mdoc_macro.c

   1 /*      $Id: mdoc_macro.c,v 1.106 2011/03/22 14:33:05 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010 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 <assert.h>
  23 #include <ctype.h>
  24 #include <stdlib.h>
  25 #include <stdio.h>
  26 #include <string.h>
  27 #include <time.h>
  28
  29 #include "mdoc.h"
  30 #include "mandoc.h"
  31 #include "libmdoc.h"
  32 #include "libmandoc.h"
  33
  34 enum    rew {   /* see rew_dohalt() */
  35         REWIND_NONE,
  36         REWIND_THIS,
  37         REWIND_MORE,
  38         REWIND_FORCE,
  39         REWIND_LATER,
  40         REWIND_ERROR
  41 };
  42
  43 static  int             blk_full(MACRO_PROT_ARGS);
  44 static  int             blk_exp_close(MACRO_PROT_ARGS);
  45 static  int             blk_part_exp(MACRO_PROT_ARGS);
  46 static  int             blk_part_imp(MACRO_PROT_ARGS);
  47 static  int             ctx_synopsis(MACRO_PROT_ARGS);
  48 static  int             in_line_eoln(MACRO_PROT_ARGS);
  49 static  int             in_line_argn(MACRO_PROT_ARGS);
  50 static  int             in_line(MACRO_PROT_ARGS);
  51 static  int             obsolete(MACRO_PROT_ARGS);
  52 static  int             phrase_ta(MACRO_PROT_ARGS);
  53
  54 static  int             dword(struct mdoc *, int, int,
  55                                 const char *, enum mdelim);
  56 static  int             append_delims(struct mdoc *,
  57                                 int, int *, char *);
  58 static  enum mdoct      lookup(enum mdoct, const char *);
  59 static  enum mdoct      lookup_raw(const char *);
  60 static  int             make_pending(struct mdoc_node *, enum mdoct,
  61                                 struct mdoc *, int, int);
  62 static  int             phrase(struct mdoc *, int, int, char *);
  63 static  enum mdoct      rew_alt(enum mdoct);
  64 static  enum rew        rew_dohalt(enum mdoct, enum mdoc_type,
  65                                 const struct mdoc_node *);
  66 static  int             rew_elem(struct mdoc *, enum mdoct);
  67 static  int             rew_last(struct mdoc *,
  68                                 const struct mdoc_node *);
  69 static  int             rew_sub(enum mdoc_type, struct mdoc *,
  70                                 enum mdoct, int, int);
  71
  72 const   struct mdoc_macro __mdoc_macros[MDOC_MAX] = {
  73         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ap */
  74         { in_line_eoln, MDOC_PROLOGUE }, /* Dd */
  75         { in_line_eoln, MDOC_PROLOGUE }, /* Dt */
  76         { in_line_eoln, MDOC_PROLOGUE }, /* Os */
  77         { blk_full, 0 }, /* Sh */
  78         { blk_full, 0 }, /* Ss */
  79         { in_line_eoln, 0 }, /* Pp */
  80         { blk_part_imp, MDOC_PARSED }, /* D1 */
  81         { blk_part_imp, MDOC_PARSED }, /* Dl */
  82         { blk_full, MDOC_EXPLICIT }, /* Bd */
  83         { blk_exp_close, MDOC_EXPLICIT }, /* Ed */
  84         { blk_full, MDOC_EXPLICIT }, /* Bl */
  85         { blk_exp_close, MDOC_EXPLICIT }, /* El */
  86         { blk_full, MDOC_PARSED }, /* It */
  87         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */
  88         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* An */
  89         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */
  90         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cd */
  91         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */
  92         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */
  93         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */
  94         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */
  95         { in_line_eoln, 0 }, /* Ex */
  96         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */
  97         { in_line_eoln, 0 }, /* Fd */
  98         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */
  99         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */
 100         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */
 101         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ic */
 102         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */
 103         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Li */
 104         { blk_full, 0 }, /* Nd */
 105         { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */
 106         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */
 107         { obsolete, 0 }, /* Ot */
 108         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */
 109         { in_line_eoln, 0 }, /* Rv */
 110         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */
 111         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */
 112         { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */
 113         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */
 114         { in_line_eoln, 0 }, /* %A */
 115         { in_line_eoln, 0 }, /* %B */
 116         { in_line_eoln, 0 }, /* %D */
 117         { in_line_eoln, 0 }, /* %I */
 118         { in_line_eoln, 0 }, /* %J */
 119         { in_line_eoln, 0 }, /* %N */
 120         { in_line_eoln, 0 }, /* %O */
 121         { in_line_eoln, 0 }, /* %P */
 122         { in_line_eoln, 0 }, /* %R */
 123         { in_line_eoln, 0 }, /* %T */
 124         { in_line_eoln, 0 }, /* %V */
 125         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ac */
 126         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ao */
 127         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Aq */
 128         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */
 129         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Bc */
 130         { blk_full, MDOC_EXPLICIT }, /* Bf */
 131         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bo */
 132         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Bq */
 133         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */
 134         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */
 135         { in_line_eoln, 0 }, /* Db */
 136         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Dc */
 137         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Do */
 138         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Dq */
 139         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ec */
 140         { blk_exp_close, MDOC_EXPLICIT }, /* Ef */
 141         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Em */
 142         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */
 143         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */
 144         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */
 145         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* No */
 146         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Ns */
 147         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */
 148         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */
 149         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Pc */
 150         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */
 151         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Po */
 152         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Pq */
 153         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Qc */
 154         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Ql */
 155         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Qo */
 156         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Qq */
 157         { blk_exp_close, MDOC_EXPLICIT }, /* Re */
 158         { blk_full, MDOC_EXPLICIT }, /* Rs */
 159         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Sc */
 160         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* So */
 161         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Sq */
 162         { in_line_eoln, 0 }, /* Sm */
 163         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sx */
 164         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sy */
 165         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */
 166         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ux */
 167         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */
 168         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */
 169         { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */
 170         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Fc */
 171         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Oo */
 172         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Oc */
 173         { blk_full, MDOC_EXPLICIT }, /* Bk */
 174         { blk_exp_close, MDOC_EXPLICIT }, /* Ek */
 175         { in_line_eoln, 0 }, /* Bt */
 176         { in_line_eoln, 0 }, /* Hf */
 177         { obsolete, 0 }, /* Fr */
 178         { in_line_eoln, 0 }, /* Ud */
 179         { in_line, 0 }, /* Lb */
 180         { in_line_eoln, 0 }, /* Lp */
 181         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */
 182         { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */
 183         { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Brq */
 184         { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bro */
 185         { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Brc */
 186         { in_line_eoln, 0 }, /* %C */
 187         { obsolete, 0 }, /* Es */
 188         { obsolete, 0 }, /* En */
 189         { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */
 190         { in_line_eoln, 0 }, /* %Q */
 191         { in_line_eoln, 0 }, /* br */
 192         { in_line_eoln, 0 }, /* sp */
 193         { in_line_eoln, 0 }, /* %U */
 194         { phrase_ta, MDOC_CALLABLE | MDOC_PARSED }, /* Ta */
 195 };
 196
 197 const   struct mdoc_macro * const mdoc_macros = __mdoc_macros;
 198
 199
 200 /*
 201  * This is called at the end of parsing.  It must traverse up the tree,
 202  * closing out open [implicit] scopes.  Obviously, open explicit scopes
 203  * are errors.
 204  */
 205 int
 206 mdoc_macroend(struct mdoc *m)
 207 {
 208         struct mdoc_node *n;
 209
 210         /* Scan for open explicit scopes. */
 211
 212         n = MDOC_VALID & m->last->flags ?  m->last->parent : m->last;
 213
 214         for ( ; n; n = n->parent)
 215                 if (MDOC_BLOCK == n->type &&
 216                     MDOC_EXPLICIT & mdoc_macros[n->tok].flags)
 217                         mdoc_nmsg(m, n, MANDOCERR_SCOPEEXIT);
 218
 219         /* Rewind to the first. */
 220
 221         return(rew_last(m, m->first));
 222 }
 223
 224
 225 /*
 226  * Look up a macro from within a subsequent context.
 227  */
 228 static enum mdoct
 229 lookup(enum mdoct from, const char *p)
 230 {
 231         /* FIXME: make -diag lists be un-PARSED. */
 232
 233         if ( ! (MDOC_PARSED & mdoc_macros[from].flags))
 234                 return(MDOC_MAX);
 235         return(lookup_raw(p));
 236 }
 237
 238
 239 /*
 240  * Lookup a macro following the initial line macro.
 241  */
 242 static enum mdoct
 243 lookup_raw(const char *p)
 244 {
 245         enum mdoct       res;
 246
 247         if (MDOC_MAX == (res = mdoc_hash_find(p)))
 248                 return(MDOC_MAX);
 249         if (MDOC_CALLABLE & mdoc_macros[res].flags)
 250                 return(res);
 251         return(MDOC_MAX);
 252 }
 253
 254
 255 static int
 256 rew_last(struct mdoc *mdoc, const struct mdoc_node *to)
 257 {
 258         struct mdoc_node *n, *np;
 259
 260         assert(to);
 261         mdoc->next = MDOC_NEXT_SIBLING;
 262
 263         /* LINTED */
 264         while (mdoc->last != to) {
 265                 /*
 266                  * Save the parent here, because we may delete the
 267                  * m->last node in the post-validation phase and reset
 268                  * it to m->last->parent, causing a step in the closing
 269                  * out to be lost.
 270                  */
 271                 np = mdoc->last->parent;
 272                 if ( ! mdoc_valid_post(mdoc))
 273                         return(0);
 274                 n = mdoc->last;
 275                 mdoc->last = np;
 276                 assert(mdoc->last);
 277                 mdoc->last->last = n;
 278         }
 279
 280         return(mdoc_valid_post(mdoc));
 281 }
 282
 283
 284 /*
 285  * For a block closing macro, return the corresponding opening one.
 286  * Otherwise, return the macro itself.
 287  */
 288 static enum mdoct
 289 rew_alt(enum mdoct tok)
 290 {
 291         switch (tok) {
 292         case (MDOC_Ac):
 293                 return(MDOC_Ao);
 294         case (MDOC_Bc):
 295                 return(MDOC_Bo);
 296         case (MDOC_Brc):
 297                 return(MDOC_Bro);
 298         case (MDOC_Dc):
 299                 return(MDOC_Do);
 300         case (MDOC_Ec):
 301                 return(MDOC_Eo);
 302         case (MDOC_Ed):
 303                 return(MDOC_Bd);
 304         case (MDOC_Ef):
 305                 return(MDOC_Bf);
 306         case (MDOC_Ek):
 307                 return(MDOC_Bk);
 308         case (MDOC_El):
 309                 return(MDOC_Bl);
 310         case (MDOC_Fc):
 311                 return(MDOC_Fo);
 312         case (MDOC_Oc):
 313                 return(MDOC_Oo);
 314         case (MDOC_Pc):
 315                 return(MDOC_Po);
 316         case (MDOC_Qc):
 317                 return(MDOC_Qo);
 318         case (MDOC_Re):
 319                 return(MDOC_Rs);
 320         case (MDOC_Sc):
 321                 return(MDOC_So);
 322         case (MDOC_Xc):
 323                 return(MDOC_Xo);
 324         default:
 325                 return(tok);
 326         }
 327         /* NOTREACHED */
 328 }
 329
 330
 331 /*
 332  * Rewinding to tok, how do we have to handle *p?
 333  * REWIND_NONE: *p would delimit tok, but no tok scope is open
 334  *   inside *p, so there is no need to rewind anything at all.
 335  * REWIND_THIS: *p matches tok, so rewind *p and nothing else.
 336  * REWIND_MORE: *p is implicit, rewind it and keep searching for tok.
 337  * REWIND_FORCE: *p is explicit, but tok is full, force rewinding *p.
 338  * REWIND_LATER: *p is explicit and still open, postpone rewinding.
 339  * REWIND_ERROR: No tok block is open at all.
 340  */
 341 static enum rew
 342 rew_dohalt(enum mdoct tok, enum mdoc_type type,
 343                 const struct mdoc_node *p)
 344 {
 345
 346         /*
 347          * No matching token, no delimiting block, no broken block.
 348          * This can happen when full implicit macros are called for
 349          * the first time but try to rewind their previous
 350          * instance anyway.
 351          */
 352         if (MDOC_ROOT == p->type)
 353                 return(MDOC_BLOCK == type &&
 354                     MDOC_EXPLICIT & mdoc_macros[tok].flags ?
 355                     REWIND_ERROR : REWIND_NONE);
 356
 357         /*
 358          * When starting to rewind, skip plain text
 359          * and nodes that have already been rewound.
 360          */
 361         if (MDOC_TEXT == p->type || MDOC_VALID & p->flags)
 362                 return(REWIND_MORE);
 363
 364         /*
 365          * The easiest case:  Found a matching token.
 366          * This applies to both blocks and elements.
 367          */
 368         tok = rew_alt(tok);
 369         if (tok == p->tok)
 370                 return(p->end ? REWIND_NONE :
 371                     type == p->type ? REWIND_THIS : REWIND_MORE);
 372
 373         /*
 374          * While elements do require rewinding for themselves,
 375          * they never affect rewinding of other nodes.
 376          */
 377         if (MDOC_ELEM == p->type)
 378                 return(REWIND_MORE);
 379
 380         /*
 381          * Blocks delimited by our target token get REWIND_MORE.
 382          * Blocks delimiting our target token get REWIND_NONE.
 383          */
 384         switch (tok) {
 385         case (MDOC_Bl):
 386                 if (MDOC_It == p->tok)
 387                         return(REWIND_MORE);
 388                 break;
 389         case (MDOC_It):
 390                 if (MDOC_BODY == p->type && MDOC_Bl == p->tok)
 391                         return(REWIND_NONE);
 392                 break;
 393         /*
 394          * XXX Badly nested block handling still fails badly
 395          * when one block is breaking two blocks of the same type.
 396          * This is an incomplete and extremely ugly workaround,
 397          * required to let the OpenBSD tree build.
 398          */
 399         case (MDOC_Oo):
 400                 if (MDOC_Op == p->tok)
 401                         return(REWIND_MORE);
 402                 break;
 403         case (MDOC_Nm):
 404                 return(REWIND_NONE);
 405         case (MDOC_Nd):
 406                 /* FALLTHROUGH */
 407         case (MDOC_Ss):
 408                 if (MDOC_BODY == p->type && MDOC_Sh == p->tok)
 409                         return(REWIND_NONE);
 410                 /* FALLTHROUGH */
 411         case (MDOC_Sh):
 412                 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok ||
 413                     MDOC_Sh == p->tok)
 414                         return(REWIND_MORE);
 415                 break;
 416         default:
 417                 break;
 418         }
 419
 420         /*
 421          * Default block rewinding rules.
 422          * In particular, always skip block end markers,
 423          * and let all blocks rewind Nm children.
 424          */
 425         if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok ||
 426             (MDOC_BLOCK == p->type &&
 427             ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)))
 428                 return(REWIND_MORE);
 429
 430         /*
 431          * By default, closing out full blocks
 432          * forces closing of broken explicit blocks,
 433          * while closing out partial blocks
 434          * allows delayed rewinding by default.
 435          */
 436         return (&blk_full == mdoc_macros[tok].fp ?
 437             REWIND_FORCE : REWIND_LATER);
 438 }
 439
 440
 441 static int
 442 rew_elem(struct mdoc *mdoc, enum mdoct tok)
 443 {
 444         struct mdoc_node *n;
 445
 446         n = mdoc->last;
 447         if (MDOC_ELEM != n->type)
 448                 n = n->parent;
 449         assert(MDOC_ELEM == n->type);
 450         assert(tok == n->tok);
 451
 452         return(rew_last(mdoc, n));
 453 }
 454
 455
 456 /*
 457  * We are trying to close a block identified by tok,
 458  * but the child block *broken is still open.
 459  * Thus, postpone closing the tok block
 460  * until the rew_sub call closing *broken.
 461  */
 462 static int
 463 make_pending(struct mdoc_node *broken, enum mdoct tok,
 464                 struct mdoc *m, int line, int ppos)
 465 {
 466         struct mdoc_node *breaker;
 467
 468         /*
 469          * Iterate backwards, searching for the block matching tok,
 470          * that is, the block breaking the *broken block.
 471          */
 472         for (breaker = broken->parent; breaker; breaker = breaker->parent) {
 473
 474                 /*
 475                  * If the *broken block had already been broken before
 476                  * and we encounter its breaker, make the tok block
 477                  * pending on the inner breaker.
 478                  * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]"
 479                  * becomes "[A broken=[B [C->B B] tok=A] C]"
 480                  * and finally "[A [B->A [C->B B] A] C]".
 481                  */
 482                 if (breaker == broken->pending) {
 483                         broken = breaker;
 484                         continue;
 485                 }
 486
 487                 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker))
 488                         continue;
 489                 if (MDOC_BODY == broken->type)
 490                         broken = broken->parent;
 491
 492                 /*
 493                  * Found the breaker.
 494                  * If another, outer breaker is already pending on
 495                  * the *broken block, we must not clobber the link
 496                  * to the outer breaker, but make it pending on the
 497                  * new, now inner breaker.
 498                  * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]"
 499                  * becomes "[A breaker=[B->A broken=[C A] tok=B] C]"
 500                  * and finally "[A [B->A [C->B A] B] C]".
 501                  */
 502                 if (broken->pending) {
 503                         struct mdoc_node *taker;
 504
 505                         /*
 506                          * If the breaker had also been broken before,
 507                          * it cannot take on the outer breaker itself,
 508                          * but must hand it on to its own breakers.
 509                          * Graphically, this is the following situation:
 510                          * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]"
 511                          * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]"
 512                          */
 513                         taker = breaker;
 514                         while (taker->pending)
 515                                 taker = taker->pending;
 516                         taker->pending = broken->pending;
 517                 }
 518                 broken->pending = breaker;
 519                 mandoc_vmsg(MANDOCERR_SCOPENEST, m->parse, line, ppos,
 520                                 "%s breaks %s", mdoc_macronames[tok],
 521                                 mdoc_macronames[broken->tok]);
 522                 return(1);
 523         }
 524
 525         /*
 526          * Found no matching block for tok.
 527          * Are you trying to close a block that is not open?
 528          */
 529         return(0);
 530 }
 531
 532
 533 static int
 534 rew_sub(enum mdoc_type t, struct mdoc *m,
 535                 enum mdoct tok, int line, int ppos)
 536 {
 537         struct mdoc_node *n;
 538
 539         n = m->last;
 540         while (n) {
 541                 switch (rew_dohalt(tok, t, n)) {
 542                 case (REWIND_NONE):
 543                         return(1);
 544                 case (REWIND_THIS):
 545                         break;
 546                 case (REWIND_FORCE):
 547                         mandoc_vmsg(MANDOCERR_SCOPEBROKEN, m->parse,
 548                                         line, ppos, "%s breaks %s",
 549                                         mdoc_macronames[tok],
 550                                         mdoc_macronames[n->tok]);
 551                         /* FALLTHROUGH */
 552                 case (REWIND_MORE):
 553                         n = n->parent;
 554                         continue;
 555                 case (REWIND_LATER):
 556                         if (make_pending(n, tok, m, line, ppos) ||
 557                             MDOC_BLOCK != t)
 558                                 return(1);
 559                         /* FALLTHROUGH */
 560                 case (REWIND_ERROR):
 561                         mdoc_pmsg(m, line, ppos, MANDOCERR_NOSCOPE);
 562                         return(1);
 563                 }
 564                 break;
 565         }
 566
 567         assert(n);
 568         if ( ! rew_last(m, n))
 569                 return(0);
 570
 571         /*
 572          * The current block extends an enclosing block.
 573          * Now that the current block ends, close the enclosing block, too.
 574          */
 575         while (NULL != (n = n->pending)) {
 576                 if ( ! rew_last(m, n))
 577                         return(0);
 578                 if (MDOC_HEAD == n->type &&
 579                     ! mdoc_body_alloc(m, n->line, n->pos, n->tok))
 580                         return(0);
 581         }
 582
 583         return(1);
 584 }
 585
 586 /*
 587  * Allocate a word and check whether it's punctuation or not.
 588  * Punctuation consists of those tokens found in mdoc_isdelim().
 589  */
 590 static int
 591 dword(struct mdoc *m, int line,
 592                 int col, const char *p, enum mdelim d)
 593 {
 594
 595         if (DELIM_MAX == d)
 596                 d = mdoc_isdelim(p);
 597
 598         if ( ! mdoc_word_alloc(m, line, col, p))
 599                 return(0);
 600
 601         if (DELIM_OPEN == d)
 602                 m->last->flags |= MDOC_DELIMO;
 603         else if (DELIM_CLOSE == d)
 604                 m->last->flags |= MDOC_DELIMC;
 605
 606         return(1);
 607 }
 608
 609 static int
 610 append_delims(struct mdoc *m, int line, int *pos, char *buf)
 611 {
 612         int              la;
 613         enum margserr    ac;
 614         char            *p;
 615
 616         if ('\0' == buf[*pos])
 617                 return(1);
 618
 619         for (;;) {
 620                 la = *pos;
 621                 ac = mdoc_zargs(m, line, pos, buf, ARGS_NOWARN, &p);
 622
 623                 if (ARGS_ERROR == ac)
 624                         return(0);
 625                 else if (ARGS_EOLN == ac)
 626                         break;
 627
 628                 dword(m, line, la, p, DELIM_MAX);
 629
 630                 /*
 631                  * If we encounter end-of-sentence symbols, then trigger
 632                  * the double-space.
 633                  *
 634                  * XXX: it's easy to allow this to propogate outward to
 635                  * the last symbol, such that `. )' will cause the
 636                  * correct double-spacing.  However, (1) groff isn't
 637                  * smart enough to do this and (2) it would require
 638                  * knowing which symbols break this behaviour, for
 639                  * example, `.  ;' shouldn't propogate the double-space.
 640                  */
 641                 if (mandoc_eos(p, strlen(p), 0))
 642                         m->last->flags |= MDOC_EOS;
 643         }
 644
 645         return(1);
 646 }
 647
 648
 649 /*
 650  * Close out block partial/full explicit.
 651  */
 652 static int
 653 blk_exp_close(MACRO_PROT_ARGS)
 654 {
 655         struct mdoc_node *body;         /* Our own body. */
 656         struct mdoc_node *later;        /* A sub-block starting later. */
 657         struct mdoc_node *n;            /* For searching backwards. */
 658
 659         int              j, lastarg, maxargs, flushed, nl;
 660         enum margserr    ac;
 661         enum mdoct       atok, ntok;
 662         char            *p;
 663
 664         nl = MDOC_NEWLINE & m->flags;
 665
 666         switch (tok) {
 667         case (MDOC_Ec):
 668                 maxargs = 1;
 669                 break;
 670         default:
 671                 maxargs = 0;
 672                 break;
 673         }
 674
 675         /*
 676          * Search backwards for beginnings of blocks,
 677          * both of our own and of pending sub-blocks.
 678          */
 679         atok = rew_alt(tok);
 680         body = later = NULL;
 681         for (n = m->last; n; n = n->parent) {
 682                 if (MDOC_VALID & n->flags)
 683                         continue;
 684
 685                 /* Remember the start of our own body. */
 686                 if (MDOC_BODY == n->type && atok == n->tok) {
 687                         if (ENDBODY_NOT == n->end)
 688                                 body = n;
 689                         continue;
 690                 }
 691
 692                 if (MDOC_BLOCK != n->type || MDOC_Nm == n->tok)
 693                         continue;
 694                 if (atok == n->tok) {
 695                         assert(body);
 696
 697                         /*
 698                          * Found the start of our own block.
 699                          * When there is no pending sub block,
 700                          * just proceed to closing out.
 701                          */
 702                         if (NULL == later)
 703                                 break;
 704
 705                         /*
 706                          * When there is a pending sub block,
 707                          * postpone closing out the current block
 708                          * until the rew_sub() closing out the sub-block.
 709                          */
 710                         make_pending(later, tok, m, line, ppos);
 711
 712                         /*
 713                          * Mark the place where the formatting - but not
 714                          * the scope - of the current block ends.
 715                          */
 716                         if ( ! mdoc_endbody_alloc(m, line, ppos,
 717                             atok, body, ENDBODY_SPACE))
 718                                 return(0);
 719                         break;
 720                 }
 721
 722                 /*
 723                  * When finding an open sub block, remember the last
 724                  * open explicit block, or, in case there are only
 725                  * implicit ones, the first open implicit block.
 726                  */
 727                 if (later &&
 728                     MDOC_EXPLICIT & mdoc_macros[later->tok].flags)
 729                         continue;
 730                 if (MDOC_CALLABLE & mdoc_macros[n->tok].flags)
 731                         later = n;
 732         }
 733
 734         if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) {
 735                 /* FIXME: do this in validate */
 736                 if (buf[*pos])
 737                         mdoc_pmsg(m, line, ppos, MANDOCERR_ARGSLOST);
 738
 739                 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
 740                         return(0);
 741                 return(rew_sub(MDOC_BLOCK, m, tok, line, ppos));
 742         }
 743
 744         if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
 745                 return(0);
 746
 747         if (NULL == later && maxargs > 0)
 748                 if ( ! mdoc_tail_alloc(m, line, ppos, rew_alt(tok)))
 749                         return(0);
 750
 751         for (flushed = j = 0; ; j++) {
 752                 lastarg = *pos;
 753
 754                 if (j == maxargs && ! flushed) {
 755                         if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
 756                                 return(0);
 757                         flushed = 1;
 758                 }
 759
 760                 ac = mdoc_args(m, line, pos, buf, tok, &p);
 761
 762                 if (ARGS_ERROR == ac)
 763                         return(0);
 764                 if (ARGS_PUNCT == ac)
 765                         break;
 766                 if (ARGS_EOLN == ac)
 767                         break;
 768
 769                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
 770
 771                 if (MDOC_MAX == ntok) {
 772                         if ( ! dword(m, line, lastarg, p, DELIM_MAX))
 773                                 return(0);
 774                         continue;
 775                 }
 776
 777                 if ( ! flushed) {
 778                         if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
 779                                 return(0);
 780                         flushed = 1;
 781                 }
 782                 if ( ! mdoc_macro(m, ntok, line, lastarg, pos, buf))
 783                         return(0);
 784                 break;
 785         }
 786
 787         if ( ! flushed && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
 788                 return(0);
 789
 790         if ( ! nl)
 791                 return(1);
 792         return(append_delims(m, line, pos, buf));
 793 }
 794
 795
 796 static int
 797 in_line(MACRO_PROT_ARGS)
 798 {
 799         int              la, scope, cnt, nc, nl;
 800         enum margverr    av;
 801         enum mdoct       ntok;
 802         enum margserr    ac;
 803         enum mdelim      d;
 804         struct mdoc_arg *arg;
 805         char            *p;
 806
 807         nl = MDOC_NEWLINE & m->flags;
 808
 809         /*
 810          * Whether we allow ignored elements (those without content,
 811          * usually because of reserved words) to squeak by.
 812          */
 813
 814         switch (tok) {
 815         case (MDOC_An):
 816                 /* FALLTHROUGH */
 817         case (MDOC_Ar):
 818                 /* FALLTHROUGH */
 819         case (MDOC_Fl):
 820                 /* FALLTHROUGH */
 821         case (MDOC_Mt):
 822                 /* FALLTHROUGH */
 823         case (MDOC_Nm):
 824                 /* FALLTHROUGH */
 825         case (MDOC_Pa):
 826                 nc = 1;
 827                 break;
 828         default:
 829                 nc = 0;
 830                 break;
 831         }
 832
 833         for (arg = NULL;; ) {
 834                 la = *pos;
 835                 av = mdoc_argv(m, line, tok, &arg, pos, buf);
 836
 837                 if (ARGV_WORD == av) {
 838                         *pos = la;
 839                         break;
 840                 }
 841                 if (ARGV_EOLN == av)
 842                         break;
 843                 if (ARGV_ARG == av)
 844                         continue;
 845
 846                 mdoc_argv_free(arg);
 847                 return(0);
 848         }
 849
 850         for (cnt = scope = 0;; ) {
 851                 la = *pos;
 852                 ac = mdoc_args(m, line, pos, buf, tok, &p);
 853
 854                 if (ARGS_ERROR == ac)
 855                         return(0);
 856                 if (ARGS_EOLN == ac)
 857                         break;
 858                 if (ARGS_PUNCT == ac)
 859                         break;
 860
 861                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
 862
 863                 /*
 864                  * In this case, we've located a submacro and must
 865                  * execute it.  Close out scope, if open.  If no
 866                  * elements have been generated, either create one (nc)
 867                  * or raise a warning.
 868                  */
 869
 870                 if (MDOC_MAX != ntok) {
 871                         if (scope && ! rew_elem(m, tok))
 872                                 return(0);
 873                         if (nc && 0 == cnt) {
 874                                 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
 875                                         return(0);
 876                                 if ( ! rew_last(m, m->last))
 877                                         return(0);
 878                         } else if ( ! nc && 0 == cnt) {
 879                                 mdoc_argv_free(arg);
 880                                 mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY);
 881                         }
 882
 883                         if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
 884                                 return(0);
 885                         if ( ! nl)
 886                                 return(1);
 887                         return(append_delims(m, line, pos, buf));
 888                 }
 889
 890                 /*
 891                  * Non-quote-enclosed punctuation.  Set up our scope, if
 892                  * a word; rewind the scope, if a delimiter; then append
 893                  * the word.
 894                  */
 895
 896                 d = ARGS_QWORD == ac ? DELIM_NONE : mdoc_isdelim(p);
 897
 898                 if (DELIM_NONE != d) {
 899                         /*
 900                          * If we encounter closing punctuation, no word
 901                          * has been omitted, no scope is open, and we're
 902                          * allowed to have an empty element, then start
 903                          * a new scope.  `Ar', `Fl', and `Li', only do
 904                          * this once per invocation.  There may be more
 905                          * of these (all of them?).
 906                          */
 907                         if (0 == cnt && (nc || MDOC_Li == tok) &&
 908                                         DELIM_CLOSE == d && ! scope) {
 909                                 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
 910                                         return(0);
 911                                 if (MDOC_Ar == tok || MDOC_Li == tok ||
 912                                                 MDOC_Fl == tok)
 913                                         cnt++;
 914                                 scope = 1;
 915                         }
 916                         /*
 917                          * Close out our scope, if one is open, before
 918                          * any punctuation.
 919                          */
 920                         if (scope && ! rew_elem(m, tok))
 921                                 return(0);
 922                         scope = 0;
 923                 } else if ( ! scope) {
 924                         if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
 925                                 return(0);
 926                         scope = 1;
 927                 }
 928
 929                 if (DELIM_NONE == d)
 930                         cnt++;
 931
 932                 if ( ! dword(m, line, la, p, d))
 933                         return(0);
 934
 935                 /*
 936                  * `Fl' macros have their scope re-opened with each new
 937                  * word so that the `-' can be added to each one without
 938                  * having to parse out spaces.
 939                  */
 940                 if (scope && MDOC_Fl == tok) {
 941                         if ( ! rew_elem(m, tok))
 942                                 return(0);
 943                         scope = 0;
 944                 }
 945         }
 946
 947         if (scope && ! rew_elem(m, tok))
 948                 return(0);
 949
 950         /*
 951          * If no elements have been collected and we're allowed to have
 952          * empties (nc), open a scope and close it out.  Otherwise,
 953          * raise a warning.
 954          */
 955
 956         if (nc && 0 == cnt) {
 957                 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
 958                         return(0);
 959                 if ( ! rew_last(m, m->last))
 960                         return(0);
 961         } else if ( ! nc && 0 == cnt) {
 962                 mdoc_argv_free(arg);
 963                 mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY);
 964         }
 965
 966         if ( ! nl)
 967                 return(1);
 968         return(append_delims(m, line, pos, buf));
 969 }
 970
 971
 972 static int
 973 blk_full(MACRO_PROT_ARGS)
 974 {
 975         int               la, nl;
 976         struct mdoc_arg  *arg;
 977         struct mdoc_node *head; /* save of head macro */
 978         struct mdoc_node *body; /* save of body macro */
 979         struct mdoc_node *n;
 980         enum mdoc_type    mtt;
 981         enum mdoct        ntok;
 982         enum margserr     ac, lac;
 983         enum margverr     av;
 984         char             *p;
 985
 986         nl = MDOC_NEWLINE & m->flags;
 987
 988         /* Close out prior implicit scope. */
 989
 990         if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
 991                 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
 992                         return(0);
 993                 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
 994                         return(0);
 995         }
 996
 997         /*
 998          * This routine accomodates implicitly- and explicitly-scoped
 999          * macro openings.  Implicit ones first close out prior scope
1000          * (seen above).  Delay opening the head until necessary to
1001          * allow leading punctuation to print.  Special consideration
1002          * for `It -column', which has phrase-part syntax instead of
1003          * regular child nodes.
1004          */
1005
1006         for (arg = NULL;; ) {
1007                 la = *pos;
1008                 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1009
1010                 if (ARGV_WORD == av) {
1011                         *pos = la;
1012                         break;
1013                 }
1014
1015                 if (ARGV_EOLN == av)
1016                         break;
1017                 if (ARGV_ARG == av)
1018                         continue;
1019
1020                 mdoc_argv_free(arg);
1021                 return(0);
1022         }
1023
1024         if ( ! mdoc_block_alloc(m, line, ppos, tok, arg))
1025                 return(0);
1026
1027         head = body = NULL;
1028
1029         /*
1030          * The `Nd' macro has all arguments in its body: it's a hybrid
1031          * of block partial-explicit and full-implicit.  Stupid.
1032          */
1033
1034         if (MDOC_Nd == tok) {
1035                 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1036                         return(0);
1037                 head = m->last;
1038                 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1039                         return(0);
1040                 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1041                         return(0);
1042                 body = m->last;
1043         }
1044
1045         ac = ARGS_ERROR;
1046
1047         for ( ; ; ) {
1048                 la = *pos;
1049                 /* Initialise last-phrase-type with ARGS_PEND. */
1050                 lac = ARGS_ERROR == ac ? ARGS_PEND : ac;
1051                 ac = mdoc_args(m, line, pos, buf, tok, &p);
1052
1053                 if (ARGS_PUNCT == ac)
1054                         break;
1055
1056                 if (ARGS_ERROR == ac)
1057                         return(0);
1058
1059                 if (ARGS_EOLN == ac) {
1060                         if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac)
1061                                 break;
1062                         /*
1063                          * This is necessary: if the last token on a
1064                          * line is a `Ta' or tab, then we'll get
1065                          * ARGS_EOLN, so we must be smart enough to
1066                          * reopen our scope if the last parse was a
1067                          * phrase or partial phrase.
1068                          */
1069                         if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1070                                 return(0);
1071                         if ( ! mdoc_body_alloc(m, line, ppos, tok))
1072                                 return(0);
1073                         body = m->last;
1074                         break;
1075                 }
1076
1077                 /*
1078                  * Emit leading punctuation (i.e., punctuation before
1079                  * the MDOC_HEAD) for non-phrase types.
1080                  */
1081
1082                 if (NULL == head &&
1083                                 ARGS_PEND != ac &&
1084                                 ARGS_PHRASE != ac &&
1085                                 ARGS_PPHRASE != ac &&
1086                                 ARGS_QWORD != ac &&
1087                                 DELIM_OPEN == mdoc_isdelim(p)) {
1088                         if ( ! dword(m, line, la, p, DELIM_OPEN))
1089                                 return(0);
1090                         continue;
1091                 }
1092
1093                 /* Open a head if one hasn't been opened. */
1094
1095                 if (NULL == head) {
1096                         if ( ! mdoc_head_alloc(m, line, ppos, tok))
1097                                 return(0);
1098                         head = m->last;
1099                 }
1100
1101                 if (ARGS_PHRASE == ac ||
1102                                 ARGS_PEND == ac ||
1103                                 ARGS_PPHRASE == ac) {
1104                         /*
1105                          * If we haven't opened a body yet, rewind the
1106                          * head; if we have, rewind that instead.
1107                          */
1108
1109                         mtt = body ? MDOC_BODY : MDOC_HEAD;
1110                         if ( ! rew_sub(mtt, m, tok, line, ppos))
1111                                 return(0);
1112
1113                         /* Then allocate our body context. */
1114
1115                         if ( ! mdoc_body_alloc(m, line, ppos, tok))
1116                                 return(0);
1117                         body = m->last;
1118
1119                         /*
1120                          * Process phrases: set whether we're in a
1121                          * partial-phrase (this effects line handling)
1122                          * then call down into the phrase parser.
1123                          */
1124
1125                         if (ARGS_PPHRASE == ac)
1126                                 m->flags |= MDOC_PPHRASE;
1127                         if (ARGS_PEND == ac && ARGS_PPHRASE == lac)
1128                                 m->flags |= MDOC_PPHRASE;
1129
1130                         if ( ! phrase(m, line, la, buf))
1131                                 return(0);
1132
1133                         m->flags &= ~MDOC_PPHRASE;
1134                         continue;
1135                 }
1136
1137                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1138
1139                 if (MDOC_MAX == ntok) {
1140                         if ( ! dword(m, line, la, p, DELIM_MAX))
1141                                 return(0);
1142                         continue;
1143                 }
1144
1145                 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1146                         return(0);
1147                 break;
1148         }
1149
1150         if (NULL == head) {
1151                 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1152                         return(0);
1153                 head = m->last;
1154         }
1155
1156         if (nl && ! append_delims(m, line, pos, buf))
1157                 return(0);
1158
1159         /* If we've already opened our body, exit now. */
1160
1161         if (NULL != body)
1162                 goto out;
1163
1164         /*
1165          * If there is an open (i.e., unvalidated) sub-block requiring
1166          * explicit close-out, postpone switching the current block from
1167          * head to body until the rew_sub() call closing out that
1168          * sub-block.
1169          */
1170         for (n = m->last; n && n != head; n = n->parent) {
1171                 if (MDOC_BLOCK == n->type &&
1172                                 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1173                                 ! (MDOC_VALID & n->flags)) {
1174                         n->pending = head;
1175                         return(1);
1176                 }
1177         }
1178
1179         /* Close out scopes to remain in a consistent state. */
1180
1181         if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1182                 return(0);
1183         if ( ! mdoc_body_alloc(m, line, ppos, tok))
1184                 return(0);
1185
1186 out:
1187         if ( ! (MDOC_FREECOL & m->flags))
1188                 return(1);
1189
1190         if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1191                 return(0);
1192         if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1193                 return(0);
1194
1195         m->flags &= ~MDOC_FREECOL;
1196         return(1);
1197 }
1198
1199
1200 static int
1201 blk_part_imp(MACRO_PROT_ARGS)
1202 {
1203         int               la, nl;
1204         enum mdoct        ntok;
1205         enum margserr     ac;
1206         char             *p;
1207         struct mdoc_node *blk; /* saved block context */
1208         struct mdoc_node *body; /* saved body context */
1209         struct mdoc_node *n;
1210
1211         nl = MDOC_NEWLINE & m->flags;
1212
1213         /*
1214          * A macro that spans to the end of the line.  This is generally
1215          * (but not necessarily) called as the first macro.  The block
1216          * has a head as the immediate child, which is always empty,
1217          * followed by zero or more opening punctuation nodes, then the
1218          * body (which may be empty, depending on the macro), then zero
1219          * or more closing punctuation nodes.
1220          */
1221
1222         if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1223                 return(0);
1224
1225         blk = m->last;
1226
1227         if ( ! mdoc_head_alloc(m, line, ppos, tok))
1228                 return(0);
1229         if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1230                 return(0);
1231
1232         /*
1233          * Open the body scope "on-demand", that is, after we've
1234          * processed all our the leading delimiters (open parenthesis,
1235          * etc.).
1236          */
1237
1238         for (body = NULL; ; ) {
1239                 la = *pos;
1240                 ac = mdoc_args(m, line, pos, buf, tok, &p);
1241
1242                 if (ARGS_ERROR == ac)
1243                         return(0);
1244                 if (ARGS_EOLN == ac)
1245                         break;
1246                 if (ARGS_PUNCT == ac)
1247                         break;
1248
1249                 if (NULL == body && ARGS_QWORD != ac &&
1250                                 DELIM_OPEN == mdoc_isdelim(p)) {
1251                         if ( ! dword(m, line, la, p, DELIM_OPEN))
1252                                 return(0);
1253                         continue;
1254                 }
1255
1256                 if (NULL == body) {
1257                        if ( ! mdoc_body_alloc(m, line, ppos, tok))
1258                                return(0);
1259                         body = m->last;
1260                 }
1261
1262                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1263
1264                 if (MDOC_MAX == ntok) {
1265                         if ( ! dword(m, line, la, p, DELIM_MAX))
1266                                 return(0);
1267                         continue;
1268                 }
1269
1270                 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1271                         return(0);
1272                 break;
1273         }
1274
1275         /* Clean-ups to leave in a consistent state. */
1276
1277         if (NULL == body) {
1278                 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1279                         return(0);
1280                 body = m->last;
1281         }
1282
1283         for (n = body->child; n && n->next; n = n->next)
1284                 /* Do nothing. */ ;
1285
1286         /*
1287          * End of sentence spacing: if the last node is a text node and
1288          * has a trailing period, then mark it as being end-of-sentence.
1289          */
1290
1291         if (n && MDOC_TEXT == n->type && n->string)
1292                 if (mandoc_eos(n->string, strlen(n->string), 1))
1293                         n->flags |= MDOC_EOS;
1294
1295         /* Up-propogate the end-of-space flag. */
1296
1297         if (n && (MDOC_EOS & n->flags)) {
1298                 body->flags |= MDOC_EOS;
1299                 body->parent->flags |= MDOC_EOS;
1300         }
1301
1302         /*
1303          * If there is an open sub-block requiring explicit close-out,
1304          * postpone closing out the current block
1305          * until the rew_sub() call closing out the sub-block.
1306          */
1307         for (n = m->last; n && n != body && n != blk->parent; n = n->parent) {
1308                 if (MDOC_BLOCK == n->type &&
1309                     MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1310                     ! (MDOC_VALID & n->flags)) {
1311                         make_pending(n, tok, m, line, ppos);
1312                         if ( ! mdoc_endbody_alloc(m, line, ppos,
1313                             tok, body, ENDBODY_NOSPACE))
1314                                 return(0);
1315                         return(1);
1316                 }
1317         }
1318
1319         /*
1320          * If we can't rewind to our body, then our scope has already
1321          * been closed by another macro (like `Oc' closing `Op').  This
1322          * is ugly behaviour nodding its head to OpenBSD's overwhelming
1323          * crufty use of `Op' breakage.
1324          */
1325         if (n != body)
1326                 mandoc_vmsg(MANDOCERR_SCOPENEST, m->parse, line, ppos,
1327                                 "%s broken", mdoc_macronames[tok]);
1328
1329         if (n && ! rew_sub(MDOC_BODY, m, tok, line, ppos))
1330                 return(0);
1331
1332         /* Standard appending of delimiters. */
1333
1334         if (nl && ! append_delims(m, line, pos, buf))
1335                 return(0);
1336
1337         /* Rewind scope, if applicable. */
1338
1339         if (n && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos))
1340                 return(0);
1341
1342         return(1);
1343 }
1344
1345
1346 static int
1347 blk_part_exp(MACRO_PROT_ARGS)
1348 {
1349         int               la, nl;
1350         enum margserr     ac;
1351         struct mdoc_node *head; /* keep track of head */
1352         struct mdoc_node *body; /* keep track of body */
1353         char             *p;
1354         enum mdoct        ntok;
1355
1356         nl = MDOC_NEWLINE & m->flags;
1357
1358         /*
1359          * The opening of an explicit macro having zero or more leading
1360          * punctuation nodes; a head with optional single element (the
1361          * case of `Eo'); and a body that may be empty.
1362          */
1363
1364         if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL))
1365                 return(0);
1366
1367         for (head = body = NULL; ; ) {
1368                 la = *pos;
1369                 ac = mdoc_args(m, line, pos, buf, tok, &p);
1370
1371                 if (ARGS_ERROR == ac)
1372                         return(0);
1373                 if (ARGS_PUNCT == ac)
1374                         break;
1375                 if (ARGS_EOLN == ac)
1376                         break;
1377
1378                 /* Flush out leading punctuation. */
1379
1380                 if (NULL == head && ARGS_QWORD != ac &&
1381                                 DELIM_OPEN == mdoc_isdelim(p)) {
1382                         assert(NULL == body);
1383                         if ( ! dword(m, line, la, p, DELIM_OPEN))
1384                                 return(0);
1385                         continue;
1386                 }
1387
1388                 if (NULL == head) {
1389                         assert(NULL == body);
1390                         if ( ! mdoc_head_alloc(m, line, ppos, tok))
1391                                 return(0);
1392                         head = m->last;
1393                 }
1394
1395                 /*
1396                  * `Eo' gobbles any data into the head, but most other
1397                  * macros just immediately close out and begin the body.
1398                  */
1399
1400                 if (NULL == body) {
1401                         assert(head);
1402                         /* No check whether it's a macro! */
1403                         if (MDOC_Eo == tok)
1404                                 if ( ! dword(m, line, la, p, DELIM_MAX))
1405                                         return(0);
1406
1407                         if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1408                                 return(0);
1409                         if ( ! mdoc_body_alloc(m, line, ppos, tok))
1410                                 return(0);
1411                         body = m->last;
1412
1413                         if (MDOC_Eo == tok)
1414                                 continue;
1415                 }
1416
1417                 assert(NULL != head && NULL != body);
1418
1419                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1420
1421                 if (MDOC_MAX == ntok) {
1422                         if ( ! dword(m, line, la, p, DELIM_MAX))
1423                                 return(0);
1424                         continue;
1425                 }
1426
1427                 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1428                         return(0);
1429                 break;
1430         }
1431
1432         /* Clean-up to leave in a consistent state. */
1433
1434         if (NULL == head) {
1435                 if ( ! mdoc_head_alloc(m, line, ppos, tok))
1436                         return(0);
1437                 head = m->last;
1438         }
1439
1440         if (NULL == body) {
1441                 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos))
1442                         return(0);
1443                 if ( ! mdoc_body_alloc(m, line, ppos, tok))
1444                         return(0);
1445                 body = m->last;
1446         }
1447
1448         /* Standard appending of delimiters. */
1449
1450         if ( ! nl)
1451                 return(1);
1452         return(append_delims(m, line, pos, buf));
1453 }
1454
1455
1456 /* ARGSUSED */
1457 static int
1458 in_line_argn(MACRO_PROT_ARGS)
1459 {
1460         int              la, flushed, j, maxargs, nl;
1461         enum margserr    ac;
1462         enum margverr    av;
1463         struct mdoc_arg *arg;
1464         char            *p;
1465         enum mdoct       ntok;
1466
1467         nl = MDOC_NEWLINE & m->flags;
1468
1469         /*
1470          * A line macro that has a fixed number of arguments (maxargs).
1471          * Only open the scope once the first non-leading-punctuation is
1472          * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1473          * keep it open until the maximum number of arguments are
1474          * exhausted.
1475          */
1476
1477         switch (tok) {
1478         case (MDOC_Ap):
1479                 /* FALLTHROUGH */
1480         case (MDOC_No):
1481                 /* FALLTHROUGH */
1482         case (MDOC_Ns):
1483                 /* FALLTHROUGH */
1484         case (MDOC_Ux):
1485                 maxargs = 0;
1486                 break;
1487         case (MDOC_Bx):
1488                 /* FALLTHROUGH */
1489         case (MDOC_Xr):
1490                 maxargs = 2;
1491                 break;
1492         default:
1493                 maxargs = 1;
1494                 break;
1495         }
1496
1497         for (arg = NULL; ; ) {
1498                 la = *pos;
1499                 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1500
1501                 if (ARGV_WORD == av) {
1502                         *pos = la;
1503                         break;
1504                 }
1505
1506                 if (ARGV_EOLN == av)
1507                         break;
1508                 if (ARGV_ARG == av)
1509                         continue;
1510
1511                 mdoc_argv_free(arg);
1512                 return(0);
1513         }
1514
1515         for (flushed = j = 0; ; ) {
1516                 la = *pos;
1517                 ac = mdoc_args(m, line, pos, buf, tok, &p);
1518
1519                 if (ARGS_ERROR == ac)
1520                         return(0);
1521                 if (ARGS_PUNCT == ac)
1522                         break;
1523                 if (ARGS_EOLN == ac)
1524                         break;
1525
1526                 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1527                                 ARGS_QWORD != ac && 0 == j &&
1528                                 DELIM_OPEN == mdoc_isdelim(p)) {
1529                         if ( ! dword(m, line, la, p, DELIM_OPEN))
1530                                 return(0);
1531                         continue;
1532                 } else if (0 == j)
1533                        if ( ! mdoc_elem_alloc(m, line, la, tok, arg))
1534                                return(0);
1535
1536                 if (j == maxargs && ! flushed) {
1537                         if ( ! rew_elem(m, tok))
1538                                 return(0);
1539                         flushed = 1;
1540                 }
1541
1542                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1543
1544                 if (MDOC_MAX != ntok) {
1545                         if ( ! flushed && ! rew_elem(m, tok))
1546                                 return(0);
1547                         flushed = 1;
1548                         if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1549                                 return(0);
1550                         j++;
1551                         break;
1552                 }
1553
1554                 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1555                                 ARGS_QWORD != ac &&
1556                                 ! flushed &&
1557                                 DELIM_NONE != mdoc_isdelim(p)) {
1558                         if ( ! rew_elem(m, tok))
1559                                 return(0);
1560                         flushed = 1;
1561                 }
1562
1563                 /*
1564                  * XXX: this is a hack to work around groff's ugliness
1565                  * as regards `Xr' and extraneous arguments.  It should
1566                  * ideally be deprecated behaviour, but because this is
1567                  * code is no here, it's unlikely to be removed.
1568                  */
1569
1570 #ifdef __OpenBSD__
1571                 if (MDOC_Xr == tok && j == maxargs) {
1572                         if ( ! mdoc_elem_alloc(m, line, la, MDOC_Ns, NULL))
1573                                 return(0);
1574                         if ( ! rew_elem(m, MDOC_Ns))
1575                                 return(0);
1576                 }
1577 #endif
1578
1579                 if ( ! dword(m, line, la, p, DELIM_MAX))
1580                         return(0);
1581                 j++;
1582         }
1583
1584         if (0 == j && ! mdoc_elem_alloc(m, line, la, tok, arg))
1585                return(0);
1586
1587         /* Close out in a consistent state. */
1588
1589         if ( ! flushed && ! rew_elem(m, tok))
1590                 return(0);
1591         if ( ! nl)
1592                 return(1);
1593         return(append_delims(m, line, pos, buf));
1594 }
1595
1596
1597 static int
1598 in_line_eoln(MACRO_PROT_ARGS)
1599 {
1600         int              la;
1601         enum margserr    ac;
1602         enum margverr    av;
1603         struct mdoc_arg *arg;
1604         char            *p;
1605         enum mdoct       ntok;
1606
1607         assert( ! (MDOC_PARSED & mdoc_macros[tok].flags));
1608
1609         if (tok == MDOC_Pp)
1610                 rew_sub(MDOC_BLOCK, m, MDOC_Nm, line, ppos);
1611
1612         /* Parse macro arguments. */
1613
1614         for (arg = NULL; ; ) {
1615                 la = *pos;
1616                 av = mdoc_argv(m, line, tok, &arg, pos, buf);
1617
1618                 if (ARGV_WORD == av) {
1619                         *pos = la;
1620                         break;
1621                 }
1622                 if (ARGV_EOLN == av)
1623                         break;
1624                 if (ARGV_ARG == av)
1625                         continue;
1626
1627                 mdoc_argv_free(arg);
1628                 return(0);
1629         }
1630
1631         /* Open element scope. */
1632
1633         if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg))
1634                 return(0);
1635
1636         /* Parse argument terms. */
1637
1638         for (;;) {
1639                 la = *pos;
1640                 ac = mdoc_args(m, line, pos, buf, tok, &p);
1641
1642                 if (ARGS_ERROR == ac)
1643                         return(0);
1644                 if (ARGS_EOLN == ac)
1645                         break;
1646
1647                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1648
1649                 if (MDOC_MAX == ntok) {
1650                         if ( ! dword(m, line, la, p, DELIM_MAX))
1651                                 return(0);
1652                         continue;
1653                 }
1654
1655                 if ( ! rew_elem(m, tok))
1656                         return(0);
1657                 return(mdoc_macro(m, ntok, line, la, pos, buf));
1658         }
1659
1660         /* Close out (no delimiters). */
1661
1662         return(rew_elem(m, tok));
1663 }
1664
1665
1666 /* ARGSUSED */
1667 static int
1668 ctx_synopsis(MACRO_PROT_ARGS)
1669 {
1670         int              nl;
1671
1672         nl = MDOC_NEWLINE & m->flags;
1673
1674         /* If we're not in the SYNOPSIS, go straight to in-line. */
1675         if ( ! (MDOC_SYNOPSIS & m->flags))
1676                 return(in_line(m, tok, line, ppos, pos, buf));
1677
1678         /* If we're a nested call, same place. */
1679         if ( ! nl)
1680                 return(in_line(m, tok, line, ppos, pos, buf));
1681
1682         /*
1683          * XXX: this will open a block scope; however, if later we end
1684          * up formatting the block scope, then child nodes will inherit
1685          * the formatting.  Be careful.
1686          */
1687         if (MDOC_Nm == tok)
1688                 return(blk_full(m, tok, line, ppos, pos, buf));
1689         assert(MDOC_Vt == tok);
1690         return(blk_part_imp(m, tok, line, ppos, pos, buf));
1691 }
1692
1693
1694 /* ARGSUSED */
1695 static int
1696 obsolete(MACRO_PROT_ARGS)
1697 {
1698
1699         mdoc_pmsg(m, line, ppos, MANDOCERR_MACROOBS);
1700         return(1);
1701 }
1702
1703
1704 /*
1705  * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1706  * They're unusual because they're basically free-form text until a
1707  * macro is encountered.
1708  */
1709 static int
1710 phrase(struct mdoc *m, int line, int ppos, char *buf)
1711 {
1712         int              la, pos;
1713         enum margserr    ac;
1714         enum mdoct       ntok;
1715         char            *p;
1716
1717         for (pos = ppos; ; ) {
1718                 la = pos;
1719
1720                 ac = mdoc_zargs(m, line, &pos, buf, 0, &p);
1721
1722                 if (ARGS_ERROR == ac)
1723                         return(0);
1724                 if (ARGS_EOLN == ac)
1725                         break;
1726
1727                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1728
1729                 if (MDOC_MAX == ntok) {
1730                         if ( ! dword(m, line, la, p, DELIM_MAX))
1731                                 return(0);
1732                         continue;
1733                 }
1734
1735                 if ( ! mdoc_macro(m, ntok, line, la, &pos, buf))
1736                         return(0);
1737                 return(append_delims(m, line, &pos, buf));
1738         }
1739
1740         return(1);
1741 }
1742
1743
1744 /* ARGSUSED */
1745 static int
1746 phrase_ta(MACRO_PROT_ARGS)
1747 {
1748         int               la;
1749         enum mdoct        ntok;
1750         enum margserr     ac;
1751         char             *p;
1752
1753         /*
1754          * FIXME: this is overly restrictive: if the `Ta' is unexpected,
1755          * it should simply error out with ARGSLOST.
1756          */
1757
1758         if ( ! rew_sub(MDOC_BODY, m, MDOC_It, line, ppos))
1759                 return(0);
1760         if ( ! mdoc_body_alloc(m, line, ppos, MDOC_It))
1761                 return(0);
1762
1763         for (;;) {
1764                 la = *pos;
1765                 ac = mdoc_zargs(m, line, pos, buf, 0, &p);
1766
1767                 if (ARGS_ERROR == ac)
1768                         return(0);
1769                 if (ARGS_EOLN == ac)
1770                         break;
1771
1772                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1773
1774                 if (MDOC_MAX == ntok) {
1775                         if ( ! dword(m, line, la, p, DELIM_MAX))
1776                                 return(0);
1777                         continue;
1778                 }
1779
1780                 if ( ! mdoc_macro(m, ntok, line, la, pos, buf))
1781                         return(0);
1782                 return(append_delims(m, line, pos, buf));
1783         }
1784
1785         return(1);
1786 }