mdoc_macro.c

   1 /*      $Id: mdoc_macro.c,v 1.122 2013/09/15 18:26:46 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010, 2012 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, MDOC_PARSED }, /* Sh */
  78         { blk_full, MDOC_PARSED }, /* 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 *mdoc)
 207 {
 208         struct mdoc_node *n;
 209
 210         /* Scan for open explicit scopes. */
 211
 212         n = MDOC_VALID & mdoc->last->flags ?
 213                         mdoc->last->parent : mdoc->last;
 214
 215         for ( ; n; n = n->parent)
 216                 if (MDOC_BLOCK == n->type &&
 217                     MDOC_EXPLICIT & mdoc_macros[n->tok].flags)
 218                         mdoc_nmsg(mdoc, n, MANDOCERR_SCOPEEXIT);
 219
 220         /* Rewind to the first. */
 221
 222         return(rew_last(mdoc, mdoc->first));
 223 }
 224
 225
 226 /*
 227  * Look up a macro from within a subsequent context.
 228  */
 229 static enum mdoct
 230 lookup(enum mdoct from, const char *p)
 231 {
 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                  * mdoc->last node in the post-validation phase and reset
 268                  * it to mdoc->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 *mdoc, 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, mdoc->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 *mdoc,
 535                 enum mdoct tok, int line, int ppos)
 536 {
 537         struct mdoc_node *n;
 538
 539         n = mdoc->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, mdoc->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, mdoc, line, ppos) ||
 557                             MDOC_BLOCK != t)
 558                                 return(1);
 559                         /* FALLTHROUGH */
 560                 case (REWIND_ERROR):
 561                         mdoc_pmsg(mdoc, line, ppos, MANDOCERR_NOSCOPE);
 562                         return(1);
 563                 }
 564                 break;
 565         }
 566
 567         assert(n);
 568         if ( ! rew_last(mdoc, 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(mdoc, n))
 577                         return(0);
 578                 if (MDOC_HEAD == n->type &&
 579                     ! mdoc_body_alloc(mdoc, 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 *mdoc, 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(mdoc, line, col, p))
 599                 return(0);
 600
 601         if (DELIM_OPEN == d)
 602                 mdoc->last->flags |= MDOC_DELIMO;
 603
 604         /*
 605          * Closing delimiters only suppress the preceding space
 606          * when they follow something, not when they start a new
 607          * block or element, and not when they follow `No'.
 608          *
 609          * XXX  Explicitly special-casing MDOC_No here feels
 610          *      like a layering violation.  Find a better way
 611          *      and solve this in the code related to `No'!
 612          */
 613
 614         else if (DELIM_CLOSE == d && mdoc->last->prev &&
 615                         mdoc->last->prev->tok != MDOC_No &&
 616                         mdoc->last->parent->tok != MDOC_Fd)
 617                 mdoc->last->flags |= MDOC_DELIMC;
 618
 619         return(1);
 620 }
 621
 622 static int
 623 append_delims(struct mdoc *mdoc, int line, int *pos, char *buf)
 624 {
 625         int              la;
 626         enum margserr    ac;
 627         char            *p;
 628
 629         if ('\0' == buf[*pos])
 630                 return(1);
 631
 632         for (;;) {
 633                 la = *pos;
 634                 ac = mdoc_zargs(mdoc, line, pos, buf, &p);
 635
 636                 if (ARGS_ERROR == ac)
 637                         return(0);
 638                 else if (ARGS_EOLN == ac)
 639                         break;
 640
 641                 dword(mdoc, line, la, p, DELIM_MAX);
 642
 643                 /*
 644                  * If we encounter end-of-sentence symbols, then trigger
 645                  * the double-space.
 646                  *
 647                  * XXX: it's easy to allow this to propagate outward to
 648                  * the last symbol, such that `. )' will cause the
 649                  * correct double-spacing.  However, (1) groff isn't
 650                  * smart enough to do this and (2) it would require
 651                  * knowing which symbols break this behaviour, for
 652                  * example, `.  ;' shouldn't propagate the double-space.
 653                  */
 654                 if (mandoc_eos(p, strlen(p), 0))
 655                         mdoc->last->flags |= MDOC_EOS;
 656         }
 657
 658         return(1);
 659 }
 660
 661
 662 /*
 663  * Close out block partial/full explicit.
 664  */
 665 static int
 666 blk_exp_close(MACRO_PROT_ARGS)
 667 {
 668         struct mdoc_node *body;         /* Our own body. */
 669         struct mdoc_node *later;        /* A sub-block starting later. */
 670         struct mdoc_node *n;            /* For searching backwards. */
 671
 672         int              j, lastarg, maxargs, flushed, nl;
 673         enum margserr    ac;
 674         enum mdoct       atok, ntok;
 675         char            *p;
 676
 677         nl = MDOC_NEWLINE & mdoc->flags;
 678
 679         switch (tok) {
 680         case (MDOC_Ec):
 681                 maxargs = 1;
 682                 break;
 683         default:
 684                 maxargs = 0;
 685                 break;
 686         }
 687
 688         /*
 689          * Search backwards for beginnings of blocks,
 690          * both of our own and of pending sub-blocks.
 691          */
 692         atok = rew_alt(tok);
 693         body = later = NULL;
 694         for (n = mdoc->last; n; n = n->parent) {
 695                 if (MDOC_VALID & n->flags)
 696                         continue;
 697
 698                 /* Remember the start of our own body. */
 699                 if (MDOC_BODY == n->type && atok == n->tok) {
 700                         if (ENDBODY_NOT == n->end)
 701                                 body = n;
 702                         continue;
 703                 }
 704
 705                 if (MDOC_BLOCK != n->type || MDOC_Nm == n->tok)
 706                         continue;
 707                 if (atok == n->tok) {
 708                         assert(body);
 709
 710                         /*
 711                          * Found the start of our own block.
 712                          * When there is no pending sub block,
 713                          * just proceed to closing out.
 714                          */
 715                         if (NULL == later)
 716                                 break;
 717
 718                         /*
 719                          * When there is a pending sub block,
 720                          * postpone closing out the current block
 721                          * until the rew_sub() closing out the sub-block.
 722                          */
 723                         make_pending(later, tok, mdoc, line, ppos);
 724
 725                         /*
 726                          * Mark the place where the formatting - but not
 727                          * the scope - of the current block ends.
 728                          */
 729                         if ( ! mdoc_endbody_alloc(mdoc, line, ppos,
 730                             atok, body, ENDBODY_SPACE))
 731                                 return(0);
 732                         break;
 733                 }
 734
 735                 /*
 736                  * When finding an open sub block, remember the last
 737                  * open explicit block, or, in case there are only
 738                  * implicit ones, the first open implicit block.
 739                  */
 740                 if (later &&
 741                     MDOC_EXPLICIT & mdoc_macros[later->tok].flags)
 742                         continue;
 743                 if (MDOC_It != n->tok)
 744                         later = n;
 745         }
 746
 747         if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) {
 748                 /* FIXME: do this in validate */
 749                 if (buf[*pos])
 750                         mdoc_pmsg(mdoc, line, ppos, MANDOCERR_ARGSLOST);
 751
 752                 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
 753                         return(0);
 754                 return(rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos));
 755         }
 756
 757         if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
 758                 return(0);
 759
 760         if (NULL == later && maxargs > 0)
 761                 if ( ! mdoc_tail_alloc(mdoc, line, ppos, rew_alt(tok)))
 762                         return(0);
 763
 764         for (flushed = j = 0; ; j++) {
 765                 lastarg = *pos;
 766
 767                 if (j == maxargs && ! flushed) {
 768                         if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
 769                                 return(0);
 770                         flushed = 1;
 771                 }
 772
 773                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
 774
 775                 if (ARGS_ERROR == ac)
 776                         return(0);
 777                 if (ARGS_PUNCT == ac)
 778                         break;
 779                 if (ARGS_EOLN == ac)
 780                         break;
 781
 782                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
 783
 784                 if (MDOC_MAX == ntok) {
 785                         if ( ! dword(mdoc, line, lastarg, p, DELIM_MAX))
 786                                 return(0);
 787                         continue;
 788                 }
 789
 790                 if ( ! flushed) {
 791                         if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
 792                                 return(0);
 793                         flushed = 1;
 794                 }
 795
 796                 mdoc->flags &= ~MDOC_NEWLINE;
 797
 798                 if ( ! mdoc_macro(mdoc, ntok, line, lastarg, pos, buf))
 799                         return(0);
 800                 break;
 801         }
 802
 803         if ( ! flushed && ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
 804                 return(0);
 805
 806         if ( ! nl)
 807                 return(1);
 808         return(append_delims(mdoc, line, pos, buf));
 809 }
 810
 811
 812 static int
 813 in_line(MACRO_PROT_ARGS)
 814 {
 815         int              la, scope, cnt, nc, nl;
 816         enum margverr    av;
 817         enum mdoct       ntok;
 818         enum margserr    ac;
 819         enum mdelim      d;
 820         struct mdoc_arg *arg;
 821         char            *p;
 822
 823         nl = MDOC_NEWLINE & mdoc->flags;
 824
 825         /*
 826          * Whether we allow ignored elements (those without content,
 827          * usually because of reserved words) to squeak by.
 828          */
 829
 830         switch (tok) {
 831         case (MDOC_An):
 832                 /* FALLTHROUGH */
 833         case (MDOC_Ar):
 834                 /* FALLTHROUGH */
 835         case (MDOC_Fl):
 836                 /* FALLTHROUGH */
 837         case (MDOC_Mt):
 838                 /* FALLTHROUGH */
 839         case (MDOC_Nm):
 840                 /* FALLTHROUGH */
 841         case (MDOC_Pa):
 842                 nc = 1;
 843                 break;
 844         default:
 845                 nc = 0;
 846                 break;
 847         }
 848
 849         for (arg = NULL;; ) {
 850                 la = *pos;
 851                 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
 852
 853                 if (ARGV_WORD == av) {
 854                         *pos = la;
 855                         break;
 856                 }
 857                 if (ARGV_EOLN == av)
 858                         break;
 859                 if (ARGV_ARG == av)
 860                         continue;
 861
 862                 mdoc_argv_free(arg);
 863                 return(0);
 864         }
 865
 866         for (cnt = scope = 0;; ) {
 867                 la = *pos;
 868                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
 869
 870                 if (ARGS_ERROR == ac)
 871                         return(0);
 872                 if (ARGS_EOLN == ac)
 873                         break;
 874                 if (ARGS_PUNCT == ac)
 875                         break;
 876
 877                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
 878
 879                 /*
 880                  * In this case, we've located a submacro and must
 881                  * execute it.  Close out scope, if open.  If no
 882                  * elements have been generated, either create one (nc)
 883                  * or raise a warning.
 884                  */
 885
 886                 if (MDOC_MAX != ntok) {
 887                         if (scope && ! rew_elem(mdoc, tok))
 888                                 return(0);
 889                         if (nc && 0 == cnt) {
 890                                 if ( ! mdoc_elem_alloc(mdoc, line,
 891                                                 ppos, tok, arg))
 892                                         return(0);
 893                                 if ( ! rew_last(mdoc, mdoc->last))
 894                                         return(0);
 895                         } else if ( ! nc && 0 == cnt) {
 896                                 mdoc_argv_free(arg);
 897                                 mdoc_pmsg(mdoc, line, ppos,
 898                                         MANDOCERR_MACROEMPTY);
 899                         }
 900
 901                         if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
 902                                 return(0);
 903                         if ( ! nl)
 904                                 return(1);
 905                         return(append_delims(mdoc, line, pos, buf));
 906                 }
 907
 908                 /*
 909                  * Non-quote-enclosed punctuation.  Set up our scope, if
 910                  * a word; rewind the scope, if a delimiter; then append
 911                  * the word.
 912                  */
 913
 914                 d = ARGS_QWORD == ac ? DELIM_NONE : mdoc_isdelim(p);
 915
 916                 if (DELIM_NONE != d) {
 917                         /*
 918                          * If we encounter closing punctuation, no word
 919                          * has been omitted, no scope is open, and we're
 920                          * allowed to have an empty element, then start
 921                          * a new scope.  `Ar', `Fl', and `Li', only do
 922                          * this once per invocation.  There may be more
 923                          * of these (all of them?).
 924                          */
 925                         if (0 == cnt && (nc || MDOC_Li == tok) &&
 926                                         DELIM_CLOSE == d && ! scope) {
 927                                 if ( ! mdoc_elem_alloc(mdoc, line,
 928                                                 ppos, tok, arg))
 929                                         return(0);
 930                                 if (MDOC_Ar == tok || MDOC_Li == tok ||
 931                                                 MDOC_Fl == tok)
 932                                         cnt++;
 933                                 scope = 1;
 934                         }
 935                         /*
 936                          * Close out our scope, if one is open, before
 937                          * any punctuation.
 938                          */
 939                         if (scope && ! rew_elem(mdoc, tok))
 940                                 return(0);
 941                         scope = 0;
 942                 } else if ( ! scope) {
 943                         if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg))
 944                                 return(0);
 945                         scope = 1;
 946                 }
 947
 948                 if (DELIM_NONE == d)
 949                         cnt++;
 950
 951                 if ( ! dword(mdoc, line, la, p, d))
 952                         return(0);
 953
 954                 /*
 955                  * `Fl' macros have their scope re-opened with each new
 956                  * word so that the `-' can be added to each one without
 957                  * having to parse out spaces.
 958                  */
 959                 if (scope && MDOC_Fl == tok) {
 960                         if ( ! rew_elem(mdoc, tok))
 961                                 return(0);
 962                         scope = 0;
 963                 }
 964         }
 965
 966         if (scope && ! rew_elem(mdoc, tok))
 967                 return(0);
 968
 969         /*
 970          * If no elements have been collected and we're allowed to have
 971          * empties (nc), open a scope and close it out.  Otherwise,
 972          * raise a warning.
 973          */
 974
 975         if (nc && 0 == cnt) {
 976                 if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg))
 977                         return(0);
 978                 if ( ! rew_last(mdoc, mdoc->last))
 979                         return(0);
 980         } else if ( ! nc && 0 == cnt) {
 981                 mdoc_argv_free(arg);
 982                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_MACROEMPTY);
 983         }
 984
 985         if ( ! nl)
 986                 return(1);
 987         return(append_delims(mdoc, line, pos, buf));
 988 }
 989
 990
 991 static int
 992 blk_full(MACRO_PROT_ARGS)
 993 {
 994         int               la, nl, nparsed;
 995         struct mdoc_arg  *arg;
 996         struct mdoc_node *head; /* save of head macro */
 997         struct mdoc_node *body; /* save of body macro */
 998         struct mdoc_node *n;
 999         enum mdoc_type    mtt;
1000         enum mdoct        ntok;
1001         enum margserr     ac, lac;
1002         enum margverr     av;
1003         char             *p;
1004
1005         nl = MDOC_NEWLINE & mdoc->flags;
1006
1007         /* Close out prior implicit scope. */
1008
1009         if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
1010                 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
1011                         return(0);
1012                 if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
1013                         return(0);
1014         }
1015
1016         /*
1017          * This routine accommodates implicitly- and explicitly-scoped
1018          * macro openings.  Implicit ones first close out prior scope
1019          * (seen above).  Delay opening the head until necessary to
1020          * allow leading punctuation to print.  Special consideration
1021          * for `It -column', which has phrase-part syntax instead of
1022          * regular child nodes.
1023          */
1024
1025         for (arg = NULL;; ) {
1026                 la = *pos;
1027                 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1028
1029                 if (ARGV_WORD == av) {
1030                         *pos = la;
1031                         break;
1032                 }
1033
1034                 if (ARGV_EOLN == av)
1035                         break;
1036                 if (ARGV_ARG == av)
1037                         continue;
1038
1039                 mdoc_argv_free(arg);
1040                 return(0);
1041         }
1042
1043         if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, arg))
1044                 return(0);
1045
1046         head = body = NULL;
1047
1048         /*
1049          * Exception: Heads of `It' macros in `-diag' lists are not
1050          * parsed, even though `It' macros in general are parsed.
1051          */
1052         nparsed = MDOC_It == tok &&
1053                 MDOC_Bl == mdoc->last->parent->tok &&
1054                 LIST_diag == mdoc->last->parent->norm->Bl.type;
1055
1056         /*
1057          * The `Nd' macro has all arguments in its body: it's a hybrid
1058          * of block partial-explicit and full-implicit.  Stupid.
1059          */
1060
1061         if (MDOC_Nd == tok) {
1062                 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1063                         return(0);
1064                 head = mdoc->last;
1065                 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1066                         return(0);
1067                 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1068                         return(0);
1069                 body = mdoc->last;
1070         }
1071
1072         ac = ARGS_ERROR;
1073
1074         for ( ; ; ) {
1075                 la = *pos;
1076                 /* Initialise last-phrase-type with ARGS_PEND. */
1077                 lac = ARGS_ERROR == ac ? ARGS_PEND : ac;
1078                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1079
1080                 if (ARGS_PUNCT == ac)
1081                         break;
1082
1083                 if (ARGS_ERROR == ac)
1084                         return(0);
1085
1086                 if (ARGS_EOLN == ac) {
1087                         if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac)
1088                                 break;
1089                         /*
1090                          * This is necessary: if the last token on a
1091                          * line is a `Ta' or tab, then we'll get
1092                          * ARGS_EOLN, so we must be smart enough to
1093                          * reopen our scope if the last parse was a
1094                          * phrase or partial phrase.
1095                          */
1096                         if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
1097                                 return(0);
1098                         if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1099                                 return(0);
1100                         body = mdoc->last;
1101                         break;
1102                 }
1103
1104                 /*
1105                  * Emit leading punctuation (i.e., punctuation before
1106                  * the MDOC_HEAD) for non-phrase types.
1107                  */
1108
1109                 if (NULL == head &&
1110                                 ARGS_PEND != ac &&
1111                                 ARGS_PHRASE != ac &&
1112                                 ARGS_PPHRASE != ac &&
1113                                 ARGS_QWORD != ac &&
1114                                 DELIM_OPEN == mdoc_isdelim(p)) {
1115                         if ( ! dword(mdoc, line, la, p, DELIM_OPEN))
1116                                 return(0);
1117                         continue;
1118                 }
1119
1120                 /* Open a head if one hasn't been opened. */
1121
1122                 if (NULL == head) {
1123                         if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1124                                 return(0);
1125                         head = mdoc->last;
1126                 }
1127
1128                 if (ARGS_PHRASE == ac ||
1129                                 ARGS_PEND == ac ||
1130                                 ARGS_PPHRASE == ac) {
1131                         /*
1132                          * If we haven't opened a body yet, rewind the
1133                          * head; if we have, rewind that instead.
1134                          */
1135
1136                         mtt = body ? MDOC_BODY : MDOC_HEAD;
1137                         if ( ! rew_sub(mtt, mdoc, tok, line, ppos))
1138                                 return(0);
1139
1140                         /* Then allocate our body context. */
1141
1142                         if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1143                                 return(0);
1144                         body = mdoc->last;
1145
1146                         /*
1147                          * Process phrases: set whether we're in a
1148                          * partial-phrase (this effects line handling)
1149                          * then call down into the phrase parser.
1150                          */
1151
1152                         if (ARGS_PPHRASE == ac)
1153                                 mdoc->flags |= MDOC_PPHRASE;
1154                         if (ARGS_PEND == ac && ARGS_PPHRASE == lac)
1155                                 mdoc->flags |= MDOC_PPHRASE;
1156
1157                         if ( ! phrase(mdoc, line, la, buf))
1158                                 return(0);
1159
1160                         mdoc->flags &= ~MDOC_PPHRASE;
1161                         continue;
1162                 }
1163
1164                 ntok = nparsed || ARGS_QWORD == ac ?
1165                         MDOC_MAX : lookup(tok, p);
1166
1167                 if (MDOC_MAX == ntok) {
1168                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1169                                 return(0);
1170                         continue;
1171                 }
1172
1173                 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1174                         return(0);
1175                 break;
1176         }
1177
1178         if (NULL == head) {
1179                 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1180                         return(0);
1181                 head = mdoc->last;
1182         }
1183
1184         if (nl && ! append_delims(mdoc, line, pos, buf))
1185                 return(0);
1186
1187         /* If we've already opened our body, exit now. */
1188
1189         if (NULL != body)
1190                 goto out;
1191
1192         /*
1193          * If there is an open (i.e., unvalidated) sub-block requiring
1194          * explicit close-out, postpone switching the current block from
1195          * head to body until the rew_sub() call closing out that
1196          * sub-block.
1197          */
1198         for (n = mdoc->last; n && n != head; n = n->parent) {
1199                 if (MDOC_BLOCK == n->type &&
1200                                 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1201                                 ! (MDOC_VALID & n->flags)) {
1202                         n->pending = head;
1203                         return(1);
1204                 }
1205         }
1206
1207         /* Close out scopes to remain in a consistent state. */
1208
1209         if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1210                 return(0);
1211         if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1212                 return(0);
1213
1214 out:
1215         if ( ! (MDOC_FREECOL & mdoc->flags))
1216                 return(1);
1217
1218         if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
1219                 return(0);
1220         if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
1221                 return(0);
1222
1223         mdoc->flags &= ~MDOC_FREECOL;
1224         return(1);
1225 }
1226
1227
1228 static int
1229 blk_part_imp(MACRO_PROT_ARGS)
1230 {
1231         int               la, nl;
1232         enum mdoct        ntok;
1233         enum margserr     ac;
1234         char             *p;
1235         struct mdoc_node *blk; /* saved block context */
1236         struct mdoc_node *body; /* saved body context */
1237         struct mdoc_node *n;
1238
1239         nl = MDOC_NEWLINE & mdoc->flags;
1240
1241         /*
1242          * A macro that spans to the end of the line.  This is generally
1243          * (but not necessarily) called as the first macro.  The block
1244          * has a head as the immediate child, which is always empty,
1245          * followed by zero or more opening punctuation nodes, then the
1246          * body (which may be empty, depending on the macro), then zero
1247          * or more closing punctuation nodes.
1248          */
1249
1250         if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, NULL))
1251                 return(0);
1252
1253         blk = mdoc->last;
1254
1255         if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1256                 return(0);
1257         if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1258                 return(0);
1259
1260         /*
1261          * Open the body scope "on-demand", that is, after we've
1262          * processed all our the leading delimiters (open parenthesis,
1263          * etc.).
1264          */
1265
1266         for (body = NULL; ; ) {
1267                 la = *pos;
1268                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1269
1270                 if (ARGS_ERROR == ac)
1271                         return(0);
1272                 if (ARGS_EOLN == ac)
1273                         break;
1274                 if (ARGS_PUNCT == ac)
1275                         break;
1276
1277                 if (NULL == body && ARGS_QWORD != ac &&
1278                                 DELIM_OPEN == mdoc_isdelim(p)) {
1279                         if ( ! dword(mdoc, line, la, p, DELIM_OPEN))
1280                                 return(0);
1281                         continue;
1282                 }
1283
1284                 if (NULL == body) {
1285                        if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1286                                return(0);
1287                         body = mdoc->last;
1288                 }
1289
1290                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1291
1292                 if (MDOC_MAX == ntok) {
1293                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1294                                 return(0);
1295                         continue;
1296                 }
1297
1298                 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1299                         return(0);
1300                 break;
1301         }
1302
1303         /* Clean-ups to leave in a consistent state. */
1304
1305         if (NULL == body) {
1306                 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1307                         return(0);
1308                 body = mdoc->last;
1309         }
1310
1311         for (n = body->child; n && n->next; n = n->next)
1312                 /* Do nothing. */ ;
1313
1314         /*
1315          * End of sentence spacing: if the last node is a text node and
1316          * has a trailing period, then mark it as being end-of-sentence.
1317          */
1318
1319         if (n && MDOC_TEXT == n->type && n->string)
1320                 if (mandoc_eos(n->string, strlen(n->string), 1))
1321                         n->flags |= MDOC_EOS;
1322
1323         /* Up-propagate the end-of-space flag. */
1324
1325         if (n && (MDOC_EOS & n->flags)) {
1326                 body->flags |= MDOC_EOS;
1327                 body->parent->flags |= MDOC_EOS;
1328         }
1329
1330         /*
1331          * If there is an open sub-block requiring explicit close-out,
1332          * postpone closing out the current block
1333          * until the rew_sub() call closing out the sub-block.
1334          */
1335         for (n = mdoc->last; n && n != body && n != blk->parent;
1336                         n = n->parent) {
1337                 if (MDOC_BLOCK == n->type &&
1338                     MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1339                     ! (MDOC_VALID & n->flags)) {
1340                         make_pending(n, tok, mdoc, line, ppos);
1341                         if ( ! mdoc_endbody_alloc(mdoc, line, ppos,
1342                             tok, body, ENDBODY_NOSPACE))
1343                                 return(0);
1344                         return(1);
1345                 }
1346         }
1347
1348         /*
1349          * If we can't rewind to our body, then our scope has already
1350          * been closed by another macro (like `Oc' closing `Op').  This
1351          * is ugly behaviour nodding its head to OpenBSD's overwhelming
1352          * crufty use of `Op' breakage.
1353          */
1354         if (n != body)
1355                 mandoc_vmsg(MANDOCERR_SCOPENEST, mdoc->parse, line, ppos,
1356                                 "%s broken", mdoc_macronames[tok]);
1357
1358         if (n && ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
1359                 return(0);
1360
1361         /* Standard appending of delimiters. */
1362
1363         if (nl && ! append_delims(mdoc, line, pos, buf))
1364                 return(0);
1365
1366         /* Rewind scope, if applicable. */
1367
1368         if (n && ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
1369                 return(0);
1370
1371         /* Move trailing .Ns out of scope. */
1372
1373         for (n = body->child; n && n->next; n = n->next)
1374                 /* Do nothing. */ ;
1375         if (n && MDOC_Ns == n->tok)
1376                 mdoc_node_relink(mdoc, n);
1377
1378         return(1);
1379 }
1380
1381
1382 static int
1383 blk_part_exp(MACRO_PROT_ARGS)
1384 {
1385         int               la, nl;
1386         enum margserr     ac;
1387         struct mdoc_node *head; /* keep track of head */
1388         struct mdoc_node *body; /* keep track of body */
1389         char             *p;
1390         enum mdoct        ntok;
1391
1392         nl = MDOC_NEWLINE & mdoc->flags;
1393
1394         /*
1395          * The opening of an explicit macro having zero or more leading
1396          * punctuation nodes; a head with optional single element (the
1397          * case of `Eo'); and a body that may be empty.
1398          */
1399
1400         if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, NULL))
1401                 return(0);
1402
1403         for (head = body = NULL; ; ) {
1404                 la = *pos;
1405                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1406
1407                 if (ARGS_ERROR == ac)
1408                         return(0);
1409                 if (ARGS_PUNCT == ac)
1410                         break;
1411                 if (ARGS_EOLN == ac)
1412                         break;
1413
1414                 /* Flush out leading punctuation. */
1415
1416                 if (NULL == head && ARGS_QWORD != ac &&
1417                                 DELIM_OPEN == mdoc_isdelim(p)) {
1418                         assert(NULL == body);
1419                         if ( ! dword(mdoc, line, la, p, DELIM_OPEN))
1420                                 return(0);
1421                         continue;
1422                 }
1423
1424                 if (NULL == head) {
1425                         assert(NULL == body);
1426                         if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1427                                 return(0);
1428                         head = mdoc->last;
1429                 }
1430
1431                 /*
1432                  * `Eo' gobbles any data into the head, but most other
1433                  * macros just immediately close out and begin the body.
1434                  */
1435
1436                 if (NULL == body) {
1437                         assert(head);
1438                         /* No check whether it's a macro! */
1439                         if (MDOC_Eo == tok)
1440                                 if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1441                                         return(0);
1442
1443                         if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1444                                 return(0);
1445                         if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1446                                 return(0);
1447                         body = mdoc->last;
1448
1449                         if (MDOC_Eo == tok)
1450                                 continue;
1451                 }
1452
1453                 assert(NULL != head && NULL != body);
1454
1455                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1456
1457                 if (MDOC_MAX == ntok) {
1458                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1459                                 return(0);
1460                         continue;
1461                 }
1462
1463                 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1464                         return(0);
1465                 break;
1466         }
1467
1468         /* Clean-up to leave in a consistent state. */
1469
1470         if (NULL == head)
1471                 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1472                         return(0);
1473
1474         if (NULL == body) {
1475                 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1476                         return(0);
1477                 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1478                         return(0);
1479         }
1480
1481         /* Standard appending of delimiters. */
1482
1483         if ( ! nl)
1484                 return(1);
1485         return(append_delims(mdoc, line, pos, buf));
1486 }
1487
1488
1489 /* ARGSUSED */
1490 static int
1491 in_line_argn(MACRO_PROT_ARGS)
1492 {
1493         int              la, flushed, j, maxargs, nl;
1494         enum margserr    ac;
1495         enum margverr    av;
1496         struct mdoc_arg *arg;
1497         char            *p;
1498         enum mdoct       ntok;
1499
1500         nl = MDOC_NEWLINE & mdoc->flags;
1501
1502         /*
1503          * A line macro that has a fixed number of arguments (maxargs).
1504          * Only open the scope once the first non-leading-punctuation is
1505          * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1506          * keep it open until the maximum number of arguments are
1507          * exhausted.
1508          */
1509
1510         switch (tok) {
1511         case (MDOC_Ap):
1512                 /* FALLTHROUGH */
1513         case (MDOC_No):
1514                 /* FALLTHROUGH */
1515         case (MDOC_Ns):
1516                 /* FALLTHROUGH */
1517         case (MDOC_Ux):
1518                 maxargs = 0;
1519                 break;
1520         case (MDOC_Bx):
1521                 /* FALLTHROUGH */
1522         case (MDOC_Xr):
1523                 maxargs = 2;
1524                 break;
1525         default:
1526                 maxargs = 1;
1527                 break;
1528         }
1529
1530         for (arg = NULL; ; ) {
1531                 la = *pos;
1532                 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1533
1534                 if (ARGV_WORD == av) {
1535                         *pos = la;
1536                         break;
1537                 }
1538
1539                 if (ARGV_EOLN == av)
1540                         break;
1541                 if (ARGV_ARG == av)
1542                         continue;
1543
1544                 mdoc_argv_free(arg);
1545                 return(0);
1546         }
1547
1548         for (flushed = j = 0; ; ) {
1549                 la = *pos;
1550                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1551
1552                 if (ARGS_ERROR == ac)
1553                         return(0);
1554                 if (ARGS_PUNCT == ac)
1555                         break;
1556                 if (ARGS_EOLN == ac)
1557                         break;
1558
1559                 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1560                                 ARGS_QWORD != ac && 0 == j &&
1561                                 DELIM_OPEN == mdoc_isdelim(p)) {
1562                         if ( ! dword(mdoc, line, la, p, DELIM_OPEN))
1563                                 return(0);
1564                         continue;
1565                 } else if (0 == j)
1566                        if ( ! mdoc_elem_alloc(mdoc, line, la, tok, arg))
1567                                return(0);
1568
1569                 if (j == maxargs && ! flushed) {
1570                         if ( ! rew_elem(mdoc, tok))
1571                                 return(0);
1572                         flushed = 1;
1573                 }
1574
1575                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1576
1577                 if (MDOC_MAX != ntok) {
1578                         if ( ! flushed && ! rew_elem(mdoc, tok))
1579                                 return(0);
1580                         flushed = 1;
1581                         if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1582                                 return(0);
1583                         j++;
1584                         break;
1585                 }
1586
1587                 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1588                                 ARGS_QWORD != ac &&
1589                                 ! flushed &&
1590                                 DELIM_NONE != mdoc_isdelim(p)) {
1591                         if ( ! rew_elem(mdoc, tok))
1592                                 return(0);
1593                         flushed = 1;
1594                 }
1595
1596                 if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1597                         return(0);
1598                 j++;
1599         }
1600
1601         if (0 == j && ! mdoc_elem_alloc(mdoc, line, la, tok, arg))
1602                return(0);
1603
1604         /* Close out in a consistent state. */
1605
1606         if ( ! flushed && ! rew_elem(mdoc, tok))
1607                 return(0);
1608         if ( ! nl)
1609                 return(1);
1610         return(append_delims(mdoc, line, pos, buf));
1611 }
1612
1613
1614 static int
1615 in_line_eoln(MACRO_PROT_ARGS)
1616 {
1617         int              la;
1618         enum margserr    ac;
1619         enum margverr    av;
1620         struct mdoc_arg *arg;
1621         char            *p;
1622         enum mdoct       ntok;
1623
1624         assert( ! (MDOC_PARSED & mdoc_macros[tok].flags));
1625
1626         if (tok == MDOC_Pp)
1627                 rew_sub(MDOC_BLOCK, mdoc, MDOC_Nm, line, ppos);
1628
1629         /* Parse macro arguments. */
1630
1631         for (arg = NULL; ; ) {
1632                 la = *pos;
1633                 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1634
1635                 if (ARGV_WORD == av) {
1636                         *pos = la;
1637                         break;
1638                 }
1639                 if (ARGV_EOLN == av)
1640                         break;
1641                 if (ARGV_ARG == av)
1642                         continue;
1643
1644                 mdoc_argv_free(arg);
1645                 return(0);
1646         }
1647
1648         /* Open element scope. */
1649
1650         if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg))
1651                 return(0);
1652
1653         /* Parse argument terms. */
1654
1655         for (;;) {
1656                 la = *pos;
1657                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1658
1659                 if (ARGS_ERROR == ac)
1660                         return(0);
1661                 if (ARGS_EOLN == ac)
1662                         break;
1663
1664                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1665
1666                 if (MDOC_MAX == ntok) {
1667                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1668                                 return(0);
1669                         continue;
1670                 }
1671
1672                 if ( ! rew_elem(mdoc, tok))
1673                         return(0);
1674                 return(mdoc_macro(mdoc, ntok, line, la, pos, buf));
1675         }
1676
1677         /* Close out (no delimiters). */
1678
1679         return(rew_elem(mdoc, tok));
1680 }
1681
1682
1683 /* ARGSUSED */
1684 static int
1685 ctx_synopsis(MACRO_PROT_ARGS)
1686 {
1687         int              nl;
1688
1689         nl = MDOC_NEWLINE & mdoc->flags;
1690
1691         /* If we're not in the SYNOPSIS, go straight to in-line. */
1692         if ( ! (MDOC_SYNOPSIS & mdoc->flags))
1693                 return(in_line(mdoc, tok, line, ppos, pos, buf));
1694
1695         /* If we're a nested call, same place. */
1696         if ( ! nl)
1697                 return(in_line(mdoc, tok, line, ppos, pos, buf));
1698
1699         /*
1700          * XXX: this will open a block scope; however, if later we end
1701          * up formatting the block scope, then child nodes will inherit
1702          * the formatting.  Be careful.
1703          */
1704         if (MDOC_Nm == tok)
1705                 return(blk_full(mdoc, tok, line, ppos, pos, buf));
1706         assert(MDOC_Vt == tok);
1707         return(blk_part_imp(mdoc, tok, line, ppos, pos, buf));
1708 }
1709
1710
1711 /* ARGSUSED */
1712 static int
1713 obsolete(MACRO_PROT_ARGS)
1714 {
1715
1716         mdoc_pmsg(mdoc, line, ppos, MANDOCERR_MACROOBS);
1717         return(1);
1718 }
1719
1720
1721 /*
1722  * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1723  * They're unusual because they're basically free-form text until a
1724  * macro is encountered.
1725  */
1726 static int
1727 phrase(struct mdoc *mdoc, int line, int ppos, char *buf)
1728 {
1729         int              la, pos;
1730         enum margserr    ac;
1731         enum mdoct       ntok;
1732         char            *p;
1733
1734         for (pos = ppos; ; ) {
1735                 la = pos;
1736
1737                 ac = mdoc_zargs(mdoc, line, &pos, buf, &p);
1738
1739                 if (ARGS_ERROR == ac)
1740                         return(0);
1741                 if (ARGS_EOLN == ac)
1742                         break;
1743
1744                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1745
1746                 if (MDOC_MAX == ntok) {
1747                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1748                                 return(0);
1749                         continue;
1750                 }
1751
1752                 if ( ! mdoc_macro(mdoc, ntok, line, la, &pos, buf))
1753                         return(0);
1754                 return(append_delims(mdoc, line, &pos, buf));
1755         }
1756
1757         return(1);
1758 }
1759
1760
1761 /* ARGSUSED */
1762 static int
1763 phrase_ta(MACRO_PROT_ARGS)
1764 {
1765         struct mdoc_node *n;
1766         int               la;
1767         enum mdoct        ntok;
1768         enum margserr     ac;
1769         char             *p;
1770
1771         /* Make sure we are in a column list or ignore this macro. */
1772         n = mdoc->last;
1773         while (NULL != n && MDOC_Bl != n->tok)
1774                 n = n->parent;
1775         if (NULL == n || LIST_column != n->norm->Bl.type) {
1776                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_STRAYTA);
1777                 return(1);
1778         }
1779
1780         /* Advance to the next column. */
1781         if ( ! rew_sub(MDOC_BODY, mdoc, MDOC_It, line, ppos))
1782                 return(0);
1783         if ( ! mdoc_body_alloc(mdoc, line, ppos, MDOC_It))
1784                 return(0);
1785
1786         for (;;) {
1787                 la = *pos;
1788                 ac = mdoc_zargs(mdoc, line, pos, buf, &p);
1789
1790                 if (ARGS_ERROR == ac)
1791                         return(0);
1792                 if (ARGS_EOLN == ac)
1793                         break;
1794
1795                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1796
1797                 if (MDOC_MAX == ntok) {
1798                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1799                                 return(0);
1800                         continue;
1801                 }
1802
1803                 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1804                         return(0);
1805                 return(append_delims(mdoc, line, pos, buf));
1806         }
1807
1808         return(1);
1809 }