man.c

   1 /*      $Id: man.c,v 1.128 2014/03/30 19:47:48 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
   5  * Copyright (c) 2011 Joerg Sonnenberger <joerg@netbsd.org>
   6  *
   7  * Permission to use, copy, modify, and distribute this software for any
   8  * purpose with or without fee is hereby granted, provided that the above
   9  * copyright notice and this permission notice appear in all copies.
  10  *
  11  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  17  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  18  */
  19 #ifdef HAVE_CONFIG_H
  20 #include "config.h"
  21 #endif
  22
  23 #include <sys/types.h>
  24
  25 #include <assert.h>
  26 #include <ctype.h>
  27 #include <stdarg.h>
  28 #include <stdlib.h>
  29 #include <stdio.h>
  30 #include <string.h>
  31
  32 #include "man.h"
  33 #include "mandoc.h"
  34 #include "mandoc_aux.h"
  35 #include "libman.h"
  36 #include "libmandoc.h"
  37
  38 const   char *const __man_macronames[MAN_MAX] = {
  39         "br",           "TH",           "SH",           "SS",
  40         "TP",           "LP",           "PP",           "P",
  41         "IP",           "HP",           "SM",           "SB",
  42         "BI",           "IB",           "BR",           "RB",
  43         "R",            "B",            "I",            "IR",
  44         "RI",           "na",           "sp",           "nf",
  45         "fi",           "RE",           "RS",           "DT",
  46         "UC",           "PD",           "AT",           "in",
  47         "ft",           "OP",           "EX",           "EE",
  48         "UR",           "UE",           "ll"
  49         };
  50
  51 const   char * const *man_macronames = __man_macronames;
  52
  53 static  struct man_node *man_node_alloc(struct man *, int, int,
  54                                 enum man_type, enum mant);
  55 static  int              man_node_append(struct man *,
  56                                 struct man_node *);
  57 static  void             man_node_free(struct man_node *);
  58 static  void             man_node_unlink(struct man *,
  59                                 struct man_node *);
  60 static  int              man_ptext(struct man *, int, char *, int);
  61 static  int              man_pmacro(struct man *, int, char *, int);
  62 static  void             man_free1(struct man *);
  63 static  void             man_alloc1(struct man *);
  64 static  int              man_descope(struct man *, int, int);
  65
  66
  67 const struct man_node *
  68 man_node(const struct man *man)
  69 {
  70
  71         assert( ! (MAN_HALT & man->flags));
  72         return(man->first);
  73 }
  74
  75
  76 const struct man_meta *
  77 man_meta(const struct man *man)
  78 {
  79
  80         assert( ! (MAN_HALT & man->flags));
  81         return(&man->meta);
  82 }
  83
  84
  85 void
  86 man_reset(struct man *man)
  87 {
  88
  89         man_free1(man);
  90         man_alloc1(man);
  91 }
  92
  93
  94 void
  95 man_free(struct man *man)
  96 {
  97
  98         man_free1(man);
  99         free(man);
 100 }
 101
 102
 103 struct man *
 104 man_alloc(struct roff *roff, struct mparse *parse, int quick)
 105 {
 106         struct man      *p;
 107
 108         p = mandoc_calloc(1, sizeof(struct man));
 109
 110         man_hash_init();
 111         p->parse = parse;
 112         p->quick = quick;
 113         p->roff = roff;
 114
 115         man_alloc1(p);
 116         return(p);
 117 }
 118
 119
 120 int
 121 man_endparse(struct man *man)
 122 {
 123
 124         assert( ! (MAN_HALT & man->flags));
 125         if (man_macroend(man))
 126                 return(1);
 127         man->flags |= MAN_HALT;
 128         return(0);
 129 }
 130
 131
 132 int
 133 man_parseln(struct man *man, int ln, char *buf, int offs)
 134 {
 135
 136         man->flags |= MAN_NEWLINE;
 137
 138         assert( ! (MAN_HALT & man->flags));
 139
 140         return (roff_getcontrol(man->roff, buf, &offs) ?
 141                         man_pmacro(man, ln, buf, offs) :
 142                         man_ptext(man, ln, buf, offs));
 143 }
 144
 145
 146 static void
 147 man_free1(struct man *man)
 148 {
 149
 150         if (man->first)
 151                 man_node_delete(man, man->first);
 152         if (man->meta.title)
 153                 free(man->meta.title);
 154         if (man->meta.source)
 155                 free(man->meta.source);
 156         if (man->meta.date)
 157                 free(man->meta.date);
 158         if (man->meta.vol)
 159                 free(man->meta.vol);
 160         if (man->meta.msec)
 161                 free(man->meta.msec);
 162 }
 163
 164
 165 static void
 166 man_alloc1(struct man *man)
 167 {
 168
 169         memset(&man->meta, 0, sizeof(struct man_meta));
 170         man->flags = 0;
 171         man->last = mandoc_calloc(1, sizeof(struct man_node));
 172         man->first = man->last;
 173         man->last->type = MAN_ROOT;
 174         man->last->tok = MAN_MAX;
 175         man->next = MAN_NEXT_CHILD;
 176 }
 177
 178
 179 static int
 180 man_node_append(struct man *man, struct man_node *p)
 181 {
 182
 183         assert(man->last);
 184         assert(man->first);
 185         assert(MAN_ROOT != p->type);
 186
 187         switch (man->next) {
 188         case (MAN_NEXT_SIBLING):
 189                 man->last->next = p;
 190                 p->prev = man->last;
 191                 p->parent = man->last->parent;
 192                 break;
 193         case (MAN_NEXT_CHILD):
 194                 man->last->child = p;
 195                 p->parent = man->last;
 196                 break;
 197         default:
 198                 abort();
 199                 /* NOTREACHED */
 200         }
 201
 202         assert(p->parent);
 203         p->parent->nchild++;
 204
 205         if ( ! man_valid_pre(man, p))
 206                 return(0);
 207
 208         switch (p->type) {
 209         case (MAN_HEAD):
 210                 assert(MAN_BLOCK == p->parent->type);
 211                 p->parent->head = p;
 212                 break;
 213         case (MAN_TAIL):
 214                 assert(MAN_BLOCK == p->parent->type);
 215                 p->parent->tail = p;
 216                 break;
 217         case (MAN_BODY):
 218                 assert(MAN_BLOCK == p->parent->type);
 219                 p->parent->body = p;
 220                 break;
 221         default:
 222                 break;
 223         }
 224
 225         man->last = p;
 226
 227         switch (p->type) {
 228         case (MAN_TBL):
 229                 /* FALLTHROUGH */
 230         case (MAN_TEXT):
 231                 if ( ! man_valid_post(man))
 232                         return(0);
 233                 break;
 234         default:
 235                 break;
 236         }
 237
 238         return(1);
 239 }
 240
 241
 242 static struct man_node *
 243 man_node_alloc(struct man *man, int line, int pos,
 244                 enum man_type type, enum mant tok)
 245 {
 246         struct man_node *p;
 247
 248         p = mandoc_calloc(1, sizeof(struct man_node));
 249         p->line = line;
 250         p->pos = pos;
 251         p->type = type;
 252         p->tok = tok;
 253
 254         if (MAN_NEWLINE & man->flags)
 255                 p->flags |= MAN_LINE;
 256         man->flags &= ~MAN_NEWLINE;
 257         return(p);
 258 }
 259
 260
 261 int
 262 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
 263 {
 264         struct man_node *p;
 265
 266         p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
 267         if ( ! man_node_append(man, p))
 268                 return(0);
 269         man->next = MAN_NEXT_CHILD;
 270         return(1);
 271 }
 272
 273
 274 int
 275 man_tail_alloc(struct man *man, int line, int pos, enum mant tok)
 276 {
 277         struct man_node *p;
 278
 279         p = man_node_alloc(man, line, pos, MAN_TAIL, tok);
 280         if ( ! man_node_append(man, p))
 281                 return(0);
 282         man->next = MAN_NEXT_CHILD;
 283         return(1);
 284 }
 285
 286
 287 int
 288 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 289 {
 290         struct man_node *p;
 291
 292         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 293         if ( ! man_node_append(man, p))
 294                 return(0);
 295         man->next = MAN_NEXT_CHILD;
 296         return(1);
 297 }
 298
 299
 300 int
 301 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
 302 {
 303         struct man_node *p;
 304
 305         p = man_node_alloc(man, line, pos, MAN_BODY, tok);
 306         if ( ! man_node_append(man, p))
 307                 return(0);
 308         man->next = MAN_NEXT_CHILD;
 309         return(1);
 310 }
 311
 312
 313 int
 314 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 315 {
 316         struct man_node *p;
 317
 318         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 319         if ( ! man_node_append(man, p))
 320                 return(0);
 321         man->next = MAN_NEXT_CHILD;
 322         return(1);
 323 }
 324
 325 int
 326 man_word_alloc(struct man *man, int line, int pos, const char *word)
 327 {
 328         struct man_node *n;
 329
 330         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 331         n->string = roff_strdup(man->roff, word);
 332
 333         if ( ! man_node_append(man, n))
 334                 return(0);
 335
 336         man->next = MAN_NEXT_SIBLING;
 337         return(1);
 338 }
 339
 340
 341 /*
 342  * Free all of the resources held by a node.  This does NOT unlink a
 343  * node from its context; for that, see man_node_unlink().
 344  */
 345 static void
 346 man_node_free(struct man_node *p)
 347 {
 348
 349         if (p->string)
 350                 free(p->string);
 351         free(p);
 352 }
 353
 354
 355 void
 356 man_node_delete(struct man *man, struct man_node *p)
 357 {
 358
 359         while (p->child)
 360                 man_node_delete(man, p->child);
 361
 362         man_node_unlink(man, p);
 363         man_node_free(p);
 364 }
 365
 366 int
 367 man_addeqn(struct man *man, const struct eqn *ep)
 368 {
 369         struct man_node *n;
 370
 371         assert( ! (MAN_HALT & man->flags));
 372
 373         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 374         n->eqn = ep;
 375
 376         if ( ! man_node_append(man, n))
 377                 return(0);
 378
 379         man->next = MAN_NEXT_SIBLING;
 380         return(man_descope(man, ep->ln, ep->pos));
 381 }
 382
 383 int
 384 man_addspan(struct man *man, const struct tbl_span *sp)
 385 {
 386         struct man_node *n;
 387
 388         assert( ! (MAN_HALT & man->flags));
 389
 390         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 391         n->span = sp;
 392
 393         if ( ! man_node_append(man, n))
 394                 return(0);
 395
 396         man->next = MAN_NEXT_SIBLING;
 397         return(man_descope(man, sp->line, 0));
 398 }
 399
 400 static int
 401 man_descope(struct man *man, int line, int offs)
 402 {
 403         /*
 404          * Co-ordinate what happens with having a next-line scope open:
 405          * first close out the element scope (if applicable), then close
 406          * out the block scope (also if applicable).
 407          */
 408
 409         if (MAN_ELINE & man->flags) {
 410                 man->flags &= ~MAN_ELINE;
 411                 if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 412                         return(0);
 413         }
 414
 415         if ( ! (MAN_BLINE & man->flags))
 416                 return(1);
 417         man->flags &= ~MAN_BLINE;
 418
 419         if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 420                 return(0);
 421         return(man_body_alloc(man, line, offs, man->last->tok));
 422 }
 423
 424 static int
 425 man_ptext(struct man *man, int line, char *buf, int offs)
 426 {
 427         int              i;
 428
 429         /* Literal free-form text whitespace is preserved. */
 430
 431         if (MAN_LITERAL & man->flags) {
 432                 if ( ! man_word_alloc(man, line, offs, buf + offs))
 433                         return(0);
 434                 return(man_descope(man, line, offs));
 435         }
 436
 437         for (i = offs; ' ' == buf[i]; i++)
 438                 /* Skip leading whitespace. */ ;
 439
 440         /*
 441          * Blank lines are ignored right after headings
 442          * but add a single vertical space elsewhere.
 443          */
 444
 445         if ('\0' == buf[i]) {
 446                 /* Allocate a blank entry. */
 447                 if (MAN_SH != man->last->tok &&
 448                     MAN_SS != man->last->tok) {
 449                         if ( ! man_elem_alloc(man, line, offs, MAN_sp))
 450                                 return(0);
 451                         man->next = MAN_NEXT_SIBLING;
 452                 }
 453                 return(1);
 454         }
 455
 456         /*
 457          * Warn if the last un-escaped character is whitespace. Then
 458          * strip away the remaining spaces (tabs stay!).
 459          */
 460
 461         i = (int)strlen(buf);
 462         assert(i);
 463
 464         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 465                 if (i > 1 && '\\' != buf[i - 2])
 466                         man_pmsg(man, line, i - 1, MANDOCERR_EOLNSPACE);
 467
 468                 for (--i; i && ' ' == buf[i]; i--)
 469                         /* Spin back to non-space. */ ;
 470
 471                 /* Jump ahead of escaped whitespace. */
 472                 i += '\\' == buf[i] ? 2 : 1;
 473
 474                 buf[i] = '\0';
 475         }
 476
 477         if ( ! man_word_alloc(man, line, offs, buf + offs))
 478                 return(0);
 479
 480         /*
 481          * End-of-sentence check.  If the last character is an unescaped
 482          * EOS character, then flag the node as being the end of a
 483          * sentence.  The front-end will know how to interpret this.
 484          */
 485
 486         assert(i);
 487         if (mandoc_eos(buf, (size_t)i))
 488                 man->last->flags |= MAN_EOS;
 489
 490         return(man_descope(man, line, offs));
 491 }
 492
 493 static int
 494 man_pmacro(struct man *man, int ln, char *buf, int offs)
 495 {
 496         int              i, ppos;
 497         enum mant        tok;
 498         char             mac[5];
 499         struct man_node *n;
 500
 501         if ('"' == buf[offs]) {
 502                 man_pmsg(man, ln, offs, MANDOCERR_BADCOMMENT);
 503                 return(1);
 504         } else if ('\0' == buf[offs])
 505                 return(1);
 506
 507         ppos = offs;
 508
 509         /*
 510          * Copy the first word into a nil-terminated buffer.
 511          * Stop copying when a tab, space, or eoln is encountered.
 512          */
 513
 514         i = 0;
 515         while (i < 4 && '\0' != buf[offs] &&
 516                         ' ' != buf[offs] && '\t' != buf[offs])
 517                 mac[i++] = buf[offs++];
 518
 519         mac[i] = '\0';
 520
 521         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 522
 523         if (MAN_MAX == tok) {
 524                 mandoc_vmsg(MANDOCERR_MACRO, man->parse, ln,
 525                                 ppos, "%s", buf + ppos - 1);
 526                 return(1);
 527         }
 528
 529         /* The macro is sane.  Jump to the next word. */
 530
 531         while (buf[offs] && ' ' == buf[offs])
 532                 offs++;
 533
 534         /*
 535          * Trailing whitespace.  Note that tabs are allowed to be passed
 536          * into the parser as "text", so we only warn about spaces here.
 537          */
 538
 539         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 540                 man_pmsg(man, ln, offs - 1, MANDOCERR_EOLNSPACE);
 541
 542         /*
 543          * Remove prior ELINE macro, as it's being clobbered by a new
 544          * macro.  Note that NSCOPED macros do not close out ELINE
 545          * macros---they don't print text---so we let those slip by.
 546          */
 547
 548         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 549                         man->flags & MAN_ELINE) {
 550                 n = man->last;
 551                 assert(MAN_TEXT != n->type);
 552
 553                 /* Remove repeated NSCOPED macros causing ELINE. */
 554
 555                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 556                         n = n->parent;
 557
 558                 mandoc_vmsg(MANDOCERR_LINESCOPE, man->parse, n->line,
 559                     n->pos, "%s breaks %s", man_macronames[tok],
 560                     man_macronames[n->tok]);
 561
 562                 man_node_delete(man, n);
 563                 man->flags &= ~MAN_ELINE;
 564         }
 565
 566         /*
 567          * Remove prior BLINE macro that is being clobbered.
 568          */
 569         if ((man->flags & MAN_BLINE) &&
 570             (MAN_BSCOPE & man_macros[tok].flags)) {
 571                 n = man->last;
 572
 573                 /* Might be a text node like 8 in
 574                  * .TP 8
 575                  * .SH foo
 576                  */
 577                 if (MAN_TEXT == n->type)
 578                         n = n->parent;
 579
 580                 /* Remove element that didn't end BLINE, if any. */
 581                 if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
 582                         n = n->parent;
 583
 584                 assert(MAN_HEAD == n->type);
 585                 n = n->parent;
 586                 assert(MAN_BLOCK == n->type);
 587                 assert(MAN_SCOPED & man_macros[n->tok].flags);
 588
 589                 mandoc_vmsg(MANDOCERR_LINESCOPE, man->parse, n->line,
 590                     n->pos, "%s breaks %s", man_macronames[tok],
 591                     man_macronames[n->tok]);
 592
 593                 man_node_delete(man, n);
 594                 man->flags &= ~MAN_BLINE;
 595         }
 596
 597         /*
 598          * Save the fact that we're in the next-line for a block.  In
 599          * this way, embedded roff instructions can "remember" state
 600          * when they exit.
 601          */
 602
 603         if (MAN_BLINE & man->flags)
 604                 man->flags |= MAN_BPLINE;
 605
 606         /* Call to handler... */
 607
 608         assert(man_macros[tok].fp);
 609         if ( ! (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf))
 610                 goto err;
 611
 612         /* In quick mode (for mandocdb), abort after the NAME section. */
 613
 614         if (man->quick && MAN_SH == tok &&
 615             strcmp(man->last->prev->child->string, "NAME"))
 616                 return(2);
 617
 618         /*
 619          * We weren't in a block-line scope when entering the
 620          * above-parsed macro, so return.
 621          */
 622
 623         if ( ! (MAN_BPLINE & man->flags)) {
 624                 man->flags &= ~MAN_ILINE;
 625                 return(1);
 626         }
 627         man->flags &= ~MAN_BPLINE;
 628
 629         /*
 630          * If we're in a block scope, then allow this macro to slip by
 631          * without closing scope around it.
 632          */
 633
 634         if (MAN_ILINE & man->flags) {
 635                 man->flags &= ~MAN_ILINE;
 636                 return(1);
 637         }
 638
 639         /*
 640          * If we've opened a new next-line element scope, then return
 641          * now, as the next line will close out the block scope.
 642          */
 643
 644         if (MAN_ELINE & man->flags)
 645                 return(1);
 646
 647         /* Close out the block scope opened in the prior line.  */
 648
 649         assert(MAN_BLINE & man->flags);
 650         man->flags &= ~MAN_BLINE;
 651
 652         if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 653                 return(0);
 654         return(man_body_alloc(man, ln, ppos, man->last->tok));
 655
 656 err:    /* Error out. */
 657
 658         man->flags |= MAN_HALT;
 659         return(0);
 660 }
 661
 662 /*
 663  * Unlink a node from its context.  If "man" is provided, the last parse
 664  * point will also be adjusted accordingly.
 665  */
 666 static void
 667 man_node_unlink(struct man *man, struct man_node *n)
 668 {
 669
 670         /* Adjust siblings. */
 671
 672         if (n->prev)
 673                 n->prev->next = n->next;
 674         if (n->next)
 675                 n->next->prev = n->prev;
 676
 677         /* Adjust parent. */
 678
 679         if (n->parent) {
 680                 n->parent->nchild--;
 681                 if (n->parent->child == n)
 682                         n->parent->child = n->prev ? n->prev : n->next;
 683         }
 684
 685         /* Adjust parse point, if applicable. */
 686
 687         if (man && man->last == n) {
 688                 /*XXX: this can occur when bailing from validation. */
 689                 /*assert(NULL == n->next);*/
 690                 if (n->prev) {
 691                         man->last = n->prev;
 692                         man->next = MAN_NEXT_SIBLING;
 693                 } else {
 694                         man->last = n->parent;
 695                         man->next = MAN_NEXT_CHILD;
 696                 }
 697         }
 698
 699         if (man && man->first == n)
 700                 man->first = NULL;
 701 }
 702
 703 const struct mparse *
 704 man_mparse(const struct man *man)
 705 {
 706
 707         assert(man && man->parse);
 708         return(man->parse);
 709 }
 710
 711 void
 712 man_deroff(char **dest, const struct man_node *n)
 713 {
 714         char    *cp;
 715         size_t   sz;
 716
 717         if (MAN_TEXT != n->type) {
 718                 for (n = n->child; n; n = n->next)
 719                         man_deroff(dest, n);
 720                 return;
 721         }
 722
 723         /* Skip leading whitespace and escape sequences. */
 724
 725         cp = n->string;
 726         while ('\0' != *cp) {
 727                 if ('\\' == *cp) {
 728                         cp++;
 729                         mandoc_escape((const char **)&cp, NULL, NULL);
 730                 } else if (isspace((unsigned char)*cp))
 731                         cp++;
 732                 else
 733                         break;
 734         }
 735
 736         /* Skip trailing whitespace. */
 737
 738         for (sz = strlen(cp); sz; sz--)
 739                 if (0 == isspace((unsigned char)cp[sz-1]))
 740                         break;
 741
 742         /* Skip empty strings. */
 743
 744         if (0 == sz)
 745                 return;
 746
 747         if (NULL == *dest) {
 748                 *dest = mandoc_strndup(cp, sz);
 749                 return;
 750         }
 751
 752         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 753         free(*dest);
 754         *dest = cp;
 755 }