man_macro.c

   1 /*      $Id: man_macro.c,v 1.43 2010/05/15 20:51:40 kristaps Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
   4  *
   5  * Permission to use, copy, modify, and distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <assert.h>
  22 #include <ctype.h>
  23 #include <stdlib.h>
  24 #include <string.h>
  25
  26 #include "libman.h"
  27
  28 enum    rew {
  29         REW_REWIND,
  30         REW_NOHALT,
  31         REW_HALT
  32 };
  33
  34 static  int              blk_close(MACRO_PROT_ARGS);
  35 static  int              blk_dotted(MACRO_PROT_ARGS);
  36 static  int              blk_exp(MACRO_PROT_ARGS);
  37 static  int              blk_imp(MACRO_PROT_ARGS);
  38 static  int              in_line_eoln(MACRO_PROT_ARGS);
  39
  40 static  int              rew_scope(enum man_type,
  41                                 struct man *, enum mant);
  42 static  enum rew         rew_dohalt(enum mant, enum man_type,
  43                                 const struct man_node *);
  44 static  enum rew         rew_block(enum mant, enum man_type,
  45                                 const struct man_node *);
  46 static  int              rew_warn(struct man *,
  47                                 struct man_node *, enum merr);
  48
  49 const   struct man_macro __man_macros[MAN_MAX] = {
  50         { in_line_eoln, MAN_NSCOPED }, /* br */
  51         { in_line_eoln, 0 }, /* TH */
  52         { blk_imp, MAN_SCOPED }, /* SH */
  53         { blk_imp, MAN_SCOPED }, /* SS */
  54         { blk_imp, MAN_SCOPED | MAN_FSCOPED }, /* TP */
  55         { blk_imp, 0 }, /* LP */
  56         { blk_imp, 0 }, /* PP */
  57         { blk_imp, 0 }, /* P */
  58         { blk_imp, 0 }, /* IP */
  59         { blk_imp, 0 }, /* HP */
  60         { in_line_eoln, MAN_SCOPED }, /* SM */
  61         { in_line_eoln, MAN_SCOPED }, /* SB */
  62         { in_line_eoln, 0 }, /* BI */
  63         { in_line_eoln, 0 }, /* IB */
  64         { in_line_eoln, 0 }, /* BR */
  65         { in_line_eoln, 0 }, /* RB */
  66         { in_line_eoln, MAN_SCOPED }, /* R */
  67         { in_line_eoln, MAN_SCOPED }, /* B */
  68         { in_line_eoln, MAN_SCOPED }, /* I */
  69         { in_line_eoln, 0 }, /* IR */
  70         { in_line_eoln, 0 }, /* RI */
  71         { in_line_eoln, MAN_NSCOPED }, /* na */
  72         { in_line_eoln, 0 }, /* i */
  73         { in_line_eoln, MAN_NSCOPED }, /* sp */
  74         { in_line_eoln, 0 }, /* nf */
  75         { in_line_eoln, 0 }, /* fi */
  76         { in_line_eoln, 0 }, /* r */
  77         { blk_close, 0 }, /* RE */
  78         { blk_exp, MAN_EXPLICIT }, /* RS */
  79         { in_line_eoln, 0 }, /* DT */
  80         { in_line_eoln, 0 }, /* UC */
  81         { in_line_eoln, 0 }, /* PD */
  82         { in_line_eoln, MAN_NSCOPED }, /* Sp */
  83         { in_line_eoln, 0 }, /* Vb */
  84         { in_line_eoln, 0 }, /* Ve */
  85         { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* de */
  86         { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* dei */
  87         { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* am */
  88         { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* ami */
  89         { blk_dotted, 0 }, /* . */
  90 };
  91
  92 const   struct man_macro * const man_macros = __man_macros;
  93
  94
  95 /*
  96  * Warn when "n" is an explicit non-roff macro.
  97  */
  98 static int
  99 rew_warn(struct man *m, struct man_node *n, enum merr er)
 100 {
 101
 102         if (er == WERRMAX || MAN_BLOCK != n->type)
 103                 return(1);
 104         if (MAN_VALID & n->flags)
 105                 return(1);
 106         if ( ! (MAN_EXPLICIT & man_macros[n->tok].flags))
 107                 return(1);
 108         if (MAN_NOCLOSE & man_macros[n->tok].flags)
 109                 return(1);
 110         return(man_nwarn(m, n, er));
 111 }
 112
 113
 114 /*
 115  * Rewind scope.  If a code "er" != WERRMAX has been provided, it will
 116  * be used if an explicit block scope is being closed out.
 117  */
 118 int
 119 man_unscope(struct man *m, const struct man_node *n, enum merr er)
 120 {
 121
 122         assert(n);
 123
 124         /* LINTED */
 125         while (m->last != n) {
 126                 if ( ! rew_warn(m, m->last, er))
 127                         return(0);
 128                 if ( ! man_valid_post(m))
 129                         return(0);
 130                 if ( ! man_action_post(m))
 131                         return(0);
 132                 m->last = m->last->parent;
 133                 assert(m->last);
 134         }
 135
 136         if ( ! rew_warn(m, m->last, er))
 137                 return(0);
 138         if ( ! man_valid_post(m))
 139                 return(0);
 140         if ( ! man_action_post(m))
 141                 return(0);
 142
 143         m->next = MAN_ROOT == m->last->type ?
 144                 MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
 145
 146         return(1);
 147 }
 148
 149
 150 static enum rew
 151 rew_block(enum mant ntok, enum man_type type, const struct man_node *n)
 152 {
 153
 154         if (MAN_BLOCK == type && ntok == n->parent->tok &&
 155                         MAN_BODY == n->parent->type)
 156                 return(REW_REWIND);
 157         return(ntok == n->tok ? REW_HALT : REW_NOHALT);
 158 }
 159
 160
 161 /*
 162  * There are three scope levels: scoped to the root (all), scoped to the
 163  * section (all less sections), and scoped to subsections (all less
 164  * sections and subsections).
 165  */
 166 static enum rew
 167 rew_dohalt(enum mant tok, enum man_type type, const struct man_node *n)
 168 {
 169         enum rew         c;
 170
 171         /* We cannot progress beyond the root ever. */
 172         if (MAN_ROOT == n->type)
 173                 return(REW_HALT);
 174
 175         assert(n->parent);
 176
 177         /* Normal nodes shouldn't go to the level of the root. */
 178         if (MAN_ROOT == n->parent->type)
 179                 return(REW_REWIND);
 180
 181         /* Already-validated nodes should be closed out. */
 182         if (MAN_VALID & n->flags)
 183                 return(REW_NOHALT);
 184
 185         /* First: rewind to ourselves. */
 186         if (type == n->type && tok == n->tok)
 187                 return(REW_REWIND);
 188
 189         /*
 190          * If we're a roff macro, then we can close out anything that
 191          * stands between us and our parent context.
 192          */
 193         if (MAN_NOCLOSE & man_macros[tok].flags)
 194                 return(REW_NOHALT);
 195
 196         /*
 197          * Don't clobber roff macros: this is a bit complicated.  If the
 198          * current macro is a roff macro, halt immediately and don't
 199          * rewind.  If it's not, and the parent is, then close out the
 200          * current scope and halt at the parent.
 201          */
 202         if (MAN_NOCLOSE & man_macros[n->tok].flags)
 203                 return(REW_HALT);
 204         if (MAN_NOCLOSE & man_macros[n->parent->tok].flags)
 205                 return(REW_REWIND);
 206
 207         /*
 208          * Next follow the implicit scope-smashings as defined by man.7:
 209          * section, sub-section, etc.
 210          */
 211
 212         switch (tok) {
 213         case (MAN_SH):
 214                 break;
 215         case (MAN_SS):
 216                 /* Rewind to a section, if a block. */
 217                 if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
 218                         return(c);
 219                 break;
 220         case (MAN_RS):
 221                 /* Rewind to a subsection, if a block. */
 222                 if (REW_NOHALT != (c = rew_block(MAN_SS, type, n)))
 223                         return(c);
 224                 /* Rewind to a section, if a block. */
 225                 if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
 226                         return(c);
 227                 break;
 228         default:
 229                 /* Rewind to an offsetter, if a block. */
 230                 if (REW_NOHALT != (c = rew_block(MAN_RS, type, n)))
 231                         return(c);
 232                 /* Rewind to a subsection, if a block. */
 233                 if (REW_NOHALT != (c = rew_block(MAN_SS, type, n)))
 234                         return(c);
 235                 /* Rewind to a section, if a block. */
 236                 if (REW_NOHALT != (c = rew_block(MAN_SH, type, n)))
 237                         return(c);
 238                 break;
 239         }
 240
 241         return(REW_NOHALT);
 242 }
 243
 244
 245 /*
 246  * Rewinding entails ascending the parse tree until a coherent point,
 247  * for example, the `SH' macro will close out any intervening `SS'
 248  * scopes.  When a scope is closed, it must be validated and actioned.
 249  */
 250 static int
 251 rew_scope(enum man_type type, struct man *m, enum mant tok)
 252 {
 253         struct man_node *n;
 254         enum rew         c;
 255
 256         /* LINTED */
 257         for (n = m->last; n; n = n->parent) {
 258                 /*
 259                  * Whether we should stop immediately (REW_HALT), stop
 260                  * and rewind until this point (REW_REWIND), or keep
 261                  * rewinding (REW_NOHALT).
 262                  */
 263                 c = rew_dohalt(tok, type, n);
 264                 if (REW_HALT == c)
 265                         return(1);
 266                 if (REW_REWIND == c)
 267                         break;
 268         }
 269
 270         /*
 271          * Rewind until the current point.  Warn if we're a roff
 272          * instruction that's mowing over explicit scopes.
 273          */
 274         assert(n);
 275         if (MAN_NOCLOSE & man_macros[tok].flags)
 276                 return(man_unscope(m, n, WROFFSCOPE));
 277
 278         return(man_unscope(m, n, WERRMAX));
 279 }
 280
 281
 282 /*
 283  * Closure for dotted macros (de, dei, am, ami, ign).  This must handle
 284  * any of these as the parent node, so it needs special handling.
 285  * Beyond this, it's the same as blk_close().
 286  */
 287 /* ARGSUSED */
 288 int
 289 blk_dotted(MACRO_PROT_ARGS)
 290 {
 291         enum mant        ntok;
 292         struct man_node *nn;
 293
 294         /* Check for any of the following parents... */
 295
 296         for (nn = m->last->parent; nn; nn = nn->parent)
 297                 if (nn->tok == MAN_de || nn->tok == MAN_dei ||
 298                                 nn->tok == MAN_am ||
 299                                 nn->tok == MAN_ami) {
 300                         ntok = nn->tok;
 301                         break;
 302                 }
 303
 304         if (NULL == nn) {
 305                 if ( ! man_pwarn(m, line, ppos, WNOSCOPE))
 306                         return(0);
 307                 return(1);
 308         }
 309
 310         if ( ! rew_scope(MAN_BODY, m, ntok))
 311                 return(0);
 312         if ( ! rew_scope(MAN_BLOCK, m, ntok))
 313                 return(0);
 314
 315         /*
 316          * Restore flags set when we got here and also stipulate that we
 317          * don't post-process the line when exiting the macro op
 318          * function in man_pmacro().  See blk_exp().
 319          */
 320
 321         m->flags = m->svflags | MAN_ILINE;
 322         m->next = m->svnext;
 323         return(1);
 324 }
 325
 326
 327 /*
 328  * Close out a generic explicit macro.
 329  */
 330 /* ARGSUSED */
 331 int
 332 blk_close(MACRO_PROT_ARGS)
 333 {
 334         enum mant                ntok;
 335         const struct man_node   *nn;
 336
 337         switch (tok) {
 338         case (MAN_RE):
 339                 ntok = MAN_RS;
 340                 break;
 341         default:
 342                 abort();
 343                 /* NOTREACHED */
 344         }
 345
 346         for (nn = m->last->parent; nn; nn = nn->parent)
 347                 if (ntok == nn->tok)
 348                         break;
 349
 350         if (NULL == nn)
 351                 if ( ! man_pwarn(m, line, ppos, WNOSCOPE))
 352                         return(0);
 353
 354         if ( ! rew_scope(MAN_BODY, m, ntok))
 355                 return(0);
 356         if ( ! rew_scope(MAN_BLOCK, m, ntok))
 357                 return(0);
 358
 359         return(1);
 360 }
 361
 362
 363 int
 364 blk_exp(MACRO_PROT_ARGS)
 365 {
 366         int              w, la;
 367         char            *p;
 368
 369         /*
 370          * Close out prior scopes.  "Regular" explicit macros cannot be
 371          * nested, but we allow roff macros to be placed just about
 372          * anywhere.
 373          */
 374
 375         if ( ! (MAN_NOCLOSE & man_macros[tok].flags)) {
 376                 if ( ! rew_scope(MAN_BODY, m, tok))
 377                         return(0);
 378                 if ( ! rew_scope(MAN_BLOCK, m, tok))
 379                         return(0);
 380         } else {
 381                 /*
 382                  * Save our state and next-scope indicator; we restore
 383                  * it when exiting from the roff instruction block.  See
 384                  * blk_dotted().
 385                  */
 386                 m->svflags = m->flags;
 387                 m->svnext = m->next;
 388
 389                 /* Make sure we drop any line modes. */
 390                 m->flags = 0;
 391         }
 392
 393         if ( ! man_block_alloc(m, line, ppos, tok))
 394                 return(0);
 395         if ( ! man_head_alloc(m, line, ppos, tok))
 396                 return(0);
 397
 398         for (;;) {
 399                 la = *pos;
 400                 w = man_args(m, line, pos, buf, &p);
 401
 402                 if (-1 == w)
 403                         return(0);
 404                 if (0 == w)
 405                         break;
 406
 407                 if ( ! man_word_alloc(m, line, la, p))
 408                         return(0);
 409         }
 410
 411         assert(m);
 412         assert(tok != MAN_MAX);
 413
 414         if ( ! rew_scope(MAN_HEAD, m, tok))
 415                 return(0);
 416         return(man_body_alloc(m, line, ppos, tok));
 417 }
 418
 419
 420
 421 /*
 422  * Parse an implicit-block macro.  These contain a MAN_HEAD and a
 423  * MAN_BODY contained within a MAN_BLOCK.  Rules for closing out other
 424  * scopes, such as `SH' closing out an `SS', are defined in the rew
 425  * routines.
 426  */
 427 int
 428 blk_imp(MACRO_PROT_ARGS)
 429 {
 430         int              w, la;
 431         char            *p;
 432         struct man_node *n;
 433
 434         /* Close out prior scopes. */
 435
 436         if ( ! rew_scope(MAN_BODY, m, tok))
 437                 return(0);
 438         if ( ! rew_scope(MAN_BLOCK, m, tok))
 439                 return(0);
 440
 441         /* Allocate new block & head scope. */
 442
 443         if ( ! man_block_alloc(m, line, ppos, tok))
 444                 return(0);
 445         if ( ! man_head_alloc(m, line, ppos, tok))
 446                 return(0);
 447
 448         n = m->last;
 449
 450         /* Add line arguments. */
 451
 452         for (;;) {
 453                 la = *pos;
 454                 w = man_args(m, line, pos, buf, &p);
 455
 456                 if (-1 == w)
 457                         return(0);
 458                 if (0 == w)
 459                         break;
 460
 461                 if ( ! man_word_alloc(m, line, la, p))
 462                         return(0);
 463         }
 464
 465         /* Close out head and open body (unless MAN_SCOPE). */
 466
 467         if (MAN_SCOPED & man_macros[tok].flags) {
 468                 /* If we're forcing scope (`TP'), keep it open. */
 469                 if (MAN_FSCOPED & man_macros[tok].flags) {
 470                         m->flags |= MAN_BLINE;
 471                         return(1);
 472                 } else if (n == m->last) {
 473                         m->flags |= MAN_BLINE;
 474                         return(1);
 475                 }
 476         }
 477
 478         if ( ! rew_scope(MAN_HEAD, m, tok))
 479                 return(0);
 480         return(man_body_alloc(m, line, ppos, tok));
 481 }
 482
 483
 484 int
 485 in_line_eoln(MACRO_PROT_ARGS)
 486 {
 487         int              w, la;
 488         char            *p;
 489         struct man_node *n;
 490
 491         if ( ! man_elem_alloc(m, line, ppos, tok))
 492                 return(0);
 493
 494         n = m->last;
 495
 496         for (;;) {
 497                 la = *pos;
 498                 w = man_args(m, line, pos, buf, &p);
 499
 500                 if (-1 == w)
 501                         return(0);
 502                 if (0 == w)
 503                         break;
 504                 if ( ! man_word_alloc(m, line, la, p))
 505                         return(0);
 506         }
 507
 508         /*
 509          * If no arguments are specified and this is MAN_SCOPED (i.e.,
 510          * next-line scoped), then set our mode to indicate that we're
 511          * waiting for terms to load into our context.
 512          */
 513
 514         if (n == m->last && MAN_SCOPED & man_macros[tok].flags) {
 515                 assert( ! (MAN_NSCOPED & man_macros[tok].flags));
 516                 m->flags |= MAN_ELINE;
 517                 return(1);
 518         }
 519
 520         /* Set ignorable context, if applicable. */
 521
 522         if (MAN_NSCOPED & man_macros[tok].flags) {
 523                 assert( ! (MAN_SCOPED & man_macros[tok].flags));
 524                 m->flags |= MAN_ILINE;
 525         }
 526
 527         /*
 528          * Rewind our element scope.  Note that when TH is pruned, we'll
 529          * be back at the root, so make sure that we don't clobber as
 530          * its sibling.
 531          */
 532
 533         for ( ; m->last; m->last = m->last->parent) {
 534                 if (m->last == n)
 535                         break;
 536                 if (m->last->type == MAN_ROOT)
 537                         break;
 538                 if ( ! man_valid_post(m))
 539                         return(0);
 540                 if ( ! man_action_post(m))
 541                         return(0);
 542         }
 543
 544         assert(m->last);
 545
 546         /*
 547          * Same here regarding whether we're back at the root.
 548          */
 549
 550         if (m->last->type != MAN_ROOT && ! man_valid_post(m))
 551                 return(0);
 552         if (m->last->type != MAN_ROOT && ! man_action_post(m))
 553                 return(0);
 554
 555         m->next = MAN_ROOT == m->last->type ?
 556                 MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
 557
 558         return(1);
 559 }
 560
 561
 562 int
 563 man_macroend(struct man *m)
 564 {
 565
 566         return(man_unscope(m, m->first, WEXITSCOPE));
 567 }
 568