mdoc.c

   1 /*      $Id: mdoc.c,v 1.241 2015/04/02 23:48:19 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010, 2012-2015 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 AUTHORS DISCLAIM ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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 "mandoc_aux.h"
  31 #include "mandoc.h"
  32 #include "roff.h"
  33 #include "mdoc.h"
  34 #include "libmandoc.h"
  35 #include "libmdoc.h"
  36
  37 const   char *const __mdoc_macronames[MDOC_MAX + 1] = {
  38         "Ap",           "Dd",           "Dt",           "Os",
  39         "Sh",           "Ss",           "Pp",           "D1",
  40         "Dl",           "Bd",           "Ed",           "Bl",
  41         "El",           "It",           "Ad",           "An",
  42         "Ar",           "Cd",           "Cm",           "Dv",
  43         "Er",           "Ev",           "Ex",           "Fa",
  44         "Fd",           "Fl",           "Fn",           "Ft",
  45         "Ic",           "In",           "Li",           "Nd",
  46         "Nm",           "Op",           "Ot",           "Pa",
  47         "Rv",           "St",           "Va",           "Vt",
  48         "Xr",           "%A",           "%B",           "%D",
  49         "%I",           "%J",           "%N",           "%O",
  50         "%P",           "%R",           "%T",           "%V",
  51         "Ac",           "Ao",           "Aq",           "At",
  52         "Bc",           "Bf",           "Bo",           "Bq",
  53         "Bsx",          "Bx",           "Db",           "Dc",
  54         "Do",           "Dq",           "Ec",           "Ef",
  55         "Em",           "Eo",           "Fx",           "Ms",
  56         "No",           "Ns",           "Nx",           "Ox",
  57         "Pc",           "Pf",           "Po",           "Pq",
  58         "Qc",           "Ql",           "Qo",           "Qq",
  59         "Re",           "Rs",           "Sc",           "So",
  60         "Sq",           "Sm",           "Sx",           "Sy",
  61         "Tn",           "Ux",           "Xc",           "Xo",
  62         "Fo",           "Fc",           "Oo",           "Oc",
  63         "Bk",           "Ek",           "Bt",           "Hf",
  64         "Fr",           "Ud",           "Lb",           "Lp",
  65         "Lk",           "Mt",           "Brq",          "Bro",
  66         "Brc",          "%C",           "Es",           "En",
  67         "Dx",           "%Q",           "br",           "sp",
  68         "%U",           "Ta",           "ll",           "text",
  69         };
  70
  71 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  72         "split",                "nosplit",              "ragged",
  73         "unfilled",             "literal",              "file",
  74         "offset",               "bullet",               "dash",
  75         "hyphen",               "item",                 "enum",
  76         "tag",                  "diag",                 "hang",
  77         "ohang",                "inset",                "column",
  78         "width",                "compact",              "std",
  79         "filled",               "words",                "emphasis",
  80         "symbolic",             "nested",               "centered"
  81         };
  82
  83 const   char * const *mdoc_macronames = __mdoc_macronames;
  84 const   char * const *mdoc_argnames = __mdoc_argnames;
  85
  86 static  void              mdoc_node_free(struct roff_node *);
  87 static  void              mdoc_node_unlink(struct mdoc *,
  88                                 struct roff_node *);
  89 static  void              mdoc_free1(struct mdoc *);
  90 static  void              mdoc_alloc1(struct mdoc *);
  91 static  struct roff_node *node_alloc(struct mdoc *, int, int,
  92                                 int, enum roff_type);
  93 static  void              node_append(struct mdoc *, struct roff_node *);
  94 static  int               mdoc_ptext(struct mdoc *, int, char *, int);
  95 static  int               mdoc_pmacro(struct mdoc *, int, char *, int);
  96
  97
  98 const struct roff_node *
  99 mdoc_node(const struct mdoc *mdoc)
 100 {
 101
 102         return(mdoc->first);
 103 }
 104
 105 const struct roff_meta *
 106 mdoc_meta(const struct mdoc *mdoc)
 107 {
 108
 109         return(&mdoc->meta);
 110 }
 111
 112 /*
 113  * Frees volatile resources (parse tree, meta-data, fields).
 114  */
 115 static void
 116 mdoc_free1(struct mdoc *mdoc)
 117 {
 118
 119         if (mdoc->first)
 120                 mdoc_node_delete(mdoc, mdoc->first);
 121         free(mdoc->meta.msec);
 122         free(mdoc->meta.vol);
 123         free(mdoc->meta.arch);
 124         free(mdoc->meta.date);
 125         free(mdoc->meta.title);
 126         free(mdoc->meta.os);
 127         free(mdoc->meta.name);
 128 }
 129
 130 /*
 131  * Allocate all volatile resources (parse tree, meta-data, fields).
 132  */
 133 static void
 134 mdoc_alloc1(struct mdoc *mdoc)
 135 {
 136
 137         memset(&mdoc->meta, 0, sizeof(mdoc->meta));
 138         mdoc->flags = 0;
 139         mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 140         mdoc->last = mandoc_calloc(1, sizeof(*mdoc->last));
 141         mdoc->first = mdoc->last;
 142         mdoc->last->type = ROFFT_ROOT;
 143         mdoc->last->tok = MDOC_MAX;
 144         mdoc->next = MDOC_NEXT_CHILD;
 145 }
 146
 147 /*
 148  * Free up volatile resources (see mdoc_free1()) then re-initialises the
 149  * data with mdoc_alloc1().  After invocation, parse data has been reset
 150  * and the parser is ready for re-invocation on a new tree; however,
 151  * cross-parse non-volatile data is kept intact.
 152  */
 153 void
 154 mdoc_reset(struct mdoc *mdoc)
 155 {
 156
 157         mdoc_free1(mdoc);
 158         mdoc_alloc1(mdoc);
 159 }
 160
 161 /*
 162  * Completely free up all volatile and non-volatile parse resources.
 163  * After invocation, the pointer is no longer usable.
 164  */
 165 void
 166 mdoc_free(struct mdoc *mdoc)
 167 {
 168
 169         mdoc_free1(mdoc);
 170         free(mdoc);
 171 }
 172
 173 /*
 174  * Allocate volatile and non-volatile parse resources.
 175  */
 176 struct mdoc *
 177 mdoc_alloc(struct roff *roff, struct mparse *parse,
 178         const char *defos, int quick)
 179 {
 180         struct mdoc     *p;
 181
 182         p = mandoc_calloc(1, sizeof(struct mdoc));
 183
 184         p->parse = parse;
 185         p->defos = defos;
 186         p->quick = quick;
 187         p->roff = roff;
 188
 189         mdoc_hash_init();
 190         mdoc_alloc1(p);
 191         return(p);
 192 }
 193
 194 void
 195 mdoc_endparse(struct mdoc *mdoc)
 196 {
 197
 198         mdoc_macroend(mdoc);
 199 }
 200
 201 void
 202 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 203 {
 204         struct roff_node *n;
 205
 206         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, ROFFT_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 roff_node *n;
 218
 219         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, ROFFT_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 != ROFFT_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 roff_node *p)
 281 {
 282
 283         assert(mdoc->last);
 284         assert(mdoc->first);
 285         assert(p->type != ROFFT_ROOT);
 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 ROFFT_BODY:
 311                 if (ENDBODY_NOT != p->end)
 312                         break;
 313                 /* FALLTHROUGH */
 314         case ROFFT_TAIL:
 315                 /* FALLTHROUGH */
 316         case ROFFT_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 ROFFT_HEAD:
 327                 assert(p->parent->type == ROFFT_BLOCK);
 328                 p->parent->head = p;
 329                 break;
 330         case ROFFT_TAIL:
 331                 assert(p->parent->type == ROFFT_BLOCK);
 332                 p->parent->tail = p;
 333                 break;
 334         case ROFFT_BODY:
 335                 if (p->end)
 336                         break;
 337                 assert(p->parent->type == ROFFT_BLOCK);
 338                 p->parent->body = p;
 339                 break;
 340         default:
 341                 break;
 342         }
 343
 344         mdoc->last = p;
 345
 346         switch (p->type) {
 347         case ROFFT_TBL:
 348                 /* FALLTHROUGH */
 349         case ROFFT_TEXT:
 350                 mdoc_valid_post(mdoc);
 351                 break;
 352         default:
 353                 break;
 354         }
 355 }
 356
 357 static struct roff_node *
 358 node_alloc(struct mdoc *mdoc, int line, int pos,
 359         int tok, enum roff_type type)
 360 {
 361         struct roff_node *p;
 362
 363         p = mandoc_calloc(1, sizeof(*p));
 364         p->sec = mdoc->lastsec;
 365         p->line = line;
 366         p->pos = pos;
 367         p->tok = tok;
 368         p->type = type;
 369
 370         /* Flag analysis. */
 371
 372         if (MDOC_SYNOPSIS & mdoc->flags)
 373                 p->flags |= MDOC_SYNPRETTY;
 374         else
 375                 p->flags &= ~MDOC_SYNPRETTY;
 376         if (MDOC_NEWLINE & mdoc->flags)
 377                 p->flags |= MDOC_LINE;
 378         mdoc->flags &= ~MDOC_NEWLINE;
 379
 380         return(p);
 381 }
 382
 383 void
 384 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 385 {
 386         struct roff_node *p;
 387
 388         p = node_alloc(mdoc, line, pos, tok, ROFFT_TAIL);
 389         node_append(mdoc, p);
 390         mdoc->next = MDOC_NEXT_CHILD;
 391 }
 392
 393 struct roff_node *
 394 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 395 {
 396         struct roff_node *p;
 397
 398         assert(mdoc->first);
 399         assert(mdoc->last);
 400         p = node_alloc(mdoc, line, pos, tok, ROFFT_HEAD);
 401         node_append(mdoc, p);
 402         mdoc->next = MDOC_NEXT_CHILD;
 403         return(p);
 404 }
 405
 406 struct roff_node *
 407 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, int tok)
 408 {
 409         struct roff_node *p;
 410
 411         p = node_alloc(mdoc, line, pos, tok, ROFFT_BODY);
 412         node_append(mdoc, p);
 413         mdoc->next = MDOC_NEXT_CHILD;
 414         return(p);
 415 }
 416
 417 struct roff_node *
 418 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, int tok,
 419                 struct roff_node *body, enum mdoc_endbody end)
 420 {
 421         struct roff_node *p;
 422
 423         body->flags |= MDOC_ENDED;
 424         body->parent->flags |= MDOC_ENDED;
 425         p = node_alloc(mdoc, line, pos, tok, ROFFT_BODY);
 426         p->body = body;
 427         p->norm = body->norm;
 428         p->end = end;
 429         node_append(mdoc, p);
 430         mdoc->next = MDOC_NEXT_SIBLING;
 431         return(p);
 432 }
 433
 434 struct roff_node *
 435 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 436         int tok, struct mdoc_arg *args)
 437 {
 438         struct roff_node *p;
 439
 440         p = node_alloc(mdoc, line, pos, tok, ROFFT_BLOCK);
 441         p->args = args;
 442         if (p->args)
 443                 (args->refcnt)++;
 444
 445         switch (tok) {
 446         case MDOC_Bd:
 447                 /* FALLTHROUGH */
 448         case MDOC_Bf:
 449                 /* FALLTHROUGH */
 450         case MDOC_Bl:
 451                 /* FALLTHROUGH */
 452         case MDOC_En:
 453                 /* FALLTHROUGH */
 454         case MDOC_Rs:
 455                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 456                 break;
 457         default:
 458                 break;
 459         }
 460         node_append(mdoc, p);
 461         mdoc->next = MDOC_NEXT_CHILD;
 462         return(p);
 463 }
 464
 465 void
 466 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 467         int tok, struct mdoc_arg *args)
 468 {
 469         struct roff_node *p;
 470
 471         p = node_alloc(mdoc, line, pos, tok, ROFFT_ELEM);
 472         p->args = args;
 473         if (p->args)
 474                 (args->refcnt)++;
 475
 476         switch (tok) {
 477         case MDOC_An:
 478                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 479                 break;
 480         default:
 481                 break;
 482         }
 483         node_append(mdoc, p);
 484         mdoc->next = MDOC_NEXT_CHILD;
 485 }
 486
 487 void
 488 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 489 {
 490         struct roff_node *n;
 491
 492         n = node_alloc(mdoc, line, pos, MDOC_MAX, ROFFT_TEXT);
 493         n->string = roff_strdup(mdoc->roff, p);
 494         node_append(mdoc, n);
 495         mdoc->next = MDOC_NEXT_SIBLING;
 496 }
 497
 498 void
 499 mdoc_word_append(struct mdoc *mdoc, const char *p)
 500 {
 501         struct roff_node        *n;
 502         char                    *addstr, *newstr;
 503
 504         n = mdoc->last;
 505         addstr = roff_strdup(mdoc->roff, p);
 506         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 507         free(addstr);
 508         free(n->string);
 509         n->string = newstr;
 510         mdoc->next = MDOC_NEXT_SIBLING;
 511 }
 512
 513 static void
 514 mdoc_node_free(struct roff_node *p)
 515 {
 516
 517         if (p->type == ROFFT_BLOCK || p->type == ROFFT_ELEM)
 518                 free(p->norm);
 519         if (p->string)
 520                 free(p->string);
 521         if (p->args)
 522                 mdoc_argv_free(p->args);
 523         free(p);
 524 }
 525
 526 static void
 527 mdoc_node_unlink(struct mdoc *mdoc, struct roff_node *n)
 528 {
 529
 530         /* Adjust siblings. */
 531
 532         if (n->prev)
 533                 n->prev->next = n->next;
 534         if (n->next)
 535                 n->next->prev = n->prev;
 536
 537         /* Adjust parent. */
 538
 539         if (n->parent) {
 540                 n->parent->nchild--;
 541                 if (n->parent->child == n)
 542                         n->parent->child = n->prev ? n->prev : n->next;
 543                 if (n->parent->last == n)
 544                         n->parent->last = n->prev ? n->prev : NULL;
 545         }
 546
 547         /* Adjust parse point, if applicable. */
 548
 549         if (mdoc && mdoc->last == n) {
 550                 if (n->prev) {
 551                         mdoc->last = n->prev;
 552                         mdoc->next = MDOC_NEXT_SIBLING;
 553                 } else {
 554                         mdoc->last = n->parent;
 555                         mdoc->next = MDOC_NEXT_CHILD;
 556                 }
 557         }
 558
 559         if (mdoc && mdoc->first == n)
 560                 mdoc->first = NULL;
 561 }
 562
 563 void
 564 mdoc_node_delete(struct mdoc *mdoc, struct roff_node *p)
 565 {
 566
 567         while (p->child) {
 568                 assert(p->nchild);
 569                 mdoc_node_delete(mdoc, p->child);
 570         }
 571         assert(0 == p->nchild);
 572
 573         mdoc_node_unlink(mdoc, p);
 574         mdoc_node_free(p);
 575 }
 576
 577 void
 578 mdoc_node_relink(struct mdoc *mdoc, struct roff_node *p)
 579 {
 580
 581         mdoc_node_unlink(mdoc, p);
 582         node_append(mdoc, p);
 583 }
 584
 585 /*
 586  * Parse free-form text, that is, a line that does not begin with the
 587  * control character.
 588  */
 589 static int
 590 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 591 {
 592         struct roff_node *n;
 593         char             *c, *ws, *end;
 594
 595         assert(mdoc->last);
 596         n = mdoc->last;
 597
 598         /*
 599          * Divert directly to list processing if we're encountering a
 600          * columnar ROFFT_BLOCK with or without a prior ROFFT_BLOCK entry
 601          * (a ROFFT_BODY means it's already open, in which case we should
 602          * process within its context in the normal way).
 603          */
 604
 605         if (n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
 606             n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
 607                 /* `Bl' is open without any children. */
 608                 mdoc->flags |= MDOC_FREECOL;
 609                 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
 610                 return(1);
 611         }
 612
 613         if (n->tok == MDOC_It && n->type == ROFFT_BLOCK &&
 614             NULL != n->parent &&
 615             MDOC_Bl == n->parent->tok &&
 616             LIST_column == n->parent->norm->Bl.type) {
 617                 /* `Bl' has block-level `It' children. */
 618                 mdoc->flags |= MDOC_FREECOL;
 619                 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
 620                 return(1);
 621         }
 622
 623         /*
 624          * Search for the beginning of unescaped trailing whitespace (ws)
 625          * and for the first character not to be output (end).
 626          */
 627
 628         /* FIXME: replace with strcspn(). */
 629         ws = NULL;
 630         for (c = end = buf + offs; *c; c++) {
 631                 switch (*c) {
 632                 case ' ':
 633                         if (NULL == ws)
 634                                 ws = c;
 635                         continue;
 636                 case '\t':
 637                         /*
 638                          * Always warn about trailing tabs,
 639                          * even outside literal context,
 640                          * where they should be put on the next line.
 641                          */
 642                         if (NULL == ws)
 643                                 ws = c;
 644                         /*
 645                          * Strip trailing tabs in literal context only;
 646                          * outside, they affect the next line.
 647                          */
 648                         if (MDOC_LITERAL & mdoc->flags)
 649                                 continue;
 650                         break;
 651                 case '\\':
 652                         /* Skip the escaped character, too, if any. */
 653                         if (c[1])
 654                                 c++;
 655                         /* FALLTHROUGH */
 656                 default:
 657                         ws = NULL;
 658                         break;
 659                 }
 660                 end = c + 1;
 661         }
 662         *end = '\0';
 663
 664         if (ws)
 665                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 666                     line, (int)(ws-buf), NULL);
 667
 668         if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
 669                 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
 670                     line, (int)(c - buf), NULL);
 671
 672                 /*
 673                  * Insert a `sp' in the case of a blank line.  Technically,
 674                  * blank lines aren't allowed, but enough manuals assume this
 675                  * behaviour that we want to work around it.
 676                  */
 677                 mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL);
 678                 mdoc->next = MDOC_NEXT_SIBLING;
 679                 mdoc_valid_post(mdoc);
 680                 return(1);
 681         }
 682
 683         mdoc_word_alloc(mdoc, line, offs, buf+offs);
 684
 685         if (mdoc->flags & MDOC_LITERAL)
 686                 return(1);
 687
 688         /*
 689          * End-of-sentence check.  If the last character is an unescaped
 690          * EOS character, then flag the node as being the end of a
 691          * sentence.  The front-end will know how to interpret this.
 692          */
 693
 694         assert(buf < end);
 695
 696         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 697                 mdoc->last->flags |= MDOC_EOS;
 698         return(1);
 699 }
 700
 701 /*
 702  * Parse a macro line, that is, a line beginning with the control
 703  * character.
 704  */
 705 static int
 706 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 707 {
 708         struct roff_node *n;
 709         const char       *cp;
 710         int               tok;
 711         int               i, sv;
 712         char              mac[5];
 713
 714         sv = offs;
 715
 716         /*
 717          * Copy the first word into a nil-terminated buffer.
 718          * Stop when a space, tab, escape, or eoln is encountered.
 719          */
 720
 721         i = 0;
 722         while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
 723                 mac[i++] = buf[offs++];
 724
 725         mac[i] = '\0';
 726
 727         tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 728
 729         if (tok == MDOC_MAX) {
 730                 mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
 731                     ln, sv, buf + sv - 1);
 732                 return(1);
 733         }
 734
 735         /* Skip a leading escape sequence or tab. */
 736
 737         switch (buf[offs]) {
 738         case '\\':
 739                 cp = buf + offs + 1;
 740                 mandoc_escape(&cp, NULL, NULL);
 741                 offs = cp - buf;
 742                 break;
 743         case '\t':
 744                 offs++;
 745                 break;
 746         default:
 747                 break;
 748         }
 749
 750         /* Jump to the next non-whitespace word. */
 751
 752         while (buf[offs] && ' ' == buf[offs])
 753                 offs++;
 754
 755         /*
 756          * Trailing whitespace.  Note that tabs are allowed to be passed
 757          * into the parser as "text", so we only warn about spaces here.
 758          */
 759
 760         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 761                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 762                     ln, offs - 1, NULL);
 763
 764         /*
 765          * If an initial macro or a list invocation, divert directly
 766          * into macro processing.
 767          */
 768
 769         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 770                 mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
 771                 return(1);
 772         }
 773
 774         n = mdoc->last;
 775         assert(mdoc->last);
 776
 777         /*
 778          * If the first macro of a `Bl -column', open an `It' block
 779          * context around the parsed macro.
 780          */
 781
 782         if (n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
 783             n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
 784                 mdoc->flags |= MDOC_FREECOL;
 785                 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
 786                 return(1);
 787         }
 788
 789         /*
 790          * If we're following a block-level `It' within a `Bl -column'
 791          * context (perhaps opened in the above block or in ptext()),
 792          * then open an `It' block context around the parsed macro.
 793          */
 794
 795         if (n->tok == MDOC_It && n->type == ROFFT_BLOCK &&
 796             NULL != n->parent &&
 797             MDOC_Bl == n->parent->tok &&
 798             LIST_column == n->parent->norm->Bl.type) {
 799                 mdoc->flags |= MDOC_FREECOL;
 800                 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
 801                 return(1);
 802         }
 803
 804         /* Normal processing of a macro. */
 805
 806         mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
 807
 808         /* In quick mode (for mandocdb), abort after the NAME section. */
 809
 810         if (mdoc->quick && MDOC_Sh == tok &&
 811             SEC_NAME != mdoc->last->sec)
 812                 return(2);
 813
 814         return(1);
 815 }
 816
 817 enum mdelim
 818 mdoc_isdelim(const char *p)
 819 {
 820
 821         if ('\0' == p[0])
 822                 return(DELIM_NONE);
 823
 824         if ('\0' == p[1])
 825                 switch (p[0]) {
 826                 case '(':
 827                         /* FALLTHROUGH */
 828                 case '[':
 829                         return(DELIM_OPEN);
 830                 case '|':
 831                         return(DELIM_MIDDLE);
 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                         /* FALLTHROUGH */
 846                 case ']':
 847                         return(DELIM_CLOSE);
 848                 default:
 849                         return(DELIM_NONE);
 850                 }
 851
 852         if ('\\' != p[0])
 853                 return(DELIM_NONE);
 854
 855         if (0 == strcmp(p + 1, "."))
 856                 return(DELIM_CLOSE);
 857         if (0 == strcmp(p + 1, "fR|\\fP"))
 858                 return(DELIM_MIDDLE);
 859
 860         return(DELIM_NONE);
 861 }
 862
 863 void
 864 mdoc_deroff(char **dest, const struct roff_node *n)
 865 {
 866         char    *cp;
 867         size_t   sz;
 868
 869         if (n->type != ROFFT_TEXT) {
 870                 for (n = n->child; n; n = n->next)
 871                         mdoc_deroff(dest, n);
 872                 return;
 873         }
 874
 875         /* Skip leading whitespace. */
 876
 877         for (cp = n->string; '\0' != *cp; cp++)
 878                 if (0 == isspace((unsigned char)*cp))
 879                         break;
 880
 881         /* Skip trailing whitespace. */
 882
 883         for (sz = strlen(cp); sz; sz--)
 884                 if (0 == isspace((unsigned char)cp[sz-1]))
 885                         break;
 886
 887         /* Skip empty strings. */
 888
 889         if (0 == sz)
 890                 return;
 891
 892         if (NULL == *dest) {
 893                 *dest = mandoc_strndup(cp, sz);
 894                 return;
 895         }
 896
 897         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 898         free(*dest);
 899         *dest = cp;
 900 }