mdoc_macro.c

   1 /*      $Id: mdoc_macro.c,v 1.120 2012/11/18 00:05:35 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                 if ( ! mdoc_macro(mdoc, ntok, line, lastarg, pos, buf))
 796                         return(0);
 797                 break;
 798         }
 799
 800         if ( ! flushed && ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
 801                 return(0);
 802
 803         if ( ! nl)
 804                 return(1);
 805         return(append_delims(mdoc, line, pos, buf));
 806 }
 807
 808
 809 static int
 810 in_line(MACRO_PROT_ARGS)
 811 {
 812         int              la, scope, cnt, nc, nl;
 813         enum margverr    av;
 814         enum mdoct       ntok;
 815         enum margserr    ac;
 816         enum mdelim      d;
 817         struct mdoc_arg *arg;
 818         char            *p;
 819
 820         nl = MDOC_NEWLINE & mdoc->flags;
 821
 822         /*
 823          * Whether we allow ignored elements (those without content,
 824          * usually because of reserved words) to squeak by.
 825          */
 826
 827         switch (tok) {
 828         case (MDOC_An):
 829                 /* FALLTHROUGH */
 830         case (MDOC_Ar):
 831                 /* FALLTHROUGH */
 832         case (MDOC_Fl):
 833                 /* FALLTHROUGH */
 834         case (MDOC_Mt):
 835                 /* FALLTHROUGH */
 836         case (MDOC_Nm):
 837                 /* FALLTHROUGH */
 838         case (MDOC_Pa):
 839                 nc = 1;
 840                 break;
 841         default:
 842                 nc = 0;
 843                 break;
 844         }
 845
 846         for (arg = NULL;; ) {
 847                 la = *pos;
 848                 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
 849
 850                 if (ARGV_WORD == av) {
 851                         *pos = la;
 852                         break;
 853                 }
 854                 if (ARGV_EOLN == av)
 855                         break;
 856                 if (ARGV_ARG == av)
 857                         continue;
 858
 859                 mdoc_argv_free(arg);
 860                 return(0);
 861         }
 862
 863         for (cnt = scope = 0;; ) {
 864                 la = *pos;
 865                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
 866
 867                 if (ARGS_ERROR == ac)
 868                         return(0);
 869                 if (ARGS_EOLN == ac)
 870                         break;
 871                 if (ARGS_PUNCT == ac)
 872                         break;
 873
 874                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
 875
 876                 /*
 877                  * In this case, we've located a submacro and must
 878                  * execute it.  Close out scope, if open.  If no
 879                  * elements have been generated, either create one (nc)
 880                  * or raise a warning.
 881                  */
 882
 883                 if (MDOC_MAX != ntok) {
 884                         if (scope && ! rew_elem(mdoc, tok))
 885                                 return(0);
 886                         if (nc && 0 == cnt) {
 887                                 if ( ! mdoc_elem_alloc(mdoc, line,
 888                                                 ppos, tok, arg))
 889                                         return(0);
 890                                 if ( ! rew_last(mdoc, mdoc->last))
 891                                         return(0);
 892                         } else if ( ! nc && 0 == cnt) {
 893                                 mdoc_argv_free(arg);
 894                                 mdoc_pmsg(mdoc, line, ppos,
 895                                         MANDOCERR_MACROEMPTY);
 896                         }
 897
 898                         if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
 899                                 return(0);
 900                         if ( ! nl)
 901                                 return(1);
 902                         return(append_delims(mdoc, line, pos, buf));
 903                 }
 904
 905                 /*
 906                  * Non-quote-enclosed punctuation.  Set up our scope, if
 907                  * a word; rewind the scope, if a delimiter; then append
 908                  * the word.
 909                  */
 910
 911                 d = ARGS_QWORD == ac ? DELIM_NONE : mdoc_isdelim(p);
 912
 913                 if (DELIM_NONE != d) {
 914                         /*
 915                          * If we encounter closing punctuation, no word
 916                          * has been omitted, no scope is open, and we're
 917                          * allowed to have an empty element, then start
 918                          * a new scope.  `Ar', `Fl', and `Li', only do
 919                          * this once per invocation.  There may be more
 920                          * of these (all of them?).
 921                          */
 922                         if (0 == cnt && (nc || MDOC_Li == tok) &&
 923                                         DELIM_CLOSE == d && ! scope) {
 924                                 if ( ! mdoc_elem_alloc(mdoc, line,
 925                                                 ppos, tok, arg))
 926                                         return(0);
 927                                 if (MDOC_Ar == tok || MDOC_Li == tok ||
 928                                                 MDOC_Fl == tok)
 929                                         cnt++;
 930                                 scope = 1;
 931                         }
 932                         /*
 933                          * Close out our scope, if one is open, before
 934                          * any punctuation.
 935                          */
 936                         if (scope && ! rew_elem(mdoc, tok))
 937                                 return(0);
 938                         scope = 0;
 939                 } else if ( ! scope) {
 940                         if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg))
 941                                 return(0);
 942                         scope = 1;
 943                 }
 944
 945                 if (DELIM_NONE == d)
 946                         cnt++;
 947
 948                 if ( ! dword(mdoc, line, la, p, d))
 949                         return(0);
 950
 951                 /*
 952                  * `Fl' macros have their scope re-opened with each new
 953                  * word so that the `-' can be added to each one without
 954                  * having to parse out spaces.
 955                  */
 956                 if (scope && MDOC_Fl == tok) {
 957                         if ( ! rew_elem(mdoc, tok))
 958                                 return(0);
 959                         scope = 0;
 960                 }
 961         }
 962
 963         if (scope && ! rew_elem(mdoc, tok))
 964                 return(0);
 965
 966         /*
 967          * If no elements have been collected and we're allowed to have
 968          * empties (nc), open a scope and close it out.  Otherwise,
 969          * raise a warning.
 970          */
 971
 972         if (nc && 0 == cnt) {
 973                 if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg))
 974                         return(0);
 975                 if ( ! rew_last(mdoc, mdoc->last))
 976                         return(0);
 977         } else if ( ! nc && 0 == cnt) {
 978                 mdoc_argv_free(arg);
 979                 mdoc_pmsg(mdoc, line, ppos, MANDOCERR_MACROEMPTY);
 980         }
 981
 982         if ( ! nl)
 983                 return(1);
 984         return(append_delims(mdoc, line, pos, buf));
 985 }
 986
 987
 988 static int
 989 blk_full(MACRO_PROT_ARGS)
 990 {
 991         int               la, nl, nparsed;
 992         struct mdoc_arg  *arg;
 993         struct mdoc_node *head; /* save of head macro */
 994         struct mdoc_node *body; /* save of body macro */
 995         struct mdoc_node *n;
 996         enum mdoc_type    mtt;
 997         enum mdoct        ntok;
 998         enum margserr     ac, lac;
 999         enum margverr     av;
1000         char             *p;
1001
1002         nl = MDOC_NEWLINE & mdoc->flags;
1003
1004         /* Close out prior implicit scope. */
1005
1006         if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) {
1007                 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
1008                         return(0);
1009                 if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
1010                         return(0);
1011         }
1012
1013         /*
1014          * This routine accommodates implicitly- and explicitly-scoped
1015          * macro openings.  Implicit ones first close out prior scope
1016          * (seen above).  Delay opening the head until necessary to
1017          * allow leading punctuation to print.  Special consideration
1018          * for `It -column', which has phrase-part syntax instead of
1019          * regular child nodes.
1020          */
1021
1022         for (arg = NULL;; ) {
1023                 la = *pos;
1024                 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1025
1026                 if (ARGV_WORD == av) {
1027                         *pos = la;
1028                         break;
1029                 }
1030
1031                 if (ARGV_EOLN == av)
1032                         break;
1033                 if (ARGV_ARG == av)
1034                         continue;
1035
1036                 mdoc_argv_free(arg);
1037                 return(0);
1038         }
1039
1040         if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, arg))
1041                 return(0);
1042
1043         head = body = NULL;
1044
1045         /*
1046          * Exception: Heads of `It' macros in `-diag' lists are not
1047          * parsed, even though `It' macros in general are parsed.
1048          */
1049         nparsed = MDOC_It == tok &&
1050                 MDOC_Bl == mdoc->last->parent->tok &&
1051                 LIST_diag == mdoc->last->parent->norm->Bl.type;
1052
1053         /*
1054          * The `Nd' macro has all arguments in its body: it's a hybrid
1055          * of block partial-explicit and full-implicit.  Stupid.
1056          */
1057
1058         if (MDOC_Nd == tok) {
1059                 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1060                         return(0);
1061                 head = mdoc->last;
1062                 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1063                         return(0);
1064                 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1065                         return(0);
1066                 body = mdoc->last;
1067         }
1068
1069         ac = ARGS_ERROR;
1070
1071         for ( ; ; ) {
1072                 la = *pos;
1073                 /* Initialise last-phrase-type with ARGS_PEND. */
1074                 lac = ARGS_ERROR == ac ? ARGS_PEND : ac;
1075                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1076
1077                 if (ARGS_PUNCT == ac)
1078                         break;
1079
1080                 if (ARGS_ERROR == ac)
1081                         return(0);
1082
1083                 if (ARGS_EOLN == ac) {
1084                         if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac)
1085                                 break;
1086                         /*
1087                          * This is necessary: if the last token on a
1088                          * line is a `Ta' or tab, then we'll get
1089                          * ARGS_EOLN, so we must be smart enough to
1090                          * reopen our scope if the last parse was a
1091                          * phrase or partial phrase.
1092                          */
1093                         if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
1094                                 return(0);
1095                         if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1096                                 return(0);
1097                         body = mdoc->last;
1098                         break;
1099                 }
1100
1101                 /*
1102                  * Emit leading punctuation (i.e., punctuation before
1103                  * the MDOC_HEAD) for non-phrase types.
1104                  */
1105
1106                 if (NULL == head &&
1107                                 ARGS_PEND != ac &&
1108                                 ARGS_PHRASE != ac &&
1109                                 ARGS_PPHRASE != ac &&
1110                                 ARGS_QWORD != ac &&
1111                                 DELIM_OPEN == mdoc_isdelim(p)) {
1112                         if ( ! dword(mdoc, line, la, p, DELIM_OPEN))
1113                                 return(0);
1114                         continue;
1115                 }
1116
1117                 /* Open a head if one hasn't been opened. */
1118
1119                 if (NULL == head) {
1120                         if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1121                                 return(0);
1122                         head = mdoc->last;
1123                 }
1124
1125                 if (ARGS_PHRASE == ac ||
1126                                 ARGS_PEND == ac ||
1127                                 ARGS_PPHRASE == ac) {
1128                         /*
1129                          * If we haven't opened a body yet, rewind the
1130                          * head; if we have, rewind that instead.
1131                          */
1132
1133                         mtt = body ? MDOC_BODY : MDOC_HEAD;
1134                         if ( ! rew_sub(mtt, mdoc, tok, line, ppos))
1135                                 return(0);
1136
1137                         /* Then allocate our body context. */
1138
1139                         if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1140                                 return(0);
1141                         body = mdoc->last;
1142
1143                         /*
1144                          * Process phrases: set whether we're in a
1145                          * partial-phrase (this effects line handling)
1146                          * then call down into the phrase parser.
1147                          */
1148
1149                         if (ARGS_PPHRASE == ac)
1150                                 mdoc->flags |= MDOC_PPHRASE;
1151                         if (ARGS_PEND == ac && ARGS_PPHRASE == lac)
1152                                 mdoc->flags |= MDOC_PPHRASE;
1153
1154                         if ( ! phrase(mdoc, line, la, buf))
1155                                 return(0);
1156
1157                         mdoc->flags &= ~MDOC_PPHRASE;
1158                         continue;
1159                 }
1160
1161                 ntok = nparsed || ARGS_QWORD == ac ?
1162                         MDOC_MAX : lookup(tok, p);
1163
1164                 if (MDOC_MAX == ntok) {
1165                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1166                                 return(0);
1167                         continue;
1168                 }
1169
1170                 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1171                         return(0);
1172                 break;
1173         }
1174
1175         if (NULL == head) {
1176                 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1177                         return(0);
1178                 head = mdoc->last;
1179         }
1180
1181         if (nl && ! append_delims(mdoc, line, pos, buf))
1182                 return(0);
1183
1184         /* If we've already opened our body, exit now. */
1185
1186         if (NULL != body)
1187                 goto out;
1188
1189         /*
1190          * If there is an open (i.e., unvalidated) sub-block requiring
1191          * explicit close-out, postpone switching the current block from
1192          * head to body until the rew_sub() call closing out that
1193          * sub-block.
1194          */
1195         for (n = mdoc->last; n && n != head; n = n->parent) {
1196                 if (MDOC_BLOCK == n->type &&
1197                                 MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1198                                 ! (MDOC_VALID & n->flags)) {
1199                         n->pending = head;
1200                         return(1);
1201                 }
1202         }
1203
1204         /* Close out scopes to remain in a consistent state. */
1205
1206         if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1207                 return(0);
1208         if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1209                 return(0);
1210
1211 out:
1212         if ( ! (MDOC_FREECOL & mdoc->flags))
1213                 return(1);
1214
1215         if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
1216                 return(0);
1217         if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
1218                 return(0);
1219
1220         mdoc->flags &= ~MDOC_FREECOL;
1221         return(1);
1222 }
1223
1224
1225 static int
1226 blk_part_imp(MACRO_PROT_ARGS)
1227 {
1228         int               la, nl;
1229         enum mdoct        ntok;
1230         enum margserr     ac;
1231         char             *p;
1232         struct mdoc_node *blk; /* saved block context */
1233         struct mdoc_node *body; /* saved body context */
1234         struct mdoc_node *n;
1235
1236         nl = MDOC_NEWLINE & mdoc->flags;
1237
1238         /*
1239          * A macro that spans to the end of the line.  This is generally
1240          * (but not necessarily) called as the first macro.  The block
1241          * has a head as the immediate child, which is always empty,
1242          * followed by zero or more opening punctuation nodes, then the
1243          * body (which may be empty, depending on the macro), then zero
1244          * or more closing punctuation nodes.
1245          */
1246
1247         if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, NULL))
1248                 return(0);
1249
1250         blk = mdoc->last;
1251
1252         if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1253                 return(0);
1254         if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1255                 return(0);
1256
1257         /*
1258          * Open the body scope "on-demand", that is, after we've
1259          * processed all our the leading delimiters (open parenthesis,
1260          * etc.).
1261          */
1262
1263         for (body = NULL; ; ) {
1264                 la = *pos;
1265                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1266
1267                 if (ARGS_ERROR == ac)
1268                         return(0);
1269                 if (ARGS_EOLN == ac)
1270                         break;
1271                 if (ARGS_PUNCT == ac)
1272                         break;
1273
1274                 if (NULL == body && ARGS_QWORD != ac &&
1275                                 DELIM_OPEN == mdoc_isdelim(p)) {
1276                         if ( ! dword(mdoc, line, la, p, DELIM_OPEN))
1277                                 return(0);
1278                         continue;
1279                 }
1280
1281                 if (NULL == body) {
1282                        if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1283                                return(0);
1284                         body = mdoc->last;
1285                 }
1286
1287                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1288
1289                 if (MDOC_MAX == ntok) {
1290                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1291                                 return(0);
1292                         continue;
1293                 }
1294
1295                 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1296                         return(0);
1297                 break;
1298         }
1299
1300         /* Clean-ups to leave in a consistent state. */
1301
1302         if (NULL == body) {
1303                 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1304                         return(0);
1305                 body = mdoc->last;
1306         }
1307
1308         for (n = body->child; n && n->next; n = n->next)
1309                 /* Do nothing. */ ;
1310
1311         /*
1312          * End of sentence spacing: if the last node is a text node and
1313          * has a trailing period, then mark it as being end-of-sentence.
1314          */
1315
1316         if (n && MDOC_TEXT == n->type && n->string)
1317                 if (mandoc_eos(n->string, strlen(n->string), 1))
1318                         n->flags |= MDOC_EOS;
1319
1320         /* Up-propagate the end-of-space flag. */
1321
1322         if (n && (MDOC_EOS & n->flags)) {
1323                 body->flags |= MDOC_EOS;
1324                 body->parent->flags |= MDOC_EOS;
1325         }
1326
1327         /*
1328          * If there is an open sub-block requiring explicit close-out,
1329          * postpone closing out the current block
1330          * until the rew_sub() call closing out the sub-block.
1331          */
1332         for (n = mdoc->last; n && n != body && n != blk->parent;
1333                         n = n->parent) {
1334                 if (MDOC_BLOCK == n->type &&
1335                     MDOC_EXPLICIT & mdoc_macros[n->tok].flags &&
1336                     ! (MDOC_VALID & n->flags)) {
1337                         make_pending(n, tok, mdoc, line, ppos);
1338                         if ( ! mdoc_endbody_alloc(mdoc, line, ppos,
1339                             tok, body, ENDBODY_NOSPACE))
1340                                 return(0);
1341                         return(1);
1342                 }
1343         }
1344
1345         /*
1346          * If we can't rewind to our body, then our scope has already
1347          * been closed by another macro (like `Oc' closing `Op').  This
1348          * is ugly behaviour nodding its head to OpenBSD's overwhelming
1349          * crufty use of `Op' breakage.
1350          */
1351         if (n != body)
1352                 mandoc_vmsg(MANDOCERR_SCOPENEST, mdoc->parse, line, ppos,
1353                                 "%s broken", mdoc_macronames[tok]);
1354
1355         if (n && ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos))
1356                 return(0);
1357
1358         /* Standard appending of delimiters. */
1359
1360         if (nl && ! append_delims(mdoc, line, pos, buf))
1361                 return(0);
1362
1363         /* Rewind scope, if applicable. */
1364
1365         if (n && ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos))
1366                 return(0);
1367
1368         /* Move trailing .Ns out of scope. */
1369
1370         for (n = body->child; n && n->next; n = n->next)
1371                 /* Do nothing. */ ;
1372         if (n && MDOC_Ns == n->tok)
1373                 mdoc_node_relink(mdoc, n);
1374
1375         return(1);
1376 }
1377
1378
1379 static int
1380 blk_part_exp(MACRO_PROT_ARGS)
1381 {
1382         int               la, nl;
1383         enum margserr     ac;
1384         struct mdoc_node *head; /* keep track of head */
1385         struct mdoc_node *body; /* keep track of body */
1386         char             *p;
1387         enum mdoct        ntok;
1388
1389         nl = MDOC_NEWLINE & mdoc->flags;
1390
1391         /*
1392          * The opening of an explicit macro having zero or more leading
1393          * punctuation nodes; a head with optional single element (the
1394          * case of `Eo'); and a body that may be empty.
1395          */
1396
1397         if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, NULL))
1398                 return(0);
1399
1400         for (head = body = NULL; ; ) {
1401                 la = *pos;
1402                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1403
1404                 if (ARGS_ERROR == ac)
1405                         return(0);
1406                 if (ARGS_PUNCT == ac)
1407                         break;
1408                 if (ARGS_EOLN == ac)
1409                         break;
1410
1411                 /* Flush out leading punctuation. */
1412
1413                 if (NULL == head && ARGS_QWORD != ac &&
1414                                 DELIM_OPEN == mdoc_isdelim(p)) {
1415                         assert(NULL == body);
1416                         if ( ! dword(mdoc, line, la, p, DELIM_OPEN))
1417                                 return(0);
1418                         continue;
1419                 }
1420
1421                 if (NULL == head) {
1422                         assert(NULL == body);
1423                         if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1424                                 return(0);
1425                         head = mdoc->last;
1426                 }
1427
1428                 /*
1429                  * `Eo' gobbles any data into the head, but most other
1430                  * macros just immediately close out and begin the body.
1431                  */
1432
1433                 if (NULL == body) {
1434                         assert(head);
1435                         /* No check whether it's a macro! */
1436                         if (MDOC_Eo == tok)
1437                                 if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1438                                         return(0);
1439
1440                         if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1441                                 return(0);
1442                         if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1443                                 return(0);
1444                         body = mdoc->last;
1445
1446                         if (MDOC_Eo == tok)
1447                                 continue;
1448                 }
1449
1450                 assert(NULL != head && NULL != body);
1451
1452                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1453
1454                 if (MDOC_MAX == ntok) {
1455                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1456                                 return(0);
1457                         continue;
1458                 }
1459
1460                 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1461                         return(0);
1462                 break;
1463         }
1464
1465         /* Clean-up to leave in a consistent state. */
1466
1467         if (NULL == head)
1468                 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok))
1469                         return(0);
1470
1471         if (NULL == body) {
1472                 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos))
1473                         return(0);
1474                 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok))
1475                         return(0);
1476         }
1477
1478         /* Standard appending of delimiters. */
1479
1480         if ( ! nl)
1481                 return(1);
1482         return(append_delims(mdoc, line, pos, buf));
1483 }
1484
1485
1486 /* ARGSUSED */
1487 static int
1488 in_line_argn(MACRO_PROT_ARGS)
1489 {
1490         int              la, flushed, j, maxargs, nl;
1491         enum margserr    ac;
1492         enum margverr    av;
1493         struct mdoc_arg *arg;
1494         char            *p;
1495         enum mdoct       ntok;
1496
1497         nl = MDOC_NEWLINE & mdoc->flags;
1498
1499         /*
1500          * A line macro that has a fixed number of arguments (maxargs).
1501          * Only open the scope once the first non-leading-punctuation is
1502          * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then
1503          * keep it open until the maximum number of arguments are
1504          * exhausted.
1505          */
1506
1507         switch (tok) {
1508         case (MDOC_Ap):
1509                 /* FALLTHROUGH */
1510         case (MDOC_No):
1511                 /* FALLTHROUGH */
1512         case (MDOC_Ns):
1513                 /* FALLTHROUGH */
1514         case (MDOC_Ux):
1515                 maxargs = 0;
1516                 break;
1517         case (MDOC_Bx):
1518                 /* FALLTHROUGH */
1519         case (MDOC_Xr):
1520                 maxargs = 2;
1521                 break;
1522         default:
1523                 maxargs = 1;
1524                 break;
1525         }
1526
1527         for (arg = NULL; ; ) {
1528                 la = *pos;
1529                 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1530
1531                 if (ARGV_WORD == av) {
1532                         *pos = la;
1533                         break;
1534                 }
1535
1536                 if (ARGV_EOLN == av)
1537                         break;
1538                 if (ARGV_ARG == av)
1539                         continue;
1540
1541                 mdoc_argv_free(arg);
1542                 return(0);
1543         }
1544
1545         for (flushed = j = 0; ; ) {
1546                 la = *pos;
1547                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1548
1549                 if (ARGS_ERROR == ac)
1550                         return(0);
1551                 if (ARGS_PUNCT == ac)
1552                         break;
1553                 if (ARGS_EOLN == ac)
1554                         break;
1555
1556                 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1557                                 ARGS_QWORD != ac && 0 == j &&
1558                                 DELIM_OPEN == mdoc_isdelim(p)) {
1559                         if ( ! dword(mdoc, line, la, p, DELIM_OPEN))
1560                                 return(0);
1561                         continue;
1562                 } else if (0 == j)
1563                        if ( ! mdoc_elem_alloc(mdoc, line, la, tok, arg))
1564                                return(0);
1565
1566                 if (j == maxargs && ! flushed) {
1567                         if ( ! rew_elem(mdoc, tok))
1568                                 return(0);
1569                         flushed = 1;
1570                 }
1571
1572                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1573
1574                 if (MDOC_MAX != ntok) {
1575                         if ( ! flushed && ! rew_elem(mdoc, tok))
1576                                 return(0);
1577                         flushed = 1;
1578                         if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1579                                 return(0);
1580                         j++;
1581                         break;
1582                 }
1583
1584                 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) &&
1585                                 ARGS_QWORD != ac &&
1586                                 ! flushed &&
1587                                 DELIM_NONE != mdoc_isdelim(p)) {
1588                         if ( ! rew_elem(mdoc, tok))
1589                                 return(0);
1590                         flushed = 1;
1591                 }
1592
1593                 if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1594                         return(0);
1595                 j++;
1596         }
1597
1598         if (0 == j && ! mdoc_elem_alloc(mdoc, line, la, tok, arg))
1599                return(0);
1600
1601         /* Close out in a consistent state. */
1602
1603         if ( ! flushed && ! rew_elem(mdoc, tok))
1604                 return(0);
1605         if ( ! nl)
1606                 return(1);
1607         return(append_delims(mdoc, line, pos, buf));
1608 }
1609
1610
1611 static int
1612 in_line_eoln(MACRO_PROT_ARGS)
1613 {
1614         int              la;
1615         enum margserr    ac;
1616         enum margverr    av;
1617         struct mdoc_arg *arg;
1618         char            *p;
1619         enum mdoct       ntok;
1620
1621         assert( ! (MDOC_PARSED & mdoc_macros[tok].flags));
1622
1623         if (tok == MDOC_Pp)
1624                 rew_sub(MDOC_BLOCK, mdoc, MDOC_Nm, line, ppos);
1625
1626         /* Parse macro arguments. */
1627
1628         for (arg = NULL; ; ) {
1629                 la = *pos;
1630                 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf);
1631
1632                 if (ARGV_WORD == av) {
1633                         *pos = la;
1634                         break;
1635                 }
1636                 if (ARGV_EOLN == av)
1637                         break;
1638                 if (ARGV_ARG == av)
1639                         continue;
1640
1641                 mdoc_argv_free(arg);
1642                 return(0);
1643         }
1644
1645         /* Open element scope. */
1646
1647         if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg))
1648                 return(0);
1649
1650         /* Parse argument terms. */
1651
1652         for (;;) {
1653                 la = *pos;
1654                 ac = mdoc_args(mdoc, line, pos, buf, tok, &p);
1655
1656                 if (ARGS_ERROR == ac)
1657                         return(0);
1658                 if (ARGS_EOLN == ac)
1659                         break;
1660
1661                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p);
1662
1663                 if (MDOC_MAX == ntok) {
1664                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1665                                 return(0);
1666                         continue;
1667                 }
1668
1669                 if ( ! rew_elem(mdoc, tok))
1670                         return(0);
1671                 return(mdoc_macro(mdoc, ntok, line, la, pos, buf));
1672         }
1673
1674         /* Close out (no delimiters). */
1675
1676         return(rew_elem(mdoc, tok));
1677 }
1678
1679
1680 /* ARGSUSED */
1681 static int
1682 ctx_synopsis(MACRO_PROT_ARGS)
1683 {
1684         int              nl;
1685
1686         nl = MDOC_NEWLINE & mdoc->flags;
1687
1688         /* If we're not in the SYNOPSIS, go straight to in-line. */
1689         if ( ! (MDOC_SYNOPSIS & mdoc->flags))
1690                 return(in_line(mdoc, tok, line, ppos, pos, buf));
1691
1692         /* If we're a nested call, same place. */
1693         if ( ! nl)
1694                 return(in_line(mdoc, tok, line, ppos, pos, buf));
1695
1696         /*
1697          * XXX: this will open a block scope; however, if later we end
1698          * up formatting the block scope, then child nodes will inherit
1699          * the formatting.  Be careful.
1700          */
1701         if (MDOC_Nm == tok)
1702                 return(blk_full(mdoc, tok, line, ppos, pos, buf));
1703         assert(MDOC_Vt == tok);
1704         return(blk_part_imp(mdoc, tok, line, ppos, pos, buf));
1705 }
1706
1707
1708 /* ARGSUSED */
1709 static int
1710 obsolete(MACRO_PROT_ARGS)
1711 {
1712
1713         mdoc_pmsg(mdoc, line, ppos, MANDOCERR_MACROOBS);
1714         return(1);
1715 }
1716
1717
1718 /*
1719  * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs.
1720  * They're unusual because they're basically free-form text until a
1721  * macro is encountered.
1722  */
1723 static int
1724 phrase(struct mdoc *mdoc, int line, int ppos, char *buf)
1725 {
1726         int              la, pos;
1727         enum margserr    ac;
1728         enum mdoct       ntok;
1729         char            *p;
1730
1731         for (pos = ppos; ; ) {
1732                 la = pos;
1733
1734                 ac = mdoc_zargs(mdoc, line, &pos, buf, &p);
1735
1736                 if (ARGS_ERROR == ac)
1737                         return(0);
1738                 if (ARGS_EOLN == ac)
1739                         break;
1740
1741                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1742
1743                 if (MDOC_MAX == ntok) {
1744                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1745                                 return(0);
1746                         continue;
1747                 }
1748
1749                 if ( ! mdoc_macro(mdoc, ntok, line, la, &pos, buf))
1750                         return(0);
1751                 return(append_delims(mdoc, line, &pos, buf));
1752         }
1753
1754         return(1);
1755 }
1756
1757
1758 /* ARGSUSED */
1759 static int
1760 phrase_ta(MACRO_PROT_ARGS)
1761 {
1762         int               la;
1763         enum mdoct        ntok;
1764         enum margserr     ac;
1765         char             *p;
1766
1767         /*
1768          * FIXME: this is overly restrictive: if the `Ta' is unexpected,
1769          * it should simply error out with ARGSLOST.
1770          */
1771
1772         if ( ! rew_sub(MDOC_BODY, mdoc, MDOC_It, line, ppos))
1773                 return(0);
1774         if ( ! mdoc_body_alloc(mdoc, line, ppos, MDOC_It))
1775                 return(0);
1776
1777         for (;;) {
1778                 la = *pos;
1779                 ac = mdoc_zargs(mdoc, line, pos, buf, &p);
1780
1781                 if (ARGS_ERROR == ac)
1782                         return(0);
1783                 if (ARGS_EOLN == ac)
1784                         break;
1785
1786                 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p);
1787
1788                 if (MDOC_MAX == ntok) {
1789                         if ( ! dword(mdoc, line, la, p, DELIM_MAX))
1790                                 return(0);
1791                         continue;
1792                 }
1793
1794                 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf))
1795                         return(0);
1796                 return(append_delims(mdoc, line, pos, buf));
1797         }
1798
1799         return(1);
1800 }