mdoc.c

   1 /*      $Id: mdoc.c,v 1.212 2014/03/30 19:47:48 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
   5  *
   6  * Permission to use, copy, modify, and distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  17  */
  18 #ifdef HAVE_CONFIG_H
  19 #include "config.h"
  20 #endif
  21
  22 #include <sys/types.h>
  23
  24 #include <assert.h>
  25 #include <ctype.h>
  26 #include <stdarg.h>
  27 #include <stdio.h>
  28 #include <stdlib.h>
  29 #include <string.h>
  30 #include <time.h>
  31
  32 #include "mdoc.h"
  33 #include "mandoc.h"
  34 #include "mandoc_aux.h"
  35 #include "libmdoc.h"
  36 #include "libmandoc.h"
  37
  38 const   char *const __mdoc_macronames[MDOC_MAX] = {
  39         "Ap",           "Dd",           "Dt",           "Os",
  40         "Sh",           "Ss",           "Pp",           "D1",
  41         "Dl",           "Bd",           "Ed",           "Bl",
  42         "El",           "It",           "Ad",           "An",
  43         "Ar",           "Cd",           "Cm",           "Dv",
  44         "Er",           "Ev",           "Ex",           "Fa",
  45         "Fd",           "Fl",           "Fn",           "Ft",
  46         "Ic",           "In",           "Li",           "Nd",
  47         "Nm",           "Op",           "Ot",           "Pa",
  48         "Rv",           "St",           "Va",           "Vt",
  49         /* LINTED */
  50         "Xr",           "%A",           "%B",           "%D",
  51         /* LINTED */
  52         "%I",           "%J",           "%N",           "%O",
  53         /* LINTED */
  54         "%P",           "%R",           "%T",           "%V",
  55         "Ac",           "Ao",           "Aq",           "At",
  56         "Bc",           "Bf",           "Bo",           "Bq",
  57         "Bsx",          "Bx",           "Db",           "Dc",
  58         "Do",           "Dq",           "Ec",           "Ef",
  59         "Em",           "Eo",           "Fx",           "Ms",
  60         "No",           "Ns",           "Nx",           "Ox",
  61         "Pc",           "Pf",           "Po",           "Pq",
  62         "Qc",           "Ql",           "Qo",           "Qq",
  63         "Re",           "Rs",           "Sc",           "So",
  64         "Sq",           "Sm",           "Sx",           "Sy",
  65         "Tn",           "Ux",           "Xc",           "Xo",
  66         "Fo",           "Fc",           "Oo",           "Oc",
  67         "Bk",           "Ek",           "Bt",           "Hf",
  68         "Fr",           "Ud",           "Lb",           "Lp",
  69         "Lk",           "Mt",           "Brq",          "Bro",
  70         /* LINTED */
  71         "Brc",          "%C",           "Es",           "En",
  72         /* LINTED */
  73         "Dx",           "%Q",           "br",           "sp",
  74         /* LINTED */
  75         "%U",           "Ta",           "ll",
  76         };
  77
  78 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  79         "split",                "nosplit",              "ragged",
  80         "unfilled",             "literal",              "file",
  81         "offset",               "bullet",               "dash",
  82         "hyphen",               "item",                 "enum",
  83         "tag",                  "diag",                 "hang",
  84         "ohang",                "inset",                "column",
  85         "width",                "compact",              "std",
  86         "filled",               "words",                "emphasis",
  87         "symbolic",             "nested",               "centered"
  88         };
  89
  90 const   char * const *mdoc_macronames = __mdoc_macronames;
  91 const   char * const *mdoc_argnames = __mdoc_argnames;
  92
  93 static  void              mdoc_node_free(struct mdoc_node *);
  94 static  void              mdoc_node_unlink(struct mdoc *,
  95                                 struct mdoc_node *);
  96 static  void              mdoc_free1(struct mdoc *);
  97 static  void              mdoc_alloc1(struct mdoc *);
  98 static  struct mdoc_node *node_alloc(struct mdoc *, int, int,
  99                                 enum mdoct, enum mdoc_type);
 100 static  int               node_append(struct mdoc *,
 101                                 struct mdoc_node *);
 102 #if 0
 103 static  int               mdoc_preptext(struct mdoc *, int, char *, int);
 104 #endif
 105 static  int               mdoc_ptext(struct mdoc *, int, char *, int);
 106 static  int               mdoc_pmacro(struct mdoc *, int, char *, int);
 107
 108 const struct mdoc_node *
 109 mdoc_node(const struct mdoc *mdoc)
 110 {
 111
 112         assert( ! (MDOC_HALT & mdoc->flags));
 113         return(mdoc->first);
 114 }
 115
 116
 117 const struct mdoc_meta *
 118 mdoc_meta(const struct mdoc *mdoc)
 119 {
 120
 121         assert( ! (MDOC_HALT & mdoc->flags));
 122         return(&mdoc->meta);
 123 }
 124
 125
 126 /*
 127  * Frees volatile resources (parse tree, meta-data, fields).
 128  */
 129 static void
 130 mdoc_free1(struct mdoc *mdoc)
 131 {
 132
 133         if (mdoc->first)
 134                 mdoc_node_delete(mdoc, mdoc->first);
 135         if (mdoc->meta.title)
 136                 free(mdoc->meta.title);
 137         if (mdoc->meta.os)
 138                 free(mdoc->meta.os);
 139         if (mdoc->meta.name)
 140                 free(mdoc->meta.name);
 141         if (mdoc->meta.arch)
 142                 free(mdoc->meta.arch);
 143         if (mdoc->meta.vol)
 144                 free(mdoc->meta.vol);
 145         if (mdoc->meta.msec)
 146                 free(mdoc->meta.msec);
 147         if (mdoc->meta.date)
 148                 free(mdoc->meta.date);
 149 }
 150
 151
 152 /*
 153  * Allocate all volatile resources (parse tree, meta-data, fields).
 154  */
 155 static void
 156 mdoc_alloc1(struct mdoc *mdoc)
 157 {
 158
 159         memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
 160         mdoc->flags = 0;
 161         mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 162         mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
 163         mdoc->first = mdoc->last;
 164         mdoc->last->type = MDOC_ROOT;
 165         mdoc->last->tok = MDOC_MAX;
 166         mdoc->next = MDOC_NEXT_CHILD;
 167 }
 168
 169
 170 /*
 171  * Free up volatile resources (see mdoc_free1()) then re-initialises the
 172  * data with mdoc_alloc1().  After invocation, parse data has been reset
 173  * and the parser is ready for re-invocation on a new tree; however,
 174  * cross-parse non-volatile data is kept intact.
 175  */
 176 void
 177 mdoc_reset(struct mdoc *mdoc)
 178 {
 179
 180         mdoc_free1(mdoc);
 181         mdoc_alloc1(mdoc);
 182 }
 183
 184
 185 /*
 186  * Completely free up all volatile and non-volatile parse resources.
 187  * After invocation, the pointer is no longer usable.
 188  */
 189 void
 190 mdoc_free(struct mdoc *mdoc)
 191 {
 192
 193         mdoc_free1(mdoc);
 194         free(mdoc);
 195 }
 196
 197
 198 /*
 199  * Allocate volatile and non-volatile parse resources.
 200  */
 201 struct mdoc *
 202 mdoc_alloc(struct roff *roff, struct mparse *parse,
 203         char *defos, int quick)
 204 {
 205         struct mdoc     *p;
 206
 207         p = mandoc_calloc(1, sizeof(struct mdoc));
 208
 209         p->parse = parse;
 210         p->defos = defos;
 211         p->quick = quick;
 212         p->roff = roff;
 213
 214         mdoc_hash_init();
 215         mdoc_alloc1(p);
 216         return(p);
 217 }
 218
 219
 220 /*
 221  * Climb back up the parse tree, validating open scopes.  Mostly calls
 222  * through to macro_end() in macro.c.
 223  */
 224 int
 225 mdoc_endparse(struct mdoc *mdoc)
 226 {
 227
 228         assert( ! (MDOC_HALT & mdoc->flags));
 229         if (mdoc_macroend(mdoc))
 230                 return(1);
 231         mdoc->flags |= MDOC_HALT;
 232         return(0);
 233 }
 234
 235 int
 236 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 237 {
 238         struct mdoc_node *n;
 239
 240         assert( ! (MDOC_HALT & mdoc->flags));
 241
 242         /* No text before an initial macro. */
 243
 244         if (SEC_NONE == mdoc->lastnamed) {
 245                 mdoc_pmsg(mdoc, ep->ln, ep->pos, MANDOCERR_NOTEXT);
 246                 return(1);
 247         }
 248
 249         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 250         n->eqn = ep;
 251
 252         if ( ! node_append(mdoc, n))
 253                 return(0);
 254
 255         mdoc->next = MDOC_NEXT_SIBLING;
 256         return(1);
 257 }
 258
 259 int
 260 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 261 {
 262         struct mdoc_node *n;
 263
 264         assert( ! (MDOC_HALT & mdoc->flags));
 265
 266         /* No text before an initial macro. */
 267
 268         if (SEC_NONE == mdoc->lastnamed) {
 269                 mdoc_pmsg(mdoc, sp->line, 0, MANDOCERR_NOTEXT);
 270                 return(1);
 271         }
 272
 273         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 274         n->span = sp;
 275
 276         if ( ! node_append(mdoc, n))
 277                 return(0);
 278
 279         mdoc->next = MDOC_NEXT_SIBLING;
 280         return(1);
 281 }
 282
 283
 284 /*
 285  * Main parse routine.  Parses a single line -- really just hands off to
 286  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 287  */
 288 int
 289 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 290 {
 291
 292         assert( ! (MDOC_HALT & mdoc->flags));
 293
 294         mdoc->flags |= MDOC_NEWLINE;
 295
 296         /*
 297          * Let the roff nS register switch SYNOPSIS mode early,
 298          * such that the parser knows at all times
 299          * whether this mode is on or off.
 300          * Note that this mode is also switched by the Sh macro.
 301          */
 302         if (roff_getreg(mdoc->roff, "nS"))
 303                 mdoc->flags |= MDOC_SYNOPSIS;
 304         else
 305                 mdoc->flags &= ~MDOC_SYNOPSIS;
 306
 307         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 308                         mdoc_pmacro(mdoc, ln, buf, offs) :
 309                         mdoc_ptext(mdoc, ln, buf, offs));
 310 }
 311
 312 int
 313 mdoc_macro(MACRO_PROT_ARGS)
 314 {
 315         assert(tok < MDOC_MAX);
 316
 317         /* If we're in the body, deny prologue calls. */
 318
 319         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 320                         MDOC_PBODY & mdoc->flags) {
 321                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADBODY);
 322                 return(1);
 323         }
 324
 325         /* If we're in the prologue, deny "body" macros.  */
 326
 327         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 328                         ! (MDOC_PBODY & mdoc->flags)) {
 329                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_BADPROLOG);
 330                 if (NULL == mdoc->meta.msec)
 331                         mdoc->meta.msec = mandoc_strdup("1");
 332                 if (NULL == mdoc->meta.title)
 333                         mdoc->meta.title = mandoc_strdup("UNKNOWN");
 334                 if (NULL == mdoc->meta.vol)
 335                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 336                 if (NULL == mdoc->meta.os)
 337                         mdoc->meta.os = mandoc_strdup("LOCAL");
 338                 if (NULL == mdoc->meta.date)
 339                         mdoc->meta.date = mandoc_normdate
 340                                 (mdoc->parse, NULL, line, ppos);
 341                 mdoc->flags |= MDOC_PBODY;
 342         }
 343
 344         return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
 345 }
 346
 347
 348 static int
 349 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 350 {
 351
 352         assert(mdoc->last);
 353         assert(mdoc->first);
 354         assert(MDOC_ROOT != p->type);
 355
 356         switch (mdoc->next) {
 357         case (MDOC_NEXT_SIBLING):
 358                 mdoc->last->next = p;
 359                 p->prev = mdoc->last;
 360                 p->parent = mdoc->last->parent;
 361                 break;
 362         case (MDOC_NEXT_CHILD):
 363                 mdoc->last->child = p;
 364                 p->parent = mdoc->last;
 365                 break;
 366         default:
 367                 abort();
 368                 /* NOTREACHED */
 369         }
 370
 371         p->parent->nchild++;
 372
 373         /*
 374          * Copy over the normalised-data pointer of our parent.  Not
 375          * everybody has one, but copying a null pointer is fine.
 376          */
 377
 378         switch (p->type) {
 379         case (MDOC_BODY):
 380                 if (ENDBODY_NOT != p->end)
 381                         break;
 382                 /* FALLTHROUGH */
 383         case (MDOC_TAIL):
 384                 /* FALLTHROUGH */
 385         case (MDOC_HEAD):
 386                 p->norm = p->parent->norm;
 387                 break;
 388         default:
 389                 break;
 390         }
 391
 392         if ( ! mdoc_valid_pre(mdoc, p))
 393                 return(0);
 394
 395         switch (p->type) {
 396         case (MDOC_HEAD):
 397                 assert(MDOC_BLOCK == p->parent->type);
 398                 p->parent->head = p;
 399                 break;
 400         case (MDOC_TAIL):
 401                 assert(MDOC_BLOCK == p->parent->type);
 402                 p->parent->tail = p;
 403                 break;
 404         case (MDOC_BODY):
 405                 if (p->end)
 406                         break;
 407                 assert(MDOC_BLOCK == p->parent->type);
 408                 p->parent->body = p;
 409                 break;
 410         default:
 411                 break;
 412         }
 413
 414         mdoc->last = p;
 415
 416         switch (p->type) {
 417         case (MDOC_TBL):
 418                 /* FALLTHROUGH */
 419         case (MDOC_TEXT):
 420                 if ( ! mdoc_valid_post(mdoc))
 421                         return(0);
 422                 break;
 423         default:
 424                 break;
 425         }
 426
 427         return(1);
 428 }
 429
 430
 431 static struct mdoc_node *
 432 node_alloc(struct mdoc *mdoc, int line, int pos,
 433                 enum mdoct tok, enum mdoc_type type)
 434 {
 435         struct mdoc_node *p;
 436
 437         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 438         p->sec = mdoc->lastsec;
 439         p->line = line;
 440         p->pos = pos;
 441         p->lastline = line;
 442         p->tok = tok;
 443         p->type = type;
 444
 445         /* Flag analysis. */
 446
 447         if (MDOC_SYNOPSIS & mdoc->flags)
 448                 p->flags |= MDOC_SYNPRETTY;
 449         else
 450                 p->flags &= ~MDOC_SYNPRETTY;
 451         if (MDOC_NEWLINE & mdoc->flags)
 452                 p->flags |= MDOC_LINE;
 453         mdoc->flags &= ~MDOC_NEWLINE;
 454
 455         return(p);
 456 }
 457
 458
 459 int
 460 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 461 {
 462         struct mdoc_node *p;
 463
 464         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 465         if ( ! node_append(mdoc, p))
 466                 return(0);
 467         mdoc->next = MDOC_NEXT_CHILD;
 468         return(1);
 469 }
 470
 471
 472 int
 473 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 474 {
 475         struct mdoc_node *p;
 476
 477         assert(mdoc->first);
 478         assert(mdoc->last);
 479
 480         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 481         if ( ! node_append(mdoc, p))
 482                 return(0);
 483         mdoc->next = MDOC_NEXT_CHILD;
 484         return(1);
 485 }
 486
 487
 488 int
 489 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 490 {
 491         struct mdoc_node *p;
 492
 493         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 494         if ( ! node_append(mdoc, p))
 495                 return(0);
 496         mdoc->next = MDOC_NEXT_CHILD;
 497         return(1);
 498 }
 499
 500
 501 int
 502 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 503                 struct mdoc_node *body, enum mdoc_endbody end)
 504 {
 505         struct mdoc_node *p;
 506
 507         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 508         p->pending = body;
 509         p->norm = body->norm;
 510         p->end = end;
 511         if ( ! node_append(mdoc, p))
 512                 return(0);
 513         mdoc->next = MDOC_NEXT_SIBLING;
 514         return(1);
 515 }
 516
 517
 518 int
 519 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 520                 enum mdoct tok, struct mdoc_arg *args)
 521 {
 522         struct mdoc_node *p;
 523
 524         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 525         p->args = args;
 526         if (p->args)
 527                 (args->refcnt)++;
 528
 529         switch (tok) {
 530         case (MDOC_Bd):
 531                 /* FALLTHROUGH */
 532         case (MDOC_Bf):
 533                 /* FALLTHROUGH */
 534         case (MDOC_Bl):
 535                 /* FALLTHROUGH */
 536         case (MDOC_Rs):
 537                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 538                 break;
 539         default:
 540                 break;
 541         }
 542
 543         if ( ! node_append(mdoc, p))
 544                 return(0);
 545         mdoc->next = MDOC_NEXT_CHILD;
 546         return(1);
 547 }
 548
 549
 550 int
 551 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 552                 enum mdoct tok, struct mdoc_arg *args)
 553 {
 554         struct mdoc_node *p;
 555
 556         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 557         p->args = args;
 558         if (p->args)
 559                 (args->refcnt)++;
 560
 561         switch (tok) {
 562         case (MDOC_An):
 563                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 564                 break;
 565         default:
 566                 break;
 567         }
 568
 569         if ( ! node_append(mdoc, p))
 570                 return(0);
 571         mdoc->next = MDOC_NEXT_CHILD;
 572         return(1);
 573 }
 574
 575 int
 576 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 577 {
 578         struct mdoc_node *n;
 579
 580         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 581         n->string = roff_strdup(mdoc->roff, p);
 582
 583         if ( ! node_append(mdoc, n))
 584                 return(0);
 585
 586         mdoc->next = MDOC_NEXT_SIBLING;
 587         return(1);
 588 }
 589
 590 void
 591 mdoc_word_append(struct mdoc *mdoc, const char *p)
 592 {
 593         struct mdoc_node        *n;
 594         char                    *addstr, *newstr;
 595
 596         n = mdoc->last;
 597         addstr = roff_strdup(mdoc->roff, p);
 598         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 599         free(addstr);
 600         free(n->string);
 601         n->string = newstr;
 602         mdoc->next = MDOC_NEXT_SIBLING;
 603 }
 604
 605 static void
 606 mdoc_node_free(struct mdoc_node *p)
 607 {
 608
 609         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 610                 free(p->norm);
 611         if (p->string)
 612                 free(p->string);
 613         if (p->args)
 614                 mdoc_argv_free(p->args);
 615         free(p);
 616 }
 617
 618
 619 static void
 620 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 621 {
 622
 623         /* Adjust siblings. */
 624
 625         if (n->prev)
 626                 n->prev->next = n->next;
 627         if (n->next)
 628                 n->next->prev = n->prev;
 629
 630         /* Adjust parent. */
 631
 632         if (n->parent) {
 633                 n->parent->nchild--;
 634                 if (n->parent->child == n)
 635                         n->parent->child = n->prev ? n->prev : n->next;
 636                 if (n->parent->last == n)
 637                         n->parent->last = n->prev ? n->prev : NULL;
 638         }
 639
 640         /* Adjust parse point, if applicable. */
 641
 642         if (mdoc && mdoc->last == n) {
 643                 if (n->prev) {
 644                         mdoc->last = n->prev;
 645                         mdoc->next = MDOC_NEXT_SIBLING;
 646                 } else {
 647                         mdoc->last = n->parent;
 648                         mdoc->next = MDOC_NEXT_CHILD;
 649                 }
 650         }
 651
 652         if (mdoc && mdoc->first == n)
 653                 mdoc->first = NULL;
 654 }
 655
 656
 657 void
 658 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 659 {
 660
 661         while (p->child) {
 662                 assert(p->nchild);
 663                 mdoc_node_delete(mdoc, p->child);
 664         }
 665         assert(0 == p->nchild);
 666
 667         mdoc_node_unlink(mdoc, p);
 668         mdoc_node_free(p);
 669 }
 670
 671 int
 672 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 673 {
 674
 675         mdoc_node_unlink(mdoc, p);
 676         return(node_append(mdoc, p));
 677 }
 678
 679 #if 0
 680 /*
 681  * Pre-treat a text line.
 682  * Text lines can consist of equations, which must be handled apart from
 683  * the regular text.
 684  * Thus, use this function to step through a line checking if it has any
 685  * equations embedded in it.
 686  * This must handle multiple equations AND equations that do not end at
 687  * the end-of-line, i.e., will re-enter in the next roff parse.
 688  */
 689 static int
 690 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
 691 {
 692         char            *start, *end;
 693         char             delim;
 694
 695         while ('\0' != buf[offs]) {
 696                 /* Mark starting position if eqn is set. */
 697                 start = NULL;
 698                 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
 699                         if (NULL != (start = strchr(buf + offs, delim)))
 700                                 *start++ = '\0';
 701
 702                 /* Parse text as normal. */
 703                 if ( ! mdoc_ptext(mdoc, line, buf, offs))
 704                         return(0);
 705
 706                 /* Continue only if an equation exists. */
 707                 if (NULL == start)
 708                         break;
 709
 710                 /* Read past the end of the equation. */
 711                 offs += start - (buf + offs);
 712                 assert(start == &buf[offs]);
 713                 if (NULL != (end = strchr(buf + offs, delim))) {
 714                         *end++ = '\0';
 715                         while (' ' == *end)
 716                                 end++;
 717                 }
 718
 719                 /* Parse the equation itself. */
 720                 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
 721
 722                 /* Process a finished equation? */
 723                 if (roff_closeeqn(mdoc->roff))
 724                         if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
 725                                 return(0);
 726                 offs += (end - (buf + offs));
 727         }
 728
 729         return(1);
 730 }
 731 #endif
 732
 733 /*
 734  * Parse free-form text, that is, a line that does not begin with the
 735  * control character.
 736  */
 737 static int
 738 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 739 {
 740         char             *c, *ws, *end;
 741         struct mdoc_node *n;
 742
 743         /* No text before an initial macro. */
 744
 745         if (SEC_NONE == mdoc->lastnamed) {
 746                 mdoc_pmsg(mdoc, line, offs, MANDOCERR_NOTEXT);
 747                 return(1);
 748         }
 749
 750         assert(mdoc->last);
 751         n = mdoc->last;
 752
 753         /*
 754          * Divert directly to list processing if we're encountering a
 755          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 756          * (a MDOC_BODY means it's already open, in which case we should
 757          * process within its context in the normal way).
 758          */
 759
 760         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 761                         LIST_column == n->norm->Bl.type) {
 762                 /* `Bl' is open without any children. */
 763                 mdoc->flags |= MDOC_FREECOL;
 764                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 765         }
 766
 767         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 768                         NULL != n->parent &&
 769                         MDOC_Bl == n->parent->tok &&
 770                         LIST_column == n->parent->norm->Bl.type) {
 771                 /* `Bl' has block-level `It' children. */
 772                 mdoc->flags |= MDOC_FREECOL;
 773                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 774         }
 775
 776         /*
 777          * Search for the beginning of unescaped trailing whitespace (ws)
 778          * and for the first character not to be output (end).
 779          */
 780
 781         /* FIXME: replace with strcspn(). */
 782         ws = NULL;
 783         for (c = end = buf + offs; *c; c++) {
 784                 switch (*c) {
 785                 case ' ':
 786                         if (NULL == ws)
 787                                 ws = c;
 788                         continue;
 789                 case '\t':
 790                         /*
 791                          * Always warn about trailing tabs,
 792                          * even outside literal context,
 793                          * where they should be put on the next line.
 794                          */
 795                         if (NULL == ws)
 796                                 ws = c;
 797                         /*
 798                          * Strip trailing tabs in literal context only;
 799                          * outside, they affect the next line.
 800                          */
 801                         if (MDOC_LITERAL & mdoc->flags)
 802                                 continue;
 803                         break;
 804                 case '\\':
 805                         /* Skip the escaped character, too, if any. */
 806                         if (c[1])
 807                                 c++;
 808                         /* FALLTHROUGH */
 809                 default:
 810                         ws = NULL;
 811                         break;
 812                 }
 813                 end = c + 1;
 814         }
 815         *end = '\0';
 816
 817         if (ws)
 818                 mdoc_pmsg(mdoc, line, (int)(ws-buf), MANDOCERR_EOLNSPACE);
 819
 820         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 821                 mdoc_pmsg(mdoc, line, (int)(c-buf), MANDOCERR_NOBLANKLN);
 822
 823                 /*
 824                  * Insert a `sp' in the case of a blank line.  Technically,
 825                  * blank lines aren't allowed, but enough manuals assume this
 826                  * behaviour that we want to work around it.
 827                  */
 828                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 829                         return(0);
 830
 831                 mdoc->next = MDOC_NEXT_SIBLING;
 832
 833                 return(mdoc_valid_post(mdoc));
 834         }
 835
 836         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 837                 return(0);
 838
 839         if (MDOC_LITERAL & mdoc->flags)
 840                 return(1);
 841
 842         /*
 843          * End-of-sentence check.  If the last character is an unescaped
 844          * EOS character, then flag the node as being the end of a
 845          * sentence.  The front-end will know how to interpret this.
 846          */
 847
 848         assert(buf < end);
 849
 850         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 851                 mdoc->last->flags |= MDOC_EOS;
 852
 853         return(1);
 854 }
 855
 856
 857 /*
 858  * Parse a macro line, that is, a line beginning with the control
 859  * character.
 860  */
 861 static int
 862 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 863 {
 864         enum mdoct        tok;
 865         int               i, sv;
 866         char              mac[5];
 867         struct mdoc_node *n;
 868
 869         /* Empty post-control lines are ignored. */
 870
 871         if ('"' == buf[offs]) {
 872                 mdoc_pmsg(mdoc, ln, offs, MANDOCERR_BADCOMMENT);
 873                 return(1);
 874         } else if ('\0' == buf[offs])
 875                 return(1);
 876
 877         sv = offs;
 878
 879         /*
 880          * Copy the first word into a nil-terminated buffer.
 881          * Stop copying when a tab, space, or eoln is encountered.
 882          */
 883
 884         i = 0;
 885         while (i < 4 && '\0' != buf[offs] &&
 886                         ' ' != buf[offs] && '\t' != buf[offs])
 887                 mac[i++] = buf[offs++];
 888
 889         mac[i] = '\0';
 890
 891         tok = (i > 1 || i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 892
 893         if (MDOC_MAX == tok) {
 894                 mandoc_vmsg(MANDOCERR_MACRO, mdoc->parse,
 895                                 ln, sv, "%s", buf + sv - 1);
 896                 return(1);
 897         }
 898
 899         /* Disregard the first trailing tab, if applicable. */
 900
 901         if ('\t' == buf[offs])
 902                 offs++;
 903
 904         /* Jump to the next non-whitespace word. */
 905
 906         while (buf[offs] && ' ' == buf[offs])
 907                 offs++;
 908
 909         /*
 910          * Trailing whitespace.  Note that tabs are allowed to be passed
 911          * into the parser as "text", so we only warn about spaces here.
 912          */
 913
 914         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 915                 mdoc_pmsg(mdoc, ln, offs - 1, MANDOCERR_EOLNSPACE);
 916
 917         /*
 918          * If an initial macro or a list invocation, divert directly
 919          * into macro processing.
 920          */
 921
 922         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 923                 if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 924                         goto err;
 925                 return(1);
 926         }
 927
 928         n = mdoc->last;
 929         assert(mdoc->last);
 930
 931         /*
 932          * If the first macro of a `Bl -column', open an `It' block
 933          * context around the parsed macro.
 934          */
 935
 936         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 937                         LIST_column == n->norm->Bl.type) {
 938                 mdoc->flags |= MDOC_FREECOL;
 939                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 940                         goto err;
 941                 return(1);
 942         }
 943
 944         /*
 945          * If we're following a block-level `It' within a `Bl -column'
 946          * context (perhaps opened in the above block or in ptext()),
 947          * then open an `It' block context around the parsed macro.
 948          */
 949
 950         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 951                         NULL != n->parent &&
 952                         MDOC_Bl == n->parent->tok &&
 953                         LIST_column == n->parent->norm->Bl.type) {
 954                 mdoc->flags |= MDOC_FREECOL;
 955                 if ( ! mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf))
 956                         goto err;
 957                 return(1);
 958         }
 959
 960         /* Normal processing of a macro. */
 961
 962         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 963                 goto err;
 964
 965         /* In quick mode (for mandocdb), abort after the NAME section. */
 966
 967         if (mdoc->quick && MDOC_Sh == tok &&
 968             SEC_NAME != mdoc->last->sec)
 969                 return(2);
 970
 971         return(1);
 972
 973 err:    /* Error out. */
 974
 975         mdoc->flags |= MDOC_HALT;
 976         return(0);
 977 }
 978
 979 enum mdelim
 980 mdoc_isdelim(const char *p)
 981 {
 982
 983         if ('\0' == p[0])
 984                 return(DELIM_NONE);
 985
 986         if ('\0' == p[1])
 987                 switch (p[0]) {
 988                 case('('):
 989                         /* FALLTHROUGH */
 990                 case('['):
 991                         return(DELIM_OPEN);
 992                 case('|'):
 993                         return(DELIM_MIDDLE);
 994                 case('.'):
 995                         /* FALLTHROUGH */
 996                 case(','):
 997                         /* FALLTHROUGH */
 998                 case(';'):
 999                         /* FALLTHROUGH */
1000                 case(':'):
1001                         /* FALLTHROUGH */
1002                 case('?'):
1003                         /* FALLTHROUGH */
1004                 case('!'):
1005                         /* FALLTHROUGH */
1006                 case(')'):
1007                         /* FALLTHROUGH */
1008                 case(']'):
1009                         return(DELIM_CLOSE);
1010                 default:
1011                         return(DELIM_NONE);
1012                 }
1013
1014         if ('\\' != p[0])
1015                 return(DELIM_NONE);
1016
1017         if (0 == strcmp(p + 1, "."))
1018                 return(DELIM_CLOSE);
1019         if (0 == strcmp(p + 1, "fR|\\fP"))
1020                 return(DELIM_MIDDLE);
1021
1022         return(DELIM_NONE);
1023 }
1024
1025 void
1026 mdoc_deroff(char **dest, const struct mdoc_node *n)
1027 {
1028         char    *cp;
1029         size_t   sz;
1030
1031         if (MDOC_TEXT != n->type) {
1032                 for (n = n->child; n; n = n->next)
1033                         mdoc_deroff(dest, n);
1034                 return;
1035         }
1036
1037         /* Skip leading whitespace. */
1038
1039         for (cp = n->string; '\0' != *cp; cp++)
1040                 if (0 == isspace((unsigned char)*cp))
1041                         break;
1042
1043         /* Skip trailing whitespace. */
1044
1045         for (sz = strlen(cp); sz; sz--)
1046                 if (0 == isspace((unsigned char)cp[sz-1]))
1047                         break;
1048
1049         /* Skip empty strings. */
1050
1051         if (0 == sz)
1052                 return;
1053
1054         if (NULL == *dest) {
1055                 *dest = mandoc_strndup(cp, sz);
1056                 return;
1057         }
1058
1059         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
1060         free(*dest);
1061         *dest = cp;
1062 }