mdoc.c

   1 /*      $Id: mdoc.c,v 1.222 2014/08/01 17:27:44 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 + 1] = {
  39         "Ap",           "Dd",           "Dt",           "Os",
  40         "Sh",           "Ss",           "Pp",           "D1",
  41         "Dl",           "Bd",           "Ed",           "Bl",
  42         "El",           "It",           "Ad",           "An",
  43         "Ar",           "Cd",           "Cm",           "Dv",
  44         "Er",           "Ev",           "Ex",           "Fa",
  45         "Fd",           "Fl",           "Fn",           "Ft",
  46         "Ic",           "In",           "Li",           "Nd",
  47         "Nm",           "Op",           "Ot",           "Pa",
  48         "Rv",           "St",           "Va",           "Vt",
  49         "Xr",           "%A",           "%B",           "%D",
  50         "%I",           "%J",           "%N",           "%O",
  51         "%P",           "%R",           "%T",           "%V",
  52         "Ac",           "Ao",           "Aq",           "At",
  53         "Bc",           "Bf",           "Bo",           "Bq",
  54         "Bsx",          "Bx",           "Db",           "Dc",
  55         "Do",           "Dq",           "Ec",           "Ef",
  56         "Em",           "Eo",           "Fx",           "Ms",
  57         "No",           "Ns",           "Nx",           "Ox",
  58         "Pc",           "Pf",           "Po",           "Pq",
  59         "Qc",           "Ql",           "Qo",           "Qq",
  60         "Re",           "Rs",           "Sc",           "So",
  61         "Sq",           "Sm",           "Sx",           "Sy",
  62         "Tn",           "Ux",           "Xc",           "Xo",
  63         "Fo",           "Fc",           "Oo",           "Oc",
  64         "Bk",           "Ek",           "Bt",           "Hf",
  65         "Fr",           "Ud",           "Lb",           "Lp",
  66         "Lk",           "Mt",           "Brq",          "Bro",
  67         "Brc",          "%C",           "Es",           "En",
  68         "Dx",           "%Q",           "br",           "sp",
  69         "%U",           "Ta",           "ll",           "text",
  70         };
  71
  72 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  73         "split",                "nosplit",              "ragged",
  74         "unfilled",             "literal",              "file",
  75         "offset",               "bullet",               "dash",
  76         "hyphen",               "item",                 "enum",
  77         "tag",                  "diag",                 "hang",
  78         "ohang",                "inset",                "column",
  79         "width",                "compact",              "std",
  80         "filled",               "words",                "emphasis",
  81         "symbolic",             "nested",               "centered"
  82         };
  83
  84 const   char * const *mdoc_macronames = __mdoc_macronames;
  85 const   char * const *mdoc_argnames = __mdoc_argnames;
  86
  87 static  void              mdoc_node_free(struct mdoc_node *);
  88 static  void              mdoc_node_unlink(struct mdoc *,
  89                                 struct mdoc_node *);
  90 static  void              mdoc_free1(struct mdoc *);
  91 static  void              mdoc_alloc1(struct mdoc *);
  92 static  struct mdoc_node *node_alloc(struct mdoc *, int, int,
  93                                 enum mdoct, enum mdoc_type);
  94 static  int               node_append(struct mdoc *,
  95                                 struct mdoc_node *);
  96 #if 0
  97 static  int               mdoc_preptext(struct mdoc *, int, char *, int);
  98 #endif
  99 static  int               mdoc_ptext(struct mdoc *, int, char *, int);
 100 static  int               mdoc_pmacro(struct mdoc *, int, char *, int);
 101
 102
 103 const struct mdoc_node *
 104 mdoc_node(const struct mdoc *mdoc)
 105 {
 106
 107         return(mdoc->first);
 108 }
 109
 110 const struct mdoc_meta *
 111 mdoc_meta(const struct mdoc *mdoc)
 112 {
 113
 114         return(&mdoc->meta);
 115 }
 116
 117 /*
 118  * Frees volatile resources (parse tree, meta-data, fields).
 119  */
 120 static void
 121 mdoc_free1(struct mdoc *mdoc)
 122 {
 123
 124         if (mdoc->first)
 125                 mdoc_node_delete(mdoc, mdoc->first);
 126         if (mdoc->meta.title)
 127                 free(mdoc->meta.title);
 128         if (mdoc->meta.os)
 129                 free(mdoc->meta.os);
 130         if (mdoc->meta.name)
 131                 free(mdoc->meta.name);
 132         if (mdoc->meta.arch)
 133                 free(mdoc->meta.arch);
 134         if (mdoc->meta.vol)
 135                 free(mdoc->meta.vol);
 136         if (mdoc->meta.msec)
 137                 free(mdoc->meta.msec);
 138         if (mdoc->meta.date)
 139                 free(mdoc->meta.date);
 140 }
 141
 142 /*
 143  * Allocate all volatile resources (parse tree, meta-data, fields).
 144  */
 145 static void
 146 mdoc_alloc1(struct mdoc *mdoc)
 147 {
 148
 149         memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
 150         mdoc->flags = 0;
 151         mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 152         mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
 153         mdoc->first = mdoc->last;
 154         mdoc->last->type = MDOC_ROOT;
 155         mdoc->last->tok = MDOC_MAX;
 156         mdoc->next = MDOC_NEXT_CHILD;
 157 }
 158
 159 /*
 160  * Free up volatile resources (see mdoc_free1()) then re-initialises the
 161  * data with mdoc_alloc1().  After invocation, parse data has been reset
 162  * and the parser is ready for re-invocation on a new tree; however,
 163  * cross-parse non-volatile data is kept intact.
 164  */
 165 void
 166 mdoc_reset(struct mdoc *mdoc)
 167 {
 168
 169         mdoc_free1(mdoc);
 170         mdoc_alloc1(mdoc);
 171 }
 172
 173 /*
 174  * Completely free up all volatile and non-volatile parse resources.
 175  * After invocation, the pointer is no longer usable.
 176  */
 177 void
 178 mdoc_free(struct mdoc *mdoc)
 179 {
 180
 181         mdoc_free1(mdoc);
 182         free(mdoc);
 183 }
 184
 185 /*
 186  * Allocate volatile and non-volatile parse resources.
 187  */
 188 struct mdoc *
 189 mdoc_alloc(struct roff *roff, struct mparse *parse,
 190         const char *defos, int quick)
 191 {
 192         struct mdoc     *p;
 193
 194         p = mandoc_calloc(1, sizeof(struct mdoc));
 195
 196         p->parse = parse;
 197         p->defos = defos;
 198         p->quick = quick;
 199         p->roff = roff;
 200
 201         mdoc_hash_init();
 202         mdoc_alloc1(p);
 203         return(p);
 204 }
 205
 206 int
 207 mdoc_endparse(struct mdoc *mdoc)
 208 {
 209
 210         return(mdoc_macroend(mdoc));
 211 }
 212
 213 int
 214 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 215 {
 216         struct mdoc_node *n;
 217
 218         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 219         n->eqn = ep;
 220
 221         if ( ! node_append(mdoc, n))
 222                 return(0);
 223
 224         mdoc->next = MDOC_NEXT_SIBLING;
 225         return(1);
 226 }
 227
 228 int
 229 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 230 {
 231         struct mdoc_node *n;
 232
 233         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 234         n->span = sp;
 235
 236         if ( ! node_append(mdoc, n))
 237                 return(0);
 238
 239         mdoc->next = MDOC_NEXT_SIBLING;
 240         return(1);
 241 }
 242
 243 /*
 244  * Main parse routine.  Parses a single line -- really just hands off to
 245  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 246  */
 247 int
 248 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 249 {
 250
 251         mdoc->flags |= MDOC_NEWLINE;
 252
 253         /*
 254          * Let the roff nS register switch SYNOPSIS mode early,
 255          * such that the parser knows at all times
 256          * whether this mode is on or off.
 257          * Note that this mode is also switched by the Sh macro.
 258          */
 259         if (roff_getreg(mdoc->roff, "nS"))
 260                 mdoc->flags |= MDOC_SYNOPSIS;
 261         else
 262                 mdoc->flags &= ~MDOC_SYNOPSIS;
 263
 264         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 265             mdoc_pmacro(mdoc, ln, buf, offs) :
 266             mdoc_ptext(mdoc, ln, buf, offs));
 267 }
 268
 269 int
 270 mdoc_macro(MACRO_PROT_ARGS)
 271 {
 272         assert(tok < MDOC_MAX);
 273
 274         /* If we're in the body, deny prologue calls. */
 275
 276         if (MDOC_PROLOGUE & mdoc_macros[tok].flags &&
 277             MDOC_PBODY & mdoc->flags) {
 278                 mandoc_vmsg(MANDOCERR_PROLOG_ONLY, mdoc->parse,
 279                     line, ppos, "%s", mdoc_macronames[tok]);
 280                 return(1);
 281         }
 282
 283         /* If we're in the prologue, deny "body" macros.  */
 284
 285         if ( ! (MDOC_PROLOGUE & mdoc_macros[tok].flags) &&
 286              ! (MDOC_PBODY & mdoc->flags)) {
 287                 mandoc_vmsg(MANDOCERR_PROLOG_BAD, mdoc->parse,
 288                     line, ppos, "%s", mdoc_macronames[tok]);
 289                 if (NULL == mdoc->meta.msec)
 290                         mdoc->meta.msec = mandoc_strdup("1");
 291                 if (NULL == mdoc->meta.title)
 292                         mdoc->meta.title = mandoc_strdup("UNKNOWN");
 293                 if (NULL == mdoc->meta.vol)
 294                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 295                 if (NULL == mdoc->meta.os)
 296                         mdoc->meta.os = mandoc_strdup("LOCAL");
 297                 if (NULL == mdoc->meta.date)
 298                         mdoc->meta.date = mandoc_normdate
 299                                 (mdoc->parse, NULL, line, ppos);
 300                 mdoc->flags |= MDOC_PBODY;
 301         }
 302
 303         return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
 304 }
 305
 306
 307 static int
 308 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 309 {
 310
 311         assert(mdoc->last);
 312         assert(mdoc->first);
 313         assert(MDOC_ROOT != p->type);
 314
 315         switch (mdoc->next) {
 316         case MDOC_NEXT_SIBLING:
 317                 mdoc->last->next = p;
 318                 p->prev = mdoc->last;
 319                 p->parent = mdoc->last->parent;
 320                 break;
 321         case MDOC_NEXT_CHILD:
 322                 mdoc->last->child = p;
 323                 p->parent = mdoc->last;
 324                 break;
 325         default:
 326                 abort();
 327                 /* NOTREACHED */
 328         }
 329
 330         p->parent->nchild++;
 331
 332         /*
 333          * Copy over the normalised-data pointer of our parent.  Not
 334          * everybody has one, but copying a null pointer is fine.
 335          */
 336
 337         switch (p->type) {
 338         case MDOC_BODY:
 339                 if (ENDBODY_NOT != p->end)
 340                         break;
 341                 /* FALLTHROUGH */
 342         case MDOC_TAIL:
 343                 /* FALLTHROUGH */
 344         case MDOC_HEAD:
 345                 p->norm = p->parent->norm;
 346                 break;
 347         default:
 348                 break;
 349         }
 350
 351         if ( ! mdoc_valid_pre(mdoc, p))
 352                 return(0);
 353
 354         switch (p->type) {
 355         case MDOC_HEAD:
 356                 assert(MDOC_BLOCK == p->parent->type);
 357                 p->parent->head = p;
 358                 break;
 359         case MDOC_TAIL:
 360                 assert(MDOC_BLOCK == p->parent->type);
 361                 p->parent->tail = p;
 362                 break;
 363         case MDOC_BODY:
 364                 if (p->end)
 365                         break;
 366                 assert(MDOC_BLOCK == p->parent->type);
 367                 p->parent->body = p;
 368                 break;
 369         default:
 370                 break;
 371         }
 372
 373         mdoc->last = p;
 374
 375         switch (p->type) {
 376         case MDOC_TBL:
 377                 /* FALLTHROUGH */
 378         case MDOC_TEXT:
 379                 if ( ! mdoc_valid_post(mdoc))
 380                         return(0);
 381                 break;
 382         default:
 383                 break;
 384         }
 385
 386         return(1);
 387 }
 388
 389 static struct mdoc_node *
 390 node_alloc(struct mdoc *mdoc, int line, int pos,
 391                 enum mdoct tok, enum mdoc_type type)
 392 {
 393         struct mdoc_node *p;
 394
 395         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 396         p->sec = mdoc->lastsec;
 397         p->line = line;
 398         p->pos = pos;
 399         p->lastline = line;
 400         p->tok = tok;
 401         p->type = type;
 402
 403         /* Flag analysis. */
 404
 405         if (MDOC_SYNOPSIS & mdoc->flags)
 406                 p->flags |= MDOC_SYNPRETTY;
 407         else
 408                 p->flags &= ~MDOC_SYNPRETTY;
 409         if (MDOC_NEWLINE & mdoc->flags)
 410                 p->flags |= MDOC_LINE;
 411         mdoc->flags &= ~MDOC_NEWLINE;
 412
 413         return(p);
 414 }
 415
 416 int
 417 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 418 {
 419         struct mdoc_node *p;
 420
 421         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 422         if ( ! node_append(mdoc, p))
 423                 return(0);
 424         mdoc->next = MDOC_NEXT_CHILD;
 425         return(1);
 426 }
 427
 428 int
 429 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 430 {
 431         struct mdoc_node *p;
 432
 433         assert(mdoc->first);
 434         assert(mdoc->last);
 435
 436         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 437         if ( ! node_append(mdoc, p))
 438                 return(0);
 439         mdoc->next = MDOC_NEXT_CHILD;
 440         return(1);
 441 }
 442
 443 int
 444 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 445 {
 446         struct mdoc_node *p;
 447
 448         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 449         if ( ! node_append(mdoc, p))
 450                 return(0);
 451         mdoc->next = MDOC_NEXT_CHILD;
 452         return(1);
 453 }
 454
 455 int
 456 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 457                 struct mdoc_node *body, enum mdoc_endbody end)
 458 {
 459         struct mdoc_node *p;
 460
 461         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 462         p->pending = body;
 463         p->norm = body->norm;
 464         p->end = end;
 465         if ( ! node_append(mdoc, p))
 466                 return(0);
 467         mdoc->next = MDOC_NEXT_SIBLING;
 468         return(1);
 469 }
 470
 471 int
 472 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 473                 enum mdoct tok, struct mdoc_arg *args)
 474 {
 475         struct mdoc_node *p;
 476
 477         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 478         p->args = args;
 479         if (p->args)
 480                 (args->refcnt)++;
 481
 482         switch (tok) {
 483         case MDOC_Bd:
 484                 /* FALLTHROUGH */
 485         case MDOC_Bf:
 486                 /* FALLTHROUGH */
 487         case MDOC_Bl:
 488                 /* FALLTHROUGH */
 489         case MDOC_En:
 490                 /* FALLTHROUGH */
 491         case MDOC_Rs:
 492                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 493                 break;
 494         default:
 495                 break;
 496         }
 497
 498         if ( ! node_append(mdoc, p))
 499                 return(0);
 500         mdoc->next = MDOC_NEXT_CHILD;
 501         return(1);
 502 }
 503
 504 int
 505 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 506                 enum mdoct tok, struct mdoc_arg *args)
 507 {
 508         struct mdoc_node *p;
 509
 510         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 511         p->args = args;
 512         if (p->args)
 513                 (args->refcnt)++;
 514
 515         switch (tok) {
 516         case MDOC_An:
 517                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 518                 break;
 519         default:
 520                 break;
 521         }
 522
 523         if ( ! node_append(mdoc, p))
 524                 return(0);
 525         mdoc->next = MDOC_NEXT_CHILD;
 526         return(1);
 527 }
 528
 529 int
 530 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 531 {
 532         struct mdoc_node *n;
 533
 534         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 535         n->string = roff_strdup(mdoc->roff, p);
 536
 537         if ( ! node_append(mdoc, n))
 538                 return(0);
 539
 540         mdoc->next = MDOC_NEXT_SIBLING;
 541         return(1);
 542 }
 543
 544 void
 545 mdoc_word_append(struct mdoc *mdoc, const char *p)
 546 {
 547         struct mdoc_node        *n;
 548         char                    *addstr, *newstr;
 549
 550         n = mdoc->last;
 551         addstr = roff_strdup(mdoc->roff, p);
 552         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 553         free(addstr);
 554         free(n->string);
 555         n->string = newstr;
 556         mdoc->next = MDOC_NEXT_SIBLING;
 557 }
 558
 559 static void
 560 mdoc_node_free(struct mdoc_node *p)
 561 {
 562
 563         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 564                 free(p->norm);
 565         if (p->string)
 566                 free(p->string);
 567         if (p->args)
 568                 mdoc_argv_free(p->args);
 569         free(p);
 570 }
 571
 572 static void
 573 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 574 {
 575
 576         /* Adjust siblings. */
 577
 578         if (n->prev)
 579                 n->prev->next = n->next;
 580         if (n->next)
 581                 n->next->prev = n->prev;
 582
 583         /* Adjust parent. */
 584
 585         if (n->parent) {
 586                 n->parent->nchild--;
 587                 if (n->parent->child == n)
 588                         n->parent->child = n->prev ? n->prev : n->next;
 589                 if (n->parent->last == n)
 590                         n->parent->last = n->prev ? n->prev : NULL;
 591         }
 592
 593         /* Adjust parse point, if applicable. */
 594
 595         if (mdoc && mdoc->last == n) {
 596                 if (n->prev) {
 597                         mdoc->last = n->prev;
 598                         mdoc->next = MDOC_NEXT_SIBLING;
 599                 } else {
 600                         mdoc->last = n->parent;
 601                         mdoc->next = MDOC_NEXT_CHILD;
 602                 }
 603         }
 604
 605         if (mdoc && mdoc->first == n)
 606                 mdoc->first = NULL;
 607 }
 608
 609 void
 610 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 611 {
 612
 613         while (p->child) {
 614                 assert(p->nchild);
 615                 mdoc_node_delete(mdoc, p->child);
 616         }
 617         assert(0 == p->nchild);
 618
 619         mdoc_node_unlink(mdoc, p);
 620         mdoc_node_free(p);
 621 }
 622
 623 int
 624 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 625 {
 626
 627         mdoc_node_unlink(mdoc, p);
 628         return(node_append(mdoc, p));
 629 }
 630
 631 #if 0
 632 /*
 633  * Pre-treat a text line.
 634  * Text lines can consist of equations, which must be handled apart from
 635  * the regular text.
 636  * Thus, use this function to step through a line checking if it has any
 637  * equations embedded in it.
 638  * This must handle multiple equations AND equations that do not end at
 639  * the end-of-line, i.e., will re-enter in the next roff parse.
 640  */
 641 static int
 642 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
 643 {
 644         char            *start, *end;
 645         char             delim;
 646
 647         while ('\0' != buf[offs]) {
 648                 /* Mark starting position if eqn is set. */
 649                 start = NULL;
 650                 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
 651                         if (NULL != (start = strchr(buf + offs, delim)))
 652                                 *start++ = '\0';
 653
 654                 /* Parse text as normal. */
 655                 if ( ! mdoc_ptext(mdoc, line, buf, offs))
 656                         return(0);
 657
 658                 /* Continue only if an equation exists. */
 659                 if (NULL == start)
 660                         break;
 661
 662                 /* Read past the end of the equation. */
 663                 offs += start - (buf + offs);
 664                 assert(start == &buf[offs]);
 665                 if (NULL != (end = strchr(buf + offs, delim))) {
 666                         *end++ = '\0';
 667                         while (' ' == *end)
 668                                 end++;
 669                 }
 670
 671                 /* Parse the equation itself. */
 672                 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
 673
 674                 /* Process a finished equation? */
 675                 if (roff_closeeqn(mdoc->roff))
 676                         if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
 677                                 return(0);
 678                 offs += (end - (buf + offs));
 679         }
 680
 681         return(1);
 682 }
 683 #endif
 684
 685 /*
 686  * Parse free-form text, that is, a line that does not begin with the
 687  * control character.
 688  */
 689 static int
 690 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 691 {
 692         char             *c, *ws, *end;
 693         struct mdoc_node *n;
 694
 695         assert(mdoc->last);
 696         n = mdoc->last;
 697
 698         /*
 699          * Divert directly to list processing if we're encountering a
 700          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 701          * (a MDOC_BODY means it's already open, in which case we should
 702          * process within its context in the normal way).
 703          */
 704
 705         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 706             LIST_column == n->norm->Bl.type) {
 707                 /* `Bl' is open without any children. */
 708                 mdoc->flags |= MDOC_FREECOL;
 709                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 710         }
 711
 712         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 713             NULL != n->parent &&
 714             MDOC_Bl == n->parent->tok &&
 715             LIST_column == n->parent->norm->Bl.type) {
 716                 /* `Bl' has block-level `It' children. */
 717                 mdoc->flags |= MDOC_FREECOL;
 718                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 719         }
 720
 721         /*
 722          * Search for the beginning of unescaped trailing whitespace (ws)
 723          * and for the first character not to be output (end).
 724          */
 725
 726         /* FIXME: replace with strcspn(). */
 727         ws = NULL;
 728         for (c = end = buf + offs; *c; c++) {
 729                 switch (*c) {
 730                 case ' ':
 731                         if (NULL == ws)
 732                                 ws = c;
 733                         continue;
 734                 case '\t':
 735                         /*
 736                          * Always warn about trailing tabs,
 737                          * even outside literal context,
 738                          * where they should be put on the next line.
 739                          */
 740                         if (NULL == ws)
 741                                 ws = c;
 742                         /*
 743                          * Strip trailing tabs in literal context only;
 744                          * outside, they affect the next line.
 745                          */
 746                         if (MDOC_LITERAL & mdoc->flags)
 747                                 continue;
 748                         break;
 749                 case '\\':
 750                         /* Skip the escaped character, too, if any. */
 751                         if (c[1])
 752                                 c++;
 753                         /* FALLTHROUGH */
 754                 default:
 755                         ws = NULL;
 756                         break;
 757                 }
 758                 end = c + 1;
 759         }
 760         *end = '\0';
 761
 762         if (ws)
 763                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 764                     line, (int)(ws-buf), NULL);
 765
 766         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 767                 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
 768                     line, (int)(c - buf), NULL);
 769
 770                 /*
 771                  * Insert a `sp' in the case of a blank line.  Technically,
 772                  * blank lines aren't allowed, but enough manuals assume this
 773                  * behaviour that we want to work around it.
 774                  */
 775                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 776                         return(0);
 777
 778                 mdoc->next = MDOC_NEXT_SIBLING;
 779
 780                 return(mdoc_valid_post(mdoc));
 781         }
 782
 783         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 784                 return(0);
 785
 786         if (MDOC_LITERAL & mdoc->flags)
 787                 return(1);
 788
 789         /*
 790          * End-of-sentence check.  If the last character is an unescaped
 791          * EOS character, then flag the node as being the end of a
 792          * sentence.  The front-end will know how to interpret this.
 793          */
 794
 795         assert(buf < end);
 796
 797         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 798                 mdoc->last->flags |= MDOC_EOS;
 799
 800         return(1);
 801 }
 802
 803 /*
 804  * Parse a macro line, that is, a line beginning with the control
 805  * character.
 806  */
 807 static int
 808 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 809 {
 810         enum mdoct        tok;
 811         int               i, sv;
 812         char              mac[5];
 813         struct mdoc_node *n;
 814
 815         /* Empty post-control lines are ignored. */
 816
 817         if ('"' == buf[offs]) {
 818                 mandoc_msg(MANDOCERR_COMMENT_BAD, mdoc->parse,
 819                     ln, offs, NULL);
 820                 return(1);
 821         } else if ('\0' == buf[offs])
 822                 return(1);
 823
 824         sv = offs;
 825
 826         /*
 827          * Copy the first word into a nil-terminated buffer.
 828          * Stop copying when a tab, space, or eoln is encountered.
 829          */
 830
 831         i = 0;
 832         while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
 833             '\t' != buf[offs])
 834                 mac[i++] = buf[offs++];
 835
 836         mac[i] = '\0';
 837
 838         tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 839
 840         if (MDOC_MAX == tok) {
 841                 mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
 842                     ln, sv, buf + sv - 1);
 843                 return(1);
 844         }
 845
 846         /* Disregard the first trailing tab, if applicable. */
 847
 848         if ('\t' == buf[offs])
 849                 offs++;
 850
 851         /* Jump to the next non-whitespace word. */
 852
 853         while (buf[offs] && ' ' == buf[offs])
 854                 offs++;
 855
 856         /*
 857          * Trailing whitespace.  Note that tabs are allowed to be passed
 858          * into the parser as "text", so we only warn about spaces here.
 859          */
 860
 861         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 862                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 863                     ln, offs - 1, NULL);
 864
 865         /*
 866          * If an initial macro or a list invocation, divert directly
 867          * into macro processing.
 868          */
 869
 870         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok)
 871                 return(mdoc_macro(mdoc, tok, ln, sv, &offs, buf));
 872
 873         n = mdoc->last;
 874         assert(mdoc->last);
 875
 876         /*
 877          * If the first macro of a `Bl -column', open an `It' block
 878          * context around the parsed macro.
 879          */
 880
 881         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 882             LIST_column == n->norm->Bl.type) {
 883                 mdoc->flags |= MDOC_FREECOL;
 884                 return(mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf));
 885         }
 886
 887         /*
 888          * If we're following a block-level `It' within a `Bl -column'
 889          * context (perhaps opened in the above block or in ptext()),
 890          * then open an `It' block context around the parsed macro.
 891          */
 892
 893         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 894             NULL != n->parent &&
 895             MDOC_Bl == n->parent->tok &&
 896             LIST_column == n->parent->norm->Bl.type) {
 897                 mdoc->flags |= MDOC_FREECOL;
 898                 return(mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf));
 899         }
 900
 901         /* Normal processing of a macro. */
 902
 903         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 904                 return(0);
 905
 906         /* In quick mode (for mandocdb), abort after the NAME section. */
 907
 908         if (mdoc->quick && MDOC_Sh == tok &&
 909             SEC_NAME != mdoc->last->sec)
 910                 return(2);
 911
 912         return(1);
 913 }
 914
 915 enum mdelim
 916 mdoc_isdelim(const char *p)
 917 {
 918
 919         if ('\0' == p[0])
 920                 return(DELIM_NONE);
 921
 922         if ('\0' == p[1])
 923                 switch (p[0]) {
 924                 case '(':
 925                         /* FALLTHROUGH */
 926                 case '[':
 927                         return(DELIM_OPEN);
 928                 case '|':
 929                         return(DELIM_MIDDLE);
 930                 case '.':
 931                         /* FALLTHROUGH */
 932                 case ',':
 933                         /* FALLTHROUGH */
 934                 case ';':
 935                         /* FALLTHROUGH */
 936                 case ':':
 937                         /* FALLTHROUGH */
 938                 case '?':
 939                         /* FALLTHROUGH */
 940                 case '!':
 941                         /* FALLTHROUGH */
 942                 case ')':
 943                         /* FALLTHROUGH */
 944                 case ']':
 945                         return(DELIM_CLOSE);
 946                 default:
 947                         return(DELIM_NONE);
 948                 }
 949
 950         if ('\\' != p[0])
 951                 return(DELIM_NONE);
 952
 953         if (0 == strcmp(p + 1, "."))
 954                 return(DELIM_CLOSE);
 955         if (0 == strcmp(p + 1, "fR|\\fP"))
 956                 return(DELIM_MIDDLE);
 957
 958         return(DELIM_NONE);
 959 }
 960
 961 void
 962 mdoc_deroff(char **dest, const struct mdoc_node *n)
 963 {
 964         char    *cp;
 965         size_t   sz;
 966
 967         if (MDOC_TEXT != n->type) {
 968                 for (n = n->child; n; n = n->next)
 969                         mdoc_deroff(dest, n);
 970                 return;
 971         }
 972
 973         /* Skip leading whitespace. */
 974
 975         for (cp = n->string; '\0' != *cp; cp++)
 976                 if (0 == isspace((unsigned char)*cp))
 977                         break;
 978
 979         /* Skip trailing whitespace. */
 980
 981         for (sz = strlen(cp); sz; sz--)
 982                 if (0 == isspace((unsigned char)cp[sz-1]))
 983                         break;
 984
 985         /* Skip empty strings. */
 986
 987         if (0 == sz)
 988                 return;
 989
 990         if (NULL == *dest) {
 991                 *dest = mandoc_strndup(cp, sz);
 992                 return;
 993         }
 994
 995         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 996         free(*dest);
 997         *dest = cp;
 998 }