man.c

   1 /*      $Id: man.c,v 1.129 2014/04/20 16:46:04 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 const struct man_meta *
  76 man_meta(const struct man *man)
  77 {
  78
  79         assert( ! (MAN_HALT & man->flags));
  80         return(&man->meta);
  81 }
  82
  83 void
  84 man_reset(struct man *man)
  85 {
  86
  87         man_free1(man);
  88         man_alloc1(man);
  89 }
  90
  91 void
  92 man_free(struct man *man)
  93 {
  94
  95         man_free1(man);
  96         free(man);
  97 }
  98
  99 struct man *
 100 man_alloc(struct roff *roff, struct mparse *parse, int quick)
 101 {
 102         struct man      *p;
 103
 104         p = mandoc_calloc(1, sizeof(struct man));
 105
 106         man_hash_init();
 107         p->parse = parse;
 108         p->quick = quick;
 109         p->roff = roff;
 110
 111         man_alloc1(p);
 112         return(p);
 113 }
 114
 115 int
 116 man_endparse(struct man *man)
 117 {
 118
 119         assert( ! (MAN_HALT & man->flags));
 120         if (man_macroend(man))
 121                 return(1);
 122         man->flags |= MAN_HALT;
 123         return(0);
 124 }
 125
 126 int
 127 man_parseln(struct man *man, int ln, char *buf, int offs)
 128 {
 129
 130         man->flags |= MAN_NEWLINE;
 131
 132         assert( ! (MAN_HALT & man->flags));
 133
 134         return (roff_getcontrol(man->roff, buf, &offs) ?
 135             man_pmacro(man, ln, buf, offs) :
 136             man_ptext(man, ln, buf, offs));
 137 }
 138
 139 static void
 140 man_free1(struct man *man)
 141 {
 142
 143         if (man->first)
 144                 man_node_delete(man, man->first);
 145         if (man->meta.title)
 146                 free(man->meta.title);
 147         if (man->meta.source)
 148                 free(man->meta.source);
 149         if (man->meta.date)
 150                 free(man->meta.date);
 151         if (man->meta.vol)
 152                 free(man->meta.vol);
 153         if (man->meta.msec)
 154                 free(man->meta.msec);
 155 }
 156
 157 static void
 158 man_alloc1(struct man *man)
 159 {
 160
 161         memset(&man->meta, 0, sizeof(struct man_meta));
 162         man->flags = 0;
 163         man->last = mandoc_calloc(1, sizeof(struct man_node));
 164         man->first = man->last;
 165         man->last->type = MAN_ROOT;
 166         man->last->tok = MAN_MAX;
 167         man->next = MAN_NEXT_CHILD;
 168 }
 169
 170
 171 static int
 172 man_node_append(struct man *man, struct man_node *p)
 173 {
 174
 175         assert(man->last);
 176         assert(man->first);
 177         assert(MAN_ROOT != p->type);
 178
 179         switch (man->next) {
 180         case MAN_NEXT_SIBLING:
 181                 man->last->next = p;
 182                 p->prev = man->last;
 183                 p->parent = man->last->parent;
 184                 break;
 185         case MAN_NEXT_CHILD:
 186                 man->last->child = p;
 187                 p->parent = man->last;
 188                 break;
 189         default:
 190                 abort();
 191                 /* NOTREACHED */
 192         }
 193
 194         assert(p->parent);
 195         p->parent->nchild++;
 196
 197         if ( ! man_valid_pre(man, p))
 198                 return(0);
 199
 200         switch (p->type) {
 201         case MAN_HEAD:
 202                 assert(MAN_BLOCK == p->parent->type);
 203                 p->parent->head = p;
 204                 break;
 205         case MAN_TAIL:
 206                 assert(MAN_BLOCK == p->parent->type);
 207                 p->parent->tail = p;
 208                 break;
 209         case MAN_BODY:
 210                 assert(MAN_BLOCK == p->parent->type);
 211                 p->parent->body = p;
 212                 break;
 213         default:
 214                 break;
 215         }
 216
 217         man->last = p;
 218
 219         switch (p->type) {
 220         case MAN_TBL:
 221                 /* FALLTHROUGH */
 222         case MAN_TEXT:
 223                 if ( ! man_valid_post(man))
 224                         return(0);
 225                 break;
 226         default:
 227                 break;
 228         }
 229
 230         return(1);
 231 }
 232
 233 static struct man_node *
 234 man_node_alloc(struct man *man, int line, int pos,
 235                 enum man_type type, enum mant tok)
 236 {
 237         struct man_node *p;
 238
 239         p = mandoc_calloc(1, sizeof(struct man_node));
 240         p->line = line;
 241         p->pos = pos;
 242         p->type = type;
 243         p->tok = tok;
 244
 245         if (MAN_NEWLINE & man->flags)
 246                 p->flags |= MAN_LINE;
 247         man->flags &= ~MAN_NEWLINE;
 248         return(p);
 249 }
 250
 251 int
 252 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
 253 {
 254         struct man_node *p;
 255
 256         p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
 257         if ( ! man_node_append(man, p))
 258                 return(0);
 259         man->next = MAN_NEXT_CHILD;
 260         return(1);
 261 }
 262
 263 int
 264 man_tail_alloc(struct man *man, int line, int pos, enum mant tok)
 265 {
 266         struct man_node *p;
 267
 268         p = man_node_alloc(man, line, pos, MAN_TAIL, tok);
 269         if ( ! man_node_append(man, p))
 270                 return(0);
 271         man->next = MAN_NEXT_CHILD;
 272         return(1);
 273 }
 274
 275 int
 276 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 277 {
 278         struct man_node *p;
 279
 280         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 281         if ( ! man_node_append(man, p))
 282                 return(0);
 283         man->next = MAN_NEXT_CHILD;
 284         return(1);
 285 }
 286
 287 int
 288 man_body_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_BODY, tok);
 293         if ( ! man_node_append(man, p))
 294                 return(0);
 295         man->next = MAN_NEXT_CHILD;
 296         return(1);
 297 }
 298
 299 int
 300 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 301 {
 302         struct man_node *p;
 303
 304         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 305         if ( ! man_node_append(man, p))
 306                 return(0);
 307         man->next = MAN_NEXT_CHILD;
 308         return(1);
 309 }
 310
 311 int
 312 man_word_alloc(struct man *man, int line, int pos, const char *word)
 313 {
 314         struct man_node *n;
 315
 316         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 317         n->string = roff_strdup(man->roff, word);
 318
 319         if ( ! man_node_append(man, n))
 320                 return(0);
 321
 322         man->next = MAN_NEXT_SIBLING;
 323         return(1);
 324 }
 325
 326 /*
 327  * Free all of the resources held by a node.  This does NOT unlink a
 328  * node from its context; for that, see man_node_unlink().
 329  */
 330 static void
 331 man_node_free(struct man_node *p)
 332 {
 333
 334         if (p->string)
 335                 free(p->string);
 336         free(p);
 337 }
 338
 339 void
 340 man_node_delete(struct man *man, struct man_node *p)
 341 {
 342
 343         while (p->child)
 344                 man_node_delete(man, p->child);
 345
 346         man_node_unlink(man, p);
 347         man_node_free(p);
 348 }
 349
 350 int
 351 man_addeqn(struct man *man, const struct eqn *ep)
 352 {
 353         struct man_node *n;
 354
 355         assert( ! (MAN_HALT & man->flags));
 356
 357         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 358         n->eqn = ep;
 359
 360         if ( ! man_node_append(man, n))
 361                 return(0);
 362
 363         man->next = MAN_NEXT_SIBLING;
 364         return(man_descope(man, ep->ln, ep->pos));
 365 }
 366
 367 int
 368 man_addspan(struct man *man, const struct tbl_span *sp)
 369 {
 370         struct man_node *n;
 371
 372         assert( ! (MAN_HALT & man->flags));
 373
 374         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 375         n->span = sp;
 376
 377         if ( ! man_node_append(man, n))
 378                 return(0);
 379
 380         man->next = MAN_NEXT_SIBLING;
 381         return(man_descope(man, sp->line, 0));
 382 }
 383
 384 static int
 385 man_descope(struct man *man, int line, int offs)
 386 {
 387         /*
 388          * Co-ordinate what happens with having a next-line scope open:
 389          * first close out the element scope (if applicable), then close
 390          * out the block scope (also if applicable).
 391          */
 392
 393         if (MAN_ELINE & man->flags) {
 394                 man->flags &= ~MAN_ELINE;
 395                 if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 396                         return(0);
 397         }
 398
 399         if ( ! (MAN_BLINE & man->flags))
 400                 return(1);
 401         man->flags &= ~MAN_BLINE;
 402
 403         if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 404                 return(0);
 405         return(man_body_alloc(man, line, offs, man->last->tok));
 406 }
 407
 408 static int
 409 man_ptext(struct man *man, int line, char *buf, int offs)
 410 {
 411         int              i;
 412
 413         /* Literal free-form text whitespace is preserved. */
 414
 415         if (MAN_LITERAL & man->flags) {
 416                 if ( ! man_word_alloc(man, line, offs, buf + offs))
 417                         return(0);
 418                 return(man_descope(man, line, offs));
 419         }
 420
 421         for (i = offs; ' ' == buf[i]; i++)
 422                 /* Skip leading whitespace. */ ;
 423
 424         /*
 425          * Blank lines are ignored right after headings
 426          * but add a single vertical space elsewhere.
 427          */
 428
 429         if ('\0' == buf[i]) {
 430                 /* Allocate a blank entry. */
 431                 if (MAN_SH != man->last->tok &&
 432                     MAN_SS != man->last->tok) {
 433                         if ( ! man_elem_alloc(man, line, offs, MAN_sp))
 434                                 return(0);
 435                         man->next = MAN_NEXT_SIBLING;
 436                 }
 437                 return(1);
 438         }
 439
 440         /*
 441          * Warn if the last un-escaped character is whitespace. Then
 442          * strip away the remaining spaces (tabs stay!).
 443          */
 444
 445         i = (int)strlen(buf);
 446         assert(i);
 447
 448         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 449                 if (i > 1 && '\\' != buf[i - 2])
 450                         man_pmsg(man, line, i - 1, MANDOCERR_EOLNSPACE);
 451
 452                 for (--i; i && ' ' == buf[i]; i--)
 453                         /* Spin back to non-space. */ ;
 454
 455                 /* Jump ahead of escaped whitespace. */
 456                 i += '\\' == buf[i] ? 2 : 1;
 457
 458                 buf[i] = '\0';
 459         }
 460
 461         if ( ! man_word_alloc(man, line, offs, buf + offs))
 462                 return(0);
 463
 464         /*
 465          * End-of-sentence check.  If the last character is an unescaped
 466          * EOS character, then flag the node as being the end of a
 467          * sentence.  The front-end will know how to interpret this.
 468          */
 469
 470         assert(i);
 471         if (mandoc_eos(buf, (size_t)i))
 472                 man->last->flags |= MAN_EOS;
 473
 474         return(man_descope(man, line, offs));
 475 }
 476
 477 static int
 478 man_pmacro(struct man *man, int ln, char *buf, int offs)
 479 {
 480         int              i, ppos;
 481         enum mant        tok;
 482         char             mac[5];
 483         struct man_node *n;
 484
 485         if ('"' == buf[offs]) {
 486                 man_pmsg(man, ln, offs, MANDOCERR_BADCOMMENT);
 487                 return(1);
 488         } else if ('\0' == buf[offs])
 489                 return(1);
 490
 491         ppos = offs;
 492
 493         /*
 494          * Copy the first word into a nil-terminated buffer.
 495          * Stop copying when a tab, space, or eoln is encountered.
 496          */
 497
 498         i = 0;
 499         while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
 500             '\t' != buf[offs])
 501                 mac[i++] = buf[offs++];
 502
 503         mac[i] = '\0';
 504
 505         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 506
 507         if (MAN_MAX == tok) {
 508                 mandoc_vmsg(MANDOCERR_MACRO, man->parse, ln, ppos,
 509                     "%s", buf + ppos - 1);
 510                 return(1);
 511         }
 512
 513         /* The macro is sane.  Jump to the next word. */
 514
 515         while (buf[offs] && ' ' == buf[offs])
 516                 offs++;
 517
 518         /*
 519          * Trailing whitespace.  Note that tabs are allowed to be passed
 520          * into the parser as "text", so we only warn about spaces here.
 521          */
 522
 523         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 524                 man_pmsg(man, ln, offs - 1, MANDOCERR_EOLNSPACE);
 525
 526         /*
 527          * Remove prior ELINE macro, as it's being clobbered by a new
 528          * macro.  Note that NSCOPED macros do not close out ELINE
 529          * macros---they don't print text---so we let those slip by.
 530          */
 531
 532         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 533                         man->flags & MAN_ELINE) {
 534                 n = man->last;
 535                 assert(MAN_TEXT != n->type);
 536
 537                 /* Remove repeated NSCOPED macros causing ELINE. */
 538
 539                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 540                         n = n->parent;
 541
 542                 mandoc_vmsg(MANDOCERR_LINESCOPE, man->parse, n->line,
 543                     n->pos, "%s breaks %s", man_macronames[tok],
 544                     man_macronames[n->tok]);
 545
 546                 man_node_delete(man, n);
 547                 man->flags &= ~MAN_ELINE;
 548         }
 549
 550         /*
 551          * Remove prior BLINE macro that is being clobbered.
 552          */
 553         if ((man->flags & MAN_BLINE) &&
 554             (MAN_BSCOPE & man_macros[tok].flags)) {
 555                 n = man->last;
 556
 557                 /* Might be a text node like 8 in
 558                  * .TP 8
 559                  * .SH foo
 560                  */
 561                 if (MAN_TEXT == n->type)
 562                         n = n->parent;
 563
 564                 /* Remove element that didn't end BLINE, if any. */
 565                 if ( ! (MAN_BSCOPE & man_macros[n->tok].flags))
 566                         n = n->parent;
 567
 568                 assert(MAN_HEAD == n->type);
 569                 n = n->parent;
 570                 assert(MAN_BLOCK == n->type);
 571                 assert(MAN_SCOPED & man_macros[n->tok].flags);
 572
 573                 mandoc_vmsg(MANDOCERR_LINESCOPE, man->parse, n->line,
 574                     n->pos, "%s breaks %s", man_macronames[tok],
 575                     man_macronames[n->tok]);
 576
 577                 man_node_delete(man, n);
 578                 man->flags &= ~MAN_BLINE;
 579         }
 580
 581         /*
 582          * Save the fact that we're in the next-line for a block.  In
 583          * this way, embedded roff instructions can "remember" state
 584          * when they exit.
 585          */
 586
 587         if (MAN_BLINE & man->flags)
 588                 man->flags |= MAN_BPLINE;
 589
 590         /* Call to handler... */
 591
 592         assert(man_macros[tok].fp);
 593         if ( ! (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf))
 594                 goto err;
 595
 596         /* In quick mode (for mandocdb), abort after the NAME section. */
 597
 598         if (man->quick && MAN_SH == tok &&
 599             strcmp(man->last->prev->child->string, "NAME"))
 600                 return(2);
 601
 602         /*
 603          * We weren't in a block-line scope when entering the
 604          * above-parsed macro, so return.
 605          */
 606
 607         if ( ! (MAN_BPLINE & man->flags)) {
 608                 man->flags &= ~MAN_ILINE;
 609                 return(1);
 610         }
 611         man->flags &= ~MAN_BPLINE;
 612
 613         /*
 614          * If we're in a block scope, then allow this macro to slip by
 615          * without closing scope around it.
 616          */
 617
 618         if (MAN_ILINE & man->flags) {
 619                 man->flags &= ~MAN_ILINE;
 620                 return(1);
 621         }
 622
 623         /*
 624          * If we've opened a new next-line element scope, then return
 625          * now, as the next line will close out the block scope.
 626          */
 627
 628         if (MAN_ELINE & man->flags)
 629                 return(1);
 630
 631         /* Close out the block scope opened in the prior line.  */
 632
 633         assert(MAN_BLINE & man->flags);
 634         man->flags &= ~MAN_BLINE;
 635
 636         if ( ! man_unscope(man, man->last->parent, MANDOCERR_MAX))
 637                 return(0);
 638         return(man_body_alloc(man, ln, ppos, man->last->tok));
 639
 640 err:    /* Error out. */
 641
 642         man->flags |= MAN_HALT;
 643         return(0);
 644 }
 645
 646 /*
 647  * Unlink a node from its context.  If "man" is provided, the last parse
 648  * point will also be adjusted accordingly.
 649  */
 650 static void
 651 man_node_unlink(struct man *man, struct man_node *n)
 652 {
 653
 654         /* Adjust siblings. */
 655
 656         if (n->prev)
 657                 n->prev->next = n->next;
 658         if (n->next)
 659                 n->next->prev = n->prev;
 660
 661         /* Adjust parent. */
 662
 663         if (n->parent) {
 664                 n->parent->nchild--;
 665                 if (n->parent->child == n)
 666                         n->parent->child = n->prev ? n->prev : n->next;
 667         }
 668
 669         /* Adjust parse point, if applicable. */
 670
 671         if (man && man->last == n) {
 672                 /*XXX: this can occur when bailing from validation. */
 673                 /*assert(NULL == n->next);*/
 674                 if (n->prev) {
 675                         man->last = n->prev;
 676                         man->next = MAN_NEXT_SIBLING;
 677                 } else {
 678                         man->last = n->parent;
 679                         man->next = MAN_NEXT_CHILD;
 680                 }
 681         }
 682
 683         if (man && man->first == n)
 684                 man->first = NULL;
 685 }
 686
 687 const struct mparse *
 688 man_mparse(const struct man *man)
 689 {
 690
 691         assert(man && man->parse);
 692         return(man->parse);
 693 }
 694
 695 void
 696 man_deroff(char **dest, const struct man_node *n)
 697 {
 698         char    *cp;
 699         size_t   sz;
 700
 701         if (MAN_TEXT != n->type) {
 702                 for (n = n->child; n; n = n->next)
 703                         man_deroff(dest, n);
 704                 return;
 705         }
 706
 707         /* Skip leading whitespace and escape sequences. */
 708
 709         cp = n->string;
 710         while ('\0' != *cp) {
 711                 if ('\\' == *cp) {
 712                         cp++;
 713                         mandoc_escape((const char **)&cp, NULL, NULL);
 714                 } else if (isspace((unsigned char)*cp))
 715                         cp++;
 716                 else
 717                         break;
 718         }
 719
 720         /* Skip trailing whitespace. */
 721
 722         for (sz = strlen(cp); sz; sz--)
 723                 if (0 == isspace((unsigned char)cp[sz-1]))
 724                         break;
 725
 726         /* Skip empty strings. */
 727
 728         if (0 == sz)
 729                 return;
 730
 731         if (NULL == *dest) {
 732                 *dest = mandoc_strndup(cp, sz);
 733                 return;
 734         }
 735
 736         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 737         free(*dest);
 738         *dest = cp;
 739 }