mdoc.c

   1 /*      $Id: mdoc.c,v 1.233 2014/11/28 06:27:05 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 #include "config.h"
  19
  20 #include <sys/types.h>
  21
  22 #include <assert.h>
  23 #include <ctype.h>
  24 #include <stdarg.h>
  25 #include <stdio.h>
  26 #include <stdlib.h>
  27 #include <string.h>
  28 #include <time.h>
  29
  30 #include "mdoc.h"
  31 #include "mandoc.h"
  32 #include "mandoc_aux.h"
  33 #include "libmdoc.h"
  34 #include "libmandoc.h"
  35
  36 const   char *const __mdoc_macronames[MDOC_MAX + 1] = {
  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         "Xr",           "%A",           "%B",           "%D",
  48         "%I",           "%J",           "%N",           "%O",
  49         "%P",           "%R",           "%T",           "%V",
  50         "Ac",           "Ao",           "Aq",           "At",
  51         "Bc",           "Bf",           "Bo",           "Bq",
  52         "Bsx",          "Bx",           "Db",           "Dc",
  53         "Do",           "Dq",           "Ec",           "Ef",
  54         "Em",           "Eo",           "Fx",           "Ms",
  55         "No",           "Ns",           "Nx",           "Ox",
  56         "Pc",           "Pf",           "Po",           "Pq",
  57         "Qc",           "Ql",           "Qo",           "Qq",
  58         "Re",           "Rs",           "Sc",           "So",
  59         "Sq",           "Sm",           "Sx",           "Sy",
  60         "Tn",           "Ux",           "Xc",           "Xo",
  61         "Fo",           "Fc",           "Oo",           "Oc",
  62         "Bk",           "Ek",           "Bt",           "Hf",
  63         "Fr",           "Ud",           "Lb",           "Lp",
  64         "Lk",           "Mt",           "Brq",          "Bro",
  65         "Brc",          "%C",           "Es",           "En",
  66         "Dx",           "%Q",           "br",           "sp",
  67         "%U",           "Ta",           "ll",           "text",
  68         };
  69
  70 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  71         "split",                "nosplit",              "ragged",
  72         "unfilled",             "literal",              "file",
  73         "offset",               "bullet",               "dash",
  74         "hyphen",               "item",                 "enum",
  75         "tag",                  "diag",                 "hang",
  76         "ohang",                "inset",                "column",
  77         "width",                "compact",              "std",
  78         "filled",               "words",                "emphasis",
  79         "symbolic",             "nested",               "centered"
  80         };
  81
  82 const   char * const *mdoc_macronames = __mdoc_macronames;
  83 const   char * const *mdoc_argnames = __mdoc_argnames;
  84
  85 static  void              mdoc_node_free(struct mdoc_node *);
  86 static  void              mdoc_node_unlink(struct mdoc *,
  87                                 struct mdoc_node *);
  88 static  void              mdoc_free1(struct mdoc *);
  89 static  void              mdoc_alloc1(struct mdoc *);
  90 static  struct mdoc_node *node_alloc(struct mdoc *, int, int,
  91                                 enum mdoct, enum mdoc_type);
  92 static  void              node_append(struct mdoc *, struct mdoc_node *);
  93 static  int               mdoc_ptext(struct mdoc *, int, char *, int);
  94 static  int               mdoc_pmacro(struct mdoc *, int, char *, int);
  95
  96
  97 const struct mdoc_node *
  98 mdoc_node(const struct mdoc *mdoc)
  99 {
 100
 101         return(mdoc->first);
 102 }
 103
 104 const struct mdoc_meta *
 105 mdoc_meta(const struct mdoc *mdoc)
 106 {
 107
 108         return(&mdoc->meta);
 109 }
 110
 111 /*
 112  * Frees volatile resources (parse tree, meta-data, fields).
 113  */
 114 static void
 115 mdoc_free1(struct mdoc *mdoc)
 116 {
 117
 118         if (mdoc->first)
 119                 mdoc_node_delete(mdoc, mdoc->first);
 120         free(mdoc->meta.msec);
 121         free(mdoc->meta.vol);
 122         free(mdoc->meta.arch);
 123         free(mdoc->meta.date);
 124         free(mdoc->meta.title);
 125         free(mdoc->meta.os);
 126         free(mdoc->meta.name);
 127 }
 128
 129 /*
 130  * Allocate all volatile resources (parse tree, meta-data, fields).
 131  */
 132 static void
 133 mdoc_alloc1(struct mdoc *mdoc)
 134 {
 135
 136         memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
 137         mdoc->flags = 0;
 138         mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 139         mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
 140         mdoc->first = mdoc->last;
 141         mdoc->last->type = MDOC_ROOT;
 142         mdoc->last->tok = MDOC_MAX;
 143         mdoc->next = MDOC_NEXT_CHILD;
 144 }
 145
 146 /*
 147  * Free up volatile resources (see mdoc_free1()) then re-initialises the
 148  * data with mdoc_alloc1().  After invocation, parse data has been reset
 149  * and the parser is ready for re-invocation on a new tree; however,
 150  * cross-parse non-volatile data is kept intact.
 151  */
 152 void
 153 mdoc_reset(struct mdoc *mdoc)
 154 {
 155
 156         mdoc_free1(mdoc);
 157         mdoc_alloc1(mdoc);
 158 }
 159
 160 /*
 161  * Completely free up all volatile and non-volatile parse resources.
 162  * After invocation, the pointer is no longer usable.
 163  */
 164 void
 165 mdoc_free(struct mdoc *mdoc)
 166 {
 167
 168         mdoc_free1(mdoc);
 169         free(mdoc);
 170 }
 171
 172 /*
 173  * Allocate volatile and non-volatile parse resources.
 174  */
 175 struct mdoc *
 176 mdoc_alloc(struct roff *roff, struct mparse *parse,
 177         const char *defos, int quick)
 178 {
 179         struct mdoc     *p;
 180
 181         p = mandoc_calloc(1, sizeof(struct mdoc));
 182
 183         p->parse = parse;
 184         p->defos = defos;
 185         p->quick = quick;
 186         p->roff = roff;
 187
 188         mdoc_hash_init();
 189         mdoc_alloc1(p);
 190         return(p);
 191 }
 192
 193 int
 194 mdoc_endparse(struct mdoc *mdoc)
 195 {
 196
 197         mdoc_macroend(mdoc);
 198         return(1);
 199 }
 200
 201 void
 202 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 203 {
 204         struct mdoc_node *n;
 205
 206         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 207         n->eqn = ep;
 208         if (ep->ln > mdoc->last->line)
 209                 n->flags |= MDOC_LINE;
 210         node_append(mdoc, n);
 211         mdoc->next = MDOC_NEXT_SIBLING;
 212 }
 213
 214 void
 215 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 216 {
 217         struct mdoc_node *n;
 218
 219         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 220         n->span = sp;
 221         node_append(mdoc, n);
 222         mdoc->next = MDOC_NEXT_SIBLING;
 223 }
 224
 225 /*
 226  * Main parse routine.  Parses a single line -- really just hands off to
 227  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 228  */
 229 int
 230 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 231 {
 232
 233         if (mdoc->last->type != MDOC_EQN || ln > mdoc->last->line)
 234                 mdoc->flags |= MDOC_NEWLINE;
 235
 236         /*
 237          * Let the roff nS register switch SYNOPSIS mode early,
 238          * such that the parser knows at all times
 239          * whether this mode is on or off.
 240          * Note that this mode is also switched by the Sh macro.
 241          */
 242         if (roff_getreg(mdoc->roff, "nS"))
 243                 mdoc->flags |= MDOC_SYNOPSIS;
 244         else
 245                 mdoc->flags &= ~MDOC_SYNOPSIS;
 246
 247         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 248             mdoc_pmacro(mdoc, ln, buf, offs) :
 249             mdoc_ptext(mdoc, ln, buf, offs));
 250 }
 251
 252 void
 253 mdoc_macro(MACRO_PROT_ARGS)
 254 {
 255         assert(tok < MDOC_MAX);
 256
 257         if (mdoc->flags & MDOC_PBODY) {
 258                 if (tok == MDOC_Dt) {
 259                         mandoc_vmsg(MANDOCERR_DT_LATE,
 260                             mdoc->parse, line, ppos,
 261                             "Dt %s", buf + *pos);
 262                         return;
 263                 }
 264         } else if ( ! (mdoc_macros[tok].flags & MDOC_PROLOGUE)) {
 265                 if (mdoc->meta.title == NULL) {
 266                         mandoc_vmsg(MANDOCERR_DT_NOTITLE,
 267                             mdoc->parse, line, ppos, "%s %s",
 268                             mdoc_macronames[tok], buf + *pos);
 269                         mdoc->meta.title = mandoc_strdup("UNTITLED");
 270                 }
 271                 if (NULL == mdoc->meta.vol)
 272                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 273                 mdoc->flags |= MDOC_PBODY;
 274         }
 275         (*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
 276 }
 277
 278
 279 static void
 280 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 281 {
 282
 283         assert(mdoc->last);
 284         assert(mdoc->first);
 285         assert(MDOC_ROOT != p->type);
 286
 287         switch (mdoc->next) {
 288         case MDOC_NEXT_SIBLING:
 289                 mdoc->last->next = p;
 290                 p->prev = mdoc->last;
 291                 p->parent = mdoc->last->parent;
 292                 break;
 293         case MDOC_NEXT_CHILD:
 294                 mdoc->last->child = p;
 295                 p->parent = mdoc->last;
 296                 break;
 297         default:
 298                 abort();
 299                 /* NOTREACHED */
 300         }
 301
 302         p->parent->nchild++;
 303
 304         /*
 305          * Copy over the normalised-data pointer of our parent.  Not
 306          * everybody has one, but copying a null pointer is fine.
 307          */
 308
 309         switch (p->type) {
 310         case MDOC_BODY:
 311                 if (ENDBODY_NOT != p->end)
 312                         break;
 313                 /* FALLTHROUGH */
 314         case MDOC_TAIL:
 315                 /* FALLTHROUGH */
 316         case MDOC_HEAD:
 317                 p->norm = p->parent->norm;
 318                 break;
 319         default:
 320                 break;
 321         }
 322
 323         mdoc_valid_pre(mdoc, p);
 324
 325         switch (p->type) {
 326         case MDOC_HEAD:
 327                 assert(MDOC_BLOCK == p->parent->type);
 328                 p->parent->head = p;
 329                 break;
 330         case MDOC_TAIL:
 331                 assert(MDOC_BLOCK == p->parent->type);
 332                 p->parent->tail = p;
 333                 break;
 334         case MDOC_BODY:
 335                 if (p->end)
 336                         break;
 337                 assert(MDOC_BLOCK == p->parent->type);
 338                 p->parent->body = p;
 339                 break;
 340         default:
 341                 break;
 342         }
 343
 344         mdoc->last = p;
 345
 346         switch (p->type) {
 347         case MDOC_TBL:
 348                 /* FALLTHROUGH */
 349         case MDOC_TEXT:
 350                 mdoc_valid_post(mdoc);
 351                 break;
 352         default:
 353                 break;
 354         }
 355 }
 356
 357 static struct mdoc_node *
 358 node_alloc(struct mdoc *mdoc, int line, int pos,
 359                 enum mdoct tok, enum mdoc_type type)
 360 {
 361         struct mdoc_node *p;
 362
 363         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 364         p->sec = mdoc->lastsec;
 365         p->line = line;
 366         p->pos = pos;
 367         p->lastline = line;
 368         p->tok = tok;
 369         p->type = type;
 370
 371         /* Flag analysis. */
 372
 373         if (MDOC_SYNOPSIS & mdoc->flags)
 374                 p->flags |= MDOC_SYNPRETTY;
 375         else
 376                 p->flags &= ~MDOC_SYNPRETTY;
 377         if (MDOC_NEWLINE & mdoc->flags)
 378                 p->flags |= MDOC_LINE;
 379         mdoc->flags &= ~MDOC_NEWLINE;
 380
 381         return(p);
 382 }
 383
 384 void
 385 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 386 {
 387         struct mdoc_node *p;
 388
 389         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 390         node_append(mdoc, p);
 391         mdoc->next = MDOC_NEXT_CHILD;
 392 }
 393
 394 struct mdoc_node *
 395 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 396 {
 397         struct mdoc_node *p;
 398
 399         assert(mdoc->first);
 400         assert(mdoc->last);
 401         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 402         node_append(mdoc, p);
 403         mdoc->next = MDOC_NEXT_CHILD;
 404         return(p);
 405 }
 406
 407 struct mdoc_node *
 408 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 409 {
 410         struct mdoc_node *p;
 411
 412         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 413         node_append(mdoc, p);
 414         mdoc->next = MDOC_NEXT_CHILD;
 415         return(p);
 416 }
 417
 418 void
 419 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 420                 struct mdoc_node *body, enum mdoc_endbody end)
 421 {
 422         struct mdoc_node *p;
 423
 424         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 425         p->pending = body;
 426         p->norm = body->norm;
 427         p->end = end;
 428         node_append(mdoc, p);
 429         mdoc->next = MDOC_NEXT_SIBLING;
 430 }
 431
 432 struct mdoc_node *
 433 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 434                 enum mdoct tok, struct mdoc_arg *args)
 435 {
 436         struct mdoc_node *p;
 437
 438         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 439         p->args = args;
 440         if (p->args)
 441                 (args->refcnt)++;
 442
 443         switch (tok) {
 444         case MDOC_Bd:
 445                 /* FALLTHROUGH */
 446         case MDOC_Bf:
 447                 /* FALLTHROUGH */
 448         case MDOC_Bl:
 449                 /* FALLTHROUGH */
 450         case MDOC_En:
 451                 /* FALLTHROUGH */
 452         case MDOC_Rs:
 453                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 454                 break;
 455         default:
 456                 break;
 457         }
 458         node_append(mdoc, p);
 459         mdoc->next = MDOC_NEXT_CHILD;
 460         return(p);
 461 }
 462
 463 void
 464 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 465                 enum mdoct tok, struct mdoc_arg *args)
 466 {
 467         struct mdoc_node *p;
 468
 469         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 470         p->args = args;
 471         if (p->args)
 472                 (args->refcnt)++;
 473
 474         switch (tok) {
 475         case MDOC_An:
 476                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 477                 break;
 478         default:
 479                 break;
 480         }
 481         node_append(mdoc, p);
 482         mdoc->next = MDOC_NEXT_CHILD;
 483 }
 484
 485 void
 486 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 487 {
 488         struct mdoc_node *n;
 489
 490         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 491         n->string = roff_strdup(mdoc->roff, p);
 492         node_append(mdoc, n);
 493         mdoc->next = MDOC_NEXT_SIBLING;
 494 }
 495
 496 void
 497 mdoc_word_append(struct mdoc *mdoc, const char *p)
 498 {
 499         struct mdoc_node        *n;
 500         char                    *addstr, *newstr;
 501
 502         n = mdoc->last;
 503         addstr = roff_strdup(mdoc->roff, p);
 504         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 505         free(addstr);
 506         free(n->string);
 507         n->string = newstr;
 508         mdoc->next = MDOC_NEXT_SIBLING;
 509 }
 510
 511 static void
 512 mdoc_node_free(struct mdoc_node *p)
 513 {
 514
 515         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 516                 free(p->norm);
 517         if (p->string)
 518                 free(p->string);
 519         if (p->args)
 520                 mdoc_argv_free(p->args);
 521         free(p);
 522 }
 523
 524 static void
 525 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 526 {
 527
 528         /* Adjust siblings. */
 529
 530         if (n->prev)
 531                 n->prev->next = n->next;
 532         if (n->next)
 533                 n->next->prev = n->prev;
 534
 535         /* Adjust parent. */
 536
 537         if (n->parent) {
 538                 n->parent->nchild--;
 539                 if (n->parent->child == n)
 540                         n->parent->child = n->prev ? n->prev : n->next;
 541                 if (n->parent->last == n)
 542                         n->parent->last = n->prev ? n->prev : NULL;
 543         }
 544
 545         /* Adjust parse point, if applicable. */
 546
 547         if (mdoc && mdoc->last == n) {
 548                 if (n->prev) {
 549                         mdoc->last = n->prev;
 550                         mdoc->next = MDOC_NEXT_SIBLING;
 551                 } else {
 552                         mdoc->last = n->parent;
 553                         mdoc->next = MDOC_NEXT_CHILD;
 554                 }
 555         }
 556
 557         if (mdoc && mdoc->first == n)
 558                 mdoc->first = NULL;
 559 }
 560
 561 void
 562 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 563 {
 564
 565         while (p->child) {
 566                 assert(p->nchild);
 567                 mdoc_node_delete(mdoc, p->child);
 568         }
 569         assert(0 == p->nchild);
 570
 571         mdoc_node_unlink(mdoc, p);
 572         mdoc_node_free(p);
 573 }
 574
 575 void
 576 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 577 {
 578
 579         mdoc_node_unlink(mdoc, p);
 580         node_append(mdoc, p);
 581 }
 582
 583 /*
 584  * Parse free-form text, that is, a line that does not begin with the
 585  * control character.
 586  */
 587 static int
 588 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 589 {
 590         char             *c, *ws, *end;
 591         struct mdoc_node *n;
 592
 593         assert(mdoc->last);
 594         n = mdoc->last;
 595
 596         /*
 597          * Divert directly to list processing if we're encountering a
 598          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 599          * (a MDOC_BODY means it's already open, in which case we should
 600          * process within its context in the normal way).
 601          */
 602
 603         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 604             LIST_column == n->norm->Bl.type) {
 605                 /* `Bl' is open without any children. */
 606                 mdoc->flags |= MDOC_FREECOL;
 607                 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
 608                 return(1);
 609         }
 610
 611         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 612             NULL != n->parent &&
 613             MDOC_Bl == n->parent->tok &&
 614             LIST_column == n->parent->norm->Bl.type) {
 615                 /* `Bl' has block-level `It' children. */
 616                 mdoc->flags |= MDOC_FREECOL;
 617                 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
 618                 return(1);
 619         }
 620
 621         /*
 622          * Search for the beginning of unescaped trailing whitespace (ws)
 623          * and for the first character not to be output (end).
 624          */
 625
 626         /* FIXME: replace with strcspn(). */
 627         ws = NULL;
 628         for (c = end = buf + offs; *c; c++) {
 629                 switch (*c) {
 630                 case ' ':
 631                         if (NULL == ws)
 632                                 ws = c;
 633                         continue;
 634                 case '\t':
 635                         /*
 636                          * Always warn about trailing tabs,
 637                          * even outside literal context,
 638                          * where they should be put on the next line.
 639                          */
 640                         if (NULL == ws)
 641                                 ws = c;
 642                         /*
 643                          * Strip trailing tabs in literal context only;
 644                          * outside, they affect the next line.
 645                          */
 646                         if (MDOC_LITERAL & mdoc->flags)
 647                                 continue;
 648                         break;
 649                 case '\\':
 650                         /* Skip the escaped character, too, if any. */
 651                         if (c[1])
 652                                 c++;
 653                         /* FALLTHROUGH */
 654                 default:
 655                         ws = NULL;
 656                         break;
 657                 }
 658                 end = c + 1;
 659         }
 660         *end = '\0';
 661
 662         if (ws)
 663                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 664                     line, (int)(ws-buf), NULL);
 665
 666         if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
 667                 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
 668                     line, (int)(c - buf), NULL);
 669
 670                 /*
 671                  * Insert a `sp' in the case of a blank line.  Technically,
 672                  * blank lines aren't allowed, but enough manuals assume this
 673                  * behaviour that we want to work around it.
 674                  */
 675                 mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL);
 676                 mdoc->next = MDOC_NEXT_SIBLING;
 677                 mdoc_valid_post(mdoc);
 678                 return(1);
 679         }
 680
 681         mdoc_word_alloc(mdoc, line, offs, buf+offs);
 682
 683         if (mdoc->flags & MDOC_LITERAL)
 684                 return(1);
 685
 686         /*
 687          * End-of-sentence check.  If the last character is an unescaped
 688          * EOS character, then flag the node as being the end of a
 689          * sentence.  The front-end will know how to interpret this.
 690          */
 691
 692         assert(buf < end);
 693
 694         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 695                 mdoc->last->flags |= MDOC_EOS;
 696         return(1);
 697 }
 698
 699 /*
 700  * Parse a macro line, that is, a line beginning with the control
 701  * character.
 702  */
 703 static int
 704 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 705 {
 706         struct mdoc_node *n;
 707         const char       *cp;
 708         enum mdoct        tok;
 709         int               i, sv;
 710         char              mac[5];
 711
 712         sv = offs;
 713
 714         /*
 715          * Copy the first word into a nil-terminated buffer.
 716          * Stop when a space, tab, escape, or eoln is encountered.
 717          */
 718
 719         i = 0;
 720         while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
 721                 mac[i++] = buf[offs++];
 722
 723         mac[i] = '\0';
 724
 725         tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 726
 727         if (tok == MDOC_MAX) {
 728                 mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
 729                     ln, sv, buf + sv - 1);
 730                 return(1);
 731         }
 732
 733         /* Skip a leading escape sequence or tab. */
 734
 735         switch (buf[offs]) {
 736         case '\\':
 737                 cp = buf + offs + 1;
 738                 mandoc_escape(&cp, NULL, NULL);
 739                 offs = cp - buf;
 740                 break;
 741         case '\t':
 742                 offs++;
 743                 break;
 744         default:
 745                 break;
 746         }
 747
 748         /* Jump to the next non-whitespace word. */
 749
 750         while (buf[offs] && ' ' == buf[offs])
 751                 offs++;
 752
 753         /*
 754          * Trailing whitespace.  Note that tabs are allowed to be passed
 755          * into the parser as "text", so we only warn about spaces here.
 756          */
 757
 758         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 759                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 760                     ln, offs - 1, NULL);
 761
 762         /*
 763          * If an initial macro or a list invocation, divert directly
 764          * into macro processing.
 765          */
 766
 767         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 768                 mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
 769                 return(1);
 770         }
 771
 772         n = mdoc->last;
 773         assert(mdoc->last);
 774
 775         /*
 776          * If the first macro of a `Bl -column', open an `It' block
 777          * context around the parsed macro.
 778          */
 779
 780         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 781             LIST_column == n->norm->Bl.type) {
 782                 mdoc->flags |= MDOC_FREECOL;
 783                 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
 784                 return(1);
 785         }
 786
 787         /*
 788          * If we're following a block-level `It' within a `Bl -column'
 789          * context (perhaps opened in the above block or in ptext()),
 790          * then open an `It' block context around the parsed macro.
 791          */
 792
 793         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 794             NULL != n->parent &&
 795             MDOC_Bl == n->parent->tok &&
 796             LIST_column == n->parent->norm->Bl.type) {
 797                 mdoc->flags |= MDOC_FREECOL;
 798                 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
 799                 return(1);
 800         }
 801
 802         /* Normal processing of a macro. */
 803
 804         mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
 805
 806         /* In quick mode (for mandocdb), abort after the NAME section. */
 807
 808         if (mdoc->quick && MDOC_Sh == tok &&
 809             SEC_NAME != mdoc->last->sec)
 810                 return(2);
 811
 812         return(1);
 813 }
 814
 815 enum mdelim
 816 mdoc_isdelim(const char *p)
 817 {
 818
 819         if ('\0' == p[0])
 820                 return(DELIM_NONE);
 821
 822         if ('\0' == p[1])
 823                 switch (p[0]) {
 824                 case '(':
 825                         /* FALLTHROUGH */
 826                 case '[':
 827                         return(DELIM_OPEN);
 828                 case '|':
 829                         return(DELIM_MIDDLE);
 830                 case '.':
 831                         /* FALLTHROUGH */
 832                 case ',':
 833                         /* FALLTHROUGH */
 834                 case ';':
 835                         /* FALLTHROUGH */
 836                 case ':':
 837                         /* FALLTHROUGH */
 838                 case '?':
 839                         /* FALLTHROUGH */
 840                 case '!':
 841                         /* FALLTHROUGH */
 842                 case ')':
 843                         /* FALLTHROUGH */
 844                 case ']':
 845                         return(DELIM_CLOSE);
 846                 default:
 847                         return(DELIM_NONE);
 848                 }
 849
 850         if ('\\' != p[0])
 851                 return(DELIM_NONE);
 852
 853         if (0 == strcmp(p + 1, "."))
 854                 return(DELIM_CLOSE);
 855         if (0 == strcmp(p + 1, "fR|\\fP"))
 856                 return(DELIM_MIDDLE);
 857
 858         return(DELIM_NONE);
 859 }
 860
 861 void
 862 mdoc_deroff(char **dest, const struct mdoc_node *n)
 863 {
 864         char    *cp;
 865         size_t   sz;
 866
 867         if (MDOC_TEXT != n->type) {
 868                 for (n = n->child; n; n = n->next)
 869                         mdoc_deroff(dest, n);
 870                 return;
 871         }
 872
 873         /* Skip leading whitespace. */
 874
 875         for (cp = n->string; '\0' != *cp; cp++)
 876                 if (0 == isspace((unsigned char)*cp))
 877                         break;
 878
 879         /* Skip trailing whitespace. */
 880
 881         for (sz = strlen(cp); sz; sz--)
 882                 if (0 == isspace((unsigned char)cp[sz-1]))
 883                         break;
 884
 885         /* Skip empty strings. */
 886
 887         if (0 == sz)
 888                 return;
 889
 890         if (NULL == *dest) {
 891                 *dest = mandoc_strndup(cp, sz);
 892                 return;
 893         }
 894
 895         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 896         free(*dest);
 897         *dest = cp;
 898 }