", n->string, addstr)) {
 597                 perror(NULL);
 598                 exit((int)MANDOCLEVEL_SYSERR);
 599         }
 600         free(addstr);
 601         free(n->string);
 602         n->string = newstr;
 603         mdoc->next = MDOC_NEXT_SIBLING;
 604 }
 605
 606 static void
 607 mdoc_node_free(struct mdoc_node *p)
 608 {
 609
 610         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 611                 free(p->norm);
 612         if (p->string)
 613                 free(p->string);
 614         if (p->args)
 615                 mdoc_argv_free(p->args);
 616         free(p);
 617 }
 618
 619
 620 static void
 621 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 622 {
 623
 624         /* Adjust siblings. */
 625
 626         if (n->prev)
 627                 n->prev->next = n->next;
 628         if (n->next)
 629                 n->next->prev = n->prev;
 630
 631         /* Adjust parent. */
 632
 633         if (n->parent) {
 634                 n->parent->nchild--;
 635                 if (n->parent->child == n)
 636                         n->parent->child = n->prev ? n->prev : n->next;
 637                 if (n->parent->last == n)
 638                         n->parent->last = n->prev ? n->prev : NULL;
 639         }
 640
 641         /* Adjust parse point, if applicable. */
 642
 643         if (mdoc && mdoc->last == n) {
 644                 if (n->prev) {
 645                         mdoc->last = n->prev;
 646                         mdoc->next = MDOC_NEXT_SIBLING;
 647                 } else {
 648                         mdoc->last = n->parent;
 649                         mdoc->next = MDOC_NEXT_CHILD;
 650                 }
 651         }
 652
 653         if (mdoc && mdoc->first == n)
 654                 mdoc->first = NULL;
 655 }
 656
 657
 658 void
 659 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 660 {
 661
 662         while (p->child) {
 663                 assert(p->nchild);
 664                 mdoc_node_delete(mdoc, p->child);
 665         }
 666         assert(0 == p->nchild);
 667
 668         mdoc_node_unlink(mdoc, p);
 669         mdoc_node_free(p);
 670 }
 671
 672 int
 673 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 674 {
 675
 676         mdoc_node_unlink(mdoc, p);
 677         return(node_append(mdoc, p));
 678 }
 679
 680 #if 0
 681 /*
 682  * Pre-treat a text line.
 683  * Text lines can consist of equations, which must be handled apart from
 684  * the regular text.
 685  * Thus, use this function to step through a line checking if it has any
 686  * equations embedded in it.
 687  * This must handle multiple equations AND equations that do not end at
 688  * the end-of-line, i.e., will re-enter in the next roff parse.
 689  */
 690 static int
 691 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
 692 {
 693         char            *start, *end;
 694         char             delim;
 695
 696         while ('\0' != buf[offs]) {
 697                 /* Mark starting position if eqn is set. */
 698                 start = NULL;
 699                 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
 700                         if (NULL != (start = strchr(buf + offs, delim)))
 701                                 *start++ = '\0';
 702
 703                 /* Parse text as normal. */
 704                 if ( ! mdoc_ptext(mdoc, line, buf, offs))
 705                         return(0);
 706
 707                 /* Continue only if an equation exists. */
 708                 if (NULL == start)
 709                         break;
 710
 711                 /* Read past the end of the equation. */
 712                 offs += start - (buf + offs);
 713                 assert(start == &buf[offs]);
 714                 if (NULL != (end = strchr(buf + offs, delim))) {
 715                         *end++ = '\0';
 716                         while (' ' == *end)
 717                                 end++;
 718                 }
 719
 720                 /* Parse the equation itself. */
 721                 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
 722
 723                 /* Process a finished equation? */
 724                 if (roff_closeeqn(mdoc->roff))
 725                         if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
 726                                 return(0);
 727                 offs += (end - (buf + offs));
 728         }
 729
 730         return(1);
 731 }
 732 #endif
 733
 734 /*
 735  * Parse free-form text, that is, a line that does not begin with the
 736  * control character.
 737  */
 738 static int
 739 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 740 {
 741         char             *c, *ws, *end;
 742         struct mdoc_node *n;
 743
 744         /* No text before an initial macro. */
 745
 746         if (SEC_NONE == mdoc->lastnamed) {
 747                 mdoc_pmsg(mdoc, line, offs, MANDOCERR_NOTEXT);
 748                 return(1);
 749         }
 750
 751         assert(mdoc->last);
 752         n = mdoc->last;
 753
 754         /*
 755          * Divert directly to list processing if we're encountering a
 756          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 757          * (a MDOC_BODY means it's already open, in which case we should
 758          * process within its context in the normal way).
 759          */
 760
 761         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 762                         LIST_column == n->norm->Bl.type) {
 763                 /* `Bl' is open without any children. */
 764                 mdoc->flags |= MDOC_FREECOL;
 765                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 766         }
 767
 768         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 769                         NULL != n->parent &&
 770                         MDOC_Bl == n->parent->tok &&
 771                         LIST_column == n->parent->norm->Bl.type) {
 772                 /* `Bl' has block-level `It' children. */
 773                 mdoc->flags |= MDOC_FREECOL;
 774                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 775         }
 776
 777         /*
 778          * Search for the beginning of unescaped trailing whitespace (ws)
 779          * and for the first character not to be output (end).
 780          */
 781
 782         /* FIXME: replace with strcspn(). */
 783         ws = NULL;
 784         for (c = end = buf + offs; *c; c++) {
 785                 switch (*c) {
 786                 case ' ':
 787                         if (NULL == ws)
 788                                 ws = c;
 789                         continue;
 790                 case '\t':
 791                         /*
 792                          * Always warn about trailing tabs,
 793                          * even outside literal context,
 794                          * where they should be put on the next line.
 795                          */
 796                         if (NULL == ws)
 797                                 ws = c;
 798                         /*
 799                          * Strip trailing tabs in literal context only;
 800                          * outside, they affect the next line.
 801                          */
 802                         if (MDOC_LITERAL & mdoc->flags)
 803                                 continue;
 804                         break;
 805                 case '\\':
 806                         /* Skip the escaped character, too, if any. */
 807                         if (c[1])
 808                                 c++;
 809                         /* FALLTHROUGH */
 810                 default:
 811                         ws = NULL;
 812                         break;
 813                 }
 814                 end = c + 1;
 815         }
 816         *end = '\0';
 817
 818         if (ws)
 819                 mdoc_pmsg(mdoc, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
 820
 821         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 822                 mdoc_pmsg(mdoc, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
 823
 824                 /*
 825                  * Insert a `sp' in the case of a blank line.  Technically,
 826                  * blank lines aren't allowed, but enough manuals assume this
 827                  * behaviour that we want to work around it.
 828                  */
 829                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 830                         return(0);
 831
 832                 mdoc->next = MDOC_NEXT_SIBLING;
 833
 834                 return(mdoc_valid_post(mdoc));
 835         }
 836
 837         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 838                 return(0);
 839
 840         if (MDOC_LITERAL & mdoc->flags)
 841                 return(1);
 842
 843         /*
 844          * End-of-sentence check.  If the last character is an unescaped
 845          * EOS character, then flag the node as being the end of a
 846          * sentence.  The front-end will know how to interpret this.
 847          */
 848
 849         assert(buf < end);
 850
 851         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 852                 mdoc->last->flags |= MDOC_EOS;
 853
 854         return(1);
 855 }
 856
 857
 858 /*
 859  * Parse a macro line, that is, a line beginning with the control
 860  * character.
 861  */
 862 static int
 863 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 864 {
 865         enum mdoct        tok;
 866         int               i, sv;
 867         char              mac[5];
 868         struct mdoc_node *n;
 869
 870         /* Empty post-control lines are ignored. */
 871
 872         if ('"' == buf[offs]) {
 873                 mdoc_pmsg(mdoc, ln, offs, MANDOCERR_BADCOMMENT);
 874                 return(1);
 875         } else if ('\0' == buf[offs])
 876                 return(1);
 877
 878         sv = offs;
 879
 880         /*
 881          * Copy the first word into a nil-terminated buffer.
 882          * Stop copying when a tab, space, or eoln is encountered.
 883          */
 884
 885         i = 0;
 886         while (i < 4 && '\0' != buf[offs] &&
 887                         ' ' != buf[offs] && '\t' != buf[offs])
 888                 mac[i++] = buf[offs++];
 889
 890         mac[i] = '\0';
 891
 892         tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 893
 894         if (MDOC_MAX == tok) {
 895                 mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
 896                                 ln, sv, "%s", buf + sv - 1);
 897                 return(1);
 898         }
 899
 900         /* Disregard the first trailing tab, if applicable. */
 901
 902         if ('\t' == buf[offs])
 903                 offs++;
 904
 905         /* Jump to the next non-whitespace word. */
 906
 907         while (buf[offs] && ' ' == buf[offs])
 908                 offs++;
 909
 910         /*
 911          * Trailing whitespace.  Note that tabs are allowed to be passed
 912          * into the parser as "text", so we only warn about spaces here.
 913          */
 914
 915         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 916                 mdoc_pmsg(mdoc, ln, offs - 1, MANDOCERR_EOLNSPACE);
 917
 918         /*
 919          * If an initial macro or a list invocation, divert directly
 920          * into macro processing.
 921          */
 922
 923         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 924                 if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 925                         goto err;
 926                 return(1);
 927         }
 928
 929         n = mdoc->last;
 930         assert(mdoc->last);
 931
 932         /*
 933          * If the first macro of a `Bl -column', open an `It' block
 934          * context around the parsed macro.
 935          */
 936
 937         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 938                         LIST_column == n->norm->Bl.type) {
 939                 mdoc->flags |= MDOC_FREECOL;
 940                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 941                         goto err;
 942                 return(1);
 943         }
 944
 945         /*
 946          * If we're following a block-level `It' within a `Bl -column'
 947          * context (perhaps opened in the above block or in ptext()),
 948          * then open an `It' block context around the parsed macro.
 949          */
 950
 951         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 952                         NULL != n->parent &&
 953                         MDOC_Bl == n->parent->tok &&
 954                         LIST_column == n->parent->norm->Bl.type) {
 955                 mdoc->flags |= MDOC_FREECOL;
 956                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 957                         goto err;
 958                 return(1);
 959         }
 960
 961         /* Normal processing of a macro. */
 962
 963         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 964                 goto err;
 965
 966         /* In quick mode (for mandocdb), abort after the NAME section. */
 967
 968         if (mdoc->quick && MDOC_Sh == tok &&
 969             SEC_NAME != mdoc->last->sec)
 970                 return(2);
 971
 972         return(1);
 973
 974 err:    /* Error out. */
 975
 976         mdoc->flags |= MDOC_HALT;
 977         return(0);
 978 }
 979
 980 enum mdelim
 981 mdoc_isdelim(const char *p)
 982 {
 983
 984         if ('\0' == p[0])
 985                 return(DELIM_NONE);
 986
 987         if ('\0' == p[1])
 988                 switch (p[0]) {
 989                 case('('):
 990                         /* FALLTHROUGH */
 991                 case('['):
 992                         return(DELIM_OPEN);
 993                 case('|'):
 994                         return(DELIM_MIDDLE);
 995                 case('.'):
 996                         /* FALLTHROUGH */
 997                 case(','):
 998                         /* FALLTHROUGH */
 999                 case(';'):
1000                         /* FALLTHROUGH */
1001                 case(':'):
1002                         /* FALLTHROUGH */
1003                 case('?'):
1004                         /* FALLTHROUGH */
1005                 case('!'):
1006                         /* FALLTHROUGH */
1007                 case(')'):
1008                         /* FALLTHROUGH */
1009                 case(']'):
1010                         return(DELIM_CLOSE);
1011                 default:
1012                         return(DELIM_NONE);
1013                 }
1014
1015         if ('\\' != p[0])
1016                 return(DELIM_NONE);
1017
1018         if (0 == strcmp(p + 1, "."))
1019                 return(DELIM_CLOSE);
1020         if (0 == strcmp(p + 1, "fR|\\fP"))
1021                 return(DELIM_MIDDLE);
1022
1023         return(DELIM_NONE);
1024 }